Vascular Endothelial Growth Factor Is Essential for ... - Semantic Scholar

J Am Soc Nephrol 12: 1734 –1741, 2001

Vascular Endothelial Growth Factor Is Essential for Hyperglycemia-Induced Structural and Functional Alterations of the Peritoneal Membrane AN S. DE VRIESE,* RONALD G. TILTON,† CLIFFORD C. STEPHAN,† and NORBERT H. LAMEIRE* *Renal Unit, Ghent University, Belgium; and †Department of Pharmacology, Texas Biotechnology Corporation, Houston, Texas.

Abstract. Long-term peritoneal dialysis is associated with the development of functional and structural alterations of the peritoneal membrane. Long-term exposure to the high glucose concentrations in conventional peritoneal dialysate has been implicated in the pathogenesis of peritoneal hyperpermeability and neoangiogenesis. Vascular endothelial growth factor (VEGF) is an endothelial-specific growth factor that potently stimulates microvascular permeability and proliferation. High glucose exposure upregulates VEGF expression in various cell types and tissues. This study investigated whether VEGF plays a pathogenetic role in hyperglycemia-induced microvascular dysfunction in the peritoneal membrane. The peritoneal microcirculation of streptozotocin-induced diabetic rats and agematched controls was studied in vivo with a combination of functional and morphologic techniques. The diabetic microcirculation was characterized by an elevated transport of small

solutes, indicating the presence of an increased effective vascular surface area. The leakage of FITC-albumin was more rapid in diabetic vessels, suggesting hyperpermeability for macromolecules. Structurally, an increased vascular density with focal areas of irregular capillary budding was found in the diabetic peritoneum. The hyperglycemia-induced structural and functional microvascular alterations were prevented by long-term treatment with neutralizing anti-VEGF monoclonal antibodies, whereas treatment with isotype-matched control antibodies had no effect. VEGF blockade did not influence microvascular density or macromolecular leakage in control rats, demonstrating specificity for the hyperglycemia-induced alterations. The present results thus support an causative link among high glucose exposure, upregulation of VEGF, and peritoneal microvascular dysfunction.

Long-term peritoneal dialysis is associated with the progressive development of functional and structural alterations of the peritoneal membrane. Longitudinal reports suggest that small solute transport tends to increase and ultrafiltration capacity tends to decrease with time on dialysis (1,2), consistent with an increase of the peritoneal vascular surface area. Morphologic changes in the peritoneal microvasculature include thickening and reduplication of the basement membrane and extensive neovascularization (3–5). Because this vasculopathy is reminiscent of the microvascular abnormalities observed in diabetes, the potential causative role of the high glucose concentrations present in conventional peritoneal dialysate solutions has become a contentious question. Two widely known characteristics of early diabetic microvascular dysfunction are hyperpermeability to macromolecules and neovascularization (6,7). Abnormal passage of plasma proteins across the endothelium and their deposition in the

basement membrane has been suggested to contribute to the basement membrane thickening typical of diabetic microangiopathy, although other mechanisms, including increased synthesis and decreased degradation of matrix proteins, may play a role as well (8). Whereas hyperglycemia is recognized as pivotal in the development of these microvascular complications, the nature of the pathogenetic link between high ambient glucose concentrations and microvascular alterations remains a matter of debate, attesting to the complexity of the processes involved rather than to a lack of candidate mechanisms. Vascular endothelial growth factor (VEGF), also known as vascular permeability factor, is a heparin-binding angiogenic growth factor that displays a high specificity for endothelial cells. VEGF plays a prominent role in physiologic and pathologic angiogenesis (9,10). In addition, VEGF dramatically increases endothelial cell permeability and is established as one of the most potent endothelial permeabilizing agents identified thus far (11). Other biologically relevant actions of VEGF include an increase in blood flow and vascular conductance (12) and modulation of leukocyte kinetics (9,10). There is compelling evidence that endothelium-derived nitric oxide acts as a downstream mediator for VEGF (12,13). Considering the cardinal features of hyperglycemia-induced microvascular dysfunction, VEGF is an attractive candidate to provide a pathogenetic link between high glucose exposure and

Received November 8, 2000. Accepted January 22, 2001. Correspondence to Dr. An De Vriese, Renal Unit, University Hospital, OK12, De Pintelaan 185, B-9000 Ghent, Belgium. Phone: ⫹32-9-2405301; Fax: ⫹32-9-2404599; E-mail: [email protected] 1046-6673/1208-1734 Journal of the American Society of Nephrology Copyright © 2001 by the American Society of Nephrology

J Am Soc Nephrol 12: 1734 –1741, 2001

the development of microvascular hyperpermeability and neoangiogenesis. In vitro exposure to high glucose concentrations rapidly induces VEGF expression in several cell types (14 –16). Increased VEGF expression has been demonstrated in the retina, kidney, and nerve fibers of humans and experimental animals with diabetes (17–21). Peritoneal generation of VEGF has been observed in patients treated with glucose-based dialysate solutions (22). Vascular dysfunction induced by topical application of elevated glucose levels in a granulation skin chamber was attenuated by administration of neutralizing VEGF antibodies (Ab) (23). Although these studies present circumstantial evidence for a pathogenetic role of VEGF in glucose-induced microvascular alterations, direct evidence is lacking. The purpose of this study was to characterize the hyperglycemia-induced functional and structural alterations in the peritoneal microvasculature of the rat. In addition, a potential causative role for VEGF in these alterations was investigated by treating the experimental animals with a neutralizing monoclonal Ab specific for VEGF.

Materials and Methods

VEGF Mediates the Vascular Effects of High Glucose

1735

the Ab and the characterization of neutralizing activity were performed as described previously (23). One mg of the anti-VEGF Ab or isotype-matched control Ab was injected intraperitoneally three times per week, starting 2 d after induction of diabetes or the injection of citrate buffer until the final experiments 6 wk later. To determine the efficiency of the anti-VEGF treatment, we obtained serum samples at baseline and after 2, 4, and 6 wk of treatment with anti-VEGF or with control Ab. The samples were drawn from the tail vein just before the next Ab administration. Binding studies were performed using a fusion protein composed of the seven-loop ectodomain of fms-like tyrosine kinase fused to the heavy chains of a mouse IgG2a Ab. The fusion protein was captured onto Immulon 4 strip wells with IgG2a-specific goat anti-mouse Ab (Sigma, St. Louis, MO). Increasing concentrations of monoclonal anti-VEGF Ab (used as standards) and increasing dilutions of serum (1:50, 1:100, and 1:250) were incubated separately with 4 ng/ml 125I-VEGF (Biomedical Technologies Inc., Stoughton, MA) for 30 min at 37°C before adding to the receptor. Nonspecific binding was defined as the binding measured in the presence of a 100-fold molar excess of unlabeled VEGF. Incubations were terminated after 60 min at 22°C by washing the wells once with 400 ␮l of ice-cold PBS, then twice more with 200 ␮l of ice-cold PBS. Bound VEGF was measured in a gamma spectrometer (Life Technology Inc., Schaumburg, IL).

Laboratory Animals The studies were performed in 60 female Wistar rats (Iffa Credo, Brussels, Belgium) that received care in accordance with the national guidelines for animal protection. Diabetes was induced by intravenous injection of streptozotocin (65 mg/kg; Pfanstiel, Davenham, UK) dissolved in citrate buffer (n ⫽ 36). Slow-release insulin implants (Linshin, Scarborough, Canada), with a release rate of 1 U/24 h, were used to maintain moderate hyperglycemia. Age-matched control rats (n ⫽ 24) received an intravenous injection of citrate buffer solution and placebo implants composed of palmitic acid. The procedures were performed under anesthesia with halothane (Fluothane; Zeneca, Destelbergen, Belgium). Experiments were carried out 6 wk after the injection of streptozotocin or citrate buffer. Diabetic rats were untreated (n ⫽ 13), treated with anti-VEGF Ab (n ⫽ 11), or treated with isotype-matched control Ab (n ⫽ 12). Control rats were either untreated (n ⫽ 13) or treated with anti-VEGF Ab (n ⫽ 11). In these experimental groups, either peritoneal transport studies (n ⫽ 29) or intravital microscopy studies (n ⫽ 31) were performed.

Anti-VEGF Ab Treatment Monoclonal anti-VEGF Ab and isotype-matched control Ab were prepared as described previously (23). Briefly, female 8-wk-old BALB/c mice (Harlan Sprague Dawley, Inc., Indianapolis, IN) were immunized, then boosted 3 times, 21 d apart, by intraperitoneal and subcutaneous injections of 50 ␮g of rhVEGF165, emulsified with an equal volume of Complete Freund’s Adjuvant for the primary immunization and Incomplete Freund’s Adjuvant for secondary immunizations. The mouse with the highest serum titer to rhVEGF165 as measured by enzyme-linked immunosorbent assay was injected intravenously with an additional 30 ␮g of immunogen in phosphatebuffered saline (PBS), 21 d after the last immunization. Three d later, spleen cells were harvested for production of hybridomas to rhVEGF165. The hybridoma cell line with the highest Ab titer and neutralizing Ab activity was selected after cloning 3 to 4 times by limiting dilution in 96-well microtiter plates, then grown in a Cellmax Bioreactor (Spectrum, Rancho Dominguez, CA) using Dulbecco’s modified Eagle’s medium culture media. Purified Ig was prepared by protein A chromatography. The isotype and light chain composition of

Peritoneal Transport Studies Rats were anesthetized with thiobutabarbital (Inactin; RBI, Natick, MA; 100 mg/kg subcutaneously). The trachea was intubated, a jugular vein was cannulated for continuous infusion of isotonic saline, and a carotid artery was cannulated for blood sampling. The saline infusion rate was matched with diuresis to maintain euvolemia. After 30 min, a silicon catheter (Venflon; Becton Dickinson, Erembodegem-Aalst, Belgium) was inserted in the abdomen, and 15 ml of 3.86% glucose peritoneal dialysate solution (Dianeal; Baxter, Nivelles, Belgium) was infused. Plasma and dialysate samples were collected at t ⫽ 0, 30, 60, and 120 min for determination of creatinine, urea, and glucose levels. Fructosamine and total protein levels were determined on the first plasma sample only. Dialysate cultures were obtained at the end of the experimental dwell, and animals were excluded from analysis if cultures were positive. After 120 min, the abdomen was opened by midline incision for collection of the dialysate fluid and for tissue sampling. The transport of low molecular weight solutes was evaluated by calculating the mass transfer area coefficient (MTAC) of urea and creatinine, using the Garred equation (24):

MTAC ⫽ volumeout/dwell time ⫻ ln[volumein ⫻ concplasma/volumein ⫻ (concplasma ⫺ concdialysateend)] The initial peritoneal concentration of urea and creatinine is set at 0. The Garred formula is a simplified approach to calculate MTAC, assuming that the reflection coefficient of the solute is 0 and that the average solute concentration in the membrane equals the plasma concentration. The magnitude of the transport of small solutes is determined by the effective vascular surface area, which is dependent on the number of perfused peritoneal capillaries (25).

Intravital Microscopy Rats were anesthetized with thiobutabarbital and cannulated as described above. Cromoglycate (cromolyn sodium salt, 10 mg/kg intravenously; Sigma) was administered 15 min before surgery, to block degranulation of mast cells induced by the surgical manipulation. A small midline abdominal incision was made, and a short

1736

Journal of the American Society of Nephrology

J Am Soc Nephrol 12: 1734 –1741, 2001

segment of the small bowel was exteriorized, carefully avoiding stretching. The visceral peritoneum was spread over a plexiglass plate and superfused continuously with an isotonic, isocolloidal solution (Hemaccel; Hoechst Marion Roussel, Marburg/Lahn, Germany) maintained at 37°C. The preparation was allowed to stabilize for 30 min after completion of surgery. Observations were made with an Axiotech Vario 100 HD microscope (Zeiss, Jena, Germany) using water immersion objectives (Achroplan 10⫻, 40⫻, 63⫻). The microscopic stage was driven by a stepping motor control MCL-2 (Lang, Hu¨ttenberg, Germany), operated by a joystick or a software program (Wincommander; Ma¨rzha¨user-Wetzlar, Wetzlar, Germany) via an RS-232 interface. The tissue was transilluminated via a fiberoptic using a light source (KL 1500; Schott, Wiesbaden, Germany) equipped with a 150-W halogen lamp. Epifluorescence was performed with a mercury lamp HBO 50 W and a FITC filter set (excitation filter BP 450 to 490, dichroic mirror FT 510, emission filter LP 520). The resulting image was displayed on a television monitor by a TK-1281 camera (Victor Company of Japan LTD-JVC, Tokyo, Japan) and recorded by a video recorder (S-VHS Panasonic AG-7355, Matsushita, Japan) for off-line analysis. All automatic gain controls were switched off during the experiments. The video images were digitized with an IP-8/AT Matrox image processing board and analyzed with image analysis software (Cap-Image; Ingenieurbu¨ro Zeintl, Heidelberg, Germany) (26). The technique allows the study of the peritoneal microcirculation, including small arteries, arterioles, capillaries, postcapillary venules, venules, and small veins (27). Larger vessels are surrounded by fat and are not routinely visualized. For the evaluation of macromolecular leakage, a venular segment with a diameter of 20 to 40 ␮m and an unbranched length of approximately 150 ␮m without any other vessels in the immediate vicinity was selected for study. FITC-bovine serum albumin (FITC-albumin, 50 mg/kg; Sigma) was administered as an intravenous bolus. When traumatic leaks were observed, the experiment was discontinued. Otherwise, epifluorescence recordings were made every 10 min for 120 min. On the digitized image of the venule under study, two intraluminal areas and two contiguous areas of perivenular interstitium were defined. The average gray scale value, ranging from 0 for black to 255 for white, was calculated for each area. As the molecule leaves the circulation, the intraluminal gray scale value falls and the perivascular gray scale value rises. Macromolecular leakage was defined as the ratio between the average gray scale value within the venule (Gv) and the average gray scale value in the perivenular interstitium (Gi) (28). To evaluate microvascular density, we inspected all segments of the visceral peritoneum of the jejunum and ileum and the images were recorded. To avoid selection bias, we performed quantification in a predefined segment. The objective (10⫻) was positioned at random in

the third segment of the distal ileum proximal from the caecum. With the aid of the Wincommander software, the microscopic stage was driven through a meander consisting of five steps of 1 mm in the X direction and five steps of 1 mm in the Y direction. The microscopic image was recorded at each of these 36 positions. The vessel length per area was determined for each microscopic image, and the average was calculated. Capillaries, defined as vessels with a luminal diameter ⬍5 ␮m (27), were analyzed separately, and their relative contribution to total vessel length was calculated.

Statistical Analyses The data are presented as mean ⫾ SEM. ANOVA and unpaired t tests were used as appropriate to test statistical significance. The significance level was set at P ⬍ 0.05.

Results Characteristics of Laboratory Animals Diabetic animals had higher plasma glucose and fructosamine levels as compared with the age-matched control rats (Table 1). There were no differences in metabolic control between Ab-treated and untreated diabetic rats. Body weights were lower in diabetic rats as compared with age-matched control rats. There were no differences in body weight between the diabetic groups. There were no differences in BP among the experimental groups. The characteristics of the animals used for peritoneal transport studies and those used for the intravital microscopy studies of the same treatment group were similar.

In Vitro Serum VEGF-Inhibitory Activity In the rats that were treated with anti-VEGF or control Ab, VEGF inhibitory activities were measured at 0, 2, 4, and 6 wk of treatment, after dilution of the serum at 1:50, 1:100, and 1:250. For the 1:50 dilution, baseline VEGF inhibitory activity was low and not different between experimental groups (Figure 1). An increased serum inhibitory activity was found at 2, 4, and 6 wk of treatment with anti-VEGF Ab in both diabetic and control rats as compared with baseline values and with diabetic rats that were treated with control Ab. At 4 and 6 wk, inhibitory activity was somewhat lower in anti-VEGF Ab-treated control rats as compared with anti-VEGF Ab-treated diabetic rats. Because the inhibitory activity of undiluted serum of antiVEGF Ab-treated rats is almost 100%, the clinical relevance of this difference may be minimal. As expected, the inhibitory

Table 1. Clinical and biochemical characteristics of the experimental groupsa Experimental Groups (n ⫽ 60)

Body Weight (g)

BP (mmHg)b

Plasma Glucose (mmol/L)

Plasma Fructosamine (␮mol/g total protein)

Control (n ⫽ 13) Diabetes (n ⫽ 13) Diabetes⫹anti-VEGFAb (n ⫽ 11) Diabetes⫹controlAb (n ⫽ 12) Control⫹anti-VEGFAb (n ⫽ 11)

289.1 ⫾ 4.3 259.3 ⫾ 5.4c 258.6 ⫾ 7.0c 260.7 ⫾ 5.1c 285.9 ⫾ 4.3

125.7 ⫾ 2.5 (n⫽7) 123.6 ⫾ 3.6 (n⫽7) 124.3 ⫾ 7.4 (n⫽6) 125.6 ⫾ 6.6 (n⫽6) 118.2 ⫾ 6.9 (n⫽5)

8.58 ⫾ 0.66 20.91 ⫾ 2.16c 21.03 ⫾ 1.87c 22.28 ⫾ 1.6c 7.86 ⫾ 0.96

1.87 ⫾ 0.25 4.14 ⫾ 0.33c 4.03 ⫾ 0.29c 4.15 ⫾ 0.13c 1.86 ⫾ 0.16

a

VEGF, vascular endothelial growth factor; Ab, antibody. BP was recorded only in rats that were studied by intravital microscopy. c P ⬍ 0.001 versus age-matched controls. b

J Am Soc Nephrol 12: 1734 –1741, 2001


1737

Peritoneal Transport Studies Small solute transport was higher in 6-wk diabetic rats than in age-matched controls, indicating the presence of a larger effective vascular surface area in the diabetic peritoneum (Figure 2). The increase in effective vascular surface area was prevented by treatment with anti-VEGF Ab but not by control Ab. Administration of anti-VEGF Ab did not significantly affect the MTAC urea and creatinine in control rats.

Intravital Microscopy: Leakage of FITC-Albumin

Figure 1. Serum vascular endothelial growth factor (VEGF) inhibitory activities in antibody (Ab)-treated rats. Serum inhibitory activities in diabetic rats ⫹ anti-VEGF Ab (n ⫽ 11, ⵧ), control rats ⫹ anti-VEGF Ab (n ⫽ 11, p), and diabetic rats ⫹ control Ab (n ⫽ 12, ■) before and 2, 4, and 6 wk after the start of treatment. Inhibitory activities are expressed as percentage of inhibition by 1:50 diluted serum of the binding between 125I-VEGF and fms-like tyrosine kinase in vitro. Comparison among time points: *, P ⬍ 0.001 versus 0 wk. Comparison among groups: #, P ⬍ 0.001 versus diabetic ⫹ anti-VEGF Ab and control ⫹ anti-VEGF Ab; §, P ⬍ 0.001 versus diabetic ⫹ anti-VEGF Ab.

activities at 1:100 and 1:250 dilution of the serum were lower than those at 1:50 dilution but followed a similar pattern (data not shown).

The leakage of FITC-albumin was quantified as the change of Gv/Gi during the observation period. Macromolecular leakage was elevated in the 6-wk untreated and control Ab-treated diabetic rats (Figure 3). The hyperpermeability for albumin was prevented by treatment with anti-VEGF Ab. After 90 min of observation, however, macromolecular leak was somewhat more pronounced in anti-VEGF Ab-treated diabetic rats than in untreated controls and in anti-VEGF Ab-treated controls but was still significantly lower than in untreated diabetic and control Ab-treated diabetic rats. VEGF blockade did not affect the permeability for FITC-albumin in control rats.

Intravital Microscopy: Vascular Density Intravital microscopy of the visceral peritoneum of control rats revealed a progressively increasing microvascular density from the proximal jejunum toward the distal ileum. In untreated and control Ab-treated diabetic rats, this gradient was not present; irregularly arranged and dense vascular networks were visible in all segments of the visceral peritoneum (Figure 4). In addition, disseminated focal areas of intense capillary proliferation that lacked uniform spacing were observed, generally along the borders of the fat axes that surround the larger

Figure 2. Peritoneal transport studies. The mass transfer area coefficient (MTAC) for urea (ⵧ) and creatinine (o) was measured after a 2-h dwell of 3.86% glucose dialysate in the peritoneal cavity in control rats (n ⫽ 6), control rats ⫹ anti-VEGF Ab (n ⫽ 6), diabetic rats (n ⫽ 6), diabetic rats ⫹ anti-VEGF Ab (n ⫽ 5), and diabetic rats ⫹ control Ab (n ⫽ 6). *, P ⬍ 0.05 versus control rats; **, P ⬍ 0.05 versus diabetic rats and diabetic rats ⫹ control Ab.

1738


J Am Soc Nephrol 12: 1734 –1741, 2001

Quantification of the vascular density with image analysis software revealed an increased vessel/area ratio in the untreated diabetic and control Ab-treated diabetic rats but not in the anti-VEGF Ab-treated diabetic rats (Figure 6). The increase in vascular density was accounted for primarily by capillary proliferation: the relative contribution of capillaries to total vessel length was elevated substantially in untreated and control Ab-treated diabetic rats (Figure 6). VEGF blockade did not reduce total vessel length in control rats, although the percentage of capillaries was slightly lower than in untreated control rats (Figure 6).

Discussion Figure 3. Macromolecular permeability. Leakage of FITC-albumin from the intravascular space into the perivascular interstitium, as defined by the ratio of the gray scale value within the vessel (Gv) and the gray scale value in the interstitium (Gi), in 6-wk control (□), diabetic (䊊), anti-VEGF Ab-treated diabetic (Œ), control Ab-treated diabetic (F), and anti-VEGF Ab-treated control rats (䡵). *, P ⬍ 0.05 versus control for all time points; **, P ⬍ 0.05 versus diabetes and diabetes ⫹ control Ab for all time points; #, P ⬍ 0.05 versus control and control ⫹ anti-VEGF Ab.

mesenteric vessels (Figure 5). These alterations were absent in diabetic rats that were treated with anti-VEGF Ab, and the peritoneal membranes of these rats were similar to those of control rats (Figures 4 and 5).

The peritoneal microcirculation in early streptozotocin-induced diabetes in the rat is characterized by the development of microvascular proliferation, associated on the functional level with an increased transport of small solutes and with an increased leakage of macromolecules. The salient observation made in the present study is that the hyperglycemia-induced microvascular alterations are largely prevented by treatment with a neutralizing monoclonal Ab to VEGF. The present results thus are the first to support a causative role for VEGF in the microvascular dysfunction induced by high ambient glucose concentrations. VEGF is known to induce angiogenesis through its potent stimulating effect on endothelial cell proliferation, migration, and tube formation (9,10). In the diabetic rats, evidence for neoangiogenesis was observed with different techniques. After 6 wk of diabetes, peritoneal vascular density was elevated

Figure 4. Intravital microscopy: microvascular density. In contrast to control peritonea (A), diabetic peritonea are characterized by the presence of dense and irregular vascular networks (B). These alterations were prevented by VEGF-blockade (C) but not by administration of control Ab (D). Magnification, ⫻64.

J Am Soc Nephrol 12: 1734 –1741, 2001


1739

Figure 5. Intravital microscopy: capillary budding. In control rats (A), arterioles divide into a thin capillary network that forms postcapillary venules and venules. In untreated and control Ab-treated diabetic rats (B, D), focal areas of dense irregular capillary budding are observed. Vascular architecture in anti-VEGF Ab-treated diabetic rats (C) is similar as in control rats. Magnification, ⫻250.

substantially, with both an increased density of the mature vascular network and the presence of focal areas of extensive capillary budding. After 1 wk of diabetes, focal capillary proliferation was already present, but the density of the established vascular network was normal (De Vriese AS, unpub-

Figure 6. Microvascular density. Recordings of 36 areas of the visceral peritoneum were made during intravital microscopy, and the total vessel and capillary length per surface area was quantified with image analysis software. The relative contribution of the capillaries is indicated by the open area of the bars. In diabetic rats, total vessel length and the percentage of capillaries were higher than in controls (44.7 ⫾ 4.5% versus 27.8 ⫾ 1.9%; P ⫽ 0.005). The vascular proliferation and increase in the proportion of capillaries was prevented by treatment with anti-VEGF Ab (29.2 ⫾ 2.8%) but not by control Ab (46.9 ⫾ 4.3%). VEGF blockade did not decrease vascular density in control rats, although the proportional contribution of capillaries was slightly lower (20.6 ⫾ 2.4%; P ⫽ 0.04 versus controls). *, P ⬍ 0.01 versus 6 wk controls; **, P ⬍ 0.01 versus diabetes and diabetes ⫹ control Ab.

lished observations), suggesting that the capillary budding precedes the development of vascular networks. Systematic quantification of the recordings made during intravital microscopy revealed that vascular density in the diabetic rats exceeded that observed in control rats and that the relative contribution of capillaries was higher. The functional consequences of the enlarged peritoneal vascular surface area were highlighted by an increased transport of small molecular weight solutes. Administration of anti-VEGF Ab dramatically inhibited the neoangiogenesis in the diabetic rats, with a commensurate effect on the transport of small solutes. These results thus implicate VEGF as a key mediator of hyperglycemia-induced neoangiogenesis. VEGF also is a very potent enhancer of microvascular permeability. The exact mechanisms and structures involved in the permeability-enhancing effect of VEGF are still controversial. Three mechanisms have been implicated: (1) induction of endothelial fenestrations, (2) functional activation of vesicularvacuolar organelles in the cytoplasm of endothelial cells, and (3) stimulation of the formation of interendothelial gaps. It has been proposed that the increase in microvascular permeability to proteins induced by VEGF is an essential step in angiogenesis, allowing the extravasation of blood-borne proteins and the formation of a matrix to support the growth of the endothelial cells and formation of tubes (9). Leakage of macromolecules was studied at the level of the individual microvessel and found to be significantly increased in the diabetic rats, as described previously (29). It should be noted that the present approach does not allow discrimination of the relative contribution of

1740


microvascular hyperpermeability and microvascular hypertension to the development of macromolecular leakage. Blockade of VEGF reduced early macromolecular leakage to values not different from those in control rats, supporting an important role for VEGF. At the end of the observation period, FITCalbumin leakage was somewhat higher in anti-VEGF Abtreated diabetic rats than in control rats, suggesting the involvement of other mechanisms. Oxidative stress (30), advanced glycation end products/receptor for advanced glycation end products interaction (31), and increased activity of the polyol pathway (32) have been implicated as mediators of increased microvascular permeability in diabetes and may have played a contributing role in our model. Application of exogenous VEGF rapidly increases permeability in different tissues (12,13), revealing the presence of functional receptors on quiescent endothelial cells. It has been suggested, therefore, that VEGF is responsible for inducing and maintaining the baseline permeability of the normal microcirculation. The present results do not support a role for VEGF in normal permeability for macromolecules, inasmuch as VEGF blockade did not affect leakage of FITC-albumin in control rats. Intravascular administration of VEGF induces endotheliumdependent vasodilation and results in a reduction of BP (12,33). VEGF is, however, primarily a paracrine mediator, and the significance of systemic VEGF levels is undefined. The presence of VEGF in serum samples is due mainly to VEGF release from platelets upon their activation during coagulation (34). True circulating VEGF levels are actually very low, which is not surprising in view of the extreme potency of the molecule and the dramatic consequences of its actions. We failed to demonstrate any effect of chronic VEGF antagonism on BP, supporting a limited hemodynamic role for endogenous free circulating VEGF. The prevention of hyperglycemia-induced microvascular changes by VEGF blockade suggests an increased expression of, activity of, or sensitivity to this growth factor. An increased expression of VEGF has been demonstrated previously in the target organs for microvascular dysfunction in humans and experimental animals with diabetes (17–21). Local generation of VEGF was reported in the peritoneum of patients who were exposed to glucose-based peritoneal dialysate (22). The cause of the upregulation of VEGF by high glucose remains speculative, but multiple factors may be implicated. It has been suggested that high glucose concentrations, through an increased flux of glucose via the sorbitol pathway, may trigger hypoxia-like alterations in cellular redox status (23). Because tissue hypoxia is a major regulator of VEGF production, the hypoxia-like redox imbalance may inappropriately upregulate VEGF expression. In addition, glucose degradation products (35) and advanced glycation end products (36) are known to induce VEGF expression in vitro. The in vivo relevance of these pathways remains to be demonstrated. An improved understanding of the role of VEGF in peritoneal microvascular alterations induced by high glucose concentrations is likely to be a source of therapeutic advance. Besides the implications for diabetic microvascular disease, the present results may be conducive to the understanding of the

J Am Soc Nephrol 12: 1734 –1741, 2001

pathogenetic role of VEGF during long-term peritoneal dialysis. The functional and structural alterations observed in the peritoneal membrane of the diabetic rats were similar to those observed in chronic peritoneal dialysis patients with loss of ultrafiltration capacity. It should be noted that in diabetes, the peritoneal membrane is exposed to high glucose concentrations from the endothelial side, whereas during peritoneal dialysis the exposure occurs from the mesothelial side. The pathophysiologic consequences may be similar, however, because VEGF specifically targets endothelial cells, as only these cells carry functional VEGF receptors in vivo. Experimental diabetes therefore may be an appropriate in vivo model to study the effects of high dialysate exposure on peritoneal function and structure. In conclusion, the present results indicate that VEGF represents a pivotal link between high glucose concentrations and microvascular hyperpermeability and neoangiogenesis in the peritoneal membrane. VEGF blockade, as currently tested in cancer patients (37), may constitute a novel, potentially useful therapeutic target to limit the microvascular complications induced by high glucose concentrations.

Acknowledgments The authors thank Tommy Dheuvaert, Julien Dupont, Mieke Van Landschoot, Pascale Vogeleere, and Marie-Anne Waterloos for their expert technical assistance. A.D.V. is supported by a grant from the Fonds voor Wetenschappelijk Onderzoek-Vlaanderen (N20/0).

References 1. Davies SJ, Bryan J, Phillips L, Russell GI: Longitudinal changes in peritoneal kinetics: The effects of peritoneal dialysis and peritonitis. Nephrol Dial Transplant 11: 498 –506, 1996 2. Selgas R, Fernandez-Reyes MJ, Bosque E, Bajo MA, Borrego F, Jimenez C, Del Peso G, De Alvaro F: Functional longevity of the human peritoneum: How long is continuous peritoneal dialysis possible? Results of a prospective medium long-term study. Am J Kidney Dis 23: 64 –73, 1994 3. Di Paolo N, Sacchi G: Peritoneal vascular changes in continuous ambulatory peritoneal dialysis (CAPD): An in vivo model for the study of diabetic microangiopathy. Perit Dial Int 9: 41– 45, 1989 4. Dobbie JW, Lloyd JK, Gall CA: Categorization of ultrastructural changes in peritoneal mesothelium, stroma and blood vessels in uremia and CAPD patients. Adv Perit Dial 6: 3–12, 1990 5. Mateijsen MA, van der Wal AC, Hendriks PM, Zweers MM, Mulder J, Struijk DG, Krediet RT: Vascular and interstitial changes in the peritoneum of CAPD patients with peritoneal sclerosis. Perit Dial Int 19: 517–525, 1999 6. Viberti GC: Increased capillary permeability in diabetes mellitus and its relationship to microvascular angiopathy. Am J Med 75: 81– 84, 1983 7. Shami SK, Chittenden SJ: Microangiopathy in diabetes mellitus: II. Features, complications and investigation. Diabetes Res 17: 157–168, 1991 8. Adler S: Structure-function relationships associated with extracellular matrix alterations in diabetic glomerulopathy. J Am Soc Nephrol 5: 1165–1172, 1994 9. Ferrara N: Role of vascular endothelial growth factor in the regulation of angiogenesis. Kidney Int 56: 794 – 814, 1999

J Am Soc Nephrol 12: 1734 –1741, 2001

10. Neufeld G, Cohen T, Gengrinovitch S, Poltorak Z: Vascular endothelial growth factor and its receptors. FASEB J 13: 9 –22, 1999 11. Senger DR, Galli SJ, Dvorak AM, Perruzzi CA, Harvey VS, Dvorak HF: Tumor cells secrete a vascular permeability factor that promotes accumulation of ascites fluid. Science 219: 983– 985, 1983 12. Tilton RG, Chang KC, Lejeune WS, Stephan CC, Brock TA, Williamson JR: Role for nitric oxide in the hyperpermeability and hemodynamic changes induced by intravenous VEGF. Invest Ophthalmol Vis Sci 40: 689 – 696, 1999 13. Wu HM, Huang Q, Yuan Y, Granger HJ: VEGF induces NOdependent hyperpermeability in coronary venules. Am J Physiol 271: H2735–H2739, 1996 14. Natarajan R, Bai W, Lanting L, Gonzales N, Nadler J: Effects of high glucose on vascular endothelial growth factor expression in vascular smooth muscle cells. Am J Physiol 273: H2224 –H2231, 1997 15. Williams B, Gallacher B, Patel H, Orme C: Glucose-induced protein kinase C activation regulates vascular permeability factor mRNA expression and peptide production by human vascular smooth muscle cells in vitro. Diabetes 46: 1497–1503, 1997 16. Sone H, Kawakami Y, Okuda Y, Kondo S, Hanatani M, Suzuki H, Yamashita K: Vascular endothelial growth factor is induced by long-term high glucose concentration and up-regulated by acute glucose deprivation in cultured bovine retinal pigmented epithelial cells. Biochem Biophys Res Commun 221: 193–198, 1996 17. Mathews MK, Merges C, McLeod DS, Lutty GA: Vascular endothelial growth factor and vascular permeability changes in human diabetic retinopathy. Invest Ophthalmol Vis Sci 38: 2729 –2741, 1997 18. Gilbert RE, Vranes D, Berka JL, Kelly DJ, Cox A, Wu LL, Stacker SA, Cooper ME: Vascular endothelial growth factor and its receptors in control and diabetic rat eyes. Lab Invest 78: 1017–1027, 1998 19. Cooper ME, Vranes D, Youssef S, Stacker SA, Cox AJ, Rizkalla B, Casley DJ, Bach LA, Kelly DJ, Gilbert RE: Increased renal expression of vascular endothelial growth factor (VEGF) and its receptor VEGFR-2 in experimental diabetes. Diabetes 48: 2229 – 2239, 1999 20. Tsuchida K, Makita Z, Yamagishi S, Atsumi T, Miyoshi H, Obara S, Ishida M, Ishikawa S, Yasumura K, Koike T: Suppression of transforming growth factor beta and vascular endothelial growth factor in diabetic nephropathy in rats by a novel advanced glycation end product inhibitor, OPB-9195. Diabetologia 42: 579 –588, 1999 21. Samii A, Unger J, Lange W: Vascular endothelial growth factor expression in peripheral nerves and dorsal root ganglia in diabetic neuropathy in rats. Neurosci Lett 262: 159 –162, 1999 22. Zweers MM, de Waart DR, Smit W, Struijk DG, Krediet RT: Growth factors VEGF and TGF-beta1 in peritoneal dialysis. J Lab Clin Med 134: 124 –132, 1999 23. Tilton RG, Kawamura T, Chang KC, Ido Y, Bjercke RJ, Stephan CC, Brock TA, Williamson JR: Vascular dysfunction induced by elevated glucose levels in rats is mediated by vascular endothelial growth factor. J Clin Invest 99: 2192–2202, 1997


1741

24. Garred LJ, Canaud B, Farrell PC: A simple kinetic model for assessing peritoneal mass transfer in chronic ambulatory peritoneal dialysis. ASAIO J 6: 131–137, 1983 25. Rippe B, Stelin G, Haraldsson B: Computer simulations of peritoneal fluid transport in CAPD. Kidney Int 40: 315–325, 1991 26. De Vriese AS, Verbeuren TJ, Vallez MO, Lameire NH, De Buyzere M, Vanhoutte PM: Off-line analysis of red blood cell velocity in renal arterioles. J Vasc Res 37: 26 –31, 2000 27. Wiedeman MP: Dimensions of blood vessels from distributing artery to collecting vein. Circ Res 12: 375–378, 1963 28. Schmidt H, Schmidt W, Muller T, Bohrer H, Gebhard MM, Martin E: N-acetylcysteine attenuates endotoxin-induced leukocyte-endothelial cell adhesion and macromolecular leakage in vivo. Crit Care Med 25: 858 – 863, 1997 29. Shostak A, Gotloib L: Increased peritoneal permeability to albumin in streptozotocin diabetic rats. Kidney Int 49: 705–714, 1996 30. Bonnardel-Phu E, Vicaut E: Reactive oxygen species and acute modulation of albumin microvascular leakage in the microcirculation of diabetic rats in vivo. J Vasc Res 37: 32–38, 2000 31. Wautier JL, Zoukourian C, Chappey O, Wautier MP, Guillausseau PJ, Cao R, Hori O, Stern D, Schmidt AM: Receptormediated endothelial cell dysfunction in diabetic vasculopathy. Soluble receptor for advanced glycation end products blocks hyperpermeability in diabetic rats. J Clin Invest 97: 238 –243, 1996 32. Tilton RG, Chang K, Pugliese G, Eades DM, Province MA, Sherman WR, Kilo C, Williamson JR: Prevention of hemodynamic and vascular albumin filtration changes in diabetic rats by aldose reductase inhibitors. Diabetes 38: 1258 –1270, 1989 33. Horowitz JR, Rivard A, van der Zee R, Hariawala M, Sheriff DD, Esakof DD, Chaudhry GM, Symes JF, Isner JM: Vascular endothelial growth factor/vascular permeability factor produces nitric oxide-dependent hypotension. Evidence for a maintenance role in quiescent adult endothelium. Arterioscler Thromb Vasc Biol 17: 2793–2800, 1997 34. Banks RE, Forbes MA, Kinsey SE, Stanley A, Ingham E, Walters C, Selby PJ: Release of the angiogenic cytokine vascular endothelial growth factor (VEGF) from platelets: Significance for VEGF measurements and cancer biology. Br J Cancer 77: 956 –964, 1998 35. Inagi R, Miyata T, Yamamoto T, Suzuki D, Urakami K, Saito A, van Ypersele de Strihou C, Kurokawa K: Glucose degradation product methylglyoxal enhances the production of vascular endothelial growth factor in peritoneal cells: Role in the functional and morphological alterations of peritoneal membranes in peritoneal dialysis. FEBS Lett 463: 260 –264, 1999 36. Lu M, Kuroki M, Amano S, Tolentino M, Keough K, Kim I, Bucala R, Adamis AP: Advanced glycation end products increase retinal vascular endothelial growth factor expression. J Clin Invest 101: 1219 –1224, 1998 37. Schlaeppi JM, Wood JM: Targeting vascular endothelial growth factor (VEGF) for anti-tumor therapy, by anti-VEGF neutralizing monoclonal antibodies or by VEGF receptor tyrosine-kinase inhibitors. Cancer Metastasis Rev 18: 473– 481, 1999