13 October 2006. Researchers from Massachusetts General Hospital in Boston and the Broad Institute, a joint venture of Massachusetts Institute of Technology and Harvard University, have come to some potentially controversial conclusions in the field of schizophrenia susceptibility genes. Writing in the November issue of The American Journal of Human Genetics, Pamela Sklar and colleagues suggest that dysbindin (DTNBP1) may not be a susceptibility gene for schizophrenia, in disagreement with a number of previous studies.
The researchers utilized the most recent version of the International HapMap to construct a high-resolution map of the DTNBP1 genomic region. When haplotypes that had been previously identified by other groups as associated with schizophrenia risk were then mapped onto the high-resolution map, the researchers found no consensus. This suggested to the authors that the role of DTNBP1 in schizophrenia should be re-evaluated in large-scale studies using their new, high-resolution map.
Previous studies of DTNBP1
A number of heritability studies have been performed among patients with schizophrenia, but it has been difficult for researchers to identify replicable susceptibility genes across studies. Currently, there are several candidate genes for which there is either compelling or promising evidence (for review, see Ross et al., 2006; Straub and Weinberger, 2006). One of the genes with the most compelling evidence is dysbindin (DTNBP1).
DTNBP1 was first identified in a study of schizophrenia-affected Irish pedigrees (Straub et al., 2002; van den Oord et al., 2003); since that time, five other studies have found an association of schizophrenia with DTNBP1 in samples of European ancestry. What has been confusing with this data, however, is that the associated alleles and haplotypes have differed among samples. No definitive causal variant of DTNBP1 that contributes to schizophrenia risk has been identified. Adding to the confusion is the fact that the same single nucleotide polymorphisms (SNPs) have not been genotyped across the six studies, preventing direct comparisons.
Lead authors Mousumi Mutsuddi and Derek Morris and colleagues attempted to make the data from these six studies directly comparable by using data from the International HapMap project (see SRF related news story). The HapMap project has genotyped SNPs from 270 individuals, grouped into four, roughly geographical locations: Yoruba in Ibadan, Nigeria; Japanese in Tokyo, Japan; Han Chinese in Beijing, China; and Utah residents with ancestry from Northern and Western Europe (abbreviation: CEU; genotyped in 1980 by the Centre d’Étude du Polymorphisme Humain). The CEU sample comprised 30 parent-offspring trios.
Mutsuddi and colleagues constructed a high-resolution haplotype map for the DTNBP1 region by using the CEU sample from HapMap. To supplement the SNPs already identified by the HapMap project in this sample (113 SNPs), SNPs from two other sources were genotyped using CEU sample DNA. First, Mutsuddi and colleagues genotyped CEU sample DNA for all the SNPs reported in the six association studies (29 SNPs). Second, they genotyped the CEU sample for SNPs across the DTNBP1 region which were taken from the Single Nucleotide Polymorphism database (24 SNPs). Thus, the total number of SNPs on the finished map was 166.
The program Tagger (Haploview) was then used to determine tagging SNPs from this map. Because SNPs in close proximity tend to be inherited together, a single tagging SNP (tSNP) can stand as proxy for a grouping of several SNPs. Based on six key tSNPs, the researchers found that the CEU sample contained five common haplotypes in the DTNBP1 region. They constructed a phylogenetic tree to show how each haplotype evolved from a common, ancestral haplotype.
Comparing the studies
To compare the association studies to each other in reference to the CEU-derived haplotype phylogenetic tree, Mutsuddi and colleagues noted the SNPs in each study that defined the “associated” haplotype for that study. They then took each associated haplotype and mapped it onto the phylogenetic tree containing the most common five haplotypes in the CEU sample. They assumed that if a haplotype were truly associated with an increased risk of schizophrenia, there would be a consensus among the six studies on the phylogenetic tree.
In fact, that is not what they found. The association from Kirov and colleagues (Kirov et al., 2004) mapped to haplotypes 1, 2, 4, and 5. The association from Bray and colleagues (Bray et al., 2005) mapped to haplotypes 2 and 5. The association from Schwab and colleagues (Schwab et al., 2003) mapped to haplotypes 1 and 2. The association from van den Oord and colleagues (van den Oord et al., 2003) mapped to haplotype 3. The strongest associations from Van Den Bogaert and colleagues (Van Den Bogaert et al., 2003) and Funke and colleagues (Funke et al., 2004) mapped onto haplotype 4. In sum, each of the five most common haplotypes identified by a high-resolution SNP map of the DTNBP1 region in the CEU sample was found to be “associated” with risk of schizophrenia in one or another of the six studies.
Mutsuddi and colleagues also found that the haplotypes analyzed in the six studies were present in the CEU sample at roughly the same frequency. “This suggests that each European-derived sample is genetically similar and that population stratification cannot explain differences in published results,” they write.
“Because we find that all of the association samples of European-derived ancestry have a similar genetic structure, the conflicting results among studies cannot simply be attributed to population differences. This calls into question the interpretation of the replication studies at this locus,” the authors conclude. They called for further, large-scale studies using their high-density map to help determine how DTNBP1 contributes to schizophrenia susceptibility.—Jillian Lokere.
Mousumi Mutsuddi, Derek W. Morris, Skye G. Waggoner, Mark J. Daly, Edward M. Scolnick, Pamela Sklar. Analysis of high-resolution HapMap of DTNBP1 (Dysbindin) suggests no consistency between reported common variant associations to schizophrenia. Published online Oct 3 in Am J Hum Genet. Abstract