Intergenerational Effects of Maternal Holocaust Exposure on FKBP5 Methylation

The results reported here replicate the previous observation of reduced methylation at a CpG site in an intronic enhancer of the FKBP5 gene in adult offspring of Holocaust survivors (5). Replication of this finding in offspring studied years apart, using DNA extracted from whole blood and peripheral blood mononuclear cells, with the use of similar pyrosequencing methods across three laboratories, increases confidence in the finding. Offspring age, childhood trauma severity, psychiatric diagnoses, suicidality, parental PTSD, psychotropic medication, FKBP5 genotype, or differential immune cell count did not account for the decrement in site 6 methylation. Rather, reduced FKBP5 methylation was associated with maternal Holocaust exposure and was particularly evident in offspring of mothers exposed as children. While the effect for the comparison between Holocaust offspring and control subjects was modest, the effect comparing offspring with maternal childhood exposure and control subjects was large.

An effect in a single area on a functionally relevant gene in peripheral blood may serve as a sentinel for significant functional effects, particularly if the observed change is reliable and correlated with changes in methylation of that gene in other important target tissues. The present finding builds on translational work demonstrating correlations of glucocorticoid-induced changes in FKBP5 enhancer methylation between rodent peripheral blood and hippocampus (22). In human tissue, DNA methylation of intron 7, site 6 was shown to be affected by glucocorticoids in peripheral blood cells and in a multipotent hippocampal progenitor cell line (7). Stable DNA demethylation of FKBP5 intron 7 in response to glucocorticoids in human hippocampal cells was present during early but not later phases of cellular proliferation and differentiation (7). In these models, FKBP5 intron 7 methylation was correlated with glucocorticoid-induced FKBP5 expression and glucocorticoid receptor sensitivity as determined in an ex vivo lymphocyte model (similar to results shown in Table 3), demonstrating the functional relevance of a methylation change in this intronic region; FKBP5 methylation was not, however, correlated with basal mRNA expression, as shown in this study and described elsewhere (23).

When bound to the glucocorticoid receptor protein complex, FKBP5 reduces ligand binding, nuclear translocation, and functional sensitivity of the glucocorticoid receptor (7, 9). The significant correlations of FKBP5 methylation in peripheral blood with endocrine measures imply that small effects observed in blood cells may reflect functionally significant alterations in other stress responsive systems. Moreover, small differences in intron 7 DNA methylation in peripheral blood have been associated with structural and functional brain alterations (7, 24). Methylation at CpG sites within the glucocorticoid response elements of intron 7 was intercorrelated and likely functions as a unit—with site 6 best representing the unit—a finding previously demonstrated in a reporter gene assay (7). Indeed, correlations between endocrine measures and mean methylation of the intron were more robust than those with site 6 methylation.

In the present study, as in the pilot (5), the effect of maternal Holocaust exposure on site 6 methylation was independent of genotype and was not associated with trauma-related psychopathology. Rather, lower site 6 methylation was associated with diminished self-reported anxiety symptoms, suggesting the possibility of a protective effect. Offspring of mothers exposed in childhood showed the greatest reductions in methylation but also the least psychopathology. Reduced intronic and site 6 methylation in offspring contrasts with increased site 6 methylation observed in Holocaust survivors (although, interestingly, parental and offspring methylation at site 6 were positively correlated) (5).

FKBP5 intron 7 methylation was not correlated with methylation of the NR3C1 gene, although both were associated with slightly different HPA axis measures, suggesting that methylation in genes that are known to interact may be independently regulated and associated with distinct aspects of trauma exposure or its sequelae (25). For instance, past work has shown that glucocorticoid receptor 1F promoter methylation predicted response to psychotherapy for PTSD, while posttreatment FKBP5 promoter methylation was associated with treatment-induced symptom change (11). Moreover, glucocorticoid receptor methylation has been shown to be differentially associated with maternal and paternal PTSD, a finding linked with increased trait anxiety, depression, and risk for PTSD (4).

Within FKBP5 intron 7, different CpG sites appear to be subject to distinct influences, as indicated by the lack of association between site 6 methylation and childhood trauma, whereas methylation of other intron 7 sites (particularly site 3) has been shown in this and other studies (5, 7, 24, 26) to interact with the rs1360780 genotype to increase risk for the development of PTSD in the presence of childhood trauma. Only one single-nucleotide polymorphism was measured in the present investigation; however, other genetic influences on site 6 methylation may be detected in future studies using genome-wide approaches. That NR3C1 and FKBP5 methylation relate to distinct clinical correlates suggests a complex interaction between the two systems that could be further explored to better understand sources of variation in molecular and endocrine phenotypes.

The finding of decreased FKBP5 methylation with earlier maternal age at Holocaust exposure is consistent with a previous observation that maternal age at exposure and PTSD were independently associated with reduced urinary cortisol levels in adult offspring (27). Offspring born to mothers with childhood exposure showed elevated 11-β-hydroxysteroid dehydrogenase (11-β-HSD-2) activity (28), a finding directionally opposite to that in Holocaust survivors (29). Survivors who were younger at age at exposure demonstrated the lowest 11-β-HSD-2 enzyme activity (29), but their adult children had the highest levels of urinary 11-β-HSD-2 activity (28). Interestingly, methylation of both FKBP5 and HSD11B2 is regulated in the placenta to moderate embryonic exposure to maternal glucocorticoids (6).

A maternal effect may be attributable to in utero glucocorticoid exposure (2, 30–32) and reflect preconception molecular effects on gametes or altered placental neuroendocrine stress physiology, with effects on fetal brain development in association with maternal childhood adversity (30, 33). Theoretically, trauma occurring prior to puberty may precipitate changes to the oocyte that are maintained throughout embryogenesis and/or are reestablished after conception, potentially influencing the intrauterine environment (2, 34). Before puberty, oocytes are still in a haploid demethylated state and are vulnerable to environmental perturbations (35). However, we know of no studies that have examined the possibility of epigenetic transmission through oocytes in humans or animals. That early maternal age at trauma exposure is a significant determinant of an intergenerational effect on site 6 (and intron 7) methylation may also imply a differential effect on subsequent interpersonal interactions between mother and child, including parental bonding and attachment (36). It is reasonable to infer that maternal age at exposure may have influenced child-rearing practices, a possibility suggested by the finding of reduced psychopathology among offspring raised by mothers exposed in childhood. Moreover, parental care and overprotection ratings differed from control subjects only for offspring whose mothers were exposed later in life. These data underscore the complexity of assigning specific mechanisms for the observed effects, because even if an epigenetic alteration involves a postconception change to germ cells or in utero perturbations, the impact of child rearing, attachment, and other postnatal environmental characteristics may be the more important drivers of change.

The lack of a finding specific to paternal exposure should not be interpreted to reflect a lack of epigenetic effects of paternal trauma. Indeed, several animal and human studies suggest that there are potent effects on offspring transmitted through sperm (26, 31, 34). It is also not possible to know, when both parents have been exposed to severe trauma at slightly different ages, what the synergistic effects of that exposure might be on their offspring.

Data from the present cross-sectional study in adult offspring cannot speak to the mechanisms through which epigenetic alterations may have been acquired. It is nonetheless interesting to speculate whether the findings help explain increased offspring vulnerability to psychopathology or reflect an adaptation to optimize offspring preparedness or response to adversity in their own lives (37). Disinhibited FKBP5 expression might promote caution or vigilance in an adverse environment, behaviors elicited by increased FKBP5 expression in animal experiments (38). Consistent with these behaviors, reduced methylation of FKBP5 intron 7 was associated with diminished anxiety following exposure-based psychotherapy (39). The significant associations of increased anxiety with higher FKBP5 mRNA levels, and of decreased anxiety with lowered FKBP5 methylation, would be consistent with such adaptive effects. The observation that lower site 6 and mean intron 7 FKBP5 methylation was associated with reduced glucocorticoid sensitivity and increased basal cortisol levels suggests that individuals with lower FKBP5 methylation display fewer PTSD-like endocrine alterations, which have been associated with elevated risk for psychopathology (8, 27). Certainly, genome-wide studies may identify broader functional epigenetic and transcriptomic profiles associated with intergenerational trauma and risk or resilience in offspring, and they would provide an important context for the present findings.

The potential functional role of any single observed epigenetic alteration requires further study, and the regulation of any specific gene transcript requires knowledge of other genes, epigenetic marks, microRNAs, and additional regulators. This study is limited by the lack of genome-wide data, the single cross-sectional observation, and the uneven sample sizes for the Holocaust offspring and comparison groups. A further limitation is that although a strong signal was detected for maternal childhood exposure, the number of participants with only fathers exposed may have been too small to test the contribution of the paternal germline to the present findings.

Nonetheless, the previous observation of an association of parental Holocaust exposure with FKBP5 DNA methylation in adult offspring has been replicated in a larger cohort. Previous work is extended by the new observation that effects on offspring FKBP5 intron 7 methylation and FKBP5 gene expression are associated with maternal exposure and, for the former, with maternal childhood trauma exposure. Our study cannot distinguish determinants of the observed effects, including a preconception effect on germ cells, differences in the in utero milieu, alterations in postnatal care, or other influences (3, 32). However, the relative stability of the observation across cohorts, in adults born years or sometimes decades after parental exposure, provides an impetus for basic science studies in this field. Prospective intergenerational studies are needed to elucidate the mechanisms for such trauma effects, as well as their functional consequences for vulnerability, adaptation, and resilience.

The authors acknowledge Jessica Johnson, B.S., for performing rs1360780 genotyping.