TGFβ signaling related genes are involved in hormonal mediation during termite soldier differentiation

Abstract

A working knowledge of the proximate factors intrinsic to sterile caste differentiation is necessary to understand the evolution of eusocial insects. Genomic and transcriptomic analyses in social hymenopteran insects have resulted in the hypothesis that sterile castes are generated by the novel function of co-opted or recruited universal gene networks found in solitary ancestors. However, transcriptome analysis during caste differentiation has not been tested in termites, and evolutionary processes associated with acquiring the caste are still unknown. Termites possess the soldier caste, which is regarded as the first acquired permanently sterile caste in the taxon. In this study, we performed a comparative transcriptome analysis in termite heads during 3 molting processes, i.e., worker, presoldier and soldier molts, under natural conditions in an incipient colony of the damp-wood termite Zootermopsis nevadensis. Although similar expression patterns were observed during each molting process, more than 50 genes were shown to be highly expressed before the presoldier (intermediate stage of soldier) molt. We then performed RNA interference (RNAi) of the candidate 13 genes, including transcription factors and uncharacterized protein genes, during presoldier differentiation induced by juvenile hormone (JH) analog treatment. Presoldiers induced after RNAi of two genes related to TGFβ (Transforming growth factor beta) signaling were extremely unusual and possessed soldier-like phenotypes. These individuals also displayed aggressive behaviors similar to natural soldiers when confronted with Formica ants as hypothetical enemies. These presoldiers never molted into the next instar, presumably due to the decreased expression levels of the molting hormone (20-hydroxyecdysone; 20E) signaling genes. These results suggest that TGFβ signaling was acquired for the novel function of regulating between JH and 20E signaling during soldier differentiation in termites.

Fig 1. Sampling points for RNA-seq analysis and gene expression profiles during each molting process.

(A) Caste differentiation observed in an incipient colony of Zootermopsis nevadensis. The oldest 3rd instar larva (No. 1 larva) molts into a presoldier, whereas the next-oldest 3rd instar larva (No. 2 larva) molts into the next instar (4th instar larva). Gut-purged individuals are always observed 3 days before the molts. Individuals were sampled at the following 5 developmental stages during each molt; pre-gut-purging (pGP), 0 days after gut-purging (GP0), 3 days after gut-purging (GP3), 0 days after the molt (M0) and 3 days after the molt (M3). (B) MDS plots of 14,204 genes detected by the RNA-seq data from the head during 3 molts in Zootermopsis nevadensis. (C) Numbers of caste-specific highly expressed genes among 3 timings in each molt (pGP, GP0, GP3). (D) Clusters of genes with highly expression levels in the head before the presoldier molt. Cluster 1 was composed of 54 genes showing no clear expression patterns with the molting events. Cluster 2 was composed of 26 genes with highly expression levels after the molt (M0) in each molting event. Cluster 3 and 4 were composed of 32 and 19 genes with highly expression levels before the molt (GP0 and GP3, respectively). The solid black line in each panel indicates the average expression pattern. The similarity was calculated based on the Jensen-Shannon distances using the RPKM (Reads Per Kilo base of exon model per Million mapped reads) values.

Fig 2. RNAi effects and expression patterns of ZnSox11 and Znev_01548.

(A) Phenotypes of GFP, ZnSox11 and Znev_01548 dsRNA injected presoldiers. Each dsRNA was injected into the side of the thorax 24 hours after the JH analog (JHA) application. The emerged presoldiers were photographed 7 days after the molt. Approximately half of the GFP dsRNA injected presoldiers (14 out of 26, 53.8%) molted into soldiers with normal phenotypes (right). Scale bar indicates 5 mm. (B) Aggression levels of GFP, ZnSox11 and Znev_01548 dsRNA injected presoldiers (n = 6–9). The levels of natural soldiers are also shown on the right (n = 9). Numbers of individuals examined are shown in parentheses. The boxes and whiskers mean median, quartiles and range. Different letters over the bars indicate significant differences in each category (Kruskal-Wallis test: P = 6.35E-04, Steel-Dwass test: P < 0.05). (C) Expression patterns of ZnSox11 and Znev_01548 in the head provided by the qPCR analysis during each molting process in an incipient colony. Relative expression levels (mean ± S.E., biological triplicates) were calibrated by the expression levels in workers (GP0) as 1.0. The statistical results of three-way ANOVA are described in each box (**P < 0.01). The data is consistent with the use of parametric statistics by the Browne-Forsythe test (ZnSox11: P = 6.37E-01 (worker), 5.08E-02 (presoldier), 7.75E-01 (soldier); Znev_01548: P = 6.77E-01 (worker), 7.45E-01 (presoldier), 9.08E-01 (soldier)) prior to the use of the ANOVAs. (D) Expression patterns of ZnSox11 and Znev_01548 under the ZnMet RNAi treatment. Gray and red lines indicate the results under the GFP and ZnMet RNAi treatments, respectively. Relative expression levels (mean ± S.E., biological triplicates) were calibrated by the expression level of intact worker (day 0) as 1.0. The statistical results of two-way ANOVA are described in each box (*P < 0.05, **P < 0.01). The data is consistent with the use of parametric statistics by the Browne-Forsythe test (ZnMet: P = 7.91E-01 (GFP), 5.90E-01 (ZnMet RNAi); ZnSox11: P = 4.99E-01 (GFP), 8.43E-01(ZnMet RNAi); Znev_01548: P = 6.46E-01 (GFP), 6.25E-01 (ZnMet RNAi)) prior to the use of the ANOVAs.

Fig 3. Gene expression during presoldier molt induced by JHA treatment under ZnSox11 and Znev_01548 RNAi treatment.

(A) Expression patterns in the whole body of ZnSox11 and Znev_01548, (B) soldier characteristic genes and (C) 20E signaling genes. Gray, red and blue lines indicate the results under the GFP, ZnSox11 and Znev_01548 RNAi treatments, respectively. Relative expression levels (mean ± S.E., biological triplicates) were calibrated by the expression level of GFP dsRNA injected workers (-10) as 1.0. The statistical results of two-way ANOVA are described in each box (*P < 0.05, **P < 0.01). Asterisks over the bars indicate significant differences in each gene compared with the GFP control (Scheffe’s F test: *P < 0.05, **P < 0.01). The data is consistent with the use of parametric statistics by the Browne-Forsythe test (ZnSox11: P = 5.08E-01 (GFP), 5.73E-01 (ZnSox11 RNAi); Znev_01548: P = 7.39E-01 (GFP), 7.94E-01 (Znev_01548 RNAi); ZnTro: P = 9.78E-02 (GFP), 5.17E-01 (ZnSox11 RNAi), 3.86E-01 (Znev_01548 RNAi); ZnLac2: P = 1.40E-01 (GFP), 6.12E-01 (ZnSox11 RNAi), 2.86E-01 (Znev_01548 RNAi); ZnEcR: P = 5.82E-01 (GFP), 8.62E-01 (ZnSox11 RNAi), 4.42E-01 (Znev_01548 RNAi); ZnE75: P = 2.97E-01 (GFP), 4.90E-01 (ZnSox11 RNAi), 7.91E-01 (Znev_01548 RNAi)) prior to the use of the ANOVAs.

Fig 4. A schematic model on the role of TGFβ signaling for soldier differentiation.

When JH titer is increased in larvae, TGFβ signaling may be involved in the formation of presoldier-specific traits (e.g. soft cuticle, low aggression) and the regulation of soldier molt via 20E signaling.