Contribution of sustained Ca2+ elevation for nitric oxide production in endothelial cells and subsequent modulation of Ca2+ transient in vascular smooth muscle cells in coculture.

To elucidate the intracellular Ca (Ca) transient responsible for nitric oxide (NO) production in endothelial cells (ECs) and the subsequent Ca reduction in vascular smooth muscle cells (VSMCs), we administrated four agonists with different Ca-mobilizing mechanisms for both cells in iso- or coculture. We monitored the Ca of both cells by two-dimensional fura-2 imaging, simultaneously measuring NO production as NO. The order of potency of the agonists in terms of the peak Ca in ECs was bradykinin (100 nM) > ATP (10 μM) > ionomycin (50 nM) > thapsigargin (1 μM). In contrast, the order in reference to both the extent of Ca reduction in cocultured VSMCs and the elevation in NO production over the level of basal release in ECs completely matched and was ranked as thapsigargin > ionomycin > ATP > bradykinin. Treatment by N-monomethyl-L-arginine monoacetate but not indomethacin or glybenclamide restored the Ca response in cocultured VSMCs to the isoculture level. In ECs, when the Ca influx was blocked by Ni or by chelating extracellular Ca, all four agonists markedly decreased NO production, the half decay time of the Ca degenerating phase, and the area under the Ca curve. The amount of produced NO hyperbolically correlated to the half decay time and the area under the Ca curve but not to the Ca peak level. Thus, the sustained elevation of Ca in ECs, mainly a result of Ca influx, determines the active NO production and subsequent Ca reduction in adjacent VSMCs. Furthermore, L-arginine but not D-arginine or L-lysine at high dose (5 mM) without agonist enhanced the NO production, weakly reduced the Ca in ECs, and markedly decreased the Ca in VSMCs, demonstrating the autocrine and paracrine effects of NO (Shin, W. S., Sasaki, T., Kato, M., Hara, K., Seko, A., Yang, W. D., Shimamoto, N., Sugimoto, T., and Toyo-oka, T.(1992) J. Biol. Chem. 267, 20377-20382).

Endothelial cells (ECs) 1 modulate the contractility of underlying vascular smooth muscle cells (VSMCs), secreting several vasoconstrictors and vasorelaxants (1,2). The principle endothelium-derived relaxing factors have been identified as nitric oxide (NO) (3) and prostacyclin. Both are regulated by intracellular calcium ions (Ca 2ϩ i ) (4). In the absence of extracellular Ca 2ϩ (Ca 2ϩ o ), NO production is greatly reduced (5,6). Recently, we have presented evidence that NO affects the handling of Ca 2ϩ i by an autocrine action in NO-producing ECs and by a paracrine action in adjacent VSMCs (7). However, little quantitative information is available on the relationship between the Ca 2ϩ i in ECs and NO production. There is also a lack of information regarding the amount of NO produced in ECs and its action to reduce the Ca 2ϩ i levels in neighboring VSMCs. The Ca 2ϩ i in VSMCs is crucial, because it is a primary factor in the regulation of muscle contractility (8). Most Ca 2ϩ i transients in ECs induced by agents that cause the release of NO consist of a peak followed by a degeneration phase (5). The peak originates from the release of Ca 2ϩ from endoplasmic reticulum. The influx of Ca 2ϩ from the extracellular medium accounts for the maintenance of the subsequent portion of degeneration phase. To determine which component of the Ca 2ϩ i transients of ECs is most significant in indicating the production of NO and to monitor the biological action of NO on VSMCs, we simultaneously measured the stable NO metabolite, nitrite (NO 2 Ϫ ), in the medium and the Ca 2ϩ i of both ECs and VSMCs in coculture by two-dimensional image analysis. We report here a unique communication between the Ca 2ϩ i in ECs and the Ca 2ϩ i in VSMCs mainly mediated by NO.

MATERIALS AND METHODS
Reagents-All reagents used were of analytical grade. Phosphatebuffered medium was utilized as the extracellular medium and was composed as follows: 137 mM NaCl, 2.7 mM KCl, 8.1 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 1.0 mM CaCl 2 , and 0.5 mM MgCl 2 , pH 7.4. Ca 2ϩ -free medium consisted of phosphate-buffered medium in which CaCl 2 was replaced by EGTA (1.0 mM). When Ni 2ϩ was employed in our protocol, the medium was switched to HEPES (Dojin, Tokyo) buffer (145 mM NaCl, 5.0 mM KCl, 1.0 mM CaCl 2 , 0.5 mM MgCl 2 , 10 mM D-glucose, and 10 mM HEPES, pH 7.4) to prevent a precipitation in phosphate-buffered medium caused by the addition of Ni 2ϩ . Thapsigargin (TG, Sigma) and ionomycin (IM, Sigma) were dissolved in Me 2 SO and then diluted with buffer. The final concentration of Me 2 SO was less than 0.1% (v/v) and therefore had no direct action on the handling of Ca 2ϩ i in either ECs or VSMCs. N G -monomethyl-L-arginine monoacetate (L-NMMA) was purchased from Calbiochem (La Jolla, CA). Bradykinin (BK), ATP, glybenclamide, and indomethacin were from Sigma. L-Arginine monohydrochloride, D-arginine monohydrochloride, and L-lysine monohydrochloride were from Wako (Tokyo, Japan). Each agent was administered by replacing half of the volume of extracellular medium with a twice concentrated solution of the reagents in incubation medium. Isoculture or Coculture of ECs and VSMCs-ECs were enzymatically isolated from bovine aorta and then cultured and identified as described previously (7). They were seeded in two kinds of dishes, one made of a fluorescence-free glass for assaying both Ca 2ϩ i and NO 2 Ϫ (1 ml in volume) and another made of plastic for the periodic sampling and measurement of NO 2 Ϫ (10 ml in volume). A cell line of VSMCs, A7r5, from American Type Culture Collection (Rockville, MD) was cultured using the same method as ECs.
To assess the biological effect of NO on the Ca 2ϩ i in VSMCs, we cultured ECs and VSMCs together (7). Three days after the seeding of VSMCs, at a density of 1.5 ϫ 10 3 /cm 2 , the culture dishes were semiconfluent. ECs were then added to the culture at the same initial density of VSMCs, 1.5 ϫ 10 3 /cm 2 . Under these conditions, both VSMCs and ECs grew symbiotically.
Measurement of Ca 2ϩ i by Two-dimensional Image Analysis-Ca 2ϩ i within individual cells was analyzed as described previously (9,10). Significant leaking of fura-2 did not occur during the measurement, evidenced by the increment of autofluorescence in the incubation medium. All procedures including the addition of drugs to the medium were performed under dark conditions, and illumination time was minimized to prevent the photobleaching of fura-2. Because the fura-2 method has several intrinsic problems in the estimation of absolute Ca 2ϩ i (11), the amplitude of Ca 2ϩ i elevation in response to each stimulant was calculated using the percentage of increase of F340/F380 with reference to F340/F380 at the resting state. This relative increase of F340/F380 after stimulation was represented by the percentage of increase in pseudocolor image. The mean percentage of increase of F340/F380 at all pixels within each cell was defined as the percentage of increase of F340/F380 of the cell.
In a pilot study for dose determination, we searched for suitable agonists and their appropriate concentrations that make it possible to analyze Ca 2ϩ i transients using different receptor/effector coupling systems. Iso-or cocultured ECs or VSMCs were stimulated by ATP (10 M) or BK (100 nM) for 3 min. In the cases of TG (1 M) and IM (50 nM), the stimulation time was extended to 40 min because the Ca 2ϩ i response to these agonists was much more gradual than ATP or BK. To quantify the Ca 2ϩ i responses in ECs, we measured three parameters: the peak level, the half decay time (HDT), and the area under the Ca 2ϩ i curve (AUC) by computer-assisted planimetry as shown in Fig. 1. To examine the reproducibility of Ca 2ϩ i responses to the same agonist, a second stimulation was performed after a 30-min equilibration period following the washout of the first application of the agonist.
NO 2 Ϫ Determination and NO Bioassay of ECs and VSMCs in Coculture-For ATP and BK studies, 0.5-1.0 ml of aliquot was sampled from plastic dishes before and 15, 30, 45, 60, 120, and 180 s after the addition of the agonist. In experiments using TG or IM, the sampling time was changed to 15, 30, 45, and 60 s and 3, 5, 10, 20, 30, and 40 min, respectively, after the administration, because of the slow Ca 2ϩ i response to these agents.
For the simultaneous measurement of both Ca 2ϩ i and NO content, 0.8 ml of medium was sampled from the glass dishes at 3 min in ATP and BK studies and at 40 min in TG and IM studies after completion of the Ca 2ϩ i measurement. Released NO was completely oxidized to NO 2 Ϫ when left standing overnight 4°C in a sealed Eppendorf tube. In a preliminary study, the amount of sealed NO 2 Ϫ was shown to be stable under this condition until assay. The amount of NO 2 Ϫ was determined by the modified method of Yui et al. (12). In short, NO 2 Ϫ was measured by colorimetry after Griess reaction. To enhance the sensitivity, light absorbancy was detected by spectrometer of HPLC at the maximum sensitivity, and full scale was set to 0.01 absorbance. Lumen of all tubings was coated by Teflon to prevent acid corrosion.
To further verify the results of NO 2 Ϫ measurement, we performed a bioassay of NO by measuring the Ca 2ϩ i reduction to below basal levels or the blunted Ca 2ϩ i rise in VSMCs caused by the native NO from cocultured ECs (7). When indicated, the substrate for NO synthase (NOS), L-arginine (5 mM), the NOS inhibitor, L-NMMA (500 M), the cyclooxygenase inhibitor, indomethacin (50 M), or the ATP-sensitive K ϩ channel blocker, glybenclamide (10 M), was added to ascertain what factor modifies the Ca 2ϩ i handling in iso-or cocultures. Data Analysis-All data were expressed by the mean Ϯ S.E. The difference between responses was analyzed either by paired t test or by analysis of variance. All differences were considered statistically significant if p Ͻ 0.05.

NO Production in ECs in Isoculture
Time Course of NO Production-The time course of NO production was determined by analyzing aliquots obtained from ECs cultures in plastic dishes containing 10 ml of medium. Under control conditions with no addition of agonist, NO assayed as NO 2 Ϫ was found to be present 10.5 Ϯ 0.6, 11.1 Ϯ 1.1, and 11.4 Ϯ 1.1 ng/ml/10 6 cells just before or 3 or 40 min after starting the incubation, respectively. These results were likely caused by the shear stress during the exchange of the medium rather than a gradual release that occurred throughout the incubation, because NO 2 Ϫ had not accumulated significantly after 40 min (Fig. 2, Ctl). After the administration of ATP, the level of NO 2 Ϫ gradually elevated from 10.8 Ϯ 0.4 to 13.5 Ϯ 0.8 ng/ml/10 6 cells (p Ͻ 0.01) at 3 min; adding BK to the medium resulted in a rise from 10.7 Ϯ 0.6 to 12.4 Ϯ 0.8 ng/ml/10 6 cells at 3 min (p Ͻ 0.05, Fig. 2A). In the experiments involving ATP and BK, measurements were taken at the end of a 3-min incubation after which the production of NO 2 Ϫ became stable. After an incubation for 40 min, the level of NO 2 Ϫ steadily increased from 10.5 Ϯ 0.5 to 23.8 Ϯ 1.5 (p Ͻ 0.01) in response to TG and from 10.9 Ϯ 0.5 to 14.0 Ϯ 0.4 ng/ml/10 6 cells (p Ͻ 0.01) in response to IM (Fig. 2B).
Dependence of NO Production on Ca 2ϩ Influx and Effect of Substrate-Another experiment showed that NO production is dependent on the entry of Ca 2ϩ into ECs. Agonist-induced NO production was enhanced relative to basal release by all four agonists in the presence of 1 mM Ca 2ϩ o . However, when Ca 2ϩ o was chelated by 1 mM EGTA without adding Ca 2ϩ to the phosphate-buffered medium, ATP, BK, and TG caused production of lower amounts of NO 2 Ϫ , close to or less than basal release. Furthermore, in media containing 1 mM Ca 2ϩ o and the nonspecific divalent cation channel blocker, Ni 2ϩ (1 mM), the agonistinduced NO production was also reduced to basal or near basal level release (Fig. 3A).
To verify the agonist-induced NO production, we used the NO synthase inhibitor, L-NMMA. Pretreatment by L-NMMA (500 M, 30 min) inhibited NO production over the basal level induced by all four agonists. However, the addition of L-arginine (5 mM) negated this inhibition and increased the production of NO after administration of all four agonists, ranging from 2-fold in the case of BK to 8-fold in the case of TG (Fig.  3B). These results indicate that NO was synthesized from L-arginine.

Ca 2ϩ i Transients
ECs in Iso-and Coculture-In the presence of Ca 2ϩ o , all four agonists raised the Ca 2ϩ i in isocultures of ECs, demonstrating different peak levels and degeneration phases (Fig. 4). Among the agonists, the Ca 2ϩ i showed rapid response to both ATP and BK, peaking at 20 s. Compared with ATP (10 M), BK (100 nM) prominently raised the Ca 2ϩ i , which then decreased at a faster rate (Fig. 4A). In contrast, the Ca 2ϩ i change was more gradual when caused by either IM or TG. Both of these agonists raised the Ca 2ϩ i within 1 min, reaching a peak of 70 Ϯ 4% at 2 min and 58 Ϯ 3% at 4 min for IM and TG, respectively. The Ca 2ϩ i level then gradually declined (Fig. 4B). As summarized in Table I, the rank order of the peak Ca 2ϩ i levels was BKϾATPϾIMϾTG. However, the HDT of Ca 2ϩ i and the AUC was ranked in the order TG Ͼ Ͼ IM ϭ ATP Ͼ BK and TG Ͼ IM Ͼ Ͼ BK Ͼ ATP, respectively. Chelation of Ca 2ϩ o by EGTA mildly decreased the peak Ca 2ϩ i caused by ATP, BK and TG and lowered the HDT and AUC to 1 ⁄3 -1 ⁄8 of the control (Table I). Ni 2ϩ pretreatment (1 mM) in the presence of 1 mM Ca 2ϩ o mildly reduced the resting Ca 2ϩ i and decreased the HDT and AUC to the same extent as chelation of Ca 2ϩ o (Table I). In cocultured ECs, the Ca 2ϩ i exhibited essentially the same transients as those in isocultured ECs (see Figs. 6 -10). All three parameters, including the peak Ca 2ϩ i level, HDT, and AUC, in coculture were comparable with those in isoculture (Table I). These results suggest that the Ca 2ϩ i handling system in ECs is not under the influence of cocultured VSMCs.
VSMCs in Iso-and Coculture-When VSMCs were cultured alone, ATP (10 M) gradually increased the Ca 2ϩ i within the first minute after stimulation and kept it above the basal level at the 3-min time point (58 Ϯ 2%; Fig. 5A and Table II). BK (100 nM) raised the Ca 2ϩ i only very weakly (8 Ϯ 3% at 5 s and 2 Ϯ 1% at 3 min; Fig. 5A), although the response was much stronger when a high dose (10 -100 M, data not shown) was used. TG (1 M) induced a unique Ca 2ϩ i response that showed a shoulder at 1 min (60 Ϯ 7%), peaked at 10 min (107 Ϯ 5%), and then declined gradually up to 40 min (16 Ϯ 2%; Fig. 5B) after stimulation. The Ca 2ϩ i rise induced by IM (50 nM) became evident at 1 min, peaked at 4 min, and then slowly decreased (43 Ϯ 2% at 40 min; Fig. 5B).
When VSMCs were cocultured with ECs, these Ca 2ϩ i changes in VSMCs were strikingly modified (Figs. 6 -10 and Table II). The Ca 2ϩ i in VSMCs after ATP stimulation did not increase but decreased to below basal level (Ϫ21 Ϯ 3% at 3 min; Fig. 6). In addition, the ATP-induced Ca 2ϩ i transient in cocul-tured VSMCs was completely dependent on the Ca 2ϩ o . The Ca 2ϩ i in VSMCs exhibited a significant reduction below the basal level in the presence of Ca 2ϩ o , whereas in the absence of Ca 2ϩ o , the VSMCs showed a Ca 2ϩ i rise to a lower peak than those in isoculture, followed by a gradual decline (Fig. 6F). BK also reduced the Ca 2ϩ i in VSMCs but to a lesser extent (Ϫ5 Ϯ 1%) than ATP ( Fig. 7 and Table II). TG caused a large Ca 2ϩ i reduction in VSMCs (Ϫ24 Ϯ 3% at 10 min; Fig. 8). This reduction was sustained up to 40 min (Ϫ25 Ϯ 3%) after the treatment. Unlike the other three agonists, IM did not decrease the Ca 2ϩ i below the basal level in VSMCs but significantly attenuated the rise of Ca 2ϩ i (Fig. 10) compared with the Ca 2ϩ i rise in isoculture. The Ca 2ϩ i increase started at 1 min (9 Ϯ 1% over the basal level), peaked at 5 min (21 Ϯ 1%), and remained above the baseline at 40 min (10 Ϯ 1%). As a control study by phosphate-buffered saline, in isocultured VSMCs as well as cocultured VSMCs and ECs, Ca 2ϩ i showed no significant changes during 40 min of follow-up (data not shown).

Identification of the Ca 2ϩ i Modulating Factor as NO in Coculture
The following four findings strongly support our scheme that the Ca 2ϩ i reduction in cocultured VSMCs is principally regulated by NO derived from ECs: (i) A pretreatment by the NOS inhibitor, L-NMMA (500 M), for 30 min attenuated the Ca 2ϩ i decrease in VSMCs caused by ATP and BK (Table II). The same treatment abolished the Ca 2ϩ i reduction in VSMCs after application of TG and IM, restoring the response to control levels ( Fig. 9 and Table II); (ii) A pretreatment by the cyclooxygenase inhibitor, indomethacin (50 M), for 15 min had no effect on the EC-dependent Ca 2ϩ i decrease in VSMCs caused by the four agonists (Table II). These results suggest that prostaglandins were not involved in the reduction of Ca 2ϩ i in VSMCs cocultured with ECs. (iii) In the case of ATP, the 30-min pretreatment by a specific inhibitor to ATP-sensitive K ϩ channels, glybenclamide (10 M), showed no effect on the ATP-induced Ca 2ϩ i reduction in cocultured VSMCs (data not shown). (iv) Surprisingly, a high concentration of L-arginine (5 mM) without agonist induced a moderate Ca 2ϩ i decrease to below basal level in ECs and a profound reduction of Ca 2ϩ i in VSMCs (Fig. 11, C and D) accompanied by a large amount of NO production (Fig.  11E). Interestingly, the L-arginine-induced NO production was not dependent on Ca 2ϩ o , because a similar NO level was produced in both the presence and the absence of Ca 2ϩ o in isocultured ECs (Fig. 11E). These results are evidence of an autocrine effect of NO on ECs that reduces their Ca 2ϩ i and a paracrine effect on VSMCs in coculture (7), whereas D-arginine (5 mM) and L-lysine (5 mM) did not induce any Ca 2ϩ i reduction in cocultured VSMCs (data not shown).   Ϯ 4 a,b 90 Ϯ 18 a,b 6978 Ϯ 2666 a,b 50 Ϯ 4 a,b 136 Ϯ 12 a,b 9278 Ϯ 2876 a,

Correlations between NO Production and Ca 2ϩ i Transients
The ATP-or BK-induced NO 2 Ϫ rise stabilized 3 min after stimulation ( Fig. 2A), and the Ca 2ϩ i response began to decline at the end of 3 min (Fig. 4A). TG-or IM-induced NO production continued to increase up to 40 min after stimulation (Fig. 2B), and simultaneous recording of the Ca 2ϩ i showed a decrease up to the 40-min time point (Fig. 4B). Accordingly, we have chosen an incubation time of 3 min for ATP-or BK-induced NO 2 Ϫ production and of 40 min for TG-or IM-induced NO 2 Ϫ production to correlate with the Ca 2ϩ i pattern. As summarized in Fig. 12A, the NO production increased sharply as the HDT rose to 200 s. After that time, further increments of NO 2 Ϫ production became reduced. The AUC was plotted in Fig. 12B. The amount of NO 2 Ϫ sharply increased until the AUC reached 20,000%⅐s and then gradually increased. Accordingly, the amount of NO 2 Ϫ hyperbolically correlated to both the HDT of Ca 2ϩ i (r ϭ 0.90, p Ͻ 0.001) and the AUC (r ϭ 0.86, p Ͻ 0.001) but did not correlate to the peak Ca 2ϩ i in ECs (r ϭ 0.17, p Ͼ 0.05; Fig. 12C). These results suggest that the NO production rate was saturated by a long sustained Ca 2ϩ i augmentation.

DISCUSSION
This is the first report to describe that (i) NO is released from ECs at the basal level even without the presence of agonist, (ii) the maintenance of Ca 2ϩ i level, which is supported by Ca 2ϩ influx, determines the rate of NO production over the basal release in ECs, (iii) NO production is not related to the peak Ca 2ϩ i but dependent on the sustained elevation of Ca 2ϩ i , as represented by the HDT or AUC of Ca 2ϩ i in ECs, (iv) the production of NO reduces the Ca 2ϩ i in cocultured VSMCs in a paracrine manner, dependent on the level of NO production, and (v) L-arginine alone enhances NO production, which has both autocrine and paracrine actions on ECs and VSMCs in coculture, respectively.
Methodologies of NO Assay-The best method to quantify the NO production in ECs would be a direct and real time measurement of NO, using an electrode specifically sensitive to NO. We tested several electrodes that have been reported to be effective (13-15) but failed to exactly measure the basal NO release and the additional production of NO after agonist application, because of the insufficient sensitivity. Very recently, Malinski's group has used BK for the simultaneous measurement of both Ca 2ϩ i and NO by electrode (16). The concentration they used was 100 times higher than that of the present study. That concentration was not suitable for studying the influence of NO on the reduction of the Ca 2ϩ i in cocultured VSMCs because the direct action of BK that increases the Ca 2ϩ i exceeded the indirect action mediated by NO (data not shown). Furthermore, the data of Blatter et al. (16) were not confirmed by us, because the transient Ca 2ϩ i was actually dependent on the Ca 2ϩ o or sensitive to a nonspecific cation channel blocker, Ni 2ϩ (Table I).
Previously, two indirect methods have been employed to quantify the NO effect on vessels. One involves the measurement of the isometric tension developed in vessel strips in the presence or the absence of ECs and the subsequent comparison of the extent of the vessel relaxation. However, the NO effect cannot be evaluated correctly in this measurement, because the vessel does not contract in the absence of Ca 2ϩ o . Another method relies on the measurement of intracellular cGMP (17,18), which is synthesized after the activation of guanylate cyclase. cGMP does not, however, explain all actions of NO (19). Both methods cannot simultaneously measure the Ca 2ϩ i dy- a Denotes statistically significant from Coculture value at p Ͻ 0.05. b Denotes statistically significant from ATP value at p Ͻ 0.05. namics in ECs and/or VSMCs. No data have been reported concerning the extent of Ca 2ϩ i elevation in ECs relative to NO production rate or the Ca 2ϩ i reduction in VSMCs. Present NO 2 Ϫ measurement was used to assay NO production mediated by inducible NOS (12), where the amount of released NO was much larger than that from eNOS and accordingly can be easily assayed. The amount of NO 2 Ϫ in the culture medium of ECs is not measurable by a simple spectrophotometry, because of its insufficient sensitivity. NO 2 Ϫ determination by HPLC, with enhancing the sensitivity after the oxidation of NO to NO 2 Ϫ (20), has made it possible to document the trace quantities of NO released from ECs mediated by eNOS for the first time. In this study, ATP and BK induced no fold increase in NO production. However, fold increase in Ca 2ϩ i or cGMP induced by ATP, BK, or A23187 as identified by Schmidt et al. (21) did not mean that the rise of NO production must be also in fold. In contrast, there was a marked and continuous rise in NO production by TG and IM in spite of the lower Ca 2ϩ i rise. These results proved that our method is reliable, although the NO measurements still required multiple sampling to minimize variation between experiments.
In addition to the NO determination, we employed biological action of native NO produced in ECs, because NO reduces the Ca 2ϩ i in cocultured VSMCs in a paracrine manner (7), dependent on the level of NO production. Using both the NO determination and the bioassay, we succeeded for the first time in developing the quantitative aspect of Ca 2ϩ i component responsible for NO production but also Ca 2ϩ i reduction in VSMCs that leads to the muscle relaxation.

Selection and Dose Determination of Agonists-All four agonists used in this study raised the Ca 2ϩ
i by different mechanisms and caused different peak Ca 2ϩ i levels and degeneration phases both in ECs and in VSMCs. ATP and BK provoked a Ca 2ϩ i rise in ECs through P 2y and B 2 receptors, respectively (22). The present results showing that ATP is more potent than BK are consistent with a previous report in its relationship to cGMP concentration in ECs (21). TG is a specific inhibitor of the Ca 2ϩ pump on endoplasmic reticulum and sarcoplasmic reticulum but is not an inhibitor of the Ca 2ϩ pump on the plasma membrane or of other ion-mobilizing ATPases (23). TG might increase the Ca 2ϩ i through a Ca 2ϩ leakage from the endoplasmic/sarcoplasmic reticulum to the cytoplasm and a secondary Ca 2ϩ influx following the depletion of the internal stores (24,25). This influx could be mediated by the Ca 2ϩ influx factor (26). IM is a Ca 2ϩ ionophore and might nonspecifically elevate the Ca 2ϩ i by the influx. The doses of these agonists were adjusted to appropriately raise the Ca 2ϩ i in ECs and produce NO within the measurable range of our system. For those results, we selected the doses determined in a preliminary study according to the following three criteria: (i) The maximum dose was not used to avoid eliciting a saturated response in both ECs and VSMCs and to accurately evaluate L-NMMA or indomethacin. The EC 50 for each agonist varied between ECs and VSMCs. For example, BK induced a larger response in ECs (EC 50 ϭ 30 nM) than in VSMCs (EC 50 ϭ 12 M). BK at 100 nM caused a significant response in ECs but a weak response in VSMCs. In contrast, the response to IM was stronger in VSMCs (EC 50 ϭ 7 nM) than in ECs (EC 50 ϭ 24 nM). The appropriate range of concentrations of IM was narrow for the Ca 2ϩ i measurement. (ii) Agonists should homogeneously excite ECs. All ECs responded to TG and IM, but only 80 -85% of the cells responded to ATP at 100 nM (EC 50 ϭ 400 nM) or BK at 10 nM. The lower the concentration of ATP or BK, the fewer the number of cells able to respond. The latent time from agonist application to the onset of the Ca 2ϩ i rise was more variable at the lower doses. (iii) The agonist concentration chosen should not induce the Ca 2ϩ i oscillation that makes the HDT measurement difficult. Low concentrations of ATP causes the oscillation, as reported by Lynch et al. (27). We used 10 M ATP that produced an exponential degeneration after the peak. BK rarely induced the oscillation, and TG or IM caused no oscillation at all.
Ca 2ϩ i Component in ECs Responsible for NO Production-The amount of NO production after the administration of each agonist was significantly higher in the presence of Ca 2ϩ o than in its absence (Fig. 3A). Many studies have shown that NO production nearly ceased without Ca 2ϩ o (5,6). The present study has indicated that the Ca 2ϩ release only from internal stores, as in the absence of Ca 2ϩ o , still has a small but significant effect on the actual production of NO after stimulation of ATP or IM (Fig. 3A), as also revealed by the blunt rise of Ca 2ϩ i in cocultured VSMCs (Fig. 5A versus Fig. 6F). The cause of the discrepancy might be due to the lower sensitivity of NO action, including cGMP assay (5,6).
Furthermore, the quantification of sustained Ca 2ϩ i elevation by the HDT or AUC revealed that the NO production rate was hyperbolically correlated to these two parameters but not to the peak Ca 2ϩ i level at all (Fig. 12). The simple Ca 2ϩ -calmodulin (CaM) scheme would not be applicable for the eNOS activation, because the scheme assumes a homogenous distribution of Ca 2ϩ , CaM, and eNOS. In the Ca 2ϩ i signaling mediated by inositol 1,4,5-trisphosphate where ATP and BK are concerned, the peak Ca 2ϩ i is supplied from the release of Ca 2ϩ from internal stores, independent of the Ca 2ϩ influx (Table I and Fig. 6F). After TG and IM stimulation, the peak Ca 2ϩ i would not be formed by the release of Ca 2ϩ from internal stores but chiefly by the entry of extracellular Ca 2ϩ , as confirmed the delay of peak time between in the presence (Fig. 4) and the absence of Ca 2ϩ o or the Ni 2ϩ treatment. 2 The Ca 2ϩ release from internal stores accounts for only 1 ⁄3 -1 ⁄8 of the Ca 2ϩ entry, as shown by the AUC (Table I). These results also suggest that the Ca 2ϩ release from internal stores play a less significant role in elevating the Ca 2ϩ i than the Ca 2ϩ influx from extracellular source in the cases of TG and IM stimulation. Compartment of either the Ca 2ϩ or NOS enzymes might explain the dissociation between peak Ca 2ϩ i and NO production; eNOS is located on the cytoplasmic membrane (28), where the entered Ca 2ϩ may di-   The activation of eNOS in ECs as well as constitutive NOS in brain is strictly Ca 2ϩ -CaM-dependent. Compared with the binding of inducible NOS to CaM, the binding of eNOS to CaM is loose and reversible (29). The Ca 2ϩ -dependent down-regulation of NOS mediated by the phosphorylation of NOS protein and the resultant decrease in the activity (30) or the dual regulation of constitutive NOS activity by Ca 2ϩ i (31) is not likely, because the Ca 2ϩ concentration they employed (10 -2000 M) exceeded the physiological Ca 2ϩ i range. Furthermore, we should be very careful if biochemical study in cell-free system does take place in vivo also.
The Ca 2ϩ i reduction in cocultured VSMCs became stable 3 min after ATP or BK stimulation (Fig. 5, 6 and 7), and it was sustained for up to 40 min after TG or IM stimulation. The long lasting action of TG and IM is compatible with a continuous production of NO (Fig. 2), because the biological lifetime of NO is very short (3). In the late phase of TG-induced Ca 2ϩ i transients, the low Ca 2ϩ i level, which was completely supported by the Ca 2ϩ influx, stimulated ECs to produce large amounts of NO. These results suggest that Ca 2ϩ influx might trigger NO production and that the activation of NOS does not require high concentrations of Ca 2ϩ i to sustain the NO production. NO Production without Agonist-High concentration of Larginine alone weakly decreased the Ca 2ϩ i in ECs and markedly reduced the Ca 2ϩ i in cocultured VSMCs. These Ca 2ϩ i modulations were mediated by NO, because NO was actually released from ECs and detected in the medium (Fig. 11). Furthermore, D-arginine or L-lysine has no NO producing action (data not shown). Thus, both the Ca 2ϩ i handling and NO production are modified specifically by the high concentrated Larginine, and these phenomena are not due to nonspecific charge effect nor basic amino acid.
In addition, the NO production induced by L-arginine was not dependent on Ca 2ϩ o (Fig. 11). The L-arginine study cannot be explained by the Ca 2ϩ -CaM theory and suggests that a basal or even lowered Ca 2ϩ i level within ECs is sufficient to activate eNOS, if a high dose of substrate is supplied. No biochemical data are available on the relationship between NOS activation and the substrate concentration. After the NO is produced, the L-arginine concentration might not significantly reduce, because it is supplied from the glutamine or recycled from Lcitrulline within a cell (32). However, several clinical reports (33,34) indicated that L-arginine might potentiate the NO production. Present studies gave direct evidence for that.
Our previous data with the measurement of cGMP (35) have suggested that NO is basically released in vivo without agonist stimulation, just under shear stress. The basal release of NO evident in this study could also be present in vivo (36), because vascular ECs are usually under shear stress. The Ca 2ϩ i in ECs is dependent on the blood flow rate (35), which could stimulate basal NO synthesis. The subsequent NO production determined by physiological flow could serve to inhibit the proliferation of ECs (18) or VSMCs (37) and prevent platelet adhesion

FIG. 12. Correlation between the HDT of Ca 2؉
i in the degeneration phase and NO production (A), AUC and NO production (B), or the peak Ca 2؉ i level and NO production (C). Each point denotes the mean Ϯ S.E. (n ϭ 12). (38), suppressing the progression of arteriosclerosis or thrombus formation in the vessel wall.
NO Action on Ca 2ϩ i Handling in VSMCs-The Ca 2ϩ i of VSMCs varies to the agonist stimulation, dependent on culture conditions (9,10). All four agonists studied directly increased the Ca 2ϩ i in isoculture of VSMCs while indirectly decreasing the Ca 2ϩ i through the action of NO in coculture with ECs. Therefore, the observed Ca 2ϩ i dynamics of cocultured VSMCs should be viewed as the net difference of the two opposite actions. ATP, BK, and TG that directly raised the Ca 2ϩ i was counterbalanced by the lowering of the Ca 2ϩ i by NO, resulting in a net Ca 2ϩ i reduction (Figs. 6 -8). IM induced no net Ca 2ϩ i reduction in cocultured VSMCs but hampered the Ca 2ϩ i rise expected in isocultures of VSMCs (Fig. 10), indicating that the direct Ca 2ϩ i raising action of IM on VSMCs exceeded the Ca 2ϩ i lowering action of NO present in cocultures with ECs. NO might have several routes to reduce the Ca 2ϩ i in VSMCs as follows: (i) NO activates a guanylate cyclase in the soluble fraction of VSMCs and increases the intracellular concentration of cGMP, which accelerates the efflux of Ca 2ϩ (39), (ii) NO may directly inhibit Ca 2ϩ entry through voltage-dependent Ca 2ϩ channels (40) or may enhance the outward K ϩ current by activating Ca 2ϩ -dependent K ϩ channels that cause hyperpolarization (19), and (iii) NO could potentiate the Na ϩ -Ca 2ϩ exchange (41).
The exogenous administration of NO gas is imprecise to exactly control NO concentration without oxidation to NO 2 Ϫ in the incubation medium. NO donors, such as sodium nitroprusside and s-nitroso-n-acetyl-D,L-penicillamine, are also inadequate for this quantitative purpose because the rate of NO supply is uncertain. This is the first report that for the most part resolves these problems by using NO naturally produced from ECs. By combining NO determination with NO bioassay measuring the Ca 2ϩ i transients of ECs and VSMCs in coculture, we have succeeded both in identifying the significant component of the Ca 2ϩ i responsible for NO production in ECs and in quantifying the relationship between NO release and the subsequent Ca 2ϩ i reduction in VSMCs. The present methods of coculturing different cell types, of measuring intracellular information by two-dimensional imaging of each cell, and of simultaneously determining released factors are of great value not only in the understanding of vascular biology but also in the examination of the relationships between a wide variety of cell populations that coexist in close proximity and require intracellular communication to regulate their mutual interaction.