Stress-related genes and heroin addiction: a role for a functional FKBP5 haplotype

Summary

Background

Stress is a critical risk factor affecting both the development of and the relapse to drug addictions. Drug addictions are caused by genetic, environmental and drug-induced factors. The objective of this hypothesis-driven association study was to determine if genetic variants in stress-related genes are associated with heroin addiction.

Methods

112 selected genetic variants in 26 stress-related genes were genotyped in 852 case subjects and 238 controls of predominantly European ancestry. The case subjects are former heroin addicts with a history of at least one year of daily multiple uses of heroin, treated at a methadone maintenance treatment program (MMTP). The two most promising SNPs were subsequently tested in an African-American sample comprising of 314 cases and 208 control individuals.

Results

Nineteen single nucleotide polymorphisms (SNPs) in 9 genes (AVP, CRHR1, CRHR2, FKBP5, NR3C2, AVPR1A, GAL, GLRA1, and NPY1R) showed nominally significant association with heroin addiction. The associations of two FKBP5 SNPs that are part of one haplotype block, rs1360780 (intron 2) and rs3800373 (the 3' untranslated region), remained significant after correction for multiple testing (P_corrected =0.03; OR = 2.35, P_corrected = 0.0018; OR = 2.85, respectively). The two SNPs also showed nominally significant association (P <0.05) with heroin addiction in an independent African-American cohort. FKBP5 is a co-chaperone that regulates glucocorticoid sensitivity. These FKBP5 SNPs were previously associated with diverse affective disorders and showed functional differences in gene expression and stress response. This study also supports our and others’ previous reports of association of the GAL SNP rs694066 and the AVPR1A SNPs rs11174811, rs1587097 and rs10784339 with heroin and general drug addiction, respectively.

Conclusions

This study suggests that variations in the FKBP5 gene contribute to the development of opiate addiction by modulating the stress response. These findings may enhance the understanding of the interaction between stress and heroin addiction.

1. Background

Addiction to opiates and the illicit abuse of prescription opioids is a growing epidemic. Addiction to drugs is a chronic relapsing brain disease caused by a combination of genetic, epigenetic, environmental and drug-induced factors. Stress is a critical risk factor affecting both the development of addictive disorders, by promoting drug seeking and excessive drug intake, and the relapse to addictive behaviors, since drug withdrawal can increase stress response, and stress increases reward-seeking behavior, such as reinstatement of drug-taking behavior (Koob and Kreek, 2007; Sinha, 2008; Ulrich-Lai and Herman, 2009; Kreek et al., 2012). Studies showed a high rate of various types of childhood trauma exposure and affective disorders comorbidity among individuals with opioid dependence (Mills et al., 2005; Nelson et al., 2006; Sansone et al., 2009). The response to stress is influenced by genetic and environmental factors and has high inter-individual variability. A plastic neural circuitry that includes the hippocampus, amygdala, hypothalamus, brainstem and prefrontal cortex coordinates the response systems (McEwen and Gianaros, 2011).

Adrenal secretion of glucocorticoids is one of the mechanisms of response to stress. Stress exposure, as well as endogenous opioids and drugs of abuse, activate the hypothalamic-pituitary-adrenal (HPA) axis. Consequently, corticotropin-releasing hormone (CRH, CRF) and arginine vasopressin (AVP) are released from the hypothalamic paraventricular nucleus (PVN) and are transported to the anterior pituitary and stimulate adrenocorticotropic hormone (ACTH) secretion, which in turn stimulates glucocorticoid synthesis and release from the adrenal cortex. CRF is also produced in other brain regions and activates the sympathetic nervous system to release epinephrine and norepinephrine from the adrenal glands. It also stimulates the mesocorticolimbic dopamine system that mediates the rewarding effects associated with drug use. High levels of glucocorticoids can “sensitize” CRF systems in the extra hypothalamic brain stress systems (extended amygdala)(Koob, 2010).

Glucocorticoids regulate the activity of the HPA axis through negative feedback via the glucocorticoid receptor (GR) and the mineralocorticoid receptor (MR). These receptors regulate the expression of genes necessary for coping with stress. The functions of the GR and the MR are partly moderated by chaperone proteins including the heat shock protein 90 (Hsp90) co-chaperone FKBP5 (FKBP51, FK506-binding protein 51). Glucocorticoids have many other effects when bound to glucocorticoid receptors (e.g., modulation of cardiovascular function, immunologic status, arousal, learning and memory). They can also alter the methylation patterns of genes (Stephens and Wand, 2012).

Numerous molecular genetic studies have evaluated the association between polymorphisms in stress-related genes and affective disorders (Domschke and Reif, 2012), but only a few studies have reported an association of variants in these genes with specific drug addictions. We have previously performed association studies of heroin addiction that include several genes related to stress response (Levran et al., 2008; Proudnikov et al., 2008; Levran et al., 2009). These studies identified, in these genes, association of SNPs in the galanin gene (GAL) in European Americans, the AVP receptor gene (AVPR1A) in African Americans, and the ACTH receptor gene (MC2R) in Hispanics. A different AVPR1A SNP was shown to be associated with general drug use disorders by another group (Maher et al., 2011), and NPY2R SNPs were associated with alcohol and cocaine dependence (Wetherill et al., 2008).

Here, we report the results of a case-control hypothesis-driven association study of 112 SNPs from 26 genes related to stress response, with heroin addiction, in a sample of 1090 subjects of predominantly European ancestry. The study is a major expansion of our previous study (Levran et al., 2008) to which 517 samples, 12 additional stress-related genes, and several new SNPs in genes included in the previous study were added. The study employed more stringent inclusion criteria for ancestry, based on biographic ancestry scores obtained by STRUCTURE analysis of 155 Ancestry Informative Markers (AIMs). This study included a validation sample of different ethnicity (African-American) for the most significant results.

2. Methods

2.1. Subjects

Discovery sample

The 1090 subjects of this study are part of a larger cohort recruited by the Kreek laboratory for the study of the genetics of specific drug addictions. There were 852 cases (33% female; mean age 40±12) and 238 controls (49% female; mean age 42±16). The subjects were selected based on phenotype (history of severe heroin addiction, normal controls) and self-identified European ancestry (including a Middle-Eastern contribution). Other ethnicities were excluded (e.g., Africans, Hispanics, Asians, Native Americans or mixed ancestry). Ancestry was verified by a family history questionnaire and STRUCTURE analysis (see below), and specific inclusion criteria were employed to obtain relative homogeneity and to limit population stratification. To be included in the discovery sample, an individual had to show at least a 50% European, Middle-Eastern or both ancestry contributions and less than 25% contribution from another specific ancestry. This study is a major expansion of our previous study (Levran et al., 2008). The current study includes a majority of the samples from the original study.

The case subjects were former heroin addicts with a history of at least one year of daily multiple uses of heroin, treated at a MMTP at the time of recruitment. The case subjects were recruited at the Rockefeller University Hospital (n = 238), the Manhattan Campus of the VA NY Harbor Health Care System (n = 55) and the Dr. Miriam and Sheldon G. Adelson Clinics for Drug Abuse Treatment and Research, in Las Vegas (n = 264) and Israel (n = 295). The control sample was mainly from the NYC community (n = 208) and was recruited at the Rockefeller University Hospital by newspaper and online advertisement. Additional controls were recruited at the Adelson clinic in Israel (n = 30). Ascertainment of cases and controls was made by personal interview performed in a similar manner at the recruiting places, using several instruments: the Addiction Severity Index (McLellan et al., 1992), Kreek-McHugh-Schluger-Kellogg Scale (KMSK) (Kellogg et al., 2003) and Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV). The following exclusion criteria from the healthy control category were used: (1) at least one instance of drinking to intoxication or any illicit drug use in the previous 30 days; (2) a history of alcohol drinking to intoxication or illicit drug use, more than twice a week, for more than six consecutive months, and (3) cannabis use for more than 12 days in the previous 30 days or past cannabis use for more than twice a week for more than four years. Subjects with active DSM-IV axis I disorder were excluded from the study. All subjects completed a family history questionnaire and relatives were excluded from the study. The Institutional Review Boards of the Rockefeller University Hospital, the VA New York Harbor Healthcare System and the Tel Aviv Sourasky Medical Center (Helsinki Committee) approved the study. All subjects signed informed consent for genetic studies.

Validation sample

The two most significant FKBP5 SNPs were subsequently tested in an independent African-American (>50% African ancestry, non-Hispanic) sample comprising of 314 (37% female) cases and 208 (52% female) control individuals. The average African contribution was ~80% in both cases and controls, with similar pattern of distribution of the other ancestry clusters (5% each of Europe, Middle East and Central Asia). The case sample was recruited at the Rockefeller University Hospital (n = 230), the Manhattan Campus of the VA NY Harbor Health Care System (n = 63) and the Dr. Miriam and Sheldon G. Adelson Clinic for Drug Abuse Treatment and Research in Las Vegas (n = 21).The control sample was recruited at the Rockefeller University Hospital. Ethnicity verification by STRUCTURE analysis, recruitment and ascertainment were as described for the discovery sample.

2.2. Genes/SNPs selection and array design

Twenty-seven genes were selected based on their known function in response to stress (Table 1, Table S1). In addition to the genes included on the original hypothesis-driven “addiction” array (GS0007064-OPA; Illumina, San Diego, CA, USA) (Hodgkinson et al., 2008) that was used in our original association studies (Levran et al., 2008; Levran et al., 2009), we have added 12 stress-related genes to the current array (GS0013101-OPA). Out of the 68 new SNPs from stress-related genes that were selected for the new array based on previous reports of potential functionality and/or association with stress response, 13 SNPs were not technically suitable for this platform based on the Illumina Assay Design Tool, and 55 SNPs were added to the new array. A total of 143 SNPs from these stress-related genes were included in the array, out of which 88 SNPs were from the original “addiction” array (Table S1).

Table 1.

Stress-related genes

Symbol	Gene name
AVP	arginine vasopressin
AVPR1A	arginine vasopressin receptor 1A
AVPR1B	arginine vasopressin receptor 1B
CARTPT	CART prepropeptide
CCK	cholecystokinin
CRH	corticotropin releasing hormone
CRHBP	corticotropin releasing hormone binding protein
CRHR1	corticotropin releasing hormone receptor 1
CRHR2	corticotropin releasing hormone receptor 2
FKBP5	FK506-binding protein 51
GAL	galanin
GALR1	galanin receptor 1
GLRA1	glycine receptor, alpha 1
HCRTR1	hypocretin receptor 1
HCRTR2	hypocretin receptor 2
MC2R	melanocortin 2 receptor
NPY	neuropeptide Y
NPY1R	neuropeptide Y receptor Y1
NPY2R	neuropeptide Y receptor Y2
NPY5R	neuropeptide Y receptor Y5
NR3C1	nuclear receptor subfamily 3, group C, member 1 (glucocorticoid receptor)
NR3C2	nuclear receptor subfamily 3, group C, member 2 (mineralocorticoid receptor)
OXT	oxytocin
OXTR	oxytocin receptor
PITX1	paired-like homeodomain transcription factor 1
SERPINA6	corticosteroid binding globulin

Genes are sorted by alphabetical order

2.3. Genotyping

Blood samples were taken and DNA was extracted and quantified using standard methods. DNA (700 ng) was precipitated as described (Levran et al., 2008). Genotyping was performed on a 1536-plex GoldenGate Custom Panel at the Rockefeller University Genomics Resource Center according to the manufacturer’s protocol. Random samples were genotyped in duplicate. Analysis was performed with BeadStudio software v2.3.43 (Illumina). The genotype data for all SNPs were visually inspected to verify and correct automatic calling. Genotype data were filtered based on SNP call rates (> 99%), minor allele frequency (MAF) in controls > 0.05, and deviation from Hardy-Weinberg equilibrium (HWE) in controls (P < 0.001) (Table S1). Of the 143 SNPs genotyped, 14 SNPs were excluded from analysis based on low quality (n = 6), more than 3 clusters (n = 4), or failure on some of the plates that did not reflect ambiguous clustering (n = 4), and 17 SNPs were excluded from analysis based on MAF < 0.05 in this sample. Since the HCRT gene was represented by only one SNP with MAF < 0.05 in this sample, this gene is not included in the final gene list. The total analysis was performed with 112 SNPs from 26 genes (Table S1).

2.4. Assessment of ancestry contribution using AIMs

Of the original 186 AIMs from the GS0007064-OPA panel (Hodgkinson et al., 2008), 171 SNPs with adequate quality were included in the new panel, and 155 AIMs with high quality scores were used for analysis. Biographic Ancestry Scores (e.g., fractions of genetic affiliation of the individual in each cluster) were estimated by STRUCTURE 2.2 with seven clusters (K). Each subject was anchored against genotypes of 1051 samples from 51 worldwide populations represented in the Human Genome Diversity Cell Line Panel (http://www.cephb.fr/HGDP-CEPH-Panel), as described (Ducci et al., 2009). To be included in the discovery sample, an individual had to show at least a 50% European, Middle-Eastern, or both ancestry contribution and less than 25% contribution from another specific ancestry. The decision to include both European and Middle-Eastern clusters was based on their low population differentiation (Tian et al., 2009). Studies showed especially close relationship between Middle Eastern and Southern European populations (Atzmon et al., 2010). To be included in the validation sample, an individual had to show at least a 50% African ancestry contribution.

In the discovery sample, the mean (SD) European ancestry contribution score was 0.73 (0.32) with median of 0.89. The mean Middle Eastern score was 0.21 (0.30) with median of 0.03. The means and medians of the other five ancestries were <0.05 (<0.05). In the validation sample, the mean African score was 0.8 (0.11) with median of 0.82. The European, Middle Eastern and Central Asia mean and median scores were 0.05±0.01 (0.04). The means of the other three ancestries’ scores were 0.01±0.01.

From the original cohort of self-identified European and/or Middle-Eastern subjects, 57 subjects were excluded because they did not meet the inclusion criteria. Seven subjects were excluded based on a major conflict between their self-identified ethnicity and STRUCTURE results. Twenty-two subjects with ambiguous self-identified ancestry and six subjects who self-identified as African Americans were included based on STRUCTURE results.

2.5. Statistical analysis

Quality control for the SNP genotypes was carried out as follows. Exact tests for deviation from HWE were performed with the PLINK program, with SNPs to be rejected based on the recommended threshold of P < 0.001 in control individuals. Pairwise linkage disequilibrium (LD) and haplotype blocks were estimated using Haploview 4.2. Analysis of block of LD was performed using the Confidence Intervals algorithm for block definition (Gabriel et al., 2002), with stringent criteria of LD (D' lower bound of 0.9). Association analysis was performed for each SNP separately by logistic regression in the PLINK program, where in different analyses the genotype was coded as a linear allelic effect (genotypes AA, AB, and BB were given numerical values 0, 1, and 2, respectively), and as two groups reflecting a dominant or recessive inheritance model. Sex was included as a covariate. Ancestry contribution scores (European and Middle Eastern for the discovery sample and African for the validation sample) were initially included as predictor covariates but had no significant effects on the dependent variable so they were not included in the final analysis. The multiple testing issue was addressed by assuming 67 independent SNPs following the method suggested by Duggal et al. (Duggal et al., 2008). Based on analysis of block of LD in controls, there are 45 “non-independent” SNPs, including seven redundant SNPs in complete LD (r² = 1) so that, with Bonferroni correction, P values were corrected by division of 0.05 by 67. P values were further adjusted for testing three models of inheritance by division by 1.7 (= √3) based on Mantel (Mantel, 1980). Thus, P = 4.4E-04 (0.05/113.9) was chosen as the threshold for significance.

3. Results

A total of 1090 subjects (852 cases and 238 controls) met the inclusion criteria and were included in the association analyses. The ethnicity of all subjects was verified as predominantly European using STRUCTURE analysis of 155 AIMs (Fig. 1). Genotypes of 112 SNPs (MAF > 0.05) from 26 stress-related genes were analyzed for association with heroin addiction (Table 1, Table S1). The European and Middle Eastern ancestry contribution scores did not show independent effects on the dependent variable indicating that the associations are not due to population stratification. No SNP showed deviation from HWE (P < 0.001) in controls. One SNP (rs10213647) showed deviation from HWE in cases only (P = 0.0004). Based on LD analysis in controls, there are 45 correlated SNPs, including seven redundant SNPs in complete LD (r² = 1), suggesting an “effective” number of 67 independent SNPs.

Figure 1. Schematic representation of the individual admixture estimates.

The estimates are based on STRUCTURE analysis using K=7. Each vertical line represents one individual, and subjects are displayed according to their predominant cluster contribution (see Methods). The clusters correspond to the geographical regions based on the HGDP sample. Color code: Africa (1) (blue), Europe (2) (red), Middle East (3) (green), Central Asia (4) (purple), Far East Asia (5) (cyan), Oceania (6) (amber), and America (7) (light blue).

Nineteen SNPs in nine genes showed nominally significant association of genotype with heroin addiction (Table 2). The top signals are from the following genes: AVP, CRHR1, CRHR2, FKBP5, NR3C2, AVPR1A, GAL, GLRA1 and NPY1R. Two tightly-linked FKBP5 SNPs, rs1360780 and rs3800373, from intron 2 and the 3' UTR, respectively, remained significant after correction for multiple testing (P_corrected = 0.03; OR = 2.35; 95% CI,1.5–3.7 & P_corrected = 0.0018; OR = 2.85; 95% CI,1.8–4.6, respectively). The two SNPs also showed nominally significant association in the same direction (P = 0.04, OR = 1.49; 95% CI, 1.01–2.18 & P = 0.03, OR = 1.54; 95% CI 1.04–2.30, respectively) with heroin addiction in an independent African-American sample comprising of 314 cases and 208 control individuals that was used for validation and generalization. The African ancestry contribution score did not show independent effects on the dependent variable indicating that the associations are not due to population stratification.

Table 2.

Details of SNPs with nominally significant associations (P < 0.05)

Gene	SNP	Location	Allelesa	MAF Case	MAF Control	P	Modeb	OR (95% CI)	Risk allele
AVP	rs2282018	intron	T/C	0.37	0.42	0.027	D	1.41 (1.04–1.92)	T
AVPR1A	rs3021529	5' UTR	C/T	0.14	0.17	0.011	D	1.50 (1.10–2.06)	C
	rs3803107	3' UTR	C/T	0.15	0.20	2.7E-03	D	1.61 (1.18–2.18)	C
	rs11174811	3' UTR	C/A	0.14	0.18	4.2E-03	D	1.57 (1.15–2.15)	C
	rs10784339	3' UTR	G/C	0.17	0.21	0.022	D	1.42 (1.05–1.92)	G
	rs1587097	3' intergenic	C/T	0.11	0.14	0.026	D	1.57 (1.15–2.15)	C
CRHR1	rs242939	intron	A/G	0.09	0.06	0.038	A	1.54 (1.02–2.33)	G
CRHR2	rs255102	intron	A/T	0.35	0.32	0.048	D	1.34 (1.00–1.80)	T
FKBP5	rs9470080	intron	C/T	0.29	0.34	3.3E-03	R	1.97 (1.25–3.09)	C
	rs1360780	intron	C/T	0.27	0.33	3.0E-04	R	2.35 (1.48–3.72)	C
	rs7757037	intron	A/G	0.51	0.44	1.1E-03	D	1.70 (1.24–2.34)	A
	rs3800373	3' UTR	A/C	0.26	0.33	1.6E-05	R	2.85 (1.77–4.58)	A
GAL	rs694066	intron	C/T	0.12	0.07	5.2E-03	A	1.72 (1.17–2.51)	T
	rs3136541	intron	T/C	0.40	0.32	3.2E-03	A	1.39 (1.12–1.73)	C
GLRA1	rs2915885	intron	T/C	0.41	0.35	0.027	A	1.27 (1.03–1.57)	C
	rs9324714	intron	T/A	0.41	0.35	0.038	A	1.25 (1.01–1.55)	A
	rs1428159	intron	A/G	0.41	0.36	0.049	A	1.24 (1.00–1.53)	G
NPY1R	rs4518200	5' intergenic	A/C	0.10	0.12	0.030	D	1.47 (1.03–2.08)	C
NR3C2	rs1040288	intron	C/G	0.45	0.52	0.011	A	1.31 (1.06–1.61)	Gc

SNPs are listed in alphabetical order. SNPs in each gene are sorted by relative position. Bolded SNPs are significant after correction for multiple testing.

^aAlleles are listed as major/minor in Europeans;

^bInheritance modes: R=recessive; D=dominant; A=additive;

^cThe risk allele is the major allele in the case sample but the minor allele in the control sample.

LD analysis revealed that some of the top results are likely to be related, as some SNPs are in strong LD. As is shown in Fig. 2, the four FKBP5 SNPs are linked (D' >.89), of which 3 SNPs (rs3800373, rs1360780, and rs9470080) are tightly linked (r² >.77). The three GLRA1 and the five AVPR1A SNPs are tightly linked (r² > .96, r² > .62, respectively). In contrast, the two GAL SNPs showed lower levels of LD.

Figure 2. Pairwise Linkage Disequilibrium (LD).

LD between SNPs in four genes was derived from genotypes of controls. The pairwise correlation between SNPs was measured as D' (in red) and r² (in black). The values are shown (x100) in each box. The color scheme indicates the magnitude of the value. When the value is equal to 1.0 no number is given. a. AVPR1A. b. FKBP5. c. GAL. d. GLRA1.

4. Discussion

The strongest results of this study are of four linked FKBP5 SNPs, two of which (rs1360780 and rs3800373, from intron 2 and the 3' UTR, respectively) remained significantly associated with heroin addiction after correction for multiple testing. The odd ratios (OR) were relatively high (> 2) and the more abundant alleles (C and A, respectively) were the “risk alleles”, while the less abundant alleles can be considered “protective alleles”. The two SNPs also showed nominally significant association in an independent African-American sample that was used for validation and generalization. These two FKBP5 SNPs are in strong LD across different ethnic groups. The FKBP5 gene was not included in our previous association study (Levran et al., 2008).

FKBP5 is considered a negative transcriptional regulator of steroid receptors. FKBP5 is a component of steroid hormones receptor hetero-complexes along with the Hsp90 and the p23 protein. It binds the immunosuppressive drugs FK506 (tacrolimus) and rapamycin. FKBP5-Hsp90 hetero-complex regulates GR sensitivity via an ultra-short negative feedback loop (Galigniana et al., 2012). When cortisol binds the GR in the cytoplasm, the complex enters the nucleus and GR regulates transcription of glucocorticoid-responsive genes. FKBP5 regulates GR sensitivity by preventing translocation of the GR complex to the nucleus.

Several rodent studies shed light on the relationship between stress, drug addiction, and FKBP5. Chronic morphine administration and precipitated withdrawal were shown to change Fkbp5 expression in the locus ceruleus and the VTA, and blockade of FKBP5 with FK506 attenuated dependence development (Homayoun et al., 2003; McClung et al., 2005). Recent studies found that Fkbp5 expression in the mouse striatum was strongly activated by acute and chronic administration of opioids (Piechota et al., 2012). Fkbp5 whole brain expression was shown to be regulated by oxycodone (Hassan et al., 2010). Fkbp5 KO mice display a reduced response to chronic social defeat stress. Under basal conditions, deletion of Fkbp5 did not change anxiety or depression-like behavior. However, exposure to stressors led to a more active coping behavior and decreased HPA axis reactivity and GR expression (Touma et al., 2011; Hartmann et al., 2012). Glucocorticoid administration was shown to decrease DNA methylation in Fkbp5 in brain and blood, and this decrease was associated with behavioral deficits (Lee et al., 2011; Yang et al., 2012).

FKBP5 SNPs were shown to be associated with stress response and affective disorders (e.g., (Binder, 2009; Willour et al., 2009; Velders et al., 2011)). Impaired GR sensitivity was detected in major depression patients that were carriers of the FKBP5 rs1360780 T allele (Menke et al., 2013). FKBP5 SNPs were associated with baseline and peak cortisol response to social stress test in healthy individuals (Mahon et al., 2013). An fMRI study of African-American subjects showed that carriers of the rs1360780 T allele had an attention bias toward threat, increased hippocampal activation to stress, and difference in hippocampal shape during performance of an attention-bias task (Fani et al., 2013).

Several studies support the hypothesis that genetic or epigenetic modifications in FKBP5 modulate the effect of the environment on the HPA axis and the risk for stress-related disorders (gene×environment interaction). These studies suggest that dysregulation of the HPA axis function in childhood may have long-lasting effects. FKBP5 variants were shown to interact with childhood trauma to predict depression, suicidality, aggression, and PTSD (e.g.,(Binder et al., 2008; Xie et al., 2010; Appel et al., 2011; Bevilacqua et al., 2012; Boscarino et al., 2012; Roy et al., 2012; Klengel et al., 2013)). Binder’s group (Binder et al., 2004; Binder et al., 2008; Klengel et al., 2013) identified the less abundant alleles of four SNPs (H2 haplotype, including the rs1360780 T allele) to be the “risk alleles” that moderate child abuse-related risk for adult PTSD, in African Americans. They also showed that this allele increased FKBP5 expression and caused relative GR resistance. Similar results were obtained by others (Xie et al., 2010; Appel et al., 2011). In contrast, Roy et al. (Roy et al., 2010; Roy et al., 2012) identified the more abundant alleles of these SNPs (H1 haplotype, including rs1360780 C allele) as the “risk alleles” for suicide attempt in substance-dependent African Americans. They found no association of FKBP5 SNPs and substance dependence (alcohol, cocaine, or opiate). A study of aggression in Italian males from the same group (Bevilacqua et al., 2012) reported a dosage effect of rs1360780 alleles with larger effect of the less abundant allele. The Bevilacqua et al. study suggested an association of the more abundant FKBP5 haplotype with substance dependence in a subgroup of 127 subjects with mixed-drug dependencies and high comorbidity with Axis I psychiatric disorders. Notably, the current study does not include subjects with active DSM-IV axis I disorder; therefore the addiction is not secondary to psychiatric disorders in this cohort. Since this study has limited information about trauma exposure, we cannot address the suggestion of gene × environment interaction at this point.

A possible mechanism of action of FKBP5 SNP rs1360780 was suggested by a recent study by The Binder’s group (Klengel et al., 2013). They found that differential DNA CpG methylation in FKBP5, in carriers of the FKBP5 rs1360780 T allele that were exposed to childhood trauma, increased the risk of developing stress-related psychiatric disorders in adulthood. A bioinformatics analysis suggested that SNP rs3800373 may cause the loss of exon splicing enhancer motive that may cause exon skipping (Liu et al., 2000).

Several other SNPs were identified by this study with only nominal significant association that may not reflect true association; nonetheless, additional support for these findings is provided by several previous association studies of related phenotypes and evidence of functionality. The AVPR1A 3' UTR SNPs rs11174811, rs1587097 and rs10784339 were previously associated with a non-specific drug use disorder in European-Americans (Maher et al., 2011). SNP rs11174811 is located in potential seed recognition sites for microRNAs (miRs) miR-526b and miR-578 and was shown to disrupt miR/mRNA interactions, in vitro (Nossent et al., 2011). The other two AVPR1A SNPs are in strong LD with SNP rs11174811. SNP rs3803107 is located very close (358 bp) to SNP rs11174811, so their association signals are most probably related. We have previously reported an association between another AVPR1A SNP (rs3759292) in the 5' flanking region and heroin addiction in African Americans (Levran et al., 2009). SNP rs3759292 is rare in European populations and was excluded from analysis in this study based on very low allele frequency (< 0.05).

The finding of this study also supports our previous report of association between GAL SNP rs694066 and heroin addiction (Levran et al., 2008). This variant was also shown to be associated with depression in a gender-dependent manner in Han Chinese (Wang et al., 2013). A GAL haplotype that includes SNPs rs694066 and rs3136541 was associated with alcoholism in Finnish and Plains American Indian men (Belfer et al., 2006).

In addition, CRHR1 SNP rs242939 was associated with depressive disorders in Han Chinese (Liu et al., 2006; Xiao et al., 2011; Liu et al., 2013) and British females (Engineer et al., 2013). NPY1R SNP rs4518200 is in strong LD with the 3' UTR SNP rs4552421 that is part of a haplotype that was associated with nicotine dependence in Han Chinese (Wei et al., 2012).

This study supports the hypothesis that atypical GR sensitivity underlies the pathophysiology of drug addiction, and contributes to addiction continuation and relapse (Kreek et al., 2005). Drug exposure generates chronic stress, activates the stress response and creates reward sensitization and a negative emotional state. Active heroin addicts have hypo-responsive (blunted) HPA system that may be caused by a preexisting trait or consequence of long term heroin abuse and/or exposure to trauma. The modulation of the stress response by FKBP5 variants, possibly in combination with other stress-related genes variants, may contribute to the development of heroin dependence by increasing sensitivity to both stress and heroin effect.

The identification of specific stress-related genetic variants involved in heroin addiction has potential clinical implications. It identifies treatment targets, supports treatment options suggested by animal studies, and has the potential to optimize therapeutic interventions by identifying patients who are at specific risk for stress-related relapse and/or patients that would benefit from specific interventions based on their genotype, especially early in the addiction cycle.

In summary, this study provides evidence for the involvement of specific genetic variants in several stress-related genes with heroin addiction. It is plausible that combinations of alleles and gene-gene interactions underlie the genetic basis of heroin addiction. Future studies with greater statistical power should examine the contributions of simultaneous variations in several genes. The study suggests that functional genetic variations in the FKBP5 gene, which is involved in glucocorticoid sensitivity, affect the risk for heroin addiction. It is important to verify whether the associations of these gene variants are specific to heroin addiction or are shared with other drug addictions. Additional association and functional studies utilizing different drug addictions, populations with different ancestries, and broader SNP coverage are required to further corroborate these findings and define the role of FKBP5 and the other genes variations indicated, as contributing factors to heroin addiction.