October 31, 2013. Several talks at the World Congress of Psychiatric Genetics, held October 17-21 in Boston, Massachusetts, focused on new mutations and their possible role in schizophrenia. “De novo” mutations arise spontaneously in the genome rather than being inherited from the parents. Though rare, these events occur more often than previously appreciated, and recent studies argue for their role in disorders such as schizophrenia (see SRF related news story and SRF news story) and autism (see SRF related news story). According to some researchers, these new mutations are leads worth following for several reasons: They offer an explanation for why schizophrenia endures even though many people with the disorder do not go on to have children; they may well be fairly potent (and thus disease related) because natural selection has not yet had a chance to temper their effects; and several speakers suggested that, though they are not inherited, they may covertly contribute to measures of heritability based on twin studies (in the case that a de novo mutation-harboring sperm or egg cell gives rise to monozygotic twins).
Sequencing as a family affair
As sequencing prices fall, this technique offers a fine-toothed way to sift the genome for mutations, including those point mutations that change a single letter in the DNA code. But it turns out that our genomes are rife with mutations, making it hard to identify which are related to a disorder. To get around this, researchers are sequencing families, whose shared genetic background should help signals pop out of the genetic noise. De novo mutations are an extreme example of this approach, since by definition they are those mutations found in a person affected by a disorder but not in either parent.
Combing through the protein-coding portion of the genome—the exome—turns up de novo mutations in schizophrenia, according to two talks given on Saturday, October 19. Dick McCombie of Cold Spring Harbor Laboratory and colleagues searched for de novo mutations among exome sequences of 42 trios consisting of a person with schizophrenia and both parents. This study revealed 47 de novo mutations, some of which constituted protein-truncating “nonsense” mutations. These nonsense mutations were enriched in schizophrenia and hit genes already implicated in autism, including a chromatin remodeling gene called CHD8 and a transcription factor gene called AUTS2. McCombie also reported a protein-altering “missense” mutation found in MECP2, the causal gene for Rett syndrome, which shares features with autism. MECP2, CHD8, and other genes fingered in this study have roles in modifying chromatin, which controls gene expression. These may flag an important contribution by chromatin remodeling to psychiatric illness, including schizophrenia.
Daniel Howrigan, a postdoc in Ben Neale’s group at the Broad Institute, Cambridge, Massachusetts, presented results from a similar de novo study of schizophrenia. Exome sequencing of a sizeable cohort of 1,135 trios from Taiwan gave rise to 1,157 de novo loss-of-function events in people with schizophrenia. This frequency was higher than expected, and the genes hit by these events overlap with those in a molecular network centered around FMRP, the gene disrupted in fragile X syndrome (Darnell et al., 2011). This again suggests a relationship between autism and schizophrenia among rare variants, which may reflect shared origins in early brain development.
On Monday, October 21, Shaun Purcell of Mount Sinai School of Medicine in New York City summarized the latest results from exome sequencing of over 600 trios from a Bulgarian cohort. Though he reported no significant increase in overall number of de novo mutations found in schizophrenia compared to controls, he did find these mutations hit disease-related gene sets more than expected by chance, including schizophrenia candidate genes, genes linked to autism or intellectual disability, and genes encoding members of the postsynaptic density (particularly those in the activity-regulated cytoskeleton-associated protein complex, or ARC). While this constitutes statistical evidence, it reinforces the notion that a diverse lot of genes scattered across the genome contribute to schizophrenia risk.
Our mutable genomes
A symposium on Sunday shifted focus from which genes are hit by de novo mutations to why new mutations happen in the first place. As our genomes are replicated and divided in meiosis and mitosis, there are many opportunities for mistakes to occur that remain uncorrected. These mistakes can contribute substantially to nervous system disorders, according to a new study of whole-exome sequencing in a clinical setting by James Lupski of Baylor University and colleagues (Yang et al., 2013). Exome sequencing in 250 people with suspected genetic syndromes but without a clear diagnosis (a majority of whom had neurological problems) returned a “molecular diagnosis” in 25 percent—meaning a mutation deemed to be highly deleterious was found. The majority of these constituted de novo mutations, which suggests they contribute to a wide swath of brain disorders.
As for the origins of these de novo mutations, some locations in the genome are repeat offenders. For example, some CNVs result during meiosis when strings of highly similar DNA sequences throw off the alignment of chromosomes before they swap DNA segments during recombination. Thus, these highly similar sequences, known as segmental duplications, constitute hotbeds for CNV mutations. Similarly, it looks as though certain single nucleotide positions within the genome are predisposed to point mutations, according to Jonathan Sebat of the University of California, San Diego. Whole-genome sequencing of 10 families consisting of two monozygotic twins both diagnosed with autism and their unaffected parents found 581 de novo point mutations, which clustered in certain locales of the genome, including those containing genes already implicated in autism. Modeling results suggested that how susceptible a single position of the genome is to mutation depends on its conservation across species, its accessibility for transcription (e.g., DNAse hypersensitivity and chromatin state), its surrounding nucleotide sequence, and recombination rates. Sebat cast these point mutation hotspots as features of the genome in general rather than being specific to autism and noted that they occurred both in low and highly conserved regions.
The next two talks focused on the discomfiting possibility of de novo somatic mutation. This type of mutation arises during the many rounds of mitosis that ensue during embryonic development. A mistake in DNA replication in one cell will then be passed on to subsequent daughter cells. This would make the embryo, and possibly the brain as well, a complicated mixture of cells, with some carrying the mutation and others not. This means that a given blood cell (the usual source for genetic studies of disease) may or may not carry a somatic mutation—a severe handicap to research if that mutation happens to be the causal variant for a disease.
To begin to define features of somatic mutations, Steven McCarroll of the Broad Institute described the vulnerable moments during cell division that are most prone to mutation. Work done with his postdoc, Amnon Koren, has identified a consistent time sequence during which different parts of the genome replicate. Those that replicate late are more prone to point mutations and CNVs arising from faulty double-strand break repairs, whereas those replicating early were predisposed to CNVs that arise from chromosome misalignment prior to recombination (Koren et al., 2012). New results suggest that replication timing varies among people at hundreds of loci and that replication timing is a genetic trait influenced by common single nucleotide polymorphisms (SNPs). This suggests the somewhat eerie idea that the genome regulates its own mutability.
Despite these measures of somatic mutation occurrence, it’s hard to know how many neurons in the brain actually harbor one. To get at this question, Christopher Walsh of Boston Children’s Hospital has been trolling single neurons isolated from postmortem brains for somatic mutations. One focus has been on L1 retrotransposons, which can insert themselves into a gene and disrupt its function, thus resulting in a somatic mutation. These “jumping genes” have been hypothesized to spur neuronal diversity in the brain (Singer et al., 2010). Walsh covered results published last year that surveyed 300 neurons isolated from postmortem tissue from three healthy people and found that less than 0.4 percent of neurons in the cortex carried an L1 insertion (Evrony et al., 2012). Those with an L1 insertion tended to cluster within a gyrus, suggesting they came from the same L1 insertion-carrying progenitor. Whether this kind of mutation contributes to psychiatric disease, though, is still an open question.—Michele Solis.