Repeated nitrous oxide exposure in rats causes a thermoregulatory sign-reversal with concurrent activation of opposing thermoregulatory effectors

Abstract

Initial administration of 60% nitrous oxide (N₂O) to rats at an ambient temperature of 21°C decreases core temperature (Tc), primarily via increased heat loss (HL). Over repeated N₂O administrations, rats first develop tolerance to this hypothermia and subsequently exhibit hyperthermia (a sign-reversal) due primarily to progressive increases in heat production (HP). When rats initially receive 60% N₂O in a thermal gradient, they become hypothermic while selecting cooler ambient temperatures that facilitate HL. This study investigated whether rats repeatedly administered 60% N₂O in a thermal gradient would use the gradient to behaviorally facilitate, or oppose, the development of chronic tolerance and a hyperthermic sign-reversal. Male Long-Evans rats (N = 16) received twelve 3-h administrations of 60% N₂O in a gas-tight, live-in thermal gradient. Hypothermia (Sessions 1–3), complete chronic tolerance (Sessions 4–6), and a subsequent transient hyperthermic sign-reversal (Sessions 7–12) sequentially developed. Despite the progressive recovery and eventual hyperthermic sign-reversal of Tc, rats consistently selected cooler ambient temperatures during all N₂O administrations. A final 60% N₂O administration in a total calorimeter indicated that the hyperthermic sign-reversal resulted primarily from increased HP. Thus, rats did not facilitate chronic tolerance development by moving to warmer locations in the gradient, and instead selected cooler ambient temperatures while simultaneously increasing autonomic HP. The inefficient concurrent activation of opposing effectors and the development of a sign-reversal are incompatible with homeostatic models of drug-adaptation and may be better interpreted using a model of drug-induced allostasis.

Abbreviations

DHL

dry heat loss

EHL

evaporative heat loss

HL

heat loss

HP

heat production

N2

nitrogen

N2O

nitrous oxide

O2

oxygen

Tc

core temperature

Tsel

selected ambient temperature

Introduction

Drug tolerance, dependence and withdrawal are hypothesized to be manifestations of a common underlying ‘adaptive’ response that develops with repeated drug use.^1-7 Most adaptation models of drug tolerance and addiction trace their origin to the concept of homeostasis.⁸ In brief, when a drug effect initially perturbs a homeostatically regulated variable, this triggers adaptive responses that oppose and eventually fully counter the drug-elicited perturbation while the drug is present (i.e., tolerance develops). With repeated drug use, the individual transitions to a dependent state wherein drug-induced perturbations are effectively countered by acquired compensatory responses. In the dependent state, if the drug effect dissipates more rapidly than the compensatory responses, symptoms of drug withdrawal occur.

As proposed by Walter Cannon, the core concept of homeostasis is captured by the title of his book, “The Wisdom of the Body”⁹; i.e., the body's homeostatic wisdom enables a coordinated array of physiological and behavioral responses to be elicited that stabilize and defend critical regulated physiological variables as they become perturbed by drugs or other stimuli.¹⁰ Several aspects of homeostasis are especially relevant to adaptation models of drug addiction. Dworkin¹¹ reiterated a widely held belief about drug tolerance by stating “… even with very many administrations, drug effects sometimes diminish to zero but do not invert to the opposite” (p. xiv). Thus, homeostatic adaptations “wisely” do not overrespond, so as to overcompensate for the disturbance. Another principle of homeostasis is the ‘wise’, efficient central coordination of corrective responses such that they work harmoniously together and are not in concurrent competition with one another.^12,13 In addition, it is commonly suggested that homeostasis preferentially recruits the least costly effector response available to correct a perturbation.

The objective of the present research was to rigorously evaluate these homeostatic concepts as they pertain to adaptation models of chronic drug use. The experimental model of tolerance development to nitrous oxide (N₂O)-induced hypothermia is well suited for this purpose. Thermoregulation is an archetype of a homeostatically regulated system. Core temperature (Tc), the regulated variable,^14,15 lends itself to accurate, continuous, and non-invasive telemetric measurement. Tc has an extensive history as a dependent measure in studies of drug tolerance.^16-18 Of great advantage for research focused on physiological regulation, much is known about the physics and physiological effects that underlie thermoregulation. In particular, the underlying determinants of Tc can be quantified accurately at the level of metabolic heat production (HP) and body heat loss (HL).¹⁹ Telemetric measurement of Tc coupled with simultaneous total calorimetry (combining indirect and direct calorimetry) allows HP and HL to be measured non-invasively and continuously in undisturbed rats.^19-22

N₂O is a pharmacologically active gas and an abused inhalant.²³ It is administered via inhalation and among other effects, causes hypothermia upon initial administration in rats.^24-26 N₂O's low solubility in blood and tissues means that a steady-state concentration can be quickly achieved and easily maintained.²⁷ Once equilibration occurs, the N₂O concentration in an animal simply equals the N₂O concentration in the chamber. The ability to “clamp” the drug concentration is an important advantage when interpreting acute and chronic adaptations to N₂O.

Acute (intrasessional) and chronic (intersessional) tolerance develop to 60% N₂O's hypothermic effect in the rat.^24,26 Total calorimetric assessments revealed that the marked drop in Tc during an initial administration of 60% N₂O is due primarily to a rapid elevation in HL that results in a state of negative heat balance.^19,20 Increases in HP can occur during the course of an initial N₂O administration and result in the development of acute tolerance,¹⁹ while progressive increases in HP over subsequent N₂O administrations result in the development of chronic tolerance.^21,22

In a previous study, a subset of rats that were relatively insensitive to the hypothermic effect of an initial 60% N₂O administration developed a hyperthermic Tc during subsequent N₂O administrations rather than merely becoming tolerant.²⁸ Subsequent research revealed that initially insensitive rats exhibited a prompt increase in HP when initially administered N₂O that was of sufficient magnitude to counter the increase in HL elicited by the N₂O²⁰; i.e., the rats considered ‘initially insensitive’ at the level of Tc were actually initially hyperresponsive at the level of HP with the consequence that there was little or no change of Tc when first exposed to N₂O. The magnitude of the HP response increases progressively over repeated N₂O administrations, which contributes to chronic tolerance development, and with additional administrations causes rats to eventually exhibit a hyperthermic overcompensation of Tc.^21,22,29

Rats recruit both autonomic and behavioral thermoeffectors to maintain Tc, although behavioral effectors often provide a quicker and more energetically efficient way to influence Tc.³⁰ Behavioral thermoregulation can be assessed by allowing animals to select their preferred ambient temperature (Tsel) in a thermal gradient.^30-32 Rats given an initial exposure to 60% N₂O while in a thermal gradient develop the usual hypothermia, and at the same time they move to a region of the gradient where the ambient temperature is cooler.³² The goal of the present study was to determine whether chronic tolerance, with or without a sign-reversal hyperthermia, develops over repeated exposures to 60% N₂O in a thermal gradient. A second goal was to determine how behavioral and autonomic effectors contribute to the restoration and/or overcompensation of Tc and whether effector interaction is compatible with a homeostatic interpretation.

Materials and Methods

Subjects

Male Long-Evans rats (Charles River, N = 16; 8 squads of 2 each) arrived in the lab at 25–28 d of age. Both rats in a squad were housed together in a polycarbonate tub with free access to water and pelleted chow (5053 PicoLab Rodent Diet 20, Animal Specialties and Provisions, Quakertown, PA). The housing room and live-in thermal gradients had a 12-h:12-h light/dark cycle (lights on at 0700 h). The housing room had an ambient temperature of 22 ± 1°C. Following surgery and recovery, experimental testing began 12 d after the rats’ arrival in the lab (141.1± 21.2 g, Mean ± SD) and lasted 31 d when the rats weighed 362.9 ± 26.7 g. All animal procedures were approved by the University of Washington Institutional Animal Care and Use Committee.

Thermal gradient

The thermal gradient system allows a rat to select its preferred ambient temperature as a function of its choice of location within an alleyway. In brief, a removable rectangular acrylic alleyway is suspended within an insulated copper shell that is cooled at one end and heated at the other end, thereby creating a temperature continuum along the length of the alleyway. Our lab's 2 thermal gradients are based on a previously published design^30,31 that we modified to make the gradients gas-tight. Pelleted chow and water were freely available in the center of the alleyway. During the present study, the alleyway had an ambient temperature range of ∼30°C, with 7.6°C ± 0.46 and 37.9°C ± 0.22 (Mean ± SD) at the 2 ends. The relationship between the temperature at each location along the length of the alleyway was similar to that described by Gordon and colleagues.³¹ [A photograph of our thermal gradient system and additional details about its design and operation are available in Part I of the online supplement.]

One of 2 gas mixtures was delivered to each thermal gradient, i.e., either control gas consisting of room air, or 60% N₂O. Specifically, the control gas was made from room air that was purified, dehumidified and compressed and then delivered to the thermal gradient at a flow rate of 10 L/min. The N₂O gas had the same flow rate and was composed of 60% N₂O, 21% oxygen (O₂), and 19% nitrogen (N₂). [A 10 L/min blend of 79% N₂O, 21% O₂, and 0% control gas was delivered for the first 12 min of the 60% N₂O gas condition to achieve the targeted 60% N₂O gas concentration more quickly.] Concentrations of N₂O, O₂, and CO₂ were measured using an infrared gas analyzer placed in the incurrent and excurrent gas lines connected to the gradient's copper shell.

Total calorimetry, Tc, and N₂O administration chambers

Independent total calorimetry chambers that also measure Tc telemetrically served as gas-tight exposure chambers for N₂O. Total calorimetry simultaneously measures the rates of total HL and metabolic HP, the 2 underlying determinants of Tc. [See Part II of the online supplement for additional details.]

Telemetric measurement of Tc, data acquisition and instrument control

Telemetric measurement of Tc was accomplished using a commercial system from Data Sciences International (Saint Paul, MN) that consists of a Data-Exchange Matrix, Physio-Tel Receiver (Model RPC-1), Dataquest ART 4.2 software, and an implantable battery-powered temperature sensor (model TA-F40) implanted in the rat's peritoneal cavity. The antenna wires surrounding the alleyway suspended inside the thermal gradient are exteriorized through a sealed port and connected to the commercial receiver base. The antenna system within the direct calorimeter consists of 2 radio ferrite coils oriented perpendicularly to each other that are epoxied underneath a Plexiglas platform that holds them ∼2 mm above the floor of the calorimeter. Wires from these coils exit the calorimeter through a sealed port and are connected to the commercial receiver base. All other instrument control and data acquisition were performed using custom programs written in LabVIEW 6.8 (National Instruments, Austin, Texas).

Surgical placement of the telemetric temperature sensor

At least one week prior to the start of testing, a telemetric temperature sensor was implanted surgically into each rat's peritoneal cavity under isoflurane anesthesia while the rat was on a 39°C heating pad. Meloxicam (an NSAID) was provided in the drinking water (0.02 mg/ml H₂O) from 1 d before to 2 d after surgery.

Procedures

Each rat received 12, 3-h administrations of 60% N₂O in the thermal gradient over a 26-d period that commenced on a Monday, 13 d after arrival in the lab. The thermal gradients’ 2 temperature-controlled recirculating water baths were set at 1°C and 42°C and circulated water around each end of the gradient's copper shell from Monday at 0900 h through Friday at 1600 h each week. Each rat was taken from the housing room and placed in the thermal gradient at 1600 h on Monday and returned to the housing room on Friday at 1600 h. Between 1200 and 1500 h on Tuesday, Thursday and Friday, 60% N₂O was administered in the thermal gradient. As a within-subject control condition, control gas instead of N₂O was administered on Wednesday between 1200 and 1500 h. Rats were briefly removed from the thermal gradient between 1600–1615 h on Wednesdays so that the waste trays could be cleaned, additional food provided as needed, and the alleyway inspected. Thermal gradient components were washed/sanitized each weekend. Rats were weighed on Monday, Wednesday and Friday.

After completing the thermal gradient phase of the experiment, rats were tested in the total calorimeters using a counterbalanced, cross-over design, so that each rat received both a control gas and a 60% N₂O exposure occurring on Tuesday and Thursday of the following week. Each rat was placed in the calorimeter at 1000 h. Control gas was delivered for a 2-h baseline period (1000–1200 h), followed by 3 h of the assigned gas condition (i.e., continued control gas or 60% N₂O). Control gas was delivered from 1500–1515 h and the rat was then returned to the housing room.

Data reduction

Thermal gradient data

The rat's position in the alleyway was recorded at 7-s intervals via infrared beam breaks from 24 locations, 7.62 cm apart. Position was computed as the average value of the location numbers of the interrupted infrared signals. Distance traveled (Dist.) was computed as the absolute value of the difference between successive time-stamped rat-position values multiplied by 7.62 cm. Distance was summed during each 6-min bin. Ambient temperature at the rat's position within the gradient (Tsel) was logged at the time the rat's position was recorded. Tsel was calculated as the mean temperature of the thermistor(s) corresponding with the interrupted infrared beam location(s). Tc data were recorded at 30-s intervals. Median Tc and mean Tsel values were computed within each 6-min bin. Gas concentration data were recorded from each gradient at 1-min intervals.

Total calorimetry and Tc

Dependent variables obtained from the calorimetry tests were Tc, HP, dry heat loss (DHL) and evaporative heat loss (EHL). Tc was recorded at 15-s intervals and mean Tc was calculated for each 6-min bin. HP and HL data were recorded at 10-s intervals. Average HP and HL were calculated for each 6-min bin. Gas concentration data were recorded from each calorimeter at 1-min intervals.

Statistical analyses

The correlated within-subjects longitudinal data were analyzed using the linear mixed-model program in SPSS Statistics 20 (IBM, Somers, NY). Session and condition were treated as fixed effects. Unless otherwise specified, unstructured covariance matrices were employed for statistical comparisons because variances for thermal outcomes differed between N₂O and control-gas conditions. For comparisons between N₂O and control-gas conditions, means and 95% confidence intervals were adjusted for baseline values.

An experimental period was defined as the 3-h interval (0 to 180 min) in which 60% N₂O or control gas was administered. For thermal-gradient studies, statistical analyses involved 4 time periods: baseline (−60 to 0 min), early-experimental period (0 to 90 min), late-experimental period (90 to 180 min), and post-experimental period (180 to 240 min). For calorimetry data, early- and late-experimental periods were defined as above, and the baseline period was the 12 min immediately prior to the experimental period (the shorter baseline allowed for abatement of the initial hyperthermia associated with handling during placement into the calorimeter 120 min prior to the experimental period). The first 2 6-min bins of HP and HL data after the onset of N₂O were omitted from analysis due to their potential for artifact.¹⁹ There was no post-experimental period for the calorimetry test sessions.

Normally-distributed data (Tc, Tsel, HP, HL and Δdistance) were summarized as means with 95% confidence intervals (CI) to convey the magnitude and uncertainty range of each outcome. Distance magnitudes were summarized in terms of medians ± 05th percentile (p05) and 95th percentile (p95). Baseline values were defined as the means over the 60 min prior to N₂O onset for normally distributed thermal gradient outcomes (median for distance) and the 12 min prior to N₂O onset for total calorimetry outcomes. The null hypothesis for baseline-adjusted comparisons was that N₂O = control. Accordingly, 95% confidence intervals for N₂O compared to control conditions that exclude zero are significant at P < 0.05, 2-tailed. We did not adjust for multiple comparisons due to the conundrums and misplaced emphasis that accompany this class of procedures when implemented in the context of basic preclinical research^33-35 [see Part III of the online supplement for additional details]. Readers are urged to judge our results on the basis of the 95% confidence intervals and their coherence across sessions.

Results

Qualitative patterns of Tc and Tsel during N₂O administrations

Figure 1 illustrates the temporal dynamics of Tc and Tsel over 3-h sessions with 60% N₂O in the thermal gradient. The evolution of patterns of Tc within and across N₂O inhalation sessions in the thermal gradient are similar to those we have observed in calorimetry experiments at typical lab temperatures (∼21–22°C).²² Specifically, in Session 1 the Tc of drug-naïve rats initially decreased rapidly from baseline, achieving a nadir by 30–45 min, and subsequently returned toward baseline, consistent with the development of acute tolerance. In subsequent sessions, rats developed chronic tolerance to the hypothermic effect of N₂O, and this stage of adaptation became fully expressed by sessions 4–6 (i.e., Tc remained commensurate with baseline during N₂O inhalation); subsequently, Tc exhibited a hyperthermic sign-reversal during N₂O inhalation (Sessions 7–12).

Figure 1.

Temporal profiles of core temperature (Tc) and selected ambient temperature (Tsel) with 95% confidence bands during 60% N₂O administrations in the thermal gradient (n = 16). Rats were housed in the active thermal gradient for a minimum of 20 h prior to N₂O administration and had ad libitum access to food and water throughout. The data collected during the 6-min time bin prior to N₂O onset (plotted at −3 min) and during the 6-min time bin after N₂O onset (plotted at +3 min) are not connected by a line segment, which facilitates visualizing the rapid and large changes that can occur with drug delivery.

In Session 1, baseline Tsel was 28.1°C ± 1.61 (95% CI). N₂O promptly resulted in the rat's selecting a cooler ambient temperature, which, after an initial sharp decline, eventually settled to approximately the typical lab temperature during the early period of N₂O administration (see Fig. 1). Tsel gradually returned toward baseline during the latter half of N₂O exposure. During subsequent administrations, rats consistently selected a cool location throughout the first hour of N₂O inhalation and then gradually selected less cool temperatures during the remainder of the N₂O exposure.

Quantitative assessment of Tc, Tsel and distance traveled during N₂O inhalation

Figure 2 depicts mean ± 95% CI for Tc, Tsel and the median ± p05 and p95 distance outcomes for the early, late and post-N₂O periods averaged within each N₂O and control gas session. Figure 3 provides the formal statistical analysis of the differences between control and N₂O sessions. Baseline Tc and Tsel did not differ between N₂O and control sessions. These baseline values gradually decreased over sessions as the rats gained body mass (as best visualized in Figure 2, baseline Tc and Tsel). On the initial N₂O inhalation, rats promptly moved to significantly cooler ambient temperatures that facilitated the development of hypothermia (Fig. 3, early period Tc and Tsel), as reported previously.³² The rats continued to select a cooler ambient temperature during the early period across all sessions (Fig. 3, early period Tsel), even as Tc was exhibiting tolerance and eventually a hyperthermic sign-reversal during the early period of N₂O inhalation (Fig. 3, early period Tc). Indeed, in the early N₂O measurement interval, Tsel was consistently and substantially depressed by 5–7°C. Tsel did not differ from control levels during the late and post-exposure periods for the first several N₂O sessions. However, by the fifth or sixth N₂O session, Tsel started to become cooler during the late and post-exposure periods (Fig. 3, late and post-exposure periods Tsel). For example, Tsel was robustly depressed compared to control values during the post-exposure period for the final 3 N₂O sessions. During the late period, although chronic tolerance developed for Tc over the first several N₂O exposure sessions, Tc remained modestly decreased relative to control levels for 5 of the last 9 sessions (Fig. 3, late period Tc). During the post-exposure period, Tc recovered over the first 5 N₂O-exposure sessions and then remained modestly, yet significantly, decreased relative to control levels for 4 of the remaining 7 N₂O administrations (Fig. 3, post-exposure period Tc). It is notable that Tsel decreased over sessions during the post-exposure period while during that same interval, Tc recovered over sessions so as to be slightly cooler than control levels. In summary, over repeated N₂O-administration sessions tolerance developed and was followed by a hyperthermic overcompensation of Tc despite the persistent selection of a cooler Tsel. During the late and post-exposure periods of the later N₂O administrations, rats had a slightly reduced Tc (i.e., they were no longer hyperthermic) while exhibiting an increasing preference for cooler ambient temperatures.

Figure 2.

Core temperature (Tc), selected ambient temperature (Tsel) and distance (Dist.) traveled during N₂O and control gas sessions. Values for Tc and Tsel are unadjusted means with 95% confidence intervals based on repeated measures multiple linear regression analysis using unstructured covariance matrices within N₂O and control gas conditions computed separately for each measurement interval. Distance was non-normally distributed and is therefore presented as median with P05 and P95 limits (n = 16).

Figure 3.

Statistical analyses involving linear mixed model repeated measures analyses of thermal and distance outcomes (n = 16). The top 2 rows depict means with 95% confidence intervals for N₂O minus control gas (C) differences for core temperature (Tc) and selected ambient temperature (Tsel) for each of the 12 N₂O sessions. For each outcome in each of the 12 N₂O sessions, the mean and confidence interval represents the contrast with the average effect of the 4 control gas sessions. Outcomes for the baseline pre-N₂O administration period were adjusted for N₂O and control gas session numbers, while outcomes for the early, late and post-N₂O intervals were adjusted for baseline values. The analysis for distance-traveled compared change (Δ) from baseline values between N₂O and control sessions adjusted for baseline (Δ distance scores were normally distributed). 95% confidence intervals that do not contain zero signify N₂O sessions that were significantly different than control gas sessions at P < 0.05.

There was a dramatic increase in locomotion in the early N₂O interval, a modest but significant increase during the late N₂O interval, and little difference from control levels in the post-exposure interval (Fig. 3, Dist.). Increased metabolic rate yoked to locomotion would presumably generate heat, but is unlikely to explain tolerance or the hyperthermic sign-reversal of Tc. This is because: (1) a significant and substantial increase of activity occurred in early sessions in which hypothermia was maximal (Fig. 3); and (2) although the increase in activity during early N₂O administration was marked, the estimated metabolic cost of transport in rats of 2–3 m of locomotion per 6-min bin (∼0.3–0.5 m/min) is estimated to be a modest ∼0.1 W.³⁶

Calorimetry testing following the thermal gradient sessions (Fig. 4) revealed a hyperthermic overcorrection of Tc during N₂O exposure similar to the hyperthermia that eventually developed during the early period of N₂O inhalation in the thermal gradient. Consistent with our previous work,²² the early-period hyperthermic Tc in drug-adapted rats was primarily underlain by an increase of metabolic HP that exceeded the effect of the drug to augment HL. However, the late-period HL was not elevated compared to control values (Fig. 4) implying the existence of a gradual within-session adaptation that reduces heat conductance, as suggested previously.²² This adaptation appears to work in concert with a gradual waning of HP during N₂O administration so as to favor an eventual rebalancing of Tc at or near its control value.

Figure 4.

Core temperature (Tc), heat production (HP) and heat loss [HL, with evaporative HL (EHL) and dry HL (DHL) depicted separately] during total calorimetry and temperature testing at the conclusion of the 16 thermal gradient sessions. The black bar indicates the interval of 60% N₂O administration. Temporal profiles and line graphs are unadjusted means with 95% confidence intervals from repeated measures linear regression analysis with inhalation condition, and for temporal profiles, time, as repeated factors. HP and HL are not depicted for the first 12 min of the N₂O administration owing to a potential for artifactual changes therein (see Methods). Text in each line graph specifies effect size in terms of the difference (Δ) and 95% confidence interval (in parentheses) between the control gas and N₂O test sessions adjusted for baseline values based on linear mixed model repeated measures analysis (n = 14).

Discussion

The present work provides novel evidence that serial administrations of an initially hypothermic drug engender an adapted biobehavioral state in which autonomic and behavioral thermoeffector responses are pitted against each other during subsequent drug administration. This scenario is inconsistent with regulatory models according to which effector responses develop and act in coordination so as to efficiently defend homeostasis in the face of a disruptive agent.

A homeostatic model that includes an adjustable regulated level or ‘set-point’³⁷ would describe N₂O as causing a “regulated hypothermia”.^32,38 In this view, the initial N₂O administration causes the set-point for Tc to be reduced, eliciting a coordinated increase of HL and lowered Tsel to efficiently facilitate decreased Tc. During the continuous steady-state N₂O exposure, acute drug tolerance develops, gradually returning the set-point toward pre-drug values. The resulting discrepancy between the recovering set-point and the hypothermic Tc activates homeostatic corrective responses that raise Tc. Specifically, autonomic effectors are recruited that increase HP, and the rats move to less cool ambient temperatures (i.e., Tsel recovers from its nadir). Both of these adaptations contribute to the intrasessional recovery of Tc (Fig. 1, Session 1).

In contrast to this homeostatic interpretation of an initial N₂O exposure, the current findings implicate a different interpretation of the regulatory changes that occur over repeated N₂O exposures. When viewed over 12 individual N₂O administrations, it becomes evident that Tc can change independently of Tsel (Fig. 1). The pattern of Tsel during a 3-h N₂O administration is remarkably similar across all sessions, decreasing promptly with the onset of N₂O, reaching a nadir within the first hour and then gradually returning toward baseline value over the subsequent 2 h. In contrast, whereas Tc attains hypothermia during the initial N₂O session, chronic tolerance with no hypothermia is seen in Sessions 4–6, and this is followed by an early hyperthermia in Sessions 7–12. However, the cost-effective behavioral strategy of moving to a warmer environment is never utilized to facilitate the recovery or elevation of Tc across sessions. Rather, Tsel opposes the recovery and ultimate sign-reversal of Tc. The final N₂O session using the total calorimeter revealed that the rats were generating increased HP that mediated the hyperthermic Tc (Fig. 4). This finding is consistent with previous calorimetric research^21,22,29 suggesting that HP is an acquired compensation that grows over repeated administrations and contributes to the development of chronic tolerance as well as to the eventual hyperthermic overcompensation of Tc. Thus, in this situation, cool-seeking behavior is dis-coordinated with HP effector activity in the regulation of Tc.

Another example of dis-coordinated effector activity relates to the progressive decease in Tsel during the post-exposure period relative to the gradual increase of Tc toward control levels that occurs during that same period over the 12 sessions (Fig. 3). In contrast to the consistent effect of early-period Tsel across sessions, post-exposure Tsel changed over sessions and eventually became a cool preference of comparable magnitude to that observed during the early period of N₂O exposure. Thus, when N₂O delivery ceases, motivated behavior for cooler ambient temperatures increases over sessions while other concurrently active regulatory influences continue to support the recovery of Tc toward control levels.

These findings are not easily reconciled with traditional homeostatic interpretations as recently reviewed.¹⁰ If Tc can be more efficiently regulated by adjusting Tsel than via changes in autonomic HP effector activity, why is HP rather than Tsel the primary mechanism accounting for chronic tolerance? Why should HP effectors and Tsel be in concurrent competition with each other when well-coordinated effector responses are a hallmark of homeostatic regulation? In Sessions 4–6, elevations in HP and possible heat-conserving adaptations are sufficient to offset the cool Tsel, thereby establishing a thermal balance that is able to maintain Tc at baseline/control levels during N₂O administration. Without a perturbation of Tc during Sessions 4–6, what drives the further adaptations that eventually lead to the thermoeffector imbalance that causes a transient hyperthermic sign-reversal of Tc during Sessions 7–12? While a transient hyperthermic overshoot could be interpreted as hysteresis resulting from time lags in homeostatic regulatory effector activity, this cannot explain why adaptations that effectively establish homeostasis during sessions 4–6 do not exhibit hysteresis.

We suggest that the explanation for these inconsistencies is that the principles of homeostatic regulation do not apply to all situations, especially to non-naturalistic experimental challenges such as those involving the delivery of pharmacological agents.^6,10,13,32 In fact, the current results are more consistent with the view that the thermoeffector loops regulating Tc are relatively independent of one another^{14,15,39,40,41} and that Tc represents a balance point rather than a set-point defended by coordinated effector activity. The point is that there are current models of regulation that incorporate the relative independence of regulatory effectors and are thus able to accommodate occurrences of dis-coordinated effector responses that work in opposition to one another.¹⁰ In this schema, aberrant challenges to evolutionarily-derived regulated systems can trigger dis-coordinated effector responses, and this has been suggested to be a characteristic of a non-homeostatic form of regulation called allostasis.¹⁰ Goldstein⁴² has described the inefficient cost of allostasis by analogy to regulating a home's temperature with competing effectors (e.g., the furnace and the air conditioner) being concurrently active. This is a fitting metaphor for the findings of the current study where concurrent motivated cool-seeking behavior opposes increased autonomic HP responses during the development of both chronic tolerance and the eventual transient hyperthermic sign-reversal. Importantly, sign-reversals of regulated variables have been suggested to reflect the existence of allostasis.¹⁰

An allostatic model of drug addiction can explain how motivational consequences arise that encourage drug-taking behavior.¹⁰ Overactive compensatory responses that lead to sign-reversals have been suggested to motivate drug-taking behavior; i.e., increased drug-taking yields a greater pharmacological effect that can oppose the sign-reversal state. In other words, the behavioral effector of drug taking can oppose the overactive effectors that caused the sign-reversal state (i.e., there is concurrent opposing-effector activity). Taking additional drug may temporarily ameliorate the sign-reversal, but it also triggers increased effector activity that eventually restores the sign-reversal. Thus, an allostatic model can include a vicious cycle hypothesis for the escalation of drug taking seen in addiction.^43-49 The findings of the current study indicate that a cool Tsel opposes the hyperthermic Tc sign-reversal and may reduce the magnitude of hyperthermia. Presumably, the hyperthermic Tc could have been reduced further if the rat had selected an even cooler Tsel (i.e., there was not a floor effect at the cool end of Tsel). Since this did not occur, the magnitude and duration of the hyperthermic Tc may reflect an allostatic balance point that develops over repeated exposure with both autonomic effectors and the thermal gradient behavioral effector being concurrently available. A subsequent study⁵⁰ assesses this hypothesis in a different way by determining whether a sign-reversal state established during N₂O administration at typical laboratory temperatures (∼21°C) is altered once a powerful behavioral effector provided by a thermal gradient becomes available during N₂O administration.

In conclusion, rats did not facilitate chronic tolerance development to N₂O-induced hypothermia by moving to warmer locations in the gradient, and instead selected cooler ambient temperatures while simultaneously increasing autonomic HP. The inefficient concurrent activation of opposing effectors and the development of a sign-reversal are incompatible with homeostatic models of drug-adaptation and may be better interpreted using a model of drug-induced allostasis.

Funding Statement

This investigation was supported by the National Institutes of Health (NIDA grant DA023484).

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

We gratefully acknowledge Christopher W Prall and Hoang Yen Ho for their technical contributions to this study. The Helen Riaboff Whiteley Center, located at the University of Washington's Friday Harbor Laboratories, provided an ideal environment for completing this manuscript. Supplemental data for this article can be accessed on the publisher's website