Independent modes of propagation of calcium waves in smooth muscle cells

Stimulation with extracellular ATP induces a nucleotide receptor-mediated rapid initial transient rise followed by asynchronous periodic oscillations of cytosolic Ca²⁺ activity ([Ca²⁺]ⁱ) in individual porcine aortic smooth muscle cells (Linderman et al., 1990; Mahoney et al., 1992). Because [Ca²⁺]_i is the primary regulator of excitation-contraction coupling in smooth muscle cells (Somlyo and Himpens, 1989), it is clear that spatiotemporal regulation of the [Ca²⁺]ⁱ responses is critical in the regulation of the contractile state of individual smooth muscle cells within a tissue.

Since the first report that [Ca²⁺]_i signals propagate as waves in oocytes (Gilkey et al., 1978), it has become increasingly clear that calcium second messenger signaling has both a temporal and spatial component. Receptor-mediated [Ca ²⁺]_i waves have been reported in glutamate-stimulated astrocytes (Cornell-Bell et al., 1990), vasopressin-stimulated hepatocytes (Rooney et al., 1990), and vasopressin-stimulated A7r5 cultured vascular smooth muscle cells (Wier and Blatter, 1991), among others. This subject has been recently reviewed by Meyer (1991) and Jaffe (1991).

Meyer notes that two essential aspects of [Ca²⁺]_i waves in cells are diffusion of Ca²⁺ and positive feedback in the amplification cascade leading to [Ca²⁺]ⁱ elevation. Models in the literature that have been applied to [Ca²⁺]_i wave signaling include the [Ca²⁺]_i amplification of phospholipase C (PLC) model of Meyer and Stryer (Meyer, 1991; Meyer and Stryer, 1988), the calcium-induced calcium release (CICR) model by Dupont et al. (1991), and models incorporating feedback due to protein kinase C (Cuthbertson and Chay, 1991). In the first model, a ligand-bound receptor activates a GTP-binding protein (G-protein) that, in turn, activates PLC. This PLC activity hydrolyzes phos-phatidylinositol (PIP²) to form inositol 1,4,5-trisphosphate (IP₃) that releases Ca²⁺ from IP ₃-sensitive Ca²⁺ stores. The initial elevation of [Ca²⁺]_i stimulates further IP₃ production by positive feedback on PLC. The second model, CICR, proposes that Ca²⁺ is first released from an IP₃-sensitive pool and, in the continued presence of IP₃, is taken up into an IP₃-insensitive pool. Overloading the IP₃-insensitive pool or elevation of cytoplasmic [Ca²⁺] triggers release of Ca²⁺ from this pool into the cytoplasm. The third model involves oscillations of [IP₃] and [Ca²⁺]_i through both negative feed-back from protein kinase C (PKC) onto G-proteins and also receptors, and positive feedback of Ca²⁺ onto PLC.

It has been recently reported that Xenopus oocytes, expressing muscarinic acetylcholine receptors, respond to acetylcholine stimulus with propagating circular and spiral [Ca²⁺]ⁱ waves (Lechleiter et al., 1991). These waves origi-nate from separate foci in the ooplasm and have been shown to propagate by a modified CICR mechanism in which elevated [Ca²⁺]_i enhances release from IP₃-sensitive calcium stores which, subsequent to Ca²⁺ release, become refrac-tory (Delisle and Welsh, 1992; Lechleiter and Clapham, 1992).

In this report, we describe two types of [Ca²⁺]_i spatial signals in ATP-stimulated pig aortic smooth muscle cells: propagating waves and transient gradients. Individual smooth muscle cells were microinjected with the Ca²⁺ indi-cator fura-2 covalently coupled to dextran (fura-dextran). Cytosolic fura-dextran cannot be compartmentalized into organelles or extruded from the cells (Miller et al., 1992), in contrast to the acetoxymethyl ester or free anion form of the dye (Moore et al., 1990; Roe et al., 1990); this allows measurement of [Ca²⁺]_i in the cytosol with no contribution from organelle-sequestered dye and without reliance on cel-lular esterase activity. The [Ca²⁺]_i waves we describe are novel in that different regions of initiation of the waves are found in a single cell and that multiple independent oscil-lators are found in a single cell. We present evidence that different modes of oscillation are mediated by different mechanisms.

Vascular smooth muscle cells were isolated from medial tissue of swine aorta as described previously (Goldman et al., 1983). For experiments, subcultured cells from passages 3-15 were grown overnight on no.1 glass coverslips in 35 cm² culture dishes in a growth medium consisting of DMEM (Dulbecco’s modified Eagle’s medium; Gibco, Grand Island, NY) supplemented with 200 i.u./ml penicillin, 0.2 mg/ml streptomycin and 10% fetal calf serum, and thereafter in defined medium (DMEM supplemented with 5.0 μg/ml transferrin, 6.0 μg/ml insulin and 0.04 mg/ml ascorbic acid). Experiments were performed approximately 96 hours after plating onto coverslips.

For fura-dextran loading, the cells were quickly rinsed in HEPES-buffered saline (HBS is, in mM, 20 HEPES (pH 7.4), 125 NaCl, 5.4 KCl, 1.8 CaCl₂, 0.8 MgSO₄, 5.5 D-glucose), pressure injected with 20 mg/ml fura-dextran (10 kDa; Molecular Probes, Inc., Eugene, OR) in 100 mM KCl, and returned to defined medium for 30 min at 37°C in a 95% air/5% CO₂ atmosphere to allow the dye to distribute homogeneously throughout the cell. For fura-2/AM loading, the cells were incubated with 2 μM fura-2/AM (from a 1 mM stock in DMSO; Molecular Probes, Inc., Eugene, OR) for 30 min at 37°C in a 95% air/5% CO₂atmosphere. The coverslip was then washed with HBS and mounted in a flow chamber held at 37°C in HBS on a Zeiss IM-35 microscope (Carl Zeiss, Inc., Thornwood, NY) in HBS. Extracellular ATP was perfused in HBS. Fluorescence images of the cells were acquired by a charge-coupled device (CCD) instrumentation camera (Photometrics, Ltd., Tucson, AZ) as previously described (Linderman et al., 1990). Images were captured by illumination at 365 nm and 334 nm with typical exposure times of 25 and 250 ms, respectively.

Extracellular ATP in HBS was perfused into the cell chamber as previously described (Linderman et al., 1990; Cheyette and Gross, 1991). Cheyette and Gross showed that the complete exchange of fluid in the cell chamber took approximately 2 s.

Using a blank-field background image void of cells, ratio (R=(F₃₃₄−F_334(blank))/(F₃₆₅−F_365(blank))) images were generated and pCa images were computed using calibration images with minor changes in the procedure described previously (Linderman et al., 1990). Briefly, fluorescence images of 0.3 ml of fura-dex-tran (40 μg/ml) or fura-2 free acid (10 μM) in a HEPES buffer (pH 7.4) containing 10 mM HEPES, 150 mM KCl, 0.2 % NaN₃ supplemented with 5 mM EGTA or 500 μM CaCl₂, or buffer alone (for background subtraction), were collected. Using these images, R_max, R_min and β were calculated: R_max = (F₃₃₄/F₃₆₅) at saturat-ing Ca²⁺, R_min = (F₃₃₄/F₃₆₅) at no Ca²⁺, and β = F₃₆₅(no Ca²⁺)/F₃₆₅(saturating Ca²⁺). The formula [Ca²⁺]_i = K^dβ[(R− R_min)/(R_max−R)], where K^d = 394 nM is the equilibrium dissoci-ation constant for Ca²⁺ and fura-dextran was used to compute [Ca²⁺]_i. Calibration corrections for high intracellular viscosity were measured in solutions containing 60% (w/v) sucrose (Poenie, 1990), as described previously (Linderman et al., 1990). Resting [Ca²⁺]_i levels calculated using these calibration images were ∼60 to 140 nM.

Calibrations for fura-2/AM-loaded cells were done by compu-tation from ratio values as described previously (Mahoney et al., 1992). The initial resting ratio value was set to pCa = 7 and [Ca²⁺]_i calculated from the antilog of the normalized ratio value.

Illustrative results are presented in the figures for 2 individual cells. Twenty two cells, 10 injected with fura-dextran and 12 loaded with fura-2/AM were examined for the presence of Ca²⁺ waves and gradients. Data from all cells were included in the com-putation of average wave speed and gradient.

We have recently shown that pig arterial smooth muscle cells respond to extracellular ATP stimulation with an ini-tial transient rise followed by oscillations of [Ca²⁺]_i that are asynchronous from cell to cell. Individual cells display dif-ferences in the details of this response in a dose-dependent manner, but more than 90% of the cells respond in this fashion with 50 μM ATP stimulation (Mahoney et al., 1992). In the experiments reported here, single arterial smooth muscle cells were loaded with fura-dextran by microinjection. In all 10 cells examined by this method, waves and gradients of [Ca²⁺]_i were found. Representative results from two of these cells are described below.

In response to 50 μM ATP, [Ca²⁺]_i averaged over a whole cell increased from ∼140 nM to ∼10 μM within 4 s of stim-ulus (Fig. 1A). Fig. 1B shows the rise and decline of the initial transient in three regions of the cell (see inset of Fig. 1A). The [Ca²⁺]_i increase in all three regions is first detected in the image captured 4 s after stimulation. Both at 4 s and at peak, the right region (▪) attained a higher [Ca²⁺]ⁱ than the center (Δ) or left region (*#x2022;). Although [Ca²⁺]ⁱ was changing rapidly during these time points, all portions of the cell were sampled in parallel, thus, the relative differ-ences in [Ca²⁺]ⁱ in different parts of the cell can be com-pared. The fact that a measured [Ca²⁺]ⁱ gradient was pre-sent at both time points supports the notion that a real gradient was present. During the declining phase of the ini-tial transient, [Ca²⁺]_i in the right region dropped below that in regions a and b until [Ca²⁺]_i reached ∼200 nM, at which point this condition reversed. Thus, in this cell, a [Ca²⁺]_i gradient from right to left was established during the rising phase of the response, it reversed transiently during the decline, and finally reversed again as [Ca²⁺]_i approached pre-stimulus levels. In all 10 ATP-stimulated arterial smooth muscle cells observed using fura-dextran, the ini-tial [Ca²⁺]_i rise began in a peripheral region of a cell and migrated away from the initiation point across the cell in a manner similar to that displayed in the cell of Fig. 1. This result contrasts with that of Wier and Blatter (1991) in which vasopressin stimulation of A7r5 cultured vascular smooth muscle cells produced an initial [Ca²⁺]_i rise at the cell perimeter that propagated inward toward the cell center.

As [Ca²⁺]_i returned to near baseline after the initial tran-sient, oscillations began (Fig. 1A). We have previously shown that ATP-stimulated oscillations in these cells arise from intracellular stores of calcium (Mahoney et al., 1992) although extracellular Ca²⁺ is required to replenish the stores. ATP-stimulated [Ca²⁺]_i oscillations showed sub-stantial variability in the spatial disposition of their initia-tion and decay (Fig. 1C). The first six oscillations initiated in the right region of the cell while the 8th oscillation ini-tiated from the left region of the cell. The 7th oscillation simultaneously initiated from both loci. In this intermedi-ate, 7th oscillation, the average peak [Ca²⁺]_i was only slightly higher than that of the previous (6th) oscillation (Fig. 1A), indicating that [Ca²⁺]_i elevation resulting from the combination of the two independently initiated waves was not additive. The first five oscillations decayed homo-geneously across the cell. Beginning with the 6th oscilla-tion, however, the decay of each oscillation developed an increasingly pronounced spatial gradient, with [Ca²⁺]_i in the right region of the cell returning to baseline after that in the left side. The 6th oscillation rises from and returns to the same region of the cell. We refer to this as a pulsing gradient. The 8th oscillation clearly represents a propagat-ing wave traveling from one side of the cell to the other. Thus, the mode of [Ca²⁺]_i propagation during this series of oscillations changed from a pulsing gradient to a propa-gating wave concomitant with the shift in the locus of ini-tiation of the [Ca²⁺]_i transient.

A second level of complexity in [Ca²⁺]_i signaling in smooth muscle cells was demonstrated in the cell shown in Fig. 2. When [Ca²⁺]_i was averaged over the whole cell, only 4 clear oscillations were observed (Fig. 2A). However, the bottom region of the cell (region d) displayed six separate oscilla-tions, while other regions of the cell displayed only four (Fig. 2B). The four oscillations in regions a, b and c prop-agated into region d as shown by the time delay of the arrival of peak [Ca²⁺]_i (Fig. 2B). Two distinct sets of oscil-lations were seen in region d as is shown in Fig. 2C. In this figure, the same lines of pixels along the axis of the cell were sampled for each of the 200 images in the 600 s time-lapse sequence, and the lines from each image were arranged horizontally in sequence to produce a one spatial dimension (vertical direction) versus time (horizontal direc-tion) image. The bottom of Fig. 2C (region d in Fig. 2A,B) shows two independent sets of oscillations, one of which propagated into region d from region c and one of which was localized to region d only.

To quantitate the magnitude and rate of the [Ca²⁺]_i wave and gradient propagations during these oscillations, [Ca²⁺]_i was measured in the three regions of the cell shown in the inset of Fig. 1A. Plots of the [Ca²⁺]_i oscillations versus time in these three subregions for the 5th through the 8th oscil-lations demonstrate that the initiation site of the oscillations changes over time and that the mode of propagation (wave versus gradient) of the [Ca²⁺]_i signal changes from oscilla-tion to oscillation (Fig. 1D).

The average speeds of the [Ca²⁺]_i wave fronts during the rising phase of the 1st through the 6th oscillations were, respectively, 4.2, 2.8, 4.7, 5.9, 8.1 and 8.6 μm/s. Wave speeds were determined from plots of [Ca²⁺]_i versus posi-tion along the direction of wave propagation for time points during the rising phase of each oscillation as shown in Fig. 3 for the 3rd and 5th oscillations. The location of the wave front for which [Ca²⁺]_i = 200 nM was chosen as the posi-tion of the wave at each time point. For the 5th oscillation (Fig. 3B), wave speed was nearly constant as the wave tra-versed the cell; however, in the 3rd oscillation (Fig. 3A), wave speed increased as the wave traversed the cell. Oscil-lations began with a shallow gradient that increased in steepness with time. Once a [Ca²⁺]_i gradient became estab-lished at the leading edge of a wave, its steepness (change of [Ca²⁺]_i over distance as in Fig. 3A,B) remained relatively constant across the cell, indicating that the [Ca²⁺]_i flux at the leading edge of a wave is approximately constant. The steepness of the [Ca²⁺]_i wave front, and thus the Ca²⁺ flux, varied from oscillation to oscillation (Fig. 3A,B). An anom-alous [Ca²⁺]_i elevation at 50 μm from the left side of the cell was seen during each oscillation as the wave front ⁱ₂₊passed this point, suggesting that this may be a region of the cell in which calcium release from internal stores is greatest.

Cells microinjected with fura-dextran or loaded with fura-2 by incubation with its acetoxymethyl ester (fura-2/AM) were carefully examined for such spatial [Ca²⁺]_i inhomogeneities. The average wave speed and gradient steepness of each oscillation in 10 fura-dextran- and 12 fura-2/AM-loaded cells were determined. All 22 cells show [Ca²⁺] gradients. [Ca²⁺] spread at an average speed of 3.9 ± 1.2 μm/s for fura-dextran-loaded cells while the speed is significantly faster, 6.5 ± 2.4 μm/s, with fura-2/AM-loaded cells (Student’s t-test, P < 0.005). Although the [Ca ²⁺]_i gra-dient steepness was slightly less in fura-2/AM-loaded cells (3.8 ± 2.4 nM/μm) versus fura-dextran-loaded cells (4.7 ± 1.8 nM/μm), the difference was not statistically significant (Student’s t-test, P > 0.1).

In fura-dextran-loaded cells, all injections were aimed towards the center of the cell where the height of the cell is greatest, while [Ca²⁺]_i gradients typically started at the periphery. Also, there was no correlation between the injec-tion sites and local regions of high basal [Ca²⁺]_i. [Ca²⁺]_i oscillations and the direction of the propagating waves in any particular cell did not correlate with [Ca²⁺]_i signals in neighboring cells. In all ATP-stimulated cells, whether loaded with fura-dextran or fura-2/AM, multiple sites of initiation and modes of propagation of [Ca²⁺]_i gradients were found similar to those shown here.

We have previously shown that pig arterial smooth muscle cells respond to nucleotide receptor stimulus by ATP with an initial transient rise in [Ca²⁺]_i that decays over time and that is followed by persistent oscillations of [Ca²⁺]_i (Mahoney et al., 1992). Both the initial transient and oscil-lations are produced by ligand-activated Ca²⁺ release from intracellular stores, although the initial transient is pro-longed in Ca²⁺-containing medium as compared to Ca²⁺-free medium and the oscillations are not established in the absence of external Ca²⁺. It appears that intracellular Ca²⁺ store refilling is necessary for oscillations to persist. In this report we show that there are two types of nucleotide recep-tor-mediated [Ca²⁺]_i spatial signals, one a gradient, similar to the [Ca²⁺]_i gradients observed in hepatocytes (Rooney et al., 1990), and the other a propagating wave, similar to those observed in cardiac cells (Takamatsu and Wier, 1990). We further show that independent sites of initiation of the [Ca²⁺]_i waves and gradients exist in an individual cell, and that these sites can act independently of each other. Within the time resolution of observation of change in [Ca²⁺]_i, the nucleotide stimulant was applied uniformly across the surface of the plasma membrane (Materials and Methods). Therefore, spatial control of [Ca²⁺]_i transients must arise from the intrinsic properties of the cell itself. Receptors, receptor-activated enzymes, intracellular Ca²⁺ stores or other components of the Ca²⁺-second messenger signaling cascade could be concentrated within subregions of cells, thus allowing spatial regulation of the [Ca²⁺]_i response.

The presence of two independent sets of [Ca²⁺]_i oscilla-tions in one region (d) of the cell shown in Fig. 2 provides insight into the nature of the mechanism of [Ca²⁺]_i propa-gation in arterial smooth muscle cells. Fig. 2C shows that the [Ca²⁺]_i oscillation localized to region d of the cell (referred to below as the local oscillator) does not propa-gate more than 30 μm away from that region while an inde-pendent wave (referred to below as the global oscillator) initiating from region b propagates across the whole cell into region d. Further, in region d, the inability of the rise and subsequent fall of [Ca²⁺]_i due to the local oscillation to block the incoming propagation of the oscillation origi-nating in region b (Fig. 2B) shows that calcium stores from which the propagating oscillations arise are either indepen-dent of those of the local oscillation or they are neither depleted nor desensitized by the local oscillation.

The data indicate that at most one of the oscillators (global or local) could be due to the proposed calcium-induced calcium release model of [Ca²⁺]_i waves (Dupont et al., 1991), since in this model elevated Ca²⁺ is a positive feedback element. If the global oscillator is controlled by CICR, then the local oscillator cannot be because it would also produce a wave that propagates across the cell. If the local oscillator is controlled by CICR, obviously only a localized pool of Ca²⁺ is sensitive to this mechanism. An analogous argument can be made for the model in which Ca²⁺-positive feedback of IP₃-mediated Ca²⁺ release pro-duces self-regenerative [Ca²⁺]_i waves in Xenopus oocytes (Girard et al., 1992), and for the model of [Ca²⁺]_i oscillations in which IP₃ levels oscillate and [Ca²⁺]_i exerts positive feedback on its own release and on IP₃ production (Harootunian et al., 1991). By analogous argument we mean that at most one of the oscillators can be explained by the model in question and if it is the local oscillator the rele-vant pool is spatially limited. No two of the above three models could be present simultaneously, because all con-tain elevated Ca²⁺ as a positive feedback element. Even though one model also contains an independently necessary element, in any combination of two of them, propagation of both would occur simultaneously.

Similarly, the observation of both pulsing gradients and propagating waves in the same cell (Fig. 1) implies that there are at least two mechanisms of control of [Ca²⁺]_i in these cells. The observation of a pulsing gradient implies at least either a localized pool of Ca²⁺ accessible to that mechanism for release, or non-uniform spatial distribution of reuptake. On the other hand, to sustain a propagating wave there must be a globally accessible release pool cou-pled with uniform distribution of Ca²⁺ removal sites.

A hypothesis that can explain this phenomenon is that the two spatially separate loci of origin of the [Ca²⁺]_i oscil-lations are responding to two different Ca²⁺ releasing agents and that one of these releasing agents produces a propa-gating [Ca²⁺]_i wave. IP₃ and cyclic ADP-ribose are two putative independent Ca²⁺-releasing agents as they have been shown to act independently to release Ca²⁺ (Dargie et al., 1990). If two different Ca²⁺-releasing agents are acting in the cell of Fig. 2, then the one responsible for the local oscillator must be incapable of Ca²⁺ release beyond the locale of that oscillator. This could be accomplished by localized production of the releasing agent, perhaps cou-pled with rapid degradation of the releasing agent, or by localization of the Ca²⁺ stores sensitive to that agent.

Our data can be used to estimate the magnitude of the Ca²⁺ flux occurring during [Ca²⁺]_i waves. The mean spatial gradient of [Ca²⁺]_i in propagating waves measured as described in the legend of Fig. 3, 4.7 nM/μm, can be used to calculate the mean calcium ion flux (obtained from Fick’s law: J_Ca =−D_Ca(∂[Ca²⁺]_i/∂x) with D_Ca = 4 × 10⁻⁶ cm²/s (Kar-gacin and Fay, 1991), where we consider only the Ca²⁺ flux in the plane of the specimen). The flux calculated is 1.8 × 10⁻²¹ mol/μm ² s = 1100 Ca ions/μm ² s in the direction of movement of the wave. This flux is a lower-limit estimate of the true calcium ion flux, since the presence of mobile cytosolic buffers will mask some of the moving Ca²⁺.

The fact that the leading edges of the [Ca²⁺]_i waves in these cells move with an average speed of 4 μm/s in fura-dextran-loaded cells raises the question of blurring of the edge of the wave during the time of image exposure. All of the data shown here were collected with exposure time of 25 ms at 334 nm excitation and 250 ms at 365 nm exci-tation. The start of the second exposure was delayed 300 ms after the end of the first exposure, so that one complete image set was collected in 575 ms. On average, a [Ca²⁺]_i wave front would have moved 2.3 μm in that time. This value represents the limiting distance over which the wave front is resolved. The data shown here were collected by combining 4 pixels in both the x and y directions on the CCD (Linderman et al., 1990) for a spatial resolution of 2.1 μm. Thus for our data, blurring during the wave move-ment is not significant.

The attenuation of measured spatial [Ca²⁺]_i inhomo-geneity in fura-2/AM-loaded cells compared to fura-dex-tran-loaded cells may result from the incorporation of fura-2 into organelles or the relatively higher mobility of fura-2 in the cytosol. The former would tend to mask cytosolic [Ca²⁺] changes while the latter would tend to dissipate [Ca²⁺]_i gradients by shuttle-buffer effects (Speksnijder et al., 1989). As discussed by Sala and Hernandez-Cruz (1990), the effect of increased mobility of a Ca²⁺ buffer is to increase the rate at which [Ca²⁺]_i gradients dissipate, i.e. to increase the effective diffusion coefficient for Ca²⁺. Allbritton et al. (1992) have shown that the diffusion of Ca²⁺ in Xenopus oocyte cytoplasmic extract is attenuated by immobile Ca²⁺ buffers. As noted by Meyer (1991), the velocity of [Ca²⁺]_i wave propagation increases as D_Ca^G; thus our finding that [Ca²⁺]_i wave speeds are greater in the pres-ence of fura-2 free acid than fura-dextran is consistent with the idea that the free acid is facilitating diffusion of Ca²⁺ in the cell.

Arterial smooth muscle cells display two unique features in the propagation of agonist-stimulated [Ca²⁺]_i oscillations, compared to the traveling waves and spatial gradients of [Ca²⁺]_i in agonist-stimulated individual cells that have been previously reported (Cornell-Bell et al., 1990; Rooney et al., 1990; Lechleiter et al., 1991; Wier and Blatter, 1991). First, these cells produce oscillations that initiate from dif-ferent loci within the cell. Second, separate regions within a single cell can sustain independent [Ca²⁺]_i oscillations that arise from separate loci within the cell. These localized [Ca²⁺]_i transients may be employed by the cell to spatially and temporally regulate [Ca²⁺]_i-dependent enzymes to effect localized control over enzyme activity.

We thank Dale A. Callaham for introducing us to fura-dextran and microinjection. We thank Clare Fewtrell, Edward W. West-head and Lionel F. Jaffe for critically reading the original manu-script and particularly the Journal of Cell Science reviewer for very helpful comments. This work was supported by the National Institutes of Health (HL 31854), and the National Science Foun-dation (DCB-9105429 and DCB-9004191).