INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

ANGIOGENESIS IS A PATHOLOGICAL neovascularization in the adult and activated only in pathological conditions such as tumor growth and metastasis, diabetic retinopathy, and wound healing (7). Degradation of the basement membrane followed by endothelial migration, proliferation, and tube formation lead to angiogenesis (9), and, therefore, the inhibition of any of these processes would be a good therapeutic strategy for these angiogenesis-related disorders (8). Actually, suppression of angiogenesis has been proven effective for inhibiting tumor growth (2) and diabetic retinopathy (6).

Ca²⁺ homeostasis plays a significant role in regulating endothelial functions including angiogenesis (12). It has been reported that nonspecific inhibition of Ca²⁺ entry pathways with carboxyamidotriazole suppressed endothelial growth, adhesion, migration, and in vitro tube formation (12). The same group also reported that endothelial spreading on type IV collagen, which is also necessary for endothelial growth, requires the activation of RhoA, and spreading induces focal adhesion kinase (FAK) phosphorylation, both of which are regulated by Ca²⁺ entry (1, 15). Therefore, Ca²⁺ entry is believed to play a critical role in angiogenesis.

Vascular endothelium forms a confluent monolayer in vivo. Wounding the confluent endothelial monolayer stimulates endothelial migration in vitro because of a release from contact inhibition, and this "wound method" has been used for the investigation of the characteristics of endothelial migration (12). It has been reported that wounding-induced intracellular Ca²⁺ concentration ([Ca²⁺]_i) elevation in the wound surface due to Ca²⁺ entry from extracellular space and the inhibition of Ca²⁺ entry with Gd³⁺ suppressed endothelial migration (23). The authors (23) speculated that the wound-induced increase in [Ca²⁺]_i stimulates the transcription of an immediate early gene, which may be important for cell motility.

It is well known that endothelial cells generate Ca²⁺ transients in response to changes in cell shape or stretch state of the membrane (5, 18). Because the shape of endothelial cells is supposedly altered dynamically during cell migration, it is conceivable that endothelial migration would be accompanied by Ca²⁺ responses. Therefore, the investigation of Ca²⁺ mobilizing properties in migrating endothelial cells would provide important information for the understanding of the physiology of angiogenesis. Brundage et al. (3) reported an intracellular gradient in Ca²⁺ distribution in eosinophils migrating toward chemotactic stimuli, i.e., [Ca²⁺]_i is low in the leading edge and high in the trailing edge. They discussed that this polarization in [Ca²⁺]_i might be related to the cytoskeletal reorganization that was involved in cell migration (3). However, endothelial migration is much slower than the movement of leukocyte, so this phenomenon has not been considered so far in relation to endothelial migration. In this study, we investigated the characteristics of Ca²⁺ mobilizing properties in migrating, confluent, and nonconfluent cells. Our results provide the first evidence for the possible relationship between endothelial migration and the alterations of Ca²⁺ mobilizations.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Culture of bovine aortic endothelial cells. The bovine thoracic aorta of a one-year-old calf was obtained from the local slaughter house. Bovine aortic endothelial cells (BAEC) were scraped off from the intima with the edge of a razor. The collected endothelial cells were cultured in DMEM (containing 1.8 mM Ca²⁺; GIBCO Life Technologies; Rockville, MD) supplemented with 10% fetal bovine serum. Cells of the second subculture were used. The identification of endothelial cells and the absence of contamination of other cell types were confirmed by the specific uptake of acetylated low-density lipoprotein (Ac-LDL; Fig. 1A, top), whereas bovine aortic smooth muscle cells, obtained by the explant method (4), did not show the uptake of Ac-LDL (Fig. 1A, bottom). Furthermore, as shown in Fig. 1B, BAEC showed honeycomb-like structure on Matrigel, thereby indicating that they could form tubular structures and be related to angiogenesis.

View larger version (102K): [in this window] [in a new window]   Fig. 1.   Bovine aortic endothelial cells (BAEC) were used in the present study. A: identification of the cells as endothelium. Cells were incubated with fluorescence-labeled acetylated low-density lipoprotein (Dil-Ac-LDL) for 4 h. BAEC show the incorporation of Dil-Ac-LDL (top), whereas bovine aortic smooth muscle cells (BASMC) do not show fluorescence (bottom). Scale, 50 µm. B: BAEC seeded on Matrigel showed a honeycomb-like network formation after 6 h. Scale, 500 µm. C: when cells were sparsely seeded (200 cells/cm2), they remained nonconfluent even on the 4th day after seeding. D: endothelial cells, which were seeded at a higher density (2,500 cells/cm2), grew in confluence after 4 days. After the confluent monolayer was wounded with a fine razor blade, cells migrated from the wound surface. a, Confluent monolayer; b, wound surface; c, migrating cells. A typical picture 10 h after wounding is shown. Scale, 50 µm. E: incorporation of bromodeoxyuridine (BrdU) into the nucleus was examined as an indicator of DNA synthesis. Cells treated with vascular endothelial growth factor (VEGF; 10 ng/ml) for 6 h showed marked incorporation of BrdU into the nucleus (a). In contrast, only several cells (arrowheads) out of more than 50 migrating cells showed BrdU incorporation 6 h after wounding (b). Scale, 50 µm.

Measurement of [Ca²+]_i. [Ca²⁺]_i was measured from BAEC using an Attofluor digital fluorescence microscopy system (Atto Instruments; Rockville, MD). Cells were loaded with fura 2 by incubating them with fura 2-AM (Dojindo; Kumamoto, Japan) for 20 min at 37°C. The coverslip with fura 2-loaded cells was placed on a chamber of 0.5-ml volume and mounted on an inverted microscope (Axiovert 135, Zeiss; Jena, Germany). Fura 2 was excited alternatively at two wavelengths (340 and 380 nm), and the emitted fura 2 fluorescence images were recorded into a rewritable optical disc recorder (LQ-4100A, Panasonic; Osaka, Japan) at a rate of ~1 Hz. From the fluorescence images of each cell, we calculated the fluorescence ratios (R = F₃₄₀/F₃₈₀, where F₃₄₀ and F₃₈₀ are the fluorescences excited with 340 and 380 nm, respectively), which were then converted into the apparent [Ca²⁺]_i using the equation

(1)

where K_d is the dissociation constant between Ca²⁺ and fura 2; S_f2 and S_b2 are the fura 2 fluorescences emitted by 380 nm at zero Ca²⁺ and saturating Ca²⁺, respectively; and R_min and R_max the fluorescence ratios at zero Ca²⁺ and saturating Ca²⁺, respectively. These constants were obtained by the in vitro calibration method, i.e., by measuring fura 2 fluorescent intensities in three standard solutions with zero, 300 nM, and 1 mM of free Ca²⁺ concentration. A transillumination picture was also taken in each experiment, and the picture was used for the analysis of [Ca²⁺]_i to determine the edge of the cells.

Recording of cell images and cell number counting. Microscopic images of endothelial cells were observed with a charge-coupled device camera mounted on a microscope (Diaphot TMD with Nomarski optics, Nikon; Tokyo, Japan), processed with an image processor (SRM-100, Nikon), and captured into computer through a video-capture board (GV-VCP/PCI, I-O Data; Kanazawa, Japan).

Measurement of DNA synthesis. Cellular DNA synthesis was examined through the incorporation of bromodeoxyuridine (BrdU) into the nucleus using a commercial kit (Cell proliferation kit, Amersham Pharmacia Biotech; Buckinghamshire, UK) according to the manufacturer's instruction.

Immunological staining of endothelial F-actin. Rearrangement of F-actin by hypotonic stress was examined immunologically using rhodamine-conjugated phalloidin (Molecular Probes; Eugene, OR) according to previously reported methods (11).

Solutions and drugs. Modified Krebs solution (1.5 mM Ca²⁺ solution) was used as the standard extracellular solution, which contained (in mM) 132 NaCl, 5.9 KCl, 1.2 MgCl₂, 1.5 CaCl₂, 11.5 glucose, and 11.5 HEPES; pH was adjusted to 7.3 with NaOH. Ca²⁺-free Krebs solution was made by substituting CaCl₂ of Krebs solution with 1 mM EGTA. All other drugs were from Sigma.

Data analysis. Pooled data are given as means ± SE. Statistical significance between two groups was determined using Student's unpaired and paired t-test for comparing different cell groups and the [Ca²⁺]_i gradient, respectively, and one-way ANOVA and Scheffe's post hoc test were used for comparing three or more groups. P < 0.05 was regarded as significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Basal [Ca²+]_i in confluent and migrating cells. When cells were sparsely seeded (100-500 cells/cm²), they were isolated from neighboring cells up to 5 days after seeding ("nonconfluent cells"; Fig. 1C). In contrast, cells were confluent after 4 days when seeded densely (2,500 cells/cm²). Wounding the confluent cell monolayer with a fine razor blade induced cell migration from the wound surface (Fig. 1D). The experiments were performed 6 h after wounding. The cellular DNA synthesis, assessed by the incorporation of BrdU into the nucleus, was not significantly accelerated in migrating cells after 6 h (Fig. 1E). Therefore, we suppose that the characteristics of migrating cells shown in the present study can be attributed to cell migration but not to cell proliferation. The values of "migrating cells" were measured from cells that were apparently discrete from the wound surface and those of "confluent cells" from cells at least two cells width inside the wound surface.

Thapsigargin-induced Ca²+ leak and Ca²⁺ entry in confluent and migrating endothelial cells. In nonconfluent cells, thapsigargin (TG; 1 µM), a specific irreversible inhibitor of sarco(endo)plasmic reticulum Ca²⁺-ATPase (SERCA) (22), evoked a transient gradual [Ca²⁺]_i increase in Ca²⁺-free solution due to Ca²⁺ leak from intracellular Ca²⁺ store sites (Fig. 2A) (22). Migrating cells showed a rapid initial [Ca²⁺]_i increase (Fig. 2C) compared with confluent (Fig. 2B) or nonconfluent cells (Fig. 2A). The time required for the initial Ca²⁺ transient to reach its peak was significantly shorter in migrating cells than confluent and nonconfluent cells (Fig. 2D). We also calculated the time integral of the initial Ca²⁺ transient per cell area as an indicator of total stored Ca²⁺. This value was significantly larger in migrating cells than in nonconfluent or confluent cells (Fig. 2E). On the other hand, the elevated initial Ca²⁺ transient declined to the same level as the basal [Ca²⁺]_i in each condition (Figs. 2, A-C).

View larger version (31K): [in this window] [in a new window]   Fig. 2.   Thapsigargin (TG)-induced intracellular Ca2+ concentration ([Ca2+]i) elevation in nonconfluent, confluent, and migrating endothelial cells. A: TG (1 µM) induced an initial Ca2+ transient in Ca2+-free solution in nonconfluent BAEC. Subsequent application of Ca2+-containing solution induced a further increase in [Ca2+]i . Trace from a representative cell is shown. B: similar TG-induced [Ca2+]i elevation was also observed in confluent BAEC. Note that the Ca2+ reapplication-induced [Ca2+]i increase was slightly smaller than that in nonconfluent cells. C: TG (1 µM) induced a steeper initial Ca2+ transient in migrating BAEC. The application of Ca2+ induced a steep increase in [Ca2+]i. Note that the basal level of [Ca2+]i was lower than in A and B. D, right: time to peak of the TG-induced initial Ca2+ transient was compared among nonconfluent, confluent, and migrating cells. D, left: how this value and the other two values were calculated. E: time integral of the initial Ca2+ transient in Ca2+-free solution was also calculated. F: maximal rate of [Ca2+]i increase of Ca2+ reapplication-induced [Ca2+]i increase (d[Ca2+]i-TG/dt) was calculated. The values of the cells on the same coverslip were averaged, and these averaged values were used for the statistical analysis in D-F. Numbers in parentheses indicate the number of experiments (coverslips). *P < 0.05 and **P < 0.01 vs. nonconfluent cells. nd, Not determined.

ATP-induced Ca²+ oscillations in confluent and migrating endothelial cells. We then examined the characteristics of agonist-induced Ca²⁺ transients in migrating and confluent cells. We (10) have previously shown that ATP induces Ca²⁺ oscillations in BAEC, which requires the integrity of various Ca²⁺ mobilizing pathways. ATP (0.3 µM) induced Ca²⁺ oscillations in both migrating (Fig. 3A) and confluent cells (Fig. 3B). However, the amplitude of the maximal Ca²⁺ peak was significantly larger in migrating cells (Fig. 3C). The ATP-induced Ca²⁺ transient was inhibited by phospholipase C inhibitors (neomycin and U-73122, data not shown), thereby indicating that the Ca²⁺ transient was due to D-myo-inositol (1,4,5)-trisphosphate-induced Ca²⁺ release from intracellular Ca²⁺ store sites.

View larger version (21K): [in this window] [in a new window]   Fig. 3.   ATP-induced Ca2+ oscillations in confluent and migrating BAEC. ATP (0.3 µM) treatment showed Ca2+ oscillations in both migrating (A) and confluent (B) cells. The peak amplitude of oscillations was significantly larger in migrating cells than in confluent cells (C). Numbers in parentheses indicate the number of experiments. *P < 0.05.

Intracellular polarization of Ca²+ in confluent and migrating endothelial cells. The results above suggest that cell migration may be accompanied by a low basal [Ca²⁺]_i, increase in stored Ca²⁺, acceleration of TG-evoked Ca²⁺ entry, and augmentation of Ca²⁺ release. Alterations in intracellular Ca²⁺ distribution have been reported in migrating eosinophils, as described in the introduction (3). We therefore examined whether such an intracellular polarization of Ca²⁺ is also observed in migrating endothelial cells.

View larger version (41K): [in this window] [in a new window]   Fig. 4.   Intracellular polarization of [Ca2+]i in migrating BAEC. A: TG (1 µM) induced a larger amplitude of Ca2+ entry in the leading edge (continuous trace) than in the trailing edge (dotted trace). Note that the basal [Ca2+]i is lower in the leading edge than in the trailing edge. B: ATP (0.3 µM) induced Ca2+ oscillations in the leading edge (continuous trace) and trailing edge (dotted trace). The amplitude of Ca2+ oscillations is different between these two conditions. Inset, raw fura 2 intensities [fluorescence at 340 nm (F340) and 380 nm (F380)] from the leading and trailing edges. Note that the trailing edge shows a larger F340 and a smaller F380 than the leading edge. C: fluorescent image of migrating cells and an example of the area from where each value was obtained. D: statistical analysis of the basal level of [Ca2+]i (a), d[Ca2+]i-TG/dt after TG-induced store depletion (b), and ATP-induced [Ca2+]i (c). [Ca2+]i-peak, peak change in [Ca2+]i. Numbers in parentheses indicate the number of cells. **P < 0.01.

View larger version (33K): [in this window] [in a new window]   Fig. 5.   Absence of the intracellular polarity of [Ca2+]i in confluent BAEC. A: TG (1 µM) induced completely identical Ca2+ responses from both sides of the cell. The Ca2+ trace from one side (edge a) is shown as a continuous trace and the other side (edge b) is shown as a dotted trace. Note that these two traces are almost completely overlapped. B: ATP (0.3 µM) also induced identical Ca2+ oscillations at both sides of the cell (continuous and dotted traces, edge a and edge b, respectively). C: fluorescent image of confluent cells and an example of the area from which the values were obtained.

Effects of SERCA inhibitors on cell migration in BAEC. The results above suggest that the intracellular polarization of Ca²⁺ signals and the alterations of the Ca²⁺ store site-related Ca²⁺ mobilizations may be related to endothelial migration. However, it is still not clear whether altered properties of Ca²⁺ are the cause or result of cell migration. We therefore examined the effects of the SERCA inhibitors TG and cyclopiazonic acid (CPA) on cell migration.

View larger version (78K): [in this window] [in a new window]   Fig. 6.   BAEC monolayer was cultured for 6 h after wounding in the absence (A) and in the presence (B) of 1 µM TG. TG suppressed the migration of BAEC almost completely. Bar, 50 µm scale.

View larger version (16K): [in this window] [in a new window]   Fig. 7.   Statistical analysis of cell migration and the effects of the sarco(endo)plasmic reticulum Ca2+-ATPase (SERCA) inhibitors TG and cyclopiazonic acid (CPA). A: TG inhibited cell migration after both 6 h () and 24 h () in a concentration-dependent manner. The number of migrated cells per 500 µm of wound surface was measured. **P < 0.01 vs. control (24 h), ¶¶P < 0.01 vs. control (6 h). B: effects of a reversible inhibitor of SERCA, CPA, on [Ca2+]i (a) and cell migration (b). [Ca2+]i was increased by application of CPA in Ca2+-free solution. Subsequent application of extracellular Ca2+ induced a sustained increase in [Ca2+]i (a). Migration of BAEC was also significantly suppressed by CPA in a concentration-dependent manner (b). Note that cell migration was restored to the control level after 24 h by washing CPA out at 6 h ().

Staining of actin filament of migrating cells. Actin fiber is one of the main cytoskeletal components of the endothelium and is related to cell motility (21). Furthermore, actin rearrangement is regulated by Ca²⁺ (16). The results above, therefore, suggest that the alterations in intracellular Ca²⁺ distribution might control actin rearrangement and thereby contribute to cell migration.

View larger version (97K): [in this window] [in a new window]   Fig. 8.   A: staining of the actin cytoskeleton in migrating cells. Note that the actin cytoskeleton is formed in the sagittal direction, and there is no difference between the leading and trailing edges. Arrow, direction of migration of the endothelial cells. Scale, 25 µm. B: confluent cells showing no apparent formation of actin fiber. Bar, 25 µm scale.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Ca²⁺ leak from the intracellular store sites was accelerated in migrating endothelial cells (Fig. 2D). The total amount of stored Ca²⁺ was also increased in migrating cells, because the time integral of TG-induced Ca²⁺ leak was augmented (Fig. 2E). Furthermore, TG-evoked Ca²⁺ entry was accelerated in migrating cells (Fig. 2F). The alteration of Ca²⁺ mobilization was also observed in ATP-induced Ca²⁺ oscillations, where the amplitude of the Ca²⁺ transient was augmented in migrating cells (Fig. 3). Because these are Ca²⁺ mobilizing properties related to the intracellular Ca²⁺ store sites, these may indicate that Ca²⁺ stores play a significant role in endothelial migration. Kohn et al. (12) reported that a nonspecific inhibition of Ca²⁺ entry with carboxyamidotriazole suppressed cell migration (12). However, in the present study, TG and CPA, both of which induce sustained activation of Ca²⁺ entry, did not accelerate but rather inhibited cell migration (Fig. 7). We therefore suppose that the acceleration of Ca²⁺ entry alone does not directly lead to cell migration but that the integrity of Ca²⁺ store site-related Ca²⁺ mobilizations may have an essential significance in endothelial migration.

In the present study, the basal [Ca²⁺]_i was lower in migrating cells than in confluent or nonconfluent cells. However, because the difference of the basal [Ca²⁺]_i was maintained even after TG-induced store depletion, it seems that the alteration of the basal [Ca²⁺]_i was not due to the TG-sensitive store site-related one but to other alterations such as TG-insensitive Ca²⁺ stores, Ca²⁺ extrusion mechanisms, cytosolic Ca²⁺-binding proteins, or mitochondrial Ca²⁺ uptake. However, mitochondria are probably not involved because ruthenium red did not change the Ca²⁺ gradient in migrating cells. Therefore, at least, there is a possibility that some of the properties of TG-insensitive Ca²⁺ stores, Ca²⁺ extrusion, or cytosolic Ca²⁺-binding properties are also altered in migrating endothelial cells.

Many of the migrating cells were isolated from neighboring cells (Figs. 1D and 6A). However, these alterations of Ca²⁺ mobilization are not due to the cell detachment itself, because nonconfluent cells, which are also isolated from neighboring cells, did not show any of these changes and most of Ca²⁺ mobilizing properties in nonconfluent cells were similar to those in confluent cells (Fig. 2). Furthermore, we suppose that the alterations of Ca²⁺ mobilization were not artifacts due to the unequal loading of fura 2 dye, because the intensities of fura 2 fluorescence were in the same range between migrating and confluent cells and between leading and trailing edges [Fig. 4B (inset)]. Also, these alterations could not be attributed to a methodological artifact that we used the in vitro calibration method for calculating [Ca²⁺]_i, because 1) the alterations of Ca²⁺ mobilization, except for d[Ca²⁺]_i-TG/dt, were observed only in migrating cells but not in nonconfluent and confluent cells (Fig. 2); 2) SERCA inhibitors that disrupt the alterations of Ca²⁺ mobilizing properties suppressed cell migration (Figs. 6 and 7); and 3) the importance of [Ca²⁺]_i alterations for endothelial migration has been suggested by other investigators with both in vivo (24) and in vitro (25) calibration methods.

We also observed an intracellular polarization of Ca²⁺ in migrating cells (Fig. 4) but not in confluent cells (Fig. 5). The leading edge of migrating cells showed a lower basal [Ca²⁺]_i and accelerated TG-evoked Ca²⁺ entry compared with the trailing edge. Therefore, we considered that the characteristic alterations of Ca²⁺ store-related Ca²⁺ mobilizations mainly occurred in the leading edge of migrating cells. An intracellular polarization of Ca²⁺ has also been reported in eosinophils moving to chemotactic stimuli, although the movement of eosinophils was much faster [11.4 µm/min (3)] than endothelial migration (~20 µm/h). It has been speculated that this eosinophil movement may be related to some Ca²⁺-sensitive mechanisms of cell migration (3). However, it has not been clarified by which mechanism this polarity was formed in eosinophils. The present study revealed the possibility that the alterations in Ca²⁺ store-related Ca²⁺ mobilizing properties may be related to the formation of this polarity in the endothelium, and, therefore, this would also provide some cue for the understanding of the formation of Ca²⁺ polarity of migrating eosinophils.

We suppose that the alterations of Ca²⁺ store site-related Ca²⁺ mobilizations are not a result of migration but are necessary for the cells to migrate, because the inhibition of Ca²⁺ sequestration by TG and CPA inhibited cell migration (Figs. 6 and 7). However, the present study cannot explain why these alterations lead to cell migration. At least the actin cytoskeleton seems to not be involved, because we did not observe a difference in actin fiber formation between the leading and trailing edge (Fig. 8). However, we can not exclude the possible involvement of other cytoskeletal components such as myosin II (13). Furthermore, the replenishment of cell adhesion receptors such as integrin may be obtained by a Ca²⁺ gradient, because the crawling of the cell requires the continuous recruitment of these adhesion receptors from the rear to the front of the cell (for a review, see Ref. 19). Further investigations are needed for revealing the precise significance for Ca²⁺ polarity and the alterations of Ca²⁺ mobilizing properties in the migration of endothelial cells.

Efficient inhibition of angiogenesis has been proposed as a therapeutic approach to solid tumors (2). The present results suggest that the disruption of Ca²⁺ store site-related Ca²⁺ mobilization by the inhibitors of SERCA and/or Ca²⁺ entry could become a novel approach for tumor therapy.

In summary, we show here the first evidence that Ca²⁺ mobilizing properties, especially Ca²⁺ store site-related ones, are altered in the migrating vascular endothelium.