Abstract
We previously showed that orexin neurons are activated by hypoxia and facilitate the peripheral chemoreflex (PCR)-mediated hypoxic ventilatory response (HVR), mostly by promoting the respiratory frequency response. Orexin neurons project to the nucleus of the solitary tract (nTS) and the paraventricular nucleus of the hypothalamus (PVN). The PVN contributes significantly to the PCR and contains nTS-projecting corticotropin-releasing hormone (CRH) neurons. We hypothesized that in male rats, orexin neurons contribute to the PCR by activating nTS-projecting CRH neurons. We used neuronal tract tracing and immunohistochemistry (IHC) to quantify the degree that hypoxia activates PVN-projecting orexin neurons. We coupled this with orexin receptor (OxR) blockade with suvorexant (Suvo, 20 mg/kg, i.p.) to assess the degree that orexin facilitates the hypoxia-induced activation of CRH neurons in the PVN, including those projecting to the nTS. In separate groups of rats, we measured the PCR following systemic orexin 1 receptor (Ox1R) blockade (SB-334867; 1 mg/kg) and specific Ox1R knockdown in PVN. OxR blockade with Suvo reduced the number of nTS and PVN neurons activated by hypoxia, including those CRH neurons projecting to nTS. Hypoxia increased the number of activated PVN-projecting orexin neurons but had no effect on the number of activated nTS-projecting orexin neurons. Global Ox1R blockade and partial Ox1R knockdown in the PVN significantly reduced the PCR. Ox1R knockdown also reduced the number of activated PVN neurons and the number of activated tyrosine hydroxylase neurons in the nTS. Our findings suggest orexin facilitates the PCR via nTS-projecting CRH neurons expressing Ox1R.
- breathing
- corticotropin-releasing hormone
- hypoxia
- nucleus of the solitary tract
- orexin
- paraventricular nucleus
Significance Statement
Previously we showed that orexin contributes to the peripheral chemoreflex (PCR), but the mechanisms underlying this effect remain unknown. Here we show that (1) orexin receptor blockade reduces the activation of the PVN and nTS; (2) hypoxia activates orexin neurons that project to the PVN, but not those projecting to nTS; (3) orexin receptor blockade reduces the activation of nTS-projecting corticotrophin releasing hormone (CRH) neurons in the PVN; (4) orexin 1 receptor (Ox1R) blockade and specific Ox1R knockdown in the PVN reduce the strength of the PCR, and (5) Ox1R knockdown reduces the number of activated PVN neurons and tyrosine hydroxylase neurons in the nTS. These findings suggest that PVN-projecting orexin neurons facilitate the PCR via Ox1R on nTS-projecting CRH neurons.
Introduction
The peripheral chemoreflex (PCR), originating from the carotid bodies, is critical for providing tonic drive to cardiorespiratory neural networks, as well as for increasing respiratory drive in response to hypoxia. Primary afferent information from the carotid bodies arrives at the nucleus of the solitary tract (nTS), where it is processed, integrated, and transmitted across cardiorespiratory networks. Neural circuits in the hypothalamus contribute significantly to cardiorespiratory function in both normal and pathophysiological conditions. Activation of the paraventricular nucleus of the hypothalamus (PVN) increases respiratory frequency in anesthetized and conscious animals under baseline conditions (Duan et al., 1997; Yeh et al., 1997; Schlenker et al., 2001). Early pharmacological and lesion studies showed that the PVN contributes to the respiratory and sympathetic responses evoked by PCR activation (Olivan et al., 2001; Reddy et al., 2005). Work from our group suggests the PVN acts through the nTS to augment the PCR. Reciprocal connections between the PVN and nTS are strongly activated by hypoxia (King et al., 2012, 2013; Ruyle et al., 2018). More recent studies showed that silencing nTS-projecting PVN neurons reduced the PCR-induced increase in phrenic and sympathetic activity, effects associated with reduced activation of nTS neurons (Ruyle et al., 2019, 2023). Together these data suggest that the ability of the PVN to facilitate the PCR heavily relies on projections to the nTS.
Orexin neurons reside in the dorsomedial (DMH), perifornical area (PeF), and lateral hypothalamus (LH), just caudal to the PVN (Tsunematsu and Yamanaka, 2012). The activity of orexin neurons is highly dependent on state (Lee et al., 2005) and circadian pattern, with their greatest activity in active wakefulness (Marston et al., 2008; Scammell and Winrow, 2011). Orexinergic neurons project widely, including to cardiorespiratory and autonomic nuclei (Carrive, 2013). They are also inherently CO2 sensitive (Nakamura et al., 2007), and several groups have shown that orexin makes a significant contribution to the hypercapnic ventilatory response (Deng et al., 2007; Nakamura et al., 2007; Li and Nattie, 2010), especially during active wakefulness when orexin neuronal activity is highest. We recently showed that hypoxia also activates orexin neurons and that orexin significantly augments the PCR in the active phase of the circadian cycle (Spinieli et al., 2022). Although the systemic administration of an orexin receptor (OxR) antagonist suppresses the PCR, the specific orexin receptors and neural circuits through which orexin acts are unknown. While direct injection of orexin into the pre-Botzinger complex (preBotC) or phrenic motor nucleus can increase phrenic nerve activity (Young et al., 2005), neural tract tracing experiments have shown that few orexin neurons project to these regions, especially the preBotC (Young et al., 2005), suggesting that endogenous orexin acts through other circuits to enhance the PCR.
Orexin neurons abundantly innervate the PVN (Peyron et al., 1998) and nTS (Zheng et al., 2005), implying that orexin has a substantial capacity for neuromodulation at these sites. In vitro studies have shown that exogenously applied orexin depolarizes both nTS (Yang et al., 2003; Yang and Ferguson, 2003) and PVN neurons (Follwell and Ferguson, 2002; Samson et al., 2002); in the PVN this includes a significant number of CRH-immunoreactive parvocellular neurons (Follwell and Ferguson, 2002) that express the orexin 1 receptor (Ox1R) (Backberg et al., 2002). Orexin depolarizes the vast majority of nTS neurons by inhibiting potassium conductance and activating a nonselective cationic conductance (Yang and Ferguson, 2003). Most of the nTS-projecting PVN neurons activated by hypoxia are immunoreactive for CRH (Ruyle et al., 2018), suggesting their involvement in the regulation of cardiorespiratory responses to hypoxia.
Here we address the hypothesis that orexin facilitates the PCR of conscious rats by facilitating the activation of the nTS via CRH neurons in the PVN. Our findings strongly suggest that orexin neurons facilitate the PCR through Ox1Rs on nTS-projecting CRH neurons.
Materials and Methods
Ethical approval
All animal experiments were approved by the University of Missouri Institutional Animal Care and Use Committee, and experiments were performed in accordance with the American Physiological Society's Guiding Principles for the Care and Use of Vertebrate Animals in Research and Training and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Animals
All experiments were conducted on adult male Sprague Dawley rats. The rats, 2–3 months of age (250–350 g), were obtained from Envigo. Rats were housed in a controlled environment with a 12 h light/dark cycle and maintained at 22°C. Rats had access to food and water ad libitum. We previously showed that orexin neurons make a stronger contribution to the PCR in the active phase of the rat circadian cycle. Therefore, all our experiments described here were conducted during the active phase. Rats were habituated to a room in which the light cycle was shifted 12 h (lights were turned on at 8 P.M., turned off at 8 A.M.). The experiments were performed between 12 P.M. and 3 P.M., which corresponds to 4–7 h into rats’ active phase (ZT 16–19). During the active phase experiments, the laboratory lights were turned off, and workstations were illuminated with red light to minimize disruption to the rat's natural circadian rhythms. In this study, we used suvorexant (Cayman Chemical Company) a potent and selective antagonist to both orexin receptor 1 (Ox1R) and 2 (Ox2R) (Cox et al., 2010). Suvorexant was administered via intraperitoneal injection at dosage of 20 mg/kg. Dose of suvorexant was based on previous studies (Sanchez-Alavez et al., 2019) as well as our own (Spinieli et al., 2022) to facilitate the interpretation of the data. Control rats were injected with vehicle (100% dimethyl sulfoxide; DMSO). We also used SB-334867 (Tocris), a selective Ox1R antagonist at a dosage of 1 mg/kg intraperitoneally; this dose was based on the original characterization of the antagonist (Smart et al., 2001).
Whole-body plethysmography
We used whole-body plethysmography to monitor respiratory and metabolic variables during experiments assessing the acute HVR, as well as for exposing rats to 2 h of hypoxia (or normoxia) to assess activation of our neurons of interest via immunohistochemistry (IHC). The volume of the chamber used was ∼15 L, constructed of plexiglass with a hole in the top to allow passage of intraperitoneal catheters for drug administration. Gas (either room air or hypoxia) was delivered to the chamber via pressurized cylinders. The pressure within the chamber was kept near-atmospheric using a standard lab bench vacuum. The plethysmographic chamber was connected to a differential pressure transducer (Validyne) to measure changes in pressure within the chamber, including those related to breathing. An O2 and CO2 gas analyzer (iWorx) continually measured CO2 and O2 concentrations within the chamber to allow calculations of metabolic CO2 production and O2 consumption. Chamber temperature and humidity were continuously monitored and values recorded for every rat to calculate tidal volume (VT). This calculation also required calibration injections of a known volume (1 ml) which were performed for each rat at the end of the experiment. All analog signals were converted to digital via a PowerLab data acquisition system (ADInstruments). All cardiorespiratory variables were monitored in real time via LabChart v8 (ADInstruments).
Ox1R knockdown
To reduce the expression of the orexin 1 receptor (Ox1R) in PVN, we nanoinjected an AAV2 expressing either a scrambled RNA (scRNA) or a short hairpin RNA (shRNA) targeting the Ox1R (Vector Biolabs) into PVN, as done previously (Zhou et al., 2019). Both AAVs also expressed GFP. We used the same surgical and microinjection procedures as used for injecting retrograde tracers. Briefly, glass micropipettes containing either the scRNA (n = 10) or shRNA (n = 13) were nanoinjected bilaterally into the PVN (200 nl/side; coordinates: 1.8–2.0 mm caudal to bregma, 0.5 mm lateral from midline, and 7.6–7.8 mm ventral to the dura). AAVs were allowed 3–4 weeks for optimal Ox1R knockdown prior to chemoreflex testing.
Chemoreflex testing
We tested the HVR in rats following acute, systemic blockade of OxRs and in separate groups of rats 3–4 weeks after having the PVN nanoinjected with sc- or Ox1R shRNA-expressing AAVs. All HVRs were tested in the active phase. Prior to testing, rats were lightly anesthetized with isoflurane to allow exteriorization of the arterial catheter. The plethysmographic chamber was maintained at ∼25°C throughout the experiment. Normoxic (21% O2, balance N2) or hypoxic gas (97% N2, 3% CO2) was delivered to the chamber from premixed cylinders. As we have shown previously (Spinieli et al., 2022), the small amount of CO2 added to the hypoxic gas minimized the hyperventilation-induced fall in arterial PCO2 which would have inhibited breathing. FIO2 fell from 0.21 to 0.10, and FICO2 increased from 0 to 0.03 over a period of ∼10 min. We utilized a fast wash-in time to isolate the PCR and avoid secondary effects of hypoxia on metabolic rate or body temperature (Spinieli et al., 2022). After a 15 min settling period, a baseline recording was made over 30 min in room air. For sc- and shRNA rats, gas entering the chamber was then switched from the room air cylinder to the hypoxic cylinder using a Hypoxydial (STARR Life Sciences). When FIO2 reached 0.09, the gas was switched back to the air cylinder to increase FIO2 back to 0.21. For global assessment of Ox1R on the HVR, rats were first injected with vehicle (∼40 µl 100% DMSO), followed by the first HVR test. Rats were allowed to recover for 30 min in room air, injected with ∼30–40 µl of 10 mg/ml SB-334867 (1 mg/kg), and then had their HVRs tested a second time.
Assessing the activation of neurons in PVN, nTS, and VRG by hypoxia
Conscious rats were first acclimated to the whole-body plethysmography chamber for 2 h/day over 2 consecutive days. Acclimation also involved intraperitoneal injections of normal saline on each day. The following day, rats were injected with either suvorexant or vehicle and then placed in the plethysmographic chamber. Rats were exposed to either normoxia (Nx; FIO2 = 0.21) or hypoxia (Hx; FIO2 = 0.11; FICO2 = 0.02–0.03) for 2 h in the active phases (ZT 16–19). Rats are exposed to Hx for 2 h to ensure adequate time for robust expression of c-Fos (Morgan et al., 1987; Ruyle et al., 2018). An FIO2 of 0.11 was selected as this is the level of hypoxia at which suvorexant had its greatest effect on the HVR (Spinieli et al., 2022).
Immunohistochemistry
In the hypothalamus, we immunostained for c-Fos (an immediate early gene and used as a marker of neuronal activation), cells immunoreactive for CRH, oxytocin (OT), vasopressin (AVP), and orexin. Within the ventral respiratory column (VRC) we also stained for c-Fos, neurokinin-1 immunoreactive regions (a general marker for respiratory regions), and tyrosine hydroxylase-immunoreactive (TH-IR) cells in the nTS to quantify the degree of activation for each neuronal phenotype. Immediately after normoxic or hypoxic exposure, rats were deeply anesthetized with 5% isoflurane followed by transcardial perfusion. Cold heparinized phosphate-buffered saline (PBS; ∼500 ml, 10 IU/ml heparin), pH 7.4, was first injected, followed by 500 ml of cold 4% paraformaldehyde (PFA), pH 7.2. After the perfusion, the brains were harvested, postfixed overnight in 4% PFA, and stored in PBS. Coronal section of the forebrain (40 µm), containing the hypothalamus, and hindbrain containing the nTS and VRC, respectively, were cut using a vibratome (1000S, Leica) with tissues stored in a cryoprotectant until time of IHC. One in every six consecutive sections was processed for immunoreactivity.
All immunohistochemical procedures were carried out using protocols similar to those previously described (King et al., 2012). In addition, all antibodies used in immunohistochemistry protocols were verified in previously published studies (King et al., 2012, 2013; Zhou et al., 2015, 2019; Coldren et al., 2017; Ruyle et al., 2018; Spinieli et al., 2022) and/or by the vendor using Western blots. Briefly, sections were rinsed with 0.01 M PBS (three times for 10 min), blocked in 10% normal donkey serum (Millipore, S30) in 0.3% Triton-0.01 M PBS (PBS-T), and incubated overnight in 3% normal donkey serum and 0.3% PBS-T containing the primary antibodies. c-Fos was detected using goat (1:500, sc-52, Santa Cruz) or rabbit anti-c-Fos (1:3,000, ab190289, Abcam). Sections from the hypothalamus were incubated with guinea pig anti-CRH (1:2,000, T-5007, Peninsula Laboratories), rabbit anti-oxytocin (1:2,000, 20068, Immunostar), guinea pig anti-vasopressin (1:1,000, T-5048, Peninsula Laboratories), or rabbit anti-orexin (1:3,000, H-003-30, Phoenix Pharmaceuticals). Ox1R in the PVN was labeled with rabbit anti-Ox1R (1:300, AOR-001, Alomone Labs). In addition to primary antibodies for c-Fos, sections from the nTS and VRG were incubated with mouse anti-TH (1:1,000, MAB318, Millipore) and rabbit anti-neurokinin-1 receptor (NK1R) immunoreactive (1:1,000, AB5060, Millipore). Because in some cases c-Fos may be expressed in astrocytes or microglia (Cruz-Mendoza et al., 2022), we qualitatively confirmed that c-Fos expression was confined to neurons but not other cell types. To do this, we incubated sections from the VLM, nTS, and PVN overnight with the c-Fos primary antibody, along with the following primary antibodies: mouse anti-NeuN to label neurons (1:500, MAB377, Millipore), guinea pig anti-glial fibrillary acidic protein (GFAP) to label astrocytes (1:500, 173004, Synaptic Systems), and guinea pig anti-ionized calcium-binding adapter molecule-1 (IBA-1) to label microglia (1:1,000, 234 004; Synaptic Systems). For all three regions, we could find no clear evidence of c-Fos expression in any cells other than neurons [see Extended Data Figs. 1-1 (VLM), 2-1 (nTS), and 4-1 (PVN)].
After 24 h with primary antibodies, sections were rinsed in PBS and incubated for 2 h with appropriate secondary antibodies conjugated to Cy2, Cy3, or Cy5 (1:200, Jackson ImmunoResearch), with 1% NDS in 0.3% Triton-0.01 M PBS. Sections then were rinsed and mounted on gel-coated slides and coverslipped with Prolong Diamond (Thermo Fisher, P36970).
Neural tract tracing
Retrograde tracers were nanoinjected either into the PVN or the nTS of separate groups of rats, in order to label orexin neurons projecting to each nucleus, similar to our previous studies (King et al., 2012, 2013; Ruyle et al., 2018). All recovery surgical procedures were performed by using an aseptic technique. Adult male Sprague Dawley rats (n = 20) were deeply anesthetized with isoflurane (5% induction; 2–2.5% maintenance, Aerane; Baxter), placed in a stereotaxic apparatus (Kopf Instruments). For injections into the PVN, a midline incision was made along the dorsal surface of the skull, with muscle and fascia were bluntly dissected to visualize bregma and lambda and the head position adjusted so that bregma and lambda were oriented in the same horizontal plane. A small hole was drilled in the skull to expose the surface of the brain. For targeting the nTS, the brainstem was exposed via a limited occipital craniotomy. Glass micropipettes containing either retrograde tracer cholera toxin subunit B conjugated to AlexaFluor 555 (CTB; 1% in deionized water, List Biological Laboratories) or latex beads (RetroBeads, Lumafluor) were nanoinjected into the PVN (n = 8) or nTS (n = 12) bilaterally (200 nl for PVN; 60 nl for nTS) using the following coordinates for PVN, 1.8–2.0 mm caudal to bregma, 0.5 mm lateral from midline, and 7.6–7.8 mm ventral to the dura (Ruyle et al., 2018), and for nTS, 0.4 mm rostral and ±0.4 lateral from the caudal pole of the area postrema, designated as calamus scriptorius (CS), and 0.4 mm ventral from the surface (Ruyle et al., 2018).
Retrograde tracers were injected over 1 min and the volume injected quantified by monitoring the movement of the meniscus within the pipette, as previously described (King et al., 2013). The pipette remained in the tissue for at least 5 min to minimize movement of tracer up the injection tract. The pipette was then removed and the incision site closed. Rats were treated postoperatively with fluids (3 ml 0.9% saline, s.c.), enrofloxacin (5 mg/kg, s.c.), and caprofen (5 mg/kg, s.c.), for hydration, prevention of infection, and pain management, respectively. Following termination of anesthesia, rats were monitored until alert and then returned to their cages. The animals were allowed 7–10 d for recovery and for transport of the retrograde tracer. Rats were then acclimated to the whole-body plethysmography chamber for 2 h/day over two consecutive days. The following day, during the active phase, different groups of rats were injected with vehicle or suvorexant (20 mg/kg, i.p.) and then placed in the plethysmograph. Rats were exposed to either room air or hypoxia (FIO2 = 0.11; FICO2 = 0.03) for 2 h to induce robust c-Fos expression. For all rats, PVN and nTS injection sites were evaluated by assessing the intrinsic fluorescence of the retrograde tracers and using standard published landmarks for each site (Paxinos and Watson, 2007).
Quantifying OxR1 expression in the PVN
Under anesthesia induced by 2% isoflurane, the rats which had been injected with either AAV-OX1R-shRNA or scrambled shRNA were quickly decapitated. Bilateral PVN tissue samples were obtained from rat brain slices using micropunch method, as done previously (Zhou et al., 2019). The proteins were extracted in the presence of a mixture of protease inhibitors. Tissue samples were homogenized, sonicated on ice, and centrifuged at 12,000×g for 30 min at 4°C. Protein concentrations were determined by a Pierce Rapid Gold BCA Protein Assay Kit (Thermo Scientific). The proteins were denatured by SDS protein gel loading solution and subjected to 5–10% SDS polyacrylamide gel electrophoresis. Separated proteins were transferred to a PVDF membrane followed by blockage with 3% BSA overnight at 4°C. The membrane was incubated with primary antibodies against OxR1 (1:200, Rabbit anti-Orexin 1 receptors; Alomone Labs) and rabbit anti-rat glyceraldehyde 3-phosphate dehydrogenase (1:500, Abcam) for 3–4 h at room temperature. Membranes were then washed four times in TBS solution and incubated with horseradish peroxidase-conjugated anti-rabbit IgG (1:8,000, Thermo Fisher Scientific, catalog #31460) for 2 h at room temperature. An ECL kit (Thermo Scientific) was used to detect and quantify the densities of protein bands using an Odyssey Fc Imager (LI-COR Biosciences).
Microscopy and image analysis
A rat brain atlas was used to determine the appropriate levels, relative to bregma, within the hypothalamus and brainstem (Paxinos and Watson, 2007). For all animals, sections containing the hypothalamus, nTS, and VRG were examined using a fluorescence microscope (BX51; Olympus) equipped with a digital monochrome camera (ORCA-ER; Hamamatsu) and a spinning disk confocal unit (Olympus). Each image consisted of stacks of 11 optical planes (2 μm between planes). Image stacks were imported into ImageJ (version 1.48v). Quantification of positively labeled cells from each of the immunohistochemical protocols was performed using ImageJ and adjusted for contrast and brightness only. We examined sections from animals exposed to either normoxia or hypoxia after either intraperitoneal injection of suvorexant or vehicle.
Cell counts
For each nucleus (VRG, nTS, and PVN), unilateral counts as a total of all sections counted or at each rostral-caudal level were averaged across all animals. c-Fos-IR neurons were manually counted using ImageJ software, identified as round- or ovoid-shaped nuclear staining with a visible nucleolus, indicating activation.
VRG
The number of VRG neurons activated (i.e., c-Fos-IR) was conducted at four different levels: caudal ventral respiratory group (cVRG), rostral ventral respiratory group (rVRG), preBotC, and Botzinger complex (BotC). These levels were determined based on their relative positions to bregma as previously described (Brown et al., 2019). In summary, ventral brainstem sections were selected based on prominent structures, including the nucleus ambiguus, pyramidal tract, inferior olive, hypoglossal nucleus, and nTS. These structures served as landmarks for accurate identification of the sections. To verify the accuracy of the selected VRG regions, staining for the NK1R, a marker of the VRG, was performed (Wang et al., 2001). NK1R-IR regions were identified by their characteristic neuronal network appearances with visible processes. We counted the number of c-Fos-IR neurons in the four regions in which there was visible NK1R-IR.
nTS
We evaluated the caudal nTS, the primary site of cardiorespiratory afferent projections. CS was used as reference point and designated as position “0.” Three sections of the nTS were evaluated relative to bregma (−14.46, −14.28, and −14.1 mm). These positions correspond to −180, 0, and +180 mm relative to CS, respectively. We focused on TH-IR neurons as they project to the PVN and contribute to the HVR (King et al., 2012). TH-IR neurons exhibited cytosolic labeling with visible processes and a blank nuclear region. Cells were considered double labeled when they met criteria for both c-Fos-IR and TH-IR under more than one filter set in the same plane of focus.
Orexin neurons
Rats were nanoinjected with retrograde tracers into either PVN or the nTS. Unilateral counts of positively labeled orexin neurons with c-Fos-IR and retrograde tracer were performed at specific levels relative to bregma (−2.76 and −3.0 mm) within DMH, PeF, and LH. Retrograde-labeled orexin neurons exhibited bright or punctate cytosolic labeling, indicating their connectivity to the PVN or nTS. Orexin neurons were considered projecting and activated when they met criteria for both c-Fos-IR and retrograde tracer under more than one filter set in the same focal plane. Orexin neurons that showed positive signal under all three filter sets (c-Fos-IR, retrograde tracer, and orexin labeling) were considered triple labeled (King et al., 2012).
PVN analysis
The evaluation of PVN neurons in the hypothalamus was conducted at four different levels: −1.4, −1.7, −1.9, and −2.1 mm from bregma as previously described (Ruyle et al., 2018). Counts for each section were averaged. CRH, OT, and AVP neurons exhibited bright cytosolic labeling with a blank nuclear region.
In separate groups of rats, the nTS was nanoinjected with retrograde tracer. Unilateral counts of positively retrograde fluorescent-labeled CRH and OT neurons with c-Fos-IR were performed at the same levels relative to bregma in PVN analysis. Cells were considered double labeled when they met the criteria for both retrograde tracer labeling and c-Fos-IR under more than one filter set in the same focal plane. Cells that showed positive signal under all three filter sets (retrograde tracer, c-Fos-IR, and CRH or OT labeling) were considered triple labeled (King et al., 2012).
Experimental design and statistical analyses
Experiments were designed to either test the HVR using plethysmography or expose rats to hypoxia for a period to induce sufficient c-Fos expression within the nuclei of interest (described above). The statistical analyses were performed using GraphPad Prism (version 8.0.2). The following tests were conducted to reveal significant differences between groups: Two-factor ANOVA was used to examine the effect of O2 level (Hx vs Nx) on (1) the average (per rat) number of activated (c-Fos-IR) neurons on VRG, nTS, and PVN neuronal phenotypes with and without OxR blockade; (2) the average number of activated, nTS-projecting CRH neurons with and without OxR blockade; and (3) the average number of activated PVN- and nTS-projecting orexin neurons. Two-factor repeated measures ANOVAs were used to evaluate the effects of OxR1 knockdown in PVN on the HVR, HR, and ABP responses to Hx, as well as the total number of activated nTS neurons, including those that are TH-IR.
Two-tailed Student's t tests were used to evaluate the effects of Ox1R knockdown on the total number of PVN neurons displaying c-Fos-IR, the number of PVN neurons expressing GFP (indicating successful transfection with sc- or shRNA expressing AAVs) as well as c-Fos. A two-factor ANOVA was used to examine the differences in the expression of OX1R protein (relative to GAPDH expression) in the PVN and hippocampus following the expression of sc- or shRNA against Ox1R in the PVN. Regression analysis was used to examine the relationship between Ox1R:GAPDH expression in the PVN and both the hypoxic ventilatory response and baseline MAP. All reported data are presented as mean values ± SD. Sidak's post hoc analyses were performed when interactions between factors existed to determine specific differences among groups. Statistically significant was considered at p values ≤0.05.
Results
Orexin facilitates the hypoxia-induced activation of neurons in the VRC
OxR blockade reduced the frequency response to hypoxia (drug effect: p = 0.0105; Fig. 1a), as we found previously (Spinieli et al., 2022). We also investigated the effect of endogenous orexin on respiratory neurons in the VRC of the medulla, utilizing NK1R-IR as a general marker of respiratory nuclei (Wang et al., 2001). We quantified c-Fos-IR in both caudal and rostral regions of the VRC, including caudal (cVRG) and rostral (rVRG) ventral respiratory groups, preBotC, and Botzinger complex (BC). OxR blockade had no significant effect on the number of activated neurons in normoxia but significantly reduced the total number of neurons activated by hypoxia throughout the VRC (drug × O2 level: p = 0.009; Fig. 1c). Analyzing the data from all regions together, hypoxia selectively increased the number of activated neurons in the VRC of rats given vehicle alone (drug × O2 level: p = 0.0094; post hoc comparing rats given vehicle in normoxia and hypoxia: p = 0.0029) and effect significantly reduced by OxR blockade (p = 0.0476; Fig. 1d). Together these data suggest that orexin facilitates the activation of neurons in the VRC during hypoxia.
Orexin receptor blockade reduces the hypoxia-induced activation of neurons in the ventral respiratory column (VRC). Rats were injected with either vehicle (DMSO) or 20 mg/kg suvorexant (Suvo) intraperitoneally and exposed to acute hypoxia (2 h at FIO2 = 0.11). a, Rats treated with suvorexant (black circles) had a reduced frequency response to hypoxia, as we have shown in previous studies. b, Hypoxia qualitatively increased the number of Fos-immunoreactive (IR) neurons (red nuclei indicated by arrows) in the neurokinin-1 receptor (NK1R)-IR regions (green) of ventrolateral medulla of rats given vehicle (top), but this effect was not apparent in rats given suvorexant (bottom; scale bar, 200 µm). Panels on the right side show higher magnification of boxed regions (scale bar, 100 µm). Tyrosine hydroxylase-IR soma and fibers are also shown (blue). c, Average number of Fos-IR neurons (per rat, unilaterally counted) in the VRC of vehicle and Suvo rats across four rostral-caudal regions of the VRC: caudal (cVRG) and rostral ventral respiratory group (rVRG), pre-Botzinger complex (PBC), and Botzinger Complex (BC). Taking all rostral-caudal levels together, hypoxia increased the average number of Fos-IR neurons in rats given vehicle but not those treated with Suvo (O2 level × drug: p = 0.009). d, Same data as in c, with the four levels combined. See Extended Data Figure 1-1, showing that c-Fos is expressed exclusively in neurons within the VRC.
Figure 1-1
Hypoxia activates neurons in the ventrolateral medulla. Shown are images of the ventrolateral medulla stained for c-Fos (red), a neuronal marker (NeuN; blue) and either (a) the astrocytic marker GFAP (green) or (b) the microglial marker IBA-1 (green). Regions denoted by dashed rectangle represented in additional images below at higher magnification, showing individual channels as well as merged images showing Fos with either NeuN or non-neuronal markers. Purple color in Fos + NeuN image indicates activated neurons. Arrows point to individual neurons showing co-labeling of Fos and NeuN. We could not identify any co-labeling of Fos with either non-neuronal marker. Scale bars = 200um. Download Figure 1-1, TIF file.
Orexin neurons facilitate the hypoxia-induced activation of the nTS
We investigated the effect of OxR blockade on the activation of neurons in the nTS. OxR blockade significantly reduced the number of activated nTS neurons in hypoxia but had no effect in normoxia [drug × O2 level: p = 0.0226; post hoc: p = 0.0290 (hypoxia); p = 0.6379 (normoxia); Fig. 2b, left]. OxR blockade had no significant effect on the absolute number of activated TH-IR neurons (Fig. 2b, middle). The proportion of TH-IR neurons activated by hypoxia was less than in controls but statistically this was not significant (drug × O2 level: p = 0.0824 Fig. 2b, right). Overall, these findings suggest that orexin facilitates the activation of nTS neurons by hypoxia.
Orexin receptor blockade reduces the activation of the nucleus of the solitary tract (nTS). Rats were injected with either vehicle (Veh; DMSO) or 20 mg/kg suvorexant (Suvo) intraperitoneally and exposed to acute hypoxia (Hx; 2 h at FIO2 = 0.11). a, Rats exposed to hypoxia following vehicle (top panels) had qualitatively more Fos-immunoreactive (IR) cells (red nuclei) compared with rats exposed to hypoxia following Suvo (bottom panels). Catecholaminergic TH-IR neurons are immunostained green. Scale bar, 100 µm. b, In normoxia (Nx) there were statistically the same number of Fos-IR (i.e., activated) nTS neurons in rats administered Veh versus those given Suvo. Following Hx, rats treated with Suvo had fewer activated nTS neurons (drug × O2 level: p = 0.0226; post hoc: p = 0.0226). b, middle, There was no difference between Veh and Suvo groups with respect to the number of activated TH-IR neurons in the nTS following Nx or Hx. b, right, Rats treated with Suvo tended to have a small proportion of activated TH-IR neurons following Hx (2-factor ANOVA; drug × O2 level: p = 0.0824). See Extended Data Figure 2-1, showing that c-Fos is expressed exclusively in neurons within the nTS.
Fig. 2-1
Hypoxia activates neurons in the nucleus of the solitary tract (nTS). Shown are images of the nTS stained for c-Fos (red), a neuronal marker (NeuN; blue) and either (a) the astrocytic marker GFAP (green) or (b) the microglial marker IBA-1 (green). Regions denoted by dashed rectangle represented in additional images below at higher magnification, showing individual channels as well as merged images showing Fos with either NeuN or non-neuronal markers. Purple color in Fos + NeuN image indicates activated neurons. Arrows point to individual neurons showing co-labeling of Fos and NeuN. We could not identify any co-labeling of Fos with either non-neuronal marker. Scale bars = 200um. Download Fig. 2-1, TIF file.
We nanoinjected CTB or retrobeads into the nTS to label projecting orexin neurons and assess the degree that hypoxia activates the nTS projection. A representative image of a coronal nTS section indicates that the injection site was localized to the nTS (Fig. 3a). We visualized and mapped the injection sites within the three targeted rostral-caudal levels of the nTS (Fig. 3a) showing that the injections were successful with some variability but minimal spread into adjacent regions. Approximately 25–30% of orexin neurons in the PeF, DMH, and LH project to the nTS (Table 1). A large proportion of nTS-projecting orexin neurons are activated even in normoxic conditions (>50% of projecting orexin neurons in DMH and PeF; Fig. 3b,c; Table 1). Hypoxia did not significantly increase the number or proportion of activated nTS-projecting orexin neurons (Fig. 3b,c; Table 1). The PeF made the largest contribution to the activated projection, with on average 13 activated out of 19 total nTS-projecting orexin neurons in normoxia and 15 activated neurons out of 21 total nTS-projecting orexin neurons in hypoxia (Table 1). Overall, in both normoxia and hypoxia, ∼70% of nTS-projecting orexin neurons in the PeF were activated (Table 1). These data suggest activated orexin neurons project to nTS but that this projection does not contribute significantly to the hypoxia-induced activation of the nTS.
Hypoxia does not increase the proportion of activated nTS-projecting orexin neurons. a, Representative image showing cholera toxin-B (CTB) injection site in the nTS (top panel; blue). AP, area postrema; cc, central canal. Bottom panel, Mapping of all injection sites at the three levels of the nTS. Coordinates (−13.8, −14.0, −14.3) are relative to bregma. b, Top left panel, The dorsomedial (DMH), perifornical (PeF), and lateral hypothalamus (LH) immunostained for Fos (red nuclei), CTB (blue) and orexin (green) following exposure to normoxia (Nx; 4× magnification; scale bar, 200 µm). Higher magnification (10×) image of the boxed region shown to the right (scale bar, 100 µm). Inset shows higher magnification (20×) of boxed region in the 10× image, showing two orexin neurons immunoreactive for Fos and CTB. Individual channels shown along the bottom. b, Bottom left panel, DMH, PeF, and LH staining in separate rats exposed to hypoxia (Hx). Panels and magnification the same as in top panels in b. c, Percentage of nTS-projecting orexin neurons activated in the DMH, PeF, and LH following Nx (open circles and bars; n = 3) and Hx (filled circles and bars; n = 4). There was no difference between Nx and Hx with respect to the proportion of nTS-projecting orexin neurons activated (2-factor ANOVA: p = 0.1161).
Summary of total activated (Fos) and/or nTS-projecting (CTB) orexin neurons
PVN-projecting orexin neurons are activated by hypoxia and facilitate the activation of the PVN
We then evaluated the effect of orexin on PVN neuronal activation by hypoxia. Qualitatively, OxR blockade reduced the number of c-Fos-IR PVN neurons following hypoxia (Fig. 4a). Acute hypoxia significantly increased the number of activated PVN neurons compared with rats exposed to room air (hypoxia: p < 0.0001; Fig. 4b). OxR blockade significantly reduced the number of activated PVN neurons in both normoxia and hypoxia (drug effect: p = 0.0029; Fig. 4a,b). These data suggest that orexin facilitates the activation of the PVN in normal conditions as well as the increase in PVN neuronal activation in response to hypoxia.
Hypoxia activates neurons in the PVN. a, Images of the PVN show activated (c-Fos immunoreactive) neurons following hypoxia in rats given vehicle (Hx-Veh; top) or 20 mg/kg suvorexant intraperitoneally (Hx-Suvo; bottom). b, Summary of the number of activated PVN neurons in normoxia (Nx; open circles; n = 7 Veh, 4 Suvo) and hypoxia (Hx; filled circles; n = 10 Veh, 10 Suvo); two-factor ANOVA; effect of drug: p = 0.0029. Data are averages ± SD. Numbers include rats used for retrograde labeling from nTS (Fig. 3). See Extended Data Figure 4-1, showing that c-Fos is expressed exclusively in neurons.
Fig. 4-1
Hypoxia activates neurons in the paraventricular nucleus (PVN). Shown are images of the PVN (region of PVN outlined by solid line) stained for c-Fos (red), a neuronal marker (NeuN; blue) and either (a) the astrocytic marker GFAP (green) or (b) the microglial marker IBA-1 (green). Regions denoted by dashed rectangle represented in additional images below at higher magnification, showing individual channels as well as merged images showing Fos with either NeuN or non-neuronal markers. Purple color in Fos + NeuN image indicates activated neurons. Arrows point to individual neurons showing co-labeling of Fos and NeuN. We could not identify any co-labeling of Fos with either non-neuronal marker. Scale bars = 200um. Download Fig. 4-1, TIF file.
To assess the possibility that an orexinergic projection facilitates the hypoxia-induced activation of the PVN, we nanoinjected fluorescent retrobeads into the PVN to label this pathway and quantified how many of these cells were activated by hypoxia. As we did for the nTS, we visualized and mapped our PVN injection sites, which consistently targeted the PVN with minimal spread into adjacent regions (Fig. 5a). A total of 12–35% of orexin neurons in the PeF, DMH, and LH project to the PVN (Table 2). In normoxia, 18 and 17% of PVN-projecting orexin neurons in the PeF and DMH, respectively, were activated. We could not identify any activated, projecting LH neurons in normoxia. Hypoxia significantly increased the proportion of activated, PVN-projecting orexin neurons in the PeF (to 65% of projecting orexin neurons), DMH (to 58% of projecting orexin neurons), and LH (to 17% of projecting orexin neurons; O2 level: p = 0.0002; Fig. 5b,c; Table 2). Taken together, our data suggest that hypoxia activates orexin neurons projecting to PVN.
Hypoxia increases the proportion of activated PVN-projecting orexin neurons. a, Representative image showing cholera toxin-B (CTB) injection site in the PVN (top panel; blue). 3V, third ventricle. Bottom panels, Mapping of all injection sites at the three levels of the PVN. Coordinates (−1.4, −1.7, −2.1 mm) are relative to bregma. b, Top left panel, The dorsomedial (DMH), perifornical (PeF), and lateral hypothalamus (LH) immunostained for Fos (red nuclei), CTB (blue), and orexin (green) following exposure to normoxia (Nx; 4× magnification; scale bar, 200 µm). Higher magnification (10×) image of the boxed region shown to the right (scale bar, 100 µm). Inset shows higher magnification (20×) of boxed region in the 10× image, showing orexin neurons immunoreactive for Fos and CTB. Individual channels shown along the bottom. b, Bottom left panel, DMH, PeF, and LH staining in separate rats exposed to hypoxia (Hx). Panels and magnification the same as in the top panel of b. c, Percentage of PVN-projecting orexin neurons activated in the DMH, PeF and LH following Nx (open circles and bars; n = 4) and Hx (filled circles and bars; n = 4). Hypoxia induced a significantly higher proportion of activated, PVN-projecting orexin neurons (2-factor ANOVA, effect of O2 level: p = 0.0002).
Summary of total activated and/or PVN-projecting orexin neurons
Orexin facilitates the hypoxia-induced activation of CRH, OT, and AVP neurons in the PVN
We investigated the effect of orexin on the activation of CRH, OT, and AVP neurons in the PVN in normoxia and hypoxia. Compared with normoxia, hypoxia activated more CRH neurons (O2 level p < 0.0001; Fig. 6a). OxR blockade significantly reduced the number of activated CRH neurons in both normoxia and hypoxia (by 37% in hypoxia; drug effect; p = 0.0035; Fig. 6a). Acute hypoxia also significantly increased the overall number of activated OT neurons (O2 level p < 0.0001; Fig. 6b). While OxR blockade had no effect on the number of activated OT neurons in normoxia, it significantly reduced their hypoxia-induced activation (drug × O2 level: p = 0.0112; post hoc in hypoxia: p = 0.0029; Fig. 6b). AVP neurons were also activated by hypoxia (O2 level p = 0.0176). Like its effect on CRH neurons, OxR blockade significantly reduced the activation of AVP neurons in both normoxia and hypoxia (drug effect: p = 0.0176; Fig. 6c). These findings demonstrate that orexin facilitates the activation of all three of these neuronal phenotypes in the PVN.
Orexin facilitates the activation of corticotropin-releasing hormone (CRH), oxytocin (OT), and vasopressin (AVP) neurons in the PVN. Shown are images (PVN level: −1.7 mm from bregma; 10× magnification) of CRH (a, pseudostained green), OT (b, yellow), and AVP (c, blue) neurons immunoreactive for c-Fos (red), following exposure to hypoxia and administered either vehicle (Hx-veh; left panels) or 20 mg/kg suvorexant intraperitoneally (Hx-Suvo; right panels). 3V, third ventricle. Scale bar, 100 µm. Individual channels and merged images shown in the insets, showing boxed region at higher magnification (40×). Right panels, Summary of the number of activated neurons of each phenotype following normoxia (Nx) or hypoxia (Hx). Suvo significantly reduced the number of activated CRH neurons in both Nx and Hx (2 factor ANOVA; drug: p = 0.0035), OT neurons in Hx (2-factor ANOVA; O2 level × drug: p = 0.0112), and AVP neurons in Nx and Hx (2-factor ANOVA; drug: p = 0.0176).
Orexin facilitates hypoxia-induced activation of nTS-projecting CRH neurons
To further address the concept that orexin facilitates the hypoxia-induced activation of the nTS via projecting CRH neurons in the PVN, we bilaterally nanoinjected retrobeads into the nTS to label projecting PVN neurons. We used IHC to stain CRH neurons and analyzed how many nTS-projecting CRH neurons were activated by hypoxia (via c-Fos-IR) and whether their activation was prevented by OxR blockade. We focused on nTS-projecting CRH neurons because, of all the PVN neuronal phenotypes, they make up the vast majority of PVN neurons activated by hypoxia (Fig. 6a). Moreover, others have shown that the majority of activated PVN neurons projecting to the nTS are immunoreactive for CRH (Ruyle et al., 2018). Consistent with Figure 6, compared with normoxia, hypoxia qualitatively and quantitatively increased the proportion of CRH neurons activated by hypoxia (compare Fig. 7a,b; Table 3), an effect that was reduced following OxR blockade (p = 0.001; Fig. 7c,d). Hypoxia significantly increased the proportion of nTS-projecting PVN neurons displaying c-Fos-IR (p = 0.0012; post hoc Nx-Veh vs Hx-Veh: p = 0.0011; Fig. 7e; Table 3), including projecting CRH neurons (p < 0.001; Fig. 7f; Table 3). In keeping with our previous data (Ruyle et al., 2018), CRH neurons made up ∼85% of the activated projection (Table 3). OxR blockade reduced the proportion of projecting neurons that were activated following hypoxia (p = 0.0149; Fig. 7e; Table 3), including projecting CRH neurons (p < 0.001; post hoc comparing Hx-Veh vs Hx-Suvo: p = 0.0003; Fig. 7f; Table 3). When examining the four levels of the PVN, OxR blockade significantly reduced the number of activated, nTS-projecting CRH neurons at levels −1.7 (post hoc comparing Hx-Veh with Hx-Suvo: p = 0.0052) and −1.9 (post hoc: p = 0.0006; Fig. 7g). A much smaller proportion of the hypoxia-activated projection to the nTS was composed of OT neurons. Across all four levels of the PVN, OxR blockade significantly reduced the proportion of nTS-projecting OT neurons activated by hypoxia (from 28 to 16%; drug: p = 0.0058).
Orexin facilitates the activation of corticotropin-releasing hormone (CRH) neurons that project to nTS. Shown are images of the PVN from rats exposed to normoxia (Nx, a), hypoxia after intraperitoneal Veh (Hx-Veh; b), and hypoxia after 20 mg/kg suvorexant (Hx-Suvo; c). Nuclei of activated (Fos-immunoreactive) cells are stained red, CRH-immunoreactive cells are stained green, and CTB-labeled cells are blue. Boxed region of 10× image shown at higher magnification (40×) in insets, showing images of individual channels and the merged image. Scale bar, 50 µm. Also shown are proportion of activated CRH neurons (d), all projecting neurons (e), and projecting CRH neurons (f) following Nx-Veh (open circles and bars), Hx-Veh (gray circles and bars), and Hx-Suvo (black circles and bars). Data in d–f are from the four PVN levels combined. g, Proportion of projecting CRH neurons at each of the four PVN levels that were activated in Nx-Veh, Hx-Veh, and Hx-Suvo. h, Proportion of neurons that did not project to NTS that were activated in each condition. i, Proportion of neurons at each of the four PVN levels that were activated in each condition. Data were analyzed with two-factor ANOVAs. Post hoc analyses indicated with p values.
Summary of number of activated and/or nTS-projecting CRH neurons
Hypoxia also activated CRH neurons that do not project to nTS (p = 0.0265; Fig. 7h). Taking all four PVN levels together, OxR blockade had no significant effect on hypoxia-induced activation of nonprojecting CRH neurons (post hoc, Hx-Veh vs Hx-Suvo: p = 0.1516; Fig. 7i). However, separate analyses of the levels individually suggest that orexin may promote the hypoxia-induced activation of nonprojecting CRH neurons at level −1.4 (post hoc: p = 0.008; Fig. 7i), where many neuroendocrine CRH neurons reside (Simmons and Swanson, 2009).
Orexin acts through Ox1R in the PVN to facilitate the HVR
Others have shown that CRH neurons in the PVN express Ox1R (Backberg et al., 2002) and that Ox1R in the PVN has a role in cardiorespiratory function (Zhou et al., 2019). We performed additional immunohistochemistry to confirm the expression of Ox1R by CRH neurons. Our immunostaining suggests that orexinergic fibers are in relatively close proximity to CRH neurons and a proportion of CRH neurons express Ox1R (Fig. 8a). To test the general function of Ox1R on the PCR, we examined the HVR in rats before and after systemic Ox1R blockade. Ox1R blockade significantly reduced the HVR (O2 level × drug: p = 0.0348; Fig. 8b), an effect that was especially evident at FIO2 of 0.13 (post hoc: p = 0.0012) and FIO2 of 0.11 (post hoc: p = 0.0009; Fig. 8b). Systemic Ox1R blockade inhibited the HVR through an effect on respiratory frequency (p = 0.0603; Fig. 8c) as well as tidal volume (p = 0.0009; Fig. 8d). Ox1R blockade had no influence on metabolic CO2 production (Fig. 8e), indicating that the effects of blockade on the HVR were not secondary to reduced metabolism.
Blockade of orexin 1 receptor (Ox1R) in PVN reduces the hypoxic ventilatory response. a, Upper and lower left, Images showing CRH neurons (green) in the PVN with orexin fibers (Ox; red) in close proximity. a, Upper right, Image shows cells that express Ox1R (red). Lower right, Merged image showing CRH neurons expressing Ox1R. b, Hypoxic ventilatory response (HVR) of rats following vehicle (Veh, open bars and circles; n = 6) was significantly higher compared with the response following administration of Ox1R antagonist, SB-334867 (SB, gray bars and black circles; same rats tested with Veh; 2-factor repeated measures ANOVA; O2 level × drug: p = 0.0348; post hoc: p = 0.0012 and p = 0.0009 at FIO2 = 0.13 and 0.11, respectively). The inhibition of the HVR following Ox1R blockade was due to the combination of a strong tendency for reduced respiratory frequency (f; drug: p = 0.0603; c) and significantly reduced tidal volume (VT; drug: p = 0.0009; d). Ox1R blockade had no effect on metabolic CO2 production (VCO2; e).
To examine the functional effect of Ox1R more specifically in the PVN, we used AAV-OxR1-shRNA to knock down the expression of Ox1R in the PVN. Western blot confirmed successful OxR1 knockdown in most, but not all, rats treated with AAV-OxR1-shRNA (Fig. 9a,b). As expected, our knockdown approach specifically targeted Ox1R in the PVN, as Western blot analysis also revealed that Ox1R expression within the hippocampus was not altered by shRNA administration in the PVN (Fig. 9a,b). We examined the effect of Ox1R knockdown in the PVN on the HVR and MAP response to hypoxia. In keeping with our data from rats treated systemically with Ox1R antagonist, rats that had shRNA against Ox1R targeted to the PVN had significantly reduced HVR (shRNA: p = 0.049; Fig. 9c). As the magnitude of the Ox1R knockdown varied somewhat among rats receiving shRNA in the PVN, we performed a regression analysis of the ventilatory response (at FIO2 of 0.13, where the effect of knockdown on the HVR appeared strongest) against the level of Ox1R expression. This analysis showed that the HVR was significantly correlated with the level of Ox1R expression (R2 = 0.50; p = 0.0153; Fig. 9d). Ox1R knockdown significantly reduced the frequency response to hypoxia (shRNA × O2 level: p = 0.0263; post hoc shRNA vs scRNA: p = 0.0008 at FIO2 = 0.13; Fig. 9e). Ox1R knockdown had no significant effect on the VT response to hypoxia (Fig. 9f). There was no influence of Ox1R knockdown on metabolic CO2 production (scRNA, 30.5 ± 2.4 ml/min/kg; shRNA, 30.3 ± 1.9 ml/min/kg), again indicating that the effects of knockdown on the HVR were not related to a reduced metabolic drive. Ox1R knockdown tended to reduce MAP at all O2 levels, but this effect did not quite reach statistical significance (shRNA, p = 0.0669; Fig. 9g). Nevertheless, there was a significant, positive correlation between Ox1R expression and MAP (R2 = 0.77; p = 0.0214; Fig. 9h). Ox1R knockdown had no influence on heart rate (data not shown). Our data suggest that orexin acting through the Ox1R (possibly on CRH neurons) facilitates the HVR and may also contribute to the maintenance of MAP.
Partial Ox1R knockdown in the PVN reduces the hypoxic ventilatory response. a, Western blot quantifying Ox1R expression in the PVN (top) and hippocampus (Hc, bottom) of rats having received scrambled RNA-expressing AAVs (sc1, sc2) or AAVs expressing short hairpin loop (sh)-RNA targeting Ox1R in the PVN. Bands are shown for both Ox1R and GAPDH, with densitometric quantification shown above each band. b, Summary data showing average Ox1R expression relative to GAPDH in the PVN and Hc (n = 5 scRNA; n = 6 shRNA) indicating successful knockdown of Ox1R in the PVN of some (but not all) rats given shRNA (2-factor repeated measures ANOVA; region × RNA: p = 0.10). c, Hypoxic ventilatory response of rats treated with shRNA against Ox1R (n = 12; gray bars and filled circles) compared with rats given scRNA (n = 10; open bars and circles); two-factor ANOVA; AAV effect: p = 0.049. d, Regression analysis indicates a significant association between Ox1R expression and the ventilatory response to hypoxia at FIO2 = 0.13 (O2 level at which the difference in the ventilatory response between the groups was greatest). Reduced response of rats given shRNA was due to a reduced respiratory frequency (f) response to hypoxia (e; 2-factor repeated measures ANOVA, AAV × O2: p = 0.0263; post hoc at FIO2 = 0.13: p = 0.0008) while the tidal volume (VT) response was the same between groups (f). Ox1R knockdown tended to decrease MAP (2-factor repeated measures ANOVA: p = 0.0669; g) and there was a significant, positive correlation between Ox1R expression and MAP (h).
Ox1R facilitates the hypoxia-induced activation of PVN neurons and tyrosine hydroxylase neurons in the nTS
We examined the effect of Ox1R knockdown in the PVN on the hypoxia-induced activation of PVN neurons. Compared with the control group, rats with Ox1R knockdown specifically in the PVN displayed a significantly reduced number of c-Fos-IR neurons in PVN following hypoxia (compare Fig. 10a, left panel, with Fig. 10b, left panel; t test: p = 0.0232; Fig. 10c). We also examined the hypoxia-induced activation of PVN neurons successfully transfected with either Ox1R scRNA or shRNA (i.e., those neurons that are GFP positive). The proportion of activated GFP positive neurons was significantly reduced in shRNA-treated rats compared with controls (compare Fig. 10a, right panel, with Fig. 10b, right panel; t test: p = 0.0083; Fig. 10d). These data strongly suggest that Ox1R activation facilitates the hypoxia-induced activation of PVN neurons.
Ox1R knockdown in PVN reduces the hypoxia-induced activation. Images show c-Fos immunoreactive cells (red nuclei) in the PVN of rats given scRNA- (a, left panel) or Ox1R shRNA-expressing (b, left panel) AAVs in the PVN (10× magnification). Right panels of a and b show GFP expression in the PVN, indicating successful targeting of the AAVs in each group (scale bar, 100 µm). Insets show higher magnification (20×) of boxed region showing cells expressing c-Fox (I), GFP (II), and merged images (III). c, Summary data showing number of activated (c-Fos-immunoreactive) cells in the PVN of rats given sc- (open bars and circles) or shRNA-expressing AAVs (gray bars and closed circles) in the PVN (Student's t test, p = 0.0232). d, Proportion of GFP-positive neurons activated (Fos-immunoreactive) in the PVN of rats given sc- (open bars and circles) or shRNA-expressing AAVs (gray bars and closed circles) in the PVN (Student's t test: p = 0.0083).
We examined the effect of specific Ox1R knockdown in the PVN on the hypoxia-induced activation of nTS neurons and specifically those nTS neurons that are catecholaminergic (TH-IR). Compared with the control group, Ox1R knockdown significantly reduced both the absolute number (shRNA: p = 0.0303; Fig. 11b), as well as the proportion (shRNA: p = 0.0038; Fig. 11c) of TH-IR neurons in the nTS activated by hypoxia. Together these data support a facilitating role of Ox1R in modulating the hypoxia-induced activation of TH-IR neurons in the nTS.
Ox1R knockdown in the PVN reduces the activation of tyrosine hydroxylase (TH) neurons in the nucleus of the solitary tract. a, Images show activated (Fos-immunoreactive; red nuclei) and TH-immunoreactive neurons (green) in a rat administered scRNA AAVs (top) and a rat given Ox1R shRNA AAVs (bottom) in the PVN. Scale bar, 100 µm. Insets show higher magnification of the boxed region, showing images of individual Fos (bottom) TH (middle) channels, and the merged image (top). b, Number of activated (Fos-immunoreactive) TH neurons at the three levels of the nTS (coordinates relative to bregma; CS, calamus scriptorius), in rats given scRNA- (open bars and circles; n = 6) or shRNA-expressing AAVs (gray bars and closed circles; n = 6) in the PVN (2-factor repeated measures ANOVA, AAV: p = 0.0303). c, Proportion of TH neurons activated (same rats as in b); two-factor repeated measures ANOVA, AAV: p = 0.0038).
Discussion
The neural circuitry through which orexin neurons facilitate the PCR-mediated cardiorespiratory responses to acute hypoxia remains unknown. Here we show the following: (1) orexin facilitates the hypoxia-induced activation of neurons in the VRG, nTS and PVN; (2) hypoxia activates orexin neurons projecting to PVN but not those projecting to the nTS; (3) orexin facilitates the activity of CRH, OT, and AVP neurons in the PVN; (4) orexin facilitates the hypoxia-induced activation of nTS-projecting CRH neurons; (5) both Ox1R blockade and specific Ox1R knockdown in the PVN reduce the HVR; and (6) Ox1R knockdown reduces the activity of PVN neurons and TH-IR neurons in the nTS. Together these data strongly suggest that in response to hypoxia, orexin neurons facilitate the HVR via nTS-projecting CRH neurons.
Orexin facilitates the activation of the VRC
Here we showed that the permissive influence of orexin on the HVR was also reflected in the activity of respiratory neurons in the VRC. We focused on regions in the VRC containing NKR-1 immunoreactive neurons, as this tends to label neurons in the ventral respiratory group, including rhythm-generating neurons in the preBotC (Wang et al., 2001). As expected, OxR blockade reduced c-Fos expression throughout the VRC in response to hypoxia. In the VRC and all regions examined, c-Fos expression was virtually exclusive in neurons. Thus, the reduced strength of the HVR following OxR blockade was reflected in reduced activation of respiratory neurons in the VRC. Others have shown relatively sparse innervation of the ventrolateral medulla by orexin neurons. For example, neural tract tracing by Young and colleagues identified only 1 out 227 orexin neurons projecting to the preBotC region (Young et al., 2005). In contrast, the nTS and PVN are densely innervated by orexin neurons (Peyron et al., 1998; Nambu et al., 1999), providing the rationale for examining the extent that orexin neurons projecting to these nuclei are activated by hypoxia, and whether OxR blockade reduces the activation of neurons in the nTS and PVN.
Orexin facilitates the hypoxia-induced activation of the nTS
We reasoned that the permissive effect of orexin on the HVR should also be reflected in an increased activation of the nTS. As expected, OxR blockade reduced the number of nTS neurons activated by hypoxia. We also examined the influence of OxR blockade on the number and proportion of TH-IR neurons in the nTS (i.e., the A2 noradrenergic neurons), given the important role this neuronal phenotype plays in the physiological and pathophysiological responses to hypoxia. There was a strong tendency for OxR blockade to reduce the proportion of TH-IR neurons activated by hypoxia, suggesting that orexin may contribute to the downstream effects of A2 neuronal activation, including sympathoexcitation (Bathina et al., 2013).
Orexin neurons project to the nTS (Peyron et al., 1998; Nambu et al., 1999), a nucleus that expresses Ox1R (Marcus et al., 2001). We examined whether nTS-projecting orexin neurons are activated by hypoxia. To our surprise, even in normoxia most projecting neurons were activated (i.e., express c-Fos), suggesting that there is normally considerable tonic orexinergic drive to the nTS, at least in the active phase. However, hypoxia did not increase the proportion of activated nTS-projecting orexin neurons. Thus, if orexin is permissive for the PCR through a direct action on nTS neurons, it is likely due to tonic orexin release rather than release induced by hypoxia.
Orexin facilitates the hypoxia-induced activation of PVN neurons, especially those immunoreactive for CRH
Broadly silencing the PVN (Ruyle et al., 2019) or its projections to nTS (Ruyle et al., 2023) blunts the cardiorespiratory responses to hypoxia, suggesting that PVN activation is necessary for full expression of the PCR, likely via the nTS. Orexin neurons project heavily to the PVN (Peyron et al., 1998; Nambu et al., 1999). Here we show that OxR blockade reduces the overall activation of the PVN in response to hypoxia and that hypoxia activates PVN-projecting orexin neurons in the PeF and DMH, supporting the concept that orexin neurons facilitate the HVR via the PVN. Our experiments do not provide any insight into the neural mechanism(s) responsible for the hypoxia-induced activation of PVN-projecting orexin neurons. Their activation may rely on the activation of the PCR itself, or orexin neurons may be intrinsically sensitive to hypoxia, as has been shown for neurons in the posterior hypothalamus (Dillon and Waldrop, 1993). Here we confirm previous reports (Coldren et al., 2017; Ruyle et al., 2018) showing that hypoxia activates CRH, OT, and AVP neurons. Hypoxia activates more CRH neurons than OT or AVP neurons, consistent with previous studies (Ruyle et al., 2018). We found that OxR blockade reduced the hypoxia-induced activation of all three neuronal phenotypes in the PVN. CRH, OT, and AVP neurons in the PVN all have effects on cardiorespiratory control. AVP and OT neurons project to the phrenic nucleus, nTS, and preBotC, and both neuropeptides have physiological effects on breathing (Kc et al., 2002; Mack et al., 2002). Thus, orexin may support multiple aspects of cardiorespiratory homeostasis and other neurophysiological and neuroendocrine responses to hypoxia via the global activation of the PVN and its multiple neuronal phenotypes.
Orexin facilitates the activation of nTS-projecting PVN neurons in response to hypoxia
The PVN and nTS are reciprocally connected (King et al., 2012, 2013; Ruyle et al., 2018). Previously we showed that the PVN→nTS projection is heavily activated by hypoxia (Ruyle et al., 2018) and contributes to the HVR (Ruyle et al., 2023). Moreover, it was found that most of the activated projection was composed of CRH neurons, with a small contribution from OT neurons (Ruyle et al., 2018). These findings suggested that CRH neurons are responsible for the strengthening of the PCR by the PVN. Our current data have largely replicated these previous findings, showing that CRH neurons projecting to nTS are heavily activated by hypoxia. Moreover, the activated PVN→nTS projection was composed mostly of CRH neurons, with a minor contribution provided by OT neurons. Here we provide the first evidence that orexin facilitates the hypoxia-induced activation of this projection to the nTS, likely enhancing the PCR. Given that 85% of the activated projection is composed of CRH neurons, we speculate that this phenotype contributes the most to the facilitation of the PCR via the nTS.
Considering the PVN globally, our data suggest that orexin does not facilitate the hypoxia-induced activation of CRH neurons that do not project to nTS. However, OxR blockade reduced the activation of neuroendocrine CRH neurons in the most rostral level of the PVN (−1.4). This suggests that in addition to the cardiorespiratory response to hypoxia, orexin also facilitates the hypoxia-induced activation of the hypothalamic-pituitary-adrenal axis (Coldren et al., 2017).
Ox1R activation in the PVN facilitates the PCR and activation of the nTS
Using a specific Ox1R antagonist, here we show that orexin acts at least partially through Ox1R to facilitate the HVR, an effect that likely involves the PVN. We provide evidence that orexinergic fibers innervate the PVN and lie in close proximity to CRH neurons expressing Ox1R, consistent with previous findings (Backberg et al., 2002). To further interrogate the function of Ox1R in the PVN, we partially knocked down Ox1R specifically in the PVN. This partial loss of Ox1R reduced the HVR (by ∼20%) but had no influence on resting metabolic drive, suggesting that the facilitation of the HVR by Ox1R in the PVN occurs independently from changes in metabolic CO2 production. Our regression analyses suggest that the HVR is significantly related to the amount of Ox1R expression and that ∼50% of the variability in the HVR can be attributed to differences in Ox1R expression. Taken together with our previously published data (Spinieli et al., 2022; Ben Musa et al., 2023), our most consistent finding is that OxR blockade reduces the respiratory frequency response to hypoxia, suggesting that orexin neurons activated by hypoxia, including those projecting to the PVN, impinge on neural circuits that influence respiratory rhythm generation. It is unlikely that orexin directly increases the excitability or activity of preBotC, given the sparse orexinergic innervation that targets it (Young et al., 2005).
Although it did not reach statistical significance, the blood pressure of rats following Ox1R blockade was ∼5 mmHg lower than that of control rats. Moreover, baseline MAP showed a statistically significant, positive interaction with the amount of Ox1R expressed. Previous research has shown that Ox1R in the PVN reduces the blood pressure of spontaneously hypertensive rats (Zhou et al., 2019). Although this study did not report an effect on controls, our data suggest that Ox1R activation in the PVN may support blood pressure homeostasis in normal animals, potentially by increasing tonic sympathetic drive to the heart or vasculature. While it is conceivable that the number of activated neurons in the nTS may have been influenced by the change in blood pressure related to Ox1R knockdown in the PVN, our group has previously demonstrated that the degree of nTS activation by hypoxia is not influenced by changes in blood pressure (King et al., 2012).
The reduced HVR in rats with Ox1R knockdown was associated with reduced c-Fos-IR within the PVN. We also counted cells having GFP fluorescence, indicative of successful transfection with either sc- or shRNA AAVs. There were more activated neurons expressing scRNA than there were cells expressing shRNA with reduced Ox1R expression, suggesting that reduced Ox1R expression on these neurons limited their activation in response to hypoxia. Knockdown of Ox1R in the PVN also selectively reduced in the activation of TH-IR neurons in the nTS. Previously studies have shown that corticotropin-releasing factor receptor 2 (CRFR2) is highly expressed throughout the nTS, including on fibers that lie near TH-IR neurons (Ruyle et al., 2018). In this way, increased CRH release in the nTS in response to hypoxia—a response enhanced by orexin acting in the PVN—may be an underlying mechanism that promotes the PCR.
Significance
Our previously published data suggested that orexin augments the PCR-mediated HVR. The current data presented here strongly suggest that during hypoxia, orexin neurons projecting to PVN are activated, with the released orexin binding Ox1R to augment the activity of nTS-projecting PVN neurons. We propose that this orexin→PVN–nTS neural circuit is important for the appropriate activation of the nTS during hypoxia, facilitating the PCR (Fig. 12). CRH neurons within the PVN that project to catecholaminergic neurons in the nTS may be especially important for the orexin-induced facilitation of the PCR-mediated HVR. Although our group has shown that orexin contributes to the behavioral response to hypoxia (Spinieli et al., 2022), whether this associated response also involve activation of the orexin→PVN–nTS circuit requires testing. Abnormally high orexinergic drive through the PVN–nTS circuit may underlie the elevated PCR that is associated with, and in some cases directly contributes to, cardiovascular diseases such as hypertension, myocardial infarction, and heart failure.
Proposed role of orexin neurons in the enhancement of the peripheral chemoreflex (PCR). Shown are orexin neurons (Ox) in the perifornical hypothalamus (PeF), projecting to corticotropin-releasing hormone (CRH) neurons in the paraventricular nucleus of the hypothalamus (PVN). Activity of CRH neurons and possibly oxytocin (OT) neurons are increased (+) by input from Ox neurons. Activated CRH neurons project to the nucleus of the solitary tract (nTS) to facilitate its hypoxia (Hx)-induced activation. Our data suggest that the Ox→CRH circuit specifically activated catecholaminergic (tyrosine hydroxylase; TH) neurons in nTS. Increase activation of the nTS promotes the activation of neurons in the ventral respiratory group (VRG) to increase the strength of the PCR.
Footnotes
This work was funded by National Heart Lung and Blood Institute grants R01HL098602-12 (D.D.K., E.M.H., and K.J.C.), and HL159157 (D-P.L.).
The authors declare no competing financial interests.
- Correspondence should be addressed to Kevin J. Cummings at cummingske{at}missouri.edu.