The exceptional longevity of the naked mole‐rat may be explained by mitochondrial antioxidant defenses

Abstract Naked mole‐rats (NMRs) are mouse‐sized mammals that exhibit an exceptionally long lifespan (>30 vs. <4 years for mice), and resist aging‐related pathologies such as cardiovascular and pulmonary diseases, cancer, and neurodegeneration. However, the mechanisms underlying this exceptional longevity and disease resistance remain poorly understood. The oxidative stress theory of aging posits that (a) senescence results from the accumulation of oxidative damage inflicted by reactive oxygen species (ROS) of mitochondrial origin, and (b) mitochondria of long‐lived species produce less ROS than do mitochondria of short‐lived species. However, comparative studies over the past 28 years have produced equivocal results supporting this latter prediction. We hypothesized that, rather than differences in ROS generation, the capacity of mitochondria to consume ROS might distinguish long‐lived species from short‐lived species. To test this hypothesis, we compared mitochondrial production and consumption of hydrogen peroxide (H2O2; as a proxy of overall ROS metabolism) between NMR and mouse skeletal muscle and heart. We found that the two species had comparable rates of mitochondrial H2O2 generation in both tissues; however, the capacity of mitochondria to consume ROS was markedly greater in NMRs. Specifically, maximal observed consumption rates were approximately two and fivefold greater in NMRs than in mice, for skeletal muscle and heart, respectively. Our results indicate that differences in matrix ROS detoxification capacity between species may contribute to their divergence in lifespan.

result, comparisons between these two species are considered to be a "gold standard" in mammalian studies of aging (Buffenstein, 2005;Dammann, 2017;Edrey et al., 2011).
According to the oxidative stress theory of aging (Barja, 2013;Harman, 1956), senescence is caused by the gradual accumulation of oxidative damage to cells, inflicted by reactive oxygen species (ROS) of mitochondrial origin. However, previous comparative studies of NMR biology mostly provided evidence that contradicted this theory. For example, comparisons of isolated heart mitochondria found no difference in the rate of H 2 O 2 efflux (i.e., the proportion of H 2 O 2 not consumed by the mitochondrion before detection, Munro, Banh, Sotiri, Tamanna, & Treberg, 2016) between NMRs and mice . In addition, extensive oxidative damage and limited antioxidant capacity have been reported in the cytosol of NMR hepatocytes (Andziak, O'Connor, & Buffenstein, 2005;Andziak et al., 2006). Taken together, these findings led to the conclusion that the longevity of NMRs occurs independently of enhanced protection against oxidative damage (reviewed in Lewis, Andziak, Yang, & Buffenstein, 2013), and this conclusion has been used repeatedly to refute the oxidative stress theory of aging (Hekimi, Lapointe, & Wen, 2011;Robb, Christoff, Maddalena, & Stuart, 2014;Stuart, Maddalena, Merilovich, & Robb, 2014).
This refined hypothesis stems from the fact that mitochondrial ROS are mostly released inside the mitochondrion (i.e., within the mitochondrial matrix), thereby directly exposing mitochondrial biomolecules to oxidative damage. According to the mitochondrial stress hypothesis, cellular senescence is primarily driven by loss of mitochondrial function with age. A central step toward testing this hypothesis would be to measure the balance between internal production and internal consumption of ROS within mitochondria themselves.
We have recently shown that traditional methodologies for detecting the rate of H 2 O 2 formation from isolated mitochondria underestimate ROS generation because of the remarkable endogenous capacity of matrix antioxidants to consume H 2 O 2 . For example, this underestimation can reach >80% in rat skeletal muscle with certain respiratory substrates (See Figure 1 in methods; Munro et al., 2016). Moreover, mitochondria can consume far more H 2 O 2 than they generate (Drechsel & Patel, 2010;Starkov et al., 2014;Zoccarato, Cavallini, & Alexandre, 2004); therefore, this capacity of mitochondria to consume H 2 O 2 putatively represents a novel and widely underappreciated test of the mitochondrial oxidative stress theory of aging in of itself. We hypothesized that differences in the capacity of mitochondria to eliminate H 2 O 2 might solve the apparent NMR oxidative stress/longevity-conundrum (Lewis et al., 2013).
To test our hypothesis, we took advantage of antioxidant inhibition methods that we developed previously (Munro et al., 2016) to measure H 2 O 2 formation rates without the confounding influence of internal consumption (Figure 1). We also compared mitochondrial H 2 O 2 clearance (i.e., maximal consumption) rates between these two species in functional isolated mitochondria (Drechsel & Patel, 2010;Lopert & Patel, 2014;Munro et al., 2016;Starkov et al., 2014;Zoccarato et al., 2004). Our results support the mitochondrial oxidative stress hypothesis of aging via a mechanism that has not been previously demonstrated: NMRs and mice do not differ in their rate of H 2 O 2 formation, but rather in the markedly greater capacity of NMR mitochondria to consume H 2 O 2 .

| Oxygen consumption
Mitochondrial oxygen consumption was measured simultaneously with H 2 O 2 formation, and these respiration rate data are reported in the Supporting Information (Figures S1 and S2). When measured at the species' respective body temperatures, the respiratory control ratio (RCR) values for NMR and mouse skeletal muscle mitochondria, respectively, were (mean ± SEM) 11.0 ± 0.4 and 9.8 ± 0.6 when fueled by malate + glutamate, and 3.7 ± 0.4 and 2.8 ± 0.1 when fueled by succinate + rotenone. These values were not significantly different between species within each substrate condition (Supporting Information Figure S2). ) but also directly as H 2 O 2 . Superoxide released inside and outside the matrix will be converted into H 2 O 2 by the Cu/ZnSOD and MnSOD, respectively. The proportion released inside is additive with the existing pool of H 2 O 2 , leading to two ultimate fates: (a) diffusion across membranes to reach the detection system, or (b) consumption by matrix-based antioxidants pathways. The reductases of the GSH-and Trx-dependent pathways are activated by provision of NADPH, when substrate is oxidized, and thus concomitantly with ROS formation. Inhibitors for the GSH (CDNB)and Trx (auranofin)-dependent pathways (also used in this study) are depicted in red heart mitochondria was significantly higher for NMRs 21.0 ± 0.7 compared to mice 7.6 ± 0.4 (p < 0.0001). Our mouse heart mitochondria nonetheless had a higher RCR for NADH-generating substrates than reported by other groups (range = 3-6; Graham et al., 1997;Hughes & Hekimi, 2011).

| The rate of H 2 O 2 formation is not consistently lower in NMR muscle or heart mitochondria
We recently validated pharmaceutical inhibition methods that compromise the glutathione (GSH)-and thioredoxin (Trx)-dependent ROS scavenging pathways in skeletal muscle mitochondria of rodents, without affecting mitochondrial energetics (Figure 1; Munro et al., 2016). By blocking the endogenous consumption of H 2 O 2 by mitochondria, this inhibitory approach allows us to accurately estimate the combined rate of superoxide and H 2 O 2 formation using fluorescence assays based on mitochondrial H 2 O 2 efflux.
When normalized to mg of protein, H 2 O 2 production rates from skeletal muscle mitochondria were higher for the mouse than for the NMR, with the exception of experiments in which malate + glutamate + succinate were supplied simultaneously in the absence of ADP (Figure 2a,b). The differences between the two species tended to be greater in conditions in which H 2 O 2 is produced at low, and therefore, more physiologically relevant rates. When normalized to the activity of the mitochondrial-specific enzyme citrate synthase, rates of H 2 O 2 production did not show any consistent trend between species and across conditions, with rates being higher for the NMR with malate + glutamate and for malate + glutamate + succinate (Figure 2c,d), but lower with malate or malate + glutamate + ADP when mitochondria of both species are measured at 37°C. The production of H 2 O 2 can also be expressed as a proportion of electrons prematurely "leaking" from the electron transport system for the single-electron reduction of oxygen (O 2 ) into superoxide (O2 •− ) or the two-electron reduction of oxygen to H 2 O 2 . The % electron leak was consistently higher from NMR mitochondria at 30°C than from mouse mitochondria at either 30 or 37°C across assay conditions, with the exception of malate + glutamate + succinate in the presence of ADP (Figure 2e,f). Therefore, depending on the denominator chosen, completely different conclusions regarding which species produces more or less H 2 O 2 may be supported by the same dataset.
To determine whether these differences persisted in other tissues, we also measured the production of H 2 O 2 from heart mitochondria at each species' physiological temperatures and normalized results to mg of protein. We found no significant differences between species except for a higher production of H 2 O 2 in NMR mitochondria during malate + glutamate oxidation (Figure 3a). Electron leakage was found to be much higher from NMR mitochondria than from mouse mitochondria in the absence of ADP, however, when ADP was present, this difference vanished, with a trend toward lower % leakage in NMR for malate + glutamate + ADP (
For skeletal muscle, the maximal rate of H 2 O 2 consumption with malate was much higher for NMR than for mouse mitochondria, irrespective of normalization to mg of protein or to the activity of citrate synthase (Figure 4a For heart mitochondria, the maximal rate of H 2 O 2 consumption, estimated with malate, was more than five times greater for the NMR than for the mouse (

| Normalizing H 2 O 2 production to the rate of consumption
The degree of oxidative insult sustained by mitochondria ultimately depends on the balance between ROS formation and elimination within the organelle. Our approach of separating H 2 O 2 formation rates from H 2 O 2 consumption rates allowed us to integrate our data for the purpose of comparing the potential for self-inflicted oxidative damage between species. Formation rates, for each respiratory substrate condition, were divided by maximal consumption rates, which is likely invariable across substrate conditions used in the current  Figure 1) were inhibited by CDNB pre-treatment and auranofin. (a, c, and e) assays for both species at the normal body temperature of the naked mole-rat (NMR; 30°C) and the mouse (37°C). (b, d, and f), results are limited to assaying each mitochondrial preparation at each species' normal body temperature. Data are normalized to mg of mitochondrial protein (a and b), to the activity of the citrate synthase measured at the same temperature (c and d), or as the % of electrons leaking from the electron transfer system toward the formation of H 2 O 2 (e and f). M: malate; G: glutamate; Rot: rotenone; S: succinate. Significant differences between species were assessed using t tests, with *p < 0.05, **p < 0.01, and ***p < 0.001. For mouse, n = 4 for M and MG, n = 9 for MGS and MGS + ADP, and n = 7 for all conditions with succinate. For NMR, n = 6 for all conditions except at 37°C (n = 2). Data are mean ± SEM, except for NMR at 37°C where bars represent range study (Munro & Treberg, 2017;Treberg et al., 2015), providing an oxidant index. Figure 6a,b shows that the oxidant index of skeletal muscle mitochondria is substantially lower for NMRs in all conditions of temperature or substrate tested. In particular, note the pronounced difference between the two species in the more biologically relevant conditions where: (a) ATP is being generated (i.e., in the presence of ADP), and (b) the electron transport system is fed mainly through complex I (malate and glutamate) or both complex I and II simultaneously (malate + glutamate + succinate). Figure 6c shows that the oxidant index difference is even larger for heart mitochondria, being up to 6.1 times lower for the NMR in the physiologically relevant condition of MGS + ADP.

| Relative rates of activity among complexes of the electron transport system
Past studies often attributed the lower rate of H 2 O 2 efflux observed in mitochondria of long-lived species (during succinate oxidation) to a lower content (or activity) of complex I of the electron transport system (Ayala et al., 2007;St-Pierre, Buckingham, Roebuck, & Brand, 2002). We measured enzymatic activity of complexes I, II, and IV in skeletal muscle mitochondria to test this possibility. Figure 7 shows the activity of complexes of the electron transport system measured at each species' normal body temperature. As predicted by the hypothesis, the activity of complex I is lower for the NMR than for the mouse when normalized to mg of protein, but the same is found for the activity of all other complexes as well as for citrate synthase ( Figure 7a). This finding could simply suggest that NMRs possess a lower content of enzymes involved in oxidative phosphorylation relative to overall mitochondrial protein. Indeed, when normalized to the activity of citrate synthase, complex I activity was higher in NMRs compared to murine mitochondria, whereas complex IV activity was no longer different between species (Figure 7b). A more consistent difference between species, however, is the remarkably lower activity of complex II in NMR muscle, irrespective of the marker used for normalization.  (Barja, 2013), without provid-

| DISCUSSION
ing a clear answer .
Here, we take a radically different approach: we use methods that

| Matrix antioxidants and aging
Antioxidant concentrations and activities do not generally correlate (negatively or positively) with longevity (reviewed in Hulbert, Pamplona, Buffenstein, & Buttemer, 2007); however, these investigations rarely focussed on mitochondrial antioxidants despite early evidence that the relationship may work better in this case. For example, looking across three age classes, it is not obvious that the activity of the cytosolic (and intermembrane space)-localized superoxide dismutase (Cu/ZnSOD), and total catalase activity are higher for the NMRs as compared to mice, and the cytosolic glutathione peroxidase is clearly more active in mice than in NMRs. However, the matrix-specific superoxide dismutase (MnSOD) is more active in NMRs than in mice for the young-and middle-aged classes (Andziak et al., 2005).
Importantly however, measuring the isolated activity of matrixspecific peroxidases or reductases is likely insufficient to draw mean- Sample sizes are n = 3 and n = 5 for NMR and mice, respectively. Each mitochondrial preparation was obtained by pooling four hearts. Other details and conditions of assay are as in Figure 2 Starkov et al., 2014) show that the relative contribution of each of the GSH-and Trx-dependent pathways is highly variable across species and tissues. Therefore, targeting one or the other of the peroxidases linked to these pathways (i.e., the GSH peroxidase or the peroxiredoxine) may be misleading. Second, in vivo, matrix antioxidant enzymes work as part of an integrated pathway (i.e., the GSHor Trx-dependent pathways) responsible for providing reducing equivalents to the peroxidase (Figure 1). To our knowledge, very little is known regarding the relative contribution to overall flux control exerted at each steps of these pathways. Indeed, this rationale can be extended to the whole mitochondrion-that is, the reductases for the GSH-and Trx-dependent pathways are exclusively dependent on the provisioning of NADPH (Drechsel & Patel, 2010;Munro et al., 2016;Starkov et al., 2014;Treberg et al., 2015;Zoccarato et al., 2004), originating from the oxidation of respiratory substrates by the Krebs cycle and directly from other matrix enzymes. . Aura = auranofin, an inhibitor of the thioredoxin reductase. For mouse skeletal muscle, n = 4 for all conditions, except for no substrate and M at 37°C, where n = 6. For NMR skeletal muscle at 37°C, n = 3; at 30°C, n = 4 for MGS +ADP, n = 5 for MGS and MGS + Aura, and n = 6 for all other conditions. For the heart, n = 4 (4 animals per preparation) for all. Other details are as in Figure 2 Third, it has been suggested that a protonmotive force across the inner mitochondrial membrane is necessary for the conversion of NADH to NADPH in certain species (Lopert & Patel, 2014 (Figure 3, . Here, we found that, for heart mitochondria, the maximal capacity for the consumption of H 2 O 2 was more than five times higher for the NMR as compared to mice, pointing to a major means of modulating the overall H 2 O 2 balance (Munro & Treberg, 2017).

Previous genetic and pharmaceutical interventions have already
suggested a potential link between mitochondrial detoxification of H 2 O 2 and lifespan in other models. For example, the overexpression of catalase targeted to mitochondria is one of the rare genetic interventions that can extend the lifespan of vertebrates (Dai et al., 2009;Schriner et al., 2005). Similarly, synthetic antioxidants chemically targeted to mitochondria (e.g., SkQ1) can extend healthspan of wild-type rodents (Skulachev et al., 2009)  Heart F I G U R E 5 Cumulative allocation of H 2 O 2 consumption among the three major matrix consumers of H 2 O 2 for skeletal muscle (a and b) and heart (c) mitochondria. Total consumption corresponds to the maximal consumption obtained during malate oxidation (from Figure 4, see text for rationale). The Trx-dependent proportion was defined by subtracting consumption with malate + auranofin from total consumption (See Figure 1). Residual consumption is the consumption in absence of exogenously added substrate. The GSH-dependent proportion was defined as the proportion of total consumption not explained by the sum of the Trx-dependent pathway and residual consumption, on the premise that only the three consumers of H 2 O 2 considered here are important in the matrix. Residual consumption can be explained by the activity of catalase as well as by residual activity of the respiration-dependent pathways for the consumption of H 2 O 2 , which are partially energized by endogenous (slow oxidizing) substrates remaining in the matrix after mitochondrial isolation (see Munro et al., 2016). Catalase was previously found to represent a negligible component of the residual consumption for rat skeletal muscle mitochondria (Munro et al., 2016), but the higher levels of residual consumption found here for the NMR in both tissues suggests at least some involvement of catalase

| Traditional measures of mitochondrial H 2 O 2 production should be interpreted with caution
Previous studies measured the proportion of H 2 O 2 diffusing out of the mitochondrion, which represent the proportion not consumed by matrix antioxidants. The problem with this approach is that most superoxide (and thus ultimately most H 2 O 2 ) is produced inside the mitochondrion, and as much as 80% may be consumed therein. As a result, this contribution remains "invisible" to the detection system. Therefore, differences reported between species using such a detection system may equally result from mitochondria having a lower H 2 O 2 formation rate per se as from mitochondria having greater H 2 O 2 consumption capacities (Munro et al., 2016;Treberg et al., 2015;Treberg, Quinlan, & Brand, 2010).
A particularly problematic aspect of the internal consumption of H 2 O 2 with traditional measures is that the lower the actual rate of H 2 O 2 formation, the greater the proportion that is consumed inside mitochondria (Munro et al., 2016). Unfortunately, experimental conditions using substrates that are generally considered more representative of the in vivo milieu (e.g., in presence of ADP and a complex assemblage of respiratory substrates at typically sub-saturating concentrations) generally tend to produce less H 2 O 2 (Goncalves, Quinlan, Perevoshchikova, Hey-Mogensen, & Brand, 2015). Hence, it is very likely that the more representative a combination of substrate and effectors is of the in vivo milieu, the greater the difference between species actually (albeit indirectly) represents a difference in matrix H 2 O 2 consumption capacities.

| The choice of denominator impacts conclusions
Previous comparative studies have mostly used protein content in the mitochondrial pellet as a means of normalizing rates of H 2 O 2 formation by mitochondria. Normalizing H 2 O 2 formation to the consumption of oxygen in the same condition of respiratory substrates is also often used. Here, we tested these two markers of mitochondrial density when normalizing H 2 O 2 formation rate. In addition, we also used the activity of citrate synthase. These three denominators provided us with three different conclusions with respect to determining which species produces more H 2 O 2 (Figure 2). The degree of

| The oxidant index as an unbiased measure of oxidative insult inside mitochondria
In this study, we divided the rate of H 2 O 2 formation by the maximal rate of H 2 O 2 consumption to obtain an oxidant index. This approach avoids any bias associated with the use of a particular marker of mitochondrial density. For instance, the oxidant index is not biased by changes in the density of electron transport system enzyme complexes, the capacity for oxygen consumption, or the protein content of mitochondria across species. Instead, the oxidant index directly assesses the ratio between the opposing processes of production and elimination of H 2 O 2 within the matrix, and thereby allows the direct determination of whether or not the matrix of one species sustains a higher degree of oxidative insult. We suggest that the oxidant index has the potential to eliminate many confounding factors in the field that have precluded the reaching of a consensus conclusion regarding a putative relationship between longevity and mitochondrial ROS in previous studies Montgomery, Hulbert, & Buttemer, 2012).

| Does the NMR biogerontology model support a role for oxidative damage in aging?
Results from the comparison between the NMR and the mouse have often been presented as evidence against the hypothesis that oxidative damage is involved in aging and longevity, and in particular (curiously) to refute the now obsolete oxidative stress theory of aging Lewis et al., 2013;Stuart et al., 2014). For example, higher steady-state levels of oxidative stress markers including protein carbonyls, urinary and cellular isoprostanes, liver 8-OHdG, malondialdehyde, and a lower activity of cytosolic glutathione peroxidase have been found in NMR hepatocytes compared to murine hepatocytes (Andziak et al., 2005(Andziak et al., , 2006. Many authors however do not distinguish between the classical oxidative stress theory of aging, first suggested by Harman in 1956 (i.e., that senescence is the result of oxidative damage to cells), and the more recent mitochondrial oxidative stress hypothesis of aging (i.e., that senescence is the result of oxidative damage to mitochondria themselves).
Whereas the latter hypothesis has received support from transgenic models (Kujoth et al., 2005;Kukat & Trifunovic, 2009;Trifunovic et al., 2004), the former has been convincingly refuted Stuart et al., 2014). Our finding of better protection against H 2 O 2 in the matrix of NMR mitochondria, as well as the previous finding of greater protection against superoxide (higher activity of the MnSOD, Andziak et al., 2005), supports the mitochondrial oxidative stress hypothesis of aging. A surprising finding, however, is that recruiting matrix antioxidants rather than downregulating basal ROS production inside the matrix of long-lived species appears to be the predominant mechanism of protection.

| Is there a longevity signature in the stoichiometry of the electron transport system?
Previous studies reported lower activity/content of complex I of the electron transport system and suggested this may represent a mechanism for increasing longevity by decreasing mitochondrial H 2 O 2 formation rate during "reverse electron transport" (Ayala et al., 2007;St-Pierre et al., 2002), with succinate as a substrate. Here, we report that complex II, and not complex I, has a distinctively lower activity in the NMR as compared to the mouse (Figure 7). The same trend is found for mantle mitochondria from the longest-lived metazoan, the  F I G U R E 7 Activity of electron carrying complexes, and the Kreb's cycle enzyme citrate synthase. Data were acquired using untreated skeletal muscle mitochondria thawed only once, except for complex I activity which requires three cycles of freeze-thaw (Spinazzi et al., 2011). Assays were conducted at the species' normal body temperature and normalized to mg of protein (a), or to the activity of citrate synthase (b). CI = complex I, CII = complex II, CIV = complex IV, CS = citrate synthase, NMR = naked mole-rat. In panel a, for mouse, n = 15, 16, 15, 10, 11 from left to right, for NMR, n = 8 except for CI, where n = 6. In panel b, for mouse n = 10 (except for CI, n = 11), for NMR, n = 8 (except for CI, n = 6). Other details of statistics are as in Figure 2 pertinent to mitochondrial oxidant burden in two ways; it can be an important source of superoxide or H 2 O 2 by itself , and complex II also supplies most of the electrons to the ubiquinone pool during succinate oxidation and the associated "reverse electron flow". Reverse electron flow is particularly sensitive to the protonmotive force, and higher level of complex II activity would maintain a higher protonmotive force for an equal content of complex I, leading to a higher ROS production rate (Lambert & Brand, 2004). Further studies are required to clarify the pattern of change in mitochondrial electron transport system between short-and longlived species.  (Shabalina et al., 2017;Skulachev et al., 2009). Future studies are required at this point to investigate whether greater mitochondrial capacity to consume H 2 O 2 is a generalized trait across long-lived species relative to their shorter-lived counterparts, which would represent a major paradigm shift in the field of aging.

| Reagents
Auranofin was obtain from Enzo Life Sciences. CDNB (1-chloro-2,4dinitro-benzene) was obtained from Milipore-SIGMA, Oakville, Canada. Amplex Ultrared was obtained from THERMO-FISHER, Waltham, MA. All other chemicals and enzymes were purchased from Milipore-SIGMA or ACROS, Geel, Belgium. Individuals were fed a standard chow diet ad libitum and sacrificed by CO 2 asphyxiation followed by rapid cervical dislocation. Male naked mole-rats of approximately corresponding biological age (i.e., between 2 and 3 years old, 38-50 g) were obtained from the colony of Dr. Matthew Pamenter, University of Ottawa, Canada. Individuals were group-housed in interconnected cages simulating burrows with tunnel systems, and kept at an ambient temperature of 30°C, and 50% humidity. NMR were fed fresh fruits, vegetables, and Pronutro cereal supplements ad libitum and sacrificed by exposure to isoflurane (skeletal muscle data) followed by cervical dislocation, or directly by cervical dislocation (heart data). For both species, hind legs and dorsal skeletal muscles were dissected and mitochondria were isolated by differential centrifugation as previously described (Affourtit, Quinlan, & Brand, 2012). Mitochondrial protein concentration was determined by the Biuret assay in the presence of 0.2% sodium deoxycholate using bovine serum albumin as a standard.

| Inhibition of matrix H 2 O 2 consumption
Measures of H 2 O 2 production rates were acquired from mitochondria treated to inhibit matrix-based H 2 O 2 consumption ( Figure 1).
Specifically, the glutathione (GSH)-dependent enzymatic pathway was inhibited by use of the GSH S-transferase substrate CDNB (1chloro-2,4-dinitro-benzene) which irreversibly binds to GSH, as explained in (Treberg et al., 2010). In accordance with preliminary tests for this study, the exposure period of mitochondria to CDNB at room temperature was reduced from 5 to 4 min which prevented changes in respiration rates in the presence of glutamate-malatesuccinate and ADP. The Trx-dependent enzymatic pathway was inhibited by 0.5 µM auranofin, an inhibitor of the thioredoxin reductase, which was added directly to the assay medium. In the absence of calcium (Rigobello, Scutari, Boscolo, & Bindoli, 2002), auranofin has no side effects on mitochondrial energetics, up to a concentration of 2 µM in rats (Munro et al., 2016) and NMRs, and up to 0.5 µM in mice (preliminary tests, present study). Further details of these methods and their degree of selectivity can be found in (Munro et al., 2016;Treberg et al., 2010).

| Consumption of H 2 O 2 (presented as change in H 2 O 2 ·min −1 in the figures)
Hydrogen peroxide consumption assays were performed with untreated mitochondria as described in (Munro et al., 2016), using between 0.05 and 0.07 (skeletal muscle); and between 0.03 and 0.04 (heart) mg mitochondrial protein.

| Enzyme assays (skeletal muscle)
Frozen (−80°C) mitochondria were thawed only once prior to measurements, except for the complex I (NADH-ubiquinone oxidoreductase) assay which requires three cycles of freeze-thaw to maximize activity according to (Spinazzi et al., 2011). Assays were conducted at 30 or 37°C (see figures), following the methodology developed by (Spinazzi et al., 2011). For complex IV (cytochrome c oxidase) activity, lauryl maltoside was used at 0.05% in order to maximize activity.

| Presentation of data
For data acquisition from heart mitochondria, animals had to be pooled by groups of 4 in each species. Due to limited availability of NMRs, we were unable to recover sufficient mitochondria to perform some of the additional assays carried out with skeletal muscle mitochondria. The means of normalizing mitochondrial rates is a matter of debate in comparative studies (Hulbert, Turner, Hinde, Else, & Guderley, 2006;Larsen et al., 2012). With the more abundant skeletal muscle mitochondria, we used a number of frequently used markers of mitochondrial density (mg of mitochondrial protein, citrate synthase activity, and oxygen consumption) to investigate whether differences between species are dependent on the use of one particular denominator or if they represent generalized trends across denominators.
The mesothermic NMR maintains its body temperature at 1-2°C above ambient temperatures when within its thermoneutral zone of 30-34°C (Buffenstein, Park, Hanes, & Artwohl, 2012). NMR colonies were housed at 28-30°C, and therefore we chose 30°C as the assay temperature at which to perform assays examining NMR mitochondria. In contrast, mice have a more typical mammalian body temperature and therefore we performed all measurements in this species at 37°C. With skeletal muscle mitochondria, we also performed several assays at both of these two temperatures (30 and 37°C) in each species. Data collected at only one temperature were measured at the respective physiological temperatures of each species to provide physiologically relevant comparisons (NMR at 30°C and mouse at 37°C).

| Statistical analysis
Significant difference between the two species was determined by Student's t test separately for each of the three biologically relevant temperature contrasts that is, NMR at 30 vs. mice at 37°C, both species at 37°C, and both species at 30°C. Homogeneity of variance was assessed with Brown-Forsyth's when n > 5, or as a plot of residual when n < 5. In the absence of homoscedasticity, the Wilcoxon test was used to distinguish between the two species. Analyses were conducted using JMP 13.0 (SAS Institute, Cary, NC). In all cases, p < 0.05 was considered significant, sample size is presented in figure legends. Data are presented as mean ± SEM.

ACKNOWLEDG MENTS
We

CONFLI CT OF INTEREST
None declared.

AUTHOR CONTRI BUTION
D.M. was responsible for conducting the experiment and analyzing data. C.B. was responsible for acquiring consumption data with heart