High throughput modular chambers for rapid evaluation of anesthetic sensitivity

The mechanisms that give rise to anesthetic-induced unconsciousness have eluded understanding for more than 150 years. Breakthroughs in circadian timekeeping and sleep neurobiology have been made in recent years using genetic approaches to unravel mutant phenotypes in mice [1–4]. Genetic approaches offer an unparalleled opportunity to elucidate complex biological problems such as anesthetic action. One characteristic of successful behavioral genetic endeavors has been the development of rapid screens to identify mutant phenotypes [5]. A common feature of these rapid screens is that they have motor-dependent endpoints. Anesthetic-induced unconsciousness is typically assessed using a loss of righting reflex (LORR) assay. As with wheel running, this endpoint depends upon movement and is amenable to full automation – another hallmark of successful forward genetic screens [5]. The LORR assay has been used with mice to identify genetic mutations that cause hypersensitivity [6, 7] or resistance [8] to various anesthetic agents.

In humans, general anesthesia is typically separated into three phases: induction, maintenance, and emergence. Animal studies of anesthetic sensitivity have failed to recognize these three phases, likely due to the inherent difficulty in maintaining intravenous or intraperitoneal steady-state levels of typical anesthetic agents such as propofol, etomidate, barbiturates, or ketamine. However, with inhaled anesthetics, steady-state drug levels are easily achieved. Multiple studies have attempted to decipher molecular mechanisms crucial for induction of anesthesia (reviewed in [9–12]), yet only rare reports characterize mechanistic changes that occur upon emergence from anesthesia [13–16]. Many clinical case reports suggest that a variety of conditions can modulate emergence from anesthesia with the most common finding being delayed emergence in specific subsets of patients [17–21]. There is a paucity of animal data investigating pharmacologic treatments which might modify emergence [22, 23]. To date, there is only a single genetics study investigating particular phenotypes of mice that exhibit altered emergence patterns following general anesthetics [24]. Greater insight into the mechanisms through which anesthetics induce unconsciousness as well as the mechanisms leading to its return are necessary to gain understanding of the entire process of general anesthesia.

In this study we report the novel design and testing of high throughput controlled environment chambers that will allow detailed phenotypic characterization of murine sensitivity to induction as well as emergence from general anesthesia. This apparatus should facilitate both unbiased forward genetic screens and putative anesthetic target reverse genetic screens to evaluate molecular mechanisms of inhaled anesthetic action. We have taken great care to control for known confounders of anesthetic sensitivity such as body temperature, circadian time, and anxiety of mice in an unfamiliar testing environment. Using these chambers we have studied the wild type sensitivity to induction and emergence from 3 volatile anesthetics in C57BL/6J male mice, aged 8–12 weeks.

Prior to putting mice in the modular anesthetizing chambers, we verified the parallel circuit design by confirming that each of the 24 chambers received equivalent gas flow, Tables 1 and 2. By placing an anesthetic vaporizer upstream of the parallel branching, identical concentrations of volatile anesthetics were delivered to each of the 24 chambers. The anesthetizing circuit was an open design predicted to have laminar flow based upon a calculated Reynolds number of 2.8. With an average of 108 ml/min fresh gas flow applied to each chamber, expired CO₂ was measured from each mouse by sampling individual chamber outflow gases. The maximal CO₂ measurement ever obtained was 14 torr (1.8%) upon the first entry of a mouse into its chamber. The average CO₂ outflow measurement was 8.4 ± 0.5 torr on the first day of habituation. Over the course of 4 successive days, CO₂ levels in chamber egress gas dropped from 8.4 ± 0.5 torr to 4.8 ± 0.1 torr. This suggests that CO₂ production decreased as the mice become accustomed to the confined chamber, figure 1.

When placed in the chambers and exposed to 100 ml/min fresh gas flow, unanesthetized mice drop their core body temperature from 37.3 ± 0.1°C to 36.6 ± 0.1°C. Without active heat transfer, mice exposed to 1.25% isoflurane in dehumidified oxygen for two hours were found to cool to 25.4 ± 0.4°C, which approaches room temperature (table 3). Since sensitivity to anesthesia is significantly modulated by temperature, we designed a means of conductive heat transfer by placing the chambers in a heated water bath. With the water bath set to 37°C, we were able maintain mice at 36.6 ± 0.1°C, eliminating hypothermia as a confounder regardless of the duration of the anesthetic exposure.

Having optimized the anesthetizing chambers with appropriate environmental controls (temperature, humidity, and inhaled gas composition) we subjected mice to MAC_LORR determinations as described in the methods. Figure 2 shows the population response for wild type C57BL/6J mice exposed to halothane, isoflurane, and sevoflurane. Best-fit parameters for data are displayed in figure 3. Emergence from anesthesia as defined by recovery of the righting reflex is shown in figure 4. Cumulative anesthetic doses prior to emergence were 2.4 ± 0.1, 2.8 ± 0.1, and 2.7 ± 0.3 MAC_LORR • hrs for halothane, isoflurane, and sevoflurane respectively.

We have constructed and validated modular chambers to conduct high throughput screens of anesthetic-induced loss of righting reflex and its subsequent return. Using inhaled anesthetic gases, we studied induction and emergence from anesthesia by measuring the following parameters: 1) MAC_LORR, defined as the minimum alveolar concentration at which 50% of the test population has lost its righting reflex; 2) Hill slope which evaluates the steepness of the transition from wakefulness to general anesthesia for the test population; and 3) emergence time_RR, defined as the duration of time before a population regains its righting reflex. Whereas most studies of anesthetic sensitivity evaluate MAC (MAC_immobility), defined as the minimum alveolar concentration required to prevent movement in 50% of a population in response to a noxious stimulus such as a tail pinch [27–29], we are primarily interested in the hypnotic component of anesthesia. As outlined by Mashour and colleagues, the hypnotic component of anesthesia is most precisely defined as a lack of "perceptive awareness" [10] and should therefore be evaluated following a non-noxious stimulus such as turning an animal prone to supine to examine its righting reflex. The righting reflex provides one behavioral measure commonly used to assess hypnosis in rodents. However, when used alone this assay could yield false positive interpretations of behavioral state. If one administered a muscle relaxant by itself, treated mice would exhibit LORR while remaining fully aware. Hence, for a formal genetic analysis, MAC_LORR and emergence time_RR data should be combined with a second index of behavioral state such as EEG-based analyses. An EEG-based secondary screen would reveal cortical signatures of anesthetic action that should compliment the righting reflex's brainstem signatures [30].

Our reported values for MAC_LORR are in good agreement with values obtained by Homanics and colleagues using a similar best-fit approach for sigmoidal dose response data [31]. The small difference in MAC_LORR for halothane (our determination is 13% greater than theirs) is easily accounted by the different genetic background of their mice which could influence anesthetic sensitivity by as much as 40% [32]. Our best fit values for Hill coefficients are within the expected range for volatile anesthetics of 20–30 [33]. Lastly, the "goodness of fit" for our behavioral data as measured by r² obtained with our apparatus is high relative to the literature (figure 3). We attribute these high best-fit values to reductions in experimental error both due to the uniform delivery of experimental conditions among all mice in our controlled environment anesthetizing chambers on a given day and to the reproducibility of conditions achieved with our chambers on different days.

Several other groups have sought to determine MAC_LORR in mice. While attempting to assess the hypnotic component of anesthesia in mice with mutations in the AMPA receptor, GluR2 [6], and also in the nicotinic acetylcholine receptor subunit, beta 2 [34], lower values for MAC_LORR were reported. Joo and colleagues quoted MAC_LORR for isoflurane at 0.57 ± 0.16%, and sevoflurane at 1.16 ± 0.19% in their wild type controls. Their range for halothane of 0.67 ± 0.20% was similar to ours. Also in agreement with Joo's MAC_LORR findings are those of Flood and colleagues [34] who found MAC_LORR of isoflurane in inbred 129 mice was 0.56 ± 0.10% and MAC_LORR of isoflurane in C57BL/6J mice was 0.63 ± 0.20%. One major methodological difference could account for the discrepancy in MAC_LORR. Both Joo and Flood used a bracketing measurement, which averages the highest anesthetic concentration at which mice still retain the righting reflex with the lowest anesthetic concentration at which mice have lost the righting reflex to estimate MAC_LORR. If the incremental anesthetic concentration step size is large, systematic error in MAC_LORR determination can't be avoided, as Hill slopes for anesthetic drugs are notoriously steep.

In creating our modular, high throughput, controlled environment, anesthetizing chambers, we were confronted with several significant design problems. The first revolved around the high metabolic rate of mice. With a minute ventilation of approximately 30 ml/min [26] and a carbon dioxide production rate, V_CO2, of 1.3 ml/min at rest [35, 36], any arrangement of mouse chambers in series would lead to rebreathing at low fresh gas flows, evaporative and convective heat loss at high fresh gas flows, and heterogeneous conditions for mice at different positions relative to the fresh gas inlet. Thus, it became clear that the mice must be arrayed in parallel. Even with each animal having its own fresh gas source, carbon dioxide handling issues remained. We reasoned that during initial exposure to the chambers, mice would experience a mild restraint stress caused by confinement in the 250 ml tubes (figure 1). This would increase carbon dioxide production. With repeated exposure during 4 daily habituation sessions, measured expired carbon dioxide levels from all mice decreased, consistent with a decrease in V_CO2 production during habituation. Further, we reasoned that at 100 ml/min fresh gas flow in each chamber, convective flow should vastly exceed the capacity of CO₂-rich expired gases to back diffuse into the inspired fresh gas flow, thus reducing rebreathing. Nonetheless, even assuming the worst-case scenario of perfect mixing in each chamber, and a maximal murine V_CO2 of 100 ml·kg^-1·min^-1, the theoretical upper limit for inspired CO₂ in our chambers for a 25 g mouse receiving fresh gas flow at 100 ml/min is 2.5%. Our data suggests that the mice only produce enough CO₂ to occupy up to 1.1% on initial entry in the chambers on day one of habituation and that this declines to 0.6% on the last day of habituation. It should be noted that these values represent the maximum theoretical upper limit for inspired CO₂. Actual inspired CO₂ concentrations will be much lower as net fresh gas flow will be laminar across any given chamber and convective flow should predominate.

Another challenge posed by this open circuit setup of our chambers is the enormous potential for convective heat loss in anesthetized mice. Unanesthetized mice, capable of normal thermoregulation, experienced only a small decline in body temperature (0.6°C), which was not significantly different if the chambers were submerged in a 37°C water bath. While the awake mouse maintains its core body temperature through vasoconstriction, anesthetic gases inhibit central and peripheral thermoregulation [37]. Therefore, in the anesthetized mouse, heat transfer to the chambers submerged in the water bath is essential to maintain normothermia.

Whereas we have chosen to use these controlled environment chambers to deliver an inhaled anesthetic and measure hypnotic responses of wild type mice, this system could easily be used to deliver any inhaled therapeutic or toxin in any setting where precise control over the testing environment is necessary. With our parallel open circuit design, gas flow is evenly divided between the 24 individual chambers. However, there is no reason why the number of chambers is limited to 24. We have recently increased the capacity to 30 chambers while still maintaining identical individual microenvironments. Gas composition, humidity, and test subjects' temperatures can be easily regulated. Moreover, by implanting telemetric transmitters in mice, it is possible to stream other physiologic data from each of the 24 pods to corresponding detectors. In this way, vital signs such as blood pressure, heart rate, respiratory rate, core body temperature, as well as electroencephalographic and electromyographic signals that may be altered by the delivery of an inhaled therapeutic can be continuously monitored. We believe that development of apparatus along these lines will enhance our ability to rapidly dissect the pharmacogenetics of volatile drug action.

This study lays the foundation for rapid screening of genes that alter anesthetic sensitivity in mice by establishing a reproducible, high throughput system for the analysis of anesthetic-induced loss and subsequent recovery of the righting reflex, a commonly used behavioral correlate of anesthetic-induced unconsciousness in rodents.