We found that healthy adults reporting a history of childhood adversity (loss of a parent, maltreatment, poor quality parenting) had increased cytosine methylation of the promoter region of the GR gene NR3C1
. In addition, methylation of this region was associated with attenuated cortisol responses to the Dex/CRH test in adulthood. This is the first study to demonstrate that variation in human parenting experiences is linked to epigenetic modulation of the leukocyte GR gene. Our results are consistent with findings of the two prior human studies (using hippocampal tissue and infant cord blood, respectively) of epigenetic modulation of this gene in association with childhood maltreatment 
and prenatal exposure to maternal depressed mood 
. In rodents, low levels of maternal care during early postnatal life are associated with increased cytosine methylation and decreased expression of the hippocampal GR (NR3C1
) promoter 
. Given previous findings on childhood maltreatment and loss on risk for psychopathology, the effect of promoter methylation on GR expression and glucocorticoid responses, and the role of HPA axis dysfunction in depressive and anxiety disorders, the present findings suggest that epigenetic changes to this gene could be a key mechanism for effects of childhood adversity on risk for these disorders.
The transcription factor NGFI-A binds at CpG 3 and 4, and we found that CpG 3 methylation was associated with loss of a parent and maltreatment in childhood. Methylation at CpG 4 was not significantly linked to these adversity measures. Our findings extend the observations of Oberlander and colleagues 
, who found that cytosine methylation at CpG 3 was associated with exposure to maternal depression in the third trimester, and with increased infant cortisol responses. Methylation at this site may alter transcription; methylation at the NGFI-A binding site in hippocampus has been shown to reduce GR expression in rodents 
and humans 
. However, this site was not methylated in mice treated with oral corticosterone 
or rat pups subjected to maternal separation 
. In humans, one study found that the NGFI-A binding site was unmethylated in post mortem hippocampal tissue from 32 individuals, most with Parkinson's or Alzheimer's disease, but no information on early environmental stressors was available in this study 
. A recent examination of post-mortem brain from donors with major depression (but without documented child abuse) found reduced expression of hippocampal NR3C1
exon 1F 
. However, promoter 1F was uniformly unmethylated and instead, NGFI-A was downregulated. This suggests that while altered GR transcription may be common to childhood abuse and major depression without early adversity, the mechanism may differ in the two conditions.
Our findings of greater methylation at CpG 1 in association with low Parental Care and with loss of a parent in childhood also extend the findings of Oberlander and colleagues 
of increased infant methylation of CpG 1 in association with maternal depressed mood in the third trimester. The additive Adversity Index was also positively correlated with both CpG 1 and 3, indicating that a greater number of different adverse childhood experiences is associated with more methylation at NR3C1. Our findings appear to be specific to early environment stress and are not accounted for by current psychopathology or subsyndromal symptoms.
None of our adversity measures was significantly associated with methylation at CpG 2 or the exploratory CpG sites (although there was a trend for low Parental Care to be associated with greater methylation at CpG 5). Cortisol response to the Dex/CRH test showed negative associations with CpG site 2 and the exploratory sites CpG 5 and the average of sites 7–13. That we found attenuated
cortisol responses contrasts with findings of enhanced
cortisol responses in rodents with increased NR3C1
promoter methylation and decreased receptor expression 
, as well as with increased cortisol responses in infants with greater cytosine methylation at CpG 3 
. It therefore seems unlikely that our cortisol findings are a direct effect of methylation-induced decreases in GR expression. Our results are consistent, however, with prior reports of blunted cortisol responses in association with early-life stress from our group 
and others (e.g., 
). The mechanism of this effect is unknown, but may involve a trajectory of initial hyper-activation of the HPA system in response to excessive and prolonged stress exposure progressing to a counter-regulatory adaptation state of chronic adrenal stress hyporeactivity 
. This may be influenced by the developmental timing or chronicity of early-life stress; unfortunately we do not have these measures in the present study. We do not have gene expression data in this sample, and our measure of HPA axis function was limited to the Dex/CRH test. This test is a sensitive measure of HPA axis dysfunction, but because it combines a large dose of dexamethasone with corticotropin-releasing hormone administration, it does not permit assessment of feedback sensitivity or other specific components of the axis that may explain the findings. Possible mechanisms include compensatory down-regulation of corticotropin-releasing hormone or adrenocorticotropic hormone receptors, or increases in GR negative feedback sensitivity, in response to or in addition to changes in GR expression.
Exposure to stress and trauma are increasingly thought to underlie the neuroendocrine abnormalities seen in some patients with depressive and anxiety disorders. Abnormal HPA axis activity may play a central role in the pathogenesis of depressive and anxiety disorders, perhaps in those with excessive stress exposure. Preclinical studies show that chronic stress and glucocorticoid administration result in hippocampal atrophy 
, which has also been documented in adults with a history of childhood maltreatment, major depression, or post-traumatic stress disorder 
. In contrast, stress and glucocorticoids can produce excitability and dendritic remodeling in the amygdala, which mediates anxiety and fear responding 
. Thus, alterations in HPA axis activation in response to early deprivation may alter the structure and activity of brain regions involved in mood and anxiety disorders. We excluded individuals with current or past Axis I psychiatric disorders in the current study. This allowed us to isolate the effects of early adversity, but it is important to note that this exclusion likely resulted in a sample of individuals who are especially resilient to psychopathology, in addition to those who are at risk but have not yet developed a disorder.
In this study we examined the leukocyte GR gene, whereas the previous human investigations of the effect of exposures examined cord blood or hippocampal tissue. The similarity of our findings to those of the prior human studies, as well as animal work on the hippocampal GR in relation to variation in maternal care, is encouraging, but data regarding the differential control of expression of the GR in various tissues are very limited. Some evidence indicates similarities in the regulation of this gene in brain and peripheral tissues. Lee and colleagues 
found that chronic oral corticosterone administration caused anxiety-like behavior and a decrease in hippocampal and blood mRNA levels of NR3C1
. Moreover, the relevance of regulation of leukocyte glucocorticoid receptor expression for stress-related disorders is supported by evidence of abnormal sensitivity and expression of this receptor in leukocytes of patients with major depression and post-traumatic stress disorder 
Our findings of age-related alterations to the extent of methylation are consistent with other reports that identify gene-specific associations between methylation and aging 
. In addition, in this sample, older subjects reported higher levels of Maltreatment and lower Parental Care than younger participants (and age was controlled in the analyses). This may be a cohort effect, but it is possible that it is due to an age-related reporting or memory bias. We studied several related childhood experiences that commonly co-occur and may have similar or redundant epigenetic effects. It is notable that the Adversity Index, the sum of the various childhood adversities, was positively associated with the degree of NR3C1
methylation. Childhood parental loss, which represents a more objective event than some other types of experiences and is less likely to co-occur with maltreatment than other forms of adversity, was the strongest predictor of CpG methylation (and was not associated with age). It should be noted that the effects of childhood parental loss may be mitigated by personal and familial characteristics that influence adaptation to loss (e.g., 
Because we studied a mixed population of leukocytes, our findings could be influenced by the relative composition of these cells, which can be altered during infection and other conditions. Glucocorticoids have strong anti-inflammatory effects, and glucocorticoid receptors are present in several leukocyte cell types, including lymphocytes, neutrophils, eosinophils, and macrophages 
. However, it is unlikely that inflammatory conditions explain our findings because 1) we excluded subjects with acute or chronic illness, allergy symptoms, abnormal blood counts, or the use of antibiotics, antihistamines, or corticosteroids, and 2) any such confounding effect would have to occur differentially in those with adversity.
In summary, we found that early-life stress, in the form of loss of a parent during childhood, maltreatment, and low parental care, was associated with epigenetic changes to the promoter region of the glucocorticoid receptor gene. In addition, methylation of the promoter region of this gene was linked to alterations in HPA axis function. These findings, together with previous research in rodents and two prior studies in humans, provide preliminary support for the hypothesis that altered expression of the glucocorticoid receptor due to cytosine methylation of the gene promoter could be a mechanism of the neuroendocrine effects of early-life stress, and could predispose to the development of major depression and post-traumatic stress disorder. However, our effect sizes tended to be modest in magnitude, different CpG sites were associated with the associations of childhood adversity measures and the cortisol response to the Dex/CRH test, and we did not have gene expression data available. This study will need to be replicated in order to draw firm conclusions about the findings. Further work is also needed to determine whether these findings are specific to lymphocytes and whether this reflects changes in central regulation of the glucocorticoid receptor in brain regions involved in stress responses and mood and anxiety disorders.