Buprenorphine Depresses Respiratory Variability in Obese Mice with Altered Leptin Signaling

THIS study was motivated by two imperatives. First, the ongoing epidemics of opiate abuse¹  and obesity²  reinforce our interest in opiate research with potential for translational relevance.^3–12  Second, opiate-induced alterations in the ventilatory response to hypercapnia are sexually dimorphic,¹³  consistent with evidence that respiratory control is altered by sex steroid hormones.¹⁴  Additional data show that opiate-induced respiratory depression is associated with increased risk among subjects who are female¹⁵  and/or obese.¹⁶  There is consensus that gaps in knowledge regarding mechanisms by which opiates alter breathing “warrant further study of the sex-dependent and sex-independent genetic factors” and that “inbred strains of mice and rats may serve as the best model.”¹³  We concur, and the present study quantified the effects of buprenorphine on minute ventilation variability in obese and normal weight male and female mice. The obese mice used for this study were characterized by either disrupted leptin signaling or diet-induced obesity. This design made it possible to use a three-way, mixed effects analysis of variance model to evaluate the two-tailed hypothesis that buprenorphine causes breathing to vary significantly due to sex, leptin status, and obesity. Buprenorphine was a logical choice because it is used for pain management in addicted patients¹⁷  and because it is one of three medications approved for treating opiate use disorder.^1,18 

Experiments and procedures were reviewed and approved by the University of Tennessee Institutional Animal Care and Use Committee (Knoxville, Tennessee). All phases of the experiments adhered to the Guide for the Care and Use of Laboratory Animals (National Academies Press, 8th edition, Washington, D.C., 2011). Mice had ad libitum access to food and water and were housed under 24-h illumination to promote free-running rhythms and offset potential circadian confounds.¹⁰  Adult mice (n = 80) were purchased from The Jackson Laboratory (USA). These congenic mice (10 males and 10 females of each line) included normal weight C57BL/6J (B6) wild-type mice that produce leptin; B6 mice with diet-induced obesity (DIO) that may lead to leptin resistance; obese B6.Cg-Lep^ob/J mice (ob/ob) that do not make leptin; and B6.BKS(D)-Lepr^db/J (db/db) mice that are obese, produce leptin, and lack functional leptin receptors. None of these mice are transgenic animals, and the ob/ob and db/db mice were derived from spontaneous mutations in the B6 line. These congenic lines are maintained at The Jackson Laboratory by a multigenerational breeding program that backcrosses a mutation of interest into different mice, thereby transferring the mutations of interest. The B6, ob/ob, and db/db mice were fed normal Teklad 8640 rodent chow comprised of 5% fat. The DIO mice were fed a 60% fat diet for 20 weeks before and during the testing period. Both rodent chows were purchased from Research Diets, Inc. (USA).

The four congenic lines of mice (B6, DIO, ob/ob, and db/db) are ideally suited to calls for studies “of the sex-dependent and sex-independent genetic factors”¹³  contributing to opiate-induced changes in breathing. Efforts to understand the sexually dimorphic response to opiates^15,19  are scientifically sound²⁰  and consistent with the 2016 policy mandating that National Institutes of Health–funded studies include sex as a biologic variable.²¹ 

Among the four mouse lines, the obesity and leptin phenotypes were regarded as sex-independent factors that may contribute to buprenorphine-induced alterations in breathing. This possibility is suggested by the fact that leptin, produced by adipocytes, covaries with obesity and modulates breathing^22–25  and nociception.^7,10,12 Table 1 summarizes similarities and differences between the obesity and leptin phenotypes in the four congenic mouse lines studied. Use of three-way analysis of variance made it possible to evaluate the effects of buprenorphine on breathing as a function of sex, leptin status, and obesity.

Mice were injected (intraperitoneally) with saline or buprenorphine (0.3 mg/kg) (Sigma–Aldrich, USA). Buprenorphine was dissolved in saline to yield the desired concentration while holding injection volume (0.3 ml) constant. The 0.3-mg/kg dose of buprenorphine has been shown to be antinociceptive in B6 mice²⁶  and in males and females of the four congenic mouse lines used in this study.¹²  Before starting data collection, the mice were habituated to being placed in whole body plethysmography chambers (Data Sciences International, USA), where they were free to move and groom. After 1 week of habituation, the mice were randomly assigned to receive either saline or buprenorphine injection and were placed in the plethysmography chambers where breathing was automatically recorded for 1 h. Testing occurred at the same time of day, and all mice had 1 week for recovery between injections. Breathing signals from mice in the plethysmography chambers were automatically digitized by the software, circumventing the need to blind experimenters to injection condition (saline vs. buprenorphine). Pioneering studies of rodent breathing across genetic lines²⁷  demonstrated that measures must be scaled to body mass, differences in CO₂ production, hypoxic and hypercapnic ventilatory responses, dead space, fat distribution, and many other factors. The present measures of the effects of buprenorphine on minute ventilation across four congenic lines of mice were therefore expressed as ml/min/g body weight, consistent with our previous studies on the respiratory effects of isoflurane²⁸  and sedative/hypnotics.²⁹ 

Sample sizes (n = 80 mice consisting of 10 males and 10 females representing each of the four congenic mouse lines) were based on previous experience.^7,10,12,29  The data were analyzed with Statistical Analysis System v9.4TS1M3 (SAS Institute, Inc., USA), using a three-way mixed-effect, repeated-measures ANOVA model with two between-subject factors (sex and mouse line) and one within subject factor of injection condition (saline vs. buprenorphine). Mouse nested within sex by mouse line was included as a random effect. After a log transformation was applied to minute ventilation values, the data satisfied all assumptions of the ANOVA model including normal distribution of residuals, similar variances, and no major outliers. No other transformations were required. Bonferroni adjustment was used to correct the P values for all post hoc comparisons.

Poincaré analysis is well established and permits novel diagnostic and prognostic insights.³⁰  The heuristic value of Poincaré analysis also makes it clinically relevant for diverse disorders ranging from locomotion³¹  to sleep.³²  Poincaré analysis can even help identify phenotypes of metabolic syndrome.³³  Relative to anesthesiology, Poincaré analysis has been used as a weaning predictor for some postoperative patients,³⁴  for estimating levels of sedation analgesia,³⁵  and to characterize anesthesia-induced changes in electroencephalogram.³⁶ 

By plotting a dependent measure relative to the previous dependent measure, Poincaré analysis makes it possible to quantify and visualize variability within a data set.³⁰  The resulting dispersion of points conveys novel information about variability in the data (fig. 1). For the present study, Poincaré analysis was used to characterize how buprenorphine alters the variability of minute ventilation. Whole-body plethysmography data were averaged every 5 min for 1 h. Poincaré analysis was used to quantify respiratory variability among male and female B6, DIO, ob/ob, and db/db mice in response to injection of saline and buprenorphine. Points in the Poincaré plots of minute ventilation represent time intervals of an inspiratory and expiratory cycle. In the present study, Poincaré short-term variability (SD1) quantified the variation in breathing during the 5-min measurement intervals. Long-term variability (SD2) quantified the overall variability in breathing throughout the 1-h measurement period. SD1 describes more rapid changes, and SD2 describes long-term changes. Short-term variability (SD1) also is described by the SD of successive differences in a breath-to-breath cycle, and long-term variability (SD2) also can be described by the SD in breath-to-breath intervals.³⁷  Both analytic approaches were applied to the present data set.

Figure 2 shows that DIO, ob/ob, and db/db mice had decreased minute ventilation under the control (saline) condition. Compared to the B6 (wild-type) mice, buprenorphine decreased average minute ventilation by 37.2% in DIO mice (t = 7.68, df = 72, P < 0.0001), 62.6% in ob/ob mice (t = 16.61, df = 72, P < 0.0001), and 65.9% in db/db mice (t = 18.21, df = 72, P < 0.0001). There was no significant sex main effect on minute ventilation. Regression analyses revealed that in no case did body weight alone account for a significant decrease in minute ventilation.

Figure 3 shows Poincaré plots illustrating features of the buprenorphine-induced decrease in minute ventilation variability. The collapse in short-term variability (SD1), visualized in figure 3 as the distribution of points perpendicular to x = y, is not apparent in figure 2. The Poincaré plots show that compared to saline, buprenorphine caused a collapse in minute ventilation variability across all four lines of mice. Three-way ANOVA indicated that short-term variability (SD1) was significantly decreased (F = 45.03; df = 3,72; P < 0.0001) as a function of mouse line and injection condition (saline or buprenorphine) (F = 148.98; df = 1,72; P < 0.0001). The only significant interaction was between mouse line and injection condition (F = 5.30; df = 3,72; P = 0.0023). Post hoc comparisons showed that buprenorphine caused a greater compression of short-term variability of minute ventilation in the DIO, ob/ob, and db/db mice than in the B6 mice. Figure 4 shows that when short-term variability was expressed as the SD of successive differences in breath-to-breath intervals (SD of successive differences), mice with disrupted leptin signaling had lower minute ventilation variability than B6 (wild-type) mice.

Long-term variability (SD2) of minute ventilation (fig. 3, dispersion of points along the line x = y) differed significantly by mouse line (F = 52.66; df = 3,72; P < 0.0001), sex (F = 9.05; df = 1,72; P = 0.0036), and injection condition (F = 184.42; df = 1,72; P < 0.0001). There was a significant (F = 4.02; df = 3,72; P = 0.0105) sex-by-mouse-line interaction for long-term changes in variability of minute ventilation. Post hoc analyses showed that long-term variability in minute ventilation of normal weight B6 mice differed from that of DIO, ob/ob, and db/db mice. Applying three-way ANOVA to long-term variability data calculated as the SD of breath-to-breath intervals revealed similar results (fig. 5). Considered together, the results of analyses illustrated by figures 4 and 5 support the conclusion that buprenorphine caused the greatest decrease in minute ventilation variability in mice with impaired leptin signaling.

Three major findings emerged from this study. First, minute ventilation was significantly altered by leptin status in the absence of buprenorphine. Second, minute ventilation did not vary significantly due to sex. Third, Poincaré plots illustrate that buprenorphine causes a collapse in minute ventilation variability. The findings are discussed below relative to the relationship between leptin and obesity, making the point that studies of congenic mice can provide unique insights into this relationship. Finally, we describe the previously unrecognized effect of buprenorphine to decrease respiratory variability. Limitations of the present study also are addressed.

The perspective that obesity is a disease satisfies multiple criteria, the most stringent of which is the increased risk that obesity conveys for premature death.^38,39  Analysis of closed claims cases identified obesity as a risk factor for opiate-induced respiratory depression resulting in severe brain damage or death.⁴⁰  Leptin covaries with obesity, making it difficult to quantify the exact degree to which breathing varies as a function of obesity, leptin, or both. The present study confronted this complexity in two ways. First, the experiments were specifically designed to compare the respiratory effects of buprenorphine in normal weight B6 mice possessing normal leptin to the respiratory effects of buprenorphine in three congenic lines of obese mice, each with different obesity/leptin profiles. The four mouse lines made it possible to relate buprenorphine-induced changes in breathing (figs. 2–5) to obesity and leptin status (table 1). B6 mice provided an ideal control group because they are among the lines of healthy mice that the Mouse Phenome Committee recommend receive the highest priority for phenotyping.⁴¹  B6 mice with diet-induced obesity are genetically identical to normal weight B6 mice. DIO mice are homologous to the vast majority of obese humans in that the mice possess both leptin and leptin receptors that, over time, may become leptin-resistant. Regression analyses provided a second approach for evaluating the degree to which buprenorphine-induced changes in breathing were altered by obesity or leptin. Those analyses showed that for none of the four mouse lines did body weight account for a significant change in minute ventilation. This finding is consistent with evidence that body weight alone does not significantly alter the antinociceptive effects of buprenorphine.¹² 

Human obesity due to congenital, monogenic leptin deficiency is rare,⁴²  and leptin-deficient ob/ob mice do not emulate most cases of human obesity. The ob/ob mice, however, provide a valuable animal model with the potential to clarify the multiple functions of leptin. The db/db mice produce leptin but fail to express the long form of the leptin receptor.^43,44  Studies by others have used DIO, ob/ob, and db/db mice to gain insights into metabolic syndrome⁴⁵  and type 2 diabetes.^46,47  To our knowledge, the present data provide the first evidence that leptin signaling alters the effects of buprenorphine on minute ventilation by reducing minute ventilation variability.

Comparing figure 3 with figures 2, 4, and 5 illustrates the ability of Poincaré plots to provide a nuanced visualization of respiratory variability. In the Materials and Methods section, the description of Poincaré as exploratory data analysis refers to an “attitude, a state of flexibility, a willingness to look for those things that we believe are not there, as well as those we believe to be there.”⁴⁸  The present application of Poincaré analysis demonstrates that buprenorphine significantly reduces minute ventilation variability. Figure 3 also makes clear that after saline administration, the mice with functional leptin systems (B6 and DIO⁴⁹ ) had significantly greater variability in minute ventilation than the mice with dysfunctional leptin transmission (ob/ob and db/db). Consequently, the decrease in minute ventilation variability caused by buprenorphine was greatest for B6 and DIO mice that express both leptin and functional leptin receptors. Statistical analyses showed that leptin status played a central role via its ordinal interaction with sex and injection of buprenorphine. This finding supports the interpretation that leptin dysfunction (ob/ob and db/db), rather than obesity in the presence of leptin (DIO), conferred the greatest risk for decreased minute ventilation variability (fig. 3). The present results are consistent with evidence that leptin status rather than obesity causes an impaired ventilatory response to hypercapnia in ob/ob mice.⁵⁰ 

Potential limitations of the present study include the murine model, unknown mechanisms of leptin signaling, and the mathematics underlying the present discovery. The limitations of mice as models of human obesity⁵¹  and obesity-related diseases derive from evolutionary differences in gene expression between mouse and human.⁵²  While acknowledging these limitations, the National Institutes of Health consider the mouse as “the premier mammalian model system for genetic research” (https://www.genome.gov/10005834/background-on-mouse-as-a-model-organism/; accessed July 23, 2017). The genetics of human obesity reveal good homology with genotypes known to promote obesity in mice.⁵³  In addition, the mechanisms by which leptin, a protein produced by adipocytes, influences breathing remain unknown.^{24,25,50,54,55}  It is clear, however, that leptin function is not limited to energy homeostasis.⁵⁶  Leptin modulates neuronal excitability,⁵⁷  and long-standing evidence shows that leptin stimulates breathing.^22,23  The present results should not be misinterpreted as indicating that leptin is the only adipocyte-derived molecule modulating the buprenorphine-induced collapse in minute ventilation variability. The potential involvement of additional cell signaling proteins is suggested by the fact that adipose tissue functions as an endocrine organ and secretes hundreds of different adipokines.⁵⁸  Furthermore, inferring that only leptin from adipocytes modulates the effects of buprenorphine may be too simplistic in view of the discovery that leptin also is produced by parathyroid glands.⁵⁹  A third potential limitation is interpreting and communicating time-domain data. The present study included Poincaré methods because the analyses (fig. 3) “correspond so conveniently to time domain summary statistics.”³⁷  The strengths and limitations of Poincaré analysis and its numeric descriptors (figs. 4 and 5) have been reviewed in detail.^30,60 

The results show that buprenorphine causes a collapse in minute ventilation variability, with the greatest collapse occurring in mice that lack normal leptin signaling. The results are consistent with evidence that leptin signaling alters the antinociceptive effects of buprenorphine.¹²  Respiratory variability is clinically relevant because such variability is an essential aspect of homeostasis (e.g., increasing tidal volume or respiratory rate) in response to challenges imposed by disease, obesity, surgical stress, and opiates. In cases where buprenorphine does cause respiratory depression, pharmacokinetic data make the cautionary point that buprenorphine-induced respiratory depression “outlasts the effects of naloxone-induced reversals from bolus injections.”⁶¹  Thus, the present findings encourage future studies to characterize the time course of the buprenorphine-induced decrease in respiratory variability. The novelty of the present findings is further illustrated by the fact that all previous clinical studies of the respiratory effects of buprenorphine have involved normal weight individuals. Maintaining ventilation within a normal range is key in the perioperative setting. However, that goal is complicated by the increasing number of obese patients and by patients who chronically use opiates or who are prescribed buprenorphine to manage their opiate use disorder.¹  Poincaré analysis of heart rate variability has a long record of diagnostic and prognostic success. The present results encourage future human studies to determine whether Poincaré plots have diagnostic or prognostic value relevant to respiratory depression caused by opiates.

Supported by National Institutes of Health (Bethesda, Maryland) grant No. HL65272, by the Departments of Anesthesiology and Psychology, and by a grant from NeuroNET (University of Tennessee, Knoxville). Also supported by the Beaman Endowed Professorship (to Dr. Baghdoyan) and Robert H. Cole Endowed Professorship (to Dr. Lydic).

The authors declare no competing interests.