The mercapturic acid pathway

Abstract

The mercapturic acid pathway is a major route for the biotransformation of xenobiotic and endobiotic electrophilic compounds and their metabolites. Mercapturic acids (N-acetyl-l-cysteine S-conjugates) are formed by the sequential action of the glutathione transferases, γ-glutamyltransferases, dipeptidases, and cysteine S-conjugate N-acetyltransferase to yield glutathione S-conjugates, l-cysteinylglycine S-conjugates, l-cysteine S-conjugates, and mercapturic acids; these metabolites constitute a “mercapturomic” profile. Aminoacylases catalyze the hydrolysis of mercapturic acids to form cysteine S-conjugates. Several renal transport systems facilitate the urinary elimination of mercapturic acids; urinary mercapturic acids may serve as biomarkers for exposure to chemicals. Although mercapturic acid formation and elimination is a detoxication reaction, l-cysteine S-conjugates may undergo bioactivation by cysteine S-conjugate β-lyase. Moreover, some l-cysteine S-conjugates, particularly l-cysteinyl-leukotrienes, exert significant pathophysiological effects. Finally, some enzymes of the mercapturic acid pathway are described as the so-called “moonlighting proteins,” catalytic proteins that exert multiple biochemical or biophysical functions apart from catalysis.

Keywords:

• Glutathione transferase
• γ-glutamyltransferase
• dipeptidase
• cysteine S-conjugate N-acetyltransferase
• NAT8
• aminoacylase
• transport
• biomarker
• mercapturomic profile
• moonlighting protein

1. Introduction

The mercapturic acid pathway plays a significant role in the detoxication of electrophiles formed by the biotransformation of xenobiotics. Indeed, the elaboration of the mercapturic acid pathway was interwoven with investigations into the metabolic fate of chemicals. Although the detoxication of xenobiotic-derived electrophiles has received the most emphasis, the mercapturic acid pathway also has a significant role in the enzymatic processing of endobiotics, particularly leukotrienes and steroids. Some enzymes of the mercapturic acid pathway are the so-called moonlighting proteins, i.e. polypeptide chains with multiple biochemical or biophysical functions (Jeffery 1999, 2018). For example, the dipeptidase aminopeptidase N is also known as CD13 and functions not only as a hydrolase but also as a viral receptor and a signaling molecule (Mina-Osorio 2008).

Mercapturic acids (S-substituted N-acetyl-l-Cys) arise from the glutathione transferase (GST)-catalyzed reaction of glutathione (GSH) with electrophilic species to give GSH S-conjugates that undergo sequential hydrolysis to l-Cys-Gly S-conjugates and thence to l-Cys S-conjugates, which undergo N-acetylation to yield the mercapturic acid (Figure 1). The metabolites formed by the mercapturic acid pathway have been proposed to constitute a “mercapturomic” profile (Gonçalves-Dias et al. 2019a, 2019b). Mercapturic acids are polar (carboxylates at physiological pH) and are readily excreted by the kidney.

Figure 1. The mercapturic acid pathway, as illustrated with the biotransformation of monochlorobimane. GSH: glutathione; GST: glutathione transferase; GGT: γ-glutamyltransferase; NAT8: Cys S-conjugate N-acetyltransferase; AcCoA: acetyl-CoA; 1, monochlorobimane; 2, l-γ-glutamyl-S-[(2,5,6-trimethyl-1,7-dioxo-1H,7H-pyrazolo[1,2-a]pyrazol-3-yl)methyl]-l-Cys-Gly; 3, S-[(2,5,6-trimethyl-1,7-dioxo-1H,7H-pyrazolo[1,2-a]pyrazol-3-yl)methyl]-l-Cys-Gly; 4, S-[(2,5,6-trimethyl-1,7-dioxo-1H,7H-pyrazolo[1,2-a]pyrazol-3-yl)methyl]-l-Cys; 5, N-acetyl-S-[(2,5,6-trimethyl-1,7-dioxo-1H,7H-pyrazolo[1,2-a]pyrazol-3-yl)methyl]-l-Cys.

Mercapturic acid formation was first reported in 1879 (Baumann and Preusse 1879; Jaffé 1879): dogs given bromobenzene or chlorobenzene excreted S-(bromophenyl)- or S-(chlorophenyl)mercapturic acid. Since these early observations, the several steps in the mercapturic acid pathway have been elucidated. The identification and characterization of GSH and the later finding that GSH is the source of the sulfur atom in mercapturic acids was intertwined with the clarification of the mercapturic acid pathway (Barnes et al. 1959; Boyland and Chasseaud 1969; Meister 1988). These observations led to the discovery of the GSTs and their central role in mercapturic acid formation (Booth et al. 1961; Habig et al. 1974b).

Baumann and Preusse (Baumann and Preusse 1879) and Jaffé (Jaffé 1879) isolated mercapturic acids from acidified urine of dogs given bromobenzene or chlorobenzene. Studies on the metabolic fate of naphthalene and several aromatic hydrocarbons in rats and rabbits showed that mercapturic acids were found in urine only after acidification (Boyland et al. 1957; Knight and Young 1957); these observations led to the concept that precursors of mercapturic acids, so-called premercapturic acids, were formed and excreted in urine. Boyland and Sims (Boyland and Sims 1958) identified the acid-labile precursor of N-acetyl-S-(naphthalen-1-yl)cysteine as N-acetyl-S-(2-hydroxy-1,2-dihydronaphthalen-1-yl)-l-cysteine; similarly, Knight and Young (Knight and Young 1958) also demonstrated the formation of premercapturic acids from benzene, naphthalene, anthracene, and halobenzenes.

The formation of an intermediate premercapturic acid is characteristic of aromatic ring conjugation by GSH, and can be illustrated by the metabolism of bromobenzene (6, Figure 2) to bromobenzene oxide 7 (Figure 2), which may react with GSH to give GSH S-conjugate 8 (Figure 2); hydrolysis and N-acetylation of the GSH S-conjugate 8 (Figure 2) gives the premercapturic acid 9 (Figure 2); acid-catalyzed dehydration of intermediate 9 (Figure 2) yields the observed S-(4-bromophenyl)mercapturic acid 10 (Figure 2).

Figure 2. Premercapturic acids as intermediates in the formation of mercapturic acids. P450: cytochromes P450; GSH: glutathione; GGT: γ-glutamyltransferase; DP: dipeptidases; NAT8: Cys S-conjugate N-acetyltransferase; 6, bromobenzene; 7, bromobenzene oxide; 8, S-(4-bromo-6-hydroxycyclohexa-2,4-dien-1-yl)GSH; 9, N-acetyl-S-(4-bromo-6-hydroxycyclohexa-2,4-dien-1-yl)-l-Cys; 10, S-(4-bromophenyl)mercapturic acid.

Structural assignments of mercapturic acids formed by the acid-catalyzed dehydration of premercapturic acids of aromatic compounds may require revision because of rearrangements of premercapturic acids after acidification (Jeffery and Jerina 1975); these authors also questioned whether premercapturic acids were the products of enzymatic reactions or whether they were formed by the nonenzymatic reaction of arene oxides with GSH.

For most compounds, mercapturic acid formation is a detoxication pathway, but some mercapturic acids or precursor Cys S-conjugates undergo bioactivation to reactive intermediates (Hashmi et al. 1992; Anders and Dekant 1998). A discussion of the bioactivation and toxicity of mercapturic acids and Cys S-conjugates is not, however, within the scope of this review. The metabolites of endogenous substrates formed by the mercapturic acid pathway may play a role in respiratory diseases, cancer, neurological disorders, and cardiometabolic diseases that contribute to the pathophysiological development of chronic inflammatory and metabolic disorders (Gonçalves-Dias et al. 2019a, 2019b).

The objective of this review is to provide a broad, detailed presentation of the formation of mercapturic acids. Descriptions of the several contributing enzymes, their discovery and characterization, substrate selectivities, inhibitors, species and tissue distribution, and biological functions are included. The review is intended to be useful to experienced investigators, students, and new investigators seeking an informative guide to the mercapturic acid pathway. Previous reviews of the mercapturic acid pathway have been published (Boyland and Chasseaud 1969; Wood 1970; Chasseaud 1973, 1976; Tate 1980; Bakke 1990; Commandeur et al. 1995; Cooper and Hanigan 2010, 2018).

2. Glutathione transferases^1,2

2.1. Introduction

GSTs catalyze the reaction of electrophilic substrates with GSH (Figure 1), the first step in the mercapturic acid pathway. The GSTs play a major role in the detoxication of toxic electrophiles: the GSH S-conjugates that are formed are typically less toxic, more water soluble, and more readily excreted than the precursor compounds.

There are three distinct mammalian GST superfamilies: (1) soluble or cytosolic GSTs, (2) mitochondrial Kappa-class GST, and (3) microsomal MAPEG (Membrane-Associated Proteins in Eicosanoid and Glutathione metabolism) (Pearson 2005; Zimniak and Singh 2007; Oakley 2011; Brown and Babbitt 2012). The cytosolic GSTs and the Kappa-class GSTs are evolutionarily derived from common ancestral proteins (Martin 1995; Frova 2006; Atkinson and Babbitt 2009). The structural differences and similarities between the cytosolic and Kappa-class GSTs led Armstrong and coworkers (Ladner et al. 2004) to propose parallel (convergent) evolution pathways for each class with thioredoxin/glutaredoxin proteins as the suggested starting points (Frova 2006). The bacterial fosfomycin resistance proteins (frp) catalyze a GST reaction and are members of the vicinal oxygen chelate (VOC) superfamily (Armstrong 2000, 2010; He and Moran 2011; Armstrong et al. 2018); these metalloenzymes are unrelated to other members of the GST superfamilies.

The GSTs were discovered during the search for the source of the Cys moiety in mercapturic acids (for a review, see Boyland and Chasseaud 1969). Two findings established GSH as the source of the Cys moiety in mercapturic acids: one, GSH S-conjugates given to animals are excreted as mercapturic acids (Barnes et al. 1959; Bray et al. 1959a). Two, liver cytosolic enzymes catalyze the formation of GSH S-conjugates (Booth et al. 1961; Combes and Stakelum 1961). Subsequent studies showed that liver enzymes catalyze the reaction of GSH with a diverse group of compounds (Booth et al. 1961; Al-Kassab et al. 1963; Boyland and Williams 1965; Johnson 1966; Suga et al. 1967; Gillham 1973; Armstrong 2010; Cooper and Hanigan 2010; Armstrong et al. 2018).

The cytosolic GSTs were the first GSTs to be identified and have been extensively studied (for reviews, see Mannervik and Danielson 1988; Coles et al. 1990; Awasthi et al. 1994; Hayes and Pulford 1995; Sheehan et al. 2001; Sherratt and Hayes 2002; Hayes et al. 2005; Josephy and Mannervik 2006; Wu and Dong 2012; Mohana and Achary 2017; Allocati et al. 2018). Later studies showed that GST activity was also present in hepatic microsomal, mitochondrial, peroxisomal, and nuclear fractions (Suga et al. 1967; Glatt and Oesch 1977; Kraus and Gross 1979; Morgenstern et al. 1979; Wahllander et al. 1979; Nishino and Ito 1990b).

This section will focus on structure and properties of GSTs along with their role in the mercapturic acid pathway. A discussion of the genetics and phylogeny of GSTs is beyond the scope of this review. Several reviews are, however, available (Nebert and Vasiliou 2004; Pearson 2005; Dourado et al. 2008a; Josephy 2010; Board and Menon 2013). The polymorphic variants in human GSTs have been summarized, and the functional effects of the polymorphisms, as well as their clinical and toxicological relevance, have been discussed (Strange et al. 2001; Holley et al. 2007; Board and Menon 2013).

2.2. Cytosolic GST superfamily (EC 2.5.1.18)³

2.2.1. Discovery and nomenclature

Early studies in which GST activities were measured with a range of substrates indicated that several GSTs were present in rat liver tissue and were assigned names based on the assay substrate used, i.e. GSH S-aryltransferase, GSH S-alkyltransferase, GSH S-aralkyltransferase, GSH S-epoxidetransferase, and GSH S-alkenetransferase (Boyland and Chasseaud 1969; Chasseaud 1974).

Gillham (1971, 1973) reported the partial purification of GSTs that catalyze the conjugation of 1-araalkyl sulfates with GSH. Jakoby and coworkers described the column chromatographic purification and preliminary characterization of rat liver GSTs (Fjellstedt et al. 1973; Pabst et al. 1973, 1974; Habig et al. 1974b, 1976); all are dimeric proteins with subunit molecular masses of approximately 25 kDa and holoenzyme molecular masses of approximately 45 kDa. When activities were measured with a panel of substrates, the purified GSTs showed broad and overlapping substrate selectivities, although some isozymes showed clear substrate preferences.

Later studies demonstrated the utility of immobilized GSH or S-hexylGSH affinity chromatography for purification of GSTs (Simons and Vander Jagt 1977; Guthenberg and Mannervik 1979). Affinity chromatography with high-pressure liquid chromatography (HPLC) and mass spectrometry for the characterization of GSTs was also reported (Rouimi et al. 1995; Listowsky et al. 1998).

Earlier, Levi et al. (1969) identified two hepatic cytosolic proteins, designated “Y” and “Z,” that bound large hydrophobic nonsubstrate compounds, e.g. sulfobromophthalein, bilirubin, indocyanine green, and other organic anions, and were termed ligandins (Habig et al. 1974a; Ketley et al. 1975). The rat liver protein “Y” was later demonstrated to be identical with GST B (Litwack et al. 1971; Habig et al. 1974a). There is considerable heterogeneity in the properties of ligandin binding sites among GSTs (Oakley 2011): with hGSTP1-1, the ligandin binding site occupies part of the second substrate binding site (“H-site”) (Oakley et al. 1999), whereas with hGSTA1-1, a large site that spans both subunits serves as a ligandin binding site (Le Trong et al. 2002). The ligandin binding site in hGSTO1-1 is found deep within the dimer interface (Brock et al. 2013).

GST Theta (GSTT) (Meyer et al. 1991), GST Zeta (GSTZ) (Board et al. 1997, 2001), and GST Omega (GSTO) (Board et al. 2000) were identified with BLAST searches of the human expressed sequence tag (EST) database. Sigma-class GSTs (GSTS) were discovered by sequence alignment and phylogenetic analysis of GST-like proteins in cephalopod tissues (Tomarev and Zinovieva 1988).

The multiplicity of GSTs and the several substrates used to measure activity led to inconsistent and confusing nomenclature. Rat GSTs can be differentiated based on column chromatographic behavior, substrate selectivity, isoelectric point, SDS-polyacrylamide gel electrophoresis, amino acid composition, and sensitivity to inhibitors (Bass et al. 1977; Jakoby 1978; Mannervik and Jensson 1982; Yälçin et al. 1983; Jakoby et al. 1984).

A standardized nomenclature for the human GSTs based on amino acid and gene sequences was described by Mannervik et al. (1992, 2005) and has been extended to other species (Hayes and Pulford 1995). The criteria for classifying mammalian GSTs have been summarized (Sheehan et al. 2001). With the standardized nomenclature, each subunit encoded by a discrete gene has its own designation. Subunits are grouped into classes or gene families of GSTs and are numbered sequentially with Arabic numerals. The recognized classes of cytosolic GSTs include: Alpha, Mu, Pi, Theta, Zeta, Omega, and Sigma, which are identified by upper-case letters, e.g. A, M. P, T, Z, O, and S. Species designations are indicated by a lower-case letter preceding the GST abbreviation, e.g. h, human; r, rat; m, murine. GSTs are dimeric proteins, and subfamily homo- and heterodimers are denoted with Arabic numerals separated by a hyphen. For example, a homodimeric rat Alpha-class GST from subfamily A would be written as rGSTA1-1. Several designations have been used in the literature for allelic variants, although currently most authors use the “star” designation, e.g. GSTM1*A-1*A.

Subunits are identified by primary DNA sequence homology. Members of the same class share 40–50% sequence identity but less than approximately 25–30% sequence identity with GSTs in other classes. The nomenclature for human, rat, and mouse GSTs is summarized in Tables 1–3. A mouse GST database is available (http://www.people.virginia.edu/∼wrp/gst_mouse/gst_mouse.html; accessed 31 October 2019).

Table 1. Human cytosolic glutathione transferases.

Class	Enzyme	Gene	Accession number*	Other names
Alpha	GST A1-1	GSTA1	NP_665683, P08263	ε, B1B1, GST2-type 2, Ha (subunit 1), αxαx
	GST A2-2	GSTA2	NP_000837, P09210	γ, B2B2, GST2-type 2, Ha (subunit 2), αyαy
	GST A3-3	GSTA3	NP_000838, Q16772
	GST A4-4	GSTA4	NP_001503, O15217
	GST A5-5	GSTA5	NP_714543, Q7RTV2
Mu	GST M1-1	GSTM1	NP_666533, P09488	μ, GST1-type 2, Hb (subunit 4)
	GST M2-2	GSTM2	NP_000839, P28161	Muscle, GST4
	GST M3-3	GSTM3	NP_000840, P21266	Brain, GST5
	GST M4-4	GSTM4	NP_671489, Q03013
	GST M5-5	GSTM5	NP_000842, P46439
Pi	GST P1-1	GSTP1	NP_000843, P09211	π, GST3
Theta	GST T1-1	GSTT1	NP_000844, P30711	GSTΘ
	GST T2-2	GST2	NP_000845, G9J6Q5
Zeta	GST Z1-1	GSTZ1	NP_665877, O43708
Omega	GST O1-1	GSTO1	NP_004823, P78417
	GST O2-2	GSTO2	NP_899062, Q9H4Y5
Sigma	PGD2/GST S1-1	PGD2	NP_055300, O60760

* NCBI and UniProt accession numbers.

Data from Mannervik et al. (1992) and Mannervik (2005).

Table 2. Mouse cytosolic glutathione transferases*.

Class	Enzyme	Gene	Accession number†	Other names
Alpha	GST A1-1	Gsta1	NP_032207, P13745	Ya1
	GST A2-2	Gsta2	NP_032208,P10648	Ya2, Yc2
	GST A3-3	Gsta3	NP_034486, P30115	GT10.6, Ya3, Yc
	GST A4-4	Gsta4	NP_034487, P24472	Yk, GST5.7
	GST A5-5	Gsta5		A5
Mu	GST M1-1	Gstm1	NP_034488, P10649	GT8.7, Yb1
	GST M2-2	Gstm2	NP_032209, P15626	Yb2
	GST M3-3	Gstm3	NP_034489, P19639	GT9.3, μ4, Yb3
	GST M4-4	Gstm4	NP_081040	Yb5, μ7
	GST M5-5	Gstm5	NP_034490, P48774	Fsc2m, mGSTM5
	GST M6-6	Gstm6	NP_032210	mGSTM5
	GST M7-7	Gstm7	NP_080948, Q80W21	μ3
Pi	GST P1-1	Gstp1	NP_038569, P19157	Yf1, piB
	GST P2-2	Gstp2	NP_861461, P46425	Yf2, piA
Theta	GST T1-1	Gstt1	NP_032211, Q64471	42129
	GST T2-2	Gstt2	NP_034491, Q61133	Yrs, 12-12
	GST T3-3	Gstt3	NP_598755, Q99L20	5-5
Zeta	GST Z1-1	Gstz1	NP_034493, QWVL0	Maleylacetoacetate isomerase
Omega	GST O1-1	Gsto1	NP_034492, O09131	p28
	GST O2-2	Gsto2	NP_080895, Q8K2Q2
Sigma	PTGDS2/GST S1-1	Ptgds2	NP_062328, Q9JHF7	GSH-dependent prostaglandin D2 synthase

* Data from: Hayes and Pulford (1995), Andorfer et al. (2004), and Mannervik et al. (2005).

Mouse Glutathione Transferase NomenclatureST (http://people.virginia.edu/∼wrp/gst_mouse/gst_mouse.html).

† NCBI and UniProt accession numbers.

Table 3. Rat cytosolic glutathione transferases*.

Class	Enzyme	Gene	Accession number†	Other names
Alpha	GST A1-1	Gsta1	NP_058709, P00502	GST B, GST 1-1, Ligandin, Ya1, 1a-1a
	GST A2-2	Gsta2	NP_001010921, P04903	Ya2, 1b-1b
	GST A3-3	Gsta3	NP_113697, P04904	GST AA, GST 2-2, Yc1
	GST A4-4	Gsta4	P14942, XP_217195	GST K, GST 8-8, Yk, Yα
	GST A5-5	Gsta5	NP_001009920, P46418	Yc2
Mu	GST M1-1	Gstm1	NP_058710, P04905	GST A, GST 3-3, Yb1
	GST M2-2	Gstm2	NP_803175, P08010	GST D, GST 4-4, Yb2
	GST M3-3	Gstm3	NP_112416, P08009	GST 6-6, Yb3, Yn1, Yβ
	GST M4-4	Gstm4	NP_065415, B29231	Yb4
	GST M5-5	Gstm5	NP_742035	GST 11-11, Yn2
	GST M6-6	Gstm6	NP_001099934	GST 9-9, Yo
Pi	GST P1-1	Gstp1	NP_036709, P04906	GST P, GST 7-7, Yf, Yp, Yδ
Theta	GST T1-1	Gstt1	NP_445745, Q01579	GST E, GST 5-5, Yrs'
	GST T2-2	Gstt2	NP_036928, P30713	GST M, GST 12-12, Yrs
Zeta	GST Z1-1	Gstz1	P57113	Maleylacetoacetate isomerase
Omega	GST O1-1	Gsto1	Q9Z339	Dehydroascorbate reductase
	GST O2-2	Gsto2	Q6AXV9
Sigma	PTGDS2/GST S1-1	Ptgds2	NP_113832, O35543	GSH-dependent prostaglandin D2 synthase

* Data from: Jakoby et al. (1984), Hayes and Pulford (1995), and Mannervik et al. (2005).

† NCBI and UniProt accession numbers.

2.2.2. Structure and properties common to cytosolic GSTs

The determination of the amino acid and DNA sequences for cytosolic GSTs was followed by crystallographic analysis, which provided insight into the relationship between structure and catalytic activity. Indeed, three-dimensional structures are available for all human cytosolic GSTs, except hGSTA5-5, for many rat and mouse cytosolic GSTs, and for Kappa-class GST (Oakley 2011; Wu and Dong 2012; Rose et al. 2017). Reviews about the structure of GSTs are available (Armstrong 1997; Salinas and Wong 1999; Sheehan et al. 2001; Pearson 2005; Frova 2006; Dourado et al. 2008a; Atkinson and Babbitt 2009; Armstrong 2010; Oakley 2011; Wu and Dong 2012; Mohana and Achary 2017; Armstrong et al. 2018).

Canonical cytosolic GSTs are homo- or heterodimeric proteins; although homodimers apparently predominate in vivo, heterodimeric GSTs, e.g. GSTA1-2, have been purified from human, rat, and hamster tissues (Mannervik and Jensson 1982; Tu and Reddy 1985; Bogaards et al. 1992; Czerwinski et al. 1996). The subunit masses are approximately 25 kDa (Eaton and Bammler 1999; Mohana and Achary 2017).

Early studies recognized that the requirement for GSH and a second electrophilic substrate likely indicated the presence of distinct substrate and electrophile binding sites (Jakobson et al. 1977). Mannervik et al. (1978, 1985) proposed the terms “G-site” and “H-site” for the binding sites for GSH and the second substrate, respectively.

Each subunit consists of distinct N- and C-terminal domains (also termed Domains 1 and 2, respectively) that are connected by a short linker sequence between helices α₃ and α₄ (Ji et al. 1992; Reinemer et al. 1992; Dirr et al. 1994b) (Figure 3). Among the cytosolic GSTs, amino acid sequence similarities are highest (∼70%) for the N-terminal domains but are lower (∼30%) and vary considerably in the C-terminal domains (Salinas and Wong 1999; Armstrong 2010); in spite of the differences in sequence similarities, the spatial structures of, for example, GSTA, -M, and -P are very similar (Dirr et al. 1994b).

Figure 3. (A) Structural representation of cytosolic GSTs. The thioredoxin fold of the N-terminal domain is represented by β₁α₁β₂α₂β₃β₄α₃ (filled arrows and rectangles). The α-helices of the C-terminal domain are shown as stippled rectangles; the number of α-helices in the C-terminal domain varies among the GST subfamilies. The broken line represents the linker between the N- and C-terminal domains. Modified from Frova (2006); (B) crystal structure of hGSTA1 in complex with S-(benzyl)GSH; (C) crystal structure of hGSTA1-1 in complex with S-(benzyl)GSH (PDB 1GUH). Panels (B) and (C) are reproduced with permission from the publisher and from Wu and Dong (2012); the authors thank Prof. Wu for providing (B) and (C).

The N-terminal domain adopts the thioredoxin fold represented by a β₁α₁β₂ motif at the N-terminus and is connected to a β₃β₄α₃ motif at the C-terminus by a loop containing the α₂ helix (Figure 3). The C-terminal domain contains five or more α-helices, depending on the subfamily, e.g. GSTA, -T and -O have an α₉ helix (Dirr et al. 1994b; Sheehan et al. 2001; Frova 2006; Dourado et al. 2008a; Oakley 2011; Wu and Dong 2012).

The H-site is typically located in the cleft between the N- and C-domains, and amino acid residues from both domains contribute to the interaction of substrates with the H-site (Mohana and Achary 2017). Crystallographic analyses of GSTs showed that amino acid residues of the G- and H-sites are found in both the N-terminal and C-terminal domains (Ji et al. 1992; Sinning et al. 1993; Ji et al. 1994; Wilce and Parker 1994; Martin 1995; Sheehan et al. 2001; Patskovsky et al. 2006; Deponte 2013). Mohana and Achary (2017) indicate that the H-site for cytosolic GSTs, except GSTK, can be divided into three regions: for GSTA, the loop connecting α₁ and β₂ is designated as H1; residues from the C-terminal portion of α₄ constitute the H2 region; and variable residues from the C-terminal α-helix constitute the H3 region. Details about the composition of the H-sites for each of the cytosolic GSTs are available (Mohana and Achary 2017). The size, topography, and amino acid residues of the H-site play a critical role in the electrophilic substrate selectivity of GSTs (Kurtovic et al. 2008; Shokeer and Mannervik 2010b; Wu and Dong 2012; Modén and Mannervik 2014; Mohana and Achary 2017). Activity-based probes that target the H- and G-sites of cytosolic GSTs have been reported (Stoddard et al. 2017).

Although the highly conserved structure and sequence of the G-site is well documented, GSTs vary in the identity and location of catalytically important amino acid residues in the G-site (for a description of amino-acid residues involved in GSH binding, see Mohana and Achary 2017). Binding of GSH to the G-site is dependent on polar interactions between GSH and the β₃β₄α₃ motif, and the thiol of GSH forms hydrogen bonds with the wall of the G-site.

A critical Tyr residue in the G-site, which serves to stabilize the GSH thiolate, is replaced in several GSTs by a Ser or Cys. Accordingly, Atkinson and Babbitt (Atkinson and Babbitt 2009) designated GSTA, GSTM, GSTP, and GSTS as “Y-GSTs” and GSTT, GSTZ, and GSTO as “S/C-GSTs.” The highly conserved G-site shows high selectivity for GSH, although some GSH analogs are substrates (Chen et al. 1988; Adang et al. 1989, 1990); the γ-Glu residue of GSH is a major binding determinant (Adang et al. 1990).

The quaternary structure of cytosolic GST dimers is characterized by a two-fold axis of symmetry that extends through the interface. The nature of the interactions between the dimers differs among the cytosolic GST subfamilies. In the Alpha-, Mu-, and Pi-class GSTs, subunit interactions are attributed primarily to a lock-and-key motif consisting of hydrophobic interactions between an aromatic amino acid (Phe52 in hGSTA, Phe56 in hGSTM, and Tyr49 in hGSTP) that protrudes from the loop connecting α₂-helix with β₃-strand in the N-terminal domain and interacts with a hydrophobic pocket formed by α₄ and α₅ helices on the C-terminal domain of the second monomer (Sinning et al. 1993; Sayed et al. 2000; Sheehan et al. 2001; Hegazy et al. 2004; Alves et al. 2006; Thompson et al. 2006). Electrostatic interactions between polar residues on the β₄ strand and the α₃ helix with residues on the α₄ helix of the second monomer also contribute to stabilization of GSTA, GSTM, and GSTP dimers (Thompson et al. 2006; Wu and Dong 2012). With Sigma-class GSTs, polar interactions between subunits serve to maintain their dimeric structure (Stevens et al. 1998), whereas with Omega-class GSTs, nonpolar interactions between subunits dominate (Board et al. 2000). With hGSTZ, nonpolar interactions between monomers prevail, but a polar interaction between Glu81 and Arg96 is also observed (Polekhina et al. 2001).

The properties of the subunit interface may exert significant effects on catalytic activity. Binding of the cosubstrate in GSTP1-1 increases the affinity for GSH by several folds (Caccuri et al. 1996). With hGSTA1-1, the side-chain of Trp21 in the N-terminal domain interacts with helices α₆ and α₈ in the C-terminal domain (Balchin et al. 2010); the W21A mutant affects the H-site (decreased k_cat^CDNB) but has little effect on the G-site (little change in k_cat^GSH).

Although cytosolic GSTs are functional dimers, the question has been raised whether catalytically competent GST monomers exist in cells. Conflicting conclusions about the existence of stable, catalytically active GSTP1 monomers have been reported, but the preponderance of the data indicate that catalytically active GSTP1 monomers are likely not present in cells (for a summary, see Fabrini et al. 2009). It was also reported that regulation of JNK signaling was attributable to a GSTP1 monomer (Adler et al. 1999); subsequent studies show, however, that dimeric GSTP1-1 rather than the monomer GSTP1 most likely serves to regulate JNK signaling (Gildenhuys et al. 2010b).

GSTs catalyze the reaction of GSH with a range of carbon-, sulfur-, and oxygen containing electrophilic substrates (Deponte 2013). In addition to catalyzing transferase reactions (S_NAr, Michael additions, epoxide ring opening), where the GSH moiety is incorporated in the product GSH S-conjugate, GSTs also catalyze isomerization reactions and the reduction of peroxides and disulfides. The catalytic mechanisms of transferase, isomerization, and reductive reactions show considerable mechanistic diversity. The mechanism of GST-catalyzed reactions has been the subject of intense investigation (for reviews, see Sheehan et al. 2001; Dourado et al. 2008a; Atkinson and Babbitt 2009; Wu and Dong 2012). Deponte (2013) has described several models of GST-catalyzed reactions and illustrated how they differ among the GST classes and substrates.

A major event in the catalytic cycle is the formation and stabilization of the GSH thiol as a thiolate (Armstrong 1997; Armstrong 2010). It was originally proposed that the Tyr7 residue of hGSTP1-1 acted as a general base in promoting the formation of the GSH thiolate (GS^-) (Karshikoff et al. 1993). Later studies showed, however, that at the pH optimum for hGSTA1-1 the Tyr residue was largely present as the conjugate acid (Björnestedt et al. 1995); moreover, the Y9F mutant retains partial catalytic activity. Rather than serving as a general base, the catalytically important Tyr is now believed to play a key role in stabilizing the reactive, negatively charged GSH thiolate. With C/S-GSTs, a hydrogen-bonding interaction between the active-site Ser or Cys plays a major role in stabilizing the GSH thiolate (e.g. Liu et al. 1992). In GSTA1-1-, GSTM1-1-, and GSTP1-1-catalyzed reactions, the binding of GSH to the G-site is accompanied by a decrease in the pK_a of the thiol group from 9.1 to 6.2–6.6 (Caccuri et al. 1999; Dourado et al. 2008a). Other studies on the binding and activation of GSH and the fate of the thiol proton in cytosolic GSTs indicate that a multistep process is involved and that a water molecule may assist in catalysis (Xiao et al. 1996; Parraga et al. 1998; Caccuri et al. 1999; Dourado et al. 2008b, 2009, 2010b). Regardless of the mechanistic details of deprotonation of GSH to form the highly nucleophilic thiolate, the most significant features of the overall GSH activation process include the rapid formation and stabilization of the thiolate upon binding of GSH to the G-site, where it is poised to react with electrophilic substrates.

In hGSTM1-1 the direct (first-sphere) interaction of the GSH thiolate with Tyr6 is supplemented by a range of second-sphere electrostatic interactions (Xiao et al. 1996). Arg15 also stabilizes the GSH thiolate in hGSTA1-1 (Björnestedt et al. 1995). The differences in catalytic rates and regio- and stereoselectivity observed among cytosolic GSTs are attributed to differences in stabilization of transition states of the specific reactions (Armstrong 2010; Deponte 2013).

The presence of H- and G-sites raised the question of whether both subunits were catalytically active (all-of-the-sites reactivity) in the holoenzyme or whether only one site was active (half of the sites reactivity). The two active sites present in each subunit are kinetically independent (Danielson and Mannervik 1985). GSTA1-1 exhibits all-of-the-sites reactivity in S_N2_Ar reactions but half-of the-sites reactivity in addition reactions (Lien et al. 2001).

As with many other xenobiotic-metabolizing enzymes, the cytosolic GSTs are induced by a range of chemicals (for reviews, see Higgins and Hayes 2011; Boušová and Skálová 2012; Ma and He 2012). A detailed discussion of the induction of GSTs by chemicals is beyond the scope of this review, but some salient points merit mention. Two xenobiotic-responsive elements (XREs) are implicated in the induction of GSTA1-1: one requires the aryl-hydrocarbon receptor (AhR) for the induction of GSTA1-1 by planar aromatic compounds, e.g. β-naphthoflavone (Rushmore et al. 1990); these compounds also induce Phase-I enzymes, e.g. cytochromes P450, and are termed bifunctional inducers. A second group of compounds, e.g. phenolic antioxidants, induce the transcriptional activation of GSTA1-1 independent of the AhR and are termed monofunctional inducers; the XRE involved was named the antioxidant-responsive element (ARE) (Rushmore and Pickett 1990). It was also demonstrated that bifunctional inducers require biotransformation by cytochromes P450 to induce GSTs. The bZIP transcription factor Nrf2 is required for the ARE-dependent transcriptional activation of GSTs (Itoh et al. 1997). In addition to Nrf2, other transcription factors regulate the activity of ARE (Motohashi et al. 2002; Hayes et al. 2005). The activity of the microsomal GSTs is also induced by xenobiotics (Higgins and Hayes 2011).

Activity-based protein profiling (ABPP) of murine cytosolic GSTs has been used to explore the effect of benzo[a]pyrene (B[a]P) induction on protein expression and on G- and H-site activities (Stoddard et al. 2019). The approach involves the use of two activity-based probes, a GSH-based photoaffinity probe that targets the G-site and a reactive GST inhibitor that binds to H-site residues, which are incubated with tissue cytosols followed by protein and proteomic analysis (Stoddard et al. 2017). In mice given B[a]P, both expression and G- and H-site activities of GSTA2, -M4, and -P1 were increased, whereas an expression-independent increase in H-site activity was found with GSTM1, -2, and -6. In contrast, an expression-independent inhibition of G- and H-site activities of GSTA4 was observed. The mechanisms underlying these observed changes have not been elucidated but the possible involvement of post-translational modification-induced conformational changes was suggested.

2.2.3. Substrates

Cytosolic GSTs exhibit broad substrate selectivity for xenobiotic substrates, although some GST classes show significant selectivity. Substrates for the several classes of cytosolic GSTs are discussed below. A summary of the catalytic activities of rat, mouse, and human GSTs has been published (Hayes and Pulford 1995). Deponte (2013) has reviewed how structural features of the cytosolic GSTs determine substrate selectivity. GSTs also catalyze reactions with physiological substrates, including steroid isomerization, maleylacetoacetic acid isomerization, prostaglandin reduction and synthesis, and organic peroxide reduction (Sharma et al. 2007). Prodrugs of cancer chemotherapeutic agents that undergo GST-catalyzed activation have also been reported (Morgan et al. 1998; Zhao and Wang 2006; Axarli et al. 2009; Johansson et al. 2011; Ruzza and Calderan 2013; Ramsay and Dilda 2014).

2.2.4. Inhibitors

Considerable effort has been expended in the development of GST (particularly GSTP) inhibitors, largely to overcome GST-dependent drug resistance (for reviews, see van Bladeren and van Ommen 1991; Tew et al. 1997; Burg and Mulder 2002; Townsend and Tew 2003; Mahajan and Atkins 2005; Mathew et al. 2006; Zhao and Wang 2006; Li et al. 2010; Thurairatnam 2012; Wu and Batist 2013; Mohana and Achary 2017). Inhibitors of the several classes of cytosolic GSTs are discussed below. Few GST inhibitors that show activity in vitro appear to have been tested for activity in vivo but compounds effective in animal models have been reported (Ouwerkerk-Mahadevan et al. 1996; Mulder and Ouwerkerk-Mahadevan 1997; Ouwerkerk-Mahadevan and Mulder 1998; Gale and Tew 2001).

2.2.5. Tissue and species distribution

The cytosolic GSTs are widely distributed in mammals, and individual GSTs often differ in tissue distribution. For example, Alpha- and Mu-class GSTs are intensely expressed in the liver, whereas Pi-class GST is scarcely expressed in the liver but is highly expressed in various nonhepatic tissues. For compilations of the tissue distribution of GSTs, see Baars et al. (1981), Mannervik (1985), Mannervik and Widersten (1995), Rowe et al. (1997), Lantum et al. (2002a), Dhanani and Awasthi (2007), Holley et al. (2007) and Mohana and Achary (2017). Activity-based probes that are targeted at the H- and G-sites of GSTs have been described (Stoddard et al. 2017); these probes allow site-specific profiling of cytosolic GST activities in multiple tissues.

2.2.6. Cytosolic GST classes

2.2.6.1. Alpha-class (GSTA)

2.2.6.1.1. Structure and properties

Alpha-class GSTs have been found in all mammalian species studied. Links to the human, mouse, and rat genes encoding Alpha-class GSTs are listed in Tables 1–3. The expression of hGSTA5 has been proposed to be limited to unidentified physiological or pathological conditions (Singh et al. 2010).

The crystal structure of hGSTA1-1 with the bound inhibitor S-(benzyl)GSH showed two domains in each monomer and identified the key amino acid residues in the G- and H-sites that form hydrogen bonds and salt links with the inhibitor (Sinning et al. 1993). Arg15, a G-site residue that is located near the juncture with the H-site, was proposed to be within hydrogen bonding distance of the GSH sulfur atom of the bound inhibitor; a more recent crystal structure of the GSH-hGSTA1-1 complex does not, however, show hydrogen bonding between Arg15 and GSH (Grahn et al. 2006, PDB: 1PKW). Arg 15 is conserved in all GSTA class members except GSTA5-5 (Singh et al. 2010); mutating Arg15 to Leu, Ala, or His in GSTA1-1 causes a substantial reduction in catalytic activity (Björnestedt et al. 1995). Arg15 has a modest influence on the affinity of GSTA1-1 for GSH, but mutation of Arg15 to Leu raises the pKa of bound GSH from 6.5 to 7.6, indicating that Arg15 contributes to the stabilization of the GSH thiolate in the active site. In the absence of hydrogen bonding between Arg15 and the GSH thiolate, the stabilization of the thiolate by Arg15 was attributed to electrostatic forces that may involve its participation in an electron-sharing network that contributes to the deprotonation of GSH and stabilization of GSH thiolate (Winayanuwattikun and Ketterman 2005; Gildenhuys et al. 2010a). A computational analysis of the activation of GSH by GSTA1-1 indicates that Arg15 forms an ion-dipole and a hydrogen bond with the Cys carbonyl of GSH that persists through the transition state and in the GSH thiolate complex (Dourado et al. 2010a).

Alpha-class GSTs, especially GSTA3-3, play an important role in steroidogenesis by catalyzing the isomerization of the C5–C6 double bond in Δ⁵-androstene-3,17-dione (Δ5-AD) (13, Figure 4) to form the testosterone precursor Δ⁴-androstene-3,17-dione (14, Figure 4) (Johansson and Mannervik 2001). The hGSTA3-3 H-site differs in three residues from the H-site of hGSTA1-1 and differs in five H-site residues from hGSTA2-2. Compared with hGSTA3-3, the catalytic efficiency, as represented by k_cat/K_m, for isomerization of the C5-C6 double bond in Δ5-AD (13, Figure 4) to the C4–C5 position is 10-fold lower for hGSTA1-1 and 5000-fold lower for hGSTA2-2. Site-directed mutagenesis in which five H-site residues in hGSTA2-2 were replaced by the corresponding five variant H-site residues (Phe10, Gly12, Leu111, Ala208, and Ala216) of hGSTA3-3 resulted in a 3500-fold increase in the hGSTA2-2 k_cat/K_m for isomerization of Δ5-AD, making it comparable to that of hGSTA3-3 (Pettersson et al. 2002).

Figure 4. GSTA substrates. 11, 1-chloro-2,4-dinitrobenzene; 12, S-(2,4-dinitrophenyl)glutathione; 13, Δ⁵-androstene-3,17-dione (Δ5-AD, carbon atoms 4, 5, and 6 are numbered); 14, Δ⁴-androstene-3,17-dione (carbon atoms 4, 5, and 6 are numbered); 15, cumene hydroperoxide; 16, 2-phenyl-2-propanol; 17, 4-hydroxy-2-nonenal (HNE); 18, l-γ-Glu-S-[2-hydroxy-1-(2-oxoethyl)heptyl]-l-Cys-Gly (HNE-GSH); 19, aflatoxin B₁-8,9-epoxide; 20, 8,9-dihydro-8-(S-glutathionyl)-9-hydroxyaflatoxin B₁; 21, busulfan; 22, γ-Glu-β-(S-tetrahydrothiophenium)-Ala-Gly; 23, azathioprine; 24, 6-mercaptopurine; 25, 1-methyl-4-nitro-5-(glutathion-S-yl)-1H-imidazole; 26, melphalan; 27, N-(2-glutathion-S-ylethyl)-N-(2-chloroethyl)-l-phenylalanine; 28, chlorambucil; 29, N-(2-glutathion-S-ylethyl)-N-(2-chloroethyl)benzenebutanoic acid; 30, (7 R,8S,9R,10S)-(-)-syn-benzo[a]pyrene-7,8-diol 9,10-epoxide ((-)-syn-BPDE); 31, (7 R,8S,9R,10R)-10-(glutathion-S-yl)-7,8,9,10-tetrahydrobenzo[a]pyrene-7,8,9-triol; 32, (8S,10S)-7,8,9,10-tetrahydro-6,8,11-trihydroxy-8-(2-hydroxyacetyl)-1-methoxy-10-[[2,3,6-trideoxy-3-[[(2,4-dinitrophenyl)sulfonyl]amino]-α-l-lyxo-hexopyranosyl]oxy]-5,12-naphthacenedione (DNS-DOX); 33, doxorubicin; 34, N-(3′-((2,4-dinitrophenyl)sulfonamido)-3-oxo-3H-spiro[isobenzofuran-1,9′-xanthen]-6′-yl)acetamide (DNs-AcRh); 35, 2-(3-(acetylimino)-6-amino-3H-xanthen-9-yl)benzoic acid (N-acetylrhodamine); 36, 2-(7-((2,4-dinitrophenyl)sulfonamido)-4-methyl-2-oxochroman-3-yl)acetic acid (DNs-Coum); 37, 2-(7-amino-4-methyl-2-oxochroman-3-yl)acetic acid; 38, (Z)-N-(9-amino-5H-benzo[a]phenoxazin-5-ylidene)-2,4-dinitrobenzenesulfonamide (DNs-CV). 39, cresyl violet; 40, 4-acetyl-2-nitro-N-(2-oxo-4-(trifluoromethyl)-2H-chromen-7-yl)benzenesulfonamide (ANs-AFC); 41, S-(4-acetyl-2-nitrophenyl)GSH; 42, 7-amino-4-(trifluoromethyl)coumarin (AFC); 43, 4-[3-(4-acetyl-2-nitrobenzenesulfonyl)phenyl]-4-methoxyspiro[1,2-dioxetane-3,2′-adamantane] (ANs-AMPD); 44, 3-(2-spiroadamantane)-4-methoxy-4-phenyl-1,2-dioxetane (AMPD).

The crystal structure of the complex of hGSTA2-2 with Δ5-AD (13, Figure 4) shows that the steroid substrate cannot achieve an orientation in the active site similar to the one seen in hGSTA3-3, due in part to unfavorable steric interactions with Phe111 and Met208, which do not occur with Leu111 and Ala208 in hGSTA3-3 (Tars et al. 2010). Examination of the crystal structure of hGSTA3-3 in a complex with both GSH and Δ5-AD revealed that Δ5-AD (13, Figure 4) is bound with a C4 hydrogen atom proximal to the GSH thiolate and the phenolic hydroxyl of Tyr9 is located near C6. A model was suggested in which Tyr9 plays an “auxiliary” role in catalysis by forming a hydrogen bond with the sulfur of GSH to facilitate its conversion to the thiolate, which functions as an acid/base catalyst in the proton transfer from C4 to C6 during isomerization of the double bond to the C4–C5 position (Tars et al. 2010).

A quantum mechanics/molecular mechanics computational analysis found that the isomerization reaction is a concerted, but asynchronous, reaction in which GSH serves as an acid/base catalyst that requires Tyr9 to act as a proton shuttle (Calvaresi et al. 2012). In this proposed mechanism, Arg15 forms a hydrogen bond with the sulfur of GSH thiolate. A similar proton donor function for Tyr9 had been suggested earlier by Gu et al. (2004). The proposal that the GSH thiolate effects both proton abstraction from C4 of Δ5-AD (13, Figure 4) and transfer of the proton to C6 (Tars et al. 2010) was modified as the result of a DFT computational analysis of the reaction (Dourado et al. 2014). The modified proposal retains the initial proton abstraction from C4 by the GSH thiolate to form GSH and the negatively charged enolate of Δ5-AD (13, Figure 4), which is followed by breaking of the weakened hydrogen bond between GSH and Tyr9 and a conformational rearrangement of Tyr9 that allows it serve as the proton donor to C6 of the enolate to form Δ⁴-androstene-3,17-dione (14, Figure 4). Following a second conformational change, Tyr9 is protonated by GSH to return the enzyme to its original state. Other studies indicate that Arg15 is as important as Tyr9 in the isomerization reaction and plays an important role in decreasing the pK_a of GSH (Robertson et al. 2017).

4-Hydroxy-2-nonenal (17, Figure 4), formed by the peroxidative degradation of arachadonic acid, is a substrate for GSTs (Ålin et al. 1985). hGSTA4-4 shows far higher catalytic activities than hGSTA1-1 with 4-hydroxy-2-nonenal (17, Figure 4) and related toxic alkenal products of lipid peroxidation (Hubatsch et al. 1998). The high catalytic activity of hGSTA4-4 for 4-hydroxy-2-nonenal (17, Figure 4) is attributed to the H-site Tyr212 residue, which enhances the electrophilicity of the aldehyde by forming a specific hydrogen bond with the carbonyl oxygen, causing increased polarization of the carbonyl group (Bruns et al. 1999).

SNPs have been identified in hGSTA1, hGSTA2, and hGSTA3 (for a summary of GSTA allelic polymorphisms and their racial distribution, see Coles and Kadlubar 2005).

Alpha-class GSTs are induced by many agents (for compilations of GSTA inducers, see Morel et al. 1994; Hayes and Pulford 1995). hGSTA1/A2 activities are increased in human subjects with high vegetable intake (Sreerama et al. 1995; Hoensch et al. 2002). Ethoxyquin and coumarin are inducers of rat GSTA5-5 (Hayes et al. 1998). 3 H-1,2-Dithiole-3-thione is an inducer of Alpha-class GSTs (Li et al. 2005b). In mice given B[a]P, activity-based protein profiling (see Section 2.2.2 above) revealed an expression-independent decrease in both G- and H-site activities in GSTA4 (Stoddard et al. 2019).

2.2.6.1.2. Substrates

Although the Alpha-class GSTs show broad substrate selectivity, some Alpha-class GSTs show substrate preferences (for compilations of Alpha-class GST substrates, see Hayes and Pulford 1995; Coles and Kadlubar 2005).

CDNB (11, Figure 4) is frequently used as an assay substrate for cytosolic GSTs; significant differences are, however, found in its reaction rates with purified human GSTs (Czerwinski et al. 1996): the reaction rates for GSTA1-2 and GSTA2-2 are much less (<10%) than the rate for GSTA1-1. With recombinant hGSTs and with CDNB (11, Figure 4) as the substrate, the reaction rates for GSTA1-1 and GSTA2-2 were similar but were larger than those found with GSTA3-3 or GSTA4-4 (Johansson and Mannervik 2001).

GSTA3-3 preferentially catalyzes the isomerization of Δ⁵-androstene-3,17-dione (13, Figure 4) to Δ⁴-androstene-3,17-dione (14, Figure 4) (a testosterone precursor) and of Δ⁵-pregnene-3,20-dione to Δ⁴-pregnene-3,20-dione (progesterone) (Benson et al. 1977; Johansson and Mannervik 2001; Raffalli-Mathieu and Mannervik 2005; Dourado et al. 2014). Δ⁵-Androstene-3,17-dione (13, Figure 4) and cumene hydroperoxide (15, Figure 4) have been proposed as class-distinguishing substrates for GSTA (Mannervik et al. 1985). hGSTA3-3-catalyzed steroid isomerase activity is 7- and 1657-folds greater than that of hGSTA1-1 and hGSTA2-2, respectively (Johansson and Mannervik 2002). hGSTA4-4 shows selectivity for 4-hydroxy-2-nonenal (HNE) (17, Figure 4) and related α,β-unsaturated aldehydes (Hubatsch et al. 1998; Balogh and Atkins 2011). Rat GSTA5-5 and the GSTA5 heterodimers GSTA1-5 and GSTA3-5 show high selectivity for aflatoxin B₁-8,9-epoxide (19, Figure 4) (Hayes et al. 1998). The resistance of mice to aflatoxin B1 (19, Figure 4) has been attributed to the high expression of GSTA3-3 (YcYc) in mouse liver (Hayes et al. 1992).

hGSTA catalyzes the reduction of cumene hydroperoxide (15, Figure 4) and phospholipid hydroperoxides (Mannervik et al. 1985; Hurst et al. 1998; Zhao et al. 1999a). Purified human GSTA1-1 selectively catalyzes the conjugation of busulfan (21, Figure 4) (Czerwinski et al. 1996); lower (<10%) activity is also observed with GSTA1-2 and GSTA2-2.

Azathioprine (23, Figure 4) is a widely used immunosuppressant that is a prodrug of 6-mercaptopurine (24, Figure 4); the drug is administered orally and is activated primarily by GST-catalyzed conjugation with GSH (Modén and Mannervik 2014). The reaction involves attack of the GSH thiolate on the imidazole ring to produce stoichiometric quantities of 6-mercaptopurine (24, Figure 4) and the imidazole–GSH conjugate (25, Figure 4) (Eklund et al. 2006). GSTA1-1, GSTM1-1, and GSTA2-2 are the principal GSTs responsible for activation of azathiopurine in human liver and intestine, with GSTA2-2 having more than twice the specific activity and twice the catalytic efficiency (k_cat/K_m) of either GSTA1-1 or GSTM1-1. GSTA2*E is an allelic variant of GSTA2 (GSTA2*A) that has 60–70% of the specific activity of GSTA2*A and a somewhat lower catalytic efficiency than GSTA2*A with CDNB (11, Figure 4) as substrate, but exhibits four-fold greater specific activity and catalytic efficiency than GSTA2*A with azathiopurine (23, Figure 4) as substrate (Ning et al. 2004; Zhang et al. 2010). Pro110 is conserved in all GSTA2 variants except GSTA2*E, which contains Ser110; this residue is near the C-terminus of α-helix 4, where it could potentially function as an H-site residue (Modén and Mannervik 2014).

Melphalan (26, Figure 4), a chemotherapeutic alkylating agent containing a reactive nitrogen mustard moiety, undergoes hGSTA1-1-catalyzed biotransformation to the monoglutatathione S-conjugate (27, Figure 4) and the diglutathione S-conjugate (Dulik et al. 1986; Hall et al. 1994a; Paumi et al. 2001). The glutathionylation reaction takes place between the GSH thiolate and the aziridinium ion intermediate formed from the nitrogen mustard (Bolton et al. 1993). The structurally similar chemotherapeutic agent chlorambucil (28, Figure 4) is a substrate for hGSTA1-1 and 2-2, and is biotransformed to the analogous monoglutathione (29, Figure 4) and diglutathione S-conjugates; monoglutathione conjugate 29 is quantitatively the most significant product (Ciaccio et al. 1990; Meyer et al. 1992). GSTA1-1 catalyzes the conversion of chlorambucil (28, Figure 4) to conjugate 29 (Figure 4) with four-fold greater efficiency than the conversion of mephalan (26, Figure 4) to its monoglutathione S-conjugate (27, Figure 4) (Paumi et al. 2001). The formation of the chlorambucil monoglutathione conjugate (29, Figure 4) results in product inhibition of GSTA1-1, whereas the mephalan conjugate 27 (Figure 4) is a less effective inhibitor of the complex (Meyer et al. 1992; Paumi et al. 2001). The bonding interactions of the monoglutathione conjugate of chlorambucil (29, Figure 4) with the active-site residues of GSTA1-1 were deduced from a crystal structure of the complex (Karpusas et al. 2013). Expression of GSTs has been associated with development of resistance to various anticancer agents. The expression of GSTA1-1 activity in MCF7 and HepG2 cells is insufficient to cause maximum resistance to chlorambucil (28, Figure 4) in the absence of efflux transporters MRP1 (ABCC1) or MRP2 (ABCC2) (Morrow et al. 1998; Paumi et al. 2001; Smitherman et al. 2004). Both transporters efficiently eliminate the chlorambucil monoglutathione conjugate (29, Figure 4) from cells, which prevents the conjugate from inhibiting the catalytic activity of GSTA1-1.

GSTs play a major role in detoxifying bay- and fjord-region diol epoxides, which are the ultimate, DNA-modifying metabolites of polycyclic aromatic hydrocarbon (PAH) carcinogens (Thakker et al. 1985). Several classes of cytosolic GSTs, including hGSTA, hGSTM, and hGSTP, catalyze the reaction of GSH with diol epoxide metabolites of PAHs (Robertson and Jernström 1986; Sundberg et al. 1997). GSTA1-1 catalyzes the conjugation of all PAH-derived diol epoxide isomers studied, although the catalytic efficiencies (V_max/K_m) differ significantly with the absolute configuration of the diol epoxide isomers (Jernström et al. 1992, 1996); the preferred site of attack is at the benzylic oxiranyl carbon with the R configuration. The catalytic efficiencies of hGSTA1-1-catalyzed conjugation of the four isomers of benzo[a]pyrene diol epoxide (BPDE) follow the order: (−)-syn (30, Figure 4) > (+)-syn > (+)-anti > (-)-anti (Jernström et al. 1996). The catalytic efficiencies for conjugation of (+)-anti-BPDE are, however, five-fold higher with hGSTA1-1 than with hGSTA2-2, even though the two isozymes differ by only four residues in the H-site (Singh et al. 2004). Mutation of H-site Ile11 to Ala11 in hGSTA2-2 results in a seven-fold increase in catalytic efficiency, which, according to molecular modeling studies, is attributable to relief of unfavorable steric interactions between the substrate and the bulky Ile11 residue.

The catalytic efficiency of murine GSTA1-1 with (+)-anti-BPDE is significantly higher than that of mGSTA2-2, -3-3, and -4-4, mGSTP1-1, and mGSTM1-1 (Xia et al. 1998); also, murine GSTA1-2 catalyzes the conjugation of (+)-anti-BPDE at rates significantly higher than other murine GSTAs, except GSTA1-1 (Hu et al. 1996; Xia et al. 1999). The mGSTA1 and mGSTA-2 subunits differ by only 10 amino acid residues; by use of chimeric enzymes, molecular modeling, and site-directed mutagenesis, it was shown that the high catalytic activity of mGSTA1-2 and, thus, the even higher activity of GSTA1-1, can be attributed to Met207 and Ile221 in the mGSTA1 subunit.

The overexpression of certain GSTs in cancer cells has led to the design of GSTA-activated prodrugs as potential chemotherapeutic agents. Sulfonamides having the general structure R-HN-SO₂-Ar, where Ar is an electrophilic aromatic ring, can be cleaved by GSTs, including GSTA1-1, to yield RNH₂, SO₂, and GS-Ar (Koeplinger et al. 1999; Zhao et al. 1999b). A sulfonamide-based prodrug of metformin (N¹,N¹-dimethyl-N⁴-(2-nitro-4-(trifluoromethyl)benzenesulfonamide)-bisguanidine) is cleaved by rat liver GSTs and by hGSTA1-1, -M2-2, and -P1-1 (Rautio et al. 2014). van Gisbergen et al. (van Gisbergen et al. 2016) reported that a sulfonamide prodrug in which R is doxorubicin and Ar is 2,4-dinitrobenzene (32, Figure 4) is metabolized to doxorubicin (33, Figure 4) at a much higher rate by GSTA1-1 than by either GSTP1-1 or by rat liver MGST1; the cytotoxicity of the prodrug in cells overexpressing GSTA1-1, however, was substantially lower than that of an analog in which Ar (2-nitro-4-acetylbenzene) was less electrophilic than 2,4-dinitrobenzene and which was converted to free doxorubicin at a much lower rate by GSTA1-1. GSTA1-1-activated, sulfonamide-based prodrugs in which R is an analog of bombesin, a peptide that is recognized and bound by tumor cell receptors, have been prepared for use in targeted drug delivery (Axarli et al. 2009).

The GSTA-catalyzed cleavage of sulfonamides was exploited to develop fluorogenic substrates (Zhang et al. 2011a). The nonfluorescent compounds DNs-AcRh (34, Figure 4), DNs-Coum (36, Figure 4), and DNs-CV (38, Figure 4) release fluorescent derivatives of N-acetylrhodamine (35, Figure 4), 2-(7-amino-4-methyl-2-oxochroman-3-yl)acetic acid (37, Figure 4), and cresyl violet (39, Figure 4), respectively, after the GSTA-catalyzed cleavage of the 2,4-dinitrophenylsulfonyl moiety. The activities of fluorogenic substrates 34, 36, and 38 (Figure 4) were at least twofold higher for GSTA1-1 than for GSTA2-2, -3-3, and -4-4, GSTM1-1 and -2-2, GSTP1-1, GSTT1-1, and MGST1.

Further studies with sulfonamide- and sulfonate ester-based GSTA substrates led to the development of ¹⁹F-NMR and bioluminescent probes (Ito et al. 2012; Shibata et al. 2013; Ito et al. 2014). Benzenesulfonamide 40 (Figure 4) was cleaved by GSTA1-1 at much higher rates than with GSTM1-1, GSTP1-1, or MGST1 to release 7-amino-4-(trifluoromethyl)coumarin (42, Figure 4); the cleavage of benzenesulfonamide 40 (Figure 4) in GSTA1-expressing Escherichia coli cells was demonstrated by ¹⁹F NMR and by fluorescent intensity. 1,2-Dioxetane 43 (Figure 4), which was designed as a bioluminescent probe for GSTA, is cleaved by GSTA to release 1,2-dioxetane 44 (Figure 4); luminescence in E. coli cells transfected with GSTA1 was readily detected.

GSTA shows a strong preference for GSH as the co-substrate. Studies with several GSH analogs and Alpha-class GSTs showed that with rat GSTA1-1 and GSTA2-2 all analogs except α-l-Glu-l-Cys-Gly were inactive (Adang et al. 1988a, 1988b).

2.2.6.1.3. Inhibitors

Indomethacin (Hall et al. 1989), curcumin and analogs (Appiah-Opong et al. 2009), and hypericin (Lu and Atkins 2004) have been reported to inhibit GSTA. Triethyllead chloride inhibits the GSTA-catalyzed conjugation of BSP, both in vitro and in vivo (Byington and Hansbrough 1979). Tributyltin acetate and triphenyltin chloride are potent inhibitors of GSTA class enzymes (Tipping et al. 1979; Warholm et al. 1986).

Increased expression of GSTs is associated with antitumor drug resistance; accordingly, there has been a search for GST inhibitors that might be used clinically to enhance the effectiveness of chemotherapeutic agents (Townsend and Tew 2003; Sau et al. 2010).

GSH analogs have been prepared and tested as GSTA inhibitors (for a review, see Burg and Mulder 2002). A series of GSH analogs based on a l-γ-Glu-D-2-aminoadipic acid backbone, e.g. (R)-5-ethyloxycarbonyl-2-γ-(S)-glutamylamino-N-2-heptylpentamide (45, Figure 5) are selective inhibitors of GSTA1-1 and -2-2, both in vitro and in vivo (Ouwerkerk-Mahadevan et al. 1995; Ouwerkerk-Mahadevan and Mulder 1998). l-γ-Glu-(S-9-fluorenylmethyl)-l-Cys-Gly and its carbamate congener (46, Figure 5), which is resistant to hydrolysis by GGT, inhibit hGSTA1-1 (also GSTM1-1 and GSTP1-1) (Cacciatore et al. 2005). In a series of 2-(pyrrolesulfonylmethyl)-N-arylimines, compound 47 (Figure 5) was the most potent inhibitor of hGSTA1-1 (K_i = 71 μM) (Koutsoumpli et al. 2012).

Figure 5. GSTA inhibitors. 45, (R)-5-ethyloxycarbonyl-2-γ-(S)-glutamylamino-N-2-heptylpentamide; 46, L-γ-(γ-oxa)glutamyl-(S-9-fluorenylmethyl)-l-Cys-Gly; 47, 2-{[(1-methyl-1H-pyrrol-2-yl)sulfonyl]methyl}-N-[(1E)-(4-nitrophenyl)methylene]aniline; 48, 2-hydroxy-4-bromo-2′-hydroxybenzophenone N-acetylhydrazone; 49, 2,2′-dihydroxy-5-phenylbenzophenone; 50, 9-oxo-9H-xanthene-4-carbaldehyde; 51, diaminobis(2-(2,3-dichloro-4-(2-methylenebutanoyl)phenoxy)acetoxy)platinum(VI) chloride (ethacraplatin); 52, ethacrynic acid.

Halogen-substituted 1,4-benzoquinones and 1,4-naphthoquinones were identified as irreversible inhibitors of rat Alpha-class GSTs (Vos et al. 1989). Further studies showed that the GSH S-conjugate of tetrachloro-1,4-benzoquinone (2-S-glutathionyl-3,5,6-trichloro-1,4-benzoquinone) and analogs are active site-directed inhibitors of rat Alpha-class GSTs that covalently modify Cys residues (van Ommen et al. 1988, 1989, 1991).

2,2′-Dihydroxybenzophenones and analogs have been investigated as GSTA inhibitors. 2-Hydroxy-4-bromo-2′-hydroxybenzophenone N-acetylhydrazone (48, Figure 5) is a potent inhibitor (IC₅₀ = 0.18 µM) of hGSTA1-1 (Perperopoulou et al. 2014). Further studies on 2,2′-dihydroxybenzophenones showed that benzophenone 49 (Figure 5) inhibited hGSTA1-1 (IC₅₀ = 1.77 µM) but was a weak inhibitor of hGSTP1A-1A (Pouliou et al. 2015). Both benzophenones 48 and 49 are weak inhibitors of GSTP1-1, although analog 48 (Figure 5) inhibits GSTM1-1 (K_i = 22.3 µM) (Georgakis et al. 2017). Xanthone 50 (Figure 5) inhibits hGSTA1-1 in colon cancer cell lysates and is weakly cytotoxic in Caco-2 cells (Zoi et al. 2013).

Ethacraplatin (51, Figure 5), which represents the GST inhibitor ethacrynic acid (52, Figure 5) tethered to the anticancer drug cisplatin, was designed to deliver ethacrynic acid (52, Figure 5) to inhibit GSTs and a Pt(IV) therapeutic agent (Ang et al. 2005). Ethacraplatin (51, Figure 5) is a potent inhibitor of both GSTA1-1 and GSTP1-1 and is a more potent cytotoxic agent than cisplatin in several cancer cell lines. Mass spectral analysis of GSTA1-1 incubated with ethacraplatin (51, Figure 5) indicated the formation of a GSTA1-1-ethacrynic acid (52, Figure 5) adduct. An analysis of the molecular interactions of ethacraplatin (51, Figure 5) with hGSTP1-1 has been reported (Parker et al. 2011).

2.2.6.1.4. Tissue and species distribution

Alpha-class GSTs are found in most tissues, although the distribution of isozymes varies with the tissue (Davies et al. 1993; Rozell et al. 1993; Sundberg et al. 1993; Forkert et al. 1999; Desmots et al. 2001; Morel et al. 2002; Coles and Kadlubar 2005; Mohana and Achary 2017). The subcellular localization of Alpha-class GSTs differs among isozymes. Activity with CDNB (11, Figure 4) as the substrate was found in cytosolic, microsomal, and mitochondrial fractions of rat liver and was ascribed to Alpha- and Mu-class GSTs (Bhagwat et al. 1998). Later studies confirmed the cytosolic and mitochondrial location of GSTA4 in mouse hepatocytes (Desmots et al. 2001). Other studies reported that GSTA4-4 was found solely in human liver mitochondria (Gardner and Gallagher 2001), but the presence of GSTA1-1 and GSTA4-4 in mouse liver cytosol and mitochondria has also been observed (Raza et al. 2002; Robin et al. 2003). Mitochondrial GSTs are encoded by nuclear DNA, synthesized in the cytosol, and apparently targeted to mitochondria by constitutive signal sequences (Raza 2011). With GSTA4-4, however, phosphorylation and binding to Hsp70 are required for efficient translocation into mitochondria (Robin et al. 2003). Western blot analysis shows that GSTA (Ya–Ya) is present in the matrix and membrane fractions of rat liver mitochondria and catalyzes the reaction of GSH with HNE (17, Figure 4) (Chen et al. 2002). GSTA4-4 is also found in sheep and human liver microsomes (Prabhu et al. 2001, 2004).

2.2.6.2. Mu-class (GSTM)

2.2.6.2.1. Structure and properties

As with other cytosolic GSTs, Mu-class GSTs are homodimeric proteins with molecular masses of approximately 50 kDa. Five human, six rat, and seven mouse Mu-class GSTM isozymes have been identified (Tables 1–3); the genes encoding hGSTM1 to hGSTM5 are located on chromosome band 1p13.

hGSTM1 has been well characterized as having two allelic variants, hGSTM1a and hGSTM1b (Widersten et al. 1991). GSTM2-2 was originally purified from skeletal muscle and the cDNA sequence reported (Board et al. 1988; Vorachek et al. 1991). hGSTM3-3 was identified in testis and brain (Campbell et al. 1990b). hGSTM4-4 is present in several organs, and its specific activity with CDNB (11, Figure 4) is lower than that of other Mu-class GSTs (Comstock et al. 1994).

GSTM1-1, along with GSTA1-1 and GSTP1-1, has an active site Tyr residue. Crystallographic and computational analysis show, however, that the binding interactions of GSH at the G-site of GSTM1-1 differ from that of other cytosolic GSTs (Dourado et al. 2010b). In contrast to GSTA1-1, Trp7 and Trp45 in GSTM1-1 form hydrogen bonds with GSH. Moreover, in GSTM1-1, proton transfer from GSH to His107 to form the GSH thiolate is assisted by two water molecules that serve as a bridge, which contrasts with the proposal that a single water molecule acts as a bridge in the transfer of a proton from the thiol group of GSH to the GSH glutamyl α-carboxylate during GSH activation by GSTA1-1 and GSTP1-1.

The most studied hGSTM1 polymorphism is a gene deletion whose frequency ranges from 50 to 78% among ethnic groups (Moyer et al. 2007). Other studies show functionally significant polymorphisms in hGSTM1 that may be associated with responses to chemotherapy (Moyer et al. 2007; Agúndez and Ladero 2008; Piacentini et al. 2013). The hGSTM1/hGSTT1 double-null genotype is associated with poor treatment outcome in acute myeloid leukemia and with increased risk of drug-induced liver injury (Lucena et al. 2008; Xiao et al. 2014).

Mice lacking the entire Gstm locus (GstmΔ/Δ) were generated by deleting all coding sequences and regulatory elements encoding GSTMs (Xiang et al. 2014). No expression of GSTM1, -2, -3, -4, -5, -6, and -7 was detected in liver, testes, heart, jejunum, and brain of GstmΔ/Δ mice, although decreased conjugation of CDNB (11, Figure 4(A)) was observed in kidney and lung cytosol. A Gstm1^−/− mouse line was also generated; the deletion of Gstm1 failed to alter the expression of the neighboring Gstm2 and Gstm4 genes.

The expression of Mu-class GSTs in rodents is induced by butylated hydroxyanisole, phenobarbital, flavonoids, and dexamethasone (McLellan and Hayes 1989; Bhagwat et al. 1998; Boušová and Skálová 2012). Mice fed an “egg white-based diet,” an “amino acid-based diet,” or lab chow containing 2% flavone for three days showed a marked increase in Ggst3 expression and a lesser increase in Gstm2 and Gstm4 (Rudolf et al. 2008); similarly, total hepatic GSTM activity in mice and rats with bromosulfophthalein as the substrate was significantly increased.

2.2.6.2.2. Substrates

Human GSTM1-1 catalyzes the conjugation of CDNB (11, Figure 4) with GSH at rates higher than those observed with GSTA1-1 or GSTP1-1 (Czerwinski et al. 1996).

Trans-4-Phenyl-3-butene-2-one (53, Figure 6) and bromosulfophthalein have been proposed as class-distinguishing substrates for GSTM (Mannervik et al. 1985). Rat GSTM1-1 and -2-2 catalyze both the addition of GSH to trans-4-phenyl-3-buten-2-one (53, Figure 6) and the elimination of GSH from 4-(glutathion-S-yl)-phenyl-2-butanone (54, Figure 6) (Chen and Armstrong 1995). Purified human GSTM1-1 catalyzes the conjugation of busulfan (21, Figure 4) with GSH but at a lower (∼50%) rate than that found with GSTA1-1 (Czerwinski et al. 1996).

Figure 6. GSTM substrates. 53, trans-4-phenyl-3-buten-2-one; 54, 4-(glutathion-S-yl)-phenyl-2-butanone; 55, 5-cyano-N-[3-[(1R)-1-[(6R)-5,6-dihydro-4-hydroxy-2-oxo-6-(2-phenylethyl)-6-propyl-2H-pyran-3-yl]propyl]phenyl]- 2-pyridinesulfonamide (PNU-109112); 56, (6R)-3-[(1R)-1-(3-aminophenyl)propyl]-5,6-dihydro-4-hydroxy-6-(2-phenylethyl)-6-propyl-2H-pyran-2-one (PNU-143070); 57, 3-cyano-1,2-dimethyl-1-nitrosoguanidine; 58, N-cyano-N′-methylpropanimidamide; 59, S-nitroso-GSH; 60, γ-Glu-α-amino-β-[[2-ethyl-N,N,N′,N′-tetrakis(2-chloroethyl)-phosphorodiamidate]-sulfonyl]propionyl-Gly; 61, tetrakis(2-chloroethyl)phosphorodiamidate; 62, N5-((R)-1-((carboxymethyl)amino)-1-oxo-3-(vinylsulfonyl)propan-2-yl)-l-Gln; 63, 5-(pentafluorobenzoylamino)fluorescein; 64, 2-(6-hydroxy-3-oxo-3H-xanthen-9-yl)-5-(2,3,5,6-tetrafluoro-4-(glutathion-S-yl)benzamido)benzoic acid; 65, 5-(3,4-dinitrobenzoylamino)-2,7-dichlorofluorescein; 66, 5-(3-nitro-4-glutathion-S-ylbenzoylamino)-2,7-dichlorofluorescein; 67, brostallicin; 68, brostallicin-GSH adduct; 69, curcumin; 70, (E)-7-(glutathion-S-yl)-5-hydroxy-1,7-bis(4-hydroxy-3-methoxyphenyl)hept-1-en-3-one; 71, O²-(2,4-dinitrophenyl)-1-[(4-ethoxycarbonyl)piperazin-1-yl]diazen-1-ium-1,2-diolate (JS-K); 72, 1-[(4-ethoxycarbonyl)piperazin-1-yl]diazen-1-ium-1,2-diolate; 73, nitric oxide; 74, ethyl piperazine-1-carboxylate.

Although sulfonamides were formerly considered to be metabolically inert, GSTs catalyze the cleavage of some sulfonamides. HIV-protease inhibitor 56 (Figure 6) is released upon cleavage of the sulfonamide group by reaction of GSH with the prodrug (PNU-109112) (55, Figure 6), which is a substrate for hGSTM1-1 and to a lesser extent hGSTA1-1 and GSTP1-1 (Koeplinger et al. 1999; Zhao et al. 1999b).

hGSTM1-1 and hGSTM2-2 show 84% sequence identity but specific activities with aminochrome and 3-cyano-1,2-dimethyl-1-nitrosoguanidine (57, Figure 6) as substrates are 100-fold greater with hGSTM2-2 than with hGSTM1-1 (Hansson et al. 1999a, 1999b).

hGSTM1a-1a catalyzes the release of the cytotoxic tetrakis(2-chloroethyl)phosporodiamidate (61, Figure 6) from the GSH-based prodrug (60, Figure 6) by a proposed elimination mechanism at a rate that is much faster than that observed with hGSTA1-1 or hGSTP1-1 (Lyttle et al. 1994b). Replacement of Gly in compound 60 (Figure 6) with phenylGly yields an analog that is a poor substrate for hGSTM1a-1a, but is readily cleaved by hGSTA1-1 and hGSTP1-1.

Rat liver GSTM1-1 catalyzes the conjugation of 5-(pentafluorobenzoylamino)fluorescein (63, Figure 6) to GSH S-conjugate 64 (Figure 6) (Arttamangkul et al. 1999); although both the substrate and product are fluorescent, they can be readily separated by HPLC or TLC. The reaction is also catalyzed by hGSTA1-1 and hGSTP1-1 but at rates lower than those by GSTM1-1. 5-(3,4-Dinitrobenzoylamino)-2,7-dichlorofluorescein (65, Figure 6) is a substrate for hGSTM1-1 and also hGSTM2-2, hGSTM3-3, hGSTA2-2, and hGSTP1-1 (Fujikawa et al. 2015).

Brostallicin (67, Figure 6) is an anticancer prodrug that is highly cytotoxic to cells that overexpress GSTM2-2 or GSTP1-1 or both. hGSTM1-1, hGSTA1-1, and hGSTP1-1 catalyze the reaction of GSH with the α-bromoacrylamido group of brostallicin (67, Figure 6) to yield the GSH S-conjugate (68, Figure 6), which is a DNA alkylating agent (Geroni et al. 2002; Pezzola et al. 2010). Brostallicin (67, Figure 6) is a potent inhibitor of GSTM2-2-catalyzed conjugation of CDNB (11, Figure 4) (Pezzola et al. 2010).

hGSTM1a-1a along with hGSTA1-1, hGSTA2-2, and hGSTP1-1, but not hGSTT1-1, catalyzes the reaction of curcumin (69, Figure 6) with GSH to form diastereomeric GSH conjugates (70, Figure 6) (Usta et al. 2007); the adducts are, however, unstable and decompose to give curcumin (69, Figure 6) and other products.

hGSTM1-1 catalyzes the activation of the diazenium diolate prodrug (JS-K, 71, Figure 6) of nitric oxide (73, Figure 6) at a much greater rate than either hGSTA1-1 or hGSTP1-1 (Shami et al. 2003); accordingly, JS-K is cytotoxic in multiple myeloma cell lines and in patient derived multiple myeloma cells (Kiziltepe et al. 2007).

2.2.6.2.3. Inhibitors

Given its putative role in the development of resistance to chemotherapeutic agents, there has been considerable interest in developing GSTM inhibitors to overcome resistance to chemotherapeutic agents (Lo and Ali-Osman 2007). 4-O-Decyl-gabosine D (75, Figure 7) inhibits hGSTM1-1 (K_i^GSH = 3.2 μM) and synergizes cisplatin cytotoxicity (IC₅₀ = 12 μM) in A549 lung cancer cells (Wang et al. 2011b).

Figure 7. GSTM inhibitors. 75, 4-O-decyl-gabosine D; 76, (S)-γ-Glu-(2RS)-(±)-2-amino-(di-n-butoxyphosphinyl)-acetyl-Gly; 77, γ-l-Glu-d-aminoadipic acid; 78, (R)-5-carboxy-2-γ-(S)-glutamylamino-N-2-heptylpentamide; 79, 6-(7-nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX); 68, N-(benzyloxy)-4-(((7-nitrobenzo[c][1,2,5]oxadiazol-4-yl)thio)methyl)benzamide; 80, N-(benzyloxy)-4-(((7-nitrobenzo[c][1,2,5]oxadiazol-4-yl)thio)methyl)benzamide; 81, 4-(glutathion-S-yl)-7-nitrobenzo[c][1,2,5]oxadiazole; 82, S-(7-nitrobenzo[c][1,2,5]oxadiazol-4-yl)-γ-(l-γ-oxaglutamyl)-l-Cys-Gly; 83, 1,5-bis(4-hydroxy-3-methoxyphenyl)-1,4-pentadiene-3-one; 84, 1-chloro-2,4-dinitronaphthalene.

Kunze and coworkers (Kunze and Heps 2000) prepared a series of phosphono analogs of GSH as potential GST inhibitors. One member of the series (76, Figure 7) is a potent inhibitor of hGSTM1-1 (IC₅₀ = 4.7 μM); GSH analog 76 (Figure 7) also inhibited porcine liver GSTA and GSTM as well as porcine lung GSTP1-1. Although compound 76 (Figure 7) was resistant to GGT-catalyzed hydrolysis and was stable in cell media, cellular uptake was poor; the methyl glycinate analog was a much weaker inhibitor of GSTM1-1 (IC₅₀ = 67 μM) but was taken up by cells and underwent intracellular hydrolysis to form 76 (Figure 7). The GSH analog γ-l-Glu-d-aminoadipic acid (77, Figure 7), which is resistant to GGT-catalyzed hydrolysis, potently inhibits rat GSTM1-1- and GSTM2-2-catalyzed reactions (K_i = 8–34 μM) (Adang et al. 1991). Related compounds, e.g. (R)-5-carboxy-2-γ-(S)-glutamylamino-N-2-heptylpentamide (78, Figure 7), are modest inhibitors of rat GSTM1-1 and -2-2 (Ouwerkerk-Mahadevan and Mulder 1998).

6-(7-Nitro-2,1,3-benzoxadiazol-4-ylthio)hexanol (NBDHEX, 79, Figure 7) is a potent inhibitor of hGSTM2-2 (IC₅₀ = ≤0.01 μM), but also inhibits hGSTP1-1 and activates JNK and apoptosis in human osteosarcoma cell lines (Ricci et al. 2005; Sau et al. 2012). The crystal structure of 7-nitrobenzoxadiazole 79 (Figure 7) bound to hGSTM2-2 shows that it forms a σ-complex with GSH and that the complex is stabilized by bonding interactions between the 7-nitro group and H-site Arg residues (Federici et al. 2009). Structure–activity studies aimed at identifying analogs of 7-nitrobenzodiazolyl 79 (Figure 7) with improved pharmacological and toxicological properties yielded additional potent inhibitors of GSTM2-2, such as compound 80 (Figure 7). Analog 80 (IC₅₀ = 0.003 uM) was a 67-fold more potent inhibitor of GSTM2-2 than of GSTP1-1 and a 3-fold more potent inhibitor of GSTM2-2 than compound 79 (Figure 7) (Rotili et al. 2015). The 7-nitrobenzoxadiazolyl (NBD) S-conjugate of GSH (81, Figure 7) and the corresponding γ-oxaglutamyl isostere (82, Figure 7) are more potent inhibitors of hGSTM2-2 compared with hGSTP1-1 (Luisi et al. 2016).

Curcumin analogs inhibit hGSTM1-1, hGSTA1-1 and hGSTP1-1 (Appiah-Opong et al. 2009); of the several compounds studied, compound 83 (Figure 7) is a potent inhibitor of hGSTM1-1 (IC₅₀ = 0.3 μM). Several 2,4-dinitronaphthalene derivatives inhibit hGSTM selectively (Groom et al. 2014): 1-chloro-2,4-dinitronaphthalene (84, Figure 7) is a much more potent inhibitor of hGSTM2-2 than of hGSTM1-1.

2.2.6.2.4. Tissue and species distribution

Mu-class GSTs are widely distributed in human tissues but are absent or weakly expressed in some tissues (Campbell et al. 1991; Awasthi et al. 1994; Sherratt and Hayes 2002; Dhanani and Awasthi 2007). The tissue distribution of GSTM1 to -5 has been summarized (Mohana and Achary 2017).

hGSTM2-2 is highly expressed in skeletal and cardiac muscle and regulates calcium signaling via ryanodine receptors (Abdellatif et al. 2007). A distinct GSTM subclass, human GSTM3 and rodent GSTM5, is selectively expressed in brain and testis (Rowe et al. 1998; Listowsky 2005). Human GSTM5 is selectively expressed in brain, lung, and testes and is not detectable in liver (Takahashi et al. 1993; Rowe et al. 1997); GSTM5 in Cynomolgus macaques, however, is expressed at higher levels in liver and adrenal gland than in testes (Uno et al. 2013).

Immunohistochemical analysis showed that Mu-class GSTs are widely expressed in rat tissues (Otieno et al. 1997). The distribution of GSTM isozymes differs among tissues: in the hamster, GSTM1 is highly expressed in liver but is absent in kidney, GSTM1–GSTM4 are expressed in testis, and GSTM4 is found in kidney (Bogaards et al. 1992). In rat kidney, Mu-class GSTs are strongly expressed in the distal tubules but are also found at other sites in the kidney (Davies et al. 1993; Rozell et al. 1993).

Catalytically active GSTM1-1 has been identified in mouse liver mitochondria (Raza et al. 2002; Sun et al. 2012).

2.2.6.2.5. Other biological functions

GSTM2-2, along with GSTA1-1 and GSTO1-1, are ryanodine receptor (RyR1, -2) modulators (Dulhunty et al. 2011). Ryanodine receptors are intracellular calcium channels; RyR1 is primarily expressed in skeletal muscle, and RyR2 is the principal RyR expressed in cardiac muscle. GSTM2-2 decreases RyR2 activity in cardiac muscle but activates RyR1 in skeletal muscle (Dulhunty et al. 2011). GSTM2-2 exerts mechanistically different effects on RyR in cardiac and skeletal muscle, which led to the proposal that the C- and N-terminal portions of GSTM2-2 interact with RyR2 and RyR1, respectively. The C-terminal domain (GSTM2C) of the GSTM2 subunit, which contains α-helices 4, 5, 6, 7, and 8 (GSTM2-2C H4-8), inhibits cardiac muscle RyR2 but fails to bind to skeletal muscle RyR1 (Liu et al. 2009). Further studies of GSTM2C H5-8 variants showed enhanced efficacy in inhibiting cardiac muscle RyR2 (Samarasinghe et al. 2015).

2.2.6.3. Pi-class (GSTP)

2.2.6.3.1. Structure and properties

As with other cytosolic GSTs, Pi-class GSTs are homodimeric proteins with molecular masses of approximately 50 kDa. One human, one rat, and two mouse Pi-class GSTM isozymes have been identified (Tables 1–3); the gene encoding hGSTP1 is located on chromosome band 11q13. As with other so-called Y-GSTs, GSTP has a Tyr residue in its active site (Atkinson and Babbitt 2009).

GSTP shares many structural features in common with GSTA and GSTM although their amino-acid sequences show little identity (20–32%) (Dirr et al. 1994a, 1994b). Crystal structures of GSTP1-1 in complex with a range of ligands have been summarized (Wu and Dong 2012).

Two amino acid substitutions (I105V/GSTP*B and A114V/GSTP*C) are known to occur in hGSTP1 (Ali-Osman et al. 1997a; Board et al. 1998; Harris et al. 1998); the I105V variant is common across racial groups, whereas the A114V variant is rare. These hGSTP1 allelic variants may determine responsiveness to N,N′N′-triethylenethiophosphoramide (thioTEPA) therapy (Srivastava et al. 1999b).

2.2.6.3.2. Substrates

The substrate selectivity of human, rat, and mouse GSTP has been summarized (Hayes and Pulford 1995; Mohana and Achary 2017); hence, only the salient characteristics of substrates or newer findings will be discussed. To test the hypothesis that relatively few amino acid residues determine substrate selectivity, the Y108V mutant of GSTP was studied (Nuccetelli et al. 1998); interestingly, the substrate selectivity of the GSTP Y108V mutant resembles that of GSTA and shows marked increases in specific activities with 4-chloro-7-nitro-2,1,3-benzoxadiazole (85, Figure 8) and cumene hydroperoxide (15, Figure 4) as substrates. The specific activities for the biotransformation of benzoxodiazole 85 (Figure 8) vary among the cytosolic GSTs (Ricci et al. 1994).

Figure 8. GSTP substrates. 85, 4-chloro-7-nitro-2,1,3-benzoxadiazole (NBD-Cl); 86, 4-(glutathion-S-yl)-7-nitro-2,1,3-benzoxadiazole; 87, O²-[2,4-dinitro-5-(N-methyl-N-4-carboxyphenylamino)phenyl]-1-(N,N-dimethylamino)diazen-1-ium-1,2-diolate (PABA/NO); 88, 2,4-dinitro-5-(glutathion-S-yl)-phenyl 4-methylaminobenzoate; 89, O²-(2,4-dinitrophenyl)diazeniumdiolate-2-(β-d-galactopyranosyl)-oleanolic acid hybrid; 90, 1-(glutathion-S-yl)-2,4-dinitro-5-{[2-(β-d-galactopyranosyl olean-12-en-28-oate-3-yl)-oxy-2-oxoethyl] piperazine-1-yl}benzene; 91, O²-(2,4-dinitrophenyl)diazeniumdiolate-2-PARP-1 inhibitor hybrid; 92, 4-[[3-(piperazine-1-carbonyl)phenyl]methyl]-2H-phthalzain-1-one; 93, 1-[(3,3-dimethyl-2-oxido-1-triazen-1-yl)oxy]-3-(glutathion-S-yl)-4,6-dinitrobenzene; 94, 1,3-(diglutathion-S-yl)-4,6-dinitrobenzene; 95, 6-chloroacetyl-2-dimethylaminonaphthalene (Cadan); 96, 6-(glutathion-S-yl)acetyl-2-dimethylaminonaphthalene; 97, 6-(5-methyl-2-nitrophenylsulfonyl)-luciferin; 98, 3-(glutathion-S-yl)-4-nitrotoluene; 99, luciferin; 100, 5-(3,4-dinitrobenzamido)-2-(6-hydroxy-3-oxo-3H-xanthen-9-yl)benzoic acid (DNAF1); 101, 5-(3-nitro-4-(glutathion-S-yl)benzamido)-2-(6-hydroxy-3-oxo-3H-xanthen-9-yl)benzoic acid.

Chlorambucil (28, Figure 4) is also a substrate for hGSTP1-1 (Ciaccio et al. 1990). The rates of biotransformation of chlorambucil (28, Figure 4) differ among the several allelic variants of hGSTP1-1 (Pandya et al. 2000): for example, the rate of biotransformation is 15-fold higher with the I104/A113 variant than with the V104/V113 variant. The crystal structure of the chlorambucil–hGSTP1-1 complex shows that the substrate binds to the H-site and that the aromatic group forms stacking pi-pi interactions with Phe8 (Parker et al. 2008); moreover, rates of biotransformation of CDNB (11, Figure 4), ethacrynic acid (52, Figure 5) and benzoxadiazole 85 (Figure 8) vary among hGSTP1-1 allelic variants.

Ethacrynic acid (52, Figure 5) has been proposed as a class-distinguishing substrate for GSTP (Mannervik et al. 1985). The k_cat/K_m is the highest for hGSTP-catalyzed conjugation of acrolein compared with the hGSTA- and hGSTM-catalyzed reactions (Berhane and Mannervik 1990); accordingly, mGstP1/P2 null mice show enhanced susceptibility to cyclophosphamide-induced urotoxicity, which is associated with acrolein formation (Conklin et al. 2009).

hGSTP1-1 catalyzes the bioactivation of the prodrug brostallicin (67, Figure 6) (Pezzola et al. 2010). The enzyme catalyzes a Michael addition of GSH to brostallicin to yield an intermediate (68, Figure 6) that alkylates DNA in a sequence-specific manner.

Several diazeniumdiolate-based nitric oxide (73, Figure 6) prodrugs have been prepared as potential cancer chemotherapeutic agents. hGSTP1-1 catalyzes the release of nitric oxide (73, Figure 6) from prodrug 87 (Figure 8) (Findlay et al. 2004; Kim et al. 2015). Nitric oxide prodrug 87 (Figure 8) shows antiproliferative effects in both in vitro and in vivo models (Kogias et al. 2012). An analog of prodrug 87 (Figure 8) that contains a 4-cyano, rather than 4-nitro, substituent is more resistant to non-enzymatic hydrolysis of the ester linkage and undergoes less non-enzymatic activation by GSH than 87 (Figure 8). Because of its antiproliferative and apoptotic potency and its potentially enhanced selectivity for hGSTP1-overexpressing cancer cells, the 4-cyano analog was proposed to be a more promising prodrug than compound 87 (Figure 8) (Kim et al. 2015). The hybrid diazeniumdiolate-oleanolic acid-based nitric oxide prodrug 89 (Figure 8) was prepared to impart hepatic selectivity and stability against non-enzymatic reaction with GSH (Fu et al. 2013). The rate of reaction of prodrug 89 (Figure 8) with GSTP was 7.5-fold greater than with GSTA. Prodrug 89 (Figure 8) released higher levels of nitric oxide (73, Figure 6) in human hepatocarcinoma cell lines than in normal liver cell lines. The hybrid diazeniumdiolate-based nitric oxide (73, Figure 6) prodrug 91 (Figure 8) was designed to release poly-ADP-ribose polymerase 1 (PARP-1) inhibitor 92 (Figure 8) (Maciag et al. 2014); PARP-1 inhibitor 92 (Figure 8) is an analog of the FDA-approved PARP-1 inhibitor olaparib. The design objective was to combine inhibition of PARP-1-dependent DNA repair with the cytotoxic effects of nitric oxide (73, Figure 6). The biotransformation of prodrug 91 (Figure 8) in A549 cells showed that the first step released nitric oxide (73, Figure 6) and PARP-1 inhibitor 92 (Figure 8); reaction of GSH S-conjugate 93 (Figure 8) with GSH gave nitric oxide (73, Figure 6) and diGSH S-conjugate 94 (Figure 8). hGSTP1-1 also catalyzed the release of PARP-1 inhibitor 92 (Figure 8) from prodrug 91 (Figure 8).

Prodrugs of alkylating agents based on a GSH backbone that undergo GSTP-catalyzed activation have been described (Lyttle et al. 1994b; Satyam et al. 1996). For example, the phenylGly analog of prodrug 60 (Figure 6) undergoes a hGSTP1-1-catalyzed reaction to release a nitrogen-mustard alkylating agent (61, Figure 6) and is cytotoxic to MCF-7 breast cancer cells that overexpress hGSTP1-1 (Satyam et al. 1996; Townsend et al. 2002); the reaction is also catalyzed by hGSTA1-1 and hGSTM1-1. GSH S-conjugate 62 (Figure 6), which is also formed upon GST-catalyzed activation of prodrug 61, is proposed to contribute to tumor cell cytotoxicity by inhibition of DNA-dependent protein kinase (Townsend et al. 2002).

Prodrugs of the cytostatic agent doxorubicin (33, Figure 4) have been developed as a strategy to overcome resistance to chemotherapeutic agents (Johansson et al. 2011). 2,4-Dinitrobenzenesulfonyl-doxorubicin (32, Figure 4) and 4-nitrobenzenesulfonyl-doxorubicin are substrates for hGSTP1-1 (and rMGST1). Doxorubicin prodrug 32 (Figure 4) is cytotoxic to V79 cells overexpressing hGSTP1-1.

There has been considerable interest in developing fluorogenic or bioluminescent substrates for cytosolic GSTs. Based on a 6-acetyl-2-dimethylaminonaphthalene scaffold, 6-chloroacetyl-2-dimethylaminonaphthalene (95, Figure 8) was an excellent substrate for GSTP1-1, although activity was also observed with GSTA and GSTM isozymes (Svensson et al. 2002). Luciferin sulfonate ester 97 (Figure 8) is biotransformed by GSTA, GSTM, and GSTP (Zhou et al. 2006); related quinolinyl luciferin derivatives are also substrates for GSTs. Fluorogenic substrates for GSTs that undergo glutathionylation and denitration have recently been reported (Fujikawa et al. 2008); fluorescein derivative 100 (Figure 8) undergoes an hGSTP1-1-catalyzed reaction to yield a fluorescent GSH S-conjugate (101, Figure 8). hGSTA1-1 and hGSTM1-1 also catalyze the reaction but at much lower rates. An analog of fluorescein derivative 100 (Figure 8) in which the carboxylic acid group was replaced with a methyl group was used to visualize GST activity in living HuCCT1 cells. Further studies with fluorescein derivative 100 (Figure 8) showed that in addition to GSTP1-1, the compound was also a substrate for GSTA2-2, GSTM1-1, and GSTM2-2 (Fujikawa et al. 2015). 3′,6′-Diacetyl fluorescein and 3′-methoxy-6′-acetylfluorescein undergo GSTP-catalyzed thiolysis to afford fluorescent products (Fujikawa et al. 2018); the reaction shows remarkable selectivity for hGSTP1-1: scant activity was observed with 12 other cytosolic hGSTs. GST-catalyzed thiolytic cleavage of esters has been previously reported (Keen and Jakoby 1978; Shyam et al. 1999). 3′,6′-Diacetyl fluorescein is also a substrate for porcine liver esterase, indicating that esterase-catalyzed hydrolysis could confound its use in imaging cellular GSTs.

2.2.6.3.3. Inhibitors

GSTP1-1 expression is increased in a range of human tumors (Mannervik et al. 1987; Hayes and Pulford 1995; O'Brien and Tew 1996). The overexpression of hGSTP1-1 is associated with resistance to chemotherapy with anticancer drugs that are substrates for hGSTP1-1 (Kodera et al. 1994; O'Brien and Tew 1996), which has led to a search for hGSTP1-1 inhibitors (Mahajan and Atkins 2005; Johansson et al. 2011).

The development and testing of hGSTP1-1 inhibitors as potential cancer chemotherapeutic agents has been reviewed (Kodera et al. 1994; Morales and Laborde 2007; Laborde 2010); three types of inhibitors have been identified: suicide inhibitors, competitive inhibitors, and hGSTP1-1-activated prodrugs. Given the potential clinical utility of hGSTP1-1 inhibitors, there has been much interest in preparing and testing compounds for their inhibitory properties.

Tributyltin acetate, triethyltin bromide, triphenyltin chloride, Rose Bengal, and Cibacron Blue are potent but nonselective inhibitors of mouse Pi-class enzymes (Warholm et al. 1986). Several 2,2′-dihydroxybenzophenones inhibit GSTP but are in general more potent inhibitors of GSTA (Pouliou et al. 2015). Curcumin (69, Figure 6) and related compounds inhibit GSTP1-1 (and GSTA1-1, GSTPA2-2, GSTM1-1, and GSTM2-2) (van Iersel et al. 1997; Hayeshi et al. 2007); of 34 curcumin analogs prepared as potential inhibitors of GSTP1-1, six were more potent than curcumin (69, Figure 6) (IC₅₀ = 15 µM) with IC₅₀ values of 0.4–1.6 µM (Appiah-Opong et al. 2009). Several compounds from the US Drug Collection (MicroSource Discovery Systems, Inc.), which includes drugs tested in clinical trials, including chlorophyllide, merbromin, ethacrynic acid (52, Figure 5), and hexachlorophene, are effective inhibitors of hGSTP1-1 (IC₅₀ = 2.3–9.7 μM), but are even more potent inhibitors of GSTA3-3 (Musdal et al. 2013). 8-Methoxypsoralen is a competitive inhibitor of hGSTP1-1 (de Oliveira et al. 2014).

Ethacraplatin (51, Figure 5), in which ethacrynic acid (52, Figure 5) is tethered to Pt(IV), inhibits hGSTP1-1 and delivers Pt to cells (Ang et al. 2005). The results of X-ray crystallographic and molecular modeling analyses led to the proposal that ethacraplatin (51, Figure 5) initially forms a complex at the dimer interface of hGSTP1-1 and undergoes reduction resulting in the release of both Pt and the two ethacrynic acid ligands, which diffuse to the H-site of each subunit where they become covalently bound (Parker et al. 2011). Ethacrynic acid tethered to p-cymene ruthenium(II) or osmium(II) complexes inhibits hGSTP1-1 and is cytotoxic to two cancer cell lines (Păunescu et al. 2017).

Several groups of investigators have prepared and tested compounds based on the GSH backbone as GSTP1-1 inhibitors. Moreover, the observations that Alpha-, Mu-, and Pi-class GSTs are subject to product inhibition led to investigation of GSH S-conjugates as GST inhibitors (Meyer 1993; Quesada-Soriano et al. 2012). A cross-docking approach has been used to investigate the binding of GSH S-conjugates to GSTP (Lannutti et al. 2012). Preparation and testing of a panel of GSH analogs as potential isozyme-selective GST inhibitors showed that compound 102 (Figure 9) is a potent inhibitor of hGSTP1-1 (K_i = 0.42 μM) and is 58-fold or 438-fold more potent as an inhibitor of hGSTP1-1 than of hGSTA1-1 and hGSTM1a-1a or hGSTM2-2 (Flatgaard et al. 1993; Lyttle et al. 1994a; Oakley et al. 1997). The S-(4-chlorobenzyl) analog of compound 102 (Figure 9) was a highly selective inhibitor of hGSTP1-1 compared with hGSTPA1-1, hGSTM1a-1a, or hGSTM2-2. Moreover, the diethyl ester (ezatiostat, TLK199) analog of compound 102 (Figure 9) potentiated chlorambucil (28, Figure 4) cytotoxicity in several cell lines, whereas the less lipophilic dicarboxylic acid 102 (Figure 9) caused negligible potentiation (Morgan et al. 1996). Phosphono analogs of GSH, in which the Cys moiety of GSH is replaced by phosphonic acid esters, inhibit the human placental GSTP-catalyzed conjugation of CDNB (11, Figure 4) (Kunze 1996).

Figure 9. GSTP inhibitors. 102, γ-glutamyl-S-(benzyl)-cysteinyl-R-(−)-phenylglycine; 103, piperlongumine (1-[(E)-3-(3,4,5-trimethoxyphenyl)prop-2-enoyl]-2,3-dihydropyridin-6-one); 104, 3-(glutathion-S-yl)-3-(3,4,5-trimethoxyphenyl)propanoic acid; 105, (4-peptidyl- or 4-glutathion-S-yl)piperidin-2-one (R = FGLCSGPADTGR or GSH); 106, 3-cinnamyl-5(E)-bromomethylidenetetrahydro-2-furanone; 107, methyl 3-({2-[3-(4-nitrophenyl)-1,2,4-trioxan-6-yl]prop-2-enoyl}-oxy)benzoate; 108, 5,8-dihydroxy-1-(hydroxymethyl)naphtho[2,3-c]furan-4,9-dione; 109, 6,7-dihydroxy-3-(3′,4′-dihydroxyphenyl)coumarin; 110, 5-[3-chloro-4-(2-methylene-1-oxopropyl)phenoxymethyl]-3-phenyl-1,2,4-oxadiazole; 111, 6-((7 nitrobenzo[c][1,2,5]oxadiazol-4-yl)thio)hexyl benzoate (MC2753); 112, methyl N-(4,6-dichloro-1,3,5-triazin-2-yl)-N-(pent-4-yn-1-yl)-l-leucinate (LAS17); 113, S-glutathionyl ethylenesulfonyl fluoride; 114, phenylarsine oxide (PAO); 115, dinitrosyl-diglutathionyl iron complex (DNDGIC).

Zompra et al. (2016) synthesized twenty-five GSH analogs and investigated their ability to serve as substrates or inhibitors of the hGSTP1-1 allozymes hGSTP1*A (Ile¹⁰⁴/Ala¹¹³), hGSTP1*B (Val¹⁰⁴/Ala¹¹³), and hGSTP1*C (Val¹⁰⁴/Val¹¹³). One analog, l-γ-Glu-l-Cys-Sar, supported the conjugation of CDNB (11, Figure 4) with all allozymes. l-γ-Glu-Thi-Gly inhibited hGSTP1*A and hGSTP1*C but not hGSTP1*B, whereas l-γ-Glu-(S-(acetamidomethyl)-l-Cys-Gly was a better inhibitor of allozymes hGSTP1*B and hGSTP1*C than hGSTP1*A.

Piperlongumine (103, Figure 9) is a natural product that interacts with GSH-binding proteins, including GSTP1-1 (Adams et al. 2012). The interaction with GSH-binding proteins was first attributed to reaction of the C2–C3 olefin of the piperidin-2-one moiety with GSH followed by covalent reaction of the C7–C8 olefin of the GSH adduct with the proteins (Adams et al. 2012). Subsequent studies failed, however, to show adduction of GSTP1-1 (Harshbarger et al. 2017). Rather studies on the mechanism of inhibition of GSTP1-1 indicate that piperlongumine (103, Figure 9) is a prodrug and that GSH S-conjugate 104 (Figure 9), formed by a reaction sequence that has not been unequivocally elucidated, binds to the H-site and, thereby, inhibits GSTP1-1 (Harshbarger et al. 2017).

Haloenol lactone 106 (Figure 9) is a highly selective inactivator of mouse liver GSTP and shows only minor inactivation of mGSTA or mGSTM (Zheng et al. 1996). Haloenol lactone 106 inactivates hGSTP1-1 by alkylation of Cys47 (Mitchell et al. 1998; Zheng et al. 2005). Several 1,2,4-trioxanes were prepared as potential inhibitors of hGSTP1-1 (Bräutigam et al. 2015); trioxane 107 (Figure 9) is a potent inhibitor of human GSTP1-1 (IC₅₀ = 1.65–2.90 μM) but a weak inhibitor of hGSTA1-1 and hGSTM2-2 (IC₅₀ = 74–98 μM).

Evaluation of a panel of natural plant products as inhibitors of recombinant hGSTP1-1 showed that isofuranonaphthoquinone 108 (Figure 9) was the most potent (Mukanganyama et al. 2011). 3-Arylcoumarin analogs, e.g. 109 (Figure 9) are strong inhibitors of human placental GSTP1-1, but their isozyme selectivity was not determined (Alparslan and Daniş 2015).

Oxadiazole analogs of ethacrynic acid, e.g. 110 (Figure 9) inhibit hGSTP1-1 and exhibit antiproliferative activity (Yang et al. 2010); interestingly, one analog was a poor inhibitor of hGSTP1-1 but still showed potent antiproliferative activity. Nitrobenzoxadiazole 79 (Figure 7) inhibits GSTP1-1 (IC₅₀ = 0.80 μM) and activates JNK and induces apoptosis in human osteosarcoma cell lines (Ricci et al. 2005; Sau et al. 2012); preparation and testing of a range of nitrobenzoxadiazole derivatives showed that all were more potent inhibitors of GSTM2-2 than of GSTP1-1 (De Luca et al. 2015; Rotili et al. 2015). Nitrobenzoxadiazole 111 (Figure 9), the benzoate ester of 79 (Figure 7), partially inhibits GSTP1-1, perhaps because of negative cooperativity (Fulci et al. 2017); benzoate ester 111 (Figure 9) undergoes little non-enzymatic reaction with GSH, and docking results indicate that it binds near the dimer interface of GSTP1-1. Benzoate ester 111 (Figure 9) undergoes rapid microsomal carboxylesterase-catalyzed hydrolysis, which limits its utility as a drug candidate (Di Paolo et al. 2019). Compared with benzoate ester 111 (Figure 9), the benzamide analog is hydrolytically stable, shows improved stability in the presence of GSH, inhibits GSTP1-1, disrupts the interaction between GSTP1-1 and TRAF2, and is cytotoxic in A375 cells.

Tyrosine-directed, irreversible inhibitors of GSTP have been reported. Leucinate ester 112 (Figure 9) inhibits hGSTP1-1 by modifying selectively Tyr108 (Crawford and Weerapana 2016). Covalent modification is attributed to the reaction of the dichlorotriazine moiety with Tyr108; no reaction was observed with the Y108F mutant. Proteomic analysis of HeLa cells incubated with ester 112 (Figure 9) confirmed the selective labeling of Tyr108 in hGSTP1-1. Binding of sulfonyl fluoride 113 (Figure 9) to the G-site of hGSTP1-1 positions an electrophilic group adjacent to the active site Tyr-108 (Shishido et al. 2017); the formation of a sulfonate ester of Tyr108 in hGSTP1-1 was confirmed by crystallographic analysis. The IC₅₀ values for the inhibition of hGSTP1-1, hGSTM2-2, and hGSTA1-1 are 12, 50, and 12 μM, respectively.

Phenylarsine oxide (114, Figure 9) reacts rapidly with GSH to form a diGSH complex that is a potent inhibitor of GSTP1-1 (K_i = 90 nM) (Parker et al. 2017). In the absence of GSH, phenylarsine oxide (114, Figure 9) forms an intermonomer crosslink between Cys47 and Cys101; accordingly, phenylarsine oxide fails to inhibit the C47S/C101S mutant. The sequestration of phenylarsine oxide (114, Figure 9) by GSTP1-1 may represent a general detoxification and inactivation mechanism for arsenic-containing drugs and environmental agents.

Negative cooperativity in hGSTP1-1 is observed with dinitrosyl-diglutathionyl iron complex (DNDGIC, 115, Figure 9), which is a potent inhibitor (K_i = <1.5 nM) of hGSTP1-1 (Cesareo et al. 2005; Bocedi et al. 2016). In the reaction of DNDGIC (115, Figure 9) with hGSTP1-1, the active-site Tyr7 displaces one of the GSH moieties to bind the complex. The sequestration of nitric oxide (73, Figure 6) by iron and GSH to give DNDGIC (115, Figure 9) followed by its reaction with GSTP1-1 may protect against nitrosylative and oxidative stress, but DNDGIC (115, Figure 9) inactivates GSH reductase, which is important in maintaining the redox state of the cell. It has been proposed that negative cooperativity serves to protect cells by maintaining residual GST activity in the presence of DNDGIC (115, Figure 9) to protect against toxins (Bocedi et al. 2013); negative cooperativity is much more prominent in Tyr- and Ser-GSTs than in Cys-GSTs.

2.2.6.3.4. Tissue and species distribution

Although GSTP is widely distributed in human tissues, it is noteworthy that expression is low in hepatocytes (Sundberg et al. 1993; Awasthi et al. 1994; Dhanani and Awasthi 2007; Mohana and Achary 2017). Indeed, immunohistochemical studies showed that hepatocytes were unstained by antibodies to GSTP but that staining was observed in the bile duct (Campbell et al. 1991); hGSTP was, however, detected in most other organs, although the intensity varied widely. GSTP is highly expressed in human placenta (Guthenberg et al. 1979; Zusterzeel et al. 1999). As with hGSTP, rat GSTP is widely expressed but its expression is low in liver (Meyer et al. 1985; Hayes and Mantle 1986; Pemble et al. 1986; Davies et al. 1993; Rozell et al. 1993). GSTP is highly expressed in mouse kidney liver and urinary bladder (Conklin et al. 2009). GSTP1-1 is highly expressed in tumor tissue (Peters et al. 1989; Tsuchida et al. 1989; Tew 1994; Jardim et al. 2012; Tew 2016).

The expression of GSTP in initiated hepatocytes after treatment with carcinogens is a biomarker of hepatocarcinogenesis (Pitot et al. 1996). GSTP1-1 is the predominant GST found in many cell lines, including those used in the National Cancer Institute Drug Screening Program (Batist et al. 1986; Tew et al. 1996). The expression of GSTP mRNA in human tissue-derived cell lines varies widely (Bauer et al. 2006): GSTP mRNA is abundant in several lung cell lines but is absent in the HepG2 cell line. In addition to the enhanced expression of genes encoding hGSTP1, epigenetic factors also influence hGSTP1 gene expression (Zhang et al. 2014b).

GSTP is also present in nuclei and mitochondria in mammalian cell lines and tissues (Hall et al. 1994b; Ali-Osman et al. 1997b; Soh et al. 2005; Allen et al. 2007; Goto et al. 2009). The nuclear localization of hGSTP is significant because its expression is associated with drug resistance and poor treatment outcomes (Huang et al. 2003; Soh et al. 2005; Allen et al. 2007; Rolland et al. 2010).

2.2.6.3.5. Other biological functions

Although a thorough treatment of the many biological functions of GSTP, particularly those involved in cancer drug resistance, is beyond the scope of this discussion, several excellent reviews are available (Tew 1994; Hayes and Pulford 1995; Lage 2008; Tew et al. 2011; Vasieva 2011; Tew and Townsend 2012; Zhang et al. 2014a; Singh 2015; Tew 2016). For example, GSTP regulates JNK signaling (Adler et al. 1999; Sau et al. 2012) and controls the S-glutathionylation of a range of proteins (Townsend et al. 2009; Tew et al. 2011). GSTP also regulates estrogen response element (ERE)-dependent estrogen receptor alpha (ERα) signaling events (Liu et al. 2014b) and is an inhibitor of TRAF2 (De Luca et al. 2014).

2.2.6.4. Theta-class (GSTT)

Three Theta-class GSTs have been identified (Landi 2000; Coggan et al. 2002) (Tables 1–3): GSTT1-1, GSTT2-2, and GSTT3-3. Theta-class GSTs are considered to be the progenitor of other GST classes and to be the most ancient of the GST classes (Pemble and Taylor 1992; Landi 2000). Theta-class GSTs are widely distributed in nature in contrast to Alpha-, Mu-, and Pi-class GSTs, which are found only in animals and fungi. Theta-class GSTs are characterized by their low activity with CDNB (11, Figure 4) as the substrate and by their failure to bind to immobilized GSH affinity matrices (Meyer et al. 1991; Hussey and Hayes 1992).

2.2.6.4.1. Structure and properties

Theta-class GSTs are homodimeric proteins with subunit molecular masses of approximately 27 kDa (Meyer et al. 1991; Tan and Board 1996). In contrast to Alpha-, Mu-, Pi-, and Sigma-class GSTs, in which Tyr is a catalytically critical active-site residue, Ser is the corresponding active-site residue in Theta-class GSTs (Wilce et al. 1995). Accordingly, site-directed mutation of the active-site Ser yields an inactive enzyme (Board et al. 1995; Caccuri et al. 1997).

Hiratsuka et al. (1990) and Ogura et al. (1991) described a rat liver GST that showed little amino-acid sequence homology (∼20%) to Alpha-, Mu-, and Pi-class GSTs, that showed high activity with reactive sulfate esters, and that was termed Yrs–Yrs. A protein purified from rat liver that showed high GSH peroxidase activity and activity with alkyl epoxides and alkyl halides was designated GST E (Meyer et al. 1984). Two proteins purified from rat liver that were designated GST 5–5 and GST 12–12 and one from human liver that was designated as human liver GST Theta (Meyer et al. 1991) were identified as Theta-class GSTs; subsequently, rat liver GST 5–5 and GST 12–12 were termed GSTT1-1 and GSTT2-2, respectively. GST Yrs–Yrs was further characterized as belonging to the Theta-class GSTs and was termed GSTT2-2 (Hussey and Hayes 1992; Tan et al. 1995). hGSTT1 and hGSTT2 subunits share approximately 54% amino-acid sequence identity (Flanagan et al. 1998). Genes encoding a third Theta-class GST, Gstt3, have been identified in mice and rats (Coggan et al. 2002) (Rat Genome Database 1562732) (Tables 2 and 3).

The crystal structures of hGSTT1-1 and hGSTT2-2 have been solved (McKinstry et al. 1998; Rossjohn et al. 1998; Tars et al. 2006). The crystal structures of hGSTT1-1 and -2-2 are similar (Tars et al. 2006); both show the canonical thioredoxin fold, but major differences are found in the C-terminus and in the loop connecting helices α4 and α5. In hGSTT1-1, the bulky side-chain of H-site residue Trp234 in C-terminus helix α9 hampers binding to both the H- and G-sites (Shokeer et al. 2005; Tars et al. 2006); in GSTT2-2 a 40-residue C-terminal extension, which does not prevent catalysis, blocks access to both the H- and G-sites, and a sulfate-binding site was identified (Rossjohn et al. 1998; Flanagan et al. 1999).

The substrate selectivities of hGSTT1-1 and -2-2 differ (Jemth and Mannervik 1997); H-site residues are poorly conserved between hGSTT1-1 and -2-2, which contributes substantially to the differences in substrate selectivity (Rossjohn et al. 1998; Tars et al. 2006). Preferred substrates for GSTT1-1 include halogenated hydrocarbons, e.g. dichloromethane (120, Figure 10) and 4-nitrobenzyl chloride (125, Figure 10), but the specific activities of the human enzyme are far lower than that of those of the rodent enzyme (Jemth and Mannervik 1997). Mutational studies that compared murine and hGSTT1-1 indicated that residue 234 (Arg and Trp in the wild-type murine and human proteins, respectively) is associated with the observed differences in specific activities (Larsson et al. 2004; Shokeer et al. 2005; Shokeer and Mannervik 2010a, 2010b); the finding that the specific activity of the W234R mutant of hGSTT1-1 with 4-nitrobenzyl chloride (125, Figure 10) as the substrate is approximately 50-fold higher than the wild-type enzyme highlights the importance of residue Trp234 (Shokeer et al. 2005). In addition, mutation of Met232 in the C-terminal helix of mGSTT1-1 l alters the specific activities for several substrates (Shokeer and Mannervik 2010a); with the M232A mutant, the specific activities with EPNP (116, Figure 10) or 4-nitrobenzyl chloride (125, Figure 10) as substrates are decreased by approximately 70% whereas the specific activity with cumene hydroperoxide (15, Figure 4) as the substrate is increased 30-fold.

Figure 10. GSTT substrates. 116, 1,2-epoxy-3-(4-nitrophenoxy)propane (EPNP); 117, S-(2-hydroxy-3-[4-nitrophenoxy]propyl)GSH; 118, phenethyl isothiocyanate; 119, S-(N-phenethylthiocarbamoyl)GSH; 120, dichloromethane; 121, S-(chloromethyl)GSH; 122, formaldehyde; 123, 1-menaphthyl sulfate; 124, S-(1-menaphthyl)GSH; 125, 4-nitrobenzyl chloride; 126, S-(4-nitrobenzyl)GSH; 127, 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU); 128, 1-(2[-glutathion-S-yl]ethyl-3-(2-chloroethyl)urea; 129, methyl parathion; 130, O-methyl O-(4-nitrophenyl) phosphorothioate; 131, S-(methyl)GSH.

In contrast to hGSTT1-1, hGSTT2-2 shows activity with cumene hydroperoxide (15, Figure 4) and 1-menaphthyl sulfate (123, Figure 10) but no detectable activity with EPNP (116, Figure 10) or dichloromethane (120, Figure 10) as substrates (Hussey and Hayes 1992; Tan and Board 1996). hGSTT2-2 catalyzes the biotransformation of 1-menaphthyl sulfate (123, Figure 10) in the absence of GSH but in the presence of S-(methyl)GSH plus 2-mercaptoethanol or L-Cys (Tan et al. 1996). Moreover, the S11A mutant shows increased specific activity with 1-menaphthyl sulfate (123, Figure 10) as the substrate (Tan et al. 1996). A sulfatase reaction mechanism was proposed in which the binding of GSH to the G-site favors the entry of the second substrate, i.e. 1-menaphthyl sulfate (123, Figure 10), which loses sulfate to form the 1-menaphthyl carbonium ion that reacts with GSH to give S-(1-menaphthyl)GSH (123, Figure 10) (Tan et al. 1996). The proposal that GSTT2-2 acts as a sulfatase rather than a GST has been challenged based on data obtained with rGSTT2-2, which exhibits highly similar residue identity with hGSTT2-2 in the sulfate-binding site (Jemth and Mannervik 2000; Moyer et al. 2007; Agúndez and Ladero 2008). The authors found that Ser11 stabilizes the GST thiolate: the lower k_cat values found for the S11A mutant of rGSTT2-2 with substrates 123 and 125 (Figure 10) compared with the wild-type enzyme are attributed to slower product release.

hGSTT1 is highly polymorphic (Thorn et al. 2012). The genes encoding hGSTT1 and hGST2 are located on band 22q11.2 (Tan et al. 1995; Webb et al. 1996). GSTT activity is found in human erythrocytes (Peter et al. 1989; Schröder et al. 1992); with methyl chloride as the substrate, approximately 40% of human subjects fail to conjugate methyl chloride with GSH and are termed “non-conjugators.” The non-conjugator phenotype is attributed to deletion of GSTT1 gene and the conjugator phenotype and GSTT1 genotype are comparable (Pemble et al. 1994). The gene-deletion genotype varies widely among ethnic groups and amounts to a low of 33.5% in Caucasian-Americans to a high of 73.5% in Han Chinese-Americans (Moyer et al. 2007; Agúndez and Ladero 2008).

Although null alleles have been the focus of most studies on GSTT polymorphisms, single-nucleotide polymorphisms (SNPs) are also known. Given the potential for GSTT-dependent bioactivation of carcinogens, the observed SNPs in GSTT1 merit investigation (Landi 2000; Agúndez and Ladero 2008). SNPs in GSTT1 affect the catalytic activity of GSTT1-1 by altering the stability and physical properties of the expressed protein (Josephy et al. 2009, 2011). Investigation of the genetic variants of GSTT1 and GSTT2 in rhesus and cynomolgus macaques identified a range of alleles (Uno et al. 2018). Activities of the D125G and R76C/D125G variants of GSTT1 were much lower than the wild-type enzyme or the R76C variant with 1-iodohexane and dibromomethane as substrates.

As with other GSTs, Theta-class GSTs are induced by a range of agents (for a review, see Higgins and Hayes 2011). In rats fed NSAIDs, including indomethacin, sulindac, ibuprofen, piroxicam, and aspirin, the expression of colonic GSTT1-1 is increased (van Lieshout et al. 1998). Among several potential chemopreventive agents studied, indole-3-carbinol is the most potent inducer of hepatic GSTT1-1 in male rats (Sherratt et al. 1998); in female rats, however, coumarin is the most potent inducer of hepatic GSTT1-1. In mice, treatment with clofibrate, 1,4-bis[2-(3,5-dichloropyridyloxy)]benzene, and phenobarbital induced expression of Gstt1, whereas clofibrate and ciprofibrate induced expression of Gstt2 (Knight et al. 2008); of 15 microsomal enzyme inducers studied, none induced expression of Gstt3. Apple extracts rich in polyphenols prevent DNA damage induced by hydroperoxide 15 (Figure 4) in colon cell lines by enhancing GSTT2-2 promoter activity (Miene et al. 2009; Petermann et al. 2009); the cytoprotective effects of the extracts may be attributed to flavan-3-ols, including procyanidins, and flavonols.

2.2.6.4.2. Substrates

The substrate selectivity of Theta-class GSTs differs among species and among the identified isoenzymes (Meyer et al. 1984; Thier et al. 1998; Landi 2000). As indicated above, although the crystal structures of hGSTT1-1 and hGSTT2-2 are similar, their substrate selectivities differ. For example, hGSTT1-1 shows high activity with 1,2-epoxy-3-(4-nitrophenoxy)propane (116, Figure 10) and phenethyl isothiocyanate (118, Figure 10) as substrates, but little or no activity with dichloromethane (120, Figure 10) or 1-menaphthyl sulfate (123, Figure 10) as substrates (Jemth and Mannervik 1997). In contrast, hGSTT2-2 shows high activity with cumene hydroperoxide (15, Figure 4) and 1-menaphthyl sulfate (123, Figure 10) as substrates (Meyer et al. 1991; Jemth and Mannervik 1997). Rat GSTT1-1 shows high activity with dichloromethane (120, Figure 10), 4-nitrobenzyl chloride (125, Figure 10), or cumene hydroperoxide (15, Figure 4) as substrates (Meyer et al. 1991). hGSTT1-1, hGST2-2, and erythrocytic hGSTT1-1 show scant or no activity with CDNB (11, Figure 4) as the substrate (Meyer et al. 1991; Hussey and Hayes 1992; Schröder et al. 1996). In contrast to GSTT1-1 expressed in solid tissues, human erythrocytic GSTT1-1 fails to accept EPNP (116, Figure 10) and 4-nitrobenzyl chloride (125, Figure 10) as substrates (Schröder et al. 1996).

hGSTT1-1 is an efficient catalyst for the denitrosation of BCNU (127, Figure 10) (Lien et al. 2002); little activity was observed with hGSTM2-2 and -3-3, and no activity was found with GSTA1-1, -2-2, and -4-4; GSTM1-1, -4-4, and 5-5; GSTP1-1; and rGSTT2-2. hGSTT1-1 is the major catalyst for the O-demethylation of methyl parathion (129, Figure 10), although lower activity was observed with hGSTA1-1, rGSTA5-5, and mGSTA3-3 (Abel et al. 2004). CDNB (11, Figure 4) is not a substrate for hGSTT2-2 and is a poor substrate for rGSTT1-1, but is a substrate for rGSTT2-2 and mGSTT3-3 (Coggan et al. 2002). Murine GSTT1-1 shows high activity with dichloromethane (120, Figure 10), EPNP (116, Figure 10) and cumene hydroperoxide (15, Figure 4), low activity with CDNB (11, Figure 4) and 1-menaphthyl sulfate (123, Figure 10), and no activity with ethacrynic acid (52, Figure 5) (Whittington et al. 1999; Coggan et al. 2002). With dichloromethane (120, Figure 10) as the substrate, the catalytic efficiency (k_cat/K_m) of murine GSTT1-1 is five-fold greater than that of human GSTT1-1 (Sherratt et al. 2002). In the biotransformation of dibromomethane to formaldehyde (122, Figure 10), GSH is required, but not consumed (Ahmed and Anders 1978).

GSTT2-2 isoenzymes show preference for reactive sulfate esters, e.g. 1-menaphthyl sulfate (123, Figure 10) and 5-sulfoxymethylchrysene, for lipid-derived alkenals, and for cumene hydroperoxide (15, Figure 4) (Hiratsuka et al. 1990, 1994; Tan and Board 1996). Rat and human GSTT2-2 show different substrate selectivities (Coggan et al. 2002): CDNB (11, Figure 4) is a substrate for rGSTT2-2 but not for hGSTT2-2, whereas ethacrynic acid (52, Figure 5) is a substrate for hGSTT2-2 but not for rGSTT2-2. Murine GSTT3-3 shows high selectivity for 4-nitrobenzyl chloride (125, Figure 10) and, similar to mGSTT1-1 and hGSTT2-2, exhibits low specific activity with 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (85, Figure 8); unlike mGSTT1-1 and hGSTT2-2, mGSTT3-3 fails to catalyze the conjugation of 1-menaphthyl sulfate (123, Figure 10) (Coggan et al. 2002).

As expected, GSTT1-1 catalyzed reactions are decreased in Gstt1-null mice (Fujimoto et al. 2007); with dichloromethane (120, Figure 10) as the substrate, activity in liver and kidney cytosols from Gstt1-null mice was only 2.2% of that observed in wild-type mice, whereas with EPNP (116, Figure 10) or BCNU (127, Figure 10) as substrates, activity in Gstt1-null mice ranged from 8.7% to 42.1% or 13.2% to 23.9%, respectively, of wild-type controls. Similarly, the biotransformation of dichloromethane (120, Figure 10) in human liver samples with the GSTT1-null genotype is significantly reduced (Arakawa et al. 2012).

2.2.6.4.3. Inhibitors

Apart from the weak inhibition of tissue and erythrocyte GSTT1-1 by GSH S-conjugates, no ground-state or active-site directed inhibitors of Theta-class GSTs have apparently been reported (Ong and Clark 1986; Meyer 1993; Schröder et al. 1996). Sulfate and S-(1-menaphthyl)GSH (124, Figure 10) inhibit the GSTT2-2-catalyzed biotransformation of 1-menaphthyl sulfate (123, Figure 10) (Gillham 1973).

2.2.6.4.4. Tissue and species distribution

The tissue-specific distribution of hGSTT2-2 has been reported (Fagerberg et al. 2014; Mohana and Achary 2017).

Immunoblot analysis shows that hGSTT1-1 is highly expressed in erythrocytes and liver and is also expressed in lung, kidney, brain, skeletal muscle, heart, small intestine, and spleen, but not in lymphocytes (Juronen et al. 1996); N-terminal amino-acid sequence analysis and similarities in immunochemical properties indicate that liver and erythrocyte GSTT1-1 are identical. Immunohistochemical analysis shows that only trace amounts of GSTT1-1 are present in human lung (Sherratt et al. 1997). Western analysis shows that hepatic expression of GSTT1-1 is greater in the mouse than in the rat (Mainwaring et al. 1998); in the lung, mGSTT1-1 is expressed in bronchiolar epithelium and in type-II alveolar cells.

With EPNP (116, Figure 10) as the substrate, GSTT activities are high in rat epididymis and adrenal glands; intermediate in liver, testis, epididymal fluid, and seminal vesicles, and not detectable in mammary gland and skeletal muscle (Meyer et al. 1984).

In the mouse, Northern-blot analysis shows that all Theta-class GSTs are highly expressed in the liver and to a lesser extent in the kidney (Coggan et al. 2002); mGSTT3, but neither mGSTT1 nor mGSTT2, expression is observed in the testes. Immunohistochemical analysis and in situ hybridization showed differences in distribution patterns and enzyme activities of GSTT1-1 and GSTT2-2 in liver and lung tissues of the mouse, rat, and human; the human tissue levels are lower than those of either the mouse or rat (Mainwaring et al. 1996).

The low expression of GSTT1-1 in human and rat lung tissue compared with mouse lung tissue, as well as the comparatively low catalytic efficiency of hGSTT1-1 with dichloromethane (120, Figure 10) as substrate, indicates little bioactivation of dichloromethane in human lung tissue (Sherratt et al. 1997, 2002).

A protein isolated from rat liver mitochondria that shows N-terminal sequence identity with Theta-class GSTs and a possible mitochondrial import signal was designated GST 13-13 (Harris et al. 1991). There apparently have been no other reports of the characterization of GSTT 13-13, although “13-13” has been used as a designation for GSTT3-3 (Table 3).

2.2.6.5. Zeta-class (GSTZ/MAAI)

2.2.6.5.1. Structure and properties

GST Zeta (GSTZ) was identified with a BLAST search of the human expressed sequence tag (EST) database with the amino-terminal sequence of a GST-like protein identified in carnation petals (Board et al. 1997, 2001). As with GSTT (Meyer et al. 1991), GSTZ binds weakly to GSH-agarose affinity columns, which may have prevented its earlier discovery (Board et al. 1997). GSTZ is identical with maleylacetoacetate isomerase (MAAI), which catalyzes the penultimate step in the tyrosine degradation pathway (Fernández-Cañón and Peñalva 1998). Several reviews about Zeta-class GSTs and MAAI are available (Seltzer 1989a, 1989b; Anders et al. 2001; Board and Anders 2005, 2011).

Tong et al. (1998a) purified a protein from rat liver that catalyzed the biotransformation of dichloroacetic acid (DCA) (132, Figure 11) to glyoxylic acid (133, Figure 11).

Figure 11. GSTZ substrates. 132, dichloroacetic acid (DCA); 133, glyoxylic acid; 134, maleylacetone; 135, fumarylacetone; 136, maleylacetoacetic acid (MAA); 137, fumarylacetoacetic acid; 138, (±)-2-bromo-3-(4-nitrophenyl)propanoic acid; 139, 2-(glutathion-S-yl)-3-(4-nitrophenyl)propanoic acid; 140, maleylpyruvic acid; 141, fumarylpyruvic acid; 142, cis-β-acetylacrylic acid ((Z)-4-oxopent-2-enoic acid); 143, (E)-4-oxopent-2-enoic acid.

Portions of the sequence of the purified protein matched that of GSTZ1-1 reported by Board et al. (1997), and antibodies raised against hGSTZ1-1 reacted with the rat liver protein. Accordingly, the GST that catalyzes the biotransformation of DCA was identified as a Zeta-class GST.

hGSTZ1 encodes a 216-residue peptide with a molecular mass of 24 212 Da (Board et al. 1997). The gene encoding hGSTZ1 maps to a single locus on chromosome 14q24.3 (Blackburn et al. 1998).

Although hGSTZ1-1 fails to catalyze the conjugation of typical GST substrates, its crystal structure indicates that it is a member of the GST superfamily (Polekhina et al. 2001; Boone et al. 2014). hGSTZ1-1 shows the N-terminal canonical thioredoxin fold characteristic of other cytosolic GSTs and a C-terminal all α-helical domain. The G-site is found in a crevice between the domains, which may explain the low affinity of GSTZ1-1 for GSH-affinity columns. The dimer interface is dominated by hydrophobic interactions. α-Haloacids, but not α-haloacid esters and amides or β-haloacids, are substrates for GSTZ1-1 (Tong et al. 1998b); the crystal structure of hGSTZ1-1 indicates that the interaction of α-haloacids with Arg175 positions them for attack by GSH, whereas binding of β-haloacids fails to orient them for attack by GSH.

Kinetic studies with hGSTZ1c-1c indicate that the active-site Cys16 is not a catalytic residue, although Cys16 is has an important role in binding of both GSH and electrophilic co-substrates (Board et al. 2003; Ricci et al. 2004). The enzyme also shows half-site reactivity: binding of GSH to one subunit effectively silences the second subunit as it is unable to bind and activate GSH (Ricci et al. 2004).

Four hGSTZ1 allelic variants have been identified (Blackburn et al. 1998, 2000, 2001; Board and Anders 2005); these are designated hGSTZ1a-1a, hGSTZ1b-1b, hGSTZ1c-1c, and hGSTZ1d-1d. (The four allelic variants have also been termed KRT, KGT, EGT, and EGM, respectively (Stacpoole et al. 2008; Shroads et al. 2012).) The frequency of hGSTZ1 variants in a normal European population is: hGSTZ1a, 0.086; hGSTZ1b, 0.285; hGSTZ1c, 0.473; hGSTZ1d, 0.156 (Blackburn et al. 2001); similar distributions have been reported in other populations (Shroads et al. 2012). Although 10 SNPs have been identified upstream of the GSTZ1 start of transcription, only two were associated with changes in promoter activity (Fang et al. 2006).

The relationship between hGSTZ1 allelic variants and rates of biotransformation has been investigated. The specific activities for biotransformation of DCA (132, Figure 11) and (R)-2-chloropropionic acid are highest with hGSTZ1a-1a, whereas the rates of biotransformation of fluoroacetic acid and (S)-2-chloropropionic acid are highest with hGSTZ1b-1b, -1c-1c, and 1d-1d (Blackburn et al. 2000, 2001). The rate constants (k_inact) for the GSH-dependent, DCA-induced inactivation of hGSTZ1-1 variants are similar for hGSTZ1b-1b, -1c-1c, and -1d-1d (∼7 × 10⁻²min⁻¹), which are higher than that of hGSTZ1a-1a (3 × 10⁻²min⁻¹) (Tzeng et al. 2000). With recombinant hGSTZ1-1, the rate constants (k_cat/K_m) for the biotransformation of maleylacetone (134, Figure 11) followed the order: 1d-1d = 1 b-1b > 1c-1c > 1a-1a, whereas the rate constants for the biotransformation of chlorofluoroacetic acid were similar among the four allelic variants (Lantum et al. 2002b). In children given DCA (132, Figure 11) for management of congenital mitochondrial disorders, the plasma drug t_½ was shorter in hGSTZ1c-1c (EGT, see above) carriers compared with hGSTZ1c-1c noncarriers (Shroads et al. 2015).

2.2.6.5.2. Substrates

As indicated above, GSTZ1-1 fails to catalyze the conjugation of typical GST substrates or catalyzes the reactions at low rates (Board et al. 1997). It was later shown that substrates for GSTZ1-1 include DCA (132, Figure 11) and a range of α-haloalkanoic acids as well as maleylacetone (134, Figure 11) (Tong et al. 1998b; Lantum et al. 2002b). hGSTZ1-1 catalyzes the reduction of organic hydroperoxides, e.g. tert-butyl hydroperoxide, and the conjugation of ethacrynic acid (52, Figure 5) and NBD-Cl (85, Figure 8), albeit at relatively low rates, but shows no activity with CDNB (11, Figure 4) and several other known GST substrates (Board et al. 1997). The rate of hGSTZ1-1 or rat GSTZ1-1-catalzyed biotransformation of DCA is, however, much greater than that of tert-butyl hydroperoxide (Tong et al. 1998a).

GSTZ catalyzes the biotransformation of DCA (132, Figure 11) to glyoxylic acid (133, Figure 11) (Tong et al. 1998a, 1998b); in this reaction, GSH is required but not consumed. In addition to DCA, GSTZ catalyzes the biotransformation of fluoroacetic acid, bromochoro-, bromofluoro-, chlorofluoro, and dibromoacetic acid, 2-bromo-, 2-chloro-, and 2-iodopropionic acid, and 2,2-dichloropropionic acid, but not difluoroacetic acid, 2-fluoropropionic acid, or 3,3-dichloropropionic acid (Tong et al. 1998b).

GSTs have been implicated in the defluorination of fluoroacetic acid (Soiefer and Kostyniak 1983, 1984; Wang et al. 1986; Tecle and Casida 1989). Tu et al. (2005) showed, however, that at least two enzymes catalyze the defluorination of fluoroacetic acid and that rat liver GSTZ1-1 is a minor catalyst. Other studies showed that recombinant hGSTZ1a-1a, -1 b-1b, -1c-1c, and 1d-1d all catalyze the defluorination of fluoroacetic acid but at rates that are much lower than those for the biotransformation of DCA (132, Figure 11) (Blackburn et al. 2000, 2001; Tu et al. 2006).

The selectivity of GSTZ for α-haloalkanoic acids was exploited to develop an assay substrate for GSTZ (Board et al. 2003): (±)-2-bromo-3-(4-nitrophenyl)propanoic acid (138, Figure 11) undergoes a GSTZ1-1-catalyzed reaction to give 2-(glutathion-S-yl)-3-(4-nitrophenyl)propanoic acid (139, Figure 11), which shows a spectral shift that allows quantification of reaction rates.

The isomerization of MAA (136, Figure 11) to fumarylacetoacetic acid (137, Figure 11) is catalyzed by GSH-requiring hepatic enzymes (Edwards and Knox 1956). Although the unstable MAA can be prepared by the homogentistate dioxygenase-catalyzed oxidation of homogentistic acid, the surrogate substrates maleylacetone (134, Figure 11), maleylpyruvic acid (140, Figure 11), and cis-β-acetylacrylic acid (142, Figure 11), which are biotransformed to the corresponding E-isomers, are most frequently used (Lack 1961; Fowler and Seltzer 1970; Lee and Seltzer 1989; Lantum et al. 2002b). The consensus mechanism of the GSTZ-catalyzed isomerization of MAA (136, Figure 11) involves the Michael addition of GSH to give a dienediolate intermediate, which allows rotation around the newly formed single bond, followed by elimination of GSH to form the trans-isomer, i.e. furmarylacetoacetic acid (137, Figure 11) (Seltzer and Lin 1979; Deponte 2013). Hence, GSH is required, but not consumed, in the isomerization reaction.

GSTZ is not detectable in Gstz1^−/− mice and accordingly the biotransformation of chlorofluoroacetic acid and maleylacetone (134, Figure 11) is not detectable (Lim et al. 2004); GSTA-, GSTM-, and GSTP-class GSTs are, however, overexpressed in female Gstz^−/− mice.

2.2.6.5.3. Inhibitors

No ground-state, competitive inhibitors of GSTZ1-1 have apparently been described. DCA (88, Figure 10) is, however, an irreversible inhibitor of GSTZ1-1: in human subjects given DCA, the t_½ is prolonged after repeated dosing (Curry et al. 1985, 1991). Hepatic levels of GSTZ1-1 are decreased in rats given DCA (Anderson et al. 1999; Guo et al. 2006).

Kinetic studies showed that DCA (88, Figure 10) inactivated GSTZ in rat liver cytosol (Anderson et al. 1999); moreover, in rats given DCA, losses in enzyme activity and in the amount of immunoreactive GSTZ1-1 are observed. From these data, the turnover of GSTZ1-1 was found to be 0.21⁻¹/day, which amounts to a t_½ of 3.3 days. In addition to DCA, bromochloroacetic acid, dibromoacetic acid, and 2,2-dichloropropionic acid also inactivate GSTZ1-1, whereas fluorine-containing dihaloacetic acids and 2-chloropropionic acid fail to inactivate GSTZ1-1. These data, along with studies with [²H]dichloroacetic acid and [²H]chlorofluoroacetic acid (Wempe et al. 1999), allowed the proposal of a mechanism for the DCA-induced inactivation of GSTZ1-1: the formation of the S-(α-chlorocarboxymethyl)GSH intermediate results in the covalent modification of a critical residue in GSTZ1-1. Indeed, mass-spectral analysis of hGSTZ1c-1c inactivated by DCA showed that Cys16 is covalently modified by the S-(α-chlorocarboxymethyl)GSH intermediate (Anderson et al. 2002). The partition ratio for the DCA inactivation of hGSTZ1c-1c was 5.7 × 10², and the k_cat for the biotransformation of DCA was 39 min⁻¹.

2.2.6.5.4. Tissue and species distribution

hGSTZ1-1 is most intensely expressed in liver, but is also found in kidney, skeletal muscle, and brain (Board et al. 1997). In mice, GSTZ-catalyzed biotransformation of DCA (88, Figure 10) is highest in liver followed by lung and kidney (Lipscomb et al. 1995). Immunohistochemical studies show that GSTZ1-1 is present in most rat tissues but that expression is highest in liver, testes, and prostate (Lantum et al. 2002a; Mohana and Achary 2017). GSTZ1-1 is also found in human and rat liver mitochondrial matrix and, as with cytosolic GSTZ1-1, is inactivated by DCA (Li et al. 2011).

2.2.6.5.5. Other biological functions

Maleylacetoacetic acid (MAA) (136, Figure 11), fumarylacetoacetic acid (137, Figure 11), and succinylacetoacetic acid are electrophilic metabolites of tyrosine, and their formation contributes to the deleterious symptoms of hereditary tyrosinemia type 1 (Lindblad et al. 1977). Gstz^−/− mice given Phe show hepatic necrosis, macrovesicular steatosis, splenic atrophy, decreased total white blood cell counts, and lowered hepatic GSH concentrations as well as increased serum succinylacetone concentrations, indicating that GSTZ1-1 plays a critical role in detoxifying Phe metabolites (Lim et al. 2004; Blackburn et al. 2006); treatment of mice with DCA (132, Figure 11), which inactivates GSTZ1-1, followed by feeding of Phe does not, however, result in decreased white blood cell counts (Theodoratos et al. 2009). The hepatotoxicity of acetaminophen is increased in Gstz^−/− mice, perhaps because of the lowered hepatic GSH concentrations seen in Gstz^-/- mice (Blackburn et al. 2006).

C/EBPα-dependent induction of GSTZ1-1 mRNA and protein is observed during adipogenesis in 3T3-L1 preadipocytes, indicating a role for GSTZ1-1 in adipocytes (Qiang and Farmer 2006).

Associations between hGSTZ1 polymorphisms and a range of human disorders have been investigated. In human subjects, no overall association between Parkinson’s Disease and hGSTZ1 polymorphisms was observed (Taylor et al. 2001); the Z1*C allele, which is the most common variant in Caucasian populations, was, however, less common in Parkinsonian subjects when stratified according to pesticide exposure. A SNP (rs2111699 located in intron 1 of GSTZ1) is associated with longevity in long-living subjects in Calabria, Italy (Di Cianni et al. 2013), but no association between hGSTZ1 polymorphisms and age-related macular degeneration was found (Othman et al. 2012). No association between hGSTZ1 allelic variants and susceptibility to breast cancer or to schizophrenia has been found (Nafissi et al. 2011; Saadat et al. 2012).

2.2.6.6. Omega-class (GSTO)

Omega-class GSTs are distinguished from most other cytosolic GSTs in that they fail to catalyze canonical conjugative reactions but catalyze reductive reactions and regulate cell-signaling pathways. Reviews about GSTO are available (Whitbread et al. 2003, 2005; Board 2011; Board and Menon 2016; Xie et al. 2018).

2.2.6.6.1. Structure and properties

Omega-class GSTs were identified by a BLAST search of the human EST database with hGSTT and hGSTZ sequences as query sequences (Board et al. 2000). These studies identified a human cDNA that was expressed, characterized, and designated as hGSTO1-1. Further studies on the genomic organization of GSTO1 revealed a second Omega-class gene, GSTO2 (Whitbread et al. 2003). The genes encoding hGSTO1 and hGSTO2 are located on chromosome band 10q25.1 (Board et al. 2000; Whitbread et al. 2003; Wang et al. 2005). Rat and mouse orthologues of hGsto2 have been described (Tables 2 and 3).

Omega-class GSTs differ from other cytosolic GSTs in that they have a Cys residue in the active-site rather than a Tyr or a Ser residue (Board et al. 2000; Board 2011). Moreover, human Omega-class GSTs show only about 20% amino-acid sequence identity with other cytosolic GSTs. hGSTO2-2 shows 64% amino-acid sequence identity with hGSTO1-1 (Whitbread et al. 2003, 2005).

The crystal structures of hGSTO1-1 and -2-2 complexed with substrates and inhibitors have been solved (Board 2011; Zhou et al. 2012). The thioredoxin fold and C-terminal domain with two helices are present in both hGSTO1-1 and -2-2, and both show an extended N-terminus loop. Although the G-sites are similar in hGSTO1-1 and -2-2, the H-site is more occluded in hGSTO2-2 due to substitution of Lys128 for Gly128.

Before the Omega-class GSTs were fully characterized, four proteins were described that were later shown to be Omega-class GSTs: S-phenacylglutathione reductase (Kitada et al. 1985); dehydroascorbic acid (144, Figure 12) reductase (Maellaro et al. 1994; Ishikawa et al. 1998); “p28,” a 28-kDa stress response protein (Kodym et al. 1999); and MMA^V (146, Figure 12) reductase (Zakharyan and Aposhian 1999).

Figure 12. GSTO substrates. 144, dehydroascorbic acid; 145, ascorbic acid; 146, methylarsonic acid (MMA^V); 147, methylarsonous acid (MMA^III); 148, dimethylarsinic acid (DMA^V); 149, dimethylarsinous acid (DMA^III); 150, 2,4-dichlorophenacyl chloride; 151, 2,4-dichloroacetophenone; 152, S-(4-nitrophenacyl)GSH; 153, 4-nitroacetophenone; 154, 2-hydroxyethane disulfide; 155, 2-hydroxyethanethiol.

hGSTO1 and hGSTO2 are polymorphic (Mukherjee et al. 2006; Board 2011). In hGSTO1, the E155/E208 variant predominates in Australian-European and Chinese population (Whitbread et al. 2003; Schmuck et al. 2005). Interestingly, thioltransferase, dehydroascorbic acid reductase, methylarsonic acid reductase, and dimethylarsinic acid reductase activities of GSTO1-1 are increased in the A140/ΔE155/E208 and A140/ΔE155/K208 allelic variants compared with similar variants that retain E155; an exception is the A140/E155/K208 variant, which exhibits dimethylarsinic acid reductase activity similar to the corresponding ΔE155 variant (Whitbread et al. 2003; Schmuck et al. 2005). The E155 deletion in hGSTO1-1 causes not only an increase in specific activity but also a substantial decrease in stability (Whitbread et al. 2003; Schmuck et al. 2005, 2008).

The association between hGSTO1 and hGSTO2 polymorphisms has been investigated and attempts to identify at-risk populations and differences in arsenic metabolism in arsenic-exposed populations have been made (Marnell et al. 2003; Tanaka-Kagawa et al. 2003; Yu et al. 2003; Schmuck et al. 2005; Whitbread et al. 2005; Chowdhury et al. 2006; Mukherjee et al. 2006; Paiva et al. 2008; Schmuck et al. 2008; Escobar-García et al. 2012). For example, in arsenic-exposed subjects in the Mexican state of Coahuila, the A140D, but not the E208K, polymorphism is associated with increased excretion of arsenic metabolites (Escobar-García et al. 2012); in addition, the A140D and E208K polymorphisms are associated with expression of genes encoding IL-8, Apaf-1, and TGF-β.

hGSTO1 polymorphisms have been associated, however inconsistently, with age-of-onset of Alzheimer’s disease and Parkinson’s disease (Kolsch et al. 2004; Whitbread et al. 2004; Li et al. 2006; Allen et al. 2012; Piacentini et al. 2012).

Of 15 microsomal enzyme inducers studied in mice, only butylated hydroxylanisole significantly increased the expression of Gsto1 RNA (Knight et al. 2008). The expression of GSTO1-1 is increased or decreased in a range of treatments or disorders, including cancer, cataract, neurodegenerative disease, pulmonary disease, and inflammation (Board and Menon 2016).

2.2.6.6.2. Substrates

GSTO catalyzes several reactions: S-phenacylglutathione reductase, thioltransferase, deglutathionylation, dehydroascorbic acid reductase, α-haloketone reductase, methylarsonic acid reductase, and dimethylarsinic acid reductase (Board et al. 2000; Zakharyan et al. 2001; Board and Anders 2007; Board et al. 2008; Board 2011; Board and Menon 2016). GSTO1-1 shows low activity with CDNB (11, Figure 4) and little or no activity with many canonical GST substrates (Board et al. 2000). hGSTO2-2 shows high dehydroascorbic acid (144, Figure 12) reductase activity compared with hGSTO1-1 (Schmuck et al. 2005; Zhou et al. 2012). In contrast to hGSTO1-1, recombinant hGSTO2-2 has been reported to show high activity with CDNB (11, Figure 4) as the substrate (Wang et al. 2005); the dehydroascorbic acid (144, Figure 12) reductase and thioltransferase activities for hGSTO2-2 reported by Wang et al. (2005), are much lower than those reported in other studies (Board et al. 2000; Menon and Board 2013).

2,4-Dichlorophenacyl chloride (150, Figure 12), a metabolite of chlorfenvinphos, undergoes GSH-dependent biotransformation to 2,4-dichloroacetophenone (151, Figure 12) (Hutson et al. 1976; Brundin et al. 1982; Kitada et al. 1985); the intermediate GSH S-conjugate S-(2,4-dichlorophenacyl)GSH, which is formed non-enzymatically, rather than α-haloketone 150 (Figure 12), is the actual substrate for the enzyme, and several thiols, e.g. GSH, 2-mercaptoethanol, L-Cys, cysteamine, can serve as second substrates. The reduction of phenacyl halides is catalyzed by hGSTO1-1, but not by hGSTO2-2 (Board and Anders 2007).

The finding that GSTO1-1 catalyzes the GSH-dependent reduction of phenacyl halides led to the development of a specific assay substrate for hGSTO1-1 (Board and Anders 2007; Board et al. 2008). hGSTO1-1 catalyzes the reduction of S-(4-nitrophenacyl)GSH (152, Figure 12) in the presence of 2-mercaptoethanol to 4-nitroacetophenone (153, Figure 12) (Board et al. 2008); the reaction is selective for hGSTO1-1: little or no reaction was observed with hGSTO2-2, hGSTP1-1, hGSTA1-1, hGSTA2-2, hGSTA4-4, hGSTM1-1, hGSTM2-2, hGSTM4-4, hGSTT1-1, hGSTT2-2, or hGSTZ1-1. Németi et al. (Németi et al. 2015b) modified the assay to increase its sensitivity by using HPLC to quantify 4-nitroacetophenone (153, Figure 12) formation and by the use of tris(2-carboxyethyl)phosphine (TCEP) as the reductant.

GSTO plays a critical role in the reduction of weakly toxic pentavalent arsenic to highly toxic trivalent arsenic (Watanabe and Hirano 2013). The reduction of methylarsonic acid (MMA^V, 146, Figure 12) to methylarsonous acid (MMA^III, 147, Figure 12) is catalyzed by MMA^V (146, Figure 12) reductase, which requires GSH as a second substrate (Zakharyan and Aposhian 1999). Later studies showed that MMA^V (146, Figure 12) reductase is identical with hGSTO1-1 (Zakharyan et al. 2001) and that both hGSTO1-1 and hGSTO2-2 catalyze the reduction of methylarsonic acid (146, Figure 12), as well as the reduction of dimethylarsinic acid (148, Figure 12), but that rates of reduction of dimethylarsinic acid (148, Figure 12) found with hGSTO2-2 are four-fold lower than those found with hGSTO1-1 (Schmuck et al. 2005). Investigation of the reduction of dimethylarsinic acid (148, Figure 12) in vivo in rats and in rat liver cytosol showed that GSH is required, but, on the basis of purification studies, the authors concluded that GSTO1-1 was not involved (Németi and Gregus 2013). Further investigation showed, however, that reduction of dimethylarsinic acid (146, Figure 12) by rat liver cytosol was inhibited by the GSTO1-1-selective substrate S-(4-nitrophenacyl)GSH (152, Figure 12) (Németi et al. 2015a); additional purification studies confirmed the role of GSTO1-1 in the reduction of dimethylarsinic acid (148, Figure 12).

In studies with Gsto1-null mice, the hepatic GSTO1-1-catalyzed reduction of methylarsonic acid (146, Figure 12) amounted to 20% that of wild-type mice, indicating that pathways in addition to GSTO may be involved (Chowdhury et al. 2006); in Gsto1-null mice given sodium arsenate, high concentrations of methylarsonous acid (147, Figure 12) were found in the kidneys. (The reduction of arsenate to arsinite is catalyzed by a range of phosphorolytic-arsenolytic enzymes, including purine nucleoside phosphorylase-a, glyceraldehyde-3-phosphate dehydrogenase, and glycogen phosphorylase (Yu et al. 2003; Németi and Gregus 2009).)

Rates for the thioltransferase reaction with 2-hydroxyethyl disulfide (154, Figure 12) as the substrate are similar with both hGSTO1-1- and hGSTO2-2 (Schmuck et al. 2005). The glutathionylation and deglutathionylation of proteins is a redox detoxification process and also acts as a regulatory switch (Board and Menon 2016). hGSTO1-1, but not hGSTO2-2, catalyzes the deglutathionylation of the glutathionylated synthetic peptide SQLWC^-SGLSN (Menon and Board 2013); expression of hGSTO1-1 in T47-D cells, which show reduced GSTO1-1 activities, decreased the total cellular glutathionylated protein concentration.

Early studies showed the presence of a GSH-dependent dehydroascorbic acid reductase in a range of rat tissues (Maellaro et al. 1994; Paolicchi et al. 1996; Ishikawa et al. 1998), an activity now attributed to GSTO. Both hGSTO1-1 and hGSTO2-2 catalyze the reduction of dehydroascorbic acid (144, Figure 12) to ascorbic acid (145, Figure 12), but rates with hGSTO2-2 are approximately100-fold higher than those with hGSTO1-1 (Schmuck et al. 2005) or with glutaredoxin-dependent dehydroascorbic acid reductase (Wells et al. 1990; Park and Levine 1996). Interestingly, the crystal structure of hGSTO1-1-C32S mutant in complex with ascorbic acid (145, Figure 12) shows that ascorbic acid is bound to the G-site (Zhou et al. 2012); this is the first report of a GST substrate or reaction product that does not contain GSH as a structural component being bound in the G-site of a GST.

2.2.6.6.3. Inhibitors

Given the putative role of Omega-class GSTs in arsenic bioactivation, cancer, and neurodegeneration, there has been considerable interest in identifying GSTO inhibitors (Board and Menon 2016). Xie et al. (2018) have reviewed the development of small-molecule inhibitors as an example of hit discovery for difficult targets, e.g. GSTO1. The role of GSTO1-1 in lipopolysaccharide-stimulated inflammation and ROS production also indicates that inhibitors may find therapeutic uses (Menon et al. 2014, 2015; Board and Menon 2016).

The cytokine release inhibitory drug CRID 2 (156, Figure 13) covalently binds to Cys32 in the active site of hGSTO1-1 and irreversibly inhibits ATP-induced IL-1β posttranslational processing (Laliberte et al. 2003); no labeling was observed with a C32A mutant and the labeling of hGSTO1-1 by CRID 2 is blocked by a range of S-substituted GSH derivatives, including S-(4-nitrobenzyl)GSH (152, Figure 12), S-(1-adamantyl)GSH, and S-octylGSH. Moreover, a GSH adduct of CRID 2 inhibits ATP-induced IL-1β posttranslational processing. The epoxyketone moiety-containing proteasome inhibitor carfilzomib covalently reacts with Cys32 and Cys192 of hGSTO1 (Federspiel et al. 2016); although carfilzomib inhibits GSTO1-1 at micromolar concentrations, the inhibition of GSTO1-1 is expected to be minimal at therapeutic doses of carbfilzomib.

Figure 13. GSTO inhibitors. 156, 1-(1,2,3,5,6,7-hexahydro-s-indacen-4-yl)-3-[2-fluoro-5-oxiranylbenzenesulfonyl]urea (CRID 2); 157, N-[3-[(2-chloroacetyl)(4-nitrophenyl)amino]propyl]-2,2,2-trifluoroacetamide (ML175); 158, 2-chloro-N-(2-(cyclohexylamino)-2-oxo-1-(pyridin-2-yl)ethyl)-N-(3-fluorophenyl)acetamide (KT53); 159, 2-chloro-N-(4-chloro-3-(N,N-dimethylsulfamoyl)phenyl)acetamide (C1-27); 160, 9-[2-carboxy-4-[[[2-[[oxo-10-[(phenylsulfonyl)oxy]decyl]amino]ethyl]amino]carbonyl]phenyl]-3,6-bis(dimethylamino)xanthylium; 161, N-[1-oxo-3-[(phenylsulfonyl)oxy]propyl]-l-isoleucylglycyl-4,4,5,5-tetradehydro-l-norvalinamide (NJP2); 162, 1-(4-amino-6-chlorohex-1,5-dione)-11,12-didehydro-5,6-dihydro-dibenz[b,f]azocine; 163, 5-chloromethylfluorescein diacetate; 164, 5-bromomethylfluorescein.

Fluorescence-based activity-based protein profiling (Fluopol-ABPP) was used to identify enzyme targets (Bachovchin et al. 2009); with this methodology, GSTO1 was found to be a target for the thiol-reactive compounds omeprazole, dioxiram, and an α-chloroacetamide. These studies were extended to focus on GSTO1-1 inhibitors and showed that α-chloroacetamide 157 (Figure 13) inhibits (IC₅₀ = 28 nM) hGSTO1-1 and covalently labels the active-site Cys32 (Tsuboi et al. 2010). Additional studies identified α-chloroacetamide 158 (Figure 13) as a potent and selective inactivator of hGSTO1-1 in vitro (IC₅₀ = 21 nM) and of hGSTO1-1 in MDA-MB-435 breast cancer cells (IC₅₀ = 35 nM) (Tsuboi et al. 2011); moreover, α-chloroacetamide 158 (Figure 13) enhanced the cytotoxicity of cisplatin in MDA-MB-435 cells.

Screening of α-chloro-N-phenylacetamide derivatives identified α-chloroacetamide 159 (Figure 13) as a potent (IC₅₀ = 31 nM), selective, and irreversible inhibitor of GSTO1-1 (Ramkumar et al. 2016). α-Chloroacetamide 159 (Figure 13), which covalently labels Cys32 in the active site of GSTO1-1, was cytotoxic at low µM concentrations to a panel of cancer cell lines, increased the cells’ sensitivity to cisplatin, and suppressed growth of a human colon cancer xenograft in vivo. Four-hour treatment of HCT116 cells with α-chloroacetamide 159 (Figure 13) causes induction of several stress response-associated genes; 24-h treatment resulted in additional transcriptional changes, many of which also occur in HCT116 GSTO1-1 knockdown cells. These studies were extended to include a panel of acrylamide-based GSTO inhibitors (Dai et al. 2019). One compound, N-((3-methylisoxazol-5-yl)methyl)-N-(4-(2-hydroxyphenyl)thiazole-2-yl)acrylamide, is a potent inhibitor of GSTO1-1 (IC₅₀ = 0.22 nM). Its selectivity for GSTO1-1 versus GSTO2-2 and other GSTs was not reported.

In addition to serving as inhibitors, compounds that target Cys32 of GSTO1 have been designed for activity-based protein profiling (ABPP), which allows quantification of protein concentrations. Although inhibition of GSTO1-1 was not explicitly reported in all experiments, adduction of Cys32 would be expected to inhibit its catalytic activity. Rhodamine-tagged phenylsulfonate ester 160 (Figure 13) targets the active-site Cys32 of hGSTO1-1 in estrogen-receptor negative cell lines (Adam et al. 2002, 2004). A library of peptides bearing electrophilic sulfonate ester or acrylamide groups targeted to Cys residues was synthesized to search for apoptotic-cell-selective inhibitors (Pace et al. 2012). Sulfonate ester 161 (Figure 13) selectively inhibits hGSTO1-1 in apoptotic cells compared with control cells. The selective internalization of sulfonate ester 161 (Figure 13) in apoptotic cells was demonstrated by use of click chemistry to attach a rhodamine moiety to the acetylene bond in sulfonate ester 161 (Figure 13) to afford a fluorescent analog; the selectivity for apoptotic cells was attributed to the increased permeability of apoptotic cell membranes.

Li et al. (2019) used the ABPP strategy to search for cellular targets of piperlongumine (103, Figure 9). With an alkyne-substituted piperlongumine analog and click-chemistry-based ABPP, GSTO1 was identified as a likely cellular target of piperlongumine (103, Figure 9); the C2-C3 olefin of piperlongumine was required for cytotoxicity.

Incubation of GSTO1-1 with dibenzazocine 162 (Figure 13) and MALDI-TOF-MS analysis demonstrated the adduction of GSTO1-1, presumably at Cys32 (Liang et al. 2015). Inductively coupled plasma mass spectrometry (ICP-MS) and tagging of the adduct by Cu-free click chemistry allowed quantification of the cellular content of GSTO1-1, which amounted to 2.4 × 10⁻²ng/cell in C7721 and MCF-7 tumor cell lines, as well as visualization of GSTO1-1 in cells. The methodology also was used to quantify the increase in GSTO1-1 content after incubation of cells with cisplatin.

A series of chemically reactive fluorescein derivatives, including benzyl halides 163 and 164 (Figure 13), were screened for their abilities to label proteins in NIH/3T3 cells (Son et al. 2010). The cell-permeable benzyl chloride-derived fluorescein diacetate (163, Figure 13), which covalently alkylates Cys32, irreversibly and selectively inhibited (IC₅₀ = 51 nM) hGSTO1-1 in HEK293 cells and allowed quantification of catalytically active hGSTO1-1 in intact NIH/3T3 cells and HEK293 cells by fluorescence intensity.

hGSTO1-1 has been proposed for use as a “self-catalytic” tag for study of membrane proteins (Lee et al. 2011). A GSTO1-1-epidermal growth factor receptor (EGFR) fusion protein was expressed in EGFR-negative HEK293 cells. Treatment of the cells with bromomethyl fluorescein 164 (Figure 13), which unlike the diacetylated fluorescein 163 (Figure 13) does not penetrate cells, allowed the visualization of EGFR both on the cell membrane and its subsequent endocytosis upon activation by an EGF ligand. No fluorescence was detected when cells expressing an EGFR fusion protein containing C32A hGSTO1-1 were treated with fluorescein 164 (Figure 13).

Studies with 4-aminopiperidine-based probes with Cys-reactive moieties identified hGSTO1 in lysates from HEK293T cells overexpressing hGSTO1 (Couvertier and Weerapana 2014). The probes also labeled hGSTO1 in intact HEK293T cells expressing hGSTO1.

In contrast to the ABPP approach discussed above to identify hit targets, the “Inverse Drug Discovery” approach uses small molecule probes with latent electrophilic reactivity to identify enzyme and non-enzyme proteins (Mortenson et al. 2018). GSTO1 was identified in HEK293T cell lysates as a target for a fluorosulfate-substituted quinoline-based probe; the Tyr229 residue, rather than the Cys32 residue, was modified by the probe. Although inhibition experiments were not reported, the bound fluorosulfate probe is located in the small-molecule binding pocket of GSTO1, which would presumably result in inhibition.

(+)-α-Tocopherol succinate inhibits the hGSTO1-1-catalyzed reduction of methylarsonic acid and dimethylarsinic acid (Sampayo-Reyes and Zakharyan 2006). In addition, GSTO1-1 is inhibited by carnosic acid and rifampicin (Board and Menon 2016).

2.2.6.6.4. Tissue and species distribution

Northern blot and immunohistochemical analyses show that GSTO1 is widely expressed in mammalian tissues but that expression is highest in liver, heart, and skeletal muscle (Paolicchi et al. 1996; Board et al. 2000; Rouimi et al. 2001; Yin et al. 2001; Whitbread et al. 2003, 2005; Mohana and Achary 2017). Northern-blot analysis showed that hGSTO2 is highly expressed in testes, prostate, and pancreas (Whitbread et al. 2003; Wang et al. 2005); further analysis by RT-PCR revealed detectable expression in most tissues, except skeletal muscle, and high expression in testes and pancreas. The expression of hGSTO2 was examined in seven tumor-derived cell lines but was detectable only in prostatic adenocarcinoma cells (Wang et al. 2005); hGSTO2-2 is found in both cytoplasm and nucleus.

In Gsto1-null mice, expression of methylarsonic acid reductase activity was reduced in all tissues studied, whereas in wild-type mice methylarsonic acid reductase activity was the highest in urinary bladder tissue (Chowdhury et al. 2006).

2.2.6.6.5. Other biological functions

In addition to its role in xenobiotic metabolism, GSTO has been implicated in several biological processes (Oakley 2005; Board and Menon 2013, 2016).

Chloride intracellular channel proteins (CLIC) are soluble and membrane-bound proteins that function as chloride-selective ion channels (Cromer et al. 2002). CLIC proteins show considerable sequence and structural similarity to cytosolic GSTs, particularly to the Omega-class GSTs (Dulhunty et al. 2001; Harrop et al. 2001; Cromer et al. 2007). CLIC proteins lack GST catalytic activity, but do show glutaredoxin-dependent thioltransferase and dehydroascorbic acid reductase activities (Al Khamici et al. 2015). GSTO1-1, along with GSTA1-1 and GSTM2-2, inhibits ryanodine receptor calcium ion channels (Dulhunty et al. 2011).

Glutathionylation of intracellular protein thiols plays an important role in regulating metabolic pathways. GSTO1-1, but not GSTO2-2, catalyzes deglutathionylation reactions, which may influence protein structure and function (Kim et al. 2012; Menon and Board 2013).

Bacterial lipopolysaccharide (LPS) induces the TLR4 signaling pathway in macrophages. GSTO1-1 is, however, required for LPS-stimulated signaling in macrophages (Menon et al. 2015). Gsto1-null mice exhibit an attenuated inflammatory response to administration of LPS and to the effect of a high-fat diet on glucose tolerance and insulin resistance (Menon et al. 2017); moreover, the response to LPS is reduced by inhibition of GSTO1-1. In addition, Gsto1-null mice show increased leakage of bacteria from the colon to the lymphatic system in a model of inflammatory bowel disease. These results indicate that inhibition of GSTO1-1 may be therapeutically useful in inflammatory diseases, including type 2 diabetes.

The demonstration that NEK7, a regulatory kinase in the NLRP3 inflammasome, is activated by GSTO1-1-catalyzed deglutathionylation provided key insight into the role of GSTO1-1 in inflammasome regulation and, ultimately, in the release of inflammatory cytokines from bone marrow-derived macrophages (Hughes et al. 2019). When macrophages were primed with lipopolysaccharide and treated with the potent GSTO1-1 inhibitor 159 (Figure 13), followed by nigericin stimulation, the release of IL-1β was reduced significantly compared with treatment with lipopolysaccharide and nigericin in the absence of the inhibitor. The level of caspase, which converts pro-IL-1β to IL-1β, was also lowered, and macrophage cell death was reduced. Inhibitor 159 (Figure 13) also caused a significant reduction in cytokine release from human peripheral blood mononuclear cells. In an in vivo model of inflammatory bowel disease, the GSTO1-1 inhibitor 159 (Figure 13) caused a lowering of IL-1β release similar to that previously observed in GSTO1-1-deficient mice (Menon et al. 2017)

Several studies have investigated the putative role of hGSTO1 and hGSTO2 polymorphisms in the age-of-onset for Parkinson’s disease and increased risk for late-onset Alzheimer’s disease, but inconsistent results have been obtained (Kolsch et al. 2004; Whitbread et al. 2004; Li et al. 2006; Allen et al. 2012; Piacentini et al. 2012).

Knockdown studies demonstrated a range of pathways regulated by GSTO1 (Ramkumar et al. 2016). Gene expression studies in HCT116 cells with siGSTO1 and RNA Bru-Seq analysis (Paulsen et al. 2014) showed that a large number of genes are up- or downregulated, including transporter-, stress-, and metabolism-related genes. A significant finding was that GSTO1 inhibition decreases the viability of HCT116 cancer cells.

2.2.6.7. Sigma-class (GSTS, PGDS)⁴

The Sigma-class GST was discovered through two lines of investigation. One, studies on the enzymology of prostaglandin biotransformation indicated the involvement of GSTs (Christ-Hazelhof et al. 1976; Urade et al. 1987; Ujihara et al. 1988b). Other studies reported that the S-crystallin family of lens proteins found in cephalopods show approximately 30% sequence identity with Alpha-, Mu-, Pi-, and Theta-class GSTs (Ji et al. 1995; Meyer and Thomas 1995; Tomarev et al. 1995; Kanaoka et al. 1997). Based on a comparison of the primary structure, gene structure, and crystal structure, the squid proteins were assigned to the Sigma-class GST (Ji et al. 1995). Sequence analysis showed that the rat spleen enzyme that catalyzes the isomerization of prostaglandin H₂ (PGH₂, 165, Figure 14) to prostaglandin D₂ (PGD₂, 166, Figure 14) is a Sigma-class GST (Meyer and Thomas 1995). A review about Sigma-class GSTs has been published (Flanagan and Smythe 2011). (A GSH-independent PGD synthase that catalyzes the isomerization of PGH₂ (165, Figure 14) to PGD₂ (166, Figure 14) and is a member of the lipocalin family has also been described (Nagata et al. 1991).)

Figure 14. PGDS substrates. 165, prostaglandin H₂ (PGH₂); 166, prostaglandin D₂ (PGD₂).

2.2.6.7.1. Structure and properties

As with other cytosolic GSTs, the mammalian GSTS is a homodimeric protein with a molecular mass of approximately 50 kDa; in common with GSTA, GSTM, and GSTP, GSTS has an active-site Tyr residue. One human gene (PGD2), one rat gene (Ptgds2), and one mouse gene (Ptgds2) encoding PGDS have been identified (Tables 1–3). The gene encoding hPGD2 has been mapped to chromosome band 4q21-22, and the murine gene encoding Ptgds2 has been mapped to chromosome band 3D–E (Kanaoka et al. 2000).

The crystal structures of rat and human PGDS have been solved (Ji et al. 1995; Kanaoka et al. 1997; Inoue et al. 2003). Crystallographic analysis of rat PGDS revealed a prominent cleft, which is the widest and deepest among the GSTs, that facilitates substrate binding and catalysis (Kanaoka et al. 1997). Both Ca²⁺ and Mg²⁺ increase the activity of hPGDS by ∼150%, apparently by altering the coordinating water structure, an effect not seen with other GSTs (Inoue et al. 2003). The effect of Mg²⁺ on the coordinated water molecules appears to enhance the stabilization of the GSH thiolate by Arg14 and to decrease the K_m of GSH; Ca²⁺, which is a much weaker activator of PGDS, does not affect the binding of GSH.

A single GSTS has been identified in mammals (Kanaoka and Urade 2003). Early studies identified a GSH-dependent enzyme from sheep lung that catalyzed the isomerization of prostaglandin H₂ (PGH₂, 165, Figure 14) to prostaglandin D₂ (PGD₂, 166, Figure 14) and the conjugation of known GST substrates, e.g. CDNB (11, Figure 4) (Christ-Hazelhof et al. 1976); several rat hepatic GSTs also catalyze the isomerization of PGH₂ (165, Figure 14) to PGD₂ (166, Figure 14), although the principal product formed by the rat hepatic GSTs was prostaglandin E₂ (PGE₂).

PGDS activity is increased in human megakaryoblastic CMK cells incubated with phorbol ester (Mahmud et al. 1997). The transcription factor MITF is required for the transactivation of PGD₂ and the expression of PGDS activity in bone-marrow-derived mast cells (Morii and Oboki 2004).

2.2.6.7.2. Substrates

GSTS (PGDS) catalyzes the isomerization of PGH₂ (165, Figure 14) to PGD₂ (166, Figure 14) (Kanaoka et al. 1997; Jowsey et al. 2001). In addition, human, rat, and mouse PGDS catalyze the conjugation of CDNB (11, Figure 4), but not 4-nitrobenzyl chloride (125, Figure 10), trans-4-phenyl-3-buten-2-one (53, Figure 6), 1,2-epoxy-3-(p-nitrophenoxy)propene (116, Figure 10), or ethacrynic acid (52, Figure 5) (Christ-Hazelhof et al. 1976; Urade et al. 1987; Flanagan and Smythe 2011). Expressed rat and human PGDS catalyze the conjugation of CDNB (11, Figure 4), 1-fluoro-2,4-dinitrobenzene, 1-bromo-2,4-dinitrobenzene, 1-iodo-2,4-dinitrobenzene, allyl isothiocyanate, benzyl isothiocyanate, but not the reduction of tert-butyl hydroperoxide (Jowsey et al. 2001; Mazari et al. 2015).

2.2.6.7.3. Inhibitors

The putative role of PGD₂ (166, Figure 14) as a proinflammatory mediator of allergic responses and the multiplicity of disease states in which inflammation is an integral component has stimulated discovery research for PGDS inhibitors (for a review, see Thurairatnam 2012).

The isomerization of PGH₂ (165, Figure 14) to PGD₂ (166, Figure 14) by rat spleen PGDS is inhibited by CDNB (11, Figure 4) and by non-substrate ligands, e.g. bilirubin and indocyanine green (Urade et al. 1987). Screening of a selection of 1040 drugs or additives as possible hPGDS inhibitors identified 23 drugs for further study (Mazari et al. 2015); the most potent inhibitors were erythrosine sodium, suramin, tannic acid, and sanguinarine sulfate (IC₅₀ = 0.2–0.6 μM).

Several PGDS inhibitors have been reported for which interactions with the H-site have been solved by crystallographic analysis. Tetrazolium 167 (Figure 15) inhibits recombinant hPGDS (IC₅₀ = 36.2 μM in the presence of EDTA and 98.1 μM in the presence of Mg²⁺) (Inoue et al. 2004). Tetrazol 168 (Figure 15) is a potent and selective inhibitor of hPGDS (K_i^PGH₂ = 5 μM) and is effective in animal models of allergic disease and suppresses microglial activation and astrogliosis (Matsushita et al. 1998; Aritake et al. 2006; Mohri et al. 2006; Takahashi et al. 2010).

Figure 15. PGDS inhibitors. 167, 2-(2′-benzothiazolyl)-5-styryl-3-(4′-phthalhydrazidyl)tetrazolium chloride (BSPT); 168, 4-benzhydryloxy-1-[3-(1H-tetrazol-5-yl)-propyl]-piperidine (HQL-79); 169, 2-phenyl-5-(1H-pyrazol-3-yl)thiazole; 170, 4-methyl-2-phenyl-N-(4-sulfamoylphenyl)-thiazole-5-carboxamide; 171, N-(5-(1H-indol-4-yl)pyridin-3-yl)-3-acetamidopropanamide; 172, 3-(3-cyclopropyl-1H-pyrazol-4-yl)-5,7-dihydropyrrolo[3,4-b]pyridine-6-carboxylic acid cyclopentylamide; 173, N-(1-amino-1-oxo-3-phenylpropan-2-yl)-6-(thiophen-2-yl)nicotinamide; 174, 6-(3-fluorophenyl)-N-(1-(2,2,2-trifluoroethyl)piperidin-4-yl)nicotinamide; 175, 4-(isoquinolin-1-yl)-N-(2-morpholinoethyl)benzamide; 176, 1-amino-4-[[4-[[4-chloro-6-[(2-sulfophenyl)amino]-1,3,5-triazin-2-yl]amino]-3-sulfophenyl]amino]-9,10-dihydro-9,10-dioxo-2-anthracenesulfonic acid (Cibacron Blue 3G-A); 177, nocodazole; 178, 1-amino-4-(4-aminosulphonyl)-phenyl-anthraquinone-2-sulfonic acid (APAS).

Fragment-based drug-design strategies have provided useful leads in the search for PGDS inhibitors. A 2D NMR fragment-based discovery approach to the design of PGDS inhibitors was used to identify thiazole 169 (Figure 15) as a potent inhibitor of hPGDS (IC₅₀ = 21 nM) (Hohwy et al. 2008). Phenylsulfonamide 170 (Figure 15), which contains the same phenylthiazole structural component as compound 169 (Figure 15), was found to be a potent hPGDS inhibitor (IC₅₀ <0.001 µM) by application of a fragment-based strategy involving a high-concentration bioassay and fluorescence-polarization analysis (Hesterkamp et al. 2007). Indole 171 (Figure 15), which inhibits both hPGDS (IC₅₀ = 0.30 μM) and the synthesis of PGD₂ (166, Figure 14) in MEG-01 cells (EC₅₀ = 0.35 μM), was identified by fragment-based focused screening of 42 000 compounds (Edfeldt et al. 2015). Fragment optimization studies led to the discovery of pyrrolopyridine 172 (Figure 15), which showed favorable pharmacodynamics and pharmacokinetic properties in a rat model (Saxty et al. 2014). A 3-methyl-5-fluoroindole-based compound (cf. 171, Figure 15) was found to be a potent inhibitor of PGD₂ (IC₅₀ = 7 nM) but was a mechanism-based inhibitor of CYP3A4 (Vaz et al. 2018); the 4-trifluoromethylindole analog, which does not contain a metabolically oxidizable 3-methyl group, failed to inactivate CYP3A4 and retained the PGD₂ inhibitory potency (IC₅₀ = 8 nM).

Screening of an ensemble of potential hPGDS inhibitors, in which the design of the screened structures was based on the hPGDS binding of fragments of known inhibitors, showed that nicotinamide 173 (Figure 15) is a potent inhibitor of hPGDS (IC₅₀ = 1.2 μM) (Christ et al. 2010). Information about the active site of PGDS and about important structural elements of previously reported PGDS inhibitors, along with the identification of new substructural components, led to the development of orally effective inhibitors of PGDS (Carron et al. 2010): nicotinamide 174 (Figure 15) is a potent and selective inhibitor (IC₅₀ = 0.5–2.3 nM) and shows favorable pharmacokinetic properties in rats. Examination of crystal structures of PGDS complexes with the potent isoquinoline inhibitor 175 (Figure 15) (IC₅₀ = 2.34 nM) and several analogs indicated that a single primary water molecule, which also interacts with Leu199, forms a critical hydrogen bond with the isoquinoline ring nitrogen of inhibitor 175 (Trujillo et al. 2012). Other investigators also have noted the significance of PGDS inhibitor interactions with the conserved water that associates with C-terminal Leu199 (Aritake et al. 2006; Hohwy et al. 2008; Carron et al. 2010; Edfeldt et al. 2015). Further studies supported the role of water molecules in the active site of hPGDS in substrate and GSH binding (Takaya et al. 2018). Substructure analysis of PGDS inhibitors indicated that a phenyl group, a heterocyclic group, e.g. 173, 174, and 175 (Figure 15), and interaction with a conserved water molecule in the substrate binding site was important for inhibition of PGDS; this information was used to develop a phenyl pyridinylpyrimidine carboxamide with high binding affinity (K_D = 0.14 nM) for hPGDS. Analysis of the crystal structure of the hPGDS complex with the phenypyrimidine carboxamide ligand identified six water molecules, including the water that hydrogen bonds with Leu199 that contribute significantly to binding of the ligand in the active site.

A fragment-based, hit-to-lead search for inhibitors of hPGDS identified 3-carboxamide-7-methoxyquinolines as a starting fragment (Deaton et al. 2019); aided by crystallographic analysis of inhibitor-PGDS complexes, a 3-carboxamide-3-difluoromethoxyquinoline hPGDS was found to be a potent hPGDS inhibitor (IC₅₀ = 9.9 nM), decreased PGD₂ (166, Figure 14) production in a murine mast cell degranulation model of inflammation, possessed favorable pharmacokinetic properties, and showed few apparent off-target liabilities.

Screening of a range of GST inhibitors showed that tetrazole 168 (Figure 15) and Cibacron Blue 3G-A (176, Figure 15) were the most potent inhibitors of hPGDS (IC₅₀ = 0.2–1.8 μM) (Weber et al. 2010). Characterization of the inhibition of hPGDS by nocodazole (177, Figure 15) (IC₅₀ ≈65 μM) by docking calculations indicated binding at the prostaglandin H-site. Kinetic and crystallographic analysis of the interaction of Cibacron Blue 3 G-A (176, Figure 15) (K_i = 87.5 nM) and APAS (178, Figure 15) (K_i = 2.5 μM) with hPGDS indicated that the former compound bound to the H-site whereas the latter compound bound to the G-site (Kado et al. 2012).

2.2.6.7.4. Tissue and species distribution

The expression of PGDS varies significantly among species and organs (Ujihara et al. 1988a, 1988b; Thomson et al. 1998; Jowsey et al. 2001; Mohri et al. 2006; Shimura et al. 2010). Immunoblot analysis shows that rat PGDS expression is highest in spleen and bone marrow. Quantitative RT-PCR shows abundant hPGDS mRNA in macrophages and placenta and, in contrast to the rat, little mRNA in spleen and bone marrow. Also, hPGDS mRNA was moderately abundant in lung, but rPGDS was not detected by immunoblot analysis in lung (Jowsey et al. 2001). In the rat, PGDS is found in histiocytes and dendritic cells in spleen, thymus, and intestinal Peyer’s patches, indicating a possible role in immune responses (Urade et al. 1989).

2.3. Mitochondrial and peroxisomal glutathione transferases (Kappa-class, GSTK, DsbA-L)⁵

2.3.1. Structure and properties

Early studies indicated the presence of enzymes in liver mitochondrial fractions that catalyzed GSH S-conjugate formation, although rates of conjugation were typically much lower in mitochondria than in cytosol (Booth et al. 1961; Suga et al. 1967; Kraus and Gross 1979; Wahllander et al. 1979). Harris et al. (1991) purified a GST (termed GST 13-13) from rat liver mitochondrial matrix whose amino acid sequence was similar to the Theta-class GSTs; the dimeric enzyme catalyzed the conjugation of CDNB (11, Figure 4) and ethacrynic acid (52, Figure 5) but lacked activity with a panel of known GST substrates. A cDNA clone that encodes GST 13 subunit was isolated from a rat cDNA library (Pemble et al. 1996); a search of a human expressed sequence tag (EST) database showed that the rat and human amino acid sequences were similar (77%) but differed from other known GSTs. The novel GST genes were assigned to a new subclass, Kappa. A mouse liver mitochondrial Kappa-class GST has also been characterized (Jowsey et al. 2003). Given its putative role in adiponectin polymerization, GSTK1-1 has also been termed disulfide bond-forming oxidoreductase A-like protein (DsbA-L) (Liu et al. 2008). A review of the Kappa-class GSTs has been published (Morel and Aninat 2011).

The crystal structure of GSTK1-1 is more similar to bacterial disulfide-bond A oxidoreductase (DsbA) and 2-hydroxychromene-2-carboxylic acid isomerase (HCCA) than to members of the canonical GST superfamily (Ladner et al. 2004; Robinson et al. 2004; Li et al. 2005a; Thompson et al. 2007). GSTK subunits show a thioredoxin-like domain that is interrupted with an α-helical domain, whereas the canonical GSH transferases have a contiguous thioredoxin-like domain at the N-terminus. A conserved serine (Ser16) is found as an active-site residue in hGSTK1-1 (Wang et al. 2011a). Although the G-sites in GSTK1-1 and in the canonical GST superfamily are similar, both GSTK1-1 and the canonical GSTs fail to catalyze DsbA-type reactions, apparently because they lack the CxxC motif found in DsbA (Atkinson and Babbitt 2009; Wang et al. 2011a). As with the cytosolic GSTs, however, the thioredoxin fold provides residues that serve to recognize GSH (Ladner et al. 2004; Li et al. 2005a). The H-site of hGSTK1-1 is a hydrophobic pocket that accommodates S-(hexyl)GSH (Wang et al. 2011a).

A single gene, GSTK1, loci located on chromosome band 7q34 encodes human GSTK1 (HGNC: 16906; https://www.genenames.org/; accessed 31 October 2019) (Morel et al. 2004). The rat (RGD: 735188; https://rgd.mcw.edu/; accessed 31 October 2019) and mouse (MGI: 1923513; http://www.informatics.jax.org/; accessed 31 October 2019; NCBI: 76263) genes Gsk1 encode r- and mGSTK1, respectively (see also Tables 1–3).

The mitochondrial localization of GSTK1-1 in mouse liver and kidney was established by electron microscopy with immunocolloidal gold staining (Thomson et al. 2004). In addition to GSTK1-1, Alpha-, Mu-, Pi-, and Zeta-class GSTs have been identified in mitochondria (Soboll et al. 1995; Gallagher et al. 2006; Goto et al. 2009; Li et al. 2011; Raza 2011; Sun et al. 2012).

The presence of GST-catalyzed GSH S-conjugate formation in peroxisomes was reported (Nishino and Ito 1990b). The amino acid sequences of rat and hGSTK1 revealed a peroxisomal targeting sequence (Ala–Arg–Leu) in the C-terminus (Pemble et al. 1996; Morel et al. 2004); Western blot analysis as well as transfection of HepG2 cells with a green fluorescent protein-hGSTK1 fusion construct and fluorescence analysis clearly demonstrated the presence of GSTK1 in peroxisomes. Western blot analysis showed a lack of hGSTK1 expression in cytosol.

In addition to its localization in mitochondria and peroxisomes, DsbA-L is also found in the endoplasmic reticulum (ER) (Liu et al. 2015). DsbA-L contains a highly conserved ER targeting sequence in the N-terminal (residues 1–12). The expression of DsbA-L in the ER is largely blocked with a mutant that lacks residues 1–6, but mitochondrial expression is not altered.

2.3.2. Substrates

The physiological function of GSTK1-1 has not been fully established, but human, rat, and mouse GSTK1-1 catalyze the conjugation of several canonical GST substrates (Jowsey et al. 2003; Morel et al. 2004; Robinson et al. 2004; Morel and Aninat 2011), including CDNB (11, Figure 4), ethacrynic acid (52, Figure 5), tert-butyl hydroperoxide, and cumene hydroperoxide (15, Figure 4). Significant species differences in substrate selectivities and in rates of biotransformation were observed: human GSTK1-1, but not rat or mouse GSTK1-1, catalyzes the conjugation of 4-nitrobenzyl chloride (125, Figure 10), 4-nitrophenyl acetate, and trans,trans-nona-2,4-dienal (Robinson et al. 2004). With CDNB (11, Figure 4) as the substrate, the relative rates of biotransformation were: mouse > rat > human. Cumene hydroperoxide (15, Figure 4) was a substrate for recombinant GSTK1-1 of all three species and the relative rates were: mouse > human > rat; tert-butyl hydroperoxide was a poor substrate for the three GSTKs.

2.3.3. Inhibitors

S-(2,4-Dintrophenyl)GSH (12, Figure 4) inhibits hGSTK1-1 (Wang et al. 2011a), but studies with other inhibitors have apparently not been reported.

2.3.4. Tissue and species distribution

Western blot analysis showed that mouse GSTK1 expression is the highest in liver and stomach and is also present in kidney, heart, large intestine, testis, and lung (Thomson et al. 2004); mGSTK1 is not expressed in small intestine, brain, spleen, or skeletal muscle. It has also been reported that mGSTK1 mRNA is expressed in all tissues examined (Knight et al. 2007); expression was higher in female mice than in male mice in heart, kidney, and lung tissue. Human GSTK1 mRNA is expressed in all tissues examined but at differing levels (Morel et al. 2004). mGSTK1 is highly expressed in adipose tissue (Liu et al. 2008); its expression in 3T3-L1 adipocytes is decreased by TNFα and increased by the PPARγ agonist rosiglitazone.

Gstk1^−/− mice lack immunologically detectable GSTK1-1 but liver and kidney mitochondria show moderately reduced enzymatic activity with CDNB (11, Figure 4) as the substrate compared with wild-type mice (Blackburn et al. 2011); the canonical GSTs found in mitochondria may account for this activity.

2.3.5. Other biological functions

The expression of DsbA-L is negatively correlated with obesity in mice and human subjects (Liu et al. 2008). Moreover, DsbA-L overexpression promotes adiponectin multimerization and silencing of DsbA-L expression by RNAi decreases adiponection multimerization and secretion in 3T3-L1 adipocytes. In Gstk^−/− mice, serum concentrations of adiponectin as well as adiponectin multimerization are not altered, thereby questioning an absolute requirement of GSTK1-1 for adiponectin multimerization (Theodoratos et al. 2012). Later studies showed, however, that colocalization of DsbA-L in both mitochondria and ER is required for promoting adiponectin multimerization and preventing ER oxidative stress (Liu et al. 2015).

The DsbA-L -1308 G>T (rs1917760) sequence variant is not associated with type-2 diabetes and obesity in Chinese subjects, whereas T allele carriers showed favorable insulin secretion function (Gao et al. 2009). The -1308 G > T genotype is associated with higher body mass indices (BMI) in Japanese male schizophrenic subjects, whereas female subjects with this genotype exhibited a lower BMI; there was no association between the -1038 T genotype and BMI in control subjects (Oniki et al. 2016). Later studies showed that the -1308 T allele modulates hGSTK1 transcription by disrupting the RXR/FXR transcription-factor binding site (Shield et al. 2010); a -1032G>C sequence variant alters a CpG methylation site and decreases transcription.

Support for the concept that redox communication between peroxisomes and mitochondria is part of a complex signaling process has been reported (Wang et al. 2013); mitochondrial GSTK1-1 may play a role in detoxifying lipid peroxides generated in response to peroxisomal oxidative stress. GSTK1 expression is decreased in five mouse models of hypertrophic cardiomyopathy and the decrease may contribute to hypertrophic cardiomyopathy through increased oxidative stress, lipid peroxidation, and mitochondrial dysfunction (Sasagawa et al. 2016).

Male Gstk1^−/− mice show shortened life span compared with female Gstk1^−/− mice; moreover, both male and female Gstk1^−/− mice show glomerulopathies similar to those seen in the human nephrotic syndrome (Blackburn et al. 2011). The pathophysiological processes that account for the shortened life span and glomerulopathies have not been elucidated.

2.4. Microsomal glutathione transferases (MGST)

2.4.1. Structure and properties

MGSTs are members of the membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG) superfamily (Jakobsson et al. 1999; Bresell et al. 2005; Pettersson et al. 2005; Martinez Molina et al. 2008). The MAPEG superfamily includes MGST1, MGST2, MGST3, leukotriene-C₄ synthase (LTC4S, EC4.4.1.20), arachidonate 5-lipoxygenase-activating protein (FLAP, EC1.13.11.34), and prostaglandin-E synthase (PGE synthase, EC5.3.99.3). All MAPEG superfamily members are GSTs and catalyze GST conjugative and peroxidative reactions. Several reviews about the MGSTs are available (Morgenstern and DePierre 1985, 1988; Andersson et al. 1994a; Morgenstern 2005; Morgenstern et al. 2011). The genes encoding hMGST1, MGST2, and MGST3 are located on chromosome bands 12p13.1-p13.2 (Kelner et al. 2000), 1q23 (Jakobsson et al. 1997), and 4q28-31 (Jakobsson et al. 1996).

Early studies provided evidence for GST activity in hepatic microsomal fractions (Booth et al. 1961; Kraus et al. 1973; Kraus 1975; Nemoto and Gelboin 1975; Glatt and Oesch 1977; Friedberg et al. 1979; Kraus and Gross 1979; Lee and McKinney 1982). Morgenstern and coworkers (Morgenstern et al. 1979, 1982) showed that microsomal GST activity was a property of GSTs localized in microsomal membranes rather than contamination by cytoplasmic GSTs. Moreover, with CDNB (11, Figure 4) as the substrate, N-ethylmaleimide (NEM) stimulated microsomal GST activity several fold, but had no effect on cytosolic GST activities.

Purification and sequence analysis indicated that rat liver MGST1 is a homotrimer with subunit masses of 17.2 kDa (Morgenstern et al. 1982, 1985; Boyer et al. 1986). Rat MGST1 contains a single Cys residue (Cys49) in each subunit, which was determined to be the site of activation by sulfhydryl reagents, e.g. NEM (Morgenstern and DePierre 1983; Morgenstern et al. 1985).

Electron crystallography has provided key structural information about MGST1; a review of the application of electron crystallography to the structural determination of MAPEG proteins, including MGST1, is available (Hebert and Jegerschöld 2007). Early structural studies of rat liver MGST1 provided details of its trimeric structure and indicated the presence of two α-helices in each subunit (Hebert et al. 1995, 1997). A 3D electron crystallographic map of rMGST1 confirmed both its trimeric structure as well as the earlier report that the binding sites for GSH and the second substrate are located on the cytoplasmic surface (Andersson et al. 1994b; Holm et al. 2002; Hebert et al. 2005). A proposed topology, based on a projection map at 6-Å resolution, identified four transmembrane α-helices in each monomer; peptides connecting α-helices 1 to 2 and 3 to 4 are located on the cytoplasmic side, and the N- and C-termini are located on the luminal side (Schmidt-Krey et al. 2000; Holm et al. 2002).

The electron crystallographic structure of the rGSTM1-GSH complex at 3.2-Å resolution revealed the first structural details about the active site and the cytoplasmic domain responsible for activation by sulfhydryl reagents (Holm et al. 2006). MGST1 lacks a thioredoxin fold, indicating that its evolutionary origin differs from that of the cytosolic GSTs. Each subunit of the homotrimer contains four left-handed α-helix bundles. Transmembrane helices 2 and 3 are connected by a short loop on the luminal side, whereas helices 3 and 4 are connected by a Pro-rich loop on the cytosolic side; domain E links helices 1 and 2 and contains the stress sensor Cys49. The monomer appears to be stabilized by interactions between transmembrane helices 2 and 3 involving polar residues near the cytoplasmic side, whereas hydrophobic Phe residues located toward the luminal side have a similar function.

Cryoelectron microscopic studies provided additional details about the structure of rat MGST1 (Kuang et al. 2017). rMGST1 is present as a dimer of trimers with phospholipids at the interface between the two trimers. The GSH binding site is located at the interface of transmembrane helices 1 and 2. In contrast to the other members of the MPAEG family, GSH is bound in an extended conformation with a hydrogen bond between the thiol group and Arg130, which is located in transmembrane 4 of an adjacent subunit; the R130A mutant is devoid of activity, indicating the critical role of Arg130 in stabilizing the GSH thiolate. Residues that either contribute, or potentially contribute, to GSH binding reside in both transmembrane helices 1 and 2 of the GSH-binding subunit and in transmembrane helices 2 and 4 of a neighboring subunit of the trimer. His76 and Glu81, which are highly conserved in the MAPEG family, are involved in intermolecular contacts between monomers and also contribute to the binding of GSH. The binding site for lipophilic substrates is located near the hydrocarbon region of the bound phospholipids.

A remarkable feature of MGST1 is activation by agents or treatments that modify Cys49. The NEM-treated enzyme was initially isolated and purified in 1982 (Morgenstern et al. 1982). Purified native rMGST1 was activated 15-fold by NEM with CDNB (11, Figure 4) as the substrate and 10-fold with cumene hydroperoxide (15, Figure 4) as the substrate, whereas no activation was seen with methyl iodide, 4-nitrobenzyl chloride (125, Figure 10), 1,2-dichloro-4-nitrobenzene, 4-nitropyridine N-oxide, and hexachlorobutadiene as substrates (Morgenstern and DePierre 1983).

Studies on the mechanism of activation of rMGST1 showed that the pK_a of Cys49 is similar to low molecular-weight thiols, e.g. GSH, that the reactivity of Cys49 is the same in all subunits, and that subunits react independently (Svensson et al. 2000). The ability of p-benzoquinones to activate rMGST1 varies with their putative reactivity and steric properties. Reaction of rMGST1 with NEM showed that modification of a single subunit gave no activation and that activation increased as two or three subunits are modified; full activation requires modification of all three subunits. Later studies showed that one GSH thiolate is formed per homotrimer (Morgenstern et al. 2001; Svensson et al. 2004).

Kinetic analyses of the binding and activation of GSH by rMGST1 showed that GSH thiolate formation, rather than the chemical reaction of the substrate with the thiolate, is the rate-limiting step for chemically reactive substrates such as CDNB (11, Figure 4) (Morgenstern et al. 2001; Svensson et al. 2004). Binding and ionization of GSH involves a slow conformational transition that results in tight binding of the GSH thiolate. Moreover, the reaction of Cys49 of MGST1 with NEM increases the rate of thiolate formation, indicating that activation is associated with a specific chemical event that facilitates the cooperative conformational transition (Busenlehner et al. 2004, 2007).

Electrospray mass spectrometric data obtained by Ålander et al. (Ålander et al. 2009b) showed, as also indicated by electron crystallographic data (Holm et al. 2006; Kuang et al. 2017), that each rMGST1 trimer binds three GSH molecules, rather than one as reported earlier (Sun and Morgenstern 1997; Lengqvist et al. 2004). Equilibrium dialysis studies revealed that there is one high-affinity and two low-affinity sites for GSH and that only the high-affinity site is catalytically competent (Ålander et al. 2009b); hence, MGST1 shows one-third-of-sites-reactivity. Kinetic studies that evaluated GSH binding, thiolate anion stabilization, and electrophile binding and reaction demonstrated that unactivated MGST1 exists in a dynamic equilibrium between activated and unactivated forms (Spahiu et al. 2017). With the chemically unactivated enzyme and reactive electrophilic substrates, the data indicate that a fraction (∼10%) of the enzyme exists in an activated form. Also, the activated enzyme favors catalytic efficiency over stability.

Cys49 serves as a stress sensor in that activation of rMGST1 may favor detoxication of cellular electrophiles (Busenlehner et al. 2004). A diverse group of agents or treatments activate MGST1 (for a review, see Shimoji et al. 2007): thiol-reactive agents (Morgenstern and DePierre 1983; Aniya and Anders 1989b), reactive oxygen and nitrogen species (Aniya and Anders 1989a, 1992; Imaizumi et al. 2006), proteolysis (Morgenstern et al. 1989; Nakama et al. 2010). Activation of MGST1 by hydrogen peroxide is accompanied by dimerization of the transferase (Aniya and Anders 1992).

Identification of hMGST2 was accomplished by a TBLASTN search of Gen Bank with the 5-lipoxygenase-activating protein (FLAP) peptide sequence (Jakobsson et al. 1996). Similarly, a TBLASTN search with leukotriene C4 synthase (LTC4S) peptides identified rMGST2 (Schröder et al. 2003). Both hMGST2 and rMGST2 catalyze the biotransformation of LTA4 (188, Figure 16) to LTC4 (189, Figure 16) and the biotransformation of 5-HPETE (190, Figure 16) to HETE (191, Figure 16) (Jakobsson et al. 1996; Schröder et al. 2003; Ahmad et al. 2013). LTC4S, which also catalyzes the conversion of LTA4 (188, Figure 16) to LTC4 (189, Figure 16), is a protein distinct from MGST2, as MGST2, but not LTC4S, catalyzes the conjugation of CDNB (11, Figure 4). Also, in contrast to MGST1, NEM fails to activate hMGST2 (Jakobsson et al. 1996). Kinetic studies show that hMGST2 exhibits one-third-of-sites reactivity and that the enzyme lowers the pK_a of GSH (Ahmad et al. 2013); with reactive substrates, e.g. CDNB (11, Figure 4), the rates of GSH thiolate anion formation and the chemical reaction step are similar, but with substrates of lower reactivity, the chemical conjugation step becomes rate limiting. Other studies established the trimeric structure of hMGST2, showed that each subunit possesses a GSH binding site, and confirmed the one-third-of-sites reactivity (Ahmad et al. 2015).

Figure 16. MGST substrates. 179, bis-(2,4-dinitrophenylsulfonyl)rhodamine; 180, 2-(2,4-dinitrophenylsulfonyl)rhodamine; 181, rhodamine; 182, (Z)-N-(9-amino-5H-benzo[a]phenoxazin-5-ylidene)-2,4-dinitrophenylsulfonamide (DNs-CV); 183, crystal violet; 184, trichloroethylene; 185, S-(1,2-dichlorovinyl)GSH; 186, chlorotrifluoroethylene; 187, S-(2-chloro-1,1,2-trifluoroethyl)GSH; 188, leukotriene A₄ (LTA4); 189, leukotriene C₄ (LTC4); 190, 5-hydroperoxyeicosatetraenoic acid (5-HPETE); 191, 5-hydroxyeicosatetraenoic acid (5-HETE).

hMGST3 was identified by a TBLASTN search of GenBank with MGST2 peptide sequences (Jakobsson et al. 1997). hMGST3 catalyzes the biotransformation of LTA4 (188, Figure 16) to LTC4 (189, Figure 16) and the reduction of 5-HPETE (190, Figure 16) but fails to catalyze the conjugation of CDNB (11, Figure 4) in the absence or presence of NEM. Rat MGST3, in contrast to hMGST3, fails to catalyze LTC4 (189, Figure 16) synthesis (Schröder et al. 2003).

Most inducers of cytosolic GSTs fail to alter the activity of MGST. Treatment of rats with phenobarbital, 3-methylcholanthrene, and trans-stilbene oxide failed to alter microsomal GST activity but increased cytosolic GST activity (Morgenstern et al. 1980). Mice fed 2(3)-tert-butyl-4-hydroxyanisole for two weeks showed a significant increase not only in cytosolic GST activity, but also in both unactivated and NEM-activated MGST activity (Morgenstern and Dock 1982). Mice fed peroxisome proliferators show an increase in MGST1 activity (Morgenstern and DePierre 1988).

MGST2 and MGST3 are also induced by several treatment regimens (Higgins and Hayes 2011). Triggering of the endoplasmic reticulum stress-response by brefeldin A and tunicamycin upregulates MGST2 and, thereby, the synthesis of LTC4 (189, Figure 16), which results in reactive oxygen species (ROS) production in non-hematopoietic cells (Dvash et al. 2015). 5-Fluorouracil (5-FU) activates the MGST2-LTC4 signaling pathway, thus producing ROS; MGST2 null mice are resistant to 5-FU-induced toxicity, and its toxicity in mice is reduced by the MGST inhibitor pranlukast (193, Figure 17).

Figure 17. MGST inhibitors. 192, (2S,3R)-3-(2-carboxyethylsulfanyl)-2-hydroxy-3-[2-(8-phenyloctyl)phenyl]propanoic acid (pobilukast, SKF 104353); 193, N-[4-oxo-2-(2H-tetrazol-5-yl)chromen-8-yl]-4-(4-phenylbutoxy)benzamide (pranlukast, ONO-1078); 194, 4-[(4S,5R,6E,8E,10Z,13Z)-1-carboxy-4-hydroxynonadeca-6,8,10,13-tetraen-5-yl]sulfanylbenzoic acid (BAYu9773); 195, 3-[3-tert-butylsulfanyl-1-[(4-chlorophenyl)methyl]-5-propan-2-ylindol-2-yl]-2,2-dimethylpropanoic acid (MK-886).

No mutations were found in MGST1 in subjects with chronic pancreatitis (Schneider et al. 2004). There are four putative promoters of MGST1 (Kelner et al. 2000); both the MGST1b and -1d promoters are active, and two variants of the MGST1b promoter lowered the transcription of the luciferase gene 1.4-fold (Iida et al. 2001; Guy et al. 2004). Thirteen SNPs were identified in an investigation of the association of MGST1 gene locus polymorphisms and colorectal cancer in Han Chinese populations; a risk of colorectal cancer was suggested for individuals having both of two specific risk alleles (Zhang et al. 2007).

Treatment of HepG2 cells with 5-aza-2′-deoxycytidine does not alter MGST1 expression, indicating that DNA methylation does not play a role in MGST1 gene expression (Björkhem-Bergman et al. 2014).

2.4.2. Substrates

In many substrate-selectivity studies, the identity of the MGST involved was not determined, but given its high expression in the liver, MGST1 was the likely catalyst. It should be noted, however, that in contrast to MGST1, the activities of MGST2 and MGST3 are not altered by the sulfhydryl reagent NEM (Jakobsson et al. 1996, 1997).

N-Acetyl-L-cysteine (NAC) can replace GSH with CDNB (11, Figure 4) as the second substrate (Weinander et al. 1994). The use of NAC rather than GSH allows determination of rates of rMGST1-catalzyed reactions in the presence of cytosolic GSTs. The GSH analogs Glutaryl-l-Cys-Gly and α-l-Glu-l-Cys-Gly also serve as substrates for rMGST1 (Mosialou et al. 1993).

rMGST1 substrates include many compounds that are substrates for the cytosolic GSTs, including 1-halo-2,4-dinitrobenzenes (Morgenstern and DePierre 1988). In contrast, rMGST1 shows low or no activity with several compounds that are known substrates for the cytosolic GSTs, including bromosulfophthalein, 1,2-epoxy-(3-nitrophenoxy)propane (116, Figure 10), ethacrynic acid (52, Figure 5), 4-nitropyridine N-oxide, trans-4-phenyl-3-buten-2-one (53, Figure 6) (Morgenstern et al. 1982; Morgenstern and DePierre 1983).

Several compounds that release fluorescent S-(2,4-dinitrophenyl)GSH (12, Figure 4) have been reported as GST substrates. The fluorogenic substrates 6-chloroacetyl-2-dimethylaminonaphthalene (95, Figure 8) and bis-2,4-dinitrophenylsulfonyl-rhodamine (179, Figure 16) are substrates for rMGST1 (and for cytosolic GSTs) (Svensson et al. 2002; Ålander et al. 2009a).

Zhang et al. (2011a) prepared a series of 4-nitro- and 2,4-dinitrophenylsulfonamide derivatives of coumarin, cresyl violet, and rhodamine (181, Figure 16) as fluorescent substrates for rMGST1 and for human cytosolic GSTs. The precursor 4-nitro- or 2,4-dinitrophenylsulfonamides (e.g. 182, Figure 16) showed very low fluorescent quantum efficiencies but GSTs catalyzed the release of the coumarin, cresyl violet (183, Figure 16), and rhodamine (181, Figure 16) moieties that showed high fluorescent quantum efficiencies. The highest specific activities for the rMGST1-catalyzed biotransformation of the fluorogenic substrates were seen with the 2,4-dinitrophenylsulfonamide derivatives of cresyl violet (182, Figure 16) and rhodamine; the 4-nitrophenylsulfonamide derivatives were poor substrates. Cell lysates of control MCF7 cells (a cell line with low GST and MGST1 activities) showed low activities with the fluorogenic substrates, whereas much higher rates were seen in lysates from rMGST1-overexpressing cells, and the rates were increased by incubation with NEM. Fluorescence microscopic studies of MCF7 cells showed that DNs-cresyl violet (182, Figure 16) allowed imaging of expressed rMGST1 in MCF7 cells.

N-(2,4-Dinitrophenylsulfonyl)doxorubicin (32, Figure 4) and the 4-nitrophenylsulfonyl analog were prepared as doxorubicin (33, Figure 4) prodrugs (Johansson et al. 2011); rMGST1-catalyzed the cleavage of the 2,4-dinitro prodrug (32, Figure 4) at a higher rate than the less electrophilic mono-nitro analog, N-(4-nitrophenylsulfonyl)doxorubicin [See also GSTA, (2), Substrates, and GSTP, (2), Substrates]. rMGST1 overexpressing MCF7 cells are resistant to doxorubicin-induced cytotoxicity, and this resistance is partially reversed by prodrug 32 (Figure 4); N-(4-nitrophenylsulfonyl)doxorubicin completely overcame the resistance but was less cytotoxic than prodrug 32 (Figure 4) to MGST1 overexpressing MCF7 cells. These studies were extended by van Gisbergen et al. (van Gisbergen et al. 2016) to include an analog of prodrug 32 (Figure 4) in which the 4-nitro group of the N-(2,4-dinitrophenylsulfonyl) moiety is replaced by the less electronegative acetyl group. The 4-acetyl-2-nitrophenylsulfonyl-doxorubicin prodrug was cleaved by rMGST1 at a much slower rate than prodrug 32 (Figure 4) (38.7 versus 463 nmol/mg/min). Similar to the slowly cleaved 4-nitrophenylsulfonyl-doxorubicin prodrug (32, Figure 4) (Johansson et al., 2011), the 4-acetyl-2-nitro analog completely reversed MGST1-induced resistance to doxorubicin in MCF7 cells. It was proposed that the less readily cleavable sulfonamide prodrugs of doxorubicin capitalize on the overexpression of MGST1 to achieve a more optimal GST-dependent accumulation of doxorubicin in the target cells than prodrugs that undergo a very rapid GST-catalyzed cleavage.

rMGST1 catalyzes the biotransformation of glyceryl trinitrate at rates greater than those observed with the cytosolic GSTs (Ji and Bennett 2006); interestingly, however, treatment with agents that activate rMGST1 activity with CDNB as substrate inhibit the biotransformation of glyceryl trinitrate.

Many haloalkenes are substrates for rMGST1 (McLellan et al. 1989; Hargus et al. 1991; Jolivette and Anders 2002); most in vitro investigations of the rMGST1-catalyzed conjugation of haloalkenes, have, however, used hepatic or renal or both microsomal fractions rather than the purified enzyme (for reviews, see Dekant et al. 1993; Dekant et al. 1994; Anders and Dekant 1998). Most chloroalkenes undergo addition–elimination reactions that yield S-(halovinyl)GSH conjugates; trichloroethylene (184, Figure 16) and tetrachloroethylene are biotransformed to S-(1,2-dichlorovinyl)GSH (185, Figure 16) and S-(trichlorovinyl)GSH, respectively (Dekant et al. 1987, 1990). Dichloroacetylene undergoes an addition reaction with GSH to give S-(1,2-dichlorovinyl)GSH (185, Figure 16) (Kanhai et al. 1989). Fluoroalkenes typically undergo addition reactions that give S-(fluoroalkyl)GSH conjugates; for example, 2-chloro-1,1,2-trifluoroethylene (186, Figure 16) and tetrafluoroethylene are biotransformed to S-(2-chloro-1,1,2-trifluoroethyl)GSH (187, Figure 16) and S-(tetrafluoroethyl)GSH, respectively (Odum and Green 1984; Dohn et al. 1985). The sevoflurane degradation product 2-(fluoromethoxy)-1,1,3,3,3-pentafluoro-1-propene undergoes both addition and addition–elimination reactions (Altuntas and Kharasch 2001).

rMGST1 shows significant peroxidase activity with organic hydroperoxides as substrates: examples include cumene hydroperoxide (15, Figure 4), linoleic acid hydroperoxide, linoleic acid ethyl ester hydroperoxide, dilinoleoylphosphatidylcholine hydroperoxide, and cholesteryl linoleate hydroperoxide (Morgenstern and DePierre 1983; Mosialou and Morgenstern 1989; Mosialou et al. 1995).

hMGST2 shows high activity with CDNB (11, Figure 4) but much lower specific activities with other substrates, including leukotriene A₄ (188, Figure 16), EPNP (116, Figure 10), several lipid hydroperoxides (e.g. 5-hydroperoxyeicosatetraenoic acid; 190, Figure 16), HNE (17, Figure 4), 4-chloro-3-nitrobenzaldehyde, 4-chloro-3-nitrobenzamide, and 4-chloro-3-nitrobenzacetophenone (Ahmad et al. 2013). HEK293 cells that highly express hMGST2 are resistant to the cytotoxicity of busulfan (21, Figure 4), indicating that hMGST2 catalyzes the conjugation of busulfan (21, Figure 4) (Harkey et al. 2005).

hMGST3 catalyzes the reduction of 5-HPETE (190, Figure 16) but CDNB (11, Figure 4) is not a substrate (Jakobsson et al. 1997).

2.4.3. Inhibitors

Rose bengal, tributyltin acetate, S-hexylglutathione, indomethacin, cibacron blue, hematin, cholic acid, and bromosulphophtalein inhibit MGST1 (Andersson et al. 1988; McLellan et al. 1989; Mosialou and Morgenstern 1990). Ethacrynic acid (52, Figure 5) and ethacraplatin (51, Figure 5) inhibit MGST1 and reverse cisplatin resistance in rMGST1 overexpressing MCF7 cells (Johansson et al. 2011). The GSH S-conjugate of CDNB (12, Figure 4) inhibits rMGST1 (Ålander et al. 2009b).

Leukotriene C₄ (189, Figure 16) is a potent (K_i = 5.4 nM), tight-binding inhibitor of rMGST1 (Bannenberg et al. 1999); the cysteinyl-leukotriene analogs pobilukast (192, Figure 17), pranlukast (193, Figure 17), and BAYu9773 (194, Figure 17) are also strong inhibitors of rMGST1.

The rat mitochondrial-membrane-bound MGST1 (mtMGST1) is present in both the outer and inner mitochondrial membranes but only the inner mtMGST1 is inhibited by the mitochondrial permeability transition inhibitors, e.g. cyclosporin A and bongkrekic acid (Ulziikhishig et al. 2010).

The FLAP inhibitors MK-886 (195, Figure 17) and LTC4 (189, Figure 17) inhibit the rat MGST2-catalyed biotransformation of LTA4 (188, Figure 17), 5-HPETE (190, Figure 17), and CDNB (11, Figure 4) (Schröder et al. 2003). Similarly, MK-886 (195, Figure 17) and LTC₄ (189, Figure 17) inhibit the rat MGST3-catalyed biotransformation of 5-HPETE (190, Figure 17) (Schröder et al. 2003).

2.4.4. Tissue and species distribution

MGST1 is an abundant enzyme and accounts for 3% of the endoplasmic reticulum protein and 5% of the outer mitochondrial membrane protein (Morgenstern 2005). MGST1 activity is widely distributed in rat tissues, but the highest activity is found in liver and testes, followed by intestine and adrenal (DePierre and Morgenstern 1983; Morgenstern et al. 1984); activation of rMGST1 by NEM varies widely among tissues with about a fivefold activation seen in liver and little activation in extrahepatic tissues. rMGST1 is also found in the inner and outer mitochondrial membranes (Morgenstern et al. 1984; Nishino and Ito 1990a; Ulziikhishig et al. 2010). In the human small intestinal tract, immunohistochemical analysis shows that MGST1 is found only in the crypts (Hayes et al. 1989). Immunoblotting studies confirm the widespread distribution of MGST1 in rat tissues (Otieno et al. 1997); in liver tissue, staining was more intense in centrilobular regions than in periportal regions. Mouse MGST1 mRNA expression is highest in liver tissue (Kelner et al. 2004; Knight et al. 2007; Fu et al. 2016). In human fetal tissue, the highest level of MGST1 mRNA is found in the adrenal gland (Björkhem-Bergman et al. 2014). Neither MGST1 mRNA nor MGST1 protein is detectable in human neuroblastoma cell lines or primary tumor tissue (Kelner et al. 2014).

MGST1 is strongly expressed in mouse hematopoietic stem and progenitor cells (HSPC) (Bräutigam et al. 2018). Treatment of HSPC cells with MGST1 shRNA significantly reduced MGST1 expression but failed to decrease cell viability.

MGST1 is overexpressed in many cancers that are associated with increased metastatic potentials and has been proposed as a biomarker for clinical staging of tumors; expression of MGST1 is associated with drug resistance in cancer chemotherapy (Morgenstern et al. 2011).

Northern blot analysis showed that hMGST2 is expressed in heart, liver, skeletal muscle, pancreas, spleen, prostate, testes, ovary, small intestine, colon, adrenal medulla and cortex, thyroid, bone marrow, mononuclear, and polymorphonuclear peripheral blood cells, and in promyelocytic and myelogenous leukemia cell lines, but not in kidney; hMGST2 is, however, expressed in fetal kidney and liver (Jakobsson et al. 1996).

MGST3 mRNA is widely expressed in human tissues, including heart, skeletal muscle, adrenal cortex, brain, placenta, liver, kidney, intestinal tract, bone marrow, and glandular tissues (Jakobsson et al. 1997); limited expression was found in lung, thymus, and leukocytes. Rat MGST3 mRNA expression in the brain is broadly distributed and is found only in neurons, although the possibility of expression in other brain cell types could not be excluded; expression was high in the hippocampus and nuclei of cranial nerves, as well as in the spinal cord, and dorsal root ganglia (Fetissov et al. 2002).

The administration of LPS failed to alter rMGST3 expression, indicating the lack of a role in the biosynthesis of proinflammatory eicosanoids.

2.4.5. Other biological functions

The MGST1 found in rat liver mitochondrial membranes, i.e. mtMGST1, appears to play a role in apoptosis (Lee et al. 2008); modification of mtMGST1 thiol groups by S-glutathionylation or dimer formation leads to cytochrome c release and opening of the mitochondrial permeability transition (MPT) pores. Incubation of mitochondria with peroxynitrite caused activation of MGST1 in the inner mitochondrial membrane and the production of a high molecular-weight protein complex consisting of disulfide-linked mtMGST1, adenine nucleotide translocator, and cyclophillin D; the newly formed protein complex is proposed to function as an MPT pore (Aniya and Imaizumi 2011; Imaizumi and Aniya 2011). Earlier studies showed that gallic acid-induced oxidative stress activates mtMGST1 and causes mitochondrial swelling and cytochrome c release, which is blocked by GST inhibitors but not by MPT inhibitors, e.g. cyclosporine A (Hossain et al. 2009); these findings indicate a role for mtMGST1 in regulating MPT.

MGST1 and mtMGST1 play a role in protecting cells against oxidative-stress-induced cytotoxicity by catalyzing the reduction of hydroperoxides (Maeda et al. 2005; Siritantikorn et al. 2007; Johansson et al. 2010; Aniya and Imaizumi 2011; Schaffert 2011).

Deletion of Mgst1, in contrast to deletion of cytosolic GSTs, is embryonic lethal in mice (Bräutigam et al. 2018). (Mgst2-deficient mice breed normally and are indistinguishable from wild-type mice (Dvash et al. 2015).) MGST1 is highly expressed in hematopoietic stem and progenitor cells (HPSC). In HPSC from wild-type mice, transfection with MGST1 shRNA suppresses the expression of MGST1 and the formation of colony-forming units, decreases mitochondrial activity, and increases the expression of genes associated with glycolysis, HIF1 signaling, and NFκB. The authors conclude that impaired hematopoiesis plays a role in the embryonic lethality of Mgst1-null mice.

3. γ-Glutamyltransferase (GGT)⁶

3.1. Introduction

Mammalian GGTs catalyze the hydrolysis and transfer of the γ-Glu moiety of GSH, GSH S-conjugates, and γ-Glu amides to amino acid or peptide acceptors or to water (for reviews, see Taniguchi and Ikeda 1998; Keillor et al. 2005; Cooper and Hanigan 2010; Castellano and Merlino 2012; Castellano and Merlino 2013). Although 13 human genes encode GGT sequences, only GGT1 and GGT5 encode catalytically active proteins (Heisterkamp et al. 2008); the genes encoding GGT1 and GGT5 map to chromosome 22q11.23 (NCBI Gene ID: 2678 and 2687). Rodents have orthologs of only four of the human GGT genes, including GGT1, GGT5, GGT6, and GGT7 (Heisterkamp et al. 2008).

GGTs catalyze three reactions: transpeptidation, autotranspeptidation, and hydrolysis (Figure 18) (Tate and Meister 1981). The terms “transpeptidation” and “γ-glutamyl transpeptidase” were first used by Fruton and coworkers (Fruton 1950; Fruton et al. 1951) and by Hanes et al. (1950, 1952). Interestingly, Bergmann and Fraenkel-Conrat (1937) had earlier proposed that proteolytic enzymes could also catalyze transamidation reactions leading to peptide synthesis. Hanes et al. (1952) clearly demonstrated the transfer of the γ-glutamyl moiety of GSH to acceptor amino acids.

Figure 18. γ-Glutamyltransferase-catalyzed transpeptidation, autotranspeptidation, and hydrolysis reactions.

3.2. γ-Glutamyltransferase 1 (GGT1; EC 2.3.2.2) (for GGT5, see Section 3.3 γ-Glutamyltransferase 5 below)

3.2.1. Structure and properties

Mammalian GGT1 is assigned to peptidase Family T3 (MEROPS ID T03.006) (Rawlings et al. 2018). GGT2 is an inactive human form assigned to peptidase Family T3 (MEROPS ID T03.015); GGT2 encodes a propeptide that fails to undergo autocleavage into heterodimers (West et al. 2013b).

GGT1 is a heterodimeric, glycosylated protein that catalyzes the biotransformation of γ-Glu compounds. GGT1 is a type II transmembrane protein and a member of the N-terminal nucleophile aminohydrolases (Ntn-hydrolase) superfamily (Brannigan et al. 1995; Inoue et al. 2000; Oinonen and Rouvinen 2000; Kinlough et al. 2005).

Translation of human GGT1 affords a propeptide (569 amino acids) that undergoes autocatalytic cleavage to give large (amino acids 1–380, 42 kDa) and small (amino acids 381-569, 20 kDa) subunits (Castonguay et al. 2007; Heisterkamp et al. 2008; West et al. 2011; Pica et al. 2016). Human GGT1 is anchored to the extracellular face of the plasma membrane by a N-terminal domain (amino acids 5–26 in hGGT1) and a cytoplasmic domain of four polar amino acids (Kinlough et al. 2005; West et al. 2013a).

The active-site nucleophile responsible for the autocatalytic cleavage of the GGT propeptide and substrate γ-Glu compounds has been identified as a Thr residue (Inoue et al. 2000; Kinlough et al. 2005). In human GGT1, the active-site nucleophile was identified as Thr381 by use of Hiratake’s phosphonate inhibitor (226, Figure 20; see Section 3.2.4, Inhibitors below) and mass spectrometric analysis as well as crystallographic analysis (Han et al. 2006; Castonguay et al. 2007; West et al. 2013a). In the rat, autocatalytic cleavage of the propeptide occurs at Thr380 (Kinlough et al. 2005). Rat and mouse GGT1 are the products of a single gene, Ggt1 (Rajagopalan et al. 1990; Shi et al. 1995; Chikhi et al. 1999).

Catalytically competent GGT1 is a membrane-bound heterodimer (Tate and Meister 1976). With hGGT1, N-glycosylation is required for maturation of the propeptide. The propeptide may be converted into two subunits either before migration to the cell surface (Nash and Tate 1982) or after transport to the brush-border membrane (Kuno et al. 1983). Rat GGT propeptide undergoes glycosylation in the Golgi apparatus (Barouki et al. 1984). Human GGT1 contains seven glycosylation sites, six in the large subunit and one in the small subunit (Castonguay et al. 2007; West et al. 2011). The crystal structure of hGGT1 revealed posttranslational modifications, e.g. N-glycosylation sites, disulfide bonds, and active-site configurations, that differ between prokaryotic GGT and hGGT1 (West et al. 2013a). Rat GGT1 also contains six potential glycosylation sites in the large subunit and one in the small subunit (Sakamuro et al. 1988; Goodspeed et al. 1989). Differing degrees of glycosylation may account for the reported range of molecular masses among preparations and species (Goodspeed et al. 1989).

Rat renal GGT1 is anchored to the extracellular side of the brush-border membrane by the glycosylated large subunit (Horiuchi et al. 1978). The large subunit of rat renal GGT1 contains a membrane-spanning domain with a hydrophobic sequence near the amino terminus, which resides on the cytoplasmic side of the membrane (Tate and Nash 1987); the protein lacks a cleavable signal sequence, and translocation and insertion are dependent on a signal recognition particle. The Cys residues in rat GGT1 do not play a role in linking the two subunits (Kinlough et al. 2005).

The crystal structure of the soluble ectodomain of deglycosylated, Glu-bound hGGT1 confirms its heterodimeric (small subunit, residues 28–380; large subunit, residues 381–569) structure (West et al. 2013a). Access to the active site is located on the small subunit with the active-site Thr381 located deep in the substrate binding site. The “lid loop” that guards the active site in prokaryotic GGT is positioned differently in hGGT1 and allows substrate access. Chloride ions bound near Thr381 may contribute to catalysis. Further studies show that the structures of apo- and Glu-bound hGGT1 are similar (Terzyan et al. 2015). These studies show that the product Glu forms hydrogen bonds with the hydroxyl oxygen of Thr381 and with the main-chain Gly474. Although small-subunit residues comprise most of the ligand-binding residues in the active site, a conserved large-subunit residue, Arg107, forms a hydrogen bond with the 5-carboxylate of l-Glu in the hGGT1–Glu complex.

3.2.2. Reactions catalyzed and substrates

GGT1 catalyzes three types of reactions: one, a transpeptidation reaction in which the γ-Glu group of γ-Glu amides is transferred to amino-acid or peptide acceptors; two, an autotranspeptidation reaction in which the γ-Glu group of GSH is transferred to GSH or to GSH S-conjugates; and, three, hydrolysis of γ-Glu amides in which water serves as the acceptor resulting in the formation of Glu (Figure 18).

The mechanism of γ-Glu transfer or hydrolytic reactions catalyzed by GGT1 is formally similar to the mechanism of the autocleavage of the propeptide that gives rise to the large and small subunits. The active-site Thr (Thr381 in hGGT1) attacks the γ-Glu amide carbonyl of the donor substrate to give a reversible, unstable tetrahedral adduct that converts to an acyl-enzyme intermediate; the acyl-enzyme reacts with water to give L-Glu or with an amino acid or peptide acceptor substrate to give a l-γ-Glu amino acid or peptide (Figure 18). Crystallographic studies with hGGT1 demonstrate that the Ser-borate complex in the active site of hGGT1 mimics the tetrahedral intermediate formed in the catalytic cycle of hGGT1 (Terzyan et al. 2015).

The concept that an acyl-enzyme intermediate is involved in the catalytic cycle of mammalian GGT1 was first proposed by Tate and Meister (1974b, 1974a). Support for this concept was provided by the finding that incubation of [glu-¹⁴C]GSH and [gly-³H]GSH with rat kidney GGT1 led to greater binding of ¹⁴C to GGT1 than of ³H, indicating the formation of an acyl-enzyme intermediate (Elce and Broxmeyer 1976). Kinetic studies are supportive of this proposed mechanism (Allison 1985). An X-ray crystal structure of the γ-glutamyl-enzyme intermediate formed by E. coli GGT, which contains active-site binding residues and a nucleophilic Thr that are conserved in rat and human GGTs, is available (Okada et al. 2006). Kinetic studies with hGGT indicate that the acylation reaction is rate limiting (Castonguay et al. 2007).

3.2.2.1. Transpeptidation

Substrate selectivity studies with rat kidney GGT1 show that recognition of the α-amino group and the pK_a of the free proton of the γ-peptide bond are critical determinants of donor substrate selectivity (Cook et al. 1987). Both l- and d-γ-Glu compounds are substrates (Tate and Meister 1985); alkyl groups in the side-chain of the γ-Glu moiety are either poorly tolerated or not tolerated (Griffith and Meister 1977), and replacement of the 4-methylene of the γ-Glu moiety with heteroatoms yields inactive donor substrates (Lherbet et al. 2003). Moreover, γ-Glu-4-nitrophenyl ester, the ester analog of the widely used assay substrate γ-Glu-4-nitroanilide (196, Figure 19), is not a substrate (Castonguay et al. 2002).

Figure 19. GGT substrates: 196, γ-Glu-4-nitroanilide; 197, 4-nitroaniline; 198, S-(4-nitrobenzyl)-l-Cys-Gly; 199, S-(4-methylbiphenylyl)GSH; 200, S-(4-methylbiphenylyl)-l-Cys-Gly; 201, clofibryl-S-acyl-GSH (l-γ-Glu-S-[2-(4-chlorophenoxy)-2-methyl-1-oxopropyl]-l-Cys-Gly); 202, clofibryl-S-acyl-l-Cys-Gly (S-[2-(4-chlorophenoxy)-2-methyl-1-oxopropyl]-l-Cys-Gly); 203, 8,9-dihydro-8-(S-glutathionyl)-9-hydroxyaflatoxin B₁; 204, 8,9-dihydro-8-(S-l-Cys-Gly)-9-hydroxyaflatoxin B₁; 205, l-γ-Glu-S-[(7R,8S,9S,10S)-7,8,9,10-tetrahydro-7,8,9-trihydroxybenzo[a]pyren-10-yl]-l-Cys-Gly ((-)-anti-BPD GSH conjugate); 206, S-[(7R,8S,9S,10S)-7,8,9,10-tetrahydro-7,8,9-trihydroxybenzo[a]pyren-10-yl]-l-Cys-Gly ((−)-anti-BPD l-Cys-Gly conjugate); 207, l-γ-Glu-S-[2-hydroxy-1-(2-oxoethyl)heptyl]- l-Cys-Gly (4-HNE GSH conjugate); 208, S-[2-hydroxy-1-(2-oxoethyl)heptyl]-l-Cys-Gly (4-HNE l-Cys-Gly conjugate); 209, γ-Glu-d,l2-(phenylthiol)Gly-Gly; 210, 2-(phenylthio)Gly-Gly ((2-amino-2-(phenylthio)acetyl)glycine); 211, l-γ-Glu-S-[2-[[4-(dihydroxyarsino)phenyl]amino]-2-oxoethyl]-l-Cys-Gly (4-(N-(S-glutathionylacetyl)amino)phenylarsonous acid; GSAO); 212, S-[2-[[4-(dihydroxyarsino)phenyl]amino]-2-oxoethyl]-l-Cys-Gly (4-(S-cysteinylacetyl)amino)phenylarsonous acid); 213, N-(γ-l-Glu)cysteamine; 214, cysteamine; 215, N⁵-(6-(benzo[d]thiazol-2-yl)naphthalen-2-yl)Gln; 216, 6-benzo[d]thiazol-2′-yl)-2-aminonaphthalene.

An investigation of the rates of transpeptidation of a series of donor GSH S-conjugates of aromatic compounds showed that the catalytic efficiency varied by a factor of four with bovine kidney GGT (Agblor and Josephy 2013). The GSH analog ophthalmic acid (l-γ-Glu-l-α-aminobutyryl-Gly) and several S-substituted GSH analogs also serve as donor substrates for rat kidney GGT (Tate and Meister 1974a); GSSG and γ-(dl-α-methylglutamyl)-l-Cys-Gly are poor donor substrates with Met as the acceptor.

Several amino acids and peptides serve as acceptor substrates for rat kidney GGT (for a compilation of donor and acceptor substrates, see Tate and Meister 1974a). With l-γ-Glu-4-nitroanilide (196, Figure 19) as the γ-Glu donor, l-cystine, l-Gln, and l-Met are the best amino-acid acceptors and Gly–Gly is the best peptide acceptor substrate (Tate and Meister 1974a; Thompson and Meister 1977). Rat GGT shows high selectivity for l-configuration amino acids or peptides as acceptor substrates but accepts donor substrates with both the l- and l-configuration (Tate and Meister 1974a; Thompson and Meister 1976, 1977).

Meister and coworkers (Orlowski and Meister 1970; Meister 1973) proposed a role for GGT in amino-acid transport: GGT could serve as a recognition site for amino acids, which could react with GGT-bound Glu; the γ-Glu amino acids would then be translocated into the cell. This proposal was controversial (Bell et al. 1987; Sastre et al. 1991; Cotgreave and Schuppe-Koistinen 1994), and current evidence does not support a role for GGT in amino-acid transport (Inoue 2016).

3.2.2.2. Autotranspeptidation

GSH and GSH S-conjugates serve as both donor and acceptor substrates; hence, the term autotranspeptidation. GGT-Catalyzed autotranspeptidation of GSH appears to be a minor reaction, particularly in the presence of alternative acceptor substrates (Elce and Broxmeyer 1976; Abbott et al. 1986). GGT catalyzes the autotranspeptidation of S-(methyl)GSH to γ-Glu-S-(methyl)GSH, albeit at a low rate (Tate and Meister 1974b). γ-Glu-GSH is formed in the isolated rat kidney perfused with GSH (Abbott et al. 1986). Although most information about autotranspeptidation reactions has been derived from in vitro studies, autotranspeptidation does occur in vivo: γ-Glu-GSH has been identified in the urine of human subjects with hereditary GGT deficiency and in the urine of mice given a GGT inhibitor (Abbott et al. 1986; Oe et al. 1998). In addition to 3-(glutathion-S-yl)acetaminophen, a known metabolite of acetaminophen, 3-(γ-Glu-glutathion-S-yl)acetaminophen and 3-(γ-Glu-γ-Glu-glutathion-S-yl)acetaminophen have been identified as minor biliary metabolites of acetaminophen in rats (Mutlib et al. 2000). (γ-Glu)_n-4-nitroanilide (n = 2, 3, and 4) has been identified in in vitro studies with GGT (Orlowski and Meister 1965); these metabolites are presumably formed by an autotranspeptidation reaction. With rat kidney GGT, autotranspeptidation is not observed with d-γ-Glu-4-nitroanilide as the substrate or when the substrate (l-γ-Glu-4-nitroanilide, 196, Figure 19) concentration is <10 μM (Thompson and Meister 1976).

3.2.2.3. Hydrolysis

The hydrolysis of γ-Glu compounds is typically observed along with transpeptidation and autotranspeptidation. The relative rates of rat renal GGT-catalyzed transpeptidation and hydrolysis are dependent on the concentration of the acceptor (Elce and Broxmeyer 1976). With GSH (50 μM) as the donor substrate and Ala (0.5 mM) as the acceptor substrate, rat kidney GGT-catalyzed hydrolysis is greater than transpeptidation from pH 6.0 to 9.0 but decreases above pH 8.5 (McIntyre and Curthoys 1979); the rate of transpeptidation increases from pH 6.5 to 9.0. Other studies with l-Glu-4-nitroanilide (196, Figure 19) as the donor substrate and a mixture of amino acids (3.11 mM) as acceptor substrates show that at pH 7.5, the rates of rat kidney GGT-catalyzed transpeptidation and hydrolysis are similar (Allison and Meister 1981); with GSH (50 μM) as the donor substrate and Ala (0.5 mM) as the acceptor substrate (pH 8.0), approximately 12% of the GSH undergoes transpeptidation. d-γ-Glu-4-nitroanilide undergoes hydrolysis, but not autotranspeptidation; at high substrate concentrations (>10 μM l-γ-Glu-4-nitroanilide, 196, Figure 19) autotranspeptidation predominates, but at lower concentrations, only hydrolysis occurs in the absence of acceptors (Thompson and Meister 1976). With rat kidney GGT, the relative contributions of transpeptidation, autotranspeptidation, and hydrolysis at saturating donor (l-γ-Glu-4-nitroanilide, 196, Figure 19) and acceptor substrate concentrations (Gly–Gly) are 77–80%, 19–22%, and 0.5–1%, respectively; in the absence of an acceptor substrate, the relative contributions of autotranspeptidation and hydrolysis are 96–98% and 2–4%, respectively (Dvoráková et al. 1996). The conclusion to be drawn from the above-mentioned studies is that the relative rates of hydrolysis and transpeptidation are highly dependent on reaction conditions but that both reactions occur in vivo.

3.2.3. Substrates

GSH S-conjugate formation is the first step in the mercapturic acid pathway and yields GSH S-conjugates that may serve as substrates for GGT. Although many GSH S-conjugates are biotransformed to mercapturic acids and are, therefore, implicitly biotransformed by GGT, few systematic structure–activity studies of GSH S-conjugates as donor substrates for GGT1 are available. It is well established, however, that a range of structurally diverse GSH S-conjugates are biotransformed by GGT1 (Moss et al. 1985; Srivastava et al. 1999a; Grillo and Benet 2001; Enoiu et al. 2003; Agblor and Josephy 2013; Grillo et al. 2013). A selected representation of that structural diversity includes S-(4-nitrobenzyl)GSH (126, Figure 10), and structures 199, 201, 203, 205, and 207 (Figure 19). Several aza- and oxaglutamyl-4-nitroanilides have been tested as substrates but all are poorer substrates than γ-Glu-4-nitroanilide (196, Figure 19) (Lherbet et al. 2003).

Studies on GGT-catalyzed hydrolytic reactions are confounded by concurrent transpeptidation reactions. To circumvent this problem and to develop an assay suitable for kinetic analyses of potential GGT inhibitors, the donor substrate γ-Glu-d,l-2-(phenylthiol)Gly–Gly (209, Figure 19), which, in the absence of an acceptor substrate, only undergoes a GGT-catalyzed hydrolytic reaction to give 2-(phenylthio)Gly–Gly (210, Figure 19) was prepared (Vergauwen et al. 2009); the 2-(phenylthio)Gly–Gly (210, Figure 19) formed undergoes chemical hydrolysis to give thiophenol, which can be detected by reaction with Ellman’s reagent. The kinetic constants for the hydrolysis of the GSH mimic (209, Figure 19) by equine kidney GGT are similar to those of the hydrolysis of GSH.

There is historical and contemporary interest in exploiting the substrate selectivity of GGT to catalyze the bioactivation of prodrugs (Magnan et al. 1982; Vamvakas and Anders 1994; Keren and Stark 1998; Ramsay and Dilda 2014). Several studies on GGT-activated prodrugs have focused on chemotherapeutic arsenicals. The trivalent arsenical prodrug 4-(N-(S-glutathionylacetyl)amino)phenylarsonous acid (GSAO, 211, Figure 19) was prepared as a potential angiogenesis inhibitor and was shown to inhibit tumor growth in mice (Don et al. 2003); the pentavalent arsenic analog was inactive. GSAO (211, Figure 19) is biotransformed by GGT to give Cys-Gly S-conjugate 212 (Figure 19), which is transported into cells by OATP, where it may be converted to the Cys S-conjugate by dipeptidases (Dilda et al. 2008). A phase 1 trial (clinicaltrials.gov, NCT01147029) of GSAO (211, Figure 19) indicated that the drug was well tolerated (Horsley et al. 2013). The arsenical prodrug S-(dimethylarsino)glutathione (darinaparsin) is biotransformed by GGT and dipeptidases to S-(dimethylarsino)-l-Cys, which is transported into cells by a cysteine/cystine transporter (Garnier et al. 2014).

Cystinosis is a rare autosomal recessive disorder in which cystine accumulates in lysosomes. Although cysteamine (214, Figure 19) is used for the clinical management of cystinosis, its adverse effects limit patient compliance. Hence, the cysteamine (214, Figure 19) prodrug N-(γ-l-Glu)cysteamine (213, Figure 19) was prepared to attempt to overcome these limitations (Anderson et al. 2006); prodrug 213 (Figure 19) and several of its analogs reduce intracellular cystine concentrations in cultured cystinotic cells. The S-acetyl, ethyl ester analog of prodrug 213 (Figure 19) and four structurally related prodrugs are biotransformed to cysteamine (214, Figure 19) in human pT37 cells (Frost et al. 2016).

There has been considerable interest in developing substrates that allow visualization of GGT activity in cells and in intact mice (for a review, see Luo et al. 2018a). The capability of GGT to utilize N⁵-Substituted l-glutamines, e.g. γ-Glu-4-nitroanilide (196, Figure 19) as substrates has been exploited in the design of a variety of GGT-activated probes. Examples of three classes of imaging probes, most of which are γ-Glu amides that are activated by GGT, have been reported: (1) probes that release a fluorescent product or precursor (Iwatate et al. 2018; Li et al. 2018; Ou-Yang et al. 2018; Liu et al. 2018a; Luo et al. 2018b); (2) probes that release a ratiometric fluorescent product (Liu et al. 2018b); and (3) probes that release bioluminescent products (Hai et al. 2017). GGT-activated fluorescent probes for imaging intracellular GGT that are suitable for one-photon confocal microscopy have been described but technical issues limit their applicability (Wang et al. 2017b, and references cited therein). N⁵-(6-(Benzo[d]thiazol-2-yl)naphthalen-2-yl)glutamine (215, Figure 19), which is biotransformed by GGT to 2-aminonaphthalene 216 (Figure 19), has been developed as a probe suitable for two-photon microscopy (Wang et al. 2017b). γ-N-Substituted Gln 215 (Figure 19) was successfully used to image GGT in HepG2 cells, in normal rat liver slices, and in mouse liver cancer tissue slices.

3.2.4. Inhibitors

Inhibitors have played a central role in elucidating the function of GGT in GSH metabolism, in amino-acid transport, and in drug resistance in tumor cells and are of interest as potential therapeutic agents (Ahluwalia et al. 1990; Hanigan 2014). The concept that GSH plays a role in tumor resistance to therapy and that GGT may serve to provide l-Cys for GSH synthesis has led to a search for GGT inhibitors with lowered toxicity (Pompella et al. 2006; Corti et al. 2010; Wickham et al. 2013). GGT is, however, variably expressed in tumor tissue: for example, GGT expression is not observed in basal cell carcinomas, Hodgkin’s disease, mesotheliomas, melanomas, non-Hodgkin’s lymphomas, and most soft tissue sarcomas, but epithelium-derived carcinomas retain their GGT-positive phenotype (Hanigan et al. 1999).

3.2.4.1. Reversible inhibitors

Early studies showed that the GGT-catalyzed biotransformation of GSH was lower in the presence of l-Ser and borate buffer than in the presence of l-Ser and Tris or Veronal (barbital) buffers (Revel and Ball 1959). Other work showed that l-Ser and borate was a competitive inhibitor of GGT (Szewczuk and Connell 1965). These studies were extended and showed that both d- and l-Ser in the presence of borate inhibited the transpeptidation of l-γ-Glu-4-nitroanilide (196, Figure 19) (Tate and Meister 1978); borate and α-methyl-dl-Ser or l-Hse were weak inhibitors of GGT. Binding studies showed that 1 mol of Ser was bound per mol of enzyme; inhibition of GGT by Ser-borate was reversed by dilution or dialysis. Given that borate forms reversible complexes with vicinal hydroxyl groups, the authors proposed that Ser-borate reacts with an active-site hydroxyl group, e.g. Ser, to form an intermediate that resembles the expected tetrahedral transition-state intermediate formed by transpeptidation or hydrolysis of the acyl-enzyme intermediate. It should be noted that a 1:1 mixture of Ser and borate is used in inhibition experiments rather than ester 217 (Figure 20); crystallographic analysis of the complex formed between hGGT1 and Ser-borate supports the formation of ester 217 (Figure 20) (Terzyan et al. 2015).

Figure 20. GGT inhibitors. 217, L-Ser-borate complex (O-(dihydroxyboryl)Ser); 218, 3-amino-3-carboxypropaneboronic acid (γ-boroGlu); 219, 1-γ-l-Glu-2-(2-carboxyphenyl)hydrazide (anthglutin); 220, 2-amino-4-[2-(carboxymethylcarbamoyl)ethylsulfinyl]butyric acid; 221, N-(5-(4-methoxybenzyl)-1,3,4-thiadiazol-2-yl)benzenesulfonamide (OU749); 222, O-(2-diazoacetyl)- l-Ser (azaserine); 223, 6-diazo-5-oxo-l-Nle (DON); 224, (αS,5S)-α-amino-3-chloro-4,5-dihydro-5-isoxazoleacetic acid (acivicin, AT-125); 225, 2-amino-4-[methyl(4-methylumbelliferyl)phosphono]butanoic acid; 226, 2-amino-4-[mono(4-cyanophenyl)phosphono]butanoic acid (Hiratake′s Phosphonate); 227, 2-amino-4-{[4-(carboxymethyl)phenyl]-(methyl)phosphono}butanoic acid (GGsTop); 228, (2 R,S)-2-amino-4-{(S)-1-[N-(carboxymethyl)carbamoyl]propyl(phenyl)-(S_p*)-phosphono}butanoic acid.

In an attempt to develop more potent inhibitors of GGT than Ser-borate, several arylboronic acids were prepared and tested; two of the monophenyl boronic acids were slightly more potent than Ser-borate (Suenaga et al. 1996). The γ-boronic acid analog of L-Glu, 3-amino-3-carboxypropaneboronic acid (γ-boroGlu, 218, Figure 20), is a potent, slow-binding inhibitor of pig kidney GGT (Stein et al. 2001).

Several γ-Glu hydrazones and hydrazides are reversible inhibitors of GGT. The γ-Glu hydrazones of several α-keto acids inhibit rat kidney GGT, but do not serve as donor or acceptor substrates; the most potent inhibitor of this series is the γ-Glu hydrazone of α-ketoglutarate (K_i = 0.4 mM) (Tate and Meister 1974a). Anthglutin (219, Figure 20), isolated from the culture medium of Penicillium oxalicum, inhibits GGT (Kinoshita and Minato 1978; Minato 1979). Both l- and d-anthglutin (219, Figure 20) inhibit rat kidney GGT (Griffith and Meister 1979). Although both l- and d-anthglutin (219, Figure 20) inhibit GGT in vivo in mice, the D isomer is less toxic, perhaps because its hydrolysis to 2-carboxyphenylhydrazine is slower than that of the l isomer or because d-anthglutin interacts with fewer enzymes than l-anthglutin (Griffith and Meister 1979; Tate and Meister 1981).

Studies on the inhibition of GGT by l-Glu analogs showed that l-Met sulfoxide is a competitive inhibitor of GGT (Lherbet and Keillor 2004). Further investigation of l-Met sulfoxide analogs led to the preparation of sulfoxide-bearing GSH analog 220 (Figure 20), which proved to be a reversible and potent (K_i = 0.053 mM) competitive inhibitor of rat kidney GGT. Further studies showed that the S_CS_S diasteroisomer of sulfoxide 220 (Figure 20) interacts with GGT (Lherbet et al. 2004). The sulfide analog of sulfoxide 220 (Figure 20) is approximately tenfold less potent (K_i = 0.52 mM) than sulfoxide 220 (Figure 20) whereas the sulfone analog is an even weaker inhibitor (K_i = 3.5 mM) of GGT.

OU749 (221, Figure 20) inhibits both GGT1-catalyzed transpeptidation and hydrolytic reactions (King et al. 2009; Wickham et al. 2012). Kinetic analysis showed that OU749 (221, Figure 20) exhibits similar potency as an uncompetitive inhibitor of both the transpeptidation and hydrolytic reactions (Wickham et al. 2012). Interestingly, OU749 (221, Figure 20) is a more potent inhibitor of isolated human kidney GGT than of rat or mouse kidney GGT and is a more potent inhibitor of GGT in human kidney 786-O and HK-2 cells than in rat NRK-52E, pig LLC-PK1, or monkey LLC-MK2 cells (King et al. 2009); OU749 (221, Figure 20) is 150-fold less cytotoxic than acivicin (224, Figure 20) to human kidney 786-O cells. Substitution of the benzenesulfonamide moiety produced divergent effects, increasing or reducing the potency for inhibition of the transpeptidation reaction, whereas substitution with electron withdrawing groups, e.g. nitro or trifluoromethyl, activated the GGT-catalyzed hydrolysis reaction (Wickham et al. 2012). The inhibitory selectivity of OU749 (221, Figure 20) for hGGT may be associated with Ala472 in the acceptor binding site compared with Ser472 in the rat, mouse, or pig (Hu et al. 2012).

3.2.4.2. Irreversible inhibitors

Azaserine (222, Figure 20) inhibits a number of glutamine-transforming enzymes, including rat and human kidney GGT, which are inactivated by compound 222 (Tate and Meister 1977; Tate and Ross 1977). Both transpeptidation and hydrolytic reactions are inhibited by azaserine (222, Figure 20); inactivation is blocked by donor, but not acceptor, substrates. The crystal structure of azaserine-inactivated E. coli GGT has been solved and shows that azaserine (222, Figure 20) forms an adduct with the active-site Thr (Wada et al. 2008). The crystal structure of azaserine-inactivated hGGT has apparently not been reported.

Although not an enzyme-specific inhibitor of GGT (Ahluwalia et al. 1990), 6-diazo-5-oxo-l-Nle (DON; 223, Figure 20) inactivates rat and human kidney GGT when incubated in the absence of a donor substrate (Inoue et al. 1977; Tate and Meister 1977; Tate and Ross 1977); incubation with the donor substrates γ-Glu-4-nitroanilide (196, Figure 19), S-(methyl)GSH and l-Gln blocks DON-induced inactivation of GGT, but the acceptor substrates L-Met and Gly-Gly fail to block inactivation. As with azaserine (222, Figure 20), both transpeptidation and hydrolytic reactions are blocked by DON (223, Figure 20). The pseudo-first-order rate constant for the inactivation of human GGT1 is approximately 0.13 min⁻¹ (Terzyan et al. 2017). When rat kidney GGT was incubated with [6-¹⁴C]DON, 1 mole of inhibitor was covalently linked per mole of GGT (Inoue et al. 1977). Maleic acid markedly increases the rate of inactivation of rat kidney GGT by DON (Inoue et al. 1977). The dipeptide analog of DON (223, Figure 20), 6-diazo-5-oxo-l-Nle-Gly, inactivates renal membrane-bound GGT but not purified GGT, indicating that hydrolysis of the dipeptide to give DON (223, Figure 20) is required for inactivation (Inoue and Morino 1981).

The crystal structure of DON-inactivated and covalently modified hGGT1 has been solved (Terzyan et al. 2017). The proposed mechanism of inactivation involves attack of the hydroxyl oxygen of Thr381 on the side-chain carbonyl carbon of DON (223, Figure 20) to give a tetrahedral intermediate followed by reaction of the α-amino group of Thr381 on C6 of DON (223, Figure 20) to form a protein-bound morpholine ring that contains the α-amino, Cα, Cβ, and Oγ atoms of Thr381 and C5 and C6 of DON (223, Figure 20).

Acivicin (AT-125; 224, Figure 20), an irreversible inhibitor of GGT, is a fermentation product of Streptomyces sviceus (Martin et al. 1973). Several early studies showed that acivicin (224, Figure 20) inhibits mammalian GGT (Allen et al. 1980; Gardell and Tate 1980; Kozak and Tate 1980; Reed et al. 1980). Studies also showed that acivicin (224, Figure 20) exhibited antitumor properties in tumor-derived cell lines and in experimental animals (Allen et al. 1980; Thornthwaite and Allen 1980; Poster et al. 1981; Lui et al. 1982; Weber et al. 1982). Accordingly, acivicin (224, Figure 20) was tested for antitumor activity in human clinical trials (see, for example: Kramer et al. 1986; McGuire et al. 1986; Earhart et al. 1989; Falkson et al. 1990); the outcomes were generally unfavorable with acivicin (224, Figure 20) showing considerable toxicity and poor tumor response. An ester-based prodrug of acivicin (224, Figure 20) that is less cytotoxic than acivicin (224, Figure 20) has been reported (Antczak et al. 2001).

Acivicin (224, Figure 20) stoichiometrically and covalently reacts with purified rat kidney GGT and is a more effective inhibitor of the enzyme than DON (Gardell and Tate 1980); l-Ser-borate, a transition-state inhibitor of GGT, protects against acivicin-induced inhibition. Although acivicin (224, Figure 20) inhibits several enzymes that utilize L-Gln as a substrate (Ahluwalia et al. 1990; Pinto et al. 2016), only GGT appears to be covalently modified by [2-³H]acivicin in rat renal brush-border membranes (Kozak and Tate 1980). Mutations S451A and S452A caused a drastic reduction in the acivicin-induced hGGT inactivation rates; Ser451 and Ser452 were suggested to be potential active-site nucleophiles (Ikeda et al. 1995). Thr381 has, however, been identified unequivocally as the active-site nucleophile in hGGT (Castonguay et al. 2007).

Incubation of acivicin-inhibited GGT with hydroxylamine leads to the release of threo-β-hydroxy-l-γ-Glu hydroxamate, which was interpreted as evidence for the displacement of chloride from acivicin (224, Figure 20) by a nucleophilic site on GGT (Stole et al. 1994; Smith et al. 1995). The crystal structures of acivicin-inactivated E. coli and H. pylori GGT have been solved, and both structures show that the adduct with the active site Thr is not an ester, as suggested in an earlier report (Smith et al. 1995). Rather, following displacement of chloride by the active-site Thr, the dihydroisoxazole ring of acivicin (224, Figure 20) remains bound to the nucleophilic oxygen of the Thr and is retained in the adduct (Wada et al. 2008; Williams et al. 2009). Other 3-halo-4,5-dihydroisoxazoles also undergo halide displacement by active-site nucleophiles, which results in adducts in which the nucleophile is directly linked to the dihydroisoxazole ring (Watts et al. 2006; Pinto et al. 2016).

Phosphono derivatives of Glu have been prepared as mechanism-based affinity labeling probes (Inoue et al. 2000; Han et al. 2006, 2007). Indeed, phosphonate diester 225 (Figure 20) is 6000-times more potent than acivicin (224, Figure 20) as an inactivator of human GGT1 (Han et al. 2007). Phosphonate 226 (Figure 20) was used with expressed hGGT and mass spectrometric analysis to provide unambiguous identification of Thr381 as the active-site nucleophile of human GGT (Castonguay et al. 2007).

Phosphonobutanoic acid 227 (GGsTop, Figure 20) is more potent and rapid inactivator of hGGT (k_on = 57 M⁻¹ s⁻¹) than acivicin (224, Figure 20) (k_on = 0.40 M⁻¹ s⁻¹) and fails to inhibit glutamine amidotransferases (Han et al. 2007; Kamiyama et al. 2016). The authors propose that the meta-carboxyl group of the phenylacetic acid moiety of GGsTop (227, Figure 20) interacts with Lys562, the recognition site for the C-terminal carboxyl group of Cys-Gly. The mechanism of inactivation of hGGT by GGsTop is similar to that of the other phosphonate inhibitors in that it involves nucleophilic attack of Thr381 on the phosphorous atom with the resulting loss of 3-hydroxyphenylacetic acid; an X-ray crystal structure of the covalently bound enzyme-inhibitor complex is available (Terzyan et al. 2015). Of the four possible stereoisomers of GGsTop (227, Figure 20), the l-S_p isomer is the most potent (Watanabe et al. 2017). Preliminary data show that GGsTop (227, Figure 20) lacks acute toxicity in rats given 100 mg/kg i.v., lacks cytotoxicity in human skin fibroblasts and rat hepatic stellate cells (≤1 mM), and is stable in solution (Yamamoto et al. 2011; Kamiyama et al. 2016). Hence, GGsTop (227, Figure 20) appears to lack the toxicity associated with acivicin (224, Figure 20), but further studies are needed to assess more fully its toxicity.

In a rat model of renal ischemia-reperfusion injury, GGsTop (227, Figure 20) attenuated acute renal injury at single doses of 1 and 10 mg/kg i.v., perhaps by preventing GSH degradation (Yamamoto et al. 2011). Similarly, GGsTop (227, Figure 20) protects against hepatic ischemic-reperfusion injury in rats, again perhaps by maintaining cellular GSH concentrations (Tamura et al. 2016). GGsTop (227, Figure 20) also protects GSH degradation in lung lining fluid, indicating a potential use in management of asthma (Joyce-Brady and Hiratake 2011; Tuzova et al. 2014). The studies with GGsTop (227, Figure 20) and analogs were extended to include a series of peptidyl phosphonate ester analogs of GSH (Nakajima et al. 2014). Peptidyl phosphonate diester 228 (Figure 20) is a potent (k_on = 145 M⁻¹ s⁻¹), irreversible inactivator of hGGT1. Its potency is attributed to the leaving group properties of the phenoxide moiety.

3.2.5. Tissue and species distribution

Mammalian GGT activities differ markedly among tissues. In all species studied, GGT activities are highest in the kidney, typically much higher than in the liver (Goldbarg et al. 1960; Hinchman and Ballatori 1990). Ultrastructural studies show that GGT is located in rat proximal tubule microvilli, endocytic vacuoles, and basolateral membranes and in the brush border of the jejunal epithelium (Marathe et al. 1979; Spater et al. 1982). In the rat renal brush-border membrane, the active site of GGT faces the luminal side of the renal tubule (Horiuchi et al. 1978). Although hepatic GGT activities are lower than renal activities, GSH secreted into the bile is almost quantitatively degraded to its constituent amino acids in the bile duct (Ballatori et al. 1986, 1988). The histochemical distribution of GGT in rat and guinea pig hepatocytes has been investigated with an activity stain (Lanca and Israel 1991). In the rat, GGT activity is found on both the bile canalicular and sinusoidal sides of hepatocytes, whereas in the guinea pig, GGT activity is strongly localized to the sinusoid side of hepatocytes but also present in the bile canalicular epithelium. Comprehensive immunohistochemical studies show widespread distribution of GGT in normal human tissues, including, for example, the luminal surface of renal proximal tubular cells, but not the glomeruli or distal tubular cells; bronchus; salivary ducts; hepatic bile ducts and canaliculi; gall bladder; pancreas; and testes, epididymis, and seminal vesicles (Hanigan and Frierson 1996).

3.2.6. Induction

As with many other drug-metabolizing enzymes, GGT activity in vivo is increased by treatment of animals with carcinogens and other chemicals. There is little hepatic expression of GGT in normal rats. Hepatic activity of GGT is, however, markedly increased after administration of a single dose of 3′-methyl-4-dimethylaminoazobenzene (3-MAB) and is increased exponentially with chronic administration of 3-MAB (Fiala et al. 1972, 1976). Long-term (≤14 weeks) administration of aflatoxin B₁ to rats led to a marked increase in the hepatic expression of GGT well before carcinoma development, indicating that the expression of GGT may serve as a marker for preneoplastic hepatocytes (Kalengayi et al. 1975). Although it is often stated that GGT is not expressed in preneoplastic lesions in mice, several reports of carcinogen induction of GGT in preneoplastic mouse hepatocytes have been published (Goldsworthy et al. 1986; Hanigan et al. 1993). The hepatic GGT induced by carcinogen treatment differs from the renal enzyme in its sugar content (Tsuchida et al. 1979).

The mechanism of GGT induction by carcinogens has not been fully elucidated. Incubation of primary cultures of rat, but not mouse, hepatocytes in the absence of carcinogens leads to a marked increase in GGT expression (Gallagher et al. 1998). In the rat, the GGT protein precursor is encoded by mRNA_III that is selectively transcribed in the liver from a single copy gene (Brouillet et al. 1994). Only mRNA_III is induced indicating that Promoter III is associated with the observed induction by carcinogens (Gallagher et al. 1998).

The significance of GGT induction in tumor cells is that it may correlate with drug resistance (Hanigan 2014); as indicated above, this has led to a search for safe and effective GGT inhibitors.

3.2.7. GGT as a biomarker and as a risk and prognostic factor in disease states

The measurement of serum GGT activities has long been used as a diagnostic test for liver disease and injury (Orlowski 1963; Zein and Discombe 1970) and is considered to be a sensitive test for disease- or drug- and chemical-induced liver damage (Sheehan and Haythorn 1979).

A relationship between serum GGT activities and a range of human disease states, including cardiovascular disease (Ruttmann et al. 2005; Emdin et al. 2006; Ndrepepa et al. 2016; Wang et al. 2017a), atherosclerosis (Paolicchi et al. 2006), diabetes (Lee et al. 2003; Kim et al. 2005; Meisinger et al. 2005), Parkinson’s disease (Sian et al. 1994; Chinta et al. 2006), chronic renal disease (Targher et al. 2010), and cancer (Strasak et al. 2008; Grimm et al. 2013; Preyer et al. 2016) has been reported. Oxidative stress has been implicated in a range of human disease states, and it has been proposed that serum GGT activity is a marker for oxidative stress (Lee et al. 2004).

The mechanism of the release of GGT from tissues, the association between elevated serum GGT activities and tissue expression of GGT, and the nature of serum GGT have not been fully clarified. Given that colorimetric assays, e.g. γ-Glu-4-nitroanilide (196, Figure 19), are commonly used to quantify serum GGT activities, serum GGT appears to retain its catalytic activity. In chemically induced cholestasis in rats, serum GGT activity is of biliary-cell origin rather than from hepatocytes (Bulle et al. 1990).

Two forms of GGT are found in human serum (Huseby 1982): a major, amphiphilic form that is present in high molecular-mass complexes and a minor, low molecular-mass (80–90 kDa) hydrophilic form that appears to be a proteolytic degradation product of GGT. Fractionation of human serum GGT by gel-filtration chromatography showed four forms (Franzini et al. 2008): b-GGT (2000 kDa), m-GGT (940 kDa), s-GGT (140 kDa), and f-GGT (70 kDa); b-, m-, and s-GGT are bound to lipoproteins, and f-GGT is an unbound form. Other studies indicate that b-, m-, and s-GGT are present as complexes with VLDL, LDL, and HDL, respectively (Fornaciari et al. 2014); f-GGT is the most abundant form in healthy adults, and b-, m-, and s-GGT are increased in liver diseases. The mechanism of transfer of GGT from plasma membranes to blood has not been fully resolved, although b-GGT is found in exosomes, implicating an exocytotic mechanism. The GGT fractions may prove to be useful in the differential diagnosis of liver disease.

3.3. γ-Glutamyltransferase 5 (GGT5, EC 3.4.19.14, leukotriene-C₄ hydrolase, γ-glutamyl leukotrienase)

3.3.1. Structure and properties

Although GGT1 had been considered to be the only enzyme that catalyzes the hydrolysis of the γ-Glu bond of GSH and GSH S-conjugates, a human enzyme with about 40% amino acid similarity to hGGT1 catalyze the hydrolysis of LTC4 (189, Figure 16) and GSH, but not γ-Glu-4-nitroanilide (196, Figure 19) (Heisterkamp et al. 1991). This GGT-related enzyme was termed GGT-rel, but has also been termed GGTLA1 (Morris et al. 1993).

Tissue homogenates and partially purified enzyme from Ggt1-null mice catalyze the hydrolysis of leukotriene C₄ (189, Figure 16) and S-decyl-GSH, but not GSH, GSSG, γ-Glu-4-nitroanilide (196, Figure 19), or γ-Glu-4-methoxy-2-naphthylamide, indicating the presence of an enzyme other than mGTT1 that catalyzes the hydrolysis of the γ-Glu bond (Carter et al. 1997); this enzyme was termed γ-Glu leukotrienase (GGL). A rat homolog of GGT-rel with 79% identity to human GGT-rel that was weakly expressed in normal trachea has been described but the hydrolysis of γ-Glu substrates was not reported (Potdar et al. 1997).

GGT5 has been proposed as the designation for GGTs formerly termed GGT-rel, GGL, and GGTLA1 (Heisterkamp et al. 2008). GGT5 (leukotriene C₄ hydrolase) is assigned to peptidase Family T3 (MEROPS ID T03.002) (Rawlings et al. 2018).

All published reports indicate that like GGT1, GGT5 is a glycosylated, extracellular heterodimeric protein (Heisterkamp et al. 1991; Carter et al. 1997; Han et al. 2002).

3.3.2. Reactions catalyzed and substrates

3.3.2.1. Transpeptidation

With mouse GGT5, addition of Gly-Gly to the reaction mixture failed to increase the hydrolysis of LTC4 (189, Figure 16), indicating that mGGT5 may not catalyze transpeptidation reactions (Carter et al. 1997). Later studies showed, however, that hGGT5 catalyzes transpeptidation reactions: Gly-Gly decreases the GGT5-catalyzed release of l-Glu from GSH, indicating that, under nonphysiological conditions, hGGT5 can catalyze a transpeptidation reaction (Wickham et al. 2011).

3.3.2.2. Autotranspeptidation

GGT5-catalyzed autotranspeptidation reactions have apparently not been reported.

3.3.2.3. Hydrolysis

Given the physiological importance of leukotrienes (Peters-Golden and Henderson 2007), the enzymatic hydrolysis of LTC4 (189, Figure 16) to LTD4 has been the subject of many investigations. Hence, as indicated above, LTC4 hydrolytic activity assigned to GGT5 has been identified in human, mouse, and rat (Heisterkamp et al. 1991; Carter et al. 1997; Potdar et al. 1997). Furthermore, Ggt5-null mice do not catalyze the hydrolysis of LTC4 (189, Figure 16) (Han et al. 2002).

Although there is general agreement that GGT5 catalyzes the hydrolysis of LTC4 (189, Figure 16) and fails to catalyze the hydrolysis of γ-Glu-4-nitroanilide (196, Figure 19), differences in the substrate selectivity of GGT5 have been reported, specifically about its ability to catalyze the hydrolysis of GSH and certain other γ-Glu compounds. It was reported that spleen homogenates from GGT1-null mice as well as partially purified enzyme from GGT1-null mouse spleen failed to hydrolyze GSH or GSSG (Carter et al. 1997; Lieberman et al. 1999), but expressed hGGT5 has been reported to catalyze the hydrolysis of GSH (Heisterkamp et al. 1991).

The substrate selectivity of hGGT5 was investigated with assay conditions that allow selective quantification of expressed hGGT5- or hGGT1-catalyzed hydrolytic reactions (Wickham et al. 2011): hGGT5 catalyzes the hydrolysis of GSH, GSSG, LTC4 (189, Figure 16), S-(4-nitrobenzyl)GSH (126, Figure 10), S-(methyl)GSH, GSH sulfonic acid, but not γ-Glu-Leu. Although both hGGT1 and hGGT5 catalyze the hydrolysis of LTC4 (189, Figure 16), V_max/K_mE_t is much greater for hGGT1 than for hGGT5 (0.49 versus 0.019 min⁻¹nM⁻¹) (Wickham et al. 2011); as noted below, however, the expression of hGGT5 is high in some tissues so that the contribution of hGGT5 may be physiologically significant in these tissues.

3.3.3. Inhibitors

3.3.3.1. Reversible inhibitors

Ser-borate (217, Figure 20) inhibits expressed hGGT5 but is a less potent inhibitor of hGGT5 than of hGGT1 (Wickham et al. 2011).

3.3.3.2. Irreversible inhibitors

GGT5 is inhibited by acivicin (224, Figure 20) (Heisterkamp et al. 1991; Carter et al. 1997).

3.3.4. Tissue and species distribution

Northern blot analysis showed that GGT5 is intensely expressed in rat tracheal epithelium, faintly expressed in trachea, and not detectable in brain, heart, kidney, liver, lung spleen, and testis (Potdar et al. 1997); after reverse transcriptase polymerase chain reaction amplification, however, a hybridization signal was detected in brain, heart, kidney, liver, lung spleen, testis, and trachea. In the mouse, GGT5 is most highly expressed in the spleen and uterus but is also expressed in thymic lymphocytes, bronchiolar epithelial cells, pulmonary interstitial cells, renal proximal convoluted tubular cells, and crypt cells of the small intestine, and in cerebral, cerebellar, and brain stem neurons but not in glial cells (Carter et al. 1998; Han et al. 2002).

Immunohistochemical studies show that both GGT5 and GGT1 are highly expressed in human kidney but that only GGT1 is expressed in proximal tubular cells (Hanigan et al. 2015). With human liver, GGT5 is strongly expressed in Kupffer cell and macrophages but is weakly expressed in hepatocytes, whereas GGT1 is expressed in bile canalicular surfaces of hepatocytes. Immunohistochemical analysis of normal human tissues showed that GGT5 is detectable in all tissues studied but that the intensity of expression varied among tissues. GGT5 is expressed in primary bronchial epithelial cells (Lukic et al. 2016) and in PC-9ER-cells derived from non-small cell lung cancer cells (Serizawa et al. 2014).

Although GGT5 is widely distributed in mouse tissues, the highest expression is found in the spleen (Carter et al. 1998; Shi et al. 2001). GGT5 is also found in the airway tissues of the rat and guinea pig (Funayama et al. 1996; Potdar et al. 1997). Both GGT1 and GGT5 are widely distributed in human tissues, but the expression in cell types differs (Hanigan et al. 2015): GGT5 is expressed in the interstitial cells of the kidney whereas GGT1 is expressed in the apical surface of the proximal tubules.

3.3.5. Induction

The hydrolysis of leukotriene C₄ (189, Figure 16) is increased in transformed and normal human bronchial epithelial cells incubated with dexamethasone, fluticasone, or beclomethasone (Zaitsu et al. 2001, 2003). Treatment with glucocorticoids also caused up-regulation of GGT5 mRNA expression, but had no effect on GGT1 activity or GGT1 mRNA expression, suggesting a significant role for GGT5 in the hydrolysis of LTC4 (189, Figure 16) in human bronchial epithelial cells.

3.3.6. GGT5 as a biomarker for disease states

In an experimental murine model of peritonitis, GGT5, but not GGT1, is responsible for the generation of LTD4 (189, Figure 16) (Shi et al. 2001). Induction of airway responsiveness with Aspergillus fumigatus culture filtrate allergen in a murine model of asthma results in an increase in pulmonary GGT5 levels in wild-type and GGT1-null mice and causes a greater increase in airway reactivity in GGT5-null mice than in wild-type mice (Han et al. 2002). The level of LTC₄ increases substantially in the lungs of GGT5-null mice with allergen-induced asthma, whereas LTD₄ is not detectable in either bronchoalveolar lavage fluid or plasma, supporting the conclusion that, in asthmatic inflammation, GGT5 plays an important role in LTD₄ generation. In contrast, GGT1 is primarily responsible for the conversion of LTC₄ to LTD₄ in lung homogenates from mice treated only with saline (Shi et al. 2001). GGT5 expression lacks prognostic value in subjects with acute myeloid leukemia (Tian et al. 2014).

4. Dipeptidases

4.1. Introduction

Three dipeptidases have been demonstrated to catalyze the hydrolysis of Cys-Gly S-conjugates to Cys S-conjugates and Gly: membrane dipeptidase (DPEP1), aminopeptidase N (APN), and leucyl aminopeptidase (LAP) (Tate 1980; Commandeur et al. 1995; Cooper and Hanigan 2010; Barrett et al. 2012; Poon and Josephy 2012). DPEP1 and APN are membrane-bound, homodimeric proteins, whereas LAP is a soluble homohexamer. All are metalloenzymes, and zinc is the most commonly found metal; DPEP1 and LAP contain two zinc ions per subunit, whereas APN contains a single zinc ion per subunit; the zinc ions are essential for catalytic activity (Auld 2013). The cytosolic nonspecific peptidase (carnosine dipeptidase II, EC3.4.13.18, MEROPS Family M20.005) has apparently not been reported to catalyze the hydrolysis of Cys-Gly S-conjugates, although Cys-Gly is a substrate (Kaur et al. 2009).

This discussion of dipeptidases will focus on the roles of DPEP1, APN, and LAP in the mercapturic acid pathway. The biotransformation of few Cys-Gly S-conjugates has been investigated with identified and purified dipeptidases. The cellular and in vivo hydrolysis of Cys-Gly S-conjugates, however, has not been studied extensively enough to allow meaningful conclusions about the relative importance of the three dipeptidases in their metabolism.

Assessment of the role of dipeptidases in the mercapturic acid pathway is complicated by overlapping substrate selectivities and by differing inhibitor susceptibilities: for example, bestatin (267, Figure 22) inhibits 12 aminopeptidases (Scornik and Botbol 2001). In addition, inhibitor potencies differ among the dipeptidases: bestatin (267, Figure 22) is a more potent inhibitor of APN than cilastatin (266, Figure 22); cilastatin (266, Figure 22) is a more potent inhibitor of DPEP1 than bestatin (267, Figure 22) (Hirota et al. 1985); amastatin (268, Figure 22) is a more potent inhibitor of APN than bestatin (267, Figure 22) (Rich et al. 1984).

4.2. Membrane-bound dipeptidases

Two membrane-bound dipeptidases catalyze the hydrolysis of Cys-Gly S-conjugates: membrane dipeptidase and aminopeptidase N.

4.2.1. Membrane dipeptidase (DPEP1, EC 3.4.13.19)⁷

DPEP1 is a zinc-dependent enzyme that catalyzes the hydrolysis of dipeptides, but not tripeptides, with N-terminal amino acids in the l-configuration and C-terminal amino acids in the d- or l-configuration (Campbell et al. 1966; Campbell 1970). DPEP1 is assigned to Family M19 (MEROPS ID M19.001; https://www.ebi.ac.uk/merops/cgi-bin/pepsum?id=M19.001; accessed 31 October 2019) (Rawlings et al. 2018). Early studies demonstrated the presence of an enzyme that catalyzed the hydrolysis of dehydropeptides (Bergmann and Schleich 1932a, 1932b). The peptidase was later purified from rat kidney and demonstrated to catalyze the hydrolysis of glycyldehydrophenylalanine (229, Figure 21), Cys-Gly S-conjugate of N-ethylmaleimide (231, Figure 21), l-cystinyl-bis-glycine (233, Figure 21), and leukotriene D₄ (234, Figure 21) (Hirota et al. 1985, 1987a). For reviews about DPEP1, see Keynan et al. (1996b) and Hooper (2013).

Figure 21. Dipeptidase substrates. 229, glycyldehydrophenylalanine; ((E)-2-[(2-aminoacetyl)amino]-3-phenylprop-2-enoic acid); 230, dehydrophenylalanine; ((Z)-2-amino-3-phenylprop-2-enoic acid); 231, S-(1-ethyl-2,5-dioxopyrrolidin-3-yl)-l-Cys-Gly; 232, S-(1-ethyl-2,5-dioxopyrrolidin-3-yl)-l-Cys; 233, cystinyl-bis-glycine; 234, leukotriene D₄ (LTD4); 235, leukotriene E₄ (LTE4); 236, S-(2,4-dinitrophenyl)-l-Cys-Gly; 237, S-(2,4-dinitrophenyl)-l-Cys; 238, imipenem (N-formimidoyl thienamycin, MK0787); 239, (αS,2R)-5-carboxy-2,3-dihydro-α-[(1R)-1-hydroxyethyl]-4-[[2-[(iminomethyl)amino]ethyl]thio]-1H-pyrrole-2-acetic acid; 240, S-(benzyl)-l-Cys-Gly; 241, S-(benzyl)-l-Cys; 242, S-(benzyl)-l-Cys-4-nitroanilide; 243, l-leu-4-nitroanilide; 244, 3-cyclohexyl-l-alanyl-O-[[[1-(2-deoxy-2-methylene-β-d-erythro-pentofuranosyl)-1,2-dihydro-2-oxo-4-pyrimidinyl]amino]carbonyl]-d-threonine; 245, 2′-deoxy-2′-methylidenecytidine (DMDC); 246, pomaglumetad methionil; ((1 R,4S,5S,6S)-4-(l-methionylamino)-2-thiabicyclo[3.1.0]hexane-4,6-dicarboxylic acid 2,2-dioxide); 247, LY404039 ((1R,4S,5S,6S)-4-amino-2-thiabicyclo[3.1.0]hexane-4,6-dicarboxylic acid 2,2-dioxide); 248, S-(methyl)-l-Cys-Gly; 249, S-(methyl)-l-Cys; 250, S-(carbamidomethyl)-l-Cys-Gly; 251, S-(carbamidomethyl)-l-Cys; 252, 2-amino-N-(4-((2-oxo-2H-chromen-7-yloxy)methyl)phenyl)propanamide hydrochloride (Ala-PABA-7HC); 253, 7-((4-aminobenzyl)oxy)naphthalen-2(1H)-one; 254, 7-hydroxycoumarin; 255, (2S)-2-amino-N-(6′-aminospiro[isobenzofuran-1(3H),9′-[9H]xanthen]-3′-yl)-4-methylpentanamide (Leu-HMRG); 256, 2-(3,6-diamino-9H-xanthen-9-yl)benzenemethanol (HMRG); 257, l-Ala anilide of amino-hemicyanine (HCAN); 258, amino-hemicyanine (HCA); 259, (R,E)-2-amino-N-(4-(2-(3-(dicyanomethylene)-5,5-dimethylcyclohex-1-en-1-yl)vinyl)phenyl)-4-methylpentanamide (TMN-Leu); 260, (E)-2-(3-(4-aminostyryl)-5,5-dimethylcyclohex-2-en-1-ylidene)malononitrile (TMN-NH2); 261, melflufen (melphalan flufenamide); 4-[bis(2-chloroethyl)amino]-l-phenylalanyl-4-fluoro-l-Phe ethyl ester); 262, 4-fluoroPhe ethyl ester; 263, 5-fu-NGR; 264, N-l-leucylglycine 2-[[5-(dimethylamino-1-naphthalenyl]sulfonyl]hydrazide; 265, glycine 2-[[5-(dimethylamino-1-naphthalenyl]sulfonyl]hydrazide.

4.2.1.1. Structure and properties

DPEP1 is a zinc-containing homodimer with a subunit M_r of 47 000 to 59 000 that is anchored to the membrane with a glycosyl-phosphatidylinositol moiety (Adachi et al. 1989; Littlewood et al. 1989; Hooper et al. 1990; Adachi et al. 1990a; Satoh et al. 1993; Brewis et al. 1995; Poon and Josephy 2012). Murine DPEP3 is a disulfide-linked homodimer with a calculated subunit M_r of 54213 (Yoshitake et al. 2011). Differences in subunit M_r among h-, p-, and mDPEP1 are attributed to differing degrees of glycosylation. A “urinary dipeptidase” has been identified as deglycosylated DPEP1 that is formed by the cleavage of glycosylated DPEP1 by a phospholipase C (We et al. 1997; Park et al. 2001, 2002).

Three genes encode human membrane dipeptidases: DPEP1, DPEP2, and DPEP3 (Adachi et al. 1990b; Austruy et al. 1993). The mouse membrane dipeptidases are encoded by Dpep1, Dpep2, and Dpep3 (Satoh et al. 1993; Habib et al. 1996, 2003); porcine, rabbit, and rat genes encoding the membrane dipeptidase have also been described (Rached et al. 1990; Igarashi and Karniski 1991; Shimoyama et al. 2015).

DPEP1 subunits are linked by interchain disulfide bonds (Keynan et al. 1996a; Nitanai et al. 2002). Although pDPEP1 contains seven Cys residues, only Cys361 is involved in homodimer formation (Keynan et al. 1996a). DPEP1 was initially reported to contain one zinc ion per subunit (Campbell et al. 1966; Armstrong et al. 1974); the crystal structure, however, clearly shows the presence of two Zn ions along with ligands for each Zn ion (Nitanai et al. 2002). DPEP1 lacks the Zn-binding HExxH motif, which is found in many zinc peptidases (Jongeneel et al. 1989; Keynan et al. 1994).

The crystal structures of oligosaccharide-trimmed, unliganded and cilastin-liganded subunits of hDPEP1 have been solved (Nitanai et al. 2002). The subunits consist of α-helices and β-strands organized into (α/β)₈ barrel structures that are linked by disulfide bonds between Cys361 to form the homodimer. The active site with its two Zn ions is located in the barrel structure. One Zn ion (Zn1) forms ligands with Asp22 and His20 whereas the other Zn ion (Zn2) forms ligands with His198 and His219; Zn1 and Zn2 are bridged by Glu125. A water molecule (Wat421) forms coordinate bonds with Zn1 and Zn2 and is proposed to undergo deprotonation to form the nucleophile that attacks the carbonyl atom of the scissile bond of the substrate. Previous site-directed mutagenesis studies had demonstrated the importance of Glu125 and His219 for catalytic activity (Adachi et al. 1993; Keynan et al. 1994, 1997).

The selectivity of DPEP1 for dipeptide substrates is demonstrated by the crystal structure of the protein in complex with cilastatin (266, Figure 22), which is considered to be a dipeptide analog: the inhibitor completely fills the active site (Nitanai et al. 2002); also, superimposition of the structures of glycyldehydrophenylalanine (229, Figure 21) and imipenem (238, Figure 21) shows that they are accommodated by the active site. An analysis of the cilastatin-DPEP1 complex indicated that the inhibitor binds in a catalytically competent manner, but tight interactions of the alkyl groups at its N- and C-termini with amino acid residues at either end of the active site prevent breakdown of the tetrahedral intermediate (Smyth et al. 2003). The substrate selectivity of DPEP1 shows a preference for dipeptides with bulky residues at the N-terminus, which were proposed to interact with a hydrophobic pocket in the active site (Campbell et al. 1988); for example, the V_max/K_m for L-Trp-Gly is larger than for Gly-l-Trp.

Figure 22. Dipeptidase inhibitors. 266, cilastatin; 267, bestatin (Ubenimex); 268, amastatin; 269, (+)-(Z)-2-(2,2-dimethylcyclopropane-1-carboxamido)but-2-enoic acid; 270, 2-[(cyanoacetyl)amino]-2-propenoic acid; 271 (2Z)-2-[[(1-amino-2-cyclohexylethyl)hydroxyphosphinyl]methyl]-3-(4-fluorophenyl)-2-propenoic acid; 272, tosedostat (CHR-2797); 2-(2-(hydroxy(hydroxycarbamoyl)methyl)-4-methylpentanoylamino)-2-phenylethanoic acid cyclopentyl ester); 273, α₁-amino-3-phenylpropyl(α₂-hydroxy-3-phenylpropyl)phosphinic acid; 274, l-Leu thiol; ((S)-2-amino-4-methylpentane-1-thiol); 275, l-leucinal; ((S)-2-amino-4-methylpentanal); 276, Leu-B(OH)₂-pinacol; (3-methyl-l-(4,4,5,5-tetramethyl-l,3,2-dioxaborolane-2-yl)-l-butylamine); 277, l-Leu phosphonic acid [(1R)-1-amino-3-methylbutyl]phosphonic acid); 278 (S)-2-[3-(2,6-dimethyl-phenyl)-ureido]-4-methyl-pentanoic acid hydroxyamide; 279, 2-amino-1,4-dihydrophenanthren-3(2H)-one; 280, 7-amino-1-bromo-4-phenyl-5,7,8,9-tetrahydrobenzocyclohepten-6-one; 281, zofenprilat; ((2S,4S)-1-[(2S)-2-methyl-3-sulfanylpropanoyl]-4-(phenylsulfanyl)pyrrolidine-2-carboxylic acid); 282, diethyl 6,8-dibenzyloxy-3,4-dihydroisoquinoline-3,3-dicarboxylate.

A computational analysis of the catalytic cycle of DPEP1 revealed substrate binding in which the N-terminal amino acid and carboxylate groups of the dipeptide are coordinated with Zn1 and Zn2, respectively (Liao et al. 2010). Deprotonation of the active-site water by Asp288 gives the oxyanion that attacks the substrate carbonyl group to form a tetrahedral intermediate that is stabilized by His152; collapse of the tetrahedral intermediate yields the hydrolysis products. These binding and mechanism results are in agreement with, and expand upon, an earlier proposal (Nitanai et al. 2002), and are consistent with the results of docking studies (Smyth et al. 2003; Kim et al. 2007).

4.2.1.2. Substrates

DPEP1 preferentially catalyzes the hydrolysis of dipeptides with l-amino acids in the N-terminal position and both D- and l-peptides at the C-terminal position, e.g. Gly-l-Ala, Gly-D-Ala (Robinson et al. 1953; Semenza 1957; Campbell et al. 1966; Campbell 1970; Kozak and Tate 1982; Tate 1985; Campbell et al. 1988; Watanabe et al. 1996; Yamada and Kera 1998). A compilation of DPEP1 substrates is available on the MEROPS website (https://www.ebi.ac.uk/merops/cgi-bin/substrates?id=M19.001; accessed 31 October 2019).

Cys-Gly is a substrate for DPEP1 but shows substrate inhibition at elevated substrate concentrations (McIntyre and Curthoys 1982). Cys-Gly inhibits the DPEP1-catalyzed hydrolysis of l-Ala-Gly (Kozak and Tate 1982). DPEP1 catalyzes the hydrolysis of Cys-Gly S-conjugates, including S-(2,4-dinitrophenyl)- l-Cys-Gly (236, Figure 21), the Cys-Gly S-conjugate of NEM (231, Figure 21), l-cystinyl-bis-glycine (233, Figure 21), and leukotriene D₄ (234, Figure 21) (Anderson et al. 1982; Kozak and Tate 1982; Hirota et al. 1985; Tate 1985; Farrell et al. 1987; Hirota et al. 1987a; Campbell et al. 1988; Adachi et al. 1989). The biotransformation of a range of S-aryl-Cys-Gly S-conjugates, e.g. 236 (Figure 21) is catalyzed by DPEP1(Poon and Josephy 2012).

The observation that the carbapenem antibiotic imipenem (238, Figure 21), which is resistant to bacterial β-lactamase-catalyzed hydrolysis, undergoes renal metabolism in several mammalian species was an unexpected finding: mammals were thought to lack β-lactamase activity. It was demonstrated, however, that DPEP1 catalyzes the hydrolysis of imipenem (238, Figure 21) (Kim and Campbell 1982; Kropp et al. 1982; Kahan et al. 1983; Campbell et al. 1984). DPEP1 is the only mammalian dipeptidase demonstrated to possess β-lactamase activity: mice with the targeted deletion of the DPEP1 gene lack β-lactamase activity (Habib et al. 1998). Formerly, Primaxin® IV, a fixed-dose preparation of imipenem (238, Figure 21) and cilastatin (266, Figure 22) to prevent DPEP1-catalyzed degradation of imipenem (238, Figure 21) was approved for human clinical use; elevated plasma Cys-Gly concentrations are observed in human subjects given Primaxin^® IV, which are attributable to the cilastatin-induced inhibition of the DPEP1-catalyzed hydrolysis of Cys-Gly (Badiou et al. 2005).

The substrate selectivity of DPEP1 has been exploited to activate prodrugs. DPEP1 is overexpressed in human colorectal and stomach tumor tissue (Kohchi et al. 2007). The DPEP1-catalyzed hydrolysis of the dipeptide moiety of prodrug 244 (Figure 21) in these tissues gives an intermediate that spontaneously releases the antitumor agent, 2′-deoxy-2′-methylidenecytidine (245, Figure 21). The methionine residue of pomaglumetad methionil (246, Figure 21) is cleaved by DPEP1 to give dicarboxylic acid 247 (Figure 21), which is a selective metabotropic glutamate 2/3 receptor agonist (Moulton et al. 2015).

Murine membrane dipeptidase-2 (MEROPS ID M19.002) hydrolyzes leukotriene D₄ (234, Figure 21), but not cystinyl-bis-glycine (233, Figure 21) whereas murine membrane dipeptidase-3 (MEROPS ID M19.004) hydrolyzes cystinyl-bis-glycine (233, Figure 21) but not leukotriene D₄ (234, Figure 21) (Habib et al. 2003).

4.2.1.3. Inhibitors

DPEP1 is inhibited by chelating agents, by transition-state analogs, and by substrate analogs. 1,10-Phenanthroline, penicillamine, and dithiothreitol inhibit porcine and human kidney DPEP1, either by chelating zinc or by binding to a zinc coordination site; activity is partially restored by incubation with ZnCl₂ (Harper et al. 1971; Kozak and Tate 1982; McIntyre and Curthoys 1982; Campbell et al. 1988; Littlewood et al. 1989). A compilation of DPEP1 inhibitors is available on the MEROPS website (https://www.ebi.ac.uk/merops/cgi-bin/inhibitors.pl?id=M19.001; accessed 31 October 2019).

Cilastatin (266, Figure 22) is a competitive, reversible inhibitor of DPEP1 (Farrell et al. 1987) and is a more potent inhibitor of DPEP1 than bestatin (267, Figure 22) or amastatin (268, Figure 22) (Hirota et al. 1985, 1986, 1987a; Campbell et al. 1990a; Watanabe et al. 1996).

The crystal structure of DPEP1 in complex with cilastatin (266, Figure 22) shows that the active site is completely occupied by the inhibitor (Nitanai et al. 2002). This is consistent with the earlier observation that DPEP1 prefers substrates with a bulky, hydrophobic group at the N-terminal position (Campbell et al. 1988). The positioning of cilastatin (266, Figure 22) in the active site was proposed to prevent the formation of a tetrahedral intermediate (Nitanai et al. 2002).

Bestatin (267, Figure 22) is a time-dependent, noncompetitive, tight-binding inhibitor of DPEP1 (Campbell et al. 1988). Butenoic acid 269 (Figure 22), which is an analog of cilastatin (266, Figure 22), inhibits the DPEP1-catalyzed biotransformation of imipenem (238, Figure 21) (Kim and Campbell 1982; Graham et al. 1987). Propenoic acid 270 (Figure 22) is a mechanism-based inactivator of DPEP1 (Wu and Mobashery 1991). Aminophosphinic acid 271 (Figure 22) and related compounds are potent (IC₅₀ ≃ 5 nM) inhibitors of DPEP1 (Gurulingappa et al. 2003, 2004). An analog of phosphinic acid 271 (Figure 22) is a slow, tight-binding inhibitor of DPEP1 that is significantly more potent (K_i = 0.52 nM) than cilastatin (K_i = 0.11 μM) (266, Figure 22) (Parsons et al. 1991).

4.2.1.4. Tissue and species distribution

DPEP1 is primarily located in the microsomal fraction of the porcine renal cortex (Harper et al. 1971). Porcine DPEP1 is expressed on the lumenal face of renal cortical tubules, but not in the medulla or glomeruli (Littlewood et al. 1989). DPEP1 activities are high in rat lung and kidney but much lower in spleen, liver, brain, pancreas, and small intestine (Hirota et al. 1986). In rat kidney, DPEP1 is found on both the brush-border and basolateral membranes (Hirota et al. 1987b). High levels of human DPEP1 activity are observed in pancreas, renal cortex, and testis, but in contrast to the rat lung, activity is very low in human lung tissue (Kera et al. 1999).

Northern-blot analysis shows that the rat DPEP1 is intensely expressed in the small intestine, lung, and kidney, weakly expressed in stomach, pancreas, and testis, and not detectable in brain, heart, liver, or spleen (Adachi et al. 1992). DPEP1 is overexpressed in both benign and malignant colon cancer epithelium (Buckhaults et al. 2001). DNA-microarray analysis of tumor tissue showed that DPEP1 is highly expressed in human colorectal and stomach tumor tissue but not in adjacent normal tissue (Kohchi et al. 2007).

DPEP1 has been identified as the target for the lung-homing peptide GFE-1 (CGFECVRQCPERC) in mice (Rajotte and Ruoslahti 1999); given that Cys-Gly is a substrate for DPEP1, the N-terminal Cys-Gly residues of GFE-1 may bind to the active site of DPEP1. These studies were extended to identify DPEP1 binding molecules and their binding mode (Kim et al. 2007).

Western-blot analysis showed that murine DPEP3 is expressed in the testicular seminiferous tubules but not in interstitial cells or in heart, liver, lung, kidney, spleen, or ovary (Habib et al. 2003; Yoshitake et al. 2011).

4.2.2. Aminopeptidase N (APN, EC 3.4.11.2)^8,9

The description of a microsomal “cysteinylglycinase” from porcine kidney is apparently the first report of an enzyme that would later be identified as APN (Olson and Binkley 1950; Binkley 1951; Binkley et al. 1957; Semenza 1957; Binkley et al. 1959). Later studies characterized a microsomal peptidase from porcine kidney that catalyzed the hydrolysis of a range of peptides along with l-Ala and l-Leu-4-nitroanilide (243, Figure 21) (Pfleiderer and Celliers 1963; Pfleiderer 1970). For reviews about APN, see Sjöström et al. (2000), Langner and Ansorge (2002), Lendeckel et al. (2004), Bauvois and Dauzonne (2006), Luan and Xu (2007), Mina-Osorio (2008), and Turner (2013).

4.2.2.1. Structure and properties

APN is a homodimeric, zinc-containing ectoenzyme assigned to MEROPS Family M1 (MEROPS ID M01.001) (Rawlings et al. 2018) that catalyzes the release of N-terminal L-amino acids, preferably Ala, from peptides, amides, or arylamides. APN contains one Zn ion per subunit and possesses the HExxH zinc-binding motif (Wacker et al. 1971; Lehky et al. 1973; Malfroy et al. 1989). Mammalian APN is a type-II membrane protein with a short intracellular tail that contains the amino terminus, a transmembrane anchoring sequence, a short extracellular stalk, and a large ectodomain with the carboxy terminus; the active site and the zinc-binding domain are located in the extracellular domain (Sjöström et al. 2002; Wong et al. 2012; Chen et al. 2012a).

ANPEP encodes a peptide of 967 amino acids (Olsen et al. 1988; Look et al. 1989; Watt and Willard 1990). The deduced subunit M_r of APN from several organs and species is approximately 110000, but APN is heavily, but variably, N- and O-glycosylated, which accounts for the range of reported molecular masses (Olsen et al. 1988; Malfroy et al. 1989; Watt and Yip 1989; O'Connell et al. 1991; Norén et al. 1997; Sjöström et al. 2002).

Aminopeptidase N is identical with CD13 and acts, therefore, not only as a hydrolytic enzyme but also plays a number of roles in cancer and virus biology (Olsen et al. 1988; Look et al. 1989; Riemann et al. 1999; Pasqualini et al. 2000; Abe and Aoyagi 2002; Mina-Osorio 2008; Haraguchi et al. 2010). APN also serves as a receptor for tumor-homing peptides that bear an Asn-Gly-Arg (NGR) motif (Liu et al. 2014a). In addition to its catalytic and tumor-homing peptide receptor functions, APN acts as a cell entry receptor for many coronaviruses (Luan and Xu 2007). Mapping of coronavirus binding sites, “virus-binding domains,” revealed their locations on the body and head domains on the surface of the APN protein (Chen et al. 2012a). Coronavirus binding may lock APN in the open conformation and, thereby, inhibit catalysis (Santiago et al. 2017). Crystal structures of complexes formed by the pig APN ectodomain with the receptor binding proteins of two coronaviruses revealed determinants of coronavirus-APN interactions (Reguera et al. 2012). Hence, due to its involvement in a myriad of cell functions, APN/CD13 is an attractive therapeutic target, which has led to a search for modulators and imaging strategies (Bhagwat et al. 2002; Langner and Ansorge 2002; Su et al. 2011; Wickström et al. 2011; Hahnenkamp et al. 2013; Li et al. 2014; Drinkwater et al. 2017; He et al. 2017a; Amin et al. 2018).

Crystal structures of the ectodomains of both human and porcine APN have been solved (Reguera et al. 2012; Wong et al. 2012; Chen et al. 2012a; Joshi et al. 2017). hAPN is organized into seven domains in each subunit: (I) cytosolic domain; (II) transmembrane domain; (III) O-glycosylation domain; (IV) N-glycosylation domain; (V and VI) active-site domain with HExxHxE motif; and (VII) dimerization domain (Sjöström et al. 2002; Luan and Xu 2007; Wong et al. 2012; Luan et al. 2012b). The ectodomain of pAPN is characterized by four domains (head, side, body, and tail) that form a spacious cavity with a wide opening that is accessible to substrates (Chen et al. 2012a); the cavity is sufficiently large to accommodate, for example, substance P (RPKPQQFFGLM). pAPN forms a head-to-head dimer each with a transmembrane anchor and can adopt open or closed conformations, the latter of which is catalytically active. The proposed catalytic mechanism of pAPN involves the binding of zinc by His383, His387, and Glu406 (Chen et al. 2012a); an active-site water molecule is activated by the Zn and positioned by Glu350 to effect the transfer of a proton to Glu384. Binding of the peptide substrate is accompanied by activation of the carbonyl oxygen by the Zn atom and Tyr472, and by activation of the scissile bond nitrogen by the carbonyl oxygen of Ala348, all of which facilitate the attack of the zinc-activated water molecule on the carbonyl carbon while a proton is transferred from Glu384 to the leaving scissile bond nitrogen. The catalytic mechanism of APN is similar to that of other metallopeptidases having 2His1Glu (or 1Asp) Zn catalytic sites (Mucha et al. 2010; Auld 2013).

4.2.2.2. Substrates

APN catalyzes the hydrolysis of N-terminal l-amino acids from a range of peptides (Pfleiderer and Celliers 1963; Pfleiderer 1970; McDonald and Barrett 1986; Turner 2013); l-amino acids with neutral, bulky hydrophobic residues are preferred substrates. Although Ala is the N-terminal amino acid most readily cleaved by APN, Phe, Met, Tyr, and Leu are also favored substrates (Huang et al. 1997; Mina-Osorio 2008). l-Amino acid amides and anilides, e.g. 243 (Figure 21), are substrates but anilides are hydrolyzed at much higher rates (Pfleiderer and Celliers 1963; Wachsmuth et al. 1966; Rankin et al. 1980; Nakanishi et al. 1989; Drag et al. 2010; Shibata et al. 2016). Glu350 in pAPN may serve as an ionic binding site for recognition of the amino group of substrates and inhibitors, as well as contributing to the stabilization of the transition-state intermediate (Luciani et al. 1998). APN catalyzes the hydrolysis of N-terminal, neutral l-amino acids from many polypeptides, including enkephalins, angiotensin III and IV, tuftsin, kinins, chemo- and cytokines, extracellular matrix proteins and more (Ward et al. 1990; Abe and Aoyagi 2002; Albiston et al. 2004; Danziger 2008). A compilation of APN substrates is available on the MEROPS website (https://www.ebi.ac.uk/merops/cgi-bin/substrates?id=M01.001; accessed 31 October 2019).

After early studies identified a porcine kidney enzyme that catalyzed the hydrolysis of Cys-Gly (see above), it was proposed that APN might catalyze the hydrolysis of Cys-Gly S-conjugates (Hughey et al. 1978). It was subsequently demonstrated that APN catalyzes the hydrolysis of many Cys-Gly S-conjugates, including S-(benzyl)-Cys-Gly (240, Figure 21), S-(benzyl)-Cys-4-nitroanilide (242, Figure 21), and cystinyl-bis-Gly (233, Figure 21), S-(methyl)-L-Cys-Gly (248, Figure 21), S-(carbamidomethyl-Cys-Gly (250, Figure 21), but shows scant or no activity with leukotriene D₄ (234, Figure 21) (Hughey et al. 1978; Rankin et al. 1980; Okajima et al. 1981; Anderson et al. 1982; Kozak and Tate 1982; McIntyre and Curthoys 1982).

The role of APN in angiogenesis and tumor biology has led to the development of strategies to image APN in cells and in vivo. Briefly, three types of imaging agents have been described: (1) substrate-based agents that are biotransformed by APN to fluorescent products; (2) inhibitor-based agents, whose inhibitory properties direct them to APN; and (3) targeted agents that exploit the affinity of APN for peptides that contain the NGR motif. Given the large number of imaging agents that has been reported, this section will focus on recently described compounds along with citations to related agents. (Note that some cited authors use the name leucine aminopeptidase (LAP) rather APN.)

l-Ala anilide 252 (Figure 21) represents an early ratiometric fluorescent probe for APN that is biotransformed to the self-immolating intermediate 253 (Figure 21) that yields fluorescent 7-hydroxycoumarin (254, Figure 21) (Chen et al. 2012b). These studies were extended to develop a naphthalimide-based probe with more favorable spectroscopic properties, which allows ratiometric fluorescent imaging of APN in cells (Chen et al. 2013b). The l-Ala anilide of 9-aminocresyl violet was used to quantify APN concentrations in urine and in HepG2 and LO2 cells (He et al. 2017b). The l-Leu anilide of hydroxymethyl rhodamine green (Leu-HMRG; 255, Figure 21) is hydrolyzed by APN to give rhodamine 256 (Figure 21) and was shown to image APN in HeLa cells (Sakabe et al. 2013). The l-Ala and l-Leu anilides of 6-aminoluciferin were used to successfully image APN in xenografted luciferase-expressing tumors in nude mice (Li et al. 2014); light production was decreased by administration of the APN inhibitor bestatin (267, Figure 22). A boron-dipyrromethylene (BODIPY) core bearing a l-Cys-l-Leu moiety (BODIPY-C-Leu) is biotransformed by APN to the BODIPY-Cys intermediate that undergoes a S to N migration to give the highly fluorescent BODIPY-N-Cys (Zhou et al. 2017); BODIPY-Leu successfully imaged APN in HeLa cells and in zebrafish, and light production was decreased by the APN inhibitor bestatin (267, Figure 22).

Near-infrared (NIR) probes can offer the advantages of large Stokes shifts, high sensitivity, water solubility, and deep tissue penetrability (Gu et al. 2016; Zhang et al. 2017; He et al. 2017a). The l-Ala anilide of amino-hemicyanine 257 (Figure 21) was used to image APN in HepG2 cells and in tumor-bearing nude mice (He et al. 2017a). NIR probe 259 (Figure 21) was used to image APN in HCT116 cells and in HCT116 tumor-bearing nude mice (Zhang et al. 2017); the intensity of the image was reduced in APN siRNA-transfected HCT116 cells and in mice given bestatin (267, Figure 22).

The selectivity of dipeptidase inhibitors has been exploited for the development of imaging probes. A probe for APN that was based on a bestatin (267, Figure 22) scaffold and included a fluorophore, biotin, and an alkyne moiety (for click-chemistry applications) has been described, but imaging results were not reported (Harbut et al. 2008). A probe based on bestatin (267, Figure 22) linked to 4-amino-1,8-naphthalimide (Bes-Green) was shown to inhibit APN and to image APN in ovarian clear cell carcinoma (ES-2) cells (Chen et al. 2015). Cyanine dye Cy 5.5 linked to a hydroxamic acid-based APN inhibitor (Cy 5.5-23) has been reported as an imaging probe (Hahnenkamp et al. 2013; Chen et al. 2013a); Cy 5.5-23 binds to APN-positive BT-549 breast cancer cells but not to APN-negative BT-20 cells, as shown by fluorescence microscopy.

APN shows high affinity for peptides bearing an NGR motif (Pasqualini et al. 2000; Corti et al. 2017). The cyclic NGR peptide cyclo-[Cys-Asn-Gly-Arg-Cys]-Gly-Lys linked to cyanine dye Cy 5.5 (NGR-Cy 5.5) was designed to bind to APN (von Wallbrunn et al. 2008); NGR-Cy 5.5 was shown by fluorescence microscopy to bind to APN-positive HT-1080 cells, but not to APN-negative MCF-7 cells, and to visualize APN in HT-1080, but not in MCF-7, tumor xenografts in nude mice by 2D planar fluorescence reflectance imaging and by 3D fluorescence-mediated tomography.

APN also catalyzes the activation of prodrugs. The melphalan (26, Figure 4) prodrug melflufen (261, Figure 21) undergoes APN-catalyzed biotransformation to release melphalan (26, Figure 4) (Wickström et al. 2010; Strese et al. 2013; Ray et al. 2016); the cytotoxicity of prodrug 261 (Figure 21) is reduced by the dipeptidase inhibitor bestatin (267, Figure 22). Clinical trials to test the efficacy of melflufen (261, Figure 21) for the treatment of multiple myeloma are currently in progress (clinicaltrials.gov) (Wickström et al. 2017). Hybrid prodrugs of 5-fluorouracil (5-FU) and gemcitabine linked to bestatin (267, Figure 22) are expected to be cleaved by plasma hydrolases to release the active antitumor agents and the APN inhibitor bestatin (267, Figure 22) (Jiang et al. 2016a, 2016b); the compounds inhibited APN and exhibited antitumor and antiangiogenic effects in several cellular and in vivo models. APN-activated glycinamide-based prodrugs of a panel of vascular disrupting agents that inhibit tubulin polymerization have been described (Devkota et al. 2016); several of these compounds were cytotoxic in cancer cell lines.

The 5-fluorouracil (5-FU) prodrug 263 (5-FU-NGR, Figure 21) is based on the affinity of NGR peptides for APN (Luan et al. 2012a; Cui et al. 2014b). The labile acyloxy-methylene linker is designed to release 5-FU (Menger and Rourk 1997); moreover, the NGR peptide was modified to include an Arg-(ω-NO₂)-COOCH₃ moiety to inhibit nitric oxide synthase. 5-FU-NGR (263, Figure 21) inhibited growth of Colo205 cells, delayed expansion of Colo205 xenografts in nude mice, induced apoptosis in Colo205 cells and xenografts, and inhibited APN expression. The affinity of APN for NGR peptides has been exploited to target peptide-conjugates, e.g. TNFα (Corti and Ponzoni 2004; Gregorc et al. 2010; Corti et al. 2017). Indeed, the NGR-hTNF conjugate is being investigated in clinical trials for the treatment of hepatocellular carcinoma (Santoro et al. 2010).

4.2.2.3. Inhibitors

The putative role of APN in tumor progression metastasis has stimulated the search for inhibitors, particularly small-molecule inhibitors. Consequently, a large number of APN inhibitors have been reported. These inhibitors should also find use in exploring the role of APN in the mercapturic acid pathway. This section will focus on more recently reported compounds. Reviews of APN inhibitors are available (Shimazawa et al. 1999; Abe and Aoyagi 2002; Fournié-Zaluski and Roques 2002; Xu and Li 2005; Bauvois and Dauzonne 2006; Luan and Xu 2007; Zhang and Xu 2008; Mucha et al. 2010; Su et al. 2011; Wickström et al. 2011; Zhang et al. 2011b; Luan et al. 2012b; Hitzerd et al. 2014; Drinkwater et al. 2017; Amin et al. 2018). A compilation of APN inhibitors is available on the MEROPS website (https://www.ebi.ac.uk/merops/cgi-bin/pepsum?id=M01.001; accessed 31 October 2019).

The structural requirements for APN inhibitors have been summarized (Drag et al. 2010; Halder et al. 2013; Amin et al. 2018): briefly, APN inhibitors are typically characterized by hydrophobic groups that are accommodated by the S₁, S_1′, and S_2′ sites and by a zinc-binding group (ZBG), including hydroxamic acids, thiols, boronic acid, phosphonic acids, and β-dicarbonyl compounds. These requirements are met by a range of scaffolds that have been exploited to identify potent and selective APN inhibitors.

APN is inhibited by a range of natural-product inhibitors, including bestatin (267, Figure 22), amastatin (268, Figure 22), actinonin, leuhistin, and probestin (Bauvois and Dauzonne 2006). Co-crystallization of zinc-bound hAPN with the inhibitors bestatin (267, Figure 22) and amastatin (268, Figure 22), both of which contain a nonhydrolyzable α-hydroxy-β-amino acid moiety, showed that the α-hydroxyl group of amastatin (268, Figure 22) coordinates with the zinc ion whereas bestatin (267, Figure 22) resides deep in the active-site pocket, where its α-hydroxyl, amino, and carbonyl groups form only water-mediated hydrogen bonds with hAPN amino-acid residues (Wong et al. 2012). LYP, the dimethylaminoethyl ester of bestatin (267, Figure 22), is a more potent inhibitor of APN than bestatin (267, Figure 22), inhibits the growth of ovarian carcinoma ES-2 and SKOV-3 cells, inhibits angiogenesis in ES-2 xenografts in mice, and delays growth of the ES-2 xenografts (Luan et al. 2009; Gao et al. 2011a, 2011b). hAPN is inhibited by the bile acids chenodeoxycholic acid and cholic acid, particularly the unconjugated acids, and by metal chelating agents (Nakanishi et al. 1989; Vanderheyden et al. 2006).

The conjugated β-dicarbonyl compound, curcumin (69, Figure 6), is a noncompetitive, irreversible inhibitor of purified APN and also inhibits APN activity in human umbilical vein endothelial cells and in HT1080 fibrosarcoma cells; curcumin (69, Figure 6) inhibits both angiogenesis and tumor cell invasion by APN-positive cells (Shim et al. 2003). Other β-dicarbonyl compounds, which were designed to act as zinc-binding agents and to interact with key active-site residues, also are effective APN inhibitors (Ma et al. 2013).

Hydroxamic acids, because of their zinc-ligand-forming ability, have been studied as inhibitors of APN (Chan et al. 1982; Flipo et al. 2007; Mucha et al. 2010). The hydroxamic acid tosedostat (CHR-2797; 272, Figure 22) is an orally bioavailable APN inhibitor, as well as an inhibitor of several other aminopeptidases. Tosedostat is biotransformed to a carboxylic acid metabolite that is also an effective aminopeptidase inhibitor. Tosedostat has been investigated in clinical trials for its antitumor activity (clinicaltrials.gov) (DiNardo and Cortes 2014). Hydroxamic acid moieties have been incorporated into a wide variety of chemical scaffolds that have yielded effective APN inhibitors (Inagaki et al. 2010; Li et al. 2012; Su et al. 2012a; Hahnenkamp et al. 2013; Jin et al. 2013; Su et al. 2013; Zhang et al. 2013; Cui et al. 2014a; Hou et al. 2016; Ma et al. 2016; Cao et al. 2018).

Phosphorous-containing inhibitors, particularly α-aminoalkylphosphonic acid and -phosphinic acids have been studied as APN inhibitors (Mucha et al. 2010; Amin et al. 2018). Phosphinic acids (phosphinates) are of interest because they function as transition-state analog inhibitors (Amin et al. 2018). For example, phosphinic acid 273 (Figure 22) is a potent inhibitor of hAPN (IC₅₀ = 60 nM) (Grzywa et al. 2010a). Other phosphinic and phosphonic acid inhibitors of APN have been reported, some of which exhibit IC_{50 or} K_i values in the low nM range (Chen et al. 1999; Drag et al. 2010; Grzywa et al. 2010b; Mucha et al. 2011; Mucha 2012; Niu et al. 2015).

Some early studies of APN inhibitors focused on agents that are Leu-derived amino acid analogs, including l-Leu thiol 274 (Figure 22), l-leucinal (275, Figure 22), and Leu-B(OH)₂-pinacol (276, Figure 22) (Andersson et al. 1982; Shenvi 1986). There has been considerable interest in l-leucine-ureido derivatives as APN inhibitors. These compounds build on the structure of bestatin (267, Figure 22), which also contains a l-Leu component whose sidechain is accommodated by the S₁′ site of APN, along with the addition of a hydroxamic acid moiety as a ZBG (Wong et al. 2012). l-Leucine-ureido analog 278 (Figure 22), for example, is a potent in vitro inhibitor of pAPN (IC₅₀ = 20 nM), inhibits hAPN in ES-2 human ovarian cells, blocks invasion by ES-2 cells, and blocks pulmonary metastasis in mice bearing H22 tumor cells (Ma et al. 2016). A number of other l-Leu-ureido-based APN inhibitors, many of which contain a hydroxamic acid zinc binding group, have been reported (Su et al. 2012a, 2012b, 2013).

3-Amino-2-tetralone derivatives have been identified as APN inhibitors. Tetralone 279 (Figure 22) is a potent, slow-binding inhibitor of APN (K_i = 20 nM) but is a much weaker inhibitor of LAP (K_i = 3 μM) (Schalk et al. 1994). The structurally related aminobenzosuberones are also selective inhibitors of APN (Albrecht et al. 2011; Maiereanu et al. 2011; Albrecht et al. 2012; Revelant et al. 2015). Aminobenzosuberone 280 (Figure 22) is a selective and potent inhibitor of pAPN and hAPN (K_i = 0.06 and 0.035 nM, respectively) (Peng et al. 2017). The amino group is required for inhibition, and binding interactions of the hydrated ketone group mimic stabilization interactions of the transition state for peptide bond hydrolysis. In hAPN, a C-Br⋅⋅⋅π interaction with F896 is thought to contribute to the high binding affinity. Studies on the cancer chemotherapeutic potential of aminobenzosuberone 280 (Figure 22) show that it inhibits capillary tube formation, lowers cell migration, and retards growth of xenograft tumors (Schmitt et al. 2013); the maximum total tolerated dose (MTTD) in mice was estimated to be 40 mg/kg when given every third day for a total of eight doses.

An additional number of APN inhibitors have been reported, but most are less potent than the compounds described above. These include flavone derivatives, cinnamoyl pyrrolidines, cyclic imides, β-amino-α-hydroxyphenylbutanoic acids, indoline-2,3-diones, and 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acids (for details about these compounds, see Mucha et al. 2010; Amin et al. 2018). Dihydroisoquinoline 282 (Figure 22) inhibits APN and shows antiproliferative effects in several cell lines (Ziemska et al. 2016).

4.2.2.4. Tissue and species distribution

APN is widely distributed across mammalian species and is also found in avian species (Sjöström et al. 2002). The membrane-bound APN is present in nearly all tissues, including kidney, intestine, heart, lung, eye, and thymus, skin and respiratory tract (Hughey et al. 1978; Sjöström et al. 1978; Hansen et al. 1987; Nakanishi et al. 1989; Dixon et al. 1994; Danielsen et al. 1995; Jardinaud et al. 2004). APN is widely expressed in brain tissue, which has led to the proposal that it may play a role in hypertension (Danziger 2008). APN is expressed in the brush-border of fetal porcine enterocytes (Danielsen et al. 1995). Immunohistochemical studies showed the presence of APN in bile canaliculi (Röcken et al. 2004, 2005). APN is found in rat and guinea pig kidney brush-border membranes (Rankin et al. 1980; Terashima et al. 1992). In rabbit kidney, in situ hybridization studies demonstrated the presence of APN mRNA in proximal convoluted tubules, but not in glomeruli (Yang et al. 1993). Immunoelectron microscopic analysis demonstrated a high level of APN expression on the lumenal surfaces of normal human proximal tubular cells and renal cancer cells (Stange et al. 2000).

4.2.2.5. Other biological functions

Given the range of functions associated with APN in addition to catalysis of the hydrolysis of Cys-Gly and Cys-Gly S-conjugates, APN (CD13) has been branded as a “…moonlighting enzyme…” (Mina-Osorio 2008). Upon ligand binding, CD13 functions as an enzyme, receptor, and signaling molecule, and these functions depend on peptide cleavage, endocytosis, and signaling. APN also serves as a receptor for a range of viruses (Yeager et al. 1992; Delmas et al. 1994). The enzymatic activity of APN is not required for interaction of the transmissible gastroenteritis virus with the virus receptor (Delmas et al. 1994).

APN was considered to be a possible binding target of the cholesterol absorption inhibitor ezetimibe (Kramer et al. 2005). Subsequent studies showed conclusively that Niemann-Pick C1-Like 1 (NPC1L1) protein is the molecular target of ezetimibe (Garcia-Calvo et al. 2005; Betters and Yu 2010; Jia et al. 2011).

4.3. Cytosolic dipeptidases

One cytosolic dipeptidase has been demonstrated to catalyze the hydrolysis of both Cys-Gly and Cys-Gly S-conjugates: leucyl aminopeptidase.

4.3.1. Leucyl aminopeptidase (LAP, EC 3.4.11.1)¹⁰

The first reference to an enzyme that would later be identified as LAP was the description of an intestinal peptidase that catalyzed the hydrolysis of Leu-Gly and other leucyl dipeptides (Linderstrøm-Lang and Sato 1929; Holter 1979).

Although APN catalyzes the hydrolysis of Cys-Gly (see above), other studies show that LAP is the major liver enzyme that catalyzes the hydrolysis of Cys-Gly and Cys-Gly S-conjugates, but its activity in other organs was not reported (Jösch et al. 1998, 2003). LAP is a metallopeptidase assigned to the Family M17 (MEROPS ID M17.001) (Rawlings et al. 2018). Reviews about LAP are available (Taylor 1993a, 1993b; Kim and Lipscomb 1994; Sträter and Lipscomb 2013).

4.3.1.1. Structure and properties

Enzymes that catalyze the hydrolysis of Leu dipeptides and Leu amides and anilides have been purified from several sources, including porcine intestine and kidney (Smith and Bergmann 1944; Spackman et al. 1955), rat liver (Tamura et al. 1973), and bovine lens (Glässer and Hanson 1964; Hanson et al. 1965; Kretschmer and Hanson 1965; Melbye and Carpenter 1971; Hanson and Frohne 1976; Cuypers et al. 1982). Human, bovine, and porcine lens LAP (hl-, bl-, and plLAP) and bovine and porcine kidney LAP (bk- and pkLAP) show similar immunoreactivity (Taylor et al. 1984a, 1984b; Oettgen and Taylor 1985). LAP protein amounts to approximately 0.05% of human, bovine, and porcine lens protein (Taylor et al. 1984a). LAP3 and Lap3 encode the human and murine LAP proteins, respectively (Cunningham et al. 2015).

LAP is a hexameric protein, whose subunit molecular mass is 52–54 kDa (Melbye and Carpenter 1971; Carpenter and Harrington 1972; Kohno et al. 1986; Burley et al. 1992); protein analysis, cross-linking studies, and X-ray crystallographic studies indicate that LAP is arranged as a dimer of two trimers. There is one active site per subunit, as shown by bestatin (267, Figure 22) binding (Taylor et al. 1993). Each of the six blLAP subunits binds two divalent metal ions: one of the metal binding sites (site 1, Zn488 in blLAP) is readily exchangeable and can bind Zn²⁺, Mn²⁺, Mg²⁺, or Co²⁺, whereas the tight-binding site (site 2, Zn489 in blLAP) binds only Zn²⁺ or Co²⁺ (Carpenter and Vahl 1973; Kim and Lipscomb 1993a; Lowther and Matthews 2002; Cappiello et al. 2006). Metal-ion occupation of both sites is required for activity, and both metal-binding sites are occupied by Zn²⁺ in native blLAP (Carpenter and Vahl 1973; Allen et al. 1983; Kim and Lipscomb 1993b). In native blLAP, Zn488 forms coordinate bonds with the side-chain carboxylate oxygens of Asp255, Asp332, and Glu334 and the carbonyl oxygen of Asp332; Zn489 forms coordinate bonds with the side-chain carboxylates of Asp255, Asp273, and Glu334 and to the side-chain amine of Lys250 (Burley et al. 1992). The distance between the zinc ions is 2.88 Å (Burley et al. 1990).

The substrate selectivity and turnover and susceptibility to inhibition of LAP is influenced by the species of origin, i.e. bovine lens or porcine kidney, and by the metals that occupy the metal-binding sites (Lowther and Matthews 2002). Purified blLAP contains two Zn²⁺ ions per subunit, whereas purified pkLAP contains only one Zn²⁺ ion per subunit (Himmelhoch 1969; Carpenter and Vahl 1973; Van Wart and Lin 1981). Although there are two metal binding sites in pkLAP, the metal affinity of the second site is apparently sufficiently low to prevent purification of the protein with two bound metals (Lin et al. 1988). A particularly interesting finding is that the hydrolysis of Cys-Gly is catalyzed by Mn²⁺Zn²⁺ blLAP but not by Zn²⁺Zn²⁺ blLAP (Cappiello et al. 2006). The k_cat for the hydrolysis of l-Leu-Gly-NHNH-Dns (264, Figure 21) is approximately threefold higher with Zn²⁺Mn²⁺ pkLAP than with Zn²⁺Zn²⁺ pkLAP (Lin et al. 1988). Metal-ion substitution affects both K_m and k_cat (Allen et al. 1983). As indicated above, rates of biotransformation vary with the metal substitution but under in vivo conditions where [S] ≪ K_m the effects of metal substitution are likely to be small (Lin et al. 1988; Lowther and Matthews 2002).

Crystal structures of native pk- and blLAP (Jurnak et al. 1977; Taylor et al. 1984b; Burley et al. 1992) and of blLAP in complex with bestatin (267, Figure 22) (Burley et al. 1990, 1991, 1992; Kim et al. 1993), amastatin (268, Figure 22) (Kim and Lipscomb 1993b), l-leucinal (275, Figure 22) (Sträter and Lipscomb 1995b; Sträter et al. 1999), l-leu phosphonic acid (277, Figure 22) (Sträter and Lipscomb 1995a; Sträter et al. 1999; Grembecka et al. 2003), zofenoprilat (281, Figure 22) (Cappiello et al. 2006), and microginin FR1 (Kraft et al. 2006) have been solved. X-Ray crystallographic studies confirm the hexameric structure of pkLAP (Taylor et al. 1984b). The N-terminal domain (160 amino acid residues) of blLAP is associated with trimer-trimer interactions but is not involved in catalysis (Burley et al. 1992); the C-terminal domain (327 amino acid residues) contains the two zinc ions and is responsible for catalysis. The N- and C-terminal domains are linked by an α-helix. The active sites are located on the interior of the hexamer and line a disk-shaped cavity (Burley et al. 1990); access to the cavity is provided by solvent channels. There are two solvent channels near the zinc ions, the larger of which affords substrate access to the active site (Erhardt et al. 2005; Jaime et al. 2007 (pt iii)).

The proposed catalytic mechanism of LAP involves the interaction of both Zn²⁺ ions, which function in a co-catalytic manner (Kim and Lipscomb 1994; Sträter and Lipscomb 1995b; Lipscomb and Sträter 1996; Lowther and Matthews 2002; Schürer et al. 2002; Weston 2005; Zhu et al. 2012). The two Zn²⁺ ions are located in close proximity (<3 Å) and are associated with a bridging water or a hydroxyl ligand that functions as a nucleophile (Paul et al. 2016). The Zn1 atom serves to polarize the substrate carbonyl group (Sträter and Lipscomb 1995b; Sträter et al. 1999). Zn2 functions to position the substrate in the active site through coordinate bonding with the terminal amino group of the substrate. Attack of the nucleophile on the carbonyl group of the substrate leads to the formation of a gem-diolate transition state, resulting in product formation (Jaime et al. 2007 (pt iii)). The nature of the metallohybrids affect only the formation of the gem-diolate but not the subsequent steps in the catalytic cycle (Zhu et al. 2012). An active site bicarbonate ion functions as a proton donor and protonates the newly released amino group of the product.

4.3.1.2. Substrates

The hydrolysis of amino terminal prolyl di- and tripeptides and polypeptides, e.g. melanostatin (Pro-Leu-Gly-NH₂), by enzymes partially purified from bovine and porcine kidney led to the proposal that a distinct prolyl aminopeptidase (EC 3.4.11.5) catalyzes these reactions (Nordwig and Mayer 1973; Khilji and Bailey 1978). Later studies showed, however, that the putative prolyl aminopeptidase is identical with LAP (Turzynski and Mentlein 1990; Matsushima et al. 1991).

LAP shows high selectivity for peptides with N-terminal, hydrophobic l-amino acid substituents, e.g. l-Met >l-Leu > l-Phe > l-Val > l-Ala > Gly (Hanson and Frohne 1976; Kohno et al. 1986; Cappiello et al. 2004). Peptides with d-amino acids in the P₁′ site, e.g. l-Leu-d-Leu, are cleaved but at very low rates (Smith and Spackman 1955; Yamada and Kera 1998). Although early studies demonstrated the role of APN and DPEP1 in particulate fractions of kidney in the metabolism of Cys-Gly and Cys-Gly S-conjugates, the finding that rat liver cytosolic LAP also catalyzes the hydrolysis of Cys-Gly and Cys-Gly S-conjugates showed that it plays a role in the γ-Glu cycle (Jösch et al. 2003). blLAP catalyzes the hydrolysis of Cys-Gly, which is attributed to the Zn²⁺Mn²⁺ metallohybrid (Cappiello et al. 2004, 2006). The results of a molecular modeling analysis attribute the failure of (Zn/Zn)blLAP to catalyze the hydrolysis of Cys-Gly to the interaction of the thiol group of Cys-Gly with Zn1, which displaces the water molecule required for catalysis; in contrast, the Mn²⁺ ion in (Mn/Zn)blLAP prefers to interact with oxygen- or nitrogen-containing ligands rather than sulfur-containing ligands, which allows catalysis to proceed.

In addition to catalyzing the hydrolysis of Cys-Gly and Cys-Gly S-conjugates, LAP shows broad substrate selectivity for dipeptides and some tripeptides (for compilations of LAP substrates, see Smith and Bergmann 1944; Smith and Slonim 1948; Smith and Polglase 1949; Smith and Spackman 1955; Kohno et al. 1986). Leucinamide is a specific substrate for porcine liver and ELD mouse ascites carcinoma cells LAP (Patterson et al. 1963; Ledeme et al. 1981). The chromogenic substrates l-Leu-4-nitroanilide (243, Figure 21) and l-Leu-β-naphthylamide are poor substrates for human or porcine liver LAP compared with l-leucinamide (Ledeme et al. 1981; Kohno et al. 1986).

Jösch et al. (1998) demonstrated the hydrolysis of the Cys-Gly S-conjugate of monochlorobimane (3, Figure 1) by rat and guinea pig liver homogenates. Studies with rat and guinea pig liver subcellular fractions showed that the highest specific activity for the hydrolysis of bimane-Cys-Gly (3, Figure 1) is found in the cytosolic fraction, although lower activity is present in the microsomal fraction (Jösch et al. 2003); immunoprecipitation, substrate selectivity, and inhibition studies identified LAP as the enzyme that catalyzes the hydrolysis of the Cys-Gly S-conjugate of monochlorobimane (3, Figure 1). The specific activity for the hydrolysis of bimane-Cys-Gly (3, Figure 1) in sinusoidal plasma membranes was low, whereas the specific activity of canalicular plasma membranes amounted to 52–67% of the cytosolic fraction.

pkLAP catalyzes the hydrolysis of Leu-Gly-NHNH-Dns (264, Figure 21) (Lin and Van Wart 1988); the fluorescence changes are attributed to direct excitation of the dansyl group rather than energy transfer between pkLAP and the substrate. The (k_cat/K_m) × 10⁻⁶ (min⁻¹ M⁻¹) for the biotransformation of hydrazide 264 (Figure 21) varies little with metal substitution, i.e. Zn, Mg, Mn, Ni, Cu (Lin et al. 1988).

4.3.1.3. Inhibitors

Bestatin (267, Figure 22) is a potent, nonhydrolyzable inhibitor of LAP (Suda et al. 1976; Umezawa et al. 1976; Rich et al. 1984; Wilkes and Prescott 1985; Kohno et al. 1986). Kinetic analysis shows that bestatin (267, Figure 22) is a slow, very tight binding inhibitor of pkLAP (K_i (k_off/k_on = 0.58 nM) (Wilkes and Prescott 1985). The crystal structure of blLAP in complex with bestatin (267, Figure 22) shows that the α-amino group of the inhibitor interacts with Zn2 and the 2-hydroxyl group binds to both Zn1 and Zn2, and that the Phe and Leu sidechains occupy the S₁ and S₁′ hydrophobic pockets near the active site (Burley et al. 1991, 1992; Kim et al. 1993; Kim and Lipscomb 1994). The photoaffinity label, p-azidobestatin is also a slow, tight binding inhibitor of blLAP (Taylor et al. 1992).

Amastatin (268, Figure 22) is a strong inhibitor of LAP (Rich et al. 1984; Wilkes and Prescott 1985); kinetic analysis shows that it is a slow, tight-binding inhibitor of pkLAP (K_i (k_off/k_on = 3.0 × 10⁻²μM) (Wilkes and Prescott 1985). The slow binding of bestatin (267, Figure 22) and amastatin (268, Figure 22) to blLAP may be attributed interactions between the α-hydroxyl-β-amino acid residue and the P₁–P_3′ residues of the inhibitor (Rich et al. 1984). The crystal structure of amastatin (268, Figure 22) in complex with blLAP is similar to that of bestatin (267, Figure 22) and is supportive of a catalytic mechanism in which the transition state involves the formation of a tetrahedral gem-diolate (Kim and Lipscomb 1993b).

l-Leucinal (275, Figure 22) is a potent transition-state inhibitor of pkLAP (K_i = 6 × 10⁻²μM) (Andersson et al. 1982). The crystal structure of blLAP complexed with l-leucinal (275, Figure 22) shows that the transition-state inhibitor forms a gem-diolate intermediate with one hydroxyl group bridging the active site metal ions and the other hydroxyl group binds to Zn1 (Sträter and Lipscomb 1995b).

L-Leu phosphonic acid (277, Figure 22) is a strong inhibitor of pkLAP (K_i = 0.23 μM) and a weak inhibitor of APN (K_i = 53 μM) whereas the phosphinic acid analog is much weaker (K_i = 87 μM) (Giannousis and Bartlett 1987; Lejczak et al. 1989). The crystal structure of blLAP in complex with phosphonic acid 277 (Figure 22) shows that the amino group interacts with Zn2 (Zn489), that one phosphoryl oxygen forms a bridge with both Zn ions, and that one phosphoryl oxygen binds to Zn1 (Zn488) (Sträter and Lipscomb 1995a). The availability of the crystal structure of blLAP in complex with phosphonic acid 277 (Figure 22) (PDB 1LCP) was exploited in a Ligand Design (LUDI) approach to design new blLAP inhibitors (Grembecka et al. 2000). An extension of these studies identified phosphinate dipeptide transition-state analogs as selective inhibitors of blLAP compared with APN (Grembecka et al. 2003). A range of theoretical studies aimed at predicting characteristics of LAP inhibitors have been reported (Grembecka et al. 1999, 2001, 2002).

3D-QSAR, docking, and toxicology prediction studies on a large panel of isoquinoline and phosphonic and phosphinic acid derivatives as inhibitors of bl- and pkLAP have been conducted, but no biological studies have apparently been reported (Ziemska et al. 2017).

Thio and thiono analogs of LAP substrates and inhibitors have been investigated. l-Leu thiol (274, Figure 22) and l-lysine thiol are weak inhibitors of pkLAP (K_i = >500 μM) (Ocain and Rich 1987). α-Thiolbestatin is a less potent inhibitor of LAP than bestatin (267, Figure 22) (Ocain and Rich 1988). The thiono analog of anilide 243 (Figure 21) is a weak inhibitor of pkLAP (K_i = 0.19 mM) but is resistant to pkLAP-catalyzed hydrolysis (Beattie et al. 1987). The effect of metallohybrid composition on the inhibitory action of the ACE inhibitor zofenprilat (281, Figure 22) on blLAP has been investigated (Cappiello et al. 2006). Zofenprilat (281, Figure 22) is a poor inhibitor (K_i = 1.3 mM) of the (Mn/Zn)blLAP-catalyzed hydrolysis of Cys-Gly but is a potent inhibitor (K_i = 0.8 μM) of the (Zn/Zn)blLAP-catalyzed hydrolysis of Leu-Gly; the inhibition of (Zn/Zn)blLAP by zofenprilat (281, Figure 22) is attributed to the interaction of the thiol group with both Zn²⁺ ions. Indeed, the crystal structure of blLAP in complex with zofenprilat (281, Figure 22) shows that its thiol group displaces the water molecule required for catalysis.

The pentapeptide microginin FR1 inhibits blLAP (IC50 = 1.3 μM) (Kraft et al. 2006). The crystal structure of blLAP in complex with microginin FR1 shows that the amino group of the 3-amino-2-hydroxydecanoic acid sidechain coordinates with one of the active-site Zn ions and that the hydroxyl group interacts with Zn2 and with the active-site water molecule.

4.3.1.4. Tissue and species distribution

LAP is present in all mammalian species studied. The LAP found in bovine and porcine lens and kidney and in human lens all share antigenic determinants, but are not identical (Taylor et al. 1984a); within organs of a single species, however, the LAP proteins appear to be identical.

Human LAP is present in kidney, liver, duodenum, ileum, colon, rectal tissue, submaxillary gland, and lung (Nachlas et al. 1960; Tamura et al. 1975). LAP activities in hepatocellular carcinoma and lung squamous cell carcinoma tissue are lower than that in normal liver and lung tissue, whereas activity in stomach adenocarcinoma tissue is higher than in normal stomach tissue (Tamura et al. 1975). LAP is also found in human peripheral lymphocytes and polymorphonuclear leukocytes (Smith et al. 1983; Kohno and Kanno 1985). LAP is present in porcine muscle but the activity is low compared with kidney (Joseph and Sanders 1966). Rabbit LAP is present in renal proximal tubules, but not in glomeruli (Yang et al. 1993).

4.4. Cellular and in vivo hydrolysis of Cys-Gly and Cys-Gly S-conjugates

Although xenobiotic-derived Cys-Gly S-conjugates are obligate intermediates in the mercapturic acid pathway, few reports are available that identify the dipeptidases that catalyze their in vivo hydrolysis to Cys S-conjugates. Given that at least three peptidases catalyze the hydrolysis of Cys-Gly and Cys-Gly S-conjugates, there is often some uncertainty about the roles of these peptidases in Cys-Gly S-conjugate metabolism.

Jones et al. (Jones et al. 1978; Moldéus et al. 1978; Jones et al. 1979) investigated the fate of GSH conjugates in isolated rat liver and kidney cells: incubation of acetaminophen with liver cells led to the formation of a GSH S-conjugate; when the liver cells were removed by centrifugation and kidney cells were added, there was a rapid formation of the Cys S-conjugate of acetaminophen, which was gradually converted to the mercapturic acid. Addition of Phe-Gly, as a competitive dipeptidase inhibitor, to the incubation mixture inhibited the conversion of the GSH S-conjugate to the cysteine conjugate and led to the accumulation of the Cys-Gly adduct, indicating a role for one or more dipeptidases in the biotransformation of GSH S-conjugates.

Studies of MBD^m1/MBD^m1 mice with the targeted deletion of the membrane dipeptidase show that multiple dipeptidases catalyze the hydrolysis of leukotriene D₄ (162, Figure 19) and cystinyl-bis-glycine (161, Figure 19): rates of hydrolysis of leukotriene D₄ (162, Figure 19) are 86% in small intestine, approximately 35% in lung, heart, and pancreas, and 16% in kidney of the rates observed in wild-type mice (Habib et al. 1998). Similarly, the rate of hydrolysis of cystinyl-bis-glycine (161, Figure 19) in MBD^m1/MBD^m1 mice is about 40% of that observed in wild-type mice.

Perfusion of rat liver with bimane-Cys-Gly (3, Figure 1) led to the formation of bimane-mercapturate (5, Figure 1) (30% of infused amount) and a trace of bimane-Cys (4, Figure 1), indicating the dipeptidase-dependent biotransformation of the Cys-Gly S-conjugate (Jösch et al. 1998).

5. Cysteine S-conjugate N-acetyltransferase (NAT8)¹¹

5.1. Introduction

NAT8 (EC 2.3.1.80) catalyzes the reaction of a Cys S-conjugate (4, Figure 1) with acetyl-CoA to give a mercapturic acid (5, Figure 1). The N-acetylation of Cys S-conjugates is a common reaction for the terminal processing of GSH S-conjugates (Chasseaud 1976). Moreover, biotransformation of nephrotoxic Cys S-conjugates to mercapturic acids is an important detoxication reaction, because the N-acetylated Cys S-conjugates do not undergo Cys S-conjugate β-lyase-dependent bioactivation (Anders and Dekant 1998).

Early studies provided indirect evidence for an N-acetylation step in mercapturic acid formation: dogs given bromobenzene (6, Figure 2) excreted S-(bromophenyl)mercapturic acid (10, Figure 2) (Baumann and Preusse 1879; Jaffé 1879). The conversion of benzyl chloride and S-(benzyl)- l-Cys (241, Figure 21) to S-(benzyl)mercapturic acid (283, Figure 23) by dogs, rabbits, and rats was also reported (Stekol 1938). When S-(4-bromophenyl)-l-Cys or S-(benzyl)-l-Cys (241, Figure 21) was given to human subjects, the corresponding mercapturic acids were excreted in the urine (Stekol 1946). The administration of several Cys S-conjugates to rats, rabbits, or guinea pigs also led to the excretion of the corresponding mercapturic acids (West and Mathura 1954; Bray et al. 1959b); moreover, incubation of the S-conjugates with liver slices also resulted in mercapturic acid formation. Elce (1970) was apparently the first to demonstrate the role of acetyl-CoA in mercapturic acid formation.

Figure 23. NAT8 and aminoacylase substrates. 283, N-acetyl-S-(benzyl)-l-Cys; 284, S-(1,2-dichlorovinyl)-l-Cys (DCVC); 285, N-acetyl-S-(1,2-dichlorovinyl)-l-Cys; 286, S-(2-chloro-1,1,2-trifluoroethyl)-l-Cys (CTFC); 287, N-acetyl-S-(2-chloro-1,1,2-trifluoroethyl)-l-Cys; 288, S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-l-Cys; 289, N-acetyl-S-[2-(fluoromethoxy)-1,1,3,3,3-pentafluoropropyl]-l-Cys; 290, S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-l-Cys; 291, N-acetyl-S-[2-(fluoromethoxy)-1,3,3,3-tetrafluoro-1-propenyl]-l-Cys; 292, 2-bromo-6-(l-cystein-S-yl)hydroquinone; 293, N-acetyl-2-bromo-6-(l-cystein-S-yl)hydroquinone, 294, S-[2-hydroxy-1-(2-oxoethyl)heptyl]-l-Cys (4-HNE l-Cys conjugate); 295, N-acetyl-S-[2-hydroxy-1-(2-oxoethyl)heptyl]-l-Cys (4-HNE l-Cys mercapturic acid); 296; N-[3-(2-furyl)acryloyl]-l-Met; 297, 3-(2-furyl)acrylic acid; 298, 2-acetamidoacrylic acid; 299, 2-aminoacrylic acid.

5.2. Structure and properties

NAT8 was purified approximately three-fold from rat kidney microsomes, but the lability of the enzyme prevented further purification (Duffel and Jakoby 1982, 1985). An approximately 6-fold purification of the enzyme with a detergent-solubilization procedure has been reported, but again considerable activity was lost during purification (Aigner et al. 1996). Protein analysis indicated that NAT8 appeared to be a homodimer with a subunit molecular mass of 34 kDa, although the results of a later study indicated that NAT8 was likely a minor component of the enzyme preparation analyzed by Aigner et al. (Veiga-da-Cunha et al. 2010). An improved, detergent-based purification procedure that yielded an approximately 60-fold purification of pig kidney NAT8 has been reported, but the preparation contained considerable acetyl-CoA hydrolase activity (Kraus et al. 2000).

A search for the N-acetyltransferase that catalyzes the N-acetylation of Lys residues in β-site amyloid precursor protein cleaving enzyme 1 (BACE1) led to the identification of NAT8B (ATase1) and a similar protein designated as NAT8 (ATase2) (Ko and Puglielli 2009). NAT8B and NAT8 catalyze the N-acetylation of Lys residues in CD133 (Mak et al. 2014). Veiga-da-Cunha et al. (2010) cloned NAT8, and its expression in HEK293T cells yielded a 25 kDa protein (227 amino acids) that catalyzed the N-acetylation of S-(benzyl)-l-Cys (241, Figure 21) and LTE₄ (235, Figure 21). hNAT8 has three domains (UniProtKB: Q9UHE5): cytoplasmic domain (residues 1–42); helical transmembrane domain (residues 43–63); and C-terminal topological domain (residues 64–227) that contains the active site. Expression of tagged NAT8 in CHO cells and confocal microscopic analysis with immunolabeling confirmed the ER location of NAT8 (Veiga-da-Cunha et al. 2010).

As many as eight NAT8-related sequences are found in some genomes and some may catalyze the N-acetylation of Cys S-conjugates (Veiga-da-Cunha et al. 2010). Two SNPs, E104K and F143S, were identified in NAT8 and their expression and assay shows activities that differ from the wild-type by no more than two-fold. The gene encoding hNAT8 is located on chromosome bands 2p12-p13.1 (Ozaki et al. 1998).

The N-acetylation of the Cys S-conjugates of the sevoflurane degradation products 287 and 289 (Figure 21) shows considerable interindividual variability: with renal tissue from 20 human subjects there were 6- and 11-fold differences in cytosolic (see below) and microsomal N-acetylation, respectively (Altuntas and Kharasch 2002).

NAT8L (NAT8-like), which shares 30% sequence identity with NAT8, catalyzes the N-acetylation of L-Asp, but Cys S-conjugates are not substrates (Wiame et al. 2010).

5.3. Substrates

A large number of diverse chemicals are metabolized to mercapturic acids in vivo (Chasseaud 1976). NAT8 catalyzes the N-acetylation of many l-Cys S-conjugates, but d-Cys S-conjugates are poor substrates (Green and Elce 1975; Duffel and Jakoby 1982). Representative examples of Cys S-conjugates that are substrates for NAT8 are shown in Figure 23 along with the corresponding product mercapturic acids (Commandeur and Vermeulen 1990; Commandeur et al. 1991; Lau et al. 1995; Birner et al. 1997; Kraus et al. 2000; Altuntas and Kharasch 2002). Although free l-Lys is not a substrate for NAT8, protein-bound l-Lys, e.g. BACE1, is acetylated (Ko and Puglielli 2009; Veiga-da-Cunha et al. 2010).

A good correlation between V_max/K_m for N-acetylation and the substituent lipophilicity of alkyl Cys S-conjugates, as modeled by Hansch π constants, was observed (Duffel and Jakoby 1982). The relevance of substrate lipophilicity in the NAT8-catalyzed acetylation reaction is supported by the linear correlation between logP and V_max/K_m for S-conjugate 286 (Figure 23) and two structurally related haloalkyl Cys S-conjugates (Kraus et al. 2000).

LTE₄ (235, Figure 21) is biotransformed to N-acetyl-LTE₄ in rats and in isolated rat hepatocytes (Denzlinger et al. 1985; Örning et al. 1986; Stene and Murphy 1988). In human subjects given LTE₄ (235, Figure 21), N-acetyl-LTE₄ is formed as a minor metabolite, and its formation was not detected in human hepatocytes (Huber et al. 1990; Sala et al. 1990). Similarly, LTE₄ (235, Figure 21) undergoes N-acetylation by expressed hNAT8 but at a rate about 100-fold lower than S-(benzyl)- l-Cys (241, Figure 21) (Veiga-da-Cunha et al. 2010). Human NAT8 expressed in HEK293T cells catalyzes the N-acetylation of three aromatic Cys S-conjugates, S-(benzyl)- l-Cys (241, Figure 21), S-(4-nitrobenzyl)- l-Cys, and S-(1-menaphthyl)- l-Cys, which exhibit similar K_m and V_max values, although the aromatic groups differ in size and electronic properties (Deol and Josephy 2017).

HNE (17, Figure 4) is biotransformed to a GSH S-conjugate (207, Figure 19), which is hydrolyzed to Cys S-conjugate 294 (Figure 23) that undergoes N-acetylation to give mercapturic acid 295 (Figure 23) (Alary et al. 2003; Mol et al. 2017).

The synthetic homocysteine S-conjugate of propachlor (2-(homocysteine-S-yl)-N-isopropyl-N-phenylacetamide), is metabolized in rats to the mercapturic acid analog 2-(N-acetylhomocysteine-S-yl)-N-isopropyl-N-phenylacetamide (Davison et al. 1992). The metabolism in rats of a GSH S-conjugate to a product that is N-acetylated on the Glu moiety has been reported (Yin et al. 2004); the reaction required microsomal enzymes and acetyl-CoA, but it was not determined that NAT8 catalyzed the reaction.

5.4. Inhibitors

Sulfhydryl reactive agents, e.g. p-chloromercuribenzoate, and N-ethylmaleimide, inhibit NAT8 (Green and Elce 1975; Aigner et al. 1996). N-Acetyl-S-(benzyl)-l-Cys (283, Figure 23) is a product inhibitor of NAT8 (K_i = 310 μM), but the GSH S-conjugate of sulfobromophthalein is not an inhibitor (Okajima et al. 1984). Several anions, including probenecid, iodipamide, sulfobromophthalein, and p-aminohippurate, are weak inhibitors of NAT8 (Inoue et al. 1984; Okajima et al. 1984; Aigner et al. 1996).

A high-throughput screen of 14 400 compounds led to the identification of two compounds, phenoxazine 300 (Figure 24) and naphthoquinone 301 (Figure 24), that inhibit NAT8 (Ding et al. 2012). Both compounds exhibited noncompetitive inhibition kinetics with respect to acetyl-CoA, and both inhibited the acetylation of BACE1 and decreased its level of expression in H4 (human neuroglioma) cells. Inhibitors 300 and 301 (Figure 24) apparently have not been tested with Cys S-conjugates as NAT8 substrates.

Figure 24. NAT8 inhibitors. 300, (6-chloro-5H-benzo[a]phenoxazin-5-one, 301, 2-chloro-3-(2-ethoxyanilino)-1,4-dihydronaphthalene-1,4-dione.

5.5. Tissue and species distribution

Mercapturic acid formation has been demonstrated in several species, including man, dogs, rats, mice, goats, and rabbits, but little mercapturic acid formation is observed in guinea pigs and calves (Stekol 1938, 1946; Bray et al. 1959b; Brüsewitz et al. 1977; Bakke et al. 1988, 1990; Skiles et al. 1991).

The rate of N-acetylation of 2-bromo-6-(l-cystein-S-yl)hydroquinone (292, Figure 23) shows significant species and strain differences: B6C3F1 mice ≈ F344 rats ≈ Sprague-Dawley rats > hamsters > BALB/c mice ≫ guinea pigs (Lau et al. 1995). The N-acetylation of S-(1,2,3,4,4-pentachloro-1,3-butadienyl)- l-Cys is somewhat greater in rat kidney microsomes than in human kidney microsomes (Green et al. 2003).

NAT8 activity is located in liver and kidney microsomes but activity is greater in the kidney than in the liver (Elce 1970; Green and Elce 1975; Hughey et al. 1978; Ko and Puglielli 2009; Veiga-da-Cunha et al. 2010), although the N-acetylation of Cys S-conjugates 288 and 290 (Figure 23) in human kidney cytosol has been reported (Altuntas and Kharasch 2002).

An enzyme located in rat liver and kidney microsomal fractions catalyzes the N-acetylation of several Cys S-conjugates, including S-(benzyl)- l-Cys (241, Figure 21), S-(butyl)-l-Cys, and S-(diphenylmethyl)- l-Cys (Green and Elce 1975). The microsomal location of NAT8 indicates that the cytosolic arylamine N-acetyltransferases NAT1 and NAT2 (EC 2.3.1.5) do not catalyze the N-acetylation of Cys S-conjugates (Weber and King 1981). The human genes encoding NAT8 are expressed predominantly in kidney and liver (Ozaki et al. 1998). NAT8 and NAT8B are expressed in mouse primary neurons and cerebral cortex and in human brain tissue (Ding et al. 2012).

The ER localization of NAT8 has been confirmed (Veiga-da-Cunha et al., 2010). Treatment of rat liver and kidney microsomal fractions with trypsin or chymotrypsin significantly decreases NAT8 activity, indicating that protease-sensitive domains of NAT8 are located on the cytoplasmic surface of the endoplasmic reticulum (ER) (Okajima et al. 1984). Furthermore, nonpermeant S-acetyldextran derivatives of Cys inhibit NAT8 activity, indicating that the active site of NAT8 faces the cytoplasmic surface of the ER. NAT8 catalyzed N-acetylation of lysine residues located in the N-terminal and globular regions of BACE1 has been localized to the lumen of the ER (Costantini et al. 2007); the luminal location requires a transporter for the polar acetyl-CoA, which has been identified as acetyl-coenzyme A transporter 1 (AT-1) (Hirabayashi et al. 2004; Jonas et al. 2010).

NAT8 activity is found, along with GGT and dipeptidase activities in the outer stripe region of the medulla of the rat kidney, which contains proximal tubular cells (Hughey et al. 1978). Immunohistochemical studies show that NAT8 is highly expressed in the proximal tubular cells of the inner and outer cortices of human kidney (Chambers et al. 2010, Supplementary Figures 4 and 5).

NAT8 activity in rat kidney is concentrated in the rat renal proximal tubule (Hughey et al. 1978; Heuner et al. 1991). Microinfusion studies with rat renal tubules and S-(benzyl)-l-Cys (241, Figure 21) showed that N-acetyl-S-(benzyl)-l-Cys (283, Figure 23) formation was greater in late proximal tubules than in early proximal tubules and that no mercapturic acid formation was detected in early distal tubules (Heuner et al. 1991). Infusion of Cys S-conjugate 241 (Figure 23) into the rat renal artery resulted in 90% conversion to mercapturic acid 283 (Figure 23).

5.6. Other biological functions

Apart from its importance in the mercapturic acid pathway, NAT8 may also play a role in regulation of blood pressure and renal function (Juhanson et al. 2008). Genome-wide association studies implicate NAT8 with susceptibility to chronic kidney disease (Chambers et al. 2010; Köttgen et al. 2010; Tin et al. 2013). A genome-wide association study revealed that a genetically determined metabotype is linked to a variation at the NAT8 locus and that this variation is associated with N-acetyl-Orn (Suhre et al. 2011); elevated blood N-acetyl-Orn concentrations are associated with decreased GFR chronic kidney disease. A missense mutation in NAT8 is a known susceptibility locus for chronic kidney disease in African Americans (Yu et al. 2014). The missense mutation is significantly associated with N-acetyl-Orn and N-acetyl-1-methylhistidine levels, which exhibit a predictive relationship with chronic kidney disease. It has been proposed that the mercapturic acid pathway may play a role in the development of diabetic kidney disease due to the tubular cell toxicity of cysteine S-conjugates; NAT8-catalyzed N-acetylation of the conjugates may provide a protective effect (Gonçalves-Dias et al. 2019a).

Hepatic NAT8 activities are decreased after hepatic ischemia/reperfusion injury in rats and can be restored by giving mesenchymal stem cells intravenously (Fu et al. 2014); furthermore, NAT8 overexpression in human hepatic L02 cells blocks hydrogen peroxide-induced apoptosis.

NAT8B- and NAT8-catalyzed transient N-acetylation of Lys residues has been proposed to regulate levels of BACE1 and, thereby, the levels of neurotoxic amyloid β-peptide (Aβ) in the brain. Both NAT8B and NAT8 levels are upregulated in the brains of Alzheimer’s Disease patients and may represent potential drug targets for the prevention of Aβ accumulation (Ding et al. 2012). Evidence also supports a role for NAT8B/NAT8-catalyzed transient N-acetylation of Lys residues in the regulation of expression and trafficking of CD133 protein to the surface of cells (Mak et al. 2014).

6. Aminoacylases

6.1. Introduction

Aminoacylases catalyze the hydrolysis of N-acyl-L-α-amino acids, including mercapturic acids (Figure 1), to yield cysteine S-conjugates, alkanoic acids, and amino acids (for a review, see Anders and Dekant 1994). Most early reports dealt with the hydrolysis of protein N-acetyl amino acids and with the use of aminoacylases to catalyze the enantiospecific hydrolysis of N-acetyl l-amino acids (Birnbaum et al. 1952; Fones and Lee 1953; Gade and Brown 1981). Birnbaum et al. (1952) were apparently the first to use the term “acylase” to describe the enzymatic hydrolysis of N-acyl-l-α-amino acids. Aminoacylases may serve to salvage N-acyl amino acids formed by protein degradation (Endo 1980; Giardina et al. 1997; Pittelkow et al. 1998; Perrier et al. 2005; Lindner et al. 2008b). The spectrum of N-acetyl amino acids found in the urine of human subjects with aminoacylase 1 deficiency (https://www.omim.org/entry/609924; accessed 31 October 2019) affords insight into the role of the aminoacylases in the processing of N-acetyl amino acids derived from N-acetylated peptides (Engelke et al. 2008; Alessandri et al. 2014, 2018).

Early in vivo studies provided inconsistent evidence for the deacetylation of mercapturic acids (Parke and Williams 1951; Marsden and Young 1958), although the hydrolysis of S-(benzyl)mercapturic acid (283, Figure 23), S-(2-chloro-4-nitrophenyl)mercapturic acid, and S-(butyl)mercapturic acid to the corresponding cysteine S-conjugates was demonstrated in rat, rabbit, and guinea pig liver tissue extracts (Bray and James 1960). In vivo studies show, however, that mercapturic acids undergo hydrolysis (and reacetylation): when [acetyl-²H₃]-labeled mercapturic acids were given to rats, a variable fraction of the dose was excreted as the unlabeled mercapturic acid (Commandeur et al. 1991; Uttamsingh et al. 1998a).

Three mammalian enzymes catalyze the hydrolysis of N-acyl α-l-amino acids: aminoacylase-1, aspartoacylase, and aminoacylase-3; all are zinc hydrolases (Hernick and Fierke 2005). Aminoacylase-1 and -3 show a preference for N-acyl aliphatic-l-amino acids and N-acyl aromatic-l-amino acids, respectively, but mercapturic acids may be substrates for either enzyme. Aspartoacylase (EC 3.5.1.15) is highly selective for N-acetyl-l-Asp and has not been reported to catalyze the hydrolysis of xenobiotic-derived mercapturic acids.

6.2. Aminoacylase-1 (Acy1, EC 3.5.1.14)¹²

Acy1 preferentially catalyzes the hydrolysis of aliphatic N-acyl α-l-amino acids, many mercapturic acids, and the l-Cys prodrug N-acetyl-l-Cys.

6.2.1. Structure and properties

An enzyme purified to homogeneity from pig kidney was termed “acylase 1” (Bruns and Schulze 1962). N-Acetyl-l-Met was the preferred substrate but a range of N-acyl-l-aliphatic amino acids were also substrates; N-Acyl-l-aromatic amino acids, e.g. N-acetyl-l-Trp, were poor substrates.

Porcine kidney and human and bovine liver Acy1 is a homodimeric, zinc metallopeptidase (Kördel and Schneider 1976a, 1977b; Gade and Brown 1981; Mitta et al. 1992; Cook et al. 1993; Palm and Röhm 1995; D'Ambrosio et al. 2003). A trimeric aggregate of rat kidney Acy1 has been reported (Giardina et al. 2000). Aminoacylase-1 is assigned to Family M20 (MEROPS ID M20.973) (Rawlings et al. 2018), whose subclasses represent degradative dipeptidases and exopeptidases as well as amidohydrolases (Biagini and Puigserver 2001).

The subunit Mr of Acy1 of human, pork, and rat kidney acylases is approximately 45 kDa (Mitta et al. 1992; Cook et al. 1993; Giardina et al. 2000; D'Ambrosio et al. 2003). Each porcine Acy1 monomer contains two domains: a catalytic domain (residues 1–188 and 311–399 or 311–406) and a dimerization domain (residues 189–310) (Palm and Röhm 1995; D'Ambrosio et al. 2003; Liu et al. 2006). The nucleotide sequences of human, porcine, mouse, and rat Acy1 have been reported (Mitta et al. 1992, 1993; Giardina et al. 2000; Durand et al. 2003; Sass et al. 2006); the deduced amino acid sequences of h-, p-, and rAcy1 show 80–90% identity. ACY1 and Acy1 encode the human and murine proteins, respectively (Cook et al. 1993; Visel et al. 2004). A baculovirus expression system for h- and pAcy1 has been described (Pittelkow et al. 1998).

The gene encoding human Acy1 is assigned to chromosome 3p21.1 (Miller et al. 1990; Cook et al. 1993). All human tissue culture cells and all mouse tissues and cells from a single strain show one electrophoretic form of Acy1 (Naylor et al. 1979). Human subjects with Acy1 deficiency express inactive Acy1 proteins and exhibit clinical symptoms associated with impaired development (van Coster et al. 2005; Sass et al. 2006, 2007; Ferri et al. 2014) (see Section 6.2.5, Other biological functions).

Early studies demonstrated the presence of several Cys residues in Acy1 and that chemical modification of the thiol residues inactivated the enzyme (Kördel and Schneider 1976a; Frey et al. 1977; Szajáni et al. 1980; Heese and Röhm 1989). Later analysis of pAcy1 by several techniques shows the presence of five Cys residues, which most studies show are present as thiols rather than disulfides (Mitta et al. 1992; Miyagi et al. 1992; Palm and Röhm 1995; Wang et al. 1995; D'Ambrosio et al. 2003). Rat and human Acy1 contain five and four Cys residues, respectively (Durand et al. 2003; Perrier et al. 2004). Although certain of the four Cys residues in hAcy1 are conserved in p- and rAcy1, there are differences in the location of the remaining Cys residues within the amino acid sequences. The Cys residues may play a structural or catalytic role (Szajáni et al. 1980; Löffler et al. 1986; Henseling and Röhm 1988; Heese and Röhm 1989). Chemical modification of Cys residues by chloromethyl ketone analogs of l-Ala and l-Leu inactivate pAcy1 (Frey et al. 1977). Cys116 in h- and pAcy1 is thought to play a role in catalysis, but Cys116 is missing in rat renal Acy1, which is much less sensitive to inhibition by mercuric chloride than h- or pAcy1.

Dialysis of Acy1 against 1,10-phenanthroline inactivates the enzyme, but activity is restored by addition of Zn²⁺ or Co²⁺ (Kördel and Schneider 1977b; Kumpe et al. 1981; Wu and Tsou 1993). Replacement of the Zn²⁺ ion with Mn²⁺, Co²⁺, Ni²⁺, or Cd²⁺ in Acy1 results in significant alterations in the kinetic properties of porcine kidney Acy1 with a panel of four N-acetyl-l-amino acids (Gilles et al. 1984). With N-chloroacetyl-l-Leu as the substrate, the K_m and V_max for the native enzyme and the Co²⁺-reconstituted (after inactivation with 1,10-phenanthroline) Acy1 are similar (Wu and Tsou 1993). The sensitivity of native and Mn²⁺-substituted Acy1 to a panel of inhibitors is similar (Heese et al. 1990).

Zinc may play a structural (Heese et al. 1990; Tang et al. 1995) or a catalytic (Szajáni et al. 1980; Lindner et al. 2003) role in Acy1-catalyzed reactions, although the latter is favored (Hernick and Fierke 2005; Auld 2013). Metal analysis indicates that Acy1 contains one Zn²⁺ ion per subunit (Kördel and Schneider 1977b; Löffler et al. 1986; Palm and Röhm 1995; Pittelkow et al. 1998), whereas the crystal structure of the zinc-binding domain of the T347G mutant of hAcy1 clearly shows the presence of two Zn²⁺ ions per subunit (Lindner et al. 2003). The crystal structure of the Zn-binding domain also shows a dinuclear zinc-binding site with a His-Asp-Glu-Glu-His motif characteristic of clan MH members (Lindner et al. 2003). Accordingly, photooxidation or chemical modification of the His residues in pAcy1 results in loss of activity (Kördel and Schneider 1977a). Chemical modification of Trp residues with 5-diazo-1H-tetrazole and N-bromosuccinimide, which do not react with His residues, also inactivates pAcy1 (Kördel and Schneider 1976b).

The reaction mechanism for Acy1 has not been fully elucidated but may resemble that found for other Zn-containing hydrolases (Hernick and Fierke 2005; Auld 2013). The rate-determining step for the Acy1-catalyed hydrolysis of a series of N-acyl-amino acids is consistent with a nucleophilic attack, perhaps assisted by a zinc ion, on the acyl carbonyl carbon of the N-acyl-amino group (Ötvös et al. 1971; Szajáni et al. 1980; Denton 2000). On the basis of mutational studies and homology modeling, Liu et al. (2006) suggest a mechanism for pAcy1 in which the active-site Glu146 acts as a general base to abstract a proton from the zinc-coordinated water; the resulting hydroxide ion may then attack the substrate acyl carbonyl to form a tetrahedral intermediate that eliminates the acyl residue; hence, the Zn²⁺ provides the water-derived nucleophile and stabilizes the active site. The proposed mechanism does not include a role in catalysis for a second active-site Zn²⁺ ion.

6.2.2. Substrates

Acy1 preferentially catalyzes the hydrolysis of N-acyl α-l-amino acids with neutral aliphatic N-acyl substituents, e.g. N-acetyl-l-Met (Birnbaum et al. 1952; Bruns and Schulze 1962; Lorentz et al. 1975; Gade and Brown 1981; Giardina et al. 2000; Durand et al. 2003; Lindner et al. 2008a); N-acetyl-l-Asp, N-acetyl-l-Phe, N-acetyl-l-Tyr, N-acetyl-l-His, and N^α-Acetyl-l-Lys are scarcely hydrolyzed. Dipeptides with uncharged N-terminal amino acids, e.g. Gly-l-Ala, are substrates (Rao et al. 1952; Kördel and Schneider 1975), but α-N-acetyl-peptides, e.g. N-acetyl-Ala-Ala-Ala, are not substrates (Gade and Brown 1978). For compilations of rates of hydrolysis of N-acyl-l-amino acids, see (Birnbaum et al. 1952; Fones and Lee 1953; Fu and Birnbaum 1953; Ötvös et al. 1971; Lorentz et al. 1975; Chenault et al. 1989; Durand et al. 2003).

The synthetic substrate N-[3-(2-furyl)acryloyl]- l-Met (296, Figure 23) is used for the assay of pAcy1 activity but is not biotransformed by r- or mAcy1(Heese et al. 1988; Giardina et al. 1997, 2000; Durand et al. 2003). Acetamidoacrylic acid (298, Figure 23) has been used as an assay substrate for pAcy1 (Cho et al. 1987).

Rat and porcine kidney Acy1 catalyze the hydrolysis of the l-Cys prodrug N-acetyl-l-Cys (Uttamsingh et al. 1998b; Yamauchi et al. 2002). The rate of hydrolysis of N-acetyl-l-Cys is highest in kidney tissue in rabbit, rat, monkey, and dog, but low activity is also present in liver, heart, small intestine, lung, brain, and pancreas in rat, monkey, and dog, but not in the rabbit (Yamauchi et al. 2002). The rate of hydrolysis of S-alkyl-N-acetyl-l-Cys compounds varies considerably: for example, the V_max/K_m (mL/mg/min) ranged from 56.9 for S-(n-propyl)-N-acetyl-l-Cys to 5 for S-(n-butyl)-N-acetyl-l-Cys, whereas the V_max/K_m for N-acetyl-l-Cys is 10.2 (Uttamsingh et al. 1998b). Moreover, the rate of hydrolysis of several S-alkyl-N-acetyl-l-Cys compounds is related to their calculated molecular volumes and log P values.

Xenobiotic-derived mercapturic acids are substrates for Acy1. N-Acetyl-S-(1,1,2,2-tetrafluoroethyl)-l-Cys, N-acetyl-S-(2-chloro-1,1,2-trifluoroethyl)- l-Cys (287, Figure 23), and N-acetyl-S-(2-bromo-1,1,2-trifluoroethyl)- l-Cys are hydrolyzed by purified rat kidney Acy1 (Uttamsingh et al. 1998b).

Rat kidney and liver homogenates catalyze the hydrolysis of N-acetyl-S-(1,2,2-trichlorovinyl)-l-Cys, N-acetyl-S-(1,2-dichlorovinyl)-l-Cys (285, Figure 23), N-acetyl-S-(1,2,3,4,4-pentachlorobutadienyl)-l-Cys, N-acetyl-S-(1,1,2,2-tetrafluoroethyl)-l-Cys, N-acetyl-S-(2,2-dichloro-1,1-difluoroethyl)-l-Cys, and N-acetyl-S-(2,2-dibromo-1,1-difluoroethyl)-l-Cys, but these results do not indicate whether Acy1 or Acy3 (see below) catalyzes the reaction (Vamvakas et al. 1987; Commandeur et al. 1988; Pratt and Lock 1988; Commandeur et al. 1989);

Studies with mercapturic acids that are formed as metabolites of the sevoflurane degradation product 1,1,3,3,3-pentafluoro-2-(fluoromethoxy)-1-propene (Compound A) showed that human kidney cytosol catalyzes the hydrolysis of N-acetyl-S-[1,3,3,3-tetrafluoro-2-(fluoromethoxy)-1-propenyl]- l-Cys (291, Figure 23), but not N-acetyl-S-[1,1,3,3,3-pentafluoro-2-(fluoromethoxy)propyl]- l-Cys (289, Figure 23) (Uttamsingh et al. 1998a); other studies showed, however, that human kidney cytosol catalyzed the hydrolysis of both mercapturic acids 289 and 291 (Figure 23) and that rates of N-deacetylation were greater than rates of N-acetylation (Altuntas and Kharasch 2002).

Human and porcine Acy1 also catalyze the hydrolysis of a mercapturic acid formed from 3,4-epoxyprecocene II epoxide as well as several other synthetic mercapturic acids. (Stocker et al. 2012).

6.2.3. Inhibitors

Acy1 is inhibited by zinc chelating agents, e.g. 1,10-phenanthroline and EDTA, by thiol reagents, e.g. 4-chloromercuribenzoic acid, 2-chloromercuri-4-nitrophenol, mercuric chloride, and 2-mercaptoethanol, by diethyl pyrocarbonate (a histidine reagent), and by 2-ethoxy-1-(ethoxycarbonyl)-1,2-dihydroquinoline (a carboxyl reagent), but not by benzamidine (a trypsin inhibitor), phenylmethanesulfonyl fluoride, iodoacetate, or iodoacetamide (Kördel and Schneider 1977b; Szajáni et al. 1980; Löffler and Schneider 1987b; Heese et al. 1988; Wang et al. 1995; Giardina et al. 1997; Durand et al. 2003). Porcine Acy1 is much more sensitive to inhibition by mercuric chloride than rat Acy1 (Durand et al. 2003). Poorly hydrolyzed Acy1 substrates, e.g. N-isovaleroyl-l-Nva, are competitive inhibitors of rapidly hydrolyzed substrates, e.g. N-chloroacetyl-l-Nva (Ötvös et al. 1971). l-α-Hydroxy- and l-α-fluorofatty acids are noncompetitive, reversible inhibitors of pAcy1 with inhibition constants in the mM range (Tamura et al. 2000).

n-Butylmalonic acid (302, Figure 25) is a slow-binding inhibitor of porcine renal Acy1, whereas benzylmalonic acid (303, Figure 25) is a weak, competitive inhibitor (Röhm 1989). Malonyl glucoside 304 (Figure 25) is a strong competitive inhibitor of pAcy1 (K_i = 11 μM) (Stocker et al. 2005). N-Hydroxyamino acids inhibit Acy1 (Löffler and Schneider 1987a; Heese et al. 1990); N-hydroxyaminobutanoic acid (305, Figure 25) is a potent inhibitor (K_i = 0.07 nM) of pAcy1.

Figure 25. Aminoacylase inhibitors. 302, n-butylmalonic acid; 303, benzylmalonic acid; 304, maakiain 3-O-(6′-O-malonyl-β-d-glucoside (malonyl glucoside); 305, N-hydroxy-α-aminobutanoic acid; 306, 2-(4-ethoxyphenyl)-2-methylsuccinic acid; 307, ebselen; 308, N-(2,3,5-trichloro-4-oxocyclohexa-2,5-dien-1-ylidene)benzenesulfonamide; 309, 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF).

6.2.4. Tissue and species distribution

Acy1 is a cytosolic enzyme (Greenhough and Turner 1991; Cook et al. 1993; Lindner et al. 2000). Although Acy1 has been reported to be associated with cellular membranes, apparently due to the presence of the hydrophobic PWW(L,A) motif (Heese et al. 1988; Palm and Röhm 1995); Acy1 is, however, not anchored to membranes (Greenhough and Turner 1991).

Investigation of the distribution of Acy1 in mammalian tissues by several techniques shows that expression is highest in the kidney (Albert and Szewczuk 1972; Naylor et al. 1979; Qavi and Kit 1980; Csaikl et al. 1986; Miller et al. 1989; Miller and Kao 1989; Lindner et al. 2000; Yamauchi et al. 2002; Cigna et al. 2007; Zhong et al. 2009). Acy1 is also present in liver (bile duct epithelium, hepatocytes), lung (bronchiolar epithelium), brain (gray, but not white, matter), skeletal muscle, stomach, small intestine, adrenal glands, pancreas, and testes, but not in serum or erythrocytes (Naylor et al. 1979; Miller and Kao 1989; Uttamsingh et al. 2000). In contrast to most species studied, little Acy1 activity or protein was found in canine liver tissue (Yamauchi et al. 2002). Acy1 is present in the distal tubules of the porcine kidney and in the glomeruli and proximal and distal convoluted tubules of the rat (Löffler et al. 1982; Uttamsingh et al. 2000).

In man and cynomolgus monkey, Acy1 is localized in the straight and convoluted renal proximal tissues (Yamauchi et al. 2002). Transcriptomic analysis (RNA-Seq) shows high expression of hAcy1 in kidney, liver, small intestine, and duodenum but detectable transcripts are found in most tissues (Fagerberg et al. 2014). Acy1 is detectable in human-mouse somatic cell hybrids but not in mitochondria from these cells (Qavi and Kit 1980).

Acy1 is an abundant protein: the immunoelectrophoretically determined concentration of Acy1 protein amounts to 1.5 mg/g fresh tissue in the porcine kidney compared with 0.4 mg/g fresh tissue in the liver (Löffler et al. 1982); moreover, in the distal proximal tubule the concentration of Acy1 is estimated to amount to 15 mg/g fresh tissue.

Acy1 expression is decreased in ferric nitrilotriacetate-induced renal cell carcinoma tissue in rats (Zhong et al. 2009) but is upregulated in SK-N-SH neuroblastoma cells exposed to neural differentiation agents (Long et al. 2011).

6.2.5. Other biological functions

Aminoacylase-1 deficiency is an autosomal-recessive inborn error of metabolism (entry 609924; https://www.omim.org/; accessed 31 October 2019) characterized by the increased excretion of N-acetyl-L-amino acids and by several developmental disabilities, including muscular hypotonia, seizures, delayed psychomotor development, and autistic behavior (van Coster et al. 2005; Sass et al. 2007; Engelke et al. 2008; Tylki-Szymanska et al. 2010; Sommer et al. 2011; Alessandri et al. 2014; Ferri et al. 2014). Mutational analysis shows loss-of-function mutations associated with abnormal splicing (Sass et al. 2006; Sommer et al. 2011; Ferri et al. 2014). (For a summary of the clinical findings and mutations in aminoacylase 1 deficiency subjects, see Sass et al. 2007; Alessandri et al. 2018).

hAcyl may play a role in cancer-cell growth regulation. Immunoassay shows that hAcy1 expression varies widely in several small cell lung cancer (SCLC)- and non-small cell lung cancer (NSCLC) cell lines (Miller et al. 1989; Scaloni et al. 1992); similarly, hAcy1 activity is positive, reduced, or undetectable in SCLC and NSCLC tissue. Inactivation of ACY1 may afford a selective growth advantage in small cell lung cancer (Cook et al. 1998). A decreased expression of Acy1 is found in lung and renal-cell carcinoma tissues and the decrease may be associated with deletions in the DNF15S2 locus on chromosome 3 (Miller et al. 1989; Naylor et al. 1989; Jones et al. 1991; Balabanov et al. 2001; Hwa et al. 2005). Acy1 may exert a tumor suppressor function in an iron-induced rat model of renal-cell carcinoma (Zhong et al. 2009); transfection of Acy1 into rat renal-cell carcinoma cell lines retards proliferation and colony formation. hAcy1 is overexpressed in colorectal cancer tissue, and its expression in colorectal cancer tissue and its presence in serum is positively correlated with tumor stage, indicating that it may play a role in tumor progression (Shi et al. 2013; Yu et al. 2017). Acy1 mRNA and protein expression are decreased in human hepatocellular carcinoma tissue (Wei et al. 2014); knockdown of ACY1 in human SMC7721 cells promotes cell viability and invasiveness whereas overexpression of Acy1 in human BEL7402 cells inhibits growth. Although Acy1 is scarcely detectable serum of normal subjects, elevated serum Acy1 concentrations are associated with delayed graft function in subjects undergoing renal transplantation (Welberry Smith et al. 2013).

Acy1 may regulate the activity of sphingosine kinase 1 (SPHK1), which promotes cell proliferation and suppresses apoptosis (Maceyka et al. 2004); Acy1 increases the anti-apoptotic effect of SPHK1 in NIH 3T3 cells. Transfection of renal-cell carcinoma-derived cell lines with Acy1 decreases cell proliferation, an effect that may be associated with stimulation of SPHK1 (Zhong et al. 2009).

6.3. Aminoacylase-3 (Acy3, EC 3.5.1.114)¹³

Acy3 preferentially catalyzes the hydrolysis of aromatic N-acetyl α-l-amino acids and some mercapturic acids. Acy3 has not been explicitly assigned to a MEROPS Family or Clan.

6.3.1. Structure and properties

An aminoacylase that preferentially catalyzes the hydrolysis of N-acetyl-l-Tyr and N-acetyl-l-Phe was purified from rat liver and kidney (Endo 1978); the enzyme was designated as Acylase III and was demonstrated to be distinct from Acy1.

An enzyme purified from rat liver and kidney that catalyzed the hydrolysis of N-acetyl aryl amino acids and S-aryl- and S-aralkyl-mercapturic acids was stated to resemble Acy3 (Suzuki and Tateishi 1981); the rat liver enzyme was likely a homotetramer. Subsequent studies identified an Acy3-like enzyme from rat kidney that catalyzed the hydrolysis of haloalkene-derived mercapturic acids (Uttamsingh and Anders 1999).

Murine Acy3 is a homotetramer of approximately 140 kDa with a subunit molecular mass of approximately 35 kDa and catalyzes the hydrolysis of N-acetyl-S-benzyl-l-Cys (283, Figure 23), N-acetyl-l-His, N-acetyl-l-Tyr, and N-acetyl-l-Phe (Pushkin et al. 2004). Murine Acy3 shows 43% and 5% amino acid sequence identity with murine aspartoacylase and Acy1, respectively. Additional studies showed that expressed murine Acy3 is present as both a homodimer and homotetramer (Newman et al. 2007). Acy3 monomers of approximately 75 kDa have been detected in human neuroblastoma-derived cell lines and in rat cortical neurons (Long et al. 2011; Tsirulnikov et al. 2012a). The specific activities of the dimeric and tetrameric forms of Acy3 were similar with N-acetyl-S-benzyl-l-Cys (283, Figure 23) as the substrate (Ryazantsev et al. 2007); electron microscopic analysis and modeling studies at 16 Å resolution indicated that tetrameric Acy3 may be attributed to a back-to-back stacking of dimeric forms of Acy3.

The gene encoding hAcy3 is located on chromosome 11q13.2 (entry 614413; www.omim.org; accessed 31 October 2019). A genome-wide association study found three SNPs on chromosome 11 that mapped to the coding and regulatory regions for ACY3 (Hiltunen et al. 2015).

The mAcy3 active-site residue Glu177 is catalytically essential (Tsirulnikov et al. 2009): the E177A mutant is catalytically inactive. The crystal structure of the E177A mutant of mAcy3 (the structure of wild-type mAcy3 was not resolvable) shows the presence of two domains: a hydrolytic domain (residues 1–120) that contains the catalytic site and a shielding domain (residues 211-318) that limits access to the active site (Hsieh et al. 2010). Acy3 possesses catalytic residues similar to aspartoacylase, and the single zinc ion is coordinated with His21, Glu24, and His116 (Tsirulnikov et al. 2009; Hsieh et al. 2010). In E177A-Acy3 in complex with N-acetyl-l-Tyr, the substrate carbonyl (acetyl) coordinates with the zinc ion and the N-acetyl-α-amino carboxylic acid moiety of the substrate forms hydrogen bonds and van der Waals interactions with active-site residues. Similar interactions are seen with N-acetyl-S-(1,2-dichlorovinyl)-l-Cys (285, Figure 23) in complex with E177A-Acy3. The substrate selectivity of Acy3 is dependent on van der Waals interactions with the substrate side-chain (Hsieh et al. 2010).

Metal analysis indicates that wild-type murine Acy3 expressed in E. coli contains 0.35 zinc ions per subunit (Tsirulnikov et al. 2009). Dialysis against o-phenanthroline inactivates Acy3, whereas incubation with Co²⁺ or Ni²⁺ activates Acy3. Consistent with the X-ray crystal structure of E177A-Acy3, which shows that the zinc ion is coordinated with His21, Glu24, and His116, mutation of these residues to Ala abolishes Acy3 activity, which is not restored by addition of cobalt (Tsirulnikov et al. 2009; Hsieh et al. 2010). It is likely that enzyme-bound zinc participates in mercapturic acid hydrolysis, perhaps by the promoted-water pathway described for carboxypeptidase A (Hsieh et al. 2010; Wu et al. 2010).

6.3.2. Substrates

Acy3 catalyzes the hydrolysis of many N-acyl amino acids, S-aryl and S-aralkyl mercapturic acids, and haloalkene-derived mercapturic acids. With purified rat kidney and liver Acy3, the substrates with the highest specific activities include: N-acetyl-l-Tyr, N-chloroacetyl-l-Trp, N-acetyl-S-(benzyl)- l-Cys, N-acetyl-S-(4-bromophenyl)-l-Cys, and N-acetyl-S-(4-nitrobenzyl)- l-Cys (Endo 1978; Suzuki and Tateishi 1981). With expressed murine Acy3, the substrates with the highest V_max/K_m ratio include: N-acetyl-S-(2,2-dichlorovinyl)- l-Cys, N-acetyl-S-(4-chlorobenzyl)- l-Cys, N-acetyl-S-(4-bromobenzyl)-l-Cys, N-acetyl-S-(pentachlorobutadienyl)-l-Cys, and N-acetyl-S-(1,2,2-trichlorovinyl)-l-Cys (Newman et al. 2007); the Acy1 substrate N-acetyl-l-Cys is not a substrate for Acy3.

Purified or expressed Acy3 catalyzes the hydrolysis of a several haloalkene-derived mercapturic acids (Uttamsingh and Anders 1999; Newman et al. 2007); mercapturic acids 285, 287, 289, and 291 (Figure 23) are representative examples. Cytosolic enzymes catalyze the hydrolysis of bromophenol-derived mercapturic acid 293 (Figure 23), but the enzyme involved was not specifically identified as Acy1 or Acy3 (Lau et al. 1995); given the structure of mercapturic acid 293 (Figure 23), it is likely that Acy3 is involved.

Acy3, but not Acy1, catalyzes the hydrolysis of the mercapturic acids of 4-hydroxynonenal (295, Figure 23) and acrolein (N-acetyl-S-(3-oxopropyl)-l-Cys) (Tsirulnikov et al. 2012a). N-Acetyl-S-farnesyl-l-Cys and N-acetyl-S-geranylgeranyl-l-Cys are substrates for hAcy3, but not hAcy1 (Tsirulnikov et al. 2018); inhibition of Acy3 in human hepatocellular carcinoma cell lines increases the levels of N-acetyl-S-farnesyl-l-Cys.

hAcy3 has been identified as a core binding protein of hepatitis C virus (Chen et al. 2009). Incubation of murine Acy3 with N-acetylated and non-acetylated “short” HCVCP peptides that correspond to the N-terminal sequence (residues 2–11) of the HCVCP and the N-terminal sequence (residues 2–11) of mAcy3 led to the formation of chain-length-shortened peptides, indicating that mAcy3 may exhibit endopeptidase activity (Tsirulnikov et al. 2012b); mAcy3 failed, however, to catalyze the N-deacetylation of N-acetylated-HCVCP and a panel of N-acetylated peptides.

6.3.3. Inhibitors

No selective inhibitors for Acy3 have apparently been reported. 4-Chloromercuribenzoic acid and Cu²⁺ inhibit, but dithiothreitol activates, rat and murine kidney Acy3 (Endo 1978; Newman et al. 2007; Tsirulnikov et al. 2009). A small-molecule screen identified inhibitors of mouse Acy3 with potencies in the nanomolar range (Tsirulnikov et al. 2012a): 2-(4-ethoxyphenyl)-2-methylsuccinate (306, Figure 25) (I₅₀ = 0.63 nM), ebselen (307, Fig 25) (I₅₀ = 1.0 nM), and N-(2,3,5-trichloro-4-oxocyclohexa-2,5-dien-1-ylidene)benzenesulfonamide (308 Figure 25) (I₅₀ = 3.20 nM) inhibited deacetylation of N-acetyl-l-Tyr; in addition, ibuprofen (K_i = 0.04 mM) is a comparatively weak inhibitor of Acy3. Molecular docking experiments indicate that these inhibitors interact with the active site of Acy3. Acy3 inhibitors 306 and 307 (Figure 25) decrease the cytotoxicity of the mercapturic acid of HNE (295, Figure 23) in primary rat cortex neurons, indicating a role for Acy3 in the cytotoxicity of HNE (Tsirulnikov et al. 2012a).

The endopeptidase activity of mAcy3 is inhibited by the serine protease inhibitor 4-(2-aminoethyl)benzenesulfonyl fluoride 309 (Figure 25) (Tsirulnikov et al. 2012b).

6.3.4. Tissue and species distribution

Acy3 is a cytosolic enzyme (Suzuki and Tateishi 1981; Lau et al. 1995; Uttamsingh and Anders 1999; Chen et al. 2009). Acy3 activity is highest in the kidney but is also present in liver, spleen, brain, and thymus (Endo 1980). Northern blot analysis shows that murine Acy3 is abundantly expressed in kidney and to a lesser extent in liver (Pushkin et al. 2004); low levels of transcripts are also present in brain, heart, lung, stomach, small intestine, and testis. Immunohistochemical studies show that Acy3 is most highly expressed in the apical membrane of the S1 segment of the renal proximal tubule, although it is detectable in the cytosol of the S2 and S3 segments of mouse kidney (Pushkin et al. 2004). Immunolocalization studies show that Acy3 is found primarily in neurons but is also present in mouse brain blood vessels (Tsirulnikov et al. 2012a). Acy3 is expressed in the nuclei of several neuroblastoma-derived cell lines, in the human adrenal gland, and in neuroblastoma tumor tissue (Long et al. 2011); Acy3 may play a role in neuroblastoma promotion, and high expression of Acy3 in human neuroblastoma tissue is associated with poor prognosis. Transcriptomic analysis (RNA-Seq) shows high expression of hAcy3 in kidney, small intestine, and duodenum but detectable transcripts are found in most tissues (Fagerberg et al. 2014).

6.3.5. Other biological functions

Acy3 is expressed in human hepatocytes but is highly expressed in hepatocellular carcinoma cell lines and in human hepatocellular carcinoma tissue (Tsirulnikov et al. 2018). Inhibition of Acy3 and suppression of Acy3 expression with siRNA decrease the viability of HuH7 and HepG2 cells. The authors propose that inhibition or silencing of Acy3 prevents Ras membrane association, which is required for cell growth.

SNPs in ACY3 are associated with the ambulatory systolic blood pressure response to bisoprolol, although ACY3 is not known to be associated with blood pressure regulation (Hiltunen et al. 2015). An extension of these studies showed that systolic and diastolic blood pressure responses to bisoprolol were significantly correlated with plasma concentrations of Phe and N-acetyl-l-Phe (Rimpelä et al. 2017); ACY3 variation rs2514036 may serve as a biomarker for response to β-blockers. A genome-wide association study and metabolomic profiling identified a variant in ACY3 that is associated with serum N-acetylphenylalanine levels in African American men (Yu et al. 2014); further analysis showed that the ACY3 SNP is associated with systolic and diastolic blood pressure responses to bisoprolol (Rimpelä et al. 2017).

7. Mercapturic acids: transport and elimination; mercapturic acids as biomarkers; bioactivation by the mercapturic acid pathway

7.1. Transport and elimination

The enzymatic synthesis of mercapturic acids and their renal elimination play central roles in the detoxication of drugs and chemicals. The enzymatic synthesis and processing of mercapturic acids has been discussed in preceding sections of this review. Brief comment about the renal transport and elimination of mercapturic acids and other S-conjugates is merited.

The terms Phase I and Phase II drug metabolism were coined by R. Tecwyn Williams (Williams 1959). Subsequently, the terms Phase 0, Phase III, and Phase IV have been proposed to describe the transport processes that surround the metabolic Phases I and II (for a review, see Döring and Petzinger 2014).

Mercapturic acid and S-conjugate transport involves at least four families of transporters: organic anion transporters (OAT), organic anion transporting polypeptide transporters (OATP), multidrug resistance ABC transporters (MDR), and amino acid transporters (for reviews, see Commandeur et al. 1995; Roth et al. 2012; Makrides et al. 2014; Yin and Wang 2016).

Members of the organic anion transport (OAT) family play a major role in the transport of organic anions in the kidney and other organs (Sweet 2005; Rizwan and Burckhardt 2007; Burckhardt and Burckhardt 2011; Wang and Sweet 2013; McLean et al. 2016; Lozano et al. 2018; Nigam 2018). Both organic anion transporter-1 (OAT1; SLC22A6) and -3 (OAT3; SLC22A8), which are located on the basolateral membrane of epithelial cells, may transport mercapturic acids. There are, however, few explicit examples of mercapturic acid transport by OAT1 or OAT3, although in some studies the inhibition of mercapturic acid transport by probenecid indicated the involvement of a member of the OAT transport family (Lock and Ishmael 1985; Zhang and Stevens 1989). Rat OAT1-expressing X. laevis oocytes facilitate the transport of the mercapturic acid N-acetyl-S-(2,4-dinitrophenyl)- l-Cys (Pombrio et al. 2001); several mercapturic acids, e.g. N-acetyl-S-(2,4-dinitrophenyl)-l-Cys and 283 (Figure 23), are competitive inhibitors of 4-aminohippuric acid transport, indicating that they are likely substrates for OAT1. The transport of mercapturic acid metabolites of alkyl polycyclic aromatic hydrocarbons in HEK293 cells expressing hOAT1 and hOAT3 has been reported (Bakhiya et al. 2007); 1-methylpyrenyl mercapturic acid was transported by both hOAT1 and -3, whereas 1,8-dimethylpyrenyl mercapturic acid was not a substrate for hOAT1 and was a poor substrate for hOAT3. The N-acetyl-l-Cys conjugate of cisplatin (NAC-1) is transported in HEK293 cells overexpressing OAT1 and -3 and is inhibited by probenecid (Hu et al. 2017); moreover, the nephrotoxicity of NAC-1 is decreased in OAT1^−/− and OAT3^−/− mice. The Tyr kinase inhibitor nilotinib and several other Tyr kinase inhibitors inhibit 4-aminohippuric acid and estrone-3-sulfate uptake in HEK293 cells overexpressing hOAT1 and hOAT3, respectively. The mercapturic acid pathway plays a significant role in the disposition of inorganic and alkyl mercury compounds and involves several transport systems, including OAT1 and OAT3 (for reviews, see Bridges and Zalups 2010; Bridges and Zalups 2017; Zalups and Bridges 2018). In addition to mercapturic acids, the Cys S-conjugates S-(1,2-dichlorovinyl)-l-cysteine (284, Figure 23) and S-(2-chloro-1,1,2-trifluoroethyl)- l-cysteine (286, Figure 23) are transported by OAT1 (Dantzler et al. 1998; Groves and Morales 1999; Groves et al. 2003).

Members of the organic anion transporting polypeptide transporter family (OATP; SLCO) are involved in mediating the transport of many drugs (for reviews, see Kalliokoski and Niemi 2009; Hagenbuch and Stieger 2013; Stieger and Hagenbuch 2014; Kovacsics et al. 2017). Given the preference of OATPs for substrates greater than 350 Da, most mercapturic acids are unlikely substrates (Roth et al. 2012). Taurocholate uptake is trans-stimulated by S-(2,4-dinitrophenyl)GSH (12, Figure 4), S-sulfobromophthalein-glutathione (BSP-SG), and S-methyl-GSH, but not by N-acetyl-l-Cys in Xenopus laevis oocytes expressing rat Oatp2 (rOATP1A4) (Li et al. 2000).

Several members of the ATP-binding cassette [ABC] family (multidrug resistant proteins, MRP; P-glycoproteins) facilitate the transport of Cys, Cys-Gly, and GSH S-conjugates (for reviews, see Deeley et al. 2006; Borst et al. 2007; Ballatori et al. 2009; Keppler 2011; Nies 2013). For example, leukotriene C₄ (189, Figure 16), S-(2,4-dinitrophenyl)GSH (12. Figure 4), and GSH S-conjugates of aflatoxin (20, Figure 4), melphalan (27, Figure 4), chlorambucil (29, Figure 4), doxorubicin, and daunorubicin serve as MRP substrates, whereas leukotriene D₄ (234, Figure 21), leukotriene E4 (235, Figure 21), and the Cys-Gly S-conjugates of dietary isothiocyanates are apparently the only examples of Cys and Cys-Gly S-conjugates serving as MRP substrates (Keppler et al. 1998; Priebe et al. 1998; Wijnholds et al. 2000; Paumi et al. 2001; Jedlitschky and Keppler 2002; Callaway et al. 2004). The chelating agents 2,3-dimercaptopropane-1-sulfonic acid (DMPS) and 2,3-dimercaptosuccinic acid (DMSA) are employed to promote the elimination of mercury; MRP2 transports the DMPS- and DMSA-S-conjugates of mercury (Bridges et al. 2008a, 2008b). MRP2 plays a role in the efflux of inorganic and methylmercury S-conjugates from renal proximal tubular cells (Bridges et al. 2011). Several Tyr kinase inhibitors inhibit ABC transporters and have been investigated as targets for anticancer chemotherapy, but inhibition of S-conjugate transport has apparently not been reported (Beretta et al. 2017; Robey et al. 2018).

Renal amino acid transporters facilitate the renal transport of protein amino acids and Cys S-conjugates (Bröer 2008; Silbernagl and Gekle 2008; Makrides et al. 2014). The Na⁺-dependent uptake of S-(1,2-dichlorovinyl)- l-Cys (DCVC) (284, Figure 23) has been demonstrated in rat luminal brush border membrane preparations and in rat renal proximal tubules (Schaeffer and Stevens 1987a; Zhang and Stevens 1989); the basolateral uptake of DCVC (284, Figure 23) in LLC-PK1 cells is attributable to a Na⁺-independent l-like amino acid transport system (Schaeffer and Stevens 1987b). In isolated rat renal proximal tubular cells, both the Na⁺-dependent OAT system and the Na⁺-independent amino acid transport system L are involved in the uptake of DCVC (284, Figure 23), whereas the uptake of the homocysteine (Hcy) analog S-(1,2-dichlorovinyl)- l-Hcy is attributable to transport system L and also to the Na⁺-dependent A and ASC amino acid transport systems (Lash and Anders 1989).

The Na⁺-dependent System B^0,+ and the Na⁺-independent System b^0,+ amino acid transporters catalyze the renal transport of several mercury S-conjugates (Bridges and Zalups 2017). Some methylmercury S-conjugates are substrates for the l-type large neutral amino acid transporters, LAT1 and LAT2, which are expressed in the kidney (Simmons-Willis et al. 2002; Park et al. 2005).

There are a few reports of the urinary excretion of Cys S-conjugates. The luminal uptake of Cys S-conjugates may explain the paucity of reports of the renal excretion of Cys S-conjugates. Calves given GSH S-conjugates excrete Cys S-conjugates, but not mercapturic acids, because of an apparent lack of NAT8 activity (Bakke et al. 1990; Davison 1991).

RLIP76, a Ral-effector binding protein, catalyzes the ATP-dependent transport of S-(2,4-dinitrophenyl)GSH (12, Figure 4), HNE-GSH (18, Figure 4), and several anticancer drugs (Awasthi et al. 2000; Singhal et al. 2009a). RLIP76 protects cancer cells from electrophile-induced apoptosis, and its inhibition or depletion may find therapeutic use in cancer chemotherapy (Singhal et al. 2009b, 2015, 2017; Awasthi et al. 2018; Nagaprashantha et al. 2018).

7.2. Mercapturic acids as biomarkers

Mercapturic acids are formed as metabolites of a host of chemicals and their formation and excretion may serve as biomarkers for exposure (De Rooij et al. 1998; Perbellini et al. 2002; Haufroid and Lison 2005; Kotapati et al. 2015). Reviews of LC-MS methods for quantifying the excretion of mercapturic acids demonstrates the potential utility of mercapturic acid formation and excretion as biomarkers (Mathias and B'Hymer 2014, 2016). LC-MS combined with metabonomic analysis has also been used to detect mercapturic acids as biomarkers (Wagner et al. 2006). A review of the methodologies for mercapturomic profiling has been published (Gonçalves-Dias et al. 2019b). Exposure may, however, be underestimated if several metabolites are excreted and the mercapturic acid is the only metabolite measured (Barr et al. 2007). Moreover, the utility of mercapturic acids as biomarkers depends on their reliable urinary excretion in surveyed populations. Considerable information is available about transporter polymorphisms and studies of associations between transporter polymorphisms and drug disposition have been published, but the effect of transporter polymorphisms on mercapturic acid excretion has apparently not been reported (Fujita et al. 2005; Erdman et al. 2006; Zhou et al. 2015; McLean et al. 2016; Yee et al. 2018). The usefulness of mercapturic acids as excreted biomarkers depends not only on the activities of renal transporters but also on the activities of Phase I and II enzymes, which could increase or decrease the metabolism of a compound to an electrophilic metabolite that undergoes GSH S-conjugate formation and processing by the enzymes of the mercapturic acid pathway.

7.3. Bioactivation by the mercapturic acid pathway

Toxicity associated with the mercapturic acid pathway often depends on the initial formation of GSH S-conjugates and further enzymatic processing by enzymes of the mercapturic acid pathway or by other xenobiotic metabolizing enzymes or both. Some S-conjugates may serve as transport or delivery agents and release toxic compounds or metabolites at the target organ (Baillie and Kassahun 1994). Cys S-conjugates, formed as intermediates in the mercapturic acid pathway or by the aminoacylase-catalyzed hydrolysis of mercapturic acids, may undergo bioactivation by Cys S-conjugate β-lyase to give nephrotoxic metabolites. An extended discussion about the bioactivation of mercapturic acids ispathway is beyond the scope of this review, but several reviews are available (Lock and Berndt 1988; Bakke 1990; Anders 1991; Monks and Lau 1992; Anders and Dekant 1994; Guengerich 2003; Anders 2008; Cooper and Hanigan 2018).

8. Conclusions

The enzymes of the mercapturic acid pathway are well characterized. DNA and protein sequences, including SNPs, along with protein crystal structures are available, as are reports on substrate selectivity, inhibitor sensitivity, and tissue, organ, and species distribution. The roles of the MA pathway enzymes in detoxication are well established.

Future investigations of the MA pathway enzymes are likely to focus on their regulatory functions in biochemical and physiological processes. The regulatory functions may involve catalysis, as in the case of the regulation of inflammasome activation by GSTO1-1 (Hughes et al. 2019), or they may involve non-catalytic processes, such as protein-protein interactions. The elucidation of the mechanisms and of the potential clinical relevance of inhibiting or activating such regulatory processes is likely to involve the design of both active site and non-active site directed inhibitors, which may also be required to be both isozyme-selective and either tissue or cell-selective.

Examples of GSTs that exhibit regulatory functions distinct from their metabolism functions include GSTM2-2, which inhibits calcium release through the cardiac RyR2 ryanodine receptor by an apparently non-catalytic mechanism (Samarasinghe et al. 2015) and GSTP1-1, which exhibits a variety of regulatory functions, including inhibition of apoptosis by forming a non-catalytic complex with non-phosphorylated JNK1α2 (De Luca et al. 2012).

Aminopeptidase N (APN, CD13) is a highly multifunctional enzyme that has been termed a “moonlighting” protein. Its regulatory, signaling, receptor, and hydrolytic functions make it an intriguing potential therapeutic target for more than a dozen human cancers, as well as lymphedema, autoimmune disorders, rheumatoid arthritis, and other pathologies associated with APN overexpression (Amin et al. 2018).

Cysteine S-conjugate N-acetyltransferase (NAT8) has been implicated in the regulation of blood pressure and kidney function, and several studies support a role for aminoacylase-1 (Acy1) as a regulator of tumor growth.

Although MA pathway research is a mature field of investigation, the above-mentioned representative examples indicate that much remains to be learned about the biological significance of MA pathway enzymes.

Declaration of interest

The authors report no conflict of interest. The authors participated in the development of the paper as individual professionals and have sole responsibility for the writing and content of the paper. Neither of the authors has been involved in the last 5 years with regulatory or legal proceedings related to the contents of the paper and neither has received funding in the last 5 years in support of the preparation of the paper.

Supplemental material

Supplemental material for this article is available online here.

Acknowledgments

The authors acknowledge with thanks the helpful comments of the independent reviewers who were selected by the Editor and who were anonymous to the authors. The authors thank Dr. Baojian Wu, Jinan University, for provision of images shown in Figure 3.

Notes

Notes

1 The accepted name is glutathione transferase (https://www.qmul.ac.uk/sbcs/iubmb/enzyme/EC2/5/1/18.html; accessed 31 October 2019), but the term glutathione S-transferase is, however, frequently used. The abbreviation GST for glutathione transferases is retained [Mannervik B, Board PG, Hayes JD, Listowsky I, Pearson WR. 2005. Nomenclature for mammalian soluble glutathione transferases. Methods Enzymol. 401: 1–8.]

2 The GSTs were designated as the Molecule of the Month for August, 2017, by the Protein Data Bank (http://pdb101.rcsb.org/motm/212; accessed 31 October 2019).

3 This section describes properties of the cytosolic GST that are similar among the several classes; properties that are distinctive of the several cytosolic GST classes are discussed in Sections 2.2.6.1 to 2.2.6.6.

4 Although identified as a Sigma-class GST, the accepted name for the enzyme (EC 5.3.99.2) that catalyzes the isomerization of PGH2 to PGD2 is prostaglandin-D synthase (PGDS) (https://www.qmul.ac.uk/sbcs/iubmb/enzyme/EC5/3/99/2.html; accessed 31 October 2019).

5 The terms GSTK and DsbA-L are used synonymously in the scientific literature. Their use in this review reflects their use in the cited publications.

6 γ-Glutamyltransferase is the IUBMB accepted name (https://www.qmul.ac.uk/sbcs/iubmb/enzyme/EC2/3/2/2.html; accessed 31 October 2019); γ-glutamyl transpeptidase is also widely used.

7 Membrane dipeptidase is the accepted name, but renal dipeptidase and dehydropeptidase I are also frequently used; for other names, see: http://www.sbcs.qmul.ac.uk/iubmb/enzyme/EC3/4/13/19.html; accessed 31 October 2019; https://www.ebi.ac.uk/merops/cgi-bin/pepsum?id=M19.001; accessed 31 October 2019.

8 Although membrane alanyl aminopeptidase is the accepted name, aminopeptidase N is widely used in the literature and is used in this review; aminopeptidase M is also used. For other names, see: http://www.sbcs.qmul.ac.uk/iubmb/enzyme/EC3/4/11/2.html; accessed 31 October 2019; https://www.ebi.ac.uk/merops/cgi-bin/pepsum?id=M01.001; accessed 31 October 2019.

9 A commercial preparation of APN (https://www.sigmaaldrich.com/catalog/substance/leucineaminopeptidasemicrosomalfromporcinekidney12345905463111?lang=en&region=US; accessed 31 October 2019) is listed as leucine aminopeptidase but is also identified as aminopeptidase N and M.

10 Leucyl aminopeptidase is the accepted name; for other names, see: http://www.sbcs.qmul.ac.uk/iubmb/enzyme/EC3/4/11/1.html, accessed 31 October 2019.

11 Although cysteine S-conjugate N-acetyltransferase is the recommended name, NAT8 is widely used in the literature and is used in this review; a range of names is found in the literature, including ATase 2, GLA, Camello-like proteins 1 and 4, and TSC501 (https://www.genenames.org/cgi-bin/gene_symbol_report?hgnc_id=HGNC:18069; accessed 31 October 2019).

12 The accepted name is N-acyl-aliphatic-l-amino acid amidohydrolase (http://www.sbcs.qmul.ac.uk/iubmb/enzyme/EC3/5/1/14.html; accessed 31 October 2019). Aminoacylase-1 and acylase-1 are, however, widely used in the literature and will be used in this review.

13 The accepted name is N-acyl-aromatic-l-amino acid amidohydrolase (http://www.sbcs.qmul.ac.uk/iubmb/enzyme/EC3/5/1/114.html; accessed 31 October 2019) but the name aminoacylase 3 is used in the literature and in this review.