Transcriptome profiling of whitefly guts in response to Tomato yellow leaf curl virus infection

Two cDNA libraries of viruliferous and nonviruliferous guts were sequenced and generated 51,471,400 and 51,554,406 raw reads, respectively. After filtering for high quality sequences, 44,362,142 and 45,610,802 clean reads were obtained for each sample (Table 1). Subsequently, the clean reads from the two samples were combined to do de novo assembly, resulting in 69,836 contigs with a mean length of 1121 bp and an N50 of 2407 bp. The lengths of contigs ranged from 200 bp to over 25,000 bp. After clustering, these contigs generated 55,124 unigenes (Fig. 1a). From these unigenes, a total of 18,574 predicted ORFs was obtained and the lengths ranged from 297 bp to 15,858 bp, with a mean length of 1049 and an N50 of 1491 (Fig. 1b). To annotate the unigenes, we searched reference sequences against NT, NR, PFAM and Uniprot databases, using Trinotate and Blast with a cut-off E-value of 10^− 5. 21,259 (38.57%) unigenes were annotated at least in one database, including 6623 in NT, 18,556 in NR, 12,457 in BLASTP and 17,772 in BLASTX against Uniprot (Fig. 2a). On the basis of NR annotation, 15.53% (n = 2882) of unigenes matched to Zootermopsis nevadensis, followed by Acyrthosiphon pisum: 8.40% (n = 1560); Ceratitis capitata: 5.53% (n = 1026); Diaphorina citri: 5.15% (n = 956); Tribolium castaneum: 4.07% (n = 755) (Fig. 2b).

Subsequently, the differentially expressed genes (DEGs) between viruliferous and nonviruliferous guts were identified using DEGseq [20]. A total of 5207 differentially expressed genes were identified, with 4014 genes upregulated and 1193 genes downregulated in viruliferous whitefly guts (Fig. 3a). The detected fold changes (log₂ ratio) of gene expression ranged from − 13.31 to 6.45, and more than 90% of the genes (4696) were up- or downregulated between 1.0- and 4.0-fold (Fig. 3b).

To further reveal their functions, GO assignments were used to classify the DEGs. A total of 1216 DEGs (840 upregulated and 376 downregulated) were categorized into 54 secondary GO categories under the ‘biological process’, ‘cellular component’ and ‘molecular functions’ divisions. While ‘binding’ and ‘catalytic’ were among the most represented ‘molecular functions’ categories, the ‘cellular component’ most represented were ‘cell part’ and ‘membrane part’. Interestingly, in ‘biological process’ category, viral infections dramatically changed the expression of genes in ‘response to stimulus’, ‘reproductive process’ and ‘immune system process’ (Fig. 4). The GO enrichment analysis showed that ‘Cargo receptor activity’ was significantly enriched with DEGs in ‘molecular functions’ (11 out of 30 genes). Exactly 6 genes were upregulated while a gene was downregulated in ‘scavenger receptor activity’ under the ‘cargo receptor activity’ (Table 2).

Among the 5207 DEGs, 414 genes were mapped to 266 pathways in KEGG database. While ‘metabolic pathways’ (169), ‘biosynthesis of secondary metabolites’ (50) and ‘microbial metabolism in diverse environments’ (37) contained most of the DEGs. Enrichment analysis demonstrated that the DEGs were only significantly enriched in ‘ATP-binding cassette (ABC) transporters’, with 16 genes upregulated and 17 genes downregulated in viruliferous whitefly guts.

Based on GO and KEGG analysis, a total of 31 genes (24 upregulated and 7 downregulated) involved in cell cycle were found in our DEGs (Table 3). For DNA repair, 2 genes were upregulated and 6 genes downregulated in viruliferous whitefly guts (Table 4). Additionally, 4 genes of cell cycle checkpoints were found in DEGs and all of them were upregulated (Table 4). We also found that 10 upregulated genes were related to ubiquitin (Table 4).

Interestingly, among the DEGs in viruliferous whitefly guts, 14 genes related to cellular and humoral immune response were upregulated (Table 5). Six genes encoding AMPs were significantly upregulated including 1 defensin and 4 knottin. For cellular responses, a total of 8 upregulated genes were found, 6 of them involved in the phagocytosis and 2 in encapsulation. In addition, 9 genes (6 upregulated and 3 downregulated) were altered in the lysosome pathway (Additional file 1: Table S1). For regulators of defense responses, 1 gene encoding secreted molecule and 9 genes encoding putative transmembrane receptors (surface receptors) were found in our DEGs (Table 5). Meanwhile, the intracellular signaling molecules or pathways, downstream of these receptors, were also activated. In viruliferous whitefly gut cells, a number of genes involved in the mitogen-activated protein kinase (MAPK), Notch, transforming growth factor beta (TGF-β), PI3K-Akt, Jak-STAT, protein kinase C (PKC), and Ras signaling molecules or pathways were upregulated (Table 6).

To validate our data, the expression levels of 10 DEGs were verified using qRT-PCR. We analyzed the expression of 5 antimicrobial peptides (Btk-1, Btk-2, Btk-3, Btk-4, DEF), 1 scavenger receptor class B (CD36), 1 regucalcin (RGN), 1 ankyrin repeat protein (Ank), 1 E3 ubiquitin-protein ligase, and 1 unnamed protein (categorized into GO biological process, defense response to virus, DRV) in both guts and whole body. Six genes were confirmed to be significantly upregulated in guts samples (Fig. 5). Moreover, the fold changes obtained by the DEG were generally more extreme than those obtained by qRT-PCR. The inconsistency may be partially due to the lower sensitivity of qRT-PCR compared tosignificant DEG. In addition, the fold changes in guts were more significant than those in whole body. These results indicate that the transcriptome profiles of whitefly guts in response to TYLCV infection are a little different from the whitefly whole body. Nevertheless, qRT-PCR analyses confirmed the direction of change detected by DEG analysis, suggesting that DEG results are reliable.

The GO and KEGG enrichment analysis showed that DEGs were significantly enriched in ‘Cargo receptor activity’ and “ATP-binding cassette (ABC) transporters”. Cargo receptor can selectively bind to an extracellular substance and deliver it into cell via endocytosis [21]. For example, cargo receptor ERGRC-53 is required for cellular glycoprotein trafficking of arenavirus, hantavirus, coronavirus, orthomyxovirus, and filovirus [22]. In human, the scavenger receptor class B type I and scavenger receptor B2 have been reported to be a receptor for the hepatitis C virus and enterovirus 71 separately [23, 24]. ABC transporters are members of a transport system superfamily and involved in diverse cellular processes such as maintenance of osmotic homeostasis, nutrient uptake, resistance to xenotoxins, antigen processing, bacterial immunity and pathogenesis [25]. Interestingly, one ABC superfamily transporter, multidrug resistance protein1 (MRP1), which is well-known to confer multidrug resistance to cancer cells through enhanced drug efflux [26], was also found in our DEGs. (c18467_g1: multidrug resistance-associated protein 1-like [Diaphorina citri], log₂ ratio = 1.87) In addition, the downregulation of MRP2, which shows structural similarity to MRP1, suppresses the transport of various genotoxic substances in the liver during hepatitis virus infection [27]. As we know, virus receptors play essential roles in the early steps of viral infection. These regulated receptors and transporters might help TYLCV cross the whitefly epithelial cells.

Viruses depend on hosts machineries to replicate and express their genomes, and altering the host cell cycle is a common strategy of viruses inside the cell [28]. Perturbation of host cell cycle has been shown to be caused by geminivirus infection. For example, Cabbage leaf curl virus (CaLCuV) can activate genes expressed during S and G₂ phases and inhibit genes active in G1 and M phases [29]. We noticed that one gene encoding cyclin D2 was upregulated in viruliferous guts. Cylin D2 (CCND2) is a member of the D-type cyclin family, which associates the extracellular signaling environment with cell-cycle progression [30]. Moreover, cyclin D2 was also upregulated in Hepatitis B virus-infected cells and has a positive role in HBV replication [31].

Virus infection can produce a large amount of exogenous DNA [32]. Several mammalian viruses have been shown to induce a cellular DNA damage response during replication, and in some cases, this response is required for optimal virus replication [33]. In addition, cell cycle checkpoints govern cell cycle progression in the presence of DNA damage and incomplete DNA replication [34]. The upregulated genes in cell cycle checkpoints suggest that TYLCV infection may induce DNA damage and affect cell cycle in whitefly gut cells. Interestingly, we also noted that all of the 10 upregulated genes encoding ubiquitin regulated E3 ubiquitin ligase activity. Furthermore, 5 of these genes accumulated in the G2 phase of the cell cycle. G2/M phase-specific E3 ubiquitin-protein ligase (G2E3) as a nucleo-cytoplasmic shuttling protein was implicated in the DNA damages response and cell cycle regulation [35], and then attenuated replicative stress [36]. Previous study also showed that Beet severe curly top virus (BSCYV) can induce RKP (a functional ubiquitin E3 ligase), which is able to interact with cell cycle inhibitor ICK/KRP proteins to regulate the cell cycle [37].

In our DEGs, 6 genes encoding AMPs were significantly upregulated. AMPs are evolutionarily ancient weapons that resist bacteria, fungi, viruses and any conceivable substance [38]. Previous study has shown that knottin genes are responsive to various stresses and related to TYLCV circulative transmission in the whitefly [10, 12]. Therefore, the observed upregulation of genes encoding the knottin proteins in our study are consistent with previous reports. For cellular responses, phagocytosis is a widely conserved cellular response and refers to the recognition, engulfment and intracellular destruction of invading pathogens and apoptotic cells [39]. Previous study in our laboratory has demonstrated that lysosome participate in autophagy, which could inhibit the efficiency of TYLCV transmission by whiteflies [11]. As stated above, all genes involved in humoral and cellular responses and most genes associated with lysosome were significantly upregulated, strongly suggesting that innate immune systems were activated in viruliferous whitefly guts.

For regulators of defense responses, the beta 1, 3-glucan recognition protein (βGRP), which can lead to melanotic encapsulation in insects [40], was upregulated in viruliferous gut. This result suggested that encapsulation might be activated after TYLCV infection. The altered genes encoding surface receptor were classified into three major classes; a) CD36 family (n = 4), b) epidermal growth factor (EGF)-like (n = 3), and c) Toll receptor (n = 2). CD36 family is class B scavenger receptor that participates in apoptotic cell removal in Drosophila embryos and in phagocytosis of S. aureus [41]. Interestingly, scavenger receptor has been noticed in GO enrichment analysis, but just 1 out of 4 of these CD36 genes was categorized into scavenger receptor activity. The EGF, a transmembrane protein, mediates phagocytosis, has been demonstrated with knockdown experiments, in a broad range of bacteria in Drosophila [42]. Spaetzle (markedly upregulated) is the ligand for Toll receptor, leading to Toll signally pathway controlling the potent resistances to bacterial, fungal, and viral infections [43]. Toll is one of the major signal transduction pathways (another one is Imd), used by antimicrobial responses in Drosophila [44].

The above results clearly show that following TYLCV infection, certain of immune responses of whitefly including humoral (antibacterial peptides) and cellular (phagocytosis, encapsulation) were activated. The regulators of phagocytosis (CD36, EGF) and encapsulation (βGRP, Toll receptor) were also activated. These immune responses may result in the degradation of virions, therefore, the activation of immune responses is probably the evolved strategy of the whitefly to protect itself from the deleterious effects of TYLCV infection.

We also observed that a number of genes involved in signal transduction of the defense response were upregulated after TYLCV invasion. Ras is a key regulator of cell growth in all eukaryotic cells and positioned centrally in signal transduction pathways that respond to diverse extracellular stimuli [45]. The medfly homologues of Ras proteins have been shown to affect phagocytosis in response to lipopolysaccharide (LPS) and E. coli [46]. PI3K-Akt pathway is involved in apoptosis and autophagy in Drosophila [47, 48], as well as in medfly apoptosis [49]. Decreases in phagocytosis produced by pathway inhibitor indicated that PI3-kinase was required for internalization of bacteria [50]. Several studies have implicated involvement of MAPKs in insect innate immune responses [51, 52], and the activation of MAPK pathway contributes to efficient infection by baculoviruses [53]. Otherwise, several studies have documented that PKC plays an important role in regulating the receptor-mediated endocytosis of virus-receptors complexes [54‐56], and specific inhibitor for PKC is highly effective in blocking West Nile virus (WNV) entry into mosquito cell line (C6/36) [57]. Here we found that all genes involved in the MAPK pathway and PKC were upregulated in DEGs.

The intracellular signaling molecules and pathways are organized as communication networks that process, encode and integrate internal and external signals. For example, activated Ras can directly interact with MAPK kinase kinase and activates distinct MAPK cascades (ERK, JNK, p38) [58]. Upon activation, the ERKs can stimulate the activity of Elk-1 transcription factor [59]. Blockage of Elk-1 like protein phosphorylation indicated that an Elk-1 like protein was candidate for the regulation of phagocytosis of bacteria [60]. Moreover, JNK was also reported to be involved in phagocytosis in mosquito cell line [61]. In our results, Ras, MAPKK7, JNK, Elk-1, even phagocytosis were upregulated. Evidently, this signaling network in whiteflies was activated by TYLCV.

The culture of B. tabaci MEAM1 (mtCO1 GenBank accession no. GQ332577) was maintained on cotton (Cossypium hirsutum cv. Zhe-Mian 1793), which is a non-host plant of TYLCV. Whiteflies were reared in a climate chamber at 26 ± 1 °C, LD 14:10 h and 60 ± 10% relative humidity. To obtain virus-infected plants, clones of TYLCV isolate SH2 (GenBank accession no. AM282874.1) were inoculated into tomato (Solanum lycopersicom L.cv. Hezuo903) at 3–4 true-leaf stage as previously described [62], TYLCV-infected and uninfected tomato plants were cultivated to 6–7 true-leaf stage when used in experiments. All plants were grown in a greenhouse at 20–30 °C, LD14:10 h and 50–70% relative humidity.

To prepare viruliferous and nonviruliferous whiteflies, approximately 2000 newly emerged (0–48 h) adult whiteflies were separately collected and released onto healthy tomato plants in different cages for 48 h. Then the whiteflies were transferred onto virus-infected and uninfected tomato plants for 48 h, respectively. After that, they were separately transferred onto another tomato (Solanum lycopersicom L.cv. Zheza502), which is a resistant variety to TYLCV [11], to eliminate effects of host differences on whiteflies. After 24 h, the whiteflies were used for gut dissection. For sample collection, approximately 1000 guts were dissected from viruliferous and nonviruliferous whiteflies in PBS respectively, and frozen in liquid nitrogen, stored at − 80 °C until RNA isolation. Additionally, all the guts were dissected in 36 h to minimize the impact of time on the results.

Total RNA was extracted from pools of approximately 1000 guts using the Absolutely RNA Nanoprep kit (Agilent, USA) according to the manufacturer’s manual with slight modifications [63]. RNA quantity and quality were assessed using 1% agarose gels, Qubit® 3.0 Flurometer (Life Technologies, CA, USA) and the RNA Nano 6000 Assay Kit of the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). RNA (700 ng per sample) was used to generate adaptor-ligated double- stranded cDNA libraries for RNA-Seq using the NEBNext® UltraTM RNA Library Prep Kit for Illumina® (New England Biolabs, USA) following the manufacturer’s protocol. The fragment size and concentration of resultant libraries were carried out using the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA) and StepOnePlus™ Real-Time PCR System (Library valid concentration > 10 nM).

The cDNA libraries were sequenced for 150 bp paired-end reads using Illumina Hiseq 4000 platform at Annoroad Gene Technology Company (Beijing, China). The total sequencing amount was 8 G for each library. To ensure the quality of data used in further-analysis, raw data were transformed into clean data through removing adaptor-polluted reads (more than 5 adaptor-polluted bases), low-quality reads (more than 15% bases with Phred threshold score ≤ 19), and reads with unknown sequences ‘N’ accounting for more than 5% using Perl scripts. As for paired-end sequencing data, both reads were filtered out if any read of the paired-end reads are adaptor-polluted. The assembler Trinity was used for de novo assembly [64]. We quantified the proportion of the clean reads that mapped to each assembly using Bowtie (v2.2.3). To test the completeness of the assemblies, we mapped the presence of highly conserved genes with 2748 core proteins from conserved regions of eukaryotes. TransDecoder was used to identify the candidate coding regions within transcript sequences, and Trinotate was used for performing the functional annotation of unigenes and ORFs.

The clean reads from viruliferous and nonviruliferous guts were separately mapped back to the assembled unigenes. For gene expression analysis, reads per kilobase million Mapped Reads (RPKM) was calculated to estimate the expression level of genes in each sample. RPKM can eliminate the effect of sequencing depth and gene length on gene expression levels, which facilitates the comparison of the number of transcript levels generated between samples. DEGseq (v1.18.0) was used to identify differentially expressed genes (DEGs) between the viruliferous and nonviruliferous guts. Genes with q ≤ 0.05 (adjusted p-value) and log₂ ratio ≥ 1 were considered differentially expressed.

To obtain an overview and further understand the biological functions of genes, all DEGs were subjected to GO functional annotation using Blast2GO and mapped to terms in KEGG pathway database using KOBAS. Enrichment analysis was used to identify the GO terms and significantly regulated KEGG pathways. We selected a corrected q ≤ 0.05 as the threshold to determine significant enrichment of the gene sets.

To confirm the result of the DEG analyses, we measured the expression of 10 selected genes using comparative C_T (ΔΔC_T) Real-time Quantitative PCR with β-actin as the internal control gene. We prepared viruliferous and nonviruliferous whiteflies as described above and gut total RNA was extracted from pools of approximately 70 guts. Three biological replicates were conducted at the same time. Moreover, 25 whiteflies were collected (24 h after beginning of the gut dissection) from each replication to compare the differences in gene expression between the gut and the whole body. Total RNA was extracted using the TRI-reagent method (Life Technologies, USA). RNA was reverse transcribed using the SYBR® PrimeScript™ RT-PCR Kit II (Takara, Japan). qRT-PCR was performed using PTC-200 Thermocycler (Bio-Rad, USA) with SYBR-Green detection (SYBR® Premix Ex Taq™ II, Takara, Japan). The primer sequences are provided in Additional file 2: Table S2.