Alterations in DNA methylation of Fkbp5 as a determinant of blood–brain correlation of glucocorticoid exposure

Summary

Background

Epigenetic studies that utilize peripheral tissues to identify molecular substrates of neuropsychiatric disorders rely on the assumption that disease-relevant, cellular alterations that occur in the brain are mirrored and detectable in peripheral tissues such as blood. We sought to test this assumption by using a mouse model of Cushing's disease and asking whether epigenetic changes induced by glucocorticoids can be correlated between these tissue types.

Methods

Mice were treated with different doses of glucocorticoids in their drinking water for four weeks to assess gene expression and DNA methylation (DNAm) changes in the stress response gene Fkbp5.

Results

Significant linear relationships were observed between DNAm and four-week mean plasma corticosterone levels for both blood (R² = 0.68, P = 7.1 × 10⁻¹⁰) and brain (R² = 0.33, P = 0.001). Further, degree of methylation change in blood correlated significantly with both methylation (R² = 0.49, P = 2.7 × 10⁻⁵) and expression (R² = 0.43, P = 3.5 × 10⁻⁵) changes in hippocampus, with the notable observation that methylation changes occurred at different intronic regions between blood and brain tissues.

Conclusion

Although our findings are limited to several intronic CpGs in a single gene, our results demonstrate that DNA from blood can be used to assess dynamic, glucocorticoid-induced changes occurring in the brain. However, for such correlation analyses to be effective, tissue-specific locations of these epigenetic changes may need to be considered when investigating brain-relevant changes in peripheral tissues.

Introduction

Levels of the glucocorticoid cortisol constitute one of the key determinants of allostasis or allostatic load, and prolonged exposure to its catabolic properties leads to numerous diseases that include diabetes, cardiovascular disease, and obesity (Karlamangla et al., 2002). For the brain, exposure to glucocorticoids is a robust risk factor for neuropsychiatric illnesses and cognitive decline, as numerous lines of evidence implicate exposure to stress or glucocorticoids in the etiology of several psychiatric disorders (McEwen and Gianaros, 2010, Fardet et al., 2012, Theall et al., 2012). In particular, hypercortisolemia or dysregulation of the hypothalamic-pituitary-adrenal-axis (HPA-axis) system is often associated with mood disorders (Gillespie and Nemeroff, 2005), and genetic association studies have linked single nucleotide polymorphisms in “HPA-axis genes” that govern the stress-response with post-traumatic stress disorder (PTSD), suicide, and depression (Binder et al., 2004, Binder et al., 2008, Roy et al., 2010, Roy et al., 2012, Sinclair et al., 2012). Large-scale epidemiological studies have also implicated iatrogenic glucocorticoid administration with psychiatric symptoms in hundreds of thousands of patients prescribed steroids for non-psychiatric diseases (Fardet et al., 2012). In addition, a more direct, causal relationship between glucocorticoid exposure and psychiatric illnesses can be found in Cushing's disease, where endogenous or iatrogenic elevation in plasma glucocorticoid levels has led to depression in 60–90% of patients (Cohen, 1980, Flitsch et al., 2000). Notably, the psychiatric symptoms are often alleviated with resolution of hypercortisolemia, strongly suggesting involvement of glucocorticoids in depressive symptoms (Starkman et al., 1986, Dorn et al., 1997). In animals, stress-induced deficits in both dopamine signaling and behavior are rescued by the glucocorticoid receptor antagonist mifepristone (RU486), further implicating glucocorticoid action (Niwa et al., 2013). Therefore, the ability to measure glucocorticoid exposure in the brain would serve as a useful gauge for assessing susceptibility to neuropsychiatric illnesses.

Previously, we found that degree of change in DNA methylation (DNAm) at a key regulatory locus within the stress-response gene Fkbp5 robustly reflected the previous 30-day mean plasma glucocorticoid values in the mouse blood (Lee et al., 2011). In fact, methylation changes in blood also correlated with anxiety-like behavior on the elevated plus maze. In addition, these epigenetic changes strongly correlated with functionally relevant physiological changes in glucocorticoid target tissues, such as atrophy of the thymus, spleen, and the adrenal glands. Thus we suggested in our previous study that using such an approach is highly practical since only one small sampling (∼20 μL of blood) accurately reflected and obviated the need for multiple daily measurements of plasma glucocorticoid levels when determining glucocorticoid burden (Lee et al., 2011).

In this study, we extend our analysis to the hippocampus, a particularly glucocorticoid-sensitive brain region (Kaouane et al., 2012), to determine whether the correlations between mean glucocorticoid exposure over a four-week treatment period and changes in DNAm and expression levels of the Fkbp5 gene in the blood accurately reflect changes in the brain. In addition to its function as a stress-response gene and a regulator of glucocorticoid signaling (Scammell et al., 2001, Wochnik et al., 2005), FKBP5 has been implicated in several candidate gene association studies of depression, bipolar disorder, and PTSD (Binder et al., 2004, Binder et al., 2008, Willour et al., 2009, Roy et al., 2012). A recent clinical study implicating methylation changes of FKBP5 in blood to childhood trauma exposure (Klengel et al., 2013) further warrants a closer examination of the relationship between blood and brain epigenetic changes in this important gene.

Given the link between glucocorticoid exposure and neuropsychiatric disorders, development of a tool that can assess exposure in the brain through the use of a surrogate, easily accessible tissue would be of great clinical utility. While reported as a case study of a single but an important gene in stress response, our results provide useful insights into the use of surrogate tissues for brain and suggest guidelines for future biomarker discoveries.