Ligation of Endothelial αvβ3 Integrin Increases Capillary Hydraulic Conductivity of Rat Lung
Abstract Complement-mediated pulmonary edema results from increases in lung capillary hydraulic conductivity (Lp), possibly by receptor-mediated mechanisms. We considered the Lp effects of vitronectin and the vitronectin-containing complement complex SC5b-9, which ligate the integrin αvβ3. Vitronectin, SC5b-9, and SC5b-9–enriched zymosan-activated serum all rapidly increased Lp, as determined by the split-drop technique in single lung capillaries of rat lung. The Lp increases were inhibited by a monospecific (LM609) and a polyclonal (R838) antibody against the αvβ3 integrin but not by an irrelevant monoclonal antibody isotype matched with LM609, by a monoclonal antibody against the αvβ5 integrin, or by preimmune rabbit serum. Vitronectin monomers failed to increase Lp. The tyrosine kinase blockers genistein and methyl 2,5-dihydroxycinnamate caused significant concentration-dependent inhibitions of Lp increases due to vitronectin and zymosan-activated serum. By contrast, the protein kinase C blocker calphostin C had no major effect. We conclude that (1) multivalent ligation of the luminally located αvβ3 integrin of lung capillary endothelium increases transcapillary liquid flux, and (2) the dominant signal transduction pathway for this effect occurs through tyrosine kinase activation.
Complement activation is a potential cause of the lung microvascular injury that underlies pulmonary edema characteristic of conditions such as ARDS. Although complement-mediated microvascular injury has been attributed to several products formed in the complement pathway,1 specific receptor mechanisms that may lead to complement-induced pulmonary edema are not well understood. In general, the increase of microvascular permeability in complement-mediated pulmonary edema may result from the ligation of endothelial receptors that affect microvascular barrier properties. In lung microvessels, the quantification of the barrier by the single capillary method2 3 4 achieves accurate determinations of the capillary Lp. This provides a sensitive index for the identification of receptor-mediated barrier regulatory mechanisms.
In the present study, we consider a new class of endothelial receptors that may play a role in the pathophysiology of complement-mediated pulmonary edema, namely, the integrins of the cytoadhesin family. Of particular interest is the integrin αvβ3, a cytoadhesin that stabilizes endothelial monolayers in vitro5 and may, by extension, stabilize the endothelial barrier in microvessels. However, relevant microvascular data are lacking. Cytoadhesins, including the αvβ3 integrin, tend to be prominent at focal adhesions at which the extracellular domain of an integrin heterodimer binds ECM components and the cytoplasmic tail binds cytoskeletal elements. Because of these linkages, the functional importance of cytoadhesins is attributed mainly to the establishment of cell anchorage on the ECM.6 Recently, integrin αvβ3 has been shown to be essential for angiogenesis, and mAb LM609, which blocks the binding of ligands to αvβ3, promoted tumor regression of angiogenic blood vessels in chick embryos.7 8
In the cultured endothelial cell, the αvβ3 integrin is located on both the abluminal and luminal surfaces.9 Since a similar distribution of the integrin probably occurs in endothelial cells of intact microvessels, a function for the lumen-facing integrin needs to be defined. Although the ECM-facing αvβ3 integrin may subserve cell anchorage, the function of a luminal αvβ3 is likely to be different. One possibility is that ligation of the luminal αvβ3 integrin by blood-borne factors induces endothelial signaling that ultimately results in increases of microvascular permeability to water and solutes. In other cell types, integrins have been shown to be capable of inducing cell signaling, as indicated in the integrin-mediated increases of cellular Ca2+,10 tyrosine phosphorylation of cell proteins,11 induction of gene expression,12 and regulation of ion transporters.13 However, specific integrin-mediated signaling events that affect endothelial barrier properties remain poorly understood.
The link between complement activation and the endothelial αvβ3 integrin may be attributable to the formation of the complement complex SC5b-9 in the terminal part of the complement pathway. SC5b-9 is known to contain several molecules of the prototypic αvβ3 ligand vitronectin, which recognizes the integrin by means of its RGD (Arg-Gly-Asp) tripeptide sequence.14 In previous experiments, we showed that complement-activated serum increases lung microvascular Lp and that the critical factor responsible is SC5b-9.3 However, no direct data attest to the importance of the endothelial luminal αvβ3 integrin in the regulation of microvascular barrier properties. In the present study, we tested endothelial barrier effects of αvβ3 ligation through determinations of lung capillary Lp by using three products, namely, vitronectin, SC5b-9, and SC5b-9–enriched ZAS, as ligands. We report below that these products caused an αvβ3-mediated Lp increase through a signaling pathway involving tyrosine kinase activation. These data are the first direct evidence that link the endothelial luminal αvβ3 integrin of lung endothelium to barrier regulation in lung microvessels.
Materials and Methods
mAbs LM609 and P1F6, directed against the human αvβ3 and αvβ5 integrins, respectively, were purchased from Chemicon. One batch of mAb LM609 was the generous gift of Dr David A. Cheresh, Scripps Institute. Both LM609 and P1F6 are isotype matched (IgG1) and grown in mouse ascites fluid. Anti-laminin mAb LAM 89, raised in mouse ascites fluid (Chemicon), was used as an irrelevant isotype-matched (IgG1) control for LM609. Purified polyclonal antibody R838 was raised against purified αvβ3 integrin isolated from human endothelial cells as previously described.15 The ability of this antibody to cross-react with αvβ3 integrin from bovine and rat cells is well characterized.16 17
Plasma vitronectin was purified by heparin affinity chromatography by the method of Yatohgo et al,18 as previously described.19 20 We used human plasma depleted of vitamin K–dependent clotting factors (barium sulfate precipitation), fibrinogen (glycine precipitation), and fibronectin (gelatin affinity). In the purified sample, we confirmed the presence of vitronectin by Western blotting with anti-vitronectin antibody (not shown). The purified vitronectin was largely in the form of high-molecular-weight disulfide-bonded multimers (Fig 1⇓, top).
To cleave disulfide bonds, the purified vitronectin was treated with 10 mmol/L dithiothreitol for 3 hours and then alkylated with 60 mmol/L iodoacetamide for 1 hour at room temperature. The precipitate was dialyzed against PBS overnight, with three changes of PBS. This procedure yielded monomeric vitronectin (Fig 1⇑, top, lane D). This reduced and alkylated vitronectin was able to support adhesion and spreading of cells when used to coat tissue culture wells (not shown), suggesting that this form of vitronectin was still capable of recognizing cell surface integrin receptors.
Human plasma fibronectin was purified from a fibronectin and fibrinogen-rich byproduct of factor VIII production by previously described methods.23 The plasma fraction was dissolved in 0.05 mol/L Tris and 0.02 mol/L sodium citrate at pH 7.5. Fibrinogen was precipitated by heating at 56°C for 3 minutes. The solution was clarified by centrifugation and chromatographed on DEAE-cellulose (not shown). The fibronectin peak was pooled, and the protein was precipitated with ammonium sulfate, dialyzed against Tris-buffered saline, and frozen at −70°C until use. For Lp measurements, the fibronectin was used at 660 μg/mL in 40 mg/mL albumin.
We have shown previously that ZAS derived from rat or human serum gives qualitatively similar Lp responses3 ; hence, we used rat serum in the present study. As before,3 we activated serum complement by using zymosan (Sigma Chemical Co) according to the method of Gawryl et al.22 Briefly, we first boiled and cooled zymosan, and then we added it to cell-free serum (10 mg/mL) prepared from freshly collected anticoagulant-free blood under cold conditions (4°C). The serum was incubated (1 hour at 37°C) and then cleared of zymosan particles by centrifugation (20 000g for 10 minutes) and filtration (0.8-μm filter, Millipore Co). Control nonactivated serum was prepared in parallel without the addition of zymosan.
Aliquots of nonactivated sera and ZAS were microscopically examined to confirm the absence of cells and then analyzed for protein oncotic pressure, protein concentration, osmolality, and electrolyte (K+ and Ca2+) concentrations, as previously described.3 These variables, which could potentially affect the interpretation of Lp, were identical for ZAS and nonactivated serum. All sera aliquots were stored in plastic containers at −70°C. Samples were used within 4 weeks of preparation; each aliquot was thawed once.
SC5b-9 was purified by the method of Gawryl et al,22 with some modifications as described. SC5b-9 was precipitated from ZAS (9 mL) by the addition of ammonium sulfate to a 37.5% concentration while the mixture was stirred in an ice bath. The precipitate was recovered by centrifugation, dissolved in PBS (pH 7.4, 2 mL) containing 1% Triton X-100, and immediately chromatographed on a column of Sepharose CL-6B equilibrated with PBS. Fractions were electrophoresed on cellulose acetate and stained (with amido black) to identify proteins (not shown). Some samples were Western blotted with anti-vitronectin antibody (Chemicon) to confirm the presence of vitronectin in SC5b-9. The product migrated to the α1 region on cellulose acetate electrophoresis and contained the appropriate peptide subunits,22 as evident on SDS-PAGE (Fig 1⇑, bottom). As before,3 SC5b-9 levels in the fractions were quantified by enzyme-linked immunosorbent assay using anti–SC5b-9 mAb (Quidel). Lp data were obtained by using the fraction containing SC5b-9 at 119 μg/mL. An aliquot of this sample was immunodepleted, as described before,3 by using beads coated with the anti–SC5b-9 antibody (Quidel) (Fig 1⇑, bottom).
Human α-thrombin (a generous gift from Dr John W. Fenton II, New York State Department of Health, Albany) was dissolved in 0.75% NaCl, dialyzed against Ringer’s lactate, and diluted in 40 mg/mL albumin to obtain required concentrations. As a vehicle for purified substances and chemicals, we used delipidated BSA (fraction V, Sigma) in Ringer’s lactate containing (mmol/L) Na+ 144, K+ 4, Ca2+ 1.5, and lactate 28, at pH 7.4 and 300 mOsm (micro-osmometer, model 3M0, Advanced Instruments, Inc). At 40 mg/mL, albumin provided the requisite split-solution oncotic pressure (21 cm H2O) for Lp measurements.4
The peptides GRGDSP (Gly-Arg-Gly-Asp-Ser-Pro) and GRGESP (Gly-Arg-Gly-Glu-Ser-Pro) (Peninsula Laboratories, Inc) were used at 1 mg/mL in 40 mg/mL albumin. In some experiments, the RGD tripeptide (Sigma) was used instead of GRGDSP. Stock solutions of the tyrosine kinase inhibitors genistein (4′,5,7-trihydroxyisofravone, Sigma) and MDC (GIBCO BRL), a stable analogue of erbstatin,24 and of the PKC inhibitor calphostin C (Kamiya Biomedical Co) were prepared in dimethyl sulfoxide (Sigma). For experiments, aliquots from the stock solutions were diluted in 40 mg/mL albumin, as required.
Cross reactivity of mAbs LM609 and P1F6 with rat lung tissue was determined by immunoprecipitation. Through a pulmonary artery cannula, the vasculature of a rat lung was filled with a membrane-impermeant biotinylating reagent, sulfosuccinimidobiotin (Pierce), at 4°C in PBS, pH 7.4. The lung was stored at 4°C overnight, and then the vasculature was washed out with PBS to remove unbound reagent. The lung was homogenized in lysis buffer and centrifuged (14 000g for 5 minutes). The supernatant was incubated (22°C for 1 hour) with beads (protein A/G–agarose) coated with anti-integrin antibodies or IgG1 (control) and centrifuged (14 000g for 5 minutes). The pellet containing the beads was washed three times with PBS, then boiled (3 minutes) with reducing buffer, and recentrifuged. The supernatant was run on SDS-PAGE using polyacrylamide gels under reducing conditions, transferred to a nitrocellulose membrane, and treated with horseradish peroxidase–streptavidin (Jackson Immuno Research Lab, Inc). Blots were developed by enhanced chemiluminescence (Dupont NEN).
The lung preparation and split-drop methods have been reported several times3 4 and are described here briefly. Lungs removed from anesthetized (2% halothane inhalation followed by 40 mg/kg IP sodium pentobarbital) rats (500 to 600 g, Harlan Sprague-Dawley Inc, Indianapolis, Ind) were pump-perfused with autologous rat blood through cannulas inserted in the pulmonary artery and left atrium. A tracheal cannula was used for lung inflation with gas (30% O2/6% CO2/64% N2) that maintained blood Po2, Pco2, and pH at, respectively, 140 mm Hg, 35 mm Hg, and pH 7.3 (178 pH/blood gas analyzer, Corning). A heat exchanger (model 44TD, Yellow Springs) maintained perfusate temperature at 37°C. Recordings of lung vascular and airway pressures (model P23 ID pressure transducer, Gould Statham), referred to the micropuncture level, were displayed on a multichannel recorder (Gould RS 3400). Lung vascular pressures were varied by adjusting the height of the venous outflow. Airway pressure was held constant at 5 cm H2O during micropuncture but was cyclically varied to induce ventilation during nonmicropuncture intervals lasting 10 to 15 minutes.
The Split-Drop Procedure
Lungs were positioned on a vibration-free air table (Micro-G, Technical Manufacturing Corp) for microscopy and videorecording (×350; microscope, Olympus model SZH; camera, Panasonic WV-CD51; monitor, Panasonic WV-5410; and videorecorder, Panasonic AG-6050). To prevent drying, the experimental lung surface was layered with a drip of silicone oil (Dow Corning 200) warmed to 37°C (Sensortek Thermalert TH-5). Micropipettes with beveled tips (diameter, ≈5 μm) prepared as previously described3 were filled with either castor oil colored with carbon black B or the experimental solution. The split-drop procedure was carried out in venular capillaries (diameter, 20 to 25 μm), which were identified by their convergent flow patterns.
The micropuncture steps of our split-drop procedure have been previously described in detail.3 4 In brief, blood flow is stopped at a defined pulmonary artery pressure, which therefore equals Pc because of stop-flow conditions. By use of successive micropunctures, the experimental capillary is filled with an oil drop, which is then split by an injection of the experimental solution (split solution) to separate the split margins by ≈60 μm (split length). Further oil injections are instituted to seal the micropuncture sites. Finally, pipettes are removed, and split-length changes are videorecorded for 1 to 1.5 minutes. The oil-endothelial contact does not deteriorate the endothelial barrier within the <2-minute duration of the split-drop procedure,4 after which blood flow is reinstated.
Split length decreases with liquid exit from the split drop (filtration) but increases with liquid entry (absorption). Microvessel diameter remains constant. Jv at time zero is estimated according to our previously reported analysis.2 4 Briefly, split length and capillary diameter measurements determined by microcaliper (Mitsutoyo Corp) from a frame-by-frame video replay (×400; resolution, 1 μm) are applied to the cylinder formula to calculate split-drop volume and surface area. Jv is calculated by differentiating, at time zero, the slope of the exponential regression of split-drop volume versus time in the first minute. Lp is determined as the slope of the linear Jv-Pc relation, calculated by linear regression of three or more Jv measurements at different Pc values. Lp remains unchanged for up to 4 hours,4 within which all determinations are completed.
Capillary fluorescence of fluorescein isothiocyanate was determined by mercury lamp (Ushio) excitation through a 510-nm barrier filter. Fluorescence collected through a ×40 objective (Nikon Fluor LWD; Olympus microscope, Vanox), an image intensifier (Videoscope KS 1381), and a high-gain video camera (Dage CCD-72) was recorded by imaging software (Imaging Research MCID-M4) on a microcomputer platform.
Statistics and Units
Data are mean±SD; n indicates the number of experiments. Paired differences were tested by the paired t test, and differences among more than two groups were determined by ANOVA (Newman-Keuls test). Significance was accepted at P<.05. NS denotes nonsignificant differences from controls that are not shown.
Integrin Distribution in the Lung Capillary
Although mAbs LM609 and P1F6 are well characterized with regard to their ability to recognize several different types of tissue, their specific cross-reactivity with rat lung is not reported. The results of immunoprecipitation shown in Fig 2⇓ confirmed that both mAbs cross-react with rat lung (Fig 2A⇓). The experiment shown in Fig 2B⇓ confirmed the antigenic specificity of mAb LM609 for the αvβ3 integrin, in that LM609 failed to recognize proteins in rat lung homogenates previously immunodepleted with the αvβ3-recognizing polyclonal antibody R838. To determine the distribution of the endothelial luminal αvβ3 integrin, we intravitally imaged αvβ3 distribution in the lung venular capillary (Fig 3A⇓). Sequential capillary infusions at 4°C of the anti-αvβ3 mAb LM609 and fluorescent anti-mouse IgG resulted in significant fluorescence of the capillary wall (Fig 3A⇓, left), whereas no capillary fluorescence was evident when fluorescent IgG alone was infused (Fig 3A⇓, right). These findings, together with the immunohistochemical data shown in Fig 3B⇓, confirm the presence of the αvβ3 and αvβ5 integrins on the capillary luminal surface of rat lung.
Histology of the Oil-Filled Capillary Segment
To determine the possible presence of blood cells in the split drop, we microinjected oil drops into the venular capillaries of three lungs, fixed the lungs in formaldehyde, and then examined the histological sections of the capillaries by light microscopy (Fig 4⇓). Although the oil drop was always lost during sectioning, the oil-filled segment appeared as an empty space, forming well-defined margins with the adjoining plasma-containing segments. No blood cells were detected by toluidine or hematoxylin-eosin staining in 11 such segments.
The single experiment shown in Fig 5⇓ (top) exemplifies our method for quantifying Lp as the regression slope of the linear Jv-Pc relation. In this example, vitronectin at 100 μg/mL increased Lp from a control of 5.1×10−7 to 12.2×10−7 mL/(cm2 · s · cm H2O). The group data in Fig 5⇓ (middle) show that the Lp increase to vitronectin was concentration dependent and that the effect was inhibited completely in the presence of a pentapeptide containing the RGD sequence. In agreement with our previous findings,3 RGD itself had no effect on control Lp (n=3). In three experiments, different concentrations of the GRGES pentapeptide, which lacks the RGD sequence, failed to block the increase of Lp (Fig 5⇓, open squares in middle panel). To rule out possible platelet effects, in three experiments we gave 100 μg/mL vitronectin in platelet-rich plasma, obtained by centrifuging blood and collecting the supernatant above the buffy coat. As we previously reported, this species of rat has a high blood platelet count.26 The Lp determined with vitronectin in platelet-rich serum [11±0.1×10−7 mL/(cm2 · s · cm H2O)] was not different from the control response to an identical vitronectin concentration [12.5±0.7×10−7 mL/(cm2 · s · cm H2O), n=3], indicating that the presence of platelets did not enhance vitronectin’s Lp effect. We conclude that these Lp increases reflect vitronectin-induced increases of transendothelial liquid flux, attributable to an endothelial integrin that recognizes the RGD sequence of vitronectin.
To determine the onset of the vitronectin-induced flux increase, lumen-directed fluxes were generated at low capillary pressure (5 cm H2O).4 Within 5 seconds of injecting vitronectin, the lumen-directed flux was significantly higher than the control value (Fig 5⇑, bottom). This rapid response attributes the vitronectin-induced flux increase to ligation of a luminal rather than an abluminal vitronectin receptor, because transendothelial ligand transport in this time scale is highly unlikely under the present conditions. Transendothelial transport processes are convective, diffusive, or transcellular. The large size of the vitronectin multimer (>200 kD) and the presence of lumen-directed convection preclude the possibility that ligand was transported by either convection or diffusion in a time scale of seconds. Transcellular transport is also unlikely, because for a much smaller molecule, albumin, transcellular transport occurs in minutes27 ; hence, for these larger ligands, transcellular effects are expected with a delay of minutes rather than seconds.
The Lp increase to vitronectin at 200 μg/mL was inhibited 88% by coinjection of the anti-αvβ3 mAb LM609 and 91% by that of the polyclonal anti-αvβ3 antibody R838 (Table⇓). No inhibitory effects were evident when we coinjected the anti-αvβ5 mAb P1F6 (20 μg/mL; n=3; Lp, 12.6±1.3×10−7 mL/(cm2 · s · cm H2O). Since both mAbs are isotype matched and raised in mouse ascites fluid, P1F6 serves as a control for LM609, and the negative response with P1F6 indicates that the inhibition by LM609 was not due to contaminants or nonspecific factors. As a second control for the LM609 effect, we coinjected the anti-laminin mAb LAM89 (30 μg/mL) with vitronectin (n=3). This mAb, which is also isotype matched with LM609 though irrelevant for the αvβ3 integrin (see “Materials and Methods”), also failed to block the Lp increases. As control for the Lp-inhibiting effect of R838, in four experiments coinjection of preimmune rabbit serum with vitronectin failed to inhibit the Lp increase. An antibody-induced Lp effect was ruled out because none of the antibodies given alone in the present concentrations increased Lp. We conclude that the antibodies LM609 and R838 competitively inhibited the ligation of vitronectin to the αvβ3 integrin and thereby blocked the vitronectin-induced Lp increase.
To determine whether vitronectin needed to be multimeric to elicit changes in Lp, vitronectin was reduced and alkylated as previously described (“Materials and Methods”). Alkylation of vitronectin resulted in a marked decrease in the proportion of high-molecular-weight species in the preparation (Fig 1⇑, top). An aliquot of this preparation (300 μg/mL) gave an Lp of 5.5±0.5×10−7 mL/(cm2 · s · cm H2O) (n=3), which was not significantly different from baseline. The adhesive glycoprotein fibronectin, which ligates the α5β1 integrin,6 also failed to increase Lp above baseline at a concentration considerably greater (660 μg/mL) than the effective vitronectin concentrations (n=3).
ZAS and SC5b-9
The Table⇑ also shows that the Lp increases to ZAS and SC5b-9 (P<.01) were inhibited 60% and 93%, respectively, by LM609 and 75% and 94% by R838. Again, no inhibitory effects were evident when we coinjected ZAS and SC5b-9 with the anti-laminin mAb LAM89 (30 μg/mL) as a control for mAb LM609 (n=1 each) or with preimmune serum as a control for R838 (n=3 each). Coinfusion of the RGD tripeptide with SC5b-9 gave an Lp of 4.3±0.5×10−7 mL/(cm2 · s · cm H2O) (n=5), which was not significantly different from baseline. This confirms that similar to our previous finding with ZAS,3 the SC5b-9–induced Lp increase is inhibited by RGD. To verify that the Lp increases to the purified products were not due to contaminants, we confirmed that aliquots of vitronectin and SC5b-9, when immunodepleted by exposure to anti-vitronectin or anti–SC5b-9 mAb, failed to increase Lp (n=3 each).
To determine evidence for signaling mechanisms, 20-minute preinfusions of capillaries with genistein or MDC were each followed by determinations of Lp responses to a fixed concentration of ZAS or vitronectin. Control Lp values for the ZAS and vitronectin groups were obtained with control serum [3.3±0.2×10−7 mL/(cm2 · s · cm H2O), n=3] and 40 mg/mL albumin [5.4±0.2×10−7 mL/(cm2 · s · cm H2O), n=3], respectively. ZAS and vitronectin (100 μg/mL) increased Lp to 8.4±1.2×10−7 and 14.5±1.4×10−7 mL/(cm2 · s · cm H2O) (n=8), respectively. Both genistein and MDC inhibited these Lp increases in a concentration-dependent manner (Fig 6⇓, top and middle). Using exponential analysis for the genistein inhibition plot in Fig 6⇓, we estimated an IC50 (concentration at 50% inhibition) of 17 μmol/L for vitronectin and 24 μmol/L for ZAS. A linear analysis for the MDC inhibition plot (Fig 6⇓) gave an IC50 of 25 μmol/L for both ligands. These effects of the tyrosine kinase inhibitors suggest that the present Lp increases were attributable to tyrosine kinase activation.
Since tyrosine phosphorylation of phospholipase C-γ may induce a diacylglycerol-mediated PKC activation, we determined the effects of PKC inhibition by calphostin C on the vitronectin-induced Lp increase. For comparison we used α-thrombin, a known PKC activator.28 Baseline Lp obtained with the vehicle, 40 mg/mL albumin, averaged 5.2±0.2×10−7 mL/(cm2 · s · cm H2O) (n=4). After 15-minute preinfusions of vehicle in paired vessels, α-thrombin at 10−7 mol/L and vitronectin at 200 μg/mL increased Lp to 15.6±1.5×10−7 and 14.5±1.4×10−7 mL/(cm2 · s · cm H2O) (n=8), respectively. In a parallel protocol, preinfusions of calphostin C markedly inhibited the Lp increase to α-thrombin but not to vitronectin (Fig 6⇑, bottom). Thus, at 1 μmol/L calphostin C inhibited the α-thrombin response by >50% but the vitronectin response by <5% (P<.01). The inhibition disparity was also evident at 20 μmol/L, at which calphostin C inhibited the α-thrombin response completely but the vitronectin response by only 20%. We conclude that PKC activation did not play a major role in the present Lp increases.
We show for the first time that ligation of the αvβ3 integrin on the endothelial luminal surface increases transcapillary liquid conductance in the lung. Vitronectin, a classic αvβ3 ligand,14 SC5b-9, which also binds the αvβ3 integrin,29 and ZAS, which is essentially SC5b-9–enriched serum,3 each markedly increased lung capillary Lp. We demonstrated the presence of the αvβ3 integrin in the lung capillary by histochemistry. Moreover, we confirmed that the integrin exists on the luminal surface of the lung capillary endothelium by imaging capillary fluorescence attributable to the distribution of anti-αvβ3 mAb LM609 (Fig 3⇑). The inhibitions of the Lp increases by two different anti-αvβ3 antibodies, namely, mAb LM609 and the polyclonal antibody R838, confirm that the Lp effects of vitronectin, ZAS, and SC5b-9 resulted from αvβ3 ligation. Involvement of the vitronectin-recognizing αvβ5 integrin,25 the existence of which we demonstrate here in lung capillary endothelium (Fig 3⇑), may be ruled out because LM609, which does not recognize αvβ5,25 effectively blocked the entire Lp responses to vitronectin and SC5b-9. The inability of anti-αvβ5 mAb P1F6 to inhibit the vitronectin-induced Lp increase is also consistent with the noninvolvement of the αvβ5 integrin. However, the αvβ5-blocking effect of P1F6, which we used essentially as a control for LM609, remains unconfirmed in rat tissues.
The Lp increases, which always occurred within 1 minute of ligand injection, indicate that ligation of the endothelial αvβ3 integrin caused rapid increases of transcapillary liquid flux. In fact, when we timed the vitronectin response more accurately by inducing lumen-directed fluxes, we recorded flux increases within seconds (Fig 5⇑, bottom). For the reasons given in “Results,” we interpret that these rapid flux increases follow αvβ3 ligation on the endothelial luminal surface and that together with the Lp data, they signify a decrease of endothelial barrier function. By contrast, the αvβ3 integrin appears to be barrier protective when ligated on the abluminal side. Inhibition of integrin-ECM binding by RGD-containing peptides or by integrin-specific antisera causes monolayer disruption, as is evident in cultured endothelial monolayers by the appearance of holes and increased permeability to tracers such as horseradish peroxidase.5 Tang et al30 concluded that maintenance of αvβ3-containing focal adhesions protects against endothelial retraction in cultured monolayers. It is likely that as part of the focal adhesion, the abluminal αvβ3 integrin affects endothelial permeability to the extent that it serves a cell-stabilizing function. In the luminal location, the αvβ3 integrin evidently behaves more as a receptor that mediates opening of the interendothelial junction and, in this role, may be important in endothelial inflammatory responses.
Our histological evidence indicates that the oil injection in the split-drop technique effectively clears the lumen of blood cells that could potentially complicate the interpretation of Lp responses. Nevertheless, we considered the possibility that the split drop contained resident endothelium-attached platelets that, if activated, could indirectly affect Lp. We believe that this is unlikely for the following reasons: (1) We have shown previously that the split-drop technique does not injure the endothelium4 ; hence, it is unlikely that platelets adhered to the normally nonthrombogenic endothelial surface. (2) Vitronectin in platelet-rich serum did not augment its platelet-free effect on Lp; hence, platelets did not contribute to the vitronectin effect. (3) We have not detected endothelial attachment of platelets or other blood cells in the oil-filled capillary segment by conventional light microscopy.
The fact that the Lp increases following αvβ3 ligation were inhibited by tyrosine kinase inhibitors provides the first evidence that signaling pathways involving tyrosine kinase activation are important in microvascular barrier regulation. The tyrosine kinase inhibitors genistein and MDC caused concentration-dependent inhibitions of the Lp increases to ZAS and vitronectin. These inhibitors have different modes of action: genistein inhibits ATP binding to tyrosine kinase,31 whereas MDC, a synthetic erbstatin analogue, inhibits by competing for the kinase substrate.24 The specificity of genistein and MDC as tyrosine kinase inhibitors has been obtained from assays of kinase activity and of kinase-dependent effects. These assays indicate that genistein inhibits tyrosine kinase at an IC50 of <50 μmol/L32 33 34 35 but serine/threonine kinases such as PKC at an IC50 of >185 μmol/L.31 The IC50 of MDC for tyrosine kinase, determined from the inhibition of tyrosine kinase–dependent effects in different cell types, is <70 μmol/L.36 37 38 Our Lp inhibition data for ZAS and vitronectin gave IC50 values for genistein and MDC that are consistent with reported data and lie in the range of tyrosine kinase but not PKC inhibition. Hence, the present Lp inhibitions by these agents were attributable to tyrosine kinase inhibition. PKC activation, if present, played a minor role in these Lp increases, as is evident in the weak inhibition of the vitronectin effect by calphostin C (Fig 5⇑). These results support tyrosine phosphorylation events as the principal postligational effect that led to the Lp increases and are consistent with our recent findings that vitronectin and SC5b-9–induced αvβ3 ligation causes protein tyrosine phosphorylation in endothelial cells.39
Since integrins are not known to possess intrinsic enzymatic activity, tyrosine phosphorylation of critical proteins is the important intermediary step that initiates signaling.6 Ligation of several integrins of the β1 family (eg, α3β1 in human carcinoma cells and α5β1 in fibroblasts) phosphorylates cytoplasmic proteins on tyrosine.11 40 In platelets, α-thrombin–induced activation of the αIIbβ3, an integrin similar to αvβ3, causes tyrosine phosphorylation of several proteins.41 αvβ3 ligation in fibroblasts mediates phosphorylation of paxillin and the kinase pp125 FAK.11 The integrin-cytoskeleton link also suggests that integrin ligation may cause tyrosine phosphorylation–induced cytoskeletal reorganization and cell-shape change,42 which in endothelium could result in the Lp increases seen here.
An important result was obtained when we used a predominantly monomeric form of vitronectin obtained by the reduction and alkylation of vitronectin multimers (see “Materials and Methods”). Although the postalkylation product was capable of recognizing integrins to the extent that it supported cell spreading and adhesion, it did not increase Lp above the control value. This result indicates that in its monomeric form, vitronectin has no detectable effect on the lung endothelial barrier. Vitronectin is monomeric in the native state, but several vitronectin molecules are incorporated into SC5b-9 following complement activation.43 The juxtaposition of multiple vitronectin molecules, as in SC5b-9 or a vitronectin multimer, is likely to increase the density of RGD sites, which may lead to ligation of multiple αvβ3 receptors, receptor aggregation, and the induction of signal transduction mechanisms.
Although we did not obtain direct evidence for receptor clustering in the lung capillary, integrins are known to cluster when ligated.40 44 In the present experiments, aggregation of the αvβ3 integrin caused by the vitronectin multimer may have induced the critical signaling that finally led to endothelial retraction45 and increased Lp. The binding studies of Zanetti et al46 and our recent findings on protein tyrosine phosphorylation,39 taken together, indicate that multivalent ligation of the αvβ3 receptor, presumably followed by receptor aggregation,44 is the prerequisite for αvβ3-induced intracellular signaling. We showed that the binding of the receptor by vitronectin multimers but not by small molecules such as the RGD peptide or by monomeric vitronectin enhanced tyrosine phosphorylation of endothelial protein substrates.39 These findings are consistent with the present negative Lp effect of monomers obtained by alkylating the vitronectin multimer or of the RGD peptide given alone.3 However, the RGD inhibition of the Lp response to αvβ3 ligation proves that RGD binding to the receptor occurred. This binding, though ineffective for Lp responses, was nevertheless sufficient to inhibit the effects of multivalent ligands, presumably because the ligands could no longer bind and aggregate the receptor. Interestingly, fibronectin, which also binds the αvβ3 integrin by an RGD mechanism, did not increase Lp, presumably because fibronectin does not form multimers and probably does not cluster the receptor. The biological potency of the multivalent form of vitronectin is further indicated in that although both the multimer and the native monomer of vitronectin bind fibroblast monolayers, only the multimer is internalized by αvβ5-induced endocytosis.20 These considerations indicate that because of its ability to form polymers, vitronectin’s effectiveness as a ligand may be conformer sensitive. Although <5% of total plasma vitronectin normally occurs in the multivalent form,14 during inflammatory conditions the increased formation of multivalent vitronectin, as in SC5b-9 or perhaps as resulting from the interaction of vitronectin with bacteria,14 may prove to be a significant factor in the increase of lung microvascular permeability.
In conclusion, our findings provide the first evidence (1) that ligation of luminal αvβ3 of the lung capillary by multimeric vitronectin and by a vitronectin-containing complement complex, SC5b-9, increases lung endothelial liquid conductance and (2) that tyrosine kinase activation by this integrin plays a role in the signaling cascade that leads to increases of Lp. In the present study, we demonstrate that in its luminal location, the endothelial αvβ3 integrin is capable of exerting a significant effect on the lung endothelial barrier; hence, the barrier regulatory function of this luminal integrin needs to be addressed separate from its ECM-linked biological functions such as blood vessel growth.7 8 Since blood levels of vitronectin-containing complement complexes increase in ARDS,47 the ligation of the endothelial luminal αvβ3 integrin by these complexes presents a new mechanism that may decrease lung microvascular barrier function and promote pulmonary edema.
Selected Abbreviations and Acronyms
|ARDS||=||adult respiratory distress syndrome|
|Jv||=||transendothelial flux (liquid flux per unit surface area)|
|PKC||=||protein kinase C|
This study was supported by grant HL-36024 from the National Institutes of Health, Bethesda, Md (Dr Bhattacharya), and by grants CA-58626 from NIH and AHA 93013270 from the American Heart Association (Dr McKeown-Longo). We thank Rashmi Patel for technical assistance.
- Received May 4, 1995.
- Accepted July 13, 1995.
- © 1995 American Heart Association, Inc.
Warren JS, Johnson KJ, Ward PA. Immunoglobulin- and complement-mediated immune injury. In: Crystal RG, West JB, eds. The Lung: Scientific Foundations. New York, NY: Raven Press Publishers; 1991:1939-1946.
Bhattacharya J. Hydraulic conductivity of lung venules determined by split-drop technique. J Appl Physiol. 1988;64:2562-2567.
Ishikawa S, Tsukada H, Bhattacharya J. Soluble complex of complement increases hydraulic conductivity in single microvessels of rat lung. J Clin Invest. 1993;91:103-109.
Qiao R-L, Bhattacharya J. Segmental barrier properties of the pulmonary microvascular bed. J Appl Physiol. 1991;71:2152-2159.
Lampugnani MG, Resnati M, Dejana E, Marchisio PC. The role of integrins in maintenance of endothelial monolayer integrity. J Cell Biol. 1991;112:479-490.
Brooks PC, Clark RA, Cheresh DA. Requirement of integrin αvβ3 for angiogenesis. Science. 1994;264:569-571.
Conforti G, Dominguez-Jiminez C, Zanetti A, Gimbrone MA Jr, Cremona O, Marchisio PC, Dejana E. Human endothelial cells express integrin receptors on the luminal aspect of their membrane. Blood. 1992;80:437-446.
Schwartz MA. Spreading of human endothelial cells on fibronectin or vitronectin triggers elevation of intracellular free calcium. J Cell Biol. 1993;120:1003-1010.
Burridge K, Turner CE, Romer LH. Tyrosine phosphorylation of paxillin and pp125FAK accompanies cell adhesion to extracellular matrix: a role in cytoskeletal assembly. Cell Biol. 1992;119:893-903.
Juliano RL, Haskill S. Signal transduction from the extracellular matrix. J Cell Biol. 1993;120:577-585.
Schwartz MA, Lechene C, Ingber D. Insoluble fibronectin activates the Na/H antiporter by clustering and immobilizing integrin αvβ1, independent of cell shape. Proc Natl Acad Sci U S A. 1991;88:7849-7853.
Albelda SM, Daise M, Levine E, Buck CA. Identification of and characterization of cell-substratum adhesion receptors on human large vessel endothelial cells. J Clin Invest. 1989;83:1992-2002.
Basson CT, Knowkes WJ, Bell L, Albelda SM, Castronovo V, Liotta LA, Madri JA. Spatiotemporal segregation of endothelial cell integrin and nonintegrin extracellular matrix-binding proteins during adhesion events. J Cell Biol. 1990;110:789-801.
Panetti TS, McKeown-Longo PJ. The αvβ5 integrin receptor regulates receptor-mediated endocytosis of vitronectin. J Biol Chem. 1993;268:11492-11495.
Panetti TS, McKeown-Longo PJ. Receptor-mediated endocytosis of vitronectin is regulated by its conformational state. J Biol Chem. 1993;268:11988-11993.
Tomasini BR, Mosher DF. Vitronectin. Prog Hemost Thromb. 1990;10:269-305.
Mosher DF, Johnson RB. In vitro formation of disulfide-bonded fibronectin multimers. J Biol Chem. 1983;258:6595-6601.
Issiki K, Imoto M, Sawa T, Umezawa K, Takeychi T, Umezawa H, Tsuchida T, Yoshioka T, Tatsuta K. Inhibition of tyrosine protein kinase by synthetic erbstatin analogs. J Antibiot (Tokyo). 1987;40:1208-1210.
Wayner EA, Orlando RA, Cheresh DA. Integrins αvβ3 and αvβ5 contribute to cell attachment to vitronectin but differentially distribute on the cell surface. J Cell Biol. 1991;113:919-929.
Yang G, Sadurski R, Bhattacharya J. Hemodynamic effects of brief pulmonary hemostasis. J Appl Physiol. 1993;75:948-954.
Ghitescu L, Bendayan M. Transendothelial transport of serum albumin: a quantitative immunocytochemical study. J Cell Biol. 1992;117:745-775.
Lynch JJ, Ferro TJ, Blumenstock FA, Brockenauer AM, Malik AB. Increased endothelial albumin permeability mediated by protein kinase C activation. J Clin Invest. 1990;85:1991-1998.
Biesecker G. The complement SC5b-9 complex mediates cell adhesion through a vitronectin receptor. J Immunol. 1990;145:209-214.
Tang DG, Chen YQ, Diglio CA, Honn KV. Protein kinase C-dependent 12(S)-HETE on endothelial cell vitronectin receptor and fibronectin receptor. J Cell Biol. 1993;121:689-704.
Akiyama T, Ishida J, Nakagawa S, Ogawara H, Watanabe S, Itoh N, Shibuya M, Fukami Y. Genistein, a specific inhibitor of tyrosine-specific protein kinases. J Biol Chem. 1987;262:5592-5595.
Honda A, Noguchi N, Takehara H, Ohashi Y, Asuwa N, Mori Y. Cooperative enhancement of hyaluronic acid synthesis by combined use of IGF-1 and EGF, and inhibition by tyrosine kinase inhibitor genistein, in cultured mesothelial cells from rabbit pericardial cavity. J Cell Sci. 1991;98:91-98.
Mustelin T, Coggeshall KM, Isakov N, Altman A. T cell antigen receptor-mediated activation of phospholipase C requires tyrosine phosphorylation. Science. 1990;247:1584-1587.
McColl SR, Dipersio JF, Caon AC, Ho P, Naccache H. Involvement of tyrosine kinases in the activation of human peripheral blood neutrophils by granulocyte-macrophage colony-stimulating factor. Blood. 1991;78:1842-1852.
Salari H, Duronio V, Howard SL, Demos M, Jones K, Reany A, Hudson AT, Pelech SL. Erbstatin blocks platelet activating factor-induced protein-tyrosine phosphorylation, polyphosphoinositide hydrolysis, protein kinase C activation, serotonin secretion and aggregation of rabbit platelets. FEBS Lett. 1990;263:104-108.
Uings IJ, Thompson NT, Randall RW, Spacey GD, Bonser RW, Hudson AT, Garland LG. Tyrosine phosphorylation is involved in receptor coupling to phospholipase D but not phospholipase C in the human neutrophil. Biochem J. 1992;281:597-600.
Bhattacharya S, Fu C, Bhattacharya J, Greenberg S. Tyrosine phosphorylation of endothelial proteins augmented by αvβ3 ligation. J Biol Chem. 1995;270:16781-16787.
Kornberg LJ, Earp HS, Turner CE, Prockop C, Juliano RL. Signal transduction by integrins: increased protein tyrosine phosphorylation caused by clustering of b1 integrins. Proc Natl Acad Sci U S A. 1991;88:8392-8396.
Golden A, Brugge JS, Shattil SJ. Role of platelet membrane glycoprotein IIb-IIIa in agonist-induced tyrosine phosphorylation of platelet proteins. J Cell Biol. 1990;111:3117-3127.
Haimovich B, Lipfert L, Brugge J, Shattil SJ. Tyrosine phosphorylation and cytoskeletal reorganization in platelets are triggered by interaction of integrin receptors with their immobilized ligands. J Biol Chem. 1993;268:15868-15877.
Miyamoto S, Akiyama SK, Yamada KM. Synergistic roles for receptor occupancy and aggregation in integrin transmembrane function. Science. 1995;267:883-885.
Wysolmerski RB, Lagunoff D. Regulation of permeabilized endothelial cell retraction. Am J Physiol. 1991;261:C32-C40.
Zanetti A, Conforti G, Hess S, Martin-Padura I, Ghibaudi E, Preissner KT, Dejana E. Clustering of vitronectin and RGD peptides on microspheres leads to engagement of integrins on the luminal aspect of endothelial cell membrane. Blood. 1994;8:1116-1123.