Quantitative Relation Between Interstitial Adenosine Concentration and Coronary Blood Flow
The effect of exogenous and endogenous adenosine in controlling coronary flow was determined using an axially distributed mathematical model of the myocardium to estimate interstitial adenosine concentration from coronary arterial and venous adenosine values. The left main coronary artery was perfused at constant pressure in closed-chest, anesthetized dogs, and exogenous adenosine was infused intracoronary to increase coronary flow. Basal interstitial adenosine was 92 nmol/L, just at the threshold for increasing coronary flow. An increase in interstitial adenosine concentration of only 62% was sufficient to increase coronary flow from 5% to 50% of maximal flow. The possible contribution of an endothelial dilator secondary to activation of adenosine receptors on endothelial cells was tested by comparing the response to exogenous intracoronary adenosine infusion with increases in endogenous adenosine produced by inhibition of adenosine kinase and adenosine deaminase. If adenosine increases coronary flow by an endothelial mechanism, then the interstitial ED50 of exogenous adenosine would be lower than that for endogenous adenosine due to the postulated additional endothelial dilator. The interstitial ED50 for exogenous adenosine was 156 nmol/L, not different from the endogenous ED50 of 150 nmol/L. In conclusion, basal interstitial adenosine concentration is at the threshold of a remarkably steep dose-response curve for increasing coronary blood flow. No evidence was found for an endothelium-mediated vasodilator mechanism secondary to adenosine receptor activation of endothelial cells in vivo. The steep adenosine dose-response curve indicates that measurements of adenosine concentration should be interpreted with caution, because small changes in adenosine concentration cause large changes in coronary flow.
The coronary circulation manifests a remarkable ability to adjust coronary blood flow to myocardial oxygen demand.1 The adenosine hypothesis proposed independently by Berne2 and Gerlach et al3 states that the intrinsic mechanism that matches coronary blood flow to myocardial metabolism is mediated by adenosine. The hypothesis is that a falling oxygen tension inside myocardial cells leads to the release of adenosine that crosses the interstitial space and activates adenosine receptors on coronary vascular smooth muscle cells to initiate vasodilation. The augmented coronary blood flow increases oxygen delivery to the tissue and restores myocardial oxygen tension to the normal operating range in a feedback manner. In effect, the concentration of adenosine in the interstitial space of the myocardium controls coronary blood flow.
Evaluating adenosine's role in controlling coronary blood flow requires an understanding of the quantitative relation between adenosine concentration in the interstitium and coronary flow. Early studies used exogenous infusions of adenosine to increase interstitial adenosine concentration4 5 and to obtain an appreciation of the relative effects of ischemic and adenosine-mediated vasodilation.6 However, Nees and coworkers7 have shown that vascular endothelial cells in vitro avidly take up adenosine, reducing the infused adenosine that reaches the interstitial space. In vivo studies have also shown that adenosine is highly extracted by endothelial cells in the coronary circulation.8 Also, Headrick and Berne9 have suggested that much of the vasodilator action of infused adenosine might be mediated by a vasodilator released from the vascular endothelium that is neither EDRF nor a prostaglandin. An alternative method of selectively increasing interstitial adenosine while minimizing endothelial uptake is to increase endogenous adenosine release from the heart itself by inhibiting adenosine kinase and adenosine deaminase with ITC and EHNA.10 11
A method has been developed for estimating interstitial adenosine concentration from venous adenosine concentration that accounts for the effects of blood flow, arterial adenosine, flow heterogeneity, and metabolism of adenosine by capillary endothelial cells.8 12 This method is based on mathematical model analysis of tracer kinetics and was validated for nontracer adenosine in vivo. The objective of the current study was to apply this method to describe the vasodilator actions of interstitial adenosine from exogenous and endogenous sources.
Briefly, the present results from experiments in anesthetized, closed-chest dogs indicate that the interstitial adenosine ED50 for coronary blood flow is in the range of 100 to 300 nmol/L, with a very steep dose-response curve. The dose-response curves for interstitial adenosine were very similar for both exogenous and endogenous adenosine, indicating there is little role for an endothelial vasodilator action in vivo.
Materials and Methods
Mongrel dogs weighing 24 to 34 kg were sedated with morphine sulfate (2.5 mg/kg SC) and anesthetized with α-chloralose (100 mg/kg IV). Anesthesia was maintained with 500-mg supplements of α-chloralose as needed. Animals were ventilated with oxygen-enriched room air by a positive-pressure respirator (model 607, Harvard Apparatus) so that arterial oxygen tension was maintained above 90 mm Hg. Body temperature was sensed with an esophageal thermistor and controlled at 37°C with a YSI model 73A controller and a heating pad. All experiments were carried out on closed-chest dogs. Aortic blood pressure was monitored with a strain gauge manometer (Statham P23 ID, Gould, Inc) connected to a catheter inserted via the right femoral artery. Sodium heparin (750 U/kg IV) was given as an anticoagulant. Ibuprofen (12.5 mg/kg) was given to prevent complement activation induced by blood-tubing interactions. Arterial and coronary venous blood gas samples were analyzed for Po2, Pco2, pH, and base excess (model 1306, Instrumentation Laboratories). Metabolic acidosis associated with chloralose anesthesia was countered with infusions of NaHCO3 as required.
Coronary Sinus Cannulation
To sample coronary venous blood for adenosine measurement, a catheter was advanced under fluoroscopic guidance into the coronary sinus via the right jugular vein. A metal ring surrounded the tip of the cannula to prevent venous collapse during sampling. The position of the catheter tip was verified postmortem and was always >30 mm inside the coronary sinus ostium.13
Left Main Coronary Artery Cannulation
The left main coronary artery was cannulated with a balloon-tipped stainless steel cannula14 inserted via the right carotid artery (Fig 1⇓). Coronary artery pressure was measured at the tip of the cannula via an inner stainless steel tube connected to a strain gauge manometer (Statham P23db). A servo-controlled roller pump maintained mean coronary artery pressure at 100 mm Hg.15 Coronary blood flow was measured with a cannulating flow transducer and an electromagnetic flowmeter (model SWF-5RD, Zepeda Instruments) placed in the extracorporeal circuit. The source of coronary perfusate blood was the left femoral artery. The seal of the balloon-tip cannula in the coronary ostium was considered satisfactory if injection of crystal violet at the end of the experiment, while coronary pressure was maintained 50 mm Hg above aortic pressure, stained perfused vessels and myocardium but not the ascending aorta. The weight of the stained tissue was used to compute the flow per gram of perfused myocardium.
Arterial and coronary venous lactate samples were drawn into vials containing NaF to prevent glycolysis and were placed on ice. Lactate concentration was measured by an automated method by injecting blood into a YSI model 23A lactate analyzer. The machine was calibrated with external lactate standards before and after analysis. Myocardial percent lactate extraction was calculated as (arterial−venous)/arterial values.
Plasma Adenosine Measurements
Plasma adenosine concentration was measured with a modified version of the Herrmann and Feigl method.16 Briefly, blood samples were collected in chilled syringes that mixed blood with ice-cold saline enzymatic stop solution containing dipyridamole, ITC, and EHNA. Dipyridamole (32 μmol/L) acts as an inhibitor of cellular adenosine uptake. ITC (1 μmol/L) is an inhibitor of adenosine kinase, preventing the incorporation of adenosine into AMP. EHNA (10 μmol/L) is an inhibitor of adenosine deaminase, preventing the degradation of adenosine to inosine. Theophylline (20 μmol/L) was included in the stop solution as an internal recovery standard. Samples were centrifuged at 15 000 rpm for 2 minutes, and 5 mL of the supernatant plasma was added to 1.8 mL of 4N perchloric acid to precipitate plasma proteins. The sample was recentrifuged and 5 mL of the acid supernatant added to 4.35 mL of a neutralizing solution containing 0.4 mmol/L KH3PO4 and 0.8 mmol/L KOH. The resulting pH was 7.0. Adenosine samples were then purified by applying the neutralized extracts to C-18 Sep-Pak cartridges and eluting with 40% methanol to separate EHNA from adenosine and theophylline. Samples were concentrated by evaporation and split into separate aliquots. One aliquot was treated with adenosine deaminase (50 U/mL; Boehringer Mannheim) and served as a blank. Samples were then separated on a Hewlett Packard 1090M HPLC, using an ion-pairing solution (tetrabutylammonium hydrogen sulfate and KH3PO4) and acetonitrile gradient. Adenosine content was determined by subtracting the blank chromatogram from the untreated sample chromatogram. Adenosine mass was determined by comparison to known adenosine standards, and concentration was quantitated by accounting for dilution steps in sample handling and normalized for recovery with the theophylline standard.
Adenosine, EHNA (Sigma Chemical), and ITC (RBI) were dissolved in isotonic saline and administered intracoronary. 8-PT (75 mg; Sigma) was prepared in a vehicle of 1 mL ethanol and adjusted with 0.1N NaOH until completely dissolved (pH 12 to 13). Final volume was adjusted to 36 mL by the addition of isotonic saline and administered intravenously over 10 minutes. The stop solution was made in isotonic saline and included 1 μmol/L ITC, 10 μmol/L EHNA, and 32 μmol/L dipyridamole (Sigma).
Exogenous Adenosine (n=10)
Before the adenosine infusions were begun, duplicate arterial and venous samples were drawn to determine the basal interstitial concentration as described below. To characterize changes in flow and interstitial adenosine during increases in arterial adenosine concentration, exogenous adenosine was infused into the left main coronary artery in graded doses over a range of infusion rates from 2.67 ng/min to 267 ng/min, a range consistent with that used by others.5 17 Once flow had reached a steady state at each level, a venous blood sample was drawn from the coronary sinus catheter for HPLC measurement of adenosine. For these exogenous adenosine experiments, the arterial plasma adenosine concentration was calculated from the adenosine infusion rate and the coronary blood flow rate using Equation 1 below.
In a subset (n=5) of the experiments with exogenous adenosine infusion described above, a second adenosine dose-response curve was obtained by increasing endogenous release of adenosine. Interstitial adenosine was increased by graded intracoronary infusions of the adenosine kinase inhibitor ITC (2.5 to 25 μg/min) and the adenosine deaminase inhibitor EHNA (5 to 95 μg/min). Once flow had reached a steady state, arterial and venous samples were drawn simultaneously for the HPLC measurement of adenosine. One exogenous (adenosine infusion) and one endogenous (ITC and EHNA infusion) dose response were performed in each animal. In each case, the exogenous dose response was done before the endogenous dose response, because it was found in pilot experiments that the effects of a maximal dose of ITC and EHNA persisted too long to permit a second dose response with exogenous adenosine.
Receptor Antagonism With 8-PT
To confirm that the coronary vasodilation in the present experiments was mediated via adenosine receptors, the response was challenged with the adenosine receptor antagonist 8-PT.18 The effect of 8-PT was tested by infusing adenosine before and after administration of 8-PT (3 mg/kg IV) in five dogs in which the left circumflex coronary artery was perfused at a constant pressure of 100 mm Hg. After determining that 3 mg/kg IV of 8-PT was an effective blocking dose for exogenous adenosine, the same dose was used for experiments with endogenous adenosine.
The effect of adenosine receptor blockade on endogenous adenosine was tested by comparing the response to left main coronary artery infusions of ITC and EHNA in four dogs before and after 8-PT (3 mg/kg) as described above. Maximal effects were not obtained in the first part of this experiment, because it was found in pilot experiments that the effects of a maximal dose of ITC and EHNA persisted too long to permit a second endogenous dose-response curve after 8-PT. It was not possible, when using the enzyme inhibitors ITC and EHNA, to increase the endogenous adenosine concentration to pharmacological levels to overcome the 8-PT blockade.
Calculation of Arterial Plasma Adenosine Concentration
During intracoronary arterial adenosine infusions, arterial plasma adenosine concentration was calculated from the adenosine infusion rate and blood flow rate:where [ADO] is adenosine concentration and Hct is hematocrit.
Adenosine concentration was determined with HPLC in a subset of arterial samples, and the calculated values were compared with the measured values. The two groups differed by less than 10%.
Hill Equation Fitting to Dose Responses
The dose-response data in the figures were fit with curves using the Hill equation:where F is coronary blood flow, Fmin is the measured resting coronary blood flow, Fmax is the maximal measured coronary blood flow, [ADO] is the calculated arterial adenosine concentration or the model-estimated interstitial adenosine concentration, and H is the Hill exponent. FigP (Biosoft), a graphics program with statistical capabilities, was used to plot the data as well as to generate the curves. For each curve, the observed values for Fmin and Fmax were fixed and the experimental values of F and [ADO] were entered. Given these fixed parameters, the best fit to the data set was then obtained by allowing the least-squares fitting routine to select the ED50 and H that provided the optimal solution to the Hill equation for that data set. The average curves were obtained by averaging the individual equation parameters (Fmin, Fmax, ED50, and H) from all data sets in an experimental group.
Statistical comparisons were performed with Student's two-tailed paired t test on Microsoft Excel. Significance was accepted at P<.05.
Steady state interstitial adenosine concentrations were estimated by using a four-region (capillary, endothelial cell, interstitial space, and parenchymal cell), axially distributed mathematical model19 to analyze measurements of coronary artery and venous adenosine concentrations and coronary blood flow (Fig 2⇓). The model describes the effects of flow, transport, and exchange between tissue regions and cellular production and consumption on the relations between arterial, venous, and interstitial adenosine concentrations. Parameters describing the consumption G and permeability surface area product PS for adenosine are first-order rate constants (units of mL/min per gram). Adenosine production S is constant (zero order) in endothelial and parenchymal cells. The effects of regional flow heterogeneity are described in the model using multiple parallel capillary flow pathways having a common arterial inlet and a common venous outlet.20 Model parameter values were obtained previously by fitting multiple indicator dilution curves using adenosine in the closed-chest dog heart under control conditions and during elevated coronary blood flow.8 The partial differential equations used in the model are given in the “Appendix.”
Constraints on Parameter Values
Plasma flow in the model is constrained to the measured blood flow times one minus the hematocrit. For plasma flows up to 1.5 mL/min per gram, flow heterogeneity was modeled using an RD (SD/mean) of 0.55 to describe the distribution of flows in the parallel flow pathways in the model. This flow distribution was modified slightly to better account for the effects of the highest flow pathways.8 For plasma flows in the range of 1.5 to 2.5 mL/min per gram, the RD of the flow distribution was decreased to 0.45, and for flows above 2.5 mL/min per gram (one case), the RD was further decreased to 0.35 to account for decreased flow heterogeneity due to increased coronary blood flow.21
During infusion of exogenous adenosine, plasma adenosine concentrations were constrained to the values calculated with Equation 1 given above. In the basal state and during increases in endogenous adenosine, arterial plasma adenosine concentrations were constrained to the values measured by HPLC.
Endothelial cell parameters
Parameters describing paracellular transport between capillary and the interstitial space (PSg), transport into capillary endothelial cells (PSecl), and consumption in endothelial cells (Gec) were assigned values based on reference values (PSg0=0.61 mL/min per gram, PSecl0=2.9 mL/min per gram, and Gec0=13 mL/min per gram) determined in tracer studies under basal conditions.8 To account for the increase in capillary exchange due to increased coronary blood flow,21 PSg, PSecl, and Gec were described as linear functions of flow.Thus, for every estimate of interstitial adenosine concentration in the present study, the values of PSg, PSecl, and Gec were specified by the plasma flow and the reference values PSg0, PSecl0, and Gec0. It was assumed that PSecl=PSeca. Volumes of tissue regions were constant: Vp=0.035 (1−hematocrit) mL/g, Vec=0.03 mL/g, and Visf=0.12 mL/g.
Parenchymal cell parameters
Uptake and metabolism of adenosine by parenchymal cells (mainly cardiomyocytes) were described by PSpc (13 mL/min per gram), Gpc (5 mL/min per gram), and Vpc (0.6 mL/g). Parenchymal cell parameters were independent of flow.
The first step in analyzing each experiment was estimating basal interstitial adenosine concentration by adjusting cellular adenosine production to fit the venous adenosine concentration. All other parameters were set at their constrained values (measured, calculated, or constant). For all fits, adenosine production in parenchymal and endothelial cells was held at a constant ratio of 20:1, reflecting the relative volume fractions of these two cell types in myocardial tissue.22
To estimate interstitial adenosine concentration (see above) during exogenous adenosine infusions, the arterial plasma adenosine concentrations and plasma flow were used in the model while constraining PSg, PSecl, and Gec calculated using equations 3, 4, and 5. The RD describing flow heterogeneity was set according to the flow. Cellular adenosine production was held constant at basal values. The method was validated in a previous study, in which the venous adenosine values predicted by the model fell within 10% on average of the measured venous adenosine values.8 No parameters were adjusted to improve the fit.
To estimate interstitial adenosine concentrations during inhibition of adenosine kinase and adenosine deaminase, arterial adenosine concentrations determined by HPLC and plasma flow were used in the model while constraining PSg and PSecl calculated using Equations 3 and 4. The RD describing flow heterogeneity was set according to the flow. Cellular adenosine consumption (Gpc and Gec) was decreased until the venous adenosine concentration predicted by the model was equal to that measured by HPLC. Gec and Gpc were held in the constant ratio of 2.5 that was observed under basal conditions.
The model assumptions are discussed in detail in the previous paper.8 Briefly, it is assumed that there is piston flow in the capillary, that there is radial diffusional equilibrium within each model region, and that the density of endothelial cell transporters and enzyme capacity is uniform along the capillary. It is also assumed that red blood cells play a negligible role in capillary adenosine kinetics in the dog. This assumption is based on previous studies that demonstrate a low red blood cell uptake in dogs,23 as well as an evaluation of tracer uptake of [3H]adenosine by red blood cells.8 It is also assumed that the enzyme inhibitors of adenosine kinase and adenosine deaminase are selective and equally effective in blocking parenchymal and endothelial cell enzymes, that adenosine production remains at basal levels independent of adenosine infusion or enzyme blockade, and that endothelial cells and cardiomyocytes contribute to myocardial adenosine production in proportion to their volume fractions in the heart.
Basal Adenosine Concentration
Basal hemodynamic and adenosine values are given in Table 1⇓. Using the modeling method described above, the estimated basal interstitial adenosine concentration was 92 nmol/L, with a CV of 30% (n=10).
Arterial plasma adenosine coronary blood flow dose-response curves were fit from each individual experiment with the Hill equation to obtain an ED50 value and H. The average ED50 for plasma adenosine was 1360 nmol/L, with a CV of 53% (Fig 3⇓). The average H for 10 curves was 5.0±2.1 SD. The estimated ED50 for interstitial adenosine from model calculations was 190 nmol/L, with a CV of 29% (Fig 3⇓). The ED50 for the interstitial curve is one seventh of the ED50 for arterial plasma.
The interstitial adenosine curve is also very steep (H=7.5±2.9; mean±SD, n=10). Using the Hill equation, ED05 and ED50 values were obtained for plasma and interstitial adenosine. These values illustrate the steepness of the adenosine dose-response curve, as a 92% increase in plasma adenosine concentration and a 62% increase in interstitial adenosine concentration were sufficient to increase flow from a threshold 5% of the maximal flow to 50% of maximal flow.
Comparison of Exogenous and Endogenous Adenosine
Endogenous adenosine concentration was increased by inhibiting adenosine kinase with ITC and adenosine deaminase with EHNA. Fig 4⇓ depicts each endogenous interstitial adenosine dose-response curve compared with its matching exogenous curve from the same animal (Fig 4A through 4E⇓) and summary curves using the average Hill equation parameters for all five dogs (Fig 4F⇓). The ED50 values were similar in both conditions (150±16 versus 156±17 nmol/L, mean±SE, P=NS), and no difference was present in H (7.61±1.63 versus 9.52±0.897, mean±SE, P=NS). These data fail to demonstrate a greater sensitivity (lower ED50) for exogenous versus endogenous adenosine, as would be expected if there were an important second dilator released from endothelial cells following activation of adenosine receptors on endothelial cells.
The HPLC-measured coronary venous adenosine concentrations for the same experiments shown in Fig 4⇑ are given in Fig 5⇓. Venous adenosine concentrations were similar during endogenous and exogenous dose-response measurements at equal levels of flow. These data also argue against an increased sensitivity for exogenous adenosine. However, the relation between venous and interstitial adenosine was different for endogenous and exogenous adenosine because enzyme inhibition with EHNA and ITC alters the endothelial cell metabolism of adenosine. The model estimates of interstitial adenosine shown in Fig 4⇑ account for the metabolic effects of EHNA and ITC.
Adenosine receptor blockade with 8-PT (3 mg/kg IV) shifted the exogenous adenosine dose-response curve to the right 12-fold, as illustrated in Fig 6⇓. The same dose of 8-PT abolished flow responses due to endogenous adenosine produced by enzyme blockade with infusions of ITC and EHNA, as shown in Fig 7⇓. The data in Fig 7⇓ indicate that 8-PT blocks coronary vasodilation due to endogenous adenosine at least as effectively as that due to exogenous adenosine. As described in “Materials and Methods,” it was not possible to obtain complete endogenous dose-response curves using enzyme inhibition with ITC and EHNA in the presence of 8-PT. These findings indicate that endogenous adenosine produces vasodilation by increasing interstitial adenosine concentration to activate adenosine receptors.
The present study is the first examination of the coronary blood flow dose response to interstitial adenosine in vivo. The major findings are as follows: The average estimated basal interstitial adenosine concentration is 92 nmol/L, which lies just below the threshold for coronary vasodilation. During exogenous adenosine infusion, the average arterial plasma adenosine ED50 is 1360 nmol/L, which is seven times greater than the corresponding interstitial ED50 of 190 nmol/L, because endothelial cells take up adenosine and form a barrier between the plasma and the interstitial space. Both the arterial plasma and interstitial dose-response curves are remarkably steep, with only a 62% increase in interstitial adenosine concentration causing coronary blood flow to increase from 5% to 50% of the maximal coronary blood flow. In a paired comparison, exogenous (intra-arterial infusion) and endogenous (inhibition of adenosine kinase and adenosine deaminase) interstitial adenosine dose-response curves were nearly identical. This finding indicates that activation of adenosine receptors on endothelial cells is not involved in increases in coronary blood flow in vivo. If an endothelial dilator were important, the exogenous dose-response curve would lie to the left of the endogenous curve. In addition, the adenosine receptor antagonist 8-PT was highly effective in blocking the effects of both exogenous and endogenous adenosine on coronary blood flow.
Estimation of Interstitial Adenosine Concentration
Because it is not feasible to directly measure adenosine concentrations in interstitial fluid, a number of techniques have been used to estimate interstitial adenosine concentration. A comparison of methods is provided in Table 2⇓.
Interstitial adenosine concentrations have been estimated by measuring adenosine concentration in buffer solutions placed in contact with the epicardial surface using a plastic well or porous disks.24 25 26 27 Some investigators have abraded off the visceral pericardium,24 28 while others have left the visceral pericardium intact.25 26 However, Imai and coworkers29 have shown that in rats treated with 6-OH dopamine (a chemical sympatholytic), the adenosine values obtained by sampling myocardial transudate are reduced compared with those obtained from untreated animals. This finding was attributed to cardiac nerves that run close to the epicardial surface that may release ATP or adenosine.29 The temporal resolution of epicardial methods is limited, as it takes about 4 minutes for the adenosine concentration in an epicardial well to come to steady state26 and 2 to 3 minutes for the epicardial disk method.30
Estimates of interstitial adenosine concentration have also been made from fluid pumped through microdialysis tubing that has been threaded through the cardiac muscle.31 The values in Table 2⇑ have been corrected for the 32.3% recovery reported with the microdialysis method.31 The microdialysis method has the disadvantage that the myocardium must be injured and the constant beating of the heart against the dialysis tubing may lead to cell leakage and very high adenosine levels.
The present distributed model method has the advantage of a rapid response time, being applicable in closed-chest animals, and it gives global values from a mixed venous sample. The values from other laboratories given in Table 2⇑ (distributed modeling open chest) were calculated with the present model using published values for coronary flow, arterial and venous adenosine concentration, and an assumed hematocrit of 45%.8
Basal Interstitial Adenosine
The estimated basal cardiac interstitial adenosine concentration in closed-chest anesthetized dogs was found to be 92 nmol/L. The observation that subthreshold doses of adenosine infusion produced measurable increases in interstitial adenosine with no change in blood flow is consistent with the results of Kroll and Feigl32 and Gewirtz et al33 that adenosine contributes little to flow during basal conditions.
Plasma Adenosine and Coronary Flow
The arterial plasma adenosine concentration was determined from the adenosine infusion rate and coronary plasma flow (Equation 1) without use of the model. The arterial plasma ED50 for adenosine infusion obtained in the present study (1360 nmol/L; Fig 3⇑) is similar to the value of 1010 nmol/L obtained by Olsson et al4 in open-chest anesthetized dogs; however, a lower value of 570 nmol/L was found in unanesthetized dogs. Thus, the present ED50 was probably increased by anesthesia. Rubio et al5 reported a high vasoactive range in their Table 3. Using their data and the Hill equation, an ED50 of ≈3000 nmol/L and an H of 6.7 were obtained. The biologic variability of the ED50 among different animals in the present study is substantial, with an SD of 721 nmol/L, which gives a CV (SD/mean) of 53% (Fig 3⇑). The steepness of the adenosine dose-response curve has been presented graphically,34 and the current study provides a quantitative description of that steepness using the Hill equation. Only a 92% increase in plasma adenosine concentration is sufficient to increase coronary blood flow from 5% to 50% of maximal flow.
Interstitial Adenosine and Coronary Flow
During intracoronary adenosine infusion, the interstitial adenosine concentration was estimated by using the distributed model. The model estimate of the ED50 for interstitial adenosine concentration was 190 nmol/L (Table 1⇑). The interstitial ED50 lies to the left of the arterial plasma ED50, because endothelial cells take up adenosine from the plasma before it can reach the interstitium via capillary intercellular clefts. The interstitial dose-response curve is remarkably steep in that only a 62% increase in adenosine concentration is needed to increase coronary blood flow from 5% to 50% of the maximal flow.
The very steep dose-response curves observed for both plasma and interstitial values in the present study indicate that caution should be used in interpreting adenosine concentrations when testing the adenosine hypothesis. Since a 62% increase in adenosine concentration will more than double coronary blood flow (Fig 3⇑), very precise measurements of adenosine will be required. With only a modest amount of variability in adenosine measurements, a significant difference will be missed, even though coronary blood flow may have doubled.
An endogenous interstitial dose-response curve for buffer-perfused in vitro guinea pig hearts has been reported by Kroll et al.11 Their ED50 value of 77 nmol/L is lower than the present value of 190 nmol/L. The steepness of the in vitro dose-response curve was less than the present in vivo result, since a 285% versus 62% increase in interstitial adenosine concentration was required to increase flow from ED05 to ED50 for the in vitro and in vivo studies, respectively.
Endothelial Contribution to Exogenous Adenosine Vasodilation
In addition to acting as a dilator of vascular smooth muscle, it has been suggested that infused adenosine acts by releasing an endothelium-dependent dilator. Headrick and Berne9 have shown that denudation of aortic rings reduced adenosine-mediated vasodilation, although enzymatic inhibition with NG-nitro-l-arginine methyl ester (nitric oxide synthase) and indomethacin (cyclooxygenase) had no effect. They concluded that adenosine promoted endothelium-dependent vasodilation independent of nitric oxide or prostacyclin. Furthermore, Balcells et al35 found that xanthine amine congener bound to 0.07-μm beads completely blocked vasodilation to infused adenosine. They concluded that the actions of infused adenosine were purely endothelium mediated, since xanthine amine congener was confined to the vascular space when bound to the beads. In contrast, Abebe et al36 found that only artificial adenosine analogues displayed endothelium-dependent vasodilation, while adenosine itself did not. In isolated coronary arteries from dogs, White and Angus37 found that adenosine relaxations were not endothelium dependent in large coronary arteries.
To test the hypothesis that an endothelial agent mediates part of the vasodilator response to exogenous adenosine, the ED50 values obtained during exogenous adenosine infusions and endogenous adenosine (ITC and EHNA) were compared. It is assumed that endothelial adenosine receptors are uniformly distributed on the luminal and abluminal surfaces of endothelial cells. Since luminal concentrations of adenosine are much higher during adenosine infusions than during production of endogenous adenosine, it is expected that endothelial cell receptors would be exposed to higher concentrations than smooth muscle cells during adenosine infusion. If so, the dose-response curve of interstitial adenosine and coronary flow during exogenous adenosine infusion would lie to the left of the curve for endogenous interstitial adenosine. To the contrary, similar dose responses would indicate that interstitial adenosine is sufficient to account for the flow response during both exogenous and endogenous elevations of adenosine concentration. As shown in Fig 4⇑, there was no significant difference between the ED50 values obtained from exogenous and endogenous adenosine. The comparison of coronary venous adenosine values during exogenous and endogenous adenosine shown in Fig 5⇑ also does not support an endothelial dilator mechanism, because the average endogenous curve lies to the left (wrong way) (P=NS) of the exogenous curve. The relation between venous and interstitial adenosine was different for endogenous and exogenous adenosine, because enzyme inhibition with EHNA and ITC interferes with endothelial adenosine metabolism. The distributed-model results in Fig 4⇑ account for this metabolic effect. Taken together, these data indicate that infused adenosine acts largely by increasing interstitial adenosine, although a small endothelial contribution might have been missed because there was a limited number of experiments. The present result should not be interpreted to mean that endothelial cells do not have adenosine receptors or that endothelial cells do not release a vasodilator when their adenosine receptors are activated. Rather, during the in vivo conditions of the present experiments, activation of endothelial cell adenosine receptors to release a second vasodilator appears unimportant as a mechanism of adenosine coronary vasodilation. Increased shear on endothelial cells secondary to augmented flow causes the release of EDRF in the coronary circulation.38 Because coronary flow increased equally during exogenous and endogenous adenosine, the present results do not pertain to whether or not increased endothelial shear secondary to adenosine infusion results in EDRF coronary vasodilation.
The quantitative relation for the increase in coronary flow due to exogenously infused adenosine and endogenous adenosine has been determined in closed-chest anesthetized dogs. A previously validated mathematical model was used to estimate interstitial adenosine concentration from experimental values of coronary blood flow, hematocrit, and arterial and coronary venous plasma adenosine concentrations. The average basal interstitial adenosine concentration of 92 nmol/L was just at the foot of a remarkably steep dose-response curve for coronary blood flow. This means that experimental adenosine concentration measurements should be interpreted with care when evaluating the adenosine hypothesis of coronary blood flow control, because small changes in adenosine concentration cause large changes in coronary blood flow. A paired comparison of the effect of exogenous and endogenous adenosine on coronary blood flow did not reveal an endothelial vasodilator mechanism secondary to adenosine receptor activation of endothelial cells in vivo.
Selected Abbreviations and Acronyms
|CV||=||coefficient of variation|
|EDRF||=||endothelium-derived relaxing factor|
|G||=||metabolic consumption of adenosine|
|PS||=||permeability surface area product|
|S||=||constant production rate of adenosine|
The model consists of a set of partial differential equations, based on the conservation of mass, that are solved using numerical techniques through time and along the length of the capillary. The equations19 and the methods of solution39 have been presented in detail previously and are summarized below. The model of a single capillary consists of equations for the capillary plasma p,interstitial region isf,and parenchymal cell pc,where subscripts refer to the model regions, x is the distance along the capillary of length L, C is adenosine concentration, fi is the plasma flow in an individual pathway relative to the overall mean flow through all the pathways, V is the region volume, S is constant production rate, PS and G are first order rate constants for the permeability surface area products for transport between regions and irreversible metabolic consumption, respectively, and PSg describes the extracellular exchange between capillary and interstitial regions (interendothelial clefts). The 10 flow pathways are fed by a common arterial inflow and drained by a common venous outflow. Each of the 10 parallel pathways uses equations that are identical to those above, except for their flow, fi. The individual pathway flows and weights are assigned according to a probability density distribution to represent the effects of regional flow heterogeneity. The overall mean flow through all the pathways is the weighted sum of the flows in individual pathways and is set to equal the plasma flow measured during the experiment. The overall venous adenosine concentration, Cout(t), is the weighted sum of the outflow concentration emerging from all the individual pathways,where Cin(t) is the arterial adenosine concentration, hcapi(t) is the impulse response of the capillary exchange unit in the ith pathway described by the equations above, wi and fi are the relative weight and flow of the ith pathway, Δfi is the width of the ith flow interval in the flow distribution (ie, wiΔfi is the fraction of the organ receiving relative flow fi), N is the number of individual pathways, and * denotes convolution. In the present study, steady state solutions were obtained, although equations are also valid for the more general case of transients.
The partial differential equations are solved numerically by dividing the capillary into a number of longitudinal segments.39 Every time step of the model solution is divided into two parts, in the first part of which the contents of every capillary segment are advanced into the next downstream segment, to represent convection. In the second part of the time step, the radial exchanges and consumption are determined analytically, using a Taylor's series approximation.
This study was supported by NIH grants HL-49822, HL-07403, and RR-01243 and American Heart Association Fellowship 95-WA-120. We thank Stephanie Belanger for expert assistance in all phases of this research.
- Received December 19, 1995.
- Accepted June 11, 1996.
Feigl EO. Coronary physiology. Physiol Rev. 1983;63:1-205.
Berne RM. Cardiac nucleotides in hypoxia: possible role in regulation of coronary blood flow. Am J Physiol. 1963;204:317-322.
Olsson RA, Khouri EM, Bedynek JL Jr, McLean J. Coronary vasoactivity of adenosine in the conscious dog. Circ Res. 1979;45:468-478.
Rubio R, Berne RM, Katori M. Release of adenosine in reactive hyperemia of the dog heart. Am J Physiol. 1969;216:56-62.
Olsson RA, Snow JA, Gentry MK. Adenosine metabolism in canine myocardial reactive hyperemia. Circ Res. 1978;42:358-362.
Kroll K, Stepp DW. Adenosine kinetics in the canine coronary circulation. Am J Physiol. 1996;270:H1469-H1483.
Headrick JP, Berne RM. Endothelium-dependent and -independent relaxations to adenosine in guinea pig aorta. Am J Physiol. 1990;259:H62-H67.
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.
Wangler RD, Gorman MW, Wang CY, Dewitt DF, Chan IS, Bassingthwaighte JB, Sparks HV. Transcapillary adenosine transport and interstitial adenosine concentration in guinea pig hearts. Am J Physiol. 1989;257:H89-H106.
Koberstein RC, Pittman DE, Klocke FJ. Right atrial admixture in coronary venous blood. Am J Physiol. 1969;216:531-535.
Mohrman DE, Feigl EO. Competition between sympathetic vasoconstriction and metabolic vasodilation in the canine coronary circulation. Circ Res. 1978;42:79-86.
Mohrman DE. A servo-controlled roller pump for cardiovascular research. Am J Physiol. 1980;238:H269-H274.
Bassingthwaighte JB, Wang CY, Chan IS. Blood-tissue exchange via transport and transformation by capillary endothelial cells. Circ Res. 1989;65:997-1020.
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.
Cousineau D, Rose CP, Lamoureux D, Goresky CA. Changes in cardiac transcapillary exchange with metabolic coronary vasodilation in the intact dog. Circ Res. 1983;53:719-730.
Anversa P, Levicky V, Beghi C, McDonald SL, Kikkawa Y. Morphometry of exercise-induced right ventricular hypertrophy in the rat. Circ Res. 1983;52:57-64.
Mo¨ser GH, Schrader J, Deussen A. Turnover of adenosine in plasma of human and dog blood. Am J Physiol. 1989;256:C799-C806.
Hanley F, Messina LM, Baer RW, Uhlig PN, Hoffman JIE. Direct measurement of left ventricular interstitial adenosine. Am J Physiol. 1983;245:H327-H335.
Herrmann SC, Feigl EO. Adrenergic blockade blunts adenosine concentration and coronary vasodilation during hypoxia. Circ Res. 1992;70:1203-1216.
Gidday JM, Hill HE, Rubio R, Berne RM. Estimates of left ventricular interstitial fluid adenosine during catecholamine stimulation. Am J Physiol. 1988;254:H207-H216.
Tietjan CS, Tribble CG, Gidday JM, Phillips CL, Belardinelli L, Rubio R, Berne RM. Interstitial adenosine in guinea pig hearts: an index obtained by epicardial disks. Am J Physiol. 1990;259:H1471-H1476.
Hanley FL, Grattan MT, Stevens MB, Hoffman JIE. Role of adenosine in coronary autoregulation. Am J Physiol. 1986;250:H558-H566.
Van Wylen DGL, Willis J, Sodhi J, Weiss RJ, Lasley RD, Mentzer RM Jr. Cardiac microdialysis to estimate interstitial adenosine and coronary blood flow. Am J Physiol. 1990;258:H1642-H1649.
Kroll K, Feigl EO. Adenosine is unimportant in controlling coronary blood flow in unstressed dog hearts. Am J Physiol. 1985;249:H1176-H1187.
Gewirtz H, Olsson RA, Brautigan DL, Brown PR, Most AS. Adenosine's role in regulating basal coronary arteriolar tone. Am J Physiol. 1986;250:H1030-H1036.
Olsson RA, Snow JA, Gentry MK, Frick GP. Adenosine uptake by canine heart. Circ Res. 1972;31:767-778.
Abebe W, Makujina SR, Mustafa SJ. Adenosine receptor-mediated relaxation of porcine coronary artery in presence and absence of endothelium. Am J Physiol. 1994;266:H2018-H2025.
Van Bibber R, Traub O, Kroll K, Feigl EO. EDRF and norepinephrine-induced coronary vasodilation in the canine coronary circulation. Am J Physiol. 1995;268:H1973-H1981.
Wang T, Mentzer RM Jr, Van Wylen DGL. Interstitial adenosine with dipyridamole: effect of adenosine receptor blockade and adenosine deaminase. Am J Physiol. 1992;263:H552-H558.
Wang T, Sodhi J, Mentzer RM, Van Wylen DGL. Changes in interstitial adenosine during hypoxia: relationship to oxygen supply:demand imbalance, and effects of adenosine deaminase. Cardiovasc Res. 1994;28:1320-1325.
Manfredi JP, Sparks HV Jr. Adenosine's role in coronary vasodilation induced by atrial pacing and norepinephrine. Am J Physiol. 1982;243:H536-H545.
Gorman MW, Ning X-H, He M-X, Portman MA, Sparks HV. Adenosine release and high-energy phosphates in intact dog hearts during norepinephrine infusion. Circ Res. 1992;70:1146-1151.
Kitakaze M, Hori M, Takashima S, Sato H, Inoue M, Kamada T. Ischemic preconditioning increases adenosine release and 5′-nucleotidase activity during myocardial ischemia and reperfusion in dogs: implications for myocardial salvage. Circulation. 1993;87:208-215.