WIKIPEDIA, RAMAMice are widely used to model human metabolism, disease, and drug response. But results published today (November 17) in PNAS reveal widespread differences between human and mouse gene expression, both in protein-coding and noncoding genes, suggesting that understanding these disparities could help explain fundamental differences in the two species’ physiology.
Michael Snyder of Stanford University and his colleagues compared how genes are expressed in 15 different human and mouse tissues, including brain, heart, liver, and kidney. They found that gene expression patterns clustered by species rather than tissues. For example, gene expression in a mouse liver more closely resembled the patterns observed in a mouse heart than those observed in a human liver. Using data from the ENCODE and modENCODE projects, among other sources, the analysis spanned “the most tissue-diverse RNA-seq dataset to date,” the authors wrote in their paper.
The results “go a little against the grain,” said bioinformatician Mark Gerstein of Yale University who was not involved in the study. “We might think that humans and mice are very similar [genetically], but when we compare their transcriptomes, they’re more different than we thought.”
Several previous studies involving a similar comparison of gene expression in human and mouse tissues have found the opposite: that gene expression in human and mouse livers was more similar compared to that of mouse liver and brain, for example. These tissue-specific gene expression patterns may have turned up because the tissues tested included the kidney, testes, brain, liver and muscle—all of which perform very specialized, parallel functions in both species. Snyder said that the limited number of highly specialized tissues included in these studies may have hidden species-specific patterns. “There’s a lot of similarity between organs within an organism that wasn’t appreciated before,” he said.
However, Nuno Barbosa Morais of the University of Lisbon in Portugal, authored coauthor on one of these earlier papers, noted in an e-mail that the groups used different algorithms to analyze their data, which may have contributed to the discrepancies.
Snyder and his colleagues found that more than 4,000 genes were differentially expressed in human and mouse tissues. Looking for epigenetic changes—such as histone modifications—underlying these differences, the researchers found that when genes were more highly expressed in one of the two species, two histones, H3K4me3 and H3K27ac, showed more active promoter marks than in the other. According to Snyder, this suggests that gene expression differences are biological and unlikely to be an artifact of the experiments or analysis.
Extending their work to noncoding RNAs, the team discovered a greater variability in these transcript levels across tissues than observed for protein-coding genes. Previously, the majority of noncoding transcripts were thought to be highly tissue-specific. But knowing how many noncoding transcripts are truly tissue-specific will require larger datasets across a wider variety of tissues. The new results are “consistent with the idea that regulatory information in general, such as transcription factor binding, is highly diverged” between the two species, the authors wrote in their paper.
The new results are “significant and complementary” to previous comparisons of gene expression patterns across species and tissues, said computational biologist Chao Cheng of Dartmouth College in New Hampshire who was not involved in the work.
These data “guide us as to where a mouse model might be useful and where to be more cautious,” said Snyder. “A mouse and human have almost the same genes. But how we express those genes differs quite a bit.”
S. Lin et al. “Comparison of the transcriptional landscapes between human and mouse tissues,” PNAS, doi:10.1073/pnas.1413624111, 2014.
Correction (November 18): This article has been updated to correct an error in a previous version: the University of Lisbon is in Portugal. The Scientist regrets the error.