Increased hypoxic stress decreases AMP hydrolysis in rabbit heart

Transcription

1 Cardiovascular Research 44 (1999) locate/ cardiores locate/ cardiores Increased hypoxic stress decreases AMP hydrolysis in rabbit heart a,b, a,c a b Lori A. Gustafson *, Coert J. Zuurbier, John E. Bassett, Jan Paul F. Barends, b a a,1 Johannes H.G.M. van Beek, James B. Bassingthwaighte, Keith Kroll a Center for Bioengineering, University of Washington, Seattle, WA 98195, USA b Laboratory for Physiology, Institute of Cardiovascular Research, Vrije Universiteit, Amsterdam, The Netherlands c Department of Experimental Anesthesiology, Universiteit van Amsterdam, Amsterdam, The Netherlands Received 24 November 1998; accepted 16 June 1999 Abstract Objective: AMP conversion to adenosine by cytosolic 59-nucleotidase (5NT) or to IMP by AMP deaminase determines the degree of nucleotide degradation, and thus ATP resynthesis, during reoxygenation. To elucidate the regulation of AMP hydrolysis during ischemia, 31 data from P NMR spectroscopy and biochemical analyses were integrated via a mathematical model. Since 5NT is downregulated during severe underperfusion (5% flow), we tested 5NT regulation during less severe underperfusion (10% flow) and then made the 31 perfusate hypoxic to see if the greater stress reactivated 5NT. Methods: P NMR spectra and coronary venous effluents were obtained from Langendorff-perfused rabbit hearts subjected to two 30-min periods of underperfusion (10% flow); the second period with or without additional hypoxia (30% O 2). Data were analyzed with a mathematical model describing the kinetics of myocardial energetics and metabolism. Results: A single 30-min period of 10% flow causes downregulation of AMP hydrolysis and the data from the second period of underperfusion are best described by lower 5NT activity, even in the presence of extra hypoxia. Thirty percent less purines appear in the venous effluent than predicted by the phosphoenergetics (PCr and ATP) when IMP is not allowed to accumulate by the model, however the model indicates that a constant accumulation of IMP via AMP deaminase could explain the discrepancy between expected and measured purines in the venous effluent. Conclusions: While AMP hydrolysis to adenosine is prominent in early ischemia and acts to preserve cellular energy potential, during a second ischemic period, nucleotides are conserved by the stable inhibition of AMP hydrolysis. Furthermore, during 10% flow conditions, nucleotides are conserved, possibly via an IMP-accumulatory pathway Elsevier Science B.V. All rights reserved. Keywords: Adenosine; Computer modeling; Energy metabolism; Hypoxia/ anoxia; Ischemia 1. Introduction myocardial contraction and relaxation. Prolonged depletion of ATP during ischemia leads to irreversible damage In the heart, an imbalance between oxygen supply and during reperfusion [1], thus the treatment of myocardial demand causes an immediate fall in PCr, followed by a net ischemia or the provision of cardiac protection during hydrolysis of ATP and an increase in the concentration of coronary bypass will be aided by preventing ATP deple- AMP. AMP can then be dephosphorylated to adenosine by tion. Adenosine is membrane permeable and its release the enzyme 59-nucleotidase (5NT, E.C ), or deami- evokes cardioprotective mechanisms such as coronary nated to IMP by the enzyme AMP deaminase (AMPD, vasodilation, improving the tissue oxygen supply [2]. E.C ). Cytosolic ATP availability is essential for Ischemic preconditioning is mediated by adenosine receptor activation. Most importantly, AMP hydrolysis to adenosine provides a mechanism whereby the phosphoryl- *Corresponding author. Laboratory for Physiology, Institute for Car- ation potential is preserved by mass balance during comdiovascular Research, Vrije Universiteit, Van der Boechorststraat 7, 1081 promised energy supply (i.e. low energy nucleotides are BT Amsterdam, The Netherlands. Tel.: ; fax: removed) [3]. While AMP hydrolysis is beneficial during address: lorig@physiol.med.vu.nl (L.A. Gustafson) 1 Deceased 15 July, Time for primary review 28 days / 99/ $ see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S (99)

2 334 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) isoform of 5NT (cn-ii), have been included in the present model metabolically depicted in Fig. 1. The results show that a single 30-min period of underperfusion at 10% flow downregulates AMP hydrolysis and that additional hypoxic stress elevates cytosolic AMP levels further but purine efflux increases only a little. The data are best described by the mathematical model with a lowered activity of 5NT (cn-i) and a constant AMP deaminase activity which allows IMP accumulation. The conclusion is that 5NT is an important regulator of AMP hydrolysis. 2. Methods 2.1. Isolated heart preparation All experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH Publication No , revised 1996) and approved by the institutional animal experimental ethics committee. New Zealand White rabbits ( kg) were sedated with acetylpromazine (0.8 mg/ kg, s.c.), and anesthetized with ketamine (45 mg/kg, i.m.) plus xylazine (4 mg/kg, i.m.). The rabbits were tracheotomized and ventilated with room air supplemented with oxygen. After opening the thorax and administration of heparin (200 U, i.v.), the aorta was cannulated in situ and perfusion was started, followed by excision of the heart. In situ cannulation was im- plemented to prevent any cardioplegic or preconditioning effects upon the myocardial metabolism. The hearts were acute ischemia, the continued loss of nucleosides during prolonged ischemia will lead to the depletion of nucleotide pools [4]. The regulation of AMP hydrolysis is therefore important for myocardial survival during prolonged ischemia and cardioplegia. AMP hydrolysis is downregulated during severe and prolonged underperfusion [5]. Purine efflux at similar cytosolic AMP concentrations was depressed during the second of two identical periods of underperfusion. Model analysis indicated that 5NT was downregulated late in the first period of underperfusion [5]. Our question is: Does this downregulation of 5NT persist if the energy supply mechanisms are stressed further by reducing the po2 during underperfusion? In order to test the hypothesis that 5NT is downregulated during prolonged underperfusion yet can become re-upregulated during greater hypoxic stress, we performed experiments designed to increase cytosolic AMP levels by using hypoxic perfusate (30% oxygen, compared to 95% in controls) during the second period of underperfusion (10% flow). The rate of adenosine formation from AMP in the cytosol is influenced by at least four factors: the AMP concentration, and the activities of three enzymes which influence its concentration: adenosine kinase (AK), which rephosphorylates adenosine to AMP, AMP deaminase, which deaminates AMP to IMP, and cytosolic, AMPpreferring 5NT (cn-i), which dephosphorylates AMP to adenosine [6 8]. Our previous mathematical model [5] did not include AMP deaminase, so this pathway, and the IMP inosine pathway catalyzed by the IMP-preferring Fig. 1. Model description of myocardial phosphoenergetics and nucleoside metabolism. Abbreviations: Cr, creatine; PCr, phosphocreatine; ATP, adenosine triphosphate; ADP, adenosine diphosphate, AMP, adenosine monophosphate, Cr kinase, creatine kinase; DrATP, rate of ATP synthesis minus rate of ATP hydrolysis; 5NT, AMP-preferring isoform of 59-nucleotidase (cn-i); 5NT-II, IMP-preferring isoform of 59-nucleotidase (cn-ii); Ado kinase, adenosine kinase; Ado deaminase, adenosine deaminase; P, inorganic phosphate; PS, permeability surface area products for membrane transport. i

3 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) Langendorff-perfused at 378C at a constant flow (perfusion analyzed by HPLC as previously described [5]. Total pressure mmhg) with non-recirculating, modified purine release was calculated by summing the purines Krebs Henseleit bicarbonate buffer containing (in mm) (adenosine1inosine1hypoxanthine) released during each 118 NaCl, 3.8 KCl, 1.2 KH2PO 4, 0.7 MgSO 4, 2.1 CaCl2 2- or 5-min period for the entire 30-min period of 0.1 EDTA, 25 NaHCO 3, 11 glucose, 5 pyruvate, and 0.1% underperfusion. Effluent lactate concentration was meabovine serum albumin, equilibrated with 95% O 2/ 5% CO2 sured with a YSI glucose analyzer equipped with a lactate using a membrane oxygenator, resulting in a ph of 7.35 membrane A fluid-filled latex balloon was inserted in the left ventricle, connected to a pressure transducer and inflated, 2.5. Model analysis yielding a systolic pressure between mmhg, with an end-diastolic pressure less than 5 mmhg. The hearts A mathematical model of myocardial phosphoenergetics were electrically paced at 180 beats per min. and nucleotide metabolism, adapted from [3] was used to Coronary venous effluent samples were collected as analyze the data. The metabolic processes described by the described previously [5]. The hearts were submerged in model are depicted in Fig. 1. A full description of the rate 378C perfusate in a cylinder encircled with a solenoid style equations, differential equations and the model parameter radiofrequency NMR coil. After the set-up was placed values can be found in the appendix of Ref. [3]. The model inside the magnet, the radiofrequency coil was tuned (81 describes the intracellular concentrations of PCr, Cr, ATP, MHz) and matched, the gradient coils were shimmed, and ADP, AMP, P i, adenosine, and inosine, the enzymes 31 a fully relaxed P NMR spectrum was acquired. Wet creatine kinase, myokinase, AMP-preferring and IMP-preventricular weight was determined after each experiment. ferring isoforms of cytosolic 5NT, AMP deaminase, adenosine kinase and adenosine deaminase, the membrane 2.2. Experimental protocol transport of adenosine and inosine, and Pi and Cr in exchange with an interstitial region. The processes where- Twelve rabbits were used to investigate AMP hydrolysis by ATP is synthesized (oxidative phosphorylation, glyduring two sequential and identical periods of underperfu- 21 colysis) and hydrolyzed (e.g. Ca and ion pumps, myofibsion. The control group of hearts (n56) were subjected to ril contraction, ATPases, homeostasis, etc.) were described two 30-min periods of underperfusion with perfusate as a continuous function, DrATP (DrATP5rate of ATP equilibrated with 95% O 2/ 5% CO2 at a flow of 10% of the synthesis2rate of ATP hydrolysis), providing a flexible baseline flow (approximately 3 ml/min, or 0.5 ml/min/g). means for describing the time-course and extent of the net The two periods of underperfusion were separated by a energy imbalance experienced during underperfusion and 20-min period of reperfusion at baseline flow. The hypoxia reperfusion [3]. The parameters describing DrATP were group (n56) underwent an identical procedure, except that estimated empirically using optimization procedures. The during the second period of underperfusion the perfusate baseline data showed a low, continuous outflow of adenowas equilibrated via the membrane oxygenator with 30% sine indicating a slightly greater hydrolysis than synthesis O 2, 5% CO2 and 65% N 2. of ATP during baseline conditions in the buffer perfused rabbit hearts. The current model allows for this small net 2.3. NMR spectroscopy ATP breakdown, differing thereby from the original model [3], which assumed baseline DrATP50. Phosphorus NMR measurements were obtained using a Permeability surface area (PS) products described trans- 4.7-Tesla superconducting magnet (Bruker) and a CSI membrane exchange of adenosine, inosine, Pi and creatine. spectrometer (GE-Omega) and analyzed using an auto- Enzyme dissociation constants and PS products were taken mated fitting routine [9] as previously described [3]. from literature values [3], with the exception of the Vmax of Intracellular P was determined and intracellular ph and 59-nucleotidase and AMP deaminase, which were deteri 21 the free intracellular Mg concentration were calculated mined by the model. The present model incorporates an as described [3]. Due to interference by extracellular Pi in alternative pathway of AMP hydrolysis to IMP by AMP the perfusion medium, baseline and reperfusion intracellu- deaminase. Literature values from the rabbit for AMP lar ph were assumed to equal 7.1 [10]. Cytosolic free deaminase [11] were used as starting values (i.e. Vmax560 [AMP] was calculated using creatine kinase and adenylate mmol/min/g, Km51.7 mm). The Km and Vmax of AMP kinase equilibrium expressions, adjusted to the calculated deaminase were then optimized prior to each simulation 1 21 H and Mg concentrations, and it was assumed that the run. The product of this reaction, IMP, was further allowed total concentration of PCr1Cr decreased linearly by 5% to be hydrolyzed to inosine by an IMP-preferring isoform during each of the two periods of underperfusion [5]. of 59-nucleotidase (cn-ii) [12]. To fit the model to the NMR and purine data, an 2.4. Determination of venous purines and lactate automated least squares optimization routine (SIMPLEX) was used to simultaneously fit the PCr, ATP, adenosine and Coronary venous effluent samples were collected and inosine curves by adjusting the Vmax of 5NT (predictions of

4 336 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) the Km were found to vary only slightly, see results) and tively, were: 145, 94, and 130 mm. Peak AMP con- the parameters of the DrATP function. All other model centrations for the first, second and second1hypoxia parameters were either held constant during fitting or were groups, respectively, were: 3.4, 1.8, and 4.5 mm. This changed according to direct measurements (flow, intracel- elevation of 5NT substrate concentration (i.e. AMP) due to 12 lular ph, Mg ). Because the model does not include hypoxia resulted in a higher purine efflux: (34.3) hypoxanthine, the coronary venous hypoxanthine data nmol/ g total purines as compared to 179 (15.0) nmol/ g were added to the inosine data for fitting with the model during the second period of underperfusion with normoxia. inosine data. Best fits were derived from the SIMPLEX Both levels of purine efflux, however, were significantly routine using various starting values to ensure avoidance of attenuated as compared to the purine efflux from the first local minima. Means and standard errors of the parameter period of underperfusion (i.e (24.8) nmol/ g). estimates were obtained by fitting the data from each Lactate efflux and the intracellular Pi concentration were individual experiment. Initial concentrations of ATP (6.0 elevated during the second period of underperfusion with mm), PCr (10.5 mm) and Cr (13.2 mm) used in the hypoxia as compared to normoxic underperfusion (P, modeling were taken from biochemical measurements of 0.05), while the cytosolic hydrogen ion concentration was freeze-clamped hearts performed previously in this labora- similar for the two periods. Diastolic pressure did not rise tory [3]. during the periods of underperfusion, but developed left ventricular pressure fell quickly to mmhg after the 2.6. Statistical analysis onset of underperfusion and was similar for the three different underperfusions. All data are presented as mean values6standard error of the mean (S.E.M. for n56). Mean values of the model results (i.e. Vmax of 5NT and the DrATP parameters) were 3.2. Model analysis determined from the separate model solutions for each experiment. Statistical comparisons between the first A second, consecutive period of underperfusion, under period of underperfusion and the second period of underperfusion, with and without hypoxia, were made using a repeated measures ANOVA with Tukey s post-hoc test for individual comparisons. A value of P,0.05 was considered to be indicative of statistical significance. normoxic conditions, caused an attenuation of purine release in the venous effluent. While this could indicate a downregulation of the activity of 5NT, under these con- ditions, the cytosolic concentration of AMP was also attenuated, thus, substrate limitation could be the cause of such attenuation. During a second period of underperfusion under hypoxic conditions, purine efflux increased, sug- 3. Results gesting a possible upregulation of 5NT. However, the simultaneous elevation of cytosolic AMP concentration 3.1. Purine efflux during increased hypoxic stress could have increased purine efflux by a simple mass action effect, without 5NT activation. We, therefore, analyzed the The NMR, purine and metabolic data for the two groups simultaneously-obtained NMR and purine data with a are presented in Fig. 2. The first period of underperfusion mathematical model which is able to differentiate between (10% of basal flow, approximately 0.5 ml/min/g) yielded the mass action effects of the AMP concentration and the very similar results for each group, therefore this data has enzymatic effects of 5NT (or AMPD). Simultaneous model been grouped (n512). The second periods of underperfu- fits to the PCr, ATP, adenosine and inosine (inosine1 sion, one under normoxic (95% oxygen) conditions, and hypoxanthine) data, using a four-region mathematical one under hypoxic (30% oxygen) conditions are, thus, model of myocardial energetics and enzyme kinetics, are directly comparable. During the first period of underperfu- given in Fig. 3. Separate fits were obtained from each of sion, PCr fell from 10.5 to 4.6 mm and then slightly the six individual experiments of the three different periods recovered toward baseline, while ATP slowly fell from 6 of underperfusion. The only parameters allowed to vary to 4.3 mm by the end of the first underperfusion period. during the fitting optimization were the Vmax of 5NT and During the second period of underperfusion, PCr fell to the the parameters describing the degree of energy imbalance same level (4.6 mm) and showed a similar recovery, while (DrATP). The means and standard errors of these analyses ATP only slightly decreased (from 4.3 to 3.9 mm). When are given in Table 1. The model analysis indicates that a hypoxia, however, was applied during the second period of 30-min period of 90% flow reduction induces a downreguunderperfusion, PCr fell to 2.7 mm, after a delay, and ATP lation of 5NT activity. Furthermore, the analysis indicates fell from 4.3 to 3.3 mm. The calculated concentrations of that, while the elevated AMP levels during increased ADP and AMP indicate that hypoxia during the second hypoxic stress do lead to an increase in purine efflux, the period of underperfusion indeed resulted in an elevation in data are best described by a lower 5NT activity. The the cytosolic concentrations of ADP and AMP. Peak ADP high-energy phosphate and purine data were best fit for the for the first, second and second1hypoxia groups, respec- first period of underperfusion with an average Vmax of 164

5 Fig. 2. Nuclear magnetic resonance, purine and metabolic data from two successive periods of underperfusion (10% baseline flow) with and without hypoxia. PCr (d) and ATP ( ). Calculated cytosolic ADP (d) and AMP ( ); AMP was multiplied by a factor of 10 to aid viewing. Intracellular P i (d). Intracellular ph (d). Lactate concentration in venous effluent (d). Purine concentration in venous effluent, adenosine (j), hypoxanthine1inosine (m). Total purine release for the first 30-min period was nmol/ g; for the second period (normoxic), nmol/ g and for the hypoxic second period, nmol/ g. Left ventricular developed pressure (d). L.A. Gustafson et al. / Cardiovascular Research 44 (1999)

6 338 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) Fig. 3. Simultaneous model fits to the PCr (d), ATP ( ), adenosine (h) and hypoxanthine1inosine (^) data using the four region model of myocardial energetics and enzymatic kinetics (Fig. 1). Data are indicated by the symbols, model fits by the lines. nmol/min/g for 5NT. The second period of underperfu- purine efflux at 10% flow was considerably lower (30%) sion, under normoxic conditions, was best fit with a than that predicted by initial runs with the model which significantly lower (P,0.05) Vmax of 129 nmol/min/g. The had been validated at 5% flow levels. This suggested that Vmax for the hypoxic second period of underperfusion was another pathway of AMP hydrolysis and accumulation was significantly lower (P,0.05) than for the first period of active at 10% flow and not at 5% flow. A post hoc aim of underperfusion, as well as for the second (normoxic) the model analysis, therefore, was to test the possibility period, and was best fit using a value of 85 nmol/min/g. that another enzyme, for example AMP deaminase, plays a Analysis of the initial part of the purine efflux curves regulatory role during ischemia-induced purine efflux. It indicated a high affinity of 5NT for its substrate during all has been shown, for example, that AMP degradation in three conditions, thus the Km for 5NT was set to 0.8 mm de-energized heart cells can occur through either deamina- during all model optimizations. Qualitatively similar re- tion (AMPD) or dephosphorylation (5NT) [14]. Thus, the sults were achieved when the V was set to a literature relative rates of the IMP and adenosine pathways reflect max value of 290 nmol/ min/ g [13], i.e. Km was the lowest the competition between AMPD and 5NT for AMP (Fig. under control conditions (3.0 mm), and highest for the 1). Since a decrease in AMPD activity during a second hypoxic group (10.9 mm) (Fig. 4). The fits obtained by period of underperfusion would also result in a decreased varying the K m, however, were poorer than the Vmax purine efflux, this pathway was added to the mathematical variations, due to a delay of the purine efflux curves. The model. model, however, best describes AMP hydrolysis, as a Exploratory modeling, performed to investigate the role whole, and has difficulty in discerning between Km or Vmax of AMP deamination to IMP by AMP deaminase, indi- effects. It remains unclear at this stage whether downregu- cated that the data were best described by allowing a lation of 5NT during underperfusion is due to a change in constant flux through AMPD, which was allowed to affinity for AMP or the V max. accumulate as IMP (Fig. 4). The analysis, which used as A surprising result of this study was that the measured determinants the ratio of inosine1hypoxanthine to adeno- Table 1 Model predictions of V max for 5NT and DrATP a Parameter 1st period of 2nd period of 2nd period of underperfusion underperfusion underperfusion1hypoxia Vmax 5NT (nmol/ min/ g) * 8565* Integral of DrATP (mmol/ l) a Values (means6s.e.) were obtained from optimized model solutions of individual experiments during the entire 30-min period of underperfusion (n56). V max, maximal reaction velocity. 5NT, 59-nucleotidase; Integral of DrATP, estimation of total net high energy phosphate breakdown during underperfusion, a negative value indicates a net breakdown of energy. * Significantly different from first period of underperfusion (P#0.05).

7 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) Fig. 4. Exploratory model fits to the PCr (d), ATP ( ), adenosine (h) and hypoxanthine1inosine (^) data using the four-region model of myocardial energetics and enzymatic kinetics with (left) and without (right) AMP deaminase activity. Data are indicated by the symbols and are identical for (A) and (B); optimized model fits indicated by the lines. Note the comparable fits to the high energy phosphate curves, but an over-estimation of purine efflux in the fits without AMP deaminase activity. sine, and the total amount of purines that appeared in the higher flow was the considerable degree of PCr recovery effluent during underperfusion, resulted in a predicted Vmax within the first 5 min of the initiation of underperfusion. for AMPD of 90 nmol/ min/ g and a Km of 30 mm for Exploration of this phenomenon with the mathematical AMP. The model predicted that IMP accumulated in the model indicated such steep recovery could not be predicted micromolar to low millimolar range during underperfusion. by a high, but constant, 5NT activity, as predicted by the Separate freeze-clamp experiments indicated, indeed, an open-adenylate hypothesis [3]. One could postulate that accumulation of 7 12 mmol/ g (1 2 mm) IMP after a 5NT activity varies strongly during this period and that 30-min period of 10% flow (unpublished observations). such rapid changes in the 5NT activity would cause such These values are in agreement with literature values. It has strong PCr recovery by a high degree of AMP hydrolysis, been shown, for example, that rat hearts accumulated 200 however, the purine efflux patterns were reasonably fit mm IMP after 45 min of ischemia, in the presence of without such rapid changes. Based upon the lactate efflux pyruvate and glucose, which increased to more than 600 data (Fig. 2), which suggest an early burst of glycolytic mm in the presence of glucose alone [15]. During the final activity shortly after the initiation of underperfusion, we optimization runs in this study, the curve fits were not explored the possibility of a short period of positive energy improved by allowing the parameters describing the balance occurring during this period. This was achieved by deamination of AMP to IMP to vary. In other words, the the implementation of a second DrATP function. It was data were best described by a constant AMPD activity, found, empirically, that the data were best fit when a small therefore, AMPD probably does not play a regulatory role positive burst of energy was allowed, measuring only under the conditions tested in these experiments. Further % of the total negative energy balance which more, that the data were best fit when purines were occurred after the onset of underperfusion (Fig. 5). Hard allowed to accumulate suggests a very low activity of the conclusions cannot be drawn from such empirical model- IMP-preferring isoform of 5NT (cn-ii) during underperfu- ing, it does suggest the possibility of a short period of sion or hypoxia. positive energy imbalance shortly after the onset of The present study was performed at a lesser degree of underperfusion. Possibly, the energy is glycolytic in naischemic stress (i.e. 10% of baseline flow) than the ture, because Janier et al. [16] showed a burst of lactate previous study (i.e. 5% flow during underperfusion) and release in glucose-perfused rabbit hearts within 10 min thus allows a comparison of the degree of hypoxic/ is- after the onset of 10% flow. Or it may be that mechanical chemic stress necessary to induce the downregulation of downregulation precedes the metabolic downregulation 5NT. An interesting and surprising result, however, of the and occurs within the first few minutes of underperfusion

8 340 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) Fig. 5. Exploratory model fits to the PCr (d), ATP ( ), adenosine (h) and hypoxanthine1inosine (^) data using the four-region model of myocardial energetics and enzymatic kinetics with (left) and without (right) extra glycolytic activity. Data are indicated by the symbols and are identical for (A) and (B); optimized model fits indicated by the lines. Note the inability to fit the high energy phosphate curves, with an over-estimation of purine efflux in the fits without the extra postulated glycolytic activity. (i.e. developed left ventricular pressure falls to lowest level mm. It is interesting to note that the hysteresis of these within 2 min), whereby the energy demands are actually relationships, caused by transport delays in the system and even less than the energy supply. This is supported by the by the phasic nature of the AMP concentration, disappears finding that the model predicts a very slight, yet positive during the hypoxic situation. This is probably due to the energy imbalance (0.3%) during the prolonged phase of fact that the AMP concentration is no longer phasic during underperfusion (i.e min). continuous hypoxia (see Fig. 2) Relationship between cytosolic AMP and purine 4. Discussion release 4.1. Downregulation of 59-nucleotidase The relation between the calculated cytosolic AMP concentration and the measured purine efflux for each The relationship between the cytosolic AMP concenseparate period of underperfusion is given in Fig. 6. The tration and adenosine release has been the topic of much data points are from the venous purine data (bottom), while research. A linear relationship between free AMP and the lines are derived from the model fits (top). The figure adenosine formation has generally been assumed and that indicates a lower purine efflux during the second period of the cytosolic concentration of AMP drives the reaction by normoxic underperfusion with similar levels of AMP mass action at the enzyme, 5NT. However, many conconcentrations, as compared to the first period of under- ditions have been observed where there has been a perfusion. For example, during the first period of under- dissociation observed between free AMP and adenosine perfusion, a capillary concentration of 40 mm is achieved release [17]. It has only recently been recognized that at a cytosolic AMP concentration of 2 mm, while during adenosine kinase inhibition amplifies the release of adenothe second period of underperfusion only 20 mm is sine during hypoxic stress, thereby explaining the enhanceachieved at a cytosolic AMP concentration of 2 mm. ment of adenosine release at only small or no increases in During the hypoxic period of underperfusion, the differ- cytosolic AMP concentrations [17,18]. The role that 5NT ences between the relationships become even more ex- may play during ischemic preconditioning has also been treme. During the hypoxic second period of underperfu- controversial. It has been proposed, for example, that 5NT sion, cytosolic AMP levels reach 4.5 mm, yet the purine becomes activated during ischemic preconditioning [19], efflux does not go above a capillary concentration of 20 yet others have shown purine release attenuation after brief

9 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) when cytosolic AMP levels are increased two-fold by hypoxia. The model-based analysis shows that the observed decrease in purine efflux can be ascribed to a lower 5NT activity. We conclude that 5NT does become downregulated during a 30-min period of 10% flow (V 5164 max vs. 129 nmol/min/g; this paper), albeit to a lesser degree than after a 45-min period of 5% flow (V 5140 vs. 67 max nmol/min/g; [5]). Since we have shown 5NT activity to be downregulated after a 20-min period of 5% flow [5], we assume here that both second periods of underperfusion described in this paper (i.e. under normoxic and under hypoxic conditions), began with similarly reduced activities of 5NT. That the second period of underperfusion with hypoxia is best fit with an even lower 5NT activity indicates modulation during the second underperfusion period. The systematic over-prediction of the purine efflux during the last 20 min of the second period of underperfusion with hypoxia is suggestive that 5NT becomes further downregulated during this phase. The peak is reasonably fit by an activity of 88 nmol/min/g, yet the purine efflux decreases during the last 10 min of the underperfusion period even while the cytosolic concentration of AMP is elevated during this period. A limitation of the current model is the inability to allow parameter values to vary during one period of underperfusion. These findings indicate the desirability to apply mathematical models in future work which have the ability to adapt 5NT levels during the underperfusion period In vivo regulation of AMP hydrolysis Fig. 6. Top panel: Model predictions of the relationship between cytosolic free AMP and capillary purine efflux for the three different periods Results presented here show a downregulation of AMP of underperfusion. (1) First period of underperfusion, (2C) second period hydrolysis during a period of underperfusion. The primary of underperfusion (normoxic), (2H) second period of underperfusion enzymes known to regulate the net hydrolysis of AMP are (hypoxic). Bottom three panels: Data points depicting the measured 59-nucleotidase, AMP deaminase and adenosine kinase. relations between the cytosolic AMP concentration and the total purine One should also not forget to take into account that release for the three different groups. Fredholm et al. in 1982 [22] showed that AMP itself is released from hearts by sympathetic nerve stimulation. exposures to global ischemia [20,21]. Possible confound- Regulation of AMP levels and adenosine efflux is, thus, ing factors in these studies, however, are mass action effect complex and finely tuned, as befitting an important regof AMP and the degree of energetic imbalance. ulator of myocardial energetics and survival. Indeed, the Previous work from this laboratory, which implemented greater tolerance of neonatal myocardium to ischemiaan integrative mathematical model to account for varying reperfusion injury has been attributed to lesser AMP degrees of energetic imbalance, has shown that 5NT, hydrolysis due to lower concentrations of 5NT [23]. In 1 indeed, becomes downregulated during prolonged under- vitro studies on highly purified 5NT indicate H and Pi to 21 perfusion (45 min, 5% flow) [5]. This can be considered to be inhibitory, while AMP, ADP and Mg are activators. be a protective mechanism whereby the cardiomyocyte A recent study by Bak and Ingwall [24] using hyperthyroid attempts to prevent depletion of the nucleotide/ nucleoside hearts to manipulate intracellular ph adds evidence that pool during prolonged oxygen deprivation. The question acidosis decreases the activity of 5NT and thus enhances remained, however, whether 5NT remains downregulated the resynthesis of ATP during reperfusion. These data in the face of additional hypoxic stress, or does it become obtained during global ischemia, however, are difficult to re-activated in order to provide adenosine-derived benefits. relate to less severe ischemic episodes, since our prepara- The data presented here show clearly lower purine efflux at tions have never shown a resynthesis of ATP during the similar cytosolic AMP levels during a normoxic second reperfusion period, even after a 45 min period of 5% flow period of underperfusion, which remains attenuated, even [5]. Evidently a different accumulatory pathway is active

10 342 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) during global ischemia than during low flow ischemia, to IMP is emerging as a complex and finely tuned system. 1 even though convincing data exist for both preparations Both H and Pi may be primary regulators of these paths. that AMP hydrolysis is downregulated during these Further work is needed to fully elucidate the underlying periods. Our results tend to support the hypothesis of Pi as mechanisms of regulation since the path that is chosen, one a regulator of 5NT, since phi was similar during both of vasodilation and nucleotide depletion or one of accumu- second periods of underperfusion, while Pi levels were lation has far-reaching ramifications for the bioenergetics higher during hypoxia, thus possibly resulting in the of the heart. lowered 5NT activity. Indeed, as shown by Itoh et al. [25], In summary, our results give strong evidence for the 5NT activity is more sensitive to changes in Pi con- persistent regulation of AMP hydrolysis during prolonged centrations at lower energy charges. ischemic conditions. Previous data show that 5NT be- In vitro data, for 5NT isolated from dog [26], rat and comes downregulated 20 min after the onset of severe human [12] heart obtain a Km of 1.5 mm under maximally ischemia. The current results point out clearly that such activated conditions. This value, however, is quite different stable downregulation is also achieved at a less severe from the model-based predicted Km for 5NT obtained in level of ischemia, but that also a nucleoside/ nucleotide this study. This model prediction of a high affinity of 5NT accumulatory pathway also is active for AMP (approximately 1mM) results from the fact that AMP hydrolysis to adenosine is quickly achieved at AMP 4.3. Clinical implications concentrations in the low micromolar range early in the underperfusion period. Since free cytosolic AMP levels The two general schools of thought regarding adenosine rarely exceed low micromolar concentrations, it would release by metabolically-perturbed myocardium have been: appear the in vitro indications that the Km of 5NT for AMP (1) adenosine release is good, therefore more will be is in the millimolar range are not physiologically realistic. better, and (2) adenosine release results in nucleotide Furthermore, exploratory modeling showed clearly that the depletion, therefore less is better. measured purine efflux rates could not be achieved when However, a more refined picture is now emerging. in vitro values ( 3mM) obtained from the literature were Severe underperfusion results in a stable downregulation of used for the Km of 5NT for AMP. AMP hydrolysis during severe ischemia, which persists In skeletal muscle, the IMP pathway is very active and through a short period of reperfusion [5]. The data remains dominant, even at high Pi concentrations. In the presented here show that the decreased AMP hydrolysis heart, however, AMPD activity is considerably lower, thus also persists during a period of even greater metabolic it has been often ignored. Data exist, however which show stress. Thus, a strategy is chosen whereby an initial, that the IMP pathway does play a regulatory role during beneficial period of high purine efflux is followed by a energetic perturbations of the heart. It has been found, for nucleotide-saving strategy whereby AMP hydrolysis to example, that energy-depleted human cardiomyocytes de- adenosine is decreased and nucleotides are allowed to phosphorylate 70% of the AMP via 5NT, and deaminate accumulate. Cardioprotective strategies need to be de- 30% via AMPD [8]. In rat heart, the regulation of AMPD veloped which will enhance and not antagonize these activity is complex, with allosteric modulation by multiple processes. One can envision, for example, pharmacological factors including ATP, GTP and P i [27]. Perfusion with interventions which aid in the accumulation of nucleotides 2-deoxyglucose, which causes a fall in ATP levels without (i.e. IMP) without adversely influencing the phosphorylaa concomitant rise in P i, induces predominantly inosine tion potential, while at the same time the importance of release, while hypoxia induces a release of a combination adenosine receptor activation should not be forgotten of adenosine and inosine [27]. Thus, the IMP pathway during the early phases of ischemia. dominates in the 2-deoxyglucose perfused heart, where ATP and Pi levels are low. Under such conditions, AMPD accounted for 97% of the AMP catabolites. In contrast, in Acknowledgements the anoxic heart, where AMPD is inhibited by high P i, the IMP pathway accounted for only 23% of the AMP flux. The authors thank Rodney Gronka for his invaluable The previously described model [5], which lacked a expertise in conducting the NMR experiments. This study purine-accumulating pathway, was able to quantitatively was supported by National Institutes of Health grants describe the purine efflux during 5% flow conditions, yet HL51152 and RR was unable to adequately describe purine efflux under 10% flow conditions. Since AMPD activity is regulated by P i [27 29], one may speculate that the IMP pathway is active References under 10% flow conditions, where Pi levels reach 5 mm, yet is inactivated by Pi during 5% flow conditions, where Pi [1] Vary TC, Angelakos ET, Schaeffer SW. Relationship between levels reached mm. The regulation of the hy- adenine nucleotide metabolism and irreversible ischemic damage in drolysis of AMP to adenosine, or the deamination of AMP isolated perfused rat heart. Circ Res 1979;45:

11 L.A. Gustafson et al. / Cardiovascular Research 44 (1999) [2] Matherne GP, Headrick JP, Berr S, Berne RM. Metabolic and tioning stimulates anaerobic glycolysis in the isolated rabbit heart. functional responses of immature and mature rabbit hearts to Am J Physiol 1994;26:H1353 H1360. hypoperfusion, ischemia and reperfusion. Am J Physiol [17] Decking UK, Schlieper G, Kroll K, Schrader J. Hypoxia-induced 1993;264:H2141 H2153. inhibition of adenosine kinase potentiates adenosine release. Circ [3] Kroll K, Kinzie DJ, Gustafson LA. Open system kinetics of Res 1997;81: myocardial phosphoenergetics during coronary underperfusion. Am [18] Wagner DR, Bontemps F, van den Berghe G. The AMP-adenosine J Physiol 1997;272:H2563 H2576. cycle is active during normoxia and impaired in ATP depletion in [4] Reimer KA, Murry CE, Yamasawa I, Hill ML, Jennings RB. Four isolated rabbit cardiomyocytes. Adv Exp Med Biol 1994;370:323 brief periods of myocardial ischemia cause no cumulative ATP loss 326. or necrosis. Am J Physiol 1986;251:H1306 H1315. [19] Kitakaze M, Hori M, Takashima S, Sato H, Inoue M, Kamada T. [5] Gustafson LA, Kroll K. Downregulation of 59-nucleotidase in rabbit Ischemic preconditioning increases adenosine release and 59-nuheart during coronary underperfusion. Am J Physiol cleotidase activity during myocardial ischemia and reperfusion in 1998;274:H529 H538. dogs. Implications for myocardial salvage. Circulation 1993;87:208 [6] Kroll K, Decking UKM, Dreikorn K, Schrader J. Rapid turnover of 215. the AMP-adenosine metabolic cycle in the guinea pig heart. Circ [20] Bradamante S, Piccinini F, Delu C, Janssen M, de Jong JW. NMR Res 1993;73: evaluation of changes in myocardial high energy metabolism [7] Newby AC. The pigeon heart 59-nucleotidase responsible for produced by repeated short periods of ischemia. Biochim Biophys ischaemia-induced adenosine formation. Biochem J 1988;253:123 Acta 1995;1243: [21] Harrison GJ, Willis RJ, Headrick JP. Extracellular adenosine levels [8] Smolenski RT, Suitters A, Yacoub MH. Adenine nucleotide catabo- and cellular energy metabolism in ischemically preconditioned rat lism and adenosine formation in isolated human cardiomyocytes. J heart. Cardiovasc Res 1998;40: Mol Cell Cardiol 1992;24: [22] Fredholm BB, Hedqvist P, Lindstrom K, Wennmalm M. Release of [9] Heineman FW, Eng J, Berkowitz BA, Balaban RS. NMR spectral nucleosides and nucleotides from the rabbit heart by sympathetic analysis of kinetic data using natural lineshapes. Magn Reson Med nerve stimulation. Acta Physiol Scand 1982;116: ;13: [23] Grosso MA, Banjeree A, St. Cyr JA et al. Cardiac 59-nucleotidase [10] Mallet RT, Kang YH, Mukohara N, Bunger R. Use of cytosolic activity increases with age and inversely relates to recovery from metabolite patterns to estimate free magnesium in normoxic myocar- ischemia. J Thorac Cardiovasc Surg 1992;103: dium. Biochim Biophys Acta 1992;1139: [24] Bak MI, Ingwall JS. Regulation of cardiac AMP-specific 59-nu- [11] Thakkar JK, Janero DR, Yarwood C, Hreniuk D. Isolation and cleotidase during ischemia mediates ATP resynthesis on reflow. Am characterization of AMP deaminase from mammalian (rabbit) J Physiol 1998;274:C992 C1001. myocardium. Biochem J 1993;290: [25] Itoh R, Oka J, Ozasa H. Regulation of rat heart cytosol 59- [12] Skladanowski AC, Hoffmann C, Krass J, Jastorff B, Makarewicz W. nucleotidase by adenylate charge. Biochem J 1986;235: Structure activity relationship of cytoplasmic 59-nucleotidase sub- [26] Darvish A, Metting PJ. Purification and regulation of an AMPstrate sites. Biochem J 1996;314: specific cytosolic 59-nucleotidase from dog heart. Am J Physiol [13] Darvish A, Pomerantz RW, Zografides PG, Metting PJ. Contribution 1993;264:H1528 H1534. of cytosolic and membrane-bound 59-nucleotidases to cardiac adeno- [27] Chen W, Gueron M. AMP degradation in the perfused rat heart sine production. Am J Physiol 1996;271:H2162 H2167. during 2-deoxy-D-glucose perfusion and anoxia. J Mol Cell Cardiol [14] Hohl CM, Wimsatt DK, Brierley GP, Altschuld RA. IMP production 1996;28: by ATP-depleted adult rat heart cells. Circ Res 1989;65: [28] Hu B, Altschuld RA, Hohl CM. Adenosine stimulation of AMP [15] De Groot MJM, Coumans WA, van der Vusse GJ. The nucleotide deaminase activity in adult rat cardiac myocytes. Am J Physiol metabolism in lactate perfused hearts under ischaemic and re- 1993;264:C48 C53. perfused conditions. Mol Cell Biochem 1992;118:1 14. [29] Spychala J, Marszalek J. Regulatory properties of AMP deaminase [16] Janier MF, Vanoverschelde JL, Bergmann SR. Ischemic precondi- from rat tissues. Int J Biochem 1991;23: