Advanced glycation end products cause epithelial-myofibroblast transdifferentiation via the receptor for advanced glycation end products (RAGE).

Tubulointerstitial disease, a prominent phenomenon in diabetic nephropathy, correlates with decline in renal function. The underlying pathogenic link between chronic hyperglycemia and the development of tubulointerstitial injury has not been fully elucidated, but myofibroblast formation represents a key step in the development of tubulointerstitial fibrosis. RAGE, the receptor for advanced glycation end products (AGEs), induces the expression of TGF-beta and other cytokines that are proposed to mediate the transdifferentiation of epithelial cells to form myofibroblasts. Here we report specific binding of (125)I-AGE-BSA to cell membranes prepared from a rat proximal tubule cell line and show that the binding site was RAGE. AGE exposure induced dose-dependent epithelial-myofibroblast transdifferentiation determined by morphological changes, de novo alpha smooth-muscle actin expression, and loss of epithelial E-cadherin staining. These effects could be blocked with neutralizing Ab's to RAGE or to TGF-beta. Transdifferentiation was also apparent in the proximal tubules of diabetic rats and in a renal biopsy from a patient with type 1 diabetes. The AGE cross-link breaker, phenyl-4,5-dimethylthiazolium bromide (ALT 711) reduced transdifferentiation in diabetic rats in association with reduced tubular AGE and TGF-beta expression. This study provides a novel mechanism to explain the development of tubulointerstitial disease in diabetic nephropathy and provides a new treatment target.

Tubular inhibition destroys charge selectivity for anionic and neutral horseradish peroxidase.

The fractional clearance of [3H]anionic HRP and [3H]neutral HRP in the isolated perfused kidney as determined by radioactivity analysis was 0.0160+/-0.0028 (n=6) and 0.0388+/-0.0076 (n=8) respectively. The apparent charge selectivity for both the anionic and neutral forms of HRP observed was destroyed with the inclusion of the tubular uptake inhibitors, 150 mM lysine and 10 mM NH4Cl, in the perfusate. In the presence of 150 mM lysine, the clearances of [3H]anionic HRP and [3H]neutral HRP were 0.0645+/-0.0110 (n=4) and 0. 0784+/-0.0120 (n=4) respectively, and 0.0564+/-0.0035 (n=4) and 0. 0694+/-0.0054 (n=4) respectively in the presence of 10 mM NH4Cl. The clearance for both the anionic and neutral HRP probes in these tubular uptake inhibited systems fits precisely the size selective characteristics of the glomerular capillary wall as determined by transport probes calibrated for hydrodynamic size by size exclusion chromatography. The tubular uptake inhibitors were observed not to alter glomerular permselectivity as determined using polydisperse dextran fractions and the behaviour of neutral HRP. This study demonstrates that charge selectivity for differently charged proteins is not as great as originally thought and suggests that the clearance differences between anionic and neutral forms may be due to differential handling by the tubules.

Albuminuria in patients with type 1 diabetes is directly linked to changes in the lysosome-mediated degradation of albumin during renal passage.

Previous studies by our group have shown that albumin is metabolized in rodents during renal passage and excreted in the urine as a mixture of intact protein and albumin-derived fragments. The aim of this study was to examine whether albumin is metabolized during renal passage in nondiabetic volunteers and in type 1 diabetic patients with varying levels of albuminuria. Nine nondiabetic normoalbuminuric volunteers and 11 type 1 diabetic patients with albumin excretion rates varying from normoalbuminuria to macroalbuminuria were studied. Each subject received an intravenous injection of tritium-labeled albumin ([3H]-albumin). Urine was collected at 4 h and 24 h after injection and analyzed by size exclusion chromatography. The amount of intact and fragmented albumin was quantified, and each fraction was analyzed by radioimmunoassay (RIA) for albumin. [3H]-albumin in nondiabetic volunteers was metabolized during renal passage to small peptide fragments not detectable by conventional RIA (only 0.05-3.8% of the total urinary radioactivity was associated with intact albumin). The process responsible for albumin fragmentation was similar in diabetic patients with normoalbuminuria (intact albumin represented 0.01-4.0% of total urinary radioactivity). However, there was a reduction in the fragmentation ratio (fragmented:intact) in diabetic patients with micro- or macroalbuminuria (intact albumin represented 2.7-55.5%, P = 0.048). This change in the fragmentation ratio was directly related to the degree of albuminuria. These results have important implications for understanding the mechanisms underlying albuminuria in nondiabetic volunteers and type 1 diabetic patients. In nondiabetic volunteers, the renal processing of albumin involves a relatively rapid and comprehensive degradation of albumin to small fragments (range 1-15 kDa). The degradation process is inhibited in diabetic nephropathy in proportion to the level of albuminuria detected by RIA.

Prevention of albuminuria by aminoguanidine or ramipril in streptozotocin-induced diabetic rats is associated with the normalization of glomerular protein kinase C.

This study examined whether the prevention of diabetes-related albuminuria by aminoguanidine (AG) or ramipril (RAM) may be mediated by a common post-glomerular basement membrane renal intracellular mechanism involving protein kinase C (PKC). The renal handling of albumin was examined over 24 weeks in control and streptozotocin (STZ)-induced diabetic rats. A radioimmunoassay (RIA) that measures intact albumin, and intravenously injected tritium-labeled rat serum albumin, was used to assess the proportion of intact albumin and albumin fragments in urine. Diabetes was induced in male Sprague-Dawley rats by the intravenous administration of STZ at a dose of 50 mg/kg. Age-matched control rats received buffer alone. Diabetes was characterized by an increase in blood glucose (>15 mmol/l), an increase in GHb (means at 24 weeks 29.3+/-1.1%; control 6.1+/-0.1%, P<0.005), an increase in glomerular filtration rate (GFR) (4.13+/-0.15 ml/min; control 3.54+/-0.19 ml/min, P<0.005), an increase in intact albumin excretion rate (expressed as geometric mean 11.64 times/divided by 2.11 mg/24 h; control 0.74 times/divided by 1.57 mg/24 h, P<0.005) as measured by RIA, and an increase in glomerular PKC activity (26.83+/-2.38 pmol x mg(-1) x min(-1); control 14.6+/-2.99 pmol x mg(-1) x min(-1), P<0.005). Treatment of diabetic rats with either AG or RAM prevented the rise in intact albuminuria and glomerular PKC activity. Renal lysosomal cathepsin activity decreased in diabetic rats and this was not prevented by AG or RAM. Neither drug affected glycemic control or GFR, but RAM reduced systolic blood pressure (BP), whereas AG did not. These data indicate that urinary excretion of intact albumin and albumin-derived fragments in diabetes may be modulated independently of glycemic control (AG and RAM) and systolic BP (RAM). While both drugs are known for their different mechanisms of action, the fact that both prevent diabetes-related increases in glomerular PKC activity and albuminuria supports the hypothesis that PKC plays a central role in the development of diabetic nephropathy.

The normal kidney filters nephrotic levels of albumin retrieved by proximal tubule cells: retrieval is disrupted in nephrotic states.

The origin of albuminuria remains controversial owing to difficulties in quantifying the actual amount of albumin filtered by the kidney. Here we use fluorescently labeled albumin, together with the powerful technique of intravital 2-photon microscopy to show that renal albumin filtration in non-proteinuric rats is approximately 50 times greater than previously measured and is followed by rapid endocytosis into proximal tubule cells (PTCs). The endocytosed albumin appears to undergo transcytosis in large vesicles (500 nm in diameter), identified by immunogold staining of endogenous albumin by electron microscopy, to the basolateral membrane where the albumin is disgorged back to the peritubular blood supply. In nephrotic rats, the rate of uptake of albumin by the proximal tubule (PT) is decreased. This is consistent with reduced expression of clathrin, megalin, and vacuolar H(+)-ATPase A subunit, proteins that are critical components of the PT endocytotic machinery. These findings strongly support the paradigm-shifting concept that the glomerular filter normally leaks albumin at nephrotic levels. Albuminuria does not occur as this filtered albumin load is avidly bound and retrieved by PTCs. Dysfunction of this retrieval pathway leads to albuminuria. Thus, restoration of the defective endocytotic and processing function of PT epithelial cells might represent an effective strategy to limit urinary albumin loss, at least in some types of nephrotic syndrome.

Protein degradation during renal passage in normal kidneys is inhibited in experimental albuminuria.

1. Tritium labelled proteins, namely bovine serum albumin ([3H]BSA), rat serum albumin ([3H]RSA), anionic horseradish peroxidase ([3H]aHRP) and immunoglobulin present in urine fractions from rat filtration studies in vivo and isolated perfused rat kidneys (IPKs) have been shown by gel chromatographic analysis to be severely degraded to small peptides. The degradation of RSA and BSA in vivo has been shown to be similar. 2. Degradation of proteins in the urine from IPK experiments was inhibited by including 150 mmol/l lysine in the perfusate. Similarly, [3H]BSA and [3H]aHRP excreted from rats with puromycin aminonucleoside nephrosis was again essentially intact for both IPK and in vivo experiments. 3. It appears that the degradation of proteins observed in urine obtained from control kidneys is due, in part, to proteolytic activity associated with the proximal tubule. Inhibition of proximal tubule function, which occurs for both lysine and puromycin aminonucleoside treatments (as calibrated by lysozyme uptake), results in inhibition of the degradation observed. Glomerular epithelial cells could also contribute to the degradation. 4. There was no generation of low-molecular-weight material in the perfusate or plasma arising from breakdown of circulating proteins or recycling of potential degradation products from the tubules.

Glomerular charge selectivity for anionic and neutral horseradish peroxidase.

Studies in isolated perfused rat kidney have demonstrated that it exhibits apparently normal charge selectivity and tubular uptake of anionic horseradish peroxidase (aHRP; pI < 4.0) and neutral horseradish peroxidase (nHRP; pI = 7.5) when these proteins are measured for their enzyme activity. The absolute fractional clearance values for aHRP and nHRP were 0.006 +/- 0.002 and 0.041 +/- 0.007, respectively. It is evident, however, that the enzyme assay for horseradish peroxidase severely underestimates the quantity of protein in urine as compared to measurement of its tritium labeled form through radioactivity. Fractional clearances estimated by radioactivity and corrected for tubular reabsorption for [3H]aHRP and [3H]nHRP were 0.040 +/- 0.029 and 0.099 +/- 0.043, respectively, compared to those estimated by enzyme activity which were 0.012 +/- 0.004 and 0.070 +/- 0.037, respectively. While charge selectivity between the anionic and neutral forms of HRP was still evident, albeit significantly reduced, the major feature of this type of analysis is that the clearance of the aHRP protein is significantly increased compared to that determined by enzyme assay. This difference correlates with the observation that the aHRP protein is markedly degraded (61 to 65%), as determined by gel chromatography, during filtration. Similar degradation was seen in urine fractions collected after the aHRP protein was administered in vivo. Degradation also occurred for the nHRP protein in both the perfused kidney and in vivo but to a far lesser extent (approximately 14 to 21%). These studies demonstrate that the anionic form of HRP was preferentially degraded during filtration and that charge selectivity for differently charged proteins is not as marked as originally thought.

Fractional clearance of high molecular weight proteins in conscious rats using a continuous infusion method.

BACKGROUND: The purported existence of "large pores" in the glomerular capillary wall has been derived primarily from studies using dextrans and Ficolls. Systematic studies using high molecular weight proteins have not been performed. One of the difficulties is that recent studies have demonstrated that albumin and other proteins undergo degradation during renal passage. Our study took into account this renal degradation in measuring the fractional clearance of various high molecular weight proteins (the hydrodynamic radii range was between 48 to 70 A). METHODS: Fractional clearances of tritium-labeled proteins were measured using ALZET osmotic pumps, which are designed to release a slow continuous infusion of tracer. Blood and urine collections were taken at 24-hour intervals over seven days and were counted for radioactivity, and glomerular filtration rate was measured by a creatinine assay. RESULTS: Steady-state levels of [3H]protein in plasma were obtained by day 6. The [3H]proteins in the plasma showed no degradation. The fractional clearances (mean +/- sd, N = 5) of the various proteins were albumin (radius = 36 A; 0.0023 +/- 0.0009), transferrin (48 A; 0.0046 +/- 0.0007), lactoperoxidase (58 A; 0. 0045 +/- 0.0005), immunoglobulin G (62 A; 0.0043 +/- 0.0009), lactate dehydrogenase (64 A; 0.0041 +/- 0.0009), and glucose oxidase (70 A; 0.0036 +/- 0.0011). CONCLUSIONS: These values suggest a weak dependence of fractional clearance on size-selective filtration, except for albumin, which undergoes a specific type of postglomerular processing. The fractional clearances were higher than expected from previous data on dextrans and Ficolls of equivalent hydrodynamic radius, and thus demonstrate that "large pores" may already exist in normal glomerular capillary walls.

Puromycin aminonucleoside nephrosis results in a marked increase in fractional clearance of albumin.

Puromycin aminonucleoside nephrosis (PAN) results in a marked increase in the fractional clearance of albumin. The increase in the fractional clearance of [(3)H]albumin to approximately 0.045, as measured both in vivo and in the isolated perfused rat kidney (IPK) with PAN, occurs without an accompanying equivalent increase in glomerular capillary wall size selectivity as previously measured with dextrans. This is very similar to the marked increase in albuminuria seen with kidneys treated with inhibitors of endocytosis by the tubular epithelium, particularly lysine (T. M. Osicka, L. M. Pratt, and W. D. Comper. Nephrology 2: 199-212, 1996). The similarity is further established that, like in the presence of lysine, [(3)H]albumin excreted in urine from rats with PAN is essentially intact whereas, in both in vivo and IPK control experiments, excreted [(3)H]albumin is heavily degraded. The same observations have also been made for (3)H-labeled anionic horseradish peroxidase. These observations suggest that the significant albuminuria that occurs in PAN is primarily post-glomerular basement membrane in origin.

A comparison of the plasma disappearance of iohexol and 99mTc-DTPA for the measurement of glomerular filtration rate (GFR) in diabetes.

BACKGROUND: Changes in glomerular filtration rate (GFR) provide a valuable indicator of the progression of diabetic nephropathy. GFR is most commonly measured by the plasma clearance of radioisotopes, however, use of iohexol, a non-ionic radiocontrast medium, is a recently described alternative and has shown good agreement with inulin clearance. A one-compartment model is used for calculating GFR in most Australian centres but a two-compartment model is more accurate. AIMS: To set up a non-radioisotopic method for assessment of GFR using iohexol, and to compare this with the currently used 99mTc-diethylene-triamine-penta-acetic acid (DTPA) method. Secondly, to compare GFR results using an unmodified one-compartment model with a one-compartment model subjected to the Brochner-Mortensen modification. METHODS: Twenty-one patients with diabetes had assessment of GFR with simultaneous measurements of 99mTc-DTPA and iohexol plasma clearance. Plasma clearance was determined by the slope intercept method and then modified according to the Brochner-Mortensen equation. Plasma iohexol concentrations were determined by capillary electrophoresis. RESULTS: There was no significant difference between iohexol and 99mTc-DTPA derived GFR values, difference 4.3+/-7.7 mL/minute (mean+/-SD). This was despite 99mTc-DTPA protein binding demonstrated in the range of 5-10%. Comparison of GFR results using an unmodified one-compartment model with a Brochner-Mortensen corrected one-compartment model showed higher GFR values with the former, in the range of 20-30% for GFR values > 100 mL/minute. CONCLUSION: Iohexol provides an efficient alternative to radioisotopic methods for serial measurement of GFR in diabetic patients with hyperfiltration, incipient and overt nephropathy. A one-compartment model with its inherent overestimation of GFR should be replaced by the Brochner-Mortensen modified one-compartment model.

Fractional clearance of albumin is influenced by its degradation during renal passage.

1. The fractional clearance of intact albumin as determined by fractionation of urine by gel chromatography gave a value of 3.9 +/- 1.6 x 10(-4) for the isolated perfused kidney and 2.1 +/- 0.6 x 10(-4) in vivo using ALZET osmotic pumps. 2. Albumin fractional clearance as measured by detection of the tritium label on the albumin molecule by radioactivity analysis gave a value of 7.5 +/- 3.9 x 10(-3) for the isolated perfused kidney and 2.3 +/- 0.9 x 10(-3) in vivo. 3. The major differences between assays that detect intact albumin compared with non-specific assays in the estimates of the fractional clearance of albumin can be explained by the degradation of approx. 90% of albumin to small peptides during its renal passage. This has been demonstrated by size exclusion chromatography of urine samples from experiments where (i) exogenous tritium-labelled albumin was used in isolated perfused kidneys, (ii) exogenous tritium-labelled albumin was administered intravenously and (iii) analysis was made with metabolically labelled endogenous albumin in vivo.

Ramipril and aminoguanidine restore renal lysosomal processing in streptozotocin diabetic rats.

AIMS/HYPOTHESIS: We aimed to examine the time course for the diabetes-related changes in renal lysosomal processing and to determine whether these changes can be prevented by aminoguanidine or ramipril treatment. METHODS: The percentage desulphation of intravenously injected tritium labelled dextran sulphate ([3H]DSO4) in the urine, as determined by ion-exchange chromatography, was used as a marker of lysosomal sulphatase activity. Sulphatase activity was determined 1, 2, 3 and 4 weeks after the onset of diabetes in rats as well as in rats treated with either aminoguanidine or ramipril for twelve weeks. RESULTS: The amount of totally desulphated [3H]DSO4 in urine collected from control rats was 65.6 +/- 0.8%. This was significantly reduced in diabetic rats two (57.4 +/- 1.4% desulphated), three (56.8 +/- 1.3 % desulphated) and four (52.9 +/- 2.2% desulphated) weeks after the onset of diabetes. The significant decrease in the amount of totally desulphated [3H]DSO4 in the urine also found at 12 weeks after the onset of diabetes was not affected by drug treatment. There was no significant difference in the amount of partially desulphated [3H]DSO4 in the urine between all the study groups. However, the increase in totally sulphated [3H]DSO4 in the urine collected from diabetic rats (8.7 +/- 1.7 % sulphated) compared with that of control rats (2.2 +/- 0.5% sulphated) was normalised by treatment with both aminoguanidine (4.8 +/- 1.6% sulphated) or ramipril (4.5 +/- 0.8% sulphated). CONCLUSIONS/INTERPRETATION: These results raise the possibility that the diabetes-induced changes in renal lysosomal processing may be one of the initial events in the development of diabetic nephropathy. Aminoguanidine and ramipril, known for their different mechanism of action, seem to prevent diabetes-induced changes in lysosomal processing either through their effects on enzyme activity within the lysosome or through their effects on the trafficking of molecules to and from the lysosome.

Aminoguanidine and ramipril prevent diabetes-induced increases in protein kinase C activity in glomeruli, retina and mesenteric artery.

This study investigated the effects of insulin therapy, inhibition of advanced glycation end-product formation with aminoguanidine and angiotensin-converting enzyme inhibition with ramipril on diabetes-related increases in protein kinase C (PKC) activity in the streptozotocin-diabetic rat. PKC activity in the glomeruli, retina and mesenteric artery was increased by 1.5-2-fold after induction of diabetes, and this increase was maintained over 24 weeks. Treatment with insulin at 2 units or 6 units per day attenuated glomerular PKC in proportion to the level of glycohaemoglobin after 4 weeks of diabetes (r=0.68, P<0.0001). The higher dose of insulin prevented the diabetes-related increase in glomerular PKC activity, although blood glucose levels were not normalized. After 8 weeks of diabetes, ramipril completely prevented the diabetes-related increases in PKC activity in the glomeruli, retina and mesenteric artery. By contrast, aminoguanidine treatment resulted in no inhibition of glomerular PKC activity, partial inhibition of retinal PKC activity and complete inhibition of mesenteric artery PKC activity. After 24 weeks of diabetes, both aminoguanidine and ramipril prevented the diabetes-related increases in PKC activity in all three tissues, in parallel with suppression of albuminuria by both agents. Aminoguanidine also prevented diabetes-related increases in retinal permeability at 16 weeks. These results suggest that the organ-protective effects of insulin, aminoguanidine and ramipril in diabetes may be mediated, at least in part, through the differential inhibition of PKC activity in various tissues.

Albumin interaction with the glomerular capillary wall in vitro.

The binding of albumin to the glomerular capillary wall was studied using albumin-gold in perfused kidneys, the interaction of [3H]albumin with isolated glomeruli at 37 degrees C and 4 degrees C and the interaction at [3H]albumin with purified basement membrane. The albumin-gold was found to bind predominantly to the basement membrane and this interaction could be dissociated with high concentrations of albumin. There was binding of albumin to isolated rat glomeruli which exhibited temperature dependence. Glomeruli exhibited a binding site at both 37 degrees C and 4 degrees C with an association constant in the range of 1 to 3 x 10(4) M-1 that bound 7 x 10(13) molecules/glomerulus. At 37 degrees C, however, there was anomalous Scatchard binding behaviour at relatively higher concentrations of albumin (30 to 50 mg/ml) which could be due to either glomerular cell uptake or the appearance of multiple binding sites or both. The binding of albumin to isolated glomeruli and the glomerular albumin levels in isolated kidney perfusion could largely be accounted for by the binding of albumin to the glomerular basement membrane. The albumin binding to glomeruli at 37 degrees C was enhanced by Pronase digestion and heparinase digestion, but remained unchanged following trypsin treatment or neuraminidase treatment. Similarly, albumin was shown to bind to purified basement membrane preparations. This binding was also enhanced (approximately 80 times) by heparinase digestion but remained unchanged after digestion with chondroitinase ABC or hyaluronidase.(ABSTRACT TRUNCATED AT 250 WORDS)

The return of glomerular filtered albumin to the rat renal vein--the albumin retrieval pathway.

BACKGROUND: Recent studies have demonstrated that the normal glomerular capillary wall (GCW) is not charge selective to albumin. This means that albumin flux across the GCW is high. This has been confirmed in studies where albumin uptake by the tubules has been inhibited. Therefore, there must be a high capacity postglomerular retrieval pathway in normal kidneys that returns filtered albumin back to the blood supply. METHODS: This study identifies the presence of glomerular filtered albumin in the renal vein from the analysis of the decrease of radioactivity in the venous effluent after the injection of a pulse of tritium labeled albumin into the renal artery in vivo and in the isolated perfused kidney (IPK). RESULTS: The glomerular filtered albumin is returned to the blood supply by a high capacity pathway that transports this albumin at a rate of 1830+/-292 microg/min rat kidney (n= 14) (mean+/-SEM). This pathway has been identified under physiological conditions in vivo and in the IPK. The pathway is specific for albumin as it does not occur for horseradish peroxidase (HRP). The pathway is inhibited in a non-filtering kidney. The pathway is also inhibited by NH4Cl, an inhibitor of protein uptake. CONCLUSIONS: The high capacity retrieval pathway for albumin is most likely associated with transtubular cell transport. It is also apparent that most albuminuric states could be accounted for by the malfunctioning of this pathway without resorting to any change in glomerular permselectivity.

The return of glomerular-filtered albumin to the rat renal vein.

BACKGROUND: Recent studies have demonstrated that the normal glomerular capillary wall (GCW) is not charge selective to albumin. This means that albumin flux across the GCW is high, and this has been confirmed in studies in which albumin uptake by the tubules has been inhibited. Therefore, there must be a high-capacity postglomerular retrieval pathway in normal kidneys that returns filtered albumin back to the blood supply. METHODS: This study identifies the presence of glomerular-filtered albumin in the renal vein from the analysis of the decrease of radioactivity in the venous effluent after the injection of a pulse of tritium-labeled albumin into the renal artery in vivo and in the isolated perfused kidney. RESULTS: The postglomerular filtered albumin is returned to the blood supply by a high-capacity pathway that transports this albumin at a rate of 1830 +/- 292 micrograms/min.rat kidney (N = 14, mean +/- SEM). This pathway has been identified under physiological conditions in vivo and in the isolated perfused kidney. The pathway is specific for albumin, as it does not occur for horseradish peroxidase. The pathway is inhibited in a nonfiltering kidney. The pathway is also inhibited by ammonium chloride (an agent that inhibits tubular protein uptake but does not alter glomerular size selectivity) and by albumin peptides (which compete for the tubular albumin receptor). CONCLUSIONS: The high-capacity retrieval pathway for albumin is most likely associated with transtubular cell transport. It is also apparent that most albuminuric states could be accounted for by the malfunctioning of this pathway without resorting to any change in glomerular permselectivity.

Angiotensin converting enzyme inhibition reduces retinal overexpression of vascular endothelial growth factor and hyperpermeability in experimental diabetes.

AIMS/HYPOTHESIS: Angiotensin converting enzyme (ACE) inhibition has been recently suggested to have retinoprotective actions in diabetic patients but the mechanism of this effect is not known. In vitro, angiotensin II stimulates expression of vascular endothelial growth factor (VEGF), a permeability-inducing and endothelial cell specific angiogenic factor which has been implicated in the pathogenesis of diabetic retinopathy in humans and in experimental animals. We sought to determine the effects of ACE inhibition on retinal VEGF expression and permeability in experimental diabetic retinopathy. METHODS: Streptozotocin-induced diabetic rats and control animals were assigned at random to receive ACE inhibitor treatment or vehicle. At 24 weeks the retinal VEGF protein gene expression was assessed by northern blot analysis and in situ hybridisation. Retinal permeability to albumin was measured using a double isotope technique. RESULTS: Experimental diabetes was associated with cell specific two to fourfold increase in retinal VEGF protein gene expression (p < 0.01) and a 2-fold increase in retinal vascular permeability to albumin (p < 0.01). The localization of VEGF expression in the retina was not altered in animals with experimental diabetes. Angiotensin converting enzyme inhibitor treatment of diabetic rats reduced diabetes-associated changes in VEGF gene expression and vascular permeability. CONCLUSION/INTERPRETATION: These findings implicate the renin-angiotensin system in the VEGF overexpression and hyperpermeability which accompany diabetic retinopathy and provide a potential mechanism for the beneficial effects of ACE inhibition in diabetic retinal disease.