Research ArticleCalcium signaling

All three IP3 receptor isoforms generate Ca2+ puffs that display similar characteristics

See allHide authors and affiliations

Sci. Signal.  18 Dec 2018:
Vol. 11, Issue 561, eaau0344
DOI: 10.1126/scisignal.aau0344

Different isoforms, similar Ca2+ puffs

Cells express three different isoforms of the inositol trisphosphate receptor (IP3R), which underlie Ca2+ signals ranging from local puffs to global waves. Lock et al. used CRISPR-Cas9 gene editing to create HEK-293 cell lines that expressed individual IP3R isoforms. Despite their reported divergent functional properties, each isoform produced Ca2+ puffs with similar characteristics. Future work is required to determine how these conserved Ca2+ puffs give rise to different global Ca2+ signals.


Inositol 1,4,5-trisphosphate (IP3) evokes Ca2+ release through IP3 receptors (IP3Rs) to generate both local Ca2+ puffs arising from concerted openings of clustered IP3Rs and cell-wide Ca2+ waves. Imaging Ca2+ puffs with single-channel resolution yields information on the localization and properties of native IP3Rs in intact cells, but interpretation has been complicated because cells express varying proportions of three structurally and functionally distinct isoforms of IP3Rs. Here, we used TIRF and light-sheet microscopy to image Ca2+ puffs in HEK-293 cell lines generated by CRISPR-Cas9 technology to express exclusively IP3R type 1, 2, or 3. Photorelease of the IP3 analog i-IP3 in all three cell lines evoked puffs with largely similar mean amplitudes, temporal characteristics, and spatial extents. Moreover, the single-channel Ca2+ flux was similar among isoforms, indicating that clusters of different IP3R isoforms contain comparable numbers of active channels. Our results show that all three IP3R isoforms cluster to generate local Ca2+ puffs and, contrary to findings of divergent properties from in vitro electrophysiological studies, display similar conductances and gating kinetics in intact cells.


Cytosolic Ca2+ signals are used by all cells of the body to regulate cellular processes as diverse as gene transcription, secretion, mitochondrial energetics, electrical excitability, and fertilization, often more than one process in the same cell (1, 2). The capacity to precisely and specifically regulate cellular events is largely attributable to an exquisite control of the spatial and temporal patterning of cytosolic free [Ca2+] transients (2). This control is exemplified by the second messenger pathway mediated by inositol 1,4,5-trisphosphate (IP3). IP3 is generated in response to activation of cell surface G protein–coupled receptors and diffuses in the cytosol to bind to IP3 receptors (IP3Rs) in the membrane of the endoplasmic reticulum (ER), causing them to open and release Ca2+ ions sequestered in the ER lumen (3). The resulting cytosolic Ca2+ signals constitute a hierarchy of events, with increasing amounts of IP3 progressively evoking Ca2+ liberation from individual IP3Rs (Ca2+ blips) (4), local Ca2+ signals arising from clusters of several IP3Rs (Ca2+ puffs) (47), and global Ca2+ waves that propagate through the cell (710).

The patterning of IP3-mediated Ca2+ signals is determined both by the functional properties of IP3Rs and by their spatial arrangement in the ER membrane. Crucially, the opening of IP3R channels requires binding of cytosolic Ca2+ in addition to IP3, leading to a phenomenon of Ca2+-induced Ca2+ release (CICR) (11, 12), such that Ca2+ diffusing from one open channel may trigger the opening of adjacent channels. The clustered distribution of IP3Rs further shapes the extent of this regenerative process. CICR may remain restricted to a single cluster containing from a few to a few tens of functional IP3Rs to produce a puff, or a global Ca2+ wave may be generated by successive cycles of CICR and Ca2+ diffusion between clusters (7, 9). The transition between these modes depends on factors including IP3 concentration and the presence of cytoplasmic Ca2+ buffers that restrict the diffusion of Ca2+ ions (13, 14). Ca2+ puffs thus serve both as local signals in their own right and as the building blocks of global cellular Ca2+ signals.

In vertebrates, three different genes encode three main types of IP3Rs—IP3R1 (15), IP3R2 (16), and IP3R3 (17)—that cotranslationally oligomerize to form tetrameric Ca2+ release channels. The three isoforms have a similar monomeric molecular mass of ~300 kDa but share only 60 to 80% amino acid homology (18). Concordant with this diversity, different isoforms are reported to exhibit distinct functional properties. For example, their binding affinities for IP3 follow a rank order with IP3R2 > IP3R1 > IP3R3 (1921), and their differential modulation by cytosolic Ca2+ (20, 2224), adenosine 5′-triphosphate (21, 25), binding proteins (26, 27), and posttranslational modifications (28, 29) further shape IP3R behavior in a subtype-specific manner. Additional complexity arises from splice variants (3032), and because most cell types express two or three different isoforms (3336) that may assemble into heterotetramers (33, 37, 38) with properties that can resemble a blend of their constituents or that are dominated by an individual isoform, depending on cellular conditions (39, 40). It has thus been proposed that each IP3R isoform functions as a specific hub to determine different trajectories of cell signaling and that different cell types express and localize a particular complement of IP3R isoforms to suit their particular needs (41).

Because of the complex and poorly determined mix of IP3R isoforms in native cells, most studies of biochemical and functional properties of specific IP3R isoforms have used experimental systems predominantly or exclusively expressing a single isoform. These approaches include the use of cell types that predominantly express a particular isoform (11, 42, 43); overexpression of recombinant IP3R isoforms in cell types that express low levels of endogenous IP3Rs (21, 24, 44) and in DT40–3KO (triple-knockout) cells where all three IP3R types have been disrupted (45); and by overexpression in an IP3R-null human cell line [human embryonic kidney (HEK)–3KO] generated by CRISPR-Cas9 technology (46).

Information regarding the functional, single-channel properties of IP3Rs derived by these approaches has primarily been obtained by electrophysiological recordings in reduced experimental systems—including isolated cell nuclei (43, 4749) and IP3Rs reconstituted into lipid bilayers (11, 12, 21, 24)—where the native cellular environment and spatial organization are disrupted. In contrast, Ca2+ imaging by total internal reflection fluorescent (TIRF) microscopy can resolve the contributions of Ca2+ flux through individual IP3Rs, providing information on their single-channel properties and their localization in intact cells with minimal perturbation (50). However, to date, all intact-cell imaging studies of local Ca2+ signals have been performed in native cells expressing their endogenous complement of IP3Rs. Although many studies used cell types [such as Xenopus oocytes (4, 5, 5153) and SH-SY5Y neuroblastoma cells (50, 5458)] that predominantly express IP3R1, the actual composition of the IP3R channels that open to generate Ca2+ puffs cannot readily be inferred. For example, only about 3% of all the IP3Rs in SH-SY5Y cells appear to be involved in generating puffs (59, 60), so those events could arise through a minor population of type 2 or 3 IP3Rs, whereas the more numerous type 1 IP3Rs remain silent.

Here, we capitalize on the availability of HEK-KO cell lines in which pairs of IP3R isoforms were knocked out by CRISPR-Cas9 technology to generate cells natively expressing exclusively type 1, 2, or 3 IP3Rs (46). Using TIRF Ca2+ imaging, we found that all three isoforms independently produced Ca2+ puffs in response to ultraviolet (UV) flash photorelease of the IP3 analog i-IP3. Puffs evoked in cells expressing each isoform remained at fixed, stationary subcellular sites, as do puffs in native, wild-type (WT) cells (58). Thus, all three IP3R isoforms can bind to some, as yet unidentified, cytoskeletal anchoring sites that define the location and composition of functional IP3R clusters (58, 61). Moreover, we found substantial redundancy in the amplitudes, single-channel Ca2+ fluxes, and spatial distribution of local signals mediated by the three IP3R isoforms, although subtle differences in kinetics were apparent that may contribute to shaping the spatiotemporal patterning of cellular Ca2+ signals in native cells. A preliminary report of this study has been published in abstract form (62).


IP3R expression in WT and KO HEK-293 cell lines

To investigate the role of individual IP3R isoforms in local Ca2+ signaling processes in mammalian cells, we used HEK-293 WT cells that were engineered by CRISPR-Cas9 technology to establish lines that endogenously express only a single IP3R type—IP3R1, IP3R2, or IP3R3—by knocking out pairs of isoforms from WT cells that regularly express all three (46). As a control, we used a 3KO cell line in which no IP3Rs were expressed (HEK-3KO). Loss of protein expression was confirmed by Western blot using isoform-specific antibodies previously shown to effectively discriminate between IP3R types (40). All three isoforms were detected by their respective type-specific antibody in WT cells, and none of three antibodies revealed immunoreactivities in the 3KO lane, confirming the complete disruption of the three IP3R genes (Fig. 1A). In contrast, cells expressing type 1, 2, or 3 IP3Rs displayed immunoreactivity to only the appropriate isoform-specific antibody and were devoid of signal from the other two antibodies (Fig. 1A). Densitometry measurements indicated that the relative abundances of individual IP3R isoforms in single isoform–expressing cells were not appreciably different from their respective expression in WT cells (Fig. 1A). These findings agree with previous results demonstrating the disruption of specific IP3R alleles by corresponding single-guide RNA (sgRNAs) and collectively affirm that these CRISPR-Cas9–modified HEK-293 cell lines natively express single IP3R isoforms (46).

Fig. 1 Protein expression of IP

3R isoforms and global carbachol- and i-IP3–evoked Ca2+ signals in HEK WT cells and CRISPR-Cas9–modified HEK-293 cells that endogenously express exclusively type 1, 2, or 3 IP3Rs. (A) Equivalent amounts of lysate proteins from HEK WT, triple IP3R-knockout (3KO) and single IP3R isoform–expressing cells were immunoblotted with isoform-specific antibodies as indicated; data are representative of three independent experiments. Bar graphs show the relative expression of each IP3R isoform in single isoform–expressing cells normalized to their respective expression in WT cells. MW, molecular weight. (B to M) Fluorescence records and analyses were made from regions of interest encompassing individual cells loaded with Ca2+ indicator Cal-520. (B to E) Traces show representative, superimposed records of global fluorescence ratio signals (ΔF/F0) evoked by bath application of 100 μM CCH (denoted by line) in WT cells (B; n = 131) and in cells solely expressing type 1 (C; n = 130), type 2 (D; n = 186), or type 3 (E; n = 148) IP3Rs. (F) Bars show the mean peak amplitudes of the fluorescence signals (ΔF/F0) evoked by 100 μM CCH in WT and single isoform-expressing cells; n = 3 imaging fields for each cell line with cell totals of 392 (WT), 383 (IP3R1), 508 (IP3R2), and 494 (IP3R3). (G) Bars show the mean maximum rate of rise of the fluorescence signal (ΔF/F0 s−1), determined from 100 randomly chosen cells for each cell line with n = 3 imaging fields. (H to K) Superimposed traces of i-IP3–mediated global fluorescence signals (ΔF/F0) evoked by a 1000-ms duration photolysis flash (marked by an arrow) in WT cells (G; n = 134) and in cells solely expressing type 1 (H; n = 117), type 2 (I; n = 140), and type 3 (J; n = 112) IP3Rs. The delayed individual-cell signals in (H) and (J) are long-latency initial responses, rather than recurrent signals in cells that had already responded. (L) Dose-response curves of the mean peak amplitudes of the fluorescence signals (ΔF/F0) after photolysis flashes of different durations for WT cells and cells expressing single isoforms (depicted by different symbols and colors as indicated). For all cell lines, data points are means ± 1 SEM of n = 3 imaging fields with totals of 336 to 465 cells. Statistical comparisons between cell lines for a given UV flash duration were determined by analysis of variance (ANOVA) with Tukey post hoc tests: (300 ms) P < 0.01 for WT versus R1, WT versus R2, and WT versus R3; (700 ms) P < 0.01 for WT versus R1, WT versus R3, R1 versus R2, and R2 versus R3; (1000 ms) P < 0.01 for WT versus R1, WT versus R3, R1 versus R2, and R2 versus R3; (2000 ms) P < 0.01 for WT versus R1, WT versus R3, R1 versus R2, and R2 versus R3. (M) Bars show the mean maximum rate of rise (ΔF/F0 s−1) of fluorescence signals after a 1000-ms photolysis flash, determined from 100 randomly chosen cells for each cell line with n = 3 imaging fields. Traces in (B) to (E) and (H) to (K) are shown on identical magnitude (ΔF/F0), but different time scales. Statistically significant differences in (F), (G), and (M) were determined by ANOVA with Tukey post hoc tests; *P < 0.05 and **P < 0.01.

Basal Ca2+homeostasis in HEK-293 cell lines

We first investigated basal Ca2+ homeostasis by imaging whole-cell Ca2+ fluorescence signals in the HEK-293 cell lines expressing different IP3R isoforms and in WT and 3KO cells. Cells were loaded with cytosolic Ca2+ indicators (63) and imaged by wide-field (WF) epifluorescence microscopy at low magnification. Basal cytosolic free [Ca2+], as determined by the mean resting fluorescence (F0) in cells loaded with the high-affinity Ca2+ indicator Cal-520 (fig. S1A) and by the mean resting fluorescence excitation ratio (340/380 nm) in cells loaded with fura-2 (fig. S1B), was similar in the different cell lines. Intracellular Ca2+ store content, as assessed by the peak ratiometric (ΔF/F0) signals of the low-affinity Ca2+ indicator Fluo-8L in cells treated with ionomycin to permeabilize intracellular organelles, was also similar between the cell lines (fig. S1C). On the basis of these data, we concluded that the flux of Ca2+ through open IP3Rs would not be appreciably affected by any differences between the cell lines in the concentration gradient driving Ca2+ efflux from the ER or by differences in basal cytosolic [Ca2+] that might affect IP3R gating. Moreover, calculation of Ca2+ signals in terms of fluorescence ratio (ΔF/F0) would be substantially unaffected by any differences in basal fluorescence (F0) among the cell lines.

Global Ca2+signals evoked by carbachol and photoreleased i-IP3

Next, we measured global, whole-cell Ca2+ signals evoked using bath application of a high concentration of the muscarinic receptor agonist carbachol (CCH) to maximally stimulate the endogenous IP3 signaling pathway. To exclude signals arising from entry of Ca2+ across the plasma membrane, cells were bathed in a solution containing no added Ca2+ and EGTA 5 min before and during imaging. Almost all (>98%) of WT and single isoform–expressing cells showed abrupt Ca2+ responses (Fig. 1, B to E), whereas 3KO cells devoid of IP3Rs showed no discernable responses. The peak responses were greatest in cells expressing type 2 and 3 IP3Rs, slightly smaller in WT cells, and substantially smaller in cells expressing type 1 IP3Rs (Fig. 1F). To obtain a surrogate estimate of the relative numbers of active IP3R/channels contributing to the initial release of ER Ca2+ in response to CCH, we measured the maximal rates of rise of fluorescence signals (ΔF/F0) during the rising phase of global Ca2+ signals in WT cells and in cells expressing single IP3R isoforms. The rates of rise were fastest for cells expressing IP3R2, intermediate for WT and IP3R3-expressing cells, and slowest for IP3R1-expressing cells (Fig. 1G).

We then applied flashes of UV light to photorelease a poorly metabolized IP3 analog, i-IP3, in cells loaded with a caged precursor together with Cal-520 (57, 64). Photoreleased i-IP3 evoked intracellular Ca2+ liberation in WT and all three isoform-expressing cells (Fig. 1, H to K), whereas 3KO cells were unresponsive. We attributed these signals to arise exclusively from Ca2+ liberation through IP3Rs, given that recordings were made in Ca2+-free bathing solution and that, in agreement with reports that HEK-293 cells lack ryanodine receptors (65, 66), i-IP3–evoked Ca2+ signals were not diminished in the presence of ryanodine (fig. S2A).

The mean amplitudes of i-IP3–evoked Ca2+ signals in WT cells showed a graded dependence on flash duration, with 100-ms flashes evoking little or no response, and flash durations of ≥700 ms evoking apparently maximal signals, with a peak ΔF/F0 of 7.6 (Fig. 1L). Mean responses in cells expressing exclusively IP3R2 resembled those of WT cells, with a similar dependence on photolysis flash duration and slightly greater maximal amplitude (9.3 ΔF/F0; Fig. 1L). In contrast, cells expressing type 1 (Fig. 1I) or type 3 (Fig. 1K) IP3Rs were less responsive; respectively, 73 and 76% of cells responded to flash durations of 1000 ms. Moreover, the fluorescence signals mediated by type 1 and 3 IP3Rs were smaller in amplitude (Fig. 1L), and their rise and fall kinetics were slower than in WT and IP3R2-expressing cells (Fig. 1, I and K). The maximum rates of rise of Ca2+ signals evoked by 1000-ms photolysis flashes in the different cell lines mirrored their peak response sizes, with rates of rise being much slower in IP3R1- and IP3R3-expressing cells (Fig. 1M).

Ca2+puffs in cell lines expressing single IP3R isoforms

To better visualize local Ca2+ puffs, we additionally loaded cells with the slow Ca2+ buffer EGTA to inhibit the formation of Ca2+ waves and sharpen the spatiotemporal profile of puffs (14, 57). Imaging was performed by TIRF microscopy at a fast frame rate from a field that generally encompassed only a part of a single cell. Transient, highly localized increases in Cal-520 fluorescence arose after photorelease of i-IP3 both in WT cells (Fig. 2A) and in cells expressing each single IP3R isoform (Fig. 2, B to D). In all cell lines, puffs occurred throughout the 30-s recording period (Fig. 2, A to D), consistent with previous findings (57, 67), indicating that photoreleased i-IP3 is not appreciably degraded during this time. The properties of Ca2+ puffs were not altered by the presence of ryanodine (fig. S2B). To avoid potential depletion of intracellular Ca2+ stores over time, we imaged puffs in a bath solution containing 2 mM Ca2+ rather than zero Ca2+. In separate experiments, we did not detect appreciable differences in puffs when imaged in a zero Ca2+, EGTA-containing solution (fig. S3, A to C).

Fig. 2 Local, subcellular Ca

2+ puffs evoked by photoreleased i-IP3 in WT cells and CRISPR-Cas9–modified HEK-293 cells exclusively expressing type 1, 2, or 3 IP3Rs. (A to D) Traces show Cal-520 fluorescence ratio measurements (ΔF/F0) from 1.5 μm by 1.5 μm regions of interest centered on puff sites in WT cells (A) and cells exclusively expressing IP3R1 (B), IP3R2 (C), and IP3R3 (D). Each panel shows records that are representative of puff sites exhibiting low and high frequencies of Ca2+ puffs; arrows and gaps in the traces mark the time of the photolysis flash. Traces in (A) were obtained using a 250-ms photolysis flash duration, and those in (B) to (D) with a flash duration of 500 ms. (E to H) Histograms depict the distributions of Ca2+ puff frequencies (number of puffs per site per 30-s recording) occurring at discrete puff sites in WT cells and cells expressing individual IP3R isoforms. (I) Mean numbers of puff sites per imaging field (37 μm by 19 μm) for WT cells and cells expressing individual IP3R isoforms; *P < 0.05 when assessed by ANOVA with Tukey post hoc tests. (J) Mean numbers of puffs per imaging field for WT cells and cells expressing individual IP3R isoforms; **P < 0.01 when assessed by ANOVA with Tukey post hoc tests. Data in (E) to (J) are from totals of 480 (WT), 272 (IP3R1), 338 (IP3R2), and 304 (IP3R3) puff sites, from 13, 8, 9, and 11 cells, respectively. Error bars in (I) and (J) depict 1 SEM. (K) Dose-response relationship for puffs generated by different IP3R isoforms. The data are from experiments in which records were acquired at a slower frame rate (151 frames s−1) than in (A) to (J) and from larger imaging fields (53 μm by 53 μm) that typically contained two to three cells. The plot shows the numbers of puffs detected throughout the imaging field in 10 s after photolysis flashes of various durations in cells solely expressing type 1, 2, or 3 IP3Rs (depicted by different symbols and colors, as indicated). Data points are means ± 1 SEM from n = 4 to 7 imaging fields.

We recorded fluorescence traces (ΔF/F0) from regions of interest centered on discrete subcellular sites and show representative records from sites exhibiting low and high frequencies of puffs in WT cells (Fig. 2A) and cells exclusively expressing type 1 (Fig. 2B), type 2 (Fig. 2C), and type 3 (Fig. 2D) IP3Rs. The numbers of puffs varied markedly from one site to another, even in the same cell (Fig. 2, A to D). Histograms plotting the numbers of puffs observed at a site (Fig. 2, E to H) displayed skewed distributions, with many sites showing only a single puff during the recording period. The mean numbers of puff sites detected per imaging field were similar in WT cells and cells expressing type 1 and 2 IP3Rs but were slightly fewer in cells expressing IP3R3 (Fig. 2I). Under matched conditions, flash durations of 500 to 700 ms evoked similar mean total numbers of puffs after photorelease of i-IP3 in cells expressing IP3R1 and IP3R2 and a greater mean number of puffs in IP3R3-expressing cells (Fig. 2J). WT cells were more responsive, such that weaker stimuli (250- to 300-ms flash durations) evoked similar or greater numbers of events than evoked by 500- to 700-ms flashes in single isoform–expressing cells (Fig. 2J).

To further examine the sensitivities of cells expressing single IP3R isoforms to different concentrations of i-IP3, in a separate set of experiments, we counted the total numbers of puffs that arose throughout the imaging field in 10 s after photolysis flashes with varying durations ranging up to 1000 ms (Fig. 2K). We could not explore stronger stimuli because, even with cell loading of EGTA, they evoked global Ca2+ rises that obscured puffs. Brief UV flashes (durations ≤100 ms) were ineffective in evoking puffs in all single isoform-expressing cell lines, with mean numbers of puffs detected (≤4.5) little greater than spontaneous events occurring during the same time without stimulation (≤2.25; Fig. 2K). Longer flash durations (≥300 ms) evoked progressively more puffs in all cell lines, and the mean numbers of puffs evoked by any given flash duration were similar for cells expressing each IP3R isoform (Fig. 2K). We thus conclude that the clusters of IP3Rs that generate local Ca2+ puffs display a similar dependence on i-IP3 concentration, irrespective of the IP3R isoform from which they are composed.

Amplitudes of puffs mediated by different IP3R isoforms

To facilitate visual comparison of the magnitudes and time courses of puffs in WT cells and in cells expressing individual IP3R isoforms, we overlaid representative fluorescence traces (ΔF/F0) on an expanded time scale after aligning puffs to the times of their peak amplitude (Fig. 3, A to D, traces). The amplitude distributions of puffs in WT cells and cells expressing each IP3R isoform followed skewed Gaussian functions, with most puffs ranging between 0.3 to 0.6 (ΔF/F0) in peak amplitude (Fig. 3, A to D, histograms). Plots of the cumulative distributions of puff amplitudes (Fig. 3E) further demonstrated the similarity in puff amplitudes between WT cells and those expressing specific IP3R isoforms. Mean peak puff amplitudes were not significantly different between the different isoforms or WT cells (Fig. 3F).

Fig. 3 Amplitudes of Ca

2+ puffs evoked in HEK WT cells and cells exclusively expressing type 1, 2, and 3 IP3Rs. (A) Left: The fluorescence traces (ΔF/F0) illustrate representative Ca2+ puffs evoked by photoreleased i-IP3 in WT cells. The panel shows several puffs, superimposed after alignment of the times of peak amplitude. The histogram plots the distribution of peak puff amplitudes (ΔF/F0) measured from 1841 events in 13 cells. (B to D) Panels show corresponding examples of puff traces and amplitude distributions in cells exclusively expressing IP3R1 (B; histogram is from 810 events from 8 cells), IP3R2 (C; 796 events from 9 cells), and IP3R3 (D; 1438 events from 11 cells). (E) Cumulative frequency curves showing the percentage of all detected events as a cumulative function of puff amplitude for WT cells and cells expressing each individual isoform (depicted by different colored symbols as indicated). (F) Bars show mean peak puff amplitudes (ΔF/F0) in WT cells and cells expressing each IP3R isoform, with n = the number of cells examined as listed above; error bars denote 1 SEM. Mean peak puff amplitudes were not significantly different between cell lines when assessed by either ANOVA or Kruskal-Wallis tests.

Quantal analysis of Ca2+flux through single channels formed by IP3R isoforms

The TIRF signal during puffs closely tracks the underlying flux of Ca2+ into the cytosol (50, 68), which, in turn, is a function of the instantaneous number of channels open and the Ca2+ flux through each channel. The similarity in mean puff amplitudes among different IP3R isoforms thus suggested that they likely involved the openings of similar numbers of channels with equivalent Ca2+ permeability but it could also be due to differing numbers of active channels if the Ca2+ flux varied between isoforms. To discriminate between these possibilities, we performed a quantal analysis by exploiting the observation that many puffs exhibit stepwise transitions in fluorescence levels that we interpreted to reflect the openings and closings of individual IP3R channels (50, 55).

Numerous puffs in WT cells and in cells expressing each individual IP3R isoform exhibited stepwise fluorescence dwell states during both their rising and falling phases, and we also observed small events with a “square” appearance that abruptly rose and fell from the baseline (Fig. 4, A to D, traces). Measurements of the amplitudes (ΔF/F0) of these fluorescence dwell-state levels in WT cells, and in cells expressing single isoforms, displayed multimodal distributions (Fig. 4, A to D, histograms). These distributions were fit well by a sum of four Gaussian functions, with successive means of about 0.13, 0.26, 0.38, and 0.5 ΔF/F0. The mean amplitude of the first Gaussian corresponded closely with our previous estimate (ΔF/F0 ~ 0.11) of the fluorescence signal resulting from opening of a single IP3R channel in a neuroblastoma cell line (50, 55), and the peaks of successive Gaussian fits recurred at near integer multiples. The peaks of these successive Gaussians incremented linearly and almost identically for all cell lines, with a regression fit giving a mean slope of 0.126 ΔF/F0 per quantal step for WT and each IP3R isoform (Fig. 4E).

Fig. 4 Quantal analysis of single-channel IP

3R Ca2+ flux during Ca2+ puffs. (A) Left: Traces show representative puffs in a WT cell, presented on an expanded time scale to illustrate stepwise increments and decrements of fluorescence (ΔF/F0) resulting from the successive openings and closings of individual IP3R channels. Fluorescence measurements were made from 1.5 μm by 1.5 μm regions of interest centered on puff sites. Horizontal lines drawn at integer multiples of fluorescence amplitude indicate the numbers of open channels during dwell times at different step amplitudes. The histogram shows the distribution of amplitudes (ΔF/F0) of visually identified mean dwell levels in WT cells. The curves show individual components of a multi (four)–Gaussian fit to the distribution. Data are from 415 WT dwell states. (B to D) Corresponding examples of stepwise puffs and of step dwell-state amplitude distributions and multi-Gaussian fits to data from cells expressing exclusively IP3R1 (B), IP3R2 (C), and IP3R3 (D). Data are from 665 (IP3R1), 659 (IP3R2), and 696 (IP3R3) dwell states. (E) The graph shows the center values of the four Gaussian distributions against their ordinate number (1 to 4). Different symbols represent data from the different cell lines, as indicated. Symbols overlap where not visible. The line is a regression fit to data from all cell types, constrained to pass through the origin, with a slope of ΔF/F0 = 0.126 per unitary step level.

These results indicated that homomeric channels formed by all three IP3R isoforms mediated similar Ca2+ fluxes, which were also indistinguishable from the unitary Ca2+ flux in WT cells. Because we found no differences in Ca2+ store content or basal cytosolic [Ca2+] among the cell lines (fig. S1, A to D), this further implied that the Ca2+ permeabilities of channels formed by each IP3R isoforms are almost identical. Moreover, the close similarity in mean puff amplitudes across the different cell lines (Fig. 3, B to F) indicated that puffs mediated by each isoform involved the openings, on average, of four channels (mean puff amplitude/unitary channel amplitude, 0.51/0.126 ΔF/F0).

Puff kinetics

The distributions of puff rise, fall, and full duration at half-maximal amplitude (FDHM) times were similar in WT cells (Fig. 5A) and in cells exclusively expressing type 1 (Fig. 5B) or type 2 (Fig. 5C) IP3Rs. In contrast, puffs in cells expressing type 3 IP3Rs displayed appreciably faster rise and fall times and shorter durations (Fig. 5D). Overlaid cumulative distribution curves of puff kinetics further highlight these observations (Fig. 5E). Quantification of mean puff rise, fall, and duration times demonstrated that the kinetics of IP3-mediated Ca2+ puffs in WT cells and in type 1 and 2 IP3R–expressing cells were similar and not significantly different from one another, whereas all parameters were significantly faster for puffs mediated by type 3 IP3Rs (Fig. 5F).

Fig. 5 Temporal characteristics of Ca

2+ puffs in WT cells and cells exclusively expressing type 1, 2, or 3 IP3Rs. (A to D) Histograms show the distributions of Ca2+ puff rise times (from 20 to 80% of peak value; left), fall times (fall from 80 to 20% of peak; center), and FDHM (right) for the different cell lines, as indicated. (E) Cumulative frequency curves show the percentage of all detected events as a function of puff rise time (left), fall time (center), and duration (right) for WT cells and cells expressing each isoform (depicted as indicated by different colored symbols). (F) Bars show mean puff rise times (left), fall times (center), and durations (right); error bars denote 1 SEM, with n = the number of cells examined. *P < 0.05 and **P < 0.01 when determined by ANOVA with Tukey post hoc tests. Data in this figure were derived from the same cells and recordings as in Fig. 3: WT, 1841 events from 13 cells; IP3R1, 810 events from 8 cells; IP3R2, 796 events from 9 cells; IP3R3, 1438 events from 11 cells.

Spatial extent of fluorescence signals during puffs

The algorithm that we used to detect and analyze puffs also determined a measure of the spatial extent of the fluorescence signal [full width at half-maximal amplitude (FWHM)] by fitting a two-dimensional (2D) Gaussian function to a temporally filtered image of the puff (Fig. 6A). Histograms (Fig. 6, B to E) and cumulative curves (Fig. 6F) plot the distributions of FWHM values for WT and individual isoform-expressing cells, with a hard cutoff at ~2.4 μm imposed by a requirement to fit a minimum of 2 pixels. Comparison of mean FWHM values (Fig. 6G) demonstrated that the spatial spread of puffs mediated WT cells and cells expressing single IP3R isoforms were generally similar, except that IP3R3-mediated puffs were slightly more restricted than puffs in IP3R1-expressing cells, possibly as a consequence of their briefer kinetics.

Fig. 6 Spatial spread of fluorescence signal during Ca

2+ puffs evoked in WT cells and cell lines exclusively expressing specific IP3R isoforms. (A) Left: Representative, temporally filtered image of a Ca2+ puff, with increasing fluorescence (F/F0) depicted on a pseudocolor scale and by increasing height. Right: 2D Gaussian fit to this puff. Scale bar, 5 μm. (B to E) Plots showing the distributions of puff spatial widths (FWHM of the Gaussian fit) measured, respectively, in WT cells and cells expressing type 1, 2, and 3 IP3Rs. The algorithm used to fit the image data truncated measurements at widths <2.4 μm. (F) Cumulative frequency curves show the cumulative percentage of all detected events as a function of increasing spatial width (FWHM) of puff fluorescence signals for WT cells and cells expressing each isoform (depicted by different colored symbols). (G) Bars show mean spatial widths; error bars denote 1 SEM, with n = the number of cells. **P < 0.01 when determined by ANOVA with Tukey post hoc tests. Data in this figure were derived from the same cells and recordings as in Fig. 3: WT, 1841 events from 13 cells; IP3R1, 810 events from 8 cells; IP3R2, 796 events from 9 cells; IP3R3, 1438 events from 11 cells.

Spatial organization of IP3R clusters as reflected by the localizations of Ca2+signals

The center of mass of the fluorescence signal during a puff is localized with a precision of several tens of nanometers by our algorithm that determines the centroids of 2D Gaussian fits to the fluorescence signals and provides a weighted estimate of the centroid of those individual IP3R channels open during that event (69). To derive information on the functional architecture of clusters of homomeric IP3Rs of different isoforms, we mapped the x and y coordinates of individual puffs recurring at given sites in cells expressing different IP3R isoforms (Fig. 7, A to D, left).

Fig. 7 Subcellular localizations of Ca

2+ puffs in WT cells and cells exclusively expressing type 1, 2, or 3 IP3Rs. (A) Left: Representative example of the locations of Ca2+ puffs recurring at an individual site in a WT cell. Circles mark the centroid localizations of Ca2+ fluorescence evoked by puffs within a 1 μm by 1 μm region of interest (outlined by the box) that was centered on the puff site. Middle: Distribution plot of neighbor-neighbor distances between every puff localization and every other localization. Distances were calculated for all puff localizations within the 37 μm by 19 μm imaging field, out to a radius of 5 μm around each localization. The black line, generated by a linear fit to the counts between 1 and 5 μm and constrained through zero, approximates the distribution expected if puff locations were randomly distributed. Right: Plot of the difference between the observed and predicted random distributions of neighbor-neighbor distances to provide a measure of the clustering of puff locations. (B to D) Corresponding examples of puff localizations and neighbor-neighbor distance distributions for cells exclusively expressing IP3R1, IP3R2, and IP3R3. (E) Neighbor-neighbor distance distributions for WT and each isoform replotted and overlaid from the right panels in (A) to (D). Data from the different cell lines are depicted by different colored symbols and lines and are shown after normalizing to their respective peaks and plotting on an expanded scale to better visualize differences in the spatial localization of puffs mediated by the different IP3R isoforms. Data in this figure were derived from the same cells and recordings as in Fig. 3: WT, 1841 events from 13 cells; IP3R1, 810 events from 8 cells; IP3R2, 796 events from 9 cells; IP3R3, 1438 events from 11 cells.

For each cell line, we used an all-neighbor distance analysis routine (56) to measure the distance of the centroid location of each puff to that of every other puff within a 5-μm radius (Fig. 7, A to D, center). If puffs arose at randomly and homogeneously distributed sites, then the distribution of neighbor-neighbor distances is expected to show a linear increase with increasing distance (70). In contrast, the histograms for all cell types showed a prominent peak at distances of <0.5 μm, consistent with successive puffs occurring repetitively in close proximity, at fixed locations. The black lines overlaid on the all-neighbor distance distributions were drawn to approximate the profiles expected from a random distribution of sites. By subtracting these predicted random distributions from the observed all-neighbor distance distributions, we obtained a measure of the clustering of puffs at specific sites. These distributions showed peaks at distances of a few hundreds of nanometers, declining toward that expected from a random distribution at distances approaching 2 μm (Fig. 7, A to D, right). To better compare the extent of clustering of puffs among the different IP3R isoforms, we replotted and superimposed the clustering data (Fig. 7, A to D, right) after normalizing the counts to the same peak value (Fig. 7E). The clustering data for WT cells and cells expressing type 2 and 3 IP3Rs showed similar profiles, whereas the broader profile for puffs in cells expressing type 1 IP3R (Fig. 7E, red curve) suggests that puffs in these cells were less tightly clustered.

Cellular locations of puff sites

TIRF microscopy provides information only on those local Ca2+ signals that arise in close proximity to the plasma membrane, adjacent to the cover glass. To determine the distribution of puff sites throughout the volume of the cell, we used lattice light-sheet microscopy to visualize Ca2+ signals with high temporal resolution in diagonal “slices” cutting through the cell interior (71). Puffs in SH-SY5Y neuroblastoma cells (57, 71) and HeLa cells (61) arise almost exclusively at sites close to the plasma membrane. However, in HEK-293 cells, we observed puffs arising at sites both close to the membrane (Fig. 8A) and deeper into the cell (Fig. 8B). Puffs in cells expressing each individual IP3R isoform and in WT cells arose primarily around the cell membrane, but in each cell line, appreciable numbers (~20%) of puffs were observed in the cell interior (Fig. 8C).

Fig. 8 Distribution of puff sites throughout the interior of WT cells and cells expressing individual IP

3R isoforms, visualized by lattice light-sheet microscopy. (A and B) The images illustrate a single, diagonal light-sheet section through a cell expressing type 3 IP3Rs. The plasma membrane is depicted in red, and local Ca2+ puffs (ΔF/F0) are depicted in green. Images of puffs arising at different times and locations are superimposed on the membrane image. Representative examples are shown of puffs arising immediately adjacent to the plasma membrane (A) and deep in the interior of the same cell (B). Scale bar, 5 μm. (C) Bar graph summarizing the percentages of puffs arising near the cell edge or deeper (>2 μm) within the cell interior in WT (n = 42), IP3R1 (n = 8), IP3R2 (n = 12), and IP3R3 (n = 9) cells.


Ca2+ puffs are localized cytosolic Ca2+ signals that arise from liberation of Ca2+ stored in the ER lumen through tight clusters comprising small numbers of IP3R/channels. Puffs function as basic building blocks of cellular Ca2+ signaling (52) and serve to trigger global, cell-wide Ca2+ waves (10). The ability to image Ca2+ signals with a single-channel level of resolution enables us to determine the properties of individual IP3Rs in an intact cellular environment. However, studies to date have been hampered because native cells express various proportions of the three IP3R isoforms (3336), so the composition of the IP3R channels underlying Ca2+ puffs could not be inferred. Although DT40 cell lines derived from avian B cells by homologous recombination to express the entire complement of IP3R types (45) have been used to investigate the behavior of select IP3R isoforms in their native environment (20), we find that the small size and limited TIRF footprint of these cells render them unsuitable for studying Ca2+ puffs. Here, we circumvented these problems by high-resolution imaging of Ca2+ puffs in HEK-293 cell lines genetically engineered to each express only a single IP3R isoform (46).

Our most obvious finding is that cells expressing exclusively type 1, 2, or 3 IP3Rs all generated puffs in response to photorelease of i-IP3. As in WT HEK-293 cells and many other native cell types, these puffs recurred at stationary subcellular sites and displayed amplitudes and properties consistent with their arising through the concerted opening of a small number of closely adjacent IP3R channels (6, 57, 61, 72, 73). Although a majority of IP3Rs (of all subtypes) are mobile within the ER membrane (61, 70, 7476), the immobility of puff sites, together with observations of stationary clusters of fluorescently tagged IP3Rs (61, 70), supports the hypothesis that puffs arise from a distinct pool of IP3Rs that are tethered to as yet unidentified cytoskeletal anchoring sites, where they are “licensed” to preferentially respond to IP3 (56, 58, 61, 77).

We thus conclude that all three IP3R isoforms express the binding site(s) that enable their scaffolding into small clusters. Because we found that the mean number of channels that opened during a puff was similar in cells expressing each isoform of IP3R, this finding further suggests that the stoichiometry of the IP3R cluster is determined by the architecture of the underlying scaffolding structure and is independent of the specific IP3R isoform. Last, we observed that the distribution of puff sites throughout the cell volume, primarily in a cortical shell adjacent to the plasma membrane, was similar for cells expressing each isoform. This again is consistent with the notion that each IP3R isoform binds to a preexisting cellular scaffold.

A further observation was that the properties of puffs generated by each isoform of IP3R were generally similar. Photorelease of equivalent amounts of i-IP3 in cell lines exclusively expressing individual isoforms evoked puffs at similar frequencies and with similar amplitudes. Moreover, the single-channel Ca2+ fluxes through all three IP3R isoforms were indistinguishable, indicating that they had similar permeabilities, given that ER Ca2+ store content and basal cytosolic [Ca2+] did not differ between the cell lines. This result agrees with a majority of studies that report similar single-channel conductances as determined from single-channel electrophysiological recordings of monovalent ion currents through different IP3R isoforms. However, there are reports of differences between different cell types, suggesting that factors such as membrane environment may affect the conductance (48).

We observed the greatest difference between puffs generated by the different IP3R isoforms in terms of their kinetics. Puffs mediated by type 3 IP3Rs were briefer than those mediated by the other isoforms, implying that in terms of cumulative cellular Ca2+ liberation, they “pack less of a punch.” The molecular basis for this kinetic difference of the type 3 IP3R remains unclear. The mechanisms underlying the termination of Ca2+ liberation during puffs are unresolved, likely involving feedback by Ca2+-dependent inhibition of IP3Rs, as well as coupled gating analogous to that seen with ryanodine receptors (78). Both of these factors may potentially differ between IP3R isoforms.

In contrast to our observations in minimally perturbed intact cells that the three IP3R isoforms mediate closely similar local Ca2+ signals, electrophysiological studies using reduced, in vitro experimental settings have shown marked differences between the single-channel gating properties of IP3R isoforms [for reviews, see (18, 48)]. Systematic comparisons between the electrophysiological data and our findings in the intact cell are confounded by several factors, which include differences in the environments to which the IP3Rs are exposed, as exemplified by the wide range of Ca2+ dependencies reported for the same recombinant IP3R expressed in different cell types (48). In light of this issue, the most straightforward data for comparison derive from analysis of homotetrameric channels of each of the three IP3R isoforms expressed in DT40-3KO cells (25, 48, 79). Those experiments showed complex differences between the isoforms in the Ca2+ dependence of their mean open probability (Po), with optimal Ca2+ concentrations over a roughly 10-fold range. In terms of IP3 sensitivity, the sequence was found to be IP3R2 > IP3R1 > IP3R3, based on observed Po values at optimal Ca2+ concentrations and respective subsaturating IP3 concentrations of 0.51, 0.27, and 0.17 (48). These findings differ from our observations that the frequencies of puffs evoked by the same photorelease of i-IP3 were similar among cells expressing each IP3R isoform.

A further problem in comparing the activity of IP3Rs during puffs with their properties as determined from electrophysiological measurements is that the latter generally report steady-state properties, whereas puffs are highly dynamic, with rapid and large changes in Ca2+ concentrations. The cytosolic Ca2+ nanodomain at a puff site likely rises to tens or even hundreds of micromolar once the first channel opens (80), so that IP3Rs in the cluster experience high concentrations that promote maximal activation and subsequent maximal inhibition. Rapid saturation of Ca2+ binding to sites on the IP3R may explain why puffs mediated by the three isoforms appear so similar, despite differences that are apparent at lower Ca2+ concentrations in steady-state electrophysiological recordings. The gating of IP3Rs during puffs likely hinges on the rapid kinetics of their modulation by Ca2+. To date, however, the dynamic responses of single IP3Rs to abrupt changes in ligand have been studied only for the single insect IP3R isoform, and no data are available for the mammalian isoforms (47).

Our results reveal intriguing discrepancies between the local (puffs) and global Ca2+ signals mediated by the different IP3R isoforms. Although the amplitudes and frequencies of puffs evoked by a given photolysis flash were closely similar among cells expressing each IP3R isoform, the global cellular Ca2+ responses evoked by strong photorelease of i-IP3 in cells not loaded with EGTA were smaller in cells expressing IP3R1 and IP3R3 than in those expressing IP3R2. We speculate that these differences may arise from differences in the functional properties of the greater fraction of IP3Rs that underlie the global Ca2+ response, given that stimulation with the receptor agonist CCH elicited much larger whole-cell Ca2+ signals and more rapid rates of rise, when compared to i-IP3, in cells solely expressing type 1 and 3 IP3Rs. Puffs are generated by the small subpopulation of IP3Rs that are anchored at discrete cluster sites and which respond preferentially at lower [IP3] than do the greater bulk of mobile IP3Rs that underlie the global cellular Ca2+ response (56, 57, 61, 77). Thus, the sensitivity to IP3 and other properties of the bulk population of IP3Rs may differ more widely between the different isoforms than is apparent for the IP3Rs clustered at puff sites. In addition, it remains possible that differences in IP3R expression levels affected the whole-cell Ca2+ responses, because assembly of IP3Rs at the cytoskeletal structures defining puff sites might remain saturated even if the expression levels of the various isoforms differed appreciably, whereas the numbers of IP3Rs available to mediate the much larger global signals would be diminished at low expression levels.

In summary, we used “optical patch-clamp” imaging of Ca2+ flux with single-channel resolution (81) to study the physiological functioning of IP3Rs in intact, minimally perturbed cells (50, 57, 58). Our results in HEK-293 cells showed that IP3Rs composed solely of type 1, 2, or 3 isoforms were equally capable of producing Ca2+ puffs that recurred at fixed sites, indicating that each isoform is able to bind to some, as yet unidentified, cytoskeletal anchoring sites that define the location and composition of functional IP3R clusters (58, 61). The mean numbers of functional IP3R channels clustered at each site and their Ca2+ permeabilities were similar for homomeric channels formed by individual isoforms, as well as for channels in WT cells that are likely heteromers composed from two or three different isoforms (33, 37, 38). In contrast to proposals that the properties of native heteromeric IP3Rs may be determined by the mix of their constituent isoforms (39, 40) and that cell types express a particular complement of IP3R isoforms to suit their particular needs (41), our results indicate substantial redundancy among the IP3R isoforms in regard to their function and spatial localization at puff sites. While this manuscript was under review, another group, using the same commercially available HEK-KO cell lines that we generated, published a study detailing similar findings to ours (82).


Generation of HEK-293 cells expressing single IP3R isoforms

sgRNAs targeting the fourth exon in IP3R1, IP3R2, or IP3R3 genes were designed using (83) as described previously (46). Oligonucleotides corresponding to these sgRNAs were synthesized by Integrated DNA Technologies and cloned into a pX458 vector encoding Cas9 nuclease and enhanced green fluorescent protein (EGFP; obtained from Addgene). sgRNA constructs in various combinations were transfected into HEK-293 WT cells using Lipofectamine 2000 following the manufacturer’s instructions. Forty-eight hours after transfection, flow cytometry was used to sort EGFP-expressing cells, and these sorted cells were grown as single cells in 96-well plates. Clones were propagated and expanded for 3 weeks and then harvested for characterization by Western blot. Potential clones were subsequently genotyped as described in (46), which confirmed the introduction of base pair insertions and deletions in both alleles.

Characterization of IP3R isoform expression by Western blot

Cells were washed with ice-cold phosphate-buffered saline and solubilized for 30 min in lysis buffer composed of 10 mM tris-HCl, 100 mM NaCl, 1 mM EDTA, 1 mM NaF, 20 mM Na4P2O7, 2 mM Na3VO4, 1% Triton X-100 (v/v), 0.5% sodium deoxycholate (w/v), and 10% glycerol with a mixture of protease inhibitors (Roche). Lysates were centrifuged at 16,000g for 10 min at 4°C, supernatants were then collected, and protein concentrations were determined using the Dc Protein Assay Kit (Bio-Rad). Equal amounts of lysate (10 μg) were fractionated by SDS–polyacrylamide gel electrophoresis and electrotransferred to nitrocellulose membranes. Blots were probed with antibodies to type 1 (rabbit polyclonal antibodies against the C-terminal 19 amino acids of rat IP3R1), type 2 (rabbit polyclonal antibodies against 320 to 338 amino acids of rat IP3R2), or type 3 (mouse monoclonal antibodies against 22 to 230 amino acids of human IP3R3) IP3Rs, as well as antibody to calreticulin to serve as a loading control. After incubation with horseradish peroxidase–conjugated secondary antibodies (Bio-Rad), blots were imaged using enhanced chemiluminescent substrate (Thermo Fisher Scientific). Antibodies against type 1 and 2 IP3Rs were from Pocono Rabbit Farm and Laboratory, and antibodies against type 3 IP3Rs were from BD Transduction Laboratories. Antibodies against calreticulin were from BD Biosciences (#612136). Band intensities were measured on a LI-COR Odyssey imaging system. For densitometry measurements, each band stained with isoform-specific IP3R antibodies was first normalized to its corresponding loading control band stained with calreticulin antibodies, before normalizing the expression of individual IP3R isoforms in single isoform–expressing cells to their respective expression in WT cells.

Cell culture and loading

CRISPR-Cas9–modified HEK-293 cell lines endogenously expressing single IP3R isoforms and an IP3R-null cell line (3KO) were generated and characterized by the Yule Lab and distributed through Kerafast ( For Ca2+ imaging experiments by the Parker Lab, these cells were purchased from Kerafast: IP3R type 1 (#EUR031), type 2 (#EUR032), type 3 (#EUR033), and 3KO (#EUR030). The parental HEK-293 WT cells used to establish the genetically engineered cells were provided directly by the Yule Lab. To confirm the identity of HEK-KO cell lines obtained through Kerafast, protein lysates of all cell lines used for imaging were prepared by the Parker Lab and sent to the Yule Lab for characterization by Western blot. All cell lines were cultured on plastic 75-cm flasks in Eagle’s minimum essential medium (#30-2003, American Type Culture Collection), supplemented with 10% fetal bovine serum (#FB-11, Omega Scientific), 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 (#25200-056, Gibco) and grown on poly-d-lysine–coated (500 μg/ml; #P0899, Sigma-Aldrich) 35-mm glass-bottom imaging dishes (#P35-1.5-14-C, MatTek) or 12-mm glass coverslips for 2 to 3 days.

Immediately before imaging, cells were incubated with membrane-permeable esters of the fluorescent Ca2+ dye Cal-520/AM (acetoxymethyl) (5 μM; #21130, AAT Bioquest) and the caged IP3 analog 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; #cag-iso-2-145-10, SiChem) for 1 hour at room temperature (RT) in a Ca2+-containing Hepes-buffered salt solution (Ca2+-HBSS). In experiments with ionomycin, cells were instead loaded with 5 μM Fluo-8L/AM (#21097, AAT Bioquest) for 1 hour at RT. In experiments using fura-2, cells were loaded with 5 μM fura-2/AM for 1 hour at RT. Where indicated, cells were loaded for an additional hour with EGTA/AM (5 μM; #E1219, Thermo Fisher Scientific). Ca2+ indicators, caged i-IP3, and EGTA/AM were all solubilized with dimethyl sulfoxide/20% pluronic F127 (#P3000MP, Thermo Fisher Scientific). Ca2+-HBSS contained 135 mM NaCl, 5.4 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 10 mM Hepes, and 10 mM glucose (pH 7.4). Zero Ca2+/EGTA-HBSS consisted of the same formulation as Ca2+-HBSS, except that CaCl2 was omitted and 300 μM EGTA was included. CCH was from Sigma-Aldrich (#C4382), and ryanodine was from Enzo Life Sciences (#ALX-630-062-M001). For lattice light-sheet imaging, the plasma membrane was stained by adding CellMask Deep Red (#C10046, Thermo Fisher Scientific) to the bathing solution in the imaging chamber (1/50,000 dilution).


TIRF and WF imaging of Ca2+ signals was accomplished using a custom-built system, based around an Olympus IX50 microscope equipped with either an Olympus 10× objective [numerical aperture (NA), 0.4] for WF microscopy or an Olympus 60× objective (NA, 1.45) for TIRF mode. Fluorescence images were acquired with an Evolve electron multiplying charge-coupled device (EMCCD) camera (Photometrics), using 488-nm laser fluorescence excitation and a 510 long-pass emission filter. WF images were captured from a 278 × 243 pixel field (1 pixel = 2.2 μm) at a rate of 10 frames s−1. TIRF images were captured using 2 × 2 pixel binning for a final field of 70 × 35 pixels (1 pixel = 0.53 μm) at a rate of ~333 frames s−1. Where indicated, some TIRF imaging was performed with a larger field of view (100 × 100 pixels or 128 × 128 pixels) at ~151 or ~125 frames s−1, respectively. To photorelease i-IP3, UV light from a xenon arc lamp was filtered through a 350- to 400-nm band-pass 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). All image data were streamed to computer memory using MetaMorph v7.7 (Universal Imaging/Molecular Devices) and stored on hard disk for offline analysis. WF ratiometric imaging of cells loaded with fura-2 was performed with a Nikon Eclipse Ti microscope equipped with a 40× objective and a Hamamatsu EMCCD camera. Light from an arc lamp was filtered through 340- and 380-nm filters and collected through a 510-nm filter. Alternating frames with 340- and 380-nm excitation (400- and 100-ms exposure times, respectively) were captured using NIS Elements software (Nikon) and stored on hard disk for offline analysis.

Lattice light-sheet imaging was performed using a custom-built system as described (71). Images were acquired with an Andor Zyla 4.2 sCMOS camera from a single, diagonal light-sheet slice (512 × 256 pixels; 1 pixel = 0.11 μm) at 250 frames s−1. Ca2+ puffs were imaged in Cal-520– and EGTA-loaded cells for several seconds after photorelease of i-IP3 by a flash from a 405-nm laser diode, using 473-nm laser fluorescence excitation and a 510- to 560-nm band-pass emission filter. A 562-nm laser and 590-nm long-pass filter were then used to image the plasma membrane stained with Deep Red Dye. The locations of puff were marked by visual inspection and overlaid on the plasma membrane image to classify puffs as near membrane (within ~2 μm) or arising deeper within the cell.

Image analysis

Image data in MetaMorph STK or TIF formats were processed using Flika (, a freely available open-source image processing and analysis software written in the Python programming language. The raw image stack was first processed to create a ratio image stack (F/F0), where the fluorescence intensity of each pixel in a given frame was calculated as that intensity (F) divided by the mean resting fluorescence intensity at the pixel averaged over 500 frames before photorelease of i-IP3 (F0). A custom plug-in (“detect puffs”) was then applied for automated detection and analysis of local Ca2+ signals (84). All Ca2+ puffs identified by the algorithm were verified by visual inspection before further analysis. Measurements of peak puff amplitudes (ΔF/F0) and kinetics were performed by the algorithm on a 3 × 3 pixel region of interest centered over the centroid of each event and were exported to Excel spreadsheets for further analysis. The spatial spread of each puff was calculated from a 2D Gaussian fitted to the mean peak signal derived from a temporally filtered image stack. Dwell-state amplitudes (ΔF/F0) of stepwise transitions during individual puffs were manually measured as the difference in fluorescence from the mean baseline preceding the puff to the mean amplitude of a fluorescence level that remained stable for ≥10 ms or as the difference between successive steps (55). Distances between puff centroids were determined from the x and y coordinates of each puff to that of every other puff in the same cell using a nearest-neighbor algorithm written in Python. The amplitudes of whole-cell Ca2+ signals were determined from regions of interest encompassing entire individual cells from the F/F0 image stack and exported to Excel spreadsheets for analysis. The maximum rate of rise of global cellular Ca2+ signals in individual cells was determined from the steepest slope during the rising phase of the F/F0 fluorescence signal. All additional analysis and graphing were performed in Microcal Origin v6.0 (OriginLab).


Mean values are expressed throughout as standardized means ± 1 SEM, where n is the number of imaging fields analyzed. Comparison of mean values between two groups was assessed by both Student’s t test (parametric) and Mann-Whitney test (nonparametric). Comparisons of mean values between three or more groups were assessed by both one-way ANOVA with Tukey post hoc tests (parametric) and Kruskal-Wallis with Mann-Whitney post hoc tests (nonparametric). Differences were considered statistically significant when indicated by both parametric and nonparametric approaches; the larger of the two P values are presented throughout. Statistical significance in figures is denoted as *P < 0.05 and **P < 0.01.


Fig. S1. Basal Ca2+ homeostasis in HEK WT and KO cell lines.

Fig. S2. Comparisons of photoreleased i-IP3–evoked global Ca2+ signals and local Ca2+ puffs in HEK WT cells in the absence and presence of 20 μM ryanodine.

Fig. S3. Comparison of Ca2+ puffs evoked by photoreleased i-IP3 in the presence and absence of extracellular Ca2+ in EGTA-loaded HEK cells solely expressing type 1, 2, or 3 IP3Rs.


Acknowledgments: We thank K. L. Ellefsen and C. A. Karsten for help with software and image analysis and S. Malik for performing the Western blots. Funding: This work was supported by NIH grants R37 GM048071 (I.P.) and R01 DE019245 (D.I.Y.). Author contributions: The study was conceptualized and designed by J.T.L. and I.P. Imaging data were collected and analyzed by J.T.L. and I.P. Generation and Western blot characterization of KO cell lines were performed by K.J.A. and D.I.Y. The manuscript was written by J.T.L. and I.P., with input from K.J.A. and D.I.Y. All authors have read and approved the published manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: Raw data are freely available to academic researchers on request. Custom software (Flika) used for data analysis is available at
View Abstract

Navigate This Article