Transcriptional neoteny in the human brain Mehmet Somela,b,1, Henriette Franzb,c, Zheng Yana, Anna Lorencb, Song Guoa, Thomas Gigerb, Janet Kelsob, Birgit Nickelb, Michael Dannemannb, Sabine Bahnd, Maree J. Webstere, Cynthia S. Weickertf, Michael Lachmannb,2, Svante Pa¨ a¨ bob,2, and Philipp Khaitovicha,b,1,2 aPartner Institute for Computational Biology, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences, 320 Yue Yang Road, Shanghai 200031, China; bMax Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, D-04103 Leipzig, Germany; cMax Planck Institute for Biophysical Chemistry, Am Fassberg 11, D-37077 Go¨ttingen, Germany; dInstitute of Biotechnology, University of Cambridge, Cambridge CB2 1TN, United Kingdom; eStanley Medical Research Institute, 9800 Medical Center Drive, Rockville, MD 20850; and fMacquarie Group Foundation Chair of Schizophrenia Research, Schizophrenia Research Institute, Prince of Wales Medical Research Institute, University of New South Wales, Sydney, NSW 2052, Australia Edited by Morris Goodman, Wayne State University, School of Medicine, Detroit, MI, and approved February 18, 2009 (received for review January 16, 2009) In development, timing is of the utmost importance, and the timing brain development between humans and apes would be crucial for of developmental processes often changes as organisms evolve. In understanding human evolution. Here we address this issue by human evolution, developmental retardation, or neoteny, has been analyzing genome-wide gene expression levels in human, chimpan- proposed as a possible mechanism that contributed to the rise of zee, and macaque brains during postnatal development. many human-specific features, including an increase in brain size and the emergence of human-specific cognitive traits. We analyzed mRNA Results and Discussion expression in the prefrontal cortex of humans, chimpanzees, and General Pattern of Expression Changes During Brain Development. We rhesus macaques to determine whether human-specific neotenic analyzed gene expression levels in the dorsolateral prefrontal cortex changes are present at the gene expression level. We show that the (DLPFC) of 39 humans, 14 chimpanzees, and 9 rhesus macaques by brain transcriptome is dramatically remodeled during postnatal de- using Affymetrix GeneChip Human Genome (GC HG)-U133 Plus velopment and that developmental changes in the human brain are 2.0 microarrays (see Materials and Methods and SI Appendix, Table indeed delayed relative to other primates. This delay is not uniform S1). For both humans and chimpanzees, the individuals’ age across the human transcriptome but affects a specific subset of genes distributions cover the entire span of postnatal ontogenesis, with a that play a potential role in neural development. particular focus on early life stages (Fig. 1A and SI Appendix, Fig. S1). In these individuals, we reliably detected and quantified the human evolution ͉ brain development ͉ gene expression ͉ heterochrony ͉ expression of 7,958 genes (Materials and Methods). Among these chimpanzee genes, we first analyzed the relative influence of 3 factors—age, sex, and species—on total expression variation among individuals. umans differ from their closest living relatives, chimpanzees, in Quantitatively, age explains the largest part of the total expression Hbrain size and numerous cognitive traits (1–5). Humans also variation at 29%, followed by species at 17% and sex at Ͻ2% (Fig. differ from chimpanzees in the timing of particular developmental 1B). Although the effects of age and species are highly significant landmarks. For example, female sexual maturity is reached between (permutation test, P Ͻ 0.001), the effect of sex is not (P ϭ 0.54) (SI 8 and 9 years of age in chimpanzees and between 13 and 14 years Appendix, Table S2). Thus, in our dataset, age has by far the greatest in humans (6, 7). Studies of comparative primate morphology, some influence on expression levels. Consistently, we find that a striking dating back to the 19th century, suggested that human ontogenesis 71% of the 7,958 genes expressed in the human brain change proceeds at a slower rate than in other primates; consequently, adult significantly during postnatal development [at a false discovery rate humans retain features characteristic of juvenile primates. This type (FDR) of 10%] (Fig. 1C). Functionally, these genes are significantly of heterochronic shift is known as neoteny (see ref. 6 and references enriched in a range of biological processes that include cell adhe- therein). Neoteny has been ascribed a central role in human sion, synaptic transmission, and axonogenesis (permutation test for evolution (8); for example, as a possible explanation for the overall enrichment, P ϭ 0.002) (SI Appendix, Table S3). emergence of human-specific cognitive abilities through an ex- Next, we estimated when during human and chimpanzee brain tended period of high neuronal plasticity (4, 6, 9). development these expression changes take place. We find that in To date, human and chimpanzee ontogenesis have mainly been both species gene expression changes occur most rapidly during the compared in terms of skeletal morphology. Results from these first few years of life. Approximately 50% of the total expression comparisons indicate that neoteny may indeed explain some human change observed between newborns and 40-year-olds occurred features, such as small jaws (10). They also show that neoteny is not within the first year of life (Fig. 1D). Furthermore, the overall a ubiquitous feature of the human phenotype (10–14). The reason trajectory of age-related expression changes in the chimpanzee EVOLUTION for the large human brain size, for example, appears to be rapid brain, although based on fewer samples, closely resembles the early postnatal brain-growth rates rather than an extended brain- growth period in human infants (3). Meanwhile, the timing of human ontogenesis relative to that in other primates at the molec- Author contributions: S.B., M.J.W., C.S.W., M.L., S.P., and P.K. designed research; H.F., Z.Y., ular and histological levels remains unexplored. For instance, it is A.L., and B.N. performed research; A.L., S.G., T.G., J.K., M.D., S.B., M.J.W., and C.S.W. contributed new reagents/analytic tools; M.S., M.L., and P.K. analyzed data; and M.S., M.L., unknown whether all genes expressed in the human brain show a S.P., and P.K. wrote the paper. consistent delay in expression timing relative to the chimpanzee or, The authors declare no conflict of interest. alternatively, whether different structures or molecular networks This article is a PNAS Direct Submission. are affected to different extents. More generally, how the transcrip- Freely available online through the PNAS open access option. tome as a whole is affected by evolutionary shifts in developmental Data deposition: The data reported in this paper have been deposited in the Gene timing is an open question. Although studies in model organisms Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (GEO accession nos. have previously documented how changes in gene expression timing GSE11528, GSE11512, and GSE15163). during development can produce morphological and functional 1To whom correspondence may be addressed. E-mail: [email protected] or novelties (15, 16), this type of evolutionary change has not yet been [email protected]. investigated on a genome-wide scale. 2M.L., S.P., and P.K. contributed equally as supervisors of this study. More than 30 years ago, M. C. King and A. Wilson (17) proposed This article contains supporting information online at www.pnas.org/cgi/content/full/ that identifying differences in the timing of gene expression during 0900544106/DCSupplemental. www.pnas.org͞cgi͞doi͞10.1073͞pnas.0900544106 PNAS ͉ April 7, 2009 ͉ vol. 106 ͉ no. 14 ͉ 5743–5748 Downloaded by guest on October 2, 2021 Fig. 1. Expression variation during pri- 60 mate and rodent brain development. (A) A 44 B The first 2 principle components of the hu- 80 7 12 denialpx 50 man and chimpanzee DLPFC dataset. The 6 351212 60 Hominids numbers represent each individual’s age in Rodents years. The first and second components ex- 40 40 e Expected 0.1 ecna plain 25% and 15% of the total variance 0.1 2CP 0 1 30 and are significantly correlated with age 20 0 0.50 47 Ϫ i (r ϭ 0.86, P Ͻ 10 16) and species identity (r ϭ r 25433846 a Ͻ Ϫ16 0 2536 % v 20 0.84, P 10 ), respectively. Red, humans; 125 2018222423 blue, chimpanzees. (B) The mean propor- 1817 −20 3 5171213 tion of the total variance explained by sex, 0.2 2 588 10 0.2 0.3 2 species identity, and age across all ex- 0.20.1 0.40.3 0.9 −40 0.1 0.80.40.50.5 pressed genes. The values for 39 humans 0 and 14 chimpanzees (orange bars, left) are −100 −50 0 50 Sex Species Age based on 7,958 genes. The values for ro- PC1 identity dents (yellow bars, right) are based on ● ● 8,362 genes measured in 18 individuals. The C D noisserp 100 expected values are calculated as the me- ● ● ● dian of 1,000 permutations of each factor. ● ● ●● ● 80 ● ●●● ● Note that the proportion of variance ex- Age+,Sp− Age+,Sp+ x ●● ● ● e ● ●● ● ● plained by sex does not exceed the random l ● ● ● %37 %34 a ●● tantso ● expectation in humans and chimpanzees, (2914) (2739) 60 ● ● ●● whereas in mice it is not estimated, because ●● only males were used. (C) Proportions of p 40 ● ni egn ni ● ● age-related genes and genes showing sig- %16 %13 ● nificant expression differences between ● Human (1254) (1051) 20 ●● humans and chimpanzees in the DLPFC a ● hc % hc Chimpanzee ϩ Ϫ ● transcriptome. Age /Age represents ● genes showing/not showing a significant Age−,Sp− Age−,Sp+ 0 ϩ expression difference with age, and Sp / 010203040 SpϪ represents genes showing/not show- ing a significant expression difference be- Age (years) tween species.