Schizophrenia Research Forum - A Catalyst for Creative Thinking

Gene Expression Study May Open Window on Brain Development

15 June 2009. A study of gene expression across the entire genome in developing human brains may lay the groundwork for an improved understanding of the origins of complex mental disorders. Nenad Sestan of Yale University in New Haven, Connecticut, and colleagues used advances in microarray processing to examine gene expression at the level of individual exons, the code-carrying parts of genes. In the May 28 issue of Neuron, the researchers report finding different patterns of expression, including alternative splicing, in different regions of the prenatal brain, even within the prefrontal cortex. The patterns differentiated themselves according to the regions’ functional roles.

The study’s first author, Matthew Johnson, of Yale University in New Haven, told SRF that the study sprang from the Sestan lab’s interest in how the cerebral cortex evolved to enable humans to do what other species cannot. He and his collaborators suspected that changes in gene expression (see SRF live discussion) might play a role in the development of humans’ advanced brain wiring. One way of changing gene expression involves alternative splicing, which makes different forms of messenger RNA from the same gene (see SRF related news story; SRF news story). The researchers wanted to know how much differential gene expression and alternative splicing occur during prenatal development in humans and whether various brain areas would differ in expression.

Expression shapes function?
Johnson and colleagues studied four apparently normal brains gathered at 18 to 23 weeks of gestation, a key time in the formation of neural circuits and structural asymmetry. Their study looked at gene expression and alternative splicing in 13 brain areas.

When they looked at the neocortex, hippocampus, striatum, thalamus, and cerebellum, the researchers found that 76 percent of genes were expressed in at least one brain region. Of these, 44 percent appeared to be regulated differently by brain region. In particular, they found differential expression of about 33 percent of the genes and signs of alternative splicing for 28 percent. Genes in the neocortex and hippocampus shared some similar expression patterns, while the cerebellum showed the most distinct ones.

Next, the researchers explored gene expression and splicing in nine neocortical areas. Those included the motor-somatosensory, parietal association, temporal auditory, temporal association, and occipital visual neocortex, as well as four prefrontal areas: specifically, the orbital, dorsolateral, medial, and ventrolateral. Following the notion that differential gene regulation molds brain development, they expected that fetal brains would show more differential gene expression than mature adult brains, which typically show the same expression across neocortical areas.

Functionally distinct areas of the prefrontal cortex showed similarities and differences that may offer a glimpse of the developmental process. For instance, the ventrolateral prefrontal cortex, which contains Broca’s speech area, correlated more strongly with the motor-somatosensory cortex than with other prefrontal areas. To the researchers, this means that these areas may resemble each other at the molecular level at this developmental stage.

The researchers further found enrichment of certain genes in the perisylvian cortical speech and language areas, which, despite being scattered across the lobes, share similar functions. They wondered whether differential gene expression could account for the emergence of between-hemisphere differences in the perisylvian areas that relate to handedness, speech, and language processing. However, they found no consistent asymmetry of gene regulation across all the brains.

The study did identify functionally defined gene clusters that were most enriched in particular areas. For instance, one group found in excess in the cerebral cortex fosters axon guidance. It includes, for instance, semaphorin 3A.

Taken together, the expression and splicing differences may explain how various parts of the brain develop their specialized roles. The findings also highlight the complexity of expression in the developing brain. As Johnson said, “There are genes expressed at this time that are critical for forming connections that are then not expressed later in the mature brain.”

“A lot of work previously—because it’s been technologically easier to do—has been focused on finding differences in the coding sequences, but not as much has been done looking at differences in the noncoding elements that control the spatial and temporal expression of genes,” Johnson said. These cis-regulatory elements reside near, but outside of, coding regions, and some are thought to have contributed to the evolution of the human brain. Consistent with that notion, Johnson and associates found that differentially expressed genes in the neocortex were twice as likely to appear near those elements that have shown faster evolution in humans than in other species.

The researchers also uncovered groups of co-regulated genes with similar connections. They suggested that genes with the most interconnections within a given network probably shape function in important ways. They hint that researchers might want to pay particular attention to these “hub genes.”

Although the study did not look at schizophrenia, the light it shines on the development of the nervous system someday may help set the stage for those who do. In their paper, Johnson and colleagues hint that differential gene expression may underlie the abnormal wiring seen in schizophrenia, especially in the prefrontal cortex. To date, only splice variation of the neuregulin 1 gene has received significant attention (see SRF related news story; SRF news story; SRF related news story).

Johnson says that these data can also help narrow the field of gene candidates that emerges from genomewide association studies. For example, he explains, if such a study implicates a stretch of DNA that holds 20 to 50 genes, “you can look at our data and see that only a handful, or only one, is highly expressed in the brain during development at the time that we think these circuits that are miswired in the disorder are being formed.” That kind of information “can start to give us a better handle on the actual processes that are going on in the human being during development and that go wrong in schizophrenia or other such disorders,” he adds.

In a commentary in the same issue of Neuron, Colette Dehay of the Stem Cell and Brain Research Institute in Bron, France, and Henry Kennedy of the Université de Lyon, France, likened the data set to the best gold mine. Fortunately, researchers enticed by these findings need not experience data envy. Rather, they can test their own ideas using the publicly available data (see Gene Expression Omnibus and USCS Genome Bioinformatics).—Victoria L. Wilcox.

Johnson MB, Kawasawa YI, Mason CE, Krsnik Z, Coppola G, Bogdanovic D, Geschwind DH, Mane SM, State MW, Sestan N. Functional and evolutionary insights into human brain development through global transcriptome analysis. Neuron. 2009 May 28; 62:494-509. Abstract

Dehay C, Kennedy H. Transcriptional regulation and alternative splicing make for better brains. Neuron. 2009 May 28; 62:455-457. Abstract

Comments on News and Primary Papers
Comment by:  Barbara Lipska
Submitted 15 June 2009
Posted 15 June 2009

In this very important and innovative study, Sestan and colleagues report a transcriptome-wide survey across multiple brain regions of the fetal mid-gestation brain. They show dramatic differences in expressed transcripts, including alternative splice variants, between brain regions, and most surprisingly, between several cortical regions. The authors have undertaken an ambitious task of further characterizing differentially expressed genes by functional clustering and co-expression clustering and comparing the results with genes identified through neurobiological experiments. They have also performed extensive validation using several additional fetal brains. Most interestingly, the authors showed that differentially expressed genes are more frequently associated with human-specific evolution of putative cis-regulatory elements. For this, they have identified genes that are near highly conserved non-coding sequences (CNSs) and found that the genes that are differentially expressed between the regions are more frequently near human-specific accelerated evolution CNSs.

The weakness of the study is a very small number of fetal brains (four) and the fact that they range in age from 18 to 23 weeks of gestation. During these several weeks of fetal life, the brain undergoes dramatic developmental changes and expression of many genes, either increases or decreases steeply. Thus, it would be critical to fully characterize these changes across fetal age. It is also crucial to explore genetic influences on fetal gene expression as it appears that in adult brain both gene expression and splicing are strongly genetically regulated. The authors have made an important contribution to our understanding of development of human brain, and further research of this type will generate the data that would help in better understanding of human brain disorders. In particular, genetic-expression effects in human brain across the entire lifespan, including fetal period, may help identify molecular mechanisms whereby candidate genes increase risk for developing the disorder. Using expression levels of transcripts and their splicing characteristics as intermediate phenotypes may yield statistically positive associations and improved understanding of the mechanisms that lead to neurodevelopmental disorders such as autism and schizophrenia, as they are the most proximal phenotypes to the risk alleles.

View all comments by Barbara LipskaComment by:  Karoly Mirnics, SRF Advisor
Submitted 15 June 2009
Posted 15 June 2009

This outstanding study reinforces how much we still do not understand about human brain development and function! It is just mind-boggling that the mid-fetal human brain expresses more than three quarters of the human genome, and that region-specific splicing appears to be an absolutely critical feature of the developing brain. Interestingly, the structural and functional interhemispheric differences do not appear to be related to gene expression differences in mid-fetal life, but rather, either they develop independently of gene expression patterns, or they are developing at later stages of cortical maturation, perhaps in a postnatal activity-driven pattern.

So, how is this developmental expression machinery related to various neurodevelopmental disorders, such as schizophrenia? Is usage of an "inappropriate" splice variant sufficient to alter the neuronal phenotypic development to a degree that would predispose the brain to developing a disease? Are environmental insults capable of disrupting this finely tuned, region-specific splicing machinery? As this is a likely possibility, we must rethink the existing disease-related gene expression findings in the context of the present study, and accept that our previous gene expression measurements may have been too crude to uncover some of the most meaningful changes that are potentially hallmarks of various brain disorders. Furthermore, as the genes that show the most widespread regional use of splice variants can be essential for proper neuronal migration or connectivity, one can argue that these genes should be the primary targets for evaluation in the various regions of postmortem tissue of diseased individuals.

Finally, there is also a minor, cautionary note arising from this study. The fact that the Affymetrix U133 and the Exon array results showed a correlation of R2 >0.5 is encouraging, but underscores that platform-dependence of the findings remains a significant interpretational challenge. Some platforms will be better suited to identify certain gene expression changes, while others will have a greater power to reveal a different (but also potentially valid!) set of mRNA alterations.

View all comments by Karoly Mirnics

Comments on Related News

Related News: Multitasking Rett Protein Shines Spotlight on RNA Splicing in Neurologic and Psychiatric Disease

Comment by:  Mary Reid
Submitted 6 February 2006
Posted 8 February 2006

Gabrielle Strobel mentions that DLX5 has been suggested as an MECP2 target. Horike and colleagues report that "transcription of DLX5 and DLX 6 was reproducibly approximately two times higher in Mecp2-null mice" (Holrike et al., 2005). In view of the fact that DLX5 is necessary for osteoblastogenesis and osteoclastogenesis, might we suspect that its increased expression may explain the bone pathology reported in Rett syndrome? Does this resemble Paget disease and if so, might we expect that biphosphonates may be beneficial in Rett syndrome?

Horike S, Cai S, Miyano M, Cheng JF, Kohwi-Shigematsu T. Loss of silent-chromatin looping and impaired imprinting of DLX5 in Rett syndrome. Nat Genet. 2005 Jan;37(1):31-40. Epub 2004 Dec 19. Abstract

View all comments by Mary Reid

Related News: Polymorphisms and Schizophrenia—The Ups and Downs of Neuregulin Expression

Comment by:  William Carpenter, SRF Advisor (Disclosure)
Submitted 22 April 2006
Posted 22 April 2006
  I recommend the Primary Papers

Related News: Polymorphisms and Schizophrenia—The Ups and Downs of Neuregulin Expression

Comment by:  Stephan Heckers, SRF Advisor
Submitted 29 April 2006
Posted 29 April 2006
  I recommend the Primary Papers

The gene Neuregulin 1 (NRG1) on chromosome 8p has been identified as one of the risk genes for schizophrenia. It is unclear how the DNA sequence variation linked to schizophrenia leads to abnormalities of mRNA expression. This would be important to know, in order to understand the downstream effects of the neuregulin gene on neuronal functioning in schizophrenia.

Law and colleagues explored this question in post-mortem specimens of the hippocampus of control subjects and patients with schizophrenia. This elegant study of the expression of four types of NRG1 mRNA (types I-IV) is exactly what we need to translate findings from the field of human genetics into the field of schizophrenia neuropathology. The findings are complex and cannot be translated easily into a model of neuregulin dysfunction in schizophrenia. I would like to highlight two findings.

First, the level of NRG1 type I mRNA expression was increased in the hippocampus of schizophrenia patients. This confirms an earlier study of NRG1 mRNA expression in schizophrenia. It remains to be seen how this change in NRG1 type I mRNA expression relates to the finer details of neuregulin dysfunction in schizophrenia.

Second, one single nucleotide polymorphism (SNP8NRG243177) of the risk haplotype linked to schizophrenia in earlier studies predicts NRG1 type IV mRNA expression. The SNP determines a binding site for transcription factors, providing clues for how DNA sequence variation may lead, via modulation of mRNA expression, to neuronal dysfunction in schizophrenia. It is exciting to see that we can now test specific hypotheses of molecular mechanisms in the brains of patients who have suffered from schizophrenia. The study by Law et al. is an encouraging step in the right direction.

View all comments by Stephan Heckers

Related News: Polymorphisms and Schizophrenia—The Ups and Downs of Neuregulin Expression

Comment by:  Bryan Roth, SRF Advisor
Submitted 5 May 2006
Posted 5 May 2006
  I recommend the Primary Papers

I think this is a very interesting and potentially significant paper. It is important to point out, however, that it deals with changes in mRNA abundance rather than alterations in neuregulin protein expression. No measures of isoform protein expression were performed, and it is conceivable that neuregulin isoform protein expression could be increased, decreased, or not changed. A second point is that although statistically significant changes in mRNA were measured, they are modest.

Finally, although multiple comparisons were performed, the authors chose not to perform Bonferroni corrections, noting in the primary paper that, "Correction for random effects, such as Bonferroni correction, would be an excessively conservative approach, particularly given that we have restricted our primary analyses to planned comparisons (based on strong prior clinical association and physical location of the SNPs) of four SNPs and a single haplotype comprised of these SNPs. Because the SNPs are in moderate LD, the degree of independence between markers is low and, therefore, correcting for multiple testing would result in a high type II error rate. The prior probability and the predictable association between the deCODE haplotype and expression of NRG1 isoforms (especially type IV, which is its immediate physical neighbor) combined with the LD between SNPs in this haplotype makes statistical correction for these comparisons inappropriate. Nevertheless, our finding regarding type IV expression and the deCODE haplotype and SNP8NRG243177 requires independent replication."

It will thus be important to determine if these changes in neuregulin mRNA isoform abundance are mirrored by significant changes in neuregulin isoform protein expression and if the findings can be independently replicated with other cohorts.

View all comments by Bryan Roth

Related News: Polymorphisms and Schizophrenia—The Ups and Downs of Neuregulin Expression

Comment by:  Patricia Estani
Submitted 9 June 2007
Posted 10 June 2007
  I recommend the Primary Papers

Related News: Neuregulin and Schizophrenia—Functional Failure Fingers Risk Allele

Comment by:  Ali Mohamad Shariaty
Submitted 14 July 2007
Posted 14 July 2007

It is really a fascinating article which is a step towards understanding the molecular mechanisms underlying phenotypes of schizophrenia. Relating genotypes to phenotypes is really necessary for untangling the puzzle of a complex disorder. However, when a regulatory SNP interferes with normal binding of a transcription factor, is it understood that the transcription factor should play a role in brain and therefore in the molecular pathology of schizophrenia? Is there any direct role for involvement of serum response factor (SRF) in brain development or any neurological process?

View all comments by Ali Mohamad Shariaty

Related News: Neuregulin and Schizophrenia—Functional Failure Fingers Risk Allele

Comment by:  Amanda Jayne Law, SRF Advisor
Submitted 14 July 2007
Posted 15 July 2007

In response to Ali Mohamad Shariaty’s comment: Serum response factor (SRF) plays a key role in regulating the transcription of a number of genes involved in brain development. Genetic manipulation of SRF has revealed a direct role for it as a regulator of cortical and hippocampal function (e.g., Etkin et al., 2006) influencing both learning and memory. At the cellular level SRF has been shown to regulate dendritic morphology and neuronal migration. Therefore, SRF is indeed an important neurodevelopmental molecule, mediated via its regulation of genes, such as NRG1. Genetic variations that are predicted to interfere with SRF binding (such as the SNP characterized in our study) may affect critical aspects of brain development and function that contribute to schizophrenia. Since SRF regulates the expression of a number of genes, beyond that of NRG1, its involvement in schizophrenia is likely mediated “indirectly” via its effects on the regulation of genes associated with the disorder.


Etkin A, Alarcón JM, Weisberg SP, Touzani K, Huang YY, Nordheim A, Kandel ER. A role in learning for SRF: deletion in the adult forebrain disrupts LTD and the formation of an immediate memory of a novel context. Neuron. 2006 Apr 6;50(1):127-43. Abstract

View all comments by Amanda Jayne Law

Related News: Neuregulin and Schizophrenia—Functional Failure Fingers Risk Allele

Comment by:  Robert Hunter
Submitted 17 July 2007
Posted 17 July 2007
  I recommend the Primary Papers