Functional architecture of inositol 1,4,5-trisphosphate signaling in restricted spaces of myoendothelial projections

Abstract

Calcium (Ca²⁺) release through inositol 1,4,5-trisphosphate receptors (IP₃Rs) regulates the function of virtually every mammalian cell. Unlike ryanodine receptors, which generate local Ca²⁺ events (“sparks”) that transmit signals to the juxtaposed cell membrane, a similar functional architecture has not been reported for IP₃Rs. Here, we have identified spatially fixed, local Ca²⁺ release events (“pulsars”) in vascular endothelial membrane domains that project through the internal elastic lamina to adjacent smooth muscle membranes. Ca²⁺ pulsars are mediated by IP₃Rs in the endothelial endoplasmic reticulum of these membrane projections. Elevation of IP₃ by the endothelium-dependent vasodilator, acetylcholine, increased the frequency of Ca²⁺ pulsars, whereas blunting IP₃ production, blocking IP₃Rs, or depleting endoplasmic reticulum Ca²⁺ inhibited these events. The elementary properties of Ca²⁺ pulsars were distinct from ryanodine-receptor-mediated Ca²⁺ sparks in smooth muscle and from IP₃-mediated Ca²⁺ puffs in Xenopus oocytes. The intermediate conductance, Ca²⁺-sensitive potassium (K_Ca3.1) channel also colocalized to the endothelial projections, and blockage of this channel caused an 8-mV depolarization. Inhibition of Ca²⁺ pulsars also depolarized to a similar extent, and blocking K_Ca3.1 channels was without effect in the absence of pulsars. Our results support a mechanism of IP₃ signaling in which Ca²⁺ release is spatially restricted to transmit intercellular signals.

The location and patterning of intracellular calcium (Ca²⁺) signals encode information that regulates distinct aspects of cell function. Two major intracellular calcium release channels in the sarco-/endoplasmic reticulum (SR/ER) of mammalian cells—inositol 1,4,5-trisphosphate receptors (IP₃Rs) and ryanodine receptors (RyRs)—have the potential to generate modulated and localized calcium signals. Elementary release events termed “calcium sparks” correspond to the activation of a spatially fixed cluster of RyRs in the SR membrane. The organization of RyRs in SR elements permits intimate communication with the closely juxtaposed plasma membrane. In smooth muscle (SM), calcium sparks activate calcium-sensitive large-conductance potassium channels in the plasma membrane, causing a transient hyperpolarizing current that decreases vasoconstriction (1). No such local calcium signaling architecture with analogous functional significance has been reported for IP₃Rs.

Elevation of Ca²⁺ in the vascular endothelium is thought to play a major role in transmitting vasoregulatory signals to adjacent SM cells in the arterial wall. Calcium-dependent endothelial signals include nitric oxide (NO), prostaglandins (PGs), and a mechanistically diverse non-NO, non-PG functionality termed endothelial-derived hyperpolarizing factor (EDHF), which is variably attributable to potassium ions, epoxyeicosatrienoic acids, and direct electrical communication via gap junctions (2, 3). Engagement of the EDHF functionality, regardless of its specific identity, depends on the activation of calcium-sensitive, small-conductance (K_Ca2.3) and intermediate-conductance (K_Ca3.1) potassium channels in the vascular endothelium (3).

Although Ca²⁺ waves and oscillations have been reported in native endothelial preparations (4–6), organized patterns of Ca²⁺ signaling have not been identified, despite the unique polarity of the vascular endothelium and known functional connections between endothelial and SM cells. To explore Ca²⁺ signaling in endothelial cells in the arterial wall, we engineered a mouse that expresses a calcium biosensor (GCaMP2) exclusively in the endothelium (7). Gene-encoded, tissue-specific expression of calcium sensors permits the examination of Ca²⁺ dynamics in multicellular preparations without prolonged periods of dye loading and confounding influence of fluorescence from multiple cell types (8).

Results

The vascular endothelium is separated from SM cells in the arterial wall by the internal elastic lamina (IEL). The IEL contains numerous holes through which endothelial membrane projections penetrate, allowing direct contact and communication between endothelial and SM membranes through myoendothelial gap junctions (9–13). To allow visualization of calcium signals while preserving the organization of this endothelial–IEL–SM structure, we exteriorized the endothelial surface of third-order mouse mesenteric artery segments by making a longitudinal incision and pinning the rectangular preparation. Individual endothelial cells with an average visible surface area of 737 ± 35 μm² were readily distinguishable (Fig. 1A). Black holes in the autofluorescence of the IEL, representing projection points of the endothelial membrane to the SM membrane and sites of connecting gap junctions (9, 11–13), also were seen [5.6 ± 0.4 holes/1,000 μm², with a mean surface area of 4.43 ± 0.09 μm²; supporting information (SI) Fig. S1Aa] (14).

Fig. 1.

Ca²⁺ pulsars colocalized with IEL holes. (A) (a) IEL autofluorescence shows the presence of “holes” in the IEL. (b) Initiation sites of Ca²⁺ pulsars from the composite image correspond to holes in the IEL (red arrows). The yellow arrows indicate pulsar sites not associated with detectable IEL holes. (Scale bar, 10 μm.) (c) Histogram illustrating the distance between Ca²⁺ pulsar initiation sites and IEL holes in endothelium (n = 357 pulsar sites). (B) Time course of a three-dimensional Ca²⁺ pulsar originating from within an IEL hole (white circle) shown in the leftmost image. (Scale bar, 5 μm.) (C) Ca²⁺ pulsars from a pressurized artery (80 mmHg) expressing GCaMP2. (Ca) An endothelial cell and its nucleus are outlined (dotted lines), with the initiation sites (Cb and Cc) indicated by red arrows. (Scale bar, 10 μm.) See also Movie S1 and Movie S2

In unstimulated arteries from GCaMP2-expressing mice, we detected calcium spark-like events in proximity to the holes in the IEL (Figs. 1 and 2 and Movie S1). The majority of these events, which we have termed “Ca²⁺ pulsars,” occurred in or within 2 μm of holes in the IEL (Fig. 1Ac). Fig. 1A shows a field in which 21 of 27 pulsar sites (78%) localized to detectable holes in the IEL. Fig. 1B illustrates a single pulsar occurring in a single hole. There was no correlation between the size of the hole and the amplitude and frequency of the associated Ca²⁺ pulsars (Fig. S1 Ab and Ac).

Fig. 2.

Kinetics and repetitive occurrences of Ca²⁺ pulsars. (A) Average of 10 images of a field of endothelial cells from mesenteric arteries of a GCaMP2-expressing mouse. The red arrow indicates the initiation site of Ca²⁺ pulsars shown in B and C. (Scale bar, 10 μm.) (B) Life span of a Ca²⁺ pulsar is shown. The field of view corresponds to the green square in A (Scale bar, 10 μm.) (C) Repetitive occurrence of Ca²⁺ pulsars at one site expressed as a line-scan analysis along the yellow line in A. (Scale bar, 5 s.) (D) Representative traces illustrating Ca²⁺ pulsar kinetics originating from two different sites (red and blue lines). See also Movie S1

Calcium pulsars occurred repetitively at the same site (Fig. 2), with a mean time of 9.8 ± 1.0 s between pulsars at a given site (Fig. S1Ba). The interval between pulsars may be determined by local calcium depletion of the involved ER element or by the amount of calcium released during an event. However, there was no correlation between the amplitude of a Ca²⁺ pulsar and the latency to the next event at the same site (Fig. S1Bb), suggesting that neither of these mechanisms determines the timing of the next event.

Fig. 2A highlights a pulsar at a hole in the IEL, indicated by the arrow, which lasted <500 ms (Fig. 2 B and D). Overall, pulsars exhibited a mean duration of 270 ms and mean area of 14 μm² (Table 1). Pulsar properties (amplitude, spatial spread, kinetics, and frequency) were essentially the same when measured using the calcium-sensitive, fluorescent dye Fluo-4 (Table 1 and Fig. S2C), indicating that calcium-induced GCaMP2 conformation changes do not distort the kinetics of pulsars (7). Importantly, with GCaMP2 detection, Ca²⁺ elevation induced by the calcium ionophore, ionomycin (10 μM), was homogenous, indicating that the calcium biosensor was expressed evenly in the endothelium (Fig. S3). Endothelium-specific expression of GCaMP2 also provided the opportunity to examine Ca²⁺ signals in pressurized arteries without distortion from SM. Indeed, we detected Ca²⁺ pulsars in intact GCaMP2 mesenteric arteries subjected to physiological transmural pressure (80 mmHg). The pulsars occurred at the IEL holes (Fig. 1C and Movie S2), and their properties were similar to those observed in open arteries (Table 1 and Table S1).

Table 1.

Comparison of Ca²⁺ sparks from vascular SM cells, Ca²⁺ puffs from Xenopus oocytes, and Ca²⁺ pulsars from mesenteric endothelium

Parameters	Ca2+ sparks (26)	Ca2+ puffs	Ca2+ pulsars
			Fluo-4	GCaMP2
Amplitude, F/Fo	2.0	50–500 nM (27)	1.77 ± 0.10	1.70 ± 0.02
Area, μm2	13.6	2–4 (16, 27)	15.9 ± 0.6	14.1 ± 0.5
Rise time, ms	22.7	<100 (28)	212 ± 11	163 ± 6
Duration, ms	n/a	1,000 (16, 27)	257 ± 12	269 ± 6
t1/2, ms	55.9	375 (15)	146 ± 9	168 ± 5
Frequency, Hz	n/a	n/a	0.10 ± 0.02	0.08 ± 0.02

Descriptive parameters of Ca²⁺ signals in terms of peak amplitude, area measured at 50% of the amplitude of the peak, rise time as measured during 10–90% of the signal, duration as measured from 50% of the signal before and after the peak, and the half-time for decay (t_1/2) of the signal. Ca²⁺ pulsars differ from Ca²⁺ puffs in that their area and rise time are significantly greater, yet their duration and half-life are significantly less. Fluo-4 values are from 743 pulsars originating from 56 sites recorded in 4 fields, and GCaMP2 values are from 450 pulsars originating from 45 sites recorded in 4 fields. n/a, not applicable.

Calcium pulsars are distinct from IP₃R-mediated “calcium puffs” previously reported in Xenopus oocytes (15) and cultured cells (16), which have significantly different kinetic properties and spatial spreads (Table 1). “Calcium puffs” represent the opening of a small number of IP₃Rs that initiates regenerative Ca²⁺ waves and global calcium signals and do not have a fixed cellular location (16–18). Distinct from Ca²⁺ pulsars, calcium waves were present in individual endothelial cells in the exteriorized endothelium preparation (Fig. S2B and Fig. S4). Calcium waves traveled over 12 ± 2 and 9 ± 2 μm at a velocity of 48 ± 4 and 58 ± 6 μm/s, as measured with Fluo-4 (n = 57) and GCaMP2 (n = 17), respectively.

Calcium pulsars resemble RyR-mediated Ca²⁺ sparks observed in SM and cardiac muscle in that they occur at fixed locations with respect to the plasma membrane and have similar spatial spreads and amplitudes. Notably, however, Ca²⁺ pulsars have slower rise times and longer durations than Ca²⁺ sparks (Table 1). Treatment with the RyR inhibitor ryanodine, which completely blocks sparks in SM (1), had no effect on Ca²⁺ pulsars or other Ca²⁺ signals in endothelial cells (Fig. 3 C and F). In contrast, the IP₃R inhibitor, xestospongin C, reduced the frequency of Ca²⁺ pulsars in endothelium to 43 ± 7% of control (Fig. 3 D and F). The presence of IP₃R-mediated calcium pulsars in the absence of stimulation suggests that endothelial cells tonically produce IP₃ under basal conditions. Indeed, inhibition of phospholipase C (PLC) with U73122 or inhibition of calcium uptake by the ER with cyclopiazonic acid (CPA) significantly reduced calcium pulsar activity (Fig. 3 B, E, and F). Removing external calcium for 5 min did not significantly alter calcium pulsar activity (Fig. 3 A and F). Consistent with these functional observations, no RyR isoforms (RyR1–3) were detected in endothelial cells whereas mRNA for all IP₃R isoforms (IP₃R1–3) was detected in both the endothelium and the SM (Fig. S5), with IP₃R2 giving the strongest signal. Collectively, these results support the concept that Ca²⁺ pulsars are localized Ca²⁺ release events that are mediated by IP₃ activation of IP₃Rs in endothelial ER.

Fig. 3.

Ca²⁺ pulsars originating from IP₃-sensitive stores. Removal of extracellular Ca²⁺ (A) or ryanodine (C) did not affect Ca²⁺ pulsars. Inhibition of SERCA with CPA (B), of IP₃Rs with xestospongin C (D), or of PLC with U73122 (E) decreased Ca²⁺ pulsars. (F) A data summary of pharmacological experiments targeting the source of Ca²⁺ pulsars is given. (n = 6, 5, 4, 6, and 3 arteries for 0 Ca²⁺, CPA, ryanodine, xestospongin C, and U73122, respectively; *, P < 0.05). For A–E, different colors represent F/F_o in ROIs over different pulsar sites in the endothelium. Calcium pulsars were recorded for 2 min, followed by variable incubation times (0 Ca, 5 min; Ry, 35 min; CPA, 15 min; xestospongin C, 40 min; U73122, 15 min) and then 2 min of recording in the drug treatment (see Methods Summary).

The effects of PLC inhibition indicate that, in the absence of stimulation, Ca²⁺ pulsars are stimulated by basal IP₃ (Fig. 3 E and F). To increase IP₃ production, the endothelial agonist, acetylcholine (ACh) was applied. ACh increased the frequency of calcium pulsars by ≈2.4-fold; inhibition of ER calcium uptake (CPA) or PLC activity (U73122) blocked this ACh effect (Fig. S6 and Table S1). ACh-induced increase in Ca²⁺ pulsar frequency was due to both the recruitment of new sites and a reduced interval between pulsars at a given site (Fig. S1B). There was also no correlation between pulsar amplitude and latency to the next event at a given site with ACh (Fig. S1B). ACh also increased the frequency of Ca²⁺ waves (≈3-fold) and global Ca²⁺ (≈1.5-fold), consistent with previous reports (19, 20).

Our findings indicate that the ER in endothelial cells forms spatially discrete IP₃-sensitive Ca²⁺ stores localized in and around distinct IEL holes. To assess the distribution of ER at the endothelium–SM interface, we probed for the ER Ca²⁺-binding protein calnexin. Calnexin staining showed that the ER was distributed in the perinuclear region, along the base of endothelial cells, and most notably within dense plaques corresponding precisely to the positions of holes in the IEL (Fig. 4 B and C). Subsequent probing revealed a similar distribution of IP₃Rs, which were highly concentrated in both large and small IEL holes (Fig. 4 E and F). Multiple images captured through the depth of the IEL from the bottom surface of the endothelium to the top surface of the first SM cells revealed the contiguous presence of IP₃Rs in the holes through the depth of the lamina (Fig. 4G). These findings suggest the existence of an IP₃R-dense, inner-membrane ER structure that protrudes with the endothelial cell plasma membrane through the IEL and interfaces with SM cell membranes at the myoendothelial junction.

Fig. 4.

Localization of ER and IP₃Rs within IEL holes. Images show immunostaining for ER protein calnexin (red) (A–C) and for IP₃Rs (red) (D–F) at the level of the IEL (green). (C and F) Superimposed images reveal that ER and IP₃R are highly concentrated inside distinct IEL holes. (Scale bar, 5 μm.) (G) A three-dimensional view along the z axis (2.9 μm) shows densities of IP₃R-positive fluorescence (white) projecting through the depth of the IEL (green). (Scale bar, 1 μm.)

Calcium-sensitive potassium (K_Ca2.3 and K_Ca3.1) channels may be key targets of discrete calcium signals in the endothelium. Importantly, K_Ca3.1 channels are in close proximity to IEL holes (11) and appear to localize specifically within endothelial projections traversing the holes (Fig. 5C). To explore the possible communication of Ca²⁺ pulsars to K_Ca3.1 channels, we measured membrane potentials of endothelial cells in intact arteries. Blocking K_Ca3.1 channels with charybdotoxin (ChTX) depolarizes this preparation by ≈8 mV (Fig. 5B) (21). Blocking K_Ca2.3 channels with apamin produces a considerably smaller effect (≈3 mV) on endothelial membrane potential (21). CPA, which blocks calcium uptake into the ER, effectively inhibited calcium pulsars (Fig. 3 B and F and Fig. S6 B and C) and caused a transient (<15 min) increase in global calcium (27 ± 6%) (22), presumably through the depletion of ER calcium. On the basis of these disparate effects, we predicted that CPA should cause endothelial membrane potential hyperpolarization if global calcium activates K_Ca3.1 channels and should cause depolarization if localized calcium pulsars activate K_Ca3.1 channels. CPA caused a ≈10-mV depolarization (Fig. 5Ab) but had no effect when K_Ca3.1 channels were blocked (Fig. 5B). CPA did not affect endothelial membrane potential in the presence of ChTx (Fig. 5B). These results suggest that Ca²⁺ pulsars, and not global Ca²⁺, directly activate K_Ca3.1 channels in the endothelium, and lend support to the concept that the K_Ca3.1 channel-dependent component of endothelial membrane potential is regulated by Ca²⁺ pulsars.

Fig. 5.

Ca²⁺ pulsars hyperpolarizing the endothelium membrane through activation of K_Ca3.1 channels. (A) (a) CPA (10 μM; black bar) depolarizes the endothelial membrane potential, and subsequent addition of ChTX (300 nM; white bar) had no effect. (b) Summary of membrane potential experiments with CPA using microelectrode and perforated patch techniques. Additionally, in microelectrode experiments, CPA exposure was followed by ChTX. Subsequent exposure to 60 mM KCl depolarized the endothelial membrane potential to −20 ± 2 mV. n = 6 and 4 for microelectrode and perforated patch recordings, respectively. (B) (a) ChTX (300 nM; black bar) depolarizes the endothelial membrane potential, and subsequent addition of CPA (10 μM; white bar) had no effect. (b) A summary of microelectrode experiments with ChTX followed by CPA is given. Subsequent exposure to 60 mM KCl depolarized the endothelial membrane potential to −22 ± 2 mV (n = 5; *, P < 0.05). All microelectrode experiments were carried out in the presence of paxilline (500 nM) and nitrendipine (100 nM). (C) (a and b) Immunostaining for K_Ca3.1 (red) at the level of the IEL (green) is shown. (c) The superimposed image reveals distinct densities of K_Ca3.1 within and around IEL holes. (Scale bar, 5 μm.) (d) A three-dimensional view along the z axis (3.1 μm) shows densities of K_Ca3.1-positive fluorescence (white) projecting through the depth of the IEL (green). (Scale bar, 1 μm.)

Discussion

Our results support the concept of a IP₃R signaling structure with a profound functional bearing on endothelial–SM intercellular communication (Fig. 6). In a process reminiscent of neuronal projections to target cells, vascular endothelial cells send projections through the IEL that contact SM cells. IP₃Rs localized to these membrane projections mediate local Ca²⁺ release events (pulsars), establishing a mechanism for Ca²⁺-dependent signaling from endothelial cells to SM cells. One target of calcium pulsars appears to be K_Ca3.1 channels that are colocalized to endothelial projections (Fig. 5C) (11). Activation of endothelial K_Ca3.1 channels (and K_Ca2.3 channels) is a common denominator of the various EDHF mechanisms that communicate dilating influences from endothelial cells to SM (3). Prominent among these mechanisms is direct electrical communication via myoendothelial gap junctions and direct activation of inward rectifier potassium channels on SM by released potassium ions (Fig. 6) (2, 3, 23). Thus, a minimal intercellular functional unit is likely composed of endothelial IP₃Rs, K_Ca3.1 channels, and gap-junction-forming connexins and inward rectifier potassium channels (Kir2.1) and voltage-dependent calcium channels (Ca_v1.2) on the SM (Fig. 6). Other endothelial calcium-dependent processes (e.g., endothelial nitric oxide synthase and phospholipase A₂) also may be activated by calcium pulsars or differentially activated by other calcium signals (global Ca²⁺ or waves). We propose that the Ca²⁺ pulsar is a fundamental endothelial Ca²⁺ signal whose activity is finely regulated by physiological agents (i.e., agonists and flow) that modulate intracellular levels of IP₃ and Ca²⁺. Given their unique, restricted localization to points of contact with vascular SM, endothelial Ca²⁺ pulsars likely encode a variety of bidirectional endothelium–SM signals (4, 13). Consequently, disruption of this IP₃R signaling process may be a hallmark of endothelial dysfunction observed in virtually all cardiovascular diseases.

Fig. 6.

Model of endothelial Ca²⁺ pulsars. IP₃R-dense ER stores follow portions of the endothelial cell membrane that evaginate through holes in the IEL and interface with underlying SM cell membranes. Repetitive localized Ca²⁺ events (pulsars) originate from these deep Ca²⁺ stores that are regionally delimited to the myoendothelial junction and the base of the endothelial cell. These ongoing dynamic Ca²⁺ signals are driven by constitutive IP₃ production and are inherently dependent on the level of endothelial stimulation. The left detail depicts a single endothelial projection through the IEL. K_Ca3.1 channels in the plasma membranes of these endothelial projections are in very close proximity to Ca²⁺ pulsars, eliciting persistent Ca²⁺-dependent hyperpolarization of the membrane potential at the myoendothelial junctions. The right detail illustrates the endothelial influence on the SM membrane potential at the myoendothelial interaction site where Ca²⁺ pulsars would activate K_Ca3.1 channels and hyperpolarize the endothelial membrane. This hyperpolarization can be transmitted to the SM through gap junction channels or by activation of SM K_ir channels by K⁺ ions released by endothelial K_Ca3.1 channels. Membrane hyperpolarization promotes relaxation of SM through a decrease in voltage-dependent calcium channel open probability.

Methods Summary

Animal Procedures.

Animal procedures used in this study are in accord with institutional guidelines and were approved by the Institutional Animal Care and Use committee of the University of Vermont. Mesenteric arteries (≈125 μm in diameter) were freshly harvested from 3- to 4-month-old C57BL6- or GCaMP2-expressing mice. We used bacterial artificial chromosome transgenesis (7, 24), expressing the circularly permutated, calcium sensor, G-CaMP2, under the control of the connexin40 (Cx40) promoter; Cx40 is expressed in vascular endothelium but not vascular SM (25). Adult female mice were euthanized by intraperitoneal injection of sodium pentobarbital (150 mg/kg) followed by a thoracotomy. Third-order mesenteric arteries from mice were cleaned of connective tissue, cut longitudinally, and pinned to a Sylgard block with the endothelium facing up.

Endothelial Ca²⁺ Imaging.

Ca²⁺ imaging was performed with a Revolution Andor confocal system (Andor Technology) with an electron-multiplying CCD camera on an upright Nikon microscope with a ×60, water-dipping objective (NA 1.0). Images were acquired at 15–30 frames per second with Andor Revolution TL acquisition software (Andor Technology). Bound Ca²⁺ was detected by exciting at 488 nm with a solid-state laser and collecting emitted fluorescence above 510 nm. Fractional fluorescence (F/F_o) was evaluated by dividing the fluorescence of a region of interest (ROI) in the collected image by an average fluorescence of 50 images without activity from the same ROI using custom-designed software (A.D.B., unpublished data). The endothelial cell surface was measured by automated analysis of the area enclosed by a freehand ROI drawn around the outline of individual endothelial cells. Global Ca²⁺ levels were measured over the entire area of a cell, defined by the freehand ROI outline. Ca²⁺ pulsars were analyzed by using an ROI defined by a 5 × 5 pixel box positioned at a point corresponding to peak pulsar amplitude. Line-scan analysis was performed offline. In preparations from wild-type mice (i.e., non-GCaMP2-expressing), endothelial cells were preferentially loaded with Fluo-4 (10 μM) for 45 min at 30°C in the presence of pluronic acid (2.5 μg/ml) before imaging, as previously described (21). The field of view was ≈115 × 137 μm, corresponding to ≈25–30 partial and whole cells and 13 active cells per field. There were 89 ± 7 holes per field, and generally between 10 and 28 active sites were identified in each field.

Solutions, Pressurized Arteries, Immunohistochemistry, Membrane Potential Recording, Reverse-Transcriptase PCR, and Statistics.

These methods are described in SI Text.