Proteomic Profiling Reveals Adaptive Responses to Surgical Myocardial Ischemia–Reperfusion in Hibernating Arctic Ground Squirrels Compared to Rats

ISCHEMIA–reperfusion (I/R) injury, a consequential component of cardiac surgery and organ transplantation, is a major determinant of morbidity and mortality. Cardioprotection after I/R remains an elusive target despite significant research and numerous compounds with promising preclinical data.¹  Mammalian hibernation is a seasonal phenomenon characterized by prolonged bouts (days to weeks) of winter torpor, a state of decreased physiologic activity, reduced body temperature, and decreased metabolic rate, interspersed with brief (hours to a day) periods of arousal when metabolism and temperature temporarily return to normal.²  The entry into and exit from torpor is akin to I/R, in that blood flow is greatly reduced and then restored. During arousal from torpor, hibernators rewarm using a combination of shivering and nonshivering thermogenesis associated with peaks in oxygen consumption, while accelerating their heart rate 100-fold with corresponding increases in organ perfusion.³  Hibernating mammals like arctic ground squirrels (AGS) show natural resistance to experimental I/R in multiple organs,^4–7  which remains yet uncharacterized for the heart. Moreover, although a suite of adaptations associated with the seasonal entry into torpor have been reported, including metabolic rate depression,⁸  increased reliance on lipid oxidation for energy production,^3,9–11  and immune suppression,¹²  the mechanisms underlying this naturally evolved organ protection remain unknown.

In the current study, we hypothesized that experimental surgical I/R will result in reduced severity of myocardial injury and dysfunction in hibernating AGS compared to a nonhibernator (rat); furthermore, we postulated that such hibernator cardioprotective phenotype will be attenuated in summer active (SA) AGS. As a first step to delineate the metabolic regulatory events involved, we employed an unbiased proteomics approach to identify and quantify differences in myocardial protein abundance between AGS and rats and further correlate with targeted metabolic changes. The potential cardioprotective role of sirtuin-3 up-regulation identified in hibernating AGS hearts was further assessed in vitro using gain-of-function (rat) and loss-of-function (AGS) experiments in adult primary ventricular cardiomyocyte (APVC) models of hypoxia–reoxygenation.

Adult male Brown Norway and Dahl/Salt Sensitive rat strains (Charles River, USA) were randomly assigned to three experimental groups: anesthetized and cannulated without undergoing cardiopulmonary bypass (CPB; sham group) and deep hypothermic circulatory arrest (DHCA; 45 min at 18°C), followed by either 3 h (DHCA R3h group) or 24 h (DHCA 24h group) of reperfusion (fig. 1).

Arctic ground squirrels (Urocitellus parryii) were trapped in July near the Toolik Field Station in northern Alaska and transported to the University of Alaska Fairbanks (Fairbanks, Alaska). In the fall, animals assigned to the hibernating group were housed in temperature- and photoperiod-controlled facilities (5°C, photoperiod, 4L:20D), after being implanted with intraperitoneal temperature-sensitive radio transmitters to monitor their body temperature and precisely determine stages of torpor and arousal, as described in the study by Yan et al.¹³  Preoperatively, torpid animals (hibernating AGS) were moved to a warm, lighted room and aroused to euthermia before each experiment. Winter experiments were conducted in January to February 2012 and 2013. Postreproductive summer euthermic animals were used as nonhibernating controls, with experiments conducted in August to September 2012 and 2013. Summer animals were housed for at least 1 month to complete quarantine and infection and parasite testing. Summer animals had normal room-temperature housing and light at least 12 h/d. AGS included in these experimental groups were composed of n = 16 males and n = 7 females. As these animals are wild caught, exact age cannot be determined; however, juvenile and aged animals were excluded.

Rats and AGS underwent an identical experimental protocol, as previously described.^14,15  Briefly, animals were anesthetized with isoflurane, endotracheally intubated and mechanically ventilated; anesthesia was maintained with isoflurane, fentanyl, and pancuronium. Surgical preparation involved cannulation of the ventral tail artery (rats) and right femoral artery (AGS; 20 gauge, arterial inflow), insertion of a multiorifice dual-stage cannula through the right internal jugular vein and advanced into the right atrium (4.5 French, venous outflow), and insertion of a 3.5-mm balloon catheter (Sprinter OTW, Medtronic, USA) via right carotid artery cutdown, positioned above the aortic valve under echocardiographic guidance. Electrocardiogram, pulse oximetry, end-tidal capnography, pericranial and rectal temperatures, and invasive arterial blood pressure were continuously recorded. Arterial blood gases were monitored throughout the experimental protocol (GEM Premier 3000, Instrumentation Lab, USA). After systemic heparinization, CPB was initiated and animals were cooled to 18°C. Cardioplegic arrest was achieved by endoaortic balloon inflation and antegrade administration of blood cardioplegia, with electromechanical arrest confirmed electrocardiographically and echocardiographically. DHCA was initiated for 45 min, at the conclusion of which the endoaortic balloon was deflated and CPB resumed. Animals were rewarmed to normothermia (1 h), weaned from CPB, and recovered until the time of euthanasia, for either 3 (R3h) or 24 h (R24h) postreperfusion. CPB, ischemia, and rewarming times are identical between R3h and R24h groups, with only duration of reperfusion being different. All experiments were approved by the Institutional Animal Care and Use Committees of Duke University and University of Alaska Fairbanks, and compliant with the Guide for the Care and Use of Laboratory Animals.

Phenotypic characterization of myocardial injury severity was performed using biochemical, echocardiographic, and histologic measurements at baseline and at the 3- and 24-h reperfusion time points. Levels of cardiac troponin I were measured at each time point in EDTA plasma using a multiplex immunoassay system (Meso Scale Discovery, USA). Serial transthoracic echocardiographic measurements of left ventricular (LV) systolic function were conducted using a Vivid i ultrasound system (GE Healthcare, USA) equipped with an M12L transducer and analyzed using Xcelera R3.3 software (Phillips, USA). LV fractional area change (LV-FAC) was calculated as (LV end diastolic area − LV end systolic area)/LV end diastolic area × 100. Percent change in LV-FAC at each reperfusion time point compared to baseline is presented for all groups. Myocardial apoptosis was assessed in LV tissue sections by terminal deoxynucleotidyl transferase dUTP nick-end labeling assays (Roche Diagnostics, USA) and double immunofluorescence staining for activated caspase-3 and troponin I, as well as by determining the levels of cleaved caspase 3 in myocardial tissue (Western blot).

At the time of euthanasia (3- or 24-h reperfusion), animals were completely exsanguinated via the jugular venous cannula and perfused with 60 ml warm phosphate-buffered saline. The heart was rapidly removed, divided using a rat heart slicer matrix into axial sections for histology, and snap frozen in liquid nitrogen. Details of sample preparation protocols for proteomic and metabolic profiling are presented in Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269.

We employed a five-fraction two-dimensional liquid chromatography, tandem mass spectrometry (2D-LC/LC-MS/MS) method to identify and quantitate proteins expressed in LV myocardium of AGS (hibernating SA) and Brown Norway rats subjected to DHCA followed by 3 or 24 h of reperfusion or in the sham group. Proteins extracted from LV homogenates were normalized, reduced, cysteine alkylated, and trypsin digested, followed by label-free quantitative 2D-LC/LC-MS/MS. Robust peak detection and label-free alignment of individual peptides across all sample injections were performed using Rosetta Elucidator v3.3 (Rosetta Biosoftware, USA) with PeakTeller algorithm. MS/MS data were searched against the NCBI RefSeq Rattus (for rat) and against a curated AGS database representing refined thirteen-lined ground squirrel (Ictidomys tridecemlineatus) and human protein sequences from Ensembl release 69, as described.¹⁰  After individual peptide scoring using the Peptide Prophet algorithm, the data were annotated at a less than 1% peptide false discovery rate (FDR). Relative peptide and protein abundance were calculated using Rosetta Elucidator. Details for mass spectrometry data collection, processing, and quality control are presented in Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269.

A key component of our analytical strategy to compare changes in myocardial protein expression across species involves the establishment of orthologous relationships among the proteins identified and quantified using label-free proteomic approaches in rat versus AGS. Commonly used methods for orthology assessment typically use Basic Local Alignment Search Tool to search for pairs of reciprocal best hits in the genomes (or proteomes) being compared. However, these methods are easily misled when additional gene duplications occur after the species in question have diverged (“in-paralogs”), leading to the presence of co-orthologs in cross-genome comparisons. We have defined sets of putatively orthologous proteins among those identified within each species by five-fraction 2D-LC/LC-MS/MS proteomics. These orthologies were established by adapting the EnsemblCompara GeneTrees resource developed by Vilella et al.,¹⁶  a system that automates the clustering, multiple alignment, and phylogenetic analysis necessary to build gene trees for large numbers of gene families represented in the Ensembl project. Given the known one-to-one, one-to-many, and many-to-many orthology relationships, we used the BioMart data-mining tool and Ensembl Genes Release 73¹⁷  (Rnor_5.0 for rat, Spetri2 for squirrel) to conduct a bidirectional orthology matching using first the rat as the reference genome and then the AGS as the reference genome. The resulting orthology map—containing 702 unique rat GenInfo Identifier protein IDs, 700 unique squirrel Ensembl Protein IDs, and 697 unique gene names—was used for subsequent cross-species comparative analyses of myocardial protein abundance. Details of the computational approach to establish the orthology map are presented in Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269.

Total DNA (mitochondrial and nuclear) was extracted from rat and AGS LV myocardium samples using a QIAamp DNA Mini Kit (Qiagen, USA). The content of mtDNA was calculated using quantitative real-time polymerase chain reaction (PCR) by measuring the threshold cycle ratio (ΔC_T) of a mitochondrial-encoded gene (Cytochrome c oxidase, subunit I, Mt-Co1) versus a nuclear-encoded gene (Cyclophilin A, Ppia) in a StepOnePlus reverse transcription (RT)-PCR System (Life Technologies USA) equipped with StepOne v2.2.3 software (see Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269, for details on primers used and RT-PCR protocol).

To validate mass spectrometry findings, Western blot imaging analysis was conducted for specific electron transport chain (ETC) proteins (NADH dehydrogenase subunit [NDUF] S8, succinate dehydrogenase complex [SDH] A, ubiquinol cytochrome c reductase [UQCR] Q, cytochrome c oxidase [COX] 5A, and adenosine triphosphate [ATP] 5A) and sirtuin-3, and all results were normalized using ratios to the constitutively expressed protein COXIV, HSP60, or β-actin depending on protein size (see Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269, for details).

To characterize the functional relevance of differences in myocardial sirtuin-3 abundance identified by proteomic analysis, we conducted additional in vitro sirtuin-3 gain- and loss-of-function studies in APVCs isolated from rat and AGS,¹⁸  respectively, and assessed the effects on cell viability after simulated I/R. Rat APVCs were treated with increasing concentrations of resveratrol (Sigma-Aldrich, USA), an activator of sirtuin-3, or nicotinamide, an inhibitor of sirtuin-3 (control, 5 μM, and 10 μM) for 1 h. AGS APVCs were transfected with sirtuin-3–siRNA (60 pmol) or scrambled RNA control (Santa Cruz, USA) overnight, according to manufacturer’s protocols. Cells were subjected to 2 h of oxygen and glucose deprivation followed by 24 h of reoxygenation and determination of apoptosis (Cell Death Detection ELISA kit, Roche) and necrosis (cardiac troponin I levels in supernatant, Life Diagnostics, USA). Sirtuin-3 expression (Western blot) and activity (fluorometric sirtuin-3 activity assay kit, Abcam, Cambridge, MA) were measured in APVC lysates. All experiments were run in triplicate.

The nuclear fraction was isolated from cryopreserved myocardial samples using a commercial nuclear extraction kit (Panomics, USA) according to manufacturer’s specifications and assayed for peroxisome proliferator–activated receptor (PPAR)-α nuclear receptor activity using ELISA (Affymetrix, USA; see Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269, for details).

Cryopreserved myocardial samples were homogenized in 50% acetonitrile supplemented with 0.3% formic acid and used to determine levels of acylcarnitines, organic acids, amino acids, and ceramides via stable isotope dilution techniques, as described previously.¹⁹  Details of the metabolomic analysis are provided in Supplemental Methods, Supplemental Digital Content 1, https://links.lww.com/ALN/B269.

Experimental sample sizes are based on previous reports of differences in protein abundance in hibernators,^3,10  comparative responses to I/R across species,²⁰  as well as efforts to conserve scant resources (wild-caught AGS).

Principal component analysis (PCA) was performed to explore sources of variation in relative protein abundance across the experimental data set, visualize whether biologic replicate samples resemble each other, and identify any outliers. PCA uses orthogonal transformation to convert relative protein abundance into a set of linearly uncorrelated principal components.²¹  The top three principal components are presented as 3D scatter plots in Supplemental Results, Supplemental Digital Content 1, https://links.lww.com/ALN/B269, with each point representing the proteomic data from one sample and samples colored by species and hibernation state. Differentially expressed proteins in this multifactor experiment were identified using mixed model ANOVA, which included species, hibernation state (winter hibernating, SA), reperfusion time (sham, reperfusion 3 h, and reperfusion 24 h), and hibernation state × reperfusion time interaction (to test whether seasonal hibernation states would result in temporal differences in myocardial protein abundance postreperfusion). Differentially expressed proteins were defined based on a fold-change more than or equal to 1.5 or less than or equal to −1.5 and a Benjamini–Hochberg FDR-adjusted P < 0.05. For this exploratory analysis, we chose FDR to adjust for multiple comparisons over the more conservative familywise error rate approach (e.g., Bonferroni correction), which would likely miss a number of potentially novel differences.²²  Exploratory PCA analysis and differential protein expression analysis were conducted using Partek Genomics Suite v6.6 (Partek, USA).

Differences in perioperative myocardial injury endpoints (biochemical, echocardiographic), as well as differences in abundance of individual proteins or metabolites between hibernating AGS, SA AGS, and rats, or between time points were assessed by one-way ANOVA with Tukey post hoc test for pairwise comparisons. Two-way ANOVA was further used to test whether differences in individual protein abundance between species and reperfusion time points were more than expected by chance. Mann–Whitney U test was used for two group comparisons of apoptosis and necrosis endpoints. Statistical analyses were conducted using GraphPad Prism v6 (GraphPad Software, Inc, USA).

Compared to rats, AGS exhibited significantly reduced myocardial injury at both reperfusion time points (mean ± SD of plasma troponin I concentrations after 3 h of reperfusion were 0.9 ± 0.2, 2.5 ± 2.1, 4.8 ± 0.8, and 6.7 ± 1.3 and after 24 h of reperfusion were 0.2 ± 0.1, 0.1 ± 0.1, 2.1 ± 0.9, and 6.3 ± 3.3 for hibernating AGS, SA AGS, Brown Norway rats, and Dahl-Salt Sensitive rats, respectively; fig. 2A). Reflecting known differences in susceptibility to myocardial I/R between the rat strains used as nonhibernator experimental model organisms, severity of myocardial injury in hibernating AGS was approximately 5-fold lower compared to Brown Norway rats (a strain characterized by resistance against myocardial I/R) and approximately 8-fold lower than in Dahl/salt-sensitive rats (a strain more susceptible to myocardial I/R). Hibernating AGS displayed the most robust cardioprotection, with SA AGS demonstrating an intermediate level of protection (fig. 2A). The biochemical results were corroborated echocardiographically, with hibernating AGS showing preserved LV systolic function after I/R, whereas LV-FAC declined in both SA AGS and rats (mean ± SD of percent changes in LV-FAC compared to preoperative baseline after 3 h of reperfusion were 7 ± 6, −16 ± 14, and −25 ± 3 and after 24 h of reperfusion were 4 ± 14, −17 ± 32, and −15 ± 4 for hibernating AGS, SA AGS, and rats, respectively; fig. 2B). Myocardial apoptosis was attenuated in hibernating AGS compared to rats after I/R, as evidenced by lower fraction of terminal deoxynucleotidyl transferase dUTP nick-end labeling–positive nuclei (fig. 2C) and reduced tissue levels of active caspase-3 (fig. 2D).

After quality control, the final quantitative data set for rat consisted of 8,482 peptides/1,320 proteins (786 proteins contained at least 2 unique peptides) and for AGS, 9,292 peptides/1,478 proteins (806 proteins contained at least 2 unique peptides). Differential expression analysis across species was conducted based on an orthology map consisting of 697 proteins. PCA was performed to identify high-level differences between sample groups (fig. S3, Supplemental Digital Content 1, https://links.lww.com/ALN/B269). Signal versus noise across all orthologous proteins for each of the factors and interactions in the ANOVA model identified that all factors contributed significant variation to the data across the set, with species (rat vs. AGS) representing the dominant source of variation (fig. S4, Supplemental Digital Content 1, https://links.lww.com/ALN/B269).

A total of 545 proteins were differentially expressed between hibernating AGS and rat based on a fold change more than or equal to 1.5 or less than or equal to −1.5 and an FDR-adjusted P < 0.05, with 255 up-regulated and 290 down-regulated proteins (fig. S5, Supplemental Digital Content 1, https://links.lww.com/ALN/B269). Of these, 191 proteins had more than a ±6-fold change between hibernating AGS and rat, and 498 had a Bonferroni-adjusted P < 0.05. By using pathway analysis, these proteins were mapped onto 77 canonical pathways. Top canonical pathways differentially expressed between hibernating and rat included oxidative phosphorylation (p = 1.1e-20), protein ubiquitination (1.4e-13), tricarboxylic acid (TCA) cycle (6e-9), branched chain amino acid metabolism (7.3e-7), and cytoskeletal remodeling (7.8e-7).

The most prominent proteomic finding is a general down-regulation of ETC proteins in the hibernator hearts (table 1). Significant reductions in levels of subunits of all 5 ETC complexes were found in hibernating AGS compared to rats, with only complex V showing increased expression of some subunits after I/R. Complex I subunits NADH-ubiquinone oxidoreductase (NDUFs) 1 to 13 were reduced 1.14- to 17.68-fold. Complex II subunits SDHA, SDHB, and SDHC declined 2.47- to 5.1-fold. Complex III subunits UQCRQ, UQCRFS1, UQCRC2, and UQCR10 declined 2.04- to 3.57-fold. Complex IV subunit COX5A declined 1.23-fold. Complex V subunits ATP synthase mitochondrial F₁/F₀ complex (ATP) AF2, ATP5F1, and ATP5B declined 1.75- to 19.11-fold; however, subunits ATP5A1-ATP5J2 increased in abundance 1.41- to 15.94-fold (table 1).

To ensure that lower quantities of ETC proteins detected in AGS were not secondary to lower mitochondrial density in the myocardium, mtDNA quantity was calculated as the ratio of COX1 to cyclophillin A DNA levels determined by RT-PCR and found to be higher for AGS compared to rat (fig. S10, Supplemental Digital Content 1, https://links.lww.com/ALN/B269), consistent with an increase in mitochondrial biogenesis in the hibernator heart. A decrease in mtDNA copy number, however, was seen in both species under the R3h condition compared to sham, coincident with peak reperfusion injury (fig. S10, Supplemental Digital Content 1, https://links.lww.com/ALN/B269).

Western blot analyses confirmed mass spectrometry findings in select ETC proteins of interest. Down-regulation of complex II protein SDHA in hibernating AGS and SA AGS compared with rat at all time points is shown in figure 3. Results for complex I protein NDUFS8, complex III protein UQCRQ, complex IV protein COX5A, and complex V protein ATP5A are shown in figure S11 and table S7, Supplemental Digital Content 1, https://links.lww.com/ALN/B269.

Sirtuin-3, the major mitochondrial protein deacetylase with multiple cardioprotective targets including ETC proteins, is significantly up-regulated in both hibernating and SA AGS compared to rat by proteomic profiling (fig. 4A). Species differences also exist in the dynamic expression changes after I/R—unlike rats, AGS recover their baseline sirtuin-3 levels by R24h (fig. 4A); Western blot analysis shows a sirtuin-3 expression pattern consistent with the proteomic findings (fig. 4B). We also corroborated the sirtuin-3 result with changes in target proteins known to be regulated by sirtuin-3, including an up-regulation in AGS hearts of manganese superoxide dismutase (total MnSOD)²³  accompanied by posttranslational deacetylation of MnSOD²⁴  (fig. 4B), known to increase its ROS scavenging activity.²⁵ 

To further explore the potential mechanistic links between sirtuin-3 expression, activity, and the hibernator cardioprotective phenotype, we conducted comparative gain- and loss-of-function experiments in APVCs cell models of hypoxia–reoxygenation. In rat APVCs, resveratrol increased sirtuin-3 expression and activity at both 5 and 10 μM doses (fig. 5, A and B), which was associated with reduced apoptotic and necrotic cell death (fig. 5, C and D). Conversely, nicotinamide reduced sirtuin-3 activity at both 5 and 10 μM (fig. 5B) and was associated with increased cardiomyocyte death (fig. 5, C and D). On the other hand, siRNA sirtuin-3 silencing in AGS cardiomyocytes resulted in significant increases of apoptotic and necrotic death compared to scrambled RNA–treated cells (fig. 5, E and F). Furthermore, the observed increase in AGS cardiomyocyte viability after hypoxia–reoxygenation stress (fig. S12, Supplemental Digital Content 1, https://links.lww.com/ALN/B269) is accompanied by reduced acetylation of mitochondrial proteins compared to the rat (fig. S13, Supplemental Digital Content 1, https://links.lww.com/ALN/B269).

The proteomic analyses revealed a sharp contrast between a general down-regulation of key metabolic enzymes associated with glycolysis, TCA cycle, ketolysis, amino acid catabolism, and ETC, whereas a significant proportion of proteins involved in fatty acid uptake, transport, and catabolism were up-regulated in the hibernator hearts (table 2). Accordingly, we used mass spectrometry–based targeted metabolomic profiling to provide additional insight into the energy metabolic changes occurring in the hibernator heart during I/R stress. The most striking distinguishing metabolite signature in the AGS compared to rat myocardium was exhibited by the acylcarnitine profile. Accummulation of toxic intermediates of lipid metabolism (acylcarnitines and ceramides) was found in LV myocardial tissue homogenates from rat compared to hibernating AGS at both reperfusion time points (fig. 6).

While the prime regulatory mechanism underlying the hibernator phenotype remains unclear, key molecules known to play a role in lipid metabolism and energy flux are significantly different between rat and hibernating AGS in our analyses. We, therefore, conducted comparative analyses of PPAR-α activity, a master regulator of mitochondrial fatty acid utilization, in the hearts of rats and hibernating AGS in response to I/R. While PPAR-α activity levels significantly declined with I/R in rats, hibernating AGS were able to maintain baseline levels of PPAR-α activity despite I/R and may be involved in the up-regulation of fatty acid oxidation enzymes observed in AGS through its transactivation effects (fig. 7).

Here, we report for the first time a cross-species analysis demonstrating cardioprotection in hibernating AGS compared with summer AGS or rat. After a clinically relevant I/R model and employing robust endpoint measures, hibernating AGS displayed significantly attenuated myocardial injury, as assessed by plasma levels of troponin I, myocardial apoptosis, and preservation of LV contractile function. Our comparative analyses employed two different strains of rat with known differences in susceptibility to I/R (Brown Norway and Dahl Salt Sensitive) and revealed a further significant reduction in susceptibility in the hibernator heart. Moreover, we were able to demonstrate a seasonal difference in responses to I/R in AGS, with increased protection in hibernating AGS. While AGS as a species seem to be resistant to I/R, many of the adaptations that allow maximum protection appear to maximize in the winter.²⁶  Our experimental model of I/R injury employs standardized temperature (18°C), anesthetic, and cannulation strategies between rat, summer AGS, and hibernating AGS, and therefore, the observed cardioprotection in hibernating AGS can be ascribed to adaptations that accompany the hibernator state, not to protective effects of hypothermia.

We undertook an unbiased approach to identify changes in metabolic enzymes and metabolite pools between species and hibernation states in response to I/R stress that may shed light onto the mechanisms underlying cardioprotection in hibernators. The most striking difference is the profound decline in expression of ETC proteins in hibernating AGS compared to rat, a pattern also associated with the most significant cardioprotection of all three groups. Recent work by an independent laboratory has characterized proteomic differences in adaptation to myocardial ischemia among nonhibernating mammals. Cabrera et al.²⁷  examined the proteomic signatures of myocardial hibernation and ischemic preconditioning in swine. Hibernating myocardium, an adaptation to chronic ischemia in which the myocardium decreases its oxygen consumption by adopting a quiescent state²⁸  (and a confusing term for hibernation biologists), was found to have decreased quantities of all five ETC complex proteins, including NADH-ubiquinone oxidoreductase (NDUFs), SDH, UQCR, COX, and ATP synthase mitochondrial F₁/F₀ complex subunits. This quiescent state allows for the preservation of viable myocardium that can regain function if blood supply is restored.²⁹  Ironically, our data demonstrate that the cardioprotective strategy adopted by hibernating mammals in response to I/R closely approximates the proteomic features of hibernating myocardium in nonhibernators, suggesting that at some level the regulatory components of the hibernator cardioprotective phenotype may elicit a conserved response to chronic ischemia in all mammals. However, hibernating AGS hearts also down-regulate the molecular machinery involved in glycolysis, TCA cycle, and the metabolism of ketones and amino acids. This is in contrast to what is known about hibernating myocardium in mammals that do not seasonally hibernate such as rats or humans, which is more reliant on glycolysis.³⁰  Our comparative analyses revealed that rats suffer more myocardial injury and have a proteomic profile characterized by maintenance of ETC protein abundance. This more closely approximates the adaptations seen in ischemic preconditioning. This occurs when the myocardium is exposed to brief nonlethal periods of ischemia and becomes resistant to further ischemic insults,^31–33  yet it does not significantly reduce ETC proteins.²⁷  In our model, rats demonstrate inferior cardioprotection to AGS associated with a pattern of maintained expression of ETC proteins and increased glycolysis, ketolysis, and amino acid metabolism. Consistent with these findings, reversible pharmacologic blockade of electron transport at the onset of reperfusion was recently reported to decrease cardiac injury in aged hearts by improving the inner mitochondrial membrane potential.³⁴ 

The cardioprotective adaptations in hibernators were associated with robust up-regulation of sirtuin-3 in AGS hearts, the main mitochondrial protein deacetylase, and key regulator of metabolic (including fatty acid oxidation) stress-response pathways in the heart.³⁵  Increasing evidence also suggests sirtuin involvement in regulating the hypometabolic states associated with hibernation. During entrance into torpor, reductions in metabolism precede reductions in body temperature (even when thermogenesis is not active) suggesting active mechanisms regulating metabolic suppression rather than passive thermal effects.^8,36  Mitochondrial respiration decreases quickly during entrance into torpor when body temperature is high (e.g., by 70% between 37°C during interbout arousal and 30°C during torpor entrance).³⁶  This occurs faster than transcriptional or translational changes; moreover, peptide elongation in hibernators ceases below 18°C,³⁷  yet in early arousal (when body temperature is much lower), mitochondrial respiration increases significantly. This pattern suggests that body temperature–sensitive, enzyme-mediated posttranslational modifications of oxidative phosphorylation complexes like differential acetylation could be involved. Although the heart accounts for less than 0.5% of basal metabolic rate, cardiac mitochondria from AGS display 60% reductions in respiration rates during torpor.³⁸  Inhibition of succinate-fueled state 3 mitochondrial respiration has been observed in three different tissues (heart, skeletal muscle, and liver), appearing to be a hallmark of hibernation in AGS. The underlying mechanisms remain unclear, but do not appear to involve inhibitions of either succinate oxidation (SDH activity)³⁹  or transport (dicarboxylate transporter). Nonetheless, the report of sirtuin-3–mediated deacetylation of SDHA regulating ETC complex II activity⁴⁰  is of particular relevance to hibernators, given the profound suppression of succinate oxidation during torpor. Although we did not specifically assess acetylation of ETC complexes in our study, increased myocardial expression of sirtuin-3 in AGS was associated with activation of key cardioprotective targets like MnSOD and consequently reduced oxidative stress. We are providing a preliminary characterization of sirtuin-3 involvement in the hibernator cardioprotective phenotype, by reporting in cardiomyocyte models of I/R the association between sirtuin-3 gain- and loss-of-function and changes in cell survival. Furthermore, consistent with the differences in sirtuin-3 expression, rat cardiomyocytes had hyperacetylated mitochondrial proteins compared to AGS, which was further increased after I/R.

A precisely controlled fuel shift from myocardial carbohydrate to fatty acid metabolism in hibernating AGS appears to be orchestrated through down-regulation of key metabolic enzymes associated with glycolysis, TCA cycle, ketolysis, and branched-chain amino acid catabolism, while increasing enzymes involved in fatty acid catabolism. This represents something of a contradiction to classic teaching for myocardial substrate utilization during ischemic stress, when the heart’s ability to process a wide variety of substrates as fuel would appear to be an advantage; moreover, oxidation of carbohydrate, ketone, and amino acid fuel sources requires less oxygen than the catabolism of fatty acids.⁴¹  However, increased expression of enzymes involved in fatty acid catabolism may help explain why hibernating AGS are able to catabolize fats without suffering ill effects, even with the addition of I/R injury. We show that increased expression levels of fatty acid oxidation enzymes is associated with preserved activity of the PPAR-α transcription factor after I/R in AGS but not in rats. The mechanism of increased myocardial injury in nonhibernators exposed to high fat may in part be due to incomplete oxidation of fatty acids. Accumulation of partially oxidized lipids and sphingolipids can give rise to lipotoxicity as these reactive intermediates accumulate in the mitochondria of organisms unable to metabolize a large fat load.⁴²  To investigate the role of lipotoxicity in our model, we used targeted metabolomic profiling to measure levels of partially oxidized lipid intermediates previously implicated as deleterious in I/R injury, like dicarboxylacylcarnitines and ceramides.^43–45  Compared to rats, winter AGS have significantly lower myocardial levels of C4-dicarboxylacylcarnitine and C-18 ceramide (fig. 6). During early reperfusion (R3h), rats experience a rise in C4-dicarboxylacylcarnitines, indicative of mitochondrial dysfunction, whereas winter AGS maintain their baseline levels.

In an attempt to employ an unbiased analysis of differences in myocardial proteins between hibernators versus nonhibernators, we have utilized a label-free proteomic profiling method. This technique has the limitation of detecting high abundance proteins over rare ones. One shortcoming of our study is that regulatory proteins may not be adequately represented. Also, in this experiment, we have focused on the identification and abundance of proteins as a first step, and many regulatory steps are undoubtedly controlled by posttranslational modifications including phosphorylation, acetylation, neddylation, modification with small ubiquitin-like modifier, and ubiquitinylation. These modifications may exert effects by modifying protein function or simply marking a protein for degradation. Oxidation and partial degradation of protein might also not be adequately detected by our technique. While we have been able to demonstrate quantitative changes in ETC proteins in whole myocardial tissue, we did not perform the analysis in isolated mitochondria and thus cannot speak to changes in protein abundance within the mitochondria specifically. Our data relate to proteomic changes within a specific quantity of myocardial protein. We did, however, demonstrate that the ratio of mitochondrial to nuclear DNA is in fact higher in AGS than rat, thus eliminating the possibility that the decline in ETC proteins is secondary to falling numbers of mitochondria.

In summary, we present the first comparative proteomics study of myocardial protein expression changes after experimental I/R in hibernating versus nonhibernating mammals. We show that several prominent features of myocardial I/R injury like myocardial necrosis, apoptosis, and mechanical stunning were significantly reduced in the hibernator heart and accompanied by differential expression of proteins mainly involved in regulation of mitochondrial fuel and energy metabolism. Our data support the idea that mammals have a dichotomous response to I/R injury, one pathway leading toward the development of hibernating myocardium and the other to ischemic preconditioning. Ironically, hibernating animals exhibit a seasonal cardioprotective phenotype that resembles hibernating myocardium with respect to the decline of ETC protein abundance. In contrast, rats maintain levels of ETC proteins consistent with what has been observed in models of ischemic preconditioning. Enhanced lipid metabolism possibly regulated by sirtuin-3, PPARs, and other master regulators of metabolism may be key to the metabolic reprogramming associated with cardioprotection in hibernators. Further understanding this unique model of extreme metabolic plasticity and developing strategies to “switch” myocardial metabolism to resemble that naturally occurring in mammalian hibernators represent a transformative approach that could ultimately have a positive impact in patients undergoing cardiac surgery and transplantation and victims of cardiac arrest, trauma, and hypothermia, in addition to fundamentally advancing cardiovascular biology.

The authors thank Jeanette Moore, Ph.D., Franziska Kohl, M.Sc., and Lori Bogren, Ph.D. (University of Alaska Fairbanks, Fairbanks, Alaska), for technical assistance with deep hypothermic circulatory arrest (DHCA) experiments; Jun Yan, Ph.D. (Chinese Academy of Sciences-German Max Planck Society, Shanghai, China), for help in building phylogenetic orthology maps for cross-species analyses; and G. Burkhard Mackensen, M.D., Ph.D. (Department of Anesthesiology, University of Washington, Seattle, Washington), for his role in developing the DHCA model in rodents.

This study was supported by the Foundation for Anesthesia Education and Research (Schaumburg, Illinois) Research Fellowship Grant, Duke University Department of Anesthesiology, primary investigator (PI): Dr. Quinones, mentor: Dr. Podgoreanu; National Institute of General Medical Sciences (Bethesda, Maryland) T32 GM08600, Duke University Medical Center, fellow: Dr. Quinones; National Institutes of Health (Bethesda, Maryland) R01-HL092071, Duke Innovation Grant Award, Duke University Medical Center, PI: Dr. Podgoreanu; and American Heart Association (Dallas, Texas) 11BGIN7620055, Beginning Grant in Aid, Duke University Medical Center, PI: Dr. Zhang.

The authors declare no competing interests.