Revisiting and expanding the largest genetic study of schizophrenia

In his short fiction “Blue Tigers”, the Argentinian writer Jorge Luis Borges tells a story about trying to understand the unexpected. While tracking a fabled tiger in a remote region of the Ganges delta, a professor finds some glittering blue stones hidden in a crevice. Though they were not the aim of his expedition, he takes a handful with him, finding later that they seem to multiply every time he tries to count them. The professor’s rational mind is challenged by this seemingly impossible occurrence, and despite all his efforts, he can find no scientific explanation for how the stones grow in number. In a passionate monologue he shows his despair, and declares that there is nothing more disturbing than discovering “that the universe can tolerate disorder”.

Researchers in schizophrenia genetics are probably familiar with some of those feelings. In 2014, an international alliance of researchers, the Psychiatric Genomics Consortium (PGC), led by Cardiff University, announced that more than a hundred locations in the human genome had been associated with this disorder, a substantial increase from only five genes previously identified. These genetic risk factors were inside large stretches of DNA called “loci” (plural of the Latin word “locus”, meaning “place”) which could span several or no genes, with few obvious or discernible patterns. From a first glance, some of the loci were exciting findings proving that susceptibility to schizophrenia could be modulated, for example, by genetic variants in neurotransmitter systems and targets of antipsychotic medication. Delving deeper, and as it often happens in research, most results were more questions than answers, and the data suggested that that there were many more loci to discover. Perhaps even thousands of them! From my perspective, as an early career researcher arriving to the discipline of psychiatric genetics at the time, there was undoubtedly much to learn from these results, but the numbers seemed overwhelming: Could we make sense of such a number of loci? Or would they be like Borges’ blue stones, the echo of a disorderly universe?

Several years later, our research group at Cardiff University has produced a study in which we reveal some of the patterns behind the PGC results. First, as predicted, their number has expanded: By adding 11,000 patient samples recruited from around the UK, we have detected an additional fifty loci implicated in schizophrenia. These loci account for some interesting findings on their own, like a gene, RBFOX1, known for its associations with Autism Spectrum Disorder or Alzheimer’s disease. In addition to identifying new potential genes, we have also been able to quantify that 64% of the genetic variants that are associated with schizophrenia are within gene boundaries, while the vast majority of our genome lies outside genes. This result is relevant as variants inside genes are more likely to affect the function of proteins in the body, and identifying changes in proteins has been a primary way of leading to the development of new medical therapies. This is not to say that variants outside genes should be disregarded: Finding out the way they affect our health might be trickier, but with time and effort we’re becoming increasingly better and better at it.

Focusing on the genes in which risk variants reside, our group found they shared another interesting characteristic: They were more likely to be genes that contain few mutations leading to a malfunctioning protein (these are so-called loss-of-function intolerant genes). If a gene is said to be “intolerant” to mutations that prevent its protein from functioning, it means that the vast majority of people don’t have such mutations in these genes, likely because their presence causes serious health problems. As these genes are implicated in important biological processes, the fact that they harbour schizophrenia risk factors is noteworthy, and narrows down our “priority list” of risk factors even further (only ~3,200 out of ~19,000 genes in the human genome are intolerant to loss-of-function mutations). Furthermore, it seems that many of these genes are particularly important for the function of the central nervous system, confirming the previous result of our group which related schizophrenia to the development of neurons and the action of neurotransmitters such as GABA and glutamate.

Finally, we also addressed a question which has puzzled psychiatrists and evolutionary geneticists alike: If people with schizophrenia have, on average, fewer children than people without this disorder, why does schizophrenia still affects so many people? Conditions which impact fecundity are, by definition, under natural selection, which should gradually eliminate related mutations from the genome. However, our genetic analyses revealed hundreds of risk loci, had they somehow escaped selection? To answer this, we directly searched in our data for signatures of natural selection, finding that regions of the genome associated with schizophrenia seemed to have been affected by an evolutionary process called “background selection”. This is a hallmark of regions of the genome with biological importance, and ensures that harmful mutations are recurrently removed. Mutations with mild effects are known to be unaffected by background selection, and our results suggest that these remain in our genome, being many of the genetic risk factors for schizophrenia that we’ve detected in our analyses so far. This reminds us that, contrary to what many of us were often taught, natural selection (or anything in evolution, really) is not a flawless process, and we cannot ignore how much its results are influenced by chance.

In summary, the results of our study help us interpret the biological significance of the large numbers of loci that, so far, we have found associated to schizophrenia. As such it owes much to those that have come before us: It would not have been possible without the herculean effort of the PGC nor without all the research that has been recently performed on statistical genetics, brain biology, protein function or molecular evolution. It is likely because of all of these precedents that we have been more fortunate than Borges’ protagonist in “Blue Tigers”, who by working in isolation was doomed to never reveal the nature of the mysterious multiplying stones. The genes and loci we find in our research might be also growing in number every time we analyse them, but there is nothing “disorderly” in them that should dishearten us.