The Journal of Neuroscience, January 17, 2007, ():
Nanodomains of Single Ca2+ Channels Contribute to Action Potential Repolarization in Cortical Neurons
J. Neurosci. Müller et al.
27: 483
Supplemental Data
Files in this Data Supplement:
- supplemental material
-
Supplementary Figure 1:
To eliminate any bias due to the data point at the origin or due to constraining a linear fit through the origin, we have fitted our data with the full function resulting from the linear approximation of the superposition of several Ca2+ sources, namely equation 20. Because the scheme in Fig. 5C contains 4 channels we have chosen n=3 (in total 4 channels). Thus, we have 4 free fit-parameters, the distances of 4 Ca2+ channels from the BK channel: r0, r1, r2 and r3. The independent variable is x=1/? which is allowed to become 0 in order to represent the 0 buffer condition (infinitely large ?). Note that the graph of this fitting function always starts at the origin irrespective of the fitting parameters. Thus, the data point at the origin does not contribute to the error optimization during the fitting procedure (zero error) and therefore also does not bias the result. If we fit this function to our data the optimization procedure indeed yields a line and all 4 parameters converge to 12.7 nm as illustrated by the black line. This result is also obtained with larger or smaller values of n. Therefore, within the theory of linear approximation of microdomain signaling our data are, in fact, best described by a line and not by a bent curve which results from superposition of multiple Ca2+ microdomains/channels.
To test the sensitivity of our approach in detecting the superposition of microdomains we created arbitrary data points centered around the theoretical curve resulting from the scheme shown in Fig. 5C (dashed black line, channels at 12.7, 50, 60 and 70 nm). These data points are illustrated as red open squares in figure “for_ref_04.jpg”. Fitting these points with equation 20 with n=3, as above (same starting parameters), clearly produces a bend curve (red line) which is typical for multiple Ca2+ channel operating mode (also see Appendix) and which well approximates the theoretical curve (dashed black). The resulting fit parameters are now different. The fit reliably detects the closest channel at around 13 nm and further 3 channels at a distance of ~70 nm. Of course, considering the deviation between the real curve and the arbitrary data points the more remote channels are not exactly located. But If we place the three data points close or onto the theoretical curve all channel distances are extracted with an accuracy of ± 3 nm (not shown). Taken together our approach discriminates reasonably well between a situation of 1:1 coupling from one in which multiple Ca2+ microdomains overlap.
