Candidate genes were examined by the best researchers using larger and larger samples and sophisticated statistics; a few were identified as prime suspects. Most results could not be replicated, but a few loci were very suspicious based on multiple studies.
Now, in an article by in this month's American Journal of Psychiatry, Saunderset al. report on 433 SNPs associated with 14 candidate genes that were prime suspects for schizophrenia in about 1900 cases and 2000 controls of European ancestry. The results? Not one of the genes was significantly associated with schizophrenia prevalence.
Even a 25% increase would have been detected with high probability.
An editorial by Steven Hamilton doi: http://dx.doi.org/10.1176/appi.ajp.2008.08020218 tries to put the best possible face on the results by noting that studies of tens of thousands of subjects were required to find genes that contribute to real but small (<25%) increases in risk for Type II diabetes. But that is not the point. Sanders, et al. deserve commendation for stating their conclusion clearly:
Our results suggest that, taken together, common DNA variants in these 14 genes are unlikely to explain a large proportion of the genetic risk for schizophrenia in populations of European ancestry. More robust findings are likely to be discovered using genome-wide association methods and, as our knowledge of the biology of mental illness continues to improve, focused studies of genes based on more precise mechanistic hypotheses. Nevertheless, although larger samples could possibly detect small genetic effects that were missed in this experiment, our findings suggest it is unlikely that true associations exist at the population level for the alleles that have formed the basis for the large candidate gene literature for these 14 postulated schizophrenia candidate genes.
Now what? Should we just do larger and larger studies with fancier and fancier bioinformatics? We have been looking for abnormal genes--mutations that cause diseases. But what if that is not the right model? That presumes that there is a normal genome and if all is in order all works fine, but when a part breaks, disease results.
A clue comes from Craig Ventner's genome. The human genome project provided sequences for haploid genomes. But the chromosomes from both Ventner's father and mother have now been sequenced. The results are a big surprise. Variation between human individuals is five times higher than we thought: 0.5% instead of 0.1%. Much of the difference is in the number of copies of a gene, and their locations. DOI: 10.1126/science.317.5843.1311
Copy number variations look likely to explain a lot. They are invisible to genetic testing that just looks for the presence of certain sequences. But they are important. Especially for mental disorders.
In this week's Science,Walsh, et al. report big differences in CNVs in people with schizophrenia: "Novel deletionsand duplications of genes were present in 5% of controls versus 15% of cases and 20% of young-onset cases" DOI: 10.1126/science.1155174 In previous work they have found similar differences in autism.
This may well explain why we have not been able to find the genes for schizophrenia--schizophrenics don't have different genes from other people, just different numbers of certain genes. This also fits with paternal age effects on schizophrenia -- the risk of schizophrenia increases as the father's --but not the mother's--age increases. (Male gamtes keep dividing throughout out life, increasing the risk of errors, while the eggs of females are all formed by birth)
So, myriads of different genetic variations may contribute to schizophrenia, many involving micro insertions and deletions. This tells us where to look.
A big piece of the puzzle remains missing, however. Why can so many different genetic variations all cause schizophrenia? Part of the answer is heterogeneity of the phenotypes--we should talk about the schizophrenias, in the plural. Nonetheless, it is remarkably that the brain fails so often in the same general ways. Why are bipolar disorder and schizophrenia so common compared to any number of other disorders, and the myriads of disorders that could exist but don't? The answer will come, I think, when we quit thinking of the body as a machine designed by engineers in which problems are caused by broken single parts. Bodies are fundamentally different from machines. Genes that make traits that on average tend to Darwinian fitness become more common. They form networks and modules, but in ways that often do not correspond to anything a sensible engineer would do. They create robust networks that are resistant to damage, until, that is, some slight variation wrecks the whole system. This may be why certain cognitive system are so vulnerable.
My best guess is that a cliff-edge effect is involved. Some trait has given such a large advantage that it has been pushed rapidly by selection to a value that is close to a cliff-edge, where the system is prone to fail catastrophically. Levels of uric acid in humans are a good example. Uric acid levels have increased in humans relative to other primates, probably because the antioxidant effects of uric acid are selected for in a a species with a long life span, despite the risk of gout. The strong correlation between uric acid levels and life span in primates is supportive evidence. For schizophrenia, Crespi summarize relevant evidence for signals of positive selection on candidate genes.
There are many other ideas out there. Bernie Crespi's work on the possibility that autism and schizophrenia are flip sides of conditions resulting from imprinted genes that advance maternal and paternal genetic interests is particularly intriguing.
We are getting there. But it is increasingly clear that it is a serious mitake to think of the brain as a machine with parts that break. The brain is, instead, an organ in an evolved soma whose information code is nothing like anything a any human programmer would write. It is not irreducibly complex, but it may well be incomprehensibly complex at the molecular level. Deeper evolutionary thinking about genomics may prove essential to understanding schizophrenia and autism.