Assembly and functional annotation
Two de novo transcriptomes were assembled from each of SOAPdenovotrans and Trinity. Overall, the Trinity assemblies contained more and longer contigs than the SOAP assemblies, and seemed to have slightly better coverage of highly conserved, single copy orthologous genes. In addition to recovering 95% of the ultra-conserved orthologs (60% of which were present as a single copy), gene candidates were identified from all canonical immune gene pathways (e.g. RNAi, Jak/STAT, Toll, IMD), supporting the relative completeness of the assembly. As expected for biological samples that were not controlled for life-history stage, experimental conditions or tissue type, when broken down into functional categories through the use of GO terms, the P. interpunctella de novo assembly differs from other lepidopteran transcriptomes (Figure 1).
Differential expression
Forty-seven P. interpunctella genes were differentially expressed 24 hours after exposure to PiGV compared to exposure to a control solution. The observed changes in expression could have several mechanistic origins, including defence, tolerance and repair. The differentially expressed genes were enriched for cuticle proteins (N = 7), 6 of which were virally downregulated and 1 was upregulated. To our knowledge, cuticle proteins have not been implicated previously from studies examining mRNA levels in cell cultures after infection with a baculovirus [14]. Cuticle proteins could play a role in defence in at least the following three ways. First, cuticle proteins form a large proportion of the peritrophic membrane (PM). The PM lines the gut and provides the first line of defence against ingested pathogens, such as PiGV and has been strongly implicated in antiviral defence [15,16]. For example, the PM of more susceptible velvetbean caterpillars, Anticarsia gemmatalis had a lower chitin content and provided a less efficient barrier against its baculovirus (AgMNPV) than more resistant larvae [15]. Furthermore, changes in the peritrophic membrane are correlated with changes in the risk of pathogen infection. For example, the thickness of the PM is also well known in Anopheles mosquitoes to increase after the ingestion of a blood meal, which is the primary source of infective pathogens [16].
The change in expression of cuticle proteins in P. interpunctella could reflect similar processes. Generally, cuticle proteins in the peritrophic membrane have a distinctive molecular signature, which is not present in any of the significantly differentially expressed cuticle proteins identified. However, since studies characterizing PMs have only been conducted in a limited number of species, which are not closely related to the lepidopteran P. interpunctella, it is possible that the molecular signature is too divergent to recognize. Furthermore, not all proteins of the PM have been characterized. For example, [17] recently identified a new PM protein in the meadow moth that is able to bind chitin, but does not contain the conserved binding domain. Secondly, expression of cuticle proteins has been monitored in insects and been shown to change when moulting takes place [18]. In Plodia, moulting includes shedding the gut lining, to which the PiGV particle may attach prior to penetration into the haemolymph. Thus, the differentially expressed cuticle proteins might correspond to Plodia moulting. Finally, the cuticle proteins may directly inhibit viral replication. For example, [19] demonstrated that a mosquito cuticle protein (AAEL011045) binds to a viral envelope and thus inhibits infection. Furthermore, they discovered that this protein was downregulated in virally exposed mosquitoes, raising the possibility that the virus was actively suppressing the expression of this molecule.
In order to discover if the degree of sequence identity amongst the 7 differentially expressed cuticle-proteins in this study could be used to explain their opposing pattern of regulation (i.e. up versus down), we aligned them with mosquito cuticle protein AAEL011045 in BioEdit v.7.2.4 [20] and built a neighbour-joining tree using CLUSTAL v1.2.0 [21] (Figure 5). This phylogenetic analysis shows that the seven moth genes cluster into three major clades; two clades of three genes each and a singleton. The singleton is the only cuticle protein that is upregulated after viral exposure. The moth genes in the clade containing the mosquito cuticle protein AAEL011045 are all candidate immune system genes that may be able to bind to the envelope of PiGV and thus inhibit its infectivity.
Two of the most up-regulated genes after viral exposure are a reverse transcriptase and a transposase. This suggests that within 24 hours of exposure to a virus, transposable elements (TE) activity, or ‘jumping’, is switched on (there were no reads from the control samples that mapped to either of these two genes). Supporting this hypothesis, a study that detected differential expression in hemocytes of the moth larvae Heliothis virescens after infection with Helicoverpa zea single nucleopolyhedrovirus found many retrotransposons to be upregulated [22]. A recent study in Drosophila showed that fragments of a virus were reverse-transcribed, producing truncated versions of the virus, which were processed by the RNAi machinery resulting in a reduction in the amount of active virus in the cells [23]. Due to the increase in transposase activity found in this system, it will be interesting to test if there is a connection between TE activity and the control of a viral infection in this system as is the case in Drosophila.
Of all of the differentially expressed genes, 36 (70%) were downregulated. The prevalence of down-regulation suggests that PiGV may directly or indirectly exert a considerable inhibitory effect on the host immune response. Indeed, late in infection, some studies in cell culture have documented a global downregulation of host genes (e.g. [24]). Similar effects of viral suppression on host immunity have been documented in other in vivo insect/viral systems (Aedes aegypti infected with Dengue-virus [19,25], West Nile Virus and Yellow Fever Virus [19]). Indeed, in our differentially expressed data set, the only canonical immune system gene, a gram-negative binding protein (GNBP), is downregulated in virally-exposed larvae compared to control larvae. GNBPs are intricately involved activating the Toll pathway, and are upregulated after exposure to bacterial pathogens. The fact that the GNBP identified in this experiment is downregulated suggests that the virus may be suppressing the TOLL pathway. This potential for a trade-off between the immune response to bacteria versus viruses has been documented in mosquitoes, where the growth of E. coli was enhanced after viral infection with dengue virus (an RNA virus) in A. aegypti hosts [25]. They went on to show that viral interference with the host immune system resulted in a decreased production of antimicrobial peptides. Our transcriptome assembly contains several putative antimicrobial peptides, but none of them were differentially expressed after infection with the virus. However, our sampling time point was very early in the infection process, and thus sampling at more time points will be necessary to test if the viral infection has an effect on antimicrobial peptide activity.
The change in GNBP transcription may support the hypothesis that the innate immune response to a viral infection is not be restricted to the RNAi pathway, but instead additionally involves the Toll pathway. However, it is difficult to distinguish if the activation of different immune pathways is a direct or indirect result of the virus. For example, if the virus crossing the gut membrane results in a wound, naturally occurring gut microbes will also pass into the haemolymph and trigger an immune reaction. Our finding that one of the differentially expressed genes in the moth larvae was of bacterial origin is consistent with this hypothesis. Because we only assessed transcription very early in infection, it is not possible to determine if other members of the Toll pathway would also be differentially expressed.
Some of the genes that were determined to be differentially expressed are not functionally annotated (N = 16, 31%). Many of these genes have high sequence similarity to other organisms, suggesting that functionally conserved, important putative immune, tolerance or repair genes have yet to be characterized in insects. Additionally, some of the genes had no known homology to other organisms, suggesting that there are also highly specific genes that are functionally uncharacterized. This study provides a solid foundation for choosing candidates for further functional characterization.