A Crystallographic Snapshot of SARS-CoV-2 Main Protease Maturation Process

Graphical abstract

Abstract

SARS-CoV-2 is the causative agent of COVID-19. The dimeric form of the viral M^pro is responsible for the cleavage of the viral polyprotein in 11 sites, including its own N and C-terminus. The lack of structural information for intermediary forms of M^pro is a setback for the understanding its self-maturation process. Herein, we used X-ray crystallography combined with biochemical data to characterize multiple forms of SARS-CoV-2 M^pro. For the immature form, we show that extra N-terminal residues caused conformational changes in the positioning of domain-three over the active site, hampering the dimerization and diminishing its activity. We propose that this form preludes the cis and trans-cleavage of N-terminal residues. Using fragment screening, we probe new cavities in this form which can be used to guide therapeutic development. Furthermore, we characterized a serine site-directed mutant of the M^pro bound to its endogenous N and C-terminal residues during dimeric association stage of the maturation process. We suggest this form is a transitional state during the C-terminal trans-cleavage. This data sheds light in the structural modifications of the SARS-CoV-2 main protease during its self-maturation process.

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the causative agent of COVID-19, a highly infectious disease that rapidly spreads causing a global pandemic. SARS-CoV-2 is an enveloped RNA virus belonging to the β-lineage of coronaviruses, which includes SARS-CoV and Middle East (MERS-CoV) respiratory viruses.1, 2, 3 The viral genome is a single-stranded positive RNA comprising about 30,000 nucleotides, that shares 82% sequence identity with SARS-CoV.⁴ The replicase gene (ORF1ab) encodes two overlapping polyproteins (pp1a and pp1ab) that are required for viral replication and transcription.⁵

The main protease (M^pro), also known as 3C-like protease (3CL^pro) is a viral cysteine protease specific for glutamine at the S1 subsite, showing variable recognition preferences at S2 (Leu/Phe/Met/Val) and S2′ subsites (Ser/Ala/Gly/Asn).⁶ M^pro is responsible for the maturation of pp1a and pp1ab in at least 11 characterized sites, including its auto-processing at the N and C terminus, which is essential for its activity and dimerization.4, 7, 8 Due to its essential role in viral replication, M^pro is one of the most well characterized non-structural proteins of SARS-CoV-2. In addition, its unique features of cleavage site recognition and the absence of closely related homologues in humans, identify M^pro as a major target for antiviral drug development.4, 9, 10

Although M^pro activity is crucial to viral biology, its self-maturation process is still poorly understood. Several biochemical and crystallographic studies on native and mutated forms of SARS-CoV M^pro tried to elucidate its maturation mechanism,¹¹ by evaluating if the N and C-terminus processing occurs within a dimer (cis-cleavage) or between two distinct dimers (trans-cleavage). The first 2005 model suggested that M^pro probably forms a small amount of active dimer after autocleavage that immediately enables the catalytic site to act on other cleavage sites in the polyprotein.¹² In 2010, based on the observation that dimerization of mature M^pro is enhanced by the presence of substrates, Li and colleagues proposed that after the translation, two M^pro protomers form a transient dimer which is stabilized by binding the N-terminal site of its substrate (another M^pro in polyprotein) and further cleave to free its N-terminus.¹³ In addition, Chen et al. suggested that the N-terminal autocleavage might only need two immature forms of M^pro in monomeric polyproteins to form an intermediate dimer that is not related to the active dimer of the mature enzyme.¹⁴

Herein, we used X-ray crystallography integrated with biochemical techniques to investigate the self-maturation process of SARS-CoV-2 M^pro. The construct of M^pro containing N-terminal insertions produced an immature form of the enzyme (IMT M^pro), unable to form a dimer, that showed a reduced enzymatic activity. We used fragment screening to probe new cavities for drug development in this construct. The inactive mutant C145S with inserted native N-terminal residues (C145S M^pro) produced a form of the protein that behaves as monomers, dimers, trimers and tetramers in solution. Crystals of the tetrameric form revealed details of the dimeric association of M^pro during self-processing of its N and C-terminal residues. All forms of the enzyme revealed important conformation changes of the enzyme during maturation, which can guide direct-acting drug development.

Activity and biochemical characterization

A general strategy to produce SARS-CoV-2 M^pro is to maintain its self-cleavage N-terminal portion and add the HRV-3C cleavage site with a histidine-tag at the C-terminal portion. We successfully used ammonium sulfate precipitation followed by ion exchange chromatography to obtain pure M^pro, simplifying the protocol to one that takes less than 8 h and with a final yield of ~2.5 mg/L of culture. The SARS-CoV-2 IMT M^pro was obtained by adding a non-cleavable sequence (Gly-Ala-Met) at the N-terminal Ser1 of M^pro, and purified by a similar protocol. The SARS-CoV-2 IMT M^pro was produced as a soluble protein, yielding ~80 mg/L of culture. To further investigate the role of N-terminal residues in the maturation of M^pro, we designed a construct containing the mutated C145S residue with its native cleavage peptide of M^pro (Ser-4, Ala-3, Val-2, Leu-1, Gln-0↓) at the N-terminal of Ser1 (Figures 1 (a) and S1(a)). During gel filtration, two M^pro peaks were identified with mass consistent with a monomer and a tetramer (Figure S1).

Figure 1.

(a) Schematic showing different constructs of M^pro. (b) Time-course reactions of M^pro constructs against fluorogenic peptide substrate. (c) Differential scanning fluorimetry of M^pro constructs. M^pro is shown as blue squares, IMT M^pro is shown as red spheres and C145S M^pro is shown as black triangles (d) SEC elution profiles with overlaid calculated molar mass from elution peaks. M^pro (blue) elutes as a single peak with a calculated molecular mass consistent with a dimer. IMT M^pro (red) exhibits a single peak with a mass compatible with a monomer in solution. The monomeric SEC peak of C145S M^pro (grey) elutes as an equilibrium between dimers and monomers in solution. The tetrameric SEC peak of C145S M^pro (black) contains peaks of monomers, dimer, trimers and tetramers. (e) SDS-PAGE of N-terminal cleavage over time from C145S M^pro. At top, reaction containing 10 µM C145S M^pro, and at the bottom the same reaction supplemented with 5 nM M^pro. Red arrows are pointing to the band of cleaved M^pro. The bar graph shows the relative band intensity of cleaved M^pro overtime for both reactions in blue and salmon, respectively. (f) SEC elution profiles of monomeric peak of C145S M^pro over time. (g) SEC elution profiles of monomeric peak of C145S M^pro over time supplemented with 10 nM M^pro. In SEC-MAL graphs, curves correspond to the change in the normalized scattered light intensity at 90° (lines) and calculated molar mass of the corresponding peak (dots) are given for each peak.

M^pro and IMT M^pro demonstrate to be active and able to recognize and cleave the fluorogenic substrate (Figure 1(b)), with K_m values of 16.4 ± 2.0 μM and 34.3 ± 2.2 μM, respectively. IMT M^pro exhibited only 0.6% of the catalytic efficiency compared with mature M^pro (Table 1 ). As previously reported, the M^pro N-terminal is fundamental for dimerization and any additional residues would reduce or even abolish its activity.4, 15, 16, 17, 18 As expected, C145S M^pro has only shown residual activity (Figure 1(b)). All three M^pro constructs exhibited similar thermal-stability profiles, indicating similar folding (Figure 1(c)).

Table 1.

Kinetic parameters of M^pro constructs. Relative efficiency is the K_m/k_cat of constructs relative to M^pro

	Km (µM)	Vmax (RFU.s−1)	kcat (s−1)	Relative efficiency
Mpro	16.4 ± 2.0	1.05 ± 0.04	28.0 ± 0.1	1
IMT Mpro	34.3 ± 2.2	0.104 ± 0.003	0.35 ± 0.07	0.006

Analysis in solution using SEC-MALS suggests that M^pro behaves as a dimer in the tested conditions, as expected (Figure 1(c)).¹⁰ For IMT M^pro, the additional residues at N-terminal seem to prevent dimerization completely (Figure 1(c)). For C145S M^pro, however, the additional residues allow the protein to adopt multiple conformational states ranging from monomers to tetramers (Figure 1(c)).

C145S M^pro in solution characterization

Despite the site-direct mutagenesis of the C145S M^pro, this enzyme exhibited residual proteolytic activity which allowed us to observe the self-processing of the monomeric peak of C145S M^pro by SDS-PAGE in the course of two days (Figure 1(e)). By quantifying the mass intensity, we estimate that at the end of two days about 30% of the protein was self-cleaved after incubation (Figure 1(e)). By using SEC-MALS, we also monitored the formation of dimers by monomeric C145S M^pro sample after 0 h, 24 h, 48 h and 72 h incubation at room temperature (Figure 1(f)). At 0 h, the mass recovery ratio between monomers/dimers was 14.9, which decreased to 1.04 at 24 h, and 0.09 at 48 h incubation, with complete degradation of the monomer peak after 72 h (Table 2 ). This data indicated that the cleavage of the N-terminal is directly proportional to the formation of dimers in solution, highlighting the importance of the N-terminal processing for the assembly of the M^pro.

Table 2.

Molecular Mass and Mass Recovery percentages for each observed peak in the SEC-MALS profiles of SARS-CoV-2 M^pro constructs. Mass Recovery Ratio is the ratio of Mass Recovery of Peak 1 and Peak 2

		Peak 1 - Monomer		Peak 2 - Dimer
Protein	Time (h)	Molar Mass (kDa)	Mass Recovery (%)	Molar Mass (kDa)	Mass Recovery (%)	Mass Recovery Ratio (Peak 1/Peak 2)
C145S Mpro	0	36.0 ± 0.1	74.5	65.7 ± 0.2	5.0	14.9
	24	35.6 ± 0.1	29.6	73.7 ± 0.1	28.4	1.04
	48	34.9 ± 0.1	7.4	69.1 ± 0.1	84.5	0.09
	72	–	–	68.7 ± 0.2	87.3	–
C145S Mpro + Mpro	0	36.4 ± 0.1	75.0	63.0 ± 0.1	5.8	12.9
	24	34.8 ± 0.2	22.8	72.9 ± 0.1	43.5	0.5
	48	36.7 ± 0.2	2.0	67.0 ± 0.1	86.7	0.02
	72	–	–	66.1 ± 0.3	91.3	–

To investigate the effect of M^pro in the N-terminal processing, we monitored the effect of adding M^pro to the C145S M^pro samples in a ratio of 1:6000. On the SDS-Page, we can see that the sample containing M^pro showed an increase ratio of protein cleavage after 20 h when compared with the previous experiment (Figure 1(e)). At the SEC-MALS, the mass recovery ratio between monomers/dimers for this sample at 0 h was 12.9, them 0.5 at 24 h, and 0.02 at 48 h incubation, also with complete degradation of the monomer peak after 72 h (Figure 1(g) and Table 2). The data suggests the addition of M^pro to the C145S M^pro sample at 1:6000 ratio increased the speed of N-terminal processing and dimer formation by the order of 50% after 24 h.

Crystal structure of M^pro in monoclinic and orthorhombic crystal system

M^pro was crystallized in the monoclinic crystal system in several conditions and its X-ray structure was determined at 1.46 Å in C2₁ space group, as the majority of the PDB deposits. All 306 residues were refined at the electron density to a final R _work/R _free of 0.16/0.18, with 99% of Ramachandran in favored positions (Table S2). The crystal asymmetric unit contains one monomer which could be symmetry expanded to the biological dimer, following the same pattern of the majority of known structures deposited in PDB (r.m.s.d of 0.2 Å vs PDB 5RGG, for all Cα 306). The M^pro protomers are formed by three domains (DI, DII and DIII), with its catalytic region located between the beta-barrels comprising DI and DII⁴ (Figure S2). Using seeds from IMT M^pro, we were able to obtain a new crystal system in orthorhombic space group P2₁2₁2₁ at final resolution of 1.86 Å. This structure was refined to final R _work/R _free of 0.19/0.22 and 98.33% of Ramachandran in favored positions (Table S2). This crystal system shows the full dimer in the asymmetric unit, and its packing appears to offer advantages for soaking compounds in M^pro active site when compared with the canonical C2₁ form, especially for compounds targeting subsites S3-S4, which are less constrained by crystal packing in the orthorombic form (Figure S14). This is being explored by the COVID Moonshot initiative, and will be latter described in a separated manuscript.

Crystal structure of IMT M^pro

The crystal structure of IMT M^pro at 1.6 Å was determined using 3 merged datasets (Figures S3, S4, Table S1) in P2₁2₁2₁ space group, with two molecules in the asymmetric unit, packed in similar shape to the known biological unit of M^pro. The structure was refined to a final R _work/R _free of 0.20/0.22, with 97% of Ramachandran in favored positions (Table S2). In the recent published structures of GM-M^pro, both apo and ligand-complexes exhibited minor differences with the mature form.¹⁵ However, in our structure there are distinguishable differences in the overall structure, especially in the position of DIII helices (Figure 2 ). Although IMT M^pro asymmetric unit resembles the biological dimer form of native protein, PISA¹⁹ analysis indicate that the dimer packing is unstable in solution, with an interface area of 1256 Å² (versus 1557 Å of M^pro), calculated free energy ΔG of −13.4 kcal/mol (versus −14.9 kcal/mol of M^pro) for 26 potential hydrogen bonds (versus 33 of M^pro) and 5 potential salt bridges (versus 10 of M^pro). The repositioning of Ser1 by itself was responsible for the breaking of five of these hydrogen bonds, maintained by Phe140, Glu166 add Gly170. Interactions between Gly2 and Ser139, Ala7 and Val124, Ser139 and Gln299 and Phe305 with Pro122 are also not present in IMT M^pro.

Figure 2.

(a) Overview of DIII region from IMT M^pro (chain A yellow and B cyan) superposed with M^pro (grey ghost). N-terminal residues are depicted as spheres. (b) Rotated view showing IMT M^pro DIII from a distinct angle. (c) Active site residues of IMT M^pro chain B (cyan cartoon) superposed with M^pro. Catalytic residues are depicted as yellow sticks. N-terminal chain A residues are depicted as spheres. M^pro structure and residues are shown as a grey ghost.

While IMT M^pro DI and DII are less affected by the N-terminal insertion (r.m.s.d of 0.34 Å vs M^pro for Cα of 1–184), DIII appears to adopt a more open conformation relative to M^pro (r.m.s.d of 1.33 Å for Cα of 201–301) (Figure 2), with the interfacing residues Ala285 at a distance of 9.9 Å in the IMT M^pro (versus 5.5 Å in M^pro) (Figure S5). This conformation is more accentuated at chain A where the electron density of the N-terminal insertion is clearly visible in the model. For this chain, the N-terminal insertion pushes chain A helices αF and αH further away from chain B active site, opening a cleft for Phe140 rises to the surface of the molecule, leading to major conformation alterations of the chain B active site souring residues, such as Glu166, Pro168 and Gln189 (Figures 2 and S6). As DIII is known for being extremely flexible,²⁰ we compared the structure of IMT M^pro with the structure of M^pro in orthorhombic crystal system in order to investigate if the dislocation of DIII was being promoted by the distinct crystal packing. In fact, the structure of M^pro in the orthorhombic crystal system is much more similar to the canonical M^pro (r.m.s.d of 0.52 Å for Cα of 604 residues) rather than IMT M^pro (r.m.s.d of 0.91 Å for Cα of 604 residues), indicating that DIII dislocation is indeed caused by the extra N-terminal residues (Figure S13).

The plasticity of SARS-CoV-2 M^pro active site was already reported when apo X-ray structures collected at cryo and room temperatures were compared,²¹ and its expected given the broad spectrum of endogenous substrates that M^pro is has to process. However, the IMT M^pro revealed major structural alterations in the oxyanion hole, likely affecting enzyme processing. The cascade effect of the steric hindrance caused by the N-terminal extra residues affects the position of Ser1, Phe140, Glu166 and Pro168, disrupting the shape of subsites S1, S2 and S4 (Figure 2). Within these the S1 seem to be most affected, assuming an unusual flattened configuration that seem to disrupt the cavity responsible for the recognition of glutamine side-chain, likely affecting substrate recognition (Figure 2). This not only explains the diminished activity of this construct, as well as shows the importance of full N-terminal processing for the correct folding of M^pro. Despite the significant changes of the active site, relative position of the catalytic dyad Cys145-His41 remains unchanged in this form (Figure 2).

Fragment screening of M^pro immature

Recently, a small-fragment library of more than 1,250 unique fragments were screened against SARS-CoV-2 M^pro, identifying 74 high-value fragment hits, including 23 non-covalent and 48 covalent hits in the active site, and 3 hits close to the dimerization interface.²² In here, we applied the same technique to probe new druggable cavities in IMT M^pro. Although the difference in scale of our experiment, we were able to identify five distinguishable sites in this form of the protein (Figure 3 ). Site #1 is the active site of chain A, in which fragment f2xe03 was identified interacting with Glu166 N and Cys145 S, corresponding to the S1 substrate subsite. In contrast, the orthorhombic crystal form of IMT M^pro seams promising for compound soaking due to the crystal packing, especially for those binding in S3-S4 subsites. Interesting, a unique cavity marked as Site #3 was identified in our experiments, bound to fragment f2xg05 by Arg4 main chain O. That cavity lies between the interface of chains A and B, and is not present in M^pro which adopt a more closed conformation (Figure S14). This new site and fragment could serve as an anchor for development of new inhibitors targeting M^pro dimerization process, a mode of action that was too date only theorized.²³ Details about data processing and statistics are given in Table S3.

Figure 3.

(a) Location of IMT M^pro probing fragments identified during screening. Chain A is colored as yellow surface, chain B as cyan surface. Fragments are depicted as red spheres. For comparison, fragments of previous manuscript using monoclinic M^pro were aligned to the structure of IMT M^pro and are depicted as grey spheres.²² (b) Contact details of identified fragments bound to IMT M^pro. Chain A is colored as yellow cartoon and chain B as cyan cartoon. Fragments are depicted as yellow sticks. Residues forming polar contacts are depicted as green lines. Contacts are depicted as black dashes.

Crystal structure of C145S M^pro in complex with N and C-terminal residues

The tetramer peaks of C145S M^pro were crystallized and its X-ray structure was determined at 2.8 Å and R _work/R _free of 0.20/0.25 (Table S2), revealing a new crystal form in which N-terminal cleaved residues (Ser-4, Ala-3, Val-2, Leu-1, Gln-0) were found trimmed in the active site of chain A, occupying subsites S1-S5 (Figure 4 (a)). Despite the site directed mutagenesis of the catalytic cysteine to serine, electron density shows that Gln0 and Ser1 are non-covalently bound in the amino region, clearly indicating that the N-terminal cleavage was completed. At the S1 subsite, Gln0 NE2 interacts with Glu166 OE1 by a hydrogen bound (2.7 Å), while Gln0 form interacts with Ser145 in the position of the native oxyanion hole (Figures 4, S8 and S9). To accommodate the hydrophobic sidechain of Leu-1 at P2, Met49 and Met165 are pushed further of each other (Figure S10), leading to a more opened groove of this subsite relatively to the apo-state, explaining the ability of this subsite to accommodate a variety of hydrophobic side chain residues, such as Leu, Met, Ile, Val and Phe.6, 24 Yet, from the eleven endogenous recognition sites of coronaviruses, S2 Leu carrying sequences are the ones in which M^pro display higher catalytic efficiency, highlighting the importance of this conformation for drug design. At subsites S3-S5, the interactions of Val-2, Ala-3 and Ser-4 are mainly maintained by hydrogen bounds between the polar residues of protein and peptide side chains (Figure 4(a)), which explains the ability of M^pro to recognize a large variety of chain sequences at those positions.

Figure 4.

(a) C145S M^pro chain A active site (cyan surface) in complex with processed N-terminal residues (yellow sticks). Main interacting residues are depicted as blue lines. (b) C-terminal peptide (yellow) main interactions with C145S M^pro chain A active site residues (blue). (c) C145S M^pro chain B active site (blue surface) in complex with processed C-terminal residues (yellow sticks). Main interacting residues are depicted as blue lines. (d) C-terminal peptide (yellow) main interactions with C145S M^pro chain B active site residues (blue). For (b) and (d), the 2mFo-DFc electron density contoured at 0.8σ. Ser1 from respective dimerization partners are depicted with green letters. *Ser145 is the site-direct mutant of Cys145. Simulated annealing omit map is available in Figure S7.

The crystal structure of C145S M^pro revealed another important step in the maturation process of M^pro, as we can see that chain B C-terminal residues (301–306) are ~180° rotated from its canonical conformation (Figure S11) and trimmed in the active site of a symmetric related chain B (Figure 4(b)), a phenomenon that was also recently observed by another group in the C145S M^pro mutant.²⁵ During this event, two C145S M^pro dimers appear to be linked by the interaction of the C-terminal and a respective active site, revealing details of the dimeric association in a non-closed complex (Figure 5 ). The electron density of this dataset indicates that chain B Ser145 OG is covalently bound to Gln306 C from crystallographic symmetric correlated chain B (distance of 1.4 Å), with the loss of one oxygen by Gln306 (Figure S9). We believe the diminished activity of the mutant as allowed the formation of these crystals after almost 20 days, from which we were able to capture this intermediary state of the maturation. We highlight here that the model deposited model does not depicted this covalent bound, as we found impossible to link two atoms outside the asymmetric unit (even after consultant with software developers). Within the active site, Gln306 occupies the respective position of Gln0 at S1, while S2 is occupied by Phe305, increasing the distance between Met49 and Met165 relatively to chain A bound to N-terminal (Figure S10). As for the N-terminal residues, subsites S3-S5 interactions with C-terminal are mainly maintained by hydrogen bounds between main chains (Figure 4(b)).

Figure 5.

Overview of the dimer-dimer association intermediary formed by C145S M^pro tetramer during self-processing. Chain A is colored as yellow surface, chain B as cyan surface. Trans-cleavage M^pro partner is show as green cartoon. N-terminal residues are depicted as blue spheres, and C-terminal residues are depicted as red spheres.

The maturation process of M^pro and its impact on drug discovery

M^pro is firstly produced as the Nsp5 domain of the viral polyproteins before they are proteolytically processed into 15 or 16 non-structural proteins.¹¹ Immediately after translation, the immature form of M^pro would contain both N and C-terminal insertions, which requires self-processing to generate the mature form of the enzyme.¹² In M^pro, the cleaved N-terminals are sandwiched between the two protomers of the dimeric enzyme, being a part not only of the dimer interface but also from the respective protomer active site. In the IMT M^pro, the N-terminal extra amino acids seem to disrupt the active site shape at S1-S3 subsites (Figure 5), affecting its capacity of recognizing the substrate and processing. Notwithstanding, the extra N-terminal residues also seem affect the enzyme ability to form dimers by pushing the reciprocal DIII further to its native conformation.

The same process appears to occurs to C145S M^pro monomers with native N-terminal inserted residues, although in this case, the slow cleavage of the N-terminus results in the formation of dimers overtime (Figure 1(e)). The incubation of this samples to allow dimerization appears to significantly enhance the enzymatic residual activity of this construct, indicating that the dimeric form is important for activity even for this serine mutant (Figure 6 ). By monitoring the formation of dimer overtime, we saw that the monomeric enzyme is capable of processing its N-terminal, suggesting that cis-cleavage as a mechanism for the first step of the maturation process (Figure 1(e)). However, when we compare the results from both experiments, we notice that the ratio of dimer formation seen to be far superior to the ratio of N-terminal processing (Figure 1(e)–(g)). This is in partial agreement with the model proposed by Li and colleagues (2010) in which two M^pro form a transient dimer that is stabilized by the binding the N-terminal site of its substrate (another M^pro in polyprotein) and further cleave to free its N-terminus.¹³ It is also another argument for our model of the immature form, in which a partially cleaved M^pro would result in a constrained packing of M^pro with diminished activity (Figure 2).

Figure 6.

Time-curse reactions of C145S M^pro monomeric construct determined after different incubation periods. Activity of the construct was monitored after 0 h (black), 24 h (red) and 48 h (blue) incubation.

During our analysis we also observed that when M^pro is added to C145S M^pro, the N-terminal cleavage and dimer formation seem to be enhanced significantly (Figure 1(e)–(g)), suggesting that this initial maturation step is a mix of cis and trans-cleavage events. Despite our efforts, we were not able to obtain a complex showing the details of an intact N-terminal protomer in complex with M^pro. Our tentative of crystallization of monomeric C145S M^pro only resulted in crystals that were identical to the canonical monoclinic crystals of M^pro, with fully cleaved N-terminal and dimeric packing. (PDBid 7N5Z, r.m.s.d of 0.25 with M^pro 7KPH for 306 Cα). Still, more studies are required to understand the exact order of the events and inter-protomer interactions involved in the N-terminal cleavage.

After the active site region is matured (or even concomitantly), dimeric M^pro _, C-terminal seem to assume an unusual rotated position (Figure S11), allowing it to bound into the active site of another mature or half-mature M^pro dimer (Figure 4(b)). In this step, the trans-cleavage processing of the C-terminal residues would serve as an anchor for a transitory dimer-dimer association state of the protein, herein captured with the construct of the mutant C145S M^pro with the processing of the N-terminal residues (Figure 5). As a result, full mature M^pro is produced and its ready to process other parts of the viral polyprotein. During all those maturation processes, both M^pro active site and surface undergo significant conformational changes, which could guide targeted drug development (Figure 7 and Supplementary Video 1). Our results not only shed light in the self-maturation process of SARS-CoV-2 M^pro, but also bring the perspective of developing drugs targeting intermediate states of this enzyme.

Figure 7.

Scheme containing steps of SARS-CoV-2 M^pro self-maturation process. (a) At first, two protomers assembly as an immature dimer during N-terminal cis and trans-cleavage. After processing, the M^pro with the matured active site permit the correct positioning of DIII, which allows the stabilization of the dimeric form. The dimer C-terminal is them trans-cleaved by another full or at least half mature dimer, forming a transient dimer-dimer association and producing the full mature form of M^pro. (b) Surface view of chain B active site from immature form. (c) Surface view of chain A active site during N-terminal residues recognition. (d) Surface view of chain B active site during C-terminal residues recognition. (e) Surface view of full mature M^pro active site.

Materials and Methods

Cloning and expression of SARS-CoV-2 M^pro forms

The viral cDNA template (GenBank MT126808.1), kindly provided by Dr. Edison Durigon (University of São Paulo, São Paulo, Brazil), was synthetized using the SCRIPT One-Step RT-PCR kit (Cellco Biotec) and random hexamers primers. For production of IMT M^pro, coding region of M^pro (residues 3264–3569) was amplified using primers: Fw 5′ CAGGGCGCCATGAGTGGTTTTAGAAAAATGGCATTC 3′ and Rv 5′ GACCCGACGCGGTTATTGGAAAGTAACACCTGAGAC 3′, and the sequence was inserted into the pET_M11 vector, which encodes an N-terminal 6xHis-tag followed by a TEV protease cleavage site (ENLYFQ↓GAM), using the LIC method,²⁶ forming the plasmid pET_M11-IMT-M^pro. To obtain the mature form of M^pro, native N-terminal residues (GAMSAVLQ↓SGFRK) were inserted into pET_M11-IMT-M^pro by inverse PCR using primers: Fw: 5′ GCTGCAGAGTGGTTTTAGAAAAATGGCATTC 3′ and Rv: 5′ ACGGCTGACATGGCGCCCTGAAAATA 3′. Amplified product was treated with T4 Polynucleotide Kinase (PNK, Thermo Fischer Scientific) and T4 Ligase (Cellco Biotec), forming plasmid pET_M11- M^pro. For C145S M^pro construct, pET_M11- M^pro was used as template for inverse PCR with primers Fw 5′ CCTTAATGGTTCATCTGGTAGTG 3′ and Rv 5′ AATGAACCCTTAATAGTGAAATTGG 3′. The PCR product was digested with DpnI (NEB), followed by treatment with PNK and T4 DNA ligase, forming the pET_M11-C145S-M^pro. All plasmids were transformed in DH5α E. coli competent cells. All PCRs were conducted with FastPol (Cellco Biotech). Positive clones were selected and confirmed by sequencing. Schematics of constructs are given in Figure S1.

For protein production, E. coli BL21 cells were transformed with respective plasmids and cultured in ZYM-5052²⁷ at 37 °C and 200 RPM to an OD₆₀₀ of 0.8, followed by expression at 18 °C, 200 RPM for 16 h. Cells were harvested by centrifugation at 5000g for 40 min at 4 °C, resuspended in lysis buffer (20 mM Tris pH 7.8, 150 mM NaCl, 1 mM DTT), disrupted by sonication and the lysate was clarified by centrifugation at 12,000g for 30 min at 4 °C.

Protein purification of IMT M^pro

After expression, a large amount of IMT M^pro had its 6xHis-tag cleaved by autoproteolytic process. The small fraction of 6xHis tagged protein was removed from the lysate using Ni-NTA resin (Qiagen). The cleaved protein was purified by adding 1 M ammonium sulfate to the cell lysate followed by incubation on ice for 10 min. The precipitated protein was recovered by centrifugation at 12,000g for 30 min at 4 °C, resuspended in gel filtration buffer (20 mM Tris pH 7.8, 150 mM NaCl, 1 mM EDTA, 1 mM DTT) and purified by size-exclusion chromatography using a HiLoad 26/100 Superdex 200 column (GE Healthcare) pre-equilibrated with gel filtration buffer. Purified fractions were aliquoted, flash-frozen and stored at −80 °C for enzymatic assays and crystallization. For crystallization, protein was concentrated to 14 mg mL⁻¹ using 10 kDa MWCO centrifugal concentrators (Vivaspin, Sartorius). Protein concentrations were determined using the measured absorbances at 280 nm and the theorical extinction coefficient of 32,890 M⁻¹ cm⁻¹. Protein purity was analyzed by SDS-PAGE (Figure S1).

Protein purification of M^pro

M^pro was purified similar to IMT M^pro, with an additional step of cation exclusion chromatography. After the size exclusion chromatography, the protein was buffer exchanged to 20 mM Tris pH 8.0, 1 mM DTT, and then injected into a Mono-Q 5/50 GL column (GE Healthcare). Protein was eluted using a linear gradient of a buffer containing 20 mM Tris pH 8.0, 1 M NaCl and 1 mM DTT. Finally, fractions containing the purified protein were buffer exchanged to gel filtration buffer. Purified fractions were aliquoted and protein was concentrated and quantified similarly to IMT M^pro (Figure S1).

Protein purification of C145S M^pro

For C145S M^pro, protein was purified by immobilized metal chromatography (IMAC) using a 5 mL HisTrap FF column (GE Healthcare). After column washing with buffer A (20 mM Tris pH 7.8, 150 mM NaCl, 25 mM Imidazole), protein was eluted with buffer A supplemented with 250 mM imidazole. Sample was buffer exchanged using a 5 mL HiTrap desalting column (GE Healthcare) equilibrated with buffer A. To remove the 6xHis-tag, 2 mg of TEV protease and 4 mM DTT were added to the sample and incubated for 16 h at 4 °C. Next day, non-cleaved protein and TEV were removed by a second step of IMAC in buffer A. Finally, the protein was purified by size-exclusion chromatography using a HiLoad 16/60 Superdex 75 column (GE Healthcare) equilibrated with gel filtration buffer. Purified fractions were aliquoted, and protein was concentrated and quantified as described for other constructions (Figure S1).

Crystallization

Crystallization screening was performed with the sitting drop vapor diffusion method in 96-well plates using a Phoenix Liquid Handling System-Gryphon LCP (Art Robbins Instruments) and commercially available kits at 20 °C. For M^pro, crystals appeared after 1 day in 0.1 M Bis-Tris pH 6.5, 25% PEG 3350, which were cryo-protected using the reservoir solution and 30% PEG 400. Crystals of IMT M^pro were observed in several conditions. After optimization, crystals grown in 0.1 M MES pH 6.5, 10% 2-propanol, 20% PEG Smear Low (BCS Screen, Molecular Dimensions) were used as seeds for the diffraction crystals, obtained in 0.1 M MES pH 6.7, 5% DMSO, 8% PEG 4000.²² Crystals of M^pro in orthorhombic crystal system were obtained using seeds of IMT M^pro combined with M^pro at the condition 0.1 M MES pH 6.7, 5% DMSO, 8% PEG 4000 at 20 °C. Crystals of C145S M^pro tetrameric were obtained after 20 days in 0.1 M phosphate/citrate, pH 5.5, 20% v/v PEG Smear High (BCS Screen, Molecular Dimensions) at 20 °C. Crystals of C145S M^pro monomeric were obtained after 20 days in 0.1 M Bis-Tris pH 6.5, 25% (w/v) PEG 3350 at 20 °C.

Data collection and processing at MANACA beamline

During the initial commissioning phase (July to October 2020) the MANACA (MAcromolecular Micro and NAno CrystAllography)²⁸ beamline adopted an emergency commissioning plan to deliver the basic instrumentation to perform data collection of SARS-CoV-2 related samples. Thus, during this period, the beamline has operated on a fixed-energy regime (9 keV) with manual crystal mounting, single-axis goniometry, beam flux estimated to be about 1 · 10¹¹ ph/s/10 mA at 9.15 keV and adjustable beam size from about 18 (h) × 20 (v) µm² to 100 (h) × 80 (v) µm² (FWHM). This project was the first external user session at MANACA beamline and the first operating beamline at Sirius (SIRIUS, Brazil). The focus was optimised to 61 (h) × 48 (v) µm² at sample position (Figure S3). Even without the full capabilities, the beamline opening was very important to SARS-CoV-2 structural biology studies.

X-ray data for apo IMT M^pro was collected from three isomorphous independent crystals, that were processed by XDS via autoPROC.29, 30 Data herein was used for confirm reliability of the beamline (Figure S4 and Table S1). Datasets were then scaled and merged using Aimless³¹, and the resulting dataset was used for structural determination of IMT M^pro by molecular replacement with Phaser³² using PDB 5RGQ as template. Model was refined with COOT³³ and BUSTER³⁴ at 1.6 Å and deposited under the code 7KFI.

X-ray data for mature M^pro and C145S M^pro monomeric and tetrameric were processed by XDS via autoPROC29, 30 and scaled using Aimless.³¹ Mature M^pro, C145S M^pro tetrameric and C145S M^pro monomeric were solved by molecular replacement with Phaser³² using template models 5RGQ, 7KFI and 5R8T respectively. Mature M^pro and C145S M^pro monomeric and tetrameric were refined with COOT³³ and phenix.refine³⁵, and are respectively deposited under the codes 7KPH (at 1.4 Å), 7N5Z (at 1.7 Å)) and 7N6N (at 2.8 Å). Details of data processing parameters and statistics are given in Table S2.

Fragment screening of IMT M^pro

For the fragment screening of IMT M^pro, we used the settled plates fragment libraries of FragMAXlib (Talibov et al., to be published) and F2XEntry,36, 37 in a total of 192 fragments tested. In those plates, the content of each drop-well was resuspended in 1.0 µL of 0.1 M MES pH 6.7, 5% DMSO (v/v), 8% PEG 4,000 (w/v), 30% PEG 400 (w/v), and crystals were added afterwards. After 4 h soaking at room temperature, crystals were manually harvested and flash cooled for data collection.

During the commissioning phase of MANACA, 166 of those crystals were tested, leading to 77 usable datasets. To analyze the data, a simplified version of FragMAXapp was configurated in our laboratory end-station computer.³⁸ Within FragMAXapp, restrictions libraries were generated by phenix.eLBOW³⁹ using rm1 force field for geometry optimization, and datasets were processed through autoPROC/STARANISO or DIALS via XIA2.29, 40, 41 Molecular replacement and initial refinement were performed using DIMPLE⁴² using PDB 7KFI as template. To highlight electron density of weak binding events, map averaging and statistical modelling were performed by PanDDA software.⁴³ Models were refined with COOT³³ and phenix.refine.³⁵ Details of data processing and refinement statistics are given in Table S3. Polder maps of fragments are available in Figure S12.⁴⁴

Activity assays

All enzymatic assays were carried out using FRET-based substrate DABCYL-KTSAVLQ↓SGFRKM-E(EDANS)–NH₂ in assay buffer (20 mM Tris pH 7.3, 1 mM EDTA, 1 mM DTT). M^pro, IMT M^pro and C145S M^pro assays were performed at final concentration of 0.14 μM, 0.3 µM and 0.3 µM, respectively. Prior to reactions, enzymes were incubated in assay buffer at 37 °C for 10 min, followed by substrate addition. To determine the kinetics parameters (K_m, V_max and k_cat), the substrate was diluted to a range of concentrations from 100 μM to 0.78 μM. Initial velocity was derived from the slope of linear phase of each time-curse reaction, and Michaelis-Menten fitting was obtained using Origin Pro 9.5.1 Software (OriginLab). Relative efficiency of constructs was calculated by comparing the K_m/k_cat relative to M^pro. Fluorescence measures were performed in SpectraMax Gemini EM Microplate Reader with λ_exc/λ_emi of 360/460 nm, every 30 s over 60 min at 37 °C. All assays were performed in triplicates.

To test C145S M^pro auto-cleavage activity, 6xHis-tagged C145S M^pro was buffer exchanged in 20 mM Hepes pH 7.3, 100 mM NaCl, 1 mM DTT. Two reactions were prepared for comparison, one containing 10 µM C145S M^pro and other containing 30 µM C145S M^pro and 5 nM of mature M^pro. Aliquots of each reaction were collected for the period of 44 h. Samples were analyzed by SDS-PAGE on a 12.5% SDS polyacrylamide gel.

Differential scanning fluorimetry

For differential scanning fluorimetry assays (DSF), SYPRO Orange at 5X final concentration was added to protein diluted to 25 μM protein in gel filtration buffer. Denaturation curves were obtained ranging temperature from 25 °C to 75 °C by increasing one degree per cycle and fluorescence was measured in the end of each cycle. Experiment was conducted in a qPCR system Mx3000P (Agilent). The melting temperature was obtained using the Boltzmann fitting on Origin Pro 9.5.1 Software (OriginLab).

In solution oligomeric state of M^pro constructs

The in solution oligomeric states of the purified samples were evaluated by size exclusion chromatography coupled with multi-angle light scattering (SEC-MALS) in running buffer composed by 20 mM Tris-HCl pH 7.8 and 100 mM NaCl. For that, 50 µL of each M^pro construct at concentration of 50 µM were injected in a Waters 600 HPLC system (Waters) coupled in-line with an UV detector, a miniDAWN TREOS multi-angle light scattering apparatus (Wyatt Technology), a column Superdex 200 Increase 10/300 GL (GE Healthcare), and a refractive index detector Optilab T-rEX (Wyatt Technology). The light-scattering detectors were normalized with bovine serum albumin (Sigma-Aldrich). Data were collected and analyzed with the ASTRA 7 integrated software provided by Wyatt. The flow rate used was 0.5 mL/min. Results are summarized in Table S4.

For dimer formation monitoring, in solution oligomeric states of C145S M^pro monomer peak were determined similarly by SEC-MALS using a Superdex 75 Increase 10/300 GL (GE Healthcare). Two reactions were conducted simultaneously one containing 60 µM C145S M^pro monomer and other containing 60 µM C145S M^pro monomer and 10 nM M^pro (ratio 6000:1). Both reactions were maintained at room temperature and samples of 50 µL were injected with 24 h intervals for the period of 72 h.

Accession numbers

Structure factors and atomic coordinates have been deposited with the protein data bank with accession codes PDB ID 7KFI, 7MBG, 7KPH, 7N6N, 7LFE, 7LDX, 7LFP, 7KVL, 7N5Z and 7KVR. Other data are available from the corresponding author upon reasonable request.

CRediT authorship contribution statement

G.D. Noske: Conceptualization, Methodology, Software, Data curation, Writing - original draft. A.M. Nakamura: Conceptualization, Methodology, Software, Data curation, Writing - original draft. V.O. Gawriljuk: Conceptualization, Methodology, Software, Data curation, Writing - original draft. R.S. Fernandes: Conceptualization, Writing - original draft. G.M.A. Lima: Software. H.V.D. Rosa: Investigation, Methodology. H.D. Pereira: Investigation. A.C.M. Zeri: Investigation. A.A.F.Z. Nascimento: Investigation. M.C.L.C. Freire: Writing - original draft. D. Fearon: Investigation. A. Douangamath: Investigation. F. von Delft: Investigation. G. Oliva: Supervision, Funding acquisition. A.S. Godoy: Conceptualization, Methodology, Software, Data curation, Investigation, Writing - original draft, Writing - review & editing, Software, Validation.

Edited by Eric O. Freed

Appendix A. Supplementary material

The following are the Supplementary data to this article: