Adapted from a story that originally appeared on the Schizophrenia Research Forum.
2 November 2011. A dynamic view of transcription in the human brain has been revealed in two papers published in Nature on 26 October 2011. Both studies tracked the expression patterns of tens of thousands of genes across the lifespan, and found that transcription is highest prenatally, slowing after birth—a pattern that presumably reflects the intricate work of building a brain. One study, led by Nenad Šestan of Yale University in New Haven, Connecticut, surveyed patterns of gene expression across 16 different brain regions at different ages. The second study, led by Joel Kleinman at the National Institute of Mental Health in Bethesda, Maryland, focused on the prefrontal cortex and found tight associations between individual gene expression and individual genetic variants but not between total genetic diversity and the entire transcription profile.
“There's a lot of genetic variation that makes us all unique, but the bottom line is that we have much more in common than not,” Kleinman told SRF.
These ambitious projects—involving thousands of genes, extensive postmortem brain collections, and plenty of analysis to make sense of it all—differ from previous surveys which were limited to a particular brain region or time point (e.g., Abrahams et al., 2007). The result is a comprehensive brain map of the ups and downs of gene expression during prenatal development, infancy, childhood, adolescence, and adulthood. This will serve as a reference for understanding the pathological events involved in brain disorders like schizophrenia. In particular, it can reveal the normal trajectories of expression for genes suspected in a disorder; it may finger new suspects by finding genes with tightly correlated patterns of expression with known risk factors; and when coupled with genetic information, it can start to outline the function of risk variants by probing their influence on gene expression.
“Ultimately, we want to find targets for treatment; this lets you take the first step in understanding how genetic variants increase risk for any brain disease,” Kleinman said.
Kleinman, whose group also collaborated with Šestan's on the first paper, was moved to look comprehensively at transcription patterns in the brain by earlier studies which linked schizophrenia-related variants to alternative transcripts that were preferentially expressed in fetal brain (e.g., Nakata et al., 2009). This made clear to him that the precise identity of a transcript, as well as its level of expression, across the lifespan would be essential for piecing together how genetic variants increase risk for a disease, he said.
Both studies have made their massive datasets freely available online. In addition, Tianzhang Ye of Kleinman’s group has created an application called BrainCloud that allows researchers to browse the expression patterns and related variants of any gene they are interested in. “There are a trillion pieces of information in the database, and we've mined it for some things that we're interested in, but we feel there is an infinite number of things it can be used for,” Kleinman said.
The first study took not one or two, but six first authors: Hyo Jung Kang, Yuka Imamura Kawasawa, Feng Cheng, Ying Zhu, Xuming Xu, and Mingfeng Li, all of Šestan's lab. The researchers assembled 57 postmortem brains from healthy donors, ranging in age from six weeks post-conception to 82 years old. Sixteen different brain regions were dissected out: 11 in the neocortex, plus the hippocampus, amygdala, striatum, thalamus, and the cerebellar cortex. RNA from these regions was extracted and quantified using a gene chip that probes 1.4 million whole transcripts or individual exons of 17,565 genes.
Apparently, change is the rule rather than the exception in the brain, with 90 percent of genes surveyed showing different expression patterns either in time, across different regions of the brain, or both. Most of this differential expression occurred prenatally, then settled down after birth. For example, while 57.7 percent of genes expressed in the neocortex changed their expression patterns temporally during fetal development, only 9 percent of these changed in postnatal development, and 0.7 percent did in adulthood. Spatially, the pattern of expression across regions became more similar after birth, with the holdout cerebellar cortex maintaining a transcription profile distinct from the rest of the brain. The researchers also monitored specific exon usage to get a handle on the dynamics of transcript diversity. They found that 90.2 percent of expressed genes showed signs of differential exon usage, with exon transcripts detected in some, but not all regions, or, conversely, at some, but not all time points. This indicates a precise orchestration of transcript type, location, and timing.
Sex, Modules, and Trajectories
A comparison of these transcription patterns between male and female brains highlighted 159 genes with sex-biased gene expression and 155 with sex-biased exon usage. The male-female differences were especially pronounced during fetal development, then faded after birth and into adulthood. Exons with sex-biased expression sometimes came from genes linked to brain diseases, including schizophrenia-related KCNH2 and autism-related NLGN4X. This suggests that the sex differences in the risk for certain disorders might stem from these sorts of transcriptional mechanisms.
Another way to distill meaning from the giant dataset is to look for groups of genes that co-vary their expression (see SRF related news story), which represent functional units in the brain (Oldham et al., 2008). The researchers identified 29 of these groups, called modules, with distinct patterns of spatial-temporal expression consistent with known developmental events. For example, one module with high expression early in fetal development in the neocortex, tapering off after birth, contained many genes involved in neural differentiation; in contrast, another module that ramped up its expression around birth was enriched for ion channels and neural ligand-receptor pairs. Interestingly, the “hub” genes in this module—those whose expression correlated the most with others in the module—included genes linked to depression (GDA) and schizophrenia (NRGN and RGS4). This suggests that these genes might be key drivers of the module’s transcriptional program.
Stepping away from this systems view, the researchers also examined the expression trajectories of individual genes implicated in autism and schizophrenia, including CNTNAP2, MET, NLGN4X, and NRGN. These revealed distinct spatial and temporal patterns of expression, which were then used to identify new suspects by looking for genes with similar patterns. For example, NRGN became highly expressed in the neocortex after birth, increased into late childhood, and remained high into adulthood; 50 transcripts with highly similar spatiotemporal expression patterns were identified (correlation coefficients greater than 0.80), and may be worth following up for a link to the disorder.
Ups and Downs in Prefrontal Cortex
In the second study, first authors Carlo Colantuoni and Barbara Lipska of the NIMH assembled postmortem samples of prefrontal cortex from an impressive 269 healthy individuals, aged between 14 weeks post-conception and 78 years old. RNA was extracted from each sample and analyzed with a microarray containing 30,176 gene expression probes. Looking globally at the rate of expression change of all genes—a measure of how much they increased or decreased expression in a year—the researchers found the highest rates in prenatal tissue, which then dropped fivefold in infancy, and dropped again 90-fold in childhood from the prenatal peak. Expression held steady in adulthood, but started to climb again with aging, starting in the fifties and eventually surpassing the rate in adolescent brains. These fluctuations seem driven by individual genes with expression trajectories having distinct turning points at the same time.
To get a sense of the kinds of genes involved in these transcriptional shifts, the researchers compared the expression patterns of the 1,502 genes experiencing significant expression change before and after the particularly abrupt fetal-infant transition. Most of these genes had high expression during fetal development which decreased in infancy, and included many genes involved in axonal function—possibly reflecting the pruning of unused axon terminals. Genes with the inverse pattern, with low fetal expression that rose in infancy, were enriched for ATP synthesis—something that might reflect the rising energy demands of a growing and maturing brain. Genes with decreases in both fetal and infant periods consisted of many related to cell division, consistent with the known tapering off of cell proliferation at this time. Genes with increases during both stages were dominated by those related to the synapse.
The second half of the study was devoted to looking for potential genetic control knobs for expression. The researchers compared 625,439 SNP genotypes from DNA samples of each donor to expression of 30,176 different probes (a possible 19 billion associations). This uncovered 1,628 significant associations between individual SNPs and individual gene expression, with the strongest between a gene called ZSWIM7 and a SNP located within the gene itself, giving a stratospheric p value of 5.4 x 10-78. The expression levels of this gene for each of the three different genotypes were nearly non-overlapping across life, suggesting a strict genetic control.
Despite this evidence of individual SNPs associated with individual gene expression, things looked different on a global scale. When the researchers calculated a genetic distance that described the genetic similarity of their donors, and a transcriptional distance that reflected the similarity in their expression patterns, this did not turn up any associations. This suggests that even diverse genomes produce a remarkably similar transcription pattern in the brain.
Both studies set the stage for more thorough examinations of the complex transcription patterns in health and disease. These descriptions of the full scale of the transcriptome's diversity in the brain are an essential first step for comprehending how dynamic transcript changes translate into protein levels, or even phenotypes.—Michele Solis
- Abrahams BS, Tentler D, Perederiy JV, Oldham MC, Coppola G, Geschwind DH. Genome-wide analyses of human perisylvian cerebral cortical patterning. Proc Natl Acad Sci U S A. 2007 Nov 6;104(45):17849-54. PubMed.
- Nakata K, Lipska BK, Hyde TM, Ye T, Newburn EN, Morita Y, Vakkalanka R, Barenboim M, Sei Y, Weinberger DR, Kleinman JE. .
- Oldham MC, Konopka G, Iwamoto K, Langfelder P, Kato T, Horvath S, Geschwind DH. Functional organization of the transcriptome in human brain. Nat Neurosci. 2008 Nov;11(11):1271-82. PubMed.
- Colantuoni C, Lipska BK, Ye T, Hyde TM, Tao R, Leek JT, Colantuoni EA, Elkahloun AG, Herman MM, Weinberger DR, Kleinman JE. Temporal dynamics and genetic control of transcription in the human prefrontal cortex. Nature. 2011 Oct 27;478(7370):519-23. PubMed.
- Kang HJ, Kawasawa YI, Cheng F, Zhu Y, Xu X, Li M, Sousa AM, Pletikos M, Meyer KA, Sedmak G, Guennel T, Shin Y, Johnson MB, Krsnik Z, Mayer S, Fertuzinhos S, Umlauf S, Lisgo SN, Vortmeyer A, Weinberger DR, Mane S, Hyde TM, Huttner A, Reimers M, Kleinman JE, Sestan N. Spatio-temporal transcriptome of the human brain. Nature. 2011 Oct 27;478(7370):483-9. PubMed.