Wednesday, December 2, 2009

Real science: Copy number variation and autism

I've posted around on different blogs about copy number variation and have edited a few papers recently about the same concept as potentially underlying autism. And now, there's a new report out that copy number variation (CNV for those of us who get tired of typing that) in neural growth genes may have a strong and inverse role in autism and schizophrenia, respectively.

What is CNV? Well, you've got a lot of DNA in each nucleus of your cells. It's packaged pretty tidily, wrapped around proteins like a spool of thread and then bundled up into those little packets we call chromosomes. Some parts of a string of DNA provide a code that the cell uses to build proteins, while other parts of the DNA are what we used to call "junk" because they weren't code for anything, but we don't think of it that way any more.

Nature builds the "string" using only four different kinds of molecules, or building blocks: A, G, C, and T. Through millennia and various accidents, sometimes this series of blocks in a segment of the string can be deleted or duplicated (doubling from, say, AGCTCCGC to AGCTCCGCAGCTCCGC). Sometimes, they can be duplicated many times over. If the duplication is of regions that hold the code for a protein (i.e., a gene), then the number of copies of that gene now varies from the usual number of copies present in the organism, and it is officially a CNV. ETA: The upshot of having more or fewer copies of the gene is that the organism may end up with more or less protein that the gene encodes.

Does it hurt the organism for these duplications to happen? Depends on what they affect. While you could probably glide through life just fine with a duplicated finger, a duplicated head might present a bit of a problem, and DNA has similarly influential and not-so-influential areas. We've already linked a few disorders to sequence duplications. Huntington's traces to having a certain number of triplet repeats of the DNA alphabet in a gene. While these might be totally innocuous somewhere else, if these CAG-CAG-CAG repeats occur in great enough numbers in the sensitive areas, the result can be disease.

Now, we've got CNVs to consider. I love this idea and strongly think it's a research avenue very much worth pursuing. In the study reported today, having a duplication can be associated with autism, while having a deletion can be associated with schizophrenia. Researchers feel that this relationship of over-presence vs. absence reflects specifically the perception of autism as resulting from neuronal overexpression and of schizophrenia arising from an underexpression. They even go so far as to point out that autistic people often have larger heads while people with schizophrenia trend to smaller head sizes, again something that intuition, at least, would say is related to brain growth.

Why do I love this research path so (outside of the fact that the PI on the study, Bernard Crespi, also studies walking sticks)? These researchers have identified four regions of the genome in which CNV was strongly associated with either autism (duplications) or schizophrenia (deletions). Four. In a disorder of triplet repeats like Huntington's, families can carry what are called premutations through generations, not having quite enough of the repeats to manifest disease. But a duplication of that repeat area in a single generation through a copying accident (they happen ALL the time) can push that over the edge and cause disease. And the more repeats, the earlier the onset of disease.

In other words, with a variation in repeats, there is a disease spectrum. If we have four areas of interest with different numbers of duplications, I would hypothesize that we'd have gradations of effects, a spectrum of manifestations related to whether all or some of those regions are affected, to how many duplications of each region exist (e.g., 4^many possible combinations). With deletions, the spectrum would be considerably more narrow, 4^4 combinations only, reflecting the relative narrowness of schizophrenia phenotypes. With this construct of CNVs, we'd have the wide variation of phenotypes we see with autism and the narrower variation with schizophrenia.

It's not The Answer. It's not The Be All and End All. But...it sure is intriguing. Now, if we could just get people to pay attention to real scientific findings like these, supported by considerable research already pointing to CNVs as relevant...we'd really be making some progress.