In skeletal and cardiac muscle, calcium release from the sarcoplasmic retic
ulum, leading to contraction, often results in calcium sparks. Because spar
ks are recorded by confocal microscopy in line-scanning mode, their measure
d amplitude depends on their true amplitude and the position of the spark r
elative to the scanned line. We present a method to derive from measured am
plitude histograms the actual distribution of spark amplitudes. The method
worked well when tested on simulated distributions of experimental sparks.
Applied to massive numbers of sparks imaged in frog skeletal muscle under v
oltage clamp in reference conditions, the method yielded either a decaying
amplitude distribution (6 cells) or one with a central mode (5 cells). Caff
eine at 0.5 or 1 mM reversibly enhanced this mode (5 cells) or induced its
appearance (4 cells). The occurrence of a mode in the amplitude distributio
n was highly correlated with the presence of a mode in the distribution of
spark rise times or in the joint distribution of rise times and spatial wid
ths. If sparks were produced by individual Markovian release channels evolv
ing reversibly, they should not have a preferred rise time or amplitude. Ch
annel groups, instead, could cooperate allosterically or through their calc
ium sensitivity, and give rise to a stereotyped amplitude in their collecti
ve spark.