Our central question was whether IP₃-evoked Ca²⁺ liberation during cell-wide Ca²⁺ signals arises through coordinated activation of pulsatile, spatially -localized events, analogous to the local Ca²⁺ puffs observed with weaker IP₃ stimulation or after loading cells with EGTA to suppress global signals (Dargan and Parker, 2003; Smith et al., 2009). To better visualize and identify transient, localized Ca²⁺ events occurring during the course of larger, global elevations of Ca²⁺, we developed an image processing algorithm to highlight and quantify temporal fluctuations of the Ca²⁺ fluorescence signal. We previously described the use of pixel-by-pixel power spectrum analysis of temporal Ca²⁺ fluctuations for this purpose (Swaminathan et al., 2020), but this is computationally intensive and unfeasible for large data sets. Here, we adopted a faster approximation, by first temporally band-pass filtering image stacks and then calculating the standard deviation (SD) of the fluorescence fluctuations at each pixel over a running time window (Ellefsen et al., 2019).

The conceptual basis of the algorithm is illustrated in Figure 1 (see also Figure 1—video 1). WT HEK293 cells were loaded with the fluorescent Ca²⁺ indicator Cal520 and imaged by TIRF microscopy during global cytosolic Ca²⁺ signals. The panels in Figure 1A show Cal520 fluorescence of individual image frames of a HEK cell captured before (i) and after (ii-v) photorelease of i-IP₃, an active, metabolically stable analog of IP₃. Photoreleased i-IP₃ evoked a widespread increase in fluorescence throughout the cell that peaked within about 5 s, during which time several transient, local ‘hot spots’ were evident. These are visible in Figure 1—video 1 but are not readily apparent in Figure 1A because of the extended grey scale required to encompass the peak global fluorescence signal. To illustrate the activity at local hot spots, we monitored fluorescence from regions of interest (ROIs) centered on 24 sites (Figure 1C). Traces from these sites showed progressive, large fluorescence increases above the baseline, with small, superimposed transients (puffs) during the rising phase. To better discriminate these localized signals, we high-pass (1 Hz) filtered the image stack, pixel-by-pixel, to strip out the slow increase in global fluorescence. Figure 1D shows mean power spectra averaged from the 24 sites during image sequences (5 s) acquired before (control; blue trace) and immediately following (red trace) photorelease of i-IP₃. The control, baseline spectrum showed substantially uniform power across all frequencies above the applied 1 Hz high-pass filter, compatible with the dominant noise source arising from ‘white’ photon shot noise. Strikingly, the spectrum obtained during the rise of the global Ca²⁺ signal showed much greater power at frequencies between about 1–20 Hz as compared to the control spectrum, rolling off at higher frequencies to a noise floor determined by photon shot noise. We thus developed an approach to isolate the low-frequency fluctuations attributable to transient Ca²⁺ puffs, while subtracting the photon shot noise that would arise in linear proportion to the overall fluorescence intensity.

Figure 1 with 4 supplements see all

Download asset Open asset

Beginning with a black level-subtracted ‘raw’ fluorescence image stack, our algorithm applied a spatial filter (Gaussian blur with sigma ~1 µm), and a band-pass temporal Butterworth filter (3–20 Hz). The resulting image stack was then processed by a running boxcar window (160 ms) that, for each pixel, calculated the standard deviation (SD) of the fluorescence signal at that pixel throughout the duration of the window. These parameters were chosen to optimally ‘tune’ the algorithm to reject slow changes in baseline fluorescence and attenuate high-frequency photon shot noise while retaining frequencies resulting from puff activity (Figure 1—figure supplement 1). Lastly, the algorithm corrected for photon shot noise by subtracting a scaled measure of the square root of fluorescence intensity at each pixel. If measurements were in terms of numbers of detected photons, the SD would equal the square root of the intensity; however, that was not the case for our records because of considerations including the camera conversion factor and the filtering applied to the image stack. We thus empirically determined an appropriate scaling factor, by determining the linear slope of a plot of mean variance vs. mean fluorescence emission from a sample of fluorescein where photon shot noise was expected to be the major noise source (Figure 1—figure supplement 2).

Figure 1B presents representative SD images calculated by the algorithm, at time points corresponding to the panels in Figure 1A, and Figure 1—video 1 shows fluorescence and SD images throughout the response. The SD signal was uniformly close to zero throughout the cell before stimulation (Figure 1B, panel i), while discrete, transient hot spots were clearly evident at several different sites during the rising phase of the global Ca²⁺ elevation (panels ii-iv), but ceased at the time of the peak response (panel v). This behavior is further illustrated by the black traces in Figure 1E, showing overlaid SD measurements from the 24 hot spots of activity. A flurry of transient events at these sites peaked during the rising phase of the global Ca²⁺ response to photoreleased i-IP₃ but had largely subsided by the time of the maximal global Ca²⁺ elevation. Even though the global Ca²⁺ level then stayed elevated for many seconds the mean SD signals at these regions remained low. Measurement of the SD signal derived from a ROI encompassing the entire cell (yellow trace, Figure 1E) closely tracked the aggregate kinetics of the individual puff sites.

To further validate the fluctuation analysis algorithm, we examined a situation where cytosolic [Ca²⁺] was expected to rise in a smoothly graded manner, without overt temporal fluctuations or spatial heterogeneities. For this, we imaged Cal520 fluorescence by TIRF microscopy in HEK293 3KO cells in which all IP₃R isoforms were knocked out (Alzayady et al., 2016). We pipetted an aliquot of ionomycin (10 μl of 10 µM) into the 2.5 ml volume of Ca²⁺-free bathing solution at a distance from the cell chosen so that the diffusion of ionomycin evoked a slow liberation of Ca²⁺ from intracellular stores to give a fluorescence signal of similar amplitude (8.3 ΔF/F₀) and kinetics to that evoked by photoreleased i-IP₃ (6.9 ΔF/F₀) in Figure 1A,C. Figure 1F shows snapshots of ‘raw’ fluorescence captured before (i) and during (ii-v) application of ionomycin. The fluorescence rose uniformly throughout the cell without any evident hot spots of local transients in the SD images (Figure 1G and Figure 1—video 1). Measurements from 24 randomly located ROIs (squares in Figure 1F) showed only smooth rises in fluorescence (Figure 1H). Mean spectra from these regions (Figure 1I) displayed flat, substantially uniform distributions of power across all frequencies, consistent with photon shot noise increasing in proportion to the mean fluorescence level. Notably, SD signals from local ROIs (Figure 1J, superimposed black traces) and from a ROI encompassing the entire cell (yellow trace) showed no increase in fluctuations beyond that expected for photon shot noise.

The SD image stacks generated by the temporal fluctuation algorithm showed transient hot spots of Ca²⁺ release associated with temporal fluctuations. However, the SD signal could also include temporal fluctuations in fluorescence that were spatially blurred or uniform across the cell. To determine whether these contribute appreciably, or whether the SD signal could be taken as a good reporter of localized puff activity, we developed a second algorithm to reveal spatial Ca²⁺ variations in Cal520 fluorescence image stacks (Figure 1—figure supplement 3).

Ca²⁺ image stacks were first temporally bandpass filtered as described above. The algorithm then calculated, frame by frame, the difference between strong and weak Gaussian blur functions (respective standard deviations of about 4 and 1 μm at the specimen), essentially acting as a spatial bandpass filter to attenuate high spatial frequencies caused by pixel-to-pixel shot noise variations and low-frequency variations resulting from the spread of Ca²⁺ waves across the cell, while retaining spatial frequencies corresponding to the spread of local Ca²⁺ puffs. The resulting spatial SD images were remarkably similar to images generated by the temporal fluctuation analysis routine (Figure 1—figure supplement 3A), and traces of mean cell-wide temporal and spatial SD signals during Ca²⁺ elevations matched closely (Figure 1—figure supplement 3B–E). We thus conclude that the temporal SD signals faithfully reflect transient, localized Ca²⁺ puff activity while minimizing confounding contributions from shot noise and slower changes in global fluorescence.

In Figure 1, we show traces from discrete subcellular regions to illustrate how temporal SD images detect transient, local Ca²⁺ elevations while being insensitive to homogeneous global Ca²⁺ elevations. However, for all the following experiments in this paper we show SD signals derived from single ROIs that completely encompassed each cell, so as to obtain an aggregate measure of puff activity throughout the cell and obviate any subjective bias that might arise in selecting smaller, subcellular regions. Unless otherwise stated, all imaging was done by TIRF microscopy with cells bathed in a zero Ca²⁺ solution including 300 µM EGTA to avoid possible complication from entry of extracellular Ca²⁺ into the cytosol.

Figure 2 and Figure 2—videos 1 and 2 present records from WT HEK cells loaded with Cal520 and caged i-IP₃ showing how the SD signal reveals the patterns of puff activity underlying global Ca²⁺ signals. Under basal conditions, the shot noise-corrected cell-wide SD signals were almost flat, with a mean around zero (Figure 2A, Figure 2—video 1), indicating a negligible level of local Ca²⁺ activity at rest. Photorelease of small amounts of i-IP₃ by brief (~100 ms) UV flashes evoked Ca²⁺ puffs - directly visible in the Cal520 fluorescence ratio movie in Figure 2—video 1, and more evident as sharp transients in the whole-cell SD trace - but without generating any appreciable global rise in basal Ca²⁺ (Figure 2B). Longer flashes (200–1000 ms) generated whole-cell elevations in cytosolic Ca²⁺ that rose and fell over several seconds, with fluorescence signals reaching peak amplitudes in rough proportion to the flash duration (smooth traces, Figure 2C–E; Figure 2—video 2). SD movies (Figure 2—video 2) and whole-cell SD traces (noisy traces, Figure 2C–E) revealed an underlying flurry of localized, transient Ca²⁺ events during the rising phase of the global Ca²⁺ responses. In instances where global Ca²⁺ signals were small and slowly rising, the SD traces showed Ca²⁺ transients persisting throughout the prolonged rising phase (Figure 2C). On the other hand, the SD traces from cells exhibiting intermediate (Figure 2D) and fast rising (Figure 2E) global signals revealed Ca²⁺ fluctuations that began almost immediately following photorelease of i-IP₃, reached a maximum during the rising phase of the global signal, but then declined almost to baseline by the peak of the response.

Figure 2 with 3 supplements see all

Download asset Open asset

The records in Figure 2A–E and Figure 2—videos 1 and 2 illustrate representative responses in individual cells. To pool data from multiple cells we grouped records into categories matching the examples in Figure 2B–E: that is responses showing puffs without an appreciable elevation of global Ca²⁺; and slow-, intermediate- and fast- rising global Ca²⁺ responses. Figure 2G shows overlaid traces depicting the mean Cal520 fluorescence ratios (ΔF/F₀) of the global Ca²⁺ responses from cells in these different categories, and Figure 2H shows the associated mean SD traces. Notably, in all three categories where global Ca²⁺ signals were evoked (Figure 2C–E) the mean SD signals were transient, indicating that puff activity was largely confined to the rising phase of the global Ca²⁺ elevation and largely ceased by the time the global signal reached a maximum. The durations of the puff flurries progressively shortened with increasing rates of rise in global Ca²⁺ and the magnitudes of the SD signal at the peak of the flurry activity increased.

UV photorelease of i-IP₃ provides a convenient tool to activate IP₃Rs with precise timing and control of the amount released. However, this IP₃ analog is slowly metabolized by the cell, remaining elevated for minutes following photo-uncaging (Smith et al., 2009; Dakin and Li, 2007), and its uniform release throughout the cell differs from endogenous generation of IP₃ at the cell membrane (Keebler and Taylor, 2017; Lock et al., 2017). We thus compared responses evoked by photoreleased i-IP₃ with those activated by the G-protein coupled muscarinic receptor agonist carbachol (CCH), locally applied through a picospritzer-driven micropipette (puffer pipet) positioned above WT HEK cells bathed in zero Ca²⁺ medium. A brief (5 s) pulse of CCH elicited a rapid, global rise in Ca²⁺ that was accompanied by an underlying burst of local Ca²⁺ signals (Figure 2F; Figure 2—video 3). As with responses evoked by photoreleased i-IP₃, fluctuations arising from local Ca²⁺ signals occurred predominantly during the initial portion of the rising phase and then subsided to near basal levels before the peak of the global response. Figure 2I shows mean traces of whole-cell global Ca²⁺ signals (ΔF/F₀) and SD signals of CCH-evoked responses from 12 cells. Peak fluorescence amplitudes were similar to mean values for 11 cells stimulated by strong photorelease of i-IP₃ (ΔF/F₀ of 8.89 ± 0.3 for CCH vs. 7.27 ± 0.4 for i-IP₃); as were the rising phase kinetics of the global Ca²⁺ signal (rise from 20% to 80% of peak 0.70 s ± 0.05 s for CCH vs. 0.80 s ± 0.06 s for i-IP₃). However, global Ca²⁺ elevations evoked by CCH decayed more rapidly than those evoked by i-IP₃ (fall from 80% to 20% of peak 6.33 s ± 0.3 s for CCH vs 20.05 s ± 3.2 s for i-IP₃) - likely because the slowly-degraded i-IP₃ evoked a more sustained release of Ca²⁺.

Puff activity (SD signal) showed a characteristic rise and fall during the rising phase of global Ca²⁺ signals, and both parameters accelerated with increasing photorelease of i-IP₃ (Figure 2). To investigate the relationship between the bulk Ca²⁺ level and puff activity in a time-independent manner, we took paired measurements of cell-wide SD signals and Ca²⁺ level (ΔF/F₀) at intervals during the rising phase of IP₃-evoked Ca²⁺ elevations. Figure 3A shows a scatter plot of SD vs. ΔF/F₀ values for measurements from the cell in Figure 2D, and Figure 3B plots corresponding mean data pooled from groups of cells that gave i-IP₃-evoked global signals with fast (pink circles), intermediate (green triangles) and slow (blue squares) rising phases. Although the amplitudes of the SD signals were greater for the faster rising responses, all cells showed similar ‘inverted U’ shaped relationships. In all three groups, the SD signal was maximal when the Cal520 fluorescence ratio reached a ΔF/F₀ value of about two and then declined progressively as global Ca²⁺ rose higher. This is illustrated more clearly in Figure 3C, where the curves for the three groups of cells superimpose closely after normalization to the same peak SD level. A closely similar inverted U relationship was observed for Ca²⁺ elevations evoked by CCH (Figure 3D).

Figure 3 with 4 supplements see all

Download asset Open asset

The decline in SD signal at higher Ca²⁺ levels during global signals cannot be attributed to a failure of our algorithm to detect local fluctuations because of saturation of the Cal520 indicator dye. Notably, maximal fluorescence responses evoked by addition of ionomycin in high (10 mM) Ca²⁺-containing medium (ΔF/F₀ of 18.93 ± 1.5; n = 32 cells) considerably exceeded the peak fluorescence level evoked by even strong photorelease of i-IP₃ (mean ΔF/F₀7.27 ± 0.4, n = 11 cells), and were greatly in excess of the fluorescence level (ΔF/F₀ ~2; Figure 3) at which the SD signal began to decline. Moreover, we observed instances of local Ca²⁺ signals even during large global Ca²⁺ elevations (ΔF/F₀ >8; Figure 3—figure supplement 1), and obtained SD signals using the lower affinity indicator fluo8L (Kd 1.86 µM vs. 320 nM for Cal520) confirming that puff activity was similarly suppressed prior to the peak of i-IP₃ evoked Ca (Berridge et al., 2000) elevations (Figure 3—figure supplement 2).

We performed the experiments in Figures 1–3 using a bathing solution containing no added Ca²⁺ together with 300 µM EGTA to specifically monitor the release of Ca²⁺ from intracellular stores without possible confounding signals arising from entry of Ca²⁺ across the plasma membrane. To explore whether these results were representative of responses in more physiological conditions, we examined Ca²⁺ signals evoked by photoreleased i-IP₃ in WT HEK cells bathed in solutions containing 2 mM Ca²⁺ (Figure 3—figure supplement 3). Cell-wide Ca²⁺ responses and flurries of local Ca²⁺ signals closely matched the patterns of activity in cells imaged in the absence of extracellular Ca²⁺ (Figure 3—figure supplement 3B–G), and scatter plots of SD signal vs. global Ca²⁺ fluorescence signal (Figure 3—figure supplement 3H,I) mirrored those in the absence of extracellular Ca²⁺ (Figure 3). Thus, the puff activity during IP₃-evoked global Ca²⁺ elevations appears independent of Ca²⁺ influx into the cell. However, global Ca²⁺ responses decayed more slowly when Ca²⁺ was included in the bath solution (fall_80-20 for strong photoreleased i-IP₃ of 35.87 s ± 3.9 s in 2 mM Ca²⁺ vs. 20.05 s ± 3.2 s in zero Ca²⁺; Fall_80-20 for CCH of 13.06 s ± 0.5 s in 2 mM Ca²⁺ vs 6.33 s ± 0.3 in zero Ca²⁺). A likely explanation is that influx through slowly activating store-operated channels prolongs the response when extracellular Ca²⁺ is present.

Mitochondria and lysosomes help shape intercellular Ca²⁺ dynamics by accumulating and releasing Ca²⁺ (Rizzuto et al., 2012; Mammucari et al., 2018; Morgan et al., 2011; Yang et al., 2019). To examine whether activity of these organelles influenced the spatial-temporal occurrence of puffs during IP₃-evoked global Ca²⁺ signals, we treated WT HEK cells for 10 min with FCCP to inhibit mitochondrial (Stout et al., 1998; Jensen and Rehder, 1991) and lysosomal (Churchill et al., 2002) Ca²⁺ uptake by dissipating the proton gradient necessary for Ca²⁺ flux. (Figure 3—figure supplement 4A,B). Mean traces of whole-cell Ca²⁺ fluorescence (ΔF/F₀) and associated SD fluctuations in FCCP-treated cells stimulated with CCH exhibited local and global Ca²⁺ signals similar to vehicle-treated controls, although with slightly smaller peak magnitudes (Figure 3—figure supplement 4C,D). Scatter plots of SD signal vs. bulk Ca²⁺ level during the rising phase of CCH-evoked Ca²⁺ elevations were closely similar in control and following FCCP application (Figure 3—figure supplement 4E).

In light of the resemblance between the inverted U relationship between puff activity and Ca²⁺ level (Figure 3) and the well-known bell-shaped curve for biphasic modulation of IP₃R channel activation by Ca²⁺(Iino, 1990; Bezprozvanny et al., 1991), we considered whether the suppression of puff activity during global elevations might result because IP₃Rs became inhibited by rising cytosolic Ca²⁺ levels. To test this, we first examined the effect of elevating cytosolic Ca²⁺ levels prior to evoking IP₃-mediated Ca²⁺ signals. We loaded HEK WT cells with caged Ca²⁺ (NP-EGTA) and delivered photolysis flashes of varying durations to cause jumps of cytosolic free Ca²⁺ of different magnitudes before locally applying CCH from a puffer pipette (Figure 4A). Although the SD signals evoked by CCH declined progressively with increasing prior photorelease of Ca²⁺, this reduction was matched by a similar diminution in peak amplitudes (ΔF/F₀) of the global Ca²⁺ signal. The open symbols in Figure 4B plot the ratio of puff activity (integral under SD traces) relative to the size of the CCH-evoked global Ca²⁺ signal in each cell, and are presented after binning according to the magnitude of the preceding Ca²⁺ jump evoked by photolysis of caged Ca²⁺. Mean ratios (Figure 4B, filled symbols) remained almost constant for all Ca²⁺ jumps; even at levels (ΔF/F₀ >6) corresponding to those where puff activity was strongly suppressed during the rising phase of global responses (Figure 3).

Figure 4

Download asset Open asset

As a complementary approach, we then examined the effect of buffering the rise in cytosolic [Ca²⁺] during global responses by strong cytosolic loading of EGTA.

Figure 4C shows representative SD and ΔF/F₀ traces in response to photoreleased i-IP₃ from a WT HEK cell that was loaded with EGTA by incubation for 1 hr with 15 μM EGTA-AM. The cell showed a typical flurry of puff activity like that in non-EGTA-loaded cells. Puffs ceased before the peak of the global Ca²⁺ signal, even though the amplitude of the signal (2.5 ΔF/F₀) was strongly attenuated. Figure 4D summarizes mean data from multiple cells, plotting paired measurements of cell-wide SD signals and Ca²⁺ level (ΔF/F₀) at intervals during the rising phase of IP₃-evoked Ca²⁺ elevations, as in Figure 3. The data again followed an inverted U relationship (solid circles), but in comparison to control, non EGTA-loaded cells (open circles) the relationship was shifted markedly to the left. Notably, the peak SD signal was attained at a fluorescence level of about 0.40 ΔF/F₀ vs. about 2 ΔF/F₀ for controls, and puffs were substantially suppressed at fluorescence levels (ΔF/F₀ ~2) where the puff activity was near maximal in control cells.

Taken together, these results demonstrate that inhibition of IP₃Rs by elevated cytosolic [Ca²⁺] is not the primary mechanism causing puff activity to terminate during whole-cell Ca²⁺ responses. They further buttress other evidence that the decline in SD signal during the rising phase of the response does not arise because the indicator dye becomes saturated, but faithfully reflects a physiological termination of puff activity.

We next considered the possibility that puff activity may terminate during the rising phase of global Ca²⁺ elevations because of falling luminal ER [Ca²⁺], rather than rising cytosolic [Ca²⁺]. We tested this idea by imaging i-IP₃ evoked global Ca²⁺ signals after partially depleting ER Ca²⁺ stores while minimizing changes in cytosolic free [Ca²⁺].

In a first approach (Figure 5A), we transiently applied cyclopiazonic acid (CPA) to reversibly inhibit SERCA activity (Uyama et al., 1992), resulting in a net leak of Ca²⁺ from the ER and a small elevation of cytosolic Ca²⁺. Following wash-out of CPA, the cell was maintained in Ca²⁺-free medium so that the cellular Ca²⁺ content (including that of the ER) gradually depleted owing to passive and active extrusion across the plasma membrane. After about 4 min the resting cytosolic Ca²⁺ level had returned close to the original baseline, and we delivered a photolysis flash to photorelease i-IP₃. This evoked a substantial elevation in global Ca²⁺, yet the SD signal showed almost no transient puff activity during this response. Similar results were obtained in a further seven cells, as shown by the mean ΔF/F₀ and SD traces in Figure 5B. To confirm that the suppression of puff activity resulted from cellular Ca²⁺ depletion, we repeated this experiment, now making a paired comparison of i-IP₃-evoked responses between cells that were bathed for 30 min after washing out CPA either in Ca²⁺-containing medium to allow ER store refilling (Figure 5C; Figure 5—video 1), or in Ca²⁺-free medium (Figure 5D; Figure 5—video 1). Cells in both groups showed substantial global Ca²⁺ responses that were not appreciably different in peak amplitudes (Figure 5E); but whereas the SD signals showed that puff activity was strongly suppressed in cells maintained in zero Ca²⁺ medium, cells in Ca²⁺-containing medium showed robust puff activity during the rising phase of the response (Figure 5F).

Figure 5 with 1 supplement see all

Download asset Open asset

As an alternative approach to partially deplete ER Ca²⁺ without pharmacological intervention, we evoked Ca²⁺ signals by repeated applications of CCH at 5 min intervals, and compared responses in cells bathed in Ca²⁺-containing (Figure 5G) and Ca²⁺-free solutions (Figure 5H). In both cases, the amplitudes of the global Ca²⁺ signals progressively declined, likely a result of inhibition of IP₃Rs. However, whereas the amplitude of puff activity reported by SD signals in cells bathed in Ca²⁺-containing medium fell roughly in proportion to the amplitude of the global fluorescence signal, puff activity in Ca²⁺-free medium declined abruptly. In the example depicted in Figure 5H, no activity was evident in the SD signal after the fifth stimulus at 20 min even though an appreciable global Ca²⁺ elevation remained. To quantify these data, we determined puff activity as the integral under the SD trace, and plotted the normalized ratio of puff activity vs. peak global Ca²⁺ amplitude (Figure 5I). For cells in Ca²⁺-containing medium, the mean ratio remained constant across successive stimuli (blue squares, Figure 5I), whereas it declined almost to zero for cells in Ca²⁺-free medium (red circles, Figure 5I).

We conclude from these results that Ca²⁺ puff activity is modulated by ER Ca²⁺ store content, and that when stores are partially depleted IP₃ can still evoke Ca²⁺ release by a process that is independent of puff activity, and occurs without detectable temporal fluctuations. We term this mode of Ca²⁺ liberation as ‘diffuse’ release and refer to Ca²⁺ puffs as a ‘punctate’ mode of Ca²⁺ liberation.

In common with many other cell types, WT HEK and HeLa cells express all three major IP₃R isoforms – types 1, 2, and 3 – that are encoded by separate genes and translated into structurally and functionally distinct proteins that co-translationally oligomerize to form heterotetrameric channels. We (Lock et al., 2018) and others (Mataragka and Taylor, 2018) recently demonstrated that all three isoforms can individually mediate Ca²⁺ puffs. We now utilized HEK cells genetically engineered to express single IP₃R isoforms to evaluate the respective roles of each isoform in liberating Ca²⁺ via punctate, localized transients versus sustained, diffuse release.

We evoked Ca²⁺ liberation in WT HEK cells and cells exclusively expressing type 1, 2, or 3 IP₃Rs by local application of CCH (Figure 6). All three single-isoform-expressing cell lines exhibited patterns of responses qualitatively similar to WT cells. The SD traces showed flurries of puffs during the foot and rising phase of global Ca²⁺ signals that ceased before the time of the peak global Ca²⁺ elevation (Figure 6A–H). Nevertheless, notable differences were apparent between the isoforms. Cells expressing IP₃R1 generated whole-cell Ca²⁺ signals having much smaller amplitudes and slower rising phases than WT and R2- and R3-expressing cells, and localized fluctuations persisted longer (Figure 6C,D,I). In contrast, IP₃R2-expressing cells displayed fast rising, large amplitude Ca²⁺ signals, with a transient flurry of Ca²⁺ fluctuations concentrated during the initial portion of the rising phase (Figure 6E,F). Ca²⁺ signals in cells expressing IP₃R3 (Figure 6G,H) were similar in amplitude to WT and IP₃R2-expressing cells, but with slower rates of rise and more prolonged flurries of puffs. Scatter plots of puff activity (SD signal) as a function of the global Ca²⁺ level (ΔF/F₀) during the rising phase of the global response further highlighted these differences (Figure 6J,K). Ca²⁺ fluctuations were maximal when Cal520 fluorescence (ΔF/F₀) rose to roughly 1.5, 6, and 3 for types 1, 2, and 3 IP₃Rs, respectively; and similarly large differences were evident in the global Ca²⁺ level attained when puff activity terminated.

Figure 6 with 1 supplement see all

Download asset Open asset

We utilized HEK cells for most experiments because of the availability of cell lines expressing individual IP₃R isoforms (Alzayady et al., 2016). The patterning of local, transient Ca²⁺ signals during IP₃-mediated whole-cell Ca²⁺ elevations was not unique to this cell type. Stimulation of HeLa cells with histamine also evoked global Ca²⁺ signals accompanied by flurries of local Ca²⁺ activity during the rising phase, which subsided as Ca²⁺ levels continued to rise (Figure 6—figure supplement 1).

The data in Figures 1–6 derive from TIRF imaging of Ca²⁺ signals in close proximity to the plasma membrane, where a majority (~80%) of puff sites in WT HEK cells are located (Lock et al., 2018). However, TIRF microscopy provides no direct information from the interior of the cell, leaving open the question as to whether slow diffusion of Ca²⁺ ions from puffs at internal sites may contribute to the diffuse component of the Ca²⁺ signal visualized in TIRF images after the puff flurry has ceased. To address this issue, we applied fluctuation analysis to images obtained using lattice light-sheet (LLS) microscopy to record Ca²⁺ signals within diagonal optical ‘slices’ through the cell volume (Ellefsen and Parker, 2018).

Figure 7A,B illustrate LLS Ca²⁺ fluorescence ratio images and corresponding SD images recorded before and after photorelease of i-IP₃ to evoke a global Ca²⁺ response. Similar to observations with TIRF imaging, the SD images revealed local Ca²⁺ transients that began soon after photorelease, and before any appreciable rise in the global Ca²⁺ signal (Figure 7A, panel ii). Discrete events then continued during much of the rising phase of the global signal (panels iii-v) but had largely ceased at the time of the peak global signal (panel vi). In this cell Ca²⁺ puffs were primarily restricted to the cell periphery, whereas Figure 7B shows an example from another cell where local activity was observed both around the periphery and in the cell interior.

Figure 7

Download asset Open asset

Figure 7C,D shows respective measurements from these two cells, plotting fluorescence ratio changes (ΔF/F₀) and SD signals from ROIs encompassing peripheral (red traces) and central (blue traces) regions of the cells, as indicated in the leftmost lower panels of Figure 7A,B. In both cells, the local Ca²⁺ activity monitored by SD fluctuations started within a few hundred ms of the photolysis flash and was maximal during the early portion of the rising phase. The SD signal then declined, returning close to baseline as the global Ca²⁺ signal approached a peak. For the cell illustrated in Figure 7A, the SD signal within the peripheral region was much greater than in the central region, even though the rise in global Ca²⁺ was slightly smaller. In contrast, the cell illustrated in Figure 7B showed a SD signal in the interior that was similar in size to the periphery (Figure 7D). On average, however, mean SD signals from the cell interior were about one quarter of that at the periphery, and fluctuations arising from interior sites followed a similar relation with bulk Ca²⁺ level as peripheral sites (Figure 7E).

Given the relatively low average level of puff activity in the cell interior, and the similar termination of internal and peripheral puff flurries during the rising of global Ca²⁺ signals, we conclude that the diffuse component of the Ca²⁺ rise observed by TIRF microscopy cannot be accounted for by Ca²⁺ spreading from punctate release at internal sites and becoming blurred by diffusion in space and time.

To assess the relative contributions of punctate versus diffuse modes of Ca²⁺ release during global Ca²⁺ signals, we derived the kinetics of Ca²⁺ flux into the cytosol through IP₃Rs on the basis that the cell-wide fluorescence signal reflects a balance between Ca²⁺ release into the cytoplasm and its subsequent removal. To obtain a rate constant for removal of cytosolic Ca²⁺ in WT HEK cells, we recorded the decline of fluorescence Ca²⁺ signals following transient photorelease of Ca²⁺ from caged Ca²⁺ loaded into the cytosol (Figure 8—figure supplement 1A), and during the final ‘tail’ of CCH-evoked Ca²⁺ signals when Ca²⁺ liberation would have almost ceased (Figure 8—figure supplement 1B). Both fitted well to single exponential decay functions, consistent with a dominantly first order removal process, with respective mean rate constants of 0.22 and 0.32 s⁻¹.

We then calculated the instantaneous Ca²⁺ release flux at intervals throughout the time course of a global Ca²⁺ response by differentiating the whole-cell fluorescence Ca²⁺ signal and adding to this the estimated rate of Ca²⁺ removal; for i-IP₃ signals we used a rate constant of 0.22 s⁻¹; for CCH evoked responses we applied rate constants (0.3 s⁻¹ to 0.6 s⁻¹) that were determined from the tail-end of the global Ca²⁺ decay for that particular cell.

We used data from the experiment of Figure 5C,D to compare the kinetics of Ca²⁺ liberation during Ca²⁺ signals under normal conditions, and when puff activity had been inhibited by partial depletion of ER Ca²⁺ store content. Figure 8A,B show records from two representative cells that gave global Ca²⁺ responses of comparable peak amplitudes (black traces). However, whereas the SD signals (grey traces) exhibited the normal flurry of puff activity in the control cell (Figure 8A) this activity was almost completely suppressed in the cell pretreated with CPA (Figure 8B). The red traces show the respective rates of Ca²⁺ release into the cytosol, revealing a larger initial transient of Ca²⁺ liberation in the control cell during the flurry of puff activity. Figure 8C shows overlaid mean traces of Ca²⁺ release from control (n = 5) and CPA-treated cells (n = 6). Colored areas indicate the relative cumulative amounts of Ca²⁺ entering the cytosol (integral under the release trace) in CPA-treated cells where puff activity was substantially abolished (blue shading), and the additional Ca²⁺ flux (pink shading) in control cells showing flurries of puffs. From these respective areas, we estimate that, in normal conditions, the punctate liberation of Ca²⁺ through puff activity contributes about 41% of the total Ca²⁺ release responsible for the initial rise of Ca²⁺ toward its peak. Figure 8D,E further illustrate representative records of SD signals (grey traces), global Ca²⁺ (black), rate of Ca²⁺ release into the cytosol (red), and cumulative amount of Ca²⁺ released (blue) during the entire time course of global Ca²⁺ signals evoked by photoreleased i-IP₃ (Figure 8D) and by CCH (Figure 8E). Because much of the cumulative Ca²⁺ release through IP₃Rs arises from a sustained, low level flux that continues after the peak, Ca²⁺ puffs on average contribute only about 13% of the total Ca²⁺ liberation during global i-IP₃-evoked signals, and about 17% during shorter-lasting responses evoked by CCH (Figure 8F).

Figure 8 with 1 supplement see all

Download asset Open asset

In agreement with previous findings (Bootman et al., 1997a; Marchant et al., 1999) our fluctuation analyses reveal flurries of puffs during the initial rise of IP₃-mediated global Ca²⁺ signals. However, although puff activity was evident during the initial foot of the response and peaked early during the rising phase, it then subsided during the later portion of the rising phase, with few or no transient, local Ca²⁺ signals evident by the time of the peak. Notably, overall Ca²⁺ levels continued to rise after puffs had largely ceased, and cytosolic Ca²⁺ remained elevated for several seconds in the face of rapid removal from the cytosol, during which time Ca²⁺ fluctuations were largely suppressed.

This ‘noise-free’ component of the fluorescence signal cannot be attributed to slow diffusion of Ca²⁺ liberated as puffs to fill in spaces between release sites. Diffusion would be rapid (e.g. mean time of ~300 ms to diffuse 5 μm assuming an effective diffusion coefficient of 20 μm² s⁻¹). In any case, the average fluorescence signal would not be expected to increase appreciably if the total amount of Ca²⁺ in the imaging volume remained constant. Utilizing lightsheet imaging we further excluded the possibility that the continuing rise in near-plasmalemmal Ca²⁺ observed by TIRF imaging might arise through diffusion of Ca²⁺ over longer distances after liberation at sites in the cell interior. Finally, as noted previously, the observation of large global Ca²⁺ signals in the absence of detectable fluorescence fluctuations (Figure 5) definitively points to a mode of Ca²⁺ liberation that is independent of puff activity.

By deriving the time course of cumulative Ca²⁺ liberation during global responses we estimated that puffs contribute only ~41% of the initial Ca²⁺ flux that drives the peak of the Ca²⁺ response, and an even smaller proportion (~15%) of the cumulative flux during its entire time course. Thus, a second component of continuous, spatially diffuse release of intracellular Ca²⁺ is responsible for generating and sustaining a large part of whole-cell Ca²⁺ signals. These two components are further discriminated by procedures that selectively promoted either puff activity (Dargan and Parker, 2003; Smith et al., 2009) (by cytosolic loading of slow Ca²⁺ buffers), or global Ca²⁺ elevations in the absence of localized Ca²⁺ transients (by partial depletion of ER Ca²⁺ store content). We term these two modes of IP₃-mediated Ca²⁺ release as ‘punctate’ (discontinuous in time and space) and ‘diffuse’ (smoothly varying in time and space).

Small elevations of [Ca²⁺] promote opening of IP₃R channels (Iino, 1990; Bezprozvanny et al., 1991) and increase puff frequency (Yamasaki-Mann et al., 2013). The accelerated puff activity during the foot and initial upstroke of the global Ca²⁺ signal is thus likely a consequence of a rising basal cytosolic Ca²⁺ level, resulting both from the puffs themselves and from diffuse Ca²⁺ liberation. This positive feedback of cytosolic Ca²⁺ to promote opening of IP₃Rs is inherently regenerative, so it is imperative that mechanisms exist to ‘put out the fire’. Our results illuminate at least two mechanisms, acting across different time scales, to terminate punctate Ca²⁺ liberation through IP₃Rs. Individual puffs terminate rapidly as IP₃R channels close within tens of ms (Parker et al., 1996; Bootman et al., 1997b) via stochastic inhibition by high local Ca²⁺ levels (Shuai et al., 2008) and potential allosteric interactions between clustered IP₃Rs at a puff site (Wiltgen et al., 2014). However, during the rising phase of a global Ca²⁺ signal, each IP₃R cluster may generate a flurry of repeated puffs - indicating that although the fast-inhibitory processes that terminates individual puffs recovers quickly, a slower process terminates the flurry. This does not appear to result from IP₃Rs becoming inactivated by the rise of bulk cytosolic Ca²⁺, because we found puff flurries were not suppressed by prior Ca²⁺ elevations, and still terminated normally during global responses when cytosolic Ca²⁺ levels were attenuated by buffering with EGTA (Figure 4C,D). Instead, our observations that partial depletion of ER Ca²⁺ stores suppressed puff activity during global Ca²⁺ responses (Figure 5) implicate the depletion of luminal Ca²⁺ as a dominant mechanism responsible for terminating the local signals; analogous to the central role of luminal Ca²⁺ depletion in terminating the Ca²⁺ sparks mediated by ryanodine receptors (Stern et al., 2013). Although individual puffs appear not to affect luminal ER [Ca²⁺] (Lock et al., 2018), it is plausible that ER depletion may occur during the rapid flurries of puffs evoked at multiple sites during the rising phase of global signals. A related question is whether the Ca²⁺ depletion is locally confined to the ER around puff sites and arises through Ca²⁺ released by the puffs themselves, or whether diffuse Ca²⁺ liberation causes global ER Ca²⁺ depletion throughout a luminally continuous ER network (Okubo et al., 2015; Park et al., 2000).

The two modes of Ca²⁺ liberation we describe – punctate and diffuse – might, in principle, arise from two functionally and physically distinct populations of IP₃Rs, or through functional regulation of the clustered IP₃Rs underlying puffs such that they switch from a pulsatile to continuous mode of release. At present, we cannot discriminate between these mechanisms, but favor the former for congruence with studies revealing two distinct populations of IP₃Rs in terms of their spatial distributions and motilities. A small (~30%) fraction of the IP₃Rs in a cell are stationary (Thillaiappan et al., 2017; Smith et al., 2014), grouped in small clusters that are anchored at fixed sites predominantly near the plasma membrane (Smith et al., 2009; Lock et al., 2018; Thillaiappan et al., 2017), whereas the majority of IP₃Rs are distributed throughout the bulk of the cytoplasm and are motile within the ER membrane (Thillaiappan et al., 2017; Smith et al., 2014; Tateishi et al., 2005; Gibson and Ehrlich, 2008; Ferreri-Jacobia et al., 2005; Fukatsu et al., 2004). Puffs are proposed to originate from IP₃Rs within the immotile clusters that are endowed with the ability to preferentially respond to low concentrations of IP₃ (Smith et al., 2009; Thillaiappan et al., 2017). In contrast, the widely distributed, motile IP₃Rs remain apparently silent under conditions when puffs are selectively activated by low concentrations of IP₃ or in the presence of slow cytosolic Ca²⁺ buffers to inhibit global Ca²⁺ elevations (Dargan and Parker, 2003; Dargan et al., 2004). The motile, distributed IP₃Rs would be an attractive candidate for the diffusive mode of Ca²⁺ liberation.

Functional differences between putative populations of IP₃Rs mediating punctate and diffuse Ca²⁺ release cannot be attributed to their being constituted of different isoforms of IP₃R, because all three isoforms mediate Ca²⁺ puffs (Mataragka and Taylor, 2018; Lock et al., 2018), and we show here that cells exclusively expressing individual isoforms generate cell-wide Ca²⁺ elevations involving both punctate and diffuse modes of Ca²⁺ liberation. Instead, the functional properties of the IP₃Rs may be affected by factors including their location in the cell, their mutual proximity to enable interactions by CICR, and by their association with modulatory and anchoring proteins (Prole and Taylor, 2016; Lock et al., 2019; Konieczny et al., 2012).

Stimulation of the IP₃/Ca²⁺ signaling pathway by activation of cell-surface receptors evokes repetitive oscillations in cytosolic Ca²⁺ in numerous cell types (Thomas et al., 1996). Signaling information is encoded in the amplitude and frequency of these Ca²⁺ oscillations, which may have periods ranging widely from a few seconds to minutes. Classical studies illustrate how different frequencies of Ca²⁺ oscillations activate distinct targets, including the selective activation of NFkB in Jurkat T cells by low frequencies (Dolmetsch et al., 1998), and the frequency-dependent control of gene expression in RBL mast cells (Li et al., 1998). However, signaling information is not restricted to frequency and amplitude components of bulk cytosolic Ca²⁺ elevations, and numerous findings implicate a component of spatial Ca²⁺ profiling (Smedler and Uhlén, 2014; Dupont et al., 2011; Berridge et al., 2003; Thul et al., 2009). Because of the restricted diffusion of free Ca²⁺ ions in the cytosol (Allbritton et al., 1992; Schwaller, 2010), the specificity of downstream signaling by Ca²⁺ liberated through IP₃Rs will be strongly influenced by the proximity of target proteins, as well as by their Ca²⁺ binding kinetics (Samanta and Parekh, 2017; Atakpa et al., 2018; Csordás et al., 2010). The two modes of IP₃-evoked Ca²⁺ liberation we describe are, therefore, likely to selectively activate different populations of effectors; those positioned close to the IP₃Rs at puff sites that experience brief, repetitive transients of high local [Ca²⁺], and others that respond to a more sustained, spatially diffuse elevation of bulk cytosolic [Ca²⁺].

Based on a close juxtaposition of stationary IP₃R clusters to ER-plasma membrane junctions where STIM and Orai interact to induce store-operated Ca²⁺ entry (SOCE) (Thillaiappan et al., 2017; Thillaiappan et al., 2019), it has been proposed that local depletion of ER Ca²⁺ content at puff sites might rapidly and selectively activate SOCE, without requiring substantial overall loss of the ER Ca²⁺ that is necessary to sustain numerous ER functions (Thillaiappan et al., 2017; Thillaiappan et al., 2019; Taylor and Machaca, 2019). On the other hand, we previously reported puffs to be unaffected by removal of extracellular Ca²⁺ (Lock et al., 2018), and we show that patterning of local puff activity during IP₃-evoked global Ca²⁺ signals is unaffected by the presence or absence of Ca²⁺ in the bath solution. Thus, influx of extracellular Ca²⁺ does not appear to contribute acutely to the initial flurry of puffs or to the rapid rise in global Ca²⁺, although the more prolonged decay phase of the Ca²⁺ signal in Ca²⁺-containing medium points to a slower activation of SOCE. The relative contributions of local puffs vs. diffuse Ca²⁺ liberation in activating SOCE remain to be determined. Another example of proximate Ca²⁺ signaling is the close tethering between ER and mitochondria (Giorgi et al., 2009) that underlies a rapid shuttling of Ca²⁺ released through IP₃Rs to the mitochondrial matrix (Csordás et al., 2010; Filadi and Pozzan, 2015), such that Ca²⁺ transients within a signaling microdomain, rather than the bulk cytosolic Ca²⁺ signal, regulate mitochondrial bioenergetics and induction of autophagy (Cárdenas et al., 2010). A similar close coupling has recently been described between IP₃Rs and lysosomes (Atakpa et al., 2018). Our description of two modes of IP₃-mediated Ca²⁺ liberation thus raises questions regarding their respective roles in downstream signaling. Is organellar Ca²⁺ signaling via structurally defined nanodomains restricted to the predominantly peripheral ER contact sites where puffs originate, or might a separate population of IP₃Rs that mediate diffuse Ca²⁺ liberation also transmit their signals via restricted domains?

HEK293 wild-type (WT) cells were kindly provided by Dr. David Yule (University of Rochester). An IP₃R null cell line (3KO; #EUR030) and cell lines natively expressing exclusively type 1 (IP₃R1; #EUR031), type 2 (IP₃R2; #EUR032) and type 3 (IP₃R3; #EUR033) isoforms generated from that parental WT HEK293 cell line by CRISPR/Cas9 genetic engineering in the Yule lab were purchased from Kerafast (Boston, MA). HEK cell lines were characterized as described in Alzayady et al., 2016 and were certified mycoplasma free prior to distribution. HeLa cells (#CCL-2) purchased from ATCC (Manassas, VA) were characterized by STR profiling and certified free of mycoplasma prior to distribution. HEK WT, 3KO, and single IP₃R isoform-expressing cell lines were cultured in Eagle’s Minimum Essential Medium (EMEM; ATCC #30–2003), and HeLa cells were cultured in Dulbecco’s modified Eagle Medium (DMEM; #11965092) from Thermo Fisher Scientific (Waltham, MA). Both EMEM and DMEM were supplemented with 10% fetal bovine serum (#FB-11) from Omega Scientific (Tarzana, CA). All cell lines were cultured in plastic 75 cm flasks and maintained at 37°C in a humidified incubator gassed with 95% air and 5% CO₂. For imaging, cells were collected using 0.25% Trypsin-EDTA (Thermo Fisher Scientific #25200–056) and grown on poly-D-lysine (Millipore Sigma #P0899; St. Louis, MO) coated (1 mg/ml) 35 mm glass bottom imaging dishes (#P35-1.5–14 C) from MatTek (Ashland, MA) or 12 mm glass coverslips for 2–3 days.

Immediately before imaging, cells were incubated with the membrane-permeant fluorescent Ca²⁺ indicator Cal520/AM (5 µM; AAT Bioquest #21130; Sunnyvale, CA) for 1 hr at room temperature (RT) in a Ca²⁺-containing HEPES buffered salt solution (Ca²⁺-HBSS). Where indicated cells were additionally loaded with membrane permeant esters of either the caged IP₃ analogue ci-IP₃/PM [D-2,3,-O-Isopropylidene-6-O-(2-nitro-4,5 dimethoxy) benzyl-myo-Inositol 1,4,5,-trisphosphate Hexakis (propionoxymethyl) ester] (1 µM; SiChem #cag-iso-2-145-10; Bremen, Germany) or the caged Ca²⁺ buffer NP-EGTA [o-nitrophenyl EGTA/AM] (200–500 nM; Thermo Fisher Scientific #N6803) in conjunction with Cal520. For the experiments in Figure 4C,D cells were additionally loaded with EGTA/AM (15 µM; Thermo Fisher Scientific #E1219) for 1 hr at RT in Ca²⁺-HBSS to attenuate global Ca²⁺ elevations. To address possible saturation of Cal520 at high Ca²⁺ levels, cells were alternatively loaded with the lower affinity Ca²⁺ indicator fluo8L/AM (5 µM, AAT Bioquest #21096) together with 1 µM ci-IP₃/PM for 1 hr at RT in Ca²⁺-HBSS. Following incubation with AM esters, cells were incubated for 30 min at room temperature in Ca²⁺-HBSS. Cal520/AM, ci-IP₃/PM, and NP-EGTA/AM, EGTA/AM, and fluo8L/AM were all solubilized with DMSO/20% pluronic F127 (Thermo Fisher Scientific #P3000MP).

FCCP (carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone) and CPA (cyclopiazonic acid), were purchased from Millipore Sigma (#C2920 and #C1530, respectively), and solubilized in 100% ethanol. Carbachol (#C4832) and histamine (#H7125), also from Millipore Sigma, were reconstituted in deionized H₂O. Ca²⁺-HBSS contained (in mM) 135 NaCl, 5.4 KCl, 2 CaCl₂, 1 MgCl₂, 10 HEPES, and 10 glucose (pH = 7.4). Nominal Ca²⁺-free HBSS consisted of the same formulation as Ca²⁺-HBSS except that CaCl₂ was omitted; for zero Ca²⁺-HBSS, 300 µM EGTA was added to nominal Ca²⁺-free HBSS. For lattice light-sheet imaging, the plasma membrane was stained by adding Cell Mask Deep Red (Thermo Fisher Scientific #C10046) to the bathing solution in the imaging chamber at 1/50,000 dilution.

Total internal reflection fluorescence (TIRF) imaging of Ca²⁺ signals was accomplished using a home-built system, based around an Olympus (Center Valley, PA) IX50 microscope equipped with an Olympus 60X oil immersion TIRF objective (NA 1.45). Fluorescence images were acquired with an Evolve EMCCD camera (Photometrics; Tucson, AZ) with a bit depth of 16 bits, using 2 × 2 binning for a final field at the specimen of 128 × 128 binned pixels (one binned pixel = 0.53 µm) at a rate of ~125 frames s⁻¹. Image data were streamed to computer memory using Metamorph v7.7 (Molecular Devices; San Jose, CA) and stored on hard disk for offline analysis.

Lattice light-sheet imaging was performed using a home-built system as previously described (Ellefsen and Parker, 2018). Images were acquired with an Andor Zyla 4.2 sCMOS camera (Oxford Instruments; Abingdon, England) from a single, diagonal light-sheet slice (512 × 256 pixels, one pixel = 0.11 µm) at 100 frames s⁻¹ and 16 bit depth. Ca²⁺ signals were imaged in Cal520-loaded cells for several seconds following photorelease of i-IP₃ by a flash from a 405 nm laser diode, utilizing 473 nm laser fluorescence excitation and a 510–560 nm bandpass emission filter. A 562 nm laser and 590 nm long-pass filter were used to image the plasma membrane stained with Cell Mask Deep Red. Images were streamed to disk using MicroManager (Vale Lab UCSF; San Francisco, CA).

Photorelease of i-IP₃ was evoked by UV light from a xenon arc lamp filtered through a 350–400 nm bandpass filter and introduced by a UV-reflecting dichroic in the light path to uniformly illuminate the field of view. The amount of i-IP₃ released was controlled by varying the flash duration, set by an electronically controlled shutter (UniBlitz; Rochester, NY). The same system was used for photolysis of NP-EGTA (i.e. caged Ca²⁺). For the local delivery of solutions to cells during imaging, glass micropipettes were prepared from borosilicate glass capillary filaments (1.5 mm x 0.86 mm, O.D. x I.D.) using a micropipette puller (Sutter Instruments; Novato, CA) to produce tip diameters of ~1–2 µm. Micropipettes were positioned above the cell understudy with local delivery controlled using a pneumatic picospritzer. The delivery of micropipette contents and the duration and intensity of the UV-flash were empirically adjusted to evoke rapid rises in whole-cell cytosolic Ca²⁺ levels.

All imaging was performed while cells were bathed in HBSS containing 2 mM Ca²⁺ or zero Ca²⁺-HBSS containing 0.3 mM EGTA and no added Ca²⁺.

Image data imported in16 bit integer MetaMorph stk or multi-plane TIF format were processed using a script written in Flika (http://flika-org.github.io), a freely available open-source image processing and analysis software in the Python programming language (Ellefsen et al., 2019; Ellefsen et al., 2014). All internal processing and data output were performed using 64 bit floating-point arithmetic.

To analyze and derive movies representing the pixel-by-pixel standard deviation (SD) of temporal fluctuations in fluorescence of the Ca²⁺ indicator dye we used a custom Flika script as described previously (Ellefsen et al., 2019). In brief, following subtraction of camera black offset level, the raw image stack from the camera was first spatially filtered by a Gaussian blur function with a standard deviation (sigma) of 2 pixels (~1 μm at the specimen). The python package (skimage) used to perform this function applies a two-dimensional Gaussian blur with a specified standard deviation (sigma) out to a range of 4 sigma. To attenuate high-frequency photon shot noise and slow changes in baseline fluorescence, it was then temporally filtered with a bandpass Butterworth filter with low and high cutoff frequencies of 3 and 20 Hz. A running variance of this temporally filtered movie was calculated, pixel by pixel, by subtracting the square of the mean from the mean of the square of a moving 20 frame (160 ms) boxcar window of the movie. The running standard deviation was calculated by taking the square root of the variance image stack to create a standard deviation (SD) stack. Finally, to remove the mean predicted photon shot noise which increases in linear proportion to the mean fluorescence intensity, the SD stack was corrected, pixel-by-pixel, by subtracting the square root of a running mean of the spatially filtered fluorescence movie multiplied by a scalar constant; derived as illustrated in Figure 1—figure supplement 2.

We also applied a second FLIKA script to analyze spatial fluctuations in Ca²⁺ image stacks. For this, black-level subtracted image stacks were first temporally band-pass filtered as before, and then new image stacks were derived by taking the difference of weak (sigma two pixel) and stronger (sigma eight pixel) Gaussian blur functions. Essentially, this functioned as a spatial bandpass filter, attenuating high frequencies caused by pixel-to-pixel shot noise variations and low frequency variations resulting from spread of Ca²⁺ waves across the cell, while retaining spatial frequencies corresponding to the spread of local Ca²⁺ puffs. Next, for each frame, we calculated the spatial variance among pixels within an imaging field, took the square root to obtain a measure of the spatial SD signal and subtracted the predicted increase in spatial shot noise with increasing fluorescence. Finally, we derived traces showing the average spatial fluctuations across a region encompassing the entire cell within a moving boxcar window of 160 ms (Figure 1—figure supplement 3).

The scripts used to perform temporal and spatial fluctuation analysis are presented as Supplementary file 1.

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Dr. Carley A Karsten for assisting in initial imaging experiments and Dr. Kyle L Ellefsen for help with software and image analysis.

© 2020, Lock and Parker

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.