During endurance training, exercising skeletal muscle experiences severe and repetitive oxygen stress. The primary transcriptional response factor for acclimation to hypoxic stress is hypoxia-inducible factor-1α (HIF-1α), which upregulates glycolysis and angiogenesis in response to low levels of tissue oxygenation. To examine the role of HIF-1α in endurance training, we have created mice specifically lacking skeletal muscle HIF-1α and subjected them to an endurance training protocol. We found that only wild-type mice improve their oxidative capacity, as measured by the respiratory exchange ratio; surprisingly, we found that HIF-1α null mice have already upregulated this parameter without training. Furthermore, untrained HIF-1α null mice have an increased capillary to fiber ratio and elevated oxidative enzyme activities. These changes correlate with constitutively activated AMP-activated protein kinase in the HIF-1α null muscles. Additionally, HIF-1α null muscles have decreased expression of pyruvate dehydrogenase kinase I, a HIF-1α target that inhibits oxidative metabolism. These data demonstrate that removal of HIF-1α causes an adaptive response in skeletal muscle akin to endurance training and provides evidence for the suppression of mitochondrial biogenesis by HIF-1α in normal tissue.
- skeletal muscle
- gene regulation
one of the greatest challenges facing skeletal muscle is the need to match energy production with energy demand during exercise. The need to maintain optimal metabolic flux within the muscle is of such great importance that it is the most plastic aspect of muscle physiology—muscles have the ability to adopt different metabolic profiles depending on the type of exercise they are most frequently called upon to perform. The ability of the skeletal muscle to keep up with increased energy demand can be greatly limited by the availability of oxygen to the tissue. In fact, exercising muscle tissue exhibits very low oxygen tensions relative to what is available in inspired air and circulating blood (25), indicating that exercising muscle is subject to severe and repeated oxygen stress.
The primary response mechanism to low oxygen concentrations in a cell is the transcription factor hypoxia inducible factor 1 (HIF-1), a heterodimeric, basic helix-loop-helix, PAS domain-containing transcription factor. The regulated subunit of HIF-1, HIF-1α, is controlled by oxygen levels within the cell; it is hydroxylated and degraded under normoxia but is stable and translocates to the nucleus under hypoxia (16). Activated HIF-1α turns on transcription of genes; this, in turn, aids cells and tissues in coping with oxygen stress. Targets include genes responsible for glycolysis, glucose uptake, and angiogenesis (27). Recent studies have examined the role of HIF-1α in the skeletal muscle. In one study, Pisani and Dechesne (23) found elevated levels of HIF-1α protein at rest under normoxia in the skeletal muscles of untrained mice, in a fiber-type-dependent manner. Ameln and colleagues (1) demonstrated that acute exercise leads to stabilized HIF-1α protein. Finally, we have previously shown that deletion of HIF-1α in the skeletal muscle of mice leads to mice with impaired glycolytic flux, enhanced aerobic metabolism, and increased submaximal endurance (20). These studies have highlighted the importance of HIF-1 signaling in the skeletal muscle in the untrained state.
A muscle's ability to improve through repeated exercise is essential to its role. The primary way through which endurance athletes improve their performance is by regular aerobic exercise, with an end result that muscles adapt through increased oxygen delivery and more efficient energy utilization. Some of the major physiological changes seen as a result of endurance training are increased muscular capillary density (24), increased mitochondrial volume (15), increased reliance on fatty acid metabolism for energy, a shift in fiber type toward oxidative fibers, decreased lactate production (13), and increased hexokinase enzyme activity (28). Because several of these changes are potentially regulated by HIF-1, we have investigated the hypothesis that HIF-1α is a necessary factor in the skeletal muscle's response to endurance training. We have trained mice lacking HIF-1α in their skeletal muscle and found that contrary to being unable to train, their muscles have already undergone an adaptive response that better prepares the muscles for endurance exercise. These changes are only matched by wild-type (WT) mice following endurance training and are linked to a constitutively activated AMP-activated protein kinase response.
MATERIALS AND METHODS
Mouse strains and crosses.
All procedures involving animals were approved by the University of California San Diego Animal Care Committee, which serves to ensure that all federal guidelines concerning animal experimentation are met. Skeletal muscle-specific HIF-1α null mice were generated from a cross of a C57BL6/J strain containing the MCK/cre transgene, as described previously (4), with a C57BL6/J strain homozygous for the HIF-1α loxP-flanked allele. WT mice were littermates that were homozygous for the loxP-flanked HIF-1α allele but did not carry the MCK-Cre transgene.
Endurance training and endurance assessment.
WT (n = 15) and HIF-1α null (n = 13) mice were trained on an enclosed chamber modular treadmill (Columbus Instruments, Columbus, OH) held at a constant 5% incline. Mice trained by running for 30 min at 18 m/min, 5 days/wk for 6 wk. Following completion of the training protocol, mice were given 2 days to recover before an endurance test. During the endurance test, trained (WT, n = 7; HIF-1α null, n = 7) and untrained (WT, n = 11; HIF-1α, null n = 6) mice ran on the treadmill at a 5° incline with an initial velocity of 12 m/min. After 5 min, velocity was increased to 16 m/min for 5 min with 4 m/min increases every 5 min until the mice reached 28 m/min. Velocity was then held constant at 28 m/min until exhaustion, which was determined as the point at which mice no longer responded to a low-voltage electrical grid at the back of the treadmill. Metabolic monitoring during the run was carried out by measuring O2 and CO2 levels in gas going into and leaving the treadmill using the Paramax O2 sensor and a CO2 sensor (Columbus Instruments). Gas exchange data were analyzed using Oxymax software (Columbus Instruments).
Whole blood analysis.
Whole blood was collected from trained (WT, n = 11; HIF-1α null, n = 11) and untrained (WT, n = 10; HIF-1α null n = 8) mice at rest and immediately following exercise via cardiac puncture. Blood was stored in EDTA citrate tubes (Terumo Medical, Elkton, MD) until analysis by the University of California, San Diego, Animal Care Program Diagnostic Laboratory (La Jolla, CA).
Histological analysis of skeletal muscle.
Gastrocnemius sections were harvested from untrained and trained mice of both genotypes and frozen in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA). Sections were cut 10 μm thick for histological analysis. Neuromuscular junction staining was carried out using a FITC-conjugated α-bungarotoxin probed (Molecular Probes, Eugene, OR). CD31 staining was carried out using a rat anti-mouse CD31 primary antibody (BD Biosciences, San Jose, CA) and a FITC-conjugated goat anti-rat secondary antibody (Pierce Biotechnology, Rockford, IL). Briefly, slides were fixed for 10 min in cold (−20°C) acetone before a 10-min rinse in PBS with 0.1% Tween-20 (PBS-T). Blocking was for 1 h in PBS-T supplemented with 3% bovine serum albumin and 10% normal goat serum (Sigma, St. Louis, MO) before a 1-h room temperature primary antibody incubation in blocking solution. Slides were then rinsed 3 times for 5 min in PBS-T before a 1-h room temperature incubation with the secondary antibody in the blocking solution. Following the secondary antibody, slides were rinsed 3 times for 5 min in PBS-T, incubated for 1 min in 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) (Sigma), rinsed once more in PBS-T for 5 min, and mounted in Vectashield mounting medium for fluorescence (Vector Laboratories, Burlingame, CA).
Measurement of mitochondrial DNA and gene expression.
Mitochondrial DNA (mtDNA) was extracted from quadriceps muscle of untrained and trained mice (n = 8 for all genotypes and conditions) using a DNeasy kit for extraction of DNA from tissue (Qiagen, Valencia, CA). DNA concentration was determined by spectrophotometry, and mtDNA was then quantified via real-time PCR. For gene expression, total RNA was isolated from quadriceps muscle from untrained and trained rested and exercised mice (minimum n = 4) using TRIzol reagent (Invitrogen, Carlsbad, CA), treated with DNase I (Invitrogen) to remove DNA, and then reverse transcribed using the Superscript III first-strand synthesis system for RT-PCR (Invitrogen). Primer sequences used were as previously described (16a, 20, 28a). Real-time PCR was carried out on an ABI Prism 7700 Sequence Detector (Applied Biosystems, Foster City, CA) using platinum quantitative-PCR supermixes for SYBR Green or Taqman real-time PCR (Invitrogen). Conditions for the PCR were 40 cycles of 95°C for 15 s (denaturing) and 60°C for 1 min (anneal/extend).
Analysis of enzyme activity.
Measurements of enzyme activity levels in untrained and trained mice were made from gastrocnemius muscles taken immediately after the endurance test. Enzymes were extracted and analyzed spectrophotometrically. Units of enzyme activity were normalized to total protein concentration using a bicinchoninic acid protein quantification kit (Pierce Biotechnology).
Primary myoblasts were isolated from freshly excised gastrocnemius muscles of mice homozygous for LoxP-flanked HIF-1α. Briefly, muscles were minced and digested in 0.2% collagenase II (Worthington Biochemical, Lakewood, NJ) in HBSS prior to being passed through a 100-μm filter. Cells were then allowed to differentially adhere to tissue culture plates for 2 h. Nonadherent cells were then pulled off, plated onto collagen-coated plates, and grown in DMEM high glucose supplemented with 10% FBS. Deletion of HIF-1α was achieved through infection with adenovirus containing the Cre recombinase transgene, and deletion efficiency (85–90% deletion) was measured via real-time PCR. Control cells were infected with adenovirus particles encoding for the βgal transgene. Cultured myoblasts were able to differentiate into myotubes expressing myosin heavy chain (data not shown), confirming that the harvested cells were indeed myoblasts. Measurements of oxygen consumption in control and HIF-1α null myoblasts were carried out using an Oxytherm electrode unit (Hansatech, Norfolk, UK) in a 0.5-ml volume in a sealed measurement chamber.
Western blot analysis.
For Western blot analysis of phospho-AMPK and phospho-acetyl-CoA carboxylase (ACC) levels, homogenates were made from quadriceps muscle from untrained resting (WT, n = 6; HIF-1 null, n = 5) and exercised (WT, n = 6; HIF-1 null, n = 6) mice. Exercise consisted of a timed run on an enclosed modular treadmill for 30 min at 24 m/min. Mice were killed, and tissue was harvested immediately after the run to prevent any time effects on AMPK phosphorylation. Muscles were homogenized in a buffer containing 20 mM HEPES, 1 mM EDTA, 1 mM Na3VO4, 2 mM EGTA, 10 mM MgCl2, 50 mM glycerophosphate, 1 mM DTT, 1 mM PMSF, 1% Triton X-100, 10% glycerol, and 1 × EDTA-free complete protease inhibitor (Roche Diagnostics, Indianapolis, IN). Following homogenization with a Polytron tissue homogenizer, samples were centrifuged for 30 min at 15,000 g. Western blot analysis was then performed on the supernatants for phosphorylated and total AMPK and ACC using antibodies from Cell Signaling Technologies (Danvers, MA).
Statistical analyses (unpaired Student's t-test and Mann-Whitney U-test) were carried out using Prism software (GraphPad Software, San Diego, CA). Cell lysates were probed with a PDK1 antibody from Stressgen/Noventa (San Diego, CA).
Response of WT and HIF-1α null mice to endurance training.
The creation and initial characterization of the skeletal muscle-specific HIF-1α null mouse has been described previously (20). To study the role of HIF-1α in the response of skeletal muscle to endurance training, WT and MCK-Cre expressing mice were trained on a treadmill for 6 wk before being given an endurance test to assess improvement. Surprisingly, as can be seen in Fig. 1A, both WT and HIF-1α null mice demonstrated a dramatic improvement in running ability following training, compared with their untrained littermates. Untrained WT mice were able to run for an average of 41.13 ± 5.03 min before exhaustion, and untrained HIF-1α null mice ran for an average of 44.42 ± 5.59 min. Intriguingly, improvement following training was similar in mice of both genotypes. Trained control mice were able to run for 67.16 ± 9.85 min, and trained HIF-1α null mice for 78.08 ± 10.65 min, improvements of 63.30% and 75.79%, respectively (Fig. 1B). These results clearly indicate that HIF-1α is not required for improvement following endurance training, contradicting our hypothesis. Despite the similar improvements in overall running ability, metabolic monitoring of mice during the endurance test revealed that WT and HIF-1α null mice responded differently to the training protocol. As mentioned earlier, a common result of endurance training is an increase in the ability of the muscles to use fatty acids for energy. Values for respiratory exchange ratio (RER) in WT mice during the endurance test show that at lower velocities, trained WT mice have a lower RER relative to their untrained littermates, indicating that they are indeed able to shift toward a more oxidative metabolic profile (Fig. 1C). Conversely, this shift was absent in the trained HIF-1α null mice; their RER values after training were nearly identical to their untrained values at all velocities (Fig. 1C). Comparison of RER values from untrained WT and HIF-1α null mice show that loss of HIF-1α has caused an adaptive response in the null muscles. As shown in Fig. 1C, untrained HIF-1α null mice already have a decreased RER at lower velocities, relative to untrained WT mice. This finding indicates that loss of HIF-1α gives rise to a phenotype mimicking exercise training.
No changes in blood oxygen carrying capacity following training.
Following training, whole blood was extracted for analysis from trained mice and their untrained littermates via cardiac puncture. As shown in Table 1, there were no initial differences in red blood cell count, hemoglobin concentration, or hematocrit between untrained WT and HIF-1α null mice. Additionally, none of these parameters changed following training in mice of either genotype, indicating that the differences in RER were due to changes in the skeletal muscle and not due to changes in the blood oxygen carrying capacity.
Training-induced changes in muscle morphology.
Following the findings of increased endurance in all mice following training, we set out to determine what caused the similar level of improvement between the two genotypes. Additionally, we looked to see what caused the RER change in the untrained HIF-1α null mice and how the trained WT mice were able to match those parameters. As one common result of endurance training is an increase in intramuscular metabolite storage, we stained gastrocnemius samples to detect stored glycogen using a PAS stain and stored triglycerides using an Oil Red O stain. No effects of training were seen on intramuscular glycogen or triglyceride storage in either WT or HIF-1α null mice (data not shown). Therefore, the improvements in mice from both genotypes following training were not a result of increased intramuscular metabolite storage. Endurance training has been shown to increase the abundance of nicotinic ACh receptors at the neuromuscular junction (8); however, α-bungarotoxin staining of gastrocnemius sections from mice involved in this study showed no changes in this parameter either (data not shown).
One of the most pronounced changes in skeletal muscle morphology as a result of endurance training is a shift in fiber-type profile: from glycolytic fibers to more aerobic fibers (13). Representative fields of myofibrillar ATPase staining from deep within the gastrocnemius muscle are shown in Fig. 2, A–D, and show that WT mice had a typical training-induced shift in fiber type. Muscle from WT mice had a significant increase of 36.4% in type I (slow-twitch oxidative) fibers and a significant 17.2% decrease in type IIB (fast-twitch glycolytic) fibers from deep sections of gastrocnemius muscle (Figs. 2, E–G). Interestingly, no significant changes were seen in the fiber-type composition of trained HIF-1α null muscles. However, trained WT mice did have a significantly greater prevalence of type IIA (fast-twitch oxidative) fibers and lower abundance of type IIB fibers relative to trained HIF-1α null mice. No changes were seen in fiber-type distribution of superficial portions of the gastrocnemius muscle (Table 2). These data indicate that while the WT muscles responded as expected in this parameter, HIF-1α null muscles were deficient in the major morphological changes typically thought necessary to achieve improvement following training.
Increases in capillary density of gastrocnemius muscle.
Another common way muscles change to meet the demands of endurance training is to increase capillary density. To measure this parameter in our mice, we stained gastrocnemius sections for the endothelial cell-specific marker CD31. Representative pictures of CD31-stained deep gastrocnemius sections from untrained and trained mice are shown in Figs. 3, A–D. In superficial gastrocnemius sections, WT muscles demonstrated a trend (P < 0.10) toward an increased capillary-to-fiber ratio. This increase was much more pronounced in the deep portions of the gastrocnemius muscle, as WT mice had a significant (P < 0.05) increase from 1.26 ± 0.04 capillaries/fiber in the untrained state to 1.46 ± 0.04 capillaries/fiber following training (Fig. 3E, Table 2). There were no training-induced increases in the capillary/fiber ratio in HIF-1α null muscles; however, the deep portions of the untrained HIF-1α null muscles already had elevated capillary/fiber ratios relative to untrained WT muscles. This finding indicates that loss of HIF-1α in the untrained skeletal muscle has caused an adaptive angiogenic response similar to that of endurance training, and the training protocol was unable to further improve upon that adaptation.
Metabolic profile of trained skeletal muscle.
As mentioned before, a hallmark of endurance training is an increase in the oxidative capacity of the skeletal muscle. An increase in oxidative capacity would also be represented by increases in mitochondrial metabolic enzymes. We measured the activities of β-hydroxyacyl-CoA dehydrogenase (βHAD) and citrate synthase (CS), two key aerobic metabolic enzymes localized to the mitochondria. βHAD catalyzes a rate-limiting step of fatty acid β-oxidation, and endurance training caused a significant up-regulation of its activity in the gastrocnemius muscles of exercised WT mice (Fig. 4A). Interestingly, as seen before in these mice (20), untrained HIF-1α null muscles already had elevated βHAD activity relative to untrained WT muscles and did not increase it as a result of training. As the gatekeeper to the citric acid cycle, CS catalyzes the addition of acetyl-CoA to oxaloacetate, producing citrate in a tightly regulated reaction. Similar to βHAD, activity of this enzyme increased as a result of endurance training in exercised WT muscles (Fig. 4B) and was already elevated in the muscles of untrained HIF-1α null mice. Training did not produce any further increase in CS activity in HIF-1α null mice. To determine whether the muscles of the trained mice increased their mitochondrial content, we measured mitochondrial DNA (mtDNA) from quadriceps muscle via real-time PCR. Training induced a dramatic increase of 65.1% in muscles from WT mice, but HIF-1α null muscles only demonstrated a nonsignificant increase of less than 20% (Fig. 4, C and D). However, posttraining levels of WT and HIF-1α null mtDNA were not significantly different, indicating that once again, the HIF-1α null muscles did not need to increase them significantly during endurance training. In fact, the difference in mtDNA from muscles of trained WT mice and untrained HIF-1α null mice is not statistically significant (P > 0.05) either. Thus, the training stimulus was not able to provoke any increase in mtDNA in the HIF-null muscles in order for them to handle the demands of repeated exertion. Here, also the data indicate that an adaptive event has geared the muscles of HIF-1α null mice toward endurance exercise, and endurance training has been unable to improve upon that adaptation.
One glycolytic enzyme that has been demonstrated to be upregulated in response to endurance training is hexokinase (28). In the skeletal muscle, hexokinase has a dual role: catalyzing the first step of glycolysis, as well as phosphorylating glucose that has entered the muscle fiber, resulting in its being unable to leave the muscle and leading to the glucose being either metabolized, or stored as glycogen. Measured levels of hexokinase in both trained WT and trained HIF-1α null exercised muscles increased significantly over that seen in untrained muscles of the same genotype (Fig. 4E). Additionally, both genotypes increased hexokinase activity to the same degree, indicating that this change likely is responsible for their similar increases in endurance.
Changes in gene expression following training.
RNA was harvested from quadriceps muscle taken at rest and immediately following the endurance test from both trained and untrained mice. Real-time PCR was then used to measure changes in gene expression in WT and HIF-1α null mice. No significant effects of training were seen on VEGF or myoglobin mRNA (data not shown). Expression of mitochondrial transcription factor A (TFAM), a key regulator of mtDNA copy number and expression of its encoded genes, was surprisingly unchanged in response to training in WT muscles (Fig. 5). However, HIF-1α null quadriceps exhibited a trend toward increased TFAM expression in the trained state, which correlated with a significant increase in TFAM mRNA seen in the gastrocnemius muscle of untrained HIF-1α null mice (data not shown). Expression of PPAR-γ coactivator-1α (PGC-1α) was not different between WT and HIF-1α null mice in resting muscle, either before or after training (Fig. 5B). Training did result in PGC-1α expression being more responsive to exercise as both trained WT and HIF-1α null muscles demonstrated a trend toward increased PGC-1α mRNA immediately after exercise. Glucose transporter 4 (Glut4) expression, the primary glucose transporter in the muscle, increased significantly following exercise in WT muscle, but was already greatly increased in HIF-1α null muscle, revealing another way in which the HIF-1α null muscles are already adapted for endurance exercise (Fig. 5C). Training did not have any significant impact on Glut4 expression in either WT or HIF-1α null muscles. Untrained WT muscles were able to significantly increase expression of hexokinase II (HKII), the predominant hexokinase isoform in the skeletal muscle, in response to exercise. However, at rest, untrained HIF-1α null muscles had elevated HKII mRNA relative to WT muscles but did not increase this further following exercise (Fig. 5D). No increases in HKII expression were seen in either trained WT or HIF-1α null muscles, indicating that the training-induced changes leading to increase hexokinase enzyme activity in the muscles were posttranscriptional. Finally, expression of pyruvate dehydrogenase 1 (PDK1), a HIF-1-inducible inhibitor of pyruvate dehydrogenase (17, 22), was significantly decreased in the muscles of resting HIF-1α null mice (Fig. 5E). Following training, WT muscles exhibited a decrease in PDK1 expression, but no significant changes were seen in HIF-1α null muscles.
HIF-1α controls O2 consumption in myoblasts in vitro.
To further investigate the importance of PDK1 in skeletal muscle, we performed an in vitro assay examining the role of HIF-1α in upregulating PDK1 in response to hypoxia in the muscle. Primary myoblasts were harvested from gastrocnemius muscles of HIF-1α floxed mice and treated with adenovirus expressing the Cre recombinase transgene to create myoblasts lacking HIF-1α. As can be seen in Fig. 6, deletion of HIF-1α attenuated the ability of the myoblasts to downregulate oxygen consumption in response to hypoxia. Instead, HIF-null myoblasts maintained high O2 consumption, demonstrating that they are unable to completely regulate aerobic metabolism. Additionally, HIF-null myoblasts failed to upregulate PDK1 under hypoxia, indicating that the HIF-1α/PDK1 pathway is crucial for muscle cells to control O2 consumption during times of hypoxic stress.
Response of AMPK to HIF-1α deletion and exercise.
AMP activated protein kinase (AMPK) acts in the cell to turn on ATP-producing pathways and turn off ATP-consuming pathways in times of metabolic need (12). It can be activated by increased cellular AMP, as well as through phosphorylation of threonine residue 172 of its alpha subunit, and exercise has been shown to increase AMPK activity (29). It has been linked to increases in many of the parameters described above and could be linked, in turn, to an energy-deficient state induced by loss of HIF-1α, since HIF-1α is linked to decreased ATP production during hypoxia (26). To see whether the response of AMPK was affected in the muscles of our HIF-1α null mice due to HIF-1α-dependent alterations in ATP production, we performed Western blot analysis on extracts from quadriceps muscles of untrained WT, and HIF-1α null mice at rest, and immediately following a timed 30-min run at 24m/min. A representative blot for phosphorylated AMPK is shown in Fig. 7A. As expected, exercise caused a threefold increase in the level of phospho-AMPK in WT muscles. However, resting HIF-1α null muscles already have a threefold greater level of phospho-AMPK relative to WT muscles and do not further increase this during exercise (Fig. 7A). The increases in phospho-AMPK were not a function of increased total AMPK as those levels were unaffected by either loss of HIF-1α or exercise (Fig. 7B). One of the primary targets of AMPK is ACC, which AMPK inhibits through phosphorylation (5). To determine whether the phosphorylation of AMPK had functional consequences, we examined the levels of ACC phosphorylation. Similar to AMPK phosphorylation, we found that the HIF-null muscles had a trend toward higher levels of ACC phosphorylation at rest, indicating that the phosphorylated AMPK is indeed active and contributing to the phenotypes seen in the HIF-null mice (Fig. 7, C and D). The increase in phospho-AMPK is further evidence that muscles lacking HIF-1α exist in a “pretrained” state (10); they further indicate that the mechanism for this preadaptation is a HIF-dependent, ATP flux-regulated increase in constitutive AMPK phosphorylation.
Exercising muscles experience severe oxygen deprivation (25), and thus it is easy to imagine a central role for the primary hypoxia response factor, HIF-1α, during the muscular response to endurance training. The primary training response is better delivery and utilization of oxygen in the muscle, which further strengthens this hypothesis arguing for the centrality of the HIF response to training. This is a hypothesis that is shared by other researchers (9, 19), and, quite surprisingly, our data do not support this hypothesis. Loss of HIF-1α did not have any visible negative impact on endurance training in the HIF-1α null mice. In fact, these mice have actually undergone an adaptive process before training, leading to their being better suited for endurance exercise; endurance training is unable to further improve upon this adaptation. As a result, they neither increased oxidative capacity following training, nor increased fatty acid oxidation, primarily because these parameters had already been elevated as a result of HIF-1α deletion. The main factor leading to the common increase in endurance in both WT and HIF-1α null mice is likely an increase in hexokinase activity, an enzyme whose activity increased to the same degree following training in both genotypes. As an entry point enzyme in glycolysis, hexokinase is of great importance to endurance. Because muscle lacks glucose-6-phosphatase, hexokinase activity ensures that glucose, once it enters the muscle fiber, cannot leave and will thus provide energy either immediately, or later, as stored glycogen. It has recently been demonstrated that knockdown of hexokinase decreases muscle glycogen stores, and hexokinase activity correlates very strongly with endurance capacity (11). These findings corroborate the conclusion that an increase in hexokinase activity is a key factor leading to the training-induced endurance increase in both genotypes studied here.
Our previous work demonstrated an increased endurance in the untrained HIF-1α null mice not shown in this study. This is due to differences in protocols used during the run tests here, where the mice spent the majority of their run at higher velocities, which, in turn provides a more intense exercise stimulus. More intense endurance exercise requires increased carbohydrate metabolism, while less intense exercises allow muscles to derive a greater portion of their energy from fatty acid oxidation (3). In our previous study, the majority of the run time was at a lower velocity, thereby allowing the muscles of the HIF-1α null mice to take full advantage of their adaptation for better fatty acid metabolism, leading to their observed increased endurance in those protocols. The most intriguing finding of this study was the degree to which loss of HIF-1α caused an endurance training-like adaptation of the skeletal muscle. Without any prior training, HIF-1α null muscles had elevated CS and βHAD activities, increased capillary-to-fiber ratios, upregulated resting Glut4 expression, and activated AMPK. No fiber type composition changes were observed in the HIF-1α null muscles, indicating that no changes were needed to keep up with the demands of the training protocol. The adaptations had the effect of lowering the RER of HIF-1α null mice during low-intensity exercise and led to the increased endurance under the submaximal exercise protocol that we described previously (20).
We have found evidence that the loss of HIF-1α in skeletal muscle leads to a constitutively elevated level of phospho-AMPK; this, in turn, can lead to the adaptive changes that otherwise occur during training. For example, activated AMPK has recently been shown to phosphorylate Glut4 enhancer factor (GEF), causing GEF to translocate to the nucleus and bind to the Glut4 promoter (14). Additionally, several studies have demonstrated an AMPK-dependent increase in CS activity (30), and mitochondrial DNA and biogenesis (2, 18). In fact, Kukidome et al. (18) saw a 22% increase in mtDNA content through constant activation of AMPK via the drug 5-aminoimidazole-4-carboxamide riboside (18). This is very similar to the nonsignificant increase of 24% in mtDNA found in the untrained HIF-null mice relative to WT mice in this study. Finally, constant activation of AMPK has been seen to lead to increased skeletal muscle angiogenesis (21); this may also explain the increased vascularization seen in the mutants described here. Coupling these previous results with our findings indicates that increased levels of phospho-AMPK may be the source of the adaptations seen as a result of the loss of HIF-1α in the skeletal muscle of these mice. That WT mice only had elevated AMPK in response to exercise reinforces the understanding that AMPK is normally only activated in response to exercise and underscores its potential to function in the muscular response to endurance training. Unregulated AMPK activity in the HIF-null mice would indeed have profound effects on the ability of the muscles to respond to endurance exercise.
Our data unequivocally show that HIF-1α signaling is not essential for the muscular response to endurance training. Recent findings lead to questions about whether HIF-1α signaling would be beneficial to endurance training and have shown a potential for HIF-1α to have a strongly negative effect on mitochondrial adaptation. In work with a pheochromocytoma line, Dahia et al. (6) have shown that constitutively active HIF-1α leads to a sharp decrease in succinate dehydrogenase B protein, indicating a likely drop in electron transport chain activity. Additionally, Papandreou et al. (22), and Kim et al. (17) have recently demonstrated that HIF-1 can directly downregulate mitochondrial oxygen consumption through increased expression of pyruvate dehydrogenase kinase I, an inhibitor of pyruvate dehydrogenase. Finally, De Palma et al. (7) have shown that hypoxic rat muscle upregulates HIF-1 and PDK1, which would be expected to impede the muscular response to endurance training. Our finding of decreased PDK1 expression in the muscles of the HIF-1α null mice is corroborated by the findings of Papandreou et al. (22), Kim et al. (17), and De Palma et al. (7) and indicates that the increased capacity for aerobic metabolism in the mutant mice is a result of this. Furthermore, our findings that loss of HIF-1α in myoblasts leads to an inability to control O2 consumption and upregulate PDK1 expression in hypoxia strengthen the hypothesis that HIF-1α is a negative regulator of oxidative metabolism in skeletal muscle during exercise. Finally, our own observations also indicate that HIF-1α is unimportant for endurance training in hypoxic conditions, as control and HIF-null mice trained in hypoxia (12% O2) are able to improve their performance to the same degree, similar to our findings for training in normoxia (S. Mason, R. Duh, R. Johnson, unpublished observation). Taken together, our data and the previous work indicate that HIF-1 signaling may not be of benefit to endurance training, since it would lead to a blockade of oxidative metabolism. Instead, training may well involve suppression of the HIF-1 signaling pathway, and, in turn, increased levels of phospho-AMPK caused by hypoxia-driven ATP flux. In support of that hypothesis, Lundby et al. (19) have recently demonstrated a decrease in the induction of HIF-1α and HIF-2α in endurance-trained muscles following exercise. Although little is known regarding the role of HIF-2α in skeletal muscles, the results of Lundby et al. (19) address the potential for HIF-2α to compensate for the loss of HIF-1α in the HIF-null mice. As the expression of both HIF-1α and HIF-2α was not upregulated in trained muscles, it is quite likely that endurance training results in a decrease of HIF signaling as a whole, indicating that removal of HIF signaling is an important event in endurance training. These recent results, when combined with the findings described above, support a key role for the suppression of HIF-1 signaling during training, ultimately leading to an adaptive response through AMPK causing increased oxidative capacity in skeletal muscle.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society