Sulfide metabolism and the mechanism of torpor

Hibernation is a seasonal physiological state of a number of mammalian species, where metabolic rate and body temperature are depressed for several days, which enables energy saving during winter (Heldmaier et al., 2004; Staples, 2014). During winter, small hibernators, such as the thirteen-lined ground squirrel (Ictidomys tridecemlineatus), reduce their energy requirement by fasting and spend most of their time in a state of deep metabolic depression named torpor, where metabolic rate decreases by ∼95% and body temperature drops down to ∼5°C (Heldmaier et al., 2004; Muleme et al., 2006; Staples, 2016). These torpor bouts are periodically interrupted by short arousals of few hours (∼12–24 h), named interbout euthermia (IBE), where metabolic rate and body temperature are restored (Fig. 1A). Compared with smaller hibernators, bears have already low mass-specific basal metabolic rates and exhibit a less marked suppression of basal metabolism during torpor (Heldmaier et al., 2004). Large hibernators such as bears maintain a fairly high body temperature (∼30°C) during torpor and do not cycle between states of torpor and IBE as regularly as squirrels, but still experience considerable decreases in metabolic rate of ∼75% (Tøien et al., 2011) (Fig. 1B). Thus, the metabolic depression of torpor is actively regulated, i.e. it does not follow body temperature passively (Heldmaier et al., 2004; Staples, 2014). For all hibernators, the changes in metabolic rate are paralleled by changes in ventilation rate and heart rate (MacCannell et al., 2018; Milsom and Jackson, 2011) to match oxygen supply with consumption, reaching a minimum level during torpor (Heldmaier et al., 2004). Thus, metabolism is essentially aerobic during hibernation and sustained by lipid catabolism of fat tissue (Hindle et al., 2014) without lactate buildup (Ma et al., 2005; Revsbech et al., 2013; Serkova et al., 2007). However, the molecular mechanisms underlying the metabolic rate suppression of torpor appear complex and are not yet fully understood (Giroud et al., 2021). In this Commentary, we highlight some recent advances indicating that hydrogen sulfide (H₂S), a widespread signaling molecule, plays a potentially key role in the suppression of metabolic rate during torpor.

The gasotransmitter H₂S is a signaling molecule continuously synthesized and degraded in vivo that can modify the function of target proteins and enzymes. It is also stored in tissues as protein-bound persulfide and polysulfide (Kolluru et al., 2013) and transported in blood (Bianco et al., 2018; Jensen and Fago, 2020). Bound and unbound H₂S are in equilibrium in vivo, with measured biological levels of free H₂S in the nanomolar to micromolar range, depending on the method used (Shen et al., 2015; Ditrói et al., 2019). In vivo levels of H₂S are also tightly controlled by several factors, including redox buffer capacity of protein cysteines reacting with H₂S (generating a pool of protein-bound sulfide, from which H₂S can be regenerated), local pH determining the ratio of H₂S/HS⁻ and the availability of oxygen for its oxidation and removal (Kolluru et al., 2013). At concentrations exceeding the low micromolar physiological levels, H₂S is a highly toxic molecule, and inhalation of >250–500 ppm ambient H₂S can be lethal for humans (Olson, 2011). This is due to the inhibitory effect of H₂S on cytochrome c oxidase, the O₂-consuming complex IV of the mitochondrial electron transport system (ETS) (Collman et al., 2009; Cooper and Brown, 2008; Nicholls et al., 2013). However, two studies found that inhalation of low levels of exogenous H₂S (80 ppm) induced a reversible torpor-like state in mice during hypoxia (Blackstone et al., 2005; Volpato et al., 2008), suggesting that H₂S suppressed metabolic rate and body temperature. These results had a strong impact but also raised some controversy, as inhaled H₂S had no effect on large mammals such as sheep and piglets (Haouzi et al., 2008; Li et al., 2008) and it was also argued that hypoxia itself could cause torpor and facilitate H₂S effects (Hemelrijk et al., 2018). In reality, many of these studies were performed under varying experimental conditions (summarized in Table 1), including different exposure protocols, ambient temperature, fasting and use of anesthesia, making it difficult to assess under which conditions externally supplied H₂S gas may work. A recent study (Marutani et al., 2021) demonstrates that prior exposures to 80 ppm H₂S gas (‘sulfide preconditioning’) eliminate the inhibitory effects on metabolic rate in mice, which might have been the case in some earlier data.

To understand the possible role of H₂S in torpor, other studies have focused on endogenous rather than exogenous H₂S. Supporting a role for H₂S in hibernation, significant in vivo changes in the levels and composition of H₂S metabolites have been detected in the plasma and liver of thirteen-lined ground squirrels (D'Alessandro et al., 2017; Jensen et al., 2021), in the plasma of brown bears (Revsbech et al., 2014) and in the lung of Syrian hamsters (Talaei et al., 2012) during winter hibernation (bears) or torpor (squirrels and hamsters). In addition, emerging evidence of increased enzymatic synthesis of H₂S (Talaei et al., 2012; D'Alessandro et al., 2017) and decreased degradation (Jensen et al., 2021) during torpor point towards a potential role of H₂S as a key physiological signaling molecule in controlling torpor in vivo. Here, we provide an overview of recent studies on how in vivo H₂S production and degradation may change during hibernation and suppress metabolic rate during torpor.

H₂S is ubiquitously produced from sulfur-containing substrates including cysteine, homocysteine and cystathionine by the cytosolic enzymes cystathionine β-synthase (CBS) and cystathionine γ-lyase (CSE) (Fig. 2). Moreover, the enzyme 3-mercaptopyruvate sulfurtransferase (MST), found in both the mitochondria and cytosol, can synthesize H₂S from 3-mercaptopyruvate in the presence of reducing co-substrates. Enzymatic H₂S production depends on several factors, such as enzyme expression and localization, substrate availability and the presence of various modulators (Giuffrè and Vicente, 2018; Kabil et al., 2011; Talaei et al., 2011). A metabolomics study on thirteen-lined ground squirrels found that substrates cysteine and cystathionine peak in the plasma during late torpor, suggesting increased H₂S production and complex IV inhibition (D'Alessandro et al., 2017). That study also proposed a model whereby CBS is activated by a nitric-oxide-dependent mechanism, suggesting the existence of a series of events in triggering and sustaining H₂S-based torpor. However, in studies by our group, we detected higher levels of plasma H₂S in summer compared with late torpor and IBE in thirteen-lined ground squirrels (Jensen et al., 2021) and in summer-active brown bears compared with winter-hibernating ones (Revsbech et al., 2014). These findings indicate that increased availability of plasma substrate for enzymatic H₂S synthesis may not necessarily reflect increased plasma H₂S levels. Plasma H₂S may derive from several sources, including extracellular enzymatic production by CSE and CBS secreted by endothelial cells and the liver (Bearden et al., 2010) and H₂S produced in the red blood cells (Revsbech et al., 2014; Vitvitsky et al., 2015) or carried by hemoglobin (Bianco et al., 2018; Jensen and Fago, 2018, 2020). Thus, an increased enzymatic production of H₂S in tissues during torpor may be difficult to detect in plasma. Also, circulating plasma cysteine in winter-hibernating brown bears is used not only for synthesis of H₂S but also for the synthesis of glutathione (Revsbech et al., 2014), a major antioxidant involved in the defense against reactive oxygen species (ROS).

In lung tissue from Syrian hamsters, H₂S synthesis has been found to increase during early and late torpor owing to increased expression of CBS, and both CBS expression and H₂S levels decreased to normal levels during IBE (Talaei et al., 2012). These changes are associated with the rapid lung remodeling taking place during torpor rather than with changes in lung metabolic rate (Talaei et al., 2012). Although this mechanism of H₂S regulation appears to be energetically costly, these results imply that CBS is partly degraded during the short period of IBE (typically <24 h) and then resynthesized during torpor in a cyclic manner, at least in the lung. In contrast, in thirteen-lined ground squirrels, a proteomic study reported lower CBS expression during torpor compared with IBE and summer in the liver, a major metabolic organ (Hindle et al., 2014). This finding aligns with the almost complete arrest of transcription and translation at the low body temperature of torpor in this tissue (Van Breukelen and Martin, 2001). Although regular changes in CBS expression may be a feasible strategy for regulating H₂S levels in the lung of hibernating Syrian hamsters, this does not seem to apply to other species or organs, particularly the liver, which undergoes significant metabolic depression during torpor (Muleme et al., 2006).

In addition to changes in the expression of H₂S-producing enzymes, changes in endogenous H₂S levels may also originate from existing bound sources of sulfur, including protein-bound persulfides and polysulfides (Kolluru et al., 2013). H₂S release from bound storage may be especially beneficial during hibernation, where free amino acids, such as cysteine, are a limited source and are largely preserved and directed away from the urea cycle to limit protein muscle breakdown (Rice et al., 2020). For instance, in the blood of winter-hibernating brown bears, levels of H₂S in red blood cells appear to derive in part from bound sulfane sulfur, allowing cysteine to be allocated to boost glutathione synthesis (Revsbech et al., 2014). The recently discovered cysteinyl–tRNA synthetase (CARS) pathway may contribute to the bound sulfur pool during protein translation. CARS can act as cysteine persulfide synthases that produce cysteine persulfides (CysSSH), which can be a source of protein persulfidation (R-SSH) (Akaike et al., 2017; Fukuto et al., 2018). Both CysSSH and protein persulfides can release free H₂S upon reduction, e.g. by thioredoxin (Dóka et al., 2016). Taken together, these data suggest that H₂S does not originate from a single source, but from a combination of enzymatic de novo synthesis, increased substrate availability and release from existing bound sulfur sources, depending on the species and the tissue. Specifically, the torpor-dependent upregulation of H₂S-producing enzymes remains to be investigated more broadly, to clarify in which hibernators and organs it may occur.

Though changes in endogenous H₂S levels and in the expression of the enzymes involved in its production provide evidence that H₂S metabolism is associated with hibernation, they do not reveal the mechanisms by which H₂S is capable of suppressing metabolic rate in torpor. To elucidate this, we will now focus on the mitochondrial targets of H₂S signaling during torpor, an aspect that has been investigated recently.

The suppression of whole-animal metabolism during torpor is reflected by suppression of mitochondrial function – a key feature for understanding mechanisms of metabolic depression (Staples, 2014). This is exemplified by liver mitochondria isolated from thirteen-lined ground squirrels, showing ∼70% lower respiration rates in torpor than in IBE (Muleme et al., 2006; Staples, 2014). This suppression of mitochondrial metabolism originates from phosphorylation-dependent changes in the activities of ETS complexes I and II (Mathers and Staples, 2019), with ETS and mitochondrial content remaining unchanged (Mathers and Staples, 2015; Mathers et al., 2017). However, other inhibitory mechanisms can contribute to depress mitochondrial respiration further, specifically a direct inhibition of cytochrome c oxidase by H₂S.

The inhibitory effect of H₂S on respiration has been widely demonstrated in purified cytochrome c oxidase, isolated mitochondria, various cell lines and animal models (Collman et al., 2009; Cooper and Brown, 2008; Leschelle et al., 2005; Módis et al., 2014; Sun et al., 2012; Wu et al., 2011). The inhibitory interaction of H₂S with cytochrome c oxidase is complex and concentration-dependent. Inhibition is reversible at low H₂S concentrations, with a binding constant of ∼0.2 µmol l⁻¹ (Cooper and Brown, 2008; Collman et al., 2009; Szabo et al., 2014), i.e. comparable to physiological levels of H₂S. However, the effect of H₂S on mitochondrial function is tightly regulated by the activity of the mitochondrial enzyme sulfide:quinone oxidoreductase (SQR), which is embedded in the inner membrane (Fig. 2). SQR is the first enzyme in the H₂S oxidation pathway that continuously oxidizes H₂S to thiosulfate, sulfite and sulfate through a series of enzymatic reactions, thereby regulating the in vivo levels of H₂S via its oxidative breakdown. Similar to cytochrome c oxidase, SQR binds H₂S tightly, with a reported binding constant of ∼0.1–2 µmol l⁻¹ (Lagoutte et al., 2010; Abou-Hamdan et al., 2015). Once H₂S binds to SQR (Olson, 2011), electrons are donated to the ETS via ubiquinone to complex III. Thus, H₂S at low levels is capable of stimulating respiration by functioning as an inorganic substrate (Goubern et al., 2007; Módis et al., 2014; Szabo et al., 2014; Yong and Searcy, 2001). However, sustaining ETS electron transport by sulfide oxidation is energetically inefficient as it consumes three times more oxygen than substrates such as NADH or succinate to generate the same proton gradient (Abou-Hamdan et al., 2015; Marutani et al., 2021). At high H₂S levels reaching SQR capacity, there is less H₂S oxidation and inhibition of ETS at complex IV increases. H₂S also inhibits its own oxidation, resulting in a positive feedback loop (Abou-Hamdan et al., 2015). Estimated values of ∼1–5 and ∼10–40 µmol l⁻¹ H₂S are considered stimulatory and inhibitory, respectively, in several cell types (Lagoutte et al., 2010), but effective ranges of values would obviously depend on the SQR concentration and specific enzymatic activity present in a given organism or tissue. Thus, H₂S inhibits mitochondrial respiration at high concentrations but stimulates it at low concentrations.

Recently, we have investigated the effects of H₂S on mitochondrial respiration in the thirteen-lined ground squirrel (Jensen et al., 2021). In that study, we isolated liver mitochondria from squirrels during summer or hibernation during torpor and IBE and found a low SQR activity and a strong, dose-dependent inhibition of mitochondrial respiration by exogenous H₂S during torpor (Jensen et al., 2021). These findings imply that during torpor, inhibition of H₂S breakdown by a less active SQR allows levels of H₂S to increase and inhibit cytochrome c oxidase. Accordingly, endogenous in vivo levels of bioavailable H₂S increased to inhibitory levels (∼90 µmol l⁻¹) in the liver during torpor in hibernating thirteen-lined ground squirrels (Jensen et al., 2021). Thus, SQR appears as a key enzyme involved in the control of mitochondrial respiration by H₂S during torpor, at least in the liver, where metabolic suppression is particularly pronounced during torpor compared with other tissues (Brown and Staples, 2014; Brown et al., 2012; Muleme et al., 2006). This mechanism of SQR-mediated regulation of H₂S is highly tissue specific, as in mice SQR is absent in neural tissue, making the brain more sensitive to H₂S (Marutani et al., 2021), but highly expressed in the intestine (Lagoutte et al., 2010; Malagrinò et al., 2019). Interestingly, a recent study (Marutani et al., 2021) reports that the brain of the thirteen-lined ground squirrel contains 100-fold higher levels of SQR protein than that of mice and rats, correlating with a higher tolerance to sulfide inhibition of oxidative phosphorylation but also with a high tolerance to ambient hypoxia. These effects were reversed by SQR silencing (Marutani et al., 2021). It appears conceivable that species- and tissue-specific variations in the level of SQR expression, intrinsic activity and the presence of available sites for post-translational modifications are crucial in setting the threshold where H₂S has inhibitory or toxic effects, or no effect at all. Possibly, these variations explain in part the variable outcome of inhaled H₂S reported in the literature on various animal models (Table 1).

In addition to complex IV, H₂S may also interact with other mitochondrial complexes. For instance, H₂S has been shown to persulfidate two reactive cysteines at complex V, which increases its enzyme activity (Módis et al., 2016). Whether this occurs in hibernators remains to be investigated.

The inhibition of mitochondrial respiration by H₂S proposed in the thirteen-lined ground squirrel during torpor (Fig. 3) involves decreased SQR activity, which hampers the ability of mitochondria to metabolize H₂S (Jensen et al., 2021). In principle, these effects can be due in part to increased SQR protein degradation and resynthesis in torpor and IBE, respectively, but in the context of energy-saving hibernation, rapid and reversible post-translational modifications may provide a suitable and energetically inexpensive mechanism for the regulation of SQR activity. Protein acetylation is a prevalent post-translational protein modification during torpor in the thirteen-lined squirrel liver (Hindle et al., 2014), suggesting that cycles of acetylation and deacetylation may control SQR activity. In support of this possibility, an unidentified liver mitochondrial protein with the same molecular mass as SQR (∼50 kDa) has been found to be acetylated during torpor, but not during IBE in thirteen-lined ground squirrels (Mathers and Staples, 2019). Moreover, an acetylation site in mouse SQR has been reported (Rardin et al., 2013), suggesting that acetylation regulates SQR activity.

Sirtuin 3 is a major mitochondrial deacetylase, and deacetylation by sirtuin 3 has been correlated with increased activity of complex I, II and IV (Ahn et al., 2008; Cimen et al., 2010; Wu et al., 2013). It is unknown whether SQR is among sirtuin 3 targets, but in the liver of thirteen-lined ground squirrels, sirtuin 3 expression is significantly lower during winter compared with summer, with no difference between torpor and IBE (Hindle et al., 2014). Further complicating the picture, sirtuin 3 may itself be activated by S-persulfidation (Yuan et al., 2019), suggesting an intricate mechanism where high H₂S may activate sirtuin 3 and trigger deacetylation of SQR, which would lead to accumulation of H₂S. Future work will clarify whether SQR is reversibly acetylated during the hibernation season and by which mechanism.

SQR expression is itself H₂S-dependent and regulated by the transcription factor nuclear factor-erythroid 2-related factor 2 (Nrf2), which also controls the expression of antioxidant genes such as thioredoxin and glutathione reductase (Harvey et al., 2009; Kensler et al., 2007). Nrf2 is tightly controlled by the repressor protein Keap1, which promotes Nrf2 degradation. Keap1 itself can be inactivated by H₂S, which persulfidates two cysteines on Keap1 (Hourihan et al., 2013; Yang et al., 2013). This alleviates repression of Nrf2, which subsequently translocates to the nucleus and facilitates gene transcription (Fig. 3). The H₂S-dependent stimulation of Nrf2 stability has been demonstrated in mice, where intravenous injection of H₂S increases Nrf2 levels in cardiac tissue after 30 min (Calvert et al., 2009). In thirteen-lined ground squirrels, levels of Nrf2 increase in the heart, liver and brown adipose tissue during torpor (Morin et al., 2008), and elevated H₂S levels during torpor could potentially contribute to Nrf2 stability and an increase in SQR expression. A study on Nrf2-deficient mouse embryonic fibroblasts found no change in the expression of SQR upon treatment with H₂S, whereas SQR expression increased in wild-type cells (Hourihan et al., 2013). Thus, increased levels of H₂S can increase Nrf2 stability, which may then contribute to enhanced SQR expression, leading to increased H₂S oxidation. This negative feedback loop acts as a safety valve that prevents H₂S levels from accumulating to toxic levels, but, as it involves protein synthesis, it would be an energetically expensive control mechanism in the context of torpor and arousal, but could be feasible in different seasons. During torpor, SQR activity is inhibited (Jensen et al., 2021), so its Nfr2-mediated expression is probably not substantial. Although SQR regulation by H₂S remains unclear in hibernators, potential mechanisms involve its post-translational acetylation and/or Nrf2-induced expression, both of which remain to be investigated further.

In addition to metabolic rate, H₂S controls a variety of physiological responses, such as vascular tone (Yang et al., 2008; Zhao et al., 2001), neuromodulation (Abe and Kimura, 1996; Kimura and Kimura, 2004) and cytoprotection (Calvert et al., 2010; Elrod et al., 2007). The redox potential and reactivity of H₂S are comparable to those of glutathione and cysteine (Carballal et al., 2011; Mishanina et al., 2015), but because in vivo levels of H₂S are much lower than those of glutathione and free cysteine (Mishanina et al., 2015), any direct ROS scavenging by H₂S is probably negligible. Instead, H₂S may be indirectly involved in protective mechanisms of ROS scavenging and enhancement of antioxidant defense through increasing activity and levels of antioxidants, such as glutathione and superoxide dismutase (Kimura and Kimura, 2004; Kimura et al., 2010; Sun et al., 2012).

Some mismatch between oxygen supply and consumption resembling ischemia/reperfusion may occur in the transitions between torpor and arousal and potentially cause a surge of mitochondrial ROS, but this does not seem to result in major oxidative stress in hibernators and effects appear to be tissue specific and transient (Staples, 2016). Hibernators appear in general more resistant to oxidative damage than mammals of similar size, suggesting some adaptive traits in these animals. Cardiac arrest does not damage hippocampal neurons (Dave et al., 2006) or the liver (Bogren et al., 2014) of Arctic ground squirrels compared with rats. Livers of thirteen-lined ground squirrels in the hibernation season tolerate ischemia/reperfusion better than those of rats or summer squirrels (Lindell et al., 2005). Arousal in Arctic ground squirrels is associated with some oxidative damage in brown adipose tissue but not in the liver (Orr et al., 2009).

In addition to metabolic suppression of torpor, H₂S could be involved in cytoprotection during the hibernation season, although more evidence is needed. Treatment with H₂S has been found to decrease ROS levels under ischemia/reperfusion of rat cardiomyocytes (Sun et al., 2012) and to prevent oxidative damage in porcine kidneys (Maassen et al., 2019). H₂S may also act by increasing antioxidant capacity through stimulation of the Nrf2 transcription pathway, as discussed above (Fig. 3). In general, H₂S has a broad range of functions, and thus H₂S signaling may be relevant in many physiological responses during hibernation, including changes in peripheral vasoconstriction, where H₂S may contribute through its ability to regulate vascular tone (Yang et al., 2008; Zhao et al., 2001).

Significant changes in the endogenous levels and composition of H₂S metabolites have been reported recently in several hibernating mammals. Although changes in the activity and expression of enzymes involved in H₂S production may be involved (Talaei et al., 2012), a key regulatory site appears to be the mitochondrial enzyme SQR, which controls H₂S levels in vivo via oxidation (Jensen et al., 2021; Marutani et al., 2021). The underlying mechanism involves decreased SQR activity during torpor, possibly mediated by acetylation, which causes a build-up in vivo of H₂S with consequent inhibition of complex IV and metabolic suppression. Several aspects need to be examined further, especially identifying the regulatory mechanism of SQR activity by post-translational modifications, possibly including reversible binding of acetyl group or H₂S. Other aspects that remain to be investigated are redox buffer capacity and the content of accessible cysteines in hibernator proteins and how these may help regulate in vivo levels of H₂S and determine thresholds of inhibitory levels and toxicity. Moreover, H₂S may be indirectly involved in other adaptive mechanisms during hibernation, such as cytoprotective antioxidant defense. Investigations on the nature of H₂S signaling and of mitochondrial regulation in comparative studies appear critical to improve our understanding on the mechanisms of torpor and hibernation.

We thank three anonymous reviewers for useful suggestions.