“Happy families are all alike; every unhappy family is unhappy in its own way”. Thus beginneth Anna Karenina. That wasn’t supposed to happen with hereditary disease. The examples we had before large scale DNA sequencing became cheap were basically one gene causing one disease. Two of the best known cases were sickle cell anemia and cystic fibrosis (CF). Not so for schizophrenia, in which a variety of individually rare genetic defects has been found each confined to 1 or 2 families. For details see my post on this subject a few months ago (https://luysii.wordpress.com/2010/04/25/tolstoy-was-right-about-hereditary-diseases-imagine-that/).
But schizophrenia at least looks pretty much like one disease (just like sickle cell anemia and CF). Autism spectrum disorder is clearly a collection of disorders with a similar constellation of symptoms. Before going much further, it’s time for a few facts about the distribution of IQ in the general population. The standard deviation of IQ is 15 points, and, except at the tails, the distribution of IQ falls on the famous Bell Curve (Gaussian distribution). This means (with an average IQ of 100 — by definition really) that between 2 and 3 % of the population will have IQs higher than the average (e.g. IQ over 130) and a similar amount of the population will have IQs of 70 or below. So there are a lot of kids around who are what used to be called mentally retarded (I’m not up on the latest euphemisms).
The definition of autism and autism spectrum disorder have radically changed over the years, and what was rare, is now common, as diagnostic criteria have become looser (or more inclusive if you wish).
Autism spectrum disorder is really a group of diagnoses. Here is the state of play 6 years ago. [ Cell vol. 119 p. 28 ’04 ]
Some of the CNVs are common, and the authors decided to look for CNVs, affecting less than 1% of the population (e.g. the rare ones), a perfect set up for Tolstoy. In the controls an average of 3.6 genes were intersected by the rare CNVs, while the autism spectrum disorder (ASD) had a higher number (4.3 genes). Not much different, but because, of the large number of people in the study, definitely significant. Because some parents of autism spectrum disorder kids were also studied, they could tell whether the CNV was new to the child (assuming BOTH parents were studied, because infidelity is always possible). Some 5.7% of the ASDs had a de novo event (a CNV which wasn’t present in the parent(s) ).
Lots of people have been looking for genes implicated in ASD, and some have been found in other studies– the names are all acronyms (SHANK2, SynGAP, DGLAP2). Even studying nearly 1000 ASD individuals CNVs in these (already known) genes were found only in one or two cases out of the nearly 1000 ASDs. So Tolstoy is right in spades, unhappy families with an ASD kid really are all different, each in their own way. Of interest to drug design chemists, NONE of the protein coding genes affected by the CNVs involved neurotransmitters or their receptors. (Probably anything messing with them would be lethal, but I’m pretty sure that some neurotransmitter receptor mutants have been found).
Since just about everyone has a rare (appearing in under 1% of the population) large (over 30,000) CNV — recall that the average number in controls was over 3, it’s going to be difficult if not impossible to pick out the culprits.
Now I’ve written a lot about how protein-centric molecular biology was. It still is. What did the authors focus the assay chip’s 1,000,000 interrogating sequences on? The sequences were not randomly distributed throughout the genome, but rather on the 20,000 or so protein coding genes that humans possess. Recall that the actual amount of DNA coding for amino acids is a mere 1.5% of the 3.2 megaBase genome. They literally ignored the dark matter of the DNA (the part not coding for protein). For details see https://luysii.wordpress.com/2010/07/14/junk-dna-that-isnt-and-why-chemistry-isnt-enough/. The editorial conflates the term gene, with protein coding genes. This just won’t do.
I’ve said this before, but it bears repeating. At one point it was stated that humans and chimps should really be considered one species, because our proteins are so similar. This is like saying Monticello and Independence Hall are just the same because they’re both made out of bricks. One could chemically identify Monticello bricks as coming from the Virginia piedmont, and Independence Hall bricks coming from the red clay of New Jersey, but the real difference between the buildings is the plan.
It’s not the proteins, but where and when and how much of them are made. The control for this (plan if you will) lies outside the genes for the proteins themselves, in the rest of the genome. The control elements have as much right to be called genes, as the parts of the genome coding for amino acids. Granted, it’s easier to study genes coding for proteins, because we’ve identified them and know so much about them. It’s like the drunk looking for his keys under the lamppost because that’s where the light is. We’re only beginning to understand the controlling elements. It would be fascinating to see how many indels the ASDs, parents and controls had in the 98.5% of the genome not studied. Since we have the sequence of the human genome in hand, it shouldn’t be too hard to design a 1,000,000 sequence chip, with the interrogating sequences spread 3,200 nucleotides apart. The problem is we’re not exactly sure what we’re looking for. The de novo CNVs may be a new lamppost. Stay tuned.