Characteristics of vsRNAs in CymMV and double infected tissues

3. Analyses of small RNAs from CymMV and mixed infected P. amabilis by deep

3.3. Characteristics of vsRNAs in CymMV and doubly infected tissues

3.3.2. Characteristics of vsRNAs in CymMV and double infected tissues

The size distribution, strand polarity, and 5’-end nucleotide preference of CymMV and ORSV vsRNAs were then analyzed. In the Ci and Cc libraries, which represented CymMV singly infected tissues, CymMV vsRNA of 21 nt (~61.2 to 63.5%) was predominant, followed by the 22 nt class (~18.3 to 22.3%) (Fig. 23-Ci/Cc). Size distribution of CymMV and ORSV vsRNAs in doubly infected tissues (Di, Dc, and Dsc) displayed similar patterns, with 21 nt as the predominant size class (~56.7 to 62.3% for CymMV vsRNAs and ~50.4 to 53.9% for ORSV vsRNAs) and 22 nt as the second most abundant (~13.3 to 22.3% for CymMV vsRNAs and ~12.2 to 15.6% for ORSV vsNRAs) (Figs. 23-24, Di/Dc/Dsi).

In regard to strand polarity, sense-stranded CymMV vsRNA proportioned to 70.3%

and 73.4% in the Ci and Cc libraries, respectively (Fig. 23-Ci/Cc). Similar ratios of (+)-stranded vsRNAs were observed in the Di, Dc, and Dsc libraries, accounting for about 71.4 to 73.4% of total CymMV vsRNAs (Fig. 23-Di/Dc/Dsi), and about 57.9-64.8% of total ORSV vsRNAs (Fig. 24-Di/Dc/Dsi).

After grouping by 5’ end nucleotide identity (A, C, G, and U), differential nucleotide preference among 21 to 24 nt vsRNAs was observed. Generally, C- and U-terminated vsRNAs were dominant in the 21 and 22 nt classes, and specifically in the 24 nt class, the proportion of 5’-A vsRNAs was elevated and became dominant (Fig. 25).

This pattern was likewise observed in the Ci, Cc, Di, and Dc libraries, and regardless of CymMV or ORSV origins in the Di and Dc libraries. With the exception of the Dsc library, a more drastic nucleotide preference was observed, however, it seemed biased from low vsRNA reads (Figs. 25-Dsc and 26-Dsc). Moreover, this pattern was inconsistent with ORSV singly infected tissues (Oi and Oc) in which ORSV vsRNAs

possessed an even distribution of 5’-A, C, and U in the 21 to 24 nt classes (Fig. 12).

Nonetheless, CymMV and ORSV vsRNAs were always less frequently 5’-G, regardless of size classes in all libraries (Figs. 25-26).

Collectively, the prevalence of 21 and 22 nt size classes and (+)-stranded polarity was found for both CymMV and ORSV vsRNAs in all libraries, indicating these features differed little between the two viruses, and were also not much affected under mixed infection. In double infections, the 5’-end nucleotide preference remained unchanged for CymMV vsRNAs, but not for ORSV vsRNAs.

3.3.3. Genome mapping and coverage of CymMV vsRNAs

To study the distribution of vsRNA origins along virus genomes, vsRNAs were mapped along each polarity of the CymMV and ORSV genome according to the 5’-end sites. Including both sense- and anti-sense stranded and all sizes of vsRNAs, the genome coverage was up to 91.04% and 98.81% for CymMV and ORSV vsRNAs in the Di library, respectively, suggesting widespread targeting of DCL enzymes. The CymMV vsRNA distribution along the CymMV genome extensively encompassed the RdRp-coding region, and obviously was seldom located in the TGBp- and CP-coding regions (Fig. 27). The distribution of CymMV vsRNA origins was less affected by co-infection with ORSV. Northern blot hybridization confirmed the accumulation of nt 2404 to 3703 RdRp region-originated CymMV vsRNAs in Ci, Di, and Dc tissues (Fig.

21A). The enrichment of the near 5’ region vsRNAs possibly reflected that more CymMV genomic RNA accumulated in these tissues, as northern blot revealed (Fig.

21A).

3.3.4. Specific ORSV vsRNA hotspots occurred in mixed infected tissues

While CymMV vsRNA hotspot distribution displayed resemblance to singly infected CymMV tissues and co-infected CymMV with ORSV tissues (Fig. 27), a sharp difference in ORSV vsRNA hotspots was observed in the Di and Dc libraries (Fig. 28) compared to ORSV singly infected tissues (Oi and Oc, as previously described) (Fig.

13). ORSV vsRNAs originated around (-)-strand nt 2640 to 2680, 2240 to 2280, and 4150 to 4180(9680) RdRp regions, comprising three prominent peaks (Fig. 28). The major vsRNA tags of these peaks were proportioned to 0.99% (5118 reads), 1.43%

(7377 reads) and 1.88% (9680 reads) total ORSV vsRNA pool in the Di library, respectively. Similar peaks were also observed in the Dc library (Fig. 28). Interestingly, the hotspot tags were also sequenced in the ORSV singly inoculated (Oi) library, but contributed to a much lower percentage than in doubly infected tissues, specifically, 0.01% (158 reads), 0.10% (2542 reads), and 0.12% (2865 reads) for the three topmost tags, respectively. The enrichment of these tags in the Di and Dc libraries indicating DCL accessibility to ORSV viral RNA may be affected under co-infection with CymMV.

3.3.5. Prediction of potential P. amabilis target transcripts of vsRNAs

Based on the sequence homology, previous research has suggested vsRNA induced symptoms form through RNA silencing mediating host mRNA degradation (Navarro et al., 2012; Shimura et al., 2011). To search transcripts that could be targeted by CymMV or ORSV vsRNAs, some vsRNA tags were aligned to the reverse complementary of P.

aphrodite EST sequences. Assuming tags with prevalent reads may have a higher chance of impacting expressions of target genes, the search was focused on the 50 most

abundant reads in the Oi, Ci, Di, and Dc libraries (Tables S3-S4). Moreover, since the necrotic symptoms only appeared in the Di tissues at 10 dpi, tags with over 50 reads and those only sequenced in the Di library were also subjected to target prediction (Tables S5-S6).

After the permissive alignment search, a position-dependent scoring matrix and MFE ratio criteria were applied to candidate vsRNA:mRNA duplexes following the method described by Allen et al., 2005. The search resulted in 78 and 48 potential targets for the 50 most abundant CymMV and ORSV vsRNAs, respectively. Of these, 49 of the CymMV and 42 of the ORSV vsRNA targets were genes predicting coding annotated proteins. Regarding the Di specific vsRNAs, 125 (88 were annotated contigs) transcripts were matched to CymMV and 22 (14 were annotated contigs) transcripts were matched to ORSV vsRNAs. The functional annotations of potential targets of Di specific CymMV vsRNAs were found to be significantly enriched (P < 0.001) in annotations of protochlorophyllide oxidoreduxtase activity, and endopeptidase activity, and chlorophyll biosynthetic process. The annotations of molecular functions of ORSV vsRNA targets were significantly enriched (P < 0.001) in ligase activity and substrate-specific transporter activity, and also enriched in biological functions such as sexual reproduction, multicellular organism reproduction, and response to biotic stimulus.

The matches of ORSV vsRNAs include ureide permease 2, ribosomal L38e protein, eukaryotic translation initiation factor (eIF) 2 beta subunit, putative cellular apoptosis susceptibility protein, UDP-glycosyltransferase superfamily protein, flavin-dependent monooxygenases (FMO) and other oxidoreductases or kinases. (Tables 8-9). CymMV vsRNAs were perfectly matched to geranyl diphosphate synthase, aspartokinase, and the

26S ribosomal RNA gene. Other potential targets included RNase III-like enzyme, chloroplastic glutaminyl-tRNA synthase, xyloglucan galactosyltransferase, abscisic acid response protein, serine/threonine kinase, light-dependent NADPH:protochlorophyllide oxidoreductase and other enzymes. (Tables 10-12).

3.4. Identification and analysis of conserved miRNAs in CymMV and mixed infected Phalaenopsis amabilis

3.4.1. miRNA populations

Known miRNAs were annotated using BLASTN search against miRBase with a tolerance of up to three mismatches. Again, the super-abundant miR166 tag was sequenced and contributed to an even higher proportion. This time, the super-abundant tag contributed to over 80% total miRNA reads and was proportioned to about 17 to 31% of total clean reads among the seven libraries (Table S7). The overwhelming preponderance of this miR166 tag was again inconsistent with qRT-PCR validation (data not shown). With the miR166 tag, conserved miRNAs were proportioned to about 21 to 36% of total clean reads within each library (Tables 6 and S7). The percentage dropped to about 4 to 6% after the miR166 tag was removed (Table S7). Nonetheless, 104,797 unique miRNA sequences corresponding to 154 families were identified in the seven libraries. The number of members within each family ranged from one to hundreds of species. The largest family was miR166, with 392 members identified (Table 13). Regardless of whether it was a single or mixed infection, the categories of miRNA families and members within each family remained similar in the different libraries (Table 7).

3.4.2. Differential expressions and predicted targets of CymMV and double infection responsive miRNAs

The expression levels of miRNA families between libraries as described were compared. In brief, read counts were normalized with miRNA pool sizes and converted to TPM for calculating the FC index between inoculated (Ci, Di), non-inoculated (Cc, Dc, Dsc) and mock-inoculated (Mi, Mc) tissues (Table 14). With a threshold of 1.5-fold, stem-loop qRT-PCR results and statistical analysis supported the up regulation of miR156, miR168, miR319 and miR5139 in inoculated tissues. MiR156 and miR168 were induced in Ci, Di, and Dsi tissues, whereas the induction of miR319 and miR5139 expression was only observed in the Di library (Table 14 and Fig. 30). Repressed expressions were observed in miR398, miR408, miR528 and miR535. The down regulation of miR398, miR408, and miR528 reached a greater extent, especially in Dsc and Dsi tissues (Table 14 and Fig. 30).

Target prediction was next performed with the focus on validated infection responsive miRNAs as described. The prediction for miR156, miR168, miR5139, miR528 and miR535 was shown in Section 2.4.3. After searching against P. aphrodite contigs, 18 potenial targets were predicted for miR396, 11 for miR319, 5 for miR408, 2 for miR393, 1 for miR394, and miR162, miR169, miR398 each with three matches (Table 15). Moreover, none of the Phalaenopsis transcript was matched to miR894 sequences, thus the target remained unknown. Integrating the prediction results, the potential target genes can be grouped into two major categories based on the functional annotations. First, several targets were involved in ABA, GA, and auxin mediated pathways and other signaling cascades, such as NUCLEAR FACTOR Y A5 (NFYA5), CYCLOIDEA AND PCF TRANSCRIPTION FACTOR (TCP) and MYB domain protein,

TRANSPORT INHIBITOR RESPONSE 1 (TIR1), LEAF CURLING RESPONSIVENESS (LCR), and Phox-associated domain protein. Second, SUPPRESSOR OF VARIEGATION 3 (SVR3) and ABA 8'-hydroxylase targeted by miR396, COPPER/ZINC SUPEROXIDE DISMUTASE 2 (CSD2) targeted by miR398, and LACCASE 3 (LAC3) targeted by miR408 were involved in oxidation processes and other metabolic pathways (Bueno et al., 1995; Yu et al., 2008). Other targets including genes functioning in developmental regulation such as plantacyanin (targeted by miR408) and PAUSED (PSD) karyopherin (targeted by miR396) played roles in vegetative and reproductive phase transition (Dong et al., 2005; Li and Chen, 2003;

Ranocha et al., 2002) (Table 15). Of the predicted targets, CSD2 has been proven to be an important antioxidant enzyme that can scavenge reactive oxygen species (ROS) in A.

thaliana (Jones-Rhoades and Bartel, 2004; Sunkar et al., 2006). The plantacyanins were associated with programmed cell death and callose deposition processes in pollen development (Dong et al., 2005), and have also been designated as stress-related proteins mediating lignin polymerization, which are thus involved in plant defense responses (Hampton et al., 2004; Nersissian et al., 1998). The target prediction revealed a broad spectrum of cellular processes potentially involved in the plant responses to virus infection.

Discussion

This study aimed to profile the small RNAs from CymMV and ORSV singly or doubly infected orchid tissues by Solexa deep sequencing. Using the leaf tip-inoculation method, two stages of infection, early and late, were distinguishable in non-inoculated and inoculated tissues. The characteristics of vsRNAs and their expressions under viral stress were further analyzed by bioinformatics. Target prediction provided information about possible roles of infection responsive small RNAs in plant-virus defense and counter-defense interactions.

1. Synergistic enhancement of CymMV infection by ORSV co-inoculation in Phalaenopsis amabilis

In double inoculation assays, synergistic effects were observed between CymMV and ORSV. First, chlorotic lesions and ringspots were noticed around 7 dpi and continued enlarging afterwards in mixed inoculated tissues (Fig. 20 and data not shown), consistent with previous studies reporting synergism between CymMV and ORSV and intensified symptoms in doubly infected plants (Hadley et al., 1987; Lawson and Brannigan, 1986; Pearson and Cole, 1991). Second, northern blot analysis revealed enhanced CymMV titer (Fig. 21A) as well as about 5-fold accumulation of CymMV vsRNAs in doubly inoculated tissues (Di) compared to singly inoculated tissues (Ci). In contrast, the viral titer of ORSV did not increase in Di tissues by CymMV co-infection and was non-detectable with northern blot in Dc tissues (Fig. 21B). The lower ORSV titer in Di than in Oi (Fig. 21B) may be due to lower amount of virions as inoculum (1 μg for the single inoculation and 0.5 μg each in the double inoculation). Previously, Hu

et al. reported that accumulation of both CymMV and ORSV progeny RNA were increased in Dendrobium protoplasts, and, in particular, the accumulation of ORSV viral RNA reached a peak earlier in double infections than when infected with ORSV alone (Hu et al., 1998). Likewise, the acceleration of ORSV genomic RNA accumulation was observed in mixed-infected N. benthamiana (Ajjikuttira and Wong, 2009). The results in this study was contradict to both reports. This discrepancy may be due to different host species, virus strains or the experimental systems used in individual studies. Different adaptability of CymMV and ORSV in P. amabilis should also be taken into consideration when interpreting these results.

Finally, accelerated spreading of CymMV in doubly infected tissues was found in this study. In contrast, spreading of ORSV was less affected by co-inoculation. Thus enhanced viral titer may facilitate the faster spreading of CymMV. In addition, the molecular interactions between CymMV and ORSV may also aid in viral movement.

Ajjikuttira et al. (2005) proved the MPs and CPs were functionally interchangeable between CymMV and ORSV, facilitating movement of these viruses in plants, with the exception of long-distance movement of ORSV RNA by CymMV CP. CymMV CP may play a dominant role in determining the host range as suggested by Lu et al. (2009) since a CymMV strain that infected Phalaenopsis but failed to accomplish systemic infection in N. benthamiana. In this study, the CymMV strain used for inoculation assays could not systemically infect N. benthamiana (data not shown). Perhaps the movement of ORSV was not greatly facilitated under double inoculation with this CymMV strain, and consequently, CymMV benefitted more than ORSV from the synergistic interactions.

2. The leading roles of DCL4 and DCL2 in P. amabilis

Characteristics of vsRNAs analyzed through deep sequencing profiles could contribute to deciphering the host defense mechanism. The size distribution of vsRNAs is an indicator of activities of multiple DCL proteins. Most of vsRNAs are 21- and 22-nt through processing of DCL4 and DCL2, respectively, as well established in A. thaliana (Blevins et al., 2006). The previous studies as well as the present results support this finding. RNA viruses from different families, such as TMV (Qi et al., 2009), CMV, MNSV, PMMoV, PVX, TRV, TuMV, TYLCV, WMV (Donaire et al., 2009), generated more 21 nt vsRNAs. In P. amabilis, ORSV and CymMV vsRNAs were mainly 21 nt, followed by 22 nt class (Figs 11, 23, and 24), mirroring the conserved roles of DCL4 and DCL2 homologs in P. amabilis antiviral RNA silencing, as in many other plants. In addition, our results showed similar size distribution patterns between inoculated and non-inoculated tissues, and these also remained unchanged in singly and doubly infected tissues. These data indicate the leading role of DCL4 homologs in P. amabilis was activated in the early stage of infection and was not much affected under conditions of mixed-infection.

3. Asymmetrical strand polarity and 5’-end nucleotide identity of CymMV and ORSV vsRNAs

Previous studies suggested vsRNAs were generated from sense and antisense annealed replicative intermediates and long dsRNA synthesized through host RDR activities using viral ssRNA templates (Ding, 2010; Ruiz-Ferrer and Voinnet, 2009;

Voinnet, 2005), thus vsRNAs derived equally from positive and negative strands, as observed in Cucumber yellow virus (Yoo et al., 2004). However, asymmetrical polarity

of vsRNAs has been reported in several studies and was also profiled in this one. In P.

amabilis, CymMV and ORSV vsRNAs mainly derived from sense strands of viral genomes and presented consistent asymmetry in strand polarity, both between single and double infections, and also regardless of late or early stages of infection (Figs 11, 23, and 24). For other viruses, more positive stranded vsRNAs were cloned in plants infected with TuMV (Ho et al., 2007), TCV (Ho et al., 2006), TMV (Qi et al., 2009), PVX, CymRSV (Molnar et al., 2005), or Cotton leafroll dwarf virus (Silva et al., 2011).

In contrast, vsRNAs of several grape viruses such as GFkV, GRGV, and GAMaV have been profiled revealing a predominance of antisense polarity (Pantaleo et al., 2010).

Collectively, these data imply vsRNAs may not solely be processed from long dsRNA templates, consistent with studies suggesting self-annealing hairpin structures formed from complementary sequences within viral ssRNA served as templates for vsRNA biogenesis (Moissiard and Voinnet, 2006; Molnár et al., 2005).

While the size distribution and strand polarity were similar for CymMV and ORSV vsRNAs, one major difference between the vsRNA of these two viruses was found in the 5’-end nucleotide composition. ORSV vsRNA generally showed a modest preference for 5’-A and 5’-U among 21 to 24 nt classes in single infections (Fig. 12), similar to vsRNAs of CMV, CymRSV, MNSV, PMMoV, PVX, TRV, TuMV and WMV (Donaire et al., 2009). In contrast, CymMV vsRNA had a strong tendency to be terminated with 5’-C or 5’-U among 21 to 23 nt species (Fig. 25), resembling GRSPaV (Pantaleo et al., 2010) and Citrus tristeza virus (CTV) (Ruiz-Ruiz et al., 2011).

Nonetheless, this study observed a common feature of underrepresented 5’-G in both CymMV and ORSV as well as many other viruses reported (Donaire et al., 2009;

Pantaleo et al., 2010; Ruiz-Ruiz et al., 2011). Since the 5’-end nucleotide identity is a

key determinant for sorting vsRNAs to associate with distinct AGO proteins (Mi et al., 2008), which have different affinities for binding small RNAs with 5’- U (AGO1), A (AGO2 and AGO4), and C (AGO5) (Mi et al., 2008; Montgomery et al., 2008; Takeda et al., 2008), thus the 5’-A and U preferred ORSV vsRNA may be recruited into AGO1, AGO2, and AGO4, whereas CymMV vsRNAs may preferentially bind to AGO1 and AGO5. Furthermore, the absence of known 5’-G preferred AGO species may explain the avoidance of 5’-G vsRNAs (Donaire et al., 2009).

The nucleotide preference did not solely represent the base composition of viral genomes [U(30.76%)>A(29.81%)>G(21.64%)>C(17.79%) for ORSV and C(29.16%)>A(26.44%)>T(24.67%)>G(19.73%) for CymMV]. The different preferences of 5’-terminal nucleotide among different size classes observed in CymMV vsRNAs was a special feature, compared to other plant viruses. In addition, the proportion of 5’-A was also elevated in the 24 nt class of ORSV vsRNAs in mixedly infected tissues (Fig. 26). In this case, unknown factors specific to CymMV may be involved in the enrichment of 5’-A in the 24 nt class, and possibly affect ORSV vsRNA generation or stabilization in the mixed infections. Moreover, the 5’-A preference of 24 nt vsRNAs seemed to coincide with features of plant endogenous small RNAs which are involved in RNA-dependent DNA methylation (RdDM) pathways (Molnar et al., 2010). Since CymMV is a cytoplasmic replicated RNA virus, further analysis is needed to determine how the 24-nt vsRNA was generated and whether these vsRNAs could be incorporated into AGO4 and act in the mode of RdDM machinery.

4. Differential distribution of CymMV and ORSV vsRNA along viral genomes One substantial difference between the characteristics of CymMV and ORSV

vsRNAs was the vsRNA origin along viral genomes. In singly infected tissues, more hotspots resided in the RdRp coding region of CymMV and the CP to 3’-UTR region of the ORSV genome, respectively (Figs. 13, 27, and 28). This discrepancy may be due to higher accumulation of ORSV CP-expressed subgenomic RNA and of more abundant CymMV gRNAs in the infected P. amabilis tissues as shown in northern blot (Figs. 8 and 21A). This may be a possible explanation as to why more vsRNAs originated from different genome sites for CymMV and ORSV.

In ORSV singly infected tissues (Oi and Oc), a specific hotspot located in 3’-UTR of the ORSV genome was found With some interesting features. First, it was constituted by one specific tag and mainly from sense strands. Second, though three consecutive homologous regions reside in the PK chain of ORSV 3’-UTR, the hotspot only peaked at the lattermost repeat, but not at the other two upstream homologous sequences. Last, modelling of the 3’-UTR secondary structure by Gultyaev et al. (1994) revealed the hotspot vsRNA resided in the first PK structure upstream to the TLS (Fig. 15), which was not a typical double stranded structure expected to be preferentially cleaved by DCL enzymes. These features suggested that the accessibility of DCL enzyme dicing at the hotspot was not merely based on the primary sequence or secondary structures, and may involve other determinant factors although the abundant hotspot vsRNAs were generated within the positive strand. Furthermore, it has been demonstrated that the highly structured PK chain and TLS ends are critical for tobamovirus replication and protein translation (Zeenko et al., 2002). However, overexpression of the hotspot vsRNA failed to interfere with ORSV replication in the co-transfection assay. Similarly, mutations in Zucchini yellow mosaic virus (ZYMV) at the loci of hotspots did not break

在文檔中以高通量定序方式分析蝴蝶蘭感染齒舌蘭輪斑病毒及蕙蘭嵌紋病毒後小型核醣核酸組成之變化 (頁 67-200)