An Olfactory Preference Test for Measuring Olfactory Hedonic Biases in Mouse Models of Depression

Summary

Here, we present a protocol of an olfactory preference test that allows for the assessment of negative olfactory biases towards both appetitive and aversive odor stimuli in mouse models of depression.

Abstract

Depression is the single largest contributor to disability worldwide, affecting 280 million people annually. Beyond reduced activity and motivation, depression has been shown to be associated with a negative bias of valence: patients find neutral and positive sensory stimuli less pleasant and negative ones more unpleasant compared to healthy individuals. In particular, olfactory hedonic bias may represent a state marker of the disease in patients and can be easily measured in mouse models. However, there are very few tests in animals to measure emotional bias in a translational way. Here we propose a protocol for a reliable evaluation of the odor valence attribution of control and depressive-like mice through the exploration of odorants associated with various hedonic values. Using this olfactory preference test and several behavioral readouts, we confirm a negative olfactory bias towards both appetitive and aversive stimuli in two distinct mouse models of depression. Importantly, this method is applicable to both male and female individuals. Finally, this protocol allows for multiple adaptations to improve the evaluation of the emotional response and can be easily combined with classical tools for live monitoring of neuronal activity.

Introduction

Depression is a serious mood disorder and a leading cause of disability worldwide. According to the World Health Organization, approximately 280 million people, i.e., 4% of the population, experience depression. It is 50% more common among women than among men, and the outbreak of COVID-19 increased by 25% the prevalence of depressive disorders in the first year of the pandemic (World Health Organization, 2022). In addition, about one-third of patients do not respond to classic antidepressant therapies, it takes weeks to observe antidepressant effects, and the physiopathology of mood disorders remains largely unknown.

Based on the DSM-5¹, a diagnosis of depression relies on several symptoms, including depressed mood, loss of interest or pleasure, significant weight variation, a physical and psychological slowing down, fatigue, feelings of worthlessness, difficulty concentrating, and recurrent thoughts of death. Clinical studies also demonstrated that mood disorders come along with sensory alterations: depressed patients find sensory stimuli less pleasant than control subjects^2,3,4.

Hedonic "valence" refers to the degree to which something is pleasurable (positive valence) or aversive (negative valence). Valence assignment to a stimulus can trigger an adapted behavior, including approach or avoidance⁵. Interestingly, this process is highly conserved between species from insects to humans⁶. Depression is thus associated with a negative bias of valence. However, clinical diagnosis relies little on sensory bias assessments yet.

Among sensory modalities, olfaction is particularly interesting as olfactory inputs largely bypass thalamic relays and project to the limbic system in mammals, including humans. From an evolutionary point of view, this system has been shown to be critical in coding for a set of primary responses to environmental stimuli^{7,8,9,10,11,12}. On top of these shared neural circuits, clinical studies reveal a bilateral relationship between olfaction and depression, with depressed patients experiencing olfactory alterations as well as olfactory disorders causing depressive symptoms¹³. In addition, olfactory function is recovered in depressed patients after medication or therapy^14,15,16,17. These results suggest that olfactory deficits might be a biomarker of state (status of clinical manifestations in patients) rather than trait (properties that play an antecedent, possibly causal, role in the pathophysiology of the psychiatric disorder and can thus be used for early, pre-symptomatic diagnosis) for the disease¹⁸. Accordingly, a better and further understanding of how the olfactory modality is altered in depression could pave the way towards new therapeutic tools for diagnosis and detection of treatment responsiveness.

Therefore, assessing olfactory hedonic bias in models of depression would be of particular interest, both to evaluate the induced depressive-like phenotype and as a tool to investigate neural mechanisms behind these changes in valence assignment. Previous studies have already used behavioral approaches to evaluate odor valence assignment in mice. However, these tests present several caveats: they require sophisticated devices¹⁹ that are mostly involving boxes closed by a lid^19,20,21, implicate constant physical contact with odor solution^20,21,22, and/or result in very short odor exploration times^21,22,23. This makes these assays incompatible with classical tools of live-monitoring of neuronal activity that require longer sampling durations in tethered mice.

Moreover, Malkesman et al. already developed the Female Urine Sniffing Test (FUST) to monitor reward-seeking activity in rodents by measuring olfactory investigation of male mice exposed to female mouse urine²⁴. However, this test only evaluates the behavioral response to one particular positive stimulus of social nature, and with short interaction times. In addition, this test was designed for male rodents exclusively, although depression is twice as prevalent among women as it is among men.

To overcome all these limitations, we developed here an Olfactory Preference Test (OPT) adapted from Bigot et al.¹⁴, where both male and female mice can be exposed to a large variety of neutral, positive, and negative odors in order to precisely assess hedonic response in the context of depression. The choice of odors can be based on previous publications^14,25, but this test also allows for exploring new social, nutritional, infection-related stimuli, using both natural components as well as monomolecular or specifically designed chemical mixtures.

Protocol

All animal care and experimental procedures followed national and European (2010/63/EU) guidelines and were approved by the French Ministry of Research (APAFiS: #16380-2018080217358599_v1; APAFIS #51636-2024101718013761_v4). C57BL/6JRj male and female mice (UCMS model) and C57BL/6NTac (CORT model) male mice (8 weeks old). The choice of the C57BL/6 substrain was based on the relevant literature²⁷. Importantly, each model was compared against their own control from the same substrain. All mice were socially housed (4-5 mice per cage) and maintained under standard conditions (23 ± 1 °C; humidity 40%) on a 14/10 h light/dark cycle (lights on at 7 a.m.) with food and water ad libitum.

1. The experimental setup

1. Installation of the arena
   1. Prepare the experimental setup, which consists of a two-compartment arena (45 cm × 50 cm × 25 cm) made of grey plexiglas walls and a white floor. Divide the two compartments (25 cm × 45 cm) by a grey plexiglas wall (37 cm × 25 cm).
   2. Place a perforated Petri dish (60 mm × 15 mm with 7 holes of 5 mm diameter) on the bottom of the wall from each compartment.
   3. Vertically fixate the Petri dishes on the wall using dual-sided adhesive or pressure-sensitive adhesive (e.g., Blu-tack).
   4. Place a camera on top of the arena to record the animals' behavior (2 m above the arena) and automatically track their movements through the experiment.
2. Choice of odors
   1. Prior to the experiment, choose a list of odors with neutral, positive, and negative valence that will be tested (see Table 1 for examples).
   2. Check that all chosen odors have similar vapor pressure. Additionally, evaluate the diffusion throughout the arena for some odors with a photoionization detector (PID)²⁶.
   3. Test each odor on a separate day.
      NOTE: It is recommended to plan the use of neutral odors first, then appetitive ones, and ending with the aversive odors to minimize exploration effects from one day to another. This order minimizes carry-over effects from one test day to another. Alternatively, a random presentation could be used, but all experimental groups must be tested with the same odor sequence.
3. Design of experimental groups
   1. Divide mice into at least two experimental groups: control mice and mice that will undergo the model of depression. House the animals from the same group in the same cages.
      CAUTION: Each experimental group should contain at least approximately 7-8 individuals to obtain statistical significance. Cages should be localized nearby in the animal house to minimize the cage effect.
   2. Induce a depressive-like phenotype in relevant mouse groups using the model of depression of choice.
      ​NOTE: Describing protocols to induce depression-like behaviors in mice is beyond the scope of this article. However, this study presents experiments using two different models (the Unpredictable Chronic Mild Stress [UCMS] and Corticosterone [CORT])^27,28 as representative results for this olfactory preference test. Shortly, the UCMS model relies on the exposition to a variety of different environmental stressors over several weeks, while the CORT model consists of chronically administering corticosterone dissolved in the drinking water of mice (about 5 mg/kg/day depending on how much mice drink).

2. Habituation phase

1. General habituation and handling
   1. Upon arrival, leave mice undisturbed for 1 week without any bedding change. Add supplementary cotton material and an igloo in the cage to reduce stress.
   2. After 1 week, gently handle all mice daily for 3-4 days, and habituate them to cup handling, preferably in the experimental room.
2. Experimental conditions of the test
   1. Transfer the mouse to the behavioral testing room 30 min prior to the start of the experiment.
      NOTE: Ideally, the behavioral room should include two separate parts allowing animals to be in one part while the tested mouse performs in the other.
   2. Beware that standard housing conditions are maintained (23 ± 1 °C; humidity 40%), and there is no disturbing sound, lighting, or odor in the room.
   3. Using a luxmeter, set up the lighting to approximately 40 lux in the arena. Make sure that illumination is homogeneous in the arena.
3. Habituation to the arena
   1. Add a clean filter paper (42.5 mm) into both Petri dishes (60 mm × 15 mm).
   2. Use transparent tape to attach the lid of the Petri dish to its bottom (because it will be placed vertically).
   3. On the first day, gently place all mice from the same housing cage together into the arena for 10 min to reduce neophobia.
      ​CAUTION: Always place mice into the arena by cup handling to avoid stress due to picking up by the tail.
   4. On the next day, introduce every mouse individually into the arena, just in between the two compartments, and record the behavior for 10 min.
      ​CAUTION: Make sure to thoroughly clean the arena and the lower part of the walls with a disinfectant solution (e.g. ethanol 70%) to avoid any lingering scents in-between individuals, and especially between male and female mice.
   5. Repeat for 2-3 days in a row.
4. Exploration criterion
   1. Define the two areas (25 cm 15 cm) in which the Petri dishes have been placed and measure the time each animal spends in each zone.
   2. Average these measures through all individual habituation steps.
   3. If the average time spent in each of the zones is higher than 50 s, the criterion is fulfilled. If not, repeat habituation sessions until reaching the criterion, or exclude the animal from the analysis.
      ​NOTE: A maximum of two habituation sessions could be performed per day (morning and afternoon).

3. Testing phase

1. Preparation of the odor
   1. On the day of the test, prepare the odor with the correct dilution under a chemical hood. Dilute the odor in mineral oil or distilled water, depending on the chemical structure. To avoid contamination, store mineral oil in a separate ventilated room away from odorants.
      NOTE: The use of a new bottle of mineral oil is recommended when starting a new experiment.
   2. Store the odorant solution in a glass vial. Avoid plastic tube containers that could add odor contamination.
   3. Right before each mouse is tested, collect 200 µL of the odor solution with a micropipette using filter tips, and place it on the filter paper in one of the two Petri dishes.
   4. Let the other filter paper in the second Petri dish remain untouched.
   5. Transfer the Petri dishes in the testing room and fixate them vertically onto the wall of the arena.
      CAUTION: Do not prepare and throw away the odorant solutions in the experimental room to avoid odor contamination. Change gloves regularly during the experiment. If it is suspected that the gloves might be contaminated during odor manipulation, change them before grabbing the mice for testing.
2. Behavioral test
   1. Place each mouse individually into the arena, just between the two compartments.
   2. Record for 10 min.
   3. Remove the mouse from the arena.
   4. Remove the odor Petri dish.
   5. Clean the arena with a disinfectant solution (e.g., ethanol 70%) and dry it. Make sure to also wipe against the lower part of the chamber walls.
   6. Throw away the filter paper and repeat odor preparation for the next mouse.
   7. Ideally, test only one odor per day. Preferentially use neutral odors first, then positive odors, and finally negative odors (see Figure 1).
      CAUTION: Conduct both habituation and testing steps approximately at the same time of the day through the days of the experiment (ideally between 9 a.m. and 3 p.m.). If the number of animals requires a longer period of time, be attentive to test different experimental groups in an alternating manner to avoid any day-time bias. If males and females are needed to be tested in the same experiment, prefer testing one sex after the other.
      NOTE: Use clean Petri dishes every day of the test to avoid odor contamination. Of note, perforated Petri dishes can be reused for several experiments after proper cleaning and odor decontamination. The control filter paper can also be renewed from one day to another if necessary.

4. Data analysis

1. Behavior tracking
   1. Use an automatic tracking program to record the position of the mouse through the tests.
   2. Define separate zones in the arena for analysis (see Figure 1):
      1. Define an "odor zone" (25 cm × 15 cm), the area where the Petri dish is placed with the filter paper soaked with the odor (i.e., the area closest to the odor representing about one-third of the surface compartment).
      2. Define a "control zone" (25 cm × 15 cm), symmetrical to the "odor zone", but without any odor added in the petri dish.
      3. Define a "behind odor zone" (25 cm × 30 cm), i.e., the rest of the compartment is where the odor petri dish is placed.
      4. Define a "behind control zone" (25 cm × 30 cm), symmetrical to the "behind control zone", consisting of the rest of the non-odor compartment.
         NOTE: Tracking of the center point of the mouse body is sufficient to analyze the olfactory preference of the animal. However, tracking the nose of each mouse specifically can illustrate more precisely the olfactory investigation by the animal.
2. Readouts of the hedonic response
   1. Using the automatic tracking, collect the following information for every mouse and every day of the test, including habituation steps:
      1. Record the total distance moved in the arena (m).
      2. Record the time spent in each zone of the arena (s).
      3. Record the number of entries in each zone of the arena.
   2. Use the total distance traveled to assess the main locomotor effects.
   3. To assess differences in olfactory preference towards various odors, use the main readouts:
      1. Use the time spent in the odor zone.
      2. Use the number of entries in the odor zone.
      3. Use a preference index, calculated as the time spent in the odor zone normalized to the time spent in this zone during habituation (normalization is performed on mice from all groups). The mean refers to the arithmetic mean of the time spent in the odor zone, and sd refers to its standard deviation.
         
      4. Use an odor exploration index, calculated as the average of the normalized values of distance moved, time spent in the odor zone, and number of entries in the odor zone as follow:
         
         
         ​NOTE: This preference index can be adapted if necessary, for example, by comparing times spent in the odor versus control zone. If groups show significant differences in their exploration of the "odor zone" during habituation, normalization can be performed with respect to the control group only, rather than to all groups combined.
   4. Visually represent these readouts for control and depressive-like mice as mean ± standard error of the mean (SEM).
   5. Use repeated measures two-way analysis of variance (ANOVA) followed by post hoc tests (here uncorrected Fisher's Least Significant Difference test) to account for both experimental groups, odors (repeated measure factor), and interaction effect factors. The interaction factor indicates whether differences between the groups are consistent across all presented odors.
   6. Verify normality and Homoscedasticity using adapted tests (Shapiro-Wilk test, Bartlett test). Statistical significance was set at *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Results

To verify that depressive-like mice present a negative olfactory bias^14,29 and that it can be reliably measured using the protocol of the olfactory preference test that we describe here, we assessed the hedonic response to different odors in two distinct mouse models of depression.

First, we used a mouse model of unpredictable chronic mild stress (UCMS) lasting 4 weeks adapted from²⁸ (see Supplementary File 1 for detailed protocol) to induce a depressive-like phenotype in young female C57BL/6JRj mice (Figure 2A,B). These female mice were tested in the olfactory preference test with male mouse urine (MU, pure) as an appetitive odor, and 2,4,5-trimethylthiazole (TMT, 5% in mineral oil) as an aversive odor. On average, every mouse spent at least 50 s in both the "control" and "odor" zone during habituation, fulfilling the criterion (Figure 2C). However, through habituation, UCMS mice spent significantly more time in the "control zone" compared to Controls, but as there is no significant difference regarding the zone where the odor will be added, this difference can be dismissed. Regarding the distance, Control mice moved significantly more than UCMS when exposed to MU, but not during habituation or in the presence of TMT (Figure 2D). Finally, all readouts of hedonic response (time spent in the odor zone, number of entries in the odor zone, and preference index) show significant differences or a statistical tendency between Control and UCMS females (Figure 2E-G). This effect is also visible through the global odor exploration index, which is a combined measure of hedonic and locomotor response (Figure 2H). Consistently, every readout thus demonstrates a negative hedonic bias in response to both appetitive and aversive olfactory stimuli. Interestingly, the emotionality score, which is calculated as a normalized value of anxiety- and depressive-like phenotype based on a large battery of behavioral tests¹⁴, is positively correlated with the odor exploration index of both MU and TMT (Figure 2I), showing a relation between the behavioral response to positive and negative odorants and the anxiety and depressive-like phenotype.

Secondly, we used the same UCMS protocol to induce a depressive-like phenotype in male mice this time. During the Olfactory Preference Test, these mice were exposed to female mouse urine (FU, pure), which is appetitive, and TMT as previously, as a negative odor (Figure 3A,B). There is no difference in the time spent in the four zones during habituation, nor in the distance moved through the test (Figure 3C,D). The time spent in the odor zone shows a significant reduction of FU and TMT exploration in UCMS compared to Controls (Figure 3E). In addition, UCMS mice enter significantly less in the odor zone, or tend to, and present a significantly reduced preference index for both FU and TMT (Figure 3F,G), confirming a negative olfactory bias in a depressive-like state. Likewise, the odor exploration index shows a negatively shifted behavioral response to FU and TMT of UCMS mice compared to Controls (Figure 3H), which tend to be positively correlated with the animals' emotionality score (Figure 3I).

In a third experiment, we resorted to another model of depression which relies on the chronic administration of corticosterone (CORT) in the drinking water of male C57BL/6NTac mice while control mice only receive the vehicle (Veh) solution (10% 2-hydroxypropyl-beta-cyclodextrin) ²⁷. After four weeks of CORT or Veh treatment, mice were habituated for four days, and then exposed to a neutral odor (OCT: 2-trans octenal, 1% in water), a positive odor (PO: peanut oil) and a negative odor (TMT: 2,4,5-trimethylthiazole, 5% in mineral oil) (Figure 4A,B). Through habituation, there were no differences between Veh and CORT (Figure 4C). However, both groups significantly spent more time in the "odor zone" compared to the "control zone". There is no differences for the distance moved, except for the positive odor (PO) for which Veh mice moved more than CORT (Figure 4D). Importantly, like for UCMS mice, the time spent in the odor zone was significantly reduced for CORT mice when exposed to a neutral and appetitive odor (Figure 4E). Reduced odor exploration is also visible through a significant reduction or a tendency to a lower number of entries in the odor zone (Figure 4F). In this case, differences between groups are revealed depending on the readout analyzed, probably due to the limitations related to the number of animals used as well as a potential floor effect caused by the low exploration time observed after TMT exposition. Overall, the preference index is significantly lower in CORT mice for all odor stimuli compared to Veh (Figure 4H). Finally, the odor exploration index for all odors is significantly correlated with the animals' emotionality score (Figure 4I), demonstrating one more time that this olfactory preference test is a suitable tool to assess negative emotional bias in several mouse models of depression.

Finally, we can rule out the possibility that the observed differences may be caused by a reduction in locomotory activity, rather than emotional biases. Indeed, locomotion does not differ between groups during habituation when evaluating total distance moved (Figure 2D, Figure 3D, Figure 4D). In addition, based on the statistical analysis, a general reduction in task engagement or a novelty effect cannot explain these biases either. In fact, we found a significant effect of the odor factor in the preference index analysis, with no significant interaction between experimental groups and odors. These results demonstrate that depressive-like mice indeed perform the task, and that their reduced exploration is specific for each olfactory cue.

Figure 1: General schematic of the timeline and setup of the olfactory preference test. Mice are first habituated to the arena. On the first day of habituation, mice from the same housing cage are placed together to reduce neophobia. The next day, they are placed individually into the arena. After habituation, an odor is added to one of the chambers. Neutral odors are preferentially tested first, followed by positive odors and finally negative odors. For analysis, the arena can be divided into four different zones. Created with Biorender. Please click here to view a larger version of this figure.

Figure 2: Negative olfactory bias in a UCMS female mouse model of depression. (A,B) Timeline and experimental groups of the study: Control and UCMS (Unpredictable Chronic Mild Stress) female mice are tested in the Olfactory Preference Test with a positive (MU: male mouse urine, pure) and negative odor (TMT: 2,4,5-trimethylthiazole, 5% in mineral oil). Created with Biorender. (C) UCMS mice spent on average significantly more time in the "control zone" and less time in the "behind odor zone" compared to Controls during Habituation (Group: F(1, 12) = 74.03, p < 0.0001; Zone: F(2.434, 29.21) = 7.688, p = 0.0012; Interaction: F(3, 36) = 5.148, p = 0.0046). (D) The total distance moved in significantly reduced in UCMS when exposed to MU, but not for Habituation (Hab) and TMT (Group: F(1, 12) = 4.664, p = 0.0518; Odor: F(1.871, 22.45) = 4.846, p = 0.0195; Interaction: F(2, 24) = 4.797, p = 0.0177). (E) UCMS mice spent significantly less time exploring MU and TMT (Group: F(1, 12) = 24.39, p = 0.0003; Odor: F(1, 12) = 59.69, p < 0.0001; Interaction: F(1, 12) = 2.267, p = 0.1580). (F) UCMS mice entered less than the odor zone or tended to (Group: F(1, 12) = 9.279, p = 0.0102; Odor: F(1, 12) = 5.100, p = 0.0433; Interaction: F(1, 12) = 1.649, p = 0.2233). (G) The preference index for MU and TMT is significantly reduced in UCMS mice (Group: F(1, 12) = 24.39, p = 0.0003; Odor: F(1, 12) = 59.69, p < 0.0001; Interaction: F(1, 12) = 2.267, p = 0.1580). (H) The odor exploration index for MU and TMT is significantly reduced in UCMS mice (F(1, 12) = 14.71, p = 0.0024; Odor: F(1, 12) = 0, p > 0.9999; Interaction: F(1, 12) = 0.02630, p = 0.8739). (I) The odor exploration index for MU and TMT is both positively correlated with the animals' emotionality score (MU: Pearson R² = 0.4837, F(1,12) = 11.24, **p = 0.0058; TMT: Pearson R² = 0.5340, F(1,12) = 13.75, **p = 0.0030). Please click here to view a larger version of this figure.

Figure 3: Negative olfactory bias in a UCMS male mouse model of depression. (A,B) Timeline and experimental groups of the study: Control and UCMS male mice are tested in the Olfactory Preference Test with a positive (FU: female mouse urine, pure) and negative odor (TMT: 2,4,5-trimethylthiazole, 5% in mineral oil). Created with Biorender. (C) There are no differences on average in the time spent in the different zones of the arena through habituation (Group: F(1, 13) = 3.921, p = 0.0693; Zoner = F(1.688, 21.94) = 34.57, p < 0.0001; Interaction: F(3, 39) = 0.7385, p = 0.5355). (D) There are no differences in the total distance moved between Controls and UCMS (Group: F(1, 13) = 2.313, p = 0.1522; Odor: F(1.722, 22.38) = 1.746, p = 0.1998; Interaction: F(2, 26) = 0.8806, p = 0.4266). (E) The time spent exploring the odor zone with FU and TMT is significantly reduced in UCMS mice (Group: F(1, 13) = 6.808, p = 0.0216; Odor: F(1, 13) = 16.21, p = 0.0014; Interaction: F(1, 13) = 0.3304, p = 0.5753). (F) UCMS mice entered significantly less times the odor zone for TMT, and tended to for FU (Group: F(1, 13) = 6.572, p = 0.0236; Odor: F(1, 13) = 3.784, p = 0.0737; Interaction: F(1, 13) = 0.2145, p = 0.6509). (G) The preference index of UCMS mice is significantly reduced for FU and TMT (Group: F(1,13 ) = 6.808, p = 0.0216; Odor: F(1, 13) = 16.21, p = 0.0014; Interaction: F(1, 13) = 0.3304, p = 0.5753); (H) The odor exploration index of UCMS is significantly reduced for both odors (Group: F(1, 13) = 12.42, p = 0.0037; Odor: F(1, 13) = 0.002580, p = 0.9603; Interaction: F(1, 13) = 0.5804, p = 0.4598). (I) The odor exploration index for FU and TMT tend to be positively correlated with the animals' emotionality score (FU: Pearson R² = 0.2573, F(1, 13) = 4.504, p = 0.0536; TMT: Pearson R² = 0.2266, F(1, 13) = 3.809, p = 0.0729). Please click here to view a larger version of this figure.

Figure 4: Negative olfactory bias in a CORT male mouse model of depression. (A,B) Timeline and experimental groups of the study: Veh and CORT male mice are tested in the Olfactory Preference Test with a neutral (OCT: trans-2-octenal, 1% in water), positive (PO: peanut oil, pure) and negative odor (TMT: 2,4,5-trimethylthiazole, 5% in mineral oil). Veh and CORT mice are administered with the Vehicle and Vehicle + Corticosterone solution, respectively, during the whole duration of the experiment. Created with Biorender. (C) Average time spent in the different zones of the arena during habituation (Group: F(1, 10) = 0.07588, p = 0.7886; Zone: F(2.224, 22.24) = 6.939, p = 0.0036; Interaction: F(3, 30) = 0.7033, p = 0.5575). (D) Total distance moved during the test for Habituation (Hab) and the different odors (Group: F(1, 10) = 4.193, p = 0.0678; Odor: F(3, 30) = 11.80, p < 0.0001; Interaction: F(3, 30) = 1.489, p = 0.2375). (E) CORT mice spend less time in the odor zone than Veh for OCT and PO (Group: F(1, 10) = 5.094, p = 0.0476; Odor: F(2, 20) = 28.87, p < 0.0001; Interaction: F(2, 20) = 3.841, p = 0.0388). (F) CORT mice enter less times the odor zone or tended to compared to Veh (Group: F(1, 10) = 6.832, p = 0.0259; Odor: F(2, 20) = 2.738, p = 0.0889; Interaction: F(2, 20) = 0.2902, p = 0.7512). (G) CORT qmice display lower preference index for OCT and PO (Group: F(1, 10) = 5.895, p = 0.0356; Odor: F(2, 20) = 20.83, p < 0.0001; Interaction: F(2, 20) = 1.437, p = 0.2611). (H) The odor exploration index is significantly reduced in UCMS mice for all odors (Group: F(1, 10) = 12.35, p = 0.0056; Odor: F(2, 20) = 0.04381, p = 0.9572; Interaction: F(2, 20) = 0.3622, p = 0.7006). (I) The odor exploration index for each odor is positively correlated with the animals' emotionality score (OCT: Pearson R² = 0.5601, F(1, 10) = 12.73, **p = 0.0051; PO: Pearson R² = 0.4495, F(1, 10) = 8.165, *p = 0.0170; TMT: Pearson R² = 0.5061, F(1, 10) = 10.25, **p = 0.0095). Please click here to view a larger version of this figure.

Name	Abbreviation	Dilution	Valence
2,4,5-Trimethylthiazole	TMT	5% in mineral oil	Negative
2-trans-octenal	OCT	1% in water	Neutral
Female mouse urine	FU	Pure	Positive
Male mouse urine	MU	Pure	Positive
Peanut oil	PO	Pure	Positive

Table 1: Examples odors with neutral, positive, and negative valence.

Supplementary File 1: Protocol for 4-week unpredictable chronic mild stress. Please click here to download this File.

Discussion

Negative emotional bias is a major characteristic of depressive disorders and is of particular interest to better understand the physiopathology of the disease. It is thus crucial to both include the evaluation of sensory biases in preclinical models to improve the characterization of the depressive-like phenotype and to investigate which biological mechanisms underlie emotional processing and its alterations in pathological conditions. Moreover, olfaction is a sense of choice to study emotional biases. From a translational perspective, it is ethologically relevant to assess emotional processing both in mice and in humans. In addition, olfactory and emotional systems are highly intertwined and share common neuronal structures throughout evolution. There is a strong, reciprocal relationship between impaired olfaction and depression¹⁰, which makes olfaction a promising and powerful tool to be used for diagnosis and treatment responsiveness, as well as in fundamental research.

Here, we developed a protocol for an olfactory preference test to assess olfactory emotional biases in two mouse models of depression. Importantly, this method allows for the evaluation of the response to a large variety of stimuli with variable valence, both positive and negative, as well as neutral stimuli that could be changed in value under pathological conditions. These later stimuli are not mandatory for the validity of the test, but they can bring interesting additional information to the experiment. This test also allows for the evaluation of the scalability of the response by using different odor concentrations. Assessing the response to stimuli with different hedonic values and separating positive from negative emotional processing in the context of depression is crucial. Indeed, preclinical results indicate that antidepressant treatment may not uniformly restore equally positive and negative olfactory processing¹⁴. Spontaneous exploration of innately salient appetitive or aversive olfactory stimuli also prevents confounding factors from cognitive or motivational deficits associated with depressive-like states^30,31. Therefore, this protocol does not require long learning phases, which has technical benefits. It is important to note that both models used in this study rely on interventions carried out in adulthood, so future experiments will be required to evaluate if this protocol is also sensitive to evaluating other depression risk factors, including early life adversity.

Importantly, this protocol is applicable for both male and female individuals, which is critical as women are twice as likely to be affected by depression as men. However, men are likely to be under-diagnosed from depression due to social expectations, shame or stigma as reviewed in Shi et al.³². Nevertheless, depression seems to differ between men and women, in terms of symptomatology (clinical manifestations), etiology (biological causes) and treatment predictivity^33,34,35, yet many models and tests of depression are still primarily designed for male subjects only. For instance, the standard female urine sniffing test²⁴ uses a similar protocol but relies on a unique social odor cue. Although very informative, this test results in a short exploration time (30 s in controls) and is mainly adapted to male mice as female urine does not have the same ethological and hedonic significance for both sexes. Finally, it appears that olfactory hedonic response is sensitive to antidepressant treatment¹⁴, making our test a tool of choice to reliably characterize depressive-like states, replacing other controversial assays like the force swimming test^36,37,38.

Importantly, the test results in longer stimulus exploration times (100-200 s) compared to previously described tests^21,22,23. Increased exploration duration is particularly important when this test is combined with live monitoring techniques of neuronal activity, for example. In addition, this test does not require food/water deprivation, avoiding metabolic confounding factors. Finally, this test can be performed during the light period, reducing experimental constraints (inverse cycle, infrared recordings).

However, to carry out this protocol successfully, some technical steps are critical. First, it is essential that experiments are performed in a very controlled and non-stressful environment: the lighting should be homogeneous and not be too bright, and there should be no disturbing sound or odor in the room. Mice can be tested in the arena once a day, and the test must be carried out approximately at the same time every day. In addition, a unique experimenter is recommended when performing these experiments to reduce animal stress.

Secondly, sufficient exploration of the arena through habituation days must be ensured. If a mouse spends less than 50 s on average in any of the zones where petri dishes are placed during habituation, one or two supplementary days of habituation can be added before testing the odors. If the mouse is still not exploring the arena long enough, it should be excluded from the analysis. In case of initial dissymmetrical exploration of the "odor zone" during habituation between experimental groups, the calculation of the preference index can be adapted to correct for differential spontaneous preference biases, for example by normalizing by the time spent during habituation in respect to each group, or even in respect to their own exploration during habituation. Indeed, another difficulty can be high variability among experimental groups. If the problem arises, to limit intra-individual variability, mice can be tested on two consecutive days with the same odor. Results can then be averaged between the two days. Importantly, consecutive testing of the olfactory response is not supposed to significantly affect explorative behavior¹⁴.

Regarding odor preparation, solutions must be prepared in a different room from the testing, and experimenters should be aware that the solution does not diffuse excessively in the behavioral room. Finally, solutions should preferably be prepared shortly before the test and stored in a non-odor glass vial so that the odor does not fade away or be contaminated. For some odors like male and female mouse urine that are supposedly very appetitive, if control mice do not present exploration readouts associated with highly positive valence, the solution should be checked. In particular, mouse urine must be freshly collected from young mice (maximum 5 months old) shortly before testing (maximum 2-3 days before) and stored at 4 °C to preserve olfactory properties. Of note, collection from several consecutive days should prevent effects due to estrous cycle stages in female mice. If necessary, urine can be kept at -20 °C for longer-term storage, although this procedure is not recommended as it sensibly reduces odor intensity. Similarly, the perceived pleasantness of peanut oil can also be variable, depending on the batch or brand, for instance. Importantly, two distinct odors or odors with different concentrations can be added in the compartments, instead of the non-odorant "control zone", to measure preference between two odorant solutions.

Finally, the olfactory preference test can be adapted in various ways to improve its design and broaden its potential applications. Indeed, apart from the motor readout, the assessment of the emotional response through this test could be improved by combining different techniques to measure various components of emotions⁶. For instance, an ultrasound microphone can be placed next to the arena to record ultrasonic vocalizations, which can be emitted in response to certain positive and negative stimuli^24,39. Body temperature can also be monitored using an infrared camera or an implantable real-time telemetry transmitter^40,41,42. Some of these transmitters are even capable of accessing cardiac and respiratory rates^43,44,45. Combining these tools would help build a more refined behavioural assessment of the emotional response. The olfactory preference test can also easily be coupled with other techniques to further elucidate neuronal mechanisms behind emotional dysregulation, like optogenetics, fiber photometry or mini-scope live-imaging or in vivo electrophysiology recordings. Lastly, this protocol can be easily translated into the clinic by integrating the assessment of sensory biases and hedonic olfactory response in human research and even as a diagnostic or treatment responsiveness tool.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
2,4,5-Trimethylthiazole	Sigma-Aldrich	W332518-100G-K
2-trans octenal	Sigma-Aldrich	52464-1ML
Blu-tack	UHU	N/A
Filter paper	Fisher Scientific	11710125	42.5 mm
Glass vial	DUTSCHER DOMINIQUE	958983
Mineral Oil	Sigma-Aldrich	M5904-500ML
Noldus Ethovision	Noldus Ethovision	Version 17.5	Alternatively, Single Mouse Tracker in the open-source ICY software or Deeplabcut could be used
Petri dish	Falcon	351007	60 mm x 15 mm Petri dish with 7 custom-made holes of 5 mm diameter
Plexiglas arena	Custom made	N/A	45 cm x 50 cm x 25 cm plexiglas arena divided in two chambers by a 37 cm x 25 cm wall in the middle.