6 December 2007. A new type of genetic analysis has uncovered novel risk factors for schizophrenia. Writing in this week’s PNAS online, researchers led by Todd Lencz and Anil Malhotra at the Zucker Hillside Hospital, Glen Oaks, New York, report that specific runs of homozygosity (ROH)—regions of DNA where individuals inherited the identical material from both parents—are more prevalent in patients with schizophrenia. Some of these runs contain or lie close to genes that have been previously implicated in the disease, while others may harbor new genetic risk factors. The new analysis, called whole-genome homozygosity association (WGHA), is better able to find associations with recessive alleles than traditional methods, according to Lencz. “The data suggest that in a certain subset of patients, schizophrenia could resemble an autosomal, recessive disorder,” he told SRF, though he cautioned that these findings need to be replicated. (These results were covered in brief in SRF meeting summaries from the 2007 WCPG and 2007 ICOSR.)
Whole-genome homozygosity association is an extension of the whole-genome association (WGA; also known as genomewide association, or GWA) studies with which researchers are already familiar. In the latter, DNA arrays, which can simultaneously identify more than 500,000 single nucleotide polymorphisms (SNPs), are used to compare genotypes among cases and controls. WGA has emerged as a powerful way to find single SNP variations that associate with disease. In fact, Lencz and colleagues were the first to apply this technology to the study of schizophrenia (see SRF related news story). But now they have taken WGA technology one step further. Instead of focusing on single SNPs, the researchers look for clusters of SNPs that show no variation in a given individual, then ask if any of these homozygosity runs are related to disease, in this case schizophrenia.
Lencz and colleagues first looked for runs of homozygosity that were common among 144 normal Caucasian controls. An ROH was defined as any segment of DNA with 100 or more consecutive SNPs that are identical in at least 10 individuals. Using this criterion, the researchers identified 339 common ROHs in their sample population. Nine of them were very common, appearing in over 25 percent of the controls.
To validate their findings, the researchers compared ROHs from their population with data from the human HapMap project (see SRF related news story). They predicted that ROHs in their Caucasian sample set should overlap with ROHs in the HapMap European sample set, but less so in the Asian and African sample sets. Sure enough, they found that of the 32 ROHs identified in the HapMap data, all but one coincided with ROHs found in the Zucker Hillside Hospital cohort. In addition, the four most common ROHs detected from the European HapMap data were also among the five most common ROHs in the hospital data set. In contrast, ROHs from the African HapMap data showed no commonality with the hospital dataset, and though there was some overlap with ROHs identified in the Asian HapMap dataset, the most common Asian ROH was only the fortieth most common in the hospital cohort. As a whole, these relationships turned out as one might predict based on differences in population ancestry and, the authors write, suggest that the ROHs identified in the Zucker Hillside Hospital cohort do not arise because of some technical artifact.
With this validation, Lencz and colleagues next asked how ROHs are represented among a sample set of 178 Caucasian schizophrenia cases that were age- and sex-matched to the 144 controls. They found that nine ROHs occurred in patients at significantly different frequencies than in controls—all nine were more represented in patients. One ROH, harboring the gene for dynein cytoplasmic 2 heavy chain 1 protein (DYNC2H1) on chromosome 11, was found exclusively in patients (in 8 percent of cases), while five other ROHs were very uncommon in controls.
What do these new data tell us about schizophrenia? For one thing, they may bolster previously reported genetic associations. A check of SchizophreniaGene showed the researchers that four of the nine ROHs that are more common in patients harbor or are in close proximity to genes previously associated with the disease. These genes include NOS1AP, coding for a protein that binds to neuronal nitric oxide synthase and which may modulate glutamate receptor activity; ATF2, coding for activating transcription factor 2, which is elevated in postmortem brain samples (see Kyosseva et al., 2000); the NSF gene, coding for N-ethylmaleimide-sensitive fusion, a protein that binds to glutamate AMPA receptors (see Mirnics et al., 2000); and PIK3C3, the gene for phosphoinositide-3-kinase, class 3. The other five ROHs may contain novel genes that could be risk factors for schizophrenia. The MAPT gene encoding the microtubule binding protein tau, is one example. Tau has been linked to numerous and neurological disorders, including Alzheimer’s disease, and supranuclear palsy.
But the research may have broader implications, too, and not just for schizophrenia. It hints that the extent to which a person is homozygous at any given locus may put them at risk, suggested Lencz. “When you are heterozygous there is a 100 percent chance that you have at least one copy of the ‘optimal’ variant in a gene; homozygosity may result in missing an allele that can be critical for healthy development,” he said. Interestingly, while 55 percent of controls had none of the nine risk ROHs, only 19 percent of patients were risk ROH free. Furthermore, the researchers calculated that as the number of risk ROHs increases, so does the relative risk of having the disease, such that in those with three risk ROHs (14 percent of cases and only 2 percent of controls), the odds ratio of having schizophrenia increased to 24.
Of course, being homozygous means that the same allele has been inherited from both parents, which is the case in many simple autosomal recessive disorders. In general, such disorders are much rarer than schizophrenia, occurring in about 0.01 percent of the population (for schizophrenia, estimates hover around 1.0 percent), and studies on genetic transmission suggest that schizophrenia does not follow a simple Mendelian pattern. But it is possible that in a subset of patients, schizophrenia is caused by a simple recessive inheritance, Lencz suggested, though he stressed that it is more likely that these recessive ROHs just increase a person’s risk for getting the disease. “The main point is that ROHs may be overrepresented in schizophrenia and that the disease might particularly relate to those regions of the genome,” said Lencz. He also said that he is eager to engage the field in replication of these studies so as to get a better estimate of effect sizes.—Tom Fagan.
Lencz T, Lambert C, DeRosse P, Burdick KE, Morgan TV, Kane JM, Kucherlapati R, Malhotra AK. Runs of homozygosity reveal highly penetrant recessive loci in schizophrenia. PNAS December 5, 2007 online.