Inhibitor recognition specificity of MERS-CoV papain-like protease differs from that of SARS-CoV

Abstract

The Middle East Respiratory Syndrome coronavirus (MERS-CoV) papain-like protease (PLpro) blocking loop 2 (BL2) structure differs significantly from that of SARS-CoV PLpro, where it has been proven to play a crucial role in SARS-CoV PLpro inhibitor binding. Four SARS-CoV PLpro lead inhibitors were tested against MERS-CoV PLpro, none of which were effective against MERS-CoV PLpro. Structure and sequence alignments revealed that two residues, Y269 and Q270, responsible for inhibitor binding to SARS-CoV PLpro were replaced by T274 and A275 in MERS-CoV PLpro, making critical binding interactions difficult to form for similar types of inhibitors. High-throughput screening (HTS) of 25,000 compounds against both PLpro enzymes identified a small fragment-like noncovalent dual inhibitor. Mode of inhibition studies by enzyme kinetics and competition surface plasmon resonance (SPR) analyses suggested that this compound acts as a competitive inhibitor with an IC₅₀ of 6 µM against MERS-CoV PLpro, indicating that it binds to the active site, whereas it acts as an allosteric inhibitor against SARS-CoV PLpro with an IC₅₀ of 11 µM. These results raised the possibility that inhibitor recognition specificity of MERS-CoV PLpro may differ from that of SARS-CoV PLpro. In addition, inhibitory activity of this compound was selective for SARS-CoV and MERS-CoV PLpro enzymes over two human homologues, the ubiquitin C-terminal hydrolases 1 and 3 (hUCH-L1 and hUCH-L3).

Graphical abstract

INTRODUCTION

Middle East Respiratory Syndrome coronavirus (MERS-CoV), previously called human coronavirus-Erasmus Medical Center (HCoV-EMC), was first reported in Saudi Arabia in 2012 and spread to twenty different countries,^1–4 resulting in 853 infections with 301 deaths as of October 2, 2014.⁵ The unusually high case-fatality rate (CFR) of MERS-CoV infections (~35%) is alarming as it far exceeds that of all other known human coronaviruses, including the human severe acute respiratory syndrome coronavirus (SARS-CoV). SARS-CoV caused a fatal global outbreak in 2003, resulting in 800 deaths (~10% CFR).⁶ There are over 20 known coronaviruses (CoV), six of which are identified as human coronaviruses (HCoV) (Supplementary Figure S1). Coronaviruses are classified into four genera (α, β, γ, and δ), and each genus can be divided into lineage subgroups. Of the six HCoVs, two (NL63 and 229E) belong to genus α, and the remaining four (HKU1, OC43, SARS-CoV, and MERS-CoV) belong to genus β. Within the betacoronavirus genus, SARS-CoV is classified as lineage group B, while MERS-CoV is categorized into lineage group C based on their genomes. Two bat CoVs from lineage group C, BtCoV-HKU4 and BtCoV-HKU5, are the most closely related to the MERS-CoV.^{2, 7–9} MERS-CoV and SARS-CoV are highly pathogenic, with evidence of person-to-person transmission via either household or hospital contacts.^{10, 11} MERS-CoV and SARS-CoV use different receptors, dipeptidyl peptidase 4 (DPP4 or CD26) and angiotensin-converting enzyme 2 (ACE2), respectively,^{12, 13} and the epidemiology of MERS-CoV is still being investigated. Both MERS-CoV and SARS-CoV exhibit as a severe respiratory infection, while MERS-CoV exhibits an additional unique symptom of renal failure.² Even though the MERS-CoV transmission rate is slower than that of SARS-CoV, the number of MERS-CoV infections continues to grow.^{11, 14, 15} Due to the recent emergence of this new coronavirus and the potential of SARS-CoV retransmission from zoonotic reservoirs to humans,^16–18 the possibility of another deadly pandemic has been seriously raised. However, there is still no effective therapeutic available against either coronavirus. Therefore, developing treatments against both coronaviruses is important.

Both MERS-CoV and SARS-CoV are single-stranded positive-sense RNA viruses with approximately 30 kb genome sizes. Each of their genes encodes two polyproteins called pp1a and pp1b (Figure 1A) that are processed by two proteases, a 3-C-like protease (3CLpro) and a papain-like protease (PLpro). Many coronaviruses contain two PLpro enzymes (PLP1 and PLP2), but MERS-CoV and SARS-CoV have only one PLpro enzyme.^{19, 20} PLpro enzymes are part of a large non-structural protein 3 (nsp3) that contains four other domains, a ubiquitin-like fold (UB1), an ADP-ribose-1d–phosphatase (ADRP) domain, a SARS-unique domain (SUD), and a transmembrane (TM) domain (Figure 1A). PLpro is responsible for cleavage of the first three positions of its polyprotein, while 3CLpro cleaves the remaining eleven locations, releasing a total of 16 non-structural proteins (nsp) in both MERS-CoV and SARS-CoV. Sequence motifs recognized by MERS-CoV PLpro (MERS-PLpro) and SARS-CoV PLpro (SARS-PLpro) are (L/I)XGG↓(A/D)X and LXGG↓(A/K)X, respectively (Figure 1B). Unlike 3CLpro, SARS-PLpro has been shown to be a multifunctional protein involved in de-ISGylation, deubiquitination, and viral evasion of the innate immune response in addition to viral peptide cleavage as a protease.^{16, 21} Researchers have discovered that the MERS-PLpro also exhibits deubiquitination and de-ISGylation functions, blocking the interferon regulatory factor 3 (IRF3) pathway.^{22, 23} Both 3CLpro and PLpro are known to be essential for viral replication, making them attractive targets in antiviral drug discovery.^{20, 24} In this study we characterized MERS-PLpro as a drug target by solving its complete structure followed by thorough analysis with four known SARS-PLpro lead inhibitors. In addition, from high-throughput screening (HTS) of a 25,000-compound antimicrobial focused library against both MERS-CoV and SARS-PLpro enzymes, we identified a low molecular weight compound that showed activity against both PLpro enzymes via two different modes of inhibition.

Figure 1. Schematics of SARS-CoV and MERS-CoV polyproteins.

(A) Cleavage positions of PLpro (pink) and 3CLpro (cyan) are shown by different colored arrows in their polyproteins. (B) Cleavage site comparison between SARS and MERS PLpro enzymes. Sequence motifs recognized by SARS-CoV PLpro (SARS-PLpro) and MERS-CoV PLpro (MERS-PLpro) are LXGG↓(A/K)X and (L/I)XGG↓(A/D)X, respectively.

RESULTS AND DISCUSSION

Overall structure comparison of MERS-CoV and SARS-CoV PLpro enzymes

We determined the complete x-ray crystal structure of unbound MERS-PLpro (PDB code: 4RNA) at 1.8 Å with R and R_free values of 19.2% and 24.5%, respectively (Supplementary Table S1) in addition to our previously released apo structure at lower resolution (PDB code: 4PT5). MERS-PLpro is composed of four distinct domains, an ubiquitin-like (Ubl)-domain, thumb, palm, and zinc fingers (Figure 2A). The overall structure of MERS-PLpro resembles that of SARS-PLpro^{21, 25} with an α carbon RMSD of 1.2 Å. There are minor structural deviations of the Ubl, thumb and palm domains between MERS-PLpro and SARS-PLpro. The majority of variation is generated from the zinc finger domain and two loop regions. The whole zinc finger domain was shifted approximately 5.9 Å when the zinc location was measured (Figure 2B). The three-dimensional arrangement of the Cys-His-Asp catalytic triad in MERS-PLpro aligns well with that of SARS-PLpro.

Figure 2. Crystal structure of MERS-PLpro.

(A) MERS-PLpro crystal structure overview (PDB: 4RNA). The overall structure is composed of four domains indicated by different colors, namely a Ubl-domain (orange), a thumb (cyan), a palm (yellow) and fingers (magenta). The protease catalytic site is located in the interface between the thumb and palm domains, indicated by a black circle. (B) Overlaid structures of SARS-PLpro (PDB: 2FE8) and MERS-PLpro (PDB: 4RNA). Detailed information for the catalytic triad, zinc-binding motif, BL1, and BL2 are shown in expanded boxes.

MERS-PLpro contains two blocking loops named BL1 and BL2, shown in red in Figure 2A, that could be structurally important. The two corresponding loops of human ubiquitin-specific protease 14 (USP14) have been proven to be crucial in blocking accessibility to the active site.²⁶ Specifically, the BL1 and BL2 loops of USP14 are positioned closed down toward its active site absent bound substrate, and those loops open away from the active site upon the substrate binding, regulating its catalytic activity.²⁶ The length of MERS-PLpro BL1 is much longer than that of SARS-PLpro BL1, and it does not overlay well with USP14 BL1. Our structure contains the second loop (BL2) that was not defined in another recently solved apo structure of MERS-PLpro (PDB ID: 4P16).²⁵ That structure and our structure are almost identical except for our structure defining the BL2 loop and two different cysteine modifications. This BL2 loop plays a crucial role in SARS-PLpro inhibitor binding according to four currently available x-ray crystal structures in complex with four different inhibitors.^27–29 Bailey-Elkin and coworkers recently determined MERS-PLpro (PDB: 4RF0 and 4RF1) bound to an ubiquitin substrate, and found the BL2 loop to be defined at a location similar to that of the ubiquitin aldehyde substrate bound SARS-PLpro.^{30, 31}

Active-Site comparison and oxyanion hole stabilization

The cysteine/serine protease mechanism involves tetrahedral intermediate formation, following nucleophilic attack by the cysteine/serine side chain. Generally, there is an oxyanion close to the catalytic site, which interacts with a negatively charged tetrahedral intermediate to stabilize it. Ratia et al. demonstrated that the SARS-PLpro W107 located below the catalytic cysteine plays this vital role in forming a hydrogen bond (H-bond) with an intermediate as an H-bond donor in the active site by showing that the SARS-PLpro W107A mutant completely lost catalytic activity.²¹ However, in the MERS-PLpro active site, the equivalent position is occupied by L106, which is not capable of being an H-bond donor (Figure 3A). Lei et al. recently demonstrated that the L106W mutation resulted in catalytic activity enhancement of MERS-PLpro,²⁵ indicating that the MERS-PLpro oxyanion hole may not be complete in comparison to that of SARS-PLpro. Interestingly, the leucine residue at this position is highly conserved in three bat coronaviruses (BtCoV-HKU4, BtCoV-HKU5, and BtCoV-133) that belong to the same lineage group C as MERS-CoV (Figure 3B). On the other hand, two human coronavirus (NL63 and 229E) PLpro enzymes have a residue (Q or T) that can be an H-bond donor similar to SARS-PLpro. Therefore, another residue must play this intermediate-stabilizing function in MERS-PLpro. Asparagine (N110) in SARS-PLpro is highly conserved among various coronavirus PLpro enzymes, and Ratia et al. suggested that this residue could be another residue contributing to the oxyanion hole stabilization in addition to W107.²¹ From the structural alignment of the active site, we noted that the N109 of MERS-PLpro overlaps with N110 of SARS-PLpro. This suggested that N109, located above the catalytic cysteine, might be the residue that plays this critical role in MERS-PLpro. We hypothesized two potential mechanisms: First, the side chain amine group of N109 could form an H-bond with the intermediate’s oxyanion as an H-bond donor (Figure 3C). Alternatively, the carbonyl group of N109 could bind to a water molecule, followed by the water forming another H-bond with the negatively charged intermediate (Figure 3D). The positions of N109 in these two scenarios could differ. We generated two MERS-PLpro mutants, N109A and N109D, to investigate these two hypotheses. Enzyme activity of the N109A mutant was completely abolished, while the N109D mutant exhibited only ~13.8% of the wild-type MERS-PLpro activity (Figure 3E). This indicates that the N109 residue is indeed crucial for stabilizing the intermediate for the enzyme to perform its catalytic function. If N109 stabilized the intermediate via the second hypothesis, the side chain of N109D could still form an H-bond with a water molecule through the carbonyl group of aspartic acid, rescuing the MERS-PLpro enzyme activity. However, the N109D mutant also showed very low enzyme activity as compared to the wild-type, suggesting that the second hypothesis is not likely to be the main stabilization mechanism. This result accordingly suggests that N109 is a critical residue for intermediate stabilization, probably through an H-bond formation with the side chain amine group of N109.

Figure 3. Active site analysis of the MERS-PLpro.

(A) Active site alignment of MERS-PLpro (tan) and SARS-PLpro (cyan). The three catalytic triad residues (C111, H278 and D293) of MERS-PLpro are aligned with the SARS-PLpro catalytic triad (C112, H273 and D287). (B) Sequence alignment of important residues near the catalytic triad between various CoV. Residue numbers are shown for MERS-PLpro. (C) Potential mechanism 1 for oxyanion hole stabilization via N109. Active site and substrate residues are shown in green and pink, respectively. (D) Potential mechanism 2 for oxyanion hole stabilization via N109. (E) Enzyme activity comparison between wild-type and two mutant MERS-PLpro enzymes.

In addition to containing a crucial residue that stabilizes the oxyanion, the small loop (residues 101–108) next to the active site in SARS-PLpro is important for its catalytic activity via controlling active site access. The hydrogen bond between D109 from this loop and W94 restrains the loop conformation, preventing it from moving to block active site access.²¹ These two residues (D108 and W93 in MERS-PLpro) are conserved in MERS-PLpro, playing the same role as that of SARS-PLpro (Figure 3B).

Assay optimization for MERS-PLpro

PLpro cysteine proteases from both coronaviruses cleave the first three positions of its polyprotein. Residues at P1 (Gly), P2 (Gly), and P4 (Leu), are highly conserved in SARS-CoV, which has a consensus sequence of LXGG. This LXGG is also the consensus sequence in the C-terminal tail of ubiquitin, which led researchers to discover that SARS-PLpro also has deubiquitinating activity.^{21, 32} On the other hand, the P4 residue is somewhat less conserved in MERS-CoV. The second cleavage site between nsp2 and nsp3, FRLKGG↓AP(I/V)K, is almost identical between SARS-CoV and MERS-CoV. Thus, we decided to use the well validated ubiquitin-derived SARS-PLpro peptide substrate, RLRGG-AMC, for MERS-PLpro as well. The Michaelis constant (K_M) was determined with this substrate for both SARS-PLpro and MERS-PLpro enzymes side by side for comparison. The substrate K_M values of SARS-CoV and MERS-PLpro were 75.9 µM and 142 µM, respectively (Supplementary Table S2). The K_M value of MERS-PLpro was 2-fold larger than that of SARS-CoV PLpro, while the k_cat (turnover number) value of SARS-CoV was ~25-fold larger than that of MERS-PLpro. Therefore, the catalytic efficiency (k_cat/K_M) of SARS-PLpro for this substrate is ~45-fold higher than that of MERS-PLpro. There are two SARS-PLpro structures in complex with an ubiquitin-derived substrate, which helped to understand how the substrate binds to SARS-PLpro. One is with a wild-type SARS-PLpro with an ubiquitin aldehyde, and the other is a C112S mutant SARS-PLpro with bovine ubiquitin as a substrate.^{31, 33} Lei and colleagues compared the C112S mutant SARS-PLpro complex structure with the apo MERS-PLpro and reported that the S2 (P163 vs. L163), S3 (F269 vs. Y265, A162 vs. E162), and S5 (R168 vs. E168) subsites of MERS-PLpro exhibited different features from those of SARS-PLpro,²⁵ which could explain the significant catalytic efficiency difference between the two PLpro enzymes with the same substrate.

SARS-PLpro lead inhibitors do not inhibit MERS-PLpro

We and others have previously identified and developed a series of non-covalent SARS-PLpro inhibitors using high-throughput screening (HTS) and structure-based drug design, that can be classified into two distinct scaffolds.^{27–29, 34} Four inhibitor structures from these two scaffolds are shown in Supplementary Figure S2, with I-1 and I-2 representing scaffold 1, and I-3 and I-4 representing scaffold 2. Inhibitors I-2 and I-3 exhibited excellent inhibitory activities with IC₅₀ values of 0.34 µM and 0.6 µM against SARS-PLpro, with SARS antiviral activities of 2 µM and 15 µM, respectively.^{27, 34} Two SARS-PLpro complex crystal structures, with lead inhibitors from each scaffold (I-2 or I-3) revealed that inhibitors bind not to the catalytic site of the PLpro enzyme, but to the BL2 loop, blocking the entrance of the active site. This appears to prevent substrate access to the catalytic site, inhibiting PLpro enzyme activity.

These four SARS-PLpro lead inhibitors (I-1 ~ I-4) were tested against MERS-PLpro to determine whether these two PLpro enzymes behave similarly or not. Surprisingly, none of them showed any inhibitory activity against MERS-PLpro. This result led us to further analyze what determined the interaction between SARS-PLpro and its inhibitors. The amide group of inhibitor I-2 forms two hydrogen bonds with D165 and Q270 (Supplementary Figure S3A). The amide group of inhibitor I-3 also forms a hydrogen bond with Q270 in the BL2 loop (Supplementary Figure S3B). The aromatic ring of Y269 forms a hydrophobic interaction with the naphthyl ring of both I-2 and I-3. Therefore, residues Q270 and Y269 form common key interactions in both scaffold 1 and 2 lead inhibitors of SARS-CoV PLpro. However, neither Q270 nor Y269 residues exist in MERS-PLpro (Figure 2B). In MERS-PLpro, A275 exists in place of Q270 of SARS-PLpro, eliminating potential hydrogen bonding with inhibitors. Additionally, the second key interaction is also impossible because T274 of the MERS-PLpro does not have the aromatic ring of Y269. Apparently due to the lack of these two key residues, none of the SARS-PLpro lead inhibitors had any inhibitory efficacy against MERS-PLpro.

In addition to the two SARS-PLpro-inhibitor complex structures discussed above, two more SARS-PLpro-inhibitor complex structures have been recently determined with two advanced compounds that were further optimized from scaffold 2 by structure-activity relationship (SAR) studies.²⁹ In order to further analyze the interaction between SARS-PLpro and its inhibitors, we aligned all four available SARS-PLpro-inhibitor complexes, one SARS-PLpro in complex with an ubiquitin aldehyde substrate, one MERS-PLpro complex with an ubiquitin substrate, and the apo structures of both SARS-PLpro and MERS-PLpro (Figure 4).^{21, 27–30, 34} All eight structures aligned well with each other, except for several distinct locations, including the zinc-binding motif and BL2 regions. The zinc atoms were shifted similar distances and locations in all five SARS-PLpro complex structures as compared to the apo structure, as noted with a red arrow pointed to the right in the bottom insert in Figure 4A. The equivalent zinc atom in MERS-PLpro was located in a different position and was shifted in a different direction from those in SARS-PLpro (dark brown arrow in Figure 4A). Residues involved in SARS-PLpro inhibitor binding are shown in Figure 4B, along with inhibitor I-3, in order to illustrate the orientation of the inhibitor, inhibitor-interacting SARS-PLpro residues, and the flexible BL2 loop. Figures 4C and 4D show the binding locations of inhibitors and ubiquitin substrates and the flexible BL2 loop in exactly the same orientation. The SARS-PLpro flexible BL2 loop blocks the entrance of the tunnel to the active site when it is unbound and becomes well-ordered upon binding of either an inhibitor or a substrate (Figure 4C and 4D) through conformational changes.^{21, 27–29, 31, 33} For MERS-PLpro, the flexible BL2 loop is positioned much further from its active site than that of SARS-PLpro when it is unbound. Upon substrate binding to MERS-PLpro, its BL2 loop moves to an orientation similar to that of the substrate bound SARS-PLpro loop, as noted with a red and dark brown arrows in Figure 4D.³⁰ Both BL1 and BL2 loops of USP14 appear to play a regulatory role in its deubiquitinating activity,²⁶ while mainly BL2 seems to serve this role in SARS-PLpro. The question of whether the BL2 loop of MERS-PLpro plays a regulatory role in its deubiquitinating activity remains to be answered.

Figure 4. Structure comparison of SARS-PLpro complexes and MERS-PLpro.

(A) Overlay of five SARS-PLpro complex structures with an inhibitor or a substrate, apo SARS-PLpro, apo MERS-PLpro, and MERS-CoV-PLpro complex with an ubiquitin. The PDB codes of aligned structures are: Apo MERS-CoV-PLpro (PDB: 4RNA); MERS-CoV-PLpro complex with an ubiquitin (PDB: 4RF1); Apo SARS-CoV-PLpro (PDB: 2FE8); SARS-CoV-PLpro inhibitor complex with inhibitor I-2 (PDB: 3E9S); SARS-CoV-PLpro complex with inhibitor I-3 (PDB: 3MJ5); SARS-CoV-PLpro complex with inhibitor 3k (PDB: 4OVZ), SARS-CoV-PLpro complex with inhibitor 3j (PDB: 4OW0), and SARS-CoV-PLpro complex with ubiquitin aldehyde substrate (PDB: 4MM3). (B) Expanded overlaid structures of BL2 and surrounding residues involved with inhibitor binding. (C) Different orientation of Figure 4B, showing catalytic residues and relative BL2 orientations. (D) Expanded overlaid structures of BL2 loops and an ubiquitin aldehyde (blue) and an ubiquitin (orange) substrates for SARS-PLpro and MERS-PLpro, respectively. Ubiquitin is hidden, and only part of the each substrate is shown in this figure due to space constraints. (E) The active-site, catalytic triad (CT), and two blocking loop (BL1 and BL2) residues of MERS-PLpro and their corresponding aligned residues in the active sites of SARS-PLpro and three bat coronaviral PLpro enzymes.

All residues involved in inhibitor binding by SARS-PLpro and their corresponding residues in MERS-PLpro are compared to three bat corona viruses because of the similarities of MERS-CoV with bat corona viruses BtCoV-HKU4 and BtCoV-HKU5. The residues compared include the active site, catalytic triad, and BL2 loop residues (Figure 4E). Two main interactions between the SARS-PLpro and its inhibitors are H-bonds and hydrophobic interactions. Three residues (D165, Y269, and Q270 shown in red in Figure 4E) form H-bonds with inhibitors. Aspartic acid (D165) is conserved in MERS-PLpro, whereas Y269 and Q270 are replaced by T274 and A275, respectively. A hydrophobic pocket is formed by two tyrosines (Y265 and Y269), two prolines (P248 and P249), and a threonine (T302) (Figure 4B). The aromatic rings of these two tyrosine residues and side chains of the latter three residues (P248, P249, and T302) make hydrophobic interactions with the naphthyl rings of both scaffold inhibitors. In the case of MERS-PLpro, two tyrosines (Y265 and Y269) are replaced by F269 and T274, removing part of the aromatic ring interactions. A threonine residue is present in place of one of the prolines (P248), which may also disturb the hydrophobic environment. Another important residue, E168, that interacts with the substrate is replaced by R168 in MERS-PLpro, making electrostatic potential less favorable for the substrate of MERS-PLpro. Together, it appears that BL2 inhibitor recognition specificity of MERS-PLpro differs significantly from that of SARS-CoV PLpro.

High-throughput screening (HTS) and hit validation

To search for MERS-PLpro inhibitors, HTS was performed with a 25,000-compound Life Chemicals antimicrobial/antiviral focused library against both SARS-PLpro and MERS-PLpro. We performed a thorough pre-screen assay optimization to determine the optimal substrate and enzyme concentrations, DMSO tolerance, reducing agent effect, additives (BSA or Triton X-100) effect, and enzyme stability at room temperature. The overall screening and hit validation process are described in Figure 5A. The primary HTS screen against SARS-PLpro was performed in duplicate, generating average Z’-factors of 0.64 ± 0.08. The MERS-PLpro primary screen was done in a single pass with Z’-factors of 0.65 ± 0.11. HTS hits with over 50% inhibition at 50 µM compound concentration were cherry picked and reanalyzed by a continuous kinetic assay to filter out false positives. The enzyme omission assay with exactly the same assay conditions, but without PLpro enzyme, was performed to remove fluorescence signal interfering compounds. Confirmed hits were repurchased, and their inhibitory activities (IC₅₀ values) were determined from full inhibition curves.

Figure 5. HTS results from Life Chemicals antimicrobial/antiviral focused library and hit validation.

(A) Schematic of HTS with 25,000 Life Chemicals compounds and hit validation process. (B) Bar graphs of IC₅₀ values and the dissociation equilibrium constants (K_D) of six hit compounds determined by fluorescence-based enzymatic assay and Surface Plasmon Resonance (SPR), respectively. All data were normalized for immobilization levels of target proteins and reference. Bars that reach the top of the graph represent either IC₅₀ or K_D values of over 200 µM (no inhibition or no binding).

Of ~25,000 compounds, four (compounds 1–4 in Figure 5B) and three (compounds 4–6 in Figure 5B) exhibited inhibitory activity with IC₅₀ values below 50 µM for SARS-PLpro and MERS-PLpro, respectively. Surface plasmon resonance (SPR) was used as a secondary orthogonal binding assay to eliminate false positives from the primary hits since our primary screen was done by a fluorescence-based enzymatic assay. Binding affinity of each hit compound can be determined by measuring the dissociation equilibrium constant (K_D) using SPR, and IC₅₀ and K_D values for the six hit compounds are compared in Figure 5B. Compound 2 from the four SARS-PLpro hits did not bind to the enzyme, indicating that it is a false positive, while the remaining three were confirmed to be binders with K_D values below 50 µM. The binding affinities of these validated hits varied from 26.3 µM to 39.9 µM and their corresponding IC₅₀ values varied from 10.9 – 31.4 µM. The IC₅₀ and K_D values of 3 are 31.4 µM and 39.9 µM, respectively, which are similar. The K_D value of 4 (26.3 µM) is approximately 2.4-fold greater than its IC₅₀ value (10.9 µM). Of the three MERS-PLpro hits, only 4 showed specific binding to the enzyme, with 18.4 µM binding affinity, while 5 and 6 were false positives. Compound 5 did not bind to MERS-PLpro at all, whereas compound 6 bound nonspecifically. These three distinct binding patterns determined by SPR: specific, no binding, and nonspecific interactions, are shown in Supplementary Figure S4A – S4C. Interestingly, compound 4 showed similar strength of inhibitory activities and binding affinity against both SARS-PLpro and MERS-PLpro enzymes.

Mechanism of inhibition and selectivity

There have been two types of noncovalent small molecule inhibitor scaffolds against SARS-PLpro previously discovered by our research group. Mechanism of inhibition studies revealed that these compounds are mixed-type inhibitors with α values greater than 1, indicating that they bind to an allosteric site other than its catalytic site, but behave as if they are competitive inhibitors ³⁵. As noted above, these lead inhibitors bind to the flexible BL2 region and induce conformational changes to block substrate access to the catalytic site of the enzyme ^{27, 28}. There have been no MERS-PLpro inhibitors published to date; we have identified one compound (4) that inhibits both SARS-PLpro and MERS-PLpro. Our mode of inhibition studies with compound 4 were done with a series of increasing substrate concentrations and enzyme-compound complexes (Figure 6A and 6B). The kinetic data were fit to four different enzyme inhibition models (competitive, non-competitive, uncompetitive, and mixed-type) using the Sigmaplot Enzyme Kinetics Module. The best fit equation was selected based on Akaike Information Criterion-corrected (AICc) values.³⁶ The equation with the lowest AICc value corresponds to the best fit, and a minimum of 2 AICc unit difference from the next lowest is required to be considered statistically significant. Interestingly, compound 4 exhibited mixed-type inhibition for SARS-PLpro, but competitive inhibition for MERS-PLpro for the same substrate. The second best fit equation of compound 4 with SARS-PLpro was noncompetitive inhibition with 7.1 AICc values lower than the best fit (mixed-type) inhibition which was also 38.6 AICc values lower than competitive inhibition, clearly indicating that compound 4 is an allosteric inhibitor of SARS-PLpro. For MERS-PLpro, the AICc value of competitive inhibition was 19.4 units lower than the next lowest, noncompetitive inhibition. Because of the large AICc value differences from the next best fit equations for both SARS and MERS-PLpro, it is clear that the same compound acts as an allosteric inhibitor for SARS-PLpro and acts as a competitive inhibitor for MERS-PLpro. Therefore, this compound inhibits the two PLpro enzymes via two different inhibitory mechanisms even though the inhibitory activities are similar, with IC₅₀ values of 10.9 µM (SARS-PLpro) and 6.2 µM (MERS-PLpro). The K_i values of compound 4 against SARS-PLpro and MERS-PLpro are 11.5 µM and 7.6 µM, respectively (Figure 6C).

Figure 6. Mechanism of inhibition.

Dixon plots of compound 4 against SARS-PLpro (A) and MERS-PLpro (B). (C) Summary table of kinetic mode of inhibition of compound 4. Mechanism of enzyme inhibition of compound 4 was determined to be a mixed inhibition for SARS-PLpro and a competitive inhibition for MERS-PLpro. Determined K_i values of compound 4 were 11.5 µM and 7.5 µM for SARS-PLpro and MERS-PLpro, respectively. (D) IC₅₀ value comparison of four SARS-PLpro lead inhibitors in combination with the newly identified compound 4 to determine if they inhibit synergistically. (E) Bar graphs of the dissociation equilibrium constants (K_D) of compound 4 in the absence (solid bars) and in the presence (striped bars) of substrate determined by Surface Plasmon Resonance (SPR).

Compound 4 appears to interact with MERS-PLpro by binding to the catalytic site since it is a competitive inhibitor with respect to the substrate. However, we were uncertain where 4 might bind to SARS-PLpro since it is an allosteric inhibitor. Our first hypothesis was that it may bind to the BL2 loop where two other SARS-PLpro lead inhibitors (I-2 and I-3) bound. We thus evaluated inhibition by 4 in the presence of each of the four SARS-PLpro lead inhibitors (I-1 – I-4) in order to see if it has an additive effect (Figure 6D). If 4 binds to the same location as I-2 and I-3, their inhibitory activities (IC₅₀ values) would not be improved by the addition of compound 4. But their IC₅₀ values should be enhanced if compound 4 binds elsewhere. Inhibitory activities of the four lead inhibitors alone varied from 0.23 µM to 2.26 µM, and their IC₅₀ values were enhanced up to almost 7-fold (0.034 µM – 0.67 µM) in the presence of 4 at a concentration of 10 µM, that is slightly lower than its IC₅₀ value. This led us to conclude that 4 binds to an allosteric site of the SARS-PLpro other than the BL2 loop. In addition to enzymatic mode of inhibition analysis, competition SPR studies of compound 4 with SARS-PLpro and MERS-PLpro were each performed in the presence and in the absence of the substrate. Binding affinity of compound 4 to SARS-PLpro was the same regardless of substrate presence, whereas that of compound 4 to MERS-PLpro was 4.5-fold weaker in the presence of the substrate than 4 alone. These results indicate that the substrate is competing with 4 for the same binding site in MERS-PLpro (Figure 6E). Therefore, both enzymatic mechanism of inhibition and SPR studies support compound 4 being an allosteric inhibitor for SARS-PLpro, while 4 is a competitive inhibitor for MERS-PLpro.

A concern of potential non-specificity was raised due to the fact that our newly identified dual inhibitor is a small fragment-like compound, and also exhibited inhibitory activity against SARS-CoV 3CLpro with an IC₅₀ value of 13.9 µM.³⁷ X-ray crystallography and mode of inhibition studies showed that 4 binds to the dimer interface of the SARS-CoV 3CLpro, inhibiting its enzyme activity by breaking the dimer since SARS-CoV 3CLpro is a functional dimer (unpublished data). Therefore, compound 4 acts as an allosteric inhibitor against both SARS-CoV proteases (3CLpro and PLpro), whereas it acts as a competitive inhibitor against MERS-PLpro. We further investigated the specificity of 4 and a lead inhibitor I-3 (control) against two human cysteine proteases also called human ubiquitin C-terminal hydrolases (hUCH-L1 and hUCH-L3) and two unrelated enzymes (Hepatitis C Virus NS3 serine protease and Bacillus anthracis dihydroorotase). The hUCH-L1 is one of the human homologues most closely related to PLpro, which makes it an excellent control to test selectivity of a newly identified inhibitor. Structural alignment of these two human homologues revealed that their catalytic triads are very similar (Figure 7A). Compound 4 was selective for SARS-PLpro and MERS-PLpro proteases over the two human cysteine proteases and both unrelated enzymes(Figure 7B).

Figure 7. Selectivity of compound 4.

(A) Structural alignment of MERS-PLpro with two human deubiquitinating enzymes. The aligned catalytic triads of two human ubiquitin C-terminal hydrolases, hUCH-L1 (green, PDB: 2ETL)³⁸ and hUCH-L3 (orange, PDB: 1UCH)³⁹, are shown with that of MERS-PLpro (tan, PDB: 4RNA) in the expanded box. (B) Selectivity of the confirmed hit compound 4. In addition to two human cysteine proteases (hUCH-L1 and hUCH-L3), two unrelated enzymes, Hepatitis C Virus NS3 serine protease (NS3) and Bacillus anthracis dihydroorotase (PyrC), were also tested along with both PLpro enzymes.

CONCLUSION

SARS-CoV and MERS-CoV cause contagious and highly virulent infectious diseases in humans, threatening the public health.^{10, 29} Both coronaviruses apparently originated from animal reservoirs such as bats or camels, but surprisingly have rapidly evolved to cause human-to-human transmission, although limited cases have been reported for MERS-CoV.¹¹ SARS-CoV has been contained by public health measures since 2003, but MERS-CoV has spread into twelve different countries so far, and the numbers of infections continues to rise. There is currently no specific treatment or vaccine available.

In this study, we determined the complete high resolution (1.8 Å) x-ray crystal structure of MERS-PLpro, a multi-functional enzyme with protease, deubiquitinating, and de-ISGylating activities. The overall MERS-PLpro structure is similar to that of SARS-PLpro including the N-terminal Ubl-domain. It was surprising to discover that SARS-PLpro has deubiquitinating function,²¹ and now it has been shown that MERS-PLpro also exhibits the same function.^{22, 23} We determined the catalytic activity of MERS-PLpro in direct comparison with that of SARS-PLpro. The catalytic efficiency (k_cat/K_M) of SARS-PLpro was ~45-fold higher (8.2 × 10⁵ M⁻¹s⁻¹) than that of MERS-PLpro (1.9 × 10⁴ M⁻¹s⁻¹). Although the deubiquitinating activity of MERS-PLpro is lower than SARS-PLpro, it is still much more active than two closely related human homologues of PLpro, herpes-associated ubiquitin-specific protease (HAUSP) and ubiquitin-specific protease 14 (USP14) which exhibit catalytic efficiencies of 2.2 × 10³ M⁻¹s⁻¹ and 107 M⁻¹s⁻¹, respectively.⁴⁰

The MERS-PLpro x-ray structure revealed crucial structural information and insights for developing inhibitors against PLpro. The flexible BL2 loop of MERS-PLpro differs significantly from that of SARS-PLpro, resulting in critically differing roles in inhibitor binding. Our structure explains the observation that all of the tested SARS-PLpro lead inhibitors were ineffective against MERS-PLpro. We performed HTS of 25,000 compounds against both PLpro enzymes and identified a dual non-covalent inhibitor that was active against both PLpro enzymes. Interestingly, this inhibitor was determined to be a competitive inhibitor against MERS-PLpro, whereas it was an allosteric inhibitor against SARS-PLpro. These results suggest that inhibitor recognition specificity of MERS-PLpro may be different from that of SARS-PLpro even though the overall structures of the whole protein and the catalytic sites are very similar. The most probable contributing factor for inhibitor selectivity of these two PLpro enzymes could be attributed to the structural differences of the BL2 loop.

MATERIALS AND METHODS

Details about cloning, expression, and purification, crystallization, confirmation assay and IC₅₀ value determination by dose response curve, and reversibility of inhibition are provided in SI Materials and Methods.

X-Ray data collection, Processing and Structure solvation

Data were collected at the LS-CAT end station 21-ID-F at the Advanced Photon Source, Argonne National Laboratory, using a wavelength of λ= 0.97872 Å, and the crystal at 100K under a dry liquid nitrogen stream. Data were recorded by a MAR CCD 225mm detector with an oscillation angle of 1.0° using a total of 190 frames. Data were processed and scaled by XDS.⁴¹

The crystal space group belonged to C2, containing one monomer in the asymmetric unit. The Matthews coefficient (V_M) was calculated as 2.5 and solvent content was estimated to be 50%. Molecular replacement was carried out using Phaser⁴² from the CCP4 package. The SARS-PLpro crystal structure (2FE8)²¹ was used as a search model. The zinc binding domain in the initial model was truncated and manually rebuilt by Coot.⁴³ Structural refinement was conducted using Refmac5.5.⁴⁴

Primary high-throughput screening

The 25,000-compound Life Chemicals library was screened against the two PLpro cysteine proteases from SARS-CoV and MERS-CoV. All assays against SARS-PLpro were done in duplicate and against MERS-PLpro were done in single pass in black 384-well plates (Matrix Technologies). The SARS-PLpro enzyme (20 nM final concentration) was prepared in assay buffer (50 mM HEPES, pH 7.5, 0.01% Triton X-100 (v/v), 0.1 mg mL⁻¹ BSA, and 2 mM GSH). The MERS-PLpro enzyme (400 nM final concentration) was prepared in the same assay buffer with 5 mM DTT in place of 2 mM GSH. 30 µL of enzyme solution was dispensed into wells, and then 200 nL of 10 mM compounds (50 µM final concentrations) were added and incubated for 5 minutes. Enzyme reactions were initiated with 10 µL of substrate Z-Arg-Leu-Arg-Gly-Gly-AMC (Bachem Bioscience) (50 µM and 75 µM for SARS- and MERS-PLpro, respectively) dissolved in assay buffer and incubated for 6 minutes, followed by adding 10 µL of 10% SDS (w/v) as a stop solution. Fluorescence intensity was monitored at 360 nm (excitation) and 450 nm (emission).

Determination of dissociation equilibrium constant (K_D) by SPR

Compound solutions with a series of increasing concentrations (0–200 µM at 1.5-fold dilution) were applied to all four channels at a 30 µL/min flow rate. Sensorgrams were analyzed using the Biacore T200 evaluation software 2.0, and response units were measured during the equilibration phase at each concentration. Each PLpro enzyme was immobilized on a CM5 sensor chip using standard amine-coupling with running buffer HBS-P (10 mM HEPES, 150 mM NaCl, 0.05% surfactant P-20, pH 7.4) using a Biacore T200 instrument. MERS-PLpro enzyme was immobilized to flow channels 2 and 3, and immobilization levels of flow channels 2 and 3 were ~16,900 RU and ~16,700 RU, respectively. SARS-PLpro was immobilized to flow channel 4 at the immobilization level of ~14,600 RU to be compared with MERS-PLpro. Data were referenced with blank (enthanolamine) RU values. SigmaPlot 12.0 was used to fit the data to a single rectangular hyperbolic curve to determine K_D values. The hyperbolic, y = y_max·x/(K_D + x), was used to plot response units and corresponding concentration, where y is the response, y_max is the maximum response and x is the compound concentration.

Mechanism of inhibition

Enzyme activities of both MERS-PLpro and SARS-PLpro were monitored in the same way as the primary screen with varying concentration of inhibitors and substrate (0–300 µM). The concentration of compounds was varied from 0 to at least 10X the IC₅₀ value of each compound. The data were fit to four equations (shown in SI Materials and Methods) using SigmaPlot Enzyme Kinetics Module 1.3 in order to determine the best fit inhibition mechanism and kinetic parameters for each compound.

Inhibitor selectivity assay

To test for selectivity, two human ubiquitin C-terminal hydrolases (UCH-L1 and UCH-L3) and two unrelated enzymes (Hepatitis C Virus NS3 serine protease and B. anthracis dihydroorotase) were tested with the top hit compound from HTS and a lead SARS-PLpro inhibitor (I-1) using a fluorometric assay. The fluorogenic substrates used in this study was ubiquitin-AMC (Boston Biochem). All assays were performed in 384-well black plates (Corning) in a total volume of 24 µL of assay buffer containing 50 mM HEPES (pH 7.5), 5 mM DTT, 0.1 mg mL⁻¹ BSA, 0.01% Triton X-100 (v/v) in triplicate. A series of compound concentrations (0 to 200 µM final concentration at 2-fold serial dilution) in 100% DMSO was prepared in a 384-well plate. Then 3X compound solutions were prepared in assay buffer prior to assays. 8 µL of each enzyme solution was distributed into wells, and 8 µL of varying concentration of compounds were added and incubated for 10 minutes. The enzyme reaction was initiated by adding 8 µL of the substrate (50 µM final concentration), and fluorescence intensity was continuously monitored at excitation/emission wavelengths of 350 nm/460 nm for 10 minutes.

Abbreviations

MERS-CoV

Middle East Respiratory Syndrome coronavirus

HCoV-EMC

human coronavirus-Erasmus Medical Center

SARS-CoV

Severe Acute Respiratory Syndrome

3CLpro

3C-like protease

PLpro

papain-like protease

nsp

non-structural protein

CFR

case-fatality rate

BL2

blocking loop 2

HTS

high-throughput screening

SPR

surface plasmon resonance

hUCH

human ubiquitin C-terminal hydrolases