Abstract
Serotonin critically modulates the activity of many brain networks, including circuits that control motivation and responses to rewarding and aversive stimuli. Furthermore, the serotonin system is targeted by first-line pharmacological treatments for several psychiatric disorders, including obsessive–compulsive disorder. However, understanding the behavioral function of serotonin is hampered by methodological limitations: the (brainstem) location of serotonergic neuron cell-bodies is difficult to access, their innervation of the brain is diffuse, and they release serotonin in relatively low concentrations. Here, we advance this effort by developing novel Tph2-Cre rats, which we utilized to study serotonin in the context of motor, compulsive, and reinforced behaviors using optogenetics in both male and female rats. Specificity and sensitivity of Cre recombinase expression and Cre-dependent processes were validated immunohistochemically, and optogenetic induction of in vivo serotonin release was validated with fast-scan cyclic voltammetry. Optogenetic stimulation of serotonin neurons in the dorsal raphe nucleus did not initiate locomotion or alter aversion-induced locomotion, nor did it elicit (real-time) place preference, and it had no measurable effect on compulsive behavior in the schedule-induced polydipsia task. In contrast, this optogenetic stimulation moderately sustained ongoing spontaneous locomotion and robustly reinforced operant lever pressing for self-stimulation of serotonin neurons, which was exacerbated by food restriction. Together, this work both introduces a novel rat Cre line to study serotonin and advances our understanding of serotonin's behavioral functions. Complementing previous findings, we find that brainwide serotonin release has an overall relatively mild effect on behavior, which manifested only in the absence of natural reinforcers and was modulated by physiological state.
- compulsive behavior
- dorsal raphe nucleus; intracranial self-stimulation
- optogenetics
- serotonin
- transgenic rats
Significance Statement
Although serotonin is produced by only a very small number of neurons, it modulates the activity of almost all brain networks and is implicated in numerous behavioral functions and many psychiatric disorders. However, our comprehension of serotonin function and dysfunction is hampered by methodological limitations, which can be improved by our novel Tph2-Cre rat line. We investigated general behavioral processes to principally understand serotonergic involvement in the many functions it has been implicated in. Complementing and furthering previous findings and consistent with its low concentrations found ubiquitously throughout the brain, we find that brainwide serotonin release has mild effects on behavior, which are modulated by hunger and manifest only in the absence of food reward in the experimental environment.
Introduction
Serotonin has been associated in the neurochemical coding of processes ranging from movement, arousal, sleep, perception, learning, decision-making, aggression, anxiety, mood, appetite, food intake, impulsivity, compulsivity, social behavior, and stress (Berger et al., 2009; Luo et al., 2016). Recent experimental work has focused mainly on the role of serotonin in neural processing and responding to rewarding and aversive stimuli (Maswood et al., 1998; Daw et al., 2002; Nakamura et al., 2008; Ranade and Mainen, 2009; Amo et al., 2014; Liu et al., 2014; McDevitt et al., 2014; Cohen et al., 2015; Fonseca et al., 2015; Hayashi et al., 2015; Li et al., 2016; Zhong et al., 2017; Iigaya et al., 2018; Ren et al., 2018; Seo et al., 2019; Nagai et al., 2020; Paquelet et al., 2022; Feng et al., 2023). In this context, several conceptualizations have emerged, attributing serotonergic control with wide-ranging functions, most prevalently in behavioral inhibition involving the control of impulsivity and patience (Miyazaki et al., 2011a,b; Fonseca et al., 2015), punishment coding and responding in opponency to dopamine function (Soubrié, 1986; Deakin and Graeff, 1991; Daw et al., 2002; Dayan and Huys, 2009), mood as influenced by an average reward rate in a given environment (Daw et al., 2002; Savitz et al., 2009), and the somewhat overlapping, broad idea of conveying overall beneficialness (Luo et al., 2016; Liu et al., 2020).
Serotonin neurons that reside in the raphe nucleus of the brainstem are the origin of widespread, long-range axons that innervate the entire mammalian brain and thereby crucially modulate the activity of numerous neuronal networks. This ubiquity is one of the reasons why it has been difficult to pinpoint the behavioral function of the serotonin system. Other obstacles in the study of serotonin function are of methodological nature: the location of serotonin–neuron cell bodies deep in the brainstem is difficult to reach, complicating selective targeting of serotonin neurons, and relatively low release concentrations (<70 nM) hinder serotonin detection and quantification (Crespi et al., 1988; Shen et al., 2004; Hashemi et al., 2009; Yang et al., 2013; Abdalla et al., 2017).
To further study the brain serotonin system, we introduce a novel transgenic rat line that targets Cre recombinase (Cre) at neurons expressing tryptophan hydroxylase 2 (TPH2), the rate-limiting enzyme for the synthesis of serotonin in the raphe nucleus. We chose to use rats because of the large range of available behavioral assays for this species [including rat-specific paradigms such as schedule-induced polydipsia (SIP)] and the larger brain size compared with mice, enabling more extensive neural implants and better targeting of the small dorsal raphe nucleus (DRN).
In this work, as a systematic, basic approach, we limited our investigation to general processes such as locomotion and reinforcement, through which we may learn to principally understand serotonergic involvement in the abovementioned diverse physiological functions. In addition, besides its physiological function, serotonin has been implicated in psychiatric dysfunction, with selective-serotonin reuptake inhibitors (SSRIs) as the first-line treatment for many disorders including depression and obsessive–compulsive disorder (OCD; Bloch et al., 2010; Jakubovski et al., 2016). Thus, we elaborated our investigation by characterizing the influence of serotonin on a well-described compulsive behavior (SIP). After verifying DRN5-HT Cre expression and functionality, we demonstrate that selective DRN5-HT activation acts robustly reinforcing for discrete operant behavior and mildly sustains ongoing spontaneous locomotion but has no detectable effect on locomotion initiation, aversion-induced locomotion, or compulsive behavior.
Materials and Methods
Generation of Tph2-Cre transgenic rats
Producing transgenic rat lines has proven difficult in the past (Hainer et al., 2015). In this study, we opted to generate a Cre rat line utilizing the serotonin-specific TPH2 regulatory sequence. This decision was based on TPH2’s involvement in catalyzing the initial, rate-limiting step of central nervous system serotonin synthesis and its exclusive expression in serotonergic neurons located in the brain and gut (Walther et al., 2003; Sakowski et al., 2006). Other commonly used regulatory gene sequences, such as SERT or Pet-1, are known to express in various peripheral cells or neurons with low transcript levels for serotonin genes (Pelosi et al., 2014; Hainer et al., 2015; Okaty et al., 2020). Therefore, our novel transgenic rat line holds promise as a more specific tool for genetics-based research on serotonin.
A P1-derived artificial chromosome (PAC; L065) which contains the full-length mouse Tph2 gene (107 kb) with 51 kb upstream and 19 kb downstream DNA sequences was modified as previously described (Weber et al., 2011) but using an Cre recombinase-encoding DNA fragment instead of a tamoxifen-inducible creERT2 recombinase. In brief, the Cre coding sequence was integrated together with a polyadenylation signal into the translation start of Exon 1 of the PAC-based Tph2 gene by recombination in E. coli. Homologous recombination resulted in the intentional deletion of 23 bp coding sequence of exon 1 containing potential alternative ATG-start sites.
The purified, linearized Tph2-Cre DNA was microinjected into the pronucleus of oocytes of Sprague Dawley rats (Charles River Laboratories). The resulting Tph2-Cre transgenic rats from founder Line 5 were bred with the Cre reporter line CAG-loxP.EGFP (Weber et al., 2011; Schönig et al., 2012) to generate double-transgenic Tph2-Cre/CAG-loxP.EGFP rats. CAG-loxP.EGFP reporter line rats harbor a loxP-flanked lacZ reporter gene, controlled by the ubiquitously active CAG promoter. The lacZ DNA fragment precludes the transcription of a second reporter gene EGFP, in which Cre-mediated recombination can be monitored by EGFP expression within the double-transgenic Tph2-Cre/CAG-loxP.EGFP rat.
An inducible version of the Tph2-Cre rat line (Tph2-CreERT2) has been introduced previously (Weber et al., 2011), as well as a comparable mouse line (Weber et al., 2009). Here, we modified the Tph2-CreERT2 construct (used for generating the tamoxifen-inducible transgenic rat line) by inserting another version of Cre recombinase in place of the CreERT2 gene. This Cre variant was designed with optimized codon usage to minimize the likelihood of epigenetic silencing (Shimshek et al., 2002). The primary motivation for developing this new line was that the CreERT2 (conditional) transgenic line requires tamoxifen to activate Cre recombinase. While tamoxifen inducibility has distinct advantages in certain experimental contexts, it is unnecessary, and often impractical, in others. The tamoxifen induction protocol is labor-intensive, and tamoxifen injections cause discomfort, preventing behavioral experiments surrounding the injection period. To our knowledge, only one other comparable mouse line with Cre expression driven by the Tph2 promoter sequence exists (Zhang et al., 2024).
It is essential to develop genetically manipulated rats due to their ability to perform in complex behavioral paradigms. For example, in the Morris water maze, a paradigm that tests cognitive/spatial learning and memory, mice have consistently performed worse than rats (Whishaw and Tomie, 1996; Frick et al., 2000; D’Hooge and De Deyn, 2001; Jonasson, 2005; Ghafarimoghadam et al., 2022). Furthermore, it has been reported that mice require more habituation and training sessions compared with rats in other cognitive behavioral tasks (Prusky et al., 2000; Colacicco et al., 2002; Bevins and Besheer, 2006; Cressant et al., 2007; Busse et al., 2011; Carandini and Churchland, 2013; Mar et al., 2013; Jaramillo and Zador, 2014). Partially, this may be due to mice experiencing more stress during handling procedures (Tabata et al., 1998; Colacicco et al., 2002; Meijer et al., 2007), which may contribute to them behaving more erratically and easily distracted compared with rats (Colacicco et al., 2002; Ellenbroek and Youn, 2016; Hok et al., 2016; Jones et al., 2017). However, whether due to handling stress or due to less developed cognitive faculties, the end result is that rats seem to be able to perform more complex behavior than mice. Together, this has led to the general conception that rats learn quicker and are better suited for complex task designs requiring large trial loads.
Immunohistochemistry
Transgenic Tph2-Cre rats were characterized by immunohistochemistry using DAB staining (VECTASTAIN Elite ABC Kit) with an α-Cre primary antibody (Millipore Sigma MAB3120, 1:1,000). We further characterized Cre expression and recombination with dual-label fluorescent immunohistochemistry in Tph2-Cre and Tph2-Cre/CAG-loxP.EGFP rats. Dissected brains from perfused animals were postfixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) at 4°C for 24–48 h and brain sections (50 μm) yielded by using a vibratome (Leica Biosystems).
The following primary antibodies were used: mouse α-Cre (Millipore Sigma 1:1,000), rabbit α-GFP (Invitrogen, 1:1,000), rabbit α-TPH2 (Dianova Laboratories, 1:5,000), and mouse α-tryptophan hydroxylase 1 (TPH1; Sigma-Aldrich, 1:2,500). TPH2 is the rate-limiting enzyme of 5-HT synthesis in the brain and specific to serotonergic neurons. The anti-TPH1 antibody cross-reacts with TPH2 and detects both isoenzymes.
Secondary antibodies were AF488 donkey α-rabbit (Invitrogen, 1:1,000 for TPH2 and 1:5,000 for GFP) and Cy3 donkey α-mouse (Jackson ImmunoResearch Laboratories, 1:200 for TPH1 and Cre). Sections were examined using a Leica SP5 confocal laser-scanning microscope. Confocal image stacks for both channels were acquired sequentially and projected on average using the Leica software.
Animals
Adult male and female Tph2-Cre rats (200–600 g; bred in-house) were housed individually and kept on a reversed light/dark cycle (light on from 21:00 to 9:00) with controlled temperature and humidity. All animal procedures were in accordance with the Dutch and European laws and approved by the Animal Experimentation Committee of the Royal Netherlands Academy of Arts and Sciences. Depending on the behavioral experiment, rats were either food-restricted to 85% of their free-feeding bodyweight or kept on an ad libitum feeding schedule, and water was always provided ad libitum.
Stereotaxic surgery: virus injection and optic fiber implantation
Rats were induced under isoflurane anesthesia (4% induction and 1–2% for maintenance) and placed into the stereotaxic frame (David Kopf Instruments) on an isothermal pad to maintain body temperature. The analgesic Metacam (0.2 mg meloxicam/100 g) was injected subcutaneously, and the shaved scalp was disinfected using 70% ethanol. Upon incision of the scalp, it was treated with lidocaine (100 mg/ml). Holes were drilled in the cranium, and the dura mater was cleared for targeting the DRN (AP, −7.6 to −7.8; ML, ±3.7). A 5 µL Hamilton syringe (Hamilton) was filled with a viral solution and lowered into the right craniotomy to the DRN (DV, −6.9 at a 30° angle). We employed an adeno-associated virus (AAV) construct to express Channelrhodopsin-2 (ChR2) in serotonergic neurons, by injecting AAVDJ-EF1a-hChR2(H134R)-eYFP (titer, 3.3 × 1012 GC/mL; Viral Vector Facility, University of Zurich; v214-DJ). For control animals, we injected AAVDJ-EF1a-eYFP (titer, 3.6 × 1012 GC/mL; Viral Vector Facility, University of Zurich; v343-DJ). The viral solution (600 nL) was injected using a pump (World Precision Instruments) at a rate of 0.1 µl/min. For fast-scan cyclic voltammetry (FSCV) experiments, optical fibers (200 µm core diameter; 8 mm) were unilaterally implanted in the DRN after a 6 min wait period. For behavioral experiments, after the 6 min wait period, two optical fibers (200 µm core diameter; 8 mm) were positioned bilaterally in the DRN, first starting with the left craniotomy (contralateral to virus injection) and then the right. The optic fibers were secured to screws in the skull using cranioplastic cement. Following surgery, rats received subcutaneous injections of 2 mL saline and were placed in a temperature-controlled cabinet to be monitored for an hour. Rats recovered for 4 weeks prior to all experiments. All animals were included that had sufficient virus expression and at least one optical fiber on target. For virus transduction quantification, we employed similar AAV constructs driven by the EF1a promoter, using both AAV-DJ and AAV-5 serotypes.
FSCV measurements and analysis
Four weeks after virus injection and unilateral optic fiber implantation, adult male and female Tph2-Cre rats (n = 9) were positioned in a stereotaxic frame under urethane anesthesia. A carbon-fiber microelectrode (glass-encased, T-650 carbon fiber, 6 µM in diameter with 150 µM exposed carbon fiber; Goodfellow) was advanced into the brain through a craniotomy and used for FSCV recordings to detect subsecond changes in extracellular concentration of serotonin, as described previously (Hashemi et al., 2009). Every 100 ms, voltammetric scans were repeated to achieve a sampling rate of 10 Hz. An N-shaped waveform from +0.2 V to +1.0 V to −0.1 V back to +0.2 V at 1,000 V/s (vs the implanted Ag/AgCl reference electrode) was applied to the working electrode to measure subsecond serotonin release before, during, and after photostimulation of the DRN. All serotonin signals were verified and amplified by the administration of escitalopram oxalate (Sigma-Aldrich), a SSRI.
The coordinates for all electrode implantations were determined relative to the bregma reference point. The amygdala recording electrode was positioned at coordinates AP, −3.5; ML,+5.2; and DV, −8.0 to −9.0, and the DRN photostimulation electrode was placed at coordinates AP, −7.6; ML, ±3.7; and DV, −6.9 using a 30° angle. An Ag/AgCl reference electrode was implanted into the contralateral hemisphere.
Recording and stimulation coordinates were carefully considered to minimize dopamine interference. We chose to target the amygdala because it is strongly innervated by the DRN (Ma et al., 1991; Asan et al., 2013). Specifically, the lateral and basolateral regions were targeted due to their relatively dense network of serotonergic fibers and high expression of serotonin transporters (Smith and Porrino, 2008). Animals were excluded from analysis due to mistargeting of either the amygdala or DRN, which led to five animals being removed.
Photostimulation
For FSCV experiments, light from a 473 nm laser was controlled by a DS8000 World Precision Instruments digital stimulator. For behavioral experiments, the blue laser was controlled directly with an Arduino (made in-house) and triggered by the Bonsai or Med Associates software. Depending on the behavioral paradigm, we chose to employ two different stimulation parameters to mimic the natural, physiological firing rates of serotonin neurons to increase serotonin: phasic, time-locked burst firing (15 Hz, 15–20 mW) and chronic, tonic-like firing (5 Hz, 1–5 mW) (Jacobs and Azmitia, 1992; Zhao et al., 2011; Liu et al., 2014). We chose to use a lower intensity (mW) for our chronic, tonic-like firing to prevent tissue damage from sustained laser output. Laser power was calibrated from the cable tip using a power meter (PM130D, Thorlabs) before every experiment.
Behavioral procedures
Behavioral training was conducted daily, with breaks (1–2 d) only on some of the weekends or for habituation to a new experimental setup. When adjusting access to food (either restriction or ad libitum feeding), we waited 1–2 weeks before training animals in the next behavioral task to allow animals to reach a stable weight (85–90% or 100% of free-feeding weight, respectively). To maintain stable behavior and minimize animal discomfort, switching between hunger states was kept to a maximum of two transitions.
To minimize sequence effects, we varied the sequence of behavioral tasks across cohorts. However, due to the training requirements for lever pressing, intracranial self-stimulation (ICSS) was always conducted in the ad libitum state first, followed by the food-restricted state. All other tasks, which did not require additional training, were randomly assigned within the sequence. Each cohort of rats completed at least three behavioral tasks. Overall, the total incorrect targeting rate across all cohorts was 14% [9/66 animals; Cohort 1, 3/23 animals (13%); Cohort 2, 2/15 (13%); Cohort 3, 4/28 (14%)].
We believe that the order of behavioral testing did not influence our results significantly, as the experimental environments were distinct [i.e., real-time place preference (RT-PP) was conducted in the open field (OF) and ICSS in the operant chamber (OC)] and important experimental cues, such as levers and water bottles, were not accessible across tasks. By utilizing the same animals for multiple experiments, we reduced the number of animals needed and strengthened the scientific rigor of our findings through within-animal comparisons.
OF: Apparatus
Rats were placed in a light-shielded, square, Perspex OF (60 × 60 × 60 cm), made in-house (Netherlands Institute for Neuroscience mechanical workshop). A camera mounted in the center above the OF recorded the position of the rat, which was tracked in real time by the open-source software Bonsai.
OF: RT-PP
Rats could freely explore the OF for 10 min while a region of interest (ROI; quadrant) was paired with blue photostimulation (5 or 15 Hz; 5 or 15–20 mW; 10 ms pulse width); no cues were provided to distinguish the photostimulation-paired quadrant. Stimulation was automatically switched on once >50% of the rat’s body was present in the chosen ROI. After habituation to the OF, rats underwent three experiments with the same ROI: 2 stimulation days, counterbalanced by order of stimulation intensity (5 or 15 Hz), followed by no stimulation. This procedure was then repeated in a second, randomly chosen quadrant, resulting in a total of two sessions per condition. Animals performed this task in both a satiated (fed ad libitum; n = 31) and hungry (food restricted; n = 22) state. We quantified the following parameters for each animal: ROI occupancy (percentage of time stimulated) and the average speed within the photostimulation-paired quadrant.
OF: locomotion
Rats (n = 28) fed ad libitum were placed into the OF and could freely explore while a trained observer manually stimulated the rats based on their activity level (in motion or at rest). Sessions lasted a total of 30 min, and no stimulation was delivered for the first and last 5 min of every session. Stimulations were separated by at least 10 s and each rat underwent two locomotion sessions. We quantified the speed of the animal before, during, and after photostimulation. We used a 0.2 cm/frame threshold, 0.5 s prior to stimulation onset to separate “trials” into “motion” and “rest” trials.
OF: approach–avoidance task
A subset of food-restricted rats (n = 11) from the white noise (WN) aversive-conditioning task (see below) performed an OF foraging task in which a moveable grid floor covered a quadrant and 10 food pellets were hidden, but accessible, underneath. The animals were placed in the middle of the OF, and the session lasted 10 min. After 1 d of habituation to the setup, rats underwent a series of sessions where they were exposed to 0, 70, or 90 dB of WN upon entering the grid-floor quadrant to forage for pellets. Rats were exposed to each WN intensity twice during foraging (in two rounds of sessions), where rats underwent exposure to all three WN intensities before exposure to each in a second round. The grid-floor position, where pellets were hidden within the OF, was changed between rounds. We quantified the percentage of time spent in the WN-paired quadrant for every animal.
OC: Apparatus
Rats were placed in a modified operant box (32 × 30 × 29 cm, Med Associates) equipped with two retractable levers, a house light, a water-spout port, a flickering blue LED light, a WN speaker, and metal grid floors (Med Associates). Each operant box was surveilled by a video camera, and the box was illuminated by a houselight for every session.
OC: ICSS
Rats (n = 38) were trained to lever press for the delivery of DRN photostimulation before undergoing five consecutive sessions in both ad libitum and food-restricted states. Two levers were extended into the box at all times: an active lever paired with 3 s DRN intracranial photostimulation (15 Hz, 15–20 mW, 45 pulses, 10 ms pulse width) and an inactive lever. Animals were trained to lever press by placing crushed sucrose pellets on both levers for the first 3 d, then only if animals performed <20 active-lever presses. The number of presses on each lever was registered via an automated procedure. The animals received no time out after a lever press and could extend their stimulation period by an additional 3 s via consecutive active-lever presses within the stimulation period. We quantified the following parameters for each animal: average and maximum number of stimulation bouts, number of active and inactive lever presses, percentage of time spent receiving stimulation, and distribution of active-lever presses within a session.
OC: WN aversive conditioning
Food-restricted rats (n = 15) were placed in an OC outfitted with a WN speaker for 30 min. WN alone (90 dB) or WN + photostimulation (15 Hz, 15–20 mW, 10 ms pulse width) was presented to the animal in randomized order for 6 s, with an average variable intertrial interval of 60 s (range, 30–90 s). Each animal performed this task twice (under food restriction). We quantified each animal’s movement (speed) for every trial type.
OC: SIP
Food-restricted rats (n = 27) were trained on the SIP paradigm as previously described (Moreno and Flores, 2012). Briefly, after 1 d of habituation to the OC setup, baseline water consumption (drinking) patterns were recorded for 2 consecutive days. Next, rats were trained daily for a minimum of 16 consecutive days or until stable drinking patterns were established (high vs low drinking) under a FI30 reinforcement schedule. Animals were divided into high versus low-drinking groups using a median split of average water consumption for 3 d prior to photostimulation. Subsequently, rats underwent 3 d blocks of on/off photostimulation. The blocks were always run in the following order: no stimulation, 15 Hz (15–20 mW) stimulation, no stimulation, 5 Hz (1–5 mW) stimulation, no stimulation. During the 15 Hz stimulation block, animals received a 5 s stimulation on every trial, occurring 2–7 s postreward delivery. During the 5 Hz stimulation block, photostimulation was delivered throughout the entire session. We quantified the following parameters for each animal (3 d average): water intake, licks at water spout, and food-magazine entries. In a small subset of animals (7/27), post-15 Hz and pre-15 Hz data were calculated using a 2 d average due to experimental challenges (i.e., disconnections, insufficient laser output).
Analysis of behavior
The DeepLabCut software (Mathis et al., 2018) was used to track rat movement in the operant box using recorded video data. Bonsai (Lopes et al., 2015) was employed to track rat movement in the OF arena. All tracking data were analyzed in MATLAB (The MathWorks) to determine speed of movement (cm/s). Water-spout licks and food-magazine entries were registered via an automated procedure during SIP, and a number of presses on each lever were registered during the ICSS task. Animals were excluded from behavioral analysis when virus and/or optical fibers were mistargeted or after loss of both optical fibers.
Histological verification of recording sites, optic fiber targeting, and virus expression
After completion of the experiments, rats were deeply anesthetized using a lethal dose of pentobarbital (14 mg/100 g) and transcardially perfused with saline followed by 4% PFA. In animals in which FSCV measurements were made, recording sites were marked by electrolytic lesion. All brains were removed and postfixed in PFA for 24 h after which they were placed in 30% sucrose for cryoprotection. The brains were rapidly frozen using an isopentane bath, sliced on a cryostat (40 μm coronal sections, −20°C) and stained with Cresyl violet (FSCV recording sites) or dual-labeled fluorescent immunohistochemistry (photostimulation sites and viral expression).
For photostimulation sites and viral expression, serial coronal sections were stored in PBS with 0.02% sodium-azide at 4°C until further use. For immunohistochemical stainings, sections were incubated in blocking buffer (5% bovine serum albumin, 5% normal donkey serum, and 0.2% Triton X-100 in PBS for 1 h), followed by overnight incubation in primary antibodies: chicken anti-GFP (1:1,000, Bio-Connect) and rabbit anti-TPH2 (1:1,000, Abcam) in a blocking buffer. All sections were washed three times (10 min each) in PBS and incubated for 1 h in a blocking buffer containing fluorescent Alexa Fluor-conjugated secondary antibodies (1:1,000, Bio-Connect, 703–545–155 and 711–585–152): donkey anti-chicken A488 (GFP) and A594 donkey anti-rabbit (TPH2). Sections were washed again three times in PBS, with the second wash containing DAPI (nuclei stain; Sigma-Aldrich D9542), and then mounted onto slides and coverslipped with Mowiol mounting medium. Slices were imaged using three fluorescent channels of the ZEISS Axio Scan.z1 slide-scanner at 10× magnification for qualitative expression and targeting verification. For quantitative expression, sections were imaged in z stacks with a confocal microscope (Leica TCS SP5) to verify specificity and sensitivity of viral transduction. Cell counts were conducted manually on confocal image stacks using the ImageJ software.
Statistical analysis
FSCV and behavioral data were analyzed using two-tailed paired or unpaired t tests, repeated-measure ANOVA, or their nonparametric equivalents when appropriate. Post hoc analyses were performed when necessary, and p values were adjusted when multiple comparisons were conducted. Statistical analyses and graphical representations were made using Prism (GraphPad Software 7.03). Statistical significance was set to p < 0.05.
Results
Novel Tph2-Cre rat line enables serotonin-selective investigation
To generate our novel Tph2-Cre rat line, an Cre coding sequence was inserted into the translation start of Exon 1 of the PAC-based Tph2 gene by recombination in E. coli, and the linearized Tph2-Cre DNA (Fig. 1A) was microinjected into the germline of Sprague Dawley rats to obtain an Cre transgenic founder line. The resulting Tph2-Cre founder line showed selective Cre immunostaining in the raphe nuclei (Fig. 1B). We further investigated tissue specificity with dual-labeled fluorescent immunohistochemistry using TPH and Cre antibodies and found their colocalization was restricted to 5-HT neurons (Fig. 1C) with 73.3% of Cre expression colocalized with the total population of TPH neurons in the DRN (specificity) and 83.6% of TPH2 neurons colocalized with the total population of Cre expression (sensitivity; n = 2 animals). Next, we tested the recombination efficiency and tissue specificity of our Tph2-Cre rat line by breeding the Tph2-Cre transgenic rats with an Cre reporter line CAG-loxP.EGFP (Weber et al., 2009; Schönig et al., 2012) to generate a double-transgenic rat line (Fig. 1D, top panel). The lacZ DNA fragment precludes the transcription of the second reporter gene EGFP; thus the visualization of EGFP is an indication of Cre-mediated recombination in double-transgenic rats, which resulted in 79.6% specificity of EYFP-labeled cells colocalized with TPH2 neurons and 85.3% sensitivity of TPH2 neurons colocalized with EYFP-labeled cells (Fig. 1D, bottom panel). In other words, we report the successful expression of Cre in ∼75% of the total number of TPH2-positive neurons, achieving similar results as the previously developed (inducible) transgenic rat line (Weber et al., 2011). Based on these positive results, we microinjected various adenovirus constructs and serotypes into the DRN of our Tph2-Cre rats to investigate the specificity and sensitivity of viral transduction. Sufficient viral transduction was obtained using serotypes AAV5 (Fig. 1E, top panel) and AAVDJ (Fig. 1E, bottom panel), which resulted in 87.7% ± 2.4 specificity and 65.6% ± 5.3 sensitivity (n = 9 animals). We found no effect of sex as a biological variable (n = 5 males; n = 4 females) for either specificity (t(3.727) = 0.692; p = 0.5296) or sensitivity (t(6.66) = 1.009; p = 0.3480).
Validation of Tph2-Cre rat line and optogenetically induced serotonin release. A, Mouse PAC-based construct for the specific expression of Cre in rat serotonergic neurons. B, Immunohistochemical detection of Cre recombinase demonstrates extensive Cre staining in serotonin neurons located in both the anterior (left) and caudal (right) raphe nuclei (dorsal/median raphe nucleus, DRN/MRN). C, Cre expression (red) is restricted to TPH2-positive (green) cells, indicated by colocalization (yellow). D, Top, Tph2-Cre rats were bred with CAG-loxP.EGFP rats to generate double-transgenic Tph2-Cre/CAG-loxP.EGFP rats. Upon Cre-mediated recombination, lacZ is replaced with the second reporter gene, enhanced green fluorescent protein (EGFP). Bottom, Merged expression of EGFP (Cre reporter; green) and TPH2 (red) demonstrate selective expression in DRN serotonin neurons via Cre-mediated recombination (EGFP+/TPH2+). E, Viral-mediated transduction of ChR2 (EGFP, green) into serotonin neurons (TPH2, red) with AAV-5 (top) and AAV-DJ (bottom) shows colocalization (merged, yellow). F, Unilateral virus injection (colocalization of ChR2 and TPH2 in yellow) and optic fiber implantation (white, dotted outline) targeting the DRN were performed at a 30° angle. G, Both FSCV recording electrodes and optical fibers were positioned unilaterally in the amygdala and DRN, respectively. H, Histological verification of FSCV-electrode placement via lesion in the amygdala (left) and optical-fiber placement in the DRN (right). I, Example of an FSCV color plot displaying photostimulation-induced serotonin release (red bar; white-dotted line marks stimulation onset). N-shaped waveform used to record serotonin release superimposed on the image in white. J, Serotonin concentration in the amygdala, before and after administration of SSRI escitalopram (escit) in EYFP (n = 5 animals; left) and ChR2 (n = 4 animals; right) animals. ***p < 0.001.
Next, to test the functionality of the Tph2-Cre line, we employed FSCV to measure DRN-photostimulated serotonin release in the amygdala of anesthetized animals (Fig. 1F,G). Histological analysis confirmed for all animals included in this experiment that the FSCV electrodes, as well as the tip of the optical fibers, were positioned in the amygdala and DRN, respectively (Fig. 1H). Serotonin was reliably released in response to our photostimulation (15 Hz, 20 mW, 30 pulses, 10 ms pulse width; representative color plot in Fig. 1I). Photostimulated release of serotonin was measured exclusively in ChR2 rats (n = 4) and not EYFP rats (n = 5), before and after the administration of SSRI escitalopram (Fig. 1J). In ChR2 animals, there was a significant increase in serotonin release with higher doses of escitalopram (F(1.296, 187.9) = 106.3; p < 0.0001). Together, these findings validate our novel Tph2-Cre rat line as a promising tool to selectively study the serotonin system in rats.
Naturally reinforced behaviors are unaffected by DRN5-HT photostimulation
Prior to all behavioral testing, viral injections and bilateral optic fiber implantations were performed at a 30° angle targeting the DRN (Fig. 2A). Histological analysis confirmed for each animal included in this study that the tips of at least one optical fiber were positioned in the DRN (Fig. 2B).
No effect of optogenetic stimulation of DRN serotonin neurons on “naturally” reinforced behavior. A, For all behavioral experiments, virus injections and bilateral optic fiber implantation (white-dotted outlines) targeted the DRN (at a 30° angle). Viral expression of ChR2 is shown in green, TPH2 neurons in red, and colocalization in yellow. B, Histological verification of optical-fiber position in the DRN of EYFP (n = 34 on-target fibers; 20 rats) and ChR2 (n = 47 on-target fibers; 27 rats) animals. C, WN is an aversive stimulus as validated using an approach–avoidance task where rats (n = 11) foraged for food pellets in a WN-paired quadrant of an OF (left): with increasing WN intensity, rats spent significantly less time foraging (middle). This behavioral response was stable across sessions (right). D, A 90 dB WN (EYFP, 7 animals; ChR2, 8 animals) administered for 6 s induces an increase in locomotion speed (left). A 15 Hz DRN photostimulation at 15–20 mW did not change WN-induced speed in ChR2 and EYFP animals (right). AUC during the 6 s epoch (0–6 s; between dotted lines) is shown in insets. E, Left, SIP was induced by an FI30 reinforcement schedule in food-restricted animals that had free access to a water spout. Right, After water intake stabilized, animals underwent two 3 d blocks of photostimulation (flanked by 3 d blocks of regular SIP). During the first photostimulation block, 15 Hz of 15–20 mW optogenetic stimulation was delivered every trial for 5 s (starting 2 s after reward delivery). During the second block, 5 Hz of 1–5 mW stimulation was delivered throughout the entire session. F, Rats were divided into high- (ChR2: 5 animals; EYFP: 8 animals) and low-drinkers (ChR2: 6 animals; EYFP: 8 animals) based on a median split of prestimulation (Pre stim) water intake (3 d average). G, The 15 Hz stimulation block (3 d average) did not alter water intake (left) or number of licks (right). H, The 5 Hz stimulation block resulted in no change in water intake (left) or licks (right). I, The 15 Hz stimulation blocks (3 d average) increased food-magazine entries in ChR2 animals (both high and low drinkers). J, The increase in food-magazine entries in ChR2 animals does not occur when the stimulation is delivered, but during an anticipatory period before the next reward delivery. *p < 0.05; **p < 0.01; ***p < 0.001; ns, not significant.
We first validated that WN was aversive in our setting by employing an approach–avoidance task in which food pellets were placed between grid-floor bars within a WN-paired quadrant in an OF (Goedhoop et al., 2022). We found that with increasing WN intensity (dB), animals spent less time in the WN-paired quadrant (F(1.836, 18.36) = 8.647; p = 0.0027), and this response was stable between exposures (Fig. 2C). Post hoc analysis using Tukey's multiple-comparison test revealed that animals spent significantly less time foraging for pellets during 90 dB WN compared with 0 and 70 dB (0 vs 90, p = 0.0022; 70 vs 90, p = 0.0404). Based on these aversive properties of WN and serotonin’s theorized role in threat responses (Seo et al., 2019), we employed 90 dB WN to examine the influence of serotonin activation on the animal's reactivity to an aversive stimulus or a primary reinforcer. In our aversive-conditioning task, animals were exposed to a 6 s epoch of 90 dB WN, paired with or without photostimulation. We observed no difference in speed between EYFP and ChR2 animals during WN-only trials [area under the curve (AUC), t(13) = 0.2477; p = 0.8083; control] or WN + photostimulation trials (AUC, t(13) = 0.5535; p = 0.5893; Fig. 2D).
Next, we investigated the role of serotonin in compulsive behavior using the SIP paradigm, in which food-restricted animals received pellets on an FI30 reinforcement schedule for 30 min daily in an OC with free access to water (Fig. 2E, left panel). Based on previous SIP studies that show pharmacologically induced increase of serotonin reduces compulsive drinking (Woods et al., 1993; Hogg and Dalvi, 2004; Navarro et al., 2015; Prus et al., 2015), we hypothesized that physiologically relevant optogenetic activation of DRN serotonin neurons may decrease water intake. After habituation to the OC and evaluation of baseline drinking, animals were trained daily until drinking patterns stabilized (minimum of 16 d), after which animals underwent 3 d blocks of on/off photostimulation. During the first photostimulation block, animals experienced a 5 s, 15 Hz photostimulation on every trial, 2–7 s after reward delivery. During the second photostimulation block, animals received 5 Hz photostimulation throughout the entire session (Fig. 2E, right panel). Animals were divided into high- and low-drinking groups using a median split of averaged water intake during their final three regular SIP sessions. To compare drinking behavior across the four groups (EYFP low, EYFP high, ChR2 low, ChR2 high), a Kruskal–Wallis test was performed and revealed a significant difference between the groups (H(3) = 31.42; p < 0.0001). Post hoc analyses by Dunn's multiple comparisons test revealed no differences between ChR2 and EYFP high (p > 0.9999) or low (p > 0.9999) drinkers. ChR2 high drinkers drank significantly more than both EYFP low (p = 0.0003) and ChR2 low (p = 0.0008) drinkers. Additionally, EYFP high drinkers drank significantly more than ChR2 low (p = 0.0007). These data demonstrate that high drinkers consumed more than low drinkers, with no differences between genotypes within each drinking group. Prior to the first stimulation block (pre stim), water intake was equal between both high (ChR2 vs EYFP: t(11) = 0.5844; p = 0.5708) and low (ChR2 and EYFP: t(12) = 1.219; p = 0.2462) pairings of animals (Fig. 2F). Under both stimulation parameters (15 and 5 Hz), we observed no effect on water intake or licks at the water spout. Furthermore, we performed a separate experiment in which we chemogenetically inhibited DRN and MRN serotonin neurons in SIP and found no effect on the activation of “inhibitory” designer receptors exclusively activated by designer drugs (DREADDs) in the MRN and DRN on compulsive drinking at both early and late time points during SIP development (Extended Data Fig. 1-1). Interestingly, we found food-magazine entries were significantly higher in ChR2 animals than EYFP during the 15 Hz stimulation block compared with prestimulation levels [percentage change (pre vs 15 Hz); ChR2 vs EYFP, t(25) = 3.121; p = 0.0045; Fig. 2I]. Further analysis (Fig. 2J) revealed that ChR2 food-magazine entries increased ∼20–30 s postreward delivery, i.e., in the anticipatory response to the next (nearing) reward pellet; there was no increased food-magazine entry during the 15 Hz photostimulation itself (2–7 s postreward delivery). These null results on compulsive behavior, together with the negative results in reactivity to aversive stimuli, suggest widespread optogenetically induced serotonin release in the DRN is insufficient to alter naturally reinforced behaviors.
DRN5-HT photostimulation produces distinct behavioral reinforcement
To study the influence of DRN activation on behaviors in the absence of natural reinforcers, we aimed to examine spontaneous locomotor activity of animals in a neutral environment. After 1 d of habituation to the setup, a trained observer manually delivered 3 s, 15 Hz photostimulation, while the animal was either at rest or in motion (Fig. 3A). When animals were at rest prior to photostimulation, ChR2 and EYFP rats increased their speed equally (t(26) = 0.5802; p = 0.5668); however, when in motion prior to photostimulation, ChR2 animals slowed down less than EYFP controls (t(26) = 3.389; p = 0.0022; Fig. 3B).
Self-stimulation of DRN serotonin neurons exclusively reinforces discrete operant behavior. A, Locomotion in an OF: 3 s of 15 Hz stimulation was applied, while animals were in motion or at rest. B, Photostimulation (EYFP, 15 animals; ChR2, 13 animals) did not differentiate speed in resting rats (left) but did sustain locomotion speed of ChR2 rats that were in motion during stimulation onset (right), relative to EYFP controls. Bar graph insets depict average change in speed during stimulation (0–3 s; gray area). C, The RT-PP task in which entry into one unmarked quadrant (shaded) induced either no (0 Hz), 5 Hz of 1–5 mW, or 15 Hz of 15–20 mW DRN photostimulation. D, The percentage of time stimulated (left) and (E) the average speed within the photostimulation-paired quadrant (right) was unaffected by DRN photostimulation in both ad libitum fed (EYFP, 13 rats; ChR2, 18 rats) and food-restricted (EYFP, 13 rats; ChR2, 9 rats) animals. F, The ICSS task in which an active-lever press triggered 3 s of 15 Hz DRN photostimulation (EYFP, 17 animals; ChR2, 21 animals). G, Left, ChR2 but not EYFP rats pressed the active lever more than the inactive one (across sessions). Right, Active-lever presses (averaged across 5 sessions) in ad libitum fed and food-restricted conditions. H, The average duration of self-photostimulation was ∼3 s (equivalent to one lever press; left) and the maximum duration of self-photostimulation bouts was ∼7.5 s (2.5 lever presses; right). Neither variable differed between EYFP and ChR2 animals. I, ChR2 animals spent significantly more time self-stimulating than EYFP animals (even more so in the food-restricted state; on average across 5 sessions), on average up to 18% of total time. J, ChR2 animals pressed the active lever consistently throughout the 30 min session in both ad libitum (left) and food-restricted states (right). **p < 0.001; ns, not significant.
Next, we investigated whether exclusive DRN photostimulation (0, 5, or 15 Hz) itself was reinforcing, by pairing it with an unmarked quadrant in the OF. To account for multiple comparisons within each hunger state, a Holm–Bonferroni’s correction was applied. Irrespective of hunger state, we observed no RT-PP (increased time spent in quadrant, i.e., being stimulated) across any of the conditions after correction, 0 Hz (ad libitum, U = 64.5; p = 0.0349; corrected p = 0.2094; restricted, U = 44; p = 0.3480; corrected p = 1.0), 5 Hz (ad libitum, U = 64; p = 0.0331; corrected p = 0.1986; restricted, U = 45.5; p = 0.4026; corrected p = 1.0), and 15 Hz (ad libitum, U = 105.5; p = 0.6587; corrected p = 1.0; restricted, U = 53.5; p = 0.7558; corrected p = 0.7558; Fig. 3D). The average speed of ChR2 animals within the photostimulation-paired quadrant was also unaffected compared with EYFP across 0 Hz (ad libitum, U = 77; p = 0.8108; corrected p = 1.0; restricted, U = 38; p = 0.9175; corrected p = 1.0), 5 Hz (ad libitum, U = 86; p = 0.9628; corrected p = 1.0; restricted, U = 12; p = 0.0178; corrected p = 0.1068), and 15 Hz (ad libitum, U = 58; p = 0.3912; corrected p = 0.3912; restricted, U = 18; p = 0.0439; corrected p = 0.2195) conditions (Fig. 3E).
Lastly, we examined whether animals would perform a discrete, operant action to receive the same DRN photostimulation. All rats were trained to press a lever in order to receive a 3 s, 15 Hz DRN photostimulation (Fig. 3F). A Holm–Bonferroni’s correction was applied to all parametric tests. ChR2 rats reliably lever-pressed more than EYFP rats when fed ad libitum (t(36) = 4.006; p = 0.0003; corrected p = 0.0006) and after food restriction (t(26) = 0.5802; p < 0.0001; corrected p = 0.0006), and this lever pressing was exacerbated by food restriction compared with ad libitum (t(20) = 5.717; p < 0.0001; corrected p = 0.0005; Fig. 3G). The average length of the animal's stimulation bouts (ad libitum, U = 139; p = 0.2556; restricted, U = 174.5; p = 0.9133) and maximum stimulation bout (ad libitum, U = 170; p = 0.8100; restricted, U = 177; p = 0.9709) did not differ between EYFP and ChR2 animals (Fig. 3H). However, ChR2 animals spent more time receiving stimulation than EYFP under both hunger states (ad libitum, t(36) = 3.885; p = 0.0004; corrected p = 0.0004; restricted, t(36) = 4.846; p < 0.0001; corrected p = 0.0004; ChR2 ad libitum vs restricted, t(20) = 5.199; p < 0.0001; corrected p = 0.0003; Fig. 3I). This increased frequency of active-lever pressing was present throughout the session (Fig. 3J). We analyzed these results further to determine whether variability in lever pressing correlated with different DRN regions. We found no detectable effects of anatomical position of photostimulation on active-lever pressing in either the medial–lateral (Niederkofler et al., 2015; ad libitum, t(10) = 1.941; p = 0.0810; food restricted, t(10) = 2.13; p = 0.0590) or rostral–caudal (Murphy and Lesch, 2008; ad libitum, t(15) = 0.4507; p = 0.6587; food restricted, t(15) = 0.2654; p = 0.7943) targeting of the DRN (Extended Data Fig. 2-1). Thus, together with the increased anticipatory food responses in SIP, these data seem to suggest that DRN photostimulation is mildly reinforcing on its own and the DRN may more generally modulate response vigor.
Discussion
In this work, we describe the development and characterization of a novel Tph2-Cre rat line, which we subsequently employ to enable selective study, on a functionally relevant timescale, the role of the DRN serotonin system in motivated behavior. We validated the selectivity of our Cre-mediated recombination in DRN serotonin neurons (1) anatomically using immunohistochemistry and (2) functionally by optogenetically stimulating serotonin neurons and measuring the resulting serotonin release in vivo using FSCV (in the amygdala). To investigate the behavioral function of serotonin, we optogenetically activated DRN serotonin neurons in several behavioral paradigms that probe prompted and unprompted locomotion, compulsive behavior (SIP), and behavioral reinforcement (RT-PP and ICSS). Our novel rat Cre line has several advantages over existing mouse lines for this type of work, as rats have a bigger brain size which improves access to the hard-to-reach brainstem location of serotonin–neuron cell bodies, as well as the capacity to carry heavier intracranial implants; additionally, rats are able to perform in more complex behavioral paradigms and develop easily detectable compulsive behavior in the SIP paradigm. Our findings demonstrate that, although optogenetic stimulation of brainwide serotonin release has an overall relatively mild effect on behavior which is detectable only in the absence of natural reinforcers, it robustly reinforces operant DRN5-HT self-stimulation behavior, an effect that was significantly modulated by physiological state.
The brain serotonin system is linked to OCD. For example, OCD is associated with a risk variant in 5-HT transporter genes, altered cortical 5-HT2A receptor availability, and modified serotonergic input into frontosubcortical circuits (Perani et al., 2008; Walitza et al., 2014; Figee et al., 2016; Derksen et al., 2020). Furthermore, SSRIs are the most commonly prescribed medication to treat OCD patients. Consistently, animal studies have reported a reduction in compulsive behavior in response to SSRI administration (Woods et al., 1993; Joel et al., 2004; Welch et al., 2007). However, the precise implication of serotonin in compulsive behavior remains unclear. Here, we studied compulsive drinking elicited behaviorally in the SIP paradigm, which can be attenuated by administration of SSRIs, 5-HT1A/B antagonists, and/or 5-HT2A/C agonists (Woods et al., 1993; Hogg and Dalvi, 2004; Navarro et al., 2015; Prus et al., 2015). To our surprise, SIP water intake was not altered by optogenetic stimulation of brainwide 5-HT release, despite the use of two sets of photostimulation parameters that were selected based on physiological firing-rate modes of serotonin neurons: one to mimic phasic, burst firing of serotonin neurons (reportedly found in response to rewards; Ranade and Mainen, 2009; Liu et al., 2014) and one to mimic an increase in the tonic, steady release of serotonin (Jacobs and Azmitia, 1992; Zhao et al., 2011). Based on the aforementioned studies as well as findings that suggest that serotonin promotes patience (Miyazaki et al., 2014, 2018) and decreases locomotion (Correia et al., 2017a), we had hypothesized that our “phasic” photostimulation would disrupt the urge to drink large amounts of water immediately following reward delivery and that our “tonic” photostimulation would more broadly decrease water intake across the behavioral session, similarly to abovementioned pharmacological manipulations. It is possible that our photostimulation was either not powerful enough, not delivered at the optimal epoch of a trial, or not delivered chronically enough to impact the serotonin system sufficiently. This would be consistent with the fact that most effective pharmacological serotonin treatments require daily administration of high doses for multiple days. Although our optical stimulation did not alter compulsive drinking, we did observe an increase in food foraging prior to reward delivery (Fig. 2I,J), indicating that photostimulation per se was successful and the serotonin system may be involved in the anticipation of reward.
Historically, the serotonin system was conceptually posited to correlate positively with locomotor activity, or wakefulness (Trulson et al., 1981; Steinfels et al., 1983; Moriya et al., 2021). In contrast, more recent investigations suggest a negative correlation (Ohmura et al., 2014; Teissier et al., 2015; Correia et al., 2017b; Kubitschke et al., 2022) or no correlation at all (Zhao et al., 2006; Zhong et al., 2017; Lottem et al., 2018; Walsh et al., 2018; Browne et al., 2019; Wang et al., 2019); for review, see Flaive et al. (2020). One possible explanation for these inconsistent findings is engagement of functionally distinct subpopulations of DRN5-HT neurons between studies (Ren et al., 2018; Paquelet et al., 2022). Alternatively, locomotor effects of serotonin might be twofold, where serotonin increases locomotion in high-threat and diminishes locomotion in low-threat situations; this possible state-dependent variation of serotoninergic locomotor effects may even extend to handling stress, explaining inconsistent findings between laboratories due to differing animal-handling procedures in their facilities. We examined the effect of DRN5-HT photostimulation on locomotor activity in different behavioral states: during (1) resting state, (2) spontaneous locomotion, and (3) locomotion induced by an aversive stimulus (“threat”). Although our DRN5-HT photostimulation did not prompt the initiation of locomotion in resting animals, it moderately sustained ongoing spontaneous locomotion. In contrast, it did not alter aversion-induced locomotion, perhaps because WN was not perceived as threatening enough. Furthermore, the MRN may play a more prominent role in aversion-induced locomotion, as it is implicated in anxiety (Ohmura et al., 2014; Abela et al., 2020) and exhibits an antagonistic relationship with the DRN (Lechin et al., 2006; Teissier et al., 2015). Together, our findings suggest mild, state-dependent stimulating effects of serotonin on locomotion.
The DRN projects extensively to circuits involved in reward processing. Through the use of viral-mediated techniques, the reinforcing effects of DRN5-HT neurons have been postulated to originate from its interconnection with the ventral tegmental area (McDevitt et al., 2014; Nagai et al., 2020). A decades-long debate over the role of serotonin in reward processing began with early pharmacological studies indicating increased electrical ICSS of brain–reward hotspots (other than the DRN) in response to administration of drugs that reduce serotonin transmission (Poschel and Ninteman, 1971; Poschel et al., 1974; Phillips et al., 1976), whereas such ICSS rates were reduced in response to pharmacological agents that increase serotonin transmission (Bose et al., 1974; Redgrave and Ian Horrell, 1976; Lee and Kornetsky, 1998). Furthermore, electrical stimulation of the DRN itself, presumably increasing brainwide serotonin release, increases ICSS (Margules, 1969; Simon et al., 1976; Van Der Kooy et al., 1978; Corbett and Wise, 1979; Rompre and Miliaressis, 1985); importantly, as only a subset of DRN neurons are serotonergic (Wang et al., 2019; Liu et al., 2020), DRN5-HT activity specifically was later found to be relevant for reward processing (Ranade and Mainen, 2009; Miyazaki et al., 2014; Li et al., 2016); for review, see Kranz et al. (2010) and Liu et al. (2020). However, optogenetic activation of DRN5-HT has yielded inconsistent results, with some studies reporting reinforcing effects [(Nagai et al., 2020) ←PP; ICSS→ (Liu et al., 2014; McDevitt et al., 2014; Lalive et al., 2018; Nagai et al., 2020)] and others not [(McDevitt et al., 2014; Fonseca et al., 2015; Walsh et al., 2018; Wang et al., 2019) ←PP; ICSS→ (Miyazaki et al., 2014, 2018)].
Our findings in this context were also mixed. On the one hand, DRN5-HT photostimulation did not produce RT-PP in rats that were food restricted. This finding was surprising given serotonin’s association with signaling satiety (Voigt and Fink, 2015), which led us to expect that it may provide relief from hunger. However, our RT-PP task assessed preference for a specific location with no distinguishing features to delineate its position, and therefore rats may not have been able to effectively learn its position. On the other hand, we observed a robust reinforcing effect of DRN5-HT ICSS that was amplified by food restriction. In contrast to RT-PP, ICSS assessed a discrete and clearly delineated operant behavior. Another factor that may have impeded RT-PP is the duration of DRN5-HT photostimulation: animals stimulated themselves for an average duration of 3 s in ICSS, whereas an entry into the stimulation-paired quadrant in RT-PP produced a 5 s stimulation (on average). It is therefore possible that serotonin may only be reinforcing in low release concentrations or for brief periods of time and/or when knowingly, voluntarily, and actively pursued. Although DRN5-HT photostimulation was self-reinforcing, behavior driven by primary reinforcers was not affected; thus, serotonin appears to have only mildly reinforcing effects. Furthermore, the effects of serotonin seem to have a strongly adaptive component, since we found that food restriction exacerbates its reinforcing effect, and others have found that stress alters its behaviorally activating effects (Seo et al., 2019). Finally, our food-restricted rats reliably increased their anticipatory response vigor to a highly predictable food reward in SIP (Fig. 2I,J), further supporting the notion that serotonin has a significant role in reward-related actions.
To summarize, our findings establish our novel Tph2-Cre rat line as a valuable tool for investigating the role of serotonin and posit serotonin release as involved in the proceeding of spontaneous movement as well as the reinforcement of discrete operant behavior as a function of behavioral state, without impacting behaviors promoted by natural reinforcers or compulsive behavior. Such maintenance of ongoing, unprompted actions and discretely reinforced actions, as well as the augmentation of anticipatory reward responses in SIP, align with contemporary theories positioning serotonin signaling as a modulator of action control and selection, behavioral reinforcement, and promoter of behavioral persistence (Cools et al., 2005; Bailey et al., 2018; Lottem et al., 2018; Miyazaki et al., 2020). To further evaluate how serotonin-promoted persistence affects both active (e.g., ongoing movement) and passive (e.g., waiting) behavior, future studies that selectively manipulate the serotonin system and its specific pathways are required to identify the mechanisms through which serotonin affects behavioral transitions.
Footnotes
We thank Ralph Hamelink and Nicole Yee for their technical support and Aishwarya Parthasarathy for the assistance with data analysis.
The authors declare no competing financial interests.
- Correspondence should be addressed to Ingo Willuhn at i.willuhn{at}nin.knaw.nl.
