CaMKII/PP1 system was simulated in a cylindrical arena discretized into 6 voxels of equal volume separated by distance = 30 nm. (A) (above) For each voxel, the trajectory of active PP1 vs. time is plotted in blue. (below) Cross-correlation matrix of PP1 activity that is the entry of the matrix shows the value of correlation coefficient of PP1 activity inside voxel and inside voxel (Pearson product-moment correlation coefficient, using numpy.corrcoef function). (B,C) Same as A but with different value of DPP1 , 0.001 µm2 s−1 and 0.1 µm2 s−1, respectively. PP1 activity reduced in all voxels with increased DPP1. The correlation of PP1 activity among voxels did not improve with increased DPP1 therefore non-uniform distribution of PP1 in voxels is unlikely to be a significant contributor to the observed loss of PP1 potency. (D) PP1 activity decreased with increased DPP1 but remained independent of diffusion coefficient of subunit (Dsub). On the y-axis, PP1 activity is measured as ratio of sum of number of all active PP1 in all voxels during the simulation divided by the simulation time in hours. On top, dashed blue line (labeled blue DPP1 = 0) represents the case where PP1 was not allowed to diffuse. At bottom, dashed blue line (labeled blue1-voxel) shows the case where the cylinder consists only of 1 voxel and diffusion is instantaneous that is it is a well-mixed system. As expected, as DPP1 increased, the six voxels system converged to a well-mixed system of 1 voxel of 6x volume. Note that a similar effect is seen for a range of Dsub, including Dsub = µm2 s−1, for which = 0.33 nm, satisfying the condition (Isaacson, 2009; Erban and Chapman, 2009; also see Appendix 1—figure 2). Thus, we do not expect that this is a numerical artifact due to our use of cross-voxel jump reactions to approximate diffusion (Isaacson, 2009).