Previous studies found that the direction and magnitude of sexual selection can partially shape the evolution of gene expression on the W chromosome . Two breeds were selected as the samples representing distinct sexual selection modes. Given its differences in reproductive performance and other female fitness traits, we expected an apparent diversity of W-linked transcriptome abundance was also exist between WL and Cor. Surprisingly, our result was the opposite of the assumption in all three tissues. Specially, only less than half of the expressed genes (44%) showed significantly different expressed in the three tissues (FDR<0.05, fold change >1.25), and 17% of these genes are WL-skewed. We inferred that there are three main reasons. First, biased expression is not necessarily a fixed property of genes, expression level can vary greatly among tissues [28–31]. Generally, somatic tissues show much less dimorphism than gonads [32–34]. We chose three somatic tissues instead of gonads as the experimental samples because the gonads of 1-day-old chicks are extremely small and may easily be mixed in by other tissues. Second, gene expression is also highly variable over the course of development. Previous evidences showed there are expression changes with minor divergence in embryonic stages and high-level divergence in sexually mature adults [35, 36]. Nonetheless, we still selected 1-day-old chicks because female-specific selection in birds is strongest during this developmental time point . Third, known W-linked genes do play an important role in sex determination, but there is no evidence that their functions are patently associated with sexual fitness , the shaping effect of sexual selection on the W chromosome may also be insignificant.
Although the assumption of parental expression divergence deviated from our original intention, it did not affect the procedure and results of our exploration of regulatory changes. Before identifying regulatory variations, we observed expression clusters between tissues, and between breeds. The different sample clusters of each tissue indicated that tissues played the most significant role in gene expression for all individuals. Within tissues, the breeds-specific patterns could both be observed clearly, of which the most typical in muscle. We speculated that this obvious pattern in muscle might be related to the disparity in growth and development performance between breeds [38, 39]. Our results suggested that maternal origin was an important contributor to the cluster in gene expression in our dataset. This clustering feature was in line with our prediction due to the maternally inherited mode of W chromosome. We also hastened to point out that this maternal origin mode is more obvious in Cor♀/CL♀, but its shaping effect in WL♀/LC♀ did not seem to be significant. This evidence reflected from the side that Cor W-linked genes had stronger hereditary capabilities. The PCA overview results proved the genetic differences between the reciprocal crosses, and further demonstrated the necessity of identifying the regulatory divergence according to it.
Regulatory changes and inheritance patterns are both based on gene expression dynamic changes in the hybridization process [12, 40]. Mechanisms of regulatory divergence may influence the inheritance of gene expression, recent studies showed that inheritance diversity may depend on the effects of trans-regulatory factors of one genome on the other genome [41–43], and hence the regulatory divergence between parental species. This assumption is based on the effect that dominance is caused by trans-regulatory factors in diploid hybrid . Under the special circumstances in W chromosome, genes locate in the non-recombination region can be regarded as ‘single allele’, so cis-/trans- regulatory elements can also lead to dominant pattern. The difference is that cis-regulatory divergence will lead to a maternal-dominant, while trans-regulatory divergence contributes independently to paternal-dominant. Nevertheless, a large proportion of expressed genes showed dominant (both including caused by cis-/trans-regulatory elements), and previously plant-based studies showed that dominant pattern is prevalent and widespread among different natural populations, and maybe also closely related to the phenotypic novelty of hybrids [44–46]. The above evidences proved two conclusions. First, the special sex-limit characteristics of the W chromosome will cause a different regulatory changes mechanism compared with autosomes and Y chromosomes. Second, the result that the novel or the superiority phenotype has a certain link to dominant patterns in offspring is consistent with the classic concept in hybrid breeding.
For all classified genes, major of them were controlled by trans-regulatory factors rather than cis-regulatory elements, the ratios of "Cis" and "Trans" are 0.55, 0.63, and 0.41 in brain, liver, and muscle respectively. These results were consistent with the previous observation in autosomal genome, in which the ratios were 0.71, 0.53, and 0.25 in brain, liver, and muscle respectively . These observations confirmed that the gene expression evolution of most W-linked genes might be controlled by loci on autosomes or Y chromosomes. The similar effect of regulatory changes on autosomal genes and sex-linked genes showed that the total efficacy of regulatory divergence along the entire genome was basically stable. To identify the relative contribution of “W single allele” originating from WL and Cor, we carefully observed whether there was a breed-skewed expression mode in dominant W-linked genes. Interestingly, all dominant genes in brain tissue exhibited Cor-skewed, regardless of the female paternal expression level. This extreme imbalance was not observed in liver and muscle tissues, the proportion of Cor-skewed expression genes only accounted for 52% and 62% respectively, only showed a slight advantage over WL-skewed expression. We previously thought that hybrids might inherit the excellent muscle growth characteristics of the Cor parent and thus be more Cor-skewed compared with the other two tissues. For this result that was contrary to our conjecture, there were at least three, non-mutually exclusive, possible explanations. First, compared with liver and muscle, brain has the most conservative expression pattern , the expression regulation was not easily affected by reciprocal crosses and has a high consistency. Second, heterosis in hybrids may be mainly reflected in over-dominant rather than dominant [47, 48]. The evidence for this inference in this study was that muscles have the most “over-dominant” genes. Finally, unlike autosomal genes, the function of W-linked genes may be less related to muscle and body development.
Taken together, this study provided a significant advance in understanding regulatory evolution on a sex-limit genomic scale. We drew on the traditional methods that distinguish between cis- and trans-acting sources on autosomes, and used a new method to evaluate the regulatory changes of W chromosome for the first time. Our results also provided a systematic look at the evolution of cis- and trans-acting, and incorporated the inheritance pattern into the same research framework with regulatory divergence. In principle, this joint analysis approach had advantages because of the causal relationship between these two concepts. Cis-regulatory elements and trans-acting factors control nearby and distant gene expression. Meanwhile, the hereditary architecture of gene expression levels determines the inheritance pattern. Using the RNA-Seq data of hybridization model, we globally identified the features at the transcriptome level of W-linked genes and visualized them through the classification of regulatory divergence. More instances of trans-regulatory divergence than instances of cis-regulatory divergence were observed in W chromosome, this might be because the relatively short divergence history of Cor and WL [23, 49]. This low genetic diversity was also a potential cause that DEGs among parents only account for a small part of expressed genes.
What is certain is that although the regulatory pattern identifies based on the transcriptome level is reasonable, it does not mean that the result is completely accurate. A small number of organized, single-point, fixed-parent studies are not enough to allow us to understand the mechanism and laws of regulatory divergence from a broader perspective, also does not allow us to locate the sequence sites of these regulatory factors. What can be encountered is that when these limitations are broken, higher throughput and more accurate sequencing methods are applied, the problem will be solved, and the research on regulatory divergence will not just stop at the stage of description and statistical analysis.
In conclusion, our research used innovative methods to identify the genetic pattern and regulatory divergence of the W chromosome, which was not limited to a single tissue, and a single set of conditions. The results revealed an autosomal-like regulatory model, which implied a robust mechanism of regulatory divergence across whole sequences. Insights on W-linked gene expression regulation and evolution would expand such research at the species (or breed) and genome levels.