Design, synthesis, and biological evaluation of tetrazole analogs of Cl-amidine as protein arginine deiminase inhibitors.

Protein arginine deiminases (PADs) catalyze the post-translational hydrolysis of arginine residues to form citrulline. This once obscure modification is now known to play a key role in the etiology of multiple autoimmune diseases (e.g., rheumatoid arthritis, multiple sclerosis, lupus, and ulcerative colitis) and in some forms of cancer. Among the five human PADs (PAD1, -2, -3, -4, and -6), it is unclear which isozyme contributes to disease pathogenesis. Toward the identification of potent, selective, and bioavailable PAD inhibitors that can be used to elucidate the specific roles of each isozyme, we describe tetrazole analogs as suitable backbone amide bond bioisosteres for the parent pan PAD inhibitor Cl-amidine. These tetrazole based analogs are highly potent and show selectivity toward particular isozymes. Importantly, one of the compounds, biphenyl tetrazole tert-butyl Cl-amidine (compound 13), exhibits enhanced cell killing in a PAD4 expressing osteosarcoma bone marrow (U2OS) cell line and can also block the formation of neutrophil extracellular traps. These bioisosteres represent an important step in our efforts to develop stable, bioavailable, and selective inhibitors for the PADs.

Introduction

Protein arginine deiminases (PADs), members of the amidinotransferase superfamily of enzymes, catalyze the hydrolysis of arginine residues to form citrulline (Figure 1).[1−3] There are five PADs (PAD1, -2, -3, -4, and -6) in humans and other mammals.[1−3] These enzymes are highly related with 50% interisozyme sequence identity and require micromolar levels of calcium for full activity; calcium binding causes a conformational change that allosterically activates the enzymes.[1−3] Although highly related, these enzymes display unique tissue distribution patterns. For example, PAD2 is expressed in most tissues and cell types, whereas PAD1 and PAD3 are generally restricted to the skin and hair follicles and PAD4 and PAD6 are primarily expressed in neutrophils and oocytes, respectively.[3] Although we are only beginning to understand how the PADs contribute to normal human physiology, it is known that these enzymes help control myelination, differentiation, the epigenetic control of gene transcription, the innate immune response, and the maintenance of pluripotency.[4,5] In addition to these processes, dysregulated PAD activity is associated with the onset and progression of multiple inflammatory diseases (e.g., rheumatoid arthritis, multiple sclerosis, lupus, inflammatory bowel disease, and ulcerative colitis) and also some forms of cancer.[3,6−14]

Figure 1

Protein arginine deiminases (PADs) convert arginine residues into citrulline.

Although it is not understood how the PADs contribute to such a disparate number of diseases, common links include their role in controlling the formation of neutrophil extracellular traps (NETs) as well as their ability to modulate the expression of both proliferative and antiproliferative genes.[15,16] The overwhelming evidence linking dysregulated PAD activity to disease pathogenesis prompted the initiation of several PAD inhibitor programs.[17,18] Prior work in this area includes the identification of several reversible PAD inhibitors (e.g., Taxol, methylarginines, and minocycline); however, these compounds are very weak PAD inhibitors or have multiple off targets.[19−27] As such, they have shown limited utility as tool compounds to study PAD biology. By contrast, the haloacetamidine class of compounds has provided key insights into PAD biology.[17,18] Cl-amidine, the most widely used member of this compound class, has been used to demonstrate PAD involvement in the maintenance of pluripotency, the epigenetic control of gene transcription, and the formation of neutrophil extracellular traps.[16,28−32] Additionally, Cl-amidine shows efficacy in multiple preclinical models of autoimmunity and cancer including ulcerative colitis, lupus, rheumatoid arthritis, and atherosclerosis as well as breast cancer.[12,33−41] To enhance the potency, selectivity, and bioavailability of Cl-amidine, we and others have generated a number of derivatives including TDFA (a tripeptide F-amidine analog) and YW3-56.[12,42] However, these compounds are all peptide based and subject to degradation by proteolysis. This is even the case for Cl-amidine (Scheme 1) because the basic chemical scaffold (i.e., benzoyl arginine amide, BAA) is a trypsin substrate.

Scheme 1

C-Terminal Bioisosteric Modification of Cl-amidine

Recently, we synthesized a series of d-ornithine based compounds hypothesizing that these analogs would be more stable relative to their l-amino acid counterparts.[43] Although subsets were potent PAD inhibitors, d-amino acid derivatives showed only minor improvements in their overall stability. In our continuous efforts to identify stable, potent, and selective inhibitors for the PADs, we describe herein, the design, synthesis, and biological evaluation of tetrazole analogs of Cl-amidine.

Results and Discussion

Chemistry

Given that tetrazoles are bioisosteres for carboxylic acids and cis/trans amides,[44−46] we hypothesized that the replacement of the C-terminal amide bond in Cl-amidine with a tetrazole ring would improve the stability of the parent compound. Notably, the tetrazolium ring tends to be ionized at pH 7.4, exhibiting a pKa value of 4.9, thereby mimicking the electronegativity of the C-terminal carboxamide. The synthesis of these tetrazole based analogs began with the corresponding Bz-Orn(Boc)-amide (Scheme 2). The C-terminal carboxamide was dehydrated with triethylamine and TFAA to give the corresponding cyano compound. The cyano ornithine was then heated with sodium azide in PrOH/H2O to generate the tetrazole. Deprotection of the Boc protecting group with TFA led to the formation of two products in one pot: the desired tetrazole and a tetrazole tert-butyl derivative. The attachment of a tert-butyl group onto a highly acidic tetrazolium ring during the course of Boc cleavage with TFA has precedence in the literature.[47] Note that these substituted tetrazoles are no longer acidic because of substitution of the acidic nitrogen. The tetrazole and tetrazole tert-butyl derived ornithines were separated by HPLC, and then the chloro- and fluoroacetamidine warheads were installed using standard protocols.[48] Purification yielded a series of tetrazole derived haloacetamidines in good yield (68–78%) (Scheme 2).

Scheme 2

Synthesis of Tetrazole Analogs of Cl-amidine and F-amidine

With the tetrazole chemistry in place, we next considered that replacement of the N-terminal benzoyl group with a biphenyl benzoyl moiety would increase the hydrophobicity of the compound to enhance cellular uptake. To this end, the commercially available H2N-Orn(Boc)-OH was treated with 4-phenylbenzoyl chloride to give biphenyl benzoyl ornithine which was converted to the corresponding amide. The amide was then further converted to the corresponding tetrazole and tetrazole tert-butyl ornithine derivatives using the methodology outlined in Scheme 3. The warheads were then attached to the individual ornithine to give another set of compounds in good yield (>68%).

Scheme 3

Synthesis of Biphenyl Tetrazole Analogs of Cl-amidine and F-amidine

Since o-Cl-amidine and o-F-amidine are highly potent PAD inhibitors,[35] we anticipated that the attachment of an o-carboxylate on the phenyl group at the N-terminus of the tetrazole based compounds would also enhance compound potency and selectivity for the PAD isozymes. The synthesis of o-carboxylates began with Fmoc-Orn(Boc)-amide which was converted to the corresponding tetrazole compound in two steps (Scheme 4). The Fmoc group was then cleaved with 20% piperidine in DMF, and the resulting compound was treated with phthalic anhydride in THF at room temperature to install the o-carboxylate on the phenyl ring. The Boc group was then cleaved with TFA, which afforded the tetrazole and tetrazole tert-butyl ornithines in good yield. The chloro- and fluoroacetamidine warheads were then installed to give the desired o-carboxylate tetrazole derivatives (Scheme 4).

Scheme 4

Synthesis of o-Carboxylate Containing Tetrazole Analogs of Cl-amidine and F-amidine

Compound Potency and Selectivity

With the tetrazole analogs in hand (4–7, 11–14, 19–22), we next evaluated their potency and selectivity by determining kinact/KI values for PAD1, -2, -3, and -4. kinact/KI is used because it is the best measure of potency for an irreversible inhibitor. PAD6 was not tested because it shows no in vitro activity. Notably, the highest potencies and selectivities were obtained for the o-carboxyl-containing tetrazoles (Figure 2, Table S1). We have previously reported that the installment of a carboxyl group ortho to the N-terminal amide increases potency,[35] likely because of the formation of favorable H-bonding and/or electrostatic interactions between the o-carboxylate and W347 and R374.[35] For Cl-ortho tetrazole (19), the kinact/KI values are 82 000 and 159 000 M–1 min–1 for PAD1 and PAD4, respectively. Cl-ortho tetrazole (19) also showed 15-fold selectivity versus PAD2 and PAD3. F-ortho tetrazole (20) was similarly potent for PAD1 and exhibits 20- to 30-fold selectivity over PAD2 and PAD3 when compared to PAD1. Interestingly, the use of the fluoroacetamidine warhead significantly reduces the potency of this compound toward PAD4, indicating that significant selectivity can be achieved by altering the identity of the warhead. Moreover, introduction of the tert-butyl moiety further altered the selectivity of the remaining compounds. For example, both Cl-ortho tetrazole tert-butyl (21) and F-ortho tetrazole tert-butyl (22) show a significant improvement in their ability to inhibit PAD2. This trend was observed for the other tert-butyl derivatives described herein. For example, the kinact/KI values of tetrazole tert-butyl Cl-amidine (6) and tetrazole tert-butyl-F-amidine (7) for PAD2 are 7500 and 6500 M–1 min–1 versus 1200 and 380 M–1 min–1 for Cl-amidine and F-amidine, respectively. In a similar fashion, the biphenyl derived tetrazole tert-butyl compounds possess higher selectivity toward PAD2. For example, tetrazole tert-butyl-F-amidine (7) and biphenyl tetrazole tert-butyl-F-amidine (14) preferentially inhibit PAD2 by 3- to 25-fold with the highest selectivity being observed for PAD2 over PAD4. It is important to note that both compounds 7 and 14 are highly selective PAD2 inhibitors relative to the other PADs. The preferential inhibition of PAD2 suggests that the fluoroacetamidine warhead is more suitably oriented within the PAD2 active site. Note that since saturation was not observed in the plots of the pseudo-first-order rates of inactivation versus concentration of inhibitor, it is difficult to ascertain whether the different potencies of the compounds toward the different isozymes are driven by effects on the reactivity of the electrophile (i.e., kinact) or differences in the inherent binding affinity of the inhibitors (i.e., KI).

Figure 2

Selectivity of tetrazole haloacetamidines: (A) selectivity of tetrazole analogs of Cl-amidine; (B) selectivity of tetrazole analogs of F-amidine.

Cell Viability Studies

To examine the antiproliferative activities of these tetrazole analogs, we next evaluated their ability to inhibit the growth of U2OS cells using a fixed concentration of inhibitor (i.e., 20 μM) (Figure 3A and Figure 3B). Cell viability was assessed with the colorimetric XTT assay. Impressively, biphenyl tetrazole tert-butyl Cl-amidine (13) completely abolished cell growth at 20 μM. Biphenyl tetrazole tert-butyl-F-amidine (14) also inhibited cell growth, although the effects of this compound were markedly attenuated (more than 50% of cells are viable when cells were treated with 20 μM of the fluoro analog) (Figure 3B). The remaining compounds showed no effects on cell viability at this concentration. To provide a more direct measure of the cellular potency of our two best compounds (13 and 14), we determined the concentration of compound that reduced cell viability by 50%, i.e., the EC50 value (Figure 3C). Consistent with our initial screening data, the EC50 of compound 13 (10 ± 2.5 μM) was significantly better than the value obtained for compound 14 (EC50 = 45 ± 1.2 μM). Importantly, the EC50 values are 16- and 3.5-fold better than those obtained for the parent compound Cl-amidine (EC50 = 160 ± 20 μM). This improved cellular efficacy is likely due to the increased hydrophobicity of the biphenyl compound enhancing cell penetration. Although the o-carboxyl containing tetrazoles are very potent in vitro PAD inhibitors, the presence of the negatively charged carboxyl group likely reduces their overall bioavailability. Overall, these data highlight the potential utility of targeting the PADs for the development of an anticancer therapeutic. This is especially true for the fluoroacetamidine containing compounds because the inherent reactivity of this warhead is quite low and shows few off targets.[49,50] We do note, however, that for the chloroacetamidine containing compounds, it is difficult to directly link their cytotoxicity to PAD inhibition versus an off target effect. We are currently using these scaffolds to develop next generation activity-based protein profiling reagents to address this possibility.

Figure 3

Cell efficacy studies. (A) Cell viability studies with the tetrazole analogs of Cl-amidine. U2OS cells were treated with compounds at final concentration of 20 μM, and cell viability was measured by the colorimetric XTT assay. Biphenyl tetrazole tert-butyl Cl-amidine (13) completely abolishes the growth of U2OS cells at 20 μM. (B) Cell viability studies with the tetrazole analogs of F-amidine. (C) EC50 of biphenyl tetrazole tert-butyl compounds. Compounds 13 and 14 have lower EC50 values compared to Cl-amidine in U2OS cells. (D) Stability of tetrazole analogs of Cl-amidine in mouse hepatic microsomes compared to Cl-amidine.

Microsomal Stability Studies

We next evaluated the stability of a subset of compounds in a murine hepatic microsome stability assay which utilizes liver microsomes. Liver microsomes possess many of the enzymes responsible for drug metabolism in vivo and are also a good predictor of drug clearance properties.[51] On the basis of these data, it is clear that the fluoroacetamidine containing compounds are significantly more stable than their chloroacetamidine containing counterparts whose half-lives are similar to that obtained for Cl-amidine (Figure 3D). The one exception is the ortho carboxyl derived tetrazole Cl-amidine (21) whose half-life mimics those obtained for the fluoro analogs, suggesting that modification of the N-terminus substantially improves the overall stability of the compounds.

Neutrophil Extracellular Trap (NET) Inhibition Studies

We and others have shown that PAD4 mediated histone hypercitrullination induces heterochromatin decondensation and chromatin unfolding to form NETs.[52−56] Neutrophil extracellular trap (NET) formation causes direct organ damage and can trigger endothelial and epithelial cell death.[37] Inhibition of PAD4 reduces NET formation in mouse neutrophils in vivo and human neutrophils in vitro.[15] We have also shown that Cl-amidine treatment blocks NET formation and modulates the lupus phenotype in animal models.[37] Furthermore, Cl-amidine mitigates atherosclerosis in the apolipoprotein-E (Apoe) murine model[40] and this activity correlates with decreased NET formation. Given these precedents, we next investigated whether our two best in vitro inhibitors, i.e., compounds 13 and 21, could block NET formation. To this end, mouse neutrophils were treated with PMA to stimulate NET formation in the absence and presence of increasing amounts of Cl-amidine, compound 13, and compound 21. Cl-amidine was used as the reference compound. NET formation was then quantified using the DNA/neutrophil elastase overlap assay. Though 21 is very potent in vitro, it inhibits NET formation only at very high concentrations. The negatively charged carboxyl group again likely limits its bioavailability. By contrast, the biphenyl derivative 13 is significantly more potent than Cl-amidine in the NET assays (Figure 4), despite its being a significantly poorer PAD4 inhibitor in vitro. The enhanced cellular activity likely reflects the hydrophobic nature of the compound which enhances cellular uptake. We also evaluated the toxicity of 13 and 21, our two best inhibitors, against human neutrophils. The results of these studies indicate that 21 displays very limited cytotoxicity (EC50 = 985 ± 20). By contrast, 13 (EC50 = 31 ± 1.0) is considerably more toxic. Nevertheless, it is noteworthy that we see considerable inhibition of NET formation at lower doses (EC50 ≈ 20 μM) than those that cause cell killing.

Figure 4

Biphenyl tetrazole tert-butyl Cl-amidine (13) and o-carboxyl tetrazole tert-butyl Cl-amidine (21) inhibit NET formation. The DNA/neutrophil elastase overlap assay suggests that compound 13 significantly reduces NET formation compared to Cl-amidine. Compound 21 also inhibits NET formation at higher doses: (∗) p < 0.05 and (∗∗) p < 0.01.

Conclusions

In summary, we identified tetrazoles as a suitable C-terminal bioisosteric modification of Cl-amidine. A subset of the analogs show enhanced potencies and selectivities relative to Cl-amidine. Importantly, we confirmed that installation of an o-carboxylate enhances the in vitro potency of the compounds by up to 30-fold highlighting the importance of this pharmacophore. We also showed that incorporation of a tert-butyl group on the tetrazolium ring markedly increased the selectivity of the compounds for PAD2. Finally, our data indicate that enhanced cellular permeability can be achieved by increasing the hydrophobicity of the compounds. These design characteristics will be incorporated into future analogs as part of our continuing efforts to develop isozyme specific inhibitors for all of the PADs.

Materials and Methods

Chemicals

Dithiothreitol (DTT), 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), ammonium iron(III) sulfate dodecahydrate, tris(2-carboxyethyl)phosphine (TCEP), and thiosemicarbazide were acquired from Sigma–Aldrich. Diacetylmonooxime (DAMO), N-α-benzoyl-l-arginine ethyl ester (BAEE), and N-α-benzoyl-l-arginine amide (BAA) were obtained from Acros. Detailed synthetic procedures are described in the Supporting Information. PAD1, -2, -3, and -4 were purified analogously to previously described methods.[35,57]

Inactivation Kinetics

The kinetic parameters of inactivation were determined by incubating PAD1, -2, or -4 (2.0 μM) or PAD3 (5.0 μM) in a prewarmed (10 min, 37 °C) inactivation mixture (50 mM HEPES, 10 mM CaCl2, and 2 mM DTT, pH 7.6, with a final volume of 60 μL) containing various concentrations of inhibitor. Aliquots were removed at various time points (0, 2, 6, 10, 15, 20, and 30 min or 0, 0.5, 1, 1.5, 2, 4, and 6 min) and added to a prewarmed (10 min, 37 °C) reaction mixture (50 mM HEPES, 50 mM NaCl, 10 mM CaCl2, 2 mM DTT, and 10 mM BAEE or 10 mM BAA in the case of PAD3; pH 7.6). After 15 min, reactions were quenched in liquid nitrogen and citrulline production quantified using the COLDER assay.[58,59] Data were plotted as a function of time and fit to eq 1,using GraFit, version 5.0.11, where v is velocity, vo is the initial velocity, kobs is the pseudo-first-order rate constant of inactivation (i.e., kobs), and t is time. The kobs values were then plotted against inhibitor concentration, and the data were fit to eq 2,using GraFit, version 5.0.11. Since saturation was not observed in any of the plots, kinact/KI was calculated from the slope of the line. kinact corresponds to the maximal rate of inactivation, and KI is the concentration of inhibitor that gives half-maximal inactivation. For a subset of the less potent compounds, kinact/KI values were determined by dividing the kobs value at a single concentration by the concentration of the inhibitor tested.

Cell Viability

The human osteosarcoma (U2OS) cell line was plated (2.5 × 106 cells/well) in a 96-well plate. The next day the cells were treated with compounds (5 μL, 20 μM final), DMSO (5 μL), Triton X-100 (5 μL), or various concentrations of 13 or 14 and incubated for 72 h. Cell viability was measured using the XTT reagent kit (ATCC) by reading the absorption at 475 nm using a Spectramax plate reader. EC50 values for cell growth inhibition were determined by fitting an eight-point dose–response curve to eq 3,using GraFit5.0.11, where range is the uninhibited value minus the background and s is the slope factor. For in vitro cytotoxicity assays with neutrophils, freshly isolated human neutrophils were resuspended in RPMI 1640 medium containing 10% fetal bovine serum and then seeded into poly-l-lysine coated 96-well plates at 40 000 cells/well. After the cells were allowed to adhere for 1 h, neutrophils were incubated for 4 h with 13 or 21 at concentrations ranging from 1 to 500 μM. Cell viability after drug exposure was measured using the XTT cell viability kit as described above.

Neutrophil Isolation

C57BL/6 mice were purchased from The Jackson Laboratory. Bone marrow neutrophils were isolated essentially as described previously.[60] Briefly, bone marrow was flushed from femurs and tibias with Hank’s balanced salt solution supplemented with 15 mM EDTA. Cells were then spun on a discontinuous Percoll gradient (52%, 69%, 78%) at 1500g for 30 min. Cells from the 69–78% interface were collected, and red blood cells were lysed. Cells were >95% Ly-6G-positive and had typical segmented nuclei by microscopy.

NET Quantification and Microscopy

A protocol similar to what we have described previously was followed.[61] Briefly, 1.5 × 105 neutrophils were seeded onto glass coverslips coated with 0.001% poly-l-lysine (Sigma). PAD inhibitors were used at the indicated concentrations, including a 30 min pretreatment in Locke’s solution (150 mM NaCl, 5 mM KCl, 2 mM CaCl2, 0.1% glucose, and 10 mM HEPES buffer, pH 7.3). Stimulation was with 100 nM PMA (Sigma) for 3–4 h in RPMI-1640 supplemented with l-glutamine, 2% BSA, and 10 mM HEPES buffer. Cells were then fixed with 4% paraformaldehyde (PFA) for 20 min, followed by blocking with 10% fetal bovine serum; cells were not specifically permeabilized. DNA was stained with Hoechst 33342 (Invitrogen), while protein staining was with a rabbit polyclonal antibody to myeloperoxidase (A0398, Dako) followed by FITC-conjugated anti-rabbit IgG (4052-02, SouthernBiotech). After staining, coverslips were mounted in Prolong Gold antifade reagent (Invitrogen). Images were collected with an Olympus microscope (IX70) and a CoolSNAP HQ2 monochrome camera (Photometrics) with Metamorph Premier software (Molecular Devices), typically at 400× magnification. Statistical background correction and image overlays were with Metamorph, and the recorded images were loaded onto Adobe Photoshop for further analysis, at which time NETs were manually quantified by two blinded observers. Decondensed nuclei that also stained positively for myeloperoxidase were considered NETs and digitally recorded to prevent multiple counts. The percentage of NETs was calculated as the average of at least five fields and normalized to the total number of cells.