Discovery of a new class of triazole based inhibitors of acetyl transferase KAT2A

Abstract

We have recently developed a new synthetic methodology that provided both N-aryl-5-hydroxytriazoles and N-pyridine-4-alkyl triazoles. A selection of these products was carried through virtual screening towards targets that are contemporary and validated for drug discovery and development. This study determined a number of potential structure target dyads of which N-pyridinium-4-carboxylic-5-alkyl triazole displayed the highest score specificity towards KAT2A. Binding affinity tests of abovementioned triazole and related analogs towards KAT2A confirmed the predictions of the in-silico assay. Finally, we have run in vitro inhibition assays of selected triazoles towards KAT2A; the ensemble of binding and inhibition assays delivered pyridyl-triazoles carboxylates as the prototype of a new class of inhibitors of KAT2A.

Keywords:

• N-pyridine triazoles
• KAT2A inhibitors
• virtual screening
• acetyl transferases
• anti-cancer

1. Introduction

1,2,3-triazoles are commonly recognised as important chemical scaffolds as well as efficient molecules in the area of medicinal chemistry^1,2. Given their versatile behaviour in acting as both Lewis acids and bases^3,4, triazoles have been used as a core structural motif in a huge variety of drug classes such as: antimicrobial^5,6, anti-inflammatory⁷, antineoplastic⁸, antiviral⁹ ,antihypertensive¹⁰ or as immunomodulatory agents¹¹.

We have recently reported a novel synthetic pathway that, by reacting β-ketoesters 1 and azides 2, provided 1,2,3-triazoles 3 or 4 (Scheme 1). The reactions employing 2-unsubstituted β-ketoesters were found to provide 5-methyl-1,2,3-triazoles 4; whereas 2-alkyl-substituted β-ketoesters provided 5-hydroxy 1,2,3-triazoles 3 in high yields and as a single regioisomer¹².

Scheme 1. Preparation of triazoles 3 and 4.

As a follow up of this work, we have posed the question of whether or not those classes of new compounds may be useful in medicinal chemistry. Triazoles were repeatedly reported as bioactive compounds^1–11 and many drug candidates containing the pyridine ring were equally described. Pyridines are commonly used in medicinal chemistry because of their ability to establish hydrogen bonds either as donors or acceptors, their water solubility, small dimensions and, most importantly, their potential to act as amide bioisosteres¹³. The latter in particular, makes pyridine pivotal in drug discovery¹⁴. Additionally, if compared to the benzene ring^15–17, the pyridine unit displays a relevant increased basicity¹⁸, an improved aqueous solubility¹⁹ and a smaller polar surface; ²⁰; all of these features consent an optimal orientation of the pyridine-containing drugs with the biological target through π-stacking interactions^21,22. According to the FDA²³, there are more than 95 approved drugs containing the pyridine moiety that are, currently, employed against tuberculosis (i.e., isoniazid^24,25 5 and ethionamide²⁶ 6), HIV/AIDS (i.e., delavirdine²⁷ 7), Alzheimer disease (i.e., tacrine²⁸ 8), Raynaud’s syndrome (i.e., nifedipine²⁹ 9), hypertension (e.g., nivaldipine³⁰ 10) and so on among the others (Figure 1).

Figure 1. Few examples of approved drugs containing a pyridine unit.

In order to answer the research question on new triazoles’ bioactivity, we submitted a small selection of pyridine-based compounds 11–18 (Figure 2) into a virtual screening study, which identified in N-(pyridin-2-yl)-4-carboxylic-5-methyl triazole 16 (Figure 2) a new chemical template with significant activity and selectivity as KAT2A inhibitor. KAT2A, also known as GCN5 (general control nonderepressible 5), is part of the HAT (histone acetyltransferase) family and, more specifically, belongs to the GCN5-related N-acetyltransferase group^31,32. KAT2A promotes the acetylation of the ε-amino group of lysines thus preventing the positive charge on the amino group to impact other proteins interaction. Furthermore, the neutralisation of lysine positive charge leads to a weaker bond with DNA strands allowing transcriptional factors to penetrate DNA itself more easily³³. Dysregulation of KATs’ activity is associated with numerous severe tumours such as small-cell lung cancer³⁴ or colon cancer³⁵, inflammatory disorders³⁶, type 2 diabetes³⁷ and so on. So far, only few HATs inhibitors are available and have been approved for therapeutical use (Figure 3). They mainly derive from natural sources such as anacardic acid³⁸ 19, garcinol^39,40 20 or curcumin⁴¹ 21 even though synthetic products, bearing thiazole⁴² 22, isothiazole^43,44 23 and pyrido/benzothiazolone⁴⁵ 24 units, have been recently introduced and proved to have a higher activity and specificity compared to the natural products.

Figure 2. Pyridine-based triazoles tested through docking screening against the BioGPS cavity database.

Figure 3. Natural and synthetic HATs inhibitors.

Unfortunately, most of the HATs inhibitors known showed lack of selectivity towards members of the HAT family and, therefore, selectivity for a specific enzyme, for example, KAT2A, is still an outstanding issue.

Herein we describe each stage of our research that led to the identification of compound 16 (Figure 2) as a new KAT2A inhibitor template. This study includes virtual screening, synthesis, binding studies and in vitro testing.

2. Results and discussion

2.1. Virtual screening

Compounds 11–18 were submitted to virtual screening towards a database of proteins with established structure and role in system biology using the BioGPS software in combination with the FLAP algorithm. The BioGPS workflow used for the docking screening of triazoles 11–18 consisted in 5 parts: i) Protein refinement, achieved by using an algorithm known as “Fixpdb” that enables the preparation of the protein structure obtained from the Protein Data Bank (PDB)⁴⁶; ii) Cavity detection, achieved by using an algorithm known as “Flapsite” that allows for the identification of pocket points located within a distance of 4 Å maximum from the closest protein⁴⁶; iii) and iv) Cavity characterisation and comparison, achieved by the FLAP algorithm (Fingerprints for Ligands And Proteins) that enables the identification of the potential complementary ligand pharmacophoric features for a protein binding site^46,47; v) Data analysis, in which each pocket/target similarity is analysed by using the global scores that attribute to 0 no similarity and to 1 maximum similarity⁴⁶. This study provided docking results and score distribution of 11–18 versus the protein panel (Table 1)⁴⁶. Importantly each of the proteins listed in Table 1 is a significant target for medicinal chemistry as for example KAT2A^31–37.

Table 1. Docking scores for triazoles 11–18 found for this dataset of twenty proteins.

The results collated in Table 1 report the following: (1) numbers are arbitrary scores comprised between 1 and 100 and reflect the fitting of molecules 11–18 to the enzyme pocket; (2) red cells indicate a bad ligand-protein fitting; (3) green cells indicate medium ligand-protein fitting meanwhile (4) blue cells indicate a good ligand protein fitting. The results obtained showed the following: (i) triazoles 11–14, bearing a hydroxyl group at position C5, did not show significant binding scores (Table 1, except for Rab7a⁴⁸ and Serine/Threonine- protein phosphatase 5 (PP5)⁴⁹; (ii) long chain acid 15 showed a good fitting with sixteen proteins (Table 1) hence lacked selectivity from the premises and for this reason was discarded. In conclusion, this dataset showed that the presence of a 5-hydroxy, or its ketol form, on the triazole generated a class of compounds that were not useful for further medicinal chemistry investigations with the 20 selected proteins. Equally, compound 17 (Table 1), showed a promiscuous behaviour with up to thirteen proteins and, therefore, was abandoned.

At this point we focussed our attention on compound 16, which docking with KAT2A was reported to be peculiarly higher than any of the other binders 11–15 and 17,18, reaching 80.13 (Table 1)¹. Considering that many of the known ligands for KAT2A comprise a carboxylate functionality (Figure 3), we hypothesised that the binding data for compound 16 were dependent on the presence of a carboxylic function at the C4 position which favoured the interaction of triazole 16 with the desired target (Figure 4), whereas triazole 18, bearing an ester function at C4, showed a lower binding score with KAT2A (Table 1). In support of our hypothesis, polar contacts connecting 16 and KAT2A occurred mainly when 16’s OH function of the carboxylic moiety and 16’s N2-N3 atoms of the triazole ring interacted with the NH functions of KAT2A Arg-558 residue (shown with blue and light blue sticks, Figure 4).

Figure 4. PyMOL outlook of triazole 16 bound to KAT2A active site pocket. Triazole’s 16 nitrogens are coloured in blue, carbons in green and oxygens in red.

In particular, 16’s N2 and N3 atoms of the triazole ring coordinate arginine NH number 20289 (2.2 Å and 2.1 Å distance respectively, Figure 4) while the N3 atom of the triazole ring and the OH function of the carboxylic moiety coordinate arginine NH number 20292 (2.3 Å and 2.5 Å distance respectively, Figure 4).

The low binding-affinity of triazole 18 with all the 20 proteins and the interactions of 16 with KAT2A identified in this work (Figure 4), clearly indicated that this selection of proteins required triazole binders to possess hydrogen-bond donor and acceptors units at C4 and C5, which may have been a discriminator in the selection of targets from the BioGPS database⁴⁶. In summary, the analysis of the docking studies identified 16 as a potential scaffold to be evaluated as KAT2A inhibitor. For this reason, we proceeded with the synthesis of a small library of triazole 16 analogs 26a-e and 27a-d.

2.2. Synthesis of pyridyl-triazoles 26a-f and 27a-d

A small library of 1,2,3-triazoles 26a-e and 27a-d bearing either a carboxylate or an ester functionality at C4 was therefore prepared (Table 2). We anticipated that the carboxylic functionality was indeed crucial to high binding, however, we also noted that compound 18, i.e., the ethyl ester of compound 16, showed a significant binding for KAT2A. Hence, both carboxylates and their esters were included in the library.

Table 2. Synthesis of pyridyl-based triazoles 26a-e^a.

Entry	Azide	R	R’	Alkyne	R’’	Product	Yield (%)^b
1	2a	H	H	25a	COOEt	26a	90
2	2a	H	H	25b	(CH2)4COOH	26b	89
3	2b	OCH3	H	25a	COOEt	26c	78
4	2c	H	CH3	25a	COOEt	26d	94
5	2d	H	Cl	25a	COOEt	26e	99

^aReaction conditions: 2a-d (1 equiv.), 25a,b (1.1 equiv.), Cu(OTf)₂*C₆H₅CH₃ (0.1 equiv.), toluene (0.25 M). ^bIsolated yields.

The synthesis of compounds 26a-e has been carried out following a procedure previously reported¹². Pyridine azides 2a-e were reacted with alkynes 25a,b to give triazoles 26a-e via a CuAAc cycloaddition protocol (Table 2)⁵⁰. Pyridine-based triazoles 26a-d were obtained in high yields (Table 2). Considering that carboxylates were found better ligands by virtual screening, as highlighted in the docking results (Table 1), we then proceeded with the hydrolysis of esters 26a,b and 26d,e to reveal the corresponding acids 27a-d (Table 3). This entailed standard saponification using a solution of potassium hydroxide, which provided 27a-d in good to high yields.

Table 3. Hydrolysis of triazoles 26a,b-d,e to reveal corresponding carboxylates 27a-d^a.

Entry	Triazole	R	R’	Product	Yield (%)^b
1	26a	H	H	27a	87
2	26b	OCH3	H	27b	70
3	26d	H	CH3	27c	71
4	26e	H	Cl	27d	64

^aReaction conditions: 26a,b-d,e (1 equiv.), KOH (1 equiv.), H₂O (1 M).

^bIsolated yields.

2.3. Preliminary binding studies through fluorescence analysis

The binding properties, alongside the ability of 16 and 27a-d to inhibit the activity of KAT2A were evaluated using a standard protocol based upon the measurement of fluorescence⁵¹. KAT2A fluorogenic assay was developed by others to screen for inhibitors of this enzyme⁵². This test is based on the transfer of an acetyl group from acetyl-CoA (acetyl coenzyme A) to a peptide substrate. After incubation with acetyl-CoA and the inhibitor, the KAT2A enzyme generates acetylated histone H3 peptide and CoASH. The thiol group of CoASH can be detected with fluorescein isothiocyanate isomer I (5-FITC), which is a reliable reaction considering the reactivity towards nucleophiles including amines and sulfhydryl groups present on proteins⁵¹. The fluorescein isothiocyanate isomer I (5-FITC) absorption is at wavelength = 495 nm and emission = 525 nm. In the experiment, which relies on the measurement of emission at 525 nm, strong emissions were observed when a poor inhibition of KAT2A was realised; on the contrary, poor emissions at 525 nm indicated high levels of inhibition (Figure 5). Each assay was run in triplicate and at increasing concentrations of 16, 26c and 27a-d (1.5 µM, 5 µM, 10 µM and 15 µM). As shown in Figure 5, compounds 27a and 27 b displayed a good inhibitory activity at as low as 5 µM (coloured in blue) while higher concentrations were required for triazoles 16 and 26c (15 µM, coloured in black, and 10 µM, coloured in red, respectively). Compound 27c also displayed a good inhibition of KAT2A acetylating activity at just 5 µΜ concentration (coloured in blue). A significant divergence was observed for the behaviour of 27d, which holds a p-Cl substitution on the pyridine ring. The results collected indicated that 27d seemed to promote rather than inhibit the enzymatic activity of KAT2A, as highlighted by the increasing level of CoASH produced vs concentration (Figure 5). This result was most unexpected in lieu of the striking similarity between 27d and 27a-c and it should be looked in more details in future work.

Figure 5. KAT2A fluorescence tests performed on triazoles 16, 26c and 27a-d.

In summary, we have demonstrated that triazoles 16, 26c and 27a-c possess inhibitory activity of KAT2A whereas only 27d proved to behave as an activator. Compounds 16, 26c and 27a-d were then subjected to in vitro assays in a cell model displaying dysfunctional activity of KAT2A to confirm the bioactivity.

2.4. In vitro testing of triazoles 16, 26c and 27a-d

Compounds 16, 26c and 27a-d were tested in U937 cell line (from human myeloid leukaemia, AML) in which KAT2A is known to be overexpressed⁵³. U937 cells were stimulated with different concentrations of compounds 16, 26c and 27a-d, starting from 200 µM. Cell viability was measured by thiazolyl blue tetrazolium bromide (MTT) assay. This assay displayed a slight reduction (10–20%) in cell viability by all the compounds tested (Figure 6).

Figure 6. MTT assay performed on triazoles 16, 26c and 27a-d. NT = non treated.

To investigate compounds 16, 26c and 27a-d induced inhibition of KAT2A acetylation levels of histone H3K9/14ac were analysed. To this end, U937 cells were treated with a 200 µM concentration of compounds 16, 26c and 27a-d for 24 h. SAHA (suberoylanilide hydroxamic acid), a known histone deacetylase inhibitor, was used as a positive control of acetylation at 5 µM concentration. Histone extraction and subsequent Western Blot (WB) analysis were carried out checking H3K9/14ac acetylation levels. Triazole 26c showed 40% reduction in the acetylation of H3K9/14ac (Figure 7) which was the strongest inhibition value obtained.

Figure 7. WB analysis (on the left) of 16, 26c and 27a-d showing H3K9/14Ac levels in U937 cells following 24 h treatment at the concentration of 200 µM; 5 µM SAHA treatment was used as a positive control of acetylation. Densitometric analysis of WB is shown on the right.

Considering the results obtained, it is possible that a lateral and flexible chain at C4 on the triazole ring is required to allow a tight interaction of triazoles and KAT2A. This in addition to the presence of a hydrogen-bond acceptor site which is represented by the carboxylate function.

3. Conclusion

In conclusion, we identified a new class of binders/inhibitors of KAT2A which comprises a pyridine and a long chain carboxylate linked via a triazole ring. The study used virtual screening to select a small library of pyridine-based 1,2,3-triazoles, including 26a-e and 27a-d. We then submitted triazoles 26a-e and 27a-d to fluorescence binding assays versus KAT2A enzyme which confirmed the binding abilities of these entities to KAT2A. Fluorescence binding assays revealed that only triazoles 27a-d and 26c interacted with KAT2A, meanwhile their correspondent ester analogs 26a,b,e,f did not show any binding. Finally, we evaluated the in vitro activity of 26c and 27a-d in U937 cell line of human AML and found out 26c to be the most active compound showing a 40% inhibition of KAT2A acetylating activity. It is noteworthy that compound 26c, having a longer chain, displayed the best inhibitory activity in vitro; meanwhile, shorter chain carboxylate, alike 16 or 27, that were demonstrated optimal binders, showed reduced activity compared to 26c. Studies are currently ongoing to (1) determine the structure-activity relationship (SAR) of 26c analogs and (2) improve potency and selectivity of this new template (26c) for KAT2A inhibition at a lower dose concentration.

Acknowledgements

The authors acknowledge IRC GOIPG/2018/3165 for support to RP and VALERE: “Vanvitelli per la Ricerca Program: EPInhibitDRUGre (CUP B66J20000680005) and Programma Operativo Nazionale (PON) Ricerca e Innovazione 2014–2020 AIM attrazione e mobilità dei ricercatori” for supporting NDG and GB research about KAT2A biological assays.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Irish Research Council (GOIPG/2018/3165), Programma Operativo Nazionale Ricerca e Competitività (Ricerca e Innovazione 2014-2020 AIM), and VALERE (EPInhibitDRUGre (CUP B66J20000680005).

Notes

1 Although the dataset predicted 16 as a good binder also for other proteins, the value of these interactions was inferior to a point that selectivity could be, at least in principle, possible.