Sex-specific expression and DNA methylation in a species with extreme sexual dimorphism and paternal genome elimination

Sexual dimorphism is exhibited in many species across the tree of life with many phenotypic differences mediated by differential expression and alternative splicing of genes present in both sexes. However, the mechanisms that regulate these sex-specific expression and splicing patterns remain poorly understood. The mealybug, Planococcus citri, displays extreme sexual dimorphism and exhibits an unusual instance of sex-specific genomic imprinting, Paternal Genome Elimination (PGE), in which the paternal chromosomes in males are highly condensed and eliminated from the sperm. P. citri also has no sex chromosomes and as such both sexual dimorphism and PGE are predicted to be under epigenetic control. We recently showed that P. citri females display a highly unusual DNA methylation profile for an insect species, with the presence of promoter methylation associated with lower levels of gene expression. In this study we therefore decided to explore genome-wide differences in DNA methylation between male and female P. citri using whole genome bisulfite sequencing. We have identified extreme differences in genome-wide levels and patterns between the sexes. Males display overall higher levels of DNA methylation which manifests as more uniform low-levels across the genome. Whereas females display more targeted high levels of methylation. We suggest these unique sex-specific differences are due to chromosomal differences caused by PGE and may be linked to possible ploidy compensation. Using RNA-Seq we identified extensive sex-specific gene expression and alternative splicing. We found cis-acting DNA methylation is not directly associated with differentially expressed or differentially spliced genes, indicating a broader role for chromosome-wide trans-acting DNA methylation in this species.

In addition to extreme sexual dimorphism, P. citri also has an unusual reproductive strategy, ). Due to the haploidization of males, PGE is often referred to as a 'pseudohaplodiploid' system. 85 Furthermore, we have previously shown P. citri females have a unique pattern of whole 86 genome DNA methylation that differs from that found in other arthropods (Lewis et al., 2020). 87 Whilst most arthropods have depleted levels of transposable element and promoter methylation, P. 88 citri has independently evolved both (Lewis et al., 2020). Interestingly, and similar to patterns shown 89 in mammals, genes with low expression in P. citri have significantly higher promoter methylation 90 than highly expressed genes (Lewis et al., 2020). It is also suggested that DNA methylation may 91 have a role in the recognition and silencing of paternally-derived chromosomes in males in the in adult Phenacoccus solenopsis (Omar et al., 2020) and Ericerus pela (Yang et al., 2015), with 96 females showing considerably higher expression compared to males in both species.

97
In order to identify sex-specific patterns of gene expression and clarify the role of DNA 98 methylation in this process, we analyse both male and female P. citri methylomes and transcriptomes.

99
This is the first genome-wide analysis of sex-specific gene expression and DNA methylation in scale 100 insects. Using RNA-seq and whole genome bisulfite sequencing (WGBS) we find clear differences 101 in gene expression and methylation profiles between the sexes. However, we find no relationship 102 between differentially expressed genes and differentially methylated genes, indicating that cis-acting 103 DNA methylation is not the sole driver of sex-specific gene expression in adult P. citri.

Differential expression and alternative splicing
Raw RNA-seq reads for each sample were trimmed for low quality bases and adapters using Fastp    We also identify 168 extremely sex-biased genes (q <0.05 and >10 fold change, Supplementary 247 1.0.2), with the majority of these showing extreme male-biased expression (140 compared to 28, 248 chi-squared goodness of fit: X-squared = 74.667, df = 1, p <0.001, Fig.2c). There were only three 249 GO terms enriched for extremely biased male genes, these were "system process" (GO:0003008),

298
The diversity within male samples may be explained by a lower input of DNA during the library 299 preparation process, resulting in possible sequencing bias.   Due to the unusual occurrence of promoter methylation in P. citri, we investigated sex-specific 344 differences in promoter methylation and exon 1-3 methylation, separately. We find 2,709 genes with 345 a differentially methylated promoter (minimum three differentially methylated CpGs and a minimum 346 overall weighted methylation difference of 15%) between males and females and 2,736 genes with 347 differentially methylated exons 1-3 (Supplementary 1.1.0). A significantly higher number of genes 348 with differential promoter methylation were hypermethylated in females compared to males (2,645 in 349 females and 64 in males, chi-squared goodness of fit, X-squared = 2459, df = 1, p < 0.001, Fig.4b and   350 4c). This was also the case for genes with differential exon methylation, with 2,709 hypermethylated 351 in females and 33 hypermethylated in males (chi-squared goodness of fit, X-squared = 2611.6, df = 352 1, p < 0.001, Fig.4b and 4d). In females, there is also a significant overlap of genes showing both 353 hypermethylation of the promoter region and exons 1-3 (hypergeometric test, p <0.001, Fig.4b).

354
As males show mostly low-to-medium levels of methylation throughout the genome, we 355 analysed the distribution of methylation levels for each feature determined as differentially methylated. 356 We found that the average level of methylation in males for male hypermethylated promoters is 0.12 ± 357 0.07 (mean ± standard deviation) and for exons is 0.14 ± 0.12 (Supplementary 2.0, Fig.S4a and S4b),

Relationship of DNA methylation and expression
Gene body DNA methylation is reported to positively correlate with gene expression in a number  Fig.5a and 5b). This is the case for both males and females as 379 there is no interaction between sex and methylation level (two-way ANOVA: F 2,3 = 0.265, p = 0.606).   , and high (0.7-1). The red dot indicates the mean with 95% confidence intervals.

Relationship of differential DNA methylation and differential expression 392
If DNA methylation is a causative driver of changes in gene expression we would expect that 393 differentially methylated genes between sexes are also differentially expressed. Given that higher 394 methylation is associated with lower expression in this species, we would also expect that down-395 regulation of gene expression is associated with higher methylation. However, on a single gene level, 396 we found there is no clear relationship between the level of differential promoter methylation and the 397 level of differential expression of the corresponding gene (Fig.6a). This is also the case for exon 1-3 398 methylation (Supplementary 2; Fig.S6a).  Table S2).

408
Therefore, whilst genome-wide higher methylation is correlated with lower expression, this trend  (Supplementary 2: Fig.S6b). 421 We then assessed the overall methylation levels of differentially expressed genes. We found

442
We also then checked to see if alternatively spliced genes were also differentially methylated 443 between sexes. We found only one significant overlap of genes which are both alternatively 444 spliced and differentially methylated (Supplementary 2: Table S3), a single gene was common 445 between alternatively spliced genes which show male expression bias and genes with male promoter 446 hypermethylation (hypergeometric test with bonferroni correction, p = 0.034). However, it is likely 447 this overlap is significant due to the small gene lists rather than due to biological significance.

Discussion
In this study, we investigated the relationship between sex-specific gene expression and DNA  464 We have identified extreme sex-specific differences in DNA methylation across the genome of P. In addition to our key findings above we have also identified sex-specific gene expression and 540 alternative splicing. P. citri have no sex chromosomes meaning that males and females share the same 541 genetic complement (Hughes-Schrader, 1948). Thus, the observed sexual dimorphism exhibited 542 must be a consequence of differences in gene expression and splicing between the sexes. Indeed, we 543 found that 54% of genes show sex-biased expression, including a subset of genes that are extremely 544 sex-biased and sex-limited. We found that both male-and female-biased genes are involved in 545 core biological processes. Sex-limited genes are likely important in the phenotypic sex differences 546 observed in P. citri, including sensory related male-limited genes that may be involved in mate 547 recognition through pheromones (Bierl-Leonhardt et al., 1981). Nasonia males also show extreme 548 sex-biased expression of pheromone genes (Wang et al., 2015). The large number of differentially expressed genes we have identified reflects the extreme sexual dimorphism shown in this species 550 (Fig.1). 551 We also identified differentially alternatively spliced genes between the sexes and found a 552 significant number of these show male-biased expression. Genome-wide sex-specific alternative 553 splicing has also been identified in aphids (Grantham and Brisson, 2018) and other insects (e.g. less genes but use more alternative transcripts. In P. citri, we found generally more female-biased 563 genes compared to male-biased genes but more male-biased alternatively spliced genes, showing 564 that P. citri sexes also employ different mechanisms to generate sex-specific phenotypes.

PGE may explain uniform DNA methylation in males
565 Surprisingly, we did not find any genes orthologous to the Drosophila doublesex gene to be 566 alternatively spliced. Alternative splicing of doublesex is ubiquitous in holometabolous insects, 567 whereas male-biased expression rather than alternative splicing has been detected in some crustaceans