Opioid suppression of an excitatory pontomedullary respiratory circuit by convergent mechanisms

  1. Jordan T Bateman
  2. Erica S Levitt  Is a corresponding author
  1. Department of Pharmacology and Therapeutics, Breathing Research and Therapeutics Center, University of Florida, United States

Abstract

Opioids depress breathing by inhibition of interconnected respiratory nuclei in the pons and medulla. Mu opioid receptor (MOR) agonists directly hyperpolarize a population of neurons in the dorsolateral pons, particularly the Kölliker-Fuse (KF) nucleus, that are key mediators of opioid-induced respiratory depression. However, the projection target and synaptic connections of MOR-expressing KF neurons are unknown. Here, we used retrograde labeling and brain slice electrophysiology to determine that MOR-expressing KF neurons project to respiratory nuclei in the ventrolateral medulla, including the preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG). These medullary-projecting, MOR-expressing dorsolateral pontine neurons express FoxP2 and are distinct from calcitonin gene-related peptide-expressing lateral parabrachial neurons. Furthermore, dorsolateral pontine neurons release glutamate onto excitatory preBötC and rVRG neurons via monosynaptic projections, which is inhibited by presynaptic opioid receptors. Surprisingly, the majority of excitatory preBötC and rVRG neurons receiving MOR-sensitive glutamatergic synaptic input from the dorsolateral pons are themselves hyperpolarized by opioids, suggesting a selective opioid-sensitive circuit from the KF to the ventrolateral medulla. Opioids inhibit this excitatory pontomedullary respiratory circuit by three distinct mechanisms—somatodendritic MORs on dorsolateral pontine and ventrolateral medullary neurons and presynaptic MORs on dorsolateral pontine neuron terminals in the ventrolateral medulla—all of which could contribute to opioid-induced respiratory depression.

Editor's evaluation

Opioid-induced respiratory depression is one of the side effects of opioid drugs. Although opioid overdose deaths are highly prevalent, our knowledge of the neural circuits underlying respiratory depression in the brainstem is far from complete. The present study used a variety of sophisticated experimental techniques to convincingly reveal the identity of brainstem components that are part of the neural circuits involved in the mediation of opioid respiratory effects, together with defining potential synaptic underlying mechanisms. They focused on two regions of the brainstem, namely the Kolliker-Fuse and the preBötzinger Complex, and proposed a combination of three complementary processes at pre- and post-synaptic sites in both KF and preBötC regions to explain respiratory depression linked to opioid exposure. This study provides very important findings on the circuitry involved in opioid-induced respiratory depression, and the present results are of broad interest to the respiratory control research community, as well as medically relevant.

https://doi.org/10.7554/eLife.81119.sa0

Introduction

With the prevalence of opioid overdose on the rise (Wilson et al., 2020; Mattson et al., 2021), understanding the network mechanisms of opioid-induced respiratory depression is of particular importance. Opioids, due to activation of the mu opioid receptor (MOR) (Dahan et al., 2001), depress breathing by inhibiting interconnected respiratory nuclei in the pons and medulla (Bateman et al., 2021; Ramirez et al., 2021). Despite significant progress, detailed mechanisms by which this occurs remain elusive, especially for the dorsolateral pons. We sought to identify mechanistic insight concerning how opioids inhibit pontomedullary respiratory neurocircuitry that gives rise to opioid-induced respiratory depression.

Respiration is generated and controlled by an interconnected pontomedullary network in the brainstem (Del Negro et al., 2018). The Kölliker-Fuse (KF) nucleus and adjacent lateral parabrachial area (LPB) of the dorsolateral pons are critical to the neural control of breathing (Lumsden, 1923; Fung and St John, 1995; Dutschmann and Herbert, 2006; Smith et al., 2007). The KF/LPB is composed of a heterogeneous population of respiratory neurons that impact respiratory rate and pattern (Chamberlin and Saper, 1994; Navarrete-Opazo et al., 2020; Saunders and Levitt, 2020) via excitatory projections to respiratory nuclei in the ventrolateral medulla, including, but not limited to the Bötzinger complex (BötC), preBötzinger complex (preBötC), and rostral ventral respiratory group (rVRG) (Song et al., 2012; Yokota et al., 2015; Geerling et al., 2017; Yang et al., 2020). The preBötC generates inspiratory rhythm (Smith et al., 1991), which is relayed to inspiratory premotor neurons in the rVRG. The BötC contains mostly inhibitory neurons that fire during expiration and is a major source of inhibition within the network (Schreihofer et al., 1999; Ezure et al., 2003). The dynamic interplay between the KF/LPB and the BötC, preBötC, and rVRG is essential for optimized respiratory output (Dutschmann and Dick, 2012; Smith et al., 2007). Unfortunately, all of these respiratory nuclei express MORs, leading to inhibition of the control of breathing network via multiple potential sites and mechanisms (Gray et al., 1999; Lonergan et al., 2003; Montandon et al., 2011; Levitt et al., 2015; Cinelli et al., 2020).

Two respiratory nuclei considered critical for opioid-induced respiratory depression are the KF/LPB of the dorsolateral pons and the preBötC of the ventrolateral medulla (Bachmutsky et al., 2020; Varga et al., 2020). The dorsolateral pontine KF/LPB is considered a key contributor of opioid-induced respiratory depression because (1) deletion of MORs from the KF/LPB attenuates morphine-induced respiratory depression (Bachmutsky et al., 2020; Varga et al., 2020; Liu et al., 2021), (2) opioids injected into the KF/LPB reduce respiratory rate (Prkic et al., 2012; Levitt et al., 2015; Miller et al., 2017; Liu et al., 2021), (3) blockade of KF/LPB opioid receptors rescues fentanyl-induced apnea (Saunders and Levitt, 2020), and (4) chemogenetic inhibition of MOR-expressing LPB neurons induces respiratory depression (Liu et al., 2021). Yet, mechanisms by which the opioid inhibition of dorsolateral pontine neurons alter neurotransmission in the respiratory circuitry and causes suppression of breathing are unknown.

MORs inhibit neurotransmission by hyperpolarizing neurons through activation of somatodendritic GIRK channels and/or inhibiting presynaptic neurotransmitter release through inhibition of voltage-gated calcium channels (Jiang and North, 1992; Chahl, 1996; Zamponi and Snutch, 1998; Al-Hasani and Bruchas, 2011). In the preBötC, presynaptic MORs inhibit synaptic transmission (Ballanyi et al., 2010; Wei and Ramirez, 2019; Baertsch et al., 2021) and are expressed more abundantly than somatodendritic MORs (Lonergan et al., 2003). These presynaptic MORs in the preBötC are poised to play a major role in the mechanism of opioid suppression of breathing within the inspiratory rhythm-generating area, but the specific origins of MOR-expressing synaptic projections remain unknown. Here, we tested the hypothesis that they are coming from the dorsolateral pons.

Opioids hyperpolarize a subset of KF neurons (Levitt et al., 2015), whose neurochemical identity and possible projection targets are unknown. Glutamatergic KF neurons project to the ventrolateral medulla (Song et al., 2012; Yokota et al., 2015; Geerling et al., 2017) and, if inhibited by opioids—either by somatodendritic activation of GIRK channels and/or presynaptic inhibition of neurotransmitter release—could depress breathing. Therefore, we hypothesized that MOR-expressing KF neurons project to and form excitatory synapses onto respiratory controlling neurons in the ventrolateral medulla (i.e. the preBötC and rVRG), and that this excitatory synapse is inhibited by presynaptic MORs on KF terminals. The results show that this excitatory pontomedullary respiratory circuit is robustly inhibited by opioids by three different mechanisms, involving presynaptic and postsynaptic opioid receptors in the dorsolateral pons and the ventrolateral medulla, revealing convergent mechanisms by which opioids can depress breathing.

Results

Oprm1 expression in dorsolateral pontine neurons

To visualize MOR-expressing dorsolateral pontine neurons, Oprm1Cre/Cre mice (Baertsch et al., 2021; Liu et al., 2021) were crossed with tdTomato Cre-reporter mice to generate Rosa26LSL-tdT/+::Oprm1Cre/+ mice (hereby referred to as Oprm1-tdT mice) that express tdTomato in neurons that also express MORs at any point during development. MOR-expressing neurons and neurites were identified in the dorsolateral pons, specifically in the lateral parabrachial area and KF (n = 3; Figure 1A–D).

Dorsolateral pontine neurons express mu opioid receptors (MORs).

(A–D) Representative images of tdTomato expression, as an indicator of MOR expression, in coronal dorsolateral pontine slices from Oprm1-tdT mice (n = 3) across the rostral to caudal Kölliker-Fuse/lateral parabrachial area (KF/LPB) axis. Fluorescent tdTomato image is overlaid onto brightfield image to show landmarks. (E–H) Representative images of GFP expression, as an indicator of MOR expression, following injection of virus encoding Cre-dependent GFP into KF/LPB to label MOR+ neurons in adult Oprm1Cre/+ mice (n = 5). Left column are slice schematics corresponding to each row. The approximate levels caudal to bregma (in mm) are to the right of each schematic. The images correspond to the solid boxed area (A–D) or the dotted boxed area (E–H) of the slice schematic. The scale bar in (D) applies to images (A–D). The scale bar in (H) applies to images (E–H). PBel, external lateral subdivision of parabrachial; SCP, superior cerebellar peduncle.

To selectively label neurons that express MORs during adulthood, a virus encoding Cre-dependent GFP expression (AAV-DIO-GFP) was injected into the dorsolateral pons of Oprm1Cre/+ 2–4-month-old mice (n = 5). MOR-expressing neurons were again identified in the lateral parabrachial and KF areas (Figure 1E–H). Neuronal cell bodies were more apparent in these images since MOR-expressing afferents into the dorsolateral pons were not labeled by this approach. These results are consistent with previous studies showing that MORs are expressed in LPB (Huang et al., 2021; Liu et al., 2021) and KF (Levitt et al., 2015; Varga et al., 2020).

Oprm1+ KF neurons project to respiratory nuclei in the ventrolateral medulla

We hypothesized that Oprm1+ KF neurons project to respiratory controlling nuclei in the ventrolateral medulla, especially the preBötC and rVRG. To determine this, retrograde virus encoding Cre-dependent expression of GFP (retrograde AAV-hSyn-DIO-eGFP) was unilaterally injected into the preBötC or the rVRG of Oprm1Cre/+ mice (Figure 2). As a control, anterograde virus encoding mCherry (AAV2-hSyn-mCherry) was co-injected to mark the injection site. The intensity of mCherry expression was measured throughout the rostral-caudal axis of the ventrolateral medulla to quantify the extent of injection spread in accordance with medullary anatomical markers (Figure 2B and C). In addition, immunolabeling for the neurokinin 1 receptor (NK1R) was used as a marker of the preBötC (Gray et al., 1999; Montandon et al., 2011) and to identify the nucleus ambiguous (NA), which was especially useful for the compact section of the NA in the preBötC region (Figure 2B). Injection sites were categorized based on the location of peak mCherry expression intensity (Figure 2C and Figure 2—figure supplement 1).

Figure 2 with 2 supplements see all
Oprm1+ Kölliker-Fuse (KF) neurons and neurites retrogradely labeled from the preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG).

(A) Schematic illustrating the approach to retrogradely label Oprm1+ KF neurons and neurites projecting to the preBötC or rVRG. (B) Images of coronal slices from the medulla with a control injection of AAV2-hSyn-mCherry into the preBötC of an Oprm1Cre/+ mouse to mark the injection site. Immunohistochemistry for the neurokinin 1 receptor (NK1R) was used as a marker for the preBötC and the nucleus ambiguous (NA). (C) Quantification of normalized AAV2-hSyn-mCherry fluorescence intensity along the rostral to caudal axis in the ventrolateral medulla of preBötC (n = 5) and rVRG (n = 5). Anatomical level relative to Bregma is indicated on the x-axis. (D–I) Representative images of GFP expression, as an indicator of retrograde-labeled Oprm1-expressing neurons and neurites, following injections into the preBötC (D–F) or the rVRG (G–I) across three levels of the dorsolateral pons. The bregma level is indicated on the schematics to the left of each row. The scale bar in (I) applies to all images (D–I). Higher magnification images of bregma –4.84 are shown in Figure 2—figure supplement 2.

Figure 2—source data 1

Quantification of spread at the injection sites.

https://cdn.elifesciences.org/articles/81119/elife-81119-fig2-data1-v2.xlsx

Oprm1+ dorsolateral pontine neurons and neurites were retrogradely labeled from both preBötC and rVRG (Figure 2D–I). Interestingly, Oprm1+ projections to the preBötC (Figure 2D–F; n = 5) and the rVRG (Figure 2G–I; n = 5) were mostly localized to the rostral and mid-rostral KF, and nearly absent in the caudal KF and lateral parabrachial area (Figure 2F and I). The majority of the retrogradely labeled dorsolateral pontine neurons and neurites were ipsilateral to the injection site, with very few or no contralateral neurons or neurites expressing GFP. Injections in three mice were located rostrally from the preBötC with the peak of mCherry expression in the BötC (Figure 2—figure supplement 1). In contrast to preBötC and rVRG projections, qualitatively fewer Oprm1+ KF neurons projected to the BötC (Figure 2—figure supplement 1E–G). Higher magnification images of retrograde-labeled GFP-expressing Oprm1+ KF neurons are shown in Figure 2—figure supplement 2.

Presynaptic opioid receptors inhibit glutamate release from KF terminals onto excitatory medullary neurons

Given that KF neurons projecting to the ventrolateral medulla are glutamatergic (Geerling et al., 2017) and express MORs (Figure 2), we hypothesized that opioids inhibit glutamate release from KF terminals onto respiratory neurons in the ventrolateral medulla, particularly the preBötC and rVRG. To test this hypothesis, we unilaterally injected a virus encoding channelrhodopsin2 (AAV2-hSyn-hChR2(H134R)-EYFP-WPRE-PA) into the KF of vglut2Cre/LSL-tdT mice (Figure 3A and B). We made whole-cell voltage-clamp recordings from tdTomato-expressing, excitatory vglut2-expressing preBötC and rVRG neurons contained in acute brain slices (Figure 3C). Because we could not determine the respiratory-related firing pattern of the neurons we recorded from in this study, we chose to target vglut2-expressing neurons since (1) this contains the population of inspiratory rhythm-generating preBötC neurons (Wallén-Mackenzie et al., 2006; Gray et al., 2010; Cui et al., 2016) and inspiratory premotor rVRG neurons, (2) KF neurons project to excitatory, more so than inhibitory, preBötC neurons (Yang et al., 2020), and (3) deletion of MORs from vglut2 neurons eliminates opioid-induced depression of respiratory output in medullary slices (Sun et al., 2019; Bachmutsky et al., 2020). Optogenetic stimulation of KF terminals drove pharmacologically isolated excitatory postsynaptic currents (oEPSCs) in excitatory preBötC and rVRG neurons (Figure 3D and I) that were blocked by the AMPA-type glutamate receptor antagonist 6,7-dinitroquinoxaline-2,3-dione (DNQX; 10 μM; Figure 3—figure supplement 1A and B, n = 11). Additionally, KF synapses onto medullary respiratory neurons are monosynaptic because oEPSCs were eliminated by tetrodotoxin (TTX; 1 μM) yet restored by subsequent application of 4-aminopyridine (4AP; 100 μM) (Figure 3—figure supplement 1A and C; n = 7). Thus, KF neurons send monosynaptic, glutamatergic projections to excitatory ventrolateral medullary neurons.

Figure 3 with 1 supplement see all
Presynaptic opioid receptors inhibit glutamate release from Kölliker-Fuse (KF) terminals onto excitatory preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG) neurons.

(A) Schematic of approach to optogenetically stimulate KF terminals and drive optogenetically evoke excitatory postsynaptic currents (oEPSCs) in excitatory preBötC and rVRG neurons in an acute brain slice. (B) Representative image of ChR2-GFP expression in the KF (injection area) of vglut2-tdT mouse. (C) tdTomato-expressing, excitatory vglut2-expressing preBötC (or rVRG) neurons were identified in acute brain slices. (D) Recording of pairs of oEPSCs (5 ms stimulation, 50 ms inter-stimulus interval) from an excitatory preBötC neuron in an acute brain slice at baseline (black), during perfusion of Met-enkephalin (ME, 3 μM) (red), and after wash (gray). (E) ME decreased oEPSC amplitude in preBötC neurons (n = 13; **p=0.007, *p=0.013 by one-way ANOVA and Tukey’s post-test). (F) ME increased the paired-pulse ratio (P2/P1) in preBötC neurons (n = 11; *p=0.001 paired t-test). (G) ME (3 μM) induced outward currents in 8 of 12 preBötC neurons. OS, opioid-sensitive; NS, non-opioid-sensitive. (H) The amplitude of the outward current (I–ME, pA) in OS preBötC neurons. (I) Recording of pairs of oEPSCs (5 ms stimulation, 50 ms inter-stimulus interval) from an excitatory rVRG neuron in an acute brain slice at baseline (black), during perfusion of ME (3 μM) (red), and after wash (gray). (J) ME decreased oEPSC amplitude in rVRG neurons (n = 9; *p=0.027 by one-way ANOVA and Tukey’s post-test). (K) ME increased the paired-pulse ratio (P2/P1) in rVRG neurons (n = 9; *p=0.043 by paired t-test). (L) ME-mediated outward currents were observed in 7 of 8 rVRG neurons. (M) The amplitude of the outward current (I–ME, pA) in OS rVRG neurons. For all graphs, bar/line and error represent mean ± SEM. Individual data points are from individual neurons.

Figure 3—source data 1

Presynaptic opioid receptors inhibit glutamate release from Kölliker-Fuse (KF) terminals onto excitatory preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG) neurons.

https://cdn.elifesciences.org/articles/81119/elife-81119-fig3-data1-v2.xlsx

To determine whether opioids inhibit glutamate release from KF terminals onto medullary respiratory neurons, pairs of oEPSCs (50 ms inter-stimulus interval) were recorded from excitatory preBötC and rVRG neurons, and the endogenous opioid agonist [Met5]enkephalin (ME) was applied to the perfusion solution. ME (3 μM) decreased the oEPSC amplitude in preBötC neurons (Figure 3D and E; n = 13) and in rVRG neurons (Figure 3I and J; n = 9), which reversed when ME was washed from the slice. In addition, ME increased the paired-pulse ratio (PPR) in both preBötC (Figure 3F; n = 11) and rVRG neurons (Figure 3K; n = 9), indicating inhibition of glutamate release by presynaptic MORs. The proportion of opioid-sensitive KF terminals was surprisingly high, considering that oEPSCs were inhibited by ME by a threshold of at least 30% in nearly all preBötC neurons (11 of 13 neurons) and all rVRG neurons. Thus, presynaptic opioid receptors inhibit glutamate release from KF terminals onto a majority of excitatory preBötC and rVRG neurons (91% [20 of 22 neurons]).

We were also able to determine whether the excitatory preBötC or rVRG neuron that received opioid-sensitive glutamatergic synaptic input from the KF was itself hyperpolarized by opioids by monitoring the holding current. ME (3 μM) induced an outward current in 68% of preBötC neurons (8 of 12 neurons) (Figure 3G and H) and 88% of rVRG neurons (7 of 8 neurons) (Figure 3L and M). There was no difference in the amplitude of the ME-mediated current in preBötC and rVRG neurons (p=0.294; unpaired t-test). Thus, a majority of excitatory preBötC and rVRG neurons that receive opioid-sensitive glutamatergic synapses from KF neurons are themselves hyperpolarized by opioids, indicating both pre- and postsynaptic suppression of this excitatory synapse by opioids.

Opioids hyperpolarize medullary-projecting KF neurons

Opioids hyperpolarize a subpopulation (~60%) of KF neurons by activating G protein-coupled inwardly rectifying potassium (GIRK) channels (Levitt et al., 2015). Given that KF neurons that project to excitatory neurons in the ventrolateral medulla express functional MORs on presynaptic terminals at a higher percentage than expected (91% [20 of 22 neurons]; Figure 3), we wanted to determine whether KF neurons also express functional somatodendritic MORs leading to hyperpolarization in a projection-specific manner. We recorded from KF neurons retrogradely labeled with FluoSpheres (580/605) that were unilaterally injected into the preBötC or rVRG of wild-type mice (Figure 4A). FluoSpheres were chosen over viral retrograde tracers for these experiments because they are highly visible in acute brain slices and do not spread as far in the injection area (Figure 4A), genetically alter neurons, require fluorescent amplification, or take long to express (2 d vs. 4 wk). Furthermore, FluoSpheres will label KF neurons regardless of Oprm1 expression status, enabling us to determine the projection pattern of both Oprm1+ and Oprm1- neurons. Whole-cell voltage-clamp recordings were made from fluorescent KF neurons contained in acute brain slices (Figure 4B). The presence of an ME-mediated outward current identified KF neurons that express functional MORs and were opioid sensitive (OS) (Figure 4C) compared to neurons that lacked an ME-mediated outward current (non-sensitive [NS]) (Figure 4D). ME induced an outward current in 59% (13 of 22 neurons) of KF neurons that project to the preBötC (Figure 4E) and 67% (12 of 18 neurons) of KF neurons that project to the rVRG (Figure 4F). The average amplitude of the ME-mediated current was not different between KF neurons that project to preBötC (n = 13) or rVRG (n = 12) (p=0.8250; unpaired t-test). Thus, both opioid-sensitive and non-sensitive KF neurons project to preBötC and rVRG, with a proportion similar to the general population of KF neurons with unidentified projection targets (Levitt et al., 2015).

Figure 4 with 1 supplement see all
Opioids hyperpolarize Kölliker-Fuse (KF) neurons that project to the preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG).

(A) Schematic (left) of approach to retrogradely label KF neurons that project to the preBötC or rVRG with FluoSpheres in wild-type mice. Images (right) of FluoSpheres in the injection area (preBötC or rVRG). The scale bar applies to both injection images. (B) A KF neuron retrogradely labeled by FluoSpheres shown with IR-Dodt and epifluorescent (FL) illumination. (C, D) Whole-cell voltage-clamp recordings from opioid-sensitive (‘OS’) and non-opioid-sensitive (‘NS’) retrogradely labeled KF neurons. Met-enkephalin (ME) (1 µM) induced an outward current in the opioid-sensitive (OS) neuron (C), but not the non-opioid-sensitive (NS) neuron (D). (E, F) Quantification of the amplitude of the ME-mediated current (I-ME [pA]) in OS and NS KF neurons that project to the preBötC (E; n = 22; ***p=0.0005; unpaired t-test) or the rVRG (F; n = 18; ***p=0.0007; unpaired t-test). ME induced an outward current in 13 of 22 KF neurons that project to the preBötC and 12 of 18 KF neurons that project to the rVRG. Individual data points are from individual neurons in separate slices. Line and error are mean ± SEM.

Figure 4—source data 1

Opioid-mediated outward currents in Kölliker-Fuse (KF) neurons that project to the preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG).

https://cdn.elifesciences.org/articles/81119/elife-81119-fig4-data1-v2.xlsx

Given the potentially lesser degree of projections from Oprm1+ KF neurons to the BötC (Figure 2—figure supplement 1) and the ability to retrogradely label Oprm1-negative neurons with FluoSpheres, we also injected FluoSpheres into the BötC (n = 11) to test the hypothesis that Oprm1-negative KF neurons project to the BötC (Figure 4—figure supplement 1). We made whole-cell voltage-clamp recordings from fluorescent KF neurons and found that ME induced an outward current in only 36% (4 of 11 neurons) of KF neurons that project to the BötC (Figure 4—figure supplement 1C). Thus, a lower proportion of opioid-sensitive neurons project to BötC compared to preBötC and rVRG.

Distribution of Oprm1+ and Oprm1- dorsolateral pontine neurons projecting to the ventrolateral medulla

To further examine the distribution of Oprm1+ and Oprm1- dorsolateral pontine neurons projecting to the ventrolateral medulla, retrograde AAV-hSyn-DIO-eGFP and retrograde AAV-hSyn-mCherry were unilaterally injected into the preBötC and rVRG of Oprm1Cre/+ mice (Figure 5A). Using this approach, projection neurons that express Oprm1 will express GFP and mCherry, whereas projection neurons that do not express Oprm1 will only express mCherry (Figure 5B). The number of mCherry and/or GFP-expressing neurons was evaluated in rostral (~bregma level –4.84 mm), mid-rostral (~bregma level –4.96 mm), and caudal (~bregma level –5.20 mm) sections of the dorsolateral pons (n = 4 mice, three slices per region per mouse). There were significantly more retrograde-labeled neurons in rostral and mid-rostral slices, regardless of Oprm1 expression status (Figure 5D). Consistent with previous observations (Figure 2), retrograde-labeled Oprm1+ neurons were mostly localized to the rostral and mid-rostral slices, and not in caudal slices or lateral parabrachial area (Figure 5C and E and Figure 5—figure supplement 1). The percentage of retrograde-labeled neurons that were Oprm1+ (co-labeled with mCherry and GFP) in rostral slices (56%) and mid-rostral slices (47%) was higher than in caudal slices (15%) (Figure 5C). Taken together, Oprm1+ and Oprm1- KF neurons that project to respiratory nuclei in the ventrolateral medulla are distributed to the rostral and mid-rostral regions of the KF of mice.

Figure 5 with 1 supplement see all
Oprm1+ and Oprm1- dorsolateral pontine neurons project to the ventrolateral medulla.

(A) Schematic of approach injecting retrograde virus encoding Cre-dependent GFP expression and a retrograde virus encoding mCherry expression into the ventrolateral medulla of Oprm1Cre/+ mice to label Oprm1+ and Oprm1- dorsolateral pontine neurons that project to these respiratory nuclei. (B) Representative images of mCherry expression (retrogradely labels neurons regardless of Oprm1 expression) and GFP expression (retrogradely labels Oprm1+ neurons) in a rostral dorsolateral pontine slice (bregma –4.84 mm). (C) Summary of percentage of retrograde-labeled neurons that were Oprm1+ (co-labeled with mCherry and GFP) in rostral (Rstrl, bregma –4.84 mm), mid-rostral (Mid, bregma –4.96 mm), and caudal (Cdl, bregma –5.2 mm) slices. (D, E) Summary of the average number of mCherry-expressing (D) or GFP-expressing MOR+ (E) dorsolateral pontine neurons per slice in rostral, mid-rostral, and caudal slices. Bar and error are mean ± SEM. Individual data points are from individual mice. N = 4 mice, three slices per region per mouse. *p<0.05, **p<0.01, ns = p>0.05 by one-way ANOVA and Tukey’s post-test.

Figure 5—source data 1

Oprm1+ and Oprm1- dorsolateral pontine neurons project to the ventrolateral medulla.

https://cdn.elifesciences.org/articles/81119/elife-81119-fig5-data1-v2.xlsx

Oprm1+, medullary-projecting KF neurons express FoxP2, but not CGRP

Rostral glutamatergic KF neurons express FoxP2 (Forkhead box protein P2) (Geerling et al., 2017; Karthik et al., 2022), whereas MOR-expressing glutamatergic neurons in the external lateral parabrachial subnucleus that project to the forebrain express Calca, a gene that encodes the neuropeptide calcitonin gene-related peptide (CGRP) (Huang et al., 2021). Considering this, we performed immunohistochemistry for FoxP2 and CGRP on Oprm1+ KF neurons projecting to the ventrolateral medulla. Oprm1+, medullary-projecting KF neurons expressed FoxP2 (n = 3; Figure 6), consistent with the population of glutamatergic FoxP2 and Lmx1b neurons in the rostral KF (Karthik et al., 2022). These are a separate population from FoxP2-expressing neurons located more dorsally and caudally in the inner portion of the external lateral parabrachial area and those activated by sodium deprivation (Geerling et al., 2011; Karthik et al., 2022). FoxP2 expression also overlapped with a smaller population of Oprm1+ medullary-projecting neurons in the caudal KF, which contains GABAergic neurons (Figure 6—figure supplement 1; Geerling et al., 2017; Karthik et al., 2022). FoxP2 was not detected in the outer portion of the external lateral parabrachial subnucleus (Figure 6—figure supplement 1), consistent with previous findings (Geerling et al., 2011; Karthik et al., 2022).

Figure 6 with 1 supplement see all
Oprm1+, medullary-projecting Kölliker-Fuse (KF) neurons express Forkhead box protein P2 (FoxP2).

Oprm1+ neurons that project to the ventrolateral medulla were retrogradely labeled by injection of retrograde AAV-DIO-GFP into Oprm1Cre/+ mice. Immunohistochemistry was used to label FoxP2. (A, B) In rostral slices (bregma –4.84), FoxP2 is expressed in Oprm1+ KF neurons that project to the ventrolateral medulla. Schematic (A) depicts the approximate bregma level and imaging area (dotted boxed area). The scale bar applies to all images. SCP, superior cerebellar peduncle.

Oprm1+, medullary-projecting KF neurons did not express CGRP (n = 3; Figure 7). Although CGRP expression was absent from the rostral KF and medullary-projecting Oprm1+ neurons and neurites, it was robust in external lateral parabrachial neurons and their axon fiber projections (Figure 7C and D).

Oprm1+, medullary-projecting Kölliker-Fuse (KF) neurons do not express calcitonin gene-related peptide (CGRP).

Oprm1+ neurons that project to the ventrolateral medulla were retrogradely labeled by injection of retrograde AAV-DIO-GFP into Oprm1Cre/+ mice. Immunohistochemistry was used to label CGRP. (A, B) CGRP is absent from rostral KF and Oprm1+ KF neurons that project to the ventrolateral medulla (Oprm1+). (C, D) CGRP marks lateral parabrachial area (LPB) neurons and their axon fiber projections, but is absent from retrograde-labeled Oprm1+ axon fiber projections in mid-rostral (C) and caudal (D) slices. The approximate bregma levels are to the right of each schematic. The images correspond to the dotted boxed area (row A) or the solid boxed area (rows B–D) of the slice schematic. The images in (A) are zoomed into the dotted boxed area of the image in (B). The scale bar in (A) applies to the images in row (A). The scale bar in (D) applies to images in rows (B–D). SCP, superior cerebellar peduncle; MCP, medial cerebellar peduncle.

Discussion

Opioid suppression of breathing could occur via multiple mechanisms and at multiple sites in the pontomedullary respiratory network. Here, we show that opioids inhibit an excitatory pontomedullary respiratory circuit via three mechanisms: (1) postsynaptic MOR-mediated hyperpolarization of KF neurons that project to the ventrolateral medulla, (2) presynaptic MOR-mediated inhibition of glutamate release from KF terminals onto excitatory preBötC and rVRG neurons, and (3) postsynaptic MOR-mediated hyperpolarization of the preBötC and rVRG neurons that receive pontine glutamatergic input (Figure 8). These mechanisms converge on a projection-specific opioid-sensitive circuit, whereby MOR-expressing excitatory KF neurons synapse onto MOR-expressing excitatory preBötC and rVRG neurons at a proportion that is higher than predicted based on MOR expression in either of these populations alone (Bachmutsky et al., 2020; Kallurkar et al., 2022; Levitt et al., 2015). We targeted the excitatory vglut2-expressing neurons in the ventrolateral medulla because they contain the populations of inspiratory rhythm-generating preBötC neurons (Wallén-Mackenzie et al., 2006; Gray et al., 2010; Cui et al., 2016) and inspiratory premotor rVRG neurons, and MOR deletion from vglut2 neurons prevents opioid-induced respiratory depression in medullary slices (Sun et al., 2019; Bachmutsky et al., 2020). Opioid inhibition of excitatory drive from KF onto these respiratory neuron populations is important for rhythm generation (preBötC) and respiratory pattern formation (rVRG). Thus, there are convergent mechanisms of opioid-induced respiratory suppression, including both presynaptic and postsynaptic opioid receptors in the dorsolateral pons and the ventrolateral medulla, resulting in distributed effects of opioids on the pontomedullary respiratory network.

Summary schematic of mu opioid receptor (MOR) regulation of excitatory pontomedullary circuitry.

Kölliker-Fuse (KF): Somatodendritic MORs hyperpolarize KF neurons that project to the ventrolateral medulla. Ventrolateral medulla: presynaptic MORs inhibit glutamate release from KF axon terminals onto glutamatergic preBötzinger complex (preBötC) and rostral ventral respiratory group (rVRG )neurons. Somatodendritic MORs hyperpolarize glutamatergic preBötC and rVRG neurons that receive KF input. Glutamatergic neurons are in green.

Opioid effects distributed throughout the pontomedullary respiratory network

The mechanistic insights shown here are parsimonious with previous studies examining the role of MORs in the dorsolateral pons (Prkic et al., 2012; Levitt et al., 2015; Miller et al., 2017; Bachmutsky et al., 2020; Saunders and Levitt, 2020; Varga et al., 2020; Liu et al., 2021) and the preBötC (Gray et al., 1999; Sun et al., 2019; Bachmutsky et al., 2020; Varga et al., 2020) in opioid-induced respiratory depression. Genetic deletion or pharmacological blockade of different subsets of pre- and postsynaptic MORs in these areas mostly resulted in partial attenuation of opioid-induced respiratory rate suppression, presumably due to redundancy from the subset(s) of MORs in this pontomedullary circuit that were not deleted or blocked. Furthermore, additional MORs outside of the dorsolateral pontine and preBötC circuit likely contribute to respiratory suppression since deletion of MORs from both dorsolateral pons and preBötC did not eliminate morphine-induced respiratory suppression (Bachmutsky et al., 2020).

Often overlooked in the context of opioids, the rVRG contains abundant opioid receptors (Lonergan et al., 2003) and application of an opioid agonist into the rVRG suppresses rate and amplitude of phrenic nerve bursting (Lonergan et al., 2003; Cinelli et al., 2020). Here, we showed that MOR-expressing KF neurons densely project to the rVRG (Figures 2 and 4) and form glutamatergic synapses onto excitatory rVRG neurons (Figure 3). Presynaptic opioid receptors inhibit glutamate release from KF terminals synapsing onto rVRG neurons (Figure 3), and the excitatory rVRG neurons that receive glutamatergic input from the dorsolateral pons are hyperpolarized by postsynaptic opioid receptors (Figure 3). The impact of this highly opioid-sensitive projection on respiration warrants further investigation. Other respiratory-related areas in the medulla, such as the retrotrapezoid nucleus and the nucleus of the solitary tract (NTS), that receive Oprm1+ pontine input (Liu et al., 2022) could also be involved, but functional connectivity and impact remains to be determined. Another potential contributor to OIRD are the caudal medullary raphe nuclei since antagonism of opioid receptors in the dorsolateral pons, ventrolateral medulla, and caudal medullary raphe was able to eliminate remifentanil-induced respiratory depression (Palkovic et al., 2022).

Opioids inhibit excitatory pontomedullary circuitry

Unexpectedly, KF neurons were more likely inhibited by presynaptic vs. somatodendritic MORs. The vast majority of KF terminals expressed presynaptic MORs since opioids inhibited glutamate release onto 91% of preBötC and rVRG neurons (Figure 3). In contrast, postsynaptic (somatodendritic) MOR-mediated outward currents were only observed in about two-thirds of medullary-projecting KF neurons (Figure 4), which matches prior studies without projection identification (Levitt et al., 2015; Varga et al., 2020). There are multiple possible reasons for this apparent heterogeneity. First, KF neurons may express MORs more abundantly on terminals than in the somatodendritic region. Second, KF neurons that did not have outward currents and were deemed not sensitive to opioids may express MORs, but lack GIRK channels, the functional readout we used to assess opioid sensitivity. MORs on these neurons could instead couple to other effectors, such as voltage-gated calcium channels (Ramirez et al., 2021). However, this seems unlikely since the percentage of retrograde-labeled neurons that were Oprm1+ (56% in rostral and 47% in mid-rostral slices; Figure 5) nearly matched the percentages of functionally identified opioid-sensitive KF neurons (59% of preBötC-projecting and 67% of rVRG-projecting neurons; Figure 4).

The last and most interesting possibility is that opioid-sensitive glutamatergic KF neurons preferentially synapse onto excitatory medullary neurons, while non-opioidergic KF neurons might preferentially synapse onto non-excitatory (i.e. inhibitory) medullary neurons. This hypothesis is consistent with anatomical-tracing studies showing that KF neurons project to excitatory and, to a lesser extent, inhibitory preBötC neurons (Yang et al., 2020), and could be tested by recording from labeled inhibitory neurons in the ventrolateral medulla. Inhibitory transmission in the medullary rhythm generator influences respiratory rate in the case of phasic inhibition or causes sustained apnea in the case of prolonged inhibition (Baertsch et al., 2018; Cregg et al., 2017; Sherman et al., 2015). We have recently found that inspiratory dorsolateral pontine neurons are silenced by fentanyl, whereas expiratory neurons are not (Saunders et al., 2022). An intriguing possibility is that opioid-insensitive pontine neurons, which have continued activity during opioid exposure, send prolonged input to inhibitory neurons in the ventrolateral medulla to promote apnea, perhaps using pathways overlapping those involved in apneas evoked by excitation of certain parts of the KF area (Saunders and Levitt, 2020; Dutschmann and Dick, 2012; Dutschmann and Herbert, 2006). This could include opioid-insensitive KF neurons that project to inhibitory neurons in the BötC since a higher proportion of opioid-insensitive pontine neurons projected to the BötC (Figure 2 and Figure 4—figure supplement 1). Inhibitory input could also come from the NTS, which contains abundant MOR-expressing afferents and non-MOR-expressing neurons that are activated by disinhibition during opioid exposure (Glatzer et al., 2007; Maletz et al., 2022).

Dorsolateral pontine subpopulations

The dorsolateral pons includes the lateral parabrachial area and the KF, both of which have been implicated in opioid-induced respiratory depression (Levitt et al., 2015; Prkic et al., 2012; Varga et al., 2020; Liu et al., 2021). Although effects of MORs in the lateral parabrachial and KF areas appear similar, mechanisms likely differ since the neuronal populations have different projection patterns (Geerling et al., 2017; Huang et al., 2021; Liu et al., 2022) and are involved in different behaviors besides breathing, especially the lateral parabrachial area, which has many different subpopulations (Campos et al., 2018; Chen et al., 2018; Liu et al., 2022; Karthik et al., 2022). In addition, the anatomical distinction between KF and lateral parabrachial area is not clear cut, though recent descriptions of transcription factor and neuropeptide/receptor expression in the dorsolateral pons provide opportunity to improve this (Karthik et al., 2022; Pauli et al., 2022).

The most well-defined area in the dorsolateral pons is the external lateral parabrachial subnucleus, which expresses Lmx1b and CGRP, but not FoxP2 (Karthik et al., 2022; Huang et al., 2021). CGRP-expressing external lateral parabrachial neurons project primarily to the forebrain (Huang et al., 2021) and are involved in pain processing, feeding, and CO2-induced arousal (Campos et al., 2018; Chen et al., 2018; Kaur et al., 2017). Although MORs are highly co-expressed with CGRP in these neurons (Huang et al., 2021), we did not observe opioid-sensitive or Oprm1+retrograde-labeled neurons in the external lateral parabrachial area. We also did not observe a ‘shell’ pattern of retrograde-labeled Oprm1+ neurons surrounding the external lateral parabrachial area, in contrast with Liu et al., 2022, which could be due to slight differences in injection location, the fluorescent probe, and/or sensitivity of the experimental design. Rather, electrophysiologically or histologically identified opioid-sensitive/Oprm1+ neurons that project to the ventrolateral medulla were found rostrally and ventrally in the area overlapping FoxP2 expression in the KF. Thus, at least two distinct subpopulations of Oprm1+ dorsolateral pontine neurons exist that can be distinguished based on CGRP expression and projection pattern: forebrain-projecting CGRP-expressing neurons and medullary-projecting neurons that do not express CGRP. Both populations are involved in pain and breathing due, at least in part, to reciprocal excitatory synaptic connections (Liu et al., 2022). Although medullary-projecting Oprm1+ pontine neurons did not express CGRP (Figure 7), they can still be involved in pain processing, just not to the same extent as forebrain-projecting Oprm1/CGRP+ pontine neurons (Liu et al., 2022).

Both populations of Oprm1+ dorsolateral pontine neurons are also likely involved in opioid-induced respiratory depression. MORs in glutamatergic medullary-projecting rostral KF neurons could reduce respiratory rate by decreasing excitatory input to the preBötC and rVRG (Figure 3), leading to a distributed blunting effect on inspiration-generating processes within the ventrolateral medulla. In contrast, MORs in forebrain-projecting pontine neurons could reduce respiratory rate through intra-pontine excitatory connections with medullary-projecting MOR+ pontine neurons (Liu et al., 2022) or through reduced excitatory input to forebrain areas involved in arousal (Kaur et al., 2017), which may be especially important in sleep-dependent effects of opioids on breathing (Montandon and Horner, 2019).

PreBötC complex mechanisms

Significant attention has been given to the mechanisms of opioid suppression of inspiratory rhythm generation in the preBötC (Sun et al., 2019; Bachmutsky et al., 2020; Baertsch et al., 2021). Presynaptic opioid receptors in the preBötC inhibit synaptic transmission and have been postulated to disrupt preBötC neuron bursting (Ballanyi et al., 2010; Wei and Ramirez, 2019; Baertsch et al., 2021) by inhibition of excitatory neurotransmission that is dominant during bursts (Ashhad and Feldman, 2020), but the projection-specific location(s) of these presynaptic MORs is unknown. Our study has revealed a projection-specific presence of presynaptic MORs on glutamatergic terminals from dorsolateral pontine inputs to the preBötC. Although other MOR-expressing glutamatergic inputs are also likely contributors, including collaterals within the preBötC (Rekling et al., 2000), the role of these specific pontine inputs on opioid inhibition of respiratory rhythm generation is worthy of further investigation.

Only a subpopulation of preBötC neurons contain MORs (Bachmutsky et al., 2020; Baertsch et al., 2021; Kallurkar et al., 2022). The population of MOR-expressing preBötC neurons is heterogeneous, including nearly equal numbers of glutamatergic, GABAergic, and glycinergic neurons (Bachmutsky et al., 2020), type 1 and type 2 Dbx1-expressing inspiratory neurons (Kallurkar et al., 2022), and pre-inspiratory, inspiratory, expiratory, and tonic neurons (Baertsch et al., 2021). We found that MOR-expressing dorsolateral pontine glutamatergic inputs seem to preferentially synapse onto MOR-expressing excitatory preBötC neurons since 68% of preBötC neurons (8 of 12 neurons) that received glutamatergic input from the dorsolateral pons were hyperpolarized by opioid (Figure 3G). This percentage is higher than even the highest estimate of MOR-expressing preBötC neurons (Baertsch et al., 2021), suggesting dorsolateral pontine neurons preferentially target MOR-expressing glutamatergic preBötC neurons, which are important mediators of inspiratory rhythm generation and opioid-induced respiratory depression in medullary slices (Sun et al., 2019; Bachmutsky et al., 2020).

Sensitivity and regulation of presynaptic and postsynaptic opioid receptors

Presynaptic and postsynaptic MORs couple to distinct effectors and are regulated differently, which can lead to differences in sensitivity that change with prolonged opioid exposure (Coutens and Ingram, 2023). For instance, postsynaptic, but not presynaptic, opioid receptors couple to GIRK channels (Lüscher et al., 1997) through binding of up to four Gβγ subunits directly to the channel (Whorton and MacKinnon, 2013). In contrast, presynaptic opioid receptors inhibit neurotransmitter release through inhibition of VGCCs (Heinke et al., 2011) or direct inhibition of vesicle release machinery (Blackmer et al., 2001; Gerachshenko et al., 2005). Coupling to these presynaptic effectors may be more sensitive since VGCCs can be inhibited by a single Gβγ subunit (Zamponi and Snutch, 1998) and vesicular release is steeply calcium dependent (Katz and Miledi, 1967). Consistent with this, presynaptic opioid receptor responses have higher sensitivity than postsynaptic responses when directly compared (Pennock and Hentges, 2011). Prolonged exposure to high doses of opioids can exacerbate differences in sensitivity since postsynaptic receptors desensitize more readily than presynaptic receptors (Blanchet and Lüscher, 2002; Fyfe et al., 2010; Lowe and Bailey, 2015; Pennock et al., 2012; Rhim et al., 1993). Thus, the responses of presynaptic receptors may predominate, especially after prolonged opioid exposure, for reasons related to receptor reserve, coupling to effectors and/or receptor regulation. The relative sensitivity of presynaptic and postsynaptic receptors in the pontomedullary circuit identified here will be important to determine, especially since postsynaptic opioid receptors on KF neurons are resistant to desensitization (Levitt and Williams, 2018), suggesting unique receptor regulation in these neurons.

In conclusion, our results show that opioids inhibit an excitatory pontomedullary respiratory circuit by three distinct mechanisms—somatodendritic MORs on dorsolateral pontine and ventrolateral medullary neurons and presynaptic MORs on glutamatergic dorsolateral pontine axon terminals in the ventrolateral medulla—all of which could influence distributed network function and contribute to the profound effects of opioids on breathing.

Methods

Animals

All experiments were approved by the Institutional Animal Care and Use Committee at the University of Florida (protocol #09515) and were in agreement with the National Institutes of Health ‘Guide for the Care and Use of Laboratory Animals.’ Homozygous Oprm1Cre/Cre mice (Liu et al., 2021) (Jackson Labs Stock #035574, obtained from Dr. Richard Palmiter, University of Washington) were crossed with homozygous Ai9-tdTomato Cre-reporter mice (Rosa26LSL-tdT/LSL-tdT) (Jackson Labs Stock #007909) to generate Oprm1-tdT mice. Homozygous vglut2-ires-Cre mice (Jackson Labs Stock #028863) were crossed with homozygous Ai9-tdTomato Cre-reporter mice (Jackson Labs Stock #007909) to generate vglut2-tdT mice. Oprm1Cre/+, Oprm1-tdT, vglut2-tdT, and wild-type C57BL/6J mice (male and female, 2–4 months old, weights commensurate with age and sex of normally developing C57BL/6J mice) were used for all experiments (Table 1). Mice were bred and maintained at the University of Florida animal facility. Mice were grouphoused with littermates in standard sized plastic cages and kept on a 12 hr light–dark cycle, with water and food available ad libitum.

Table 1
Mice used in this study.
StrainReferenceSource informationKey gene
Oprm1-creLiu et al., 2021.Jax 035574
https://www.jax.org/strain/035574
Dr. Richard Palmiter (University of Washington)
Cre recombinase expressed in neurons with mu-opioid receptors
Vglut2-creVong et al., 2011Jax 028863
https://www.jax.org/strain/028863
Cre recombinase expressed in excitatory glutamatergic neurons
Ai9, tdTomato Cre-reporterMadisen et al., 2010Jax 007909
https://www.jax.org/strain/007909
LoxP-flanked STOP cassette preceding transcription of CAG promoter-driven red fluorescent protein variant (tdTomato) inserted into the Gt(ROSA)26Sor locus
C57BL/6J (wild-type)Simon et al., 2013Jax 000664
https://www.jax.org/strain/000664

Stereotaxic injections

Mice (1–4 months old) were anesthetized with isoflurane (2–4% in 100% oxygen; Zoetis, Parsippany-Troy Hills, NJ) and placed in a stereotaxic alignment system (Kopf Instruments model 1900, Tujunga, CA). The dorsal skull was exposed and leveled horizontally in preparation for a small, unilateral craniotomy targeting either the KF (y = –5 mm, x = ±1.7 mm, z = - 3.9 mm from bregma), BötC (y = –6.6 mm and x = ±1.3 mm from bregma, z = - 5.625 mm), preBötC (y = –6.9 mm and x = ±1.3 mm from bregma, z = –5.625 mm), or rVRG (y = –7.2 mm and x = ±1.3 mm from bregma, z = –5.625 mm). Virus (undiluted) or FluoSpheres (580/605, 0.04 µm, diluted to 20% in saline, Invitrogen) were loaded into freshly pulled glass micropipettes and injected using a Nanoject III pressure injector (Drummond Scientific Company, Broomall, PA) at a rate of 10 nl every 20 s (100–200 nl total) (Table 2). Following the injection, the pipette was left in place for 10 min and slowly retracted. The wound was closed using Vetbond tissue adhesive (3M Animal Care Products, St Paul, MN). Mice received meloxicam (5 mg kg–1 in saline, s.c.) and were placed in a warmed recovery chamber until they were ambulating normally. Mice were used either 2–6 d (FluoSpheres) or 4–5 wk (virus) later for electrophysiology, microscopy, or immunohistochemistry.

Table 2
Key resources.
InjectateStrain usedInjection targetFigureSource Information
FluoSpheres
580/605, diameter: 0.04 µm
C57BL/6JBötC, preBötC, or rVRGFigure 4 and Figure 4—figure supplement 1Invitrogen
Retrograde AAV-hSyn-DIO-EGFPOprm1Cre/+BötC, preBötC, or rVRGFigure 2 and 
Figure 2—figure supplement 1
Addgene
AAV2-hSyn-mCherryOprm1Cre/+BötC, preBötC, or rVRGFigure 2 and 
Figure 2—figure supplement 1
UNC Vector Core
Retrograde AAV-hSyn-mCherryOprm1Cre/+BötC, preBötC, and rVRGFigure 5Addgene
AAV2-hSyn-DIO-EGFPOprm1Cre/+KF/PBFigure 1E–HAddgene
AAV2-hSyn-hChR2(H134R)-EYFP-WPRE-PAvglut2-tdTKF/PBFigure 3UNC Vector Core
  1. PreBötC, preBötzinger complex; rVRG, rostral ventral respiratory group; KF, Kölliker-Fuse; PB, parabrachial area.

For retrograde labeling in Oprm1Cre/+ mice, a 1:1 mixture (100 nl total) of either retrograde AAV-hSyn-DIO-eGFP (Addgene) and AAV2-hSyn-mCherry (UNC vector core) (Figures 2, 6, and 7, Figure 2—figure supplement 1) or retrograde AAV-hSyn-DIO-eGFP (Addgene) and retrograde AAV-hSyn-mCherry (Addgene) (Figure 5) was injected into the BötC, preBötC, and/or the rVRG. For labeling Oprm1+ pontine neurons, AAV2-hSyn-DIO-EGFP (Addgene; 100 nl) (Figure 1E–H) was injected into the dorsolateral pons of Oprm1Cre/+ mice. Vglut2-tdT mice received AAV2-hSyn-hChR2(H134R)-EYFP-WPRE-PA (Addgene; 100 nl) injections targeting the KF (Figure 3). Lastly, FluoSpheres (580/605, diameter: 0.04 µm, 20% in saline; 100 nl) were unilaterally injected into the BötC (Figure 4—figure supplement 1), preBötC or rVRG of wild-type C57BL/6J mice (Figure 4).

The correct placement of injections into the either the KF, BötC, preBötC, or rVRG was verified by anatomical landmarks, immunohistochemistry, and fluorescence in free-floating coronal brain slices (40–100 µm) using a MultiZoom microscope (Nikon AZ100). The BötC, preBötC ,and rVRG are located bilaterally in a rostro-caudal column in the ventrolateral medulla, just ventral to the nucleus ambiguous. The BötC, preBötC, and rVRG can be distinguished using the inferior olives, nucleus ambiguous, and nucleus tractus solitarius as medullary landmarks (Franklin and Paxinos, 2008; Varga et al., 2020). The KF is located bilaterally in the dorsolateral pons, just ventrolateral to the tip of the superior cerebellar peduncle and medial of the middle cerebellar peduncle (Varga et al., 2020; Karthik et al., 2022).

Brain slice electrophysiology

Brain slice electrophysiology recordings were performed from KF neurons in acute brain slices from wild-type C57BL/6J mice (2–4 months old) or from vglut2-expressing preBötC and rVRG neurons in acute brain slices from vglut2-tdT mice (2–4 months old) injected with AAV2-hSyn-hChR2(H134R)-EYFP-WPRE-PA into the KF. Mice were anesthetized with isoflurane, decapitated, and the brain was removed and mounted in a vibratome chamber (VT 1200S, Leica Biosystems, Buffalo Grove, IL). Brain slices (230 µm) containing either the KF, BötC, preBötC, or rVRG (identified based on anatomical landmarks and coordinates from Franklin and Paxinos, 2008) were prepared in warmed artificial cerebrospinal fluid (aCSF) that contained the following (in mM): 126 NaCl, 2.5 KCl, 1.2 MgCl2, 2.4 CaCl2, 1.2 NaH2PO4, 11 d-glucose, and 21.4 NaHCO3 (equilibrated with 95% O2–5% CO2). Slices were stored at 32°C in glass vials with equilibrated aCSF. MK801 (10 µM) was added to the cutting and initial incubation solution (at least 30 min) to block NMDA receptor-mediated excitotoxicity. Brain slices were transferred to a recording chamber and perfused with 34°C aCSF (Warner Instruments, Hamden, CT) at a rate of 1.5–3 ml min−1.

Cells were visualized using an upright microscope (Nikon FN1) equipped with custom-built LED-based IR-Dodt gradient contrast illumination and DAGE-MTI IR1000 camera. Cells containing FluoSpheres (580/605) or tdTomato were identified using LED epifluorescence illumination and a Texas Red filter cube (ex 559 nm/ em 630 nm). Whole-cell recordings were made using a Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA) in voltage-clamp mode (Vhold = −60 mV). Glass recording pipettes (1.5–3 MΩ) were filled with internal solution that contained (in mM) 115 potassium methanesulfonate, 20 NaCl, 1.5 MgCl2, 5 HEPES(K), 2 BAPTA, 1–2 Mg-ATP, 0.2 Na-GTP, adjusted to pH 7.35 and 275–285 mOsM. The liquid junction potential (10 mV) was not corrected. Data were low-pass filtered at 10 kHz and collected at 20 kHz with pCLAMP 10.7 (Molecular Devices), or collected at 400 Hz with PowerLab (LabChart version 5.4.2; AD Instruments, Colorado Springs, CO). Series resistance was monitored without compensation and remained <15 MΩ for inclusion. For optogenetic experiments, ChR2-expressing KF terminals were stimulated using 470 nm LED illumination (5 ms duration) through a ×40 objective to optogenetically evoke excitatory postsynaptic currents (oEPSC) in preBötC and rVRG neurons. A pair of optical stimuli (5 ms pulse, 50 ms interval) was delivered every 20 s. Blockers of glycine (strychnine, 1 µM) and GABA-A (picrotoxin, 100 µM) receptors were added to the aCSF to isolate excitatory neurotransmission. Peak amplitudes were determined in Clampfit 10.7 (Molecular Devices), and paired-pulse ratios (peak 2/peak 1), were determined in Microsoft Excel. All drugs were applied by bath perfusion. Bestatin (10 µM) and thiorphan (1 µM) were included with ME to prevent degradation.

Immunohistochemistry and microscopy

Mice (2–4 months old) were anesthetized with isoflurane and transcardially perfused with phosphate-buffered saline (PBS) followed by 10% formalin. The brains were removed and stored at 4°C in cryoprotectant (30% sucrose in 10% formalin). A vibratome (VT 1200S, Leica Biosystems) was used to prepare free-floating coronal brain slices (40–100 µm) for microscopy or immunohistochemistry.

Free-floating slices were stained for forkhead box P2 (FoxP2), calcitonin gene-related peptide (CGRP), or neurokinin 1 receptor (NK1R) (Table 3). Slices were washed in diluting buffer (TBS with 2% bovine serum albumin, 0.4% Triton X-100, and 1% filtered normal goat serum) for 30 min, blocked in TBS and 20% normal donkey serum for 30 min, and incubated in primary antibody for 24 hr at 4°C. Primary antibodies included sheep polyclonal anti-FoxP2 (Cat# AF5647; R&D Systems, Minneapolis, MN; 1:1000 in diluting buffer), rabbit polyclonal anti-CGRP (Cat# T-4032; Peninsula, San Carlos, CA, 1:1000 in diluting buffer), and rabbit polyclonal anti-NK1R (Cat# S8305; Sigma-Aldrich; 1:1000 in diluting buffer). Slices were washed in diluting buffer and then incubated in secondary antibody (goat anti-rabbit 647 [Cat# A32733; Thermo Fisher Scientific, Waltham, MA] or donkey anti-sheep 647 [Cat# A21448; Thermo Fisher Scientific; 1:500]) in diluting buffer. Finally, slices were rinsed with TBS and ddH20 and mounted onto glass slides with Fluoromount-G DAPI (Thermo Fisher Scientific). A confocal laser scanning microscope (Nikon A1R) with a ×10 objective (N.A. 0.3) or a multizoom microscope (Nikon AZ100) with a ×1 objective (N.A. 0.1) were used to image sections. All images were processed in Fiji (Schindelin et al., 2012).

Table 3
Antibodies used in this study.
AntigenImmunogen descriptionSource, host species, RRIDConcentration
Forkhead box P2 (FoxP2)Targets human and mouse FoxP2R&D Systems, sheep polyclonal, Cat# AF5647, RRID:AB_21071331:1000
Calcitonin gene-related peptide (CGRP)Targets alpha-CGRP in canine, mouse, and ratPeninsula, rabbit polyclonal, Cat# T-4032,
RRID:AB_518147
1:1000
Neurokinin 1 receptor (NK1R)Targets C-terminal of NK1R in mouse, guinea pig, and humanSigma-Aldrich, rabbit polyclonal, Cat# S8305
RRID:AB_261562
1:1000

To determine the spread and intensity of mCherry expression in the BötC, preBötC, and rVRG, serial coronal brain slices (50 µm) were collected and every slice containing mCherry expression was imaged in sequential order with a multizoom microscope (Nikon AZ100) at 500 ms exposure. Mean fluorescence intensity was determined for a region of interest drawn ventral to the NA to encompass the 7N/pFRG, BötC, preBötC, or rVRG in sequential slices. Mean intensity data were background subtracted and normalized to the peak intensity per injection. Bregma level was assigned using anatomical landmarks, including the inferior olives, nucleus ambiguus, and nucleus tractus solitarius (Franklin and Paxinos, 2008; Varga et al., 2020).

Drugs

ME ([Met5]-enkephalin acetate salt), bestatin, DL-thiorphan, strychnine, picrotoxin, DNQX, 4-aminopyridine (4AP), and MK801 were from Sigma-Aldrich (St Louis, MO). Tetrodotoxin and ML-297 was from Tocris Bio-Techne (Minneapolis, MN). All drugs were applied by bath perfusion. Bestatin (10 µM) and thiorphan (1 µM) were included with ME to prevent degradation. ME is an endogenous opioid peptide agonist for mu and delta opioid receptors. Delta opioid receptors are not expressed in KF or preBötC neurons (Varga et al., 2020) and do not cause opioid-induced respiratory depression (Dahan et al., 2001). An EC80 concentration of ME (1–3 µM) was used to ensure robust and reliable responses but avoid acute receptor desensitization that occurs with higher concentrations (Levitt and Williams, 2018).

Statistics

All statistical analyses were performed in GraphPad Prism 8 (La Jolla, CA). All error bars represent SEM unless otherwise stated. Replicates are biological replicates. Data with n > 8 were tested for normality with Kolmogorov–Smirnov tests. Comparisons between two groups were made using paired or unpaired two-tailed t-tests. Comparisons between three or more groups were made using one-way ANOVA followed by Tukey’s post hoc test.

Data availability

Data generated or analyzed during this study are included in the manuscript and supporting files.

References

  1. Book
    1. Franklin K
    2. Paxinos G
    (2008)
    The Mouse Brain in Stereotaxic Coordinates, Compact
    Acad Press.
    1. Katz B
    2. Miledi R
    (1967) The release of acetylcholine from nerve endings by graded electric pulses
    Proceedings of the Royal Society of London. Series B, Biological Sciences 167:23–38.
    https://doi.org/10.1098/rspb.1967.0011
    1. Rhim H
    2. Glaum SR
    3. Miller RJ
    (1993)
    Selective opioid agonists modulate afferent transmission in the rat nucleus tractus solitarius
    The Journal of Pharmacology and Experimental Therapeutics 264:795–800.

Decision letter

  1. Muriel Thoby-Brisson
    Reviewing Editor; CNRS Université de Bordeaux, France
  2. Timothy E Behrens
    Senior Editor; University of Oxford, United Kingdom
  3. Gaspard Montandon
    Reviewer; Unity Health Toronto, Canada

Our editorial process produces two outputs: i) public reviews designed to be posted alongside the preprint for the benefit of readers; ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Opioid suppression of an excitatory pontomedullary respiratory circuit by convergent mechanisms" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Ronald Calabrese as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Gaspard Montandon (Reviewer #2). We strongly apologize for the abnormally long reviewing process. This was due in part to the difficulty to find reviewers available during the summertime and also due to personal problems that one reviewer got in the middle of the process, resulting in the need for us to give him some extra time allowing this reviewer to be able to provide his/her full review.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

This study explores the mechanisms underlying inhibition by opioid drugs of two regions, the preBötzinger Complex, and the Kolliker-Fuse, well-known to mediate opioid-induced respiratory depression and respiratory rhythmogenesis. Based on sophisticated experiments three mechanisms are proposed to occur: postsynaptic inhibition of excitatory KF projection neurons, reduced glutamate release of KF projection neurons via MOR mediated pre-synaptic inhibition, and postsynaptic inhibition of medullary preBötC and VRG neurons. Despite the fact that conclusions are well supported by the data several important concerns must be addressed by the author:

1. Justify the dose (provide a dose-response curve) and the relevance (why not DAMGO) for having chosen ME at 3µM.

2. A detailed description of the recorded neuronal firing patterns should be provided in order to testify their belonging to the respiratory network.

3. Providing a summary diagram or a cartoon of the neural circuits between KF, preBötC, rVRG, and types of projections revealed here would greatly help following the demonstration and summarizing the present findings.

4. Orientate the discussion more towards broader network implications and not mainly focus on preBötC.

5. The possibility that other mechanisms may underlie inhibition by opioids that are not involving potassium current, such as inhibition of voltage-gated calcium channels should be considered.

6. The specificity and potential roles of the different sub-population of KF/LPB group should be more detailed.

7. The presentation (zoom) of the images of cells of interest should be improved.

8. Please provide a better description of the relationship between CGRP and FoxP2 and recorded neurons.

Reviewer #1 (Recommendations for the authors):

The 3 proposed mechanisms: postsynaptic inhibition of excitatory KF projection neurons, reduced glutamate release of KF projection neurons via MOR mediated pre-synaptic inhibition, and postsynaptic inhibition of medullary preBotC and VRG neurons are well supported by the data. However, most of the data presented were predictable and the postsynaptic mechanisms of MOR activation of KF and pre-BotC neurons were described before. That KF projection neurons are predominantly glutamatergic was also predictable from previous studies. Thus, it would have been of major interest to study whether the identified MOR-dependent pre- and post-synaptic mechanisms underlying opioid respiratory depression have different sensitivity to exogenous opioids.

The patch clamp recordings of either pontine or medullary neurons identify MOR effects via valid experimental protocols. However, only a single dose of 3uM ME was used. I think ME dose-response curves for all the patch clamp experiments would allow for a comparative analysis of the opioid sensitivity of post and pre-synaptic inhibition.

The ME dose response curves would increase the value of your study significantly and would help to straighten the discussion – at its present stage, the discussion is largely around in vitro mechanisms and specific subsets of pre-BotC neurons which is not much advancing the understanding of the network mechanisms of opioid respiratory depression. The discussion is hard to understand in the context of the designated roles of excitatory and inhibitory neurons in terms of concepts of rhythm generation and pattern formation. The essential roles of ponto-medullary synaptic interactions and distributed network mechanisms are largely avoided. Your work is clearly supporting that opioid respiratory depression is affecting distributed network functions and thus the extensive discussion concerned with the pre-BotC could be significantly shortened and the discussion then could be more broadly focused on the broader network implications.

Figure 1

Please check scale bars for C and E (-5.02). Also, the labelling pattern between C and E does not match – does Oprm1 Cre/tdT vs Oprm1 Cre/+ change the expression level? Please comment.

Figure 6

Foxp2 expression appears to be very high – the photographs indicate very high numbers of Foxp2 expression neurons in the KF that do not match previously published data. Did you validate the anti-body?

I suggest deleting the figure – at the end, it adds very little to the story.

Reviewer #2 (Recommendations for the authors):

This study explores the mechanisms underlying inhibition by opioid drugs of two regions, the preBötzinger Complex, and the Kolliker-Fuse, well-known to mediate opioid-induced respiratory depression. It shows pre-synaptic inhibition of Kolliker-Fuse glutamatergic neurons which projects to the preBötzinger Complex and rostral VRG. It also suggests that inhibition occurs at the preBötzinger Complex levels.

There are a few comments that the authors should address:

1) Pre-synaptic versus postsynaptic opioid effects. Presynaptic inhibition by ME is clearly demonstrated in this study. Postsynaptic inhibition is identified with outward currents. It is possible that other mechanisms may underlie inhibition by opioids that are not involving potassium current, such as the inhibition of voltage-gated calcium channels. The authors should acknowledge other potential mechanisms that may be involved.

2) Are the mechanisms described and identified here also mediating opioid-induced respiratory depression in-vitro (brainstem preparation) or in-vivo (anesthetized or freely-behaving animals)?

3) Activation of opioid receptors by MET5-enkephalin. Why did the authors use MET-enkephalin which is mostly acting on δ-opioid receptors with lower effects on mu-opioid receptors?

Why not use DAMGO to activate mu-opioid receptors or morphine to mimic opioid drug effects? How was the concentration chosen?

Do we know the affinity of ME on mu and δ receptors at this concentration?

4) FoxP2 and CGRP identify different types of KF/parabrachial neurons. What are the functional roles of these KF subpopulations?

If CGRP KF neurons are involved in co2-induced arousal and express opioid receptors, it may provide a better understanding of the effects of opioids on arousal (sedation) and the relationship between respiratory depression and sleep-wake states or sedation. This could be elaborated.

5) I would suggest the authors prepare a summary diagram or cartoon summarizing the neural circuits between KF, preBötC, rVRG, types of projections, etc… This would provide a clearer picture of the circuitry.

6) The size of microscopy images could be substantially increased for clarity and visibility. Zoom-in images could also help see cell bodies etc. Please see the comments below.

Page 2, line 5. The projection target and synaptic connections are (?) unknown.

Figure 1. Larger images of oprm1 expression could be provided on the right side of the figures, so individual cells could be easily visualized.

Figure 2A. The diagram could be simplified and could show only the projections identified in the study.

Figure 2B. Is it coexpressed with NK1-R? A larger version of NK1-R could be provided to better see cells.

Figure 3A. Separate images with NK1R and oprm1 could be provided. It is difficult to see the different expressions with images with 3 colors.

Page 5. Figure 3E-G. The results describing Figure 3E-G are not consistent with the legend of Figure 3. The effect of ME should be presented before Figure 3K etc. This section needs to be reorganized for clarity and consistency.

Figure 3E. Please define ME in legend.

Figure 4. Why is the concentration of ME lower in these experiments?

Figure 7. The absence of co-expression of oprm1 and cgrp is difficult to visualize with the current figures.

Page 9. Line 24. Considering the MOR inhibits neuronal activity through other mechanisms than GIRK channels, it is possible that MOR inhibition is due to calcium channel inhibition in some neurons.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Opioid suppression of an excitatory pontomedullary respiratory circuit by convergent mechanisms" for further consideration by eLife. Your revised article has been evaluated by Timothy Behrens (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

Overall the manuscript appears to be somewhat improved.

General comment:

Unfortunately, the authors did not address my main suggestions to test whether the different neuronal populations in pons and medulla or the different mechanisms linked to somatic or pre-synaptic MOR may have different sensitivities. The argument that lower doses of morphine showed inconsistencies indicates that there might have been more to explore.

Specific comments:

Introduction:

"With the prevalence of opioid overdose on the rise (Wilson et al., 2020; Mattson et al., 2021), understanding the mechanisms of opioid-induced respiratory depression is of particular importance to aid in development of countermeasures and/or analgesics that do not affect breathing".

How does your study aid the development of countermeasures when systemic opioids act on all 3 mechanisms at the same time?

Methods:

Please provide the age and weight of the mice used for slice preparation and tracing experiments.

Results:

Figure 1 there appears to be two populations of MOR-expressing neurons – for the non-experts please indicate the location lateral parabrachial (external lateral nucleus) and Kolliker-Fuse nucleus in the photographs and sematic drawings.

Figure 2D-I the staining seems diffuse on the photographs (soma, axons/dendrites, and terminals?) – the figure legend states retrograde labeled Oprm1-expressing neurons. Please provide higher magnification pictures to illustrate labelled neurons.

Figure 3? NK1R labelling for the NA? please clarify, – I thought NK1R label pre-BotC neurons while ChAT would mark the NA.

Figure 6

The density of FoxP2 labeled cells indicated in the seems to be very high. Compared to previous reports. While the transcription factor FoxP2 labels cell nuclei it seems that the soma size of Oprm1+ labelled neurons appear in a very similar range l? Please clarify.

Discussion

Paragraph Opioids inhibit excitatory pontomedullary circuitry

"The last and most interesting possibility is that opioid-sensitive glutamatergic KF neurons preferentially synapse onto excitatory medullary neurons, while opioid-non-sensitive KF neurons are GABAergic and/or synapse onto non-excitatory (i.e. inhibitory) medullary neurons".

As I understand you patched MOR-positive cells in the rostral mid-rostral KF while the literature suggests that GABAergic KF neurons are located in the caudal KF so what is the basis of your speculation/hypothesis? I think this part of the discussion is not supported by data from the present study nor by the literature and should be at least toned down.

"An intriguing possibility is that opioid insensitive pontine neurons, which have continued activity during opioid exposure, send prolonged inhibitory input to the ventrolateral medulla to promote apnea".

Any evidence in the literature to support this speculation? What are the mechanisms for apnea mediated via descending inhibitory projections? Since there are only small fractions of inhibitory neurons in the caudal KF I would be cautious to suggest that these can mediate apnea via shifting excitatory-inhibitory balance in the pre-BotC microcircuit.

Para Dorsolateral pons: You should be firm with a statement that you have no evidence to support the findings of Lui et al., 2021.

https://doi.org/10.7554/eLife.81119.sa1

Author response

Essential revisions:

This study explores the mechanisms underlying inhibition by opioid drugs of two regions, the preBötzinger Complex, and the Kolliker-Fuse, well-known to mediate opioid-induced respiratory depression and respiratory rhythmogenesis. Based on sophisticated experiments three mechanisms are proposed to occur: postsynaptic inhibition of excitatory KF projection neurons, reduced glutamate release of KF projection neurons via MOR mediated pre-synaptic inhibition, and postsynaptic inhibition of medullary preBötC and VRG neurons. Despite the fact that conclusions are well supported by the data several important concerns must be addressed by the author:

1. Justify the dose (provide a dose-response curve) and the relevance (why not DAMGO) for having chosen ME at 3µM.

Relevance: ME is the endogenous opioid peptide agonist for mu and delta opioid receptors. ME and DAMGO are both full agonists for mu opioid receptors. When we have compared ME and DAMGO using dose-response curves in KF neurons they were equally effective with EC50 values in the expected range, similar to neurons in the LC, periaqueductal gray and ventral tegmental area (Levitt et al., J Physiol, 2015). The main difference between DAMGO and ME is selectivity. DAMGO is selective for mu opioid receptors, while ME has equal potency for mu and delta opioid receptors. Delta opioid receptors are not expressed on KF neurons or preBötC neurons, since when MORs were genetically deleted from these neurons the effects of ME at these concentrations were eliminated (Varga et al., J Physiol, 2020). Delta opioid receptor activation does not cause respiratory depression, since respiratory depression caused by morphine, which activates both mu and delta opioid receptors, is eliminated in MOR knockout mice in our hands and as previously reported (Dahan et al., 2001). Thus, the selectivity of DAMGO is not needed for the experiments in this study making ME and DAMGO functionally identical. We prefer to use ME for brain slice experiments for several reasons. (1) ME is an endogenous opioid ligand and activation of opioid receptors by the endogenous ligand sheds light on opioid receptor functions independent of the exogenous opioid use context. (2) From a technical standpoint ME is superior to morphine, DAMGO or fentanyl because it rapidly washes from the slice allowing multiple applications and analysis of the “wash” period in comparison to the baseline period to determine if technical artifacts have occurred during the course of the recording.

Concentration: The concentration of ME (3 µM) is approximately the EC80, based on our previous concentration-response experiments in KF neurons (Levitt et al., 2015). We chose this concentration because we knew from previous work that it would give a reliable response in both KF and preBötC neurons (Varga et al., 2020) and avoid causing acute desensitization which can occur with concentrations at and above 10 µM (Levitt and Williams, 2018). In pilot experiments we used lower concentrations of ME (0.3 – 1 µM) followed by 3 µM. The response to these lower concentrations was not as consistent. Since our goal for these experiments was to determine opioid-sensitive versus non-sensitive responses, we wanted to use a high enough concentration to ensure robust and reliable responses, but not too high to avoid receptor desensitization. Clarifying text was added to the Drugs section of the Methods.

2. A detailed description of the recorded neuronal firing patterns should be provided in order to testify their belonging to the respiratory network.

These recordings were performed in non-rhythmic slices, so the firing pattern of the recorded neurons does not provide the respiratory-related information required to determine the respiratory phenotype of each neuron. The viral injections used to achieve regional specificity necessitate using adult mice, which precluded the use of rhythmic slices. To our knowledge, medullary rhythmic slices for preBötC neuron recording are still only possible from early postnatal mice. Rhythmic slices for the pons are not possible, and en bloc preparations are technically not feasible to combine with the recording and optical stimulation techniques used here. It is precisely for this reason that we recorded from vglut2-expressing preBötC and rVRG neurons, as a surrogate method of cell-type identification. This is an important caveat and we have added additional clarification of this limitation to the Results describing these experiments (Page 5, line 30-31).

“Because we could not determine the respiratory-related firing pattern of the neurons we recorded from in this study, we chose to target vglut2-expressing neurons since (1) this contains the population of inspiratory rhythm-generating preBötC neurons (Wallén-Mackenzie et al., 2006; Gray et al., 2010; Cui et al., 2016) and inspiratory premotor rVRG neurons, (2) KF neurons project to excitatory, more so than inhibitory, preBötC neurons (Yang et al., 2020), and (3) deletion of MORs from vglut2 neurons eliminates opioid-induced depression of respiratory output in medullary slices (Sun et al., 2019; Bachmutsky et al., 2020).”

3. Providing a summary diagram or a cartoon of the neural circuits between KF, preBötC, rVRG, and types of projections revealed here would greatly help following the demonstration and summarizing the present findings.

We made a summary schematic highlighting the major findings (Figure 8).

4. Orientate the discussion more towards broader network implications and not mainly focus on preBötC.

We have reorganized the Discussion to lead with expanded sections on implications of pontomedullary circuitry and dorsolateral pontine subpopulations. The section on preBötC mechanisms has been shortened and moved to the end of the Discussion.

5. The possibility that other mechanisms may underlie inhibition by opioids that are not involving potassium current, such as inhibition of voltage-gated calcium channels should be considered.

Inhibition of voltage-gated calcium channels is almost certainly involved in the presynaptic inhibition of neurotransmitter release by opioids. It is possible that opioid inhibition of dendritic VGCCs could affect other processes, such as synaptic integration or back-propagation of action potentials, but these have not been described for KF neurons. The possible involvement of VGCCs has been added to the Introduction and the Discussion.

Intro: “MORs inhibit neurotransmission by hyperpolarizing neurons through activation of somatodendritic GIRK channels and/or inhibiting presynaptic neurotransmitter release through inhibition of voltage-gated calcium channels (Jiang and North, 1992; Chahl, 1996; Zamponi and Snutch, 1998; Al-Hasani and Bruchas, 2011).”

Discussion: “Second, KF neurons that did not have outward currents and were deemed not sensitive to opioids may express MORs, but lack GIRK channels, the functional readout we used to assess opioid sensitivity. MORs on these neurons could instead couple to other effectors, such as voltage-gated calcium channels (Ramirez JM et al., 2021).”

6. The specificity and potential roles of the different sub-population of KF/LPB group should be more detailed.

We have added a section on “Dorsolateral pontine subpopulations” to the Discussion. New Text is copied below:

“The most well-defined area in the dorsolateral pons is the external lateral parabrachial subnucleus, which expresses Lmx1b and CGRP, but not FoxP2 (Karthik et al., 2022; Huang et al., 2021). CGRP-expressing external lateral parabrachial neurons project primarily to the forebrain (Huang et al., 2021), and are involved in pain processing, feeding, and CO2-induced arousal (Campos et al., 2018; Chen et al., 2018; Kaur et al., 2017). Although MORs are highly co-expressed with CGRP in these neurons (Huang et al., 2021), we did not observe opioid-sensitive or Oprm1+ retrograde labeled neurons in the external lateral parabrachial area. Rather, electrophysiologically or histologically identified opioid-sensitive/Oprm1+ neurons that project to the ventrolateral medulla were found rostrally and ventrally in the area overlapping FoxP2 expression in the KF. Thus, two distinct subpopulations of Oprm1+ dorsolateral pontine neurons exist that can be distinguished based on CGRP expression and projection pattern: forebrain-projecting CGRP-expressing neurons and medullary-projecting neurons that do not express CGRP. Both populations are involved in pain and breathing due, at least in part, to reciprocal excitatory synaptic connections (Liu et al., 2022). Although medullary projecting Oprm1+ pontine neurons do not express CGRP (Figure 7), they can still be involved in pain processing, just not to the same extent as forebrain projecting Oprm1/CGRP+ pontine neurons (Liu et al., 2022).

Both populations of Oprm1+ dorsolateral pontine neurons are also likely involved in opioid-induced respiratory depression. MORs in glutamatergic, medullary-projecting, rostral KF neurons could reduce respiratory rate by decreasing excitatory input to the preBötC and rVRG (Figure 3) leading to a distributed blunting effect on inspiration generating processes within the ventrolateral medulla. In contrast, MORs in forebrain-projecting pontine neurons could reduce respiratory rate through intra-pontine excitatory connections with medullary-projecting MOR+ pontine neurons (Liu et al., 2022), or through reduced excitatory input to forebrain areas involved in arousal (Kaur et al., 2017), which may be especially important in sleep-dependent effects of opioids on breathing (Montandon and Horner, 2019).”

7. The presentation (zoom) of the images of cells of interest should be improved.

Images in Figures 1, 2, 3, 6-supplement 1, and 7 have been improved according to suggestions specific to each Figure (described below).

8. Please provide a better description of the relationship between CGRP and FoxP2 and recorded neurons.

Recordings were made from retrogradely labeled neurons located rostral and ventral to the CGRP population of neurons, and in a similar location to the rostral/ventral population of FoxP2 neurons. We did not fill and identify recorded neurons post-hoc, so cannot say with certainty that they expressed CGRP or FoxP2.

We have added these sentences to the Discussion (page 10, lines 18-23):

“Although MORs are highly co-expressed with CGRP in these neurons (Huang et al., 2021), we did not observe opioid-sensitive or MOR+ retrograde labeled neurons in the external lateral parabrachial area. Rather, electrophysiologically or histologically identified opioid-sensitive MOR+ neurons that project to the ventrolateral medulla were found rostrally and ventrally in the area overlapping FoxP2 expression in the KF.”

Reviewer #1 (Recommendations for the authors):

The 3 proposed mechanisms: postsynaptic inhibition of excitatory KF projection neurons, reduced glutamate release of KF projection neurons via MOR mediated pre-synaptic inhibition, and postsynaptic inhibition of medullary preBotC and VRG neurons are well supported by the data. However, most of the data presented were predictable and the postsynaptic mechanisms of MOR activation of KF and pre-BotC neurons were described before. That KF projection neurons are predominantly glutamatergic was also predictable from previous studies. Thus, it would have been of major interest to study whether the identified MOR-dependent pre- and post-synaptic mechanisms underlying opioid respiratory depression have different sensitivity to exogenous opioids.

The patch clamp recordings of either pontine or medullary neurons identify MOR effects via valid experimental protocols. However, only a single dose of 3uM ME was used. I think ME dose-response curves for all the patch clamp experiments would allow for a comparative analysis of the opioid sensitivity of post and pre-synaptic inhibition.

As described above (Essential revision #1), ME (3 µM) is approximately the EC80, based on our previous dose-response experiments in KF neurons. We chose this concentration because it would provide robust and reliable responses and avoid causing acute desensitization which can occur with concentrations at and above 10 µM. We agree that it would be beneficial to examine the difference in sensitivity of pre-synaptic and post-synaptic responses. However, we think it would be most valuable to examine this difference in sensitivity in the context of acute versus chronic opioid exposure, which is out of scope of the current study and is best examined in a future study dedicated to this goal.

The ME dose response curves would increase the value of your study significantly and would help to straighten the discussion – at its present stage, the discussion is largely around in vitro mechanisms and specific subsets of pre-BotC neurons which is not much advancing the understanding of the network mechanisms of opioid respiratory depression. The discussion is hard to understand in the context of the designated roles of excitatory and inhibitory neurons in terms of concepts of rhythm generation and pattern formation. The essential roles of ponto-medullary synaptic interactions and distributed network mechanisms are largely avoided. Your work is clearly supporting that opioid respiratory depression is affecting distributed network functions and thus the extensive discussion concerned with the pre-BotC could be significantly shortened and the discussion then could be more broadly focused on the broader network implications.

We have significantly revised and reoriented the Discussion towards distributed pontomedullary mechanisms and circuitry. We have added a discussion of excitatory/inhibitory balance and expanded discussion of potential sources of opioid influence, including BötC, NTS, rVRG and raphe. We added a section on dorsolateral pontine subpopulations, including roles in breathing and pain, to address other reviewer comments. We shortened the section on preBötC mechanisms and moved it to the end of the Discussion.

Figure 1

Please check scale bars for C and E (-5.02). Also, the labelling pattern between C and E does not match – does Oprm1 Cre/tdT vs Oprm1 Cre/+ change the expression level? Please comment.

The bregma level (-5.02) for C and E (now G) is accurate according to the mouse brain atlas (Franklin and Paxinos, 2008). The scale bars are different, as indicated in the figure legend, because the image in E (now G) is more zoomed in than C.

The fluorescent labeling in C and E (now G) are different because the experimental design was different. In figures A-D, Oprm1-Cre mice were bred with Ai9 tdT-Cre reporter mice. In the resulting offspring (Oprm1 Cre/tdT), Cre drives tdT expression in all Oprm1-expressing neurons in the brain throughout development, leading to tdT expression in both cell bodies and axon terminals projecting to the dorsolateral pons shown in these images. In new figures E-H, an AAV encoding Cre-dependent expression of GFP was injected into the KF/parabrachial area in adult Oprm1-Cre/+ mice, leading to GFP expression only in the neurons in the injection area (KF/PB), and not in axon terminals projecting to the KF/PB from elsewhere in the brain. The cell bodies are more visible in E-H because they are not obscured by the dense innervation from Oprm1-expressing axons into the area.

Figure 6

Foxp2 expression appears to be very high – the photographs indicate very high numbers of Foxp2 expression neurons in the KF that do not match previously published data. Did you validate the anti-body?

The expression of FoxP2 we observed closely matches the recent comprehensive work from Karthik et al., 2022. FoxP2 expression in rostral PB/KF (bregma -4.9) is quite abundant, similar to Figure 6. In contrast, FoxP2 is excluded from the external lateral parabrachial subnucleus, which contains cells that project to the forebrain and express CGRP and Lmx1b and is located more caudally (bregma -5.2). We also observe this exclusion of FoxP2 from the external lateral parabrachial subnucleus, even though FoxF2 is detected in the surrounding PB. We have added representative images in Figure 6 —figure supplement 1. These images also show FoxP2 expression in caudal KF, consistent with Karthik et al., 2022 and overlapping with a small population of medullary projecting Oprm1+ cells.

I suggest deleting the figure – at the end, it adds very little to the story.

We hope that the additional figure supplement and text in the Results and discussion related to dorsolateral pontine subpopulations adds enough to the story to justify keeping these data in the manuscript.

Additional text in Results:

“Oprm1+, medullary-projecting KF neurons expressed FoxP2 (n=3; Figure 6), consistent with the population of glutamatergic FoxP2 and Lmx1b neurons in the rostral KF (Karthik et al., 2022). These are a separate population from FoxP2 expressing neurons located more dorsally and caudally in the inner portion of the external lateral parabrachial area and those activated by sodium deprivation (Geerling et al., 2011; Karthik et al., 2022). FoxP2 expression also overlapped with a smaller population of Oprm1+ medullary projecting neurons in the caudal KF containing GABAergic neurons (Figure 6 – Figures supplement 1) (Geerling et al., 2017; Karthik et al., 2022). FoxP2 was not detected in the outer portion of the external lateral parabrachial subnucleus (Figure 6 —figure supplement 1), consistent with previous findings (Geerling et al., 2011; Karthik et al., 2022).”

Reviewer #2 (Recommendations for the authors):

This study explores the mechanisms underlying inhibition by opioid drugs of two regions, the preBötzinger Complex, and the Kolliker-Fuse, well-known to mediate opioid-induced respiratory depression. It shows pre-synaptic inhibition of Kolliker-Fuse glutamatergic neurons which projects to the preBötzinger Complex and rostral VRG. It also suggests that inhibition occurs at the preBötzinger Complex levels.

There are a few comments that the authors should address:

1) Pre-synaptic versus postsynaptic opioid effects. Presynaptic inhibition by ME is clearly demonstrated in this study. Postsynaptic inhibition is identified with outward currents. It is possible that other mechanisms may underlie inhibition by opioids that are not involving potassium current, such as the inhibition of voltage-gated calcium channels. The authors should acknowledge other potential mechanisms that may be involved.

See response to Essential revision #5.

2) Are the mechanisms described and identified here also mediating opioid-induced respiratory depression in-vitro (brainstem preparation) or in-vivo (anesthetized or freely-behaving animals)?

Given the density of the projections we identified here and the importance of these brainstem areas in OIRD, we predict that these mechanisms would mediate OIRD in vitro and in vivo. We hope this study lays the groundwork for future studies assessing the impact of these projections on respiratory activity.

3) Activation of opioid receptors by MET5-enkephalin. Why did the authors use MET-enkephalin which is mostly acting on delta-opioid receptors with lower effects on mu-opioid receptors?

Why not use DAMGO to activate mu-opioid receptors or morphine to mimic opioid drug effects? How was the concentration chosen?

Do we know the affinity of ME on mu and delta receptors at this concentration?

See response to Essential revision #1. In addition, morphine is also a delta opioid receptor agonist, with similar affinity for mu and delta receptors, so it would have similar selectivity considerations as ME. Morphine is also a partial agonist (Levitt and Williams, Mol Pharmacol, 2012), which makes it difficult to see small effects. GIRK currents in mouse KF neurons can be relatively small and we did not know what size of effect would occur pre-synaptically. Since the goal of these experiments was to identify MOR+ projections, we were looking for robust results that would differentiate opioid-sensitive vs. non-opioid-sensitive responses, so we chose to use a full agonist (ME) at an ~EC80 concentration.

4) FoxP2 and CGRP identify different types of KF/parabrachial neurons. What are the functional roles of these KF subpopulations?

If CGRP KF neurons are involved in co2-induced arousal and express opioid receptors, it may provide a better understanding of the effects of opioids on arousal (sedation) and the relationship between respiratory depression and sleep-wake states or sedation. This could be elaborated.

We have added two paragraphs to the Discussion on dorsolateral pontine subpopulations, including functional roles. The new paragraphs are below:

“The most well-defined area in the dorsolateral pons is the external lateral parabrachial subnucleus, which expresses Lmx1b and CGRP, but not FoxP2 (Karthik et al., 2022; Huang et al., 2021). CGRP-expressing external lateral parabrachial neurons project primarily to the forebrain (Huang et al., 2021), and are involved in pain processing, feeding, and CO2-induced arousal (Campos, 2018; Chen, 2018; Kaur et al., 2017). Although MORs are highly co-expressed with CGRP in these neurons (Huang et al., 2021), we did not observe opioid-sensitive or Oprm1+ retrograde labeled neurons in the external lateral parabrachial area. Rather, electrophysiologically or histologically identified opioid-sensitive/Oprm1+ neurons that project to the ventrolateral medulla were found rostrally and ventrally in the area overlapping FoxP2 expression in the KF. Thus, two distinct subpopulations of Oprm1+ dorsolateral pontine neurons exist that can be distinguished based on CGRP expression and projection pattern: forebrain-projecting CGRP-expressing neurons and medullary-projecting neurons that do not express CGRP. Both populations are involved in pain and breathing due, at least in part, to reciprocal excitatory synaptic connections (Liu et al., 2022). Although medullary projecting Oprm1+ pontine neurons do not express CGRP (Figure 7), they can still be involved in pain processing, just not to the same extent as forebrain projecting Oprm1/CGRP+ pontine neurons (Liu et al., 2022).

Both populations of Oprm1+ dorsolateral pontine neurons are also likely involved in opioid-induced respiratory depression. MORs in glutamatergic medullary projecting rostral KF neurons could reduce respiratory rate by decreasing excitatory input to the preBötC and rVRG (Figure 3) leading to a distributed blunting effect on inspiration generating processes within the ventrolateral medulla. In contrast, MORs in forebrain-projecting pontine neurons could reduce respiratory rate through intra-pontine excitatory connections with medullary-projecting MOR+ pontine neurons (Liu et al., 2022), or through reduced excitatory input to forebrain areas involved in arousal (Kaur et al., 2017), which may be especially important in sleep-dependent effects of opioids on breathing (Montandon and Horner, 2019).”

5) I would suggest the authors prepare a summary diagram or cartoon summarizing the neural circuits between KF, preBötC, rVRG, types of projections, etc… This would provide a clearer picture of the circuitry.

See response to Essential revision #3. Hopefully this new summary schematic (Figure 8) provides a clear picture of the circuitry and MOR effects.

6) The size of microscopy images could be substantially increased for clarity and visibility. Zoom-in images could also help see cell bodies etc. Please see the comments below.

Images have been enlarged and images displaying cell bodies have been added, as described in comments below.

Page 2, line 5. The projection target and synaptic connections are (?) unknown.

Corrected.

Figure 1. Larger images of oprm1 expression could be provided on the right side of the figures, so individual cells could be easily visualized.

Unfortunately, the density of MOR/tdT positive axonal projections into the pons obscures the ability to visualize individual cells in the Oprm1 Cre/tdT mice, even with higher magnification. This is not an issue for the experiments using viral mediated labeling of oprm1-expressing neurons, since only neurons in the injection area express GFP. We have now added a column to this figure showing labeling of oprm1 expressing neurons using the viral approach at all bregma levels, so that individual neurons can be visualized.

Figure 2A. The diagram could be simplified and could show only the projections identified in the study.

Figure 2B. Is it coexpressed with NK1-R? A larger version of NK1-R could be provided to better see cells.

The diagram in 2A has been simplified and the NK1-R image in 2B has been enlarged.

Figure 3A. Separate images with NK1R and oprm1 could be provided. It is difficult to see the different expressions with images with 3 colors.

Separate images have been provided.

Page 5. Figure 3E-G. The results describing Figure 3E-G are not consistent with the legend of Figure 3. The effect of ME should be presented before Figure 3K etc. This section needs to be reorganized for clarity and consistency.

Figure 3 was reorganized to improve clarity and consistency with the text. The results of TTX and DNQX experiments have been moved to a supplemental figure to reduce the density of the figure and highlight the major opioid-related findings.

Figure 3E. Please define ME in legend.

Met-enkephalin (ME) has been defined in the legend.

Figure 4. Why is the concentration of ME lower in these experiments?

These experiments were done first using a concentration that we knew from prior experiments (Varga et al., 2020) would evoke reliable GIRK currents in KF neurons. The concentration of ME was increased for the experiments recording from medullary neurons to a concentration that would ensure reliable and robust responses in these neurons based on previous experience (Varga et al., 2020).

Figure 7. The absence of co-expression of oprm1 and cgrp is difficult to visualize with the current figures.

Zoomed in images have been added to help with visualization. Note, there is no CGRP expression in the image once we zoom in on the Oprm1+ neurons.

Page 9. Line 24. Considering the MOR inhibits neuronal activity through other mechanisms than GIRK channels, it is possible that MOR inhibition is due to calcium channel inhibition in some neurons.

See response to Essential Revision # 5 above. Specifically, this point in the Discussion was revised as follows: “Second, KF neurons that did not have outward currents and were deemed not sensitive to opioids may express MORs, but lack GIRK channels, the functional readout we used to assess opioid sensitivity. MORs on these neurons could instead couple to other effectors, such as voltage-gated calcium channels (Ramirez JM et al., 2021).”

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

Overall the manuscript appears to be somewhat improved.

General comment:

Unfortunately, the authors did not address my main suggestions to test whether the different neuronal populations in pons and medulla or the different mechanisms linked to somatic or pre-synaptic MOR may have different sensitivities. The argument that lower doses of morphine showed inconsistencies indicates that there might have been more to explore.

While there might be more to explore, these experiments are not necessary to justify our conclusions that opioids inhibit an excitatory pontomedullary respiratory circuit via three mechanisms that are newly defined in a projection-specific manner, which were “well supported by the data”. In addition, our ability to do these experiments in a timely manner is constrained by a lack of mice and personnel since we just moved institutions and are still getting the lab back to normal operations. We have added a paragraph to the Discussion (below) to address the potential for differences in sensitivity based on presynaptic and postsynaptic receptors at other synapses. We ask for understanding and lenience from the reviewer regarding the necessity of these experiments and look forward to exploring this further in the future.

“Sensitivity and regulation of presynaptic and postsynaptic opioid receptors

Presynaptic and postsynaptic mu opioid receptors couple to distinct effectors and are regulated differently, which can lead to differences in sensitivity that change with prolonged opioid exposure (Coutens and Ingram, 2023). For instance, postsynaptic, but not presynaptic, opioid receptors couple to GIRK channels (Luscher et al., 1997) through binding of up to four Gβγ subunits directly to the channel (Whorton et al., 2013). In contrast, presynaptic opioid receptors inhibit neurotransmitter release through inhibition of VGCCs (Heinke et al., 2011) or direct inhibition of vesicle release machinery (Blackmer et al., 2005; Gerachshenko et al., 2005). Coupling to these presynaptic effectors may be more sensitive, since VGCCs can be inhibited by a single Gβγ subunit (Zamponi and Snutch, 1998) and vesicular release is steeply calcium dependent (Dodge and Rahamimoff, 1967). Consistent with this, presynaptic opioid receptor responses have higher sensitivity than postsynaptic responses when directly compared (Pennock and Hentges, 2011). Prolonged exposure to high doses of opioids can exacerbate differences in sensitivity, since postsynaptic receptors desensitize more readily than presynaptic receptors (Blanchet and Luscher, 2002; Fyfe et al., 2010; Lowe and Bailey, 2015; Pennock et al., 2012; Rhim et al., 1993). Thus, the responses of presynaptic receptors may predominate, especially after prolonged opioid exposure, for reasons related to receptor reserve, coupling to effectors and/or receptor regulation. The relative sensitivity of presynaptic and postsynaptic receptors in the pontomedullary circuit identified here will be important to determine, especially since postsynaptic opioid receptors on KF neurons are resistant to desensitization (Levitt and Williams, 2018), suggesting unique receptor regulation in these neurons.”

We were also prompted by the reviewer’s comment to carefully check the language about sensitivity throughout the manuscript and changed “sensitivity” to “proportion” in some instances and added quantification of these proportions. For instance: (Results, page 6, line 9) “presynaptic opioid receptors inhibit glutamate release from KF terminals onto a majority of excitatory preBötC and rVRG neurons (91 % 20 of 22 neurons).” And (Discussion page 8, lines 17-18): “These mechanisms converge on a projection-specific opioid-sensitive circuit, whereby MOR-expressing KF neurons synapse onto MOR-expressing excitatory preBötC and rVRG neurons at a proportion that is higher than predicted based on MOR expression in either of these populations alone (Bachmutsky et al., 2020; Kallurkar et al., 2022; Levitt et al., 2015).” We would like to thank the reviewer for emphasizing this comment and think the additional text has helped further “straighten the Discussion”.

Specific comments:

Introduction:

"With the prevalence of opioid overdose on the rise (Wilson et al., 2020; Mattson et al., 2021), understanding the mechanisms of opioid-induced respiratory depression is of particular importance to aid in development of countermeasures and/or analgesics that do not affect breathing".

How does your study aid the development of countermeasures when systemic opioids act on all 3 mechanisms at the same time?

As Reviewer 1 noted in the original round of reviews “The study advances our knowledge of network mechanisms that mediate opioid respiratory depression and may provide interesting frameworks for the development of therapies to counteract or prevent opioid respiratory depression.” However, we understand the point that we did not explore countermeasures (or analgesia), so we have revised this sentence as follows: "With the prevalence of opioid overdose on the rise (Wilson et al., 2020; Mattson et al.,2021), understanding the network mechanisms of opioid-induced respiratory depression is of particular importance".

Methods:

Please provide the age and weight of the mice used for slice preparation and tracing experiments.

Mice used for brain slice recordings and tracing experiments were 2-4 months old. This is in the Methods animals section and has been added to the beginning of the Methods sections on brain slice electrophysiology and immunohistochemistry and microscopy.

Mice used in this study are normally developing mice with weights commensurate with age and sex during normal development of C57BL/6J mice (20-30 g for 2–4 month-old mice). Since we did not routinely record body weight, we don’t feel comfortable stating a weight range in the Methods, and instead state “weights commensurate with age and sex of normally developing C57BL/6J mice”.

Results:

Figure 1 there appears to be two populations of MOR-expressing neurons – for the non-experts please indicate the location lateral parabrachial (external lateral nucleus) and Kolliker-Fuse nucleus in the photographs and sematic drawings.

We are glad the reviewer noticed two populations of MOR-expressing neurons. We agree and have updated Figure 1 to indicate locations of the KF, LPB and PBel on the schematics and images.

Figure 2D-I the staining seems diffuse on the photographs (soma, axons/dendrites, and terminals?) – the figure legend states retrograde labeled Oprm1-expressing neurons. Please provide higher magnification pictures to illustrate labelled neurons.

We have added higher magnification pictures showing retrograde labelled neurons from injections into the preBotC, rVRG and BotC in Supplemental Figure 2-2. We also agree that the most intense area of GFP fluorescence is diffuse and contains a significant amount of cell processes. We have changed the figure legend to “Oprm1+ KF neurons and neurites retrogradely labeled from the preBotC and rVRG” to reflect this. We also refer to retrograde labeled Oprm1-expressing “neurons and neurites” later in the legend and in the main text of the manuscript.

Figure 3? NK1R labelling for the NA? please clarify, – I thought NK1R label pre-BotC neurons while ChAT would mark the NA.

NK1R expression is a commonly used a marker for preBotC, but NK1Rs are also found in the NA (as well as ChAT). To facilitate the review process, please see NK1R expression (message or protein) in NA and preBotC from the following sources: image in Allen Brain Atlas, Figure 2 from Montandon, Liu and Horner (Scientific Reports, 2016), Figure 5 from McKay and Feldman (Am J Resp Critical Care Med, 2007), Figure 1 from Gray et al., (Nature Neuroscience, 2001) and Figure 2 from Wang, Stornetta, Rosin and Guyenet (JCN, 2001). NK1R labeling was not the only marker we used for the NA (we also used cytoarchitecture), but the NK1R labeling was obvious in the slices that were immunostained for NK1R, which are in Figure 2B and Figure 3. Regarding Figure 3, we removed the panels in question since they do not contribute much and were a source of confusion.

Figure 6

The density of FoxP2 labeled cells indicated in the seems to be very high. Compared to previous reports. While the transcription factor FoxP2 labels cell nuclei it seems that the soma size of Oprm1+ labelled neurons appear in a very similar range l? Please clarify.

We also observe that FoxP2 labels only the cell nuclei, and importantly does not fill the entirety of the Oprm1+, GFP labelled neurons. We have improved and enlarged the higher magnification images in Figure 6 to make this more apparent. The overlap of FoxP2 and DAPI is best observed in GFP negative neurons. The exclusion of FoxP2 from the cytosol of GFP labelled neurons is harder to visualize in many of the neurons because the nucleus takes up a large portion of the soma. The uppermost neuron indicated by an arrowhead has the largest soma in this image and is the easiest to see that FoxP2 does not fill the entire soma.

Because of the difficulty to qualitatively visualize the nuclear restriction of FoxP2, we also quantified the relative intensities of the DAPI, FoxP2 and GFP signals across labelled somas. Example profile plots for the indicated neuron (yellow line) is shown in Author response image 1. Notice that FoxP2 overlaps with the DAPI signal, marking the nucleus, and is excluded from the cytosolic portion of the soma (labeled with GFP).

Author response image 1
Example intensity profile plots of the GFP, FoxP2 and DAPI signal across the neuron, indicated by the yellow line on the image.

In addition, we performed control experiments, where the primary antibody against FoxP2 was omitted, but all other conditions were identical and performed in parallel (Author response image 2). FoxP2 specific labeling was in neurons (labeled with Neurotrace) and lacking from slices where the primary antibody was omitted.

Author response image 2
FoxP2 immunolabeling (magenta) in the KF area with (top row) and without (bottom row) the primary anti-FoxP2 antibody.

Neurotrace (green) is a fluorescent Nissl stain and labels neurons.

I am wondering which previous reports the reviewer is referring to regarding the density of FoxP2 labeled cells. Our reading of the prior literature reports that FoxP2 expression is lesser in caudal PB areas, but the density increases in the dorsolateral pons in more rostral slices (see Karthik et al., 2022). The density of FoxP2 labeled cells we observe in the rostral KF area matches the Allen Brain Atlas and Karthik et al., JCN, 2022 at the rostral-caudal level we are investigating (bregma -4.84). To help facilitate the review process, we added low resolution widefield images (admittedly low quality) showing that FoxP2 staining is focally restricted, in a manner similar to the Allen Brain Atlas (Author response image 3).

Author response image 3
FoxP2 immunolabeling in the rostral dorsolateral pons (~bregma -4.84).

See also, ISH for Foxp2 from the Allen Brain Atlas.

Also see pertinent Figure panels from Karthik et al., JCN, 2022 that show high density of FoxP2 expressing cells in the rostral KF (> 500 neurons at bregma -4.9).

Discussion

Paragraph Opioids inhibit excitatory pontomedullary circuitry

"The last and most interesting possibility is that opioid-sensitive glutamatergic KF neurons preferentially synapse onto excitatory medullary neurons, while opioid-non-sensitive KF neurons are GABAergic and/or synapse onto non-excitatory (i.e. inhibitory) medullary neurons".

As I understand you patched MOR-positive cells in the rostral mid-rostral KF while the literature suggests that GABAergic KF neurons are located in the caudal KF so what is the basis of your speculation/hypothesis? I think this part of the discussion is not supported by data from the present study nor by the literature and should be at least toned down.

Good point regarding the GABAergic KF neurons. We have removed the possibility that non-opioidergic KF neurons are GABAergic, and revised the text as follows.

New text: “The last and most interesting possibility is that opioid-sensitive glutamatergic KF neurons preferentially synapse onto excitatory medullary neurons, while non-opioidergic KF neurons might preferentially synapse onto non-excitatory (i.e. inhibitory) medullary neurons. This hypothesis is consistent with anatomical tracing studies showing that KF neurons project to excitatory and, to a lesser extent, inhibitory preBötC neurons (Yang et al., 2020), and could be tested by recording from labeled inhibitory neurons in the ventrolateral medulla.”

"An intriguing possibility is that opioid insensitive pontine neurons, which have continued activity during opioid exposure, send prolonged inhibitory input to the ventrolateral medulla to promote apnea".

Any evidence in the literature to support this speculation? What are the mechanisms for apnea mediated via descending inhibitory projections? Since there are only small fractions of inhibitory neurons in the caudal KF I would be cautious to suggest that these can mediate apnea via shifting excitatory-inhibitory balance in the pre-BotC microcircuit.

We agree that direct inhibitory descending projections that promote apnea are unlikely. More likely, descending excitatory projections onto inhibitory medullary neurons could promote apnea, possibly through mechanisms overlapping those involved in post-inspiratory apneas observed by excitation of the KF. We have rephrased to indicate this possibility:

“An intriguing possibility is that opioid insensitive pontine neurons, which have continued activity during opioid exposure, send prolonged excitatory input to inhibitory neurons in the ventrolateral medulla to promote apnea, perhaps using pathways overlapping those involved in apneas evoked by excitation of certain parts of the KF area (Saunders and Levitt, 2020; Dutschmann and Dick, 2012; Dutschmann and Herbert, 2006). This could include opioid insensitive KF neurons that project to inhibitory neurons in the BötC, since a higher proportion of opioid insensitive pontine neurons projected to the BötC (Figure 2 and 4 supplements).”

We also removed the sentence regarding a shift in the excitatory-inhibitory balance. I think it may be possible that opioids could shift the excitatory-inhibitory balance if opioid-insensitive neurons are synapsing onto inhibitory neurons in the medulla (BotC or preBotC), but this is highly speculative and has been removed from the manuscript.

Para Dorsolateral pons: You should be firm with a statement that you have no evidence to support the findings of Lui et al., 2021.

I think the assertion that we have no evidence to support the findings of Lui et al. is a little extreme, since we were focusing on slightly different areas and many of the functional effects on breathing align with what we would have predicted from manipulating opioidergic neurons in the dorsolateral pons. The main difference between our study and theirs is that we don’t see labeling of the PBN “shell” with retrograde injections into the preBotC area. This could be due to differences in the injection location and/or the experimental design. Liu et al. injected rgAAV-EF1a-DIO-FLPo into the preBotC and AAV9-EF1a-fDIO-ChR2-eYFP into the PB. Because only a few copies of recombinase are needed to drive expression of flp, which will be amplified by fDIO-ChR2-YFP, this approach has higher sensitivity but with the potential for false positives (ie. if the retrograde AAV spreads outside of the intended injection area). In contrast, our approach (rgAAV-DIO-GFP) will have less sensitivity but with less potential for false positives and more potential for false negatives. It remains to be determined which is the case. In addition, the cellular distribution of YFP tagged ChR is different from soluble GFP, leading to differences in the fluorescence appearance in labeled neurons and dendrites. We have added a shortened version of these points to the Discussion: “We also did not observe a “shell” pattern of retrograde labeled Oprm1+ neurons surrounding the external lateral parabrachial area, in contrast with (Liu et al., 2022), which could be due to slight differences in injection location, the fluorescent probe and/or sensitivity of the experimental design.”

https://doi.org/10.7554/eLife.81119.sa2

Article and author information

Author details

  1. Jordan T Bateman

    Department of Pharmacology and Therapeutics, Breathing Research and Therapeutics Center, University of Florida, Gainesville, United States
    Contribution
    Formal analysis, Funding acquisition, Investigation, Visualization, Writing - original draft, Writing – review and editing
    Competing interests
    No competing interests declared
  2. Erica S Levitt

    Department of Pharmacology and Therapeutics, Breathing Research and Therapeutics Center, University of Florida, Gainesville, United States
    Present address
    Department of Pharmacology, University of Michigan, Ann Arbor, United States
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Writing – review and editing
    For correspondence
    elsawyer@umich.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3634-6594

Funding

National Institute on Drug Abuse (R01DA047978)

  • Erica S Levitt

National Institute on Drug Abuse (F31DA053798)

  • Jordan T Bateman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by the National Institutes of Health Grant R01DA047978 (ESL). JTB was supported by F31DA053798. We would like to thank Dr. Richard Palmiter (University of Washington) for generously providing the Oprm1-Cre mice and Keiko Arakawa for technical assistance. We thank Drs. Gordon Mitchell, John Williams, and Adrienn Varga for comments on the manuscript.

Ethics

All experiments were approved by the Institutional Animal Care and Use Committee at the University of Florida (protocol #09515) and were in agreement with the National Institutes of Health "Guide for the Care and Use of Laboratory Animals."

Senior Editor

  1. Timothy E Behrens, University of Oxford, United Kingdom

Reviewing Editor

  1. Muriel Thoby-Brisson, CNRS Université de Bordeaux, France

Reviewer

  1. Gaspard Montandon, Unity Health Toronto, Canada

Version history

  1. Received: June 16, 2022
  2. Preprint posted: June 26, 2022 (view preprint)
  3. Accepted: June 13, 2023
  4. Accepted Manuscript published: June 14, 2023 (version 1)
  5. Version of Record published: July 3, 2023 (version 2)

Copyright

© 2023, Bateman and Levitt

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 450
    Page views
  • 63
    Downloads
  • 2
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Jordan T Bateman
  2. Erica S Levitt
(2023)
Opioid suppression of an excitatory pontomedullary respiratory circuit by convergent mechanisms
eLife 12:e81119.
https://doi.org/10.7554/eLife.81119

Share this article

https://doi.org/10.7554/eLife.81119

Further reading

    1. Neuroscience
    Peibo Xu, Jian Peng ... Yuejun Chen
    Research Article

    Deciphering patterns of connectivity between neurons in the brain is a critical step toward understanding brain function. Imaging-based neuroanatomical tracing identifies area-to-area or sparse neuron-to-neuron connectivity patterns, but with limited throughput. Barcode-based connectomics maps large numbers of single-neuron projections, but remains a challenge for jointly analyzing single-cell transcriptomics. Here, we established a rAAV2-retro barcode-based multiplexed tracing method that simultaneously characterizes the projectome and transcriptome at the single neuron level. We uncovered dedicated and collateral projection patterns of ventromedial prefrontal cortex (vmPFC) neurons to five downstream targets and found that projection-defined vmPFC neurons are molecularly heterogeneous. We identified transcriptional signatures of projection-specific vmPFC neurons, and verified Pou3f1 as a marker gene enriched in neurons projecting to the lateral hypothalamus, denoting a distinct subset with collateral projections to both dorsomedial striatum and lateral hypothalamus. In summary, we have developed a new multiplexed technique whose paired connectome and gene expression data can help reveal organizational principles that form neural circuits and process information.

    1. Neuroscience
    Maureen van der Grinten, Jaap de Ruyter van Steveninck ... Yağmur Güçlütürk
    Tools and Resources

    Blindness affects millions of people around the world. A promising solution to restoring a form of vision for some individuals are cortical visual prostheses, which bypass part of the impaired visual pathway by converting camera input to electrical stimulation of the visual system. The artificially induced visual percept (a pattern of localized light flashes, or ‘phosphenes’) has limited resolution, and a great portion of the field’s research is devoted to optimizing the efficacy, efficiency, and practical usefulness of the encoding of visual information. A commonly exploited method is non-invasive functional evaluation in sighted subjects or with computational models by using simulated prosthetic vision (SPV) pipelines. An important challenge in this approach is to balance enhanced perceptual realism, biologically plausibility, and real-time performance in the simulation of cortical prosthetic vision. We present a biologically plausible, PyTorch-based phosphene simulator that can run in real-time and uses differentiable operations to allow for gradient-based computational optimization of phosphene encoding models. The simulator integrates a wide range of clinical results with neurophysiological evidence in humans and non-human primates. The pipeline includes a model of the retinotopic organization and cortical magnification of the visual cortex. Moreover, the quantitative effects of stimulation parameters and temporal dynamics on phosphene characteristics are incorporated. Our results demonstrate the simulator’s suitability for both computational applications such as end-to-end deep learning-based prosthetic vision optimization as well as behavioral experiments. The modular and open-source software provides a flexible simulation framework for computational, clinical, and behavioral neuroscientists working on visual neuroprosthetics.