key: cord-0257267-5hlzqmuz
authors: Średniawa, Władysław; Wróbel, Jacek; Kublik, Ewa; Wójcik, Daniel Krzysztof; Whittington, Miles Adrian; Hunt, Mark Jeremy
title: Network and synaptic mechanisms underlying high frequency oscillations in the rat and cat olfactory bulb under ketamine-xylazine anesthesia
date: 2020-07-24
journal: bioRxiv
DOI: 10.1101/2020.07.23.217604
sha: b83cb27a32b2d098bab704ca8eee5ba5b1dd2dcc
doc_id: 257267
cord_uid: 5hlzqmuz

High frequency oscillations (HFO) are receiving increased attention for their role in health and disease. Ketamine-dependent HFO have been identified in cortical and subcortical regions in rodents, however, the mechanisms underlying their generation and whether they occur in higher mammals is unclear. Here, we show under ketamine-xylazine anesthesia, classical gamma oscillations diminish and a prominent > 80 Hz oscillation emerges in the olfactory bulb of rats and cats. In cats negligible HFO was observed in the thamalus and visual cortex indicating the OB was a suitable site for further investigation. Simultaneous local field potential and thermocouple recordings demonstrated HFO was dependent on nasal airflow. Silicon probe mapping studies spanning almost the entire dorsal ventral aspect of the OB revealed this rhythm was strongest in ventral areas of the bulb and associated with microcurrent sources about the mitral layer. Pharmacological microinfusion studies revealed HFO was dependent on excitatory-inhibitory synaptic activity, but not gap junctions. Finally, we showed HFO was preserved despite surgical removal of the piriform cortex. We conclude that ketamine-dependent HFO in the OB are driven by nasal airflow and local dendrodendritic interactions. The relevance of our findings to ketamine’s model of psychosis in awake state are also discussed.

1 Introduction 1 Local field potential (LFP) oscillations reflect synchronous activity of neuronal assemblies and are thought to 2 play a crucial role in information processing [Engel and Singer, 2001, Colgin and Moser, 2010] . Recent years 3 have witnessed a surge of interest in high frequency oscillations (HFO), also known as ripples, considered 4 important for their roles in health and disease [Buzsáki et al., 2012 , Khodagholy et al., 2017 . HFO have 5 been investigated most notably in the rodent hippocampus, but are also found in diverse cortical, olfactory 6 and limbic areas [Haufler and Pare, 2014 , Zhong et al., 2017 , Vaz et al., 2020 , and have been linked to 7 near-death states [Borjigin et al., 2013] , seizures [Zijlmans et al., 2012] , Parkinson's disease [Foffani et al., 8 2003], and models of psychoses [Hunt et al., 2006 ]. Collectively, these results point to regional and state 9 dependent mechanisms underlying HFO generation in the brain. N-methyl-D-aspartate receptors (NMDAR) 10 are expressed widely in the CNS, and are well-known to be involved in learning and memory processes 11 [Shimizu et al., 2000] . NMDAR antagonists like ketamine or phencyclidine, induce short-lasting psychotic 12 states and is used to study synaptic mechanisms of psychoses in animal models [Frohlich and Van Horn, 13 2014]. To date, many groups, including our own, have shown that low-doses of ketamine (or related NMDA 14 receptor antagonists) produce HFO in many cortical and subcortical areas [Hunt et al., 2006, Phillips et Glutamate is the major excitatory neurotransmitter in the OB, and its NMDA receptors are expressed by 19 the major cell types [Nagayama et al., 2014] . NMDA receptors are important at reciprocal dendrodendritic 20 mitral-granule cell synapses, considered to underlie the generation of fast rhythms in the OB [Neville and 21 Haberly, 2003, Fourcaud-Trocmé et al., 2014, Osinski and Kay, 2016] . We showed recently that the olfactory 22 bulb (OB) is a strong generator of ketamine-dependent HFO in the brain of rodents [Hunt et al., 2019] . The 23 OB appears to have a particularly privileged position since it can impose both fast and slow oscillatory 24 activity in distant regions [Ito et al., 2014] . Although ketamine-dependent HFO has been well-documented, 25 little is known about its mechanisms of generation or its occurrence in higher mammals. The OB is the first 26 relay station of the olfactory system and receives direct sensory input from the olfactory nerve. Since their 27 discovery in 1942, fast oscillations in the OB [Adrian, 1942] , in particular gamma, have been the subject of 28 growing investigations in vivo, in vitro and in modelling studies [Rojas-Líbano et al., 2014] . Indeed, it is 29 well elucidated that NMDA receptors at reciprocal mitral-granule cell dendrodendritic synapses [Schoppa 30 et al., 1998 , Isaacson and Strowbridge, 1998 , Chen et al., 2000 ] underlie gamma rhythmogenesis [Osinski and 31 Kay, 2016] . Even in the absence of odors, in humans and rodents nasal respiration can powerfully entrain 32 fast and slow oscillations in the OB, and other brain regions [Grosmaitre et al., 2007 , Zelano et al., 2016 Zhong et al., 2017], for example delta oscillations during rodent anesthesia and quiet waking. Anesthetized 34 states offer distinct advantages when studying fundamental neuronal processing, since core networks and 35 respiration are spared. Although anesthesia is usually associated with reduced power [Bagur et al., 2018] , 36 and frequency [Ylinen et al., 1995] of fast oscillations, there is some evidence that HFO can be observed 37 under ketamine-xylazine anesthesia [Grenier et al., 2001 , Chery et al., 2014 . This provides an opportunity 38 to probe the mechanisms of ketamine-dependent HFO rhythmogenesis in a straightforward manner devoid of 39 behavioural confounds.

In this study, we found ketamine-xylazine anesthesia in rats and cats was associated with a switch 41 from gamma activity and emergence of HFO in the OB. We found ketamine-xylazine HFO resembled the 42 ketamine-dependent HFO found in the wake-related state, although of slower frequency. This rhythm 43 was strongest in ventral parts of the OB, entrained by nasal respiration, and mediated by excitatory and 44 inhibitory synaptic transmission. It was preserved despite removal of the piriform cortex indicating HFO 45 arises through intrinsic OB circuitry. We conclude that ketamine-dependent HFO is a fundamental brain 46 rhythm in mammals. Given the reduction in frequency we observed previous reports on classical gamma in 47 some ketamine-xylazine anesthetised studies may have inadvertently detected ketamine-dependent HFO.

2 Results 49 2.1 Ketamine-xylazine and subanesthetic ketamine differentially affect gamma 50 (30-65 Hz) and faster rhythms. 51 We compared fast oscillatory activity recorded from LFP's in the OB after a subanesthetic dose of ketamine 52 25 mg/kg and under KX anesthesia (ketamine 100 mg/kg + xylazine 10 mg/kg) and ketamine 200 mg/kg 53 anesthesia (n=9 rats) (Fig. 1) .

Immediately after injection of 25 mg/kg ketamine we observed substantial fast activity at 152.37±10.21 Hz 55 and a concomitant reduction in power of 30-65 Hz activity ( Fig. 1 A-C, early HFO p=0.0042, early gamma 56 p=0.049, paired t-test, n=8). KX anesthesia, confirmed by loss of tail-pinch reflex and eyeblink reflexes, was 57 also associated with initial fast activity ( Fig. 1 E-H , early HFO p=0.0046, early gamma p=0.0077, paired 58 t-test, n=8), which over the course of an hour gradually slowed in frequency to 124.43 ± 10.07 Hz (Fig. 1 I, 59 p< 10 − 4, paired t-test, n=8). Like 25mg/kg ketamine, KX anesthesia also produced a concomitant reduction 60 in 30-65 Hz activity ( Fig. 1 E-H ). An anesthetic dose of ketamine alone was associated with a reduction 61 in fast activity which lasted around 45-60 min followed by emergence of HFO during the recovery phase, 62 typically associated with recovery of the righting reflex ( Fig. 1 E-H , late HFO p=0.023, late gamma p=0.32, 63 paired t-test, n=8). HFO that emerge during recovery of the was slower 125.83 ± 8.66 in frequency than 64 initial burst after injection ( Fig. 1 O, p=0 .0029, paired t-test, n=8). Attenuation of most EEG oscillatory 65 activity (described as EEG holes) following an anesthetic dose of ketamine has been reported recently in 66 sheep [Nicol and Morton, 2020] and we have observed previously that anesthetic ketamine attenuates HFO 67 in the nucleus accumbens of rodents [Hunt et al., 2006] . 68 We noticed that both fast oscillations after subanesthetic ketamine and KX anesthesia occurred as discrete 69 bursts, lasting around 100 ms, and were nested towards the peaks of slow frequencies ( Fig. 1 B,F and K) . 70 In rats injected with 25 mg/kg ketamine coupling occurred at theta frequencies, whereas in KX coupling 71 occurred at slower delta frequencies.

In the same group of rats we injected xylazine 10 mg/kg in 4 rats and we observed no effect on 80-180 73 Hz activity (Fig. S1 A and B) , and a trend for increased 30-65 gamma power was observed. This indicated 74 that ketamine rather than xylazine was responsible for the effect on the high frequencies we observed. A fundamental issue when examining under-reported brain activity is whether electrophysiological signatures 78 are present in a higher mammals. We therefore extended our studies to include cats (n=3) and recorded 79 simultaneously from the olfactory bulb, the lateral geniculate nucleus (LGN) of the thalamus, and from the 80 visual cortex (Fig. 2 A1-A3 ). In OB of two cats ∼ 90 Hz oscillation was recorded under KX anesthesia ( Fig. 81 2 B2-B3). In the third cat histology revealed electrode placement at the edge of the OB which was associated 82 with fast activity of comparable frequency but much smaller power (Fig. 2 B1 ). Consistent with our findings 83 from rats, this fast activity occurred as discrete bursts (Fig. 2 A1) . Importantly, HFO activity was not 84 present in thalamus or visual cortices, demonstrating certain neuroanataomical selectivity to the OB under 85 KX anesthesia. However, note that in awake rats, where the power of fast activity is higher, subanesthetic 86 ketamine can induce HFO in many subcortical areas [Hunt et al., 2006 [Hunt et al., , 2009 ]. In line with our previous 87 findings that ketamine-related fast oscillations can be attenuated by various types of anesthesia [Hunt et al., 88 2009], here, we found that propofol anesthesia also attenuated fast oscillations associated with KX ( Fig. 2 89 B2 and B3 insert). We also calculated the modulation index score (Fig. 2 C1 -C3) for OB channels in each 90 cat, see methods for computational details. Results strongly indicate that KX HFO in cats is coupled to 91 local slow oscillations in OB (strong blue spot around 1 Hz-90 Hz pixel).

KX was infused intravenously at a rate of 0.2 ml every 20 min which provided a window to determine 93 any electrophysiological changes just after administration of KX. Prior to administration gamma ∼ 60 Hz 94 oscillations were present in the OB, then immediately after infusion a clear 90 Hz oscillation was visible ( Fig. 95 2 D). This effect was reproducible across different infusions (Fig. 2 E) . Nasal respiration is known to drive 96 fast and slow oscillations in the OB [Neville and Haberly, 2003 ]. To test if the same held true for KX fast 97 A: Example raw waveforms from one cat from three different brain regions. Raw signal is presented in top rows and 80-130 filtered signal is in bottom row. B: Power spectrum from 10 min of recordings. We observed ∼ 90 Hz HFO separate band in all the recorded cats. In the cat with the electrode in posterior OB HFO was smaller. There was no activity of this type in other simultaneous recorded structures (Thalamus, Visual Cortex) nor in OB under propofol anesthesia condition (insets). C: Modulation index score computed from single OB channels for all three cats. Color strength of the 'pixel' represents power of the modulation for a given slow (driving frequency) and fast (modulated frequency) oscillation extracted from the raw signal. D and E show transition from low ketamine-xylazine to higher dose. We reported increase in 90 Hz band with parallel decrease of the power of gamma activity. The same phenomena were observed in rats; compare Fig. 1 . We checked what happens if we unilaterally block naris of the cat and we saw rapid reduction in HFO power (F) but not 160 Hz activity of unknown origin (supplementary figure). oscillations we also applied short unilateral naris blockade, in two cats, and found this was associated with 98 a reduction in 90 Hz oscillation power (Fig. 2 F) . In one cat, in addition to the 90 Hz activity (described 99 above) we also observed a strong ∼ 160 Hz oscillation( Fig. S2 A and B) present exclusively in the bulb. We 100 do not understand the origin of this oscillation which was not responsive to unilateral naris block ( Slow oscillations in the mammalian OB have long been known to be closely related to nasal respiration. We 105 investigated the relationship between nasal respiration and KX fast oscillations further using thermocouples 106 implanted in the naris of rats. These KX experiments, and those described hereafter, were carried out after 107 initial isoflurane anesthesia for surgical procedures. As expected, shortly after injection of KX, fast rhythmic 108 activity was visible in the raw LFP, nested on a slower oscillatory rhythm ( Fig. 3D and E). However, prior 109 isoflurane exposure reduced mean frequency of KX-HFO compared to KX alone (Fig. S3B , p=0.0028, paired 110 one-way ANOVA)). For this reason, in these experiments the 80-130 Hz band was used. Although gamma 111 oscillations in the OB can reach 80 Hz, considering that gamma power is largely attenuated under KX 112 anesthesia, there would be little gamma contamination.

Consistent with our findings from cats, unilateral naris blockade immediately reduced the power of 80-130 114 Hz activity ( Fig. 3A -C, p=0.0013, one-way ANOVA). The reduction of 80-130 Hz activity power in the OB 115 occurred exclusively on the ipsilateral side and quickly recovered when blockade was removed. Interestingly, 116 a small increase in 80-130 Hz activity power was recorded in the contralateral side. 117 Fig. 3D shows an example recording of nasal respiration, and its relation to < 2Hz, and 80-130 Hz 118 activities recorded in the OB. Note that bursts temporally correlate with peaks of the local delta oscillation 119 and troughs from thermocouple signal ( Fig. 3D and G). We computed correlation coefficients between 120 nasal respiration, <3 Hz filtered signal and the 80-130 Hz envelope (Fig. 3G) . Interestingly, phase analysis 121 (Fig. 3 E and F) showed that the peak of the envelope of fast oscillatory bursts were time-locked to slower 122 oscillations on the descending phase for KX (Fig. 3 F) . The peak phase of local slow LFP rhythm corresponds 123 to inhalation-exhalation transition in breathing cycle (Fig. 3E ). We also calculated the modulation index 124 score ( Fig. 3H and I), see methods for computational details, which confirmed that 80-130 Hz oscillations 125 are driven by nasal respiration. 

To determine if 80-130 Hz activity within the bulb was localised to particular layers we mapped 80-130 Hz 129 activity in the OB using 32 site linear silicon probes of two densities (3.2 mm long with 100 µm spacing and 130 0.64 mm long with 20 µm spacing) ( Fig. 4 A and E). 80-130 Hz oscillations were largely coherent across the 131 OB including contacts up to 3.2 mm apart, the greatest distance we measured. Notably, 80-130 Hz power 132 was substantially larger deep in the granule layer which was clearly visible in the 3.2 mm probe recordings 133 (Fig. 4 B) . We observed a sharp reduction in power close to the mitral layer (Fig. 4 C and G) where 80-130 134 Hz activity shifted in phase (Fig. 4 D, F and H) . For electrodes that did not cross the mitral layer 80-130 135 Hz oscillations were synchronous across all contacts, on a cycle by cycle basis. We found that the slow 2Hz 136 oscillation also reversed phase, but more ventrally and within the EPL/glomerular layers (Fig. 4 D and H) . 137 Since field potentials reflect neuronal activity from a broad area we next reconstructed the underlying sources 138 of 80-130 Hz oscillations and the 2 Hz oscillation. We used the kCSD method to accurately identify the 139 spatial and temporal profiles of changes in membrane current, underlying the field potential (see Methods 140 for computational details). In the raw signal, triggered on 80-130 Hz oscillatory events, we found dipole-like 141 spatiotemporal profiles across the mitral and granule cell layers (Fig. 4 I) . CSD reconstruction (average for 142 N=8) was filtered for slow frequencies (0.3-8 Hz) and we observed dipole-like structure that propagates from 143 glomerular layer, before emergence of 80-130 Hz activity, to EPL layer as the 80-130 Hz power rises in time 144 (Fig. 4 J) . We next filtered the CSD in the 80-180 Hz band and found strong dipoles around the mitral 145 layer ( Fig. 4 anesthesia, we gradually removed cortical tissue, starting laterally and moving medially. We did this until 178 the base of the skull was exposed to ensure that a large amount of the piriform cortex has been extracted 179 (Fig. 6A) . Comparison of LFP oscillations from the dissected vs. intact side had similar power of 80-130 180 Hz activity (Fig. 6B -D, p=0.64 and p=0.37, paired t-test, n=9). There was neither significant change in 181 frequency after brain removal (Fig. 6E, p=0.34 and p=0 .55, paired t-test, n=9). We did not dissect right up 182 to the midline due to the presence of major vessels. However, in four rats the anterior commissure (which 183 carries centrifugal fibers to the OB) was also partially or fully transected. 184 We were not able to dissect to the midline due to the presence of the anterior rostral nerve, which is 185 necessary for circulation of blood to the OB and other anterior brain regions. Unexpectedly, on occasions 186 when this vessel was punctured we observed a transient increase in 80-130 Hz activity followed by attenuated 187 activity. We do not currently understand this hemodynamic effect but suspect it is associated with anoxia, 188 and appears in line with another study showing cardiovascular arrest also induced a transient burst of fast 189 oscillatory activity [Borjigin et al., 2013] . 190 Figure 5 . HFO under ketamine-xylazine are inhibition and excitatory-based but not dependent on gap junctions. A to C show results for NBQX local infusion to the bulb (red lines and red bars) which caused significant reduction in power of the HFO (B and C, n=7). Picture A is an example spectrogram and dotted line refers to the moment of the infusion. D to F and G to I show the same results and analysis as the top row. GABAA receptor blockade with bicuculline infusion (red lines and bars) caused the same effect -reduction in power of the HFO (red lines and bars in E and F, n=5). Example waveforms for a given drug after infusion are presented in the second column. However, gap junction blockage with carbenoxolone seems to have no effect on HFO (H and I, n=4). . HFO do not depend on the piriform cortex or centrifugal input. Picture A shows example conceptual histology of rat in which we removed part of frontal lobe. B and C presents example waveforms at the contralateral and ipsilateral side after the cut was made. We did not see any differences after brain removal neither in power nor in frequency of the oscillations (D and E, n=9). This support the theory that HFO oscillations are generated locally in olfactory bulb by neuronal circuits.

Nasal respiration is critical for ketamine 80-130 Hz activity.

Here we show that a 80-130 Hz rhythm emerges under KX anesthesia in the OB of rats and cats. This 193 rhythm shows some degree of neuroanatomic specificity, being large in amplitude in the OB, relatively weak 194 at the OB border, and absent in thalamic and visual cortices. 80-130 Hz activity occurs as discrete bursts 195 frequently coupled to nasal respiration, indeed naris occlusion attenuates this rhythm. The olfactory bulb 196 receives direct input from the olfactory nerve and nasal respiration has long been known to entrain the 197 rhythmic activity in olfactory networks [Adrian, 1942, Macrides and Chorover, 1972] . We found that KX 198 mitral/tufted neurons fire in phase with 80-130 Hz, consistent with a previous finding under KX reporting 199 mitral/tufted firing over 100 Hz and associated with inhalation/exhalation transitions [Burton and Urban, 200 2014]. Our previous work in freely moving rats given a subanesthetic dose of ketamine also showed spiking 201 phase locked to a fast 130-180 Hz activity [Hunt et al., 2019] . Nasal respiratory therefore provides a gate 202 for the firing of OB projection neurons [Wachowiak, 2011] and the emergence of ketamine-dependent fast 203 rhythms. Using a 3.2 mm probe we mapped, almost entirely, the dorsal-ventral aspect of the rat OB. analyses, revealed the presence of two spatially and temporally distinct current dipoles. A sink, around 1 Hz, 208 followed the breathing cycle, was localized close to the EPL layer and a faster 80-130 Hz microcurrent dipole 209 was found closer to the mitral layer. Current sinks indicate flow of the positive ions (i.e., potassium, calcium) 210 into neurons [Nicholson and Freeman, 1975, Mitzdorf, 1985] . Stimulation of the olfactory nerve to mimic 211 afferent input, generates a sink in the glomerular and EPL layers [Neville and Haberly, 2003, Kay, 2014] Both NMDA blockade and nasal respiration are prerequisites for the 80-130 Hz rhythm we recorded. 222 Importantly, this activity was not present prior to NMDA blocker injection and was attenuated by isoflurane 223 (rats) and propofol (cats) anesthesia. Xylazine alone had no significant impact on 80-130 Hz power indicating 224 dependence of this rhythm on NMDA receptor blockade. Fast oscillations typically require interplay between 225 excitatory and inhibitory transmission [Buzsáki et al., 2012] . Within the OB fast oscillations can be generated 226 through glutamatergic release by mitral cell dendrites onto granule cell spines and reciprocal GABA release 227 to locally inhibit depolarization spread [Chen et al., 2000] . Our microinfusion studies showed that inhibitory 228 and excitatory components are required for the emergence of 80-130 Hz activity. By contrast, gap junctions, 229 which can also underlie both physiological and pathological fast rhythmogenesis [Traub et al., 2002] [Chen et al., 2000] . Thus other excitatory-inhibitory networks would be predicted to underlie 248 the generation of ketamine-dependent fast rhythms. There is an abundance of other types of inhibitory 249 interneurons within the OB which can shape M/T behavior [Burton, 2017] . How might such fast oscillations 250 be achieved in the OB? Clues may be found in other areas known to generate both gamma and faster 251 oscillations. A good example is the hippocampus, where inhibitory PV-positive basket cells generate a ripple 252 frequency. PV positive interneurons have been identified in the EPL layer of the OB, which make reciprocal 253 contacts the perisomatic region of mitral cells [Kosaka et al., 1994] . They have morphological similarities to 254 basket cells [Crespo et al., 2013] . In the OB, as elsewhere in the brain, PV interneurons are fast spiking 255 [Mountoufaris et al., 2017] . In the OB, these cells can fire around 170 Hz, and are modulated by respiration 256 [Kato et al., 2013] . They are thought to mediate broad lateral inhibition since they receive input and inhibit 257 MC along their long secondary dendrites [Miyamichi et al., 2013] Importantly, excitation of PV-positive 258 interneurons by mitral cells is chiefly mediated by AMPA receptors, with only a weak contribution by NMDA 259 receptors [Kato et al., 2013] . Although PV-positive cells innervate mitral cells more densely than granule 260 cells, very little is known about the capacity of this network to generate oscillations, however, given the 261 precedent that these cells can generate fast rhythms (>100 Hz) we hypothesise that this circuit underlies 262 the fast rhythm observed here [Kato et al., 2013] . We speculate that the movement of air across nasal 263 mechanoreceptors stimulates the olfactory nerve, which in turn depolarizes MC via their dendrodendritic 264 synapses, to induce burst firing in PV cells, which generates an E-I rhythm, which is turned off during 265 expiration until the next breathing cycle. Reduced inhibitory tone at GC-MC synapses produced by NMDA 266 blockade further increases the excitability of MC possibly enhancing this rhythm.

The OB also contains deep short axon interneurons which can modulate the firing of projection neurons, 268 12/24 however, their firing frequencies would be unlikely to support fast rhythm . Depolarization 269 of mitral cells, expected to occur following disinhibition by granule cells, is associated with progressive increases 270 in frequency and power of subthreshold oscillatory activity; however, since field activity is independent of 271 excitatory and inhibitory transmission, and below 50 Hz, it would not account for the ketamine-dependent 272 rhythm we observed [Desmaisons et al., 1999] . 273 Given the above, we propose dynamical independent excitatory-inhibitory networks govern gamma 274 (granule-mitral) and higher frequency (PV-mitral) oscillations in the OB. Differential sensitivities of these 275 circuits to NMDA receptor blockade permit emergence or inhibition of fast brain rhythms. [Ylinen et al., 1995] . Additionally, 287 Neville and Haberly also reported that the frequency of discrete gamma and beta oscillatory bands is slower 288 under urethane that the corresponding oscillations in awake rats [Neville and Haberly, 2003 ]. Finally, we 289 have shown previously, the frequency of ketamine-dependent fast oscillations are highly dynamic and can 290 drop by as much as 80 Hz after antipsychotic injection [Olszewski et al., 2013] and are likely to be related in 291 awake and KX-anesthetised states. In summary, ketamine-dependent HFO is a highly dynamic band whose 292 frequency varies according to state, and HFO recorded in awake and KX-anesthetized states appear to be 293 related.

Surgery and chronic recordings 296 All experiments were conducted in accordance with the European community guidelines on directive 297 2010/63/UE on the protection of animals used for scientific purposes and approved by a local ethics 298 committee. Thirty male Wistar rats (250-350 g) were used in this study. In 8 rats, tungsten electrodes 299 (125 µm, Science Products, Germany) were implanted bilaterally in OB (AP+7.5, ML=+0.5, DV=3-3.5 300 mm). LFP recordings were carried out before and after injection intraperitoneal of ketamine 25 mg/kg, and 301 anesthetic doses of ketamine 100 mg/kg + xylazine 10 mg/kg (KX), and ketamine 200 mg/kg with 3-4 302 days separating each recording session (n=9). Under KX anesthesia the left or right naris was occluded 303 using a soft piece of silicon rubber to block nasal respiration. At the end of the experiments the brains were 304 post-fixed in 4% paraformaldehyde solution. Brains were dissected and placed in a 10% followed by 30% 305 sucrose solution for 2-4 days. Electrode locations were determined on 40 µm Cresyl violet (Sigma, UK) or 306 Hoechst (Sigma, UK) stained sections.

Thermocouple and LFP recordings 308 For acute studies, 8 rats were initially anesthetized using isoflurane during which time electrodes were 309 implanted in the olfactory bulb and thermocouples in the frontal recess of the naris on the ipsilateral side. 310 When electrodes were in place isoflurane anesthesia was replaced with KX by gradual injection of the cocktail 311 and removal of isoflurane. Initial isoflurane exposure was necessary due to well-documented variable surgical 312 plane anesthetic responses in rats compared to KX alone [Struck et al., 2011] .

Silicon probe recordings 314 A total of 14 rats were used for spatial mapping of oscillatory activity in the rodent OB. Rats were prepared 315 for acute recordings as described above. Recordings were carried out in the OB using 32-channel silicon 316 probes Edge-10mm-100-177 (N=6 rats, Neuronexus) and Edge-10mm-20-177 (N=8 rats). The electrodes 317 were separated by an interelectrode distance of 100 (long probe) and 20 µm (short probe). Prior to recording 318 electrodes were dipped in a 5% solution of DIi (Sigma) dissolved in DMSO (Sigma). The track of the 319 electrode was visualized using a fluorescent microscope.

Local drug infusion 321 Rats were initially anesthetized using isoflurane and implanted with a guide-electrode complex in the left 322 and right OB. Following implantation, isoflurane was substituted for KX anesthesia (see above for further 323 information). LFPs were recorded bilaterally and when a stable fast oscillation was visible (80-130 Hz). 324 NBQX (2 µg, Sigma, n=7), Bicuculline methiodide (0.05 µg, Sigma, n=5) and Carbenoxolone disodium 325 salt (1 µg, Sigma, n=4) were infused into the bulb. Saline was infused to the opposite bulb. For infusion, 326 internal cannulae (28 gauge, Bilaney) that extended 2mm below the tip of the guides were inserted for 60s 327 followed by 60s infusion of CNQX, bicuculline, CBX or saline (volume 0.5 µl). Rats were recorded for 30 328 min post infusion and were then humanely killed using an overdose of anesthesia and their brains dissected 329 for histological processing.

Dissection of brain tissue 331 Rats were initially anesthetized by isoflurane for implantation of electrodes in the left and right OB. Following 332 removal of the overlying skull isoflurane anesthesia was was replaced by KX. Under stable KX anesthesia 333 we drilled the perimeter of a large cranial window approx. 7mm x 7mm (left hemisphere) from the midline 334 to the lateral edge of the skull and removed the overlying bone. The exposed brain was dissected (using 335 aspiration and a scalpel) in a lateral-medial direction until the base of the skull had been reached. LFPs 336 from the left and right OB were recorded prior to and immediately after dissection.

Cat experiments 338 All experiments were conducted in accordance with the European community guidelines on directive 339 2010/63/UE on the protection of animals used for scientific purposes and approved by a local ethics 340 committee. Three healthy, mature cats (1 female, 2 male) were used. Cats were initially anesthetised using 341 a bolus ketamine-xylazine i.p. injection . An intravenous catheter was implanted in a saphenous vein for 342 supplementary ketamine-xylazine administration and 0.1-0.2 ml was administered every 20 min. Cat's were 343 placed in a stereotaxic frame and electrodes (32-channel silicon probes or in-house 16-channel electrodes made 344 of 25 mm tungsten wire) implanted in the OB (AP+16,5, ML=-1,5, DV=3-4 mm), Thalamus nucleus (AP 345 6,5, ML=-9, DV=13,5 mm) and visual cortex (AP -16, ML=-1-9, DV=1-2 mm). Following acquisition of 346 electrophysiological data the naris was briefly closed for 10 seconds (in 2 cats), anesthesia was then replaced 347 with propofol for visual presentation experiments (not presented here). Recorded signals were processed using SciPy signal and NumPy Python libraries. Analysis included 351 bandpass filtering using Butterworth filters. Power of dominant frequency and dominant frequency were 352 evaluated using Welch transform from 60 seconds windows. To establish phase relation in KX HFO we first 353 used Hilbert transform to find a maximum activity of HFO burst and then computed shift in time relative to 354 peak of delta oscillations (score is rescaled to radians). Several hundred HFO bursts were used to compute 355 intertrial phase clustering (ITCP) defined as IT P C = |n −1 n r=1 e ik |, where k is the relative phase of the 356 burst and n is the number of trial.

To study correlation of thermocouple's rhythm and LFP signal oscillation we filtered the signal in delta 358 frequencies 0.3-5 Hz and KX HFO 80-130 Hz. We computed a Pearson correlation score between the delta 359 band of thermocouple and the envelope of HFO signal computed with Hilbert transform. To confirm our 360 hypothesis that HFO is modulated by breathing rhythm we used comodulogram analysis for the two signals. 361 Comodulogram matrix was computed using open-source Python library described in [Dupré la Tour et al., 362 2017]. We used the "Tort" method from pactools Python package, which seems to find a compromise for 363 proper resolution in 'phase' and 'amplitude' signal and is based on classic phase-amplitude coupling method. 364 We evaluated statistical significance of the coupling using resampling test of MI martices between groups of 365 rats that were under KX and isoflurane anesthesia.

For multielectrode recordings, KX-HFOs significant bursts were detected using 3 standard deviation 367 threshold from top (short silicon probe) and middle (long silicon probe) channels used as a reference. We 368 computed phase shift between channels using maximum correlation score in respect to reference channel and 369 averaged the score across rats. CSDs were reconstructed using kCSD algorithm method from [Potworowski 370 et al., 2012 , Chintaluri et al., 2019 and available at https://github.com/Neuroinflab/kCSD-python. We 371 reconstructed CSD first and then filtered spatio-temporal CSD picture in delta 0.3-5 Hz and KX-HFO 372 80-130 Hz frequency bands. For Multiunit activity analysis we first filtered the signal above 500 Hz for 373 every HFO burst/event. Then we extracted candidates for spikes with 3 standard deviation criterion and 374 represented them as discrete events in time. As a final step we made a histogram from aggregated (across 375 HFO events) spikes and computed Pearson correlation coefficient between spike histogram and average HFO 376 waveform. We repeated this kind of analysis for all channels independently.

Shaded regions in all plots and whiskers of the bar plots represent standard deviation of the mean (s.e.m). 378 For statistical analysis we used one-way ANOVA test for independent experiments or different electrode 379 channels and Student's paired t-test for different timepoints in a given channelf oneway and ttestrel, 380 respectively, from SciPy Python library. All sample groups were tested with Shapiro-Wilk's test for normality. 381 All data sets passed normality testing and therefore parametric statistics were used. For all figures we used 382 *** for pvalue < 0.001, ** for pvalue < 0.01 and * for pvalue < 0.05 convention. Additionally we used 383 resampling test (100 000 draws with return) to compare modulation index matrices for KX and isoflurane 384 anesthetised rats (n=8). Figure S3 . A: Example spectrogram of the chronically implanted rat that was exposed first to isoflurane and then got KX injection. B: Analysis of the frequency reduction in rats that were shortly exposed to isoflurane before KX injection. 

Olfactory reactions in the brain of the hedgehog

Modulation of 398 thalamo-cortical activity by the NMDA receptor antagonists ketamine and phencyclidine in the awake

Harnessing olfactory 404 bulb oscillations to perform fully brain-based sleep-scoring and real-time monitoring of anaesthesia depth

Surge of neurophysiological coherence and connectivity in the dying brain

Cortical Feedback Control of Olfactory 411 Bulb Circuits

Broadcasting of cortical activity to the 415 olfactory bulb

Inhibitory circuits of the mammalian main olfactory bulb

Greater excitability and firing irregularity of tufted cells underlies distinct 422 afferent-evoked activity of olfactory bulb mitral and tufted cells

Olfactory bulb deep short-axon cells 426 mediate widespread inhibition of tufted cell apical dendrites

The origin of extracellular fields and currents -EEG

Ketamine alters oscillatory 432 coupling in the hippocampus

Analysis of relations between NMDA receptors and GABA 434 release at olfactory bulb reciprocal synapses

Anesthetic regimes modulate the temporal dynamics of local field 437 potential in the mouse olfactory bulb

kCSD-python, a tool for reliable Current Source Density estimation

Lateral excitation within the olfactory bulb

Gamma oscillations in the hippocampus

The circuits of the olfactory bulb: The 448 exception as a rule

Control of action potential timing by intrinsic subthreshold 451 oscillations in olfactory bulb output neurons

Molecular signatures of neural connectivity in the olfactory cortex

Electrical coupling underlies high-frequency 458 oscillations in the hippocampus in vitro

Non-linear auto-regressive models for cross-frequency coupling in neural time series

Temporal binding and the neural correlates of sensory awareness

A PK-PD Model of Ketamine-Induced High-466 Frequency Oscillations

469 300-Hz subthalamic oscillations in Parkinson's disease

Two distinct olfactory bulb sublaminar networks involved 473 in gamma and beta oscillation generation: a CSD study in the anesthetized rat

Reviewing the ketamine model for schizophrenia

Two Distinct Channels of Olfactory 477 Bulb Output

Independent control of gamma and 479 theta activity by distinct interneuron networks in the olfactory bulb

Correlation-induced synchronization of 482 oscillations in olfactory bulb neurons

Focal synchronization of ripples (80-200 Hz) in neocortex and 486 their neuronal correlates

Dual functions of mammalian olfactory 489 sensory neurons as odor detectors and mechanical sensors

High-frequency oscillations are prominent in the extended amygdala

Comprehensive connectivity of the mouse main olfactory bulb: analysis and online 496 digital atlas

NMDA receptor hypofunction produces opposite effects on prefrontal 500 cortex interneurons and pyramidal neurons. The Journal of neuroscience : the official journal of the 501 Society for

Ketamine Dose-Dependently Induces High-Frequency Oscillations 505 in the Nucleus Accumbens in Freely Moving Rats

State-dependent changes in high-frequency 508 oscillations recorded in the rat nucleus accumbens

511 The olfactory bulb is a source of high-frequency oscillations 130 -180 Hz associated with a subanesthetic 512 dose of ketamine in rodents

Olfactory reciprocal synapses: Dendritic signaling in the CNS

Whisker 517 barrel cortex delta oscillations and gamma power in the awake mouse are linked to respiration

Parvalbumin-expressing interneurons 520 linearly control olfactory bulb output

Circuit oscillations in odor perception and memory

The effect of NMDA-R antagonism on simultaneously acquired 526 local field potentials and tissue oxygen levels in the brains of freely-moving rats

Learning-enhanced coupling between ripple oscillations 530 in association cortices and hippocampus

Calcium-binding protein parvalbumin-immunoreactive neurons 533 in the rat olfactory bulb -1. Distribution and structural features in adult rat

Interplay between Local GABAergic Interneurons and Relay Neurons 536 Generates γ Oscillations in the Rat Olfactory Bulb

NMDA receptor antagonists disinhibit rat posterior 540 cingulate and retrosplenial cortices: A potential mechanism of neurotoxicity

Olfactory Bulb Units: Activity correlated with inhalation cycles and odor 543 quality

Current source-density method and application in cat cerebral cortex: investigation of evoked 545 potentials and EEG phenomena

Dissecting local circuits: 547 parvalbumin interneurons underlie broad feedback control of olfactory bulb output

Interneurons of the 556 neocortical inhibitory system

Neuronal organization of olfactory bulb circuits

Beta and Gamma Oscillations in the Olfactory System of the Urethane-564

Theory of current source-density analysis and determination of conductivity 567 tensor for anuran cerebellum

Characteristic patterns of EEG oscillations in sheep (Ovis aries) induced by 570 ketamine may explain the psychotropic effects seen in humans

Disruption of centrifugal inhibition 573 to olfactory bulb granule cells impairs olfactory discrimination

Antipsychotic compounds differentially 578 modulate high-frequency oscillations in the rat nucleus accumbens: A comparison of first-and second-579 generation drugs

Granule cell excitability regulates gamma and beta oscillations in a model of 582 the olfactory bulb dendrodendritic microcircuit

Pharmacological manipulation of the olfactory 585 bulb modulates beta oscillations: Testing model predictions

Differential effects of NMDA antagonists on high frequency and gamma EEG oscillations 590 in a neurodevelopmental model of schizophrenia

Kernel current source density method

The olfactory bulb theta rhythm follows all 596 frequencies of diaphragmatic respiration in the freely behaving rat

Dendrodendritic inhibition in 599 the olfactory bulb is driven by NMDA receptors

NMDA receptor-dependent synaptic reinforcement as 602 a crucial process for memory consolidation

Effect of a Short-term Fast on Ke-605 tamine-Xylazine Anesthesia in Rats

Axonal 608 gap junctions between principal neurons: A novel source of network oscillations, and perhaps epileptogenesis. 609 Reviews in the Neurosciences

Fast Oscillations and Synchronization Examined 611 with In Vitro Models of Epileptogenesis

Replay of cortical spiking sequences during human 614 memory retrieval

All in a Sniff: Olfaction as a Model for Active Sensing

NMDA-receptor modulation of lateral inhibition 619 and c-fos expression in olfactory bulb

Sharp wave-associated 622 high-frequency oscillation (200 hz) in the intact hippocampus: Network and intracellular mechanisms

Nasal respiration 625 entrains human limbic oscillations and modulates cognitive function

Selective entrainment of gamma subbands by different slow network oscillations

High-frequency oscillations 632 as a new biomarker in epilepsy

A: Spectrogram of the example rat after xylazine injection. B: Extraction of the power of dominant frequency for Gamma and 80-180 Hz HFO. We did not see any change in HFO after xylazine application alone. C: Spectrogram of the example rat with anesthetic ketamine injection. D: Same type of analysis as B but for ketamine anesthetized rat