A low-cost mouse cage warming system provides improved intra-ischemic and post-ischemic body temperature control – application for reducing variability in experimental stroke studies

Abstract

Background:

Brain temperature is a strong determinant of ischemic stroke injury. For this reason, tight management of brain or body temperature (Tcore) in experimental rodent stroke models is recommended to improve the rigor and reproducibility of outcomes. However, methods for managing Tcore during and after stroke vary widely in approach and effectiveness.

New Method:

We developed a low-cost warm ambient air cage (WAAC) system to provide improved temperature control during the intra-ischemic and post-ischemic recovery periods. The system is incorporated into standard holding cages for maintaining Tcore during the intra-ischemic period as well as for several hours into the recovery period.

Results and Comparison with Existing Methods:

We compared the WAAC system with a commonly used heat support method, consisting of a cage on a heating pad. Both heat support systems were evaluated for the middle cerebral artery occlusion (MCAo) stroke model in mice. The WAAC system provided improved temperature control (more normothermic Tcore and less Tcore variation) during the intra-ischemic period (60 min) and post-ischemic period (3 hrs). Mean infarct volume was not statistically different by heat support system, however, standard deviation was 54% lower in the WAAC system group.

Conclusions:

Mice and other small rodents are highly vulnerable to heat loss during and after the MCAo procedure. The WAAC system provides more precise and controlled Tcore maintenance compared with frequently used induction heating methods in mice undergoing the MCAo stroke model. The improved temperature control should enhance experimental rigor and reduce the number of experimental animals needed.

Introduction:

Many neuroprotective agents or strategies have shown significant benefit in experimental stroke models, only to fail to demonstrate similar benefit in clinical trials. As a result, an evolving set of experimental guidelines have been proposed by multiple investigators or groups in an effort to increase the rigor of stroke studies and thereby increase the probability for translation into effective clinical therapies(Dirnagl, 2006; Fisher et al., 2009; Kahle and Bix, 2012; Liu et al., 2009; Percie du Sert et al., 2017; Stroke Treatment Academic Industry Roundtable, 1999). One significant component of these recommendations includes more rigorous monitoring and control of key physiological variables (including body and brain temperature) in stroke models. Because body/brain temperature is highly correlated with stroke outcome(Buchan and Pulsinelli, 1990; Busto et al., 1987; Busto et al., 1989; Cao et al., 2014; Cao et al., 2017; Liu and Yenari, 2007), variability in temperature (intra-ischemic and post-ischemic periods) can lead to significant variability in outcomes. These effects can be especially pronounced in small rodents (rats and mice) due to their potential for rapid and significant temperature changes (Hankenson et al., 2018). However, current recommendations are vague or insufficient regarding temperature measurement or maintenance during the intra-ischemic or post-ischemic period. For example, in the middle cerebral artery occlusion (MCAO) model, it is recommended to provide feed-back controlled heat support during surgery to introduce and remove the occluding filament, but temperature management during the intra-ischemic awake portion or during the first few hours of reperfusion are not specifically addressed. Given the high dependence of ischemic injury on intra- and post-ischemic brain and body temperature, we suggest that more consistent measurement and tighter control of body temperature should be a prominent component of the overall experimental design for reducing variability in experimental stroke studies.

The present study sought to evaluate the effect of tight temperature control during the full intra-ischemic period and the first 3 hours of reperfusion on stroke outcome variability. We developed a novel low-cost surgical recovery cage, consisting of a modified programmable warm air circulator that can be placed in a standard rodent cage. We compared this warmed ambient air cage (WAAC) to a commonly used temperature support approach, consisting of a cage placed partially on a heating pad. The WAAC system provided more consistent and precise temperature support versus the conventional “cage on a heating pad” strategy. We found that housing mice in these recovery cages provided superior control of body temperature (measured by wireless transponder) and less variance in outcome measures compared with heating pad temperature support, particularly in mice with more severe stroke injury. Our results support the benefit of tighter body temperature control for reducing variability of stroke outcome measures. Lastly, we describe how to produce this low-cost cage warming device to facilitate its use in any relevant laboratory.

Materials and Methods:

Warmed Ambient Air Cage (WAAC) Design.

The warmed ambient air cage (WAAC) was created by installing a regulated heating module on the wire rack of a ‘shoe box’ mouse cage. The heating module consisted of a thermostat, heater, and small circulating fan designed for temperature control of reptile habitats (IncuKit MINI, IncubatorWarehouse.com) (Fig 1). The fan was configured to direct air upward and then around the cage (i.e. not directly blowing on the mouse). However, as provided, the fan was too strong for the small mouse cage. We therefore modified the module by adding a 3D printed baffle to reduce the circulating air flow (Fig 1B). The effect of the added baffle was evaluated by measuring the clearance rate of dry ice vapors in the cage (not shown). The feedback-controlled temperature on the module was set to 33 °C. The temperature probe for the heating module was placed beneath the wire rack approximately centered over the cage.

Figure 1:

Assembly of warm ambient air cage (WAAC). (A) Top view of WAAC demonstrating location of heating module. (B) Heating module top view and bottom view with 3D-printed plastic fan baffle. (C) Side view of WAAC with indicated location of thermocouples for temperature characterization experiments. One thermocouple was placed 2 cm above the bedding (T_air) and another placed under the top layer of the bedding (T_bedding).

Heating Pad Warmed Cage Design.

A widely used method for providing post-surgical heat support to rodents is to place the cage half on and half off a heating pad (see Fig 2A). By doing so, it is assumed that the animal can warm on the heating pad side and then “escape” to the unheated side once the animal is either warm enough or too warm. To replicate this commonly used set up, we placed the cage half on a standard heating pad which was set to the “low” setting (Sun Beam).

Figure 2:

Comparison of heating characteristics for the WAAC and heating pad cage designs. (A) Drawing of the WAAC and heating pad systems. (B) Rate of ambient air warming (T_air, measured above bedding). Time 0 represents the point where both systems were turned on. (C) Rate of bedding warming (T_bedding, measured under top layer of bedding). For the heating pad system, bedding temperature was measured on the warming side (placed over the heating pad) and the unheated “escape” side (off the heating pad). The arrows represent the time for each system to reach within 1 °C of the steady-state temperature. Data are presented as mean ± SD from 3 separate measurements.

Cage Temperature Characterization.

For cage temperature characterization, we measured temperature at two points where a mouse would experience heat transfer – at the surface layer of the cage bedding and in the air just above the bedding. The location of these two thermocouples can be seen in Fig 1C. Temperature at these two locations was monitored by thermocouple and a temperature logger (HH806AU, Omega Engineering, Stamford, CT). Temperatures were recorded at 5 min intervals for up to two hours after turning on heating module or heating pad.

Animal preparation and experimental groups.

The experimental protocols were approved by the Institutional Animal Welfare Committee at the McGovern Medical School at UTHealth. A total of 41 male C57BL6/J mice (12 weeks old, Jackson Laboratory) were included in the study. The animals were allowed free access to food and water. Animals were randomly assigned to four experimental groups according to heat support method (WAAC or heating pad design) and severity of stroke model (mild or severe stroke). Mice were implanted with a small wireless capsule (~16x8 mm) in the peritoneal cavity (Anipill, BodyCAP) (see Supplemental Fig 1). Upon activation, the capsule logged core temperature at 15 min interval. Mice were allowed at least 7 days to recover before additional experiments. Mice were also implanted with a small temperature transponder (IPTT-300, BioMedic Data Systems, Seaford, DE USA), which permitted the wireless measurement of body temperature. The transponder temperature reading could be performed through the wall of the cage and thus avoided the effect of stress (handling and restraint, rectal probe insertion) on body temperature(Hankenson et al., 2018). The wireless temperature measurement also eliminated the effect of repeated opening of the cage and exposure of the animal to cooler ambient air. Temperature measurements were made by an investigator blinded to the stroke severity grouping.

Middle Cerebral Artery Occlusion (MCAo) Model of Mild and Severe Stroke.

Mice were initially anesthetized by inhalation of 5% isoflurane in 30% O₂/70% N₂O, and then maintained with 1.5% isoflurane during the procedure. The neck region was prepared and bupivacaine was injected subcutaneously at the surgical site 5 min before incision.

Rectal temperature was monitored and maintained at 37 °C by a feedback-controlled heating pad throughout the surgical portions of the procedure (Omega Engineering). During all surgical periods, the ability of the feedback-control unit to maintain rectal temperature within 0.5 °C of the 37 °C setpoint was confirmed by the digital readout. Outside of the surgical procedure times, core temperature (Tcore) was recorded and logged by an implanted wireless temperature telemeter (Anipill) or transponder (IPTT-300). The Anipill capsule was implanted in the peritoneal cavity and the temperature transponder (Implantable Programmable Temperature Transponder; IPTT-300, BioMedic Data Systems, Seaford, DE USA) was implanted subcutaneously on the lower dorsolateral region of the mouse. This location placed the transponder in direct contact with the thin tissue layer surrounding the peritoneal cavity and thus provides good correlation with core body temperature(Feketa et al., 2014). The correlation between Anipill (intraperitoneal) and IPTT-300 (subcutaneous) temperatures was also evaluated in the present study (Supplemental Fig 2). Linear regression of the two temperature methods showed good agreement with the two measurement methods, with a slope of 1.103 (P<0.0001).

Transient ischemic stroke was induced by occluding the middle cerebral artery (MCA) with a silicone-coated monofilament (MCAo model)(Fasipe Titilope et al., 2018). In brief, the occluder was introduced into the right external carotid artery (ECA), advanced back to the common carotid artery (CCA), and then advanced up the internal carotid artery (ICA) into the circle of Willis where it occluded the origin of the right MCA. The monofilament was left in place for 60 min and then removed to initiate reperfusion. Two occluder filaments were used to achieve the mild (0.21 mm diam, 2-3 mm coating length, 602123PK10Re, Doccol) or severe stroke injury (0.21 mm diam, 5-6 mm coating length, 602156PK10Re, Doccol). Once the occluder was in place, the mice were placed in either the WAAC or heating pad recovery cage according to their assigned experimental group. Successful occluder placement was confirmed by neurologic deficit test (see below) during the intra-ischemic period. At 60 min following occluder placement, mice were again anesthetized and the monofilament withdrawn. Mice were again returned to their respective heat support recovery cages for 3 hours. At the end of the 3 hour temperature support period, mice were placed in standard mouse “shoe box” cages on a static rack. Mice were singly housed to avoid variation of body temperature due to huddling or other group-related influences on temperature (Hankenson et al., 2018). Mice were given wet mash food and received daily saline for hydration (1ml, intraperitoneal). Mice were sacrificed for brain collection and analysis at post-stroke day 3.

Heat support during intra-ischemic and post-ischemic periods.

Temperature support during the intra-ischemic period (time between monofilament placement and removal) and in the post-stroke period was provided by the WAAC or the heating pad design (described above). Mice were maintained in the heat support cages for 60 minutes during the occlusion period and for 3 hours following removal of the monofilament (i.e. reperfusion phase). Mice assigned to the heating pad system were placed in a region of the cage that was centered over the heating pad. Body temperature was measured via the wireless transponder and Anipill at 15 min intervals in the intra-ischemic period and at 30 min intervals in the post-ischemic period. After 3 hours of heat support, mice were transferred to standard cages without heat support in a static rack. The housing room ambient temperature was maintained at 25-26°C.

Behavioral test (Neurologic deficit test).

Behavioral testing was performed to evaluate the degree of neurologic deficits based on a 12-point scoring system (12, maximal deficits; 0, no deficit)(Belayev et al., 1999). The test assessed postural reflex, forelimb placement by visual and tactile stimuli, and proprioceptive responses. We performed a neurologic deficits test just before reperfusion in order to confirm the successful occlusion. A priori exclusion criteria were set for mice a) not demonstrating a neurologic deficit score (NDS) of 12 at this time point, b) mouse barrel-rolling during MCAo or c) demonstrating evidence of vessel puncture by brain inspection at time of death. Two mice (one mouse in the heating pad system group with mild injury and one mouse in the WAAC system group with severe stroke) were excluded based on these criteria. Behavioral tests were additionally conducted at 24, 48 and 72 h after stroke.

Histopathology.

Mice were euthanized by overdose of isoflurane at day 3 after stroke. Brains were removed, sectioned at 1 mm thick coronal sections using a mouse brain matrix, and then incubated in 1% TTC (2,3,5-Triphenyltetrazolium Chloride, Sigma, in PBS) for 15 min at room temperature to evaluate infarct size. The stained sections were then fixed in 10% neutral buffered formalin before imaging with a stereomicroscope at 0.8X magnification (Olympus SZ-40). Cortical and subcortical infarcts in each section were outlined and measured by Image J. Infarct volume in the cortex and subcortex were calculated by multiplying infarct area in cortex or subcortex by sectional thickness and corrected for brain swelling (Belayev et al., 2003) as follows: Corrected Infarct Area = Infarct Area X (1- [(Ipsilateral Hemispheric Area – Contralateral Hemispheric Area)/Contralateral Hemispheric Area]). Severity of hemorrhagic transformation (HT) was graded on a 0 to 5 point scale by an investigator blinded to grouping (see Supplemental Fig 3). The severity of HT was scored from 0 (none) to 5 (severe) for each section and then expressed as a cumulative total HT score.

Statistical analysis.

Data was expressed as a mean ± standard deviation (SD) or as median with interquartile range. The mean body temperature during intra-ischemic and post-ischemic periods was analyzed by generalized estimating equation (GEE) method to address the within-subject correlation, followed by post-hoc group comparison at each time point. The homogeneity of variance in the two groups at each time point were evaluated by Bartlett test or Levene test depending on the normality. All p values were adjusted for multiple tests. NDS and HT score were analyzed by GEE method similarly, followed by post-hoc group comparison at each time point and comparisons between day 1, 2, 3 and baseline after MCAo. All tests were adjusted for multiple tests. For single time measurements, groups were compared by two sample t test or Wilcoxon rank sum test. Data analyses were performed in GraphPad Prism 8 (San Diego, CA) and SAS 9.4 software (Cary, NC).

Results:

In total, 41 mice were used in this study. Eight mice died prior to the planned end point at 3 days poststroke, including 4 mice from the WAAC group and 4 mice from the heating pad groups (see Table 1). The mild stroke injury resulted in very low mortality, with only one death in the heating pad group during the 3 days following stroke. The severe stroke injury group resulted in 3 deaths in the heating pad group (all on post-stroke day 1) and 4 deaths in the WAAC group (spread among post-stroke days 1 to 3). As noted in Methods, two mice were excluded from the study (n=1, heating pad group with mild stroke; n=1, WAAC group with severe stroke). No differences were observed in initial weight or weight loss after stroke among heating pad and WAAC groups (Supplemental Fig 4).

Table 1:

Number of deaths before experiment end on post-stroke day 3. Two mice (n=1, heating pad group with mild stroke; n=1, WAAC group with severe stroke) were excluded due to surgical causes.

	Number of premature deaths
Group (n)	Day 1	Day2	Day 3
Heating pad, mild stroke (9)	0	0	1
Heating pad, severe stroke (12)	3	0	0
WAAC system, mild stroke (9)	0	0	0
WAAC system, severe stroke (11)	1	1	2

The WAAC system provides more rapid and precise cage temperature equilibration compared with the heating pad system.

The rate of temperature stabilization was compared between the WAAC and heating pad cage systems (Fig 2A). To record temperature of bedding (T_bedding) and ambient air (T_air) from where a mouse would normally rest, one thermocouple was placed just under the top layer of bedding and another elevated two centimeters above the bedding. For the heating pad system, another thermocouple was placed in the bedding on the unheated “escape” side of the cage. Measurements were initiated at the same time that the respective heat support systems were powered on (time 0). The WAAC system heating module was set to 33 °C and the heating pad was adjusted to the “low” setting. The T_air in the WAAC reached the set point in under 10 min and remained extremely stable for the 2 hour period (Fig 2B). In contrast, the T_air in the heating pad cage responded more slowly, requiring 45 min to reach ±1 C of the steady-state temperature. The T_bedding increased to 30.85 °C in the WAAC, reaching ±1 °C of the equilibrated temperature in 30 min (Fig 2C). The T_bedding in the heating pad cage reached a higher temperature (41.5 °C) and took 75 min to reach ±1 °C of the steady-state temperature. The bedding temperature on the “escape” side of the cage remained largely unchanged.

The WAAC system provides more precise and stable control of mouse body temperature during the intra-ischemic and post-ischemic periods.

Given the strong correlation between brain temperature and ischemic injury, it is critical to take steps to reduce the variability of this physiologic variable in experimental stroke studies(Busto et al., 1987; Busto et al., 1989; Dirnagl, 2006; Fisher et al., 2009; Kahle and Bix, 2012; Liu et al., 2009; Percie du Sert et al., 2017; Roundtable, 1999). In this study, we measured body temperature as an easily accessible alternative to direct measurement of brain temperature in freely moving conscious mice. Brain temperature is influenced by multiple local factors (such as brain metabolic activity and blood flow) and can therefore differ in absolute temperature compared with core temperature or temporalis muscle temperature(DeBow and Colbourne, 2003; Jackson-Friedman et al., 1997; Kiyatkin et al., 2002). However, in non-stimulated conscious rodents, brain temperature generally parallels core temperature(DeBow and Colbourne, 2003; Kiyatkin, 2019, 2010), thus allowing body temperature measurements to be used as a reasonable approximation of brain temperature.

We measured Tcore in mice during and after 60 min of MCAO of either “mild” or “severe” stroke. The severity of stroke depended on the type of occluder used – the mild occluder had a 2-3 mm length of silicone tip, whereas the severe occluder had a 5-6 mm length. The longer silicone tip produces additional occlusion of the posterior cerebral artery (PCA) (Supplemental Fig 5), which results in the additional injury within the PCA territory. Heat support during surgery was provided by feed-back controlled heating pad and then by either WAAC or heating pad during the 60 minute intra-ischemic period and the first 3 hours of the post-ischemic (reperfusion) period.

In the severe stroke group, the WAAC system provided more effective and consistent heat support during the intra-ischemic period (Fig 3A, top). The mean Tcore (from 15 to 60 min) for the WAAC system mice was 36.9 ± 0.3 °C versus 36.1 ± 1.3 °C in the heating pad system. While the mean temperatures were not statistically different (P=0.154, t test), the variance in the mean temperature was significantly less in the WAAC group (Bartlett’s test, p =0.004). During the post-ischemic phase, the WAAC system produced significantly better Tcore maintenance within the normothermic range and lower Tcore variability compared with the heated cage system for the severe stroke group mice (Fig 3B, top). Repeated measures analysis showed superior temperature maintenance during the post-ischemic period, with statistical significance at all time points (P<0.05, Bartlett’s test). The mean Tcore over the entire reperfusion period was 35.9 ± 0.4 °C versus 34.2 ± 1.7 °C for WAAC and heating pad groups, respectively. The mean Tcore differed significantly (p=0.029, t-test), and the variability (SD value) was significantly lowered in the WAAC group (p=0.004, Barlett’s test). To avoid skewing the data, only mice that survived through PSD3 were included in the temperature analysis. Mortality rates during 3 days of MCAo were similar between groups (4 vs. 3, respectively). However, while mortality occurred exclusively in the first 24 hours in the heating pad group, mortality was spread across the 72 hour period in the WAAC group (see Table 1).

Figure 3:

Comparison of core temperature (Tcore) management by heat support system during the intra-ischemic and post-ischemic phase for mice that underwent the severe stroke protocol (A,B) or the mild stroke protocol (C,D). (A,C) Summary of intra-ischemic Tcore for WAAC and heating pad groups at different time points during the occlusion (left) and the mean Tcore over the entire occlusion period (right). The * indicates group differences in mean temperature; the ^† indicates significantly lower variance in the WAAC group (GEE method). (B,D) Summary of post-ischemic Tcore for the WAAC and heating pad groups at different time points (left) and over the entire reperfusion period (right). The * indicates group differences in mean temperature. All-time points in the WAAC system with severe stroke reached statistical significance. The † indicates significantly lower variance in the WAAC group (Bartlett’s test or Levene’s test, P values were adjusted for multiple tests). All data are presented as mean ± SD.

In the mild stroke group, mean Tcore over the entire intra-ischemic period (15 to 60 min) was 36.4 ± 0.5 versus 35.9 ± 0.7 °C in WAAC system and heating pad groups, respectively. Mean differences were not significant (t test, P=0.119), nor was variance (Fisher’s F test, P=0.296). During the post-ischemic period, repeated measures analysis did not show differences by heating system at any time point. Furthermore, the variation at each time point was similar between groups. The mean Tcore during the reperfusion period (15 to 165 min) was 35.6 ± 0.4 versus 35.2 ± 0.4 °C, respectively (t test, P=0.131). As above, only mice surviving through three days of stroke were included in the analysis. Mortality was low in the mild stroke groups (0 vs. 1, respectively) (see Table 1).

The WAAC system does not alter variability in neurologic deficit score following stroke.

Mice were evaluated by neurologic deficit score (NDS) at the end of the occlusion period (60 min) and at day 1, 2 and 3 after MCAo. All mice demonstrated an NDS of 12 (full deficit) at the point just prior to removal of the occluder (Fig 4). Mice of all groups showed progressive functional improvement (lower NDS) through the three days following stroke. As expected, the severe stroke group showed a correspondingly worse NDS compared with the mild stroke group. There was no overall group effect (across all days) by heat support system. Only mice that survived through three days of experimental period were included in the analysis.

Figure 4:

Comparison of neurological deficit score (NDS) by heat support system through three days after MCAo. Data are presented as mean ± SD; the # reflects significant intra-group difference from 24 hours post-stroke (GEE method).

Effect of WAAC system on infarct volume variability.

Brains were harvested at three days after MCAo and analyzed for infarct volume as cortical infarct, subcortical infarct, and total infarct (with and without correction for swelling) (Fig 5). In the severe stroke group, no differences were found in mean infarct volumes between heat support systems. Infarct volumes in cortical, subcortical, total and corrected infarct volume (in mm³ for WAAC and heating pad, respectively) were: 63.0 ± 20.2 vs. 54.9 ± 39.0 (cortex), 29.0 ± 5.8 and 30.1 ± 11.5 (subcortex), 92.0 ± 24.4. vs. 58.0 ± 50.0 (total), and 88.7 ± 18.2 vs. 80.1 ± 43.3 (corrected total). The variability of the infarct volume data (SD values) was consistently lower in the WAAC group (1.9 to 2.4 fold reduced), but did not reach statistical significance in these limited group sizes (Levene’s test; P=0.1880, 0.0932, 0.1410, and 0.1030, respectively). In the mild stroke group, neither the infarct volumes nor the infarct volume variability differed by heat support system. Infarct volumes in cortical, subcortical, total and corrected infarct volume (in mm³ for WAAC and heating pad, respectively) were: 43.8 ± 28.1 vs. 36.8 ± 29.5 (cortex), 19.6 ± 7.7 vs. 17.8 ± 8.5 (subcortex), 62.0 ± 34.3 vs. 53.1 ± 35.5 (total), and 59.5 ± 32.0 vs. 52.4 ± 34.7 (corrected total). In the severe injury group, regression analysis of Tcore and resulting infarct volume demonstrated that mean Tcore during the intra-ischemic period positively correlates with final infarct volume in the pooled data with both heating systems (P=0.0084, difference from zero slope) or with the heating pad group data alone (P=0.031, difference from zero slope) (Fig 6A). The mild injury group, which demonstrated less range in Tcore temperatures, showed no correlation with mean Tcore during the intra-ischemic period and final infarct volume (Fig 6B). Regression analysis with post-ischemic mean Tcore showed no significant correlation with final infarct volume in severe or mild injury groups. To evaluate the effect of tight temperature regulation during the intra-ischemic and post-ischemic periods (independent of temperature control system), we grouped infarct volume according to whether mean Tcore was within 35.7-37.3 °C during the intra-ischemic period alone (intra only), intra-ischemic and post-ischemic period (intra and post), or outside of the range for both (outside range). Fig 6C shows the mean ± SD for severe and mild stroke groups. While the mean values did not statistically differ by temperature range group, the SD was significantly reduced in the “intra and post” subgroup in the severe stroke group (P=0.0373, Bartlett’s test). These findings highlight the significant influence of intra-ischemic Tcore on infarct volume (Fig 6A) and the value of tight regulation of both intra-ischemic and post-ischemic Tcore to reduce infarct variability in severe stroke mice (Fig 6C).

Figure 5:

Comparison of infarct at day 3 after MCAo between heating systems. (A) Representative TTC stained brain sections for WAAC and heating pad systems for severe and mild stroke models. Severe stroke brains show infarct in posterior sections, reflecting injury to both the middle cerebral artery (MCA) and posterior cerebral artery (PCA) territories. Mild stroke brains show injury to the MCA territory only. (B) Summary of infarct volume between groups, plotted as all points, mean volume, and SD for cortical, subcortical, total, and total corrected for swelling infarct volume. No group differences in mean values were found; however, SD values were consistently lower in the WAAC heat support group with severe stroke.

Figure 6.

Relationship between mean Tcore during the intra-ischemic and post-ischemic periods and final infarct volume. (A) Correlation between Tcore and total infarct volume in severe injury model. Linear regression is shown from all data (black line) or heating pad system only (red line). The P values reflect difference from zero slope. (B) Correlation between Tcore and infarct volume in mild injury model. (C) Summary of mean Tcore (independent of heating system) on variability of total infarct volume. Data are presented as mean ± SD grouped according to whether the mouse Tcore was within 35.7 and 37.3 °C during both intra-ischemic and post-ischemic periods (intra and post), during intra-ischemic period only (intra only), or neither (outside range). † indicates difference in SD compared with other groups (P=0.0373, Bartlett’s test).

Effect of WAAC system on hemorrhagic transformation score.

Hemorrhagic transformation (HT) was evaluated at day 3 after MCAo by gross inspection of high-resolution images of the TTC-stained sections. Representative sections showing the presence of hemorrhage (yellow arrows) in severe and mild stroke brains are shown in Fig 7A. HT was scored (0-5 scale) by blinded investigator for each section and then summed to obtain a total score for each brain. There was no overall group effect of heat support system in either severe or mild stroke models (Fig 7B). However, regression analysis of Tcore with resulting HT score demonstrated a significant correlation between the intra-ischemic Tcore and the resulting HT score in the severe injury group (P=0.033, difference from zero slope) (Fig 8A). No significant correlation was evident between Tcore during the post-ischemic period and HT score. Additionally, no correlation was found between Tcore (either period) and HT score in the mild injury group (Fig 8B). Analysis of HT score variability by temperature grouping, as described above, showed no difference in mean HT score or significant reduction in SD values (Fig 8C). These findings demonstrate a modest (R squared = 0.3696), but statistically significant relationship between intra-ischemic temperature and resulting HT in the severe injury group.

Figure 7:

Comparison of hemorrhagic transformation (HT) at day 3 after MCAo between heating systems. (A) Representative TTC stained brains showing evidence of hemorrhage in the infarct region (yellow arrows). Enlarged regions corresponding to the yellow box are provided to better visualize regions of hemorrhage. Scale bar is 100 μm. (B) Summary of total HT score for severe and mild stroke. Data is presented as mean ± SD. Mean HT score did not differ by heating system for severe or mild stroke groups.

Figure 8.

Relationship between mean Tcore during the intra-ischemic and post-ischemic periods and hemorrhagic transformation (HT) score at day 3 after MCAo. (A) Correlation between Tcore and HT score in the severe stroke group. Linear regression is shown from all data (black line). P value reflects difference from zero slope. (B) Correlation between Tcore and HT in mild stroke group. (C) Summary of HT mean and variability grouped by Tcore variability of HT score (i.e. independent of heating system). Data are presented as mean ± SD grouped according to whether the mouse Tcore was within 35.7 and 37.3 °C during both intra-ischemic and post-ischemic periods (intra and post), during intra-ischemic period only (intra only), or neither (outside range). † indicates difference in SD compared with other groups (P=0.0373, Bartlett’s test). Comparison of SD values in the severe stroke group showed a non-significant trend for difference in SDs (P = 0.067, Bartlett’s test).

Discussion:

The primary findings of this study are 1) intra-ischemic body temperature positively correlates with infarct volume and HT score, 2) tighter control of intra-ischemic and post-ischemic core temperature reduces variability in infarct volume in more severely injured mice, and 3) the warmed ambient air cage (WAAC) system provides significantly more precise control of intra-ischemic and post-ischemic mouse body temperature compared with traditional heating pad warming system. These findings re-affirm the critical influence of body temperature on stroke outcomes and thus support the experimental value of tightly controlling body temperature during the intra-ischemic and post-ischemic periods. These studies additionally describe the fabrication of a low-cost and highly effective cage warming system that can be assembled and applied in any stroke laboratory.

Our findings showed that the WAAC system was significantly more stable than the heating pad system, reaching steady-state cage temperature more quickly and then maintaining cage temperature very precisely. Although our study did not require frequent opening of the lid, we found that the WAAC system was able to quickly recover temperature after lid opening within (2-3) minutes (Supplemental Fig 6). The WAAC system is easy to assemble and the components are relatively inexpensive. With the cost of the heater/circulator unit at <$60 USD, a laboratory can assemble multiple recovery cages to accommodate high-throughput studies. Although not shown, the heater system also provided good temperature control for a rat cage.

We compared our WAAC system to a widely used form of post-surgical warming (cage on heating pad) to evaluate its effectiveness on animal temperature stability as well as variability in stroke outcomes. We used a common model of stroke (MCAo) with two versions of commercially available occluder devices. The occluder devices consisted of a 6-0 monofilament with the occluding tip coated with silicone to a diameter of 0.21 mm. In the C57BL/6 strain of mice, the coating length affects whether the occluder blocks flow to the middle cerebral artery (MCA) alone (i.e. 2-3 mm occluder) or to the MCA and the posterior cerebral artery (PCA) (i.e. 5-6 mm occluder) (see Supplemental Fig 5). As a result, these occluders produce either a more mild stroke (2-3 mm, MCA territory only) or a more severe stroke (5-6 mm, MCA and PCA territory).

Mice and other mammals maintain body temperature through a balance of metabolic heat production and temperature exchange with the environment. During times of reduced metabolic activity (e.g. anesthesia, stroke, etc.), this balance is shifted and body temperature can fall. This drop in temperature can be both rapid and extreme in mice due to their lower mass to body surface area relationship(Gordon, 2012). In relation to body warming, two important temperature ranges are the thermoneutral zone (TNZ) and the ambient temperature range for normothermia. The TNZ reflects the ambient temperature range over which mice are metabolically neutral (i.e. increased metabolic activity is not needed to maintain core temperature). In mice, the TNZ varies according to conditions, but is typically 28-32 °C(Gordon, 2012). The ambient temperature range for normothermia reflects the point at which further increased temperature results in an increase in core temperature(Gordon, 2012). The ambient temperature of the WAAC system was set to 33 °C, which falls just above the thermoneutral zone (TNZ) in mice of ~28-32 °C and within the ambient temperature range for normothermia (32 – 34 °C). By providing an elevated ambient temperature, our WAAC system compensates for the loss in metabolic heat production by the post-surgery/post-stroke mice, and thus allows the mice to maintain a more consistent and normothermic body temperature. Furthermore, by setting the ambient temperature (33 °C) only slightly above the TNZ, mouse temperature stays within the normothermia range – and importantly, below the temperature that results in hyperthermia.

Our choice to measure body temperature (instead of direct brain temperature) was based on practical issues, and thus on what could be easily applied in stroke studies in any laboratory. While measurement and control of brain temperature would be ideal, its direct measurement requires insertion of a temperature probe into the brain. This procedure adds cost, an additional surgical procedure, and produces brain tissue damage. Given that intra-ischemic and post-ischemic body temperature is not currently widely reported in experimental stroke studies, the expectation that the field would adopt routine brain temperature measurements is low. While not a perfect reflection of absolute brain temperature, core temperature has been shown to track reasonably well with brain temperature in non-stimulated conditions(DeBow and Colbourne, 2003; Jackson-Friedman et al., 1997; Kiyatkin and Wise, 2002). In mice, the small subcutaneous transponders show good correlation with intraperitoneal temperature(Hankenson et al., 2018; Kort et al., 1998) (Supplemental Fig 2), and thus offer a simple and inexpensive option for wirelessly monitoring core temperature. The transponders used in the present study cost approximately $10 USD per unit and can be sterilized and reused several times. The wand reader is the most expensive item, costing around $2500 USD. However, since brain temperature is one of the most influential variables in brain injury, tightening this variable through more precise heat support should enable stroke studies to be performed with fewer mice while maintaining equal or better statistical power. As an example, using infarct volume as the measured variable in a power analysis calculation, the WAAC system would reduce the number of mice needed by >4 fold compared with the heating pad system (based on a 50% change in means).

Our findings showed that the primary benefit of the WAAC system was in the mice which underwent the severe stroke protocol. By comparison, the traditional heating pad system mice in the severe stroke group demonstrated larger deviation of body temperature from normothermia compared with the WAAC mice. In addition, the heating pad mice showed greater variation in temperature, evaluated in both the intra-ischemic and post-ischemic periods. The mild stroke mice showed no statistically significant effect by heat support method. The reason for the difference between stroke injuries is presumably due to a greater reduction in activity-dependent heat production in the more severely injured animals. Note that the hypothalamus, a key brain region in thermoregulation(Morrison and Nakamura, 2011; Tan and Knight, 2018), was not injured in either stroke protocol (see Fig 5). From these findings, the WAAC system should also provide improved temperature control in other stroke models where stroke injury is more severe, such as in aged mice or mice with exacerbating co-morbidities.

Limitations: As noted above, our studies utilized core temperature as a surrogate of brain temperature. While brain temperature often parallels body temperature in conscious rodents, absolute brain temperature can range from ~1 °C above or below core temperature(DeBow and Colbourne, 2003; Kiyatkin et al., 2002) and can further deviate during specific physiological challenges (e.g. tail pinch, social stress) and pharmacological challenges (Kiyatkin and Wise, 2002). The brain itself is a significant source of metabolic heat, and thus brain temperature can further dissociate from core temperature at times when increased brain activity is not adequately accounted for by increased heat removal via the circulatory system. Measurement of temporalis muscle temperature is sometimes used as an approximation of brain temperature in rodents, particularly in rats(Jackson-Friedman et al., 1997; Kiyatkin, 2010). The temporalis muscle is physically associated with the head and is also supplied by the carotid artery. Thus, the temporalis muscle temperature is considered to be a reflection of both the temperature of arterial blood supplying the brain as well as the underlying brain, which is separated by the skull. Unfortunately, the MCAo stroke model involves permanent ligation of the external carotid artery, which is the branch of the carotid that supplies much of the scalp and the temporalis muscle(Ku and Choi, 2012; Vaas et al., 2017). In addition, the temporalis muscle in mice is of significantly less mass than in rats(Liu et al., 2009). These factors make the temporalis muscle less ideal (or at least far more complicated) as a proxy of brain temperature in mice. An additional limitation of our study is that rectal temperature during the surgical periods of occluder placement and removal were not recorded (logged) for subsequent analysis. While rectal temperature was set to 37 °C on the feedback controller and the actual temperature periodically affirmed by the surgeon, we cannot rule out slight temperature variance between mice during the surgical periods. As a solution to this issue, we suggest using a temperature logging control unit or capturing temperature data by a 3^rd party logging system. As a final limitation of our study, we did not use laser Doppler (LD) or laser speckle contrast imaging (LSCI) to measure restoration of perfusion following occluder removal. We chose to omit additional measures that extended anesthesia time or resulted in increased skull exposure and cooling. In our experience with several hundred mouse MCAo surgeries in which we did measure LD or LSCI, we have always obtained reperfusion – except in the rare case where the filament caused a puncture in the circle of Willis. In such a situation, the puncture is often evident at the time of filament removal or by the much more severe injury in the mouse. These mice frequently die before 3 days post-stroke and the evidence of puncture is typically evident at the time of brain removal. Despite the abovementioned limitations, our approach based on providing tighter core temperature control appears to have produced the desired effect of reducing variability in stroke outcome.

In summary, we describe a low-cost cage warming system (WAAC) that reflects a significant improvement over the widely used “cage on a heating pad” approach. The WAAC T_air and T_bedding can be more precisely controlled, which leads to tighter body temperature control and reduced variability in stroke outcome. In addition, the WAAC system prevents accidental hyperthermia, as can occur with heating pads without feedback control systems. These refinements in intra-ischemic and post-ischemic temperature management are consistent with the goals of the Stroke Therapy Academic Industry Roundtable (STAIR), the Ischaemia Models: Procedural Refinements Of In Vivo Experiments (IMPROVE), and other recommendations for improving translation of stroke findings(Dirnagl, 2006; Kahle and Bix, 2012; Percie du Sert et al., 2017; Roundtable, 1999). The described approach of wireless temperature monitoring and precise temperature support could easily be applied in stroke research laboratories.

Highlights:

• Description for fabricating a low-cost mouse cage warming system

• Improved body temperature control compared with conventional warming methods

• Tight body temperature control during experimental stroke reduces data variability