A Subregion of the Parabrachial Nucleus Partially Mediates Respiratory Rate Depression from Intravenous Remifentanil in Young and Adult Rabbits

OPIOIDS are standard treatment to reduce perioperative and chronic pain; however, their use is limited by respiratory depression.^1–5  Respiratory depression is primarily mediated by μ-opioid receptors,^6,7  which are widely expressed throughout the brainstem respiratory network.^3,4,8–16  The typical pattern of clinical, opioid-induced respiratory depression is a decrease in respiratory rate and even apnea. Because respiratory rhythm is generated by the central pattern generator (CPG)^17–20  in the brainstem, studies have looked for opioid effects in this area.^4,10,14  Local application of μ-opioid receptor agonists at pharmacologic, micromolar concentrations causes significant depression of neuronal activity in several areas of the CPG, that is, the ventral respiratory column^21,22  including the pre-Bötzinger complex (preBötC)^9,10,23,24  in the rostral medulla and the parabrachial nucleus (PBN)¹⁴  and Kölliker–Fuse nucleus (KFN)²⁵  in the rostral pons.

There are conflicting results regarding the brainstem location where clinically relevant, nanomolar opioid concentrations depress respiratory rate.^26–29  Since the discovery of pacemaker-like, opioid receptor–containing neurons in the preBötC, investigators have focused on this area.³⁰  The importance of the preBötC regarding clinical opioid effects was called into question when in an in vivo decerebrate dog model local injection of naloxone into the preBötC did not reverse the bradypneic effects of an IV remifentanil infusion.⁹  In contrast, sequential injections of naloxone into the PBN region significantly reversed respiratory rate depression in the same model.¹⁴  Our previous study in the in vivo decerebrate rabbit model showed that IV remifentanil infusion indeed affected inspiratory and expiratory phase timing in the preBötC but that reversing these effects with local naloxone injections did not reverse the respiratory rate depression.²³  These seemingly contradictory results suggest that systemic opioids at clinically relevant concentrations affect more than one area within the respiratory network but that not all effects necessarily lead to changes in respiratory rate.

Our current study focused on the rabbit PBN. Previous studies had achieved respiratory rate depression with a μ-opioid receptor agonist bath applied to the dorsal surface of the pons in cats³¹  and with grid-wise injections of μ-opioid receptor agonists into an area caudal of the inferior collicle and several millimeters lateral from midline in dogs.¹⁴  We sought to determine the following: (1) is there a subregion of the PBN where μ-opioid receptor agonists depress respiratory rate; (2) can this location be identified by the tachypneic response to local injection of the glutamate agonist α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA); (3) what is the maximal respiratory rate depression that pharmacologic concentrations of μ-opioid receptor agonists can achieve in this area; and (4) what is the degree by which respiratory rate depression from systemic, clinically relevant concentrations of μ-opioid receptor agonists can be reversed in this area? Because the contribution of the pons to the respiratory pattern seems to change throughout development,³²  experiments were performed in young and adult rabbits. Finally, we compared the results from the PBN with our previous study investigating opioid effects on the preBötC²³  to determine any differences in opioid effects on the two brainstem sites.

This research was approved by the subcommittee on animal studies of the Zablocki VA Medical Center, Milwaukee, Wisconsin, in accordance with provisions of the Animal Welfare Act, the Public Health Service Guide for the Care and Use of Laboratory Animals, and U.S. Department of Veterans Affairs policy. The surgical procedures were similar to previous studies in this laboratory.²³  In short, adult (>6 months; 3 to 4 kg) and young (14 to 25 days; 200 to 500 g) New Zealand White rabbits of either sex were induced with 5% sevoflurane via facemask. Animals were tracheotomized and then ventilated with an anesthesia machine ventilator (Ohmeda CD, GE Datex Ohmeda, USA) or at a weight less than 400 g, with a small animal ventilator (SAR-830 ventilator, CWE, USA). Anesthesia was maintained with 1 to 2% (young) or 1.5 to 3% (adult) isoflurane.³³  Anesthetic depth was increased for signs of inadequate anesthesia, such as increased heart rate, blood pressure, or lacrimation. Fractional inspired oxygen tension, expiratory carbon dioxide, and expiratory isoflurane concentration were continuously recorded with an infrared analyzer (POET II, Criticare Systems, USA). Skin was infiltrated with 1% lidocaine before each skin incision. Femoral arterial and venous lines were used for blood pressure monitoring and infusion of solutions, respectively. Lactated Ringer’s solution with 2 μg/ml epinephrine was continuously infused at 1 ml/h. This infusion rate did not result in appreciable changes in heart rate and blood pressure. Infusion rate was increased as needed for hypotension in response to drug injections or from blood loss. Rectal temperature was monitored and maintained at 37.0 ± 0.5^oC with a warming blanket. The animal was placed in a stereotaxic frame (David Kopf Instruments, USA), and a pneumothorax was performed to prevent ventilator artifact during neuronal recording. Blunt precollicular decerebration with complete removal of the forebrain was performed through a parietal craniotomy. After decerebration, isoflurane was discontinued or continued at subanesthetic levels (0.3 to 0.4%) for additional blood pressure control. In the latter case, isoflurane concentration was not changed throughout the experimental protocol. The brainstem was exposed via occipital craniotomy, and partial (young) or complete (adult) removal of the cerebellum was performed. Animals were paralyzed with vecuronium (initially 1 mg/kg and redosed as needed) to avoid motion artifacts during neural/neuronal recording. The vagal nerves were left intact. Phrenic nerve activity was recorded with fine bipolar electrodes through a posterior neck incision. Throughout the experiment animals were ventilated with a hyperoxic gas mixture (fractional inspired oxygen tension = 0.6) to achieve functional denervation of the peripheral chemoreceptors and at mild hypercapnia (expiratory carbon dioxide = 45 to 55 mmHg) to ensure sufficient respiratory drive (i.e., above the apneic threshold), including during systemic opioid infusion. At the end of the experiment the animals were euthanized with IV KCl. In a subgroup of animals the brainstem was fixed via transcardial perfusion for histologic analysis. The brainstem tissue was cryoprotected, frozen, and serially sectioned for Nissl staining and identification of fluorescent tracer injection sites.

The neuronal recording and microinjection techniques have been described previously in detail.^34,35  In short, extracellular neuronal recordings were obtained using glass multibarrel micropipettes (20- to 40-μm tip diameter) consisting of three drug barrels and a recording barrel containing a 7-μm–thick carbon filament (~0.5 MΩ). Barrels were filled with AMPA (50 μM), the μ-opioid receptor agonist [D-Ala,² N-MePhe,⁴ Gly-ol]-enkephalin (DAMGO; 100 μM), and the opioid receptor antagonist naloxone (1 mM), which were dissolved in artificial cerebrospinal fluid (aCSF). The microinjected volume was determined via height changes in the meniscus in the respective pipette barrel with a 100× monocular microscope and calibrated reticule (resolution = approximately 3.5 nl). Respiratory neurons were classified according to their discharge pattern and their temporal relationship relative to the phrenic neurogram. The neuronal and pressure microejection marker signals were continuously displayed and recorded along with the phrenic neurogram, respiratory rate meter, arterial blood pressure, airway pressure, and expiratory carbon dioxide on a computerized chart recorder (Powerlab/16SP; ADInstruments, Australia).

Post hoc analysis averaged respiratory cycles from the phrenic neurogram. For each study protocol, steady-state conditions were obtained for respiratory parameters both before and after drug injection. Based on the phrenic neurogram, between 10 and 50 consecutive respiratory cycles were averaged over 1 to 2 min, with the number of cycles dependent on the respiratory rate (breaths/min). We determined peak phrenic activity (PPA), respiratory rate, and inspiratory duration (T_I) and expiratory duration (T_E). Respiratory drive was calculated as PPA/T_I.

Chamberlin et al.¹⁵  described various changes of the respiratory pattern with glutamate injection into the PBN/KFN region in rats studied in vivo with an increase in respiratory rate elicited mostly in the PBN region. Prkic et al.¹⁴  showed in dogs studied in vivo that grid-wise injection of DAMGO into the PBN area, that is, between the inferior collicle and superior cerebellar peduncle, decreased respiratory rate. Guided by these studies we performed grid-wise microinjections of AMPA (70 nl) with step size 0.5 mm into the equivalent area in adult (n = 12) and young (n = 14) rabbits. Recorded neuronal activity was used to guide the depth of injection. We identified a location in the rostral pons, where AMPA microinjection caused an increase in respiratory rate (tachypnea). On average, in young rabbits this area was located 0.5 ± 0.5 mm caudal to the inferior collicle, 2.0 ± 0.5 mm lateral to midline, and 6.0 ± 1.0 mm below the dorsal surface of the brainstem/residual superior cerebellar peduncle (n = 26). The cerebellum was not completely removed in animals less than 400 g. In adult rabbits the area was located 0.5 ± 0.2 mm caudal to the inferior collicle, 2.7 ± 0.1 mm lateral of midline, and 7.4 ± 0.2 mm below to the dorsal surface of the brainstem/residual superior cerebellar peduncle (n = 35). Often a lesser tachypneic response was observed with AMPA injection 0.5 mm rostral and/or caudal and 0.5 mm medial to this area (fig. 1A). We injected the μ-opioid receptor agonist DAMGO bilaterally into the area of greatest tachypnea and found that an injection volume of 350 nl in young and 700 nl in adult rabbits was necessary to achieve maximal respiratory rate depression (n = 6) but that with these volumes additional injections around this area did not produce any additional rate depression (n = 12). The majority of neurons that we recorded in this area had no modulation with respiratory phase, that is, a tonic discharge pattern (young: tonic neurons n = 411, expiratory neurons n = 4; adult: tonic neurons n = 444, expiratory neurons n = 6, inspiratory neurons n = 2). Postmortem histology located the area in the PBN region (fig. 1B). We will refer to this area as the “tachypneic area” of the PBN (tPBN). Before each study protocol, the tPBN was located bilaterally according to stereotaxic coordinates, presence of neuronal discharge activity, and maximal tachypneic response to AMPA injection.

In four adult animals, the tPBN was functionally identified and then marked by injection of a fluorescent tracer (700 nl, 5% Red Retrobeads, Lumafluor Inc., USA). The animals were then transcardially perfused with phosphate-buffered saline and 4% paraformaldehyde in phosphate-buffered saline, followed by extraction of the brainstem tissue. The tissue was cryoprotected in 30% sucrose for at least 72 h, frozen, and serially sectioned (25 μm) in the transverse plane from the superior cerebellar peduncle to 2 mm rostral to the caudal border of the inferior colliculus (approximately 5 mm). Each section was adhered to electrostatically treated slides and was Nissl stained for identification of gross anatomic structures and the relative location of fluorescent tracer (fig. 1B).

For Nissl staining, after a 10-min drying period, the tissue was cleared in Histoclear (Sigma, USA) for 1 h followed by sequential rehydration in 100%, 95%, and 70% ethanol before a 10-min distilled, deionized water rinse. The tissue was exposed to 4% cresyl violet for 12 min followed by sequential dehydration in ethanol, including exposure to 0.5% acetic acid ethanol. The tissue was again cleared in Histoclear for 1 h followed by coverslipping. Tissue was stained at 100-μm intervals, and images were captured at 4,000 DPI (Nikon Super Coolscan 9000, Nikon, Japan). MetaMorph imaging software (Molecular Devices, USA) was used to spatially calibrate each image.

To identify the location of the fluorescent tracer, each section was examined with a Nikon Eclipse E600 fluorescent microscope (Nikon) and photographed with a Hamamatsu ORCA-FLASH 4.0 LTS SCMOS camera (Hamamatsu Photonics, Japan). Images were acquired at a resolution of 2,048 × 2,048 pixels, 6.5 × 6.5 μm pixel size. The location of the fluorescent tracer relative to midline and the ventral surface of the brainstem were recorded. Distance from the brainstem dorsal surface was not used due to variation in the amount of cerebellar tissue remaining in each animal. The mediolateral and dorsoventral coordinates of the tracer were marked on the spatially calibrated Nissl-stained image.

The tachypneic area in the PBN is in close anatomic proximity to other respiratory-related areas that also contain opioid receptors, that is, the locus coeruleus (LC)^36,37  and the KFN.^25,38  To verify that the drug effects observed with injection into the tPBN were not due to diffusion into these adjacent areas, we performed a separate set of experiments in 8 adult animals where we injected DAMGO into the bilateral tPBN and then attempted to reverse the DAMGO effect with naloxone injections into the bilateral LC and KFN.

Grid-wise injections of AMPA (50 µM, 70 nl) into the LC area to identify an area involved in respiratory rate control did not reveal consistent changes in respiratory rate (n = 6). In two animals we observed transient slowing of respiratory rate in a location 1.0 mm medial, 0.5 mm rostral, and 3.0 mm dorsal from the tachypneic area of the PBN, that is, approximately 1.5 mm lateral from midline, 0.5 mm caudal from inferior collicle, and 3.5 mm ventral to the dorsal surface. DAMGO (100 μM, 700 nl) injection into this area did not have any effect. Dye injection confirmed that this area was within the LC. For the control protocol (see below: Location of the KFN), we thus determined the location of the tPBN first and used the coordinates 1.0 mm medial, 0.5 mm rostral, and 3.0 mm dorsal to this area for our LC injections. Dye injection into these areas after completion of the protocol confirmed that this area was always in the LC.

We identified the KFN by injecting AMPA (50 µM, 70 nl) in grid-wise fashion lateral, caudal, and ventral to the tPBN where it was expected per histology. We found an area where AMPA injection caused bradypnea, as described by Dutschmann and Herbert³⁹  and Levitt et al.,²⁵  which was located on average 0.5 mm caudal, 0.5 to 1.0 mm lateral, and 2.0 mm ventral to the tPBN. Dye injection after completion of the control protocol confirmed that this area was in the KFN.

After determining the location of the tPBN, the LC, and the KFN, we performed bilateral DAMGO injections (100 µM, 700 nl) into the tPBN analogous to protocol 1 (see below: Main Studies). Three minutes after the second DAMGO injection, that is, when steady-state respiratory depression was reached and when we would inject naloxone into the tPBN in protocol 1 (see below: Main Studies), we instead injected naloxone (1 mM, 700 nl) into the bilateral LC and the bilateral KFN. Finally, we injected naloxone (1 mM, 840 nl) into the bilateral tPBN. Bilateral DAMGO injections into the tPBN decreased the respiratory rate from 24 ± 12 breaths/min to 12 ± 7 breaths/min. Naloxone injections into the bilateral LC and KFN did not have any reversal effect (10 ± 8 breaths/min; P = 0.503). However, final injection of naloxone into the bilateral tPBN reversed the depression (18 ± 11 breaths/min; P = 0.026, all 1-way repeated measures ANOVA). We conclude that the respiratory depression achieved with local DAMGO injection in protocol 1 (see below: Main Studies) and any local naloxone reversal of IV remifentanil-induced respiratory depression in protocol 2 (see below: Main Studies) indicate opioid effects solely mediated by the PBN.

The experimenters were not blinded to the experimental conditions. Animals were not randomized to the protocols, because New Zealand White rabbits are a purebred strain with little physiologic variation between animals. Only one complete protocol was performed per animal to avoid any confounding effects from residues of the locally injected drugs.

To determine the effect of maximal μ-opioid receptor activation, 350 nl (young) or 700 nl (adult) DAMGO (100 μM) was microinjected bilaterally into the tPBN. Injections were spaced 3 min apart, which was sufficient to accomplish steady-state respiratory rate depression. Subsequently, the effect was reversed with bilateral injections of the competitive opioid antagonist naloxone (1 mM; young = 420 nl; adult = 840 nl) at the same coordinates.

To determine the contribution of the tPBN to systemic opioid-induced respiratory depression we injected naloxone into the tPBN during continuous IV infusion of the potent µ-agonist remifentanil. The tPBN was identified on both sides of the brainstem as described above. Then remifentanil was infused intravenously at 0.08 to 0.50 μg · kg⁻¹· min⁻¹ until respiratory rate was depressed by approximately 50%. These infusion rates match the clinically relevant analgesic dose rates described for rabbits.⁴⁰  Remifentanil was chosen for its short onset time and short half-life (approximately 4 min)⁴¹  that remains independent of the duration of the infusion.^42,43  After reaching steady-state respiratory depression for at least 5 min, 1 mM naloxone (young = 420 nl; adult = 840 nl) was injected bilaterally into the tPBN with injections spaced 3 min apart. After the respiratory pattern had again reached a steady state, a single IV injection of naloxone (30 to 80 μg/kg) was given to completely reverse any residual systemic opioid effect. Only then was the remifentanil infusion discontinued.

To ensure that the naloxone effect represented reversal of remifentanil-induced respiratory depression rather than reversal of intrinsic opioidergic tone,¹⁵  we injected naloxone (1 mM, 840 nl) into the bilateral tPBN without remifentanil infusion. Similarly, to rule out an independent effect of aCSF, which was used as solvent for all of the injected drugs, aCSF (840 nl) was injected into the bilateral tPBN without remifentanil infusion. In the interest of reducing animal use, control experiments were only conducted in adult rabbits.

Statistical analysis was performed using SigmaPlot 11 (Systat Software, USA) for ANOVA with Tukey test for pairwise multiple comparisons in the exploratory study 2 and the paired t tests in protocol 3. Data sets were tested for normal distribution (Kolmogorov–Smirnov test). R software (R package version 3.1 to 128, http://CRAN.R-project.org/package=nlme; accessed November 28, 2016) was used for the linear mixed-effect model with Bonferroni correction for multiple comparisons for the results for protocols 1 and 2 and the comparison between tPBN and preBötC. We did not perform a formal a priori power analysis. Sample sizes for each protocol were based on previous studies after an initial review of the first 5 or 6 animals per protocol, and no adjustments were made for interim analyses. Comparable studies have used 8 to 15 rabbits,²³  4 to 9 rats,¹⁰  or 10 to 21 dogs per protocol.^9,14  For all of the protocols, statistical tests were performed on raw data except for PPA and respiratory drive, where activity is measured in arbitrary units and normalization to control is necessary to allow for comparison between animals. The effects of opioid agonists/antagonists on all of the respiratory parameters were determined using a linear mixed model including fixed effects for age (adult/young), type of drug (protocol 1: control, local DAMGO, local naloxone; protocol 2: control, IV remifentanil, IV remifentanil plus local naloxone, IV remifentanil plus IV naloxone), and study (tPBN/preBötC). Interaction terms measured any interaction among type of drug, type of study, and the factor age. Other interaction terms were considered but found to be nonsignificant. The intercept consisted of an overall intercept term, that is, the average response for an adult control animal in the preBötC study, a random effect for each individual animal, and the model noise term. Similarly, the inputs to inspiratory and expiratory duration as determined from the two studies were compared using a linear mixed-effect model with fixed effects for age, study and input source, and Bonferroni correction for multiple comparisons.

Average respiratory rate in all of the experiments was 27 ± 10 breaths/min. Bilateral injection of DAMGO into the tPBN resulted in a significant decrease in respiratory rate from 26.0 ± 10.0 breaths/min by 15.5 ± 9.1 breaths/min (62 ± 20%) in young rabbits (n = 14) and a decrease from 23.0 ± 13.0 breaths/min by 11.0 ± 7.2 breaths/min (45 ± 26%) in adults (n = 11; P < 0.001; fig. 2). This was due to an increase in both T_I (young: 1.2 ± 2.0 s; adult: 1.0 ± 2.2 s; P = 0.004) and T_E (young: 6.3 ± 5.7 s; adult: 3.1 ± 4.4 s; P < 0.001). Bilateral DAMGO injection also decreased PPA (young: −17 ± 32%; adult: −25 ± 29%; P = 0.007) and respiratory drive (PPA/T_I; young: −49 ± 26%; adult: −38 ± 56%; P < 0.001). All of the DAMGO effects were near completely reversed by bilateral injection of naloxone at the same coordinates (P < 0.05). There was no significant difference in DAMGO effect between young and adult rabbits (P > 0.05) for any of the parameters. For full disclosure, one young animal (23 days, 350 g) was removed from above analysis as DAMGO injection resulted in severe, naloxone-reversible inspiratory apneusis with T_I increased 50 times more than average. The effect suggested incorrect injection into the KFN (see Discussion for KFN characteristics).

In a separate group of 12 young and 12 adult rabbits, naloxone was injected bilaterally into the tPBN during systemic IV remifentanil infusion (rate: 0.08 to 0.5 μg · kg⁻¹· min⁻¹). Baseline respiratory rate was 31 ± 4 breaths/min in young and 27 ± 11 breaths/min in adult animals. IV remifentanil depressed respiratory rate by 17.0 ± 3.4 breaths/min (55 ± 9%) in young rabbits and by 12.8 ± 7.0 breaths/min (46 ± 20%) in adults (P < 0.001; fig. 3). This was due to an increase in T_I (young: 0.5 ± 0.4 s; adult: 0.3 ± 0.6 s; P < 0.001) and T_E (young: 2.3 ± 1.6 s; adult: 2.5 ± 3.3 s; P < 0.001). IV remifentanil also decreased PPA (young: −24 ± 22%; adult: −38 ± 18%; P < 0.001) and respiratory drive (young: −48 ± 21%; adult: −51 ± 15%; P < 0.001).

Bilateral injection of naloxone into the tPBN resulted in a partial recovery of respiratory rate by 10.8 ± 5.8 breaths/min, which was within 20 ± 14% of the control rate in young rabbits, and by 7.2 ± 11.0 breaths/min, which was within 18 ± 27% of the control rate in adults (P < 0.001; fig. 3). This was due to a decrease in T_I (young: −0.4 ± 0.2 s; adult: −0.2 ± 0.5 s; P = 0.02), as well as T_E (young: −1.8 ± 0.4 s; adult: −2.0 ± 1.2 s; P < 0.001). Bilateral naloxone injection also resulted in a recovery in PPA (young: 21 ± 26%; adult: 24 ± 40%; P = 0.001) and respiratory drive (young: 35 ± 26%; adult: 30 ± 47%; P < 0.001). IV naloxone infusion (30 to 80 μg/kg) completely reversed any residual remifentanil effects. There were no significant differences between young and adult rabbits in either the IV remifentanil effect or the local naloxone reversal (P > 0.05 for all parameters).

In six adult rabbits, bilateral injection of naloxone into the tPBN under control conditions had no significant effect on respiratory rate T_I, T_E, PPA, or respiratory drive (all P > 0.05; data not shown). In a separate set of animals, injection of 700 nl of aCSF into the tPBN did not have any effect on respiratory rate and pattern (n = 8, all P > 0.05; data not shown).

The linear mixed model allowed for comparison of our current data with data from our previous study, which investigated opioid effects on the preBötC with similar protocols. Pharmacologic concentrations of DAMGO affected the respiratory pattern in both areas; however, there were some distinct differences (fig. 4): DAMGO depressed respiratory rate significantly more in the tPBN than the preBötC (difference: 7 ± 3 breaths/min; P = 0.012). Although the increase in T_E was similar (P = 0.456), DAMGO caused an increase in T_I in the tPBN but shortened it in the preBötC (difference: 2.2 ± 0.6 s; P < 0.001). DAMGO injection decreased respiratory drive in the tPBN but not in the preBötC (difference: 60 ± 14%; P < 0.001).

Clinical concentrations of remifentanil increased T_E more in the tPBN than in the preBötC (difference: 5.0 ± 1.4 s; P = 0.001; fig. 5), and there was an increase in T_I in the tPBN but a decrease in the preBötC, resulting in a significant difference (1.4 ± 0.4 s; P = 0.002). The depression of respiratory rate was not statistically significant between the tPBN and the preBötC (P = 0.491). Remifentanil depressed respiratory drive more in the tPBN than in the preBötC (difference: 41 ± 15%; P = 0.005).

This experimental protocol allowed us to estimate the contributions of the tPBN versus non-tPBN areas to inspiratory and expiratory phase duration and to compare them with the contributions of the preBötC versus non-preBötC areas obtained in our previous study.²³  The calculations are described in detail in the appendix, and the resulting input values are summarized in table 1. Intrinsic activity of inspiratory and expiratory neurons is modulated by opioid-sensitive inputs that originate from the preBötC area, the tPBN area, and potentially additional areas. Opioid-sensitive input from the preBötC increased T_I, although input from the tPBN decreased T_I. Inputs from the preBötC and the tPBN shortened T_E. PreBötC input to T_I was significantly different from non-preBötC inputs (P = 0.024). Pontine input to T_E was significantly larger than medullary input (P = 0.0003). The effects are summarized in a hypothetical model (fig. 6).

In a developmental rabbit model we identified a subregion of the PBN (tPBN) that is involved in respiratory timing and where opioids cause respiratory rate depression at clinical concentrations in vivo. Specifically, we showed the following: (1) pharmacologic concentrations of the μ-opioid receptor agonist DAMGO decreased respiratory rate in the tPBN; (2) local μ-opioid receptor antagonism in the tPBN substantially reversed respiratory rate depression from systemic opioids; (3) clinical dose rates of IV remifentanil affected respiratory phase timing differently in the tPBN and the preBötC; (4) clinical dose rates of IV remifentanil depressed respiratory drive more in the tPBN than in the preBötC; and (5) there was no difference in the opioid effect on the tPBN between young and adult rabbits.

In this study, bilateral microinjection of DAMGO (100 μM) into the tPBN decreased respiratory rate (fig. 2). Similar effects have been observed in the medullary raphe in rats^12,13,44  and the preBötC in rats,^4,10  as well as in our previous study in rabbits²³  (fig. 4). Thus, pharmacologic doses (μM to mM) of μ-opioid receptor agonists can affect respiratory rate at multiple sites within the respiratory network. However, the much lower plasma and effect site concentrations (approximately 10 nM in humans⁴¹  and approximately 20 nM in dogs^41,43 ) that are achieved with clinically relevant, analgesic doses may not affect these areas. In this study, respiratory rate depression from IV remifentanil could be substantially reversed with localized naloxone injection into the tPBN confirming the clinical relevance of the tPBN in opioid-induced respiratory depression.

We propose a hypothetical model where opioid-sensitive inputs from several brainstem areas modulate intrinsic inspiratory and expiratory phase duration (fig. 6). The anatomic correlate for intrinsically active, opioid-sensitive inspiratory neurons may be type 1, inspiratory preBötC neurons.²⁴  The expiratory correlate may be preinspiratory neurons with intrinsic phase duration,⁴  although they appear less opioid sensitive than the inspiratory neurons. Our data suggest that opioid-sensitive inputs from the preBötC are different from tPBN in their magnitude, as well as effect on phase duration (table 1). Also, although the differences were not statistically significant, non-preBötC inputs seemed larger than tPBN inputs. We have thus added an additional opioid-sensitive input to our model. This may reflect chemodrive from the retrotrapezoid nucleus⁴⁵  or the caudal medullary raphe where systemic opioids decrease respiratory rate.¹² 

A limitation of our studies was that we investigated only one dose (clinical target: approximately 50% rate depression) of remifentanil. This did not allow us to determine whether inhibition of the tPBN is the main cause for apnea with opioid overdoses. The relative magnitude of opioid-induced inhibition of the different areas may be dose dependent and may not follow a linear dose–effect relationship. For example, in an in vivo rat model, Montandon et al.¹⁰  could completely prevent the approximately 25% decrease in respiratory rate from 1 μg/kg fentanyl IV with microdialysis of naloxone into the preBötC. Similarly, when respiratory rate was depressed approximately 30% with IV DAMGO in in vivo rats, microinjection of the opioid antagonist CTAP into the caudal medullary raphe partially reversed the respiratory rate depression.¹²  In summary, systemic opioid concentrations affect respiratory phase timing in multiple areas of the respiratory network, but specifically the effect on the tPBN leads to marked respiratory rate depression.

Extensive afferent and efferent projections exist between the PBN/KFN region, LC, retrotrapezoid nucleus, and rhythmogenic neurons within the ventral respiratory column in the medulla.^{21,26,27,34,36,37,43,45,46}  Much work has focused on KFN contribution to the inspiratory off-switch, which determines inspiratory phase duration.⁴⁷  Glutamatergic excitation of the KFN results in bradypnea from expiratory phase prolongation, while inhibition through the N-methyl-d-aspartate antagonist MK-801 severely prolongs inspiratory duration (apneusis).^39,47–49  Injection of high concentrations of DAMGO (1 mM) into this area caused prolongation of T_I and T_E in spontaneously breathing, anesthetized rats but resulted in apneusis in the decerebrate in situ rat preparation.²⁵  Neuronal discharge patterns were described as tonic with inspiratory, expiratory, or phase-spanning modulation.^39,47,49  Exploratory study 2 showed that our injection site was distinctly different from the KFN. Postmortem histology (fig. 1B) placed the functionally defined tPBN in the medial PBN region, that is, rostral, medial, and dorsal of the KFN. Injection of the glutamate agonist AMPA into the tPBN resulted in tachypnea with shortening of the expiratory phase (data not shown). This matched the description by Dutschmann and Herbert³⁹  of an area rostral to the KFN, where glutamate injection caused transient tachypnea. Anatomic and functional projections to the medial PBN have been shown from the nucleus solitarius and ventral respiratory group,⁵⁰  and neuronal projections have been shown from the PBN to the rostral ventral respiratory group,⁵¹  the raphe magnus,⁵²  and the Bötzinger complex.⁵³  Immunohistology demonstrated intense stain for μ-opioid receptors in this area.^14,16 

The tPBN injections also did not affect the LC: numerous enkephalin receptors on LC dendrites³⁷  and the involvement of the LC in multiple regulatory functions including arousal and nociception⁵⁴  make the LC a theoretical target during our opioid protocols. The importance of awake drive for respiratory rate has been shown recently in pediatric patients.⁵⁵  However, the lack of effect of AMPA injections in exploratory study 2 suggests a lack of direct involvement of the LC in respiratory pattern control in our decerebrate preparation, which may be similar to the lack of hypothalamic drive during nonrapid eye movement sleep.⁵⁶ 

In contrast to similar studies in the in vivo dog preparation that used grid-wise drug injections over an area of several millimeters,¹⁴  we found that the tachypneic response to AMPA injection reliably identified the tPBN and that a single injection of the opioid agonist or antagonist was sufficient to produce the maximal effect. The area contained approximately 98% nonrespiratory modulated neurons, that is, did not receive pulmonary afferent-mediated inputs or phasic feedback from the medullary preBötC/Bötzinger complex. AMPA and DAMGO injection affected respiratory rate mainly through a change in T_E. We propose that the tPBN provides tonic excitatory input primarily to neurons mediating the expiratory off-switch (e.g., preinspiroatory neurons) and that inhibition of this area depresses respiratory rate predominantly through an increase in expiratory duration. Smaller increases in T_I may be due to decreased excitation of inspiratory off-switch neurons or to intrinsic network properties where increases in T_E cause an increase of the subsequent T_I.⁵⁷  Although the tPBN does not contain the phasic neurons necessary to generate the respiratory pattern like the preBötC/Bötzinger complex, the significant respiratory slowing that can be achieved through inhibition of tPBN neurons suggests that this area is highly relevant for a mature respiratory pattern.

Systemic and local opioid administration in the tPBN also depressed respiratory drive. This suggests that some neurons in this area also provide excitatory drive to inspiratory neurons in the ventral respiratory group or possibly to chemosensitive areas,^58,59  which provide drive to respiratory neurons in the medulla and pons.¹⁸  Although respiratory drive appeared completely restored after local naloxone reversal, other studies suggest that systemic opioids inhibit respiratory drive in multiple areas, including chemoreceptive areas.^12,13 

We have discussed limitations of our experimental technique in a previous publication.²³  Our studies use a complex in vivo setup with multiple drug injections over several hours. Despite a stable preparation, respiratory parameters at baseline, as well as the response to systemic and locally applied drugs, vary between animals. This limits the power of our analysis to identify small drug effects within each study and also smaller differences in respiratory parameters between this study and our previous study.²³  Using the linear mixed model allowed us to directly compare our current data on the tPBN with our historical data on the preBötC while accounting for random variation between studies. This approach makes our analysis the largest in vivo study to date to directly compare opioid effects with the same protocols on two different brainstem sites.

Developmental studies suggest that respiratory rhythm originates within one or two neuronal oscillators located in the preBötC and parafacial respiratory group.³⁰  The age where pontine inputs begin to shape respiratory pattern remains poorly defined. Dutschmann et al.³²  showed that the role of the KFN in the pulmonary stretch receptor–mediated component of the inspiratory off-switch (Hering–Breuer reflex) was mature by postnatal day 15 in rats. Technical difficulty currently prevents us from expanding our experiments to neonatal rabbits (less than 7 days). However, the similar results in rabbits age 2 to 3 weeks (i.e., preweaning) and adults suggests that the role of the tPBN in respiratory pattern generation is already established at a relatively early stage of development.³² 

μ-Opioid receptors within a functionally identified PBN subregion play a major role in mediating respiratory rate depression during the administration of systemic opioids at clinically relevant dose rates in young and adult rabbits. Pharmacologic manipulation of this area mainly affects expiratory duration, suggesting that this area provides excitatory drive to neurons of the expiratory off-switch. This is consistent with the observation that systemic opioids depress respiratory rate predominantly by increasing expiratory duration.

The authors thank Jack Tomlinson (biological laboratory technician) and Jennifer Callison, B.S., of the Medical College of Wisconsin, Milwaukee, Wisconsin, for excellent technical assistance.

Support was provided by the Foundation for Anesthesia Education and Research, Schaumburg, Illinois (FAERMRTG-BS-02-15-2010 to Dr. Stucke), the National Institutes of Health, Bethesda, Maryland (R01GM112960-02 to Dr. Stucke), and the Department of Veterans Affairs, Washington, DC (VA Merit Review Biomedical Laboratory Research and Development Award No. 2 I01 BX000721-05 to Dr. Zuperku). This publication was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, Bethesda, Maryland, through grant No. UL1TR001436.

The authors declare no competing interests.

Model to assess tPBN, preBötC, and additional contributions to respiratory phase timing.

The timing of respiratory phase switch can be described with a physical timer model where the sum of the inputs (∑) determines the time (T) until the threshold (V_threshold) is reached and the phase is terminated.

The values for inspiratory (T_I) and expiratory (T_E) phase duration obtained with protocol 2 allow us to estimate the magnitude and polarity of inputs from the naloxone injection site (F_i) as compared with all other brainstem sites (F_o). Intravenous remifentanil reduces all inputs by the factor “r.”

Inputs under control conditions result in control phase duration T_C. Systemic remifentanil inhibits opioid-sensitive inputs to all areas of the brainstem resulting in T_R. Local naloxone injection into the study area, that is, the tPBN or preBötC, restores the input at the injection site (F_i) to control values, although all other sites (F_o) are still inhibited by remifentanil, resulting in phase duration T_RN. These calculations apply for inspiratory and expiratory duration.

When V_thr = 1,

We used the actual values for T_IC, T_IR, and T_IRN, or T_E resp., for each individual animal from protocol 2 in the current and our previous study²³  to determine I, F_i, and F_o. We assumed r = 0.5 for the reduction in respiratory rate of approximately 50%. For the current study, F_i described input from the tPBN and F_o all inputs outside the tPBN, while in our previous study F_i described input from the preBötC and F_o all inputs from outside the preBötC.²³  Differences between F_i (tPBN) and F_o (preBötC) or F_i (preBötC) and F_o (tPBN), respectively, suggest additional inputs to phase duration from outside of these two areas.