The role of adenosine in myocardial and vascular protection is highly complex. Comprehensive experimental studies on the isolated heart, vessels, and cell systems have shown that the nucleotide is distributed in different regions and coupled to a multitude of processes. Thus, even simple flow experiments with input-output measurements of adenosine concentrations are difficult to interpret. To elucidate the mechanisms governing an important part of the adenosine metabolism - the extracellular hydrolysis of AMP to adenosine - a mathematical model analysis was performed, based on data from experiments with isolated human internal mammary artery. The vessel grafts were mounted in a flow chamber setup and superfused with HEPES buffer at a constant flow of 0.2 ml min^-1. Infusing 5 mM etheno-labeled AMP throughout the experiments, samples of the effluent perfusate were taken at a time resolution of 2.0 min and analyzed for concentrations of etheno-AMP and etheno-adenosine with fluorescence HPLC. To assess the relative importance of the conversion pathways of AMP to adenosine via CD73 and alkaline phosphatase (AP), respective competitive inhibitors, alpha-beta-methylene-adenosine-diphosphate (AOPCP; 50 mM) and levamisole (500 mM), were applied for defined periods. The developed mathematical model of the adenosine metabolism simulated the etheno-AMP and etheno-adenosine concentrations at the outflow as function of time. The conversion was accounted for by introducing Michaelis-Menten kinetics for each pathway, delivering parameter estimates for the K_i of the inhibitors and for the V_max of CD 73 and AP. The experimental geometry simulated as a sequence of well-stirred tanks was validated via a series of test experiments with non-reactive tracer substances. The application of Bayesian data analysis techniques allowed a logically consistent inclusion of prior knowledge about the model parameters, and delivered a reliable estimation of parameter values and errors. Computations were performed using present-day Monte-Carlo-Markov-Chain algorithms.

On the basis of set initial K_m values for CD 73 at 25 mM and for AP at 500 mM, respective V_max of ~ 2 nmol ^. min^-1 and ~ 10 nmol ^. min^-1 resulted for the analyzed vessel samples. Typical estimations of K_i values were 35 mM for AOPCP and 1mM for Levamisole. The calculated time-dependent output concentrations of the mathematical model were in good accordance with the experimental data. In summary, the mathematical model of the setup permits the quantitative evaluation of the assumed intrinsic metabolic processes. It allows to critically examine the underlying concepts in the light of experimental data.