| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychology (K.C.G.), American University; Department of Psychiatry (T.S.C., J.C., H.R.K.), University of Connecticut School of Medicine; Department of Psychology (S.A.), Fairleigh Dickinson University; Department of Community Medicine (H.T.), University of Connecticut School of Medicine.
Address correspondence and reprint requests to Howard Tennen, Department of Community Medicine, MC 6325 263 Farmington Avenue, Farmington, CT 06030-6325. Email: tennen{at}nso1.uchc.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
Methods: A total of 350 college students recorded their daily stressors and mood for two 30-day periods, separated by 1 year.
Results: Across both years, diathesis-stress patterns were observed for reports of anxious mood as a function of 5-HTTLPR. Individuals with at least one copy of the S or LG allele at 5-HTTLPR experienced elevated anxious mood on days with more intense stressors, as compared with those who were LA homozygotes. Genotype differences in anxiety were less apparent on low stress days. No consistent allelic association of 5-HTTLPR was observed with any other mood states, trait anxiety, or neuroticism.
Conclusion: Our findings highlight the potential value of focusing on genetic vulnerability in the context of everyday environmental triggers.
Key Words: serotonin transporter • anxiety • stress • reactivity • daily diary
Abbreviations: 5-HTTLPR = serotonin transporter gene promoter polymorphism; 5-HTT = serotonin transporter protein; fMRI = functional magnetic resonance imaging; STAI = State-Trait Anxiety Inventory.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
Lesch et al. (3) first showed that individuals with one or two copies of the S allele (either L/S or S/S genotype) reported higher trait levels of neuroticism than those homozygous for the long allele (L/L). However, subsequent studies of the relationship between 5-HTTLPR and anxiety-related personality traits have produced mixed results (4–9). Mixed results could be due to a combination of factors, including the type of personality measure used. For example, two meta-analyses found no overall association between 5-HTTLPR and anxiety-related personality traits in 26 studies (with approximately 17 overlapping across the two meta-analyses); however, both analyses found significant associations in those studies using the NEO neuroticism measure, although the associations were modest (10–12).
Mixed results regarding the relationship between genotype and anxiety-related traits may also occur because 5-HTTLPR is functionally triallelic, in contrast to the biallelic categorization often reported (13–17). Due to the presence of a single nucleotide polymorphism within the L allele, it can be further divided into one of two variants: LA or LG. The lower expressing LG allele is functionally equivalent to the S allele in terms of promoter activity and RNA levels produced in cell culture systems (16), and is therefore appropriately grouped with the S rather than the LA variant. A number of researchers have argued that the heavy reliance on the biallelic approach in the literature might explain some of the discordant findings with respect to the association of 5-HTTLPR with anxiety and depression phenotypes (13,15–18). Recent research using the triallelic approach has shown that 5-HTTLPR variation is associated with severity of depression (13), response to treatment with a selective serotonin reuptake inhibitor in patients with social anxiety (17), and response of remitted depressed patients to tryptophan depletion (15). Among individuals with no psychiatric history, tryptophan depletion resulted in increased anxiety scores among those with at least one copy of the S or LG allele, but not those with the LALA genotype. Grouping the S and LG alleles together could improve the prediction of anxiety-related outcomes as a function of 5-HTTLPR.
Another possible explanation for the mixed results is that genetic vulnerability might manifest in subjective markers of anxiety, but only when activated by stress. Many studies of anxiety assess functioning under conditions of no or little stress. However, animal research as well as functional magnetic resonance imaging (fMRI) investigations in humans strongly suggest that 5-HTTLPR differences in anxiety may be best be observed under conditions of immediate threat. In 5-HTT knockout mice, for example, there is evidence of exaggerated observable stress responses and anxiety-like behaviors under conditions of danger (19). In humans, the 5-HTTLPR S allele predicts greater fMRI-observed amygdala reactivity in response to threatening visual stimuli (20,21–24). Given that the amygdala plays a key role in behavioral arousal and emotional states such as fear, this work suggests that variation in 5-HTTLPR may be associated with acute increases in anxiety triggered in response to stress. In nonstressful conditions (e.g., a neutral laboratory situation), the association between 5-HTTLPR genotype and anxiety could be missed.
Recent research provides increasing support for this diathesis-stress framework, showing that variation at 5-HTTLPR is associated with adverse clinical outcomes, but only in the context of elevated life stress (25,26). Whereas most of this work on gene-environment interplay has focused on "macro-level" stressors (e.g., large-scale events such as divorce, job loss), we offer yet another possibility—that diathesis-stress processes might also occur in the context of "micro-level" stressors, i.e., events that happen as individuals navigate their everyday lives. According to this perspective, although on low stress days, individuals with at least one copy of the S or LG allele should experience levels of anxiety similar to LA homozygotes, genotype differences in anxiety should emerge as individuals encounter and react to stressful circumstances. To date, no research has tested the association between 5-HTTLPR and the subjective experience of anxiety in this manner.
This gap in the literature is important, particularly in light of research indicating that daily stressors are often more influential with respect to ongoing psychological health and well-being than major life stressors (27–29). The chronic activation of anxiety through low-grade, everyday stressors might be just as "risky" to vulnerable individuals (S and LG carriers) as the experience of major life stressors. Furthermore, a twin study recently showed that this everyday mood reactivity is heritable (approximately 12% of variance attributable to genetics) (30). Although no study has yet documented specific genetic susceptibility to excessive stress reactivity, from the twin study, it was suggested that 5-HTTLPR genotype should be explored as a potential underlying vulnerability (30).
Overview
Our research aimed to fill these gaps in the literature by examining the association of the triallelic 5-HTTLPR with reports of anxious reactions to stressors in daily life. We used a daily process approach (31), in which we assessed participants’ experiences repeatedly over time in daily life. This strategy allows us to understand not just average levels of mood in daily life but also how mood changes in association with naturally occurring stressors. Specifically, we are able to track how anxiety changes on days with more intense versus less intense stressors, called "mood reactivity," and to test whether variation at 5-HTTLPR moderates stress reactivity patterns.
We hypothesized that, compared with LALA individuals, those with S or LG alleles would report elevated daily anxiety on days with greater daily stress (i.e., greater "anxiety reactivity"). In addition to anxiety, we assessed other daily mood states to test whether genotype differences extend to a range of emotions or were specific to anxiety. We also included trait-level measures of anxiety and neuroticism to compare the relative sensitivity of trait versus daily process measures of anxiety and their association with 5-HTTLPR. Finally, given that the literature on 5-HTTLPR and anxiety is marked by inconsistent research findings, it is important to demonstrate consistency of findings across time and contexts. For this reason, we repeated the daily diary portion of our design with our participants 1 year later. Demonstrating that genotype is associated with anxiety reactivity across years will allow us to establish the stability of the genotype-reactivity relationship over time.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
2 = 6.04) and slightly younger (18.7 versus 19.0 years, p < .01). To avoid potential confounds due to population stratification, i.e., racial/ethnic differences in allele frequencies and outcomes, we limited our analysis to non-Hispanic Caucasian subjects (n = 361), as is common practice (17). We also excluded an additional 11 participants for insufficient data.2 Of the 350 participants with sufficient data, 306 provided data in both years of the study, 27 participants provided data in year 1 only, and 17 participants provided data in year 2 only. Therefore, the sample size was 333 in year 1, and 323 in year 2. At study entry, the average age of the 350 participants was 18.7 ± 0.9 years and most were freshman (60%) or sophomores (33%). After describing the behavioral study to the subjects, we obtained written informed consent. Separate informed consent was obtained for the genetic substudy.
Procedure and Measures
The data were collected across two academic years: 2003 to 2004 and 2004 to 2005. Study procedures were approved by the Institutional Review Board of the University of Connecticut. In each of two years, participants completed a baseline assessment 1 to 2 months after the start of the fall or spring semester by logging onto a secure Web site. The baseline assessment included demographics and several personality and health questionnaires, including the 60-item NEO Five-Factor Inventory (32), which we used to compute the 12-item measure of neuroticism (items were on a 1–7 scale and were averaged, 
year 1 = 0.87; year 2 = 0.88), as well as the State-Trait Anxiety Inventory (STAI) (33), a 20-item measure of general and long-standing anxiety (items were on a 1–4 scale and were averaged, year 1 = 0.93; year 2 = 0.92).
Approximately 2 weeks after completing this initial assessment, participants began the daily diary procedure. Using a secure Web site, they logged on daily to complete a 5-minute survey between 2:30 PM and 7 PM for 30 days. Each diary contained questions about a range of health-related behaviors that were rated as 1 (very slightly/not at all) to 5 (extremely). We averaged reports of "jittery" and "nervous" for a measure of current anxiety (year 1 = 0.68; year 2 = 0.75); "hostile" and "angry" for hostility ( = 0.73, 0.78); "sad" and "dejected" for depressed mood ( = 0.72, 0.76); and "enthusiastic," "cheerful," "happy," "relaxed," "excited," and "content" for positive mood ( = 0.91, 0.92). Later, in the diary, participants were asked to think about their "most negative experience that day, no matter how small." There was a wide range of events reported, from more intense events such as receiving a bad grade or having a fight with a friend/romantic partner to more minor events such as waking up late. For a measure of negative event stress, they were asked to rate how stressful this event was from 1 (not stressful at all) to 7 (extremely stressful). The same protocol was followed in the second year of the ongoing study. Our sample consisted of participants who completed the initial battery and at least one diary phase to criterion (i.e., with 
15 complete records).
After the diary procedure in year 2, all participants were sent an e-mail inviting them to participate in a genetic substudy. Participation in the genetic study was voluntary and did not affect eligibility to continue in the larger study. Small group sessions of six to eight participants were held for DNA collection. After providing informed consent, participants were asked to "swish" Scope mouthwash (Proctor and Gamble Co., Cincinnati, OH) for 20 to 30 seconds, and then to spit the mouthwash into a collection tube, which was coded with a unique ID different from participants’ study ID.
Genotyping Procedure
Genomic DNA was extracted from the mouthwash-stabilized samples using a commercial DNA isolation protocol (PureGene, Gentra Systems, Minneapolis, Minnesota). The 5-HTTLPR polymorphism was genotyped using a TaqMan 5'nuclease assay modified from that originally described by Hu and colleagues (34). A more detailed description of our genotyping procedure can be found in the work of Covault et al (35). The number of l-alleles (0, 1, or 2) for each subject was identified by examination of scatterplots of end-point Fam versus Vic fluorescence levels captured using an ABI 7500 Sequence Detection System (Applied Biosystems, Foster City, California). Fifteen percent of samples were repeated with no discrepancy in genotype between assays. A second TaqMan (Applied Biosystems) 5'nuclease allelic discrimination assay served to distinguish LA versus LG alleles by using the same primers and amplification conditions as for the L versus S allele assay but using LA versus LG allele-specific probes, (6FAM-CCCCCCTGCACCCCCAGCATCCC-MGB and VIC-CCCCTGCACCCCCGGCATCCCC-MGB, respectively). We did not observe the G allele in samples from S allele homozygotes, consistent with the findings of Hu and colleagues (34).
Statistical Analysis
Because the LG and S alleles are functionally equivalent, they were grouped together as S’ (lower expressing allele); the LA allele was designated as L’ (13,16,17). To test for genotype differences in trait variables (neuroticism/anxiety), a linear regression analysis was performed in SPSS with a Helmert coding procedure. Two codes were entered as simultaneous predictors, yielding two separate comparisons: 1) the L’/L’ group versus L’/S’ and S’/S’ groups, coded –2, 1, 1 respectively; and 2) the L’/S’ group versus the S’/S’ group, coded 0, –1, 1. This coding approach was more conservative than treating genotype as a single quantitative predictor with equal intervals, which assumes an additive relationship between genotype and outcomes. The test of the first coefficient (L’/L’ versus S’ groups) reflects the common finding in the literature that the lower expressing allele has a dominant mode of action (16). The second contrast code, however, allowed us to test whether the L’/S’ and S’/S’ groups also differed.
To analyze the effect of genotype on average daily mood and daily mood reactivity, we relied on a multilevel modeling approach, using Hierarchical Linear Modeling software (with restricted maximum likelihood estimation procedures) (36). The daily diary data have a nested structure, in which the 30 repeated daily observations (level 1) are nested within persons (level 2). To test for differences in average daily stress and mood, we entered each daily variable as a level 1 outcome with no level 1 predictors using an intercept-only model. Genotype was entered as a level 2 predictor of the level 1 intercepts using the two Helmert codes (L’/L’ versus S’ groups; L’/S’ versus S’/S’).
For the mood reactivity analyses, we used multilevel modeling procedures to determine the association between daily stressors and mood, and how this association varied as a function of genotype. A regression equation was generated for each individual at level 1, with daily reports of event stress as the level 1 predictor and anxiety (or hostility, or depressed mood, etc.) as the level 1 outcome.3 This within-person association, or slope, b1, indicated the change in anxiety for every 1-unit increase in perceived stress. Steeper slopes indicated greater mood reactivity (e.g., more observed changes in anxiety as a function of perceived stress). At level 2, we modeled individual differences in the level 1 slopes, b1, as a function of the 5-HTTLPR genotype Helmert codes. We predicted stronger anxiety reactivity for the S’ groups. To account for the number of analyses conducted, we used an level of 0.01 to designate significant findings, and we report trends at the p < .05 level.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
2 = 0.05, p = .83). The mean neuroticism score across all participants was 3.57 (standard deviation (SD) = 1.05; t score = 57.6), and the mean STAI score was 2.05 (SD = 0.45; t score = 55.4), with similar values in year 2. Table 1 presents the mean scores for trait and daily variables for each of the genotypes, as well the genotype main effects. Because we were most interested in the L’/L’ versus S’-carrier group comparison, we report the results of this contrast in all tables (L’/S’ and S’/S’ genotypes are labeled "S’ groups"). Where the second contrast (L’/S’ versus S’/S’) is significant, we note the findings in the text. As shown in Table 1, S’ carriers did not differ from L’L’ individuals on either neuroticism or trait anxiety as measured by the STAI. In fact, the S’/S’ genotype seemed to have the lowest N scores in year 1, although this difference was not statistically significant when comparing the S’/S’ against the L’/S’ group (Helmert code 2; b = –0.132, p = .06).4 For the daily variables, 5-HTTLPR genotype was not associated with perceived intensity of daily stressors in either year, nor was it associated with average ratings of anxiety or hostility. Interestingly, there was a trend suggesting that S’/S’ individuals reported lower depressed mood (Helmert code 2: b = –0.053, p = .03, 0.5% variance explained) and higher positive mood (Helmert code 2: b = 0.102, p = .01, 1.3% variance explained) than L’/S’ individuals in year 1. However, these effects were not significant in year 2 (p = .18 and .40, respectively).
|
Table 2 presents the results for the reactivity analyses. The typical within-person association between stress and mood outcomes (b slopes) are presented separately by genotype group. The t coefficients reflect the test for Helmert code 1, comparing the L’/L’ group with those in the S’ groups (L’/S’ and S’/S’). As predicted, genotype moderated the mood reactivity relationship for anxiety in the expected direction across both years. As seen in data rows 1 and 5 of Table 2, L’/S’ and S’/S’ individuals reported greater increases in anxiety on days with more intense stressors, as compared with individuals with the L’/L’ genotype, who showed weaker increases in anxiety in the face of similarly intense daily stressors. The second contrast code was nearly significant in year 1, b = 0.013, p = .04 (in year 2, b = 0.014, p = .051), suggesting that S’/S’ individuals were more reactive than L’/S’ individuals. Figure 1 illustrates these patterns of diathesis-stress. In the context of low daily stress, the three genotype groups reported similar anxiety levels. However, as daily stress increased, genotype differences emerged. On high stress days, individuals with at least one S’ allele reported higher anxiety than the L’/L’ group. 5-HTTLPR genotype accounted for 5.6% of the variance in these stress-anxiety slopes in year 1, and 7.2% of the variance in year 2. It is noteworthy that when these analyses are repeated using a biallelic model (L/L versus L/S and S/S), the variance explained is considerably lower (3.2% in year 1 and 2.2% in year 2).
|
|
Furthermore, patterns associated with 5-HTTLPR seemed specific to anxious mood. There were only trends suggesting that the L’/L’ group was less reactive for hostile and depressed mood in year 2, but not in year 1, and there were no genotype differences in positive mood reactivity.
Supplemental Analyses
Although we did not find an association between genotype and neuroticism, neuroticism itself has been consistently related to daily anxiety and excessive mood reactivity in response to daily stressors (37,38). Thus, in supplementary analyses, we assessed whether genotype differences continued to explain variance in daily mood and reactivity above and beyond that accounted for by neuroticism. We repeated all analyses with genotype and neuroticism as simultaneous level 2 predictors. Neuroticism was associated with greater emotional reactivity to daily stress for most mood states in the expected direction. Genotype, however, continued to predict anxiety reactivity across both years even after we statistically controlled for neuroticism (5.6% and 7.8% of the variance in years 1 and 2, respectively). Results were similar when we entered STAI as a control variable instead of neuroticism. Furthermore, neuroticism did not interact with genotype in predicting daily mood levels or mood reactivity.
Finally, there were no effects of gender on the relationship between genotype and daily mood reactivity.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
Our findings with respect to 5-HTTLPR and daily stress-reactivity complement emergent research regarding the significance of environmental circumstances in the expression of genetic differences. Evidence suggests that 5-HTTLPR variation moderates the relationships between affective outcomes and major life events that have occurred over the lifetime (25), as well as those that have occurred over a more circumscribed time period (e.g., over the last month (26) or 6 months (13)). This diathesis stress pattern occurs in studies using both the biallelic and triallelic approaches to 5-HTTLPR genotype. Our study extends this work to show that genetic vulnerability might also moderate the impact of stressors at the daily or "micro" level in a diathesis-stress fashion. Our data showed that S’ groups and LA homozygotes reported similar levels of anxiety on days with less severe stressors. Genotype differences only emerged on days with more severe stressors.
This work also highlights the importance of using a triallelic approach to understanding 5-HTTLPR-related differences in anxiety. Although an association between 5-HTTLPR and anxiety reactivity was found using the biallelic approach, the triallelic model resulted in greater variance explained for anxiety reactivity (an average across the 2 years of 2.7% for the biallelic model and 6.4% for the triallelic model).
In terms of mechanism, the observed relationship between variation in 5-HTTLPR and anxiety reactivity may reflect the byproduct of heightened amygdala activity among those with the lower expressing S’ allele. Research on emotional endophenotypes has shown that S-carriers have excessive amygdala activity in response to fear stimuli (22), and dysfunctional coupling between circuits involving the amgydala and other reflexive cortical structures (ventromedial prefrontal cortex (24); perigenual cingulate (39)). Such processing differences could translate into heightened anxiety responses as one confronts a perceived threat in real-world contexts (40).
The present results also dovetail with a recent quantitative genetics study of stress-reactivity. Jacobs et al. (30) compared monozygotic and dizygotic twins in terms of their daily negative affect reactivity, as measured over 5 days using an experience sampling procedure. In their study, undefined genetic factors accounted for approximately 12% of the variance in mood reactivity to daily stressors. Our findings suggest that the 5-HTTLPR polymorphism may be an important contributor to this genetic variance, potentially contributing up to 50% of the known genetic differences in reactivity. Such conclusions are made with caution, however, given that the variance estimate from Jacobs et al. reflected general negative mood reactivity (anxiety + sadness).
It is interesting that we did not find a consistent association between 5-HTTLPR genotype and reactivity with respect to depressed mood. Research suggests that at a macro-level, in the context of life stress, S and LG carriers are susceptible to the development of depressive symptoms (13,25). Yet, our S’ group reacted with only slightly greater increases in depressed mood in year 2 (trend), and showed no differences in year 1. It is possible that over the long term, excessive anxious reactions on an everyday basis may leave S and LG carriers more vulnerable to episodes of depression. In the future, it will be helpful to clarify whether reactions to minor daily stressors affect S and LG carriers’ ability to stave off depression in the longer term.
More broadly, our research suggests that daily process approaches may be a valuable tool in behavior genetics research. In our study, variation in the gene encoding 5-HTT was not related to differences in a trait measure of anxiety or neuroticism, but was related to differences in state anxiety experienced in response to daily stress. These findings suggest that intensive, longitudinal sampling of acute affective reactions in daily life, by virtue of their capacity to capture mood that is in close temporal association to daily stressors, may be more sensitive than one-time trait reports for detecting the subjective correlates of genetic variation, at least with respect to the 5-HTTLPR polymorphism.
Limitations in the present study include the fact that, because we assessed stress and mood once per day, directionality is ambiguous in our analyses of stress reactivity. Although we interpreted significant relationships to suggest that stress results in a change in mood, it is also possible that negative moods influence perceptions of stress. Future genetic association studies of daily affective experience might profitably include other indicators of fluctuating anxiety responses, such as salivary cortisol levels. It is also important to replicate these findings in more diverse samples, as the majority of studies investigating the association of 5-HTTLPR with behavioral outcomes have been limited to European Americans.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
We gratefully acknowledge the assistance of Nicholas Maltby for Web programming. We thank David Haaga for his valuable feedback on an earlier draft of this paper.
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
2One participant could not be assigned a 5-HTTLPR genotype; nine participants were excluded for providing too few diary records (<15 records each year); one participant was excluded for noncompliant responding in the diary procedure (responding identically to all items).
3The daily event stress variable was individual-mean centered, so that changes were in relation to each individual’s mean stress level.
4A separate test using a different contrast coding procedure to compare the S’/S’ against the L’/L’ and L’/S’ groups combined showed that this difference was a trend (b = –.089, p = .049).
This article reflects joint first authorship of Gunthert and Conner.
Received for publication October 26, 2006; revision received July 16, 2007.
This study was supported by Grants P50AA03510, M01 RR06192 (University of Connecticut GCRC), K24 AA13736 (to H.R.K.), and T32 AA07290 (NIAAA postdoctoral training grant) from the National Institutes of Health.
DOI:10.1097/PSY.0b013e318157ad42
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONNOTESREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  M. C. Pfaltz, T. Michael, P. Grossman, J. Blechert, and F. H. Wilhelm Respiratory Pathophysiology of Panic Disorder: An Ambulatory Monitoring Study Psychosom Med, October 1, 2009; 71(8): 869 - 876. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
