Epigenetics: the light and the way?

October 24, 2010 • 6:44 am

How often do you see an editor of a scientific journal complain that a field is overhyped?   Well, you can see it this week in Current Biology, where Florian Maderspacher, the senior reviews editor, takes out after the current penchant of  journalists to see epigenetics as the Great Missing Piece of Biology—a field that will completely revolutionize Darwinism and our view of inheritance. (I take epigenetics to mean “inheritance not based on coding changes in the DNA”.)  The title of Maderspacher’s piece pretty much says it all: “Lysenko rising.” (Sadly, it seems to be behind a paywall.)

This isn’t the first time we’ve encountered journalistic hype about epigenetics: last March there was a dire puff-piece in the Guardian asserting that epigenetics was the death knell of Darwinism.  I went after it, arguing that while epigenetics was a novel and important new phenomenon in genetics and development, it wasn’t poised to completely revise our view of evolution for three reasons.

First, epigenetically inherited changes in DNA and protein, like methylated bits of DNA, ultimately rest on “normal” mutations in DNA that affect those changes. Things get methylated because the nucleotide bases in DNA code for that methylation.  How can “nongenetic” changes in DNA reside in the DNA? Here’s one way.  There are genes whose DNA sequence tells them to do this: “put methyl groups on another bit of DNA if you detect that you’re in the body of a male.  Don’t do that if you’re in the body of a female.”  Males and females would thus have the same DNA code, but it would be used differently depending on the DNA’s “environment”—for example, different hormone titers in males vs. females.  The two sexes would then have produce different types of modified DNA even though their primary DNA sequences were identical.  These modifications usually last only one generation, and then are reset when the DNA finds itself in a new body that could be of a different sex.

Second, as I just noted, in nearly all cases the epigenetic modifications are not inherited past one or two generations, so they can’t serve as lasting templates for evolutionary change.  Insofar as those changes are important in evolution, they must ultimately reside in the primary nucleotide sequence of DNA, the genetic material.

Finally, those who tout the importance of epigenetics in evolution, most notably Eva Jablonka and Marion Lamb, keep trotting out the same handful of tired examples, like changes in toadflax and mouse coat color, that are inherited only temporarily and have nothing to do with evolution.

Maderspacher was cheesed off because the latest issue of his German magazine Der Spiegel devoted ten pages to epigenetics, including a racy cover of a nude nymph whose naughty bits were conveniently occluded by DNA-shaped splashes:

Florian translates this as “THE VICTORY OVER THE GENES. Smarter, healthier, happier: how we can outwit our genome.”  And he explains that this is not a one-off bit of hype:

Epigenetics is of course being considered ‘sexy’ in vast circles of the scientific world (and has attracted the funding to go with it), but that Spiegel cover was a different type of ‘sexy’. This kind of public attention seemed unusual: molecular biology rarely makes it to the front page. And what’s more, this wasn’t just some German oddity: Newsweek had last year a similar cover story, touting a revolution in biology in gonzo-journalism style: “Roll over, Mendel. Watson and Crick? They are so your old man’s version of DNA”. Likewise, the New York Times is in tune, as a news piece last year celebrated the role of the ‘epigenome’ in controlling “which genes are on or off”; nor is the hype confined to the popular press, as a recent editorial in Nature also noted that: “genome sequences, within and across species, were too similar to be able to explain the diversity of life. It was instead clear that epigenetics — those changes to gene expression caused by chemical modification of DNA and its associated proteins — could explain much about how these similar genetic codes are expressed uniquely in different cells, in different environmental conditions and at different times”.

And the wonders of epigenetics, at least in this piece, came down to the same tired old data:

The article itself was mainly concerned with listing examples supporting the notion that ‘genes aren’t everything’: on the one hand, cases where genetic predisposition, e.g. for adiposity, does not lead to the development of that phenotype, as well as the much-discussed weaknesses in genome-wide association studies to pick up causative genetic agents for common diseases; on the other hand, examples of how the environment can influence the genome, evident for instance as differences in DNA modifications between monozygotic twins in different environments and lifestyles. The piece culminated in bold statements like: “Epigenetics is the long sought link through which the environment influences the hereditary material [… and it] currently leads to a dramatic new understanding of human biology”.

In other words, neo-Lamarckism: the inheritance of acquired characteristics.  Well, in one way that’s true: stuff like methylation is an “acquired” characteristic that can be passed on for one or a few generations.  But it’s acquired via genetic instructions in the DNA, and it’s inherited for only a handful of generations. So, while important, it’s not a dramatic new paradigm of genetics. As Maderspacher says,

There is thus no need to construe a dichotomy between the power of the genes and the power of the environment — a molecular version of the ancient nature vs. nurture debate. The environment influences the phenotype through the genes. There is no contrast, no one over whom to achieve ‘victory’.

Indeed.  I’m not sure I agree with Maderspacher’s analysis of the reasons why this misconception is so important.  He floats the idea that Germans are particularly fond of  it “for historic reasons,” that is, because it apparently contradicts the hegemony of genetic determinism that undergirded Nazi racist ideology. But ultimately he lays the hype at the feet of Marxism, which trumpets the malleability of the individual by the environment. (This is where Lysenko comes in—the Russian agronomist whose fraudulent claim that one could permanently modify crops by environmental manipulation so impressed Stalin, and so ruined Russian agriculture.)  Maderspacher sees epigenetics as “a kind of lysenkoism for the molecular age.”

Well, maybe the popularity of epigenetics is a vulgarised environmentalist response to the “vulgarised genetic determinism” that so dominates our times.  And maybe that’s why journalists love it, though, as Maderspacher notes, they love anything that smacks of an overthrown paradigm—especially Darwinian evolution.  Regardless, he sees this kind of buzz-journalism as injurious to the public understanding of science, and I agree 100%:

Therefore, a larger frame has to be invoked, far-fetched as it may be. Building around the story is a legitimate literary technique to some extent, but becomes dangerous when the frame interferes with the presentation and interpretation of empirical data. In effect, it’s not far from what Lysenko did, and makes the whole purpose of science journalism questionable. It won’t cost lives as Lysenko’s mad ideas — after all, it’s only molecular biology — but the public have a right to be informed correctly. First, because they pay for the research. Second, because at the very least they need to know that science, and genetics in particular, cannot give them simple answers about who they are and how they should live, and neither can epigenetics. They’ll have to work that out for themselves and let Lysenko lie.

Well done!  But I’d go further and include among the miscreants those scientists—especially those evolutionists—who argue in the face of the data that epigenetics will overthrow conventional ideas about evolution and natural selection. Unlike journalists, they know better.

40 thoughts on “Epigenetics: the light and the way?

  1. Wow. I just did a PubMed search on “epigenetics” and got 39,157 hits, including 6,567 reviews and 16,930 free text articles.

    Any way to help the clueless but curious in the crowd (ie, me) separate the wheat from the chaff?

    I can’t read 16,930 articles between now and kick-off.

  2. Excellent article! I too, wouldn’t know where to start when it comes to epigenetics. It does appeal as an elegant explanation for environmental influences. However, I’m skeptical enough of my own, limited knowledge of genetics to think that any misgivings I might have are due to a misunderstanding or incomplete knowledge of ‘traditional’ evolutionary theory…

    You know, it would be a great topic for a book… 😉

  3. “…maybe the popularity of epigenetics is a vulgarized environmentalist response to the ‘vulgarized genetic determinism’ that so dominates our times.”

    Right. Anything that seems to defeat the imagined specter of determinism will be eagerly snapped up by a culture wedded to the myth of contra-causal freedom. But whether genetic or environmental or their interaction, it’s all deterministic, at least for practical purposes having to do with behavior (and indeterminism doesn’t help establish originative agency). To point this out isn’t being vulgar, but simply stating the obvious.

    Eventually we’re going to have to come to terms with determinism – the fact of our complete inclusion in natural cause and effect – as science closes gaps in our understanding. Killing off the “little god” of libertarian free will is the next step for atheists wanting to be pro-active in developing and promoting a fully naturalistic worldview. Sam Harris does a nice job of it pp. 102-112 in The Moral Landscape, plus he draws out some of the progressive implications for criminal justice. Hope his fans (and critics) take note, whatever the merits of his thesis about science and morality.


    1. Quantum Mechanics shows that the world is fundamentally stochastic not deterministic. DNA is molecular and molecules are in the quantum domain, this is why Schrödinger in the forties was able to predict that genetic information was coded at the molecular level rather than at the level of super-macromolecular assemblies, as most biologists believed at the time. He merely deduced that this was a consequence of X-rays causing mutations,

      As a chemist we tend to use the time-independent Schrödinger equation as a starting point for our calculations so only see a time averaged result in order to obtain molecular properties. This is fine systems in equilibrium. Biological systems are in principle out of equilibrium. So what we see is dependent upon the evolution of the state vector in time. It doesn’t matter if you believe in state vector reduction or take it’s unitary evolution seriously as Everett did in the relative state formulation of QM, what we see is fundamentally stochastic.

      Our world is not deterministic. Determinism is dead.

      1. Indeterminism may not of itself necessarily establish “originative agency” but it does establish potential and in a sense really existing contrafactual histories. This has important implications.

      2. Our world is not deterministic. Determinism is dead.

        Doesn’t that depend on your interpretation of quantum mechanics.

        My understanding is that on interpretations like Many Worlds, the randomness is purely subjective—it’s a matter of which universe you’re in, and the seemingly random events always happen and don’t happen in different universes.

        As I understand it, this general kind of interpretation is not dead, and is common among quantum cosmologists in particular.

        1. The ensemble as a totality is deterministic just as the unitary evolution of the state vector is deterministic. That is the bird’s eye view as Tegmark puts it. For any individual history it appears to be stochastic, this is the frog’s eye view as Tegmark puts it.

          Personally I favour the MWI (the Everett relative state formulation). However it still comes down to the fact the world we live in is stochastic.

          This is the lesson of QM and what separates it from classical theories like relativity.

          1. Oh yes, as the adventures of cGh Tompkins show, you just never know what might happen next in this world of quantum randomness.

          2. Personally I favour the MWI (the Everett relative state formulation). However it still comes down to the fact the world we live in is stochastic.

            Oops… I wasn’t reading carefully enough, and missed that you made clear what you meant by “fundamentally,” which was not what I meant by “fundamentally.”

            I’m still confused about
            what you find particularly interesting about whether the “world we see” is fundamentally random, or just pseudo-random.

            I think we need some clearer vocabulary for this sort of thing, but I don’t know what it would be.

            Many people seem to think that it’s a big philosophical deal whether things are “fundamentally” deterministic vs. nondeterministic, strictly speaking.

            I don’t see it. If my behavior is driven by a random number generator, it doesn’t generally matter whether it’s really, really fundamentally random, e.g., quantum noise, or if it’s pseudo-random.

            Either way, it’s noise for my purposes, and dependency on noise is not the same thing as any useful concept of freedom, or any particularly interesting high-level concept like that.

            I’m not clear on what “interesting implications” you see in quantum randomness. Whether things are effectively random in the middle seems much more important than whether they’re random at the bottom.

            Either way, higher-level systems may be resilient in the face of the underlying or external noise, or sensitive to it, and that’s usually the difference that matters.

      3. Quantum Mechanics shows that the world is fundamentally stochastic not deterministic.

        Curiously this is an analog to the genetic-epigenetics discussion.

        As you yourself admit, QM is a deterministic theory par excellence where ‘stochastic modifications are not inherited past one or two Planck times’. It is inherent in the unitary evolution that we can’t loose causality. And it is explicit determinism in the MW theory, which is the realistic QM.

        At the same time the fact that we live in one world of the many means that indeed what we see is fundamentally stochastic (in the “one or two Planck times” sense).

        However it still comes down to the fact the world we live in is stochastic. This is the lesson of QM and what separates it from classical theories like relativity.

        And this is the, again curious, analog to the nature-nurture discussion.

        It is actually classical systems which are unboundedly stochastic, as demonstrated by deterministic chaotic systems exponential divergence.

        Conversely non-classical quantum systems can demonstrate but limited chaos from other sources (say, “hockey rink” geometric effects) as they diverge much more orderly in their linear fashion.

        So long-time environmental stochasticity is overruling inherent short-time stochasticity, and it has nothing to do with QM. In fact QM tries its best to avoid it and it mostly slips in through the classical regime.

        I put that the ability of classical systems to act stochastic is, or should be, well known since the discovery of probability theory. And that, to continue on the string of analogs, the promotion of QM as the origin of stochasticity is overhyped.

        This hype is further a source quantum woo, and should be discouraged for the same reason as the overhyping of epigenetics should be. To paraphrase Coyne, one can see this “this kind of buzz-philosophizing as injurious to the public understanding of science”.

      4. Just to make myself clear, I note that I myself used to argue inherent stochasticity of QM specifically, out of irreversible environmental interaction during decoherence as I understand it today.

        But never, I hope, argued that “the world is fundamental stochastic not deterministic”. Even including decoherence and uncertainty, even including stochastic distributions, all of our laws evolve deterministically. (Or we would loose causality.)

        If anything causality and the concordant determinism is fundamental AFAIU, since having energy is to evolve “in time” or time and energy wouldn’t be classically complementary. For other systems that putatively jump all over time, see supernaturalism. 😀

        Also, speaking of QM overhyping, the sibling to overselling its stochasticity is to oversell its discreteness.

        From a classic axiomatic instrumentalist theory of QM one can see what is taking place is actually a seamless stitching together of discrete and continous states by way of their boundedness. Bounded systems exhibit discreteness (say, electron energy levels in atoms), unbounded systems exhibit continous properties (say, energy of free electrons).

        More fundamentally, the work of Lucien Hardy, translating classical and quantum theory into classical respectively quantum probability theory, shows that it is the classical world that is, this time genuinely, discrete.

        While quantum physics is the continous physics, necessitating a continous transformation between pre- and post observation states. Talk about putting old prejudices on their head!

        Interestingly, and tieing back to stochasticity, I believe Hardy’s work has been implicitly tested by modern measurements of, tentatively suggested as, decoherence. It turns out that decoherence may reject the “Copenhagen collapse” of states (and the whole Copenhagen theory with it).

        It seems one can take systems gradually in and out of decoherence. (Measurements on photon traps, IIRC. I’ll have to get back to you if asked for the exact reference.) This would, as I understand it, directly correspond to Hardy’s continous transformation of pre to post observation states.

        Now I ask, do this gradual, more precisely deterministic, irretriveable loss of information to the environment remind of stochasticity? To me it suggest that decoherence is but an entropic process.

        To wax philosophically, since the 2nd law of thermodynamics that allows entropy to increase fundamentally goes back to the inflationary caused cosmological expansion, it smacks of determinism to me.

        Determinism isn’t dead. Neither is stochasticity. They are married. (Puns on marital states vs death aside.)

  4. I think epigenetics is important! But you’re right, it’s been overhyped and isn’t going to revolutionize evolutionary theory at all.

    My usual answer answer when people ask me about epigenetics is that it’s a phenomenon that blurs the effect of the genotype over several generations…but fundamentally it’s still all about the inheritance of genetic traits.

    Did anyone freak out over maternal effect genes and decide they were going to overturn evolutionary biology? No. They’re also important but they aren’t going to change the way we think about evolution.

    1. Oh, I absolutely agree about their importance, even in evolution (that’s why I said it twice). After all, differential imprinting by males and females may be an evolutionary strategy to alter gene expression in adaptive ways. But that’s a fillip of evolution, not a new paradigm.

        1. I think you mean the now famous “Beard & Boot”.

          The other paraphernalia is “Tentacle & Paw”. Or should be paraphernalia, I’m not so sure in PZ’s case. … no, wait, there is “Puss in Boots”. Have anyone seen Jerry’s feet? His cordwainer, perhaps!?

  5. I think its currently very important in medical research since it is involved in the ongoing evolution of malignant cells.
    In terms of species evolution, well there is the problem that you tend to lose the epigenetic imprint during meiosis. A Jerry mentioned, it is very likely that you will completely lose any trace of a previous epigenetic imprint within a couple of generations. The current model of how this affects evolution is more along the lines of the imprint itself increasing particular mutations. In other words, a specific epigenetic imprint usually means a ‘mark’ or some sort, either a methylation of a cytosine in a CpG dinucleotide or a histone modification, within a gene promoter region. This mark has the function, while it is present, of changing the rate of transcription of the gene in question. Cytosine methylation also has the effect of altering the mutation rate at that nucleotide – thus the epigenetic mark may lead to a nucleotide mutation at the same point and therefore the ‘temporary’ epigenetic signal may be altered to a permanent ‘genetic’ signal. Obviously this is much less efficient than a straightforward mutation at the same point but it has the advantage that the mutation is ‘targeted’ for a promoter that seems to be affected by epigenetic alteration. This model at least provides some degree of plausibility as to how an environmental condition might ‘quickly’ cause a mutation in genes that respond to it.
    I’m not sure how well all of this has been proven experimentally but I think I’ve outlined the current model.

    1. I don’t know a real lot about evolution (I really only come to this blog because of Prof Coyne’s Puss ‘n’ Boots), but it did always strike me that organisms – especially big, complex organisms with complex needs – wouldn’t be ideally placed for ‘thriving’, or even surviving, if they couldn’t make some kind of short-term temporary response to a world that can change far more rapidly than either ‘natural selection’ or ‘gene drift’ or whatever could possibly respond to (even in a very short punctuation of an equilibrium).

      So firstly I though, aha, epigenetics is that very much needed ‘rapid response’ mechanism. And, or so I thought, that the response was short-lived was a good thing (who knows how the world will change in the next 5 minutes in this totally random quantum world ?).

      But, if the world doesn’t change back real quick, then an epigenetic change would be lost and the organisms would have to ‘reinvent’ it. Except that you now say that an epigenetic change can end up being ‘fixed’.

      Most interesting.


  6. How about epigenetic gene therapy?

    Transform my mass (and I have enough for two!) into the nymph(s) and I would change my name to “Doc Nymph.” Sort of like Thing 1 and Thing 2.


  7. But I’d go further and include among the miscreants those scientists—especially those evolutionists—who argue in the face of the data that epigenetics will overthrow conventional ideas about evolution and natural selection. Unlike journalists, they know better.
    Do they know better? Perhaps they should, but (to mention names, but who did you mean?) Eva Jablonka and Massimo Pigliucci are not getting off their hobby horse.

  8. “Epigenetics is the long sought link through which the environment influences the hereditary material”

    Well, it’s the sort of thing Lamarck and later Lysenko would dream about. I’ve met a few Lamarckians (and biologists at that) and am shocked that such people still exist. Oh well, I hear there are still flat-earthers out there.

      1. It would be nice to have a term for these new pseudoscientists of old dead theories. Say, if we call them “phlogindeadhorseys”, we would include them all, wouldn’t we?

        1. Very good: much more poetic than archosis and much less sordidly pedestrian, than ‘zombie’.

          Carried by acclaim !

  9. Maderspracher says “Science journalism, where it still exists,
    is part of the news industry, and thus
    needs to be newsy; ironically, that
    the environment can influence the
    phenotype and the genes is terribly
    old news, no news at all, really. […] Therefore,
    a larger frame has to be invoked,
    far-fetched as it may be. Building
    around the story is a legitimate
    literary technique to some extent, but
    becomes dangerous when the frame
    interferes with the presentation and
    interpretation of empirical data. In
    effect, it’s not far from what Lysenko
    did, and makes the whole purpose
    of science journalism questionable.”

    Well, all true – but it is not just the the journalists – scientists & universities need to shoulder some blame. This links in with competition for research grants & getting funding plus staying ‘contemporary’. People want all research to be ‘ground breaking’ rather than ‘ground covering’ as it should be as well, if you see what I mean.

    1. Possibly, but it should be a safe environment outside of press releases and interviews IMHO. But it is your show.

      (Btw, as it is today, I believe I will settle for “grounded” or at least “ground pointing”. (¬_¬) )

  10. Not surprised at all. I am always wondering why the SPIEGEL is hailed as Germany’s best and most sophisticated magazine. All their articles are based on argumentation by carefully selected anecdote, and whenever they report about a topic that you are familiar with yourself, you scratch your head and go: “What are they talking about? And did they not do any research whatsoever apart from interviewing one or two people promoting an isolated extreme view of the matter?”

  11. As a computer scientist viewing the genome as a kind of program—not a von Neumann program, of course—I’m wondering what computational abilities are added by things like methylation.

    There’s already a capacity for passing state information from one generation of cells to the next, from he basic way genes work and the way that cells divide to make two—right?

    Genes interact by producing transcription products that switch other genes on or off. (Or change their rates of firing in an analog way.) Those transcription products diffuse around the surrounding plasm and dock to repressor and promoter sites of other genes to influence their rates of transcription.

    When the plasm is split along with the cell, some of those products go into each of the resulting cells, and hey, presto, that state information is transferred from the original cell to both copies. Right?

    That’s the usual thing that happens by default, right?

    And it’s under genetic control, too—you can have genes switch on or off and quiesce before the cell divides, to make sure that state information is transferred neatly.
    (Ensuring that the relevant transcription products concentrations are appropriately high or appropriately low for whatever state you want to transfer—e.g., that certain genes will stay on, and others will stay off, across the split.)

    In computer nerd terms, this is like the fork() operation in UNIX, which creates a new process by copying the entire state of the parent process, and setting the copy running. (But passing it a one piece of information to tell it it’s a new copy, rather than the parent. Other than that, all the state information for the parent process is transferred to the child process by default.

    Given my mental model of these things, the ability to pass state information from one generation of cells to the next is the furthest thing from new—it’s the absolutely normal, default thing that happens when cells divide, just as it’s the default when forking processes in UNIX. Methylation and whatnot just give you another way of doing something you could do before.

    Am I mistaken?

    What is the significance of using those mechanisms rather than the default one, i.e., transfer of transcription products through the plasm surrounding the genes?

    (E.g., if I were designing a synthetic organism, would it be neater to do it all the default way, such that methylation is just a redundant kludge that evolution happened to come up with, or is their some good reason to transfer some information one way and other information the other?)

  12. I’d like to point out that there IS a form of non-genomic inheritance that may actually be more prevalent than currently known (as no one really looked) – cellular (cytoplasmic and cortical) inheritance of organisation of components (eg some cytoskeletal systems). Sonneborn and Beisson (1965, PNAS) have a classic experiment wherein a row of cilia in paramecium was inverted surgically, and inherited in the following generations (for thousands of them – the original strain may well still be alive, haven’t checked though), even though no genetic content was changed. Subsequent experiments showed that modifying the genetic content via multiple crosses, etc (in case somehow the genome was ‘altered’ by the surgery – which would also have been amazing) did not alter the inverted row; meanwhile, this row was only inherited cortically, so the modified cilia were not transmitted to the other sex partner. Thus, it is a solid and fascinating example of non-genomic inheritence in the most definite sense.

    Sadly, since most cell biologists work on big things and could care less about evolution, this phenomenon has been only barely investigated, and mostly in ciliates.

    Speaking of ciliates, their nuclear dimorphism is absolutely the most amazing example of epigenetics, ever. I highly recommend reading up on the effect of the old [somatic] macronucleus on the development of the new macronucleus from the [germline] micronucleus.

    And the gene-silencing methylation is just boring. Mostly restricted to embryonic plants and animals, from what I know, possibly as a result of conflict between the male and the female parents, if such theories are to be believed. I’m annoyed they hijacked “all” of epigenetics that way…

    I do agree with the hype aspect. They hype the wrong things. Cellular inheritance, on the other hand, is completely obscure and desperately in need of attention.

    (You can also see the book by Grimes & Aufderheide 1991 “Cellular aspects of pattern formation: the problem of assembly”; as well as Cavalier-Smith 2002 “he membranome and membrane heredity in development and evolution” in ” Organelles, genomes and eukaryote phylogeny” for further discussion)


  13. @MadScientist I’m shocked people make such a big deal over “Lamarckian” vs “Darwinian”. Both terms, and associated accusations, are bullshit. Both of the guys are long dead, and have little to offer to modern evolutionary biology. We have moved on. I hope it doesn’t shock you that I am apparently “Lamarckian” in some ways, for reasons described in my post above. But seriously, who cares? The data is what it is, no need to soak everything in philosophy and intradisciplinary politics!

Leave a Reply