Mammalian hibernation involves periods of substantial suppression of metabolic rate (torpor) allowing energy conservation during winter. In thirteen-lined ground squirrels (Ictidomys tridecemlineatus), suppression of liver mitochondrial respiration during entrance into torpor occurs rapidly (within 2 h) before core body temperature falls below 30°C, whereas reversal of this suppression occurs slowly during arousal from torpor. We hypothesized that this pattern of rapid suppression in entrance and slow reversal during arousal was related to changes in the phosphorylation state of mitochondrial enzymes during torpor catalyzed by temperature-dependent kinases and phosphatases. We compared mitochondrial protein phosphorylation among hibernation metabolic states using immunoblot analyses and assessed how phosphorylation related to mitochondrial respiration rates. No proteins showed torpor-specific changes in phosphorylation, nor did phosphorylation state correlate with mitochondrial respiration. However, several proteins showed seasonal (summer vs. winter) differences in phosphorylation of threonine or serine residues. Using matrix-assisted laser desorption/ionization-time of flight/time of flight mass spectrometry, we identified three of these proteins: F1-ATPase α-chain, long chain-specific acyl-CoA dehydrogenase, and ornithine transcarbamylase. Therefore, we conclude that protein phosphorylation is likely a mechanism involved in bringing about seasonal changes in mitochondrial metabolism in hibernating ground squirrels, but it seems unlikely to play any role in acute suppression of mitochondrial metabolism during torpor.
- protein phosphorylation
- thirteen-lined ground squirrel
- mitochondrial respiration
- posttranslational modification
hibernation in small mammals, such as the thirteen-lined ground squirrel (Ictidomys tridecemlineatus), is characterized by repeated bouts of torpor during which whole animal metabolic rate (MR, approximated by mass-specific oxygen consumption rate, V̇o2) is suppressed, presumably to conserve energy during the winter (Fig. 1A). During these torpor bouts, core body temperature (Tb) falls toward ambient levels and can fall as low as 4°C in I. tridecemlineatus. Every 5–14 days, torpor bouts are spontaneously interrupted by arousals, during which Tb and V̇o2 return to levels that approximate those seen during the summer activity (SA) period and are maintained at these levels for ∼8 h. These periods are known as interbout euthermia (IBE). Transitions between IBE and torpor are of great interest due to the large difference in Tb and V̇o2 between these states (37°C to 4°C respectively, V̇o2 suppressed to 1% of IBE levels), the rate at which these changes occur (<8 h), and the fact that changes in V̇o2 precede changes in Tb, which is indicative of active suppression of metabolism.
In the attempt to identify mechanisms that hibernators employ to suppress MR, much research has focused on mitochondria, as they are responsible for consuming 90% of respired O2 in mammals (40). Particular attention has been paid to mitochondria originating from metabolically active visceral organs, such as the liver, which accounts for 18% of mammalian (mouse) resting MR but only 6% of total body mass (33). Although passive thermal effects clearly alter mitochondrial metabolism at low Tb (12, 34), active suppression of metabolism almost certainly plays a role (reviewed in Ref. 45). Comparisons of succinate-fueled liver mitochondrial O2 consumption (measured in vitro at 37°C) between euthermic (i.e., SA and IBE) and torpid ground squirrels have repeatedly shown that respiration is suppressed by up to 70% (4, 12, 14, 18, 34). In characterizing mitochondrial respiration during arousal from torpor, Armstrong and Staples (4) determined that alleviation of this suppression occurs slowly, with only 50% recovery by the time Tb increased from 5°C to 30°C. Conversely, during entrance into a torpor bout Chung et al. (14) observed that suppression of mitochondrial respiration occurs rapidly, with maximal suppression seen before Tb fell below 30°C (Fig. 1B).
Recently we have been attempting to connect the “fast in, slow out” pattern (46) of mitochondrial respiration with a mechanistic explanation. We have found that inhibition of succinate dehydrogenase by oxaloacetate occurs during torpor (4, 11a) but cannot fully account for the observed suppression of mitochondrial metabolism. Also, mitochondrial membrane phospholipids are remodeled between entrance, torpor, and IBE (5, 14), but these effects do not correspond with the fast-in, slow-out pattern of mitochondrial suppression, and other mechanisms must be involved.
Any mechanisms involved in the reversible suppression of mitochondrial metabolism should meet several criteria. First, they are likely temperature dependent, because the initiation of suppression occurs rapidly at high Tb (30–37°C) but only slowly during arousal, when Tb begins near 5°C. Second, any mechanism must be fairly rapid, as full suppression of mitochondrial respiration during entrance occurs within 2 h (14), likely excluding changes in gene transcription and translation. Finally, the mechanism must be readily reversible because suppression of mitochondrial respiration (and MR) is repeatedly reversed throughout the hibernation season. Posttranslational protein modification by kinase-mediated phosphorylation and/or phosphatase-mediated dephosphorylation meets these criteria, and, as such, our goal with this study was to investigate their role as potential regulators of mitochondrial respiration during hibernation.
The regulatory capacity and pervasiveness of reversible protein phosphorylation are well documented, but only recently has its role in mitochondrial regulation started to receive attention. Relatively recent studies have established the existence of mitochondria-associated protein kinases and phosphatases and reversible phosphorylation of mitochondrial enzymes (7, 23, 28, 36, 38, 39, 42, 48, 52), with many of these changes being linked to alterations in mitochondrial oxidative phosphorylation capacity. For example, Acin-Perez et al. (1) demonstrated that phosphorylation of ETC complexes I and IV, by an endo-mitochondrial cAMP-dependent protein kinase (PKA) signaling cascade, significantly increases mitochondrial respiration in isolated HeLa cells. In addition, studies focused on the effects of posttranslational modification of specific electron transport chain (ETC) complexes, such as cytochrome c oxidase, have shown that tyrosine residue phosphorylation occurs and has an inhibitory effect (29, 51).
Reversible phosphorylation has been shown to play a role in metabolic regulation in mammalian hibernation. Many changes in phosphorylation state and the enzymes that mediate these changes have been linked to energy saving requirements and the well-described shift in fuel use toward lipids during hibernation. For example, enzymes associated with glycolysis (11), actin filament turnover (22), Na+-K+ transport (31), and amino acid metabolism (6) have been shown to change phosphorylation state during hibernation. However, the potential for reversible phosphorylation to mediate the profound changes in mitochondrial metabolism that occur over the course of a single torpor bout have yet to be explored.
The objective of the present study was to test the prediction that changes in mitochondrial protein phosphorylation state correspond to changes in mitochondrial respiration that are known to occur between IBE and torpor. Using a phosphoproteomic approach, we determined that the phosphorylation state of several mitochondrial proteins does change in the thirteen-lined ground squirrel. These changes, however, appear more related to seasonal changes than to changes in mitochondrial metabolism within torpor bouts. Thus changes in the mitochondrial phosphoproteome may be important in preparing mitochondria for the shift in fuel use seen in the winter season.
Unless specified, all reagents were purchased from Sigma-Aldrich (Oakville, ON, Canada).
All procedures were approved by the University of Western Ontario Animal Use Subcommittee (protocol number 2008-055-06). Adult I. tridecemlineatus were captured from the wild (Carman, MB, Canada; 49°30′N, 98°01′W) during a 2 wk period in early May. Daily animal care procedures have been described previously (12). During the summer (May to August), animals were kept in a conventional animal care room [ambient temperature (Ta) = 22 ± 3°C, photoperiod adjusted weekly to match that in Carman, MB].
To facilitate hibernation in early October, animals were transferred to an environmental chamber and kept under low Ta and short photoperiod conditions (Ta = 4 ± 2°C; photoperiod = 2 h light-22 h dark, lights on at 8:00 AM EST). Hibernation was observed in all individuals within 2 wk, at which point food, but not water, was removed, since these animals do not eat during hibernation.
Core body temperature monitoring.
Temperature-sensitive radiotelemeters (model TA-F20; Data Sciences International, St. Paul, MN) were implanted abdominally to monitor hibernation bouts via measurement of Tb via telemetry receivers (model RA1010) as previously described (33).
Liver mitochondria isolation.
Experimental groups were classified as follows: summer active (one male, three females) I. tridecemlineatus were sampled in the morning (8:30 AM ± 30 min EST) during July (Tb = 37 ± 1°C); winter sampling of torpid (four males, four females) and IBE (four males, five females) ground squirrels occurred during January and February. IBE was defined as the period following spontaneous arousal where Tb = 37 ± 1°C for no less than 1 h, and torpor was the period where Tb = 5 ± 1°C for at least 48 h. Sampling of all torpid individuals occurred at 8:30 AM EST ± 30 min; however, due to the spontaneous nature of interbout arousals, IBE individuals could not be sampled at a standardized time and were sampled between (8:30 AM and 2:30 PM EST).
Summer and IBE individuals were euthanized by anesthetic overdose (intraperitoneal injection of Euthanyl; 270 mg/ml, 0.2 ml/100 g; Ref. 32). Euthanyl has been shown to have no effects on mitochondrial metabolism (47). Animals sampled during torpor were euthanized by cervical dislocation to prevent arousal. Following excision, the whole liver was weighed, placed in ice-cold liver homogenization buffer [LHB; 250 mM sucrose, 10 mM HEPES, 1 mM EGTA, 1% BSA (wt/vol), pH 7.4 at 37°C], and minced into 1 mm3 pieces.
Isolation of purified mitochondria followed methods outlined by Brown et al. (12). Isolation involved the homogenization of liver followed by differential centrifugation. The resulting crude mitochondrial preparation was purified by Percoll density gradient centrifugation, which yielded a final mitochondrial pellet that was suspended in 1 ml LHB and kept on ice until respiration rates were measured (within 4 h).
Respiration rates of purified liver mitochondria (25 μl aliquot, ∼0.1 mg protein) were measured with a Clark-type polarographic O2 electrode (Oxygraph-2k; Oroboros, Innsbruck, Austria) in 2 ml (total volume) of mitochondrial respiration assay buffer [MiR05; 110 mM sucrose, 0.5 mM EGTA, 3 mM MgCl2, 60 mM K-lactobionate, 20 mM taurine, 10 mM KH2PO4, 20 mM HEPES, pH 7.1 at 30°C, 1% (wt/vol) BSA]. All substrates were dissolved in MiR05, except rotenone and oligomycin, which were dissolved in ethanol, and final concentrations represent those in the Oxygraph-2k chamber. The O2 response of the Oxygraph-2k was calibrated to air-saturated and oxygen-depleted buffer (obtained with the use of a yeast suspension) levels based on published O2 solubilities (19), corrected for local atmospheric pressure. All respiration rate measurements were performed at 37°C with constant stirring (750 rpm).
Succinate-fueled state 3 (ADP-phosphorylating) and state 4 (nonphosphorylating) mitochondrial respiration rates were measured as follows. Rotenone (0.5 μM), an ETC complex I inhibitor, was added to the mitochondrial suspension prior to the addition of succinate (10 mM) to observe flux through ETC complex II. ADP (2 mM) was introduced to stimulate state 3 respiration, and state 4 respiration rate was estimated by the addition of oligomycin (2 μg/ml), which inhibits ATP synthase.
All respiration rates were standardized to mitochondrial protein concentration, determined by Bradford assay (10). Standards consisted of BSA dissolved in LHB. Remaining mitochondrial samples were divided into pellets containing ∼1 mg mitochondrial protein by centrifugation (21,000 g for 10 min at 4°C) and stored at −80°C until further use. To mitigate any change in protein phosphorylation due to freeze-thaw cycles, samples remained frozen until solubilized.
Mitochondria protein sample preparation.
Frozen mitochondrial proteins were suspended in 500 μl of SDS-PAGE sample buffer [20% (vol/vol) glycerol, 4% (wt/vol) SDS, 100 mM Tris pH 6.8, 0.002% (wt/vol) bromophenol blue, 100 mM dithiothreitol (DTT)]. Samples were shaken for 30 min (200 rpm) at 25°C and subsequently centrifuged at 21,000 g for 5 min at 25°C. The supernatant was removed for SDS-PAGE.
Each gel was loaded with one liver mitochondrial protein sample (15 μg) from each metabolic stage (SA, IBE, Torpor; all liver mitochondrial samples were included as biological replicates, one technical replicate for each sample), and a single SA external control (15 μg of protein from the same SA individual loaded onto all gels to allow for standardization and intergel comparisons; Fig. 2A). SDS-PAGE was performed using a Mini-Protean II apparatus (Bio-Rad, Mississauga, ON, Canada) with a 12.5% (wt/vol) bis-acrylamide resolving gel [6 M urea, 0.66 M Tris pH 8.8, 0.6% (wt/vol) SDS, 0.3% (vol/vol) APS, 0.2% (vol/vol) TEMED] and a 5% (wt/vol) bis-acrylamide stacking gel [0.125 M Tris pH 6.8, 1% (wt/vol) SDS, 1% (vol/vol) APS, 0.1% (vol/vol) TEMED]. Electrophoretic running buffer (0.25 mM Tris pH 8.6, 1.92 mM glycine, 0.1% SDS) was based on Laemmli (27). Separation of proteins was completed under constant voltage (100 V, 1 h 45 min at 25°C). Total protein was visualized with a Coomassie G-250 stain [50% (vol/vol) methanol, 10% (vol/vol) acetic acid; Fig. 2A].
Protein transfer and immunodetection.
After SDS-PAGE, proteins were transferred to nitrocellulose membranes (0.2 μm pore diameter, Bio-Rad) using a Mini Trans-blot system (Bio-Rad). Immunoblotting buffer [25 mM Tris pH 8.6, 1.92 mM glycine, 20% (vol/vol) methanol, 0.3% (wt/vol) SDS] and transfer protocol were taken from Towbin et al. (49). Proteins were transferred under constant voltage (100 V, 1 h at 4°C).
Membranes were exposed overnight at 4°C to blocking solution [2 mM Tris pH 7.5, 15 mM NaCl, 0.05% (vol/vol) Tween 20 containing 5% (wt/vol) BSA]. Blocked membranes were rinsed in membrane wash buffer [2 mM Tris pH 7.5, 0.05% (vol/vol) Tween 20] and probed for 1 h at 25°C with the following primary antibodies diluted in antibody buffer [2 mM Tris pH 7.5, 15 mM NaCl, 0.25% (vol/vol) Triton X-100] containing 2% (wt/vol) BSA, rabbit anti-phosphothreonine (P-Thr, product number 71-8200) 1:250, rabbit anti-phosphoserine (P-Ser, product number 61-8100) 1:250, rabbit anti-phosphotyrosine (P-Tyr, product number 61-5800) 1:250 (Invitrogen, Camarillo, CA), and the mitochondria-specific loading control mouse anti-voltage dependent anion-selective channel protein 1 (VDAC1) 1:1,000 (Abcam, Cambridge, MA; product number ab14734). Probed membranes were exposed to their respective horseradish peroxidase-conjugated secondary antibodies (anti-rabbit, product number a9169; anti-mouse, product number 31430; Thermo Scientific, Rockford, IL) and exposed to ECL chemiluminescent detection reagent (Amersham, Amersham, UK). Antibody-protein complexes were visualized on X-ray film (see Fig. 2, B–D; Fujifilm, Tokyo, Japan). Digitizing of X-ray film at a resolution of 600 dpi was completed with a flatbed scanner (7400c; Hewlett Packard, Mississauga, ON, Canada).
Image conversion to grayscale (8 bit) and densitometric analyses were completed using ImageJ v1.46h software (National Institutes of Health, Bethesda, MA). Background correction was completed using the rolling ball function in ImageJ (resolution 300 pixels). Briefly, vertical gel lanes of equal dimensions were converted to density maps, where area under the peak indicates protein quantity. When necessary, peaks of interest were isolated from surrounding signals with vertical lines that extended from the base of the peak to the base of the density map. We performed densitometric comparisons among all detectable phosphoproteins, while keeping our analyses conservative by strictly using protein bands that were clearly distinguished across all samples. This approach was used due to the occurrence of multiple immunoblot signals from an individual sample (i.e., within an individual lane), arising from the use of nonprotein-specific phosphorylated amino acid antibodies. While this has the potential to result in biologically relevant changes being overlooked, it was ultimately necessary in order to prevent the misinterpretation of an incomplete dataset. The area under the curve for a given protein of interest [identified by apparent molecular weight (MW)] was expressed as a percentage of the sum of peak area across all loaded samples (for the same protein, within the same gel). This percentage was then corrected to the SA external control peak.
Two-dimensional blue-native PAGE of mitochondrial phosphoproteins.
Separation of liver mitochondrial proteins for identification (one biological replicate from each metabolic state) was achieved by a modified two-dimensional blue-native PAGE (2D BN-PAGE) technique based on Schägger and von Jagow (43). Mitochondrial proteins (1 mg) were solubilized in mitochondrial extraction buffer (750 mM aminocaproic acid, 50 mM bis-Tris pH 7.0) and 10% (vol/vol) lauryl maltoside solution (β-DDM) on a shaker for 20 min at 4°C. Solubilized samples were centrifuged for 20 min at 21,000 g at 4°C, and the resulting supernatant was combined with BN-PAGE sample buffer [50 mM bis-Tris·HCl pH 7.2, 50 mM NaCl, 10% (wt/vol) glycerol, 0.001% (wt/vol) Ponceau S], and a 5% (wt/vol) Coomassie G-250 suspension (750 mM aminocaproic acid pH 7.5).
Mitochondrial protein (200 μg) from each metabolic state was loaded onto a 4–16% (wt/vol) bis-acrylamide precast gradient gel (Fig. 3A; Invitrogen, Carlsbad, CA). We used an XCell Surelock Mini-Cell (Invitrogen) at 4°C under a constant current (17 mA for 1.5 h). The buffer system consisted of a cathode [0.002% (wt/vol) Coomassie G-250, 50 mM bis-Tris pH 7.0, 50 mM tricine], and anode buffer (50 mM bis-Tris, 50 mM tricine pH 7.0). Protein separation was completed on three separate gels for immunoblot detection of each phospho-amino acid residue, and a fourth gel was used for Coomassie staining and protein identification by matrix-assisted laser desorption/ionization-time of flight/time of flight mass spectrometry (MALDI-TOF/TOF MS).
Prior to protein separation in the second dimension, ETC complexes from each metabolic state were excised and solubilized with a denaturing solution [1% β-mercaptoethanol, 1% (wt/vol) SDS] for 15 min at 25°C. Solubilized gel pieces were sealed in a 5% (wt/vol) bis-acrylamide stacking gel with agarose [10% (wt/vol) dissolved in SDS-PAGE running buffer] and placed on top of a 12.5% (wt/vol) bis-acrylamide resolving gel. Proteins from each ETC complex, from animals of each metabolic state, were subsequently separated via SDS-PAGE by methods described above (Fig. 4A).
2D BN-PAGE technique confirmation.
To confirm that all ETC complexes were present following 2D BN-PAGE separation, an immunoblot analysis for one subunit unique to each ETC complex was completed (Fig. 3B). ETC subunit antibodies (Mitosciences, Eugene, OR) and dilutions were as follows: mouse anti-CI NDUFA9 subunit 1:1,000 (MS111), mouse anti-CII 30 kDa Ip subunit 1:200 (MS203), mouse anti-CIII Rieske subunit 1:1,000 (MS305), mouse anti-CIV subunit IV 1:1,000 (MS407), and mouse anti-CV subunit d 1:1,000 (MS504).
Mitochondrial proteins (one biological replicate from each metabolic state) separated by 2D BN-PAGE were immunoblotted with antibodies specific to phospho-amino acid residues, by techniques described above (Fig. 4). This method allows for isolation and observation of ETC complex subunits that undergo changes in phosphorylation between different stages of hibernation. Automated in-gel digestion of mitochondrial proteins (modified from Ref. 41; Waters MassPREP Station; PerkinElmer, Milford, MA), MALDI-TOF/TOF MS (4700 Proteomics Analyzer; Applied Biosystems, Foster City, CA), data acquisition (4000 Series Explorer, Applied Biosystems), and data processing (Data Explorer, Applied Biosystems) were completed at the UWO Functional Proteomics Facility (London, ON, Canada). Protein spots of interest were excised using an Ettan spot picker (GE Healthcare, Buckinghamshire, UK). Briefly, in-gel digestion involved repeatedly washing gel pieces with 50% (vol/vol) acetonitrile (ACN) followed by rehydration with 100 mM NH4HCO3. This washing procedure was followed by a treatment with 50 mM DTT at 56°C for 45 min; subsequently, proteins were alkylated with 200 mM iodoacetamide for 30 min 25°C in the dark. Gel pieces were again repeatedly washed with 50% (vol/vol) ACN and 100 mM NH4HCO3.
Proteins were digested overnight at 37°C, with modified trypsin (12.5 ng/μl; Promega, Fitchburg, WI) suspended in digestion buffer (25 mM NH4HCO3 pH 7.8, 2.5 mM NaCl). Gel pieces were sequentially washed with a 50% (vol/vol) ACN, 25 mM NH4HCO3 solution and a 50% (vol/vol) ACN, 5% (vol/vol) formic acid solution, with supernatant being extracted following each wash. Prior to MS analysis, dried peptide samples were redissolved in a 10% (vol/vol) ACN and 0.1% (vol/vol) trifluoroacetic acid solution.
MALDI matrix, α-cyano-4-hydroxycinnamic acid, was prepared as 5 mg/ml in 6 mM ammonium phosphate monobasic, 50% (vol/vol) ACN, 0.1% (vol/vol) trifluoroacetic acid and mixed with the sample at 1:1 ratio (vol/vol). This mixture (0.7 μl) was spotted on the MALDI target. The MALDI-TOF/TOF MS was equipped with a 355 nm Nd/YAG laser; the laser rate was 200 Hz. Reflectron and linear positive ion modes were used. Reflectron mode was calibrated at 50 ppm mass tolerance. Each mass spectrum was collected as a sum of 1,000 shots. Peptide sequences obtained from MALDI-TOF/TOF MS were submitted to the NCBInr database, which provided putative protein identities. A minimum threshold protein score confidence interval of 95% was set, such that any protein identities inferred by peptide sequence matching exceeding this value were deemed statistically significant. All detected 2D BN-PAGE phosphoproteins were submitted for identification, but only those proteins with confirmed identity, and an apparent MW that matched proteins with an observed change in phosphorylation state (determined through densitometric analyses of liver mitochondrial samples separated by SDS-PAGE) were given further consideration (Table 1).
Confirming putative identities provided by MALDI-TOF/TOF MS.
We performed immunoblot analyses to confirm the putative identity of one of our proteins, long chain-specific acyl-CoA dehydrogenase (LCAD). This analysis used mouse anti-LCAD (1:1,000, Abcam, ms1068) and mitochondrial proteins separated via modified 2D BN-PAGE and SDS-PAGE as described above. An identical apparent MW of the protein spot picked for MALDI-TOF/TOF MS and the SDS-PAGE immunoblot analysis was used to verify our technique (Fig. 5).
In all instances, α was set at 0.05, and data presented are means (SD). Differences in succinate-fueled mitochondrial respiration rates (state 3, state 4), and respiratory control ratio (state 3/state 4, an indicator of the mitochondrial coupling) among metabolic states were assessed by a one-way ANOVA, followed by a Tukey's post hoc test.
Changes in mitochondrial protein-specific phosphorylation state [values obtained from the one-dimensional (1D) SDS-PAGE densitometric analyses] were assessed by multivariate statistical analyses. In this way, seasonal changes in phosphorylation, among groups of proteins (based on MW), could be elucidated. For each phosphorylated amino acid, a principal component analysis (PCA) was used to compare the relative density of proteins of the same MW among metabolic states. “PCscores2” from each PCA was extracted for further analysis and used to represent metabolic state-dependent phosphorylation changes unique to each individual.
In the case of P-Thr, we excluded proteins with PCA correlation coefficients between −0.3 and 0.3 (i.e., Thr 67, 57, and 29), because correlation coefficients that fall within this range can be considered to not contribute greatly to observed effects (25) and there was no a priori reasoning to include all phosphoproteins in the initial P-Thr PCA.
PCscores2 for each phospho-amino acid was analyzed by one-way ANOVA to assess differences in the pattern of protein phosphorylation among metabolic states. This was followed by a Tukey's post hoc test. A Pearson's coefficient correlation comparing PCscores2 and succinate-fueled state 3 mitochondrial respiration was used to determine whether observed changes in groups of phosphorylated proteins correlate with changes in flux through the ETC.
Differences in the relative density of VDAC1, among each metabolic state were assessed by a one-way ANOVA. All statistical analyses were completed with SPSS 16.0 (IBM, Armonk, NY).
State 3 succinate-fuelled liver mitochondrial respiration did not differ significantly between IBE and SA (P = 0.544). There was, as has been shown previously, a significant suppression of state 3 respiration in torpid animals compared with both IBE (57%) and SA (48%, Fig. 6, P < 0.001). State 4 succinate-fueled liver mitochondrial respiration did not differ significantly among the metabolic states (P = 0.512).
Patterns of mitochondrial protein phosphorylation.
The mitochondrial content of VDAC1, as assessed by densitometry, did not differ significantly among metabolic states (Fig. 7, P = 0.911), which indicates that our protein quantification and equal loading methodology is accurate. To assess changes in protein phosphorylation among metabolic states, we performed a PCA using 1D SDS-PAGE densitometry values. Across all phosphorylated amino-acid residues, PC1 accounted for the majority of observed variation (P-Thr = 53.85%, P-Ser = 56.02%, P-Tyr = 45.97%). However, PC1 only accounts for differences in absolute protein phosphorylation. Unlike PC1, PC2 (or “metabolic state-dependent phosphorylation”) accounts for variation in phosphorylation among metabolic states. Eigenvalue variation explained by PC2 for each phosphoprotein was as follows: P-Thr (34.9%), P-Ser (19.2%), P-Tyr (40.3%). In the case of P-Thr PC2, a 49 and 41 kDa protein band had negative loading values, whereas a 37 and 26 kDa protein had positive loading values (Fig. 8A). P-Ser PC2 exhibited a similar trend with a 47 and 38 kDa band having positive loading values, and a 34, 26, and 21 kDa band having negative loading values (Fig. 8C). In the case of P-Tyr PC2 a 36 kDa band had a positive loading value and a 28 kDa protein was negatively loaded (Fig. 8E).
Relative density of phosphoprotein among metabolic states.
Regardless of which phosphorylated amino acid residue is considered, absolute phosphorylation (PC1) did not differ among metabolic states (P-Thr P = 0.255, P-Ser P = 0.286, P-Tyr P = 0.228). P-Thr metabolic state-dependent phosphorylation (PC2) differed between seasons (i.e., SA vs. IBE and torpor; P < 0.01) but not within the winter season (i.e., torpor vs. IBE; Fig. 8B). A similar pattern was noted in P-Ser, but a significant difference was evident only between the SA and torpor states (Fig. 8D, P = 0.020). P-Tyr metabolic state-dependent phosphorylation did not differ among metabolic states (Fig. 8F, P = 0.508).
Mitochondrial protein phosphorylation and mitochondrial respiration.
Neither total phosphorylation (PC1) nor metabolic state-dependent phosphorylation (PC2) correlated significantly with succinate-fueled state 3 liver mitochondrial respiration rate (PC1: P-Thr P = 0.338, P-Ser P = 0.135, P-Tyr P = 0.999; PC2: P-Thr P = 0.638, P-Ser P = 0.68, P-Tyr P = 0.692; Fig. 9). We examined the relationship between the densitometric signal of individual phosphorylated protein bands (proteins that change phosphorylation among metabolic states, determined by PCA) and succinate-fueled mitochondrial respiration and determined there was no significant correlation (P-Thr: 49 kDa P = 0.395, 41 kDa P = 0.229; P-Ser: 47 kDa P = 0.667, 38 kDa P = 0.704; P-Tyr: 28 kDa P = 0.257).
Identification of proteins that undergo changes in phosphorylation state.
We attempted to identify mitochondrial proteins that changed phosphorylation state in a season-dependent fashion (P-Thr: 49, 41, 37, and 26 kDa; P-Ser: 47, 38, 26, and 21 kDa), using MALDI-TOF/TOF MS on phosphoproteins picked from 2D BN-PAGE separated gels (phosphoproteins identified by phospho-amino acid-specific immunoblotting, Fig. 4). This analysis yielded nine putative protein identities (Table 1). Of these proteins, three had an apparent MW that matched the SDS-PAGE immunoblot bands that changed phosphorylation among metabolic states (Fig. 8). Significant seasonal changes in phosphorylation occurred in proteins identified as F1-ATPase α-chain (SA > IBE, torpor), ornithine transcarbamylase (SA > torpor), and long chain-specific acyl CoA dehydrogenase (SA > torpor).
Confirming MALDI-TOF/TOF MS analysis.
To confirm the putative identities provided by MALDI-TOF/TOF MS we performed an immunoblot analysis of 2D BN-PAGE, and SDS-PAGE separated mitochondrial proteins using antibodies specific to LCAD. We determined that the putative identification provided by MALDI-TOF/TOF MS was correct, as the protein's location matched the apparent MW of the 2D BN-PAGE picked sample as well as the SDS-PAGE anti-LCAD immunoblot (Fig. 5).
The aim of this project was to assess the role of protein phosphorylation in the regulation of mitochondrial respiration during hibernation. We predicted that protein phosphorylation plays a role in altering mitochondrial respiration during torpor. The absence of a significant correlation between state 3 respiration and protein phosphorylation among metabolic states and the fact that changes in mitochondrial protein phosphorylation and O2 consumption occur in different patterns (seasonal vs. metabolic state, respectively) do not support our prediction.
The observed seasonal changes in phosphorylation may, nonetheless, be related to the regulation of mitochondrial metabolism during hibernation. Previous studies on the thirteen-lined ground squirrel liver metabolome (35, 44), plasma proteome (16), and cardiac proteome (22) observed that when changes do occur, they are consistent with a “Two-Switch” pattern of hibernation (44). The Two-Switch pattern describes hibernation as a cycle of torpor bouts within a larger seasonal cycle. This construct recognizes that hibernators must undergo preparatory changes prior to winter to express the hibernation phenotype and, like any nonhibernating mammal in winter, cope with low Ta and high Tb during IBE. Our experimental design, comparing summer animals that do not hibernate with winter animals that cycle between torpor and IBE allowed us to observe the effects of protein phosphorylation without being limited to torpor-arousal cycles. Our data showed seasonal differences in mitochondrial protein phosphorylation (i.e., SA differs from one or both winter states) rather than differences related to metabolism within torpor bouts. In instances where seasonal patterns occur in hibernator proteomes (15) and metabolomes (16, 44), changes often occur in proteins and metabolites associated with energy homeostasis. As such, the seasonal changes in phosphorylation state observed in this study may facilitate the ability to undergo torpor cycles in the winter.
Proteins that changed phosphorylation state seasonally were putatively identified as ornithine transcarbamylase, the F1-ATPase α-chain, and LCAD. Hibernation-specific fuel usage, which includes a shift away from carbohydrate metabolism during winter, has been repeatedly demonstrated with metabolomic and proteomic studies (17, 34, 44). It may be the case that the changes in phosphorylation we observe are linked to seasonal shifts in fuel usage.
The effect of phosphorylation of ornithine transcarbamylase, a urea cycle enzyme, has, to our knowledge, not been studied. Nonetheless, plasma metabolomics have demonstrated higher circulating urea concentration during summer in I. tridecemlineatus (16). While protein metabolism does not typically represent a large fuel source for mammals, protein catabolism nevertheless occurs and requires enzymes such as ornithine transcarbamylase. Protein and amino acid metabolism may be important during the winter hibernation season, when this species does not feed, by supplying carbon skeletons for gluconeogenesis (reviewed in Ref. 9). Protein metabolism during the winter is critical to the hibernator's survival, so the activity of enzymes involved in this metabolism may undergo alteration which is perhaps facilitated by phosphorylation.
Much like ornithine transcarbamylase, the F1-ATPase α-chain underwent season-dependent changes in phosphorylation state [i.e., SA > (IBE, torpor)]. This ETC complex V component is one of two subunits directly involved in the catalytic synthesis of ATP. Although changes in the phosphorylation state of the F1-ATPase α-chain have, to the best of our knowledge, no currently defined functional role, a reduction in the activity of the ADP phosphorylation system (i.e., ATP synthase, the adenine nucleotide transporter, and the phosphate transporter) has been demonstrated between IBE and torpor in I. tridecemlineatus, (12). We hypothesize that seasonal changes in the phosphorylation state of the F1-ATPase α-chain may permit torpor bout-specific changes in ADP phosphorylation kinetics.
It is well established that hibernators undergo a seasonal fuel shift during winter away from carbohydrates and toward lipids (reviewed in Refs. 3, 13). As such, our observed changes in mitochondrial protein phosphorylation may help to explain how these shifts are regulated. Epperson et al. (16) show that plasma metabolite concentrations were indicative of increased lipid usage in winter states (e.g., increased palmitic acid during torpor). We hypothesize that the seasonal change (SA > torpor) in LCAD phosphorylation facilitates the shift in fuel use. Although the literature is sparse in demonstrating changes in phosphorylation of LCAD (which catalyzes the first step in β-oxidation), other members from this protein class (e.g., very long-chain acyl-CoA dehydrogenase) exhibit changes in phosphorylation state that correspond to changes in activity. Studying medium-chain acyl-CoA dehydrogenase isolated from pig kidneys, Macheroux et al. (32) observed that reversible phosphorylation does occur but were unable to assign a functional role to this change. Kabuyama et al. (26) demonstrated that serine-dependent phosphorylation of very long-chain acyl-CoA dehydrogenase is required for proper β-oxidation and mitochondrial electron transfer ability. In their study of the golden-mantled ground squirrel (Spermophilus lateralis) liver proteome, Epperson et al. (15) observed a small but significant increase in LCAD during the winter state. This increase reflects the increased lipid metabolism known to occur during the hibernation season; however, changes in phosphorylation state as well as an increase in the quantity of LCAD may afford hibernators greater control over β-oxidation during torpor.
Although changes in LCAD phosphorylation are most likely related to shifts in fuel use during hibernation, they may also impact mitochondrial metabolism. Pande and Blancher (37) demonstrated that long-chain acyl-CoA esters inhibit rat heart mitochondrial respiration, possibly through inhibition of adenine nucleotide translocase. Additionally, Ventura et al. (50) demonstrated that a buildup of long-chain acyl-CoA esters inhibits succinate and glutamate transport in rat liver mitochondria. Lerner et al. (30) previously demonstrated that the inhibitory effect of long-chain acyl-CoA esters on liver mitochondrial function occurs in thirteen-lined ground squirrels in hibernation. The changes we have observed in phosphorylation state of LCAD may be a prerequisite for the suppression of mitochondrial respiration during hibernation.
The findings of this study present many opportunities for future research, in particular, identification of torpor-specific metabolic switches. Phosphoproteomics is still a relatively young and emerging field with the potential to improve links between described patterns and functional biological changes. It is important to note that although we were unable to detect changes in phosphorylation state associated with changes in mitochondrial respiration during torpor, our findings do not preclude the possibility that these changes occur. Indeed, the difficulty of isolating membrane bound mitochondrial proteins, let alone observing the phosphorylation status of these proteins is well understood. In addition to providing novel insight into the role of phosphorylation in regulating changes in a hibernator, this study highlights the necessity for additional work. Thus, it may be informative to specifically target membrane-embedded proteins (such as cytochrome c oxidase) employing refined methodologies such as those described by Helling et al. (24).
Utilizing powerful molecular tools, such as 2D difference gel electrophoresis (8, 21), can build upon this project's findings to determine how mammalian metabolic suppression is achieved during hibernation. Additionally, an in vitro manipulation of phosphorylation state of our identified proteins and an analysis of the resultant enzyme activity in different metabolic states will help us to understand the functional effects of the changes in phosphorylation state observed in this study. The potential effects of other forms of posttranslational modification (e.g., acetylation, glycosylation) of mitochondrial proteins and function should be examined, especially since most subunits of ETC complexes have been demonstrated as targets for acetylation (2).
Funding was provided by the Natural Sciences and Engineering Research Council of Canada in the form of Discovery and Research Tools and Infrastructure Grants to J. F. Staples. Funding was provided to D. J. Chung in the form of an Ontario Graduate Research Scholarship.
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: D.J.C., B.S., J.C.L.B., N.P.A.H., and J.F.S. conception and design of research; D.J.C. and B.S. performed experiments; D.J.C. and J.C.L.B. analyzed data; D.J.C. and J.F.S. interpreted results of experiments; D.J.C. prepared figures; D.J.C. drafted manuscript; D.J.C., J.C.L.B., N.P.A.H., and J.F.S. edited and revised manuscript; J.F.S. approved final version of manuscript.
Alvin Iverson at the Carman and Area Research Center (University of Manitoba) for assistance with animal trapping, and Manitoba Conservation for providing permission to trap animals.
- Copyright © 2013 the American Physiological Society