Longevity gene klotho (KL) has been linked to multiple biomarkers of cellular aging and adverse health outcomes in both preclinical and clinical studies [1,2,3]. Of particular interest are studies associating KL genotype, expression, and protein levels to phenotypes such as Alzheimer’s disease [4], Alzheimer’s related biomarkers (e.g., amyloid-beta; [5,6,7]), memory and cognitive performance [8,9,10,11,12], prefrontal cortical brain volume and neural connectivity [13, 14], hippocampal synaptic plasticity and neurogenesis [8, 9, 15], and axonal myelination [16, 17]. Evidence suggests that KL expression serves a protective function against age-related cognitive decline, with effects most evident among older individuals [11, 15, 18], though null effects have also been reported (e.g., [19]).

A separate line of research suggests that traumatic stress is associated with various biomarkers of cellular aging, including advanced epigenetic age, also known as DNA methylation (DNAm) age, relative to chronological age [20,21,22,23,24]. That said, these findings also have not been uniform across studies [25, 26]. Recently, in a sample of veterans with a high prevalence of posttraumatic stress disorder (PTSD), we found that the association between PTSD and several biomarkers of accelerated cellular aging was dependent on the presence of the minor allele at a single nucleotide polymorphism (SNP) in the KL gene [27]. Specifically, we examined 43 common SNPs in the KL gene and found that carriers of the rs9315202 minor allele (A) evidenced a strong association between PTSD symptom severity and advanced epigenetic age in blood (per the Horvath index), increased peripheral inflammation (C-reactive protein), and decreased white matter integrity in right-lateralized tracts connecting prefrontal to limbic regions. These effects were accentuated in subjects who were aged 30 years or greater (the median age in that sample of veterans returning from the Global War on Terror), consistent with other research suggesting that the effects of KL are more evident with increasing age [11, 15, 18].

Though numerous studies have found evidence of associations between traumatic stress and advanced DNAm age measured in blood (e.g., [22]), no study to date has examined associations between traumatic stress and epigenetic aging in brain tissue. Studies of advanced epigenetic age in postmortem brain tissue in association with other psychiatric conditions have yielded mixed results. Specifically, significant associations were reported between a novel epigenetic age acceleration measure and major depression in prefrontal and subgenual regions (n = 141; [28]), but not between the Horvath index and depression in prefrontal cortex (two samples of n = 40 and 58; [29]). Null effects were reported for bipolar disorder with Horvath-derived estimates (n = 32 cases and 32 controls; [30]), though there was evidence of bipolar-related advanced DNAm age in hippocampal tissue among those >45 years of age). Studies of schizophrenia [31, 32] and alcohol-use disorders [33] reported no significant associations in frontal and/or temporal regions. Finally, heroin-use was associated with slowed epigenetic age in orbitofrontal cortex relative to controls and non-heroin-related suicide cases [34]. The equivocal pattern of findings across studies may be related to sample size, variability in brain regions evaluated, lack of consideration of genetic (such as KL) and demographic (such as age) modifiers of these associations, and/or differences arising from the DNAm age algorithm or psychiatric diagnosis under evaluation. The varied results of these studies highlight the need for further evaluation of stress-related epigenetic age acceleration in brain tissue.

Aims and hypotheses

Based on the foregoing, we advanced the following aims and hypotheses.

Aim 1

The first aim was to test the main and interactive effects of PTSD and the KL SNP rs9315202 on advanced epigenetic age in postmortem brain tissue obtained from the VA National PTSD Brain Bank. Our primary hypothesis was that rs9315202 would moderate the association between PTSD and epigenetic age (relative to chronological age) in older decedents in the brain bank, consistent with the pattern of results we previously observed in blood and neuroimaging data [24]. To evaluate this, we conducted regression analyses focused on this candidate SNP X PTSD as a predictor of DNAm age residuals, evaluating effects in a subset of the cohort who were age 45 years or greater (the median). Models were also examined in those below the median age and in the sample as a whole. Three brain regions were evaluated: dorsolateral prefrontal cortex (dlPFC), ventromedial prefrontal cortex (vmPFC), and motor cortex. Alterations in dlPFC and vmPFC have commonly been associated with PTSD [35, 36] and aging [37, 38]. Moreover, KL has shown particular associations with right dlPFC [13, 14], suggesting this may be a primary locus for KL-related neuropathology risk versus resilience. Motor cortex was included given evidence of notable age-related cortical thinning in this area [37] and the role of KL in motor disturbances. Specifically, animal studies suggest that genetic manipulation of KL leads to early onset of motor pathology, such as gait disturbance [2], and that alterations in motor-related cells (e.g., anterior horn and Purkinje cells) in the central nervous system are key features of KL knockout mice [39, 40].

Aim 2

The second aim was to examine the main effects of diagnoses commonly comorbid with PTSD (major depression and alcohol abuse/dependence, representing internalizing and externalizing pathology, respectively; [41]) on epigenetic age in brain tissue relative to chronological age. Based on prior research in blood [24] and brain [28], we hypothesized that PTSD and both depression and alcohol-use disorders would be associated with advanced epigenetic age.

Aim 3

The third aim was to test the hypothesis that advanced epigenetic age in brain tissue is associated with altered KL mRNA expression and that the effects of rs9315202 and PTSD on altered KL expression are indirect, via advanced epigenetic age. We used mediation analysis to evaluate if rs9315202 X PTSD was associated with altered KL expression via advanced epigenetic age in the older cohort.

Aim 4

The final aim was to examine the molecular effect of the candidate SNP, rs9315202, on KL regulation, which is thus far unknown. To do so, we cloned genomic DNA containing this SNP into a reporter system, and evaluated the effects of rs9315202 on KL gene expression. We hypothesized that the SNP would act as an enhancer that regulates KL gene expression.


PTSD brain bank

Participants and procedure

Postmortem tissue from N = 117 brains was obtained from the VA National PTSD Brain Bank housed at the VA Boston Healthcare System [42], which acquired the tissue and clinical information from the Lieber Institute for Brain Development at Johns Hopkins University [43]. Of this group, 2 were missing genotypes due to quality control concerns and were removed from analyses. A total of n = 85 were determined to be of white, non-Hispanic ancestry (per genotype in concert with medical record information) and were the primary sample evaluated in analyses involving KL genotype (due to concerns of population stratification). Analyses that did not include genotypes (i.e., evaluation of main effects of PTSD comorbidity and main effects of DNAm age residuals on KL expression) were based on the full cohort with complete data regardless of ancestry (n = 114 for methylation-only analyses and n = 92 for analyses including RNA expression data) to maximize statistical power and generalizability. Left dlPFC (Brodmann Area [BA] 9/46) was taken at the level of the genu of the corpus callosum, left vmPFC (BA 12/32) was taken at the level of the genu of the corpus callosum, and left motor cortex (BA 4) was taken at the superior central sulcus. Neuropathological exams and toxicology evaluations were performed by board-certified neuropathologists. Cause and manner of death were determined by Maryland state medical examiners.

The neuropathology examination and diagnostic assessment included medical record review and next-of-kin interviews to determine medical and psychiatric diagnoses, including administration of MINI International Neuropsychiatric Interview 6.0, the PTSD checklist for DSM-5 (adapted for postmortem studies), and the Lieber Psychological Autopsy Interview, which included interviews with mental health clinicians where possible [43]. Psychiatric diagnoses were further reviewed by at least two independent board-certified psychiatrists, with confidence in PTSD determinations rated on a 1−5 scale [44]. Diagnoses rated 3 or higher were classified as PTSD cases. From these sources, we obtained diagnoses of PTSD, major depression, and alcohol abuse or dependence. Exclusion criteria were neurodegenerative disease, history of severe traumatic brain injury (or neuropathological evidence of such), and neuritic pathology [44].

The full sample was 61.2% male with a mean age at death of 44.79 years (SD: 13.93, range: 17.35–92.25; Table 1). The median age at death was 45 years. This median point was used to a priori divide the sample into older vs. younger cohorts for age-group specific analyses.

Table 1 Demographic characteristics of the sample.

DNA, DNAm, and related statistical procedures

Procedures for DNA extraction are described in Supplementary Materials. Genotypes were determined from DNA extracted from motor cortex samples and were interrogated on the Illumina HumanOmni2.5−8 array. Genotype imputation procedures (for secondary analyses) are described in the Supplementary Materials. DNAm was assessed in the dlPFC, vmPFC, and motor cortex with the Illumina Infinium MethylationEPIC array (Supplementary Materials). The Horvath [45] DNAm age calculation was based on the 335 probes on the EPIC array that overlap those on the 450k array used to develop the algorithm [46]. Estimated percentage of neurons from each region was computed using the CETS method [47] and included as covariates to control for cellular heterogeneity (Supplementary Materials). Two sets of ancestry principal components (PCs) were calculated, each from 100,000 randomly chosen common SNPs (at least 5% minor allele frequency; see Supplementary Materials for details). The first set was developed to capture ancestral variation across all ancestries in the sample (global ancestry) and the second set was developed within the white non-Hispanic subsample (the largest in the cohort) to represent any cryptic population substructure within that cohort. The top three PCs were included in each set of analyses as covariates, as appropriate to the composition of the cohort in a given analysis.

RNA sequence data and related statistical procedures

RNA was extracted from each frozen tissue section, and library preparation conducted using the Illumina TruSeq Stranded total RNA kit with globin depletion. Libraries were then sequenced using HIseq 2500 (Supplementary Materials). The pipeline for determining gene expression included adapter removal and quality filter of reads with trimmomatic [48], alignment to the hg38 genome with STAR [49], transcript quantification with Kallisto [50, 51], and quantification of KL gene expression with the tximport Bioconductor package and log transformed using the regularized log transformation (rLog) method implemented in DEseq2 [52]. Cell types (astrocytes, endothelial cells, microglia, mural cells, neurons, oligodendrocytes, and red blood cells) were estimated using BrainInABlender and controlled for in all RNA analyses.

Data analyses

We first calculated the Horvath DNAm age estimates and compared them with age at death (i.e., chronological age) using correlation. We then regressed age at death out from the DNAm age estimates to generate “DNAm age residuals” to index a dimension of residuals reflecting advanced to slowed cellular aging for each brain region (i.e., the primary dependent variable in subsequent analyses) and compared these values across brain regions via correlation. Next, to test Aim 1, we conducted multiple linear regression analyses in the white non-Hispanic decedents (separately for each brain region). These analyses were conducted in the older cohort (where we hypothesized an effect) with multiple testing correction across the three regions (see below). We also tested for association in the younger cohort and in the full cohort for completeness, using the same multiple testing correction. In each of these models, we included demographic covariates (the first three ancestry substructure PCs, sex), methodological covariates (postmortem interval, RNA integrity [RIN] value, percentage of neurons the DNA was extracted from as estimated from the DNAm data), and PTSD diagnosis in the first step of the model. The second step added the SNP, and the third step added the interaction term for PTSD X rs9315202. The multiple testing correction used a procedure that permutes PTSD case status and adjusts the p value based on the correlational structure among the DNAm age residuals across the three brain regions (e.g., [24]). Significant results were further evaluated in separate models by including additional covariates in the model related to manner of death (death by suicide, death by substance use). Although we were focused on rs9315202 in interaction with PTSD, we wondered if other KL variants might also relate to DNAm age residuals or account for effects attributable to the candidate SNP. Therefore, we conducted follow-up regressions (parallel to the models described above) examining common variants in the gene (n = 211 variants beyond rs9315202) to determine if they showed significant interactions with PTSD on DNAm age residuals (Supplementary Materials).

To test Aim 2, we entered the main effects of several clinical variables including PTSD comorbidity (major depression and alcohol abuse/dependence diagnoses) and other lifestyle/behavioral factors (body mass index [BMI], cigarette use) that have previously been associated with epigenetic aging in blood [53, 54] in a regression in the full sample, as we did not have a priori age-related hypotheses regarding the main effects of psychopathology and lifestyle factors on DNAm age (age-specific hypotheses were only relevant to KL analyses and KL was not included in this model). We followed the same permutation procedure for multiple testing correction across the three regions as described above.

To test Aim 3, we obtained KL expression levels and examined the association between DNAm age residuals and KL expression, controlling for 7 cell types the RNA was extracted from, age at death, sex, RIN, PMI, and 3 ancestry PCs in the all-ancestry older (hypothesized) cohort. To be thorough, we also conducted the analysis in the full cohort and the younger cohort. We then conducted a follow-up mediation model (path analysis) in which we examined the indirect effects of rs9315202, PTSD, and their interaction on KL expression via DNAm age residuals in the older white non-Hispanic cohort.

Effects of rs9315202 on KL regulation

To test Aim 4 and determine the effects of rs9315202 on KL regulation, we cloned the rs9315202 containing region into the enhancer region of KL 4 kb promoter in firefly luciferase (FLuc) and NanoLuc luciferase (NLuc) coincidence reporter [55]. To do so, the 3.5 kb genomic DNA region flanking the rs9315202 variant was PCR amplified from the human kidney cell line HK-2 genomic DNA and cloned to replace the SV40 enhancer region of pNLCoI1-SV40-KL4000 using Clontech HiFi and In Fusion cloning kit according to the manufacturer’s protocol using the following primers:




5′-GGATCCTTATCGATTTTACCAC-3′ (Vector reverse).

The mutagenesis to produce rs9315202 was performed using Clontech HiFi and In Fusion cloning kit using the following primers:



Transfection was accomplished via HEK and HK-2 cells grown on poly-D-lysine-coated plates in 96, 12, or 6-well formats. Twenty-four hours after plating, cells reached 70−80% confluency and were transfected with reporter plasmid or control plasmid. Transfections were carried out using Mirus TransIT-X2 with 100 ng, 1 μg, or 2 μg of total plasmid DNA per well in 96, 12, or 6-well plates, respectively. Transfection medium was removed and replaced with fresh medium after 5 h.

For measurement of FLuc and NLuc expression, the coincidence reporter vector under the KL promoter, Nano-Glo® Dual-Luciferase® Reporter Assay System (cat. N1620, Promega) was used according to manufacturer’s instructions. The luminescence was measured using a plate reader (GloMax® Discover System, Promega).


PTSD brain bank: DNAm age estimates

The correlation between Horvath DNAm age estimates and age at death ranged from r = 0.92 to r = 0.93 (ps < 0.001) across the three brain regions. The estimates of DNAm age correlated with each other across brain regions also at r = 0.92 to 0.93 (ps < 0.001). The DNAm age residuals correlated with each other at r = 0.50 to 0.51 (ps < 0.001) across the three regions.

Aim 1: rs9315202 X PTSD as a predictor of DNAm age residuals

To test our hypothesis for Aim 1, we entered PTSD, rs9315202, and their interaction as terms in a regression predicting age residuals in each region in the older white non-Hispanic cohort. After controlling for covariates (Table 2), we found a corrected significant positive interaction between PTSD and rs9315202 on DNAm age residuals in motor cortex in the older cohort (p = 0.002, corrected p = 0.014; Table 2, Fig. 1). The positive association between PTSD status and DNAm age residuals was evident among those with the minor allele (A) but not those with the major allele (G). In follow-up analyses, we additionally controlled for manner of death covariates (death by suicide, death by substance use), and found that neither was associated with DNAm age residuals in motor cortex (p = 0.15 and 0.17, respectively), while the interaction term remained significant (p = 0.005). As expected, the effect of rs9315202 X PTSD on DNAm age residuals in the motor cortex was specific to the older subset: This effect was not observed in the younger cohort (pmotor = 0.78), nor was it evident in the combined all-ages sample (pmotor = 0.38, pdlPFC = 0.57, pvmPFC = 0.56; Table 2). We then examined additional common variants in KL (n = 211), and found that while several yielded similar or identical significance values, these were all in high or complete linkage disequilibrium (LD) with rs9315202 and none yielded meaningfully more significant interactions with PTSD (or main effects) in any region or cohort than the rs9315202 X PTSD interaction effect in the older cohort in the motor cortex (Fig. S1 and Supplementary Materials).

Table 2 Interactions between rs9315202 and PTSD on DNAm age residuals across three brain regions.
Fig. 1: PTSD is associated with advanced epigenetic age in motor cortex as a function of rs9315202.
figure 1

The figure shows mean (error bars represent +/−1.96 SE) DNAm age residuals in the motor cortex in the older cohort as a function of PTSD diagnosis and genotype at rs9315202.

Aim 2: PTSD comorbidity as a predictor of DNAm age residuals

We next examined the main effects of PTSD comorbidity and behavioral factors in association with DNAm age residuals in the full sample in each region. Results revealed a nominally significant main effect for alcohol abuse/dependence diagnoses in association with DNAm age residuals in motor cortex (p = 0.039, corrected p = 0.100; Table S1; Fig. S2). Based on these results, we ran additional analyses in which we included the aforementioned potential time-of-death confounds in the model and found that none of the covariates were significantly associated with DNAm age residuals in this region (smallest p = 0.102) while alcohol abuse/dependence remained (nominally) significantly associated (p = 0.029). We also examined the same model separately in men and women in motor cortex because our main effects for psychopathology on DNAm age residuals in our previous studies were conducted in veteran samples that were nearly entirely male. Results suggested that the nominally significant effect for alcohol abuse/dependence in the full sample was likely driven by the men as alcohol abuse/dependence was nominally associated with DNAm age residuals among the men (p = 0.021; Table S2) but not the women (p = 0.53, Table S2). The effect was still significant (p = 0.012) when the aforementioned covariates were added to the male-only model.

Aim 3: advanced epigenetic age as a predictor of KL expression

We then examined the association between DNAm age residuals in motor cortex and KL expression in the same region (i.e., the region where we found an effect for PTSD X rs9315202 on DNAm age residuals), controlling for 7 cell types the RNA was extracted from, age, sex, RIN, PMI, and 3 ancestry PCs in the older cohort (including all subjects, regardless of race as genotype was not included in the model). We found that DNAm age residuals were associated with decreased KL expression in motor cortex in the older cohort (β = −0.022, std β = −0.434, p = 0.032, Table 3, Fig. S3). We next proceeded to test our planned mediation model to examine the indirect effect of PTSD, rs9315202, and their interaction on KL expression in motor cortex via DNAm age residuals in motor cortex in the subset of older, white non-Hispanic decedents. We found significant indirect effects of the interaction term on KL expression through DNAm age residuals (indirect B = −0.116, std indirect β = −0.198, p = 0.027; Fig. S4), controlling for 3 ancestry PCs, age, sex, RIN, and PMI.Footnote 1,Footnote 2 In total, the model explained 37% of the variance in DNAm age residuals in motor cortex and 43% of the variance in KL expression in motor cortex.

Table 3 Effects of DNAm age residuals on KL expression in motor cortex.

The effect that we observed for DNAm age residuals as a predictor of KL expression in motor cortex in the older cohort was not evident when we ran the same regression model in the younger cohort (β = −0.008, std β = −0.178, p = 0.265, Table 3). The effect was significant in the combined-age sample (β = −0.010, std β = −0.216, p = 0.042, Table 3), but this was clearly driven by the older members of the cohort given that the effect was not significant in those below the median age.

Aim 4: molecular effects of rs9315202 on regulation of KL expression

The KL variant (rs9315202) is located 1734 bps downstream of the KL coding region (Fig. S5A). The region that contains the SNP of either wild type (WT, i.e. common allele) or rs9315202 mutant (i.e. minor allele) was cloned into a dual-luciferase enhancer reporter system to evaluate the effects of allelic variation at rs9315202 on KL gene regulation. In this system, Fluc and PEST-destabilized NlucP are expressed off the same KL 4 kb promoter using ribosome skipping mediated by the P2A peptide (Fig. S5B). We amplified 3.5 kb of genomic DNA sequence from HK-2 cells and cloned into the enhancer region of the FLuc and NLuc luciferase coincidence reporter containing the 4 kb or the 1.8 kb KL promoter (Fig. S5B). We examined the enhancer activities of rs9315202 minor allele and WT constructs by transfecting the dual-luciferase reporter into HK-2 cells that express KL endogenously. The results showed that single point C > T mutation (reverse strand as that reported for brain bank cohort) at the SNP position resulted in significant reduction in dual-luciferase reporter activities for both KL 4 and 1.8 K reporter systems (Fig. 2). These results demonstrate that the rs9315202 variant drives a change in KL transcriptional activities.

Fig. 2: Analysis of the effects of mutation on rs9315202 by a dual-luciferase reporter system in HK-2 cells.
figure 2

Firefly (FLUC), NLUC coincidence reporter system under control of either KL 4Kb (a) or 1.8 kb (b) promoter. The enhancer activities of WT and rs9315202 T variant (Tovar) constructs were analyzed by dual-luciferase assay.

The 3.5 kb genomic DNA from HK-2 cells was then sequenced (Fig. S6). By searching ncRNA database in using the sequence flanking SNP rs9315202, we found an ncRNA with ID HSALNG0096254 that is 1729 bases in length (Fig. S6). We refer to this ncRNA as ncRNA1729 for its length.Footnote 3 These data indicate that the ncRNA1729 within the enhancer region alters KL gene expression, suggesting that variation at rs9315202 may affect the ncRNA function in KL gene expression.


The KL gene was named for the Greek goddess, Clotho, who was said to spin the web of life and thereby determine human lifespan. Our results underscore the apt naming of the gene, as Clotho transcends mythology to promote accelerated epigenetic aging in brain tissue. Specifically, we showed that the KL SNP rs9315202 moderated the association between PTSD and advanced epigenetic age in motor cortex, such that PTSD was associated with advanced DNAm age among those with the minor frequency allele (A) at this locus. The effect was evident only in relatively older decedents in the sample, consistent with our prior results in living subjects which found that the impact of rs9315202 on the association between PTSD symptom severity and advanced epigenetic age in blood was accentuated in older adults [27]. No other KL variant (out of an additional 211 imputed genotypes) evidenced a more significant association with DNAm age residuals (alone or in interaction with PTSD) in any brain region or cohort. Rather, variants in high LD with the candidate SNP showed equivalent patterns of interaction with PTSD. These follow-up analyses bolster our decision to focus on our candidate SNP in the older cohort.

The interaction between rs9315202 and PTSD in these data was indirectly linked to reduced KL mRNA expression in motor cortex via advanced epigenetic age in this same region. This effect, too, was driven by older decedents in the sample. Reduced KL expression is thought to be a risk factor for weakened synaptic activity and cognitive decline (e.g., [9]), particularly at older ages where KL enhancement is needed to protect against age-related neurocognitive loss. Examination of the molecular effects of allelic variation at rs9315202 further supported our hypothesis that this SNP exerts regulatory effects on KL expression as the polymorphism was found to be associated with alterations in KL transcription via ncRNA. Specifically, the 3.5 kb downstream region of KL may act as an enhancer to regulate KL gene expression.

Long ncRNAs are significant in that they regulate gene expression without directly transcribing the gene. These transcripts are common throughout the genome and one of their major roles is to alter chromatin structure, which has downstream effects on gene expression [56]. This commonly occurs at DNAaseI hypersensitivity sites, which are located at open (and therefore modifiable) ends of chromatin, known for their role in transcriptional binding and gene regulation [57]. Thus, rs9315202 may exert effects on KL expression in older, stressed populations through intermediate effects on chromatin structure modifications that promote accelerated aging. This is likely to be tissue-specific given that the effect of this SNP on KL expression in the brain bank data was indirect and moderated by PTSD. Additional molecular work, such as identifying the expression patterns and cell types of ncRNA1729 in brain tissue and using CRISPR gene editing technology to study the effect of rs9315202 on KL gene regulation via ncRNA1729 in cell lines, is needed to further characterize the effects of rs9315202 on ncRNA1729 function and on KL gene regulation.

In addition to replicating the results from our prior work in blood demonstrating a PTSD X rs9315202 interaction on advanced epigenetic age, we also replicated results from a distinct longitudinal study [24] in which we found effects for alcohol abuse/dependence on an accelerated pace of epigenetic aging over time [24]. In this study, there was a nominally significant association between alcohol abuse/dependence diagnoses and advanced epigenetic age in motor cortex. Other labs have also found evidence (albeit not consistently so) of associations between alcohol misuse and advanced epigenetic age in blood [33, 58, 59] and these results suggest that this effect may be parallel across blood and brain tissue.

All the effects we observed were only in motor cortex, which was not expected given prior evidence of associations between both KL [13, 14] and PTSD [60] with dlPFC. However, motor cortex shows substantial age-related changes, such as decreased cortical thickness [37, 61]. As well, KL has known effects on age-related motor phenotypes [2], including in a mouse study of Parkinson’s disease [62] which found that KL augmentation ameliorated Parkinson’s disease pathophysiology and improved performance on motor tasks in rodents. Similarly, enhancing KL expression in a mouse model of amyotrophic lateral sclerosis, a fatal disease defined by progressive and rapid skeletal muscle wasting and paralysis, resulted in delayed disease onset, slowed disease progression, longer lifespan, and reduced expression of a number of inflammatory genes in motor cortex via reduced microglial activation, both of which are thought to be major contributors to progression in the disease [63]. The combined results highlight the importance of examining multiple brain regions in postmortem evaluation of accelerated biological aging, as differences in the pace of epigenetic aging across regions may help to explain the inconsistent pattern of results in prior studies of psychiatric disorders and epigenetic age in postmortem brain tissue.

Study limitations and strengths

These results should be interpreted in light of several study limitations. First, though our sample size in the PTSD brain bank was among the largest to date to examine associations between psychiatric conditions and epigenetic age in postmortem tissue, the sample size was, overall, small and most of our effects were in yet smaller subsets of the sample (e.g., the older cohort), which can increase the risk of false-positive (and negative) results. We attempted to mitigate this concern by testing specific hypotheses in our analyses that were based on the results of prior studies. Similarly, given the difficulty in obtaining PTSD-associated postmortem brain tissue, we did not have access to a second cohort to test for replication. We were also limited in the brain regions available for analysis and, thus, we could not examine other regions that might be of particular interest to aging, PTSD, and KL, such as the hippocampus. Only left hemisphere tissue was obtained in the brain bank, thus we could not examine potential right-lateralized effects. We also did not conduct cloning experiments outside the rs9315202 region, thus it is possible that other genotypes are also critical for KL expression in the brain. These limitations are arguably off-set by the strengths of the study, including analysis of genotype, DNAm, and mRNA expression in a single study, evaluation of multiple psychiatric and behavioral factors thought to relate to cellular aging, consideration of potential confounds and covariates that might account for our primary associations of interest, and the examination of the molecular effects of rs9315202 on KL regulation.


Results further highlight the importance of KL and rs9315202 for the study of accelerated cellular aging in stressed populations. The SNP moderated the association between PTSD and advanced epigenetic age in those 45 years of age and older and indirectly predicted reduced KL mRNA expression in motor cortex via advanced epigenetic age, with experimental evidence suggesting that the SNP influenced KL regulation. KL augmentation holds promise as a therapeutic intervention to slow the pace of cellular aging and stem the tide of premature disease onset and neuropathology [55, 64, 65], especially in the populations at greatest risk by virtue of their advanced age and psychiatric history. The evidence that psychiatric stress accelerates aging and is associated with early onset of age-related disease is substantial. The time has come to extend beyond these observations to determine the pathophysiology of advanced cellular aging and to develop personalized therapeutics that will effectively lengthen Clotho’s thread and enhance health and wellness.

Funding and disclosures

This work was supported by the National Institute on Aging grant number 1R21AG061367 to EJW, and VA BLR&D Merit Award grant number 1I01BX003477 to MWL. This work was also supported by a Presidential Early Career Award for Scientists and Engineers (PECASE 2013A) to EJW, as administered by U.S. Department of Veterans Affairs Office of Research and Development and by the National Center for PTSD. FGM’s contribution to this work was supported by National Institute of Mental Health award number 5T32MH019836-16. Genotype and methylation data were generated with the support of resources at the Pharmacogenomics Analysis Laboratory (Research Service, Central Arkansas Veterans Healthcare System, Little Rock, Arkansas), a core research laboratory funded by the Cooperative Studies Program, Research and Development, Department of Veterans Affairs. The contents of this manuscript do not represent the views of the U.S. Department of Veterans Affairs, the National Institutes of Health, or the United States Government.

All named authors report no financial or other conflicts of interest in relationship to the contents of this article. FGM contribution to this work was completed as a post-doctoral fellow at Boston University School of Medicine and the National Center for PTSD. Dr. Morrison is currently an employee of BlackThorn Therapeutics.

Traumatic Stress Brain Research Group member Ronald Duman has received consulting, speaking fees and/or grant support from Naurex, Taisho, Johnson & Johnson, Lilly, Lundbeck, Sunovion, Navitor, and Forest. Traumatic Stress Brain Research Group member John Crystal was a consultant for AstraZeneca Pharmaceuticals, Biogen, Idec, MA, Biomedisyn Corporation, Bionomics, Limited (Australia), Boehringer Ingelheim International, Concert Pharmaceuticals, Inc. Epiodyne, Inc., Heptares Therapeutics, Limited (UK), Janssen Research & Development, L.E.K. Consulting, Otsuka America Pharmaceutical, Inc., Perception Neuroscience Holdings, Inc., Spring Care, Inc., Sunovion Pharmaceuticals, Inc., Takeda Industries, and Taisho Pharmaceutical Co., Ltd. He is on the scientific advisory board of Bioasis Technologies, Inc., Biohaven Pharmaceuticals, BioXcel Therapeutics, Inc. (Clinical Advisory Board), Cadent Therapeutics (Clinical Advisory Board), PsychoGenics, Inc., Stanley Center for Psychiatric research at the Broad Institute of MIT and Harvard, and Lohocla Research Corporation. He owns stock in ArRETT Neuroscience, Inc, Biohaven Pharmaceuticals, Sage Pharmaceuticals, and Spring Care, Inc. He has stock options in Biohaven Pharmaceuticals Medical Sciences, BlackThorn Therapeutics, Inc., Storm Biosciences, Inc. He is an editor for Biological Psychiatry, and has received research support from AstraZeneca Pharmaceuticals provides the drug, Saracatinib, for research related to NIAAA grant “Center for Translational Neuroscience of Alcoholism [CTNA-4]. He has the following patents: (1) Seibyl JP, Krystal JH, Charney DS. Dopamine and noradrenergic reuptake inhibitors in treatment of schizophrenia. US Patent #:5,447,948.September 5, 1995. (2) Vladimir, Coric, Krystal, John H, Sanacora, Gerard—Glutamate Modulating Agents in the Treatment of Mental Disorders US Patent No. 8,778,979 B2 Patent Issue Date: July 15, 2014. US Patent Application No. 15/695,164: Filing Date: 09/05/2017. (3) Charney D, Krystal JH, Manji H, Matthew S, Zarate C., - Intranasal Administration of Ketamine to Treat Depression United States Application No. 14/197,767 filed on March 5, 2014; United States application or Patent Cooperation Treaty (PCT) International application No. 14/306,382 filed on June 17, 2014. (4): Zarate, C, Charney, DS, Manji, HK, Mathew, Sanjay J, Krystal, JH, Department of Veterans Affairs “Methods for Treating Suicidal Ideation”, Patent Application No. 14/197.767 filed on March 5, 2014 by Yale University Office of Cooperative Research. (5) Arias A, Petrakis I, Krystal JH.—Composition and methods to treat addiction. Provisional Use Patent Application no.61/973/961. April 2, 2014. Filed by Yale University Office of Cooperative Research. (6): Chekroud, A., Gueorguieva, R., & Krystal, JH. “Treatment Selection for Major Depressive Disorder” [filing date 3rd June 2016, USPTO docket number Y0087.70116US00]. Provisional patent submission by Yale University. (7) Gihyun, Yoon, Petrakis I, Krystal JH—Compounds, Compositions and Methods for Treating or Preventing Depression and Other Diseases. U. S. Provisional Patent Application No. 62/444,552, filed on January 10, 2017 by Yale University Office of Cooperative Research OCR 7088 US01. (8) Abdallah, C, Krystal, JH, Duman, R, Sanacora, G. Combination Therapy for Treating or Preventing Depression or Other Mood Diseases. U.S. Provisional Patent Application No. 047162-7177P1 (00754) filed on August 20, 2018 by Yale University Office of Cooperative Research OCR 7451 US01. None of the authors or remaining members of the Traumatic Stress Brain Research Group have any conflicts to disclose. The authors have no conflict of interest.