Abstract
In ulcerative colitis (UC), an increased expression of vascular endothelial growth factor (VEGF) correlates with disease activity, but a causal relationship is unknown. We tested the hypothesis that VEGF plays a mechanistic role in the pathogenesis of experimental UC and that VEGF neutralization may exert therapeutic effect. UC was induced in Sprague-Dawley rats by 6% iodoacetamide given intracolonically. Neutralizing anti-VEGF antibody (50 μg/rat), nonspecific IgG, or saline (0.1 ml/rat) was injected intramuscularly on the 3rd and 5th days after iodoacetamide enema. Rats were euthanized on the 7th day. We examined the extent of macroscopic, histologic, and clinical features of colitis and colonic vascular permeability. Colonic VEGF mRNA and protein expressions increased as early as 0.5 h after iodoacetamide enema and remained elevated in the active phase of colitis. Treatment with anti-VEGF antibody markedly improved the clinical and morphologic features of UC. Colonic lesion area was significantly reduced from 370 ± 140 or 311 ± 170 mm2 in saline- or IgG-treated groups to 122 ± 57 mm2 in the anti-VEGF-group (p < 0.05). Increased colonic vascular permeability was decreased by the anti-VEGF antibody (p < 0.05) and the Src inhibitor PP1 [pyrazolopyrimidine, 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]-pyrimidine] (p < 0.01). The number of acute and chronic inflammatory cells in the lesion area was significantly reduced in anti-VEGF-treated rats. In the anti-VEGF-treated group, mucosal levels of VEGF, platelet-derived growth factor, and basic fibroblast growth factor were also reduced. In conclusion: 1) Neutralizing anti-VEGF antibody significantly ameliorates experimental UC in rats in part by reducing excessive vascular permeability and decreasing inflammatory cells infiltration; and 2) VEGF seems to mediate increased colonic vascular permeability in experimental UC via the Src-dependent mechanism.
Inflammatory bowel disease (IBD), including UC and Crohn's disease (CD), are chronic inflammatory disorders that arise from a complex interplay among immunological (Yen et al., 2006), genetic (Hugot et al., 2001), and environmental (Lodes et al., 2004) factors. IBD affects approximately 1 million patients in the United States and has a major clinical and economic impact (Sandler et al., 2002).
Recent publications suggest that altered angiogenesis may be a critical component of IBD pathogenesis (Spalinger et al., 2000; Danese et al., 2006; Koutroubakis et al., 2006). Clinical studies and animal models of experimental colitis showed increased microvascular density in the mucosal and submucosal tissue (Spalinger et al., 2000; Chidlow et al., 2006; Danese et al., 2006) and up-regulation of proangiogenic factors such as VEGF (Griga et al., 1998, 1999, 2002; Kanazawa et al., 2001; Sandor et al., 2006), basic fibroblast growth factor (bFGF) (Kanazawa et al., 2001), and platelet-derived growth factor (PDGF) (Kanazawa et al., 2001). Extracts from IBD mucosa induced a potent angiogenic response in both the corneal and chorioallantoic assays (Danese et al., 2006).
Angiogenesis in the area of chronic inflammation, such as that observed in IBD, is abnormal and characterized by distorted vasculature, increased permeability, and thrombogenic potential (Carmeliet, 2003). Moreover, abnormal angiogenesis may facilitate migration of inflammatory cells to the site of inflammation, leading to the perpetuation of chronic inflammation (Chidlow et al., 2006).
VEGF is a fundamental regulator of angiogenesis (Carmeliet et al., 1996). It was originally discovered as a potent “vascular permeability factor” (VPF) (Senger et al., 1986), which rapidly increases permeability and causes vascular leakage through VEGF receptor-2-Src-dependent mechanism (Dvorak et al., 1999; Eliceiri et al., 1999) by promoting the β-arrestin2-dependent endocytosis of vascular endothelial (VE)-cadherin (Gavard and Gutkind, 2006).
Increased expression of VEGF in mucosal endothelial cells has been demonstrated in IBD (Griga et al., 1998, 1999, 2002; Kanazawa et al., 2001; Chidlow et al., 2006; Sandor et al., 2006). Griga et al. (1998) reported increased serum VEGF level in patients with active UC and CD but not in patients with inactive disease. In a subsequent study, the same group identified the colonic mucosa as the source of the increased serum level of VEGF (Griga et al., 1999, 2002). Kanazawa et al. (2001) reported increased VEGF level in the serum and colonic tissue of patients with active UC. Likewise, Bousvaros et al. (1999) showed that VEGF-A serum levels in children and young adults are elevated during active CD. Furthermore, patients with CD who responded to infliximab (anti-TNF-α antibody) treatment showed a rapid and sustained reduction in VEGF serum levels (Di Sabatino et al., 2004). It is surprising that little is known about VEGF levels in experimental models of colitis. In our previous study, we found increased levels of VEGF in the early stages of iodoacetamide- and trinitrobenzosulfonic acid-induced UC in rats (Sandor et al., 2006). Chidlow et al. (2006) showed significant up-regulation of VEGF-A to D genes in distal colonic tissues from the CD4+CD45RBhigh mouse model of colitis.
Despite clinical reports indicating increased levels of VEGF in IBD, the mechanistic role of VEGF expression in the pathogenesis of UC in vivo remains unclear. The most important question, whether inactivation (neutralization) of elevated endogenous VEGF will affect experimental UC, has not been tested before. In this study, we examined whether neutralization of endogenous VEGF by anti-VEGF antibody to rats with experimental UC will reduce increased vascular permeability, leukocyte infiltration, and severity of morphologic and clinical features of colitis.
Materials and Methods
Animals. Female Sprague-Dawley rats (170-200 g) were obtained from Harlan (Indianapolis, IN) and housed in the animal research facility at the Veterans Affairs Medical Center (Long Beach, CA) under standard environmental conditions. All animals had unlimited access to Purina chow and tap water. They were randomly divided into groups of five rats each. These studies were approved by the Subcommittee for Animal Studies of the Research and Development Committee of the Veterans Affairs Medical Center in Long Beach, CA.
Iodoacetamide-Induced Colitis Model. Experimental UC was induced in rats by the sulfhydryl alkylator iodoacetamide (Satoh et al., 1997). This model using iodoacetamide has several advantages over other models of chemically induced UC. A single dose of 6% iodoacetamide is sufficient to induce well reproducible colonic lesions, with the initial manifestations (e.g., increased vascular permeability, massive mucosal edema) seen 1 to 2 h after iodoacetamide enema, leading to erosions and ulcers (6-12 h), followed by extensive acute and chronic inflammation (7-14 days). Thus, tissue samples may be obtained for biochemical and molecular biologic studies before the development of morphologic lesions. The pathogenesis of sulfhydryl alkylator-induced colonic lesions is relatively well understood. Iodoacetamide produces lesions by several sequential pathways, involving both endothelial and epithelial components. After depletion of endogenous protective glutathione in the mucosa, these chemicals alkylate cellular proteins and, hence, initiate irreversible cell damage in colonic mucosa, resulting in increased vascular permeability, formation of erosions and ulcers, and acute and chronic inflammation.
In brief, 0.1 ml of 6% iodoacetamide (Sigma-Aldrich, St. Louis, MO), dissolved in 1% methylcellulose (Sigma-Aldrich) or the vehicle 1% methylcellulose, was given once by enema (7 cm from anus) via rubber catheter. Rats were euthanized 0.5, 1, 2, 6, and 24 h and 3, 7, and 14 days after intracolonic administration of iodoacetamide.
Treatment with Neutralizing Anti-VEGF Antibody. To evaluate the effects of VEGF neutralization on severity of experimental UC, we used neutralizing mouse monoclonal anti-VEGF antibody (Ab-3) (Lab Vision, Fremont, CA), which was generated against recombinant human VEGF-A and neutralizes bioactivity of all VEGF-A isoforms (Gupta et al., 1999). To demonstrate that this antibody interacts with rat colonic tissue, we have performed Western blot with total proteins isolated from rat colon.
After iodoacetamide enema rats were divided into three groups (n = 5 in each group). The first group was given neutralizing mouse monoclonal anti-VEGF antibody (50 μg/rat i.m.), the second group received control mouse IgG (50 μg/rat i.m.) (Lab Vision), and the third group received saline (0.1 ml/rat i.m.) on the 3rd and 5th days after iodoacetamide enema. Rats were euthanized on the 7th day. Clinical parameters such as body weight changes, lethargy, and diarrhea were recorded during the treatment period, as described by Szabo et al. (1998) (Table 1). The experiments were repeated two times, and data were pooled.
Macroscopic and Histologic Evaluation of Colonic Lesions. Rats were euthanized by CO2 inhalation and cervical dislocation. The colon (7 cm from anus) was removed, opened longitudinally, and rinsed with saline. The colonic lesion areas (millimeters squared), colonic dilatation (millimeters), and loss of rugae (millimeters squared) were measured. The wet weights of standardized colonic specimens were also recorded and expressed as milligrams per 100 g of body weight. To assess histologic damage, full-thickness colonic tissue samples were fixed in 10% buffered formalin, embedded in paraffin, sectioned, stained with hematoxylin and eosin, and coded for blind microscopic measurement of colonic lesions. Histologic slides were examined by two experienced pathologists who were unaware of the treatment. The histologic evaluation was performed according to the scoring system described before (Satoh et al., 1997). Mucosal scrapings of the colon were harvested and frozen (-80°C) for the study of gene and protein expressions.
Measurement of In Vivo Microvascular Permeability. For the quantitative studies of colonic vascular permeability, we used Evans blue, which binds to albumin. Its leakage reflects increased vascular permeability of macromolecules (Rogers et al., 1989). Rats were anesthetized with inhalation of isoflurane, and Evans blue (0.4 mg/100 g in PBS; Sigma-Aldrich) was injected intravenously 15 min before autopsy; rats were killed at 0.5 h after 6% iodoacetamide enema or vehicle (1% methylcellulose). The abdomen was opened, and 7 cm of colon was removed, rinsed in saline, gently blotted with filter paper, and weighed. Evans blue was extracted from the tissue using chloroform and measured by spectrophotometry at 610 nm as described by Baluk et al. (1999). Results were expressed as milligrams of dye per gram wet weight of colon.
Pretreatment. To determine the role of VEGF on early vascular permeability during experimental UC, we injected intramuscularly anti-VEGF antibody (Ab-3; 100 μg/rat), vehicle mouse IgG (100 μg/rat) (Lab Vision), or saline 1 h before iodoacetamide enema. To determine whether the mechanism of VEGF-induced vascular permeability is mediated via the Src-dependent pathway (Eliceiri et al., 1999; Chou et al., 2002; Gavard and Gutkind, 2006), we also investigated the effect of the Src family selective tyrosine kinase inhibitor PP1 (BIOMOL Research Laboratories, Plymouth Meeting, PA), which was shown to prevent VEGF-induced vascular permeability in cardiac and neural tissue (Paul et al., 2001; Weis et al., 2004). Based on the effective dose of PP1 on infarct volumes in vivo (Paul et al., 2001), rats were treated subcutaneously with 0.2 mg/100 g PP1 or vehicle (0.1 ml of dimethyl sulfoxide) 1.5 h before iodoacetamide enema. Rats were euthanized 0.5 h after iodoacetamide.
Real-Time PCR. The total RNA was isolated from frozen colonic tissue by TRIzol reagent (Invitrogen, Carlsbad, CA) and purified by using an RNA isolation kit (Clontech, Mountain View, CA) according to the manufacturer's instructions. Expression profiles of the VEGF gene were examined by using an iQ5 real-time PCR detection system (Bio-Rad iCycler Real-Time PCR instrument; Bio-Rad, Hercules, CA), as in our previous study (Khomenko et al., 2006). The level of target gene mRNA measured by threshold cycle number was compared with GAPDH, which was used as an internal control to correct variability in starting mRNA concentration. The fold changes of treated groups over the control group were calculated.
Western Blotting. Total proteins (100 μg), which were extracted from colonic mucosa in a lysis buffer containing protease inhibitors (Sigma-Aldrich), were processed routinely for Western blot as described previously (Khomenko et al., 2006). The primary antibodies were used against VEGF-A (C-1), PDGF-B (H-55), bFGF (147), β-arrestin2 (H-9), VE-cadherin (H-72), and actin (1:500; Santa Cruz Biotechnology, Inc., Santa Cruz, CA); and Src and phospho-Src (Tyr416) (1:1000; Cell Signaling Technology Inc., Danvers, MA). The loading controls were performed by using a mouse monoclonal antibody to GAPDH (1:2000; EnCor Biotechnology Inc., Alachua, FL). Every Western blot was repeated at least two times using proteins from three different rats per group.
Enzyme-Linked Immunosorbent Assay. The rat VEGF immunoassay kit for cell and tissue lysate (RayBiotech, Inc., Norcross, GA) and serum (RayBiotech, Inc.) were used according to the manufacturer's instruction for the measurement of VEGF in colonic mucosa and serum, respectively. We calculated the concentration at a ratio of endogenous VEGF versus total proteins (picograms per milligram).
Immunohistochemistry. Immunostaining was performed using paraffin-embedded 5-μm-thick intestinal sections. Sections were deparaffinized, hydrated, blocked for endogenous peroxidase using 3% H2O2/H2O, and subsequently subjected to microwave antigen retrieval using a Dako target retrieval solution (BD Pharmingen, San Diego, CA) at pH 10.00. An overnight incubation with primary antibodies rabbit polyclonal anti-von Willebrand factor (1:100; Millipore Bioscience Research Reagents, Temecula, CA), rabbit polyclonal anti-VEGF-A (147) (1:200; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), and mouse monoclonal anti-CD34 (QBEnd/10), anti-CD45 (2B11&PD/76), and anti-CD-138 (B-A38) (Ventana Medical Systems, Tucson, AZ) at 4°C was followed by biotinylated secondary antibodies and a peroxidase-labeled streptavidin-biotin and then developed using the ABC detection method for examination under a Nikon microscope (Nikon, Tokyo, Japan). To ensure specificity of the antibody, immunoabsorption of the antibodies was performed to provide controls.
Morphometric Analysis of Colonic Microvasculature. Computerized morphometric analysis was used for quantification of angiogenesis. In brief, immunostained colonic sections with anti-von Willebrand factor or anti-CD34 were examined under a Nikon microscope at low power (40×) to detect the most vascularized area in the granulation tissue of the ulcer base. At least five microphotographs at 200× magnification of the mucosa were obtained per each specimen. The number of vessels/field (expressed as mean vascular density) was measured using the MetaMorph 7.0 video image analysis system (Molecular Devices, Sunnyvale, CA).
Analysis of Number Leukocytes in Colonic Mucosa. Immunostained colonic sections with anti-CD45 and anti-CD138 antibodies were used for counting lymphocytes and plasma cells. Neutrophils were counted on hematoxylin and eosin-stained slides. Leukocytes were counted on coded slides by two independent investigators experienced in histopathology. Colonic sections were examined under a Nikon microscope at low power (40×) to study the lesion area beneath the necrotic tissue layer, i.e., the number of leukocytes was counted at higher (400×) magnification in 10 fields of the granulation tissue and in the adjacent areas.
Statistical Analysis. Quantitative results are expressed as mean ± S.D. The statistical significance was determined by the nonparametric Mann-Whitney U test or Student's t test where appropriate, and p values < 0.05 were considered statistically significant.
Results
VEGF Protein and mRNA in Iodoacetamide-Induced UC in Rats. As shown in Fig. 1A, VEGF mRNA expression in colonic mucosa was increased 1.7-fold (p < 0.01) at 1 h after iodoacetamide. Maximal elevation was observed at 2 h (3.5-fold, p < 0.05). VEGF protein levels were significantly increased as early as 0.5 h after iodoacetamide enema and remained elevated during disease progression, e.g., at 24 h, necrotic ulcers with acute inflammatory cell infiltration were seen. Gradual replacement of necrotic tissue in the ulcer bed by granulation tissue (rich in newly formed blood vessels) from the 3rd to 14th days was associated with the second pick of VEGF up-regulation. (Fig. 1, B and C). We observed similar changes in the activity of Src (phosphorylation of Tyr416) but not in the expression of the Src protein family after 0.5 h of iodoacetamide enema. Activated Src remained elevated during disease progression. From the 3rd to 14th days, the phosphorylation of Src was associated with VEGF up-regulation (Fig. 1D).
We next examined the cell types producing VEGF in inflamed colonic mucosa using immunohistochemical staining. In normal colon, VEGF was predominantly localized to the endothelial cells of mucosal and submucosal blood vessels. In inflamed colonic mucosa (6 h after iodoacetamide enema), VEGF staining was increased versus normal mucosa; in addition, epithelial cells and numerous infiltrating leukocytes were also stained for VEGF (Fig. 2).
Effect of Neutralizing Anti-VEGF Antibody on Clinical and Morphologic Features of Experimental UC. To determine the effect of VEGF neutralization on experimental UC, rats were treated with neutralizing anti-VEGF antibody, vehicle mouse IgG, or saline. During iodoacetamide-induced colitis, well formed necrotic lesions, extending into the submucosa and muscle layers, develop within 1 to 3 days. Therefore, we injected the first dose of anti-VEGF antibody or vehicle on the 3rd day after iodoacetamide enema. At that time, rats with experimental colitis had watery diarrhea (scale, 1.5-3), loss of body weight of approximately 10 to 15%, and histologic examination of colonic tissue showed subacute stages of colitis. Administration of the anti-VEGF antibody improved clinical signs of experimental UC. Namely, 77% of rats from the saline-treated group and 70% from the IgG-treated group still had diarrhea on the 7th day, with scores of 1.1 and 0.9, respectively, compared with 50% of anti-VEGF-treated rats that had diarrhea with a score of 0.6. These changes were also reflected in the lethargy scores, e.g., animals in the anti-VEGF group demonstrated a score of 0.6 versus 1.1 and 0.7, respectively, in saline- and IgG-treated groups.
To examine whether neutralization of VEGF affects the extent of colonic injury, we measured the area of macroscopic mucosal ulcers. As shown in Fig. 3, VEGF-neutralizing antibody significantly reduced the size of colonic lesions (p < 0.001 versus saline; p < 0.01 versus control IgG), colonic dilatation (p < 0.01 versus saline; p < 0.05 versus IgG), loss of colonic rugae (p < 0.001 versus saline; p < 0.01 versus IgG), and ratio of colon wet weight/100 g of body weight (p < 0.01 versus saline; p < 0.05 versus IgG).
Histologic evaluation revealed that rats treated with either saline or IgG after iodoacetamide enema had extensive ulcers, infiltration of acute and chronic inflammatory cells, i.e., neutrophils, lymphocytes, and plasma cells, and no or minimal mucosal regeneration. Rats treated with the anti-VEGF antibody had only small or no colonic ulcers, with minimal inflammation, and granulation tissue was partly covered by regenerating mucosa (Fig. 3F).
Effect of Neutralizing Anti-VEGF Treatment on Colonic Levels of VEGF, bFGF, and PDGF in Iodoacetamide-Induced UC. Because up-regulation of other potent angiogenic growth factors such as bFGF and PDGF was also detected in IBD, we examined whether the reduction in colonic injury by anti-VEGF antibody treatment is associated with changes in levels of other angiogenic growth factors. As shown in Fig. 4, VEGF, bFGF, and PDGF protein levels in colonic mucosa of the IgG-treated control group were similar to those seen in the saline-treated control group during iodoacetamide-induced UC. In contrast, the anti-VEGF-treated group had significantly reduced VEGF, bFGF, and PDGF levels.
Effect of Anti-VEGF Treatment on Blood Vessel Density in the Colon during Experimental UC. We used immunohistochemical staining for anti-von Willebrand factor, a highly selective endothelial cells marker to assess angiogenesis. Despite the marked differences in morphologic signs of UC after anti-VEGF treatment, we did not find any significant changes in blood vessel density among groups. Namely, the saline-treated group had 9 ± 5, the IgG-treated group had 10 ± 2, and the anti-VEGF-treated group had 10 ± 2 vessels/field in the ulcer base. Because not all endothelial cells may synthesize the von Willebrand factor, a small percentage of blood vessels might not stain with this antibody. To exclude this possible technical error, we also performed immunohistochemical staining for anti-CD34. We obtained similar results.
The Role of VEGF in Increased Vascular Permeability in the Pathogenesis of Experimental UC. VEGF is a potent vascular permeability factor that contributes to development of edema and swelling in many diseases, such as myocardial infarction (Weis et al., 2004), stroke (Paul et al., 2001), and age-related macular degeneration (Scheppke et al., 2008). However, it is unknown whether elevated VEGF can mediate vascular permeability in colon. We next studied whether the VEGF plays a role in increased vascular permeability during iodoacetamide-induced UC in rats. In our study, pretreatment of rats with the neutralizing anti-VECF antibody but not vehicle-control IgG or saline had significantly reduced vascular permeability (p < 0.05, versus saline or IgG) (Fig. 5A), indicating for the first time mediation by VEGF of early colonic vascular permeability.
Src(-/-) deficiency or blockade of Src activity decreased VEGF-induced vascular permeability in the brain (Paul et al., 2001) and retina (Scheppke et al., 2008). Therefore, the Src-dependent mechanism was implicated as a key signal pathway for VEGF-induced vascular permeability in these tissues. Gavard and Gutkind (2006) recently demonstrated that the endpoint of this signaling pathway is the β-arrestin2-dependent endocytosis of VE-cadherin, thereby disrupting the endothelial barrier function.
We investigated whether Src-dependent downstream mechanism mediates VEGF-induced colonic vascular permeability during experimental UC. In our study, we observed enhanced tyrosine phosphorylation of Src in colonic mucosa after iodoacetamide enema that was associated with increased interaction of β-arrestin2 with VE-cadherin, which was detected by Western blot with coimmunoprecipitation (Fig. 5B).
To investigate whether Src-dependent downstream mechanism mediates VEGF-induced colonic vascular permeability during experimental UC, we examined the effect of the Src inhibitor PP1. Pretreatment with PP1 1.5 h before iodoacetamide enema had significantly reduced Evans blue extravasation (p < 0.01, versus vehicle) (Fig. 5C), indicating reduction of increased vascular permeability. The plasma levels of VEGF after iodoacetamide enema have increased by approximately 50% (p < 0.01, versus vehicle-methylcellulose group). Pretreatment with PP1 did not change this level (Fig. 5D).
Effect of Anti-VEGF Treatment on Number of Leukocytes in the Colon during Experimental UC. VEGF stimulates the adhesion of leukocytes via increased expression of adhesion molecules and by increasing vascular permeability promotes tissue infiltration by these cells. We examined the number of acute (neutrophils) and chronic inflammatory cells (lymphocytes and plasma cells) in lesion areas and in adjacent tissue after anti-VEGF treatment. As shown in Fig. 6, VEGF neutralization markedly reduced the number of neutrophils (p < 0.001 versus saline; p < 0.003 versus control IgG) and chronic inflammatory cells (p < 0.001 versus saline; p < 0.02 versus control IgG) in the area of colonic lesions. The number of inflammatory cells in adjacent nonulcerated areas was not different between these groups.
Discussion
The present study demonstrated, for the first time, an important causal role of VEGF in the pathogenesis of UC, especially in its early phase. The anti-VEGF antibody significantly attenuated the clinical and morphologic features of iodoacetamide-induced colitis. This effect was accompanied by decreased expression of not only VEGF but also two other potent angiogenic growth factors, such as bFGF and PDGF. Moreover, treatment with the anti-VEGF antibody significantly decreased colonic vascular permeability and reduced the number of acute and chronic inflammatory cells in the ulcer base. These findings provide compelling evidence for the importance of an elevated level of VEGF in the pathogenesis of UC.
VEGF is the most potent and endothelial-specific angiogenic growth factor, but VEGF also has proinflammatory properties. In addition to its well known role as a potent vascular permeability factor (Senger et al., 1986), VEGF accelerates inflammatory cells adhesions via up-regulation of adhesion molecules expression (Barleon et al., 1996; Goebel et al., 2006). Clinical and experimental studies demonstrated elevated serum and tissue levels of VEGF/VPF in patients with active UC, implicating it (by association) in the pathogenesis of this disease (Griga et al., 1998, 1999, 2002; Kanazawa et al., 2001; Chidlow et al., 2006; Tsiolakidou et al., 2008). However, the mechanistic role of VEGF/VPF in UC was uncertain.
Here, we demonstrated a strong causal association between increased VEGF expression and progression of experimental UC. Chidlow et al. (2006) found in the dextran sulfate sodium colitis model that increased angiogenic activity was associated with active inflammation, and it was diminished during tissue restitution and repair. Increased VEGF protein levels can be because of up-regulation of mRNA expression and an increased number of inflammatory cells infiltrating ulcerated mucosa. The increased leukocyte infiltration is a characteristic feature of UC and CD, with their contributions to disease initiation and subsequent tissue damage (Abreu, 2002). Activated monocytes and/or macrophages alone are sufficient to induce angiogenesis in the avascular cornea (Koch et al., 1986). Moreover, Griga et al. (1999) demonstrated that peripheral blood mononuclear cells of both active CD patients and active UC patients produced significantly higher amounts of VEGF (versus healthy volunteers). In addition, there was a significantly increased VEGF production by peripheral blood mononuclear cells of patients with active disease compared with peripheral blood mononuclear cells of patients with quiescent CD. In our study, we also found strong positive staining for VEGF in leukocytes in inflamed colonic tissue, whereas in normal tissue, VEGF was mostly localized to the endothelial cells. In inflamed colonic tissue, in addition to leukocytes, VEGF was also highly expressed by epithelial cells. Our observation is consistent with data (Cane et al., 2007) showing that activation of decay accelerating factor signaling (part of the lipopolysaccharide-induced receptor complex) leads to up-regulation of VEGF in human colonic epithelial cells.
To determine directly a possible mechanistic role of VEGF in the pathogenesis of UC, we neutralized VEGF with the specific anti-VEGF antibody. Treatment with the anti-VEGF antibody had a beneficial effect on the course of iodoacetamide-induced UC, reducing diarrhea and lethargy scores and the macroscopic and microscopic pathologic changes in the colon.
Neutralization of VEGF caused reduction of not only VEGF levels but also PDGF and bFGF levels. It is more likely that these results might be explained by a decreased number of infiltrated inflammatory cells, which express and release VEGF, bFGF, and PDGF in the pathogenesis of IBD. Moreover, we found a significantly decreased number of leukocytes in lesion areas after anti-VEGF therapy. Matsuura et al. (2005) found that rectal administration of human recombinant bFGF to normal mice significantly increased expression of VEGF in colonic tissue, but expression of VEGF in mice with experimental UC after bFGF treatment was lower than in the nontreated mice with colitis in a dose-dependent manner and directly correlated with a beneficial dose of bFGF.
VEGF is a potent vascular permeability factor, 50,000 times more potent than histamine. In fact, VEGF is unique in this regard because other growth factors (e.g., bFGF, PDGF) can induce neovascularization but do not induce vascular permeability (Dvorak et al., 1999; Weis et al., 2004). Moreover, exogenous VEGF administration during embryonic vasculogenesis or VEGF overexpression in various tissues in transgenic animals results in increased angiogenesis but with malformed leaky vessels with unusually large irregular lumina (Drake and Little, 1995; Thurston et al., 1999). Thus, VEGF is a “capricious” molecule, with a narrow range between its effective concentrations promoting normal or pathologic angiogenesis.
We proposed that the beneficial effect of anti-VEGF therapy in the pathogenesis of experimental UC is because of attenuation of VEGF-induced vascular permeability, which results in reduced vascular leakage and inflammatory cell infiltration. In our study, pretreatment with the neutralizing anti-VEGF antibody significantly attenuated vascular permeability. We also showed for the first time that VEGF-induced colonic vascular permeability is Src-dependent. Pretreatment with the Src inhibitor PP1 almost prevented iodoacetamide-induced vascular permeability without affecting serum levels of VEGF. It was especially interesting that the total Src protein expression (Western blot) did not change during this experimental colitis, but Src phosphorylation was markedly increased. Furthermore, we also found an enhanced coimmunoprecipitation of β-arrestin2 with VE-cadherin that represents a Src-dependent downstream mechanism in the VEGF-induced increased vascular permeability.
Goebel et al. (2006) showed on activated mouse colonic endothelial cells that VEGF-A increased adhesions of neutrophils and T cells to these cells and decreased rolling. Because neutralization of VEGF attenuated vascular permeability in colonic mucosa, we expected a decreased number of inflammatory cells in the colonic lesions. Treatment with the anti-VEGF antibody significantly decreased the number of acute (neutrophils) and chronic inflammatory cells (lymphocytes and plasma cells) in lesion areas.
Our studies demonstrate for the first time that VEGF plays a prominent role in the pathogenesis of experimental UC. Up-regulated VEGF increases vascular permeability in colonic mucosa; thus, it facilitates inflammatory cell infiltration at the site of injury and promotes persistent chronic inflammation. Beneficial action of anti-VEGF treatment shown in this study provides a rational for testing this new therapeutic approach in clinical settings, especially that UC patients have increased levels of VEGF.
Acknowledgments
We thank Dr. Maria Dacosta-Iyer and histologist Marie Marcher for technical assistance in histology and immunohistochemistry studies.
Footnotes
-
This work was supported by the Department of Veterans Affairs, Veterans Health Administration Merit Review [Grant VAMR0710-580].
-
This work was previously presented in abstract form: Tolstanova G, Khomenko T, Deng X, Chen L, Tarnawski A, Ahluwalia A, Szabo S, and Sandor Z (2007) Neutralizing anti-vascular endothelial growth factor (VEGF) antibody reduces severity of experimental ulcerative colitis in rats: direct evidence for the pathogenic role of VEGF. Digestive Disease Week 2007; 2007 May 19-24; Washington DC. American Gastroenterological Association, Bethesda, MD.
-
doi:10.1124/jpet.108.145128.
-
ABBREVIATIONS: IBD, inflammatory bowel disease; UC, ulcerative colitis; CD, Crohn's disease; VEGF, vascular endothelial growth factor; bFGF, basic fibroblast growth factor; PDGF, platelet-derived growth factor; VPF, vascular permeability factor; VE, vascular endothelial; PP1, [pyrazolopyrimidine, 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine]; PCR, polymerase chain reaction.
- Received August 20, 2008.
- Accepted December 4, 2008.
- U.S. Government work not protected by U.S. copyright