Characterization of Self-Processing Activities and Substrate Specificities of Porcine Torovirus 3C-Like Protease

Currently, the active-site residues and substrate specificities of 3C-like protease (3CL^pro) differ among nidoviruses, and the detailed catalytic mechanism remains largely unknown. Here, porcine torovirus (PToV) 3CL^pro cleaves 12 sites in the polyproteins, including its N- and C-terminal self-processing sites. Unlike coronaviruses and arteriviruses, PToV 3CL^pro employed His53 and Ser160 as the active-site residues that recognize a glutamine (Gln) at the P1 position. Surprisingly, mutations of P1-Gln impaired the C-terminal self-processing but did not affect N-terminal self-processing. The “noncanonical” substrate specificity for its N-terminal self-processing was attributed to the phenylalanine (Phe) residue at the P4 position in the N-terminal site. Furthermore, a double glycine (neutral) substitution at the putative P4-Phe-binding residues (P62G/L185G) abolished the cleavage activity of PToV 3CL^pro suggested the potential hydrophobic force between the PToV 3CL^pro and P4-Phe side chains.

ABSTRACT

The 3C-like protease (3CL^pro) of nidovirus plays an important role in viral replication and manipulation of host antiviral innate immunity, which makes it an ideal antiviral target. Here, we characterized that porcine torovirus (PToV; family Tobaniviridae, order Nidovirales) 3CL^pro autocatalytically releases itself from the viral precursor protein by self-cleavage. Site-directed mutagenesis suggested that PToV 3CL^pro, as a serine protease, employed His53 and Ser160 as the active-site residues. Interestingly, unlike most nidovirus 3CL^pro, the P1 residue plays a less essential role in N-terminal self-cleavage of PToV 3CL^pro. Substituting either P1 or P4 residue of substrate alone has little discernible effect on N-terminal cleavage. Notably, replacement of the two residues together completely blocks N-terminal cleavage, suggesting that N-terminal self-cleavage of PToV 3CL^pro is synergistically affected by both P1 and P4 residues. Using a cyclized luciferase-based biosensor, we systematically scanned the polyproteins for cleavage sites and identified (FXXQ↓A/S) as the main consensus sequences. Subsequent homology modeling and biochemical experiments suggested that the protease formed putative pockets S1 and S4 between the substrate. Indeed, mutants of both predicted S1 (D159A, H174A) and S4 (P62G/L185G) pockets completely lost the ability of cleavage activity of PToV 3CL^pro. In conclusion, the characterization of self-processing activities and substrate specificities of PToV 3CL^pro will offer helpful information for the mechanism of nidovirus 3C-like proteinase’s substrate specificities and the rational development of the antinidovirus drugs.

IMPORTANCE Currently, the active-site residues and substrate specificities of 3C-like protease (3CL^pro) differ among nidoviruses, and the detailed catalytic mechanism remains largely unknown. Here, porcine torovirus (PToV) 3CL^pro cleaves 12 sites in the polyproteins, including its N- and C-terminal self-processing sites. Unlike coronaviruses and arteriviruses, PToV 3CL^pro employed His53 and Ser160 as the active-site residues that recognize a glutamine (Gln) at the P1 position. Surprisingly, mutations of P1-Gln impaired the C-terminal self-processing but did not affect N-terminal self-processing. The “noncanonical” substrate specificity for its N-terminal self-processing was attributed to the phenylalanine (Phe) residue at the P4 position in the N-terminal site. Furthermore, a double glycine (neutral) substitution at the putative P4-Phe-binding residues (P62G/L185G) abolished the cleavage activity of PToV 3CL^pro suggested the potential hydrophobic force between the PToV 3CL^pro and P4-Phe side chains.

INTRODUCTION

Toroviruses (order Nidovirales, suborder Tornidovirineae, family Tobaniviridae, subfamily Torovirinae, genus Torovirus) are enveloped positive-stranded RNA viruses with genomes ranging from 23 to 33 kb that cause a variety of diseases in animals and humans (1–4). Toroviruses (ToVs) are potentially zoonotic, although the precise mechanisms of transmission are still unknown. The existence of human ToV or ToV-like particles and the emergence in humans of coronavirus disease 2019 (COVID-19), caused by infection with another member of the Coronaviridae family, highlight the need to further understand the biology of ToVs (3, 5–7).

Porcine torovirus was first detected and characterized in feces samples from piglets in the Netherlands by reverse transcription-PCR (RT-PCR) and immunoelectron microscopy in 1998 (8). Limited epidemiological studies suggest that PToV appears to be widespread in the global pig population and swine herds have a high seroprevalence, although its pathogenicity remains unclear (9–14). Although PToV is clinically believed to cause some subclinical symptoms and has not yet caused significant economic damage, recent studies have shown that the PToV genome is highly variable and susceptible to recombination with viruses of other genera, poses a risk of cross-species transmission, making the virus of increasing concern (15–20). The inability to isolate PToV in cell culture severely limits our current understanding of the biological properties and pathogenicity of the disease. Before the successful isolation of the virus achieved, the pathogenicity of PToV could be better understood with detailed research on the function of the PToV-encoded proteins.

As with other nidoviruses, two-thirds of the PToV genome contains two long open reading frames (ORFs), ORF1a and ORF1ab, which encode the replicative polyproteins (pp1a and pp1ab) (21). The polyproteins pp1a and pp1ab are cleaved posttranslationally by virus-encoded 3C-like protease (3CL^pro; also known as main protease) to produce most of the functional nonstructural proteins that play important roles in regulating viral replication and pathogenesis (22, 23). Given its crucial role in viral replication, 3CL^pro is an important target for the design and development of potent antiviral agents (24). However, the 3CL^pro sequences of nidoviruses vary widely among themselves and are only reasonably conserved in the region of the active site (25–28). For example, arterivirus 3CL^pros employ a canonical Ser-His-Asp triad, which is responsible for the recognition of cleavage sites with a glutamic acid (Glu, E) at the P1 position (22, 29–31). However, coronavirus 3CL^pros use a Cys-His catalytic dyad that recognizes a glutamine (Gln, Q) at the P1 position (32–34). In this study, we characterized the self-processing activities and substrate specificities of the PToV 3CL^pro. In contrast to coronaviruses and arterivirus, PToV 3CL^pro employed His53 and Ser160 as the active-site residues that recognizes a glutamine (Gln, Q) at the P1 position. Surprisingly, the phenylalanine (Phe, F) at the P4 position and the Gln at the P1 position had important synergistic effects on the N-terminal self-cleavage of PToV 3CL^pro, since mutation of either residue (Q-P1-A or F-P4-A) alone affects only partial cleavage activity. However, replacement of the two residues (Q-P1-A/F-P4-A) together completely blocked N-terminal cleavage. Subsequently, homology modeling and biochemical experiments showed that the potential interactions between the protease’s S1/S4 pockets and P1/P4 positions of the substrate are related to the loss of the N-terminal self-cleavage of PToV 3CL^pro.

RESULTS

Complete genome sequencing and genetic analysis of porcine torovirus.

To characterize the PToV 3CL^pro and the proteolytic processing of PToV polyproteins, we first collected PToV RT-PCR-positive fecal samples from pigs. Briefly, total RNA was extracted from pig feces, and the complete PToV genome was determined from a collection of cDNA clones derived from reverse-transcribed RNA. The whole-genome sequence comprises 28,276 nucleotides, excluding the 3′-terminal poly(A) tail, and has been deposited in the GenBank database under accession number MH603532 (named HB-1, Fig. 1A). Sequence similarity analysis showed that the complete genome of PToV HB-1 strain shared 88.24 to 92.07% nucleotide similarity with those of previously sequenced PToVs (GenBank accession no. JQ860350, KM403390, LC483442, and LT900503). Furthermore, we evaluated the phylogenetic position of PToV in relation to other nidoviruses. Sequence comparison and phylogenetic analysis revealed PToV and other toroviruses formed a separate lineage (suborder Tornidovirineae) that separated viruses of the suborders Coronavirieae and Arterivirieae (Fig. 1B), which led us to believe that toroviruses might represent a separate cluster of nidoviruses that linked arteriviruses and coronaviruses.

FIG 1.

Complete genome sequencing and genetic analysis of porcine torovirus. (A) Schematic diagram of the PToV (HB-1) genome structure, the genome contains six open reading frames (ORFs) flanked by a long 5′ untranslated region (UTR) (796 nucleotides [nt]) and a short 3′ UTR (194 nt). The predicted PToV 3CL^pro specific region is marked by the dark area; all the predicted cleavage sites listed above are shown in the green triangle. (B) 3CL^pro-based phylogeny of representatives of established nidovirus species. The evolutionary history was inferred by using the maximum-likelihood method and the Le_Gascuel_2008 model. The amino acid sequences of nidovirus 3CL^pro were aligned using MAFFT. Different viruses are indicated at the terminal nodes. Different colors represent different suborders: Ronidovirineae and Mesnidovirineae (orange), Cornidovirineae (blue), Tornidovirineae (red), Arnidovirineae (green), and Monidovirineae (purple).

Identification of a PToV ORF1a-encoded 3C-like serine protease.

Previous studies have shown that the central and C-terminal portions of nidovirus replicase polyproteins are processed extensively by the viral 3CL^pro (29, 35, 36). Within the replicase polyprotein, nidovirus 3C-like proteases (also known as main proteases, M^pro) are generally flanked by two transmembrane (TM)-spanning domains (TM-3CL^pro-TM) (37, 38). When transmembrane prediction was performed using TMHMM on the pp1a region (encoded by ORF1a) based on the multiple alignment of toroviruses sequences, some conserved transmembrane regions were observed (data no shown). For putative PToV 3CL^pro, two transmembrane domains (show as red) were predicted at its N- and C-terminal borders (Fig. 2A), which is consistent with previous studies (37). First, we used a “self-cleavage” approach to identified the PToV 3CL^pro domain. To avoid the possible cytotoxicity caused by the expression of hydrophobic sequences in E. coli, we expressed a fusion protein comprising PToV pp1a residues 3012 to 3392 (pp1a-3012-3392, which represent the putative 3CL^pro domain, together with the N- and C-terminal flanking sequences, except for the predicted transmembrane domains). Considering the expected small size of the potential C-terminal autocleavage product probably prevented its detection, a C-terminal glutathione S-transferase (GST) was fused to pp1a-3012-3392, resulting in the expression construct pp1a-3012-3392-GST (named WT-GST, Fig. 2B). The predicted molecular mass of this fusion protein is 69 kDa, of which 26 kDa is contributed by the GST protein. Based on the conservation of putative active-site residues in many viral 3C and 3C-like proteases [including GX(C/S)G motif], the active-site nucleophile Ser-3325 was predicted in the PToV 3CL^pro domain. To confirm that any detected cleavage products were due to the proteolytic activity of the 3CL^pro, we also generated a mutant derivative of the pp1a-3012-3392-GST construct in which the conserved serine of the active-site nucleophile (Ser-3325) was mutated to alanine (named S3225A-GST and served as a control). As a second control, the GST fusion protein encoded by the pET-28a plasmid (empty vector) was expressed. When the expression of the 3CL^pro WT-GST construct was induced by IPTG (isopropyl-β-d-thiogalactopyranoside), we observed three proteins of ∼10 kDa (predicted N-terminal cleavage product, N-term CP), 27 kDa (predicted C-terminal cleavage product, C-term CP), and 32 kDa (predicted mature 3CL^pro) in the Coomassie blue-stained SDS-polyacrylamide gel, respectively (Fig. 2C, lane 6). These cleavage products were not detected in pET-28a-GST, 3CL^pro mut-S3225A-GST, and noninduced cells (the controls). In contrast, the induction of S3225A-GST protein expression led to one additional overexpressed protein with the expected size of 69 kDa for the unprocessed pp1a-3012-3392-GST fusion protein (Fig. 2C, lane 7). Collectively, these results indicate that the PToV-3CL^pro WT-GST construct had the predicted proteolytic activity, while this proteolytic activity was abolished after the assumed active-site nucleophile, Ser3225, was replaced with alanine (Ala).

FIG 2.

Identification of a PToV ORF1a-encoded 3C-like serine protease. (A) As predicted by TMHMM, the predicted 3CL^pro region was flanked by two transmembrane domains (red). (B) Schematic representation of PToV pp1a/1ab 3012–3392 and fusion protein constructs used in this experiment. The putative 3CL^pro domain is indicated in yellow, and flanking regions are shown in white. The GST tag is shown in green. (C) Prokaryotic expression analysis of GST (lanes 2 and 5), WT-GST (lanes 3 and 6), or S160A-GST (lanes 4 and 7) fusion proteins. The bacteria were mock induced (lanes 2, 3, and 4) or induced with 0.8 mM IPTG for 5 h at 37°C (lanes 5, 6, and 7). Total cell lysates were separated by SDS-PAGE in a 12% polyacrylamide gel and then stained with Coomassie brilliant blue R-250, followed by destaining with 50% methanol for a short time (Fig. 2C, upper) or long time (Fig. 2C, lower). The positions of the fusion protein and cleavage products are indicated by arrowheads.

To further determine the N- and C-terminal PToV 3CL^pro self-processing sites, the predicted mature 3CL^pro (32 kDa) and C-terminal cleavage product (27 kDa) were cut from the SDS-PAGE gel, and N-terminal sequencing was performed by Edman degradation. The mature 3CL^pro protein contained the sequence Ser3096-Val-Phe-Ser-Lys at its N termini (Fig. 3A), indicating that PToV 3CL^pro N-terminal cleavage occurred between Gln3095 and Ser3096 in the sequence context SNFSFQ³⁰⁹⁵↓S³⁰⁹⁶VFSKV. In addition, the sequence Ser3383-Val-Ser-Thr-Asn at its N termini of PToV 3CL^pro C-terminal cleavage product indicated that the 3CL^pro C-terminal cleavage in the PToV pp1a occurred at QPFKKQ³³⁸²↓S³³⁸³VSTNV (Fig. 3B). Taken together, these data suggest that PToV pp1a/pp1ab residues 3096 to 3382 region contain an active serine protease that undergoes self-processing at both N- and C-terminal cleavage sites (SNFSFQ³⁰⁹⁵↓S³⁰⁹⁶VFSKV and QPFKKQ³³⁸²↓S³³⁸³VSTNV, respectively) with a conserved P1-Gln residue.

FIG 3.

Characterization of N-terminal and C-terminal PToV 3CL^pro self-processing sites determined by protein sequencing. (A and B) Results of N-terminal sequencing. N-terminal (A) and C-terminal (B) self-processing sites were determined by protein sequencing. Chromatograms of PTH amino acids from reaction cycles 1 to 5. Specific peaks are indicated by a single-letter code.

Identification of catalytic residues in the PToV 3CL^pro active site.

Previous studies suggested that the nidoviruses 3CL^pro are serine/cysteine proteases that mostly utilize a Ser-His-Glu/Asp catalytic triad or a Cys-His catalytic dyad (27, 30, 39–42). For the PToV 3CL^pro, a comparative sequence analysis of nidovirus 3CL^pro, together with the identified PToV 3CL^pro, was performed to further identify the active site residues of PToV 3CL^pro. As shown in Fig. 4A, residues H53, S160, and E91 (the numbers indicated here represent their positions in mature PToV 3CL^pro) were conserved with the identify active-site Ser, His, and Asp/Glu of other nidovirus 3C-like proteases, suggesting that these three residues might be catalytic residues in the PToV 3CL^pro active site. To verify these predictions, the PToV 3CL^pro domain, together with a flanking pp1a sequence (pp1a-3012-3392), was fused with both N- and C-terminal GST tags using a mammalian expression vector pCAGGS as a backbone (Fig. 4B). Then, the 3CL^pro domain was expressed in HEK-293T cells, and proteolytic processing was analyzed by Western blotting of the total lysates with GST antibody. As expected, two faster-migrating protein bands were detected in WT 3CL^pro (GST-WT-GST)-transfected cells at 34 kDa (predicted N-terminal self-cleavage product) and 27 kDa (predicted C-terminal self-cleavage product) (Fig. 4C, lane 2). However, the substitution of Ser160 and His53 with Ala resulted in an unprocessed protein (corresponding to a molecular mass of 92 kDa) that could be detected by Western blotting with a GST-specific antibody (Fig. 4C, lanes 3 and 5), which is consistent with the results obtained from another fusion protein construct expressed in an Escherichia coli expression system (Fig. 2C). In contrast, the mutation of Glu91 (E91A) did not seem to influence the autoprocessing release of the 3CL^pro domain from the GST-E91A-GST fusion protein (Fig. 4C, lane 4), which contradicted its assumed role as a third catalytic residue. Interestingly, some additional smaller processing products of <92 kDa (unprocessed protein) were also observed in GST-H53A-GST- and GST-S160A-GST-transfected cells using a GST-specific antibody by Western blotting, as previously shown for alphamesonivirus 3CL^pro (25). These results may due to nonspecific proteolytic cleavage by host cellular proteases and deserve further investigation.

FIG 4.

Identification of catalytic residues in the PToV 3CL^pro active site. (A) Multiple sequence alignment of arterivirus and torovirus 3CL^pro domains. Clustal Omega was used to multiple sequence alignment of torovirus and arterivirus 3CL^pros (64). Abbreviations of virus names and GenBank accession numbers for the sequences are as follows: EAV, equine arteritis virus (NC_002532); LDV; lactate dehydrogenase-elevating virus neurovirulent type C (NC_02534); PRRSV, porcine reproductive and respiratory syndrome virus strain Lelystad (AY588319); EToV, equine torovirus strain Berne (X52374); PToV, porcine torovirus strain HB-1 (MH_603532); and BRV, Breda virus (NC_007447). The secondary structure element of the EAV 3CL^pro is shown above the sequence (PDB 1MBM). This was extracted from the PDB file and added to the alignment using ESPript 3.0 (65). Residues shared by porcine torovirus 3CL^pro and any of the other sequences are indicated by dark arrowheads. (B) Schematic representation of the eukaryotic expression construct used in this experiment. The putative 3CL^pro domain is indicated in yellow, and flanking regions are shown in white. The two GST tags between PToV pp1a/1ab 3012–3392 are shown in blue. 3CL^pro mut, putative 3CL^pro domain containing a Ser-160-to-Ala change in the expressed sequence. (C) HEK-293T cells were transfected with wild-type p-GST-WT-GST or p-GST-WT-GST mutants, including p-GST-H53A-GST, p-GST-S160A-GST, and p-GST-E91A-GST. Cells were then lysed after 28 h transfection and evaluated by Western blotting. (D) HEK-293T cells were transfected with p-GST-S160A-GST, along with HA-tagged PToV 3CL^pro expression plasmid or PToV 3CL^pro mutants, including pH53A, pS160A, and pE91A. The cells were then lysed after 28 h of transfection and evaluated by Western blotting.

We further performed a trans-cleavage assay using a GST-S160A-GST as a substrate to verify catalytic residues (H53 and S160) in the active site of PToV 3CL^pro. To this end, the wild-type and mutant putative catalytic residues in mature 3CL^pro (pp1a-3096-3382) were fused with pCAGGS-HA (N-terminal hemagglutinin [HA] tag) and transfected together with the GST-S160A-GST construct in HEK-293T cells. In contrast to 3CL^pro mutants (H53A and S160A), the overexpression of WT and E91A mutant of 3CL^pro significantly induced GST-S160A-GST cleavage (Fig. 4D). Together, the data showed that the PToV 3CL^pro uses Ser160 and His53 as active-site residues in both cis and trans cleavage.

Characterization of self-cleavage substrate specificities of PToV 3CL^pro.

To better understand the specific substrate properties for efficient cleavage by the PToV 3CL^pro, mutagenesis of the N- and C-terminal self-processing sites was performed. Since most nidovirus 3CL^pros have a marked preference for Gln/Glu at the P1 position of the substrate (21, 22, 26, 43–45), we sought to abolish PToV 3CL^pro self-cleavage by replacement of the P1-Gln residue by Ala (Q3095A and Q3382A). As a control, the two other glutamines around the N- and C-terminal PToV 3CL^pro self-processing sites were also replaced with alanine (Q3088A and Q3377A) (Fig. 5A). As expected, the expression of the fusion proteins GST-WT-GST, GST-Q3088A-GST, and GST-Q3377A-GST led to two faster-migrating protein bands that migrated with the marker proteins GST-N and C-GST (Fig. 5B and C). Moreover, the fusion proteins GST-Q3382A-GST resulted in a putative C-terminal unprocessed protein (Fig. 5B, lane 6), suggesting that Gln3382 at the C-terminal P1 position caused the loss of its C-terminal self-processing site. Surprisingly, although a putative N-terminal unprocessed protein (corresponding to a molecular mass of 66 kDa) was detected from the GST-Q3095A-GST protein, N-terminal cleavage also occurred with nearly the same efficiency as the wild-type construct (Fig. 5B, lane 4), suggesting that the N-terminal P1-Gln residue might be not a sufficient determinant for N-terminal cleavage by PToV 3CL^pro.

FIG 5.

Characterization of self-cleavage substrate specificities of PToV 3CL^pro. (A) Schematic representation of the eukaryotic expression constructs used in this experiment. Positions of individual cleavage sites are indicated, and calculated molecular masses of processing products are given. Cleavage products are represented by symbols on the right of each lane, as shown in the schematic representation. (B) HEK-293T cells were transfected with wild-type p-GST-WT-GST or p-GST-WT-GST mutants, including p-GST-Q3088A-GST, p-GST-Q3095A-GST, p-GST-Q3377A-GST, and p-GST-Q3382A-GST. Cells were then lysed at 28 h posttransfection and evaluated by Western blotting. (C) As shown in panel A, two p-GST-WT-GST truncates were constructed, including GST-N and C-GST. The truncates were then transfected with HEK-293T cells and lysed for Western blotting at 28 h posttransfection.

To further investigate the possible influences of single or multiple residues at the cleavage sites, we expressed a series of mutant derivatives at the N-terminal self-cleavage of PToV 3CL^pro containing appropriate amino acid substitutions (Fig. 6A and B). As shown in Fig. 6C, the N-terminal self-cleavage products was no longer detectable after deletion of P6-P6′ or P6-P1 at the N-terminal self-cleavage region (Fig. 6C, lanes 3 and 4). In contrast, the deletion of P1′-P6′ did not prevent the release of N-terminal self-cleavage product (Fig. 6C, lane 5), speculating that one or more residues at P6-P1 positions might be sufficient determinants for the N-terminal self-cleavage of PToV 3CL^pro. To test this hypothesis, consecutive Ala substitutions were introduced at the P6-P1 positions, named (P2-P1)-A, (P3-P1)-A, (P4-P1)-A, (P5-P1)-A, and (P6-P1)-A (Fig. 6D). In contrast to consecutive Ala substitutions at the P2-P1 and P3-P1 positions, the (P4-P1)-A mutant prevented the release of N-terminal cleavage products (Fig. 6D, lane 6), suggesting the Phe at the P4 position might be important for the effective N-terminal cleavage. Similar results were obtained from constructs containing additional substitutions at the P5 and P6 positions (Fig. 6D, lanes 7 and 8). To further test whether Phe at the P4 position is a sufficient determinant for N-terminal self-cleavage, a single Ala substitution at the P4 position (F-P4-A) and a double Ala substitution at the P1 and P4 positions (F-P4-A/Q-P1-A) were constructed. A single point mutation at the P4 or P1 position only partially affect, whereas double point mutations at the P4 and P1 position abolish the ability of N-terminal self-cleavage (Fig. 6E). Similar results were obtained in the trans-cleavage assay, which suggested that Phe at the P4 and Gln at the P1 position synergistically control cleavage at the N-terminal 3CL^pro self-processing site (Fig. 6F).

FIG 6.

P4-F plays a key role in the N-terminal self-cleavage of PToV 3CL^pro. (A) Schematic representation of the eukaryotic expression constructs used in this experiment. Positions and sequences of individual cleavage sites are indicated in left, and the calculated molecular masses of processing products are given. Cleavage products are represented by symbols to the right of each lane. (B) Schematic representation the sequences of deletions or mutations in the N-terminal cleavage region of PToV 3CL^pro. (C to E) The p-GST-WT-GST was selected as the template, and various p-GST-WT-GST mutants were constructed as shown in panel B. HEK-293T cells were transfected with p-GST-WT-GST or various p-GST-WT-GST mutants. The cells were lysed at 28 h psottransfection and then analyzed by Western blotting. (F) The p-GST-S160A-GST was selected as the template, and the P1 or P4 position was single site mutated or double site mutated to A. HEK-293T cells were transfected with HA-tagged pCAGGS-PToV 3CL^pro, along with p-GST-S160A-GST or p-GST-S160A-GST mutants, including S160A+F-P4-A, S160A+Q-P1-A, and S160A+(F-P4-A/Q-P1-A). The cells were then lysed at 28 h psottransfection and evaluated by Western blotting.

Identification of cleavage sites recognized by PToV 3CL^pro within polyproteins.

To further investigate whether proteolytic processing can occur at other PToV polyprotein cleavage sites, 12 putative cleavage sites of PToV 3CL^pro were predicted based on alignments of pp1ab polyproteins among PToV isolates and other related viruses in the Torovirus genus. Subsequently, we performed a trans-cleavage assay using the cyclized luciferase reporter gene vector, which has been used to identify cleavage sites recognized by viral protease in previous studies (46). To this end, the P6-P6′ positions of PToV 3CL^pro potential recognition regions were introduced into the cyclized luciferase reporter. The identified PToV 3CL^pro N-terminal autocleavage sequence (named CP4/nsp2) and the C-terminal autocleavage sequence (named nsp2/nsp3) served as positive controls, whereas the tobacco etch virus protease recognition sequence ENLYFQYS (named 233D) and the putative PToV 3CL^pro internal recognition sequence (named nsp2 internal, the same location which can be cleaved by equine torovirus [EToV] 3CL^pro) were used as negative controls (27). As shown in Fig. 7A, the luciferase activity of positive controls (CP4/nsp2 and nsp2/nsp3) were markedly induced in PToV 3CL^pro-cotransfected cells, whereas no activity of negative controls was detected. Also, 10 additional putative cleavage sites were identified to be cleaved by PToV 3CL^pro (Fig. 7B), as determined by the cyclized luciferase reporters. Consistent with previous study, the 12 cleavage sites of PToV 3CL^pro shared a high similarity with putative cleavage sites in pp1a/pp1ab for the EToV and BRV 3CL^pro (Fig. 7B). Interestingly, a predicted cleavage site (CP3/CP4, Fig. 7A and B) is cleaved by PToV 3CL^pro, which is typically cleaved by the papain-like protease. This cleavage site might be a conserved cleavage site for torovirus 3CL^pros, as it was also predicted to be cleaved by EToV 3CL^pro (27, 43).

FIG 7.

Identification of cleavage sites recognized by PToV 3CL^pro within polyproteins. (A) HEK-293T cells cultured in 24-well plates were transfected with various luciferase reporter plasmids and pRL-TK (Promega), together with 3CL^pro expression plasmid pCAGGS-PToV 3CL^pro or the empty control plasmid. At 28 h posttransfection, the cells were lysed for dual-luciferase assay. (B) The results show the identified sequence recognized by PToV 3CL^pro. Amino acids highlighted in red represents highly conserved positions. (C) The identified sequences in panel B were submitted to PSSMSearch for modeling and visualization of the determinants of substrate specificities of the identified proteins (66). (D) HEK-293T cells in 24-well plates were transfected with CP4/nsp2 reporter or CP4/nsp2 mutants with single amino acid substitutions at the P4 position and pRL-TK, together with 3CL^pro expression plasmid pCAGGS-PToV 3CL^pro or the empty control plasmid (named HA). At 28 h posttransfection, the cells were lysed for dual-luciferase assay.

Comparative sequence analysis of the 12 PToV 3CL^pro cleavage sites in pp1ab revealed that Gln is strictly conserved at the P1 position. The logo representations revealed that Phe at the P4 position was also highly conserved (Fig. 7C) but replaced with Tyr, Leu, Ile, or Met in some of the PToV 3CL^pro cleavage sites. To investigate the possible effects of the Phe-to-Tyr/Leu/Ile/Met substitution at P4, different single amino acid substitutions at the P4 position were constructed using the CP4/nsp2 reporter as a backbone (F-P4-Y, F-P4-L, F-P4-I, F-P4-M, and F-P4-A). As shown in Fig. 7D, this pronounced preference for Phe at the P4 position seems to in favor of efficient cleavage, since substitution of P4-Phe by Ala yielded a 10-fold reduction in cleavage rate, whereas the introduction of natural amino acids Tyr, Leu, Ile, or Met at P4 reduced the rate of cleavage by 2-fold. Taken together, the data suggested that macrobenzene side chains of hydrophobic residues at the P4 position might be preferred in substrates of PToV 3CL^pro.

Investigation of the putative S1 and S4 pockets within PToV 3CL^pro.

To provide a structural rationale for PToV 3CL^pro’s preference at the P4 and P1 positions, we attempted to build a structural model of PToV 3CL^pro. As PToV 3CL^pro and EAV 3CL^pro share the highest amino acid sequence similarities (30.1%), we generated a homology model of PToV 3CL^pro in Modeller 9 using the X-ray crystal structure of EAV 3CL^pro (PDB 1MBM) as a template. The model showed that PToV 3CL^pro adopts a fold similar to that of EAV 3CL^pro and is predicted to has a three-domain structure, with a two-β-barrel structure with His53 and Ser160 as the active-site residues (Fig. 8A). To study the molecular mechanism of PToV 3CL^pro’s preference for the P4 and P1 positions, the structure of Streptomyces griseus proteinase E (SGPE) in complex with a tetrapeptide (PDB 1HPG) was used as the reference structure for PToV 3CL^pro complex construction, since among all known crystal complex structures EAV 3CL^pro shared a high similarity with SGPE structure, which was reported in previous studies (29, 44). We then established a protease-substrate complex system consisting of PToV 3CL^pro and N-terminal self-cleavage sequence SNFSFQ↓SVFSKV and performed molecular dynamics simulations to improve structure accuracy and study the binding model between PToV 3CL^pro and substrate. As shown in Fig. 8B, two putative pockets S1 (Ser-153, Asp159, and His-174) and S4 (Lys-61, Pro-62, Leu-59, Leu-185, and Gly-176) are formed between PToV 3CL^pro and the substrate, where the S4 pocket is composed mainly of several hydrophobic amino acids (Fig. 8C and D). To test the coupling of atomic fluctuations in the S4 and S1 pockets, we calculated the correlation coefficient between atomic fluctuations in the main chain of PToV 3CL^pro and the P4 and P1 positions of substrate. As shown in Fig. 8E, the dynamics of the P4 and P1 positions are highly correlated with the movement of amino acids in the S4 and S1 pockets, which proves the stability of the pockets.

FIG 8.

Investigation of the putative S1 and S4 pockets within PToV 3CL^pro. (A) Homology model of PToV 3CL^pro generated in Modeller 9 using EAV 3CL^pro (PDB 1MBM) as a template. (B) For ease of observation, the simulated substrate in the complex is shown in green. (C) Interaction between the P1 position and each amino acid in the S1 pocket. Light blue or green carbon atoms represent proteases and substrates, respectively. (D) Interactions between the P4 position and each amino acid in the S4 pocket. Red represents Van der Waals force, and blue represents static electricity. (E) Correlation between atomic fluctuations of protease active site residues and the substrate. A warm color in the matrix indicates an increase in correlation. The color of the surface residue is used to indicate its position in the active site.

To further verify the results of the molecular dynamics simulations, several amino acids in the S1 and S4 pockets of PToV 3CL^pro were selected and mutated to alanine. Using pCAGGS-PToV 3CL^pro as a backbone, the putative P1-Gln-binding residues in S1 pocket, Glu159 and His174, were mutated to Ala and studied in a peptide-based cyclized luciferase reporter system (contain N-terminal self-cleavage sequences SNFSFQ↓SVFSKV). When analyzed by dual-luciferase assay and Western blotting, in contrast to the wild-type PToV 3CL^pro (named WT), the D159A and H174A mutants were virtually inactive to induce the activity of luciferase and cleave the substrate, which confirmed that D159 and H174 in S1 pocket affect the effectiveness of protease cleavage (Fig. 9A). To further verify the role of hydrophobic forces in the S4 pocket, the effect of substitutions of the putative P4-Phe-binding residues, Pro62 and Leu185, with Ala (hydrophobic) or Gly (neutral) were also analyzed in this experiment. As shown in Fig. 9B and C, the mutants (P62A, L185A, P62G, L185G, and P62A/L185A) exhibited lower cleavage activities compared to the wild-type 3CL^pro. Nevertheless, the double mutant (P62G/L185G) completely abolished the ability of cleavage activity of PToV 3CL^pro (Fig. 9C). This means that the potential hydrophobic force between the PToV 3CL^pro and P4-Phe side chains is critical for substrate binding, for a double mutant of P62 and L185 to Gly will destroy the substrate binding between PToV 3CL^pro and P4-Phe side chains (Fig. 9C). To further investigate the role of potential interaction in the S4 pocket or S1 pocket in binding different substrate peptide, nsp4/5 (PTIMWQ↓SDDVAD) and nsp7/9 (PIYQPQ↓SVRYAD) of PToV 3CL^pro was used for further assay. As shown in Fig. 10 and 11, the protease loses the ability to cleave the substrate nsp4/5 and nsp7/9 after mutating the S1 and S4 pocket key amino acids.

FIG 9.

Identification of key amino acid within putative PToV 3CL^pro S1 and S4 pockets. (A) HEK293T cells were cotransfected with CP4/nsp2 and pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S1 pocket key amino acid mutants D159A and H174A. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting. (B) HEK293T cells were cotransfected with CP4/nsp2 and pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S4 pocket key amino acid mutants P62A, L185A, P62G, and L185G. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting. (C) HEK293T cells were cotransfected with CP4/nsp2 and pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S4 pocket key amino acid mutants P65A/L185A and P65G/L185G. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting.

FIG 10.

Identification of key amino acid within PToV 3CL^pro S1 and S4 pockets in binding substrate peptide nsp4/5. (A) The luciferase vector contains substrate peptide nsp4/5 (PTIMWQ↓SDDVAD) was cotransfected with pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S1 pocket key amino acid mutants D159A and H174A into HEK293T cells. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting. (B and C) The luciferase vector contains substrate peptide nsp4/5 (PTIMWQ↓SDDVAD) was cotransfected with pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S4 pocket key amino acid mutants P62A, L185A, P62G, L185G, P65A/L185A, and P65G/L185G into HEK293T cells. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting.

FIG 11.

Identification of key amino acid within PToV 3CL^pro S1 and S4 pockets in binding substrate peptide nsp7/9. (A) The luciferase vector contains substrate peptide nsp7/9 (PIYQPQ↓SVRYAD) was cotransfected with pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S1 pocket key amino acid mutants D159A and H174A into HEK293T cells. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting. (B and C) The luciferase vector contains substrate peptide nsp7/9 (PIYQPQ↓SVRYAD) was cotransfected with pRL-TK, along with pCAGGS-PToV 3CL^pro (named WT) or S4 pocket key amino acid mutants P62A, L185A, P62G, L185G, P65A/L185A, and P65G/L185G into HEK293T cells. At 28 h posttransfection, the cells were lysed for dual-luciferase assay and Western blotting.

DISCUSSION

In nidoviruses, 3CL^pro cleaves the viral polyproteins at several conserved sites and is involved in translational processing of the viral nonstructural proteins (22, 45). It has therefore also been referred to as a potential target for the development of antiviral agents. In this study, we sequenced the complete genome of PToV strain (HB-1) and first sought to refine the putative PToV 3CL^pro. We confirmed the cis and trans activity of PToV 3CL^pro and identified 12 cleavage sites on the polyproteins of PToV. The results identify a consensus cleavage sequence for PToV 3CL^pro and help clarify the genomic organization of PToV.

Using a bacterial expression system, we confirmed the putative self-processing activity of the PToV 3CL^pro domain and identified N- and C-terminal 3CL^pro self-processing sites by N-terminal sequencing of candidate processing products. Sequence comparisons suggested that the PToV 3CL^pro active site may employs a catalytic triad comprising the putative Ser-His-Asp residues. However, our results suggested that 3CL^pro employs Ser-160 and His-53 as the active-site residues. The mutagenesis data showed that the replacement of the presumed third catalytic residue, Asp91, had no detectable effect on PToV 3CL^pro activity. It should also be noted that the third catalytic residue of the chymotrypsin catalytic triad is absent in coronavirus 3CL^pro (42, 47). In fact, catalytic residues in different nidovirus 3CL^pro active site differ greatly between them. For example, coronavirus 3CL^pro is a catalytic dyad (Cys-His) with a conserved P1-Gln residue (32), while arterivirus 3CL^pro is a catalytic trimer (Ser-His-Asp) with a conserved P1-Glu residue (31). In this study, we found that PToV 3CL^pro is a serine protease (as is the arterivirus 3CL^pro) that employs Ser-160 and His-53 as the active-site residues with a conserved P1-Gln residue (similar to the coronavirus 3CL^pro). Interestingly, phylogenetic analysis of polyproteins among nidoviruses revealed that the coronaviruses represent the most closely related family to PToV (38, 48, 49). In the case of the 3CL^pro, however, the most significant matches were found in homologs from arteriviruses. Taken together, our results provide an evolutionarily link between the 3CL^pros of arteriviruses, toroviruses, and coronaviruses.

Similar to some other nidovirus 3C-like proteases, PToV 3CL^pro has a pronounced substrate preference for Gln at the P1 position. Surprisingly, Ala substitutions at P1-Gln impaired the C-terminal self-processing but only slightly affect N-terminal self-cleavage. The subsequent mutagenesis of substrate revealed that this “noncanonical” substrate specificity for its N-terminal self-processing required both P1-Gln and P4-Phe. Interestingly, the Phe mediated-noncanonical substrate specificity was not unique to torovirus. Previous studies have shown that Phe is required at both the P2 position and the P3′ position for the C-terminal self-cleavage of SARS-CoV 3CL^pro, which is also conserved in SARS-CoV-2 3CL^pro (50, 51). Unlike SARS-CoV or SARS-CoV-2, based on the 12 cleavage sites of PToV 3CL^pro determined in this study, Phe was also found to be highly conserved at the P4 position recognized by PToV 3CL^pro within polyproteins. Moreover, it has been shown that EToV 3CL^pro cleaves its own internal self-sequence SEFATQ↓AWQTVN, and this cleavage apparently downregulates protease activity, but this internal self-cleavage is not present in PToV 3CL^pro (Fig. 2C and 7A) (27). Sequence comparison showed a high degree of protease similarity within torovirus; however, in the same region the PToV 3CL^pro base sequence was SELATQ↓AWQTVN, where the Phe at the P4 position of EToV 3CL^pro became the Leu in PToV 3CL^pro (data not shown). This mutation might be responsible for the disappearance of internal cleavage of PToV 3CL^pro, which may be beneficial for the proliferation of PToV. Consistent with our results, in the case of coronavirus, mesonivirus, and ronivirus 3CL^pros (22, 25, 33, 52, 53), residues at the P4 position are also thought to be additional specificity determinants.

The general mechanism of catalysis of 3CL^pro relies on the nucleophilic attack of the peptide bond of the substrate by a catalytic Cys/Ser residue in the active site (34, 54). The substrate specificity of the 3CL^pro, on the other hand, is primarily determined by the interactions between the side chains of the P1 residue of the substrate that is accommodated in the S1 pocket of the active site (23, 39). With respect to conserved residues potentially involved in forming important specificity pocket, mutagenesis and molecular dynamics investigation in this study suggested that the Asp159 and His174 residues in PToV 3CL^pro might play an important in binding potential P1-Gln residues. In addition to the conserved residues in putative S1 pocket, many viral 3CL^pros possess other conserved residues that assist in the binding of other residues in substrates (22, 23, 25, 55–57). In the case of PToV 3CL^pro, the conserved residues in putative S4 pocket (Pro-62, Leu-185) are also considered an important determinant for the PToV 3CL^pro-mediated cleavage. All of these residues in the putative S1 and S4 pockets are highly conserved in the 3CL^pros of toroviruses but not the 3CL^pros of coronaviruses and arteriviruses (data not shown). The composition of different residues in the S1 and S4 pockets may be a key factor in the substrate specificities of different residues at the P1 and P4 positions by different nidovirus 3CL^pros. It should be noted that although alanine scanning has been widely used to identify side chains that play important roles in protein-peptide interactions, alanine substitutions at key residues within the protease can affect its structural integrity. Thus, relative to alanine scanning, the scanning other amino acid residues (e.g., hydrophile scanning) may provide new insights regarding the interactions experienced by specific side chains upon protease-substrate complex. Meanwhile, the determination of crystal structures of PToV 3CL^pro-substrate will provide more details on the interactions of PToV 3CL^pro-substrate and the molecular determinants of substrate specificities of PToV 3CL^pro, which deserve further investigation.

In summary, this first characterization of PToV 3CL^pro supported and developed previous findings for other nidoviruses. It appears that torovirus 3CL^pros have evolved properties that distinguish them from other nidovirus proteases. In particular, this applies to the unique substrate specificity identified for the PToV 3CL^pro in this study. The use of Ser-160 and His-53 as the active-site residues and a P4 residue with Phe as an additional specificity determinant of PToV 3CL^pro provides additional insights for understanding the diversity of nidovirus 3CL^pros.

MATERIALS AND METHODS

Sequence analysis of the PToV genome.

The positive PToV feces samples from pigs with severe diarrhea were diluted in phosphate-buffered saline (PBS [pH 7.4]) solution and finally clarified by centrifugation at low speed (4,000 × g) for 10 min. The supernatants were then collected and subjected to RNA extraction. The total RNA was extracted with the RNA Solv reagent (Omega Bio-Tek) and dissolved in diethyl pyrocarbonate-treated water. The cDNA was synthesized with reverse transcriptase using the Roche reverse transcription system. For genome sequence analysis, oligonucleotide primer pairs targeting the genome of PToV were used, and the resulting amplicons were cloned using pGEM-T vector according to the manufacturer’s protocols (Promega).

Inferring virus evolutionary history.

Virus 3CL^pros were aligned by using the E-INS-i algorithm implemented in the program MAFFT (58). The evolutionary history was then inferred by using the maximum-likelihood method and the Le_Gascuel_2008 model (59). The tree with the highest log likelihood (−36,582.01) is shown. Initial trees for the heuristic search were obtained automatically by applying neighbor-joining and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model and then selecting the topology with superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites (5 categories [+G, parameter = 1.9218]). The rate variation model allowed for some sites to be evolutionarily invariable ([+I], 0.62% sites). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. This analysis involved 90 amino acid sequences. There were 633 positions in the final data set. Evolutionary analyses were conducted in MEGA X (60).

Plasmids.

The complete genomic sequence of porcine torovirus isolate HB-1 (GenBank accession number MH603532) was used for the source for all constructs described below. First, to construct pET28a-GST, the GST coding sequence GST tag was amplified from pET-42b and cloned into the C-terminal of the pET28a polyclonal site. Then, to generate pET28a-WT-GST and pET28a-S3225A-GST, the predicted coding sequence of PToV 3CL^pro, together with short flanking sequences (pp1a/pp1ab 3012–3392), were amplified by PCR and subclone into pET28a-GST.

To construct p-GST-GST, the GST tag was amplified from pET-42b and then cloned into pCAGGS vector. Subsequently, the pp1a/pp1ab 3012–3392 was amplified from HB-1 and the resulting PCR products were ligated with p-GST-GST by using T4 DNA ligase to generate p-GST-WT-GST. To construct pCAGGS-PToV-3CL^pro, the coding region for pp1a/pp1ab amino acids Ser-3096 to Gln-3382 was amplified from p-GST-GST-WT and subcloned into pCAGGS vector.

To construct a total of 13 reporter plasmids, the predicted peptides were amplified from pp1a/pp1ab of PToV HB-1 and subcloned into the 233D reporter vector, including CP3/CP4, CP4/nsp2, nsp2/nsp3, nsp3/nsp4, nsp4/nsp5, nsp5/nsp6, nsp6/nsp7, nsp7/nsp9, nsp9/nsp10, nsp10/nsp11, nsp11/nsp12, nsp12/nsp13, and nsp2 internal.

Bacterial expression and N-terminal sequencing.

To express recombinant proteins encoded by pET28a-GST, pET28a-WT-GST, and pET28a-S160A-GST, E. coli BL21(DE3) was used. Freshly transformed cells were cultured in Luria-Bertani medium containing 50 mg/liter kanamycin until the optical density at 600 nm reached 0.6 to 0.9. The cultures were then separated, and protein expression was induced with 0.8 mM IPTG in one of the cultures. The induced and uninduced cultures were incubated at 37°C for an additional 5 h under severe shaking (225 rpm). After 5 h of incubation, the cells were collected in PBS and lysed by sonication. The total cell lysates were obtained by adding Laemmli sample buffer, followed by incubation at 95°C for 5 min. Protein expression was analyzed by SDS-PAGE and Coomassie staining.

An appropriate amount of pET-WT-28a protein sample was separated onto a 12% SDS-polyacrylamide gel and transferred onto polyvinylidene fluoride membranes (Millipore, Burlington, MA). The membranes were then stained with 0.025% Komas Blue R-250 containing 40% methanol and subsequently decolorized with 50% methanol. The area containing the desired protein was excised, and the protein was subjected to N-terminal sequencing by Edman degradation (Biotech Bioengineering Co., Shanghai, China).

Dual-luciferase assay.

HEK-293T cells were obtained from the China Typical Culture Collection Center (Wuhan, China) and grown at 37°C and 5% CO₂ in Dulbecco modified Eagle medium (Invitrogen, Madison, WI), to which 10% fetal bovine serum was added. The luciferase-based reporter constructs and negative controls (233D and nsp2 internal, respectively) were used to identify the putative recognition sites of PToV 3CL^pro. Briefly, HEK-293T cells in 24-well plates were transfected with various luciferase-based reporter plasmids and pRL-TK (Promega), together with 3CL^pro expression plasmids or the empty control plasmid. At 28 h posttransfection, the cells were lysed, and a luciferase reporter assay system (Promega) was used to test the luciferase activities in the lysed cells. The activities were normalized to the corresponding Renilla luciferase activities.

Western blotting.

HEK-293T cells cultured in six-well plates were transfected with various plasmids. At 28 h after transfection, the cells were collected by using lysis buffer (Beyotime) and subsequently added to sample upload buffer (Beyotime) and then boiled for 10 min. Subsequently, the samples were gummed with SDS-PAGE and transferred to polyvinylidene fluoride membranes (Millipore). Anti-GST antibody (MBL, Nagoya, Japan), anti-β-actin antibody (Antgene, Wuhan, China), and anti-HA antibody (MBL) were used to detect the respective proteins.

Homology modeling.

A homology model of PToV 3CL^pro was generated with Modeller 9 using EAV 3CL^pro (PDB 1MBM) as the principal template (31, 61). PToV 3CL^pro in complex with substrate peptides was generated using the structure of SGPE in complex with a tetrapeptide (PDB 1HPG) as the reference structure, because EAV 3CL^pro shared a high similarity with SGPE structure among all known crystal complex structure (29, 44). The substrate in the complex was replaced with the CP4/nsp2 autocleavage sequence of PToV 3CL^pro (SNFSFQ↓SVFSKV) using SYBYL-X (v.2.0; https://omictools.com/sybyl-x-tool). In order to improve structure accuracy, MD simulations were performed using the Amber ff14SB forcefield implemented in the GROMACS 2018 software package as previously described (62, 63).

Data availability.

The complete genomic nucleotide sequence of porcine torovirus isolate HB1 has been deposited in GenBank under accession number MH603532.