Dynamic Changes Impact the Plum Pox Virus Population Structure During Leaf and Bud Development

Abstract

Plum pox virus (PPV) is a worldwide threat to stone fruit production. Its woody perennial hosts provide a dynamic environment for virus evolution over multiple growing seasons. To investigate the impact seasonal host development plays in PPV population structure, next generation sequencing of ribosome associated viral genomes, termed translatome, was used to assess PPV variants derived from phloem or whole leaf tissues over a range of plum leaf and bud developmental stages. Results show that translatome PPV variants occur at proportionately higher levels in bud and newly developing leaf tissues that have low infection levels while more mature tissues with high infection levels display proportionately lower numbers of viral variants. Additional variant analysis identified distinct groups based on population frequency as well as sets of phloem and whole tissue specific variants. Combined, these results indicate PPV population dynamics are impacted by the tissue type and developmental stage of their host.

1. Introduction

The emergence of new viral strains is an important yet poorly understood aspect of virus evolution. New strain appearances are generally associated with genetic mutations or recombination events that confer enhanced virus infection, spread, or adaptation to new hosts, vectors, and environments (Duffy et al., 2008; Geoghegan and Holmes, 2018; Mustroph et al., 2009; Sanjuán and Domingo-Calap, 2016). For RNA viruses, the evolution of new variants is often linked to several factors that include the error-prone nature of the viral RNA polymerase, recombination, high generation rates, and large population sizes that can rapidly produce an assortment of genome variants or quasi-species upon infection (Dolja and Koonin, 2018; Duffy et al., 2008; Geoghegan and Holmes, 2018; Sanjuán and Domingo-Calap, 2016). For plant RNA viruses, especially those viruses that infect woody perennials, the impact of these factors is potentially multiplied over numerous growing seasons resulting in a continually evolving reservoir of viral variants having unique infection and disease-causing capabilities. How these genetic variants arise and are maintained within a population is not well understood and is of critical importance to developing a more comprehensive understanding regarding the evolution and emergence of new virus isolates.

Plum pox virus (PPV), a member of the Potyviridae family, is a worldwide threat to stone fruit production, causing economic losses in the hundreds of millions of dollars (Cambra et al., 2006). PPV has a ~10 kb long positive sense RNA genome that generates a 350 kDa polyprotein encoding ten distinct proteins and one frameshift protein (García et al., 2014; Sochor et al., 2012). PPV is capable of infecting a range of woody perennial hosts including plum, peach, almonds, cherry and other stone fruits and causes symptoms that include mottled misshapen leaves, reduced fruit quality and premature fruit drop; though some strain-host combinations remain asymptomatic (Levy et al., 2000; Milošević et al., 2019; Usenik and Marn, 2017). There are ten described PPV strains that share between 70 to 97% sequence homology (Chirkov et al., 2017; Hajizadeh et al., 2019; Sihelská et al., 2017). These strains appear to have arisen by a combination of coevolution with their hosts and recombination events between co-infecting isolates and strains (Chirkov et al., 2017; García et al., 2014; Hajizadeh et al., 2019; James et al., 2015; Sihelská et al., 2017). In addition, disease symptoms as well as host and vector transmission rates vary widely depending on the viral strains present (Subr and Glasa, 2008). Thus, the maintenance of PPV within its perennial hosts has provided an ideal environment for the mixing and development of new genome variants with potentially unique disease-causing abilities.

The perennial nature of stone fruit has provided a unique system for the study of intravirus populations within individual trees over extended periods of time. Jridi et al. (Jridi et al., 2006) extensively sampled a single PPV infected peach tree that had been maintained under vector free conditions for 13 years. Results from single strand conformation polymorphism analysis of 333 samples derived from leaves, branch, trunk bark and roots identified 33 distinct PPV haplotypes. The distributions of these haplotypes were spatially unique, suggesting that as new variants emerge, they become independently established within developing tissues. In contrast, Predajna et al. (Predajňa et al., 2012) sampled a PPV infected plum tree seven years post infection with three PPV isolates. Total RNA samples from leaves and fruit produced 105 PCR derived PPV sequences. Sequence analysis revealed 51 different PPV-M haplotypes. Interestingly, Predajna et al. (Predajňa et al., 2012) did not observe the spatial distribution of these haplotypes within the infected tree as found by Jridi et al. (Jridi et al., 2006) possibly due to vector derived mixing as aphids were not controlled in this study.

While previous studies provided snapshots into the composition of PPV populations, their dynamics during cycles of growth and dormancy have not been investigated. Further, plants have evolved diverse organs and tissues to carry out specific functional tasks. For example, within leaves there are lamina associated cells that function in photosynthesis and vascular cells that conduct water and nutrients. These tissues can be further divided into subsets of cell types each with their own contributions to plant structure and physiology. Due to these differences, each cell type represents a potentially unique environment that could affect the makeup of the virus quasispecies. Unfortunately, it is technically challenging to separate these tissues from each other, making tissue specific investigations into the dynamics of virus populations difficult. In this study we profiled PPV populations in both phloem, whole leaf and bud tissues using translating ribosome affinity purification (TRAP) (Reynoso et al., 2015). In the TRAP approach, transgenic or transient introduction of an immuno-tagged ribosomal protein (RPL18) under control of a tissue specific promoter allows for the purification and subsequent profiling of ribosome-associated mRNAs (termed translatome) directly from whole tissue samples. An advantage of this approach is that ribosome associated viral RNAs are likely to be undergoing translation and can be separated from inactive or packaged RNAs. Here we used previously developed plum trees expressing a tagged RPL18 under either the phloem-specific promoter pSUC2 or the ubiquitous cauliflower mosaic virus 35S promoter (Collum et al., 2019; Collum and Culver, 2017). Translatome RNAs were sampled from leaves at 2, 4, 6, and 12 weeks after cold induced dormancy as well as pre- and post-chilling vegetative buds to evaluate PPV population dynamics in whole tissues and phloem of individual plum trees.

2. Results

2.1. Characterization of plant lines and the translatome system.

The translatome system used in this study including the PPV infected leaf translatome data sets have been previously described (Collum et al., 2020, 2019; Collum and Culver, 2017). This system utilizes tissue specific expression of hexa-histidine and FLAG-tagged ribosomal protein 18 (HF-RPL18) for the immunocapture of ribosomes and their associated RNAs. The A. thaliana HF-RPL18 protein used in this study shares 87% amino-acid identity and 95% similarity with the Prunus RPL18 and associates with the polysome fraction when expressed as a transgene in plums (Collum et al., 2019). As a result, high-quality mRNA from all plant lines expressing HF-RPL18 but not from non-transgenic controls is recoverable from the polysome fraction. In addition, the A. thaliana phloem-specific Sucrose 2 promoter (pSUC2) and the ubiquitous 35S promoter were used to drive expression of the HF-RPL18 construct. Previous studies have shown that the pAtSUC2 promoter successfully confers phloem expression in a range of plants including sweet oranges, strawberries, and most recently plums while the 35S promoter expresses in all tissues including phloem (Collum et al., 2020; Miyata et al., 2017; Mustroph et al., 2009; Zhao et al., 2004). For this study, PPV infected transgenic trees expressing the HF-RPL18 construct from either three pSUC2 trees or three p35S trees were used to generate translatome RNA at 2, 4, 6 and 12 weeks post cold induced dormancy and from buds just prior to and after chilling for subsequent next generation sequencing (NGS) analysis (Fig. 1). Each tree was treated as an independent biological replicate with leaf tissue surveyed over two growth periods and buds analyzed from one chilling event. It should be noted that expression of the HF-RPL18 transgene was previously found not to be altered in response to PPV infection (Collum et al., 2020).

Figure 1.

Representative photographs of plum pox virus infected leaves collected at 2, 4, 6 and 12 weeks as well as buds used for translatome analysis. Bar represents 1 cm.

2.2. Temporal and spatial accumulation of the PPV translatome in phloem and whole leaf tissue.

The number of PPV mapped reads per sample ranged from a low of 588 for buds to 3,069,012 for six-week old leaves with an average of 802,673 PPV specific reads per sample (Table S1). To compare PPV levels within each translatome sample, total PPV reads for each tree and time point were normalized to the total translatome reads averaged across all time points. For p35S whole leaf tissues, PPV levels were similar for the two monitored growth periods, rising rapidly between two and four weeks and peaking at six weeks and then falling slightly in 12-week old leaves with the lowest levels observed in bud tissues (Fig. 2A & B). PPV levels within the pSUC2 phloem tissues were more variable, peaking at four weeks in the first growth periods and at six weeks in the second period. However, within 12-week old leaves levels of PPV in the phloem translatome were significantly reduced compared to the whole leaf translatome (t-Test: p=0.05) in both growth periods (Fig. 2A).

Figure 2.

Plum pox virus (PPV) levels in plum whole-leaf and phloem translatomes. (A) PPV translatome associated leaf and bud RNA collected over two growth periods. (B) Expanded view of bud translatome RNA collected from one chilling period. Bars represent the mean of three independent trees ± standard error, except for the pSUC2 two week time point for growth period two, which represents the mean of two biological replicates.

We also investigated the levels of PPV RNA participating in translation as ribosome associated PPV RNAs are likely to represent only a fraction of the total viral transcripts present within a sample, since those packaged as virions or not associated with ribosomes would be largely excluded. To assess the proportion of total PPV reads within the translatome we calculated the ratio of PPV RNA associated with the translatome-purified ribosomes versus the PPV RNA present in the same sample prior to ribosome extraction. This was done for the three p35S::RPL18 trees at both two and six weeks post bud-break. NGS results for these samples were then compared for the number of PPV reads from total RNA versus translatome RNA isolations. Results indicate that on average only ~0.2% of the PPV RNA in these samples is associated FLAG-tagged RPL18 (Table S1). This fraction of ribosome associated PPV RNA would represent only those virus RNAs present on ribosomes containing the FLAG-tagged RPL18 and not those associated with native non-tagged ribosomes. Thus, few viral genomes participate in protein production.

2.3. PPV variant levels display dynamic changes throughout the growth cycle.

For variant identification, regions of the PPV genome with at least 10-fold read coverage were used for variant calling. Furthermore, only nucleotide variants that occurred twice or more within a sample were selected for study. Based on these criteria we identified a total of 2747 unique variants from all samples and time points (Table 1). Of these variants only 283 were shared between pSUC2 phloem and p35S whole tissue translatomes with 1290 being specific to the phloem and 1174 specific to whole tissue (Table S2). The total set of 2747 variants spanned 25% of the PPV genome with the majority, 65%, being nonsynonymous and impacting PPV protein coding. Within individual samples, the number of unique variants observed ranged from 0 to 286 (Table S2). To examine the dynamics of PPV variants over time we determined the percent of PPV reads that contained a variant for each sampling time. The PPV-D Penn 7 genome was used as a reference genome for this study (Schneider et al., 2009). However, variants with frequencies >95%, listed as group 1 in Table 1, were considered part of the consensus sequence for this isolate and excluded from this analysis. For these comparisons, the number of variant PPV reads were normalized to the average PPV translatome reads. Interestingly, we observed an inverse relationship between PPV translatome levels and the number of variants (Fig. 3). Specifically, as a proportion of the total PPV reads both two-week old leaves and bud samples contained higher percentages of virus variants than observed in four and six-week-old leaves where virus levels are at their highest (Fig. 3A & B). This inverse relationship between variant and virus levels is most significant between bud tissues and week 4 and 6 leaf tissue (t-Test: p=0.002). In addition, the number of variants within buds is likely to be greater than reported here as the low virus levels within these samples produced low PPV genome coverage levels, often less than the 10-fold cutoff used in this study.

Table 1.

Tissue distribution of translatome identified PPV variants within four frequency groups.

Maintenance behavior1	Shared, Found in Both pSUC2 and p35S		Phloem (pSUC2) Only		Whole Tissue (p35S) Only
		Breakdown of variants present in the six tested trees2		Breakdown of variants present in the three SUC2 trees		Breakdown of variants present in the three p35S trees
Group 1 Variants with >95% frequency in all trees	15	15 in 6/6	0	0 in 3/3	0	0 in 3/3
		0 in 3-5/6		0 in 2/3		0 in 2/3
		0 in 2/6		0 in 1/3		0 in 1/3
Group 2 Variants that vary frequency between samples	28	10 in 6/6	14	1 in 3/3	2	0 in 3/3
		16 3-5/6		5 in 2/3		0 in 2/3
		2 in 2/6		8 in 1/3		2 in 1/3
Group 3 Variants with consistently low frequencies	61	9 in 6/6	21	3 in 3/3	12	1 in 3/3
		35 in 3-5/6		12 in 2/3		1 in 2/3
		17 in 2/6		6 in 1/3		10 in 1/3
Group 4 Variants that appear nonconsecutively in one or two samples	179	0 in 6/6	1255	0 in 3/3	1160	0 in 3/3
		26 in 3-5/6		57 in 2/3		35 in 2/3
		153 in 2/6		1198 in 1/3		1125 in 1/3

¹Classificaiton of variants based on frequency and presence within individual trees.

²Number of times a variant was identified in and individual tree. For example, within the shared variants of Group 1, 15 out of 15 variants were found in six of six tested trees.

Figure 3.

Plum pox virus (PPV) variant diversity in plum whole-leaf and phloem translatome RNA. Variants include nucleotide polymorphisms, indels and replacements with SNPs making up the majority, (87%). Diversity of PPV populations as a percent of normalized PPV reads containing variants in collected plum tissues over two growth periods. Bars represent the mean of three biological replicates ± standard error, except for the pSUC2 two week time point for growth period two, which is the mean of two biological replicates.

To determine if the inverse relationship between virus levels and variants is specific to the translatome we also compared percent variant levels in total RNA samples. This was done for two- and six- week old leaves. Total RNA was isolated from a portion of homogenized samples prior to ribosome isolation and submitted for NGS followed by RNAseq analysis to identify PPV variants. This was done for all three p35S::RPL18 trees at two and six weeks post bud-break. Results indicate that at both sampling times, variant levels were similar for both the 35S translatome and total RNA samples (Fig. 4). However, the level of variants at two weeks was ~30 fold greater than observed at six weeks for both translatome and total RNA samples (Fig. 4). Thus, the inverse relationship observed between virus and variant levels is observed in both translatome and total RNA samples.

Figure 4.

Plum pox virus (PPV) variant levels in total vs translatome purified RNA. Total RNA was isolated from the same sample prior to translatome purification. Samples were taken from three independent PPV infected trees at 2 and 6 weeks post chilling. Bars represent the mean of normalized PPV reads from three biological replicates ± standard error.

2.4. Contributions of individual variants to the PPV population.

To better understand the contributions of specific variants to the PPV population, we determined the average frequency of individual variants at each sample time point. Results indicated that variants could be grouped into four general categories (Tables 1, S3 - S6). Group 1 included 15 variants that were present in at a frequency of >95% for all eight leaf samples and a majority of bud samples (Fig. 5A, Tables 1 & S3). Interestingly, some group 1 variants were not detectable in some individual bud samples but subsequently reappeared upon leaf development. The inability to detect these variants in buds is likely associated with the low virus levels that occur in these tissues, making detection difficult. Group 2 included 44 variants that were highly variable in their frequency from one sample point to the next but were still present at 3 to 8 sample times (Fig. 5B, Tables 1 & S4). Group 3 was comprised of 94 variants that were maintained in at least 3 of the 10 samples and had low average frequency, varying between 0.5% and 20% (Fig. 5C, Tables 1 & S5). Finally, group 4 represented 2592 unique variants that were found sporadically at no more than two sampling times with 2296 of these variants being unique to only one time point (Fig. 5D, Tables 1 &S6). The majority of group 4 variants appeared at low frequency (Fig. 5D). However, seven of these time-point specific variants yielded frequencies above 50%, suggesting they are efficient in replication and local movement (Table S7).

Figure 5.

Maintenance of individual plum pox virus (PPV) variants over sampling times. The percent frequency of individual variants at each sample time is shown for the four identified variant groups. A, Group 1 variants appearing at high frequencies at most time points. B, Group 2 variants displaying variable frequencies. C, Group 3, variants consistently showing frequencies <50%. D, Group 4 variants showing no consistent maintenance pattern. Graphs represent variants identified in one PPV infected translatome tree.

Comparisons of phloem versus whole tissue translatomes did not show differences in the makeup of group 1 variants, as all 15 variants that make up this group were found in both the phloem and whole tissue translatome samples (Tables 1 & S3). Among 44 group 2 variants, 14 were phloem specific and two were found only in whole leaf tissues leaving 28 shared between translatomes. Of the group 3 variants 21 were specific to the phloem and 12 were whole leaf tissue specific. Variants belonging to group 4 comprised 179 shared variants with 1255 and 1160 found only in the phloem or whole tissue translatomes, respectively (Tables 1 & S6).

2.5. Impact of translatome variants on the PPV genome.

To assess the distribution of variants and variant groups across the genome, the nucleotide positions of the viral variants were mapped to the PPV genome. Results show that the identified variants are dispersed across the PPV genome at 2443 of the 9786 PPV nucleotides. A majority of these nucleotide positions are impacted by only one variant. However, we did identify 281 nucleotide positions that contained more than one unique variant. Of these PPV nucleotide positions with multiple variants, 94% exhibited two different variants, 5% exhibited three and the remainder displayed between four and seven different variants.

To investigate the relationship between variant frequency and genome position, individual nucleotide variants were mapped onto the PPV genome based on the four frequency groups outlined in Tables 1 & S3 - S6. Group 1 high frequency variants are all single nucleotide transitions that affect six of the 10 PPV cistrons (leaving HC-pro, 6K1, 6K2, and NIa-VPG unaffected). Three of the group 1 variants, as mentioned earlier are nonsynonymous, affecting the P3 protein at two positions and the coat protein at one (Fig. 6A). Interestingly, all three of the high frequency nonsynonymous variants result in the substitution of a glycine or proline residue with a serine amino acid. Group 2 variants with varying frequencies affect 9 of the 10 PPV cistrons except for NIa-Pro (Fig. 6B). Fifteen of the group 2 variants are nonsynonymous, affecting P1, HC-Pro, P3, 6K1, CI, 6K2, NIb, and CP (Table S4). Low frequency group 3 variants affect nine polyprotein products except 6K2. Of the group 3 variants, 37 % are nonsynonymous, affecting 9 protein cistrons (Fig 6C; Table S5). The 2594 group 4 variants impacted all PPV cistrons with 67% being nonsynonymous (Fig. 6D).

Figure 6.

Genomic positions of identified plum pox virus (PPV) translatome variants. Percent frequency of PPV variants identified as (A) group 1, (B) group 2, (C) group 3 and (D) group 4. PPV genomic nucleotide position of each variant is indicated by the position of the vertical bar. Bar height indicates variant frequency. Individual PPV coding domains are color coded.

In total, 2747 variants, including single and multiple nucleotide polymorphic variations, insertions, deletions, and replacements, were found with frequencies ranging from 0.5 to 100% (Table S8). The majority of these variants, 86%, are single nucleotide variations (SNV). Of the 2346 single nucleotide polymorphisms found in groups 1 - 4, we found transitions (1360 variants) were more common than transversions (986 variants). Although, G>T transversions made up more than 50 percent of the transversions, outnumbering any of the individual transitions. Multiple nucleotide polymorphisms and replacements were also detected within this dataset. However, they all fall into the low frequency group 4 variants. Finally, 51 insertions and 313 deletions were also detected, primarily in the low frequency single sample group 4.

3. Discussion

Results demonstrate that dynamic changes within the PPV population occur during leaf and bud development. Pre- and post-chill bud tissues were found to have relatively low levels of virus yet contained proportionately higher numbers of PPV sequence variants than found in the more highly infected leaf tissues at later stages of development. This suggests that a proportionally larger pool of viral variants is maintained through periods of bud dormancy while only a few select variants become dominant during leaf development. Furthermore, in older leaf tissues, virus levels drop significantly within the phloem translatome but not within other leaf tissues, suggesting that with age phloem tissue becomes less conducive to viral translation, perhaps due to the higher activation of RNAi associated host defenses within developing leaf phloem (Collum et al., 2020). Analysis of individual variants identified four categories based on frequency and maintenance between sampling times. In addition, comparisons of whole leaf and phloem specific translatomes revealed sets of variants unique to each tissue. The impact of these variants on the dynamics and structure of the PPV population is discussed.

In this study, the profiling of ribosome isolated virus RNAs provided a unique snapshot of the viral genomes that are likely involved in translation and thus actively engaged in the infection process. We observed that less than 0.02% of the total viral genomes were associated with FLAG-tagged ribosomes and thus directly contributing to the infection process. Virus translatome levels reach their maxima within developing four and six-week old leaves and then declined as the leaves matured to 12 weeks. This corresponded with the appearance of virus leaf symptoms, which were most evident in the developing leaves (Fig. 1). Both buds and two-week old leaves contained reduced virus translatome titers indicating lower levels of active virus infection in these tissues. The most notable tissue specific variation in virus levels occurred in older 12-week leaves where the PPV associated phloem translatome was significantly reduced in comparison to the whole tissue translatome over both growth periods (Fig. 2). Interestingly, the induction of plant defense responses within the phloem including RNAi associated viral defense genes RDR1, DCL2 and HEN2 corresponded with the accumulation of phloem associated PPV translatome RNA, with these defense genes being significantly up regulated in four and six week old infected leaves followed by reduced expression levels in 12 week old leaves (Collum et al., 2020). This finding suggests that although initially ineffective, host defense responses are induced during leaf development, particularly in the phloem, and thus may be responsible for reducing phloem virus levels late in infection. Combined, these findings are consistent with studies that have shown that in perennial hosts such as plum, PPV infection typically displays high levels of virus replication and more severe disease symptoms in newly developing tissues (Levy et al., 2000; Marini et al., 2015).

In woody perennials, the passage of virus from one growing season to the next involves maintenance of the virus population within overwintering tissues. For PPV, leaf buds represent one possible tissue for virus maintenance. Here, we determined that tissues derived from both apical and axillary buds just prior to and after chilling contain low levels of PPV translatome RNA, <1% of the average observed in leaves. Despite these low virus levels, the number of observed variant-containing reads in these samples, measured as a percent of the total PPV translatome, was greater in diversity within bud tissue and to lesser extent within newly emerging two week old leaf tissues than in any of the other leaf samples (Fig. 3). We also found that anti-viral RNAi associated genes, previously identified as upregulated in leaf tissue in response to infection, are not differentially regulated in either pre- or post-chilled PPV infected buds (Collum et al., 2020) (Table S9). Low levels of virus replication and the lack of anti-viral host defenses within infected buds could remove key selective pressures on the PPV genome imposed by RNAi defenses and/or super-infection exclusion (SIE). SIE is a process whereby the accumulation of viral products are thought to block infection by superinfecting or progeny virus genomes from subsequent rounds of virus replication and thus acts to prevent the buildup of deleterious mutations (Folimonova, 2012; Zhang et al., 2017). Low PPV translatome levels in these tissues likely limit the production of viral proteins associated with SIE and limit its effectiveness in repressing mutant virus accumulation or maintenance. Based on the findings presented here it is likely that buds provide a unique environment for the maintenance of variant populations from one growing season to the next.

While buds may provide a permissive cellular environment for the development of variant populations, it is likely that upon bud break, increases in virus replication and translation along with the activation of host defense processes leads to the accumulation of specific high frequency variants. As frequency can be considered an estimate of fitness we assume the high frequency variants that emerge after bud break represent the most fit viral genomes (García- Arenal et al., 2001). These genomes primarily encompass the 15 group 1 variants that display a >95% frequency and makeup the consensus sequence for this isolate of PPV-D, Penn7. Of these 15 variants only three were nonsynonymous, yielding amino acid substitutions in P3 and CP proteins. Interestingly, these three substitutions all resulted in proline or glycine to serine conversions. Serine’s small side chain and the ability of its side-chain hydroxyl to hydrogen bond with the protein backbone make it an effective mimic for proline and glycine residues, suggesting that these substitutions may only subtly impact protein structure (Barnes and Gray, 2003). It is important to note that the short-read technology used here prevented us from evaluating whether these 15 variants occurred in linkage or not. Use of long read technology in the future may enable such an assessment. Further, we do not have the genetic makeup of the Penn7 isolate from P. domestica, which was used as inoculum in these studies. However, the high frequency and presence of these variants in all trees suggests they were likely present in the inoculum.

Variants that displayed variable changes in frequency over multiple time points represented a second category, group 2. This group includes variants that while identified in a significant number of the sampled time points, (three to eight), displayed highly variable frequencies from one sample time to the next (Table 1 and Fig. 6B). This variability suggests that environmental conditions such as leaf developmental age or competition from other variants differentially impacts their ability to consistently establish themselves across the range of sample times. Similarly, variants in group 3 are present at only low levels in most tested trees and samples, making up less than half of the variant population in a given sample (Table 1; Fig. 6C). It is likely that these minor variants while being maintained do not compete efficiently within the virus population, thus reducing their frequency. The majority of identified PPV variants are found in group 4, primarily at only one sample time and in only one or two trees. Such variants likely represent localized expansions of specific mutations that are not systemically viable components of the virus population and thus not maintained. Alternatively, the error rate associated with NGS and the selection criteria used in this study requiring reads for 10 fold coverage of a genome position with the mutation occurring on at least two of the reads could have limited our ability to consistently identify rare variants that persist from sample to sample at extremely low frequencies (Pfeiffer et al., 2018; Whitfield and Andino, 2016).

Within the PPV genome we observed 1780 nonsynonymous variants, 874 synonymous and 93 variants in the untranslated region. The combined distribution of these variants across the PPV genome revealed regions of distinct complexity. For nonsynonymous changes, the higher frequency variants found in groups 1 and 2 included groups proximally located in the N termini of CP and the C terminus of P3 (Table S3 and S4). Variability within these cistron regions has previously been noted (Carbonell et al., 2013; Nigam et al., 2019). In fact, mapping the 1780 nonsynonymous variants against the PPV genome show notably less variants covering the genome save for at the N termini of HC-Pro and CP, and the C terminus of P3, all of which have been previously noted as being variable (Carbonell et al., 2013; García et al., 2014; Maliogka et al., 2012; Nigam et al., 2019).

The structure of the PPV population within phloem and whole tissues identified 1290 phloem specific and 1174 whole tissue specific variants. The majority of these variants, 1255 in the phloem and 1160 in whole tissue samples, occurred in group 4 and were identified in only one sample. Interestingly, only 179 group 4 variants were shared between whole tissue and phloem translatomes (Table 1). In comparison, of the 153 variants that make up groups 1 through 3, 104 are found in both phloem and whole tissues (Table 1). It is likely that these group 4 variants lack the ability to spread and thus do not significantly accumulate, essentially making them one-offs within the cells and tissues they arise in. In contrast, the higher frequency variants identified in groups 1, 2 and 3 share a higher percentage across the two translatome tissues, indicating the ability to move within these tissues and thus a lack of tissue specificity (Table 1). However, in groups 2 and 3 several phloem (35) and whole tissue (14) specific variants were observed (Table 1). Of the 14 variants found only in the whole tissue translatome and thus non-phloem associated, 6 are nonsynonymous mutations affecting the following amino acid positions: P1 K188E, P1 K195N, P3 P252L, NIa-Pro S214A, Nib D504N, and CP D6A. The 35 phloem specific variants are mainly synonymous with the nonsynonymous mutations affecting only three amino acid positions: P3 A291V, NIa-VPg G188L and CP T75H. What role these variants have in conferring tissue specificity will require further investigation. However, the use of tissue specific translatome data for their identification represents a unique starting point for such investigations.

4. Conclusion

In this study, we sought to identify population structures that occur in specific tissues and at specific developmental times within a woody perennial host. Our findings indicate that the population structure of PPV consists of distinct variant categories that display high, low and fluctuating frequency levels during leaf development. Furthermore, leaf buds exhibit reduced infection levels yet maintain proportionately higher levels of variants indicating they may provide an environment permissive to the maintenance of diverse virus populations that could provide a yet unknown adaptive advantage.

5. Materials and Methods

5.1. Plant Growth Conditions and Virus Inoculations

A total of three pSUC2::HF-RPL18 and three p35S::HF-RPL18 transgenic European plum trees were previously inoculated by aphid transmission with the Penn7 isolate of the PPV-D strain (Schneider et al., 2011). This PPV isolate was sourced from an infected peach tree in York County, Pennsylvania (NCBI reference number EF640935) and has been maintained in P. domestica “Stanley” since 2009 (Schneider et al., 2009). Aphid transmission was carried out using a colony of the green peach aphid Myzus persicae fed on PPV-D Penn7 infected, detached symptomatic plum leaves (P. domestica ‘Stanley’) and then placed on the transgenic plum seedlings for an inoculation access period of 72 hrs. Inoculated trees were maintained under BSL-3P greenhouse conditions at 27°C for a duration of two 12 week-long growth periods to insure the establishment of PPV infections. PPV infections were verified by RT-PCR using PPV coat protein specific primers (Schneider et al., 2004).

5.2. Translatome and Total RNA Isolations

Leaf and bud samples were taken over two 3-month long vegetative growth cycles separated by a 60-day chilling period at 4°C. Leaf samples from individual trees were taken and processed for translatome RNA at 2-, 4-, 6- and 12-weeks growth while lateral and axial buds were sampled immediately before and after a chilling period (Fig. 1). Specifically, leaf composites of 10-15 randomly selected leaves from each tree were collected and flash frozen in liquid nitrogen. Similarly, lateral and apical buds excised from the budwood of each tree were flash frozen. Translatome RNA was extracted from frozen tissues as previously reported while total RNA was isolated using the RNeasy Plant Mini Kit per the manufacturer instructions (Qiagen, Valencia, CA, USA) (Collum et al., 2020, 2019).

5.3. Next Generations Sequencing and Analysis

Translatome and total RNA were submitted for cDNA library preparation and paired-end read sequencing using Illumina Hi-Seq 2500 system’s Rapid Run Mode (Genewiz, South Plainfield, NJ, USA). Output libraries were composed of 151 nucleotide paired-end reads. For each sample time, the RNA isolated from an individual tree was treated as a biological replicate. In total 59 cDNA libraries were generated as one two-week pSUC2 tree sample was lost. On average the libraries yielded between 33 to 105 million 151bp paired-end reads via Illumina NGS. Using CLC Genomics Workbench v. 10.0.1. (Qiagen, Valencia, CA, U.S.A.) reads were trimmed for quality and mapped to the PPV strain isolate Penn7 genome (NCBI Reference Sequence EF640935). Datasets are available under GEO accession number GSE131832 (Collum et al., 2020) and GSE150606 for PPV analysis.

5.4. Statistical analysis

Data were analyzed by t-tests or ANOVA, as appropriate, and statistical significance was defined as p<0.05. PPV levels were analyzed by two-sample student's t-Test assuming unequal variances in Microsoft Excel (2007). Percent variant reads were analyzed by one-way ANOVA for specific time points using Microsoft Excel (2007).

Supplementary Material

1

Table S1 PPV read numbers in plum whole-leaf translatome vs total RNA.

Table S2 PPV reads and variants in whole-leaf translatome and phloem translatome total RNA.

Table S3 Group 1 Variants

Table S4 Group 2 Variants

Table S5 Group 3 Variants

Table S6 Group 4 Variants

Table S7 Group 4 Variants with High Frequencies

Table S8 Nucleotide Variations Identified

Table S9 RNAi regulated genes in leaf and buds