To examine potential changes in feeding-related neuromodulatory molecule expression with odour-food association, mice were divided into a group trained to associate a particular odour with a food reward and a control group. C57BL/6 male mice at 8 weeks of age (n = 9 mice) were food restricted to 80–90% of their original body weight, and were presented with an odour molecule, eugenol, combined with a sugar reward four times per day (Fig. 1A). Control mice (n = 9) were similarly food-restricted and presented with eugenol alone. This procedure was conducted for 8 successive days. On day 8, all the trained mice located the eugenol-scented sugar under the bedding very rapidly. Quantification of the behaviour from randomly sampled mice (n = 5 mice) showed their short latency for odour source detection (latency on the first trial, 5.8 ± 1.6 s; latency on the fourth trial, 3.0 ± 0.7 s; average ± SEM). On the other hand, no control mice dug into the bedding over the eugenol during the 2-min observation period.Figure 1

Odour-food reward association training and dissection of olfactory areas. (A) Protocol for odour-food association training. For trained mice, food was restricted and odour (eugenol) was presented with sugar four times per day from day 1 to day 8. On day 9, the mice were presented with the odour but no sugar for 30 min, and their digging behaviour over the odour stimulus was evaluated. Control mice were subjected to the same food restriction, odour presentation without sugar from day 1 to day 8, and behavioural analysis on day 9. (B) Digging behaviour of control and trained mice. On day 9, the digging behaviour of mice over the odour stimulus during the 30-min test period was evaluated for control and trained mice. Data show the average ± SEM. n = 5 for each treatment; **p < 0.01 (t-test). (C) Dissection of olfactory areas from the mouse brain. The ventral view of the mouse brain is shown. The olfactory bulb (OB), three areas of the olfactory tubercle (amOT, cOT, lOT), and olfactory cortical area (OC) were dissected as indicated. (D) Histological examination of the OT and OC dissection. Coronal sections of intact brain (left) and dissected brain (right) were stained with DAPI. The OT and piriform cortex (PC) were appropriately dissected. (E) Molecular expression of dissected olfactory areas. Protein extracts from the OT, OC, and neocortical area (NC) were electrophoresed and examined for DARPP32 (left) and RGS14 (right) expression using immunoblots. Images of blots have been cropped and full images are included in the Supplementary Information. (F) Blood glucose and plasma insulin concentrations in control and trained mice. After behavioural tests on day 9, the blood was collected and the blood glucose (left) and plasma insulin (right) were measured. Data show the average ± SEM. n = 6 for control and 5 for trained groups. n.s., not significant (t-test).

On day 9, mice were presented with eugenol alone without a sugar reward under the bedding. All the trained mice showed extensive food searching behaviour toward the odour source. Quantification of the digging behaviour over the odour source during the 30-min test period for the sampled mice (n = 5 for each group) showed that trained mice extensively dug the bedding over the odour stimulus to search for the sugar reward, while control mice showed only a short period of digging behaviour over the odour stimulus (Fig. 1B). Digging behaviour of trained mice was observed within 10 seconds after their introduction into the test cage, while that in control mice was delayed by least several minutes (data not shown). These behaviours clearly showed that trained mice associated the odour cue with a food reward, while control mice did not.

After the 30-min test period, mice were deeply anesthetized and sacrificed, and the brain tissues were recovered. The OB, OT, and olfactory cortical area except the OT (referred to as OC; major structure was likely the piriform cortex) were dissected (Fig. 1C). Figure 1D shows the remaining brain structure after dissection of the olfactory areas. The OT and piriform cortex were removed, while deeper brain structures remained intact. Western blot analysis showed that DARPP-32, a signalling molecule downstream of dopamine receptors²³, was enriched in the OT sample (Fig. 1E, left). RGS14, a regulator of G-protein signalling that is highly expressed in the piriform cortex and hippocampus²⁴, was enriched in the OC sample but not in the OT or neocortex (NC) sample (Fig. 1E, right). These results confirmed the proper dissection of olfactory areas.

The OT contains functional domains, with the anteromedial domain being responsible for attractive behaviour and the lateral domain for aversive behaviour²². Thus, the OT was further divided into the anteromedial region (amOT), lateral region (lOT) and remaining central region (cOT) (Fig. 1C; see Methods). Accordingly, the five olfactory areas of OB, OC, amOT, cOT, and lOT were each collected from control and trained mice, and then processed for mRNA expression analysis.

Trained mice were supplied with small amount of sugar reward during 8 training days (~200 mg per day). To examine whether the sugar supply altered the metabolic state of mice, glucose and insulin contents in the peripheral blood were examined in different sets of mice that received same food restriction and training processes as described above. On day 8, all the trained mice (n = 5) located the eugenol-scented sugar under the bedding very rapidly (latency on the first trial, 4.8 ± 1.5 s; latency on the fourth trial, 3.0 ± 1.2 s; average ± SEM), while control mice (n = 6) did not during the 2-min observation period. On day 9, trained mice but not control mice exhibited extensive digging behaviour over the odour source during the 30-min test period (trained mice, 44.4 ± 7.8 s, n = 5; control mice, 6.2 ± 2.9 s, n = 6; average ± SEM). After the behaviour test they were deeply anesthetized and sacrificed, and their blood was collected. The blood glucose and plasma insulin concentrations did not differ between the two groups (Fig. 1F), suggesting that the metabolic state of trained mice was not significantly altered by the sugar supply.

mRNA collected from the olfactory areas was subjected to quantitative real-time PCR to determine the expression levels of feeding-related neuromodulatory molecules. We selected 23 molecules related to feeding, including neuropeptides, growth factors, endogenous opioids, cannabinoids, and their receptors, as well as related signalling molecules (Table 1). The expression levels were normalized to that of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh), which was used as the internal control because its expression level did not differ significantly among the five olfactory areas based on quantitative real-time PCR (data not shown).Table 1

Neuromodulatory signalling molecules examined and PCR probes used in this study.

Firstly, expression levels in different olfactory areas were compared in control mice, which were subjected to food restriction and odour (eugenol) presentation that was not associated with a food reward. The data are shown in Table 2 as mean ± SD, in Supplementary Table S1 as F and p values, in Table 3 as post-hoc Tukey-Kramer comparison tests, and as representative figures in Fig. 2, and figures for all 23 molecules in Supplementary Fig. S1. In each group of data, 7–9 samples from different mice were examined.Table 2

Expression of neuromodulatory molecules in individual olfactory areas of control mice. Data show the amount of each molecule relative to Gapdh (×10⁻³; mean ± standard deviation (SD)).

Most molecules showed differing expression levels among olfactory areas. Generally, expression was relatively high in the OC and OT compared to the OB. For most of the 23 molecules examined, expression in the OC and/or OT was higher than or equivalent to that in the OB (Tables 2 and 3, Supplementary Fig. S1). Only ghrelin (Ghrl) and opioid receptor delta 1 (Oprd1) showed higher expression in the OB compared to the OC (see Fig. 2 for Oprd1 expression).

Further, many molecules showed higher expression in the OT than in the OB and OC. Orexin receptor 2 (Hcrt2), melanocortin 4 receptor (Mc4r), neprilysin (Mme), opioid receptor kappa 1 (Oprk1), prodynorphin (Pdyn), preproenkephalin (Penk) and tachykinin 1 (Tac1) showed higher expression in all three OT areas than in the OB or OC. A few molecules, such as brain-derived neurotrophic factor (Bdnf) and oxytocin receptor (Oxtr), had their highest expression levels in the OC (see Fig. 2 for Oxtr expression). Among the three OT areas, many molecules showed higher expression in the anteromedial and central OT. Expression levels of arginine-vasopressin receptor 1a (AVP1a), CART peptide (Cartpt) and orexin receptor 1 (Hcrtr1) were highest in the anteromedial OT among the three OT areas (see Fig. 2 for Cartpt and Hcrtr1 expression). Expression levels of cannabinoid receptor 2 (Cnr2), ghrelin (Ghrl), melanocortin 4 receptor (Mc4r), neuropeptide Y receptor type 2 (Npy2r), opioid receptor delta 1 (Oprd1), preproenkephalin (Penk) and tachykinin 1 (Tac1) were highest in the central OT (see Fig. 2 for Npy2r expression). Expression levels of ghrelin receptor (Ghsr), orexin (Hcrt), orexin receptor 2 (Hcrtr2), opioid receptor kappa 1 (Oprk1) and prodynorphin (Pdyn) were higher in the anteromedial and central OT than in the lateral OT (see Fig. 2 for Oprk1 expression). Conversely, neprilysin (Mme) showed higher expression in the lateral OT than in the anteromedial OT.

Taken together, while each neuromodulatory molecule examined in this study had differing expression profiles among olfactory areas, the overall tendency was that many molecules showed higher expression in the OT than in the OB and OC, especially in the anteromedial and central OT.

Molecular expression in the five olfactory areas was compared between control mice and odour-food reward association-trained mice to determine whether the expression of feeding-related neuromodulatory molecules is regulated by odour-related food experience. The data were first normalized to Gapdh expression, and then evaluated using the 2^−ΔΔCT method²⁵. The relative expression in control and trained mice (mean ± SD) was compared. Given the examination of 23 different molecules, the p values of t-test were adjusted by adopting Storey’s method for multiple comparisons²⁶. Their statistical significance based on the adjusted p values are shown in Table 4 (significance, p < 0.05); representative figures are shown in Fig. 3, figures for all 23 molecules in Supplementary Fig. S2, and all the unadjusted and adjusted p values in Supplementary Table 2. In each group of data, 7–9 samples from different mice were examined.Table 4

Expression of neuromodulatory molecules in individual olfactory areas of control and trained mice. Expression of each molecule was first normalized to Gapdh, and the relative values in control and trained mice were then calculated based on the averaged ΔCt of control mice. Data show the mean ± SD. Symbols indicate Storey-adjusted p values. ^#0.05 < p < 0.1; *p < 0.05. Bold letters indicate data sets with significant p values (p < 0.05).

Many molecules showed differential expression between control and trained mice, and the difference was most conspicuous in the anteromedial OT. Expression of arginine-vasopressin (AVP), cannabinoid receptor 1 (Cnr1), ghrelin (Ghrl), leptin receptor (Lepr), melanocortin 4 receptor (Mc4r), neprilysin (Mme), opioid receptor delta 1 (Oprd1) and opioid receptor kappa 1 (Oprk1) were elevated in the anteromedial OT of trained mice (Table 4, see Fig. 3 for Avp, Cnr1, Ghrl, Lepr and Oprd1 expression). Among these compounds, ghrelin (Ghrl) was also enhanced in the lateral OT (see Fig. 3 for Ghrl expression), while the other seven molecules specifically increased in the anteromedial OT among the three OT areas.

Differential expression was observed in other olfactory areas. Expression of leptin receptor (Lepr) increased in the OB of trained mice (see Fig. 3 for Lepr expression). Decreased expression in the trained mice was also observed. In the OC, cannabinoid receptor 2 (Cnr2) and oxytocin receptor (Oxtr) decreased in the trained mice. In the central OT, cannabinoid receptor 2 (Cnr2) decreased in the trained mice (see Fig. 3 for Cnr2 expression).

These results indicate that expression levels of neuromodulatory molecules in the olfactory areas were significantly altered by odour-food reward association training. The alteration was most conspicuous in the anteromedial OT, where many molecules showed increased expression in the trained mice. In other olfactory areas, not only increased expression but also decreased expression was observed in the trained mice.

While not attaining the authentic statistical significance level, some molecules showed p values below 0.1 (0.05 < p < 0.1; shown in Table 4 as #) for the differential expression in trained mice. Their alteration toward increase or decrease appeared to differ among olfactory areas. In the OB, expression of BDNF (Bdnf), cannabinoid receptor 1 (Cnr1), ghrelin (Ghrl), neprilysin (Mme), opioid receptor kappa 1 (Oprk1), prodynorphin (Pdyn) and preproenkephalin (Penk) altered toward increase in trained mice. Intriguingly, in the OC, expression of BDNF (Bdnf), cannabinoid receptor 1 (Cnr1), ghrelin (Ghrl), orexin receptor 2 (Hctr2) and neuropeptide Y receptor Y2 (Npy2r) altered toward decrease, while only arginine-vasopressin (AVP) toward increase in trained mice. In the central OT, expression of ghrelin (Ghrl), melanocortin 4 receptor (Mc4r), neprilysin (Mme), opioid receptor delta 1 (Oprd1), prodynorphin (Pdyn) and tachykinin 1 (Tac1) altered toward decrease in trained mice. In the lateral OT, cannabinoid receptor 1 (Cnr1), leptin receptor (Lepr) and melanocortin 4 receptor (Mc4r) altered toward increase and POMC (Pomc) toward decrease in trained mice. In the anteromedial OT, oxytocin (Oxt) altered toward increase and CART peptide (Cartpt) toward decrease in trained mice. The overall tendency of expression pattern may help speculating the possible roles of neuromodulatory molecules in these olfactory areas (see Discussion).

All experiments were conducted in accordance with the guidelines of the Physiological Society of Japan and were approved by the Kochi Medical School Animal Care and Use Committee. Male C57BL/6 mice (Japan SLC Inc., Sizuoka, Japan) were housed individually in plastic cages (24 × 17 × 12 cm) with wood shavings at 26 °C under a 12-h light/dark cycle (light on at 21:00 and off at 9:00). Odour-food association training was initiated when mice were 8 weeks of age.

For odour and sugar association training, mice were food-restricted to achieve 80–90% of their ad libitum feeding body weight. Food pellets were removed 2 days prior to the start of odour-sugar association training. A limited amount of food pellets (2.7–3.3 g per day) was then delivered, to maintain 80–90% body weight during the training period. Water was available ad libitum throughout the experiment. Association training and behavioural tests were conducted during dark phase (between 9:00 and 12:00; dark phase was 9:00–21:00). Association training, of odour with the sugar reward, was conducted in a plastic conditioning cage (24 × 17 × 12 cm) with 2-cm-deep paper bedding (Japan SLC Inc.). During the first trial on day 1, mice learned to eat sugar; they were presented with sugar (granulated sucrose; 20–40 mg) together with powdered diet (20–40 mg) on a petri dish, with holes exposing a filter paper (2 cm × 2 cm) soaked with 10 µl of eugenol (Tokyo Chemical Industry, Tokyo, Japan) in the lower compartment. In the three successive trials of day 1, the mice underwent association training with the sugar reward (50 mg; without powdered diet) and eugenol in the same dish. The dish was hidden under the bedding (2 cm depth) so that the mouse could locate it by smell, and the position of the hidden dish within the conditioning cage was randomly chosen. Because mice could find and eat the sugar within 2 min, the duration of one trial was fixed at 2 min. Four training trials per day were conducted for a further 7 consecutive days (days 2–8). For control mice, the same food restriction was applied, and a holed petri dish containing filter paper soaked with 10 µl of eugenol without sugar was presented for one 2-min trial. Four trials per day were conducted from day 1 to day 8. For the first and fourth trials on day 8, the latency period to the start of digging over the dish with eugenol (with or without sugar) was measured for randomly sampled trained and control mice. Controlled amount of food pellet was delivered after the each day’s training.

On day 9, mice were habituated to a larger test cage (30 × 20 × 13 cm) with paper bedding (2 cm depth) for 30 min. Then, a dish containing eugenol, but without sugar, was buried under the bedding of the test cage and presented to both trained and control mice. The behaviour of sampled mice for 30 min after odour presentation was video-recorded, and the total duration of digging behaviour over the scented dish over 30 min was evaluated.

Mice were deeply anesthetized through intraperitoneal injection of sodium pentobarbital. They were transcardially perfused with phosphate-buffered saline (PBS) and then decapitated. The brain was removed from the skull, and the OB, amOT, lOT, cOT and OC, aside from the OT (major structure was likely the piriform cortex), were dissected and collected separately. The boundary between the amOT and cOT was a line extending mediocaudally from the rostral pole of the OT to separate semicircular anteromedial OT region, and that between the cOT and lOT was a line connecting the rostral and lateral poles of the OT (see Fig. 1C). The OC was dissected from the lateral and caudolateral area to the OT, including the piriform cortex.

For quantitative real-time PCR analysis, each brain tissue specimen was soaked in 1.5 ml of RNAlater solution (Thermo Fisher Scientific, Waltham, MA, USA) for at least 8 h at 4 °C, and then held at −80 °C in the same solution until analysis. For western blot analysis, the OB, OT (recovered as one block without separation), OC, and dorsal area of the neocortex (major structure was likely the somatosensory cortex) were recovered. These specimens were soaked in protein solution, as described in the following section.

After dissection of the olfactory areas, the brain was immersed in 4% paraformaldehyde in 0.1 M phosphate buffer (PB) overnight, and then transferred to 30% sucrose in 0.1 M PB. The brain was then embedded in optimal cutting temperature (OCT) compound (Sakura Finetechnical, Tokyo, Japan), frozen at −80 °C, and sliced into coronal sections with a thickness of 50 μm using a cryotome. The sections were rinsed with PBS, mounted on slide glasses, and stained with DAPI (1 µg/ml) for 5 min. After washing in PBS, sections were mounted in Prolong Gold (Thermo Fisher Scientific). The intact brain without olfactory area dissection was processed using the same procedure. Images were acquired using fluorescence microscopy.

Dissected brain tissues were soaked in lysis buffer containing 1× RIPA buffer (Cat# 89900; Thermo Fisher Scientific), 0.5 M EDTA and 1× protease and phosphatase inhibitor cocktail (Cat# 78440; Thermo Fisher Scientific) at 4 °C. The tissues were sonixged at 14,000 g for 20 min at 4 °C; their supernatants were then collected. Brain extracts containing 20 µg protein were fractionated through SDS–PAGE using a polyacrylamide gel (Cat# 4561096; Bio-Rad, Hercules, CA, USA) according to the manufacturer’s protocol. The proteins were transferred to polyvinylidene fluoride (PVDF) membranes blocked with 4% skim milk in TTBS (in mM: 150 NaCl, 20 Tris, pH 7.5, 0.1% Tween-20) overnight at 4 °C, and incubated with a primary antibody, anti-DARPP32 (1:400; Cat# ab40801; Abcam, Cambridge, UK) or anti-RGS14 (1:500; Cat# GTX15147; GeneTex, Irvine, CA, USA), for 1 h at room temperature. After three rinses with TTBS for 5 min, the membranes were incubated with the secondary antibody, horseradish peroxidase (HRP)-conjugated anti-rabbit IgG (1:5000, Cat# 401315; Millipore, Billerica, MA, USA), for 30 min at room temperature. After three rinses with TTBS for 5 min, the blots were developed with Luminata^TM Forte Western HRP Substrate (Millipore) and the Lumino Image Analyzer System LAS4000 (Fujifilm, Tokyo, Japan).

Different sets of mice were similarly food-restricted and received odour-food association procedure. After the behavioural test on day 9, they were deeply anesthetized through intraperitoneal injection of sodium pentobarbital. Then the thoracic cavity was opened, the body fluid in the cavity was removed by using cotton bud, and the right atrium of the heart was cut by scissors. The blood leaked into the thoracic cavity was recovered and measured for glucose concentration by using a blood glucose meter (FreeStyle, Abbott, Canada). For the measurement of insulin concentrations, the blood was collected in a tube containing EDTA (final concentration, 1 mg/ml). The samples were ice-cooled and centrifuged at 1,800 g for 10 min at 4 °C, and the plasma was collected. The plasma insulin concentrations were determined using a Mouse/Rat Insulin ELISA kit (Ultra sensitive kit, Morinaga Institute of Biological Science, Japan) according to the manufacturer’s instruction.

The corresponding right and left olfactory areas of each mouse brain were collected as one tissue sample. Total RNA from each tissue sample was extracted using the RNeasy Lipid Tissue Mini Kit (Qiagen Ltd., Tokyo, Japan). The tissue samples were placed into 200 μl of QIAzol Lysis Reagent, homogenized with a disposable pestle (Sansyo Co., Ltd., Tokyo, Japan), and combined with 800 μl of QIAzol Lysis Reagent. Then, the samples were processed using the RNeasy Lipid Tissue kit, following the manufacturer’s instructions. On-column DNase digestion (RNase-Free DNase Set; Qiagen Ltd.) was conducted to remove genomic DNA.

The RNA from OC and OB samples was eluted with 30 μl of nuclease-free water, and that from OT samples was eluted with 20 μl of nuclease-free water. The quantity and quality of the extracted RNA was measured using a Nanodrop spectrophotometer (Thermo Fisher Scientific) and Agilent 2100 Bioanalyzer (Agilent Technologies. Ltd., Santa Clara, CA, USA). The OC and OB samples were evaluated with the Agilent RNA 6000 Nano Kit, and OT samples with the Agilent RNA 6000 Pico Kit. Only RNA samples with an RNA integrity number (RIN) of at least 7.5 were used in this study. When needed, RNA was concentrated using a SpeedVac (Thermo Fisher Scientific) to achieve a final concentration of 12.5 ng/μl.

cDNA was synthesized using SuperScript™ IV VILO™ Master Mix (Thermo Fisher Scientific) according to a modified version of the manufacturer’s instructions. First, 2 μl of SuperScript™ IV VILO™ Master Mix was added to 8 μl of extracted RNA (100 ng total RNA) in reaction tubes. Reverse transcription was conducted by heating the reaction mixture to 25 °C for 10 min, 50 °C for 10 min, and 85 °C for 5 min, before cooling to 4 °C with the Applied Biosystems GeneAmp PCR system 9700 (Thermo Fisher Scientific). As the majority of samples were small, the minimum amount of cDNA that could be detected by PCR probes was determined in advance, as was the dilution concentration. Synthesized cDNA was diluted 10-fold with 10 mM Tris-HCl (pH 8.0) and stored at −70 °C.

The q-PCR assay was performed using a QuantStudio 5 Real-time PCR system (Thermo Fisher Scientific) according to the manufacturer’s instructions. The reaction was carried out with TaqMan™ Gene Expression Master Mix (Thermo Fisher Scientific) in a 20-μl volume containing 2 μl of cDNA. The protocol took approximately 60 min to complete, including a 2-min incubation step at 50 °C and an enzyme activation step for 10 min at 95 °C, followed by 40 15-s cycles of denaturation at 95 °C and an annealing and extension step for 1 min at 60 °C. Reactions were performed in duplicate and the threshold cycle (Ct) values were averaged among replicates. Negative controls were included to detect possible contamination. The q-PCR probes used are listed in Table 1.

An amplification curve was generated after analysing the raw data and adjusting the Ct values. Glyceraldehyde 3-phosphate dehydrogenase (Gapdh) was used as the internal control gene for normalizing the relative expression levels of genes of interest. Relative expression was evaluated using 2^−ΔCT for Tables 2, 3, Fig. 2 and Supplementary Fig. S1, and the relative expression in control and trained mice was compared using the 2^−ΔΔCT method²⁵ for Table 4, Fig. 3, Supplementary Fig. 2 and Table 2. In the 2^−ΔΔCT analysis, the average of ΔCT values in control mice was determined and the ΔΔCT value in each control and trained mouse was caleraged ΔCT of control mice. Then the 2^−ΔΔCT values were calculated and subjected to statistical analyses.

Statistical analyses were performed using R software (version 3.4.3; R Core Team, 2017). One-way ANOVA, followed by Tukey-Kramer’s multiple comparison test (expression in control mice relative to Gapdh gene expression) was used for comparisons among the five olfactory areas. P values for the comparisons between control and trained mice were calculated using a two-tailed Student’s t-test. The p values were further adjusted by using Storey’s method for multiple comparisons²⁶. Statistical significance was set at p < 0.05. Graphs were constructed using GraphPad Prism software (version 8.0; GraphPad Software Inc., La Jolla, CA, USA). In the box plot, the box indicates the 25^th and 75^th percentiles, defined as the interquartile range (IQR), while whiskers indicate the minimum and maximum values, and the line inside the box represents the median.

Supplementary Information.