DIABETIC NEPHROPATHY is a serious complication

Transcription

1 X/00/$03.00/0 Vol. 85, No. 3 The Journal of Clinical Endocrinology & Metabolism Printed in U.S.A. Copyright 2000 by The Endocrine Society Proteolysis of Insulin-Like Growth Factor-Binding Protein-3 Is Increased in Urine from Patients with Diabetic Nephropathy* M. SHINADA, A. AKDENIZ, S. PANAGIOTOPOULOS, G. JERUMS, AND L. A. BACH University of Melbourne, Department of Medicine, and Endocrinology Unit (A.A., S.P., G.J.), Austin and Repatriation Medical Center, Heidelberg, Victoria 3084, Australia ABSTRACT The insulin-like growth factor (IGF) system has been implicated in the development of experimental diabetic nephropathy. IGF-binding protein-3 (IGFBP-3) modulates IGF actions, and proteolysis decreases its binding affinity for IGFs. The aim of this study was to explore the possibility that proteolysis of IGFBP-3 may be altered in diabetic nephropathy and may therefore modify the intrarenal effects of IGFs. IGFBP-3 proteolysis in urine from diabetic patients with normo- [albumin excretion rate (AER), 20 g/min], micro- (AER, g/min), and macroalbuminuria (AER, 200 g/min) was studied in 34 patients with noninsulin-dependent diabetes mellitus (NIDDM), 14 patients with insulin-dependent diabetes mellitus, and 9 controls. Urine samples were analyzed by Western ligand blotting and IGFBP-3 immunoblotting. Protease activity was quantitated using [ 125 I]IGFBP-3 as a substrate. WLB showed three main bands (40 46, 35, and 26 kda) in control urine and a fainter 18-kDa band. All but the 35-kDa band were immunoreactive with the IGFBP-3 Received September 10, Revision received November 23, Accepted December 4, Address all correspondence and requests for reprints to: Dr. Leon A. Bach, University of Melbourne, Department of Medicine (Austin Campus), Austin and Repatriation Medical Center, Studley Road, Heidelberg, Victoria 3084, Australia. bach@austin.unimelb.edu.au. * This work was supported in part by a grant from the Novo-Nordisk Regional Diabetes Support Scheme. antiserum. The same pattern of IGFBPs was seen in urine from normoalbuminuric diabetic patients. However, the urine of diabetic patients with micro- and macroalbuminuria contained little or no intact 40- to 46-kDa IGFBP-3. In patients with noninsulin-dependent diabetes mellitus, urinary IGFBP-3 protease activity in micro- (n 13) and macroalbuminuric patients (n 12; mean SD[SCAP], 75 25% and 84 24%) was significantly higher than that in normoalbuminuric patients (29 9%; P ). Similar results were observed in patients with insulin-dependent diabetes mellitus. Proteolytic activity in diabetic urine was due to a serine protease. In conclusion, diabetic nephropathy was associated with IGFBP-3 proteolysis in urine. As similar changes were not observed in patients sera, this is likely to reflect changes in the kidney or urinary tract, resulting in increased local IGF bioavailability, and therefore may contribute to the structural changes of diabetic nephropathy. (J Clin Endocrinol Metab 85: , 2000) DIABETIC NEPHROPATHY is a serious complication that affects up to 40% of patients with diabetes. Clinically, diabetic nephropathy is characterized by increased urinary albumin excretion, with the subsequent development of renal impairment. Although the mechanisms underlying the development of diabetic nephropathy are incompletely understood, renal hypertrophy is a feature of this condition. Animal studies have suggested that activation of a number of growth factor systems, including the insulin-like growth factors (IGFs), may be involved in the development of diabetic nephropathy (1). IGF-I and IGF-II are widely expressed peptides that are essential for normal development (2). The actions of IGF are modulated by a family of at least six specific, high affinity binding proteins (IGFBPs) (2, 3). At a cellular level, IGFBPs may inhibit or potentiate IGF actions (2, 3). In serum, IGFs are predominantly found within 150-kDa ternary complexes that contain IGFBP-3, an approximately 40-kDa N-glycosylated protein that binds IGFs with high affinity, and an 85-kDa acid-labile subunit (4). IGFs within the 150-kDa complex remain in the vascular compartment, serving as a potential reservoir of IGFs while protecting the host from possible acute insulin-like effects of free IGFs. Circulating IGFs also bind to IGFBPs in binary complexes that are able to leave the vascular compartment. Binary, but not ternary, complexes are found in tissues. Limited proteolysis of IGFBPs has recently emerged as a physiological mechanism by which the IGF binding affinities of IGFBPs are reduced (5). This may lead, in turn, to increased availability of IGFs for binding to IGF-I receptors. A wide range of IGFBP proteases, including serine and aspartic and metalloproteases, have been described. Increased IGFBP-3 proteolysis has been reported in serum from pregnant women and patients with catabolic conditions, including diabetes (6, 7). IGF-I and IGFBP-1 to -6 are all expressed in normal rat kidney in distinct distributions (8, 9). A number of roles for the IGF system in renal development and normal physiology have been described (10). Humans and animals deficient in GH (and therefore IGF-I) have decreased kidney size, glomerular filtration rate, and renal plasma flow, and infusion of IGF-I increases these parameters. The IGF system has been implicated in animal models of diabetic nephropathy (11). Kidney enlargement during the early stage of experimental diabetes is associated with transiently increased IGF-I levels (12, 13), whereas renal IGF-I receptor levels are increased for a more sustained period (14). Changes in levels of renal IGFBPs have also been described in diabetic rats (15 18). However, there have been few stud- 1163

2 1164 SHINADA ET AL. JCE&M 2000 Vol 85 No 3 ies of the role of the IGF system in human diabetic nephropathy. Serum IGF-I levels are not substantially altered in diabetic nephropathy (11), but these may not reflect local renal changes, which may be significant because the IGF system has important paracrine/autocrine actions. Indeed, this is exemplified by diabetic adolescents with microalbuminuria who had higher urinary, but not plasma, IGF-I levels than those with normoalbuminuria (19). As urinary IGFBPs are altered in some, but not all, kidney diseases (20, 21), we hypothesized that the development of diabetic nephropathy might be associated with changes in urinary IGFBPs. We found that intact IGFBP-3 levels were decreased due to increased proteolysis in the urine of patients with diabetic nephropathy. Subjects and Methods Patients and control subjects Urine was examined from diabetic patients attending the Endocrinology Clinic at the Austin and Repatriation Medical Center with noninsulin dependent diabetes mellitus (NIDDM; n 34; age, yr, mean sd), without a urinary tract infection or history of other kidney diseases. The diagnosis of NIDDM was defined by clinical characteristics: age of diagnosis more than 35 yr, no history of ketoacidosis, and no absolute insulin requirement. Nephropathy was classified by albumin excretion rate (AER; normoalbuminuria, 20 g/min; microalbuminuria, g/min; macroalbuminuria, 200 g/min). To be classified as micro- or macroalbuminuric, AERs had to be elevated on at least two of three occasions. We also examined urine from a smaller number of patients with insulin-dependent diabetes mellitus (IDDM; n 14; age, yr) without a urinary tract infection or history of other kidney diseases. The diagnosis of IDDM was defined by a history of ketoacidosis and/or absolute insulin requirement. Nine healthy control subjects (n 9; age, yr) without a urinary tract infection, hypertension, or history of other kidney diseases were also studied. Reagents IGF-II was a gift from Eli Lilly & Co. (Indianapolis, IN). Nonglycosylated recombinant human IGFBP-3 was a gift from Dr. C. Maack, Celtrix Pharmaceuticals, Inc. (Santa Clara, CA). Rabbit antihuman IGFBP-3 antiserum was a gift from Dr. Janet Martin, Sydney University (Sydney, Australia). Biotinylated goat antirabbit antibody and streptavidin-horseradish peroxidase conjugate were purchased from Amersham Pharmacia Biotech (Castle Hill, Australia). Protease inhibitors [ethylenediamine tetraacetate (EDTA), 1,10-phenanthroline, leupeptin, aprotinin, and pepstatin] were purchased from Sigma (St. Louis, MO). Preparation of urine samples First voided morning urine samples or aliquots from 24-h urine collections were centrifuged at 1200 rpm for 10 min at 4 C to separate cellular and other sediments. Supernatants were stored at 20 C for up to 2 weeks before dialysis. Stored urine specimens were thawed slowly and centrifuged at 2500 rpm for 10 min at 4 C, and the pellet was discarded. Supernatants were dialyzed three times against distilled water for 24 h at 4 C using 12,000 14,000 molecular mass cut-off membranes (Union Carbide, Chicago, IL). Aliquots (1 ml) were freeze-dried and stored at 20 C. Iodination of IGF-II and recombinant human IGFBP-3 IGF-II and IGFBP-3 were iodinated by the chloramine-t method to specific activities of and 15 Ci/ g, respectively, as previously described (22). Western ligand blotting (WLB) Dialyzed urine samples (1 ml original volume) were diluted with nonreducing sample buffer (0.5 mol/l Tris, ph 6.8; 10% glycerol; 2% SDS; and 0.05% bromophenol blue) and loaded onto a 0.75-mm discontinuous SDS-12% PAGE at 200 V for h. Electrophoresed proteins were electroblotted onto a nitrocellulose membrane that was blocked with 1% BSA, incubated in buffer containing [ 125 I]IGF-II ( cpm), washed, and exposed to BioMax x-ray film (Eastman Kodak Co., Rochester, NY) for 5 7 days. Western immunoblotting (WIB) WIB was performed as previously described (22), using enhanced chemiluminescence with minor modifications. Dialyzed urine samples were separated by SDS-12% PAGE under nonreducing conditions and then transferred to nitrocellulose membranes as described above. Alternatively, membranes used for WLB were incubated in 10 mmol/l Tris-HCl and 150 mmol/l NaCl (ph 7.4) containing 3% Nonidet P-40 for1hatroom temperature to remove bound ligand. Membranes were blocked in Tris-buffered saline with 5% nonfat dry milk and incubated with a polyclonal rabbit antihuman IGFBP-3 antibody (1:3000). Membranes were then successively incubated with biotinylated antirabbit Ig and streptavidin-horseradish peroxidase conjugate followed by detection using enhanced chemiluminescence (Super Signal, Pierce Chemical Co., Rockford, IL). IGFBP-3 proteolysis assay To further examine the presence of proteolytic activity toward IGFBP-3, urine samples collected from healthy controls and diabetic patients were analyzed according to the method of Lamson et al. (23). Digestion mixtures were made up to a total volume of 20 L. 125 I-Labeled nonglycosylated IGFBP-3 ( 30,000 cpm) was suspended in phosphatebuffered saline with 0.5 mmol/l calcium chloride and incubated for 5 h at 37 C with 2.5 L serum or 50-fold concentrated urine (250 L original volume) in the absence of protease inhibitors or in the presence of EDTA (5 mmol/l), 1,10-phenanthroline (5 mmol/l), aprotinin (50 U/mL), pepstatin (2.5 g/ml), or leupeptin (5 mmol/l). Laemmli sample buffer was added to stop digestion before separation by SDS-12% PAGE. Gels were fixed with 10% acetic acid for 15 min, dried, and subjected to autoradiography for 24 h. The amounts of low molecular mass bands and intact tracer in each lane were determined by laser densitometry (model , Ultro Scan laser densitometer, LKB, Bromma, Sweden). Results were corrected for tracer degradation by subtracting the amount of lower molecular mass bands in control samples containing tracer alone, and were expressed as a percentage of the total optical density per lane that occurred in lower molecular mass bands representing proteolytic cleavage products rather than intact IGFBP-3. Urinary albumin excretion The urinary albumin concentration was measured in aliquots of 24-h urine specimens using immunoturbidimetry (Turbitimer, Behringwerke AG, Marburg, Germany). Intra- and interassay coefficients of variation were 4.5% and 3.0%, respectively (24). Statistical analysis Results are shown as the mean sd. Differences among diabetic patients with normo-, micro-, and macroalbuminuria were compared using one-way ANOVA, followed by Fisher s protected least significant difference test for post-hoc comparisons. Relationships between the extent of IGFBP-3 proteolysis and other variables were examined both by simple regression and by stepwise multiple linear regression. AER data were log-transformed before regression analysis. Results Clinical characteristics of patients with normo-, micro-, and macroalbuminuria Patients with NIDDM were grouped according to degree of albuminuria: normoalbuminuria (n 9; AER, 20 g/

3 IGFBP-3 PROTEOLYSIS IN DIABETIC NEPHROPATHY 1165 min), microalbuminuria (n 13; g/min), and macroalbuminuria (n 12; AER, 200 g/min; Table 1). Groups did not significantly differ in age, duration of diabetes, renal function, blood pressure, or treatment modality. There was a marginal difference between groups in body mass index (P 0.054). The clinical characteristics of 14 patients with IDDM were also grouped according to degree of albuminuria, as shown in Table 2. Patients with macroalbuminuria were older than those in other groups (P 0.02). Patients with nephropathy had a greater prevalence of hypertension (P 0.02). There were nonsignificant trends toward increased duration of diabetes (P 0.07), lower creatinine clearance (P 0.06), and higher systolic blood pressure (P 0.052) in patients with nephropathy, although these findings must be interpreted with caution given the small number of patients. Urinary IGFBPs in control and diabetic patients with and without nephropathy Figure 1A shows a typical pattern of IGFBPs in urine (lanes 1, 2, 4, and 6) and in serum (lanes 3, 5, 7, and 8) of control and diabetic patients as demonstrated by WLB. In control serum (lane 8), the predominant band was kda, with less intense bands of kda. In control urine (lane 1) the predominant band was approximately 35 kda, with other bands of and 26 kda. The patterns of IGFBPs in serum from diabetic patients with normoalbuminuria (lane 3), microalbuminuria (lane 5), and macroalbuminuria (lane 7) were similar to that seen in control serum. The pattern of urinary IGFBPs in normoalbuminuric diabetic subjects was also similar to that in control subjects. In contrast, urine from patients with micro- (lane 4) and macroalbuminuria (lane 6) lacked the 40- to 46-kDa band, and an approximately 18-kDa band was the most predominant IGFBP in the latter sample. To further investigate the nature of the IGFBPs in urine from diabetic subjects, IGFBP-3 in serum and urine samples was assessed by immunoblotting (Fig. 1B). This confirmed that the 40- to 46-, 30-, and 18-kDa bands were forms of IGFBP-3. Urine from micro- and macroalbuminuric patients (lanes 4 and 6) contained less intact 40- to 46-kDa IGFBP-3, whereas levels of 18-kDa IGFBP-3, presumably a fragment, were increased. The 18-kDa IGFBP-3 band was not seen in serum from the same subjects. The 35-kDa band seen by WLB was identified as IGFBP-2 by immunoblotting (results not shown). There was substantial variability in the forms of IGFBP-3 present in the urine of control subjects (Fig. 2A). Nevertheless, 40- to 46-kDa intact IGFBP-3 was present in significant amounts in 5 of 6 samples and was still detectable in the other (lane 4). Substantial amounts of intact IGFBP-3 were also present in 15 of 17 normoalbuminuric diabetic urine samples (e.g. Fig. 2B, lanes 1 and 2), but IGFBP-3 was reduced or absent in all 26 micro- (e.g. lanes 3 and 4) or macroalbuminuric (e.g. lane 5) urine samples. These observations were true regardless of whether patients had NIDDM or IDDM. Urinary IGFBP-3 proteolysis is increased in urine from patients with diabetic nephropathy To investigate whether proteolysis of IGFBP-3 was altered in the urine of patients with diabetic nephropathy, urine samples were incubated with 30-kDa nonglycosylated [ 125 I]IGFBP-3 and analyzed by SDS-PAGE (Fig. 3). After incubation for 24 h at 37 C with micro- or macroalbuminuric diabetic urine samples, little or no intact tracer remained, and the intensity of lower molecular mass bands was increased (lanes 4 and 5). In contrast, the tracer was only partially proteolyzed in the presence of urine from a normoalbuminuric diabetic (lane 3) or a control (lane 7) subject. Only minimal proteolysis was seen in the presence of serum from normal or diabetic subjects (lane 2 and results not shown). These findings suggest that proteolysis of IGFBP-3 is largely confined to the renal system in patients with diabetic nephropathy. The proteolysis observed is not a postvoiding phenomenon, because there was no difference in the urinary IGFBP-3 patterns between samples stored immediately at 20 C and those left at room temperature for 24 h as assessed by immunoblotting (results not shown). The extent of [ 125 I]IGFBP-3 proteolysis in NIDDM subjects was quantitated by laser densitometry as the percentage of tracer in each sample that was present in low molecular mass bands (Fig. 4A). Levels of proteolysis of IGFBP-3 in the healthy control group and the normoalbuminuric group were 49 15% and 29 9%, respectively. The activities in the microalbuminuric and macroalbuminuric groups were significantly increased to 75 25% and 84 14%, respectively (P ). TABLE 1. Clinical characteristics of patients with NIDDM Normoalbuminuria Microalbuminuria Macroalbuminuria P n Sex (F/M) 4/5 8/5 5/ Age (yr) (9) (13) (12) 0.57 Duration of diabetes (yr) (9) (12) (12) 0.64 AER ( g/min) (9) (13) (12) Creatinine clearance (ml/s) (9) (13) (12) 0.34 HbA 1c (%) (9) (13) (12) 0.64 BMI (8) (10) (10) Systolic BP (9) (11) (12) 0.61 Diastolic BP (9) (11) 84 8 (12) 0.10 Therapy (diet/oha/insulin) 1/5/3 1/5/6 1/5/ Hypertension 4 (9) 3 (12) 1 (12) 0.16 AER, Albumin excretion rate; BMI, body mass index; OHA, oral hypoglycemic agent. The number of patients for whom data was available is shown in parentheses.

4 1166 JCE & M 2000 Vol 85 No 3 SHINADA ET AL. TABLE 2. Clinical characteristics of patients with IDDM n Sex (F/M) Age (yr) Duration of diabetes (yr) AER ( g/min) Creatinine clearance (ml/s) HbA1c (%) BMI Systolic BP Diastolic BP Hypertension Normoalbuminuria Microalbuminuria Macroalbuminuria P 7 2/ (7) (7) (7) (7) (7) (6) (7) 79 9 (7) 0 (7) 4 3/ (4) (4) (4) (4) (4) (2) (4) 80 8 (4) 2 (4) 3 0/ (3) (3) (3) (3) (3) (2) (2) 84 6 (2) 2 (2) AER, Albumin excretion rate; BMI, body mass index. The number of patients for whom data was available is shown in parentheses. FIG. 1. Urinary and serum IGFBPs in patients with diabetic nephropathy. Urine (lanes 1, 2, 4, and 6) and serum (lanes 3, 5, 7, and 8) from control (lanes 1 and 8) and normo- (lanes 2 and 3), micro(lanes 4 and 5), and macroalbuminuric (lanes 6 and 7) diabetic subjects were separated by SDS-12% PAGE and analyzed by WLB using [125I]IGF-II (A). The membrane was then stripped and reanalyzed by IGFBP-3 immunoblotting (B). Migration of molecular mass markers is shown on the left. Urinary protease activity for IGFBP-3 was also examined in the three diabetic groups with IDDM. The extents of IGFBP-3 proteolysis in the normo-, micro-, and macroalbuminuric groups were 44 18%, 91 19%, and 83 29%, respectively (P 0.008). The extent of IGFBP-3 proteolysis in urine was therefore similar regardless of the type of diabetes. The extent of IGFBP-3 proteolysis was correlated with AER in patients with NIDDM (r 0.61; P ; Fig. 4B and Table 3). A similar relationship was observed in patients with IDDM (r 0.56; P 0.04; Table 3). When IDDM and NIDDM patients were combined, there was also a positive FIG. 2. Urinary IGFBP-3 in control (A) and diabetic (B) subjects. Urine samples were analyzed by IGFBP-3 immunoblotting. A, Control subjects (lanes 1 6). B, Normo- (lanes 1 and 2), micro- (lanes 3 and 4), and macroalbuminuric (lane 5) urine. Urine from a nondiabetic control is shown in lane 6. Migration of molecular mass markers is shown on the left. correlation between the extent of IGFBP-3 proteolysis and the duration of diabetes (r 0.33; P 0.02; Table 3) and a negative relationship with creatinine clearance (r 0.31; P 0.03; Table 3). After stepwise regression analysis, a final equation was derived, including significant positive associations with AER and duration of diabetes and an inverse association with age in NIDDM patients alone (overall: r 0.75; P ) and in NIDDM and IDDM patients combined (overall: r 0.69; P ). Stepwise regression analysis was not performed with IDDM patients alone because of the small sample size. There was no correlation between IGFBP-3 proteolysis and blood pressure or hemoglobin A1c by either simple or stepwise regression analysis.

5 IGFBP-3 PROTEOLYSIS IN DIABETIC NEPHROPATHY 1167 TABLE 3. Correlations between IGFBP-3 proteolysis and other variables FIG. 3. Proteolysis of urinary IGFBP-3 is increased in diabetic nephropathy. Incubation of [ 125 I]IGFBP-3 was performed for5hat37c with pooled serum from control subjects (lane 1), serum from a diabetic patient with macroalbuminuria (lane 2), or urine from diabetic patients with normo- (lane 3), micro- (lane 4), or macroalbuminuria (lane 5). Tracer incubated with buffer alone is shown in lane 6, and urine from a control subject is shown in lane 7. Migration of molecular mass markers is shown on the left. NIDDM IDDM Combined Albumin excretion rate 0.61 a 0.56 b 0.58 a Duration of diabetes b Creatinine clearance b Age Systolic BP HbA 1c BMI Results are shown as simple correlation coefficients. Albumin excretion rates were log-transformed before analysis. BMI, Body mass index. a P b P FIG. 5. Urinary IGFBP-3 is proteolyzed by a serine protease. [ 125 I] IGFBP-3 was incubated at 37 C for 5 h with aliquots of macroalbuminuric diabetic urine in the presence or absence of protease inhibitors. Lane 1, Urine only; lane 2, PBS buffer only; lane 3, pepstatin (2.5 mol/l); lane 4, EDTA (5 mmol/l); lane 5, 1,10-phenanthroline (10 mmol/l); lane 6, aprotinin (50 U/mL); lane 7, leupeptin (5 mmol/l). This gel is representative of five experiments. FIG. 4. Proteolysis of urinary IGFBP-3 is increased in patients with NIDDM and diabetic nephropathy. A, The amount of [ 125 I]IGFBP-3 proteolysis was determined by laser densitometry as the percentage of the total optical density per lane that occurred in bands representing proteolytic cleavage products rather than intact IGFBP-3. Results are shown as the mean SD. Con, Control subjects; Normo, normoalbuminuric; Micro, microalbuminuric; Macro, macroalbuminuric. ***, P vs. normoalbuminuria;, P 0.05 vs. control;, P vs. control. B, Relationship between extent of IGFBP-3 proteolysis and log-transformed urinary albumin excretion (AER) in patients with NIDDM. The regression line shown represents the equation y x (r 0.61; P ). IGFBP-3 is proteolyzed by a serine protease in the urine of patients with diabetic nephropathy Figure 5 shows degradation of recombinant human [ 125 I]IGFBP-3 by urine from a macroalbuminuric subject incubated for 5hat37Cwith (lanes 3 7) and without (lane 2) protease inhibitors. Aprotinin (lane 6; a serine protease inhibitor) and leupeptin (lane 7; a serine and thiol protease inhibitor) inhibited degradation of IGFBP-3. In contrast, pepstatin (lane 3; an aspartic protease inhibitor), EDTA (lane 4; a chelator of divalent cations), and 1,10-phenanthroline (lane 5; a matrix metalloproteinase inhibitor) did not inhibit IGFBP-3 degradation. Leupeptin and aprotinin inhibited IGFBP-3 proteolysis to similar extents in urine samples from five subjects. After incubation with leupeptin and aprotinin, proteolysis levels were 36 24% and 36 20%, respectively, compared with 70 19% in untreated samples. Discussion In this study, we have shown that levels of intact IGFBP-3 are decreased in urine from patients with diabetic nephropathy due to increased serine protease activity. Urinary IGFBP-3 proteolysis was not increased in normoalbuminuric diabetic patients, indicating that diabetes alone is not responsible for this finding. However, increased IGFBP-3 proteolysis was observed even in patients with microalbuminuria and normal renal function, indicating that it occurred relatively early in the disease process. The effect was observed to the same extent in patients with IDDM and NIDDM, indicating that the type of diabetes is not a significant factor. Previous studies have shown the presence of intact IGFBP-2 ( 35 kda) and IGFBP-3 ( 40 kda) as well as smaller IGFBPs in urine from normal subjects (20). The IGFBP-2/

6 1168 SHINADA ET AL. JCE&M 2000 Vol 85 No 3 IGFBP-3 ratio was higher in urine than in serum, a finding confirmed in the present study. More detailed studies showed proteolysis of IGFBP-3 in urine from control subjects (21, 25). Using the same assay as that used in the current study, the extent of IGFBP-3 proteolysis was 10 50% in these studies. One study showed that the predominant form of IGFBP-3 in urine from children and adults was 17.7 kda in size (26); this form may have been missed or understated in studies using WLB for detection, because this fragment binds IGFs poorly. Urinary IGFBPs have also been studied in a variety of kidney diseases. In patients with glomerulonephritis but normal renal function, urinary intact IGFBP-3 levels were increased (20). The urine of children with nephrotic syndrome contains increased amounts of intact as well as fragmented IGFBP-3, as determined by size exclusion chromatography (27). Another study showed increased IGFBP-3 proteolysis in urine from patients with Alport s syndrome, but not in those with IgA nephropathy, systemic lupus erythematosus, or focal segmental glomerulosclerosis (21). In patients with acute or chronic renal failure (serum creatinine, mmol/l), there was no intact IGFBP-3 in urine (20, 25), and urinary IGFBP-3 protease activity was greater than 80%, as measured by the same assay as that used in the present study (25). In the present study even diabetic patients with microalbuminuria and normal renal function had urinary IGFBP-3 protease activities in this range. These findings suggest that the induction of urinary IGFBP-3 proteolysis may be disease specific rather than being related to renal dysfunction per se. Based on the pattern of inhibition of protease activity by protease inhibitors, the IGFBP-3 protease detected in the urine of five patients with diabetic nephropathy in the present study is a serine protease. Serine protease activity was also observed in patients with renal failure, although three distinct activities with different protease inhibitor profiles were observed in those patients (25). IGFBP-3 proteolysis is increased in serum from patients with untreated insulin-dependent diabetes mellitus due to the activity of cation-dependent serine proteases (7). Insulin treatment restored protease activity to control levels. Consistent with this observation, levels of intact and proteolyzed IGFBP-3 did not differ in serum from treated diabetic and control subjects in the present study. Further, the presence of diabetic nephropathy did not affect levels of intact IGFBP-3 or its fragments in serum from diabetic patients. This is in contrast to findings in patients with renal failure, who had decreased intact IGFBP-3 levels and increased levels of low molecular mass IGFBP-3 fragments in serum (25). The source of urinary IGFBP-3 has not been previously defined. Some studies have shown correlations between urinary and serum IGFBP-3 levels and between urinary IGFBP-3 levels and glomerular filtration rate, suggesting that filtration of serum IGFBP-3 may be important (28, 29). IGFBP-3 messenger ribonucleic acid is expressed in human kidney (30), so that local expression may also contribute to urinary IGFBP-3 levels. In patients with renal failure, filtration of serum fragments may contribute to the increased urinary levels of IGFBP-3 fragments. In contrast, the present study suggests that fragmentation of IGFBP-3 occurs within the renal tract in diabetic nephropathy. Firstly, the presence of nephropathy had no effect on levels of IGFBP-3 fragments in serum. Secondly, increased proteolytic activity was detected in urine. We therefore postulate that proteolysis is occurring within the kidney or urinary tract. This may be due to the induction of proteases within the renal tract. In the present study there was a significant correlation between IGFBP-3 proteolysis and urinary albumin excretion. During this process, glomerular leakage of albumin and other proteins is increased; it is possible that this may include filtration of circulating proteases or secretion of proteases from renal tubules into urine. In diabetic nephropathy, accumulation of mesangial components, such as collagen, laminin, and heparan sulfate proteoglycans, is due to alterations in the balance between protein synthesis and degradation (31). In streptozotocininduced diabetic rats, matrix metalloproteinase activity in the kidney is decreased (32, 33). In patients with NIDDM, Del Prete et al. has reported down-regulated matrix metalloproteinase-2 and increased tissue inhibitor of matrix metalloproteinase gene expression (34). In diabetic rats, there is decreased albumin degradation at a postglomerular site compared with that in control rats (35). In contrast to the present findings, renal protease activity therefore appears to be generally decreased. This indicates that increased urinary IGFBP-3 proteolysis is not due to a generalized increase in protease activity, but is more likely to be a specific effect. Proteolysis of IGFBPs has emerged as a possible mechanism by which IGFs are released from high affinity complexes and are subsequently able to bind to IGF-I receptors (5). In the present study, comparison of WIB and WLB clearly shows that the smaller IGFBP-3 fragment has decreased IGF binding affinity, which is consistent with the observations of others (5). A previous study has shown increased urinary, but not plasma, IGF-I levels in patients with microalbuminuria (19). It is possible that the combination of increased IGFBP-3 proteolysis and IGF-I in the urine of patients with diabetic nephropathy results in increased free IGF-I levels. As IGF-I in the glomerular ultrafiltrate of nephrotic rats stimulates proximal tubular collagen secretion (36), increased IGF-I activity may then contribute to the structural changes associated with diabetic nephropathy. Acknowledgments We thank David Casley, Department of Medicine, for iodinations, and Peter McRae and Kelli Panagopoulos, Clinical Pathology Department, for assisting with sample collection. References 1. Gambaro G, Baggio B Growth factors and the kidney in diabetes mellitus. Crit Rev Clin Lab Sci. 35: Jones JI, Clemmons DR Insulin-like growth factors and their binding proteins: biological actions. Endocr Rev. 16: Bach LA, Rechler MM Insulin-like growth factor binding proteins. Diabetes Rev. 3: Baxter RC, Martin JL, Beniac VA High molecular weight insulin-like growth factor binding protein complex. J Biol Chem. 264: Fowlkes JL Insulinlike growth factor-binding protein proteolysis: an emerging paradigm in insulinlike growth factor physiology. Trends Endocrinol Metab. 8: Bang P, Brismar K, Rosenfeld RG Increased proteolysis of insulin-like growth factor-binding protein-3 (IGFBP-3) in noninsulin-dependent diabetes mellitus serum, with elevation of a 29-kilodalton (kda) glycosylated IGFBP-3

7 IGFBP-3 PROTEOLYSIS IN DIABETIC NEPHROPATHY 1169 fragment contained in the approximately 130- to 150-kDa ternary complex. J Clin Endocrinol Metab. 78: Bereket A, Lang CH, Blethen SL, Fan J, Frost RA, Wilson TA Insulin-like growth factor binding protein-3 proteolysis in children with insulin-dependent diabetes mellitus: a possible role for insulin in the regulation of IGFBP-3 protease activity. J Clin Endocrinol Metab. 80: Bortz JD, Rotwein P, DeVol D, Bechtel PJ, Hansen VA, Hammerman MR Focal expression of insulin-like growth factor I in rat kidney collecting duct. J Cell Biol. 107: Price GJ, Berka JL, Edmonson SR, Werther GA, Bach LA Localization of mrnas for insulin-like growth factor binding proteins 1 to 6 in rat kidney. Kidney Int. 48: Hammerman MR, Miller SB The growth hormone insulin-like growth factor axis in kidney revisited. Kidney Int. 265:F1 F Bach LA, Rechler MM Insulin-like growth factors and diabetes. Diabetes Metab Rev. 8: Flyvbjerg A, Thorlacius-Ussing O, Naeraa R, Ingerslev J, Orskov H Kidney tissue somatomedin C and initial renal growth in diabetic and uninephrectomised rats. Diabetologia. 31: Bach LA, Jerums G Effect of puberty on initial kidney growth and rise in kidney IGF-1 in diabetic rats. Diabetes. 39: Werner H, Shen-Orr Z, Stannard B, Burguera B, Roberts Jr CT, LeRoith D Experimental diabetes increases insulinlike growth factor I and II receptor concentration and gene expression in kidney. Diabetes. 39: Bach LA, Cox AJ, Mendelsohn FAO, Herington AC, Werther GA, Jerums G Focal induction of IGF-I binding proteins in proximal tubules of diabetic rat kidney. Diabetes. 41: Flyvbjerg A, Kessler U, Dorka B, Funk B, Orskov H, Kiess W Transient increase in renal insulin-like growth factor binding proteins during initial kidney hypertrophy in experimental diabetes in rats. Diabetologia. 35: Landau D, Chin E, Bondy C, et al Expression of insulin-like growth factor binding proteins in the rat kidney: effects of long-term diabetes. Endocrinology. 136: Price GJ, Berka JL, Werther GA, Bach LA Cell-specific regulation of mrnas for IGF-I and IGF-binding proteins-4 and -5 in streptozotocin-diabetic rat kidney. J Mol Endocrinol. 18: Cummings EA, Sochett EB, Dekker MG, Lawson ML, Daneman D Contribution of growth hormone and IGF-I to early diabetic nephropathy in type 1 diabetes. Diabetes. 47: Hasegawa Y, Cohen P, Yorgin P, Rosenfeld RG Characterization of urinary insulin-like growth factor binding proteins. J Clin Endocrinol Metab. 74: Gargosky SE, Hasegawa T, Tapanainen P, MacGillivray M, Hasegawa Y, Rosenfeld RG Urinary insulin-like growth factors (IGF) and IGF-binding proteins in normal subjects, growth hormone deficiency, and renal disease. J Clin Endocrinol Metab. 76: Bach LA, Leeding KS, Leng SL Cyclic AMP agonists increase levels of insulin-like growth factor (IGF) binding protein-6 in PC12 rat pheochromocytoma cells. Growth Horm Insulin-Like Growth Factor Res. 8: Lamson G, Giudice LC, Rosenfeld RG A simple assay for proteolysis of IGFBP-3. J Clin Endocrinol Metab. 72: Gilbert RE, Akdeniz A, Jerums G Semi-quantitative determination of microalbuminuria by urinary dipstick. Aust NZ J Med. 22: Lee DY, Park SK, Yorgin PD, Cohen P, Oh Y, Rosenfeld RG Alteration in insulin-like growth factor-binding proteins (IGFBPs) and IGFBP-3 protease activity in serum and urine from acute and chronic renal failure. J Clin Endocrinol Metab. 79: Spagnoli A, Gargosky SE, Spadoni GL, et al Characterization of a low molecular mass form of insulin-like growth factor binding protein-3 (17.7 kilodaltons) in urine and serum from healthy children and growth hormone (GH)-deficient patients: relationship with GH therapy. J Clin Endocrinol Metab. 80: Haffner D, Tonshoff B, Blum WF, et al Insulin-like growth factors (IGFs) and IGF binding proteins, serum acid-labile subunit and growth hormone binding protein in nephrotic children. Kidney Int. 52: Tonshoff B, Blum WF, Vickers M, Kurilenko S, Mehls O, Ritz E Quantification of urinary insulin-like growth factors (IGFs) and IGF binding protein 3 in healthy volunteers before and after stimulation with recombinant human growth hormone. Eur J Endocrinol. 132: Gill MS, Whatmore AJ, Tillman V, et al Urinary IGF and IGF binding protein-3 in children with disordered growth. Clin Endocrinol (Oxf). 46: Hintz RL, Bock S, Thorsson AV, et al Expression of the insulin like growth factor-binding protein 3 (IGFBP-3) gene is increased in human renal carcinomas. J Urol. 146: Rabkin R, Shechter P, Shi J, Boner G Protein turnover in the hypertrophying kidney. Miner Electrolyte Metab. 22: Reckelhoff J, Tygart V, Racusen L, Dzielak D Glomerular metalloproteinase activity in streptozotocin- treated rats and in spontaneously diabetic rats (BB/DP). Life Sci. 55: Reckelhoff J, Tygart V, Mitias M, Walcott J STZ-induced diabetes results in decreased activity of glomerular cathepsin and metalloproteinease in rats. Diabetes. 42: Del Prete D, Anglani F, Forino M, et al Down-regulation of glomerular matrix metalloproteinase-2 gene in human NIDDM. Diabetologia. 40: Burne MJ, Panagiotopoulos S, Jerums G, Comper WD Alterations in renal degradation of albumin in early experimental diabetes in the rat: a new factor in the mechanism of albuminuria. Clin Sci. 95: Hirschberg R Bioactivity of glomerular ultrafiltrate during heavy proteinuria may contribute to renal tubulo-interstitial lesions: evidence for a role for insulin-like growth factor I. J Clin Invest. 98: