Photochemical Control of Protein Arginine Deiminase (PAD) Activity

Abstract

Protein Arginine deiminases (PADs) play an important role in the pathogenesis of various diseases, including rheumatoid arthritis, multiple sclerosis, lupus, ulcerative colitis and breast cancer. Therefore, the development of PAD-inhibitors has drawn significant research interest in recent years. Herein, we describe the development of the first photoswitchable PAD-inhibitors. These compounds possess an azobenzene photoswitch to optically control PAD activity. Screening of a series of inhibitors structurally similar to BB-Cl-Amidine afforded compounds 1 and 2 as the most promising candidates for the light-controlled inhibition of PAD2; the cis-isomer of 1 is 10-fold more potent than its trans-isomer, whereas the trans-isomer of 2 is 45-fold more potent than the corresponding cis-isomer. The altered inhibitory potency upon photoisomerization has been confirmed in a competitive activity-based protein profiling (ABPP) assay. Further investigations indicate that the trans-isomer of 2 is an irreversible inhibitor, whereas the cis-isomer acts as a competitive inhibitor. In cells, the trans-isomer of compound 1 is completely inactive, whereas the cis-isomer inhibits histone H3-citrullination in a dose-dependent manner. Taken together, 1 serves as the foundation for developing photopharmaceuticals that can be activated at the desired tissue, using light, to treat diseases where PAD activity is dysregulated.

INTRODUCTION

Protein arginine deiminases (PADs) are cysteine hydrolases that mediate the conversion of arginine to citrulline (Figure 1A).^1,2 Five isozymes (PAD1, 2, 3, 4 and 6) are known but only four of them (PADs 1–4) are catalytically active.³ The deiminase activity of PADs 1–4 is strictly regulated by the presence of Ca²⁺ ions, and the holoenzymes are known to bind four (PAD1), five (PADs 3 and 4) and six (PAD2) calcium ions at distinct sites.^4,5,6 Although calcium does not play a direct role in catalysis, recent crystal structures indicate that it induces a series of structural rearrangements that leads to the formation of the catalytically competent state. Particularly, the movement of the nucleophilic cysteine (Cys647 and Cys645 in PAD2 and PAD4, respectively) into the active site is triggered by calcium.⁶ PAD-mediated protein citrullination is known to regulate various cellular processes, including neutrophil extracellular trap (NET) formation, the epigenetic regulation of gene transcription, and the maintenance of pluripotency. Moreover, aberrant protein citrullination is associated with various autoimmune diseases, including rheumatoid arthritis (RA), lupus, multiple sclerosis, ulcerative colitis, and certain cancers.^{1, 2, 7–14}

Figure 1.

(A) PAD-catalyzed citrullination of peptidyl-arginine. (B) Chemical structures of pan-PAD inhibitors and the k_inact/K_I values for BB-Cl-Amidine. (C) Cocrystal structure of PAD4 bound to BB-F-Amidine indicating the covalent modification of Cys645 by the inhibitor (PDB code 5N0M).¹⁵

Given the potential role of PADs in human pathology, the development of PAD inhibitors has attracted significant attention.^{1, 2, 15} The most widely-used pan-PAD inhibitor is Cl-amidine (Figure 2B), which possesses a chloroacetamidine warhead that mimics the substrate guanidinium and irreversibly inhibits the PADs. Notably, Cl-amidine shows excellent efficacy in animal models of rheumatoid arthritis (RA), lupus, ulcerative colitis, spinal cord injury, breast cancer and atherosclerosis.^{13, 16–18} Structure-activity relationships ultimately led to the development of BB-Cl-amidine, a second-generation PAD inhibitor in which the phenyl and carboxamide groups of Cl-Amidine are replaced with a biphenyl and benzimidazole group, respectively (Figure 1B).¹⁷ BB-Cl-amidine exhibits at least 10-fold more potency than Cl-amidine in cell-based and animal experiments.^{17, 19, 20} However, the fluoroacetamidine counterparts of Cl-amidine and BB-Cl-amidine, i.e. F-Amidine and BB-F-amidine, exhibit significantly less inhibitory activity, likely due to the lower electrophilicity of the fluoroacetamidine warhead (Figure 1B).^{21, 22} In agreement with the proposed mechanism of inhibition,²³ the cocrystal structure of PAD4 bound to BB-F-amidine indicates that nucleophilic attack of the active site cysteine on the haloacetamidine warhead displaces the halide and irreversibly modifies the enzyme active site (Figure 1C).¹⁵ Recently, we showed that BB-F-Yne, an alkyne bearing fluoroacetamidine based PAD inhibitor, is remarkably selective for the PADs in cell based assays.²⁴ By contrast, BB-Cl-yne, possesses a number of off targets including, β-tubulin, β-Actin, clathrin heavy chain 1 (CLTC), bifunctional glutamate/proline-tRNA ligase (EPRS), heterogeneous nuclear ribonucleoprotein U (HNRPU).²⁴ These observations suggest that localized activation or deactivation of chloroacetamidine-containing inhibitors in the target tissue may be an alternative approach to limit potential systemic toxicities.

Figure 2.

(A) The chemical structures of trans- and cis-isomers of compounds 1-8 (see Scheme 1 for variable groups). (B) The change in the UV-vis spectra of compound 1 upon irradiation with UV-A, demonstrating the photoisomerization of 1T into 1C. (C) Optimisation of UVA-exposure time from the change in absorbance of 1 at 324 nm. (D) Photoisomerization of compound 2 monitored by UV-Vis spectra. (E) Thermal stability of 1C in aqueous buffer at 37 °C. Assay conditions: compound 1 (30 μM), 100 mM TRIS pH 7.4, 50 mM NaCl.

Photopharmacology, which relies on the activation or inactivation of drug molecules in the targeted tissue by light, has led to the development of several light-controlled ion-channel blockers, receptor antagonists, antibiotics, and enzyme inhibitors.^25–32 Light serves as a suitable external, noninvasive stimuli to modulate the potency of drug molecules and it can be delivered with high spatiotemporal precision and with a wide range of intensities as well as wavelengths. Photopharmaceuticals are generally developed by incorporating photoresponsive groups into bioactive molecules. The simplest and most widely-used example of such a photoresponsive element is an azobenzene. Azobenzenes undergo a trans-cis reversible isomerisation in the presence of light. While the thermodynamically more stable trans-isomer can be switched to the less stable, bent cis-isomer by UV light (i.e., 350 nm), the reverse process can be achieved by shining blue light (450 nm).^{29, 30} The excitation wavelengths and the half-life of the photoisomers can also be modulated by incorporating different substitutions on the azobenzene scaffold. The two isomers of azobenzene mainly differ from each other in shape and polarity and consequently, they exhibit different biological activities. Therefore, we hypothesized that the incorporation of azobenzene switches in BB-Cl-amidine, by replacing the biphenyl group, would result in a PAD-inhibitor that is structurally-similar to BB-Cl-amidine but can be activated or inactivated by light. Herein, we report the development of photoresponsive PAD inhibitors that can be used to optically control the inhibition of PADs in vitro and in cell-based assays.

RESULTS AND DISCUSSION

To achieve the optical control of PAD-activity, we synthesized various azobenzene-containing compounds carrying either a fluoro- or chloroacetamidine warhead (compounds 1-8, Scheme 1). While compounds 1 and 2 are the simple azobenzene-modified analogues of BB-Cl-Amidine and BB-F-Amidine, respectively, compounds 3-8 carries various substitutions on the azobenzene as well as the benzimidazole moiety. We hypothesized that the p-hydroxyl and p-dimethylamino substitutions on the terminal aromatic ring of the azobenzene moiety in compounds 3-6 may alter the binding affinity of different photoisomers for the enzyme active site by both electronic and steric influence. The N-methyl and 4-methoxy substitutions on the benzimidazole ring in compounds 7 and 8 were incorporated because we previously showed that these modifications significantly increase the potency and isoform-specificity of benzimidazole-based PAD inhibitors.¹⁵ The synthesis of compounds 1-8 is depicted in Scheme 1 and it involves the common precursors 9 and 10, which were synthesized by following a previously reported procedure.^{15, 17}

Scheme 1.

Chemical structures and the synthesis of compounds 1-8.

With this family of the photoresponsive PAD-inhibitors in hand, we next investigated their trans-cis isomerisation by UV-visible spectroscopy (Figure 2A and Figure S1). The thermodynamically more stable trans-isomer of compound 1 (1T) exhibits maximal absorption (λ_max) at 324 nm (Figure 2B). However, upon irradiation with UV-A (λ = 320–380 nm), this peak disappears with time, indicating the formation of the cis-isomer of this compound (i.e., 1C). A plot of absorbance at 324 nm versus time indicates that 40–60 s of UV-A exposure is sufficient to convert more than 80 % of 1T into 1C (Figure 2C). A similar phenomenon was also observed for the fluoroacetamidine warhead-containing compound 2 (Figure 2D), indicating that the halogen substitution on the electrophilic warhead does not impact the photoisomerisation of compounds 1 and 2. We also investigated the thermal stability of the cis-isomer by UV-vis spectroscopy, as fast back conversion to the trans-isomer (from the thermodynamically less stable cis-isomer) would limit their utility. Both 1C and 2C exhibit remarkable stability in aqueous buffer at 37 °C as the change in absorbance at 324 nm over 12 h is negligible after irradiating 1T or 2T with UV-A lamps (Figure 2E and Figure S2). Both 1C and 2C can be switched back to their trans-form, i.e. 1T and 2T, respectively, by irradiating with blue light (λ = 400–450 nm) (Figure S1G and H) similarly to other azobenzene-containing photopharmaceuticals.^25–32 This phenomenon, however, is not important for our purposes because compounds 1-8 are irreversible PAD-inhibitors and, as such, once the trans- or cis- isomer reacts with the active site cysteine, photoswitching will not turn off the inhibitor to regenerate the active enzyme. The photoswitching properties of compounds 3-8 were also characterised by UV-vis spectroscopy (Figure S1) and the observations are in agreement with the reported photoswitching behaviour of hydroxy- and diaminomethyl-substituted azobenzene photopharmaceuticals.^25–32

Having established the optimum photoswitching conditions (excitation wavelengths and exposure time), we set out to evaluate the inhibitory potency of the trans- (T) and cis- (C) isomers of the inhibitors in in vitro PAD-inhibition assays. PADs 1–4 were expressed and purified following standard protocols established in the Thompson lab.^{4, 5} Inhibition of citrulline production was quantified using the COLDER assay (see experimental section). For PAD1, replacement of the biphenyl group in BB-Cl-amidine with the azobenzene decreases the potency of 1T by almost 8-fold, while the fluorinated analogue, 2T exhibits almost 6-fold higher activity than BB-F-amidine for this isozyme (Table 1). Interestingly, compound 3T, with a p-hydroxy substitution, exhibits almost 13-fold higher activity than the unsubstituted analogue 1T, while the fluorinated analogue 4T exhibits 2-fold lower potency than 2T for PAD1. Furthermore, compounds 5T, 7T and 8T are inactive towards PAD1 (Table 1), indicating that the p-dimethylamino substitution on the azobenzene in combination with the N-methyl and 4-methoxy substitutions on the benzimidazole ring disfavors binding of the inhibitor to the PAD1 active site. Next, we evaluated the inhibitory potency of the cis-isomers of 1-8. The changes in the potency upon photoisomerisation of compounds 1-8 were generally found to be within 1–2-fold, although compound 6T exhibits 5-fold higher activity than 6C.

Table 1.

k_inact/K_I vales for the inhibition of PADs 1–4 by BB-Cl-Amidine, BB-F-Amidine, and trans- (T) and cis- (C) isomers of compounds 1-8 (see Scheme 1 for variable groups).

Compounds & photoisomers		kinact/KI (M−1min−1)
		PAD1	PAD2	PAD3	PAD4
BB-Cl-Amidine		16000a	5000a	6000a	14000a
BB-F-Amidine		900a	1200a	3400a	3750a
1	T	2300 ± 200b	600 ± 30b	1000 ± 60b	10510 ± 590c
	C	2000 ± 100b	5970 ± 670c	1500 ± 40b	4900 ± 100b
2	T	4920 ± 760c	4520 ± 760c	500 ± 77b	3880 ± 1320c
	C	2850 ± 320c	≤100 ± 20de	1100 ± 40b	4150 ± 420c
3	T	28770 ± 7020c	2100 ± 100b	6500 ± 100b	19920 ± 2170c
	C	12000 ± 570c	7330 ± 1280c	6300 ± 400b	27940 ± 180c
4	T	2130 ± 510c	900 ± 40b	2020 ± 210c	3640 ± 100c
	C	1030 ± 380c	4410 ± 980c	1300 ± 50b	5580 ± 350c
5	T	No Inhibition	230 ± 70d	No Inhibition	2440 ± 760c
	C	No Inhibition	≤100 ± 40d	No Inhibition	3070 ± 50c
6	T	870 ± 360c	4340 ± 650c	350 ± 50d	450 ± 40d
	C	180 ± 70d	10490 ± 1460c	510 ± 50d	450 ± 40d
7	T	≤40 ± 10d	≤100 ± 30d	≤130 ± 30d	600 ± 30b
	C	≤80 ± 30d	220 ± 30d	170 ± 20d	600 ± 100b
8	T	No Inhibition	360 ± 100d	No Inhibition	≤60 ± 20d
	C	No Inhibition	80 ± 40d	No Inhibition	≤80 ± 20d

^ak_inact/K_I values were taken from the published results (ref. 15).

^bk_inact/K_I was determined from a linear fit of the k_obs versus [I] data.

^ck_inact and K_I was determined from a nonlinear fit of the k_obs versus [I] data.

^dA single k_obs was determined.

^eDetailed kinetic studies indicate that 2C is actually a reversible inhibitor of PAD2 with a K_i value of 25.2 μM (see Figure S7).

Figure 5.

Inhibition of histone H3 citrullination in HEK293T/PAD2 cells by compound 1. Inhibitor concentrations [I] are given under each lane of the western-blot image. Citrullinated H3 (H3Cit) and H3 are shown in red and green, respectively. Quantification of each band yielded the H3Cit/H3 ratio, from which the % relative H3 citrullination was calculated.

Next, we tested their ability to inhibit PAD3 and PAD4. The potencies of the chloro- and fluoroacetamidine warhead-containing compounds 1-8 for PAD3-inhibition were generally found to be lower than BB-Cl-Amidine and BB-F-Amidine, respectively. Compound 3T exhibited the highest potency in the series towards PAD3 (Table 1). However, upon photoswitching to the cis-isomer (3C), there was no change in potency. For other compounds in the series, photoisomerisation had modest (1.5–2-fold) effects on potency. For PAD4, compounds 1T and 2T exhibit similar potency to BB-Cl-amidine and BB-F-amidine (Table 1). As observed for PAD1 and PAD3, compound 3T exhibited the highest inhibitory activity towards PAD4, which is almost 2-fold higher than that of 1T. However, its fluorinated analogue, 4T, is only just as potent as compound 2T for PAD4 inhibition. While 5T (with a p-dimethylamino substitution) exhibits reasonable activity towards PAD4, the fluorinated analogue 6T is 5-fold less active (Table 1). It should be noted that 5T is completely selective to PAD4 over PAD1 and PAD3, and such selectivity over PAD1 is rare in the literature. However, compounds 7T and 8T exhibit poor inhibitory activity for PAD4, indicating that the benzimidazole-ring substitutions disfavor binding of the inhibitor to the active site of PAD4. Similar to the inhibition of PAD1 and 3, the cis-conformation of these compounds are equipotent to the trans-conformation, e.g. 2C (k_inact/K_I = 4150 ± 420 M⁻¹min⁻¹) inhibits PAD4 as well as 2T (k_inact/K_I = 3900 ± 1300 M⁻¹min⁻¹). Moreover, there was only a two-fold change in potency between compounds 1T and 1C (Table 1).

Finally, we investigated the ability of these compounds to inhibit PAD2. Dysregulated PAD2 activity is strongly associated with multiple sclerosis, RA and breast cancer.^{1, 2} PAD2 is known to be secreted in the synovial fluid of arthritic joints, where it citrullinates numerous extracellular proteins, leading to the development of anticitrullinated protein antibodies (APCA), a diagnostic marker of RA.^{3, 12} PAD2 is also responsible for the citrullination of histone H3 at R26 position, which triggers the localized decondensation of chromatin and activates gene transcription.^{13, 33} Notably compound 1T is ~10-times less active than BB-Cl-amidine, indicating that replacement of the biphenyl group with the azobenzene disfavors interactions between the trans-form of the inhibitor and PAD2. Gratifyingly, its potency increases by 10-fold upon excitation to the cis-isomer, 1C (Table 1). Moreover, the k_inact/K_I value of 1C (6000 ± 680 M⁻¹min⁻¹) is almost identical to that of BB-Cl-amidine (5000 M⁻¹min⁻¹). The increased potency upon photoisomerisation is most likely due to enhanced binding to the PAD2 active site because the plots of k_obs versus inhibitor concentration move from linear (1T) to hyperbolic (1C) (Figure 3A), indicative of a lower K_I value for the cis-isomer. The k_inact and K_I values for 1C are 0.15 min⁻¹ and 25.2 μM, respectively (Figure 3B). These results indicate that the less potent compound 1T can be activated to the more potent compound 1C for the inhibition of PAD2 in the presence of UV-A radiation. Surprisingly, 2T, the fluorinated analogue of 1T, is almost 45-fold more active than its cis-isomer 2C (Table 1 and Figure 3C), indicating that the nature and size of the halogen atom in the haloacetamidine warhead plays an important role in the binding and reactivity of the inhibitor. 2T shows saturation kinetics with k_inact and K_I values of 0.3 min⁻¹ and 66.4 μM, respectively (Figure S4C). By contrast, 2C showed very little, if any, enzyme inactivation, as such a single k_obs was used to estimate the k_inact/K_I of 2C (k_inact/K_I ≤ 100 ± 20 M⁻¹min⁻¹) (Figure S4D). In comparison to compounds 1 and 2, compounds 3C and 4C are ~ 4- and 5-fold, respectively, more active than 3T and 4T, indicating that the p-hydroxy substitution in the azobenzene ring suppresses the optical control of the inhibition of PAD2 activity (Table 1 and Figure 3C). Compound 6T is also ~2-times more potent than 6C. Interestingly, 6C is ~57-, 20- and 23-fold more selective for PAD2 than PAD1, PAD3 and PAD4, respectively. Such isozyme selectivity has rarely been observed amongst previously developed PAD inhibitors. Furthermore, compound 8T is ~5-times more active than 8C, although the potency of 8T is very low (Table 1 and Figure 3C). Taken together, compounds 1 and 2 are the most promising candidates in the series for the optical control of PAD2 activity – compound 1 can be photo-activated by 10-fold, whereas compound 2 can be photo-deactivated by 45-fold.

Figure 3.

Concentration dependence of pseudo-first-order rate constant of inactivation (k_obs) of PAD2 by 1T (A) and 1C (B). (C) Potencies (k_inact/K_I) of trans- and cis-isomers of compounds 1-8 for the inhibition of PAD2. Assay conditions: PAD2 (2 μM), 100 mM TRIS pH 7.4, 50 mM NaCl, 2 mM DTT, 10 mM CaCl₂, and 10 mM BAEE.

To further confirm the results of our in vitro inhibition assays, we used a previously established competitive activity-based protein profiling (ABPP) assay to visualise the change in inhibitory potency upon photoisomerisation (Figure 4A and B). For these studies, PAD2 was treated with inhibitors in microplates for 30 min prior to the addition of rhodamine-conjugated fluoroamidine (RFA), a PAD-targeted ABPP. For strong inhibitors, RFA-labelling will be blocked, whereas for weak inhibitor a smaller proportion of PAD2 would be occupied, and RFA-labelling is enhanced (Figure 4B).

Figure 4.

(A) Chemical structure of RFA. (B) Workflow of labelling of PAD2 with RFA in the presence of inhibitors. RFA-labelling of PAD2 in the presence of 1T and 1C (C), 2T and 2C (D), 3T and 3C (E). F and C stands for fluorograph and coomassie stain, respectively. Concentrations of inhibitors [I] are given in micromolar (μM). Intensities of each band in the fluorograph were measured using ImageJ software and the relative decrease in the fluorescence intensity has been plotted against the concentration of inhibitor to obtain the IC₅₀ values. Assay conditions: PAD2 (2 μM), 100 mM TRIS pH 7.4, 50 mM NaCl, 2 mM DTT, and 10 mM CaCl₂.

Consistent, with the k_inact/K_I data, 1T is a very weak PAD2 inhibitor (Figure 4C); the IC₅₀ is >100 μM. By contrast, the IC₅₀ of 1C, 9.1 μM, is more than 10-fold lower than that of 1T. Despite the fact that 2T is 45-fold more potent than 2C, these two isomers yielded similar IC₅₀ values, i.e. 11.8 μM and 23.8 μM respectively, suggesting that 2T is only 2-times more active than 2C (Figure 4D). To reconcile this contradiction, we considered the possibility that 2C binds to the PAD2 active site in such a conformation that precludes its reaction with C647, the active site cysteine, but still inhibits the enzyme reversibly. To test this possibility, we determined the K_m of N_α-benzoyl-L-arginine ethyl ester (BAEE), a known substrate of PAD2, in the presence of increasing concentrations of 2C. The resulting inhibition patterns indicate that 2C is a reversible, competitive inhibitor, with a K_i value of 25.2 μM (Figure S7). Since 3C is almost 4-fold more potent than 3T (Table 1), we also assayed this compound in this competitive ABPP assay. Interestingly, the IC₅₀ of 3C is >8-fold lower than that of 3T, roughly in agreement with the results obtained using the COLDER assay.

Motivated by these promising inhibition data, we next evaluated their ability to inhibit histone H3 citrullination in cells; histone H3 is citrullinated at multiple positions, including R2, R8, R17 and R26 by PAD2.^33–35 For these experiments, HEK293T/PAD2 overexpressing cells^{36, 37} were incubated with the inhibitor, CaCl₂ and ionomycin (a calcium ionophore) for 3 h, and then the cells were lysed, the soluble proteins in the lysate resolved by SDS-PAGE, and transferred to a PVDF membrane. The membrane was then incubated with primary antibodies for histone H3 and citrullinated histone H3 (Cit 2, 8, 17), followed by incubation with appropriate secondary antibodies and imaged by Licor analysis. Consistent with our in vitro data, 1T does not inhibit histone H3 citrullination even at 100 μM. By contrast, the light-activated isomer 1C inhibits citrullination in a dose-dependent manner (Figure 5), suggesting that compound 1 can be photoactivated to inhibit histone H3-citrullination in HEK293T/PAD2 cells. By contrast, 2T only showed modest inhibition (~33% at 100 μM) and 2C did not show any activity (Figure S8A). Similarly, 3C was slightly more active than 3T in inhibiting H3 citrullination by PAD2 (Figure S8B). These results suggest that in addition to inhibitory potency, cell-permeability of the different isomers plays an important role in inhibiting PAD2 activity in the HEK293T/PAD2 cells.

We also evaluated the cytotoxicity of the photoisomers of compounds 1-3 and BB-Cl-amidine in the HEK293T/PAD2 cells. Compound 1, which exhibited the best results for the photoactivation of inhibition of PAD2 was as cytotoxic as BB-Cl-amidine (Figure S9). Notably, the trans- and cis isomers of 1 do not show any difference in the cytotoxicity, suggesting that the cytotoxicity of compound 1 does not originate from the inhibition of PAD2. By contrast, compounds 2 and 3 are less toxic than 1.

Conclusions

Herein, we report the development of the first photoswitchable PAD inhibitor. A series of compounds, 1-8, carrying an electrophilic warhead were evaluated and in vitro inhibition experiments indicate that these compounds can optically control PAD2 activity, with 1 and 2 being the most promising candidates. While the cis-isomer (1C) exhibits 10-fold higher potency than the trans-isomer (1T) of 1, the trans-isomer (2T) exhibits 45-fold more potency than the cis-isomer (2C) of 2. Furthermore, alterations in the inhibitory potency upon photoisomerisation were confirmed using a competitive ABPP assay. This study indicates that the IC₅₀ of 1C is >10-fold lower than that of 1T. In contrast to the k_inact/K_I data, compound 2C is only 2-fold less potent than 2T in the ABPP-based assay. Detailed enzyme kinetics in the presence of increasing concentrations of 2C revealed that 2C is a reversible, competitive inhibitor of PAD2. Cell-based studies reveal that 1T is inactive, whereas the light-activated isomer 1C inhibits histone H3 citrullination in a dose-dependent manner in HEK293T/PAD2 cells.

In total, the present data sets a foundation for developing photopharmaceuticals to optically control PAD activity. Nevertheless, further research is required to overcome the limitations of the current molecules, including the poor tissue penetration of 350 nm light and modest isozyme selectivity. With regard to the former, introduction of fluoro- or chloro substitutions into the azobenzene scaffold will shift the excitation wavelength from UV to green or red light, respectively.³⁸ With regard to the latter, the PADs generally show tissue or disease specific expression patterns.³ For example, PAD2 is overexpressed in luminal breast cancer cells and its expression correlates with the level of the HER2 protooncogene.^{13, 33} Thus, compound 1 represents a promising starting point for the development of photopharmaceuticals against diseases in which PAD2 activity is dysregulated. Nevertheless, improvements in potency and interisozyme selectivity will likely be necessary and could be achieved by modifying the chromophore.

EXPERIMENTAL SECTION

Materials and methods

Fmoc-Orn(Boc)-OH was purchased from Chem-Impex International, Inc. 4-(phenylazo)benzoyl chloride, p-(p-dimethylaminophenylazo)benzoate dehydrate, DIPEA, anhydrous methanol, DMF, dichloromethane, piperidine, triethylamine, trifluoroacetic acid, chloroacetonitrile, fluoroacetonitrile and HPLC-grade acetonitrile were obtained from Sigma-Aldrich. 4′-Hydroxyazobenzene-4-carboxylic acid was bought from TCI Chemicals. Precoated silica gel plates were purchased from Merck. Deuterated solvents were acquired from Cambridge Isotope Laboratories. Mouse monoclonal anti-histone H3 (catalogue no. ab10799) and rabbit polyclonal anti-H3 (Citrulline R2, R8 and R17) (catalogue no. ab5103) were purchased from Abcam. NMR spectra were recorded in d₄-MeOH or d₆-DMSO as solvent. ¹H and ¹³C NMR spectra were recorded using a Bruker 500 MHz NMR spectrometer. Chemical shift values are cited with respect to SiMe₄ as the internal standard. Column chromatography was carried out in glass columns. Final compounds were purified by reverse-phase HPLC using a semi-preparative C18 column (Agilent, 21.2 × 250 mm, 10 μm) and water/acetonitrile gradient supplemented with 0.05% trifluoroacetic acid.

Synthesis

Detailed procedures for the synthesis of compounds 1-8 are given in the supporting information. All the compounds were isolated as trifluoroacetate salts and were characterized using ¹H, ¹³C NMR spectroscopy and ESI-Mass spectrometry. The purity of the compounds was analyzed by analytical HPLC.

Inactivation kinetics

Inactivation kinetic parameters for compounds 1-8 were determined by following reported procedures.^{22, 23} Briefly, PAD1, PAD2, PAD4 (2 μM) or PAD3 (5 μM) was added to a prewarmed (10 min, 37 °C) inactivation mixture (100 mM TRIS pH 7.4, 50 mM NaCl, 10 mM CaCl₂, 2 mM DTT, with a final volume of 50 μL) containing various concentrations of inhibitors. Aliquots (6 μL) of the inactivation mixture were removed at various time points and were added to a prewarmed (10 min, 37 °C) reaction mixture (100 mM TRIS pH 7.4, 50 mM NaCl, 10 mM CaCl₂, 2 mM DTT, and 10 mM BAEE or 10 mM BAA for PAD3, with a final volume of 60 μL). The reactions were quenched with liquid nitrogen after 15 min and citrulline production was quantified using the COLDER assay.^{5, 39} The time-dependence of PAD inhibition was fit to equation 1,

$$\mathrm{\nu }={\mathrm{\nu }}_0{\mathrm{e}}^{-k\mathrm{t}},$$ (1)

using Grafit, version 5.0.11, where ν is velocity, ν₀ is initial velocity, k (or k_obs is the pseudo-first-order rate constant of inactivation, and t is the time. If the rates of inactivation reached saturation, the concentration dependence of k_obs was fit to equation 2,

$$k_{\text{obs}}=k_{\text{inact}}[\mathrm{I}]/(K_{\mathrm{I}}+[\mathrm{I}]),$$ (2)

using Grafit, version 5.0.11, where k_inact is the maximal rate of inactivation, K_I is the concentration of inhibitor that yields half-maximal inactivation, and [I] is the concentration of inhibitor. If the plot of k_obs versus [I] did not saturate and was linear, then the value of k_inact/K_I was determined from the slope of the line. All the experiments were carried out at least in duplicate.

RFA-labelling

The labeling of PAD2 with RFA in the presence of inhibitor was carried out by following a previously reported procedure.^{36, 40, 41} Briefly, PAD2 (2 μM final) was added to the reaction mixture (100 mM TRIS pH 7.4, 50 mM NaCl, 10 mM CaCl₂, and 2 mM DTT in a final volume of 20 μL) containing DMSO or various concentrations of the inhibitor. The mixture was incubated at 37 °C for 30 min, followed by addition of RFA (10 μL, 2 μM final). Then the reaction mixture was incubated at 37 °C for a further 2 h, quenched with 5X SDS-PAGE dye, incubated at 95 °C for 15 min and then loaded onto a 10% SDS-PAGE gel. After electrophoresis, fluorescence of the protein bands was recorded using a typhoon scanner (excitation/emission maxima of ~546/579, respectively). Fluorescence intensity of the protein bands was quantified using ImageJ software and was plotted against the concentration of the inhibitor to obtain the IC₅₀ values.

Histone H3 Citrullination in HEK293TPAD2 Cells

HEK293T cells stably expressing human PAD2 (HEK293T/PAD2) were cultured as reported earlier.^{36, 37} Cells were grown to ~90% confluence, trypsinized, quenched with complete media and harvested by centrifugation at 3,000 rpm for 3.5 min. Then the cells were washed with 1X HBS four times and were resuspended in 1X HBS at 8 × 10⁶ cells/mL. 4 × 10⁵ cells were used for subsequent assays with each inhibitor concentration. Calcium chloride (1 mM), ionomycin (10 μM), and either DMSO or various concentrations of inhibitor were incubated with the cells for 3 h. The final concentration of DMSO in each sample was 1%. Triton X-100 (1% final) was then added to each sample and the suspension was sonicated at 4 °C for 1 h, followed by centrifugation at 21,000g for 15 min. Lysates were collected, soluble proteins in the lysates were quantified using the DC-assay (Bio-Rad) and were normalized. 2 μg of protein was loaded onto a 4–15% gradient SDS-PAGE gel, separated by electrophoresis and proteins were then transferred to PVDF membrane (Bio-Rad) at 80 V for 50 min. The membrane was then blocked by treating with phosphate buffered saline containing 0.1% Tween-20 (PBST) and 5% BSA, and the blocked membrane was incubated with primary antibodies for histone H3 (1:1,000) and histone H3Cit 2,8,17 (1:1000) in PBST with 5% BSA for 12 h at 4 °C. After washing the membrane with PBST, it was incubated with anti-mouse and anti-rabbit IgG Licor conjugate (1:5000) in PBST and 5% BSA for 1 h at 25 °C. The membrane was then washed with PBST and imaged by Licor analysis.

Cytotoxicity Studies

HEK293T/PAD2 cells were plated (2 × 10⁴ cells/well) in a 96-well plate and were allowed to grow for 24 h. DMSO or various concentrations of inhibitor were added to the wells and incubated for 24 h. Cell viability was measured using the XTT reagent kit (ATCC) by recording the absorbance at 475 nm and 660 nm. An eight-point dose-response curve was fit to the following equation 3 using GraphPad Prism 7.03 to determine the EC₅₀ values for cell-growth inhibition,

$$\mathrm{Y}=\text{Bottom}+(\text{Top-Bottom})/[1+{10}^{((\text{logEC}50\text{-X})\ast \text{Hillslope})}],$$ (3)

where Top and Bottom are plateaus of the dose-response curve, X is the log of inhibitor-concentration, Hillslope is the slope factor or Hill slope.

ABBREVIATIONS

RA

Rheumatoid arthritis

PAD

Protein arginine deiminase

ACPA

Anti-citrullinated protein antibodies

NET

Neutrophil extracellular trap

RFA

Rhodamine-conjugated F-amidine

ABPP

Activity-based protein profiling