key: cord-1011914-vgnxhksv
authors: Lee, Ting-Wai; Cherney, Maia M.; Liu, Jie; James, Karen Ellis; Powers, James C.; Eltis, Lindsay D.; James, Michael N.G.
title: Crystal Structures Reveal an Induced-fit Binding of a Substrate-like Aza-peptide Epoxide to SARS Coronavirus Main Peptidase
date: 2007-02-23
journal: J Mol Biol
DOI: 10.1016/j.jmb.2006.11.078
sha: 0f351d5be0c24d0b2e5dc0b8d941ce63bcbda449
doc_id: 1011914
cord_uid: vgnxhksv

The SARS coronavirus main peptidase (SARS-CoV M(pro)) plays an essential role in the life-cycle of the virus and is a primary target for the development of anti-SARS agents. Here, we report the crystal structure of M(pro) at a resolution of 1.82 Å, in space group P2(1) at pH 6.0. In contrast to the previously reported structure of M(pro) in the same space group at the same pH, the active sites and the S1 specificity pockets of both protomers in the structure of M(pro) reported here are in the catalytically competent conformation, suggesting their conformational flexibility. We report two crystal structures of M(pro) having an additional Ala at the N terminus of each protomer (M(+A(-1))(pro)), both at a resolution of 2.00 Å, in space group P4(3)2(1)2: one unbound and one bound by a substrate-like aza-peptide epoxide (APE). In the unbound form, the active sites and the S1 specificity pockets of both protomers of M(+A(-1))(pro) are observed in a collapsed (catalytically incompetent) conformation; whereas they are in an open (catalytically competent) conformation in the APE-bound form. The observed conformational flexibility of the active sites and the S1 specificity pockets suggests that these parts of M(pro) exist in dynamic equilibrium. The structural data further suggest that the binding of APE to M(pro) follows an induced-fit model. The substrate likely also binds in an induced-fit manner in a process that may help drive the catalytic cycle.

Severe acute respiratory syndrome (SARS) is a highly transmissible, infectious and often fatal disease (World Health Organization-Severe acute respiratory syndrome †). Since its outbreak in 2002 and rapid spread throughout early 2003, efforts in the development of anti-SARS vaccines and drugs have taken on paramount importance. SARS is caused by a coronavirus (CoV); 1−3 it is an enveloped, positive-sense single-stranded RNA virus. Anti-SARS therapeutics could target any one of several major steps in the viral life-cycle, such as virus-cell interactions, virus entry, intracellular viral replication, virus assembly and exit. 4 Extensive studies have been carried out on the proteins involved in these steps. 5 The intracellular replication of CoV is mediated by a replicase complex derived from two virally coded polyprotein precursors, pp1a (486 kDa) and pp1ab (790 kDa). 6, 7 The formation of this replicase complex requires the extensive processing of the two polyproteins by two cysteine peptidases encoded within them; namely, the main peptidase (M pro ), also known as the 3C-like peptidase (3CL pro ) because of its similarity to the 3C peptidases of Picornaviridae, 8 and the accessory papain-like peptidase 2 (PL2 pro ), 7 which cleaves at three sites in the N-proximal regions of the two polyproteins. By contrast, M pro cleaves at 11 sites in the central and C-proximal regions of the two polyproteins, releasing key viral replication pro-teins, such as an RNA polymerase and a helicase. 7 As an essential protein, M pro is an attractive molecular target for the development of anti-SARS drugs.

SARS-CoV M pro is catalytically active only in the homodimeric form; each protomer has a molecular mass of 33.8 kDa. 9−14 Considerable efforts have been directed to the X-ray structural studies of M pro , resulting in the availability of many crystal structures of M pro and its variants over a pH range of 5.9-9.0 in a variety of space groups. All of these structures show that the two protomers of M pro are oriented almost perpendicular to each other. The Nterminal residues 1-7 of each protomer constitute the N-finger, of which Arg4 was shown to be mandatory for the dimerization and the exhibition of the catalytic activity of M pro . 14 Beyond the Nfinger, each protomer consists of three domains. Domain I (residues 8-101) and domain II (residues 102-184) comprise a two-β-barrel fold similar to that of the chymotrypsin-type serine peptidases. Domain III (residues 201-300) has five α-helices and is connected to domain II by a long loop (residues 185-200). Each protomer has its own substratebinding region situated in the cleft between domains I and II. Recent mutagenesis studies have confirmed that, similar to the main peptidases from human coronavirus strain 229E, 15, 16 and porcine transmissible gastroenteritis coronavirus, 17 SARS-CoV M pro is a cysteine peptidase with a Cys-His catalytic dyad at the active site. 16, 18 As suggested by the structure-based sequence alignment of the main peptidases (including their flanking residues in the polyproteins) from SARS-CoV and other coronaviruses, 16 and confirmed by in vitro studies, 7, 9 M pro cleaves preferentially at a consensus sequence for the P4 to P1′ residues of substrates (the nomenclature is based on that used by Schechter & Berger 19 with the arrow indicating the cleavage site): (amino acid with a small side-chain)-(any amino acid)-Leu-Gln↓(Ala, Ser, Gly).

A number of M pro inhibitors have been proposed using structure-based discovery 20, 21 and experimental screening. [22] [23] [24] The crystal structures of M pro inhibited by some peptidomimetic compounds have been determined. 25−27 Recently, virtual screening followed by experimental evaluations have identi-fied the old drug cinanserin as a strong inhibitor of the replication of SARS-CoV; cinanserin likely targets M pro . 28 We have reported on the kinetic and structural characterization of the inhibition of SARS-CoV M pro by an aza-peptide epoxide (APE, Figure 1 ). 29 APEs were synthesized as a new class of inhibitors apparently specific for clan CD cysteine peptidases, 30, 31 including the legumains, 32 and the caspases. 33 Each APE has an aza-peptide component, with an epoxide moiety attached to the carbonyl group of the P1 residue. The side-chain of the P1 residue predominantly determines the target-peptidase specificity of an APE. The substituent on the epoxide C2 atom also allows some tuning of both the inhibitory activity and specificity of APE towards a particular target peptidase. In the APE, the C α atom of the P1 residue is replaced by a nitrogen atom, yielding an aza-amino acid residue. This introduces trigonal planar geometry to the α-atom of the P1 residue and reduces the electrophilicity of the carbonyl C atom of the P1 residue, thereby making the carbonyl group of the P1 residue resistant to nucleophilic attack. 34 It has been proposed that APEs inhibit their target peptidases irreversibly by a mechanism in which the catalytic Cys S γ atom nucleophilically attacks one of the two epoxide carbon atoms (C2 or C3) of APE. 30, 32, 33 This results in the opening of the conformationally strained epoxide ring, with the formation of a covalent bond between the Cys S γ atom and the attacked APE atom. An APE containing Gln at P1 inhibits M pro with a k inact /K i = 1900(±400) M −1 s −1 . In this reaction, the catalytic Cys145 S γ atom attacks the epoxide C3 atom of the APE. 29 

The parameters and statistics derived from data processing and structure refinement are summarized in Table 1 . SARS-CoV M pro crystallizes in space group P2 1 in conditions slightly different from Figure 1 . Inhibition of SARS-CoV M pro by the aza-peptide epoxides (APEs) synthesized for our study, Cbz-Leu-Phe-AGln-EP-Coo-Et. The epoxide carbon atoms are numbered and their stereochemistry is omitted for simplicity. The proposed mechanism for the irreversible inhibition of clan CD cysteine peptidases by APEs is indicated by arrows. In the inhibition of M pro , route I was adopted. Cbz, the benzyloxycarbonyl group; AGln, aza-glutamine; EP, epoxide; COOEt, ethyl ester. those reported previously. 25 Each asymmetric unit contains both protomers (A and B) of the physiological dimer. In the electron density maps of the unbound M pro , residues 1A-44A, 50A-305A and 1B-302B could be identified. In the Ramachandran plot of this structure, Asp33A, Asn84A and Asn84B are in the generously allowed regions, and Asp33B, Glu47B, Tyr154A and Tyr154B are in the disallowed regions. In both independent protomers, the Asp33 N atom forms a hydrogen bond with the Thr98 O γ1 atom, and the Asp33 carbonyl O atom forms hydrogen bonds with the Trp31 N ε1 and the Asn95 N atoms. Also, the Asn84 N δ2 atom forms a hydrogen bond with the Glu178 carbonyl O atom. The poorly defined electron densities of the sidechains of Glu47B, Tyr154A and Tyr154B indicate dynamic disorder.

SARS-CoV M +A (-1) pro crystallized in space group P2 1 2 1 2 1 29 as well as P4 3 2 1 2. Each asymmetric unit of the latter contains only one protomer of the dimer; the two protomers of each dimer are related by the crystallographic 2-fold symmetry axis parallel with the C-face diagonal of the unit cell. In the electron density maps of the unbound M +A(-1) pro , residues 3-300 of the protomer were clearly identified. In the Ramachandran plot of this structure, Asp33 and Ser139 are in the generously allowed regions, and Asn84 and Tyr154 are in the disallowed regions. The side-chain of Tyr154 makes contact with those of Ile78 and probably Arg76 from a neighboring asymmetric unit. The electron density of Ser139 is not well defined. In the electron density maps of the M +A(-1) pro :APE complex, residues 2-300 of the protomer were identified. In the Ramachandran plot of this structure, Asp33 and Asn277 are in the generously allowed regions, and Asn84, Tyr154 and Ile286 are in the disallowed regions. The side-chains of Thr285 and Ile286 from opposite protomers of the dimer contact each other. The electron density of Asn277 is not well defined.

Superpositions of the crystal structures reported here with all of those previously reported show no significant difference in the protomer orientation and the overall fold.

The previously reported crystal structure of SARS-CoV M pro in space group P2 1 at pH 6.0 showed the collapse of the active site and S1 specificity pocket of one of the protomers, whereas the structures in the same space group at pH 7.6 and at pH 8.0 showed all the active sites to be in the catalytically competent conformation ( Table 2) . On the basis of this trend, a pH-triggered switch for the catalytic activity of M pro was proposed. 25 We have now grown crystals of M pro in the same space group (P2 1 with the same unit-cell constants) at pH 6.0 under slightly different conditions; the active sites and the S1 specificity pockets of both protomers are in the catalytically competent conformation (Figures 2(a) and (b), 3(a) and (b), 4(a) and (b)). More specifically, superposition of protomers A and B of the resulting M pro structure (rmsd 0.50 Å for 1070 out of 1172 main-chain atoms) shows good agreement in most atomic positions. In both independent protomers, the catalytic dyad has a distance of 3.6 Å between the His41 N ε2 atom and the Cys145 S γ atom, and the Cys145 S γ atom is coplanar with the atoms of the His41 imidazole ring. Residues Gly143 to Cys145 are in the proper conformation to form the oxyanion hole that 

where | F o | and | F c | are the observed and calculated structure factor amplitudes of a particular reflection, respectively. The summation is over 95% of the reflections in the specified resolution range. The remaining 5% of the reflections were selected randomly before the structure refinement and are not included in the structure refinement. R free was calculated over these reflections with the equation used for R work . 54 accommodates the carbonyl O atom of the scissile peptide bond of the substrate; the position to be occupied by the carbonyl O atom is occupied by a water molecule. The O atom of the water molecule forms hydrogen bonds at distances of 3.0 Å with the Gly143 N atom and with the Cys145 N atom (Figure 3 (a) and (b)). Previous studies suggest strongly that the predominant S1 specificity of SARS-CoV M pro for Gln is determined primarily by the conserved residue His163. 15, 16, 35 In both independent protomers, the orientation of the imidazole ring of His163 is determined by the hydrogen bonding between its N δ1 atom and the OH group of Tyr161 (3.0 Å) and by its π-stacking with the phenyl ring of Phe140 (the distance between the geometric centers of the aromatic rings is 3.9 Å). The position to be occupied by the side-chain carbonyl O atom of P1-Gln of the substrate is occupied by a water molecule, whose O atom forms a hydrogen bond with the His163 N ε2 atom (2.9 Å); this water molecule is coplanar with the atoms of the imidazole ring of His163. In both protomers, Phe140 and Glu166 interact with Ser1 of the opposite protomer to form the "floor" of the S1 specificity pocket (Figure 3 (a) and (b)). In protomer A, the Glu166A O ε1 atom forms a hydrogen bond (2.8 Å) with the His172A N ε2 atom, thereby constituting a "side-wall" of the S1 specificity pocket similar to that in protomer A of previously reported structures and the molecular-dynamic simulation model of M pro at pH 6.0 (Figure 3(a) ). 25, 36 However, in protomer B of the structure of M pro reported here, the side-chain of Glu166B rotates away and the His172B N ε2 atom forms a hydrogen bond (3.0 Å) with the Ser1A O γ atom of protomer A instead. The Ser1A O γ atom also forms a hydrogen bond (2.8 Å) with the Gly170B carbonyl O atom (Figure 3 (b)). The difference in the side-chain conformation of Glu166 in protomers A and B of the structure of M pro reported here indicates the weakness of the interaction between the side-chains of Glu166 and His172, even though this interaction may acquire some ionic character by the protonation of the His172 N δ1 atom. Contrary to those in protomer B of both the previously reported structure and the moleculardynamic simulation model of M pro at pH 6.0, 25,36 the side-chains of His163B and Glu166B do not interact in protomer B of the structure of M pro reported here, probably because His163B is protonated only at its N ε2 atom and carries no charge.

As in the previously reported structure of M pro at pH 6.0, 25 protomers A and B in the current structure show good agreement in the rest of the substratebinding region, including the S2, S4 and S1′ specificity pockets. Superpositions with the crystal structures of M pro in the other conditions show that the rest of the substrate-binding regions in the structure of M pro reported here are in the catalytically competent conformation.

Active sites and substrate-binding regions of the unbound SARS-CoV M +A(-1) pro We have reported the crystal structure of the SARS-CoV M +A (-1) pro :APE complex in space group P2 1 2 1 2 1 , whose asymmetric unit contains both protomers of the M +A(-1) pro dimer. 29 In that structure, the active sites and the substrate-binding regions of both protomers are in the catalytically competent conformation ( Table 2 ). In both independent protomers, the additional Ala at the N terminus blocks Ser1 and disrupts its interactions with Phe140 and Glu166 of the opposite protomer; however, the floor of the S1 specificity pocket is only partly disrupted. More importantly, the presence of a ten-residue affinity tag at the N terminus of both independent protomers reduces the specific activity of M pro by less than an order of magnitude. 29 Similar observations are given by the crystal structure of M pro with additional residues Ser-Leu at the N terminus of both independent protomers (PDB accession code 1Q2W; J. B. Bonanno et al., unpublished results).

Attempts to crystallize M +A(-1) pro in space group P2 1 have not been successful. Interestingly, M +A(-1) pro crystallizes in space group P4 3 2 1 2 as well as in space group P2 1 2 1 2 1 under the same conditions. In space group P4 3 2 1 2, each asymmetric unit has one protomer of M +A (-1) pro . Superpositions of the resulting structure of the unbound M +A(-1) pro with the crystal structures of M pro in the other conditions show no difference in the overall fold or the protomer orientation. Crystal contacts of the unbound M +A(-1) pro in space group P4 3 2 1 2 do not involve any residues forming the active sites and the S1 specificity pockets of the peptidase. The structure of the unbound M +A(-1) pro in space group P4 3 2 1 2 shows good agreement in most atomic positions with the structure of the APE-bound M +A(-1) pro in space group P2 1 2 1 2 1 . In the structure of the unbound M +A(-1) pro reported here, the catalytic dyad has a distance of 3.9 Å between the His41 N ε2 atom and the Cys145 S γ atom, and the Cys145 S γ atom is coplanar with the atoms of the His41 imidazole ring. However, the oxyanion hole and the S1 specificity pocket are distorted (Figures 2(c) and 3(c) ). The ϕ and ψ angles of residues Lys137 to Ser144 show dramatic differences compared with those of the previously reported structure of the APE-bound M +A(-1) pro 29

( Figure 4 (c)). The above-average B-factors of these residues in the unbound M +A(-1) pro indicate their high mobility relative to the rest of the peptidase ( Figure  5 (c)). The oxyanion hole is not distorted as much; the N atoms of Gly143 and Cys145 are still oriented to donate hydrogen bonds that would stabilize the negatively charged carbonyl O atom of the scissile peptide bond of the substrate, although no water molecule is found at the position to be occupied by the carbonyl O atom (Figure 3(c) ). The hydrogen bond between the N δ1 atom of His163 and the OH group of Tyr161 is preserved (3.1 Å); however, the imidazole ring of His163 is no longer π-stacked with the phenyl ring of Phe140; it makes contacts with the side-chain of Leu141 instead (Figure 3(c) ). Here the phenyl ring of Phe140 makes contacts with the side-chains of Val114, Tyr126, Ile136 and His172, and the mainchain atoms of Lys137 and Gly138 from the parent protomer, and with the side-chain of Arg4 from the opposite protomer. The Phe140 carbonyl O atom also forms a long hydrogen bond with the OH group of Tyr118 (3.4 Å). The electron density maps show two possible conformers of the side-chain of Glu166; the occupancies of both conformers were fixed at 0.5 Figure 2 (legend on next page) without refinement. In conformer 1, the Glu166 O ε1 atom forms a hydrogen bond with the His172 N δ1 atom (2.3 Å), and the Glu166 O ε2 atom forms a hydrogen bond with the His163 N ε2 atom (2.8 Å); whereas in conformer 2, the side-chain of Glu166 protrudes into the solvent (Figure 3(c) ). Probably because of their high mobility, the additional Ala at the N terminus, Ser1 and Gly2 of the protomer could not be identified in the electron density maps of the unbound M +A(-1) pro . Interactions similar to those forming the floors of the S1 specificity pockets of M pro are not observed in the unbound M +A (-1) pro . The rest of the substrate-binding region in the structure of the unbound M +A(-1) pro reported here is in the catalytically competent conformation, in good agreement with those in the previously reported structures of the APE-bound M +A(-1) pro and of M pro , except for some side-chains whose conformational rearrangements are necessary in order to accommodate APE or the other inhibitors.

Binding of APE to SARS-CoV M +A(-1) pro Both the P2 1 crystals of SARS-CoV M pro and the P4 3 2 1 2 crystals of SARS-CoV M +A(-1) pro were soaked in solutions of the APE synthesized for this study in three stereochemical versions: the two diastereomers Cbz-Leu-Phe-AGln-(2S,3S)EP-COOEt and Cbz-Leu-Phe-AGln-(2R,3R)EP-COOEt, and the racemic mixture of these two diastereomers. Outstanding electron density for APE was observed only in the electron density maps of M +A(-1) pro ; it could be fit only by the 2S,3S diastereomer ( Figure 6(a) ). The latter is consistent with the results of the studies on the inhibition of M pro by the APE in these versions. These results can be explained with the models of all four of the possible diastereomers of the APE binding to M pro . 29 to the speculation that the APE could act as a reversible inhibitor of M pro .

The conformation of the opened epoxide moiety and the main-chain conformation of P1-AGln of the APE in the structure of the M +A (-1) pro :APE complex reported here are essentially the same as those in the previously reported structures of the M pro :APE and the M +A(-1) pro :APE complexes (Figure 7(a)-(d) ). Unlike the P1 C α atom of other inhibitors, the P1-AGln N α atom of the APE is sp 2 -hybridized and has a trigonal planar geometry; in order to be accommodated by the S1 specificity pocket of the peptidases, the sidechain of P1-AGln has to adopt a different conformation, in particular the equivalent to χ (N-N α -C β -C γ ). This angle is −123.6°in the structure of the M +A(-1) pro : APE complex reported here; by contrast, in the crystal structures of the complexes of an M pro variant with a series of peptidomimetic inhibitors, this angle is in the range of −65°to −80°. 27 Interestingly, the binding of the APE induces the recovery of the catalytically competent conformation of the oxyanion holes and the S1 specificity (Figure 4(d) ). In contrast to those in the structure of the unbound M +A(-1) pro , the B-factors of these residues in the structures of the APE-bound M +A(-1) pro are close to the averages, indicating that the mobility of these residues is reduced upon the binding of the APE (Figure 5(d) ).

The hydrogen bond between the N δ1 atom of His163 and the OH group of Tyr161 is preserved (3.3 Å), and the π-stacking of the imidazole ring of His163 with the phenyl ring of Phe140 is recovered (distances between the geometric centers of the aromatic rings: 3.7 Å) (Figure 3(d) ). The His163 N ε2 atom no longer interacts with the Glu166 O ε2 atom, but forms a hydrogen bond (2.7 Å) with the sidechain carbonyl O atom of P1-AGln of the APE instead (Figure 6(b) and (c) ). The side-chains of Glu166 and His172 interact with each other (3.0 Å; Figure 3 (d)). The additional Ala at the N terminus and Ser1 of the protomer could not be identified in the electron density maps of the M +A(-1) pro :APE complex. Interestingly, the side-chain amide N atom of P1-AGln of the APE forms hydrogen bonds, though not in ideal geometry, with the Phe140 carbonyl O atom (3.4 Å) and the Glu166 O ε1 and O ε2 atom (3.5 Å and 2.9 Å, respectively) ( Figure 6(b) ). Therefore, without the participation of Ser1 of the opposite protomer, Phe140 and Glu166 are tied together to form parts of the floors of the S1 specificity pockets of M +A (-1) pro . Similar to the APE in protomer A of the previously reported structure of the APE-bound M +A(-1) pro , the benzyloxycarbonyl (Cbz) group of the APE in the structure of the APE-bound M +A(-1) pro reported here squeezes into and slightly widens the S4 specificity pockets of M +A (-1) pro and, as a result, making contacts with Leu167, Pro168, Gln192 and Ala193 of M +A (-1) pro . Otherwise, the interactions of the APE with the rest of the substrate-binding region observed in the Dynamic equilibrium for the conformation of the active sites and the S1 specificity pockets of SARS-CoV M pro

In contrast to the previously reported crystal structure of SARS-CoV M pro in space group P2 1 at pH 6.0, 25 the crystal structure of M pro reported here (in the same space group and at the same pH) shows that the active sites and the S1 specificity pockets of both protomers are in the catalytically competent conformation ( Table 2 ). In the determinations of both structures, the complete wild-type sequence (residues 1-306) of M pro was over-expressed, purified and crystallized. Although different strategies might have been used in the X-ray diffraction data collection and processing, and in the structure solution and refinement, this could not result in the dramatic structural differences observed. The structural differences probably arise from the differences in the conditions of the preparation and the crystallization of M pro . Similarly, the crystal structure of M pro in space group P2 1 2 1 2 at pH 6.5 shows that both protomers are in the catalytically competent conformation, 37 whereas that in the same space group P2 1 2 1 2 and at a slightly higher pH (6.6) shows that both protomers of the latter are in the catalytically incompetent conformation. 36 The conformations of the collapsed active sites and S1 specificity pockets observed in the catalytically incompetent protomers show some variability. This variability indicates that the active site and the S1 specificity pocket of each protomer of M pro do not adopt a single conformation in solution at pH 5.9-6.6, but instead there is an ensemble of conformations. A particular conformation might be favored by a particular set of crystallization conditions. However, the various conformations of the collapsed active sites and S1 specificity pockets share a common feature: the imidazole ring of His163 is not π-stacked with the phenyl ring of Phe140. π-stacking is an example of aromatic interactions. 38 In the structures of the unbound M pro , an aromatic interaction is observed between these two rings in an offset-stacked (i.e. π-stacking) or an edge-to-face fashion wherever the protomer is in the catalytically competent conformation (Figure 8 ). In both orientations, one or two hydrogen atoms (with partial positive charge) on the phenyl ring of Phe140 are positioned near the central region (with partial negative charge) of the imidazole ring of His163. According to the results of previous studies, 39,40 a single aromatic interaction as such is weak (the interaction energy may be in the range of 1-2 kcal/ mol only), in contrast to the clusters of aromatic interactions commonly involved in the stabilization of protein structures. 41 Therefore, this interaction is susceptible to disruptions that could be caused by changes in a number of factors; the formation of this interaction can be viewed as a reversible process in dynamic equilibrium:

The position of the equilibrium could be determined, in part, by pH. At low pH (near or below the pK a1 of His163; Figure 9 ), the His163 N δ1 atom in a significant number of the M pro protomers is protonated, thereby introducing a positive charge on the imidazole ring of His163. This would disfavor its aromatic interaction with the phenyl ring of Phe140, and the position of the equilibrium would shift to the right. This is consistent with the trend exhibited by the structures of the unbound M pro at various pH values ( Table 2 ). In the pH range of 7.0-9.0, most of the M pro protomers have the aromatic interaction and are in the catalytically competent conformation (the left-hand side of the equilibrium predominates); whereas in the pH range of 5.9-6.6, some of the M pro protomers lose the aromatic interaction and are in the catalytically incompetent conformation (the position of the equilibrium shifts to the right).

A second factor in determining the position of the equilibrium could be the integrity of the interactions among Phe140 and Glu166 of the parent protomer, and Ser1 of the opposite protomer. In the structure of M +A(-1) pro reported here, the additional Ala at the N terminus of the opposite protomer blocks Ser1 and disrupts its normal interactions with Phe140 and Glu166 of the parent protomer. This probably weakens the conformational anchor of Phe140, as indicated by its above-average B-factors ( Figure  5(c) ), making the aromatic interaction of its phenyl ring with the imidazole ring of His163 vulnerable to disruption. In all of the crystal structures of the unbound M pro (both the wild-type and the variants) determined so far, wherever Phe140 and Glu166 of the parent protomer, and Ser1 of the opposite protomer do not interact normally, the B-factors of residues Lys137 to Ser144 of the parent protomer are above the average, even if the parent protomer is in the catalytically competent conformation (e.g. the M pro structure of M pro at pH 6.5 in the space group P2 1 2 1 2); 37 in contrast, wherever the three residues interact normally, the B-factors of residues Lys137 to Ser144 of the parent protomer are close to the average (e.g. protomer A of the M pro structure reported here; Figure 5 (a)). These observations suggest that the interactions among the three residues can immobilize residues 137 to 144. Apparently, both the ionic interaction between the N terminus of Ser1 and the side-chain of Glu166 of opposite protomers, and the amide hydrogencarbonyl oxygen hydrogen bonds between Ser1 and Phe140 of opposite protomers contribute to the immobilization of residues Lys137 to Ser144. The 

Induced-fit Binding of APE to SARS-CoV M pro M pro structure at pH 7.0 in space group P2 1 2 1 2 42 and protomer B of the M pro structure reported here ( Figure 5(b) ) suggest that the absence of either interaction slightly compromises the immobilizing effect.

The increased vulnerability of the aromatic interaction to disruption would shift the position of the equilibrium to the right. This probably explains why the active sites and the S1 specificity pockets of both protomers are collapsed in the crystal structure of M +A(-1) pro reported here. M pro with its three N-terminal residues truncated, M pro Δ (1-3) , exists mainly as a dimer in solution but its activity is 24% lower than that of M pro . 14 Without the interactions among Phe140 and Glu166 of the parent protomer, and Ser1 of the opposite protomer, the position of the equilibrium would shift to the right, resulting in the decreased availability of the active sites and the S1 specificity pockets in the catalytically competent conformation. The binding of a substrate to M pro Δ(1-3) may induce the recovery of the catalytically competent conformation of the active sites and the S1 specificity pockets (to be discussed below), at an energetic cost accounted for by the reduced activity of M pro Δ (1-3) .

The crystal structures of the unbound and the APE-bound SARS-CoV M +A(-1) pro reported here show that the binding of the APE to M +A(-1) pro follows an induced-fit model, not the lock-and-key model. More particularly, the binding of the APE induces the recovery of the catalytically competent conformation of the oxyanion holes and the S1 specificity pockets of both protomers of M +A (-1) pro . This induced fit is likely relevant to the binding of APE to M pro , because of the conformational flexibility of the active sites and the S1 specificity pockets of M pro discussed above. In all of the crystal structures of the inhibitor-bound M pro (both the wild-type and the variants) determined so far, [25] [26] [27] 29, 42 except for protomer B in the complex of M pro with a chloromethyl ketone (CMK), 25 the active sites and the S1 specificity pockets are in the catalytically competent conformation ( Table 2 ). The induced-fit binding of the APE to M +A(-1) pro probably is driven by the formation of the covalent bond between the Cys145 S γ atom of M +A(-1) pro and the epoxide C3 atom of the APE. It would be energetically costly to break this C-S bond in order to repel the APE and preserve the conformational flexibility of the active sites and the S1 specificity pockets of M +A (-1) pro . Therefore, the active sites and the S1 specificity pockets of M +A(-1) pro have to adopt the catalytically competent conformation in order to accommodate a substrate mimic like APE. This explanation probably applies to the actual catalytic cycle of M pro as well: the acylation step (i.e. formation of As the side-chain of P1-AGln of the APE occupies the S1 specificity pockets of M +A(-1) pro , the side-chain amide N atom of P1-AGln of the APE donates hydrogen bonds to the Phe140 carbonyl O atom and to the Glu166 O ε1 and O ε2 atoms of M +A (-1) pro , in addition to the hydrogen bond between the sidechain carbonyl O atom of P1-AGln of the APE and the His163 N ε2 atom of M +A (-1) pro . Similarly, additional hydrogen bonds are observed in the crystal structures of the M +A (-1) pro :APE complex in space group P2 1 2 1 2 1 , the M pro :APE complex, 29 protomer A of the M pro :CMK complex, 25 and the complex of M pro with a peptidomimetic inhibitor. 26 In the crystal structures of the complexes of an M pro variant with a series of peptidomimetic inhibitors, these hydrogen bonds seem to be water-mediated. 27 Consequently, even without the participation of Ser1 of the opposite protomer of M +A(-1) pro , Phe140 and Glu166 are tied together by these hydrogen bonds to form parts of the floors of the S1 specificity pockets. These hydrogen bonds also conformationally anchor Phe140, as indicated by its close-to-average B-factors ( Figure 5(d) ), and make the π-stacking of its phenyl ring with the imidazole ring of His163 resistant to disruption. As a result, the catalytically competent conformation of the active sites and the S1 specificity pockets of M +A(-1) pro is rigidified; the position of the equilibrium described by equation (1) would shift to the left. This probably serves as an additional explanation for the predominant S1 specificity of M pro or M +A(-1) pro for Gln, but not for Glu. Comparison of the structure of the unbound M pro with that of the APE-bound M pro , both in space group C2 at pH 6.5, 29 does not show an obvious reduction in the Bfactors of residues Lys137 to Ser144 upon the binding of the APE to M pro , probably because Phe140 and Glu166 of the parent protomer, and Ser1 of the opposite protomer in the unbound M pro interact normally and immobilize these residues.

In protomer B of the M pro :CMK complex, the binding of the CMK does not induce the recovery of the catalytically competent conformation of the active site and the S1 specificity pocket of M pro ( Table 2 ). The side-chain of P1-Gln of the CMK cannot be accommodated by the collapsed S1 specificity pocket, so it protrudes into the solvent. 25 This could be attributed to the covalent bond between the Cys145 S γ atom of M pro and the methylene C atom of the CMK, whose rotation can orient the side-chain of P1-Gln of the CMK towards the solvent, thereby avoiding the steric hindrance due to the collapsed S1 specificity pocket. In protomer A of the M pro :CMK complex, although the active site and the S1 specificity pocket of M pro are in the catalytically competent conformation (Table 2) , allowing the side-chain of P1-Gln of the CMK to occupy the S1 specificity pocket of M pro , the unoccupied space in the rest of the substrate-binding region of M pro is large enough for the conformational rearrangements in the rest of the CMK, resulting in an unexpected binding mode of the CMK to protomer A of M pro as well. 25 These unexpected binding modes are not observed in the crystal structures of the complexes of M pro (both the wild-type and the variants) with the APE, 29 nor with the peptidomimetic inhibitors, 26, 27 probably because their P' ester groups sterically hinder the rotation of the C-S covalent bond formed between the peptidases and the inhibitors. In the case of the APE, the π-conjugation of the N α atom and the carbonyl group of P1-AGln restricts the orientation of the side-chain of P1-AGln to point towards the S1 specificity pockets of M pro or M +A (-1) pro . In the case of the peptidomimetic inhibitors, the C atom attacked by the Cys145 S γ atom corresponds to the carbonyl C atom of the scissile peptide bond of a substrate. Therefore, the resulting C-S bond pulls the inhibitor towards the back wall of the substrate-binding region. This reinforces the steric hindrance due to Figure 9 . Protonation/deprotonation states of His163 hydrogen bonded to Tyr161 in SARS-CoV M pro (wild-type and variants). K a1 is the dissociation constant for the protonation of the His163 N δ1 atom; K a2 is the dissociation constant for the deprotonation of the His163 N ε2 atom. the P' ester group of the inhibitor against the rotation of the C-S bond.

Thus, a refined explanation could be provided for the pH-dependence of the activity of M pro12,18,25,36 : At low pH, a significant number of the M pro protomers are in the catalytically incompetent conformation. However, these protomers are conformationally flexible; the recovery of their catalytically competent conformation can be induced by the binding of a substrate. The activity of M pro is still compromised, because of the energetic cost associated with this induced-fit binding. It would be interesting to determine this energetic cost quantitatively, and compare it with the experimentally determined decrease in the activity of M pro . Before carrying out such studies, the possibility of a second mechanism, such as the change in the protonation/ deprotonation state of the catalytic dyad with pH, being involved in the pH-dependence of the activity of M pro should not be ruled out.

The compositions of the dimer interfaces observed in the structures reported here are essentially identical with those observed in all of the previously reported structures. Extensive studies have been carried out on the role of domain III in the dimerization of M pro . With domain III truncated, the protomer of M pro does not dimerize over a wide range of concentrations, 10,14 whereas domain III alone forms a stable dimer even at very low concentrations. 10 Truncation of the final C-terminal helix (residues 293-306) strongly disfavors the dimerization of M pro ; 14,43 analysis of the dimer interfaces observed in the structures of M pro found that most of the residues of domain III involved in dimer interactions are parts of the final C-terminal helix. Superpositions of the previously reported structures of M pro in space group P2 1 at pH 6.0, 7.6 and 8.0, 25 show that domain III of both independent protomers rotates in a rigid-body manner as the pH increases from 6.0 to 8.0, whereas the remainder of the M pro dimer remains essentially in the same atomic positions (Figure 10 ). Domain III of protomer B rotates through a slightly greater angle than does domain III of protomer A. This rotation was regarded as a consequence of the recovery of the catalytically competent conformation of the active sites and the S1 specificity pockets of M pro as the pH increased from 6.0 to 8.0. However, the structure of M pro in space group P2 1 at pH 6.0 reported here, having the active sites and the S1 specificity pockets of both protomers in the catalytically competent conformation, shows good agreement in the atomic positions of domain III of both independent protomers with the previously reported structure of M pro in space group P2 1 at pH 6.0, but not with those at pH 7.6 or pH 8.0. The interactions between the domain-III residues of opposite protomers, and the interactions of the residues of domain III with those of domains I and II of the same protomer have been analyzed, but the mechanics of this rotation is still unclear. The structures of M pro in the other space groups have not been studied over such a wide range of pH. It would be interesting to explore whether this rotation is an artifact associated with crystal packing in space group P2 1 , or whether similar rotations occur in the structures of M pro in the other space groups as well, indicating some biochemical significance of this rotation. For example, the rotation may represent a mechanism underlying the pH-dependence of the dimerization behaviors of M pro . 12

Preparation of SARS-CoV M pro , M +A(-1) pro and APE SARS-CoV M +A(-1) pro was cloned, over-expressed and purified as described. 24 A clone expressing M pro was Figure 10 . C α traces showing the orientations of domain III of the previously reported structures of SARS-CoV M pro in space group P2 1 at pH 6.0 (red), 7.6 (yellow) and 8.0 (blue) 25 , and the structure of M pro in space group P2 1 at pH 6.0 reported here (green). Domain III of both independent protomers rotates as pH increases from 6.0 to 8.0. generated using oligonucleotide-directed evolution to delete the codon corresponding to the N-terminal Ala of M +A (-1) pro . Using this clone, M pro was over-expressed and purified essentially as described for M +A (-1) pro . Cbz-Leu-Phe-AGln-(S,S)EP-COOEt was synthesized using the methods established to synthesize other APEs 30,32,33 with minor modifications.

Crystallization, crystal soaking and cryo-protection Before crystallization, both SARS-CoV M pro and M +A(-1) pro were dialyzed against 20 mM Tris-HCl (pH 7.5), 20 mM NaCl and concentrated to 10 mg/ml. All crystals were grown at ambient temperature by the hanging-drop, vapor-diffusion method. For the P2 1 crystals, the reservoir solution contained 0.1 M Mes (pH 6.0), 2% (w/v) polyethylene glycol 20,000, 3% (v/v) dimethyl sulfoxide, and 1 mM dithiothreitol. The final crystallization drop contained equal amounts of the M pro solution and the reservoir solution. Crystals of a thick-plate habit grew in three to five days to a size of about 0.3 mm × 0.3 mm × 0.1 mm. For the P4 3 2 1 2 crystals, the growth conditions were the same as those for the P2 1 2 1 2 1 crystals; 29 the reservoir solution contained 0.1 M Mes (pH 6.5), 50 mM ammonium acetate, 6% (w/v) polyethylene glycol 8000, 3% (v/v) ethylene glycol, and 1 mM dithiothreitol. The drop contained equal amounts of the M +A(-1) pro solution and the reservoir solution. Bipyrimidal crystals grew in three to five days to a size of about 0.1 mm × 0.1 mm × 0.1 mm. Macroseeding was carried out to improve the quality of the crystals. Crystals of good quality were selected and soaked overnight in drops having the same compositions as their reservoir solutions plus the APE chosen for this study present at 3 mM. Cryoprotectants had essentially the same compositions as reservoir solutions, except for the inclusion of 25% (v/v) ethylene glycol and the exclusion of dimethyl sulfoxide and dithiothreitol. Crystals were soaked for about 10 s and then immediately shock-cooled in liquid nitrogen for storage and shipment to the synchrotron beamline.

Data collection and processing, structure solution, refinement and analysis The X-ray diffraction data from all crystals were collected at the synchrotron Beamline 8.3.1 (equipped with an ADSC-Q210 CCD detector) at the Advanced Light Source in the Lawrence Berkeley National Laboratory. All data sets were indexed, scaled and merged using DENZO and SCALEPACK. 44 Structure solution and refinement were carried out in CCP4. 45, 46 All structures were solved by the molecular replacement method using MOLREP, 47 with the structure of unbound SARS-CoV M pro in space group C2 (PDB accession code 2A5A) being used as the search model. 29 In the structure of the M +A(-1) pro :APE complex, APE was located as outstanding electron densities in the substrate-binding region of M +A(-1) pro in both the |F o |-|F c |,α c (contoured at 3σ and 4σ) and the 2|F o |-|F c |,α c (contoured at 1σ) maps. A restraint was applied to the covalent bond between the Cys145 S γ atom of M +A(-1) pro and the epoxide C3 atom of the APE, on the basis that the length of a C-S single bond is normally about 1.8 Å. All structures were then iteratively refined using REFMAC, 48 and manually adjusted when needed using XtalView/Xfit. 49 The stereochemical qualities of the final structures were assessed using PROCHECK. 50 Graphical representations of the structures were prepared using PyMOL ‡. Superimpositions of the various structures were carried out using ALIGN, 51,52 based on the main-chain atoms (amide N, C α , and carbonyl C and O). The surface areas of the structures were calculated using NACCESS § Dimer interactions and M +A(-1) pro :APE interactions were analyzed using DIMPLOT' 53 and LIGPLOT, 53 respectively.

The atomic coordinates and the structure factors of all the structures have been deposited in the RCSB Protein Data Bank. The accession code is 2GT7 for the structure of the unbound SARS-CoV M pro , 2GT8 for the structure of the unbound M +A(-1) pro , and 2GTB for the structure of the M +A(-1) pro :APE complex.

associated with severe acute respiratory syndrome

Identification of a novel coronavirus in patients with severe acute respiratory syndrome

Coronavirus as a possible cause of severe acute respiratory syndrome

Molecular mechanisms of severe acute respiratory syndrome (SARS)

Structural insights into SARS coronavirus proteins

Viral replicase gene products suffice for coronavirus discontinuous transcription

Mechanisms and enzymes involved in SARS coronavirus genome expression

Virus-encoded proteinases and proteolytic processing in the Nidovirales

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

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

Characterization of SARS main protease and inhibitor assay using a fluorogenic substrate

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

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 dynamic simulations

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

Biosynthesis, purification, and characterization of the human coronavirus 229E 3C-like proteinase

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

Structure of coronavirus main proteinase reveals combination of a chymotrypsin fold with an extra alpha-helical domain

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

On the size of the active site in proteases

Identification of novel inhibitors of the SARS coronavirus main protease 3CLpro

Synthesis and evaluation of keto-glutamine analogues as potent inhibitors of severe acute respiratory syndrome 3CLpro

Small molecules targeting severe acute respiratory syndrome human coronavirus

Identification of novel small-molecule inhibitors of severe acute respiratory syndrome-associated coronavirus by chemical genetics

High-throughput screening identifies inhibitors of the SARS coronavirus main proteinase

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

Design and synthesis of peptidomimetic severe acute respiratory syndrome chymotrypsin-like protease inhibitors

Design of wide-spectrum inhibitors targeting coronavirus main proteases

Cinanserin is an inhibitor of the 3C-like proteinase of severe acute respiratory syndrome coronavirus and strongly reduces virus replication in vitro

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

Aza-peptide epoxides: a new class of inhibitors selective for clan CD cysteine proteases

Evolutionary lines of cysteine peptidases

Aza-peptide epoxides: potent and selective inhibitors of Schistosoma mansoni and pig kidney legumains (asparaginyl endopeptidases)

Design, synthesis, and evaluation of aza-peptide epoxides as selective and potent inhibitors of caspases-1

Irreversible inhibitors of serine, cysteine, and threonine proteases

Mutational analysis of the active centre of coronavirus 3C-like proteases

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

Structure of the SARS coronavirus main proteinase as an active C2 crystallographic dimer

Aromatic interactions

Dimerization energetics of benzene and aromatic amino acid sidechains

Aromatic-aromatic interactions and protein stability. Investigation by double-mutant cycles

Aromaticaromatic interaction: a mechanism of protein structure stabilization

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

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

Processing of Xray diffraction data collected in oscillation mode

The CCP4 suite: programs for protein crystallography

A graphical user interface to the CCP4 program suite

MOLREP: an automated program for molecular replacement

Refinement of macromolecular structures by the maximum-likelihood method

XtalView/Xfit-a versatile program for manipulating atomic coordinates and electron density

PROCHECK: a program to check the stereochemical quality of protein structures

Phosphocholine binding immunoglobulin Fab McPC603. An X-ray diffraction study at 2.7 A

ALIGN: a program to superimpose protein coordinates, accounting for insertions and deletions

LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions

Free R value: a novel statistical quantity for assessing the accuracy of crystal structures

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/ j.jmb.2006.11.078