Survival of near-freezing body temperatures and reduced blood flow during hibernation is likely the result of changes in the expression of specific genes. In this study, we described a comprehensive survey of mRNAs in the heart of the thirteen-lined ground squirrel (Spermophilus tridecemlineatus) before and during hibernation. The heart was chosen for this study because it is a contractile organ that must continue to work despite body temperatures of 5°C and the lack of food for periods of 5–6 mo. We used a digital gene expression assay involving high-throughput sequencing of directional cDNA libraries from hearts of active and hibernating ground squirrels to determine the identity and frequency of 3,532 expressed sequence tags (ESTs). Statistical analysis of the active and hibernating heart expression profile indicated the differential regulation of 48 genes based on a P ≤ 0.03 threshold. Several of the differentially expressed genes identified in this screen encode proteins that likely account for uninterrupted cardiac function during hibernation, including those involved in metabolism, contractility, Ca2+ handling, and low-temperature catalysis. A sampling of genes showing higher expression during hibernation includes phosphofructokinase, pancreatic triacylglycerol lipase, pyruvate dehydrogenase kinase 4 (PDK4), aldolase A, sarco(endo)plasmic reticulum Ca2+-ATPase 2a (SERCA2a), titin, and four-and-a-half LIM domains protein 2 (FHL2). Genes showing reduced levels of expression during hibernation include cyclin-dependent kinase 2-associated protein 1 (CDK2AP1), troponin C, phospholamban, Ca2+/calmodulin-dependent protein kinase II (CaMKII), calmodulin, and four subunits of cytochrome c oxidase.
- cardiac genes
- expression profiling
several species of hibernating mammals survive the cold, foodless months of temperate-zone winters by depressing metabolic rates to a mere 2–4% of active values. Upon arousal from hibernation during early spring, these animals display fewer signs of atrophy associated with the long-term disuse of muscle systems seen in nonhibernating species and display no signs of ischemia-reperfusion injuries (for a review, see Ref. 10). Interestingly, hibernating species are found scattered throughout various orders of class Mammalia. Current research suggests that physiological adaptations that occur in hibernators just before and during torpor are caused by changes in the expression of specific genes common to all mammals rather than through the activation of a set of genes unique to the hibernating species (4, 33, 35).
In this study, a cDNA-based digital gene expression assay was performed to identify genes that are differentially regulated during hibernation in the heart of the thirteen-lined ground squirrel, Spermophilus tridecemlineatus. This assay was performed by randomly sequencing over 4,000 cDNA clones from two nonnormalized cDNA libraries: one library created from the hearts of active ground squirrels and the other from hearts of hibernating animals. The expressed sequence tags (ESTs) were then identified via the BLASTn local alignment program (1), and those ESTs matching or exceeding specific alignment criteria were tabulated to determine the total number for each gene in each physiological state. The probability that specific genes were differentially expressed between the two physiological conditions was calculated using a statistical test designed to detect differential regulation in digital gene expression assays (3). At least one representative EST sequence from every unique gene identity as determined by the alignment criteria for positive gene identification was submitted to the database for ESTs [dbEST (6)].
Our analysis indicated that 48 genes are differentially regulated in the hibernating heart; 37 genes showed higher expression in active animals, and 11 genes indicated higher expression during hibernation. A complex heart function model has developed as a result of the assay, indicating a role for these differentially expressed genes in altering contractility, glucose metabolism, and Ca2+ handling in the hibernating heart.
MATERIALS AND METHODS
The animal care and use described were approved by the Institutional Animal Care and Use Committee of the University of Minnesota at the time of publication. All male and female thirteen-lined ground squirrels were obtained in the summer from TLS Research (Bartlett, IL) within 3–4 days of wild capture or live trapped in either Mount Pleasant, MI, or Moorhead, MN. Squirrels were housed individually in plastic top-load cages filled with pine shavings and fed standard laboratory rodent chow (Purina no. 5001) supplemented with black oil sunflower seeds and water ad libitum. The squirrels were maintained in a temperature-controlled environment where food availability and light-dark photoperiods were altered artificially to mimic environmental conditions that induce hibernation. In August, the squirrels were housed at 23°C, in September at 17°C, in October at 11°C, and from November to late March at 5°C. Light was maintained at a 12:12-h light-dark cycle from August until November 1, when the squirrels were housed in complete darkness and all food was removed with water available ad libitum. These conditions were sufficient for all squirrels to hibernate. In late March, spring arousal was stimulated by increasing the housing temperature to 23°C, reintroducing food, and reinstating the 12:12-h light-dark cycle.
The hibernating state of thirteen-lined ground squirrels was monitored from October through March using the sawdust technique (29), and organs and tissues were harvested at times that reflected different physiological states based on animal activity, body temperature, ambient temperature, and month of the year. Squirrels were killed via decapitation, and body temperatures were measured rectally. Harvested organs and sera were placed in cryogenic vials and quick frozen by liquid nitrogen submersion. Frozen organs and serum samples were kept in liquid nitrogen tanks or −80°C freezers for long-term storage.
Heart RNA was isolated using the ToTALLY RNA Kit (Ambion; Austin, TX). Whole heart samples were disrupted using an electronic razor-stator tissue homogenizer (Kinematica Polytron, Brinkmann Instruments; Westbury, NY). Phase Lock Gel heavy tubes (Eppendorf; Westbury, NY) were used to facilitate phase separation and minimize protein and genomic DNA contamination. RNA was dissolved in an appropriate amount of diethylpyrocarbonate (DEPC)-treated water-EDTA (0.1 mM EDTA, ∼200 μl) and aliquoted for storage at −80°C. The concentration of RNA samples was determined by absorbance at 260 nm; 260-to-280-nm absorbance ratios were calculated to determine RNA purity. Denaturing gel electrophoresis was performed for all samples according to the ToTALLY RNA Kit protocol to establish RNA integrity. Briefly, 3× volumes of formaldehyde load dye provided by the kit were added to RNA samples. These samples were heated at 70°C for 15 min and then loaded onto a 1% agarose gel containing a final concentration of 1× MOPS running buffer [0.4 M MOPS (pH 7.0), 0.1 M sodium acetate, and 0.01 M EDTA] and 37% formaldehyde. The gel was stained with ethidium bromide and viewed over an ultraviolet transilluminator.
cDNA library construction.
Two nonnormalized, directional cDNA libraries were created from S. tridecemlineatus heart tissue via the SuperScript Plasmid System for cDNA synthesis and plasmid cloning (GIBCO-BRL Life Technologies/Invitrogen; Carlsbad, CA); the hibernating library was generated with heart RNA from three male and three female torpid [body temperature (Tb) ≈ 5°C] ground squirrels in December, and the active library was generated with heart RNA from three male and three female active (Tb ≈ 37°C) ground squirrels in August. First-strand cDNA synthesis was performed using SuperScript II Reverse Transcriptase and a NotI-poly(A) primer adapter to add a Not I restriction endonuclease cloning site. Second-strand cDNA synthesis was catalyzed by Escherichia coli DNA polymerase I, E. coli RNase H, and E. coli DNA ligase. T4 DNA polymerase was added to the reaction to create blunt ends, after which SalI adapters (precut at a SalI restriction enzyme recognition site) were added to both ends of the cDNA molecules. The cDNA molecules were size fractionated by column chromatography to reduce the likelihood of cloning smaller (≤500 bp) inserts. NotI restriction enzyme digestion of cDNAs and ligation to NotI- and SalI-digested pSPORT1 vectors ensured a 5′ to 3′ orientation of each cDNA with respect to the plasmid vector. Recombinant clones were selected via blue-white colony selection after transformed bacteria were plated on Luria-Bertani (LB) agar plates containing 100 μg/ml ampicillin, 100 μM isopropyl-β-d-thiogalactopyranosid (IPTG), and 0.002% Xgal. These recombinant colonies were isolated and frozen at −80°C in 96-well plates containing LB media with 100 μg/ml ampicillin and 15% glycerol.
cDNA propagation and sequencing.
cDNA clones were propagated for sequencing by transferring recombinant E. coli colonies via a 96-well prong stamp to 96-well growth blocks (Qiagen; Valencia, CA) containing 1.3 ml LB + 100 μg/ml ampicillin per well. Bacteria were grown at 37°C in a shaking incubator at 300 rpm for 20 h. Growth blocks were then centrifuged at 4°C for 10 min at 3,500 g to pellet the cells. Growth media were decanted, and plasmid preparations were performed in a 96-well format using the REAL Prep 96 Plasmid Kit (Qiagen) in conjunction with the Qiagen 8000 BioRobot. Plasmid preparations were then subjected to sequencing reactions using BigDye terminator cycle sequencing chemistry (Applied Biosystems; Foster City, CA) with the Perkin-Elmer GeneAmp 9700 thermal cycler (Perkin-Elmer; Boston, MA) at the University of Minnesota Advanced Genetic Analysis Center. Sequencing of plasmid inserts used the M13/pUC reverse sequencing primer to take advantage of the directional cloning and thus generate the sequence from the 5′-end of all cDNAs. PCR fragments generated from each sequencing reaction were separated via the ABI PRISM 377 DNA Sequencer (Applied Biosystems) according to the manufacturer’s protocol. Sequence traces were then verified for correct base calling, converted to FASTA file format, and trimmed for pSPORT1 and contaminant sequences using Lasergene 99 sequence-analysis software (DNASTAR; Madison,WI) at default settings.
Each cDNA sequence was subjected to BLASTn (version 2.2.6) alignment against the nonredundant nucleotide database (1). cDNA identifications were assigned based on high sequence identity to previously characterized genes submitted to the nonredundant nucleotide database; only alignments generating a BLASTn score ≥50 and an expectation value ≤10−5 were considered positive identifications, and those cDNA sequences that did not meet the identification criteria were omitted from the expression profiles and statistical analysis. The identified cDNAs were considered ESTs of genes transcribed in the heart at the hibernating or active state of tissue collection. ESTs that were identified only as “mitochondrial genome” or “mitochondrial DNA”, rRNA, or E. coli sequences were eliminated from the expression profiles and omitted from overall expressed sequence statistical analysis. A summary of all cDNA and EST counts for the active and hibernating libraries is shown in Table 1.
Differential gene expression probabilities.
Tags of the same gene identity (isoform specific) were counted to determine gene-specific tag frequencies in each library. The Audic and Claverie test (3) was used to establish the statistical significance of the difference in tag frequencies for an individual gene between the active and hibernating heart libraries. The test was computed according to: with where the variables x and y are tag frequencies in the hibernating and active libraries and N1 and N2 are the total number of ESTs identified in the hibernating and active libraries, respectively (23). A website (http://igs-server.cnrs-mrs.fr/∼audic/winflat.cgi) was used to compute the probability of differential regulation. The Audic and Claverie statistical test was designed specifically for predicting differential gene expression based on the comparison of EST counts generated from digital expression analyses and was found to produce critical values within 1.5% of the SAGE300, Fischer’s exact test, and Z-test in side-by-side statistical comparisons of identical serial analysis of gene experssion (SAGE) library data regardless of whether or not the sequenced libraries were of equal or different sizes (31). Additionally, the use of Fischer’s exact test remains somewhat controversial, as values that are presumed to be fixed based on the Fischer’s model are in fact not fixed in an actual EST experiment (31). The Audic and Claverie statistical test also allows the calculation of differential expression probabilities derived from unequal EST pools (3, 31).
A P value of ≤0.03 was chosen as a threshold for determining significant differential gene expression in our assay to accommodate for the increase in predicted false positive rate, or type I error, associated with performing multiple pairwise comparisons. A pairwise comparison was performed on a specific gene if more than one tag was detected between each library. For example, a statistical analysis would be performed if a gene was detected once in each library or if more than one gene-specific EST was collected in either library. Singletons, or gene-specific ESTs that were detected only once among the two libraries, were omitted from pairwise comparisons but were included in the overall number of ESTs for their respective library. The predicted false positive rate of the pairwise comparison of the 304 gene tag counts at a P value threshold of ≤0.03 is 9.12 and was chosen as a compromise between the increase in potential type I error and type II error, the predicted false negative rate associated with multiple pairwise comparisons. It is important to note that the P value in this study was not used to determine the statistical significance of differential regulation or imply fold regulation of a specific gene at the mRNA level within the actual tissue; rather, the probability generated through this statistical analysis reflects a significant discrepancy between gene-specific EST counts that were randomly sequenced from cloned mRNA populations of differing physiological conditions. An overview of the entire digital expression profile assay and its analysis is depicted in Fig. 1.
EST dbEST submission.
All singleton cDNA clone sequences that could be identified based on the BLASTn alignment guidelines stated previously were submitted to the dbEST according to the NCBI protocol (6). In the case of multiple ESTs from the same gene, there was a single dbEST submission that represented the highest score and E value among the gene-specific ESTs collected. All ESTs of any gene that was determined to be differentially regulated based on the Audic and Claverie statistical test were submitted to the dbEST except for S. tridecemlineatus ATP synthase 6, a gene that has been entirely sequenced (GenBank Accession No. AF411434); one representative EST transcript was submitted to the dbEST in this case. Accession numbers for submitted ESTs are shown in the complete active and hibernating digital expression profiles found in Supplemental Tables S1 and S2, respectively (available at the Physiological Genomics web site).1 Those sequences that were represented more than once between the active and hibernating heart libraries were assembled into contigs using the Lasergene 99 software suite (DNASTAR) at default settings. As expected, some gene-specific tag groups produced more than one contig and others produced none due to a lack of overlap, because individual sequencing reactions differ in the length and region of an individual clone. A total of 114 unique contigs was created and can be accessed at the following URL: http://www.d.umn.edu/∼mandrews/HeartExpressionContigs.htm.
Serum glucose determinations.
Whole blood collected at death was spun at 3,500 g for 10 min, and partially purified serum was transferred to 1.5-ml tubes and spun at 3,500 g for 15 min at 4°C. Serum was divided into aliquots, frozen in liquid nitrogen, and stored at −80°C until the time of assays. Serum was thawed on ice, and glucose concentrations were obtained using Infinity Glucose Reagent (Sigma Diagnostics; St. Louis, MO). Three microliters of ground squirrel serum were mixed with 300 μl of reconstituted reagent, in triplicate, and added to a 96-well plate. Samples were incubated for 3 min at 37°C and immediately analyzed by spectrophotometry at a wavelength of 360 nm. A glucose/urea nitrogen standard (Catalog No. 16-300, Sigma) was used to calibrate serum samples. Tris buffer (pH 7.6) was used as a negative control. The concentration of glucose was determined by dividing the absorbance of the unknown sample by the absorbance of the glucose/urea nitrogen standard and multiplying that value by the calibrator value (5.56 mmol/l).
We used high-throughput sequencing of cDNAs to generate a gene expression profile for the hearts of active and hibernating thirteen-lined ground squirrels. Directional cloning allowed us to position our sequencing primer proximal to the 5′-end of the cDNA inserts in both the active and hibernating libraries. Sequence generated from the 5′-end of the cDNAs increased the likelihood of identifying open reading frames and avoided difficulties associated with sequencing through poly-A regions and generating sequence from lengthy, and often uninformative, 3′-untranslated regions. This approach of confining sequence reactions to the 5′-end allowed for easy alignment and identification of the cDNAs using the BLASTn algorithm (1). Identification of the protein coding potential of the cDNA was vital for inferring the functional significance of the expression profile.
In total, 1,283 unique, isoform-specific GenBank identifications were made from a total of 3,532 ESTs based on BLASTn scores of ≥50 and expectation values of ≤10−5 identification thresholds. Of these gene identifications, 1,188 ESTs were represented only once in either the active and hibernating cDNA libraries, and 2,344 ESTs were identified as genes that were represented more than once in both libraries. A summary of the distributions of all cDNA data is shown in Table 1.
The complete digital gene expression profiles from active and hibernating hearts are shown in Supplemental Tables S1 and S2, respectively. Both expression profiles specify the orthologous GenBank Accession number used to identify all or a portion of each EST. In addition, both tables show the S. tridecemlineatus dbEST accession number(s) for each gene, the BLASTn identification score or score range based on alignment to the nonredundant nucleotide database, and the total number of gene-specific ESTs in the respective library. The hibernating expression profile consisted of 2,186 individual ESTs. Of these, 909 recovered from the hibernating population were identified only once (singletons). The active expression profile is comprised of a total of 1,346 ESTs, 279 of which were singletons recovered from the active population.
Statistical analysis of the hibernating and active gene expression profiles indicated that 48 of 304 pairwise comparisons were significantly different based on a P ≤ 0.03 threshold. This value was chosen as a compromise to limit increases in type I error due to the multiple pairwise comparisons performed on the data and to accommodate increases in type II, or false negative, errors. This P ≤ 0.03 threshold reflects the probability of differential gene expression in the active and hibernating ground squirrel heart. Eleven genes were found to be expressed at higher levels during hibernation based on statistical comparison of hibernating and active EST counts, and 37 genes were found at higher levels in active animals. Summaries of the tag count data for these differentially regulated genes are shown in Tables 2 and 3 for elevated expression in active and hibernating animals, respectively. Information regarding the GenBank and dbEST Accession numbers and BLASTn-derived score ranges can be referenced in the complete active and hibernating expression profiles, shown in Supplemental Tables S1 and S2, respectively.
Several genes showing differential regulation were confirmed as differentially expressed at the RNA level based on previous Northern blot or quantitative RT-PCR studies. For example, pyruvate dehydrogenase kinase isoform 4 (PDK4) mRNA and protein showed higher levels in S. tridecemlineatus hearts during hibernation versus August active levels (2, 8). Upregulation of pancreatic triacylglycerol lipase (PTL), as shown by our digital screen, was also confirmed in the same species by a fourfold increase in mRNA and a concomitant increase in PTL protein and enzymatic activity in the heart during hibernation compared with the active state (2, 34). The α-isoform of myosin heavy chain (α-MHC) was found to have higher expression during hibernation in our screen, an observation supported by the increased expression of α-MHC protein in hibernating versus active European hamster heart ventricles (26). Sarco(endo)plasmic reticulum Ca2+-ATPase 2a (SERCA2a), a protein responsible for the reuptake of Ca2+ from the cardiomyocyte into the sarcoplasmic reticulum, was found to be upregulated at the RNA and protein levels in hibernating woodchuck (Marmota monax) hearts compared with active counterparts (40). Also, in agreement with our study, phospholamban (PLB) was found to be downregulated during hibernation at both the mRNA and protein levels (40). Finally, NADH-ubiquinone oxidoreductase subunit 2, a gene showing upregulation in hibernators based on our screen, displayed twofold higher levels in Spermophilus lateralis hibernating hearts via Northern blot hybridization (13).
Phosphofructokinase (PFK) and aldolase A, two enzymes catalyzing sequential reactions in the metabolism of glucose, showed higher mRNA levels during hibernation in our screen (Table 3). PFK catalyzes the rate-limiting step in glycolysis, and aldolase generates the initial 3-carbon intermediates that eventually lead to the formation of pyruvate. An increase in the expression of genes encoding these two enzymes was surprising because hibernators survive prolonged periods without food by minimizing carbohydrate oxidation (for a review, see Ref. 19). However, Dark and Miller (11) have shown that hibernating ground squirrels rely heavily on glucose oxidation during periodic arousal from torpor, particularly during the rewarming phase. We hypothesized that the transcriptional upregulation of PFK and aldolase A would poise the heart to utilize glucose when needed during interbout arousals (IBAs). In this scenario, we would predict that an elevation in serum glucose during IBAs would provide the necessary fuel for the transition to euthermic metabolism. We tested this hypothesis by comparing serum glucose levels in hibernating and IBA animals near the beginning (December) and end (March) of the hibernation season. We found that IBA animals showed higher levels than hibernators at both time points (Fig. 2). Amazingly, squirrels that had not been fed for periods up to 4–5 mo (March hibernators) could still elevate their serum glucose levels significantly higher than torpid animals during the same month. This result suggests that the observed increases in cardiac PFK and aldolase A places the hibernating heart in a position to quickly metabolize the higher levels of circulating glucose and thus provide energy for rewarming and other cellular activities during arousal.
In this study, we profiled genes expressed in the hearts of active and hibernating thirteen-lined ground squirrels. The differential expression of specific genes identified in our screen indicated adaptive mechanisms that maintain heart function during hibernation. Understanding how the heart works in natural hibernators can lead to improved prevention and treatment of cardiovascular diseases that are a primary cause of death in humans. Myocardial infarction, for example, produces necrosis when the need for oxygenated blood in heart tissue exceeds the oxygen being supplied by the blood. This necrosis can lead to a number of other heart problems, including arrhythmias, decreased contractility, and, ultimately, heart failure. Analysis of the differential regulation of genes responsible for cardiac function and providing protection during the physiological extremes of hibernation is therefore of important biomedical interest. Because specific organs in hibernators must adjust uniquely to changes in metabolism and decreased temperature and oxygen supply, an organ-specific analysis of differential gene expression is intuitively advantageous.
In this project, high-throughput DNA sequencing was used to assay differential gene expression by randomly sequencing nonnormalized cDNA libraries created from the hearts of active and hibernating thirteen-lined ground squirrels. Sequencing and subsequent identification of cDNA clones allowed a digital count of ESTs that are cloned directly from original mRNA populations, thus providing an estimate of the relative amounts of gene-specific transcripts that could be isolated from a given tissue in a specific physiological state (20, 28). This “digital” approach of assaying differential gene expression altogether bypasses several of the hybridization-associated drawbacks of microarray technologies, such as the normalization procedures necessary for the standardization of gene expression signal data. A distinct advantage of our approach over microarrays is that direct sequencing of cDNAs captures a specific mRNA population and is therefore not limited to those sequences that are immobilized on the array. Although it is not possible to quantify relative levels of expression or differences in gene regulation between individual specimens, our approach eliminates false positives that can occur on microarrays resulting from hybridization with nonexpressed sequences that share sufficient homology with labeled probes. EST sequencing also has advantages over SAGE because the long sequence-read lengths provided by ESTs allow for absolute identification of a cDNA sequence based on local alignments of hundreds of bases rather than the average of 15–20 bases used to identify SAGE tags (21, 36). This advantage is underscored when investigating organisms lacking a sufficient nucleic acid sequence database because SAGE tags can frequently align to common gene motifs present on a variety of different genes, making absolute tag identification unlikely. Our EST-based assay also provides a wealth of additional species-specific gene sequence information, which can lead to the discovery of novel genes or splice variants as well as previously unknown temporal gene expression. Multiple overlapping sequences encoding the same protein may also be assembled to generate cDNA contigs such as those we have posted at http://www.d.umn.edu/∼mandrews/HeartExpressionContigs.htm.
We found significant changes in the level of cDNAs for proteins that handle Ca2+ and increase contractile efficiency under physiological conditions not tolerated by nonhibernators (39). This includes a significant increase in SERCA2a and a significant decrease in its membrane-bound negative regulator PLB during hibernation. These same two proteins have been recently found to be differentially regulated in the hearts of hibernating marmots (40), and their levels suggest enhanced cytoplasmic clearance of Ca2+ during relaxation, possibly accounting for the uninterrupted rhythmic contractility in hibernators at temperatures that cause rat hearts to fail (18). This differential gene expression provides an explanation for the increased rate of Ca2+ reuptake and larger calcium stores observed in the hibernating Spermophilus richardsonii sarcoplasmic reticulum (5). It may also partially explain the similar calcium transients that have been recorded for active and hibernating woodchucks despite a decrease in the amount of calcium entering the hibernating myocyte to trigger sarcoplasmic release of Ca2+ via the ryanodine receptor (40). A higher density of SERCA2a in the sarcoplasmic reticulum membrane not only increases Ca2+ uptake, but the concomitant hydrolysis of ATP also provides a mechanism for regional endothermy (12) despite core body temperatures of 5–6°C during hibernation. Ca2+/calmodulin protein kinase II and its activator calmodulin are two genes found in this study to be expressed at higher levels in active ground squirrels. When Ca2+/calmodulin protein kinase II is activated by Ca2+-bound calmodulin, the kinase phosphorylates PLB (32), inactivating the protein and enhancing SERCA2a uptake of Ca2+ into the sarcoplasmic reticulum (16).
The rate-limiting glycolytic enzyme PFK and the enzyme that catalyzes the next step in the pathway, aldolase A, showed increased mRNA levels during hibernation based on our screen. This potential enhancement of glycolysis was surprising because hibernators show both a metabolic rate reduction and a long-term reliance on lipids as the primary source of fuel. However, serum glucose measurements (shown in Fig. 2) suggest that these enzymes would be poised to catabolize the elevated levels of glucose that are available during the periods of increased metabolic activity seen during IBAs. Aldolase A generates two 3-carbon intermediates that can eventually lead to the formation of pyruvate and/or the glycerol moiety of triacylglycerols that accumulate in the heart of hibernators (9). Lipolysis of triacylglycerols is achieved in the hibernating heart at temperatures as low as 0°C (2, 34) by another upregulated enzyme, PTL. PDK4 has been shown in our screen and elsewhere (2, 8) to be upregulated during hibernation. PDK4 reduces carbohydrate oxidation by blocking the conversion of pyruvate to acetyl-CoA so that glycolytic products cannot enter the tricarboxylic acid (TCA) cycle.
We propose that rewarming during IBA may further increase activity of PFK and aldolase A, thereby increasing pyruvate production from glucose. PDK4 activity is influenced by the production of pyruvate, acetyl-CoA, and NADH; pyruvate accumulation acts by inactivating the enzyme and promoting glucose utilization, whereas acetyl-CoA and NADH production activates PDK4 and inhibits glucose oxidation (for a review, see Ref. 15). Temporary inactivation of PDK4 during arousal by declining acetyl-CoA-to-CoA and NADH-to-NAD+ ratios may allow additional energy production via glucose-derived carbon utilization via the TCA cycle, complementing ATP production by β-oxidation of fatty acids during this time of high energy demand. Once euthermic temperatures have been achieved, accumulating acetyl-CoA and NADH may work to once again activate PDK4, allowing the tissue to decrease glucose oxidation and once again resume fatty acid breakdown for required energy in the heart. Additionally, subunit B of lactate dehydrogenase, an enzyme involved in the conversion of pyruvate to lactate, and the β-subunit of phosphorylase kinase, an inhibitory subunit of the enzyme responsible for the breakdown of glycogen into glucose (for a review, see Ref. 7), were both found to have decreased expression during hibernation in our screen.
α-MHC is elevated during hibernation and confers a higher shortening velocity and greater ATPase activity than β-MHC (for a review, see Ref. 25). Interestingly, hibernators have demonstrated increased contraction amplitude, or percent cell shortening, in myocytes during torpor (38). Different ratios of the two MHC isoforms can contribute to the differing contractile properties of various muscles. Ventricular myosin light chain (VLC-1) was observed to be downregulated during hibernation in our screen. The exact function of VLC-1 has remained obscure; however, studies aimed at weakening the actin interaction with VLC-1, which creates a tether between MHC and actin, observed increased force production in human ventricular fibers (27) and increased myofibrillar ATPase activity (30). Troponin C was additionally found to be downregulated during hibernation in our study; interestingly, this protein is responsible for changing the conformation of the entire troponin complex in response to calcium, which allows myosin cross-bridges to interact with actin (for a review, see Ref. 22).
Titin, an extremely large protein with a molecular mass of ∼3,000 kDa, was upregulated at the mRNA level in our digital assay. Titin is an important determinant in diastolic filling of the heart because it serves as a molecular spring that gives rise to elasticity and passive muscle stiffness (14) that may be critical to heart function at extremely low temperatures. A recent study (37) has suggested that the 3,400-kDa isoform of titin is expressed predominanatly during hibernation in both skeletal and cardiac muscle of Citellus undulatus. Four-and-a-half LIM domains protein 2 (FHL2), an adaptor protein shown to bind creatine kinase, adenylate kinase, and PFK to titin in the rat heart (17), showed increased RNA expression during hibernation in our screen. In the heart, it has been proposed that FHL2 concentrates ATP-producing enzymes in the vicinity of the contractile apparatus (17). Telethonin, a protein thought to function in myofibrillogenesis and sarcomere organization (24), was found to have greater expression levels in active animals based on our screen.
In summary, our cDNA-based strategy for assaying the differential expression of heart genes between the active and hibernating state has provided an informative approach to study the molecular basis of hibernation in mammals. Our results offer new insight into the biochemical mechanisms that underlie cardiac contractility, metabolism, and improved Ca2+ handling. Further research will undoubtedly clarify the phenotypic ramifications of these differences in RNA expression and possibly reveal potential therapeutic targets for heart disease and related illnesses. cDNA sequences generated in this project can also serve as genetic markers for the recently constructed bacterial artificial chromosome library of the the S. tridecemlineatus genome (www.genome.gov/10001852).
This study was funded by United States Army Research Office Grant DAAD19-01-1-0014 and National Heart, Lung, and Blood Institute Grant HL-081100.
The authors thank M. Tredrea for assistance in experimental design, A. Klein for assistance with the glucose assays, I. Efimov and K. Russeth for insightful discussion and review of the manuscript, members of the Andrews Lab for expression profile organization and contig assembly, and the Visualization and Digital Imaging Lab (University of Minnesota-Duluth) for computer and software access and assistance.
↵1 The Supplementary Material for this article (Supplemental Tables S1 and S2) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00076.2005/DC1.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: M. T. Andrews, Dept. of Biology, Univ. of Minnesota, Duluth, MN 55812 (e-mail:).
- Copyright © 2005 the American Physiological Society