Interaction of brain 5-HT synthesis deficiency, chronic stress and sex differentially impact emotional behavior in Tph2 knockout mice

While brain serotonin (5-HT) function is implicated in gene-by-environment interaction (GxE) impacting the vulnerability-resilience continuum in neuropsychiatric disorders, it remains elusive how the interplay of altered 5-HT synthesis and environmental stressors is linked to failure in emotion regulation. Here, we investigated the effect of constitutively impaired 5-HT synthesis on behavioral and neuroendocrine responses to unpredictable chronic mild stress (CMS) using a mouse model of brain 5-HT deficiency resulting from targeted inactivation of the tryptophan hydroxylase-2 (Tph2) gene. Locomotor activity and anxiety- and depression-like behavior as well as conditioned fear responses were differentially affected by Tph2 genotype, sex, and CMS. Tph2 null mutants (Tph2−/−) displayed increased general metabolism, marginally reduced anxiety- and depression-like behavior but strikingly increased conditioned fear responses. Behavioral modifications were associated with sex-specific hypothalamic-pituitary-adrenocortical (HPA) system alterations as indicated by plasma corticosterone and fecal corticosterone metabolite concentrations. Tph2−/− males displayed increased impulsivity and high aggressiveness. Tph2−/− females displayed greater emotional reactivity to aversive conditions as reflected by changes in behaviors at baseline including increased freezing and decreased locomotion in novel environments. However, both Tph2−/− male and female mice were resilient to CMS-induced hyperlocomotion, while CMS intensified conditioned fear responses in a GxE-dependent manner. Our results indicate that 5-HT mediates behavioral responses to environmental adversity by facilitating the encoding of stress effects leading to increased vulnerability for negative emotionality.


Introduction
Serotonin (5-hydroxytryptamine (5-HT)) is implicated in the pathophysiology of a wide spectrum of neuropsychiatric conditions, including severe depression and anxiety disorders as well as in the mechanism of antidepressant/anxiolytic pharmacotherapies. Its involvement in fundamental processes of brain development, connectivity, and plasticity identified 5-HT as a critical factor in the modulation of cognition and emotion as well as physiological and behavioral responses to stress (Lanfumey et al. 2008). While there is evidence that 5-HT moderates gene-by-environment interactions (GxE) impacting on the vulnerability-resilience continuum in neuropsychiatric disorders (for review (Lesch 2011)), the mechanisms by which environmental inputs are encoded in the brain to modify its function and behaviors are not fully understood (McEwen et al. 2012). Studies in humans report an association of functionalTPH2 variants with personality traits (Gutknecht et al. 2007) and various neuropsychiatric disorders (for review (Waider et al. 2011)). To further dissect the role of 5-HT in complex behavior and its relevance for GxE, we generated a mouse model featuring brain-specific 5-HT deficiency resulting from a targeted inactivation of neuronal tryptophan hydroxylase-2 (Tph2) gene (Gutknecht et al. 2008). Tph2 null mutant (Tph2 −/− ) mice lack the ability to synthesize 5-HT specifically in the brain (Savelieva et al. 2008;Alenina et al. 2009;Gutknecht et al. 2009Gutknecht et al. , 2012. In contrast to other genetically engineered mice with brain 5-HT deficiency such as Pet1 KO (Hendricks et al. 2003), Lmx1bc KO (Song et al. 2011), or Tph2 R439H (Beaulieu et al. 2008), Tph2 −/− mice completely lack brain 5-HT synthesis, while their "serotonergic-like" raphe neurons differentiate and remain electrophysiologically functional (Gutknecht et al. 2012).
Here, we investigated Tph2 mutant mice with emphasis on the effect of central 5-HT deficiency on behavioral and neuroendocrine responses to environmental adversity. As an integral part of the neural circuitry underlying stress regulation, the 5-HT system has commonly been viewed to exert modulatory functions gender dependently (Jones and Lucki 2005). Tph2 −/− mice, lacking brain 5-HT synthesis and Tph2 +/− mice, with up to 30 % reduction of 5-HT concentrations in the raphe nuclei (Gutknecht et al. 2012;Waider et al. 2013), were subjected to unpredictable chronic mild stress (CMS (Willner 2005)) to assess the impact of deficient serotonergic neurotransmission in the brain on anxiety-and depression-like behavior, conditioned fear responses, and aggression as well as hypothalamic-pituitary-adrenal (HPA) axis reactivity and adaptation to CMS.
Sucrose preference (SP) was measured for five consecutive days before (baseline (BL)) and after behavioral testing and CMS re-exposure (BT/CMS) to assess anhedonia-like behavior. The experimental design (Fig. 1a) included elevated plus maze (EPM) and open field (OF) to investigate exploratory activity and anxiety-like behavior, stress-induced hyperthermia (SIH) to determine increases in body temperature following acute stress, a delay fear conditioning (FC) paradigm to assess associative learning and memory, a test for behavioral despair (Porsolt swim test (PST)) to analyze depression-like behavior, and the resident-intruder (RI) test to assess aggression in males (for details, see text S1).
To monitor stress responsiveness, fecal corticosterone metabolites (Cm) were analyzed according to (Touma et al. 2004) at five different time points: at BL, after 1 and 2 weeks of CMS (CMS1 and CMS2, respectively), after SIH, and after re-exposure to CMS for 1 week (CMS3). After completion of the experiments, mice were sacrificed and blood samples for plasma corticosterone (pCort) quantification were collected. An additional cohort of naive mice was used to determine baseline pCort levels.

Statistical analysis
Data were evaluated via three-way ANOVA with sex, group, and genotype as between-subject factors followed by Bonferroni post hoc tests for significant main effects and independent t tests for significant interactions. Repeated measures ANOVAs (with time as withinsubject factor) were performed when appropriate. Data were analyzed using IBM SPSS Statistics 21 and expressed as mean±SEM; p<0.05 was considered significant. Data were illustrated with GraphPad Prism version 5.03 for Windows, GraphPad Software, La Jolla, California USA.
A separate cohort of naive mice was used to assess daily food and water intake under baseline conditions (Fig. S2). ANOVA revealed a highly significant genotype effect for food (F (2,114) = 31.38, p < 0.001) and water (F (2,114) = 43.03, p<0.001) intake (Fig. S2), reflecting that both male and female Tph2 −/− mice consumed more food and water as compared to Tph2 +/− and Tph2 +/+ mice (all p<0.001). This finding is in line with increased total fluid intake in Tph2 −/− mice in the sucrose preference test.
Sex-specific HPA response to CMS

Retention of cued fear memory
Retention of cued fear memory was tested 24 h and 7 days after conditioning. Repeated measures ANOVA revealed a significant phase×sex (F (2,142) =7.54, p=0.001) and phase× genotype (F (4,142) =4.07, p=0.004) interaction for freezing in the 24-h cue test (Fig. 3b) and a significant phase×sex interaction (F (2,142) =5.94, p=0.003) for freezing in the 7-day test (data not shown). Thus, data were analyzed separately for males and females according to each phase of testing. In the 24 h test, Tph2 −/− males tended to freeze more than Tph2 +/+ males during and after the tone presentation (both p<0.1). In the 7-day test, however, enhanced freezing in Tph2 −/− males was specifically observed in the tone phase (p<0.1) (data not shown). In females, stressed Tph2 −/− mice displayed increased freezing during the adaptation period (pre-tone) in the 24-h test (p<0.05 compared to CTRL Tph2 −/− as well as CMS Tph2 +/+ and CMS Tph2 +/− females). Moreover, Tph2 −/− females of both groups froze significantly more than Tph2 +/+ mice during tone presentation and the post-tone phase (both p<0.05). Similar results were observed for the 7-day test.

Retention of contextual fear memory
Retention of contextual fear memory was tested 48 h and 8 days after conditioning (Fig. 3c). Three-way ANOVA Activity in the open field (OF) measured by the total distance travelled (e), number of rears (f), center time (g), and center crossings (h) in male (upper panel) and in female (lower panels) Tph2 −/− mice. a-d Males: n=8 per group×genotype condition except for CTRL Tph2 −/− and CMS Tph2 +/+ (n=7); females: n=8 per group× genotype condition except for CMS Tph2 +/+ (n=7). e-h Males: n=8 per group×genotype condition except for CTRL Tph2 +/+ (n=7); females: n=8 per group×genotype condition. Data are shown as means±SEM.

5-HT and sex-dependent HPA response to CMS
Sex-specific functions of 5-HT have previously been reported (Bethea et al. 1998(Bethea et al. , 2009Rubinow et al. 1998;Goel and Bale 2010;Hall and Steiner 2013). Our results show that lifespanning inactivation of brain Tph2 synthesis differentially impacts behavior in both male and female Tph2 −/− mice and their response to CMS. Sex-specific stress responses due to CMS were described in both rodents and humans, suggesting an interaction between the HPA and hypothalamic-pituitarygonadal (HPG) systems (Goel and Bale 2010) and references therein. Previous reports of pCort levels following CMS in rodents yielded conflicting results with either reduced (Murison and Hansen 2001), unchanged (Azpiroz et al. 1999), or increased concentrations after 3 weeks (Bielajew et al. 2002) or after 1 and 2 weeks but with return to baseline after 3-6 weeks (Silberman et al. 2002). In patients with severe depressive disorders, sex-related dysregulation of the HPA system with higher circadian cortisol secretion and impaired negative feedback regulation was generally but not consistently reported (Posener et al. 2000;Gold and Chrousos 2002;Stewart et al. 2005;Lanzenberger et al. 2010;Parihar et al. 2011;Wardenaar et al. 2011); for review (Pariante and Lightman 2008). In this study, Tph2 −/− females showed a strong rise in Cm levels during CMS while Tph2 −/− males remained insensitive, suggesting increased stress vulnerability and activation of the HPA system in Tph2 −/− females. Furthermore, stressnaive Tph2 −/− females displayed low Cm levels, suggesting hypoactivity of the HPA system under basal conditions due to the lack of 5-HT. Thus, CMS produced an overactivation of the HPA axis, which is able to adapt after multiple CMS exposures. Interestingly, 5-HT transporter (5-Htt) deficient mice also showed stress-induced alterations in the adrenocortical stress response as measured by Cm excretion levels (Jansen et al. 2010). As 5-HT was shown to alter HPA system activity and brain corticosteroid receptor expression (Goel and Bale 2010;Heydendael and Jacobson 2010), an increased HPA c Fig. 4 Sex, stress, and Tph2 genotype differentially affect temperature control and depression-like responses. a, b Stress-induced hyperthermia of chronic mild stressed (CMS) Tph2 −/− mice compared toT ph2 +/+ and Tph2 +/− mice and their respective controls (CTRL) with basal body temperature for males (left panel) and females (right panel) (a) and the increase in body temperature (dT) (b). Latency to float (c) and total floating time (d) in the Porsolt swim test in male (left panel) and female Tph2 −/− (right panel) mice; n=8 per sex×group× genotype condition except for female CMS Tph2 +/+ and Tph2 −/− (n=7). Data are shown as means± SEM. ( * ) p<0.1, *p<0.05, and **p<0.01 compared to respective controls; (#) p<0.1 and ### p<0.001 Tph2 −/− compared to Tph2 +/+ and Tph2 +/− ; (+) p<0.1 CMS vs CTRL; xxx p<0.001 males vs females activation in Tph2 −/− females is likely due to a cumulative effect of 5-HT deficiency and gonadal steroids, supporting the notion of hormone-dependent adaptive mechanisms to stress in females vs males. In line with this, corticotropinreleasing factor (CRF) injections increased cFos expression in males but decreased cFos in females in the dorsal raphe (Howerton et al. 2014).
Depression-like behavior in Tph2 −/− females Discrepant results were reported for depression-like behavior in Tph2 −/− mice (Savelieva et al. 2008;Mosienko et al. 2012) as well as in Tph2 R439H mutant mice (Beaulieu et al. 2008). Here, we report a reduced latency to float and a trend to lower swimming time in Tph2 −/− mice in the PST, suggesting a tendency to behavioral despair. However, the total floating duration was not altered in Tph2 −/− mice. Although consistently supported in a review by Willner (Willner 2005), we did not detect an effect of CMS on floating time in the PST, which was conducted at the end of BT. However, the preference for sucrose decreased in CMS females after BT/CMS re-exposure, thus arguing for an anhedonia-like effect of CMS in females independent of genotype. In contrast to females, sucrose preference was increased in Tph2 −/− males, whereas CMS had no effect. Interestingly, Tph2 −/− mice of both sexes showed a general increase in total liquid and food intake but remained thinner compared to controls. This can be explained by an effect of lifelong Tph2 deficiency on metabolism, with Tph2 −/− mice displaying an increased energy metabolism (Yadav et al. 2009). Because a high metabolism and an increased level of activity in the open field were associated with a long latency to float in the PST representing active stress coping behavior (Campbell et al. 2003), Tph2 −/− mice might display a shorter latency to float due to their altered metabolism rather than a depressive-like phenotype (Mosienko et al. 2014), which manifests itself differentially in males and females.

5-HT deficiency prevents males from stress-induced hyperlocomotion
Previous studies provided inconsistent evidence for altered anxiety-like behavior in Tph2 −/− mice (Angoa-Perez et al. 2012;Mosienko et al. 2012). Our results demonstrate that anxiety-like behavior in Tph2 mice is moderated by sex, stress, and genotype. Interestingly, Tph2 −/− mice showed reduced anxiety-like behavior in the EPM indicated by reduced time spent in the closed arms. Thus, 5-HT seems to tonically reduce risk assessment behavior, defined as frequent approach and withdrawal movements providing information of potential dangerous situations leading to either further defensiveness or a return to non-defensive behaviors (Blanchard and Blanchard 1989). Based upon lifelong 5-HT depletion in the brain, Tph2 −/− males do not respond to CMS with increased locomotor activity like Tph2 +/+ and Tph2 +/− males. Thus, CMS may induce hyperlocomotion through 5-HT-dependent mechanisms (Strekalova et al. 2005;Hale et al. 2008;Couch et al. 2013), potentially involving different subgroups of 5-HT neurons targeting different brain regions involved in locomotor control (Hale et al. 2012). In line with present and other studies, deficiency of 5-HT in males may shift defensive avoidance behavior toward active avoidance (Angoa-Perez et al. 2012;Mosienko et al. 2012). While male Tph2 −/− showed reduced anxiety in the EPM and OF, Tph2 −/− females exhibited increased anxiety, as reflected by novelty-induced locomotor suppression and increased stress-related autonomic reactivity in the OF but reduced anxiety in the EPM without an increase in locomotion. Thus, 5-HT-dependent mechanisms seem to act in females via tonic suppression of anxiety in escapable contexts like the EPM, while in inescapable contexts, 5-HT serves to suppress an increased stress response and passive coping behaviors, which is supported by increased baseline immobility of Tph2 −/− females during the adaptation period in fear conditioning.
In contrast to unconditioned anxiety-like behavior in the EPM, fear conditioning provides an inescapable aversive physical stimulus evoking an adaptive freezing response to the novel context (Homberg 2012) similar to the OF. Furthermore Tph2 −/− mice showed increased shock reactivity independent of CMS, which indicates that 5-HT is implicated in the suppression of flight responses Spannuth et al. 2011;Gutknecht et al. 2012), as thermal and tactile pain sensitivity was not changed in Tph2 −/− mice.
Increased cue fear learning in male and female Tph2 −/− mice is in line with findings in other models of 5-HT deficiency (Dai et al. 2008;Kiyasova et al. 2011) and argue for specific effects of 5-HT neuron subgroups, which balance CMSevoked changes in limbic brain areas involved in cued conditioning (Mongeau et al. 1997;Sullivan et al. 2004;Pape and Pare 2010;Hale et al. 2012).
However, long-term contextual fear conditioning was influenced by the lack of 5-HT in stressed mice, arguing for 5-HT as important factor in the inhibition of contextual longterm memory formation after CMS. Furthermore, the correlation of Cm with an increase in MR expression in the frontal cortex of Tph2 −/− females after 1 week of CMS indicates a reciprocal modulation between the HPA axis and the 5-HT system mediated by differential expression of GR and MR receptors of males and females in the frontal cortex and the hippocampus. This, in turn, may modulate Tph2 and 5-HT receptor expression (Lanfumey et al. 2008;Nexon et al. 2011;Gutknecht et al. 2012), which are involved in contextual memory formation and points toward a similar effect of CMS and inescapable stress on the prefrontal cortex, which may inhibit 5-HT mediated stress-related adverse effects (Amat et al. 2005;Baratta et al. 2009).
Here, we used unpredictable CMS in combination with a behavioral testing battery. From a clinical perspective, it appears counterintuitive that 5-HT deficiency simultaneously results in anxiolytic effects and in depression-like behavior. Nevertheless, mirror models like the 5-HT 1A KO mouse also display increased anxiety-like and reduced depressive-like behavior (Heisler et al. 1998). In patients, depression is frequently accompanied by anxiety symptoms, also being responsive to drugs targeting 5-HT neurotransmission. Lemogne and associates recently suggested that cognitive appraisal modulates brain responses to emotional stimuli and can counteract genetic and environmental susceptibility factors in humans (Lemogne et al. 2011). Furthermore, CMS may induce diverse compensating effects based on the lifelong inactivation of brain 5-HT synthesis involving activity-induced adult neurogenesis (Parihar et al. 2011;Klempin et al. 2013). Thus, a simplified, more accessible and controllable model such as the Tph2 −/− mouse may help deciphering basic mechanisms and neuronal circuits involved in the different roles of 5-HT likely to be operant in humans as well. Taken together, our findings suggest that 5-HT depletion throughout life blocks the CMS effect on locomotion and anxiety in males and in females. Thus, 5-HT seems to mediate the locomotor stress effects and that it is of clinical relevance to understand the neural network and adaptive mechanisms between sexes, which allow anxiety, fear, and depressive symptoms to be ameliorated by mild stressors of an enriched environment like CMS.
Impulsive and hyperaggressive behavior in Tph2 −/− males In the RI paradigm, Tph2 −/− males exhibited highly increased aggressive behavior, particularly increased impulsivity reflected by decreased latency to the first attack and increased duration of fighting, which was further aggravated by CMS. Aggression-like behavior of a male resident toward male intruders is an ethologically determined response to territorial threats. Overwhelming evidence links 5-HT to impulsive and aggressive behavior as the primary determinant of aggression control (Lesch and Merschdorf 2000). Several regions of the frontal and cingulate cortices, amygdala, septum, hypothalamus, and periaqueductal gray matter are among the best documented to be involved in the neural circuitry of aggression. Serotonergic fibers extensively project to each of these regions, and it is well established that both aggressiveness and increased impulsivity are associated with brain 5-HT deficiency. The impulsive and hyperaggressive behavior of Tph2 −/− mice resembles the increased defensiveness reported for Pet1 KO (Hendricks et al. 2003) and Tph2 R439H mutants (Beaulieu et al. 2008). Acute treatment with 5-HT 1A and 5-HT 1B receptors agonists (or 5-HT 2A/2C antagonists) via their inhibitory action on neurotransmission (presynaptically or postsynaptically) was reported to reduce aggressive behavior, and it was suggested that low 5-HT levels in the brain are associated with maladaptive forms of excessive violence rather than with natural defensiveness (Miczek et al. 2007;de Boer et al. 2009). Furthermore, chronic reduction in serotonin neuron firing through overexpression of Htr1a in adulthood, but not during development, was sufficient to increase aggression (Audero et al. 2013). Since 5-HT deficiency is likely to result in impaired inhibition of engagement and sustainment of aggressive behavior, it may explain that Tph2 −/− mice display exaggerated aggressive behavior as a consequence of failing 5-HT-mediated inhibitory control, thus rendering these mice inert to acquire social abilities.
In contrast to Tph2 −/− mice, no effects of sex and CMS on behavior were detected in Tph2 +/− mice, which are characterized by up to 30 % reduction in 5-HT in the raphe but unaltered concentrations of 5-HT in target regions of 5-HT neurons (Gutknecht et al. 2012). This provides evidence for a specific cellular compensatory mechanism during development in Tph2 deficient mice, which affects GABAergic networks via upregulation of 5-HT1A and 5-HT2 receptors (Jorgensen et al. 2013;Waider et al. 2013).
In conclusion, converging lines of evidence suggest that the balance of the 5-HT system plays a critical role in synaptic plasticity of a multitude of neuronal networks, thus setting the stage for expression of complex traits and their associated behaviors throughout development and adult life depending on the individual 5-HT metabolism . Moreover, variation in genes moderating 5-HT system function, in conjunction with other rare and common variants of the genetic background and with inadequate adaptive responses to environmental stressors, is also likely to contribute to negative emotionality and aggression-related behavior emerging from compromised brain development and from highly efficient neuroadaptive processes across the life cycle.