De novo transcriptome assembly reveals sex-specific selection acting on evolving neo-sex chromosomes in Drosophila miranda

Abstract

Background: The Drosophila miranda neo-sex chromosome system is a useful resource for studying recently evolved sex chromosomes. However, the neo-Y genomic assembly is fragmented due to the accumulation of repetitive sequence. Furthermore, the separate assembly of the neo-X and neo-Y chromosomes into genomic scaffolds has proven to be difficult, due to their low level of sequence divergence, which in coding regions is about 1.5%. Here, we de novo assemble the transcriptome of D. miranda using RNA-seq data from several male and female tissues, and develop a bioinformatic pipeline to separately reconstruct neo-X and neo-Y transcripts.

Results: We obtain 2,141 transcripts from the neo-X and 1,863 from the neo-Y. Neo-Y transcripts are generally shorter than their homologous neo-X transcripts (N50 of 2,048-bp vs. 2,775-bp) and expressed at lower levels. We find that 24% of expressed neo-Y transcripts harbor nonsense mutation within their open reading frames, yet most non-functional neo-Y genes are expressed throughout all of their length. We find evidence of gene loss of male-specific genes on the neo-X chromosome, and transcriptional silencing of testis-specific genes from the neo-X.

Conclusions: Nonsense mediated decay (NMD) has been implicated to degrade transcripts containing pre-mature termination codons (PTC) in Drosophila, but rampant description of neo-Y genes with pre-mature stop codons suggests that it does not play a major role in down-regulating transcripts from the neo-Y. Loss or transcriptional down-regulation of genes from the neo-X with male-biased function provides evidence for beginning demasculinization of the neo-X. Thus, evolving sex chromosomes can rapidly shift their gene content or patterns of gene expression in response to their sex-biased transmission, supporting the idea that sex-specific or sexually antagonistic selection plays a major role in the evolution of heteromorphic sex chromosomes.

Figure 1

Neo-Y assembly pipeline. (a) Trinity de novo transcripts were broken into exons using genomic read mapping information; (b) transcripts were modified to only harbor neo-Y specific variants; (c) only sections of neo-Y transcripts were kept if they were supported by RNA-Seq reads and carried at least one variant compared to the neo-X; (d) transcripts were scaffolded using the STM method and D. pseudoobscura protein alignments.

Figure 2

Density plot of the total divergence between the assembled neo-Y transcripts compared to neo-X (green) and D. pseudoobscura (black) transcripts, respectively.

Figure 3

Histogram of the position of the first pre-terminal codon (PTC) within the neo-Y coding sequence, relative to the length of the neo-Y coding region.

Figure 4

No evidence for nonsense-mediated decay in reducing neo-Y transcript abundance. (a) Neo-Y FPKM levels for transcripts with intact ORFs, transcripts containing a PTC and transcripts containing a frame-shift mutation only (and no PTC). Transcript levels for both classes of putative non-functional genes were significantly decreased, compared to those with intact ORFs (Wilcoxon test: W = 132628, p < 0.001 and W = 234562, p < 0.001). (b) Transcript abundance of the neo-Y in males, relative to the neo-X homolog in females, as a function of the length of the 3'UTR (estimated as the sum of the corresponding neo-X UTR length, plus the distance from the first PTC to the end of the CDS).

Figure 5

Deletions on the neo-X chromosome in D. miranda. Shown are flybase gbrowser plots of genomic regions on Muller element C in D. pseudoobscura (genome assembly version 3); red bars indicate deletions in the homologous regions on the neo-X in D. miranda, and red arrows indicate insertion on the neo-X. See Additional file 2 for a detailed description of the inferred deletions.

Figure 6

Transcript abundance (FPKM) for different classes of neo-sex-linked genes. a) Neo-Y FPKM-levels in D. miranda males; in testis, neo-Y transcripts whose homologues on the neo-X have been deleted or silenced are expressed at significantly higher levels compared to transcripts whose neo-X homologues are expressed (Wilcoxon test: W = 63263, p < 10⁻⁸ for neo-X deleted genes; W = 42141, p < 0.01 for neo-X silenced genes). b) Neo-X FPKM-levels in D. miranda females. c) FPKM-levels of D. pseudoobscura transcripts which are homologues to neo-sex linked D. miranda genes; in testis, D. pseudoobscura transcripts whose homologues on the neo-X in D. miranda have been transcriptionally silenced are expressed at significantly higher levels compared to transcripts whose homologues on the neo-X are expressed (W = 42141, p < 0.01); the same observation holds true when the three D. pseudoobscura transcripts with neo-X deleted homologues are added to a combined “neo-X silenced or deleted” category (W = 44476, p < 0.01).