MicroRNA profiling of the whitefly Bemisia tabaci Middle East-Aisa Minor I following the acquisition of Tomato yellow leaf curl China virus

Ago1 and Dcr1 genes are key elements involved in miRNA-mediated silencing. To determine whether the core RNA-induced gene silencing machinery is present in B. tabaci MEAM1, we cloned partial sequences for Ago1 and Dcr1 orthologs. A partial fragment of a putative whitefly Ago1 gene (encoding 743 amino acids) was sequenced and compared to orthologs from other species, including Drosophila melanogaster, Tribolium castaneum, Bombyx mori, Rattus norvegicus and Homo sapiens. As expected, all the sequences analyzed were conserved and the typical PAZ and Piwi-like domains reported for other Argonaute family proteins were also present in the whitefly Ago1 (Fig. 1). The entire 743 amino acids of the cloned B. tabaci MEAM1 Ago1 exhibited approximately 90 % sequence identity to Ago1 from other insects (Fig. 1a), and approximately 95 and 90 %, within the PAZ and Piwi-like motifs respectively (Fig. 1b). Similarly, the partial sequence of the whitefly Dcr1 gene (encoding 480 amino acids) was also conserved amongst different insects and the typical Ribonulease III domain reported for other Dicer family proteins was also present in whitefly Dcr1 (Fig. 2). The 480 amino acids of B. tabaci MEAM1 Dcr1 exhibited 70 % sequence identity to Dcr1 from other insects (Fig. 2a) and approximately 77 % identity for the Ribonulease III motif (Fig. 2b). These results indicate that two of the core components of the miRNA-mediated silencing system exist in the whitefly.

The presence of this system in whiteflies has implications regarding recent studies demonstrating differences in the global gene expression profile in viruliferous and nonviruliferous whiteflies [37]. It has been found that after TYLCCNV infection a number of genes involved in cell cycle regulation, primary metabolism, the immune response, Toll-like signaling and mitogen-activated protein kinase (MAPK) pathways were differentially regulated in the viruliferous whiteflies [37]. As miRNAs are now recognized as critical regulators of gene expression, we suggest that identification and comparison of miRNAs in viruliferous and nonviruliferous whiteflies could provide new information concerning biological changes in the vector upon virus infection.

In order to identify differentially expressed miRNAs involved in begomovirus-whitefly interactions, we constructed small RNA libraries from uninfected and TYLCCNV-infected whiteflies. High throughput Solexa sequencing of these two small RNA libraries was performed and low-quality sequences and sequences <18 nt or >31 nt were eliminated. A total of 1,910,274 reads (948,290 unique sequences) in the nonviruliferous library and 5,663,142 reads (1,910,584 unique sequences) in the viruliferous library were obtained. Analysis of the size distribution indicated that the highest percentage of small RNAs in both libraries were 21–23 nt (26.9 and 47.5 % of all reads in nonviruliferous and viruliferous libraries respectively) and 28–30 nt (47.5 and 27.6 % of all reads in nonviruliferous and viruliferous libraries respectively) in length (Fig. 3a). The small RNAs in the 21–23 nt range are consistent with that observed for miRNAs in animals [38], and small RNAs in the 28–30 nt range are consistent with pi-RNA-like sequences. A previous study on identification of miRNAs in B. tabaci B and Q has showed that the distribution of sequence lengths from both B and Q libraries were enriched with small RNAs of 21–23 and 28–30 nt [39]. Our data are consistent with this previous report for nonviruliferous whiteflies, which provide confidence in the reliability of the data.

Subsequent sequence analysis (NCBI GenBank and Rfam Version 10.1) indicated that, among these small RNAs, a total of 128,523 (nonviruliferous) and 613,517 (viruliferous) were rRNAs (71,238/3.73 % for nonviruliferous library and 356,499/6.30 % for viruliferous library), snRNAs (13/0.00; 35/0.00 %, respectively), or tRNAs (57,272/3.00; 256,983/4.54 %, respectively) (Fig. 3b). After eliminating reads corresponding to these RNAs, both libraries contained a large fraction of reads derived from unannotated genome sites (82.87 and 66.95 %, respectively) and miRNAs (9.44 and 21.27 %, respectively) (Fig. 3b). These unique data sets and read counts were used to identify conserved and novel miRNAs in whiteflies. The coverage of small RNAs is also consistent with previous report for nonviruliferous whiteflies [39]. It is interesting that the corresponding coverage of reads for rRNAs, snRNAs or tRNAs shows a similar trend in both libraries except for the miRNAs, which were expressed at a much higher level in the library from viruliferous whiteflies.

To identify conserved miRNAs in B. tabaci MEAM1, all clean small RNA tags were annotated into different categories to remove rRNAs, tRNAs, snRNAs, and snoRNAs using the Rfam database (Version 10.1). The remaining small RNAs from the nonviruliferous and viruliferous whitefly libraries were used to identify conserved miRNAs in B. tabaci MEAM1 by comparison to known miRNAs in the miRBase database (Version 19.0). Sequences in our libraries identical to or related to (having four or fewer nucleotide substitutions) miRNA sequences of D. melanogaster or other insects (Aedes aegypti, Apis mellifera, B. mori, and T. castaneum) were considered to be potentially conserved miRNAs. After BLASTn searches and further sequence analysis, a total of 112 conserved miRNAs were identified from the nonviruliferous whitefly library and 136 conserved miRNAs were identified from the viruliferous whitefly library (Additional file 1: Table S1).

We then compared expression levels of putative conserved miRNA between nonviruliferous and viruliferous whitefly libraries. For the comparison, we first normalized the expression of miRNAs (normalized expression = actual miRNA count/total count of clean reads × 100000), and then set a threshold of a 2-fold difference in normalized expression and a representation of 0.1 % (actual miRNA count/total count of clean reads) in both miRNA libraries. Between the nonviruliferous and viruliferous libraries, we identified 52 miRNAs that were differentially expressed. Among these, 26 miRNAs were unique to the viruliferous library and 2 miRNAs were unique to the nonviruliferous library (Additional file 2: Table S2). Among the 24 miRNAs that were identified in both libraries, 15 miRNAs were up-regulated and 9 reduced relative to the nonviruliferous library (Fig. 4). The strongest relative induction was observed for bantam (43-fold), let-7a-5p (26-fold), miR-1175-3p (13-fold) and miR-219 (10-fold), while the strongest reduction was observed for miR-306-5p (6-fold), miR-993a-5p (5-fold), miR-2779 (4-fold) and miR-307 (3-fold) (Fig. 4).

Bantam miRNA has been reported to simultaneously stimulate cell proliferation and prevent apoptosis in response to patterning cues in Drosophila [40]. In our study, bantam miRNA was significantly up-regulated in the viruliferous whiteflies as compared to the nonviruliferous controls. It has been reported that TYLCCNV can activate whitefly immune responses, including autophagy. The induction of autophagy can inhibit cell growth and induce apoptotic cell death, which might lead to a gradual decrease of viral particles within the body of viruliferous whiteflies [37, 41]. An enhanced level of bantam in the viruliferous whiteflies suggests that bantam may act to arrest the apoptotic response and help to maintain homeostasis in the presence of virus.

The let-7 miRNA family, which includes let-7a, let-7b, let-7c, let-7d, let-7e, let-7f, has been known to play an important role in cell cycle, proliferation and apoptosis. Moreover, let-7 has been implicated in post-transcriptional control of responses to pathogenic agents [42, 43]. Significant up-regulation of let-7 was also detected in viruliferous whiteflies, suggesting that this miRNA may act to perturb the cell cycle in the whitefly, thus offering one explanation for the negative effect of this virus on the longevity and fecundity of B. tabaci MEAM1. Recently, miR-219 has been connected with NMDA receptor signaling in humans, and it has been shown that deregulation of this miRNA can lead to the development of mental disorders such as schizophrenia [44]. Viruliferous whiteflies also showed a significant increase in miR-219 expression level, although how this is relevant to the presence of virus in the whitefly is unknown. While the physiological functions of other up- and down-regulated miRNAs are still unknown, their specific expression patterns indicate that they are also likely to play critical roles in hypometabolic processes in whiteflies. Further experiments are needed to elucidate their potential roles in this process.

Next, we used the miRNA prediction software miRDeep2 [45] to identify putative novel miRNAs by searching against the previously published B. tabaci MEAM1 transcriptome database [46]. In total, 7 potential novel miRNA candidates were identified from both the nonviruliferous and viruliferous libraries (Table 1). The length of the 7 predicted novel miRNA candidates ranged from 18 to 23 nt. The miRDeep2 scores for these novel miRNA candidates were all ≥4. Interestingly, we found that the expression levels of all 7 novel miRNA candidates were up-regulated in the viruliferous as compared to the nonviruliferous library. To investigate whether these novel miRNA candidates are conserved in a wide range of animal species, we used these miRNAs as query sequences to perform BLASTn search against all nucleotide sequences in miRBase Version 19.0. The results indicated that these miRNAs candidates were not found in other species, possibly indicating that they are specific to whiteflies and thus have a species-specific role(s). Interestingly, we found the sequence of bta-miRn16 also map into the genome of a primary endosymbiont, Candidatus Portiera aleyrodidarum [47], which has been found to accompany whiteflies in our laboratory without any treatment [48, 49]. Furthermore, to evaluate the novel miRNA candidates represent true miRNAs, the hairpin structures for putative pre-miRNA hairpins were generated using RNAfold [50]. As shown in Fig. 5, although differing complexities, all novel miRNA candidate precursor sequences fold into hairpin structures characteristic of miRNA precursors.

To further validate the miRNA sequencing data, we used qRT-PCR to examine the expression levels of four up-regulated miRNAs (bantam, miR-1, −2b, and −124) and four down-regulated miRNAs (miR-306-5p, −307, −317, and −993a) in the viruliferous relative to nonviruliferous library. As shown in Fig. 6, these miRNAs showed an expression pattern consistent with the results obtained from Solexa sequencing (Fig. 4), except miR-306-5p that exhibited similar levels of expression in the two libraries.

For the potential novel miRNA candidates identified in both nonviruliferous and viruliferous libraries, we also examined the expression level of four novel miRNAs by qRT-PCR. All of them could be amplified by qRT-PCR using specific primers (Fig. 7). Three of the potential novel miRNA candidates showed higher expression in viruliferous than in nonviruliferous whiteflies, which was the same trend observed in the sequencing data of novel miRNA candidates (Table 1). However, one novel miRNA, bta-miRn17, exhibited a decrease in expression in viruliferous as compared to nonviruliferous whiteflies (Fig. 7a), which is the opposite to that of Solexa sequencing. The reason for these differences is currently unknown.

The function of a miRNA is ultimately defined by its effects on the expression of target genes. To reveal the possible functions of the miRNAs identified in whiteflies, potential targets of conserved and novel miRNAs were predicted using the previously published B. tabaci MEAM1 transcriptome database [46] with the miRNA target prediction algorithm miRanda 3.1. In total 193,090 targets were obtained. The length of the target sequences varied from 200 to 5926 bp, and the binding energy between the miRNAs and the target varied from −20.01 to −48.05 kCal/mol. All of the miRNAs had more than 100 predicted targets. Some miRNAs had more than 1000 predicted targets, and some target genes were putatively regulated by more than two miRNAs. For a better understanding of the functions of the miRNAs, we analyzed the number of miRNA targets in a specific GO. Predicted targets covered three main categories: biological process, cellular component, and molecular function (Additional file 3: Figure S1).

To gain insight into the potential functions of the differentially expressed miRNAs of the nonviruliferous and viruliferous libraries, GO analysis was also used to classify the potential enriched functional groups of their putative targets. For both up- and down-regulated miRNAs in the target libraries, the biological processes most represented are cellular and metabolic processes, whereas binding and catalytic activity are among the most represented molecular function categories, which is similar to the GO analysis of all conserved and novel miRNA targets. To identify the relatively enriched GO terms for up- and down-regulated miRNAs, we first normalized the up- and down-regulated miRNA target genes (normalized target genes = up- or down-regulated miRNA target gene count/total up- or down-regulated miRNA target gene count), and then set a threshold of a 1.5-fold difference in normalized target genes (Table 2). The target genes of up-regulated miRNAs were significantly enriched in eight GO terms, including antioxidant activity, translation regulator activity, protein binding transcription factor activity, structural molecule activity, growth, immune system process, negative regulation of biological process and membrane-enclosed lumen. The targets of down-regulated miRNAs were enriched in nine GO terms, including channel regulator activity, metallochaperone activity, virion, virion part, synapse, synapse part, cell junction, transporter activity and biological adhesion (Fig. 8). Of particular interest is the enrichment of the targets of up-regulated miRNAs in growth. This may be linked to the global transcriptional depression of growth-related genes and contribute to the reduced fecundity and longevity in viruliferous whiteflies [37, 51]. For the targets of novel miRNAs, as compared with the targets of conserved miRNAs, there was a very similar GO distribution (Additional file 4: Figure S2). This suggests that the novel miRNAs might be functionally divergent.