| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Pathology, Yale University School of Medicine, New Haven Connecticut, USA; and
* Max-Planck Institute for Vascular Biology, Muenster, Germany
1Correspondence: Department of Pathology, Lauder Hall 115, Yale University School of Medicine, New Haven, CT 06510, USA. E-mail: joseph.madri{at}yale.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. We found that endothelial cells lacking PECAM-1 exhibit increased single cell motility and extension formation but poor wound healing migration, reminiscent of cells in which Rho activity has been suppressed by overexpressing a GTPase-activating protein (3)
. The ability of PECAM-1 to restore wound healing migration to PECAM-1-deficient cells was independent of its extracellular domain or signaling via its immunoreceptor tyrosine-based inhibitory motif. PECAM-1-deficient endothelial cells had a selective defect in RhoGTP loading, and inhibition of Rho activity mimicked the PECAM-1-deficient phenotype of increased chemokinetic single cell motility at the expense of coordinated wound healing migration. The wound healing advantage of PECAM-1-positive endothelial cells was not only Rho mediated but pertussis toxin inhibitable, characteristic of migration mediated by heterotrimeric G-protein-linked seven-transmembrane receptor signaling such as signaling in response to the serum sphingolipid sphingosine-1-phosphate (S1P) (4
, 5)
. Indeed, we found that the wound healing defect of PECAM-1 null endothelial cells is minimized in sphingolipid-depleted media; moreover, PECAM-1 null endothelial cells fail to increase their migration in response to S1P. We have also found that PECAM-1 localizes to rafts and that in its absence heterotrimeric G-protein components are differentially recruited to rafts, providing a potential mechanism for PECAM-1-mediated coordination of S1P signaling. PECAM-1 may thus support the effective S1P/RhoGTP signaling required for wound healing endothelial migration by allowing for the spatially directed, coordinated activation of Galpha signaling pathways.—Gratzinger, D., Canosa, S., Engelhardt, B., Madri, J. A. Platelet endothelial cell adhesion molecule-1 modulates endothelial cell motility through the small G-protein Rho.
Key Words: CD31 • vascular endothelium • cell movement • small G-protein • sphingosine-1-phosphate
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 7)
. The cytoplasmic domain of PECAM-1 contains an immunoreceptor tyrosine-based inhibitory motif (ITIM) that binds and activates SH2 domain-containing proteins when phosphorylated (8
, 9)
. PECAM-1 modulates endothelial cell migration in vitro and angiogenesis both in vitro and in vivo (1
, 2)
. The PECAM-1 knockout mouse is viable and fertile, with no overt vascular abnormalities (10)
. However, defects in angiogenesis into a polyvinyl acetate sponge are seen in a model of foreign body inflammation in the PECAM-1 knockout mouse (11)
. At the same time, some aspects of endothelial cell motility are enhanced in the absence of PECAM-1: in a model of cardiac cushion development (12)
, migration of individual endocardial cells onto a type I collagen gel is preserved in the presence of high glucose in explants derived from the PECAM-1 knockout mouse despite decreased VEGF-mediated signaling (13)
.
Rho family GTPases play prominent roles in vascular physiology, including the rearrangement and migration of confluent endothelial cells and angiogenesis in vitro and in vivo (14
15
16
17)
. Rho family GTPases including Rac, Rho, and Cdc42 act in a coordinated manner to regulate cell extension formation, motility, and directionality by modulating actin cytoskeletal and focal contact dynamics (18
, 19)
. S1P is a platelet-released lipid mediator of endothelial chemotaxis, wound healing endothelial migration, VEGF- and bFGF-induced angiogenesis, and vascular maturation that represents the majority of the endothelial chemotactic activity present in serum (4
, 20
, 21)
. S1P engages Endothelial differentiation gene (EDG) or S1P seven-transmembrane receptors to mediate endothelial cell migration and angiogenesis through heterotrimeric G-protein signaling. Coordinated S1P signaling via EDG1/SIP1 activates Galphai2, localizing and activating focal contact components (22)
as EDG3/S1P3 activates Galpha13 to appropriately remodel the actin cytoskeleton via Rho signaling (23
, 24)
.
Lipid rafts are cholesterol-rich, liquid-ordered subdomains within the liquid-disordered plasma membrane (25)
. Such lipid raft domains have been found to help segregate associated proteins to the leading edge vs. the rear of certain polarized migrating cells (26)
. Some signaling proteins including heterotrimeric G-proteins, src, and H-ras localize to these endothelial subdomains (27
, 28)
, providing spatially localized signaling information that could potentially contribute to polarized cellular behaviors, including migration. Caveolin-positive, low density lipid raft domains have been found to interact with various cell–cell junctional proteins including components of adherens junctions (29)
, tight junctions (30)
, and even gap junctions (31)
, and as such could play a role in the loosening of cell–cell adhesions that permits coordinated migration of cells into a wound to occur.
We have examined the role of PECAM-1 in promoting coordinated wound healing migration vs. the chemokinetic motility of individual endothelial cells in the absence of a signaling gradient via modulation of RhoGTP. Single cell motility is prominently involved in developmental and physiologic processes such as angioblast migration during neural tube vasculogenesis (32)
and migration of endothelial cells through areas of local basement membrane degradation during sprouting angiogenesis (33)
. We have further examined the contribution of S1P and heterotrimeric G-protein-mediated signaling to the PECAM-1-mediated promotion of coordinated wound healing migration. Coordinated wound healing migration necessarily involves response to multiple directional signals, as wounding induces release of chemotactic factors at the leading edge and uncovers haptotactic factors previously laid down by the endothelial monolayer. For example, injury induces endothelial release of fibroblast growth factor-2 (FGF-2), which contributes to the endothelial wound healing response in vitro (34
, 35)
. In vivo, denuded basement membrane promotes platelet aggregation and release of multiple chemotactic factors, including S1P, which also promotes endothelial wound healing migration (4)
. Finally, exposed extracellular matrix components such as fibronectin themselves provide strong haptotactic cues to endothelial wound healing and angiogenesis (36)
. Wound healing migration models physiologically relevant in vivo cell behaviors such as the reendothelialization of denuded vasculature following balloon angioplasty or the remodeling of vascular endothelium under variant flow conditions.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and human (2)
PECAM-1 were generated in our lab. mAb to ERK2 and caveolin and pAb to paxillin PY118 were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). mAbs to Rac1, Rho, and FAK were purchased from BD Biosciences, Transduction Laboratories (San Diego, CA, USA). mAbs to RhoGAP p190 and pAbs to FAK PY397 and RasGAP p120 were purchased from Upstate (Lake Placid, NY). mAb to ß actin was purchased from Sigma-Aldrich (St. Louis, MO, USA). Anti-phosphoERK pAb was purchased from Cell Signaling Technology (Beverly, MA, USA). pAbs to heterotrimeric G-protein subunits Galphai2 and Galpha13 were purchased from Transduction Laboratories. Rhodamine-phalloidin and methyl ß cyclodextrin was purchased from Sigma-Aldrich. Botulinum exoenzyme C3, pertussis toxin, S1P, and Y27632 were purchased from Calbiochem-Novabiochem (San Diego, CA, USA). Complete Protease InhibitorTM (CPI) was purchased from Roche Diagnostics (Mannheim, Germany). Cell culture and transfection reagents were from Life Technologies (Grand Island, NY, USA).
Full-length human PECAM-1 (see Fig. 4
c) in the mammalian expression vector APEX-1 (Alexion, New Haven, CT) and its Y663F mutant have been described (2)
. An additional mutation at Y686F was made using the QuikChangeTM Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions to produce an ITIM-defective construct. A myc-tagged truncated construct in APEX-1 lacks most of the ectodomain (38)
. The Rho binding domain of Rhotekin in a GST fusion protein expression construct (pGEX2T, Pharmacia, Piscataway, NJ, USA) was a gift from X. D. Ren and M. A. Schwartz (39)
. A GST fusion protein expression construct of the Rac binding region of PAK, pGEX4T-1 
PAK70-106, was a gift from F. Giancotti (40)
. pIRES-EGFP was from BD Biosciences, Clontech (Palo Alto, CA, USA). pCDNA3-VCAM mammalian expression construct was made by subcloning the full-length cDNA purchased from Biogen (Cambridge, MA, USA). Antisense oligonucleotide 5' TCCTTCCAGGGATGTGATC 3' for human PECAM-1 and control scrambled oligonucleotide 5' TTCTACCTCGCGCGATTTAC 3' were the gift of F. Bennett and T. Condon at IsisTM Pharmaceuticals (Carlsbad, CA, USA).
|
Cell culture, immunostaining, and transfections
Endothelioma cell line luEnd.PECAM-1.1 (PecamKO) was established by retroviral transduction of primary lung endothelial cells derived from the PECAM-1 knockout mouse with the polyoma virus middle T oncogene (41)
. PecamKO cells were retrovirally transduced with full-length murine PECAM-1 cDNA to generate a PecamRC (reconstituted) cell line (41
, 42)
. Cells were cultured in luEnd media [Dulbecco’s modified Eagle medium (DMEM), 10% fetal bovine serum (FBS), 10 mM HEPES, 2 mM L-glutamine, 1% nonessential amino acids, pyruvate, 10-5 M 2-mercaptoethanol, and antibiotics]. Selection of PECAM-1 expression was maintained with 1 µg/mL puromycin. Human umbilical vein endothelial cells (HUVEC) were purchased from the BCMM cell culture core (Yale Medical School) and cultured on gelatin in HUVEC media [M199 medium, 20% FBS, 50 mg/mL endothelial cell growth supplement (ECGS), 50 mg/mL heparin, 10 mM HEPES, 2 mM L-glutamine, and antibiotics]. For immunostaining, PecamRC and PecamKO cells were incubated overnight in luEnd media on 8-chamber fibronectin-coated glass culture slides (Falcon, Becton Dickinson, Franklin Lakes, NJ, USA). Cells were fixed in Streck’s Tissue Fixative (STF, Streck Laboratories, La Vista, NE, USA), permeabilized with 0.5% Triton X-100 in TBS, and stained with rhodamine-phalloidin (Sigma) 1:200 or anti-phosphopaxillin (Santa Cruz) 1:5000, followed by donkey anti-rabbit Cy3 (Jackson Immunoresearch Labs, West Grove, PA, USA) at 1:200. After coverslipping with Antifade Mounting Media (Molecular Probes, Eugene, OR, USA) cells were photographed using a Carl Zeiss Research microscope (Göttingen, Germany) and SPOT digital camera (Diagnostic Instruments Inc., Sterling Heights, MI, USA). For transient transfections, subconfluent PecamKO cells were transfected using LipofectAMINE2000 according to the manufacturer’s instructions. For antisense oligonucleotide transfections, passage 3 HUVEC at 70–80% confluence was transfected using Lipofectin according to the manufacturer’s instructions.
Migration assays
Single cell chemokinetic assay
To measure chemokinesis or single cell motility in the absence of a directional signal, the motility of cells to the bottom surface of a porous filter was quantitated in the absence of chemotactic or haptotactic stimuli. TranswellsTM (8 µm pore size; Corning Inc., Corning, NY, USA) were coated on both sides overnight with 12.5 µg/mL fibronectin and blocked with 5% BSA as described (43)
. Endothelial media (100 µL) was added to the top well and 500 µL to the bottom well. Endothelial cells were briefly trypsinized, washed twice in endothelial media, and 100 µL of a 106 cell/mL single cell suspension was added to the top well. Lung endothelial cell transmigrations were performed in luEnd media that contained 10% FBS. HUVEC transmigrations were performed in HUVEC media lacking ECGS and with FBS decreased to 2%, because transmigration in the presence of ECGS and higher concentrations of FBS were so high that transmigrated cells were difficult to count. After 2.5 h of incubation at 37°C, the cells were fixed in STF and stained with crystal violet. Cells on top of the filter were removed with a cotton swab and cells on the bottom surface were quantitated.
Wound healing assay
Falcon® Petri dishes (60 mm; Becton Dickinson) were coated with fibronectin as above. Butanol-extracted fibronectin and butanol-extracted BSA were used for coating in experiments involving S1P. Confluent PecamKO and PecamRC monolayers grown on fibronectin-coated plates in luEnd media and confluent HUVEC grown on gelatin-coated plates in HUVEC media were scraped with a 15-well minigel comb leaving concentric rings of cells, washed with PBS, and incubated 24 h. Distance migrated was quantitated by taking pictures at 0 and 24 h with a Nikon Coolpix995 Digital Camera (Nikon Corp., Tokyo, Japan) on an Olympus IM light microscope (Melville, NY, USA) and measuring distance of wound edge from a mark on the bottom of the plate.
luEnd and HUVEC media contained FBS (10% and 20%, respectively) and thus contained sphingolipids, including S1P (44)
; for S1P experiments, therefore, FBS was depleted of sphingolipids through charcoal stripping (CS-FBS). For S1P experiments, CS-FBS was prepared by incubating 50 mL FBS with 5 g activated charcoal (Sigma) overnight at 4°C (45)
. The CS-FBS was centrifuged (10 min at 1000 g) and filtered through a 0.22 µm sterile Nalgene® filter (Sybron Corp., Rochester, NY, USA).
To quantitate extension formation, 104 PecamRC or PecamKO cells/well in luEnd media were plated in a fibronectin-coated 24-well plate. Cells were fixed in STF and stained with crystal violet, photographed as above, and quantitated using the public domain NIH Image program (National Institutes of Health, Bethesda, MD, USA). The theoretical radius (r) had the cell been a perfect circle was determined according the formula rperimeter = perimeter/2
and rarea = 
(area/), and the ratio rperimeter to rarea was used to determine deviation from circularity.
Western blot and immunoprecipitation
Confluent monolayers of PecamRC and PecamKO cells cultured in luEnd media on fibronectin-coated plates were washed twice in ice-cold PBS containing 1 mM orthovanadate and scraped into lysis buffer (20 mM Tris pH 7.5, 100 mM NaCl, 10 mM EDTA, 1% Nonidet P-40, 1% deoxycholate, 0.1% SDS, 50 mM sodium fluoride, 1 mM sodium orthovanadate, and CPI). Lysates were centrifuged at 14,000 g at 4°C for 10 min and supernatant protein concentration was determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories, Hercules, CA, USA) as directed. For Western blots, 20 µg protein was boiled with 4x sample buffer (240 mM Tris pH 6.8, 40% glycerol, 8% SDS, 0.002% bromophenol blue, 0.002% ß-mercaptoethanol). For immunoprecipitation, 200 µg protein was precleared with protein A/G-Sepharose (Santa Cruz); the cleared supernatant was incubated with the appropriate antibody for 1 to 2 h turning end-over-end at 4°C, then precipitated with A/G-Sepharose. Beads were washed three times in lysis buffer and boiled in 4x sample buffer.
Samples were separated by SDS PAGE, transferred to an Immobilon P membrane (Millipore, Bedford, MA, USA), probed with anti-ERK 1:10,000, anti-phosphoERK 1:1000, anti-FAK 1:10,000 and anti-FAK PY397 1:2000, anti-p190 RhoGAP 1:5000, anti-p120 RasGAP 1:5000, or anti-phosphotyrosine 1:10,000 followed by horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit IgG (Promega, Madison, WI, USA), and developed using Chemiluminescent Reagent Plus (PerkinElmer Life Sciences, Boston, MA, USA). Blots were scanned on an Arcus II scanner (Agfa, Mortsel, Belgium) and quantitated using BioMax 1D software (Kodak).
RhoGTP and RacGTP pulldown assays
GST-Rho binding domain (BD) was expressed and purified using glutathione-agarose beads (Sigma-Aldrich) as described (39)
. GST-p21 activated kinase (PAK) Rac BD was expressed and purified as described (46)
. Freshly confluent PecamRC or PecamKO cells grown in luEnd media on fibronectin-coated 100 mm plates were lysed, 20 µL lysate set aside for normalization, and RhoGTP or RacGTP pulled down using glutathione/agarose-bound GST-Rho BD or GST-PAK Rac BD (39
, 46)
. Samples were separated by SDS PAGE, transferred to an Immobilon P membrane, probed with anti-Rho 1:100 or anti-Rac 1:100, and developed and quantitated as above.
Raft fractionation
Sucrose density gradient lipid raft fractionation was carried out in a detergent-free sodium bicarbonate buffer essentially as described in ref 47
. Confluent endothelial cells were scraped into lysis buffer (500 mM sodium bicarbonate) and sonicated. 85% sucrose in MBB (25 mM MES, pH 6.5, 150 mM NaCl, and 25 mM sodium bicarbonate) was added to bring the sample to 42.5% sucrose. The 2 mL sample was placed in an ultracentrifuge tube on ice for 4 h, layered with 8 mL of 30% sucrose in MBB, followed by 2 mL 5% sucrose in MBB layered to make a discontinuous gradient. The sample was centrifuged at 39,000 rpm for 18 h at 4°C in an SW41 rotor (Beckman Instruments, Palo Alto, CA, USA), at which time an opalescent band representing the caveolin-positive raft fraction was visible at the 30%/5% sucrose interface. Fractions of 1 mL were carefully pipetted; 20 µL samples were boiled in 4x sample buffer for analysis, separated by SDS-PAGE, and transferred to an Immobilon P filter. Blots were probed with anti-PECAM-1, anti-caveolin, or anti-ß actin at 1:10,000 or anti-Galphai2 or anti-Galpha13 at 1:250, developed, then quantitated as above.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 42)
. Lung endothelial cell culture media (luEnd media) contains 10% FBS, and all results presented were acquired in cells cultured in the presence of serum containing multiple growth factors and sphingolipids. PecamRC cells form cobblestone-like monolayers; PecamKO cells form similar tightly compact monolayers, but the cells are more spindle-shaped. Figure 1
a demonstrates this morphology at the migrating front of a wound healing endothelial monolayer. Phalloidin staining for filamentous actin (F-actin) revealed stress fibers throughout the cytoplasm of PecamRC cells compared with sparser cortical actin bands in PecamKO cells (Fig. 1b
). Immunostaining for phosphopaxillin (48)
, a component of focal contacts and adhesions, revealed typical elongated focal adhesions in PecamRC cells; in PecamKO cells, phosphopaxillin staining prominently highlighted the leading edges of lamellipodial extensions. Individually plated PecamKO cells form multiple extensions whereas PecamRC spread more evenly (Fig. 2
a). As a surrogate measure of extension formation or deviation from circularity, I derived theoretical radii (r) from the measured perimeter and the measured area, respectively, as if they had been perfect circles. For a perfect circle the ratio of rperimeter to rarea will be 1.0; any deviation from circularity will produce higher ratios. This number is dimensionless and therefore independent of any variation in size distribution between cell populations. The ratio of rperimeter to rarea is in fact significantly increased in PecamKO cells, confirming the apparent increase in extension formation seen when cells are individually plated (Fig. 2b
).
|
|
Two aspects of endothelial cell migration were assessed. Single cell motility in the absence of a chemotactic or haptotactic gradient (chemokinesis) was assessed by quantitating cells reaching the underside of a fibronectin-coated 8 micron pore TranswellTM membrane in 2.5 h. Coordinated migration of endothelial cells was assessed using a wound healing assay, measuring the distance migrated into a scraped wound over 24 h. Previous studies have reached conflicting conclusions about the role of PECAM-1 in endothelial migration. Antibody blocking studies have demonstrated decreases in wound healing migration and single cell motility in anti-PECAM-1-treated human umbilical vein endothelial cells (49)
. The direct effect of PECAM-1 expression on migration has been tested in nonendothelial cell lines with varying results, including decreased wound healing migration in epithelial cells (50)
and increases in both wound healing and single cell motility in mesothelial cells (49)
. In endothelial cells the role of PECAM-1 in the two types of migration are discordant: PecamKO cells were found to have significantly enhanced single cell motility (Fig. 3
a) but decreased wound healing migration (Fig. 3b
) with respect to PecamRC cells. To rule out the possibility that decreased wound healing in PecamKO cells is secondary to a proliferation rather than a migration defect, the wound healing assay was also carried out in the presence of mitomycin C to block proliferation (Fig. 3b
). Proliferation does contribute to wound healing migration, and overall distance migrated during mitomycin C treatment dropped by 
50% for both PecamRC and PecamKO endothelial cells. However, the differential in migration between the two cell types remained, implying that a proliferation defect does not account for the retardation of wound healing migration in PECAM-1-deficient endothelial cells.
|
The role of PECAM-1 in endothelial cell motility was confirmed in primary HUVEC using an antisense approach. Western blot analysis confirmed stable depletion of PECAM-1 expression over 48 h in HUVEC treated with antisense oligonucleotide compared with scrambled oligonucleotide control (Fig. 4
a). Single cell motility was significantly increased and wound healing migration decreased with antisense treatment (Fig. 4b
), confirming the findings in PecamKO and PecamRC endothelial cells. The effect of PECAM-1 on wound healing could require the ectodomain, which mediates homophilic cell–cell interactions and whose engagement modulates integrin activation (51
, 52)
and/or the cytoplasmic domain. The ITIM domain is dephosphorylated during wound healing migration (2)
but phosphorylated during flow-induced rearrangement (53)
, modulating binding and activation of SH2-containing proteins. PecamKO cells were therefore transiently transfected with control vectors expressing GFP or vascular cell adhesion molecule (VCAM); full-length PECAM-1; PECAM-1 lacking its ectodomain; or PECAM-1 mutated in its ITIM domain to eliminate phosphorylation (Fig. 4c
). Unlike GFP or VCAM, full-length PECAM-1 restored wound healing migration to PecamKO cells (Fig. 4d
). Neither the ectodomain nor the intact ITIM domain was required to confer the PECAM-mediated wound healing advantage, however, implying that PECAM-1 cytoplasmic domain signals to modulate migration independent of phosphorylation-mediated recruitment of SH2 domain-containing proteins.
PECAM-1 null endothelial cells are deficient in RhoGTP, and this deficiency accounts for their migratory phenotype
PecamKO cells show pronounced extension-formation and enhanced single cell motility but blunted wound healing migration. Lamellipodia are RacGTP-driven membrane extensions (18)
involved in cell motility; endothelial wound healing in contrast is RhoGTP mediated (15)
. The bivalent migratory phenotype of PECAM-1 null endothelial cells may thus reflect an imbalance of Rho and Rac activation. To assess levels of activated Rho and Rac, the G-proteins were pulled down with recombinantly expressed proteins that have a high affinity for their respective GTP-bound forms (Fig. 5
a, b). GTP-bound protein was quantitated and normalized to total expression to reveal a significant defect in RhoGTP loading in PecamKO compared with Pecam RC endothelial cells (Fig. 5
c); there was by comparison no significant difference in RacGTP loading. Thus, overall Rac activation is normal in PecamKO cells but RhoGTP levels are depressed, potentially lowering the threshold for Rac-mediated extension formation and motility while interfering with wound healing migration.
|
The GTPase-activating protein (GAP) p190RhoGAP inhibits Rho signaling upon integrin engagement in a src-dependent manner (3)
, allowing for extension formation and cell motility. Tyrosine phosphorylation of p190RhoGAP mediates p120RasGAP association, which is necessary for GAP activation (54)
. p120RasGAP was immunoprecipitated to determine levels and tyrosine phosphorylation of associated p190RhoGAP (Fig. 5d
). Quantitation revealed significantly increased association and tyrosine phosphorylation of p190RhoGAP with p120RasGAP (Fig. 5e
), consistent with p190RhoGAP activation and the decrease observed in RhoGTP in PecamKO cells. In the converse experiment, immunoprecipitation of p190RhoGAP also revealed increased p190 tyrosine phosphorylation and p120RasGAP association (data not shown).
Depletion of RhoGTP was noted in HUVEC treated with PECAM-1 antisense oligonucleotide compared with a scrambled oligonucleotide control (Fig. 5f, g
), confirming the applicability of the findings to primary as well as immortalized endothelial cells.
To determine whether in fact decreased Rho activity accounts for the migration characteristics of PecamKO endothelial cells, migrating cells were treated with both a direct inhibitor of Rho, the C3 exoenzyme (exoC3) of C. botulinum, and an inhibitor of the Rho effector p160ROCK, Y27632 (16)
. ExoC3 abrogated the wound healing advantage of PecamRC endothelial cells and significantly increased their single cell motility (Fig. 6
a, b). Y27632 had effects similar to those of ExoC3 in PecamKO and PecamRC cells (Fig. 69c, d). Extension formation was significantly increased, further mimicking the effect of PECAM-1 deficiency in endothelial cells (Fig. 6
e). Y27632 also significantly increased single cell motility and extension formation in bovine aortic endothelial cells, confirming the applicability of the findings to primary as well as immortalized endothelial cells (Fig. 6f, g
).
|
PECAM-1 is required for effective signaling and wound healing migration in response to sphingosine-1-phosphate
S1P, a serum sphingolipid, signals through heterotrimeric G-proteins associated with its seven-transmembrane EDG/S1P family receptors. S1P-mediated endothelial wound healing and chemotaxis is Rho dependent and pertussis toxin (PTX) inhibitable (Fig. 6)
; (4
, 5)
. PTX treatment abrogated the wound healing advantage of PecamRC cells; no significant difference in distance migrated remained between PTX-treated PecamRC and PecamKO endothelial cells (Fig. 7
a). PTX inhibits Galphai2-mediated activation of extracellular signal-regulated kinase (ERK), FAK, and src downstream of the EDG1/S1P receptor. Western blot with phosphospecific antibodies revealed increased activation of ERK and phosphorylation of FAK at tyrosine 397 (FAK PY397) in confluent PecamKO endothelial cells compared with PecamRC cells (Fig. 7b
).
|
ERK activation normally occurs only at the leading edge of migrating endothelial monolayers, where it mediates migration in response to wounding (35)
. Indeed, ERK phosphorylation increases as expected in wounded, migrating PecamRC monolayers (data not shown), whereas no further increase was observed in the already high levels of ERK activation in migrating PecamKO monolayers. Inappropriate activation of FAK and ERK in confluent PecamKO cells indicates dysfunctional signaling in the absence of an appropriate migratory stimulus, and this signaling proves ineffective in coordinating migration of PecamKO monolayers when that wounding stimulus is provided.
To assess S1P-mediated migration, sphingolipid-depleted media was prepared using FBS that had been charcoal-stripped to deplete it of sphingolipids (CS-FBS) without significantly effecting levels of nonlipid growth factors such as VEGF and FGF (21)
. Wound healing migration in CS-FBS was compared with migration in media supplemented with 100 nM S1P, about half the concentration of S1P in normal human plasma (44)
, and in sphingolipid-replete media (Fig. 7c
). PecamRC cells responded to S1P supplementation and to sphingolipid-replete media with significantly increased migration whereas PecamKO cells had no such increase; two-way ANOVA revealed a statistically significant cell type by treatment effect. There was a nonsignificant decrease in migration of PecamKO cells with S1P supplementation and a significant decrease in wound healing in sphingolipid-replete media. Thus, S1P-mediated signaling and migratory responses are severely dysregulated in PECAM-1-deficient endothelial cells.
PECAM-1 localizes to caveolin-positive lipid rafts
We used sucrose density gradient fractionation to isolate the low density, caveolin-positive lipid raft fraction of PecamRC and PecamKO endothelial cell monolayers (Fig. 8
a). Methyl ß-cyclodextrin (mßcd) disperses low density lipid rafts by depleting them of cholesterol. Treatment with mßcd eliminated the low density, caveolin-positive fraction, confirming that it consists of lipid rafts, including caveolae. PECAM-1 was found to colocalize with caveolin-1 to the lipid raft fraction (Fig. 8b
); mßcd treatment also disrupted the low density fractionation of PECAM-1.
|
PECAM-1 modulates the relative recruitment of Galphai2 and Galpha13 to rafts: PECAM-1 cofractionates with low density, caveolin-positive lipid rafts, and PECAM-1 expression is needed for an effective coordinated migration response to S1P. Galpha subunits localize to low density lipid rafts and move in and out of caveolae (55)
. Caveolin-1 interactions may modulate heterotrimeric G-protein signaling, directly by binding the inactive, GDP-bound version of Galpha heterotrimeric protein subunits and inhibiting GDP/GTP exchange (56)
or indirectly by transiently sequestering dissociated Galpha subunits, desensitizing signaling via heterologous seven-transmembrane receptors that share the same G-protein (57)
. Therefore, the relative recruitment of Galpha subunits to caveolin-positive rafts and their accessibility to either EDG family receptors or to inhibitory associations with caveolin-1 may have a direct bearing on Galpha subunit localization and degree of activation.
S1P-mediated migration has a pertussis-inhibitable, EDG-1/Galphai2-mediated component that results in activation of ERK signaling and FAK phosphorylation and an EDG-3-mediated component that couples to Galpha13, contributing to Rho activation. The ERK, FAK, and Rho signaling characteristics of PecamKO compared with those of PecamRC endothelial cells suggest a model in which the Galphai2-mediated signal is promiscuously activated, and the Galpha13 component is defective in the absence of PECAM-1. Given the potential role of raft localization in modulating heterotrimeric G-protein-mediated signaling, we assessed the relative targeting of Galphai2 vs. Galpha13 to rafts in PecamRC vs. PecamKO endothelial cells (Fig. 9
a). The results of these experiments were consistent with a role for PECAM-1 in supporting the targeting of Galpha subunits to lipid rafts: the ratio of Galphai2 to Galpha13 in the low density, caveolin-positive fraction was decreased by more than half in PecamKO compared with PecamRC endothelial cells (Fig. 9b
).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. An in vitro model of angiogenesis reveals a prominent defect in tube formation in immortalized lung endothelial cells derived from the PECAM-1 knockout mouse (PecamKO) compared with cells reconstituted with full-length PECAM-1 cDNA (PecamRC cells) (11)
. Embryonic endocardial cells derived from the cardiac cushion of the PECAM-1 knockout mouse are aberrantly resistant to glucose-mediated inhibition of VEGF-induced scattering motility (13)
. PECAM-1-deficient endothelial cells are defective in forming a stable new vasculature and, conversely, are resistant to inhibition of endocardial cell motility. Consistent with the dichotomous migratory behavior in vivo, we have found that PecamKO endothelial cells exhibit enhanced single cell single cell motility and lamellipodial extension formation at the expense of wound healing migration (Figs. 2
, 3)
. The contribution of PECAM-1 to this migratory phenotype was confirmed by depleting PECAM-1 in primary endothelial cells using antisense technology (Fig. 4a, b
). The PECAM-1 cytoplasmic domain was sufficient to restore wound healing migration independent of a functional ITIM domain (Fig. 4c, d
). The migratory differences are accounted for by a selective deficiency of activated Rho (RhoGTP) as opposed to RacGTP in PecamKO cells (Fig. 5a-c
), since inhibition of either Rho itself or the Rho effector p160ROCK reproduced the PECAM-1-deficient migratory phenotype (Fig. 6)
.
The decrease in RhoGTP loading could be at least partially attributed to increased activation of p190RhoGAP, a GTPase-activating protein, as marked by tyrosine phosphorylation and association of p190RhoGAP with p120RasGAP (Fig. 5c, d
). Figure 10
a delineates growth factor and integrin engagement-mediated lamellipodial extension and initiation of motility: some growth factors such as VEGF (58)
activate Rac, which itself inhibits Rho (59)
; other growth factors (54)
and integrins (3)
activate src family kinases that phosphorylate and activate p190RhoGAP, inhibiting Rho indirectly. In fact, overexpression of p190RhoGAP alone promotes Rac-mediated lamellipodial extension formation and enhances motility (3)
. The insensitivity of PECAM-1-deficient cardiac cushion endocardial cells to glucose-mediated suppression of VEGF-induced motility may thus be attributable to baseline low RhoGTP. Like primary PECAM-1-deficient cardiac cushion endocardial cells, PecamKO endothelial cells are resistant to glucose-mediated inhibition of single cell motility; pretreatment with an inhibitor of the Rho effector p160ROCK abrogates the inhibitory effect of glucose on PecamRC endothelial cell motility (data not shown).
|
S1P is a major endothelial chemotactic component of serum that engages EDG1/S1P1 and EDG3/S1P3 seven-transmembrane receptors to mediate endothelial migration and VEGF and bFGF-induced angiogenesis (20
, 45)
via downstream signaling through Galphai2 to FAK, ERK, and src (22
, 60)
and through Galpha13 to Rho (Fig. 10b
) (23
, 24)
. In addition to direct EDG/S1P-receptor-mediated signaling, S1P likely signals via transactivation of other known angiogenic growth factor receptors such as the VEGF receptor (61)
. The wound healing advantage of PecamRC cells is not only Rho dependent but PTX inhibitable (Fig. 7a
), a pattern characteristic of S1P-mediated endothelial migration. Although physiologic levels of S1P increase wound healing migration of PecamRC endothelial cells, as expected, PecamKO endothelial cells are not responsive to S1P; indeed, the wound healing advantage of PecamRC cells is much decreased in the presence of sphingolipid-depleted serum (Fig. 7
c). Elevated threonine/tyrosine phosphorylation of ERK and tyrosine phosphorylation of FAK in confluent PecamRC cells (Fig. 6b, c
) is consistent with dysregulated and ineffective Galphai2 signaling downstream of EDG1/S1P1 receptor. Src activity is likely increased in PecamKO cells: increased FAK at PY397 indicates that more src is likely bound and activated at focal contacts (62)
, and tyrosine phosphorylation of p190RhoGAP, which is also increased, is mediated by src (3)
.
Global activation of these pathways is insufficient to permit directed migration and wound healing; EDG1/S1P1 null fibroblasts, like PecamKO endothelial cells, display aberrant FAK and src activation but a decreased wound healing response to S1P, whereas local FAK and src activation at the leading edge promotes directed extension formation and migration of wild-type fibroblasts (22)
. The timing and localization of signaling are crucial to promoting coordinated, directional migration. Local suppression of RhoGTP signaling (63)
and activation of FAK, src, and ERK kinase activity (22
, 35)
promotes directed extension formation as a component of a coordinated endothelial migratory response to mechanical wounding. In PecamKO endothelial cells, it appears that global suppression of Rho signaling and activation of Galphai2-mediated signaling pathways promote the formation of multiple lamellipodial extensions and single cell motility at the expense of coordinated migration. Another immunoglobulin family molecule containing ITIM sequences in its cytoplasmic domains, SHPS-1, may play a role similar to that of PECAM-1 in fibroblasts: expression of SHPS-1 lacking its cytoplasmic domain in fibroblasts blocks RhoGTP activation and wound healing migration, increases extension formation, and increases ERK and FAK phosphorylation (64)
.
We propose a model of defective sphingosine-1-phosphate-mediated signaling and Rho activation in PECAM-1-deficient endothelial cells (Fig. 10b, c
). In PECAM-1-positive endothelial cells, S1P signals through Galphai2 and Galpha13 in a coordinated manner to produce directed extension formation and directed wound healing migration rather than single cell motility (Fig. 10b
). PECAM-1 modulates the selective recruitment of Galphai2 vs. Galpha13 to caveolin-positive, low density lipid raft fractions, where their activity might be modulated via association with EDG receptors and caveolin-1 (Fig. 9)
. In PECAM-1-deficient endothelial cells, EDG1/S1P1 and Galphai2 are activated in a noncoordinated manner, leading to baseline activation of ERK, FAK, and src (Fig. 10c
). Phosphorylation activates p190RhoGAP and suppresses RhoGTP; EDG3/S1P3-mediated signaling to Galpha13 is insufficient to activate Rho, favoring extension formation and single cell motility over directed wound healing migration. This lack of coordinated responsiveness to serum sphingolipid chemotactic signals may account for the poor angiogenesis into polyvinyl acetate sponge implants in the PECAM-1 knockout mouse in a model of the chronic phase of foreign body inflammation (11)
.
Indeed, S1P is far from alone in regulating endothelial migration through seven-transmembrane-linked heterotrimeric G-protein signaling. For example, CXCRs, seven-transmembrane receptors for cysteine-X-cysteine motif containing chemokines, including SDF-1 (CXCR4) (65)
and interleukin-8-related cytokines (CXCR1/2) (66)
, play an important role in angiogenesis in the settings of injury, inflammation, and neoplasia. The role of PECAM-1 in modulating S1P-mediated wound healing migration may extend to modulation of signaling via Galpha subunits downstream of multiple seven-transmembrane receptors. Preliminary studies indicate that both thrombin and histamine activate Rho to a significantly lower extent in PECAM-1 null as opposed to PECAM-1 reconstituted endothelial cells (data not shown).
Histamine receptor subtype H1, which promotes vascular permeability, is a seven-transmembrane receptor, although unlike the thrombin receptor it is ligand activated and signals primarily through Galphaq/11 (67)
. Histamine promotes a p160ROCK-mediated increase in endothelial permeability (68)
. The thrombin receptor belongs to the seven-transmembrane heterotrimeric G-protein linked family of protease activated receptors (PARs) and promotes endothelial cell migration and angiogenesis through pathways involving Rho signaling downstream of Galpha13 (69
, 70)
. Thrombin mediates endothelial cell contraction through RhoGTP-mediated activation of p160ROCK, myosin light chain kinase, and consequently actin/myosin-mediated contractility (71)
.
This may prove to be of significance in explaining the prolonged bleeding time phenotype of PECAM-1 null mice. Reciprocal bone marrow engraftment experiments between PECAM-1 null and wild-type mice have shown that the bleeding time is attributable to some deficiency of the PECAM-1 null vasculature rather than to a defect of bone marrow derived components such as platelets (72)
. To stimulate platelet plug formation, endothelial cells must contract to reveal the underlying thrombogenic collagen surface in response to platelet-released factors such as thrombin (73)
. Thus, one model to explain a prolonged bleeding time in PECAM-1 null mice despite the ability of PECAM-1 null platelets promote adequate hemostasis in the presence of a wild-type vasculature may be that PECAM-1 null endothelium fails to properly recruit platelet thrombi thanks to an impaired ability to contract in response to heterotrimeric G-protein-mediated signaling.
|
ACKNOWLEDGMENTS |
|---|
Received for publication November 13, 2002. Accepted for publication April 13, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  Y. Wu, K. Tworkoski, M. Michaud, and J. A. Madri Bone Marrow Monocyte PECAM-1 Deficiency Elicits Increased Osteoclastogenesis Resulting in Trabecular Bone Loss J. Immunol., March 1, 2009; 182(5): 2672 - 2679. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Garnacho, V. Shuvaev, A. Thomas, L. McKenna, J. Sun, M. Koval, S. Albelda, V. Muzykantov, and S. Muro RhoA activation and actin reorganization involved in endothelial CAM-mediated endocytosis of anti-PECAM carriers: critical role for tyrosine 686 in the cytoplasmic tail of PECAM-1 Blood, March 15, 2008; 111(6): 3024 - 3033. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. F. Vernon-Wilson, F. Aurade, L. Tian, I. C. M. Rowe, M. J. Shipston, J. Savill, and S. B. Brown CD31 delays phagocyte membrane repolarization to promote efficient binding of apoptotic cells J. Leukoc. Biol., November 1, 2007; 82(5): 1278 - 1288. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Kaneko-Tarui, L. Zhang, K. J Austin, L. E Henkes, J. Johnson, T. R Hansen, and J. K Pru Maternal and Embryonic Control of Uterine Sphingolipid-Metabolizing Enzymes During Murine Embryo Implantation Biol Reprod, October 1, 2007; 77(4): 658 - 665. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Wu, T. Welte, M. Michaud, and J. A. Madri PECAM-1: a multifaceted regulator of megakaryocytopoiesis Blood, August 1, 2007; 110(3): 851 - 859. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Kondo, E. A. Scheef, N. Sheibani, and C. M. Sorenson PECAM-1 isoform-specific regulation of kidney endothelial cell migration and capillary morphogenesis Am J Physiol Cell Physiol, June 1, 2007; 292(6): C2070 - C2083. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. J. Newman PECAMigration Blood, May 15, 2007; 109(10): 4113 - 4114. [Full Text] [PDF]
|
|
|
|
 |
|
 P. Viegas, N. Chaverot, H. Enslen, N. Perriere, P.-O. Couraud, and S. Cazaubon Junctional expression of the prion protein PrPC by brain endothelial cells: a role in trans-endothelial migration of human monocytes J. Cell Sci., November 15, 2006; 119(22): 4634 - 4643. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Mansouri, J. Douglas, P. P. Rose, K. Gouveia, G. Thomas, R. E. Means, A. V. Moses, and K. Fruh Kaposi sarcoma herpesvirus K5 removes CD31/PECAM from endothelial cells Blood, September 15, 2006; 108(6): 1932 - 1940. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Wang, Y. Gao, E. Scully, C. T. Davis, J. F. Anderson, T. Welte, M. Ledizet, R. Koski, J. A. Madri, A. Barrett, et al. {gamma}{delta} T Cells Facilitate Adaptive Immunity against West Nile Virus Infection in Mice J. Immunol., August 1, 2006; 177(3): 1825 - 1832. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Y. Kim, G. H. Liang, J. A. Kim, Y. J. Kim, S. Oh, and S. H. Suh Sphingosine-1-phosphate activates BKCa channels independently of G protein-coupled receptor in human endothelial cells Am J Physiol Cell Physiol, April 1, 2006; 290(4): C1000 - C1008. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. M. DeLisser, B. P. Helmke, G. Cao, P. M. Egan, D. Taichman, M. Fehrenbach, A. Zaman, Z. Cui, G. S. Mohan, H. S. Baldwin, et al. Loss of PECAM-1 Function Impairs Alveolarization J. Biol. Chem., March 31, 2006; 281(13): 8724 - 8731. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Cernuda-Morollon and A. J. Ridley Rho GTPases and Leukocyte Adhesion Receptor Expression and Function in Endothelial Cells Circ. Res., March 31, 2006; 98(6): 757 - 767. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C. Ford, J. P. Bertram, S. R. Hynes, M. Michaud, Q. Li, M. Young, S. S. Segal, J. A. Madri, and E. B. Lavik Tissue Engineering Special Feature: A macroporous hydrogel for the coculture of neural progenitor and endothelial cells to form functional vascular networks in vivo PNAS, February 21, 2006; 103(8): 2512 - 2517. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
I. Fleming, B. Fisslthaler, M. Dixit, and R. Busse Role of PECAM-1 in the shear-stress-induced activation of Akt and the endothelial nitric oxide synthase (eNOS) in endothelial cells J. Cell Sci., September 15, 2005; 118(18): 4103 - 4111. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Papanikolaou, A. Papafotika, C. Murphy, T. Papamarcaki, O. Tsolas, M. Drab, T. V. Kurzchalia, M. Kasper, and S. Christoforidis Cholesterol-dependent Lipid Assemblies Regulate the Activity of the Ecto-nucleotidase CD39 J. Biol. Chem., July 15, 2005; 280(28): 26406 - 26414. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Krump-Konvalinkova, S. Yasuda, T. Rubic, N. Makarova, J. Mages, W. Erl, C. Vosseler, C. J. Kirkpatrick, G. Tigyi, and W. Siess Stable Knock-Down of the Sphingosine 1-Phosphate Receptor S1P1 Influences Multiple Functions of Human Endothelial Cells Arterioscler. Thromb. Vasc. Biol., March 1, 2005; 25(3): 546 - 552. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Carrithers, S. Tandon, S. Canosa, M. Michaud, D. Graesser, and J. A. Madri Enhanced Susceptibility to Endotoxic Shock and Impaired STAT3 Signaling in CD31-Deficient Mice Am. J. Pathol., January 1, 2005; 166(1): 185 - 196. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. D. O'Brien, G. Cao, A. Makrigiannakis, and H. M. DeLisser Role of immunoreceptor tyrosine-based inhibitory motifs of PECAM-1 in PECAM-1-dependent cell migration Am J Physiol Cell Physiol, October 1, 2004; 287(4): C1103 - C1113. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
