Literature review · methylation series

Can a methyl group carry a memory across generations?

A guided tour of what the human and animal literature actually reports about DNA methylation and inherited trauma — with the original papers, short quotes, and the caveat that belongs next to each one.

The honest version: the animal work is strong and controlled; the human work is suggestive, small, and hard to disentangle from everything else parents pass to children. This page keeps those two apart on purpose.

01 · The distinction everything hinges on

Intergenerational is not transgenerational

Almost every argument in this field is really an argument about which generation you are looking at. If a generation was physically present during the exposure — as a fetus, or as a germ cell inside that fetus — then finding a mark in it is not evidence of inheritance. It is evidence of exposure.

F1 / F2 · Intergenerational

The exposed individual and any generation that was directly present during the exposure. Real, measurable, biologically important — but it does not require anything to be transmitted through the germline.

F2 / F3 · Transgenerational

The first generation with no direct exposure at any point, including as a germ cell. This is the only generation that can demonstrate inheritance through the germline — and it is the generation most human studies never reach.

The arithmetic is concrete. If a pregnant woman (F0) lives through a famine or a genocide, her fetus (F1) is already alive — and so are that fetus's own developing germ cells, which will become F2. Three generations are physically present for one event. Only F3 is conceived from material that was never there. On the paternal side the chain is shorter: a man's sperm is exposed, so his child F1 is exposed, and F2 is the first clean generation.

02 · Counting the generations

Which generation is the first clean one?

The answer differs depending on which parent was exposed, because germ cells are already present in a fetus. Pregnant exposure reaches three generations at once. Paternal preconception exposure reaches two.

Hatched — directly exposed Solid + ✓ — first unexposed generation Dashed — confirmatory, not required

03 · Human cohorts

What has been measured in people

Human studies are observational, the cohorts are small, and the tissue is almost always blood rather than the tissue anyone cares about. They are still the reason the field exists — read each finding with its caveat attached.

Quotes are short excerpts in the original English, linked to the source. Everything else is a paraphrase.

◆ Human F1 · in utero IGF2 DMR

Persistent epigenetic differences associated with prenatal exposure to famine in humans

Heijmans, Tobi, Stein, Putter, Blauw, Susser, Slagboom & Lumey · PNAS 2008 · Dutch Hunger Winter · 60 sibships

“less DNA methylation of the imprinted IGF2 gene”— Heijmans et al., PNAS 2008

FoundPeople conceived during the 1944–45 Dutch famine still differed in IGF2 methylation from their own unexposed same-sex siblings six decades later. The effect showed up only for periconceptional exposure, not for exposure late in gestation.

CaveatThis is a mark in the directly exposed generation. It shows an environmental exposure can leave a durable methylation difference in a human — it says nothing about that difference being inherited by anyone.

pnas.org · 10.1073/pnas.0806560105 ↗

◆ Human F1 · adult offspring FKBP5 intron 7

Holocaust exposure induced intergenerational effects on FKBP5 methylation

Yehuda, Daskalakis, Bierer, Bader, Klengel, Holsboer & Binder · Biological Psychiatry 2016 · 32 survivors, 22 offspring, 8 + 9 controls

“Holocaust exposure had an effect on FKBP5 methylation”— Yehuda et al., Biol Psychiatry 2016

FoundThe first study to test directly in humans whether a parent's methylation response to trauma reappears in their adult children at the same genomic site. At one CpG in FKBP5 intron 7 — the glucocorticoid-response region already tied to PTSD risk — survivors were more methylated than controls while their adult children were less methylated than control offspring. The offspring effect survived controlling for the children's own FKBP5 genotype and their own trauma history, so it is not simply inherited sequence or a hard life of their own.

CaveatThe authors flag it themselves: at n = 22 offspring you cannot separate a mark carried in the gametes from an in utero effect, or from simply growing up with a parent who had PTSD. One CpG site, in blood. The opposite-direction result has no agreed mechanism. A hypothesis generator, not a demonstration.

sciencedirect · 10.1016/j.biopsych.2015.08.005 ↗

◆ Human F1 · adult offspring NR3C1 exon 1F

Influences of maternal and paternal PTSD on epigenetic regulation of the glucocorticoid receptor gene in Holocaust survivor offspring

Yehuda, Daskalakis, Lehrner, Desarnaud, Bader, Makotkine, Flory, Bierer & Meaney · American Journal of Psychiatry 2014 · 80 Holocaust offspring vs 15 comparison subjects

“offspring with paternal PTSD showed higher GR-1F promoter methylation”— Yehuda et al., Am J Psychiatry 2014

FoundWhich parent had PTSD changed the direction of the effect. With paternal PTSD alone, offspring had higher methylation at the NR3C1 exon 1F promoter; when both parents had PTSD, methylation was lower. Lower methylation went with stronger cortisol suppression after dexamethasone.

CaveatParental PTSD also changes how a child is raised. No observational design can separate a germline mark from a childhood spent with a traumatised parent.

psychiatryonline · 10.1176/appi.ajp.2014.13121571 ↗

◆ Human F1 · in utero NR3C1 exon 1F

The Tutsi genocide and transgenerational transmission of maternal stress: epigenetics and biology of the HPA axis

Perroud, Rutembesa, Paoloni-Giacobino, Mutabaruka, Mutesa, Stenz, Malafosse & Karege · World J Biol Psychiatry 2014 · 25 exposed mothers + children vs 25 unexposed pairs

“higher methylation of the NR3C1 exon 1F than non-exposed groups”— Perroud et al., World J Biol Psychiatry 2014

FoundWomen pregnant during the 1994 genocide were compared with Rwandan women of the same ethnicity pregnant in the same period but not exposed. Exposed mothers and their children both showed higher PTSD and depression scores, lower cortisol and glucocorticoid receptor levels, higher mineralocorticoid receptor levels, and higher NR3C1 exon 1F methylation — the same HPA-axis signature appearing in both generations.

CaveatThe children were in utero during the genocide, so this is F1 programming, not inheritance — it cannot separate a mark transmitted through the germline from one induced directly by maternal cortisol crossing the placenta. n = 25 per group, cross-sectional, and the exposed families also lived through very different postnatal conditions.

tandfonline · 10.3109/15622975.2013.866693 ↗

▲ Review Open access

Intergenerational transmission of trauma effects: putative role of epigenetic mechanisms

Yehuda & Lehrner · World Psychiatry 2018 · 17(3):243–257

“The most compelling work to date has been done in animal models”— Yehuda & Lehrner, World Psychiatry 2018

FoundNotably more hedged than the primary papers that made this author's name in the area. The review separates two mechanisms that popular coverage conflates — developmental programming through maternal care and the in utero environment, versus preconception changes in the parental germline — and walks through the candidate windows: oocytes, which sit in a demethylated and vulnerable state from birth until puberty; sperm marks; and in utero programming via maternal cortisol.

CaveatIt is also where the field's central limitation gets said plainly: parental sex, PTSD status and the offspring's developmental timing all shift the direction and size of the methylation difference between studies, in ways that do not yet resolve into one mechanistic story. A position paper as well as a survey — and still the best single entry point.

PMC6127768 · full text ↗

04 · Animal models

What has been measured in mice and rats

Here you can control the genome, the environment, and who raises whom — and you can go one step further and move the candidate molecule into a fresh embryo. This is where the mechanistic claims actually come from.

● Rat Developmental, not germline GR exon 17 promoter

Epigenetic programming by maternal behavior

Weaver, Cervoni, Champagne, D'Alessio, Sharma, Seckl, Dymov, Szyf & Meaney · Nature Neuroscience 2004

“differences emerged over the first week of life, were reversed with cross-fostering”— Weaver et al., Nat Neurosci 2004

FoundPups of high-licking, high-grooming mothers ended up with a demethylated CpG in the NGFI-A binding site of the hippocampal glucocorticoid receptor promoter, more GR expression, and a calmer stress axis in adulthood. Infusing an HDAC inhibitor erased the group difference — methylation, transcription factor binding, GR levels and stress response all moved together.

CaveatCross-fostering reversed it, which is the point: this is behaviour writing methylation in the pup's own brain each generation, not a mark travelling through sperm or egg. It is the cleanest evidence that experience becomes methylation — and the clearest example of transmission without inheritance.

nature.com · 10.1038/nn1276 ↗

● Mouse F1 sperm → F2 brain MeCP2 · CB1 · CRFR2

Epigenetic transmission of the impact of early stress across generations

Franklin, Russig, Weiss, Gräff, Linder, Michalon, Vizi & Mansuy · Biological Psychiatry 2010 · unpredictable maternal separation (MSUS), P1–P14

“DNA methylation is altered in both directions and in a gene-specific manner”— Franklin et al., Biol Psychiatry 2010

FoundEarly separation stress produced depressive-like behaviour that persisted into the next generations. In F1 sperm, methylation went up at MeCP2 and CB1 and down at CRFR2 — and the MeCP2 and CRFR2 changes reappeared in the brains of F2 females, alongside reduced expression of those genes.

CaveatA matching mark in sperm and in the next brain is a correlation across tissues, not proof the sperm mark caused the brain phenotype. Group sizes are small (often n = 3–6 for methylation), and the effects were largely female-specific in F2.

sciencedirect · 10.1016/j.biopsych.2010.05.036 ↗

● Mouse Causal test Sperm sncRNA

Implication of sperm RNAs in transgenerational inheritance of the effects of early trauma in mice

Gapp, Jawaid, Sarkies, Bohacek, Pelczar, Prados, Farinelli, Miska & Mansuy · Nature Neuroscience 2014

“Injection of sperm RNAs from traumatized males into fertilized wild-type oocytes”— Gapp et al., Nat Neurosci 2014

FoundSame MSUS trauma paradigm, but this time the candidate carrier was purified and moved: RNA from the sperm of stressed males, injected into normal fertilised oocytes, reproduced the behavioural and metabolic phenotype in animals with no stressed ancestor at all.

CaveatThis is the strongest causal design in the whole field — and the carrier it identifies is small non-coding RNA, not DNA methylation. Worth sitting with: the best-evidenced germline vehicle for inherited stress effects so far is not a methyl mark.

nature.com · 10.1038/nn.3695 ↗

● Mouse F1 + F2 Olfr151 · IVF control

Parental olfactory experience influences behavior and neural structure in subsequent generations

Dias & Ressler · Nature Neuroscience 2014 · odour fear conditioning before conception

“increased behavioral sensitivity to the F0-conditioned odor, but not to other odors”— Dias & Ressler, Nat Neurosci 2014

FoundMales were shock-conditioned to acetophenone — a cherry-and-almond-like smell that activates one identified receptor, Olfr151 — then bred, both by natural mating and separately by IVF, to rule out learned behaviour and the in utero environment. Pups that had never met the odour startled more to it and not to others, and so did the grand-pups. Both generations had more neurons detecting that odour and an enlarged brain structure devoted to it, and the Olfr151 region was hypomethylated in the fathers' sperm.

CaveatThe single most striking result here, and the most contested: independent replication has been patchy, the proposed route from a fear memory to a specific sperm locus has no accepted mechanism, and the reported methylation differences are small. Ressler himself, speaking to the press at the time, put the human extension only as speculation — that a parent's anxiety might reach later generations through marks on stress-hormone receptor genes. Cite it as an open question, not a settled fact.

nature.com · 10.1038/nn.3594 ↗

05 · Three routes, often confused

How an experience can reach a grandchild

Popular coverage collapses these into one story. Only the third one requires anything to survive in the germline, and it is the only one that would count as epigenetic inheritance.

ROUTE 1 · social

Care and environment

A traumatised parent parents differently. The child's own methylation changes in response to their own childhood. Weaver 2004 is the mechanistic version of this — and cross-fostering switches it off.

ROUTE 2 · in utero

Fetal programming

Maternal stress hormones, nutrition and placental signalling act on a developing fetus during the window when methylation patterns are being set. Heijmans 2008 and Perroud 2014 sit here.

ROUTE 3 · germline

Through sperm or egg

A mark survives in the gamete and shapes the next embryo. The obstacle is that mammalian germlines erase and rewrite most methylation twice — after fertilisation and again during germ cell development. Small RNA has cleared this bar experimentally; methylation has not, convincingly.

06 · The bar

What a convincing result would have to include

Bernhard Horsthemke laid out the requirements in a 2018 comment in Nature Communications. It is short, free to read, and the most useful thing to hand anyone who sends you a headline about inherited trauma.

▲ Comment Open access

A critical view on transgenerational epigenetic inheritance in humans

Bernhard Horsthemke · Nature Communications 2018 · 9:2973

“Lamarck has never been dead and every so often raises his head”— Horsthemke, Nat Commun 2018

FoundTransgenerational epigenetic inheritance is established in plants, nematodes and flies. In mammals — and in humans especially — the evidence is entangled with ordinary genetic inheritance, shared environments, and culture. Horsthemke argues that most human findings labelled transgenerational never actually reach the F2/F3 standard, and sets out what a study would have to do to untangle the three.

nature.com · 10.1038/s41467-018-05445-5 ↗

Why the germline is the hard part

Two structural facts stand between a life event and a grandchild's genome. First, the Weismann barrier: in mammals, germ cells are set aside early and are not updated by what happens to the body's somatic cells. Second, the genome is stripped and rewritten twice — once in the primordial germ cells, and again just after fertilisation. For an acquired methylation mark to be inherited it has to survive both erasures, and only a small set of loci, mostly imprinted ones, is known to do that. This is why a mark measured in blood is not yet an inheritance claim, and why sperm small RNA, which is not erased the same way, has been an easier carrier to demonstrate.

Use inbred strains and strictly controlled environments, so a genetic difference cannot masquerade as an epigenetic one.
After maternal gestational exposure, study F3 or later. After paternal exposure, study at least F2. Anything earlier was directly exposed.
Use IVF, embryo transfer and foster mothers where possible, so the phenotype cannot travel through pregnancy or parenting.
Identify the actual carrier in the germ cells — and check that germ cell preparations are not contaminated with somatic cells.
In humans, accept that ecological and cultural inheritance cannot be ruled out — but genetic effects can and should be excluded explicitly.

What the field does not need is another intergenerational cohort reporting a p-value at one CpG; it has several of those, pointing in interesting but mechanistically ambiguous directions. What is missing is human data at the F2 or F3 standard, paired with a germline carrier that plausibly survives both rounds of reprogramming — which at the moment looks more like sperm small RNA than like methylation itself.

07 · At a glance

Every study on this page, side by side

The column that decides how much any of this can prove is the fourth one: which generations were actually tested, and whether the last of them was ever exposed.

Study	Species	Locus measured	Generations tested	Type
Heijmans 2008	human	IGF2 DMR	F0 → F1	direct exposure (in utero)
Yehuda 2016	human	FKBP5 intron 7	F0 → F1	intergenerational
Yehuda 2014	human	NR3C1 exon 1F	F0 → F1	intergenerational
Perroud 2014	human	NR3C1 exon 1F	F0 → F1	intergenerational
Weaver 2004	rat	GR exon 1₇ promoter	within one generation	developmental, reversible
Franklin 2010	mouse	Mecp2 · Cnr1 · Crhr2	F0 → F1 → F2	transgenerational (paternal)
Gapp 2014	mouse	sperm small RNA (not DNAm)	F0 → F1 → F2	transgenerational + causal
Dias & Ressler 2014	mouse	Olfr151	F0 → F1 → F2	transgenerational (paternal)

How many people each human study rests on

Heijmans 2008 — sibships compared

Yehuda 2014 — Holocaust offspring

Perroud 2014 — mother–child pairs per group

Yehuda 2016 — Holocaust offspring

Units differ per study, so compare each bar to its own label rather than to the others. The ceiling is structural: a human trauma cohort is limited by how many people lived through one specific historical event and could later be reached for a blood draw. Mouse studies breed purpose-built cohorts across generations instead — which is exactly why the mechanistic claims come from there.

08 · Where the evidence stands

Claim by claim

Claim	In animals	In humans
Stress changes DNA methylation in the exposed individual	Well established, with causal manipulation	Well supported, mostly in blood
Parental stress changes the offspring's methylation and behaviour	Established, including via parenting behaviour	Reported repeatedly; small cohorts, mixed directions
A mark survives in the germline and reaches an unexposed generation	Demonstrated for sperm small RNA; contested for methylation	Not demonstrated — the required generations have not been studied
A specific methylation change causes a specific inherited phenotype	Not yet shown for a trauma phenotype	Not shown

09 · Why this sits next to my thesis

The missing experiment is a writing experiment

Every study above is correlational at the level that matters most: someone observed a methylation difference and a phenotype together. Nobody put the methyl group there on purpose and watched what followed.

That is precisely what epigenome editing changes. CRISPRoff and ZFPoff deposit methylation at a chosen promoter without cutting DNA, and nanopore sequencing reads the resulting pattern molecule by molecule. On this site, that combination is applied to CD55 — but the general capability is the one this literature has been waiting for: write a mark, read it back, and test whether it does anything.

The honest limit: my work is somatic, in cultured human cells, over days and weeks. It can test whether a methylation mark is stable and consequential in a cell lineage. It cannot test whether one survives a germline. Those are different questions, and this page exists partly so the difference stays visible.

10 · Sources

Papers referenced on this page

1

Persistent epigenetic differences associated with prenatal exposure to famine in humans

Heijmans BT, Tobi EW, Stein AD, Putter H, Blauw GJ, Susser ES, Slagboom PE, Lumey LH · PNAS 2008;105(44):17046–17049 · 10.1073/pnas.0806560105
2

Holocaust exposure induced intergenerational effects on FKBP5 methylation

Yehuda R, Daskalakis NP, Bierer LM, Bader HN, Klengel T, Holsboer F, Binder EB · Biological Psychiatry 2016;80(5):372–380 · 10.1016/j.biopsych.2015.08.005
3

Influences of maternal and paternal PTSD on epigenetic regulation of the glucocorticoid receptor gene in Holocaust survivor offspring

Yehuda R, Daskalakis NP, Lehrner A, Desarnaud F, Bader HN, Makotkine I, Flory JD, Bierer LM, Meaney MJ · Am J Psychiatry 2014;171(8):872–880 · 10.1176/appi.ajp.2014.13121571
4

The Tutsi genocide and transgenerational transmission of maternal stress: epigenetics and biology of the HPA axis

Perroud N, Rutembesa E, Paoloni-Giacobino A, Mutabaruka J, Mutesa L, Stenz L, Malafosse A, Karege F · World J Biol Psychiatry 2014;15(4):334–345 · 10.3109/15622975.2013.866693
5

Intergenerational transmission of trauma effects: putative role of epigenetic mechanisms

Yehuda R, Lehrner A · World Psychiatry 2018;17(3):243–257 · open access · PMC6127768
6

Epigenetic programming by maternal behavior

Weaver ICG, Cervoni N, Champagne FA, D'Alessio AC, Sharma S, Seckl JR, Dymov S, Szyf M, Meaney MJ · Nature Neuroscience 2004;7(8):847–854 · 10.1038/nn1276
7

Epigenetic transmission of the impact of early stress across generations

Franklin TB, Russig H, Weiss IC, Gräff J, Linder N, Michalon A, Vizi S, Mansuy IM · Biological Psychiatry 2010;68(5):408–415 · 10.1016/j.biopsych.2010.05.036
8

Implication of sperm RNAs in transgenerational inheritance of the effects of early trauma in mice

Gapp K, Jawaid A, Sarkies P, Bohacek J, Pelczar P, Prados J, Farinelli L, Miska E, Mansuy IM · Nature Neuroscience 2014;17(5):667–669 · 10.1038/nn.3695
9

Parental olfactory experience influences behavior and neural structure in subsequent generations

Dias BG, Ressler KJ · Nature Neuroscience 2014;17(1):89–96 · 10.1038/nn.3594
10

A critical view on transgenerational epigenetic inheritance in humans

Horsthemke B · Nature Communications 2018;9:2973 · open access · 10.1038/s41467-018-05445-5

Last checked August 2026. Open-access links are marked; the rest lead to the publisher's abstract.