Contribution
A mechanistic review of how epigenetic states could pass between generations, and of the reprogramming that stops them. It sets out what is well shown (plants, nematodes, some fruit-fly cases), what is rare (mammalian epialleles at transposons), and what is not shown (environmentally induced, adaptive inheritance in mammals and humans). It also fixes the definition the later debate uses: an effect is transgenerational only in a generation never exposed to the trigger.
Key insight
Epigenetic inheritance and germline reprogramming are “two sides of the same coin” (Introduction). Reprogramming exists so that each embryo starts from the genome, not from its parents’ lives. In mammals it happens twice, in the germline and after fertilization, and “these two rounds of epigenetic erasure leave little chance for inheritance of epigenetic marks, whether programmed, accidental or environmentally induced”. Where marks do escape, they are mostly at transposons, which the cell silences for its own defence. So inheritance, where it occurs, looks like a by-product of genome defence, not a channel for a parent’s experience.
Assumptions
- A conservative definition of epigenetics (Box 1): the perpetuation of gene expression and function across cell divisions without a change in DNA sequence. Waddington’s developmental sense and Bird’s 2007 “structural adaptation” sense are noted and set aside.
- Sequence first. A heritable difference counts as epigenetic only “when individuals that are truly genetically identical” show it (Epigenetic variation). Where the gene is unknown, a mutation can hardly be ruled out.
- Exposure defines generations (Box 1, Figure 1). Any generation whose cells were present during the exposure is excluded from the test.
Key results
- Definitions (Box 1, Figure 1).
- Intergenerational (parental) effects: the exposed parent, and any embryo or germ cells exposed with it.
- Transgenerational effects: “found in generations that were not exposed to the initial signal”.
- An exposed pregnant female exposes F1 (the fetus) and F2 (the fetus’s germ line). “Only F3 individuals can be considered as true trans-generational inheritance.”
- An exposed male exposes himself (F0) and his germ line (the future F1). F1 is intergenerational; only F2 and later count.
- Mammalian reprogramming (Germline reprogramming and imprinting; Box 2; Figure 2A). Two waves: in the zygote and early cleavage stages (imprints survive it) and in primordial germ cells (imprints are erased and the inactive X reactivated). Mouse genome-wide profiling found 4,730 loci escaping germline demethylation (more than 40% methylated), mostly repeat elements, and 233 single-copy loci, which the authors call “prime candidates” for transgenerational inheritance in mammals.
- Plants reprogram less (Box 2; Figure 2B). More than 80% of methylated cytosines are kept in pollen sperm cells; only asymmetric CHH methylation is reset. Epialleles can persist “for literally hundreds of years”. In Arabidopsis epiRILs, hypomethylated segments were inherited for eight or more generations and account for up to 90% of the heritability of flowering time and root length.
- Worms and flies. Small-RNA silencing in C. elegans lasts more than 20 generations, and an olfactory imprint more than 40. A piRNA-based trans-silencing effect in Drosophila lasts more than 50.
- Mammalian cases are few and short.
- Agouti viable yellow and Axin fused: bona fide transgenerational epialleles, both at retrotransposon (IAP) insertions; Agouti is inherited through the female germ line only.
- A diet of methyl donors changes Agouti offspring coat colour, but “this effect is only transmitted over two generations and is lost by the third”.
- The Kit paramutation-like effect disappeared by F4.
- Human familial epimutations (MLH1, DAPK1) turned out to depend on DNA sequence polymorphisms: the silencing is re-established each generation and erased in the germ line.
- Rodent exposure studies (Epigenetic variation; Implications for
human health; Box 3).
- Vinclozolin’s multigenerational effect on male fertility was seen in outbred rats, not in an inbred strain, so genetic variation may explain it.
- Paternal diet effects in mice: transmitted, probably via sperm, by an unknown agent. No global change in sperm methylation was found, and most paternal RNA and histones are replaced after fertilization.
- Paternal transmission of social-defeat phenotypes was traced to maternal provisioning, not epigenetic inheritance (Dietz et al.).
- Human evidence (Implications for human health). Discussed only for
nutrition.
- Dutch famine: those exposed in gestation had poorer glucose tolerance; offspring of prenatally undernourished fathers, not mothers, were heavier. “No evidence of transgenerational effects of grandmaternal under-nutrition during gestation” was found.
- Most such studies are “inter-generational (parental or grand-parental exposure) rather than truly trans-generational”, and cannot rule out postnatal nutrition or missing covariates.
- Bona fide transgenerational inheritance over several generations “seems unlikely given the robust reprogramming found in the mammalian germ line”.
- Not discussed. Överkalix, Holocaust offspring and FKBP5 are not mentioned. Stress and behaviour appear only in Box 3, as a “very seductive but highly controversial topic”.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Only generations never exposed to the trigger can show transgenerational inheritance: F3 after an exposed pregnant female, F2 after an exposed male | strong as a definition; it follows from which cells were present at the exposure | Box 1, Figure 1 |
| C2 | Transgenerational epigenetic inheritance is well documented and relatively common in plants and nematodes, mostly at transposons, transgenes and viruses | strong: many independent systems and genetic screens | RNAi, paramutation, histone and DNA-methylation sections |
| C3 | Mammals reset epigenetic marks twice, in the germline and the early embryo, which leaves little chance for inheritance | strong for DNA methylation (genome-wide mouse profiling); weaker for histones and RNA, whose fate the authors call unclear | Germline reprogramming; Box 2; Figure 2A |
| C4 | Mammalian epialleles exist but are extremely rare and are found at retrotransposons | moderate: a handful of loci; argued by absence elsewhere | transposon section; Conclusions |
| C5 | Environmentally induced epigenetic changes are rarely inherited across generations and still more rarely adaptive, even in plants | moderate: failed attempts to show adaptive stress tolerance; vernalization reset each generation | Epigenetic variation section |
| C6 | Human evidence for transgenerational epigenetic inheritance is little: the studies are intergenerational and confounded | moderate as an assessment of the nutritional literature only; other human claims are not examined | Implications for human health; Conclusions |
Concepts
- intergenerational (parental) effect: an effect on the exposed parent or on offspring whose cells were exposed with it, in utero or as germ cells.
- transgenerational effect: an effect in a generation never exposed to the trigger.
- germline reprogramming: erasure and resetting of DNA methylation, histone marks and small RNA in germ cells, and again after fertilization, so that the embryo is totipotent.
- epiallele: a heritable variant in a gene’s epigenetic state without a change in sequence.
- paramutation: one allele heritably silences another in trans, via small RNA.
Connections
- The other two sceptics. Horsthemke (LIT-1762) takes the F2/F3 rule from here and adds the secondary epimutation. Bird (LIT-1708) cites this review for directionless plant epialleles. Bird is thanked here for comments on the manuscript; the three are not independent.
- The mouse fear study (Dias and Ressler, LIT-1746) appeared too late to be discussed; Bohacek and Mansuy (LIT-1738) is the record’s review on the other side.
- Inherited state in bacteria (THEORY-179). E. coli cells pass a physiological state to descendants for about four generations, then lose it. Box 3 lists exactly this kind of route, a self-sustaining feedback loop or a metabolite, as epigenetic inheritance without chromatin, and doubts it could last through development in complex organisms. Bacteria have no germline to reset; mammals do. The two together are a contrast, not a parallel: the same short, decaying persistence (Agouti diet: two generations; Kit: gone by F4) is what survives in mammals.
- Cultural persistence (THEORY-889). The authors’ last word, that human culture is Lamarckian, is the route THEORY-889 documents for attitudes.
Bearing on the record
- The question (QUESTION-297). On the germline route this review is a strong negative for mammals in general and a weak one for humans in particular. It examines only nutritional human studies, and finds them intergenerational. It never addresses trauma in humans.
- A consequence for the question’s framing (an inference). By Figure 1’s rule, a parent traumatized before conception exposes only their own germ cells, so their children are F1, intergenerational. The grandchildren are the first generation that could show germline transmission. Any study of survivors’ children, whatever it finds, can test only the intergenerational effect. The question’s “germline” route should be split: a parent’s exposed gametes (intergenerational, not excluded by this review) and inheritance through untouched generations (what reprogramming makes unlikely).
- The owner’s thesis (CLAIM-491). Indirect. If marks are reset each generation, what reaches grandchildren who never met the trauma must travel by care, behaviour or story. That leaves more of “intergenerational trauma”, past the first generation, to the social routes, and the identity route is one of them. The review does not choose among those routes, and it treats culture as transmitted teaching and habit, not as identity. No note on the claim is proposed.
- Proposed dated note on QUESTION-297 (below, section 3).
- No THEORY filed.
- No ML instruction.
Limitations
- Humans get one section, on nutrition. Stress, trauma and behaviour are confined to a paragraph in Box 3.
- The Dutch famine passage is unclear. It first reports “increased neonatal adiposity among the grandchildren of women who had been undernourished during pregnancy”, then says no transgenerational effect of grandmaternal undernutrition was found. By the review’s own Figure 1, those grandchildren are F2 of an exposed pregnant woman, so any effect in them would be intergenerational anyway: the passage uses “transgenerational” in the loose sense Box 1 warns against.
- The mammalian negative is argued from mechanism and from absence. The 233 single-copy loci that escape reprogramming are left as candidates.
- Figures not seen. Only their legends are in the copy.
- Typos in the manuscript.
- “penotype” (Paramutation section); “bone fide” for “bona fide” (twice); “specfically”; “extrambryonic”; “quantititative”; “inlcuding”; “sterotypical”; “Single nucleotie polymorphisms”; “descendents” in the abstract.
- “IAPTR1” for what is probably IAPLTR1.
- “This mechanism highly reminiscent” lacks “is”.
- “No evidence … were found”.
- The same NIH grant number, GM067014, is listed twice.
- The text names SET-24 and SET-25 for C. elegans; the Figure 4 legend names SET-25 and SET-32.
Open questions
- Do any of the 233 single-copy loci that escape germline demethylation carry an environmentally induced state to F2 or F3?
- Does the F2/F3 rule hold for small RNA in sperm, whose fate after fertilization the review calls unclear?
- Human trauma is not examined here. Do the human studies the question cites (LIT-1764, LIT-1692) meet even the intergenerational standard, or only the loose one?