Decorin, a Novel Player in the Insulin-like Growth Factor System*

Decorin is a multifunctional proteoglycan that is expressed by sprouting endothelial cells. Its expression supports capillary formation and cell survival. Previously, it was shown that some effects of decorin are mediated by protein kinase B and the cyclin-dependent kinase inhibitor, p21. However, the cell surface receptor responsible for these effects was unknown. We demonstrate that decorin binds to the insulin-like growth factor-I (IGF-I) receptor on endothelial cells with an affinity in the nanomolar range (KD = 18 nm), which is comparable with IGF-I (KD = 1.2 nm). Furthermore, decorin can bind IGF-I itself, but with a lower affinity (KD = 190 nm) than classical IGF-I-binding proteins. Decorin addition causes IGF-I receptor phosphorylation and activation, which is followed by receptor down-regulation. These effects are caused by the core protein of decorin, and the binding region could be mapped to the N terminus of the molecule. The physiological relevance of the decorin/IGF-I receptor interaction was corroborated in two animal models (e.g. inflammatory angiogenesis in the cornea and unilateral ureteral obstruction). In both models the IGF-I receptor was up-regulated in decorin-deficient mice compared with controls and the up-regulation could not compensate the decorin deficiency in the disease models. These data indicate that decorin is an important player in the IGF system and its loss cannot fully be compensated in different types of diseases.

Decorin is the most thoroughly investigated member of the still growing family of small leucine-rich proteoglycans. Core proteins of these proteoglycans are characterized by leucinerich repeat motifs flanked by cysteine-rich clusters. In addition, they carry at least a single glycosaminoglycan chain. Recently, a crystal structure for bovine decorin has been published (1) that together with earlier x-ray scattering data (2) suggests that decorin is a dimeric protein with 12 leucine-rich repeat motifs that include the N-and C-terminal cysteine-rich regions. This numbering of leucine-rich repeats is subsequently used in this report. Each monomer adopts a curved structure, and an antiparallel dimerization occurs through the ␤-sheet on the concave surface of the monomer. Decorin has been shown to have a variety of different functions that can be mediated by the core protein as well as the glycosaminoglycan chain. It regulates collagen fibril formation and stabilization, and it modulates cell adhesion as well as transforming growth factor-␤ activity (for review see Refs. 3 and 4). More recently it was found to influence directly the behavior of several types of cells. It can interact with members of the ErbB receptor family in tumor cells where it leads to a more differentiated phenotype (5) or to apoptosis of these cells (6). Decorin expression in endothelial cells, hepatocytes, fibroblasts, or macrophages causes differentiation, but it does not lead to apoptosis. In contrast, it protects cells from programmed cell death (6 -8).
Endothelial cells cultured inside a collagen lattice can form capillary-like structures. We showed that both capillary formation and protection from apoptosis were only observed in decorin-synthesizing cells. A decorin-containing matrix was not sufficient to induce these changes. Adenoviral gene transfer of decorin into endothelial cells confirmed that decorin was instrumental for tube formation and cell survival (7). The physiological and pathophysiological relevance of these observations in culture was further supported by the finding that interleukin-6 and -10, two cytokines that are released during inflammation, induce decorin synthesis in endothelial cells growing in a collagen lattice (9). These cell culture results corroborate earlier observations in vivo, where decorin was found during inflammation-induced angiogenesis, i.e. in newly formed capillaries in granulomatous tissue (7) or in neovessels in temporal arteritis, but not in capillaries of the ovary in different phases of follicle and corpus luteum formation (10). The importance of decorin in angiogenesis could also be demonstrated in a model of inflammatory angiogenesis in the cornea, as blood vessel growth was significantly reduced in decorin knock-out mice compared with wild-type mice (11). However, decorin signaling is involved not only in angiogenesis but also in renal inflammation and fibrosis, as enhanced apoptosis and atrophy of tubular epithelial cells as well as an increased infiltration of mononuclear cells were observed in unilateral ureteral obstruction (UUO) 1 in decorin-deficient mice (12).
How decorin affects differentiation and apoptosis is not completely understood. In tumor cells, decorin binds to the epithelial growth factor (EGF) receptor or ErbB4 and leads to activation of the mitogen-activated kinase (MAPK) pathway, Ca 2ϩ influx, induction of the cyclin-dependent kinase inhibitor, p21, and subsequently to down-regulation of the receptor (13)(14)(15). In endothelial cells decorin affects different pathways. It enhances the phosphorylation of protein kinase B (Akt) and subsequently induces p21 by a MAPK-independent pathway. In addition to p21 the related cyclin-dependent kinase inhibitor, p27, is also induced, but by an Akt-and MAPK-independent mechanism (16).
In this study we have searched for potential cell surface receptors that could be involved in decorin signaling. Inhibitor studies indicated that the insulin-like growth factor-I receptor (IGF-IR) was a likely candidate. The IGF-IR is a ligand-activated tyrosine protein kinase highly homologous to the insulin receptor. It consists of a heterotetramer (␣2␤2) linked by disulfide bridges. Via its extracellular ␣2 units, the receptor binds IGF-I and IGF-II with high affinity. This interaction leads to autophosphorylation of the receptor and downstream proteins, including insulin receptor substrate-1. In the next step phosphoinositide 3-kinase is activated, which increases the levels of phosphoinositol-3,4,5trisphosphate. Phosphoinositol-3,4,5-trisphosphate binds to the pleckstrin homology domains of phosphoinositide-dependent kinase I and Akt, thereby initiating the phosphorylation of Akt and several other signaling molecules such as protein kinases C and A. These signaling events finally lead to the inhibition of proapoptotic factors, the activation of survival factors, and cell differentiation or proliferation, depending on the cell type and tissue investigated (Ref. 17 and references therein). The general importance of the IGF-IR is implicated by the phenotype of IGF-IR knock-out mouse, which exhibits severe growth retardation and dies at birth (18). IGF-I is mainly produced by the liver, but expression in other types of tissue has been shown (19). The availability of IGF-I and IGF-II for receptor binding is regulated extracellularly by six high affinity-binding proteins and several low affinity-binding proteins. In addition, some of the IGF-binding proteins can exert biological functions independently of their role in the IGF system (20).
In this report we show for the first time that decorin can bind the IGF-IR and also interacts with IGF-I itself. This interaction regulates IGF signaling and may be responsible for the retardation of angiogenesis in the cornea of decorin knock-out mice and for the atrophy in later stages of tubulointerstitial fibrosis observed after UUO in these mice.

EXPERIMENTAL PROCEDURES
Materials-Antibodies used were sc-713 (Santa Cruz Biotechnologies) against the ␤-chain of the IGF-IR, rabbit anti-phospho-Akt (T308), rabbit anti-Akt (Cell Signaling Technologies, Beverly MA), and rat anti-CD31 (ERMP12; BMA Biomedical, Augst, Switzerland). Secondary antibodies labeled with horseradish peroxidase used were porcine antirat immunoglobulin (DAKO, Hamburg, Germany), goat anti-rabbit-IgG (Bio-Rad), and horse anti-mouse IgG (Alexis, Grü neberg, Germany). Tyrphostin AG1048 and AG1478 were obtained from Alexis. IGF-I and chemicals not specifically indicated were purchased from Sigma. Decorin and its mutants were purified from conditioned medium from HEK-293 cells or human skin fibroblasts as previously described (21). The purity of the preparation was checked by sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and silver staining.
Cell Culture-Cells of the permanent human endothelial cell line EA.hy 926 (22) were permanently grown in MCDB 131 supplemented with 10% fetal calf serum, hypoxanthin/aminopterin/thymidine supplement, and antibiotics (all from Invitrogen). Hydrophilic cell culture dishes (3 cm) were coated with a neutralized collagen mixture (7). After collagen gel formation (30 min, 37°C), EA.hy 926 (500,000 cells) were plated on this substrate in 1 ml of Waymouth MAB 87/3 medium containing 1% heat-inactivated fetal calf serum. After 5 h of culture, 1 ml of medium without serum was added. After 48 h, decorin and/or the respective inhibitors were added, and cells were harvested at the indicated times. Adenoviral infections were performed as previously described (7,16).
Immunoblotting-For the detection of phosphorylated proteins, cells from a 3-cm-dish were harvested in 100 l of cold lysis-buffer (10 mM sodium phosphate, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 2 mM sodium vanadate, proteinase inhibitors) and sonicated. Aliquots (equal protein concentration/lane) were applied on SDS-PAGE gels. In addition, immunoprecipitations with sc-713 were carried out (described below), and precipitated proteins were separated by SDS-PAGE. After transfer to nitrocellulose, proteins were detected as previously described (7). For the detection and quantification of IGF-IR in mouse kidneys, homogenates of whole kidneys were prepared and applied to SDS-PAGE as previously described (12). Western blots were performed using sc-713 according to the recommendations of the manufacturer. An antibody to ␤-tubulin was used as control for equal loading. Bands were quantified with the ImageQuant 5.0 software program (Amersham Biosciences).
Decorin (10 g) with and without chondroitin ABC lyase digestion (Seikagaku Kogyo, Tokyo, Japan) was separated by SDS-PAGE and transferred to nitrocellulose. Blots were blocked for 2 h with 50 mM Tris/HCl, pH 7.4, 3% Nonidet P-40, 0.5 mg/ml NaN 3 and probed with 125 I-IGF-I in binding buffer (50 mM Tris/HCl, pH 7.4, 0.1% Tween 20, 1 mg/ml bovine serum albumin) overnight at 4°C. After washing with binding buffer the nitrocellulose was dried and exposed to x-ray films (X-OMAT; Kodak). A Sepharose CL-4B gel filtration column (20 cm long) was equilibrated with binding buffer with or without decorin, and 125 I-IGF-I (500,000 cpm) was applied. The elution profiles were monitored by scintillation counting.
To determine binding of decorin to the IGF-IR, the receptor was immunoprecipitated from EA.hy 926 cells (7 ϫ 10 6 cells) with antibody sc-713. Cells were harvested in lysis buffer, and a complex of normal rabbit serum bound to protein A-Sepharose was added for 6 h to remove non-specifically binding proteins. Then 10 g of the anti-IGF-IR were added, and after 16 h the antibodies were precipitated with protein A-Sepharose. As control, a similar amount of protein A-Sepharose was incubated with the antibody but without cell lysate. After washing (3 ϫ lysis buffer, 2 ϫ phosphate-buffered saline), the material was divided in 7 aliquots and incubated in 250 l of Tris-buffered saline, 0.1% bovine serum albumin, 0.1% Tween 20 with labeled decorin or IGF-I (as indicated). For competition experiments unlabeled wild type decorin or mutants were added together with the labeled molecules. Immune complexes without receptor were used as controls. After washing five times with binding buffer, bound protein was removed by boiling in 1% SDS and aliquots were counted in a scintillation counter. Dissociation constants were determined with GraphPad Prism3 software using the subroutine for fitting one site binding curves (GraphPad software, San Diego, CA).
Binding of decorin to IGF-I was determined with 125 I-decorin in a solid phase assay. Shortly, IGF-I (4 g/ml, 50 l/well) was used for coating microtiter plates (Immobilon 2; Dynatec); after blocking with Tris-buffered saline, 3% bovine serum albumin, 0.05% Tween 20 for 4 h, labeled decorin was applied in the same buffer (4 h, 37°C). After washing, the bound material was released and measured by scintillation counting. Dissociation constants were determined as described above.
Animal Experiments-All animal work was done in accordance with the German Animal Protection Law. Decorin-deficient mice were generated as described previously (23). Angiogenesis assays in mice were performed by chemical cauterization of the cornea as described (11). The eyes were removed and fixed in 4% phosphate-buffered paraformaldehyde for paraffin embedding. Unilateral uretral obstractions were done as in previous experiments (12). For the quantification of the IGF-IR, whole kidneys were homogenized in 1 ml of extraction buffer (final volume), and equal volumes were applied for SDS-PAGE. The protein content/aliquot of ligated kidneys was ϳ50 g, and the protein content/aliquot of unligated kidney was ϳ80 g/lane. Equal volumes were applied because the kidneys change their weight during the procedure and equal volumes reflect more accurately the situation in the organ. Western blots were carried out as described elsewhere (12) with the exception that the blots were incubated with the antibody sc-713 (1:500) against IGF-IR as primary antibody overnight at 4°C. Immunohistochemistry was performed as described elsewhere (11,12).

RESULTS
Previous experiments have shown that decorin can induce tube formation in endothelial cells and inhibit apoptosis. The phosphorylation of Akt was one step in the pathway(s) (16), but the cell surface receptor for decorin in endothelial cells was unknown. To analyze how decorin activates Akt, different re-ceptor tyrosine kinase inhibitors were tested. The EGF receptor inhibitor tyrphostin AG1478 (10 M) had no effect on decorin-induced Akt phosphorylation in endothelial cells, but preincubation with tyrphostin AG1024 (10 M), an inhibitor of the insulin receptor and the IGF-IR tyrosine kinase, was able to inhibit Akt phosphorylation (Fig. 1A) and p21 expression (result not shown).
IGF-I led to a dose-dependent induction of p21 mRNA in EA.hy 926 cells after 16 h with maximal expression levels (5-fold) at a concentration of 100 ng/ml (result not shown). Therefore, 100 ng/ml IGF-I were used in the following experiments. Immunoprecipitations of the IGF-IR from endothelial cells that had been stimulated for 5 min with IGF-I or decorin, followed by Western blots with an antibody recognizing phosphotyrosine residues, showed phosphorylation of the receptor after addition of decorin or IGF-I (Fig. 1B). After normalization for the amount of IGF-IR, IGF-I increased receptor phosphorylation by 42% and decorin by 47%, respectively. Reprobing the same blot with an antibody to pY1131 of the IGF-IR indicated that the receptor tyrosine kinase was active. These data demonstrate that decorin is involved in signal transduction via the IGF system in endothelial cells.
To investigate whether decorin and IGF-I interfere with each other's effects on downstream signals, two types of experiments were performed. In one type, optimal concentrations of IGF-I (100 ng/ml) and decorin (80 g/ml) were combined and added together to endothelial cells; in the second type of experiment, cells were preincubated at 4°C with decorin or IGF-I and then the alternative ligand was added for 10 min at 37°C. Neither of these experiments (n ϭ 2) led to a significant decrease or increase in IGF-IR or Akt phosphorylation compared with the addition of a single ligand at optimal concentration (result not shown).
To investigate whether decorin can down-regulate the IGF-IR, we used a replication-deficient adenovirus containing the human decorin cDNA to induce decorin expression. Previous studies had shown that decorin-mediated effects were observed 2 days after infection and subsequent culture of endothelial cells in a collagen lattice (16). Therefore, endothelial cells treated with a decorin-containing adenovirus or control virus were harvested after 2 and 4 days of culture on a collagen lattice. The IGF-IR was expressed in cells treated with the control virus, whereas overexpression of decorin drastically reduced the amount of receptor found in the cells after 2 days, when the decorin expression reached its peak. There was no recovery after prolonged decorin expression for 4 days (Fig. 1C). These data show that decorin leads to a sustained down-regulation of the IGF-IR.
To characterize which part of the decorin core protein is important for signaling, Akt phosphorylation and/or p21 induction were analyzed after decorin mutants were applied to endothelial cells. For these experiments the following mutants were used: DCN-Q153, a mutant that contains the N terminus and the first 5 leucine-rich repeats of decorin, and DCN-E180K, a mutant with a point mutation reducing its affinity to collagen type I (21). These two mutants were chosen because previous studies (15) had shown that the N-terminal part of decorin is not involved in binding the EGF receptor but leucine-rich repeat 7, which contains the point mutation (DCN-E180K). Our investigation demonstrated that DCN-Q153 alone was still able to induce Akt phosphorylation ( Fig. 2A) and p21 expression (result not shown), albeit to a lesser extent than wild type decorin. DCN-E180K induced p21 as strongly as wild type decorin, and it also triggered its translocation to the nucleus (Fig. 2B). These results imply that a different part of the decorin core protein interacts with the IGF-IR than with the EGF receptor. After preincubation with the indicated inhibitors for 1 h, decorin (80 g/ml) was added for 10 min. Cells were harvested, and equal amounts of protein were separated by SDS-PAGE. Phosphorylation of Akt at Thr-308 was determined by Western blotting. Only the IGF-IR/insulin receptorspecific inhibitor (AG1024), not the EGF receptor-specific reagent (AG1478), inhibited phosphorylation. B, EA.hy 926 cells (3 ϫ 10 Ϫ6 ) were cultured as in panel A, and IGF-I (100 ng/ml) or decorin (80 g/ml) was added for 4 min. Immunoprecipitations for the IGF-IR were carried out with antibody sc-713. The precipitates were separated by SDS-PAGE, and Western blots were probed for phophotyrosine (pY), the tyrosine kinase activation-specific phosphorylation site at Tyr-1131, and IGF-RI itself. Decorin leads to a strong tyrosine phosphorylation and activation of the IGF-IR. C, EA.hy 926 cells were infected with an adenoviral vector for decorin (AdvDCN) and a control virus (AdvCo) and seeded after 24 h on fibrillar collagen. After 2 and 4 days, the medium and the cells were harvested. Decorin in the medium was determined after anion exchange chromatography and Western blotting with an antibody specific for human decorin. Aliquots of cells (40 g of protein/ lane) were separated by SDS-PAGE, and IGF-IR and ␤-tubulin (control for loading) were determined by Western blotting. Decorin down-regulates the IGF-IR in endothelial cells.

FIG. 2. Effects of decorin mutants on Akt and p21 in endothelial cells.
A, EA.hy 926 cells were plated as in Fig. 1A. Wild type decorin (40 g/ml ϳ1 M) or DCN-Q153 (25 g/ml ϳ2 M) were added for 10 min. Akt phosphorylation was determined as in Fig. 1A. The fragment DCN-Q153 is sufficient to induce Akt phosphorylation. B, EA.hy 926 cells were plated as in Fig. 1A. Wild type decorin or DCN-E180K (both 80 g/ml) was added to the cells. After 20 h, cells were harvested and nuclear (N) and cellular (C) fractions were prepared. The cyclin-dependent kinase inhibitor, p21, and ␤-tubulin were determined by Western blotting. The mutant DCN-E180K induces p21 synthesis and translocation to the nucleus as efficiently as wild type decorin.
The binding properties of decorin to IGF-IR and IGF-I were investigated by using iodinated ligands. The purity of decorin for all experiments was checked by SDS-PAGE and silver staining (Fig. 3A). The binding properties of the IGF-IR from endothelial cells to decorin were not determined by direct binding to cells, because it is not known yet whether the IGF-IR is the only decorin binding receptor on the endothelial cell surface. Therefore, the receptor was immunoprecipitated with antibody against its intracellular domain (␤-subunit). The same antibody was also used in the cell culture studies (Fig. 1, B and  C), which confirm its specificity. 125 I-IGF-I and 125 I-decorin were applied as soluble ligands. Ligands bound to the receptor were quantified by scintillation counting and correlating the amount of 125 I[iodine] to the protein concentration of IGF-I or decorin. K D values of 1.2 ϫ 10 Ϫ9 M and K D of 18 ϫ 10 Ϫ9 M were calculated for IGF-I and decorin, respectively. These determinations indicated that the affinity of decorin for the IGF-IR is only about 10-fold less than that of IGF-I (Fig. 3, B and C). In a further experiment, 125 I-IGF-I was bound to the receptor and displaced by unlabeled IGF-I or decorin. Using IGF-I as competitor, an IC 50 of 3 ϫ 10 Ϫ7 M was determined. The decorin proteoglycan and the decorin core protein without glycosaminoglycan chain (result not shown) gave the same IC 50 of 5 ϫ 10 Ϫ7 M (n ϭ 4), whereas the N-terminal fragment DCN-Q153 inhibited with an IC 50 of 20 ϫ 10 Ϫ7 M (Fig. 3D). These data show that decorin and IGF-I can compete for the IGF-IR, and they corroborate the cell culture results that the N-terminal part of decorin is sufficient to bind and activate the IGF-IR.
To investigate whether decorin can also interact with IGF-I, decorin core protein and proteoglycan were separated by SDS-PAGE, transferred to nitrocellulose, and 125 I-IGF-I was used as probe. Both bound 125 I-IGF-I (Fig. 4, insert). The core protein showed increased binding compared with the proteoglycan, indicating that the glycosaminoglycan chain is not responsible for the interaction but may even inhibit binding of decorin to IGF-I. To investigate whether 125 I-IGF-I also binds to decorin in solution, a Sepharose CL-4B gel filtration column was prepared and equilibrated with buffer containing different concentrations of decorin. When 125 I-IGF-I was applied, it eluted in the absence of decorin in the V t of the column. In the presence of decorin a dose-dependent formation of high molecular mass complexes was observed that eluted in the included volume of the column (Fig. 4). Solid phase binding assays of IGF-I with 125 I-decorin gave a dissociation constant of about 190 ϫ 10 Ϫ9 M, which shows a 10-fold lower affinity of decorin to IGF-I itself compared with its affinity to the IGF-IR (K D of 18 ϫ 10 Ϫ9 ). In comparison to IGF-binding proteins this affinity is about 1000fold lower, indicating that decorin is more likely to compete with IGF for binding to the IGF-IR than for the interaction with binding proteins that have K D values in the range of 10 Ϫ10 M (20).
In previous experiments (11) we could show that after chemical cauterization angiogenesis is reduced in the cornea of the decorin-deficient mouse compared with control mice (Fig. 5, A  and B). Sections from these mice showed that after injury the invading cells including the endothelial cells contain more IGF-IR in the decorin-deficient animals compared with the wild type mice (Fig. 5, C and D). Similar observations were made 48, 72, and 96 h after injury (results not shown). In addition, an increase in expression of IGF-IR in the epithelial cell layer of the cornea was observed in the decorin-deficient animals. This was found in both the injured (Fig. 5, C and D) and non-injured epithelium (result not shown). Thus, there is a significant overexpression of IGF-IR in decorin-deficient mice.
Using a different injury model in the same type of mice, based on UUO, we found an enhanced rate of apoptosis of tubular epithelial cells in ligated kidneys from DcnϪ/Ϫ compared with wild type animals up to day 14, with a maximum at day 7 (12). Immunohistochemical analysis using an antibody against the A gel filtration column (Sepharose CL-4B) was equilibrated without decorin (squares), 2 g/ml (triangles), 5 g/ml decorin (circles), and 125 I-IGF-I (500,000 cpm) was applied. Increasing amounts of decorin lead to the formation of a higher molecular mass complex. Insert, a blot with decorin proteoglycans (ϳ97 kDa) and core protein (ϳ45 kDa) was incubated with 125 I-IGF-I. Decorin binds IGF-I with its core protein. The glycosaminoglycan chain is not necessary for binding. IGF-IR in these kidneys at day 7 showed an increase of staining in the tubuloepithelium of obstructed kidneys from both DcnϪ/Ϫ and wild type mice (Fig. 5, G and H) compared with the respective non-ligated kidneys (Fig. 5, E and F). However, ligated kidneys from DcnϪ/Ϫ mice displayed a much stronger increase in IGF-IR (Fig. 5G) compared with ligated kidneys from wild type animals (Fig. 5H). There were no genotypespecific differences in the location and intensity of the IGF-IR staining between non-ligated kidneys (Fig. 5, E and F).
Analysis of IGF-IR expression by Western blot analysis at different time points confirmed the immunohistological findings, showing that IGF-IR is induced in the affected kidney from wild type and DcnϪ/Ϫ mice (Fig. 6A) 7 days after UUO. The maximal difference in IGF-IR expression between obstructed kidneys from decorin-deficient compared with wild type mice was 2.1-fold (n ϭ 3, p Ͻ0.00014) at day 7 (Fig. 6B). At day 14 the difference had declined to 1.8-fold (n ϭ 3, p Ͻ0.018) (Fig. 6C). The expression of IGF-IR in the contralateral unligated kidney was only marginally affected with a 1.3-fold increment (n ϭ 3, p Ͻ0.064). These data indicate that in the kidney IGF-IR is induced by UUO and that significant overexpression occurs in decorin-deficient mice. Both of these models show that IGF-IR is up-regulated, presumably as a compensatory mechanism for the decorin deficiency, and suggest that in vivo decorin and the IGF system work together during signaling in endothelial and as well as epithelial cells. DISCUSSION Our results show for the first time that decorin is involved in IGF-I signaling. It binds to the IGF-IR and activates its tyrosine kinase activity in endothelial cells. Previous results demonstrated that decorin inhibits apoptosis in endothelial cells (7). Other researchers have shown the same for IGF-I (24). In our experiments both decorin and IGF-I led to a phosphorylation of the IGF-IR after 4 min, an early stimulation of Akt phosphorylation after 10 min, and an induction of p21 after 20 h. On first view, concentrations of 100 ng/ml IGF-I compared with 80 g/ml decorin added to the cell cultures may look very different, but on a molar basis the concentration for IGF-I (13 nM) is even higher than the concentration for decorin (2 nM). In addition, one has to consider that the endothelial cells are grown on a collagen matrix, which binds major amounts of decorin. We also found that the medium as well as the cell layer/collagen matrix contained a whole variety of IGF-I-binding proteins (data not shown). Furthermore, we could show in two separate types of experiments that at optimal concentration decorin and IGF-I do not interfere with each other's downstream signaling. From these data we conclude that decorin stimulates the IGF-IR and does not inhibit signaling as has been shown for its interaction with receptors of the ErbB family (5,15).
Furthermore, our binding studies show only a 10-fold lower affinity of decorin to the IGF-IR than IGF-I. This affinity was calculated with the assumption that decorin interacts as a monomer. If a dimeric form of decorin (1) binds to the receptor the dissociation constants will be even closer. Our assays using immunoprecipitated native IGF-IR from the same endothelial cells show a lower affinity (K D ϭ 1.2 ϫ 10 Ϫ9 M) for binding of IGF-I to its receptor than the values determined for IGF-I and recombinant soluble IGF-IR previously published (25,26). These differences may be due to the different methodologies and the presence of the intracellular ␤-subunit that is used for  1). B, after 14 days, ligated kidneys (n ϭ 3), unligated kidney (n ϭ 1). C, quantitation of IGF-IR in the kidney after UUO (n ϭ 3). Wild type mice (black bars), DcnϪ/Ϫ mice (white bars). After 7 and 14 days about twice as much IGF-IR was found in DcnϪ/Ϫ compared with wild type animals. the immunoprecipitation. However, the affinities of decorin and IGF-I to the receptor only differ ϳ10-fold under similar binding conditions, showing that binding of decorin to the IGF-IR on the cell surface of endothelial cells is very likely.
Previous investigations have shown that decorin can interact via its core protein with the EGF receptor and the ErbB4 receptor (5,15). In the case of the EGF receptor the binding has been investigated in more detail, and a stretch of amino acids in the L2 domain of the extracellular EGF binding site (H394-I402) was identified as critical for the interaction. On the decorin core protein, a binding region comprising leucine-rich repeat 7 was identified (15). The IGF-I binding domain of the IGF-IR is structurally related to the binding domain of the EGF receptor (27), but in the homologous region of the L2 domain of the IGF-IR (Phe-371-Leu-377) compared with the L2 domain of the EGF receptor (His-394-Ile-402) the acidic amino acids (Glu-397, Glu-400; EGF receptor) are substituted by basic amino acids (Lys-373, Arg-376; IGF-IR), indicating that this stretch of amino acids in the IGF-IR may have different properties. In addition, the site of interaction for the IGF-RI on the decorin core protein is very likely to be different. We showed that the truncated DCN-Q153, which only contains the N-terminal 5 leucine-rich repeats, was competing for binding to the IGF-IR with IGF-I almost as efficiently as full-length decorin. This truncated mutant could also induce Akt phosphorylation and p21 expression. Furthermore, we could show that a decorin mutant with a point mutation of glutamate 180 (DCN-E180K), which has been shown to have a reduced affinity to collagen type I (21), induced Akt phosphorylation as effectively as wild type decorin. This mutation is in the region that has been shown to be critically involved in EGF receptor binding (15).
Decorin can also bind to IGF-I itself, but the affinity for IGF-I is ϳ1000-fold below the affinities of the typical IGFbinding proteins (20). Therefore, only in situations in which decorin is expressed abundantly may it compete with the classical binding proteins. We did not test whether decorin-bound IGF-I can interact with collagen, because of the low affinity of the observed interaction.
The physiological relevance of the interaction of decorin with the IGF-IR can be clearly seen in the decorin-deficient mouse. The mouse shows no growth retardation like the IGF-I-or IGF-IR-deficient animals (18), but when the mouse is challenged, as during inflammation-induced angiogenesis in the cornea, blood vessel growth is reduced and the IGF-IR is more strongly up-regulated in capillary-forming endothelial cells and invading cells in decorin-deficient mice than in wild type mice. The importance of IGF-I for angiogenesis has been shown in several different models ranging from normal development (28) to tumor angiogenesis (29) and angiogenesis in the retina (30). Some effects of IGF-I are transmitted by the induction of VEGF (30), but direct activation of downstream signaling of the IGF-IR has also been implicated for cell survival and differentiation (31). Moreover, ligation of kidneys from decorin null mice resulted in enhanced apoptosis of tubular epithelial cells. Simultaneously, a more pronounced up-regulation of IGF-IR occurred in these cells. Increased expression of decorin in ligated kidneys from wild type mice and rats (12,32), as well as accelerated tubular atrophy and infiltration during UUO in decorin-deficient mice, suggests an important regulatory role of decorin in this model. In agreement with our findings that decorin and IGF-I stimulate a common signaling pathway, treatment of UUO with IGF-I has been shown to reduce tubular apoptosis and atrophy in the neonatal rat (33). The elevated tubuloepithelial expression of IGF-IR protein in ligated kidneys from DcnϪ/Ϫ mice as compared with wild type animals might represent a compensatory mechanism for the lack of decorin to protect tubular epithelial cells from apoptosis. Not much is known about the expression of IGF-IR protein in the UUO. In terms of IGF-IR mRNA, expression was unchanged up to 7 days of UUO in neonatal rat (33) and markedly downregulated in hydronephrotic kidneys with fibrotic progression in humans (34).
In conclusion, our findings indicate that decorin can bind both IGF-I and its receptor and that this interaction leads to Akt phosphorylation and p21 induction in endothelial cells. Further studies, however, are needed to clarify whether the interaction of decorin with IGF-I and its receptor is important for other types of diseases in which the IGF system is involved.