Hypoxia-inducible factor 1α (HIF-1α) transcriptionally activates multiple genes, which regulate metabolic cardioprotective and cross-adaptive mechanisms. Hypoxia and several other stimuli induce the HIF-1α signaling cascade, although little data exist regarding the stress threshold for activation in heart. We tested the hypothesis that relatively mild short-cycle hypoxia, which produces minimal cardiac dysfunction and no sustained or major disruption in energy state, can induce HIF-1α activation. We developed a short-cycle hypoxia protocol in isolated perfused rabbit heart to test this hypothesis. By altering cycling conditions, we identified a specific cycle with O2 content and duration that operated near a threshold for causing functional injury in these rabbit hearts. Mild short-cycle hypoxia for 46 min elevated HIF-1α mRNA and protein within 45 min after reoxygenation. Expression also increased for multiple HIF-1α target genes, such as VEGF and heme oxygenase 1. After mild hypoxia, VEGF protein accumulation occurred, although HIF-1α and VEGF protein accumulation were suppressed after more severe hypoxia, which also caused depletion of ATP and nondiffusible nucleotides. In summary, these results indicate that mild near-threshold hypoxia induces HIF-1α cascade, but more severe hypoxia suppresses protein accumulation for this transcription factor and the target genes. Posttranscriptional suppression of these proteins occurs under conditions of altered energy state, exemplified by ATP depletion.
- heme oxygenase
- oxygen consumption
- vascular endothelial growth factor
hypoxia induces adaptive responses in heart, which promote resistance to other environmental stressors (28). This cross-tolerance is due to the activation of common protective pathways, including induction or reprogramming of gene signaling pathways and modification of posttranscriptional mechanisms (14). Cross-tolerance may provide a therapeutic mode of reducing effects of noxious stimuli, such as ischemia during cardiac surgery.
Many of these adaptive responses are mediated by hypoxia inducible factor 1α (HIF-1α). Oxygen concentration influences HIF-1α by multiple mechanisms, including posttranslational modifications, nuclear translocation, heterodimerization with the HIF-1β subunit, and target gene trans-activation (4). A decrease in ATP concentration or production, generally accompanying hypoxia, putatively triggers some of these mechanisms (14). However, oxygen-independent pathways, such as hormones, cytokines, and growth factors, can all trigger accumulation of HIF-1α and its transcriptional activation. HIF-1α also responds to reactive oxygen species and nitric oxide signaling. In the intact heart, HIF-1α activation and induction have been studied after prolonged heat stress (30 days), severe intermittent hypoxia (2), or ischemia with myocardial infarction (14). Furthermore, HIF-1α contributes to the mechanism for cross-tolerance between heat shock and ischemia. Although these chronic forms of stress induce HIF-1α and promote HIF-1α protein stabilization, they would provide limited clinical benefit because of their severe and prolonged nature.
The principal objective of this study was to determine whether mild acute hypoxic stress induces a modification in HIF-1α signaling. We therefore designed an experiment in which we first determined the cardiac functional threshold for oxygen deprivation using short-cycle hypoxia. The cycle durations used in this study have direct relevance to clinical scenarios as well as to high-altitude medicine. For instance, cycling (15–90 s) in arterial oxygen saturation occurs during nocturnal periodic breathing induced by high altitude (4,500–8,000 m) (9, 16, 23). Similarly, patients with congestive heart failure exhibit 60-s cycles during sleep (8, 12). Using the short-cycle hypoxia protocol, we determined whether the HIF-1α mRNA and protein responses occur at oxygen-deprivation levels, which do not surpass the injury threshold, and whether the response was divorced from major metabolic or functional disturbances.
Preparation of isolated hearts.
Rabbit hearts (male or female rabbits, 2.3–2.8 kg body wt, anesthetized with 45 mg/kg pentobarbital sodium and heparinized with 700 U/kg iv pentobarbital sodium) were rapidly excised and momentarily immersed in ice-cold physiological salt solution (PSS). Procedures followed were in accordance with the “Guide for the Care and Use of Laboratory Animals” published by the National Institutes of Health (publication 85-23, revised 1996) as well as approved by the University of Washington Institutional Animal Care and Use Committee (no. 3107). The preparation has been reported previously (17). In brief, the aorta was cannulated with the use of the Langendorff mode perfused with PSS equilibrated with 95% O2-5% CO2 at 37°C and pH 7.4. The PSS contained (in mmol/l) 118.0 NaCl, 4.0 KCl, 22.3 NaHCO3, 11.1 glucose, 0.66 KH2PO4, 1.23 MgCl2, and 2.38 CaCl2 and was passed twice through filters with 3.0-μm pore size. Perfusion pressure was maintained at 90 mmHg (17, 18). A pressure transducer was connected to a balloon, which was placed in the left ventricle through the mitral orifice to measure left ventricular pressure and its first derivative with respect to time (dP/dt). The balloon volumes were varied over a range of values to construct left ventricular pressure-volume curves to define an optimal function condition with developed pressure between 100 and 140 mmHg and end-diastolic pressures <8 mmHg at baseline. This volume remained unchanged during the entire experiment. The coronary flow was measured with a flow meter (T201; Transonic Systems, Ithaca, NY) that was connected to a cannula in the pulmonary artery. Myocardial oxygen consumption was calculated as CF × [(PaO2 − PvO2) × (c/760)]/VM, where CF is coronary flow (ml·min−1·g wet wt−1), (PaO2 − PvO2) is the difference in Po2 (Torr) between the coronary affluent and effluent; c is the Bunsen solubility coefficient of O2 in perfusate at 37°C (22.7 μl O2 × atm−1·ml−1 perfusate), and VM is the molar volume (22.4 ml/mmol). Lactate production was the difference between coronary effluent and coronary affluent concentration times coronary flow.
Short-cycle hypoxia was induced by timed alteration of hypoxic PSS infusion with normal oxygenated PSS (19 μl/ml). The hypoxic PSS O2 content was varied by mixing PSS bubbled with 95% nitrogen-5% CO2 with oxygenated PSS. We used a 120-s full cycle containing hypoxia and reoxygenation components and exposed each heart to a single variation of this cycle with respect to hypoxia duration. Accordingly, hearts underwent one of the following hypoxia-reoxygenation (H/R) protocols: 10 s/110 s, 30 s/90 s, and 60 s/60 s. Preliminary data showed that at least 60 s is required for recovery within each individual cycle. The H/R duration and the cycle intervals for this study were based on clinically relevant hypoxia cycling documented at high altitude (16, 23) and during congestive heart failure (12). Forty-two hearts were used in this entire study. Most underwent the short-cycle hypoxia protocols defined above. Functional analyses for four or more hearts were evaluated at each time or intensity point. Twelve hearts were perfused under control conditions without hypoxic exposure (C group).
Two other groups (A and S) were defined in the first stage of experiments (shown in results) by their cardiac function response to short-cycle hypoxia and their relationship to the functional injury threshold (A hearts were exposed to H/R cycling protocol of 10 s/110 s; S hearts were exposed to H/R cycling protocol of 60 s/60 s). In this study, subthreshold is defined as oxygen deprivation inadequate to elicit a decrease in cardiac function. Accordingly, the oxygen content would be higher than oxygen content that would surpass the threshold. Hearts in the A group (H/R 10 s/110 s) received subthreshold doses of oxygen deprivation, and hearts in the S group (H/R 60 s/60s) received hypoxic exposure, which surpassed the injury threshold. Hearts from both A and S groups were perfused with oxygen content of 2.2 μl/ml for 23 cycles. Samples for energy metabolites were collected after 15 min of reoxygenation, following completion of the 23rd cycle or at the comparable time in C group.
Energy metabolite measurements.
Hearts were rapidly frozen with tongs and stored in liquid N2. Tissue was lyophilized for 48 h at −40°C under 200-Torr vacuum. An aliquot (10 mg) of the dried tissue was homogenized with 800 μl of 0.73 M TCA. After centrifugation (7,000 rpm, 2 min) at 4°C, the supernatant (400 μl) was removed and added to a new Eppendorf tube containing an equal volume of tri-n-octylamine and Freon (1:1, vol/vol). The sample mixture was then vortexed and centrifuged as before. The aqueous phase was analyzed by HPLC with Waters 484 ultraviolet absorbance detector for nucleotides (ATP, ADP, AMP, IMP) and nucleosides (hypoxanthine, xanthine, adenosine, inosine) (20, 21). Analysis was performed with Waters Maxima 820 software and NEC Power Mate 1.
Total RNA was extracted with a RNA isolation kit (Ambion, Austin, TX) from an aliquot (100 mg) of pulverized, homogenized frozen tissue. RNA samples were tested by ultraviolet absorption at 260 nm to determine the concentration, and the quality was further confirmed by electrophoresis on denatured 1% agarose gels (22, 24).
Northern blot analysis.
Fifteen micrograms of RNA were denatured and electrophoresed in a 1% formaldehyde agarose gel, transferred to a nitrocellulose transfer membrane (Micron Separations, Westboro, MA), and cross-linked to the membrane with short-wave UV cross-linker. The hybridizing solutions contained 50% formamide, 1× Denhardt solution, 6× sodium chloride-sodium phosphate-EDTA, and 1% SDS. mRNA levels of the 70-kDa heat shock protein (HSP70–1) were detected with a 1.7-kb cDNA fragment cloned from human hippocampus (ATCC, Rockville, MD) (17, 19). mRNA levels of glucose-regulated protein 94 (Grp94) were detected with a 427-bp cDNA fragment cloned from rabbit heart (made by our laboratory). cDNA probes were labeled with [32P]dCTP by random primer extension (PRIME-IT II; Stratagene, La Jolla, CA) and added to the hybridizing solution to a specific activity. Hybridization was carried out at 42°C for 18 h followed by several blot washings with a final wash in 1× standard sodium citrate and 0.1% SDS at 65°C. The blots were exposed on PhosphorImager (model 400S; Molecular Dynamics, Sunnyvale, CA) and Kodak Biomax MS film (Eastman Kodak, Rochester, NY) at −70°C. The RNA loading was normalized by comparison to that of GAPDH (17, 19). To compare different mRNA levels in the same myocardial sample, mRNAs were analyzed by means of sequentially reprobing the membranes for GAPDH, HSP70–1, and Grp94 cDNA probes (sequenced in our laboratory).
Hypoxia microarray analysis.
A hypoxia signaling pathway cDNA microarray was used to compare gene expression between rabbit heart samples taken from the three groups (A, S, and C). Microarrays were used to identify genes with potential for response to short-cycle hypoxia. The GEArray Q series gene expression array contains 96 genes presumably regulated by hypoxia (Super Array Bioscience, Frederick, MD). Total RNA from four heart samples within each group were pooled to provide an aliquot of 3 μg per group that could be used to make Biotin 16-dUTP-labeled cDNA, which was then used as a probe of gene expression on the microarray. The labeled cDNA from each group was hybridized overnight to a hypoxia microarray. The next day, the microarrays were washed to remove any unhybridized probe. Chemiluminescence was used to detect gene expressed, and the results were recorded on Kodak BioMax Light-1 film. All procedural steps in the GEArray expression array kit provided by the supplier were followed. Specific array signal spots were analyzed with ImageQuant quantitation software (Molecular Dynamics, Sunnyvale, CA) (18).
Specific proteins were selected for content determination, based primarily on results from the hypoxia microarray. Fifty micrograms of total protein extracts from rabbit heart tissue were electrophoresed along with one lane containing 30 μg of human HeLa cells as a positive control and one lane of molecular weight size markers (chemichrome Western control; Sigma) in 4.5% stacking and either a 7.5, 10, or 12% running SDS-polyacrylamide gels. The gels were then electroblotted onto polyvinylidene difluoride plus membranes. Western blots were blocked for 1 h at room temperature with 5% nonfat milk in Tris-buffered saline plus Tween 20 (TBST) (10 mM Tris·HCl, pH 7.5, 150 mM NaCl, and 0.05% Tween 20), followed by overnight incubation at 4°C with each primary antibody diluted in the appropriate blocking solution as recommended by the supplier. The primary antibodies glucose transporter 4 (400064) and Grp94 (368675) were obtained from Calbiochem and EMD Biosciences. The primary antibodies IL-6 signal transducer (gp130, sc-656), heme oxygenase 1 (sc-1797), HSP70 (sc-1060), HIF-1α (sc-12542), peroxisomal proliferator-activated α-receptor γ-coactivator 1 (sc-5816), and VEGF (sc-152) were obtained from Santa Cruz Biotechnology. After two 5-min washes with TBST and one 5-min wash with TBS, membranes were incubated at room temperature for 1 h with the appropriated secondary antibody conjugated to horseradish peroxidase. The membranes were washed twice for 10 min with TBST and visualized with enhanced chemiluminescence after exposure to Kodak Biomax Light ML-1 film. The membranes were stripped by washing them two times for 30 min with 200 mM glycine, 0.1% SDS, and 1% Tween 20 (pH adjusted to 2.2), followed by three 10-min washes with TBS. The membranes were again blocked for 1 h as above, followed by overnight incubation at 4°C with a β-actin antibody (Santa Cruz Biotechnology; sc-1616) diluted 1:200 in blocking solution. The next day, the membranes were washed (as above), the appropriate secondary-horseradish peroxidase antibody was applied, and the remaining procedures as described above were followed. The β-actin was used to verify protein lane loadings.
The reported values are means ± SE. The S-PLUS version 6.0 program (Insightful, Seattle, WA) was used for statistical analysis. Data were evaluated with repeated-measures ANOVA within groups and single-factor ANOVA across groups. When significant F values were obtained, individual group means were tested for differences. The criterion for significance was P < 0.05 for all comparisons.
To determine hypoxia threshold, we needed to define a set value for contractile function impairment. Our preliminary studies indicated that we could confidently resolve a 10% decrease in left ventricular developed pressure. Therefore, left ventricular developed pressure < 90% baseline was defined as functional impairment.
Developed pressure shows recovery to <90% baseline during reoxygenation after 23 cycles (Fig. 1). Thus this particular heart displayed contractile impairment after the short-cycle hypoxia protocol. To determine the hypoxia intensity threshold, the hearts were exposed to a given exposed duration and cycles, 10-s hypoxia followed by 110 s reoxygenation and total of 23 cycles. The hearts displayed no deviation from baseline developed pressure throughout the protocol until oxygen content in the perfusate decreased to below 12 μl/ml for five cycles from the normal O2 content of 19 μl/ml. This response was transient for hearts exposed to 7.4, 4.6, 2.4, and 2.2 μl/ml for 23 cycles (46 min), and full recovery was achieved rapidly with reoxygenation for each cycle. Cycling with an O2 content ≤2.0 μl/ml impaired functional recovery during the cycle's reoxygenation portion, thereby defining oxygen content between 2.0 and 2.2 μl/ml as the injury oxygen threshold for these cycling conditions (Fig. 2).
Cycle length and number.
The next phase was to define the number of cycles operating at previously defined oxygen content (2.2 μl/ml), which would reasonably maintain cardiac function and presumably energy status. The principal goal was to achieve at least 45 min of stability and allow a 45-min recovery period before sampling for metabolic state and mRNAs (see discussion). Perfusing at oxygen content of 2.2 μl/ml maintained cardiac function for as long as 23 cycles (46 min). Extending the hypoxia cycle duration to 30 s impaired contractile function during the reoxygenation phase (Fig. 3). Doubling the hypoxia cycle duration (60 s) did not exacerbate functional impairment, implying that oxygen content and not hypoxia duration within the cycle determined functional impairment, once threshold was surpassed (Fig. 3). The short duration of the hypoxia period in the H/R 10 s/110 s (A group) and 30/90 cycles precluded accurate sampling of steady-state oxygen consumption during the hypoxic and recovery portion of these cycles. Details for cardiac function, coronary flow, and myocardial oxygen consumption during cycling and recovery are shown in Table 1 for S and C groups. Data were obtained for each portion of the 1st cycle, hypoxia (H1) and reoxygenation (R1), and for the final, 23rd cycle (H23 and R23). Significant differences (P < 0.05) between these groups occur for most of these parameters by the end of the 23rd hypoxic cycle. However, reoxygenation abrogates many of these differences, implying that lack of profound persistent impairment was caused even by H/R 60/60 s (S group).
The relative coronary flow responses during the hypoxic portions for the 1st and 23rd cycles in C, A, and S groups are shown in Fig. 4. Coronary flow did not increase in A compared with C group during the hypoxic portions of the cycling periods, although the potential for increasing coronary flow exists as shown by the data for the S group (Table 1 and Fig. 4). The dramatic coronary flow response in the S group (160 ± 2.0%) demonstrates that the C and A groups have not accessed their full coronary reserve. The coronary flow returned to baseline during the reoxygenation portion of the cycles. High coronary flow apparent in S distinguishes this model from ischemic variants and provides a mechanism for energy metabolite washout (discussed later). For all cycle lengths with oxygen content beyond the injury intensity threshold, function progressively deteriorated as cycle number increased (Fig. 3), although not significantly until reoxygenation cycle 10. The progressive deterioration due to increasing cycle number indicates that cycle effect is additive once intensity threshold is surpassed.
Energy metabolism and metabolic stress.
Myocardial ATP, total nondiffusible nucleotides, and total diffusible nucleosides were measured for the three heart perfusion conditions (Fig. 5). ATP and total nondiffusible nucleotides were not noticeably different between C and A but significantly decreased in S compared with C and A (P < 0.05). The concentrations of total diffusible nucleosides were not statistically different between the groups. Minimal lactate production occurred during control conditions and cycling in group A (Fig. 6). However, group S showed substantial increases in lactate production throughout the protocol.
Gene expression during mild and severe short-cycle hypoxia.
We analyzed mRNA expression in groups C, A, and S as previously defined. We compared C with a single group exhibiting metabolic stress and functional injury (S) and a group exhibiting no significant ATP depletion and functional injury (A) but exposed to hypoxic cycling. Samples were extracted from hearts after 45 min of reoxygenation following completion of 23 cycles. Our initial Northern blot analyses revealed that expression for two stress-related genes, HSP70 and Grp94, relative to GAPDH were significantly altered by both mild and more severe hypoxia (Fig. 7). GAPDH mRNA content relative to β-tubulin was not different among the groups (data not shown). Because Grp94 is putatively regulated by HIF-1α, we then performed more extensive analyses of the HIF-1α signaling pathway using the hypoxia microarray to identify target genes and their proteins, which might show responses during our protocol. We defined change in gene expression by comparisons among groups relative to GAPDH and β-actin as indicated in Fig. 8. Because array samples were pooled and therefore could not be subjected to statistical analyses, we used strict criteria for defining differences in expression. Upregulation in groups A or S was defined as change >2.0-fold compared with C, whereas downregulation was defined as change <0.6-fold. The HIF-1α signal within the chips relative to signal for many genes was relatively low. Nevertheless, we measured these changes by our criteria in both A and S relative to C. Table 2 shows results for known HIF-1α target genes. Also shown are results for other important hypoxia-regulated genes, which demonstrated differences among groups. All other genes on the microarray chip are included in the supplemental Table 1. (The online version of this article contains supplemental data.) These showed no change in expression among the groups. The HIF-1α target gene response to these protocols was highly variable as short-cycle hypoxia induced VEGF, glucose transporter 1 (Glut1), and glucose transporter 2 but did not alter expression for erythropoietin (EPO). Genes related to other signaling pathways are also included on the chip. For example, the apoptosis inhibitor Survivin (BIRC5) is downregulated by severe short-cycle hypoxia.
Protein expression response to short-cycle hypoxia.
Immunoblots revealed a significant increase in HIF-1α and VEGF protein in group A compared with C (Fig. 9). Although a modest increase in these proteins also occurred in group S, differences from C did not reach significance. Relative expression for eight proteins is included in Table 3. The selection of proteins for these immunoblot experiments was based on positive results from the microarray and availability of antibodies. Peroxisome proliferator α-receptor γ-coactivator-1 was included as a primary metabolic regulator. The antibody for Glut1 did not provide adequate signal for analysis.
The principal objective of this study was to determine whether a relatively mild or subthreshold dose of hypoxia could suddenly trigger the HIF-1α cascade in heart. Promotion of HIF-1α protein accumulation and concomitant increases in expression for various target genes and their proteins occurred within a time frame measured in minutes. To our knowledge, no prior study has detailed the rapid triggering of the HIF-1α signaling pathway in heart by a relatively mild hypoxic stress. Stroka et al. (28) showed organ variable response of HIF-1α to changes in ambient oxygen in vivo over 1 h. In their study, brain exhibited the greatest sensitivity with marked induction at 18% inspired oxygen fraction, whereas kidney and liver showed no HIF-1α accumulation until exposed to inspired oxygen fraction of 6%. Although those authors indicated that HIF-1α appeared in hypoxic heart, no details regarding the experimental conditions were provided. In the present study, we avoided systemic induction of the HIF-1α pathway by employing an isolated perfused rabbit heart, which showed minimal expression of this protein in the control group. Appearance of protein is stimulated by the mild dose of hypoxia provided by the 10/110 s short-cycle hypoxia protocol. Because HIF-1α mRNA was barely detectable in the control hearts, the array data imply that mild short-cycle hypoxia increased mRNA content by promoting transcription rather than altering mRNA stability. Rapid transcriptional regulation of this protein represents a novel mechanism for heart, as changes in steady-state level have only been described as occurring after many hours or even days (14).
HIF-1α mRNA appeared higher in the microarrays, although no HIF-1α protein accumulated during the more severe hypoxic conditions (H/R 60/60 s), which also produced a detectable drop in myocardial ATP concentration. Within the confines of this study performed in the intact heart, we are not certain whether lack of protein response reflects a transcriptional or posttranslational disturbance. However, degradation of the HIF-1 proteins is triggered by hydroxylation of two key proline residues in the oxygen-dependent degradation domain-containing proteins via prolyl hydroxylase domain-containing proteins. Most studies show suppression of prolyl hydroxylase domain-containing protein activity by severe hypoxia, thereby inhibiting, but not suppressing, HIF-1α degradation (3, 15). Therefore, our results suggest that inhibition of translation, and not accelerated protein degradation, contributes to the reduced HIF-1α protein accumulation in the HR 60/60 s group.
ATP depletion during severe hypoxia in HEK-293 cells directly limits protein synthesis by inhibiting activity of the mammalian target of rapamycin (5), which positively regulates key enzyme controllers of translation initiation. We therefore considered whether a relationship existed between ATP and HIF-1α protein accumulation in the intact heart. Although ATP depletion occurs generally only during the most severe and often irreversible conditions of ischemia, the same does not hold true during decreases in ambient O2 concentration, which are accompanied by maintained or increased coronary flow rates. Accordingly, prior studies in hearts in vivo (25, 26) and ex vivo (10) have shown that reducing oxygen concentration to just surpass the critical threshold causes parallel depletion in phosphocreatine and ATP stores. Although phosphocreatine levels fully recover immediately with reoxygenation, ATP stores remain depressed because of accelerated purine degradation and loss during hypoxia. Degradation and washout of the degradation products are illustrated by the concomitant decrease in ATP and nondiffusable nucleotides shown in group S, while maintaining the low levels of diffusible nucleotides. Although we did not measure phosphocreatine in this study, the stability of ATP in the rabbit hearts subjected to the more modest HR 10/110-s cycle implies that no major or sustained disruption in high-energy phosphate metabolism occurred during the cycling or afterward. Additional support for limited disruption in energy status is the lack of increased lactate production throughout the protocol in group A, implying that anaerobic metabolism was not stimulated. Thus the finding that HIF-1α accumulated after hypoxia with stable ATP, but not under the more severe conditions accompanied by ATP depletion, indicates that ATP either directly or indirectly enhances accumulation of this protein after hypoxia.
Rapid induction of HIF-1α target genes in heart by a reduction in oxygen concentration has not been previously examined in detail. Previously, Cai et al. (2) demonstrated that exposure of mice to systemic intermittent hypoxia resulted in protection of isolated hearts against ischemia-reperfusion 24 h later. The mRNA for the HIF-1α target genes VEGF, Glut1, and EPO did not exhibit changes after rather profound and prolonged systemic hypoxia (5 cycles of 6% O2 for 6 min alternating with 21% O2 for 6 min) compared with the hypoxic conditions in our present study. However, kidney EPO mRNA increased and presumably stimulated observed elevations in plasma EPO levels. Cardiac protection and EPO elevation were abrogated in mice heterozygous for a knockout allele at the locus encoding HIF-1α. Accordingly, those authors attributed cardiac protection primarily to hypoxic induction of renal EPO mRNA.
Some of our mRNA results were obtained by pooling samples from microrarray chips to identify potential targets for further evaluation, including immunoblot analyses of respective proteins. We used strict criteria, including twofold differences in expression for defining upregulation. Nevertheless, we cannot validate these array results through statistical methods, although they were for the most part consistent with Northern blot and protein analyses. Our protocol using mild short-cycle hypoxia, free of systemic influence, produced protein elevations for multiple HIF-1α targets. Additionally, we statistically validated mRNA elevations for the putative HIF-1α target Grp94 and for the stress-related gene HSP70. Protein for the HIF-1α target gene VEGF rapidly accumulates, although levels are lower in hearts exposed to the more severe protocol. Therefore, the data imply that mild short-cycle hypoxia rapidly initiates the HIF-1α signaling cascade through both transcriptional and posttranscriptional modes. As expected, the transcriptional and posttranscriptional responses vary according to target gene. This variation may be related to target gene sensitivity or differences in translational efficiency, which are uncovered by this short protocol. Additionally, significant protein accumulation for HIF-1α and VEGF may be technically easier to detect, as control levels for these are relatively low compared with the other proteins evaluated. The EPO response to protocols, which presumably activate HIF-1α cascade, has varied (2, 14). Some report that HIF-1α dramatically induces responses to prolonged heat stress (14). Our findings with respect to EPO are consistent with results from previous studies employing longer and more severe intermittent hypoxia (2). Accordingly, results from prior work and our own present experiments together imply that hypoxia does not promote changes in cardiac EPO mRNA.
Although our study was directed toward evaluation of the HIF-1α signaling cascade, the commercial arrays also provide some interesting potential information regarding other important pathways. We did not intend to pursue these serendipitous results within the scope of the present study, although some of these results point to directions for further research. For instance, we noted a marked decrease in mRNA for EEF1A1 in the severe hypoxia group. Unlike other peptide elongation or initiation factors (13), the response of this particular eukaryotic elongation factor to hypoxia has not been previously reported. Our finding with respect to mRNA expression for this factor, although preliminary, represents a potential mechanism for reduction in protein synthesis after severe hypoxic insult. Additionally, we noted that severe hypoxia decreased expression for Survivin (BIRC5), a known inhibitor of apoptosis (1). Changes in mRNA for EEF1A1 and Survivin did not occur during the modest hypoxia, which elicited response in the HIF-1α pathway.
The nature of the short-cycle hypoxia as performed in these protocols precluded measurement of many parameters, which serve as energy sensors within cardiomyocytes. Conceivably, undetected and transient alterations in cellular redox state may trigger activation of HIF-1α signaling cascade. AMP-activated protein kinase (AMPK) serves as an energy-sensing protein, which responds to transient metabolic stresses and modulates multiple metabolic and gene signaling pathways (27). AMPK activity in particular can be triggered rapidly by hypoxia in heart (6), although increases in activity for this enzyme can take as long as 30 min (7). The relationship between AMPK and HIF-1α has not been clearly established in heart, although this enzyme displays a critical function for stimulating HIF-1 transcriptional activity in cancer cells.
In summary, we have shown that HIF-1α cascade responds to relatively modest short-cycle hypoxia. We define this modest short-cycle hypoxia as subthreshold because this degree of oxygen deprivation does not elicit major changes in cardiac function or ATP stores. The subthreshold definition is supported further by diminished stimulation for coronary vasodilatation and anaerobic metabolism, apparent in hearts exposed to mild short-cycle hypoxia. In contrast, slightly more severe short-cycle hypoxia or oxygen deprivation, which surpasses the threshold, is accompanied by decreases in ATP and nucleoside pool and marked elevations in coronary blood flow and lactate production. These metabolic changes occur with reductions in the HIF-1α and VEGF protein accumulation, suggesting that protein synthesis is inhibited, although an increase in protein degradation remains a possibility. Induction of HIF-1α by subthreshold hypoxia shows promise as a mechanism for cross-adaptation and protection against subsequent ischemic injury. Thus mild short-cycle hypoxia shows clinical relevance or potential, as this protocol stimulates this signaling cascade within a brief time without causing detectable changes in cardiac function or energy status. The data also highlight the exquisite sensitivity of the HIF-1α pathway to subthreshold oxygen deprivation.
This work was supported in part by Children's Hospital and Regional Medical Center Grant HR5836, National Science Foundation Grant 38970307 (P. R. China), and National Heart, Lung, and Blood Institute Grant R01 HL-60666.
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 by the American Physiological Society