Subnanomolar Cathepsin S Inhibitors with High Selectivity: Optimizing Covalent Reversible α‐Fluorovinylsulfones and α‐Sulfonates as Potential Immunomodulators in Cancer

The cysteine protease cathepsin S (CatS) is overexpressed in many tumors. It is known to be involved in tumor progression as well as antigen processing in antigen‐presenting cells (APC). Recent evidence suggests that silencing CatS improves the anti‐tumor immune response in several cancers. Therefore, CatS is an interesting target to modulate the immune response in these diseases. Here, we present a series of covalent‐reversible CatS inhibitors based on the α‐fluorovinylsulfone and ‐sulfonate warheads. We optimized two lead structures by molecular docking approaches, resulting in 22 final compounds which were evaluated in fluorometric enzyme assays for CatS inhibition and for selectivity towards the off‐targets CatB and CatL. The most potent inhibitor in the series has subnanomolar affinity (Ki=0.08 nM) and more than 100,000‐fold selectivity towards cathepsins B and L. These new reversible and non‐cytotoxic inhibitors could serve as interesting leads to develop new immunomodulators in cancer therapy.


Introduction
Cysteine cathepsins are ubiquitous papain-like proteases, in mammalians mainly located in the lysosome, involved in extracellular matrix degradation and intracellular protein processing. [1] They have various functions in cells and, above all, share a high structural similarity. [2] However, cathepsin S (CatS) differs from other cysteine cathepsins in its stability at neutral pH and its limited tissue distribution (mainly in antigenpresenting cells, e. g. macrophages). [3,4] CatS is known to be overexpressed in many tumors (e. g., follicular lymphoma, breast, gastric, colon, pancreatic cancer). [5][6][7][8] To date, various mechanisms how CatS is involved in tumor progression are known. For example, CatS is known to turn over extracellular matrix proteins and to drive tumor angiogenesis. [7,9] Additionally, Riese and co-workers showed that CatS regulates antigen processing and presentation in antigen-presenting cells (APC). [10][11][12][13][14][15] With this important role in immune cells, cathepsin S intervenes in the body's immune response also to tumors. It shifts MHC-II expression to MHCÀ I, resulting in a favored activation of CD4 + T cells (e. g., regulatory T cells) over cytotoxic CD8 + T cells. [5,6,16] Jakoš and co-workers and Wilkinson and co-workers also stated that CatS polarizes APCs from M1 to M2 phenotype that is associated with tumor progression, supporting the proliferation of myeloid-derived suppressor cells (MDSC) and tumor-associated macrophages. [9,16] This shift results in a suppressed T cell-induced immune response. [17][18][19] Data from murine models also indicates that CatS inhibition reduces the overall T cell immunity in healthy mice but enhances the CD8 + T cell immunity in mice with cancer. [20] Cytotoxic CD8 + T cells can attack tumor cells and thus lead to tumor volume reduction. [20] Experiments with small-interfering RNA (siRNA) targeting CatS mRNA and thus, reducing CatS expression, resulted in tumor volume and invasion reduction as well as increased apoptosis and attenuated angiogenesis. [21,22] Burden and co-workers used inhibitory CatS antibodies and observed an increased effect of chemotherapeutics plus a significant tumor growth limitation. [23,24] Furthermore, CatS overexpression occurs in follicular lymphoma including the expression of an overactive mutant (Y132D) with enhanced auto-activation. [5] Knocking down the protease leads to an improved immune response towards lymphoma cells. [6] Overall, CatS is an interesting new target to enhance anti-tumor immunogenicity and thus, stop tumor growth, especially in case of resistances to current tumor immunotherapies. [6,10,25,26] Figure 1 summarizes the mentioned effects of CatS in the tumor microenvironment (TME).
CatS is a papain-like protease expressed as an inactive zymogen. [27,28] After cleaving off the propeptide, the mature enzyme consists of 217 residues and a catalytic dyad (Cys25, His164) in the active site. [3,4,29] Despite the high structural similarity to other human cathepsins, there are various residues in the S1' to S3 subsites that differ and can be addressed to gain selectivity. [4,29] The first selective CatS inhibitors were published in the early 2000s based on a publication by Pauly et al. that described the binding site and especially the differences towards other cysteine cathepsins. [1,29] The S2 subsite contains a flexible Phe211 residue that can flip and open up to Phe70 from the S3 site, creating space for bulkier residues in S2. Furthermore, it allows ligand π-stacking with these Phe residues. [1,30] During the last 20 years, many cathepsin S inhibitors have been developed, including noncovalent as well as covalent ones (e. g. vinylsulfones, nitriles, aldehydes). [1,[30][31][32][33][34][35][36][37][38] One nitrile-based inhibitor has already been tested in vivo. The compound led to significant reduction of tumor volume in murine models. [39] The structures of three advanced CatS inhibitors are summarized in Scheme 1. [30,40,41] Here, we focus on developing new selective cathepsin S inhibitors based on the structure of the well-known pancathepsin inhibitor K11777. This compound with an electrophilic vinylsulfone warhead is known to be a covalent irreversible cathepsin inhibitor ( Figure 2).
The active site cysteine undergoes a Michael-type addition and cannot dissociate from the inhibitor after the covalent bond formation. Since irreversible inhibition has several drawbacks, e. g., off-target effects, toxicity, haptenization, we have recently developed modified K11777 derivatives by introducing a fluorine atom at the α-position of the vinylsulfone double bond. [42][43][44] The generated α-fluorovinylsulfone (1 a) undergoes a reversible Michael-type addition with thiols ( Figure 2). [45] With this reversibility we maintain the benefits of covalent inhibition, e. g., longer residence times, higher potency, thus possible dosage reduction, and a lower pharmacokinetic sensitivity, without the drawbacks of irreversible inhibition mentioned above. [46][47][48] Our most recent findings suggest that modifying the warhead from an α-fluorovinylsulfone (1 a) to an α-fluorovinylsulfonate (1 b) results in slowly reversible cathepsin inhibitors, further prolonging the target residence time (Figure 2). [49] Since covalent-reversible inhibition has many benefits, we chose previously described fluorinated derivatives from Schirmeister et al. and Jung, Fuchs et al. as initial starting structures (1 a, 1 b) for the development of new CatS inhibitors. [45,49] Results of molecular docking studies combined with literature-known motifs resulted in 22 new compounds ( Figure 3) that were tested in fluorometric enzyme assays against cathepsins S, B, and L. We evaluated their potency and selectivity profiles in a structure-activity relationship study backed up by molecular docking results. Finally, their cytotoxicities against the breast cancer cell line MDA-MB-231 and murine bone marrow-derived dendritic cells were tested in cell viability assays.

Chemistry
The synthesis route allowed many combinations of the two warhead variations with various dipeptides, resulting in the final inhibitors 1 a-6 c.

Warhead preparation
To prepare the warhead, we first synthesized fluorinated phosphonates 11 and 12 (Scheme 3) as previously published. [49] To synthesize aldehyde 14, we prepared the Weinreb amide 13

Dipeptide synthesis
For ester-protected dipeptides 19, 21, 23 and 41, we prepared isocyanates which then reacted with piperazine derivatives or morpholine. We synthesized the other ester-protected dipeptides using standard amide coupling reactions. The cleavage of the ester moieties by hydrolysis under basic conditions gave access to the final dipeptides 20-42 (Scheme 4).

Amide couplings
Warheads 17 or 18 were coupled with dipeptides 20-42 in standard amide coupling reactions (Scheme 5). The resulting inhibitors 1 a-6 c were purified by HPLC (> 95 % purity in all cases). For some compounds (4 c, 6 a, 6 b) the formation of diastereomers could not be avoided, but the (E)-configurations of the isolated, purified and tested inhibitors were confirmed in all cases using the coupling constants in the NMR.

Fluorometric enzyme assays
The synthesized compounds' inhibitory activities against CatS and the off-targets CatB and CatL were tested using well-known fluorometric enzyme assays (Table 1). [50] The inhibitors were initially screened at different concentrations (20 μM, 1 μM) followed by IC 50 /K i value determination if > 50 % inhibition at 20 μM. For more information regarding assay procedures and calculations of inhibition constants see 'Fluorometric enzyme assay' in the Experimental Section.
We started by varying the P3 position of lead compounds 1 a and 1 b ( Figure 3) and prepared two compound sets, namely the corresponding fluorovinylsulfones/-sulfonates (2 a-6 c), and evaluated their inhibitory activities and selectivities (Table 1). We generally observed all α-fluorovinylsulfonates (3 a-3 f & 5 a-5 c) to be more potent cathepsin S inhibitors than the corresponding α-fluorovinylsulfones (2 a-2 e & 4 a-4 d) which already improves their selectivity towards CatB and CatL (Table 1).
Next, we altered the P2 position while maintaining a 4pyridyl residue in P3, an effective strategy to achieve the most potent and selective P2 residues. The 4-pyr moiety was among the top two residues in P3 regarding potency and selectivity with a better synthetic accessibility than morpholyl. [35] Here, we also prepared corresponding α-fluorovinylsulfones/-sulfonates (4 a-5 d) and determined their inhibitory activities and selectivity profiles ( Table 1). The cyAla residue in P2 seems to be the most favorable, with K i values in the low nanomolar range (5.9-7.9 nM) for both warheads (4 a, 5 a, see Table 1). The resulting selectivity towards CatB and CatL is > 1,000-fold for both compounds and enzymes. Leucin in P2 combined with an α-fluorovinylsulfonate (5 b) also shows good CatS inhibition (K i = 35 nM) and high selectivity, but lacks selectivity compared to 5 a with cyAla in P2. Homophenylalanine and tryptophane are not suitable as they lack affinity and selectivity compared to the most favorable compounds.
Based on the results from these first optimizations, we prepared three additional compounds (6 a-6 c) with cyAla in P2 and modifications in P3. For future attachments of our inhibitors onto nanodelivery systems via various linkers, we chose to introduce an amino-substituted phenyl ring in the P3 position (6 b). We also tested the N-boc protected intermediate 6 a. Since we had found the morpholyl substituted compounds 2 c, 3 c to be very potent, with K i values of 0.9 nM (3 c) and 40 nM (2 c) respectively, we prepared a final inhibitor combining the favorable cyAla residue in P2 with the morpholyl moiety in P3 (6 c). The results shown in Table 1 reveal that all three moieties are suitable for the P3 position with K i values in the low nanomolar or even subnanomolar range and high selectivities towards cathepsins B and L. However, it should be noted that combining suitable residues in P2, P3 and the most potent warhead is essential to achieve a highly active and selective inhibitor. Compound 6 c (morpholyl in P3, cyAla in P2, F-vinylsulfonate warhead) with a K i value around 90 pM and more than 100,000-fold selectivity towards the other two cathepsins is the most potent and selective inhibitor of the series.
To verify the extraordinarily high inhibition potency of 6 c, we also used an alternative substrate (Z-Phe-Arg-AMC) and  [a] See Schirmeister et al. [45] [b] See Jung, Fuchs, et al. [49] [c] Time-dependent inhibition.
repeated the inhibition experiments, resulting in a similar K i (120 pM, see Figure 5 and Fluorometric enzyme assay). Generally, the progress curves for α-fluorovinylsulfones are linear, indicating that the inhibition is not time-dependent ( Figure 4). For the α-fluorovinylsulfonates, we observed timedependent inhibition ( Figure 5) with biphasic binding behavior as we have reported previously for α-fluorovinylsulfonate inhibitors of the cysteine protease rhodesain. [49] Thus, we also determined further inhibition constants, such as k 3 , k 4 and K i * (dissociation constant of final covalent complex) using the slowbinding equation for these compounds [1 b, 3 a-f, 5 a-6 c, Table 2, and Equation (1), which depicts enzyme-inhibitor complex formation and the relevant constants]. [51] (1) We found the rate constant of the dissociation of the final complex (k 4 ) to be significantly lower than the association rate constant (k 3 ) for all time-dependent compounds ( Table 2), suggesting tight-binding behavior. The dissociation constants of the final covalent complexes (K i *) are in the low nanomolar  50 calculation. K i value was calculated using the Cheng-Prusoff equation. [51]  [a] k 3 , k 4 and K i * calculated with the slow-binding equation. [52] to subnanomolar range proving the very tight binding of the inhibitors.

Dilution assay
With dilution assays, we proved that the inhibitors are reversible as expected from our previous experiences with such compounds. [49] For the experiments, two compounds (F-vinylsulfonate 3 c, F-vinylsulfone 4 a) were incubated with cathepsin S, followed by 100-fold dilution with substrate-containing assay buffer. In case of reversible inhibition, the enzyme activity should recover. Furthermore, we used the pan-cathepsin inhibitor K11777 as an irreversible control. [53] Figure 6 shows that enzyme activity can be recovered for Fvinylsulfones and -sulfonates, suggesting reversible inhibition as anticipated.

SAR discussion
Comparing the K i values of the lead compounds F-vinylsulfone 1 a (K i = 1.2 μM) and F-vinylsulfonate 1 b (K i = 0.010 μM) with P3 substituted inhibitors 2 a-2 e and 3 a-3 f shows significantly enhanced affinity towards the target enzyme CatS in all cases except for piperidyl substituted vinylsulfone 2 b (K i = 1.3 μM) and 3-thiophenyl substituted vinylsulfonate 3 e (K i = 0.011 μM). The non-covalent docking results show scores in the same range or even higher compared to the leads 1 a & 1 b. For compound 3 f the predicted score for the stability of the covalent complex is even higher compared to the lead compounds (see Table B, Supporting Information). Additionally, plots resulting from assays with substrate Z-Val-Val-Arg-AMC for K i calculation using the slow-binding equation. [52] (C) k obs [s À 1 ] vs. Inhibitor concentration [μM] plots resulting from assays with substrate Z-Phe-Arg-AMC for K i calculation using the slow-binding equation. [52] Figure 6. Dilution assays. DMSO as control (black) and K11777 as irreversible control (red). Incubation of CatS with compounds 3 c (F-vinylsulfonate, blue) and 4 a (F-vinylsulfone, green) followed by 100-fold dilution results in an enzyme activity recovery. the selectivities for CatS increased significantly towards CatB for all P3 modified compounds (2 a-3 f) by up to 220-fold for inhibitor 2 a vs 1 a. The selectivity towards CatL also improved in all cases (2 a-3 f) from 0.02-fold for 1 a to > 71-fold for 2 a (vinylsulfones) and from 2.4-fold for 1 b to > 1,100-fold for 3 a (vinylsulfonates). Overall, the vinylsulfonate warhead resulted in more potent inhibitors as exemplified for compounds 2 b vs 3 b. Superposition of the covalent docking poses of both compounds revealed that the phenylalanine substituent of 3 b leads to a different orientation of the phenyl ring inside the S2 subpocket of the active site, where additional face to edge and face to face π-stacking interactions with Phe211 and Phe70 could be possible, possibly leading to tighter binding of 3 b compared to 2 b (Figure 7).
Replacing phenylalanine in P2 with four different amino acids while maintaining the 4-pyridyl substituent in P3 lead to the result that the cyAla-residue is best suited to address the S2 pocket of the enzyme with both warheads. This is highlighted by the increase in potency for the vinylsulfone-based inhibitor 4 a (K i = 0.006 μM) compared to 2 d (K i = 0.10 μM) and a better selectivity towards both off-targets (> 2,500-fold vs CatB and > 1,700-fold vs CatL). For the vinylsulfonate-based inhibitor series, the same exchange resulted in a potency drop of about 10-fold (3 d vs 5 a) but a big jump in selectivity (> 1,900-fold vs CatL and > 1300-fold vs CatL) for inhibitor 5 a. This might be due to the sp 3 -hybridization of the carbon atoms of the cyAla substituent, which better fills the S2 subpocket of CatS and presumably generates new non-polar interactions with the subpocket atoms compared to the sp 2 -hybridized planar phenyl ring ( Figure 8). The S2 pockets of both off-target cathepsins lack the depth for accepting bulky residues. Additionally, a water molecule present in the S2 pocket, could putatively be expelled by the hydrophobic cyAla residue and thus lead to a change in entropy and thereby have an impact on the binding free energy.
Combining the cyAla motif in P2 with morpholine in P3 and the vinylsulfonate warhead yielded the most potent inhibitor 6 c, with a K i -value in the picomolar range and excellent selectivities over CatB (> 150,000) and CatL (> 125,000). Superposition of the non-covalent docking pose of 6 c with the covalent enzyme-inhibitor complex shows that all polar interactions between the non-covalently bond inhibitor and the enzyme should still be intact after the covalent bond formation ( Figure 9). Compound 6 c also has one of the highest scores for the stability of the covalent enzyme inhibitor complex (Affinitỹ G, MOE-score = À 6.0 kcal/mol) as well as the second highest HYDE-score of all inhibitors with À 50 kJ/mol (Table B, Supporting Information).
The incorporation of an amine functionality in the P3 site for compound 6 b (K i = 9 nM) and its boc-protected intermediate 6 a (K i = 6 nM) did not deteriorate the affinity or the selectivity for the target enzyme (Table 1). Therefore, 6 b can be used in future studies with nanodelivery systems.

Cell viability
Selected compounds were tested in a CellTiter-Glo Luminescent Cell Viability assay to assess their cytotoxicity. We used MDA-MB-231 cells which are breast cancer cells that compensate the inhibition of CatS and other cathepsins. Therefore, only unspecific cytotoxic effects, that are not related to CatS inhibition, are detected. [54] We did not observe significant cytotoxicity ( Figure 10) after 24 h treatment at concentrations > 1,000-fold higher (20 μM) than the compounds' K i values in the low nanomolar range. Only compounds 5 a and 6 b exhibited low cytotoxic effects at the highest concentration applied (100 μM), which was > 10,000-fold the compounds' K i values. In conclusion, the exemplarily selected compounds do not affect cell viability of MDA-MB-231 cells at their biologically active concentrations.
In addition, the cytotoxic effect of several compounds was also tested on single cell level using murine DC. The various compounds (1 μM) were applied alone or followed by administration of the DC activator lipopolysaccharide (100 ng/mL) required to achieve robust T cell stimulatory activity. Neither compound exerted major cytotoxic activity on CD11c + DC at concentrations about 1,000-fold higher than their K i values as assessed using membrane impermeable fixable viability dye, which binds to amines of cytoplasmic proteins of dead cells with a porous cell membrane ( Figure 11).

Conclusions
Here, we have demonstrated that α-fluorovinylsulfones and -sulfonates are potent covalent-reversible cathepsin S inhibitors. Both warheads are well suitable for the target enzyme, with the α-fluorovinylsulfonates being more effective. Starting from the K11777 scaffold, we replaced residues in the P2 and P3 positions, resulting in high affinity compounds, some of them being highly selective against off-target cathepsins. In the P3 position, we observed a morpholyl (3 c) or 4-pyridyl (3 d) residue to be most suitable with K i values in the subnanomolar range and moderate selectivity. In the P2 position, we found that cyAla (4 a, 5 a) increased the selectivity immensely with K i values in the low nanomolar range for the on-target CatS. Combining the best-performing residues of P2 and P3 to form a morpholyl-cyAla-hPhe-F-vinylsulfonate-Ph motif (6 c) proved to be most effective with subnanomolar affinity (K i = 0.09 nM, K i * = 0.01 nM) and exceptional selectivity towards cathepsins B and L (> 150,000/125,000-fold). The time-dependent inhibition enables slow-tight binding, thus prolonging target residence times.   Therefore, compound 6 c will be an excellent candidate for further optimizations regarding new small molecule immunomodulators in cancer therapy, where already resistances to existing immunotherapies are known. [26] Cell viability experiments using a non-CatS sensitive cancer cell line and murine derived dendritic cells both did not show cytotoxic effects for all tested inhibitors at relevant concentrations (> 1,000-fold K i ). The next steps include immunoassays with macrophages or dendritic cells to evaluate the potential of our CatS inhibitors, e. g., for immune cell polarization. This could involve markers like MHC-I and MHC-II expression as well as functional assays for T cell activation.
Moreover, development of inhibitor-nanocarrier constructs is possible with these compounds. By attaching cathepsin S inhibitors to nanocarriers, their efficacy could be further enhanced through specific targeting, e. g., to dendritic cells or other APC. [25] Compound 6 b with a free amino moiety (K i = 9 nM, SI > 1,000) allows the attachment of various functionalities, such as linkers or nanodelivery systems.

Experimental Section General
All reagents and solvents were purchased from Sigma-Aldrich, Alfa Aesar, Acros, TCI, BLD Pharmatech, Carbolution or Carl Roth in analytical or HPLC grade quality. Chemicals were used without further purification, whereas solvents were distilled and desiccated by standard methods if necessary. 1 H and 13 C spectra were recorded on a Bruker Fourier 300 device using DMSO-d 6 or CDCl 3 as solvents. Chemical shifts δ are given in parts per million (ppm) using residual proton peaks of the solvent as internal standard ( 1 H/ 13 C: DMSO 2.50/39.52 ppm; CHCl 3 7.26/77.16 ppm). The compound purity was determined via HPLC-MS at λ = 254 nm using an Agilent 1100 series HPLC with an Agilent Poroshell 120 EC-C 18 column (150 × 2.10 mm, 4 μm) coupled with an Agilent 1100 series LC/MSD Trap with electron spray ionization (ESI) in positive mode. All compounds tested in enzymatic assays are �95 % pure by HPLC analysis. The mobile phase consisted of a variable mixture of ACN and H 2 O with 0.01 % formic acid. For purification we used a Varian PrepStar system (model 218) with a MZ-Aqua Perfect C 18 column (250 × 20 mm, 7 μm) by MZ-Analysentechnik. Column chromatography was performed with silica gel (0.040-0.063 mm) and all reactions were monitored by thin-layer chromatography using Macherey-Nagel Alugram Xtra SIL G/UV254 silica gel 60 plates for detection at λ = 254 nm. Melting points were determined in open capillaries with a Schorpp Device Technology MPM-H3 instrument. Optical rotation a ½ � 22 D was measured on a Krss P3000 polarimeter (c = 10 mg/mL in MeOH) at 22°C.

Procedure B (amide couplings)
The carboxylic acid (1.2 eq) was dissolved in DCM or a mixture of DCM/DMF, and cooled to 0°C. Then, HOBt (1.2 eq), TBTU (1.2 eq), and DIPEA (3.5 eq) were added, and the mixture was stirred for 20 min until all components dissolved. The respective amine (1.0 eq) was added, and the mixture was stirred for an additional 12-24 h, then stopped by adding water. The mixture was extracted with DCM (2 ×) and the combined organic extracts were washed with water (2 ×), sat. aq. NaHCO 3 (2 ×), and brine (2 ×). After drying the crude product over Na 2 SO 4 , it was purified by column chromatography.

Procedure C (boc deprotection)
HCl (4 M) in dioxane was added dropwise to the boc-protected amine (1.0 eq) until all components dissolve. The mixture was stirred for 2-12 h and the product was precipitated with diethyl ether and lyophilized afterwards.

Procedure D (alkaline hydrolysis)
The ester (1.0 eq) was dissolved in THF. LiOH monohydrate (4.0 eq) was dissolved in water and added to the reaction dropwise. The mixture was stirred for 12-24 h, then the solvent was removed under reduced pressure. The pH of the aqueous phase was adjusted with KHSO 4 to 5, giving the products as solids that were further lyophilized.

Molecular docking
Since the inhibitors were designed to react covalently with cysteine-25 of CatS, two different docking approaches were followed. First, a conventional non-covalent docking was performed, to estimate affinity and geometry of the pre-organized enzyme-inhibitor complex, secondly a covalent docking was used to determine the final covalent enzyme-inhibitor complex. In both docking setups a crystallographic reference ligand was used for validation via redocking (Table A, Supporting Information). Molecular docking experiments were performed using the following crystal structure freely available in the protein data bank (PDB): [55] Cathepsin S covalently bound to N-2-(morpholin-4-ylcarbonyl)-N-[(3S)-1-phenyl-5-(phenylsulfonyl)pentan-3-yl]-l-leucinamide (C1P), PDB entry 1NPZ. [29] For both docking approaches, chain A of the dimer of 1NPZ was extracted via PyMOL 2.5.2. [56] All ligands were energetically minimized prior docking with Molecular operating environment (MOE Version 2020.09) [57] using the MMF94x force field. [58] For visual presentation of the top binding poses, PyMOL 2.5.2 was used. [56]

Docking approach A: non-covalent docking with LeadIT
The non-covalent docking was performed with LeadIT 2.3.2. [59] The receptors were prepared in MOE with the protonate3D functionality and the covalent bond between the co-crystallized ligand and the corresponding protease was untethered via the Builder tool in MOE. For the receptor the binding site was defined as a 6.5 Å shell around the bound reference ligand. Water molecules that form at least three hydrogen bonds with the receptor and ligand were kept as part of the binding site. The docking was performed under default settings using the enthalpy-entropy hybrid approach with 2,000 solutions per iteration and fragmentation. Only the top pose of the initial docking was kept and re-scored using the HYDE scoring function. [60] For the docking, pharmacophore constraints needed to be included to obtain reasonable binding modes. The nitrogen atoms of the peptide backbone were therefore defined as H-bond donors with a 1 Å sphere radius.

Docking approach B: covalent docking with MOE
Covalent docking was performed with MOE. The receptor was prepared using the 3D protonation tool inside MOE. For the covalent reaction of the different warheads, the already existing template reactions were used. Initial 30 poses from the triangle match placement with London~G scoring were re-scored using the Affinity~G scoring function and induced fit refinement