Global Transcriptional Profiling of Diapause and Climatic Adaptation in Drosophila melanogaster

Abstract

Wild populations of the model organism Drosophila melanogaster experience highly heterogeneous environments over broad geographical ranges as well as over seasonal and annual timescales. Diapause is a primary adaptation to environmental heterogeneity, and in D. melanogaster the propensity to enter diapause varies predictably with latitude and season. Here we performed global transcriptomic profiling of naturally occurring variation in diapause expression elicited by short day photoperiod and moderately low temperature in two tissue types associated with neuroendocrine and endocrine signaling, heads, and ovaries. We show that diapause in D. melanogaster is an actively regulated phenotype at the transcriptional level, suggesting that diapause is not a simple physiological or reproductive quiescence. Differentially expressed genes and pathways are highly distinct in heads and ovaries, demonstrating that the diapause response is not uniform throughout the soma and suggesting that it may be comprised of functional modules associated with specific tissues. Genes downregulated in heads of diapausing flies are significantly enriched for clinally varying single nucleotide polymorphism (SNPs) and seasonally oscillating SNPs, consistent with the hypothesis that diapause is a driving phenotype of climatic adaptation. We also show that chromosome location-based coregulation of gene expression is present in the transcriptional regulation of diapause. Taken together, these results demonstrate that diapause is a complex phenotype actively regulated in multiple tissues, and support the hypothesis that natural variation in diapause propensity underlies adaptation to spatially and temporally varying selective pressures.

Keywords: Drosophila melanogaster; RNAseq; climatic adaptation; diapause.

Fig. 1.

Log2-fold change of expression levels versus mean normalized RPKM values in head and ovary. Genes that are significantly differentially expressed (FDR < 0.05) in D and ND are shown in red. The expression levels are normalized by library sizes. Genes with relatively higher expression levels in diapausing samples as compared with nondiapausing samples have positive log2-fold change values, and genes with lower expression levels in D have negative log2-fold change values. In the head, 6,237 genes fall above the line of y = 0 and 5,967 genes fall below the line. In the ovary, 5,272 genes fall above the line and 5,584 genes fall below.

Fig. 2.

Venn diagrams of numbers of genes identified from the gene-level and isoform-level differential expression tests.

Fig. 3.

The null distributions (shown as curves) and real (shown as vertical lines) Moran’s I of log2-fold change of D over ND on the major chromosome arms in head and ovary. A positive Moran’s I indicates positive spatial autocorrelation. The null distribution was generated by scrambling the starting sites of genes within each chromosome 1,000 times.

Fig. 4.

Clinal and seasonal enrichment of differentially expressed genes. Enrichment (log₂ odds ratio) of differentially expressed genes among genes that harbor, or are near clinally varying (left) and seasonally oscillating (right) polymorphisms. Error bars represent 95% confidence intervals based on 500 block bootstrap resampling.