The complete genomic sequence of SMoV Nova Scotia isolate NSPer3 (accession numbers, KU200456-KU200457">KU200456-KU200457) has been described previously (Bhagwat et al., 2016) and this isolate was the source for all constructs described below. Reverse transcription was conducted using SuperScript IV (Thermo Fisher) and primer P610R (see Table 1 for primers) to generate cDNA which was then used as a template for PCR amplification using Q5 HF DNA polymerase (New England Biolabs). RT-PCR fragments corresponding to constructs NTB’-Pol’, 365-735, 1-865 and 501-1691 were generated using specific primers that included additional restriction sites for cloning into vector pCITE4a (Novagen) (Table 1). Fragments for all other constructs were synthesized commercially by GeneArt (Thermo Fisher) and were subsequently subcloned into pCITE4a. Fragments were inserted into the BamHI-XhoI (NTB’-Pol’, 365-735, 1-865, and 501-1691) or the NcoI-BamHI restriction sites (most other constructs) of the pCITE4a polylinker, resulting in an N-terminal in frame fusion to the S-tag contained within the vector. The only exception was the VPg-Pro construct, which was inserted into the MscI-BamHI sites of pCITE4a, allowing the synthesis of a viral protein with only three additional amino acids at its N-terminus (including one additional methionine as a start codon). Therefore, this construct (and mutated derivative) was not fused to the S-tag so as to facilitate expression of the native protease. In all cases, a stop codon was inserted immediately downstream of the viral sequence. Mutations were inserted into the parent constructs either by site-directed mutagenesis (Fisher and Pei, 1997) using specific primers or using the mutagenesis service of GeneArt. DNA sequence for all constructs was verified by Sanger Sequencing using the ABI 3500 series Genetic Analyzer (Thermo Fisher).

The deduced amino acid sequence of the P1 and P2 polyproteins from SMoV NSPer3, BRNV isolate Alyth (accession numbers, FN908128-FN908129">FN908128-FN908129), tomato ringspot virus isolate Rasp2 (ToRSV-Rasp2, NC_003839-NC_003840), arabis mosaic virus isolate NW (ArMV-NW, NC_006056-NC_006057), chocolate lily virus A (CLVA, NC_016443-NC_016444) and dioscorea mosaic-associated virus (DMaV, KU215538-KU215539) were used for Table 2 and for the alignments shown in the Supplementary Figures S1, S2. Multiple amino acid sequence alignments were produced using Clustal Omega (Sievers et al., 2011).

For in vitro translation assays, the rabbit reticulocyte system was chosen because wheat germ extracts were previously reported to contain inhibitors of 3C-like proteases (Shih et al., 1987; Margis et al., 1991). Protein translation reactions were carried out using the TnT Quick coupled transcription/translation rabbit reticulocyte system (Promega) as previously described (Wetzel et al., 2013). Briefly, protein labeling with EasyTag L-[³⁵S]-methionine (PerkinElmer) was carried out at 29°C for 90 min followed by translation termination by the addition of an RNase A and cold methionine mix. Cis-cleavage reactions were directly diluted in an equal volume of protease buffer [10 mM HEPES, pH 6.5, 0.1% CHAPS (3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate hydrate), 10 mM DTT (dithiothreitol) and 30% glycerol] (May et al., 2014) and incubated at 16°C overnight to facilitate proteolytic processing. For trans-cleavage reactions, samples were prepared by mixing the unlabelled VPg-Pro translation product with [³⁵S]-methionine labeled products from RNA2 construct(s) at a ratio of 5:1. The mixture was then diluted in an equal volume of protease buffer and incubated at 16°C overnight. Following overnight incubation, an equal volume of 2X SDS protein loading buffer was added (Laemmli, 1970). Samples were heated at 60°C for 10 min followed by separation by 10 or 12% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Protein bands were visualized using a phosphoimager (Cyclone Plus, PerkinElmer).

Based on alignments of P1 polyproteins amongst Canadian SMoV isolates and with related secovirids, putative cleavage sites were previously predicted including Q⁴⁶⁵/G, Q⁹⁶⁴/G, Q⁹⁸⁹/G, and Q¹²²⁰/G (Figure 1, numbering correspond to the amino acid position starting from the beginning of the polyprotein) (Thompson et al., 2002; Bhagwat et al., 2016). The last three proposed cleavage sites would delineate the VPg, Pro, and Pol domains, while the first cleavage site was tentatively proposed to be upstream of the NTB domain. Re-examination of amino acid sequence alignments revealed that two additional possible cleavage sites were present: Q¹⁴⁶/G and Q³⁴⁸/S, which correspond approximately to the positions of the previously characterized X1-X2 and X2-NTB cleavage sites of two nepoviruses [ToRSV and ArMV] (Wang and Sanfacon, 2000; Wetzel et al., 2008) (Supplementary Figure S1).

Proteolytic processing of SMoV P1 was initially investigated using a partial polyprotein precursor containing the entire VPg and Pro domains as well as a portion of the C-terminal region of the NTB domain (NTB’) and N-terminal region of the Pol domain (Pol’). This construct, referred to as NTB’-Pol’, included the predicted NTB-VPg, VPg-Pro and Pro-Pol cleavage sites (Figure 2A, i). To confirm that any detected proteolytic cleavage was due to the activity of the 3C-like Pro, we also generated a mutant derivative of the NTB’-Pol’ construct in which the conserved cysteine of the catalytic triad (C1171) was mutated to alanine (referred to as Pro-null). Results from the in vitro translation assays showed accumulation of the expected precursor polyprotein (34 kDa) after a 90 min translation reaction (Figure 2B, 0 h, lane 1). A few smaller minor bands were also observed but were likely not due to a specific proteolytic event directed by the 3C-like protease, since they were also observed in the Pro-null derivative (Figure 2B, compare lanes 1 and 3). Rather, they were likely the result of internal translation initiation or premature translation termination events. After an overnight incubation in the protease buffer (see Materials and Methods), a predominant cleavage product was observed for the wild-type NTB’-Pol’ but not for the Pro-null derivative (Figure 2B, 16 h, compare lanes 2 and 4). This band had an apparent molecular mass of 30 kDa, which is close to the calculated size for the predicted VPg-Pol’ cleavage product (30.3 kDa) after processing at the NTB-VPg cleavage site (Figure 2B, lane 4). Other possible cleavage products that could arise from secondary cleavage at the remaining VPg-Pro and/or Pro-Pol cleavage sites (Pro-Pol’, VPg-Pro, and Pro) were not confidently detected over the background.

Next, we tested a truncated precursor (NTB’-Pro) which lacks the Pol’ sequence, and therefore the Pro-Pol cleavage site (Figure 2A, ii). As above, primary cleavage at the predicted NTB-VPg site was detected resulting in the accumulation of a protein corresponding in size to the VPg-Pro cleavage product after an overnight incubation (Figure 2B, lane 6). In addition, smaller amounts of another cleavage product were detected that likely corresponded to the mature Pro (25.1 kDa) after secondary processing at the VPg-Pro cleavage site. The other expected cleavage product, i.e., the mature VPg protein, was not detected from these gels due to its small size (2.8 kDa).

We further characterized the processing of NTB’-Pol’ by introducing mutations in potential cleavage sites. We chose to mutate the conserved glutamine (Q) of the -1 position of the cleavage site to an alanine (Figure 2A, iii–v). A similar mutation in ToRSV cleavage sites was previously shown to completely abolish cis- or trans-cleavage (Carrier et al., 1999). Single mutation of the predicted Pro-Pol cleavage site (A1220 mutant, Figure 2A, v) did not drastically change the cleavage pattern compared to the wild-type polyprotein and resulted in accumulation of the VPg-Pol’ product after cleavage at the predominant NTB-VPg site (Figure 2B, compare lanes 4 and 12). As expected, single mutation of the predicted NTB-VPg cleavage site (A964 mutant, Figure 2A, iv) prevented the accumulation of the VPg-Pol’ cleavage product (Figure 2B, compare lanes 4 and 10). Instead, the mutation resulted in cleavage at the remaining VPg-Pro and Pro-Pol sites (Figure 2A, iv). This was evidenced by the accumulation of cleavage products corresponding in size to NTB’-Pro (31.5 kDa), Pro-Pol’ (27.5 kDa) and Pro (25.1 kDa) (Figure 2B, lane 10). Mutation of both the NTB-VPg and Pro-Pol cleavage sites (A964 + A1220 double mutant, Figure 2A, iii) resulted in cleavage at the only available cleavage site between VPg and Pro resulting in the accumulation of the Pro-Pol’ cleavage product (27.5 kDa) (Figure 2B, lane 8). The presence of a large amount of the precursor polyprotein after an overnight incubation for all tested wild-type or mutated NTB’-Pol’ derivatives (Figure 2B, lanes 2, 4, 6, 8, 10, and 12) suggests that in vitro cleavage by the 3C-like Pro is not efficient. Despite this observation, our results confirm cleavage of the NTB’-Pol’ construct by the 3C-like Pro at the three predicted cleavage sites, with the NTB-VPg cleavage site being recognized the most efficiently at least in vitro.

To determine whether proteolytic processing can occur at other predicted P1 polyprotein cleavage sites, we generated a construct that spans the entire N-terminal region of the P1 polyprotein up to and including the Pro domain (X1-Pro; 137 kDa) (Figure 3A, i). Although the X1-Pro precursor included five possible cleavage sites, the only detectable cleavage observed was between NTB and VPg. This was determined based on the accumulation of cleavage products that corresponded in size to X1-NTB (109.1 kDa) and VPg-Pro (27.9 kDa) (Figure 3B, lane 2). These cleavage products were not detected in the Pro-null derivative (Figure 3B, lane 4), confirming that cleavage was due to the activity of the 3C-like Pro. Taken together, these results are similar to those observed with the NTB’-Pol’ construct and indicate that the SMoV 3C-like Pro preferentially cleaves at the NTB-VPg site in vitro.

Since suboptimal cleavage events were detectable when one or more cleavage sequences were mutated in the NTB’-Pol’ construct (Figure 2), we used a similar strategy to complete the mapping of the N-terminal domain(s) of P1. Using the X1-Pro construct, a series of constructs with mutations at one or more cleavage sites were generated (Figure 3A, ii–iv). For each mutant tested, translation reactions were ran simultaneously on a 12% SDS-acrylamide gel to visualize smaller cleavage products (Figure 3C, upper panel) and on a 10% gel to optimize the separation of larger cleavage products (Figure 3C, lower panel). Compared to the wild-type, mutation of the NTB-VPg cleavage sequence (A964 mutant, Figure 3A, ii) prevented the release of the X1-NTB and VPg-Pro cleavage products (Figure 3C, compare lanes 2 and 4). Instead, new cleavage products were observed. A band with an apparent molecular mass of approximately 100 kDa likely corresponds to NTB-Pro, suggesting that cleavage upstream of the NTB domain occurred at Q³⁴⁸/S (calculated molecular mass for NTB-Pro of 98.3 kDa) rather than at the originally predicted Q⁴⁶⁵/G (calculated molecular mass for NTB-Pro of 84.8 kDa). Two other major cleavage products were also detected. The first cleavage product migrated at approximately 38 kDa and could correspond to the entire N-terminal region of P1 upstream of Q³⁴⁸/S (calculated molecular mass of 43 kDa). The second one was approximately 20 kDa and could result from cleavage at Q¹⁴⁶/G, which would define two small protein domains upstream of NTB. We will refer to these domains as X1 (20.6 kDa) and X2 (22.3 kDa), by analogy to the X1 and X2 domains mapped in the N-terminal region of the P1 polyproteins of two nepoviruses (Wang and Sanfacon, 2000; Wetzel et al., 2008). Introduction of the Pro-null mutation in the X1-Pro A964 mutant (double mutant A964 + Pro-null) prevented the accumulation of these new cleavage products, confirming that the activity of the 3C-like Pro is required (Figure 3C, lane 14).

In contrast to results with the smaller NTB’-Pol’ precursor (Figure 2), cleavage was not observed between the VPg and Pro domains in the X1-Pro polyprotein after mutation of the NTB-VPg site, as release of the mature Pro was not detected for the A964 mutant (Figure 3C, lane 4). This suggests that the VPg-Pro cleavage site may be suboptimal in the presence of the X1-X2 and X2-NTB cleavage sites. Indeed, introducing a second mutation in the VPg-Pro cleavage site in addition to the NTB-VPg cleavage site mutation (double mutant A964 + A989) did not alter the cleavage product banding pattern (Figure 3C, compare lanes 4 and 6). To confirm that cleavage occurred between X2 and NTB at Q³⁴⁸/S, and between X1 and X2 at Q¹⁴⁶G, we introduced mutations of these cleavage sites in the X1-Pro double mutant (A964 + A989), creating triple and quadruple mutants. Mutation of the X2-NTB Q³⁴⁸/S cleavage site (triple mutant A348 + A964 + A989, Figure 3A, iii) resulted in the loss of the X1-X2 cleavage product and processing at the cleavage site between X1 and X2. This was evidenced by the accumulation of two cleavage products corresponding to X2-Pro (120.8 kDa) and X1 (Figure 3C, lane 8). Similarly, mutation of the X1-X2 Q¹⁴⁶/G cleavage site (triple mutant A146 + A964 + A989, Figure 3A, iv) resulted in the loss of the X1 product and processing at the X2-NTB cleavage site (Q³⁴⁸/S), resulting in the accumulation of NTB-Pro and X1-X2 (Figure 3C, lane 10). A quadruple mutant with simultaneous mutation of the Q¹⁴⁶/G, Q³⁴⁸/S, Q⁹⁶⁴/G, and Q⁹⁸⁹/G cleavage sites was not cleaved by the 3C-like protease (Figure 3C, lane 12) suggesting that the putative Q⁴⁶⁵/G site was not recognized. Taken together, our results identify a total of five cleavage sites in the P1 polyprotein with the consensus sequence AxEQ/(G or S) (Table 2). Similar to what was observed for two nepoviruses, these cleavage sites define six protein domains in the P1 polyprotein, namely X1, X2, NTB, VPg, Pro, and Pol.

Scanning of P2 for putative cleavage sites did not reveal any sites that fit the consensus sequence established for the P1 polyprotein above. However, a related AYEE⁴⁵²/G sequence was previously identified as a putative cleavage site between the MP and CP domains (Thompson et al., 2002; Bhagwat et al., 2016). This sequence would meet the P1 consensus with the exception of the presence of a glutamate (E) rather than a glutamine (Q) at the -1 position. In addition, a DIEE⁴³⁶/G sequence was also found in the P2 polyprotein of SMoV NSPer3 and NSPer51, although this sequence is altered to DIDE⁴³⁶/G in all other isolates. We first performed a trans-cleavage assay using a partial P2 polyprotein precursor termed 365-735 (numbering refers to the amino acids from P2 that are included in the precursor; Figures 4A,B). This precursor overlapped both putative MP-CP cleavage sites (Figure 4B). The VPg-Pro was used as a source of active RNA1-encoded protease to be provided in trans and was synthesized by in vitro translation in the presence of unlabelled methionine. Processing of the 365-735 precursor by VPg-Pro resulted in the accumulation of two cleavage products after an overnight incubation. These products corresponded in size to the C-terminal region of the MP (14.3 kDa) and the N-terminal region of the CP (30.7 kDa) (Figure 4B,C, lane 2). Using mutagenesis, we investigated which of the two possible cleavage sites between the proposed MP and CP domains was recognized by VPg-Pro for processing. As above, we mutated the glutamate at the -1 position of the E/G dipeptide to an alanine. Mutation of the first putative cleavage site (A436 mutant) did not prevent the release of the cleavage products (Figure 4C, lane 4). In contrast, mutation of the second putative cleavage site (A452 mutant) abolished the processing (Figure 4C, lane 6). As a control, we also tested a derivative of the VPg-Pro that incorporated the Pro-null mutation. In vitro translation with labeled methionine confirmed that both the wild-type and mutant derivative of VPg-Pro were expressed to similar levels (Figure 4D). As expected, cleavage of the 365-735 precursor was not observed when incubated with the Pro-null derivative of VPg-Pro (compare Figure 4E, lane 4 to Figure 4F, lane 4).

Next, we tested whether VPg-Pro could cleave the P2 polyprotein at other cleavage sites. All potential Q/G, Q/S, or E/G dipeptides were considered, even if they did not entirely meet the consensus for an E or Q at the -2 position and an A at the -4 position. Dipeptides Q³¹⁴/G, Q⁹⁴⁹/G, E¹⁴¹²/G, and Q¹⁵⁸⁹/S were identified as putative cleavage sites (Figure 4A). In addition, since some picornavirus 3C proteases show relaxed specificity for the +1 position (Seipelt et al., 1999), we also considered dipeptides Q⁸⁷⁵/N and Q¹⁰⁸⁶/T. We tested a collection of partial P2 polyprotein precursors that covered these cleavage sites (Figure 4A). However, we could not detect any processing events that could be attributed to the activity of a functional VPg-Pro (compare Figures 4E, F).

To rule out the possibility that the protein conformation of the partial P2 precursors may have affected the proper presentation of the cleavage sites to the VPg-Pro, we generated two larger overlapping partial P2 precursors that included either the N-terminal region of the polyprotein (1-865), or the C-terminal region starting after the predicted CP(s) domain (501-1691) (Figures 5A,B). Similar to the 365-735 precursor, cleavage was detected between the MP and CP domains in the larger 1-865 precursor and was dependent on the catalytic activity of the VPg-Pro supplied in trans (Figure 5C, lanes 4 and 10). However, no other cleavage events were detected in this precursor. Precursor 501-1691 was not cleaved by VPg-Pro since we did not observe a different banding pattern when the wild-type or mutated VPg-Pro were supplied in trans (Figure 5C, lanes 6 and 12). Together these results suggest that the RNA1-encoded VPg-Pro intermediate is active on a single trans-cleavage site over the entire RNA2 polyprotein. The results also provide an updated cleavage consensus sequence for SMoV cleavage sites of AxE (Q or E)/(G or S) (Table 2).