Wednesday, February 17, 2016

Somatic mutation and neurological traits. Part II. Relevant Somatic Mutation discovered?

The first installment of this brief series discussed some possible manifestations of somatic mutation in helping to account for the biology and epidemiology of epilepsies.  From a cellular point of view, epilepsies seem to be functionally close to gene action, but mostly have eluded gene mapping in families, GWAS, or mouse models.  Yesterday, I referred to a paper that I published in 2005 which mused about the possibility that some epilepsies and other behavioral traits might be due to genetic effects that arose through somatic mutation rather than being inherited in the parents' germline, and I tried to suggest reasons why that idea seems plausible  But finding direct evidence for relevant somatic mutation in relevant neural (brain) tissue, just a subset of cells, would have been technically very challenging.  That was then, however, and technology marches on.

A new paper by Wei et al. in the February 2016 issue of Cell ("Long Neural Genes Harbor Recurrent DNA Break Clusters in Neural Stem/Progenitor Cells") may suggest that we're getting to a point where it might be possible to address my speculation directly--and perhaps go far beyond that to understand epilepsies much better but, even more, to address variation in normal behavior and abnormal psychological function in general.

Wei et  al. report using a clever, sophisticated, if very technical method to make a systematic search for a class of somatic mutations in neural cells.  The experiments are done in mouse cell culture and, basically, Wei et al looked genomewide for chromosomal changes or rearrangements that arise out of double-strand chromosomal breaks.  The detail is far beyond what I could competently describe here, and would be out of place in a blog.  However, the upshot of their work is that the authors expanded on what had been known, that there are very large mutational rearrangements detectable in repeatable locations in the genomes of early neural precursor cells.

As shown in the figure below, the rearranged locations are found genomewide, and on most chromosomes. The changes varied among cells tested.  I am not qualified either to defend or critique the method itself, but have no reason to doubt the findings, and if anything it must be fair to assert that because the method only finds one sort of mutation, the finding implies that other methods would find expanded evidence for somatic mutation of various sorts in these cells, as are clearly known to happen in other cell types. The authors used methods that have passed muster before, they cite ample other literature on somatic rearrangements generally.  Experiments like this are artificial in various ways, and theirs doesn't prove that the mutations they detected do in fact have phenotypic effects in vivo, but the case for that is certainly and strongly plausible.

Neural and other genes undergoing SoMu in (mouse) brain. Circle numbers are chromosomes  From Wei et al., Cell, 2016
The importance of this finding, and what justifies its being published in Cell in my view, is that it shows not just tolerance of somatic mutation, but perhaps even active genomic modification in the early mouse brain. Moreover, the authors find that a major fraction if perhaps even preponderance, of the genes altered in this process, whatever other roles they may serve, are involved in neural cell adhesion and synapse formation and/or function.  One can be properly skeptical about assigning function to genes (gene ontologies are often quite shaky and genes typical serve many functions), but even if there is a sort of after-the-fact bias in interpretation, the involvement in the modified genes in synapse function seems well established.

For a partial summary of the approach, the original authors as well as a short, cogent and somewhat more digestible accompanying commentary by Weissman and Gage ("A Mechanism for Somatic Brain Mosaicism"), explain the potential this may have for the determination of individuality. This is speculation in detail, naturally, but the proto-neurons develop into relevant areas of the adult brain, and the authors note that the affected cells are, or reasonably may be, involved in a variety of higher functions.  They mention learning and mental functions such as autism, bipolar depressive disorders, schizophrenia, and intellectual disability.  They also speculate that the somatic changes could be relevant to various forms of cancers of brain cells.  And, perhaps an artifact of the way science is done these days, I saw no mention of the obvious likely fact, that if these changes can be involved in disease they can also be involved in any other psychological traits.

The authors also did not mention epilepsies, nor whether genes known or suspected of being involved in some epilepsies were detected in any of these rearranged regions. I'm not any sort of expert on that, but on a very cursory check, I did not spot any specific known epilepsy related genes in the above figure, but one can hardly take that as in any way definitive. Since epilepsy is itself relatively rare, only rare somatic mutational changes need be involved, for epilepsy to be an occasional consequence and to help account for the variation in the types and parts of the brain affected, the seriousness, age of onset, and so on.  We'll see.

Phenotype amplification
In the previous post, I referred to to means of phenotypic amplification, a process by which mutations in a single or small subset of cells could reach organismal (whole-person) detectability or effect.  One means would be if a somatic mutation happens early enough in development for a large enough set of descendant cells to be affected, almost as if the mutation were inherited in the germ line.  Given the n numbers of at-risk embryonic cells and the number of people produced each year, and the surely large number of genes that could have an effect on epilepsy (or any other psychological trait), this simply must occur!  Depending on when in development the somatic mutation occurred, its descendant set could be general and bilateral, or could be restricted to a small part of the brain on just one side, depending on the way these tissues differentiate during development.  That is, the types of epilepsy that are focal, affecting only a restricted, identifiable part of the brain, might be susceptible to genomic analysis to identify the mutations (illustrated in the figure in yesterday's post).  That is, the rest of the person's neural cells would not be mutated.

The other method of phenotype amplification I mentioned yesterday could occur when one or just a few mutant, misfiring neurons misfires, and that then induces a cascade of firing in the many other neurons to which it synapses.  This kind of phenotype amplification might not be easily detectable, because the single or few mutant culpable neurons might be invisible in sequences of the entire affected brain region, because most of the misfiring cells would, themselves, be genetically normal. That is, their misbehavior would be induced by the abnormal neuron(s) to which they were synapsed.

The variable and usually highly restricted, focal nature of epilepsies (i.e., suggested in yesterday's figure) cries out for explanation that involves the unique features of these small areas.  The methods in the Cell paper are very complex, technical, and certainly nothing I ever did in my lab, so I must acknowledge that my thoughts could be off the speculative mark.  Still, the paper and others it cites show that the ideas have more than totally circumstantial support.  There are lots of mutations in neural cells, as indeed, there are in any set of somatic cells.  They affect the individual in which they occur but are not transmitted in the germ line and hence not discoverable by GWAS and family studies.

No matter what one may think of the idea, a point worth making is that there is now at least one major study systematically documenting the pervasive frequency of somatic mutation of particular types (rearrangements),  that is active, common, and genomewide in developing neural precursor cells (and the authors cite other relevant results).  How or even whether a role of somatic mutation in epilepsy can be shown in real mice, much less humans, and not just cell culture or sequencing of the brains of deceased patients, is of course an open question.

However, unlike 2005 when I wrote my speculations, high throughput and even single-cell sequencing are now at least beginning to be practicable in mouse models and deceased humans who had epilepsies of known focality and expression.  If the ideas are cogent, investigators may find them compelling subjects to study.

Searching for somatic mutational causes of traits that seem to have specific tissue locations, seems now at least more possible, inroads might be possible and could, in principle at least, lead to substantial advances in understanding the mechanisms involved, as well as the epidemiology and phenotypes of epilepsy.


Michael Finfer, MD said...

I am surprised that this study has not been done. Some patients with a single epileptic focus in a non-critical area will have the focus resected. Someone needs to collect some of those cases, pull the paraffin blocks, and sequence the genomes of the neurons in the epileptic foci. I am sure that the results would be fascinating one way or the other,

Ken Weiss said...

I hadn't thought of that possibility, since I'm not a clinician of any sort. But testing the idea could have even broader implications--even if it showed that the idea was just baloney.

Peter said...

Is this mutation, or is it programmed in an analogous way to antibody VDJ recombination? If there were VDJ-like "shuffling" of modular genes in particular cell lineages, how would we know?

Ken Weiss said...

This is a bit beyond what I ca comment knowledgeably on, but reading the paper might clarify for you. However, as I understand it the author used some recombination-repair-deficient mouse cells, to see the rearrangements that occurred, but did not interpret the results to mean that in 'normal' cells all the double-strand breaks were repaired. You should check the paper or the commentary. If you can't get access I can email pdf's to you.