The magical world of epigenetics

Part I: The didactic bit for those who don't know what epigenetics is

Epigenetic inheritance is a simple enough concept, but man oh man the baggage that gets loaded onto it…in many ways, the assumptions people make about it are more interesting than the phenomenon itself. Epigenetics is always in the news lately, but rather than just digging into the details of the paper that prompted all the media coverage, I want to step way back to the beginning and try to defuse a few misconceptions, and also say a few things about popular wrong ideas, and why they might be popular.

Epigenetics refers to mechanisms that modify the expression of a genetic signal. It's a separate set of signals from the sequence of nucleotides in the genome; for instance, you might have a gene for an enzyme, in which the sequence of amino acids in the protein is specified by a chain nucleotides, but the cell also has ways to modify that chain that don't change the sequence, but do affect whether the enzyme gets made or not. One method is to add a methyl group -- a single carbon and 3 hydrogens -- to the sugar backbone of the DNA for the gene. That leaves the sequence intact, but acts as a signal to the RNA synthetic machinery to ignore that piece of DNA. Methylation is one mechanism of transcriptional silencing.

There are many other mechanisms of epigenetic modification than methylation: there are DNA associated proteins, like the histones, that can be chemically modified to affect gene activity, and some of these can activate instead of silencing genes. There are transcription factors that can control gene activity by their presence or absence. It gets extremely complicated when trying to track all the components that modulate a gene's pattern of expression.

Furthermore, epigenetic modifications are passed on from generation to generation. When a cell divides and replicates its DNA, it also dutifully copies the same pattern of methylation from the old DNA strand to the new strand. This is important in maintaining a hierarchical pattern of cell types: when one of your liver cells divides, the new daughter cell does not have to go through a complicated series of differentiation steps or in any way struggle to set what kind of cell type it is. It inherits all the epigenetic marks of its parent cell, all the methylation, all the histone modifications, all the transcription factors, that specify that it is also a liver cell.

For all you coders out there, it's like object-oriented programming. When you create a new instance, all the code (the DNA) is the same, but we also have an initiation routine that duplicates all the local variables. (That probably doesn't help at all if you don't program, but I hope it makes sense to all the computer nerds reading about biology.)

Does epigenetic inheritance have evolutionary implications? Yes, it does. It represents a kind of memory that is passed down for one or a few generations. Imagine a protist that has switched on a set of genes that help it adapt to a particular environment. It would be advantageous for all of its daughter cells, which are existing in that same environment, to carry that appropriate pattern of gene expression right from the beginning -- to be preloaded with properties that immediately adapt them.

Epigenetics also confers greater plasticity on individuals. An organism could, for instance, be able to thrive in both cool and warm temperatures by switching appropriate genes on and off. In this case, though, the specific state of those genes doesn't have to be transmitted from parent to offspring -- it's a kind of inherited flexibility.

But otherwise, it's hard to see a specific evolutionary effect. The whole point of epigenetic modification of gene expression is that it is responsive to the environment! The state of gene expression can be rewritten during any generation by a new set of conditions.

Another problem is that most of the popular interest in epigenetic inheritance focuses on multicellular organisms. The only cells that matter in this case are germ line cells -- that is, guys, only physiological effects that modify cells in your testes can be passed on, and ladies, it's only your ovaries that count. These are highly specialized cells, with a great many epigenetic modifications specific to their functions as gametes. Their most important role is to be prepped to be totipotent, to be primed to follow a broad developmental pathway, and the epigenetic state of many genes will be overwritten to condition them to do that job. If a person has gene X methylated in their tissues, and their child is found to have gene X methylated as well, it's not necessarily a case of epigenetic inheritance -- that gene might have been demethylated in the gamete, and the methylated state in the adult is a consequence of modification during their life history.

So it's kind of weird to see studies of epigenetic states in one generation of adults, comparing them to the epigenetic states in a different generation of adults. Those individuals went through a single-celled bottleneck, where the epigenetic marks on many genes were explicitly reset to a specific condition appropriate for gametogenesis and embryogenesis, and not to, for instance, a state appropriate to the neurons relevant to a behavior. The genes that have a variable pattern of epigenetic marks at this point may be irrelevant to the essential activities that are going to occur during development, and what these researchers studying epigenetic inheritance are actually looking at may, in many cases, be pure noise.

Part II: the bit about silly misconceptions

I've been gathering stories about epigenetics from science and pop culture for a while now. It's a painful exercise in witnessing science harnessed to support magical thinking -- it seems a lot of people want a natural, material explanation to justify their belief that behavior now will modify their biological legacy in a significant way, so they are happy to embrace nonsensical interpretations.

So, for instance, epigenetics is apparently an explanation for racial memory.

May our DNA Carrying also spiritual and cosmic memories passed down in genes from our ancestors?

Sounds ridiculous, I know, but the commenters at that link are lapping it up.

I have written about this for a long time. I did believe in reincarnation, but the more I looked into it the more it was obvious that it was genetic memory. Good to see that I was right.

Great! We're going to replace reincarnation pseudoscience with genetic memory pseudoscience.

Even once you dismiss the bogus kookery, though, there are still real concerns. One is that there is a bizarre trend towards blaming cultural, sociological, and psychological problems on biology, like this claim that trauma may be woven into DNA of native Americans.

Folks in Indian country wonder what took science so long to catch up with traditional Native knowledge. “Native healers, medicine people and elders have always known this and it is common knowledge in Native oral traditions,” according to LeManuel “Lee” Bitsoi, Navajo, PhD Research Associate in Genetics at Harvard University during his presentation at the Gateway to Discovery conference in 2013.

According to Bitsoi, epigenetics is beginning to uncover scientific proof that intergenerational trauma is real. Historical trauma, therefore, can be seen as a contributing cause in the development of illnesses such as PTSD, depression and type 2 diabetes.

No. Just no. The oppression of American Indian populations is terrible and real, but some kind of poorly understood claim of warping of their DNA is a damaging distraction. The problem is poverty, exploitation, and neglect, which has more direct and immediate effects on the well-being of people. Indians are not suffering because of genetic damage, but because our society values their lives less and traps them in difficult and unsupportive environments.

I don't think that article intends it that way, but this is a variant of the old canard that alcoholism in Indian populations can be blamed on their genetic inferiority at metabolizing alcohol, rather than on despair and deprivation. They do have differences in the frequency of many alleles from European populations, but the disreputable interpretation that those variations represent differences in quality is simply not true. Alcoholism is a complex problem that cannot be reduced to a single enzyme.

In fact, there’s no evidence that Native Americans are more biologically susceptible to substance use disorders than any other group, says Joseph Gone, associate professor of psychology at the University of Michigan. American Indians don’t metabolize or react to alcohol differently than whites do, and they don’t have higher prevalence of any known risk genes.

It's all just naive reductionism, and it's really annoying. It's also being applied trivially to all kinds of behaviors.

Why can’t your friend “just get over” her upbringing by an angry, distant mother? Why can’t she “just snap out of it”? The reason may well be due to methyl groups that were added in childhood to genes in her brain, thereby handcuffing her mood to feelings of fear and despair.

Aaaargh. Really? We're going to go there? One of my earliest memories is of my mother showing me how to use a toy microscope. Obviously, this removed methyl groups from the genes responsible for enjoying lenses and light, while selectively adding a -CH₃ to those genes that might have misdirected me towards astronomy. I don't actually have any evidence for that, but hey, it makes sense, right, and you can't argue with chemistry!

Yes, gene expression is constantly being tweaked and modified throughout your life, different cell types have different epigenetic marks, and experiences do change the pattern of activity and molecular behavior of your neurons. Changes in the way you think don't occur without changes in the physical properties of your brain. But it is comically absurd to think you can explain a complex, higher-level psychological phenomenon like resentment of your mother to a facile comment about one kind of chemical reaction.

Epigenetics has become the new "gene for X" of the 21st century. I recall (usually with a barely suppressed glower and curl of the lip) all the sensationalism of researchers looking for the "gay gene" way back when I was in grad school. Even then, it made no sense: I'd been working on flies and fish, and the lessons we were getting over and over is that behavior is an extremely complex emergent property of interactions between multiple genes and their environment, and yet we've got people thinking they can find single genes that are responsible for phenomena as complex and diverse as human sexual attraction? I cringed when I saw the t-shirts thanking Xq28 for their gayness. That one's sexual preferences were not consciously chosen is not synonymous with simple allelic inheritance directly generating them -- your brain is shaped by genes, development, hormones, and experience.

So no "gay gene" has been found. So instead, I think in frustration that their jejune genetic determinism is unsatisfied by correlated patterns of nucleotides, some researchers have turned to searching for epigenetic causation. There are no gay genes. There are also no gay epigenetic marks. That is not to say that there are no differences in how gays vs. straights view sexual behavior, but that you can't simply say that it's due to histone differences or carbons tagged onto a sugar backbone. It's like trying to explain a car accident with "It's like F=ma, dude. I didn't hit your car, blame that douche, Newton."

Also, every time we look at the studies behind these claims, they always seem to have tiny sample sizes and statistically minuscule effects.