Tissue-driven hypothesis of genomic evolution and sequence-expression correlations

Abstract

To maintain normal physiological functions, different tissues may have different developmental constraints on expressed genes. Consequently, the evolutionary tolerance for genomic evolution varies among tissues. Here, we formulate this argument as a ''tissue-driven hypothesis'' based on the stabilizing selection model. Moreover, several predicted genomic correlations are tested by the human-mouse microarray data. Our results are as follows. First, between the human and mouse, we have elaborated the among-tissue covariation between tissue expression distance (E ti) and tissue sequence distance (Dti). This highly significant Eti ؊ D ti correlation emerges when the expression divergence and protein sequence divergence are under the same tissue constraints. Second, the tissue-driven hypothesis further explains the observed significant correlation between the tissue expression distance (between the human and mouse) and the duplicate tissue distance (T dup) between human (or mouse) paralogous genes. In other words, between-duplicate and interspecies expression divergences covary among tissues. Third, for genes with the same expression broadness, we found that genes expressed in more stringent tissues (e.g., neurorelated) generally tend to evolve more slowly than those in more relaxed tissues (e.g., hormone-related). We conclude that tissue factors should be considered as an important component in shaping the pattern of genomic evolution and correlations.expression divergence ͉ tissue expression ͉ mammalian genomics ͉ gene duplications

Tissue-driven Hypothesis with Gene Ontology (GO) Analysis

“…The strong correlation shows that in the tissues where the stabilizing constraint for expression divergence between species is weaker, its tolerance to expression divergence between duplicated genes is also larger; and vice versa. In the previous study, 7 we have shown this on a tissue level. Here we show that sub-grouping the genes by their GO categories does not break this correlation.…”

“…In our previous study, 7 we have shown this constraint exists on the tissue level. Here, we further show that this correlation does not break when the genes expressed in a tissue are subgrouped into GO categories.…”

“…However, since all the expression data are from adult tissues, it is possible that divergence in expression among these tissues may be profound in other developmental stages, such as embryogenesis. The tissuespecific stabilizing constraints we shown here and in the previous report 7 are based on the adult tissue, although the whole set of methods can be readily applied to tissues in other developmental stages once the data is available.…”

“…The tissues are ordered in the same order as is in Fig. 3 of Gu and Su 7 (the general distribution), where the mean tissue expression distances " E ti are in descending order among the tissues. Comparison of the histograms of " E ti;cat and " E ti displays the deviation of individual " E ti;cat distribution from the general distribution.…”

“…Meanwhile, the stabilizing selection model is also a valid null hypothesis for testing positive selection. In a recent study by Gu and Su, 7 we proposed a statistical model of gene expression divergence driven by stabilizing selection. It is also proposed in that study a ''tissue-driven'' hypothesis stating that the selective pressure on the evolution of sequence and expression are correlated and can be partitioned among tissues, and tissues differ in their tolerance to gene expression variances.…”