Sex differences in risk-based decision-making and the modulation of risk preference by dopamine-2 like receptors in rats

Abstract

Heightened risk-based decision-making is observed across several neuropsychiatric disorders including schizophrenia, bipolar disorder, and Parkinson’s disease, yet no treatments exist that effectively normalize this aberrant behavior. Preclinical risk-based decision-making paradigms have identified the important modulatory roles of dopamine and sex in the performance of such tasks, though specific task parameters may alter such effects (e.g., punishment and reward values). Previous work has highlighted the role of dopamine 2-like receptors (D2R) during performance of the Risk Preference Task (RPT) in male rats, however sex was not considered as a factor in this study, nor were treatments identified that reduced risk preference. Here, we utilized the RPT to determine sex-dependent differences in baseline performance and impact of the D2R receptor agonist pramipexole (PPX), and antagonist sulpiride (SUL) on behavioral performance. Female rats exhibited heightened risk-preference during baseline testing. Consistent with human studies, PPX increased risk-preference across sex, though the effects of PPX were more pronounced in female animals. Importantly, SUL reduced risk-preference in these rats across sexes. Thus, under the task specifications of the RPT that does not include punishment, female rats were more risk-preferring and required higher PPX doses to promote risky choices compared to males. Furthermore, blockade of D2R receptors may reduce risk-preference of rats, though further studies are required.

1. Introduction

Engaging in risky behavior is increased in several neuropsychiatric disorders, including bipolar disorder (BD) and schizophrenia (Adida et al., 2011; Brand et al., 2005; Pedersen et al., 2017; Premkumar et al., 2008; Sevy et al., 2007), and is a proposed risk factor for suicidality in such populations (Jollant et al., 2005; Richard-Devantoy et al., 2016;Sastre-Buades et al., 2021). Risk-taking is also linked to substance misuse and poorer treatment adherence (Baeza-Velasco et al., 2020; Remien et al., 2007; Stevens et al., 2013), further implicating the adverse consequences of such behavior in neuropsychiatric populations. There are no treatments that alleviate aberrant risk-taking behavior; in fact, some medications may exacerbate risk-taking, such as dopamine (DA) agonists (e.g. pramipexole; PPX), which induce compulsive and gambling problems in people with BD and Parkinson’s disease (Dodd et al., 2005; Voon et al., 2006; Weintraub et al., 2006). Hence, it is important to determine the mechanisms underlying high risk-taking so-as to develop targeted therapeutics that might alleviate such behavior in these populations.

It is well-established that dopamine (DA) plays a crucial role in risk-taking behavior. For one, the aforementioned people with psychiatric disorders e.g. schizophrenia, BD, and Parkinson’s have altered dopamine functioning and perform worse on risk-based decision-making tasks (Adida et al., 2011; Brand et al., 2005; Pedersen et al., 2017; Premkumar et al., 2008; Sevy et al., 2007). Pathological gamblers also demonstrate risk-based decision-making impairments and exhibit altered DA dynamics during task performance compared to healthy controls (Linnet et al., 2011a, 2011b). Even amongst healthy populations, amphetamine-induced elevations in DA correlate with risk-taking behavior in the Iowa Gambling Task (IGT; Oswald et al., 2015), a commonly used risk-based decision-making paradigm. Finally, preclinical pharmacological studies in animal models support causal links between DA and risk-behavior, in that DA manipulations change risk-based decision-making in rodents (Simon et al., 2011; St Onge et al.,2010). Thus, parsing the role of DA in risk-taking behavior remains vital.

There are two main classes of DA receptors, DA-1-like and DA-2-like receptors (D1R and D2R respectively), both of which contribute to, or have been associated with, risk-taking behavior in a variety of tasks across species (see Soutschek et al., 2023 for a review of human studies; Ishii et al., 2018; Simon et al., 2011; Soutschek et al., 2023; St Onge et al., 2010). However, the specific contribution of each DA receptor class to risk-taking appears to be mediated by several factors including task specifications (Soutschek et al., 2023; Winstanley and Floresco, 2016). In particular, the contribution of DA receptor subtypes to risk-based decision-making in tasks that vary reward and punishment values is unclear (Winstanley and Floresco, 2016). Thus, utilizing tasks that hold one of these variables constant (i.e. reward or punishment value), may help better disentangle the neural mechanisms underlying risk-taking behavior. Furthermore, sex also appears to mediate risk-based decision-making, as well as the contribution of DA, and specific DA receptor subtypes, to risk-taking behavior (Georgiou et al., 2018; Hynes et al., 2020; Ishii et al., 2018). Indeed, females exhibit higher risk-taking in the IGT (Overman and Pierce, 2013; van den Bos et al., 2013), which is similarly observed in studies utilizing rodent analogs of the IGT (i.e. rIGT; Overman and Pierce, 2013; van den Bos et al., 2013). Further, in the rIGT risk-based decision-making is impaired by administration of the D2R agonist quinpirole selectively in females, whereas administration of the D2R antagonist eticlopride disturbed risk-based decision-making selectively in males (Georgiou et al., 2018). Hence, sex must also be considered alongside interpretations of DA’s role in risk-based decision-making.

Recently, Zalocusky et al. (2016) designed the risk preference task (RPT) where the D2R, but not the D1R, heavily influenced task performance in male rats. In the RPT, rats initiated a trial via a 1-sec nosepoke, termed the decision-making period, then choose between one port with a certain medium sized reward, or another port associated with a high probability of a small reward coupled with a low probability of a large reward. This task is unique in that the expected value across choice aperatures is identical with sufficient trials, and therefore choice of the ‘risky’ option does not lead to explicit loss or punishment as typically found in other risk-based decision-making tasks (Rivalan et al., 2009; Simon and Setlow, 2012; Zeeb et al., 2009). Thus, the RPT presents particular research value in its ability to parse apart behavior controlled by sensitivity to reward values that are not influenced by punishment or explicit loss outcomes.

Using in vivo fiber photometry Zalocusky et al. (2016) demonstrated that D2R-, but not D1R-, expressing neurons in the nucleus accumbens core (NAcc) were activated by a loss outcome following a risky choice and the magnitude of this signal correlated with future risk-taking behavior. Optogenetic manipulation of D2R-expressing neurons during the decision-making period, but not during other events, was capable of altering choice (safe versus risky) behavior. Consistently, systemic or intra-NAcc injection of the D2R-like agonist PPX increased risk-taking behavior, whereas a D1R agonist had no effect. This work suggested that risk-taking in the RPT correlated with reduced sensitivity to losses that were encoded by activity of NAcc D2R-expressing neurons, which in turn led to more risk-taking. Thus, risk-taking under conditions that do not include absolute loss or punishment outcomes, may be more heavily influenced by the D2R than conditions devoid of such factors. More work is needed however, to substantiate such claims given that: 1) limited data has been collected using the RPT especially identifying the impact of blocking D2Rs; 2) reproducibility of such findings are important; and 3) sex differences in the RPT have yet to be reported.

Here, we utilized the RPT to determine sex-dependent differences in baseline performance, and impact of the D2R-like agonist PPX, and antagonist sulpiride (SUL), on behavioral performance. We hypothesized that: 1) females would show a higher risky decision-making profile relative to males; 2) PPX would increase risk-based decision-making irrespective of sex; 3) SUL would reduce risk-based decision-making irrespective of sex.

2. Materials and methods

2.1. Animals

16 female and 16 male adult Long Evans rats (Charles River) were used in this study. Animals were pair-housed in ventilated shoebox cages with standard environmental enrichment (plastic tube housing and nesting material; LWH: 15.5” x 11.5” x 8”) in a temperature-controlled room on a reversed light-dark cycle (07:00/19:00). Food and water was provided ad libitum, except during operant training and testing when animals were food restricted to 85% of their free-feeding body weight. All behavioral testing began at least 2 h into the animal’s dark (active) phase and only occurred during their active period. Rats were maintained in a University of California San Diego (UCSD) animal facility which meets all federal and state requirements for animal care, and all procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at UCSD.

2.2. Drugs

Treatments were administered 10 min prior to cognitive testing. In Experiment 1, pramipexole dihydrochloride (PPX; Sigma-Aldrich, A1237) was dissolved in saline at 0.15 and 0.3 mg/ml and injected intraperitoneal (i.p.) at a volume of 1 ml/kg to produce 0.15 mg/kg and 0.3 mg/kg doses, respectively. These doses of pramipexole elevated risky decision-making in males performing the RPT (Zalocusky et al., 2016). In Experiment 2, sulpiride (SUL; Sigma-Aldrich, S8010) was dissolved in saline at 60 mg/ml and injected i.p. at a volume of 1 ml/kg to produce a 60 mg/kg dose. This dose of SUL was chosen as it has disrupted reward-based decision-making in rats (Groman et al., 2018). Both PPX and SUL have a half-life of up to 4 h in rats (Benakis et al., 1976; Ferger et al., 2010), therefore we were confident that drug was onboard throughout the 10 min preinjection times and 2 h of testing.

2.3. Apparatus

RPT choice behavior was measured using 9-choice nosepoke operant chambers (Med Associates Inc., St. Albans, VT and Lafayette Instrument Company, Lafayette, IN; LWH: 10.5–12″ x 10.5 “x 12”), located in larger sound-attenuating cabinets (LWH: 23″ x 18″ x 18.5), with built-in fans to mask noise and provide ventilation. Nosepoke apertures consisted of an LED light for stimulus presentation and an infrared beam to detect responding. Reward (strawberry milkshake; Nesquik in nonfat milk), was delivered into a magazine on the opposite wall of the chamber that contained an LED light to signal reward delivery and an infrared beam to detect reward collection. Stimulus outputs and response inputs were managed by a SmartCtrl Package (8-In/16-Out) with additional interfacing by MED-PC for Windows (Med Associates Inc., St. Albans, VT) using custom programming.

2.4. Risk preference task

The RPT created by Zalocusky et al. (2016) was adapted for these experiments. Rats were first conditioned to associate the magazine with reward delivery during a 20-min session wherein 50 μl strawberry milkshake was dispensed on a 15-sec fixed interval schedule into an illuminated magazine. Rats were maintained on this program until at least 60 reward collections were made across two consecutive days. Next, animals moved on to RPT specific training which consisted of three phases that preceded the final version of the RPT (see Fig. 1A). Each training phase consistent of 90-min sessions and animals were moved on to subsequent phases once 200 rewards were collected within a single session for two consecutive days. In phase 1 (free-choice training), rats could make an operant response (nosepoke) into either of the two illuminated choice apertures (#3 and #7) to earn 50 μl of reward on a fixed-ratio 1 (FR1) schedule of reinforcement. In phase 2 (forced-choice training), rats could nosepoke in the randomly selected illuminated choice aperture (#3 or #7) to earn 50 μl of reward on an FR1 schedule. In the third and final phase of training (initiate trial training), rats could initiate a trial by maintaining a 1-sec nosepoke in the middle aperture (#5). On the first trial of this phase rats were required to hold a 250 ms nosepoke to activate the illumination of the two adjacent choice apertures (#3 and #7), which rats could then respond in either for reward delivery. Each subsequent trial added an additional 5 ms to the nosepoke hold length required to initiate a trial initiation until a full 1-sec hold length was achieved by the end of session.

Fig. 1. Illustration of the RPT and Experimental Timeline.

A) Three training phases precede the RPT: 1) in the free-choice FR1 phase nosepokes into one of two illumination choice apertures (3 and 7) were rewarded on an FR1 schedule of reinforcement; 2) In the forced-choice FR1 phase one of the two choice apertures were illuminated and a nosepoke into this aperture were rewarded on an FR1 schedule of reinforcement; 3) In the initiate trail training phase the middle aperture (5) was illuminated and rats were required to hold a 250 ms nosepoke to initiate a trial. Upon trial initiation both adjacent choice apertures became illuminated and a nosepoke response in either aperture led to reward delivery (i.e. a free-choice trial). The hold time required to initiate a trial increased by 5 ms with each subsequent trial until a 1-sec nosepoke hold was achieved by the end of the session. All training sessions were 90 min in length and animals were moved onto the next phase once 200 reward collections were achieved across two consecutive sessions. The RPT consisted of 400 trials wherein the first 50 trials were forced-choice trials and the remaining 350 free-choice trials. Across both trial types, one choice aperture was set to deliver 50 μl of reward with a 100% probability (defined as the safe option), and the other a 75% probability of a 10 μl reward, and a 25% probability of 170 μl reward (defined as the risky option). Trials did not time out across all training and testing phases, requiring choice before succession to the next trial.

Following training, rats moved onto the RPT that consisted of a block of 50 forced-choice trials followed by a block of 350 free-choice trials over 120-min. Rats initiated a trial by holding a 1-sec nosepoke in the middle aperture (#5). During forced trials, rats were required to respond in the randomly selected illuminated choice aperture now set to deliver 50 μl of reward with a 100% probability (defined as the safe option), or a 75% probability of a 10 μl reward coupled with a 25% probability of 170 μl reward (defined as the risky option). During the remaining 350 choice trials, both choice apertures (#s 3 and 7), were illuminated following trial initiation and rats could choose which aperture to respond in for the same associated reward contingencies used during forced trials. Trials did not time out during training and testing sessions, requiring choice behavior prior to trial succession. The choice aperture associated with the risky option was counterbalanced across animals.

The primary outcome measure was percent risk choice (%RC; percentage of trials where the risky aperture was selected). Secondary outcome measures include decision-making metrics (percentage [%] of trials with a safe-stay, risky-win stay, risky-lose stay), total trials completed, choice latency (duration to choose a choice aperture following trial initiation), and reward collection latency.

2.5. Experimental timeline

Fig. 1B illustrates the experimental timeline in this study. All rats completed training and were then exposed to the RPT for 14 days. Baseline task performance was calculated by averaging behavior across the final three sessions preceding testing. On day 15 rats were exposed to an injection of saline prior to the RPT to reduce injection stress. Consistent with Zalocusky et al. (2016), both experiments utilized a within-subjects drug regimen that consisted of 3 drug exposures, inter-leaved between 4 saline washouts. In Experiment 1, rats were assigned to one of two doses of PPX (doses counterbalanced using baseline %RC performance). Rats received 6 drug-free sessions following Experiment 1. Six animals were removed from the study prior to Experiment 2 as they were required to be used elsewhere. In Experiment 2, all rats received the same dose of SUL using the drug testing regimen described above.

2.6. Statistical analyses

The first 50 forced-choice trials were not included in any data analyses. In cases of extreme choice preference (rats that only chose safe or risky options), decision-making metrics (%safe-stay trials, %safe-shift trials etc.), could not be computed and were therefore not analyzed. Baseline data included all rats (n = 32) and were calculated as averaged performance across the last three days of RPT prior to drug exposure. Independent-samples t-tests were used to compare baseline performance between sexes. In both experiments, data were collapsed by drug across sessions and analyzed as repeated measures mixed analysis of variance (ANOVA), with session (i.e. drug), as a within-subjects factor, and dose and sex as between-subjects factors. Mauchly’s test was utilized to report sphericity and degrees of freedom were corrected using Greenhouse-Geisser when sphericity violations were identified. The primary outcome variable %RC, and total trials completed were also assessed across all 7 individual test sessions. Given the experimental design utilized, only differences between a PPX test session and its respective saline control day (preceding saline administration day), were reported. Animals that did not complete at least 20 trials on any test session (Experiment 1, n = 9; Experiment 2, n = 0) were removed from drug testing analyses. Based on a priori hypotheses, animals were removed from the analysis if their baseline training performance (i.e. averaged three days prior to each drug exposure) indicated a %RC ceiling risk (%RC >95%; n = 2) in Experiment 1, or a floor risk (%RC <5%; n = 11) in Experiment 2. Furthermore, animals with data that were >two standard deviation from the mean (extreme responders), were removed from statistical analyses (Experiment 1, n = 2; Experiment 2, n = 2). Statistical analyses were conducted using SPSS v 27 (Chicago, IL), with alpha set at p < 0.05, though trend effects (p < 0.1) were reported where observed.

3. Results

3.1. Females exhibited higher baseline risky decision-making relative to males

During initial training, sex did not impact succession through the first (female: mean = 2.88, SD = 1.2, male mean = 3.13, SD = 1.45) or second phases of task training (female and male mean = 2.0, SD = 0), however females tended to require more sessions to reach nosepoke training criterion [t(30) = 2.0, p = 0.054; data not shown]. Fig. 2A illustrates the number of rats grouped within each %RC category by sex. Most rats fell into the risk-adverse category (<20 %RC score), while females were more risk-preferring than males as measured by elevated %RC [t(18.684) = 2.474, p = 0.023; Fig. 2B]. Females and males did not differ in their mean choice [t(30) = 0.649, p = 0.522; Fig. 2C] or reward [t(24.588) = 1.128, p = 0.270; Fig. 2D] latencies, though females completed fewer trials than males [t(30) = 2.689, p = 0.012; Fig. 2E]. The sex differences in %RC were likely driven by differences in decision-making metrics observed, wherein males exhibited increased %safe-stay trials [t(19.337) = 2.634, p = 0.016; Fig. 2F], as well as decreased % risky-lose stay [t(17.249) = 2.162, p = 0.045; Fig. 2G] and %risky-win stay [t(16.263) = 2.246 p = 0.039; Fig. 2H] trials. This pattern of responding suggested that males prefer the safe option, while females prefer the risky option, regardless of outcome.

Fig. 2. Females exhibited higher %RC relative to males at baseline in the RPT.

The number of rats within each %RC category revealed that females were more risk-prone and less risk-averse than males (A). Overall, females exhibited higher %RC than males (B). Mean choice (C) and reward (D) latencies did not differ by sex. Females completed less trials than males (E). In-terms of decision-making metrics, %safe-stay trials were lower in females (F), while %risk-lose stay trials (G) and % risky-win stay trials (H) were higher in females. Data presented as individual data-points, plus mean ± S.E.M., * = p < 0.05 as indicated.

3.2. PPX treatment increased risky decision-making in rats

A total of 8 females and 13 males were included in the analysis based on a priori test requirements of baseline pre-PPX ceiling risk (%RC > 95%). Within animals that were included in this analysis, an independent samples t-test determined %RC did not differ between male and female animals [t(19) = 1.67,p = 0.13], likely because extreme responders were excluded from analyses.

Consistent with the original RPT study, we collapsed data across drug test sessions and observed that %RC was higher after PPX vs. saline sessions [F(1,17) = 15.817, p < 0.001; Fig. 3A]. As with baseline assessment, females tended to display higher mean %RC than males [F(1,17) = 3.793, p = 0.068; Fig. 3A]. A session * sex interaction [F(1,17) = 5.496, p = 0.031] revealed PPX increased risky choices in females (p < 0.001), but not males (p = 0.202), and females exhibited higher %RC relative to males during PPX (p = 0.030) but not saline (p = 0.272) sessions. A session * sex * dose interaction [F(1,17) = 5.136, p = 0.037] indicated however that high dose PPX (0.3 mg/kg) increased %RC in females (p < 0.001), while low dose PPX (0.15 mg/kg) increased %RC in males, though this effect was not significant (p = 0.14). Also, within the high dose PPX group, females exhibited higher mean %RC compared to males on PPX (p = 0.021) but not saline (p = 0.395) sessions, which likely drove the session * sex effect observed.

Fig. 3. PPX administration increased %RC and altered decision-making metrics.

When %RC was collapsed across test sessions high dose PPX increased %RC in females; in males low dose PPX increased %RC in a manner that did not reach significance (p = 0.14; A). When %RC was analyzed across test sessions, high dose PPX increased %RC on its second and third session (PPX-2 and PPX-3) and there was a trend (p = 0.089) for low dose PPX to also increase %RC on its second session in females; in males, low dose PPX increased %RC on its first session (PPX-1; B). PPX reduced %safe stay trials (C). PPX increased %risky-win stay trials in females, but not males (D). PPX increased %risky-lose stay trials, with a trending dose interaction (p = 0.053) suggesting this was driven by high-dose PPX (E). Data presented as individual data-points, plus mean ± S.E.M.,; *** = p < 0.001, ** = p < 0.01, * = p < 0.05; # = p < 0.1; relative to previous saline treatment day or as indicated, + = p < 0.05 relative to sex saline control.

When data were split and analyzed across the 7 test sessions, an effect of session [F(1.882,32.002) = 7.582, p = 0.002; Fig. 3B] revealed that PPX increased %RC following its second (PPX-2) and third (PPX-3) administration compared to their relative saline control sessions (ps < 0.01) and the first saline session (ps < 0.05; Fig. 3B). However, PPX increased %RC in a manner that interacted with both sex and dose [(F (1.882, 32.002) = 3.773, p = 0.036]. Specifically, 0.3 mg/kg PPX increased %RC following its second (PPX-2; p < 0.001) and third administration (PPX-3; p < 0.05) compared to all other test sessions in females, but not in males (p > 0.1). Within the 0.3 mg/kg groups there were no other significant differences between adjacent test sessions. 0.15 mg/kg PPX increased %RC on its first administration session (PPX-1) compared to its previous saline session (Sal-1) in males (p = 0.015), whereas there was a statistical trend suggesting it increased %RC on its second administration session (PPX-2) compared to its subsequent saline session (Sal-2) in females (p = 0.089). Collapsed data revealed that PPX-induced changes in %RC across test sessions were likely driven by reduced %safe-stay trials (session: [F(1,17) = 5.625, p = 0.03; Fig. 3C]) and increased %risky-win stay (session: [F(1,12) = 17.374, p = 0.001; Fig. 3D]), and %risky-lose stay (session: [F(1,15) = 12.998, p = 0.003; Fig. 3E] trials. A session * sex interaction [F(1,15) = 6.721,p = 0.02] on %risky-lose stay trials revealed PPX only increased this metric in females (p < 0.001), and not males (p = 0.418), and that females had higher % risky-lose stay trials than males during PPX (p = 0.042), but not saline (p = 0.816), sessions. A trending session by dose interaction on %risky-win stay trials [F(1,12) = 4.615,p = 0.053], revealed that high dose (p = 0.002) but not low-dose PPX (p = 0.129), drove the increased %risky-win stay trials on PPX test days.

Analyses of total trials on collapsed data indicated PPX lowered completed trials (session: [F(1,21) = 128.73, p < 0.001; Fig. 4A]). Irrespective of test day, females completed fewer trials than males (sex: [F(1,21) = 10.221, p = 0.004; Fig. 4B]). When data were split and analyzed across the 7 test sessions, the same session [F(3.192,67.028) = 59.810, p < 0.001] and sex effects [F(1,21) = 11.121, p = 0.003] were observed on total trials completed. Irrespective of dose or sex, PPX slowed choice [F(1,17) = 5.823,p = 0.027] and reward [F(1,17) = 27.367,p < 0.001] latencies.

Fig. 4. PPX reduced trials and slowed latencies in the RPT secondary outcome variables.

Consistent with baseline testing, females completed less trials than males overall (sex effect not illustrated; A). When trials completed were analyzed across test sessions, PPX reduced trials completed on each PPX test session relative to saline sessions (B). PPX slowed mean choice latency (C) and mean reward latency (D). Data presented as individual data-points, plus mean ± S.E.M., *** = p < 0.001, *p < 0.05 relative to previous saline treatment day or as indicated.

3.3. SUL treatment reduced risky decision-making in rats

A total of 6 females and 6 males were included in the analysis based a priori test criteria of baseline pre-SUL floor risk (%RC<5%). Within animals that were included in this analysis, an independent samples t- test determined baseline %RC did not differ between male and female animals [t(10) = 1.16,p = 0.27], likely because extreme responders were excluded from analyses. Further, despite PPX-induced %RC alterations in rats, pre-SUL %RC baseline was strongly correlated with pre-PPX % RC baseline, [r(24) = 0.78, p < 0.001].

When data were collapsed across drug test sessions, %RC was lower following SUL compared to saline sessions [F(1,10) = 35.512, p < 0.001; Fig. 5A]. When data were split and analyzed across the 7 test sessions, a trending effect of session [F(2.453,24.531) = 2.747, p = 0.074; Fig. 5B] revealed that SUL decreased %RC following its first (SUL-1) and second (SUL-2) sessions compared to their previous saline sessions (ps < 0.01). Collapsed data revealed the SUL-induced changes in %RC were likely driven by increased %safe-stay trials [session: F(1,10) = 14.815, p = 0.003; Fig. 5C], reduced %risky-lose stay trials (session: [F(1,10) = 15.043,p = 0.003; Fig. 5D]), and reduced %risky-win stay trials particularly in females (p = 0.002), but not males (p = 0.466), as revealed by a session * sex interaction [F(1,10) = 12.841,p = 0.005; Fig. 5E].

Fig. 5. SUL decreased %RC and altered decision-making metrics.

When %RC was collapsed across test sessions SUL decreased %RC (A). When %RC was analyzed across test sessions, SUL decreased %RC on its first and second session (SUL-1 and SUL-2; B). SUL increased %safe stay trials (C). SUL decreased %risky-lose stay trials (D). SUL decreased %risky-win stay trials in females but not males (E). Trials completed did not differ across sessions; females continued to complete less trials than males (sex effect not illustrated; F). Reward latency and choice latency did not differ across sessions (G, H). Data presented as individual data-points, plus mean ± S.E.M.; *** = p < 0.001, ** = p < 0.01, * = p < 0.05; + = difference from respective sex saline control p < 0.05 as indicated.

Collapsed data also indicated SUL did not affect total trials [session: F (1,10) = 0.147, p = 0.709], though a sex effect revealed that again females completed less trials than males [F(1,10) = 7.286, p = 0.022; Fig. 5F]. SUL treatment also did not impact mean choice [session: F(1,10) = 0.289, p = 0.602; Fig. 5G] or reward [session: F(1,10) = 0.275, p = 0.611; Fig. 5H] latencies.

4. Discussion

To our knowledge, these data are the first report on sex-based differences in rats performing the RPT. At baseline, most rats preferred the safe option, consistent with previously reported findings in males (Zalocusky et al., 2016). The percentage of rats choosing the riskier option was twice as many in females relative to males, with some females preferring the riskier option (i.e. %RC > 50%). Furthermore, treatment with the D2-like receptor agonist PPX increased risky choice in rats, as previously demonstrated in males (Zalocusky et al., 2016). The effects we report however, were dose- and sex-dependent. More specifically, although the high and low dose increased risky choice in both sexes, this effect was larger in females. Additionally, we found that treatment with the D2-like receptor antagonist SUL decreased %RC constistently across sex. Together, these results highlight sex-based differences in the RPT and provide further support for the role of D2-like receptors in risk-based decision-making irrespective of sex, as measured by the RPT.

At baseline males preferred the safe option in the RPT, as previously reported. In contrast to previous findings however, here no males exhibited risk preferences >40%. Compared to males, risk preference was twice as high in females with several reaching >40% risk preference categories. Preclinical studies assessing sex differences in risk-based decision-making are scarce and to our knowledge, we are the first to report behavioral differences in male and female rats performing the RPT. These results are similar to the sex differences reported in the cross-species IGT paradigm wherein females select disadvantageous options in a gambling task more often than males (Georgiou et al., 2018; Overman and Pierce, 2013; van den Bos et al., 2012), though not always (Peak et al., 2015). In the IGT, disadvantageous (i.e. riskier) choices are associated with high payouts of money or food reward coupled with high probabilities of losing such rewards. Like the RPT, some rodent IGTs are performed over multiple sessions, therefore baseline sex differences in the RPT seems to be consistent with this cross-species work in a task that provides varying magnitudes of reward and reward loss. Other risk-based paradigms exist however, wherein females exhibit less risk-preference than males, e.g., the risky decision-making task wherein rats have learned that a risky choice is coupled with increasing probability of coupled footshock (Orsini et al., 2016, 2021). Hence, females may take fewer risks when severe punishments contingencies for risky choices have been learned prior to testing. Indeed, female rodents are more sensitive to probabilistic punishment and display faster punishment avoidance learning than males (Chowdhury et al., 2019; Jacobs and Moghaddam, 2020), highlighting the importance of learning and level of punishment. In our RPT study, female rats exhibited decreased sensitivity to losses (i.e. decreased %risky-lose stay trials), at odds with literature that suggests they are more sensitive to losses than males (Cazzell et al., 2012; Cross et al., 2011; Dhingra et al., 2021; Eneva et al., 2017; Ishii et al., 2018; Lee et al., 2009). For example, less risk-taking and increased lose-shift behavior was observed in water deprived female rats compared to males in an operant-based task where animals could lose the chance to earn water reinforcement (Ishii et al., 2018). Similarly, brain imaging studies report increased brain activation in females relative to males in response to punishment feedback during risk-based decision-making tasks (Cazzell et al., 2012; Lee et al., 2009). It appears these contrasts are likely a result of task design, as the RPT includes no explicit loss associated with the risky choice, only lesser reward. Thus, in the absence of explicit loss or severe punishment, as in the RPT, females may be more risk-prone. Irrespective of methodological drivers, these data further highlight sex differences in risk-based decision-making and support the importance of considering sex as a factor in future studies utilizing the RPT.

The current findings also provide reproducibility of prior findings Zalocusky et al. (2016) as we describe that D2-like receptor activation (via PPX treatment) increased risky choice in male rats performing the RPT. Furthermore, we extended these findings by demonstrating that PPX also increased risky preference in female rats. Decision-making metrics revealed that these PPX-induced increases in risk preference were driven by reduced sensitivity to losses (i.e. increased %risky-lose stay trials), and smaller risk rewards (i.e. decreased %safe stay trials), in addition to increased sensitivity to higher risk rewards (i.e. increased % risky-win stay trials). There were notable differences between our findings and that of Zalocusky et al. (2016), however. For one, neither dose of PPX affected %RC in males when data was collapsed by drug across test sessions. This discordance was likely because low-dose PPX only increased risky choice in males following its first, but not second or third, administration, potentially suggestive of tolerance effects. Instead, high-dose PPX increased %RC in females specifically after its second and third administration. This delayed risk-taking profile development over multiple exposures with PPX is consistent with chronic PPX treatment inducing gambling behavior in human populations (Dodd et al., 2005). The important role of the D2-like receptor family in RPT performance was further supported by our novel findings that decreased risky choice preference was observed following blockade of the D2-like receptors via SUL, an effect that was consistent across sex. Decision-making metrics suggested the effects of SUL were driven by increased sensitivity to losses (i.e. decreased %risky-lose stay trials), and reduced sensitivity to higher risk rewards (i.e. decreased %risky-win stay trials). Since drug order was consistent across all animals, it should be acknowledged that preexposure to PPX may have altered the sensitivity of animals to the effects of SUL on %RC behavior. To address this, a large wash-out period (15 days, including 8 days of behavioral baseline training) was employed between the two drug exposures, however, future work should aim to replicate the observed effects without such drug-order confound. Nonetheless, our data support the importance of D2R functioning on risk-taking decision-making across the sexes.

The loci of these pharmacological effects likely includes the NAcc given this region’s heavy involvement in risk-based decision-making processes and dense expression of D2Rs (Cardinal and Howes, 2005; Floresco et al., 2018; Le Moine and Bloch, 1995; Lee et al., 2004; Mai et al., 2015). Previous results also indicated D2, but not D1, NAcc signaling modulates risk preference in male rats performing the RPT (Zalocusky et al., 2016). These findings however were only in male subjects, thus necessitating confirmation in female animals – especially given our data suggesting sex-dependent behavioral effects in this task and differential D2-like receptor activation sensitivity. The NAcc may also be the region for the observed sex differences since gonadal hormones modulate DA dynamics in this region (Landry et al., 2002; Le Saux et al., 2006; Purves-Tyson et al., 2014). Thus, while these data support the importance of the D2-like receptors for performance in the RPT and its associated sex-differences, further studies are required.

Our observed results are highly novel in their contributions to the literature of dopaminergic and sex-dependent modulations of risk taking in the RPT. First, female animals are sparsely used to assess the effects of sex and dopamine on risk-taking behavior. To our knowledge, only one study has tested the interaction of these factors, which revealed sex-dependent drug effects on risk-taking behavior, highlighting the importance of such work (Georgiou et al., 2018). More specifically, in the rodent gambling task (rGT) treatment with a D2 antagonist (eticlopride) increased risky choices selectively in males, whereas a D2 agonist (quinpirole) increased risky choices in females only. Previous research from this group however, revealed that the D2 antagonist (eticlopride) decreased risky choices in male rGT performance (Zeeb et al., 2009). Thus, even within task studies, there are contrasting results within sex. Our data reveal that in the RPT the D2 agonist PPX increased risky choices across the sexes (as seen in males in the original RPT study and in humans), and that the D2 antagonist decreased risky choices, again across the sexes. As previously highlighted, these discrepancies may be the result of punishments (time-out periods), associated with the risky options in the rGT, not included in the RPT. The Risky Decision-Making Task utilizes physical footshock as punishment paired with risky choices and D2 agonist (bromocriptine) treatment decreased risk-taking behavior in male rats (Simon et al., 2011), in contrast to the rGT and RPT studies. Our results however, do correspond to drug-induced increases and decreases of the large/risky choice options in male rats in the probability discounting paradigm, by D2 agonists and antagonists, respectively which also do not incur punishments (long time-outs or footshocks) (St Onge and Floresco, 2009). Hence, data in decision-making paradigms that do not include overt punishments appear to reproduce D2-treatment effects, at least in males, with consistent directionality of effects in females as we showed in the RPT.It should be noted that the RPT does not measure risk-based decision--making in a manner similar to other clinical or preclinical models of risk-taking behavior. For one, animals never experience explicit loss or punishment following a risky choice in the RPT. We raise the possibility that this is a strength of this task, as many real world risk-based decisions may lead to less reward rather than loss or punishment. In fact, many real world risk-based decisions seldom lead to explicit punishment, limiting the value of risk-based decision making tasks that include such factors, particularly those utilizing physical punishments. Further, since averaged reward magnitudes across choices are identical in the RPT, this raises the possibility that choice of the risky option by animals may instead reflect a preference for variable outcomes, rather than risk per se. To futher tease apart this meaning of “risk choice” in the RPT, we suggest future studies test animals concurrently in the RPT and other risk-based decision-making tasks, such as the IGT, to determine if risk performance correlates across tasks. While more work is needed to further elucidate the implications of risk-taking in this task, the RPT provides particular research value in its ability to parse apart behavior controlled by sensitivity to reward values that are not influenced by punishment or explicit loss outcomes, currently unavailable in other tasks of risk-based decision making.

The results of this study must be considered alongside its experimental limitations. For one, the loci of pharmacological effects were not determined and while likely mediated by the NAcc, future work must verify the importance of this region in female rats. Though the importance of D2, but not D1, receptor modulation on RPT task performance has been observed in male rats, no study has tested this in female rats and should in future work. Discrepancies between observed findings and results reported by Zalocusky et al. (see above; 2016) were not likely due to small task differences between laboratories (i.e. levers vs. nosepoke and sucrose vs. strawberry milkshake). However, PPX significantly decreased trial count across testing, with some animals ceasing to perform the task, though it is unclear whether a similar reduction in completed trials was also evident in the previous study by Zalocusky et al. (2016). We therefore removed all animals with less than 20 trials completed, as to preclude disproportionately calculated %RC outcomes, and had lower sample sizes as a result. The previous report by Zalocusky et al. (2016) did not discuss PPX effects on trial count nor whether low trial count animals were included in analyses. Thus, slight analytical differences, and inclusion criteria for testing, may have contributed to the minor discrepancies observed across studies. Importantly, however, the main outcomes of our study were consistent with those reported by Zalocusky et al. (2016). Additionally, based on exclusion criteria constraints different sets of rats were analyzed in the PPX and SUL experiments (though 6 overlapped), which raises interpretational challenges when comparing the effects of both drugs across Experiments. It is therefore possible that the opposing effects of PPX and SUL on %RC behavior only occurred in a subset distribution of our rat population. When the 6 overlapping rats were examined however, PPX and SUL moved %RC in the same direction (i.e., increased and decreased %RC, respectively), making this interpretation unlikely.

5. Conclusions

Together, both our and prior studies highlight the modulatory role of the D2-like receptor in risk-based decision-making as measured by the RPT. More specifically, we replicated previous evidence that PPX increased risk preference in the RPT in male rats, and extended these finding to females which demonstrated more pronounced effects. Moreover, we extended these findings by demonstrating that pharmacological blockade of D2-like receptor activity (via SUL) decreased risk choice in the RPT. While the RPT has not been extensively utilized, it provides opportunities to assess risk-based decision-making under conditions of no absolute loss or punishment. Our results suggest that under such conditions, female rats may be more risk-preferring, and differentially respond to the effects of D2-like receptor modulation when compared to males in the RPT. These results strongly encourage the use of both sexes in further studies assessing dopaminergic modulation of risk behavior as well as those utilizing the RPT paradigm.

Data availability

Data will be made available on request.