Or inactivation, but there was still a large area where alternans ispresent. This indicated that recovery of the RyR2 from inactivation was able to sustain alternans in that region. On the other hand, when the fraction of recovered RyR2s was 22948146 clamped (Figure 5C), calcium alternans was also maintained in a large area. Therefore, combining Figures 5A, B, and C allowed us to identify the regions where (see Table 1): 1) alternation in SR calcium load is the only UKI-1 web mechanism underlying calcium alternans (region “L”); 2) recovery of the RyR2 from inactivation is the responsible mechanism (region “R”); 3) both mechanisms are SMER 28 necessary (region “R+L”); 4) either mechanism is able to sustain alternans (region “R, L”). Figure 5D shows how these four regions are distributed as a function of activation and inactivation rates for a pacing frequency of 3 Hz. To further understand the presence of alternans when SR load does not alternate, we considered an idealized situation where: 1) stimulation was done using an action potential clamp, and 2) the SR calcium and 3) the subsarcolemmal calcium were fixed at a constant concentration at all times. This ensures that, if alternans still appears, the RyR2 dynamics is its only possible source. From a mathematical analysis of this case (see Section 2 in Appendix S1) we demonstrate the presence of an instability that gives rise to alternans, through a period-doubling bifurcation (Figure S4 in Appendix S1). The instability is inherent to the RyR2 dynamics and requires a stimulation period shorter than its recovery time from inactivation (Figure S5 in Appendix S1). We then investigated how the stimulation frequency affects the relative relevance of the different mechanisms, recalculating Figure 5D at different pacing rates (2 Hz, 3 Hz and 4 Hz) and the results are summarized in Figure 6A.Effect of Changes in the Recovery Time of the RyR2 from InactivationFigure 6B shows that the boundaries of calcium alternans enlarge as the time for recovery of the RyR2 from inactivation increases from 200 ms to our standard value of 750 ms, andCa2+ Alternans 
and RyR2 RefractorinessFigure 3. Slowing of RyR2 activation or inactivation induces calcium alternans at physiological pacing rates. A) The effect of increasing the stimulation frequency from 3 Hz to 5 Hz on trasmembrane potential (top panel), fraction of recovered RyRs (top middle panel), SR calcium load (lower middle panel) and cytosolic calcium (lower panel) for fixed activation and inactivation rates of ka = 8.5 mM22 ms21, ki = 0.17 mM21 ms21 with a recovery time from inactivation of tr = 1/kim = 750 ms. B), C), and D) 
Color-code graphs showing the amplitude of alternations in the calcium transient amplitude as a function of RyR2 activation and inactivation at a pacing rate of 1 Hz (B), 2 Hz (C), and 3 Hz (D). The horizontal axis represents the RyR2 inactivation rate, while the vertical axis represents the RyR2 activation rate. The alternans amplitude, defined as the difference in peak cytosolic calcium between two consecutive beats, is given in color code with blue representing no alternans and dark red corresponding to strong alternations in peak values. The gray area represents cases where a complex beat-to-beat behavior is observed, including 3:1 or 4:1 rhythms, or seemingly chaotic dynamics. E) Borders for the transition to cytosolic calcium alternans obtained with different pacing frequencies. doi:10.1371/journal.pone.0055042.gfurther to 1500 ms. To expand t.Or inactivation, but there was still a large area where alternans ispresent. This indicated that recovery of the RyR2 from inactivation was able to sustain alternans in that region. On the other hand, when the fraction of recovered RyR2s was 22948146 clamped (Figure 5C), calcium alternans was also maintained in a large area. Therefore, combining Figures 5A, B, and C allowed us to identify the regions where (see Table 1): 1) alternation in SR calcium load is the only mechanism underlying calcium alternans (region “L”); 2) recovery of the RyR2 from inactivation is the responsible mechanism (region “R”); 3) both mechanisms are necessary (region “R+L”); 4) either mechanism is able to sustain alternans (region “R, L”). Figure 5D shows how these four regions are distributed as a function of activation and inactivation rates for a pacing frequency of 3 Hz. To further understand the presence of alternans when SR load does not alternate, we considered an idealized situation where: 1) stimulation was done using an action potential clamp, and 2) the SR calcium and 3) the subsarcolemmal calcium were fixed at a constant concentration at all times. This ensures that, if alternans still appears, the RyR2 dynamics is its only possible source. From a mathematical analysis of this case (see Section 2 in Appendix S1) we demonstrate the presence of an instability that gives rise to alternans, through a period-doubling bifurcation (Figure S4 in Appendix S1). The instability is inherent to the RyR2 dynamics and requires a stimulation period shorter than its recovery time from inactivation (Figure S5 in Appendix S1). We then investigated how the stimulation frequency affects the relative relevance of the different mechanisms, recalculating Figure 5D at different pacing rates (2 Hz, 3 Hz and 4 Hz) and the results are summarized in Figure 6A.Effect of Changes in the Recovery Time of the RyR2 from InactivationFigure 6B shows that the boundaries of calcium alternans enlarge as the time for recovery of the RyR2 from inactivation increases from 200 ms to our standard value of 750 ms, andCa2+ Alternans and RyR2 RefractorinessFigure 3. Slowing of RyR2 activation or inactivation induces calcium alternans at physiological pacing rates. A) The effect of increasing the stimulation frequency from 3 Hz to 5 Hz on trasmembrane potential (top panel), fraction of recovered RyRs (top middle panel), SR calcium load (lower middle panel) and cytosolic calcium (lower panel) for fixed activation and inactivation rates of ka = 8.5 mM22 ms21, ki = 0.17 mM21 ms21 with a recovery time from inactivation of tr = 1/kim = 750 ms. B), C), and D) Color-code graphs showing the amplitude of alternations in the calcium transient amplitude as a function of RyR2 activation and inactivation at a pacing rate of 1 Hz (B), 2 Hz (C), and 3 Hz (D). The horizontal axis represents the RyR2 inactivation rate, while the vertical axis represents the RyR2 activation rate. The alternans amplitude, defined as the difference in peak cytosolic calcium between two consecutive beats, is given in color code with blue representing no alternans and dark red corresponding to strong alternations in peak values. The gray area represents cases where a complex beat-to-beat behavior is observed, including 3:1 or 4:1 rhythms, or seemingly chaotic dynamics. E) Borders for the transition to cytosolic calcium alternans obtained with different pacing frequencies. doi:10.1371/journal.pone.0055042.gfurther to 1500 ms. To expand t.