o slice-select readout allows an independent estimate of the exchange rates to be calculated each TR interval. In modeling the data, we ignored the backward reactions from lactate- or alanine-to-pyruvate. In addition, only pyruvate is assumed to flow into the slice each TR. In practice this second assumption was violated, as significant lactate, produced primarily in the heart, was also observed to flow into the targeted slice. To eliminate this undesired lactate inflow, we added a lactate-selective saturation pulse immediately after each readout. The overall pulse sequence is shown in Figure 1 (A). To simplify modeling, we assume that the inflow rate of pyruvate was constant within each TR interval, hence effectively modeling the pyruvate input function to the slice as a piecewise linear function. The choice of TR is critical. If TR is too small, the constant pyruvate inflow rate assumption well approximates the true inflow curve, but the SNRs of the received signals are low because there is little time for the metabolic conversion. If TR is too large, the SNRs of the pyruvate, lactate, and alanine signals are higher, but the piecewise linear model of the pyruvate bolus may be violated. Based on simulations, an optimal TR value of 5 second was used for the experiments described here. Each TR, the estimated metabolic rate constants are given by Equation 1. According to the Michaelis-Menten enzyme kinetics, the rates of enzyme- catalyzed first-order reactions only proceed linearly with respect to substrate concentration when there is an abundance of enzyme as shown in Equation 2. For high substrate concentrations, the reaction rates approaches a constant, limited by the available enzyme activity as shown in Equation 3. Here, we only give the equations for lactate, as the analogous alanine equations are similar. 1