Data-based theoretical identification of subcellular calcium compartments and estimation of calcium dynamics in cardiac myocytes.

In cardiac cells, Ca(2+) release flux (J(rel)) via ryanodine receptors (RyRs) from the sarcoplasmic reticulum (SR) has a complex effect on the action potential (AP). Coupling between J(rel) and the AP occurs via L-type Ca(2+) channels (I(Ca)) and the Na(+)/Ca(2+) exchanger (I(NCX)). We used a combined experimental and modelling approach to study interactions between J(rel), I(Ca) and I(NCX) in porcine ventricular myocytes.We tested the hypothesis that during normal uniform J(rel), the interaction between these fluxes can be represented as occurring in two myoplasmic subcompartments for Ca(2+) distribution, one (T-space) associated with RyR and enclosed by the junctional portion of the SR membrane and corresponding T-tubular portion of the sarcolemma, the other (M-space) encompassing the rest of the myoplasm. I(Ca) and I(NCX) were partitioned into subpopulations in the T-space and M-space sarcolemma. We denoted free Ca(2+) concentrations in T-space and M-space Ca(t) and Ca(m), respectively. Experiments were designed to allow separate measurements of I(Ca) and I(NCX) as a function of J(rel). Inclusion of T-space in themodel allowed us to reproduce in silico the following important experimental results: (1) hysteresis of I(NCX) dependence on Ca(m); (2) delay between peak I(NCX) and peak Ca(m) during caffeine application protocol; (3) delay between I(NCX) and Ca(m) during Ca(2+)-induced-Ca(2+)-release; (4) rapid I(Ca) inactivation (within 2 ms) due to J(rel), with magnitude graded as a function of the SR Ca(2+) content; (5) time delay between I(Ca) inactivation due to J(rel) and Ca(m). Partition of 25% NCX in T-space and 75% in M-space provided the best fit to the experimental data. Measured Ca(m) and I(Ca) or I(NCX) were used as input to the model for estimating Ca(t). The actual model-computed Ca(t), obtained by simulating specific experimental protocols, was used as a gold standard for comparison. The model predicted peak Ca(t) in the range of 6­25 µM, with time to equilibrium of Ca(t) with Ca(m) of ~350 ms. These Ca(t) values are in the range of LCC and RyR sensitivity to Ca(2+). An increase of the SR Ca(2+) load increased the time to equilibrium. The I(Ca)-based estimation method was most accurate during the ascending phase of Ca(t). The I(NCX)-based method provided a good estimate for the descending phase of Ca(t). Thus, application of both methods in combination provides the best estimate of the entire Ca(t) time course.