Sister grouping of chimpanzees and humans as revealed by genome-wide phylogenetic analysis of brain gene expression profiles

Abstract

Gene expression profiles from the anterior cingulate cortex (ACC) of human, chimpanzee, gorilla, and macaque samples provide clues about genetic regulatory changes in human and other catarrhine primate brains. The ACC, a cerebral neocortical region, has human-specific histological features. Physiologically, an individual's ACC displays increased activity during that individual's performance of cognitive tasks. Of approximately 45,000 probe sets on microarray chips representing transcripts of all or most human genes, approximately 16,000 were commonly detected in human ACC samples and comparable numbers, 14,000-15,000, in gorilla and chimpanzee ACC samples. Phylogenetic results obtained from gene expression profiles contradict the traditional expectation that the non-human African apes (i.e., chimpanzee and gorilla) should be more like each other than either should be like humans. Instead, the chimpanzee ACC profiles are more like the human than like the gorilla; these profiles demonstrate that chimpanzees are the sister group of humans. Moreover, for those unambiguous expression changes mapping to important biological processes and molecular functions that statistically are significantly represented in the data, the chimpanzee clade shows at least as much apparent regulatory evolution as does the human clade. Among important changes in the ancestry of both humans and chimpanzees, but to a greater extent in humans, are the up-regulated expression profiles of aerobic energy metabolism genes and neuronal function-related genes, suggesting that increased neuronal activity required increased supplies of energy.

Fig. 1.

The optimal maximum parsimony trees inferred by phylogenetic analysis of primate ACC microarray data. Trees are shown in phylogram format (except the lineage leading to the macaques), with branch lengths proportional to the assigned changes for that branch as determined with the DELTRAN option in parsimony analysis. Numbers above a branch indicate the assigned branch length. Number below an internode indicates the bootstrap support value for that node (500 replicates). (A) Topology obtained from Data Matrix A, the detection call analysis. Tree length, 24,266. Detection calls were generated for each taxon by assigning as present or absent only those probe sets consistently called present or absent among all three biological replicates or between duplicates. Probe sets with variable detection calls within species, or those consistently detected as marginal within a species, were assigned a call of marginal. (B) Topology obtained from Data Matrix B, the binned signal values of commonly detected probe sets. Tree length, 46,838. Continuous characters were converted into one of 25 discrete bins before analysis (see Materials and Methods for details). (C) Topology obtained from Data Matrix C, the combined data matrix. Tree length, 71,104. Bins from Data Matrix B were retained and combined into an alignment with Data Matrix A. Values assigned by the latter on a per-taxon basis were further assigned to each individual (or replicate of an individual) included in that taxon.

Fig. 2.

Number of probe sets changing unambiguously on each stem lineage and/or internode. Shown are the number of up- and down-regulated probe sets analyzed for functional information using oe (Ver. 2.0) (28, 29). Because expression levels at the ancestral catarrhine node are unknown, the direction of expression change was not reconstructed for the African ape–macaque internode.