Caveat: All findings discussed here were presented at a conference and have not undergone peer review.
What is the use of animal models?
I understand the use of animal models like mice to figure out how gene changes affect outcomes in a whole animal, rather than, say, in cells in a dish. I’ve used them myself. Knocking out a gene of interest in a mouse strain, applying an environmental exposure of interest, observing the behavior of the mouse involved — these tactics can be revealing, sometimes. Say a mouse with a nonfunctioning partner in a gene pair shows a specific behavior — like vocalizing less to its mother — and maybe we can interpret that in human terms as being inhibited social communication and assume the gene in question is involved. The idea is that observing changes linked to the absence or overabundance of a particular gene product can help us home in more on that gene, see what its usual job is and in what pathways, and determine if it could rationally be linked to the human condition in question–in this case, autism.
Studies like this are called “mechanistic” studies because they are the scientific version of opening up the hood and checking the different working parts to see what their function is. Adding a part or taking away a part can often tell us quite a bit about what that part does, what its mechanistic role is in the whole. But I’m feeling a little jaded about animal models in autism because of the genetics and genomics data I saw presented at the conference. With a few exceptions, nothing seems to have emerged as a clear new contender for knocking out or otherwise manipulating in mice. Some of the usual suspects, like SHANK, were there. But the genome-wide association studies, intended to examine a genome for changes associated with a disorder or other condition, are not kicking out a lot of obvious single candidates for genes associated with autism. It’s almost looking like we’d have to make about a thousand animal models of autism to tease out various associations between a gene change and a specific autism-related endpoint.
Genetics and genomics
What studies are kicking out is a mix of information that’s conflicting, suggestive, intriguing, and requiring a lot more refinement. I saw data implicating genes specific to neuronal development or connectivity and data suggesting genetic and biological marker overlap with autism and schizophrenia (and then one suggesting no common, shared gene variants between the two), but possibly tracing to different changes in the genes relative to each condition. The most interesting genetics-related talk I saw referenced microRNA, the tiny molecules in the cell that get to a gene product before the cell can actually use it, and silence it. They’re like the molecular hitmen of the cell. More microRNA generally means less functional product of a given gene. The implication? A gene sequence may be intact and unchanged, but if the microRNA is more abundant, it could be interfering with the pathway well after the gene in question has already completed its work. It is regulation beyond the gene.
Another finding of interest related to maternal genes and their influence in the development of autism. These “maternally acting gene alleles” or MAGAs, have been linked to a number of disorders, including autism, and one conference presentation described using genome-wide association to identify MAGAs potentially involved in autism. That’s an intriguing path to follow.
Awhile back, everyone got pretty excited about genome-wide association studies. These broad-scale but fine-grained analyses were supposed to compare genomic changes between populations and pull out differences that were population specific. The great hope was that these studies would find the needles in the haystack of human gene sequences. Based on what I saw at the conference, in many cases, they’ve just added to the haystack of genes and gene variants implicated autism. On the one hand, given how clearly multifactorial autism is and how little evidence there is for a single gene or even a few genes to be associated with all instances of it, it makes sense that these studies would have confusing results. On the other hand, genome-wide association studies haven’t yielded the clearly writ signposts pointing out research directions that everyone had hoped they would, with some exceptions that apply to a limited segment of the autistic population. That’s not to say we should abandon them. It may be that to really pull the signal from the abundant noise, the number of replications and studies will need to approach biblical proportions before we can see the same candidate genes emerging repeatedly.
Another great hope for tracing the genetic factors of autism was copy number variation (CNV), and that remains in the running. As the name implies, the number of copies of a given DNA sequence can vary from person to person, and sometimes, a difference in copy number can be linked to an outcome, like autism — and sometimes not. Some of the presentations at the conference offered hints that an accumulation of differences in the genome might be more relevant in autism than one specific gene variant or CNV.
Epigenetics is hot in autism research, as it is in seemingly every other relevant research field. The term refers to the chemical tagging of DNA that either silences the sequence that is tagged or causes it to be used more often. Environment can affect this tagging, so much so that identical twins essentially become quite different in their genetic expression through life, not because their actual gene sequences change, but because lifestyle differences result in different tagging patterns on the DNA. Their sequences are the same, but what they use of those sequences can become quite different. Results in epigenetics studies were mixed and often just the beginning of the adventure. This field looks wide open and poised to complicate autism research even more.
The genetics of autism are of interest to me primarily because pinpointing gene variants, gene silencing, and copy number variation as being associated with autism largely extends the host of studies establishing the primarily genetic nature of autism. That’s a purely scientific interest. But with the big analyses, there seems to be a whole lot of noise in the results without much to latch onto. What we need is for genetics studies to focus on the behavioral endpoints of autism that most negatively affect an autistic person’s quality of life. What are the genetic links, for example, to autistic people who engage in self-injurious behaviors? What are the genetic links among autistic people who are nonverbal (some hints here, perhaps)? Most of the populations in these studies are not stratified to that level–they simply include autistic people generally or are divided — still — into “low-functioning” and “high-functioning” individuals, which precludes homing in a specific gaps that both groups may share, another draw back of relying on these “functional” divisions.
A few studies at the conference involved efforts to link genes and outcomes, or phenotypes, and even to use this knowledge therapeutically. I’d like to see a lot more of this greater specificity in these analyses because overall, I don’t see “curing” autism as a realistic or even desirable goal, but addressing gaps is both. In addition, the clear heterogeneity of the condition along with still significant overlaps among populations as researchers currently divide them does nothing to clarify all the noise in the results. Rather than taking autism as a complete entity, I’d like to call for research to home in on specific features that directly and negatively influence quality of life of autistic people or even other populations that share these gaps and look for the mechanisms and pathways associated with them. I am not alone in this idea. At the most, a more focused approach could lead to real interventions for the actual negatives associated with autism without necessarily effacing the positives. At the very least, it could save the lives of a whole lot of mice.
Next from Emily: Research with “human subjects,” the research that deserves a “Most Pointless Award,” and why Simon Baron-Cohen (yes, it’s true) may have presented the most interesting results at this conference