The notion that nature and nurture interact to produce the phenotype of an individual is a very old one. Modern techniques of molecular biology and the mapping of the human genome have led to multiple studies approaching gene-environment interaction as more than a metaphor. A part of that revolution was led by Caspi and associates in their studies of depression. They examined a gene linked to the serotonin transporter mechanism, which influences how long the neurotransmitter serotonin remains in the synapse between nerve cells. Many anti-depressant medications inhibit the re-uptake of serotonin by the transmitting cell. In some way, this helps to alleviate depressive symptoms. Caspi and associates found that risk of depression was enhanced under the condition of individuals possessing a specific polymorphism or gene variant for the serotonin transporter gene and having experienced recent stressful life events. For individuals with other variants, risk of depression associated with the experience of stressful life events was reduced.
This was indeed revolutionary work, inspiring an explosion of papers examining gene-environment interaction (or “GxE” as it is abbreviated). It also generated its own controversy and lively debates. One of these involves whether or not GxE effects have been detected at all, based on two issues. First, replication of results in GxE research is highly variable. Second, the “candidate gene” approach, where one gene (or sometimes a multiple polymorphism index) is examined, has been severely criticized by researchers in favor of an approach involving genome-wide scans, a method in which, for example, respondents are divided into groups who are depressed and not depressed, and then the entire genome for each individual in each group is scanned to see if there is some collection of genes that differ between them. The argument for this approach is that any complex phenotype, like depression, is likely to be the result of multiple—literally dozens—of interacting genes. It is then argued that candidate gene researchers are finding spurious results because they are capitalizing on a chance outcome, given the hundreds of possible combinations of genes that could be detected.
While the candidate gene versus genome-wide scan arguments are interesting, what seems more relevant to me is the question forming the title of this essay: what is the “environment” in gene-environment interaction research? As I noted, there has been an explosion of papers in GxE research, and with only a few exceptions, the ”environment” examined has consisted solely of the occurrence of stressful life events. These have been assessed both during childhood (referred to as “childhood adversity”) and as standard concurrent stressful events (e.g., unemployment, death of a close family member); but typically, that’s the “environment.”
There are very few anthropologists involved in GxE research, at least in the study of mental health outcomes. It would, nevertheless, seem an area of research ripe for our attention, because if anthropologists know about anything, it is context. And if the environment in GxE research refers to anything, it must refer to context. We know that the cultural context within which individuals function can be configured in a myriad of ways. Therefore, it is entirely plausible that the variable replication of results in GxE research is a result of the ways in which the environment changes from one study to the next, in subtle and nuanced ways.
Why have GxE researchers stuck to an impoverished view of the human environment? I think because most of them have been more interested in the genes than in the environment (and that is, of course, an important way to proceed). Since a number of researchers have found interesting results employing a measure of experience in the social environment as easy and straightforward as stressful life events, it simplified research design to stick to that. It does, however, leave the environment in GxE work under-specified.
My colleagues and I ventured into this area some years ago when we found, in a small subsample from a larger study in Brazil, an interaction between cultural consonance in family life and a polymorphism in the 2A receptor for serotonin. Those individuals with a single variant of the gene, if they had low cultural consonance, reported exceptionally high depressive symptoms; those individuals with that same variant, if they had high cultural consonance, reported exceptionally low depressive symptoms. Persons with other gene variants exhibited the same inverse association of cultural consonance and depressive symptoms, but not as strong.
When later we tried to replicate the results on a larger and more representative sample, as is some common in this literature, we could not. We did, however, find an interaction between childhood adversity and the 2A receptor polymorphism. This in turn was mediated by cultural consonance in family life, and the interaction and mediating effects were most pronounced among lower social class respondents.
This results in a model that looks like this:
This is a far cry from a simple interaction between life events and a gene. And, given that there is a cultural mediation of a gene-environment interaction, which in turn is moderated by social class, the sociocultural context looms large in the overall biocultural process. This approach does not lend itself well to the genome-wide scan, either, since instead of comparing two groups, you would end up having a complex, multiple group comparison.
I won’t argue that more complex models of GxE, specified within a sociocultural context, is the only way to proceed. I only want to emphasize is that this is an important way to proceed, and that biocultural anthropology is uniquely situated to contribute to it.
Biocultural Systematics is written by members of the University of Alabama Biocultural Medical Anthropology program.
Bill Dressler, a professor in the department, has conducted research in social epidemiology in Brazil and the U.S.