Why is the Omicron main protease of SARS-CoV-2 less stable than its wild-type counterpart? A crystallographic, biophysical, and theoretical study

Abstract

During the continuing evolution of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the Omicron variant of concern emerged in the second half of 2021 and has been dominant since November of that year. Along with its sublineages, it has maintained a prominent role ever since. The Nsp5 main protease (M^pro) of the Omicron virus is characterized by a single dominant mutation, P132H. Here we determined the X-ray crystal structures of the P132H mutant (or O-M^pro) as a free enzyme and in complex with the M^pro inhibitor, the alpha-ketoamide 13b-K, and we conducted enzymological, biophysical, as well as theoretical studies to characterize the O-M^pro. We found that O-M^pro has a similar overall structure and binding with 13b-K; however, it displays lower enzymatic activity and lower thermal stability compared to the WT-M^pro (with “WT” referring to the prototype strain). Intriguingly, the imidazole ring of His132 and the carboxylate plane of Glu240 are in a stacked configuration in the X-ray structures determined here. Empirical folding free energy calculations suggest that the O-M^pro dimer is destabilized relative to the WT-M^pro due to less favorable van der Waals interactions and backbone conformations in the individual protomers. All-atom continuous constant-pH molecular dynamics (MD) simulations reveal that His132 and Glu240 display coupled titration. At pH 7, His132 is predominantly neutral and in a stacked configuration with respect to Glu240 which is charged. In order to examine whether the Omicron mutation eases the emergence of further M^pro mutations, we also analyzed the P132H+T169S double mutant, which is characteristic of the BA.1.1.2 lineage. However, we found little evidence of a correlation between the two mutation sites.

Graphical Abstract

INTRODUCTION

In December 2019, the severe acute respiratory syndrome outbreak caused by a novel coronavirus initiated a global epidemic of hitherto unknown dimensions. The new coronavirus is ~82% identical to severe acute respiratory syndrome coronavirus (SARS-CoV) at the amino-acid level; hence it is referred to as SARS-CoV-2 [1,2]. Globally (as of August 2023), there have been over 770 million confirmed cases of SARS-CoV-2 infection and 6.9 million deaths [3]. In November 2021, the lineage BA.1 (earlier B.1.1.529 [4]), the Omicron variant, arose in South Africa and Botswana, and since then, new sublineages are springing off from this Omicron variant [5,6].

In the host cell, the viral proteins are translated from the viral RNA into two huge polyproteins (pp1a, pp1ab), which are processed into 16 nonstructural proteins (Nsps) by the main protease (M^pro) and the papain-like protease (PL^pro) [7]. The M^pro cleaves the polyproteins at 11 sites that contain the recognition sequence Leu (in one case: Phe)- Gln↓(Ser, Ala, in one case: His) (↓ marks the cleavage site).

M^pro is a homodimeric cysteine protease, and each monomer comprises three domains. Domains I and II (residues 10 to 99 and 100 to 182, respectively) consist of β-barrels. A cleft between the two domains harbors several subsites (S1’ and S1–S5) that provide the binding site for the substrate. Unlike most proteases of the host, the S1 pocket has a substrate preference for glutamine, which makes M^pro an attractive antiviral drug target [8]. Finally, domain III (residues 198 to 303) is a cluster of five α-helices crucial for regulating the dimer formation of the enzyme. Dimerization of the M^pro monomer is necessary for the catalytic activity as it allows the interaction between the N-finger (residues 1–9) of one protomer with Glu166 of the second protomer. The N-terminus/Glu166 interaction is essential for forming the S1 subsite [9].

According to GISAID [10], the Omicron variant started replacing the Delta variant around December 2020. By the end of 2021, it became the dominant strain globally, and its evolution is still ongoing, with several main lineages (BA.1, BA.2, BA.3, BA.4, BA.5) and their sublineages (e.g., BA.2.75, BQ.1, XBB, XBB.1.5, XBB.1.9, XBB.1.16, EG.5.1, HK.3, JN.1) [10,11]. The omicron variant accumulated up to 50 mutations throughout the genome [11]. One of the mutations of concern, P132H, is in the Nsp5, affecting the main protease. The P132H mutation maintains extremely high frequencies in all Omicron lineages (over 99.8%) [12], and the reason this mutation has become dominant remains unexplained. The other mutation site considered here, Thr169, is a mutation hot spot of the M^pro. Its codon is ACU in the prototype strain of SARS-CoV-2 but ACG, ACC, or ACA, all coding for Thr, are also found at this site in different M^pro sequences (GISAID [10]). In the T169S mutant of Omicron, the codon is UCU; therefore, the T169S mutation is the result of an A>U transversion. The P132H+T169S double mutant emerged more or less in parallel to or slightly after the P132H single mutation (there are only very few M^pro [< 110] sequences that have the T169S mutation but lack the main O-M^pro mutation, P132H; all data extracted from GISAID [10]). However, the P132H+T169S double mutant never reached the frequency of P132H. At its peak of occurrence in January 2022, there were 34,931 cases of the double mutation collected in one month, and almost all of them belonged to the lineage BA.1.1.2 (which was most prevalent in Japan) [10,13,14]. At this time, the double mutant comprised ~2.6% of all Omicron M^pro sequences and was the most common M^pro mutation after P132H. Following its peak occurrence, the P132H+T169S double mutant started to decline along with the entire lineage BA.1.1.2. Its last occurrence to date was in a BA.5-like virus in Hong Kong, China in August 2023. We include the P132H+T169S double mutant in our considerations here because as a short-lived phenomenon, it forms an interesting contrast to the omnipresent P132H mutant, which became the most common M^pro mutation after late 2021.

In total, there are so far 6 publications [15–20] that describe crystal structures of complexes between the P132H mutant (O-M^pro) and different inhibitors (Table S3). Among these, 5 papers very briefly present the O-M^pro structure along with the crystal structures of other mutants. Only the paper by Sacco et al. exclusively deals with the O-M^pro structure, in complex with the inhibitor GC-376 [17]. However, the reason for the reduced thermal stability of the O-M^pro is still unknown; it has only been speculated that minor residue adjustments around the bulkier His132 side chain might be the cause [17].

The P132H mutation site is located at the interface of domains II/III within the same protomer, ~21 Å away from the active site, and also far away (over 12 Å) from the other monomer within the M^pro dimer. Several studies suggest that residue 132 participates in forming the so-called distal site, along with Arg131, Thr196, Asp197, Thr198, Asn238, Tyr239, Leu287, and Asp289, which has been proposed to be an allosteric site [21,22]. Another study proposes a long-distance dynamic communication between the distal and active sites that might influence the catalytic activity [23].

Subtle structural changes are important because the concern is that when they accompany single-site mutations, they may ease the emergence of secondary mutations that could reduce the affinity of the few approved inhibitors (nirmatrelvir, ensitrelvir) for the M^pro and thereby cause resistance towards these drugs. Therefore, we investigate the effect of the mutation, P132H, as well as of the double mutation, P132H+T169S, on the binding of our M^pro inhibitor 13b-K (Figure 1) [9,24] using X-ray crystallography.

Figure 1. Structure of the alpha-ketoamide 13b-K.

In addition, we perform molecular dynamics (MD) simulations to understand the possible effects of the P132H mutation on the conformation and structural stability of the M^pro. We further investigate the residues most influenced by the P132H mutation, near or at the active site, that might be subject to future potential accumulated mutation, as possibly exemplified by T169S. One focus of our study is the protonation state of His132 and its potential hydrogen bonding with Glu240.

RESULTS AND DISCUSSION

Enzymatic Activity and Thermal Stability of the Omicron M^pro

A well-established Förster resonance energy transfer (FRET) assay using a fluorescent substrate, which harbors the cleavage site (indicated by the arrow ↓) of SARS-CoV-2 M^pro (Dabcyl-KTSAVLQ↓SGFRKM-E[Edans]-NH₂; Biosyntan), was used to evaluate the enzymatic activity of WT-M^pro, the P132H mutant (O-M^pro), and the P132H+T169S double mutant. The kinetics of peptide cleavage by the O-M^pros is considerably slower compared to that of the wild-type enzyme. We determined the k_cat/K_m for the wild-type enzyme as 6752±1304 M⁻¹·s⁻¹. In our hands, the k_cat/K_m for the O-M^pro is 4915±1278 M⁻¹·s⁻¹, i.e., 27% lower. This difference is mainly due to k_cat, which is 0.58 s⁻¹ for the WT and 0.40 s⁻¹ for the O-M^pro, indicating that the mutation does not appreciably affect the binding of the substrate but the velocity of its conversion. k_cat/K_m is nearly the same for the P132H+T169S double mutant, but with slightly lower k_cat (0.35 s⁻¹), compensated by a reduction in K_m (72 μM) (Figure 2A, Table 1). These results are largely in agreement with data presented by Greasley et al. [16] and Chen et al. [12], both of whom report a k_cat/K_m reduction by ~34% for the P132H mutant, but at variance with Sacco et al. (~9% reduction in k_cat/K_m) [17], Lin et al. (~3% reduction) [20], and Ullrich et al. (~43% increase) [25]. These variations are probably caused by differences in the temperature of the measurements and by the usage of different FRET substrates.

Figure 2. Biochemical and empirical stability characterization of O-M^pro and P132H+T169S double mutant.

(A) The Michaelis–Menten kinetics of WT-M^pro, O-M^pro, and P132H+T169S double mutant. (B) Melting temperature (T_m) measurements using NanoDSF for WT-M^pro (red lines), O-M^pro (blue lines), and P132H+T169S double mutant (green lines). The measurements were repeated using the same assay to determine the T_m after incubation with the 13b-K inhibitor, with identical color coding. (C) Folding free energy difference (ΔΔG_fold) between the WT-M^pro and O-M^pro dimers (solid), and monomers (striped) calculated from the Rosetta empirical energy functions. Positive ΔΔG_fold indicates decreased stability. The total ΔΔG_fold is broken down into van der Waals (VdW), torsion angle (Torsion), hydrogen bonds (H. bonds), electrostatic interactions (Elect.), solvation free energy (Solvation), and other (Other) terms. Monomer terms are the sums of two monomer energies. (D, E, F) The inhibition of O-M^pro (D), WT-M^pro (E), and P132H+T169S double mutant (F) by 13b-K.

Table 1.

Summary of the kinetic parameters and melting temperatures of the M^pro-WT, O-M^pro, and P132H+T169S double mutants

Mpro	kcat (s−1)	Km (μM)	kcat/Km (M−1·s−1)	Activity (%)	Tm of Mpro	Tm of Mpro with 13b-K
WT	0.58±0.04	85.11±14.06	6752.05±1303.87	100.0	57.44±0.03	68.41±0.12
O-Mpro	0.40±0.04	81.45±19.55	4915.08±1278.03	72.8	55.33±0.03	66.41±0.07
P132H + T169S	0.35±0.04	71.91±24.59	4927.46±1776.58	73.0	55.88±0.05	65.82±0.10

The inhibitory activities of our ɑ-ketoamide 13b-K [9,24] versus SARS-CoV-2 O-M^pro, the double mutant P132H+ T169S, and WT-M^pro were determined to assess its activity against these enzyme species. The half-maximal inhibitory concentrations (IC₅₀) of 13b-K against WT-M^pro and the two O-M^pros were very similar, 0.164±0.030, 0.171±0.034, and 0.209±0.040 μM, respectively (Figure 2D–2F). No substantial differences can be observed in terms of the potency of inhibition by 13b-K. In line with this observation, nirmatrelvir (PF-07321332) was reported to have a similar inhibition potency for WT-M^pro and O-M^pro [15,16,20].

We also assessed the thermal stability of the O-M^pro mutant and that of the P132H+T169S double mutant by nano-differential scanning fluorometry (nano-DSF). In agreement with earlier reports [12,17], we find that the “melting temperature” of the O-M^pro is reduced by 2.1 °C, compared to the WT enzyme. The double mutant, P132H+T169S, appears to be slightly more stable than O-M^pro (difference: 0.55 °C). Coincubation of each enzyme sample with our inhibitor 13b-K increases the melting temperature by 10–11 °C, but the differences seen between the various forms of the free enzyme (WT, P132H, P132H+T169S) remain the same in the inhibitor complexes (Figure 2B).

Structural Analysis of the O-M^pro and Double-mutant Free Enzymes and Their Complexes with 13b-K at Different pH Values

The crystal structure of the free O-M^pro enzyme (as crystallized at pH 8.2) was determined at 1.91 Å resolution in monoclinic space group C2 with one M^pro chain (half an M^pro dimer) in the asymmetric unit. The complex between O-M^pro and 13b-K was crystallized at two different pH values (8.5, 6.5) in order to visualize pH-dependent local differences in the interaction between Ser1* of one O-M^pro protomer and Glu166 of the other protomer (to emphasize the intermolecular character of this interaction, Ser1* will be marked by an asterisk in what follows). Crystals grew in triclinic space group P1 with four O-M^pro chains per asymmetric unit and diffracted X-rays to 2.30 Å (pH 8.5) and 2.48 Å (pH 6.5). In addition, the free enzyme of the double mutant P132H+T169S and its complex with our inhibitor 13b-K was crystallized (at pH 8.0 and 7.5) in space groups C2 and P2₁, respectively, with half an M^pro dimer and a complete M^pro dimer in the asymmetric unit. These crystals diffracted X-rays to 1.8 Å and 1.7 Å, respectively. All structures were refined to good stereochemistry and yielded reasonable statistics (see Table S1). Structure factors and atomic coordinates have been deposited in the Protein Data Bank with accession codes 8R19, 8R26, 8R0V, 8R24, and 8R1Q.

The overall structure of the O-M^pro enzymes is very similar to the WT-M^pro (Figure 3A), albeit small changes can be observed near the mutation site. Close to His132, there is a salt bridge between Asp289 and Arg131. This salt bridge has been suggested to be essential for the intramolecular domain II/domain III interface in SARS-CoV M^pro [26]. Both WT-M^pro and O-M^pro feature this salt bridge, although the presence of His132, which potentially has a charged side chain due to its close proximity to Glu240 (see below), might influence it directly or indirectly and reduce the protein stability (Figure 3B). In addition, the three-dimensional structure of the free enzyme of the P132H+T169S double mutant is largely unchanged compared to the O-M^pro. This mutation has almost no structural effect, as the residue is fully solvent-exposed on the surface of the protein. In contrast, in the structure of the complex between the double mutant and our inhibitor 13b-K, the loop 167–170 has a different conformation in protomer A. In this structure, there is a flip of the Pro168 peptide bond relative to the other structures and the ψ angle of Pro168 is 148°, whereas in the free enzyme, it is −29°, very similar to that in the wild-type M^pro. The Pro168 conformation is less clear in protomer B due to the poorly resolved electron density of the loop. In summary, however, there are no substantial structural changes in any of the X-ray structures, neither in the P132H mutant nor in the P132H+T169S double mutant. Based on these data, and on the lack of coupling between His132 and Ser169, the double mutant P132H+T169S is unlikely to drive further mutation of the Omicron Nsp5 sequence.

Figure 3. X-ray structure of the free O-M^pro and P132H+T169S double mutant in comparison to the free WT-M^pro.

(A) Overview of WT-M^pro (wheat), O-M^pro (blue), and P132H+T169S M^pro (green). The active site (yellow ball) is shown in panel A, the mutation sites P132H (blue ball) and T169S (red ball) are also indicated. (B, C) A closer view of the O-M^pro mutation includes the 2F_o-F_c map contoured at 1.2 σ at His132 (mutation site) and Glu240. (B) Shows the close proximity of the mutation site from the Glu289-Arg131 salt bridge. (C) Shows the rotation of Thr198 and the movement of Asp197, with the WT-M^pro depicted in gray.

Regarding the destabilization of the M^pro by the P132H mutation, it has to be considered that His132 occupies a larger space than its ancestor, Pro132, and consequently, the side chain of Glu240 is shifted to accommodate the newcomer (Figure 3C). Notably, the water-mediated hydrogen bond between His132 and Glu240 observed in the structure published by Sacco et al. [17] is absent in our O-M^pro and P132H+T169S structures, as well as in all free O-M^pro enzyme structures in other studies [12,20]. If His132 was charged, it would likely influence the hydrogen-bonding network in the vicinity. Based on this reasoning and the His132-Glu240 distance, we expected to observe an “edge-to-edge” configuration between the two side chains, in which the imidazole donates a charged hydrogen bond to one of the carboxylate oxygens. Surprisingly, all five X-ray structures, including the free enzyme of both O-M^pro and the double mutant, as well as 13b-K-bound forms obtained at different pH conditions, show a “stacking” configuration between the His132 and Glu240 side chains, hindering the formation of an H-bond in both protomers. This stacking configuration has also been observed in the X-ray structure of O-M^pro in complex with the inhibitor GC376 (PDB accession code 7TOB [17]). On the other hand, the X-ray structure of O-M^pro in complex with nirmatrelvir (PDB accession code 7TLL [16]) shows an edge-to-edge configuration, with an imidazole-to-carboxylate distance of 3.2 and 3.8 Å for protomer A and B, respectively, indicating the possibility of forming an H-bond. To provide a mechanistic understanding of the unexpected relative orientations of the His132 and Glu240 side chains and the mutation-induced destabilization, we performed constant-pH and fixed-charge MD simulations as well as empirical protein stability calculations.

Constant-pH MD Simulations Reveal that His132 is Neutral and Prefers the Stacking Configuration

To understand why His132 and Glu240 engage in a stacking configuration, we performed pH-titration simulations of the free O-M^pro using the newly developed graphical processing unit (GPU)-accelerated all-atom continuous constant-pH MD (CpHMD) method [27] in the Amber22 package [28]. The simulations were initiated from the pH 8.5 inhibitor-removed X-ray structure (which was determined first, PDB accession code 8R26) and made use of the asynchronous pH replica exchange algorithm for accelerated convergence [29] (Table S2). Surprisingly, the simulations show that His132 is neutral and Glu240 is charged at pH 7.0, which contradicts our initial hypothesis that His132 is charged and forms a charged H-bond (salt bridge) with Glu240. Furthermore, the titration of the two residues is coupled. To understand the coupled titration, we calculate the pH-dependent probabilities of the combined protonation states: the doubly protonated state, His132(+)/Glu240(0), the singly protonated states, His132(+)/Glu240(−) and His132(0)/Glu240(0), and the doubly deprotonated state, His132(0)/Glu240(−). Figure 4 demonstrates that at pH 7, the doubly deprotonated state His132(0)/Glu240(−) is predominant for both protomers. Furthermore, as pH decreases, the probabilities of the two singly protonated states increase; however, the His132(0)/Glu240(0) state dominates over the His132(+)/Glu240(−) state (Figure 4A–4B). Thus, the CpHMD simulations suggest that the singly protonated His132(+)/Glu240(−) pair, which would promote the edge-to-edge configuration through a charged H-bond, is an unlikely state. From the CpHMD simulations, macroscopic pK_a’s can be obtained by fitting the average number of protons bound to His132 and Glu240 at different pH to a coupled two-proton model [30,31]. The two stepwise pK_a’s are estimated as 4.0/6.1 for protomer A and 4.2/5.9 for protomer B (Figure S1). Since the His132(0)/Glu240(0) state is the more probable singly protonated state, the higher pK_a can be assigned to Glu240 (6.1 and 5.9 for protomer A and B, respectively) while the lower pK_a can be assigned to His132 (4.0 and 4.2 for protomer A and B, respectively). The 0.2-pH unit difference between the pK_a’s of the two protomers is within the statistical error of the simulations [27].

Figure 4. Protonation and configuration states of H132/E240 from the all-atom CpHMD simulations of the free O-M^pro (with inhibitor removed, PDB 8R26).

(A, B) Probabilities of the four protonation states of the H132/E240 pair at different pH in protomer A (A) and B (B). (C) Overlay of the MD snapshots of the edge-to-edge (cyan) and stacking (magenta) configuration. His132 and Glu240 side chains are shown as sticks. The normal vectors of the imidazole ring and the carboxylate plane are shown as arrows. (D) Percentage probability of forming the stacking configuration in the presence of charged His132(+) or neutral His132(0) for protomer A (blue) and B (orange).

Next, we examine the effect of the His132 protonation state on the relative His132/Glu240 side chain conformations using the normal vectors of the imidazole ring of His132 and the carboxylate plane of Glu240 (Figure 4C). The configuration is considered edge-to-edge, if the angle between the two normal vectors is between 60° and 120°, i.e., the imidazole and carboxylate planes are approximately perpendicular, and otherwise stacking (Figure 4C). Based on this definition, the CpHMD simulations demonstrate that His132/Glu240 can adopt either the edge-to-edge or the stacking configuration. The stacking configuration is observed in the X-ray structures obtained in this work and the crystal structure of the O-M^pro in complex with GC376 (PDB accession code 7TOB [17]). On the other hand, the edge-to-edge configuration is found in the X-ray structure of the O-M^pro complex with nirmatrelvir (PDB accession code 7TLL) [16]. Interestingly, the CpHMD data show that the relative His132/Glu240 configuration is dependent on the protonation state of His132. In the presence of His132(+), the probability of forming the stacking configuration is about 50%, i.e., stacking and edge-to-edge configuration are equally probable, whereas, in the presence of neutral His132(0), the probability of the stacking configuration is about 70%, i.e., stacking is preferred (Figure 4D). However, since the His132(+) state has a low probability across the simulation pH range (2.5–8), the stacking configuration is always preferred. This is in agreement with the stacking configuration being observed in both of our X-ray structures of O-M^pro in complex with 13b-K at pH 6.5 and 8.5.

Fixed-charge MD Simulations Support the Preference for the Stacking Configuration of His132 and Glu240

Although the CpHMD data demonstrate the preference for the stacking configuration, the simulation length was limited. Thus, to confirm the finding, we conducted two runs of 500-ns fixed-charge MD simulations starting from the pH 8.5 X-ray structure of the free O-M^pro (with inhibitor removed, PDB accession code 8R26) with His132 in the neutral state (Table S2). Additionally, four runs of 1-μs fixed-charge MD simulations were performed based on the published X-ray structure with the PDB accession code 7TOB [17] (inhibitor removed, Table S2), whereby His132 was either charged (two runs) or neutral (two runs). The simulations based on our X-ray structure model show that the imidazole ring of His132 and the carboxylate plane of Glu240 are predominantly stacked, as the angle between the two normal vectors (defined above) is either 30° or 150° (Figure 5A). In contrast, the edge-to-edge state is infrequently sampled, as the angle distribution plot displays a valley between 60° and 120°. Note, that the second trajectory gave very similar results (Figure S2). The simulations based on 7TOB [17] confirm that the stacking configuration is favored when His132 is neutral whereas the edge-to-edge configuration is favored when His132 is charged (Figure S3).

Figure 5. Configuration of H132/E240 and the effect on the water-mediated D197-H132 interaction from the fixed-charge MD simulations.

(A) Distributions of the angle between the normal vectors of the His132 imidazole and Glu240 carboxylate planes for protomer A (blue) and B (red). The grey area represents the edge-to-edge (EtE) configuration, while the other parts represent the stacking (S) configuration. (B) Distributions of the minimum distance between the carboxylate oxygens of Asp197 and the backbone carbonyl oxygen of His132 for the His132/Glu240 stacking (solid lines) and edge-to-edge (dotted lines) configuration. Data for protomer A and B are shown in blue and red, respectively.

MD Analysis Suggests that the Local Electrostatic and H-bond Environment of His132 is Largely Unperturbed

Our X-ray structure of the free O-M^pro shows a salt bridge between Asp289 and Arg131, which is also present in the WT-M^pro and has been hypothesized to be important for the intramolecular domain II/III interface in SARS-CoV(−1) M^pro [26]. We calculated the probability distribution of the minimum Asp289-Arg131 charge center distance based on the fixed-charged MD, which confirms that the salt bridge is stable regardless of the His132/Glu240 configuration (the most probable distance is 2.8 Å for both stacking and edge-to-edge configurations, Figures S2 and S4). This is consistent with the CpHMD finding that His132 is neutral, which does not perturb this salt bridge interaction. Another residue in the vicinity of His132 is Thr198. In our X-ray structure of the free O-M^pro, the distance from the His132 imidazole nitrogen to the Thr198 hydroxyl oxygen is 3.3 Å (Figure 3C), suggesting that they might form a dynamical H-bond in solution. To examine this possibility in light of the two different His132/Glu240 configurations, we calculated the distribution of the minimum distance from the imidazole nitrogen to the hydroxyl oxygen for the stacking and edge-to-edge configurations of the free O-M^pro (Figures S2 and S4). Although there is a minor peak at about 3.0 Å, the major peak (the most probable distance) is at about 5.5 Å for the His132/Glu240 stacking configuration. For the edge-to-edge configurations, the major peak is at about 4.5 Å. Thus, our simulations do not support an H-bond between Thr198 and His132 in O-M^pro.

MD Analysis Suggests that the His132/Glu240 Stacking Configuration Gives Space for Water to Mediate the Side Chain-to-backbone Interaction between Asp197 and His132

In our crystal structure of the free O-M^pro, the Oδ1 atom of Asp197 shifts by 2.2 Å and a water molecule appears to mediate an H-bond between the main-chain nitrogen of His132 and the carboxylate of Asp197 (Figure 3C). This water is absent in the X-ray structures of the WT-M^pro. To examine if this water-mediated interaction is possibly related to the interaction between His132 and Glu240, we calculate the minimum distance between the carboxylate oxygens of Asp197 and the backbone carbonyl oxygen of His132 for the His132/Glu240 stacking and edge-to-edge configurations based on the fixed-charge MD simulations. For both configurations, the Asp197-His132 distance has two most probable values, about 5.5 Å and 7.5 Å (two peaks in the distributions of Figure 5B); however, the relative probability of the two distances is dependent on the His132/Glu240 configuration. When His132 is in the stacking configuration with respect to Glu240, the larger distance (7.5 Å) is sampled with a much higher probability than the smaller distance (5.5 Å); in contrast, when His132 is in the edge-to-edge configuration, the two distances are sampled with equal probabilities (Figure 5B). Since a larger distance between the Asp197 side chain and the His132 backbone makes space for the entrance of water, the simulation data suggests that the stacking configuration may promote the water-mediated interaction between Asp197 and His132.

Empirical Folding Free-energy Calculations Predict that the P132H Mutation Destabilizes M^pro

To understand the mechanism of the stability decrease of O-M^pro relative to the WT, we estimate the folding free energy change ΔΔG_fold for both the dimers and monomers using the ddG_monomer application in the Rosetta software suite [32]. The calculation gives a positive ΔΔG_fold of 7.8±0.8 kcal/mol, indicating that the P132H mutation destabilizes the M^pro dimer, which is consistent with the melting temperature reduction of 2.1 °C. The destabilization is mainly driven by unfavorable van der Waals interactions and backbone torsion energies, indicating that the P132H mutation induces steric repulsion and unfavorable conformations (Figure 2C). To test if the dimer interface is involved in the stability change, we examine ΔΔG_fold of the monomers. If the sum of the monomer ΔΔG_fold, and the individual contributions are the same as the corresponding ΔΔG_fold of the dimer, the stability changes of the monomers are solely responsible for the stability change of the dimer, i.e., no change in the dimer interface plays a role. On the other hand, if ΔΔG_fold of the dimer is greater (more positive) than that of the monomers, the difference can be attributed to the destabilization of the dimer interface. Following this reasoning, we examine the differences between the dimer and monomer calculations. The sum of monomer ΔΔG_fold is 6.8±0.8 kcal/mol; the difference from the dimer ΔΔG_fold is 1.0 kcal/mol, which is within the error bar of 1.1 kcal/mol (Figure 2C). The greatest difference between the individual energy terms of the monomers and the dimer is in the electrostatic and solvation energies. The P132H mutation is predicted to have stabilizing electrostatic (−3.0±1.1 kcal/mol) and destabilizing solvation energies (4.3±1.7 kcal/mol) for the monomers, whereas these terms are nearly unchanged for the dimer. It is noteworthy that the stabilization of electrostatics and destabilization of solvation largely cancel out, making the net difference between the monomers and the dimer negligible. Taken together, these data suggest that the destabilization of O-M^pro relative to the WT is mainly due to the destabilization of the individual protomers, and although the electrostatics and solvation of the dimer interface are affected by the mutation, the net effect on the stability of the dimer is negligible.

The P132H Mutation Has No Negative Impact on 13b-K Binding

The active site of the M^pro is ~21 Å away from the discussed mutation (Figure 3A), and the mutation site is > 12 Å from the monomer–monomer interface. No substantial changes were observed between the overall structures of WT-M^pro and O-M^pro. The root mean square deviation (RMSD) values of the structures were calculated for the complete chain against the WT-M^pro (PDB 6Y2E [9]), additionally for each domain separately. With four copies (two dimers) of the O-M^pro-13b-K complex in triclinic space group P1, such calculations potentially offer interesting results regarding subtle differences between individual M^pro complex molecules (such information is not available from the common M^pro complexes crystallized in space group C2, which has a two-fold axis of symmetry). Thus,a difference in the RMSD of each chain within the model of the O-M^pro-13b-K complex at pH 8.5 is observed. Chains A and C have higher RMSD values (0.43 Å and 0.47 Å) than chain B and chain D (0.37 Å and 0.40 Å; values are for all non-hydrogen atoms). On the other hand, within the model of the O-M^pro-13b-K complex at pH 6.5, the four chains have almost identical RMSD values (0.37 Å, 0.38 Å, 0.37 Å, and 0.37 Å) for chains A, B, C, and D, respectively. For the high-pH form, calculation of RMSD values for the individual amino-acid residues (using the WT-M^pro [PDB 6Y2E] as a reference) for each of the four chains, A - D, reveals that in specific regions, i.e., in the vicinity of Ser1*, Asn27, Gln74, Pro168, or Thr279, a considerable increase in the RMSD values exists for the A/C chains but not the B/D chains (Figure S5). One of these differences is due to a different arrangement of the interaction between Ser1* and Glu166 in chains A and C, compared to B and D. In the former, the N-terminus (Ser1*) and Glu166 are connected via a salt bridge, whereas in the latter, this salt bridge is absent and there is instead an H-bond between the side-chain hydroxyl of Ser1* and the carboxylate of Glu166 (Figure 6E–6F). The reason for this feature could be that at the pH of crystallization (8.5), the free amino group of Ser1* may be partly deprotonated, as the solution pK_a of the N-terminal amino group of a protein is ~8.0–8.5. This may vary in the M^pro dimer, as the N-terminal Ser1* is only partially solvent-accessible due to its engagement with Glu166, and further modulation may be caused by the nearby His172 [33]. When the N-terminal Ser1* is deprotonated, the salt bridge with Glu166 cannot be formed and is replaced by the H-bond between Ser1* Oγ and the Glu166 carboxylate. Along with these changes in the H-bonding pattern, the side-chain torsion angles χ1 and χ2 of Glu166 are in the antiperiplanar (ap)/ap range in protomer A and in the synclinal (sc)/ap range in protomer B. This conformational change of the Glu166 side chain in protomer B enables it to form a new hydrogen bond with His172 Nϵ2. At elevated pH, each O-M^pro dimer would thus consist of one protomer (A or C) comprising the salt bridge between Ser1* and Glu166 (with Glu166 in ap/ap conformation) and one protomer (B or D) lacking the salt bridge (and Glu166 in sc/ap conformation). This interpretation is supported by the fact that the crystals of the O-M^pro/13b-K complex grown at pH 6.5 show negligible RMS differences between all four chains, and the salt bridge is fully formed in all four copies of the Ser1*-Glu166 pair. In the 13b-K complex of the double mutant, P132H+T169S, which was crystallized at pH 7.5, the situation is comparable to the high-pH form of the P132H single-mutant/13b-K complex, i.e., the dimer also comprises one protomer with a salt bridge between Ser1* and Glu166 and one having the Ser1* Oγ-Glu166 H-bond instead (Figure 6G–6H).

Figure 6. X-ray structures of O-M^pro/P132H+T169S M^pro in complex with 13b-K.

(A, B) The substrate-binding cleft of the O-M^pro crystallized in a complex with 13b-K at two different pH values, 6.5 (A) and 8.5 (B). The carbon atoms of the inhibitor are colored green, and those of the protein are colored cyan, except for Ser1*, which are in magenta. The binding pockets (S1′-S2) of the protein and the binding moieties (P1′-P4) of the inhibitor are labeled. The 2F_o-F_c map carved around the inhibitor is depicted as mesh and contoured at 1.0 σ. (C–F) Interactions of the N-terminus in the two protomers of the same dimer of the pH 6.5 crystal (C and D) and of the pH 8.5 crystal (E and F). Red arrows indicate the difference between the two protomers in the latter case. The represented maps are polder maps calculated in phenix.polder [36], by omitting the complete residues of interest, Glu166 and Ser1*, contoured at 3.0 σ. (G, H) The substrate-binding cleft of the P132H+T169S double mutant co-crystal structure with 13b-K, including protomer A (G) and protomer B (H).

The discussed interaction between Ser1* and Glu166 (a salt bridge at pH 6.5 and either a salt bridge or an H-bond at pH 8.5) is crucial for forming the S1 pocket (Figure 6A–6B, and 6E–6F). We have previously observed similar differences at Ser1* in the crystal structure of the high-pH form of the SARS-CoV-2 WT M^pro (PDB 6Y2G, molecule A) [9], although it has to be mentioned here that the originally deposited high-pH structure of the SARS-CoV-2 WT M^pro structure had to be corrected, as the interaction between Ser1* and Glu166 had been modeled incorrectly. We have recently applied this correction and uploaded the revised coordinates to the PDB. A similar asymmetry in the arrangement of the Ser1*-Glu166 pair has also been observed in the neutron diffraction study of the telaprevir-M^pro complex described by Kneller et al. [34].

On the other hand, the O-M^pro model at pH 6.5 behaves differently; all monomers have similar salt bridge interactions between Ser1* and Glu166 (Figure 6C–6D). Despite the discussed differences, it is essential to emphasize that independently of the pH of crystallization, our inhibitor, 13b-K, binds efficiently to all four protomers without significant structural differences (Figure 6A–6B). And yet, the structural differences around the Ser1*-Glu166 pair become evident only in the structures of the 13b-K complexes at elevated pH (≥ 7.5), not in the free enzyme. This is in agreement with the observation by Kneller et al. [34], who showed by neutron crystallography that binding of the HCV NS3/4A inhibitor telaprevir to the M^pro leads to a shift in the protonation status of several histidine residues near the active site, along with the aforementioned conformational change of the side chain of Glu166 in one of the protomers within the dimer. This phenomenon could well be related to the observed half-site reactivity of the SARS-CoV M^pro described by Chen et al. [35].

CONCLUSION

In this work, X-ray crystallography, biochemical experiments, and MD simulations are employed to characterize the structure and dynamics of the P132H mutant (or O-M^pro) in complex with the α-ketoamide-based covalent inhibitor 13b-K. The FRET assay for the O-M^pro shows that its enzymatic activity is 27% lower than that of the WT-M^pro. Our kinetic results are in agreement with some reports [12,16] but at variance from others that describe nearly equal kinetics for O-M^pro and WT-M^pro [15,17,20]. The nano-DSF results show that the O-M^pro has substantially lower thermal stability than the WT-M^pro, in agreement with previously reported characterizations of the O-M^pro [12,17]. His132 is located at the intramolecular domain II/III interface and in close proximity to the charged residue Glu240. In addition, it is close to the Asp289-Arg131 salt bridge. Thus, we originally hypothesized that the P132H mutation would introduce a new salt bridge with Glu240 and perturb the local electrostatic and H-bonding networks. However, the crystallographic results at pH 6.5 and 8.5 show that the structures of O-M^pro and WT-M^pro are very similar, with the Asp289-Arg131 salt bridge remaining intact. Curiously, His132 and Glu240 are in a stacking arrangement between the imidazole ring and carboxylate plane as opposed to the edge-to-edge configuration if His132 was charged and donated an H-bond to the carboxylate of Glu240. The constant-pH MD simulations find that the titrations of His132 and Glu240 are coupled and His132(0)/Glu240(−) is the most probable protonation state, which explains why the stacking configuration is observed in the X-ray structures. The fixed-charged MD simulations confirm that the stacking configuration is favored over the edge-to-edge configuration for the His132(0)/Glu240(−) pair. Moreover, the stacking configuration results in more space for water to mediate an H-bond between the backbone of His132 and the side chain of Asp197, which is consistent with the X-ray structures. To explain the observed thermal stability difference between the O-M^pro and WT-M^pro, empirical folding free-energy calculations are employed, which show that the O-M^pro is less stable due to unfavorable van der Waals and backbone torsion energies, suggesting that the P132H mutation induces steric repulsion and unfavorable conformations.

We do not observe any negative impact of the P132H mutation on the binding of the inhibitor 13b-K. In addition, the IC₅₀ measurements show similar values compared to the WT-M^pro. In all structures, 13b-K shows covalent binding to Cys145 and a clear electron density in all binding sites. The assessment of the RMSD against WT-M^pro (PDB 6Y2E) [9] shows that at pH 8.5, each dimer of the two in the asymmetric unit contains one protomer that has a higher RMSD value than its partner. This difference is manifested mainly in the interaction of the N-terminus, Ser1*, with Glu166. A similar phenomenon has been seen earlier in our high-pH crystal structure of the 13b-K complex of WT-M^pro (PDB 6Y2G) [9] and in the neutron structure of the WT-M^pro in complex with telaprevir [34].

Interestingly, the structure of the 13b-K co-crystal at pH 6.5 does not show similar behavior. We also determined the crystal structure of the double mutant P132H+T169S. In the period from December 2021 to April 2022, T169S was recorded as the most frequent companion of the P132H mutation (occurring in ~2.6% of the O-M^pro sequences) and was almost exclusively restricted to lineage BA.1.1.2. After the decline of this lineage, the P132H+T169S double mutant has only rarely been detected. We did not find indications for a coupling between the two mutations, neither in the amino-acid sequences nor in the three-dimensional structures, suggesting that they arose independently of each other. Overall, further investigation is needed to understand better the mutation dynamics and the potential danger from Omicron-based M^pro mutations that might lead to drug resistance.

MATERIALS AND METHODS

Cloning, Gene Expression, and Purification of SARS-CoV-2 M^pro and Mutants

The P132H mutation was inserted by overlap extension-polymerase chain reaction (PCR) reaction. A pair of special primers, P132H_forward (AGTGTGCTATGCGTCATAACTTCACGATCAAA; the underlined sequence corresponds to the mutated P132H codon) and P132H_reverse (TTTGATCGTGAAGTTATGACGCATAGCACACT) was designed. The initial PCR reaction was carried out to generate two splice fragments with 5′ overhangs, using the wild-type M^pro construct as the template. In a subsequent PCR, these spliced fragments served as templates to produce the final PCR product encoding the P132H mutation in M^pro. The resulting PCR product was digested with the restriction enzymes BamHI and XhoI and subsequently ligated into the pGEX-6P-1 vector (GE Healthcare, Uppsala, Sweden), which had been digested with the same enzymes. The mutation T169S was inserted using another pair of primers and the M^pro P132H construct was used as the template. The T169S primers were, T169S-forward: ACATGGAATTGCCGAGCGGTGTACAT; T16 9S-reverse: ATGTACACCGCTCGGCAATTCCATGT.

The gene sequence of the mutated M^pro constructs was verified by sequencing (MWG Eurofins). The expression and purification of sequence-verified SARS-CoV-2 M^pro wild-type and mutated M^pro were performed as described previously [9]. The proteins were concentrated to around 15 mg/mL in the crystallization buffer (20 mM Tris, 150 mM NaCl, 1 mM athylene diamine tetracetic [EDTA], 1 mM dithiothreitol [DTT], pH = 7.5).

Assessment of Protein Stability via Nano-differential Scanning Fluorimetry (nanoDSF)

Thermal-shift assays of SARS-CoV-2 M^pro and its mutants, or the protein-inhibitor complexes, were carried out using the nanoDSF method as implemented in the Prometheus NT.48 (NanoTemper Technologies, Free State of Bavaria, Germany). This method leverages the intrinsic fluorescence of tryptophan (and tyrosine) residues to track protein unfolding. As the temperature rises, the protein unfolds, exposing hydrophobic residues, which alters the autofluorescence ratio at 350 nm and 330 nm. The melting temperature (T_m) is determined from the first derivative of the 350/330 nm ratio. The WT or mutant proteins at 10 μM and 13b-K at 100 μM were mixed in a running buffer (20 mM HEPES, 120 mM NaCl, 0.4 mM EDTA, 4 mM DTT, 20% glycerol, pH 7.0), in a final volume of 15 μL for each measurement. 10 μM of proteins were diluted in the same buffer without inhibitor, to prepare the free enzyme measurement. The free enzyme samples were supplemented with 1% dimethyl sulfoxide (DMSO) to achieve the same final concentration as in the protein-inhibitor mixtures. All samples were loaded into Prometheus NT.48 nanoDSF Grade Standard Capillaries (PR-C002, NanoTemper Technologies, Free State of Bavaria, Germany). Fluorescence signals were recorded across a temperature gradient of 25–90 °C with a flow rate of 0.5 °C/min. Melting curves were generated using GraphPad Prism 7.0 software (Graphpad, California, USA), plotting the first derivative of the 350/330 nm ratio on the Y-axis. The melting temperature (T_m) was determined as the midpoint of the melting curve using PR.ThermControl software (NanoTemper Technologies, Free State of Bavaria, Germany).

Enzyme Kinetics Assays

The enzymatic activity and IC₅₀ of SARS-CoV-2 M^pro and its mutants were measured using a FRET assay with a fluorescent substrate (Dabcyl-KTSAVLQ↓SGFRKM-E[Edans]-NH₂) in a buffer (20 mM HEPES, 120 mM NaCl, 0.4 mM EDTA, 4 mM DTT, 20% glycerol, pH 7.0) [9]. Fluorescence was monitored at 360 nm (excitation) and 460 nm (emission) using a SPARK Multimode Microplate Reader (TECAN, Zurich, Switzerland). For enzymatic activity, 50 nM of SARS-CoV-2 M^pro WT or mutant was mixed with substrate at varying concentrations (10, 20, 40, 80, 120, 160, 240, 320 μM). Initial velocities (ΔRFU/s) were converted to substrate cleavage rates (μM/s) using a calibration curve of free Edans, and were corrected for inner-filter effect [37]. Kinetic constants (V_max, K_m) were derived from the Michaelis–Menten equation and k_cat/K_m was calculated. Results were reported as mean ± standard deviation (SD) from triplicate experiments.

For IC₅₀ determination, 50 nM of SARS-CoV-2 M^pro or O-M^pro, or 100 nM of the P132H+T169S mutant, was incubated with 13b-K (0–100 μM) for 10 min at 37 °C. The reaction was initiated by adding the FRET substrate (10 μM). IC₅₀ values were calculated using GraphPad Prism 7.0, and results were presented as mean ± SD from triplicate experiments.

Crystallization of the SARS-CoV-2 M^pro Mutants

A freshly purified protein solution at 15 mg/mL concentration was centrifuged at 12,000 × g and used to crystallize M^pro mutants as a free enzyme and in complex with 13b-K. For the co-crystallization of the 13b-K/M^pro complex, the protein was mixed with 13b-K (dissolved in 100% DMSO) at a molar ratio of 1:5. Then, the mixture was incubated at 4 °C overnight. The next day, centrifugation was applied (12,000 × g) to remove any precipitate. Subsequently, four commercially available screening kits, PACT premier, SG1 (ShotGun), Morpheus, and Ligand-Friendly Screen (LFS) from Molecular Dimensions (Yorkshire, UK), were used for crystallization using a Crystal Phoenix robot (Art Robbins, California, USA). The sitting-drop vapordiffusion method was used at 25 °C, and 0.3 μL of protein solution was mixed with 0.3 μL of the reservoir and left to equilibrate against a 40-μL reservoir solution. Crystals appeared within three days under several conditions. First, the free enzyme crystals of O-M^pro were obtained from 200 mM NaF, 20% polyethylene glycol 3350 (PEG 3350), and 10% ethyleneglycol at pH 8.2. Next, 13b-K co-crystals were obtained under two different conditions at two different pH values: 25% PEG 3350 and 100 mM Bis-Tris pH 6.5; and 200 mM MgCl₂, 25% PEG 3350, and 100 mM Tris pH 8.5. The free-enzyme crystals of P132H+T169S M^pro were obtained from 100 mM SPG (succinic acid, sodium phosphate monobasic monohydrate, glycine), 25% PEG 1500, pH 8.0. The 13b-K co-crystals with P132H+T169S M^pro were also obtained under two different conditions: 200 mM NaF, 20% PEG 3350, 100 mM Bis-Tris propane, pH 7.5; and 100 mM CHES (N-cyclohexyl-2-aminoethanesulfonic acid), 20% PEG 8000, pH 9.0. Crystals were fished from the drops and cryo-protected by mother liquor plus varied concentrations of glycerol (10%–20%). Subsequently, fished crystals were flash-cooled in liquid nitrogen.

Diffraction Data Collection, Phase Determination, Model Building, and Refinement

All diffraction data sets were collected using synchrotron radiation of wavelength 1.033 Å at beamline P11 of DESY (Hamburg, Germany), using an Eiger 2× 16M detector (Dectris, Zurich, Switzerland) [38]. Five data sets of the mutated SARS-CoV-2 M^pro, the free enzymes of O-M^pro and P132H+T169S mutants, as well as three datasets in complex with 13b-K, were collected from the crystals grown under the aforementioned conditions. XDSapp [39], Pointless [40,41], and Scala [41] were used for processing the datasets.

The datasets of the free enzymes of O-M^pro and P132H+T169S mutants were processed at resolutions of 1.91 Å and 1.80 Å, respectively, in monoclinic space group C2 with one protomer in the asymmetric unit. Datasets of the complexes with 13b-K were processed at resolutions of 2.3 Å and 2.48 Å for the O-M^pro crystals grown at pH 8.5 and 6.5, respectively, in triclinic space group P1 (four protomers in the asymmetric unit). A dataset for the 13b-K complex of P132H+T169S mutant was collected at a resolution of 1.80 Å in monoclinic space group P2₁ (two protomers in the asymmetric unit). Molrep [42] from the CCP4 suite [43] was used for molecular replacement using coordinate set 6Y2E [9] as a starting model. Electronic Ligand Builder and Optimization Workbench (eLBOW) [44] from the Phenix suite [45] was employed for the generation of the geometric restraints for 13b-K, and the inhibitor was built into the F_o-F_c density by using the Coot software [46]. The initial rounds of refinement of the five structures were performed with Refmac5 [47]. Then, the initial models were refined further using Phenix [45], after adding the solvent and the inhibitor, 13b-K. Statistics of diffraction data processing and the model refinement are given in Table S1.

All-atom PME Continuous CpHMD Titration Simulations

To determine the protonation states of O-M^pro, we performed titration simulations using the newly implemented all-atom particle-mesh Ewaald (PME) CpHMD [27] method in the Amber22 program [28] and the asynchronous pH-replica exchange protocol [29] to enhance sampling and accelerate convergence. Starting from the pH 8.5 X-ray structure of O-M^pro (with the inhibitor removed), the LEAP utility in Amber22 [28] was used to add hydrogens and dummy hydrogen for all Asp, Glu, and His side chains and construct the free N- and C-termini. The protein was solvated in an octahedral water box with a minimum distance of 11 Å between the protein heavy atoms and water oxygen atoms at the edges of the box. Sodium and chloride ions were added to neutralize the system (with titratable sites in the default or model protonation states) and reach an ionic strength of 150 mM. The ff14SB force field [48] and TIP3P model [49] were used to represent the protein and water, respectively. The system underwent energy minimization with the steepest descent for the first 200 steps and conjugate gradient algorithm for the last 300 steps; a harmonic restraint with the force constant of 100 kcal/mol/Ų was applied to the protein-heavy atoms. With the same harmonic restraint, the system was heated from 100 K to 300 K using a Langevin thermostat for 100 ps with a time step of 1 fs. After heating, CpHMD was turned on and a two-step equilibration was performed at pH 7, whereby the harmonic force constant was 100 kcal/mol/Ų in the first 250 ps and 10 kcal/mol/Ų in the last 250 ps. Next, the system was further equilibrated in four steps at 21 pH conditions of the replica-exchange runs (see below), whereby the restraint force constant was reduced from 10 to 1, 0.1, and 0 kcal/mol/Ų, and each step lasted 500 ps. The production run utilized 21 pH replicas in the pH range of 2.5–8 with an interval of 0.25 pH units. An exchange between adjacent pH replicas was attempted every 2 ps (1000 MD steps) and the simulation lasted 45 ns per pH replica, which resulted in an aggregate sampling time of 945 ns. During the equilibration and production runs, the temperature was maintained at 300 K using the Langevin thermostat with a collision frequency of 1 ps⁻¹. The isotropic Berendsen barostat was used to maintain a pressure of 1 bar with a relaxation time of 1 ps. The PME method with a real-space cutoff of 8 Å and grid space of 1 Å was used to treat long-range electrostatic interactions. The SHAKE algorithm was used to constrain the bonds involving hydrogen atoms to allow the integration step of 2 fs. Except for His132 and Glu240, the protonation states of all Asp, Glu, and His at pH 7 are in agreement with our previous simulations of the WT-M^pro [33].

Fixed-charge MD Simulations

The fixed-charge MD simulations of O-M^pro were performed using the Amber20 program [50]. The solvated protein system was prepared as in the all-atom CpHMD simulation above. The protonation states were fixed as determined by the all-atom CpHMD for the pH 7 condition. Neutral histidines were set in the HIE state. In the simulations (two independent runs) starting from our pH 8.5 X-ray structure, His132 was set in the neutral state. In the simulations (four independent runs) starting from 7TOB, His132 was set in the neutral (two independent runs) or charged (two independent runs) state. The system was first energy minimized under the harmonic restraints of 100 kcal/mol/Ų placed on the protein heavy atoms for a total of 20,000 steps using the steepest descent (first 10,000 steps) followed by 10,000 steps using the conjugate gradient algorithm. The system was gradually heated from 0 K to 300 K in 1 ns under the same harmonic restraints. Next, five stages of equilibration (every 10 ns) were conducted, whereby the harmonic restraints were gradually reduced from 10 to 5, 2, 1, and 0.1 kcal/mol/Ų. All other settings were the same as in the CpHMD simulations. Two 500-ns production runs were performed based on our pH 8.5 X-ray structure of O-M^pro. Four 1-μs production runs were performed based on the X-ray structure 7TOB, two of them with His132(0) and two of them with His132(+).

Empirical Calculations of Protein Stability Change upon Mutation Using Rosetta

The changes in the stability of the free M^pro dimer and monomer upon the P132H mutation were estimated separately using the ddG_monomer application in the Rosetta software suite [32]. In this method, an ensemble of structures of the P132H mutant was generated from an input WT structure (PDB ID 7VH8 [51]; nirmatrelvir was removed). The change in folding free energy was calculated as the difference in the energy scores (using the REF2015 energy function [52]) between the WT and P132H M^pro structures. The high-resolution protocol was followed, which allowed for both backbone and side-chain adjustment. Rosetta’s standard side-chain optimization module first optimized the structure, then three sequential minimization calculations were performed where the Lennard-Jones potential was scaled by 0.1, 0.33, and 1.0, respectively. This was repeated 50 times for both the WT and mutant proteins, and then the average score for each system was calculated. To prevent the backbone from deviating from the initial structure, distance restraints were placed on the alpha carbon atoms following the default protocol [32].

Supplementary Material

SUPPLEMENTARY DATA

Supplementary data to this article can be found online at https://doi.org/10.1016/j.hlife.2024.06.003.

HIGHLIGHTS.

• The enzyme activity and thermal stability of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Omicron M^pro and the P132H+T169S double mutant are reduced.

• Crystal structures of the Omicron M^pro and the double mutant show a stacked configuration of H132 and E240.

• The stacked His132/Glu240 may promote a water-mediated interaction between Asp197 and His132.

• The Omicron M^pro is less stable due to steric repulsion and unfavorable conformations of His132.

DATA AVAILABILITY

Crystal structures have been deposited in the Protein Data Bank with accession codes 8R19 (P132H, pH 8.2), 8R24 (P132H+T169S, pH 8.0), 8R0V (P132H in complex with 13b-K, pH 6.5), 8R26 (P132H in complex with 13b-K, pH 8.5), 8R1Q (P132H+T169S in complex with 13b-K, pH 7.5). Supplementary Figure S6 shows the structure of the P132H+T169S double mutant in complex with 13b-K at pH 9.0, to enable a comparison with the pH 7.5 structure. Supplementary Figure S7 shows 2F_o-F_c electron density maps of O-M^pro or M^pro-P132H+T169S in complex with 13b-K.