From Dynamic Combinatorial Chemistry to in Vivo Evaluation of Reversible and Irreversible Myeloperoxidase Inhibitors

Abstract

The implementation of dynamic combinatorial libraries allowed the determination of highly active reversible and irreversible inhibitors of myeloperoxidase (MPO) at the nanomolar level. Docking experiments highlighted the interaction between the most active ligands and MPO, and further kinetic studies defined the mode of inhibition of these compounds. Finally, in vivo evaluation showed that one dose of irreversible inhibitors is able to suppress the activity of MPO after inducing inflammation.

Neutrophils represent the first line of the human innate immune defense system by phagocytosing and killing invading pathogens.¹ Optimal antimicrobial action in neutrophils relies on the action of hypochlorous acid (HOCl), the product of the myeloperoxidase (MPO, EC 1.11.2.2)–hydrogen peroxide–chloride system.²

In certain inflammatory events, MPO and/or HOCl are released from neutrophils causing oxidative damage of host tissue and modification of biomolecules.²⁻⁴ Consequently, MPO has become a new target for designing anti-inflammatory drugs.⁵⁻⁷

From a general point of view, the development of pharmacophores typically proceeds according to a conventional pathway, namely, the structural design and synthesis of analogues from a “hit” molecule followed by the evaluation of structure–activity relationships.^6,8 However, this classical method is particularly costly and time-consuming. Another innovative strategy consists in the generation and screening of a dynamic combinatorial library (DCL).⁹ In the realm of dynamic combinatorial chemistry (DCC), DCL constitutes a rational alternative in drug discovery, opening thus new horizons for medicinal chemists. Indeed, the in situ reaction of simple building blocks is able to give rise to a wide range of new molecules through reversible covalent bond formation. In the last 10 years, this strategy allowed for the creation and the identification of ligands that specifically recognize targets such as proteins and nucleic acids.¹⁰

With this in mind, we decided to apply this approach in order to develop new irreversible inhibitors of MPO. Recently, we evaluated a new family of scaffolds, i.e., hydralazine¹¹ and isoniazid, endowed with the ability to inhibit MPO irreversibly but with high IC₅₀ values (0.9 and 5 μM, respectively) (Figure 1). Keen to improve these substrates, we decided to take advantage of the high reactivity of hydrazine and hydrazide functionalities toward aldehyde partners in order to prepare and evaluate a library of ligands by a dynamic combinatorial strategy.

Figure 1.

Structures of aromatic aldehydes 1A–24A, aliphatic aldehydes 1B–14B, and hydrazine/hydrazide derivatives 1C–6C.

A set of aldehydes and hydrazine derivatives was selected to compose the building blocks as follows: group A contained aromatic aldehydes 1A–24A, group B comprised aliphatic aldehydes 1B–15B, and group C consisted of hydralazine, isoniazid, and some other hydrazines 1C–6C (Figure 1). The selected aldehydes have a molecular weight (Mw) lower than 160 g/mol in order to achieve ligands with Mw < 320 g/mol since the active site of MPO is located at the end of a narrow tunnel.¹² At first, the inhibitory ability of groups A and B was assessed against MPO, but none of the aldehydes had an activity at a 1 μM concentration. In contrast, hydrazines of group C were capable of inhibiting 61% of MPO activity at 1 μM. Next, more efficient ligands were designed according to a dynamic combinatorial approach. In substance, MPO was incubated with two mixtures A–C and B–C composed of 1 μM of each building block A/C and B/C, respectively. From this, the complete suppression of activity of MPO (>96%) using both libraries A/C and B/C (Figure 2, step 1) was observed.

Figure 2.

Determination of the most active inhibitors of MPO by dynamic combinatorial chemistry using aromatic aldehydes (group A), aliphatic aldehydes (group B), and hydrazine derivatives (group C).¹³

The results clearly indicated that new scaffolds can be formed and that the resulting inhibitors have a good affinity toward MPO, even better than the hydrazines of group C (Figure 2). A step further, a new experiment was set up in order to determine the best aldehyde/hydrazine partners that cause the highest inhibitory effect. First, in a 96-well plate, each aldehyde A and B (1 μM each) was challenged with all hydrazines of group C through DCC in the presence of MPO. The resulting DCLs highlighted an increased inhibitory activity in most cases, but ligands obtained from vanilline 1A, 3-hydroxy-4-methoxybenzaldehyde 6A, 4-dimethylaminobenzaldehyde 13A, and glycolaldehyde 13B provoked a high inhibition of the enzyme (>82%). Therefore, potent inhibitors of MPO were formed from these building blocks. Subsequently, the remaining experiments have focused on the determination of the best aldehyde/hydrazine couple by the reaction of each hydrazine of group C (1 μM) with each aldehyde 1A, 6A, 13A, and 13B. It could be demonstrated that hydralazine 1C, 4-fluorophenylhydrazine 2C and isoniazid 3C (Figure 2) gave rise to scaffolds with a high inhibitory effect toward MPO (>82%), but the hydrazone derivative 13A–1C was able to suppress the activity of MPO at 100% (Figure 2, step 2).

In order to prevent any bias in the previous DCL results, the correlation between the increased inhibitory activity and the hydrazone content was investigated by ¹H NMR. Hence, equimolar mixtures of complementary randomly chosen active (13A–1C) and inactive (10A–1C and 17A–4C) building blocks were incubated in the presence of MPO. After 15 min, the disappearance of the aldehyde peak (CHO) and the increase of hydrazone proton (CH=N−) clearly suggested the formation of hydrazones in variable amounts (5–13%), this being observed both for active and inactive couples. More importantly, the activity appeared independent of the hydrazone quantity. Note that in all instances, the low conversion rate in hydrazone provides a sufficient amount of compound for a complete suppression of activity. As a consequence, this experiment confirmed the formation of active and inactive pairs, consistent with those obtained by DCC.

Next, the active hydrazones were synthesized in order to determine their level of activity. Under a classical condition, the reaction between equimolar amounts of hydrazine and aldehyde derivatives proceeded in refluxing ethanol for 4 h. After treatment, the resulting hydrazones were recovered in excellent yields (>90%) as solids. Subjecting MPO to these ligands allowed the confirmation of the previous results, i.e., the best inhibitors were derived from hydralazine 1C and the lowest IC₅₀ value (79 nM) was obtained with compound 13A–1C (Table 1). Although couples 13A–1C and 13B–1C are endowed with good inhibitory effects toward the enzyme, the association of aldehydes 13A and 13B with hydrazines 2C and 3C gave rise to moderate and weak inhibitors (IC₅₀ value > 1.60 μM).

Table 1. Values of IC₅₀, Free Energies of Binding ΔG Predicted from Docking Experiments and Residual Activity of MPO after Diluting 100 Times the Active Hydrazone Compounds^a.

code	IC50 (μM)	ΔG (kcal/mol)	residual activity of MPO without H2O2 (%)	residual activity of MPO with H2O2 (%)
1C	0.90 ± 0.2	–9.3	92	19
1A-1C	0.34 ± 0.07	–17.1	87	22
6A-1C	0.15 ± 0.04	–18.4	93	25
13A-1C	0.08 ± 0.03	–23.2	93	92
13B-1C	0.11 ± 0.06	–19.0	82	18
2C	>5	–7.6	103	98
1A-2C	1.2 ± 0.1	–15.6	89	98
6A-2C	1.13 ± 0.05	–17.5	81	92
13A-2C	3.1 ± 1.0	–12.9	87	84
13B-2C	>5	–8.9	107	97
3C	4.7 ± 1.1	–9.4	98	64
1A-3C	0.46 ± 0.12	–15.4	96	91
6A-3C	0.43 ± 0.16	–15.6	88	87
13A-3C	1.62 ± 0.4	–11.2	91	93
13B-3C	>5	–12.4	100	94

^aIC₅₀ values are given as mean ± SD, n = 3.

These encouraging results have convinced us to implement a comprehensive study of the inhibitory activity by molecular docking experiments. A comparison of binding prediction for active hydrazones 1A–1C, 6A–1C, 13A–1C, and 13B–1C and starting hydrazine 1C highlighted additional interactions assigned to the structural features of the aldehydes (Table 1). Hence, methoxy and hydroxy functions of 1A and 6A, respectively, made hydrogen bonds with Glu102, which plays a pivotal role in the interaction with the inhibitor (see SI). Moreover, 13A–1C is doubly bonded to Glu102 through phtalazine and NH groups of 1C.

Compounds 1A–1C, 6A–1C, and 13B–1C were predicted to stack on the active site of MPO through the aromatic ring of hydralazine, as seen on Figure 3. In contrast, the docking pose with ligand 13A–1C emphasized an interaction involving the aromatic group of 13A (Figure 3A).

Figure 3.

Comparison of the best-scored docking poses of hydrazones 13A–1C (A), 1A–1C (B), 6A–1C (C), and 13B–1C (D) derived from hydralazine 1C (left). Structures of the same ligands are given on the right side.

For the sake of comparison, additional docking experiments were carried out with ligands based on pharmacophores 2C and 3C. Interestingly, all ligands derived from isoniazid 3C make a π–π stacking with the heme of the enzyme through the pyridyl group of 3C. In opposite, hydrazones formed from 2C showed stacking poses with the π-system of the aldehydes, except for 13A–2C since no interaction was observed (see SI). According to the predicted models, molecules bearing both aromatic and polar functionalities have a greater binding affinity toward MPO and therefore a higher inhibition ability.

Finally, we elucidated the mechanism of action of these hydrazones and their capacity to act as irreversible inhibitors. These experiments were performed in the presence and absence of hydrogen peroxide, which is necessary to initiate the catalytic cycle of MPO. In practice, mixtures of ligands and enzymes were diluted 100-fold, followed by measurement of the residual enzymatic activity. In general, in the absence of H₂O₂, inhibitors provoked inactivation up to 19%, whereas hydrogen peroxide alone minimally affected the activity (Table 1). In the presence of hydrogen peroxide 13A–1C and 2C- and 3C-based ligands, the activity decreased by 10–15%. By contrast, 1A–1C, 6A–1C, and 13B–1C were able to cause an almost complete inhibition of MPO in the presence of hydrogen peroxide (residual activity in the range 18–25%, Table 1). Interestingly, these levels of inhibition correlated well with the predicted models. Indeed, the docking poses of hydralazine- and isoniazid-based hydrazones underline the important role played by the nitrogen heterocycles of 1C and 3C, with hydralazine being a better inhibitor than isoniazid. Hence, in most of the molecules, the aromatic groups governed interactions with the heme, but additional contacts due to aldehyde moieties seemed to lock the system. This aspect is probably responsible for the higher affinity and inhibition rate induced by ligands derived from hydralazine. Moreover, the variation of inhibitory effect between 1A–1C, 6A–1C, 13A–1C, and 13B–1C might be reflected by the distance between hydralazine and heme groups (Figure 3). Hydrazone 13A–1C acted as a reversible inhibitor while 1A–1C, 6A–1C, and 13B–1C were irreversible.

The mechanism of MPO inhibition was subsequently investigated by the multimixing stopped-flow technique. Native ferric MPO [Fe(III)···Por] is oxidized (k₁) by H₂O₂, producing water and Compound I {oxoiron(IV) combined with a porphyrin cation radical: [^+•PorFe(IV)=O]} (Figure 4). In the halogenation cycle, Compound I is directly reduced back to the resting state (k₂) by chloride [or other (pseudo)halides], thereby releasing hypochlorous acid. Alternatively, in the presence of one-electron donors, the peroxidase pathway is followed, including Compound I reduction to Compound II [PorFe(IV)–OH] (k₃) and Compound II reduction to the ferric state (k₄) (Figure 4).¹⁴ Here, Compound I reduction was evaluated with the following ligands: 1A–1C, 13A–1C, and 13B–1C. In all cases, a direct and fast transition of Compound I to Compound II (Soret maximum at 456 nm) was observed with clear isosbestic points (see Figure 4, middle, SI). The determined values of k₃ were 2.8 × 10⁵, 7.1 × 10⁵, and 3.3 × 10⁵ M^–1 s^–1 for 1A–1C, 13A–1C, and 13B–1C, respectively. The results clearly indicated that all selected molecules behaved as good one-electron donors of Compound I reacting similar to hydralazine alone (k₃ = 7.1 × 10⁵ M^–1 s^–1). In contrast, the reaction of the tested molecules with Compound II gave variable results. Hydrazone 13A–1C showed a direct transition of Compound II to ferric MPO (Soret maximum at 429 nm) with clear isosbestic points, but this reaction was slow. The apparent bimolecular rate constant (k₄) was found to be 89.3 M^–1 s^–1, indicating a high k₃/k₄ index of 7951. This high index suggested that inhibitor 13A–1C induced the (reversible) accumulation of Compound II, which is outside the halogenation cycle (Figure 4). By contrast, during the reactions of hydralazine and substrates 1A–1C and 13B–1C with Compound II, a steady-state shift to Compound III was observed (Figure 4 bottom). Compound III can only be formed from ferric or ferrous MPO with activated oxygen (k₇) or dioxygen (k₆), respectively (Figure 4). It seems that the substrate radicals (AH^•, Figure 4), which are generated during Compound I and II reduction, react with ferric MPO and reduce it to the ferrous state (k₅). Alternatively, substrate radicals could activate dioxygen and the resulting superoxide reacts with ferric MPO to generate Compound III. Interestingly, only partial recovery of ferric MPO after complete consumption of hydrogen peroxide was observed, suggesting that these molecules might act as suicide inhibitors (see SI). This kinetic study supports previous findings that demonstrated that ligands 1A–1C and 13B–1C inhibit MPO irreversibly, while substrate 13A–1C behaves as a reversible inhibitor.

Figure 4.

Reactions catalyzed by human myeloperoxidase. In the halogenation cycle Compound I is reduced by halides (X^–) directly to the ferric state [thereby releasing hypohalous acids (HOX)], whereas in the peroxidase cycle Compound I is reduced in two one-electron steps via Compound II to the resting state (A). Reaction of MPO Compound I with 20 μM 13B–1C (B). Reaction of MPO Compound I with 500 μM 13B–1C (C). Red spectra correspond to Compound II formation; light blue spectra correspond to Compound III formation, and light green spectra show the decay to incomplete ferric MPO. Insets show the time traces at 430 and 456 nm.

Finally, the inhibition of MPO was tested in vivo using hydrazones 6A–1C, 13A–1C, and 13B–1C. After inducing inflammation in Wistar Han male rats by intraperitoneal injection of carrageenan, 10 mg/kg of the selected ligands were injected intravenously.

Twenty-four hours after drug administration, a slight decrease in enzyme release with molecule 6A–1C was measured. As concerns 13A–1C and 13B–1C, the concentration of MPO was similar to that found in rats treated only with carrageenan (Figure 5).¹⁵ A dramatic decrease of MPO activity was observed with compounds 6A–1C, 13A–1C, and 13B–1C. When 13A–1C was used, the enzyme activity dropped to the same level as measured for reference rats (untreated with carrageenan). After 48 h, MPO was collected in the peritoneal liquid and its activity was determined for all groups of rats. When 13A–1C was administered, the enzyme recovered its activity, whereas the enzyme activity remained inhibited with 6A–1C and 13B–1C to a large extent (Figure 5).

Figure 5.

Determination of MPO concentration collected in the peritoneal liquid of rats (top). Measurement of MPO activity collected in peritoneal liquid after 48 h of drug administration (bottom). (*) MPO concentration in ref group is significantly lower than the other groups, and the activity of MPO in the carrageenan group is significantly higher than in the other groups (P < 0.001, Shapiro–Wilk test).

In summary, new potent reversible and irreversible inhibitors of MPO were developed through the implementation of dynamic combinatorial libraries. Starting from a series of aldehydes and hydrazines, a three-step procedure allowed to select four couples (1A–1C, 6A–1C, 13A–1C, and 13B–1C), which demonstrated high activity as individual ligands, the lowest IC₅₀ value (79 nM) being attained with compound 13A–1C. Docking predictions highlighted the interaction between the hydralazine-based substrates and MPO, and further mechanistic investigations correlated their mode of inhibition with the predicted models. According to the kinetic study, 1A–1C, 6A–1C, and 13B–1C belong to the class of irreversible inhibitors, while ligand 13A–1C suppresses the activity reversibly. Hence, 13B–1C features the lowest IC₅₀ values reported to date for an irreversible MPO inhibitor. At last, in vivo evaluation demonstrated that one dose of irreversible inhibitors is able to suppress the activity of MPO released upon proceeding inflammation.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.6b00417.

• Experimental data, kinetic experiments, in vivo tests, and docking poses (PDF)

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

J.S. is a Research Fellow of the Belgian National Fund for Scientific Research (FRS-FNRS).

The authors declare no competing financial interest.