The June issue of Cell has 3 articles on this subject, and an overview commentary about those papers by von Mayenn and Reik, titled "Forget the Parents." Experiments on mice and data on humans show that, first of all, a new embryo quickly isolates its future germline (sperm or egg) cell lineage as primordial germ cells (PGCs) from the rest of its cells. Those germline cells divide and differentiate into ovary and testis tissue, to be used as gametes later in life. During this time the differentiation of the cells, and just their maintenance, require interaction with other cells (such as those that supply blood and hence nutrition), so they aren't entirely isolated, but from a genetic point of view, that is, the stream of developmental cell division and hence genetic inheritance, they are separate lineages of cells.
The papers show that by and large the epigenetic marking, that is, the instructions for which genes to use or to silence, in the germline cells is basically stripped away early in their lineage formation. The phrase they use is 'global erasure.' Early on, the authors of one reviewed paper show that this marking is 'progressively erased genome wide...to the lowest levels...observed in the human genome to date." Another paper shows that a similar pattern occurs in mice, so that the reprogramming of the germline is highly conserved in mammals. While not 100%, the evidence shows the "erasure of epigenetic memory is a key purpose" of this stripping away of marking. Forget your parents, indeed!
This can be seen, in principle at least, as a great thing! It means that we are not hide-bound by our parents' experience in a Lamarckian way. It means that expression effects that accumulated during their whole lives before they made us do not force us into any particular path, except that basically enabled by our genes proper. That's good, because our genes are the result of 3+ billion years of evolution and there's a reason that they are as they are, a reason one would not want to trifle with based our parents' particular experiences. We each get a fresh start, by forgetting our parents, which lets evolution screen our genomes for fitness to current situations, that is, enables our patrimony to evolve. Or do we each start life fresh?
These regions are reported as including segments associated with some brain and growth-related physiological function and one paper suggests they are also associated with obesity, schizophrenia, and multiple sclerosis. What are we to make of such statements and findings? My first temptation is to say that this is cherry-picking based on known associations, many of which have proven to be minor or fickle, ignoring other things that are simply not known but could be important. The size of genomes is such that any substantial subset will likely include just by chance what may in retrospect seem to be of particular importance. They may be, but with limited knowledge such appearance of major findings may be inevitable, unless we resist the temptation to tell Just-So adaptive stories.
Furthermore, there are reports that much or even most of the genome is transcribed into various types of non-coding RNA (reports from the ENCODE project, for example), so the conservation/non-conservation distinction becomes rather unclear: it is taken as remarkable that some fraction of 'real' genes are not erased but the bulk of the genome that many are arguing does have function is erased. And Tang et al report that these escapee regions mainly became re-marked later in development...and might contribute to transgenerational epigenetic inheritance. Why? One can speculate this is because they can be marked, during gestation, as result of the mother's environmental exposures.
Making things even more curious, most of these regions were found not to have sequence conservation. Since sequence conservation is widely taken as the criterion for evolutionary importance, the Just-So story temptation must be tempered. In essence, this can be seen as correctly reporting data, but trying too hard to give importance explanations to them.
Obvious questions remain, as to the mechanisms by which various levels and patterns of developmental erasure are managed and how they evolve (these issues were discussed with substantial findings, in the Cell papers, but are beside our point here).
Playing word games?
In a sense the findings seem really to say "Forget your parents....except when you remember them!" It is a post hoc explanation offered, in a sense, to rescue a theory. But if a theory says something's true except when it isn't, that isn't much of a theory: it's a tautology. It may be true, but what it means is that there is a mix of effects that we don't yet understand.
The enthusiasm for this phenomenon, in which in utero experiences based on the mother's state affect the later lives of exposed fetuses has generated lots of excitement, claims for paradigm shifts and the like. This is how our society works. But as noted by Heard and Martienssen, such effects have not yet convincingly been shown to be truly transgenerational, that is, to affect unexposed descendant generations. So, the epigenetic-claims temperature probably needs to be cooled--which is not how our hyperbolic society treats things these days.
No one can be faulted from interest in a subject or from our inadequate knowledge, since the unkown is what science is about. However, what we need is some formal, regular understanding of what 'except' means, that is, of when and where in genomes we should find marking conserved and when/where stripped or re-set. Clearly something more specific in the way of knowledge is needed. This is not to denigrate the science, because we can only know what we've learned so far. But it shows our general unwillingness to consider our imperfect knowledge, and the temptation to overstate what we do know. In fact, it would be important enough just to say that epigenetic inheritance seems possible and can be important when it occurs.
Whatever's here isn't Lamarckian
In this case, the caveat offered by the Tang paper's authors that the escapee regions "may be sensitive to environmentally induced variations in methylation [marking] in individual embryo [sic] that could persist over a short term or might even become heritable. As such, they are potential candidates for transgenerational epigenetic inheritance." This is unexceptionable, but again it is the same tautology, phrased to hint at something remarkable.
The idea is often raised that such transgenerational phenomena, affected by experience, would overthrow Darwin, cause a 'paradigm shift' and be truly remarkable if not dramatic. The phenomenon is interesting, and important to the extent that it happens, and would change some aspects of evolution from our usual idea of it, there is nothing all that new and nothing at all 'Lamarckian' about it. So to whatever extent marking patterns are inherited and a response to experience, they don't challenge evolutionary theory in that way.
For example, an important fact about evolutionary generalization that probably only a small fraction of biologists expounding on the heresy of Lamarckism or proclaiming a conceptual revolution, is that plants do regularly have transmission of acquired traits. Germ lines in many plants develop from the meristem cells, at the ends of the same individual plant's branches. Whatever environmentally based genome marking, or mutations and their effects, have occurred during the life of a given branch will be transmitted to its gametes. What happens in the other meristems of the plant may or may not be similar as they are independent cell lineages branching--literally!--from the original seed.
But this, while important and different from how we usually think of inheritance is not all that different from our own condition. Germline lineages divide into separate egg or sperm cells, and mutations can occur in one but not the others of these cells in an individual. Presumably, marking can do that as well.
A core epigenetic question is how permanent are these settings and how are they set and removed specifically. Hard-to-change settings, like hard to mutate genetic variants, could be part of evolution that would modify our theory in what may seem like a kind of Lamarckian way, but in fact they are still 'Darwinian' in that they are set by environmental conditions imposed on the individual, that is, the set-determining conditions are external rather than internal.
Right now, enough is unknown that we cannot take seriously any claim that this is a fundamentally different aspect of inheritance that will modify evolutionary theory except, at most, in some details. But such modification is always happening to any scientific theory, and is healthy for science. Epigenetic inheritance would join DNA-sequence inheritance as the mechanism of evolutionary change. We'd just have to welcome it and wonder at why we didn't think it happened. But there would be little if any real 'paradigm shift'. Future work will show what's true.
Indeed, if we think about this carefully, we might not so easily dismiss things, sneeringly, as being 'Lamarckian'. Tomorrow, we'll examine a bit more closely what it was that Lamarck actually said, and to what extent he deserves the ridicule that is so consistently directed at him.