Synthesis and Evaluation of Tiaprofenic Acid-derived UCHL5 Deubiquitinase Inhibitors

Abstract

The ubiquitin-proteasome system (UPS) plays an important role in maintaining protein homeostasis by degrading intracellular proteins. In the proteasome, poly-ubiquitinated proteins are deubiquitinated by three deubiquitinases (DUBs) associated with 19S regulatory particle before degradation via 20S core particle. Ubiquitin carboxyl-terminal hydrolase L5 (UCHL5) is one of three proteasome-associated DUBs that control the fate of ubiquitinated substrates implicated in cancer survival and progression. In this study, we have performed virtual screening of an FDA approved drug library with UCHL5 and discovered tiaprofenic acid (TA) as a potential binder. With molecular docking analysis and in-vitro DUB assay, we have designed, synthesized, and evaluated a series of TA derivatives for inhibition of UCHL5 activity. We demonstrate that one TA derivative, TAB2, acts as an inhibitor of UCHL5.

Graphical Abstract

1. Introduction

The ubiquitin proteasomal system (UPS) constitutes the major intracellular protein degradation system in eukaryotic cells. The UPS controls the lifetime of most intracellular proteins and eliminates misfolded and damaged proteins.^1–3 Consequently, the UPS activity is essential to maintain the protein homeostasis and regulates various cellular processes.^4–7 To be processed by the UPS, proteins destined for degradation are covalently conjugated to ubiquitin (Ub), via a series of enzymes, including ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2) and ubiquitin ligases (E3), in a multimeric form of Ub chains, namely polyubiquitination.^{8, 9} The polyubiquitinated conjugate is then recognized and degraded by the proteasome.

The 26S proteasome is composed of 20S proteolytic core particle (CP) and 19S regulatory particle (RP). 19S RP is responsible for binding to the polyubiquitinated substrate, deubiquitination, ATP-driven unfolding of the substrate, and translocation into 20S CP proteolytic chamber.^10–16 In this process, deubiquitination is mediated by three deubiquitinases (DUBs) associated with the 19S RP, including metalloprotease RPN11 at the base, and ubiquitin-specific protease 14 (USP14) and ubiquitin C-terminal hydrolase L5 (UCHL5) at the lid.^{17, 18} RPN11 cleaves at the base of Ub chains, which is coupled to unfolding and degradation of target proteins. On the other hand, USP14 and UCHL5 trim a distal tip of Ub chains, which has shown to rescue target proteins from proteolysis. ^{19, 20} Therefore, the DUB activities are important not only to remove and recycle ubiquitin from target proteins, but also to decide fate of ubiquitinated substrates to be processed for proteolysis or rescued from degradation.^{21, 22}

UCHL5 has two major domains, catalytic domain and C-terminal domain (CTD). The catalytic domain retains a close homology to the same family of DUBs, including UCHL1 and UCHL3. The CTD causes auto-inhibition of UCHL5 activity, which is reactivated when CTD binds to Ub receptor, Rpn13, at the 19S RP.^{23, 24} Therefore, UCHL5 in complex with Rpn13 at the proteasome displays higher DUB activity than UCHL5 alone. In addition to its role at the proteasome, UCHL5 plays a role in DNA repair or transcription by interacting with Ino80 chromatin-remodeling complex at the nucleus.²⁵ Although UCHL5 substrates are not yet completely understood, multiple lines of biological evidence support its role in cancer. UCHL5 overexpression is observed in many types of cancers, including esophageal squamous cell carcinoma and epithelial ovarian carcinoma.^{26, 27} Its high level of expression is a predictor of recurrent hepatocellular carcinoma.²⁸ UCHL5 deubiquitinates Smoothened, TGF-β, and NF-κB, whose stabilizations contribute to tumor cell survival, migration and invasion.^28–31 Despite its importance in cancer, there are only a few inhibitors developed to target UCHL5, many of which suffer from low selectivity.^32–35

In this report, throughout virtual screening, we discovered that tiaprofenic acid (TA) shows weak inhibition of UCHL5 activity. We have therefore designed and synthesized derivatives of TA guided by computational docking analyses. Our in vitro evaluation supports the conclusion that one TA derivative, TAB2, inhibits UCHL5 activity both in the absence and presence of Rpn13.

2. Results and Discussion

2.1. Virtual screening

We have initially performed virtual screening with an FDA approved drug library of 1,217 compounds (DSSTOX and ZINC database) against the active site of UCHL5 (PDB: 3IHR) via DOCK Blaster to discover potential UCHL5 inhibitors (Figure 1A).^36–39 Among a list of hit compounds obtained, (R)-tiaprofenic acid (TA) was selected after considering binding energies and contact points around the active site (Supplementary Figure 1). It was noted that several hits from virtual screening have a common structural feature, such as a hydrophobic core structure with a terminal carboxylic acid (examples in Supplementary Figure 1). Molecular docking analysis with AutoDock Vina program predicted that TA interacts with a few hydrophobic or basic residues in a pocket next to the active site that consists of a catalytic triad (C88, H164, D179) and an oxyanion hole (Q82) (Figure 1B and Supplementary Figure 2A).⁴⁰ To examine its experimental inhibitory potential, TA was obtained as a racemic mixture and evaluated with in vitro DUB assay using purified UCHL5 and Ub-rhodamine 110 (Ub-Rho) as a substrate, which restores fluorescence upon cleavage of a bond between Ub and rhodamine. TA has shown weak but apparent inhibition against UCHL5 activity with ca. 15-20% inhibition at 200 μM concentration (IC₅₀ = ~3 mM) (Figure 1C and Table 1).

Figure 1. Virtual screening and analysis of tiaprofenic acid (TA) with UCHL5.

(A) Virtual screening with UCHL5. TA was selected as a hit. (B) Molecular docking of TA with UCHL5 (PDB: 3IHR). Residues in a catalytic triad and oxyanion hole are colored in magenta. (C) Inhibition curve of UCHL5 activity by TA. UCHL5 activity was measured by using Ub-Rho as a substrate. Data represent the mean ± SD, n = 2 independent experiments.

Table 1.

IC₅₀ values for TA and its derivatives

Name	IC50 (μM)	Name	IC50 (μM)
TA	~2,946	TAB5	251.8 ± 1.8
TAA1	89.1 ± 1.5	TAB6	96.2 ± 1.6
TAA2	75.7 ± 1.1	TAB7	80.9 ± 1.6
TAB1	80.5 ± 1.8	TAB8	215.3 ± 1.7
TAB2	32.9 ± 1.5	TAB9	84.5 ± 1.7
TAB3	284.5 ± 1.5	TAB10	116.1 ± 1.8
TAB4	72.6 ± 1.6

2.2. Design, synthesis, and in vitro inhibitory assay of TAA derivatives

Encouraged by weak but apparent UCHL5 inhibitory activity of TA, we have designed derivatives of TA considering its molecular docking analysis. To expand potential interactions of TA around the active site of UCHL5, we added an aromatic ring (R₁) that is directly fused to the TA, resulting in TAA derivatives (TAA1 and TAA2) (Figure 2A and Scheme 1). TAA derivatives contain a carboxylic acid group in R₁ (red, Figure 2A), which was predicted to interact with the active site residues in Autodock modeling. For example, TAA2 was predicted to form hydrogen bonding interactions with H164 and D179 in a catalytic triad and Q82 in an oxyanion hole (Figure 2B and Supplementary Figure 2C). TAA1 and TAA2 were then synthesized in similar steps to synthesis of TA, as summarized in the Scheme 1. Boronic acid ester 1 was conjugated to aryl halide (2-3) by palladium-catalyzed Suzuki coupling to produce 4-5 in relatively good yields (55-70%). 4 and 5 were then converted to the corresponding acyl chloride and reacted with 6 via Friedel-craft acylation, which proceeded with low yields (10-20%). The resulting 7-8 were then hydrolyzed under a basic condition to produce TAA1 and TAA2.

Figure 2. Design of TA derivatives with molecular docking analysis.

(A) Design of TA derivatives (TAA and TAB series) with different R groups. (B) Autodock analysis of (R)-TAA2 with UCHL5. (C) Autodock analysis of TAB2 with UCHL5. The active site residues in magenta.

Scheme 1.

Synthesis of TAA derivatives. Reagents and conditions: (a) Pd(PPh₃)₄, K₂CO₃, MeOH/THF, 65 °C, 55–70% yields. (b) COCl₂, DCM, cat. DMF (c) AlCl₃, DCM, 10–20% yields over two steps. (d) LiOH, THF/MeOH/H₂O, 15–20% yields.

Next, TAA1 and TAA2 were evaluated by in vitro DUB assay. Encouragingly, TAA1 showed more significant inhibition of UCHL5 activity with the IC₅₀ value of 89 μM, which is ca. 33-fold improvement when compared to TA (Table 1), which suggests that the benzoic acid group in R₁ of TAA1 (red, Scheme 1) makes an additional favorable interaction with UCHL5, as predicted in our docking analysis (Supplementary Figure 2B). TAA2 with 3-furoic acid in R₁ (Scheme 1) showed slightly improved potency with the IC₅₀ value of 76 μM when compared to TAA1 (Table 1). Interestingly, an ester form of TAA1 (7), which is same as TAA1 but without hydrolysis of ortho-benzoate methyl ester in R₁, showed 2-3 fold reduced inhibitory potency (Supplementary Figure 3), supporting that the carboxylic acids in R₁ of TAA derivatives contribute to the improved potency.

2.3. Design, synthesis, and in vitro inhibitory assay of TAB derivatives

Next, throughout docking analysis, TAA derivative was further modified to include additional R₂ and R₃ groups while a thiophene ring of TA was replaced by R₄ group, such as carboxylic acid, which resulted in TAB derivatives (Figure 2A). For example, docking analysis of TAB2 with UCHL5 suggests that TAB2 has binding energy comparable to TAA2 (−8.1 and −6.6 kcal/mol for TAB2 and TAA2, respectively) and it is positioned to interact with H164 and D179 in the catalytic triad in a similar manner to TAA2 (Figure 2C and Supplementary Figure 2D). In addition, we envisioned that the R₂ group can be introduced in facile synthesis by click chemistry.

Various TAB derivatives were synthesized as shown in Scheme 2. Aryl bromide 9 was coupled with various aryl borates (R₁) (10-12) by Suzuki coupling. The resulting azide derivatives (13-15) were conjugated with R₂-containing alkynes via copper-mediated click chemistry to produce triazole derivatives (16-20). The methyl ester groups of 16-20 were then hydrolyzed to produce TAB1-TAB5. Additionally, the aniline groups in TAB1 and TAB4 were further acylated with R₃ groups to generate TAB6-TAB9. With the similar chemistry, TAB10 with R₄ group was synthesized.

Scheme 2.

Synthesis of TAB derivatives. Reagents and conditions: (a) Pd(PPh₃)₄, K₂CO₃. DMF, 52–83% yields. (b) 4-ethynylaniline, CuSO₄, Na-ascorbate, DMF, 45–80% yields. (c) LiOH·H₂O, THF/water, 44–71% yields. (d) R₂-ethyne, CuSO₄, Na-ascorbate, DMF, 51–80% yields. (e) R₃-Cl or (Ac)₂O, K₂CO₃, THF, 13–83% yields.

First, we synthesized TAB1 that has a terminal para-aniline group and evaluated it by in vitro DUB assay. TAB1 showed a comparable IC₅₀ value versus TAA2 (81 and 76 μM for TAB1 and TAA2, respectively, Table 1), supporting that TAB derivatives can inhibit UCHL5 with potency comparable to TAA derivatives. Next, TAB derivatives with different R₂ groups (TAB4 and TAB5 in Scheme 2) were synthesized and evaluated, which showed that TAB1 and TAB4 with terminal para- and meta-aniline groups at R₂ position, respectively, show comparable IC₅₀ values (81 and 73 μM for TAB1 and TAB4, Table 1). However, TAB5 containing terminal pyridine at R₂ position showed much weaker inhibitory potency (IC₅₀ = 252 μM), suggesting that the terminal aniline groups at R₂ in TAB1 and TAB4 are important for their inhibitory activity. To improve the inhibitory potency of TAB1 and TAB4, several R₃ groups were introduced on the aniline group of TAB1 and TAB4, producing TAB6-TAB9 (Scheme 2). TAB6 and TAB7, which are acetylated derivatives of TAB4 and TAB1, respectively, retained similar or improved potency when compared to their non-acetylated derivatives (IC₅₀ = 96 and 81 μM for TAB6 and TAB7 vs. 73 and 81 μM for TAB4 and TAB1, respectively, Table 1). Similarly, TAB9, a sulfamylated derivative of TAB1, did not significantly change the potency (IC₅₀ = 84 and 81 μM for TAB9 and TAB1, respectively). However, TAB8, which is a benzoyl derivative of TAB1, significantly lost its potency (IC₅₀ = 215 μM for TAB8), suggesting that a large size of modification at R₃ group leads to a loss of inhibitory activity. Interestingly, Autodock analysis suggests that the aniline group of TAB1 (IC₅₀ = 81 μM) would be positioned at a binding pocket where a only small size of R₃ group, such as acetyl group in TAB7 (IC₅₀ = 81 μM), may be tolerated to retain binding interactions, as opposed to a bulky size, such as benzoyl group in TAB8 (IC₅₀ = 215 μM) (Supplementary Figure 2E).

Next, to examine the importance of carboxylic acid position at R₁ of TAB derivatives, we synthesized TAB1-TAB3 that contains ortho-, meta-, or para-carboxylic acid in R₁ (Scheme 2) and evaluated them by in vitro DUB assay. Interestingly, TAB2 containing meta-carboxylic acid at R₁ showed ca. 2.5-fold improved potency (IC₅₀ = 33 and 81 μM for TAB2 and TAB1, respectively). In contrast, TAB3 with para-carboxylic acid at R₁ lost its potency dramatically (IC₅₀ = 285 μM for TAB3), suggesting that the position of carboxylic acid at R₁ is important for their potency. Autodock analysis of TAB1, TAB2, and TAB3 predicted their similar positioning at the active site, but with different orientations of the carboxylic acid group in R₁ (Supplementary Figure 4). Despite our docking analysis of TAB derivatives at the active site, an additional binding site was also predicted that may contribute to inhibitory activity of TAB derivatives (Supplementary Figure 5). In addition, the importance of carboxylic acid at R₄ was evaluated by synthesizing TAB10, which has the same structure as TAB1 but without carboxylic acid at R₄. TAB10 showed slightly reduced inhibitory potency (IC₅₀ = 116 and 81 μM for TAB10 and TAB1, respectively), suggesting that the carboxylic acid at R₄ also contributes to inhibitory potency.

Lipophilic ligand efficiency (LLE) estimates specificity of a compound to the target protein.⁴¹ Thus, we have calculated lipophilicity (cLogP) and LLE values for all TA derivatives (Supplementary Table 1). cLogP values were in the range of 2-4. LLE values were in the range of 0-2 (LLE for TAB2 = 0.87), suggesting that further improvement of specificity would be necessary.

2.4. Evaluation of TAB derivatives for inhibition of UCHL5 in complex with Rpn13 and other UCH family enzymes

UCHL5 is known to be activated upon binding to Rpn13, which apparently causes conformational change at the UCHL5 active site to accommodate ubiquitin substrates.⁴² Therefore, we examined the inhibitory potential of TAB inhibitors with an activated form of UCHL5 in complex of Rpn13. UCHL5 activity was increased about 2.5-fold with a saturating amount of Rpn13 (Figure 3A and Supplementary Figure 6), which shows apparent binding between UCHL5 and Rpn13 (Supplementary Figure 6). TAB compounds showed the inhibitory activity even with the UCHL5/Rpn13 complex (Figure 3A), among which TAB2 showed the highest inhibitory activity (~70% inhibition at 100 μM). The dose-dependent inhibition curves showed slightly reduced potency of TAB2 with UCHL5/Rpn13 (IC₅₀ = 52 μM) versus UCHL5 alone (IC₅₀ = 33 μM) (Figure 3B). These data support that TAB2 can inhibit the UCHL5/Rpn13 complex as well as UCHL5.

Figure 3. Evaluation of TAB derivatives for inhibition of UCHL5 in complex with Rpn13 as well as UCHL1 and UCHL3.

(A) Inhibition of UCHL5 activity in complex with Rpn13 by TAB derivatives. UCHL5 was included in all samples. TAB derivatives were used at a concentration of 100 μM. (B) Inhibition curves of UCHL5 activity by TAB2 with and without Rpn13. In all conditions, Rpn13 was used at a concentration of 0.92 μM. (C) Ub-VS conjugation competition assay with TAB2. Ub-VS (1 μM) was added to cell lysates (prepared from MDA-MB-231 cell line) after pre-incubation of lysates with TAB2. UCHL5 was analyzed by western blot. The blot is a representative of 2 independent experiments. (D) Inhibition curves of UCHL1 and UCHL3 activity by TAB2. In all enzyme assays, Ub-Rho (1 μM) was used as a substrate. Data represent the mean ± SD, n = 2 independent experiments.

To further demonstrate UCHL5 inhibition by TAB2, the ubiquitin vinyl sulfone (Ub-VS) assay was performed. Ub-VS is an irreversible inhibitor of DUB that binds to the active site of UCHL5 and shifts its size in gel analysis.⁴³ This shift can be inhibited and examined upon addition of UCHL5 inhibitors. We evaluated TAB2 by its incubation with cell lysate prepared from human breast cancer MDA-MB-231 cell line. The Ub-Vis conjugation followed by western blot analysis of UCHL5 showed that TAB2 inhibits UCHL5/Ub-VS conjugation at low concentrations (25-50 μM) (Figure 3C), which largely agrees with its in vitro inhibitory potency.

Lastly, TAB2 was evaluated for its inhibition of other UCH family enzymes, UCHL1 and UCHL3. In the DUB assays, TAB2 showed dose-dependent inhibition of both UCHL1 and UCHL3 (IC₅₀ = 92 and 74 μM, respectively) (Fig 3D). These data suggest that TAB2 has relatively limited selectivity for UCHL5 versus UCHL1 and UCHL3.

3. Conclusion

The proteasome and its associated DUBs are emerging as potential therapeutic targets in cancer.^44–46 UCHL5 is one of three DUBs on the proteasome that control the fate of poly-ubiquitinated proteins associated with cancer. Thus, the potent and selective inhibitors for UCHL5 would be important to understand UCHL5 in cancer biology. In this report, we have used the virtual screening to find TA as a weak inhibitor of UCHL5. A combination of docking analysis, synthesis, and DUB assay was then used to develop TAB2, which inhibits UCHL5 activity with and without Rpn13 in vitro. TAB2 could be further optimized to improve potency for UCHL5 inhibition in the future.

4. Experimental Section

4.1. Computational Studies

In-silico molecular modeling and virtual screening were performed with an FDA-approved drug library of 1,217 compounds from DSSTOX and ZINC database by using DOCK Blaster.^37–39 Specifically, each of the drugs were docked around the active site of UCHL5, which resulted in a list of candidates. Compounds were ranked by the number of contact points between each drug and UCHL5 as well as their binding free energies. Computational docking analysis was performed using AutoDock Vina 1.1.2 and AutoDock tools 1.5.6 software.^{40, 47} The receptor was used with UCHL5 (PDB: 3IHR).³⁶ The receptor file for docking assessment was prepared by removing any ligands, metal ions and solvent molecules using UCSF Chimera 1.13.1 software.⁴⁸ The Dock prep function was used to further remove any solvent molecule, add hydrogen atoms and assess charges. The inhibitors were drawn in ChemDraw and transformed to the 3D coordinates using Chem3D. After energy minimization, the inhibitor structure was converted to PDB file. In Autodock Vina, a grid box with a search space of 20 Å × 20 Å that centers on the active site of UCHL5 was used.

4.2. Chemical synthesis and compound characterization

All TA derivatives were synthesized according to the synthetic Scheme 1–2. Synthesis and characterization of compounds are described in Supplementary Information.

4.3. Purification of UCHL5 and RPN13

Plasmid pET151-hUCH37 (Addgene, plasmid ID 61929) was transformed to BL21(DE3) codon+ (RIL) E.coli cells (Agilent, #230245) in LB medium. A starter culture of 10 mL was inoculated in 1 L LB medium and cells were grown at 37°C until optical density at 600 nm (OD₆₀₀) reached 0.6 and then induced with 1 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG). After incubation overnight at 18°C, cells were harvested by centrifugation at 5,000 rpm. The pellet was resuspended in a lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10 mM imidazole, 10% glycerol, 1% Triton X-100, 2 mM DTT and 1 mM PMSF). Cells were lysed by double passing through French press at 1,000 psi. Cell debris were removed by centrifugation at 14,000 rpm for 30 min. The clarified lysate was incubated with Ni-NTA resin (QIAGEN) for 2 h. The beads were washed three times with a wash buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 20 mM imidazole, 10% glycerol, 2 mM DTT and 1 mM PMSF). Protein was eluted via elution buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 250 mM imidazole, 10% glycerol, 2 mM DTT). Eluted fractions were collected and dialyzed with a buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10% glycerol, 2 mM DTT) overmght and concentrated on a centrifugal filter device (Millipore). Protein concentration was determined by Bradford assay. For purification of Rpn13, plasmid pET151-RPN13 (Addgene plasmid ID 73741) was transformed to BL21 (DE3) E. coli cells in LB medium. Rpn13 was expressed and purified by Ni-NTA resin in the same manner to UCHL5.

4.4. Enzyme assay

The DUB activity was measured with ubiquitin-rhodamine (Ub-Rho) (SouthBay Bio) by detecting fluorescence with the excitation (485 nm) and emission (535 nm) wavelengths for 30 min at RT. The reaction contains 50 mM Tris HC1, pH 7.5, 150 mM NaCl, 0.464 nM UCHL5, and 1 μM Ub-Rho, 0.1 mg/ml ovalbumin, 0.5 mM EDTA, 2.5 mM DTT. Reaction was performed in a reaction volume of 50 μL in a 96-well black plate and was initiated by addition of UCHL5. The data were plotted with the normalized rates (y-axis) versus concentrations (x-axis) to determine the IC₅₀ values using GraphPad. To measure UCHL5 activity in the presence of Rpn13, a series of RPN13 (0-1 μM) were titrated with UCHL5 (0.464 nM). To evaluate TA derivatives with a complex of UCHL5 and Rpn13, UCHL5 (0.464 nM) and RPN13 (915 nM) were used in the presence of individual TA derivative (100 μM). Enzyme assays for UCHL1 and UCHL3 (Boston Biochem) were performed in the same conditions as UCHL5, except using different enzyme concentrations, UCHL1 (1.2 nM) and UCHL3 (32 pM).

4.5. Ub-Vis assay

MDA-MB-231 cells grown at 80-90% confluency in a 10 cm dish were washed with cold PBS once and lysed in a lysis buffer (50 mM Tris, pH 7.5, 250 mM Sucrose, 5 mM MgCl₂, 2 mM ATP, 1% NP-40, 1 mM PMSF). Protein concentrations in lysates were determined by Bradford assay. Lysates (20 μg of proteins) were pre-incubated with TAB2 in a reaction buffer (30 μL, 50 mM Tris, pH 7.5, 250 mM Sucrose, 5 mM MgCl₂, 2 mM ATP) for 30 mm at RT. Ub-VS (1 μM) (South Bay Bio) was added to the reaction, which was incubated for 11 min at RT. The reaction was quenched by adding an SDS-loading buffer. Proteins were resolved by SDS-PAGE (12% gel) and probed by western blot with anti-UCHL5 antibody (Santa Cruz biotechnology, sc-271002, 1:1000 dilution) and β-actin (Cell signaling, #3700, 1:1000 dilution).