OPTOGENETIC STIMULATION OF C1 AND RETROTRAPEZOID NUCLEUS NEURONS CAUSES SLEEP STATE DEPENDENT CARDIORESPIRATORY STIMULATION AND AROUSAL IN RATS

Abstract

C1 catecholaminergic neurons and neurons of the retrotrapezoid nucleus are integrative nodes within the brainstem network regulating cardiorespiratory reflexes elicited by hypoxia and hypercapnia, stimuli that also produce arousal from sleep. In the present study, Channelrhodopsin2 was selectively introduced into these neurons with a lentiviral vector in order to determine if their selective activation also produces arousal in sleeping rats. Sleep stages were identified from electroencephalographic and neck muscle electromyographic recordings. Breathing was measured using unrestrained whole body plethysmography and blood pressure by telemetry. During non-rapid eye movement sleep, unilateral photostimulation of the C1 region caused arousal in 83.0 ± 14.7% of trials and immediate and intense cardiorespiratory activation. Arousal during photostimulation was also observed during rapid eye movement sleep (41.9 ± 5.6% of trials), but less reliably than during non-rapid eye movement sleep. The cardiorespiratory responses elicited by photostimulation were dramatically smaller during rapid eye movement sleep than non-rapid eye movement sleep or wakefulness. Systemic alpha1-adrenoreceptor blockade reduced the cardiorespiratory effects of photostimulation, but had no effect on the arousal caused by photostimulation during non-rapid eye movement sleep. Postmortem histology showed that neurons expressing Channelrhodopsin2-mCherry were predominantly catecholaminergic (81%). These results show that selective activation of C1 and retrotrapezoid nucleus neurons produces state dependent arousal and cardiorespiratory stimulation. These neurons, which are powerfully activated by chemoreceptor stimulation, may contribute to the sleep disruption associated with obstructive sleep apnea.

INTRODUCTION

Hypoxia, hypercapnia and asphyxia during sleep produce both cardiorespiratory stimulation and arousal. The respiratory stimulation and arousal are life-saving in case of airway obstruction but the sleep fragmentation and intermittent hypoxia associated with chronic obstructive sleep apnea (OSA) create a cohort of acute and chronic cardiovascular and other health problems ¹. The mechanisms responsible for the cardiorespiratory stimulation cause by hypoxia and hypercapnia are reasonably well understood but the neural mechanisms of arousal are still obscure ².

The C1 adrenergic neurons and the retrotrapezoid nucleus (RTN) located in the rostral ventrolateral medulla (RVLM) play a pivotal role in the cardiorespiratory responses to hypoxia and hypercapnia ^{3, 4}. C1 neurons regulate sympathetic vasomotor tone and are excited by hypoxia and to a lesser extent by hypercapnia ^{3, 5}. C1 neurons have extensive central nervous system (CNS) projections ⁶ to regions of the brain that regulate sleep and arousal ⁷ and therefore could contribute to hypoxia-induced arousal. RTN neurons are putative central chemoreceptors ⁴ that innervate respiratory centers ⁸ and mediate around 60% of the hypercapnic respiratory chemoreflex in conscious rats ⁹. RTN neurons are also activated by hypoxic stimulation of the carotid bodies ¹⁰.

Combined optogenetic stimulation of C1/RTN neurons elicits a rapid, robust cardiorespiratory response in awake rats ¹¹ that is similar to the response elicited by asphyxia in humans ¹². The present study seeks to determine whether such stimulations also produce arousal from sleep. The second objective is to test whether the cardiorespiratory responses elicited by activating these neurons are sleep state-dependent.

METHODS

All experiments were conducted using male Sprague-Dawley rats (N=30; 364 ± 7 gm at the time of experimentation, Taconic, USA) in accordance with NIH Guide for the Care and Use of Laboratory Animals and approved by the University of Virginia Animal Care and Use Committee. An expanded methods section is available in the online data supplement.

RESULTS

Photostimulation of ChR2⁺ C1/RTN neurons during NREMS causes arousal

Unilateral photostimulation (20 Hz, 20 s trains, 5–10 ms pulses) of C1/RTN neurons in ChR2⁺ rats (injected with PRSx8-ChR2-mCherry¹³ into RVLM) during non-rapid eye movement sleep (NREMS) produced immediate respiratory stimulation followed by arousal (Figure 1A). Arousal events consisted of an abrupt and sustained decrease in electroencephalographic (EEG) slow wave activity and an increase in high frequency oscillations (Figure 1A, S1A). Neck electromyographic (EMG) activation was not always detected, but in many cases either brief EMG bursts associated with large augmented inspiratory efforts (i.e. sighs) and/or persistent increases in EMG tone were observed. In control rats (injected with saline or PRSx8-AllatoR-eGFP¹³ into RVLM), respiration was unaffected by RVLM photostimulation during NREMS (Figure S1B). The arousal probability during photostimulation was significantly greater in ChR2⁺ rats than controls for frequencies of 10 Hz or greater (P<0.001 for the interaction between the presence of ChR2⁺ and stimulation frequency, N=17 ChR2⁺ and 8 controls) (Figure 1B). At a stimulus frequency of 20 Hz, 83.0 ± 14.7% of trials in ChR2⁺ rats resulted in arousals within the 20 s stimulus, in contrast to 23.1 ± 14.3% of trials in control rats (P<0.001) (Figure 1C). The earliest signs of cortical desynchronization could be observed within 1 s of the stimulus onset in some trials, with 68.6 ± 2.8% of arousal events occurring in the first 10s of 20 Hz photostimulation trials (Figure 1C). In control rats, arousal events were evenly distributed across the photostimulation period (Figure 1C), implying that these arousals were spontaneous (i.e. unrelated to the photostimulation).

Figure 1. Photostimulation of ChR2⁺ C1/RTN neurons during non-rapid eye movement sleep (NREMS) causes arousal.

A, Photostimulation trial during NREMS in a ChR2⁺ rat. Open arrowheads indicate first signs of cortical desynchronization. Inset. Expanded view of the initial respiratory effects and cortical desynchronization in A. Flow- barometric plethysmographic recording of breathing, EEG- electroencephalograph, EMG-neck electromyograph. Y-axis: Flow- 50 ml/s, EEG and EMG- 0.5 mV. B, The probability of arousal during photostimulation in NREMS in ChR2⁺ and control rats (N= 17 ChR2⁺ and 8 control rats, **P<0.01, ***P<0.0001). C, Cumulative frequency of arousals during photostimulation in NREMS (1 s bins, N= 12 ChR2⁺ and 8 control rats).

Photostimulation of ChR2⁺ C1/RTN neurons during REMS causes arousal

Photostimulation during REMS produced arousal in ChR2⁺ rats but less reliably than during NREMS. Arousal events during REMS trials consisted of abrupt EEG desynchronization, increases in neck EMG tone and EMG bursts associated with movement, and in some cases a delayed sigh (Figure 2A, S1C). Arousal events during photostimulation in REMS were rare in control rats (Figure S1D). At a stimulus frequency of 20 Hz, 41.9 ± 5.6% of photostimulation trials in ChR2⁺ rats resulted in arousal from REMS within the 20 s stimulus (vs. 16.3 ± 2.8% in control rats; P<0.01, N=13 ChR2⁺ and 8 control rats) (Figure 2B). However, the effectiveness of high frequency photostimulation in evoking arousal in ChR2⁺ rats was significantly less during REMS than NREMS (P<0.001 for the interaction effect between sleep state and stimulation frequency in ChR2⁺; compare Figure 1B,C with Figure 2B,C). Unexpectedly, there was no correlation between the probability of arousal during photostimulation in NREMS and REMS in ChR2⁺ rats (Pearson r=0.08, P=0.12) raising the possibility that different arousal mechanisms or pathways are engaged by the photostimulation during NREMS vs. REMS.

Figure 2. Photostimulation of ChR2⁺ C1/RTN neurons during rapid eye movement sleep (REMS) causes arousal.

A, As in Figure 1A except during REMS. B, The probability of an arousal during photostimulation in REMS in ChR2⁺ and control rats (N= 13 ChR2⁺ and 8 control rats, **P<0.01). C, Cumulative frequency of arousals during photostimulation in REMS (1 s bins, N=12 ChR2⁺ and 8 control rats).

In quietly resting awake rats, photostimulation did not produce overt behavior or detectable changes EEG or EMG activity. Specifically, photostimulation in awake rats produced no change in total EEG power (P=0.68) or within any spectral band evaluated (P=0.99).

The respiratory stimulation produced by activating ChR2⁺ C1/RTN neurons is sleep state dependent

Consistent with other reports ^{14, 15}, we observed significant sleep state differences in respiratory frequency (f_R), estimates of tidal volume (V_T) and minute volume (MV) measured using whole-body unrestrained plethysmography (Table S1). Photostimulation in ChR2⁺ rats during quiet wakefulness and NREMS increased f_R, V_T and MV to a similar extent (episode 1 (NREMS) and episode 5 (quiet wakefulness) in Figure 3A, S2A), but these effects were greatly suppressed during REMS (episodes 2 and 3 in Figure 3A, S2A; group data in Figure 3B, S2B). On average, photostimulation at 20 Hz increased MV by 132 ± 50 % of resting values during quiet wakefulness and 124 ± 40% during NREMS but only by 32 ± 29% during REMS (P<0.0001 for REMS vs. NREMS and quiet wakefulness at 20 Hz between sleep state and stimulation frequency) (Figure 3B). Interestingly, stimulus-evoked respiratory activation instantly reappeared with the first signs of cortical desynchronization during REMS trials resulting in arousal (inset of Figure 2A). This observation shows that the absence of the respiratory response to photostimulation of ChR2⁺ neurons is temporally linked with the cortical indications of REMS.

Figure 3. The respiratory effect of ChR2+ C1/RTN neuron photostimulation is sleep state dependent.

A, Photostimulation trials (20 Hz) spanning 3 arousal states (1: NREMS to wake, 2: REMS, 3: REMS, 4: REMS to wake, 5: wake). Note that the ventilatory response is attenuated during REMS trials. Y-axis: Flow- 50 ml/s, EEG and EMG- 0.5 mV. B, Changes in minute volume (MV) during photostimulation in waking, NREMS, and REMS in ChR2⁺ and control rats (*P<0.05, **P<0.01, ***P<0.001, repeated measures two-way ANOVA comparing sleep state & stimulus frequency in ChR2⁺ rats. ### P<0.001, two-way ANOVA comparing the presence of ChR2 & sleep state at a stimulus frequency of 20 Hz.

Arousal from NREMS is correlated with the changes in ventilation

Prior studies have shown a high degree of correlation between increased inspiratory efforts and arousal from sleep ^{2, 15}. Therefore, we determined whether the probability of arousal during photostimulation of ChR2⁺ C1/RTN neurons was correlated with degree of respiratory activation. These two variables were correlated during NREMS (Pearson r=0.78, P<0.0001), but not during REMS (Pearson r=0.24, P=0.21) (Figure 4A, B).

Figure 4. Arousal during ChR2⁺ C1/RTN neuron photostimulation is correlated with changes in ventilation during NREMS, but not REMS.

A–B, X-Y scatter plots of the change in MV and the probability of arousal during 2, 10 and 20 Hz photostimulation trials in ChR2⁺ rats during NREMS (A) and REMS (B). (N=12 for NREMS and 10 for REMS). Correlation determined using Pearson’s correlation calculation.

The cardiovascular effects of ChR2⁺ C1/RTN neuron photostimulation are sleep state dependent

These studies were performed in 6 fully instrumented ChR2⁺ rats in which arterial pressure (AP) was recorded from the femoral artery with an implanted telemetry device. We applied 3 s high frequency stimulus trains at 0.01 Hz to compare the AP and heart rate (HR) response elicited during waking, NREMS and REMS. These trains produced biphasic reciprocal changes in mean AP (MAP) and HR (Figure 5A). The initial pressor effect of photostimulation was significantly smaller during REMS than during NREMS (P<0.01) and quiet wakefulness (P<0.001) (Figure 5B, S3A). All other features of the MAP and HR response were similar between arousal states. This evidence shows that the cardiovascular effects elicited by photostimulating ChR2⁺ C1/RTN neurons are depressed during REMS in a similar fashion to the breathing effects.

Figure 5. The cardiovascular effect of ChR2+ C1/RTN neuron photostimulation is sleep state dependent.

A, Effects of 3 s photostimulation trials (20Hz, 5 ms) on arterial pressure (AP) and heart rate (HR) during each arousal state in ChR2⁺ rats. Y-axis: Flow- 50 ml/s, EEG and EMG- 0.5 mV. B, Normalized time course changes in MAP and HR during a 3 s stimulus trains during each arousal state in ChR2+ rats (N=6).

Photostimulation produces sighs associated with signs of arousal

Details available in an online data supplement.

Alpha1-adrenoreceptor blockade does not prevent arousal, but attenuates the cardiorespiratory activation during photostimulation

To test whether the rise in AP might contribute to arousal during RVLM photostimulation, we administered the selective α1-antagonist prazosin (1mg/kg, I.P.) (Figure 6). Administration of prazosin blocked the stimulus-evoked increase in MAP (P<0.0001, Figure 6A,B), had no significant effect on HR (P=0.09) (Figure 6A,B), and reduced the increase in MV by 43.8% (P<0.01), owing primarily to a 46.5% reduction in the f_R response (P<0.05) (Figure 6A,C). Treatment with prazosin did not significantly affect the occurrence of arousals during photostimulation (pre-drug vs. post-drug: 92.1 ± 5.0% vs. 76.9 ± 10.4% of trials, P=0.27) (Figure 6D). Thus central and peripheral α1-adrenoreceptor blockade attenuates the cardiorespiratory effects, but not the arousal effects of ChR2⁺ C1/RTN neuron activation.

Figure 6. Alpha1-adrenoreceptor antagonism does not prevent arousal, but attenuates the cardiorespiratory activation during photostimulation.

A, Effects of photostimulation before (left) and after (right) administration of prazosin (1mg/kg, I.P.). Y-axis: Flow- 35 ml/s, EEG and EMG- 0.25 mV. B–D, Average cardiovascular (B), respiratory (C) and arousal response to photostimulation before (open columns) and after prazosin treatment (closed columns) (N=6, * P<0.05, ** P<0.01, *** P<0.001, paired Student’s t-test).

Histological results

The number and location of ChR2⁺ neurons was mapped in each rat by identifying ChR2-mCherry by immunohistochemistry (Figure 7A, S5). On average, mCherry was detected in 132 ± 10 neurons counted in a 1:6 coronal series (N=22, approximately 792 neurons per rat without correction). Of the neurons expressing ChR2-mCherry, 80.7 ± 2.1 % had detectable tyrosine hydroxylase (TH)-immunoreactivity (range: 57.0–94.6%) accounting for 59.8 ± 2.5% of TH-immunoreactive neurons in counted sections (range: 28.3–80.5%). The number of ChR2⁺ neurons was weakly correlated with the change in MV caused by photostimulation during NREMS (Pearsons r=0.37, P=0.034), but there was no correlation between the number of ChR2⁺ neurons and the probability of arousal during NREMS (Pearsons r=0.01, P=0.63) or REMS (Pearsons r=0.01, P=0.76). The fiber optic tip was typically located within 500 μm of the ventral surface of the brain and dorsal to the bulk of the ChR2-mCherry neurons (Figure 7B).

Figure 7. ChR2-mCherry is preferentially expressed by C1 and RTN neurons.

A, Rostro-caudal distribution of neurons expressing ChR2-mCherry, tyrosine hydroxylase (TH) or both in the rostral ventrolateral medulla (N=22 ChR2⁺ rats). B, Location of the fiber placement in all experiments (N= 22 ChR2⁺ and 8 control rats). Scale bar: 0.5 mm.

Discussion

The key novel observation is that selective stimulation of C1/RTN neurons is sufficient to produce cortical desynchronization in sleeping rats in addition to cardiorespiratory activation. Both effects (arousal and cardiorespiratory stimulation) were attenuated during REMS.

Limitations

In this study, arousal was defined as an all-or-none event according to established criteria¹⁶, providing a proof of principle that C1/RTN neuron activation interrupts sleep in rodents. Activating C1/RTN neurons may also evoke graded activation of sub-cortical and cortical regions that would not have been reflected in gross EEG recordings. Variations in sleep history (e.g. sleep deprivation), sleep-epoch duration, and response habituation due to repeated photostimulation trials could also influence the probability of arousal during photostimulation.

ChR2-based optogenetics allows reproducible and precisely timed activation of subsets of genetically coded neurons in conscious rats, which could not be accomplished with prior technology, especially in the lower brainstem. However, this approach has limitations. For example, excitation imposed by optogenetic stimulation may not precisely replicate the effect of a naturalistic stimulus, e.g. hypoxia. Also, it is possible that a change in the excitability of C1/RTN neurons between sleep states modulates ChR2-mediated excitation. Furthermore, our ChR2-targeting strategy relies on the established selectivity of the PRSx8 promoter for Phox2-expressing neurons ^{13, 17}. Most ChR2⁺ neurons were TH-immunoreactive (80.6%), therefore C1 neurons. TH-negative neurons were presumed to be RTN cells based on location and appearance. We found no correlation between the relative number of each population expressing ChR2-mCherry and the effects of photostimulation. Thus, the physiological effects observed in this study could be driven by activation of either C1 or RTN neurons.

Mechanisms of arousal during activation of ChR2⁺ C1/RTN neurons

The arousal-promoting effects of ChR2⁺ C1/RTN neuron photostimulation could be due to two classes of mechanisms, singly or in combination: sensory feedback (e.g. baroreceptors, airway and lung afferents or chest proprioceptors) or direct activation of CNS mechanisms promoting arousal².

Acute changes in AP cause arousal from sleep in lambs and humans ^{18, 19}, which can be eliminated by sinoaortic denervation ¹⁸ implicating the baroreceptors or the carotid bodies in this effect. In the present study, arousal elicited by photostimulating ChR2⁺ C1/RTN neurons was not significantly reduced by eliminating the AP rise with prazosin suggesting that baroreceptor activation contributed minimally to the arousal. Furthermore, the persistence of the arousal in prazosin-treated rats suggests that this effect does not rely principally on the release of catecholamines, unlike the arousal caused by selective stimulation of the locus coeruleus ²⁰.

The correlation between the intensity of the breathing response and the arousal caused by photostimulation during NREMS suggests that sensory afferent activation secondary to increased breathing effort could be responsible for arousal during NREMS. However, direct CNS connections between C1/RTN neurons and wake-promoting networks would provide an equally satisfactory explanation for the correlation between arousal and changes in breathing during stimulation. In support of a CNS mechanism, prazosin reduced the respiratory effects of photostimulation by half without significantly changing arousal probability during NREMS, and photostimulation during REMS produced arousal despite greatly attenuated respiratory stimulation. This evidence is suggestive but insufficient to parse out the relative contribution of peripheral respiratory feedback vs. CNS mechanisms in the arousal elicited by activating C1/RTN neurons.

Arousal and cardiorespiratory activation evoked by ChR2⁺ C1/RTN neuron stimulation is sleep state dependent

REMS is generated by reciprocally connected ponto-medullary neurons ⁷ and is characterized by distinctive cortical, autonomic and respiratory activity and motor atonia ²¹. During REMS, the arousal response to hypoxia, hypercapnia, and upper-airway stimulation is attenuated in dogs ^22–24. The arousal response to hypoxia, hypercapnia, and upper-airway stimulation is not consistently blunted during REMS in healthy humans ^25–27, but obstructive events are more common and oxygen desaturation more severe during REMS in OSA patients ^{28, 29}. This has been attributed to an increase in arousal threshold during REMS relative to NREMS ³⁰. Here we show that direct activation of C1/RTN neurons is a less effective arousal stimulus during REMS. The reduced arousal during photostimulation in REMS may be related to a reduction in the activity of arousal-promoting networks, e.g. serotonergic neurons ³¹, or the absence of somatic feedback related to the blunted cardiorespiratory stimulation during REMS.

The notion that central cardiovascular and respiratory networks are less excitable during sleep is based on evidence that chemo- or somatic reflexes are state-dependent ^{2, 32}. Our study provides direct evidence that the excitability of central cardiorespiratory networks is markedly attenuated during REMS by directly monitoring the consequences of activating specific neurons involved in cardiorespiratory regulation.

During REMS, decreased excitability of respiratory and somatic motor neurons is caused by reduced monoaminergic tone and active inhibition mediated by cholinergic and GABA/glycine mechanisms ^{33, 34}. Motor atonia during REMS disproportionately affects non-diaphragmatic respiratory muscle groups, such as cranial motor neurons, and motor neurons controlling chest compliance, upper airway tone and active expiration ^{14, 15, 33}. Reductions in motor neuron excitability may explain the attenuated effects of photostimulation on tidal volume during REMS. However, the markedly reduced effect of photostimulation on breathing frequency suggests that the respiratory rhythm generator is also less excitable during REMS.

The pattern of sympathetic outflow changes predictably during REMS; muscle sympathetic nerve activity is elevated ^{35, 36}, whereas the splanchnic, renal and cardiac sympathetic nerve activity is reduced ^{37, 38}. Muscle sympathetic nerve activity is recruited during obstructive events in OSA patients ³⁵; however studies in lambs indicate that the hypertension caused by hypercapnia is reduced during REM-like sleep states ³⁹. Consistent with the latter, the pressor effect of photostimulation was attenuated during REMS. As stated earlier, these changes may reflect a reduction in the excitability of critical pathways in the effects of photostimulation or C1/RTN neurons themselves. Changes in monoaminergic tone at the level of the sympathetic preganglionic neurons (SPNs) ^{40, 41} during REMS may also explain the reduced hypertensive effects of photostimulation during this sleep state.

Perspectives

C1 and RTN neurons mediate a major portion of the autonomic and respiratory responses to hypoxia and hypercapnia in anesthetized rodents. We suggest here that the same neurons may also cause arousal, an effect that facilitates circulatory and respiratory responses to hypoxia and hypercapnia. By extension, we propose that C1/RTN neurons may be important in the cardiorespiratory effects and sleep fragmentation caused by sleep apnea in man. Furthermore, the blunted effects of C1/RTN neuron stimulation during REMS imply that the integration of excitatory drive in central cardiorespiratory networks is attenuated during REMS. A similar mechanism may explain the increased severity of apneic events during REMS in man.

Novelty and Significance.

What is new?

Selective optogenetic activation of C1and retrotrapezoid nucleus (RTN) neurons causes both cardiorespiratory stimulation and arousal in sleeping rats. The study also shows the reduced excitability of lower brainstem cardiorespiratory network during rapid eye movement sleep (REMS).

What is relevant?

Obstructive sleep apnea (OSA) causes hypoxia and hypercapnia resulting in cardiorespiratory activation and sleep disruption. Also, the duration of apnea and oxygen desaturation is both greater when apneas occur during REMS. The present study presents a brainstem pathway that may underlie the cardiorespiratory and sleep disruption associated with OSA and increased severity of apneic events during REMS.

Summary

Selective optogenetic activation of C1 and RTN neurons causes arousal and cardiorespiratory activation in rats. Both responses were attenuated during REMS. Arousals were correlated with the degree of respiratory activation during non-rapid eye movement sleep, and arousal persisted after pharmacological blockade of the blood pressure effects of stimulation.