Experiments were performed on male and female mice expressing Cre recombinase (Cre) in defined neural populations. Mouse strains used were: Pcdh21-Cre [Tg(Pcdh21-cre)BYoko], Gensat Stock #030952-UCD; OMP-Cre [Tg(Omp-tm4-Cre)Mom], JAX Stock #006668, Tbet-Cre [Tg(Tbx21-cre)1Dlc], JAX Stock #024507, CCK-IRES-Cre [Tg(CCK-IRES-Cre)Zjh], JAX Stock #012706 (Haddad et al., 2013), and Thy1-jRGeCO1a Tg(Thy1-jRGECO1a)GP8.31Dkim/J, JAX Stock #030526 (Dana et al., 2018). Mice ranged from 3 to 8 months in age. Mice were housed up to four/cage and kept on a 12/12 h light/dark cycle with food and water available ad libitum. All procedures were carried out following the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Utah Institutional Animal Care and Use Committee.

Viral vectors were obtained from the University of Pennsylvania Vector Core (AAV1 or 5 serotype, AAV.hSynap-FLEX.iGluSnFR, AAV1 serotype AAV.hSynap-FLEX.GCaMP6f), HHMI Janelia Campus or Vigene (AAV1 or 5 serotype, pAAV.hSynap-FLEX.SF-iGluSnFR.A184V, pAAV.hSynap-FLEX.SF-iGluSnFR.A184S). Virus injection was done using pressure injections and beveled glass pipettes, as described previously (Rothermel et al., 2013; Wachowiak et al., 2013; Short and Wachowiak, 2019). After injection, mice were given carprofen (Rimadyl, S.C., 5 mg/kg; Pfizer) as an analgesic and enrofloxacin (Baytril, I.M., 3 mg/kg; Bayer) as an antibiotic immediately before and 24 h after surgery. Mice were singly housed after surgery on ventilated racks and used 21–35 days after virus injection. In some mice, viral expression was characterized with post hoc histology using native fluorescence.

Two-photon imaging in anesthetized mice was performed as described previously (Wachowiak et al., 2013; Economo et al., 2016). Mice were initially anesthetized with pentobarbital (50–90 mg/kg) then maintained under isoflurane (0.5–1% in O₂) for data collection. Body temperature and heart rate were maintained at 37°C and ~400 beats per minute. Mice were double tracheotomized and isoflurane was delivered passively via the tracheotomy tube without contaminating the nasal cavity (Eiting and Wachowiak, 2018). Two-photon imaging occurred after removal of the bone overlying the dorsal olfactory bulb.

Imaging was carried out with a two-photon microscope (Sutter Instruments or Neurolabware) coupled to a pulsed Ti:Sapphire laser (Mai Tai HP, Spectra-Physics; or Chameleon Ultra, Coherent) at 920–940 nm and controlled by either Scanimage (Vidrio) or Scanbox (Neurolabware) software. Imaging was performed through a 16×, 0.8 N.A. objective (Nikon) and emitted light detected with GaAsP photomultiplier tubes (Hamamatsu). Fluorescence images were acquired using unidirectional resonance scanning at 15.2 or 15.5 Hz. For dual-color imaging, a second laser (Fidelity-2; Coherent) was utilized to optimally excite jRGECO1a (at 1,070 nm) and emitted red fluorescence collected with a second PMT, as described previously (Short and Wachowiak, 2019; Moran et al., 2021).

in vivo pharmacology was carried out after removing the bone and dura overlying the dorsal olfactory bulb, using protocols described previously (Pírez and Wachowiak, 2008; Brunert et al., 2016). For experiments where both OBs were imaged sequentially in a single mouse, we performed pharmacology on one bulb while the other bulb remained encased in bone. After the experiment was complete on the initial OB, the bone overlying the contralateral OB was repeatedly rinsed with Ringers solution, thinned then craniotomized, the underlying dura removed, and the experiment repeated. Drug solutions (1 mM CGP35348 and 0.5 mM NBQX/1 mM APV) were dissolved in Ringers solution, pre-warmed on a heating block, and applied in bulk to the dorsal OB without the use of agarose or coverslip. For higher concentration experiments, we utilized 2.5 mM NBQX/5 mM APV dissolved in Ringers immediately prior to use. We waited at least 10 min after drug application to allow for absorption and temperature equilibration across the tissue. For the experiments where we applied ionotropic glutamate blockers following GABA_B blockade, NBQX + APV was applied immediately after CGP35348 without an intervening control wash. Due to the extremely slow washout of CGP35348 in vivo, we considered CGP35348 to still be present during the subsequent NBQX + APV application.

In most experiments, odorants were presented as precise dilutions from saturated vapor (s.v.) in clean, desiccated air using a custom olfactometer under computer control, as described previously (Bozza et al., 2004; Economo et al., 2016). Odorants were presented for durations ranging from 2 to 8 s for most experiments, with inhalation rates ranging from 1 to 2 Hz, as specified; for experiments used to obtain inhalation-triggered averages (ITAs), odorant duration was 70 s with 0.25 Hz inhalations. Inter-trial intervals were 20 s for 2-s odorant presentations and 36 s for 4- or 8-s presentations. Clean air was passed across the nostrils in between trials to avoid contribution from extraneous odorants in the environment. Odorants were pre-diluted in solvent (1:10 or 1:25 in mineral oil or medium chain triglyceride oil) to allow for lower final concentrations and then diluted to concentrations ranging from 1% to 3% s.v. Relative increases in concentration were confirmed with a photoionization detector (miniRAE Lite, PGM-7300, RAE Systems) 3 cm away from the flow dilution port. Estimated final concentrations of odorants used ranged from 0.03 to 48 ppm, depending on vapor pressure and s.v. dilution (Table 1). For experiments testing a larger panel of 12 odorants (e.g., Figure 6), we used a novel olfactometer design that allowed for rapid switching between odorants with minimal cross-contamination (Burton et al., 2019). Here, odorants were presented for 3 s, in random order, using 10 s interstimulus intervals (Table 1) and repeated three times each. The odorant presentation was as described in a previous publication (Burton et al., 2019), using an eductor nozzle for additional mixing in a carrier stream of filtered air. The end of the eductor was placed 5–7 cm from the nose. With the configuration used, estimated dilutions of odorant were approximately 1.5% s.v.; odorants were prediluted to achieve relatively sparse activation of dorsal glomeruli (Burton et al., 2019). Estimated final concentrations ranged from 0.03 to 15 ppm.

Image analysis was performed using custom software written in MATLAB (Mathworks). For display, odorant response maps were displayed using ΔF values rather than ΔF/F to minimize noise from nonfluorescent regions. Activity maps were scaled as indicated in the figure and were kept to their original resolution (512 × 512 pixels) and smoothed using a Gaussian kernel with σ of 1 pixel. For time-series analysis, regions of interest (ROIs) were chosen manually based on the mean fluorescence image and were further refined based on odorant response ΔF maps, then all pixels averaged within an ROI to generate time-series for analysis. Time-series were computed and displayed as ΔF/F, with F defined as the mean fluorescence in the 1–2 s prior to odorant onset, and upsampled to 150 Hz for analysis using the MATLAB pchip function.

To analyze response patterns, traces were averaged across four-eight presentations of each odorant. Responses were classified as having significant excitatory and/or suppressive components (excitatory for Figures 1–6, suppressive only for Figure 2) as follows. First, each averaged response in the Predrug condition was z-scored using a baseline from 0.5 to 2 s before odorant onset, with z defined as the SD of the baseline period concatenated for each odorant response spanning each ROI. Peak excitation was measured as 95% of the maximum signal during the duration of the odorant presentation (4–8 s; when applicable, suppressive responses (i.e., Figure 2) were measured as the 15th percentile of all values in a time window from odorant onset to 500 ms after odorant offset. We then used a conservative criterion for the significance of z = ± 4 SD for identifying significant excitatory or suppressive responses for further analysis.

For analysis of inhalation-linked dynamics, in most experiments, responses to the first inhalation after odorant onset were analyzed, after averaging across 4–8 presentations. In other experiments (e.g., Figure 5) inhalation-triggered average (ITA) responses were generated by averaging each inhalation, repeated at 0.25 Hz, over a 70 s odorant presentation (17 inhalations averaged in total), as described previously (Short and Wachowiak, 2019). Onset and peak latencies, relative to the start of inhalation, were calculated from the ΔF/F time-series using the “risetime” function in MATLAB to fit a rising sigmoid to the baseline and peak ΔF/F levels and return the time to 10% of the rise from baseline to peak (onset latency) and the time to 90% of the rise from baseline to peak (peak latency). Latency values were calculated after upsampling but with no additional temporal filtering.

For analysis of odorant-evoked dynamics across multiple sniffs (i.e., Figure 6), responses to 4–8 presentations of odorant were averaged before analysis. Changes in response amplitude over time, or T₂ − T₁/T_max, were calculated as the difference in amplitude between the peak ΔF/F following the first (T₁) and the second-to-last (T₂) inhalation during the odorant presentation, divided by the maximum ΔF/F during the 4-s odor presentation.

For statistical analysis of drug effects, each experiment was treated as an independent observation, since drug treatment was applied across all ROIs in an experiment. For datasests with five or more experiments, summary statistics are reported as the mean (or median, for non-normally-distributed data) ± standard deviation (SD) effect across all imaged ROIs, and comparisons made on these mean/median values. For datasets with fewer than five experiments, mean (or median) effects across ROIs are reported for each experiment and, when possible given the number of ROIs, tests for significance performed on each experiment, using ROIs as the independent measure, and p-values adjusted for multiple comparisons based on the number of experiments per dataset. Statistical tests were performed in Origin (OriginLab Corp.). All measurement of response parameters was done using analysis code that was independent of treatment or comparison condition.

Data and analysis code underlying the results of this study are available from the corresponding author upon request.

Neurotransmitter release from OSNs can be modulated by presynaptic dopamine (D2) and GABA_B receptors, which modulate Ca²⁺ influx into the presynaptic terminal and subsequent release of the neurotransmitter (Ennis et al., 2001; Murphy et al., 2005; Wachowiak et al., 2005; Vaaga et al., 2017). GABA_B-mediated presynaptic inhibition is activated both tonically and via sensory-evoked, intraglomerular circuits that have been proposed to mediate gain control in vivo in a manner independent of inhalation frequency (McGann et al., 2005; Pírez and Wachowiak, 2008; Shao et al., 2009; Vaaga and Westbrook, 2017). Thus we first sought to characterize how GABA_B-mediated presynaptic inhibition modulates the gain and dynamics of the glutamatergic drive onto MT cells in vivo.

First, we confirmed effective blockade of GABA_B-mediated presynaptic inhibition by imaging Ca²⁺ influx into OSN terminals in dorsal OB glomeruli using the genetically-encoded calcium reporter GCaMP6f expressed exclusively in OSNs (Wachowiak et al., 2013; Rothermel and Wachowiak, 2014; Short and Wachowiak, 2019) and comparing responses before and after topical application of the GABA_B receptor antagonist CGP35348 (1 mM) to the dorsal OB, as done previously (Wachowiak et al., 2005; Brunert et al., 2016). Odorant was presented at suprathreshold concentrations (17–48 ppm; Table 1) to anesthetized mice using an artificial inhalation paradigm and evoked signals imaged from the dorsal OB under epifluorescence. Robust odorant-evoked presynaptic Ca²⁺ signals were apparent in multiple foci representing glomeruli across the dorsal OB (Figure 1A). Temporal response patterns during 2 Hz inhalation consisted of inhalation-linked transients apparent atop a sustained signal (Figure 1B). CGP35348 application caused an increase in the odorant-evoked Ca²⁺ signal that was apparent as early as the first inhalation of odorant and which persisted throughout the odorant presentation (Figure 1B). The magnitude of the increase varied across glomeruli and in different mice (Figure 1C), with a mean increase of 97% across the five mice tested (mean ± SD of median CGP35348/Predrug ratio per experiment: 1.97 ± 0.34; p = 0.003, t-test on median ratios with ratio = 1 as null hypothesis; df = 4). CGP35348 did not appear to change the dynamics of inhalation-linked GCaMP6f transients (Figure 1B, bottom). These results confirm previous findings using calcium-sensitive dyes (Wachowiak et al., 2005; Pírez and Wachowiak, 2008).

Next, we used the second-generation iGluSnFR, SF-iGluSnFR.A184V, to directly monitor the impact of GABA_B receptor blockade on odorant-evoked glutamate signaling onto MT cells. As expected from an earlier characterization using first-generation iGluSnFR (Brunert et al., 2016), CGP35348 led to an increase in the amplitude of inhalation-driven glutamate transients (Figures 1D,E). Here, in contrast to the GCaMP6f imaging results, we observed a difference in the magnitude of the effect of GABA_B receptor blockade on responses to the first and second inhalations of odorant. Overall, CGP35348 caused a significant increase in response to the second inhalation in all four mice but increased responses to the first inhalation of odorant in only three of the four mice (Figure 1F; see Figure 1 legend for statistical tests). The impact of CGP35348 on responses to the third and fourth inhalations appeared similar to that on the second inhalation (not shown).

However, GABA_B receptor blockade had little to no impact on the inhalation-linked dynamics of glutamatergic signaling onto MT cells. Neither onset latencies nor peak latencies (defined as time to 10% of peak and time to 90% of peak, respectively) of the iGluSnFR transient evoked by the first inhalation of odorant changed substantially after CGP35348 application (Δ onset latency, −0.3 ± 2.6 ms; Δ peak latency, 3.1 ± 4.2 ms; mean ± SD of median values from four mice; Figure 1G). While two of the four mice showed a significant change in onset or peak latency (see Figure 1G legend), the magnitude of the change in these experiments (<10 ms) was negligible compared to the range of onset and peak latencies seen across different glomeruli in the same preparation, which ranges over a span of 200–250 ms (Moran et al., 2021). These results confirm that GABA_B-mediated inhibition regulates the strength of glutamatergic signaling from presynaptic sources onto MT cells in vivo (McGann et al., 2005; Wachowiak et al., 2005; Brunert et al., 2016), and suggest that presynaptic inhibition contributes little to the inhalation-linked timing of these signals.

To assess the impact of this GABA_B-mediated modulation of glutamatergic drive on postsynaptic responses of MT cells in vivo, we returned to Ca²⁺ imaging, expressing GCaMP6f in MT cells and comparing odorant-evoked responses before and after CGP35348 application. GCaMP6f signals were imaged from the apical tufts of MT cells across the glomerular layer using two-photon imaging, as done previously (Economo et al., 2016; Short and Wachowiak, 2019; Figures 2A,B). CGP35348 had variable effects on MT cell Ca²⁺ responses in different glomeruli, with increased responses in some glomeruli and decreased responses in others (Figures 2B,C). Suppressive responses reflected as a decrease in GCaMP6f fluorescence and presumably reflecting inhibition of ongoing MT cell spiking (Economo et al., 2016), persisted after application of CGP35348 (Figures 2B,C) indicating that MT cell suppression is largely not mediated by presynaptic inhibition of glutamatergic drive from OSNs. Overall, there was no significant net change in response magnitude, either for peak responses to the first inhalation of odorant, or averaged across the duration of odorant presentation (CGP35348/Predrug ratios, first inhalation of odorant (median [quartile range] of medians: 0.75 [0.64 ± 1.05], p = 0.11, one-sample Wilcoxon Signed Rank test, n = 12; odor duration: 0.91 [0.77 ± 1.33], p = 0.85). Thus, increasing the magnitude of excitatory input to the glomerulus by the enhanced release of glutamate from OSN terminals does not uniformly translate into increased excitation of MT cells.

Sensory-evoked glutamatergic input onto MT cells can arise from OSNs or from multisynaptic excitation involving dendritic glutamate release from ET cells and, possibly, superficial tufted cells (sTCs; Hayar et al., 2004a; De Saint Jan et al., 2009; Gire et al., 2012; Vaaga and Westbrook, 2016; Sun et al., 2020). In addition, inhibitory circuits can modulate multisynaptic glutamate signaling (Gire and Schoppa, 2009; Shao et al., 2012; Gire et al., 2019). Glomerular glutamate signals may also arise from glutamate release by MT cell dendrites (Isaacson, 1999; Najac et al., 2015), and thus at least partially reflect MT cell excitation itself. To isolate the direct contribution of OSN inputs to MT cell excitation dynamics, we compared evoked glutamate dynamics before and after pharmacological blockade of postsynaptic activity with ionotropic glutamate receptor antagonists (Gurden et al., 2006; Pírez and Wachowiak, 2008; Lecoq et al., 2009). Because iGluSnFRs are insensitive to these antagonists (Marvin et al., 2013, 2018), we reasoned that this approach would largely prevent OSN-driven excitation of postsynaptic neurons and allow us to image evoked MT cell glutamate signals arising solely from OSN inputs.

To confirm our ability to block ionotropic glutamate receptors with the in vivo drug application protocol, we first continued the GCaMP6f imaging from MT cells in the same mice used for the GABA_B blockade experiments, applying a cocktail of APV+NBQX (1 mM/0.5 mM) following the initial CGP35348 treatment (n = 4 mice). APV+NBQX sharply reduced odorant-evoked GCaMP6f signals, such that distinct inhalation-linked transients were eliminated in most glomeruli and replaced with a weak tonic glutamate signal (Figure 2B); this signal might reflect modulation by metabotropic glutamate receptors (Matsumoto et al., 2009; Dong and Ennis, 2014), which we did not attempt to block in these experiments. Excitatory responses to both the first inhalation and over the duration of odorant presentation were reduced to 36 ± 20% (p = 3.5 × 10⁻¹⁰, n = 29 ROIs from four mice, paired t-test); and 36 ± 15% (p = 1.2 × 10⁻⁶, n = 23 ROIs, paired Wilcoxon Signed Ranks) respectively, of their baseline values (Figures 2B,C; see Figure legend for summary statistics). Suppressive responses were eliminated after APV+NBQX application and were replaced with a slow, tonic increase (Figure 2B).

We next tested the ability of APV+NBQX to block activation of cholecystokinin (CCK)-expressing sTCs, as these, like ET cells, are strongly driven by monosynaptic OSN input (Isaacson, 1999; Najac et al., 2015). We imaged activation of CCK+ sTCs at their dendrites in CCK-IRES-Cre mice crossed to a Cre-dependent GCaMP6f reporter line (n = 3 experiments from two mice, see “Materials and Methods” section). We first blocked GABA_B-mediated presynaptic inhibition with CGP35348 to further enhance transmitter release from OSNs. As expected, CGP35348 led to a modest increase in the peak CCK+ GCaMP6f response, with median increases of 33, 9, and 24%, respectively, in the three experiments (Figures 3A–E). Subsequent application of APV+NBQX (1 mM/0.5 mM) had mixed effects in different glomeruli, with responses completely or nearly eliminated in 8 of 14 glomeruli (97 ± 3% median reduction in peak response to 1st inhalation) but only partially reduced in the remaining six (38 ± 16% median reduction; Figures 3A–E). In these six glomeruli, responses to successive inhalations after odorant onset were reduced even further or eliminated (e.g., Figure 3E). We performed additional experiments using a 5× higher concentration of antagonist (5 mM APV/2.5 mM NBQX) and without preapplication of CGP35348. In these, APV+NBQX completely blocked postsynaptic activation in 11 of 13 glomeruli (from three experiments), with strongly reduced responses (66 and 75% reduction) in the remaining two (Figure 3F).

Importantly, in glomeruli with persistent (but weakened) responses after APV+NBQX application, inhalation-linked Ca²⁺ transients were significantly delayed (Figures 3B,D), with a mean increase in onset and peak latency of 255 ± 79 ms and 257 ± 51 ms, respectively (Figure 3G). There was no change in latencies after CGP35348 application alone (Figure 3G; see Figure legend for summary statistics). An explanation for this substantial delay is that, in glomeruli receiving the strongest sensory input, glutamate concentrations are eventually able to overcome APV+NBQX blockade sufficiently to trigger spike bursts in sTCs. Overall, these results demonstrate that APV+NBQX blocks postsynaptic activation in glomeruli receiving all but the strongest OSN inputs, and even in those glomeruli, substantially weakens and delays activation of monosynaptically-driven sTCs.

We next used in vivo pharmacology to test the contribution of mono- vs. di- or polysynaptic glutamate signaling to inhalation-linked glutamate transients onto MT cells. We focused on MT cells projecting to the piriform cortex (pcMTs) as the multisynaptic excitatory drive is thought to predominately impact this subpopulation (Fukunaga et al., 2012; Gire et al., 2012; Igarashi et al., 2012). We selectively expressed SF-iGluSnFR.A184V in pcMTs via retrograde viral infection by targeting virus injections to the anterior piriform cortex of Tbet-Cre mice (n = 2 mice), as described previously (Rothermel et al., 2013). We have established in a recent study that odorant- and inhalation-evoked glutamate transients in pcMTs are indistinguishable from those measured from sTCs or from the general MT cell population (Moran et al., 2021). As expected, inhalation-linked glutamate transients on pcMTs were increased by CGP35348 (Figures 4A,B), with mean increases of 35% and 90% in the two mice, and no impact on inhalation-linked glutamate dynamics (Δ onset and peak latencies, mean ± SD, paired t-tests, n = 10 ROIs pooled across two mice: Δ onset, 8 ± 11 ms, p = 0.07; Δ peak, 8 ± 15 ms, p = 0.2). Surprisingly, subsequent application of APV+NBQX (1 mM/0.5 mM) had no significant impact on response amplitudes (Figures 4A–C; see Figure 4C legend for summary statistics). iGluR blockade did cause a very small, but statistically significant, increase in the latency of the inhalation-linked glutamate transient (Δ onset latency, 13 ± 12 ms, p = 0.01; Δ peak, 34 ± 16 ms, p = 1.2 × 10⁻⁴; Figure 4D); however, the magnitude of this change was very small relative to the overall dynamics of the inhalation-linked transient (Figure 4B).

Multisynaptic glutamatergic excitation may preferentially drive MT cell responses to weak inputs (Najac et al., 2011; Vaaga and Westbrook, 2016; Gire et al., 2019), and the odorant concentrations used in the preceding experiments were relatively high (~20–50 ppm; Table 1). Removal of presynaptic inhibition with CGP35348 may also enhance OSN input strength to physiologically excessive levels. Thus, we next tested the impact of glutamate receptor blockade on inhalation-linked transients in pcMT cell responses to odorants presented at 10–100× lower concentrations (0.1–5 ppm), using a higher concentration of APV+NBQX (5 mM/2.5 mM) and omitting CGP35348. In addition, to allow for greater sensitivity of glutamate detection and to focus on inhalation-linked dynamics of the glutamate signal, we used the higher-affinity SF-iGluSnFR.A184S and modified the artificial inhalation protocol to generate averaged responses to a single inhalation repeated at 0.25 Hz (Figure 5A). This protocol yielded inhalation-triggered average waveforms that showed substantial variation in onset and peak latency as well as duration, as we have reported previously (Short and Wachowiak, 2019; Moran et al., 2021). Finally, in a subset of mice (n = 2), we used dual-color imaging to simultaneously image presynaptic glutamate transients and postsynaptic Ca²⁺ signals in the apical glomerular tufts of overlapping MT cell populations (Figure 5B), using Tbet-Cre mice crossed with the Thy1-jRGeCO1a reporter line (Dana et al., 2018; see “Materials and Methods” section).

In the two dual-color imaging preparations, APV+NBQX completely blocked postsynaptic MT cell activation in 7 of 11 glomeruli analyzed, as reflected in the peak jRGECO1a Ca²⁺ signal imaged from the glomerular layer, with a reduction to 9, 12, 12 and 34% of pre-drug levels in the remaining four glomeruli. APV+NBQX had minimal impact on the glutamate transients measured in these same glomeruli (Figure 5C). Overall, across all aPC-injected SF-iGluSnFR.A184V mice in this dataset (n = 6), the impact of APV+NBQX on the magnitude of inhalation-linked glutamate transients was modest and variable, causing a statistically significant reduction in three of six experiments (see Figure 4 legend) and a 27 ± 26% reduction in peak inhalation-triggered average (ITA) amplitude across all experiments (mean ± SD of mean ratios, p = 0.06, t-test, n = 6; Figure 5D). However, APV+NBQX did not impact the dynamics of inhalation-linked glutamate responses, with no significant change in peak times across the dataset (mean ± SD of Δ peak latencies, 9 ± 21 ms, p = 0.34, paired t-test, n = 6; Figure 5E). Overall, these results suggest that multisynaptic pathways modestly contribute to the strength of glutamatergic excitation of MT cells in response to low-intensity stimulation, but do not uniquely shape the inhalation-linked dynamics of this excitation.

Finally, we sought to further examine the role of direct vs. multisynaptic excitation in shaping the dynamics of glutamatergic input to MT cells during repeated odorant sampling. We have shown previously that glutamate signals on MT cells show diverse temporal patterns across repeated inhalations of the odorant that include adaptation, facilitation, and even suppression depending on the odorant identity and concentration (Moran et al., 2021). To capture this diversity, we used a 12-odorant panel (delivered concentrations ranging from 0.03 to 15 ppm) and 3 Hz artificial inhalation, comparing response patterns before and after application of APV+NBQX (1 mM/0.5 mM). This approach yielded significant odorant responses in 83 glomerulus-odorant pairs across three mice and included diverse excitatory response patterns (Figure 6A). Consistent with the earlier experiments, APV+NBQX application alone had little overall impact on the magnitude of evoked glutamate signals: while there was variability in the effect for certain glomerulus-odor pairs, neither mean nor peak excitatory response showed a significant overall change after APV+NBQX application (paired t-tests, p = 0.147 for mean amplitude, p = 0.605 for peak amplitude, n = 83; Figure 6B).

APV+NBQX did, however, impact the temporal patterns of the glutamate signal for a subset of responses. In particular, application of APV+NBQX alone tended to reduce adaptation that occurred following the initial inhalation of odorant (e.g., vinyl butyrate response; Figure 6A), or it enhanced long-lasting, tonic-type responses (e.g., ethyl butyrate response; Figure 6A). We used a simple metric, T₂ − T₁/T_max, (Moran et al., 2021) to quantify and compare these adapting or facilitating dynamics before and after drug application (Figure 6C, see also “Materials and Methods” section). APV+NBQX caused an overall increase in T₂ − T₁/T_max values (Figure 6C), with significant increases seen in each of the three preparations. These results suggest that multisynaptic pathways contribute to the slower temporal patterning of MT cell activity by modestly suppressing the excitatory drive of MT cells across repeated samples of odorant.