The Role of Dorsal Raphe Nucleus Serotonergic Systems in Emotional Learning and Memory in Male BALB/c Mice

—Selective serotonin reuptake inhibitors (SSRIs) are the ﬁrst-line pharmacological treatment for a variety of anxiety, trauma-and stressor-related disorders. Although they are eﬃcacious, therapeutic improvements require several weeks of treatment and are often associated with an initial exacerbation of symptoms. The dorsal raphe nucleus (DR) has been proposed as an important target for the modulation of emotional responses and the therapeutic eﬀects of SSRIs. Using a fear-conditioning paradigm we aimed to understand how SSRIs aﬀect emotional learning and memory, and their eﬀects on serotonergic circuitry. Adult male BALB/c mice were treated with vehicle ( n = 16) or the SSRI ﬂuoxetine (18 mg/kg/d) acutely ( n = 16), or chronically (21d, n = 16), prior to fear conditioning. Treatment was stopped, and half of the mice ( n = 8/treatment group) were exposed to cued fear memory recall 72 h later. Activation of DR serotonergic neurons during fear conditioning ( Experiment 1 ) or fear memory recall ( Experiment 2 ), was measured using dual-label immunohistochemistry for Tph2 and c-Fos. Acute and chronic ﬂuoxetine treatment reduced associative fear learning without aﬀecting memory recall and had opposite eﬀects on anxiety-like behaviour. Acute ﬂuoxetine decreased serotonergic activity in the DR, while chronic treatment led to serotonergic activity that was indistinguishable from that of control levels in DRD and DRV sub-populations. Chronic ﬂuoxetine facilitated fear extinction, which was associated with rostral DRD inhibition. These ﬁndings provide further evidence that SSRIs can alter aspects of learning and memory processes and are consistent with a role for discrete populations of DR serotonergic neurons in regulating fear-and anxiety-related behaviours. (cid:1) 2023 The Author(s). Published by Elsevier Ltd on behalf of IBRO. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


INTRODUCTION
An appropriate behavioural response to threat is fundamental for survival (Fanselow, 1994) and can include active or passive strategies, depending on the proximity of the threat.These strategies can be innate or learned (i.e., conditioned) to form part of the animal's defence mechanisms (Bowen et al., 2013;Koutsikou et al., 2014), but can also become maladaptive, leading to increased risk for anxiety and trauma-and stressorrelated disorders (Hartley and Phelps, 2010;Homberg, 2012;Parsons and Ressler, 2013).Clinically, patients with generalised anxiety disorder, panic disorder and phobia often exhibit inappropriate processing of potentially threatening stimuli, which might include exaggerated fear responses (Lang et al., 2000;Lissek et al., 2008;Cryan and Sweeney, 2011;American Psychiatric Association, 2013).Similarly, patients with posttraumatic stress disorder (PTSD) often present with strong, persistent memories of traumatic experiences (Diamond et al., 2007) and exhibit heightened fear conditioning in laboratory settings (Glover et al., 2011;Roy et al., 2013).This suggests that cognitive deficits associated with altered conditioned fear learning, and in particular dysfunctional acquisition and extinction of conditioned fear memories, are core https://doi.org/10.1016/j.neuroscience.2023.10.0030306-4522/Ó 2023 The Author(s).Published by Elsevier Ltd on behalf of IBRO.This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).processes in the development and maintenance of anxiety and trauma-and stressor-related disorders.
Selective serotonin reuptake inhibitors (SSRIs) are the first-line pharmacological treatment for a variety of anxiety-, trauma-, and stressor-related disorders, such as PTSD (Ravindran and Stein, 2009;Koen and Stein, 2011).The primary pharmacological action of SSRIs is to block the serotonin transporter that is responsible for the reuptake of serotonin.As a result, the concentration of serotonin in the synaptic cleft is increased, at least temporarily (Bel and Artigas, 1992;Malagie´et al., 1995;Wong et al., 2005).However, the acute effects of SSRI administration are often an initial exacerbation of clinical symptoms (Teicher et al., 1990;Masand and Gupta, 1999;Fergusson et al., 2005), while the improvements in anxiety-related symptomology do not appear until after several weeks (Patel et al., 2018;Jakubovski et al., 2019;Lewis et al., 2019).In clinical laboratory settings using classical fear conditioning paradigms, a single dose of the SSRI citalopram enhances fear-potentiated startle responses when taken immediately prior to fear acquisition (Browning et al., 2007;Grillon et al., 2007).In contrast, 2 weeks of continuous treatment with escitalopram facilitates extinction of the conditioned fear response (Bui et al., 2013).A similar pattern of effects can be seen in preclinical studies using rats, whereby conditioned fear behaviour is enhanced by acute SSRI administration, but reduced with chronic SSRI administration (Burghardt et al., 2004;Burghardt et al., 2007;Burghardt and Bauer, 2013).Thus, with prolonged use, SSRIs seem to engage compensatory or adaptive changes that occur within anxiety and fear networks.Multiple hypotheses for the mechanisms underlying the acute effects of SSRIs on fear learning and memory have been proposed (for a review, see Ravinder et al., 2013).However, despite significant research, the mechanisms underlying the effects of chronic SSRI treatment on emotional learning and memory circuitry, and the reason why long-term treatment is necessary remain largely unknown.
The midbrain dorsal raphe nucleus (DR) is an important locus for serotonin neurotransmission as well as a major source of serotonin innervation to the forebrain (Vertes, 1991;Vertes et al., 1999;Ren et al., 2018), and has been proposed as an important target for the modulation of emotional responses and the therapeutic effects of SSRIs.Previous studies have shown that DR serotonergic neurons are active during emotionally salient aversive stimuli (Grahn et al., 1999;Takase et al., 2004;Schweimer and Ungless, 2010;Cohen et al., 2015;Baratta et al., 2016;Seo et al., 2019), including fear conditioning and fear-potentiated startle (Ishida et al., 2002;Liu et al., 2003;Spannuth et al., 2011;Shikanai et al., 2012), while lesion studies have implicated the DR in threat prediction (Berg et al., 2014).Importantly, the DR is not a homogeneous structure but contains subsets of serotonergic neurons that are anatomically and functionally distinct (Hale and Lowry, 2011;Hale et al., 2012).Distinct subgroups of serotonergic neurons originating from these DR subregions are selectively engaged by anxiety-and fear-related stimuli (Grahn et al., 1999;Abrams et al., 2005;Spannuth et al., 2011;Shikanai et al., 2012;Lawther et al., 2015).Moreover, differential regulation of serotonergic clusters in the DR has been implicated in the opposing effects of acute and chronic SSRI treatment on approachavoidance behaviour to social stimuli (Payet et al., 2018;Payet et al., 2021).However, while there is some overlap, social approach-avoidance paradigms elicit a set of defensive behaviours and engage brain systems that are distinct to those elicited by immediate threat-related paradigms, such as fear conditioning.In addition, fear conditioning paradigms involve cognitive processes associated with learning and memory that are not assessed in social approach-avoidance.
In the present study, we tested the hypothesis that serotonergic subsystems in the DR are involved in the effects of SSRIs on conditioned fear learning, memory and extinction processes.Adult male BALB/c mice, which are commonly used as a model of high anxiety, emotionality and stress-sensitivity and possess a polymorphism of the tryptophan hydroxylase 2 (Tph2) gene, were first exposed to acute and chronic systemic treatment with the SSRI fluoxetine, then to fear conditioning followed by cued, auditory fear memory recall and extinction.Dual-label immunohistochemistry for the protein product of the immediate-early gene c-fos (c-Fos; used as measure of cellular activation) and Tph2 (used to identify serotonergic neurons) in the DR was used to investigate the adaptive mechanisms through which chronic SSRI treatment influences fear learning and memory.

EXPERIMENTAL PROCEDURES Animals
Four-week-old male BALB/c mice (N = 48) were sourced from the Animal Resources Centre (Murdoch, WA, Australia) and given a 7-day acclimation period upon arrival.All mice were 5 weeks old at the start of treatment, and 8 weeks old when exposed to fear conditioning.Mice were housed on a 12:12 h reverse light:dark cycle (lights on at 7:00 pm) upon arrival.These mice were housed four per cage in individually ventilated cages (21 °C ± 2 °C, 45% humidity), and were provided with standard chow and water ad libitum.While it is important to consider sex as a biological variable in preclinical research on anxiety-, trauma-, and stressorrelated disorders, only male mice were used in the current study based on previous research demonstrating that female BALB/c mice fail to respond to chronic fluoxetine treatment in the drinking water (Payet et al., 2021).This experiment was approved by the La Trobe University Animal Ethics Committee and follows the Australian code for the care and use of animals for scientific purposes (8th edition; National Health and Medical Research Council, 2013) as well as the reporting requirements of the ARRIVE 2.0 guidelines (Percie du Sert et al., 2020).

Pharmacological treatment
All animals received an acute dose (18 mg/kg; i.p.) or a chronic dose (18 mg/kg/d; delivered via the drinking water) of fluoxetine (Merck,Lot No. LRAA9180) or vehicle, based on a random allocation of cages.These doses of fluoxetine were selected based on previous studies in this strain (Dulawa et al., 2004;Payet et al., 2018;Payet et al., 2021).For the chronic treatment group (n = 16), fluoxetine was dissolved at a concentration of 0.14 mg/mL and administered in the drinking bottles for 21 consecutive days.Following 21 days of treatment, these mice were administered 100 mL of 0.9% sterile saline via the intraperitoneal (i.p.) route, 30 min prior to fear conditioning.Drinking bottles containing fluoxetine were replaced by plain drinking water for the remainder of the experiment.For the acute fluoxetine treatment group (n = 16) mice were given untreated drinking water for 21 consecutive days, followed by 100 mL of fluoxetine (i.p.) 30 min prior to fear conditioning.Mice in the vehicle control group (n = 16) received untreated drinking water for 21 consecutive days, followed by 100 mL of 0.9 % sterile saline (i.p.), 30 min prior to fear conditioning.

Behavioural apparatus
Fear conditioning chambers (30 cm L Â 24 cm W Â 21 cm H; Med Associates, St Albans, VT, USA) with two different contexts, context A and context B, were used to measure fear learning acquisition and cued fear memory recall and extinction learning, respectively.Context A consisted of aluminium walls with a flat rectangular rear wall and overhead near infrared lighting providing low, red-light condition, and peppermint essence solution applied to the chamber prior to every new trial.Context B consisted of white acrylic walls with a curved rear wall, sawdust bedding, and white-light condition.Both contexts had a metal-grid floor with stainless steel rods connected to a scrambled electric shock generator.The fear conditioning chambers contained a wall speaker connected to an exterior tone generator and were placed in a sound-attenuating box.Mice were placed in the conditioning chambers one at a time, for a total duration of 6 min/mouse/day, during the dark phase of the 12:12 light:dark cycle.All chambers were thoroughly cleaned using F10 TM veterinary disinfectant between every trial to avoid animal scent carryover, and all testing took place over three days, between 10:00 am and 3:00 pm (3-8 h after lights off), with treatment groups counterbalanced across times.

Details of individual experiments
Experiment 1.To determine the effects of acute and chronic fluoxetine treatment prior to fear conditioning on fear learning and serotonergic activity in the DR, mice (n = 8 per treatment group) were treated with fluoxetine either acutely or chronically as described above.Thirty minutes after the administration of fluoxetine or vehicle via i.p. injection, mice were exposed to fear conditioning in context A. After a 90 s acclimatisation to the chamber, mice received three pairings of a 30 s tone (conditioned stimulus; CS) of 5 kHz and 75 dB sound pressure level (SPL), that co-terminated with a 0.7 mA electric stimulus (unconditioned stimulus; US) that lasted for the last 1 s of the tone.Each CS-US pairing was separated with a 60 s inter-pairing interval (IPI).Sixty seconds after the final CS-US pairing, mice were returned to their home cage.Ninety minutes later (120 min post-treatment with fluoxetine or vehicle), mice were deeply anesthetized using sodium pentobarbital and transcardially perfused with 0.05 M phosphatebuffered saline followed by 4% paraformaldehyde fixative solution in preparation for immunohistochemistry.
Experiment 2. To determine if the effects of acute and chronic fluoxetine treatment given prior to fear conditioning were associated with later changes to fear memory recall and extinction learning, as well as lasting or adaptive changes to the activation of serotonergic neurons in the DR, BALB/c mice (n = 8 per treatment group) were treated with fluoxetine either acutely or chronically and exposed to fear conditioning in context A, as described above.Treatment with fluoxetine or vehicle was stopped, and 72 h later mice were placed into a conditioning chamber containing context B. After a period of 3 min of free exploration, a 5 kHz, 75 dB SPL tone was delivered continuously for 3 min in the absence of electric stimulus.Mice were then returned to their home cage and, 90 min later, they were deeply anesthetized and brains were fixed by transcardial perfusion, as described above for Experiment 1, in preparation for immunohistochemistry.In this experiment, the c-Fos expression observed is in response to exposure to fear memory recall, drug-free, although differences between groups may indicate an effect of pre-treatment with an acute dose or a chronic dose of fluoxetine on later cell activation to memory recall processes.

General methods
Behaviour.The six-minute duration of each testing session was recorded by a near-infrared video camera (Video-Freeze, Med Associates) set in a fixed position in front of the chambers.A measure of the percent of time spent freezing in each component of the test was computed by the VideoFreeze software (Med Associates, USA), whereby the amount of time spent immobile in a component of the test is divided by the total duration of that component of the test.VideoFreeze-derived freezing scores have been previously validated as a measure of freezing behaviour in mice (Anagnostaras et al., 2010).Fear conditioning paradigms also allow for the measurement of anxietyrelated behaviour, such as exploratory-driven rearing behaviour.Rearing, which is defined as an instance of upright, bipedal standing, either free standing or against a wall, is indicative of exploratory behaviour and lower levels of anxiety (Rodgers and Johnson, 1995).The number of rearing bouts was manually counted using the event-logging software BORIS by an experimenter blind to treatment conditions.
Cell counting.Three rostro-caudal levels (-4.42 mm, -4.78 mm and -5.14 mm bregma) of the DR were selected for analysis of the cell count data in each animal.Illustrative photomicrographs of the anatomical subdivisions of the DR selected can be found in our previously published papers (Payet et al., 2018;Payet et al., 2021).In the current study, photomicrographs were taken using an Aperio AT2 whole slide scanner (Leica Biosystems, Mt Waverley, VIC, Australia) at 20x.ImageJ (version 1.52p) was used to define the DR subdivisions in each photo, and for manual cell counting using the cell counter function.

Statistical analysis
Behaviour.All data were analysed using IBM SPSS Statistics version 25 for Windows (SPSS Inc., Chicago, IL, USA).Freezing and rearing data during fear acquisition and cued fear memory recall were analysed for each experiment separately using mixed-design repeated-measures analysis of variance (ANOVA).For the fear acquisition data in Experiment 1 and Experiment 2, treatment was the between-subjects factor (three levels; vehicle, acute fluoxetine or chronic fluoxetine) and time was the within-subjects factor (made up of the different components of the fear acquisition procedure, seven levels; baseline, CS-US 1, IPI 1, CS-US 2, IPI 2, CS-US 3, End).For the fear memory recall data in Experiment 2, treatment was again the between-subjects factor (three levels; vehicle, acute fluoxetine or chronic fluoxetine) and time was the within-subjects factor (made up of the different components of the fear memory recall procedure, four levels; baseline, tone min 1, tone min 2, tone min 3).When the assumption of sphericity was not met, a more conservative Greenhouse-Geisser correction epsilon (e) was used.Significant ANOVA effects were followed by two-sided Dunnett's tests, with the vehicle group as control.For the fear acquisition data, the Grubbs' test (Grubbs, 1969) identified 29 outliers (5.8% of total data) in Experiment 1, and 22 outliers (4.4% of total data) in Experiment 2. For the cued fear memory recall data in Experiment 2, nine outliers (4.7% of total data) were identified.Outliers were Winsorized (Dixon, 1960) in order to perform mixed-design repeated-measures ANOVAs.No outlier or Winsorized value was included in the post hoc testing or graphical representations of the data.
Paired-samples t-tests were conducted on freezing behaviour between the first and last minute of the tone during the fear memory recall test in order to investigate the extent of fear memory extinction within the three treatment groups.Two-tailed significance was set a priori at p < 0.05.
Cell counting.Cell count data were analysed for Experiment 1 and Experiment 2 separately.A series of mixed-design repeated-measures ANOVAs were computed for; 1) c-Fos-ir/Tph2-ir neurons; 2) c-Fos-ir non-serotonergic cells; and 3) total number of Tph2-ir neurons.In all comparisons, the between-subject factor was treatment (three levels; vehicle control, acute fluoxetine or chronic fluoxetine) and the within-subjects factor was brain region (eight levels; DRD and DRV at -4.42 mm bregma; DRD, DRV, DRI, and DRVL at -4.78 mm bregma; DRI and DRC at -5.14 mm bregma).All significant interaction effects and main effects were followed by post hoc testing using two-sided Dunnett's tests or Dunnett's T3 test in case of heteroscedasticity.In cases where the assumption of sphericity was violated, a more conservative Greenhouse-Geisser correction epsilon (e) was used.In Experiment 1, 12 outliers (2.1% of dataset) were identified.In Experiment 2, five outliers (0.9% of dataset), plus 24 missing values due to one brain with poor tissue quality that could not be imaged (4.2% of dataset) were identified.Missing values were removed from the dataset, while outliers were Winsorized (Dixon, 1960) in order to perform mixed-design repeated-measures ANOVAs.As with the behaviour data, no outlier, missing value or Winsorized value was included in the post hoc calculations or graphical representations of the data.
Correlation analyses.A set of Pearson's correlations between fear-related freezing and rearing behaviour and the number of c-Fos-ir/Tph2-ir neurons in the DRD and DRV were conducted.These subregions of the DR were selected a priori for the correlation analyses based on previous research (Payet et al., 2018;Payet et al., 2021).The percent of time spent freezing and number of rearing bouts were collapsed into one 'Tone' and 'Post-Tone' value for the fear acquisition phase in Experiment 1, and one 'Pre-Tone' and 'Tone' value for the fear memory recall phase in Experiment 2. Two-tailed significance was set a priori at p < 0.05.

RESULTS
Experiment 1: Effects of acute and chronic fluoxetine treatment on fear learning acquisition and serotonergic activity in the DR Body weight and water intake.Weekly mouse weights were recorded over the 21 days of treatment with fluoxetine-treated or untreated water.There was a main effect of time on body weight (F(1.67,35.09) = 465.77,p < 0.001, partial g 2 = 0.96), but no main effect of treatment condition (F(2, 21) = 1.61, p = 0.224, partial g 2 = 0.13), or an interaction between treatment and time (F(3.34,35.09) = 0.95, p = 0.437, partial g 2 = 0.08; Supplementary Fig. 1(A")), which suggests that chronic fluoxetine treatment did not affect typical growth or health during the study.
The amount of water consumed by each mouse was estimated based on the daily weight of the bottles in each cage of four.There was a main effect of treatment condition (F(2, 21) = 55.58,p < 0.001, partial g 2 = 0.84), such that mice in the chronic fluoxetine group (mean ± SEM = 2.58 ± 0.01 mL) consumed less water than mice in the control group (3.47 ± 0.12 mL, p < 0.001) and in the acute fluoxetine group (3.47 ± 0.01 mL, p < 0.001; Supplementary Fig. 1(A')).
Correlation analyses.Pearson's correlations were conducted to determine the association between c-Fosir/Tph2-ir neurons in the rostral and mid-rostrocaudal DRD and DRV and behaviour during fear learning acquisition.In the vehicle condition, activation of serotonergic neurons in the rostral DRV negatively correlated with rearing behaviour during the tone (r = -0.75,p = 0.032), while activation in the midrostrocaudal DRD positively correlated with freezing behaviour post-tone (r = 0.81, p = 0.014), but negatively correlated with rearing behaviour post-tone (r = -0.81,p = 0.014).In the acute fluoxetine condition, serotonergic neuron activity in the midrostrocaudal DRV negatively correlated with freezing behaviour post-tone (r = -0.76,p = 0.029).In the chronic fluoxetine group, the number of c-Fos-ir/Tph2-ir neurons in the rostral DRD and DRV positively correlated with freezing behaviour post-tone (r = 0.92, p = 0.001; r = 0.89, p = 0.003; Supplementary Fig. 2).
Correlation analyses.Pearsons' correlations were conducted to determine the association between c-Fosexpressing serotonergic neurons in the DRD and DRV and behaviour during fear memory recall.In the vehicle control condition, the number of c-Fos-ir/Tph2-ir neurons in the mid-rostrocaudal DRD was positively correlated with rearing behaviour pre-tone (r = 0.74, p = 0.036).In the acute fluoxetine condition, serotonergic activity in the rostral DRD was positively correlated with freezing pre-tone (r = 0.81, p = 0.016), while in the midrostrocaudal DRV it was negatively correlated with freezing during the tone (r = -0.84,p = 0.017).In the chronic fluoxetine condition, freezing pre-tone was negatively correlated with serotonergic activity in the mid-rostrocaudal DRD (r = -0.85,p = 0.032), while rearing pre-tone was positively correlated with serotonergic activity in the mid-rostrocaudal DRD and DRV (r = 0.80, p = 0.031; r = 0.90, p = 0.015; Supplementary Fig. 3).

DISCUSSION
The SSRI fluoxetine had differential effects on emotional learning and extinction of fear memory, anxiety-driven exploratory behaviour, and serotonin activity in the DR when given acutely or chronically.Both acute and chronic doses of fluoxetine disrupted aspects of associative fear learning, and chronic fluoxetine facilitated the extinction of fear memory.The immediate response to fluoxetine administration resulted in reduced exploratory rearing behaviour during fear learning, suggesting an anxiogenic effect of the drug, and an inhibition of most serotonergic activity in the DR.The continuous exposure to fluoxetine over 3 weeks, in contrast, resulted in increased exploratory rearing behaviour during fear learning, suggestive of an anxiolytic effect, and levels of serotonergic activity in the rostral part of the DR that were comparable to controls.Serotonergic activity in these neuronal clusters was associated with freezing behaviour during fear acquisition.Together these findings provide further evidence that SSRI-driven changes to serotonin neurotransmission can alter learning and memory processes and are consistent with a role for a unique subpopulation of serotonergic neurons that originate in the DR in the regulation of fear and anxiety-related behaviours.
Three weeks of fluoxetine treatment prior to fear conditioning disrupted aspects of associative fear learning and was associated with anxiolytic effects on exploratory rearing behaviour.The disruption of emotional learning by chronic fluoxetine is consistent with previous research using the SSRI citalopram, whereby 22 consecutive days of treatment reduced the acquisition of cued fear conditioning in rats (Burghardt et al., 2004).Previous research using rats has further demonstrated that daily fluoxetine administration impaired the acquisition of a footshock escape response, which is another fear-related task (Nelson et al., 1997), and prevented the re-acquisition of cued fear conditioning (Deschaux et al., 2013).Our results are consistent with the effects reported in rats and support the hypothesis that chronic treatment with SSRIs has an inhibitory effect on associative fear learning.Further, chronic fluoxetine treatment increased exploratory rearing behaviour, which is indicative of an anxiolytic effect (Rodgers and Johnson, 1995).This effect is consistent with previous research in rats, where 4 weeks of fluoxetine treatment alleviated the effects of stress on exploratory activities (Rygula et al., 2006).The anxiolytic effects of long-term treatment with fluoxetine and other SSRIs have also been demonstrated in patients with anxiety disorder (Romano et al., 2004;Baldwin et al., 2006;Baldwin et al., 2016;Gomez et al., 2018).
Acute fluoxetine administration prior to fear conditioning disrupted fear acquisition and was associated with anxiogenic effects on exploratory rearing behaviour.The observation that a single dose of fluoxetine disrupted the acquisition of fear learning is inconsistent with previous research showing that an acute administration of citalopram 60 min before fear conditioning increases fear-related freezing (Burghardt et al., 2004).However, other studies have reported an overall lack of effect of acute SSRIs on freezing behaviour during fear conditioning (Ravinder et al., 2013), or a trend towards impaired acquisition of fear learning (Gravius et al., 2006).The discrepancy between these findings could be explained by differences in methodology, including the animal model and strain used.For example, the serotonin transporter knockout model (SERT -/-) shows deficits in fear acquisition processing, resulting in blunted cued fear-potentiated startle, but enhanced contextual fear (Bijlsma et al., 2015).This behavioural profile is indicative of fear generalisation, which is also seen in some patients with panic disorder who present with deficits in associative fear learning in the clinic (Jovanovic et al., 2012;Lopresto et al., 2016).The BALB/c strain possesses a polymorphism in the Tph2 gene, which is associated with altered serotonin availability and increased  anxiety and stress reactivity, which may underlie the differences in response to acute SSRI administration observed in our study.Furthermore, there are reports that different drugs of the same class, such as citalopram and fluoxetine, may not affect learning and memory in the same way despite similar effects on anxiety-and depression-like behaviours (Karpova et al., 2011;Burghardt and Bauer, 2013;Burghardt et al., 2013;Pedraza et al., 2019).A recent study looking at the effects of acute SSRIs in mice has found increased anxiety-like behaviour in the elevated zero-maze in the absence of fear-related freezing during fear learning (Marcinkiewcz et al., 2016).Consistent with this observation, acute fluoxetine inhibited exploratory rearing behaviour during fear conditioning in our study, which is indicative of an anxiogenic effect (Brookshire and Jones, 2009).This finding is also consistent with the effects of acute fluoxetine on tests of anxiety-like behaviour (Belzung et al., 2001;Birkett et al., 2011), and consistent with the the clinical effects of SSRI treatment (Bandelow et al., 2017).Thus, in the absence of fear-related freezing, acute fluoxetine might be affecting anxiety in BALB/c mice, which might have consequences for aspects of memory processing.
Together our results suggest that fluoxetine effects on anxiety-like behaviour may underlie, at least in part, the effects on fear learning acquisition.Acute treatment and exposure to fear conditioning was associated with a reduction in serotonergic activity across the DR, except for the mid-rostrocaudal DRD and the caudal DRC where activation of serotonergic neurons remained unchanged compared to vehicle controls.The inhibitory effect of a single dose of fluoxetine on serotonergic activity in the DR is consistent with previous research (Payet et al., 2018) and is consistent with the hypothesis that acute SSRI administration inhibits the firing of serotonergic neurons (Pin˜eyro and Blier, 1999).This effect is likely mediated by the inhibitory 5-HT 1A autoreceptor via a negative feedback loop.In line with this, pre-treatment with the 5-HT 1A receptor agonists F13714 or R(+)-8-OH-DPAT, which preferentially activate the somatodendritic autoreceptors, prior to fear acquisition training reduced fear potentiation (Zhao et al., 2019).Together these data suggest that the activation of somatodendritic 5-HT 1A autoreceptors by acute fluoxetine administration mediates cued fear learning disruption and anxiety-like behaviour.
Chronic fluoxetine and exposure to fear conditioning altered activation of serotonergic neurons in a manner that was dependent on the subregion of the DR.The mid-rostrocaudal levels of the DRVL and DRI showed a reduction in serotonergic neuron activity that was comparable to that observed following a single dose of fluoxetine.This finding is consistent with previous research (Payet et al., 2018;Payet et al., 2021), and suggests that 3 weeks of fluoxetine treatment produces a sustained decrease in serotonergic activity in these regions, or alternatively that these serotonergic subpopulations failed to adapt to the fluoxetine-induced decrease in serotonergic neurotransmission with chronic exposure.In contrast, subdivisions of the DR that include the DRD and DRV at both rostral and mid-rostrocaudal levels, and the caudal DRC and DRI, showed levels of serotonergic activity that were comparable to controls.Additionally, serotonergic activity in the rostral DRD and DRV positively correlated with freezing behaviour during fear conditioning.Previous research suggests that clusters of serotonergic neurons that originate in the DRD and DRV subdivisions and project to the amygdala nuclei are involved in the control of both conditioned fear and anxiety-like behaviours.Serotonergic neurons make up more than 90% of the neuronal input of the DR into the basolateral nucleus of the amygdala (BLA; Bernabe et al., 2020), and optogenetic activation of BLAprojecting DR neurons increases fear-related freezing and anxiety-like behaviour (Bernabe et al., 2020), an effect that was associated with increased serotonin release in the BLA (Zanoveli et al., 2009;Ravinder et al., 2013;Bernabe et al., 2020).In contrast, reducing serotonin concentration in the BLA reduced cue-induced conditioned fear response and anxiety-like behaviour (Johnson et al., 2015;Bernabe et al., 2020), an effect that was also associated with reduced number of serotonergic neurons in the DRD and DRV (Bernabe et al., 2020).Together these results provide support to the hypothesis that discrete populations of serotonergic neurons in the DRD and DRV that innervate the BLA are responsible for the control of anxiety-like behaviour and conditioned fear response, and that chronic fluoxetine treatment likely leads to a modulation of serotonin neurotransmission along this pathway.
Pre-treatment with the SSRI fluoxetine administered acutely or chronically prior to fear conditioning did not affect cued fear memory recall but disrupted extinction learning.While acute and chronic fluoxetine treatment both impaired associative fear learning in the first experiment, only acute fluoxetine disrupted fear-related freezing in experiment two, while chronic fluoxetinetreated mice showed a trend towards enhanced fear acquisition that is still suggestive of a disruption in associative learning processes.Nevertheless, the differences in their pattern of effects 72 h later suggest acute and chronic fluoxetine may alter learning and memory processes through different mechanisms.Upon re-exposure to the conditioned stimulus (i.e., the tone), mice in all treatment groups had an immediate increase in freezing behaviour to the same extent, suggesting intact memory of the paired CS-US association.In other words, neither acute nor chronic fluoxetine had any effect on initial fear memory recall.However, after three minutes of continuous presentation of the tone, chronic fluoxetine-treated mice showed a reduction in the conditioned freezing response, indicative of fear extinction (Chang et al., 2009;Morrissey et al., 2011).This effect is consistent with previous research reporting that fourteen days of chronic fluoxetine treatment facilitates the extinction of a conditioned fear response (Karpova et al., 2011).In contrast, mice that received an acute dose of fluoxetine prior to fear conditioning maintained an elevated freezing response throughout the 3min duration of the tone, which suggests interference with fear extinction learning.The finding that acute and chronic fluoxetine differentially affect the extinction of the fear memory without affecting the initial recall suggests an effect of SSRI treatment on fear learning through mechanisms involved in long-term memory consolidation processes.Consolidation is a process that takes place during the first few hours after fear learning acquisition, which allows the stabilisation of the temporary memory trace into a persistent long-term memory (Johansen et al., 2011), and serotonin neurotransmission has been implicated in the consolidation of memories (Meneses, 2007;Hart et al., 2011;Liu et al., 2011), including fear memories (Zhang et al., 2013).Evidence from previous research supports the hypothesis that long-term treatment with SSRIs alters aspects of fear memory by interfering with consolidation processes (Burghardt et al., 2004;Camp et al., 2012).Specifically, it was recently shown that chronic fluoxetine enhances fear memory extinction by slowing down systems consolidation (Pedraza et al., 2019).During consolidation, memories are re-organised into a more generalised state that is more resilient to change.Thus, slowing down consolidation processes prevents fear generalisation, which in turns facilitates fear memory extinction (Moscovitch et al., 2005;Winocur et al., 2007;Clem and Huganir, 2010;Gra¨ff et al., 2014).Together, these data suggest that chronic fluoxetine disrupts memory consolidation processes to facilitate extinction of the conditioned fear memory, while acute fluoxetine interferes with extinction learning.
Chronic pre-treatment with the SSRI fluoxetine and exposure to cued fear memory recall and extinction learning reduced activation of serotonergic neurons in two subregions of the DR, namely the rostral DRD and mid-rostrocaudal DRVL.In contrast, acute fluoxetine pre-treatment had no effect on DR serotonergic populations, which suggests that the fear extinction impairments in this group are likely to be independent of serotonergic neurotransmission in the DR.Previous research suggests that consolidation processes require serotonergic activity in the DR during fear learning and 5-HT 2C receptor signalling in the BLA immediately after the learning experience (Vertes, 1991;Kennett et al., 1997;Michelsen et al., 2007;Cools et al., 2008;Hale et al., 2012;Baratta et al., 2016;Sengupta and Holmes, 2019).There is a sustained increase in extracellular serotonin level in DR projection regions such as the BLA in the hour following conditioned fear learning (Yokoyama et al., 2005), which corresponds to the consolidation window, where activation of 5-HT 2C receptors promotes fear memory consolidation (Baratta et al., 2016).This suggests that 2 weeks of fluoxetine exposure may have disrupted this process, by altering expression or activation of 5-HT 2C membrane receptors in the BLA to slow down or inhibit memory consolidation, and this process was likely mediated by the reduction in serotonin neurotransmission in the DRD.
Alternatively, numerous studies have also implicated a role for serotonergic modulation of N-methyl-Daspartate (NMDA) receptors in the amygdala in extinction learning.Chronic SSRI treatment alters NMDA receptor structure and function (Paul et al., 1994;Skolnick et al., 1996;Boyer et al., 1998) and inhibition of NMDA receptors, and specifically of the GluN2b subunit of the NMDA receptor, in the amygdala is associated with impaired acquisition of fear extinction (Falls et al., 1992;Sotres-Bayon et al., 2007;Radulovic et al., 2019).Importantly, modulation of GluN2b in the amygdala has been proposed to mediate the effects of chronic SSRI treatment on extinction learning (Burghardt et al., 2013), and this effect is likely driven by the serotonergic 5-HT 1A and 5-HT 2A receptors.There is a high density of the inhibitory 5-HT 1A receptors in the amygdala (Pazos and Palacios, 1985) and activation of these receptors is associated with suppression of NMDA receptor function (Yuen et al., 2005).In contrast, the excitatory 5-HT 2A receptors are present on both excitatory, glutamatergic, pyramidal cells and inhibitory, GABAergic, non-pyramidal neurons in the amygdala (for a review, see Bombardi, 2014), and are involved in both the facilitation and inhibition of NMDA-induced responses (Arvanov et al., 1999;Clinard et al., 2015;Barre et al., 2016;Becamel et al., 2017).It has been proposed that the relative balance of 5-HT 1A and 5-HT 2A receptor actions on excitatory and inhibitory modulation of NMDA receptor function will determine the net effect on fear-related behaviour (see Bennett, 2010).Thus, chronic SSRI treatment may modify this balance by enhancing excitatory 5-HT 2A receptors over that of the inhibitory activity of 5-HT 1A receptors, leading to net increases in NMDA receptor activity to promote extinction learning.Future research looking at the activation of the DRD-BLA pathway and serotonin gene expression in the amygdala and the DR would help to elucidate the role of the 5-HT 1A , 5-HT 2A and 5-HT 2C receptors in the chronic fluoxetine effects on memory consolidation and extinction processes.
This study has some limitations.To minimise the effects of stress caused by repeated injections or isolation from conspecifics, given that stress exposure is known to impact responses to later fear conditioning (Baratta et al., 2016), mice were group-housed and chronic fluoxetine was administered via the cages' drinking bottles.As a result, the volume of fluoxetine that was consumed by each mouse could only be estimated based on the total volume consumed by the whole cage.While this method could lead to some variability in the group, the estimated dose received by each mouse was calculated based on the mouse weights, and all estimates fell close to the desired dose.Additionally, the mice in our study took perhaps longer than expected to learn the CS-US association as suggested by the lack of freezing to the second CS-US pairing during the conditioning phase.It is possible that including a training session prior to testing of the conditioned fear response would have improved the learned association and increased our capacity to detect an effect of fluoxetine treatment on fear-related freezing.Finally, as only male mice were used in our study, the generalisability of our findings to females remains unknown.Sex differences in the therapeutic effects of SSRIs have been proposed, although it remains controversial (Sramek et al., 2016).Future research using both male and female models may elucidate these effects.
In conclusion, the present study demonstrated that acute and chronic doses of the SSRI fluoxetine administered prior to fear conditioning disrupted processes involved in associative learning and consolidation of fear memory as well as having differential effects on anxiety-like behaviour in male BALB/c mice.The acute administration of fluoxetine primarily led to anxiogenic effects and had an inhibitory effect on serotonergic neurons across the DR, while serotonergic activity in DRD and DRV subpopulations following chronic fluoxetine exposure were associated with modulation of fear acquisition and anxiolytic effects.Further, reduced activation of subgroups of serotonergic neurons in the DRD was associated with extinction or consolidation processes 3 days later in the chronic fluoxetine-treated group, while impaired extinction seen in the acute fluoxetine-treated group did not seem to involve changes in serotonin neurotransmission at the level of the DR.These findings provide support for a distinct group of serotonergic neurons in the DR that have a unique role in the mechanisms of action of SSRIs.Further, while fluoxetine and other SSRIs are effective in reducing symptoms of anxiety with long-term treatment, their immediate and long-lasting effects on emotional learning and memory processes should be taken into consideration by clinicians before implementing learning-based therapies, such as cognitive behavioural therapy.

Fig. 1 .
Fig. 1.Acute and chronic fluoxetine administered prior to fear conditioning differentially affected fear acquisition behaviour and activation of serotonergic neurons in the dorsal raphe nucleus of adult male BALB/c mice.(A) Schematic illustration of the timeline of the fear conditioning acquisition session.(A', A") Line graphs illustrating the percent component of time spent freezing and the number of rearing bouts of mice in the three treatment conditions during fear learning acquisition.(B-B") Representative brain sections and (C-C") low-power photomicrographs of c-Fos immunoreactivity in serotonergic neurons and non-serotonergic cells in the mid-rostrocaudal DR (-4.78 mm bregma) in; B-C) a representative vehicle-treated mouse; B'-C') a representative mouse with acute fluoxetine treatment; and B"-C") a representative mouse with chronic fluoxetine treatment, respectively.Black boxes in C-C" indicate the subregion of the dorsal raphe nucleus, ventral part (DRV) shown at higher magnification in panels D-D".White boxes in panels D-D" indicate regions shown at higher magnification in insets at the lower right of each panel.c-Fosimmunoreactive (ir)/ Tph2-ir double immunostained serotonergic neurons are indicated by black arrowheads; c-Fos-immunonegative/Tph2-ir serotonergic neurons are indicated by white arrowheads; c-Fos-ir/Tph2-immunonegative cells are indicated by black arrows.Scale bar = 250 lm (C-C"), 50 lm (D-D"), 25 lm (for insets).(E) Box plots illustrating the effects of treatment with acute or chronic fluoxetine, or vehicle, and fear conditioning on c-Fos-ir/Tph2-ir neurons in subregions of the dorsal raphe nucleus.Vehicle control, n = 8; acute fluoxetine, n = 8; chronic fluoxetine, n = 8. *p < 0.05, ***p < 0.001, chronic fluoxetine versus vehicle control; + p < 0.05, +++ p < 0.001, acute fluoxetine versus vehicle control.Abbreviations: CS, conditioned stimulus; DRC, dorsal raphe nucleus, caudal part; DRD, dorsal raphe nucleus, dorsal part; DRI, dorsal raphe nucleus, interfascicular part; DRV, dorsal raphe nucleus, ventral part; DRVL, dorsal raphe nucleus, ventrolateral part; FLX, fluoxetine; IPI, interpairing interval; ir, immunoreactive; Tph2, tryptophan hydroxylase 2; US, unconditioned stimulus; VEH, vehicle.

Fig. 2 .
Fig. 2. Acute and chronic fluoxetine administered prior to fear conditioning differentially affected extinction of the fear memory and activation of serotonergic neurons in the dorsal raphe nucleus of adult male BALB/c mice.(A, A') Line graphs illustrating the percent of time spent freezing during the fear acquisition phase (A) and the fear memory recall phase 72 h later (A'), in Experiment 2. (A") Box plot illustrating the difference in freezing behaviour between the first and the last minute of the tone in each of the vehicle control, acute fluoxetine and chronic fluoxetine group.(B, B') Line graphs illustrating the number of rearing bouts during fear acquisition (B), and during fear memory recall (B').Representative brain sections (C-C") and low-power photomicrographs (D-F) of c-Fos immunoreactivity in serotonergic neurons and non-serotonergic cells in the mid-rostrocaudal DR (-4.78 mm bregma) in; D-D' a representative vehicle-treated mouse; E-E' a representative mouse with acute fluoxetine treatment; and F-F' a representative mouse with chronic fluoxetine treatment, respectively.Black boxes in D-F indicate a region of the dorsal raphe nucleus, ventrolateral part (DRVL) shown at higher magnification in the panels D'-F'.White boxes in panels indicate regions shown at higher magnification in insets at the lower right of each panel.The c-Fos-immunoreactive (ir)/ Tph2-ir double immunostained serotonergic neurons are indicated by black arrowheads; c-Fos-immunonegative/Tph2-ir serotonergic neurons are indicated by white arrowheads; c-Fos-ir/Tph2-immunonegative cells are indicated by black arrows.Scale bar = 250 lm (D-F), 50 lm (D'-F'), 25 lm (for insets).(G) Box plots illustrating the effects of treatment with acute or chronic fluoxetine, or vehicle, and cued fear memory recall on c-Fos-ir/Tph2-ir neurons in subregions of the dorsal raphe nucleus.Vehicle control, n = 8; acute fluoxetine, n = 8; chronic fluoxetine, n = 7. *p < 0.05, chronic fluoxetine versus vehicle control; + p < 0.05, acute fluoxetine versus vehicle control.Abbreviations: CS, conditioned stimulus; DRC, dorsal raphe nucleus, caudal part; DRD, dorsal raphe nucleus, dorsal part; DRI, dorsal raphe nucleus, interfascicular part; DRV, dorsal raphe nucleus, ventral part; DRVL, dorsal raphe nucleus, ventrolateral part; FLX, fluoxetine; IPI, interpairing interval; ir, immunoreactive; Tph2, tryptophan hydroxylase 2; US, unconditioned stimulus; VEH, vehicle.

Table 1 .
Effects of acute or chronic fluoxetine treatment on the number of c-Fos-ir/Tph2-immunonegative cells and total number of Tph2-ir neurons in subregions of the dorsal raphe nucleus following fear acquisition DR, dorsal raphe nucleus; DRC, dorsal raphe nucleus, caudal part; DRD, dorsal raphe nucleus, dorsal part; DRI, dorsal raphe nucleus, interfascicular part; DRV, dorsal raphe nucleus, ventral part; DRVL, dorsal raphe nucleus, ventrolateral part.n = 7-8 mice in each group.Values are presented as mean (SEM). Abbreviations:

Table 2 .
Effects of acute or chronic fluoxetine treatment on the number of c-Fos-ir/Tph2-immunonegative cells and total number of Tph2-ir neurons in subregions of the dorsal raphe nucleus following fear memory recall DR, dorsal raphe nucleus; DRC, dorsal raphe nucleus, caudal part; DRD, dorsal raphe nucleus, dorsal part; DRI, dorsal raphe nucleus, interfascicular part; DRV, dorsal raphe nucleus, ventral part; DRVL, dorsal raphe nucleus, ventrolateral part.n = 7-8 mice in each group.Values are presented as mean (SEM). Abbreviations: