One noted reason for varied findings concerns the stress-diathesis model. This theory suggests that an underlying predisposition exists for certain disorders/diseases, but the presence of an environmental stressor may be necessary in order for the disorder to manifest itself. Accordingly, failure to consider gene by environment interactions may lead to the erroneous conclusion that an effect is not present. The work of Caspi and Moffitt (Caspi et al. 2003; Moffitt et al. 2005) has illustrated this point with findings indicating that allelic variation in the functional polymorphism 5-HTTLPR interacts with stressful life events to predict the occurrence of depression. Specifically, carriers of the less transcriptionally efficient 5-HTTLPR short (s) allele had increased rates of major depression as a function of increased numbers of varied past life stressors. Others have replicated this interaction by examination of environmental stressors such as unemployment (Grabe et al. 2005) and low socio-economic status (SES) (Manuck et al. 2004).

Gender should also be considered when examining the genetic effects of 5-HTTLPR on psychological outcomes (Du et al. 2000). At present it is difficult to draw conclusions about the effects of gender with respect to associations among personality traits, affective disorders, and 5-HTTLPR because the consideration of gender has varied markedly across studies. Some have failed to specifically address gender differences—e.g., gender was modeled as a covariate, the participants were predominately or exclusively male or female, or there was no consideration of potential gender effects (Deary et al. 1999; Gonda et al. 2005; Hamer et al. 1999; Jacobs et al. 2006; Lesch et al. 1996; Nakamura et al. 1997). The results have been inconsistent among studies that have examined potential gender differences. In three instances (Ball et al. 1997; Gelertner et al. 1998; Greenberg et al. 2000) no significant effects for gender were found when examining the association between NEO-PI scores and the 5-HTTLPR genotype; however, one study did report that higher Harm Avoidance was associated with the 5-HTTLPR s allele in males, whereas, the s allele was associated with lower scores for females. Du et al. (2000) reported a positive association between neuroticism and the presence of short (s) alleles, but only for males. Conversely, in two different adult samples, anxiety was significantly lower among males with the s variant of 5-HTTLPR, whereas, no such association was demonstrated in females (Flory et al. 1999). Furthermore, other findings indicate that the 5-HTTLPR genotype may interact with both gender and environmental stress with respect to psychological out-comes (Grabe et al. 2005, Sjoberg et al. 2006).

Race is yet another characteristic that must be taken into account when examining the effects of genes on personality traits and psychiatric disorders (Gelernter et al. 1999). Gelertner et al. (1998) reported that Caucasians with the 5-HTTLPR s allele had higher scores on neuroticism compared to those with the l/l genotype, whereas the opposite pattern of findings was reported for African Americans. Results from samples of Polish (Samochowiec et al. 2001) and Japanese individuals (Katsuragi et al. 1999) suggest that the 5-HTTLPR s allele is associated with higher scores on harm avoidance. Conversely, no association was found in Finnish (Mazzanti et al. 1998), Israeli (Ebstein et al. 1997), and Japanese (Nakamura et al. 1997) samples. Finally, increased anxiety has been significantly related to the 5-HTTLPR s allele in a Swedish study (Melke et al. 2001), while results from an Australian study (Jorm et al. 1998) did not show this association.

Related work from our lab has examined whether the functional polymorphism 5-HTTLPR is associated with CNS serotonin turnover—as indexed by cerebrospinal fluid levels of 5-hydroxyindoleacetic acid (5-HIAA) (Williams et al. 2003). A significant race by genotype interaction was found, indicating homozygosity for the 5-HTTLPR s allele is associated with higher CSF 5-HIAA levels in African Americans, but with lower levels in Caucasians. In addition, a significant gender by genotype interaction indicated that the 5-HTTLPR s/s genotype is associated with higher 5-HIAA levels in women, but with lower levels in men. These findings suggest that effects of the 5-HTTLPR polymorphism on CNS serotonergic function varies as a function of race and gender, further indicating that, as highlighted above, race and gender should be taken into account in research evaluating effects of 5-HTTLPR on behavioral functions that are regulated by CNS serotonergic function. We are aware of only one study that has included gender as a potential moderator of the interaction between 5-HTTLPR and stress as predictors of depressive symptoms (Sjoberg et al. 2006). Findings from their sample of boys and girls aged 16–19, indicated that females homozygous for the 5-HTTLPR s allele, in conjunction with childhood stress, were more likely to develop depression as compared to girls with no such history, or the presence of the long (l) allele. Moreover, their results also suggested that males with the s allele and traumatic backgrounds are protected from depression.

In the present paper, we examined the association of 5-HTTLPR genotype with depressive symptoms in the following two independent samples: (1) Study 1: participants who were caregivers of a relative with Alzheimer’s Disease or other major dementia and non-caregiver controls, and (2) Study 2: participants of varying childhood SES as indexed by father’s educational background. Consistent findings demonstrate that the burden associated with providing care for a loved one suffering from dementia is considerable (Schulz et al. 1995), and has been associated with a number of negative health outcomes (Baumgarten et al. 1992), including worse ratings of physical health, significant decreases in cellular immunity (Kiecolt-Glaser et al. 1991), higher levels of triglycerides (Vitaliano et al. 1995), and increased rates of early mortality, (Schulz and Beach 1999). We have also previously shown, in participants from Study 1, that caregiving is associated with increased symptoms of depression, poorer sleep quality, and increased glucose, as compared to non-caregivers (Brummett et al. 2006, 2005).

The stressor examined in Study 2 pertains to the negative impact of adversity during childhood. Because the risk of poor health is cumulative, starting during childhood and ultimately resulting in disease during adulthood (Power et al. 1999), consideration of factors that occur throughout the lifecourse is important. Low SES during childhood is a stressor that has been associated with poor health, e.g., depression, substance abuse, and early mortality (Bergstrom et al. 1996; Garner 1997; Shilling et al. 2007). Moreover, there is evidence to suggest that father’s educational level is an index of childhood SES that predicts cardiovascular mortality and ischemic heart disease, independently of adult SES (Strand and Kunst 2006).

In Study 1, the mean score for female caregivers homozygous for the s allele was in the clinical range for depressive symptoms (i.e., 17.8); and was almost 1 standard deviation higher than the mean score for all other females in the study combined; Cohen’s (Cohen 1988) d = 0.78. Similarly, male caregivers homozygous for the l allele had CES-D ratings that were just below the clinical range (i.e., 15.9), and the mean CES-D score for this group was greater than 1 standard deviation higher than all other males in the study (Cohen’s d = 0.94). For female caregivers, genotype accounted for 4% of the variance in CES-D scores; and for male caregivers, genotype accounted for 13.2% of the variance in CES-D scores.

This study expands upon a growing literature showing that 5-HTTLPR genotype can have an opposite effect on behavioral characteristics in men and women. In a sample of 128 females, participants carrying the s allele had significantly higher ratings on a measure of subthreshold depression, as compared with those who carried the ll genotype (Gonda et al. 2005). Although not present in males, Eley et al. (2004) found a significant 5-HTTLPR × environmental stress interaction in females such that the proportion of depressed subjects who were carrying the s allele, coupled with high environmental stress, was higher than the proportion of non-s allele carriers and those with low environmental stress. Flory et al. have shown that anxiety was significantly lower among males with the short variant of 5-HTTLPR, whereas, no association was present in females (Flory et al. 1999). Likewise, in prior work on another sample we found that men, but not women, with the s variant of the 5-HTTLPR reported lower Anxiety (Brummett et al. 2003). In another study of 374 females, stressful life events interacted with the 5-HTTLPR genotype to predict symptoms of depression such that those with the s allele had significantly higher depression ratings, as compared to those with the l allele (Jacobs et al. 2006). Similarly, in a predominately female sample, lower expressing 5-HTTLPR alleles were associated with increased severity of major depression in those with moderate to severely stressful life events (Zalsman et al. 2006b). Additionally, results from a general population study indicate that for females, but not males, the s allele was associated with perceived mental and physical distress in unemployed participants as compared to those who were employed (Grabe et al. 2005). Finally, in a sample of 81 boys and 119 girls, Sjoberg et al. (2006) found results strikingly similar to those of the present study. Specifically, their findings suggest that females homozygous for the 5-HTTLPR s allele, in conjunction with the stressor of family conflict, are more likely to develop depression than girls with no family conflict, or the presence of the l allele; whereas, for males the s allele in conjunction with a traumatic background serves to protect an individual from depression. Thus, a similar pattern of gender X 5-HTTLPR X environmental stress effects on depression, and/or perceived stress, has now been demonstrated in 4 independent samples. The present results do, however, contrast with those of Caspi et al. (2003) who have shown in a sample of both males and females, that the s allele interacts with environmental stress to increase probability of depression later in life.

Females have higher rates of depression than males (Piccinelli and Wilkinson 2000) and a number of findings suggest that sex differences in serotonergic function may contribute to the enhanced risk in women. Selective serotonin re-uptake inhibitors (SSRI) efficacy is higher in females than in males (Kornstein et al. 2000), and conversely, the emergence of depressive symptoms after tryptophan depletion is significantly greater in women (Booji et al. 2002; Moreno et al. 2006). One meta analysis has even suggested that tryptophan depletion causes depressive symptoms only in women (Jans et al. 2007). The specific sex differences in serotonergic function which contribute to vulnerability to depressive symptoms are not clear, although there are some plausible candidates. Rates of central nervous system (CNS) serotonin synthesis are approximately 50% higher in males than in females (Nishizawa et al. 1997). Our studies and others (Jonsson et al. 2000; Williams et al. 2003) have shown that CSF 5 HIAA is higher in women than in men, although this could reflect either enhanced serotonin neuron firing or enhanced metabolism. Both 5-HT_1A and 5-HT₂ receptor density is lower in the brains of women than men (Biver et al. 1996; Costes et al. 2005), although such a difference could reflect either lower receptor expression or lower serotonin release to compete with radioligand for binding. How these characteristics combine to produce serotonergic “vulnerability” that leads to depression is not clear, but the consistency of the sex differences is suggestive.

The initial findings of Lesch et al. (1996), that polymorphisms in the serotonin transporter may be associated with normal variation in personality traits, has prompted an exciting area of research. One ongoing controversy about the mechanism by which the 5-HTTLPR influences behavior is the relative importance of neurodevelopmental effects that influence the development of serotonergic circuitry for current transporter expression. Reports vary from the s/s genotype exhibiting more (van Dyck et al. 2004), less (Heinz et al. 2000) or no difference (Shioe et al. 2003; Willeit et al. 2001) from s/l or l/l genotypes in serotonin transporter binding. A pair of exciting recent studies from one research group have particular relevance for the present study. In the first study, they confirmed their previous finding that overall 5-HTTLPR genotype is not associated with serotonin transporter expression in human brain, even when considering the recent triallelic categorization of genotypes (Parsey et al. 2006). However, this same laboratory reported a strong association between the low-expressing 5-HTTLPR genotype and increased vulnerability to depression in persons experiencing high levels of stressful life events. (Zalsman et al. 2006a). Although they did not report a sex difference in this association, 71% of their depressed subjects were female and this association was detected with patients describing either early life and current stress, as in the present study. It is tempting to speculate the neurodevelopmental events as well as transporter expression may play a critical role in the association we report here among sex, stressful life events and depression vulnerability.

Studies of ovarian steroid modulation of serotonergic function in non-human primate brain further support the existence of functionally important sex differences in females that could influence the emergence of depressive symptoms. In female Macaques, estrogen administration has been shown to decrease mRNA levels of the 5-HT_1A autoreceptor, monoamine oxidase A and the 5-HTT and increase mRNA of tryptophan hydroxylase in dorsal raphe and hypothalamic nuclei (Bethea et al. 2002; Gundlah et al. 2002; Pecins-Thompson and Bethea 1999) Using protein levels to index expression, estrogen treatment was found to increase tryptophan hydroxylase 2 and 5-HTT expression and decrease 5-HT_1A and 5-HT₂ receptor expression (Lu and Bethea 2002; Smith et al. 2004). While more research will be required to fully understand effects of ovarian steroids on serotonergic functions, there are clearly effects on serotonin synthesis, its removal from the synapse, its metabolism and also its stimulation of pre and postsynaptic receptor populations that strongly support a role for ovarian steroids in regulating almost every gene, its mRNA and protein that controls extracellular serotonin.

Research in rodent models also has documented sex differences in the effects of variation in serotonin transporter function on serotonergic function that could influence the emergence of depressive symptoms. In serotonin transporter knockout mice, the 5-HT_1A autoreceptor is desensitized more extensively in females than males (Bouali et al. 2003; Li et al. 2000). Similarly, in 5-HT_1B knockout mice, females exhibit greater disinhibition of serotonin release which manifests as less depressive behavior than 5-HT_1B knockout males. These findings suggest that the presynaptic 5-HT_1A and 5-HT_1B autorecpetors contribute differentially more to regulation of extracellular serotonin in intact females than in intact males.

Finally it is important to consider the possibility that sex differences in effects of 5-HTTLPR interactions with stress on depression could reflect sex differences in stress reactivity that affect serotonergic function. In humans, an extensive literature supports the contribution of early life stressors to the development of depression, an association that is particularly strong in women who have been sexually or physically abused (Heim et al. 2004). Similarly, female macaques who carry the 5-HTTLPR s allele and who have been exposed to early adversity (peer rearing) exhibit lower cortisol responses to stress, a pattern that has been associated with certain stress-related neuropsychiatric disorders (Barr et al. 2004). More research will be required to determine how these mechanisms might account for the opposite effects of the 5-HTTLPR s allele on depressive symptoms that we observe in men vs. women exposed to stressful life circumstances.

Several factors should be noted with respect to the present findings. In addition to moderation by environmental stress, we hypothesized a main effect for 5-HTTLPR with respect to depressive symptoms that was not supported in either study. This highlights the importance of continued examination of potential moderators of genetic effects. Although a significant 3-way interaction was observed with respect to gender × genotype × stressor, this finding resulted under conditions of limited power. Importantly, our inability to detect a similar 3-way interaction with respect to race may have been due to the fact that the present samples were underpowered with regard to detection of potential race effects. Finally, it should be noted that the allele frequencies differed somewhat between the two present studies, a factor that was likely influenced by the differences in racial composition. Thus, these results are to be interpreted with caution prior to replication in additional samples.

Recent research (Hu et al. 2006) has revealed the existence of a common single base substitution (A → G) within the 5-HTTLPR L allele, with the rarer (10–15% in Caucasians, 24% in African Americans) L_G allele showing reduced transcriptional efficiency, comparable to that of the S allele, while the L_A allele is about twice as transcriptionally efficient as the S or L_G alleles. The presence of the less functional L_G allele within the LL or LS subjects of our study would dilute the effects of the more active L_A allele, making it harder to find an association between 5-HTTLPR genotype and depressive symptoms. It is likely, therefore, that our findings represent a conservative estimate of the impact of 5-HTTLPR genotype on depressive symptoms in persons exposed to stressful life circumstances.

The present study contributes to an emerging literature showing that the 5-HTTLPR genotype and stress interact to affect depression vulnerability by demonstrating that this association is moderated by gender. Our findings, in two independent samples, combine with those of Sjoberg et al. (2006) to suggest that—whether the stressor is early in life or in adulthood—in women, the 5-HTTLPR s/s genotype confers increased risk of depression following exposure to stressful life circumstances, while in men the l/l genotype confers increased risk. It will be important to pursue this finding by exploring the relative importance of current versus developmental contributions of 5-HTTLPR genotype.