Hyperpolarized States

Overall, the proportion of MPSO-exhibiting cells increased when the membrane potential was hyperpolarized. In H3, the oscillating cell population represented 47.8% of the cells (n = 11/23), whereas it represented only 27.1% of the cells (n = 13/48) at the resting membrane potential. In addition, the membrane potential hyperpolarization led to an increase in the proportion of cells exhibiting an MPSO+.

Depolarized States

In contrast to hyperpolarization, the membrane potential depolarization did not appear to affect the frequency of the MPSO occurrence; the proportion of oscillating cells was 30.8% (n = 4/13) in the depolarized state (D2) and 27.1% (n = 13/48) at the resting membrane potential (Figure 5A). However, an examination of the MPSO evolution at the single-cell level indicated that the membrane potential depolarization could also trigger an MPSO (Figure 6A, hatching). In the depolarized state, the MPSO+ type disappeared in favor of the MPSO- and MPSOsym types (Figure 5A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Unmasking of silent membrane potential slow oscillations.

A) A synthetic view of the membrane potential slow oscillations and respiration-related discharge patterns from 17 cells in the H3, H2, RMP and D2 excitability states. Each line corresponds to a single cell. Of these 17 cells, 9 were also examined with odor stimulation and are indicated with the same numbers (between 1 and 36) as those analyzed in Figure 2. The remaining 8 cells were numbered from 37 to 44. Of the 17 cells, 6 were not depolarized. The activity patterns are indicated by color as follows: blue indicates a complex synchronized pattern (Sc), red indicates an excitatory-simple-synchronized pattern (S+), yellow indicates a non-synchronized pattern (NoS) and white represents null activity (Null). The different types of hatch marks represent the presence and types of membrane potential; the right hatch marks indicate positive membrane potential slow oscillations (MPSO+), the left hatch marks indicate negative membrane potential slow oscillations (MPSO-), the cross hatches indicate symmetric membrane potential slow oscillations (MPSOsym) and a lack of hatches indicates cells without oscillations (NoMPSO). B) Computational simulation of intracellular activity with rhythmic inputs at the resting membrane potential (RMP-like state, 2) and during negative current injection (hyperpolarized-like state, 3). 1) Schematic representation of the simulation configuration. The synaptic currents corresponding to a mixture of excitation from olfactory receptor neurons and inhibition from periglomerular cells were injected into the tuft. The simulation parameter details are provided in the Material and Methods section and in Supporting Information S1. 2/Simulated recording of a mitral soma model at the resting membrane potential indicating no intracellular oscillations. 3/Simulated recording of a mitral soma model with an additional negative current. This signal showed an MPSO. In 2 and 3: top-left: intracellular activity simulation; bottom-left: simulated input; top-right: Fourier power spectrum plot of the membrane potential and respiration-triggered membrane potential average; bottom-right: the resulting respiration-triggered histogram of action potentials. The vertical dotted line indicates the inspiration/expiration transition.

https://doi.org/10.1371/journal.pone.0043964.g006

In addition, we analyzed the effects of intracellular excitability changes on discharge activity. Overall, we observed that the proportions of cells exhibiting S+, S- and Sc were only slightly influenced by excitability changes (Figure 5B). We found that the S+ pattern was a particularly stable pattern because it persisted regardless of the excitability state, except in H4 (Figure 5B, Figure 6A). Furthermore, the Sc and S- patterns were rarely present regardless of the excitability level. Indeed, the Sc pattern was observed only in H1 and at the resting membrane potential in a small proportion of cells (H1∶5.3%, n = 1/19; resting membrane potential: 14.6%, n = 7/48; Figure 5B). The S- pattern appeared only twice (once at the resting membrane potential and once in D2) (Figure 5B). As expected, the more hyperpolarized the cell was, the more frequent the occurrence of the null pattern (Figure 5B).

These results suggest that intracellular excitability changes profoundly affected the MPSOs, whereas the respiration-synchronized discharge patterns were only slightly sensitive to the intracellular excitability level. Importantly, this weak effect was not due to a general lack of effect on the generation of action potentials because, as expected, the cells’ firing rates varied with the intracellular excitability state (Figure 5C). Indeed, the membrane potential depolarization that was caused by positive current injection led to an increased mean firing rate, whereas progressive hyperpolarization caused by injection with increasing amounts of negative current induced a significant, progressive decrease in the discharge activity (Mann-Whitney test: p<0.001 between the resting membrane potential and the three most hyperpolarized states).

We then combined these new data (i.e., the results obtained from changes in the excitability level) into a matrix in which the MPSO type and spike pattern were represented for each of the 17 cells that exhibited a NoMPSO at the resting membrane potential as recorded under the 4 excitability states (Figure 6A). This representation revealed that all of the cells manifesting a non-congruent association (a respiration-synchronized discharge pattern associated with a NoMPSO) at the resting membrane potential developed an MPSO when the intracellular state was modified (Figure 6A, cells 24, 31, 37, 38 and 39). Conversely, all of the cells exhibiting a NoS pattern did not exhibit an MPSO when they were hyperpolarized or depolarized (Figure 6A, cells 33, 35, 36, 40, 41, 42 and 43). Indeed, with the exception of one cell, these cells maintained a NoS/NoMPSO combination or acquired a null pattern associated with a NoMPSO (Figure 6A). Taken together, these results indicate that the following are true: 1) the MPSO was actually masked at the resting membrane potential in all of the cases in which we observed an unexpected combination of a synchronized spiking pattern with a NoMPSO and 2) an MPSO is never actually expressed in combination with a NoS pattern.

We then attempted to strengthen our conclusion by using a modeling approach, which allowed us to explore how a silent oscillation may be generated and how it can induce a synchronized discharge pattern (see the Materials and Methods section and Supporting Information S1 for details). First, we observed that when the cell membrane potential is close to the reversal potential of combined excitatory and inhibitory synaptic inputs, assuming that excitation and inhibition have equal time course, the net synaptic input current is null, which leads to a silent oscillation, characterized by oscillating conductance inputs, and a flat membrane potential. Such a silent oscillation can be unmasked by changing the cell excitability. Second, assuming that larger synaptic conductances are associated with larger conductance noise and thus larger membrane potential noise, we used a one-compartment integrate-and-fire model to demonstrate that the amplitude of membrane potential noise can be oscillatory and entrains a cell rhythmic discharge while the average membrane potential is still relatively flat. We confirmed this result using a modeled mitral cell (see the Materials and Methods section and Supporting Information S1 for the model’s details) in both a resting-membrane-potential-like state and a hyperpolarized-like state (Figure 6B). We examined the resulting activity of the modeled cell (MPSO and spiking activity) when the cell was subjected to rhythmic input in the tuft (Figure 6B1). As in the simple model, we used a combined synaptic reversal potential that was similar to the cell resting potential. We then observed that, in the resting membrane potential-like state, a 2-Hz rhythmic input resulted in synchronized discharge activity, even in the absence of an MPSO (Figure 6B2). In the hyperpolarized-like state, the same rhythmic input condition resulted in the appearance of an MPSO and decrease in action potential occurrence (Figure 6B3). Note that the injection of oscillatory conductances only in the tuft was simpler to fulfill our hypothesis of equal time course for excitation and inhibition, in particular if we take into account the slow time constant observed for lateral dendrite inhibition (50 to 100 ms). This suggests that silent MPSO generation mechanism is mainly located in mitral cell tuft.

Taken together, our experimental and modeling results demonstrated that cells without an observable MPSO but with a respiratory-synchronized discharge pattern likely express a silent oscillation (silent MPSO).

As a final analysis, we included the silent oscillations revealed by intracellular current injection in matrices similar to those presented in Figure 3 which already showed that a NoS pattern was always exhibited in combination with a NoMPSO type. Figure 7 now shows that each type of S pattern is always associated with the occurrence of an MPSO. In addition, Figure 7 indicates that the S+ pattern was never associated with an MPSO- and the Sc pattern was never associated with an MPSO+. Finally, we observed that olfactory stimulation increased the probability (from 38.5% to 50%) that the M/T cells would exhibit a specific association between a respiration-synchronized discharge pattern and an MPSO.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. Exclusive associations between membrane potential slow oscillations and respiration-related discharge patterns.

The cell percentages exhibiting an association between each type of membrane potential slow oscillation (including silent membrane potential slow oscillations, silent MPSOs) and each type of respiration-related discharge pattern are shown for the control (n = 26) and odor periods (n = 30). The cells that were used to create theses matrices are the same as those used in Figures 2 and 3. For the cells with NoMPSOs or silent MPSOs, only those that were tested using the intracellular excitability protocol were conserved. The cell percentages are represented using a grayscale code, in which white indicates a null percentage and black indicates 30% of the entire population. NoMPSO: no oscillation; silent MPSO: silent membrane potential slow oscillation; MPSO+: positive membrane potential slow oscillation; MPSO-: negative membrane potential slow oscillation; MPSOsym: symmetric membrane potential slow oscillation; NoS: non-synchronized pattern, S+: excitatory-simple-synchronized pattern; S-: suppressive-simple-synchronized pattern; Sc: complex synchronized pattern; Null: null activity.

https://doi.org/10.1371/journal.pone.0043964.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Schematic representation of our data interpretation.

Case A: A mitral/tufted cell related to a glomerular unit receiving no input. This cell exhibits neither a membrane potential slow oscillation (NoMPSO) nor a respiration-synchronized discharge pattern (NoS). Gl: glomerulus; M: mitral cell; Olf Cx: olfactory cortex. Case B: A mitral/tufted cell related to a glomerular unit receiving a strong respiration-related input. This cell exhibits a positive membrane potential slow oscillation (MPSO+) that exhibits an excitatory-simple-synchronized pattern (S+). This S+ spike pattern induces, via granular activation, lateral inhibition of the mitral/tufted cells connected to neighboring glomeruli. Therefore, this cell exhibits a negative membrane potential slow oscillation (MPSO-), which is associated with a suppressive-simple synchronized pattern (S-). Gr: granule cell. Case C: A mitral/tufted cell related to a glomerular unit receiving a weak respiration-related input. Because of the weak peripheral input, this cell is strongly influenced by the cellular excitability state, which is dependent on the characteristics of the animal’s internal state (such as neuromodulation and attention). The synergistic effect of the peripheral input and the intracellular excitability can result in several different types of membrane potential slow oscillations (MPSO+, -, sym or silent), which can induce different respiration-synchronized discharge patterns.

https://doi.org/10.1371/journal.pone.0043964.g008

In summary, we found that the cells exhibiting an unsynchronized discharge pattern never exhibited an MPSO, whereas the cells with a respiration-synchronized discharge pattern always exhibited an MPSO. Some spiking pattern/MPSO association have never been observed such as S+/MSPO-, S−/MPSO+ and Sc/MPSO+.

The Origin of the Olfactory Bulb Slow Rhythm

The origin of the slow rhythm has been primarily described as peripheral, occurring via the rhythmic stimulation of the olfactory epithelium at each inspiration. Indeed, the local field potential signal has been shown to not exhibit respiration modulation when air flow does not penetrate the nasal cavity [28] or in the absence of mechanical stimulation of the olfactory neuroreceptor [29]. Similarly, the respiratory patterning of the olfactory bulb single-unit activity reflects the phasic stimulation of the olfactory receptors because slow patterning was not observed during continuous odor stimulation [30] or in the absence of air stimulation [28]. Our study demonstrated that the occurrence of respiration-synchronized discharge patterns is enhanced by odor stimulation, which is consistent with the results of previous extracellular studies [13], [31]. Interestingly, we have demonstrated for the first time that these patterns were poorly influenced by intracellular excitability changes, confirming that respiration-related spike activity primarily originates from afferent input activation. However, the origin of MPSOs is not well established. Several studies have proposed an exclusively peripheral origin because MPSOs disappear when no airflow exists in the nasal cavity [22], [25]. However, we observed that odor stimulation modulated the occurrence of MPSOs and improved the synchronization of MPSOs with slow oscillations of the local field potential (data not shown), indicating a common peripheral input.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 10. Characterization of membrane potential slow oscillations.

A) Intracellular recordings of 1) a cell with a positive oscillation (MPSO+), 2) a cell with a negative oscillation (MPSO -) and 3) a cell without oscillation (NoMPSO). Top: intracellular signal; bottom: respiratory signal with expiration (Expi) and inspiration (Inspi) separated by a dotted line. The gray circles indicate the inspiration/expiration transitions. The action potentials have been truncated to improve membrane potential visibility. B) Membrane potential Fourier power spectrum analysis. Cells with membrane potential slow oscillations (1 and 2) exhibited a higher power in the 2 Hz frequency range (corresponding to the respiratory rhythm) than a cell without oscillation (3). C) Autocorrelograms of the membrane potential. Cells with oscillations (1 and 2) present an oscillating autocorrelogram, while a cell without oscillation (3) shows a flat autocorrelogram. D) Respiration-triggered membrane potential average. A positive membrane potential slow oscillation was characterized by an up-point proportion <42% (1), whereas a negative oscillation manifested an up-point proportion >58% (2) (please refer to the text for details). In the cells without oscillations, the plots were relatively flat. The inspiration/expiration transition is marked by a vertical dotted line, and the mean membrane potential is represented by a horizontal dotted line. E) Histograms of membrane potential distribution. The mean membrane potential is represented by a vertical dotted line.

https://doi.org/10.1371/journal.pone.0043964.g010

The peripheral input is not likely the only contributing factor to the slow rhythm; various voltage-dependent conductances also appear to be involved in MPSO expression (See Figure S1). Calcium and sodium fluctuations have been shown to reflect slow membrane potential dynamics, and the blockade of some calcium or sodium voltage-dependent conductances result in decreased MPSO amplitude [15], [32]–[34]. In addition, the blockade of some potassium voltage-dependent conductances in olfactory bulb slices has been shown to delay the falling phase of oscillations [33], [35]. Our observation that MPSOs are highly sensitive to excitability changes supports the assumption that MPSOs are also likely shaped by intrinsic components, including voltage-dependent conductances.

The Relationship between Respiration-related Discharge Patterns and Membrane Potential Slow Oscillations

Our study is the first to dissect the relationships between the various respiration-related discharge patterns of M/T cells and the different types of slow oscillations of their membrane potentials. We demonstrated that the cells that lacked a synchronized discharge pattern never exhibited an MPSO; additionally, S- patterns were always combined with an MPSO- (or MPSOsym), whereas cells with S+ patterns never exhibited an MPSO-. However, some MPSO/discharge activity associations were rather unexpected; for example, the observation of a rhythmic discharge activity, synchronized to the respiratory signal, superimposed onto a non-oscillating membrane potential. We can explain such a discrepancy by our observation that MPSOs can be silenced in some excitability states and revealed in others.

The response patterns of M/T cells are closely correlated with the spatial patterns of glomerular activation. Luo and Katz [27] have demonstrated that M/T cells are robustly excited when their corresponding glomeruli are activated by an odorant but are strongly inhibited when neighboring glomeruli are activated, showing a clear center-surround organization. We used this observation to propose the following interpretation of our data (Figure 8). We assumed that the association between MPSO and respiration-related discharge pattern depends on the cell’s location within the spatial pattern of the odor-activated glomeruli. An M/T cell expressing NoMPSO and a spike pattern unrelated to respiration is likely connected to a non-activated glomerulus (either by pure airflow or odor) (Figure 8A). In contrast, an M/T cell exhibiting an MPSO is likely related (directly or indirectly) to a rhythmically activated glomerulus (Figure 8B, C). A cell located in a strongly activated area receives a strong, direct activating input if the cell is connected to a center-activated glomerulus (Figure 8B, left glomerulus) or an indirect inhibiting input if the cell is related to the surrounding glomerulus (Figure 8B, right glomerulus). In our model, strong, direct inputs corresponding to excitatory responses were characterized by the presence of an MPSO+. Observations of membrane potential values of MPSO+ induced by odor stimulation in relation to membrane potential values recorded in the control period supported our model. Indeed, the up-value of the MPSO+ is more depolarized than that of the mean membrane potential value of the NoMPSO recorded in the control period, indicating strong excitation (Figure S1). This MPSO type has been proposed to be initiated by the activation of AMPA/kainate, NMDA and metabotropic glutamate receptors [15], [22], [25], [32], [36], [37]. The observation that neuroreceptor activation reaches its maximum at the end of inspiration [38] could explain why MPSO+ peaks after the inspiration/expiration transition (data not shown). We additionally demonstrated that MPSO+ imposes an S+ pattern, which constrains the spike discharge to the restricted time window of the inspiration/expiration transition epoch [16], [39]. In this case, the strong afferent input overrides the influence of the cellular excitability, and the spike pattern does not depend on the excitability state. Propagation of the S+ spike discharge along a lateral dendrite may facilitate the rhythmic activation of connected granule cells [16], [25], [40], [41], which could thus induce the rhythmic inhibition of the M/T cells in neighboring glomeruli. This rhythmic inhibition would likely be reflected by an MPSO- (see Figure S1) and would impose an S- pattern on these M/T cells, resulting in surrounding inhibition [27]. Our model is also supported by the demonstration that the inhibitory discharge responses of M/T cells depend on granule cell-mediated lateral inhibition [42].

In the glomerular maps described by Luo and Katz [27], some glomerular columns are more weakly activated. In this case, M/T cells projecting into such glomeruli are likely more sensitive to intracellular excitability changes (Figure 8C). In vivo, the cellular excitability state of M/T cells largely depends on an animal’s internal state because neuromodulator systems, such as cholinergic, noradrenergic or serotonergic inputs, act directly or indirectly on M/T cells [43]. According to our results, an M/T cell connected to a weakly activated glomerulus should exhibit a MPSO type which depends on its excitability state. Indeed, we demonstrated that the artificial modification of the intracellular excitability state unmasked the presence of a silent MPSO and led to the modification of the exhibited MPSO type (notably, MPSO+ was promoted by hyperpolarization). We also analyzed the effect of the intracellular excitability state using a mitral cell model. This model confirmed and complemented our experimental results by demonstrating the existence of silent MPSOs, which were predominant at a specific excitability state. Taken together, these data suggest that both the peripheral input strength and the intracellular excitability state determine the MPSO type exhibited by an M/T cell, and the MPSO type constrains the expression of a respiration-related discharge pattern.

Conclusion

Our M/T cell intracellular recordings obtained from freely breathing anesthetized rats allowed us to demonstrate, for the first time, the specific sensitivity of MPSOs and respiration-related discharge patterns to odor stimulation and intracellular excitability changes. We explained the specific and narrow relationships that exist between these two activity levels. In addition, these specific arrangements can be integrated into the spatio-temporal coding scheme of olfactory processing. We propose that, via dual modulation by input and network excitability, the rough glomerular activation map can be refined as a function of the animal's state.

Animal Preparation

Male Wistar rats (240 to 350 g, Janvier, Le Genest-Saint-Isle, France) were anesthetized with an intraperitoneal injection of urethane (1.5 g/kg). Local field potential oscillations were used to monitor the depth of anesthesia, and supplemental doses were delivered when necessary. Each animal was immobilized in a stereotaxic apparatus. Respiration was unimpeded and was measured with a homemade bidirectional pressure sensor, which was placed in front of the right nostril, as described by Roux et al. [44]. This sensor measured respiratory airflow changes as a depression during inspiration and a super-pressure during expiration. Viscous xylocaine (2%) was applied to the skin before surgery, and a craniotomy was performed above the dorsal surface of the left olfactory bulb. The dura was removed, and the olfactory bulb was protected by a 3% agar Ringer’s lactate solution. The cisterna magna was then drained. The animal’s temperature was maintained at 37°C with a heating blanket (Harvard Apparatus, Holliston, MA, USA). At the end of the experiments, the rats were euthanized by a lethal intraperitoneal injection of pentobarbital.

All of the experiments were conducted in strict accordance with the European Community Council directive of November 24, 1986 (86/609/EEC) and those of the French Ethical Committee and French Legislation. These experiments were approved by the Ethics Committee of the Université de Lyon 1 (authorization number: BH2008–08).

In vivo Electrophysiological Recordings

Intracellular recordings were performed in the lateral or medial area of the left olfactory bulb using borosilicate glass micropipettes (o.d. = 1.5 mm; i.d. = 0.86 mm; Harvard Apparatus, Holliston, MA, USA) pulled with a horizontal puller (model P-97, Sutter Instruments, Novato, CA, USA). The micropipettes were filled with a solution of 2 M potassium acetate. To allow the morphological identification, we also included 2% biocytin (Sigma, Saint Louis, MO, USA) in the micropipette solution during some of the recordings. The electrode resistances ranged from 55 to 180 MΩ. The electrophysiological signal was amplified and low-pass filtered on-line at 10 KHz by an intracellular amplifier (Axoclamp 2B, Axon Instruments, Foster City, CA, USA). The signal was then digitized at 20 KHz (PCI-DAS 1602/16, Measurement Computing, Norton, CA, USA) and stored on a personal computer using the ELPHY software program (Sadoc G., Centre National de la Recherche Scientifique). During the experiment, the resting membrane potential was recorded with no current injection in 33/48 cells and with small stabilizing currents (−0.14±0.03 nA, mean ± SEM) in 15/48 cells. Intracellular excitability changes were produced by progressive direct current (DC) injections with an intracellular amplifier. The membrane potential was hyperpolarized at 4 distinct levels via negative current injections. The first level, H1, was obtained via a −0.14±0.03 nA (n = 19) DC injection and caused a 20% decrease in discharge activity relative to the spike discharge activity that was recorded at the resting membrane potential. The second level (H2) corresponded to the potential level at which the discharge activity decreased by 50% via a −0.22±0.04 nA DC injection (n = 24). The third and fourth hyperpolarized levels (H3 and H4) were defined as the potential levels that resulted in a level of discharge activity in which few or no action potentials were observed, respectively. These potential levels were obtained using current injections (for H3: −0.30±0.04 nA, n = 23; for H4: −0.41±0.07 nA, n = 10) carrying membrane potentials just below (H3) or well below (H4) the action potential threshold. Two depolarized states were also used to analyze the intracellular excitability effects. D1 corresponded to the level at which a 10% increase in discharge activity was induced and was evoked by a positive DC injection of 0.11±0.02 nA (n = 10). D2 corresponded to a 0.18±0.03 nA injection (n = 13) and induced a 30% increase in discharge activity. The input resistance of the cells was measured by applying brief hyperpolarizing current pulses with a programmable stimulator (Master-8, A.M.P. Instruments LTD, Jerusalem, Israel). Before the impalements were performed, the electrode tip potential was controlled by measuring the DC offset of the electrode in the extracellular medium. This measurement was performed again at the end of the experiment. For all of the cells analyzed, the electrode tip potential was not modified between these two measurements.

The local field potential recordings were performed simultaneously using silicon probes (Neuronexus Technology, Ann Arbor, MI, USA) placed in medial or lateral positions in the left olfactory bulb at a depth of approximately 1500 µm. The broadband signal (0.1 Hz to 5 KHz) was amplified with a homemade amplifier; it was digitized at 20 KHz (PCI-DAS 1602/16, Measurement Computing, Norton, USA) and stored on a personal computer.

Odor Stimulation and Acquisition Protocol

Odor was delivered in front of the left nostril using a homemade olfactometer linked to a respiratory sensor; odor onset was triggered at the end of an expiration period. Three different odorant molecules, isoamyl acetate, heptanone and ethyl pentanoate, were used separately (Sigma-Aldrich, Saint Louis, MO, USA). A single odor was delivered at 9% of the saturated vapor pressure, and 2 consecutive stimulations were separated by at least 1.5 min. The recording session consisted of 4 seconds of pre-stimulus activity, 3 seconds of odor-evoked activity and 5 seconds of post-stimulus activity (n = 97 recordings with odor stimulation). Fifty 12-second recordings were performed without odor stimulation.

Histology

Biocytin was injected into some of the recorded cells using 0.2 nA depolarizing current pulses of 500 ms every 1 second for 10 minutes. At least 1 hour after the biocytin injection, the rat was transcardially perfused with Ringer’s solution, followed by a 4% paraformaldehyde solution. After the animal was decapitated, the olfactory bulb was removed and placed in cold 30% sucrose in 0.1 M phosphate buffer. After the freezing process was complete, 80-µm-thick coronal sections were cut with a cryostat (Reichert-Jung, NuBlock, Germany). Biocytin was detected after treatment via the avidin-biotin-peroxidase complex (ABC Elite Kit, Vector Laboratories, Burlingame, CA, USA) and incubation with 3, 3′ diaminobenzidine-tetra-hydrochloride (Sigma-Aldrich, Saint Louis, MO, USA). The sections were stained with cresyl violet, dehydrated and mounted with DPX. The cells were visualized using an optical microscope.

Data Analysis

Data processing was performed using the OpenElectrophy open-access homemade software program ([45]; freely available at http://neuralensemble.org/trac/OpenElectrophy). This program is associated with a MySQL database, and specific analyses were performed with scripts written in Python.

Mitral Cell Model

In Figure 6B and Supporting Information S1, we used a compartmental mitral cell model to infer the effects of inhibitory and excitatory oscillatory conductance injections into the tuft and lateral dendrites on the soma membrane potential. We used a mitral cell model that was developed by Migliore and Shepherd ([47]; ModelDB access number: 97263). Briefly, the model contained a soma, a primary dendrite with a tuft and two secondary dendrites. Spikes were generated by a sodium current, and a delayed rectifier potassium current induced the spike repolarization. A slow inactivating potassium A-current was also present.

Noisy oscillatory conductances were injected into the model as follows:where g is the injected conductance, τ is the synaptic time constant, g₀ is the average conductance, a is the modulation amplitude, f is the modulation frequency, η is a white noise process and σ is the noise amplitude. The final injected current was calculated using the following equation:

where V_syn is the synaptic reversal potential. We simultaneously injected an excitatory conductance (V_E = 0 mV, τ_E = 3 ms) and an inhibitory conductance (V_I = −80 mV, τ_I = 10 ms) into the mitral cell tuft (see Supporting Information S1 for other configurations of synaptic inputs). We used g_I,0 = 3 pS and g_E,0 = 1 pS, and σ was set at 50% of g_E/I,0. Finally, we used f = 2 Hz and a = 0.7.

Twelve seconds of real time were simulated. During the simulations, we monitored the membrane potential in the soma, the tuft and the middle of a secondary dendrite.

The simulations were performed on a standard Linux workstation using the NEURON simulator.

Statistical Analysis

Statistical tests were performed using R combined with Python scripts. The level of significance was set at p<0.05 for all of the statistical tests (p<0.05*; p<0.01**; p<0.001***). The mean values are accompanied by SEM values. The McNemar test was applied to compare the proportion of oscillating cells between the control and odor stimulation conditions. The change in occurrence of an MPSO between the excitability states was compared using Fisher’s exact test. The proportion of cells of the various MPSO types (+, − or sym) was compared between the control and odor period and then between the excitability states using a χ² test. The differences in the MPSO frequencies between the control and odor conditions were compared using a Mann-Whitney test. The correlations between the respiratory frequency and MPSO frequency were quantified using Pearson’s correlation coefficient. The M/T cell activity patterns were compared between two excitability levels, or between control and odor conditions, using a χ² test, and the significant difference between the firing rates across the two excitability states was measured using a Mann-Whitney test.