Chondroitin 4-Sulfate Disaccharide-Based Inhibitors of Cathepsin S

Abstract

Cathepsin S (Cat S) is a cysteine protease involved in several human diseases (i.e., autoimmune, inflammatory and cardiovascular disorders, cancer, and psoriasis) and is an important target in drug development. Emerging evidence highlights the potential of inhibiting Cat S by glycosaminoglycans, particularly chondroitin sulfates (CSs), as a promising therapeutic strategy. Given the limited and heterogeneous GAG materials from animal sources, a series of synthetic biotinylated non- or sulfated chondroitin oligomers were synthesized and assessed for their ability to inhibit Cat S. The biotinylated disaccharide C4S displayed in vitro potent inhibitory activity toward Cat S with IC₅₀ value in the micromolar range and showed selectivity over cathepsins K and L. Molecular modeling studies suggested that only C4S dp2 but not C6S, C4,6S or non-sulfated chondroitin binds selectively to the active site of Cat S. In addition, a synthetic multivalent C4S dp2 glycosylated BSA was shown to be more efficient towards Cat S inhibition (nanomolar range) than the monovalent parent C4S dp2. Our findings also indicated that this new neoglycoconjugate displayed selectivity for Cat S vs. cysteine cathepsins expressed by differentiated THP-1 cells. This study reports a new approach for designing selective and potent inhibitors of Cat S using multivalent C4S derivatives as a molecular scaffold.

1. Introduction

Cathepsin S (Cat S, EC 3.4.22.27) is a papain-like cysteine protease member of the lysosomal cathepsin family (B, C, F, H, K, L, O, S, V, W, and X/Z), which is implicated in physiological and pathophysiological processes, including autoimmune disorders, atherosclerosis, inflammation, nervous system diseases, psoriasis, and cancers (for review [1]). These findings underscore the critical role of Cat S in regulating inflammatory responses and immune modulation. Especially, in vitro and in vivo studies reported that the overexpression of active Cat S contributes to the etiology of the aforenamed diseases, particularly by its ability to degrade efficiently a variety of extracellular matrix (ECM) and basement membrane (BM) components, including elastin, fibronectin, collagen IV and nidogen, which are crucial for tissue homeostasis. For many years, Cat S has been widely studied for its clinical significance and remains a promising therapeutic target for the pharmaceutical industry and academic laboratories seeking to develop selective and specific synthetic inhibitors [2,3].

Chondroitin sulfates (CSs) belong to the family of glycosaminoglycans (GAGs) and are natively presented as long linear negatively charged oligosaccharides, composed of repeating β-D-glucuronic acid (GlcA) and N-acetyl-β-D-galactosamine (GalNAc) units linked by GlcA-β(1→3)-GalNAc-β(1→4) glycosidic bonds, with variable high and low sulfation patterns. It was concomitantly found that beside their disaccharide composition, degree and patterns of sulfation, the size of CS is also critical for their functional properties, particularly in terms of binding affinity and selectivity [4]. CS play pivotal roles in many biological processes such as cell division, neuronal development, viral invasion, and cancer metastasis, and thus constitute an interesting therapeutic option for several pathologies [5]. We previously reported a potent and selective inhibition of Cat S by chondroitin 4 sulfated (C4S) among GAGs of animal origin tested (heparin, heparan sulfate, C6S, dermatan sulfate, and hyaluronic acid) [6]. Nevertheless, current sources of CS from extracts of tissues of animal origin do not guarantee sufficient homogeneity and purity to fully reproduce their biological activities or for therapeutic application. An alternative is the development of synthetic well-defined modified short CS oligomers, having the same characteristics as their counterpart of animal origin. To facilitate direct qualitative and quantitative GAG-binding proteins analysis, several methods to label GAGs were used, including ³H, ¹²⁵I, ³⁵S, and fluorophores that were coupled directly to the GAGs [7,8]. Interestingly, biotinylated GAG have been developed to study protein–carbohydrate interactions and were reported to retain most of their biological activity but also their binding affinity [9,10,11]. On the other hand, multivalent binding is also an effective and widely employed method of attaining high affinity between GAG and proteins [12]. However, to our knowledge, none have paid attention to developing Cat S inhibitors using short CS oligomers and multivalent CS.

In the present study, we extend our previous investigations on the inhibitory potency of C4S on Cat S [6,13]. Since 4-sulfation pattern (C4S) is important for the inhibition of Cat S and that short C4S oligosaccharides are efficient to inhibit Cat S, we hypothesized that chemical analogs of C4S (e.g., biotinylation and multivalency) may strengthen its inhibitory functionality. To study the role of the sulfate, we evaluated the potency of desulfated chondroitin (obtained after desulfation of C4S) to inhibit Cat S. Then, we examined the inhibitory potency of short C4S oligomers produced by methanolysis of mammalian C4S. In addition, we evaluated the inhibition of Cat S by synthetic biotinylated 4-sulfated and non-sulfated chondroitin oligosaccharides of different sizes (dp2-dp4). Cat S-C4S dp2 interactions were further scrutinized and characterized by molecular modeling studies. Additionally, a multivalent C4S dp2 analog showed a higher inhibitory efficiency than the monovalent parent C4S dp2 and was selective towards Cat S. It is anticipated that future studies of such multivalent system will be beneficial for the design of potent and selective C4S-based inhibitors of Cat S.

2. Materials and Methods

2.1. Enzymes, Substrates, and Inhibitors

All the chemicals were of analytical grade. The expression of the recombinant Cat S in Escherichia coli, extraction, as well as renaturation, purification, and processing into active form were carried out as previously described [14]. Recombinant human cathepsin K (Cat K) was produced in Pichia pastoris as reported previously [15]. Human cathepsin L (Cat L) purified from liver, were supplied by Calbiochem (VWR International S.A.S., Rosny–sous–Bois, France). Their active site titration was determined using L-3-carboxy-trans-2, 3-epoxy-propionylleucylamide-(4-guanido)-butane (E-64) (Sigma-Aldrich, Saint-Quentin Fallavier, France). Activity assays were performed in the following buffer: 100 mM sodium acetate buffer, pH 5.5, 2 mM dithiothreitol (DTT), and 0.01% Brij35. Enzymatic assays were carried out at 37 °C in the activating buffer, using peptidyl-AMC substrate, Z-FR-AMC (benzyloxycarbonyl-Phe-Arg-7-amino-4-methyl coumarin, R&D System Europe, Abingdon, UK) (λ_ex = 350 nm, λ_em = 460 nm). DQ-elastin (λ_ex = 505 nm, λ_em = 515 nm) was purchased from Molecular Probes (Life Technologies, Saint Aubin, France). Fluorogenic assays were monitored using a 96-well microplate reader spectrofluorometer (SpectraMax Gemini, Molecular Devices, Saint Gregoire, France). Chondroitin sulfate C4S from bovine trachea (C4S_T) and cartilage (C4S_C) (Mw: 20–30 kDa) were obtained from Sigma-Aldrich. C4S was diluted (3%, weight/volume, w/v) in sodium acetate buffer 0.1 M, pH 5,5. Morpholinurea-leucine-homophenylalanine-vinyl sulfone phenyl (LHVS) was a kind gift from Dr. J. H. McKerrow (University of California, San Francisco, CA, USA) and is available commercially (MedChemExpress, Princetown, NJ, USA).

2.2. Synthetic Chondroitin Oligomers

The synthesis of chondroitin oligomers and biotinylated chondroitin oligomers have been described elsewhere [16,17,18]. Briefly, the oligomers of well-defined size and sulfation patterns have been synthesized following a straightforward strategy starting from a key disaccharide building block. The carboxylic acid function of D-GlcA was protected by a methyl ester (CO₂Me) and the amine function of D-GalNAc was protected by a trichloroacetyl group (NHTCA). Hydroxyls of future sulfated position (position 4 and/or 6 of the GalNAc) were protected by chloroacetyl groups. Hydroxyl at position 4 of the GlcA was protected with levulinoyl group thus allowing the elongation at the non-reducing end. Activation of the anomeric position by a trichloroacetimidate function allowed the elongation and the introduction of the aminoethyl chain protected by a benzyloxycarbonyl (CBz) group which served as a primer for the spacer arm linked to a biotin. All other hydroxyl groups were protected by ester (benzoyl or acetyl) groups and were deprotected at the end of the synthesis. After sequential glycosylation steps, selective deprotection of ClAc, sulfation and full deprotection of ester and CBz, well-defined chondroitin derivatives were biotinylated on the nitrogen of the primer via an amidohexanoyl spacer. All intermediates have been isolated, purified by flash-silica gel column chromatography and have been fully characterized by ¹H and ¹³C NMR, HRMS, and elemental analysis.

2.3. Synthesis of Neoglycoprotein Chondroitin 4 Sulfate (NeoC4S dp2)

A solution of 3-(2-Propyn-1-yloxy)propanoic acid 2,5-dioxo-1-pyrrolidinyl ester (2 mg, 60 eq.) was added in a solution of bovine serum albumin (BSA, 10 mg) in PBS and the reaction mixture was stirred for 5 h at room temperature. BSA-alkyne was purified on a Sephadex G25. All chemical reagents were purchased from Sigma-Aldrich (Saint Quentin Fallavier, France). The purity of functionalized BSA alkyne was controlled by SDS-PAGE and the number of alkynes group by BSA was determined by MALDI-TOF (UltrafleXtreme, Bruker, Vienna, Austria). Azidopropyl C4S dp2 was synthesized following a method previously reported by the group for the synthesis of the non-sulfated analog [19] (unpublished data, see Supplementary Materials for ¹H and ¹³C NMR spectra, Figure S1).

The conjugation of BSA-alkyne with azidopropyl C4S dp2 using copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) was performed as followed: TBTA (5 mg/mL), L-ascorbic acid (3 mg/mL), CuSO₄.5H₂O (3 mg/mL) and C4S dp2 propyl azide (10 mg/mL, 50 eq.) was added to a solution of BSA-alkyne (2 mg/mL) in PBS. The solution was stirred 24 h at room temperature then the crude mixture was purified by size-exclusion chromatography on a Sephadex G25 gel. The purity of the neoglycoprotein was controlled by SDS-PAGE (Figure S2) and the ratio of C4S dp2 by BSA determined by MALDI-TOF was calculated as 31:1.

2.4. Desulfation of Mammalian Chondroitin 4 Sulfate

Bovine C4S_T and C4S_C were chemically desulfated according to the protocol adapted from [20]. Briefly, C4S_T was diluted in a 0.5% methanol solution of acetyl chloride at 5 mg/mL (the solution was prepared at least 48 h in advance for complete methanolysis). Sample was incubated at room temperature for 24 h. The solution was then centrifuged at 3000× g for 5 min at 25 °C. The pellet was retained and mixed again in a 0.5% methanol solution of acetyl chloride and stirred at room temperature for 24 h. This procedure was repeated over 7 days. After centrifugation at 3000× g for 10 min at 25 °C, the pellet was solubilized in 5% water then precipitated with absolute ethanol and then centrifuged again at 3000× g for 10 min at 25 °C. The pellet was then washed with ethanol and ethyl ether (v/v). The solution was then dried in an extractor hood until complete evaporation (3 days). The sample obtained was weighed and mixed in a 0.1 M potassium hydroxide (KOH) solution to obtain a concentration of 25 mg/L and stirred slowly at room temperature for 24 h. The KOH was neutralized with 1 M potassium acetate supplemented with 10% acetic acid (4 mL of solution per gram of starting product). The mixture was then centrifuged for 10 min at 3000× g, the pellet was taken up, washed in absolute ethanol and ethyl ether (v/v) and dried in an extractor hood. The powder was then stored at −20 °C. GAG desulfation was monitored using the dimethylmethylene blue (DMMB) colorimetric assay (Sigma-Aldrich), which binds only to the sulfate groups of GAGs [21]. Different volumes of GAG diluted in water (1%: w/v) were placed in a transparent 96-well plate to which a DMMB solution is added. Absorbance was measured at 530 nm using a microplate spectrophotometer (VersaMax Plus ROM). Data were processed with Softmax pro 7.2 software.

2.5. Acid Methanolysis of Mammalian Chondroitin 4-Sulfate

C4S_T (1 mg) was mixed with 1 mL of an anhydrous acidic methanolic solution (3 N HCl with 50 µL of 2,2-dimethoxypropane to reduce any side reactions with water during methanolysis) and heated for 90 min in a water bath (65 °C) [22]. The solution was then neutralized with 3 M NaOH and evaporated by freeze-drying. The powder was diluted in 500 µL of ultrafiltered water and then evaporated again by freeze-drying. The sample was solubilized in 0.1 M sodium acetate buffer, pH 5.5. The absorbance spectrum (200 to 500 nm) of the sample obtained after methanolysis was measured at different times (0–240 min) using a Varian Cary 100 Scan spectrophotometer and Cary Win UV Scan software 4.2 (Agilent Technologies, Courtaboeuf, France). In parallel, the sample was analyzed by SDS-PAGE 15% [23]. Sample was diluted in TBE buffer (0.1 M Tris-borate, pH 8.3 1 mM EDTA), with 2 M sucrose (GAG: sucrose ratio of 1:1) and bromophenol blue (0.02%). Electrophoresis was carried out for 45 min at 20 mA with cold TBE solution as running buffer. After electrophoresis, the gel was stained for 45 min with 0.5% alcian blue solution (alcian blue 8G, Sigma-Aldrich) with 2% acetic acid. The gel was then destained for 20 min with 2% acetic acid solution, then with PBS solution.

2.6. Inhibition of Cat S by Mammalian and Synthetic Biotinylated CS and NeoC4S dp2

Cat S (1 nM) was incubated in 96-well plates in the activity buffer: 0.1 M sodium acetate, pH 5.5, 0.01% Brij35, 2 mM DTT for 10 min at 37 °C in the presence of increasing concentrations of untreated bovine C4S_T and desulfated C4S_T (C0S_T) (0.001–0.5%, w/v). The Z-FR-AMC substrate (20 µM) was then added to measure enzyme activity, with stirring at 5-sec intervals for 15 min at 37 °C. The slope values measured by SoftMax Pro 5.3 software were normalized to the activity of Cat S alone (control). The activity was expressed as a percentage and compared to the control without the inhibitor (100% activity). Data analysis is performed by generating a dose–response curve, and inhibition is calculated by determining the median effective concentration (IC₅₀) at the inflection point (GraphPad Prism 9, Irvine, CA, USA). The same method was used for synthetic biot-C4S dp2 and NeoC4S dp2. The enzymatic activity of Cat S (1 nM) was also measured in the activity buffer in the presence of C4S_T submitted to acid methanolysis (0.15%, w/v), using Z-FR-AMC as substrate (20 µM). Similar assays were performed with biotin-labeled C4S, C6S, C4,6S, non-sulfated chondroitin (C0S) with different size (dp2, dp3, dp4) at 100 µM. Elastinolytic activity of Cat S (1 nM) was measured with DQ-elastin (4 µg/mL, λ_ex = 495 nm, λ_em = 515 nm) in the absence and presence of biot-C4S dp2 (100 µM).

2.7. Cell Culture

Wild-type (WT) and Cat S-knockout (CTSS^−/−) THP-1 monocytes were obtained from Abcam (Cambridge, UK) and cultured in RPMI 1640 medium (Gibco A1049101) supplemented with 10% heat-inactivated fetal bovine serum, 1% penicillin/streptomycin, and 50 µM of β-mercaptoethanol, as described before [24,25]. Briefly, cells were incubated at 37 °C in a water-saturated atmosphere containing 5% CO₂. THP-1 cells were seeded in 6-well plates (2 million cells/well) and differentiated into macrophages by adding phorbol 12-myristate 13-acetate (162 nM) before being weaned (day 2) with RPMI 1640 medium without growth factors or antibiotics. At day 6, macrophages-free culture supernatants were collected, supplemented with 0.1 M sodium acetate buffer, pH 5.5, containing 1 mM S-methyl methanethiosulfonate (MMTS, Sigma-Aldrich) and concentrated approximately thirty times (Vivaspin 20, 10 kDa resolution, PES, Sigma-Aldrich). Protein concentration was determined by the Bradford method (Bio-Rad, Marnes-la-Coquette, France) by measuring the absorbance with a spectrophotometer at 595 nm.

2.8. Selectivity Inhibition of Cat S by Biot-C4S dp2 and NeoC4S dp2

Inhibition of the peptidase activity of Cat L, Cat K, and Cat S (1 nM) with biot-C4S dp2 (100 µM) was measured in the activity buffer with Z-FR-AMC (20 µM) as a substrate. Similar assays were performed with NeoC4S dp2 (1 µM). The activity was expressed as a percentage and compared to the control (enzyme) without biot-C4S dp2 or NeoC4S dp2 (100% activity). The activity of cysteine cathepsins in THP-1 supernatants (0.5 µg of total protein) was measured in 0.1 M sodium acetate buffer, pH 5.5, 10 mM DTT, 0.01% Brij-35 using Z-FR-AMC (20 µM), in the presence and absence of NeoC4S dp2 (1 µM), LHVS (5 nM) and E-64 (100 µM).

2.9. Molecular Modeling

For molecular docking, the structure of Cat S catalytic domain (PDB: 2H7J) and the four GAG oligosaccharides (C4S, C6S, C4,6S and C0S) were used as receptor and ligands, respectively. The unbound structure of Cat S was obtained by removing the inhibitor from the complexed structure and by subsequent minimization of the binding site using the AMBER 99 force field as implemented in MOE (Molecular Operating Environment, Chemical Computing Group, Montreal, QC, Canada), as reported elsewhere [6]. The flexible docking of GAG oligosaccharides to Cat S was performed with Autodock 3 (AD3) and further clustered as described before [26]. 100 ns of molecular dynamics (MD) simulations of the representative Cat S/GAG complexes obtained by molecular docking were performed in AMBER 20 [27] and binding free energy calculations were then performed using protocol as reported previously [26].

2.10. Statistical Analysis

Data were expressed as median ± interquartile range and analyzed statistically with a non-parametric Mann–Whitney test. Analyzes were performed with GraphPad Prism software 6.0 (GraphPad Software Inc., San Diego, CA, USA). Differences at a p value < 0.05 were considered significant.

3. Results

3.1. Importance of Sulfation of C4S for Cat S Inhibition

To assess the importance of the sulfate group of C4S in the inhibition of Cat S, bovine tracheal C4S (C4S_T) was desulfated by chemical treatment, according to the described method [20]. The desulfation of C4S_T was monitored using a colorimetric test using dimethylmethylene blue (DMMB). This cationic dye binds to the sulfate group, causing a metachromatic reaction that shifts the absorbance of the complex to 530 nm [21]. The absorbance increases proportionally with the mass concentration of unmodified C4S_T (Figure 1A). Acidic methanol treatment of C4S_T after 7 days, resulted in completed desulfation of C4S_T (C0S_T), as indicated by the slope close to horizontal line (x-axis) in the desulfated chondroitin curve measured by the DMMB assay.

Subsequently, the inhibitory potential of C0S_T (0.1%) was measured against Cat S and compared to its sulfated counterpart. Our results indicated that unmodified C4S_T strongly inhibited the peptidase activity of Cat S (64% apparent inhibition) toward Z-FR-AMC, as previously reported [6]. Desulfation of C4S_T resulted in a significant decrease in Cat S inhibition by a factor of ~1.6. The median inhibitory concentrations (IC₅₀) of C4S_T and C0S_T were then determined. The dose–response curve for C4S_T yielded an apparent IC₅₀ of 0.60 ± 0.08 g/L. Considering the average molecular mass of C4S_T (~25 kDa) according to the supplier, the IC₅₀ was estimated to be 24.0 ± 3.4 µM. Previous work in our laboratory reported that bovine C4S_T inhibited Cat S with the inhibition constant (Ki) being 16.5 ± 6 µM [6]. Although the IC₅₀ is not a direct indicator of affinity, it can be converted into Ki according to [28]. The Ki of Cat S by C4S_T was then estimated at 12 ± 1.7 µM (with S: Z-FR-AMC = 20 µM and the Km = 18.2 µM), which corroborates this earlier finding [6]. Desulfation of C4S_T reduced its inhibitory potential against Cat S by a factor of ~3 (IC₅₀ = 68.4 ± 4.7 µM), supporting that sulfation is important in the interaction with the enzyme. Similar results were obtained with another commercial source of C4S from bovine cartilage (C4S_C) (IC₅₀ = 1.05 ± 0.05 g/L, corresponding to a Ki of 21 ± 2 µM) (Figure S3).

3.2. Efficacy of Acid Catalyzed Methanolysis of C4S_T on Cat S Inhibition

To assess the importance of the minimal size-fragments of C4S_T required to inhibit Cat S, acid catalyzed methanolysis method [22] was performed to obtain disaccharide derivatives. Complete cleavage of C4S_T into disaccharide units was followed by spectral analysis (Figure 2A). The appearance of an absorbance peak at 232 nm over time corresponds to the presence of an unsaturated Δ4,5 double bond of GlcA (chromophore) at the end of the newly formed disaccharides [29,30]. Absorbance spectra at t = 120 min revealed no difference with that obtained at 90 min, which indicates complete hydrolysis of C4S_T into disaccharides after 90 min of incubation. In addition, complete hydrolysis of C4S_T was confirmed by polyacrylamide gel electrophoresis (Figure S4). Once C4S_T sample was hydrolyzed into disaccharides, their ability to inhibit Cat S was tested (Figure 2B). To a lesser extent than the parent C4S_T (0.15%, w/v), the resulting C4S disaccharides (C4S dp2) from C4S_T after methanolysis (C4S_m) significantly inhibited Cat S, suggesting that sulfated disaccharides are sufficient to inhibit the enzyme.

Nevertheless, the heterogeneity of animal-extracted C4S, which can vary from batch to batch, in terms of molecular mass, oligosaccharide sequence, or variable degree of sulfation, makes it impossible to accurately characterize the structural elements of bioactive CS sequences required to inhibit Cat S. In this context, we tested two synthetic C4S oligosaccharides of well-defined structure (dp2 and dp4, Supplementary Materials Figures S1 and S4) that were prepared as previously reported [16]. When performing a dose–response curve of both synthetic C4S oligosaccharides (100–600 µM), only the C4S dp2 caused 20% Cat S inhibition at 200 µM (p = 0.0159), while no inhibition was measured for C4S dp4 (p > 0.5) (Figure 2C,D). In our assay conditions, we could not evaluate IC₅₀ of C4S dp2 due to limited inhibition at highest concentration tested (600 µM), supporting its weak potency to inhibit Cat S.

3.3. Inhibition of Cat S by Synthetic Biotinylated Chondroitin Oligomers

While studying the specific inhibition of Cat S by C4S oligosaccharides, assays were carried out with a library of thirteen synthetic biotin-labeled C4S, C6S, C4,6S (biot-C4S, biot-C6S, biot-C4,6S) and non-sulfated CS (biot-C0S) derivatives of different sizes [17,18] (Figure S5). In the oligosaccharide series, biotinylated C4S dp2 (biot-C4S dp2) was found to display potent inhibition of 50% at 100 µM and inhibited better than its C6S counterpart (p < 0.01) (Figure 3A,B).

We previously reported that mammalian C4S and to a lesser extent C6-S inhibit Cat S activity, emphasizing that the sulfate at the C4 position of GalNAc is critical for inhibition [6]. The extension of fragment size (dp3 and dp4) to the C4S disaccharide scaffold (dp2) substantially led to a decreased inhibition of Cat S. In addition, sulfation exclusively on the first GalNAc unit of C4S dp3_H and dp4_H reduced significantly the inhibition of Cat S compared to the sulfation of all GalNAc residues in dp3 and dp4 structures, respectively. Furthermore, the results showed that the disulfated C4,6S (biot-C4,6S) derivatives have little effect on inhibition, suggesting that disulfated disaccharide sequence is unfavorable regardless of the size of the oligosaccharide tested (Figure 3C).

Moreover, 0. reduction in Cat S inhibition compared to biot-C4S dp2, corroborating the difference observed between desulfated C4S_T (C0S_T) results and its native form (Figure 3D). Moreover, biot-C4S dp2 significantly inhibited the elastolytic activity of Cat S, using the fluorogenic substrate DQ-elastin (Figure S6A). Subsequently, the inhibition of Cat S by biot-C4S dp2 was evaluated over a concentration range and IC₅₀ was found to be 149 ± 1 µM (Figure S6B). Compared to the IC₅₀ obtained for bovine C4S_T, the inhibitory potential of biot-C4S dp2 was reduced by a factor of ~6, supporting the difference in inhibition observed between disaccharide derivatives from decomposed C4S_T and untreated C4S_T (reduction by a factor of ~3). Noteworthy, the biotin molecule had no effect on the fluorescence spectrum of AMC (Figure S7A) and did not modulate Cat S peptidase activity, even at high concentrations (500 µM) (Figure S7B). The biological significance of the impact of biotin with the spacer for the inhibition of Cat S is not yet understood, though it could interact on the surface of Cat S, stabilizing chondroitin oligosaccharides, particularly C4S dp2 in the active site of Cat S and thus enhances its inhibitory potential.

3.4. Selective Inhibition of Cat S by Biot-C4S dp2

Cat S displays a high similarity with Cat K and Cat L in terms of amino acid sequences (58% and 55% of identity, respectively) and folding making it difficult to develop potent and selective inhibitors. Interestingly, we did not observe any inhibitory effect of biot-C4S dp2 (100 µM) against Cat K and Cat L (Figure 4).

3.5. Molecular Modeling Studies of the Interactions Between Cat S and C4S dp2

To better understand the differences in Cat S inhibition with the CS derivatives at the molecular level, we applied molecular docking approach followed by molecular dynamic (MD) simulations to Cat S with C4S, C6S, C4,6S, and non-sulfated chondroitin (C0S) dp2 (Figure 5).

According to MD studies, only C4S dp2 (Figure 5A) bound primarily in the active site of Cat S (ΔG = −18.5 ± 4.7 kcal/mol) by interacting with residues G23, C25, N67, and R141 via hydrogen bonds (Figure 5E). Residues Q19 and N163 also appeared to be important for the interaction with the disaccharide. Conversely, C6S, C4,6S, and C0S dp2 interacted only outside the active site (Figure 5B–D). These results partially corroborate the experimental inhibition results obtained with biotinylated C4S, C6S, C4,6S, and C0S derivatives.

3.6. Inhibition of Cat S by NeoC4S dp2

In addition to the biotinylated derivatives, the conjugation of C4S dp2 to bovine serum albumin (BSA) scaffold has been investigated for the inhibition of Cat S. Herein, a multivalent C4S dp2-linked BSA neoglycoconjugate (henceforth NeoC4S dp2) was synthesized and evaluated as an inhibitor of Cat S. The multivalent conjugate was prepared by the ligation of thirty-one C4S dp2 fragments to an alkyne-modified BSA, thus leading to the loading of 31 disaccharides per BSA with a triazole linker (Figure 6A and Figure S2A,B). A detailed biophysical characterization of NeoC4S dp2 will be the object of future research. While IC₅₀ for biotinylated C4S dp2 was in the micromolar range, the multivalent NeoC4S dp2 displayed nanomolar Cat S inhibitory activity with a ~1000-fold enhancement of the potency (IC₅₀ = 155 ± 64 nM) (Figure 6B). Of note, SDS-PAGE analysis revealed that NeoC4S dp2 and BSA were both relatively resistant to degradation after 120 min of incubation with Cat S (Figure S8A,B). Likewise, control experiments with Cat S incubated either with BSA or BSA-alkyne revealed that none of the compounds led to a similar inhibition potential as compared to NeoC4S dp2, supporting that C4S is an important structural requirement for high Cat S inhibition (Figure S8C). In addition, NeoC4S dp2 was selective for the inhibition of Cat S vs. Cat K and Cat L. Surprisingly, the endopeptidase activity of both Cat K and Cat L was substantially enhanced in the presence of NeoC4S dp2 (Figure 6C), suggesting that negatively charged surface of NeoC4S dp2 induces a conformational change in the active site of both enzymes that affect positively the enzyme activity.

The potent and selective Cat S inhibitory activity of NeoC4S dp2 encouraged us to further evaluate the compound in THP-1 cell culture expressing several cysteine cathepsins [24,25], including Cat S (wild-type, WT) and in Cat S deficient THP-1 (CTSS^−/−) (Figure 6D). In this regard, peptidase activity of overall cysteine cathepsins was performed using Z-Phe-Arg-AMC, as standard substrate for cysteine protease. Both WT and CTSS^−/− supernatants (0.5 µg of total protein/assay) hydrolyzed Z-FR-AMC. Proteolytic activity was significantly higher in WT than in CTSS^−/− samples and was fully abolished by the addition of E-64, indicating that hydrolysis of Z-FR-AMC related specifically to cysteine cathepsins. NeoC4S dp2 significantly reduced the peptidase activity of cysteine proteases in WT THP-1 supernatants, contrary to Cat S deficient THP-1. Moreover, a non-significant difference (p = 0.1) was observed between NeoC4S dp2-treated WT supernatants and WT supernatants treated with LHVS, a selective inhibitor of Cat S [31], supporting that NeoC4S dp2 is a potent and selective inhibitor of Cat S.

4. Discussion

The outcomes of this in vitro pilot study expand earlier results of the impact of GAGs on Cat S activity [6]. Indeed, we reported that across mammalian-extracted GAGs (heparin, heparan sulfate, chondroitin 4/6 sulfate, dermatan sulfate and hyaluronic acid) tested, only bovine tracheal C4S (C4S_T, ~25 kDa) reduced substantially in a dose dependent manner the activity of Cat S and to a lesser extent cathepsins B, K, and L [6,13]. Unfortunately, C4S from animal origin and commercial sources shows significant constraints for therapeutic applications due to its heterogeneity of both purity (poor batch reproducibility) and structure [32,33]. For instance, commercial source of C4S from bovine tissue contains ~60% C4S and ~30% C6S (Sigma-Aldrich). Thereby, engineering synthetic or semi-synthetic polysaccharides may solve these problems and help to develop new potent and selective Cat S inhibitors.

As such, our results herein showed for the first time that the C4S disaccharide unit obtained after methanolysis of C4S_T slightly reduced the activity of Cat S, as confirmed by the chemically synthetized C4S dp2. Fortuitously, among a library of thirteen synthetic high purity chondroitin derivatives functionalized by a biotinylated linker and having different lengths or sulfation patterns, with homogeneous and heterogeneous sulfation [17,18], we found that biot-C4S dp2 was more efficient to inhibit Cat S compared to its unlabeled counterpart. Addition of biotin with the linker potentially improved the inhibition of C4S dp2. The chemical polar structure of biotin consists of a tetrahydroimidizalone ring fused with a tetrahydrothiophene ring. It is likely that there are important bonds (i.e., hydrogen bonds or hydrophobic interactions) that favor biot-C4S dp2 interactions with the highly complementary contours of the binding active site of Cat S. In-depth analysis of biot-C4S dp2/Cat S interactions will be the matter of subsequent studies. Independent of sulfation, our results indicated that the disaccharide unit GlcA-GalNAc (biot-C0S dp2) was sufficient to inhibit Cat S. Elongation of the disaccharide unit (biot-C0S dp3 and dp4) had no effect on the potency but surprisingly altered the inhibition for biot-C4S and biot-C6S (dp3 and dp4). This finding that focused on the active site binding may contrast somehow with a previous in silico study that we performed on cathepsins-GAG interactions, using the global sequence information of the enzyme [13], in which the stability of the Cat S/C4S and Cat S/C6S complexes, through computational binding free energy analysis, increased with the chain length of C4S and C6S, respectively. Nevertheless, a possible explanation relies on the fact that in terms of binding regions, we predicted several clusters of C4S or C6S, located either outside or inside the substrate binding cleft of Cat S. Herein, we focused on the inhibition of the protease by targeting its active site. Our data also showed that substituting a sulfate specifically at the C4 of GalNAc (biot-C4S dp2) reduced significantly the activity of Cat S compared to biot-C6S dp2, supporting our previous investigations on mammalian GAGs [6]. Moreover, the sulfation degree is also a determining factor for the inhibition of Cat S, as illustrated by the low inhibitory potency of biot-C4,6S dp2 in comparison to biot-C4S dp2. In addition, the importance of sulfate position on C4 and sulfation degree in Cat S inhibition was consolidated by MD simulations, demonstrating that despite the complex nature of protein-GAG interactions, molecular modeling tools offer useful and reliable approaches for studying biomolecular systems. The identification of a highly selective inhibitor of Cat S with great potency is of paramount importance. Notably, biot-C4S dp2 was endowed with significant selectivity for Cat S, since it was inactive against related cathepsins K and L. The amino acid residues that constitute the catalytic triad for Cat S, K, and L are Cys25, His159, and Asn175 [1]. Differences mainly occur in the substrate-binding sites (Sn’-Sn), particularly in the residues of S1’ to S3 sites, which determine the binding specificity of the cysteine cathepsins. Our finding indicated that residues of the S1’ binding pocket of Cat S, in particular Arg141 and Asn163 constituted the main amino acid residues responsible for the binding of C4S dp2.

Following C4S dp2 modification and to improve the potency of the inhibition of Cat S, a synthetic multivalent C4S dp2-BSA neoglycoconjugate (NeoC4S dp2) was developed to leverage the micromolar IC₅₀ measured for biot-C4S dp2. The compound NeoC4S dp2 was found to be superior to the other monovalent inhibitors investigated, with three orders of magnitude, and is selective for Cat S vs. cysteine cathepsins expressed by differentiated THP-1 cells [24,25]. On the other hand, elevated endopeptidase activity of both Cat K and Cat L was observed with NeoC4S dp2. It is becoming established that GAGs are important regulators of papain-like lysosomal cysteine proteases, resulting in different effects on their activity and stability (for review: [34,35]). Interestingly, Brömme and colleagues reported that the residual activity of human Cat K was increased specifically by C4S, favoring its stability and catalytic activity [36,37]. This complied with studies by Novinec et al., showing that C4S and other GAGs increased the activity of the enzyme by promoting a conformational change in the enzyme [38,39]. It was suggested that C4S and other GAGs serve as natural allosteric modifiers of Cat K and this coupling on the enzyme surface potentiates the endopeptidase activity of the enzyme. Except for Cat K, no available experimental evidence of allosteric effectors of related-human cysteine cathepsins have yet been detailed. Nevertheless, GAGs have been shown to modulate the activity of several mammalian and non-mammalian cathepsins L- and B-like endopeptidases, possibly by allosteric mechanisms [40,41,42,43,44]. Despite C4S has no apparent stabilizing effect on the activity of human Cat L in vitro [37], we observed several C4S binding sites on the surface of Cat L that are located outside the active site [13]. This result suggests that C4S may lead to a likely conformational change in Cat L, as supported by another study using molecular dynamics simulations [39]. A likely hypothesis is that the endopeptidase activity of both cathepsins K and L could be enhanced in vitro by NeoC4S dp2, as previously established with C4S. Nevertheless, additional investigations are required to elucidate the molecular mechanism underlying the structural and functional consequences of Cat K and Cat L in the presence of NeoC4S dp2.

The spatial distribution and orientation of the thirty-one C4S dp2 units on BSA scaffold is likely to be of importance to enhance Cat S inhibition. Modeling of the multivalent C4S dp2 is likely to be an interesting and highly fruitful avenue of investigation. A better understanding of synthetic or hemi-synthetic biotin-labeled and multivalent sulfated C4S disaccharides structure–function relationship may lead to the discovery of novel selective Cat S inhibitors as potential therapeutic agents for the treatment of lung, cardiovascular and autoimmune diseases, associated with an aberrant expression and activity of Cat S.