key: cord-271259-6kkzh1tp
authors: Chen, Shuai; Jonas, Felix; Shen, Can; Higenfeld, Rolf
title: Liberation of SARS-CoV main protease from the viral polyprotein: N-terminal autocleavage does not depend on the mature dimerization mode
date: 2010-01-01
journal: Protein & Cell
DOI: 10.1007/s13238-010-0011-4
sha: 
doc_id: 271259
cord_uid: 6kkzh1tp

The main protease (M(pro)) plays a vital role in proteolytic processing of the polyproteins in the replicative cycle of SARS coronavirus (SARS-CoV). Dimerization of this enzyme has been shown to be indispensable for transcleavage activity. However, the auto-processing mechanism of M(pro), i.e. its own release from the polyproteins through autocleavage, remains unclear. This study elucidates the relationship between the N-terminal autocleavage activity and the dimerization of “immature” M(pro). Three residues (Arg4, Glu290, and Arg298), which contribute to the active dimer conformation of mature M(pro), are selected for mutational analyses. Surprisingly, all three mutants still perform N-terminal autocleavage, while the dimerization of mature protease and transcleavage activity following auto-processing are completely inhibited by the E290R and R298E mutations and partially so by the R4E mutation. Furthermore, the mature E290R mutant can resume N-terminal autocleavage activity when mixed with the “immature” C145A/E290R double mutant whereas its trans-cleavage activity remains absent. Therefore, the N-terminal auto-processing of M(pro) appears to require only two “immature” monomers approaching one another to form an “intermediate” dimer structure and does not strictly depend on the active dimer conformation existing in mature protease. In conclusion, an auto-release model of M(pro) from the polyproteins is proposed, which will help understand the auto-processing mechanism and the difference between the autocleavage and trans-cleavage proteolytic activities of SARS-CoV M(pro).

The global outbreak of the highly infectious Severe Acute Respiratory Syndrome (SARS) between November, 2002, and June, 2003 , was caused by a new human coronavirus, SARS coronavirus (SARS-CoV) (Drosten et al., 2003; Peiris et al., 2003) . SARS-CoV is an enveloped, positive-sense RNA virus and involves the largest viral RNA genome known to date, encoding several structural and auxiliary proteins as well as two large overlapping polyproteins, pp1a (replicase 1a, around 450 kDa) and pp1ab (replicase 1ab, around 750 kDa) necessary for viral RNA synthesis (Marra et al., 2003; Thiel et al., 2003) . These two polyproteins are cleaved extensively by the main protease (M pro , also often called 3Clike protease) and the papain-like cysteine protease (PL2 pro ), both of which are encoded by the viral genome, to yield a multi-subunit protein complex termed "viral replicasetranscriptase" Ziebuhr, 2004; Groneberg et al., 2005) . Because of its functional indispensability in viral replication, the main protease (M pro ) has become an attractive target in developing inhibitors directed at SARS-CoV and other coronaviruses .

SARS-CoV M pro exists as a dimer in all crystal structures of the wild-type enzyme determined so far (Yang et al., 2003; Hsu et al., 2005a; Lee et al., 2005; Tan et al., 2005) and dimerization is also observed in solution, in a concentrationdependent manner (Chou et al., 2004; Fan et al., 2004; Shi et al., 2004; Chen et al., 2005; Graziano et al., 2006b) . Structurally, two monomers orient perpendicular to one another to form a dimer. Each monomer contains three domains ( Fig. 1A ) (Yang et al., 2003) : domains I (residues 8-101) and II (residues 102-184) form a chymotrypsin fold and domain III (residues 201-306) is an antiparallel globular cluster of five α-helices connected to domain II by a long loop (residues 185-200) . The substrate-binding pocket is located in a cleft between domains I and II and the active site consists of a Cys145-His41 catalytic dyad . In the active conformation of the dimer, the pocket is accessible for interaction with the respective amino-acid residues of the substrate and the oxyanion loop (residues 138-145) has the correct shape to donate two hydrogen bonds from main-chain amides to stabilize the tetrahedral transition-state of the proteolysis reaction. To date, numerous experimental results have indicated that only the dimer is the biological functional form of SARS-CoV M pro and the monomer is considered enzymatically inactive (Chou et al., 2004; Hsu et al., 2005a, b; Barrila et al., 2006; Graziano et al., 2006a; Chen et al., 2008b) . It has also been revealed that the activity loss of the dissociated monomer is mainly due to the collapse of the oxyanion hole and the S1 substrate-binding subsite .

Since the dissociated monomer of M pro is inactive, the dimer interface has been suggested as another potential target for inhibitor design (Shi et al., 2004; Hsu et al., 2005a) . As revealed by the crystal structure, the dimer interface of the protease mainly involves the following interactions : (i) between the N-terminal finger (residues 1-7) of one monomer and residues near the S1 subsite of the other Figure 1 . The overall dimeric structure and detailed dimer interface of mature SARS-CoV M pro . (A) A ribbon diagram for the dimer structure of mature SARS-CoV M pro (PDB code: 1UK4). Monomers A and B are represented in green and cyan, respectively, and the three domains are labeled. The dimer interface mainly involves the interactions: (B) between the S1 subsite of monomer A (green) and the N-terminal finger (cyan) of monomer B, (C) between the N-terminal finger and the S1 subsite of monomer A and the helical domain III of monomer B, (D) between the two A' α-helices of each monomer. The residues selected for site-directed mutations in this study are marked in red. monomer, in particular the oxyanion loop, (ii) between the Nterminal finger and the helical domain III from each monomer, (iii) between the two A' α-helices (residues 10-15), one from each monomer, that immediately follow the N-terminal finger. The contributions of several individual residues on the dimer interface to dimerization and enzymatic activity of M pro have been identified by several groups (Chou et al., 2004; Hsu et al., 2005b; Barrila et al., 2006; Shi and Song, 2006; Wei et al., 2006; Chen et al., 2008a, b; Lin et al., 2008) . Firstly, residues on the N-terminal finger were considered to play an important role in both dimerization and activity (Chen et al., 2005; Hsu et al., 2005b; Chen et al., 2008b) . In particular, Arg4 of the N-terminal finger can form a salt-bridge with residue Glu290 of the other monomer, which is vital for stabilizing the dimer structure to maintain the correct conformation of the active site (Chou et al., 2004) . Furthermore, residues of domain III have also been found to extensively mediate dimerization and to be responsible for positioning the N-terminal finger to interact with the active site of the neighboring monomer (Shi and Song, 2006; Lin et al., 2008) . Residue Arg298 in the C-terminal helix (residues 293-301) has been identified as a key component for maintaining the dimer conformation and its mutation was found to trigger a structural switch from dimer to monomer (Shi et al., 2008) . In addition, the A' α-helix (residues 10-15) is another critical part of the dimer interface since mutation of Gly11 can also result in complete dimer dissociation, as shown by X-ray crystallography (Chen et al., 2008a) .

Before proteolytic processing of the viral polyproteins pp1a and pp1ab into a total of 15 or 16 non-structural proteins (Nsp) occurs, SARS-CoV M pro itself is embedded in these polyproteins as the Nsp5 domain. On both sides, it is flanked by putative transmembrane (TM) domains (Nsp4 and Nsp6) that are anchored to the double-membrane vesicles where viral replication takes place in the infected host cell (Snijder et al., , 2006 Knoops et al., 2008) . Therefore, the M pro has to first liberate itself from the polyproteins through autocleavage, and then the self-released mature M pro would form a dimer and trans-cleave pp1a and pp1ab at other sites. Despite being the first and essential step for viral maturation, the autocleavage mechanism of M pro has only been rarely addressed Shan et al., 2004; Hsu et al., 2005a) and remains poorly characterized. On the other hand, the relationship between dimerization and enzymatic activity of the mature M pro is well documented (Chou et al., 2004; Shi et al., 2004; Hsu et al., 2005b; Graziano et al., 2006a, b; Chen et al., 2008a; Lin et al., 2008) . Since most of these studies use a protease preparation that resembles the sequence of mature M pro after auto-processing and formation of the final dimer structure, all these results are only relevant for the trans-cleavage activity of the mature enzyme.

In this paper, we focus on the correlation between the Nterminal autocleavage activity and dimerization of "immature" SARS-CoV M pro , which has not been reported yet. A total of three residues involved in maintaining the dimer conformation of mature M pro are selected for mutational analyses, i.e., Arg4, Glu290, and Arg298 ( Fig. 1B and 1C ). In the crystal structure of the mature dimer, the side-chain guanidyl of Arg4 forms a salt bridge with the side-chain of Glu290 of the neighboring monomer in the dimer. Mutation of the conserved Glu290 has been reported to cause dimer dissociation of M pro , leading to an inactive monomer in solution, while mutation of the not absolutely conserved Arg4 shifts the dimer-monomer equilibrium and induces a significant decrease of trans-cleavage activity (Chou et al., 2004) . Furthermore, mutation of Arg298 has also been shown to trigger dimer dissociation in both solution and crystal as well as a complete loss of activity (Shi et al., 2008) . As the construction strategy of all the plasmids used in the study presented here, a SARS-CoV M pro autocleavage site (AVLQ↓S) was introduced between the N-terminal tag and the N-terminal residue (Ser1) of the protease. Thus, the Nterminal autocleavage activity of wild type and mutated M pro s can be evaluated by the extent to which the N-terminal tag is removed. Our results surprisingly reveal that all three mutants can still perform autocleavage during protein production and purification. In the following, we also investigate the effects of these mutations on dimerization and trans-cleavage activity of mature SARS-CoV M pro . Consistent with the published results, dimer formation of mature protease in solution is completely abolished by the E290R and R298E mutations and partially so by the R4E mutation, resulting in an entire or dramatic loss of trans-cleavage activity. Furthermore, the reconstructed cleavage assay indicates that the mature E290R mutant can resume N-terminal autocleavage activity when mixed with the "immature" C145A/E290R double mutant whereas its trans-cleavage activity remains absent. These results indicate that N-terminal autocleavage of SARS-CoV M pro from the polyproteins only requires two "immature" proteases approaching one another to form an "intermediate" dimer structure and does not depend on the active dimer conformation existing in the mature protease. The present study is expected to help us better understand the maturation mechanism and the difference between autocleavage and trans-cleavage proteolytic processing of SARS-CoV M pro .

Autocleavage of the N-terminally tagged wild-type and mutated SARS-CoV M pro during lysate preparation

The goal of this study was to elucidate the relationship between the N-terminal autocleavage activity and dimerization of "immature" SARS-CoV M pro . Three residues (Arg4, Glu290, and Arg298), which are involved in maintaining the integrity of the dimer interface of mature SARS-CoV M pro ( Fig. 1B and 1C ), were selected for mutational analysis. The Arg4…Glu290 salt bridge has been demonstrated to be a key element in dimerization of mature M pro and residues on domain III (e.g., Arg298, Gln299 et al.) have also been shown to be vital regulators of dimerization and trans-cleavage activity (Chou et al., 2004; Shi and Song, 2006; Lin et al., 2008) . Structurally, domain III of monomer B in the dimer not only directly interacts with the opposite S1 subsite (Gln299B…Ser139A, Arg298B…Ser123A; Fig. 1C ), but also helps its own N-terminal finger to properly insert into the counterpart monomer A by making both inter-monomer (Arg4B…Glu290A) and intra-monomer (Arg298B…Met6B) interactions (Fig. 1B ). Since these residues contribute to the active dimer conformation of mature M pro mainly by electrostatic interactions, we mutated them into oppositely charged residues and evaluated the influence of these mutations on the N-terminal autocleavage activity of M pro . An M pro autocleavage site (AVLQ↓S), corresponding to the Cterminus of Nsp4 which precedes the M pro (Nsp5) in the viral polyproteins pp1a and pp1ab, was inserted between the Nterminal GST tag and the first residue (Ser1) of the protease ( Fig. 2A ) to resemble the "immature" M pro before autocleavage. In addition, 24 nucleotides coding for eight extra residues (GPH 6 ) were added at the C-terminus of M pro for purification convenience. Using this construct, the N-terminal autocleavage activity of M pro was evaluated in terms of the extent of removal of the GST tag during gene expression and protein purification. As shown by SDS-PAGE analysis ( Fig. 2B) , only one purified protein band with an M r of approximately 33.0 kDa was obtained in the eluants of wildtype (WT) M pro from the Ni-NTA affinity column, well in agreement with the molar mass calculated from the aminoacid sequence (33.8 kDa) of mature M pro . This indicated that the N-terminal GST fusion tag (~26 kDa) had been removed by WT protease through autocleavage. As a negative control (Fig. 2C) , we mutated the catalytic residue Cys145 to Ala. The C145A mutant exhibited a band with an M r of around 59 kDa, suggesting that it still existed as a GST-fusion protein and excluding the possibility that the autocleavage of WT protease was caused by non-specific side reactions in the E. coli expression system used to produce the proteins. Subsequently, autocleavage assays were carried out for all three mutants using the same procedure. For the R4E mutant ( Fig. 2D ), the autocleavage activity was not affected since the GST tag was cleaved off completely. It has been reported (Chou et al., 2004 ) that mutation of the highly, but not absolutely conserved Arg4 to Ala can result in unstable dimers while the mutant still maintains a trans-cleavage activity comparable to that of WT protease. Thus, it is possible that the "immature" R4E mutant can still perform efficiently in N-terminal autocleavage during expression, as a small amount of unstable dimers might form. On the other hand, mutation of the totally conserved Glu290 to Ala was reported to induce complete dimer dissociation of M pro in solution (Shi and Song, 2006) and the Arg298Ala mutation also produced a monomeric structure in the crystal (Shi et al., 2008) , thereby inducing complete loss of trans-cleavage activity. These results clearly indicated that Glu290 and Arg298 are key factors in maintaining the dimeric form of mature M pro . Surprisingly, as shown in Fig. 2E and 2F, both the E290R and R298E mutants displayed no obvious difference of autocleavage behavior compared to WT protease. Since mutation of either of these two residues were reported to completely abolish the dimer of mature M pro , our finding raises the intriguing question of how the E290R and R298E mutants can auto-process their N-terminal GST tags when they are unable to form the active dimer structure. In order to better explain these results and elucidate the autocleavage mechanism of M pro , the contributions of these residues to dimerization and trans-cleavage activity of mature M pro needed to be further investigated.

The GST protein alone also exists as a homodimer in the crystal (Nishida et al., 1998) and the N-terminal autocleavage of the E290R or R298E mutant might possibly be due to dimerization of the GST-fused M pro through the N-terminal GST tag. For excluding this possibility, WT M pro and all the mutants were also subcloned into the vector pET-32a ( Fig. 2A) . This construct encodes an N-terminal thioredoxin (Trx) tag, which does not form dimers by itself (Jeng et al., 1994) . As shown by SDS-PAGE ( Fig. 2G and 2H), two purified protein bands corresponding to cleaved M pro (~33 kDa) and Trx (~19 kDa including an additional in-frame sequence from the vector) were observed for both WT and R298E mutant, suggesting that the Trx-tagged R298E mutant maintains autocleavage activity comparable to WT protease. These findings therefore clearly indicate that the N-terminal fusion tag has no impact on autocleavage of M pro and further validate the reliability of our autocleavage assay.

Folding behavior and dimerization features of mature WT and mutated SARS-CoV M pro

As described above, all three mutants maintain efficient Nterminal autocleavage activity despite the residues we selected for mutation, especially E290 and R298, have been reported to be indispensable for dimerization of mature M pro (Shi and Song, 2006; Shi et al., 2008) . In view of this apparent inconsistency, we further investigated the folding behavior and dimerization features of all mature mutants after autocleavage. Fig. 3A shows the fluorescence emission spectra of mature WT and the mutants of M pro . The emission λ max of WT M pro is around 330 nm. Similar to WT protease, all three mutants show only minor differences in the emission λ max (varying from 329 to 331 nm), demonstrating that mutation of a single residue on the dimer interface into one carrying the opposite charge has not changed the fold of the protease. In addition, the far-UV CD spectra of mature WT and mutated M pro are also similar to one another (Fig. 3B) . All spectra feature a positive peak around 200 nm and dual negative peaks at 209 and 222 nm, typical of proteins containing significant amounts of α-helix and β-sheet. These data suggest that all three mutants have well-defined secondary structure and exclude the possibility of misfolding caused by the mutation of individual residues.

The dimerization of mature M pro has been successfully characterized by various biochemical and biophysical methods (Shi et al., 2004; Chen et al., 2005; Hsu et al., 2005b; Graziano et al., 2006b) . According to a published method (Prakash et al., 2002) , we performed a chemical cross-linking analysis of mature WT M pro . When incubated with 0.01% glutaraldehyde, the protease at a concentration of 0.1 mg/mL predominantly displayed the monomeric form near 33.0 kDa, with a minor band around 66.0 kDa corresponding to the dimer (Fig. 4A, lane 5) . With increasing protein concentration, the dimeric form increased and was almost equivalent to the amount of the monomeric form at 1 mg/mL (Fig. 4A , lanes 6-8). When using a higher concentration of glutaraldehyde (0.1%), a similar cross-linking pattern of the protease was observed with slightly higher efficiency, as the dimer had become the predominant species at the same protein concentrations (Fig. 4A, lanes 1-4) . These results further demonstrate that mature WT M pro possesses a dimermonomer equilibrium in solution and its dimerization is concentration-dependent, in good agreement with the literatures (Chou et al., 2004; Shi et al., 2004; Chen et al., 2005; Graziano et al., 2006b) . Nevertheless, the possibility of minor artificial cross-linking effects might still exist due to the appearance of high-order multimers in SDS-PAGE (Fig. 4A) . However, since the cross-linking analyses of all three mutants can be performed under exactly the same experimental conditions, this method is still very useful for examining the effects of these mutations on dimerization of mature M pro . As indicated in Fig. 4B , the R4E mutant showed an obvious tendency to monomer and the amount of the dimer decreased significantly at lower protein concentrations compared to WT protease, which is consistent with the published result that mutation of Arg4 would result in weakened dimerization of M pro (Chou et al., 2004) . For the E290R and R298E mutants ( Fig. 4C and 4D) , dimerization seemed to be disrupted even more severely than for the R4E mutant since almost no dimer can be observed at low protein concentrations, further demonstrating the indispensability of Glu290 and Arg298 for maintaining the dimeric structure. However, a very small amount of dimeric form of the E290R and R298E mutants were visible in SDS-PAGE at relatively high protein concentrations ( Fig. 4C and 4D, lanes 7, 8) . This might be the real dimer of M pro or just a dimeric product caused by minor artificial cross-linking effects as mentioned above.

Subsequently, size-exclusion chromatography (SEC) analyses were carried out to evaluate the dimer-monomer equilibria of the three mutants more precisely. In view of the appearance of a dimeric form of the mutants at high concentration in the cross-linking analyses, we used a protein concentration of 5 mg/mL for each SEC run, which is much higher than the highest concentration used for cross-linking. The physical state corresponding to the native monomer or dimer of M pro in the gel-filtration column was then monitored. As shown in Fig. 5 , WT M pro eluted as a single peak with a retention volume of 57.01 mL. The elution profiles of four molecular-mass marker proteins confirmed that the peak represented the dimeric species of M pro (Table 1 , estimated M r :~66.7 kDa), implying that in solution, WT protease exists almost exclusively as a dimer at relatively high protein concentration. In comparison, the R4E mutant eluted at a higher volume (61.86 mL), consistent with it being a mixture of dimer and monomer, with the monomer being the predominant species (Table 1 , estimated M r :~47.6 kDa). This further supports the cross-linking result and the literature report that Arg4 affects dimerization only to a moderate extent (Chou et al., 2004) . For E290R and R298E mutants, the elution volumes dramatically shifted to 66.47 and 64.79 mL, corresponding to estimated M r values of 34.6 kDa and 38.9 kDa, respectively, clearly indicating a monomeric state in solution. The data therefore demonstrate that individual mutation of Glu290 or Arg298 is sufficient to prevent dimerization of mature M pro even at a high protein concentration, which convincingly supports the published results (Shi and Song, 2006; Shi et al., 2008) and also suggests that the very minor amounts of E290R and R298E dimers that appeared in SDS-PAGE after cross-linking ( Fig. 4C and 4D) is likely due to a small extent of nonspecific cross-linking. 

Numerous published data have proposed that only the dimer is the biologically functional form of mature M pro and the dissociated monomer has no trans-cleavage activity. Since mutation of a single residue on the dimer interface could result in partial or complete dimer dissociation of mature M pro as demonstrated above, the three mutants were expected to show decreased or no trans-cleavage activity. To verify this prediction, we performed trans-cleavage assays of mature WT protease and the mutants using a fluorogenic peptide substrate (Verschueren et al., 2008) . As shown in Fig. 6 , the fluorescence increase following hydrolysis of the substrate by WT M pro was significant and time-dependent, implying that the protease trans-cleaved the substrate efficiently, whilst the C145A mutant showed negligible activity against the substrate, thus proving the validity of the trans-cleavage assay. Similar to the C145A mutant, the trans-cleavage activities of the three "dimerization mutants" at a concentration of 0.1 μM were all below the detection limit of the assay during the entire measuring process (Fig. 6) , even when the reaction time was prolonged to overnight (data not shown). Next, we increased the concentration of the mutants up to 1 μM, and the data showed that the R4E mutant displayed a slight increase of activity but still no measurable trans-cleavage reactivity was detected for the E290R and R298E mutants (see inset of On the other hand, the R4E mutant exhibits a weak activity when increasing protein concentration, well in agreement with its unstable dimerization in solution.

Reconstruction of N-terminal autocleavage of M pro with mature E290R and "immature" C145A/E290R mutants According to all the results described above, mutation of residues E290 or R298 seemed to have no obvious influence on the N-terminal autocleavage of "immature" M pro , although it did cause complete dimer dissociation and disruption of trans-cleavage activity of the mature form of the mutated M pro . This surprising observation led to the interesting conclusion that the N-terminal autocleavage of SARS-CoV M pro is not dependent on dimerization, at least it does not require the final dimer conformation existing in the mature protease. On the other hand, this final dimer conformation is indispensable for trans-cleavage activity of mature M pro . In order to prove this hypothesis, an N-terminal autocleavage assay was constructed. The inactive C145A/E290R double mutant, which cannot perform autocleavage and still contains the Nterminal GST tag, was prepared as an "immature" substrate for mature WT M pro and the E290R mutant. The cleavage process can be easily followed from the appearance of free GST in SDS-PAGE. As shown in Fig. 7A , GST-C145A/E290R was cleaved efficiently by mature WT M pro . The band (~26 kDa) corresponding to the cleaved GST tag appeared after 5 min of incubation time (lane 2) but not in the substrate-only lane (lane 1), and the cleavage was totally completed within 2 hours (lanes 3-6). Since mature WT M pro can form the active dimer even at a low concentration in solution, the free GST tag is the product of trans-cleavage of the "immature" substrate by the WT M pro dimer (Fig. 7C) . As a control, an unrelated protein (the effector domain of influenza A virus nonstructural protein 1, H1N1 NS1 ED, Fig. 7A , lane 8) containing an N-terminal His tag and an M pro cleavage site (AVLQ↓S), was also prepared as a trans-cleavage substrate for WT M pro . Compared to GST-C145A/E290R, a similar cleavage pattern was observed for His-H1N1 NS1 ED and the cleaved substrate (~17 kDa) was clearly visible in SDS-PAGE (lanes 9-13), further demonstrating the high trans-cleavage activity of the WT M pro dimer. With the mature E290R mutant, quite a different cleavage behavior was observed between His-H1N1 NS1 ED and GST-C145A/E290R (Fig. 7B) . The mature E290R mutant showed no detectable cleavage activity with His-H1N1 NS1 ED during the entire measuring procedure (lanes 9-13). As His-H1N1 NS1 ED is a transcleavage substrate for M pro (see above; Fig. 7D ), this result is well in agreement with the conclusion that the mature E290R mutant exists as a monomer in solution as revealed by SEC analysis (Fig. 5) , without any trans-cleavage activity (Fig. 6) .

On the other hand, when mixing the E290R mutant with GST-C145A/E290R, the cleaved GST tag appeared after 2 h (lane 3) and cleavage was almost complete after overnight incubation (lanes 4-6), indicating that the mature E290R mutant can still cleave "immature" GST-C145A/E290R, although the efficiency is much lower than for WT protease.

Since the mature E290R monomer cannot trans-cleave His-H1N1 NS1 ED (Fig. 7B ) nor the M pro peptide substrate (Fig. 6) at all, the cleavage of GST-C145A/E290R by the mature E290R monomer might be due to the reconstruction of only the N-terminal autocleavage by these two M pro monomers (Fig. 7D ). This strongly supports the idea that autocleavage of coronavirus M pro occurs in trans (Anand et al., 2003) . Furthermore, as the E290R mutation completely abolishes formation of the mature dimer in solution (Fig. 5) , these data strongly support the hypothesis that the active dimer conformation existing in mature WT protease is unnecessary for N-terminal autocleavage of M pro .

How does "immature" M pro release itself from the precursor polyproteins without forming the mature dimer structure?

What we find puzzling is why the N-terminal autocleavage activity of M pro is independent from the dimer structure that is vital to trans-cleavage activity of the protease? What is the difference between the proteolytic autocleavage and transcleavage mechanisms of SARS-CoV M pro ? It has been reported (Shi et al., 2004 ) that domain III of M pro itself possesses an intrinsic tendency to form a dimer even at a very low concentration. Recently, the crystal structure of domain III alone has been determined, revealing a novel 3D domain-swapped dimer structure (Zhong et al., 2009) . Furthermore, It was reported (Zhong et al., 2008) that the Nterminal finger deletion mutant of M pro also forms a novel stable dimer through domain III-domain III interactions, which is different from the dimer interface observed in the WT M pro dimer. These results indicate that in addition to the dimer conformation in the crystal structure of WT protease, M pro might have other possible dimerization modes. M pro is still a part of pp1a and pp1ab when the polyproteins are produced in coronavirus-infected cells. Through the hydrophobic domains of Nsp4 and Nsp6 that flank the M pro (Nsp5) domain, the unprocessed polyproteins are anchored to the double-membrane vesicles found in infected host cells. Subsequently, when these precursor proteins accumulate to high concentration, the "immature" M pro can release itself from the polyproteins by inter-molecular cleavage, and then the self-released mature M pro triggers the trans-cleavage processing of the polyproteins (Anand et al., 2003) . In the dimer structure of mature M pro , the N-terminal finger of one monomer can form intensive interactions with the S1 subsite and Domain III (the C-terminal helix) of the other monomer ( Fig. 1B and 1C) , which are key residues for maintaining the active dimer conformation. It is very unlikely that the "immature" M pro would form such dimer because of the two TM domains flanking it at both N-and C-termini before selfreleasing (Oostra et al., 2008) . In our proposed model (Fig. 8) , the N-terminal autocleavage might only need two "immature" M pro domains in monomeric polyproteins approach one another to form an "intermediate" dimer, possibly through Domain III-III dimerization, which also agrees well with the plausible dimer formation pathway proposed for the M pro previously (Chen et al., 2008a; Hu et al., 2009 ). The formation of the "intermediate" dimer could trigger the rotation of their domains I/II relative to domains III and thereby insert their "uncleaved" N-termini into the substrate-binding pockets of the opposite monomer, which might induce the active conformation of the S1 subsites through an induced-fit catalytic mechanism. Once the autocleavage is finished, the "cleaved" N-terminal fingers should slip away from the active sites and switch to their final spatial positions, which is observed in the mature dimer of WT M pro (Fig. 1B) , thereby locking the dimer in a catalytic competitive state. The conformational shift of the N-terminal finger from the auto-processed state to the final "cleaved" state is also supported by the crystal structure of the M pro H41A mutant in complex with an N-terminal auto-cleavage substrate (Xue et al., 2008) , indicating an estimated distance of~20 Å between residues P1'-P5' (SGFRK) of the bound substrate and the free N-terminal finger of the mature dimer (Fig. 9A) . Meanwhile, Domains III should also glide to their "final" positions to produce the mature active dimer since the last helices of domains III are tethered with the N-terminal fingers in the mature dimer structure (Fig. 1C) . The dimer conformation existing in mature M pro is not the cause but more likely the product of N-terminal auto-processing of "immature" protease from polyproteins. Thus mutation of residues that are indispensable for dimerization of mature M pro has little effect on N-terminal autocleavage since the conformation of the "intermediate" dimer is different from that of mature dimer. On the other hand, the trans-cleavage activity of the mature dimer is much higher than the autocleavage activity of the "intermediate" dimer, as indicated in Fig. 7A and 7B. However, the "intermediate" dimer might be the only feasible structure for "immature" M pro to form within pp1a and pp1ab. In order to perform the first step of autocleavage from the polyproteins, "immature" M pro possibly tolerates a less-than-ideal dimer.

Once the mature dimer is formed after N-terminal selfreleasing, the unprocessed C-terminus of one mature dimer can insert into an active site of another mature dimer for trans-cleavage (Fig. 8) , as revealed by a product-bound crystal structure of the M pro C145A mutant (Hsu et al., 2005a) (Fig. 9B and 9C ). After the formation of the final mature M pro dimer with authentic N-and C-termini, transcleavage of the polyproteins at other sites could subsequently be triggered at a high rate. In conclusion, our current work is expected to provide more insights into the auto-processing mechanism of SARS-CoV M pro . In addition, the proposed auto-releasing model should be further verified by X-ray crystallographic determination of the "intermediate" dimer structure. Figure 8 . A proposed mode of SARS-CoV M pro auto-release from the precursor polyproteins. The chymotrypsin fold (domains I and II) and domain III of M pro monomers in two polyproteins are shown as boxes and cylinders, respectively. The N-and C-termini are also labeled. The auto-release mode involves four steps.

Step 1, two "immature" M pro monomers approach one another and their domains III form an "intermediate" dimer structure, which triggers the rotations of their chymotrypsin folds and insert their "uncleaved" N-termini into the substrate-binding pockets of the opposite monomers.

Step 2, with an induced-fit catalytic mechanism, the active site of M pro could be activated and catalyze the N-terminal autocleavage. Afterwards, the "cleaved" Nterminal fingers slip away from the active sites and switch to their final spatial positions, as well as domains III, locking the mature dimer with "uncleaved" C-termini at the active state.

Step 3, the "uncleaved" C-terminus of one mature dimer can insert into an active site of another mature dimer, as supported by a product-bound crystal structure of the M pro C145A mutant (Hsu et al., 2005a) .

Step 4, once the C-terminus is processed by trans-cleavage, the final mature dimer with authentic N-and C-termini is formed, which is observed in the crystal structure of WT M pro (Xue et al., 2007) .

The plasmid of WT SARS-CoV M pro was a gift from Prof. Z. Rao (Xue et al., 2007) . Briefly, 12 nucleotides coding for the four amino acids AVLQ (corresponding to the P4-P1 residues of the N-terminal autocleavage site of SARS-CoV M pro ) were added to precede the N-terminal Ser1 residue of the protease. 24 nucleotides coding for eight extra residues (GPH 6 ) were added at the C-terminus of M pro . The resulting coding sequence was inserted into BamHI and XhoI sites of the pGEX-6P-1 plasmid. The plasmid was then transformed into E.coli BL21 (DE3) cells, and the N-terminal GST fusion protein with a C-terminal His tag was produced by introducing IPTG to 0.5 mM at 25°C for 6 h. Because of the plasmid construction strategy, the N-terminal GST tag can be cleaved off by autocleavage of active M pro . The purification of the C-terminally GPH 6 -tagged product was performed according to the published method (Xue et al., 2007) . The protein was first subjected to Ni-NTA affinity chromatography and concentrated in PreScission protease cleavage buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, and 1 mM DTT). The PreScission protease was then added to cleave off the C-terminal GPH 6 tag, producing an authentic mature SARS-CoV M pro . For the inactive M pro C145A mutant, which cannot perform autocleavage, the product was first purified by Ni-NTA column, followed by adding the PreScission protease to remove both the N-terminal GSTand the C-terminal GPH 6 tags. All the cleaved proteins were further purified using anionexchange chromatography. Finally, the purified and concentrated protease was dialyzed against 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM DTT, 1 mM EDTA and stored at K20°C.

Site-directed mutagenesis of SARS-CoV M pro was processed by a modified recombinant PCR method. The mutated proteases (Cys145Ala, Arg4Glu, Glu290Arg, Arg298Glu single mutants, and Cys145Ala/Glu290Arg double mutant) were prepared with the QuikChange II site-directed mutagenesis kit (Stratagene) using pGEX-6p-1-SARS-CoV M pro as a template. The nucleotide sequences of the primers used for mutation are given in Table 2 . The plasmids encoding mutated forms of M pro were verified by sequencing, and then E. coli BL21 (DE3) cells were transformed by the resulting plasmids. The mutated proteins were produced and purified using a procedure similar to that described above for WT protease. The purity and structural integrity of the mutated proteases were analyzed by SDS-PAGE.

All the gene fragments coding for WT and mutated proteases were also cleaved off from the plasmids of pGEX-6p-1-SARS-CoV M pro using BamHI and XhoI, and subcloned into the vector pET-32a with N-terminal thioredoxin (Trx) tag. The resulting plasmids were verified by sequencing and transformed into E. coli BL21 (DE3) cells. The proteins were produced and purified using a procedure similar to that for the GST fusion protease mentioned above.

According to the construction strategy of all the plasmids, a SARS-CoV M pro autocleavage site (AVLQ↓S) was located between the N-terminal GST or Trx tag and the first Ser1 residue of the protease (Fig. 2A) . The N-terminal GST or Trx tag can be removed by autocleavage of active M pro . Thus, the N-terminal autocleavage activity of WT and mutated M pro s can be evaluated by the removal extent of GST or Trx tag during the process of protein production and purification. The tag-cleaved protease was then checked using SDS-PAGE after Ni-NTA affinity chromatography purification.

The fluorescence measurements were performed on a Cary Eclipse fluorescence spectrophotometer. The protease samples were prepared in 50 mM Tris-HCl pH 7.5, 150 mM NaCl,1 mM DTT, and 1 mM EDTA, with a protein concentration of 5 μM. The fluorescence emission spectra from 300 to 420 nm were recorded after excitation at 280 nm, with a spectral slit width of 5 nm for excitation and emission. Fluorescence spectra of mature WT and mutated SARS-CoV M pro s were measured in a 1-mL quartz cuvette at 25°C. All final spectra were corrected for the buffer contribution, and were the average of three parallel measurements.

Far-UV CD spectra from 195 to 250 nm were recorded on a JASCO-715 spectropolarimeter. The protein samples were prepared in 50 mM Tris-HCl pH 7.5, 1 mM EDTA at 25°C, with a concentration of 10 μM. The spectra were collected with 1 nm band width using a 0.1-cm path length cuvette, and normalized by subtracting the baseline recorded for the buffer. Each measurement was repeated five times and the final result was the average of five independent scans. The CD spectra of mature mutated proteases were compared to that of WT SARS-CoV M pro to exclude the possibility of misfolding caused by mutation.

Chemical cross-linking was performed for mature WT and mutated SARS-CoV M pro s (final concentrations from 0.1 to 1.0 mg/mL in 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM DTT, and 1 mM EDTA) with glutaraldehyde. An aliquot of 25% (v/v) glutaraldehyde was added to the protein samples to give a final concentration of 0.1% or 0.01%, and the samples were incubated at 25°C for 20 min followed by quenching the reaction with the addition of 1.0 M Tris-HCl pH 8.0 (0.5%, v/v). Orthophosphoric acid was thereafter added to the Note: The mutant codons in the oligonucleotide sequences are highlighted in boldface. SARS-CoV M pro amino acids are numbered continuously from the N-terminal residue, Ser1, to the C-terminal residue, Gln306.

reaction mixture to result in precipitation of the cross-linked proteins. After centrifugation (12,000 rpm, 4°C), the precipitate was redissolved in loading buffer and heated at 70°C for 10 min. SDS-PAGE was run with 12.5% acrylamide gels.

The dimer-monomer equilibria of mature WT and mutated SARS-CoV M pro s were measured using size-exclusion chromatography on a HiLoad 16/60 Superdex 75 prep grade column through an ÄKTA FPLC system. Buffer used was 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM DTT, and 1 mM EDTA. Protein samples with a concentration of 5 mg/mL were loaded onto the column and then eluted with the buffer by detection of absorbance at 280 nm. The column was calibrated with four molecular-mass marker proteins (labeled in the inset of 

A FRET-based assay was used to assess the trans-cleavage activity of mature WT and mutated SARS-CoV M pro s (Verschueren et al., 2008) . The assay made use of a 14-amino acid fluorogenic substrate, Dabcyl-KTSAVLQ↓SGFRKME-EDANS (95% purity, Biosyntan GmbH, Berlin, Germany), which contains an optimal M pro cleavage site (indicated by the arrow). The enhanced emission fluorescence due to cleavage of this substrate as catalyzed by the protease was monitored as a function of time at 490 nm with excitation at 340 nm, using a Cary Eclipse fluorescence spectrophotometer. The experiments were performed in a buffer consisting of 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM DTT, and 1 mM EDTA at 25°C. The reaction was initiated by adding M pro (final concentration 0.1-1 μM) in a 96-well black microplate containing a final fluorogenic substrate concentration of 3 μM. The incubation of the substrate in the assay buffer without the protease was performed as a control and the transcleavage activity was the average of three parallel assays.

Coronavirus main proteinase: target for antiviral drug therapy

Coronavirus main proteinase (3CL pro ) structure: basis for design of anti-SARS drugs

Long-range cooperative interactions modulate dimerization in SARS 3CL pro

Severe acute respiratory syndrome coronavirus 3C-like proteinase N terminus is indispensable for proteolytic activity but not for enzyme dimerization. Biochemical and thermodynamic investigation in conjunction with molecular dynamics simulations

Mutation of Gly-11 on the dimer interface results in the complete crystallographic dimer dissociation of severe acute respiratory syndrome coronavirus 3C-like protease: crystal structure with molecular dynamics simulations

Residues on the dimer interface of SARS coronavirus 3C-like protease: dimer stability characterization and enzyme catalytic activity analysis

Quaternary structure of the severe acute respiratory syndrome (SARS) coronavirus main protease

Severe acute respiratory syndrome: identification of the etiological agent

Biosynthesis, purification, and substrate specificity of severe acute respiratory syndrome coronavirus 3C-like proteinase

Enzymatic activity of the SARS coronavirus main proteinase dimer

SARS-CoV main proteinase: The monomer-dimer equilibrium dissociation constant

Molecular mechanisms of severe acute respiratory syndrome (SARS)

Mechanism of the maturation process of SARS-CoV 3CL protease

Critical assessment of important regions in the subunit association and catalytic action of the severe acute respiratory syndrome coronavirus main protease

Two adjacent mutations on the dimer interface of SARS coronavirus 3C-like protease cause different conformational changes in crystal structure

3C-like proteinase from SARS coronavirus catalyzes substrate hydrolysis by a general base mechanism

High-resolution solution structures of oxidized and reduced Escherichia coli thioredoxin

SARS-coronavirus replication is supported by a reticulovesicular network of modified endoplasmic reticulum

Crystal structures of the main peptidase from the SARS coronavirus inhibited by a substrate-like aza-peptide epoxide

Characterization of trans-and ciscleavage activity of the SARS coronavirus 3CL pro protease: basis for the in vitro screening of anti-SARS drugs

Correlation between dissociation and catalysis of SARS-CoV main protease

The genome sequence of the SARS-associated coronavirus

Three-dimensional structure of Escherichia coli glutathione S-transferase complexed with glutathione sulfonate: catalytic roles of Cys10 and His106

Topology and membrane anchoring of the coronavirus replication complex: not all hydrophobic domains of nsp3 and nsp6 are membrane spanning

Coronavirus as a possible cause of severe acute respiratory syndrome

Unique oligomeric intermediates of bovine liver catalase

A novel auto-cleavage assay for studying mutational effects on the active site of severe acute respiratory syndrome coronavirus 3C-like protease

The catalysis of the SARS 3C-like protease is under extensive regulation by its extra domain

Mechanism for controlling the dimer-monomer switch and coupling dimerization to catalysis of the severe acute respiratory syndrome coronavirus 3C-like protease

Dissection study on the severe acute respiratory syndrome 3C-like protease reveals the critical role of the extra domain in dimerization of the enzyme: defining the extra domain as a new target for design of highly specific protease inhibitors

Unique and conserved features of genome and proteome of SARS-coronavirus, an early split-off from the coronavirus group 2 lineage

Ultrastructure and origin of membrane vesicles associated with the severe acute respiratory syndrome coronavirus replication complex

pH-dependent conformational flexibility of the SARS-CoV main proteinase (M pro ) dimer: molecular dynamics simulations and multiple X-ray structure analyses

Mechanisms and enzymes involved in SARS coronavirus genome expression

A structural view of the inactivation of the SARS coronavirus main proteinase by benzotriazole esters

The N-terminal octapeptide acts as a dimerization inhibitor of SARS coronavirus 3C-like proteinase

Production of authentic SARS-CoV M pro with enhanced activity: application as a novel tagcleavage endopeptidase for protein overproduction

Structures of two coronavirus main proteases: implications for substrate binding and antiviral drug design

The crystal structures of severe acute respiratory syndrome virus main protease and its complex with an inhibitor

C-terminal domain of SARS-CoV main protease can form a 3D domain-swapped dimer

Without its N-finger, the main protease of severe acute respiratory syndrome coronavirus can form a novel dimer through its C-terminal domain

Molecular biology of severe acute respiratory syndrome coronavirus

This work was supported, in part, by the Sino-European Project on SARS Diagnostics and Antivirals (SEPSDA, contract NO. SP22-CT-2004-003831) and by VIZIER (contract no. LSHG-CT-2004-511960), both funded by the European Commission. We acknowledge support from the Sino-German Center for the Promotion of Research, Beijing (grant no. 233 (202/6)), the Schleswig-Holstein Innovation Fund, and the DFG (Hi 611/4 and Cluster of Excellence "Inflammation at Interfaces"). We thank Professor Z. Rao (Tsinghua University, Beijing) for providing the plasmid of WT SARS-CoV M pro , Dr. Jinzhi Tan for discussion, and Silke Schmidtke for expert technical assistance.

ABBREVIATIONS CD, circular dichroism; Dabcyl, phenyl]azo] benzoic acid; EDANS, 5-[(2'-aminoethyl)-amino] naphthelene sulfonic acid; FRET, fluorescence resonance energy transfer; GST, glutathion S-transferase; M pro , main protease; Nsp, non-structural protein; SARS-CoV, severe acute respiratory syndrome coronavirus; SEC, size-exclusion chromatography; TM, transmembrane; Trx, thioredoxin; WT, wild type