IP3 mediated global Ca2+ signals arise through two temporally and spatially distinct modes of Ca2+ release
- Department of Neurobiology & Behavior, UC Irvine, United States
- Department of Physiology & Biophysics, UC Irvine, United States
Abstract
The ‘building-block’ model of inositol trisphosphate (IP3)-mediated Ca2+ liberation posits that cell-wide cytosolic Ca2+ signals arise through coordinated activation of localized Ca2+ puffs generated by stationary clusters of IP3 receptors (IP3Rs). Here, we revise this hypothesis, applying fluctuation analysis to resolve Ca2+ signals otherwise obscured during large Ca2+ elevations. We find the rising phase of global Ca2+ signals is punctuated by a flurry of puffs, which terminate before the peak by a mechanism involving partial ER Ca2+ depletion. The continuing rise in Ca2+, and persistence of global signals even when puffs are absent, reveal a second mode of spatiotemporally diffuse Ca2+ signaling. Puffs make only small, transient contributions to global Ca2+ signals, which are sustained by diffuse release of Ca2+ through a functionally distinct process. These two modes of IP3-mediated Ca2+ liberation have important implications for downstream signaling, imparting spatial and kinetic specificity to Ca2+-dependent effector functions and Ca2+ transport.
Introduction
Cytosolic Ca2+ signals generated by the liberation of Ca2+ ions sequestered in the endoplasmic reticulum (ER) through inositol trisphosphate receptor (IP3R) channels regulate ubiquitous cellular processes as diverse as gene transcription, secretion, mitochondrial energetics, electrical excitability and fertilization (Clapham, 2007; Berridge et al., 2000). Cells achieve such unique repertories of Ca2+-dependent functions by generating a hierarchy of cytosolic Ca2+ signals with markedly different spatial scales and temporal durations, ranging from brief, localized Ca2+ transients called puffs (Parker and Yao, 1991; Yao et al., 1995) to larger and more prolonged Ca2+ elevations that engulf the cell. Global elevations in cytosolic Ca2+ typically last several seconds and may appear as waves that propagate throughout the cell (Woods et al., 1986). They can recur as oscillations with periods between a few seconds and a few minutes, and are thought to encode information in a ‘digital’ manner, whereby increasing stimulus strength results predominantly in an increase in frequency rather than amplitude (Parekh, 2011; Smedler and Uhlén, 2014). Puffs, on the other hand, are tightly localized elevations in cytosolic Ca2+ generated by stationary clusters containing small numbers of IP3Rs, which last only tens or a few hundreds of milliseconds and remain restricted within a few micrometers (Bootman et al., 1997a; Parker et al., 1996).
The patterning of cellular Ca2+ signals evoked by IP3 is largely determined by the functional properties of the IP3Rs and by their spatial arrangement in the ER membrane. Crucially, the opening of IP3R channels is regulated by cytosolic Ca2+ itself, in addition to IP3. Low concentrations of Ca2+ increase the open probability of the channel whereas high concentrations favor a closed state (Parker and Ivorra, 1990; Iino, 1990; Bezprozvanny et al., 1991). This biphasic modulation of IP3Rs by Ca2+ leads to the phenomenon of Ca2+-induced Ca2+ release (CICR). Ca2+ diffusing from one open channel may thus trigger the opening of adjacent channels, with self-reinforcing CICR countered by Ca2+ feedback inhibition. The clustered distribution of IP3Rs further shapes the extent of this regenerative process. CICR may remain locally restricted to a single cluster, containing from a few to a few tens of functional IP3Rs, to produce a puff; whereas it is proposed that a global response is generated by successive cycles of CICR and Ca2+ diffusion acting over longer spatial ranges to recruit successive puff sites (Bootman et al., 1997a; Parker et al., 1996; Berridge, 1997; Bootman and Berridge, 1996; Callamaras et al., 1998; Dawson et al., 1999; Marchant, 2001; Marchant et al., 1999). However, the transition between these modes remains an area of active investigation (Rückl and Rüdiger, 2016; Miyamoto and Mikoshiba, 2019; Sneyd et al., 2017a; Sneyd et al., 2017b; Rückl et al., 2015), and recent theoretical simulations have questioned whether Ca2+ released through puff activity is alone sufficient to propagate global cytosolic Ca2+ signals (Piegari et al., 2019).
Here, we examined the nature of Ca2+ liberation through IP3Rs during global cellular Ca2+ signals, asking whether this accords with the widely-accepted ‘building block’ model (Bootman et al., 1997a; Parker et al., 1996; Berridge, 1997; Bootman and Berridge, 1996; Marchant, 2001; Marchant et al., 1999; Mataragka and Taylor, 2018) in which global signals are constructed by the summation of coordinated, pulsatile activation of Ca2+ release at puff sites; or whether global signals also involve an additional mode of Ca2+ liberation that is more homogeneous in space and time. Although Ca2+ puffs are often evident during the initial rising phase of global Ca2+ signals (Bootman et al., 1997a; Bootman and Berridge, 1996; Marchant, 2001; Marchant et al., 1999), a challenge in answering this question arises because puffs become obscured as the overall cytosolic Ca2+ level continues to increase. To reveal and monitor temporally rapid and spatially confined Ca2+ transients (puffs) during even large amplitude global Ca2+ elevations we developed image processing and analysis routines to analyze local fluctuations in Ca2+ fluorescence signals (Ellefsen et al., 2019). We applied these routines to Ca2+ recordings obtained both by total internal reflection fluorescence (TIRF) microscopy to resolve signals arising near the plasma membrane, and by lattice light-sheet (LLS) microscopy to acquire optical sections through the cell interior. We find that rapid flurries of Ca2+ puffs accompany the rising phase of global Ca2+ signals evoked by photoreleased IP3 and by agonist stimulation of the IP3 signaling pathway, but these rapidly terminate before the peak of the response through a mechanism regulated by ER Ca2+ store content. The punctate liberation of Ca2+ via transient, localized Ca2+ puffs contributes only a small fraction of the total Ca2+ liberated during global Ca2+ signals, which are instead sustained by diffuse Ca2+ liberation through a functionally distinct mode of release. These two modes of IP3-mediated Ca2+ release will likely selectively activate different populations of effectors; those positioned close to the IP3R clusters that mediate puffs and which respond to brief, repetitive transients of [Ca2+], and others that respond to a more sustained, spatially diffuse elevation of bulk cytosolic [Ca2+].
Results
Fluctuation processing of Ca2+ images highlights transient signals
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
Fluctuation analysis of Ca2+ signals.
(A–D) Records from a single WT HEK cell loaded with Cal520 and stimulated by photorelease of i-IP3 to evoke a global Ca2+ elevation. (A) Panels show ‘raw’ TIRF fluorescence images of the cell before (i), during the rising phase (ii-iv) and at the peak (v) of the global Ca2+ signal. Images are Gaussian-blurred (sigma ~1 µm) single frames (8 ms exposure time) captured at times as marked in C. Grey scale intensities depict fluorescence in arbitrary camera units, as indicated by the bar at the right. The yellow outline marks the TIRF footprint of the cell. (B) Panels show corresponding standard deviation (SD) images at the same times as in A, highlighting hot spots of local, transient Ca2+ release. Grey scale intensities (arbitrary units; A.U.) represent the shot noise-corrected standard deviation of fluorescence fluctuations within a 160 ms running time window. (C) Overlaid black traces show fluorescence monitored from 24 regions of interest (ROIs; marked by squares in panel Ai) placed on areas of local Ca2+ activity. The arrow indicates the time of the photolysis flash. (D) Power spectra of Ca2+ fluorescence fluctuations averaged from the 24 ROIs at baseline (blue trace) and during the rising phase of the global Ca2+ signal (red trace). Spectra were calculated from recordings during the respective times indicated by the colored bars in C, after low-pass (1 Hz) filtering of the fluorescence image stack to strip out the slow rise of the global signal. (E) Overlaid traces show shot noise-corrected SD signals from the 24 ROIs centered on hot spots of Ca2+ activity. The thicker yellow trace shows the mean SD signal monitored from a ROI encompassing the entire cell and is depicted after scaling up by a factor of 10 relative to the traces from small ROIs. (F–I) Corresponding images and plots from an HEK cell devoid of IP3Rs (3KO) in which a global Ca2+ elevation was evoked by pipetting a 10 µl aliquot of ionomycin into the bathing solution at a distance from the cell when marked by the arrow in H. In this case no hot spots or increased low-frequency fluctuations accompanied the elevation in cytosolic [Ca2+], and the ROIs (marked by squares in panel Fi) used to derive the data in H-J were placed randomly. The yellow trace in J depicting the mean SD signal from the entire cell is scaled up by a factor of 10 relative to the traces from small ROIs. Fluorescence and SD magnitudes are expressed in arbitrary units consistent with those in A-D.
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.
Temporal fluctuations reflect spatially localized Ca2+ signals
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.
Fluctuation analysis reveals a transient flurry of puffs during global Ca2+ signals
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
Localized fluctuations in cytosolic [Ca2+] occur predominantly during the rising phase of global Ca2+ elevations.
Representative records show the Cal520 fluorescence ratio (ΔF/F0; smooth traces) and the associated SD fluctuation measurements (noisy traces) from ROIs encompassing single WT HEK cells bathed in Ca2+-free medium. (A) Record obtained under basal conditions without stimulation. (B–E) Responses evoked by progressively longer photolysis flashes to release increasing amounts of i-IP3 in cells loaded with caged i-IP3. The SD signals are presented in arbitrary units (A.U.) but are consistent throughout all panels. To better display responses to weaker stimuli, the y-axes are scaled differently between panels. (F) Responses evoked by application of carbachol (CCH; 10 µM) when indicated by the bar. (G, H) Pooled data plotting, respectively, means of the global Ca2+ fluorescence signals and SD signals of cells stimulated with progressively increasing photorelease of i-IP3 to evoke predominantly local Ca2+ signals (yellow traces; n = 7), and global elevations with slow (blue; n = 9), medium (green; n = 13), and fast rising Ca2+ signals (pink; n = 11). (I) Mean Cal520 fluorescence ratio signal (cyan trace) and SD signal (black trace) averaged from 12 cells stimulated by local application of 10 µM CCH when marked by the bar.
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.
Ca2+ signals evoked by agonist activation and photoreleased i-IP3 show similar patterns of puff activity
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+.
Ca2+ puff activity terminates during the rising phase of global Ca2+ signals
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
Relationship between Ca2+ fluctuations and Ca2+ level during the rise of global Ca2+ signals.
Scatter plots show measurements of the SD signal at intervals during the rising phase of global Ca2+ response against the magnitude of the global Ca2+ elevation (ΔF/F0) at that time. Data were binned at intervals of (0.1 ΔF/F0). (A) Measurements from the same cell as in Figure 2D. (B) Data from the same groups of cells as in Figure 2G,H, plotting mean SD signal amplitude as a function of mean Ca2+ level during global responses for cells exhibiting slow (blue squares), intermediate (green triangles) and fast rising responses (pink circles). (C) The same data as in B, after normalizing to the respective maximum SD signals for each group of cells. (D) Scatter plot of mean SD signal amplitude as a function of Ca2+ level during global responses for 12 cells stimulated by local application of CCH, as in Figure 2I.
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).
Ca2+ puffs are independent of extracellular Ca2+
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 (fall80-20 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+; Fall80-20 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.
Patterns of Ca2+ release are largely unaffected by inhibition of mitochondrial and lysosomal Ca2+ uptake
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).
Ca2+ puffs do not terminate because of rising cytosolic Ca2+ during cell-wide elevations
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
The suppression of Ca2+ puffs during global signals does not result because of elevated cytosolic [Ca2+].
(A,B) IP3-evoked Ca2+ puffs are not suppressed by prior photorelease of Ca2+. (A) Traces depict fluorescence ratios (black; ΔF/F0) from WT HEK cells and corresponding SD signals (grey; in arbitrary units, A.U.). Records, from left to right, show responses from individual cells loaded with NP-EGTA (caged Ca2+) that were unstimulated or exposed to increasing UV flash durations (marked by arrows) to photorelease progressively increasing amounts of free Ca2+ before challenging cells with CCH (100 µM) locally delivered by a puffer pipette when indicated by the bars. Traces are blanked out during the artifact caused by the photolysis flash. (B) Data points from traces like those in A show the integral under SD trace (a measure of puff activity) as a ratio of the change in global Ca2+ signal (ΔF/F0) evoked by CCH. The data are binned in terms of the jump in Cal520 fluorescence (ΔF/F0) evoked by photolysis of caged Ca2+. Open symbols are from individual cells, and filled symbols are means for each group (respective n numbers for different bins; 20, 4, 4, 6, 8). Data are normalized with respect to the mean ratio without prior photorelease of Ca2+. There was no significant difference between control CCH responses and CCH responses following Ca2+ jumps (evaluated by Student T-test; p values between 0.17 and 0.66 for the different binned groupings). (C,D) Termination of puff activity is unaffected when global cytosolic Ca2+ signals are attenuated by buffering with EGTA. (C) Traces showing the Cal520 fluorescence ratio (ΔF/F0; smooth trace) and SD signal (noisy trace) in response to photoreleased i-IP3 in a representative WT HEK cell that was incubated with 15 µM EGTA/AM to buffer cytosolic Ca2+ and attenuate the amplitude of the global Ca2+ signal. (D) Scatter plots show measurements of the SD signal at intervals during the rising phase of global Ca2+ responses against the magnitude of the global Ca2+ elevation (ΔF/F0) at that time. Measurements were binned at intervals of (0.1 ΔF/F0) and SD data are normalized to a peak value of 1. Solid circles show mean data from 14 EGTA-loaded cells. For comparison, open circles present data reproduced from Figure 3C showing measurements from 11 control cells that gave fast rising responses to photoreleased i-IP3.
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.
Partial depletion of ER Ca2+ selectively inhibits Ca2+ puffs
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
Ca2+ puffs are selectively depressed by reduced ER Ca2+ content.
(A–F) Selective depression of puffs during i-IP3-evoked global Ca2+ signals following depletion of ER Ca2+ content using transient application of cyclopiazonic acid (CPA; 50 µM) (A) The smooth trace shows fluorescence ratio (ΔF/F0) from a WT HEK cell, and the noisy trace the corresponding SD signal (in arbitrary units). The cell was bathed throughout in solution containing no added Ca2+ and 300 µM EGTA, and CPA was locally applied from a puffer pipette during the time indicated by the bar. A UV flash was delivered when marked by the arrow to photorelease caged i-IP3 loaded into the cell. (B) Mean ΔF/F0 and SD signals from seven WT HEK cells in response to photoreleased i-IP3 following CPA treatment and wash in Ca2+-free medium as in A. (C,D) Representative ΔF/F0 and SD responses to photoreleased i-IP3 in individual cells that were bathed, respectively, in Ca2+-containing or Ca2+-free medium for 30 min following treatment with CPA as in A. (E) Peak amplitudes of global fluorescence signals evoked by photoreleased i-IP3 in experiments like those in C,D, for cells bathed in Ca2+-containing (n = 8 cells; blue squares) or Ca2+-free medium (n = 6; red circles). Open symbols denote measurements from individual cells; filled symbols are means. No significant difference between peak amplitudes (ΔF/F0) of cells bathed in Ca2+-containing and Ca2+-free medium (Student T test; p=0.72). (F) Corresponding measurements of integral under SD traces (puff activity) during the time from the photolysis flash to the peak global fluorescence signal. SD integrals were significantly different between cells bathed in Ca2+-containing and Ca2+-free medium (Student T test; p=0.012). (G–I) Selective depression of puffs by depleting ER Ca2+ content by repeated applications of CCH in zero Ca2+ bathing solution. (G,H) Global Ca2+ signals (smooth traces; ΔF/F0) and SD signals (noisy traces) evoked by successive, identical applications of CCH at 5 min intervals in two representative cells bathed, respectively, in medium containing 2 mM Ca2+ or 300 µM EGTA with no added Ca2+. Amplitudes of the SD signals are depicted after normalizing to the peak amplitude of the first response for each cell. (I) Data points show the ratio of puff activity (integral under the SD trace) vs. peak magnitude of the global Ca2+ signal (ΔF/F0) for successive responses evoked by CCH application at 5 min intervals. Blue squares are data from cells bathed in medium containing 2 mM Ca2+ and red circles are from cells in Ca2+-free medium; open symbols are ratios from individual cells and filled symbols are means. Data are plotted after normalizing to the mean SD integral and peak ΔF/F0 evoked by the initial stimulus in each condition. Responses were significantly different between cells bathed in the presence and absence of external Ca2+ for times ≥ 10 min (Student T test; p=0.000008).
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.
All three IP3R isoforms mediate punctate and diffuse modes 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
Cell-wide Ca2+ elevations and SD fluctuations in WT HEK cells and cells exclusively expressing single IP3R isoforms.
(A–H) Traces show whole-cell Cal520 fluorescence ratio (smooth colored traces; ΔF/F0) and SD fluctuations (noisy black traces) of HEK cells locally stimulated with CCH locally delivered in a Ca2+-containing bath solution when indicated by the bars. Panels on the left are representative records from individual cells, and panels on the right show mean traces from 7 (B) or 3 (D,F,H) cells. (A, B) Records from HEK WT cells. (C–H) Records from HEK cells solely expressing IP3R1 (C,D), IP3R2 (E,F), or IP3R3 (G,H). (I) Overlaid mean Cal520 fluorescence ratio traces, aligned to their rising phase, in WT HEK cells (cyan; n = 7), and HEK cells solely expressing IP3R1 (gold; n = 3), IP3R2 (green; n = 3), and IP3R3 (pink; n = 3). (J) Scatter plots of SD signal vs. fluorescence ratio during the rising phase of the Ca2+ responses in WT HEK cells (cyan diamonds) and HEK cells solely expressing IP3R1 (gold circles) IP3R2 (green squares) or IP3R3 (pink triangles). Data points are means from the same cells as in I. (K) The same data as in J, after normalizing to the same peak SD values.
HeLa and HEK cells exhibit similar patterns of Ca2+ signals
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).
Diffuse Ca2+ signals in TIRF do not reflect punctate release in the cell interior
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
Lightsheet imaging of global Ca2+ elevations evoked in HEK cells by photoreleased i-IP3.
(A) Upper panels show 45o diagonal image ‘slices’ through the center of a WT HEK cell imaged by lattice light-sheet microscopy. Grey scale intensities correspond to increases in fluorescence (ΔF) of Cal520 relative to the mean intensity (F0) averaged over 100 frames before stimulation (ΔF/F0). Each panel is a single 10 ms exposure, captured at times before and after stimulation, as indicated by the Roman numerals in C. The cell outline is marked in yellow. Lower panels show corresponding SD images, at times corresponding to the upper panels. Colored outlines mark ROIs used to derive ΔF/F0 and SD traces from peripheral (red) and center (blue) regions of the cell. (B) Corresponding ΔF/F0 and SD lightsheet images from a different HEK cell that showed more prominent puff activity in the center of the cell. (C) Measurements of ΔF/F0 (smooth traces) and SD (noisy traces) from the cell illustrated in A. Traces in red show average measurements from the peripheral region of interest marked in the bottom left panel of A, and traces in blue show measurements from the central region of interest. (D) Corresponding measurements of ΔF/F0 and SD from the cell illustrated in B. (E) Scatter plot of SD signal versus Ca2+ fluorescence (ΔF/F0) at intervals during the rising phase of global Ca2+ signals. Data are from eight cells, with measurements binned at intervals of 0.1 ΔF/F0.
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.
Puff activity contributes only a fraction of the total Ca2+ liberated during global signals
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
Relative proportions of Ca2+ released by punctate versus diffuse modes of Ca2+ liberation during an IP3-evoked global Ca2+ signal.
(A,B) Whole cell Ca2+ fluorescence responses (black traces) and associated SD signals (grey traces) during the initial phase of a Ca2+ response evoked by photoreleased i-IP3 in representative WT HEK cells. The red traces show the estimated rate of Ca2+ efflux, derived as described in the text. Both panels show responses from cells pretreated with CPA as in Figure 5A that were treated identically, except that the cell in A was incubated in Ca2+-containing medium to allow refilling of ER Ca2+, whereas the cell in B was incubated in Ca2+-free medium to maintain the ER Ca2+ in a partially depleted state and suppress puff activity. The SD and Ca2+ efflux traces in A are scaled to similar peak height for clarity; traces in B are scaled the same as in A. (C) Mean Ca2+ efflux traces from five cells in Ca2+-containing medium that showed robust puff activity (top) and six cells in Ca2+-free medium where puff activity was almost absent (lower). The area shaded blue reflects the relative amount of Ca2+ released when puff activity was absent, and the pink area reflects the additional amount of Ca2+ release attributable to Ca2+ puffs. (D,E) Traces show SD signal (grey), global Ca2+ fluorescence ratio (black) and calculated Ca2+ efflux rate (red) for the entire duration of responses evoked by photoreleased i-IP3 (D) and by CCH (E). Blue traces additionally show the cumulative percentage of Ca2+ released, derived by integrating under the red Ca2+ efflux traces. For clarity of presentation all traces are shown scaled to the same peak height. (F) Bars show mean percentages of total Ca2+ release during i-IP3-evoked (left; n = 8 cells) and CCH-evoked signals (right; n = 7 cells) under control conditions attributable to punctate (pink) and diffuse (blue) modes of Ca2+ liberation. Data were calculated from the cumulative Ca2+ release at the time puff activity had ceased in each cell, assuming 41% of that release was due to punctate release.
Discussion
Ca2+ puffs are transient, localized elevations in cytosolic Ca2+ that arise from concerted opening of small numbers of IP3Rs clustered at fixed intracellular sites (Parker and Yao, 1991; Thillaiappan et al., 2017). Puffs are apparent as discrete events superimposed on a steady basal Ca2+ level when cytosolic IP3 concentrations are modestly elevated (Parker and Yao, 1991; Yao et al., 1995; Parker et al., 1996), whereas higher concentrations of IP3 evoke global, cell-wide Ca2+ signals on which puffs are evident on the rising phase (Bootman et al., 1997a; Marchant et al., 1999). Puffs have been proposed as fundamental building blocks of IP3-mediated Ca2+ signaling (Bootman et al., 1997a; Parker et al., 1996; Berridge, 1997; Bootman and Berridge, 1996; Marchant, 2001; Marchant et al., 1999; Mataragka and Taylor, 2018); acting as local signals in their own right at low [IP3] and mediating global Ca2+ signals at higher [IP3] by a fire-diffuse-fire mechanism whereby Ca2+ released by a puff site diffuses to activate CICR at neighboring sites (Bootman et al., 1997a; Parker et al., 1996; Berridge, 1997; Dawson et al., 1999). However, it has been difficult to definitively test this ‘building block’ model because puffs become obscured by the large global Ca2+ elevations; and recent theoretical simulations have questioned whether the summation of Ca2+ released through coordinated puff activity at multiple sites is alone sufficient to propagate global cytosolic Ca2+ signals (Piegari et al., 2019). Here, we addressed this topic by analyzing temporal and spatial fluctuations in Ca2+ image data to resolve local Ca2+ transients during global signals (Swaminathan et al., 2020). Our main conclusion is that global Ca2+ signals involve two modes of Ca2+ liberation through IP3Rs: ‘punctate’ release as a flurry of transient, local events, and a more sustained, ‘diffuse’ release mode.
As with any new approach, we first needed to validate the ability of our algorithm to faithfully report local Ca2+ transients during even large global Ca2+ elevations, when resolution may be impaired by factors including the dynamic range of the indicator dye and by increased photon shot noise at high fluorescence levels. A particular concern was whether the indicator (Cal520) we used for most experiments may have approached saturation, thus ‘clipping’ the signals to artifactually suppress the temporal SD signal and giving a false impression that puff activity terminates as the Ca2+ level and fluorescence rise during global signals. Several lines of evidence convincingly argue that this is not the case. Notably: (i) maximal, saturating signals evoked by ionomycin (~19 ΔF/F0) were much higher than mean peak IP3-evoked fluorescence signals (~7 ΔF/F0), and puff activity began to decline as fluorescence rose above ~2 ΔF/F0; (ii) we observed patterns of puff activity using the low affinity indicator fluo-8L (Kd1.86 µM) that closely matched those obtained with Cal520 (Kd320 nM) (Figure 3—figure supplement 2); (iii) we were able to resolve instances of local puff activity even at the peak of IP3-evoked global fluorescence elevations (Figure 3—figure supplement 1); (iv) the kinetics of puff activity were closely similar in cell lines individually expressing single IP3R isoforms, despite large differences in the amplitudes of the global Ca2+ signals (Figure 6); (v) the onset and termination of puff activity during the rise of IP3-mediated global Ca2+ signals were little altered when the initial basal Ca2+ level was elevated (Figure 4A,B) or, conversely, when the global Ca2+ rise was attenuated by buffering with cytosolic EGTA (Figure 4C,D). Finally, the suppression of punctate Ca2+ liberation throughout all phases of IP3-evoked Ca2+ responses when ER Ca2+ stores were partially depleted (Figure 5) strongly supports our proposal that Ca2+ liberation can arise in a diffuse manner, independent of local puff events.
Puff activity during global Ca2+ signals
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).
Two functionally distinct modes of IP3R-mediated Ca2+ liberation
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).
Consequences of bimodal Ca2+ signals for downstream signaling
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?
Materials and methods
Cell culture and loading
Request a detailed protocolHEK293 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.
Ca2+ imaging
Request a detailed protocolTotal 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).
Photo-uncaging and local application of agonist
Request a detailed protocolPhotorelease 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 analysis
Request a detailed protocolImage 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.
Data availability
Algorithms used to generate SD fluctuation images from Ca2+ image recordings are provided in the Supplementary file 1.
References
-
Elementary and global aspects of calcium signallingThe Journal of Physiology 499:291–306.https://doi.org/10.1113/jphysiol.1997.sp021927
-
The versatility and universality of calcium signallingNature Reviews Molecular Cell Biology 1:11–21.https://doi.org/10.1038/35036035
-
Calcium signalling: dynamics, homeostasis and remodellingNature Reviews Molecular Cell Biology 4:517–529.https://doi.org/10.1038/nrm1155
-
Imaging the hierarchical Ca2+ signalling system in HeLa cellsThe Journal of Physiology 499:307–314.https://doi.org/10.1113/jphysiol.1997.sp021928
-
Spatiotemporal patterning of IP3-mediated Ca2+ signals in Xenopus oocytes by Ca2+-binding proteinsThe Journal of Physiology 556:447–461.https://doi.org/10.1113/jphysiol.2003.059204
-
Buffer kinetics shape the spatiotemporal patterns of IP3-evoked Ca2+ signalsThe Journal of Physiology 553:775–788.https://doi.org/10.1113/jphysiol.2003.054247
-
Calcium oscillationsCold Spring Harbor Perspectives in Biology 3:a004226.https://doi.org/10.1101/cshperspect.a004226
-
Applications of FLIKA, a Python-based image processing and analysis platform, for studying local events of cellular calcium signalingBiochimica Et Biophysica Acta (BBA) - Molecular Cell Research 1866:1171–1179.https://doi.org/10.1016/j.bbamcr.2018.11.012
-
Translational mobility of the type 3 inositol 1,4,5-trisphosphate receptor Ca2+ release channel in endoplasmic reticulum membraneJournal of Biological Chemistry 280:3824–3831.https://doi.org/10.1074/jbc.M409462200
-
Lateral diffusion of inositol 1,4,5-trisphosphate receptor type 1 is regulated by actin filaments and 4.1N in neuronal dendritesJournal of Biological Chemistry 279:48976–48982.https://doi.org/10.1074/jbc.M408364200
-
Structural and functional link between the mitochondrial network and the endoplasmic reticulumThe International Journal of Biochemistry & Cell Biology 41:1817–1827.https://doi.org/10.1016/j.biocel.2009.04.010
-
Biphasic Ca2+ dependence of inositol 1,4,5-trisphosphate-induced ca release in smooth muscle cells of the guinea pig Taenia caeciThe Journal of General Physiology 95:1103–1122.https://doi.org/10.1085/jgp.95.6.1103
-
Endogenous signalling pathways and caged IP3 evoke Ca2+ puffs at the same abundant immobile intracellular sitesJournal of Cell Science 130:3728–3739.https://doi.org/10.1242/jcs.208520
-
Spatial organization of intracellular Ca2+ signalsSeminars in Cell & Developmental Biology 23:172–180.https://doi.org/10.1016/j.semcdb.2011.09.006
-
Spatial-temporal patterning of Ca2+ signals by the subcellular distribution of IP3 and IP3 receptorsSeminars in Cell & Developmental Biology 94:3–10.https://doi.org/10.1016/j.semcdb.2019.01.012
-
Mitochondrial calcium uptake in organ physiology: from molecular mechanism to animal modelsPflügers Archiv - European Journal of Physiology 470:1165–1179.https://doi.org/10.1007/s00424-018-2123-2
-
Initiation of IP(3)-mediated ca(2+) waves in Xenopus oocytesThe EMBO Journal 18:5285–5299.https://doi.org/10.1093/emboj/18.19.5285
-
All three IP3 receptor subtypes generate Ca2+ puffs, the universal building blocks of IP3-evoked Ca2+ signalsJournal of Cell Science 131:jcs220848.https://doi.org/10.1242/jcs.220848
-
Molecular mechanisms of endolysosomal Ca2+ signalling in health and diseaseBiochemical Journal 439:349–378.https://doi.org/10.1042/BJ20110949
-
Decoding cytosolic Ca2+ oscillationsTrends in Biochemical Sciences 36:78–87.https://doi.org/10.1016/j.tibs.2010.07.013
-
Inositol 1,4,5-trisphosphate receptors and their protein partners as signalling hubsThe Journal of Physiology 594:2849–2866.https://doi.org/10.1113/JP271139
-
Mitochondria as sensors and regulators of calcium signallingNature Reviews Molecular Cell Biology 13:566–578.https://doi.org/10.1038/nrm3412
-
Modulation of elementary calcium release mediates a transition from puffs to waves in an IP3R cluster modelPLOS Computational Biology 11:e1003965.https://doi.org/10.1371/journal.pcbi.1003965
-
Calcium waves in a grid of clustered channels with synchronous IP3 binding and unbindingThe European Physical Journal E 39:108.https://doi.org/10.1140/epje/i2016-16108-4
-
Spatial Ca2+ profiling: decrypting the universal cytosolic Ca2+ oscillationThe Journal of Physiology 595:3053–3062.https://doi.org/10.1113/JP272860
-
Cytosolic Ca2+ buffersCold Spring Harbor Perspectives in Biology 2:a004051.https://doi.org/10.1101/cshperspect.a004051
-
Frequency decoding of calcium oscillationsBiochimica Et Biophysica Acta (BBA) - General Subjects 1840:964–969.https://doi.org/10.1016/j.bbagen.2013.11.015
-
Modeling calcium waves in an anatomically accurate three-dimensional parotid acinar cellJournal of Theoretical Biology 419:383–393.https://doi.org/10.1016/j.jtbi.2016.04.030
-
Life and death of a cardiac calcium sparkThe Journal of General Physiology 142:257–274.https://doi.org/10.1085/jgp.201311034
-
Glutamate-induced neuron death requires mitochondrial calcium uptakeNature Neuroscience 1:366–373.https://doi.org/10.1038/1577
-
Noise analysis of cytosolic calcium image dataCell Calcium 86:102152.https://doi.org/10.1016/j.ceca.2019.102152
-
Cluster formation of inositol 1,4,5-Trisphosphate receptor requires its transition to open stateJournal of Biological Chemistry 280:6816–6822.https://doi.org/10.1074/jbc.M405469200
-
IP3 receptors and store-operated Ca2+ entry: a license to fillCurrent Opinion in Cell Biology 57:1–7.https://doi.org/10.1016/j.ceb.2018.10.001
-
IP3 receptors and Ca2+ entryBiochimica Et Biophysica Acta (BBA) - Molecular Cell Research 1866:1092–1100.https://doi.org/10.1016/j.bbamcr.2018.11.007
-
Spatial and temporal aspects of cellular calcium signalingThe FASEB Journal 10:1505–1517.https://doi.org/10.1096/fasebj.10.13.8940296
-
Effects of cyclopiazonic acid, a novel ca(2+)-ATPase inhibitor, on contractile responses in skinned ileal smooth muscleBritish Journal of Pharmacology 106:208–214.https://doi.org/10.1111/j.1476-5381.1992.tb14316.x
-
Cytosolic [Ca2+] regulation of InsP3-evoked puffsBiochemical Journal 449:167–173.https://doi.org/10.1042/BJ20121271
-
Quantal puffs of intracellular Ca2+ evoked by inositol trisphosphate in Xenopus oocytesThe Journal of Physiology 482:533–553.https://doi.org/10.1113/jphysiol.1995.sp020538
Decision letter
-
Kenton J SwartzSenior and Reviewing Editor; National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
-
Colin W TaylorReviewer; University of Cambridge, United Kingdom
-
Grant HennigReviewer; Univ. Vermont, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Acceptance summary:
Parker's pioneering work established that the initial responses to low concentrations of IP3 comprise Ca2+ puffs, wherein a few IP3Rs within an immobile cluster almost simultaneously open to evoke a local and short-lived cytosolic Ca2+ signal. It has been widely assumed that Ca2+ released by these events recruits further Ca2+ puffs (by Ca2+-induced Ca2+ release) to give global Ca2+ signals. It has, however, been difficult to examine this hypothesis because as Ca2+ puffs become more frequent, it becomes impossible to resolve them from background signal. The present work applies an elegant noise analysis method (the principles of which were described in a recent short report from the group in Cell Calcium) to assess the subcellular heterogeneity of Ca2+ signals evoked by IP3. The results are both unexpected and important in that they suggest that most cytosolic Ca2+ signals are not associated with the noise expected from Ca2+ puffs. The results convincingly challenge prevailing dogma by suggesting that Ca2+ puffs contribute only a small fraction the cytosolic Ca2+ signals evoked by IP3, with the remainder likely arising from openings of single IP3Rs. The authors have done an excellent job of revising the manuscript to address the technical concerns of the reviewers, and overall this is a rigorous, well done and significant body of work that justifies publication in eLife.
Decision letter after peer review:
[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]
Thank you for submitting your work entitled "Inositol trisphosphate mediated Ca2+ spikes arise through two temporally and spatially distinct modes of Ca2+ release" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Colin W Taylor (Reviewer #1); Grant Hennig (Reviewer #3).
Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
The reviewers agreed that your work is interesting and has the potential to be quite important, but there were major concerns with the SD approach, on which most of the presented findings hinge, involving the dynamic range of the Ca2+ indicators and the spatial and temporal Ca2+ release patterns.
The first concern is that after uncaging or after the addition of agonists, the cytoplasmic background fluorescence in cells increases to a seemingly very high level. As punctate release events are only observed when background levels are low to moderate, it brings into question whether the Ca2+ indicators have become saturated at peak cytoplasmic background Ca2+ levels (the author's have previously assessed a variety of Ca2+ indicators with Kd values in the 150-400nM range). This would make it essentially impossible to resolve punctate events, even though punctate release events may still be occurring – akin to "signal clipping". While SD analysis may provide some additional resolution in situations with narrow dynamic fluorescence ranges, it still depends on the indicator having some dynamic range. This concern needs to be carefully addressed so that readers are confident that the presented findings are accurate and biological.
Some ways that could be used to examine the potential dynamic range issue of the indicators include:
• use a ratiometric indicator (Fura) to estimate actual cytoplasmic Ca2+ levels during uncaging/agonists which could be compared to Kd/saturation mM of Cal-520 and GCamP etc.
• pharmacologically limit the increase in cytoplasmic Ca2+ using low/moderate levels of EGTA (to preserve the dynamic range of the indicator) and examine whether punctate release patterns are still reduced after uncaging etc.
• purposefully increase cytoplasmic Ca2+ levels to near maximum with ionomycin then initiate an uncaging event to see puncate release sites can still be resolved (temporally synchronized release).
• use quantitative shape analysis (not ROIs) of transient bright regions at high cytoplasmic Ca2+ concentrations. At these levels where there is extremely limited dynamic range and greater shot noise, noise aggregates and punctate release sites will appear similar – but they can be separated using perimeter:area ratios (high for noise aggregates), or Gaussian fitting (better for release sites). This will require no Gaussian filtering of the data and some form of deconvolution to reduce the smoothing of objects due to Z-smear (where appropriate).
The second concern is discussed adequately in the reviews below, but questions whether the SD approach can detect synchronous, widespread puffs/ release events, or high frequency puffs/release events that appear and behave like changes in cytoplasmic background. It would be useful to the reader to see how different duration timespans affect the variance signal (which is currently tailored for the timecourse of individual puffs), to assess what range of Ca2+ behaviors are included or filtered by varying the space/time variance parameters.
Without additional information to address the major concerns of the reviewers, it is not possible to assess whether the work represents a sufficient advance to warrant publication in eLife.
Reviewer #1:
Parker's pioneering work established that the initial responses to low concentrations of IP3 comprise Ca2+ puffs, wherein a few IP3Rs within an immobile cluster almost simultaneously open to evoke a local and short-lived cytosolic Ca2+ signal. It has been widely assumed that Ca2+ released by these events recruits further Ca2+ puffs (by Ca2+-induced Ca2+ release) to give global Ca2+ signals. It has, however, been difficult to examine this hypothesis because as Ca2+ puffs become more frequent, it becomes impossible to resolve them from background signal. The present work applies an elegant noise analysis method (the principles of which were described in a recent short report from the group in Cell Calcium) to assess the subcellular heterogeneity of Ca2+ signals evoked by IP3. The results are both unexpected and important in that they suggest that most cytosolic Ca2+ signals are not associated with the noise expected from Ca2+ puffs. The results convincingly challenge prevailing dogma by suggesting that Ca2+ puffs contribute only a small fraction the cytosolic Ca2+ signals evoked by IP3, with the remainder likely arising from openings of single IP3Rs. The rigour and significance of the work justify publication in eLife; none of my comments should detract from that recommendation.
1) Figure 1 provides compelling evidence that noise decreases as the IP3-evoked Ca2+ signals develop and that a comparable increase in global cytosolic [Ca2+] evoked by ionomycin is “noise-free”, as expected for homogenous Ca2+ release. My understanding is that after some filtering and spatial blurring, most analyses calculate the SD for each pixel as a 160-ms (ie 20 frames) boxcar average. The original view posits that a global Ca2+ signal reflects most Ca2+ puff sites firing very frequently (perhaps 20 or so sites active at each instant), each spreading a µm or more from its source. Might this cause elevations in cytosolic [Ca2+] that are less noisy than the infrequent Ca2+ puffs observed initially, because each pixel would be influenced by Ca2+ reaching it from several puffs? It's perhaps too tall an order to require quantitative simulation of this on SD measurements, but it would be helpful to have the authors consider the issue explicitly.
2) From results in Figure 5, the authors propose that loss of Ca2+ from the ER causes puffs to terminate without appreciably abrogating the global increase in cytosolic [Ca2+]. That proposal would align with a long-standing, but contentious, suggestion that luminal Ca2+ supports IP3R activity, but it seems difficult to align with other work from this lab:
A paper from this lab and under review with eLife (Mak et al) proposes that luminal Ca2+ inhibits IP3R activity, such that loss of ER Ca2+ would be expected to enhance Ca2+ puffs.
Substantial evidence, much of it form this lab, indicates that individual Ca2+ puffs do not detectably affect ER luminal [Ca2+], since steps during the falling phase of a Ca2+ puff (reflecting channel closures) are of fixed amplitude. It is not thereby clear that the flurry of puff activity would cause the loss of ER Ca2+ proposed to terminate Ca2+ puffs.
The authors should discuss both issues.
Reviewer #2:
In this manuscript, Lock and Parker expand the ongoing work in the Parker lab to examine the temporal and spatial properties and their regulation of Ca2+ release by the inositol trisphosphate receptor (IP3R) Ca2+ release channel. They use HEK and HeLa cells to examine by TIRF microscopy (the primary approach here, and previously used extensively by this group) and light sheet microscopy, and a heavy dose of image processing abd fluctuation aanlyses, to address the issue of the sources of Ca2+ release and their mechanisms of transitioning from discrete Ca2+ signals, referred to as puffs, to global ones.
They find that IP3-mediated Ca2+ release is associated with a flurry of noisy events during the rising phase of the global Ca2+ signal that terminates before the peak of the global Ca2+ signal is achieved. They suggest that the noisy events are Ca2+-release events from IP3R clusters, i.e. Ca2+ puffs, on a background of IP3R-mediated more diffuse release. Using procedures to partially deplete Ca2+ stores, they conclude also that the Ca2+ puffs terminate because of depletion of the ER stores, although they cannot determine whether it represents local or global store depletion, whereas the diffuse release, from what they consider to be a separate population of IP3R from those in the Ca2+-puff inducing clusters, continue to release Ca2+. By measuring the rate of cytoplasmic Ca2+ concentration relaxation following a pulse of cytoplasmic Ca2+, they calculate the relative contributions of Ca2+ puffs (punctate release) and diffuse release during the rising phase and during the entire Ca2+ transient.
This is an elegant analysis and an interesting set of studies, and an not least of all because they address and challenge a major model of IP3-mediated local and global Ca2+ signaling (mostly developed by Parker); and because they address a conundrum in the field regarding the role and existence of "silent" IP3Rs that are localized throughout the cell but were previously (Parker; Taylor) considered to not participate. Here, the suggestion is that in addition to active IP3R in clusters near the plasma-membrane, these other IP3R channels might indeed be active and contribute to the diffuse release that continues after puffs have stopped firing.
I have a few comments that might suggest that the authors could clarify a few points.
1) Why aren't local release events observed in the raw ratio images, as Parker has described many times?
2) Figure 3—figure supplement 1 shows the sites with high standard deviation (SD) to be rather large, ~10 uM, a size that is inconsistent with the size of clusters of ~15 IP3R that generate the Ca2+ puffs. Rather their size suggests to me that these are entire regions of the endoplasmic reticulum (ER) that are near the plasma membrane in the TIRF evanescent field. If this is the case, are the authors selecting small sites within the large "contact" sites for the SD analyses? And, I would think that there is a "global"/homogenous signal at these sites that is contributed by Ca2+ that has diffused into these "contact" sites. This might also be consistent with the observation that the TIRF-detected high-SD signal terminates before the peak signal is achieved, because the peak is contributed by slow Ca2+ diffusion that persists after release has terminated.
3) The authors describe an inverted U relationship between the amplitude of the SD signal and the amplitude of the global response. What does it mean?
4) The authors conclude that elevated cytoplasmic Ca2+ doesn't inhibit the IP3R as a primary mechanism to terminate Ca puffs. Why doesn't it inhibit?
5) The authors examine the decay in cytoplasmic Ca2+ concentration after an elevation. They refer to this as sequestration. I'm not sure of their meaning. Although they suggest that the kinetics can be described by a single exponential, there must be at least two processes involved: sequestration by the ER, and extrusion by the plasm membrane Ca2+ pump. Indeed, in the discussion they use the phrase "efflux rate from the cytoplasm". I think what the authors mean to suggest in the sequestration away from the cytoplasm. If so, this should be stated more clearly.
Reviewer #3:
In this manuscript, the authors present a method that enables some resolution and measurement of small, punctate Ca2+ events in the context of global cytoplasmic Ca2+ levels. While previous studies used pharmacological tools to largely prevent global changes in Ca2+ in order to better reveal small discrete Ca2+ release events, the current study uses a variety of software routines that utilize fluorescence variance changes – from which punctate and global Ca2+ event behaviors are derived from. The authors surmise that punctate and global Ca2+ events are likely produced by 2 separate populations of IP3Rs. The experimental manipulations to investigate depletion of stores, Ca2+ entry, IP3R channel KO etc are well thought out and appropriate – but some concerns remain as to the ability of the authors to resolve puffs once background/cytoplasmic Ca2+ levels reach a certain intensity.
The use of signal variance to isolate different components of mixed fluorescence signals relies on those components having distinct (and variable) spatio-temporal signatures. However, if Ca2+ release events reach a steady state, then the variance technique becomes essentially useless. This is my main concern with some of the conclusions drawn from the analysis.
The ability to detect puffs is greatest during low background conditions, but the as global Ca2+ levels increase, the ability to detect puffs is hampered by i) closing of the max difference between background and puff sites and ii) an increase in the amplitude of shot noise that may start to mimick fast and small puff events. To resolve between genuine puff events, that are inherently Gaussian (1-2 µm) and random noise aggregates that occur during high background levels, is difficult – but is made worse by applying Gaussian filters to the data that reshape noise aggregates into more Gaussian shapes. Employing shape analysis before smoothing (perimeter:area ratios) is an effective way of resolving puffs (low P:A – smooth circles) even when surrounded by noise aggregates of equal intensity and size (high P:A – leafy looking). This may assuage some doubts as to the authors ability to resolve punctate events once background/global levels have reached moderate to high levels.
Similarly, the use of user-placed (or randomly positioned) ROIs based on visual inspection of bright spots in the videos is subjective and can heavily bias results to what the user wants to show. Many "hot spots" don't appear to have ROIs drawn over them and some ROIs appear to be sampling the same site. Small ROIs are more sensitive to spatial shifts in the position of the underlying fluorescent sites. The "flickering" variance may be a function of out-of-sync release events from a number of channels/clusters covered by each ROI. When these channels syncronize (in the open position?) or fluorescence sites increase in size, this will dramatically reduce the signal variance, even though puffs may still be occurring.
The shape of the cell and the distance and size of the cytoplasm located away from the evanescent field can dramatically affect the blurring of the background (cytoplasmic signal). How do the authors discriminate between "global" cytoplasmic changes and large puff sites that are heavily blurred being out of focus (see Figures 1 and 7). This effect still remains using lightsheet imaging.
The pixel resolution used during TIRF imaging was 0.5µm and is unlikely to fully resolve individual puffs, especially after smoothing. This may bias analysis to release events that are comprised of multiple channel clusters.
The use of black level subtraction can normalize absolute intensity levels, but does not compensate for the higher shot noise in areas with high static background. Was this taken into account?
Gaussian blur is reported as 2 pixels (in text) but as an SD of 2 pixels in the Materials and methods. What was the actual kernel size in pixels (3x3, 5x5?) and the SD.
Random ROIs were used for HEK cells devoid of IP3Rs. What does the data look like if the authors select the brightest spots in the 3KO video – or use random ROIs in controls?
Figure 3—figure supplement 3C appears to show a stochastic build up of high frequency events correlated to baseline increases – until background fluorescence levels may start to swamp out the ability to resolve small punctate events.
[Editors’ note: further revisions were suggested prior to acceptance, as described below.]
Thank you for choosing to send your work entitled "Inositol trisphosphate mediated Ca2+ spikes arise through two temporally and spatially distinct modes of Ca2+ release" for consideration at eLife. Your letter of appeal has been considered by a Senior Editor and the reviewers, and we are prepared to consider a revised submission.
The reviewers have also provided additional feedback below after reading your rebuttal to hopefully better convey the technical concerns.
Specific points:
1) Indicator Saturation/Clipping concerns:
In your rebuttal, you state (i) that "…maximal signal evoked by ionomycin is>2 fold higher than peak IP3-evoked fluorescence signals"
However, in Figure 1, the maximum raw signal during uncaging (1C) and after ionomycin (1F) looks exactly the same – both topping out at 8500 a.u – and suggests that IP3-stimulated Ca2+ release may be reaching levels that saturate the dye. Similarly, the large areas that show uniform white intensity (Figure 1, image v) are either due to: i) over zealous contrast/brightness adjustment, ii) indicator saturation @ max or iii) camera chip saturation (clipping). Assuming reasonable contrast adjustment and digitizing in 16-bits? (please add digitizer bit depth in Materials and methods), this suggests that saturation is still occurring. As none of the results figures presented data in absolute intensity values and used relative a.u. or F/F0 ratios which can give different "max levels" depending on the F0, it is hard for a reviewer to determine what the actual peak/max levels are and if indicator saturation was reached.
The importance of indicator saturation and/or Ca2+ release uniformity is magnified in using variance as the primary measure to resolve Ca2+ release events. Large, saturated areas have essentially no variance (after subtracting shot noise), and could dramatically underestimate SD values and obscure Ca2+ events. In a similar fashion, simultaneous firing of multiple puffs throughout a cell could also be difficult to interpret, as one would lose the ability to spatially discriminate individual events. This, also, would reduce SD values and make it hard to distinguish between cytoplasmic and "punctate" events.
While the use of Fluo8L, which has a much larger dynamic range in response to Ca2+ concentrations goes some way to assuage concerns over indicator saturation, Ca2+ levels may still be saturating this dye given the nature of the abrupt IP3-induced Ca2+ release. Out of curiosity, what does an IP3-uncaging event look like in a cell treated with ionomycin?
If the saturated/clipped regions from your recordings were dynamically filtered out using image processing and SD/intensity recalculated in parts of the cell in which the indicator is presumably unsaturated, this would help to allay saturation concerns on SD measurements.
2) Figure 1, ROIs
Towards the end of the rebuttal ("Summarizing key points…"), where you indicate the attention drawn to subjectively-placed ROIs was based on a "misconception", then one has to question ROIs were used at all. ROIs do illustrate important features of the Ca2+ release events, such as an apparent reduction in intensity variability in traces that reach near-maximal intensity (see first point).
3) 2 populations of IP3Rs…
Concerning point v) Figure 4 (we think you meant Figure 6), if potential methodological concerns are addressed – and indeed the punctate responses do drop out as global levels rise – the concluding statement that: "Our findings of a diffusive mode of Ca2+ liberation implicate a second population of IP3Rs with properties distinct from those clustered at puff sites" is too strong a statement given the evidence presented.
We think there needs to be experiments/analysis to show that punctate and global Ca2+ responses can be unrelated for this to be the case. In most examples presented the pattern of punctate release events appear to correlate with changes in cytoplasmic/global Ca2+.
For example, in Figure 2, the rate and amplitude of the SD events appears to be mirrored in the shape of the global responses. In E (puce color), the abrupt increase in SD is reflected as a sharp transition in global Ca2+ from baseline. In F (blue) the ramping up of SD events is reflected as a more gradual, sloping increase in global Ca2+. These relationships are illustrated nicely in G.
The maximum slope of global Ca2+ increases seems to occur at the point of maximum SD events. See Figure 4, Figure 5G and H (albeit de-amplified), Figure 6, Figure 8 (A: see 2 peaks in SD reflected in 2 steps in rate of Ca2+ intensity, like in Figure 3—figure supplement 3).
To disprove that there is just one population of IP3Rs, comparisons of the rate change of SD and global fluorescence should show little (or highly variable) correlation strength. Cumulative integration of SD and cumulative global fluorescence signals (akin to Ca2+ mass / SD mass) followed by regression, cross-correlation (with ∆ time offsets) and/or inflection analysis would be appropriate to examine those potential relationships. Performing this analysis on specific regions in cells to elucidate any potential zone-of-influence between punctate and global Ca2+ characteristics.
https://doi.org/10.7554/eLife.55008.sa1Author response
[Editors’ note: The authors appealed the original decision. What follows is the authors’ response to the first round of review.]
The reviewers agreed that your work is interesting and has the potential to be quite important, but there were major concerns with the SD approach, on which most of the presented findings hinge, involving the dynamic range of the Ca2+ indicators and the spatial and temporal Ca2+ release patterns.
The first concern is that after uncaging or after the addition of agonists, the cytoplasmic background fluorescence in cells increases to a seemingly very high level. As punctate release events are only observed when background levels are low to moderate, it brings into question whether the Ca2+ indicators have become saturated at peak cytoplasmic background Ca2+ levels (the author's have previously assessed a variety of Ca2+ indicators with Kd values in the 150-400nM range). This would make it essentially impossible to resolve punctate events, even though punctate release events may still be occurring – akin to "signal clipping". While SD analysis may provide some additional resolution in situations with narrow dynamic fluorescence ranges, it still depends on the indicator having some dynamic range. This concern needs to be carefully addressed so that readers are confident that the presented findings are accurate and biological.
We agree that this is a crucial issue. Our original submission already included several lines of evidence indicating that fluorescence signals were well below the saturating level, even at the peak of large IP3-evoked responses. However, these data may not have been presented with sufficient clarity. We describe below our arguments and call out specific figures and places in the text where they are presented in our revised manuscript.
i) The maximal, saturating signals evoked by ionomycin were substantially higher than peak IP3-evoked fluorescence signals, and suppression of puff activity occurred at yet lower levels (paragraph two of subsection “Ca2+ puff activity terminates during the rising phase of global Ca2+ signals”). See also the sample traces we show in Author response image 1.
Author response image 1
ii) We observed the same patterns of Ca2+ liberation using the low affinity indicator fluo-8L (Kd 1.86 µM) as compared to Cal520 (Kd 320 nM) which we used for the majority of our experiments (Figure 3—figure supplement 2).
iii) We were able to resolve instances of local puff activity even at the peak of IP3-evoked global fluorescence increases (Figure 3—figure supplement 1).
iv) Puff activity (SD signal) in cells exclusively expressing type 1 IP3R was suppressed during the rising phase of global Ca2+ signals and followed a similar pattern to WT cells, even though the magnitude of the peak fluorescence increase was <40% of that in WT cells (Figure 6).
v) The onset and termination of puff activity during the rise of IP3-mediated global Ca2+ signals were little altered when the initial basal Ca2+ level was elevated (Figure 4A,B) or, conversely, when the global Ca2+ rise was attenuated by buffering with cytosolic EGTA (new Figure 4C,D).
vi) The suppression of punctate Ca2+ liberation throughout all phases of IP3-evoked Ca2+ responses evoked when ER Ca2+ stores are partially depleted (Figure 5) strongly supports our proposal that Ca2+ liberation can arise in a diffuse manner that sustains Ca2+ signals after puff activity terminates during the rising phase of the response.
We thus conclude that the termination of puff activity during IP3-evoked signals reflects the physiological functioning of IP3Rs and is not an artifact of indicator saturation.
Some ways that could be used to examine the potential dynamic range issue of the indicators include:
• use a ratiometric indicator (Fura) to estimate actual cytoplasmic Ca2+ levels during uncaging/agonists which could be compared to Kd/saturation mM of Cal-520 and GCamP etc.
We do not feel this approach is appropriate, with problems including uncertainties in determining the Kd of both Fura-2 and Cal520 in the cytosolic environment. Further, accurate calibration of Fura-2 signals could be impaired because Fura-2 itself may be nearly saturated; especially given that the affinity of Fura-2 is higher than Cal520. Instead, we believe our approach of comparing the peak amplitude of IP3-evoked Cal520 signals with the maximal (saturating) amplitude evoked by ionomycin provides a more direct and convincing evidence that IP3-evoked Cal520 signals were well below saturation.
• pharmacologically limit the increase in cytoplasmic Ca2+ using low/moderate levels of EGTA (to preserve the dynamic range of the indicator) and examine whether punctate release patterns are still reduced after uncaging etc.
This is a good suggestion. Thank you! We performed this experiment (luckily before our lab was shut down in response to the Covid19 crisis!) and incorporate the new data as Figure 4C,D. Our results show the same pattern of SD activity in EGTA loaded cells as in non-loaded cells – i.e. puff activity ceased before the peak of the global IP3-evoked Cal520 signal even though the peak amplitude of the global signal was much (~40%) lower than in control, non-EGTA-loaded cells (paragraph two subsection “Ca2+ puffs do not terminate because of rising cytosolic Ca2+ during cell-wide elevations.”).
• purposefully increase cytoplasmic Ca2+ levels to near maximum with ionomycin then initiate an uncaging event to see puncate release sites can still be resolved (temporally synchronized release).
Although performed for a different purpose, our experiments (Figure 4A,B) using photorelease from caged Ca2+ (rather than ionomycin) to elevate cytosolic [Ca2+] prior to evoking an IP3-mediated response already address this issue. Our results showed that puff activity was still triggered even with prior [Ca2+] elevations to levels (Cal520 DF/Fo of 6-8) at which puffs are suppressed during the rising phase of IP3-evoked responses beginning from baseline cytosolic [Ca2+].
• use quantitative shape analysis (not ROIs) of transient bright regions at high cytoplasmic Ca2+ concentrations. At these levels where there is extremely limited dynamic range and greater shot noise, noise aggregates and punctate release sites will appear similar – but they can be separated using perimeter:area ratios (high for noise aggregates), or Gaussian fitting (better for release sites). This will require no Gaussian filtering of the data and some form of deconvolution to reduce the smoothing of objects due to Z-smear (where appropriate).
This is an interesting suggestion, but developing, verifying and applying appropriate algorithms and software would incur a substantial time delay. We further contest the statement that dynamic range would be extremely limited. Mean peak Cal520 fluorescence signals evoked by i-IP3 were <40% of the maximal response evoked by ionomycin (paragraph two subsection “Ca2+ puff activity terminates during the rising phase of global Ca2+ signals”).
The second concern is discussed adequately in the reviews below, but questions whether the SD approach can detect synchronous, widespread puffs/ release events, or high frequency puffs/release events that appear and behave like changes in cytoplasmic background. It would be useful to the reader to see how different duration timespans affect the variance signal (which is currently tailored for the timecourse of individual puffs), to assess what range of Ca2+ behaviors are included or filtered by varying the space/time variance parameters.
We now include (new Figure 1—figure supplement 1) an evaluation of how varying the space/time variance parameters of the algorithm affect the resulting SD fluctuation signal. Other concerns are dealt with below in our responses to specific comments by the reviewers.
Without additional information to address the major concerns of the reviewers, it is not possible to assess whether the work represents a sufficient advance to warrant publication in eLife.
A major concern of the reviewers was that the decline of the SD signal (termination of puff activity) during the rise of global Ca2+ elevations could be attributed to an artifact of indicator saturation. We feel that this issue, and other concerns of the reviewers, are fully addressed by our responses detailed here, and that our findings provide convincing support for our proposal of two different functional modes of IP3-mediated signaling involving both punctate and diffuse Ca2+ liberation. Our model represents a major revision of the widely accepted “fire-diffuse-fire” notion by which global Ca 2+ signals are generated by coordination of transient, punctate release events. This represents an important advance and re-evaluation of current dogma in the field of Ca2+ signaling that we hope would warrant publication in eLife.
Reviewer #1:
1) Figure 1 provides compelling evidence that noise decreases as the IP3-evoked Ca2+ signals develop and that a comparable increase in global cytosolic [Ca2+] evoked by ionomycin is “noise-free”, as expected for homogenous Ca2+ release. My understanding is that after some filtering and spatial blurring, most analyses calculate the SD for each pixel as a 160-ms (ie 20 frames) boxcar average.
We now include data (Figure 1—figure supplement 1) illustrating how the SD signal is affected by varying parameters for spatial filtering and boxcar averaging.
The original view posits that a global Ca2+ signal reflects most Ca2+ puff sites firing very frequently (perhaps 20 or so sites active at each instant), each spreading a µm or more from its source. Might this cause elevations in cytosolic [Ca2+] that are less noisy than the infrequent Ca2+ puffs observed initially, because each pixel would be influenced by Ca2+ reaching it from several puffs? It's perhaps too tall an order to require quantitative simulation of this on SD measurements, but it would be helpful to have the authors consider the issue explicitly.
We interpret the essence of this concern as to whether the continued, “noise-free” rise in cytosolic Ca2+ after puff activity ceases might arise as Ca2+ released during puffs diffuses into spaces between puff sites – i.e. that all the Ca2+ released during global responses originally derives from puffs. Perhaps the strongest argument supporting a distinct, “diffuse” mode of release comes from our experiments with partial depletion of ER Ca2+ content (Figure 5), where we show Ca2+ responses that arise in the absence of detectable puff activity. Moreover, a simple calculation indicates that Ca2+ diffusion would be too fast to account for the continuing smooth rise in Ca2+. Assuming a value of D = 20 mm2 s-1 for the effective diffusion coefficient of Ca2+, the mean time t for a Ca2+ ion to diffuse midway between puff sites separated by 10 mm ( d =10/2 mm) would be about 300 ms (t = d2/4D). In contrast, global Ca2+ signals typically continued to rise and were maintained for several seconds in the face of SERCA uptake after puff activity ceases (Figure 8). We had further addressed this issue in the context of whether Ca2+ diffusing from release sites deep in the cell interior might underlie the noise-free rise observed by TIRF imaging, but our lightsheet imaging indicated this was not the case (Figure 7).
We now explicitly discuss this issue in the revised manuscript (paragraph two subsection “Puff activity during global Ca2+ signals”).
2) From results in Figure 5, the authors propose that loss of Ca2+ from the ER causes puffs to terminate without appreciably abrogating the global increase in cytosolic [Ca2+]. That proposal would align with a long-standing, but contentious, suggestion that luminal Ca2+ supports IP3R activity, but it seems difficult to align with other work from this lab:
A paper from this lab and under review with eLife (Mak et al) proposes that luminal Ca2+ inhibits IP3R activity, such that loss of ER Ca2+ would be expected to enhance Ca2+ puffs.
The proposal in the Mak et al. paper does not appear to be incompatible with our model. The proposed luminal regulation of IP3R activity would create a positive feedback loop, such that any local depletion of ER Ca2+ would speed Ca2+ liberation through puffs and more rapidly terminate that liberation if local stores at puff sites became more completely emptied. We prefer not to include discussion of this point at present, as the Mak et al. paper is not yet published.
Substantial evidence, much of it form this lab, indicates that individual Ca2+ puffs do not detectably affect ER luminal [Ca2+], since steps during the falling phase of a Ca2+ puff (reflecting channel closures) are of fixed amplitude. It is not thereby clear that the flurry of puff activity would cause the loss of ER Ca2+ proposed to terminate Ca2+ puffs.
The authors should discuss both issues.
Findings of uniform step decrements of Ca2+ during the falling phase of puffs do suggest that there is little decline of ER Ca2+ during individual puffs evoked at relatively low frequency. However, 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. We now incorporate discussion of this point (paragraph four subsection “Puff activity during global Ca2+ signals”).
Reviewer #2:
[…]
I have a few comments that might suggest that the authors could clarify a few points.
1) Why aren't local release events observed in the raw ratio images, as Parker has described many times?
Local release events are indeed present in the raw ratio images. They are not readily visible in the figures because the grey scale range was adjusted to encompass the much higher fluorescence levels during global responses, and because puffs in the HEK cells we used for most experiments are smaller than in cell lines such as SHSY-5Y and HeLa. We now include a note in the text (Results paragraph two) to clarify the point for readers who may also have this question.
2) Figure 1—figure supplement 1 shows the sites with high standard deviation (SD) to be rather large, ~10 uM, a size that is inconsistent with the size of clusters of ~15 IP3R that generate the Ca2+ puffs. Rather their size suggests to me that these are entire regions of the endoplasmic reticulum (ER) that are near the plasma membrane in the TIRF evanescent field. If this is the case, are the authors selecting small sites within the large "contact" sites for the SD analyses? And, I would think that there is a "global"/homogenous signal at these sites that is contributed by Ca2+ that has diffused into these "contact" sites. This might also be consistent with the observation that the TIRF-detected high-SD signal terminates before the peak signal is achieved, because the peak is contributed by slow Ca2+ diffusion that persists after release has terminated.
All of our SD analyses (excepting Figure 1, where we show small regions of interest for purposes of illustrating the technique) were done using regions of interest encompassing the entire cell. There is thus no issue of “selecting small sites … for the SD analyses”. We now emphasize this point more clearly in the manuscript (subsection “Fluctuation analysis reveals a transient flurry of puffs during global Ca2+ signals”).
Please see our response to major point #1 of reviewer #1 regarding whether the peak is contributed by slow Ca2+ diffusion that persists after release has terminated.
3) The authors describe an inverted U relationship between the amplitude of the SD signal and the amplitude of the global response. What does it mean?
We show plots of amplitude of the SD signal and the amplitude of the global response as a convenient means of summarizing and depicting the relationship between these parameters in a time-independent manner. As described in the manuscript we originally thought the inverted U relationship might correspond to the “bell-shaped curve” of Ca2+ dependence of activation of IP3Rs. However, further experiments (Figure 4, now incorporating new data) indicated that is not the case. Our interpretation (Discussion, subsection “Puff activity during global Ca2+ signals”) is that the cytosolic Ca2+ signal is an inverse mirror of the local ER Ca2+ store level, and that the decline in SD signal arises as puff activity declines owing to falling ER store content.
4) The authors conclude that elevated cytoplasmic Ca2+ doesn't inhibit the IP3R as a primary mechanism to terminate Ca puffs. Why doesn't it inhibit?
Although high cytosolic [Ca2+] would be expected to inhibit opening of IP3R channels, a likely explanation is simply that depletion of ER Ca2+ stores is the more dominant mechanism to terminate puff activity during global signals. This interpretation is buttressed by new data we present (Figure 4C,D), demonstrating that puffs still terminate even when the global cytosolic Ca2+ elevation is substantially reduced by loading with EGTA.
5) The authors examine the decay in cytoplasmic Ca2+ concentration after an elevation. They refer to this as sequestration. I'm not sure of their meaning. Although they suggest that the kinetics can be described by a single exponential, there must be at least two processes involved: sequestration by the ER, and extrusion by the plasm membrane Ca2+ pump. Indeed, in the discussion they use the phrase "efflux rate from the cytoplasm". I think what the authors mean to suggest in the sequestration away from the cytoplasm. If so, this should be stated more clearly.
We agree that our use of the term “sequestration” was imprecise. We have replaced it by terms such as “removal from the cytosol”. Our experimental data fit well to a single exponential process, likely because sequestration by the ER is much faster than loss across the plasma membrane (e.g. Figure 5H)
Reviewer #3:
[…] The experimental manipulations to investigate depletion of stores, Ca2+ entry, IP3R channel KO etc are well thought out and appropriate – but some concerns remain as to the ability of the authors to resolve puffs once background/cytoplasmic Ca2+ levels reach a certain intensity.
The use of signal variance to isolate different components of mixed fluorescence signals relies on those components having distinct (and variable) spatio-temporal signatures. However, if Ca2+ release events reach a steady state, then the variance technique becomes essentially useless. This is my main concern with some of the conclusions drawn from the analysis.
By the statement “Ca2+ release events reach a steady state” we interpret the reviewer to mean that Ca2+ release through the IP3Rs that mediate puffs becomes continuous; i.e. that some or all of the IP3R channels remain open all the time, resulting in a “noise-free” signal. If this indeed occurs, we believe it is valid and useful to regard such continuous release as a mode distinct from the transient puffs. The issue, then, is whether the same IP3Rs underlie puffs and continuous release, or whether these modes arise from separate populations of IP3Rs. Although we favor the notion that distinct populations of IP3Rs mediate the two modes of Ca2+, we now explicitly discuss the possibility the two modes may instead arise through a switch of the same IP3Rs from one form of regulation to another (Discussion and see response to similar query of reviewer #1).
The ability to detect puffs is greatest during low background conditions, but the as global Ca2+ levels increase, the ability to detect puffs is hampered by i) closing of the max difference between background and puff sites and ii) an increase in the amplitude of shot noise that may start to mimick fast and small puff events. To resolve between genuine puff events, that are inherently Gaussian (1-2 µm) and random noise aggregates that occur during high background levels, is difficult – but is made worse by applying Gaussian filters to the data that reshape noise aggregates into more Gaussian shapes. Employing shape analysis before smoothing (perimeter:area ratios) is an effective way of resolving puffs (low P:A – smooth circles) even when surrounded by noise aggregates of equal intensity and size (high P:A – leafy looking). This may assuage some doubts as to the authors ability to resolve punctate events once background/global levels have reached moderate to high levels.
Use of shape analysis is an interesting suggestion; but developing, verifying and applying appropriate algorithms and software would incur a substantial time delay. We feel that the data we already present convincingly demonstrate that the fall in SD signal during the rising phase of global Ca2+ responses truly represents a decline in puff activity, not a failure to detect puffs in the face of elevated Ca2+levels. Notably; (i) Puff activity (SD signal) was still observed in response to IP3 activation when the basal cytosolic Ca2+ was substantially elevated by prior photorelease of Ca2+ (Figure 4A,B). (ii) We now include new data showing that puff termination follows similar kinetics when peak global Ca2+ elevations were reduced by EGTA buffering (Figure 4C,D). Moreover, we present additional evidence in our rebuttals to concerns of the other reviewers that our ability to detect puffs at elevated Ca2+ levels may have been impaired by limited dynamic range of the indicator.
Similarly, the use of user-placed (or randomly positioned) ROIs based on visual inspection of bright spots in the videos is subjective and can heavily bias results to what the user wants to show. Many "hot spots" don't appear to have ROIs drawn over them and some ROIs appear to be sampling the same site. Small ROIs are more sensitive to spatial shifts in the position of the underlying fluorescent sites. The "flickering" variance may be a function of out-of-sync release events from a number of channels/clusters covered by each ROI. When these channels syncronize (in the open position?) or fluorescence sites increase in size, this will dramatically reduce the signal variance, even though puffs may still be occurring.
We show SD traces from small, subcellular ROIs only in Figure 1, for the purpose of better illustrating the basis of our technique. For all the other experiments – from which our conclusions are drawn – we used a single ROI encompassing the entire cell. There is thus no issue of subjective bias resulting from user-placed ROIs. The reviewer appears to have missed this methodological point, which we now emphasize more clearly in the text (subsection “Fluctuation analysis reveals a transient flurry of puffs during global Ca2+ signals”).
Our SD signal represents a measure of temporal fluctuations aggregated across all pixels covering the cell. It will thus be substantially unaffected whether local puffs are synchronized or asynchronous. Of course, if channels all “synchronize” in a continuously open state there would be no fluctuations – but as described above, we would regard this as a distinctly different mode of release.
The shape of the cell and the distance and size of the cytoplasm located away from the evanescent field can dramatically affect the blurring of the background (cytoplasmic signal). How do the authors discriminate between "global" cytoplasmic changes and large puff sites that are heavily blurred being out of focus (see Figures 1 and 7). This effect still remains using lightsheet imaging.
Again, all our analyses are based on SD and DF/F0 traces derived from the same, single ROI encompassing each cell. They represent an average of the temporal fluctuations and Ca2+ level across the entire TIRF footprint of the cell. We do not attempt to discriminate on the basis of spatial patterns between global signals and large puffs.
The pixel resolution used during TIRF imaging was 0.5µm and is unlikely to fully resolve individual puffs, especially after smoothing. This may bias analysis to release events that are comprised of multiple channel clusters.
The camera pixel resolution, subsequent Gaussian smoothing, and boxcar calculation of SD were chosen to optimally “tune” for the spatio-temporal characteristic of puffs while minimizing stochastic photon shot noise. The SD signal simply reports an average of pixel-by-pixel temporal fluctuations across the entire cell. We do not attempt to discriminate between individual or multi-cluster events.
In response to the editor’s comment we now include a supplementary figure (Figure 1—figure supplement 1) illustrating how different parameters in our algorithm affect the SD signal. The spatial smoothing we applied (σ corresponding to ~1 mm at the specimen) effectively reduced photon shot noise see without smoothing, but did not otherwise appreciably alter the SD signal arising from puff activity.
The use of black level subtraction can normalize absolute intensity levels, but does not compensate for the higher shot noise in areas with high static background. Was this taken into account?
Our algorithm subtracts the mean shot noise level predicted for the intensity level on a pixel-by-pixel basis (subsection “Fluctuation processing of Ca2+ images highlights transient signals” paragraph three).
Gaussian blur is reported as 2 pixels (in text) but as an SD of 2 pixels in the Materials and methods. What was the actual kernel size in pixels (3x3, 5x5?) and the SD.
Spatial smoothing was accomplished using a Python routine (skimage) that applies a two-dimensional Gaussian blur with specified standard deviation (σ) out to a range of 4 σ. We now clarify this description, and state σ in terms of both pixels and corresponding micrometers at the specimen (subsection “Image Analysis”). See new Figure 1—figure supplement 1 for an illustration of how the SD signal is affected by different values of Gaussian blur.
Random ROIs were used for HEK cells devoid of IP3Rs. What does the data look like if the authors select the brightest spots in the 3KO video – or use random ROIs in controls?
As noted above, we show traces from small, subcellular ROIs only in Figure 1, for the purpose of helping explain the basis of our SD approach. Use of random ROIs for the 3KO cell would seem appropriate for that purpose. Excepting the introductory Figure 1, all data were acquired from single regions of interest encompassing the entire cell (yellow outlines and traces in Figure 1).
Figure 3—figure supplement 3C appears to show a stochastic build up of high frequency events correlated to baseline increases – until background fluorescence levels may start to swamp out the ability to resolve small punctate events.
As described in responses to several of the reviewers’ comments, we present convincing evidence that fluctuations from punctate events do not become “swamped out” at higher background fluorescence levels, but rather that the punctate release activity itself declines.
[Editors’ note: what follows is the authors’ response to the second round of review.]
The reviewers have also provided additional feedback below after reading your rebuttal to hopefully better convey the technical concerns.
Specific points:
1) Indicator Saturation/Clipping concerns:
In your rebuttal, you state (i) that "…maximal signal evoked by ionomycin is>2 fold higher than peak IP3-evoked fluorescence signals"
However, in Figure 1, the maximum raw signal during uncaging (1C) and after ionomycin (1F) looks exactly the same – both topping out at 8500 a.u – and suggests that IP3-stimulated Ca2+ release may be reaching levels that saturate the dye.
We apologize that we had not fully described the procedure used to evoke the ionomycin response in Figure. 1, leading to the reviewer’s confusion. The experiment in Figure 1F,H was done under conditions deliberately chosen to evoke a Ca2+ fluorescence signal matching the amplitude and kinetics of the IP3-evoked signal in Figure 1A,C. Our intent was to compare the output of our SD analysis routine with signals of similar size where punctate release was either present or absent. For this purpose, we delivered a small aliquot of ionomycin at a distance (determined by trial and error) from cells imaged in zero Ca2+ solution, so that the slow diffusion and dilution of ionomycin evoked gradual release of Ca2+ from intracellular stores.
Separately, we used ionomycin to determine the saturating fluorescence of the indicator by complete bath exchange of medium containing high (10 mM) calcium with ionomycin. We now clarify these points in the text, and also state the peak amplitudes of the ionomycin-and i-IP3-evoked evoked response in Figure 1 as ΔF/F0 values as well as in fluorescence units.
Similarly, the large areas that show uniform white intensity (Figure 1, image v) are either due to: i) over zealous contrast/brightness adjustment, ii) indicator saturation @ max or iii) camera chip saturation (clipping). Assuming reasonable contrast adjustment and digitizing in 16-bits? (please add digitizer bit depth in Materials and methods), this suggests that saturation is still occurring.
Camera digitization was 16 bit; the camera was operated far below full-well capacity; and image processing was done using 64 bit floating point arithmetic. No clipping occurred anywhere through the entire imaging pipeline. This is now stated in Materials and methods.
Areas of uniform white in Figure 1A,F resulted because we chose a grey scale that would still display some data at low fluorescence levels, and this inadvertently clipped the higher intensity values. We have now re-made this figure with a more extended grey scale so that the Cal520 fluorescence image panels at the peak of the response are no longer clipped.
As none of the results figures presented data in absolute intensity values and used relative a.u. or F/F0 ratios which can give different "max levels" depending on the F0, it is hard for a reviewer to determine what the actual peak/max levels are and if indicator saturation was reached.
The term “absolute intensity value” (even if expressed as photons/pixel/unit time) has little meaning in this context. The fluorescence intensity will depend on factors including excitation laser intensity, concentration of indicator loaded in the cytosol, TIRF angle setting, etc.; factors that will vary between cells and between experiments. Instead, we follow the widely used practice of expressing data as ΔF/F0 values to normalize for these factors. The only crucial underlying assumption is that the basal free Ca2+ level (setting F0) is consistent between cells prepared under the same conditions. From many years’ experience, including comparison measurements with fura-2, we find this to be the case.
For reference, we show in Author response image 1 representative traces of strong responses to i-IP3 in 3 cells black), and maximal responses to ionomycin in 3 cells (red). These data are displayed as both “raw” fluorescence in arbitrary camera units (left) and as F/F0 values (right).
The importance of indicator saturation and/or Ca2+ release uniformity is magnified in using variance as the primary measure to resolve Ca2+ release events. Large, saturated areas have essentially no variance (after subtracting shot noise), and could dramatically underestimate SD values and obscure Ca2+ events.
While we agree that our data and conclusions would be dramatically affected if the indicator became saturated, we present numerous lines of evidence (including new data – Figure 4C,D – incorporated in our revised manuscript) clearly demonstrating that IP3-evoked Ca2+ signals were well below the level of indicator saturation, and that our key conclusion that puff activity terminates before the peak of global signals is not an artifact of indicator saturation.
In a similar fashion, simultaneous firing of multiple puffs throughout a cell could also be difficult to interpret, as one would lose the ability to spatially discriminate individual events. This, also, would reduce SD values and make it hard to distinguish between cytoplasmic and "punctate" events.
Our SD signal reflects the cell-wide aggregate of fluctuations at each pixel, and is independent of any requirement to spatially discriminate individual events.
While the use of Fluo8L, which has a much larger dynamic range in response to Ca2+ concentrations goes some way to assuage concerns over indicator saturation, Ca2+ levels may still be saturating this dye given the nature of the abrupt IP3-induced Ca2+ release.
We do not understand this point. Ca2+ binding to the indicator (and its fluorescence) will equilibrate with free [Ca2+] on a timescale vastly more rapidly than the timescale of seconds for the IP3-evoked signal.
Out of curiosity, what does an IP3-uncaging event look like in a cell treated with ionomycin?
Because ionomycin will cause uncontrolled loss of Ca2+ from the ER, we would not expect any response to uncaged IP3 if no Ca2+ remained to be released. In Figure 4A,B we show i-IP3-evoked responses after cytosolic [Ca2+] was elevated using caged calcium.
If the saturated/clipped regions from your recordings were dynamically filtered out using image processing and SD/intensity recalculated in parts of the cell in which the indicator is presumably unsaturated, this would help to allay saturation concerns on SD measurements.
Again, we present compelling evidence that our IP3-evoked signals are well below the level of indicator saturation. Our revised manuscript includes new experimental data, and a new section in the Discussion summarizing and evaluating this evidence.
As shown in (revised) Figure 1A and Video 1, the fluorescence at the peak of IP3-evoked signals is quite homogeneous throughout the cell (TIRF footprint), without evidence of saturated/clipped regions.
2) Figure 1, ROIs
Towards the end of the rebuttal ("Summarizing key points…"), where you indicate the attention drawn to subjectively-placed ROIs was based on a "misconception", then one has to question ROIs were used at all. ROIs do illustrate important features of the Ca2+ release events, such as an apparent reduction in intensity variability in traces that reach near-maximal intensity (see first point).
We show traces from subcellular ROIs in Figure 1 as a way to illustrate the behavior of puff sites (particularly to provide helpful background for readers who may not be familiar with this literature), and to show how the aggregate behavior of multiple sites results in the whole-cell ROI traces we present in the rest of the manuscript. We have reworded the text to better communicate this intent. We agree that there is valuable information to be derived from analysis of individual release events, but leave that analysis for a future publication.
3) 2 populations of IP3Rs…
Concerning point v) Figure 4 (we think you meant Figure 6), if potential methodological concerns are addressed – and indeed the punctate responses do drop out as global levels rise – the concluding statement that: "Our findings of a diffusive mode of Ca2+ liberation implicate a second population of IP3Rs with properties distinct from those clustered at puff sites" is too strong a statement given the evidence presented.
We agree that our statement was too strong. In our revised manuscript we now refer to two “modes” of Ca2+ release, and in the Discussion speculate as to whether these modes may arise from two separate populations of IP3R, or through a modulation of the functional properties of a single population. We further address this point in our detailed responses to the original comments of the reviewers.
We think there needs to be experiments/analysis to show that punctate and global Ca2+ responses can be unrelated for this to be the case.
Figure 5A,B,D,F,H,I present clear-cut cases where a global Ca2+ signal is evoked without detectable puff activity (SD signal); i.e. in this situation (partial depletion of ER Ca2+) punctate and diffuse modes of Ca2+ release are indeed unrelated.
In most examples presented the pattern of punctate release events appear to correlate with changes in cytoplasmic/global Ca2+.
For example, in Figure 2, the rate and amplitude of the SD events appears to be mirrored in the shape of the global responses. In E (puce color), the abrupt increase in SD is reflected as a sharp transition in global Ca2+ from baseline. In F (blue) the ramping up of SD events is reflected as a more gradual, sloping increase in global Ca2+. These relationships are illustrated nicely in G.
The maximum slope of global Ca2+ increases seems to occur at the point of maximum SD events. See Figure 4, Figure 5G and H (albeit de-amplified), Figure 6, Figure 8 (A: see 2 peaks in SD reflected in 2 steps in rate of Ca2+ intensity, like in S5).
This is consistent with, and predicted by, our proposal that punctate Ca2+ release though puffs contributes a substantial (~40%; Figure 8A-C) fraction of the total Ca2+ liberation during the rising phase of global Ca2+ signals. We thus expect that the rising phase of the global signal will reflect both punctate and diffuse modes of Ca2+ liberation.
To disprove that there is just one population of IP3Rs, comparisons of the rate change of SD and global fluorescence should show little (or highly variable) correlation strength. Cumulative integration of SD and cumulative global fluorescence signals (akin to Ca2+ mass / SD mass) followed by regression, cross-correlation (with ∆ time offsets) and/or inflection analysis would be appropriate to examine those potential relationships.
As noted above, our revised manuscript refers to two modes of Ca2+ release, and we discuss whether these may arise from two populations of IP3Rs or a single population. Because we conclude that punctate Ca2+ release makes up an appreciable part of the total Ca2+ released during the rising phase of the cell-wide response, we do expect correlation between the SD signal and the global fluorescence.
Performing this analysis on specific regions in cells to elucidate any potential zone-of-influence between punctate and global Ca2+ characteristics.
This suggestion appears predicated on a proposal that punctate Ca2+ and diffuse Ca2+ release arise at different sites in the cell. An interesting question, but one we feel is beyond the scope of the present manuscript, and a topic for future experiments once our lab reopens after the Covid-19 pandemic.
https://doi.org/10.7554/eLife.55008.sa2Article and author information
Author details
Funding
National Institute of General Medical Sciences (R37 GM048071)
- Ian Parker
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Acknowledgements
We thank Dr. Carley A Karsten for assisting in initial imaging experiments and Dr. Kyle L Ellefsen for help with software and image analysis.
Senior and Reviewing Editor
- Kenton J Swartz, National Institute of Neurological Disorders and Stroke, National Institutes of Health, United States
Reviewers
- Colin W Taylor, University of Cambridge, United Kingdom
- Grant Hennig, Univ. Vermont, United States
Publication history
- Received: January 9, 2020
- Accepted: May 12, 2020
- Accepted Manuscript published: May 12, 2020 (version 1)
- Version of Record published: May 27, 2020 (version 2)
Copyright
© 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.
Metrics
-
- 3,341
- Page views
-
- 356
- Downloads
-
- 20
- Citations
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
IP3 mediated global Ca2+ signals arise through two temporally and spatially distinct modes of Ca2+ release
eLife 9:e55008.
https://doi.org/10.7554/eLife.55008
Further reading
-
- Cell Biology
- Immunology and Inflammation
Tristetraprolin (TTP) is a critical negative immune regulator. It binds AU-rich elements in the untranslated-regions of many mRNAs encoding pro-inflammatory mediators, thereby accelerating their decay. A key but poorly understood mechanism of TTP regulation is its timely proteolytic removal: TTP is degraded by the proteasome through yet unidentified phosphorylation-controlled drivers. In this study, we set out to identify factors controlling TTP stability. Cellular assays showed that TTP is strongly lysine-ubiquitinated, which is required for its turnover. A genetic screen identified the ubiquitin E3 ligase HUWE1 as a strong regulator of TTP proteasomal degradation, which we found to control TTP stability indirectly by regulating its phosphorylation. Pharmacological assessment of multiple kinases revealed that HUWE1-regulated TTP phosphorylation and stability was independent of the previously characterized effects of MAPK-mediated S52/S178 phosphorylation. HUWE1 function was dependent on phosphatase and E3 ligase binding sites identified in the TTP C-terminus. Our findings indicate that while phosphorylation of S52/S178 is critical for TTP stabilization at earlier times after pro-inflammatory stimulation, phosphorylation of the TTP C-terminus controls its stability at later stages.
-
- Cell Biology
We show that TANGO2 in mammalian cells localizes predominantly to mitochondria and partially at mitochondria sites juxtaposed to lipid droplets (LDs) and the endoplasmic reticulum. HepG2 cells and fibroblasts of patients lacking TANGO2 exhibit enlarged LDs. Quantitative lipidomics revealed a marked increase in lysophosphatidic acid (LPA) and a concomitant decrease in its biosynthetic precursor phosphatidic acid (PA). These changes were exacerbated in nutrient-starved cells. Based on our data, we suggest that TANGO2 function is linked to acyl-CoA metabolism, which is necessary for the acylation of LPA to generate PA. The defect in acyl-CoA availability impacts the metabolism of many other fatty acids, generates high levels of reactive oxygen (ROS), and promotes lipid peroxidation. We suggest that the increased size of LDs is a combination of enrichment in peroxidized lipids and a defect in their catabolism. Our findings help explain the physiological consequence of mutations in TANGO2 that induce acute metabolic crises, including rhabdomyolysis, cardiomyopathy, and cardiac arrhythmias, often leading to fatality upon starvation and stress.
