Minireview: Transgenerational Epigenetic Inheritance: Focus on Endocrine Disrupting Compounds

Abstract

The idea that what we eat, feel, and experience influences our physical and mental state and can be transmitted to our offspring and even to subsequent generations has been in the popular realm for a long time. In addition to classic gene mutations, we now recognize that some mechanisms for inheritance do not require changes in DNA. The field of epigenetics has provided a new appreciation for the variety of ways biological traits can be transmitted to subsequent generations. Thus, transgenerational epigenetic inheritance has emerged as a new area of research. We have four goals for this minireview. First, we describe the topic and some of the nomenclature used in the literature. Second, we explain the major epigenetic mechanisms implicated in transgenerational inheritance. Next, we examine some of the best examples of transgenerational epigenetic inheritance, with an emphasis on those produced by exposing the parental generation to endocrine-disrupting compounds (EDCs). Finally, we discuss how whole-genome profiling approaches can be used to identify aberrant epigenomic features and gain insight into the mechanism of EDC-mediated transgenerational epigenetic inheritance. Our goal is to educate readers about the range of possible epigenetic mechanisms that exist and encourage researchers to think broadly and apply multiple genomic and epigenomic technologies to their work.

The word epigenetics is loosely used and its meaning is still being debated (1). Here, we are using epigenetics to mean “DNA sequence-independent changes in chromosomal function that yield a stable and heritable phenotype” (2). Thus, transgenerational epigenetic inheritance covers changes in biological traits that are not mediated by changes in the DNA primary sequence and are passed on to subsequent generations. The result is a change in gene expression, which can be brought about by chemical modifications on DNA or histone proteins and/or by one of an increasing number of different noncoding RNAs (ncRNAs) that alter chromatin structure and DNA accessibility.

In this review, we will focus primarily on endocrine-disrupting compounds (EDCs) as the environmental agents causing transgenerationally inherited traits and epigenetic modifications (3, 4). Like other forms of inheritance, transgenerational inheritance must be coded in the genome and epigenome of primordial germ cells (PGCs, oocytes, and spermatozoa) (oocytes and spermatozoa). To be called transgenerational, expression of the trait has to persist for at least 3 generations after the initial exposure to the environmental agent. For transmission via either the maternal or the paternal parental generation (P0), a phenotype may be present in the first filial generation (F1), but this does not ensure that the trait is heritable. In maternal transmission, if a pregnant female is exposed to an EDC that passes through the placenta, her embryos (F1) and the germ cells they possess (F2) are also directly exposed. Thus, the transgenerational phenotype has to be displayed by F3 and subsequent generations (4–6). Importantly, a phenotype may be expressed to different degrees in F1, F2, and F3 generations. Exposures in the F1 can produce modifications caused by direct interactions of the EDC with the developing tissue (brain, mammary tissue, or prostate). Changes in F2 phenotypes may be influenced by temporary modification in germ cell genomes and epigenomes, again produced by direct contact with the EDC. Modifications observed in the F2 need not be identical to those in the F1 or to the final transgenerational phenotypes noted in F3 (and beyond) generations (7–9).

In contrast, if the trait is passed down via the paternal lineage, the transgenerational phenotype can be established earlier, in the F2 generation. For example, if an adult male (P0) is exposed to an EDC, it can produce epigenetic alterations in germ cells due to ongoing spermatogenesis, which represent the F1 generation. Because the F1 germ cells are not directly exposed when the P0 is exposed, alterations that are expressed in the F2 generation can be attributed to transgenerational epigenetic inheritance because the F2 is the first generation that is not directly exposed to EDC (10) (Figure 1). It is important to understand these definitions because there are many examples of reports claiming “transgenerational” effects, which in fact are inter-generational (F1 or F2) but are not strictly transgenerational.

Figure 1.

A cartoon of the multiple generations comprising transgenerational inheritance. A pregnant woman (P0) carries genetic and epigenetic information for up to the F2 generation. Her fetus (F1) and its germ cells (F2) may be affected by environmental exposures such as EDCs. For any changes in the F1 phenotype or epigenetic state to be considered transgenerational, the aberrant epigenetic state has to be sustained until at least the F3 generation. In contrast to females, EDC exposure in males (P0) will potentially change his germ cells, which will be the F1 generation. However, because the F2 germ cells are not exposed, any phenotypic or epigenetic changes noted in F2 can be considered transgenerational.

In addition to target tissues in the developing embryo, the pregnant female is also exposed to EDCs, which can change maternal behavior directed to the F1 offspring. Such context-dependent changes in behavior may last for the lifetime of the individual, or not, but do not persist for more than a few generations (11). The best-known example is maternal licking and grooming of rat pups, which changes the pups' own maternal behaviors when they themselves are dams. Fostering pups to dams that show high or low levels of the licking and grooming behaviors dictates the pups' maternal behaviors in adulthood, regardless of the phenotype of their biological dam (12). Although behaviors transmitted to offspring in this manner are not accepted as transgenerational epigenetic inheritance, they might be caused by epigenetic alterations of chromatin states of steroid hormone receptors and other related neurotransmitter genes in the brain (13). Thus, changes in parental behavior can produce transient changes in F1 behaviors, which are not heritable. In the next section, we will discuss and summarize the major mechanisms of epigenetic regulation.

Major Epigenetic Mechanisms

DNA modifications

DNA methylation is the most widely studied epigenetic modification. As the name suggests, it is a biochemical process in which “methyl” chemical groups are covalently attached to “cytosine” (C) residues by DNA methyltransferase enzymes. Although it was initially thought that only cytosines in cytosine-guanosine dinucleotide pairs (CpG) could be modified, later studies demonstrated that other non-CpG cytosine residues are methylated as well (14, 15). DNA methylation is historically associated with transcriptional repression. However, it should be noted that in addition to methylation, the cytosine residues are also hydroxymethylated, formylated, and carboxylated (16, 17). Although the functional roles of these modifications are not fully understood, they are considered intermediate modifications during the DNA demethylation process. The formation of these alternate modifications is facilitated by ten-eleven translocation (TET) enzymes (18). Transgenerational alterations in DNA methylation have been observed globally and at specific genomic loci in studies using EDCs and other environment-related factors in rodents (19–22). It is likely that DNA methylation has received more attention than other epigenetic modifications because, before discovery of TET enzymes, this modification was considered to be highly stable and permanent.

Histone modifications

Histone proteins are the building blocks of genomic DNA organization. Importantly, histone proteins undergo various posttranslational modifications in their tails (23). The modifications change the local chromatin structure by altering the electrical charge of the histone tails. At last count, there are >100 histone modifications, only a handful of which have been studied. Recent mapping of various histone modifications at the whole-genome level reveal important insights into the biological function of these modifications (24–26). Unique sets of histone modifications are associated with distinct regulatory regions and genomic activity (Figure 2). For example, active enhancer regions are marked with histone 3 lysine 27 acetylation and histone 3 lysine 4 monomethylation (H3K4me1), active promoters are marked with histone 3 lysine 4 trimethylation (H3K4me3), and repressive genomic regions are marked with H3K27me3 (24, 27, 28). Importantly, mutations in histone modifiers (enzymes that add or remove various histone marks) are recurrently found in many diseases states including cancer (29–31).

Figure 2.

Epigenomic features of expressed and silenced chromatin states. Genomic positions and patterns of various histone modifications are based on chromatin immunoprecipitation-sequencing–mediated epigenome mapping. Active genes are marked with H3K4me3 in their promoters and H3K36me3 in the gene body. Active enhancers are marked with H3K4me1 and H3K27 acetylation (ac). On the other hand, silenced genes have repressive H3K9me3 and H3K27me3 marks as well as DNA methylation in their promoter regions.

Probably the most complete dataset on transgenerational epigenetic inheritance of histone modifications comes from studies in Caenorhabditis elegans. It is important to keep in mind that these animals do not have DNA methylation. Genetically wild-type descendants of mutants for the H3K4me3 methyltransferase complex (COMPASS) display significantly longer life spans (32, 33). Another set of data implicates mutations in the worm homolog of lysine specific demethylase 1 (LSD1), the H3K4me2 histone demethylase enzyme. This mutation causes progressive sterility and smaller brood sizes over many generations of worms (34). Moreover, in the LSD1 mutant animals, the H3K4me2 mark increases in germ cells, leading to aberrant regulation of spermatogenesis genes and sterility over generations. LSD1 interacts with the androgen receptor, and interestingly, when C. elegans develop on medium containing testosterone, the behavioral responses of the offspring to gentle touch are reduced (31). This behavioral modification persists for 4 generations after the initial exposure. Several potential androgen receptor gene sequences were identified in this organism, and after consumption of bacteria containing RNA interference (RNAi) for a subset of these genes, the behavioral effects of testosterone are blocked (35). These data suggest that natural hormones, such as EDCs, can have transgenerational actions.

In Drosophila, stress-induced alterations in dATF2 (Drosophila activating transcription factor-2) function, which involves HP1 (heterochromatin protein 1), results in a defective chromatin state over multiple generations (36). HP1 is an essential mediator of heterochromatin formation, requiring H3K9me3 methylation by the enzyme SUV39 methyltransferase (37). Polycomb-induced repressive chromatin formation by H3K27me3 is another pattern altered over many generations in response to diet or stress in flies and mice (38, 39). Thus, there is precedence for a role of the histone modification proteins and enzymes in transgenerational inheritance.

ncRNAs

ncRNAs are being increasingly recognized as major players in epigenetic gene regulation and global chromatin organization. Thousands of short and long ncRNAs have been discovered; however, the molecular function of the vast majority of these is yet to be discovered (40–42). Seminal work on several long ncRNAs has shown their vital functions in gene regulation and chromatin organization. The best example is Xist ncRNA, an essential player in X-chromosome inactivation in mammals (43). X-chromosome inactivation is a dosage compensation process in females (XX) by which one of the X chromosomes is randomly silenced by Xist-mediated polycomb group proteins. The orchestrated recruitment of key components of polycomb group proteins such as enhancer of zeste homolog 2 (EZH2) by Xist is one of best characterized repressive epigenetic mechanisms coordinated by a ncRNA and a histone modifier, EZH2 (44). EZH2 is the major enzymatic subunit of the polycomb repressive complex that catalyzes the H3K27me3 modification.

In addition to long ncRNA, small ncRNAs such as short interfering RNA, microRNA (miRNA), small nucleolar RNA, and piwi-interacting RNA are important regulators of gene expression in various organisms ranging from yeast to human (45). Whether small ncRNAs (small interfering RNA and miRNA) mediate classic epigenetic mechanisms in mammalian systems is still being debated (46, 47). However, there is compelling evidence in plants that they mediate epigenetic gene regulation through DNA methylation (48, 49). For example, in higher plants interactions between short RNA and genomic DNA triggers de novo DNA methylation, a process known as RNA-directed DNA methylation (50). Furthermore, RNAi-directed assembly of heterochromatin formation is well established in fission yeast (51) and C. elegans (52). Importantly, small RNA-mediated H3K9me3 methylation is implicated in the transgenerational inheritance of the RNAi mechanism in C. elegans (53, 54). These examples highlight how the epigenetic regulation of gene expression is orchestrated with layers of mechanisms coordinated through ncRNAs, DNA methylation, and histone modification.

Transmission of Epigenetic Information

Epigenetic mechanisms control how DNA sequence information is being used. Modifications in DNA and histone proteins and in ncRNA molecules are implicated in chromatin structure and function and are major forms of epigenetic information. The exact mechanism(s) of transmission for epigenetic information through mitotic or meiotic cell cycle division is yet to be understood (55). DNA methylation is mitotically transmitted in a semiconservative fashion (56). However, there is no clear evidence that histone modifications are transmitted the same way. Indeed, recent studies suggest an alternative scenario that the histone modifiers, ie, the enzymes that catalyze the modification, are transmitted: they remain bound to DNA during DNA replication and hence reestablish the histone marks on the sister chromatids (57, 58).

For germline-dependent transmission, epigenetic information must be transmitted during meiosis. Thus, the mechanism of transgenerational epigenetic inheritance is more complex than mitotic transmission. This is because the transmitted epigenetic information has to be maintained during the 2 major waves of global reprograming events at the onset of each generation, fertilization, and PGC development, associated with epigenome resetting events. Studies of DNA methylation patterns shed important light on the dynamics of these epigenomic-resetting events. For example, right after fertilization, before the sperm and oocyte pronuclei merge, DNA methylation is actively erased globally from the paternal genome (59, 60). Moreover, the maternal genome undergoes a replication-dependent, passive DNA demethylation process that is completed by the blastocyst stage. As a result of this major wave of DNA methylation erasure, the genome of the inner cell mass is substantially hypomethylated. In the inner cell mass, only 20% of CpG sites are methylated, but in sperm and oocytes nearly 90% and 40% of the CpG sites (respectively) are methylated (56). As embryonic development progresses from the blastocyst to epiblast stage (around embryonic day [E] 6.5 in the mouse), the genome undergoes de novo methylation whereby nearly 70% of the genome is methylated. Around this developmental stage, subsets of epiblast cells follow specific sets of developmental cues and become PGCs in which somatic transcription is silenced (61, 62). The second major period for epigenomic reprograming occurs during development of the PGCs. As the PGCs begin to migrate through the developing hindgut to colonize the undifferentiated gonads (about E7.25 in mice), the genome undergoes a global DNA methylation erasure (61). These 2 waves of global epigenomic reprograming during preimplantation development and PGC specification are well orchestrated. During this process, major components of de novo DNA methylation are inhibited and key players of DNA demethylation program such as the recently discovered TET enzymes are activated (56). Similar to DNA methylation, there is extensive reprograming of histone modifications during preimplantation and PGC development (63, 64).

Because preimplantation and PGC development are coupled with extensive epigenomic reprograming, it is important to ask how an epigenetic chromatin state is transmitted from parent to offspring. A partial answer to this question comes from genomic imprinting studies in which only 1 allele is expressed based on its parent of origin. The epigenetic state of imprinted genes escapes from both waves of epigenome reprograming (55, 65). Interestingly, in addition to the imprinted genes, the chromatin state of pericentromeric heterochromatin and certain genomic repeat regions such as intracisternal A particles (IAPs) escape the global reprograming process (66).

Indeed, the best-studied mouse model of epigenetic inheritance includes insertion of an IAP element distal to the Agouti locus (67). These studies suggest that certain genomic regions, such as imprinted genes and IAP elements, are more susceptible to epigenetic alterations that persist in subsequent generations. Several EDCs can disrupt the normal imprinting process when exposure occurs during either the fertilization and/or germ cell migratory periods. Two (Igf2r and Peg3) of 3 imprinted genes examined are hypomethylated in germ cells from gestational day 12.5 in mice treated with a low dose of diethylhexyl phthalate or bisphenol A (BPA) (68, 69). These genes are still hypomethylated in 21-day-old oocytes compared with controls. Even more interestingly, hypomethylation persists in F2 offspring germ cells. Rats treated with BPA for the first 5 days after birth possess hypomethylation at the imprinted H19 gene promoter in their spermatozoa (70). Unfortunately, transcript levels of these genes have not been quantified. The only truly transgenerational data (F1, F2, and F3) on imprinted genes in sperm show that vinclozolin decreases the methylation of H19, Peg3, and Snrpn (71).

One aspect of sperm maturation that complicates identification of epigenetic modifications produced by EDCs is that the chromatin undergoes a remarkable remodeling process during spermatogenesis. During this process, protamine proteins replace the vast majority of histone proteins. The current view is that only a fraction of the nucleosomes (1%) are retained in mouse spermatozoa; however, in humans a greater percentage (approximately 15%) of nucleosomes are retained (72). Importantly, recent epigenome mapping studies show unique patterns of histone modifications in human and mouse spermatozoa (72–74). For example, H3K27me3 and H3K4me2 are retained at the promoters of developmentally critical genes (73, 74). Moreover, detailed mapping studies also show that the retained nucleosomes are particularly enriched in CpG-rich gene promoters that lack DNA methylation in mouse spermatozoa (72). Whether these histone modifications are transmitted transgenerationally is yet to be determined. However, whole-genome mapping studies suggest that the H3K27me3 mark in sperm is implicated in gene repression during early stages of preimplantation development (72). During fertilization, the entire sperm is fused to the ovum. Therefore, in addition to the histones, coding and ncRNAs such as miRNAs present in sperm are also transferred to the ovum (75). It is also likely that sperm RNA and protamine modifications contribute to intergenerational as well as transgenerational epigenetic inheritance (75–77). Further studies are needed to elucidate the epigenetic roles for histone, DNA methylation, and other mechanisms.

EDCs, Steroids, and Epigenetics

The classic case of an EDC is diethylstilbestrol (DES), an estrogen agonist and androgen receptor antagonist synthesized first in the 1930s and prescribed to at least 5 million women at risk for miscarriage or experiencing other reproductive problems, from 1938 up to 1975. Instead of the desired effects, use of this compound lead to increased incidence of breast, vaginal, and cervical cancers (78–81). In addition, maternal exposure has documented adverse affects on daughters. These include the same types of cancers as well as a variety of difficulties conceiving and maintaining pregnancies, reproductive tract malformations, abnormal menstrual cycles, early puberty, and behavioral issues (81–83). The vast variety of effects is probably related to complexities introduced by the timing of DES treatment and doses. For example, a recent large study of women exposed prenatally to DES revealed a strong correlation between DES, particularly in the first trimester, and noncancerous uterine fibroids (84). Offspring of rodents exposed to DES during pregnancy recapitulate many of these effects (85–87).

Although the initial clinical studies were limited to female offspring, correlations between DES exposure and hypospadias, cryptorchidism, and testicular cancer have been reported in F1 and F2 sons and grandsons of women given DES (88–90). Analyses of the clinical studies suggest that the male reproductive illnesses are related to but not necessarily caused by estrogen actions. An alternative hypothesis is that DES produces low-birth-weight babies, and these infants are more prone to testicular dysgenesis syndrome (90, 91). Multigenerational work in mice has demonstrated that high, but clinically relevant, doses of DES increase the incidence of uterine and other reproductive tract tumors in females and lesions in the male rete testes in F2 offspring (86, 92, 93). Few data on the critical F3 generation in humans are available nor are there experimental data from rodent models (81).

EDCs, such as DES, share many properties with steroid hormones: they act at low doses (picograms) and can act in a nonmonotonic manner (94, 95). Like hormones, they are particularly effective during development, at which time they can modify the course of reproductive tract and brain development (96–98). Importantly, the EDCs are more promiscuous than steroids and bind to a larger variety of receptors than normal ligands, albeit with reduced affinities (95).

Steroid receptor proteins are transcription factors (TFs) and typically act in association with clusters of protein complexes that include coactivators or corepressors, all of which are required to access DNA. Many of these coregulators (ie, NCor [nuclear receptor corepressor], CBP [cAMP response element–binding protein–binding protein], and others) are also epigenetic factors (99). Numerous interactions between steroids, their receptors, and epigenetic mechanisms are known. For example, estradiol stimulation of breast cancer cells alters a number of chromatin modifications as well as DNA methylation and chromosome interactions (100). Estrogen receptor α interacts with and suppresses the Drosha enzyme, which in turn affects miRNA expression (101), and estradiol affects methylation of its α receptor (102). Therefore, EDCs may alter the epigenetic state by regulating the activity of various TFs, thereby changing expression of chromatin regulators such as DNA or histone methyl transferases and/or ncRNAs. Moreover, TFs can also directly change the local chromatin states by recruiting epigenetic regulators (Figure 3).

Figure 3.

A general model depicting epigenetic alterations upon exposure to EDCs. EDCs bind to steroid receptors, and other TFs, thereby changing the local chromatin states and expression of various chromatin modifiers (CM) such as histone and DNA modifiers or ncRNAs. Moreover, EDCs can change the composition or the activity of epigenetic chromatin regulator complexes and thereby act directly on the global epigenome.

BPA is one of the most heavily manufactured EDCs worldwide (103). BPA is commonly used as a plastic hardener in bottles and for epoxy resins that line metal-based food and beverage cans, and most recently it has been added to thermal cash receipts. Globally, approximately 6 billion tons are produced each year. BPA has been shown to alter various epigenetic mechanisms, for example, it promotes hypomethylation and hypermethylation (104, 105). Moreover, supplementing mice with a methyl donor–rich diet can reverse the ability of BPA ability to hypomethylate. BPA also modifies expression of a number of the estrogen receptors in brain (7, 106) and DNA methylation of both Esr1 and Esr2 (107, 108). Immunocytochemical analysis has been conducted on testes from F1 and F3 rats, steroid receptors, and several coactivators (androgen receptor, estrogen receptor β, steroid receptor coactivator 1 [SRC1], and NCor) are reduced by BPA exposure across all generations (109). These types of relationships have been noted for other EDCs. Given these interactions between the epigenome and steroid hormones and their receptors, it is not surprising that EDCs can promote transgenerational change.

EDCs and Transgenerational Epigenetic Inheritance

EDCs have transgenerational actions on multiple target tissues in rodents, particularly germ cells, gonads, and brain (4, 19). Exposure in midgestation to the phthalate diethylhexyl phthalate produces puberty delay in males both in the F1 and F3 generations, and F3 males have lower sperm counts, testicular germ cell function, and increased incidence of abnormal seminiferous tubules (8). These phenotypes are noted in F3 males regardless of the parent of origin. Dioxins (eg, 2,3,7,8-tetrachlorodibenzo-p-dioxin) are another class of EDCs found throughout the world. In mice, exposure to 2,3,7,8-tetrachlorodibenzo-p-dioxin during pregnancy reduces fertility in F1 and F3 female offspring (110). Mixtures of EDCs can also have transgenerational actions (111, 112). The combination of phthalates and BPA, at 2 doses, several dioxins, and jet fuels causes puberty advancement female rats 3 generations removed from the initial gestational exposure and reduces the numbers of ovarian follicles. In F3 males, serum testosterone is lower in adults of all EDC lineages, with the exception of the pesticides.

The first EDC shown to possess transgenerational actions was vinclozolin, a fungicide in use since 1984 in high-production agriculture (9). Aged male rats and mice exposed to high levels in utero have higher rates of germ cell death and disease-like phenotypes in prostate, kidneys, and other tissues (through at least the F3 generation). In rats, the effects on sperm are more extreme than those in mice, including decreased numbers and motility (9, 113). Female F3 rats and mice exposed to vinclozolin as embryos in the P0 generation have increased numbers of ovarian cysts (113, 114). Vinclozolin lineages have fewer numbers of newborn rat oocytes and fewer primary follicles in mature ovaries, suggesting that germ cell death in embryonic PGCs is increased by this EDC (114). Behavioral effects are present as well and include an effect of vinclozolin on female mate preference and heightened anxiety-like behavior in females but less anxiety in males, and this phenotype is modified by stress during the pubertal period (115, 116).

DNA methylation analyses have been conducted with materials from F3 vinclozolin and control rats and mice (9, 21, 113, 117). Immunoprecipitation (methylated DNA immunoprecipitation-chromatin immunoprecipitation) followed by hybridization to promoter tiling arrays, bisulfite, and/or pyrosequencing has produced a series of articles showing differentially methylated regions (DMRs) between material from F3 control and vinclozolin lineages. Vinclozolin has both hypomethylating and hypermethylating actions. Many of the identified DMRs contain known genes, and recently one such region displayed a copy number variation (117). A consensus sequence motif, which is associated with genes containing low CpG density (1%–8%), is present in most of the confirmed promoters. The DMRs in F3 vinclozolin-exposed mice and rats are not conserved, but the motif identified in rats is also present in affected mouse DMRs (113). Interestingly, the largest gene pathway affected in E13 germ cells and in brain is an olfactory transduction pathway (116, 117). This may be related to some of the transgenerational behaviors reported.

BPA has transgenerational effects on behavior and sperm in rodents (7, 118, 119). Pregnant mouse dams ingested BPA at a low concentration in their chow throughout pregnancy and on the day of birth all pups were transferred to control dams. This was done to block the actions of BPA on maternal behavior and to ensure that exposure was ended at birth. F1 and either F3 or F4 juveniles were tested on several types of behavioral tasks. The social interaction tasks provided to be sensitive to BPA. One of the tasks used was an unrestrained interaction between nonfamilial juveniles of the same age and sex and given the same treatment. In this task, F1 mice were less socially interactive than controls if they had been exposed to BPA in utero. However, in the F2 and F4 generations, the BPA lineage mice were more interactive than the controls (7). In the well-established social recognition task, juveniles exposed to BPA in utero did not habituate as rapidly to repeatedly presented adult females compared with controls, and this behavior persisted in F3 mice (118). This hyperactivity phenotype is noted by others also giving mice low doses of BPA (120). Finally during the dishabituation portion of the social recognition task, control juveniles display heightened interested in a novel adult, whereas F3 mice in the BPA lineage fail to do so. This recognition failure is not caused by lack of olfactory abilities, and taken together, the data show that social interactions are affected transgenerationally by BPA exposure during gestation.

BPA can increase or decrease DNA methylation, depending on the dose (94). In young girls, urinary concentrations of BPA are associated with hypermethylation and hypomethylation of DNA, and, as in mice, higher BPA levels are correlated with hypomethylation (121). In addition, BPA increases miRNA expression in immortalized placenta cells (122). BPA enhances expression and protein for the histone methyltransferase, Ezh2 in human cancer cells and mouse mammary tissues (123). This methyltransferase is part of the polycomb group complex 2, which is essential for genetic imprinting during germ cell development. Enhanced H3K27me3 is likewise a result of BPA exposure. Although the studies are in their infancy, it is certain that more than one epigenetic mechanism underlies the transgenerational actions of BPA.

Moving Ahead

As described above for BPA and vinclozolin, EDCs can alter the epigenome via multiple pathways. Likewise the known examples of epigenetic transgenerational inheritance use a number of epigenetic mechanisms. Of course, although the types of epigenetic information are typically explained separately, these have to act together to establish a particular chromatin and cellular state. For example, the epigenetic mechanisms in X chromosome inactivation require interaction between Xist ncRNA and EZH2, which recruits polycomb group proteins and then deposits repressive H3K37me3 histone marks to silence the inactive X chromosome via DNA methylation (44). Therefore, it is safe to say that it is the total net effect of the combination of these epigenetic processes that establishes a unique chromatin structure and function.

Further studies are needed to address several important questions in the field. The major question that remains unanswered is how to detect transgenerational changes. We suggest that unbiased whole-genome approaches using next-generation or newer sequencing technologies, are essential for discovery (also see Ref. 124). Because the epigenetic mechanisms work together, whole-genome mapping of histone modifications and DNA methylation will reveal aberrant chromatin state(s) at specific loci in normal as well as malignant development (24, 27, 125). To this end, studying ncRNAs is an essential complementary approach. Although it is relatively straightforward to measure differentially regulated coding and ncRNAs across many generations, identification of genomic regions regulated by RNA continues to present a major technical challenge. Unlike histone and DNA modifications, which act in cis, RNA may act in both cis and trans. Therefore, quantification of an ncRNAs does not necessarily reveal the essential genomic region. However, DNA or histone modification mapping will reveal specific positional information, and when combined with RNA expression profiles (RNA sequencing), these unbiased genomic approaches will yield a complete picture of both the epigenetic and the expression states of genic as well as nongenic genomic regions.

Many have shown that EDCs mediate both short-term direct effects as well as long-term transgenerational effects. Utilization of state-of-the-art genomic and epigenomic approaches in well-controlled treated and nontreated experimental settings will allow identification of such effects. Moreover to truly assert that an effect is transgenerational, F3 and later generations have to be compared with F1, a step that is often skipped. EDCs and other environmental toxicants may alter the chromatin state in critical genomic regions by changing both histone modifications and DNA methylation patterns. Furthermore, the chromatin features at some of these differentially regulated genomic regions (DRGRs), if not all, may persist over generations, inherited transgenerationally. We would like to use the term “transgenerational epigenetically regulated regions” (TERRs) to describe such critical genomic sites. Our working model is illustrated in Figure 4. Unbiased whole-genome genomic and epigenomic data acquisition and comparative analyses are required to identify both DRGRs and TERRs associated with EDC exposure. Such studies will also reveal the epigenetic mechanisms contributing to establishment of DRGRs and TERRs.

Figure 4.

Hypothetical model for potential short-term and long-term effects of EDCs on the genome of germ cells. Direct exposure (in P0) to EDCs induces DRGRs. The chromatin features at some or all of these DRGRs may persist over generations (as pictured here in the F3) and be inherited transgenerationally. We are calling such regions “transgenerational epigenetically regulated regions” (TERRs). In example 1, a DRGR is present in both sexes after exposure to an EDC and retained transgenerationally. Example 2 illustrates a sex difference in establishment of a TERR. Example 3 is a transient DRGR that is not retained as a TERR.

Other mechanistic questions that remain are how epigenomic changes at specific loci are maintained via the germline and what determines the duration of transgenerational epigenetic traits. To this end, epigenome mapping in the germ cells is essential. However, this also poses significant technical challenges due to limited material. Therefore, new genomic and epigenomic methods that require less material are in great need (27, 126). Another important area of research needed is comparative epigenetic mechanisms of action. Most transgenerational studies use mice, flies, or worms and so far even the limited data available suggest divergence of mechanisms. If improving human health is our ultimate goal, we must be able to pinpoint which mechanisms are conserved in different organisms. The time is ripe for technology development, comparative epigenetic studies, and their applications to transgenerational inheritance of traits in human health and disease.