Carotid Bodies and the Integrated Cardiorespiratory Response to Hypoxia

Abstract

Advances in our understanding of brain mechanisms for the hypoxic ventilatory response, coordinated changes in blood pressure, and the long-term consequences of chronic intermittent hypoxia as in sleep apnea, such as hypertension and heart failure, are giving impetus to the search for therapies to “erase” dysfunctional memories distributed in the carotid bodies and central nervous system. We review current network models, open questions, sex differences, and implications for translational research.

Introduction

Carotid body chemoreceptors are sensors of arterial O₂, CO₂, and pH (31, 123, 143, 236). These peripheral receptors operate cooperatively with central chemoreceptors in the brain that monitor CO₂-pH or O₂ (22, 43, 78, 87, 180, 195, 211, 259). Carotid chemoreceptors contribute to the drive to breathe. Their activity is enhanced during hypoxemia, a reduction in the Po₂ in the blood that commonly occurs at high altitude, during apneas associated with sleep disordered breathing or dysautonomia (80, 107, 214), with ventilation-perfusion mismatching in the lungs, or as a consequence of other disorders (201, 255). Numerous medical conditions are associated with altered carotid body function (see discussion in Refs. 143, 154).

Hypoxia stimulates oxygen-sensing mechanisms in the carotid bodies, leading to reduced potassium currents in type I glomus cells (29, 30, 123, 141–143, 208, 212). The resulting depolarization and transmitter release evoke action potentials in the glossopharyngeal nerve, exciting neurons in the nucleus of the solitary tract (NTS). The brain stem circuits targeted by these NTS neurons generate a ventilatory response, expressed as increases in tidal volume and breathing frequency, along with coordinated changes in sympathetic and parasympathetic outflows that increase cardiac output and vascular tone to maintain or reestablish tissue Po₂ sufficient to meet metabolic demands (114, 152, 209, 236, 249, 260). Reciprocal interactions between the respiratory and cardiovascular control systems through brain networks and via sensory systems mediate this “cardiorespiratory coupling” (14, 56, 91, 165, 265).

Carotid chemoreceptor activity evoked by hypoxia is sufficient to cause the reemergence of breathing during hypocapnic apneas (69, 185) and to maintain ventilation under conditions associated with the congenital loss of central chemoreceptors (213). Repeated episodes of transient hypoxia, as during sleep apneas, can induce a distributed “memory” in circuits of the brain stem and spinal cord, termed “long-term facilitation” or LTF, expressed as increases in respiratory drive and blood pressure that persist well beyond the evoking perturbations (17, 51, 77, 113, 160, 161, 171, 172). Parallel mechanisms evoked by disruption of vagal feedback during repeated obstructive apneas, but independent of hypoxia, can also trigger potentiated hypoglossal motor neuron activity (243, 248).

Chronic intermittent hypoxia, depending on its severity, duration, and patterning, can trigger a time-dependent and escalating cascade of physiological responses (206, 235) that, although initially adaptive, can cause or exacerbate hypertension, heart failure, and other disorders (143, 207, 209, 256). There is a growing body of preclinical and clinical research targeting the carotid bodies with the goal of developing therapies for these disorders (45, 106, 153, 179, 237). Recent papers have described hypoxia-sensing transduction mechanisms in the carotid bodies (143, 208, 212, 236), the ventilatory response to hypoxia (192, 249), adaptations to chronic sustained or intermittent hypoxia (17, 72, 82, 84, 101, 143, 193, 196, 205, 220, 222), and cardiorespiratory coupling (56, 91), including pulmonary receptor-evoked cardiovascular effects triggered by air pollution (12, 103).

This review focuses on the brain stem network through which the carotid bodies enhance the drive to breathe and blood pressure, and induce LTF. Some open questions on mechanisms, sex differences, and translational research are also considered. The models and hypotheses reviewed are based on results from many laboratories and diverse experimental and computational approaches, as described in the cited literature and in other recent complementary reviews (17, 56, 80, 81, 91, 131, 162, 164, 216, 265).

Brain Stem Circuits for the Integrated Cardiorespiratory Response to Transient Hypoxia

Neurons in the NTS and adjacent medial medulla have heterogeneous phenotypes and diverse responses to carotid body stimulation (2, 4, 36, 116, 120, 157, 194, 265). These “chemoresponsive” neurons operate through multiple circuit pathways to regulate the depth and frequency of breathing and, concurrently, cardiac output and vascular tone (FIGURE 1) (91, 177).

FIGURE 1.

Continued

Inspiratory Drive and the Hypoxic Ventilatory Response

A rostral cluster of pre-inspiratory “I-Driver” neurons in the pre-Bötzinger complex (128, 131, 162, 229, 231, 240) of the ventral respiratory column (VRC) excites a downstream “chain” of premotor and motor neurons for the diaphragm, other inspiratory pump muscles, and muscles that influence airway resistance, such as those of the tongue and oropharynx (FIGURE 1A) (19, 67, 73, 83, 246). Carotid chemoreceptor stimulation can shorten the inspiratory burst duration of some I-Driver neurons, which otherwise exhibit little or no change in their firing rate, whereas their targets, including bulbospinal premotor neurons, exhibit larger rate increases, enhancing inspiratory drive. These observations led to a model with parallel circuit mechanisms for tuning tidal volume and breathing frequency (173, 177).

Chemoresponsive NTS neurons operate on a subset of rostral I-Driver neurons and at multiple sites along the inspiratory neuron chain to increase neuronal firing rates (11, 171, 173). Evidence for excitatory synaptic interactions between some responsive NTS and VRC neurons supports a model for inspiratory drive amplification that includes positive feedback circuits constrained by synaptic depression in the recurrent arm of the loop (FIGURE 1B; see Fig. 11B in Ref. 177). This model is consistent with the observation that some chemoresponsive NTS neurons have an evoked or enhanced inspiratory-modulated firing pattern during carotid body stimulation (194).

A second stage of amplification mediated by a disinhibitory microcircuit downstream from the pre-Bötzinger complex has also been proposed (233). An efference copy of inspiratory drive excites inhibitory inspiratory neurons, which in turn inhibit pericolumnar tonic expiratory (t-E) neurons. This sequence results in disinhibition of their excitatory inspiratory neuron targets (FIGURE 1C). These t-E neurons operate as network “hubs” for integration of convergent baroreceptor, peripheral and central chemoreceptor, and motor efference copy inputs that cooperatively tune inspiratory drive and, as considered subsequently, autonomic influences on cardiovascular function (135, 177, 190, 233). During exercise, carotid chemoreceptors contribute to tonic vasoconstriction and locomotor blood flow; their role in the enhancement of sympathetic activity and their interactions with other afferent and feed-forward central command mechanisms remain incompletely understood (46). The convergence of chemoreceptor and efference copy motor influences at VRC t-E hub neurons during rhythmic behaviors like breathing and coughing suggests that these hubs may have a role in the generation of hyperpnea and other physiological responses to exercise (186, 234).

Via NTS neurons, carotid chemoreceptors also cooperate with central CO₂-pH chemoreceptors in the retrotrapezoid nucleus-parafacial region (FIGURE 1D) (22, 89, 91, 94, 253, 259). Recent evidence suggests that chemoresponsive neurons in this area influence inspiratory drive via multi-path modulation of t-E neuron nodes, supporting a complex, partly hierarchical architecture for cooperative interactions between peripheral and central chemoreceptor drives and a quasi-periodic tuning of VRC circuits by coordinated clusters of adjacent tegmental field neurons (177, 189, 190, 233).

FIGURE 1.

Medullary circuit mechanisms in contemporary models of carotid chemoreceptor and baroreceptor modulation of breathing and cardiovascular functions

A–K: schematic representations of specific functional interactions described in the text. Neuron populations are grouped by brain stem location and labeled by their respiratory modulated firing rates: inspiratory (I), post-inspiratory (post-I), or expiratory (E), according to the phase of greatest average activity, and as decrementing (Dec) or augmenting (Aug) if the average peak firing rate occurs during the first or second half of the phase, respectively. I-Driver neuron activity begins slightly before phrenic nerve discharge, with firing rates that peak in the early I phase and then slowly decrease before abruptly decreasing at the end of inspiration. Tonic E cells (t-E) are active throughout the respiratory cycle, with their greatest rate during expiration. Black lines and “synapses” indicate inferred functional connectivity discussed in the text. 1, Cross-correlogram with offset peak consistent with RTN-pF region neuron exciting a putative premotor caudal VRC augmenting expiratory neuron. Figure adapted from Fig. 5D in Ref. 190 with permission from Journal of Neurophysiology. Both neurons responded to transient hypoxia with increased firing rates (177). Binwidth = 15.5 ms, 71,132 trigger neuron spikes, 25,541 target neuron spikes. Böt, Bötzinger complex; BS, bulbospinal; FTL/RTN-pF, lateral tegmental field/retrotrapezoid n.-parafacial n; IML, intermediolateral column of the spinal cord; IVLM, intermediate ventrolateral medulla; Lum, lumbar nerve; NA, n. ambiguus; NTS-DMM, n. tractus solitarius-dorsal medial medulla; Phr, phrenic nerve; pre-Böt, pre-Bötzinger complex; RVLM, rostral ventrolateral medulla; VRC, ventral respiratory column.

Tuning Breathing Rate

Chemoresponsive NTS neurons evoke changes in breathing frequency through mechanisms that reduce or limit the durations of both inspiratory and expiratory phases (Ti and Te, respectively). Truncation of I-Driver activity and the inspiratory phase is mediated in part by enhanced post-inspiratory (decrementing expiratory) neuron inhibition caused by carotid body stimulation (FIGURE 1E) (125, 128, 138, 139, 173, 177, 216, 226). Teasing apart the physiological contexts and respective contributions of inhibition and intrinsic mechanisms for the I-Driver “off switch” remains an active and controversial area of research, in part because of the different preparations and model systems studied and the complex consequences of pharmacological manipulations used (7, 10, 37, 150, 216, 225).

A shortened expiratory phase, although not obligatory, when present, may reflect enhanced rostral Bötzinger complex augmenting expiratory neuron inhibition of the relevant inhibitory decrementing expiratory (E-Dec) cells (FIGURE 1F) (129, 139, 140), overcoming chemoreceptor-driven enhancement of E-Dec neuron activity (11, 96, 170, 173, 187, 238). Regulation of the expiratory phase of the respiratory cycle remains an area of active research, and additional network mechanisms proposed to tune breathing frequency, including circuit loops through the pons, are considered subsequently.

Enhancement of Expiratory Drive

Mechanisms for the enhanced expiratory effort associated with hypoxia are less well understood. Active expiration, including excitation of abdominal and intercostal expiratory muscles, is present during resting breathing under some conditions (1, 35, 42, 105, 185, 219) and dramatically enhanced during hypoxia (140). It remains to be determined whether this increased expiratory drive reflects, in part, NTS chemoresponsive neurons acting directly on caudal bulbospinal expiratory premotor neurons. However, there is evidence for the recruitment of both Bötzinger and adjacent RTN-parafacial neurons during chemoreceptor stimulation (75, 108, 109, 191). The parafacial-lateral tegmental field region, a site with complex neuronal interactions (177, 189, 190) and where peripheral and central chemoreceptor influences converge, as noted above, has been strongly implicated as a source of expiratory drive (20, 94, 131, 164). Hypercapnia leads to disinhibition of non-chemosensitive late-E neurons proposed to excite downstream expiratory neurons (41). This observation and other recent results collectively suggest parallel push-pull excitatory and disinhibitory control circuits for expiratory drive tuning, including tightly coordinated control of premotor inspiratory and expiratory drives by shared RTN-pF region neurons (FIGURE 1G) (104, 162, 177, 189, 190, 233).

Central Oxygen Sensing and Gasping

Selective stimulation of carotid chemoreceptors may evoke somewhat different motor patterns and autonomic responses than transient systemic hypoxia, which stimulates oxygen sensors within the carotid bodies and in the brain, all of which can potentially contribute to the hypoxic ventilatory response via parallel actions on pre-Bötzinger complex I-Driver neurons (FIGURE 1H) and other elements of the brain stem respiratory network (87, 236).

Under conditions of severe hypoxia, breathing is first enhanced and then depressed until apnea or the cessation of breathing (217, 218), followed by the emergence of a simplified reconfigured respiratory network that generates an autoresuscitative gasping motor pattern. The network states and biophysical processes required for this behavior remain incompletely understood, as does the extent to which efferent copies of this drive engage autonomic circuits (57, 64, 82, 122, 197, 226, 245, 251). Gasping is common in patients in cardiac arrest with ventricular fibrillation and is associated with successful resuscitation (61).

Efference Copy and Hypoxia-Evoked Changes in Autonomic Activity

The brain stem respiratory network has a profound influence on the patterns (121) and magnitude (8) of autonomic nervous system signaling that modulates cardiac output and blood pressure in response to hypoxia (FIGURE 1I). Carotid chemoreceptors evoke coordinated changes through other routes, including direct NTS neuron projections to pre-sympathetic C1 and non-C1 neurons in the rostral-ventrolateral medulla (RVLM) (116, 120). These RVLM neurons excite spinal pre-ganglionic sympathetic neurons that drive vasoconstriction and stimulate the heart in response to hypoxia (93, 156, 188, 265). The RVLM C1 and non-C1 neurons are phenotypically heterogeneous, with attributes including three distinct classes of respiratory modulated discharge patterns: inspiratory-excited [presumably excited by pre-Bötzinger complex I-Driver neurons or downstream inspiratory chain follower neurons (166)], inspiratory-inhibited, and those with a post-inspiratory modulation (166, 168). C1 neurons are involved in inspiratory modulation of sympathetic activity enhanced by carotid chemoreceptors (167). The post-inspiratory subset includes neurons that also exhibit late expiratory neuron excitation following chronic intermittent hypoxia, an example of plasticity considered further in a subsequent section of this review (168). RVLM neurons are modulated by baroreceptors indirectly through inhibitory neurons (FIGURE 1I) in the intermediate ventrolateral medulla (IVLM, “alias” caudal ventrolateral medulla) (93), which also receive post-inspiratory excitation during hypoxia (168).

Parasympathetic cardiac vagal pre-ganglionic neurons in the nucleus ambiguus (NA) receive convergent carotid chemoreceptor and baroreceptor influences via the NTS; they are inhibited during inspiration, presumably by neurons in the VRC (FIGURE 1J). The resulting respiratory sinus arrhythmia (RSA) caused by reduced vagal inhibition of the heart can contribute to increased cardiac output (85, 88). Moderate hypoxia diminishes the magnitude of the RSA (28, 60, 254, 262, 263). The extent to which changes in the breathing pattern and autonomic modulation of the heart and circulation influence venous return and cardiac output depends on various factors (147). Left ventricular stroke volume decreases during inspiration, although right ventricular filling pressure increases (221). Moreover, venous flow from the legs during inspiration varies, depending on the relative contributions of the diaphragm and intercostal and abdominal muscle activity (258). Respiratory sinus arrhythmia does not appear to improve gas exchange efficiency, but rather minimizes the work done by the heart while maintaining physiological levels of arterial carbon dioxide or Paco₂ (23, 24). An enhancement of firing rate during the post-inspiratory interval also has been observed in vivo, supporting the hypothesis that slow breathing, which increases the post-inspiratory phase duration, slows the heart rate by increasing the firing rate of cardiac vagal pre-ganglionic neurons (65, 88).

Cardiorespiratory coupling is also apparent in the arterial pulse modulation of respiratory premotor and motor neurons (52, 74). Baroreceptors provide a major coordinating influence on cardiorespiratory coupling during hypoxia through their actions on breathing and via coordinated feedback regulation of cardiovascular function. Functional connectivity of medullary raphe circuits supports baroreceptor modulation of inspiratory drive and expiratory phase duration via push-pull tuning of VRC t-E and E-Dec neurons, respectively (9, 135). Raphe circuits are dynamically reconfigured over the course of the respiratory cycle and baroreceptor and carotid chemoreceptor stimulation (6, 32, 127, 172). These circuits (FIGURE 1K) have been proposed to incorporate internal equilibrium-seeking mechanisms that help to stabilize breathing and regulate reflex gain during perturbations of blood pressure, for example, during cough (130, 134, 135, 202), and to contribute to the respiratory modulation of the sympathetic baroreceptor reflex (9).

The Role of the Pons in the Hypoxic Ventilatory Response and Cardiorespiratory Coupling

The Kölliker-Fuse/parabrachial complex of the dorsolateral pons receives inputs from NTS neurons that relay signals from carotid body afferents (FIGURE 2A) (98, 244). Neurons of the Kölliker-Fuse (K-F) region contribute to cardiorespiratory coupling and are activated during hypoxia and carotid sinus nerve stimulation (63). They project to several brain stem sites that participate in the control of tidal volume and breathing frequency and directly to brain stem sympathetic circuits that regulate the cardiovascular network (13, 15, 16, 53, 59). Pharmacological perturbations of K-F neurons disrupt the hypoxic ventilatory response (38), possibly, in part, through reciprocal connections with the RTN and parafacial region (59, 223, 239).

FIGURE 2.

Simplified functional connectivity of hypoxia-responsive pontine neurons

A–E: interactions inferred from various experimental approaches reviewed in the text that contribute to the hypoxic ventilatory response and changes in cardiorespiratory coupling during hypoxia in current models. F and G: MWRO-triggered firing rate histograms of a pontine neuron before and after bilateral vagotomy and phase resetting of MWROs by withholding lung inflation in vagus-intact cat. Figure derived from Fig. 5A in Ref. 176 with permission. See text for details. KF, Kölliker-Fuse n.; LPBM, lateral parabrachial n.; MWRO, Mayer wave-related oscillations; NPBM, medial parabrachial n.; Ti (Te), duration of inspiratory (expiratory) phase.

The dorsolateral pons influences breathing frequency during hypoxia through inspiratory-expiratory phase switching mechanisms. These circuits remain incompletely understood; models (5, 17, 178, 203) include pontine neurons that operate to shorten inspiration and expiration (e.g., FIGURE 2B). Inhibition of the K-F region with muscimol, a GABA-A receptor agonist, reduces the enhancement of phrenic and sympathetic drive (FIGURE 2C) and breathing frequency responses to hypoxia (39). The Kölliker-Fuse nucleus also provides a pathway for post-inspiratory modulation of cardiac vagal parasympathetic activity (FIGURE 2D) (65). K-F connectivity to post-inspiratory or E-Dec neurons could also contribute to modulation of augmenting or late expiratory neuron activity and the enhancement of active expiration with increased chemical drive (18).

Neurons of the lateral parabrachial nucleus (LPBN) may also enhance breathing frequency during hypoxia by reducing expiratory phase duration (242). Bilateral lesions of K-F nuclei provoke apnea, suggesting that neurons there promote inspiration (204). This conclusion is supported by the finding that both pharmacological and patterned electrical microstimulation within the pontine medial parabrachial nucleus promote inspiratory burst onset (FIGURE 2E) and can evoke a premature onset of the next inspiratory phase, thereby shortening Te and increasing breathing frequency (266).

The A5 cell group of noradrenergic and glutamatergic neurons in the ventrolateral pons modulates breathing frequency and cardiovascular function during and following hypoxia (118, 119). Neurons in the A5 group project to the VRC, have post-inspiratory and expiratory modulated firing patterns, and are responsive to hypoxia (50, 59, 92). Micro-injections of muscimol into the ventrolateral pons prolongs inspiration (110), whereas application of glutamate prolongs expiration (111). This expiratory-facilitating function is apparent during and following episodes of transient severe hypoxia: respiratory frequency first increases, then slows in a “post-hypoxic frequency decline,” followed by a gradual return to control levels. A similar sequence is observed during and following carotid sinus nerve stimulation under hyperoxic conditions, suggesting that the frequency decline is of central origin (97).

Raphe-Pontine Circuits, Slow Breathing, and Mayer Wave-Related Oscillations

Slow (~0.1 Hz) oscillations in respiratory drive and blood pressure, termed Mayer waves, can be triggered by hypoxia or hemorrhage, conditions that increase carotid body chemoreceptor activity and sympathetic drive (3, 33, 68, 124, 262). Oscillations evoked by carotid artery occlusion with hemorrhage can be eliminated by section of the carotid sinus nerves or destruction of carotid body chemoreceptors, suggesting that hypoxia or under-perfusion of the carotid body can evoke this category of cardiorespiratory coupling (3).

Bilateral vagotomy commonly results in a prolongation of the inspiratory phase. The loss of the Breuer-Hering reflex (96) together with the aforementioned inspiratory off-switch function of the pons (55) contributes to this slowing of breathing frequency. A second mechanism involving cardiorespiratory coupling has been proposed to play a role in the generation of the slower breathing rhythm. The VRC is embedded in a highly interconnected pontomedullary network with medullary raphe circuits providing parallel signaling routes between the pons and the VRC (184, 230, 232). Pontine and raphe neurons with Mayer wave-related oscillations (MWROs) can become synchronized 1:1 with the breathing rhythm after vagotomy (FIGURE 2F) (176). Computational models show that neurons with inspiratory efference copy input and feed-forward inhibition produce similar rate profiles (55). Prior to vagotomy, withholding lung inflation during inspiration both prolongs the inspiratory phase and resets the MWROs (FIGURE 2G) (176). Increased cardiorespiratory coupling can persist following slow deep breathing in human subjects (54). Functional connectivity inferred from multi-array recordings suggests that pontine-raphe circuits contribute to this aspect of cardiorespiratory coupling (176).

Distributed Memories Induced by Intermittent Hypoxia

In 1980, David Millhorn and colleagues reported that episodic stimulation of carotid body chemoreceptors can induce a respiratory memory expressed as an increase in phrenic nerve inspiratory burst amplitude and breathing frequency. This “long-term facilitation” (LTF) persists well beyond the period of stimulation (160, 161) and can also be evoked by direct electrical stimulation of n. raphe obscurus but not by central CO₂ chemoreceptors; it is prevented or attenuated by serotonin (5-HT) antagonists (76, 159, 160). Hypoxia and electrical stimulation of the carotid sinus nerve also induces Fos-like immunoreactivity in serotoninergic and catecholaminergic neurons at multiple sites within the brain stem (63). Raphe stimulation releases 5-HT in the cervical ventral horn of the spinal cord, the site of phrenic motor neurons (27, 215). Subsequent research has identified stimulus parameters and conditions necessary for enhancement of specific motor and autonomic outflows by intermittent hypoxia (76, 117, 169, 257), along with insights into the network, and cellular and molecular mechanisms for a memory system distributed among multiple sites in the brain stem (26, 77, 126, 137, 174, 175), spinal cord (34, 70, 247), and carotid bodies (236). These multi-scale mechanisms for adaptive physiological responses alter cardiorespiratory coordination and can exacerbate or result in various pathophysiological conditions (143, 207, 209, 256, 261).

Cardiorespiratory-related neurons of the medullary raphe nuclei have dynamic functional associations and interact with the VRC, pons, RTN, and other brain stem sites for the coordinated regulation of breathing and the cardiovascular system (6, 32, 91, 132–134, 136, 172, 173, 177, 184, 210). Some raphe neurons respond to repeated stimulation of the carotid chemoreceptors with step-like increases in firing rate (FIGURE 3A). A “ratchet-like” circuit mechanism has been proposed to contribute to this incremental induction of respiratory LTF: periods of evoked high firing rate in some neurons are limited by the delayed inhibitory actions of other neurons (FIGURE 3B), until the firing rates of the early responders increase and “saturate” (149, 171, 172). Some raphe neurons excited by peripheral carotid chemoreceptor stimulation are functionally inhibited by central chemoreceptor activation (FIGURE 3C) (183), a result consistent with the inability of central chemoreceptor stimulation to evoke LTF (159). Indeed, repeated episodes of hypercapnia have been shown to induce a long-term depression of phrenic nerve activity (115) through processes involving activation of 5-HT_1A and α2-adrenergic receptors (227).

FIGURE 3.

Distributed network sites for induction and expression of carotid body- and intermittent hypoxia-induced long-term facilitation

A: responses of raphe neurons and integrated phrenic nerve activity during repeated selective stimulation of carotid chemoreceptors and induction of long-term facilitation (LTF). Figure is derived from Ref. 174 with permission from Respiration Physiology. B: raphe circuit mechanism proposed to contribute to ratchet-like incremental induction of LTF (171, 172). C: firing rate histogram and by-cycle rate plots show opposite responses of a raphe neuron to selective stimulation of central and peripheral carotid chemoreceptors. Figure derived from Fig. 2A, C in Ref. 183 with permission from The Royal Society. D: increased number of black bins in the top right quadrant along the diagonal of the joint peri-stimulus time histogram (JPSTH) documents a significant increase in spike synchrony and effective connectivity for two raphe cells following induction of LTF. Figure adapted from Fig. 4A in Ref. 171 with permission from Journal of Physiology. E: offset cross-correlogram peak shows effective connectivity between a pre-Bötzinger I-Driver neuron and a VRG inspiratory target cell; binwidth = 0.5 ms, 8,216 trigger neuron spikes, 4,569 target neuron spikes. F: non-uniform distribution of statistically significant black bins along the diagonal of the JPSTH for the same neurons shown in E documents enhanced effective connectivity between this pair of neurons following induction of LTF. Figure derived from Fig. 3, B and C, in Ref. 171 with permission from Journal of Physiology. G–L: additional sites for distributed LTF expression following repeated episodes of intermittent hypoxia. See text for details.

Intermittent or sustained hypercapnia can have a profound effect on the responses to intermittent hypoxia (155). Although intermittent hypercapnia may depress ventilatory LTF, sustained hypercapnia facilitates it (90, 95). Sustained hypocapnia prevents the occurrence of LTF, hence LTF can be difficult to observe during sleep-disordered breathing. Such breathing is made up of cycles of hyperventilation followed by apneic hypoxic hypercapnia followed by arousal, hyperventilation, etc. (155).

In rats, both intermittent and sustained hypoxia increase sympathetic nerve activity after 2 wk, but they affect sympatho-respiratory coupling differentially (56). Intermittent hypoxia enhances sympatho-respiratory coupling and decreases ventilatory variability. Constant hypobaric hypoxia replaces normal 1-to-1 coupling between bursts of sympathetic and phrenic nerve activity with 2-to-3 coupling with increased ventilatory variability. Even a single session of hypoxic exposure is capable of increasing tonic sympathetic nerve output (sympathetic long-term facilitation) and altering chemo- and baroreflexes (260).

Several lines of evidence collectively support an important role for distributed brain stem circuits in the induction and expression of LTF, including carotid chemoreceptor-evoked responses, changes in functional connectivity among raphe neurons (FIGURE 3D), and evidence of increased effective connectivity among I-Driver neurons and their premotor targets (FIGURE 3, E–G) (17, 62, 79, 127, 171, 172, 257). How chronic intermittent hypoxia impacts different states of inspiratory activity in the pre-Bötzinger complex remains an active area of research (81). Cell signaling pathways for LTF of inspiratory drive have been most thoroughly investigated in the spinal cord (FIGURE 3H), where serotonin-dependent induction of facilitation in phrenic motor neurons depends on competing cell signaling mechanisms that interact conditionally with adenosine-dependent pathways for motor facilitation (47–49, 71, 77, 200, 252).

FIGURE 3.

Continued

Serotonergic mechanisms are also implicated in the induction of expiratory LTF (126) at sites in the RTN-pF respiratory group (FIGURE 3I) (91). This enhanced expiratory drive, together with sensitization to central CO₂ chemoreception with chronic intermittent hypoxia, contribute to increased sympathetic nerve activity (51, 163, 264). A subset of non-C1 RVLM presympathetic neurons (FIGURE 3J) has been implicated as a source of this enhancement (168). Recruitment of 5-HT systems by hypoxia and hypercapnia also modulates the excitability of parasympathetic cardiac vagal neurons (FIGURE 3K) (113). An altered balance of hypoxia-inducible factor-α isoforms and redox state were identified in the NTS, RVLM, and adrenal medulla of rats exposed to chronic intermittent hypoxia. These changes were prevented by carotid body ablation, suggesting that carotid-evoked changes in neural activity lead to these alterations and the associated development of hypertension and elevated sympathetic activity (199).

Chronic intermittent hypoxia enhances the sensitivity and gain of carotid body chemoreceptors in an age- and history-dependent manner (FIGURE 3L). Hypoxic sensitization requires ~10 times the number of episodes of intermittent hypoxia in adult compared with neonatal rats. Moreover, subsequent acute intermittent hypoxia evokes LTF in the carotid body of previously exposed adults but not of neonates (196, 198). Hypoxia-induced changes in the redox state of the carotid bodies via an altered balance of HIF-1α-dependent pro-oxidant and HIF-2α-dependent anti-oxidant activities may play a role in the induction of these changes (235, 236); a recent study reports that HIF-2α is essential for carotid body function and development (146). Acute hypoxia with concurrent hypercapnia also evokes sensory LTF in carotid bodies (224).

The hypoxemic carotid body expands far beyond its normal size (142). Adult neural stem cells contribute to carotid body growth during hypoxia acclimatization, and these new cells contain voltage-dependent ion channels (193). This adaptation to chronic hypoxia requires HIF-2α for oxygen-sensitive glomus cells to develop and survive within the carotid body (146) and for normal responses to hypoxia (101). Contrary to maladaptive responses outlined previously, other studies have suggested therapeutic benefits of hypoxic or hypercapnic regimes in spinal cord injury (34, 181), obstructive sleep apnea (153), and other cardiovascular, respiratory, cognitive, and metabolic consequences of intermittent hypoxia (154).

Future Directions for Basic and Translational Science

Sex Differences in Responses to Hypoxia

Further research is needed to better understand sex differences in the hypoxic ventilatory response and their impact during development. Prenatal nicotine exposure alters the postnatal heart rate response to hypoxia and respiratory sinus arrhythmias in young male but not female rats (25). The hypoxic ventilatory response is greater in men than in women (86), and is differentially depressed more in women by morphine (228). Age- and sex-dependent differences in serotonin-dependent hypoxia-evoked neural plasticity have also been identified: LTF increases with age in female rats but declines in male rats (21).

The “Erasure” of Dysfunctional Cardiorespiratory Network Memories

An emerging goal of contemporary translational research is to identify safe and effective therapeutic strategies for management and treatment of the long-term consequences of adaptations to chronic intermittent hypoxia (106, 143, 153, 236). Pharmacological and immunological approaches and gene therapy are being actively considered (145, 148, 182, 236, 243). Ablation or denervation of the carotid bodies for resistant hypertension is also being evaluated in preclinical research (106). A prospective human feasibility and safety trial on unilateral ablation has been reported (179), and similar approaches for heart failure are being explored (44, 151).

The ablation approach is controversial. Carotid body ablation was used in humans as an experimental treatment for bronchial asthma and chronic obstructive pulmonary disease but did not produce significant benefits for patients (112). Patients with previous bilateral carotid body ablations to treat asthma were subsequently evaluated for responses to hypoxia and hypercapnia; whereas ventilation in room air was normal, hyperpnea in response to hypoxia was absent, and the response to hypercapnia was reduced (144). These observations suggest that central hypoxia-sensing mechanisms cannot fully compensate for the loss of carotid O₂ chemoreception (78, 87), and it remains an open question as to whether central mechanisms can “appropriate” network mechanisms used by carotid chemoreceptors to reestablish adaptive (or maladaptive) set points and system gains (100, 158, 241). Several recent reviews and perspectives (112, 143) and preclinical studies (58, 66, 102, 237) have noted potential risks and adverse consequences of carotid body removal, including baroreceptor failure, exacerbation of co-morbidities, altered blood glucose regulation, reduced ability for acclimatization to high altitude, and irregular breathing rhythms.

Significant gaps remain in our understanding of brain networks and cell signaling mechanisms for cardiorespiratory regulation and hypoxia sensing. The development, organization, plasticity, and state-dependent functionality of these systems and the integration of peripheral and central chemoreceptor influences are areas of active investigation (259). These lines of inquiry also have a broader relevance beyond cardiorespiratory integration with the growing appreciation that other brain regions, rhythms, and functions are influenced by breathing (40, 99, 233, 250).