3.1 It is impossible to separate epigenetics from genetics.
In May 2016, an article in the New Yorker about how environmental factors can change the activity of genes without altering the DNA sequence, written by the cancer researcher and author Siddhartha Mukherjee [37], stirred a strong critical response by geneticists, epigeneticists, and other biologists [38]. They criticized that Mukherjee, by emphasizing histone modification and DNA methylation, ignored the primary role of transcription factors and RNA in the transcription process. Mark Ptashne and John Greally pointed to the significance of specificity: "Development requires the highly specific sequential turning on and off of sets of genes. Transcription factors and RNA supply this specificity, but enzymes that impart modifications to histones cannot: every nucleosome (and hence every gene) appears the same to the enzyme." Not only the specificity of cellular identity but also the response to stress "has been known for decades to be due to the actions of specific DNA binding proteins (and, more rarely, RNA molecules) that regulate gene transcription" [39]. In an exceptional and most commendable response, Mukherjee thanked his critics for their “immensely detailed comments”, admitting to having erred by "omitting key areas of the science" [38].
The exclusion of mechanisms based on DNA sequence as primary cause of gene expression changes is a severe flaw in the arguments of social epigeneticists, too. Emphasizing the supposed role of DNA methylation and histone acetylation in the control of gene regulation, they do not mention transcription factors. As Stephen Henikoff and John Greally commented, transcription factors "actually have many of the required properties of a regulator of cellular memory or a mediator of environmental influences" [4].
Histone modifiers or enzymes that transfer methyl groups to DNA (methyltransferases) lack specific DNA binding domains and are not specifically directed to certain genes. Thus, transcription factors are necessary to target transcriptional regulatory events to specific DNA sequences, sometimes binding to long non-coding RNAs in these events [40]. They also mediate environmental influences on gene activity and maintain cellular memory in a sequence-specific way. In addition, transcription factors are involved in the initial stages of X chromosome inactivation and imprinting, which is then maintained by DNA methylation [12]. Adrian Bird believes that this close interaction between epigenetic marks and genetics is dissolving the distinctiveness of epigenetics. "And I think that’s a good thing” [41].
3.2 The function of DNA methylation is controversial
The role of DNA methylation and histone modification in the biochemical events that regulate genes is still not clearly established and is controversial. There are groups of organisms such as nematodes and certain insects, such as Drosophila, which do not methylate their genomes. It is undisputed that DNA methylation does not silence active promoters of genes, but affects genes that are already silent [42,43]. It is also generally accepted that one of the major functions of DNA cytosine methylation is its crucial involvement in processes such as transposon silencing, imprinting, and X chromosome inactivation [44-47]. Here, too, DNA sequence-specific factors such as transcription factors or RNAs target the methyltransferases to the respective parts of the genome.
Several authors suggest that the cytosine methylation of repeated DNA sequences and transposons presents a genomic defense system [44, 46]. Timothy Bestor and his co-authors made it clear that despite many correlations between transcriptional activation and demethylation, causation has not been demonstrated, and the available data do not support "the existence of a biochemical system that regulates embryogenesis by programmed methylation and demethylation of regulatory sequences." They also hold that, "to date there is no reasonable proof of the existence of a complex biochemical system that activates and represses genes via reversible DNA methylation" [46]. The authors criticize the lack of robust criteria in the studies purporting that genes are regulated "by dynamic programmed DNA methylation and demethylation during development."
Bestor et al. suggest that "mammalian genomic methylation patterns represent an evolutionary adaptation of a genome defense system that endows genomes with the ability to inactivate specific genomic regions in a self-perpetuating manner which is essentially irreversible over the lifespan of the organism." They agree with Ptashne and Greally [39] that gene activation and repression during development are controlled by well-established and conserved protein - and RNA - based mechanisms. Thus, DNA methylation, emphasized most strongly by social epigeneticists as mechanisms of gene regulation, does not appear to play a role in switching genes on and off.
3.3 The environment has no lasting impact on the change of epigenetic marks
According to Adrian Bird, there is no hard data on the influence of the environment on the human "epigenome" [41]. The response to environmental signals is usually mediated by specific proteins such as transcription factors or, in the terminology of Mark Ptashne, recruiters [48].
DNA methylation patterns, once established within a cell by transcription factors, can be replicated and transmitted to daughter cells by the DNMT1 enzyme independently of transcription factors; the same may be true with certain histone marks. This opens the possibility that the methylome may be affected directly by the environment, for example by a severe shortage of enzymatic co-factors such as methyl donors required by the methyltransferases, or by the presence of enzymatic inhibitors such as 2-hydroxyglutarate, which inhibits demethylases. Social epigeneticists may have used this fact for their reasoning, but they have ignored all other facts and contexts, in particular that in almost all cases the environment acts on the phenotype through transcriptional regulation and cellular differentiation. Much of stability and cellular memory is based on gene regulatory networks involving feedback loops [I am grateful to an anonymous reviewer for this information].
According to Edith Heard and Robert Martienssen, epigenetic variation can respond to the environment, but this does not mean that it has any impact on adaptive fitness. Thus, in Drosophila, heat shock or osmotic stress-induced white gene repression can be maternally and paternally inherited for several generations, but then returns to the normal state. In the Agouti mouse, mothers can modulate the coat color of their progeny through a specific diet of methyl donors, but this effect gets lost by the third generation, indicating that the influence of diet is not stable or truly transgenerational [49].
Organisms respond to the environment through the interaction of many factors, most notably specific DNA binding proteins. In yeast, it was shown that environmental stress such as heat, oxidation, acidity, or starvation, affects various genes in different ways, i.e. the response is DNA sequence-specific, with transcription factors and a multiprotein chromatin modifying complex upregulating stress-sensitive genes in response to the stressors [50]. This multiprotein complex is evolutionary conserved; in yeast it acetylates and deubiquitinates histones [51].
Some studies point to changes of cell fate decisions as a response to the deficiency of micronutrients, or to endocrine disruptors in mice through transcription factors. Endocrine-disrupting chemicals modify the function of the normal endocrine system and represent a major area of interest in epigenetics research [12]. In a well-studied case, mice that were exposed in utero to certain chemicals (of the organotin family, members of which are used as pesticides) accumulated fat from birth to adulthood. These phenotypic effects appeared to be mediated by receptors that cause mesenchymal stem cells to differentiate preferentially into the adipocyte (fat cell) lineage. This means that they do not require the reprogramming of a specific cell type [12].
Testing the hypothesis that victimization of young people across childhood and adolescence is associated with DNA methylation, Marzi et al. showed that such analyses suffered from severe methodological flaws (they were confounded by tobacco smoking and/or did not survive co-twin control tests) [52]. Analyses of six candidate genes in the stress response (NR3C1, FKBP5, BDNF, AVP, CRHR1, SLC6A4) did not reveal predicted associations with DNA methylation. Concluding that their epidemiological analysis of epigenetic effects of early-life stress did not support the hypothesis of robust changes in DNA methylation in victimized young people, the authors recommended that "we need to come to terms with the possibility that epigenetic epidemiology is not yet well matched to experimental, nonhuman models in uncovering the biological embedding of stress" [52]. Heeding this advice would greatly reduce confusion about epigenetic marks.
A recent study showed that, indeed, changes in smoking behaviors were linked to changes in DNA methylation that were dependent on the stimulus across the human genome but independent of genetic and environmental risk factors, as data from twins discordant for smoking behavior did not match [53]. Based on these findings and pointing to methodical problems in social epigenomic studies in general, such as their low statistical power, some sociologists recommend to rely instead on genomic methodology: "With the advent and growing robustness of genomic methodologies, sociologists are in an enviable position to adopt these tools and integrate them into their research" [54].
3.4 There is no evidence for transgenerational epigenetic inheritance in humans
- Epigenetic inheritance in plants and nematodes
Most scientists reviewed in this article agree about the existence of transgenerational inheritance of acquired traits through RNA in nematodes and through methylation in plants. But proof that transgenerational inheritance has an epigenetic basis in mammals is rare [49].
Transgenerational epigenetic inheritance of unclear function is common in plants. To date, there is no evidence that the inherited traits are adaptive. Epigenetic inheritance in plants is usually associated with transposable elements, viruses, or transgenes and might be, as was suggested for mammals, a byproduct of germline defense strategies [49]. In recent years, a new political movement, which is accompanied by growing sympathy for Stalin, has invoked epigenetics to rehabilitate the flawed experiments on vernalization by agronomist Trofim Lysenko, a protege of Stalin [55]. Vernalization, the influence of temperature and season on the flowering time of plants, was discovered by the German botanist Gustav Gassner in 1918 and then widely applied by Lysenko, who claimed that the effects of vernalization were inherited [56]. The lack of scientific rigour in his work has been analyzed elsewhere, as have the devastating political and economic consequences of Lysenko's practices (see e.g. [55,57].
It has been shown that vernalization that occurs after prolonged periods of cold, results in epigenetic silencing of a floral repressor in a complicated process that involves two protein complexes and methylation. But in contrast to the claims by the new pro-Lysenko movement, the memory of vernalization is not retained in the next generation, because it is robustly reset in the germline and early embryo [49].
Transgenerational epigenetic inheritance has been most reliably demonstrated by many researchers in the nematode C. elegans, where small RNAs can enter the germline and mediate heritable transcriptional silencing in subsequent generations (nematodes do not methylate their genomes). An example is the transgenerational inheritance over many generations of small interfering RNAs that target genes that are relevant for the worm's chemotaxis, nutrition, or virus genome silencing [58-60]. In these studies, Oded Rechavi and his co-authors envisage - but so far are unable to show - that the mechanisms they discovered might provide adaptive advantages for the worm. The mechanisms are gene-based and thus subject to natural selection. These genes, which are "essential for this multigenerational effect" of the transmission of RNAs, target other genes with roles in nutrition [59]. The small RNAs are transcribed and, unlike methyl groups, contain genetic information. For this reason, and because of the hitherto lack of their proven adaptiveness, Rechavi et al.'s statement that "our results therefore support the Lamarckian concept of the inheritance of an acquired trait" [60] is not appropriate. Transgenerational inheritance of acquired traits does not have to be Lamarckian, i.e. adaptive and evolutionary meaningful. Results in nematodes cannot easily be applied to humans. Nematodes have a very short generation time and unlike higher animals possess RNA-dependent RNA polymerases that can copy small RNA molecules for many generations. In addition, unlike in C. elegans, most of the alleged transgenerationally inherited traits in humans, such as the effects of starvation, are detrimental.
- The lack of evidence for transgenerational epigenetic inheritance in humans and its rare occurrence in other mammals
Many of the potential examples of epigenetic inheritance that have been proposed for humans, concern inter- rather than transgenerational effects and rarely exclude DNA sequence changes as the underlying cause for heritability [49, 61]. Parental or intergenerational effects occur when the uterus is exposed to toxins, viruses (such as Rubella), detrimental nutritional, or hormonal environments that directly affect the developing embryo and its germline. This exposure usually impacts the first generation, but occasionally also grandchildren. In contrast, transgenerational effects relate to generations that were not exposed to the initial environmental trigger, i.e. to great-grandchildren and beyond.
Intergenerational effects occur in humans and other mammals, but there are two rounds of efficient reprogramming and erasure of DNA methylation in the development of totipotent cells in the early embryo as well as during germ cell differentiation. It is widely believed that this reprogramming prevents the inheritance of most of the epigenetic marks, though some gene loci escape it. Some researchers attribute an evolutionary meaning to it: "Evolution appears to have gone to great lengths to ensure the efficient undoing of any potentially deleterious bookmarking that a parent’s lifetime experience may have imposed," and they conclude that "although much attention has been drawn to the potential implications of transgenerational inheritance for human health, so far there is little support" [49].
More recently, John Edwards et al. have demonstrated that the dynamics of demethylation and remethylation during early development are more complex than previously assumed [47]. They showed that only sequences that appear to have little evidence of biological function, such as old and inactive transposon remnants, satellite and other repeated DNA, undergo the double wave of demethylation and remethylation. In contrast, other sequences, such as the large majority of CpG island promoters are not subject to these waves of methylation and demethylation because they are unmethylated at all stages. The sex-specific methylation at imprinting control regions are demethylated only in the first round, whereas the small population of young, CpG-rich transposons largely escapes both rounds of demethylation.
The authors showed, moreover, that genomic methylation patterns at regulatory sequences are essentially static during development, and that the demethylation of promoters upon transcriptional activation is likely a consequence rather than a cause of the activation. Citing evidence that only about 10% of the mammalian genome is functional and that among the primary biological functions of DNA methylation are the heritable transcriptional repression of retrotransposons and X chromosome inactivation in female cells, the authors hold that "most DNA methylation is also likely to be without significant biological function" [47].
According Bernhard Horsthemke, the majority of studies that claim to have demonstrated transgenerational epigenetic inheritance through DNA methylation or sperm RNA - studies that showed responses to environmental metabolic factors (high-fat diet, obesity, diabetes, undernourishment, and trauma) in mice and rats - still await independent confirmation [61]. It is very difficult, Horsthemke says, to provide conclusive proof for transgenerational epigenetic inheritance in mammals, especially humans, because its study is confounded by genetic inheritance, and the impacts of ecology and culture. Some studies, such as those on the transgenerational effects of endocrine disruptors and high-fat diet on the DNA methylome have been challenged by others.
A key study about the allegedly long-lasting effects of endocrine disruptors reported that the exposure of pregnant female rats to the endocrine disruptor vincozolin affected male fertility in subsequent generations and that it was associated with epigenetic changes in the germline [62]. Emma Whitelaw drew attention to studies that refuted such claims [63]. A meanwhile widely cited study by Iqbal et al. showed conclusively that these epigenetic changes are corrected by germline reprogramming events in the next generation [64]. According to Whitelaw, the evidence of epigenetic effects lasting for more than one generation as purported in studies on transgenerational effects of the Dutch hunger winter and of PTSD after the world trade center attacks [65,66] has been inconclusive. She adds the disquieting observation that studies refuting this idea are mainly absent from the literature: "It is very difficult to publish negative results, no matter how important those negative results might be." As a result, the positive studies "seem to be uncontested to those outside the field" [63].
According to Horsthemke, the increased incidence of cardiovascular and metabolic diseases in the adult offspring of pregnant women who were affected by severe undernourishment during the Dutch "Hongerwinter" was not caused by the transmission of epigenetic information through the maternal germline, but a direct consequence of the exposure in the uterus [61]. He cites studies showing that abnormal DNA methylation patterns can be the result of a mutation in a neighboring gene that affects abnormal promoter methylation in that gene. Since it is dependent on DNA sequence, the transmission of this methylation pattern into the next generation is not an example of transgenerational epigenetic inheritance.
Mayumi Iwasakia, and Jerzy Paszkowskia argue that the prospect of environmental factors, including stress and maternal care, being inherited via epigenetic changes and influencing subsequent generations "are as intriguing as they are troubling, since it is possible to imagine that accumulation of stress memories over several generations could make life decisions difficult" [67]. Investigating the release of detrimental epigenetically suppressed transposons through abiotic stress, they found a mechanism that renders this activation only transient by rapidly resetting stress induced epigenetic states, thus erasing “epigenetic stress memory” and therefore preventing their mitotic propagation and transgenerational inheritance. They showed that this mechanism is conserved between plants and mammals.
- Methodological problems of epigenome-wide association studies
Epigenome-wide association studies (EWAS) i.e. studies of the changes of DNA methylation in individual genomes or genomes of populations, are widely used to investigate whether DNA methylation changes can be linked to the correlation of disease phenotypes with environmental exposures, in particular those occurring a long time before the phenotype emerged. Statistical problems and the problems of the irreproducibility of such studies are not dealt with here. Instead, this section illuminates the hitherto unresolved problems regarding the interpretation of EWAS, as determined by Tuuli Lappalainen and John Greally [12]. Focusing on the interpretability of even clearly-demonstrated DNA methylation changes, the authors reveal multiple problems, including the following:
- "The often vague definitions and terminologies" that are used when discussing epigenetics;
- The fact that DNA methylation can change "in response to a diverse range of influences," among them the presence of systematic differences in cell-subtype proportions between the groups tested.
- The fact that a large proportion of differences of DNA methylation between individuals can be attributed to DNA sequence. A study by Gertz et al. of a three-generation family and unrelated individuals showed that DNA sequence accounted for up to 80% of the DNA methylation variability [68]. According to the authors, "the majority of variation in DNA methylation can be explained by genotype," whose influence on patterns of DNA methylation "greatly exceeds the influence of imprinting on genome-wide methylation patterns." They conclude that the genotype will need to be taken into account when assessing DNA methylation in the context of disease.
- Reverse causation, i.e. the change of DNA methylation as a consequence of transcription, reflecting rather than causing the differences in gene expression. This has been observed in a number of cases (e.g. [69]; and in general, genetic and epigenetic factors are closely interlaced (see section 3.1). Lappalainen and Greally conclude that the fact that "many EWAS do not measure or account for genetic effects on DNA methylation," is one of the reasons for the current problems of interpretability of EWAS studies [12].
Statistical flaws of studies claiming to have demonstrated transgenerational epigenetic inheritance in humans are indicated by Kevin Mitchell, who points for example to noise being interpreted as evidence, or to the justification of sweeping general claims of transgenerational epigenetic effects by tiny statistical differences [70]. Statistical problems are not examined in this article.