Intra- and Interspecific Variation in Primate Gene Expression Patterns

See allHide authors and affiliations

Science  12 Apr 2002:
Vol. 296, Issue 5566, pp. 340-343
DOI: 10.1126/science.1068996


Although humans and their closest evolutionary relatives, the chimpanzees, are 98.7% identical in their genomic DNA sequences, they differ in many morphological, behavioral, and cognitive aspects. The underlying genetic basis of many of these differences may be altered gene expression. We have compared the transcriptome in blood leukocytes, liver, and brain of humans, chimpanzees, orangutans, and macaques using microarrays, as well as protein expression patterns of humans and chimpanzees using two-dimensional gel electrophoresis. We also studied three mouse species that are approximately as related to each other as are humans, chimpanzees, and orangutans. We identified species-specific gene expression patterns indicating that changes in protein and gene expression have been particularly pronounced in the human brain.

Striking differences in morphology and cognitive abilities exist between humans and their closest evolutionary relatives, the chimpanzees. At least some of these differences can be assumed to form the basis for the complex and rapid cultural evolution and demographic explosions that have characterized recent human evolution (1). In addition, humans and chimpanzees differ in several other traits that are of medical interest, such as susceptibility to AIDS, epithelial neoplasms, malaria, and Alzheimer's disease (2, 3). Although it was pointed out 25 years ago (4) that many of these differences may be due to quantitative differences in gene expression rather than structural changes in gene products, nothing is known about how gene expression profiles differ between humans and chimpanzees. In order to take a first step toward understanding the evolution of the mammalian transcriptome and proteome, we studied mRNA expression levels, as well as protein expression patterns, in different tissues of humans, chimpanzees (Pan troglodytes), orangutans (Pongo pygmaeus), and rhesus macaques (Macaca mulatta). For comparative purposes, we performed similar studies in rodent species that have diverged from each other approximately as much as humans and the great apes.

First, we compared mRNA levels in brain and liver of humans, chimpanzees, and a orangutan using Affymetrix U95A arrays (5), which contain oligonucleotides that examine approximately 12,000 human genes. From the brain, gray matter from the left prefrontal lobe (Brodmann area 9) was removed at autopsies from three adult male humans, three adult male chimpanzees, and one adult male orangutan. For brain and liver, two independent isolations of RNA from adjacent tissue samples were performed for each individual and analyzed independently (5).

All possible pairwise comparisons among the six human, six chimpanzee, and two orangutan samples were made for each tissue, and the differences in apparent expression levels were used to calculate an overall distance summarized over all genes (6). For the brain samples, the distances measured among the duplicates from the same individual constituted less than 14% of the distances between individuals. For the liver samples, the corresponding value was less than 12%. Because experimental variation between the tissue samples from the same individual was small, the averages of the pairwise distances measured between the duplicates for each tissue sample were used to estimate a tree depicting the overall differences in gene expression measured between individuals. The results (Fig. 1A) show that the variation in gene expression between individuals within the species is substantial, relative to the differences between humans and chimpanzee. For example, one human brain sample differs more from the other human samples than the latter differ from the chimpanzee samples. However, for both the brain and liver samples, the humans, as well as the chimpanzees, fall into two mutually exclusive groups when their gene expression patterns are related to that seen in the orangutan, which is evolutionarily further removed from humans and chimpanzees than these are from each other. When statistically tested by a bootstrap approach, this observation is supported in both liver and brain (7). Thus, a number of gene expression differences between humans and chimpanzees are shared among all individuals analyzed from each species. The amount of gene expression differences shared among all humans is larger than those shared among all chimpanzees. One likely factor contributing to this is that oligonucleotides complementary to human cDNAs are used to assay RNA levels not only in humans but also in chimpanzees and orangutans. Thus, nucleotide sequence differences between the last-named species and humans can be expected to reduce the apparent expression levels measured in the apes. Such differences will be assigned to the human lineage. However, the apparent acceleration on the human lineage is larger in the brain (3.8-fold) than in the liver (1.7-fold), raising the possibility that gene expression patterns may have changed more in the brain than in the liver during recent human evolution.

Figure 1

Distance trees representing the relative extent of expression changes in brain and liver among (A) three primate and (B) three mouse species: MUS., M. musculus; SPR., M. spretus; and CAR, M. caroli (6). Numbers refer to the ratio between the changes common to humans and chimpanzees, and M. musculusand M. spretus, respectively.

To investigate the latter possibility, we performed a second set of experiments using membrane-based cDNA arrays carrying 21,504 DNA sequences of an average length of ∼1,000 bp, amplified from 17,997 human genes of the Unigene set (8). When such long target sequences are used, the average nucleotide sequence difference of around 0.8% between human and chimpanzee cDNAs (9) is not expected to influence the results significantly. For these experiments, brain neocortex samples from the autopsies of seven humans, four chimpanzees and two macaques were used, as well as liver samples from six humans, five chimpanzees, and four macaques. In addition, blood samples were collected from 10 humans, 10 chimpanzees, and 10 rhesus macaques. To allow the same filter arrays to be used throughout the experiments, equal amounts of RNA from a given species and tissue were pooled, labeled, and hybridized to the cDNA arrays (10).

The relative rates of evolutionary change in the transcriptomes of the three tissues were estimated (11), using the macaque as an out-group (Fig. 2). For both blood leukocytes and liver, the human expression patterns are more similar to those of the chimpanzees than to those of the macaques, reflecting the evolutionary relationships of the species. Furthermore, the extent of change on the lineages leading to the chimpanzees and the humans are equal in leukocytes and 1.3-fold different in liver. In stark contrast, the expression pattern in the chimpanzee brain cortex is more similar to that of the macaques than to that of humans. This is due to a 5.5-fold acceleration of the rate of change in gene expression levels on the lineage leading to humans. Thus, the results show that the rate of evolutionary change of gene expression levels in the brain is accelerated in the human evolutionary lineage relative to the chimpanzee, whereas no such acceleration is evident in liver or blood. It should be noted, however, that the extent of the acceleration is highly dependent on the metric used.

Figure 2

Distance trees representing the relative extent of expression changes among three primate species and three tissues as assayed by the cDNA arrays (11). Numbers refer to the ratio between the changes common to humans and chimpanzees.

To gauge whether the observations made among the primate species are typical of mammals, we investigated the three mouse species, Mus spretus, M. caroli, and M. musculus, among which the former two species differ from M. musculus at silent sites, i.e., at sites that do not change the encoded amino acids, by approximately 2.5% and 4.5%, respectively (12). Thus, their extent of divergence from M. musculus is in the same order of magnitude as that of chimpanzees (1.08%) and orangutans (2.98%), respectively, from humans (13,14). Affymetrix arrays carrying oligonucleotides specific for 12,000 M. musculus genes (5) were used to analyze samples from the frontal part of the brains and livers from three individuals of M. musculus, three individuals ofM. spretus, and one individual of M. caroli. To make the experiments as comparable as possible to the analysis of the humans and apes, outbred mice were used, and only gray matter was sampled from the frontal cortex. As in the primates, the gene expression patterns within species show great variation (Fig. 1B), as recently reported even for inbred mice (15). However, when the more distantly related M. caroli is taken into account, it is clear that all M. musculus and M. spretusindividuals share gene expression patterns that separate them from the other species, as is the case for humans and chimpanzees. When these species-specific differences are compared, it is found that the change on the line to M. musculus is 2.1-fold and 2.3-fold that in brain and liver, respectively. Thus, as in the case of the primate analyses, the species for which the oligonucleotide array was designed shows an apparent acceleration, which is likely to be due to nucleotide sequence differences between the species analyzed. However, in the rodents, this acceleration is of similar magnitude in brain and liver, and as expected from the slightly higher genomic divergence, it is slightly higher than that seen in primate liver. Thus, these results show that gene expression differences are substantial between closely related mammalian species and supports the notion that changes in gene expression levels in the brain may have been especially pronounced during recent human evolution.

Differences in mRNA levels do not necessarily translate into differences in protein levels. Therefore, we investigated whether quantitative changes not only in RNA levels but also in protein levels are especially pronounced in the brain during recent human evolution. We studied protein patterns in the brains of humans and chimpanzees, as well as in M. musculus and M. spretus to put the primate differences into perspective (16). In each case, the tissue samples were removed from sites adjacent to the ones used in the first set of mRNA analyses from the same individuals. Soluble proteins were isolated by differential centrifugation, separated on two-dimensional (2D) polyacrylamide gels, and visualized by silver staining (Fig. 3). Two types of differences were scored: (i) shifts in the migration positions of proteins, which represent a shift in size or charge of the protein, i.e., covalent difference that in most cases are changes in amino acid sequence; (ii) differences in quantity of proteins without a shift in migration position which represent differences in amounts of protein expressed in the tissue. The relative amounts of qualitative protein differences observed between humans and chimpanzees and between M. musculus and M. spretus, respectively, are similar (Table 1), as expected from the similar extent of genomic DNA sequence differences between the two species pairs. For the two rodents, the relative amounts of quantitative protein differences are similar to the qualitative differences. In contrast, quantitative differences are approximately 6 times as common as qualitative differences when chimpanzee and human brains are compared. Thus, the human brain has probably experienced more evolutionary changes in gene expression both at the mRNA and protein levels than the two mouse species. In this regard, a recent comparison of human and great ape blood plasma proteins (17) found only one human-specific difference. This is in contrast to the many differences found here for soluble brain proteins and supports a more rapid rate of evolution of protein expression levels in the brain.

Figure 3

Two-dimensional gel electrophoresis of proteins from the cytosolic fraction of human (A) and chimpanzee (B) cortex frontalis. From the total pattern, comprising about 8500 protein spots, a representative section consisting of about 200 spots is shown. Protein patterns from human and chimpanzee were compared, and changes between homologous spots found in all three human-chimpanzee pairs were scored. Three different types of variations were registered: (i) variations in electrophoretic mobility of spots (↔), most likely due to mutations affecting the structure of proteins (e.g., amino acid substitutions); (ii) variations in spot intensity (↑ or ↓) reflecting alterations in protein amount, possibly due to mutations in regulatory genes; and (iii) presence or absence variations (+ or −), which may also result from quantitative changes. For each type of variation, a few examples are indicated. From identification by mass spectrometry in both humans and chimpanzees, these proteins spots are 1, aldose reductase (gi|576365); 4, carbonic dehydratase (gi|4502517); 5, electron transfer flavoprotein (gi|2781202); 6, hypothetical protein DKFZp564D1378 (gi|14149777); 10, δ-aminolevulinate dehydratase (gi|2118316); 11, CGI-105 protein (gi|11431155); 13, hypothetical protein XP_047816 (gi|14743583); 14, malate dehydrogenase 2 (gi|5174541); 15, MAWD-binding protein (gi|16307296); 16, uncharacterized hypothalamus protein HCDASE (gi|8923864); 30, purine nucleoside phosphorylase (gi|4557801); 31, purine nucleoside phosphorylase (gi|4557801); and 35, aldehyde reductase (gi|1633300).

Table 1

Brain protein pattern differences between humans and chimpanzees as analyzed by 2D gel electrophoresis (16). Differences between humans and chimpanzees were scored if confirmed in three individual human-chimpanzee pairs and were analyzed in the same way as in a larger mouse study comparing M. musculus andM. spretus (23). Qualitative differences represent changes in electrophoretic mobility of spots, which likely result from amino acid substitutions, whereas quantitative differences reflect changes in the amount of protein.

View this table:

Our results show that that large numbers of quantitative changes in gene expression can be detected between closely related mammals. They furthermore suggest that such changes have been particularly pronounced during recent evolution of the human brain. The underlying reasons for such expression differences are likely to be manifold, for example, duplications and deletions of genes, promotor changes, changes in levels of transcription factors, and changes in cellular composition of tissues. A challenge for the future is to investigate the relative contributions of these factors to the expression differences observed. A further challenge is to clarify how many of the differences have functional consequences.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: paabo{at}


Stay Connected to Science

Navigate This Article