Focal adhesion kinase mediates porcine venular hyperpermeability elicited by vascular endothelial growth factor


  • Mack H. Wu,

    Corresponding author
    1. Cardiovascular Research Institute and Departments of Medical Physiology and Surgery, College of Medicine, Texas A&M University System Health Science Center, 702 Southwest HK Dodgen Loop, Temple, TX 76504, USA
    • Corresponding author
      M. Wu: Department of Medical Physiology, Texas A&M University Health Science Center, 702 Southwest HK Dodgen Loop, Temple, TX 76504, USA. Email:

    Search for more papers by this author
  • Mingzhang Guo,

    1. Cardiovascular Research Institute and Departments of Medical Physiology and Surgery, College of Medicine, Texas A&M University System Health Science Center, 702 Southwest HK Dodgen Loop, Temple, TX 76504, USA
    Search for more papers by this author
  • Sarah Y. Yuan,

    1. Cardiovascular Research Institute and Departments of Medical Physiology and Surgery, College of Medicine, Texas A&M University System Health Science Center, 702 Southwest HK Dodgen Loop, Temple, TX 76504, USA
    Search for more papers by this author
  • Harris J. Granger

    1. Cardiovascular Research Institute and Departments of Medical Physiology and Surgery, College of Medicine, Texas A&M University System Health Science Center, 702 Southwest HK Dodgen Loop, Temple, TX 76504, USA
    Search for more papers by this author


Focal adhesion kinase (FAK) is known to mediate endothelial cell adhesion and migration in response to vascular endothelial growth factor (VEGF). The aim of this study was to explore a potential role for FAK in VEGF regulation of microvascular endothelial barrier function. The apparent permeability coefficient of albumin (Pa) was measured in intact isolated porcine coronary venules. Treating the vessels with VEGF induced a time- and concentration-dependent increase in Pa. Inhibition of FAK through direct delivery of FAK-related non-kinase (FRNK) into venular endothelium did not alter basal barrier function but significantly attenuated VEGF-elicited hyperpermeability. Furthermore, cultured human umbilical vein endothelial monolayers displayed a similar hyperpermeability response to VEGF which was greatly attenuated by FRNK. Western blot analysis showed that VEGF promoted FAK phosphorylation in a time course correlating with that of venular hyperpermeability. The phosphorylation response was blocked by FRNK treatment. In addition, VEGF stimulation caused a significant morphological change of FAK from a punctate pattern to an elongated, dash-like staining that aligned with the longitudinal axis of the cells. Taken together, the results suggest that FAK contributes to VEGF-elicited vascular hyperpermeability. Phosphorylation of FAK may play an important role in the signal transduction of vascular barrier response to VEGF.

The wall of capillary and postcapillary venules consists of a layer of endothelial cells that are tethered to extracellular matrix (ECM). The attachment of the endothelial lining to ECM is mediated by focal adhesions composed of transmembrane receptors defined as integrins and multiple intracellular proteins that link integrins to the cytoskeleton (Aplin et al. 1998; Geiger et al. 1998). Dynamic interactions between the integrins, linkage proteins and cytoskeletal elements control the assembly and distribution of focal adhesions and thus cell morphology and motility (Parsons et al. 2000). During this process, integrins function not merely as adhesion receptors, but transmit chemical signals and mechanical forces between the matrix and cytoskeleton (Aplin et al. 1998; Geiger et al. 1998). Evidence is emerging that integrin-mediated focal adhesions also participate in the endothelial contractile response to shear stress, growth factors and inflammatory mediators (Girard & Nerem, 1995; Soldi et al. 1996; Abedi & Zachary, 1997; Takahashi et al. 1999; Garcia et al. 2000). We have previously reported that blocking integrin binding to fibronectin and/or vitronectin induces a dramatic increase in the permeability of venules, suggesting that the adhesive interaction between endothelial cells and ECM plays an essential role in the maintenance of microvascular barrier integrity (Wu et al. 2001).

Because integrins lack catalytic activity, the physical forces or chemical signals are transduced via a network of integrin-associated proteins (Aplin et al. 1998; Geiger et al. 1998). Within this context, the focal adhesion complex contains a host of signalling molecules, of which focal adhesion kinase (FAK) is the major kinase capable of catalysing various downstream signalling reactions leading to integrin engagement and focal adhesion assembly (Schlaepfer et al. 1999; Schaller et al. 2000; Schaller, 2001). The activity of FAK is mainly regulated through phosphorylation. Inhibition of FAK tyrosine phosphorylation prevents, whereas tyrosine phosphatase inhibitors promote, focal adhesion formation and associated cellular responses (Schlaepfer et al. 1999; Schaller, 2001). Overexpression of dominant negative FAK in endothelial cells inhibits FAK phosphorylation-induced cell contraction (Schaller et al. 2000). Activation of protein tyrosine phosphorylation with tyrosine phosphatase inhibitors causes an increase in transendothelial permeability coupled with focal adhesion tyrosine phosphorylation (Garcia et al. 2000). In human pulmonary arterial endothelial cells, actin-guided FAK translocation to focal adhesions modulates the changes in transendothelial electrical resistance in the presence of inflammatory mediators (Mehta et al. 2002). In agreement with these reports, our previous experiments (Yuan et al. 1998) have revealed an association between FAK tyrosine phosphorylation and microvascular hyperpermeability.

Although the critical role of FAK-signalled focal adhesion formation in angiogenesis has been well recognized, it is not clear whether FAK serves as a signalling molecule in the mediation of VEGF-elicited microvascular leakage, an initial reaction of the angiogenic response to the growth factor. Therefore, the aim of this study was to evaluate the signalling effect of FAK on microvascular barrier function during stimulation by VEGF. To achieve this objective, we utilized a recently developed protein transfer technique (Tinsley et al. 1998, 2001) to introduce FAK-related non-kinase (FRNK) (Schaller et al. 2000) directly into the endothelium of coronary venules and human umbilical vein as a means of blocking the participation of FAK in VEGF-induced signalling.



An albumin physiological salt solution (APSS) was used as a bathing solution while the microvessels were being dissected. It contained the following (mm): NaCl 145.0, KCl 4.7, CaCl2 2.0, MgSO4 1.17, NaH2PO4 1.2, glucose 5.0, pyruvate 2.0, EDTA 0.02 and 3-N-morpholino propanesulfonic acid buffer 3.0. After adding 1 % bovine serum albumin, the solution was buffered to a pH of 7.40 at 4°C and then filtered through a Millex-PF 0.22 μm filter unit (Millipore, Bedford, MA, USA). The APSS used to perfuse the vessels during permeability measurements had the same composition as mentioned above, but was buffered to a pH of 7.40 at 37°C. The chemicals used to make the perfusate, including fluorescein isothiocyanate (FITC)-albumin, were purchased from Sigma (St Louis, MO, USA). Bovine serum albumin was obtained from United States Biochemical (Cleveland, OH, USA). Cell culture supplies including the culture media Dulbecco's modified Eagle's medium (DMEM) and fetal bovine serum were from Gibco (Gaithersburg, MD, USA). VEGF was from R&D System (Minneapolis, MN, USA). Antibodies directed against phosphotyrosine (PY20) and anti-FAK were from Transduction Laboratory (Lexington, KY, USA) and anti-IgG conjugated with horseradish peroxidase was from Jackson Immuno Research Laboratory Inc. (West Grove, PA, USA).

Expression and purification of FRNK

Escherichia coli transformed with pET-histidine-tagged FRNK was a generous gift from Dr J. T. Parsons (University of Virginia). The bacteria (250 ml) were cultured in 0.3 mm isopropyl-1-thio-d-galactopyranoside (Amersham Pharmacia Biotech, Piscataway, NJ, USA) for 3 h, the culture was centrifuged and the pellet was frozen at −80°C overnight. The sample was lysed in B-PER (Pierce, Rockford, IL, USA) containing 300 mm NaCl and 1 μl ml−1 protease inhibitor mixture (Calbiochem, La Jolla, CA, USA) and then cleared by centrifugation at 27 000 g. Protein purification was conducted using a Ni-NTA Spin Columns Kit (Qiagen, Valencia, CA, USA). FRNK was eluted from each column using a buffer containing 50 mm NaH2PO4, 300 mm NaCl and 20 mm imidazole at pH 8 and dialysed in a pre-washed dialysis tube (Gibco BRL, Gaithersburg, MD, USA) for 48 h. Following SDS-polyacrylamide gel analysis, the protein was transferred to a polyvinylidene difluoride membrane. The product was confirmed by protein sequencing based on the N-terminal sequence of FRNK (Protein Sequencing Facility, Texas A&M University).

Animal preparation

Yorkshire pigs weighing 9-13 kg were anaesthetized with sodium pentobarbital (30 mg kg−1, i.v.) and heparinized (250 units kg−1, i.v.). Following a tracheotomy and intubation, the animal was ventilated with room air. A left thoracotomy was performed and the heart was electrically fibrillated, excised and placed in 4°C physiological saline. The pigs were killed through surgical removal of the heart. The animals were housed and handled in accordance with the protocols approved by the Institutional Animal Care and Use Committee of the Texas A&M University and Scott and White Hospital.

Isolation and perfusion of coronary venules

The coronary sinus was cannulated and 5 ml India ink- gelatin- physiological salt solution was infused to clearly define venular microvessels. This solution was prepared by adding 0.2 ml of India ink (Koh-I-Noor, Bloomsbury, NJ, USA) and 0.35 g of porcine skin gelatin to 10 ml of warm physiological salt solution and filtered through P8 filter paper (Fisher Scientific, Pittsburgh, PA, USA). Information regarding the validation and limitation of the ink-perfusion procedure has been provided in our previous publications (Yuan et al. 1993a,b). The method for isolating and cannulating coronary venules is also described in detail in these studies. Briefly, in a dissecting chamber containing APSS at 4°C, a suitable venule (length 0.8-1.2 mm, diameter 20-60 μm) was dissected from the surrounding myocardium with the aid of a Zeiss stereo dissecting microscope. The vessel was transferred to a cannulating chamber which was mounted on a Zeiss axiovert microscope. The isolated vessel was cannulated with an inflow and outflow micropipette (tip diameter 30-40 μm) on each end and secured with 11-0 suture (Alcon, Fort Worth, TX, USA). A third smaller pipette (tip diameter 20-25 μm) was inserted into the inflow pipette. The vessel was perfused with either APSS through the outer inflow pipette or APSS containing FITC-albumin through the inner inflow pipette. Each cannulating micropipette was connected to a reservoir and the vessel was perfused at a relatively constant intraluminal pressure and flow rate. The bath solution in the chamber was maintained at 37°C and pH 7.4 throughout the experiments. The image of the vessel was projected onto a Hamamatsu charged coupled device-intensified camera and was displayed on a high resolution monochromatic video monitor and recorded onto a VHS video recorder. The diameter of the vessel was measured on-line with a video caliper (Cardiovascular Research Institute, Texas A&M University, College Station, TX, USA).

Measurement of venular permeability

The permeability of the vessel was measured with a fluorescence ratio technique (Huxley & Curry, 1987). Using an optical window of a video photometer positioned over the venules and adjacent space on the monitor, the fluorescent intensity from the window was measured and digitized on-line by a Power Macintosh computer. In each measurement, the isolated venule was first perfused with APSS through the outer inflow pipette to establish baseline intensity. The venular lumen was then rapidly filled with APSS containing FITC-albumin by switching the perfusion to the inner inflow pipette. This produced an initial step-increase, followed by a gradual increase, in fluorescence intensity. There was a step-decrease of intensity when the fluorescence-labelled molecules were washed out of the vessel lumen by switching the perfusion back to the outer inflow pipette. The apparent solute permeability coefficient of albumin (Pa) was calculated using the equation Pa= (1/ΔIf) (dIf/dt)0(r/2), where, ΔIf is the initial step increase in fluorescent intensity, (dIf/dt)0 is the initial rate of gradual increase in intensity as the fluorescence-labelled solutes diffuse out of the vessel into the extravascular space and r is the venular radius.

In each experiment, the cannulated venule was perfused at a constant perfusion pressure of 20 cmH2O. The preparation was equilibrated for 45-60 min after cannulation and the measurements were conducted at 36-37°C and a pH of 7.35-7.45. In each vessel, a limited number (two to four) of interventions were applied. The preparations were washed three times and allowed to equilibrate for 10-15 min between interventions. In some vessels, the permeability was monitored over 6 h to ensure that the permeability properties of the venules were not significantly altered with time.

Treatment of venules

Our previous study demonstrated that VEGF increased the permeability of isolated coronary venules in a time- and concentration-dependent fashion (Wu et al. 1996, 1999). A maximum hyperpermeability response was observed within 3-5 min after application of VEGF at 10−10m. This provides a basis for the selection of dose and time course of VEGF treatment in the current study. In the control group, the Pa values were measured under the basal condition first and then at 3-5 min after topical administration of VEGF (10−10m). To prevent activation of FAK, FRNK protein was inserted into the venular endothelial cells by perfusion with a polyamine transfer reagent containing the blocking protein. In separate groups, the vessels were perfused for 60 min with a perfusate containing the transfection reagent TransIT (10 μl ml−1) with FRNK (1-3 μg ml−1) or vehicle followed by VEGF stimulation. Technical details and validation of the microvessel protein transfer approach has been provided in our previous publications (Tinsley et al. 2001; Yuan et al. 2002). The Pa values were again measured before and after application of the same dose of VEGF in the presence of the inhibitor. In each intervention, the diameter of venules was monitored to ensure that the changes in Pa were not due to alterations in vessel diameter.

Cell culture and treatment

Human umbilical vein endothelial cells (HUVEC) were ordered from Clonetics (San Diego, CA, USA). Cells were routinely maintained in gelatin-coated dishes containing EGM-2 culture media with 6 % fetal bovine serum (Clonetics). Cells were used at low (4-5) passages. Protein assays indicated that confluent HUVEC grown on a 60 mm dish contained proteins at a level of 150-170 μg.

To examine VEGF-induced phosphorylation of FAK, cells were incubated with VEGF (10−10m) for different times (0-15 min). Immunoprecipitation with the phosphotyrosine antibody PY20 was followed by immunoblotting with an anti-FAK antibody. In separate dishes, cells were first incubated with culture medium containing FRNK (1-3 μg ml−1) and TransIT (10 μl ml−1) for 1 hs. The effect of VEGF on tyrosine phosphorylation of FAK was compared between FRNK, vehicle-treated and untreated cells.

Immunoprecipitation and Western blot analysis

After the above treatment, confluent cell monolayers (passage 4-5) in 60 mm dishes were lysed by incubation for 20 min in 1 ml of ice-cold lysis buffer (20 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1 % Triton X-100, 1 mm phenylmethylsulfonyl fluoride, 0.15 u ml−1 aprotinin, 10 μg ml−1 leupeptin, 10 μg ml−1 pepstatin A, 1 mm sodium orthovanadate and 10 % glycerol). The lysate was clarified by centrifugation at 14 000 g for 10 min at 4°C. For immunoprecipitation, cell lysate containing 100 μg protein was incubated with PY20 for overnight at 4°C. Protein concentrations in cell lysates were determined with Bradford's method using the Bio-Rad protein assay reagent. The level of proteins loaded to each sample was carefully controlled to ensure equal amount of loading. The extract was incubated with protein G-Sepharose 4B (Zymed, San Francisco, CA, USA) for 2 h at 4°C. The immunocomplex was collected by centrifugation at 15 000 g for 10 s and washed three times with cold immunoprecipitation buffer containing 0.1 % Triton X-100 and once with 10 mm Tris-HCl at pH 7.4. Proteins were fractionated by SDS-polyacrylamide gel electrophoresis on precast 4-12 % gradient gels and transferred to nitrocellulose sheets for immunoblotting. Then the blots were incubated for 1 h with an anti-FAK antibody followed by incubation with a secondary antibody conjugated to horseradish peroxidase. Immunoreactive bands were detected by enhanced chemiluminescence. Images of the bands were scanned by reflectance scanning densitometry and the intensity of the bands was quantified using NIH Image software. To ensure that changes in the phosphoprotein content were not attributed to the difference in protein content, immunoprecipitated proteins were incubated in parallel with antibodies directed against the same specific proteins as controls.

In vitro permeability assay

HUVECs were seeded at a density of 105 cells cm−2 on gelatin-coated Costar Transwell membranes (VWR, Houston, TX, USA) and grown to confluence. Fluorescence-labelled bovine serum albumin was added to the top (luminal) chamber at a concentration of 10 mg ml−1, and samples were collected from both the luminal and abluminal (bottom) chambers and then analysed with a fluorescence microplate reader. Sample readings were converted with a standard curve to albumin concentration. These concentrations were used in the following equation to determine the permeability coefficient of albumin: Pa= ([a]/t)(1/A)(V/[L]), where [a] is albumin concentration, t is time in seconds, A is the area of membrane in cm2, V is the volume of the abluminal chamber and [L] is the luminal concentration. Control experiments were performed to measure tracer flux through the gelatin-coated microporous filter without cells. Monolayers that failed to form an effective barrier as indicated by a <20-fold decrease in Pa were discarded.


HUVECs were grown to confluence on gelatin-coated coverslips. After treatment, cells were immediately fixed with 2 % paraformaldehyde for 15 min and permeabilized with 0.2 % Triton X-100 for 5 min. The FAK protein was labelled by incubation of the cells for 1 h with a monoclonal anti-FAK antibody (Transduction Laboratory, Lexington, KY, USA) at 1:100 dilution followed by an FITC-labelled anti-mouse antibody. Cell nuclei were labelled with Hoechst 33342. The coverslip was mounted on a glass slide and cells were observed using a Zeiss Axiovert 300M fluorescence microscope at ×40. Images were collected through an AxioCam MRm digital camera (Carl Zeiss, Thornwood, NY, USA) and processed with Zeiss Axiovision 4.0 software.

Data analysis

In the intact vessel studies, Pa was measured two to three times for each venule at each experimental intervention and the values were averaged. For all experiments n is given as the number of vessels studied, with each vessel representing a separate animal. For each experimental condition, the values of Pa from different venules were averaged, normalized to the basal values obtained before drug treatments, and reported as percentages of basal (means ±s.e.m.). Analysis of variance was used to evaluate the significance of intergroup differences. A value of P < 0.05 was considered significant for the comparisons.


Our previous experiments have shown a dose- and time-dependent increase in the permeability of porcine coronary venules upon VEGF stimulation, in which a maximal response (2- to 3-fold increase in Pa) has been observed at 10−10m of VEGF at 3-5 min. The current study used the same isolated coronary venule approach. Under the basal condition, the Pa value of five vessels was 2.69 ± 0.16 × 10−6 cm s−1. The transfer reagent per se did not alter either the basal permeability (Pa= 2.72 ± 0.17 × 10−6 cm s−1) or the stimulated permeability (Pa= 5.50 ± 0.24 × 10−6 cm s−1 after VEGF, which was 206 % of basal, P= 0.0001vs. basal) (Fig. 1). In contrast, the permeability response to VEGF was significantly reduced in FRNK-treated venules in a dose-related manner. At the concentration of 1 μg ml−1 (n= 5), FRNK treatment produced Pa values of 2.96 ± 0.23 × 10−6 cm s−1 before VEGF and 4.46 ± 0.47 × 10−6 cm s−1 after VEGF (151 % of basal, P= 0.01vs. basal, P= 0.07vs. inactive control). At the concentration of 3 μg ml−1 (n= 6), the Pa values of FRNK-treated venules were 2.59 ± 0.05 × 10−6 cm s−1 before VEGF and 3.20 ± 0.35 × 10−6 cm s−1 (123 % of basal, P= 0.125vs. basal, P= 0.000004vs. inactive control), reflecting a 80 % inhibition of the VEGF permeability response.

Figure 1.

Topical application of vascular endothelial growth factor (VEGF; 10−10m) significantly increased albumin permeability in intact perfused coronary venules under control conditions

Inhibition of focal adhesion kinase (FAK) by direct delivery of FAK-related non-kinase (FRNK) peptide into endothelial cells of the venules attenuated VEGF-induced hyperpermeability in a dose-related manner. Numbers in parentheses represent the numbers of vessels studied. * Significant difference vs. basal; † significant difference vs. vehicle control (TranIt only).

The cultured HUVEC monolayers displayed a similar pattern of permeability response to VEGF and FRNK treatments (Fig. 2). The basal Pa value of monolayers was 1.94 ± 0.08 × 10−5 cm s−1. In the absence of FRNK (n= 4), VEGF (10−10m) increased Pa to 3.46 ± 0.05 × 10−5 cm s−1 which was 178 % of the basal value (P= 0.00008vs. basal). In FRNK (3 μg ml−1)-transfected cells, the permeability response to VEGF was reduced to 2.50 ± 0.18 × 10−5 cm s−1 (128 % of basal value, P= 0.004vs. basal, P= 0.0007vs. VEGF treated without FRNK).

Figure 2.

VEGF (10−10m) induced an increase in albumin flux across cultured human umbilical vein endothelial cell monolayers in the same time course (3-5 min) as seen in venules

The hyperpermeability response was reduced in cells pre-treated with FRNK; n, number of dishes studied. * Significant difference vs. basal; † significant difference vs. vehicle control.

Western blot analyses showed a significant increase in the phosphotyrosine content of FAK (Fig. 3A), which occurred upon stimulation, reached the maximum at 3-5 min, and started to decline after 5 min treatment with VEGF (10−10m). A dose-dependent effect of VEGF on tyrosine phosphorylation of FAK was also observed (Fig. 3B). The time course and dose dependency of VEGF-induced FAK phosphorylation corresponded to VEGF-induced venular hyperpermeability. More importantly, transfer of FRNK protein into the cells greatly reduced VEGF-induced FAK phosphorylation in a dose-related manner (Fig. 3C).

Figure 3.

Western blots showing the effect of VEGF (10−10m) on tyrosine phosphorylation of FAK

A, human umbilical vein endothelial cells were stimulated with VEGF for different times and then lysed. The lysates were immunoprecipitated with an anti-phosphotyrosine antibody and followed by blotting with an anti-FAK antibody. B, cells were treated for 3-5 min with different concentrations of VEGF. Immunoprecipitation followed by immunoblotting was performed in cell lysates as described above. The second panel in B shows the protein content of FAK in each group. C, cells were pretreated with protein transfer vehicle alone or in combination with FRNK (1 or 3 μg) and then stimulated with VEGF (10−10m). Bar graphs show statistically evaluated optical densities of the protein bands. * Significant difference vs. basal.

In addition, immunofluorescence analysis of the cell monolayers revealed a significant change in the FAK morphology and distribution after VEGF (Fig. 4). Under control conditions, FAK staining showed a punctate pattern which was distributed across the cells. Upon VEGF (10−10m, 3-5 min) stimulation, the staining became elongated in a dash-like pattern which seemed to align with the longitudinal axis of the cells, indicating a possible relationship between focal adhesion redistribution and cell contractile force development. The VEGF-induced FAK conformational changes were diminished in FRNK (3 μg ml−1)-treated cells.

Figure 4.

Immunofluorescence staining of FAK in human umbilical vein endothelial cell monolayers

A, control cells under the basal condition. B, cells exposed to VEGF (10−10m) for 5 min. C, FRNK (3 μg ml−1)-pretreated cells subjected to the same concentration of VEGF. Note that the upon VEGF stimulation FAK staining changed from a punctate pattern to an elongated, dash-like staining which seems to align with the longitudinal axis of the cells. The morphological change was diminished in cells transfected with FRNK.


The importance of FAK in the signal transduction of endothelial adhesion, contraction, and migration during angiogenesis and inflammation has recently been recognized. However, the direct effect of FAK on the functional regulation of microvascular endothelial barrier property remains to be established. The shortcoming of direct evidence confirming the signalling role of FAK in microvascular permeability is largely due to the lack of specific FAK inhibitors and the technical difficulty of quantifying permeability in living microvessels. In this study, we used the single isolated venule model for a quantitative assessment of microvascular barrier function under conditions where the intact endothelium adhered to its native extracellular matrix. We also cloned and purified an endogenously expressed FAK competitive inhibitor, FRNK, and transferred it directly into the endothelium of venules for specific blockage of FAK function. Our major finding is that FAK activation plays an important role in VEGF-induced increases of venular permeability. To the best of our knowledge, this is the first report on the functional significance of FAK signalling in VEGF regulation of microvascular permeability. While there is a recent paper describing a modulatory role of antisense inhibition of FAK expression in agonist-induced hyperpermeability in cultured endothelial cell monolayers (Mehta et al. 2002), the current study presents new evidence suggesting that functional activation of FAK mediates VEGF-elicited endothelial barrier dysfunction. More importantly, our conclusion was based on data obtained from intact exchange microvessels where the endothelium was studied under native conditions.

VEGF-induced microvascular hyperpermeability not only represents an integral component of the angiogenic process (Dvorak et al. 1995; Senger, 1996), but also contributes to the development of many pathophysiological conditions, such as inflammation, diabetic retinopathy, ischaemic heart disease, reperfusion injury, rheumatoid arthritis, and tumour growth and metastasis (van Bruggen et al. 1999; Ferrara, 2001; Zachary & Gliki, 2001). The molecular mechanism by which VEGF increases microvascular permeability is not fully understood. Our previous experiments using the isolated coronary venule preparation indicate that VEGF induces a rapid and transient increase in venular permeability; the effect is mediated by a tyrosine kinase receptor triggered intracellular signalling pathway involving the activation of phospholipase C, cytosolic calcium, protein kinase C (PKC), nitric oxide (NO) synthase, guanylate cyclase and protein kinase G (Wu et al. 1996, 1999). The results are in agreement with previous work in frog mesenteric microvessels (Bates & Curry, 1996) showing that VEGF perfusion results in an acute transient increase in hydraulic conductivity which lasts a few minutes and is followed by a chronic response in the next few days. These investigators further suggest that VEGF increases permeability acutely by acting on its receptor KDR to activate phospholipase C, which stimulates a diacylglycerol-dependent calcium influx through store-independent cation channels. The production of diacylglycerol and elevation in intracellular calcium further activate the PKC pathway and the NO-signalling cascade, leading to endothelial hyperpermeability (Bates & Harper, 2003). The mechanisms underlying the transient nature of the acute permeability response to VEGF remain to be identified, so do the pathways responsible for its chronic effect.

Recent experimental evidence suggests FAK as an important component of VEGF-stimulated signalling pathway. It is known that FAK activation participates in VEGF-elicited cell proliferation, adhesion, migration and contraction (Takahashi et al. 1999; Rousseau et al. 2000; Eliceiri et al. 2002). In vascular endothelial cells, VEGF is able to cause FAK tyrosine phosphorylation and translocation to the focal adhesion complex, an important structure that not only tethers the endothelial monolayer to the basement membrane but also serves as an transducer in mediating endothelial response to physical stress and chemical stimuli (Abedi & Zachary, 1997; Qi & Claesson-Welsh, 2000; Abu-Ghazaleh et al. 2001). Overexpression of dominant negative FAK in endothelial cells attenuates VEGF-induced activation of intracellular signalling molecules (Qi & Claesson-Welsh, 2000). However, despite the fact that the FAK signalling in VEGF regulation of cell morphology and motility has been well recognized, whether the same mechanism is involved in the endothelial permeability response to the growth factor remains a question. In this regard, previous studies by us (Yuan et al. 1998) and others (Soldi et al. 1996; Abedi & Zachary, 1997) have shown that histamine, thrombin, phorbol esters and platelet activating factors, which have been defined as typical permeability-increasing factors, cause FAK phosphorylation coupled with microvascular barrier dysfunction. A recent study in cultured human pulmonary artery endothelial cells treated with FAK antisense oligonucleotides has demonstrated a modulatory role of FAK in regulating endothelial barrier function (Yuan et al. 1998). While these reports provide a close association between FAK activation and endothelial permeability during inflammatory stimulation, there has been no evidence that can directly support a role for FAK in the functional regulation of endothelial barrier, especially under the stimulation of VEGF. The difficulty of specifying the FAK mechanism lies in the lack of specific pharmacological or molecular approaches for single venules.

Recent advances in molecular biology have led to the discovery of the FAK-related nonkinase FRNK. FRNK is the C-terminal non-catalytic domain of FAK; it contains the focal adhesion target sequence but is devoid of any kinase activity. FRNK is autonomously expressed in many cell types including vascular endothelial cells and is frequently used as an endogenous dominant negative regulator of FAK (Schaller, 2001; Mortier et al. 2001). Due to the relatively large size and cell-impermeable nature of this molecule, its functional importance has not been tested in intact microvessels. In most of the in vitro experiments, dominant-negative inhibition of FAK was induced through viral transfection of FRNK genes (Schaller et al. 2000; Qi & Claesson-Welsh, 2000; Schaller, 2001; Mehta et al. 2002), an approach that is not suitable for in vivo or in situ study of microvascular function. In this study, we took advantage of our recently developed microvessel protein transfer technique (Tinsley et al. 1998, 2002) to efficiently deliver the intact FRNK peptide into the endothelium. As shown in the current results, the transfer reagent per se did not alter the basal and VEGF-stimulated permeability; in contrast, transfer of the FAK inhibiting peptide FRNK into the venular endothelial cells significantly attenuated the increase in permeability caused by VEGF, suggesting that FAK contributes to VEGF-induced microvascular hyperpermeability. The specific role of FAK in the VEGF signalling was further supported by the observation that VEGF-stimulated FAK phosphorylation and redistribution were diminished in FRNK-treated endothelial monolayers.

Several pathways may be considered for potential mechanisms underlying the effect of FAK on endothelial barrier function. On one hand, FAK-signalled focal adhesion disassembly and redistribution may reduce cell-matrix adhesion forces, leading to rounding of the endothelial cell and widening of interendothelial channels. This possibility is consistent with an emerging view that FAK activation leads to disassembly, rather than assembly, of focal adhesion complexes (Crowley & Horwitz, 1995; O'Neill et al. 2000; Schaller, 2001). On the other hand, activated FAK may directly participate in the development of contractile force in the endothelial cells resulting in intercellular gap formation. In support of this notion, FAK phosphorylation has been linked to actin polymerization and stress fibre formation in endothelial cells treated with VEGF (Rousseau et al. 2000). Alternatively, FAK could indirectly affect endothelial barrier function through coordinating and activating other intracellular signalling molecules, including the Src-family of tyrosine kinases and the Rho-family of GTPases (Parsons et al. 2000; Abu-Ghazaleh et al. 2001; Eliceiri et al. 2002). Biochemical experiments indicate that FAK can directly phosphorylate Src and activate Rho (Berk et al. 1995; Schlaepfer et al. 1999; Schaller, 2001). It is also possible that FAK functions as a scaffolding protein that recruits and activates Src, Rho, and other focal adhesion-associated proteins (Schlaepfer et al. 1999; Schaller, 2001), which in turn cause changes in the cytoskeleton or cell-cell junctions, leading to endothelial hyperpermeability. The roles of Src and Rho in endothelial cell contraction and junctional disorganization have been documented (Thomas & Brugge, 1997; Garcia et al. 1999; Tinsley et al. 2002).

Further studies are required to test the above hypotheses in vivo and to explain how FAK activation is related to other signalling reactions known to be involved in VEGF regulation of microvascular permeability (Bates & Harper, 2003; Yuan, 2003). Currently, technical limitations, such as the lack of specific molecular probes and the difficulties in site-specific delivery of proteins or peptides, impose a great challenge to investigators who intend to use methodologies that can quantify microvascular permeability directly in vivo. Within this context, intravital microscopic measurements of protein extravasation across a microvascular bed reflect the global changes in microvascular transport influenced by haemodynamic factors. The cultured endothelial monolayer model, on the other hand, may not fully represent the in vivo physiological system. Indeed, we have found that the basal permeability coefficient of albumin is at least one magnitude higher than the value obtained from in vivo preparations. Compared with cultured cells, the isolated venule possesses a tighter barrier property and its permeability coefficient is comparable to that obtained from the intact heart (Yuan et al. 1993a). Furthermore, transfection of FRNK appeared to be more effective in isolated venules than in cultured cells. A possible explanation for this is the presence of intraluminal flow and pressure in isolated vessels.

In summary, we have examined the signaling effect of FAK on the permeability response of coronary venules and HUVEC monolayers to VEGF. Inhibition of FAK with FRNK greatly attenuated VEGF-induced FAK phosphorylation and hyperpermeability. We suggest that FAK activation plays an important role in VEGF regulation of microvascular permeability.


We thank Dr J. T. Parsons for providing Escherichia coli transformed with pET-histidine-tagged FRNK. This work was supported by the National Heart, Lung, and Blood Institute grants HL21498, HL58062 and HL61507.