Omptin proteins: an expanding family of outer membrane proteases in Gram-negative Enterobacteriaceae (Review)

Abstract

The Escherichia coli K-12 outer membrane protein OmpT is a prototype of a unique family of bacterial endopeptidases known as the omptins. This family includes OmpT and OmpP of E. coli, SopA of Shigella flexneri, PgtE of Salmonella enterica, and Pla of Yersinia pestis. Despite their sequence similarities, the omptins vary in their reported functions. The OmpT protease is characterized by narrow cleavage specificity defined by the extracellular loops of the β-barrel protruding above the lipid bilayer. It employs a distinct proteolytic mechanism that involves a histidine and an aspartate residue. Most of the omptin proteins have been implicated in bacterial pathogenesis. As a result, the omptins are potential targets for antimicrobial drug and vaccine development. This review summarizes recent developments in omptins structure and function, emphasizes their role in pathogenesis, proposes evolutionary relation among the existing omptins, and offers possible directions for future research.

• outer membrane protein
• OmpT
• aspartate protease
• Gram-negative bacteria
• Enterobacteriaceae
• β-barrel

Introduction

Omptins have been identified in several Gram-negative bacteria. These proteins are embedded in the bacterial outer membrane and folded in a conserved 10-stranded β-barrel structure [1]. Their activities and specificities are determined by the long, extracellular loops that connect the strands of the β-barrel [2]. Most of the known omptins are bacterial virulence factors and function as either proteases, adhesins, or invasins (Table I) [3], [4]. E. coli K12 expresses the temperature-regulated OmpT [5] and its close homologue OmpP [6]. Yersinia pestis, the causative agent of plague, possesses a highly virulent plasminogen activator Pla [7]. PlaA, a close homologue of Pla, is found in the plant pathogen Erwiniapyrifoliae. Three similar forms of SopA (IcsP) are present in three oral-faecal pathogens: enteroinvasive E. coli, a major cause of diarrhea and death among children in the developing countries, Shigella flexneri and Shigella dysenteriae which cause bacterial dysentery [8]. Salmonella enterica expresses the two homologous forms of PgtE (also referred to as the E protein) [9].

Table I.  Characteristics of the proteins of the omptin family.

Protein	Host organism	Identity to OmpT (%)	Size of mature protein (kDa)	Roles in pathogenesis
OmpT	Escherichia coli K12	100	33.4	Cleaves proteamine, the antimicrobial peptide, secreted by the urinary tract epithelial cells; processes some autotransporter proteins
OmpP	Escherichia coli K12	73	33.1	Processes some autotransporter proteins
SopA/IcsPSopA2/SopA3	Shigella flexneri enteroinvasive E. coli Shigella dysenteriae	60	33.4	Cleaves IcsA autotransporter, ensuring polar localization of IcsA; required for proper actin-based motility inside host eukaryotic cells
PgtE/PgtE2	Salmonella enterica	50	32.8	Cleaves α-helical cationic antimicrobial peptides secreted by epithelial cells
Pla	Yersinia pestis	50	32.6	Activates plasmin; inactivates α2-antiplasmin (plasmin inhibitor); cleaves Complement C3 component; mediates bacterial adhesion to ECM of eukaryotic cells; mediates bacterial invasion into eukaryotic cells
PlaA	Erwinia pyrifoliae	47	32.3	N/A
Q8ZGQ6	Yersinia pestis	40	32.4	Contributes to infection establishment in C. elegans

Thus, omptin proteins are present in a number of human and plant (e.g., Erwinia) bacterial pathogens. Many are important contributors to bacterial virulence. A decreased virulence of the bacterial pathogen was seen upon the loss of a gene coding for Pla omptin [10–12]. Because of their pathogenic potency, the design of specific protease inhibitors against omptin proteases would offer promising leads for new antimicrobial therapeutics. Besides proteolysis, proteolytically-inactive omptins can carry functions such as adhesion and invasion [13]. Therefore, vaccine development against omptins is another fascinating avenue of research. A recent x-ray crystal structure of one of the omptins [1] offers an opportunity for structure-based drug design.

This review discusses the crystal structure of OmpT, which led to elucidation of its proteolytic mechanism and the characterization of its cleavage specificity. The review proceeds to the overview of diverse functional characteristics of the omptin family members. Then, the contribution of the omptins to the bacterial pathogenesis is provided. This review continues with the evolutionary analysis of the omptins. Finally, the review concludes by laying out the directions for future research.

Structural characteristics of OmpT, Pla, and PgtE

OmpT is the only member, whose structure has been elucidated [1]. Its crystal structure, resolved at 2.6 Å, reveals a transmembrane β-barrel (Figure 1A). The β-barrel structure is a conserved feature among the integral membrane proteins found in the outer membrane of the Gram-negative bacteria. This vase-shaped barrel consists of 10 antiparallel β-strands connected by four short periplasmic loops (T1–T4) with no known activity and five longer extracellular loops (L1–L5) [1], [14]. The extracellular loops (Figure 1B) are responsible for protein's activity and specificity [2]. The active site of OmpT is located in the grove formed by these extracellular loops (Figure 1A). Like other transmembrane proteins, OmpT has two rings of aromatic amino acid residues at the water-lipid interface [1]. The protein extends about 40 Å above the lipid bilayer with several extracellular loops extending just above the outer edge of LPS.

Figure 1.  Crystal structure and membrane topology of OmpT (1I78). Adapted and reproduced with permission from reference [1]. (A) Front and side views of OmpT based on its crystal structure. (B) Topology model of OmpT. Amino acids in squares form β-strands. Periplasmic loops are labeled T1–T5. Loops exposed to the extracellular space are labeled L1–L5. The position of outer membrane is highlighted in gray. This Figure is reproduced in color in Molecular Membrane Biology online.

The association with LPS molecules is an absolute requirement for the proteolytic activities of OmpT, Pla, and PgtE [15], [16]. Activation by LPS might be a safety mechanism to ensure that the protease does not become active in the bacterial cytoplasm or periplasm, and is only active when inserted into the outer membrane [15]. Circular dichroism (CD) studies [15] indicate that OmpT-LPS interaction does not cause major changes in the protease structure. The interactions between Arg₁₃₈, Arg₁₇₅ and the 4′ phosphate of lipid A result in a proper alignment of the residues of the active site [15]. To bring OmpT to its full enzymatic potential, the LPS molecule must have heptose-bound phosphates and a fully acylated lipid A [14], [15].

Recent research [16], [17] has indicated that LPS may have a dual function: it can activate and inactivate the omptin proteins. O-antigen in the outer core of LPS sterically inhibits Pla and PgtE proteins from cleaving larger substrates. However, OmpT targets short peptides which are small enough not to encounter steric hindrance from O-antigen. Indeed, Y. pestis, a host organism of Pla, does not produce O-antigen, due to mutations in its genes coding for the biosynthesis of O-antigen pathway [16]. S. enterica, a host organism of PgtE, reduces the length of O-antigen and regulates the expression of pgtE gene during its intracellular growth in macrophages [18].

The FhuA-LPS crystal structure, the only outer membrane protein structure crystallized with LPS, reveals an existence of the conserved LPS-binding motif present on the surface of FhuA [19]. The comparison between FhuA and OmpT structures [1] lead to a hypothesis that the OmpT amino acids associating with LPS are Tyr₁₃₄, Glu₁₃₆, Arg₁₃₈, Arg₁₇₅, and Lys₂₂₆ (Figure 1A). The accuracy of this theoretical prediction is supported by mutation data [16]. Other omptins have slightly different LPS binding regions (Table II). Even the closest homologue of OmpT, OmpP, has His instead of Arg₁₇₅ and Ile instead of Lys₂₂₆. Both forms of SopA have Leu in place of Arg₁₇₅. Pla and PlaA have Ser in place of Lys₂₂₆. Two forms of PgtE have Glu in place of Lys₂₂₆. Of all the amino acids mediating LPS association, Tyr₁₃₄ and Glu₁₃₆ are the most conserved and are present in nine out of 11 omptins. Arg₁₃₈ is conserved in eight of the omptins. The least conserved amino acids are Arg₁₇₅ and Lys₂₂₆.

Table II.  Amino acids with the known functions among the omptin proteins.

	OmpT	OmpP	SopA 1/2/3	Pla/PlaA/AAN04526.1	PgtE/PgtE2	lpl2311	EAN05038.1	CAG37431.1	mlr6132	Q8ZGQ6/CAH20506.1	AAW85741.1	AAG31039
Catalytic amino acids
Asp 83	+	+	+	+	+	+	+	+	+	Asn	Glu	Ile
Asp 85	+	+	+	+	+	+	+	+	+	+	Arg	Gln
Asp 210	+	+	+	+	+	+	+	+	+	+	+	Ser
His 212	+	+	+	+	+	+	+	+	+	+	+	Ile
Active site amino acids
Glu 27	+	+	+	+	+	+	Ala	+	+	+	+	+
Ser 40	+	+	+	+	+	+	+	+	+	+	+	+
Leu 42	+	+	+	+	+	+	+	+	+	+	+	Lys
Met 81	+	+	+	+	+	+	+	+	+	+	His	Val
Asp 97	+	+	+	+	+	His	His	His	Asn	+	+	Gln
Ser 99	+	+	+	+	+	+	+	+	+	+	+	+
His 101	+	+	+	+	+	+	+	Asp	+	+	Ser	+
Ala 143	+	+	+	+	+	+	Ser	Ser	+	+	Asp	+
Tyr 150	+	+	+	+	+	+	+	+	+	+	+	+
Ile 170	+	+	+	+	+	+	+	+	+	Val	+	Val
Asp 208	+	+	+	+	+	+	+	+	Ala	Ala	+	+
Ser 223	+	+	Asp	Glu	Glu	Glu	Asp	Lys	Asp	Asp	Lys	Asp
Thr 263	+	+	+	+	+	+	+	+	Tyr	+	Met	Val
Ala 280	+	+	+	+	+	+	+	+	Ile	Ser	+	Asp
Ile 282	+	+	+	+	+	Leu	Ala	Gly	Gly	+	Leu	+
LPS Binding region
Tyr 134	+	+	+	+	+	+	Leu	Leu	Phe	+	Phe	Phe
Glu 136	+	+	+	+	+	Trp	Tyr	Gln	Tyr	+	Tyr	Lys
Arg 138	+	+	+	+	+	Ser	Asp	Asn	Ser	Asn	Gln	Asn
Arg 175	+	His	Leu	+	+	Lys	+	Trp	Lys	Lys	Glu	Asp
Lys 226	+	Ile	+	Ser	Glu	Asn	+	Asp	Glu	+	Glu	Thr

Proteolytic characteristics of the omptins

OmpT as an aspartate protease

Earlier investigations [20] supported the hypothesis that OmpT was a member of serine protease family because (i) it could be partially inhibited by serine protease inhibitors, such as diisopropylfluorophosphate (DFP) and phenylmethanesulphonylfluoride (PMSF) [21], and (ii) a Ser₉₉→Ala mutation decreased OmpT proteolytic activity 500 fold. However, based on the crystal structure analysis [1], OmpT was reclassified as an aspartate protease. The analysis of structural and mutagenesis studies suggested that the omptin family proteases utilize a unique mechanism that combines the elements of both serine and aspartate proteases [22]. On one hand, OmpT catalytic couple Asp₈₃-Asp₈₅ shares similarity to a catalytic site of aspartate proteases (Figure 1 A, B). On the other hand, the OmpT catalytic couple Asp₂₁₀-His₂₁₂ is comparable to the catalytic triad of serine carboxyl proteases (Figure 1 A, B). It has been proposed that the role of the catalytic site Ser, absent in omptins, is performed by a water molecule present in the catalytic site of the omptins [1]. In the proposed mechanism, catalytic residues of OmpT activate a water molecule that acts as a nucleophile to attack the scissile bond of the substrate. This water molecule plays the role of Oγ of the catalytic serine in the serine proteases. The Asp₈₃-Asp₈₅ couple contributes to a proton translocation or a stabilization of an oxyanion intermediate [1], [22].

The sequence alignment analysis of omptin proteases shows that the catalytic amino acids are conserved among the omptins. However, there are several homologues, which have mutations in the catalytic site residues (Table II). Overall, the proteolytic mechanism is highly conserved among the proteases of the omptin family.

The active site of OmpT is surrounded by the following amino acids Glu₂₇, Ser_40, and Leu₄₂ in L1; Met₈₁, Asp₉₇, Ser_99, and His₁₀₁ in L2; Ala₁₄₃, Tyr₁₅₀, and Ile₁₇₀ in L3; Asp₂₀₈ in L4; Thr₂₆₃, Ala₂₈₀, and Ile₂₈₂ in L5, which are well conserved (Table II). However, different amino acids in the active site can lead to a difference in the cleavage specificity. A recent report [23] states that a change of the amino acid from Ser to Arg at the position 223 alters drastically cleavage specificity of OmpT from Arg-Arg to Ala-Arg Comparison of amino acid residues at this position among the omptins shows variation (Table II) in agreement with the experimental findings.

Ionic interactions appear important in cleavage specificity because the negative environment surrounding the active site leads to a high selectivity of OmpT towards more basic targets [14]. Significance of the electrostatic interaction between OmpT and its substrates is highlighted by the evidence that increased ionic strength, high pH, and acidic amino acids at the cleavage site of a substrate are able to decrease the OmpT affinity for its substrate [14], [24], [25].

Narrow cleavage specificity of OmpT

The OmpT protease has a narrowly-defined cleavage specificity. OmpT targets sites that consist of two basic amino acids, its main substrate is proteamine P1 [26]. Proteamine P1 is 50 amino acids long peptide with 24 arginines and no lysines. The high sequence homogeneity among its targets is consistent with the narrow cleavage specificity of OmpT.

Using Schechter and Berger nomenclature [27], the OmpT recognition site on the surface of a substrate protein can be designated as … P3-P2-P1-P1’-P2’-P3’ … and so on, where the proteolytic cleavage occurs between P1 and P1’. OmpT has an absolute requirement for a basic amino acid at position P1 [24], [25], [28]. On the other hand, the amino acid at the P1’ position is more flexible. Although Lys and Arg are the most preferable for cleavage, amino acids with small non-acidic side chains are also good substrates. The flexibility at the P1’ position helps to explain the capability of OmpT to cleave plasminogen at the Arg-Val site [28], while D-Arg at P1’ is an inhibitor of OmpT activity [24]. Selectivity of OmpT diminishes with the distance from the scissile bond. At the P2’ site, OmpT prefers hydrophobic amino acids, such as Ile or Val. OmpT prefers Ala at the P2 position, but a variety of non-acidic amino acids are tolerated. At the P3 site, Arg and Trp are the residues of choice. The P3’ site can accommodate large amino acids like Trp, Arg, and Thr. Both P4 and P6 sites prefer a basic amino acid and disallow acidic residues [25]. Overall, OmpT has a strong preference for the positively charged amino acids (Table III). Such specificity strengthens the hypothesis that the long-range electrostatic interactions between OmpT and its substrates play an important role in the recognition of substrates.

Table III.  Summary of preferable amino acids in the recognition site of OmpT substrate.

Position of substrate amino acid	Amino acids permitted in a descending order.	Comments
P6	Lys, Arg > >Asn, Gln > Ala.	Positively charged amino acids are preferred; acidic amino acids are not allowed.
P4	Arg, Lys > >Gly.	Positively charged amino acids are preferred; acidic amino acids are not allowed.
P3	Arg > Trp > Val
P2	Ala > Ile, Phe > Leu > Val > Pro > His > Gly > Arg.	Broad tolerance; acidic amino acids are not allowed.
P1	Arg or Lys.	Highly conserved.
P1’	Lys > Ile > His > Arg.	Acidic amino acids are not allowed.
P2’	Ile, Val > Ala.	Amino acids with small side chains are preferred.
P3’	Trp > Arg > Thr.	Amino acids with large side chains are preferred.

OmpT-Pla chimeric proteins [2] demonstrate that the substrate specificity varies from one omptin to another via the sequence variability in the outer loops. The substitution of the C-terminal part of L4 of OmpT with the Pla sequence confers a Pla-like ability to efficiently cleave plasminogen to OmpT. The additional substitution of L3 and L5 gives OmpT capability to inactivate α₂-antiplasmin via cleavage [2]. Substrate's size also influences the rate of cleavage. Consequently, OmpT prefers to cleave small peptides, while Pla targets larger proteins, and PgtE cleaves both [29].

Adhesive properties of the omptins

Several omptins have been implicated in adherence to eukaryotic cells. Pla of Y. pestis, PgtE of S. enterica, and OmpT of E. coli are the best characterized omptin adhesins [13], [16], [30–32]. The adhesive activity of Pla is distinctly separated from its proteolytic activity because proteolytically inactive Pla Ser₉₉→Ala and Asp₂₀₆→Ala mutants still retain their ability to adhere to human endothelial cell lines [13]. Y. pestis binds to the cells lines derived from human epithelial and endothelial tissues through Pla. Even the cell components, such as basement membrane preparations (Matrigel) and mammalian extracellular matrix (ECM) preparations, are sufficient for Pla-expressing Y. pestis or E. coli cells to bind [13], [29], [31]. Pla exhibits a strong adherence to a murine laminin, a medium adherence to a murine heparan sulfate proteoglycan, and a weak adherence to human collagen types IV, I, V, human cellular, and plasma fibronectins [31]. Inhibition studies [30] reveal a lectin-like nature of the Pla-mediated adherence to galactose-rich carbohydrates, present in the glycoproteins, laminin and collagen IV, and the glycolipid globotetraosylceramide. Adherence to laminin is significant not only because it is very strong, but also because laminin is a highly abundant component of the extracellular matrix. Laminin provides an ample opportunity for bacterial adhesion to occur during infection.

PgtE of S. enterica is also shown to promote bacterial adhesion to a human endothelial cell line, almost as efficiently as Pla [16]. OmpT-directed bacterial adherence is four-five times lower than that of Pla-expressing E. coli cells but is still significantly higher than seen in E. coli cells lacking OmpT [31]. Also, OmpT-mediated E. coli adhesion to basement membrane preparation is significantly higher than to albumin-coated surfaces [16]. Furthermore, studies have shown that the purified OmpT binds weakly to purified human laminin and fibronectin, the two extracellular proteins commonly found coating human cells [32]. From the data it appears that Pla and PgtE are strong adhesins, while OmpT exhibits the least adhesive activity among the three omptins examined [16], [29–31].

Invasin activity of Pla of Y. pestis

Pla of Y. pestis is the only omptin shown to be an invasin (Table I). It has been shown that Pla is capable of promoting bacterial invasion of the human epithelial and endothelial cells [16]. Gentamycin protection assays [33] show that the majority of Pla-expressing Y. pestis cells invade human pneumocyte and HeLa cell lines within 1 h of infection. The presence of the Pla-encoding plasmid, pPCP1, is responsible for 90–95% of Y. pestis invasiveness measured against a strain that lacks the pPCP1 plasmid [33]. The pivotal role of Pla in the bacterial invasion is highlighted by the ability of Pla to confer invasiveness to non-invasive E. coli strains. E. coli cells with cloned Pla invade HeLa monolayers in a time-dependent manner, similar to Y. pestis. Moreover, proteolytically inactive Pla mutants Ser₉₉→Ala and Asp₂₀₆→Ala retain their ability to confer invasiveness to E. coli cells [13]. As in the case of its adhesion properties, Pla's ability to invade is independent of the protein's proteolytic activity.

The internalization of Y. pestis depend on actin and less on microtubules [33]. Studies with Pla-expressing E. coli have begun to dissect the bacterial invasion mechanism by determining the host factors involved. Therefore, a several fold decrease in the bacterial invasion following the staurosporin, a broad-spectrum protein kinase inhibitor, treatment suggests the involvement of several classes of host protein kinases. Further work with a more specific protein kinase inhibitor, genistein, indicates that some tyrosine protein kinases are involved. Phosphatidyl-inosinol-4-kinase, small GTPase Rho, and 5-lipoxygenase have been implicated in Pla-promoted bacterial internalization. A drastic effect of Cytochalasin D, an inhibitor of actin polymerization, on invasion strongly implies the involvement of actin in Pla-promoted invasion. Furthermore, invasion is immediately preceded by a burst of cytoskeletal rearrangement inside the target human epithelial cells [34]. Pla targets eukaryotic signal transduction pathways responsible for transducing extracellular signals to the eukaryotic cell and reacting to these signals by cytoskeleton rearrangement.

PgtE of S. enterica and OmpT of E. coli are the other omptins, whose invasin properties have been tested. Neither can invade endothelial cell lines, despite the ability of PgtE to mediate significant bacterial binding to human endothelial cells [29].

Coagulase activity of Pla of Y. pestis

The only omptin whose coagulase activity has been investigated is Pla of Y. pestis. Pla has been reported to coagulate rabbit plasma [35], [36]. The positive results require a long incubation time and a significant number of Y. pestis cells, which are unlikely during a rapidly disseminating plague infection [7]. Insensitivity of Pla-promoted coagulation to thrombin inhibitors suggests that Pla initiates the process not by thrombin activation, but by the direct cleavage of rabbit fibrinogen. It is likely that Pla fibrinogen cleavage sites are unique to rabbits, because Pla cannot cleave and coagulate plasma from human, rats, or mice [7]. The Pla coagulase activity is unlikely to be significant to the virulence of Y. pestis in these hosts.

Role of the omptins in bacterial pathogenesis

OmpT of E. coli

OmpT is an outer membrane protein T (temperature-regulated), previously known as protein A and protease VII [21] whose expression is upregulated at the higher temperatures. OmpT is detected in both nonpathogenic (e.g., K12), and human uropathogenic E. coli strains (UPEC). The ompT gene is detected in 83.1–88.5% of E. coli isolates from urinary tract infections (UTI) and in 67.7% of faecal E. coli isolates [37]. All UTI E. coli isolates had a statistically higher prevalence of ompT gene compared to the faecal E. coli isolates, based on a pairwise z-test with a 5% margin of error. The majority (>50%) of human E. coli isolates have the ompT gene. However, the proteolytic activity of OmpT is detected in E. coli samples from urine (13%) and wounds (19%) [38]. A comprehensive study [39] of E. coli isolates from neonatal bacterial meningitis in the Netherlands finds ompT DNA in 96% of the isolates. In the UPEC strains, OmpT is reported to cleave proteamine P1, an antimicrobial peptide is secreted by epithelial cells of the urinary tract. Protamine associates with DNA in the nuclei of sperm of various animal species and is a part of innate immunity. This peptide is secreted by host cells in the effort to kill the invading bacterial infection. It is thought that it acts on the bacterial cytoplasmic membrane and targets phospholipids in the membrane. However, the details of its mechanism are still unknown. Studies have shown that it disrupts bacterial protein synthesis, diminishes cellular ATP content, inhibits amino acid transport, disrupts oxygen consumption, and causes leakage of intracellular components [40], [41]. OmpT-mediated deactivation of proteamine P1 contributes to bacterial survival inside the host and classifies OmpT as a virulence factor (Table I) [26].

As an outer membrane protease, OmpT processes some of the autotransporters that are shown to be virulence factors [42–44]. Reported examples of such autotransporters are IgAp, IcsA, PrtS [42], and YapA [45]. When autotransporters, with the OmpT recognition site, are introduced into the OmpT-expressing E. coli strains, they are processed and released into the extracellular space (Table I).

Purified OmpT can slowly cleave plasminogen [28], releasing active host serine protease plasmin [46]. However, the plasmin activated by OmpT is quickly inhibited by α₂-antiplasmin, a natural plasmin inhibitor [2]. The limited plasmin activation might be sufficient for the UPEC cells to penetrate beneath the upper layer of cells of the urinary tract to establish an infection [38], but insufficient to promote an invasive infection.

The ability of OmpT to adhere to basement membrane preparations [16], extracellular matrix proteins [32], and human epithelial cells [13] may contribute to bacterial cell attachment to host epithelial tissues (e.g., urinary tract) and the establishment of a persisting bacterial infection. Moreover, it has been recently shown that the purified OmpT induces the production of cytokine TNFα in mononuclear cells [47]. Identity of this receptor on the surface of mononuclear cells has not been achieved.

Pla of Y. pestis

Plasminogen activator (Pla) is a 292 amino acid protein encoded by the pPCP1 plasmid of Y. pestis[13]. Despite its small size (9.6 kbp), this plasmid is crucial for a rapid dissemination of bacteria throughout a host body from a subcutaneous infection. Bacteria that express Pla are highly virulent yielding an LD₅₀ increase of 10⁶ fold in the absence of Pla [7], [10], [11]. It has been recently shown that the expression of Pla is needed for establishment of pneumonic plague [12] and is necessary to initiate bubonic plague [11]. Furthermore, pla has been found to be one of the most up-regulated genes of Y. pestis isolated from the bubo [48].

The proteolytic activity of Pla may help Y. pestis to overcome barriers, such as fibrin clots set up by a host to limit the spread of the infection. Pla is not able to dissolve fibrin clots, cleave laminin, nor cause a degradation of ECM human preparations and basal membrane murine preparations. At the same time, Pla can induce these processes by activating plasmin. Pla activates the human plasmin from a plasminogen precursor via proteolytic cleavage [31]. Pla cleaves plasminogen at the same site and with the same efficiency as the host plasmin activators (e.g., urokinase) [7]. The host has to tightly control the amount of active plasmin since plasmin has a broad proteolytic specificity. Host α₂-antiplasmin and aprotinin deactivate free plasmin, however, to extend plasmin activity, Pla cleaves and inactivates α₂-antiplasmin [31]. Furthermore, the plasmin is likely to remain bound to the bacterial surface to ensure its activity. This suggests that Y. pestis uses Pla to overcome host barriers and to facilitate bacterial spread from a subcutaneous infection site [31]. Pla protease is unregulated by host protease inhibitors. As a result, it induces uncontrollable activation of proteolysis leading to detrimental effects of infection (Table I).

Adhesive properties of Pla contribute to the bacterial colonization in damaged tissue, such as epithelium after a flea-bite. This protein also helps the bacteria invade human endothelial cells [13]. Lesions at the subcutaneous infection sites of Pla-expressing Y. pestis cells have much lower counts of inflammatory cells than the sites infected with Y. pestis cells lacking Pla. Pla cleaves the C3 component of the complement system on the bacterial surface. Since C3 component is one of the key factors in the complement activation, its deactivation leads to a reduced production of chemoattractants and subsequent suppression of the local inflammation [7].

SopA of S. flexneri

SopA (Shigella outer membrane protease) is encoded by S. flexneri virulence plasmid. The main function of SopA (IcsP) protease is to cleave autotransporter IcsA (VirG) between Arg₇₅₈ and Arg₇₅₉ and release the passenger domain (IcsAα) from the bacterial surface [8]. The localization of accessible IcsAα at one pole of the bacterial cell is essential for the formation of actin ‘comet’ tails on the bacterial surface [49]. These tails propel the bacterial pathogen within the host cell as well as enable the bacilli to spread to neighboring cells [50]. Interestingly, in the presence of OmpT, IcsA is degraded and the virulence of S. flexneri is suppressed [44]. IcsA is encoded by the same virulence plasmid as SopA. Moreover the expression of both genes is controlled by VirF. The expression of SopA is additionally controlled by VirB. VirB and VirF proteins regulate the virulence phenotype in S. flexneri[51]. SopA is an important virulence factor (Table I) since it plays an important role in the exclusive polar localization of IcsA, [52].

PgtE of S. enterica

PgtE is a chromosomally-encoded protein of S. enterica. The activity of PgtE is controlled by the SlyA and PhoP/PhoQ mechanisms known to partially control some of Salmonella virulence and resistance to antimicrobial peptides [3], [53], [54].

Like Pla, PgtE can activate plasminogen to plasmin, which degrades fibrin clots and extracellular matrices of eukaryotic tissues. Just like Pla, PgtE can be deactivated by long O-antigen chains of LPS. Like OmpT, PgtE degrades small peptides. PgtE can cleave alpha-helical cationic antimicrobial peptides, such as C18G (Table I). Thus PgtE contributes to bacterial survival inside the host during infection [53]. PgtE expression is upregulated in S. enterica inside vacuoles of murine macrophages and the length of O-antigen chain is reduced, thus increasing PgtE protease activity [18], [55]. Therefore it might contribute to the bacterial survival inside the host cells.

Evolutionary aspects

Recently new members of the omptin family have been identified, as summarized in MEROPS database under family A26 [56]. Plasmids encode genes for most of the identified omptins. Notable exceptions are ompT of E. coli, pgtE of S. enterica, Q8ZGQ6 of Y. pestis, and CAH20506.1 of Y. pseudotuberculosis which are found on chromosomes. The ompT gene is flanked by repeats, suggesting that it was encoded by a lambiod prophage [44]. Based on the proteins’ similarity, an evolutionary tree of the omptin family proteases was constructed (Figure 2). Known omptins fall into two major groups, one group contains PgtE, Pla and PlaA, while the other group contains OmpT, OmpP, and SopA [3]. Each group is characterized by 60 to 84% sequence similarity among its members. Moreover, members in each group are functionally similar. PgtE and Pla are both strong adhesins [16], [31] able to cleave large substrates [29], while the O-antigen of LPS inhibits the proteolytic activities of both Pla and PgtE [16]. PlaA of Erwinia spp. is likely to play an important role in bacteria-plant interaction.

Figure 2.  Phylogenetic tree of the omptin family members from theA26 family from the MEROPS database [56]. Scale bar refers to time in million years. MEGA3 software and UPGMA method were used to contract the tree [68].

Y. pestis evolved from a mild Y. pseudotuberculosis some time between 1,500 to 20,000 years ago [57]. Both species have chromosome-encoded omptin homologs (Q8ZGQ6 and CAH20506.1) with Asn instead of Asp in their catalytic sites. This mutation probably renders them proteolytically inactive. Nevertheless, studies have shown that Q8ZGQ6 of Y. pestis contributes to the establishment of infection in the soil nematode, Caenorhabditis elegans[58]. Y. pestis pla gene most likely evolved from an ancestral S. enterica chromosomal pgtE gene, found in two homologous forms, one in S. enterica serovar Typhimurium and the other one in serovar Typhi [7]. Functional similarities between Pla and PgtE support their common ancestry [2], [16]. There are several homologues of Pla found in plant pathogens Erwinia spp. PlaA and AAN04526.1 have a conserved catalytic site and are likely to be functional proteases and may play a role in establishment of Erwinia infection in plants. Another Erwinia omptin homolog is AAG31039, most likely a nonsecreted nonproteolytic omptin homologue since all four catalytic site residues are mutated and SignalP software found no signal sequence [59]. Another omptin belonging to this group is lpl2311 of Legionella pneumophila, a cause of Legionnaires’ disease. Based on SignalP predictions and sequence analysis, lpl2311 contains a signal peptide, necessary to the secretion and conserved catalytic site residues. Lpl2311 may play a role in bacterial survival inside the phagocytic cells because lpl2311 sequence resembles that of PgtE and L. pneumophila like S. enterica, proliferates inside macrophages. Alternatively lpl2311 can also contribute to L. pneumophila survival inside protozoa.

The second group includes OmpT and SopA, which process autotransporter proteins [8], [42] important in virulence. SopA1 of S. flexneri, SopA2 of enteroinvasive E. coli, and SopA3 of S. dysenteriae are all located on a virulence plasmid that arose in ancestral E. coli that subsequently gave rise to Shigella and enteroinvasive E. coli[60]. Interestingly, enteroinvasive E. coli strains have lost the ompT gene, through the prophage excision, but gained the sopA2 gene, possibly, through the horizontal gene transfer [44]. If ompT gene is introduced in either enteroinvasive E. coli strains or Shigella, these bacteria lose the ability of intercellular spreading [44].

E. coli OmpT is a housekeeping protein that plays a role in terminating SOS response to UV radiation. OmpT cleaves and deactivates UvrB, preventing UvrABC endonuclease from hydrolyzing undamaged DNA [61]. OmpT has been observed to cleave bound colicins [62] and it is thought to contribute to bacterial survival in the presence of colicins. Also, OmpT has an affinity for denatured substrates such as over-expressed proteins in inclusion bodies under denaturing conditions (up to 5 M urea) [63]. This property of OmpT plays a role as a housekeeping protein in denatured protein turn-over at elevated temperatures and processing of proteins in the membrane. This omptin also contributes to the pathogenicity of UPEC strains by cleaving protamine [26] and by processing certain autotransporters.

Other omptins do not fall into either of the two groups described above, such as Vibrio fischeri AW85741 a nonproteolytic omptin homologue and CAG37431 from Desulfatalea psychrophila, a sulfate-reducing bacterium from Arctic permafrost, contains a conserved catalytic site [64]. Other omptin family members are the outer membrane proteases (Q98A65) of Mesorhizobium loti and EAN05038 of Mesorhisobium sp. BNC1 the symbiotic plant bacteria. Computational analysis with SignalP software identified a cleavable signal peptide [59], [65]. Analysis of the region upstream of the structural genes revealed a Shine-Dalgarno sequence implying that these genes are likely to be expressed. The most conserved amino acid in the active site is Ser₉₉, a crucial active site residue for proteolytic activity (Table II). This characteristic is shared by both proteolytic and nonproteolytic omptins, and initially even lead researchers classify OmpT as a serine protease. Ser₄₀ and Tyr₁₅₀ are the other residues that are conserved among all the known omptins. High conservation of serines and omptins’ partial susceptibility to serine inhibitors [20] suggest that omptins might have evolved from serine proteases.

Concluding remarks and future research

There are many bacterial infections without effective treatment or prevention. Death from an infectious disease reached the top five causes of death in the United States in the mid-1990s [66]. One of the several reasons is the increasing population of the elderly, who are particularly susceptible to infections. Another reason is that the achievements in cancer therapy and organ transplantation increased the population of immunosuppressed patients. Moreover, old infectious diseases (e.g., plague) are now classified as re-emerging infections. The omptins provide hope as new potential targets for drug and vaccine development.

Several proteolytic studies with OmpT [24], [25], [28], [67] provide numerous clues about the proteolytic mechanism of the OmpT protease and its inhibition. This could be used as a starting point of computer-based search for OmpT-inhibiting drugs. Narrow cleavage specificity of OmpT could be exploited. Modified OmpT could be a specific protease to release engineered proteins [23], [25]. Besides the E. coli OmpT, there are other omptins whose proteolytic properties need to be explored. One of them is a potent protease Pla. The inhibition of this protein is crucial in preventing a septisemic plague infection. Recently identified omptins provide many other directions of research. It would be interesting to see if PlaA contributes to the pathogenesis of E. pyrifoliae, and if the protease of M. loti plays a role in bacterium-plant symbiosis. Also the ability of the chromosomal Y. pestis proteolytically inactive omptin to contribute to pathogenesis needs to be explored. It is still unknown if the chromosomal omptins of Y. tuberculosis or V. fischeri have any activities. The function of lpl2311 and its possible role in L. pneumophila infection in humans or protozoa remains to be explored. The relation between the protein's activity and the length of O-antigen would be interesting to explore.

Acknowledgements

Authors would like to thank Dr Mohammed Shayib for his help with statistical analysis and Dr William Widger for reviewing the manuscript. Authors are grateful to the anonymous reviewers for their critical comments which greatly improved the quality of the manuscript.