Schizophrenia Research Forum - A Catalyst for Creative Thinking

GWAS and Anatomy—Pooled Data Finger Genes Driving Brain Size, IQ

Adapted from a story that originally appeared on the Alzheimer Research Forum.

8 May 2012. Several international teams have pooled genomic data and brain scans from thousands of people, and report several new loci linked to hippocampal and intracranial volumes in four papers posted online April 15 in Nature Genetics. Besides demonstrating the power of collaborative research with massive datasets, the findings point to new common variants in regulating brain size, with potential relevance to a number of brain disorders, and particular relevance to the study of intermediate or "endo"-phenotypes in psychiatric illness.

Structural neuroimaging has a long history in schizophrenia, driven by the hope that it might guide the research lens to the elusive lesion in the disorder. The jury is still out on whether and where structural changes occur in the disorder (see review of the evidence by Shepard et al., 2012), but since the hippocampus is among the better supported structures, these new data may be of interest to the schizophrenia community.

The current research was initiated by two multinational consortia. Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) began as a collection of U.S. and European elderly cohorts for genomewide association studies in cardiovascular health, and later grew to include neuroimaging. Sudha Seshadri at Boston University, Massachusetts, leads the CHARGE neurology subgroup and was a principal investigator on the new work. Enhancing Neuro Imaging Genetics Through Meta-Analysis (ENIGMA) focuses on brain disorders, but looks across the lifespan, and so includes a wider spectrum—from healthy young adults to seniors with dementia. Paul Thompson of the University of California, Los Angeles, headed the ENIGMA team.

Each consortium scoured its participants’ genomes for single nucleotide polymorphisms (SNPs) associating with brain measures on magnetic resonance imaging (MRI) scans, then sent their top hits to the other consortium, and additional researchers, for validation. No SNP associated with total brain volume, but several gene variants tracked with hippocampal size or intracranial volume, which measures total space within the skull, irrespective of brain size.

The CHARGE team—led by Seshadri, along with Charles DeCarli of the University of California, Davis, and M. Arfan Ikram of Erasmus Medical Center in Rotterdam, the Netherlands—analyzed genotyping and MRI data from 9,232 cognitively intact seniors. They found two loci on chromosome 12 that correlate with hippocampal volume. The stronger hit was a 12q24 SNP between genes HRK and FBXW8, which encode proteins that regulate apoptosis and ubiquitination, respectively. The second locus, at 12q14, included two SNPs—one in the intron of an enzyme (MSRB3) involved with oxidative stress, the other between genes encoding a Wnt signaling protein (WIF1) and transforming growth factor-β antagonist (LEMD3). People with one of these SNPs had, on average, smaller hippocampi equivalent to an extra four to five years of aging.

The ENIGMA effort—headed by Thompson and first author Jason Stein of the University of California, Los Angeles—also turned up a chromosome 12q24 SNP that associates with reduced hippocampal volume. This polymorphism lands in between genes, the closest being one called TESC, which is expressed in the hippocampus and influences brain development. Moreover, the 12q24 variant not only knocks off the equivalent of three years of brain aging, it also correlates with TESC expression. The latter was demonstrated by collaborators in London who measured TESC levels in hippocampal tissue from epilepsy patients.

The other major hit from ENIGMA—which analyzed more than 21,000 people—was a chromosome 12q14 polymorphism that associates with greater intracranial volume and brain size. Found near the 3’ untranslated region of HMGA2, a protein important for stem cell renewal, this SNP seems to influence not just brain size but also function. “By having a single switch from a T to a C, overall brain size was boosted about 0.5 percent—and if you had two C alleles, it would go up by 1 percent,” Thompson told Alzforum. C carriers also had higher IQ scores by about 1.3 points per allele. An SNP near HGM2A also came up as a top hit in an independent search for genes associated with infant head circumference, also reported this week in Nature Genetics. Greater head growth during the first year of life has been linked to higher childhood IQ (see Gale et al., 2006).

The infant study—a meta-analysis of seven genomewide association studies involving 10,768 people in pregnancy and/or birth cohorts—came up with two additional hits. One SNP, at 12q24, falls within a gene implicated in cancer (SBNO1). The second lies within the chromosome 17q21 inversion that includes MAPT and GRN. These genes encode tau and granulin, which have mutations linked to neurodegenerative diseases, including AD and frontotemporal lobar dementia.

The CHARGE study also picked up a signal in 17q21, which correlated with intracranial volume in healthy elderly. Taken together, the infant study and CHARGE data suggest that genes in this chromosome 17 region might link early brain growth with neurological disease, such as Alzheimer’s, in adulthood. However, such a direct relationship cannot be established at present, suggested H. Rob Taal of Erasmus Medical Center in an e-mail to Alzforum. The findings “give clues for further research to investigate exact underlying consequences of these associations, investigate if specific parts of the brain or brain development are involved, and how this affects neurodevelopment in early and later life,” noted Taal, who was first author on the infant study. A second CHARGE SNP linked to intracranial volume came up in a known height-associated locus on chromosome 6q22.

Though most were replicated in the other consortium, the key SNPs from ENIGMA did not initially come up as major hits in CHARGE, and vice versa. This could have stemmed from differences in the cohorts (e.g., ENIGMA included people with dementia and psychiatric disease, as well as younger individuals, whereas CHARGE was all dementia-free elderly) as well as procedures. For example, ENIGMA used automated tools (FreeSurfer) to calculate intracranial volume, while CHARGE used mostly hand-traced methods, DeCarli noted. Seshadri told Alzforum that a combined meta-analysis of ENIGMA and CHARGE data is in the works.—Esther Landhuis.

Stein JL et al. Identification of common variants associated with human hippocampal and intracranial volumes. Nat Genet. 2012 Apr 15. Abstract

Bis JC et al. Common variants at 12q14 and 12q24 are associated with hippocampal volume. Nat Genet. 2012 Apr 15. Abstract

Ikram MA et al. Common variants at 6q22 and 17q21 are associated with intracranial volume. Nat Genet. 2012 Apr 15. Abstract

Taal HR et al. Common variants at 12q15 and 12q24 are associated with infant head circumference. Nat Genet. 2012 Apr 15. Abstract

Comments on News and Primary Papers

Primary Papers: Identification of common variants associated with human hippocampal and intracranial volumes.

Comment by:  Patrick Sullivan, SRF Advisor
Submitted 20 April 2012
Posted 23 April 2012
  I recommend this paper

This is one of a series of three papers in Nature Genetics of relevance to students of the brain. It uses genomewide association studies of MRI phenotypes, with N's well in excess of 10,000.

This must be one of the most complex meta-analyses done in genetics. In addition to the now familiar issues of analyzing genetic data across multiple sites, the authors tackled harmonizing imaging phenotypes across many different scanners.

They identify robust and replicated candidate loci at levels of significance far smaller than chance. As is now familiar, the loci that emerge were rather unexpected, and not part of a typical list of "the usual suspects." How these loci influence MRI phenotypes could contain interesting biology.

It will be interesting to see if their findings do or do not line up with those from large studies of psychiatric disorders.

A much-debated issue in the field is whether imaging endophenotypes would prove to have simpler genetic architectures, perhaps characterized by larger effect sizes. The present studies do not support this conjecture. Indeed, imaging phenotypes seem to be rather typical complex traits. As the authors wrote:

"It has previously been hypothesized that brain imaging endo-phenotypes would have large effect sizes; however, this has proven not to be the case for the specific volumetric traits measured here, which had comparable effect sizes to those observed in other genome-wide association studies of complex traits (ref 40). Notably, the discovery sample had 99.92% power to detect variants with effect sizes of 1% of the variance for MAF ≥ 0.05."

View all comments by Patrick Sullivan