Is epigenetics totally gay?

In the light of much of what we know about evolution, human homosexuality doesn’t make a lot of sense. Available data suggests that sexual orientation has some inborn, probably genetic, basis. But it’s hard to reconcile that with the fact that gay men and lesbians aren’t, by definition, particularly interested in doing what it takes to pass on any genes that might have contributed to creating their orientation. Natural selection is, all things being equal, pretty good at eliminating genes that make people less likely to make babies.

I’m gay. I’m also an evolutionary biologist. You could say this particular puzzle is tailor-made to attract my interest.

It turns out that there are a number of ways that human populations might accommodate gene variants for same-sex attraction without suspending the rules of natural selection. But it’s also possible that human sexual orientation has a biological basis without being genetic. Natural selection can’t do anything about a trait if variation in that trait isn’t linked to variation at the genetic level. So I was immediately interested by the recent announcement that a team of biologists at NIMBioS, the National Institute for Mathematical and Biological Synthesis, had found that human homosexuality is due not to genetics, but to epigenetics.

However, as soon as I secured a copy of the study itself (available in PDF format here), I was disappointed to find out that the reports of a solution to this particular evolutionary enigma are somewhat exaggerated. The paper doesn’t present any new data that directly links a specific developmental process to human sexual orientation — it’s a review article, gathering existing results in support of a hypothesis that isn’t, at its most basic level, entirely new. But it’s not the job of a review article to present new data; reviews are supposed to gather up what is already known on a topic and identify what new research could do to better answer the questions that remain. And that’s exactly what the new study does.

The paper’s authors are William R. Rice, Urban Friberg, and Sergey Gavrilets — all are evolutionary geneticists who don’t particularly specialize on humans, and Gavrilets especially is best known (to me, anyway) as a theoretician, testing hypotheses using mathematical models and computer simulations. Gavrilets and Rice have previously published one of the most thorough theoretical analyses [PDF] of how gene variants that promote homosexuality might remain in human populations in spite of their selective downside, so they have some established expertise on this topic. In the new paper with Friberg, they introduce a new factor, as mentioned above: epigenetics.

A role for epigenetics

Epigenetics refers to a class of chemical changes to the packaging of DNA that can alter how the genetic code is translated and expressed in visible traits, but without changing genetic code itself. Epigenetic markers, or epi-marks, can attach to the genetic code to “turn off” genes, or make them more active, or change their responses to the activity of other genes. You could think of epi-marks as annotations to the genetic code, like notes in the margins of a book that help a reader remember what passages to return to, or how different parts of the text connect to each other. The prefix epi means “above” or “upon,” so epigentics is a code “upon” the genetic code. Here, I’ll give you a thematically appropriate illustration:

The epigenetic mark-up of the genome is erased and reset very early in development — when an embryo is still a small cluster of cells, not yet implanted in the wall of the uterus. But — and this is the intriguing part — that early epi-mark erasure isn’t perfect. Epi-marks that escape erasure can be passed from parent to offspring, like marginal notes on a photocopied page.

There isn’t any direct evidence that epigenetic markers play a role in human sexual developent. But Rice, Friberg, and Gavrilets argue that what we know so far about the process by which an embryo develops male or female traits suggests that some sort of epi-marks play a role.

In humans, as in most mammals, hormones called androgens (testosterone and its relatives) play a big role in the development of sexual characteristics — embryos carrying an X chromosome and a Y chromosome develop testes, which produce androgens to promote development of male genitals and reproductive anatomy, and eventually sexual attraction to women; embryos carrying two X chromosomes develop ovaries instead of testes, have consistently lower androgen levels throughout development, and typically develop female genitalia and anatomy, and attraction to men.

XY embryos produce, and are exposed to, higher concentrations of androgens than XX embryos — but that’s on average. Individually, the low end of the range of androgen concentrations recorded for XY embryos overlaps with the upper limit of the range of androgen concentrations recorded for XX embryos. Yet the number of embryos in that “ambiguous” range of androgen concentrations is greater than would account for the frequency of children born with ambiguous sexual anatomy or same-sex orientations. In other words, the minimum concentration of androgens necessary to develop a XY embryo into a male is too low to interfere with the normal development of a XX embryo into a female. That suggests that the key difference in male and female development isn’t just hormone levels, but how sensitive embryos are to those levels. Think of it this way: if androgen concentrations are like the thermostat in a room, XY embryos are wearing a couple extra layers of clothing — turn up the heat, and the XYs will start sweating well before XXs start to feel too warm.

That extra clothing could very well be epi-marks that either act to make XX embryos less sensitive to androgens, or to make XYs more sensitive, or both. Rice et al. argue that such markers would help to “canalize” sexual development, making the process robust to variation in the environment encountered by the embryo, and that should be advantageous in most cases. Indeed, there are a few studies that have found differences in epi-marks between XX and XY embryos — though none directly connected to sexual development. If there are in fact sex-specific epi-marks that determine sexual development, and these epi-marks are sometimes transmitted from a parent of one sex to an embryo of the opposite sex — well, maybe things get complicated.

If the model fits

So far Rice et al. have established that the available data are at least consistent with some sort of role for epigenetics in developing human sexual anatomy and psychology. But how would this work evolutionarily? Epigenetic markers are part of biological responses to the environment, but at some level they’re still created by genes. So the authors build and analyze a simple mathematical model to see what might happen to a gene variant that acts the way they propose that sexual development epi-marks might, promoting typical development of one sex, but sometimes promoting atypical development of the other sex.

In broad strokes, this model is very similar to one in the paper Gavrilets and Rice published earlier, in which a gene variant promotes greater fitness when carried by a female, but makes her sons more likely to develop same-sex orientation. The takeaway from that model is that it’s quite possible for the benefits to a mother’s fitness conferred by a variant to outweigh the increased possibility that one or more of her sons might be gay.

In the new model, Rice et al. fold in epigenetics by simulating a gene variant that creates sex-dependent epi-marks — such as epi-marks that reduce an embryo’s sensitivity to androgens. The simulated epi-marks increase the fitness of one sex, but has some probability of carrying over to the next generation, where it might interefere with the development of the opposite sex. In the simplest case they consider, such a gene variant would be favored by selection as long as the benefits it confers to one sex are at least four times as great as the risks for the opposite sex — and if the epi-marks are transmitted less than one hundred percent of the time (which is usually the case), that ratio can be smaller. In fact, if the carryover probability is sufficiently small, selection can favor the epi-marking variant even if the cost to one sex exceeds the benefits conferred to the other.

Better than pure genetics?

Epigenetics is an appealing explanation for same-sex attraction because we have, at best, a fuzzy picture of the genetic basis of sexual orientation. Homosexuality definitely “runs in families”. That is, people with gay or lesbian parents, siblings, aunts, or uncles are more likely to be gay or lesbian themselves; and pairs of identical twins, who share pretty much all their genetic code, are more likely to have the same sexual orientation than pairs of fraternal twins, who share only half their genes.

Yet more sophisticated methods to identify specific genes associated with sexual orientation have failed to find any consistent candidates. (Though, as a caveat, the only genetic association study [PDF] I’ve seen suffers from small sample size and considers a very small number of markers by modern standards.) Moreover, while identical twins share sexual orientation more than fraternal twins, they don’t share it with complete fidelity — only about 20% of gay men who are identical twins have twin brothers with the same orientation.

Epi-markers, which are not so much heritable as somewhat sticky, might explain that fuzziness. For a final twist, Rice et al. used an estimate of the frequency of gay men in the general population (about 8%, which seems high to me) and that 20% “concordance” rate for identical twins to calculate the transmission rate for a hypothetical “feminizing” epi-mark that makes men more likely to be gay: about 50%. That seems high, but the authors argue that we don’t yet now enough about the range of epigenetic transmissibility in humans to rule it out as unrealistic.


So what do we take from this new paper? It’s certainly not positive proof that accidential transmission of epi-markers for sexual development causes same-sex orientations. As I said above, this is a review article, doing what reviews are supposed to do — gather existing evidence, make the case for a hypothesis, and point the way toward fruitful directions for new empirical research.

I came away from the paper more convinced that (as I suggested back in that earlier article) that it may be better to think about the evolution of same-sex orientations not in terms of the selective fitness of gay men and lesbians, but in terms of our parents’ fitness. The new epi-mark model by Rice et al. frames things in just those terms — the gene variant it considers improves the fitness of offspring of one sex, but poses some risk to the fitness of offpsring of the other sex.

Under the Rice et al. model, there’s still a role for gene variants that might be affected by natural selection — but gene variants carried by parents, not necessarily offspring. Epigenetics allows parents to shape their offsprings’ traits in a more subtle fashion than direct genetic inheritance, and Rice, Friberg, and Gavrilets make a convincing case that we’ll need to take this into account as we search for the evolutionary origins of human sexuality.


Bermejo-Alvarez, P., Rizos, D., Lonergan, P. & Gutierrez-Adan, a. 2011. Transcriptional sexual dimorphism during preimplantation embryo development and its consequences for developmental competence and adult health and disease. Reproduction 141: 563–70. doi: 10.1530/REP-10-0482

Gavrilets, S. & Rice, W.R. 2006. Genetic models of homosexuality: generating testable predictions. Proceedings of the Royal Society B 273: 3031–8. doi: 10.1098/rspb.2006.3684

Morgan, D.K. & Whitelaw, E. 2008. The case for transgenerational epigenetic inheritance in humans. Mammalian Genome 19: 394–7. doi: 10.1007/s00335-008-9124-y

Mustanski, B.S., Dupree, M.G., Nievergelt, C.M., Bocklandt, S., Schork, N.J. & Hamer, D.H. 2005. A genomewide scan of male sexual orientation. Human Genetics 116: 272–8. doi: 10.1007/s00439-004-1241-4

Ngun, T., Ghahramani, N., Sanchez, F.J., Bocklandt, S. & Vilain, E. 2011. The genetics of sex differences in brain and behavior. Frontiers in Neuroendocrinology 32: 227–246. doi: 10.1016/j.yfrne.2010.10.001

Pillard, R.C. & Bailey, J.M. 1998. Human sexual orientation has a heritable component. Human Biology 70: 347. PMID: 9549243

Rice, W.R. & Friberg, U. 2012. Homosexuality as a consequence of epigenetically canalized sexual development. Quarterly Review of Biology 87: 343–368. doi: 10.1086/668167