Transgenerational epigenetic inheritance: Perspectives and challenges

The prevalence of asthma and allergies has continued to increase in the past decades.¹ Genome-wide association studies have linked relatively few genes to asthma and allergies,² underscoring the critical role of the environment and its interaction with genetics. Importantly, recent animal and human studies have linked asthma and allergies to parental, grandparental, or even great-grandparental exposures in the absence of direct exposure of the offspring themselves.³ Such phenomena suggest possible heritable mechanisms that go beyond gene-environment interactions, such as epigenetic modifications (hereinafter, epimutation) that include DNA methylation alteration, histones and chromatin modification, small regulatory RNA changes, and more.⁴ Although there is ample evidence for direct environmental effects on epigenetic modifications, whether environmentally induced epigenetic changes can be inherited across generations in humans remains an ongoing debate. However, if substantiated, it could help to explain the increasing rates of asthma.

Transgenerational inheritance of disease is defined as an increased disease risk in the unexposed offspring generation owing to exposure in earlier generations. This is distinct from an intrauterine effect in which the fetus is directly exposed via maternal routes or an intergenerational effect in cases where prefertilization parental germ cells have been exposed and then develop into the offspring. In some cases, there is exposure to persistent chemicals such as polychlorinated biphenyls, which can pass from the mother to the female fetus and stay in the offspring’s body long enough to directly expose the offspring’s own fetus when she is pregnant.⁵

Although not fully elucidated, transgenerational epigenetic inheritance is biologically plausible. A majority of the mammalian DNA methylation undergoes reprogramming during early embryogenesis and germ cell development, but some DNA methylation signals may escape, similar to imprinted genes.⁶ Small regulatory RNAs could also facilitate the escape of DNA methylation reprogramming via ancestry extracellular vesicle RNA packages delivered to early embryos for global methylation reprogramming. Other epigenetic mechanisms such as histone and ribosome modification have the potential to directly pass through the maternal side.

Whereas transgenerational epigenetic inheritance has been well studied in animal models,⁷ human studies on transgenerational epigenetic inheritance are still lacking. Nevertheless, numerous studies have provided evidence of intergenerational effects of grandparental or nonpregnant parental smoking on asthmatic and allergic disease activity.⁴ With many ongoing decades-long multigenerational population studies, breakthroughs in human transgenerational epigenetic inheritance are expected. Here we provide a conceptual framework to investigate transgenerational epigenetic inheritance, with adoption of a recently proposed “induced epigenetic transmission” pathway.⁸ We also discuss the challenges and potential solutions in transgenerational studies.

Examining the role of epigenetics in transgenerational inheritance requires a comprehensive analysis of the environment and the epigenome (Fig 1). It is essential to control the confounding effect from multigenerational environmental exposures (eg, shared social and cultural environment), although this can be very difficult to accomplish in population studies. With confounding controlled, transgenerational epigenetic inheritance exists if the following 2 conditions are met: (1) the specific environmental exposure in the nonpregnant grandparental (or pregnant great-grandparental) generation has significant associations with epimutation of both the grandparental and the offspring generations, although their epimutations need not be identical; and (2) in the offspring generation, the epimutation is associated with the specific phenotypic disease risk. In the grandparental generation, a direct link between the exposure and the specific phenotypic disease is not necessary because the exposure may happen outside the biologically relevant window for the disease but still leave marks on the germ cells that carry over to later generations, particularly given that germ cells have been suggested to be more sensitive to environmental insults during reprogramming phases.

FIG 1.

Conceptual framework for transgenerational epigenetic inheritance. The traditional pathway (Path A) involves only inherited germline epimutation passing along generations. An alternative Path B involves parental somatic epimutation induced by grandparental germline epimutation. Parental somatic epimutation induces parental phenotypic changes that induce de novo parental germline epimutation. A germline epimutation, either inherited or de novo, could pass on to the offspring and the offspring disease phenotype. Confounding pathways involve shared environment.

The traditional pathway (Path A in Fig 1) of transgenerational epigenetic inheritance focuses on a germline epimutation passing along generations. However, it is also biologically plausible that the manifestation of a grandparental germline epimutation results in parental epimutation on both the germline and the somatic cells, in which case the epimutation in the somatic cell induces parental phenotypic changes (the alternative Path B in Fig 1). Such phenotypic changes could serve as an exposure that induces parental de novo germline epimutation. Both inherited and de novo parental germline epimutations affect the offspring generation’s germline or somatic epigenetics that eventually increase disease risk. Notably, parental phenotypical changes must be associated with grandparental-specific exposure, whereas parental de novo germline epimutation should also be associated with offspring disease. Control of any relevant genomic modification is needed if the grandparental exposure potentially modifies the germline genome. In addition, the whole genome should be carefully examined to tease out methylation quantitative trait loci and genetic mutations that remove the transcription termination signal and thus mimic an epimutation on neighborhood genes. Such an alternative induced epigenetic transmission pathway has been found in mouse studies in which pregnant F0 exposures to traffic particles have been associated with differentially methylated regions in F1, F2, and F3.⁹ Whereas these differentially methylated regions across generations were not overlapped, enrichment analyses revealed transgenerational epigenetic inheritance via chromatin remodeling factors and changes outside gene bodies.⁹

Investigating this alternative path in humans is challenging given its requirement for multigenerational measures on the epigenome, the genome, and (preferably) the transcriptome.

A major difficulty in transgenerational studies is the lengthy follow-up needed across at least 3 populations and the ability to adequately measure confounder information across all generations. Additional challenges in the study of transgenerational epigenetic inheritance are summarized in Table I. Recruiting all grandparents, parents, offspring, and their siblings is an ideal but often unrealistic scenario in epidemiologic studies. Another consideration is subfertility, which should be carefully assessed to control its influence on study population selection, measurement error, and the estimation of true population-level health effects.

TABLE I.

Challenges in the study of transgenerational epigenetic inheritance

Source of errors	Potential solutions
Selection bias
Sampling strategy in population: sampling maternal vs paternal ancestries answers different questions, whereas subsampling siblings in the offspring generation may introduce bias	Select all members in all generations; control for number of offspring siblings if needed
Subfertility or infertility could affect selection into the study	Consider using IPW to investigate the effect on the health of the overall population, recognizing that bias is less relevant in assessing exposure-disease association
Cohort retention is likely to be influenced by socioeconomic and other confounding factors	Collect information and control in the analysis; consider using IPW as well
Information bias
Retrospectively collected information could induce ascertainment and recall bias, especially when the offspring is diseased	Use objective measures such as biomarkers or historical records
Sample storage and degradation from long-term, unfavorable conditions	Use statistical analysis and control of laboratory variation
Batch effect	Use consistent intraplate and interplate controls; include samples from the same family in the same plate; randomize plate
Cell type composition could dilute the association of epimutations with an exposure if more cells than the target cell are included	Isolate cell; perform cell type estimation; control in analysis
Confounding effect
Shared environment and lifestyle factors across generations	Comprehensively measure environment information from multiple generations; careful analysis of familial clusters
Genetics	Control genetic variation; use a multi-omics approach
Reverse causation
Offspring disease status can induce epimutation	Use prospective design and mQTL-based mendelian randomization
Statistical consideration
When the middle generation’s contribution is analyzed, special consideration of confounders between F1 and F2 is needed	Use causal mediation
Clustering in the analysis with multiple offspring included in the analysis needs to be considered because of the violation of independent observations	Use random effect modeling

IPW, Inverse probability weighting; mQTL, methylation quantitative trait loci.

In biospecimen assays, it is preferable that the duration and condition of sample storage and batch effect be controlled in both the design and analysis phases. Because the epigenome is cell type–specific, isolation and purification of the target cells from the target organ are preferred, although most epidemiologic studies are limited to noninvasive collection (eg, blood, urine). In constructing historical exposure in the grandparental generation, use of objective measures such as biomarkers or historical records is potentially less prone to recall bias than either self-report or index report by the offspring generations. Samples and information collected from the middle generations are also necessary to control for potential confounding effects.

Constructing the alternative pathway requires causal mediation analysis of the role of parental phenotypical changes in the association between grandparental exposure and grandchild disease risk as the first step. Confounders in the parental generation should be carefully controlled, particularly when they are also induced by grandparental exposure. In this case, marginal structural modeling may be useful to control collider stratification biases. Methylation quantitative trait loci–based mendelian randomization can be used as an instrumental variable of offspring epimutation to assess unmeasured confounders and reverse causation.¹⁰

In conclusion, animal studies have provided compelling evidence for transgenerational epigenetic inheritance, whereas evidence in humans is still lacking. Here we have discussed a traditional germline-based pathway and presented an alternative non–germline-induced epimutation-phenotype pathway as a potential conceptual framework underlying transgenerational inheritance. Although the challenges to this research are numerous, multigenerational human studies are under way, and we anticipate that new evidence will come to light in subsequent years as more standardized protocols are put in place, more statistical approaches are developed, and greater efforts to integrate multiple layers of omics data and to disentangle multigenerational confounding are made.