Decorin Evokes Protracted Internalization and Degradation of the Epidermal Growth Factor Receptor via Caveolar Endocytosis *

Decorin inhibits the epidermal growth factor receptor (EGFR) by down-regulating its tyrosine kinase activity, thereby blocking the growth of a variety of transformed cells and tumor xenografts. In this study we provide evidence that decorin directly binds to the EGFR causing its dimerization, internalization, and ultimately its degradation. Using various pharmacological agents to disrupt clathrin-dependentand-independentendocytosis,wedemonstrate that decorin evokes a protracted internalization of the EGFR primarily via caveolar-mediated endocytosis. In contrast to EGF, decorin targets the EGFR to caveolae, but not to early or recycling endosomes. Ultimately, however, both EGF- and decorin-induced pathways converge into late endosomes/lysosomes for final degradation. Thus, we have discovered a novel biological mechanism for decorin that could explain its anti-proliferative and anti-oncogenic mode of action.

Decorin, the prototypic member of an expanding family of small leucine-rich proteoglycans (1), is implicated in modulating collagen fibrillogenesis (2,3) and cell growth and survival (4 -8). The mechanism of decorin action has begun to be elucidated by the discovery that decorin interferes with epidermal growth factor receptor (EGFR) 3 signaling (9,10). Decorin leads to a protracted down-regulation of EGFR tyrosine kinase (11) and other members of the ErbB family of receptor tyrosine kinase (12), and causes an attenuation of the EGFR-mediated mobilization of intracellular calcium (11). Decorin induces expression of the endogenous cyclin-dependent kinase inhibitor p21 WAF1 (13,14) and a subsequent arrest of the cells in the G 1 phase of the cell cycle (15). These cytostatic effects occur in a wide variety of tumor cell lines (5,14) and can also affect murine tumor cells (5) and normal human cells, such as endothelial cells (16) and macrophages (17). During quiescence, decorin expression is markedly up-regulated in most normal diploid cells, whereas its expression is nearly abolished in most transformed cells (2, 18 -20). For example, transformation induced by the activating transcription factor-3 and the nuclear v-Src and v-Jun oncoproteins cause a marked suppression of decorin gene expression (21)(22)(23). Lack of decorin expression is permissive for tumor development insofar as mice with a targeted ablation of both decorin and the tumor suppressor gene p53, develop lymphomas at accelerated rates as compared with the p53 null animals (24). Consistent with these findings, decorin expression is differentially down-regulated in hepatocellular (25), lung (26), and ovarian (27) carcinomas, and reduced expression of decorin is associated with a poor prognosis in invasive breast carcinoma (28). Finally, adenovirus-mediated gene transfer of decorin causes a significant growth inhibition of various tumors (6,29), and its de novo expression prevents metastastic spreading in a breast carcinoma model (30).
The similarity of the response to decorin in several human cell lines with a diverse histogenetic background and the induction across species suggested that there is a general, well conserved signal transduction system that is operational in mammalian cells. The existence of a "decorin receptor" has been previously reported (31,32). However, this receptor shows very slow kinetics with no apparent associated signaling and could represent a scavenger receptor, insofar as Class A scavenger receptors bind both decorin and biglycan (33). Thus, we hypothesized that decorin might induce growth suppression by activating a cell surface receptor, such as members of the ErbB family of receptor tyrosine kinase (34, 35). The first evidence in support of decorin/EGFR interaction came from cell-free kinase experiments in which immunopurified EGFR was activated in vitro by recombinant decorin proteoglycan or protein core (10). This biological activity was independently confirmed by expression cloning using as a soluble ligand a decorin/alkaline phosphatase chimeric protein (36). Two species of EGFR were isolated: one containing the full-length EGFR and one that lacked the intracytoplasmic domain. A physiologically relevant interaction between decorin and EGFR was further confirmed using the yeast two-hybrid system (36).
In this study we investigated the uptake, internalization, and intracellular fate of the EGFR evoked by decorin, vis à vis the natural ligand EGF. For this purpose, we generated a biologically active proteoglycan chimera containing decorin fused to the enhanced yellow fluorescence protein (DCN-EYFP). Using a combination of biochemical and quantitative immunological approaches, we demonstrate that decorin causes a protracted internalization and degradation of the EGFR primarily through a caveolar-mediated pathway. These findings provide a plausible explanation for the down-regulation of EGFR and the anti-oncogenic properties of decorin, and perhaps other related molecules. and Dulbecco's phosphate-buffered saline (DPBS) were purchased from Mediatech (Herndon, VA). All chemicals, unless otherwise specified, were purchased from Sigma. Recombinant human EGF was purchased from Promega (Madison, WI). Tetramethylrhodamine-conjugated EGF (EGF-Rhodamine) was purchased from Molecular Probes Inc. (Eugene, OR). Recombinant PhiYFP fluorescent protein was obtained from Evrogen (Moscow, Russia); bis-(sulfosuccinimidyl) suberate and sulfosuccinimidyl-4-(p-maleimidophenyl)butyrate (Sulfo-SMPB) were purchased from Pierce. Solutions of cross-linkers were freshly prepared before each experiment. Nitrocellulose membrane was purchased from Bio-Rad. Methyl-␤-cyclodextrin, filipin, and chlorpromazine were purchased from Sigma. Antibodies include polyclonal rabbit antibodies recognizing the N-terminal region of decorin (37), the C-terminal region of the EGFR (sc-03; Santa Cruz Biotechnology, Santa Cruz, CA), caveolin-1 (BD Transduction Laboratories, San Diego, CA), and monoclonal antibodies recognizing ␤-actin (Sigma), phosphotyrosine (PY20; BD Transduction Laboratories), EGFR (Ab-12; NeoMarkers Inc., Union City, CA), early endosome antigen 1 (EEA1), human CD63, Rab11, and p230 trans-Golgi (all from BD Transduction Laboratories). Rhodamineor fluorescein isothiocyanate-conjugated anti-mouse and anti-rabbit IgG were purchased from Santa Cruz Biotechnolog; Cy5-conjugated anti-mouse and anti-rabbit IgG were from Chemicon (Temecula, CA). Horseradish peroxidase-conjugated donkey anti-rabbit and sheep antimouse were purchased from Amersham Biosciences. SuperSignal West Pico chemiluminescent substrate was purchased from Pierce.
Plasmid Construction and Transfection-The enhanced YFP (pIRES-EYFP, Clontech) gene was fused to the C terminus of the human decorin sequence obtained from pDCN1 (38) by standard recombinant techniques. Briefly, the pIRES-EYFP plasmid was used as a template for PCR amplification using the following synthetic oligonucleotide primers: upstream primer with a XhoI site (underlined), 5Ј-CCGCTCGAGAT-GGTGAGCAAGGGCGAGGA-3Ј; downstream primer with a BamHI site (underlined) and a stop codon (bold), 5Ј-CGCGGATCCTTACT-TGTACAGCTCGTCCATGC-3Ј. The 729-bp PCR fragment was ligated into the pCEP-Pu mammalian expression vector (39,40) using XhoI and BamHI restriction sites to create the pCEP-Pu-EYFP vector. pDCN1 was used as a template for PCR amplification with the following primers: upstream primer with a KpnI site (underlined) before the insulin signal peptide sequence, 5Ј-CGGGGTACCATGGCCCTCTG-GATGCGCCT-3Ј; downstream primer with a XhoI site (underlined), 5Ј-CCGCTCGAGCTTATAGTTTCCGAGTTGAATGG-3Ј. The 1113-bp PCR fragment was ligated into the pCEP-Pu-EYFP vector using KpnI and XhoI restriction sites to create the pCEP-Pu-DCN-EYFP vector. The construct was fully sequenced. Human embryonic kidney (293-EBNA) cells were transfected with pCEP-Pu-DCN-EYFP using Lipofectamine (Life Technologies, Inc.). Stable transfectants were obtained by selection with G418 and puromycin (250 and 500 g/ml, respectively). Positive clones expressing high levels of DCN-EYFP were confirmed by detecting endogenous fluorescence and by Western immunoblotting.
Expression and Purification of Recombinant Decorin and DCN-EYFP-Recombinant decorin and DCN-EYFP were produced using either a recombinant vaccinia virus expression system or a stably transfected 293-EBNA cell line. Generation of the recombinant vaccinia virus encoding the mature form of decorin (38) and decorin purification have been described previously (41). DCN-EYFP-expressing 293-EBNA cells were grown to saturation in a CELLine AD 1000 Bioreactor (IBS Integra Biosciences AG, Switzerland), and protein production was achieved by switching to serum-free medium. Details of protein production, purification, and collection are the same as previously published for decorin (41).
Immunofluorescence and Confocal Microscopy-Approximately 5 ϫ 10 4 A431 cells were plated on 4-well chamber slides (Nalge Nunc, Naperville, IL) and grown for 48 h in Dulbecco's modified Eagle's medium, 10% fetal bovine serum at 37°C. Cells were washed once with warm serum-free medium and then incubated with DCN-EYFP, DCN, EGF-Rhodamine, or EGF in serum-free medium at 37°C. Concentrations and incubation times are detailed in the figure legends. Slides were rinsed twice with DPBS and fixed with ice-cold 4% paraformaldehyde in DPBS for 20 min. Cells were permeabilized with DPBS, 0.25% Triton X-100 for 2 min and washed with DPBS, 1% bovine serum albumin. Subsequently, slides were subjected to standard immunofluorescence protocols, and mounted with Vectashield medium (Vector Laboratories Inc., Burlingame, CA). Images were acquired using an Olympus BX51 microscope driven by SPOT Advanced version 4.0.9 imaging software (Diagnostic Instruments Inc.), and a Zeiss Axiovert 200M microscope (Zeiss LSM 5, Carl Zeiss Inc., Oberkochen, Germany). Metamorph 5.0 imaging software (Universal Imaging Corp.) was used for image acquisition and analysis. All dual-or triple-labeled samples were analyzed using filters set at 484 (YFP), 555 (rhodamine), and 620 nm (Cy5). Confocal analysis was performed on a Zeiss LSM 510 META laser scanning confocal microscopy system (Carl Zeiss Inc.) driven by the Zeiss LSM 510 image software. Beam splitters were set at 405 (4,6-diamidino-2phenylindole), 488 (YFP), 543 (rhodamine), and 633 nm (Cy5). All the images were analyzed using the software mentioned above along with Adobe Photoshop (version 9.0, Adobe Systems, San Jose, CA).
Image Quantification-Collected images were analyzed using Adobe Photoshop. Briefly, images of single cells were isolated, and total fluorescent signal was quantified for individual fluorescence channels using the Threshold and Histogram tools. Single cells were then operationally divided into three regions of interest: 1) plasma membrane; 2) intracellular region; and 3) perinuclear region (nucleus included). These areas were subsequently masked and each region of interest was individually quantified. Results were plotted as a percentage of total signals.

RESULTS AND DISCUSSION
Decorin Causes a Decrease in EGF-independent (Preformed) EGFR Dimers as Well as EGFR Monomers-It is well established that EGFR dimers represent the active subpopulation of this receptor tyrosine kinase family (35,42,43). Upon EGF binding, EGFR monomers dimerize, leading to autophosphorylation of various tyrosine residues within the cytoplasmic domain, and subsequent activation of the signal transduction cascade. We investigated whether decorin would affect EGFR dimerization in the well studied A431 human squamous carcinoma cells, which overexpress wild type EGFR. Decorin has been shown to be effective on these cells both in vitro, by down-regulating EGFR phosphorylation (11), and in vivo, by retarding the growth of A431 tumor xenografts (6,30). Cells were exposed to recombinant decorin protein core (0.67 M) or EGF (16 nM) for various intervals and then treated for the last 20 min with the membrane-impermeable chemical cross-linker Sulfo-SMPB (0.5 mM) (41). Decorin protein core caused a reduction in the amount of both monomers and dimers (Fig. 1A). Quantitative analysis of three independent experiments showed that decorin protein core caused a significant and progressive decline in the amount of ligandindependent (preformed) dimers, with a t1 ⁄ 2 of ϳ45 min (Fig. 1B). The EGFR dimer levels reached a plateau at 1 h (70% reduction) and this did not change for up to 6 h. Interestingly, the EGFR monomer amount also declined with similar kinetics; however, its levels partially recovered and reached a plateau at later time points remaining at ϳ60% of control EGFR levels (Fig. 1C). Comparable results were obtained using recombinant decorin proteoglycan (not shown). In contrast, EGF caused a rapid (t1 ⁄ 2 ϳ 2 min) and substantial (ϳ7-fold) increase in EGFR dimers, and this did not appreciably change at later time points (Fig. 1D). In the presence of EGF, the amount of EGFR monomers declined in the first 30 min, but subsequently increased and thereafter remained relatively constant at ϳ90% of control levels (Fig. 1E). These data indicate that decorin protein core, independent of its glycosaminoglycan moiety, affects the intracellular fate of EGFR monomers and dimers in a way that differs from EGF.
Decorin Causes a Slow and Progressive EGFR Degradation-A plausible interpretation of the results presented above is that decorin affects the equilibrium of ligand-independent EGFR population by decreasing the amount of dimers and monomers via intracellular degradation. To further investigate this possibility, we incubated A431 cells with decorin core or EGF for the same time points as above (0.5-6 h) but without any cross-linking reagent. Surprisingly, total EGFR did not rapidly decline upon incubation with the decorin core ( Fig. 2A). In the constant presence of decorin, total EGFR levels declined slowly during the first 2 h of incubation and later on began to drop significantly, reaching a plateau at ϳ3 h (Fig. 2B). Notably, the presence of exogenous decorin core led to a protracted reduction by ϳ40% of total EGFR content that lasted for up to 6 h. In contrast, EGF treatment caused a significant and faster decline in total EGFR followed by a nearly full recovery (Fig. 2, A and B). The decorin-induced reduction in EGFR is in agreement with the estimates derived from A431 clones stably expressing decorin. In this study we found three individual clones with ϳ40% EGFR levels of control wild-type cells (11). In addition, we found that the average number of 125 I-EGF binding sites in three independent clones expressing decorin was 1.3 ϫ 10 6 Ϯ 0.2 receptors per cell as compared with 2.4 ϫ 10 6 Ϯ 0.1 receptors per cells in the wild-type A431 (11). The dissociation constants determined by Scatchard analysis were not significantly different among the control and decorin-expressing A431 cells, indicating that decorin did not appreciably change EGFR affinity for EGF. Collectively, these results indicate that decorin induces EGFR degradation via a slow pathway that, in contrast to that evoked by EGF, does not lead to recovery.
Decorin Can Be Cross-linked to the EGFR and Forms Large Supramolecular Aggregates-After comparing the results shown in Fig. 2, A and B, with the cross-linking data shown in Fig. 1, an apparent discrepancy was noted in the amount of total EGFR detected upon decorin core treatment, especially at the earlier time points, 30 min and 1 h (compare Figs. 1C and 2B). Upon cross-linking and Western immunoblotting analyses, ϳ50% of total EGFR was unaccounted for. These data suggest that in the cross-linking experiments a proportion of the EGFR could not be detected, possibly because it was migrating in larger polymeric complexes, which either did not penetrate the 3-15% gradient SDS-PAGE or did not efficiently transfer onto the membrane. Cross-linking experiments were repeated and, indeed, a longer exposure revealed an EGFR immunoreactive band migrating slower than the EGFR dimer ( Fig. 2C, arrow). It is well known that EGFR exists in a monomeric form (175 kDa) and in several aggregated states ranging between 350 and Ͼ500 kDa (44). Thus, to determine whether decorin core could also be a component of EGFR-containing high molecular mass complexes, we probed the high molecular mass region (Ͼ350 kDa) with an antidecorin antibody. A significant amount of decorin was cross-linked with EGFR dimers (Fig. 2D, arrow), and higher order oligomers. These decorin-containing aggregates were not present in the absence of decorin or when the cells were challenged with EGF ( Fig. 2D), indicating specificity of the decorin signal. Therefore, our results show for the first time a covalent cross-linking between decorin protein core and EGFR dimers and oligomers. Considering that the cross-linker used (Sulfo-SMPB) has an arm length of 14.5 Å, decorin core had to be in very close proximity to the ligand-binding region of the EGFR. By plotting the relative mobility of protein standards from several blots, we estimated the molecular mass of cross-linked decorin-EGFR complexes to be ϳ450 kDa (Fig. 2D, arrow). This would correspond to an EGFR dimer (ϳ350 kDa) and two decorin core molecules (ϳ100 kDa), suggesting a stoichiometry of 1:1 decorin core:EGFR monomer. These data provide an explanation for the apparent loss of EGFR after cross-linking in the presence of decorin core. Most likely, EGFR epitopes would be masked in the immunoblottings by the formation of EGFR-decorin complexes. Our results, however, cannot rule out the presence of other cross-linked cell surface proteins in addition to decorin and EGFR in the larger supramolecular aggregates with molecular mass Ͼ500 kDa.
Decorin Inhibits EGF-dependent EGFR Dimerization-The previous results raised the possibility that decorin could directly interact with the EGFR by forming supramolecular complexes. Previous studies using the yeast two-hybrid system have shown that decorin binds to the EGFR within its ligand-binding L2 domain (36). The binding site partially overlaps with but is distinct from the EGF-binding domain. Thus, we determined whether decorin would compete with EGF in inducing the initial step of EGFR activation, that is, receptor dimerization. In these experiments the cells were first challenged with decorin core (0.6 M) for either 2 or 5 min and then with EGF (16 or 8 nM) for 10 min. After incubation, cells were exposed to the impermeable cross-linker bis-(sulfosuccinimidyl) suberate (0.5 mM). A representative immunoblotting is shown in Fig. 3A. Densitometric analysis of the immunoblot bands revealed that EGFR dimer was reduced by ϳ30 and 20% at 8 and 16 nM EGF, respectively, with an average ϳ25% inhibition (Fig. 3B). In another set of experiments, cells were pretreated with decorin core (0.6 M) for 30 min and then with EGF (8 or 1.6 nM) for 10 min. In this case, there was 30 and 70%, respectively, inhibition of EGFR dimer formation (Fig. 3C). Collectively, these results indicate that decorin inhibits EGFdependent EGFR dimerization in a manner dependent on EGF concentration. These data raise the possibility that decorin could compete with traditional ligands, such as EGF or TGF-␣, for their binding to EGFR. This might represent an additional mechanism through which decorin could control tumor cell growth.
DCN-EYFP Induces EGFR Internalization and Partially Co-localizes with the Receptor in Intracellular Vesicles-To further investigate the role of decorin in EGFR internalization, we generated a chimeric pro- teoglycan containing decorin fused to an EYFP moiety (Fig. 4A), and used it to visualize the interaction of decorin on the plasma membrane and inside the cells by fluorescence microscopy. After selection, the EBNA-293 clone DY8 showed the highest EYFP expression (Fig. 4B). The secreted chimeric proteoglycan was of the predicted size migrating just before decorin proteoglycan (Fig. 4C, lane 3) with a molecular mass of ϳ130 kDa (ϳ100 kDa decorin and ϳ30 kDa EYFP). Western immunoblotting, using an anti-decorin antibody, recognized both decorin and DCN-EYFP (Fig. 4C, lanes 4 and 5), and the latter could be detected by fluorography (Fig. 4C, lane 6). Next, we tested the biological activity of DCN-EYFP by measuring its ability to suppress EGFR phosphorylation (11,41). Western immunoblotting and quantitative analysis showed that DCN-EYFP was fully functional and capable of inhibiting EGFR phosphorylation to levels comparable with those evoked by the native decorin proteoglycan (Fig. 4, D and E).
The DCN-EYFP was then tested in A431 cells, which show a rapid EGFR internalization induced by EGF (45,46). In the absence of exogenous DCN-EYFP, most of the EGFR was localized on the plasma membrane (Fig. 5A, 0 min panel). After 30 min of DCN-EYFP incubation, most signal was diffusely present on the plasma membrane and within punctuate intracellular regions. Notably, some of the EGFR-containing vesicles clearly co-localized with DCN-EYFP (Fig. 5A, arrows). After 60 min of continuous incubation, DCN-EYFP was mostly concentrated in the perinuclear area and some endocytosed EGFR still co-localized with the DCN-EYFP signal. Similar results were obtained after a 120-min incubation: co-localization of EGFR and DCN-EYFP in small vesicles was observed near the inner leaflets of the plasma membrane (Fig. 5B,  upper panel). These vesicles became larger around the nucleus and showed EGFR surrounding DCN-EYFP, forming a "doughnut-shaped" structure (Fig. 5B, bottom panels, arrows). These results suggest the possibility that the EGFR might still be intercalated in the membrane of the endocytic vesicles, whereas decorin is located within their lumina.  Similar results were obtained by pulsing with DCN-EYFP for 30 min followed by chasing for 10 or 30 min with serum-free medium (not shown).
Next, we addressed the issue of specificity, insofar as the EYFP moiety, linked to the C terminus of decorin, could have been dissociated from the decorin protein core resulting in a misleading interpretation of our results. To this end, cells that were incubated with DCN-EYFP for 60 min were reacted with an antibody against the N terminus of decorin followed by fluorescence-labeled secondary antibody. Signal merge showed that the pattern of fluorescence using the antibody against decorin was identical to the one from the EYFP channel (Fig. 5C). Note that nonspecific staining, because of the secondary antibody, was present in the nucleus. The fact that the EYFP signal co-localized with decorin led us to conclude that decorin is responsible for the effects shown in Fig. 5. Furthermore, DCN-EYFP appears to be relatively stable after internalization.
To further confirm the issue of specificity, we performed competition experiments by incubating A431 cells with a constant amount of DCN-EYFP (0.4 M) and 20-and 30-fold molar excess of decorin protein core. The results showed a nearly complete block of binding and internalization when decorin protein core was used, indicating that the major binding was indeed provided by decorin protein moiety (Fig. 6). Finally, we incubated A431 cells with recombinant PhiYFP, a yellow fluorescent protein highly homologous to EYFP, and found no significant binding (not shown). In summary, using DCN-EYFP chimeric proteoglycan and fluorescence microscopy, we show for the first time in living cells that decorin causes a protracted internalization of the EGFR, that this process is mediated by decorin protein core, and that both the receptor and DCN-EYFP co-localize within endocytic vesicles.
Reduced EGFR Levels Diminish DCN-EYFP and EGF Binding and Endocytosis-To further test the specificity of the DCN-EYFP/EGFR interaction, we selected A431 clones expressing low EGFR, by chronically exposing them to a high concentration (100 ng/ml) of EGF. These concentrations are known to be toxic for wild-type A431 cells and allow for selection of clones that express low levels of EGFR (47,48). After several weeks of continuous culture in the presence of EGF, several surviving clones were isolated. Western immunoblotting showed that clones A10 and B6 contained ϳ40% of the EGFR expressed by the wildtype cells (Fig. 7A). Wild-type A431 and clone A10 cells were incubated with DCN-EYFP for 30 min at 37°C or EGF-Rhodamine for 10 min. Binding and internalization of both DCN-EYFP and EGF were significantly reduced in clone A10 (Fig. 7B). Similar results were obtained with clone B6 (not shown). Thus, decorin binding and internalization is quantitatively dependent on EGFR expression in a way similar to EGF. These data represent further evidence that EGFR is a direct partner for decorin binding on the cell surface. (35, 49 -51). In general, EGFR movement from caveolae to clathrin-coated pits is considered to be the initial event in internalization and signaling (35), and there is evidence that its receptor FIGURE 5. DCN-EYFP induces EGFR internalization and partially co-localizes with the receptor in intracellular vesicles. A, A431 cells were incubated with DCN-EYFP (0.78 M) for 30 and 60 min at 37°C, fixed with 4% paraformaldehyde, permeabilized with 0.25% Triton X-100, and then incubated with a monoclonal anti-EGFR antibody, followed by a rhodamine-conjugated secondary antibody. Rhodamine signal (red) and YFP fluorescence (green) were visualized using a live cell fluorescence microscopy system. Arrows indicate vesicles in which EGFR and DCN-EYFP signals co-localize or merge (orange/yellow). B, high magnification images of an experiment as in A, after a 120-min DCN-EYFP incubation. Arrows indicate vesicles in which EGFR and DCN-EYFP signals co-localize. C, EYFP moiety co-localizes with decorin protein core. Following a 60-min incubation with DCN-EYFP, the cells were incubated with an anti-decorin antibody followed by a Cy5-conjugated secondary antibody. The merge image shows co-localization of EYFP and decorin core signals (orange-yellow). Note that the EYFP moiety is at the C terminus of the DCN-EYFP proteoglycan chimera, whereas the anti-decorin antibody recognizes an N-terminal epitope in the decorin core protein. Scale bars ϭ 10 m. tyrosine kinase activity is required for proper EGFR internalization (52). Therefore, we tested whether EGFR internalization evoked by decorin would also require EGFR tyrosine kinase activity, by using the tyrphostin AG1478, a highly specific and selective inhibitor of EGFR tyrosine kinase when used at low dosages (53). We have previously shown that incubation of A431 cells with AG1478 (1-5 M) completely blocked EGF-induced EGFR phosphorylation without affecting platelet-derived growth factor receptor phosphorylation induced by platelet-derived growth factor (9). A431 cells were preincubated with AG1478 (5 M) for 60 min, and then incubated with EGF-Rhodamine or DCN-EYFP. Following a 10-min pulse with EGF-Rhodamine and a 10-min chase, the internalization of both ligand and receptor was almost completely blocked by AG1478 (Fig. 7C, lower panels). Both EGF and EGFR were located at the plasma membrane in contrast to control samples incubated without AG1478 (Fig. 7C, arrows). These results are in agreement with the AG1478-mediated inhibition of both EGF and EGFR internalization in Chinese hamster ovary cells transfected with EGFR (54). In this case, treatment with AG1478 caused a stepwise and dose-dependent inhibition of EGFR internalization as monitored by quantification of both endocytosed EGF and remaining cell surface EGFR (54).

Decorin-induced EGFR Internalization Does Not Require EGFR Tyrosine Kinase Activity-In eukaryotic cells, internalization of cell surface receptors and their ligands occurs via clathrin-dependent and -independent pathways
In contrast, AG1478 did not appreciably block decorin-induced EGFR internalization. Following a 30-min pulse with decorin-EYFP and a 30-min chase in serum-free medium (longer incubation periods were used because of the slower kinetics of decorin), numerous large vesicles containing both decorin and EGFR were noted; these vesicles were primarily located in the perinuclear area (Fig. 7D, arrows). These results indicate that the EGFR internalization evoked by decorin differs from that induced by EGF, and that EGFR tyrosine kinase is not required for EGFR endocytosis following decorin challenge.
Decorin Causes Internalization of EGFR via Caveolae-The previous results raised the possibility that EGFR internalization evoked by decorin could involve a clathrin-independent pathway. Indeed, the association of the EGFR with lipid rafts and caveolae can affect receptor sequestration and trafficking in a number of cells (55)(56)(57)(58). Caveolae are cholesterol-rich, detergent-insoluble microdomains of the plasma membrane that are enriched in cell surface receptors, such as EGFR (58), although in A431 cells EGFR may not be concentrated in caveolae but rather randomly distributed in the plasma membrane (59). Interestingly, AG4178-mediated inhibition of EGFR tyrosine phosphorylation in glioblastoma cells increases receptor association with caveolae (57). . Effects of EGFR density and tyrosine kinase activity on decorin-and EGF-induced EGFR internalization. A, Western immunoblotting of total EGFR from A431 cells chronically exposed to high (100 ng/ml) EGF concentration. Notice that clones A10 and B6 contain ϳ40% of the EGFR expressed by the wild-type A431. B, fluorescence microscopy showing DCN-EYFP and EGF-Rhodamine internalization in wild type A431 cells and clone A10. Notice the marked reduction in binding and uptake of both ligands in the cells with lower EGFR numbers. C, confocal microscopy images showing that AG1478 inhibits EGF/EGFR endocytosis. A431 cells were preincubated with AG1478 (5 M) for 60 min and then incubated with EGF-Rhodamine (3 nM) for 10 min, followed by a 10-min chase at 37°C. Cells were fixed, permeabilized, and then incubated with a monoclonal anti-EGFR antibody, followed by a fluorescein isothiocyanate-conjugated secondary antibody. Pretreatment with AG1478, a specific inhibitor of EGFR tyrosine kinase, prevents the internalization of EGF-EGFR complexes (arrows) but not the binding of the EGF to the cell surface. D, confocal microscopy images showing that AG1478 does not significantly inhibit DCN-EYFP/EGFR endocytosis. A431 cells were preincubated with AG1478 (5 M) for 60 min and then incubated with DCN-EYFP (0.78 M) for 30 min, followed by a 30-min chase at 37°C. Cells were fixed, permeabilized, and processed as in C. Notice that in contrast to EGF, decorin-induced endocytosis of the EGFR is not appreciably inhibited by the tyrphostin (arrows). Scale bars ϭ 10 m.
Thus, we first investigated whether exogenous decorin would cause a redistribution of caveolin-1, one of the major integral caveolar proteins that functions as a scaffolding matrix to organize and concentrate numerous signaling and adaptor molecules within caveolae (58). A431 cells were treated with DCN-EYFP at 37°C for different time points. At time 0, most caveolin-1 was present in the perinuclear area, presumably in the trans-Golgi network, and some was diffusely distributed on the plasma membrane and cytoplasm as small scattered spots (Fig. 8A). After incubation with DCN-EYFP for 15 min, most caveolin-1 migrated to the inner side of the plasma membrane (Fig. 8A, 15 min, arrows) and some accumulated to form larger spots (arrowheads). After 30 min, large vesicles were internalized and began accumulating in the perinuclear region (Fig. 8A, 30 min, arrowheads). After 60 min, most vesicles accumulated in the perinuclear area, few were located in the cytoplasm, and even fewer were associated with the plasma membrane. Quantification of the changes in caveolin-1 distribution showed a rapid decline (15 min) of the perinuclear caveolin-1-rich endocytic vesicles and a concurrent enrichment of cell surface vesicles (Fig. 8B). Within the next 45 min, the distribution of the perinuclear and cell surface caveolin-1positive vesicles returned to nearly basal levels. In contrast, the intracellular (i.e. cytoplasmic) fraction of the caveolin-positive vesicles did not appreciably change with time (Fig. 8B). In summary, decorin treatment causes a rapid (15 min) displacement of caveolin-1 from the trans-Golgi area to the plasma membrane, followed by formation of large caveolinpositive structures inside the cell. These results support the concept that decorin-induced caveolin-1 redistribution might be closely linked to the EGFR internalization pathway. Indeed, confocal microscopy analysis showed partial, but significant, co-localization of DCN-EYFP, EGFR, and caveolin-1 in the perinuclear area (Fig. 8, C and D, arrows). In contrast, little or no co-localization of EGFR with caveolin-positive vesicles was observed when EGF was used (not shown). The results clearly show that the endocytic vesicles containing decorin and EGFR are enriched in caveolin-1 and are located in the perinuclear region.
Cholesterol Depletion Affects Decorin-induced EGFR Internalization-To further address the role of caveolae in the internalization of decorin-EGFR complexes, we utilized methyl-␤-cyclodextrin (M␤CD). M␤CD is a water-soluble cyclic heptasaccharide that binds cholesterol with high specificity and, because of its ability to deplete plasma membrane cholesterol, has been utilized as an agent to disrupt membrane rafts (60). Caveolae formation on the cell surface is critically dependent on the presence of cholesterol and its compartmentalization. Therefore, in the presence of M␤CD, caveolin-dependent endocytosis can be selectively inhibited (51). Interestingly, M␤CD triggers ligand-independent activation of the EGFR in several cells including A431 cells (57,59,61). Utilizing relatively low concentrations of M␤CD (2.5 mM) (higher concentrations, 5 and 10 mM, were found to be toxic to the cells) we discovered that, in contrast to EGF, decorin-induced internalization of the EGFR was markedly blocked by the drug (Fig. 9). Following a 30-min incubation in the presence of M␤CD, most of decorin-EGFR complexes were located along the plasma membrane (Fig. 8, lower middle panel); in contrast, following a 10-min incubation with EGF, the EGF-EGFR complexes were properly internalized and reached perinuclear vesicles (Fig.  9, top middle panel).
To further investigate decorin-induced caveolae-mediated EGFR internalization, we used the drug filipin, which is known to bind and sequester cholesterol in the plasma membrane and impair the invagination of caveolae, thereby inhibiting their internalization (62,63). Pretreatment of cells with 5 g/ml filipin for 60 min blocked endocytosis of both EGF-Rhodamine and DCN-EYFP, without blocking the binding to EGFR on the cell surface (not shown). When we lowered the filipin concentration to 2 g/ml (30 min pretreatment), we obtained a result comparable with M␤CD. DCN-EYFP co-localized with EGFR on the plasma membrane and the decorin-EGFR complexes were not internalized. In contrast, EGF-induced EGFR endocytosis was not affected (not shown).
Decorin-induced Internalization of the EGFR Is Not Affected by Chlorpromazine-Next, we tested the effects of chlorpromazine, a cationic amphiphilic drug that prevents the assembly of clathrin adapter protein AP-2 on clathrin-coated pits, and also blocks receptor recycling (64). Chlorpromazine causes the loss of coated pits from the surface of human fibroblasts and the concurrent appearance of clathrin coats on endosomal membranes (64). In A431 cells, the internalization of EGF-EGFR complexes was markedly reduced by 30 min incubation with relatively low dosages (7.5 M) of chlorpromazine (Fig. 9, top right panel). Higher concentrations (15 and 30 M) of the drug induced significant morphological changes and, therefore, were not utilized. In contrast, the decorin-EGFR complexes were properly internalized and reached the perinuclear region within 30 min of incubation (Fig. 9,  bottom right panel). Collectively, these results support the concept that cholesterol-rich membrane microdomains and caveolae are necessary for proper EGFR endocytosis induced by decorin, and that this pathway, in contrast to EGF, does not significantly involve clathrin-coated pits.
Decorin Does Not Target the EGFR to Either Recycling Endosomes or the Golgi-It has been reported that caveolae-mediated endocytosis can directly target the vesicles to the Golgi apparatus (65). Therefore, we tested whether decorin would target EGFR to the Golgi apparatus and then to secretory vesicles for recycling to the cell surface. A431 cells were incubated with EGF (3 nM) or DCN-EYFP (0.78 M) for 10 and 30 min, respectively, at 37°C, and then chased for various time points. The cells were then reacted with antibodies directed against EGFR and Rab11, a 24-kDa GTPase originally isolated from the Golgi-microsomal FIGURE 9. Effects of cholesterol depletion and inhibition of clathrin-mediated endocytosis on the EGF-and decorin-induced EGFR internalization. Immunofluorescence images of A431 cells preincubated with either methyl-␤-cyclodextrin (2.5 mM) or chlorpromazine (7.5 M) for 30 min, and then treated with EGF-Rhodamine (3 nM) or DCN-EYFP (0.78 M) for 10 and 30 min, respectively. Incubation was performed in the presence of the inhibitors. Cells were fixed, permeabilized, and subjected to immunofluorescence using an anti-EGFR monoclonal antibody, followed by fluorescein isothiocyanate-or rhodamine-conjugated secondary antibodies. For simplicity, only the merged images are shown with EGF-Rhodamine and DCN-EYFP (red) and EGFR (green). The orange-yellow staining is indicative of co-localization. Notice the differential blockage of EGFR internalization (arrows) induced by the two drugs in the EGF-(top panels) and decorin-treated (bottom panels) cells. Scale bar ϭ 10 m. fraction and commonly used as a marker for both trans-Golgi network and recycling endosomes (66). Confocal microscopy revealed that EGF induced extensive localization of EGFR into the recycling Rab11 ϩ compartment (Fig. 10A). In contrast, DCN-EYFP evoked no significant colocalization of EGFR and Rab11 (Fig. 10B). These findings were further corroborated by the observation that DCN-EYFP did not co-localize with the protein p230 (67), a specific marker for the trans-Golgi network (Fig. 10C). This is a key compartment for the endocytic/secretory pathway from which proteins are sorted either for constitutive transport to the late endosomes or for recycling to the cell surface. Thus, decorin targets the EGFR to a compartment other than the Golgi apparatus (most likely the "caveosome," see below) and does not cause significant recycling of the receptor to the cell surface, even after a 2-h, 30-min pulse and 90-min chase (not shown).
Internalized EGFR Induced by Decorin Does Not Localize to Early Endosomes-To further investigate the decorin-induced EGFR endocytosis pathway, we tested for the presence of EGF-Rhodamine⅐EGFR and DCN-EYFP⅐EGFR complexes in early endosomes. To label early endosomes, we used an antibody against EEA1, a 180-kDa peripheral membrane protein that resides within early endosomes (68). Typically, entrance to EEA1 ϩ compartments occurs through clathrin-mediated endocytosis (68), and transmembrane proteins in the EEA1 ϩ compartment can either enter the late endosomes/lysosomes for degradation or the Rab11 ϩ recycling endosomes (69). A431 cells were incubated with EGF-Rhodamine (3 nM) for 10 min and chased for 10 min, or DCN-EYFP (0.78 M) for 30 min followed by a 30-min chase. Confocal microscopy revealed that both EGF-Rhodamine and EGFR largely co-localized within EEA1 ϩ endosomes (Fig. 11A, merge). This is in line with the fact that EGF remains associated with the endocytosed EGFR down to late endosomes, and that it is released from the receptor at very low pH within lysosomes (35). In contrast, DCN-EYFP and the internalized EGFR evoked by DCN-EYFP only minimally localized within EEA1 ϩ vesicles (Fig. 11B). This corroborates our finding presented above that DCN-EYFP does not appreciably target the endocytosed EGFR to the Rab11 ϩ recycling compartment (Fig. 10B).
Decorin and Late Endosomes-Next, we investigated whether decorin-induced EGFR internalization would lead to late endosome localization, and subsequently to receptor degradation. As a marker for the late endosome/lysosome compartment, we used an antibody against CD63, a 53-kDa type III lysosomal glycoprotein (also known as gp55). CD63 is a member of the tetraspan transmembrane 4 superfamily and has been previously described to be a specific protein of the late endosome/lysosome compartment (70). Following a 30-min incubation and 90-min chase, DCN-EYFP induced a significant co-localization of EGFR within CD63 ϩ endosomes, in a manner comparable with that evoked by EGF-Rhodamine (Fig. 12). Further confocal microscopy analysis showed that both EGF-Rhodamine and DCN-EYFP were also present in late endosomes, presumably targeted for final degradation in the lysosomes (not shown). These results indicate that decorin evokes a protracted down-regulation of the EGFR via a caveosome/late endosome/ lysosome axis and provide a plausible explanation for the reduced EGFR . Immunofluorescence was performed using antibodies against the EGFR or Rab11, a marker of recycling endosomes, followed by appropriate secondary antibodies. EGFR and Rab11 co-localization is shown in the merged image (orange-yellow). Insets: higher magnification of areas in the white square. C, confocal microscopy showing lack of DCN-EYFP localization in the Golgi apparatus. A431 cells were incubated with DCN-EYFP (0.78 M) for the times indicated. To visualize the Golgi we used a monoclonal antibody against p230, a specific marker for the trans-Golgi network, followed by 4,6-diamidino-2-phenylindole (DAPI) staining. Images represent sequential sections along the z axis (1.74, 2.18, 3.49, 3.92 and 4.79 m from left to right, respectively). Notice the lack of co-localization of endocytosed decorin within the Golgi even after 2 h. Scale bars ϭ 5 m. FIGURE 11. Internalized EGFR induced by decorin does not localize to early endosomes. A and B, confocal microscopy images of A431 cells incubated with EGF-Rhodamine or DCN-EYFP (same concentrations as in Fig. 9) for the times indicated. Cells were immunostained for EGFR (red) or EEA1 (green), a marker for early endosomes. Fluorescent EGF-Rhodamine or DCN-EYFP were detected using appropriate filters. The merged image shows co-localization vesicles (yellow). Insets: high magnification images of the areas in the white rectangles. Both EGF and EGFR extensively co-localize with EEA1 ϩ endosomes (A), in contrast to decorin-EGFR complexes, which only minimally co-localize with EEA1 ϩ endosomes (B). Nu, nucleus. Scale bars ϭ 10 m. content in A431 cells exposed to recombinant decorin or when decorin is endogenously produced by transgenic experiments (11).
Conclusions-In this study, we report a novel mechanism of action induced by a soluble proteoglycan, decorin, in controlling EGFR, an important receptor tyrosine kinase widely implicated in the pathogenesis and progression of cancer. Collectively, our results demonstrate that decorin is internalized through a unique endocytic pathway that is qualitatively and topographically different from that induced by EGF in the same in vitro system. A working model summarizing the present results is provided in Fig. 13. Accordingly, decorin causes a slow and sustained internalization of the EGFR primarily via caveolae-mediated endocytosis. Decorin targets the EGFR to a distinct intracellular compartment, most likely the "caveosome," because decorin-induced EGFR internalization does not appreciably involve EEA1 ϩ early or Rab11 ϩ recycling endosomes. Eventually, EGFR is transferred to the CD63 ϩ late endosome compartment, presumably for final degradation by acidic hydrolases. In contrast, EGF induces the canonical and well established EGFR endocytosis primarily via the clathrin-dependent pathway that utilizes EEA1 ϩ early endosomes and finally CD63 ϩ late endosome/lysosomes. Some internalized EGFR can then be recycled to the plasma membrane via Rab11 ϩ endosomes.
We note that decorin causes an inhibition of EGFR phosphorylation (11) in addition to receptor down-regulation. This is noteworthy because a low phosphorylation state of the EGFR is associated with increased receptor localization to caveolae and attenuation of the transformed phenotype (57). Thus, strategies that induce EGFR association with caveolae may be beneficial in suppressing aberrant signaling events in various tumor systems. Our findings confirm and expand this concept and further stress that caveolar-mediated endocytosis of the EGFR, and perhaps other receptors, can be an additional anti-oncogenic mechanism that could be operational in vivo. The recent discovery that decorin also signals via the insulin-like growth factor receptor (7) suggests that this proteglycan may have a broader bioactivity.
There are two recent examples of caveolar involvement in receptor control. The first is provided by the transforming growth factor-␤ (TGF-␤) receptor, which utilizes both caveolae and clathrin-coated pits for its internalization (71). Notably, the latter pathway promotes TGF-␤ signaling, whereas the former pathway leads to receptor degradation, as in the case of decorin. Thus, segregation of the TGF-␤ receptor within distinct endocytic compartments regulates TGF-␤/Smad signaling and receptor turnover. The second example is provided by the EGFR, which can be diverted to either clathrin-or caveolar-dependent endocytosis based on the amount of exogenous EGF. At low (0.25 nM) EGF concen-trations the EGFR is preferentially internalized via the clathrin pathway and is not ubiquitinated, whereas at higher (3 nM) EGF concentrations the receptor is preferentially endocytosed via caveolae and is ubiquitinated (72). We should note that although we used high (3 nM) EGF concentrations in most of the confocal and immunofluorescence studies, we still found that the majority of the EGFR was preferentially endocytosed via a clathrin-dependent pathway. This might be because of the differences in receptor numbers between the A431 cells used in the present study, which we showed previously to express 2.4 ϫ 10 6 receptors per cell (11), and HeLa cells, which express ϳ5 ϫ 10 4 EGFRs/cell (72). Whereas "quantitative" differences in ligand might evoke diverse endocytic pathways, we show that "qualitative" differences in ligand are also important in dictating the intracellular fate of the EGFR. We conclude that intracellular routing of a receptor tyrosine kinase can be modulated by the specificity of the ligand and the diverse events emanating from the ligand/receptor interaction. FIGURE 13. Schematic model depicting the divergent effects of decorin and EGF on the internalization and sorting of the EGFR. The intracellular EGFR fate evoked by exogenous decorin differs significantly from that induced by EGF in A431 squamous carcinoma cells. Decorin causes a slow and sustained internalization of the EGFR primarily via caveolae-mediated endocytosis (Cav1 ϩ , EEA1 Ϫ endosomes). From here the decorin-EGFR complexes are transferred to large perinuclear vesicles, which we interpret as representing caveosomes. Ultimately, the EGFR is transferred to late endosomes (CD63 ϩ ) for final degradation. In contrast, EGF induces the canonical EGFR endocytosis, primarily via a clathrin-dependent pathway that includes early endosomes (Cav1 Ϫ , EEA1 ϩ ) and late (CD63 ϩ ) endosomes. Some of the EGFR also enters the recycling pathway via Rab11 ϩ endosomes. One of the major consequences of this is that decorin causes a protracted internalization and degradation of the EGFR (a constant ϳ40% reduction in total EGFR for several h) and decreases the number of EGF binding sites on the cell surface, thereby promoting growth suppression and anti-oncogenic activities.