Exogenous hydrogen sulfide gas does not induce hypothermia in normoxic mice

Hydrogen sulfide (H2S, 80 ppm) gas in an atmosphere of 17.5% oxygen reportedly induces suspended animation in mice; a state analogous to hibernation that entails hypothermia and hypometabolism. However, exogenous H2S in combination with 17.5% oxygen is able to induce hypoxia, which in itself is a trigger of hypometabolism/hypothermia. Using non-invasive thermographic imaging, we demonstrated that mice exposed to hypoxia (5% oxygen) reduce their body temperature to ambient temperature. In contrast, animals exposed to 80 ppm H2S under normoxic conditions did not exhibit a reduction in body temperature compared to normoxic controls. In conclusion, mice induce hypothermia in response to hypoxia but not H2S gas, which contradicts the reported findings and putative contentions.

anapyrexic agents. Of these, exogenous H 2 S has received the most attention in the last few years in response to the Science publication by Blackstone et al. 20 . However, our experiments in mice, which duplicated the experiments by Blackstone et al. 20 , revealed that exogenous H 2 S does not induce hypothermia at normoxic conditions. Instead, the hypothermia observed in the experiments emanates from a hypoxia-induced anapyrexic response, which is a natural response in mice to hypoxic stress 2,3 . The results are described in this paper and addressed in the context of artificial hypometabolism.
Exogenous H 2 S has been proposed to induce hypometabolism that is associated with a state of suspended animation 20 . Mice that were subjected to a gas mixture composed of 17.5% O 2 , 80% nitrogen (N 2 ), and 80 ppm H 2 S exhibited a 22 °C reduction in T b (Fig. 1A) after 4 h of exposure, yielding a T b that was slightly above the T a of 13 °C. At this point CO 2 production and O 2 consumption had decreased by approximately 90%, suggesting that the animals had reached a state of hypometabolism by anapyrexia. Moreover, this state was reversible inasmuch as all metabolic parameters reverted to baseline within 4 h after the exposure to H 2 S was abrogated. During this recovery period the T b also gradually restored to baseline at a T a of 24 °C. In another study by Volpato et al., inhalation of air containing 17.5% O 2 and 80 ppm H 2 S induced similar anapyrexic effects in mice at a T a of 27 °C as well as 35 °C (Fig. 1B) 21 , altogether suggesting that inhaled H 2 S reduces the T b to T a levels.
The mechanism behind exogenous H 2 S-induced suspended animation 20,21 is generally ascribed to the direct inhibition of oxidative phosphorylation 22 and consequent histotoxic hypoxia. Because of is its high membrane permeability, H 2 S is readily delivered to tissues via the circulation where it transgresses cell membranes and localizes to various intracellular organelles, including mitochondria 23 . H 2 S binds cytochrome c oxidase (complex IV) in the electron transport chain in a reversible and noncompetitive fashion. As a result, H 2 S prevents O 2 binding to cytochrome c oxidase and thereby interferes with the reduction of O 2 to water. Concurrently, H 2 S interferes with the production of adenosine triphosphate (ATP) by ATPase due to H 2 S-induced perturbation of electron transfer and proton gradient over the mitochondrial inner membrane 24,25 . It should be noted, however, that H 2 S-mediated histotoxic hypoxia has never been proven to directly translate to H 2 S-induced hypothermia. Similarly, experimental evidence that H 2 S triggers a downward shift of the thermoneutral zone directly remains at large.
Although the hypometabolic effects of exogenous H 2 S seem convincing, the putative mechanism for the hypometabolic state induced by exogenous H 2 S, i.e., cytochrome c oxidase inhibition 22 , may not account for the observed effects. As H 2 S is a toxic, irritant gas 22 , inhalation is known to provoke epithelial damage in the upper 26 and lower respiratory tract 27,28 in rats and pulmonary edema in pigs 29,30 . The pulmonotoxicity of exogenous H 2 S may therefore be associated with hypoxemic hypoxia.
Hypoxia, on the other hand, is a very potent inducer of anapyrexia, hypothermia, and hypometabolism and, thereby, of suspended animation 3 . Several hibernating and non-hibernating mammalian species, including mice, exposed to different degrees of hypoxic atmospheres (i.e., F i O 2 5-10%) immediately drop their T b to enter a reversible state of hypometabolism [31][32][33][34] . The hypothermic effects of hypoxia are known to be caused by downward adjustment of the 'internal thermostat, ' and involve the preoptic anterior hypothalamus (POAH), as has been demonstrated in thermobehavioral experiments in rodents 2 . Consequently, we proposed that the hypothermia in exogenous H 2 S-exposed mice, which constitutes a hallmark feature of hypometabolism, emanated from the combination of mild hypoxia (17.5% O 2 ) and inhalation of H 2 S gas, and not the exogenous H 2 S gas per se.

Results
To test the hypothesis that exogenous H 2 S-induced hypothermia emanates from hypoxia and not H 2 S, we performed experiments in 48 female C57BL/6 mice using a similar approach as was employed by Blackstone et al. 20 . The experiments, which are outlined in Fig. 2   . The effects of exogenous H 2 S and a hypoxic atmosphere on T b at a T a of ~21 °C were measured non-invasively with a thermographic camera and the locomotor activity of the animals was quantitated with dedicated motion analysis software.
As shown in Fig. 3 and Supplemental Video S1, hypothermia and reduction in locomotor activity only occurred in mice subjected to hypoxic conditions. 5% O 2 -exposed animals immediately dropped their T b to approximately 2 °C above the T a ( Fig. 3B, P < 0.0001) and reduced their locomotor activity to nearly nil compared to the H 2 S in F i O 2 21% and normoxia groups ( Fig. 3C, P < 0.0001) during the entire exposure period. The exogenous H 2 S in 21% F i O 2 group did not differ from the normoxia group during 6 h of 80 ppm H 2 S gas exposure in neither superficial temperature nor locomotor activity. At 3 h of exposure, however, animals in the H 2 S in 17% F i O 2 group started to drop their T b to approximately 4 °C above T a , in contrast to F i O 2 17%-exposed control animals ( Fig. 3B, P < 0.0001). Alleviation of the hypoxic conditions during the restoration phase resulted in complete reversal of the superficial temperature to baseline levels within 1 h in the F i O 2 5% group, which is in agreement with previous reports 20,21 ( Fig. 1). During the 3 h of restoration at normoxic atmosphere, the H 2 S in F i O 2 17% -exposed animals remained hypothermic and only restored T b to the level of the F i O 2 17% and 21% control groups at 9 h ( Fig. 3B, P < 0.01). Mice in the H 2 S groups exhibited some discomfort during H 2 S exposure, as evidenced by the cringed posture, which occasionally concurred with vigorous locomotion (Supplemental Video S1).
Peripheral vasodilation is one of the cooling mechanisms that is autonomically regulated in response to a mismatch between the T b and the internal thermostat (i.e., T b > thermoneutral zone) 3,35,36 . Peripheral vasodilation is integral to anapyrexia 3 , which enables cooling. The cooling process is in turn facilitated by the blockade of thermogenic effectors and the enabling of peripheral vasodilation [36][37][38] . Therefore, the extent of peripheral vasodilation was determined by measuring the change in tail temperature at baseline and at approximately 4 min after initiation of H 2 S-or hypoxia exposure.
The tail of 5% O 2 -exposed animals warmed up right after the start of exposure (+2.1 ± 0.5 °C, N = 3, P < 0.05 versus the H 2 S group, unpaired student's t-test), while the tail of H 2 S in 21% O 2 -exposed animals (−1.3 ± 1.1 °C, N = 3) and normoxia-exposed animals (+0.1 °C, N = 1) did not exhibit changes in temperature (P > 0.05, unpaired student's t-test) (Fig. 4). These results provide compelling evidence for the induction of peripheral vasodilation by hypoxia but not exogenous H 2 S, and hence for hypoxia-mediated anapyrexic signaling. The absence of a vasodilatory response in the exogenous H 2 S group is in agreement with the surface temperature data, which encompassed an absence of hypothermia (Fig. 3).

Discussion
Based on the experimental evidence, namely T b , tail temperature, and locomotion, it can be concluded that inhalation of H 2 S gas at 80 ppm in a native atmosphere of 21% O 2 and 79% N 2 does not induce hypothermia in mice, which contradicts what has been reported previously 20,21 . Hypoxia, on the other hand, is a very potent inducer of hypothermia that, given the peripheral vasodilation observed in the tail vasculature, may comprise part of an anapyrexic response 2,3 . The subclinical thermal effects of mild hypoxia, however, are potentiated by combined 80 ppm H 2 S gas exposure.
One consistent finding in mouse studies on the pharmacological induction of hypothermia is that the animal's T b or surface temperature approximates the T a and subsequently enters a plateau phase that is sustained in the  20 and Volpato et al. 21 , the T b was in all instances downmodulated to a depth at which the T b was more or less in equilibrium with the T a , irrespective of the magnitude of the T a (i.e., 13 °C, 27 °C, or 35 °C). The same pattern was observed in our experiments (T a = 21 °C), suggesting that the hypothermia may have been mediated via a common mechanism. Moreover, this decline-plateau pattern suggests that the cooling process is passive once the thermogenic effectors have been shut off. The cooling is halted upon reaching a thermodynamic equilibrium where T b = T a , i.e., a point at which the organism is not equipped to cool further. Unlike under normophysiological circumstances, where T b is tightly regulated via engagement of cooling effectors or thermogenic effectors 35,39 , the hypothermic state seems to sustain itself through passive heat transfer only.
The main differences between the results of Blackstone et al. 20 , Volpato et al. 21 , and our results are the rate of cooling and subsequently the time required to reach the plateau phase (T b = T a ). The cooling rate was approxi-  20 , ~4 h in the study of Volpato et al. 21 , and 2 h in our study (Fig. 3). The same animal species with similar animal weights were employed in all studies. Hence, it is unlikely that these discrepancies arose from differences related to physical laws such as Galilei's square-cube law 40 , the implication of which is that animals with a large body surface:mass ratio (i.e., small animals) cool faster than animals with a small body surface:mass ratio (i.e., large animals) 1 . The discrepancies in cooling rate also did not emanate from differences in metabolism in accordance with Kleiber's law, which states that small animals exhibit a relatively higher metabolic rate to maintain euthermia compared to larger animals 1,41 .
In light of the finding that exogenous H 2 S is not an inducer of hypothermia, the question that remains to be answered is "why did Blackstone et al. and Volpato et al. observe hypothermia in H 2 S-exposed mice?" Volpato et al. was able to reproduce the hypothermic effects of 80 ppm H 2 S of Blackstone et al. Consequently, we do not question the methodology and validity of their results. In our opinion, the answer lies in the hypoxic conditions that were induced by the combination of subatmospheric F i O 2 and the various mild forms of exogenous H 2 S-induced hypoxia. The 3.5% lower F i O 2 versus native atmospheric F i O 2 (17.5% versus 21%, respectively) is, in itself, not sufficient to trigger anapyrexia in mice, unless such mild hypoxic conditions are exacerbated by exogenous H 2 S. In line with our results obtained in the 17% F i O 2 groups, the exacerbation likely occurred in the experiments by Blackstone et al. and Volpato et al. for four possible reasons. First, as explained in the Introduction section, H 2 S can induce histotoxic hypoxia by inhibiting cytochrome c oxidase and corollary ATP production, resulting in reduced metabolic supply (energy). Consequently, the organism is forced to adapt its metabolic demand to survive by means of e.g., hypothermia (Arrhenius' law). Secondly, H 2 S can limit the binding of O 2 to hemoglobin's O 2 binding sites 42 , thereby causing O 2 affinity hypoxia 14 . Thirdly, H 2 S reduces cardiac output through its deregulatory and negative chronotropic effects on cardiac rhythm 21,28 , which leads to circulatory hypoxia 43 . Fourthly, H 2 S is pulmonotoxic [26][27][28] and may impair pulmonary O 2 /CO 2 exchange and the extent of O 2 saturation, which in turn may aggravate the circulatory hypoxia caused by the cardiovascular effects. In addition, based on ex vivo experiments, H 2 S seems to play an essential role in hypoxic pulmonary vasoconstriction 44 . Therefore, administration of exogenous H 2 S to the lungs may further compromise pulmonary blood flow during hypoxic conditions, which can augment hypoxemic hypoxia. Accordingly, all these forms of H 2 S-mediated hypoxia may add to the mild hypoxia caused by subatmospheric F i O 2 levels and culminate in a hypoxic state that is considerable enough to trigger anapyrexia. As addressed in Dirkes et al. 45 , circulatory hypoxia is sensed through carotid bodies located in the carotid artery 46,47 that, under non-hypometabolism-inducing, hypoxic conditions, relay arterial O 2 tension (P a O 2 )-related information to the brain. The brain subsequently (hyper)activates certain physiological functions to remediate the hypoxia 48 , which include panting [49][50][51][52][53] and tachycardia 53,54 . How this is blocked during the induction of anapyrexia is currently unclear. Endogenous H 2 S as well as intracerebrally administered exogenous H 2 S analogues inhibit the ventilatory and thermal response to hypoxia in the hypothalamus and brain stem. Contrastingly, microinjection of Na 2 S (H 2 S precursor) in the anteroventral preoptic hypothalamus of rats potentiates hypothermic signaling by hypoxia, but does not alter T b under normoxic conditions 55 . Microinjection of the endogenous H 2 S production inhibitor amino-oxyacetate in the sympathetic excitatory rostral ventrolateral medulla of rats attenuates hypoxia-induced hypothermia 56 . As H 2 S passes the blood-brain barrier freely, central effects of inhaled H 2 S could have contributed to hypoxia-induced anapyrexia via the hypothalamus or brain stem, 22 albeit an unequivocal mechanistic explanation remains warranted in light of the contrasting results.
In the experiments of Blackstone et al. and Volpato et al., T b was determined by telemetry devices that record the core temperature (i.e., intra-abdominal temperature). In our experiments, the superficial temperature was determined. We believe that this approach is valid for the purpose of this study inasmuch as we were interested in temperature trends as a function of exposure time and gas composition, and not the real T b per se. Since all groups were thermographically analyzed in the same manner, the resulting data yield credence to our conclusions. Moreover, the use of thermographic imaging has some benefits over intra-abdominal temperature determination, such as the determination of thermoregulatory vasoactivity by tail temperature measurement (Fig. 4).
Although this paper focused on the hypometabolic properties of H 2 S gas, several animal studies on the effects of liquid H 2 S analogues NaHS and Na 2 S have been published. After inhalation, H 2 S gas diffuses freely across the alveolar membrane and enters the blood as predominantly HS − and H 2 S 22 . Accordingly, intravenous administration of solubilized H 2 S precursors/analogues is believed to follow the same pharmacodynamics as administration through inhalation, only without the detrimental effects on local pulmonary physiology and toxicity. The hypothermic effects of NaHS and Na 2 S in small as well as in large animals have been reviewed before 57 . Continuous administration of NaHS is assumed to induce hypothermia in anesthetized rats, although these studies lack essential control groups 58,59 . The evidence considering the hypothermic and hypometabolic effects of NaHS in large animals has been conflicting: in a pig study a small hypothermic effect was observed following 8 continuous hours of NaHS administration 29 , whereas in several other studies in pigs 45,60 and sheep 61 such hypothermic effects were not reproducible. The differences between the effects of H 2 S in small and large animals have been contemplated by Dirkes et al. and are explained by the inability of large animals to lose heat sufficiently due to the low body surface:mass ratio 45 .
In this paper, the tail temperature was used as a measure of central activation of peripheral cooling mechanisms (i.e., peripheral vasodilation), as has been used before in the determination of thermoregulatory peripheral vasoactivity in pyrexic mice 38 . However, as reviewed by Liu et al., H 2 S has biphasic effects on the vascular tone: at low concentrations H 2 S induces vasoconstriction and at higher doses vasodilation is induced, as evidenced in mouse and rat aortic tissue [62][63][64] . Consequently, the absence of thermoregulatory vasodilation and a consequent increase in the tail temperature of 3 animals (Fig. 4) could also be a direct vasoconstrictive effect of low-dose H 2 S. Nevertheless, H 2 S-induced vasoconstriction is unlikely to be responsible for the absence of H 2 S-induced hypothermia in our experiments. A 'masked' thermoregulatory vasodilative response would be accompanied by deactivation of brown adipose tissue (BAT) and shivering thermogenesis (i.e., major source of heat in mice at a T a of 21 °C) 1,39 . Subsequently, the cessation of thermogenesis would be reflected in the T b /superficial temperature of H 2 S-exposed animals, which was not observed (Fig. 3).
In conclusion, exogenous H 2 S is not a hypometabolism-inducing agent. The hypometabolism induced in mice that were subjected to exogenous H 2 S was caused by hypoxia. At subatmospheric F i O 2 levels, exogenous H 2 S exacerbates the hypoxic conditions to such a degree that anapyrexia and hypothermia are triggered. Accordingly, exogenous H 2 S is a hypometabolic adjuvant rather than a hypometabolism-inducing agent.
An experimental setup was custom-built to allow controlled gas exposure while unobtrusively assessing body temperature (T b ) with a thermographic camera (ThermaCAM SC2000, FLIR Systems, Wilsonville, OR) in non-anesthetized mice. The setup consisted of gas-tight polypropylene chambers (Fig. 5A, length × depth × height of 109 mm × 109 mm × 61 mm) that were sealed at the imaging end with a thin, infrared light-permeable polyethylene sheet to permit thermal imaging from outside (Fig. 5B). Metal wires were secured longitudinally so that the animals could not reach the polyethylene sheet (Fig. 5C). Gas inflow and outflow tubes were connected to each box at the posterior end for modulation of experimental conditions (Fig. 5E). The gas permeability of the chambers was tested by air pressure decline experiments. Also, a thermistor (Fluke 51 II, Fluke Corporation, Everett, WA) was secured in the posterior wall (Fig. 5D) to facilitate the measurement of the temperature in the chamber. The thermistor was used as a calibrator for the thermographic camera images, as the thermographic images display the temperature of the copper bolt retaining the thermistor. To ascertain sufficient inflow of gas in all experiments and prevent CO 2 accumulation, the flow rates were controlled on the basis of CO 2 outflow concentrations (<600 ppm, CO 2 Meter, Ormond Beach, FL). The system was also connected to an O 2 and H 2 S meter (model OdaLog 7000, App-Tek International, Brendale, Australia), which was calibrated by a certified company prior to the experiments (Carltech, Maarheeze, the Netherlands). The O 2 and H 2 S meter was post hoc tested for measurement accuracy. The experiments were performed at a mean ± SD T a of 21.2 ± 0.6 °C.
Experimental procedure. To test the hypothesis that H 2 S-induced hypothermia emanates from hypoxia and not H 2 S, all 48 animals were randomly divided among 5 experimental groups. Group A was exposed to 80 ppm H 2 S in 21% O 2 and 79% N 2 (H 2 S in 21% O 2 group, N = 12), group B was exposed to 80 ppm H 2 S in 17% O 2 and 83% N 2 (H 2 S in 17% O 2 group, N = 6), group C was exposed to 5% O 2 and 95% N 2 (5% O 2 group, N = 12), group D was exposed to 17% O 2 and 83% N 2 (17% O 2 group, N = 6), and group E was exposed to 21% O 2 and 79% N 2 (normoxia group, N = 12).
Mice were placed in the chambers individually. After 1 h of exposure to normoxia (21% O 2 and 79% N 2 ), the mice were exposed to one of the gas mixtures (A -E) for 6 h, after which 6 of the animals per group were allowed to recover at normoxic conditions for 3 h before being terminated. The other 6 animals of group A, C and E were terminated immediately after the 6 h of exposure for another study.
No anesthetics were used before or during the experimental procedure.
Thermal imaging and data processing. Animals were filmed every hour for 10 min with a thermographic camera (ThermaCAM SC2000, FLIR Systems, Wilsonville, OR) (Fig. 2). Thermographic camera images (3 images per second) were processed and analyzed in ThermaCAM Researcher 2001 (FLIR Systems). The mean maximum superficial temperature was calculated per time point per group. The tail temperatures of animals in group A (N = 3), C (N = 3), and E (N = 1) were obtained from the thermographic camera images at 0 h, just before the start of exposure, and approximately 4 min after the start of exposure. We noticed the intergroup differences in tail temperature during the experiments, as a result of which the tail temperature was measured in only 7 animals. The mean difference in tail temperature between both time points was calculated and compared for group A and C.
Locomotor activity was assessed per time point using the same thermal images as were used for the calculation of superficial temperature. An analytics program was written in LabVIEW (LabVIEW, National Instruments, Austin, TX). The thermographic camera images were converted to grayscale images and loaded into LabVIEW. Locomotor activity was calculated per animal per time point (−1 up to 9 h) on the basis of fluctuations in pixel intensity. A pixel was considered to reflect 'motion' when the grayscale intensity difference between direct temporally consecutive pixels exceeded 7 on a scale of 0 to 255. The intensity difference of at least 7 was based on the disappearance of background scatter present as intensity differences between 1 and 6. Values were expressed as the mean ± SEM amount of pixels with 'motion' per group per time point. Statistical analysis. Statistical analyses were performed using MatLab 2013a (MathWorks, Natick, MA).
Homogeneity of variance in each group was tested using the Bartlett's test. Based on equality of variances, either a one-way ANOVA or a Kruskal-Wallis test was performed, followed by a Tukey's range test or Dunn's test, respectively, to compare ordinal variables related to maximum superficial temperature and locomotor activity between groups. Tail temperature values were compared using an unpaired student's t-test. P-values less than 0.05 were considered significant. All values were presented as mean ± SEM, unless otherwise mentioned.