Somatic, germline and sex hierarchy regulated gene expression during Drosophila metamorphosis

Background Drosophila melanogaster undergoes a complete metamorphosis, during which time the larval male and female forms transition into sexually dimorphic, reproductive adult forms. To understand this complex morphogenetic process at a molecular-genetic level, whole genome microarray analyses were performed. Results The temporal gene expression patterns during metamorphosis were determined for all predicted genes, in both somatic and germline tissues of males and females separately. Temporal changes in transcript abundance for genes of known functions were found to correlate with known developmental processes that occur during metamorphosis. We find that large numbers of genes are sex-differentially expressed in both male and female germline tissues, and relatively few are sex-differentially expressed in somatic tissues. The majority of genes with somatic, sex-differential expression were found to be expressed in a stage-specific manner, suggesting that they mediate discrete developmental events. The Sex-lethal paralog, CG3056, displays somatic, male-biased expression at several time points in metamorphosis. Gene expression downstream of the somatic, sex determination genes transformer and doublesex (dsx) was examined in two-day old pupae, which allowed for the identification of genes regulated as a consequence of the sex determination hierarchy. These include the homeotic gene abdominal A, which is more highly expressed in females as compared to males, as a consequence of dsx. For most genes regulated downstream of dsx during pupal development, the mode of regulation is distinct from that observed for the well-studied direct targets of DSX, Yolk protein 1 and 2. Conclusion The data and analyses presented here provide a comprehensive assessment of gene expression during metamorphosis in each sex, in both somatic and germline tissues. Many of the genes that underlie critical developmental processes during metamorphosis, including sex-specific processes, have been identified. These results provide a framework for further functional studies on the regulation of sex-specific development.


Statistical analyses of microarray data
All arrays were scanned using the GenePix 4100A scanner and GenePix Pro 5.0 software from Axon Instruments (Molecular Diagnostics, Sunnyvale, CA). Visual inspection of the microarray images filtered out florescence most likely not due to labeled cDNA binding; the data from these array elements was flagged as absent. Array elements were only considered for further analysis if at least one channel (Cy3 or Cy5) had greater than 75% of the pixels with intensity values one standard deviation above background levels (columns "B532+1SD" and "B635+1SD" in gpr file, respectively) and if they were not flagged by either GenePix Pro or visual inspection. All microarray normalization and statistical analyses were performed using the LIMMA package of BioConductor in the program R [1][2][3][4]. Global-loess normalization was used for all arrays, and significance was converted to q values using the q value application for R [5]. In the analyses, the design matrix for the analyses labeled the microarray experiments in the following manner: wild type males from 0, 24, 48, 72, and 96 hour APF were labeled as wtM_0hr, wtM_24hr, wtM_48hr, wtM_72hr, and wtM_96hr, respectively. Similar labeling was used for wild type females, tud progeny males, and tud progeny females.
Two-hundred-fifty-eight genes were thus identified as showing sex-differential expression in the somatic tissues during metamorphosis. To determine how these genes were sex-differentially expressed at the five time points examined in this study (0, 24, 48, 72, and 96 hour APF), moderated t-tests were performed in R comparing the mean expression in tud progeny males and females at each time point separately. The resulting P values of the 258 genes were converted to q values using the q value application for R [5], and significance was declared at a q<0.15 level.

Identification of genes expressed in the male or female germlines
To identify genes with expression differences due to the presence of a germline, Fstatistics were also implemented using contrasts in LIMMA. Again, contrast matrices for the appropriate analysis were created and then fitted to the model. The P values for the resulting Fstatistic were then converted into q values. Genes were first identified with sex-differential expression between wild type males and females using the following contrast design: [FvM0=wtF_0hr-wtM_0hr, FvM24=wtF_24hr-wtM_24hr, FvM48=wtF_48hr-wtM_48hr, FvM72=wtF_72hr-wtM_72hr, FvM96=wtF_96hr-wtM_96hr]. Genes with significant (q<0.15) and at least a 1.2 fold-change expression differences between the sexes were kept, leaving 3194 genes to be analyzed (2320 and 883 female-and male-biased genes, respectively; nine genes have one male-biased isoform and one female-biased isoform).
To identify genes expressed in the male germline, genes with significant expression differences in males according to the genotype factor (wild type vs. tud progeny) were determined by the following contrast design: [wtVtud0hr=wtM_0hr-tudM_0hr, wtVtud24hr=wtM_24hr-tudM_24hr, wtVtud48hr=wtM_48hr-tudM_48hr, wtVtud72hr=wtM_72hr-tudM_72hr, wtVtud96hr=wtM_96hr-tudM_96hr]. Male-biased genes (of the 883 found above) with significant (q<0.15) and at least a 1.2 fold-change higher expression in wild type males as compared to tud progeny males were declared as being expressed in the male germline (586 genes). In addition, to avoid false negatives, genes were included in the male germline that were expressed in at least four time points in wild type males and that had no expression in wild type females and tud progeny males at all five time points examined in this study. Seventy-three additional genes were thus included. See Additional file 8 for list of the 659 genes expressed in or as a consequence of the male germline.
To identify genes expressed in the female germline, genes with significant expression differences in females according to the genotype factor (wild type vs. tud progeny) were determined by the following contrast design: [wtVtud0hr=wtF_0hr-tudF_0hr, wtVtud24hr=wtF_24hr-tudF_24hr, wtVtud48hr=wtF_48hr-tudF_48hr, wtVtud72hr=wtF_72hr-tudF_72hr, wtVtud96hr=wtF_96hr-tudF_96hr]. Female-biased genes (of the 2320 found above) with significant (q<0.15) and at least a 1.2 fold-change higher expression in wild type females as compared to tud progeny females were declared as being expressed in the female germline (342 genes). In addition, to avoid false negatives, genes were included in the female germline that were expressed in at least four time points in wild type females and that had no expression in wild type males and tud progeny females at all five time points examined in this study. No such genes were identified for the female germline. See Additional file 9 for list of the 342 genes expressed in or as a consequence of the female germline.