Previous studies have shown that the rate of evolution among brain-expressed genes is probably lower (or at most equal) in humans compared with chimpanzee and old world monkeys (for instance, most recently in [31
]). Slower sequence evolution of tissue/region-specific genes is a probable indicator of stronger selective constraints operating on the region in hand. Hence, the overall sequence conservation of highly expressed brain genes makes them an interesting subject for the further study of the basic relation between gene expression and ER. To this end, we find that cortically expressed genes are more conserved than subcortical ones, and that gene expression levels exert stronger constraints on sequence evolution in cortical versus subcortical regions. Taken together, these findings support the view that cortically expressed genes are under stronger selective pressure than subcortically expressed genes.
One possible mechanism that can partially explain these findings is the overall broader tissue distribution of cortically expressed genes, but other nonexclusive mechanisms may take part. For instance, it is possible that there are more frequent genetic and protein interactions among highly expressed genes in the cortical regions, which are known to be correlated with reduced ER levels [7
]. The cellular complexity (types of cells and their distribution) of the regions studied is different, which may further determine different and complex evolutionary constraints in each region. Another factor potentially influencing these regional differences is the sex bias of genes, because it has been suggested that the expression of genes that are more pleiotropic (or, in terms of our work, that have a greater tissue expression breadth) is less sex biased [32
], and that sex-dependent allelic effects cannot maintain polygenic variation [33
]. Thus, the exact mechanisms underlying our findings are probably subject to quite complex interplay that remains to be further explored.
The magnitudes of some of the ER/expression correlations reported here are lower than in yeast (see Figure for gene expression in the prefrontal cortex versus their ERs), even though these correlations are highly significant. In the case of the yeast (for example [5
]), the respective correlation found is around 1.5 times higher. There are three main reasons that may explain this difference. First and foremost, in contrast to the yeast, humans are multicellular organisms with hundreds of distinct cell types and diverse tissues; thus, gene ER in humans is likely to be under a large variety of (sometimes perhaps counteracting) selection forces, resulting in a lower correlation with gene expression in any single specific cell/tissue type [28
]. Second, because this study focuses on human brain regions, we have estimated ER values along shorter evolutionary time periods (the past 6.5 to 10 million years of the human lineage, after the human-chimp split [34
], and the 50 to 100 million years corresponding to the human-mouse split), which is in contrast to the much longer time spans employed for estimating ER in the yeast studies. Indeed, when using ER estimates using the human-mouse lineage, we obtain ER/expression correlations that are two times higher than those obtained when using ER estimates from the shorter period, human-chimp lineage. Third, the sets of genes studied differ markedly, with the number of genes included in this study being two to three times higher than the number of genes examined in previous yeast studies (larger datasets usually increase the correlations but may decrease their significance).
Cortical regions, at least in their extensive mammalian form, are more recent than subcortical regions, which have a broader phyletic distribution. The ER of cortically expressed genes is yet slower than that of subcortically expressed genes. This is in contrast to the findings at the gene level, at which the ER of younger genes is higher than that of older ones [21
]. This appears paradoxical at first, because one would perhaps expect that genes that are highly expressed in the more recently evolving cortical brain regions would be younger than the genes that are highly expressed in subcortical regions. However, this is not the case; highly expressed cortical genes tend also to be highly expressed in many subcortical regions, and thus both types of regions are composed of both younger and older highly expressed genes (with cortical areas being actually composed of older genes than subcortical regions, on average). Furthermore, although we find that cortically expressed genes are more conserved than subcortical ones, this does not necessarily imply that cortical regions offer more stringent 'environments' for gene evolution than subcortical regions, because this excess conservation may arise from their broader, somatic tissue distribution. However, the tighter correlation between ER and expression levels that characterizes cortically expressed genes does point to the fact that the cortex may form a more stringent environment for gene evolution than other brain and somatic tissues, as one may intuitively expect [19
]. (Obviously, in turn, it is also possible that the rates of gene evolution may play an important role in shaping their expression profiles in the cortex.)
There are many definitions for the tissue/region specificity of genes (for example, based on expressed sequence tags data, serial analysis of gene expression data, literature [36
], or gene expression [as was adopted here]). Each of the definition may give rather different sets of genes. Currently, there are no available datasets based on expressed sequence tags, serial analysis of gene expression, or literature that provide information about brain regional specificity. Hence, we have focused on the gene expression definition of region specificity. A comparison of our results with those based on other tissue specificity definitions will have to be deferred until the corresponding biologic information becomes available.
Finally, the results reported in Figure are intriguing, generalizing in a way the results reported in Figures to . Whereas Figures to report that cortical regions exhibit a correlation between ERs and tissue gene expression levels in cortical versus subcortical regions, Figure shows that this correlation tends to be stronger for vertically higher regions in the developmental axis. each point in Figure corresponds to the correlation between ER and expression levels of 10,594 genes in all of the regions in a developmental area, the reported correlation values are highly robust. Thus, drawing an analogy from the observation that cortical regions are evolutionary more recent than subcortical ones [24
], one may (perhaps boldly) speculate that regions located higher on the vertical axis at brain development are also more evolutionarily recent. However, because even the basic claim that cortical regions are more recent is not accepted by everyone, care should obviously be taken with formulating such hypotheses. Their examination should await the accumulation of additional gene expression samples from more brain tissues and from more mammalian species.