Verge: A Novel Vascular Early Response Gene

Materials and Methods

Reagents. PMA, human recombinant fibroblast growth factor (FGF), RNase A, kainic acid, pentylene tetrazole (PTZ), thrombin, transforming growth factor-β (TGF-β), and horseradish peroxidase-conjugated Bandeirae Simplicifolia lectin (BS-1) were purchased from Sigma (St. Louis, MO). Angiopoietin-1 (Ang-1), angiopoietin-2 (Ang-2), and hepatocyte growth factor (HGF) were purchased from R & D Systems (Minneapolis, MN). U0126, bisindolylmaleimide I (BIM), SB203580, Calpain inhibitor I (ALLN), and Calpain inhibitor II (ALLM) were obtained from Calbiochem (La Jolla, CA)-Novabiochem (San Diego, CA). 4′,6-diamidino-2-phenylindole, dihydrochloride (DAPI) and mouse monoclonal antibody (Ab) against hemagglutinin (HA) were purchased from Roche Diagnostics (Mannheim, Germany). Fluorescent secondary antibodies were purchased from Jackson ImmunoResearch (West Grove, PA). Fluorescein isothiocyanate (FITC)-labeled phalloidin was purchased from Molecular Probes (Eugene, OR).

cDNA subtraction and isolation of full-length verge cDNA. A subtracted cDNA library was constructed as described previously (Yamagata et al., 1993). Full-length verge cDNA clones were isolated by screening an unsubstracted, oligo (dT)-primed cDNA library prepared from the hippocampus of rats that had been given cycloheximide (20 mg/kg) and 4 hr of multiple maximal electroconvulsive shock (MECS).

Northern blot analysis. Total RNA was isolated using standard CsCl density centrifugation and assayed for purity and yield by ultraviolet spectroscopy. Ten micrograms of RNA per lane was subjected to denaturing agarose gel electrophoresis. Gels were stained with ethidium bromide, and the ribosomal bands were visualized to assess equal loading of RNA. After transfer to Hybond N⁺ nylon membranes, Northern blot analysis was performed with a probe corresponding to the open reading frame (ORF) of rat verge or to a 2070 bp XhoI/HindIII cDNA fragment of human Verge. A probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an RNA loading control. Quantification of the hybridization signal was performed using the Typhoon imaging system (Amersham Biosciences, Piscataway, NJ) and normalized to GAPDH.

Generation of anti-Verge antibodies. Synthetic peptides representing the N-terminal and C-terminal 22 amino acids of the rat Verge were conjugated to thyroglobulin (m-maleimidobenzoyl-N-hydroxysuccinimide ester; Pierce, Rockford, IL) and used as immunogens in rabbits (Covance, Princeton, NJ). Rabbit polyclonal Abs were similarly generated using a glutathione S-transferase (GST) fusion protein containing the C-terminal 73 amino acids of rat Verge (Covance). Rabbit polyclonal antibodies were similarly generated using human Verge sequences. Human Verge peptides comprising amino acids 3-21 [Ab 30, 36-51 (Ab 25), and 230-248 (Ab 23)] were conjugated via an additional cysteine to keyhole limpet hemocyanin (Biotrend Chemikalien GmbH, Koln, Germany).

Isolation of rat brain microvessels. Capillary segments were isolated from PBS-perfused rat brain as described previously (Goldstein et al., 1975). Isolated microvessels were air dried onto superfrost plus-coated slides (VWR Scientific Products, West Chestire, PA) and stored at -20°C before analysis.

In situ hybridization. Colorimetric in situ hybridization was performed on fresh-frozen sectioned tissue using nonradioactive digoxigenin-labeled cRNA probes and 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium (Roche) colorimetric detection as described previously (Blackshaw and Snyder, 1997). Fluorescent in situ hybridization was performed using identical hybridization and wash conditions as for colorimetric in situ. Fluorescent-tyramide kit (NEN, Boston, MA) was used for detection.

Immunohistochemistry. Brains were removed from control rats or rats treated for 2 hr with kainic acid (11 mg/kg, i.p.). Brains were frozen in optimal cutting temperature compound (Sakura, Tokyo, Japan) and sectioned at 20 μm on superfrost-coated slides. Sections were immersion fixed in 3.7% paraformaldehyde (PFA; in PBS, pH 7.4) for 10 min, permeabilized for 15 min in 0.25% Triton X-100/PBS, and then treated for 15 min with 3% H₂O₂. Antigen retrieval was performed by placing the slides in 100 mm Citrate, pH 4.0, and heating to ∼95°C by microwave for 30 sec. Slides were blocked with 10% normal goat serum (NGS) in PBS at room temperature. Primary Ab was added to the slides and incubated at 4°C overnight. Slides were washed thoroughly, and bound antibodies were detected using ABC kit (Vector Laboratories, Burlingame, CA) followed by DAB (Sigma) colorimetric staining.

Cell culture and transfection. Primary endothelial cells from adult mouse lung and heart were isolated as described previously (Allport et al., 2002). Neonatal human dermal microvascular endothelial cells (HMVECs) were purchased from Cambrex BioScience (Walkersville, MD) and cultured in the endothelial growth medium 2-microvascular (EGM2-MV) media from the manufacturer, supplemented with EGM2 single quots. Rat brain endothelial cell line 4 (RBE4) cells have been described previously (Roux et al., 1994) and were a generous gift from John Laterra (Johns Hopkins University, Baltimore, MD). These cells were grown on gelatin-coated tissue culture dishes in RBE4 media (DMEM, 10% fetal bovine serum (FBS), 1% penicillin-streptomycin, 20 mm HEPES, pH 7.4, and 1 ng/ml hrbFGF). Stable cell lines were established that express 5′-HA-Verge transgene. For transfection, the plasmids pRK5-5′HA-Verge and pPUR (containing the puromycin resistance gene; Clontech, Palo Alto, CA) were mixed together in a molar ratio of 10:1. RBE4 cells were electroporated with this plasmid mixture and allowed to recover for 48 hr. Stably transfected cells were selected with 10 μg/ml puromycin (Clontech) and screened for transgene expression by Western blot analysis. Four stable lines were selected and termed clones 2, 38, 41, and 42. Nontransgene-expressing, puromycin-resistant clones were also selected as negative controls (termed clones A and B). COS-1 cells were purchased from American Type Culture Collection (Manassas, VA) and grown in DMEM, 10% FBS, and 1% penicillin/streptomycin and were transiently transfected using calcium phosphate.

Hypoxia treatment. HMVECs of passage 6 were seeded with 1.2 × 10⁶ cells per 10 cm dish and grown to confluence in supplemented EGM2-MV media. On day 4, cells covered by 3 ml of medium per plate were placed into a 37°C tempered hypoxia chamber. For catalytic removal of residual oxygen, a package of BBL GasPak replacement charges (Becton Dickinson, Cockeysville, MD) was placed into the chamber and then it was constantly floated with a mixture of N₂ (90% volume), CO₂ (5% volume), and H₂ (5% volume) at 37°C. Plates were removed from the chamber at indicated time points, and RNA preparation and Northern blot analysis was performed as described.

Immunofluorescence. HMVECs of passage 8 were grown on fibronectin-coated glass coverslips. Cells were washed three times with PBS and fixed for 15 min at 37°C in 4% PFA in PBS. Cells were rinsed twice with PBS and washed three times for 5 min with glycine (50 mm)/PBS. After permeabilization on ice with 0.15% Triton X-100/PBS for 7 min, cells were blocked with 2% NGS/PBS for 1 hr at room temperature. Incubation with anti-Verge antisera was performed in a 1:200 dilution in 1.5% (v/v) NGS/PBS for 2 hr at room temperature. Cells were washed four times with washing buffer [1.5% (v/v) NGS, 100 mm NaCl, 10 mm Tris-HCl, pH 7.6] and incubated with FITC-anti-rabbit IgG for 1 hr. Cells were washed three times for 5 min with washing buffer, three times with PBS, and rinsed four times with Tris-HCl (10 mm, pH 7.6). Coverslips were mounted with Aqua Poly/Mount (Polysciences, Warrington, PA), and cells were visualized using a confocal Leica (Bensheim, Germany) microscope.

Electrical cell-substrate impedance-sensing experiments. RBE4 cells were grown to confluence on small gold electrodes (Applied Biophysics, Troy, NY) in RBE4 media (Tiruppathi et al., 1992; Verin et al., 2000; Garcia et al., 2001). Electrical resistance of endothelial cell monolayers was measured with an electrical cell-substrate impedance-sensing (ECIS) system (Applied Biophysics). Before experimentation, cells were incubated with fresh media, and resistance across the resting monolayer was monitored for 1 hr to ensure a stable baseline. A 4000 Hz AC signal with a 1 V amplitude was applied to the RBE4 monolayer through a 1 MΩ resistor, creating an approximate constant-current source of 1 μA. The small electrode and larger counter electrode (1 cm²) are connected to a phase-sensitive lock-in amplifier with a built-in differential preamplifier (5316A; EG & G Instruments, Princeton, NJ). The in-phase and out-of-phase voltages between the electrodes were monitored in real time with the lock-in amplifier and were converted to scalar measurements of transendothelial impedance. Resistance values from each microelectrode (measured in ohms) were normalized as the ratio of measured resistance to baseline resistance and plotted versus time and statistically assessed using Epool software, kindly provided by Dr. K. L. Schaphorst (Johns Hopkins University School of Medicine, Baltimore, MD). At time 0, either agonist or vehicle was added to the cells, and resistance was assayed over the course of several hours. Resistance data were normalized to the initial voltage and plotted as normalized resistance. An independent t test was performed to analyze the differences between means of electrical resistance. Results are expressed as means ± SD. Differences in two groups are considered statistically significant when p < 0.05.

Focal cerebral ischemia model. For middle cerebral artery occlusion and reperfusion, the filament model was used as described previously (Schneider et al., 1999).

Quantitative PCR. cDNA was synthesized using oligo-dT primers and Superscript II reverse transcriptase (RT) and purified by using silica-based columns (Qiagen, Hilden, Germany). Quantitative PCR was performed using the LightCycler system (Roche) as described previously (Brambrink et al., 2000). In brief, cDNAs were serially threefold diluted to obtain four to five concentration points for quantification. The concentrations of genes of interest in different cDNA samples were normalized to levels of the housekeeping gene cyclophilin; 60°C was chosen as the annealing temperature. Measurements of cyclophilin and Verge amplicons were performed at 84°C to exclude potential PCR by-products from the analysis. The following primers were used for PCR amplification: cyclophilin: cyc5(s), 5′-ACCCCACCGTGTTCTTCGAC-3′; acyc300(as), 5′-CATTTGCCATGGACAAGATG-3′, Verge: 119-17(s), 5′-GGGTCTGAATAGGAA GGGAGTCTG-3′; L119-19(as), 5′-ATAGGACATCAGGTTTCCAAGGTC-3′. The specificity of the PCR reaction was checked by determining the melting point and by size using agarose gel electrophoresis. Values were averaged from at least three different dilution measurements per sample and are given as means ± SD.

Verge is a novel immediate early gene

We used a differential cloning strategy to identify mRNAs that are rapidly induced in rat brain by electrical seizure (Yamagata et al., 1993). A novel cDNA (verge) corresponding to a ∼3.0 kb mRNA was isolated, and three independent full-length cDNAs were sequenced. The size of the full-length verge cDNA (2947 bp) corresponds closely to the estimated size of the rat mRNA identified by Northern blotting (Fig. 1A). The longest ORF with an initiator methionine predicts a protein of 246 amino acids. This initiator methionine is an acceptable Kozak consensus sequence (Kozak, 1981) and is downstream of an in-frame stop codon (Fig. 2A). The 3′ noncoding region contains eight AUUUA (ATTTA in cDNA) mRNA instability motifs (Shaw and Kamen, 1986). The predicted molecular mass of rat Verge is 26.8 kDa, and the pI is 9.60. Analysis of Verge amino acid sequence predicts an N-terminal leucine-zipper motif (amino acids residues 15-43), two regions of hydrophobicity [amino acids residues 50-97 and 157-179, as predicted using the TMpred program (Kyte and Doolittle, 1982)], and a C-terminal coiled-coil domain [amino acid residues 188-240 predicted by the Lupas method (Lupas et al., 1991)] (Fig. 2B). There are several predicted phosphorylation sites including a PKC site at residue 153 and PKA sites at residues 50 and 237.

Verge sequences and predicted ORFs were subsequently identified for both mouse and human in the National Center for Biotechnology Information database (NCB AX_665396 and NM_030817, respectively), and three independent mouse cDNAs obtained by screening a brain library were confirmed to match the NCBI database. The human verge mRNA is ∼4.7 kb and encodes an ORF predicted to be 248 amino acids, with two additional amino acids at the N terminus. The amino acid residues are 87% identical between rat, mouse, and human species (Fig. 2B).

Basic Local Alignment Search Tool (BLAST) analysis of the verge ORF identified several proteins that contain regions of modest homology (Fig. 2C). In mammals, all six members of the apolipoprotein L (ApoL) family (NP_003652.2, NP_6663612.2, XP_235464.1, NP_085146.2, NP_085145.1, NP_085144.1) encode a ∼70 amino acid region that is homologous to Verge. Two putative Fugu rubripes proteins (NCBI accession numbers SINFRUP053682 and SINFRUP055309) and two putative C. elegans proteins (NCBI accession numbers T15963 and NP_740988.1) also show homology to Verge. One of the F. rubripes proteins (SINFRUP053682) appears to be a fish homolog of mammalian ApoL, and the second F. rubripes protein (SINFRUP055309) appears to be a partial sequence. In all cases, homology is limited to amino acid residues 35-108 of Verge (Fig. 2C). Within this region, Verge shows 30% identity/55% amino acid homology with the ApoL family members, 30% identity/61% homology with F. rubripes (SINFRUP055309), and 25% identity/42% homology with C. elegans. In comparison, members of the ApoL family show ∼50% identity/∼72% amino acid homology in this region with each other and 30% identity/50% homology with the putative F. rubripes or C. elegans proteins. PsiBlast analysis did not identify additional proteins that contain the putative ApoL domain.

Analysis of expressed sequence tag (EST) and serial analysis of gene expression (SAGE) databases revealed verge mRNA to be abundantly expressed in a large number of tissues sources, including fetal tissues, placenta, and solid tumors (data not shown). BLAST analysis indicates that the genomic location of human verge is chromosomal position 12p13.2.

Verge is rapidly induced in endothelial cells of cerebral vessels by seizures

Verge mRNA expression was analyzed for regulation by seizure evoked by MECS or by chemical inducers of seizures, including kainic acid and PTZ. We also examined the dependence of verge mRNA induction on de novo protein synthesis using cycloheximide, which blocks protein translation (Rajagopalan and Malter, 1996). Northern blot analysis demonstrates that both MECS seizure and cycloheximide induce verge mRNA (Fig. 1A). Induction by MECS is rapid and transient with peak mRNA expression occurring 20-30 min after seizure. Verge mRNA expression returns to baseline by 1 hr. Induction by cycloheximide is also rapid but continues up to 6 hr (final time point) after administration. The level of verge mRNA after 6 hr of cycloheximide is estimated to be ∼20-fold higher than the peak level induced by MECS (Fig. 1A). Similarly, treatment of rats with the epileptogenic excitatory amino acid kainic acid (15 mg/kg, i.p.) or PTZ (50 mg/kg, i.p.) resulted in a rapid transient induction of verge mRNA in brain (Fig. 1B). This temporal pattern of induction by seizure, and induction in rats treated with cycloheximide, indicates that verge is regulated as an immediate early gene in adult brain.

In situ analysis indicates that Verge mRNA is expressed in capillaries and in the cells lining the inner surface of veins and arteries (Fig. 3A-C), suggesting expression by endothelial cells. Verge mRNA was not detected in smooth muscle cells. Verge mRNA was also detected in the endocardium that lines the atria and ventricles of the heart (Fig. 3D). To confirm that verge mRNA is expressed in endothelial cells, dual fluorescent labeling was performed using the BS-1, which selectively binds endothelial cells (Thurston et al., 2000). Verge mRNA colocalized with BS-1 in isolated brain capillaries prepared from rats treated with cycloheximide (Fig. 3E). At high magnification, verge mRNA is visible in the perinuclear region, whereas the BS-1 lectin stains plasma membrane (Fig. 3F). Verge mRNA that is induced by cycloheximide was similarly detected in vessel endothelium of all tissues sampled (lung, liver, spleen, brain, retina, kidney, heart, adrenal gland, skin; data not shown).

Antibodies generated against N- and C-terminal peptides or GST fusion proteins detected Verge transgene expressed in COS cells. Verge Abs were specific and at least as sensitive as the best commercial anti-HA Abs in detecting the epitope-tagged Verge transgene by Western blotting (Fig. 4A). However, native Verge protein was not clearly detected in lysates from hippocampus or other tissues of control rats. We next examined Verge protein expression in hippocampus of naive and kainate-treated rats. N- and C-terminal Verge antibodies recognized a kainate-inducible protein of the correct molecular weight (Fig. 4B). Immunohistochemistry was performed using the same antibodies and revealed a specific signal in kainate-treated brain (Fig. 4D,E) that was not present in naive brain (Fig. 4C). Preimmune serum failed to detect a specific signal in kainate-treated brain (data not shown). Staining was restricted to small blood vessels, consistent with verge mRNA expression in endothelial cells. We conclude that Verge protein is induced by kainate seizures and is expressed in vascular cells in the adult brain.

Verge is induced in cultured endothelial cells by TNFα, FGF2, and hypoxia

To investigate the effects of vascular bioactive molecules on verge expression, primary cultured mouse lung endothelial cells (MLECs) were exposed to either Ang-1 (50 ng/ml), Ang-2 (50 ng/ml), FGF2 (50 ng/ml), HGF (40 ng/ml), TNF-α (25 ng/ml), PMA (300 nm), or cycloheximide (10 μg/ml). Northern blot analysis was performed, and hybridization signals were quantified (Fig. 5A,B). Similar to in vivo studies, cycloheximide induced continuous verge mRNA expression. FGF2 and TNF-α each induced verge expression ∼20-fold within 30 min after initial exposure. Verge mRNA levels returned to near baseline levels by 60 min. HGF induced verge expression approximately twofold to threefold. PMA proved to be the most potent inducer of verge as mRNA levels increased 55-fold after 30 min of stimulation and were decreased at 60 min. Ang-1 and Ang-2 showed subtle, if any, effects on verge expression at the concentrations and time points studied. Identical Northern blots were probed for c-fos, zif268, and jun B mRNA, and their signal was quantified (Fig. 5B). Of these, c-fos showed the most similar profile of responses suggesting a shared signaling pathway leading to transcriptional activation. Verge mRNA is also rapidly induced by FGF2 in primary mouse heart endothelial cells (Fig. 5C). These studies provide precedent in which verge can be induced in endothelial cells by membrane receptor signaling pathways.

We next examined verge mRNA expression in primary cultured human endothelial cells. Northern analysis confirmed that verge mRNA is expressed by neonatal HMVECs, and the ∼4.5-5.0 kb size corresponds well with estimated size of the human verge cDNA (4.6 kb) present in the NCBI database and ESTs identified by our laboratory (Fig. 5D). To investigate the effects of hypoxia on verge expression, HMVECs were exposed to hypoxic conditions. Verge mRNA expression revealed a time-dependent increase over the 3 hr studied (Fig. 5D). Note that hypoxia and cycloheximide are the only stimuli identified to produce persistent upregulation of verge mRNA.

Verge mRNA is focally induced by cerebral ischemia

Based on the preceding observation that verge is differentially induced by defined growth factors or hypoxia, we examined the hypothesis that verge expression may be regulated by local conditions of the tissue. We examined the effect of focal ischemia on the expression of verge using a well characterized stroke model in which the middle cerebral artery of an adult mouse is transiently occluded by a filament that is threaded into the lumen of the vessel (Schneider et al., 1999). This model results in a transient disruption of the blood-brain barrier and increased vascular permeability (Cole et al., 1990; Betz et al., 1994). Transient focal ischemia (1.5 hr ischemia and 2 hr reperfusion) resulted in a threefold increase in verge mRNA expression in the ischemic hemisphere compared with the nonischemic hemisphere of the same animal (Fig. 6A). Similarly, immunofluorescence studies revealed increased Verge protein to be present in blood vessels in ischemic portions of brain (Fig. 6C), relative to nonischemic portions of the same brain (Fig. 6B). The increase in the ischemic hemisphere suggests that local tissue conditions can influence the induction or stability of verge mRNA and protein.

Verge is developmentally regulated in the rat heart

Although verge expression is low in normal adult vasculature, the abundance of verge ESTs from developing tissues suggested that it may be developmentally regulated. An initial survey of tissues from postnatal day (P) 0 rat confirmed prominent expression of verge mRNA in heart. Detailed in situ analysis of developing heart is presented in Figure 7. Verge mRNA is evident at high levels in heart from embryonic day (E) 17 to P12 (Fig. 7A-D). Verge expression is restricted to vessels. There is a marked decrease in verge expression between P12 (Fig. 7D) and P17 (Fig. 7E), and by P30 verge expression is no longer detected (Fig. 7F). The developmental progression of angiogenic activity is well documented in heart (Baldwin, 1996; Tomanek, 1996; Tomanek et al., 1996, 1999) and parallels the expression of verge.

Verge protein is expressed in primary cultured endothelial cells and localizes to focal regions at the periphery of the cell

Using conventional approaches, we were not able to detect Verge protein in HMVECs under normoxic or hypoxic conditions by Western blotting. This stands in contrast to the striking induction of verge mRNA by hypoxia and growth factors (Fig. 5). IEG proteins are often in rapid turnover (Kruijer et al., 1984; Murphy et al., 2002), and treatment of cells with protease inhibitors can lead to accumulation of rapidly degraded proteins (Wang et al., 1994; Blagosklonny et al., 1999). Accordingly, we examined the effect of treating endothelial cells with protease inhibitors. Verge protein was detected in HMVEC lysates after treatment with the cysteine protease inhibitors ALLN and ALLM (25 μg/ml of each added directly to media for 3-5 hr) (Fig. 8A). Western blots revealed a prominent band of ∼26 kDa that was present in total lysates prepared from ALLN-ALLM-treated cells but not untreated cells. Immunofluorescence was performed to examine the sub-cellular localization of Verge in HMVECs. Staining with two independent Verge antibodies showed Verge protein to be enriched in cells treated with ALLN-ALLM and localized to the periphery of cells in close proximity to the plasma membrane (Fig. 8C,E). Verge immunoreactivity frequently showed focal accumulations along the periphery of cells. Peripheral staining was not seen in control HMVECs stained with preimmune serum (Fig. 8B,D) (data not shown). Confocal images confirmed that focal puncta of Verge localize to basolateral periphery, particularly at sites where the membrane displayed outward convexity (Fig. 8F,G). These data indicate that Verge protein is rapidly synthesized and degraded in HMVECs and suggest that Verge functions at or near the plasma membrane.

Stable Verge-expressing RBE4 cells show increased sensitivity to PKC stimulation in an in vitro assay of monolayer permeability

To examine the function of Verge in endothelial cells, we generated stable cell lines that express HA-tagged full-length Verge. RBE4 cells, an immortalized cell line derived from rat brain microvasculature (Roux et al., 1994), were transfected with separate expression constructs for puromycin resistance and HA-tagged full-length Verge. Although low levels of endogenous verge mRNA were detected in RBE4 cells by RT-PCR, endogenous Verge protein was not detected in Western blots of total lysate (Fig. 9C, top panel). Four cell lines were selected and confirmed to express Verge transgene. In these cells, Verge transgene was evident in total lysates and migrated at the anticipated molecular mass. Two cell lines from the same screen that did not express Verge transgene were maintained as controls.

Because Verge accumulates at the periphery of HMVECs, we hypothesized that it might function to regulate endothelial cell barrier function. Accordingly, we performed assays of cellular permeability using Verge-expressing endothelial cell lines. Confluent monolayers of RBE4 clones were grown on gold electrodes in chambers designed to measure the electrical resistance of the layer using the method of Tiruppathi et al. (1992). Growth properties and measures of base line electrical resistance were identical for Verge-expressing and nonexpressing (control) RBE4 cell lines (Table 1). We then examined the effect of adding agents that are reported to influence monolayer resistance. Thrombin and TGF-β failed to affect permeability characteristics of either control or stable Verge-expressing monolayers (Table 1). However, the PKC-activating phorbol ester PMA decreased monolayer resistance of stable Verge-expressing cells but not control cells (Table 1). In all four stably expressing Verge RBE4 clones, but not in two nonexpressing clones, PMA (100 nm) treatment induced a rapid decline in transendothelial electrical resistance (TER). A reduction of resistance was statistically significant within 10 min, and ∼90% of the response was complete by 30 min after agonist stimulation (Fig. 9A). The rapid kinetics and magnitude of PMA-induced increase in permeability is similar to that reported using primary endothelial cells (Verin et al., 2000). In contrast, cell lines that do not express Verge transgene showed <10% reduction in monolayer resistance over the 3.5 hr time course of the experiment. PMA effects were dose dependent and maximal at 100-300 nm (Fig. 9B).

We next examined the molecular mechanism of the PMA effect. Preincubation of 5′HA-Verge-expressing RBE4 cells with the PKC inhibitor BIM (Toullec et al., 1991) (1 μm) did not alter basal electrical resistance but did block the PMA-induced decrease in electrical resistance (Table 2). Pretreatment with either U0126, a potent MAP kinase kinase (MEK) 1/2 inhibitor (Favata et al., 1998), or SB203580, which specifically inhibits p38 (Lee et al., 1994), did not affect basal resistance or responses to PMA (Table 2). We conclude that in this in vitro model of monolayer permeability, increased expression of Verge leads to an increased sensitivity to PMA that requires PKC. This effect does not appear to be dependent on MEK1/2 and p38.

To further assess the mechanism of increased response to PMA, we performed Western blots on RBE4 cell lysates to assess expression of PKC isoforms that are reported to be expressed in endothelial cells and that are stimulated by PMA (Ferro et al., 2000; Idris et al., 2001; Siflinger-Birnboim and Johnson, 2003). No significant change in expression levels of PKC-α, -β, -γ, -δ, or -ϵ were detected between Verge expressing and nonexpressing RBE4 cell lines (Fig. 9C). Thus, differences in response to PMA are not likely attributable to changes in expression of PKCs.

PMA treatment of Verge-expressing RBE4 cells leads to a rapid and transient remodeling of the actin cytoskeleton and the formation of paracellular gaps

PMA-induced increases in endothelial cell permeability are often associated with changes in cell shape, reorganization of the cytoskeleton, and the formation of para-cellular gaps (Antonov et al., 1986; Grigorian and Ryan, 1987; Lynch et al., 1990; Garcia et al., 1992). As noted, control and Verge-expressing RBE4 cell lines form similar monolayers, and staining of nuclei confirm their cell density to be approximately equal (Fig. 10A,B). Visualization of F-actin with FITC-labeled phalloidin showed that the actin stress fiber network is also similar in control and Verge-expressing RBE4 cells (Fig. 10C,D, respectively). In Verge-expressing RBE4 cells, treatment with PMA (100 nm, 1 hr) led to a reorganization of the actin cytoskeleton and the formation of paracellular gaps (Fig. 10F). PMA failed to induce a change in actin cytostructure or formation of paracellular gaps in the control, non-Verge-expressing lines (Fig. 10E). Remodeling of the actin cytoskeleton and the formation of paracellular gaps was transient as normal monolayer morphology returned to stable expressing cell lines 24 hr after continuous PMA treatment (data not shown).