NOTE-1265: A critical view on transgenerational epigenetic inheritance in humans

Contribution

A short sceptical statement by a human geneticist, aimed at human studies. Its distinctive contributions are two. It names the commonest false positive in families, the secondary epimutation. And it gives a three-step roadmap for proving germline epigenetic inheritance, then concedes that in humans the first and last steps cannot be completed.

Key insight

Offspring inherit from their parents genes, an environment and a culture, and the environment and culture were partly built by earlier generations (Fig. 1a). Shared epigenetic marks between parent and child are therefore expected without any germline transmission: “it is extremely difficult to divide out whether these features have been transmitted through the germline or established anew in each generation by the action of shared genes and shared environments” (Secondary epimutations). Of the three rivals, genes can be excluded in humans; environment and culture cannot.

Assumptions

• Epigenetics is cellular inheritance of chromatin and expression states, by feedback loops, chromatin marks and long-lived non-coding RNA. Transmission through the germline is an extension that must be proved, not a part of the definition.
• The Weismann barrier separates mammalian germ cells from somatic ones, so an acquired somatic mark would have to appear in germ cells as well.
• Proof means a causal factor in germ cells, not a correlation between an ancestor’s exposure and a descendant’s outcome.

Key results

• The barriers (Defining transgenerational epigenetic inheritance). The Weismann barrier, and the reshaping of chromatin in germ-cell differentiation and again after fertilization. Some loci escape germline reprogramming. Long-lived RNA is “less affected by these barriers” and so a likelier carrier, though its mechanisms are “largely unsolved”.
• Rodent evidence (Evidence). Reports in mice and rats over ten years involve endocrine disruptors, high-fat diet, obesity, diabetes, undernourishment and trauma. Two are described:
  • prediabetes in male mice altered sperm methylation and the islets of F1 and F2 offspring (Wei et al.);
  • early-life trauma in male mice altered offspring behaviour and metabolism, and injecting their sperm RNA into normal fertilized eggs reproduced the changes (Gapp et al.). Most such studies are “technically sound” but “still await independent confirmation”, and the endocrine-disruptor and high-fat-diet methylome results have been challenged.
• Human evidence (Evidence).
  • Epidemiological studies link grandparental food supply to grandchildren’s health. The one cited is Kaati, Bygren, Pembrey and Sjöström 2007. The text does not name Överkalix; that the paper uses the Överkalix records is known from outside this Comment.
  • An indirect study suggests sporadic Prader–Willi imprinting defects come from a grandmaternal imprint passed through the male germline.
  • Both rest on unique cohorts and await replication.
• The Dutch Hunger Winter (Fetal programming and intergenerational inheritance). Undernourishment reached pregnant women, their fetuses and the fetuses’ germ cells. The F1 adults’ cardiovascular and metabolic disease “is not due to the transmission of epigenetic information through the maternal germline, but a direct consequence of the exposure in utero”: fetal programming, or, if F2 is affected, “intergenerational inheritance”.
• Secondary epimutations (Secondary epimutations; Fig. 1b). In several families an abnormal promoter methylation co-segregates with a rare disease over generations. In each, a mutation in the neighbouring gene removed a transcription termination signal, and read-through transcription methylated and silenced the gene. If the neighbouring gene is expressed in the germline, the epimutation is in germ cells too. That is genetic inheritance. A true case needs a primary epimutation, independent of sequence, in germ cells.
• The roadmap (Roadmap).
  1. Rule out genetic, ecological and cultural inheritance. In animals: inbred strains, controlled environments; F3 and later after a pregnant female is exposed, F2 after a male; better still IVF, embryo transfer and foster mothers. In humans “it is impossible to rule out ecological and cultural inheritance … but genetic effects should and can be excluded”: suspect a Mendelian pattern, check the haplotype, sequence the whole genome. Even then “you still cannot be 100% sure that a genetic variant does not exist”.
  2. Identify the factor in germ cells, purified (swim-up, micromanipulation) and tested for somatic contamination. Female germ cells are “scarse or unavailable”.
  3. Show the factor causes the effect: remove it and lose the effect, add it to control germ cells and gain it. Done for sperm RNA in mice; DNA methylation and histone marks cannot easily be manipulated; and “all of these experiments can hardly be done in humans”. This “might currently be too much to ask for”.
• Evolution (In the light of evolution). Almost all mouse models and human observations concern deleterious traits; the one exception cited is hepatic wound healing. Reprogramming may serve to block the transmission of a parent’s environmental insults. For long-lived humans a prediction of the offspring’s environment is likely to be wrong.
• Conclusion. Even if mechanisms exist, cultural transmission “by communication, imitation, teaching and learning surpasses the effects of epigenetic inheritance and our ability to detect this phenomenon”. “For now, I remain skeptical.”
• Not discussed. Holocaust offspring, FKBP5 and the Dias and Ressler odour study are not mentioned. Trauma appears only in the rodent list and in Gapp et al.

Claims

id	claim	strength	support
C1	Human studies of germline epigenetic inheritance cannot exclude ecological and cultural inheritance, and can only partly exclude genetic	strong as an argument from design; observational human data cannot separate shared environment and culture from germline	Key insight; Roadmap step 1
C2	The Dutch Hunger Winter’s effects on F1 are fetal programming, not germline transmission	strong as a classification: the fetus was itself exposed	Fetal programming section
C3	Most familial epimutations are secondary to a nearby genetic variant and mimic epigenetic inheritance	moderate: three cited family studies and Guéant et al.; “most” is not quantified	Secondary epimutations; Fig. 1b
C4	Proof requires excluding other inheritances, locating the factor in pure germ cells, and showing by removal and addition that it causes the effect	a proposed standard, not a finding	Roadmap
C5	Reported transgenerational effects in mammals are almost all deleterious, which argues against an adaptive role	weak to moderate: reporting bias, which the author grants, could produce the pattern	In the light of evolution
C6	Cultural transmission very likely outweighs epigenetic inheritance in humans	an opinion, stated as such; no evidence is given for the comparison	Conclusion

Concepts

• intergenerational inheritance: here, an effect reaching F2 through fetal germ cells exposed in utero; distinct from fetal programming of F1.
• transgenerational epigenetic inheritance: transmission of epigenetic information through the germline to unexposed generations.
• primary epimutation: an abnormal epigenetic state independent of any DNA sequence change.
• secondary epimutation: an abnormal epigenetic state caused by a sequence change, here read-through transcription from a mutated neighbouring gene; it is re-created each generation and so passes as genetic inheritance.
• Weismann barrier: the separation of germ cells from soma.
• niche construction: the author’s suggestion of how culture, if it affects the epigenome at all, would do so indirectly.

Connections

• Heard and Martienssen (LIT-1745) is cited for the rodent literature; the F2/F3 rule is theirs. Bird (LIT-1708) cites this Comment as “a recent critical review”. Horsthemke and Bird later co-authored a paper on mutation-induced CpG island methylation (2023, in Bird’s references), so the sceptics are not independent voices.
• The Överkalix study (Pembrey et al., LIT-1764) is in the record; the Comment cites its companion paper (Kaati et al. 2007), which the record does not hold.
• The womb (Heijmans et al., LIT-1751). The record already reads the Dutch Hunger Winter as intergenerational; this Comment agrees and says why.
• Lehrner and Yehuda (LIT-1793), who join group narrative to the biology, meet Fig. 1a’s statement that “at present there is no evidence for a direct effect of culture on the epigenome”.
• Inherited state in bacteria (THEORY-179). A single-celled organism has no Weismann barrier; whatever state a mother cell holds, its daughters receive. The barrier is what makes the mammalian case hard.

Bearing on the record

• The question (QUESTION-297). The Comment does two things for it.
  • It answers the question’s “What would count as an answer” on routes, in the negative for humans. The question asks for a human study that separates germline from uterine, parenting and cultural routes, with fathers’ children raised apart or adoptees. Horsthemke’s roadmap says culture and ecology cannot be excluded in humans at all, and his causal step cannot be done in humans. Adoption and paternal-only designs would weaken the confounds, not remove them. The question’s “or evidence that no such separation is achievable” is the answer he gives, for now.
  • It turns the three human claims into specific failures: Överkalix (unique cohort, unreplicated); the Dutch Hunger Winter (fetal programming, so not a germline case); familial epimutations (genetic).
• The owner’s thesis (CLAIM-491). Indirect, and an inference. The author’s last word is that culture, transmitted by communication, imitation, teaching and learning, very likely outweighs epigenetic inheritance. That places the bulk of what passes between human generations on the social side, where the identity route lies. But his “culture” is learned behaviour, not identification with ancestors; the Comment gives no support to the identity route over the parenting route. No note on the claim is proposed.
• Proposed dated note on QUESTION-297 (below, section 3).
• No THEORY filed.
• No ML instruction.

Limitations

• A Comment, not a review. Two rodent studies and two human ones are described; the rest are referenced in bulk.
• The cultural conclusion is unsupported here. No evidence is offered that cultural transmission is larger; it is the author’s opinion, labelled as such.
• The deleterious-bias argument is admitted to be open to reporting bias.
• Gapp et al.’s sperm-RNA result is described without criticism, and it is the closest the Comment comes to a trauma mechanism; it is not weighed against the roadmap.
• Typos. “scarse” for “scarce” (Roadmap); a stray space in “germ cells ,” (Secondary epimutations).

Open questions

• If culture and ecology cannot be excluded in humans, what result could move the human question at all? Perhaps only a mechanism shown in animals and then found, molecularly, in human germ cells.
• Are any trauma-related human methylation findings, such as FKBP5 in Holocaust offspring (LIT-1692), secondary to genotype? Step 1’s haplotype test is the one step that can be done in humans.
• How much of what descendants carry is the “culture” this Comment means, learned habits and teaching, and how much is identification with the ancestors? The Comment does not separate the two.