Monday, March 24, 2014

What is epigenetics?

Several students in my class decided to write essays on epigenetics. This was very brave of them since nobody seems to have a good definition of epigenetics and much of the hype about epigenetics is not very scientific. I'm also more than a little skeptical about some of the claims that have been made.

Here's a video. What do you think? Is this a useful contribution to our understanding of a complex issue? Is the inheritance of methylation sites at restriction/modification loci in bacteria an example of epigenetics? After E. coli divides, both cells inherit some lac repressor molecules and the lac operon is not expressed provided the parent wasn't exposed to lactose. Is this epigenetics?



61 comments :

  1. As a scientist you can not be skeptical because you will be closing the doors for new discoveries.
    Credulity and incredulity, belief and unbelief, are both equally biased ways of thinking.

    ReplyDelete
  2. I think you get some awesome points for linking to SciShow. :)

    ReplyDelete
  3. It looks as if these days any non-genetic inheritance counts as epigenetics; in fact, that's pretty much all it seems to mean now. All very fine unless someone wants to claim it means much for evolution.

    ReplyDelete
    Replies
    1. Why don't you admit it...? You have no clue and even if you do, you will still deny it... You were programmed that way... Whatever contradicts your believes is has to be rejected... Scientific facts don't matter...

      Delete
    2. What is it about Creationists and their tics? Now, there's just no way Quest could be John Witton. But compare and contrast...

      "I'm a "hair-stylist"...So, what do you want from me...? A blog devoted to my humble personality tells me that I must have stepped on some toes...What worries me is what is going to happen, when I begin to post the real stuff... I had to restrain myself from writing the details even though I was soooo tempted... I'm sure you will understand when this contest is over...
      PS. I love you Larry... you are so predictable... ;)"

      Delete
    3. Yes, I think that is one of the two meanings of epigenetics nowadays. Either epigenetics is used to specifically indicate DNA methylation, histone modification and chromatin remodelling, etc. Or it just means any heritable factor that is not coded in the DNA sequence. That's not quite the same thing.

      Delete
    4. Seems pretty convincing they are the same person, yes. The same compulsion for telling lies and abuse elipse usage is certainly there.

      Delete
    5. I think there exists some confusion about what exactly the central dogma is and what the central dogma is not.

      For example – here is an interesting blog entitled: A Challenge to the Supremacy of DNA as the Genetic Material

      http://blogs.plos.org/dnascience/2014/03/20/challenge-supremacy-dna-genetic-material/

      Check out the original article

      As Larry & PZ Myers have pointed out on many occasions, It has been long understood that a simplistic interpretation of the central dogma is a straw man. This is nothing new!

      One cannot deduce phenotype from gene sequence because the cell as an interacting chemical system is too complex.

      This blog’s citation is kinda neat is the physics vocabulary to reassess the problem from a slightly different angle.

      Of course, tying this in with epigenetics, we can consider a long observed phenomenon of imprinting.

      For example, hinnies are not different from mules because nucleosome modifications are different in the zygote. A hinny zygote has is diploid cell with two sets of chromosomes: one from a donkey and another from a horse; ditto the mule. Any differential nucleosome modifications should be identical in both zygote scenarios, yet hinnies and mules are quite different. That must be due to maternal the transcription factors in the female horse that gave birth to a mule are different from the transcription factors in the female donkey that gave birth to a hinny.

      In conclusion nucleosome modifications may be necessary for epigenetic responses, but they are not sufficient. Kudos to Mark Ptashne for forcing everybody to refocus.

      Recent advances in technology have given rise to a whole new generation of “DNA–chompers” who have had little, if any, training in classical genetics. We are essentially reinventing the wheel! According to Ptashne, this younger generation of DNA-chompers would not be getting some of their “facts” wrong (in this case the Epigenetics story) if only they paid better attention to the work of their predecessors.

      Delete
    6. Clearly from Larry's thesis title, Larry also numbers among those dwindling few still adept in Classical Genetics.

      Larry - je vous tire mon chapeau jusqu'au terre!

      Bravo for drawing this non-issue of epigenetics to eveybody's attention.

      BTW - There was a recent seminar in Dalhaousie that discussed this very problem (among others)

      Hype In Science http://www.situsci.ca/event/hype-science

      Ford Doolittle came up with a clever title picked up by Florian Maderspacher,

      "Epigenetics and the New Lysenkoism."

      Delete
    7. oops - typo correction to the penultimate post above

      Any differential nucleosome modifications should be identical in both zygote scenarios, yet hinnies and mules are quite different. That must be due to maternal CYTOPLASMIC INHERITANCE, i.e the transcription factors in the female horse that gave birth to a mule are different from the transcription factors in the female donkey that gave birth to a hinny.

      apologies

      Delete
    8. "any non-genetic inheritance counts as epigenetics" .... "All very fine unless someone wants to claim it means much for evolution."

      I'm curious about how this argument works so I'll have a go.

      First I want to exclude gut flora and the like from "non-genetic inheritance", difficult to argue evolutionary effects there, although there might be some directed feedback to the genome. I'll stick with (copied from Corneel's post): DNA methylation, histone modification and chromatin remodelling.

      Taking methylation as the model I can conceive three levels of analysis:
      (1) The possibility that methylation can occur (or not) in a genome.
      (2) The heritability of specific methylation features down a generation of two
      (3) The heritability of the ability to inherit as per (2)

      Presumably (3) the ability to pass certain methylation features down a few generations is itself a heritable characteristic, and could (in principle at least) help family lines survive in environments where for example: climate remains relatively constant for tens to hundreds of years and then changes.

      So, it seems to me that the ability of an organism to modify the characteristics of its descendants, and the specific ways that modification occurs could be selected for by the continued survival of some family groups versus others and thus have "meaning for evolution".

      And having got this far I just realized that I'm talking about #666 group selection.

      Delete
    9. @ Keith Elias

      I think numbers (1) and (2) are now established quite well, but there are other issues with evolution by epigenetic mechanisms. New epimutations are not stable in evolutionary time. The majority of these revert back to the ancestral allele in a few generations, and it is very unlikely they can contribute to differentiation between species. However, some people argue that plastic responses can become fixed in a lineage. In the movie there was an example where people born in periods of food scarcity were more thrifty. This is just developmental plasticity, but could in principle become hardwired in the genome if the phenotype is adaptive (e.g. there is now permanent food shortage). The mechanism by which this should be accomplished has never been explained to my satisfaction, but there it is; evolution by epigenetics.

      Delete
    10. Yes, I think that is one of the two meanings of epigenetics nowadays. Either epigenetics is used to specifically indicate DNA methylation, histone modification and chromatin remodelling, etc. Or it just means any heritable factor that is not coded in the DNA sequence. That's not quite the same thing.
      Epigenetics as a concept has been around much longer than knowledge of histone modification, DNA methylation, and chromatin remodeling. Personally I favor the original meaning, not the more limited one that I believe has been promoted by cancer biologists who fail to realize biological concepts exist before they think of them. (ENCODE is a prime example of this.)

      Delete
  4. Short Answer – if by Epigenetics we ONLY mean eukaryotic nucleosome modification, then yes, the inheritance of methylation sites at restriction/modification loci in bacteria an excellent comparison to epigenetics’ nucleosome modification in eularyotes.

    Restriction Enzymes are homodimers that recognize their sites recognize specific stem-loop structures permitting site-specific endonuclease cutting of DNA. In other words, they recognize palindromes or stem-loop DNA. That is only half the story. Every restriction enzyme is paired with a methylase enzyme that methylates exactly the same DNA palindrome thereby making it immune to restriction enzyme cleavage.

    So why is host-bacterial DNA immune to restriction-endonuclease cleavage while invading phage DNA is so sensitive to restriction endonucleases? Bear in mind that bacterial host DNA is constantly undergoing replication, which means that bacterial host DNA is constantly presenting unmodified DNA to its own restriction endonucleases. Werner Arber worked that story out: http://library.cshl.edu/oralhistory/speaker/werner-arber/

    Werner Arber determined that partially-or hemi-methylated DNA is NOT a substrate for restriction endonucleases. Parental strands are always protected, leaving invading unmethylated virus vulnerable to restriction enzymes.

    On January 2011, Pope Benedict XVI appointed Arber as the first Protestant to hold the office of President of the Pontifical Academy of Sciences… go figure!

    So - propagation of nucleosome modification by enzyme machinery copying the DNA –methylation/Histone-acetylation status of the parental strand would at first glance be a great introductory model to indtroduce epigenetics.

    But alas, the story is not so simple!

    ReplyDelete
  5. ... con't

    Eukaryotic epigenetics is akin to

    1.- restriction-modification system of bacteria thereby globally shutting domains of genes on the understanding that comparable eukaryotes nucleosome modifications “decay” and require “maintainance”

    2. – repressible operons in bacteria such as the Trp Operon (or better yet lambda phage lysogenesis)

    We need to remember that transcription factors are required for global shut down gene expression. Similarly eukaryotes require similar transcription factors to maintain nucleosome modification (amoung other modifications to DNA architecture let us not forget).

    Mark Ptashne described it best in a recent letter to PNAS. http://www.pnas.org/content/110/18/7101.full#ref-1

    it is patently incorrect to :
    “… refer to nucleosome modifiers as “epigenetic”—they, like the very many proteins recruited to genes by specific transcription regulators, are parts of a response, not a cause, and there is no convincing evidence they are self-perpetuating.” Mark Ptashne

    ReplyDelete
  6. When discussing metabolic pathways, “positive feedback” is often misconstrued with “precursor activation” to maintain homeostasis. In fact, “Positive feedback” and “homeostasis” are generally considered mutually exclusive. During “Positive feedback” in a metabolic pathway, the end products would “feedback” to increase the concentration of the end product even more. This scenario results in an ever accelerating deviation from a set point! Positive feedback in Biological systems leads to “boom and bust” or “fill and expel” scenarios; in other words, an escalating (often oscillating instability) which constitutes the antithesis of homeostasis.

    One example would be the Positive Feedback regulation of lymphocyte signaling which requires an urgent amplified and fast response to pathogen attack. Other examples include blood clotting, the “all-or-nothing” action-potentials of neurons, and nervous regulation of defecation or urination, lactation precipitated by Oxytocin response to suckling, child-birth precipitated by ever increasing uterine contractions.

    In all cases, the response amplifies the initial stimulus in a single pathway leading to an escalating instability that resolves itself quickly… called “terminal events” of positive feedback.

    However, Mark Ptashne explains that suggestion that genetic control in bacterial/bacteriophage systems maintaining homeostasis are often examples positive feedback as well.

    This is a subtle point: inducible vs. repressible gene regulation should be considered separately from positive vs. negative feedback. Gene regulation can invoke positive feedback without necessarily resulting in a “terminal event”. Positive feedback is important in a variety of biological scenarios such as commitment steps in the cell cycle, the menstrual cycle or even cell differentiation. Epigenetics also includes positive feedback control at the level of gene regulation.

    ReplyDelete
  7. Lysogeny is akin to the inducible Lac-operon. The cI protein (called the λ repressor) is distributed to daughter cells during binary fission. Exposure to a signal (e.g., UV light) inactivates the repressor, viral genes are turned on and viral replication together with lysis ensues.

    In fact, many parallels with the Trp Operon also exist. Expression of the Tryptophan Repressor protein is regulated by its own protein product by a process called “autogenous regulation”. The Trp repressor binds to an operator that precedes the Trp gene itself thereby keeping repressor levels low (about 20-30 molecules per cell) allowing the system to be very responsive to Tryptophan fluctuations.

    Back to λ bacteriophage cI proteins: As in the Trp operon, “Autogenous regulation” again occurs such that the prophage dictates the synthesis of a low level of cI repressor protein binding to a promoter site called PRM (Promoter for Repressor Maintenance).

    This where it gets tricky: cI protein is simultaneously a repressor in one DNA direction as well as transcription activator on the same stretch of DNA, but in the other direction.

    While the cI protein is activating expression of its own gene (a positive feedback reaction) it is simultaneously repressing transcription of viral genes to the right (including the cII gene that initiated cI’s expression to begin with). Ptashne is correctly citing an example of gene regulation that simultaneously demonstrates repressible gene regulation (lysogeny), inducible gene regulation (lysis) and positive feedback (cI autogenous regulation). (BTW, cI also prevents superinfection by any λ virus that happens to enter a cell already possessing a propage – neat or what?!)

    (Aside Quick question: If positive feedback is occurring, why do cl levels not rise until the entire cell fills up with cI protein? Ans: that's where cII comes into play: Overshoots of cI supress expression of cII which in turn lowers levels of cI.)

    Ptashne nailed it! Classical Geneticists have known for decades that there exist gene regulatory models where Positive Feedback control in fact maintains homeostasis, if by homeostasis we understand “status quo”! It gets better: irreversible commitment points often occur in Biology. Self-perpetuating responses are then required long after the triggering stimulus is removed and these responses occur as a result of positive feedback as well as double-negative feedback mechanisms as first described by Jacob and Monod. http://tinyurl.com/pqx4jom

    James Ferrell does an excellent job of elucidating these difficult concepts.
    http://www.medicine.mcgill.ca/physio/mackeylab/courses_mackey/pdf_files/ferrell_2002.pdf

    To summarize in terms as the Jesuits would phrase it: In conclusion nucleosome modifications may be necessary for epigenetic responses, but they are not sufficient.

    Kudos to Mark Ptashne!

    Bottom line, Epigenetics is more about Classical Genetics and requisite transcription factors controlling gene regulation than standard textbooks would have us understand.

    ReplyDelete
  8. Final Word to Mark Ptashne:

    “… all attempts to show that such [nucleosome] modifications are “copied along with the DNA,” … , have, to my knowledge, failed”

    So how does Epigenetics happen?

    Again to Mark Ptashne:

    "The answer—where we know it—is positive feedback, a process understood for years for bacterial regulatory systems.”

    ReplyDelete
  9. OOPs - I almost forgot:

    I direct everyone's attention to PZ Myers' latest & greatest in his cancer series that directly contradicts many current over-simplifications of Epigenetics that often consitutes textbook dogma.

    http://scienceblogs.com/pharyngula/2013/10/12/micrornas-and-cancer

    Myers picks up where Ptashne left off - without continual recruitment of transcription factors, nucleosome modifications will inexorably decay. That is what makes epigenetics “epi” in the sense of “nearly” or “not quite” irrevokable inheritance in the classical sense of genetics as William Bateson first imagined.

    To understand epigenetics, we need first to understand the cytoplasmic transcription factors (and other regulatory elements) that maintain epigenetic memory and how this memory fades as transcription factors and other regulatory elements can dilute.

    One key candidate mentioned by Ptashne and examined in great detail by Myers would be miRNAs.

    As luck would have it - I am about to cover this topic with my AP Biology class. I basically posted my lesson plan. I really appreciate feedback; suggetions for improvement or any correction before passing this on to my students.

    ReplyDelete
  10. I would love to see some numbers about how much of the heritable variation turns out to be epigenetic. My guess would be that it is a small fraction, but I am not aware of any studies that tried to quantify that. Anybody knows?

    BTW since when were genes considered to be the "be-all and end-all of who you got to be"? With regard to the intelligence example, I believe that 50 years ago behaviorism was the norm in psychology, not genetic determinism.

    ReplyDelete
  11. Still – I remain unclear on one nagging question:

    How do new gene imprints become properly reestablished in the Primordial Germ Cell genomes? How exactly does nucleosome modification track one pattern for male gametes and a very different one for female gametes?

    Let’s redirect our attention from mules and hinney’s and focus on the murine model.

    In spermatogenesis, all imprints are erased and rewritten in the paternal pattern, even for genes that came from the mother.

    Meanwhile, in oogenesis, all imprints are erased and rewritten in the maternal pattern, even for genes that came from the father.

    That is tantamount to a male hiney producing nucleosome modifications equivalent to horse sperm whereas a female hiney only produces the nucleosome modifications equivalent for donkey ova…

    … things that make you go hmmm! Very curious.

    Maternal cytoplasm was identical in both scenarios. Clearly something else is also going on.

    http://learn.genetics.utah.edu/content/epigenetics/imprinting/

    So how does this differential genomic imprinting occur? It must be due to the XX vs XY status of the primordial germ cells. I am only guessing here, perhaps there is something to do with Xist.

    In any case, the obvous erasure of imprints in Primordial Germ Cells seems to contradict the naïve explanation of epigenetics in the original video.

    OK – now I am in over my head. Suggestions?

    ReplyDelete
    Replies
    1. Tom Mueller How do new gene imprints become properly reestablished in the Primordial Germ Cell genomes?

      In a context dependent way.

      DNA methylation is erased (global demethylation) from all single copy genes when PGC migration reaches the (future) gonads. There are exceptions of course. After sex-independent demethylation, imprinting/reprogramming is established in a sex-dependent manner.

      As to the factors involved, one of the biggest players is DNA methyltransferases. Loss of some Dnmt genes results in various failures to imprint in certain chromosomal locations called Imprinting Control Regions. At the beginning of imprint erasing, Dnmt3 transcripts are absent in the nucleus. During methylation establishment, transcript (and protein) increases greatly in the PGC nucleus. Methylation levels of ICRs increase for the first time as well. It all suggests the Dnmt3s are responsible for imprinting. But it's actually more complicated than that because during methylation reprogramming, de novo methyltransferases are variable between male and female PGCs. Context dependent.

      Delete
    2. @ caynazzo

      I must be missing something. Help me out here.

      I reread you reply and for the life of me I cannot understand how this vexing question is answered:

      In spermatogenesis, all imprints are erased and rewritten in the paternal pattern, even for genes that came from the mother.

      Meanwhile, in oogenesis, all imprints are erased and rewritten in the maternal pattern, even for genes that came from the father.


      I understand from Hein that

      Normally, methylation patterns are copied when cell divides

      Sperm and egg cells have (almost) normal methylation patterns

      After fertilization:
      – the Paternal genome is (almost) demethylated within 4 hours
      – the Maternal genome more slowly (passively?) demethylated
      – From 120-cell stage, DNA is slowly re-methylated until birth


      I guess my question is:

      Exactly how do your Dnmt3s behave differently in a nucleus with an XY complement than in the nucleus with an XX complement? What are the contextual determinants?

      What am I missing?

      Delete
    3. I recommend this paper: Hajkova,et al., 2002, Epigenetic reprogramming in mouse primordial germ cells.

      Methylation imprints in male mice PGCs coincides with spermiogenesis (16 days post partum). In females it coincides with oogenesis...much later.

      The de novo methylatransferases are sex specific.

      Refer to Kenada et al., 2004 Essential role for de novo DNA methyltransferase Dnmt3a in paternal and maternal imprinting.

      Dnmt3a is required for methylation imprints in both males and females. Dnmt3l is exclusive to methylation imprinting on female germ cells.

      Delete
    4. @ caynazzo

      My problem appears to have been very naïve in the extreme! If both paternal and maternal chromosomes are completely erased, there is no way for enzyme machinery to distinguish between the two

      Here is the salient quote right from the introduction:

      The fertilized egg first undergoes a wave of demethylation during preimplantation development, which erases part of the inherited parental methylation pattern. The DNA of blastocysts is thus relatively undermethylated. After implantation, the embryo undergoes a wave of de novo methylation that establishes a new embryonic methylation pattern.

      So in fact, paternal and maternal chromosomes can still carry different epigenetic tags, even after "demethylation", which is in fact only “partial”.

      All of a sudden, this is beggining again to bear an uncanny resemblance to Werner Arber's endonuclease story above.

      I hope I got that correct.

      Thanks for your help.

      Delete
    5. All single copy genes in the PGCs are demethylated as far as I know. The "part" you're referring to are mostly centromeric and pericentric repeats which are involved in genomic integrity. Although there seems to be some evidence for repeats (which are targeted for methylation) establishing methylation imprints.

      Lees-Murdock et al., 2003. Methylation dynamics of repetitive DNA elements in the mouse germ cell lineage.

      There is however something called allele discriminating signals that "sense" (based on transgenic implant studies) the parental origin of an allele and protect against de novo methylation. Some sort of trans-factors interact with the allele discriminating signal which are present in whole protein nuclear extracts from ES cells, but I don't think anyone has figured out what they are.

      Delete
    6. @caynazzo

      Very Very interesting!!! Thank you!

      Which brings right back to Mark Ptashne's interpretation. I was wondering about that.

      But I am still wondering - all the paternal alleles will occur only on the one paternal chromosome homologue (ditto maternal alleles) so why invoke trans-factors if chromosome identity (paternal vs. maternal) could already occur with differential residual methylation?

      Delete
  12. Ask a developmental biologist. We say epigenetics is really important in development and in physiological adaptation -- it's good to know more about it, and is essential for understanding the state of the organism. But evolution? Meh. Acquiring the process of semi-permanently modifying the cell state is something that was a key innovation (OK, many innovations) in evolution, but it's been overhyped as an information transfer process on evolutionary timescales.

    ReplyDelete
    Replies
    1. We're only going to cover epigenetics for a few minutes in today's class because there aren't any serious arguments in favor of changing our view of evolution because of epigenetics.

      However, we are going to spend a lot more time learning that evo-devo is just as stupid because there are some seriously misled developmental biologists who think that discoveries in their field change how we should view evolution. :-)

      Delete
    2. If I may be so bold? " What epigenetic modification does is broaden the range of phenotypes produced by a given genotype, allowing more genetic variation to persist in the population." This is pretty much my nonprofessional understanding. But what does this imply about evolutionary theory?

      It seems to me that there is a seemingly overpowering tendency for the ordinary population to see genes as determining everything. They act more or less as material equivalents of a soul. This kind of thing justifies seeing a zygote as a human being. This view may not be claimed by professional scientists as a part of evolutionary theory (and may even be explicitly denied.)

      But it's like popular usage of "theory" as "opinion" instead of "scientific explanation." Popular usage of evolution tends to incorporate this kind of genetic determinism. It seems to me that practically every outrageous "revolution" against the Modern Synthesis that has come down the pike has
      been a revolution against this genetic determinism. Unfortunately, the fury unleashed by the counterrevolutionary scientists against this misunderstanding of evolutionary theory proper comes across as defense of genetic determinism.

      Worse, the many occasions on which professional scientists talk as if neutral evolution doesn't even exist, strongly suggests there is some unconscious equivocation between genetic determinism and evolutionary theory proper. I shouldn't want to dwell on the sore topic of evolutionary psychology. But the furious denunciations of every scientist who talks about this or that development that refutes genetic determinism as requiring a modification of evolutionary theory comes across as defense of genetic determinism as exemplified in things like evolutionary psychology.

      Things like epigenesis do not much affect evolutionary theory about speciation or the origin of novel adaptations. But these things do strongly affect notions about evolution as the inevitable unrolling of the effects of genes. There is even a strain in popular thought in which evolution in effect plays the role of a material force whose decrees may be "natural" but are even more compulsory.

      Delete
    3. @S Johnson
      This is more of a philosophical point you're making, but I think there is some truth to it. One of the proponents of evolution by epigenetic means, Mary Jane West-Eberhard, famously claimed that genes are followers, rather than leaders, in phenotypic evolution. That indeed seems to indicate a certain distaste for genetic determinism.

      Delete
    4. As I mentioned above -

      "Larry & PZ Myers have pointed out on many occasions, It has been long understood that a simplistic interpretation of the central dogma is a straw man. This is nothing new!"

      "One cannot deduce phenotype from gene sequence because the cell as an interacting chemical system is too complex."

      And to cite PZ Myers immediately above:

      "[epigenetics is] overhyped as an information transfer process on evolutionary timescales."

      Leading me to a citation of Nature Magazine's latest on epigenetics:

      http://www.nature.com/news/how-to-build-a-neanderthal-1.15063

      salient quote:

      "Epigenetic differences between humans and their ancient relatives may explain differences in physical traits, or phenotypes, such as the jutting brow ridge of Neanderthals. Yet various obstacles still hinder the study of ancient epigenomes, and some researchers are not yet sure if the approach will yield insights."

      uhmm... "some"?!

      that's my post - I await riposte

      Delete
  13. When it comes to explaining evolutionary changes by epigenetic changes, an important issue is the rate at which the epigenetic changes revert. Evidence so far is that it is in few enough cell generations that these changes revert in a few (organism-) generations.

    That implies that to keep an epigenetic change, you need natural selection -- a lot of it -- to kill off the individuals that have reverted the epigenetic change. You need selection coefficients of a size that is about the probability that the epigenetic change will revert. So if it reverts in, on average, 3 generations, you need selection coefficients favoring it of about 1/3.

    The alternative is to have substitutions in DNA that stabilize the epigenetic change so that it does not revert. And those substitutions, of course, are not epigenetic changes. There is already a name for such changes -- genetic assimilation. This term was invented by C.H. Waddington to denote genetic changes that take a response to an environmental perturbation and make it the normal phenotype. In our case the environmental perturbation is an "epigenetic" change. (By the way, the term "epigenetics" was also invented by Waddington).

    The discussion above by Corneel and Keith Elias raised the issue of genetic changes that stabilize an epigenetic change. Without such genetic changes, PZ's point is, if anything, understated. You need incredibly strong ongoing natural selection to keep epigenetic changes present. So without genetic assimilation, epigenetic changes can't account for evolution over even modest time scales.

    ReplyDelete
    Replies
    1. It seems the a predictable and recurrent environmental stressor could be enough to maintain epigenetic changes across generations, not sure what one would be, perhaps seasonal droughts?

      But one thing that intrigues me is how epigenetic effects can produce cohort effects in a population dynamic context. For example, the Dutch Hunger Winter produced a delayed generational effect that resulted in cohorts of adults that had lower fitness. Such cohort effects can either produce stabilizing or destabilizing effects on population size, which in turn can influence selection pressures.

      In my mind, population dynamic studies of cohort effects are very interesting, but there was never a good mechanism to explain what produced a cohort apart from sets of infants being born in a good versus bad year. Epigenetics promises to provide a new set of mechanisms (and even if epigenetic effects revert, the population dynamic consequences they produce would be interesting to track).

      Delete
    2. rich lawler says,

      For example, the Dutch Hunger Winter produced a delayed generational effect that resulted in cohorts of adults that had lower fitness.

      To me, that seems like an extraordinary claim. I'm going to reserve judgement until the result has been reproduced. I don't think we should assume that it has been proven that children inherit the effects of starvation on their parents.

      It's worth keeping in mind that not everything published in the scientific literature is correct.

      Delete
    3. Joe Felsenstein You need incredibly strong ongoing natural selection to keep epigenetic changes present.

      Any thoughts on this paper?

      Primate CpG islands are maintained by heterogeneous evolutionary regimes involving minimal selection.
      Cell. 2011 May 27;145(5):773-86. doi: 10.1016/j.cell.2011.04.024.

      Delete
    4. I'm no expert, I'm just going with what the numerous studies have reported. A summary paper can be found here:

      Roseboom et al. 2011. Hungry in the womb: What are the consequences? Lessons from the Dutch famine. Maturitas 70: 141– 145

      I don't have any reason to doubt the conclusions, though different studies varied in sample sizes from 100 to 700.

      Delete
    5. @caynazzo: Any thoughts on this paper?

      OK, I've gotten in over my head here. I do know that any methylation pattern (or histone modification pattern) that tends to be lost in a few generations needs either strong mutation to it, or strong natural selection maintaining it, to explain its long-term persistence.

      And if a lineage acquires such a pattern when its relatives don't have it, that argues for either it having acquired the strong mutation process or the strong selection against losing it.

      Delete
    6. The Cell paper summarizes two previous models for conserved methylation patterns and lays out a third hypothesis involving CpG island classification. Adrian Bird's early work suggests a conservation of unmethylated CpG islands (avoiding the hypermutability of methylated CpGs) that does not assume any function or purifying selection (neutral) (also if CpG islands are found in gene bodies then by dint of their location they are conserved). This explanation is likely incomplete since we know CpG islands are important in development, effecting gene expression in a cell-type specific way of hyper and hypo-methylation. That suggests strong selection against losing CpG islands comes in. The paper, whose modeling I can't really speak on, sees no constraint on tissue-specific methylated CpG islands vs global unmethylated counterparts.

      Delete
  14. @ Larry & Corneel

    Corneel brings up an interesting point about the "Thrifty Phenotype"

    advocatus diaboli ON:

    According to some, the thrifty phenotype hypothesis suggests that early-life metabolic adaptations help in survival of the organism by selecting an appropriate trajectory of growth in response to environmental cues. That means phenotypic change in response to the environment generate epigenetic changes that prepares a population of adult organisms for better survival in the long term and (this is the important bit) these changes are heritable from one generation to the next. In other words, Epigenetic's impact on Evolutionary theory could be very important.

    That idea is mind-boggling… in other words Novo-Lamarckism (not Lamarckism,not Neo-Lamarkism) would be back in vogue. (Or as Doolittle & Maderspacher phrased it: "New Lysenkoism")

    In Darwinian terms – isolated individual variants are not always grist enough for Natural Selection's mill, rather entire populations can present a "tentative" new phenotype (presenting greater opportunity for mutation the old fashioned way) in response to environmental change that can subsequently respond to Natural Selection by altering gene regulation for the long term. OOPs - so sorry - My German predilection for run-on sentences is showing again!

    ITMT

    (Yikes: Incoming! Baton down the hatches, Was that epigenetics as a putative mechanism for "group selection"?!)

    Explained another way: some populations have a thrifty phenotype as a default setting – other populations have a spendthrift phenotype as a default setting. Default settings can be switched on and off in response to the environment and these switches are heritable. Eventually populations can change their default settings in response to Natural Selection.

    advocatus diaboli OFF

    Is this really epigenetics? I dunno... even so, it seems to me any such phenomenon would be limited to a very small repertoire of genes restricted to growth and nutrition.

    ReplyDelete
  15. @ PZ Myers

    Sir - My apologies to you... Oh my gosh I mistook you for LARRY!

    In fact, my last response was directed to you and not to Larry.

    FTR - I am a big fan of yours!

    Best regards

    ReplyDelete
  16. @ Joe - we cross-posted again

    I was still typing when you made your observations.

    Does not my tentative "advocatus diaboli" in fact slip between the horns of your posed dilemma?

    ReplyDelete
  17. A Joe

    ... I am of course being contrary and contentious here.

    My apologies in advance

    ReplyDelete
  18. @Tom Mueller

    I am not sure I understand your suggestion, but it seems to rely on genetic assimilation. Right?

    ReplyDelete
  19. Eva Jablonka, while at times overstating what epigenetics is, makes a good point here [transcript from interview]
    "Well, I think that we have to reconsider it, and that is why I think evolutionary theory – evolutionary studies has been lagging behind all other domains in biology in terms of the impact that epigenetic research is having on them. This is changing. I think what we need and what need really badly, in addition, of course, to empirical studies, are good models.

    We need people who systematize the knowledge that we have at present, and the theoretical implications of variable transmissibility of epigenetic variation, of the inducibility of epigenetic variation, and the variable expressivity of epigenetic variations which can be dependent on the number of generations of exposure on how many generations a particular chromosome or part of a chromosome had undergone sequentially oogenesis, or spermatogenesis on the genetic background. All these things we have to understand to model, and once we will have good models we will also be able to do good experiments, both in the lab and in populations. Of course, this doesn’t mean that we should not do experiments now: we must have the ecological data, and this is being done. Ecological epigenetics is beginning to take off, especially in plants, but not only.

    So I think that what we need is a lot of population studies, and a lot of modeling, because although people are doing modeling – individual people are doing modeling and there are some very, very interesting models in the field already, it is not all put together, and certainly there is not one person who has taken it upon himself to systemize this theoretical framework, and we need this.


    I think a simple word search of bisulfite sequencing in methods sections of pop-gen papers would make her point.

    ReplyDelete
    Replies
    1. Why the martyr syndrome about people being unimpressed by epigentics? Joe Felsenstein and PZ Myers have laid out good reasons to be skeptical of this "revolution". And despite her caricature, Pop-gen'ers (even those writing the textbooks) are looking into this.

      ex.
      http://www.oeb.harvard.edu/faculty/hartl/lab/Evolutionary.html

      Delete
    2. Show me where anyone said "nobody is interested in population level epigenetics"?

      Delete
    3. I am having a difficult time taking your bisulfide comment any other way. Maybe you can explain what you meant.

      Delete
    4. Bisulfite conversion is a straightforward way to deaminate cytosines that aren't "protected" by a methyl group to determine DNA methylation profiles. It's a semi- facetious proxy for determining the research interests one field, say pop-gen, has compared with another field, say cancer biology.

      Delete
    5. Speaking of bisulfite sequencing and population genetics, the inventor of bisulfite sequencing is Marianne Frommer. She worked at the University of Sydney (until her retirement). A noted theoretical population geneticist there (until his retirement) was John Sved.

      I know them both, mostly through John. And that meant I got to meet Marianne too, because they both stayed in our house -- as they are a married couple. So I suspect each has heard about the other's field.

      Delete
    6. Interesting, I always assumed it was your colleague Charles Laird who invented bisulfide sequencing. He teaches a great upper level epigenetics course btw.

      Delete
    7. Charles has a 1992 paper with Scott Hansen and Stan Gartler which is one of the earliest using bisulfite sequencing, but Marianne's 1992 paper on it seems to precede theirs. I can't find the Hansen/Gartler/Laird 1992 paper online to see whether they cite Marianne or attribute the sequencing method to her. I do recall that Stan Gartler told me that Marianne invented the bisulfite sequencing method. Marianne and colleagues are credited in the Wikipedia page for bisulfite sequencing.

      Delete
    8. ... and here's an Acknowledgements section from a 2008 paper by Genereux et al. (the et al. includes Charles Laird):

      "We thank Audrey Fu, Stanley Gartler, Bruce Godfrey, Scott Hansen and Matthew Stephens for their many helpful suggestions; the developers of the bisulfite-treatment protocols evaluated here, especially Marianne Frommer, Susan Clark and Hikoya Hayatsu, for their encouragement and input over many years; and Ali Javed of GeneLink for providing detailed information on oligonucleotide synthesis and the purity of isobutyrl-methylcytosine stocks."

      Delete
    9. Thanks for that. Charles's UW bio, credits him with the invention of hairpin-bisulfite sequencing.

      Laird C.D., et al. Hairpin-bisulfite PCR: assessing epigenetic methylation patterns on complementary strands of individual DNA molecules.
      Proc. Natl Acad. Sci. USA, 101, 204–209.

      Delete
    10. Yes, in 2004. The original bisulfite method is Frommer in two papers in 1992. The 2004 method is a further development.

      Delete
  20. @ Joe

    Genetic Assimilation? Canalization?

    Confession #1: Remember I am a hopelessly out of date aging high school teacher attempting to keep his head above water.

    Confession #2: The advocatus diaboli rant above was what I actually taught my AP Bio class until I stumbled across Mark Ptashne’s letter in PNAS. That would also explain my gratitude to Larry Moran, PZ Myers and altruists such as you who bring me back up to speed.

    Let’s see if I can get this right: Genetic Assimilation is initially an organism’s response to some environmental stress and that response can eventually become genetically determined. So yes, I was in fact talking about “…substitutions in DNA that stabilize the epigenetic change so that it does not revert”. That is in fact what I meant by changing “default settings”.

    Canalized characters occur as underlying variable gene expression is masked when the organism expresses a common phenotype that only changes when some environmental stress breaks down canalization… or as I phrased it above , “temporarily switching off the default setting”.

    I guess the important bit is that the F1 (and maybe even many of the F2) generation must also get the message to temporarily “continue switching off the default setting” in anticipation of continued environmental stress.

    So without realizing it, my advocatus diaboli inadvertently did express epigenetics in the same terms as Waddington (unless I am helplessly confused and have got it all wrong).

    You mentioned “You need incredibly strong ongoing natural selection to keep epigenetic changes present. So without genetic assimilation, epigenetic changes can't account for evolution over even modest time scales.”

    But that was exactly my point – Metabolic responses in times of feast vs. times of famine should have extremely high selection coefficients.

    And it gets better, if such environmental stresses are transitory and frequent, the ideal scenario would be some reactive mechanism that permits transitory toggling back and forth between “default settings”. In other words, transitory epigenetic memory (that decays with a half-life of 1-3 generations) would be in itself an adaptation subject to natural selection. Meta-selection as it were.

    I confess, that was my naïve understanding of epigenetics as I attempted to wrap my head around the data presented in the documentary “Ghost in the Machine” (recommended resource for AP Biology) describing the Överkalix study.
    http://www.pbs.org/wgbh/nova/transcripts/3413_genes.html

    In the Överkalix study, Marcus Pembrey and colleagues observed that the paternal (but not maternal) grandsons of Swedish men who were exposed during preadolescence to famine in the 19th century were less likely to die of cardiovascular disease. If food was plentiful, then diabetes mortality in the grandchildren increased, suggesting that this was a transgenerational epigenetic inheritance. The opposite effect was observed for females—the paternal (but not maternal) granddaughters of women who experienced famine while in the womb (and therefore while their eggs were being formed) lived shorter lives on average.

    Transgenerational epigenetics is real and more than just a little intriguing and the Överkalix study supports the Dutch Hunger Winter study mentioned above. The devil is in the details! The confounding inheritance pattern has me mystified.

    Let’s remember that in fact the sperm do transfer quite a bit of cytoplasm to the zygote (including mitochondria which are later destroyed by autophagy). So, Mark Ptashne’s transcriptional factors can still come into play here.

    That said, Epigenetic inheritance patterns still have me stymied! Human Genetics seem to mirror equine and murine epigenetics as detailed above in a most peculiar fashion.

    I can sympathize with those who feel that epigentics still play an important (albeit diminished) role in evolution.

    ReplyDelete
  21. @ PZ Myers

    Before signing off, I revisited your post:

    "... Acquiring the process of semi-permanently modifying the cell state is something that was a key innovation (OK, many innovations) in evolution, but it's been overhyped as an information transfer process on evolutionary timescales."

    WOW - as they say in my part of the woods, "...even a blind pig can pick up an occasional acorn."

    By pure serendipity on my part, it would appear my most recent summary of epigenetics in AP Bio mirrors your more succinct assessment.

    BTW – your blog has long inspired many changes in my classroom.

    Best and grateful regards

    ReplyDelete
  22. ISTM (as someone who knows next to nothing about it!) that the methylation state 'unmethylated' has equal claim to being 'epigenetic'. There appear to be coarse systems of wiping and remethylating, with subtle overlays of resistance to and promotion of both systems, differentially in somatic and germ cell lines. Germ line cells take steps to avoid persistence of prior methylation states. Those that get through are predominately gender-specific, and relate to active mechanisms, under the scrutiny of 'conventional' natural selection. It's the control of methylation state, not the methylation state itself, that seems the likelier candidate for evolution.

    ReplyDelete
    Replies
    1. Hi Allan,

      Your thinking coincides with Ptashne's who maintains it is patently incorrect to
      “… refer to nucleosome modifiers as “epigenetic”—they, like the very many proteins recruited to genes by specific transcription regulators, are parts of a response, not a cause, and there is no convincing evidence they are self-perpetuating.” Mark Ptashne

      The Methylation status of CpG islands together with their peculiar propensity to mutation seems to me to represent an excellent candidate for Joe Felsenstein's "Canalization" and surprisingly rapid "Genetic Assimilation".

      The ubiquity of presumed “functional” CpG islands as upstream regulatory elements would also seem to indicate that they may play a significant role in evolution; yes, an information transfer process even on evolutionary timescales… maybe.
      (apologies to PZ Myers )

      I am not saying it is so… just saying I really need to rethink this again!

      Delete
  23. @caynazzo @ Joe Felsenstein

    Jotun Hein - Professor of Bioinformatics, Fellow of University College Oxfoed came up with an interesting talk on the subject of Non-genic evolution and selection in the human genome
    http://www.stats.ox.ac.uk/~hein/HumanGenome/hg.pdf

    CpG islands are very intriguing

    First CpG islands are very ubiquitous and second many seem to have functionality

    The problem is when Cytosine is methylated and subsequently deaminated. The resulting Thymidine transition represents an illegal paring of two legal nucleotides. Any repair would randomly substitute either the mismatched T or the G, resulting in rapid degeneration of the CpG island and loss of function.

    In neutrally evolving DNA – 10% of mutations occur in 1% of CpG Wow!

    If I read Hein correctly, it behooves a zygote to demethalate asap in order to sequester and protect Primordial Germ Cells from the enhanced mutational effects of methylation before differentiating into gametes and restoring methylation patterns.

    This all would seem to directly contradict the interpretation of the paper caynazzo cites. It would appear that some functional CpG is subject to remarkably strong selection!

    It gets better - 30% - 50% of disease causing mutations occur in CpG dinuceotides!!!

    This all reopens the question of the importance of “epigenetics” in evolutionary terms!

    I had some problems with the caynazzo’s paper – for example it did not seem to account of the CpG of Alu. But then, I admit I am in over my head. BTW, Hein also addresses Alu.

    Hein also addresses how much DNA is functional by aligning mouse and human sequences. Very Interesting!

    5% of the genome is functional while only 1.5% codes for protein leaving another 3.5% to be accounted for.

    In your face ENCODE!

    IMHO - That Hein talk is really worth perusing!

    ReplyDelete