The A113V mutation within the p28 portion of p28/p88 diminishes sgRNA1 levels and delays TCV systemic spread

We previously reported the isolation of numerous TCV mutants containing varying numbers of single-nt mutations [21]. One such mutant, designated Mut17, had four different single-nt mutations within the p28/p88 coding region. Among them, only one – C972U – did not alter the aa sequence of p28/p88. The remaining three mutations, C401U, G674A, and G2332U, altered identities of three aa residues (Fig 1B). Specifically, alanine, arginine, and glycine residues at positions 113, 204, and 757 of p28/p88 were mutated to valine, histidine, and a premature stop codon, hence the names A113V, R204H, G757*, respectively (Fig 1B).

Mut17 was originally identified in cells infected by a TCV replicon known as TCV-dMP_sg2R. This replicon harbored a deletion within p8/p9 movement protein (MP) genes, and an mCherry gene in place of the capsid protein gene (Fig 1B). As a result, it replicated robustly in single cells, but could neither spread cell-to-cell nor assemble virus particles. When the four Mut17 mutations were re-introduced into TCV-dMP_sg2R together, the resulting mutant was unable to replicate in single cells, as judged by Northern blot hybridizations (Fig 1C, lane 3). Note that Mut17 arose from single-cell infections with a movement-defective virus. Such a mutant, despite its apparent replication defect, could have replicated to high abundance in a cell with the help of wildtype (wt) RPs translated from co-infecting genome copies that were themselves blocked from replication by reproductive bottlenecks, as predicted by the Bottleneck, Isolate, Amplify, Select (BIAS) hypothesis [22].

When each of the mutations was introduced into TCV-dMP_sg2R separately, the resulting A113V and N303N mutants replicated to levels similar to the mutation-free TCV-dMP_sg2R (Fig 1C, compare lanes 4 and 6 with lane 2). By contrast, the R204H mutant replicated to very low, barely detectable levels (Fig 1C, lane 5), whereas the G757* mutant RNA was undetectable (Fig 1C, lane 7). Intriguingly, while a faint sgRNA1 (sg1) band was detected in both TCV-dMP_sg2R and N303N samples (Fig 1C, lanes 2 and 6), this band appeared to be missing in the A113V sample (Fig 1C, lane 4). This observation prompted us to re-examine the impact of A113V mutation in wt TCV infections. To this end, A113V was introduced into an Agrobacterium (agro)-infiltration-compatible infectious clone of wt TCV and the resulting mutant was brought into N. benthamiana plants, with the wt TCV clone as control (Fig 2). Moreover, to assess whether the inoculum dose had any effect on infection outcomes, we tested three different agro concentrations, with the respective optical densities (OD; at λ = 600 nm) adjusted to 1, 0.1, and 0.01, respectively. We were surprised to find that at 4 days post agro-infiltration (4 dpai), lower agro concentrations consistently correlated with higher viral gRNA levels in the agro-infiltrated local leaves (IL), for both wt TCV and the A113V mutant (Fig 2A, compare lanes 2–4, 5–7). Therefore, lower inoculum dose probably facilitated more efficient replication in primary cells that received viral cDNAs.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. A113V diminishes sgRNA1 levels and delays TCV systemic spread.

A. The A113V mutant failed to produce sgRNA1 in 4 dpai inoculated leaves (ILs) to levels detectable with NB. Agrobacterium suspensions harboring the A113V mutant, and the wt TCV control, were delivered into N. benthamiana leaves at three different concentrations (1, 0.1, and 0.01 OD). Relative sg mRNA1 (sg1) and sg mRNA2 (sg2) levels were calculated for ILs as ratios of sg mRNA levels to their cognate genome (g) levels. B. Viral RNA of A113V mutant was undetectable with NB in 7 dpai SLs. C. Viral RNA of A113V was marginally detectable with NB at 14 dpai (SLs) only in plants receiving the mutant at a very low dose (0.01 OD). D - F. MP+ plants expressing a heterogenous (RCNMV) MP mitigate the A113V defect in ILs and SLs. D. NB detection of wt TCV and A113V viral RNA levels in MP+ and wildtype N. benthamiana leaves at 4 dpai. Note the lower levels of sgRNA1 in lanes with the A113V mutant (lanes 4,5,9,10). E and F. wt TCV and A113V viral RNA levels in MP+ and wildtype N. benthamiana SLs at 7 dpai (E) and 14 dpai (F).

https://doi.org/10.1371/journal.ppat.1013842.g002

In contrast to wt TCV, the A113V mutant failed to produce detectable levels of sgRNA1 in ILs at all three inoculation doses (Fig 2A, lanes 5–7). Consistent with the crucial role of sgRNA1 as mRNA for p8 and p9 MPs, viral RNAs of the A113V mutant were undetectable in systemic leaves (SLs) at 7 dpai (Fig 2B, lanes 5–7). Furthermore, at 14 dpai, A113V RNAs were present only in SLs of plants inoculated with the 0.01 OD dose, and at low titers (Fig 2C, lane 7; also see later). Consistent with the diminished viral RNA levels, the A113V-inoculated plants were symptom-free until at least 13 dpai, though some of them developed delayed symptoms afterwards (see later). Notably, the control plants inoculated with wt TCV developed clearly discernable symptoms by 7–8 dpai. Together, these results strongly suggest that while the A113V mutant replicated robustly in individual primary cells, its systemic spread was severely compromised, likely due to diminished sgRNA1 synthesis or stability, hence dearth of p8 and p9 MPs.

The delayed systemic spread of the A113V mutant could have been caused by other replication defects in addition to lower sgRNA1 levels. To assess this possibility, we tested the A113V mutant in transgenic N. benthamiana plants expressing MP of red clover necrotic mosaic virus (RCNMV), hereafter referred to as MP+ plants [23]. The MP+ plants have been previously shown to complement TCV mutants with deletions in p8 and p9 [24]. As shown in Fig 2D, 2E, and 2F, in MP+ plants the A113V mutant yielded wt levels of viral gRNA in 4 dpai ILs as well as 7 and 14 dpai SLs. Of note, the sgRNA1 levels in A113V-infected tissues remained low or undetectable until at least 14 dpai. These results strongly suggested that reduction in sgRNA1 levels, and hence shortage of p8 and p9 MPs, were the primary culprit of systemic movement delay observed in A113V-infected wt N. bethamiana plants.

A113V-infected plants exhibiting delayed symptoms contain viral variants harboring second-site mutations

By 21 dpai, some, but not all, A113V-infected plants developed systemic symptoms of varying severity (Fig 3A). Some of the plants exhibited clear curling and crinkling of the top young leaves, whereas others had only mild vein clearing on older leaves. The extent of stunting also differed among the plants. We hence evaluated TCV RNA levels in SLs of these plants. For this analysis we selected four plants each from the 0.1 OD and 0.01 OD inoculum groups. As shown in Fig 3B, viral RNA accumulation varied considerably in these plants, ranging from barely detectable in some plants (lanes 6–9), to wt levels in others (lanes 4, 5, 11). These results led us to speculate that the dominant TCV variants in some of the plants might have reverted to wt sequence or acquired compensatory second-site mutation(s).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Identification of second-site mutations from SLs of A113V-infected N. benthamiana plants.

A. Systemic symptoms of wt TCV and A113V-infected plants at 21 dpai. Bottom images reflect the actual size difference of two systemically infected leaves. B. NB detection of A113V viral RNA levels at 21 dpai, in 8 different plants. C – F. Chromatographs showing the emergence of reversion and/or second-site mutations in 4 different plants at 21 and 42 dpai. G. Summary of mutations identified from six different A113V-infected plants at 21 and 42 dpai.

https://doi.org/10.1371/journal.ppat.1013842.g003

To test this, we subjected the viral RNA in 3 plants of the 0.1-OD group and 3 plants of the 0.01-OD to reverse transcription (RT)-PCR to amplify the entire p28/p88 coding region. The RNA samples selected corresponded to lanes 4–6 and 8–10 in Fig 3B. The amplified cDNA fragments were subsequently sequenced to identify dominant variants in the respective samples. The results showed that TCV cDNA of sufficient quantities was obtained in five of the six plants (except for plant #3; Fig 3G). At 21 dpai, just two plants (#1 and 6) contained the original mutant with C401U (A113V) as the sole mutation (Fig 3C and 3D; the U/C residues correspond to A/G peaks in antisense reads. Highlighted with magenta boxes). However, the mutant was not stable in plant #1 because, by 42 dpai, the C401U mutation became a minor peak co-existing with the dominant C peak of wt TCV (Fig 3C, bottom graph). On the other hand, although the C401U mutation remained stable in plant #6 by 42 dpai, an additional G1237A mutation emerged, altering aa position 392 from a glutamate to a lysine (E392K) (Fig 3D; blue box).

Strikingly, the E392K mutation in plant #6 was also detected in plant #2, at the earlier 21 dpai time point (Fig 3E). In this plant the E392K (G1237A) mutation was accompanied by another de novo mutation, K389E (A1228G; yellow box), as well as the original A113V (C401U) mutation. More intriguingly, by 42 dpai, a minor C401 (wt) peak emerged under the dominant C401U peak (Fig 3E). Meanwhile, a dominant A1228 (wt) peak overshadowed the A1228G (K389E) de novo mutation by 42 dpai (Fig 3E). Nonetheless, the G1237A (E392K) de novo mutation was still stable (Fig 3E). The fact that the E392K mutation emerged in two independent plants (#6 and 2. Fig 3D and 3E) strongly suggested that this mutation may have compensated for the defect associated with the A113V mutation (more later).

The dominant variant in plant #5 was likewise noteworthy (Fig 3F). At 21 dpai, a minor, yet unmistakable G2321U peak co-existed with the wildtype G2321 (green box). The G2321U mutation would have changed aa residue 753 from tryptophan to leucine (W753L). By 42 dpai, the G2321U peak rose to co-dominance with G2321 (wt). Curiously, by then the original C401U (A113V) mutation was also beginning to be accompanied by a minor C401 (wt) peak. Thus, W753L was probably another mutation that compensated for the A113V defect. In summary, two new mutations, E392K and W753L, may have been selected for their ability to rescue the loss of sgRNA1 in the A113V mutant infections.

Finally, the dominant variant in plant #4 incurred two second-site mutations, G1434A and U2097G, in addition to a minor C401 (wt) peak under the dominant C401U peak, at 21 dpai (Fig 3G). However, neither of them altered any aa residue. Moreover, in this plant the C401U mutation reverted to wt by 42 dpai. Together these data suggested that the sgRNA1 loss caused by the A113V mutation required additional compensatory mutations, or simple reversion to wt, in order to restore systemic spread to the virus. It is further noteworthy that these compensatory mutations occurred within p88 RdRp, at positions distant from the original A113V mutation at the primary aa sequence level.

E392K and two double mutants yield sgRNA1-like alternative sgRNAs and alleviate A113V-caused delays in TCV systemic spread

To resolve whether the second-site mutations could indeed rescue the A113V defects, we introduced these mutations, including K389E, E392K, and W753L, as well as K389E-E392K double mutations (designated KEEK for simplicity), into wt TCV and A113V backbones (Fig 4A). Northern blot hybridization results revealed that in 4 dpai ILs, these mutants produced drastically different levels of viral gRNA and sgRNAs, and novel sgRNA species in some cases. It should be noted that in this set of experiments the A113V mutant did accumulate low levels of sgRNA1 (Fig 4B, compare lanes 13–14 with 3–4). Consistent with earlier observations, A113V systemic movement was delayed as its RNAs were undetectable in 7 dpai SLs, and detectable in just one of the two plants examined at 14 dpai (Fig 4C and 4D, lanes 13–14). More tellingly, sequence analysis revealed that in this plant the A113V mutation has reverted to wt TCV (Fig 4E, sample #14).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. In planta fate of second-site mutations by themselves, and in combination with A113V.

A. A list of 8 mutants created. B – D. NB determination of viral RNA levels for the 8 mutants in ILs at 4 dpai (B), in SLs at 7 (C) and 14 (D) dpai. Relative sg mRNA levels were calculated for ILs as described in Fig 2. E. Dynamic changes of the mutations in 14 dpai descendants of the 8 mutants as determined by sequencing of RT-PCR products.

https://doi.org/10.1371/journal.ppat.1013842.g004

The K389E mutation by itself completely abolished sgRNA1, but also substantially decreased the levels of gRNA and sgRNA2 in 4 dpai ILs (Fig 4B, compare lanes 5–6 with 3–4). Combining K389E with A113V failed to resurrect sgRNA1, and instead decreased gRNA and sgRNA2 levels further (Fig 4B, lanes 15–16). Moreover, neither K389E single mutant nor K389E-A113V double mutant could be detected in SLs at 7 dpai, reflecting delayed systemic spread (Fig 4C, lanes 5–6, 15–16). We hasten to note that the K389E mutation was never isolated as a stand-alone mutation in earlier experiments. Rather, it was detected in one of the A113V-infected plants along with E392K and A113V (Fig 3). Interestingly, the K389E single mutant, but not the A113V-K389E double mutant, accumulated high levels of TCV RNA in 14 dpai SLs (Fig 4D, lanes 5–6, 15–16).

Sequence analysis of RT-PCR products revealed that by 14 dpai, in one of the K389E-infected plants the K389E mutation reverted to wt, and a new E392K mutation simultaneously emerged (Fig 4E, sample #5). In the second plant, the partially reverted K389E was accompanied by the same E392K mutation (Fig 4E, sample #6). Finally, the A113V-K389E double mutant replicated poorly in 14 dpai SLs and, in the sample subjected to sequence analysis (Fig 4E, sample #15), it likewise acquired the E392K mutation. Therefore, the K389E mutation by itself was likely severely deleterious. Conversely, the E392K mutation was likely enriched due to its ability to compensate for the defects of K389E, as well as that of A113V.

The E392K single mutant was even more deleterious than K389E (Fig 4B, compare lanes 7–8 with 5–6) in 4 dpai ILs, as mutant gRNA was barely detectable, and its sgRNA2 levels were lower still (Fig 4B, lanes 7 and 8). More strikingly, sgRNA1 was replaced by two new sgRNA species, migrating slightly slower and faster than wt sgRNA1, respectively (Fig 4B, lanes 7–8). For convenience, hereafter we refer to the slower-migrating species as sgRNA1a, the faster-migrating one sgRNA1b. Detailed characterization of sgRNA1a and 1b will be described later. Notably, the sgRNA1b band intensity was higher than sgRNA2, reflecting a perturbation of sgRNA relative abundances. Both sgRNA1a and 1b were also evident in ILs receiving KEEK and A113V-W753L mutants at 4 dpai (Fig 4B, lanes 9–10; 21–22), and they were consistently observed in at least three independent repeat experiments.

Contrary to K389E, the E392K defect was dramatically corrected in the A113V-E392K double mutant (Fig 4B, compare lanes 17–18 with 15–16). Specifically, viral RNAs of A113V-E392K double mutant accumulated to substantially higher levels than the E392K single mutant, and the sgRNA1a and 1b bands also disappeared (Fig 4B, compare lanes 7–8 and 17–18). Instead, the wt sgRNA1 band reappeared at a relative abundance similar to wt TCV. Accordingly, the A113V-E392K double mutant, but neither the A113V nor E392K single mutant, accumulated viral RNA levels in 7 dpai SLs to near wildtype levels (Fig 4C, compare lanes 17–18 with 7–8 and 13–14). Of interest, the E392K single mutant, though delayed, was able to reach 14 dpai SLs at relatively high levels, with the original mutation stably retained (Fig 4D and 4E, lanes 7–8). Likewise, the A113V-E392K mutant also retained both mutations for at least 14 days (Fig 4D and 4E, lanes 17–18). In short, the A113V-caused defects, both the loss of sgRNA1 and delay in systemic spread, were successfully rescued by the E392K mutation.

The KEEK double mutant accumulated viral RNAs to levels higher than K389E and E392K alone, but still lower than wt TCV, in 4 dpai ILs (Fig 4B, lanes 3–10). Intriguingly, KEEK retained the sgRNA pattern of E392K (sgRNA1a, 1b, sgRNA2) but not K389E (sgRNA2 only). Thus, it appears likely that K389E was primarily enriched in the E392K context to compensate for the E392K-caused replication inefficiency. Consistent with this interpretation, in 4 dpai ILs the A113V-KEEK triple mutant accumulated more gRNA and sgRNA2 than the A113V-E392K double mutant but modestly less sgRNA1 (Fig 4B, compare lanes 19–20 with 17–18). Perhaps due to the lower sgRNA1 levels, the A113V-KEEK triple mutant was less efficient at systemic spread, and gradually diminished in SLs (Fig 4C and 4D, lanes 19–20).

The W753L single mutant accumulated the same RNA pattern – gRNA, sgRNA1, and sgRNA2 – as the wt TCV in 4 dpai ILs. However, the sgRNA1 levels rose substantially to levels exceeding that of sgRNA2 (Fig 4B, lanes 11–12). Among the four single mutants (A113V, K389E, E392K, W753L), W753L was the only one detectable in 7 dpai SLs, albeit at lower levels than wt TCV. Notably, the relative over-abundance of sgRNA1 of W753L persisted in SLs until at least 14 dpai (Fig 4C and 4D, lanes 11–12). On the other hand, the A113V-W753L double mutant, contrary to A113V-E392K and A113V-KEEK, replicated much worse than A113V or W753L single mutant (Fig 4B, compare lanes 21–22 with 11–14). Curiously, the A113V-W753L also accumulated sgRNA1a and 1b, with their sizes and relative ratio resembling that of E392K and KEEK mutants (Fig 4B, compare lanes 21–22 with 7–10). Thus, loss of wt sgRNA1 and concurrent emergence of sgRNA1a and 1b appeared to be a shared response to certain types of structural disturbance, rather than a specific response to particular aa mutations.

Conversely, the W753L mutation in the A113V background was probably a temporary compensation for the loss of sgRNA1. This is because, unlike W753L single mutant, the A113V-W753L double mutant still exhibited delayed systemic infection (Fig 4C, lanes 21–22). Nevertheless, this mutant accumulated to wt levels in 14 dpai SLs (Fig 4D, lanes 21–22), and retained both mutations (Fig 4E, samples #21 & 22). In summary, W753L appears to compensate for the A113V defects by increasing the levels of sgRNA1 only.

sgRNA1a and sgRNA1b have different 5’ proximal sections, but nearly identical remaining sequences

We next wondered how various second-site mutations compensated for reduced sgRNA1 levels and alleviated the delay in viral systemic spread caused by the A113V mutation. Since three different sets of mutations – E392K, KEEK, and A113V-W753L – led to the emergence of two alternative sgRNA species – sgRNA1a and sgRNA1b – in lieu of wt sgRNA1, we sought to uncover possible compensation mechanisms by resolving the identities of these novel sgRNAs. We hasten to note that these alternative sgRNA species were detectable in ILs only, despite the persistence of some of the mutations in 14 dpai systemic leaves (Fig 4D and 4E, lanes 7–8, 9–10, 21–22). Nevertheless, their identities should still be informative as they probably represented immediate, though not necessarily lasting, natural selection response to the original A113V mutation.

Previous studies suggested that a TCV-borne noncoding RNA, designated ttsgRNA, produced circularized or dimeric replication intermediates during replication by forming covalent linkages between 5’ and 3’ termini of the same or two different ttsgR copies [25]. Our ongoing research further found that similar covalent linkages were also prevalent in (-)-strands of TCV sgRNAs, accounting for up to 30% of such (-) strands. We thus took advantage of these end-joined forms of sgRNAs for simultaneous mapping of the 5’ and 3’ ends of alternative sgRNAs, using a pair of divergent RT-PCR primers (TCV-3872F and -2741R. Fig 5C, orange arrowheads). Fig 5A shows the RT-PCR products of RNA samples extracted from ILs of wt TCV and two double mutants – KEEK and A113V-W753L. The 3872F and 2741R primer pair should generate RT-PCR products only when the (-) sgRNA templates were in circular or dimeric forms, yielding a 594-bp fragment from sgRNA1 (Fig 5B & 5C; the 183-bp dull blue fragment at the 3’ end fused to the 411-bp red fragment corresponding to the 5’ end of sgRNA1), and a 319-bp fragment for sgRNA2 (Fig 5B & 5C, the 183-bp dull blue fragment fused to the136-bp pink fragment).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Molecular characterization of alternative sgRNA1 (sg1) species.

A. RT-PCR products generated with a pair of primers that enrich circular or dimeric forms of sgRNAs, from three representative virus-infected plant RNA samples, infected with wt TCV, KEEV, and A113V-W753L, respectively. The tissues were ILs collected at 4 dpai. B. Summary of sequencing results, showing various unique clones and their respective numbers. See C and D for the composition of three representative classes of clones. C. Schematic depiction of expected RT-PCR fragments derived from wildtype sgRNA1 and 2 (sg1 and sg2). D. Assembly of sgRNA1a, 1a*, 1b, and 1c based on sequencing results.

https://doi.org/10.1371/journal.ppat.1013842.g005

Notably, the KEEK and A113V-W753L samples both yielded 3 products of similar sizes – approximately 750, 500, and 300 bp. Note the sizes of these RT-PCR products were consistent with them being derived from sgRNA1a, 1b, and sgRNA2 (Fig 4), with the 300-bp product migrating at the position of sgRNA2-specific product in the wt TCV lane (Fig 5A, lane 2–4; and Fig 5C). We then cloned these products into a plasmid vector and sequenced multiple clones to resolve the identities of these fragments. As expected, most clones derived from wt TCV control samples contained cDNA fragments with 5’ and 3’ ends matching authentic TCV sgRNA1 and 2 (6 each out of a total of 15 clones. Fig 5B).

Surprisingly, the similar-sized products derived from KEEK and A113V-W753L samples shared nearly identical sequences. Several classes of clones have been recovered. The first class consisted of 8 clones – 3 from the KEEK sample and 5 from the A113V-W753L sample, harboring nearly identical inserts of 745–749 bp (Fig 5B). Four of the 8 inserts were 747 bp, and the minor size variations in remaining 4 clones were due to differences at the 3’-5’ junctions. Excluding the183–186 bp originating from the TCV 3’ terminus (position #3872–4054 plus a few extra C or Gs), the remaining 563 bp was made up of three partially repetitive, discontinuous TCV sections (Fig 5D, top diagram): a 303-bp 5’ proximal section (Fig 5D, orange line) that began at the authentic sgRNA1 start site (position 2331) and ended at position 2633 of TCV gRNA; followed by a 35-bp section spanning positions 2390–2424, thus internal to the first section (Fig 5D, light blue lines); as well as a 225-bp 3’ proximal section spanning positions 2517–2741 (Fig 5D, gray lines). Assuming no additional changes downstream of position 2741, these chimeric inserts would have been derived from an sgRNA1a of 1,876 nt, surpassing wt sgRNA1 (1,724 nt) by 152 nt.

The second class consisted of 3 clones – 2 from the KEEK sample and 1 from the A113V-W753L sample – had inserts with a shorter, 38-bp 3’ terminal section (positions 3872–3909), but the same 5’ 563 bp as the first class of clones. The corresponding alternative sgRNA would be of same size as sgRNA1a, hence was named as sgRNA1a* (Fig 5B and 5C). Collectively 5 of the 26 KEEK clones, and 6 of 25 A113V-W753L clones, arose from the nearly identical, 1,876-nt sgRNA1a. Finally, it is worth reiterating that the 5’ termini of sgRNA1a and 1a* were identical to that of wildtype sgRNA1.

The third class of clones were recovered 13 times – 6 from KEEK and 7 from A113V-W753L samples (Fig 5B and, sg1b diagram). Their inserts ranged from 470-475 bp, with the minor size differences again mapping to the 3’-5’ junctions. Excluding the 183-to-188-bp section originating from TCV 3’ end, the remaining 287 bp also comprised three discontinuous sections. The 28-bp 5’ proximal section (Fig 5D, light purple line) initiated at the authentic sgRNA2 start site (position 2606) and, intriguingly, ended at the same 2633 position as sgRNA1a and sgRNA1a* (Fig 5D). Moreover, the next 35-bp section (Fig 5D, light blue line) and the last 225-bp section (Fig 5D, grey line) were likewise shared by sgRNA1a/1a*-derived cDNA. These inserts would predict an alternative sgRNA1b of 1,601 nt, 123-nt shorter than wt sgRNA1, but 152-nt longer than wt sgRNA2 (Fig 5D).

The fourth class of clones were recovered twice, from the KEEK sample only. Their insert of 623-bp was particularly interesting (Fig 5D, sg1c diagram). Excluding the 183-bp section derived from TCV 3’ terminus, the remaining 440 bp represented reiterated RNA recombination events at the same hotspots. It contained the same two 5’ sections as sgRNA1b. But the next, 117-bp section represented yet another pause at position 2633, and a second recruitment of the 35-bp spanning positions 2390–2424. These additional recombination events gave rise to the alternative sgRNA1c that was 1,753 nt in size, 29-nt longer than wt sgRNA1. Taken together, these results resolved the identities of sgRNA1a and 1b that emerged in cells infected by two distinct double mutants – KEEK and A113V-W753L.

The deduced sequence of sgRNA1a could readily template the translation of wt p8 (S1A Fig). Furthermore, despite the 35-nt insertion (light blue lines in Fig 5D) and partial duplication of the p9 coding sequence, it could support undisrupted translation that began with an AUG internal to original p9 coding sequence (but in a different reading frame. S1A Fig), read through the 35-nt insertion, and then continue into the N-terminally truncated p9 coding sequence downstream. The resulting p9 variant replaced the N-terminal 7 aa of wt p9 with a 29-aa novel N-terminus. Therefore, sgRNA1a could potentially rescue the loss of sgRNA1 by templating the translation of a wt p8 and an N-terminally altered p9. By contrast, sgRNA1b (and sgRNA1c) was unlikely to have rescued the loss of sgRNA1 (S1B Fig). This is because it lacked most of the p8 coding sequence. Even though a putative coding frame for an N-terminally truncated and mutated p9 variant could be identified (S1B Fig), that frame did not contain any AUG start codon, hence was unlikely to translate this p9 variant.

Recombination events giving rise to alternative sgRNA1a and 1b likely have occurred during (-)-strand synthesis

We showed in Fig 5 that sgRNA1a, 1a*, and 1b shared the same two recombination junctions that fused three discontinuous sections of TCV genome. Their size difference can be straightforwardly attributed to their different 5’ termini: sgRNA1a and 1a* possessed the 5’ terminus of wt sgRNA1, whereas sgRNA1b possessed that of wt sgRNA2. These characteristics are in turn best explained if recombination took place during (-)-strand synthesis [27–29]. As schematically depicted in Fig 6, mutations like E392K, KEEK, and A113V-W753L probably caused (-) strand synthesis to pause more frequently at certain sites of TCV genome. One such site is position 2517, where strong base-pairing between GGGGG in the template and the newly copied CCCCC may stall RdRp (Fig 6, Step 1). The fragment released, possibly after adding an extra, non-templated C, would then re-attach to position 2424 where its 3’-terminal CCCCCC could anneal to GGCGG with one mismatch (Fig 6, Step 2). After a short, 35-nt extension, the (-) strands apparently encountered a second site of frequent pauses, releasing a product with a 12-nt terminal motif (5’-AGAGAGUUGUAG-3’) (Fig 6, Step 3). Incidentally, this 12-nt motif was 100% complementary to another template location near position 2633, permitting the released (-) strands to re-anneal at this downstream position, followed by continued copying to the authentic termination signals of wt sgRNA1 and 2, thereby completing the synthesis of (-)-strand intermediates of sgRNA1a and 1b, respectively (Fig 6, Step 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. A model for the genesis of alternative sgRNA1 species.

https://doi.org/10.1371/journal.ppat.1013842.g006

How might the A113V, K389E, E392K, W753L mutations, or combinations of them, alter the pattern of (-) sgRNA synthesis?

Replication of TCV gRNA, as well as transcription of its sgRNAs, are known to take place inside membranous spherules formed by invagination of mitochondrial outer membrane [30,31]. These replication spherules morphologically resemble those described for brome mosaic virus (BMV) and flock house virus (FHV), which have been more thoroughly studied [32–37]. Such spherules are typically induced by virus-encoded RPs, such as p28 and p88 in TCV, 1a and 2a in BMV, and Protein A in FHV, possibly along with viral RNA. Recent cryo-EM studies of FHV replication complexes revealed a crown-like structure in which 24 copies of protein A organize into two stacks of 12-mer rings, forming the neck channel of spherules [34,35]. It has also been speculated that the interiors of spherules are lined by single-layer lattices of RPs, such as BMV 2a [32]. Importantly, spherules are known to vary in size depending on the size of viral RNAs being replicated (e.g., FHV RNA1 or RNA2) [35]. Thus, different-sized viral RNAs probably recruit different numbers of RP copies, or different ratios of two RPs, to from different-sized spherules. Indeed, we found earlier that in TCV infections, overexpressing p88 relative to p28 increased the accumulation of sgRNA1, 2, and ttsgR, but decreased the level of TCV gRNA, suggesting a reduction of large-sized spherules capable of supporting gRNA replication [25,38].

The exact p28:p88 ratios in spherules replicating/transcribing TCV gRNA, sgRNA1, sgRNA2, and ttsgR are yet unknown. However, it is not unreasonable to assume that these ratios differ for spherules housing these different-sized RNAs, possibly entailing different numbers of p28 copies to pair with a set number (monomer or dimer) of p88. As a result, the A113, K389, E392, and W753 mutations could have altered intra-spherule interactions of p28 and p88 in subtle ways, thereby slightly distorting the size of spherules suited for transcribing sgRNA1, allowing them to stabilize and/or transcribe sgRNA1a and 1b variants.

In addition, compensation conferred by the W753L mutation might not be solely due to the W-to-L aa change – the underlying G2321U nucleotide change might have contributed as well. As noted earlier, viruses like TCV synthesize (-) sgRNA intermediates via a premature termination mechanism that relies on RNA secondary structures in the (+)-stranded templates to stall (-) strand elongation, thereby defining the 3’ termini of these (-) strands. It is thus reasonable to assume that stronger RNA secondary structures cause more frequent stalling of (-) strands, leading to more (-) strands that template more (+) strands. Incidentally, the single G2321U nt mutation underlying W753L also converted a G-G mismatch to a G:U base pair in the RNA secondary structure responsible for generating sgRNA1 (-) strands (S2 Fig) [39], causing modest stabilization of this structure. Such structural stabilization could conceivably lead to more sgRNA1 (-) strands.

Why were sgRNA1a and 1b absent in systemic leaves of plants infected with E392K, KEEK, and A113V-W753L?

Interestingly, the triple-sgRNA pattern observed in ILs of E392K, KEEK, and A113V-W753L infections was not detected in SLs of the same plants, despite the apparent stability of some of the mutations (Fig 4E). We also failed to detect any compensatory mutations outside the RP coding region of TCV genome despite additional sequencing efforts. To account for this discrepancy, it should be noted that sgRNA1a and 1b produced in ILs by the mutant viruses could not have served as efficient mRNA for the p9 MP (S1 Fig in Supplementary Data). Lack of p9 MP would in turn render mutant viruses incapable of cell-to-cell movement, restricting them in single cells. We hence reason that, due to cell-level heterogeneity [40], wt sgRNA1 could have been transcribed in a small fraction of IL cells, producing low levels of wt sgRNA1 that evade Northern blot detection. Nonetheless, only these few cells could produce enough p9 and be expected to export mutant viruses to neighboring cells to commence cell-to-cell movement. Upon receiving the mutant viruses, a similarly small fraction of new cells would produce p9 MP and facilitate secondary cell-to-cell movement. Reiteration of such low efficiency spread could explain the delayed systemic spread of E392K, KEEK, and A113V-W753L mutants, and may also explain why in SLs wt sgRNA1 became more abundant. Additionally, absent of compensatory mutations, a fraction of mutant viruses might have acquired epigenetic modifications in RPs and/or viral RNAs that favor the transcription of wt sgRNA1 by mutant viruses.

Finally, we wish to reiterate that the rescue afforded by compensatory mutations such as E392K and W753L was by no means perfect. Rather, they provide transient relief for the A113V defects. This then allows their descendants to replicate to higher copy numbers in the subsequently invaded cells than sister lineages not containing the compensatory mutations. Such temporary solutions need not to be perfect – they simply need to be comparatively more fit then and there, permitting emergence of revertant or additional new mutants with superior compensatory capacities. On the other hand, embarking on the trajectory of harboring both original (A113V) and compensatory (e.g., E392K) mutations also means that the new mutant is now more difficult to revert, thus possibly trapping it in a suboptimal state for some lengths of time. To summarize, our current study identified several naturally occurring mutations that specifically affect one of the viral sgRNAs, and examined the natural selection remedies that compensated for their defects. These findings are expected to inform strategies that target viral sgRNA synthesis for controlling viral diseases of plants, animals, and humans.

Data sharing plan.

All data collected throughout the current study have been provided in the manuscript text and Supplemental Information (SI) Appendix. Readers are encouraged to contact us with any questions.