A colorimetric assay adapted to fragment screening revealing aurones and chalcones as new arginase inhibitors

Abstract

Arginase, a difficult-to-target metalloenzyme, is implicated in a wide range of diseases, including cancer, infectious, and cardiovascular diseases. Despite the medical need, existing inhibitors have limited structural diversity, consisting predominantly of amino acids and their derivatives. The search for innovative arginase inhibitors has now extended to screening approaches. Due to the small and narrow active site of arginase, screening must meet the criteria of fragment-based screening. However, the limited binding capacity of fragments requires working at high concentrations, which increases the risk of interference and false positives. In this study, we investigated three colorimetric assays and selected one based on interference for screening under these challenging conditions. The subsequent adaptation and application to the screening a library of metal chelator fragments resulted in the identification of four compounds with moderate activity. The synthesis and evaluation of a series of compounds from one of the hits led to compound 21a with an IC₅₀ value of 91.1 μM close to the reference compound piceatannol. Finally, molecular modelling supports the potential binding of aurones and chalcones to the active site of arginase, suggesting them as new candidates for the development of novel arginase inhibitors.

Three colorimetric assays were investigated, and the selected one was applied to the screening of a library of metal chelator fragments. One of the hits was optimised to give the polyphenolic chalcone 21a with an IC₅₀ value of 91.1 μM.

Introduction

Arginase (EC 3.5.3.1) is a binuclear manganese cluster metalloenzyme that hydrolyses l-arginine into l-ornithine and urea. Two isoforms are encoded in mammals. Hepatic and cytosolic arginase I ensures nitrogen detoxification via the urea cycle. Extrahepatic and mitochondrial arginase II provides l-ornithine for polyamine and l-proline synthesis and maintains nitric oxide (NO) homeostasis by competing with NO synthase, which shares l-arginine as a substrate.¹ Arginase upregulation is implicated in a wide range of pathologies. High level of arginase I, secreted by myeloid-derived suppressor cells (MDSCs) in most of the tumour microenvironment, depletes l-arginine and limits T-cells proliferation and function by altering T-cell receptor synthesis.² Decreased NO levels impair the effectiveness of the immune response and macrophage cytotoxicity. In this aim, several pathogens, including Helicobacter pylori, Staphylococcus aureus and Mycobacterium tuberculosis, produce their own arginase and/or modulate the host arginase to divert l-arginine from NO synthesis.³ In addition, in leishmaniasis and trypanosomiasis, parasite arginase contributes to the synthesis of trypanothione, a powerful antioxidant essential for parasite growth. Finally, the decrease of NO synthesis, reinforced by rigidification and fibrosis, due to the overproduction of polyamines and l-proline, is a common mechanism to cardiovascular and pulmonary damage, observed with endothelial dysfunction.⁴

Arginase inhibitors are almost exclusively amino acids, and the first effective inhibitor was N^ω-hydroxy-nor-arginine (nor-NOHA) which displays micromolar inhibition.⁵ Several clinical trials of nor-NOHA have been reported in diabetes, hypercholesterolemia and cardiovascular diseases.⁶ Unfortunately, the pharmacokinetic parameters of nor-NOHA limit its administration to the injectable route or in situ administration. Nowadays, the development of arginase inhibitors is dominated by boronic acid derivatives of α-amino acids.⁷ The prototype 2(S)-amino-6-boronohexanoic acid (ABH)⁸ and S-(2-boronoethyl)-l-cysteine (BEC) analogue show micromolar inhibition.⁹ Extensive work has been carried out to optimise the affinity of the amino acid moiety for the arginase active site and now α,α-disubstituted and cyclic constrained amino acid derivatives of ABH are the most potent arginase inhibitors with nanomolar activity.⁷ Among these, CB-1158 (numidargistat) is currently in clinical trials for cancer immunotherapy (NCT02903914, NCT03314935, NCT03910530, NCT03837509) and CB-280 is in clinical trials for cystic fibrosis (NCT04279769). With the exception of ABH and CB-1158, boronic based arginase inhibitors suffer from low oral bioavailability and fast clearance.¹⁰ ABH carries an unmodified α-amino acid moiety and CB-1158 carries a dipeptide moiety, suggesting that they both use active transport. Polyphenolic compounds from ref. 11 and inspired from ref. 12 and 13 natural origins are moderate micromolar to millimolar inhibitors. Polyphenols also suffer from fast metabolism,¹⁴ and prolonged release via prodrugs is currently being developed to overcome poor pharmacokinetic parameters.^15–17

Amino acid-based arginase inhibitors have been extensively studied^18–23 and the search for innovative arginase inhibitors has now extended to screening.²⁴ To this end, methods for the direct measurement of l-ornithine based on HPLC-MS,²⁵ MALDI-TOF,²⁶ and, a fluorescent probe²⁷ have been developed and applied to high-throughput screening of 18 000, 175 000 and 93 000 compounds respectively. Of these, Grobben et al.²⁷ describe a high interference rate and no hit. Finally, only the approach of Asano et al.²⁵ who performed an assay with a high concentration of the tested compounds, led to the identification of a new inhibitor. The arginase active site is narrow and small, which makes screening challenging because it involves testing small molecules with low binding capacity. Consequently, screening has to meet the criteria of fragment-based screening²⁸ in which high concentration of the tested compounds is required. Under these conditions, interference is a major concern, especially in colorimetric assays because high concentration compounds may interfere with the colorimetric process. Herein, we investigated interference in three colorimetric assays of arginase using a small set of fragments and compared their interference rates with the colorimetric reaction. One of these assays with a low interference profile was selected and the screening approach was adapted to eliminate false positives at early stage. A metal chelator fragment library was chosen due to the binuclear metal cluster within the arginase active site and four low active hits were identified, including an aurone. To complete the fragment-based drug discovery, a concise medicinal chemistry optimisation process was then initiated starting from the aurone hit. The synthesis and evaluation of a series of aurones and chalcones led to improved inhibition. Finally, the putative binding of the optimised compound to the arginase active site was supported by quantum chemistry calculations.

Results and discussion

Colorimetric assay selection

Most colorimetric arginase assays²⁴ are based on the formation of an adduct with urea produced by arginase (Fig. S1†). Among these, α-isonitrosopropiophenone (INPP)^29–31 reacts with urea by cyclisation at 100 °C, while ortho-phthalaldehyde–primaquine combination (OPA–p)³² forms an adduct with urea at room temperature and the latter is becoming the standard.^33–35 Behind these, an emerging assay called the thioornithine generating assay (TOGA)³⁶ based on thioarginine hydrolysis, produces 2-amino-5-mercaptopentanoic acid which is revealed by the well-known Ellman reagent (DTNB, 5,5′-dithio-(bis-2-nitrobenzoic acid)). A comparative evaluation among these assays was first engaged to assess the effect of a manganese supplementation and the interference with a small set of fragments.

In practice, amino acid-based (ABH, BEC, and nor-NOHA) and polyphenolic arginase inhibitors (verbascoside, piceatannol, and chlorogenic acid), along with a set of 32 compounds from the OTAVA® fragment library (Fig. S2†) were screened at a final concentration of 500 μM in the presence of substrate (l-Arg 14.3 mM, final concentration; thio-arginine 7.71 mM, final concentration) in a pH 7.5 buffer solution (Tris-HCl 28.6 mM, final concentration). Arginase (5.7 × 10⁻⁴ I.U. per μL, final concentration) is replaced in half of the wells by the enzyme product used in the colorimetric reaction (urea 2.14 mM or thio-Orn 1.01 mM). Therefore, concomitantly to the percentage of inhibition of the enzymatic reaction, the percentage of interference between the test compound and the colorimetric reaction was measured. Therefore, to assess influence of the test compound on substrates, reagents, and intermediates of the colorimetric reaction, arginase was substituted with the enzymatic reaction product (urea or thio-ornithine) at a concentration equivalent to that produced by the enzyme (Table 1). Because arginase, which is a binuclear manganese metalloenzyme, remains active in a mononuclear form,³⁷ the experiments were performed twice: once with manganese supplementation (4 mM, final concentration) to ensure the formation of the binuclear manganese cluster^38,39 and to exclude direct sequestration of manganese by the tested compounds, and once without supplementation to maintain a mononuclear form of arginase.⁴⁰

Comparison of colorimetric reactions included in arginase assays.

Substrate, reagents conditions		Thioornithine, DTNB B(OH)3, PBS, room temp., 5 min		Urea, INPP H2SO4, H3PO4, H2O 100 °C, 45 min		Urea, OPA–p B(OH)3, H2SO4, H2O, room temp., 15 min
Mn2+ supplementation		4 mM	0 mM	4 mM	0 mM	4 mM	0 mM
Reference inhibitors: arginase inhibition [inhibition of the colorimetric reaction] in %	Nor-NOHA	100 ± 3 [2 ± 2]	93 ± 2 [0 ± 0]	100 ± 1 [4 ± 1]	100 ± 1 [5 ± 1]	100 ± 3 [0 ± 0]	99 ± 0 [0 ± 1]
	BEC	100 ± 1 [11 ± 5]	97 ± 2 [0 ± 1]	100 ± 1 [0 ± 0]	98 ± 1 [2 ± 1]	96 ± 1 [0 ± 0]	96 ± 1 [0 ± 0]
	ABH	99 ± 1 [13 ± 4]	100 ± 1 [0 ± 0]	100 ± 1 [2 ± 1]	100 ± 1 [0 ± 1]	97 ± 2 [0 ± 1]	100 ± 1 [4 ± 1]
	Verbascoside	66 ± 4 [16 ± 2]	19 ± 3 [0 ± 0]	98 ± 0 [76 ± 2]	60 ± 1 [50 ± 12]	85 ± 1 [22 ± 8]	11 ± 3 [3 ± 1]
	Chlorogenic acid	77 ± 3 [0 ± 3]	17 ± 2 [0 ± 1]	86 ± 1 [81 ± 4]	37 ± 2 [46 ± 9]	48 ± 2 [4 ± 1]	9 ± 4 [0 ± 0]
	Piceatannol	100 ± 1 [33 ± 6]	9 ± 4 [24 ± 8]	98 ± 1 [60 ± 1]	82 ± 0 [89 ± 6]	92 ± 3 [23 ± 3]	22 ± 2 [15 ± 6]
Set of 32 fragments
Mean inhibition in %a		37 ± 21	24 ± 11	14 ± 14	19 ± 17	15 ± 13	15 ± 9
Interference, n (%)		3 (9.4)	3 (9.4)	10 (31)	9 (28)	0 (0)	0 (0)

^aWithout considering false positives.

The conditions for quenching and for the colorimetric reaction follow the literature (Fig. S1†). After a quench with boric acid (23 mM, final concentration), a solution of DTNB (0.7 M, final concentration) in phosphate buffer at pH 7.8 was added and the reading was taken at 412 nm after 5 min at room temperature.³⁶ After quenching with a solution of sulfuric acid (1.12 M, final concentration) and phosphoric acid (3.45 M, final concentration), an ethanolic solution of INPP (15 mM, final concentration) was added and the reading was taken at 550 nm after 60 min at 100 °C.³⁰ An extemporaneous mixture of OPA (2.3 mM, final concentration) in sulfuric acid (0.35 M, final concentration) and primaquine diphosphate (0.9 mM, final concentration) in a sulfuric acid and boric acid solution (0.35 M and 30 mM, final concentrations) was added directly and the reading was taken at 450 nm after 15 min at room temperature.³²

For amino acid-based reference compounds, nor-NOHA, BEC and ABH, manganese supplementation had no effect on inhibition and no interference with the colorimetric reaction was observed even at high concentration (Table 1). In contrast, inhibition by polyphenolic compounds, piceatannol, verbascoside, and chlorogenic acid, was strongly dependent on manganese supplementation and these compounds interfered particularly with the INPP reagent. From the set of fragments (Table 1), the average inhibitions, excluding false-positive compounds, were low and showed high variability. In addition, TOGA seemed to overestimate the percentage of inhibition. Regarding false positives, DTNB, INPP, and OPA–p reagents generated 9%, 31%, and none, respectively (Table 1). In more details, DTNB reacted with benzenedithiol 3 (BDT), 2-aminopyridin-3-ol 1 (I077) and phenazine-2,3-diamine 2 (I294) (Fig. 1). Interference with benzenedithiol 3 (BDT) was expected due to the analogy with mercaptonitrobenzoic acid. Interference with 2-aminopyridin-3-ol 1 and phenazine-2,3-diamine 2 is probably related to their reducing potential, which unbalances the thiol-disulfide exchange.⁴¹ The compounds interfering with INPP share a nucleophilic function adjacent to an electrophilic function (4, 8, 9, 12) or adjacent to another nucleophilic function (2, 3, 5–7, 10, 11). This suggests degradation of urea and/or a reaction intermediate by a hydrolysis process involving the test compound. Since urea is present in excess of the test compound, the process is necessarily catalytic, and the drastic acidic conditions and heating also accelerate the degradation. Furthermore, these results point to catechol as a cause for concern and the need for careful interference control throughout the study.

Fig. 1. False positive compounds. A) from thioornithine generating assay (TOGA), B) from α-isonitrosopropiophenone (INPP) assay.

Full screening

The comprehensive screening of the OTAVA® fragment library for metalloenzymes was then performed twice, with and without manganese supplementation, using the OPA–p colorimetric assay and included a step to check for false positives. In details, the inhibition percentage of each of the 857 compounds was measured once at 500 μM (Fig. 2A). Compounds showing more than 45% inhibition progressed to the second step, which included a triplicate of the initial measurement and simultaneous measurement of the interference percentage as described above. Compounds that confirmed an inhibition greater than 50% without interference were engaged in the IC₅₀ determination and compounds showing a dose effect with an IC₅₀ lower than 1 mM were considered as hits.

Fig. 2. Screening of the OTAVA® fragment library for metalloenzymes. A) Strategy and results; B) percentage of inhibition without manganese supplementation; C) percentage of inhibition with manganese supplementation; D) resynthesis of 14 and 15a: a) tBuOK, MeOH, room temp., 15 h, (92%); b) H₂N–NH₂, H₂O, EtOH, reflux, 30 min, (100%); c) KOH aq., MeOH, room temp., overnight, (55%); E) hits and IC₅₀ (see Fig. S3 for S23–S27†).

Without manganese supplementation, six compounds passed the first screening stage, but only one was confirmed by triplicate measurement and interference checking (Fig. 2B). In contrast, with manganese supplementation, four compounds passed the first screening stage and one was then discarded (Fig. 2C). This difference is consistent with the high level of background noise that may be due to the presence of mononuclear arginase, and as mononuclear arginase is already 50% less active than binuclear arginase, more weak inhibitors appeared. Prior to progressing to hit-to-lead studies, compounds 14 and 15a were resynthesised and retested (Fig. 2D). The activity of both 14 and 15a was validated. The overall hits and corresponding IC₅₀ are shown in Fig. 2E.

Synthesis and evaluation of aurones and chalcones

A hit-to-lead optimisation stage was conducted from 15a by preparing a series of aurones and a subsequent series of chalcones. The 6,7-dihydroxybenzofuran-3(2H)-one 6 (Fig. 3a) is obtained starting from pyrogallol as previously described⁴² and then eight aurones are prepared in moderate to good yields by Claisen–Schmidt condensation⁴³ using potassium hydroxide at room temperature. Polyphenolic chalcone synthesis (Fig. 3b) requires protection of the catechol starting materials with a methoxymethyl protecting group.⁴⁴ Polyphenolic chalcones are also obtained in high global yields by Claisen–Schmidt condensation using potassium hydroxide at room temperature followed by deprotection by treatment with aqueous hydrochloric acid.

Fig. 3. Synthesis of aurones 15a–h and chalcones 21a–j. a) i) chloroacetyl chloride, POCl₃, reflux, 2 h, (51%); ii) AcONa, EtOH, reflux, 5 h, (71%); b) KOH aq., MeOH, room temp., overnight; c) MOMBr, DIPEA, DCM, 0 °C to room temp.; d) HCl aq., MeOH, 0 °C to room temp., 30 min.

These two series, supplemented by related natural products, were evaluated by determination of IC₅₀ against bovine arginase 1 (Table 2). Firstly, the type and position of the substituents of the benzylidene ring of aurones were investigated. Compared to the hit 15a with an IC₅₀ of 507 μM, compound 15b, bearing an additional fluorine in ortho position, displays a slightly improved activity with an IC₅₀ of 312 μM. However, as shown by the pentafluoro derivative 15c with an IC₅₀ of 1091 μM, the addition of more fluorine decreases the activity suggesting that rather specific interactions are involved in the activity. Therefore, we turned our attention to the hydroxyl substituents and the compound 15d with an IC₅₀ of 254 μM bears two catechol moieties and shows a similar activity to that of the difluoro analogue 15b. Again, additional hydroxyl does not improve the activity or reduces it as in the case of the compounds 15e and 15f with IC₅₀ of 287 μM and 532 μM respectively. Interestingly, the comparison of compound 15d with compound 15g with a unique para-hydroxy group and compound 15h with a unique meta-hydroxy group with IC₅₀ of 1925 and 120 μM, respectively, suggests a predominant effect of the meta-hydroxy group in the activity. The natural product sulfuretin which has a lower inhibitory activity than 15d with an IC₅₀ of 437 μM confirms a predominant effect of the catechol moiety.

b-ARG1 inhibition by compounds 15a–h.

	Cpd	R1	R2	R3	R4	R5	IC50 ± SE (μM)
1	15a	—	—	F	—	—	507 ± 10
2	15b	—	F	F	—	—	312 ± 28
3	15c	F	F	F	F	F	1091 ± 87
4	15d	—	OH	OH	—		254 ± 42
5	15e	—	OH	OH	OH	—	287 ± 20
6	15f	OH	OH	OH	—	—	532 ± 83
7	15g	—	—	OH	—	—	1925 ± 111
8	15h	—	OH	—	—	—	120 ± 12
9	Sulfuretin						437 ± 95

In order to investigate a series of compounds without the cycle constraint, a series of polyphenolic chalcones were evaluated (Table 3). The two-catechol bearing chalcone 21a which has an IC₅₀ of 91.1 μM, is better than the constrained aurone analogue 15d. Compounds 21b and 21g each bearing a unique catechol on one of the cycles display similar activity with IC₅₀ of 384 and 351 μM, respectively, which do not show a predominant effect of one of either a catechol. However, comparison of these compounds with those bearing a unique para-hydroxyl substituent on the R ring, such as 21c with an IC₅₀ of 302 and on the R′ ring, such as 21h with an IC₅₀ of 354 μM, and a unique meta-hydroxyl substituent on the R ring, as 21d with an IC₅₀ of 135 μM and on the R′ ring, as 21i with an IC₅₀ of 102 μM, confirms a predominant effect of a meta-hydroxyl substituent in addition to the catechol. Interestingly, on the R′ ring, the 3,5-dihydroxy system from 21j (IC₅₀ of 136 μM) and the 2,4,6-trihydroxy system from eriodictyol chalcone (IC₅₀ of 98 μM) show similar inhibition than the catechol from 21a (IC₅₀ of 91 μM). By contrast, on the R ring, the 3,5-hydroxy system from 21e (IC₅₀ of 364 μM) and the 3,4,5-hydroxy system from 21j (IC₅₀ of 764 μM) reduce the activity.

b-ARG1 inhibition by compounds 21a–j.

	Cpd	R′ ring					R ring			IC50 ± SE (μM)
							R1	R2	R3
10	21a	—	OH	OH	—	—	OH	OH	—	91.1 ± 13.2
11	21b	—	OH	OH	—	—	—	—	—	384 ± 133
12	21c	—	OH	OH	—	—	—	OH	—	302 ± 24
13	21d	—	OH	OH	—	—	OH	—	—	135 ± 25
14	21e	—	OH	OH	—	—	OH	—	OH	364 ± 25
15	21f	—	OH	OH	—	—	OH	OH	OH	764 ± 141
16	21g	—	—	—	—	—	OH	OH	—	351 ± 32
17	21h	—	—	OH	—	—	OH	OH	—	354 ± 42
18	21i	—	OH	—	—	—	OH	OH	—	102 ± 16
19	21j	—	OH	—	OH	—	OH	OH	—	136 ± 15
20	Butein	OH	—	OH	—	—	OH	OH	—	1551 ± 370
21	Eriodictyol chalcone	OH	—	OH	—	OH	OH	OH	—	98.0 ± 22.0
22	Chlorogenic acid									271 ± 31
23	Piceatannol	—	—	—	—	—	—	—	—	71.6 ± 7.0
24	BEC	—	—	—	—	—	—		—	23.0 ± 4.6

Quantum chemistry modelling

Because of the high degree of homology between mammalian arginases, particularly in terms of the active site,⁴⁵ quantum chemistry calculations on our previously described model¹³ were used to identify the most probable position of compound 21a in the enzyme active site. The binding mode of lowest energy (Fig. 4B) displays several interesting features. At the bottom of the active site and as already identified,¹³ the catechol ring binds to the binuclear manganese cluster by a double binding between oxygen atoms of the catechol ring and the manganese ions, and two H-bonds with neighbouring groups. At the entrance of the active site, a second catechol ring displays a pair of two strong H-bonds and the primary amide function of the aspartate 183. Another H-bond links the ketone moiety of compound 21a and the alcohol function of the serine 137. As compound 21a bears two catechol rings, the reverse complex has been also investigated (Fig. 4B). The catechol rings bind the enzyme active site in a very similar way. However, this position prevents the ketone from binding to the serine 137 but forms an aromatic H-bond with the outer catechol ring. This bond is evidently not as energetic as the ketone–serine bond and this complex is 8.5 kcal mol⁻¹ less stable than the other orientation. These results strongly suggest that the serine 137 side chain should be used to increase the interaction between the inhibitor compound and the enzyme.

Fig. 4. Most favourable pose (A) and reverse pose (B) of compound 21a inside the arginase active site. For clarity, enzyme atoms are displayed in white, except the manganese atoms (in pink). H-bonds are shown in green dotted lines.

The rigidity of aurone compounds prevents the formation of the hydrogen bond with the serine 137, which largely accounts for the weak affinity of this series of molecules. Their complexation energies are consistently worse than that of piceatannol, by at least 2 kcal mol⁻¹. In particular, the comparison between the binding structures of chalcone 21a and aurone 15d, that feature the same rings but not the same central part, sheds some light on their large complexation energy difference (14.4 kcal mol⁻¹ stronger for 21a than for 15d). The impossibility of binding the serine 137 side chain is not the only hindrance for the aurone: its lack of flexibility disturbs the free formation of four of the six bonds with the enzyme. One of the bonds with a manganese ion is stretched by more than 0.10 Å, while the lengths of three of the four hydrogen bonds are longer by 0.04 to 0.07 Å than in the case of the corresponding chalcone. These results strongly suggest that arginase inhibitors should remain rather flexible to allow the formation of strong bonds with the enzyme pocket.

Discussion and conclusion

The therapeutic interest of arginase inhibitors has led several teams to screen large chemical libraries. Nevertheless, sophisticated methods of high throughput screening are restricted by the concentration of compounds tested preventing the identification of weak inhibitors that could be further improved. With this mind, we examined three of the main colorimetric assays for arginase to assess their compatibility with high concentrations, especially in terms of interferences. The reference inhibitors show no interference, so the choice between the three assays for the evaluation of submicromolar inhibitors is not limited. However, these colorimetric tests generate significant differences in terms of interference when evaluating fragments. The α-isonitrosopropiophenone (INPP) based assay, inconvenient because of the time required and the heating to 100 °C, generated numerous interferences and therefore appeared unsuitable for screening. The thioornithine generating assay (TOGA) and the assay based on the combination of ortho-phthalaldehyde and primaquine (OPA–p) can be read after a few minutes at room temperature and show low level of interference. TOGA appears to slightly overestimate activity, which may be advantageous for screening purposes, but thioarginine is slightly more expensive and thioornithine, used for interference checking, is not commercially available. However, TOGA may be suitable for cross-confirmation with INPP, which measures urea. As the arginase active site is narrow, hydrophilic and contains a binuclear manganese cluster, a chemical library of metal chelator fragments was selected and screened at high concentrations due to the moderate affinity potential of the fragments. Due to the risk of interference, the OPA–p method was chosen, and a false positive control step was added. With manganese supplementation, three out of four hits passed this step, whereas without manganese supplementation, only one out of six hits passed this step. As screening in the absence of manganese is less discriminating and gives few results, it is proposed to make it optional. Furthermore, as mononuclear arginases have a reduced yield, it has been proposed that this is a form of physiological regulation³⁷ and that the more efficient binuclear arginases are therefore involved in overactivation. The poor success of the screening without manganese suggests that it may be more efficient to stick to a single screening with manganese and increase the number of compounds tested. On the other hand, the false positives obtained in the full screening confirm the importance of a control step to early discriminate false positive compounds that interfere with the colorimetric reaction when screening highly reactive or high concentration compounds. Four hits were validated, one of which was then further optimised to give compound 21a with an IC₅₀ of 91.1 μM and binding to the arginase active site was supported by molecular modelling. Although halogenated chalcones have recently been identified as leishmaniasis arginase inhibitors,⁴⁷ aurones and polyphenolic chalcones have been identified for the first time as arginase inhibitors. Very few polyphenols are active against mammalian arginase and compound 21a therefore complements the polyphenolic arginase inhibitors. Under the same evaluation conditions,⁴⁶ quercetin heterosides showed inhibitory activity between 100 μM and 300 μM and chlorogenic acid displayed an IC₅₀ of 271 μM. Piceatannol, with an IC₅₀ of 71.6 μM, is slightly more active than compound 21a and remains the reference compound. However, polyphenolic aurones and chalcones represent a new molecular scaffold with pharmacological potential and a promising avenue for the development of new arginase inhibitors. Additionally, the predominance of OPA–p assay for the fragment-based screening and optimisation sequence in the context of arginase inhibitor screening was confirmed.

Author contributions

JM and RA performed screening; JM, LM, RM, MV performed synthesis and evaluation; AZ performed arginase assay; DH performed analysis; CR, SF, BC performed molecular modelling; FS, BC, CG, MP wrote original draft; all authors reviewed the manuscript; CG and MP designed, supervised, and administered the project.

Conflicts of interest

There are no conflicts to declare.