Hypoxia-Induced Inhibition of Adenosine Kinase Potentiates Cardiac Adenosine Release
Abstract To elucidate the physiological role of the AMP-adenosine metabolic cycle and to investigate the relation between AMP and adenosine formation, the O2 supply of isolated guinea pig hearts was varied (95% to 10% O2). The net adenosine formation rate (AMP→adenosine) and coronary venous effluent adenosine release rate were measured; free cytosolic AMP was determined by 31P-nuclear magnetic resonance. Switching from 95% to 40% O2 increased free AMP and adenosine formation 4-fold, whereas free cytosolic adenosine and venous adenosine release rose 15- to 20-fold. In the AMP range from 200 to 3000 nmol/L, there was a linear correlation between free AMP and adenosine formation (R2=.71); however, adenosine release increased several-fold more than formation. At 95% O2, only 6% of the adenosine formed was released; however, this fraction increased to 22% at 40% O2, demonstrating reduced adenosine salvage. Selective blockade of adenosine deaminase and adenosine kinase indicated that flux through adenosine kinase decreased from 85% to 35% of adenosine formation in hypoxia. Mathematical model analysis indicated that this apparent decrease in enzyme activity was not due to saturation but to the inhibition of adenosine kinase activity to 6% of the basal levels. The data show (1) that adenosine formation is proportional to the AMP substrate concentration and (2) that hypoxia decreases adenosine kinase activity, thereby shunting myocardial adenosine from the salvage pathway to venous release. In conclusion, because of the normal high turnover of the AMP-adenosine metabolic cycle, hypoxia-induced inhibition of adenosine kinase causes the amplification of small changes in free AMP into a major rise in adenosine. This mechanism plays an important role in the high sensitivity of the cardiac adenosine system to impaired oxygenation.
Whenever the O2 supply to the myocardium is critically impaired, there is an increase in adenosine formation, and in consequence, cytosolic and interstitial adenosine increases.1 Subsequently, mediated by different purine receptors, adenosine is assumed to play an important role in regaining the balance between energy supply and demand,2 eg, by its potent vasodilatory3 and antiadrenergic4 5 effects. Adenosine may also be involved in ischemic preconditioning.6 Although a large number of studies have focused on the cardiac adenosine metabolism, the precise mechanism linking myocardial oxygenation to adenosine formation and release is not yet fully understood.
In the normoxic heart, a close match of ATP formation and consumption maintains stable free cytosolic concentrations of both ATP (5 to 10 mmol/L) and ADP (40 to 60 μmol/L). Whenever the O2 supply is inadequate for a given level of cardiac work, the cardiac energy status is compromised, and free cytosolic ADP increases. Because of the myokinase equilibrium, this is translated into a rise of free AMP. Conventionally, it is assumed that the increase in AMP, the substrate for 5′-nucleotidase, results in a rise in adenosine formation. Consistent with this concept, a simultaneous rise of myocardial free AMP and coronary venous release of adenosine from the heart were observed in hypoxia in many previous studies (eg, see References 7 through 97 8 9 ). It is unclear, however, whether the rate of AMP dephosphorylation to adenosine in the heart is indeed controlled by the substrate concentration. In addition, the 10- to 30-fold rise seen in cytosolic adenosine10 and adenosine release7 8 9 in the presence of much smaller (2- to 4-fold) changes in free AMP remains largely unexplained.
In the heart, adenosine is predominantly produced from free cytosolic AMP via cytosolic 5′-nucleotidase. Since the Km of purified cytosolic 5′-nucleotidase is at least two orders of magnitude above the cytosolic AMP concentration,11 12 a linear relation between free AMP and the rate of adenosine formation might be expected. However, the enzyme activity of cytosolic 5′-nucleotidase is also dependent on ATP, ADP, and Pi as well as free Mg2+ and pH.11 13 Moreover, adenosine-induced deactivation of the enzyme has also been described.14 To test the regulation of 5′-nucleotidase in vivo in the past, several indirect indexes of adenosine formation have been used: Activation of 5′-nucleotidase in the hypoxic heart has been postulated on the basis of myocardial adenine nucleotide loss.15 Others have observed a linear correlation of adenosine release and free AMP and have suggested control of adenosine formation by substrate concentration.9 Exponential16 17 and hyperbolic relations7 8 between free AMP and coronary venous adenosine or purine release have also been described. Thus, it is presently unclear whether adenosine formation is indeed primarily controlled by the AMP concentration.
Our laboratory has recently demonstrated that in the normoxic heart ≈80% of adenosine formed from AMP is intracellularly rephosphorylated by AK.18 The remainder is either converted to inosine by ADA or released from the heart. Because of the high turnover of the AMP-adenosine cycle, intracellular adenosine formation is much greater than coronary adenosine release. Similar findings have been reported by others.19 20 Since the rapid turnover of this AMP-adenosine metabolic cycle continuously consumes ATP, the function of this cycle is an intriguing question. At a given rate of adenosine formation, a high activity of AK limits purine loss from the heart. Using a comprehensive model of adenosine metabolism, we predicted21 that a high turnover of the AMP-adenosine metabolic cycle accelerated the response of cytosolic adenosine to changes in free AMP, such as those that occur during hypoxia. In addition, it was postulated that endogenous inhibition of AK might transform this cycle into an amplification system, translating small changes in adenosine formation into major changes in cytosolic adenosine and, ultimately, adenosine release.
In view of these unresolved issues, the goals of the present study were 2-fold: (1) to test the hypothesis that there is a linear relation between free cytosolic AMP concentration and the rate of adenosine formation in the hypoxic guinea pig heart and (2) to investigate whether hypoxia causes endogenous inhibition of AK, augmenting the net flux through the AMP-adenosine cycle and increasing coronary adenosine release.
Materials and Methods
Guinea pigs weighing 300 to 350 g were killed by cervical dislocation (n=47), and the hearts were rapidly excised and perfused by a nonrecirculating Langendorff technique. A modified Krebs-Henseleit solution was used, which contained (mmol/L) NaCl 116, KCl 4.6, MgSO4 1.1, NaHCO3 24.9, CaCl2 2.5, KH2PO4 1.2, glucose 8.3, and pyruvate 2, equilibrated with 95% O2 and 5% CO2 (pH 7.4, 37°C). Coronary flow and perfusion pressure were continuously monitored by an ultrasound transit time flow probe (Transonics) and pressure transducers (Braun Melsungen), respectively. The mitral valve was cut to vent the left ventricle. The pulmonary artery was cannulated, and the caval veins were ligated. A silver electrode was gently pierced into the left ventricle, and the other was fixed immediately above the heart. The heart was placed inside a 20-mm NMR tube, immersed in perfusion buffer, and transferred into a temperated probe inside a 9.4-T NMR magnet. During transit from the medium reservoir outside the magnet to the heart, the perfusate was maintained at 37°C. Adequate choice of tubing (Tygon) ensured a transit time well below 2 minutes and a Po2 of ≈600 mm Hg at the aortic cannula.
Initially, hearts were perfused at a constant pressure of 7 kPa (52 mm Hg). After hearts had stabilized inside the magnet, cardiac pacing (300 min−1) was initiated and continued throughout. Five minutes later, the coronary perfusion rate was fixed to the steady state flow achieved during pacing and was maintained constant. A partially relaxed baseline 31P-NMR spectrum was acquired (see below), and two serial coronary venous effluent samples were collected (2 minutes each) for determination of adenosine release. After each subsequent intervention, hearts were allowed to stabilize for 5 minutes before acquisition of a further NMR spectrum and collection of venous effluent.
Free Cytosolic Adenosine, Adenosine Release, and Adenosine Formation
To determine the concentration of free cytosolic adenosine, the SAH technique10 was used. Although under physiological conditions in the heart the breakdown of SAH catalyzed by SAH hydrolase results in the net formation of adenosine and homocysteine, in the presence of 500 μmol/L homocysteine the net flux through this pathway is reversed, and the accumulation of SAH is a measure of free cytosolic adenosine (see below).
To obtain coronary venous effluent, the pulmonary artery was cannulated and connected to a narrow tubing that ultimately opened outside the magnet ≈8 cm below the heart, thus exerting a small negative pressure. For sample collections, the transit time of <1 minute was taken into account. Coronary venous adenosine release was calculated as concentration×flow.
To determine total cardiac adenosine formation, coronary venous adenosine release was measured while AK and ADA were blocked by iodotubercidin and EHNA, respectively, since in the presence of blockade of all pathways metabolizing adenosine, its release provides an estimate of adenosine formation.18 19 22 In a previous study,18 we had demonstrated that iodotubercidin was maximally effective in blocking AK at 1 μmol/L; furthermore, we had shown that in the presence of 5 μmol/L EHNA, infused tritiated adenosine was not converted to inosine, suggesting complete inhibition of ADA. In the present study, 10 μmol/L iodotubercidin and 5 μmol/L EHNA were used.
Adenosine and Free AMP in Moderate Hypoxia
In a first series of experiments, the alterations in cardiac adenosine metabolism and energy status induced by moderate hypoxia (40% O2) were investigated (n=8). Coronary venous adenosine release was determined during normoxic perfusion; subsequently, hypoxic perfusion was initiated, and a further set of venous effluent samples was obtained. During each collection of venous effluent, 31P-NMR spectra were acquired. In a second set of hearts (n=8), adenosine formation (measured in the presence of 10 μmol/L iodotubercidin and 5 μmol/L EHNA) was assessed in normoxia and during subsequent hypoxic perfusion. To determine the changes in free cytosolic adenosine induced by hypoxia, SAH accumulation (see above) and cardiac energy status were determined in two additional groups of hearts. Our laboratory had previously demonstrated that cardiac SAH content (1.1 nmol·min−1·g wet wt−1, corresponding to 8.9 pmol/mg protein) in the absence of homocysteine infusion is independent of cardiac energy status and adenosine release.10 In the present study, hearts were either perfused with normoxic medium only (n=4) or switched from normoxic to hypoxic medium after the stabilization period (n=4). After attaining steady state, 500 μmol/L homocysteine was infused for 11 minutes (see above) in both groups; subsequently, hearts were freeze-clamped and extracted for analysis of SAH content. The SAH accumulation rate was then calculated as an index of free cytosolic adenosine.
Flux Through CK and ATP Synthase
To test whether CK is in equilibrium, both in normoxia and hypoxia (see below), O2 consumption and CK flux rates were measured, and the ratio between ATP synthase and CK flux was calculated. O2 consumption was determined as previously described in detail23 and converted to ATP formation rate, assuming a P/O ratio of 3. The unidirectional flux through CK was determined by using 31P-NMR magnetization transfer techniques.24 After measuring the decrease in the phosphocreatine signal during 5-second irradiation of the γ-ATP resonance, the pseudo–first-order rate constant (kfor) for the CK reaction was calculated according to the following relation: In this equation, M0 and M∞ indicate the magnetization of phosphocreatine without and with long-time (>4-second) irradiation of the γ-ATP resonance, respectively, and T1 is the intrinsic longitudinal relaxation time of phosphocreatine and was assumed to be 3.55 seconds. This intrinsic relaxation time had been observed in a variety of species (eg, hamsters, rats, and turkeys)24 and was consistent with an apparent T1 of 2.58±0.26 seconds, determined by us in separate experiments in the normoxic guinea pig heart (n=5). The product of kfor and the phosphocreatine concentration determined in the absence of γ-ATP irradiation gave the flux through CK.
Relation Between Free AMP, Adenosine Release, and Adenosine Formation
To investigate the relation between free cytosolic AMP and adenosine formation and release, isolated hearts were exposed to different levels of hypoxic perfusion (60%, 40%, 20%, or 10% O2; each n=4 or 5). In this experimental series, left ventricular pressure was measured by an intraventricular balloon connected to a pressure transducer (Braun Melsungen) located outside, close to the magnet. Each heart was initially perfused with normoxic medium for the assessment of basal adenosine release. Subsequently, hearts were switched to one level of hypoxia only, and adenosine release and, subsequently, adenosine formation (iodotubercidin+EHNA) were determined before hearts were perfused again with normoxic medium for the measurement of adenosine formation in normoxia. NMR spectra were acquired at each of these four consecutive experimental conditions. This design permitted the comparison of (1) adenosine release in normoxia versus hypoxia, (2) adenosine release versus adenosine formation in the hypoxic heart, and (3) adenosine release versus formation in the normoxic heart.
ADA and AK Activity in Hypoxia and Normoxia
To estimate the importance of ADA and AK in adenosine metabolism, adenosine release of isolated hearts was determined in the absence and presence of either 5 μmol/L EHNA or 10 μmol/L iodotubercidin. These experiments were performed in hearts perfused with normoxic (n=10) or hypoxic (40% O2) (n=8) medium. After measurement of adenosine release, either AK or ADA was blocked before inhibition of both enzymes enabled the determination of adenosine formation.
Mathematical Model Analysis of Adenosine Metabolism in Hypoxia
A mathematical model was used21 ; this model comprehensively took into account the present knowledge of the different pathways of adenosine metabolism, specifically, the known kinetic parameters of the relevant enzymes in adenosine formation, degradation, and rephosphorylation, the compartmentalization of the various metabolic pathways, and the parameters governing tissue-capillary exchange of adenosine. At a given coronary flow and adenosine synthesis rate, the model predicted the steady state coronary venous effluent release of adenosine, as well as cytosolic and interstitial adenosine concentrations. AK and ADA were modeled by Michaelis-Menten kinetics. As described previously,21 the Km of AK and ADA was taken to be 2.5 and 83 μmol/L, respectively.25 Vmax of AK was taken to be 100 nmol·min−1·g−1 in the parenchymal cell compartment (cardiomyocytes) and 37 nmol·min−1·g−1 in the endothelial cell compartment. On the basis of histo-chemical evidence,26 ADA was assumed to be almost exclusively located in the endothelial compartment (Vmax, 950 nmol·min−1·g−1).
All 31P-NMR spectra were acquired on an AMX400 WB pulsed Fourier transform NMR spectrometer (Bruker) as described in detail previously.23 In brief, the magnetic field was adjusted to achieve a line width of the water proton resonance peak of <25 Hz. Partially saturated 31P spectra were accumulated at 161.97 MHz using a dedicated, variable-temperature, 20-mm 31P probe (Bruker) with a π/2 pulse length of 75 microseconds, 70° pulses, a sweep width of 35 ppm, 2048 data points in the time domain, and a pulse interval of 3 seconds. Each spectrum was a time average over 128 scans, with the total spectral acquisition time being 6.5 minutes. Spectra were processed using NMR1 software (Tripos). Zero filling to 4096 and exponential multiplication of the data (5-Hz line broadening) was followed by Fourier transformation, automatic phasing, and baseline correction. Phosphocreatine and the γ-ATP peak were automatically integrated over a region of 7.5×peak line width; a curve-fitting routine was used for the integration of the cytosolic Pi resonance peak, which partially overlapped with the extracellular Pi resonance.
On the basis of previous experimental data,23 basal and normoxic concentrations for ATP, phosphocreatine, and creatine were taken to be 7.08, 13.3, and 8.9 mmol/L, respectively. Relative changes in the peak areas of ATP and phosphocreatine were converted to the respective changes in metabolite concentrations, and total creatine was assumed to be constant throughout the experiment. Peak areas of phosphocreatine and intracellular Pi were compared, and the Pi concentration was calculated, taking the appropriate saturation factors into account. Cytosolic free Mg2+ concentration was estimated from the difference in chemical shift of the α- and β-phosphate resonances of ATP.27 pHi was determined from the chemical shift difference of phosphocreatine and intracellular Pi (δ) (ppm) according to the following relation: Cytosolic free ADP concentration was calculated from the CK equilibrium as follows: where Cr is creatine, PCr is phosphocreatine, and Kobs is the observed equilibrium constant of the CK reaction. Cytosolic free AMP concentration was calculated from the myokinase equilibrium as follows: where KMK is the equilibrium constant of the myokinase reaction.
Coronary venous effluent adenosine samples were measured by reverse-phase HPLC at 254 nm, as described in detail previously.18 28 Samples were desalted, and the nucleosides were concentrated. They were injected on a 150-mm C-18 Bondapak HPLC column (Waters) and eluted using a gradient changing from 5% methanol/95% ammonium acetate (26 mmol/L, pH 5) to 35% methanol/65% ammonium acetate before switching to 67% methanol/33% H2O. Chromatogram peaks were identified by comparing the retention times with those of external standards and quantified by comparison of the integrated peak areas with those of the standards after interactive baseline correction.
To determine cardiac SAH content, freeze-clamped cardiac tissue (≈200 mg) was extracted as described previously.23 The neutralized extract was injected on a 300-mm C-18 Bondapak HPLC column (Waters). The elution gradient was identical to that used for adenosine determinations (see above); again, UV absorbance at 254 nm was measured, and peak areas were related to those of external standards.
Chemicals and Statistical Analysis
Iodotubercidin and EHNA were obtained from RBI. l-Homocysteine was obtained from Sigma. The remaining chemicals were purchased from Merck.
All results are expressed as mean±SD. The number of experiments averaged is indicated by n. For the evaluation of significant differences of data obtained under different conditions, paired and unpaired Student’s t tests were used. A value of P<.05 was taken to indicate a significant difference.
To estimate the goodness of fit of model predictions to measured data, CVw was calculated according to the following formula: Here, n is the number of individual observations in a total of four separate experimental approaches (no blockade or blockade of either ADA or AK or of the two enzymes), and npar is the number of parameters that are optimized (two parameters, ie, adenosine formation rate and AK activity). ¯yi is the model prediction for each of these approaches; yi, the individually measured values; and wi, a weighing factor (ie, the reciprocal of the respective mean for each approach). Thus, in principle, sums of squares were determined for the differences between the individually measured and predicted data, summed, and divided by the degrees of freedom to calculate a variance, which was then converted to a coefficient of variation. Since the mean data of the four approaches were very much different, the experimental data sets with the smaller means would have carried too small a weight. Therefore, each individual difference was multiplied by the weighing factor wi, and the sums of squares were divided by the sum of the four wi values.
Adenosine and Free AMP During Moderate Hypoxia
Basal coronary venous adenosine release from saline-perfused guinea pig hearts perfused with a normoxic (95% O2) medium was 59±28 pmol·min−1·g−1 (n=8). It rose 20-fold to 1164±415 pmol·min−1·g−1 after hypoxic (40% O2) perfusion had been initiated (Fig 1⇓). In normoxic hearts (95% O2), SAH accumulation as a measure of the concentration of free cytosolic adenosine was 4.78±2.52 pmol·min−1·mg−1. In hypoxic hearts (40% O2), it was 17 times higher (each n=4) (Fig 1⇓). In contrast to the substantial changes in adenosine release and cytosolic concentration, hypoxia had a lesser impact on cardiac energy status. Free cytosolic AMP increased 3.7-fold from 234 to 876 nmol/L (Fig 1⇓), Pi rose 3.8-fold from 0.76±0.33 to 2.87±0.75 mmol/L, and free energy of ATP hydrolysis declined from −62.7±1.2 to −57.3±0.8 kJ/mol when switching from normoxic to hypoxic perfusion (n=8).
When AK and ADA are effectively blocked by EHNA and iodotubercidin (5 and 10 μmol/L, respectively), adenosine release provides an estimate of the rate of adenosine formation (AMP→adenosine). Hypoxia induced a 4-fold increase in adenosine formation (n=8), similar to the increase observed in free AMP but much smaller than the rise seen in adenosine release in the absence of the blockers (Fig 1⇑). Compared with basal adenosine release (59 pmol·min−1·g−1) in the normoxic heart, adenosine formation (1615 pmol·min−1·g−1) was 25 times higher. Thus, the majority of adenosine formed is metabolized or rephosphorylated under normoxic conditions; only a minor part is released from the heart.
Flux Through CK and ATP Synthase
Determination of free cytosolic AMP depends on CK and myokinase equilibrium, since neither free ADP nor free AMP can be directly measured. Flux through CK was compared with oxidative ATP synthesis to verify CK equilibrium. Using 31P-NMR magnetization transfer techniques (see “Materials and Methods”), flux through CK in the normoxic heart was determined to be 62±14 μmol·min−1·g−1 (n=3). This contrasted with an O2 consumption of 5.4±2.4 μmol·min−1·g 1, which was equivalent to an ATP synthesis rate of 32±15 μmol·min−1·g−1. In hypoxia (40% O2), flux through CK was unchanged (76±19 μmol·min−1·g−1), whereas O2 consumption (2.2±1.1 μmol·min−1·g−1) and, in consequence, ATP synthesis were reduced by >50%. Thus, the unidirectional CK flux measured was several-fold greater than oxidative ATP synthesis, even in hypoxia, consistent with CK equilibrium.
Relation Between Free AMP and Adenosine Release and Formation
Hearts were perfused at different levels of O2 supply (medium equilibrated with 95%, 60%, 40%, 20%, or 10% O2; for each level of hypoxia, n=4 or 5) to test (1) whether there was a linear relationship between free AMP and adenosine formation and (2) whether the relative increase in coronary venous adenosine release was always much greater than the increase in free AMP. At each level of hypoxia, after the determination of adenosine release and cardiac energy status, adenosine formation was measured in the presence of iodotubercidin and EHNA. As shown in the Table⇓, a stepwise reduction of O2 supply from 95% to 20% O2 resulted in a significant and monotonic increase in adenosine release and in free cytosolic AMP, ADP, and Pi, whereas ATP, the free energy of ATP hydrolysis, and left ventricular developed pressure declined. pHi and free cytosolic Mg2+ remained unchanged (Table⇓). Iodotubercidin and EHNA had no impact on myocardial phosphocreatine, had only a small effect on ATP, and decreased free AMP by <20%. When reducing the O2 supply to 10% O2, adenosine formation, but not adenosine release or free AMP, increased further (Table⇓). Only at this level of hypoxia did iodotubercidin and EHNA induce a small rise in free AMP (+20%).
When varying the O2 supply from 95% to 10%, there was a close linear relation between adenosine formation and free AMP. Regression analysis revealed a regression coefficient R2 of .71 (P<.0001) and a y-axis intercept that was not significantly different from 0 (Fig 2⇓, top). Thus, a doubling of free AMP resulted in an approximate doubling of adenosine formation (Fig 2⇓, middle). In striking contrast to this finding, the relative increase in adenosine release was much higher than the increase in free AMP at all levels of hypoxia in every single experiment (Fig 2⇓, bottom).
In the normoxic heart, only a small fraction of the adenosine formed within the heart was released into the venous effluent, mainly because of the intracellular metabolism of adenosine via AK (see above). When calculating the percentage of adenosine released at different levels of O2 supply, it became apparent that hypoxia induced a substantial rise in the fractional release of adenosine. The fraction reached a maximum at a free AMP of 1 μmol/L (Fig 3⇓). While in the normoxic heart (95% O2) only 6% of adenosine formed was released, this percentage rose >3-fold in moderate hypoxia (40% O2).
Model Analysis I
The decrease in adenosine metabolism during hypoxia could have been due to saturation of AK or ADA or to inhibition of enzyme activity, eg, of AK. To exclude the possibility of enzyme saturation in the hypoxic myocardium, a comprehensive mathematical model of cardiac adenosine metabolism21 that predicted coronary venous adenosine release at a given adenosine formation rate was used. The modeled rates of the AK and ADA reactions were controlled by mass action effects that included partial enzyme saturation (Michaelis-Menten kinetics), without any change in enzyme activity. As can be seen in Fig 3⇑, the model adequately predicted the fractional release of adenosine in normoxia (95% O2). However, when adenosine formation was increased in proportion to the increase in AMP in the hypoxic myocardium (60% to 10% O2), the observed rise in the fractional adenosine release was not at all predicted. Thus, the measured rise in the fraction of adenosine released could not be explained by enzyme saturation.
ADA and AK in Normoxia and Hypoxia
To test whether endogenous inhibition of AK could explain the decrease in adenosine metabolism induced by hypoxia, AK and ADA were selectively blocked in two groups of hearts. In normoxic hearts (95% O2), blockade of AK (+iodotubercidin) alone induced a 7-fold increase in adenosine release, whereas blockade of ADA (+EHNA) alone had a significantly smaller effect (2.5-fold) (Fig 4⇓). Combined blockade of AK and ADA increased adenosine release at least 20-fold. Thus, in normoxic hearts, adenosine release represented only 5% of adenosine formation, and the enzyme mainly responsible for adenosine metabolism was AK.
The results were quite different in hypoxic hearts (40% O2), where basal adenosine release was much higher than in normoxic control hearts (1160 versus 69 pmol·min−1·g−1). In hypoxic hearts (Fig 4⇑), blockade of AK alone resulted only in a 2.4-fold increase in adenosine release, whereas inhibition of ADA induced a 3.1-fold rise. Blocking both enzymes showed that adenosine formation during hypoxia was 6490 pmol·min−1·g−1, ie, 5.6-fold greater than adenosine release in the absence of the blockers. These results show that the AK pathway metabolized fractionally less of the total adenosine formed in the hypoxic heart than in the normoxic heart. Under these conditions, the percentage metabolized by ADA exceeded that metabolized by AK.
Model Analysis II
Since the experimental data suggested hypoxia-induced inhibition of AK, the comprehensive model of adenosine metabolism21 was used to assess the extent of enzyme inhibition. First, as shown in Fig 5⇓, top, the rates of adenosine release measured in the normoxic heart (open bars) were compared with model predictions, and the summed error of the model prediction was expressed as CVw (see “Materials and Methods”). At a given adenosine formation rate of 2.3 nmol·min−1·g−1 and 100% activity of AK (hatched bars), the model accurately predicted adenosine release in the absence and presence of iodotubercidin and EHNA; the CVw was 8.3%. However, reducing Vmax of AK to 6% of its normal value (crosshatched bars) decreased the accuracy of the model predictions for basal adenosine release and the effects of EHNA (see arrows), increasing the summed error to 49%. This suggested that a high activity of AK in the normoxic heart was mandatory for the observed low adenosine release rates.
Second, measured and predicted release rates were compared in the hypoxic myocardium (Fig 5⇑, bottom). Adenosine formation was taken to be 6.5 nmol·min−1·g−1, consistent with the rate of adenosine formation measured in hypoxia (see Fig 1⇑). When 100% of normal activity of AK was assumed, the model failed to predict both the high adenosine release observed in hypoxia and the major effect of ADA blockade by EHNA (Fig 5⇑; see arrows); the summed error was 26%. However, when decreased activity of AK was modeled (6% of normal), the predictions were consistent with the hypoxia data, reducing the summed error to 13%.
To obtain the single model solution providing the best fit of the hypoxia data (Fig 5⇑, bottom), adenosine formation was systematically varied from 1 to 15 nmol·min−1·g−1, and the Vmax of AK was varied from 100% to 0.1% of normal, while all other model parameters were held constant. The summed errors (CVw) between the predicted and the measured data were plotted as a three-dimensional surface (Fig 6⇓). A low CVw represents a low error and hence an accurate fit. Both high (>10 nmol·min−1·g−1) and low (<5 nmol·min−1·g−1) adenosine formation rates resulted in increased errors. Full AK activity (100%) also caused a major error. Only when AK activity was reduced to well below 20% of its normal value was the error decreased significantly. The best fit to the hypoxia data, as indicated by the lowest CVw of 13%, was observed at an adenosine formation of 6.5 nmol·min−1·g−1 and an AK activity of 6% of normal. At still lower AK activities, the error increased again. Thus, the model results indicate that AK is inhibited by ≈94% because of hypoxia (40% O2).
There are two main findings in the present study: (1) Adenosine formation, measured as the coronary venous adenosine release in the presence of blockers of ADA (EHNA) and AK (iodotubercidin), was proportional to free cytosolic AMP (with AMP ranging from 200 to 3000 nmol/L), suggesting that 5′-nucleotidase is not modulated by hypoxia and that flux through 5′-nucleotidase is almost exclusively dependent on AMP substrate concentration. (2) Hypoxia causes a pronounced inhibition of the enzyme AK, resulting in substantially increased cellular and coronary venous efflux of adenosine formed by the heart. Thus, the physiological purpose of the normally high turnover rate of the substrate cycle between AMP and adenosine is that hypoxic inhibition of AK transforms this cycle into an amplification system that can translate small changes in AMP and adenosine formation into major changes in intracellular and extracellular adenosine.
Regulation of 5′-Nucleotidase
The present study provides the first direct estimates of total adenosine formation in the intact heart both during normoxia and hypoxia. Previously, we had demonstrated that upon blockade of all pathways metabolizing adenosine, ie, AK and ADA, with highly specific and efficient inhibitors, coronary venous effluent adenosine release is a direct measure of total adenosine formation.18 We have now applied this technique to the hypoxic myocardium and combined it with 31P-NMR measurements of high-energy phosphates and pHi to determine simultaneously free cytosolic AMP and the rate of adenosine formation. By use of this approach, a close linear relation between free AMP and adenosine formation was observed, indicating that flux through 5′-nucleotidase is controlled by mass action at the enzyme over a wide range.
This conclusion rests on three assumptions: (1) The adenosine formation rate determined, ie, adenosine release in the presence of ADA and AK blockade, represents cytosolic 5′-nucleotidase activity. (2) The enzyme blockade applied was both selective and efficient. (3) The free AMP calculated from 31P-NMR data indeed represents the cytosolic AMP concentration. These assumptions appear to be valid as discussed below:
(1) All available evidence indicates that flux through cytosolic 5′-nucleotidase contributes at least 70% to total adenosine formation; the remainder is being produced by ecto-5′-nucleotidase and SAH-hydrolase. In a previous study, in the normoxic heart, blockade of SAH-hydrolase had no effect on the measured adenosine formation rate, indicating that the contribution of this pathway to total adenosine formation is rather small.18 Blocking ecto-5′-nucleotidase in the normoxic heart had either no effect29 or only a small effect30 on adenosine release during normoxia and hypoxia, indicating that extracellular adenosine formation is low (0.4 nmol·min−1·g−1 in normoxia).29
(2) To determine total adenosine formation, coronary venous adenosine release (calculated as venous adenosine concentration×coronary flow) was measured in the presence of 10 μmol/L iodotubercidin to block AK and 5 μmol/L EHNA to block ADA. Both inhibitors are well characterized and have been used by other investigators. They have no impact on the cardiac energy status and do not interfere with cardiac adenosine receptors.18 In the normoxic guinea pig heart, iodotubercidin was already fully effective in blocking AK at a concentration as low as 1 μmol/L.18 This is consistent with an apparent Ki of ≈9 nmol/L reported by others.31 By increasing the iodotubercidin concentration 10-fold, a fully effective blockade, ie, >99% inhibition, can be assumed also in the hypoxic myocardium. Similarly, EHNA had been fully effective in the normoxic myocardium at 5 μmol/L,18 consistent with a low Ki observed by others (1.2 nmol/L).32 On the basis of the low Ki, 5 μmol/L EHNA should secure a >99% inhibition, even at cytosolic concentrations of up to 2 μmol/L.
(3) Since myocardial AMP is partially bound to proteins, the concentration of free cytosolic AMP cannot be directly measured in tissue extracts but has to be determined assuming myokinase and CK equilibrium. In the present study, using 31P-NMR saturation transfer techniques, we demonstrated that the CK flux is several-fold greater than the ATP turnover rate, both in the normoxic and the hypoxic myocardium. Therefore, CK equilibrium is well supported. Myokinase Vmax is high, and equilibrium has been assumed by many investigators in the field (eg, see References 7 through 97 8 9 , 15, and 33). Since the net flux from ADP to AMP is relatively small during steady state conditions, a low turnover of myokinase would already ensure equilibrium.
The conclusion that flux through cytosolic 5′-nucleotidase in the intact heart is regulated by mass action, ie, the free cytosolic AMP, at concentrations ranging from 200 to 3000 nmol/L, is consistent with studies of the purified enzyme.11 13 25 These studies reported Km values of cytosolic 5′-nucleotidase activity ranging from 55 μmol/L to 2.6 mmol/L, thus, at least two orders of magnitude higher than cytosolic free AMP. Since neither pHi nor Mg2+ (Table⇑) changed when switching from normoxic to hypoxic medium, the dependence of 5′-nucleotidase activity on these parameters does not affect our results. In the enzyme purified from dog heart, doubling free ADP from 40 to 80 μmol/L increased the enzyme activity by 50%, suggesting enzyme activation when starting hypoxic perfusion.11 However, we could not find any evidence for this conclusion in the intact guinea pig heart.
When relating the rate of adenosine formation (1.7 nmol·min−1·g−1) in the normoxic heart to the substrate concentration (AMP, 240 nmol/L), an index of cytosolic 5′-nucleotidase activity can be calculated (7 mL·min−1·g−1). Since AMP concentration is well below the reported Km for guinea pig heart (55 μmol/L),25 this index is equivalent to Vmax/Km. In dog and guinea pig hearts, Vmax (24.9 and 14.3 nmol·min−1·mg protein−1, respectively) and Km (2.6 and 0.055 mmol/L, respectively) values of cytosolic 5′-nucleotidase that result in Vmax/Km terms of 1.20 and 32.5 mL·min−1·g−1, respectively (assuming 125 mg protein/g wet wt), have been reported.11 25 Thus, the activity terms determined for the intact guinea pig heart and the enzyme in vitro agree well, supporting the concept that the adenosine formation measured indeed represents the flux through cytosolic 5′-nucleotidase at a given substrate concentration.
Hypoxia-Induced Inhibition of AK
Since the data of the present study rule out activation of 5′-nucleotidase in the hypoxic guinea pig heart (see above), the major rise in adenosine release seen in the presence of only small changes of cytosolic AMP could in principle be due to either saturation of AK or ADA or to hypoxia-induced inhibition of AK.
When switching to hypoxia, the fraction of adenosine released into the coronary venous effluent increased >3-fold (Fig 3⇑), indicating a substantial decrease in adenosine metabolization in the hypoxic myocardium. This could be due to enzyme saturation. However, the enzyme activities (Vmax) for both AK (ranging from 74 to 228 nmol·min−1·g−1) and ADA (2700 nmol·min−1·g−1) measured in guinea pig heart homogenates25 34 were several-fold higher than the observed adenosine formation rate even in severe hypoxia (Table⇑), making enzyme saturation rather unlikely. A similar conclusion can be reached using a comprehensive model of adenosine metabolism.21 Augmenting adenosine formation in the model up to 8-fold did not reduce the fraction being salvaged by the two enzymes (Fig 3⇑). For this analysis, Vmax and Km of AK and ADA were set to 137 and 950 nmol·min−1·g−1 and 2.5 and 83 μmol/L, respectively, consistent with in vitro data.25 Thus, a model that did not include direct inhibition of AK could not describe the hypoxia data in Fig 3⇑.
The model accurately predicted the very low coronary venous adenosine release of the normoxic myocardium (Fig 5⇑). However, reducing Vmax of AK by only 50% or doubling Km increased the predicted adenosine release and induced a major error compared with the measured results (data not shown). Therefore, in the normoxic heart, high enzyme activities of AK are a prerequisite to explain the low adenosine release observed.
Both measured data and model predictions strongly suggest that hypoxia-induced inhibition of AK induces the observed potentiation of adenosine release. In the normoxic heart, adenosine release was increased 7-fold by pharmacological blockade of AK but only 2.4-fold in the hypoxic heart. In contrast, blockade of ADA by EHNA increased adenosine release 2.5- to 3-fold, both in the normoxic and hypoxic heart. Thus, in the normoxic heart, AK was the enzyme mainly responsible for adenosine metabolization, whereas in the hypoxic heart, its relative importance was considerably reduced. Since blocking an enzyme that is already endogenously inhibited (ie, AK in hypoxia) will have a lesser effect on adenosine release than blocking an enzyme that is not inhibited (ie, AK in normoxia), it follows that the more marked increase in adenosine release with AK blockade during normoxia versus hypoxia reflects intrinsic inhibition of AK with hypoxia.
To assess the degree of inhibition necessary to explain the observed results, a mathematical model of adenosine metabolism21 was used. The experimentally measured hypoxia data in the absence and presence of selective pharmacological blockade of ADA and AK (Fig 4⇑) were fitted by the model. The best fit was obtained when setting adenosine formation during hypoxia to 6.5 nmol·min−1·g−1 and reducing Vmax of AK to 6 nmol·min−1·g−1, ie, 6% of its normal value (Fig 6⇑). It is also possible that hypoxia causes inhibition of AK via a decrease in its affinity for adenosine substrate, since reducing Vmax to 6% or increasing Km 16-fold to 40 μmol/L (data not shown) fitted the data equivalently. Thus, the model predicted a 94% inhibition of AK in hypoxia (40% O2).
What are the factors that mediate inhibition of AK by hypoxia? AK could in principle be inhibited by ADP, AMP,35 Pi,36 adenosine substrate inhibition,37 lack of ATP substrate,34 35 rise in cytosolic Mg2+ concentration,35 and alkalinization.35 In the present study, free ADP rose from 40 to maximally 90 μmol/L (Table⇑). A similar rise in ADP resulted in a small inhibition of purified AK from human placenta35 ; in the same preparation, AMP acted as a competitive inhibitor, albeit at concentrations two orders of magnitude above the measured cytosolic AMP concentrations. Partial substrate inhibition of purified AK was observed37 when adenosine exceeded 5 to 10 μmol/L; however, in the present study, cytosolic adenosine rose to 1 to 2 μmol/L only. The small decrease in ATP seen during hypoxia is unlikely to influence AK, since the Km of AK for ATP is extremely low (0.1 mmol/L).34 Lack of ATP might, however, explain the blockade of AK seen in rat cardiomyocytes after energy deprivation by iodoacetate.20 Recently, a rise in Pi from 0 to 16 mmol/L was shown to reduce the activity of purified AK by 50%.36 However, in the present study, Pi increased from 2.4 to only 4.6 mmol/L (40% O2). Mg2+ is a powerful inhibitor of AK35 ; however, free cytosolic Mg2+ as determined by 31P-NMR did not change in the present study, nor did pHi (Table⇑). It remains unclear whether the combined effect of substrate inhibition and rise in free ADP and Pi can explain the hypoxia-induced inhibition of AK observed or whether other factors are involved.
Role of the Substrate Cycle Between AMP and Adenosine
The data of the present study enable a comprehensive description of flux rates and cytosolic concentrations of the cardiac adenosine metabolism in the normoxic and hypoxic guinea pig heart. As summarized in Fig 7⇓, in the normoxic heart, >80% of adenosine formed within the heart was rephosphorylated to AMP, ≈15% was deaminated to inosine, and only 5% was released into the coronary venous effluent. Under these conditions, the cytosolic adenosine concentration was ≈60 nmol/L, in good agreement with the findings of others10 ; the interstitial concentration predicted by the model was 40 nmol/L; and the coronary venous concentration measured 9 nmol/L. Switching to hypoxia (40% O2) induced a 3- to 4-fold rise in free AMP and adenosine formation. However, as a result of hypoxia-induced inhibition of AK, flux through this enzyme hardly increased, whereas cytosolic adenosine rose 17-fold, leading to a similar rise in flux through ADA and adenosine release. Thus, the hypoxia-induced inhibition of AK ultimately translated a small change in cytosolic AMP and adenosine formation into a major rise in interstitial adenosine and venous adenosine release.
When free AMP was further increased, eg, by severe hypoxia (10% O2), both cytosolic AMP and adenosine release increased to a similar extent (Table⇑), indicating that hypoxia-induced inhibition of AK had attained a maximum at moderate hypoxia (40% O2) (see Fig 3⇑). Consequently, the further increase of adenosine release upon more severe hypoxia (10% to 40% O2) was mainly caused by the increase of the substrate AMP.
We have previously suggested that the AMP-adenosine metabolic cycle couples cytosolic adenosine closely to the cardiac energy status.18 We now provide first evidence that this “futile cycle” serves an important physiological role: it amplifies small changes in free AMP into a major rise in adenosine. Endogenous inhibition of the enzyme translates a minor decrease in the cardiac energy status into a rise in cytosolic adenosine, which very likely is the molecular basis for the known high sensitivity of the cardiac adenosine system to changes in the O2 supply-to-demand ratio.23 In addition, the enhanced net degradation of AMP due to AK inhibition may serve as an important sink for cytosolic ADP during ischemia, resulting in improved free energy of ATP hydrolysis at a level that is sufficient to maintain ion homeostasis.38
The principles of regulation of the adenosine metabolism elaborated in the guinea pig heart appear to have a broader application. Aside from the heart, AK is found in almost every organ and cell.39 40 In liver and brain, pharmacological blockade of AK was already shown to result in a significant rise in adenosine, suggesting a rapid turnover of the AMP-adenosine metabolic cycle.41 42 Endogenous inhibition of AK through small changes in energetics may therefore be a general mechanism of rapid signal amplification for adenosine during impaired tissue oxygenation.
Selected Abbreviations and Acronyms
|CVw||=||weighted coefficient of variation|
|HPLC||=||high-performance liquid chromatography|
|NMR||=||nuclear magnetic resonance|
This study was supported by the Deutsche Forschungsgemeinschaft (SFB 242, E2) and by the Center for Biological and Medical Research (Biomedizinisches Forschungszentrum) of the Heinrich-Heine-University Düsseldorf. Development of the mathematical model was supported by National Institutes of Health grant RR-01243. A NATO travel grant to J. Schrader and K. Kroll facilitated the trans-Atlantic collaboration in this project. The authors thank Eva Bergschneider for her expert help in the biochemical analysis of myocardial tissue and coronary venous effluent.
Presented in part in abstract form at the annual Experimental Biology meeting, Washington, DC, April 14-17, 1996 (FASEB J. 1996;10:A327).
- Received February 10, 1997.
- Accepted April 30, 1997.
- © 1997 American Heart Association, Inc.
Deussen A, Borst M, Kroll K, Schrader J. Formation of S-adenosyl-homocysteine in the heart, II: a sensitive index for regional myocardial underperfusion. Circ Res.. 1988;63:250-261.
Schrader J. Adenosine: a homeostatic metabolite in cardiac energy metabolism. Circulation.. 1990;81:389-391.
Stepp DW, Van Bibber R, Kroll K, Feigl EO. Quantitative relation between interstitial adenosine concentration and coronary blood flow. Circ Res.. 1996;79:601-610.
Dobson JG Jr, Ordway RW, Fenton RA. Endogenous adenosine inhibits catecholamine contractile responses in normoxic hearts. Am J Physiol.. 1986;251:H455-H462.
Neumann J, Gupta RC, Jones LR, Bodor GS, Bartel S, Krause EG, Pask HT, Schmitz W, Scholz H, Watanabe AM. Interaction of b-adrenoceptor and adenosine receptor agonists on phosphorylation: identification of target proteins in mammalian ventricles. J Mol Cell Cardiol.. 1995;27:1655-1667.
Headrick JP, Willis RJ. Adenosine formation and energy metabolism: a 31P-NMR study in isolated rat heart. Am J Physiol.. 1990;258:H617-H624.
Deussen A, Borst M, Schrader J. Formation of S-adenosylhomocysteine in the heart, I: an index of free intracellular adenosine. Circ Res.. 1988;63:240-249.
Darvish A, Metting PJ. Purification and regulation of an AMP-specific cytosolic 5′-nucleotidase from dog heart. Am J Physiol.. 1993;264:H1528-H1534.
Skladanowski AC, Smolenski RT, Tavenier M, De Jong JW, Yacoub MH, Seymour AML. Soluble forms of 5′-nucleotidase in rat and human heart. Am J Physiol.. 1996;270:H1493-H1500.
Itoh R, Oka J, Ozasa H. Regulation of rat heart cytosol 5′-nucleotidase by adenylate energy charge. Biochem J.. 1986;235:847-851.
Kitakaze M, Hori M, Minamino T, Takashima S, Komamura K, Node K, Toshinao K, Morioka T, Sato H, Inoue M, Kamada T. Evidence for deactivation of both ectosolic and cytosolic 5′-nucleotidase by adenosine A1 receptor activation in the rat cardiomyocates. J Clin Invest.. 1994;94:2451-2456.
Bak MI, Ingwall JS. Acidosis during ischemia promotes adenosine triphosphate resynthesis in postischemic rat heart: in vivo regulation of 5′-nucleotidase. J Clin Invest.. 1994;93:40-49.
Headrick JP, Willis RJ. 5′-Nucleotidase activity and adenosine formation in stimulated, hypoxic and underperfused rat heart. Biochem J.. 1989;261:541-550.
Headrick JP, Matherne GP, Berr SS, Han DC, Berne RM. Metabolic correlates of adenosine formation in stimulated guinea pig heart. Am J Physiol.. 1991;260:H165-H172.
Kroll K, Decking UKM, Dreikorn K, Schrader J. Rapid turnover of the AMP-adenosine metabolic cycle in the guinea pig heart. Circ Res.. 1993;73:846-856.
Wagner DR, Bontemps F, van den Berghe G. Existence and role of substrate cycling between AMP and adenosine in isolated rabbit cardiomyocytes under control conditions and in ATP depletion. Circulation.. 1994;90:1343-1349.
Kroll K, Deussen A, Sweet IR. Comprehensive model of transport and metabolism of adenosine and S-adenosylhomocysteine in the guinea pig heart. Circ Res.. 1992;71:590-604.
Szondy Z, Newsholme EA. The rate of the AMP/adenosine substrate cycle in concanavalin-A-stimulated rat lymphocytes. Biochem J.. 1989;261:739-742.
Decking UKM, Arens S, Schlieper G, Schulze K, Schrader J. Dissociation between adenosine release, cardiac energy status and O2 consumption in working guinea pig hearts. Am J Physiol.. 1997;272:H371-H381.
Schrader WP, West CA. Localization of adenosine deaminase and deaminase complexing protein in rabbit heart: implications for adenosine metabolism. Circ Res.. 1990;66:754-762.
Decking UKM, Reffelmann T, Schrader J, Kammermeier H. Hypoxia-induced activation of KATP-channels limits energy depletion in the guinea pig heart. Am J Physiol.. 1995;269:H734-H742.
Borst MM, Schrader J. Adenine nucleotide release from isolated perfused guinea pig hearts and extracellular formation of adenosine. Circ Res.. 1991;68:797-806.
Hurley MC, Lin B, Fox IH. Regulation of deoxyadenosine and nucleoside analog phosphorylation by human placental adenosine kinase. J Biol Chem.. 1985;260:15675-15681.
Kang YH, Mallet RT, Bünger R. Coronary autoregulation and purine release in normoxic heart at various cytoplasmic phosphorylation potentials: disparate effects of adenosine. Pflügers Arch.. 1992;421:188-199.
Palella TD, Andres CM, Fox IH. Human placental adenosine kinase: kinetic mechanism and inhibition. J Biol Chem.. 1980;255:5264-5269.
Gorman MW, He M-X, Hall CS, Sparks HV. Inorganic phosphate as regulator of adenosine formation in isolated guinea pig hearts. Am J Physiol.. 1997;272:H913-H920.
Miller RL, Adamczyk DL, Miller WH. Adenosine kinase from rabbit liver, I: purification by affinity chromatography and properties. J Biol Chem.. 1979;254:2339-2345.
Kroll K, Gustafson LA, Kinzie DJ. Open system kinetics of myo-cardial phosphoenergetics during coronary underperfusion. Am J Physiol. In press.
Arch JRS, Newsholme EA. Activities and some properties of 5′-nucleotidase, adenosine kinase and adenosine deaminase in tissues from vertebrates and invertebrates in relation to the control of the concentration and the physiological role of adenosine. Biochem J.. 1978;174:965-977.
Bontemps F, van den Berghe G, Hers HG. Evidence for a substrate cycle between AMP and adenosine in isolated hepatocytes. Proc Natl Acad Sci U S A.. 1983;80:2829-2833.