**3. Results**

To test the odor responses of OB neurons under satiated and fasted states, di fferent types of odorants were used. Isoamyl acetate and 2-heptanone were used as neutral odorants, peanut butter [24–26] and food odor were used as appetitive odorants, and 2,4,5-trimethylthiazole and peppermint oil [27] were used as aversive odorants. To confirm that the mice had a preference for the appetitive odorants and avoided the aversive odorants, we performed a preference/avoidance test (Figure 1A). An example is illustrated in Figure 1B: although the mouse had no preference or avoidance for 2-heptanone versus mineral oil, it demonstrated a preference for food odor and avoidance of 2,4,5-trimethylthiazole. Further analysis across all the mice tested demonstrated that isoamyl acetate/2-heptanone, peanut butter/food odor, and 2,4,5-trimethylthiazole/peppermint oil were neutral, appetitive, and aversive odorants for mice, respectively (Figure 1C–E).

To compare the neural activity and sni ffing patterns in satiated and fasted states, signals were recorded before the removal of food, and 12 hours and 24 hours after the removal of food (Figure 1F). Sni ffing signals and neural activity, including spikes and LFP, were recorded simultaneously in awake, head-fixed mice (Figure 1G).

#### *3.1. Baseline Firing Rate and Odor-Evoked Responses are Both Enhanced in a Fasted State*

First, we investigated the spontaneous neural activity of single M/Ts under di fferent nutritional states. Extracellular microelectrodes were placed into the mitral cell layer and single M/T units were isolated and sorted as described previously [18,19,23]. As in previous studies, we observed strong spontaneous firing of M/Ts in awake mice (Figure 2A). Figure 2B shows examples of two single M/Ts sorted from microelectrode recordings. The shapes of these units were similar across satiated and fasted states, indicating that the signals were likely collected from the same units under di fferent states. We performed further analysis on all recorded unit to double check that the spikes were from the same units (repeated measures ANOVA on the spike amplitude and half-width). The data showed that signals recorded at di fferent stages were not significantly di fferent (Friedman's test, for amplitude, x2(2,248) = 4.501, *p* = 0.11; for half-width, x2(2,248) = 4.36, *p* = 0.11), indicating they were likely from same units. Compared with the satiated state, the spontaneous firing of M/Ts was significantly increased at 12 hours after removal of food, and further increased at 24 hours after removal of food (Figure 2C). In addition, we also performed control experiment in which we recorded data at di fferent time points (0 h, 12 h, 24 h), but the food was not removed. The data showed that spontaneous firing of M/Ts at di fferent time point with food were not significantly di fferent (Two-sample *K-S* test, 0 h vs. 12 h, *p* = 0.18, 0 h vs. 24 h, *p* = 0.10, 12 h vs. 24 h, *p* = 0.99; Friedman's test, x2(2,248) = 0.4, *p* = 0.82).

Next, we investigated how fasting a ffects the odor-evoked responses of M/Ts. Consistent with the findings from previous studies, M/Ts showed both excitatory and inhibitory responses to odor stimulation in the satiated state (Figure 2D). Compared with the satiated state, the number of units showing excitatory responses was significantly increased 12 h after the removal of food and the number of units showing inhibitory responses was significantly decreased, for all three types of odorant (neutral, appetitive, and aversive) (Figure 2(E1)). Interestingly, this tendency was not observed 24 h after the removal of food (Figure 2(E1)). Control experiment showed that excitatory responses at di fferent time point with food was not significantly di fferent, for all three types of odorant (Chi-Square Tests, all *p* > 0.05). Since the number of responsive units under satiated state was similar with 24 h after the removal of food, this raises the question that whether they were the same set of units. We provided further presentations of the odor-evoked responses (Figure 2(E2,E3)). Interestingly, we found that most of the units showing excitatory responses under over-fasted state (24 h after removal of food) were not the same units showing excitatory responses under satiated state, indicating that the neural connectivity was reconfigured under over-fasted state (Figure 2(E2,E3)).

To further compare the amplitude of odor-evoked responses under different states, we analyzed the normalized odor response. We found that, compared with the satiated state, the odor response was increased 12 h after the removal of food for all odorants tested, and recovered 24 h after the removal of food (Figure 2F). This finding is consistent with the changes in the number of responsive units under different fasting states. Together, these results from single unit recordings indicate that the excitability of M/Ts is enhanced in fasted mice.

#### *3.2. Neural Discrimination of Odors is Slightly Decreased in the OB of Fasted Mice*

The significant difference in the excitability of M/Ts in satiated and fasted states raises the question of whether odor discrimination by single M/Ts is different under these two states. To investigate this, we characterized the ability of single-unit M/Ts to discriminate the odors. To compare the classification of odor-evoked responses under different nutritional states, we calculated the receiver operating characteristics (ROC) [19,28]. Figure 3A shows two example ROC plots—whereas the ROC curves were similar under different states for the odor pair of peanut butter and food odor (Figure 3A, left), the ROC curves were different under different states for the odor pair of food odor and 2,4,5-trimethylthiazole (Figure 3A, right). ROC analysis of all animals showed that the auROC values for the peanut butter/food odor pair were similar for satiated and fasted states (Figure 3B, left) but the auROC values for the food odor/2,4,5-trimethylthiazole pair were significantly different in different states (Figure 3B, right). We analyzed all odor pairs and found that most of the odor pairs showed smaller auROC values under fasting (Figure 3C,D, left), although no specific odor pair other than food odor/2,4,5-trimethylthiazole had a significant difference in auROC values between the satiated and fasted states (Figure 3C,D, right). The cumulative probability of auROCs analysis of all odor pairs from all animals showed that the auROC values under the satiated state were significantly larger than under the fasted states (Figure 3E) both 12 h and 24 h after the removal of food. Thus, these results indicate that, compared with the satiated state, there was a tendency that the neural discrimination was slightly decreased in the OB of fasted mice under awake, head-fixed conditions.

#### *3.3. Odor-Evoked Gamma Responses are Decreased in a Fasted State*

Whereas single-unit spiking reflects the activity in a single M/T cell, oscillations in the LFP reflect the neural activity in the population of cells surrounding the recording site [29]. Oscillations recorded from the OB contain important information relating to the chemical properties of odors, olfactory learning, and odor discrimination [30,31]. Thus, we next compared the LFP signals in awake, head-fixed mice in satiated and fasted states. As in previous studies, the raw LFP signals were divided into different frequency bands: theta, 2–12 Hz; beta, 15–35 Hz; low gamma, 36–65 Hz; and high gamma, 66–95 Hz (Figure 4A). No significant differences were found between the satiated and fasted states for any frequency band of the ongoing, baseline LFP (Figure 4B).

Next, we investigated whether odor-evoked LFP responses differed with fasting state in awake, head-fixed mice. Figure 5A–C show the LFP response to isoamyl acetate in a single mouse across different fasting states. In all states, there was a strong beta response to the odor and a high gamma response. However, although the amplitude of the isoamyl acetate-induced beta-band response was similar in the fasted state and the satiated state, the amplitude of the isoamyl acetate-induced high-gamma response decreased with fasting (Figure 5B,C). This phenomenon was also observed for other odorants (e.g., Figure 5D–I).

**Figure 3.** Nutritional status influence odor representation in M/Ts. (**A**) Example receiver operating characteristic (ROC) plots of the neural responses to peanut butter vs. food (left) and food vs. 2,4,5-trimethylthiazol (right) when mice had been fasted for 0 h (black), 12 h (orange), or 24 h (red). (**B**) Comparison of areas under the ROC (auROCs) for the neural responses to peanut butter vs. food (left) and food vs. 2,4,5-trimethylthiazol (right). Each gray circle represents the auROC value for a single unit. Left. Friedman's test: χ2 (2,248) = 0.36, *p* = 0.83. Right. Friedman's test: χ2 (2,248) = 14.06, *p* = 0.00089, 0 h vs. 12 h, *p* = 0.0059, 0 h vs. 24 h, *p* = 0.0020. (**C**,**D**) Pseudocolor plots of the D-value and the *p* value when auROC12 was compared with auROC0 (**C**) or when auROC24 was compared with auROC0 (**D**). (**E**) Cumulative probability of auROCs. Two-sample *K-S* test: 0 h vs. 12 h, *p* = 0.0092, 0 h vs. 24 h, *p* = 0.0024, 12 h vs. 24 h, *p* = 0.90. \*\* *p* < 0.01.

**Figure 4.** Nutritional status has no significant effect on the ongoing LFP in the OB. (**A**) Examples of ongoing baseline LFP signals from a single mouse fasted for 0 h (black), 12 h (orange), and 24 h (red). The first row shows 6 s of the raw trace; the second to fifth rows show the filtered signal (theta, beta, low gamma, and high gamma, respectively). (**B**) The averaged power spectrum of the ongoing LFP signals. (**B1**–**B4**) show the averaged power spectrum in the theta (**B1**), beta (**B2**), low gamma (**B3**), and high gamma (**B4**) bands across the group of mice. (**B1**). Friedman's test: χ2 (2,38) = 0.57, *p* = 0.75. (**B2**). Friedman's test: χ2 (2, 80) = 2.06, *p* = 0.36. (**B3**). Friedman's test: χ2 (2, 118) = 2.85, *p* = 0.24. (**B4**). One-way rANOVA: *F* (2,116) = 1.09, *p* = 0.34. Error bars show the SEM.

**Figure 5.** Nutritional status modulates odor-evoked LFP responses. (**A**) Responses of the raw LFP trace and the filtered beta and high gamma bands to odor stimulation under different nutritional states. Black bars indicate odor stimulation. (**B**,**C**) Top: Example power spectra for odor-evoked beta (**B**) and high gamma oscillations in the OB when mice were fasted for 0 h (black), 12 h (orange), or 24 h (red). Bottom: Trial-averaged normalized traces of odor-evoked beta (**B**) and high gamma (**C**) responses. The red dotted lines indicate the period of odor stimulation. Error bars show the SEM. (**D**–**F**) Comparison of the power in the normalized odor-evoked beta band evoked by neutral (**D**), appetitive (**E**), or aversive odorants (**F**) under different fasting states (n = 12, neutral odorants, Friedman's test: χ2 (2,46) = 1, *p* = 0.61; appetitive odorants, Friedman's test: χ2 (2,46) = 0.58, *p* = 0.75; aversive odorants, Friedman's test: χ2 (2,46) = 4.08, *p* = 0.13). (**G**–**I**) Comparison of the power in the normalized odor-evoked high gamma evoked by neutral (**G**), appetitive (**H**), or aversive odorants (**I**) under different fasting conditions (n = 12, neutral odorants, Friedman's test: χ2 (2,42) = 8.27, *p* = 0.016, 0 h vs. 24 h, *p* = 0.012; appetitive odorants, Friedman's test: χ2 (2,42)= 7.64, *p* = 0.022, 0 h vs. 24 h, *p* = 0.018; aversive odorants, Friedman's test: χ2 (2,42) = 8.45, *p* = 0.015, 0 h vs. 24 h, *p* = 0.018). \* *p* < 0.05.

When the odorant-induced changes in LFP power under satiated and fasted states were surveyed across all mice and odors (Figure 5D–I), we found that there was no significant difference between satiated and fasted states for the odor-evoked beta response (Figure 5D–F), but the high gamma response was significantly reduced for all the three types of odorants with the longest fasting duration (48 h after the removal of food, Figure 5G–I). These results indicate that odor-evoked inhibition of the neural network in the OB is reduced in a fasted state in awake, head-fixed mice.

#### *3.4. Slight Decrease in the Sni*ffi*ng Volume Between Satiated and Fasted States*

Since mice rely on respiration/sniffing to sample odors, sniffing plays an important role in odor processing and representation in the OB [32]. Thus, we next investigated whether sniffing changes under satiated and fasted states. The raw sniffing patterns recorded under different states are shown in Figure 6A. We analyzed different aspects of the sniffing signal, including sniffing frequency, mean inhalation duration (MID), and volume (Figure 6B). As shown in Figure 6C, there were no significant changes in ongoing sniffing frequency, MID, or volume in the fasted state versus the satiated state. Analysis of the odor-evoked sniffing data across all animals supports the result that only sniffing volume changed with fasting (24 h after the removal of food); this phenomenon was observed consistently for all three types of odorant (Figure 6D–F). Thus, taken together, these results indicate that there is only a weak difference in the sniffing pattern between satiated and fasted states, limited to a slight decrease in sniffing volume.

**Figure 6.** Change in the sniffing volume under different nutritional states. (**A**) Raw sniff trace recorded from a representative mouse fasted for 0 h (black), 12 h (orange), or 24 h (red). (**B**) Diagram illustrating the extraction of sniff frequency, mean inhalation duration (MID), and volume from a sample nasal flow trace. (**C**) Sniffing recorded under baseline conditions (no odor presentation) after fasting for different durations. (**C1**). Mean sniff frequency. One-way ANOVA: *<sup>F</sup>*(2,22) = 1.30, *p* = 0.29. (**C2**). Sniff mean inhalation duration (MID). Friedman's test: χ2 (2,22) = 2.17, *p* = 0.34. (**C3**). Sniff volume. Friedman's test: χ2 (2,22) = 3.17, *p* = 0.21. (**D**) Odor-evoked mean sniff frequency recorded when mice were fasted for 0 h (black), 12 h (orange), or 24 h (red). Neutral odorants: Friedman's test: χ2 (2,42) = 3.91, *p* = 0.14. Appetitive odorants: Friedman's test: χ2 (2,42) = 3.27, *p* = 0.19. Aversive odorants: Friedman's test: χ2 (2,42) = 4.73, *p* = 0.094. (**E**) Fasting has no effect on odor-evoked sniff MID. Neutral odorants: Friedman's test: χ2 (2,42) = 1.91, *p* = 0.38. Appetitive odorants: Friedman's test: χ2 (2,42) = 2.45, *p* = 0.29. Aversive odorants: Friedman's test: χ2 (2,42) = 1.91, *p* = 0.38. (**F**) Odor-evoked sniff volume changes with fasting. Neutral odorants: Friedman's test: χ2 (2,42) = 10.18, *p* = 0.0062, 0 h vs. 24 h, *p* = 0.042, 12 h vs. 24 h, *p* = 0.0072. Appetitive odorants: Friedman's test: χ2 (2,42) = 6.91, *p* = 0.032, 0 h vs. 24 h, *p* = 0.042. Aversive odorants: Friedman's test: χ2 (2,42) = 9.91, *p* = 0.0071, 0 h vs. 24 h, *p* = 0.028, 12 h vs. 24 h, *p* = 0.012. \* *p* < 0.05, \*\* *p* < 0.01.
