Fentanyl-Induced Respiratory Depression and Locomotor Hyperactivity Are Mediated by μ-Opioid Receptors Expressed in Somatostatin-Negative Neurons

Abstract Opioid drugs are widely used as analgesics but cause respiratory depression, a potentially lethal side effect with overdose, by acting on μ-opioid receptors (MORs) expressed in brainstem regions involved in the control of breathing. Although many brainstem regions have been shown to regulate opioid-induced respiratory depression, the types of neurons involved have not been identified. Somatostatin is a major neuropeptide found in brainstem circuits regulating breathing, but it is unknown whether somatostatin-expressing circuits regulate respiratory depression by opioids. We examined the coexpression of Sst (gene encoding somatostatin) and Oprm1 (gene encoding MORs) mRNAs in brainstem regions involved in respiratory depression. Interestingly, Oprm1 mRNA expression was found in the majority (>50%) of Sst-expressing cells in the preBötzinger Complex, the nucleus tractus solitarius, the nucleus ambiguus, and the Kölliker–Fuse nucleus. We then compared respiratory responses to fentanyl between wild-type and Oprm1 full knock-out mice and found that the lack of MORs prevented respiratory rate depression from occurring. Next, using transgenic knock-out mice lacking functional MORs specifically in Sst-expressing cells, we compared respiratory responses to fentanyl between control and the conditional knock-out mice. We found that respiratory rate depression by fentanyl was preserved when MORs were deleted only in Sst-expressing cells. Our results show that despite coexpression of Sst and Oprm1 in respiratory circuits and the importance of somatostatin-expressing cells in the regulation of breathing, these cells do not mediate opioid-induced respiratory rate depression. Instead, MORs found in respiratory cell populations other than Sst-expressing cells likely contribute to the respiratory effects of fentanyl.


Introduction
Opioid drugs are widely used as pain medications because of their potent analgesic effects. Opioid drugs act on m-opioid receptors (MORs) in spinal and brainstem regions involved in pain, but also on regions regulating breathing (Stein, 2016;Montandon, 2022). Through their actions on MORs in respiratory circuits, opioid drugs can cause respiratory depression, characterized by slow and shallow breathing, with severe cases involving inhibition of the hypoxic and hypercapnic ventilatory responses and even death with overdose (Dahan et al., 2001;Ramirez et al., 2021). To identify new therapeutic strategies aimed at reducing the side effects of opioid drugs, a better understanding of the mechanisms and respiratory circuits underlying opioid-induced respiratory depression is needed. Many brainstem regions contribute to the respiratory side effects of opioids (Palkovic et al., 2020;Ramirez et al., 2021) including the preBötzinger Complex (preBötC; Montandon et al., 2011;Bachmutsky et al., 2020;Varga et al., 2020a), an inspiratory rhythm-generating site (Smith et al., 1991), the Kölliker-Fuse (KF) nucleus/parabrachial nucleus (PBN; Prkic et al., 2012;Levitt et al., 2015;Miller et al., 2017;Saunders and Levitt, 2020), and the rostral ventromedial medulla (Phillips et al., 2012). The caudal medullary raphe nuclei, including the raphe obscurus (ROb) and raphe pallidus (RPa; Zhang et al., 2007;Phillips et al., 2012;Palkovic et al., 2022) and the nucleus tractus solitarius (NTS; Zhang et al., 2011;Zhuang et al., 2017) may also mediate the severity of opioid-induced respiratory depression because of changes in chemosensory respiratory responses. As well, the nucleus ambiguus (NA) may contribute to changes in the motor control of upper airway muscles (Bieger and Hopkins, 1987;Pascual-Font et al., 2011) associated with opioid drugs (Hassen et al., 1984). Although various brain regions are modulated by opioid drugs and mediate respiratory depression, the identity of the cells mediating these effects is not known.
Brainstem respiratory circuits are identified by various neurotransmitters and neuromodulators. For instance, preBötC neurons express the G-protein-coupled receptor neurokinin-1 receptor and the neuropeptide somatostatin (Stornetta et al., 2003). Although somatostatin is expressed throughout the central and peripheral nervous systems, it acts as a neuromodulator in many brainstem regions controlling breathing (Llona and Eugenín, 2005;Tan et al., 2008;Montandon et al., 2016). In the brainstem, somatostatin is expressed in the preBötC (Stornetta et al., 2003;de Sousa Abreu et al., 2022), KF (De León et al., 1992;Bou Farah et al., 2016), NTS (Helke, 1984;Johansson et al., 1984;Kalia et al., 1984), and NA cells , with little expression seen in the raphe nuclei . Somatostatin-expressing cells within the preBötC are primarily glutamatergic and involved in modulating the respiratory pattern (Cui et al., 2016;de Sousa Abreu et al., 2022), while their inhibition leads to apnea (Tan et al., 2008), suggesting a critical role of these cells in breathing. Moreover, preBötC somatostatin-expressing cells project to the KF nucleus (Yang and Feldman, 2018), both of which are involved in opioid-induced respiratory depression (Bachmutsky et al., 2020;Varga et al., 2020a). Together these findings suggest that somatostatin-expressing cells are critical for normal respiration and are found within brainstem sites involved in the control of breathing and opioid-induced respiratory depression. We hypothesized that somatostatin and MORs are coexpressed in respiratory circuits and that somatostatin-expressing cells contribute to the regulation of opioid-induced respiratory depression.
Using in situ hybridization, we first quantified the expression of Sst (gene encoding somatostatin) and Oprm1 (gene encoding MORs) mRNAs in brainstem regions involved in the control of breathing and respiratory depression. Interestingly, Oprm1 mRNA expression was observed in the majority (.50%) of Sst mRNA-expressing cells in the preBötC, NTS, NA, and KF regions. Next, we determined the role of MORs in the regulation of respiratory depression by opioid drugs by assessing the respiratory responses to fentanyl in wild-type and full MOR knock-out mice (Oprm1 À/À ; Matthes et al., 1996). We then generated conditional transgenic knock-out mice that lacked functional MORs only in Sst-expressing cells (Sst-Oprm1 À/À ). Locomotor responses to fentanyl were also quantified since opioid drugs, such as highly potent fentanyl, have been shown to increase movement in mice, an effect that can impact respiration (Murphy et al., 2001;Contarino et al., 2002;Smith et al., 2009;Varshneya et al., 2019). We found that fentanyl induced a significant respiratory rate depression in wild-type mice, the severity of which was correlated with locomotor hyperactivity, meaning that respiratory rate decreased despite increases in locomotor movements with fentanyl. These responses were absent in Oprm1 À/À mice, suggesting that MORs fully mediate the effects of fentanyl on respiratory rate and locomotor activity. Surprisingly, mice that lacked functional MORs in Sst-expressing cells still presented a significant decline in respiratory rate following systemic fentanyl administration. Our results suggest that, despite coexpression of Sst and Oprm1 mRNAs and the role of Sst-expressing cells in the regulation of breathing, these cells are not involved in respiratory rate depression by opioid drugs. Rather, our results suggest that Sst-expressing cells involved in breathing may be spared from the effects of opioid drugs.

Animals and drugs
Mice used in the study were obtained from The Jackson Laboratory and included C57BL/6 (stock #000664), Oprm1 À/À (stock #007559; Matthes et al., 1996;Martin et al., 2003), Sst-Cre (stock #013044; Taniguchi et al., 2011), and Oprm1 fl/fl (stock #030074). A Cre-loxP recombination strategy was used to generate transgenic mice lacking functional MORs in Sst-expressing cells by crossing Sst-Cre and Oprm1 fl/fl mice to produce Sst-Oprm1 À/À mice. Custom primers were designed and used to confirm knockout of exons 2 and 3 in the Oprm1 gene of Sst-Oprm1 À/À mice using touchdown PCR (Korbie and Mattick, 2008). The Oprm1 gene in Oprm1 fl/fl mice was floxed in regions upstream and downstream of exons 2 and 3. Primers were designed to target regions flanking the floxed exon segment. Without the removal of exons 2 and 3 by Cre, the amplicon would be too large (;3000 bp) to amplify using Taq DNA polymerase when combined with a short synthesis time and therefore no band would be produced (Arezi et al., 2003). After the removal of exons 2 and 3 by Cre, the amplicon would be shorter (;470 bp), allowing for synthesis and amplification to occur. Different types of tissues were sampled in Sst-Oprm1 À/ À mice: ear notch skin tissue as a control and medullary brain tissue, which expresses Sst and Oprm1 (Mansour et al., 1994;Stornetta et al., 2003). Only tissues that coexpress Sst (and therefore Cre) and Oprm1 would have a band, indicating the removal of Oprm1 exons 2 and 3. Genotyping of Sst-Oprm1 À/ À mice confirmed the presence of Cre and loxP in ear skin and medullary brain tissue, and the deletion of Oprm1 exons 2 and 3 in medullary brain tissue, but not in ear skin tissue. Experimental animals were male, 3-4.5 months old, and weighed 20-40 g. Mice were housed with free access to food and water under a 12 h light/dark cycle (lights on at 8:00 A.M.). Fentanyl citrate (50 mg/ml) was purchased from Sandoz.

In situ hybridization
In situ hybridization was performed in Sst-Cre mice to determine expression of Sst and Oprm1 mRNAs in brainstem regions involved in respiration. Mice were perfused with PBS followed by formalin, and the brain was harvested. Brains were soaked in 20% sucrose in PBS for 24 h followed by 30% sucrose in PBS for 24 h. Fixed brains were frozen using Tissue-Tek O.C.T. Compound (Sakura) and dry ice and were stored at À80°C. Coronal sections were cut at 25 mm thickness using a cryostat (model CM3050S, Leica Biosystems). The manufacturer protocol was used to perform in situ hybridization using the RNAscope Multiplex Fluorescent Reagent version 2 Assay (ACD; Wang et al., 2012), and sections were counterstained with DAPI. Target probes used were Mm-Sst (catalog #404631-C3, ACD) targeting Sst gene mRNA and Mm-Oprm1 (catalog #315841, ACD) targeting Oprm1 gene mRNA. Tissue sections 100 mm apart were scanned using the Axio Scan.Z1 slide scanner (Zeiss).

mRNA quantification
To quantify mRNA expression, two to three sections containing each region of interest were exported from Zen (Zeiss) to Adobe Illustrator (Creative Suite 5, Adobe).
Regions where mRNA was quantified include the preBötC, NTS, ROb and RPa nuclei, NA, and KF nucleus. To identify these areas in tissue sections, The Mouse Brain in Stereotaxic Coordinates (Franklin and Paxinos, 2008) was consulted, and anatomic markers included the NTS, the NA, the facial nucleus, the hypoglossal nucleus, the cerebellum, and the external cuneate nucleus. Regions of interest were drawn, and images were exported to Fiji (ImageJ) for counting. We used the manufacturer Guide for RNAscope Data Analysis (ACD) to count and distinguish cell mRNA expression from background expression. Cells were considered to express mRNA if four or more dots (one dot represents one mRNA molecule) and/or one or more clusters of dots overlapped with or were adjacent to a DAPI-stained cell nucleus. Coexpression of Sst and Oprm1 mRNA was considered if four or more dots and/or one or more clusters of dots for both Sst and Oprm1 mRNA overlapped with or were adjacent to a DAPI-stained cell nucleus. Counts were obtained for total DAPI, Oprm1, Sst, and Oprm1 1 Sst cells, and expression was presented as percentages of total DAPI, total Oprm1, and total Sst cells.

Whole-body plethysmography
Flow-through whole-body plethysmography was used to measure respiratory activity in freely moving mice (Buxco Bias Flow Manual, DSI), according to previous studies (Montandon et al., 2006;Bachmutsky et al., 2020). Eight C57BL/6, seven Oprm1 À/À , nine Sst-Cre, and eight Sst-Oprm1 À/À mice were used. Plethysmography chambers were ventilated with a constant airflow of 0.9 L/min at room temperature and measured 21.5 cm in diameter, allowing enough room for mice to move freely during recordings. Pressure changes inside the chamber were recorded with a pressure transducer, amplified (model PS100W-2, EMKA Technologies), and digitized using PowerLab 4/26 in LabChart Version 8 (ADInstruments). Data were extracted from LabChart Version 8 (ADInstruments) and exported to Microsoft Excel for analysis. To approximate tidal volume, we calculated the area under the curve during inspiration using pressure traces. Since accurate approximation of tidal volume requires measurements of body temperature, chamber humidity, and temperature (Mortola and Frappell, 1998), which is challenging and often inaccurate in small rodents (Enhorning et al., 1998), we normalized tidal volume according to baseline values, expressed it as percentage of baseline, and labeled it relative tidal volume. We thereby assumed that temperatures and humidity did not change substantially during the experiment. In addition, considering the behavioral changes observed following saline or fentanyl injection, relative tidal volume measured with this approach includes behavioral artifacts such as sniffing and movements that may affect the reliability of tidal volume measurements. We measured respiratory rate as the number of breaths per minute. Relative minute ventilation was calculated as the product of relative tidal volume and respiratory rate.
Mice were acclimatized to the plethysmography chamber for 3 d before experiments between 10:00 A.M. and 1:00 P.M. Experiments took place over 2 d at the same time of day as the acclimatization period. During the 2 experimental days, mice were placed in the plethysmography chamber, and baseline measures were recorded. At 12:00 P.M., mice received an intraperitoneal injection of either saline or fentanyl citrate (0.3 mg/kg; Fujii et al., 2019). The volume injected to achieve a dose of 0.3 mg/ kg fentanyl was calculated to be 120 ml/20 g mouse. The injected volumes of saline and fentanyl were calculated on each respective experimental day based on the weight of the mouse obtained at 10:00 A.M. and ranged between 120 and 240 ml, depending on the weight of the mouse. Treatments were randomized between experimental days. Data for 1 h preinjection and 1 h postinjection were divided into 1 min average time bins and used to assess respiratory parameters over time. Respiratory variables were averaged for the 1 h period leading up to injection on the first experimental day, to determine baseline measures. Averages of respiratory rate, relative tidal volume, and relative minute ventilation were obtained for minutes 5-10 and 50-55 postinjection to assess representative 5 min windows during "early" and "late" phases, respectively, of the postinjection response. Representative 5 s traces of respiratory recordings were extracted beginning at minute 7 postinjection, during the peak effect of fentanyl.

Locomotor activity
Whole-body plethysmography chambers with transparent bottom platforms were mounted on a box with a 1080p high-definition camera placed facing upward at the bottom of the structure. This camera allowed for simultaneous video recording of mouse movements during respiratory recordings. Mouse movements were tracked during both experimental days for the entire 3 h duration. Videos were recorded using Pinnacle Studio 24 MultiCam Capture software (Corel), resized, and exported to EthoVision XT Version 14 (Noldus) for movement analysis. Mouse velocity was calculated for minutes 5-10 and 50-55 postinjection in Microsoft Excel, and video data were aligned with respiratory recordings using video time stamps.

Statistics
Statistical analyses were performed using GraphPad Prism 9. Data in all figures and text are represented as the mean 6 SEM with individual data points displayed. Data were tested for normality using the Shapiro-Wilk test. Normally distributed data were analyzed using an unpaired t test with Welch's correction, two-way repeatedmeasures ANOVA, or a mixed-effects model, to account for missing values. The repeated measure in this study was drug treatment (saline or fentanyl). A Sidak multiplecomparisons post hoc test was used following the twoway repeated-measures ANOVA or mixed-effects analysis. Non-normally distributed data were analyzed using the Mann-Whitney test or the Wilcoxon matched-pairs signed-rank test with the Holm-Sidak multiple-comparisons post hoc test. Pearson's correlation coefficient was used to determine the association between respiratory rate and velocity. All tests were two tailed with a significance level of a = 0.05.

Expression of Sst and Oprm1 mRNAs
To determine the expression of Sst and Oprm1 mRNAs, we performed in situ hybridization in medullary and pontine regions of control (Sst-Cre) mice (n = 3). We first analyzed sections located 6.8-7.1 mm caudal to bregma containing the preBötC, ROb, RPa, and the NTS (Fig. 1A, B). In the preBötC, 33.6 6 6.8% of DAPI-stained cells expressed Oprm1, 23.1 6 7.7% expressed Sst, and 15.2 6 4.7% coexpressed Oprm1 and Sst (Fig. 1C). In the ROb, Oprm1 mRNA was expressed in 11.0 6 6.9% of DAPI-stained cells, and Sst was expressed in 11.7 6 6.3% of DAPI-stained cells; 1.7 6 1.4% of DAPI-stained cells in the ROb coexpressed both Oprm1 and Sst. In the RPa, 7.5 6 4.5% of DAPI-stained cells expressed Oprm1, 9.8 6 4.3% expressed Sst, and 2.6 6 1.4% coexpressed Sst and Oprm1. In the NTS, Oprm1 mRNA was expressed in 68.6 6 4.5% of DAPI-stained cells, whereas Sst mRNA was expressed in 25.5 6 7.0% of DAPI-stained cells. Coexpression of both Oprm1 and Sst mRNAs in the NTS was found in 22.2 6 6.6% of DAPI-stained cells. We then looked at the NA, located 6.6-6.7 mm caudal to bregma ( Fig. 2A-C). Oprm1 was expressed in 88.5 6 2.8% of DAPI-stained cells, Sst was expressed in 33.5 6 16.1% of DAPI-stained cells, and both Oprm1 and Sst were found in 32.7 6 15.7% of DAPI-stained cells. Last, we looked at mRNA expression in the KF nucleus, 4.8-5.3 mm caudal to bregma ( Fig. 2A-C): 40.7 6 6.9% of DAPI-stained cells expressed Oprm1, 17.9 6 5.0% expressed Sst, and 11.7 6 4.1% expressed both Oprm1 and Sst. In summary, relatively high expression of Oprm1 mRNA was found in the NTS and the NA, with moderate expression in the preBötC and the KF nucleus, and low expression in the raphe nuclei. Sst mRNA was moderately expressed in the preBötC, the NTS, the NA, and the KF nucleus, with relatively low expression in the caudal raphe nuclei.
To better assess the relationship between Oprm1 and Sst expression, we looked at the relative proportion of Oprm1 1 Sst cells among Oprm1-expressing cells (Fig.  2D). In other words, how many Oprm1 cells expressed Sst mRNA? A total of 43.4 6 13.5% of Oprm1-expressing cells coexpressed Sst in the preBötC, 11.2 6 4.4% in the ROb, 21.1 6 15.0% in the RPa, 31.0 6 8.1% in the NTS, 35.4 6 17.0% in the NA, and 28.3 6 8.9% in the KF nucleus. We also looked at the relative proportion of Oprm1 1 Sst mRNAs among Sst-expressing cells ( Fig.  2E): 64.4 6 9.5% of Sst-expressing cells coexpressed Oprm1 in the preBötC, 19.1 6 14.1% in the ROb, 14.7 6 7.3% in the RPa, 84.0 6 4.1% in the NTS, 98.7 6 0.8% in the NA, and 63.2 6 9.3% in the KF nucleus. In conclusion, relatively high expression of Oprm1-expressing cells was found in the NTS and NA, moderate expression in the preBötC and the KF nucleus, and low expression in the ROb and RPa nuclei. While Sst expression was low to moderate in all regions, a majority (.50%) of Sst-expressing cells in the preBötC, the NTS, the NA, and the KF   . Opioid-induced respiratory depression in wild-type and Oprm1 À/À mice. A, Using a combination of whole-body plethysmography and video recordings, respiratory and locomotor activities were recorded in wild-type and Oprm1 À/À mice. B, On day 1, mice were placed in the chamber and respiratory rate was recorded for 2 h. An intraperitoneal injection of either saline or fentanyl (0.3 mg/kg) was administered after 2 h of baseline activity. Responses were then recorded for 1 h. On day 2, mice were placed in the chamber and recorded for 2 h followed by an intraperitoneal injection of either fentanyl (0.3 mg/kg) or saline, and respiratory activity was recorded for 1 h. C, Representative traces of respiratory responses to saline and fentanyl in wild-type and Oprm1 À/À mice. D, Relative minute ventilation following injection of saline and fentanyl in wild-type and Oprm1 À/À mice. E, Relative minute ventilation averaged over a 5 min period during the early phase and late phase of the responses to saline and fentanyl. Data are represented as the mean 6 SEM. *p , 0.05. Panel A was created with BioRender.com. nucleus coexpressed Oprm1. Interestingly, .20% of Oprm1-expressing cells in the preBötC, the NTS, and the KF nucleus coexpressed Sst. Considering the role of the preBötC and the KF nucleus in mediating opioid-induced respiratory depression (Montandon et al., 2011;Varga et al., 2020a), our findings suggest that the Sst cell populations in these regions may contribute to respiratory depression by opioid drugs.

Deletion of MORs and respiratory depression by the opioid fentanyl
To characterize the effects of fentanyl on respiratory activity, we administered a single dose of fentanyl (0.3 mg/ kg; Fujii et al., 2019) in wild-type (C57BL/6) and Oprm1 À/À knock-out mice, and measured breathing using wholebody plethysmography (Fig. 3A,B). The first 5 min following injection were omitted from all analyses to allow for the drug to take effect and for the mouse to adjust following injection. Respiratory responses to fentanyl and saline were calculated and analyzed for the early and late phases of the response. Systemic injection of fentanyl decreased relative minute ventilation compared with saline in wild-type mice, but not in Oprm1 À/À mice (Fig. 3C,D). During the early phase (minutes 5-10 postinjection), there was a significant interaction between treatment (saline or fentanyl) and genotype (p = 0.0041, mixed-effects model with Sidak multiple-comparisons test; Fig. 3E). Fentanyl induced a significant decrease in relative minute ventilation compared with saline in wild-type mice (p = 0.0022) but not in Oprm1 À/À mice (p = 0.7535). During this early phase, relative minute ventilation was significantly lower in wild-type mice compared with Oprm1 À/À mice following fentanyl injection (p = 0.0026). During the late phase (minutes 50-55 postinjection), there were no significant differences between relative minute ventilation in response to fentanyl compared with saline in wild-type mice (p . 0.9999, Wilcoxon matched-pairs signed-rank test with Holm-Sidak multiple comparisons) or Oprm1 À/À mice (p = 0.9961).
Systemic injection of fentanyl decreased respiratory rate relative to saline injection in wild-type mice but not in Oprm1 À/À mice (Fig. 4A). During the early phase, there was a significant interaction between treatment (saline or fentanyl) and genotype (p , 0.0001, mixed-effects model with Sidak multiple comparisons). Fentanyl induced a significant decline in respiratory rate relative to saline in wild-type mice (p , 0.0001) but not in Oprm1 À/À mice (p = 0.9845; Fig.  4B). During this early phase, respiratory rate was significantly greater in wild-type mice compared with Oprm1 À/À mice following saline injection (p = 0.0038), but was significantly greater in Oprm1 À/À mice compared with wild-type mice following fentanyl injection (p = 0.0005). During the late phase, there were no significant differences between respiratory rate in response to fentanyl compared with saline in wild-type mice (p . 0.9999, Wilcoxon matched-pairs signed-rank test with Holm-Sidak multiple comparisons) or Oprm1 À/À mice (p = 0.5056). No significant differences were found between baseline respiratory rates of wild-type and Oprm1 À/À mice (p = 0.7789, Mann-Whitney test).
Interestingly, relative tidal volume increased in response to fentanyl when compared with saline in wild-type mice, but not in Oprm1 À/À mice (Fig. 4C). During the early phase, there was a significant interaction between treatment (saline or fentanyl) and genotype (p = 0.0059, mixedeffects model with Sidak multiple comparisons; Fig. 4D), where fentanyl significantly increased relative tidal volume in wild-type mice (p = 0.0019) but not in Oprm1 À/À mice (p = 0.9514). Relative tidal volume was also significantly greater in wild-type mice compared with Oprm1 À/À mice in response to fentanyl (p = 0.0339). No significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.0573, mixed-effects model) during the late phase in wild-type or Oprm1 À/À mice, and no significant effects of treatment (p = 0.2740) or genotype (p = 0.1160) were found. Overall, our findings demonstrate that MORs regulate respiratory depression by fentanyl in freely behaving mice, mainly because of reductions in respiratory rate induced by fentanyl.

Deletion of MORs and locomotor activity
Opioid drugs such as fentanyl produce profound effects on locomotor activity in mice (Murphy et al., 2001;Contarino et al., 2002;Smith et al., 2009;Varshneya et al., 2019). One limitation associated with respiratory recordings in freely behaving mice is that the plethysmography recording system also captures movements and cannot distinguish between various behavioral states, which may directly impact breathing (Montandon and Horner, 2019). To better understand the effects of fentanyl on respiration, we quantified its effects on locomotion and examined the relationship between locomotor activity and respiration in wildtype and Oprm1 À/À mice (Fig. 5A). Baseline average velocities (movement of the mouse per second) showed no significant differences between wild-type and Oprm1 À/À mice (p = 0.4634, Mann-Whitney test). Representative movement traces of wild-type and Oprm1 À/À mice following injection of either saline or fentanyl are shown in Figure 5B. Velocity increased in wild-type mice following systemic injection of fentanyl when compared with saline, but no change in velocity was observed in Oprm1 À/À mice (Fig. 5C). The postinjection response was divided into early and late phases (Fig. 5D). During the early phase (minutes 5-10 postinjection), a significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.0192, mixed-effects model with Sidak multiple comparisons). Fentanyl significantly increased velocity compared with saline in wild-type mice (p = 0.0069), but not in Oprm1 À/À mice (p = 0.9898). Velocity was also significantly increased following fentanyl injection in wild-type mice compared with Oprm1 À/À mice (p = 0.0001). During the late phase (minutes 50-55 postinjection), no significant differences were found for velocity in response to fentanyl when compared with saline in wild-type or Oprm1 À/À mice (p = 0.1787 and p = 0.9375, respectively, Wilcoxon matchedpairs signed-rank test with Holm-Sidak multiple comparisons). Overall, fentanyl significantly increased locomotion in wild-type mice and had no effect on locomotion in Oprm1 À/ À mice, suggesting that MORs mediate the locomotor response to fentanyl.
Next, we examined the association between velocity and respiratory rate in response to the injection of saline and fentanyl (Fig. 5E). Significant positive correlations were found between velocity and respiratory rate in both . Respiratory rate and relative tidal volume in response to fentanyl in wild-type and Oprm1 À/À mice. A, Respiratory rate following the injection of saline or fentanyl in wild-type and Oprm1 À/À mice. B, Respiratory rate averaged over a 5 min period during the early and late phase of the responses to saline and fentanyl. C, Relative tidal volume following the injection of saline or fentanyl. D, Relative tidal volume averaged over a 5 min period during the early phase and late phase of the responses to saline and fentanyl. wild-type mice (r = 0.8496, p , 0.0001) and Oprm1 À/À mice (r = 0.7258, p , 0.0001) in response to saline injection, for which increased velocity was associated with increased respiratory rate. Following fentanyl injection, a significant negative correlation was found between velocity and respiratory rate in wild-type mice (r = À0.4274, p , 0.0001), where greater velocity was associated with lower respiratory rate. However, a positive correlation was found between respiratory rate and velocity in Oprm1 À/À mice following fentanyl injection (r = 0.7739, p , 0.0001). These results show that, despite substantial locomotor activity in mice following fentanyl injection, respiratory rate was depressed, and, interestingly, its severity increased in association with increased movement. By contrast, a positive correlation was found between velocity and respiratory rate following fentanyl injection in Oprm1 À/À mice, suggesting that the loss of functional MORs abolished the effects of fentanyl.

Deletion of MORs in Sst-expressing cells and respiratory depression by fentanyl
To determine the role of Sst-expressing cells in opioidinduced respiratory depression, we produced conditional transgenic knock-out mice that lacked functional MORs in Sst-expressing cells only (Fig. 6). To delete exons 2 and Figure 5. Opioid-induced locomotor hyperactivity in wild-type and Oprm1 À/À mice. A, Videos were recorded from the bottom of the plethysmography chamber to assess mouse movements. B, Representative tracings and heat maps of mouse movements during the 1 h postinjection period. C, Average velocity for 1 h postinjection of saline and fentanyl in wild-type and Oprm1 À/À mice. D, Velocity during early and late responses to injections. E, Relationships between velocity and respiratory rate in wild-type and Oprm1 À/À mice following saline and fentanyl injections. Data are represented as the mean 6 SEM. *p , 0.05. Panel A was created with BioRender.com.
3 of the Oprm1 gene in Sst-expressing cells, we used a Cre-loxP recombination strategy and bred Sst-Cre and Oprm1 fl/fl mice to produce Sst-Oprm1 À/À mice. Although breeding strategies with Cre-loxP recombination are well established and are commonly used, especially in the context of Oprm1 knockout (Severino et al., 2020), we designed a genotyping protocol to confirm that the Oprm1 gene was deleted in Sst-expressing cells. Breeding of Sst-Cre and Oprm1 fl/fl mice aimed to remove exons 2 and 3 of the Oprm1 gene (Fig. 6A). In Oprm1 fl/fl mice, the amplicon of the floxed Oprm1 exons 2 and 3 was ;3000 bp in length (Fig. 6B). When exons 2 and 3 of the Oprm1 gene were removed, the amplicon was reduced to ;470 bp (Fig. 6C). Thus, the size of the amplicon should be 470 bp in Sst-expressing cells of Sst-Oprm1 À/À mice. Considering that Taq DNA polymerase can synthesize DNA at a rate of ;1000 bp/min (Arezi et al., 2003;Korbie and Mattick, 2008), we used a short synthesis time (,1 min) during the PCR protocol, so only the short amplicon of 470 bp was amplified (Fig. 6D). Electrophoresis gel from tissue samples from an ear notch and the medulla of a Sst-Oprm1 À/À mouse indicated the presence of Cre (;200 bp) and loxP (;450 bp) in both ear and medullary samples (Fig. 6E). Importantly, a mutant band using our primer (;470 bp) was found in medullary tissue but not ear notch tissue, suggesting tissue-specific excision based on the coexpression of Sst and Oprm1. In conclusion, our genotyping approach confirmed that Sst-Oprm1 À/À mice lacked Oprm1 exons 2 and 3 in Sst-expressing cells.
A systemic injection of fentanyl (0.3 mg/kg; Fujii et al., 2019) was administered and respiratory activity was recorded in control (Sst-Cre) and Sst-Oprm1 À/À mice using Figure 6. Genotyping of conditional Sst-Oprm1 À/À mice. A, To produce Sst-Oprm1 À/À mice, we bred Oprm1 fl/fl and Sst-Cre mice. B, In Oprm1 fl/fl mice, the sequence of the floxed Oprm1 exons 2 and 3 was ;3000 bp in length. C, When exons 2 and 3 of the Oprm1 gene were removed, the sequence was reduced to ;470 bp. Taq DNA polymerase can synthesize DNA at a rate of ;1000 bp/min (Arezi et al., 2003;Korbie and Mattick, 2008). When provided with a short synthesis time (,1 min) during the PCR protocol, the long fragment would be too long to amplify, therefore resulting in no band being produced. D, To determine whether the Oprm1 gene was deleted specifically in Sst-expressing cells, we genotyped tissue with coexpression of Sst and Oprm1, such as medullary brain tissue, and tissue without coexpression, such as the ear skin in Sst-Oprm1 À/À mice. Using custom primers, the short amplicon of 470 bp was amplified by Taq DNA polymerase using a short synthesis time. Sequences are provided for the forward and reverse primers designed to target regions upstream and downstream of the floxed exon 2 and 3 fragment of the Oprm1 gene. The protocol used for touchdown PCR is shown along with associated cycle numbers and durations (Korbie and Mattick, 2008). E, Electrophoresis gel from an Sst-Oprm1 À/À mouse with tissue samples from an ear notch and the medulla indicates bands with Cre (;200 bp) and loxP (;450 bp) in both ear and medullary samples. A band for loxP was produced in Sst-Oprm1 À/À mice since the sequence was still present in non-Sst-expressing cells, and one loxP sequence also remained after the excision of Oprm1 exons 2 and 3 by Cre. The presence of a mutant band using our custom primers (;470 bp) indicated that the floxed Oprm1 exon 2 and 3 fragment was excised in medullary tissue but not in ear notch tissue. Negative controls (NCs) are indicated. whole-body plethysmography (Fig. 7B,C). Systemic fentanyl injection decreased relative minute ventilation compared with saline in both control and Sst-Oprm1 À/À mice (Fig. 7D,E). During the early phase (minutes 5-10 postinjection), a significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.0408, two-way repeated-measures ANOVA with Sidak multiple comparisons; Fig. 7F). Fentanyl significantly reduced relative minute ventilation compared with saline in both control mice (p = 0.0058) and Sst-Oprm1 À/À mice (p , 0.0001). During the late phase of the response (minutes 50-55 postinjection), relative minute ventilation was significantly higher in response to fentanyl compared with saline in Sst-Cre mice (p = 0.0078, Wilcoxon matched-pairs signed-rank test with Holm-Sidak multiple comparisons) and Sst-Oprm1 À/À mice (p = 0.0391).
Fentanyl depressed respiratory rate in control and Sst-Oprm1 À/À mice compared with saline (Fig. 8A). During the early phase, no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.4581, two-way repeated-measures ANOVA with Sidak multiple comparisons; Fig. 8B). There was a significant effect of treatment (p , 0.0001), where fentanyl induced a significant respiratory rate depression compared with saline injection in both control (p , 0.0001) and Sst-Oprm1 À/À mice (p , 0.0001). During the late phase of the response, no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.0710, two-way repeated-measures ANOVA with Sidak multiple comparisons), though a significant effect of genotype was found (p = 0.0096) with a significant difference between the respiratory rates of control and Sst-Oprm1 À/À mice Figure 7. Opioid-induced respiratory depression in control and Sst-Oprm1 À/À mice. A, Generation of Sst-Oprm1 À/À mice using a Cre-loxP recombination strategy by breeding Oprm1 fl/fl and Sst-Cre mice. B, C, Respiratory parameters were measured using whole-body plethysmography recording in freely behaving control (Sst-Cre) and Sst-Oprm1 À/À mice. D, Representative traces of respiratory responses to saline and fentanyl injections. E, Relative minute ventilation following the injection of saline and fentanyl in control and Sst-Oprm1 À/À mice. F, Relative minute ventilation averaged over a 5 min period during the early and late phase responses to injection. Data are represented as the mean 6 SEM. *p , 0.05. Panels A and B were created with BioRender.com. Figure 8. Respiratory rate and relative tidal volume in response to fentanyl in control and Sst-Oprm1 À/À mice. A, Respiratory rate following the injection of saline or fentanyl in control and Oprm1 À/À mice. B, Respiratory rate averaged over a 5 min period during the early and late phases of the response to saline and fentanyl. C, Relative tidal volume following the injection of saline or fentanyl. D, Relative tidal volume averaged over a 5 min period during the early and late phases of the response to saline and fentanyl.
following fentanyl injection (p = 0.0028). No significant differences were found between baseline respiratory rates in control and Sst-Oprm1 À/À mice (p = 0.4830, unpaired t test with Welch's correction).
Relative tidal volume was increased in both control and Sst-Oprm1 À/À mice following fentanyl injection, when compared with saline (Fig. 8C). During the early phase, no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.7965, two-way repeatedmeasures ANOVA with Sidak multiple comparisons; Fig. 8D). There was a significant effect of treatment (p = 0.0002), where fentanyl induced a significant increase in relative tidal volume compared with saline in both control mice (p = 0.0043) and Sst-Oprm1 À/À mice (p = 0.0138). During the late phase of the response, no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.5354, two-way repeated-measures ANOVA with Sidak multiple comparisons), though a significant effect of treatment was found (p , 0.0001) where relative tidal volume was significantly increased in response to fentanyl in control mice (p = 0.0023) and Sst-Oprm1 À/À mice (p = 0.0006), compared with saline. Overall, our results show that mice lacking functional MORs in Sst-expressing cells presented similar ventilatory depressions compared Figure 9. Opioid-induced locomotor hyperactivity in control and Sst-Oprm1 À/À mice. A, Locomotor activity was recorded using a video camera positioned below the chamber. B, Representative tracings and heat maps of mouse movements during the 1 h postinjection period. C, Average velocity for 1 h after injection of saline or fentanyl in control (Sst-Cre) and Sst-Oprm1 À/À mice. D, Velocity in response to treatment during early and late phases of the response. E, Relationships between velocity and respiratory rate in control and Sst-Oprm1 À/À mice following saline and fentanyl injection. Data are represented as the mean 6 SEM. *p , 0.05. Panel A was created with BioRender.com. with control mice following fentanyl administration, suggesting that Sst-expressing cells are not involved in respiratory depression by the opioid fentanyl.

Deletion of MORs in Sst-expressing cells and locomotor activity
We examined the locomotor effects of fentanyl in control (Sst-Cre) and Sst-Oprm1 À/À mice (Fig. 9A). No significant differences were found between baseline velocities of control and Sst-Oprm1 À/À mice (p = 0.0592, Mann-Whitney test). Representative movement traces of control and Sst-Oprm1 À/À mice following injection of either saline or fentanyl are shown in Figure 9B. Mouse velocity increased in both control and Sst-Oprm1 À/À mice following fentanyl injection when compared with saline (Fig. 9C). The postinjection period was divided into early and late phases (Fig. 9D). During the early phase (minutes 5-10 postinjection), no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.7501, two-way repeated-measures ANOVA with Sidak multiple comparisons). There was a significant effect of treatment (p , 0.0001) where fentanyl significantly increased velocity compared with saline both in control mice (p , 0.0001) and Sst-Oprm1 À/À mice (p = 0.0001). Similarly, during the late phase (minutes 50-55 postinjection), no significant interaction occurred between treatment (saline or fentanyl) and genotype (p = 0.1567, two-way repeated-measures ANOVA with Sidak multiple comparisons). There was, however, a significant effect of treatment (p = 0.0001) where velocity was significantly increased in response to fentanyl compared with saline in control mice (p = 0.0328) and Sst-Oprm1 À/À mice (p = 0.0007). We repeated our analysis examining the relationship between respiratory rate and velocity in control and Sst-Oprm1 À/À mice (Fig.  9E). Following saline injection, a significant positive correlation was found between velocity and respiratory rate in both control mice (r = 0.7748, p , 0.0001) and Sst-Oprm1 À/À mice (r = 0.7888, p , 0.0001), where increased velocity was associated with increased respiratory rate. Following fentanyl injection, a significant negative correlation was found between velocity and respiratory rate in control mice (r = À0.6311, p , 0.0001) and a weak negative correlation in Sst-Oprm1 À/À (r = À0.3452, p , 0.0001), as previously shown in wild-type mice, where greater velocity was associated with lower respiratory rate. Overall, these results show that mice lacking functional MORs in Sst-expressing cells and control mice presented with locomotor hyperactivity negatively correlated with the depression of respiratory rate in response to fentanyl.

Discussion
Opioid drugs have caused countless deaths because of their addictive properties, leading to overdose and severe respiratory depression. Respiratory depression is characterized by slow and shallow breathing that can lead to hypoxemia and death during an overdose (Dahan et al., 2010). Understanding the mechanisms underlying respiratory depression by opioids will ultimately aid in the development of safer opioid pain therapies without potentially lethal side effects. Here, we aimed to identify the circuits mediating the effects of opioid drugs on respiratory rate. Somatostatin-expressing cells are critical for normal breathing (Tan et al., 2008;Cui et al., 2016;de Sousa Abreu et al., 2022) and are found within brainstem respiratory regions shown to contribute to respiratory depression by opioids (Helke, 1984;Johansson et al., 1984;Kalia et al., 1984;De León et al., 1992;Stornetta et al., 2003;Bou Farah et al., 2016;de Sousa Abreu et al., 2022). Importantly, silencing of somatostatin-expressing neurons in key respiratory circuits such as the preBötC, leads to prolonged apneas (Tan et al., 2008); therefore, suggesting that somatostatin-expressing preBötC cells constitute vital components of rhythmic breathing that could contribute to respiratory depression by fentanyl. In our study, we first examined expression of Sst and Oprm1 mRNAs in the brainstem. We found that a majority (.50%) of Sst mRNA-expressing cells expressed Oprm1 mRNA in the preBötC, the NTS, the NA, and the KF nucleus, supporting the hypothesis that these cells may be vulnerable to inhibition by opioids. To determine the role of Sst-expressing cells in opioid-induced respiratory depression, we developed conditional knock-out mice that lacked functional MORs in Sst-expressing cells. We found that respiratory rate was still depressed by fentanyl in these mice, suggesting that Sstexpressing cells are not required for opioid-induced respiratory rate depression.
Sst and Oprm1 coexpression in brainstem respiratory regions Both somatostatin and MORs are expressed throughout the brainstem De León et al., 1992;Mansour et al., 1994). Using in situ hybridization to quantify Sst expression, we found moderate expression of Sst mRNAs in the preBötC, the NTS, the NA, and the KF nucleus, with low expression in the ROb and RPa nuclei. Previous studies have found relatively high expression of somatostatin in the preBötC (Stornetta et al., 2003), the NTS (Kalia et al., 1984), and the KF nucleus (Viollet et al., 2008). This difference may be because of our cell counts being expressed as a percentage of total DAPI-stained cells. Since DAPI stains both neuronal and non-neuronal cell nuclei, counts of total DAPI may include non-neuronal cells, which would lower the percentage of cells expressing Sst, Oprm1, or both. On the other hand, we found high expression of Oprm1 in the NTS and the NA, moderate expression in the preBötC and the KF nucleus, and low expression in the raphe nuclei. This is consistent with previous studies showing that the NTS (Cerritelli et al., 2016;Zhuang et al., 2017), the NA (Xia and Haddad, 1991), the preBötC (Hayes et al., 2017), and the KF nucleus (Varga et al., 2020b) express MORs. We found that a majority of Sst mRNA-expressing cells (.50%) in the preBötC, the NTS, the NA, and the KF nucleus coexpressed Oprm1 mRNA. Interestingly, these brainstem regions also contribute to opioid-induced respiratory depression. For instance, the preBötC (Montandon et al., 2011;Bachmutsky et al., 2020;Varga et al., 2020a), the NTS (Zhang et al., 2011;Zhuang et al., 2017), the NA (Hassen et al., 1984), and the KF nucleus (Prkic et al., 2012;Levitt et al., 2015;Miller et al., 2017;Saunders and Levitt, 2020) mediate important components of the respiratory side effects of opioids. Our results therefore support the idea that, because of the high expression of Oprm1 in Sst mRNA-expressing cells, these cells may be vulnerable to opioid drugs and contribute to respiratory depression.

Role of MORs in opioid-induced respiratory rate depression
Most opioid ligands are selective for a wide-range of opioid receptors (Montandon, 2022). For example, fentanyl and morphine bind strongly to the MOR, while morphine also moderately binds to k -opioid receptors. More specifically, the respiratory depressant effects of fentanyl and morphine are mediated by MORs (Dahan et al., 2001;Hill et al., 2020). Here in our study, we first characterized the respiratory and locomotor response to the opioid fentanyl in wild-type and Oprm1 À/À mice. During the baseline period, no significant difference in respiratory rate was found between wild-type and Oprm1 À/À mice. This is in contrast to earlier findings (Dahan et al., 2001) showing a slight increase in respiratory rate in Oprm1 À/À mice compared with controls. However, this study used hybrids of 129/SV and C57BL/6 mice as controls, whereas we used C57BL/6. We then assessed the response to a relatively high dose of fentanyl (0.3 mg/kg; Fujii et al., 2019) in freely behaving, nonanesthetized wild-type and Oprm1 À/À mice. Fentanyl decreased relative minute ventilation compared with saline in wild-type mice, but not in Oprm1 À/À mice, an effect attributed to significant respiratory rate depression following the administration of fentanyl. Fentanyl moderately increased relative tidal volume in wild-type mice, but not in Oprm1 À/À mice, which is likely because of increased behavioral artifacts collected with plethysmography. In previous studies, fentanyl at this dosage consistently decreased minute ventilation in wild-type mice (Kuo et al., 2015;Varshneya et al., 2022). By contrast, Oprm1 À/À mice did not show any significant changes in respiratory rate and relative tidal volume following fentanyl injection compared with saline. This is consistent with fentanyl inhibiting neuronal activity by acting on MORs (Montandon, 2022). There were also significant differences in respiratory rates in response to saline during the early postinjection response between wild-type and Oprm1 À/À mice, despite no differences in baseline breathing rates before injection. It is plausible that the lack of MORs may affect the behavioral and respiratory responses to the stress associated with intraperitoneal injection. In fact, endogenous MOR circuits are involved in stress (Du et al., 2022), and the absence of MORs may affect the stress response following injection. Overall, our results showed that respiratory rate depression by fentanyl was mediated by MORs in freely behaving mice.

Role of Sst-expressing cells in opioid-induced respiratory rate depression
After showing that functional MORs mediate the respiratory rate depressant effects of the opioid fentanyl, we then determined whether MORs in Sst-expressing cells contribute to respiratory rate depression. Using a Cre-loxP recombination approach, we produced conditional knock-out mice that lacked functional MORs selectively in Sst-expressing cells. We compared respiratory responses in control and Sst-Oprm1 À/À mice. In control mice, fentanyl injection induced a significant decline in respiratory rate compared with saline, as observed in wild-type mice. Surprisingly, fentanyl also decreased respiratory rate and relative minute ventilation in Sst-Oprm1 À/À mice when compared with saline. Although we found significant differences in respiratory rates between control and Sst-Oprm1 À/À mice during the late phase following fentanyl injection, the physiological mechanisms mediating these differences are unclear. In addition, fentanyl moderately increased tidal volume in control and Sst-Oprm1 À/À mice, similar to wild-type mice. Overall, our findings suggest that Sst-expressing cells are not involved in opioid-induced rate respiratory depression. These findings are surprising given the coexpression of Sst and Oprm1 mRNAs in regions involved in mediating respiratory depression by opioid drugs and the importance of somatostatin-expressing cells in the modulation of breathing (Tan et al., 2008;Cui et al., 2016;de Sousa Abreu et al., 2022). These results suggest that respiratory depression by fentanyl may involve a population of somatostatin-negative neurons in key medullary regions regulating opioid-induced respiratory depression.

Opioid-induced locomotor hyperactivity
To ensure that the respiratory effects of opioids were not concealed by movement artifacts in freely behaving mice, we assessed locomotor activity (quantified as velocity) in response to saline and fentanyl. Fentanyl injection induced a significant increase in locomotor activity when compared with saline injection in wild-type, control (Sst-Cre), and Sst-Oprm1 À/À mice, but not in Oprm1 À/À mice. These findings suggest that while MORs mediate the locomotor effects of opioids, as previously shown (Contarino et al., 2002), the lack of functional MORs in Sst-expressing cells does not affect this response. Next, we correlated respiratory rate and velocity to determine whether changes in locomotion may affect respiratory rate. If locomotor activity were a confounding factor for our breathing rate results, we would expect to see a positive correlation between velocity and respiratory rate in response to fentanyl in wild-type mice. Instead, we found a negative correlation in these mice, for which severe respiratory rate depression was correlated with greater locomotor hyperactivity. Interestingly, respiratory rate during opioid-induced respiratory depression and simultaneous locomotor hyperactivity was comparable to respiratory rate during the preinjection baseline where animals were undisturbed for 2 h and velocity was near zero. Our study is the first, to our knowledge, to demonstrate the inverse correlation between the severity of respiratory rate depression and locomotor hyperactivity. In Oprm1 À/À mice exposed to fentanyl, respiratory rate and locomotor hyperactivity were positively correlated, where increased locomotor activity was associated with increased respiratory rate. Positive correlations were also observed in wild-type mice following saline administration. In both control and Sst-Oprm1 À/ À mice, the severity of respiratory rate depression and locomotor hyperactivity were negatively correlated following fentanyl administration, similar to findings in wild-type mice. This once again shows that the lack of functional MORs in Sst-expressing cells did not change the locomotor or respiratory responses to fentanyl. In conclusion, our results demonstrate that, even with increased locomotion following fentanyl administration, respiratory rates were depressed in wild-type, control, and Sst-Oprm1 À/À mice, but not in Oprm1 À/À mice.
In summary, we characterized respiratory and locomotor responses to determine the cell types mediating respiratory rate depression by opioids and found that MORs unequivocally mediate the respiratory rate depressive effects of the opioid fentanyl. Considering the role of somatostatin-expressing cells in the control of breathing, we quantified the expression of Oprm1 mRNA in Sst-expressing cells of multiple respiratory circuits involved in mediating the effects of opioids on respiration. We found that a majority (.50%) of Sst-expressing cells coexpressed Oprm1 mRNA in the preBötC, the NTS, the NA, and the KF nucleus. Despite this, the deletion of MORs in Sst-expressing cells did not affect respiratory rate depression by fentanyl when compared with control mice. Considering that we measured respiratory rate depression in freely behaving mice by quantifying respiratory activity combined with measures of locomotion, we are confident that respiratory rate depression by fentanyl was accurately assessed. Our results suggest that other cell types may be involved in mediating respiratory rate depression by the opioid fentanyl. MORs are indeed expressed in other cell populations within the brainstem, including cells expressing the neurokinin-1 receptor (Gray et al., 1999). In mice lacking the Tac1 gene, which encodes the peptide substance P, the endogenous ligand for neurokinin-1 receptors, morphine induced a reduced respiratory depression compared with wild-type mice (Bilkei-Gorzo et al., 2010). In addition, neurokinin-1 receptor-expressing preBötC cells are preferentially inhibited by the MOR agonist DAMGO (Montandon et al., 2011). In conclusion, our study showed that Sst-expressing cells are spared from the effects of fentanyl on respiratory rate depression and may constitute a robust neuronal population that can be targeted to stimulate breathing.