Sigh generation in preBötzinger complex

To test whether selective activation of pF NMB or GRP neurons, i.e., bombesin peptide-expressing pF neurons generates sighs, we established Grp-ChR2 or Nmb-ChR2 mouse lines (see Methods) by crossing floxed-ChR2-tdTomato mice with Grp^Cre and Nmb^Cre mice (Figure 1a and c). We then measured the effects of targeted pF photostimulation (long pulse photostimulation, LPP, single bilateral 4–10 s 5 mW pulse or short pulse photostimulation, SPP, single bilateral 100–500 ms pulse; see Methods) Cui et al., 2016 in anesthetized Grp-ChR2 or Nmb-ChR2 mice.

Figure 1 with 2 supplements see all

Download asset Open asset

In brainstems of adult Grp-ChR2 mice at the rostrocaudal level of the facial nucleus (7 N), tdTomato fluorescence is expressed mainly in the dorsomedial part of retrotrapezoid nucleus/parafacial respiratory group (RTN/pFRG; Figure 1—figure supplement 1a), consistent with previous data in newborn mice (Li et al., 2016). In brainstems of adult Nmb-ChR2 mice, Nmb-tdTomato is mainly expressed ventral and lateral to 7 N, with a few of these neurons in the 7 N, consistent with previous data in mice (Shi et al., 2017; Figure 1—figure supplement 1b).

In Grp-ChR2 mice, LPP in pF elicited a large breath equivalent to an endogenous sigh in both amplitude and tidal volume (V_T) when initiated during phase 0.3–1.0 of the sigh cycle (see Methods); we refer to such events as ectopic sighs (Figure 1a, bottom). Ectopic sighs in Grp-ChR2 mice were of partial doublet or augmented breath shape (See Appendix 1—figure 1) and were indistinguishable from endogenous sighs in both V_T and inspiratory (T_I) and expiratory (T_E) duration and exhibited a postsigh apnea (T_E immediately following ectopic sighs increased to 216 ± 35% of T_E of eupneic breaths, Figure 1b). During sigh phase 0.3–0.9, the interval from the previous endogenous sigh to the LPP-induced ectopic sigh, i.e., the perturbed sigh cycle, was shorter than the prior control sigh cycle. The interval between an ectopic sigh and the next (endogenous) sigh was, on average, indistinguishable from the interval between endogenous sighs, indicating that ectopic sighs reset the sigh cycle (Figure 1a). No ectopic sighs could be generated by bilateral pF SPP in Grp-ChR2 mice during inspiration or expiration (phase: 0.0–1.0).

In Nmb-ChR2 mice, the effects of pF SPP and LPP were similar to those in Grp-ChR2 mice, as SPP (phase: 0.0–1.0) had no effect on sighing while LPP induced ectopic sighs, which appeared as a partial doublet or augmented breaths, and reset the sigh cycle during 0.4–0.9 sigh phase (Figure 1c). Ectopic sighs were also indistinguishable from spontaneous sighs in V_T, T_E, and T_I and they exhibited a postsigh apnea (T_E after evoked sighs increased to 151 ± 22% of T_E of eupneic breaths, Figure 1d).

Notably, in either Grp-ChR2 or Nmb-ChR2 mice, pF LPP did not elicit ectopic sighs when applied very shortly after an endogenous sigh (phase range: 0.0–0.2 and 0.0–0.3 of endogenous sigh cycle, respectively; Figure 1e), indicating a postsigh refractory period for sigh induction. LPP in these cases had no subliminal effect on the sigh cycle, as the expected time to the next endogenous sigh was unchanged.

When pF LPP was initiated immediately following the end of the postsigh refractory period, the perturbed sigh cycle was significantly shorter than the control sigh cycle. For pF LPP later in the sigh cycle, i.e., closer to when the next endogenous sigh was expected, ectopic sighs occurred at even shorter latencies (Figure 1e). Thus, after the refractory period during which ectopic sighs could not be elicited, this inverse relationship between photostimulation phase and sigh latency indicated that acute activation of the peptidergic microcircuit was more effective as the sigh phase advanced.

To understand how the activity of peptide-expressing pF neurons results in the generation of sighs, we next examined expression pattern of Nmbr- and Grpr-expressing cells in the preBötC using fluorescence in situ hybridization (FISH; Figure 2). The vast majority of Nmbr and Grpr cells (>97%) did not colocalize with the specific astrocytic marker Aldh1l1, indicating that these cells are neurons (Figure 2—figure supplement 1). Grpr neurons were present caudal to pF, in a stream continuing dorsocaudally from pF to preBötC. In contrast, with the exception of sparse Nmbr/ChAT neurons in nucleus ambiguus (Amb. ~4 neurons in each section) that were excluded from subsequent analysis, Nmbr neurons were mainly found in preBötC. As previously reported (Li et al., 2016), some neurons coexpressed Nmbr and Grpr (Figure 2b). In our sample, 57% of Nmbr neurons coexpressed Grpr, which constituted 32% of the Grpr population (Figure 2b and c).

Figure 2 with 1 supplement see all

Download asset Open asset

Next, we assessed colocalization with the vesicular glutamate transporter VGlut2 to determine whether Nmbr and Grpr neurons are glutamatergic. Approximately 85% of Nmbr and Grpr neurons co-expressed VGlut2 (Figure 2b and c). Finally, we asked whether Nmbr and Grpr were expressed on preBötC Sst neurons. We found scant colocalization of Sst and Nmbr or Grpr (~10%), indicating that the substantial majority of these peptide receptors are not on Sst neurons (Figure 2b and c).

Activation of Nmb- or Grp-expressing pF neurons generates sufficient input to preBötC to generate ectopic sighs, but is activation of neuropeptide-receptor expressing preBötC neurons via their cognate receptors necessary for sigh production? We tested whether bypassing the peptidergic receptors and directly activating preBötC Grpr- or Nmbr-expressing neurons could also generate sighs. To do this, we expressed ChR2 or the excitatory DREADD hM3Dq in discrete subsets of preBötC neurons (see Methods).

To express ChR2 in preBötC GRPR neurons, we microinjected a Flp-dependent ChR2 virus Fenno et al., 2014 into the preBötC of Grpr^Flp mice (Figure 3—figure supplement 1a). While no ectopic sighs could be generated by bilateral preBötC SPP in these mice (phase: 0.0–1.0), preBötC LPP during 0.2–0.9 sigh phases elicited ectopic sighs which appeared as partial doublet or augmented breaths, and reset the sigh cycle (Figure 3a, bottom). Ectopic sighs elicited by preBötC LPP of GRPR neurons were no different from spontaneous sighs in V_T and T_E (Figure 3b) and exhibited a postsigh apnea (T_E following evoked sighs increased to 194 ± 22%, Figure 3b).

Figure 3 with 3 supplements see all

Download asset Open asset

Similarly, in Nmbr-ChR2 mice (Figure 3—figure supplement 1b), preBötC SPP (phase: 0.0–1.0) had no effect on sighing, while LPP during 0.4–0.9 sigh phases elicited ectopic sighs which appeared as a partial doublet or an augmented breath, and reset the sigh cycle (Figure 3c, bottom). V_T and postsigh T_E of ectopic sighs were no different from that of spontaneous sighs (Figure 3d). Ectopic sighs elicited by preBötC LPP of NMBR neurons exhibited a postsigh apnea (postsigh T_E increased to 225 ± 33%, Figure 3d).

As with photostimulation of pF NMB and GRP neurons, there was a postsigh refractory period during which we could not generate ectopic sighs in preBötC NMBR or GRPR neurons. Again, the latency from LPP to ectopic sigh onset was inversely related to phase (Figure 3e).

We also tested whether activation of preBötC NMBR neurons using an excitatory DREADD could induce sighing. To do this, we transduced preBötC Nmbr-expressing neurons by microinjection of Cre-dependent AAV-hM3Dq in Nmbr^Cre mice (Figure 3—figure supplement 1c). In these mice, application of the selective agonist clozapine-n-oxide (CNO) to the brainstem surface increased sigh frequency ~threefold (Figure 3f; see Figure 3—figure supplement 2 for control CNO application). Sighs induced by CNO in Nmbr^Cre mice were of the doublet shape. Thus, increasing the excitability of preBötC NMBR neurons increases sighing.

We next asked: what are the electrophysiological properties of the Grpr- and Nmbr-expressing preBötC neurons? We took several approaches. First, using inspiratory rhythmic slices Smith et al., 1991 from neonatal Grpr-EGFP or Nmbr-tdTomato reporter mice, we obtained whole-cell patch-clamp recordings from preBötC GRPR or NMBR neurons while monitoring hypoglossal nerve (XII) inspiratory activity. In Grpr-EGFP mice, only 3 of 21 preBötC GRPR neurons were inspiratory-modulated, i.e., fired during XII inspiratory bursts. In contrast, in Nmbr-tdTomato mice, the majority of preBötC NMBR neurons (10/16) were inspiratory-modulated (Figure 4a). This finding was confirmed using two-photon calcium imaging of rhythmic slices from Nmbr-GCaMP6f mice, in which we could monitor the activity of up to 10 neurons simultaneously (Figure 4b). The majority of preBötC NMBR neurons (35/58; n=6) exhibited inspiratory-modulated Ca²⁺ oscillations coincident with XII bursts (Figure 4b); however, episodes of non-breathing-modulated episodic bursting, i.e., bursts not in phase with XII, were also seen (Figure 4c) and varied in frequency from 0.025 to 0.1 Hz (Figure 4—figure supplement 1). Almost all NMBR neurons that had Ca²⁺ elevations during inspiratory bursts also had elevations during sighs (34/35), but not during the pre-I period, consistent with whole-cell recordings (Figure 4a), indicating that they are likely Type II, i.e., presumptive non-rhythm generating, neurons (Gray et al., 1999; Figure 4d). The burst duration of the GCaMP6f signal associated with sighs was longer-lasting than those associated with eupneic events in the same neuron (Figure 4d and e).

Figure 4 with 1 supplement see all

Download asset Open asset

In typical recording conditions in vitro ([K⁺]_ACSF = 9 mM), the preBötC population activity burst and the periodic XII discharges are in sync. Under conditions of low excitability ([K⁺]_ACSF <7 mM), however, in addition to these coincident high-amplitude bursts in preBötC and XII, there are low-amplitude burstlets in preBötC but not in XII (Smith et al., 1991). We have hypothesized that the burstlets represent the rhythmogenic kernel (Smith et al., 1991; Feldman and Kam, 2015; Ashhad and Feldman, 2020). NMBR neurons had a Type II firing pattern, i.e., no preinspiratory spiking, suggesting that they are not rhythmogenic.

We reliably observed both burstlets and bursts in population activity when we lowered [K⁺]_ACSF to 6 mM (Shi et al., 2017). When we did so, NMBR neurons were still active during bursts but never during burstlets (Figure 4f). When we applied NMB (30 nM), only bursts were observed, while both burst and sigh frequency in both preBötC and XII increased (Figure 4g). The duration of sighs and eupneic bursts were similar in [K⁺]_ACSF = 6 mM vs 9 mM (Figure 4h), while the percentage of sighs increased by 30% during NMB (30 nM) application in 6 mM K⁺ (Figure 4i), similar to the increase elicited by 30 nM NMB at 9 mM K^{+ 4}.

Photoactivation of preBötC GRPR or NMBR neurons can induce sighs. Given that there are three bombesin-related peptide receptor-expressing subpopulations in preBötC, i.e., neurons that express Nmbr but not Grpr (NMBR-only), neurons that express Grpr but not Nmbr (GRPR-only) and neurons that express both Nmbr and Grpr, does activation of neurons expressing one but not the other of these receptors affect sighs differently? We used an intersectional genetic approach to target neurons that express only one of these receptors, excluding double-positive (GRPR/NMBR) neurons (Fenno et al., 2020). Thus, using Cre-on/Flp-off and Cre-off/Flp-on viral vectors for the delivery of ChR2 in Nmbr^Cre;Grpr^Flp double mutant mice (Figure 5—figure supplement 1), we expressed ChR2 in preBötC NMBR-only (Figure 5a) or GRPR-only neurons (Figure 5d), respectively. We examined the effect of preBötC LPP and SPP of these neurons on sighing. When targeting NMBR-only neurons, preBötC SPP elicited ectopic bursts which appeared similar to partial doublets or augmented breaths with smaller V_T compared to spontaneous sighs due to a decrease in T_I to 64 ± 8% (Figure 5c). T_E immediately following these ectopic bursts were longer than the T_E of eupneic breaths (Figure 5c). Note that the time between the SPP-induced ectopic burst and the next endogenous sigh was shorter than the time between endogenous sighs, indicating that SPP-induced burst did not reset the sigh cycle (Figure 5b, top); SPP-induced ectopic burst could also occur in a sigh refractory period right after the previous eupneic sigh (Figure 5b, bottom). When targeting GRPR-only neurons, while ectopic sighs could not be generated by SPP at any phase of the sigh cycle, preBötC LPP in the late phase of the sigh cycle elicited ectopic sighs with a partial doublet or augmented breath shape and reset the sigh cycle (Figure 5e). T_I, T_E, and V_T of LPP-induced sighs were indistinguishable from spontaneous sighs (Figure 5f). The ectopic sighs exhibited a postsigh apnea, T_E after evoked sighs increased to 171 ± 12% of the T_E of eupneic breaths (Figure 5f).

Figure 5 with 1 supplement see all

Download asset Open asset

These data suggest that GRPR and NMBR neurons play different overlapping roles in sigh generation.

In addition to generating sighs, activation of preBötC NMBR and/or GRPR neurons also affected eupneic breathing frequency (f) and V_T. In Grpr^Flp mice expressing ChR2 in preBötC, LPP during the sigh refractory period did not generate ectopic sighs but increased f to 131 ± 9% of baseline (Figure 6a), and increased V_T to 117 ± 4% of baseline (Figure 6a), however, when excluding double-positive (GRPR/NMBR) neurons, preBötC SPP or LPP of GRPR-only neurons did not change the inspiratory burst amplitude in stimulus cycles (Figure 6b).

Figure 6 with 1 supplement see all

Download asset Open asset

In contrast, in Nmbr-ChR2 mice, preBötC LPP delivered during a postsigh refractory period did not generate ectopic sighs but decreased f to 83 ± 4% (Figure 6a) and increased V_T to 140 ± 12% (Figure 6a) of baseline. preBötC SPP or LPP of NMBR-only neurons increased V_T to 274 ± 41% or 264 ± 17% of control, respectively (Figure 6b).

The effect on inspiratory burst amplitude was also observed in Nmbr^Cre mice that had preBötC NMBR neurons transfected with Cre-dependent AAV-HM3Dq; CNO applied to the brainstem surface significantly increased V_T by 343 ± 56% (Figure 6c) and decreased f by 63 ± 8.4% (Figure 6c). Note that although such breaths were of the same V_T as eupneic sighs before administration of CNO (Figure 3c), in line with our definition of sighs (Appendix 1—figure 1), we classify this outcome as large amplitude breaths rather than ‘all sigh’ breathing rhythm, according to our definition of sighs (Appendix 1—figure 1).

Thus, stimulation of preBötC NMBR and/or GRPR neurons also has a profound effect on the eupneic, nonsigh, breathing pattern, and GRPR and NMBR neurons appear to differently affect inspiratory burst amplitude.

Expression of Grpr and/or Nmbr defines subsets of preBötC neurons that affect sigh generation. Is activation of these peptide receptors necessary for sigh generation?

For that, we investigated the effects of activating peptide-expressing pF neurons in the presence of bombesin receptor antagonists. First, we confirmed that bilateral microinjection of the NMBR antagonist BIM23042 and GRPR antagonist RC3095 (300 µM each, 50 nl/side) into the preBötC in anesthetized mice eliminates spontaneous sighs (Figure 6—figure supplement 1). Next, we determined the effects of such NMBR and GRPR antagonism on LPP-evoked sighs. After blockade, no sighs were elicited by pF LPP in Nmb-ChR2 mice; in contrast, pF LPP of GRP neurons still elicited ectopic sighs of a doublet shape. The V_T was not significantly different from that of spontaneous sighs; T_E of the following respiratory cycle was unaffected (Figure 6d). Thus, when pF NMB neurons were stimulated, preBötC NMBR neurons require NMBR activation to contribute to sigh generation, whereas when pF GRP neurons were stimulated, sighs could still be generated after antagonism of preBötC GRPR receptors.

We next investigated whether explicit activation of preBötC NMBRs and/or GRPRs is necessary to produce sighs, i.e., could sighs be generated simply by depolarizing these preBötC neurons to fire action potentials? In ChR2-transduced Grpr^Flp or Nmbr-ChR2 mice after antagonism of both receptors sufficient to eliminate spontaneous sighs (Figure 6—figure supplement 1), LPP of preBötC GRPR or NMBR neurons still elicited sighs of doublet shape. V_T and T_I of the sighs elicited in Grpr^Flp or Nmbr-ChR2 mice after blockade of both receptors was no different from spontaneous sighs (Figure 6d). Postsigh apnea was observed in Grpr^Flp (T_E immediately following the sigh increased to 132 ± 24%), but not in Nmbr-ChR2 mice (Figure 6d).

Thus, initiating or increasing the activity of pF GRP neurons, or preBötC GRPR or NMBR neurons, can generate sighs without activation of these receptors, either endogenously or exogenously.

Can activity of preBötC neurons other than those expressing Grpr or Nmbr generate sighs? To address this question, we virally expressed ChR2, hM3Dq, or the PSAM4 (Figure 7—figure supplement 1) in preBötC SST neurons. In ChR2-transduced Sst^Cre mice, bilateral preBötC SPP during inspiration elicited ectopic sighs of partial doublets or augmented breath shapes and reset the sigh cycle (Figure 7b). V_T, T_I, and T_E of the induced sighs were no different from those of spontaneous sighs (Figure 7c). The ectopic sighs elicited by photostimulation of preBötC SST neurons exhibited a postsigh apnea, T_E of the following respiratory cycle increased by 210 ± 19% (Figure 7c). After BIM23042 and RC3095 microinjection (300 µM each, 50 nl/side), spontaneous sighs were eliminated (Figure 6—figure supplement 1). In the presence of antagonists, preBötC SPP during inspiration could still elicit an ectopic sigh (Figure 7d). V_T and T_I of ectopic sighs elicited after blockade of both receptors were no different from spontaneous sighs, or from ectopic sighs elicited before blockade (Figure 7c). Ectopic sighs elicited after blockade also had a postsigh apnea, T_E of the following respiratory cycle increased by 236 ± 49% (Figure 7c).

Figure 7 with 2 supplements see all

Download asset Open asset

In hM3Dq-transduced Sst^Cre anesthetized mice (Figure 7e), excitation of preBötC SST neurons by CNO application onto ventral brainstem surface affected eupneic breathing: f decreased by 60 ± 6.2%, V_T increased by 222 ± 21%. CNO application increased sigh rate ~10 fold with sighs appearing as doublets (Figure 7f). In these mice, subsequent microinjection of BIM23042 and RC3095 into preBötC did not affect f, V_T, or sigh rate induced by CNO administration (Figure 7f). Thus, excitation of preBötC SST neurons can produce sighs in an NMBR- and GRPR-independent manner.

Next, we investigated whether sighs induced by the NMBR- and GRPR-dependent pathway require activity of preBötC SST neurons. For that, we transfected preBötC SST neurons with Cre-dependent ultrapotent inhibitory DREADD AAV-PSAM4-GlyR Magnus et al., 2019 in Sst^Cre mice (Figure 7g). The sigh rate was then increased by microinjection of peptides NMB and GRP (250 µM each, 50 nl/side) into preBötC (Figure 7g and h) in ketamine/xylazine anesthetized mice, which also decreased f, and increased V_T (Figure 7g and h). Subsequent application of the PSAM4-GlyR ligand uPSEM817 to the brainstem surface did not affect f or V_T, but completely eliminated all sighs (Figure 7g and h). See Figure 7—figure supplement 2 for control applications of uPSEM817 and saline. Note that similar to excitatory DREADD effects in Nmbr^Cre mice, manipulations increasing V_T of all breaths were classified to affect V_T of eupneic breaths rather than induce an ‘all sigh’ rhythm, according to our definition of sighs (Appendix 1—figure 1).

Grp^Cre was obtained from Jackson Labs (Strain #033174). Grpr^flp, Nmb^Cre, and Nmbr^Cre animals were designed and generated by Cyagen Biosciences Inc (see below). Transgenic ChR2 mice were generated by crossing Cre mice with floxed-ChR2-tdTomato mice (Jackson Labs Strain #012567). These crosses generated mice expressing ChR2 in GRP neurons (Grp-ChR2), NMB neurons (Nmb-ChR2), or NMBR neurons (Nmbr-ChR2). To express ChR2 selectively in preBötC neurons, AAV injections were performed on adult Sst^Cre mice (Jackson Labs Strain #013044) or Grpr^flp mice. All in vivo experiments were performed on adult male or female mice (10–14 wk old, 24–32 g). Unless otherwise specified, mice were anesthetized with ketamine/xylazine. For in vitro experiments, transgenic Grpr-EGFP, Nmbr-tdTomato, and Nmbr-GCamp6s were used. These mice were generated by crosses between Nmbr^Cre or Grpr^Flp mice and one of the following lines: Ai14 (Jackson Labs Strain #007914), RCE:FRT (MMRRC Strain #032038-JAX), GCaMp6f (Jackson Labs Strain #029626).

To generate Nmb^Cre mice the open reading frame of Cre recombinase together with SV40 polyA signal was cloned downstream of the mouse Nmb promoter, such that the expression of Cre mimicked the endogenous mouse Nmb gene. The PiggyBac ITRs was inserted into the BAC backbone flanking the genomic insert, to facilitate transposes mediated BAC integration. The modified BAC was co-injected with transposes into single-cell stage fertilized eggs from C57BL/6 mice. Resulting pups were then genotyped by PCR for the presence of the modified BAC and the Cre cassette. To generate Grpr^Flp mice, the Grpr gene was first located on mouse chromosome X. Three exons have been identified, with the ATG start codon in exon 1 and TAG stop codon in exon 3; the TAG stop codon was then replaced with the Flp cassette. To engineer the targeting vector, homology arms were generated by PCR using BAC clone RP23-402H23 and RP23-50M18 from the C57BL/6 J library as template. In the targeting vector, the Neo cassette was flanked by FRT sites. DTA was used for negative selection. The constitutive knock-in allele was obtained after FRT-mediated recombination. C57BL/6 ES cells were then used for gene targeting to introduce knock-in alleles into host embryos followed by transfer into surrogate mothers. Resulting pups were then genotyped by PCR for the presence of the Flp cassette. To generate Nmbr^Cre mice, the Nmbr gene was first located on mouse chromosome 10. Three exons have been identified, with the ATG start codon in exon 1 and TGA stop codon in exon 3; the TGA stop codon was then replaced with the Cre cassette. To engineer the targeting vector, homology arms were generated by PCR using BAC clone RP24-232I16 and RP24-124K10 from the C57BL/6 J library as template. In the targeting vector, the Neo cassette was flanked by LoxP sites. DTA was used for negative selection. The constitutive knock-in allele was obtained after Cre-mediated recombination. C57BL/6 ES cells were then used for gene targeting to introduce knock-in alleles into host embryos followed by transfer into surrogate mothers. Resulting pups were then genotyped by PCR for the presence of the Cre cassette.

pAAV-hSyn Con/Foff hChR2(H134R)-EYFP (Addgene plasmid # 55646; http://n2t.net/addgene:55646; RRID:Addgene_55646) and pAAV-hSyn Coff/Fon hChR2(H134R)-EYFP (Addgene plasmid # 55648; http://n2t.net/addgene:55648; RRID:Addgene_55648) were gifts from Karl Deisseroth (Fenno et al., 2020). pAAV-hSyn-DIO-hM3D(Gq)-mCherry was a gift from Bryan Roth (Addgene plasmid # 44361; http://n2t.net/addgene:44361; RRID:Addgene_44361). pAAV-Syn-flex-PSAM4-GlyR-IRES-EGFP was a gift from Scott Sternson (Magnus et al., 2019) (Addgene plasmid # 119741; http://n2t.net/addgene:119741; RRID:Addgene_119741). All viruses were stored in aliquots at –80 °C until use.

Mice were anesthetized with isoflurane (4% for induction and 2% for maintenance) and placed in a stereotaxic apparatus (David Kopf Instruments) with Bregma and Lambda skull landmarks level. Two holes were drilled in the skull 6.80 mm caudal to Bregma and 1.2 mm lateral to the midline. Virus was delivered 4.65 mm from the dorsal surface of the brain into the preBötC through a glass pipette using either a pressure ejection system (Picospritzer II; Parker Hannafin) or Micro4 microinjection system (World Precision Instruments). For ChR2 experiments, we used 100–200 nl of virus solution (1–6 × 10¹² vg/ml) per side, and for DREADD experiments, 30–50 nl of virus solution (4.6 × 10¹² vg/ml) per side. The pipettes were left in place for 5 min after injection to minimize backflow. The wound was closed with 5–0 gauge non-absorbable sutures. Mice were returned to their home cage and given 2–3 wk to recover to allow for sufficient levels of protein expression. All virus injection sites were subsequently confirmed by immunostaining and only mice with sufficient and localized transfection in preBötC were used for all analysis.

Adult mice were anesthetized via intraperitoneal injection of ketamine and xylazine (100 and 10 mg/kg, respectively). Isoflurane (1–2% volume in air) was administered throughout an experiment. The level of anesthesia was assessed by the suppression of the withdrawal reflex. A tracheostomy tube was placed in the trachea through the larynx, and respiratory flow was detected with a flow head connected to a transducer to measure airflow (Mouser Electronics). The mice were placed in a supine position in a stereotaxic instrument (David Kopf Instruments). The larynx was denervated, separated from the pharynx, and moved aside. The basal aspect of the occipital bone was removed to expose the ventral aspect of the medulla. The canal of the hypoglossal nerve (XII) served as a suitable landmark. The preBötC was 0.15 mm caudal to the hypoglossal canal, 1.2 mm lateral to the midline, and 0.24 mm dorsal to the ventral medullary surface.

A 473 nm laser (OptoDuet Laser; IkeCool) was targeted bilaterally to the preBötC or pF with a branching fiber patch cord (200 µm diameter; Doric Lenses) brought to the exposed ventral medullary surface. Laser power was set at 5 mW. Short Pulse Photostimulation (SPP; 200 ms) and Long Pulse Photostimulation (LPP; 5 s) waveforms were delivered under the command of a pulse generator (Pulsemaster A300 Generator; WPI) connected to the laser power supply. In instances where breathing frequency was decreased, e.g., after BIM23042 and RC3095 microinjections, SPP and LPP of longer durations (500 ms and 8 s, respectively) were used to confirm consistency of results. Stimulating 500 µm rostral to preBötC or pF in mice expressing ChR2 that served as a control did not produce significant output effects (Figure 1—figure supplement 2 & Figure 3—figure supplement 3). No output effects and no sighs were produced by preBötC or pF photostimulation in Grp-RFP, Grpr-EGFP, Nmbr-EGFP, Nmb-RFP reporter mice (Figure 1—figure supplement 2 & Figure 3—figure supplement 3).

The NMBR antagonist BIM23042 (Tocris Bioscience) and the GRPR antagonist RC3095 (Sigma-Aldrich) were injected together (300 µM each, 50–60 nl/side) to block NMBR and GRPR activation. Injections were made using micropipettes (~40 µm tip), placed bilaterally into the preBötC. Injections targeted to the center of the preBötC were placed 0.15 mm caudal to the hypoglossal canal, 1.2 mm lateral to the midline, and 0.24 mm dorsal to the ventral medullary surface (Cui et al., 2016). Small corrections were made to avoid puncturing of blood vessels on the surface of the medulla. All injections were made using a series of pressure pulses (Picospritzer, Parker-Hannifin; or Micro4, World Precision Instruments). For activation of hM3Dq receptors, 50 µL of CNO solution was applied to the ventral surface of the brainstem. CNO (Tocris Biosciences) was dissolved in saline at a concentration of 1 mg/ml with 0.5% dimethylsulfoxide. Solution was prepared fresh daily. For activation of PSAM4-GlyR receptors, 30 µL of uPSEM817 tartrate solution (10 mM) was applied to the ventral surface of the brainstem. In chemogenetic experiments CNO or uPSEM817 were applied to the ventral surface of the brainstem instead of microinjection in order to avoid preBötC damage from multiple microinjections; in control experiments application of the same dose of CNO or uPSEM817 does not produce responses in any breathing parameters.

Nmbr^Cre, Sst^Cre, or wild-type mice (10–14 wk) were euthanized with isoflurane overdose, their brainstems were rapidly removed and flash-frozen in dry ice. Fresh frozen brainstems were sectioned sagittally on a cryostat (CryoStar NX70, Thermo Scientific), mounted on SuperFrost Plus slides (Fisher Scientific) and stored at –80 °C for at least 1 d. They were then processed according to the manufacturer’s protocol (RNAscope version 1, Advanced Cell Diagnostics). Briefly, tissue samples were postfixed in 10% neutral buffered paraformaldehyde, washed, and dehydrated in sequential concentrations of ethanol (50, 70, and 100%). Samples were treated with protease IV and incubated for 2 hr at 40 °C in the HybEZ Hybridization Oven (Advanced Cell Diagnostics) in the presence of target probes. We used combinations of the following probes for hybridization: Mm-ChAT, Mm-Grpr, Mm-Nmbr, Mm-Sst, Mm-Slc17a6 (VGluT2), Mm-Aldh1l1, Cre, and Flp. A probe for Mm-ChAT was used in all probe combinations to mark the location of nucleus ambiguus, which is necessary for determining the location of preBötC. After a 4-step amplification process, samples were counterstained with DAPI and coverslipped with ProLong Gold (Invitrogen) used as a mounting agent. Images were acquired on a confocal laser scanning microscope (LSM710 META, Zeiss). High-resolution z-stack confocal images were taken at 1 µm intervals and then merged using ImageJ software.

Immunohistochemistry was performed according to the following protocol. Free-floating sections were rinsed in PBS and incubated with 10% normal donkey antiserum (NDS) and 0.2% Triton X-100 in PBS for 60 min to reduce nonspecific staining and increase antibody penetration. Sections were incubated overnight with primary antibodies diluted in PBS containing 1% NDS and 0.2% Triton X-100. The following day, sections were washed in PBS, incubated with the specific secondary antibodies conjugated to the fluorescent probes diluted in PBS for 2 hr. Sections were further washed in PBS, mounted, and coverslipped with Fluorsave mounting medium (Millipore). The primary antibodies used for this study were as follows: rabbit polyclonal anti-somatostatin-14 (1:500; Peninsula Laboratories), mouse monoclonal anti-NeuN (1:500; Millipore; MAB377), goat anti-ChaT (1:500, Chemicon), and chicken polyclonal anti-GFP (1:500; Aves Labs). DyLight488 donkey anti-chicken, Rhodamine Red-X donkey anti-rabbit, and Cy5 donkey anti-mouse conjugated secondary antibodies (1:250; Jackson ImmunoResearch) were used to detect primary antibodies. Slides were observed under an AxioCam2 Zeiss fluorescent microscope connected with AxioVision acquisition software or under a LSM510 Zeiss confocal microscope with Zen software (Carl Zeiss). Images were acquired, exported in TIFF files, and arranged to prepare final figures in Zen software (Carl Zeiss) and Adobe Photoshop (Adobe).

Neonatal mice (P0-5) of either sex were anesthetized with isoflurane. The brainstem was isolated from the pons to the rostral cervical spinal cord under cold ACSF containing (in mM): 124 NaCl, 3 KCl, 1.5 CaCl₂, 1 MgSO₄, 25 NaHCO₃, 0.5 NaH₂PO₄, and 30 D-glucose; a single transverse slice (600 μm) was cut on a vibratome (Leica VT1200S).

Slices were placed in a recording chamber with the rostral face of preBötC facing up and continuously superfused with warmed ACSF (at 28–30°C) equilibrated with 95% O₂ and 5% CO₂, and extracellular K⁺ was increased to 9 mM to elevate preBötC excitability. Inspiratory-related motor output was recorded from the hypoglossal nerve (XII) using a suction electrode and a differential AC amplifier (AM systems), filtered at 2–4 kHz, integrated, and digitized at 10 kHz. Whole-cell recordings were made using a MultiClamp 700 A (Molecular Devices), filtered at 2–4 kHz, and digitized at 10 kHz. Borosilicate glass recording electrodes (G150-4, Warner Instruments) had tip resistances of 4–8 MΩ and were filled with intracellular solution containing (in mM): 140 potassium gluconate, 10 HEPES, 5 NaCl, 1.1 EGTA, 2 Mg-ATP, and 0.1 CaCl₂ (~305 mOsm; pH 7.3). In two-photon imaging experiments, field potentials were recorded from the contralateral preBötC, using an extracellular electrode and a differential AC amplifier (AM systems), filtered at 2–4 kHz, integrated, and digitized at 10 kHz. Digitized data were analyzed offline using custom procedures written for IgorPro (Wavemetrics). Agonists were bath applied at the specified concentrations while monitoring field potentials and XII output.

Intracellular Ca²⁺ was imaged using a multi-photon resonant scanner (3i) and a microscope equipped with a water immersion 20 x, 1.0 objective. Illumination was provided by a laser with a power output of 1050 mW at 970 nm (Coherent Chameleon UltraI). Identified preBötC neurons were scanned at 10–39 Hz and fluorescence data were collected using 3i software and analyzed using SlideBook, Image J, and IgorPro. Regions of interest (ROIs) were manually detected and Ca²⁺ transients were plotted as F/F₀, where F₀ is the average fluorescence intensity of all pixels within a given ROI averaged over the entire time series.

In vitro data were analyzed using IgorPro (Wavemetrics) and GraphPad Prism (GraphPad Software) was used for statistical analysis. Paired T-tests or Kruskal-Wallis one-way ANOVA, adjusted for multiple comparisons, was used to calculate significance.

For in vivo experiments, data were recorded on a computer using LabChart 7 Pro (AD Instruments) and analyzed using LabChart 7 Pro (ADInstruments), Excel, and Igor Pro (Wavemetrics, Inc) software. The flow signal was high-pass filtered (>0.1 Hz) to eliminate DC shifts and slow drifts, and was digitally integrated with a time constant of 0.05 s to calculate breathing frequency, period, V_T, T_I, and T_E during eupnea or sigh (Appendix 1—figure 1d). Breathing frequency is measured by the time interval between peaks. T_I was calculated by the time interval from start to the maximum of the peak, T_E was calculated by the time interval from the previous inspiratory peak to the onset of the next inspiratory burst and V_T was calculated from the inspiratory peak. Due to the high reproducibility of effects of photostimulation, V_T, T_I, and T_E during ectopic sighs were calculated as an average of 3–5 sighs, paired t-tests were used to determine statistical significance of changes before and after pharmacological injections or photostimulation. For statistical comparisons of more than two groups, repeated-measures (RM) ANOVAs were performed. For one-way RM ANOVAs, post hoc significance for pairwise comparisons was analyzed using the Holm-Sidak method. Significance was set at p<0.05. Data are shown as mean ± SE.

Newly created materials (i.e. Grpr^flp, Nmb^Cre, and Nmbr^Cre mice) are currently maintained at the UCLA animal facility and are available from the corresponding author, JLF, upon reasonable request.

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

This work was funded by NIH Grants HL135779, NS72211 and NSFC Grant 37101012.

The study and experimental protocols were in accordance with the guidelines approved by the University of California, Los Angeles (UCLA) Institutional Animal Care and Use Committee and Animal Research Committee (#1994-159-83 and #2009-107).

1. Sent for peer review: June 5, 2024
2. Preprint posted: June 9, 2024
3. Reviewed Preprint version 1: August 7, 2024
4. Reviewed Preprint version 2: January 3, 2025
5. Version of Record published: June 24, 2025

You can cite all versions using the DOI https://doi.org/10.7554/eLife.100192. This DOI represents all versions, and will always resolve to the latest one.

© 2024, Cui, Bondarenko et al.

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.