Waves of Calcium Depletion in the Sarcoplasmic Reticulum of Vascular Smooth Muscle Cells: An Inside View of Spatiotemporal Ca2+ Regulation

Agonist-stimulated smooth muscle Ca2+ waves regulate blood vessel tone and vasomotion. Previous studies employing cytoplasmic Ca2+ indicators revealed that these Ca2+ waves were stimulated by a combination of inositol 1,4,5-trisphosphate- and Ca2+-induced Ca2+ release from the endo/sarcoplasmic reticulum. Herein, we present the first report of endothelin-1 stimulated waves of Ca2+ depletion from the sarcoplasmic reticulum of vascular smooth muscle cells using a calsequestrin-targeted Ca2+ indicator. Our findings confirm that these waves are due to regenerative Ca2+-induced Ca2+ release by the receptors for inositol 1,4,5-trisphosphate. Our main new finding is a transient elevation in SR luminal Ca2+ concentration ([Ca2+]SR) both at the site of wave initiation, just before regenerative Ca2+ release commences, and at the advancing wave front, during propagation. This strongly suggests a role for [Ca2+]SR in the activation of inositol 1,4,5-trisphosphate receptors during agonist-induced calcium waves. In addition, quantitative analysis of the gradual decrease in the velocity of the depletion wave, observed in the absence of external Ca2+, indicates continuity of the lumen of the sarcoplasmic reticulum network. Finally, our observation that the depletion wave was arrested by the nuclear envelope may have implications for selective Ca2+ signalling.


Introduction
In vascular smooth muscle cells (VSMCs), fluctuations in cytoplasmic Ca 2+ concentration ([Ca 2+ ] i ) selectively control multiple functions, including contraction-relaxation, energy metabolism, proliferation, migration and apoptosis, in health and disease [1,2,3,4,5]. It is generally accepted that functional selectivity of Ca 2+ signals is encoded in their spatial and temporal characteristics [6,7,8]. In this context, we support the view that the ubiquitous asynchronous Ca 2+ waves in VSM are optimally suited to couple agonist-mediated stimulation to vasoconstriction in healthy blood vessels, while avoiding recruitment of stress related functions [2,3,9], [10]. Neylon and coworkers were the first to report that receptor stimulation of cultured human VSM did not simply elevate [Ca 2+ ] i to induce activation, but in fact caused a wave of elevated [Ca 2+ ] i to travel across the cell from an initiation site [11]. They further proposed a mechanism involving stimulation of inositol 1,4,5trisphosphate receptors (IP 3 R) and Ca 2+ induced Ca 2+ release (CICR) [11]. In 1994, Iino et al. made the next advance by recording [Ca 2+ ] i in the intact rat tail artery smooth muscle [4]. Iino had earlier discovered that IP 3 Rs were sensitive to Ca 2+ , such that IP 3 required Ca 2+ as a co-activator, and that the elevation of [IP 3 ] i facilitates CICR at IP 3 Rs [12]. Sympathetic nerve stimulation initiated asynchronous repetitive waves of [Ca 2+ ] i elevation, traveling in both directions along the length of the spirally arranged smooth muscle fibres. The unexpected aspect of this discovery was that the tonic increases in both the bulk smooth muscle [Ca 2+ ] i and force were based on wave-like [Ca 2+ ] i oscillations in individual VSMCs. Because of the asynchronous nature of these cellular Ca 2+ waves, summation over thousands of cells in the vascular media results in maintained average [Ca 2+ ] i elevation and vascular tone. Nevertheless, the repetitive Ca 2+ waves would still confer the advantage of added informational content due to frequency encoding and prevention of potential harm due to prolonged elevated [Ca 2+ ] i .
Although a great deal of valuable information has subsequently accumulated on the mechanism of Ca 2+ waves, previous studies were limited, to some degree, by the fact that they reported changes in cytoplasmic Ca 2+ , which represent the result of Ca 2+ release from the sarcoplasmic reticulum (SR) rather than the process itself. In this study, in order to obtain more direct insight into regenerative Ca 2+ release in VSMCs, we have used a specific Ca 2+ indicator targeted to calsequestrin in the SR lumen that we refer to as D1SR. The VSMCs used in this study constituted a primary culture, which remained highly contractile during the study. Here, we report a wave of Ca 2+ depletion from the SR itself, which in this case is activated by applying the vascular autocoid endothelin-1 (ET-1). The observed agonist-induced SR Ca 2+ The zoomed-in images highlight areas of SR lumen negative for D1SR signal, but positive for ER-Tracker. In all experiments, such areas were excluded from data analyses (scale bars, 10 mm unless indicated otherwise). C) SMCs are loaded with Golgi-GFP CellLightH solution for 16

Buffers & Reagents
HEPES-PSS containing (in mmol?L 21 ) NaCl 140, glucose 10, KCl 5, HEPES 5, CaCl 2 1.5 and MgCl 2 1 (pH 7.4) was used for all calcium measurements and confocal microscopy. The nominal zero-Ca 2+ PSS was prepared in the same way as normal PSS without the addition of calcium. Endothelin-1 (ET-1) was obtained from Sigma-Aldrich (ON, Canada). Thapsigargin (Tg), a cell permeable inhibitor of sarcoplasmic reticular Ca 2+ -ATPase (SERCA), and xestospongin C (Xes-C), a potent membranepermeable blocker of IP 3 -mediated Ca 2+ release, were purchased from EMD Chemicals (NJ, USA). Sodium orthovanadate, an inhibitor of secretory pathway Ca 2+ ATPase (SPCA) pumps, was purchased from Sigma-Aldrich. Stock solutions of ET-1 and Tg were prepared in dimethyl sulfoxide (DMSO). For all experiments, vehicle-treated (0 mM) groups were incubated with equal volume of DMSO, respectively (the maximum volume of solvents used with the highest concentration of drugs). Further dilutions of reagents were made in zero-Ca 2+ PSS buffer. The calcium dye Fluo-4 AM, ER-Tracker, Mito-Tracker Green FM, BODIPYH TR-X thapsigargin, and CellLight TM Golgi-GFP BacMam 2.0 were all purchased from Invitrogen (ON, Canada).

Cell Culture
Rat aortic SMCs were originated from the laboratories of Drs. Urs Ruegg and Nicolas Demaurex (University of Geneva, Geneva, Switzerland), and prepared from aorta of male Wistar Kyoto rats (200-300 g) as previously described [13]. Cells between passage 9 to 13 were cultured in Matrigel-coated (BD Sciences; ON, Canada) 35 mm glass bottom culture dishes (MatTek Co., MA, USA), and grown in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated newborn calf serum, Penicillin G (100 mg.mL 21 ), and streptomycin (100 mg.mL 21 ) (Invitrogen, ON, Canada) at 37uC, in a humidified incubator in 5% CO 2 . See following sections for a detailed description of the transfection protocol and buffer/reagents used in the study.

Transient Transfection with D1SR Constructs
D1SR indicator, a modified variant of the D1ER cameleon [14,15,16], is kindly provided by Dr. Wayne Chen (University of Alberta, Canada). In the D1SR indicator, the original calreticulin signal sequence in D1ER has been replaced by the mutant calsequestrin sequence with a reduced binding ability to calcium (to eliminate the competition with endogenous calsequestrin in binding to calcium ion within SR). The D1SR construct consists of a truncated enhanced CFP and YFP that are joined by a linker containing modified calmodulin (CaM) and M13 (the 26-residue CaM-binding peptide of myosin light-chain kinase) sequences. The CaM-M13 modifications prevent M13 from binding endogenous calmodulin. SR retention is achieved by calcequestrin sequences on the 59 end of CFP. Following binding to Ca 2+ , conformational changes in the CaM-M13 domain increase the fluorescence resonance energy transfer (FRET) between the flanking CFP and YFP yielding a Ca 2+ response. SMCs were transfected with adenoviral D1SR constructs at a multiplicity of infection of 100 (MOI = 100). Following overnight incubation at 37uC, cells were replenished with fresh medium. Fluorescence microscopy was used to assess transfection efficiency and cellular

SR Luminal Ca 2+ Measurement
Ratiometric FRET images were acquired with a 636 oilimmersion objective (Leica DM16000 inverted microscope) and a cooled Hamamatsu 9100-02 electron multiplier CCD camera. Cells were excited at 440 nm, and 513 nm; and imaged using 488 nm/535 nm (for donor and FRET channels) and 535 nm (for acceptor channel) band-pass filters (three images/time point). The acceptor channel was simultaneously recorded to monitor photobleaching. We set the intensity of light at 15% transmitted light, and excitation exposure times at 150 ms with 10 s intervals.

Cytoplasmic Ca2+ Measurement
During in vitro measurement of cytoplasmic Ca 2+ signals, all parameters (laser intensity, gain, etc.) were maintained constant during the experiment. The cell culture was illuminated using an Argon-Krypton laser (488 nm) and a high-gain photomultiplier tube collected the emission (505-550 nm). The customized Hamamatsu 9100-02 electron multiplier CCD camera delivers 100061000 pixels, imaged to Shannon-Nyquist specifications for the 636 objectives, providing a larger field of view. The representative fluorescence traces shown reflect the averaged fluorescence signals from 15 regions of interest (ROIs) in each cell. The measured changes in Fluo-4 AM fluorescence level are proportional to the relative changes in cytoplasmic Ca 2+ ([Ca 2+ ] i ). The confocal images were analyzed off-line with the Improvision Volocity software (Perkin-Elmer). Fluorescence traces were extracted from the movies and were normalized to initial fluorescence values.

Image Analysis
All data used for Ca 2+ traces were analyzed by Improvision Volocity software, using built-in regions-of-interest (ROI) function to select the areas of interest that is at least 30 pixels long in length and width. For traces involving specific sites, shape and size of ROI were adjusted to avoid artifacts and saturated areas. For distance and velocity analysis, one ROI was selected per frame (time point) based on the movement of the Ca 2+ wave. The resulting data were formatted on Microsoft Excel 2007, and analyzed by GraphPad Prism 5.0. Pseudo-colour visualization was performed by ImageJ, using customized lookup tables to assign colour for each pixel intensity values. Line scan was performed first by analyzing pixel intensity of a series of small ROIs (1-3 pixels in length per ROI) along a line using a customized Python script to output change in intensity over time along the line. The script was confirmed to output same values when analyzing same ROI on Volocity. The resulting values were then graphed using Gnuplot 4.4 with custom lookup table.

Statistical Analysis
Data are presented as means 6 SEM of at least four independent experiments. Significance was determined using Student's t-test with two-tailed distribution using GraphPad Prism 5 (P#0.05 was considered as statistically significant).

Distribution of D1SR Ca 2+ Indicator in Rat Aortic SMCs
To monitor fluctuations in [Ca 2+ ] SR , we transfected rat aortic SMCs with an adenoviral vector expressing the specific FRETbased SR Ca 2+ indicator, D1SR. Figure 1A shows the individual and merged fluorescence images of three recorded channels (CFP: 440 nm/488 nm; FRET: 440 nm/535 nm; YFP: 513 nm/ 535 nm) for D1SR indicator in cultured VSMCs (see Movie S1). We also compared the distribution of D1SR with that of ER-Tracker, commonly used for localization of either ER or SR. As expected, D1SR co-localizes with ER-Tracker except for a region close to the nucleus that is negative for D1SR fluorescence, but positive for ER-Tracker (Fig. 1B). Specific staining with Golgi-Tracker indicated that the Golgi apparatus is typically located in this region (Fig. 1C).

D1SR Calibration in situ
The D1SR indicator was calibrated in situ in semi-permeablilized rat aortic SMCs in intracellular HEPES solution (135 mM KCl, 10 mM NaCl, 1 mM MgCl 2 , 20 mM HEPES, 20 mM sucrose, 0.01 mM digitonin, 0.01 mM ionomycin, 0.005 mM CCCP, PH 7.2) with 5 mM HEDTA as described before [16]. The intracellular solution was blended with the stock solution of 100 mM CaCl 2 to prepare buffers with free Ca 2+ concentrations ([Ca 2+ ] free ) ranging from 1 mM to 10 mM (Max Chelator v2.40, C. Patton, Standford University, USA, maxchelator.stanford.edu). D1SR fluorescent ratio (R/R 0 ) was plotted against [Ca 2+ ] free in 12 cells from four independent cultures and an exponential fit was applied to the data to determine the apparent dissociation constant K d (197647 mM) and the Hill slope (n = 0.8760.19) ( Fig. 2A).
[Ca 2+ ] SR was calculated by calibrating normalized D1SR ratio against the standard calibration curve as described previously [16]. To calculate [Ca 2+ ] SR, normalized ratio values were fitted to: where R min and R max were obtained by measuring signal intensity in cells perfused with intracellular HEPES solution containing 5 mM HEDTA (Calcium free buffer) and 10 mM free calcium (15 mM CaHEDTA), respectively.

Effects of Endothelin-1 and Thapsigargin on SR and Cytoplasmic Ca 2+ Signals
The traces in Figure 2B illustrate the effects of receptor activation with ET-1 (100 nM) and SERCA inhibition with 2 mM thapsigargin (Tg) on [Ca 2+ ] SR in the peripheral (marked with red circle) and peri-nuclear (marked with black circle) areas of SMCs.  [Ca 2+ ] SR is that in these cells the SR is contiguous with membranous organelles or endosomes, which accumulate Ca 2+ via the Tg-insensitive ''secretory pathway Ca 2+ ATPase'' (SPCA). An alternative explanation is provided by a recent report from Ledeen's group [17], which showed that NCX located in the inner membrane of the nuclear envelope is able to take up Ca 2+ after SERCA blockade. Furthermore, since the nuclear envelope and the SR are confluent, this Ca 2+ , taken up by the NCX is capable of diffusing into the peri-nuclear SR.
In order to compare the changes in [Ca 2+ ] SR with fluctuations in [Ca 2+ ] i , we used the cytoplasmic Ca 2+ indicator fluo-4 AM in the same VSMC preparation (Fig. 3). In these cells, ET-1 (100 nM) appears to be a weak agonist eliciting only a transient increase in [

ET-1-induced Regenerative SR Ca 2+ Depletion Waves in the Absence of Extracellular Ca 2+
In order to distinguish the effects of ET-1 on luminal SR Ca 2+ from those related to ET-1 stimulated Ca 2+ entry from the extracellular space, we removed Ca 2+ from the bathing solution, but without adding a chelator. Under these conditions, which inhibit SR Ca 2+ refilling [18], ET-1 caused a large and delayed drop in [Ca 2+ ] SR (Fig. 4). The difference between ET-induced responses in the presence and absence of external Ca 2+ is not surprising, because in the presence of external Ca 2+ , there is rapid Ca 2+ cycling between the SR lumen and extracellular space [19], which appears to protect the SR from excessive Ca 2+ depletion. In contrast, in nominally Ca 2+ free condition, which inhibits SR refilling, ET-1 causes a much larger drop in [Ca 2+ ] SR .
In about 80% of the cells, application of ET-1 after removal of external Ca 2+ revealed a novel phenomenon that we refer to as the ''SR Ca 2+ depletion wave'' ( Fig. 4 and Movie S2). At 300 s, Ca 2+ is removed from the bathing solution, and at 600 s, ET-1 (100 nM) is added. In the presence of ET-1, focal fluctuations in [Ca 2+ ] SR can be observed, and just before the onset of the Ca 2+ depletion wave, [Ca 2+ ] SR is elevated at the site of wave initiation (Fig. 4,  panel 1200 s). The SR depletion waves routinely originated close to the peri-nuclear region, rather than in the cell periphery.
Measurements of changes in the R/R 0 values at the point of wave initiation, confirmed a transient increase in [Ca 2+ ] SR to 7006100 mM immediately prior to wave initiation (Fig. 5A). Subsequently a depletion of [Ca 2+ ] SR occurs at the site of initiation (from 7006100 mM to 60620 mM), from which a wave of Ca 2+ depletion radiates out into the surrounding SR network (Fig. 4, panels 1210-1400 s). In the absence of external Ca 2+ the extent of ET-1-induced [Ca 2+ ] SR depletion is similar to that induced by Tg. The depletion wave progresses in a roughly circular fashion and is always surrounded by a rim of elevated [Ca 2+ ] SR . It thus appears that some of the Ca 2+ released at the wave front is taken up by an adjacent, but not-yet-activated SR locus. Figure 5B shows that at any given time point of wave expansion, there are significant differences in [Ca 2+ ] SR between the depleted SR region (marked as A), the transiently refilled neighbouring SR locus (marked as B), Endothelin-1 Induced Calcium Depletion Wave PLOS ONE | www.plosone.org and the next SR locus further ahead of the depletion wave (marked as C). After passage of the ET-induced Ca 2+ depletion wave, the SR was capable of refilling upon subsequent reperfusion with normal (1.5 mM) Ca 2+ solution, confirming the reversibility of the observed SR depletion wave (Fig. 6A). Furthermore, the depletion waves recorded herein did not cause any long-lasting structural changes in SMCs, and the SR structure remained intact during the experiment (Fig. 6B).

Ca 2+ Depletion Waves in SMCs Propagate via IP 3 Receptors
A plausible explanation for the initiation of SR Ca 2+ depletion waves is that after the IP 3 concentration builds up to a critical level and the cytoplasmic and SR Ca 2+ concentrations at the initiation site reach a certain threshold values, the IP 3 Rs on the SR membrane open in a regenerative fashion; the wave is then propagated by CICR at IP 3 Rs. To test this hypothesis, we blocked the IP 3 Rs with a specific inhibitor xestospongin C (Xes-C) prior to the application of ET-1 (Fig. 7). Addition of Xes-C (1 mM) completely blocked ET-induced depletion waves, confirming that as for other VSM Ca 2+ waves [8,18,20], the underlying mechanism is indeed mediated by CICR at IP 3 R. This notion was further corroborated by the observation that our cultured SMCs failed to respond to caffeine (data not shown), confirming the lack of functional RyRs in these cells, which is in agreement with a previous report by Vallot et al [21].

Dynamics of the Ca 2+ Depletion Waves in the SR Lumen
Another instructive way of analyzing the [Ca 2+ ] SR depletion wave is to identify a line intersecting the site of wave initiation and recording the changes in R/R 0 values along this line vs. time (line scan). The resulting ''heat map'' illustrates that initially the area of [Ca 2+ ] SR depletion expands rapidly, but as time proceeds, it asymptotically approaches a final limit (Fig. 8A). Analysis of a number of such heat maps yields the average distance vs. time, as well as the velocity vs. time curves (Fig. 8B). Extrapolation of the velocity curve to the time of wave initiation (0 s), shows that the SR Ca 2+ depletion wave has an initial velocity of about 0.7 mm/s, which falls below the range (2-30 mm/s) reported previously for intact smooth muscle Ca 2+ waves in the presence of extracellular Ca 2+ [22]. This could be explained by the fact that in our cultured VSMCs, ET-1 is a weak agonist, and a number of studies have shown that both the velocity and frequency of smooth muscle Ca 2+ waves increase with the level of activation and the concentration of IP 3 [8,23]. The profound decrease in velocity over time is likely related to the decline of the [Ca 2+ ] SR at the rim as the wave progresses (Fig. 8C).

Quantitative Model for Propagation of Ca 2+ Depletion Waves
The observed transient increase in [Ca 2+ ] SR at both the origin and the rim of the depletion wave ( Fig. 5A and 5B) suggests an important role for local [Ca 2+ ] SR elevation during initiation and propagation of regenerative IP 3 R-mediated Ca 2+ release. Since in cytoplasmic Ca 2+ oscillations, the latency period is related to the inter-spike interval [24], which is, in turn, regulated by the SR Ca 2+ content [24], the VSMCs activity may well be controlled by focal fluctuations in [Ca 2+ ] SR . In line with previous reports [25,26], we propose that these fluctuations are generated in part by a differential distribution of SR luminal Ca 2+ sinks (clusters of IP 3 Rs), and Ca 2+ sources (SERCA) (Fig. 9A). Stochastic opening of individual IP 3 Rs (which yield cytoplasmic Ca 2+ blips) [25,26] would add variability to focal [Ca 2+ ] SR . Whenever an increase in local [Ca 2+ ] SR exceeds the threshold for regenerative opening of an IP 3 R cluster, sufficient Ca 2+ would be released to initiate a regenerative Ca 2+ wave.
Generation of a wave according to this mechanism, and possessing the features of deceleration and decreasing intensity at the progressing rim as we observe, implies a continuity of the SR lumen. Since under control conditions the non-activated SR remains loaded with Ca 2+ because of continual activity of SERCA, the observed decline in [Ca 2+ ] SR at the rim of the depletion wave (Fig. 8C) is a strong indication that a re-equilibration of the SR Ca 2+ content is taking place within the SR lumen. Only in this manner, when removal of extracellular Ca 2+ blocks SR refilling, then opening of IP 3 Rs at the wave front would partially deplete the SR located just ahead of the wave.
To analyze this proposed mechanism, we developed a preliminary and simplified two-dimensional quantitative model for the propagation of the observed depletion waves, which is based on previous observations and suggested models arguing for 1) a continuous SR lumen in regards to Ca 2+ transport, 2) the appearance of functionally segregated compartments in the SR of SMCs, and 3) the existence of a luminal Ca 2+ binding site of the IP 3 R [27,28,29]. The essential steps of this model are presented in Figure 9B, and the generated trace for the velocity of the depletion wave is presented in Figure 9C.
Using the symbols and data in Table 1, our model assumes that, before any Ca 2+ depletion wave (CDW) event, the [Ca 2+ ] SR is at a normal resting level (C SR ) and that it needs to reach a critical level (C * SR ) for release. Then, for example, following the depletion of a given SR compartment, for a nearest neighbouring compartment to release, its [Ca 2+ ] SR needs to change by C * SR 2C SR = 200 mM, which translates to about 1500 Ca 2+ . For the sake of order-ofmagnitude calculations, let us say that SERCA pumps operating at 300 s 21 would take a time interval Dt 1 = 5 s to cause a 200 mM [Ca 2+ ] SR change. We calculate the depletion wave velocity, v CDW , as the inter-IP 3 R-cluster distance, d, divided by the time, Dt, taken to raise the [Ca 2+ ] SR in the new compartment from normal to critical: v CDW = d/Dt. For the sake of completeness, the wave velocity should also include a component for the Ca 2+ diffusion time from cluster to cluster. This component is however of the order of milliseconds and therefore negligible when compared to SR compartment refill times.
In this manner, for one depletion wave ''step'' in our model (from compartment 0 to 1 in Figure 9B), v CDW,1 = d/Dt 1 = 0.1 mm/   A wave initiating at compartment 0 in this model would move both left and right, but we focus only of the right-moving part for simplicity. Assuming that the [IP 3 ] cyt is such that the IP 3 R is already Ca 2+ -sensitized, a [Ca 2+ ] i fluctuation near the SR compartment 0 at t = t 0 raises the [Ca 2+ ] SR from a resting level to a threshold level, and thereby, causes the observed transient rise (B-1). This rise is followed by depletion of Ca 2+ from compartment 0. Partial equilibration of luminal Ca 2+ then takes place (blue block arrows between compartments), accompanied by an increase in [Ca 2+ ] SR in compartment 1 (B-2). Subsequently, [Ca 2+ ] SR in compartment 1 reaches a threshold value leading to the releases of (some of) its Ca 2+ , which goes on to partially refill its nearest neighbouring locus (compartment 2), and so on (B-3). Ca 2+ release from compartment 1 can trigger release from compartment 2, only after [Ca 2+ ] SR will have reached a critical level for Ca 2+ release, which will take a time interval Dt 2 = t 2 2t 1 . Due to the Ca 2+ passage from compartment 2 to 1, Dt 2 must be greater than Dt 1 , the interval it took to raise [Ca 2+ ] SR in 1 to the release level. Successive compartment refilling times to release level will get progressively longer because of the ability to re-equilibrate Ca 2+ depletions. Assuming now that the CICR sustaining IP 3 R clusters are on average equidistant at a length d (500 nm) from one another, the Ca 2+ depletion wave velocity (v CDW ), as the wave reaches successive compartments, is given by s. If we now assume that SR compartments 2, 3, etc lose part of their Ca 2+ toward the depleted neighboring SR compartments by the percentages indicated in Table 1 (these values are arbitrarily chosen), their ''normal'' [Ca 2+ ] SR will be C 2,SR , C 3,SR , etc., where these values are lower than C SR . The same line of reasoning then suggests that, for compartment 2 to release, its [Ca 2+ ] SR needs to change by C * SR 2 C 2,SR .200 mM, which will take a time Dt 2 .Dt 1 and, in turn, yields a velocity for the second step as v CDW,2 = d/ Dt 2 ,v CDW,1 . To consider what would happen in a portion of SR with Ca 2+ tight compartments, to a first degree of approximation, we can re-use the model above with 0% Ca 2+ communication between compartments. It should be evident then that the interval needed to refill each compartment before release is the same as the putative wave progresses. Since the average inter-cluster distance is the same, the resulting wave velocity would be a constant value v CDW = d/Dt at each step. The results reported in Figure 9C were obtained by following this step-by-step procedure for about 100 seconds, which confirms that introduction of such luminal Ca 2+ flux into the model indeed caused incremental decrease of the simulated Ca 2+ depletion wave velocity (Fig. 9C, red trace).
Our proposed model clearly depends on populations of IP 3 Rs and SERCA spread over the entire SR network. This appears to be the case for the SMCs used in this study as both SERCA (Fig. 10A) and IP 3 Rs (Fig. 10B) display the same diffuse distribution as the D1SR and ER-Tracker.

Arrest of the ET-induced SR Depletion Wave at the Border of the Nuclear Envelope
Close inspection of the SR depletion waves also provides novel insight into sub-cellular differential Ca 2+ signalling. Figure 11A presents snapshots of a depletion wave, which originates very close Table 1. Symbols and data for the quantitative Model.

Quantity Data References
SR lumen cross-section 50 nm650 nm [46,47] IP 3 R cluster-cluster distance, d 500 nm [46] SR compartment volume 50 nm650 nm6500 nm [46,47,48] SERCA refill rate 300 s 21 [46] Normal  to the nucleus (see Movie S3). In this particular case, the shape of the wave front becomes oblong instead of circular, as it fails to involve the nuclear envelope (white arrows), and proceeds in the opposite peripheral direction. The asymmetry of this particular wave is clearly demonstrated by the heat map in Figure 11B. As shown, the distance traveled by the wave and its velocity drop dramatically when the wave collides with the nuclear envelope, without depleting its luminal Ca 2+ content. In contrast, the pattern Endothelin-1 Induced Calcium Depletion Wave of two neighbouring SR depletion waves colliding with each other is clearly different from the case when a depletion wave encounters the nuclear envelope. As shown in Figure 12, in some cells, multiple Ca 2+ depletion waves develop independently. In these cells, a line scan of the Ca 2+ signal through the point of origin and through the border between neighbouring waves (a & b) shows that the depleted areas of colliding waves coalesce (white arrow) without a dividing line of Ca 2+ loaded SR as seen at the periphery of both depletion waves (red arrows).

Discussion
In this first report of Ca 2+ depletion waves in the smooth muscle sarcoplasmic reticulum, we describe a crucial role for luminal [Ca 2+ ] SR in the initiation and propagation of cytoplasmic Ca 2+ waves. This additional insight was gained because the dynamics of SR Ca 2+ depletion do not simply mirror [Ca 2+ ] i elevations, but display their own unique characteristics. The disparity between the [Ca 2+ ] i and [Ca 2+ ] SR transients is most likely due to significant contributions by plasma membrane Ca 2+ fluxes to smooth muscle Ca 2+ homeostasis and excitation.
Iino's hypothesis that Ca 2+ waves in VSM are propagated by regenerative CICR at the IP 3 Rs [4] was recently corroborated by a well-controlled study by McCarron and collaborators on freshly isolated SMCs, voltage clamped and incubated in a Ca 2+ free medium [22]. Progression of the wave front by CICR requires that the released Ca 2+ raises the local [Ca 2+ ] i in the neighbouring nano-domain occupied by IP 3 Rs, which are about to be activated. Although such spatio-temporal resolution of cytoplasmic Ca 2+ waves has not yet been reported, our data conclusively show that this is indeed the case. The only possible explanation for the increased [Ca 2+ ] SR at the wave rim is that the SERCA pumps respond to the local [Ca 2+ ] i elevations before activation of IP 3 R ahead of the Ca 2+ wave. Our observation of a diffuse distribution of both SERCA and IP 3 R over the entire SR (see Fig. 10) corroborates this view. Moreover, our conclusions show Ca 2+ diffuses from an elevated [Ca 2+ ] i site, to a neighbouring site containing an IP 3 R cluster more rapidly than the propagation of a depletion wave. This supports the mechanism of a wave of CICR, and is experimentally confirmed by the increased [Ca 2+ ] SR at the wave rim due to SERCA-mediated Ca 2+ uptake in response to the local [Ca 2+ ] i elevation before activation of IP 3 Rs, ahead of the Ca 2+ wave.
The transient elevations of [Ca 2+ ] SR at the site of wave initiation and at the wave front, directly demonstrate local interactions between adjacent SR sites, which depend on SR luminal continuity and in principle could be readily influenced by other Ca 2+ transporting organelles, such as mitochondria and lysosomes in shaping the smooth muscle Ca 2+ signal [30,31,32,33,34]. It is evident that the elevated [Ca 2+ ] SR at the origin and at the rim of the wave plays a crucial role in wave initiation and propagation, underscoring the importance of previous reports demonstrating that [Ca 2+ ] SR has a stimulatory role in IP 3 R-mediated Ca 2+ release [35,36,37].
In fact, several reports make a strong case for the existence of a luminal excitatory Ca 2+ -binding site on IP 3 R, which would provide a straightforward explanation for the stimulating effects of [Ca 2+ ] SR [27,28,38,39]. Regulation of IP 3 R by luminal Ca 2+ suggests a mechanism for Ca 2+ wave initiation based on local [Ca 2+ ] SR fluctuations caused by random opening of IP 3 Rs and separation of Ca 2+ sources (SERCA) and sinks (IP 3 Rs) in the walls of the SR lumen (see Fig. 9). It is conceivable that the ''Frequent Discharge Sites'' observed by Bolton and coworkers [40] were due to specific local ultra-structural characteristics of the SR which favour focal [Ca 2+ ] SR fluctuations. One of these is closeness to the nucleus, which has been shown in HeLa cells to delay the decay of IP 3 R-generated Ca 2+ puffs, due to decreased sequestering/ buffering [25,26].
A relevant question is why these waves were observed only in the absence of external Ca 2+ . One plausible answer is that removal of external Ca 2+ causes marked lowering of [Ca 2+ ] i and thereby delays SR Ca 2+ release stimulated by ET-1. In our VSMC preparation, we have shown that the release is mediated by opening of IP 3 Rs (see Fig. 7). It is likely that the reduced [Ca 2+ ] i due to removal of external Ca 2+ lowers the rate of IP 3 synthesis by phospholypase C, and thus increases the time required for IP 3 concentration to reach a threshold value [41,42]. The latency period would be further increased because the low [Ca 2+ ] i would also raise the threshold for IP 3 -mediated activation. The consequently slower rates of spontaneous discharge combined with the lack of SR refilling would greatly enhance the development of Ca 2+ depletion waves. On the other hand, in the presence of extracellular Ca 2+ , rapid SR refilling could obscure the depletion process. In addition, activation of clusters of IP 3 Rs at much higher frequency would prevent the development of full Ca 2+ depletion waves. Since it was essential to separate trans-plasma membrane fluxes from the Ca 2+ release process, we simply removed the extracellular Ca 2+ component, a procedure which we and numerous previous studies have shown to not have deleterious effects on the SR [16,20,43,44,45]. Thus, we conclude that the observed Ca 2+ depletion waves provide valuable new insight into a regulatory role of [Ca 2+ ] SR in the processes of Ca 2+ wave initiation and propagation.
Finally, the arrest of the SR depletion wave at the border of the nuclear envelope indicates that, as has been shown previously [17], the two membrane systems may differ with respect to their respective Ca 2+ transport mechanisms. Figure 10B strongly suggests that the nuclear envelope does contain IP 3 R, but whether these are located in the membrane facing the cytoplasm or the inner membrane facing the nucleoplasm is an open question. Nevertheless, the observation that the wave does not deplete the nuclear envelope, could be of physiological importance in terms of Endothelin-1 Induced Calcium Depletion Wave segregation of dynamic Ca 2+ signalling of vasoconstriction from functions such as gene transcription.
In conclusion, the findings presented in this study provide new insight into the mechanisms whereby the ER/SR, the main Ca 2+ regulatory organelle in living cells, determines spatial and temporal characteristics of cellular Ca 2+ signalling, which differentially control cellular migration, growth, proliferation and apoptosis in health and disease.

Supporting Information
Movie S1 3-D structure of rat SMCs transfected with D1SR.

(WMV)
Movie S3 Arrest of depletion wave at the nuclear envelope. (WMV)