RNA-sequencing and statistics of transcript expression
To explore the molecular mechanism of testicular fusion during the larva-to-pupa metamorphosis in S. litura, the samples of the testes were collected at L6D4, L6D6 and P4D (Fig. 1). The mRNA was extracted and sequenced using Illumina HiseqTM 2000 platform. In total, 4,780,380, 4,891,555 and 4,854,931 clean reads were obtained respectively for three groups of samples (Addition file 1). These clean reads were mapped to reference sequences from S. litura genome database  using TOPHat software . At L6D4, L6D6 and P4D stages 4,524,698, 4,602,274 and 4,594,593 unique reads were mapped and the mapping ratios were higher than 94% (Addition file 2). All the mapped clean reads were assembled by Cufflinks software  and the known genes from S. litura genome were obtained. Novel transcripts were detected using Cuffcompare. There were 12,339 transcripts expressing in all the three samples, while 9,856 transcripts were co-expressed at the three stages, and 10,995, 11,389 and 10,879 transcripts were specifically expressed at L6D4, L6D6 and P4D stages, respectively (Fig. 2).
The expression levels of transcripts were calculated by using the Fragments Per Kilobase of transcript per Million mapped reads (FPKMs) [30–31]. According to FRKM, the differentially expressed genes (DEGs) were obtained comparing the samples between L6D4 and L6D6 (L6D4 vs L6D6), and between L6D6 and P4D (L6D6 vs P4D) by using the edgeR platform (http://www.bioconductor.org/packages/release/bioc/html/edgeR.html) . In total, 1,676 transcripts were identified as DEGs for L6D4 vs L6D6, of which 1,387 were up-regulated and 289 were down-regulated. For L6D6 vs P4D, 3,365 transcripts showed differential expression levels, of which 884 were up-regulated and 2,481 were down-regulated (Fig. 3). The results indicated a large number of the transcripts were up-regulated during the testis fusion, but after the fusion, the major of DEGs were down-regulated. It is speculated that those transcripts that were expressed in such a trend may involve in the testis fusion.
Expression of the 20E and JH signaling pathway related genes during the testicular fusion
The testicular fusion occurs during the larva-pupae metamorphosis, which is regulated by 20-hydroxyecdysone (20E) and juvenile hormone (JH). In this study, the 20E signaling pathway related genes USP, EcR, HR38, βFTZ-F1, BRC-Z2, BRC-Z3 and BRC-Z4 were highly expressed at L6D6 and SRC and Kr-h1 in the JH signaling pathway were up-regulated at L6D6 (Fig. 4A, B). The results implied that the 20E signaling pathway may mainly regulate or initiate the testicular fusion, while JH signaling may also involve in the process.
Expression of transcription factors (TFs) during the testicular fusion
Transcription factors are important for the growth, development and differentiation of organisms. By using Pfam (http://pfam.xfam.org/), a total of 187 transcripts with potential transcription factor activity were predicted in DEGs, 12 of which were specifically highly expressed at L6D6 (when the testicular fusion occurred) including the transcription factors Ken1, chorion specific C/EBP, βFTZ-F1, spalt, Elbow, Ets, trachealess, kayak (Fig. 5A). This expression pattern was verified by qRT-PCR (Fig. 5B). It is hypothesized that highly expressed transcription factors at L6D6 may be involved in testicular fusion.
Expression of the cytoskeleton and chitin metabolism related genes during the testicular fusion
The testicular fusion occurs during the larva-to-pupa metamorphosis, during which many of the tissues and organs are remodeled. The cytoskeletal proteins and insect chitin are involved in the remodeling of the tissue and organ [33–34]. To study whether cytoskeleton and chitin are involved in testicular fusion, the expression of these genes was analyzed. The analysis found that the cytoskeletal proteins, including actin, tubulin, actin binding protein, microtubule-associated proteins, were up-regulated from L6D4 to L6D6, and down-regulated from L6D6 to P4D (Fig. 6A). The expression pattern of chitinases, chitin deacetylases, chitin synthase A was similar to that of the cytoskeletal proteins (Fig. 6B). The results indicated that the testis fusion during the larva-to-pupa metamorphosis involved in testis cell remodeling through the roles of cytoskeletal proteins and chitin metabolism-related proteins.
Expression of ECM component proteins
Extracellular matrix (ECM) proteins usually consist of collagens, proteoglycans and glycoproteins and these proteins function in physical support for tissue integrity and elasticity, as well as the microenvironment of a cell, which influences cell behaviors, such as cell proliferation, adhesion and migration . To study the roles of the ECM components in the testicular fusion, the expression of collagens, proteoglycans and glycoproteins were analyzed. Most of these genes were up-regulated at L6D6, when the testicular fusion occurred (Fig.7). Two collagen genes (SWUSl10005600, XM_022978189.1, collagen alpha–1(IV) chain; SWUSl10005599, XM_022978190.1, collagen alpha–2(IV) chain) and the three glycoprotein laminin genes (SWUSl10014957, XM_022976705.1, laminin subunit gamma–1; SWUSl10015737, XM_022973420.1, laminin subunit beta–1; and SWUSl10010299, XM_022964712.1, laminin subunit alpha–1) were selected for qRT-PCR verification (Fig. 8), the results revealed that ECM component proteins may be involved in the testicular fusion in S. litura.
Expression of ECM-associated proteins: mucins and lectins
In this study, ECM-associated proteins: mucins and lectins were analyzed, these genes were up-regulated significantly during the testicular fusion (from L6D4 to L6D6), and down-regulated after the testicular fusion (from L6D6 to P4D) (Fig. 9). The expression pattern of these genes is consistent with that of the ECM component proteins, implying that the ECM-associated proteins and ECM component proteins may work together in the testicular fusion in S. litura.
Expression of ECM receptor integrins
The ECM-receptor interaction plays important roles in controlling cytoskeletal dynamics and regulating diverse functions including cell survival, differentiation, migration, attachment, focal adhesion assembly . In this study, the ECM receptors integrins were significantly up-regulated at L6D6 as detected by both transcriptome and qRT-PCR analyses (Fig. 10). These results implied that all of the ECM receptor integrins and ECM component proteins, ECM-remodeling enzymes, ECM-related proteins may jointly participate in the testicular fusion in S. litura.
Expression and function analysis of ECM-remodeling enzymes: Matrix metalloproteinases (MMPs)
In this study, the genes related to the ECM remodeling were found and analyzed, MMPs and ADAMs were up-regulated significantly at L6D6 (Fig. 11A). To study the function of ECM-remodeling enzymes, MMPs were selected because three MMP transcripts were found in the testes (SWUSl10007900, XM_022966048.1, matrix metalloproteinase–14 isoform X3; SWUSl10005499, XM_022958284.1, matrix metalloproteinase–2-like and SWUSl10009480, XM_022964227.1, matrix metalloproteinase–2-like). Two Slmmps were significantly up-regulated during the testicular fusion process (Fig. 11B). Phylogenetic tree analysis was performed for the three MMP proteins. These three MMP proteins of S. litura were highly homologous to the Bombyx mori MMPs, BmMMP1, BmMMP2 and BmMMP3, respectively (Additional file 3). Thus, MMP SWUSl10007900, SWUSl10005499 and SWUSl10009480 were named SlMMP1, SlMMP2 and SlMMP3, respectively, in this study.
To study the relationships between SlMMP proteins and the testicular fusion, the expression patterns of Slmmps were tested by qRT-PCR, the results indicated that Slmmp1 was not differentially expressed, while Slmmp2 and Slmmp3 were up-regulated significantly during the testicular fusion (Fig. 11B), which was consistent with the transcriptomic analysis data. To study the location of SlMMPRNA in the testis, the peritoneal sheath and sperm cellswere isolated from the testis, the results indicated that Slmmp2 and Slmmp3 had a higher expression level in the peritoneal sheath than in sperm cells (Fig. 11C), while Slmmp1 was not significantly differentially expressed in the peritoneal sheath and sperm cells (Fig. 11C). These results implied that SlMMP1 may not be related to the testicular fusion, whereas SlMMP2 and SlMMP3 may play roles in the peritoneal sheath for testicular fusion.
To study the function of the SlMMPs during the testicular fusion, a broad-spectrum inhibitor, GM6001, which is found to inhibit MMP function in some studies [36–37], was applied to inhibit the SlMMPs action. GM6001 was injected into the 5th abdominal segment near the testis at the early L6D6 stage and then the testicular fusion process was statistically analyzed at the middle L6D6, late L6D6 stage, white pupae stage (P0) and 1-day-old pupae stage (P1D) (Fig. 12 and Table 1). The results indicated that almost 57% larvae were not able to develop into normal pupae and out of these larvae, 50% individuals had their testes separated and were not able to fuse (Table 1), as compared to the control DMSO, in which only 20% larvae did not develop into normal pupae. In all of the normal control pupae, the testis was able to get to close, adhere and fuse, however the fusion did not happen in the both normal and abnormal treatment pupae, (Table 1). Thus, these results implied that SlMMPs play a vital role in the testicular fusion.