Regulatory, Integrative and Comparative Physiology

Seasonal proteomic changes reveal molecular adaptations to preserve and replenish liver proteins during ground squirrel hibernation

L. Elaine Epperson, James C. Rose, Hannah V. Carey, Sandra L. Martin


Hibernators are unique among mammals in their ability to survive extended periods of time with core body temperatures near freezing and with dramatically reduced heart, respiratory, and metabolic rates in a state known as torpor. To gain insight into the molecular events underlying this remarkable physiological phenotype, we applied a proteomic screening approach to identify liver proteins that differ between the summer active (SA) and the entrance (Ent) phase of winter hibernation in 13-lined ground squirrels. The relative abundance of 1,600 protein spots separated on two-dimensional gels was quantitatively determined using fluorescence difference gel electrophoresis, and 74 unique proteins exhibiting significant differences between the two states were identified using liquid chromatography followed by tandem mass spectrometry (LC-MS/MS). Proteins elevated in Ent hibernators included liver fatty acid-binding protein, fatty acid transporter, and 3-hydroxy-3-methylglutaryl-CoA synthase, which support the known metabolic fuel switch to lipid and ketone body utilization in winter. Several proteins involved in protein stability and protein folding were also elevated in the Ent phase, consistent with previous findings. In contrast to transcript screening results, there was a surprising increase in the abundance of proteins involved in protein synthesis during Ent hibernation, including several initiation and elongation factors. This finding, coupled with decreased abundance of numerous proteins involved in amino acid and nitrogen metabolism, supports the intriguing hypothesis that the mechanism of protein preservation and resynthesis is used by hibernating ground squirrels to help avoid nitrogen toxicity and ensure preservation of essential amino acids throughout the long winter fast.

  • Ictidomys tridecemlineatus
  • metabolism
  • starvation
  • tricarboxylic acid cycle

hibernating mammals exhibit extreme plasticity in many basic physiological parameters, including metabolic, heart, and respiratory rates and core body temperature (Tb). In winter, ground squirrels and other small circannual hibernators become heterotherms; in summer, they behave as more typical mammalian homeotherms (Fig. 1). The heterothermic pattern of winter is characterized by oscillations between multiday periods of torpor and short (<1-day) periods of euthermy, known as interbout arousal. In torpor, Tb is maintained at ∼0°C, and heart, respiratory, and metabolic rates, as well as DNA, RNA, and protein synthesis rates, are also dramatically reduced. All these physiological and biochemical parameters return to or exceed typical summertime euthermic values during each interbout arousal (for review see Ref. 6).

Fig. 1.

Seasonal core body temperature (Tb) transitions in ground squirrels. Graph of Tb over time shows summer homeothermy and winter heterothermy (hibernation). Summer active (SA) animals maintain Tb at 37 ± 1°C until late in the fall. Hibernation consists of multiday periods of torpor, during which Tb is near ambient, 4°C, punctuated by short interbout arousals to ∼37°C. Animals at the entrance (Ent) phase of winter hibernation were reentering torpor after a spontaneous interbout arousal.

Hibernation is a complex phenotype that must be orchestrated from the hibernator's genotype. In an earlier study (25), we hypothesized that the hibernating phenotype is achieved by reprogramming of expression of genes common to mammals, rather than the expression of novel genes. This hypothesis was based on the broad phylogenetic distribution of hibernating species among mammals and a limited assessment of differential gene expression in the liver of golden-mantled ground squirrels (25). Data from subsequent screens support this hypothesis and provide evidence for a limited reprogramming of liver gene expression in winter, which is evident at mRNA (9, 32, 33) and protein (8) levels. Results obtained using other tissues and hibernating species are generally consistent with those from the liver in golden-mantled ground squirrels (5, 8, 19, 20, 23, 3234).

It is noteworthy that the majority of gene products do not change as a function of hibernation status in any tissue. Those that do change appear to be predominantly seasonal; that is, the steady-state level of a given mRNA or protein is typically elevated or reduced in summer animals compared with all winter-stage animals and does not vary between interbout arousal and torpor. Transcription and translation virtually cease during torpor but are reactivated during each interbout arousal (see Ref. 27 and references therein; also see Ref. 28). Consistent with this observation, during the torpor phase of hibernation, there is some evidence for loss of specific mRNAs (2, 9, 33) and proteins (8, 18), which are restored during interbout arousal.

Here we apply a fluorescence-based two-dimensional (2-D) difference gel electrophoresis (DiGE) approach to quantitatively assess protein differences between summer active (SA) 13-lined ground squirrels and those at the entrance (Ent) phase of winter hibernation (Fig. 1). Differentially expressed or modified proteins between the two states are then identified using liquid chromatography followed by tandem mass spectroscopy (LC-MS/MS). Our choice of sampling is based on the rationale that the proteins necessary to orchestrate and withstand torpor must be fully restored as the hibernator reenters this state following an interbout arousal but are not needed in summer, when the animals are homeothermic and do not use torpor. Similarly, proteins that are critical in summer but not needed in winter may be decreased during hibernation. This work extends the results of a similar analysis of liver proteins that differ between SA and Ent in golden-mantled ground squirrels (8) and provides the first opportunity to compare the effect of hibernation season on the liver proteome of two relatively distantly related (13) sciurid rodent hibernators: 13-lined (Ictidomys tridecemlineatus) and golden-mantled (Callospermophilus lateralis) (12) ground squirrels.



The use of all animals was approved by the University of Wisconsin Institutional Animal Care and Use Committee. Ground squirrels were trapped in the wild and then housed individually with free access to water and food (Purina rodent chow 5001, supplemented with sunflower seeds) at 22°C with a 12:12-h light-dark cycle. Six SA squirrels were killed in August, 1 mo after they were trapped. In August or September, six additional ground squirrels were implanted with radiotelemeters (Minimitter) for Tb monitoring and allowed to recover for ≥3 wk. Squirrels were then transferred to a cold room maintained at 4°C; all exhibited torpor within 3 days, at which time food and water were removed. Ent animals were killed 1–4 mo after the first bout of torpor at Tb between 27 and 21°C, as they reentered torpor following a spontaneous interbout arousal to 37°C.

Protein extracts.

The liver was frozen in liquid N2 immediately after dissection, shipped on dry ice, and stored at −80°C. Protein extracts were prepared as described previously (8); briefly, 100 mg of tissue were homogenized in 100 μl of buffer [0.5 M sucrose, 100 mM potassium phosphate (pH 6.7), 5 mM MgCl2, 1 mM phenylmethylsulfonyl fluoride, and 10 μg/ml protease inhibitor cocktail (catalog no. P8340, Sigma)] with a Polytron (Brinkmann Instruments). The homogenate was centrifuged at 4°C for 10 min at 500 g. The postnuclear supernatant was recovered, frozen in 15-μl aliquots using liquid N2, and stored at −80°C. Each aliquot was thawed and used only once. Protein concentration was determined by bicinchoninic acid assay (Pierce).

DiGE labeling.

DiGE labeling was done as described previously (19). Briefly, equal amounts (μg) of liver protein extract from all 12 animals (6 SA and 6 Ent) were combined into a single tube and mixed to create a reference standard. The pooled mixture was frozen in liquid N2 in small aliquots and stored at −80°C. For labeling, 90 μg of each sample were denatured overnight at room temperature in 8 M urea, 2 M thiourea, 4% CHAPS, and 25 mM Tris (pH 8.8). Denatured proteins were labeled on the following day with Cy2, Cy3, or Cy5 (CyDye DiGE Fluors, GE Healthcare, Piscataway, NJ). Cy2 was always used to label the pooled reference mixture; Cy3 and Cy5 were alternated: three samples from each group of six were labeled with Cy3, and the other three samples were labeled with Cy5, to control for any bias in dye labeling.

2-D gels.

For each gel, one of the individual samples labeled with Cy3 (e.g., SA) and another from the second state labeled with Cy5 (e.g., Ent) were combined with the reference standard (Cy2). The mixed, labeled proteins for each gel (270 μg) were coprecipitated with methanol-chloroform (31); ∼150 μg of total protein were recovered after precipitation. Precipitated proteins were resuspended in 150 μl of a solution containing 9 M urea, 4% CHAPS, 65 mM DTT, 35 mM Tris base, and 0.0025% bromophenol blue and 150 μl of a solution containing 7 M urea, 4% CHAPS, 100 mM DTT, 0.0025% bromophenol blue, 2 M thiourea, and 0.8% 3.5–10 ampholytes “Resolyte” (Gallard-Schlesinger, Plainview, NY) and absorbed into Immobiline DryStrips (pH 3–10 NL, 18 cm; GE Healthcare). After 18–24 h, the strips were transferred to a Multiphor II isoelectric focusing apparatus (GE Healthcare) and focused at 2 mA and 5 W for all steps: 1) ramp to 500 V over 30 min, 2) ramp to 3,500 V over 4 h, and 3) hold at 3,500 V for 14–17 h. The strips were then incubated for 15 min each in reducing [50 mM Tris (pH 6.8), 2% SDS, 15% glycerol, 6 M urea, and 1% DTT] and alkylating [50 mM Tris (pH 6.8), 2% SDS, 15% glycerol, 6 M urea, 1.25% iodoacetamide, and 0.05% bromophenol blue] buffers before SDS-PAGE through 9–16% acrylamide gradient gels. Each gel was scanned in the glass plates with three lasers (Typhoon 9400, GE Healthcare) to collect the Cy2, Cy3, and Cy5 images within 4 h of completion of electrophoresis.

Gels used for spot picking were run as described above, except the starting sample was 270 μg of unlabeled reference standard. The second-dimension gel was poured onto a bind-silane (PlusOne, GE Healthcare)-treated plate. After electrophoresis, these gels were fixed for ≥1 h in 10% methanol-7.5% acetic acid, stained in SYPRO Ruby gel stain (Bio-Rad, Hercules, CA) overnight, and destained in 10% methanol-7.5% acetic acid before they were scanned with the green laser.

Quantitative analysis of 2-D gels.

For analysis of differences in the liver proteome between SA and Ent ground squirrels, 18 images (6 gels with 3 images each: Cy2, Cy3, and Cy5) were analyzed using DeCyder 2-D (version 6.5, GE Healthcare). Approximately 100 spots were manually matched (“landmarks”) on all the Cy2 images in the biological variation analysis module. The software was then used to match the remaining spots. t-Tests were run using the CyDye images in which the Cy3 and Cy5 spot values (pixel volumes) were normalized to their corresponding Cy2 spot value. The biological variation analysis module also includes a statistical post hoc algorithm “false discovery rate” (4), which is a stringent modifier of P values (q) and greatly reduces false positives in a large set of comparisons with a relatively small sample size, in this case, six animals in each state.

Protein recovery and identification.

Each spot with q < 0.05 was examined individually for robustness on 10 gel images (all 6 Cy2 images and 4 pick gel images). After this comprehensive examination, 180 spots that appeared reproducible and robust (i.e., present on ≥4 of 6 Cy2 images and 2 of 4 pick gel images) were chosen as suitable for attempted protein identification by LC-MS/MS.

Protein spots were picked into 96-well plates by an Ettan robotic spot picker (version 1.10, GE Healthcare) using a 1.5-mm-deep, 1.4-mm-diameter picker head. The plates with picked spot plugs were transferred to a second robot (Ettan Digester, version 1.10) for digestion with trypsin. Robotic picks and digests were done in the University of Colorado Health Science Center Proteomics Shared Facility (, and the digested gel spots were kept in 96-well plates at −20°C until analysis by LC-MS/MS. Each sample was applied by HPLC (1100 series pump, Agilent Technologies, Wilmington, DE) onto a 5-μm C18 reverse-phase trapping column (Zorbax 300SB, Agilent Technologies) for aqueous wash before flowing to a 3.5-μm C18 analytical column (Zorbax 300SB). Peptides were eluted by nanospray over a 60-min gradient from 0 to 90% buffer B, where buffer A is 0.1% formic acid in HPLC-grade water (Burdick and Jackson) and buffer B is 90% acetonitrile (HPLC grade, Fisher Scientific) and 0.1% formic acid in HPLC-grade water. Full and tandem mass spectra were collected for each spot using an Agilent XCT Plus ion trap set for Ultra Scan. Each full MS scan was followed by MS/MS scans of the two highest peaks with a 1-min dynamic exclusion. Raw mass spectra data were analyzed using Spectrum Mill MS Proteomics Workbench (revision A.03.02.060a ETD-65). The range of mass limits for the precursor ions was 600–4,000 Da, parent and fragment masses were set to monoisotopic, precursor peptide mass tolerance was 2.5 Da, fragment ion tolerance was 0.7, the enzyme specified was trypsin, the maximum number of internal missed cleavage sites was two, and cysteines were given a fixed modification of +57. An in-house compilation of all National Center for Biotechnology Information (NCBI) mammal sequences in January 2007 was used as the database. It contained 456,753 entries. Because the number of 13-lined ground squirrel sequences in the database is small, most data files were reanalyzed in homology mode using a saved results file; this approach typically increased protein coverage, often substantially. An in-house program, ExtracTags (available on request), was written to collate homology peptides onto a single sequence; the data tables report the best score from all the mapped peptides (see supplemental Table S1 in the online version of this article). Tables 1 and 2 report only identifications (IDs) that are supported by at least two peptides and have a score >30 after elimination of keratin. In many cases, multiple protein IDs were recovered for a single gel spot. Some of these were resolved to a unique ID by consideration of the relative average intensities, scores, and number of peptides in support of each of the multiple protein IDs; the protein supported by the greatest of any of these three criteria appears as the unique ID in Tables 1 and 2, provided the value of at least one of these parameters was at least fourfold greater than that of the other potential IDs. If these criteria were not sufficient to resolve the ambiguity, that spot was not reported in Tables 1 and 2. All the peptide matches recovered by Spectrum Mill for each spot are reported in supplemental Table S1. Spectral data from the same spot from at least two pick gels were combined to increase confidence in protein identifications.

View this table:
Table 1.

Proteins identified in 2-D gel spots that were significantly increased in Ent liver

View this table:
Table 2.

Proteins identified in 2-D gel spots that were significantly decreased in Ent liver

Western blotting.

Forty micrograms of each liver protein extract, the same samples used for DiGE analysis, were fractionated by electrophoresis through 10% SDS-polyacrylamide gels in a Mini-PROTEAN III (Bio-Rad) and then transferred to Sequi-blot polyvinylidene difluoride (Bio-Rad) for 1 h using a GENIE blotter (Idea Scientific), following the manufacturers’ recommendations. The membrane was cut into two pieces to separate all >75-kDa proteins from <75-kDa proteins to allow for detection of ATP citrate lyase (ACLY) and 3-hydroxy-3-methylglutaryl-CoA synthase 2 (HMGCS2) transferred from the same gel. Membranes were blocked for 1 h in 5% milk in Tris-buffered saline-Tween 20 and incubated with primary antibody overnight at 4°C and then with secondary antibody for 1 h at 21°C. Anti-ACLY (Cell Signaling Technology) was used at 1:800 dilution and detected with horseradish peroxidase-linked goat anti-rabbit IgG (GE Healthcare); HMGCS2 (Santa Cruz Biotechnology) was used at 1:500 dilution and detected with horseradish peroxidase-linked donkey anti-goat IgG (Santa Cruz Biotechnology). Blots were developed with ECL Plus (GE Healthcare), chemifluorescence was imaged on a Typhoon 9400 (GE Healthcare), and relative ACLY and HMGCS2 were quantified using ImageQuant 5.2 (GE Healthcare).

Protein function and pathway analysis.

The NCBI Entrez gene database ( and Ingenuity Pathways Analysis ( were used to obtain functional information for the identified proteins. Because human genes generally contain the most detailed information, human gene annotations were used. If the source of the original top-matching protein homolog was not human, the highest-scoring human homolog of the top-matching protein was identified using the BLink link in the sequence record (NCBI). Kyoto Encyclopedia of Genes and Genomes (KEGG) spider (3) was used to assess whether canonical metabolic pathways or gene ontology (GO) categories were significantly enriched in the lists of proteins elevated in SA or Ent ground squirrels. Analyses were carried out with human as the reference organism, with the maximum number (200) of random networks used to determine P values. Only categories with P ≤ 0.05 are reported.


Liver proteins from six SA and six winter-hibernating 13-lined ground squirrels in the Ent phase of the torpor-arousal cycle (Fig. 1) were compared. DeCyder identified 2,485 protein spots on the master gel and statistically analyzed 2,396 of these spots for differences in intensity between the sample groups. The relative intensity of 465 spots was significantly different (P < 0.05) between SA and Ent livers by t-test, and 269 of these protein spots remained significantly different after multiple-test correction (q < 0.05) (4). On inspection of the 269 individual spots, however, several could not be found on the Cy2 images of the majority of gels and, therefore, were eliminated from further consideration. One hundred eighty of the significantly different protein spots were reproducible and robust; extrapolation of this proportion to the entire analysis implies that 1,603 spots were useful for rigorous evaluation of protein differences between Ent and SA. The 180 spots were picked for attempted identification by LC-MS/MS: 51 (28%) of the picked spots were increased in Ent, and 129 (71%) were decreased. The relative differences of the picked spots between SA and Ent were 1.1- and 8.3-fold (Tables 1 and 2).

Unambiguous protein identifications were obtained by LC-MS/MS for 31 of the 51 spots that were elevated in Ent compared with SA (Fig. 2A, Table 1; also see supplemental Table S1). Seven of the remaining spots gave ambiguous IDs, i.e., two or more protein IDs were recovered from 1 gel spot, and the spectra obtained from the remaining 13 gel spots failed to meet the criteria for a positive identification. The same protein was found in several of the 31 spots with unique IDs; therefore, we identified 23 unique proteins that increased in Ent. Of these 23 proteins, 17 were found in 1 spot, 4 were found in 2 spots, and 2 were found in 3 spots (Table 1).

Fig. 2.

Liver proteins fractionated by 2-dimensional (2-D) gel electrophoresis. In a representative pick gel stained with SYPRO Ruby, protein spots for which a unique identification (ID) was obtained are indicated with arrows and their gene ID is given. A: protein spots elevated in Ent. B: protein spots elevated in SA. Approximate molecular mass (in kDa) is shown at left, and approximate isoelectric point is shown at top. •, Reflective reference markers necessary for robotic picking. See Tables 1 and 2 for explanation of abbreviations.

The proteins in 84 of the 129 2-D gel spots that decreased in Ent compared with SA were unambiguously identified. Thirty-one spots gave ambiguous IDs, and 14 were not identified. Fifty-one unique proteins were decreased in Ent compared with SA livers (Fig. 2B, Table 2; also see supplemental Table S1). The remainder of the spots with unambiguous IDs contained proteins that were identified in more than one 2-D gel spot (Table 2).

All but 1 of the 25 proteins that were identified in multiple spots were coordinately increased or decreased in Ent compared with SA. For example, three spots containing transketolase, spots 621, 622, and 623, were decreased 1.4-fold in Ent compared with SA livers. Conversely, three spots of transferrin were increased 1.3- to 1.7-fold in Ent compared with SA livers. The one exception to this pattern was found for the protein HMGCS2, which was identified in four spots: three increased and one decreased, in Ent compared with SA (Tables 1 and 2).

Two of the proteins identified in this screen were selected for validation using Western blotting: ACLY was identified in three spots, which decreased 4.0- to 7.2-fold in Ent (Table 2, Fig. 3A), and HMGCS2 was found in four spots, three of which increased 1.6- to 2.9-fold and one decreased 1.3-fold, in Ent (Tables 1 and 2). Interestingly, ACLY ranged from not detectable by Western blotting in all six Ent samples to varying degrees of greater intensity in the six SA samples (Fig. 3B, top), correlating perfectly with the values obtained using DiGE (Fig. 3A). These results demonstrate the power of the DiGE approach; although Western blotting confirms the protein changes originally revealed by DiGE, the latter method provides the enhanced information of independent quantitative values for multiple isoforms, as well as robust quantification, including greater sensitivity and dynamic range, of the differences between Ent and SA liver. In the case of HMGCS2, again only a single band was resolved in the Western blot (Fig. 3B, bottom), in contrast to the multiple forms recovered by DiGE followed by LC-MS/MS. The size variation of HMGCS2 observed in the 2-D gels is consistent with that of human and rodent HMGCS2s in the NCBI protein database, e.g., GenBank GI numbers 555833, 555835, and 21758044. The single band on the Western blot may be due to a failure of the antibody to recognize the shorter forms or their relative abundance being beneath the level of detection by Western blotting. Nevertheless, the HMGCS2 band was detected in all 12 liver samples. Quantitative analysis of this band revealed a 2.01-fold overall relative increase of the protein during the Ent phase of hibernation compared with SA (Fig. 3C), consistent with DiGE results (Tables 1 and 2). Thus the Western blot results for ACLY and HMGCS2 confirm the changes in the liver proteome that were identified using our broad screening approach and illustrate that DiGE is a robust quantitative method with enhanced quantitative and qualitative capabilities compared with conventional Western blotting.

Fig. 3.

Fold changes revealed by difference gel electrophoresis (DiGE) and Western blot results. A: DiGE quantification of ATP citrate lyase (ACLY) in 12 individual 13-lined ground squirrels (6 SA and 6 Ent). Cy2-normalized value for each of 3 spots (186, 193, and 197) is plotted. B: Western blot of ACLY showing the gel region from 75 to 150 kDa for 12 animals in A (top) and Western blot of 3-hydroxy-3-methylglutaryl-CoA synthase 2 (HMGCS2) showing the region from 37 to 50 kDa of the gel used for ACLY detection (bottom). Only a single band with chemifluorescent product was detected with either antibody, but the reverse image artifact bands observed with anti-HMGCS2 confirm equal protein loading for each sample. M, mol wt marker.C: quantification of HMGCS2 from the bottom Western blot in B. Average values of SA and Ent were normalized to SA. Error bars, SD. HMGCS2 is increased 2.01-fold in Ent compared with SA. ***P < 0.001 (Student's t-test).

Comparison of the functions of the proteins that are increased in Ent hibernators (Table 1) with those that are decreased (Table 2) reveals strikingly little overlap in the types of proteins represented in these two lists. For example, proteins involved in protein synthesis, stability, and folding, RNA metabolism, and fatty acid metabolism and transport dominate the list of proteins that are elevated in Ent hibernators. In contrast, the proteins that decrease in Ent hibernators include many metabolic enzymes, particularly those involved in amino acid and carbohydrate metabolism. Using a more rigorous means to compare the two protein lists, KEGG spider (3) revealed no significantly overrepresented metabolic pathways in the list of proteins elevated in Ent hibernators. In contrast, several metabolic pathways are significantly overrepresented in the list of proteins that decreased in Ent hibernators, including the tricarboxylic acid (TCA) and urea cycles, and pyruvate, amino acid, butanoate, glutathione, and xenobiotic metabolism (Table 3). Remarkably, 72% (38 of 53) of the identified proteins that decreased in Ent livers mapped to these canonical metabolic pathways, a striking contrast to the small number [2 of 23 (9%)] of metabolic proteins that increased in Ent. These changes indicate substantial shifts in liver metabolism during winter hibernation compared with summer.

View this table:
Table 3.

Overrepresented genes in the list of proteins that are decreased in Ent hibernators

KEGG spider also reports significantly overrepresented GO categories, divided among biological process, molecular function, and cellular component. No GO biological processes or molecular functions were significantly overrepresented by the 23 proteins in the list of proteins elevated in Ent hibernators by this analysis. However, proteins belonging to one cellular compartment, endoplasmic reticulum, were significantly overrepresented in the list of liver proteins elevated in Ent hibernators (Tables 1 and 4). In contrast, several GO biological process categories were significantly overrepresented in the list of liver proteins that decreased in Ent, most of which were shared with the overrepresented KEGG pathways [e.g., glycolysis, TCA cycle, and metabolic process (not shown)]. Additionally, proteins with oxidoreductase activity (molecular function) and those residing in the mitochondrial matrix, mitochondrion, and cytosol (cellular components) were significantly overrepresented in the list of proteins that decreased during the hibernation season (Tables 2 and 4).

View this table:
Table 4.

Overrepresented GO categories in proteins increased and decreased in Ent hibernators

The results of these interaction analyses provide strong support for major metabolic alterations between SA and winter-hibernating ground squirrels, as well as significant changes in proteins involved with protein synthesis, stability, and folding. Although these observations are not surprising, given the dramatic phenotypic differences between animals in these two states, the results of this proteomic screen provide specific molecular support for known physiological differences, as well as new insights into the identity of the gene products involved.


This study used a nonbiased screening approach to identify differences in the abundant soluble component of the liver proteome between SA and winter-hibernating 13-lined ground squirrels. The hibernators were in the process of reentering torpor after a spontaneous arousal, a time when all the protein components required for orchestrating and surviving torpor must be present. By far, the vast majority of the protein spots surveyed by this approach remained constant between these two dramatically different physiological states; significant differences were detected in the relative abundance of only 11% of the ∼1,600 (180/269 × 2,396) proteins that were surveyed reliably. Of the 180 protein spots that differed, 28% were increased and 72% were decreased in Ent compared with SA animals. This proportion is retained in the 74 proteins that were unambiguously identified as increased (31%) or decreased (69%) in the Ent compared with the SA state. The subset of liver proteins that differ between the SA and Ent animals provide insight into the biochemical mechanisms underlying the unique physiology of hibernation.

A dramatic shift in fuel and metabolic precursor utilization in liver between these SA and Ent ground squirrels is revealed by the large number of decreased metabolic enzymes in winter that directly and indirectly involve the TCA cycle (Fig. 4). Many hibernators, including 13-lined ground squirrels, fast during the hibernation season and rely on massive quantities of stored fat to fuel their metabolism throughout the winter (6). Thus winter hibernation is characterized by a shift away from oxidation of dietary carbohydrates and toward oxidation of stored fat for energy (16). Levels of β-hydroxybutyrate are elevated in liver (24) and blood (15) of winter season ground squirrels; β-hydroxybutyrate and other ketone bodies serve as an important fuel for brain, heart, and muscle during the hibernation season (1, 15). Proteomic changes between SA and Ent revealed by our study identify specific proteins involved with this fuel alteration: the enzyme responsible for the last irreversible step in glycolysis, pyruvate kinase [pyruvate kinase, liver and red blood cell (PKLR)], was decreased in Ent hibernators, as were two subunits of the pyruvate dehydrogenase complex [pyruvate dehydrogenase β-subunit (PDHB) and dihydrolipoamide dehydrogenase (DLD)], the critical enzyme for conversion of pyruvate into acetyl CoA for entry into the TCA cycle. Two cytoplasmic enzymes of the malate:aspartate shuttle, which bring the reducing equivalents from glycolytic NADH into the mitochondria, malate dehydrogenase (MDH1) and glutamic-oxaloacetic transaminase (GOT1), are also decreased in the Ent phase. In contrast, two proteins involved in fatty acid metabolism and transport, solute carrier family 27 (SLC27A2) and the liver form of fatty acid-binding protein (FABP1), were increased in Ent livers, along with three isoforms of a critical enzyme in ketone body synthesis, HMGCS2.

Fig. 4.

Metabolic enzyme differences between SA and Ent livers centered on the tricarboxylic acid (TCA) cycle. Schematic depicts TCA cycle intermediates within the mitochondrion. Liver proteins, differing between SA and Ent animals (see Tables 1 and 2), that connect directly (solid arrow) or indirectly (dashed arrow) to the indicated metabolites are identified by their Entrez gene identifiers (in italics). If underlined, the enzyme is increased in Ent liver; if not, it is decreased, i.e., elevated in SA compared with Ent. FA, fatty acid; PEP, phosphoenolpyruvate; see Tables 1 and 2 for explanation of other abbreviations.

Additional alterations in the processing of fuel and biosynthetic intermediates are apparent in the lists of proteins that differ between Ent and SA animals. In August, ground squirrels are fattening in preparation for the long fast that characterizes winter hibernation. The liver is the primary site for fatty acid biosynthesis; many of the enzymes that are relatively elevated in the SA livers of this study are consistent with an active conversion of dietary carbohydrate and amino acid precursors into fatty acids (Fig. 4). The need for these enzymes diminishes in winter, when the stored fats are oxidized. ACLY, one of the most strongly regulated proteins revealed by this study, is represented by three spots, which are elevated 4- to 7.2-fold in the SA animals, a finding confirmed by Western blotting. ACLY provides the acetyl CoA for fatty acid biosynthesis, and increased cytoplasmic isocitrate dehydrogenase (IDH1), transaldolase, and transketolase ensure the supply of NADPH required for fatty acid biosynthesis.

The active amino acid catabolism of summer is ramped down to conserve valuable amino acids during winter hibernation by reduced levels of glutamic-oxaloacetic transaminase (GOT1), glutamic-pyruvate transaminase, and numerous others (Table 2, Fig. 4) aided by reduced arginase (ARG1), the rate-limiting enzyme of the urea cycle. Significantly, the lone protein involved in amino acid metabolism that was elevated in Ent liver, acireductone dioxygenase 1 (Table 1), functions in methionine salvage (14). Methionine, an essential amino acid, serves a unique role in initiation of protein synthesis. The reduced level of proteins involved with amino acid catabolism and the urea cycle during winter hibernation is consistent with previous reports of decreased steady-state levels of urea cycle mRNAs (e.g., ARG1) during ground squirrel hibernation (32, 33).

The list of proteins that increased during winter hibernation is striking for its paucity of metabolic enzymes compared with the list of proteins that decreased. Instead, the proteins increased in winter are dominated by those involved in protein synthesis, folding, transport, and stability and RNA-interacting proteins (Table 1). mRNAs encoding several components of the protein biosynthetic machinery were also reported to be elevated in a recent study of hibernation in black bears (10); however, this functional group was not revealed in hibernating ground squirrel liver in studies using similar transcript screening strategies (32, 33). Our results do not address the levels of mRNAs corresponding to these proteins in 13-lined ground squirrels; however, changes in mRNA levels must lead to protein differences, as demonstrated here, to exert an effect on the hibernating phenotype. Thus it appears that adaptations to conserve proteins and facilitate their biosynthesis are a more general feature of hibernating mammals than heretofore appreciated. It has been known for some time that protein synthesis in hibernators is restored to SA levels or greater during each interbout arousal, after nearly ceasing during torpor (28, 35). Increased abundance of the protein biosynthetic machinery throughout winter may be critical to allow cells in aroused hibernators to quickly replenish the proteins that are damaged or depleted during torpor (8, 18). Although increased protein biosynthetic machinery could lead to increased energy consumption and heat production and, therefore, could appear to be contraindicated in hibernation, the activity of this machinery can be controlled by posttranslational mechanisms to reversibly inactivate key components during torpor (7, 11, 29). Reversal of the inactivating modifications during arousal would allow rapid resumption of enhanced protein synthesis during each interbout arousal.

Other protein differences between the SA and Ent livers revealed by this work initially appear paradoxical as well. Hibernating ground squirrels show strong indication of transient hypoxia during arousal (17); yet catalase and two forms of glutathione S-transferase, which could help counter oxidative damage, each decrease in Ent compared with SA liver. A similar decrease in catalase activity was reported in a recent study to assess antioxidant capacity during hibernation in 13-lined ground squirrels (21). All three of these enzymes may decrease in winter hibernators, because their fasted status alleviates the need for detoxification of dietary compounds.

Using comparable conditions of tissue sampling, protein extraction, and protein fractionation through 2-D gels, we previously examined seasonal changes associated with the hibernating liver proteome (8). Our previous study differs from the present study in two important ways: 1) the species and treatment of animals used and 2) the method of quantification. In our previous study, proteomic differences between SA and Ent golden-mantled ground squirrels were examined in livers from SA animals in May, June, or July (2, 4, and 3 animals, respectively) within 12–36 h after they were trapped. The golden-mantled ground squirrels were earlier in the summer phase of their circannual rhythm and less well-acclimated to laboratory conditions than the 13-lined ground squirrels used in the present study. Additionally, the golden-mantled ground squirrels were in the reproductive or early postreproductive phase of their circannual cycle, rather than in the actively fat-storing phase, as was the case for the 13-lined ground squirrels used in the present study. The second difference that distinguishes the two proteomic screens is that proteins fractionated by 2-D gel electrophoresis were quantified after exposure to SYPRO Ruby stain in our previous study, whereas covalent modification with fluorescent dyes was used to quantify proteins in 13-lined ground squirrels in the present study. The present method allows comparison of two samples on each gel, along with a common reference standard, which greatly reduces problems associated with gel-to-gel variability and facilitates between-gel spot matching, leading to greater sensitivity and reproducibility of the quantitative analysis. Both studies reported increased steady-state levels of heat shock 70-kDa protein 5, HMGCS2, ribosomal protein SA, C19orf10 protein, and liver fatty acid-binding protein in Ent animals. Similarly, these two independent proteomic screens revealed decreased levels of the cytoplasmic C-1-tetrahydrofolate synthase, formyltetrahydrofolate dehydrogenase, heat shock 60-kDa protein 1, dihydrolipoamide branched-chain transacylase E2, and the β-subunit of the GDP-forming succinate-CoA ligase in the Ent phase. Seven additional proteins that were identified in both studies were increased in the Ent phase in golden-mantled ground squirrels but decreased in the Ent phase in 13-lined ground squirrels: α-aminoadipate-semialdehyde synthase, phosphoenolpyruvate carboxykinase 2, antiquitin, transaldolase, malate dehydrogenase, glutathione S-transferase M2, and aldehyde dehydrogenase 1B1. Another noteworthy difference between the results of the SA-Ent comparisons of liver proteins in the two species is that approximately two-thirds of the protein spot differences were elevated in SA compared with Ent liver in the 13-lined ground squirrels, whereas the opposite was found in the earlier study; i.e., about two-thirds of the protein spot differences were elevated in Ent compared with SA liver in golden-mantled ground squirrels. Further work is required to determine whether the differences between these two data sets are due to the differences in physiological status of the animals, the methods used, or bona fide species differences.

Perspectives and Significance

Surviving winter hibernation presents numerous challenges to mammalian physiology that are met, at least in part, by altered metabolism. The proteomic changes between summer and winter that we identified by this unbiased screening approach provide insight into the specific mechanisms and pathways involved. In circannual hibernators, including the ground squirrels studied here, the long winter fast with concomitant reliance on stored fat and ketone bodies as metabolic fuel likely reflects the use of long-term starvation pathways that are evolutionarily conserved among terrestrial vertebrates (see Refs. 22 and 30 and references therein). Importantly, proteins are conserved by reduction of the metabolism of amino acids as a source of metabolic fuel and heavy reliance on fat catabolism and ketone body formation. Mammals typically do not store amino acids, except in the form of proteins, but proteins throughout the body are normally considered necessary for function, rather than as reservoirs of stored amino acids. It has long been known that aggressive protein synthesis occurs during the interbout arousals that punctuate hibernation (35). This finding is generally attributed to the need to replenish proteins and restore function during interbout arousal (26, 28, 35). If proteins are to be replenished throughout the winter in the absence of feeding, it is also crucial to avoid catabolism of amino acids, particularly essential amino acids that are not resupplied over the many months of hibernation. However, in a closed system, such as a hibernating mammal, where little to no nitrogen is excreted, the ability to avoid toxic nitrogen accumulation also requires reduction or cessation of amino acid catabolism. Additional studies are needed to determine whether the increased protein biosynthetic capabilities of hibernating mammals represent an adaptation to avoid toxic nitrogen accumulation, to store amino acids for later use, or to provide a means to restore the winter proteome or, most likely, a combination of these mechanisms.


This work was supported by Defense Advanced Research Projects Agency Grant W81XWH-05-2-0016 (to S. L. Martin and H. V. Carey) and National Heart, Lung, and Blood Institute Grant HL-089049 (to S. L. Martin).


No conflicts of interest are declared by the author(s).


We thank Dr. C. Nelson and members of S. L. Martin's laboratory for helpful comments regarding the manuscript.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
View Abstract