Impulsivity in intertemporal choice has been operationalized as “delay discounting”, referring to the preference for a sooner, smaller reward. FK506 binding protein 5 (FKBP5) is a co-chaperone of the glucocorticoid receptor (GR). FKBP5 overexpression causes GR resistance, resulting in increased plasma cortisol levels. High cortisol levels are associated with low impulsivity in intertemporal choice. The aim of this study was to explore the effect of single nucleotide polymorphisms (SNPs) in FKBP5 on delay discounting.
The participants consisted of 91 healthy Japanese people (66 males and 25 females with a mean age of 40.9 ± 6.9 years). Each participant completed Kirby's monetary choice questionnaire (MCQ) and donated a whole blood sample. Five SNPs in FKBP5 were genotyped using the DigiTag2. SNP linear regression analyses with 100,000 permutations were conducted for the hyperbolic time-discount rate (k).
Two SNPs were excluded from analysis because of their low minor allelic frequencies. The SNP rs1360780 showed a significant association; participants with more minor alleles (T) were less impulsive in intertemporal choice for delayed gain (multiplicity-corrected P = 0.047).
The significant SNP rs1360780 is located in the region adjacent to the hormone response element (HRE)-binding sequence where transcription factors bind and alter the transcription of FKBP5. A minor allele (T) of rs1360780, which causes FKBP5 overexpression, may reduce impulsivity in intertemporal choice (i.e. delay discounting) via GR resistance and the subsequent high cortisol levels. This is the first study to demonstrate an association between FKBP5 and impulsivity in intertemporal choice.
Investigations of the relationship between impulsivity and stress regulation are important for a better understanding of the role of stress in addiction (Takahashi, 2010; Takahashi et al., 2010; Bruijnzeel, 2012), in anti-social behavior (Vaillancourt and Sunderani, 2011), and in suicidal behavior (Kamali et al., 2012). The key system in the stress response is regulated by the faster activity of the adrenergic neurotransmitters (i.e. the sympatho-adrenomedullary [SAM] system) and the slower activity of the glucocorticoid hormone system (i.e. the hypothalamic-pituitary-adrenal [HPA] system). Glucocorticoid sensitivity reportedly plays a pivotal role in mood disorders (Spijker and van Rossum, 2012). Cortisol is the main steroid hormone of the HPA system in humans. It exerts its action via two distinct types of steroid receptors: the glucocorticoid receptor (GR) and the mineralocorticoid receptor (MR). The GR is distributed in the hippocampus, the prefrontal cortex, the paraventricular nucleus of the hypothalamus, and the amygdala. The MR is found in the hippocampus, the prefrontal cortex, and the amygdala (Patel et al., 2000). FK506 binding protein 5 (FKBP5), which is a co-chaperone of the GR, modulates GR activity (Pratt and Toft, 1997).
In the rapidly developing field of neuroeconomics (Glimcher and Rustichini, 2004; Camerer et al., 2005; Takahashi, 2009; Hasler, 2012), impulsivity in intertemporal choice has been operationalized as “delay discounting”, referring to the decrease in the subjective value of a reward as the delay until its receipt increases (i.e. the preference for a sooner, smaller reward rather than a later, larger reward) (McClure et al., 2004, 2007; Kable and Glimcher, 2007; Takahashi, 2009; Hasler, 2012; Yu, 2012). There have been reports that impulsivity in intertemporal choice is associated with the dopaminergic and serotonergic systems (Takahashi, 2009; Sellitto et al., 2011). In regard to the relationship between stress and an impulsive decision over time, we previously demonstrated that cortisol (which is a marker of HPA system activation) and salivary alpha-amylase (which is a non-invasive biomarker of SAM system activation [Nater and Rohleder, 2009]) are associated with reduced impulsivity in intertemporal choice (Takahashi, 2004; Takahashi et al., 2007, 2010). Low cortisol levels are associated with risky decision-making (van Honk et al., 2003), habitual violence (Virkkunen, 1985), and anti-social behavior (McBurnett et al., 2000). Recent studies further report that impulsivity in intertemporal choice is related to suicidal behavior (Dombrovski et al., 2011, 2012; Takahashi, 2011).
FKBP5 regulates the activity of the GR. A number of genetic studies have also linked it to stress-related diseases such as major depression and posttraumatic stress disorder (PTSD) (Roy et al., 2012). When FKBP5 is bound to the GR complex via the heat shock protein 90 (Hsp90), the receptor has a lower affinity for cortisol, resulting in GR resistance to glucocorticoid activation. Variations in the FKBP5 gene have been associated with the response to antidepressants (Sarginson et al., 2010), the recurrence of depressive episodes (Binder et al., 2004), completed suicide in the Japanese population (Supriyanto et al., 2011), and dysregulation of cortisol secretion (Velders et al., 2011). Because suicidal behavior (Dombrovski et al., 2011, 2012; Takahashi, 2011) and addiction (Takahashi, 2009; Monterosso et al., 2012) are related to impulsivity in intertemporal choice, it is important to examine the role of genetic variations in the FKBP5 gene for a better understanding of the relationship between stress and impulsivity.
The aim of the present study is to explore the effect of single nucleotide polymorphisms (SNPs) in the FKBP5 gene on impulsivity in intertemporal choice (i.e. delay discounting).
All participants were recruited in 2010 from Kanagawa Prefecture, which is adjacent to Tokyo, Japan. They comprised 104 genetically unrelated, non-clinical Japanese white-collar workers in a large corporation, and they represented a high-functioning, non-clinical adult population. There were 20 ex-smokers and 17 current smokers. The participants completed the Japanese version of Kirby's monetary choice questionnaire for delayed gain (MCQ) (Kirby et al., 1999). Trained psychiatrists and psychologists conducted a short structured diagnostic interview in accordance with the Mini-international Neuropsychiatric Interview (M.I.N.I.) (Sheehan et al., 1998). This was to confirm lifetime diagnoses of affective, anxiety, and psychotic disorders, based on the criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) (American Psychiatric Association, 2000). Each participant donated a whole blood sample for DNA analysis. Thirteen of the 104 participants were excluded because of a current or past DSM-IV diagnosis, degenerated DNA, or inconsistent rating on the MCQ. Four males had been diagnosed with major depressive disorder; four participants (three males and one female) were diagnosed with bipolar II disorder; DNA samples of four participants (three males and one female) were degenerated; and one male had an inconsistent rating.
After the previously described exclusions, the study participants comprised 91 high-functioning Japanese workers, which included 19 ex-smokers and 14 current smokers. (Of the 66 male participants, 19 were ex-smokers and 12 were current smokers; of the 25 female participants, two were current smokers.) The mean age of the participants was 40.9 years (standard deviation [SD] = 6.9) and the age range was 28 to 56 years. The mean age of the males and females was 42.0 (SD = 6.6) and 38.0 (SD = 6.8) years, respectively, with an age range of 28 to 54 years for the males and 28 to 56 years for the females. There were no significant differences between the 91 participants and the 13 excluded participants in sex, age, smoking status, or MCQ scores.
The aim of the present study was clearly explained to all participants. They provided written, informed consent. The ethics committee of the Faculty of Medicine at the University of Tokyo approved the study, and the study conformed to the provisions of the Declaration of Helsinki.
Measurement of time-discount rate with Kirby's monetary choice questionnaire (MCQ)
Researchers in neuropsychopharmacology, psychoneuroendocrinology, and behavioral neuroeconomics studies have repeatedly observed that the hyperbolic discount function (Takahashi, 2009) well describes delay-discounting behavior (i.e. intertemporal choice) in human and animal subjects. The hyperbolic discount function is expressed as V(D) = 1/(1 + kD), in which V(D) is the subjective value of a delayed reward at delay time (D) and the hyperbolic time-discount rate (k) is a free parameter indicating a subject's impulsivity in intertemporal choice. Large k values correspond to more rapid discounting, whereas small k values indicate self-control in intertemporal choice.
The MCQ is a self-administered instrument for measuring time-discount rates (Kirby et al., 1999). It consists of 29 statements and features choices between two hypothetical sums of money of varying sizes and delays. The MCQ includes two validity statements to eliminate random responders (e.g. “[question 28] Would you prefer \1,000,000 now or \0 in 15 days?” and “[question 29] Would you prefer \0 now or \1,000,000 in a day?”). In the MCQ, 27 statements require choosing a small amount now or a large amount in the future; each choice helps in estimating the respondent's time-discount rate. High time-discount rates on the questionnaire are associated with self-reported impulsivity and real-life impulsive behaviors (Hirsh et al., 2008). The one-year temporal stability of the MCQ has been confirmed (Kirby, 2009).
In the current study, the Japanese version of the MCQ used the same procedure as previous studies (Kirby et al., 1999; Takahashi et al., 2006, 2007, 2010) for assessing time-discount rates. As in our previous studies, $1 was converted to \100 (e.g. “Would you prefer \2,700 now or \5,000 in three weeks?”) (Takahashi et al., 2007, 2010). Three time-discount rates – small, medium, and large gains, and represented by k(S), k(M), and k(L), respectively – were obtained for each participant. The geometric-mean time-discount rate, k(mean), for the three rates of the different sizes was calculated by following Kirby's procedure (Kirby et al., 1999; Kirby, 2009). We then examined the relationship between the hyperbolic time-discount rates (i.e. k[mean]) and the variations in the FKBP5 gene. Because the distribution of the time-discount rate, k, is known to be skewed, we used the log of k (i.e. ln k) in the following analysis, based on a standard analytical procedure (Kirby et al., 1999; Kirby, 2009). The same assessment procedures for time-discount rates have recently been utilized in a structural neuroimaging study (Yu, 2012).
Single nucleotide polymorphism (SNP) selection and genotyping
Genomic DNA was isolated from leukocytes in whole blood by using the Wizard genomic DNA purification kit (Promega Corporation, Madison, WI, USA) (Promega Corporation 2005). In the present study, we selected five SNPs in the FKBP5 gene: rs3800373, rs755658, rs1360780, rs1334894, and rs4713916. These SNPs have been associated with depression, antidepressant response, and hippocampal volume in previous studies (Zobel et al., 2010). They were genotyped at the Human SNP-Typing Center (University of Tokyo, Tokyo, Japan) by using the DigiTag2 assay (Nishida et al., 2005, 2007), which is designed for multiplex SNP typing. The assay was used in accordance with the manufacturer's protocol (Olympus, Tokyo, Japan).
The Hardy-Weinberg equilibrium (HWE) for genotype distributions was assessed using χ2 test with PLINK version 1.07 (Purcell et al., 2007). As the SNP-based quantitative trait association analysis for an additive model, linear regression analyses using single markers for the ln k(mean) values of the MCQ were performed by using PLINK. Family-wise (i.e. SNP-wise) corrected empirical P-values were calculated on the basis of 100,000 permutations (Good, 2000) under the control of the family-wise error rate (FWER) for each SNP examined (Hochberg and Tamhane, 1987) as a function of PLINK.
The coefficient of determination (R2) was obtained through regression analysis. R2 expresses the contribution ratio and represents the effect size in regression analysis (Cohen, 1988; Field, 2005). R2 was categorized approximately as small (0.01 ≤ R2 < 0.1), medium (0.1 ≤ R2 < 0.3), or large (0.3 ≤ R2). Statistical power was estimated by using QUANTO version 1.2.4 (Gauderman, 2002). Statistical analyses of mean values, SD, R2, χ2 tests, t-tests, and regression tests for demographic data and the time-discount rates were conducted with SPSS 16.0.2J for Windows (SPSS Incorporated, Chicago, IL, USA) (SPSS Incorporated 2007). Statistical tests were two-tailed and the significance level was set at P < 0.05.
Table 1 lists the five genotyped SNPs, and Figure 1 depicts the linkage disequilibrium (LD) plots. All SNPs had genotyping call rates greater than 0.90. Their HWE P-values were greater than 0.01, indicating that none of their genotype distribution severely deviated from the equilibrium. Most SNPs were found in strong LD. The minor allelic frequencies (mAFs) of rs755658 and rs1334894 were less than 0.05. These two SNPs were therefore excluded from all subsequent analyses. As a result, the three remaining SNPs (i.e. rs3800373, rs1360780, and rs4713916) were tested.
Table 1. Results of FKBP5 SNP genotyping
SNPs rs755658 and rs1334894 were excluded from the subsequent analysis because their mAFs were less than 0.05.
FKBP5, FK506 binding protein 5; mAF, minor allelic frequency; P-HWE, P-value of the Hardy-Weinberg equilibrium; SNP, single nucleotide polymorphism; UTR, untranslated region.
Table 2 lists the results of the SNP linear regression analysis. The SNP rs1360780 showed an SNP-wise (i.e. multiplicity-corrected) empirically significant association with the ln k(mean) of the MCQ (P = 0.047 with a small effect size), indicating that participants having more minor alleles (T) in rs1360780 are less impulsive in intertemporal choice for delayed gain. The other two SNPs exhibited nonsignificant levels (P > 0.1). No significant association was observed between haplotypes comprising the three SNPs and values of ln k(mean) of the MCQ after correcting for multiple testing by the haplotype-based quantitative trait association analysis by using PLINK (data not presented).
Table 2. Results of SNP linear regression analysis for the hyperbolic time-discount rate
Note: The SNP rs1360780 shows a family-wise-corrected significant association with the hyperbolic time-discount rate, ln k(mean).
β, regression coefficient; 95% C.I., 95 percent confidence interval of β; R2, coefficient of determination; SNP, single nucleotide polymorphism; PEMP1, empirical P-value on the basis of 100,000 permutations; PEMP2, family-wise (SNP-wise) corrected PEMP1.
The statistical power of rs1360780 was estimated at 0.66 for n = 91, mAF = 0.274, R2 = 0.060, mean ln k (SD) = −5.681 (1.971), and two-tailed nominal P < 0.05. Furthermore, the smoking status did not affect the time-discount rate (i.e. ln k[mean]) in the present population (β = 0.095; P > 0.7). This was consistent with our previous report indicating that the intertemporal choice did not differ between mild smokers and nonsmokers in the Japanese population (Ohmura et al., 2005).
Our study is the first to demonstrate that a genetic variation in the FKBP5 gene is associated with impulsivity in intertemporal choice (i.e. delay discounting). We observed that a minor allele (T) in rs1360780 was negatively associated with impulsivity in intertemporal choice for delayed gain.
The significant SNP rs1360780 is adjacent to the hormone response element (HRE)-binding sequence where transcription factors bind and alter the transcription of the FKBP5 gene (Hubler and Scammell, 2004). The protein FKBP5 is a co-chaperone of the GR and plays a role in regulating the stress hormone-regulating HPA system. Previous studies indicate that the minor homozygous genotype (T/T) of rs1360780 was associated with increased FKBP5 expression in human blood monocytes (Binder et al., 2004) and that the in vitro overexpression of human FKBP5 reduced hormone binding affinity (Denny et al., 2000) and nuclear translocation of the GR (Wochnik et al., 2005). Therefore, rs1360780-associated FKBP5 overexpression causes GR resistance (Scammell et al., 2001; Binder et al., 2004), which is accompanied by increased plasma cortisol levels (Binder et al., 2008). High cortisol levels have been associated with a small time-discount rate, indicating that subjects with high cortisol levels are less impulsive in intertemporal choice (Takahashi, 2004). Taking these findings together, it is rational to deduce that participants with the more minor alleles (T) of rs1360780 in the FKBP5 gene are less impulsive in intertemporal choice (i.e. delay discounting).
The FKBP5 protein is expressed in the striatum (Nilsson et al., 2007), the prefrontal cortex (Costin et al., 2012), and the hippocampus (Scharf et al., 2011). These brain regions reportedly mediate impulsivity in intertemporal choice (Takahashi, 2009; Sellitto et al., 2011; Yu, 2012). It is possible that the genetic variance in the FKBP5 gene regulates intertemporal choice by modulating stress systems and neural functioning in brain regions such as the striatum and the hippocampus.
The genetic polymorphism of rs1360780 in the stress-related FKBP5 gene is reportedly associated with the vulnerability of adults to suffer PTSD symptoms after child abuse (Binder et al., 2008). Furthermore, rs1360780 has been implicated in the relationship between childhood trauma and aggressive behavior (Bevilacqua et al., 2012). Because evolutionary psychological studies imply that male aggression may be associated with impulsivity in intertemporal choice in a complex manner (Wilson and Daly, 2004, 2006; Daly and Wilson, 2005), future studies should examine these associations. Such investigations may help to establish effective treatment for stress dysregulation in aggressive and impulsive patients.
We now address the limitations of the present study. The first limitation is that the study participants were employees of a major corporation and therefore were not necessarily representative of the general community. Our study participants likely represent a high-functioning segment of the population and the score distribution of Kirby's MCQ could be different in the general population. The second limitation is that the significance has a small effect size because of the small sample size. A larger sample will be needed to show a large effect size in future molecular neuroeconomic studies.
In conclusion, we have demonstrated that variation in the FKBP5 gene is associated with impulsive decision-making. By utilizing neuroeconomic theory, future psychiatric studies could investigate the role of the FKBP5 gene in impulsive problem behavior observed in anti-social personality disorder, addiction, and suicide.
This study was supported by Grants-in-Aid for Scientific Research (numbers 17019029 and 21300242) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.