Wednesday, July 23, 2008


Epigenetics is one of those words that means entirely different things to different people. P.Z. Myers has put up a nice description of the term on his blog [Epigenetics]. Here's how he defines epigenetics ...
Epigenetics is the study of heritable traits that are not dependent on the primary sequence of DNA.
In fairness, he then goes on to explain that this is an unsatisfactory definition. That's an understatement.

Now, as it turns out, those scientists who work on animal development employ a definition of epigenetics that looks very much like what we used to call developmental regulation of gene expression. That's why PZ can say ...
... developmental biology basically takes epigenetics entirely for granted — development is epigenetics in action! Compare an epidermal keratinocyte and a pancreatic acinar cell, and you will discover that they have exactly the same genome, and that their profound morphological, physiological, and biochemical differences are entirely the product of epigenetic modification. Development is a hierarchical process, with progressive epigenetic restriction of the fates of cells in a lineage — a dividing population of cells proceeds from totipotency to pluripotency to multipotency to a commitment to a specific cell type by heritable changes in gene expression; those cases where there is modification of the DNA, as in the immune system, are the exception.
Here's the problem. If this is epigenetics then what's the point? When I was growing up we had a perfectly good term for these phenomena—it was regulation of gene expression. Why is there a movement among animal developmental biologists to use "epigenetics" to refer to a well-understood phenomenon?

I've been bugging my colleagues today by asking them to tell me whether certain examples of gene regulation are epigenetic or not.1 The answers are mixed so I thought I'd submit the questions to Sandwalk readers. Which of the following are "epigenetic"?
  1. Consider an E. coli cell that grows and divides for hundreds of generations in the absence of any exogenous β-galactosides (e.g. lactose). Under those conditions the lac operon is repressed and this state is heritable from generation to generation due to the presence of lac repressor.
  2. Consider mating type in yeast. In an α cell the a gene is suppressed from generation to generation. This is heritable regulation of gene expression. All daughter cells inherit the ability to express the α gene and suppress the a gene.
  3. During a bacteriophage infection certain genes are turned on in a definite sequence. In the simplest cases there is a set of "early" genes that are expressed as soon as the 'phage DNA enters the cell. After a few minutes the expression of the "early" genes triggers the expression of the "late" genes. Note that the "late" genes are not transcribed initially even though they are present.
  4. Right now your major heat shock genes (e.g. Hsp70 genes) are transcriptionally silent. However, if you are stressed by heat those genes will become active and will be transcribed at a very high rate.
  5. During oogenesis in fruit flies the bicoid gene is expressed in nurse cells and bicoid mRNA is deposited in the egg. In males, the bicoid gene is never expressed.
  6. One of the nucleotides at an EcoR1 restriction endonuclease site in E. coli is methylated. This blocks cleavage at that site, thus protecting the bacteria from degrading its own genome. The methylation pattern is inherited from generation to generation by the action of a methylase enzyme.
  7. Globin genes are expressed in erythroblasts but not in brain cells. During development the globin genes are activated in erythroblast stem cells because certain activator proteins are synthesized. The globin genes are not activated in any other tissues.
  8. During development in mammalian females one of the X-chromosomes is randomly inactivated [Calico Cats]. Once this occurs the pattern is inherited in (almost) all cells that descend from the initial embryonic cell where the inactivation first occurred. The same X-chromosome is inactivated in all daughter cells.
I'm interested in two questions. First, is it possible to define epigenetics in a rigorous manner so that we can decide whether certain cases are "epigenetic" or not? Second, what, if anything, is the difference between "epigenetics" and "developmental regulation of gene expression"?

1. And they are quite annoyed about it. Many of them are avoiding me because they don't know how to answer the questions.

[Image Credit: The cartoon is from Mark Hill's website at the University of New South Wales, Australia. It appeared originally in Nature. The figure represents a different definition of "epigenetics"—one that focuses on modifications to DNA and histones.]


  1. My definition (mine because that's what I believe, not because I came up with it) of "epigenetic" requires covalent modification of either DNA directly, or of nucleosomes. This includes CpG methylation as well as histone acetylation and methylation only. I haven't made up my mind whether or not phophorylation counts. How that applies to your list is left as an exercise for the reader. :)

  2. I searched PubMed for the term 'epigenetic*'. One of the earliest hits was DL Nanney (1958), Epigenetic Control Systems, PNAS 44(7):712-717. It's available as full text from PubMed Central.

    I don't know how authoritative Nanney is, but what he calls epigenetics would probably encompass all of Dr. Moran's examples. So even if that's not the most widely accepted view today, there's clearly precedence for it.

  3. So, development is "regulation" of gene expression?
    Who regulates the regulators, Larry?

    That is just flushing biological complexity down the toilet in favor of a lousy abuse of the term "regulation". If genes are expressed differently, it is also because of cytoplasmic regionalization and assymetric division, differential environmental exposure...STRUCTURAL processes, laws of physics, chemistry and form (geometry)

  4. Let me attempt to answer.

    1. Not epigenetic. The activity of the lac operon is dependent on the presence or absence of β-galactosides. Thus the sate is not stably heritable and rather is dependent on the environment.

    2. Not epigenetic. Mating type in yeast is actually dependent on the DNA sequence - mating type switching involves DNA recombination. So, by definition, mating type in yeast per se is not epigenetic. On the other hand, the way the silent mating cassettes are silenced is epigenetic.

    3. Not epigenetic. Expression of such genes are not stable, but is sequential.

    4. Not epigenetic. It is not stably inherited and is dependent on the environment.

    5. I'm not sure. I think it depends on how much of the nurse cell identity and bicoid expression are dependent on cell-cell communication and how stable they are.

    6. Epigenetic. It does not involve the DNA sequence itself and is heritable.

    7. It's hard to say. I don't want to call it epigenetic. But I can see the argument that it is epigenetic.

    8. A classic example of epigenesis.

    I don't think there is a definition of epigenesis that everyone can agree on. For me, what is important is that it is stably heritable and cell autonomous.

    I do think that there is difference between "epigenetics" and "developmental regulation of gene expression." The latter is often dependent on cell-cell communication and is not necessarily cell autonomous. But epigenesis does play a role in development.

    I disagree with the annonymous' comment that epigenesis requires covalent modifications of DNA or nucleosomes. Yeast prion (PSI+) is an example of epigenetic phenomenon that involves neither.

  5. I would call 6 and 8 epigenetic, the other cases examples of differential gene expression.

    Epigenetics originated with C.H. Waddington, as far as I know, in his 1957 book "The strategy of the genes". Waddington gives a well-known figure of an epigenetic landscape, that determines the phenotype through gene interactions. That book must have been long before any molecular example of gene regulation. As far as I can see, what Waddington said has been superseded by molecular biology and developmental biology, and is no longer of any but historical interest.

    Eva Jablonka & Marion Lamb posit that epigenetic inheritance systems are an important addition to evolutionary biology,and that Waddington’s ideas have been sadly ignored. Their book “Evolution in Four Dimensions” is partly devoted to arguing that epigenetics has been almost criminally neglected and that therefore evolutionary biology is woefully inadequate. J&L seem to suggest some neo-darwinian genetic establishment is opposing their ideas, but then, their description of standard evolutionary biology seems not to reach beyond taking the first-year one locus model as the only evolutionary model available.

  6. The term epigenesis in development originates in reference to the debate against preformism. The basic idea of epigenesis is from aristotle. Unfortunately
    Aristotle considered this to be evidence some "final cause" was "giving shape" to initially shapeless early stages. A similar role was given to entelechy by the vitalists. The modern notion substituting for a final cause is that of a "genetic program".

    Though preformism is definitely not the case, the zygote is not "formless", either.

    Let's invert the question to place the burden on the geneticists. Which of the cases above involves genetics? Probably all of them, since involve genes. But is everything genetic? If not, how do we call what isn't?

    Whta in the end we have is a poor acknowledgement of the fact that it's not all genes, or simply "a matter of gene regulation"

  7. Not to be picky but EcoR I is not blocked by methylation in E.coli. EcoR I recognition site does not overlap with neither Dam not DCM methylase recognition sites. If it were, EcoR I wouldn't be the cheapest and popular restiction enzyme used in Molecular Biology.

    And it woldn't be epigenetic anyway because there are no environmental factors that would affect methylation state in such a way that the modification is inherited by a daughter cell even without such invironmetal factors.
    (I.e., you can inhibit methylases but the moment you remove them, the progeny will have all its DNA methylated).

  8. dk says,

    Not to be picky but EcoR I is not blocked by methylation in E.coli.

    Yes, it is—in strains that have the restriction endonuclease. Almost all restriction endonucleases are associated with a corresponding methylase and EcoR1 is no exception. The methylase is EcoR1 methylase and it methylates the second adenylate residue in the EcoR1 site [Restriction, Modification, and Epigenetics].

    EcoR I recognition site does not overlap with neither Dam not DCM methylase recognition sites.

    That's correct, but irrelevant.

    If it were, EcoR I wouldn't be the cheapest and popular restiction enzyme used in Molecular Biology.

    I don't get your point. All restriction enducleases are purified from organisms (usually bacteria) whose genomic DNA is methylated to prevent cleavage.

    When you use E. coli as a host to clone DNA you have to use a strain that lacks the restriction enzyme. It is relatively simple to create strains lacking the EcoR1 restriction-modification system since it's on a plasmid. It is more difficult to eliminate the other restriction endonucleases in E. coli. You have to create strains with mutated methylase and restriction endonuclease genes.

  9. Larry, you are right of course. My mistake. I was thinkng of all the E.coli strain that lack EcoRI restriction-modification plasmid - which, I believe, are the majority of strains.

  10. Re: what, if anything, is the difference between "epigenetics" and "developmental regulation of gene expression"?

    Epigenetics is "the study of the mechanisms of temporal and spatial control of gene activity during the development of complex organisms." - according to R Holiday (1990). It is thus the name of an area of study.

    Some modern folk have hijacked the term and used it to mean something quite different - but that's another story.