Late Ca2+ Sparks and Ripples During the Systolic Ca2+ Transient in Heart Muscle Cells

Supplemental Digital Content is available in the text.

The fractional block of LTCC by Cd 2+ was calculated from the Hill equation with half-maximal inhibition occurring at 2.14 μmol/L and a Hill coefficient of 0.74 (2). The onset of LTCC block during solution exchange was calculated from the normalised change in the sulforhodamine B fluorescence signal. Note that Cd 2+ is a weak blocker of NCX (by <1 % and 15 %, at 10 and 100 μmol/L respectively (2)) and TTXsensitive Na + channels (by <1 % and 23 %, at 10 and 100 μmol/L respectively (3)). Cd 2+ was chosen over organic pharmacological blockers because LTCC block by Cd 2+ is rapid and not voltage or usedependent.
Confocal Ca 2+ line scan recording Cells were loaded with 5 μmol/L Fluo-4-AM for 15 min, washed in NT and then allowed to rested for >10 min to allow time for de-esterification. Ca 2+ sparks and transients were recorded in line scan mode using an inverted confocal microscope (LSM 880, Zeiss) with a 1.4 NA 63x oil immersion lens. Excitation light was provided by a 488 nm argon laser and fluorescence emission collected at 492-600 nm. Ca 2+ line scans were recorded with the pinhole set to <2 Airy units, at a pixel size of 0.1-0.2 μm/pixel and with a scan speed of 1-2 ms per line. GaAsP photodetectors were used to increase the sensitivity of Ca 2+ spark detection. Ca 2+ line scans were recorded with the pinhole set to <2 Airy units, pixel size <0.2 μm/pixel and scan speed of 1-2 ms/line. The local concentration of Cd 2+ was measured from the included sulforhodamine-B fluorescence which was excited at 543 nm and emission at >600 nm.

Fluorescence image processing
Non-cell background fluorescence from an area adjacent to the cell was subtracted from recordings. Variations in fluorescence due to dye loading was minimised by normalising fluorescence (F) to resting fluorescence during a 100 ms quiescent period immediately before stimulation (F 0 ). The F/F 0 recording was converted into units of [Ca 2+ ] using the self-ratio method (4): Where K is the in vivo affinity of Fluo-4 for Ca 2+ (K d ~1000 nmol/L), R is the self-ratio fluorescence (F/F 0 ), and [Ca 2+ ] rest is the resting Ca 2+ concentration (~100 nmol/L) (5).
During systolic Ca 2+ transients, the increased cytosolic [Ca 2+ ] presents a challenge for the detection of LCS due to the reduced contrast of fluorescent Ca 2+ dyes at high cytosolic [Ca 2+ ]. To partially ameliorate this problem, the low-frequency time-averaged fluorescence in xt line scan recordings was subtracted from the F/F 0 recording. A low-pass quadratic Savitsky-Golay filter (window size ~301 ms) was applied along the t dimension, for every point in the x dimension. These filter values were found to effectively suppress background fluorescence variation due to the underlying Ca 2+ transient, while preserving the morphology and enhancing detectability of LCS.
Ca 2+ spark detection Ca 2+ sparks were detected using an automated optimal filter algorithm implemented in MATLAB (described in detail elsewhere (6)). Briefly, the algorithm cross-correlated the flattened line scan recording with a model Ca 2+ spark and the location with the greatest correlation was identified. Following a test for significance, the centroid and amplitude of the underlying Ca 2+ spark was then measured and recorded. The Ca 2+ spark identification process was repeated until the significance of the maximum correlation fell below the threshold of significance. Ca 2+ spark full-width at half maximum and fullduration at half maximum were measured in the flattened line scan recordings. Ca 2+

T-tubule imaging and processing
The t-tubule system in the area surrounding the Ca 2+ line scan recording was imaged by labelling the sarcolemma with di-8-ANEPPS from a stock 1 mmol/L solution (in DMSO) added directly to the cell recording chamber (final concentration 1 μmol/L) for 2-3 min. A stack of xy images above and below the recording focal plane was recorded using 488 nm excitation and emission collected at >600 nm.
3D stacks of t-tubule images were deconvolved using a model point spread function for the microscope objective which was derived from images of 100 nm fluorescent beads. A 3D t-tubule detection algorithm was implemented in MATLAB by the authors (who can be contacted for further details). The resulting data was then skeletonized in 3D in MATLAB (Skeleton 3D, version 1.12).
The origin of Ca 2+ sparks in three dimensions (xyz) is uncertain in line scanning due to the spatial spread of Ca 2+ sparks from sites just outside the scanned line. To minimize this problem, the t-tubule skeleton was collapsed to form a maximal intensity projection of the region ±2 μm above and below, and ±1 μm adjacent to the line scan region to capture the possible location of all closely coupled jSR release sites. The Euclidean distance from the apparent LCS centroid to nearest t-tubule was calculated from this maximal intensity projection.

LCS propagation/interaction analysis using 2D autocorrelation
2D autocorrelation was used to analyze the temporal and spatial relations between late Ca 2+ sparks and Ca 2+ ripples in Ca 2+ line scan records. The autocorrelogram axes represent time (lag) and space (displacement), hence the distance and angle of structures relative to the origin of the autocorrelogram corresponds to the dominant frequency and velocity of LCS propagation in the form of Ca 2+ ripples. The autocorrelation values in the time domain were measured from the origin and normalised to the autocorrelation value at the origin. The angle from the origin to the peaks of the autocorrelation correspond to propagation velocities of Ca 2+ ripples (~50-250 μm/s).
To test whether the apparent peaks in the autocorrelogram were significant, we scrambled the data and repeated the autocorrelation. This was repeated 5 times to compute a mean and standard deviation for equivalent uncorrelated data.

Statistical Analysis
Data were tested for normality using the Shapiro-Wilk test and in cases where data were skewed, the test was reapplied to log-transformed data. Paired t-tests were performed on normally distributed original data or the transformed data. Differences between cumulative frequencies were tested using the Kolmogorov-Smirnov test. Curve fits of LCS probability (P LCS ) were compared using the extra sum-of-squares F test. Results are presented as mean ± SEM. The number of cells (n) and animals (N) used for each experiment are given in figure legends as n/N. p<0.05 was considered to be the limit of statistical confidence. A nested hierarchical approach was used to examine possible clustering effects in Fig. 1 and the ICC was found to be only 0.14 (7).

Effects of BDM.
BDM is not a specific inhibitor of muscle contraction (8) and partially inhibits many of the key Ca 2+ transport systems in cardiomyocytes. It has been shown to affect NCX (9), Ca 2+ channels (10) and weakly enhances PP1 and/or PP2A phosphatase activity (the latter at 100 mmol/l) (11). A reduction in LTCC availability should increase the probability of LCS as shown in Fig. 3a. On the other hand, 10 mM BDM can reduce SR Ca 2+ content with only small effects on the amplitude of the Ca 2+ transient (12). Therefore, BDM may alter the exact relationship between triggers and the consequent Ca 2+ release but is clearly not the cause of LCS activity seen here (see Online Figure I Figure 1. On the right, the pacing rate was increased to 2Hz at 37 o C (with BDM to avoid the movement artifact) and, while the increased Ca 2+ transient amplitude decreases contrast and increases Poisson noise, LCS can still be discerned.

Online Figure II. Mechanisms for interactions between Ca 2+ sparks, LCS and the ECC cycle.
During the cardiac AP, Ca 2+ enters the cell via LTCC (seen macroscopically as I Ca ) which causes the nearsynchronous release of Ca 2+ from jSR by Ca 2+ -induced Ca 2+ release (CICR). Activated SR junctions enter a refractory state where further release is prevented due to depletion of SR Ca 2+ . Ca 2+ -dependent inactivation (CDI) of LTCC may also decrease the probability of LCS. Cytosolic Ca 2+ is re-sequestered by SERCA2a, refilling the SR. During the AP plateau, LTCC may (re-)open and may trigger LCS if the SR release sites were triggered during the AP or else recovered from the refractory state. Inward current generated by NCX will prolong the AP and increase the duration of I Ca . By delaying the decline of the Ca 2+ transient, LCS may promote additional LCS which may take the form of Ca 2+ ripples, if SR load is sufficient and SR release not refractory. The non-LTCC triggered probability of LCS triggering can be described by an equation of the form f<Ca 2+ cyt > x g< Ca 2+ SR > where f and g describe the cytosolic and SR dependencies of Ca 2+ spark initiation. This may account for ~60% of LCS in normal cells (the remainder being due to LTCC activity). In disease, the pathological changes in the ECC cycle can increase the probability of LCS which, in turn, may prolong the duration of the Ca 2+ transient and AP duration forming a new positive feedback pathway.