Dissociation of inositol 1,4,5-trisphosphate from IP₃ receptors contributes to termination of Ca²⁺ puffs

Abstract

Ca²⁺ puffs are brief, localized Ca²⁺ signals evoked by physiological stimuli that arise from the coordinated opening of a few clustered inositol 1,4,5-trisphosphate receptors (IP₃Rs). However, the mechanisms that control the amplitude and termination of Ca²⁺ puffs are unresolved. To address these issues, we expressed SNAP-tagged IP₃R3 in HEK cells without endogenous IP₃Rs and used total internal reflection fluorescence microscopy to visualize the subcellular distribution of IP₃Rs and the Ca²⁺ puffs that they evoke. We first confirmed that SNAP-IP₃R3 were reliably identified and that they evoked normal Ca²⁺ puffs after photolysis of a caged analog of IP₃. We show that increased IP₃R expression caused cells to assemble more IP₃R clusters, each of which contained more IP₃Rs, but the mean amplitude of Ca²⁺ puffs (indicative of the number of open IP₃Rs) was unaltered. We thus suggest that functional interactions between IP₃Rs constrain the number of active IP₃Rs within a cluster. Furthermore, Ca²⁺ puffs evoked by IP₃R with reduced affinity for IP₃ had undiminished amplitude, but the puffs decayed more quickly. The selective effect of reducing IP₃ affinity on the decay times of Ca²⁺ puffs was not mimicked by exposing normal IP₃R to a lower concentration of IP₃. We conclude that distinct mechanisms constrain recruitment of IP₃Rs during the rising phase of a Ca²⁺ puff and closure of IP₃Rs during the falling phase, and that only the latter is affected by the rate of IP₃ dissociation.

Intracellular Ca²⁺ signals regulate many cellular processes. Most Ca²⁺ signals are initiated by inositol 1,4,5-trisphosphate (IP₃) produced when cell-surface receptors stimulate phospholipase C. IP₃ binds to the four subunits of a tetrameric IP₃ receptor (IP₃R), priming it to bind Ca²⁺, and triggering the opening of an intrinsic Ca²⁺-permeable channel, through which Ca²⁺ flows rapidly from the lumen of the endoplasmic reticulum to the cytosol (1, 2, 3). Ca²⁺ puffs are evoked by low stimulus intensities, similar to those likely to occur under physiological conditions. These Ca²⁺ puffs are brief, localized increases in cytosolic free Ca²⁺ concentration ([Ca²⁺]_c) that arise from the coordinated opening of a few channels within a cluster (4, 5, 6, 7). Compelling evidence suggests that the amplitude of a Ca²⁺ puff reports the number of open IP₃Rs. This includes evidence that the amplitudes of steps within the falling phase of a Ca²⁺ puff match those of the very smallest events (‘Ca²⁺ blips’), which report the opening of a single IP₃R (4). The coordinated openings of IP₃Rs are thought, at least in part, to arise from costimulation of all IP₃Rs by IP₃ and Ca²⁺ (1). An additional, but essential, level of regulation is provided by Kras-induced actin-binding protein (KRAP), which licenses clustered IP₃Rs to respond to IP₃ (8).

Most IP₃Rs within a cell are mobile, but Ca²⁺ puffs arise preferentially from immobile clusters of IP₃Rs which are tethered near to endoplasmic reticulum-plasma membrane junctions by KRAP (7, 8). These subcellular sites are the same whether Ca²⁺ puffs are evoked by IP₃ produced by endogenous signaling pathways or by photolysis of a caged analog of IP₃ (ci-IP₃) (6, 9). Ca²⁺ puffs may both allow local regulation of Ca²⁺ sensors and contribute to the genesis of global Ca²⁺ signals, although the role of Ca²⁺ puffs in the latter is unresolved (10).

Ca²⁺-induced Ca²⁺-release between IP₃Rs is potentially explosive, but the mechanisms that terminate the activity of Ca²⁺ puffs are incompletely understood. Delayed feedback inhibition of IP₃Rs by substantial local increases in [Ca²⁺]_c probably contributes (11, 12, 13, 14), but it may not be the only mechanism. Evidence for coupled closing of IP₃Rs during the falling phase of a Ca²⁺ puff (15) and conflicting evidence implicating luminal Ca²⁺ in the regulation of IP₃Rs (16, 17) suggest additional factors that may contribute to termination of Ca²⁺ puffs. The effect of cytosolic Ca²⁺ on IP₃Rs is determined by whether they have IP₃ bound (1). The simplest scheme suggests that IP₃ binding allows Ca²⁺ to stimulate channel opening, while Ca²⁺ inhibits IP₃Rs without IP₃ bound (18, 19, 20). This scheme suggests that if local increases in [Ca²⁺]_c contribute to terminating Ca²⁺ puffs, it may be necessary for IP₃ to first dissociate from at least one IP₃R subunit; in that case, the rate of IP₃ dissociation may be an important determinant of how quickly Ca²⁺ puffs terminate. Furthermore, we might expect coupled closing and inhibition by local increases in [Ca²⁺]_c to be influenced by the density of IP₃Rs within the clusters that evoke Ca²⁺ puffs. These considerations prompted our analyses of the effects of varying IP₃R expression and IP₃ affinity on Ca²⁺ puffs.

All three subtypes of IP₃R evoke similar Ca²⁺ puffs (21, 22), but IP₃R3 offers advantages for experimental analyses. Plasmids encoding IP₃R3 are easy to manipulate, there is an excellent IP₃R3-selective antibody, and there are high-resolution structures of IP₃R3 in different states (23, 24, 25). We therefore chose human IP₃R3 for our analyses of Ca²⁺ puffs. By expressing SNAP-tagged IP₃R3 in cells devoid of native IP₃R, we were able to visualize both IP₃Rs and the Ca²⁺ signals they evoke. Our results establish that functional interactions between IP₃Rs constrain their recruitment during the rising phase of a Ca²⁺ puff and that IP₃ dissociation from the IP₃R contributes to termination of each Ca²⁺ puff.

Results

SNAP-tagged IP₃R3 are reliably identified and evoke normal Ca²⁺ puffs

To define the subcellular distribution of IP₃R3 while measuring the Ca²⁺ puffs they evoke, human IP₃R3 fused to a fast-labeling SNAP-tag (SNAP-IP₃R3, Fig. 1A) was expressed in human embryonic kidney (HEK) cells lacking endogenous IP₃Rs (HEK-3KO cells) (2). We used an N-terminal tag because it does not disrupt IP₃R function (7) and a SNAP-tag (19.4 kDa; smaller than GFP) that allows versatile covalent labeling with fluorophores (Fig. 1B) (26). Our use of HEK-3KO cells (2) and transient expression of SNAP-IP₃R3 under control of an inducible promoter ensured that functional responses were entirely mediated by SNAP-IP₃R3 homotetrameric channels expressed at appropriate levels (Fig. 1B). Our optimized methods, which required labeling of cells in suspension before plating for analyses using fluorescence microscopy (Fig. 1, A–D), allowed SNAP-IP₃R3 to be identified in cells, most of which expressed IP₃R3 at modest levels (Fig. 1E) and with the punctate distribution of IP₃R clusters typical of native IP₃Rs (Fig. 2Aii and see Fig. 3A) (7).

Figure 1.

Controlled expression and visualization of SNAP-IP₃R3.A, the SNAP-IP₃R3 construct. A fast-labeling SNAP-tag (SNAPf) is attached to the N-terminus of human IP₃R3 (hIP₃R3) via a short peptide linker. B, expression and labeling of SNAP-IP₃R3. HEK-3KO cells are transiently transfected with plasmids encoding TetR and SNAP-IP₃R3 under a tetracycline-inducible CMV promoter. Addition of doxycycline (1 μg/ml, 48 h) then derepresses expression of SNAP-IP₃R3. HEK-SNAP-IP₃R3 cells are irreversibly labeled in suspension with a fluorescent SNAP-tag substrate (SNAP-Cell 647-SiR, 1 μM) and plated onto imaging dishes before noninvasive loading with Cal520-AM, ci-IP₃/PM, and EGTA-AM for TIRF imaging of Ca²⁺ puffs evoked by photolysis of ci-IP₃. C, TIRF images of HEK-SNAP-IP₃R3 cells labeled with SNAP-Cell 647-SiR (1 μM) in the imaging dish according to the manufacturer’s instructions or in-suspension (see Experimental procedures) and recorded under identical conditions. Scale bars represent 10 μm. D, signal-to-noise ratio (SNR) of intracellular/extracellular fluorescence measured from small ROI (5.6 μm²) for the two labeling methods shown in C. Individual values from 53 cells (points) with mean values from three independent experiments (squares, color-coded) and mean ± S.E.M., ∗p < 0.05, unpaired Student’s t test. E, frequency distributions of whole-cell epifluorescence intensities for HEK-SNAP-IP₃R3 cells (3892 cells from one experiment) and mock-transfected HEK-3KO cells (2798 cells from one experiment). The results indicate that ∼40% of cells express detectable SNAP-IP₃R3. FU, fluorescence unit. HEK-3KO cell, HEK cell with no IP₃Rs; HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI, region of interest; TIRF, total internal reflection fluorescence.

Figure 2.

SNAP-tag labeling reliably reports IP₃R distribution.A, TIRF images of HEK-SNAP-IP₃R3 cells immunostained with Ab-IP₃R3 and labeled with SNAP-Cell 647-SiR (SNAP-647) at high (i) and low (ii) levels of SNAP-IP₃R3 expression. Scale bars represent 10 μm, 5 μm in enlargements of boxed areas. B, Manders’ split coefficient values for colocalization of Ab-IP₃R3 fluorescence with SNAP-647 fluorescence (M1) and vice versa (M2) for cells expressing different amounts of IP₃R3 (see panel C). Individual values from 22 cells from two independent analyses (color-coded) and mean ± S.D. C, relationship between whole-cell fluorescence intensities from the entire TIRF footprint for SNAP-647 and Ab-IP₃R3 staining. Pearson correlation coefficient r = 0.87, p < 0.0001, n = 34 cells from two independent analyses. D, TIRF images of a HEK-SNAP-IP₃R3 cell showing punctate Ab-IP₃R3 and SNAP-647 fluorescence. Scale bars represent 10 μm, 5 μm in pseudo-colored enlargements of boxed area. E, fluorescence intensity profiles along the lines (1, 2, 3, 4) shown in D. F, relationship between fluorescence intensities of individual SNAP-647 and Ab-IP₃R3 puncta. 2310 puncta from 10 cells from two independent analyses. Pearson correlation coefficient r = 0.85, p < 0.0001. Ab-IP₃R3, IP₃R3-selective antibody; HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; TIRF, total internal reflection fluorescence.

Figure 3.

Similar Ca²⁺puffs are evoked by endogenous IP₃R3 and SNAP-IP₃R3.A, typical TIRF image of SNAP-IP₃R3 in a live cell selected for analysis because the IP₃R expression reveals mobile and immobile puncta typical of native expression (7) without the clear delineation of reticular ER that occurs with over-expression. Boxed area (enlarged below) shows the region (19.2 × 19.2 μm) from which Ca²⁺ puffs were recorded (ROI^puffs). The time-overlay of images captured at 0 s (green) and 60 s (magenta) show immobile SNAP-IP₃R3 puncta (white). Scale bars represent 10 μm, 5 μm in enlargements. B, fluorescence traces show Ca²⁺ puffs evoked by photolysis of ci-IP₃ (250-ms UV flash, arrows) in a HEK-IP₃R3 or HEK-SNAP-IP₃R3 cell. Cal520 fluorescence (F) was recorded in TIRF from a small region (1.76 × 1.76 μm), wherein Ca²⁺ puffs occurred repeatedly and is expressed relative to fluorescence recorded before the UV flash (F₀). Enlargements of boxed areas show individual Ca²⁺ puffs. C and D, frequency and latency (time from UV flash to first Ca²⁺ puff) of Ca²⁺ puffs evoked through SNAP-IP₃R3 and endogenous IP₃R3. Results show individual cells and mean ± S.D. E, measured properties of Ca²⁺ puffs (F/F₀): amplitude (baseline to peak), rise time (20% to 100% peak amplitude), decay time (100% to 20% peak amplitude), and duration (at half-maximal amplitude). F and G, mean amplitudes (F) and mean kinetic properties (G) of Ca²⁺ puffs evoked by IP₃R3 and SNAP-IP₃R3. Results show mean values from single cells and mean ± S.E.M. Results (C, D, F, and G) are from 19 cells from five independent experiments for IP₃R3 and from 18 cells from three independent experiments for SNAP-IP₃R3. ns p > 0.05, unpaired Student’s t test. HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded; TIRF, total internal reflection fluorescence.

We confirmed the specificity of an antibody to IP₃R3 (Ab-IP₃R3) (27) in immunocytochemical analyses by demonstrating that Ab-IP₃R3 stains puncta in HEK-SNAP-IP₃R3 cells, but not in HEK-3KO cells (Fig. S1). HEK-SNAP-IP₃R3 cells labeled with SNAP-Cell 647-SiR (hereafter, SNAP-647) were immunostained with Ab-IP₃R3 and imaged using total internal reflection fluorescence (TIRF) microscopy. The results demonstrate colocalization of immunostaining and SNAP-647 fluorescence (Fig. 2, A, B, D, and E), a tight linear correlation between the intensities of whole-cell immunostaining and SNAP-647 fluorescence (Fig. 2C), and a linear correlation for individual puncta between Ab-IP₃R3 and SNAP-647 fluorescence intensities (Figs. 2F and S2). The results so far (Figs. 1 and 2) demonstrate that transfection of HEK-3KO cells with SNAP-IP₃R3 and subsequent labeling with SNAP-647 allows near-native levels of IP₃R expression and reliable detection of all IP₃R3.

We used TIRF microscopy to compare Ca²⁺ puffs evoked by photolysis of ci-IP₃ in HEK-IP₃R3 and HEK-SNAP-IP₃R3 cells. Since the latter were transiently transfected (Fig. 1B), individual cells differed in their expression of SNAP-IP₃R3 (Fig. 1E), we therefore selected cells in which the distribution of IP₃R resembled that observed in HeLa cells with tagged endogenous IP₃R (7), namely cells with comparatively low SNAP-647 fluorescence intensities in which most IP₃R puncta were mobile and a smaller fraction were immobile (Fig. 3A). The results demonstrate that Ca²⁺ puffs occurred with indistinguishable frequencies and after similar latencies in the two cell types (Fig. 3, B–D). Furthermore, the properties of individual Ca²⁺ puffs (mean amplitudes, rise times, decay times, and durations) were also indistinguishable for HEK-IP₃R3 and HEK-SNAP-IP₃R3 cells (Figs. 3, E–G and S3).

Our results establish that HEK-SNAP-IP₃R3 cells after labeling with SNAP-647 allow reliable identification of all IP₃Rs within a cell and unperturbed IP₃-evoked Ca²⁺ puffs. We use these cells to explore the effects of varying IP₃R expression and the rate of IP₃ dissociation from IP₃R on the properties of Ca²⁺ puffs.

Increased IP₃R expression increases the frequency of Ca²⁺ puffs and the size of IP₃R clusters without affecting the properties of Ca²⁺ puffs

We used SNAP-647 fluorescence intensity measured from the TIRF footprint of the region of interest (ROI) from which Ca²⁺ puffs were recorded (ROI^puffs, 19.2 × 19.2 μm) to report SNAP-IP₃R3 expression in individual HEK-SNAP-IP₃R3 cells and compared it with the properties of the Ca²⁺ puffs evoked by photolysis of ci-IP₃. The results demonstrate that over about a 60-fold range of expression there is a positive correlation between SNAP-IP₃R3 expression and the frequency of Ca²⁺ puffs (Fig. 4A): Ca²⁺ puffs are more frequent in cells with more IP₃Rs. The latency to the first detected Ca²⁺ puff after the photolysis flash decreased as SNAP-IP₃R3 expression increased (Fig. 4B). However, the mean properties of individual Ca²⁺ puffs (amplitude, rise and decay times, and durations) were similar at all levels of SNAP-IP₃R3 expression (Fig. 4, C–F).

Figure 4.

Ca²⁺puffs are more frequent in cells over-expressing IP₃R, but the properties of individual Ca²⁺puffs are unaffected.A, relationship between SNAP-647 fluorescence intensity (recorded from the 19.2 × 19.2 μm region where Ca²⁺ puffs were recorded, ROI^puffs) and the frequency of Ca²⁺ puffs evoked by photolysis (250 ms) of ci-IP₃. Regression line shows a positive correlation between the frequency of Ca²⁺ puffs and SNAP-IP₃R3 expression. Pearson correlation coefficient r = 0.91, p < 0.0001. Results show 33 cells from five experiments. B, relationship between SNAP-IP₃R3 expression and latency to the first Ca²⁺ puff detected after the UV flash; r = −0.37, p < 0.05. C–F, relationships between SNAP-IP₃R3 expression and mean puff amplitude (C), mean rise time (D), mean decay time (E), and mean duration (F) of individual Ca²⁺ puffs. Results show 33 cells from five experiments. FU, fluorescence unit; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded.

We next asked how IP₃R are distributed within cells expressing different numbers of IP₃R. We restricted this analysis to HEK-SNAP-IP₃R3 cells in which the SNAP-647 fluorescence intensity measured from ROI^puffs was <20,000 fluorescence units because at higher levels of expression, it was impossible to reliably distinguish individual puncta. The results demonstrate that both the number of SNAP-IP₃R3 puncta (Fig. 5A and B) and their mean fluorescence intensity (Fig. 5, A and C) were increased in cells expressing more SNAP-IP₃R3. The rightward shifts in the fluorescence intensity distributions for individual puncta as IP₃R expression increased indicates that when cells express more IP₃Rs, most puncta contain more IP₃Rs (Fig. S4). This analysis indicates that clusters of IP₃Rs within the TIRF field are more abundant and each cluster includes more IP₃Rs in cells with more IP₃Rs.

Figure 5.

IP₃R clusters are more abundant and contain more IP₃Rswith increased expression of IP₃R.A, live-cell TIRF images of HEK-SNAP-IP₃R3 cells expressing low (i) or high (ii) levels of SNAP-IP₃R3. Scale bars represent 5 μm. Images were captured under identical conditions and with identical display values. B and C, relationships between overall SNAP-IP₃R3 expression (SNAP-647 fluorescence intensity measured from ROI^puffs) and the number of SNAP-IP₃R3 puncta (B, Pearson correlation coefficient r = 0.71, p < 0.001) and mean fluorescence intensity of individual puncta (C, r = 0.98, p < 0.0001). Results show 23 cells from four experiments. Analysis was restricted to cells in which puncta could be resolved (SNAP-647 fluorescence <20,000 FU). D, fluorescence intensity distributions for individual puncta from cells expressing the lowest (blue) and highest (red) levels of SNAP-IP₃R3 (SNAP-647 fluorescence = 1133 ± 350 and 15,226 ± 1950 FU, respectively; n = 3 cells for each category). Results show number of puncta identified within ROI^puffs (mean ± S.D., n = 3). Arrows indicate the brightest 30% of puncta in cells with fewest IP₃Rs and the dimmest 30% in cells with most IP₃Rs (about 30% of puncta are expected to be active Ca²⁺ puff sites, Fig. S5C). The lack of overlap between these two populations indicates that Ca²⁺ puffs in cells with most IP₃Rs are evoked by clusters that contain more IP₃Rs. FU, fluorescence unit; HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded; TIRF, total internal reflection fluorescence.

These results demonstrate that across a wide range of SNAP-IP₃R3 expression (∼60-fold), the properties of individual Ca²⁺ puffs, including their mean amplitudes, are preserved, but they occur after shorter latencies and they are more frequent in cells with more IP₃Rs. Since only a fraction of IP₃R puncta are licensed to respond (7), we had to consider whether in cells with most IP₃Rs, a few puncta with a normal complement of IP₃R might lurk beneath the substantial increase in the average number of IP₃Rs per punctum and thereby explain the unaltered properties of individual Ca²⁺ puffs. To address this issue, we recognize that about 30% of IP₃R puncta are competent to evoke Ca²⁺ puffs (Fig. S5) (7) and therefore compared the brightest 30% of puncta in cells with the fewest SNAP-IP₃R3 to the dimmest puncta in cells with most SNAP-IP₃R3. The results demonstrate that in cells with most IP₃Rs, the number of puncta with fluorescence intensities comparable to the brightest puncta of cells with fewest IP₃Rs is far too low to underlie the observed number of sites at which Ca²⁺ puffs occur (Fig. 5D). This analysis demonstrates that in cells expressing many IP₃Rs, puncta that contain more IP₃Rs evoke Ca²⁺ puffs of unaltered amplitude.

We conclude that the amplitude of Ca²⁺ puffs, indicating the number of open IP₃Rs (4, 5, 6, 7), is similar in cells expressing very different numbers of IP₃Rs. Since IP₃R clusters are larger in cells with more IP₃Rs, functional interactions between IP₃Rs must determine the number of IP₃Rs that contribute to a Ca²⁺ puff. This conclusion prompted further analysis of the mechanisms that might terminate the activity of IP₃R during a Ca²⁺ puff.

IP₃R with reduced affinity for IP₃ generate briefer Ca²⁺ puffs

Since the mechanisms that terminate Ca²⁺ puffs are unresolved (15, 28), we asked whether IP₃ dissociation from IP₃R contributes to termination. A point mutation (R568Q) within the conserved IP₃-binding core (IBC) of IP₃R1 causes a 9-fold reduction of its affinity for IP₃ (29). We introduced the same mutation into SNAP-IP₃R3 and used it to explore the effects of increasing the rate of IP₃ dissociation on Ca²⁺ puffs (Fig. 6A). We confirmed that SNAP-IP₃R3^RQ had the same subcellular distribution as SNAP-IP₃R3 (Fig. 6B). Since the IBC is highly conserved in all IP₃R subtypes and each IBC binds IP₃ with the same affinity (30), we expect the IP₃R3^RQ mutation to exactly replicate the 9-fold decrease in affinity reported for IP₃R1^RQ (29). The assumption is supported by evidence that the frequency of Ca²⁺ puffs evoked by photolysis of ci-IP₃ is reduced in cells expressing IP₃R3^RQ (Fig. 6C).

Figure 6.

Ca²⁺puffs are less frequent in cells expressing IP₃R with reduced affinity for IP₃, but their amplitude is unaffected.A, structure of human IP₃R3 showing R568 in the IP₃-binding core (IBC) coordinating IP₃ (Protein Data Bank, 6DQN) (23). Mutation of R568Q within the IBC of SNAP-IP₃R3 reduces its affinity for IP₃. B, live-cell TIRF image of a HEK-SNAP-IP₃R3^RQ cell, enlarged in lower panel. Scale bars represent 5 μm. The punctate distribution is similar to that of SNAP-IP₃R3 (Figs. 1C, 2, A and D, and 3A). C–E, relationships between SNAP-IP₃R3^RQ expression (from ROI^puffs) and the frequency of Ca²⁺ puffs evoked by photolysis of ci-IP₃ (C) (r = 0.75, p < 0.0001), latency to the first Ca²⁺ puff after the photolysis flash (D), and the mean amplitude of Ca²⁺ puffs (E) (slopes not significantly different from 0). Results are from 32 cells from five experiments. Linear regression analyses for SNAP-IP₃R3 (from Fig. 4, A–C) show that frequencies are lower, latencies are longer, and mean amplitudes of Ca²⁺ puffs are unaffected in SNAP-IP₃R3^RQ cells. FU, fluorescence unit; HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded; TIRF, total internal reflection fluorescence.

We expected that reducing the affinity of an IP₃R for IP₃ while maintaining the submaximal concentration of i-IP₃ delivered after the UV flash to reduce the number of IP₃R occupied by i-IP₃ and so reduce the frequency of Ca²⁺ puffs and increase the latency before the first observed Ca²⁺ puff. Our results confirm these expectations. As with SNAP-IP₃R3, the frequency of Ca²⁺ puffs increased with increased expression of SNAP-IP₃R₃^RQ, the latencies shortened with increased expression, and the mean amplitude of individual Ca²⁺ puffs was unaffected by the overall level of SNAP-IP₃R3^RQ expression (Fig. 6, C–E). However, across a wide range of IP₃R3 expression levels, the frequencies of Ca²⁺ puffs were lower, and the latencies were longer for SNAP-IP₃R3^RQ than for SNAP-IP₃R3 (Fig. 6, C and D).

To allow direct comparisons of cells expressing SNAP-IP₃R3 and SNAP-IP₃R3^RQ, we pooled data from cells in which the SNAP-647 fluorescence intensity measured from ROI^puffs fell within a defined range (1800–5900 FU), which coincided with cells in which the subcellular distribution of IP₃R3 resembled that of native IP₃Rs. The pooled data confirmed that the average level of IP₃R3 expression (Fig. 7A) and mean puff amplitude (Fig. 7C) were indistinguishable for HEK-SNAP-IP₃R3 and HEK-SNAP-IP₃R3^RQ cells, but the frequency of Ca²⁺ puffs was significantly reduced in the HEK-SNAP-IP₃R3^RQ cells (Fig. 7B).

Figure 7.

IP₃R with reduced affinity for IP₃evoke Ca²⁺puffs that terminate more quickly.A, SNAP-647 fluorescence intensity measured from ROI^puffs for SNAP-IP₃R3 and SNAP-IP₃R3^RQ cells selected for analyses of kinetics of Ca²⁺ puffs. Selection criteria (1800–5900 FU) were chosen to include cells with near-endogenous expression of IP₃R3. B and C, frequency (B) and mean amplitude (C) of Ca²⁺ puffs evoked by photolysis of ci-IP₃ in cells that satisfied the selection criteria. D–F, mean rise times (D), decay times (E), and durations (F) of Ca²⁺ puffs in the analyzed cells, where Ca²⁺ puffs with ‘square’ temporal profiles were excluded (Fig. S6, A and B). Results show mean ± S.D. (A and B) or S.E.M. (C–F) for 11 cells from four experiments for SNAP-IP₃R3 or 11 cells from five experiments for SNAP-IP₃R3^RQ. ns p > 0.05, ∗∗p < 0.01, ∗∗∗∗p < 0.0001, unpaired Student’s t test. FU, fluorescence unit; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded.

We next compared the kinetic properties of individual Ca²⁺ puffs in cells expressing SNAP-IP₃R3 or SNAP-IP₃R3^RQ. Automated analysis of the rise times of Ca²⁺ puffs is distorted when Ca²⁺ puffs have a sustained plateau phase (‘square’ puffs) (15) because stochastic fluctuations in fluorescence during the plateau can delay the peak amplitude and thereby erratically extend the computed rise time (Fig. S6A). We therefore excluded ‘square’ Ca²⁺ puffs from our analyses of the kinetics of individual Ca²⁺ puffs. The exclusion rate was similar (∼18%) for analyses of SNAP-IP₃R3 and SNAP-IP₃R3^RQ (Fig. S6B), confirming that it did not bias our analyses. We note in passing that although we have not specifically addressed the mechanisms underlying ‘square’ Ca²⁺ puffs, their similar abundance in cells expressing IP₃R3 and IP₃R3^RQ indicates that ‘square’ Ca²⁺ puffs are not selectively affected by changes in IP₃R affinity. Rise times were indistinguishable in cells expressing SNAP-IP₃R3 and SNAP-IP₃R3^RQ (Fig. 7D), but decay times and the duration of Ca²⁺ puffs were significantly shorter for SNAP-IP₃R3^RQ (Figs. 7, E and F and 6, C–E).

If the effects of reducing the affinity of the IP₃R for IP₃ were solely attributable to reduced occupancy of IP₃R by IP₃, we would expect Ca²⁺ puffs evoked by a 9 to 10-fold lower concentration of i-IP₃ through normal IP₃R to have the same properties as Ca²⁺ puffs evoked by IP₃R^RQ. We therefore compared Ca²⁺ puffs evoked by photolysis of ci-IP₃ in HEK-SNAP-IP₃R3^RQ cells with the usual UV flash duration (250 ms) to those evoked in HEK-SNAP-IP₃R3 cells exposed to a 10-fold shorter flash (25 ms). The frequency of the Ca²⁺ puffs and the relationship between IP₃R expression and frequency were similar under the two conditions (Fig. 8, A and B). These results confirm that for the frequency of Ca²⁺ puffs, a 9 to 10-fold reduction in IP₃-binding affinity can be compensated by a comparable increase in i-IP₃ concentration, suggesting that steady-state occupancy of IP₃R by IP₃ may determine the frequency of Ca²⁺ puffs. As expected, the amplitudes of individual Ca²⁺ puffs recorded under the two conditions were indistinguishable (Fig. 8C). The more important result is that neither the shorter duration of Ca²⁺ puffs nor the faster decay times in cells expressing IP₃R3^RQ were reproduced by exposing IP₃R3 to a 10-fold lower concentration of i-IP₃ (Fig. 8, D–F). Our results show that in cells expressing SNAP-IP₃R3, all measured properties of individual Ca²⁺ puffs are indistinguishable whether evoked by low concentrations of i-IP₃ (25-ms flash) or a 10-fold higher concentration (250-ms flash), but the decay times and durations are longer than for IP₃R with reduced affinity for IP₃ (SNAP-IP₃R3^RQ) (Fig. 8G).

Figure 8.

Shortened Ca²⁺puffs from IP₃R3^RQare not a consequence of reduced occupancy of IP₃R by IP₃.A, relationship between SNAP-IP₃R3 expression (SNAP-647 fluorescence intensity measured from ROI^puffs) and the frequency of Ca²⁺ puffs recorded after photolysis of ci-IP₃ from HEK-SNAP-IP₃R3 cells stimulated with a brief UV flash (25 ms). Results are from 30 cells from four experiments; Pearson correlation coefficient r = 0.60, p < 0.001. Regression lines (from Figs. 4A and 6C) show results from similar analyses of SNAP-IP₃R3 and SNAP-IP₃R3^RQ stimulated with a 10-fold longer flash (250 ms). B–F, frequency of Ca²⁺ puffs (B), and mean amplitude (C), rise time (D), decay time (E), and duration (F) for individual Ca²⁺ puffs. Results (mean ± S.E.M., or S.D. for B) are from 11 cells from four experiments for SNAP-IP₃R3 (25-ms UV) or 11 cells from five experiments for SNAP-IP₃R3^RQ (250-ms UV). Selection criteria (SNAP-647 fluorescence intensity of 1800–5900 FU, measured from ROI^puffs) were chosen to include cells with near-endogenous expression of IP₃R3 (B–F), and square Ca²⁺ puffs were excluded from analyses of kinetics (D–F). ns p > 0.05, ∗p < 0.05, ∗∗p < 0.01, unpaired Student’s t test. G, overall summary (individual values, mean ± S.E.M.) shows the kinetic properties of Ca²⁺ puffs evoked by flash photolysis of ci-IP₃ with the indicated flash durations for HEK-SNAP-IP₃R3 or HEK-SNAP-IP₃R3^RQ cells (compiled from Figs. 7, D–F and 8, D–F). FU, fluorescence unit; HEK, human embryonic kidney; IP₃R, inositol 1,4,5-trisphosphate receptor; ROI^puffs, ROI from which Ca²⁺ puffs were recorded.

Discussion

We developed methods that allowed homomeric IP₃R3 to be expressed in HEK cells at levels comparable to expression of endogenous IP₃R and, by attaching a SNAP-tag to the expressed IP₃R3, we were able to reveal both the subcellular distribution of IP₃R3 and the Ca²⁺ puffs they evoke in response to photolysis of ci-IP₃ (Figs. 1 and 2). The expressed SNAP-IP₃R3 and native IP₃R3 evoked indistinguishable Ca²⁺ puffs (Fig. 3).

We established that cells with more IP₃Rs have more IP₃R puncta within the TIRF field, each punctum included more IP₃Rs (Fig. 5), and Ca²⁺ puffs occurred more frequently and after shorter latencies. However, the properties of individual Ca²⁺ puffs were indistinguishable at all levels of IP₃R expression (Fig. 4). We confirmed, by analysis of the fluorescence intensity distributions of puncta in cells expressing different numbers of IP₃Rs, that the mean peak amplitude of a Ca²⁺ puff remained constant even as the number of IP₃Rs within a cluster increased (Figs. 4 and 5). These observations demonstrate that functional interactions between IP₃Rs within a cluster must constrain the number of IP₃Rs activated during the rising phase of a Ca²⁺ puff.

Local increases in [Ca²⁺]_c are thought to contribute to both rapid recruitment of IP₃R activity during the rising phase of a Ca²⁺ puff (through costimulation of IP₃R by Ca²⁺ and IP₃) and to termination of Ca²⁺ puffs (through delayed inhibition of IP₃R by Ca²⁺). It is, therefore, relevant that the biphasic effects of [Ca²⁺]_c on IP₃R is itself regulated by IP₃. We, for example, have suggested that IP₃ binding to IP₃R primes them to open by facilitating the binding of Ca²⁺ to a stimulatory site; while in the absence of IP₃, Ca²⁺ binds to an inhibitory site that prevents channel opening (18, 19, 20). Others suggest variants of this simple scheme that also have IP₃ regulating the effects of Ca²⁺ on IP₃R gating (1, 31). The key point is that the rate of IP₃ dissociation from an active IP₃R may influence its susceptibility to Ca²⁺ inhibition. We therefore examined the effect of manipulating the rate of IP₃R dissociation on the properties of Ca²⁺ puffs. We demonstrated that IP₃R3 with about a 10-fold lower affinity for IP₃ (SNAP-IP₃R3^RQ) evoked Ca²⁺ puffs in response to photolysis of ci-IP₃. As expected, the Ca²⁺ puffs mediated by SNAP-IP₃R3^RQ were less frequent than those mediated by SNAP-IP₃R3, consistent with the steady-state occupancy of IP₃R by IP₃ determining the number of active IP₃Rs (Fig. 6). More importantly, while the mean peak amplitudes of the Ca²⁺ puffs were similar for SNAP-IP₃R3 and SNAP-IP₃R3^RQ, indicating similar numbers of open IP₃Rs, the Ca²⁺ puffs decayed more rapidly for SNAP-IP₃R3^RQ (Fig. 7). Furthermore, the disparity in decay times between SNAP-IP₃R3 and SNAP-IP₃R3^RQ remained when the intensity of the photolysis flash (and so the intracellular concentration of i-IP₃) was reduced for SNAP-IP₃R3 cells to provide matched frequencies of Ca²⁺ puffs in the two cell lines (indicative of comparable IP₃R occupancies by i-IP₃) (Fig. 8). These analyses establish that the affinity of an IP₃R for IP₃ affects the rate at which Ca²⁺ puffs terminate: reducing the affinity for IP₃ (increasing the rate of IP₃ dissociation) causes Ca²⁺ puffs to terminate more quickly. We draw two important conclusions from these data. Firstly, the rate of IP₃ dissociation from IP₃R contributes to termination of Ca²⁺ puffs. Various mechanisms might align with this conclusion, but it is consistent with a local increase in [Ca²⁺]_c contributing to termination by closing IP₃R but only after IP₃ has dissociated from active IP₃R (19, 20).

Our second conclusion relates to whether the same mechanism might account for the constraint on the number of IP₃Rs recruited within a cluster during the rising phase of a Ca²⁺ puff and the closure of active IP₃Rs during the falling phase. The key observation here is that although Ca²⁺ puffs evoked by IP₃R with reduced affinity for IP₃ (SNAP-IP₃R3^RQ) terminate more quickly, their mean peak amplitude is indistinguishable from that of Ca²⁺ puffs evoked by normal IP₃Rs (Figs. 7 and 8). This implies that different mechanisms constrain the opening of IP₃Rs during the rising phase of a Ca²⁺ puff and closure of open IP₃Rs during the falling phase. A speculative explanation is that during the rising phase of a Ca²⁺ puff, Ca²⁺ released by active IP₃Rs rapidly inhibits neighboring IP₃Rs that have not bound IP₃, while during the falling phase, dissociation of IP₃ (perhaps from only a single IP₃R subunit) causes both channel closure and its immediate susceptibility to Ca²⁺ inhibition, which might then prevent reopening.

We conclude that when presented with more IP₃Rs, cells incorporate more of them into clusters, assemble more IP₃R clusters, and generate more frequent Ca²⁺ puffs in response to IP₃, but the properties of individual Ca²⁺ puffs, including their mean peak amplitudes, are unchanged. We suggest that different mechanisms constrain recruitment of IP₃Rs during the rising phase of a Ca²⁺ puff and cause closure of IP₃Rs during the falling phase; only the latter is affected by the rate of IP₃ dissociation.

Experimental procedures

Materials

Cal520-AM and Calbryte590-AM were from AAT Bioquest. 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) was from SiChem. 35-mm glass-bottom imaging dishes (14-mm micro-well, #1 cover glass) were from Cellvis (IBL Baustoff+Labor GmbH). EGTA-AM, human plasma fibronectin, Pluronic F-127, and doxycycline hydrochloride were from Merck. SNAP-Cell 647-SiR, BsrGI-HF restriction enzyme, Q5 High-Fidelity Master Mix, and Gibson Assembly Master Mix were from New England Biolabs. pcDNA3.2/V5-DEST, pDONR221, pcDNA6/TR and pT-Rex-DEST30 vectors, LR and BP Clonase II enzyme mixes, Pfl23II (a.k.a. BsiWI) and CpoI (a.k.a. RsrII) restriction enzymes, and custom oligonucleotides were from Thermo Fisher Scientific.

Sources of additional materials are provided in relevant sections of Experimental procedures.

Generation of SNAP-IP₃R3 constructs

A plasmid encoding human IP₃R3 was obtained from Harvard PlasmID Database (Clone ID: HsCD00399229, GenBank Accession: BC172406) and cloned from pENTR223 into pcDNA3.2/V5-DEST using LR Gateway cloning. We tagged human IP₃R3 with a SNAP-tag, a self-labeling enzyme tag based on the human DNA repair protein, O⁶-alkylguanine-DNA-alkyltransferase, which reacts with O⁶-benzylguanine derivatives conjugated to fluorescent dyes (32). To generate the N-terminally SNAP-tagged human IP₃R3 construct (SNAP-IP₃R3), PCR was used to amplify an IP₃R3 fragment (forward primer: CTCGGCGGTGGTTCTGGTGGTGGTTCTGGTATGAGTGAAATGTCCAGC, reverse primer: GAACCGCGGGCCCTCTAGATCAACCACTTTTGTACAAGAAAGCTGGGC), and BsrGI restriction digest was used to generate a backbone fragment. Gibson assembly was used to introduce a DNA string (GeneArt Strings, Thermo Fisher Scientific) encoding SNAPf (fast-labeling SNAP-tag) with a GGSGGGSG peptide linker between the backbone fragment and the IP₃R3 fragment (Fig. 1A). For expression of SNAP-IP₃R3 under control of a tetracycline-inducible promoter, SNAP-IP₃R3 was transferred from pcDNA3.2 to pDONR221 and then to pT-REx-DEST30 using BP and LR Gateway cloning, respectively.

To generate the construct encoding IP₃R3 with reduced affinity for IP₃ (SNAP-IP₃R3^RQ, Fig. 6A), a DNA string was synthesized encoding a region of SNAP-IP₃R3 between two restriction sites, BsiWI and RsrII, which contained an Arg to Gln mutation at residue 568 (R568Q, CGC to CAG). A BsiWI/RsrII-restriction digest of SNAP-IP₃R3 in pDONR221 was used to generate a fragment encoding the remainder of the construct. Gibson assembly was used to assemble the two fragments, and SNAP-IP₃R3^RQ was transferred to pT-REx-DEST30 using LR Gateway cloning. Sequences of all coding regions were confirmed using Sanger sequencing (Source BioScience). A silent point mutation was detected in bases (ATC to ATA) encoding Ile²⁵¹² in both constructs.

Cell culture

HEK cells lacking all IP₃R subtypes (HEK-3KO, #EUR030) or expressing only IP₃R3 (HEK-IP₃R3, IP₃R1/2 knock-out, #EUR033) were generated by CRISPR/Cas9 (2) and obtained from Kerafast. HEK cells were cultured in Dulbecco’s modified Eagle’s medium/F-12 with Gluta-MAX (Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich). Cells were maintained at 37 °C in humidified air with 5% CO₂ and passaged every 3 to 4 days using Gibco TrypLE Express (Thermo Fisher Scientific). Regular screening confirmed that cells were free of mycoplasma.

To generate cells (HEK-SNAP-IP₃R3 cells) expressing SNAP-IP₃R under control of the Tet repressor (TetR), we used pcDNA6/TR, a plasmid encoding TetR, and the Invitrogen T-REx expression system. HEK-3KO cells were seeded in 25-cm² flasks and after 24 h, when they were ∼70% confluent, they were transiently transfected using TransIT-LT1 Transfection Reagent (3 μl/μg DNA; Mirus Bio) with SNAP-IP₃R3 in pT-REx-DEST30 (2.6 μg/T25 flask) and pcDNA6/TR (15.8 μg). After 24 and 48 h, medium was exchanged for fresh complete medium containing doxycycline (1 μg/ml) to block activity of TetR and derepress the expression of SNAP-IP₃R3 (Fig. 1B).

Labeling of cells with SNAP-tag

HEK-SNAP-IP₃R3 cells were used 72 h after transfection for labeling of SNAP-IP₃R3 with SNAP-Cell 647-SiR, a cell-permeable, far-red fluorescent label (excitation 645 nm; emission 661 nm). To avoid background labeling, it was essential to label cells in suspension before reattaching them to imaging dishes for microscopy (Fig. 1, C–E). Transfected cells were detached using TrypLE Express, resuspended in fresh complete medium (3 × 10⁵ cells in 300 μl) containing SNAP-Cell 647-SiR (1 μM), and incubated for 15 min at 37 °C with 5% CO₂. Cells were then washed three times (650g, 2 min), resuspended in fresh complete medium, plated into a 35-mm glass-bottomed imaging dish (#1 cover glass) coated with 10 μg/ml human fibronectin, and used after 12 to 14 h (Fig. 1B).

Fluorescence microscopy

Fluorescence microscopy used an inverted Olympus IX83 microscope equipped with a 100× oil-immersion TIRF objective (numerical aperture, NA = 1.49) or (for Fig. S1) a 60× oil-immersion objective (NA = 1.45). Illumination for TIRF microscopy was via a Cairn MultiLine LaserBank (488 and 647 nm) and an iLas2-targeted laser illumination system (Cairn Research), through which the excitation angle for TIRF was adjusted to achieve a theoretical penetration depth of 80 to 100 nm. Excitation light of 488 or 647 nm was passed through a quad-band filter set (TRF89902-EM, Chroma Technology), and emitted light was passed through an appropriate band-pass filter within a high-speed filter wheel (Cairn Optospin; peak/bandwidth: 525/50 or 700/75 nm). A 395-nm LED (SPECTRA X-light engine, Lumencor) was used for flash photolysis of ci-IP₃. Detection of emitted light was via an EMCCD camera (iXon Ultra 897, Andor; 512 × 512 pixels; pixel size = 0.16 μm at 100× magnification). Image capture used MetaMorph Microscopy Automation and Image Analysis Software (version 7.10.1.161, Molecular Devices), through which image capture (control of shutters, etc.) was fully automated. Both live and fixed cells were imaged at 20 °C.

Immunostaining

About 14 h after labeling, HEK-SNAP-IP₃R3 cells in imaging dishes were fixed (20 min) with 4% paraformaldehyde (Alfa Aesar) in PBS (Thermo Fisher Scientific), washed twice in PBS, permeabilized (5 min) with Triton X-100 (0.25% in PBS, Thermo Fisher Scientific), washed twice in PBS, and blocked (1 h) in PBS with 5% bovine serum albumin (BSA; Europa Bioproducts). Cells were then incubated (12 h, 4 °C) with primary monoclonal mouse anti-IP₃R3 antibody (Ab-IP₃R3, 1:200, BD Biosciences, #610313, RRID: AB_397705) in PBS with 3% BSA, washed three times in PBS, and incubated (1 h, 20 °C) with secondary rabbit anti-mouse antibody conjugated to Alexa Fluor 488 (1:400, Thermo Fisher Scientific, #A11059, RRID: AB_2534106) in PBS with 3% BSA for 1 h. Cells were then washed five times in PBS before imaging. All steps were performed under conditions that minimized exposure to light. We confirmed the specificity of Ab-IP₃R3 by demonstrating the absence of immunostaining in HEK-3KO cells (Fig. S1).

Quantification of fluorescence images

All TIRF images were first background-corrected by subtracting the gray value of an area from outside a cell using MetaMorph or FIJI (https://fiji.sc/) (33). To quantify the expression of SNAP-IP₃R3, fluorescence (647 nm) was measured as the mean gray value from a region (19.2 × 19.2 μm) within which Ca²⁺ puffs were recorded. We refer to this measure of SNAP-IP₃R3 expression as SNAP-647 fluorescence from ROI^puffs. To allow comparisons between experiments, fluorescence (F_TetraSpeck) from TetraSpeck beads (Thermo Fisher Scientific, excitation 660 nm, emission 680 nm) was recorded and used to normalize measurements of SNAP-IP₃R3 fluorescence (F_SNAP). For experiments addressing the effects of varying SNAP-IP₃R3 expression (Figure 4, Figure 5, Figure 6, Figure 7, Figure 8), normalized values (10, 000 × F_SNAP/F_TetraSpeck) were used for quantitative analyses.

For comparisons between IP₃R3 immunostaining (488 nm) and SNAP-IP₃R3 labeling (647 nm) (Fig. 2C), the mean gray value from a ROI demarcating the entire TIRF footprint of the cell was measured. Pixel-based colocalization analysis used the FIJI plugin JACoP (34). Images were manually thresholded before computing Manders’ split coefficients (M); M1 reports the fraction of IP₃R3 immunofluorescence coinciding with SNAP-IP₃R3 fluorescence, and M2 reports the fraction of SNAP fluorescence coincident with immunofluorescence (Fig. 2B). M values of 1.0 indicate perfect colocalization. Costes’ randomizations (100 iterations, pixel-block size = 5) were generated for each image and used to assess the statistical significance of any colocalization (35).

For fluorescence intensity measurements of individual puncta, ROI (19.2 × 19.2 μm) were background-subtracted (rolling ball = 50 pixel) in FIJI to improve signal-to-noise. The FIJI plugin TrackMate (v 6.0.1) was used for automated detection of IP₃R3 immunofluorescent puncta (Fig. 2F) or SNAP-IP₃R3 puncta (Fig. 5, B and C) (36).

Within TrackMate, we used an estimated punctum size of 600 nm for spot detection and applied a Laplacian of Gaussian filter to detect spot maxima with sub-pixel accuracy. Briefly, the Gaussian filter smooths noise, while the Laplacian filter is used to identify edges (36). Detected spots were filtered by ‘Quality’ until only genuine puncta (confirmed by manual inspection) were detected; the selection criteria require a punctum to both exceed a critical fluorescence intensity and to fall below a critical size (diameter < 600 nm). To compare fluorescence from immunostaining and SNAP-tag labeling (Fig. 2F), the coordinates of the pixel with the maximum fluorescence intensity for each detected spot in the IP₃R3 immunofluorescence image were mapped onto the corresponding pixel of the SNAP-IP₃R3 image. We used these point fluorescence intensity measurements to compare expression determined by immunostaining and SNAP-tag labeling. The approach is a valid measure of the fluorescence intensity of the entire punctum because the point intensity is linearly related to total intensity of the spot for the IP₃R3 immunofluorescence image (Fig. S2).

High-resolution imaging of Ca²⁺ puffs

To allow imaging of Ca²⁺ puffs, cells were loaded with a Ca²⁺ indicator (Cal520), EGTA to restrict Ca²⁺ diffusion (4), and ci-IP₃ to allow controlled release of intracellular i-IP₃ by photolysis. The slow on-rate for Ca²⁺ binding to EGTA is sufficient to allow EGTA to buffer Ca²⁺ diffusing between Ca²⁺ puff sites, which are typically a few μm apart. EGTA thereby restrains propagation of global Ca²⁺ signals, but the on-rate is too slow for EGTA to intercept Ca²⁺ diffusing between IP₃Rs within sites (200–400 nm) (4, 5, 6, 7). EGTA does not, therefore, significantly affect the properties of individual Ca²⁺ puffs (4, 37, 38). HEK-SNAP-IP₃R3 cells in imaging dishes were washed with Hepes-buffered saline (HBS: 135 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl₂, 1.5 mM CaCl₂, 11.5 mM glucose, 11.6 mM Hepes, pH 7.3; Merck), incubated (1 h) with Cal520-AM (5 μM) and ci-IP₃/PM (1 μM) in HBS with pluronic acid (0.02%, v/v). Cells were then washed and incubated with EGTA-AM (5 μM) in HBS with pluronic acid (0.02%). After 45 min, cells were washed and incubated in HBS for a further 30 min before imaging. All steps were performed at 20 °C with minimal exposure to light.

TIRF microscopy was used to record Cal520 fluorescence (488 nm) within ∼100 nm of the plasma membrane. To achieve the high temporal resolution required to resolve the kinetics of Ca²⁺ puffs, image streams were captured at 200 frames s^-1 (5-ms intervals) from a 19.2 × 19.2 μm ROI (∼25–50% of the TIRF footprint of a cell, referred to as ROI^puffs). Image stacks typically comprised 6000 to 10,000 frames (∼30–50 s). After 2 s of recording, a UV flash lasting 250 ms (25 ms in parts of Fig. 8) was delivered uniformly to the entire field to uncage ci-IP₃. A 50:50 mirror in the light path allowed simultaneous excitation by the 488-nm laser and 395-nm LED. Image stacks were streamed directly to RAM in MetaMorph and then exported as.tif files.

Detection and analysis of Ca²⁺ puffs

Raw image stacks were background-corrected in MetaMorph and frames containing the UV flash artefact were removed in FIJI. Automated detection and analysis of Ca²⁺ puffs were performed using the detect_puffs plugin in FLIKA, an open-source Python-based image processing and analysis software (39, 40). An F/F₀ stack was generated by dividing the fluorescence intensity of each pixel from each frame (F) by the average intensity of that pixel over the first 390 frames before photorelease of i-IP₃ (F₀). Image stacks (F/F₀) were then divided by the SD of the baseline F/F₀ images, a Gaussian blur (σ = 2) was applied, and these stacks were then used to identify Ca²⁺ puffs and measure their properties.

When a global increase in fluorescence obscured Ca²⁺ puffs, a Butterworth temporal bandpass filter (low frequency cut-off = 0.01) was applied to remove the slow increase in baseline fluorescence. In some cases, global increases in [Ca²⁺]_c completely obscured Ca²⁺ puffs causing premature termination of the analysis; the shortest of these terminated recordings was ∼12 s. The recording interval was defined as the time from the first Ca²⁺ puff to either the end of the recording or the time when it became impracticable to resolve Ca²⁺ puffs against an elevated global [Ca²⁺]_c.

FLIKA uses a threshold-cluster algorithm (41) to identify pixels brighter than a user-defined threshold (0.25 for our analysis). The brightest pixels are considered cluster centers, and adjacent bright pixels are included in the cluster. Each cluster represents a Ca²⁺ puff, to which a 2D Gaussian is fitted to identify its centroid. The centroid, defined with sub-pixel resolution, reports the center-of-mass of the clustered IP₃Rs underlying the Ca²⁺ puff (37). The centroids of all Ca²⁺ puffs were mapped onto the TIRF footprint of the cell. To measure the amplitudes and kinetics of individual Ca²⁺ puffs, an F/F₀ trace was produced from a small ROI (1.76 × 1.76 μm) centered on the centroid of the Ca²⁺ puff. Detected puffs were manually inspected to exclude noisy traces and to ensure the entire F/F₀ trace was included in the analysis. In some cases, ‘square’ puffs (detected by eye) were manually removed from analyses of kinetic properties (Fig. S6). Files containing puff properties generated by FLIKA were exported to Microsoft Excel for further analysis.

Identification of Ca²⁺ puff sites

Ca²⁺ puffs recur at the same stable sites over many minutes, reflecting Ca²⁺ release from immobile clusters of IP₃Rs (7, 8). In our analyses, Ca²⁺ puffs that lie within 1 μm (6.25 pixels) of each other were considered to belong to the same Ca²⁺ puff site. FLIKA successfully identifies these sites when Ca²⁺ puffs are relatively infrequent and sparsely distributed. However, when Ca²⁺ puffs are more frequent and more densely packed, FLIKA merges large numbers of Ca²⁺ puffs into fewer sites, causing the number of sites to be underestimated (Fig. S5, A and B). This occurs because FLIKA assigns Ca²⁺ puffs to the same site if each successively assessed Ca²⁺ puff is < 1 μm from the last; hence, a site can grow in a chain-like fashion as densely packed Ca²⁺ puff centroids become linked across considerable distances. To overcome this difficulty, we used a MATLAB script, ClusterXYpoints (version 1.5.0.0, https://www.mathworks.com/matlabcentral/fileexchange/56150-distance-based-clustering-of-a-set-of-xy-coordinates) (42), to identify Ca²⁺ puff sites. The XY coordinates of Ca²⁺ puff centroids generated by FLIKA were saved as .csv files and imported into MATLAB (version R2021a). ClusterXYpoints starts by sorting the XY coordinates of Ca²⁺ puffs (a.k.a. points) in ascending order. The first point is assigned as a cluster, and the cluster centroid is calculated. If the second point is < 1 μm from the first cluster centroid, it is added to the cluster and the centroid is recalculated. Subsequent points are added to the cluster if they are < 1 μm from the updated centroid of the cluster. If a point is > 1 μm from the current cluster centroid, a new cluster centroid is generated. The process continues until all Ca²⁺ puffs have been assigned to a site.

ClusterXYpoints detected the same sites as FLIKA when Ca²⁺ puffs were infrequent, but it identified more sites than FLIKA when Ca²⁺ puffs were more frequent. At the highest frequencies of Ca²⁺ puffs, when FLIKA merged most Ca²⁺ puffs into a single site, ClusterXYpoints continued to identify discrete sites (Fig. S5, A and B). We therefore used ClusterXYpoints to identify the number of Ca²⁺ puff sites. The problems encountered with ‘coalescing’ Ca²⁺ puff sites in FLIKA analyses arise only because the centroids of many temporally separated Ca²⁺ puffs are overlaid for analysis, they therefore have no impact on the analyses of individual Ca²⁺ puffs.

Epifluorescence imaging of cell populations

For measuring SNAP-IP₃R3 expression in cell populations (Fig. 1E), HEK-SNAP-IP₃R3 or mock-transfected HEK-3KO cells were labeled with SNAP-Cell 647-SiR. Cells were washed with HBS and loaded with Calbryte590-AM (2 μM) in HBS with pluronic acid (0.02%) for 1 h at 20 °C in darkness. Here, Calbryte590 is used to define cell boundaries rather than as a Ca²⁺ indicator. Cells were washed and incubated in HBS for a further 1 h before imaging.

Epifluorescence imaging used an EVOS M7000 Cell Imaging System equipped with a 10× objective and a CMOS camera with a large field of view (Thermo Fisher Scientific). Calbryte590 and SNAP-Cell 647-SiR fluorescence were captured using EVOS LED cubes for Texas Red and Cy5, respectively. For image analysis, Calbryte590 fluorescence was used to demarcate whole cells because it is bright and uniformly distributed in the cytosol, making it ideal for automated segmentation. A custom-written ImageJ macro was used for segmentation (available from H.A.S.); the detected cell boundaries were then confirmed by manual inspection. Raw mean SNAP-IP₃R3 fluorescence intensity was measured from individual whole-cell ROIs.

Statistical analyses

Most data are presented as mean ± SD or S.E.M. Numbers of cells, Ca²⁺ puffs, and independent analyses are reported in figure legends. Statistical comparisons used unpaired Student’s t tests; correlation analyses used Pearson’s correlation coefficient; comparisons of frequency distributions used χ² test for trend (GraphPad Prism 5, GraphPad). The statistical significance of colocalization analyses (Manders’ split coefficients) was determined after Costes’ randomization, where a Costes’ p value > 0.95 indicates that < 0.05 of the randomized images had a Pearson’s correlation coefficient value greater than the nonrandomized image (i.e., p < 0.05) (35).

Data availability

All data required to support the conclusions presented are contained with the article.

Supporting information

This article contains supporting information (Figs. S1–S6) (39, 40).

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Biography

Holly A. Smith is a graduate student in the Department of Pharmacology at the University of Cambridge, where she is completing her PhD with Professor Colin Taylor. Her research explores the function of IP₃ receptors and the Ca²⁺ signals they evoke using high-resolution fluorescence microscopy. She is particularly interested in understanding the mechanisms controlling activity of local Ca²⁺ signals known as Ca²⁺ “puffs”. Twitter: @HollyASmith_488

Edited by Roger Colbran