Investigating hormone-induced changes in affective state using the affective bias test in male and female rats.

Recent clinical and pre-clinical research suggests that affective biases may play an important role in the development and perpetuation of mood disorders. Studies in animals have also revealed that similar neuropsychological processes can be measured in non-human species using behavioural assays designed to measure biases in learning and memory or decision-making. Given the proposed links between hormones and mood, we used the affective bias test to investigate the effects of different hormone treatments in both male and female rats. Animals were pre-treated with acute doses of hormone or vehicle control prior to learning each of two independent substrate-reward associations. During a subsequent choice test, positive or negative biases were observed by animal’s preference towards or away from the substrate learnt during drug treatment respectively. In both sexes, oestradiol and the oestrogen-like compound bisphenol A induced positive biases, whilst blockade of oestrogen hormones with formestane induced a negative bias. Progesterone induced a negative bias in both sexes, but testosterone only induced a negative bias in males. Blocking testosterone with flutamide induced a positive bias in both sexes at the higher dose (10mg/kg). The oxytocin analogue, carbetocin induced positive biases in both sexes but the vasopressin analogue, desmopressin, induced a positive bias in male rats only. These results provide evidence that modulating levels of hormones using exogenous treatments can induce affective biases in rats. They also suggest that hormone-induced affective biases influence cognitive and emotional behaviour and could have longer-term effects in some mood disorders.


INTRODUCTION
Recent findings suggest that affective biases (the process whereby cognitive functions such as learning and memory and decision-making are modified by emotional state), may play an important role in the cause and treatment of depression (Harmer et al., 2017;Robinson, 2018).
It is hypothesised that negative affective biases contribute to the reinforcement of negative feelings and beliefs in depression (Disner et al., 2011). Antidepressants have been shown to induce positive affective biases which is thought to influence the reversal of symptoms of depression over time (Harmer et al., 2017). Studies in both human and non-human species have shown that affective biases can influence many different cognitive domains (Roiser et al., 2012;Robinson, 2018). These include findings that people with anxiety and/or depression tend to exhibit biases in attention, memory and decision-making particularly in relation to ambiguous Since the first report of an affective bias task for rats (Harding et al., 2004), two types of behavioural assay for non-human species have been developed. The judgement bias task (also called the ambiguous cue interpretation task) is designed to quantify affective biases in a twochoice decision-making task (for review see, Hales et al., 2014). Animals are first trained to associate two distinct cues with positively or negatively-valenced outcomes. Biases in decision-making are then quantified by presenting the animals with intermediate, ambiguous cues and recording their responses. More positive affective states are associated with a greater number of responses in anticipation of positive outcomes during ambiguous cue presentation.
Studies using pharmacological and phenotypic models suggest that these biases are generally consistent with predicted changes in the underlying affective state, although there are exceptions (for review and more detailed discussion see, Hales et al., 2014). The affective bias test (ABT), first reported in 2013, is designed to quantify biases linked to learning and memory (for review see, Hales et al., 2014;Robinson, 2018). The assay uses a within-subject study design where animals encounter two distinct and independent learning experiences under treatment or control conditions. The value of the experiences is kept constant and, using a choice test, biases resulting from the treatment can be quantified. Treatments inducing a positive bias result in a preference for the treatment-paired experience, whilst negative biases are observed as a preference for the control paired experience. The ABT has been extensively validated using pharmacological, psychosocial and neurobiological studies with results consistent with predicted effects on affective state (Stuart et al., 2013(Stuart et al., , 2015Hinchcliffe et al., 2017;Stuart et al., 2017).
In this study, we used the ABT to investigate the effects of the gonadal hormones, oestrogen, progesterone and testosterone, and peptide hormones, oxytocin and vasopressin on biases in learning and memory in both male and female rats. The gonadal hormones and the peptide hormones have been linked to changes in emotional processing and mood (Swaab et (Buddenberg et al., 2009). On the other hand, high levels of testosterone increase aggression behaviours in men and in male rats (Finkelstein et al., 1997;Rosell and Siever, 2015). Testosterone can also modulate the stress response, although with different outcomes reported. For example, some studies report increased responsiveness (Mehta et al., 2008), whilst others suggest a reduction in the stress response (Stephens et al., 2016). Although previous studies have shown links between hormone levels and emotional behaviour, their effects on affective biases is unknown. There have also been issues in the interpretation of previous studies in rodent models of depression given the limitations associated with these approaches (Slattery and Cryan, 2017). In particular, conventional methods of assessing emotional behaviour in rodents have limited translational validity and generally rely on stress related methods. Furthermore, there is a lack of consistency in terms of the depression-like phenotype which develops in males versus females which do not necessarily reflect sex differences in clinical populations (Kokras and Dalla, 2014). Thus, to further understand the effects of the gonadal hormones on affective biases, we tested the both androgen hormones and androgen receptor antagonists, flutamide, and aromatase inhibitor, formestane. There is also a detailed literature investigating the role of oxytocin and vasopressin and how they contribute to emotional behaviour (Neumann and Landgraf, 2012). It has previously been reported that oxytocin has antidepressants effects in the forced swim test (FST) (Balmus et al., 2018). In   Table   S1). The experimental training and testing for ABT were similar to those previously described containing sucrose, casein, maltodextrin, corn starch, corn oil, minerals, vitamins, magnesium stearate, DL-methionine). Once each animal was able to find the food pellet within 30 s on 10 consecutive trials, the digging training was complete. On the last day of training, animals underwent a discrimination session allowing them to explore two bowls with two novel digging substrates (reward-paired substrate, CS+, with single pellet versus unrewarded substrate CS-).
A reward pellet was crushed into the bowl and mixed within the unrewarded substrate, to avoid choices based on odour. On each trial, the animal was individually placed in front of the two bowls. Once the animal made a choice and started digging in one bowl, the other bowl was removed by the experimenter. Choice of the reward-paired substrate was marked as a 'correct' trial, digging in the CS-substrate was classified as an 'incorrect' trial and if an animal failed to approach and explore the bowls within 30 s, the trial was recorded as an 'omission'. Trials were continued until the rat achieved six consecutive correct choices for the reward-paired substrate. All animals completed training and were included in the subsequent studies (n = 12 per cohort).

Hormone dose-response studies
All studies were based on four pairing sessions (one per day) followed by a choice test on the fifth day of that week. Each pairing session followed the same procedures as the discrimination session detailed above. During four pairing sessions, each animal learnt to associate two different digging substrates with acquiring a food reward under vehicle or treatment conditions. Each trial involved a choice between two bowls containing two different digging substrates, Affective biases were quantified during the choice test on day 5 in which the two previously rewarded substrates ('CS + A' and CS+'B') were presented at the same time for 30 trials. Trials were reinforced using a random schedule with a single-pellet reward baited in either bowl with a probability of one in three. Both bowls contained a crushed pellet to reduce the likelihood of the animal using odour to find the reward. The animals' choices and latency to dig were recorded.

Consumption test
To assess whether gonadal hormone treatments have any effects on appetite and food intake, we used a consumption test where the quantity of food (reward pellets, 45 mg pellets, Test Diet, UK) consumed by an animal within 10 min was measured. To match experimental conditions with the ABT, animals subjected to this consumption test were mildly food restricted and hormones were injected 30 min prior to testing. The consumption test was carried out in the ABT arena with one pottery bowl (Ø 5 cm). The study took place over four nonconsecutive days using a fully counterbalanced design with the dose of each hormone that induced the largest affective bias being administered on each day followed by a food low doses of desmopressin has been found to activate the HPA axis and lead to cortisol release in human participants suggesting it crosses the blood brain barrier (Scott et al., 1999). Vehicle solution for oestradiol, progesterone, testosterone and formestane was 5% DMSO and 95 % sesame oil; for flutamide and bisphenol A was 1% ethanol in strawberry milkshake, and for carbetocin and desmopressin was saline. Prior to the start of the experiments, rats were trained to drink milkshake (Frijj, UK, 0.5 ml) from a 1 ml syringe to facilitate oral drug dosing. On each day of treatment, drugs and vehicle solutions were freshly prepared. All studies used a within-subject design and there was a minimum of 7 days drug free before commencing a new treatment. All subcutaneous injections were performed with minor animal restraint and injected on their left or right flank (changing daily) to minimise the stress associated with restraint. All experiments were carried out with the experimenter blind to treatment.

DATA ANALYSIS
Data were analysed and the graphs were created using GraphPad Prism 6.0 (GraphPad Software, USA). Choice bias was calculated as the number of choices made for the treatmentpaired substrate divided by the total number of trials (treatment-paired substrate + vehiclepaired substrate) multiplied by 100 to give a percentage value. A value of 50 was then subtracted to give a % choice bias score where a bias towards the treatment-paired substrate gave a positive value and a bias towards the vehicle-paired substrate gave a negative value.
The % choice bias results from the dose response studies were analysed using a repeated measures ANOVA with dose (oestradiol, formestane, bisphenol A, progesterone, testosterone and flutamide study) or treatment (carbetocin and desmopressin study) as the within-subject factor, and one-sample t-test against the null hypothesised mean of 0% choice bias as post-hoc tests. A Shapiro-Wilk test was used to determine a normal distribution, the Huynh-Feldt correction was used to adjust for violations of the sphericity assumption, and Levene's test was used to correct for inequality of variances for the % Choice bias. A repeated measures ANOVA with treatment as the within-subject factor was used to analyse the results from consumption test. Analysis of the trials to criterion and response latency utilised a paired t-test, comparing vehicle vs treatment for each animal during the pairing sessions.

DISCUSSION
Our studies have shown that acute hormone treatments can induce affective biases in male and female rats. The effects, in terms of induction of a positive or negative bias, were similar for both sexes but with some exceptions. In both sexes, treatment with oestradiol and bisphenol A and the oxytocin analogue, carbetocin, induced a positive bias. The vasopressin analogue, desmopressin, induced a positive bias but only in male rats. The lower dose of progesterone induced a negative bias in male rats, but only the higher dose was significantly effective in female rats. Testosterone treatment induced a negative bias in male rats but had no effect in females. When the effects of the gonadal hormones were blocked, the opposite effects were To further investigate the effects of endogenous oestradiol we used, formestane, a selective aromatase inhibitor. Acute administration of formestane induced a dose-dependent negative bias, the opposite effect to oestradiol treatment suggesting inhibition of endogenous oestradiol has negative effects on emotional behaviour in both sexes. It previously has been shown that formestane inhibits oestrogen synthesis in the brain, and by inhibiting the conversion of androgens into oestrogens (Dowsett, 1994). Acute formestane treatment lacks effects in FST Formestane was used in the treatment of breast cancer and these data suggest that long term use may have caused mood related side effects (Wiseman and Goa, 1996). In the ABT, we observed no effect with testosterone treatment in female rats but a negative bias following acute dosing in males. These findings are opposite to those previously reported for the forced swim test where an antidepressant like effect was observed (Buddenberg et al., 2009). Castrated male rats were also observed to have more depression-like behaviours in this assay suggesting low testosterone levels may be pro-depressant however, there are limitations associated with the forced swim test (McHenry et al., 2014). The antidepressant-like effects of testosterone have been linked to its conversions into oestradiol, as treatment with dihydrotestosterone, a testosterone metabolite that cannot be aromatised to oestradiol, did not induce the same effect (McHenry et al., 2014). However, we previously observed positive biases with oestradiol in the ABT in male rats and therefore, this does not appear to be a mechanism contributing to the effects observed in this assay. We also observed that blocking the effects of testosterone induced a positive bias suggesting these effects were hormone specific. Again, these effects are not generally the same as has previously been reported although aspects of the experimental protocols do differ.  (Kis et al., 2015). A study in humans has also shown that oxytocin can reduce cortisol levels in humans during stress conditions (Cardoso et al., 2013).
We also observed positive biases in male but not female rats, when animals were treated with desmopressin. These effects are contrary to previous findings, where anxiogenic effects have been reported (Mak et al., 2012) and treatment with vasopressin antagonists have had antidepressant effects (Iijima et al., 2014). Desmopressin interacts with AVP1 receptors, which would be expected to increase aggressive behaviour and cause psychosocial stress and negative biases. However, we used a very low dose of desmopressin because of its powerful antidiuretic properties. If it could be done safely (e.g. avoiding water retention, low blood sodium and seizures, Verbalis, 1993), use of a higher dose would be desirable to investigate dose dependency of any induced biases. It would also be interesting to test the vasopressin receptor 1 antagonists, which have previously been found to have anxiolytic and antidepressant effects and are selective for the AVP1 receptor thus reducing effects associated with fluid homeostasis.
The ABT is an appetitive task, however, we have previously reported effects for a wide range of compounds which are in a direction that does not consistently relate to changes in motivation The results from our voluntary consumption test did not find any effects of oestrogen, progesterone or testosterone on appetite when compared to the vehicle group although there was some variation in the data and differences between vehicle and progesterone in absolute consumption per body weight female rats. This finding seems to be the opposite of that observed in previous preclinical studies. Authors have demonstrated that acute progesterone treatment (10 mg/kg) causes an increase in appetite in female mice (Kaur and Kulkarni, 2002). There was also no effect on latencies during the task consistent with a lack of effect on motivation or general locomotor function.
In summary, this study has shown that acute manipulation of gonadal hormone levels using exogenous administration or treatment with antagonists can induce biases in reward learning and memory in the ABT. Both male and female rats were similarly affected by the peptide hormones and female gonadal hormones but with differences in their response to testosterone.
There is growing interest in the relationship between affective biases and mood disorders and these data suggest that hormones may also be an important modulator of these biases (Hales et    . Data shown as mean % choice bias ± SEM, *p < 0.05, **p < 0.01, ***p < 0.001, one sample t-test against a null hypothesised mean of 0% choice bias.