Olfactory Bulb Field Potentials and Respiration in Sleep-Wake States of Mice

It is well established that local field potentials (LFP) in the rodent olfactory bulb (OB) follow respiration. This respiration-related rhythm (RR) in OB depends on nasal air flow, indicating that it is conveyed by sensory inputs from the nasal epithelium. Recently RR was found outside the olfactory system, suggesting that it plays a role in organizing distributed network activity. It is therefore important to measure RR and to delineate it from endogenous electrical rhythms like theta which cover similar frequency bands in small rodents. In order to validate such measurements in freely behaving mice, we compared rhythmic LFP in the OB with two respiration-related biophysical parameters: whole-body plethysmography (PG) and nasal temperature (thermocouple; TC). During waking, all three signals reflected respiration with similar reliability. Peak power of RR in OB decreased with increasing respiration rate whereas power of PG increased. During NREM sleep, respiration-related TC signals disappeared and large amplitude slow waves frequently concealed RR in OB. In this situation, PG provided a reliable signal while breathing-related rhythms in TC and OB returned only during microarousals. In summary, local field potentials in the olfactory bulb do reliably reflect respiratory rhythm during wakefulness and REM sleep but not during NREM sleep.


Introduction
Neuronal network oscillations and their interactions are thought to be important for perception [1], sensory-motor coordination [2], and memory processes [3]. According to a recent hypothesis, the breathing rhythm serves as a clock for binding several orofacial senses to a common percept [4]. In line with this hypothesis, it was found that whisking synchronizes with breathing [5,6] and that slow oscillations in the whisker barrel cortex of waking mice are locked to respiration [7]. Recently, we demonstrated in the dentate gyrus of anaesthetized [8] and awake [9] mice a respirationrelated rhythm (RR) clearly distinct from theta oscillations. Discrimination of the RR from other slow rhythms requires registration of respiration, preferably as an electrical field potential which can be easily recorded together with other electrophysiological data. There are several other more direct ways to measure respiration in rodents. Intranasal pressure sensors or thermocouples implanted into the nasal cavity [10][11][12] are commonly used in rodents but are invasive devices which may compromise natural behavior. As a less invasive alternative, whole-body plethysmography [13][14][15] provides a reliable, state-independent biomechanical measure of respiration. However, plethysmography requires housing of the animals in specialized and closed small chambers, making it unpractical for use in combination with tethered brain recordings or complex behavioral tasks. For decades, it had been known that local field potentials (LFP) of the olfactory bulb (OB) of mammals are entrained by nasal respiration [16][17][18][19]. This signal has already been used to measure respiration rate in freely moving rats [20]. However, the reliability of OB LFP for estimating breathing and sniffing frequencies in mice has not been systematically studied and verified by alternative approaches, especially not through different states of vigilance. In the present study, we simultaneously used plethysmography, nasal thermocouple, and LFP recordings in OB during various sleep-wake states of mice.
for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Research [22]) and were approved by the Governmental Supervisory Panel on Animal Experiments of Baden Württemberg at Karlsruhe (G84/13). All efforts were made to minimize animal suffering and to reduce the number of animals used.

Animal
Care and Housing Conditions. C57BL/6N mice were purchased at 70 or 84 d of age from Charles River (Sulzfeld, Germany). Animals were housed in groups of four inside a ventilated Scantainer (Scanbur, DK) on an inverted 12 h/12 h light/dark cycle (light on 8:00 PM) for a minimum of 2 weeks. Mice had free access to food and water. After chronic electrode implantation, mice were kept individually throughout the experiment. The animals were killed with an overdose of isoflurane during brain perfusion.

Surgery.
A total of 17 C57BL/6N mice (16 female and one male; 13-29 weeks old; weight: 22-27 g) were anesthetized with isoflurane in medical oxygen (4% isoflurane for induction, 1.5%-2.5% for maintenance, flow rate: 1 L per minute). For analgesia, 0.1 mg/kg of buprenorphine was injected subcutaneously before and 8 h after surgery (for details see [9]). After exposure of the skull, holes of 0.5-1.0 mm in diameter were drilled above the OB, the parietal cortex, and cerebellum according to stereotaxic coordinates; see below for coordinates based on [23]. Two stainless steel watch screws (1 × 3 mm) over the cerebellum served as ground and reference electrode. For monitoring the temperature of nasal air flow, precision fine bare wire thermocouples (80 m diameter, Omega Engineering, part #5TC-TT-KI-40-1 M) were implanted into both nasal cavities (−11 mm anteroposterior (AP), 0.5 mm mediolateral (ML)). A pair of varnish-insulated tungsten wires (50 m) was implanted into the granule cell layer of left OB (AP: +4.5; ML: 0.8; 1.3 mm ventral). For restraining the head of the animal, a steel pin (2.1 mm diameter, 14 mm length) was attached to the skull under an angle of 75 degrees.

Electrophysiology.
One week after surgery, experiments began with a 1 h recording session in the plethysmograph (see below) or in a custom-built head-fixed setup with a treadmill for voluntary running [9]. Extracellular signals were filtered (1-500 Hz), amplified (RHA2000, Intan Technologies, Los Angeles, USA), digitized (2.5 kHz), and stored for offline analysis. A three-dimensional accelerometer mounted on the amplifier board at the animals head allowed movement detection. Mice habituated quickly to the head-restraining conditions and tolerated it well for up to 1 h. The whole-body plethysmograph (PG) consisted of a transparent cylindrical box (78 mm inner diameter, 165 mm height) which was connected to a reference chamber and adapted for collection of LFP (EMKA Technologies, SAS, France). Mice habituated well to the PG apparatus within the first session.   [24].
Data was analyzed in MATLAB (The Mathworks Inc., Natick, MA) using built-in and custom-written routines (for more details see [9]).

Spectral and Coherence Analysis.
Power spectral density was calculated by means of the Welch periodogram method using the pwelch.m function from the Signal Processing Toolbox (50% overlapping, 4 s Hamming windows). Timefrequency power decomposition was obtained by means of the spectrogram.m function; sliding windows of 2 s and 50 ms time steps were used; phase coherence was obtained by means of the mscohere.m function from the Signal Processing Toolbox, using 2 s windows with 50% overlap. For all types of data analysis, LFP signals of 25 s duration were used.

Histology.
After the experiments, animals were deeply anesthetized with isoflurane and perfused transcardially with PBS and subsequently with 4% paraformaldehyde (PFA). Brains were carefully dissected and stored in PFA overnight, and coronal sections were cut (50 m), mounted, and stained with cresyl violet. The electrode position was then verified by light microscopy.  Friedman test, Mann-Whitney test, or the Wilcoxon signed rank test.

In Awake Mice Plethysmography, Nasal Thermocouple and Olfactory Bulb Local Field Potentials Reveal Similar
Respiration Signals. Respiration was recorded using a thermocouple (TC) chronically implanted into the nasal cavity. The TC signal reflects temperature changes of nasal air flow. Simultaneously, thoracic movements were detected with whole-body plethysmography (PG) indicating relative values of respiratory air flow (mL/s). In addition, local field potentials (LFP) from the granular cell layer of the olfactory bulb (OB) showed fluctuations entrained by nasal respiration as described earlier [18]. Figure 2 illustrates similarity of the three methods in a representative example of waking immobility (Figure 2

During Waking the Power of Respiration-Related Rhythm in Olfactory Bulb Decreased with Increasing Respiration Rate.
We next investigated how peak power of the RR in OB varied with changing breathing frequency. Respiration rate in waking mice varies strongly, especially during interspersed spouts of sniffing [9,11]. Figure 3 shows a representative example of an abrupt transition from slow breathing to fast sniffing. At high sniffing rates, the amplitude of the PG signal increased while the amplitude of the RR in OB decreased (Figure 3(a), raw signals). Despite the amplitude differences similar prominent rhythms were found for all three signals in the spectrograms (Figure 3(b)) and in power spectral density (PSD) plots, at low (Figure 3(c), black) and high frequencies (Figure 3(c), blue). For quantitative analysis, periods of waking were sorted into frequency bins of breathing/sniffing obtained by PG (Figure 3(d)). PSDs of the three simultaneously recorded respiration signals and coherence between PG

Negative Correlation between Power of Respiration-Related Rhythm in OB and Breathing Frequency during
Treadmill Running. To reduce respiratory variability, we next performed head-fixed experiments where the behavior of the mouse was basically clamped to either immobility or running on a circular treadmill and minimized sniffing [9]. Only periods of running were used here and divided into 1 Hz bins of increasing breathing frequency obtained by simultaneously recorded TC. Figure 3(e) displays -scored averages of RR peak power in OB ( = 11 mice) which linearly decreased with increasing respiration frequency ( = 7; = −0.958; < 0.001, Pearson's correlation). Both RR in OB and respiratory signal in TC regularly returned during microarousals (see asterisks in Figure 4(b)). During parts of REM sleep, especially at low breathing frequencies, RR in OB recovered ( Figure 5(a), low; Figure 5(b)), whereas the TC signal remained absent. Clear power peaks were found for PG and OB for both low ( Figure 5(c), upper and lower panel, black) and high ( Figure 5(c), blue) breathing frequencies, in contrast to TC where no corresponding peaks were discernible in the power spectral density graph ( Figure 5(c), middle panel). During REM sleep, RR in OB was detectable, but the signal was less reliable as compared to awake animals, especially at high breathing frequencies (Figures 5(a) Figure 5(d) shows peak power averages of six animals of PG (red) and OB (black) recorded during REM sleep binned into three breathing frequency ranges as above. PG peak power did not depend on breathing frequency, in contrast to RR peak power in OB (Figure 5(d); black; = 6; * < 0.01; Friedman's test). Likewise, coherence between PG and OB signals decreased between the lowest and highest breathing frequency bins ( Figure 5(e); = 6; * < 0.01; Friedman's test). Coherence  between PG and OB was also lower in REM compared to waking (not shown) for the two highest frequency bins ( = 6; < 0.05 and < 0.01; Wilcoxon test). In summary, plethysmographic data provided the only reliable measure of respiration throughout NREM sleep and was more reliable during REM sleep at high breathing frequencies compared to OB LFP which was a good indicator of respiration during REM sleep with low respiration rates. Figure 6 illustrates representative examples of coherence and its harmonics between all three signals: PG-OB (red), TC-OB (blue), and PG-TC (green) at different ranges of respiration during waking (a), NREM sleep (b), and REM sleep (c). In waking (Figure 6(a)), large peaks of coherence between all signals correspond to respiration frequency (PG PSD: black) indicating the reliability of all three signals to measure respiration. During waking and REM sleep, there is no delta activity in OB; therefore, the LFP is a good indicator of respiration and its peak in PSD corresponds to the respiration rate. However, in NREM sleep, peaks at respiration rate remained only for coherence between PG and OB. Coherence analysis indeed demonstrates that the respiration-related rhythm is preserved in OB during sleep. However, for coherence analysis, a second source of respiration is required during NREM sleep; the OB LFP alone is not sufficient as unequivocal signal of respiration due to multiple equivocal peaks in the power spectral density of the OB LFP.

Discussion
In the present study, we report slow local field potentials (LFP) in the granular cell layer of olfactory bulb (OB) which reliably indicate respiration in waking mice. This was verified by simultaneous measurement of nasal thermocouple (TC) signals and plethysmography (PG). Our results confirm reports from several decades ago which described slow rhythmic LFPs in the OB related to nasal air flow and breathing [16][17][18]. At the time, however, these data could not be independently verified due to lack of adequate digital analysis tools. During subsequent years, the concept of respirationrelated rhythmic fluctuations of LFPs in OB was progressively replaced by a terminology which subsumed slow OB rhythms as olfaction-related theta oscillations [25][26][27][28]; see also discussion in [9]. From a different viewpoint, however, 8 Neural Plasticity theta oscillations are not just rhythmic fluctuations in the respective frequency range (4-12 Hz) but share the characteristic behavioral features of hippocampal theta rhythms. Theta oscillations occur during locomotion, exploration [29], arousal, or fear [30,31] or in REM sleep [32]. In contrast, respiration-entrained rhythms in OB, if not masked by delta waves during sleep, are present in any behavioral state. They reflect respiration, and their amplitude is largest during relaxed wakefulness [9,33], a state when theta rhythm is absent [29,34]. Indeed, we have recently demonstrated that a clearly distinct respiration-related rhythm (RR) occurs alone or simultaneously with theta oscillations both in OB and in upstream regions like the dentate gyrus of the hippocampal formation [8,9]. Hippocampal RR is generated by different mechanisms than theta. For example, tracheotomy abolishes RR but not theta oscillations, whereas atropine diminishes theta but does not affect RR [8].
In the present study, we report that only PG, but not TC or OB LFPs, is trustworthy proxy for respiration during NREM sleep. While RR in OB repeatedly returned during NREM and especially during REM sleep, the TC signal was absent throughout NREM and REM sleep. This may be in part explained by a decrease in nasal air flow during these states. Indeed, total inspiratory flow in mice decreases continuously from quiet waking to NREM and REM sleep [35]. Other factors, like redirection of air flow inside the nasal cavity [36], may contribute to the disappearance of respiration signals at the TC. Interestingly, both signals, TC and RR in OB, regularly reappeared during microarousals, providing a potential indicator of these brief states with enhanced vigilance.
In summary, investigating respiration and its related oscillations in the mouse brain during NREM sleep requires PG or other signals different from TC and RR in OB. In waking states of freely moving and behaving mice and in parts of REM sleep with low respiration rates, however, OB LFPs can be reliably used to measure respiration. In addition, we report that RR peak power decreases with increasing respiration rate under several behavioral conditions in wakefulness and REM sleep. Our findings from mice are in good agreement with earlier reports on anesthetized rats [37,38]. In any case, measuring respiratory rhythm during NREM sleep and separating it from delta activity require great caution.

Conclusions
Local field potentials in the olfactory bulb of waking but not sleeping mice reliably reflect respiration. This holds also true for nasal thermocouples. In sleeping mice, the most accurate method to record breathing is whole-body plethysmography. Nasal thermocouples are completely unsuitable. Local field potentials from olfactory bulb are reliable in parts of REM sleep with low respiration rates but alone are not sufficient in NREM sleep.