In this study we have identified a large number of eQTL that regulate the transcriptome in cerebrum tissue, with the majority of these eQTL being trans-acting. The overall regulation of gene expression variation in the cerebrum was markedly different to the hypothalamus, with relatively few eQTL in common between these two brain regions. In comparison, we find that cerebrum tissue shared more eQTL with a distant tissue (liver) than with the hypothalamus. Furthermore, we also identify one candidate gene that putatively regulates cerebrum size (MTF2) and two that potentially regulate overall body mass.
Although of obvious interest, relatively few genes responsible for brain size variation have been identified (chiefly in humans (20–22), mice (23) and chickens (5)). Research into the genes underlying variation in human brain size has identified 2 loci (24), whilst a large-scale GWAS for intra-cranial volume (maximum brain size) discovered a further five loci (25). These five loci were identified using extremely large sample sizes (32438 individuals) to identify association signals in several genes, with these genes involved in neural stem cell proliferation (FOXO3) and neurodegeneration (MAPT) amongst others. The larger sample size here is somewhat offset by the relatively small variation present in humans compared to the extreme variation generated by comparing wild and domestic birds (5). Here we have combined this genetics approach, using the loci detected previously, with transcriptomic eQTL to combine actual gene expression with brain size parameters to detect a single candidate for cerebrum size. This is a large advantage over studies that solely rely on association with no gene expression data, as these are unable to distinguish trans-acting effects over local cis-acting effects when determining the actual causal genes. The candidate gene identified, MTF2 (also known as Pcl2), is a metal response element binding transcription factor that selectively binds to unmethylated DNA (26) and has been found to control development of the anterior central nervous system (27). Therefore, this demonstrates a previous role in brain development, making it a good candidate for further investigation as a regulator of inter-population variation in cerebrum size. The candidate genes identified for regulations of overall body mass were PCBD2 and an EST (603848039F1). PCBD2 is a target of the microRNA miR-3174, with this miRNA potentially having a role in rectal cancer through its action on PCB2 (28). The EST mapped to an exon on the gene RCHY1 (ring finger and CHY zinc finger domain containing 1) that has been shown to have DNA damage response and cell cycle activity (29). Additionally, evidence show that HOXA2 induces RCHY1 degradation (30), indicating that RCHY1 might play a role in brain development as the HOX-genes are involved in embryonic development for both the axial skeleton and the hindbrain (31).
Although we did identify one candidate gene for cerebrum size (and two for body size), certainly many more were invariably failed to be identified. The fact that only one gene responsible for the variation in cerebrum size was identified could be due to the fact that the cerebral hemisphere is a large brain structure that consist of several substructures (13). The functional constraints hypothesis (the mosaic hypothesis) predicts a more independent genetic basis for brain substructures than the concerted hypothesis and suggests that the correlated coevolution of brain regions sizes reflects the action of selection on functional systems connecting the different sub-components. The fact that we find very few correlations between the transcriptome of the cerebrum and the regulation of inter-population variation in size might therefore be due to the fact that the size of different sub-regions in the cerebrum is controlled by different genes and separate genetic architectures. Similarly, the effect size of these genes regulating this variation may also be too small to be detectable in our study. However, the link we do find between body-size and gene-expression in the cerebrum suggest a link to the regulation of an individual’s overall size is partially regulated by the cerebrum. It was outside the scope of this study to do more detailed dissections on the cerebrum to measure size and gene-expression in the smaller substructures, and our discussion on this matter therefore remains speculative.
To test the two different brain evolution hypotheses, a comparison between cerebrum and hypothalamus tissue was conducted. Relatively little overlap was identified between the two brain regions in terms of shared eQTL identified. In fact, both brain regions had more in common with the liver tissue, respectively, than each other. Similarly, when looking at shared eQTL hotspots, only two eQTL were present in hotspots in both types of brain tissue, with only one of those genes (RHCY1 - EST 603848039F1) actually having a similar genotypic effect (i.e. the Red junglefowl allele increasing expression) in both tissues. These results would seem to give greater evidence to the mosaic evolution hypothesis. In particular, the lack of shared eQTL between the two brain regions implies that genetic regulation of gene expression appears to be relatively independent between the two brain regions. Given the large differences in both overall brain size and composition between wild and domestic chickens, this large variation appears to share distinct regulatory patterns. However, there are several caveats to this interpretation. eQTL analysis, and indeed any expression analysis is of course dependent not only on the tissue but also on the time-point of sampling. As such it is possible that earlier sampling (for example immediately pre-hatch or early post-hatch) may have revealed a different genetic architecture of gene regulation, with more shared eQTL. In particular we know that brain composition changes from hatch to early adolescence to adulthood in chickens (5). Similarly, an inherent problem with QTL mapping is that eQTL of small effect may well be missed, and as such some overlaps may be missed as a result.
The lack of shared eQTL hotspots between hypothalamus and cerebrum tissue also has some bearing on the Neural Crest Cell (NCC) hypothesis of domestication (32). The NCC hypothesis conjectures that the migration and distribution of neural crest cells control a whole range of domestication-related traits, therefore pleiotropy is to be expected for multiple domestication-related QTL traits where they are all underpinned by loci controlling neural crest cell parameters. However, we find that rather few eQTL overlap between multiple tissue types (especially when we consider overlap between all three tissues). When we consider shared eQTL hotspots between the different brain tissues, there is some overlap, however only two actual eQTL are shared between the hotspots present in these two tissues. Interestingly, one of these genes, GEMIN2 (also known as SIP1) has been associated to neural crest activity (33,34) and motor neuron activity (35). The other gene, C5H15orf41, has an unknown cellular function, though the open reading frame in C5H15orf41 is highly conserved in vertebrates and has orthologs in human (C15orf41), rats (RGD1563680), mouse (BC052040), dog (C30H15orf41) and zebrafish (zgc:154061). Therefore, although the eQTL evidence overall does not appear to support neural crest cells as a mechanism of domestication, the presence of one gene does warrant further investigation in the future for a potential pleiotropic role.