Distinct epigenomic and transcriptomic modifications associated with Wolbachia-mediated asexuality

Abstract

Wolbachia are maternally transmitted intracellular bacteria that induce a range of pathogenic and fitness-altering effects on insect and nematode hosts. In parasitoid wasps of the genus Trichogramma, Wolbachia infection induces asexual production of females, thus increasing transmission of Wolbachia. It has been hypothesized that Wolbachia infection accompanies a modification of the host epigenome. However, to date, data on genome-wide epigenomic changes associated with Wolbachia are limited, and are often confounded by background genetic differences. Here, we took sexually reproducing Trichogramma free of Wolbachia and introgressed their genome into a Wolbachia-infected cytoplasm, converting them to Wolbachia-mediated asexuality. Wolbachia was then cured from replicates of these introgressed lines, allowing us to examine the genome-wide effects of wasps newly converted to asexual reproduction while controlling for genetic background. We thus identified gene expression and DNA methylation changes associated with Wolbachia-infection. We found no overlaps between differentially expressed genes and differentially methylated genes, indicating that Wolbachia-infection associated DNA methylation change does not directly modulate levels of gene expression. Furthermore, genes affected by these mechanisms exhibit distinct evolutionary histories. Genes differentially methylated due to the infection tended to be evolutionarily conserved. In contrast, differentially expressed genes were significantly more likely to be unique to the Trichogramma lineage, suggesting host-specific transcriptomic responses to infection. Nevertheless, we identified several novel aspects of Wolbachia-associated DNA methylation changes. Differentially methylated genes included those involved in oocyte development and chromosome segregation. Interestingly, Wolbachia-infection was associated with higher levels of DNA methylation. Additionally, Wolbachia infection reduced overall variability in gene expression, even after accounting for the effect of DNA methylation. We also identified specific cases where alternative exon usage was associated with DNA methylation changes due to Wolbachia infection. These results begin to reveal distinct genes and molecular pathways subject to Wolbachia induced epigenetic modification and/or host responses to Wolbachia-infection.

Fig 1. Crossing scheme for creating homozygous introgressed isofemale lines and fitness at each introgression generation.

A) After seven generations of introgression, between 95–99% of the Wolbachia-dependent asexual nuclear background was replaced by the sexually reproducing nuclear background. At each generation of the introgression, we used virgin wasps to assay for sex ratio (a proxy for successful Wolbachia-mediated parthenogenesis) and fecundity prior to mating with a sexual male. All offspring were screened for zygosity, allowing for the identification of females produced via fertilization rather than Wolbachia-mediated parthenogenesis. This scheme was performed 3 independent times to generate 3 isofemale lines. B) Wasp fitness and C) efficiency of Wolbachia-mediated parthenogenesis at each generation of introgression. Sex ratio is denoted as the proportion of female offspring.

Fig 2. Differential methylation and expression between Wolbachia infected and cured lines.

A) Heatmap of 340 significantly differentially methylated positions (DMPs, found within 84 genes) detected using RADMeth. More sites exhibit increased DNA methylation in the infected lines (239 out of 340 DMPs are hypermethylated in the infected lines compared to cured lines). B) Heatmap of 59 differentially expressed genes, the majority (39 out of 59; χ², P = 3.13x10^-6) of which were up-regulated in the infected lines. C) Gene body fractional methylation and gene expression relationship between DEG, DMG and other genes (genes that are neither DEG or DMG) accompanied by gene body fractional methylation densities of each gene class. Gene body fractional methylation is calculated as the mean fractional methylation of all CpGs within each gene, and shows a bimodal distribution as in other species [46]. Gene body fractional methylation and gene expression are log₁₀ and log₂ transformed, respectively, to improve normality. D) Average gene lengths and E) density plot of gene lengths for each gene class. Error bars represent standard deviations.

Fig 3. Transcriptional noise is negatively correlated with gene body methylation, gene expression, gene length, and Wolbachia infection.

A) Summary of multiple linear regression modeling transcriptional noise, represented as the percent coefficient of variation of gene expression. B) Transcriptional noise is lower in infected lines and decreases more sharply with increased gene expression (Regression coefficients for cured and infected lines are -0.83 (P < 10⁻¹⁵) and -0.88 (P < 10⁻¹⁵), respectively). Transcriptional noise and gene expression are log₁₀ and log₂ transformed to improve clarity (see Methods). C) Violin plot with an imbedded boxplot demonstrating that transcriptional noise in infected lines is lower than that in cured lines (Student’s t-test, P < 10⁻¹⁵).

Fig 4. Representative genes showing differential exon usage between cured and infected Trichogramma.

Boxes filled in purple are differentially used exons. Among the differentially used exons, those that contain DMPs are shown in yellow-filled boxes. A) Gene TPRE001057, ortholog of D. melanogaster CG14299, contains 2 differentially used exons, exons 1 and 23 (exon numbers are shown as integers on the x-axis). The exon 23 contains 6 DMPs is highlighted with a yellow-filled box. B) Gene TPRE010061, ortholog of D. melanogaster Mzt1, contains 4 differentially used exons (exons 5, 7, 8, and 10). The exon 10 has 8 DMPs and shown in yellow-filled box.