Neuronal mechanisms underlying opioid-induced respiratory depression: our current understanding

Abstract

Opioid-induced respiratory depression (OIRD) represents the primary cause of death associated with therapeutic and recreational opioid use. Within the United States, the rate of death from opioid abuse since the early 1990s has grown disproportionally, prompting the classification as a nationwide “epidemic.” Since this time, we have begun to unravel many fundamental cellular and systems-level mechanisms associated with opioid-related death. However, factors such as individual vulnerability, neuromodulatory compensation, and redundancy of opioid effects across central and peripheral nervous systems have created a barrier to a concise, integrative view of OIRD. Within this review, we bring together multiple perspectives in the field of OIRD to create an overarching viewpoint of what we know, and where we view this essential topic of research going forward into the future.

INTRODUCTION

In the late 1990s, a large increase in prescriptions for opioid-based medications spurred by misunderstanding and a lack of transparency between pharmaceutical companies and the medical community led to a rapid expansion of opioid abuse. From 2011 to 2018, the rate of overdoses involving prescription opioids began to plateau (https://www.drugabuse.gov/related-topics/trends-statistics/overdose-death-rates); however, the rate of heroin overdose more than quadrupled, whereas the overdose rate from synthetic opioids (other than methadone), such as fentanyl, increased ∼×9 during the same time frame (1–3). In 2016, two-thirds of all drug overdoses involved opioids (4), and in 2017, the “opioid epidemic” was deemed a public health emergency in the United States. The opioid epidemic continues to claim the lives of an estimated 130 people per day in the United States (https://www.hhs.gov/opioids/about-the-epidemic/index.html), and new, relatively easy to synthesize, opioid drugs are reaching the streets with ever increasing potencies.

The primary cause of death associated with opioid-based analgesics and drugs of abuse is opioid-induced respiratory depression (OIRD). Although on average the risk of mortality associated with opioid use increases in a dose-dependent manner, no amount of opioid is without risk (5–7). Within and among individuals, variability in the response to opioids makes these drugs particularly dangerous and is a major contributor to overdose. One factor underlying variability in OIRD is that the respiratory response to opioids is state dependent. Opioid sensitivity is heightened during different neurological and physiological states including concurrent drug administration, such as anesthesia (8–10) and benzodiazepines (11), as well as sleep disordered breathing (SDB, Ref. 12). Indeed, SDB is prevalent among opioid users (13–15), and this clinical population is particularly vulnerable to OIRD since hypercapnic (16, 17) and hypoxic (18, 19) conditions present during SDB exaggerate opioid responses. Although numerous studies using animal models including mice, rats, dogs, cats, rabbits, and goats (5, 20–28) suggest that variability is a hallmark of OIRD, this variability is often overlooked, poorly understood, and has proven to be a significant barrier to forming a consensus on the specific causes and prevention of OIRD.

This review brings together multiple research teams to discuss the current state of knowledge regarding key brain regions and mechanisms responsible for OIRD. Breathing is a highly complex and integrated behavior (29–32), and there is growing evidence that its underlying rhythmogenic and modulatory networks are highly dynamic and plastic (30, 33–36). As a result, unraveling the underlying cause of OIRD is not a simple task. Although there is general consensus that the breathing rhythm itself is generated within the brainstem (32), numerous regions and modulatory systems distributed throughout the central nervous system modulate breathing (37), and breathing in turn affects many areas distributed throughout the brain (31, 38). Breathing is also highly regulated by chemo- and mechano-feedback mechanisms that are critical for maintaining homeostasis. The respiratory response to opioids is similarly complex since OIRD ultimately emerges from the combined effects of opioids on mechanisms of respiratory rhythm generation, modulation, and sensory feedback (Fig. 1) at sites distributed throughout the nervous system. Here, we highlight key respiratory centers in the brainstem (Fig. 1), as well as cortical regions (Fig. 6) and components of the peripheral nervous system (PNS), that are strongly implicated in the failure or impairment of breathing due to OIRD. In addition, we emphasize that the variability and state-dependence of OIRD likely reflects the dynamic nature of the mechanisms that generate and regulate breathing (30). By embracing this variability, we hope to better understand how opioids affect these dynamic mechanisms to more effectively identify risk factors for, and develop treatments and countermeasures against, OIRD.

Figure 1.

A: normal breathing requires the integration of rhythmogenic, modulatory, and sensory feedback mechanisms. Opioid overdose can suppress all of these important mechanisms of respiratory control, leading to OIRD. B: human brainstem schematic illustrating key sites in the pons and medulla that mediate OIRD. Ventral view is shown on the left, and transverse sections are shown on the right. Colors in B correspond to mechanisms depicted in A. NTS, nucleus tractus solitarius; OIRD, opioid-induced respiratory depression; preBOTc, preBӧtzinger complex.

MEDULLARY NETWORKS

It is widely accepted that the respiratory rhythm emerges from a highly interconnected brainstem network located within the medulla (39–52). Neurons within this wider respiratory network are synchronized to generate the specific phases and patterns of breathing (42, 45, 46, 49, 50, 53–55). For example, synchrony critically contributes to network stability, rhythmogenesis, and transmission of motor drive to inspiratory muscles (56–58). One manifestation of synchrony is high frequency oscillations (HFOs) (50–100 Hz) (59) (i.e., periodicities) present in neurograms and neuron firing rates (60). Morphine administration depressed HFOs to lower frequencies (61), consistent with disruption of inspiratory network synchrony. However, HFOs can be nonspecific metrics of overall circuit behavior. Lalley and Mifflin (62) showed similar effects of the opioid fentanyl on autocorrelations of respiratory neurons (61), which can show oscillations due to HFOs These studies support the concept that synchrony in the brainstem inspiratory circuit is sensitive to opioids. Interestingly, the magnitude of coherence between right and left phrenic nerves was not changed by morphine (61), suggesting that crossed pathways at the brainstem and spinal level were not altered by this drug.

Opioids can change discharge identity (respiratory phenotype) of neurons that participate in the respiratory network. Existing evidence is consistent with significant alterations in discharge patterns of breathing- and nonbreathing-modulated (NBM) neurons in the ventral respiratory column (VRC). This is illustrated in the respiratory cycle-triggered histograms before and after intra-arterial administration of the µ-opioid receptor agonist, codeine, as shown in Fig. 2. Seven examples of NBM, expiratory-tonic (E-T), and inspiratory-tonic (I-T) neurons are shown. In the first three examples, NBM neurons that were not modulated during breathing became respiratory-modulated (Fig. 2, 1–3). In two cases, inspiratory or expiratory neurons changed their main phase of activation during breathing (Fig. 2, 4 and 5). In most cases, the baseline discharge rate of these neurons increased or decreased. These changes are consistent with altered discharge identity of these neurons. Discharge identity is the activity pattern of a given neuron relative to the breathing cycle (63).

Figure 2.

Examples of neurons with altered discharge identity or discharge rate after codeine administration from several animals. Several of the neurons shown became breathing modulated (examples 1–3). In others, the time of maximum discharge rate during the respiratory cycle changed (4 and 5). Discharge rates of most of the neurons shown increased (1, 2, and 7) or decreased (4 and 5) after codeine. Cycle-triggered histograms of neuronal activity (colors) overlying phrenic discharge (gray). E-P, expiratory phasic; E-T, expiratory tonic; I-T, inspiratory tonic; n, neuron; NBM, nonbreathing modulated. Anesthetized, paralyzed, and artificially ventilated cats. Codeine (1–3,000 µg/kg) was administered via the vertebral artery.

Alterations in discharge identity in the respiratory network may affect respiratory rhythmogenesis and pattern formation, since these alterations can result in differing temporal delivery of postsynaptic potentials to their targets. The timing of arrival of synaptic volleys has an important role in excitability and phase switching in the respiratory network (64). Even subtle changes to these volleys from multiple presynaptic terminals would alter the spatial and temporal summation of PSPs and thus could greatly alter the dynamic membrane potential of the target neuron. Some neurons in the breathing network can change discharge identity during non-breathing behaviors, such as cough, and this collective process has been termed reconfiguration (39, 52, 65, 66). Reconfiguration is hypothesized to contribute to large changes in motor patterns of respiratory muscles that are necessary for execution of ballistic-like behaviors (65–67). It is conceivable that opioids, in addition to altering the firing rates of respiratory neurons, induce far reaching changes in the organization of the respiratory network through reconfiguration. This proposed opioid-induced reconfiguration represents a higher scale effect of these drugs than at individual cells and integrates alterations in the excitability of individual members of the entire network into an aggregate response that determines behavior. Given that the network is a higher-level scale than that of individual cells, it can be hypothesized that unique opioid-sensitive mechanisms exist that are not easily detectable at the cellular scale (67). As noted above, there is evidence that network synchrony, a mechanism that is a feature of network behavior, is sensitive to opioids (61). Unraveling the mechanisms underlying synchrony is an evolving research field and further investigation will reveal additional manifestations of the effects of opioids at the network scale. Discovery of these proposed additional mechanisms will lead to a more comprehensive, multiscale hypothesis for the mechanisms of opioid-induced respiratory depression.

Rhythmogenetic Networks within the Medulla

Although neurons within the medullary network are highly connected and interacting, these neurons are not evenly distributed, and anatomically distinct, yet interconnected rhythmogenic brainstem microcircuits can be identified. These microcircuits are capable of maintaining respiratory rhythmic activity even when isolated from the remaining respiratory network (68, 69). So far, three such microcircuits have been identified: the preBӧtzinger Complex (preBӧtC), which generates inspiration (70); the retrotrapezoid/parafacial nucleus (RTN/pFRG) critical for active expiration (71–73) and sensing hypercapnia (74–76); and the postinspiratory complex (PiCo) (77) associated with postinspiratory activity. These networks receive descending inputs from various areas including the Kölliker-Fuse/Parabrachial (KF) nuclei which exert significant additional regulatory control. As will be discussed later in this review, lesions of this region result in apneustic breathing characterized by prolonged inspirations and extended expiratory pauses (78). Respiratory activity in each of these brainstem centers, with the exception of RTN/pFRG, is depressed by opioids. However, so far only the preBӧtC has been demonstrated to be essential for in vivo breathing and survival in mouse models (79–82), but see Lalley et al. (83), making this region a critical focus for understanding the depressive action of opioids on respiratory motor generation.

The PreBötzinger Complex

Although multiple brainstem sites are involved in OIRD (11), the preBötzinger Complex (preBötC) is particularly important, because inhibiting this medullary region in vivo leads to respiratory failure (84, 85). The mouse preBötC microcircuit is composed of ∼800–1,000 glutamatergic neurons bilaterally distributed to two ventrolateral nuclei (32, 69), which are part of the larger ventral respiratory group (VRG) column composed of the Bӧtzinger complex (BC) (rostral to the preBӧtC), the rostral ventral respiratory group (rVRG), and the caudal ventral respiratory group (cVRG) (progressively more caudal to the preBӧtC). The transiently expressed embryonic homeobox transcription factor Dbx1 provides a useful molecular marker for a significant fraction of these glutamatergic neurons in Cre/loxP transgenic mouse lines (86). However, additional molecular markers may also define other nonoverlapping subsets of these excitatory neurons (87–89). Based on computational models, these neurons are thought to be connected in a sparse network composed of multiple hubs, each enriched in local short-range synaptic interactions (90–93). Because rhythmic activity persists following pharmacological blockade of inhibitory transmission (GABAergic and glycinergic), it has been concluded that inspiratory rhythmogenesis primarily depends upon excitatory glutamatergic transmission (94, 95). However, inhibitory transmission during both inspiratory and expiratory phases of the rhythm plays a critical role in controlling respiratory frequency, and shaping the amplitude and response dynamics of the respiratory rhythm (36, 96).

Inspiratory bursts emanating from the preBötC are highly synchronous rhythmic bursts generated over a substantial fraction of neurons in the network, apparently arising from a threshold assembly of less completely synchronized “burstlets” generated by small subsets of neurons within the network (97, 98). Only fully synchronous inspiratory bursts are transmitted in part through a pool of somatostatin-positive (Sst⁺) preBötC interneurons, to distinct secondary pools of premotor interneurons interposed between the preBӧtC and respiratory-driven motoneuronal pools in the brainstem and spinal cord (57, 99). Inspiratory bursts result from the dynamic interactions of several functional classes of preBӧtC neurons, defined by their firing patterns from single cell recordings including: 1) quiescent cells, whose activity is recruited during strong respiratory bursts, 2) tonic cells with firing frequencies modulated by respiratory phase, 3) conditional bursters with firing timed with inspiration, but which fail to burst upon pharmacological blockade of synaptic transmission, and 4) intrinsic bursters or “pacemaker” neurons which fire bursts during inspiration, and which persist in isolation from all synaptic transmission. The topology of this network is thought to allow for the flexibility needed to make rapid and reversible adaptive changes in cycle-to-cycle burst phasing or scaling of inspiratory output, in response to acute sensory and neuromodulatory input (90, 91). Synaptic dynamics may play an important role in shaping and constraining the oscillatory properties of this network (100). Recurrent excitatory connections are likely another important general feature of this network, since simultaneous photoactivation of a small randomly-selected subset (<1%) of preBӧtC neurons suffices to initiate full inspiratory bursts, measured by propagation to the XII hypoglossal nucleus (101). This nonlinear dynamic behavior is remarkably reminiscent of similar findings in the cortex (102, 103). Conversely, random photoablation of 10%–17% of neurons in this network is enough to silence rhythmic activity (92, 104–106). These findings highlight the critical role and the unique vulnerability of glutamatergic transmission, in the context of this network architecture for inspiratory burst generation. Overlaid on this requirement for glutamatergic transmission is the contribution from intrinsic bursters, although the relative importance of these neurons to network activity remains unresolved (90, 107, 108). Intrinsic bursters possess specific ion channels required for autonomous bursting, many of which may be modulated by opioid receptor activation. Thus, both the molecular components underlying glutamatergic synaptic transmission and ion channels necessary for pacemaker activity present plausible molecular targets for modulation by opioids, with relevance for understanding and reversing opioid-mediated depression of respiratory output from the preBӧtC.

Opioid Action on the PreBötC—Cellular Mechanisms

Opioids act through binding to a family of four G-protein coupled receptors (GPCRs) (mu-, delta-, kappa-, and nociceptin/orphanin FQ), which signal through the canonical Gα_i/o intracellular signaling pathway (Fig. 3). Mouse KO studies have implicated the mu-opioid receptor (encoded by Oprm1, hereafter referred to as MOR) as the primary receptor type underlying both antinociceptive and adverse nonanalgesic side effects, including opioid-induced respiratory depression (OIRD) and constipation due to suppression of gastric motility (109, 110). In addition to canonical signaling via Gα_i/o signaling, MOR is also capable of noncanonical signaling by the recruitment of β-arrestin2 to its phospho-C-terminal tail, following activation. Recruited β-arrestin2 mediates a well-characterized role for recycling ligand-bound receptors from the plasma membrane, via clathrin-mediated endocytosis. However, β-arrestin2 may also serve as a scaffolding platform for a variety of signaling molecules, including kinases, phosphatases, and phosphodiesterases (111, 112). These assembled signaling complexes provide the basis for non-canonical intracellular signaling events, independent and exclusive of G-proteins (113, 114). Early reports using β-arrestin2 KO mice suggested that beneficial opioid-mediated analgesia could be dissociated from adverse OIRD and constipation side effects in the absence of β-arrestin2, implicating noncanonical intracellular signaling as a key mechanism underlying these adverse side effects (115). This led to substantial interest in the development of “biased agonist” drugs to promote receptor signaling exclusively through the canonical G-protein pathway, with the aim of providing analgesia while minimizing adverse side effects (116, 117). However, the validity of this concept has been questioned by recent studies, using transgenic mice engineered to express phosphorylation-defective MOR, and by a multilaboratory re-examination of OIRD and constipation phenotypes in β-arrestin2 KO mice, both of which failed to show any significant difference between WT and genetically altered mouse lines (118, 119). Presently, it remains unclear how much non-canonical signaling contributes to OIRD, or whether unique effectors of non-canonical signaling may be identified and harnessed to provide analgesia without adverse side effects. However, these conflicting recent studies re-emphasize the importance of more fully examining well-established mechanisms and molecular effectors of canonical Gα_i/o signaling as potential mediators of OIRD.

Figure 3.

Schematic illustrating the canonical MOR signaling pathways via G_i/o-protein coupling. G_βγ can directly activate GIRK currents and directly inhibit voltage-gated Ca²⁺ currents. G_αi inhibits adenylyl cyclase, reducing the intracellular concentration of cyclic AMP. Reduced cAMP concentrations inhibit voltage-activated cation currents, such as HCN and can also indirectly influence the activity of numerous downstream effectors through inhibition of PKA. The predicted effects of these mechanisms on preBötC network function and the inspiratory rhythm depend on their somatic and/or synaptic sites of action. Predicted effects are color coded to the canonical pathways outlined above. GIRK, G-protein-activated inward rectifier K⁺ channels; MOR, mu-opioid receptor; preBOTc, preBӧtzinger complex.

The two best characterized molecular effectors downstream of MOR activation are ion channels: 1) G-protein-activated inward rectifier K⁺ channels (GIRK), opened by direct binding of G_β/γ released from activated MOR (120–124), and 2) Presynaptic voltage-gated calcium channels (N-, P/Q- and R-type) whose voltage-dependent gating is suppressed by the binding of G_β/γ to the second intracellular loop between homology domains I and II (125–130). Canonical opioid-mediated Gα_i/o intracellular signaling to either of these ion channels would be expected to suppress excitability (Fig. 3). A role for GIRK potassium channels activated through MOR signaling was examined by Montandon and colleagues, and proposed as the primary mechanism for OIRD (131–133). However, this may be an oversimplification, and there is also increasing evidence for a significant role for presynaptic voltage-gated calcium channel inhibition in OIRD. A recent study suggests that MOR-mediated suppression of presynaptic calcium channels, and the integrity of glutamatergic transmission more generally, is a major mechanism underlying OIRD in the preBötC (Fig. 4). This conclusion is based on the following observations: 1) Heterozygous loss of a single copy of Cacna1a (encoding the P/Q-type calcium channel) sensitized preBötC slices to OIRD by DAMGO ([D-Ala2, N-MePhe4, Gly-ol]-enkephalin, a MOR-specific agonist), 2) DAMGO reduces the frequency of spontaneous miniature excitatory postsynaptic currents (mEPSCs) recorded from Dbx1⁺ respiratory neurons in the preBӧtC, consistent with reduced presynaptic calcium channel activity and lower intracellular Ca²⁺ concentrations (20). Moreover, this study also demonstrated that at the presynaptic terminal, KCNQ potassium channels play a critical role in modulating the depressive effect of DAMGO on synaptic transmission. Application of XE991, a blocker of KCNQ channels, increased the frequency of mEPSCs after suppression by DAMGO, in a large fraction of recorded neurons (86%, 6/7). This effect was corroborated at the network level by population recordings of inspiratory bursts from isolated in vitro preBӧtC slice preparations. Blockers of KCNQ channels (Chromanol 293B, XE991) reversed DAMGO-mediated OIRD in slices, and KCNQ activators (ICA69673, Retigabine) suppressed inspiratory rhythms, mimicking OIRD in the absence of MOR activation (20).

Figure 4.

A: schematic illustrating the proposed effect of opioids at excitatory synapses on rhythmogenic Dbx1 neurons. In response to MOR activation, G_βγ-mediated inhibition of voltage-gated Ca²⁺ currents reduces the probability of vesicular release in response to bouts of action potentials that occur during inspiratory bursts. B: PreBötC slices from mice with a heterozygous deletion of Ca_V 2.1 encoding P/Q-type Ca²⁺ channels are ∼10-fold more sensitive to the MOR agonist DAMGO than WT controls (data shown as median ± interquartile range). C: representative voltage clamp recordings of mEPSCs from a Dbx1 neuron following blockade of action potentials with TTX (1 µM). The frequency of mEPSCs is reduced in the presence of DAMGO (100 nM), and subsequent blockade KCNQ channels with XE991 (20 µM) restores mEPSC frequency to baseline levels. Data adapted with permission from Wei and Ramirez (20). KO, knockout; mEPSCs, miniature excitatory postsynaptic currents; MOR, mu-opioid receptor; P, probability; preBOTc, preBӧtzinger complex; WT, wild type.

Other potential OIRD-related molecular effectors are suggested by studies which report reversal of OIRD by pharmacological manipulations related by their actions to increase intracellular cAMP. These include: 1) agonists of Gα_s-coupled GPCRs (5-HT4a receptors, Ref. 24; D1 dopamine receptors, Refs. 134–136), 2) Methylxanthines, caffeine, and theophylline given to counter apneas of prematurity (137–139), and 3) Phosphodiesterase 4 (PDE4) inhibitors (rolipram) (140–142). Each of these manipulations would be expected to counter cAMP decreases due canonical Gα_i/o signaling downstream of activated MOR (143). We speculate that molecular machinery affecting glutamatergic synaptic transmission or electrical excitability, modulated directly by cAMP or indirectly by PKA phosphorylation, could thus be implicated as possible effectors and contributors to OIRD. Among potential ion channels, HCN (hyperpolarization-activated cyclic nucleotide-gated) cation channels are plausible OIRD-related targets because they are modulated by directly binding cAMP and play an important role in controlling neuronal excitability (144–146). These ion channels underlie slow depolarizing I_h currents triggered upon after-hyperpolarization during interburst intervals, implicated in pacemaker activity (147). They also regulate dendritic input resistance and thus synaptic efficacy, by shunting ligand-gated postsynaptic currents (148, 149). Another attractive class of ion channels is TWK-type two-pore potassium channels (KCNK) which mediate K⁺ leak conductance through weakly voltage-dependent, outwardly-rectifying K⁺ currents, many of which are modulated by both volatile anesthetics that depress respiration, and PKA phosphorylation (150). The respiratory stimulant Doxapram acts through blocking TASK1 and TASK3 channels in peripheral carotid bodies, increasing the gain of chemosensory input to the brainstem (151, 152). Although these channels are expressed in the preBӧtC (153), a central role for TASK channels in OIRD has not been examined.

G-protein modulation of voltage-gated sodium channels is another well-characterized general phenomenon within the nervous system, mediated in part by PKA phosphorylation of sites between homology domains I and II, affecting both peak transient and persistent components of sodium conductance (154, 155). A possible role for opioid modulation of voltage-gated sodium channels in OIRD has yet to be investigated. However, a critical role for persistent sodium currents has been demonstrated for sustaining the rhythmicity of perinatal brainstem respiratory circuits, carried by the NALCN “sodium leak” channel (156). These channels were first described to be modulated by CaSR, a Gα_q-coupled GPCR which senses extracellular Ca²⁺(157), but recent reports also suggest that NALCN channels are susceptible to modulation by Gα_i/o-coupled D2 dopamine receptors, suggesting a similar possible susceptibility to MOR (158, 159).

Additionally, G-protein signaling has been shown to control synaptic efficacy by modifying both presynaptic and postsynaptic molecular targets. At some presynaptic terminals, vesicular release is suppressed by phosphorylation-independent binding of G_β/γ to SNAP-25 (160–164). Postsynaptically, constitutive PKA-dependent phosphorylation maintains AMPA glutamate receptors at postsynaptic densities, underlying long-term potentiation (LTP) and long-term depression (LTD) (165–168). These observations suggest additional molecular targets for investigation as mediators of OIRD, consistent with the demonstrated ability of ampakines such as CX1942 to reverse centrally derived OIRD (21, 169, 170).

In summary, multiple molecular and cellular mechanisms likely contribute to OIRD in the preBӧtC, constrained by the network architecture and specific electrical properties of its constituent neurons. Current evidence suggests that MOR-mediated inhibition of presynaptic calcium channels is a significant component of OIRD, which is consistent with the demonstrated critical role for glutamatergic transmission in sustaining rhythmic network activity. However, other mechanisms likely contribute to OIRD. New molecular components contributing to OIRD may emerge from future studies, focusing on candidate ion channels and effectors affecting glutamatergic transmission known to be modulated downstream of canonical Gα_i/o signal transduction.

Opioid Effect on the PreBötC In Vivo

Multiple studies in vitro have shown that administration of the synthetic opioid, DAMGO into the isolated brainstem slice preparation generally results in a reduction in fictive respiratory frequency (20, 21, 171), suggesting the importance of this medullary subregion in OIRD. However, when opioids are administered locally into this region in vivo, effects on ventilation have been shown to be variable, resulting in either a decrease, no-change, or an increase in respiratory activity. For example, Montandon et al. (131) showed that dialysis of DAMGO into the preBötC of rats during wakefulness and sleep resulted in depression of respiratory activity. By contrast, Langer et al. (172) and Krause et al. (173) demonstrated that injection or dialysis of DAMGO into the preBötC of awake and alert goats resulted in either no change or an increase in respiratory activity. Furthermore, Mustapic et al. (174) demonstrated that similar DAMGO dialysis into the preBötC of decerebrate dogs led to an increase in overall respiratory activity. If the opioid-sensitive preBötC is indeed necessary for eupneic ventilation in the intact in vivo system, this raises the question of why there is such discordance in the effects of opioid administration into this medullary region in intact physiological preparations.

One hypothesis proposed for the variable effect of in vivo opioid administration into the preBötzinger complex is a concept termed “interdependence of neuromodulators.” This concept, as proposed by Doi and Ramirez (175) states that, “A modulator’s action is determined by the concurrent modulation and interaction with other neuromodulators.” In other words, the action of a single neuromodulator, such as a therapeutic opioid may be compensated for, or influenced by the action of other endogenous neuromodulators within the network. This hypothesis was based on studies comparing the effects of Substance P receptor (NK-1) antagonism on respiratory network activity, when administered in vitro on rhythmic brainstem slice preparations and in vivo locally into the preBötC of intact animal preparations. These studies demonstrated that NK1 antagonism in the isolated in vitro brainstem preparation resulted in a reduction in fictive respiratory frequency, while in vivo local NK1 receptor antagonism within the intact preBötC urethane-anesthetized preparation failed to reduce spontaneous breathing rate. However, in vivo respiratory activity could be reduced following concurrent antagonism of NK1, α₁-noradrenergic, and 5-HT_2A receptors within the preBötC (175). This suggested that changes in one neuromodulator within the preBötC are likely influenced by compensatory changes in one or more neuromodulatory systems in order to maintain stable eupneic ventilation. Similar observations have been suggested in earlier studies of neural networks in the crustacean stomatogastric ganglion (176), and the neuromuscular junction of Aplasia by Marder and Brezina, et al., who described this phenomenon as “redundancy” or “degeneracy” of neural control networks (177–179).

The presence of “redundancy,” “degeneracy,” or “interdependence of neuromodulators” within the preBötC creates a barrier for direct interpretation of studies examining the effects of opioids on the preBötC in vivo, because direct agonism of opioid-receptors may be partially or completely compensated by changes in other neuromodulators. Indeed, this was shown to be the case when Langer et al. (172) observed a paradoxical increase in ventilation when DAMGO was dialyzed into the preBötC of awake and alert goats. In these studies, goats were given 10, 50, and 100 μM of DAMGO into the preBötC through microdialysis. When dialyzed unilaterally, DAMGO resulted in no-change in ventilation. However, it was found that DAMGO dialysis led to a dose-dependent reduction of the inhibitory neuromodulator GABA, suggesting that a DAMGO-dependent reduction in GABA-mediated inhibition offset the inhibitory effects of DAMGO within the preBötC. Similarly, bilateral dialysis of DAMGO into the preBötC of awake and alert goats resulted in an overall increase in ventilation, however with no change of any neuromodulators measured in effluent dialysate. Additionally, despite no-change or an increase in ventilatory rate, it was shown that respiratory rhythm was destabilized following both unilateral and bilateral dialysis of DAMGO into the preBötC. These studies are important because they: 1) Further highlight the possible variability of OIRD following opioid administration directly into the preBötC of awake and alert animals, 2) Provide evidence of neuromodulatory mechanisms within the preBötC, capable of compensating for the depressant effects of opioids, and 3) Suggest the need to understand how these compensatory neuromodulator systems may be harnessed across different physiological conditions to prevent or reverse OIRD.

If indeed there are intrinsic mechanisms within the preBötC capable of compensating for, or reversing OIRD, then understanding the conditions and states of the respiratory network that lead to resistance or reversal of OIRD will be key in understanding the specific mechanisms leading to OIRD, and for developing strategies to protect against OIRD. For example, neuromodulator compensation, or more generally, neuromodulation has been shown to be state-dependent across various brain states such as sleep and wakefulness (180). Additionally, within the preBötC, chronic intermittent hypoxia has been shown to destabilize rhythm generation, within the isolated brainstem (84). These studies strongly suggest that mechanisms of respiratory rhythmogenesis within the preBötC may be dynamically set across varying physiological conditions. Further, it is likely that the sensitivity of the respiratory network to perturbations such as opioid exposure likely changes under different “states” of rhythmogenesis. Thus, future studies aimed at understanding the relationship between state-dependency, neuromodulation, and rhythm generation will be necessary for unraveling the specific action of opioids within the brainstem and for creating strategies to prevent or reverse OIRD.

PONS

The pontine respiratory group in the dorsolateral pons was shown long ago to be important for normal breathing and opioid-induced respiratory depression (181, 182). Key structures that comprise the pontine respiratory group are the lateral parabrachial nucleus (LPB), medial parabrachial nucleus (MPB), and Kölliker-Fuse (KF). The KF/PB significantly contributes to the control of respiratory rate (183–186), regularity (187), eupneic pattern formation (188–191), and chemosensory reflexes (192–196). In addition, the KF plays a major role in coordination of the upper airways in breathing and swallowing (189, 197–199).

During opioid-induced respiratory depression, the majority of these functions are compromised (200–202), including impairment of the upper airways at therapeutically relevant doses (203–205). The correlation between the clinically observed effects of opioids and the known functions of the KF/PB is striking. However, due to the abundance of mu opioid receptors in the majority of the respiratory control areas of the brainstem, it has proven challenging to resolve the neural substrates of respiratory malfunctions during overdose. Further complications arise when one considers the numerous reciprocal projections and feedback loops present in the respiratory network.

Opioid Effect on the KF/PB In Vivo

Direct administration of opioid agonists locally into the KF/PB results in suppression of respiratory rate (Fig. 5), due to increases in the duration of both inspiration and expiration (182, 184, 206, 207). This reduction in respiratory rate is consistently observed across species, including cat, dog, rabbit and rat. This is in contrast to the conflicting results observed following local opioid administration within the ventral respiratory column, particularly the preBötC, as described above (131, 172–174, 208, 209).

Figure 5.

Opioid modulation of pontine networks. A: whole cell voltage-clamp recording from a KF neuron in brain slice illustrating activation of GIRK current by opioid agonist Met-enkephalin. Modified from Levitt et al. (184). B: schematic of the in situ working heart-brainstem preparation of the rat. On the right, phrenic bursts indicating inspiratory activity during baseline, following systemic fentanyl administration, and following systemic fentanyl with opioid antagonist, CTAP microinjection into the KF. Antagonism of KF MORs eliminates rate depression by fentanyl; however, the apneustic pattern characteristic of opioid effects persists. Modified from Saunders and Levitt, 2020 (211). C: schematic of in vivo plethysmograph. Graph on the top right demonstrates significant attenuation of morphine-induced respiratory rate depression in vivo following bilateral MOR deletion from KF neurons compared to control animals. Below, example time course of average respiratory rate following saline and morphine (10 mg/kg, i.p.) in control mice (left) vs. mice following MOR deletion from KF neurons (right). Modified from Varga et al. (212). GFP, green fluorescent protein; GIRK, G-protein-activated inward rectifier K⁺ channels; KF, Kölliker-Fuse; MOR, mu-opioid receptor. *,**,***Statistically significant (P < 0.05) in order of least to most significant.

Systemic administration of opioids causes reductions in respiratory rate due to small increases in inspiration and large increases in expiration (11, 169, 210). In the presence of systemic opioid administration, antagonism of MORs (both pre- and postsynaptic receptors) located in the KF/PB at least partially restores respiratory rate, mostly due to reductions in expiratory duration (206, 207, 211). Antagonism of opioid receptors in the KF/PB also prevents and reverses sustained apnea following systemic fentanyl or remifentanil (206, 211). Similarly, deletion of MORs from a large portion of KF neurons reduces the number of apneas induced by morphine in awake mice (212). Together, these studies suggest that opioids in the KF/PB enhance tonic expiratory drive leading to prolonged expiration and apnea (211). Expiratory drive from KF has also been implicated in generating apnea in Rett Syndrome and nasotrigeminal reflex (213, 214). The mechanisms within the KF generating apnea are likely different because enhanced postinspiratory discharge accompanies the apneas of RTT and nasotrigeminal reflex, whereas postinspiratory discharge is abolished by opioids.

In awake mice, removal of MORs from KF neurons significantly rescues morphine-induced respiratory rate depression (87, 212). This occurs at a therapeutically relevant analgesic dose and, importantly, also at a very high dose that is near overdose (212). The improvement in rate is due to shortening of both inspiration and expiration. By comparison, removal of mu opioid receptors from preBötzinger complex neurons only rescues morphine-induced rate depression at analgesic doses, but not at a high dose (212).

The rate changes caused by opioid modulation come hand-in-hand with alterations in respiratory pattern, which are easily observed and analyzed in the in situ working heart brainstem preparation of the rat (215). The in situ preparation produces a eupneic respiratory cycle with a clear readout of three respiratory phases in nerve recordings in a nonanesthetized, vagotomized setting. Local injection of the opioid agonist DAMGO into the KF of the in situ preparation leads to the emergence of an apneustic 2-phase respiratory pattern, exhibiting loss of post-inspiration and lengthened, low amplitude inspiratory bursts (184). This pattern is very similar to bilateral pontine lesion or pharmacological inhibition of the KF in the in situ preparation (189, 216) and has implications for both the quality of breathing and patency of the upper airways. This two-phase respiratory pattern is also present, though at a higher frequency, in preparations with systemic fentanyl and antagonism of KF/PB opioid receptors (211), indicating that blockade of receptors in the KF/PB is not sufficient, on its own, to restore a normal breathing pattern. Local injection of DAMGO into the KF also increased the variability of phrenic nerve bursting (36), but deletion of MORs from KF neurons did not improve the increased variability of breathing rate caused by systemic morphine in awake mice (69).

Opioids also decrease tidal volume—an effect often attributed to muscle rigidity and decreased chest wall compliance (217–219). Opioids, especially fentanyl, cause muscle rigidity, which is blocked by opioid receptor antagonist and alpha-2 adrenergic receptor agonist, and proposed to be due to central opioid receptor-mediated increase in adrenergic activity (220–223). However, in paralyzed preparations fentanyl causes a decrease in the amplitude of phrenic motor output, indicating that there is also a central component to the reduction of tidal volume (169, 211). Antagonism of opioid receptors in the KF/PB does not restore the amplitude of phrenic motor output (211), indicating that the reduction in tidal volume does not occur in the dorsolateral pons.

Opioid Action on the KF/PB—Cellular Mechanisms

The effects of opioid administration into the KF/PB largely mimic the effects of global inhibition of the KF/PB by pharmacological inhibition or ablation (186, 188, 189, 193, 216, 224, 225), indicating that opioids have a strong inhibitory action in the KF/PB. The KF/PB express an abundance of MORs (226–228). A proportion of KF neurons (∼60%) contain somatodendritic MORs that directly hyperpolarize KF neurons by activation of GIRK conductance (184, 212). Opioid-sensitive and nonsensitive KF neurons have different intrinsic properties (firing rate, action potential duration, and afterhyperpolarization), indicating that they are indeed separate populations of neurons serving distinct functions (184). Lateral parabrachial complex neurons also contain somatodendritic MORs that activate GIRK conductance (229). The sensitivity of the somatodendritic opioid receptors on KF/PB neurons is similar to somatodendritic opioid receptors on other neuronal populations involved in pain or reward (184, 229).

MORs are also expressed on neuronal processes in the KF (226–228). These processes could include reciprocal projections to the KF from medullary respiratory structures, including preBötzinger complex, Bötzinger complex, or NTS (82, 230–232). The functional contribution of presynaptic opioid receptors in the KF/PB is yet to be determined.

The KF/PB contains a heterogeneous population of respiratory patterned neurons with various projection targets. Glutamatergic KF neurons send dense projections to the majority of the ventral respiratory column (198, 233–237), as well as the nucleus tractus solitarius (NTS) and hypoglossal motor nucleus (197, 236, 237). Many KF neurons fire phasically relative to the respiratory cycle, including neurons with phasic firing (discharge identity) in inspiration, postinspiration and expiration (189, 233, 238–240). The discharge identity of some KF neurons changes following exposure to mu opioid agonist DAMGO (238), but a full understanding of the discharge identity and projection target of opioid-sensitive KF neurons is lacking. This understanding would help predict the effect of modulating KF circuits in respiratory depression.

Chemosensory and Arousal Pathways in the Pons

In addition to rate and pattern changes, opioids reduce hypercapnic and hypoxic ventilatory responses (11), due to activation of mu opioid receptors (109). In opioid overdose, limited, or even lack of hypoxic/hypercapnic arousal responses can occur when respiratory cessation is abrupt, such as with high-dose fentanyls (11). Under normal conditions, changes in blood gases lead to shifts in behavioral state toward arousal to increase ventilatory drive and stabilize blood gas concentrations (241). Respiratory chemosensory nuclei in the pons can either serve as relay centers for blood gas information, or serve as arousal centers that directly or indirectly feed-back to respiratory control centers to increase respiratory drive.

The PB/KF contains a population of glutamatergic neurons that are part of a chemosensory relay circuit projecting to higher order arousal centers (194, 236, 242–245). Stimulation of PB, especially external lateral PB neurons that express CGRP (calcitonin gene-related peptide), causes cortical arousal, meanwhile lesions of the glutamatergic relay population leads to increased amount of sleep and decreased hypercapnic arousal responses (194, 243, 246, 247). These CGRP-expressing external lateral PB neurons contain mu opioid receptors (227), thus opioid-induced inhibition of these neurons likely contributes to the sedative effects of opioids.

The locus coeruleus (LC) is another pontine region critical for a wide range of homeostatic functions, including arousal, which is studied mainly in the context of sleep apnea and sleep-wake states (248–250). The LC is directly involved in inducing arousal when blood CO₂ levels rise (251–255). Since all LC neurons contain mu opioid receptors and are hyperpolarized by opioids (256), this region is also a prime candidate for investigating opioid-induced loss of hypercapnia reflexes.

Endogenous Opioid Expression in Pons

A large population of KF/PB neurons express pre-pro-enkephalin mRNA and produce enkephalin, an endogenous opioid peptide that activates mu and delta opioid receptors (257–260). KF neurons containing pre-pro-enkephalin project to the ventral medulla and the spinal cord (257). Enkephalin-producing KF neurons may be one source of endogenous opioids important in regulating respiration. The endogenous opioid peptide endomorphin-2 is also expressed in neuronal fibers of the KF/PB and overlaps with MOR-expressing fibers (227). Global MOR knockout mice not only lack respiratory depression in response to exogenously administered opioids, but also breathe at a substantially higher rate at baseline than littermate controls, indicating an endogenous opioid tone in the respiratory network (109). However, MOR removal from KF and/or preBötzinger complex neurons does not increase baseline respiratory rate, indicating that the KF and preBötzinger complex may not play a significant role in maintaining this baseline inhibitory tone (212).

Pons—Conclusions

Given the high proportion of mu opioid receptor expression and the wide range of respiratory-related roles played by the pontine respiratory group, it is predictable that the pons has a substantial contribution to OIRD. In particular, opioid actions in the pons contribute to reduction in respiratory rate, loss of upper airway patency, and degenerate pattern formation. The KF appears to be especially important for apnea generation following high-dose opioids. We infer that countering opioid effects in the KF is especially important in overdose. Some countermeasure targets with potential activity in the KF/PB include orexin and 5-HT1A receptors (261–264). Further work into understanding the mechanisms of opioids in pontine circuits will hopefully produce additional countermeasure targets.

PREFRONTAL CORTEX

Although breathing is known to be generated by the pontomedullary brain stem (33, 232), the goal of this section is to highlight evidence that the prefrontal cortex (PFC) is a brain region mediating OIRD (265). The PFC also contributes to pain processing (266) and addiction (267). Functional studies reviewed elsewhere propose that PFC dysfunction underlies loss of inhibitory control over appetitively motivated behaviors, such as drug abuse (268). In support of this concept, recent studies show that transcranial, direct current stimulation over human dorsolateral PFC causes a significant decrease in drug craving and reduces deficits in cognitive control (269).

The forgoing findings and scientific constructs raise many questions regarding the mechanisms through which the PFC modulates a wide range of complex phenotypes. Partial answers include the fact that the PFC comprises multiple subregions (270) containing cells that project to and receive projections from widely distributed neuronal networks (271). Anatomical connections between PFC and respiratory nuclei in the pontine and medullary brain stem have been documented (272). One example of the diverse connectivity through which the PFC might exert effects on breathing is the discovery of monosynaptic projections from medial regions of rat PFC to the thoracic spinal cord (273). Although the PFC contains no respiratory neurons, it has been known since the 19th century that breathing frequency can be altered by electrically stimulating the PFC (see Lépine, 1875 in Ref. 274).

Contemporary pharmacological data show that microinjecting the GABA-A receptor antagonist bicuculline into rat PFC causes site-specific, apneic breathing (275). Medial regions of the rodent PFC are organized into four layers comprised of up to 90% excitatory pyramidal neurons and 10 to 20% inhibitory GABAergic interneurons (276). Systematic reviews of PFC neurochemistry (277) indicate the contribution of at least eight neurotransmitter systems to PFC function. Pharmacological studies confirm the presence of opioid receptors in PFC of rodent (278) and human (279). Mu, kappa, and delta subtypes of opioid receptors are widely and differentially distributed throughout the rodent (278) and human (280) brain. Thus, connectivity and opioid receptor data provide ample substrates through which systemically administered opioids can simultaneously inhibit and disinhibit widely distributed neuronal networks. The functional significance of these receptor and network data is clear from the observation that systemically administered opioids alter the cortical electroencephalogram (EEG) and corresponding states of consciousness (281). Figure 6 illustrates that in mice, cortical EEG power and frequency are differentially altered by antinociceptive doses of morphine, fentanyl, and buprenorphine (282).

Figure 6.

Examples of how tapered spectrograms reveal that administering saline (A), morphine, (B), buprenorphine (C), or fentanyl (D) differentially changes cortical EEG power and frequency in mice. E illustrates how EEG power can be quantified to show opioid-specific alterations. These spectrograms demonstrate the importance of understanding nuanced effects of opioids on cortical excitability. Modified from O’Brien et al. (282).

The cortical EEG data and recent studies of PFC (283) encourage evaluation of the hypothesis that administering opioids directly into mouse PFC causes respiratory depression. Figure 7 shows that microdialysis delivery of fentanyl into the PFC of intact, behaving mice depresses breathing (284). Additional preliminary data show that microinjection of morphine into PFC of awake mice decreases minute ventilation (285).

Figure 7.

An example of how breathing in mice is depressed by direct administration of fentanyl into the prefrontal cortex A: adequate respiratory control includes the ability to vary breathing in response to environmental challenges. B shows how Poincaré analyses can be used to visualize the decrease in breathing variability caused by delivering fentanyl into mouse prefrontal cortex. Modified from Zhang et al. 2019 (284). PFC, prefrontal cortex. *P < 0.05.

In summary, the PFC has long been known to modulate breathing (274), but evidence that the PFC may contribute to addiction (268, 269) and OIRD (265) has appeared only recently. The preliminary data of Figs. 6 and 7 encourage future studies designed to specify which regions of the PFC modulate breathing and OIRD. Such studies should also clarify the mechanisms through which the PFC interacts with other neural networks involved in respiratory control. The pontine parabrachial nuclei modulate the timing of breathing and the PFC projects to (286) and receives projections from (272) the parabrachial respiratory nuclei. Considered together, these data support the view that future studies of the interaction between the PFC and parabrachial nuclei will enhance understanding of OIRD.

PERIPHERAL ACTIONS OF OPIOIDS

Although central nervous system (CNS) mechanisms are certainly critically involved in OIRD, it is also important to consider the action of opioids on the peripheral nervous system. To our knowledge, the first evidence that opioids can have peripheral actions to alter breathing was published by Gruhzit (287) in the late 1950s. This investigator showed that bolus i.v. administration of codeine, morphine or the nonopioid drug, dextromethorphan could induce rapid shallow breathing and apnea in dogs or cats (287). Vagotomy or vagal cooling sufficient to block c-fibers eliminated this effect (287). Since that time, a wealth of evidence has emerged that strongly supports the concept that peripherally administered opioid drugs, especially µ-opioid agonists, can induce apnea and rapid shallow breathing (288–296).

Pulmonary Opioid Receptors

MORs are found in lung tissue, on airway sensory nerves, and in vagal and spinal ganglia (297–299). Utilizing autoradiography, it was shown that the highest density of binding sites for [³H] morphine was in the alveolar walls (300) with lower densities in the larger airways. In lung tissue preparations, the binding affinities of tritiated morphine and naloxone were lower than in the brain, but the binding site densities were up to two orders of magnitude higher (299). These data suggest a high level of importance for opioid receptor-mediated mechanisms in the lung.

The mechanism by which MOR agonists are thought to alter the breathing pattern is through excitation of pulmonary c-fibers (290). Excitation of these c-fibers with peripheral vascular administration of TRPV-1 agonists and other drugs (e.g., capsaicin, phenylbiguanide) is well-known to elicit apnea and rapid shallow breathing in a variety of species (301). It is presumed that MOR agonists act on opioid receptors located on sensory terminals of c-fibers. However, local application of MOR agonists in the nodose ganglion also can elicit apnea and/or tachypnea consistent with stimulation of airway c-fibers (289). The functional relevance of opioid receptor modulation of vagal sensory traffic at the level of the nodose ganglion is not well understood.

Opioids elicit local effects in the airways that may alter the pattern of breathing through changes in pulmonary mechanics. The efferent function of c-fibers to release the pro-inflammatory neuropeptide, substance P, can be inhibited by a MOR agonist (302). Both mucus secretion and cholinergic-mediated contraction of tracheal smooth muscle are inhibited by MOR agonists as well (303, 304).

Carotid Body Opioid Receptors

Endogenous opioids (leu- and met-enkephalin) are present in carotid body glomus cells (305, 306). Intracarotid injections of MOR agonists inhibit spontaneous chemoreceptor discharge but have smaller depressive effects on acetylcholine-evoked discharge of these sensory receptors (307). However, an in vivo rank order potency study showed that carotid chemoreceptor discharge was more affected by delta opioid receptor agonists than those with high affinity for the MOR (308).

In anesthetized cats, the peripheral influence of morphine on the isocapnic hypoxic ventilatory response may be limited to depression of CO₂ sensitivity during hyperoxia (309). In conscious rats, there is evidence that morphine more strongly depresses the hypoxic and hypercapnic ventilatory responses after section of the carotid sinus nerve (310), suggesting that MOR-related mechanisms in the carotid body may ameliorate opioid-related respiratory depression. Hyperoxia in spontaneously breathing humans enhances opioid-induced respiratory depression, presumably via profound reductions in carotid body afferent feedback (311, 312). It is possible that this straightforward hypothesis needs revision. The effect of opioids on carotid body sensory afferents in concert with recent preclinical in vivo results strongly suggest that the peripheral effects on hypoxic and hypercapnic ventilatory mechanisms are likely conditional and more complex that previously appreciated.

Opioid Receptors and Actions in the Nucleus of the Tractus Solitarius, as the Integration/Relay Station for Peripheral Inputs

NTS interneurons mediating airway sensory information include pump cells (313, 314), Iβ cells (313), laryngeal interneurons (315, 316), neurons mediating pulmonary c-fiber reflexes (313, 316), and neurons mediating rapidly adapting receptor information (317). Extra-NTS targets of these second-order populations include the pons, VRC, and cervical spinal cord (313). Within the NTS, slowly adapting receptor (SAR) second-order interneurons monosynaptically inhibit their rapidly adapting receptor (RAR) counterparts in the rat (318). Pump cells can project contralaterally to excite or inhibit Iβ cells (313). A recent study by McGovern et al. (319) using herpes simplex virus mapped central pathways from the extrathoracic trachea and identified a large neuronal network in the NTS and surrounding areas that may exceed the substrate of second-order NTS interneurons by an order of magnitude. Results from these studies strongly support complex processing of sensory feedback from the lungs and airways by NTS circuits.

Opioid receptors are found in the NTS (320). Application of opioids to the caudal NTS can mimic the respiratory effects of peripheral administration of these drugs (321). Relatively little is known regarding the distribution of MOR receptor afferent projections to the caudal NTS, which processes airway vagal c-fiber sensory feedback. Interneurons in the caudal NTS also process sensory feedback from the carotid sinus nerve related to hypoxic responses (322), indicating that this region of the NTS represents a critical area in understanding the anatomical and functional elements that are responsible for the peripheral actions of these drugs on breathing.

OVERALL CONCLUSIONS

Currently there are limited treatment options available for acute opioid overdose and long-term opioid addiction. Following acute overdose, administration of the competitive opioid receptor antagonist, naloxone, remains the standard of care (323). This treatment has proved effective in rapidly reversing the fatal respiratory effects of an opioid overdose (323–325). However, the effects are not limited to the respiratory system and simultaneously cause reversal of analgesia (326), resulting in unpredictable effects and acute withdrawal symptoms (327–329) that may become life threatening. Conversely, there are currently three available medication assisted therapies available for long-term opioid addiction (330). These include the use of controlled opioid-based replacements such as methadone and buprenorphine, or opioid receptor antagonists such as naltrexone. Although these therapies have proven to be effective in a small proportion of patients, there remain major hurdles in the use of these therapies due to poor patient adherence and abuse of opioid-based replacement drugs. Accordingly, alternative therapies targeted at preventing overdose, specifically reversing respiratory suppression, and mitigating long-term addiction are desperately needed.

The NIH initiative to help end addiction over the long-term (HEAL) has called for an “all hands on deck” effort to improve treatments for addiction and to enhance strategies for pain management (331). The HEAL initiative is a logical extension of earlier calls (332) for the development of pharmacological interventions to counter opioid-induced respiratory depression (OIRD). The “all hands” call can be extended to “all brain regions” for research on OIRD. In this review, we provided an overview of the different CNS and PNS regions involved in the neuronal control of breathing and their potential contribution to OIRD. From this overview it is clear, that opioids inhibit breathing at multiple, and highly distributed sites. Inhibition of the different regions is manifested in a variety of differential effects on rhythmogenesis, pattern formation, and neuromodulation. The opioid action on these sites is highly dependent on the metabolic and modulatory state of the organism, thus explaining why OIRD is so variable and unpredictable. Although OIRD is on average dose dependent, no dose is safe. Even small opioid doses under certain metabolic conditions can lead to an overdose. Unraveling these state-dependent mechanisms will be critical to find effective countermeasures to cope with the opioid crisis.

GRANTS

This work was supported by R01 HL144801, R01 HL126523, and P01 HL090554 to J-M. Ramirez; and R01 DA047978, R00 DA038069, and Grant 3608 from IRSF (International Rett Syndrome Foundation) to E. S. Levitt. A. V. Varga was supported by T32 HL134621.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J-M.R., E.S.L., N.J.B., A.D.W., N.A.B., A.G.V., H.A.B., R.L., K.F.M., and D.C.B. drafted manuscript; J-M.R., E.S.L., N.J.B., A.D.W., N.A.B., A.G.V., H.A.B., R.L., K.F.M., and D.C.B. edited and revised manuscript; J-M.R., E.S.L., N.J.B., A.D.W., N.A.B., A.G.V., H.A.B., R.L., K.F.M., and D.C.B. approved final version of manuscript.