Extracellular Matrix Metalloproteinase Inducer Stimulates Tumor Angiogenesis by Elevating Vascular Endothelial Cell Growth Factor and Matrix Metalloproteinases

Matrix metalloproteinases (MMPs) are endopeptidases that play pivotal roles in promoting tumor disease progression, including tumor angiogenesis. In many solid tumors, MMP expression could be attributed to tumor stromal cells and is partially regulated by tumor-stroma interactions via tumor cell–associated extracellular matrix metalloproteinase in-ducer (EMMPRIN). The role of EMMPRIN during tumor angiogenesis and growth was explored by modulating EMMPRIN expression and activity using recombinant DNA engineering and neutralizing antibodies. In human breast cancer cells, changes in EMMPRIN expression influenced vascular endothelial growth factor (VEGF) production at both RNA and protein levels. In coculture of tumor cells and fibroblasts mimicking tumor-stroma interactions, VEGF expression was induced in an EMMPRIN-and MMP-dependent fashion, and was further enhanced by over-expressing EMMPRIN. Conversely, VEGF expression was inhibited by suppressing EMMPRIN expression in tumor cells, by neutralizing EMMPRIN activity, or by inhibiting MMPs. In vivo, EMMPRIN overexpression stimulated tumor angiogenesis and growth; both were significantly inhibited by antisense suppression of EMMPRIN. Expression of both human and mouse VEGF and MMP, derived from tumor and host cells, respectively, was regulated by EMMPRIN. These results suggest a novel tumor angiogenesis mechanism in which tumor-associated EMMPRIN functionally mediates tumor-stroma interactions and directly contributes to tumor angiogenesis and growth by stimulating VEGF and MMP expression.


Introduction
Tumor angiogenesis is a complex and multistep process (1).Critical steps during tumor angiogenesis are the outgrowth of endothelial cells from preexisting capillary vessels initiated by the migration of endothelial cells away from the parental vessels.Only after the dissociation of these endothelial cells can they proliferate in response to various angiogenic growth factors, including vascular endothelial growth factor (VEGF).Proliferating endothelial cells subsequently remodel the extracellular matrix around neovasculature sites, align into tube-like structures, and eventually form new functional blood vessels.Extracellular matrix remodeling occurs continuously throughout the tumor angiogenic process in a well-orchestrated fashion involving numerous extracellular matrixdegrading enzymes.Among them, matrix metalloproteinases (MMPs) are believed to be a critical group of enzymes that affect tumor angiogenesis, tumor growth, local invasion, and subsequent distant metastasis (2,3).MMPs are a family of metal-dependent endopeptidases sharing a common modular domain structure, which are capable of cleaving all of the extracellular matrix components of the parenchymal and vascular basement membranes, normally mechanical barriers to cell migration and invasion.MMPs are often overproduced in the tumor local environment and high levels of MMPs have been correlated with tumor vascular density (2,3).In experimental animal models, MMP activity has been directly associated with tumor angiogenic potential (4,5).
Examination of MMP expression patterns in clinical tumor specimens revealed that the majority of these enzymes were produced by peritumoral fibroblasts in the stromal compartment rather than by tumor cells.These fibroblasts produce tumor-associated MMP-1, MMP-2, MMP-3, and MMP-11 in breast, colon, lung, skin, and head and neck cancers (6)(7)(8).This expression pattern could be at least partially attributed to a 58 kDa tumor cell surface glycoprotein named extracellular matrix metalloproteinase inducer (EMMPRIN; ref. 9), originally purified from the plasma membrane of cancer cells and designated tumor collagenase-stimulating factor because of its ability to stimulate fibroblast synthesis of MMP-1 (10).Subsequent studies showed that EMMPRIN is highly expressed in different cancer types (8,(11)(12)(13)(14)(15)(16), and it also stimulates fibroblast synthesis of multiple MMPs, including membrane type 1-and type 2-MMP, the endogenous activators of MMP-2 (17).Most recently, the biological activity of EMMPRIN has been linked to stimulation of MMP production by endothelial cells, therefore suggesting its potential involvement in regulating tumor angiogenesis (18).
To investigate the biological significance of increased expression of EMMPRIN in tumors and its potential role during tumor angiogenesis, we recombinantly engineered MDA-MB-231 human breast cancer cells to express different levels of EMMPRIN.The effects of EMMPRIN expression on angiogenic potential were determined both in vitro and in vivo.Our findings show that EMMPRIN positively regulates the production of VEGF in tumor cells.In vivo, increased EMMPRIN expression accelerated tumor growth and enhanced tumor angiogenesis, partially due to a significant up-regulation of VEGF and MMPs in both tumor and stromal compartments.Our results support a new paradigm in which tumor cell surface EMMPRIN plays a key role in regulating tumor angiogenesis and growth.

Materials and Methods
Materials.DMEM and fetal bovine serum were from Invitrogen (Carlsbad, CA).RNeasy RNA kits were purchased from Qiagen (Valencia, CA).ELISA kits for human and mouse MMP-2, MMP-9, and VEGF were from R&D Systems, Inc. (Minneapolis, MN).Zymography gels were from Bio-Rad (Hercules, CA).The neutralizing anti-human EMMPRIN monoclonal antibody was purchased from Research Diagnostics, Inc. (Flanders, NJ).
Cell culture.Methods for transfection and establishment of MDA-MB-231 cells stably expressing different levels of EMMPRIN have been described previously (19).Normal human dermal fibroblasts were obtained from Clonetics (Walkersville, MD).Coculture studies of cancer cells and fibroblasts were carried out as previously described (19).
Angiogenesis antibody array analysis.The expression profile of 20 angiogenic growth factors by engineered MDA-MB-231 cells expressing different levels of EMMPRIN was analyzed with angiogenic antibody array (Ray Biotech, Inc., Norcross, GA http://www.raybiotech.com).The protein concentration of serum-free media conditioned by these cells was determined using a bicinchoninic acid protein assay kit (Pierce, Rockford, IL).Blocking, hybridization of the array filters, washing conditions, and chemiluminescent detection steps were done according to the manufacturer's instructions.
Gelatin zymography.MMP-2 and MMP-9 in serum-free conditioned medium or tumor extracts were detected using substrate zymography as previously described (20).
ELISA.ELISA measurements of human or mouse MMP-2, MMP-9, and VEGF concentrations were done using Quantikine ELISA kits from R&D Systems.Triplicates of each sample were analyzed using VersaMax Tunable MicroPlate Reader equipped with Softmax Pro 3.1 software (Molecular Devices, Sunnyvale, CA).EMMPRIN ELISA assays of conditioned medium or tumor extracts were carried out as previously described (19).
Endothelial cell migration assay.Endothelial cell migration was evaluated using QCM-Collagen I Quantitative cell migration assay kit (Chemicon International, Temecula, CA).Human microvascular endothelial cells from Clonetics (10 5 in 300 AL serum-free medium) were added to the top compartment.Serum-free media conditioned by tumor cells were used as chemoattactant sources in the bottom compartment with or without anti-VEGF monoclonal antibodies (mAb; R&D Systems).Cell migration assays were carried out at 37jC for 6 hours.Cells that remained in the top compartment were removed by cotton swabs and insert filters were fixed with 3% paraformaldehyde and stained with 1% (v/v) crystal violet in acetic acid.The number of migrated cells was determined using Image Pro-Plus 3D Imaging System (Apparatus Co., Ardmore, PA).
Animal studies.All procedures involving animals and their care were conducted in conformity with the Institutional Animal Care and Use Committee guidelines.Four-week-old female CD1 nu /nu mice were obtained from Charles River Laboratories (Raleigh, NC).At approximately 6 weeks of age, animals were inoculated with 5 Â 10 6 cells s.c.Tumor growth was monitored weekly and tumor volume (mm 3 ) was calculated based on the formula (length Â width Â width) / 2. At termination of the study, tumors were excised, weighed, rinsed in ice-cold PBS, and processed for histologic/microscopic examination.Tissue specimens were also snapfrozen in liquid nitrogen for protein extraction and biochemical analysis.
Immunohistochemical and morphometric analyses.Tumors were fixed with modified Beckstead's fixative [1% zinc acetate, 0.05% calcium chloride, 0.5% formalin in 0.1 mol/L Tris (pH 7.4)], processed, and cut into 4-Am-thin sections (21).Immunohistochemistry was carried out using goat anti-murine EMMPRIN (0.2 Ag/mL, R&D Systems), rat antimurine CD31 (1.6 Ag/mL, BD Biosciences, PharMingen, San Diego, CA), and goat anti-murine MMP-9 (1.7 Ag/mL, R&D Systems).Tissue sections were incubated with primary antibodies for 3 hours at room temperature, followed by incubating with biotin-conjugated secondary antibodies, goat anti-rat immunoglobulin (BD Biosciences, PharMingen), or biotin-conjugated rabbit anti-goat (BioGenex, San Ramon, CA) for 30 minutes.Streptavidin-horseradish peroxidase conjugate was added and the peroxidase activity was made visible with diaminobenzidine.Counterstaining was done with hematoxylin.Morphometric analysis was done in five 20Â fields in each tumor to determine the area occupied by CD31-positive endothelial cells using Image-Pro Plus software version 5.1 (Media Cybernetics, Silver Spring, MD).The vasculature density was calculated as the ratio of the area occupied by endothelial cells divided by the total area.
Statistical analysis.Data analysis was done using GraphPad Prism software (GraphPad Software, Inc., San Diego, CA).P < 0.05 was considered as significant in unpaired t test analysis.Similar results were obtained from three individual experiments unless otherwise stated.(19).In coculture of tumor cells with fibroblasts, MMP-2 and MMP-9 production was regulated in an EMMPRINdependent fashion (Fig. 1).The induction of MMP-2 was most apparent when EMMPRIN was overexpressed in tumor cells, and both MMP-2 and MMP-9 were suppressed when EMMPRIN expression was inhibited (Fig. 1).These results were confirmed with ELISA quantitative measurement (data not shown).

Functional characterization of engineered tumor cells expressing different levels of EMMPRIN. To investigate the relationship between tumor cell EMMPRIN expression and tumor angiogenic potential, we created tumor cells expressing different levels of EMMPRIN
EMMPRIN expression modulates vascular endothelial growth factor expression.To determine the direct impact of EMMPRIN expression levels on the production of angiogenic growth factors, conditioned media from the engineered cells

Cancer Research
Cancer Res 2005; 65: (8).April 15, 2005 were analyzed using antibody arrays.Among 20 different angiogenic growth factors analyzed, only VEGF was significantly modulated by EMMPRIN expression (Fig. 2A).Soluble VEGF production was further quantitatively assessed by ELISA analysis (Fig. 2B).Consistent with antibody array results, EMMPRIN overexpression resulted in a significant increase in VEGF protein expression, from 208 pg/mL in wild-type tumor cells and 176 pg/mL in vector control cells to 310 pg/mL in sense EMMPRIN cells (P < 0.05 compared with wild-type).Conversely, soluble VEGF production was substantially suppressed to 65 and 109 pg/mL in two antisense tumor cell lines when EMMPRIN expression was inhibited (P < 0.01).The effects of EMMPRIN on VEGF gene expression were further analyzed at the mRNA level (Fig. 2C).VEGF mRNA level increased by approximately 3-fold in EMMPRIN-overexpressing cells as compared with that in either wild-type or vector control cells (P < 0.01), and conversely, was reduced by 78% and 70% in two antisense clones (P < 0.01).These data suggest that EMMPRIN modulates VEGF gene expression and consequently affects the production of VEGF protein in tumor cells.
EMMPRIN-stimulated vascular endothelial growth factor is biologically active.The biological activity of VEGF induced by EMMPRIN expression was determined by testing the chemotactic activity of conditioned medium in an endothelial cell migration assay.Consistent with increased VEGF levels determined by ELISA analysis, conditioned medium of EMMPRIN-overexpressing tumor cells exhibited a 69% increase in endothelial cell chemotactic activity as compared with that derived from wildtype tumor cells (P < 0.01, Fig. 2D).This chemotactic activity was suppressed by 45% and 57% in two antisense tumor cell lines, concomitant with the inhibition of EMMPRIN expression (P < 0.01).This chemotactic activity was substantially inhibited by antibody blockade of VEGF activity in a dose-dependent fashion (Fig. 2E).Approximately 50% of cell migration was suppressed when the anti-VEGF mAb was used at 1.0 Ag/mL (P < 0.01), and to near background level at 10 Ag/mL (P < 0.01), confirming that VEGF is a major endothelial cell chemotactic activity in this system.Based on these results, we conclude that tumor cellassociated EMMPRIN modulates the production of biologically active VEGF.
Tumor cell-associated EMMPRIN modulates vascular endothelial growth factor expression in the coculture of tumor cells and fibroblasts in a matrix metalloproteinaseand EMMPRIN-dependent fashion.Since tumor cell-associated EMMPRIN is believed to be a critical mediator of the interactions between tumor cells and surrounding stromal fibroblasts, we studied whether EMMPRIN expression also influenced VEGF production in the coculture of tumor cells and fibroblasts.A substantial increase in VEGF expression was detected in the coculture (Fig. 2F).The level of soluble VEGF was 306 pg/mL in the coculture of normal human dermal fibroblasts and wild-type tumor cells, representing a 35% increase over the combined levels of 20 and 208 pg/mL in the monocultures, respectively (P < 0.01).This induction was EMMPRIN-dependent as evidenced by an enhanced stimulation of VEGF to 416 pg/mL when normal human dermal fibroblasts were cocultured with tumor cells overexpressing EMMPRIN (P < 0.01 compared to coculture with wild-type tumor cells).Conversely, in coculture of normal human dermal fibroblasts with two antisense tumor cell lines, VEGF levels were only 135 and 154 pg/mL, respectively (P < 0.01).When EMMPRIN-mediated tumor-stroma interaction was interrupted using a neutralizing mAb, a considerable reduction in VEGF production was observed, further supporting the role of EMMPRIN in modulating VEGF production (P < 0.01, Fig. 2G).In addition, it seemed that the induction of soluble VEGF in coculture could be due to increased MMP activity produced in coculture.When MMP activity was inhibited using 1,10-phenanthroline, a partial but significant decrease in soluble VEGF level was also detected (P < 0.01, Fig. 2G).However, this reduction was not seen when a nonchelating 1,7-phenanthroline was used (data not shown).
Taken together, our findings suggest that tumor cell surface EMMPRIN not only modulates MMP expression, but also regulates VEGF production by mediating tumor-stromal cell interactions.
EMMPRIN regulates vascular endothelial growth factor and matrix metalloproteinase expression in xenograft tumors and stimulates tumor angiogenic potential and growth rate.The effects of EMMPRIN on tumor angiogenesis and growth were investigated in vivo using tumor cells expressing different levels of EMMPRIN.Compared with tumors derived from wild-type or vector control tumor cells, a 5-fold increase in final tumor weight was seen in sense 1-3 tumors derived from EMMPRIN-overexpressing cells (P < 0.05, Fig. 3A).When tumor expression of EMMPRIN was inhibited in AS1-5 and AS2-5 cells, the tumorigenic potential of these cells was greatly suppressed, resulting in tumors of significantly smaller size (P < 0.05).
Furthermore, morphometric analysis showed an increase in the mean density of CD31-positive microvasculature in EMMPRINsense tumors as compared with wild-type tumors (0.094 F 0.052 versus 0.047 F 0.009, P = 0.037; Fig. 3B).Because only a very small number of animals developed antisense tumors, we were not able to analyze the consequences of EMMPRIN antisense inhibition on tumor angiogenesis.

Cancer Research
Cancer Res 2005; 65: (8).April 15, 2005 Consistent with our in vitro observation of VEGF regulation by EMMPRIN, the enhanced tumor angiogenesis in EMMPRINoverexpressing tumors could be attributed to an increase in tumor VEGF expression (Fig. 3C and D).A 2.6-fold augmentation of human VEGF level was detected in tumor extracts from EMMPRIN sense tumors (2.4 pg/Ag total protein) compared with wild-type or vector control tumors (0.9 pg/Ag, P < 0.01).More significantly, the impact of tumor-associated EMMPRIN on VEGF expression was not limited to tumor cells.Concomitant with stimulation of tumor VEGF production, host-derived mouse VEGF escalated from 0.23 and 0.24 pg/Ag in wild-type and vector control tumors to 0.48 pg/Ag in sense tumors (P < 0.01).Conversely, suppression of tumor EMMPRIN expression resulted in 40% and 57% reduction in tumor and host cell-derived VEGF levels (P < 0.01).
These observations support a new paradigm in which tumor EMMPRIN mediates active interactions between tumor and stromal compartments to stimulate VEGF production and subsequently tumor angiogenesis and growth in vivo.
Tumor cell-associated EMMPRIN stimulates matrix metalloproteinase expression in both tumor and stroma compartments.The influence of tumor cell-associated EMMPRIN on MMP expression was determined in order to further explore the potential involvement of the EMMPRIN-MMP system during tumor angiogenesis and growth.Human EMMPRIN levels increased considerably in sense tumors (110 pg/Ag of total protein), and conversely were suppressed greatly in antisense tumors (26 pg/Ag), compared with 59 pg/Ag in wild-type tumors (P < 0.01, Fig. 4A).This stable effect of EMMPRIN expression subsequently influenced MMP expression in vivo.Both human MMP-2 and MMP-9 expression levels in EMMPRIN overexpressing tumors were elevated by approximately 2-to 2.5-fold (P < 0.01), and conversely inhibited by 2-fold in antisense tumors (P < 0.01, Fig. 4B and C).The effect of tumor EMMPRIN expression on host MMP-9 expression was even greater than that of tumor MMPs, resulting in a 3.3-fold increase or a 59% decrease in mouse MMP-9 expression associated with stromal cells in sense or antisense tumor nodules, respectively (P < 0.01, Fig. 4D).
The effect of tumor EMMPRIN expression on the host EMMPRIN-MMP system was studied with immunohistochemical analysis (Fig. 5).In EMMPRIN sense tumors, up-regulation of both mouse MMP-9 and EMMPRIN was seen in stromal cells in tumor capsules or those cells infiltrated into tumor tissue, and did not seem to be associated with xenograft human tumor cells.In contrast, there were only minimal levels of MMP-9 and EMMPRIN expression in vector control tumors, mainly detected in tumor stroma.
These results show that tumor-associated EMMPRIN actively mediates the cross-talk between tumor and host cells, resulting in stimulation of MMPs as well as EMMPRIN expression in the stromal compartment of the host, and may therefore contribute to tumor angiogenesis and growth.

Discussion
The importance of the microenvironment at the tumor-host interface during malignant disease progression is becoming increasingly appreciated.Comprised of a variety of cell types including immune cells, inflammatory cells, muscle and myofibroblasts, vascular cells, as well as different extracellular matrix molecules, tumor stroma is now being viewed as one of the critical elements which promotes the transition from carcinoma in situ to invasive cancer (22,23).Cells and extracellular matrix molecules in the tumor stroma are intimately involved in tumor cell proliferation, tumor angiogenesis, as well as tumor cell dissemination and metastasis.Interactions between tumor cells and surrounding stromal cells are mediated by various soluble and cell surface molecules, including EMMPRIN.
EMMPRIN expression in primary breast cancer tissue correlates with tumor size and staging, and is predictive of poor prognosis (24).The abnormally high levels of EMMPRIN expression in cancer have recently been attributed to dysregulation of epidermal growth factor receptor signaling (25).Overexpression of EMMPRIN in a breast cancer cell line had been previously shown to stimulate tumor growth in vivo (26).The functional importance of EMMPRIN during tumor disease progression was thought to be mainly due to its stimulatory effects on fibroblast-derived MMP expression in tumor tissue.The results presented here support the additional roles of EMMPRIN in tumor angiogenesis and growth.
In this new paradigm, tumor angiogenesis is stimulated by elevated levels of tumor cell-derived VEGF as a direct consequence of increased EMMPRIN expression, and is further enhanced by host cell-derived VEGF production induced by EMMPRIN-mediated tumor-stroma interactions.EMMPRIN-mediated VEGF production likely occurs at different levels.Increased EMMPRIN expression results in an immediate stimulation of VEGF transcription and accompanying VEGF protein production in tumor cells.In addition, EMMPRIN also seems to modulate VEGF secretion in an MMPdependent fashion.Recent findings suggest that MMPs, i.e., membrane-type 1 MMP may directly stimulate VEGF expression via the Src tyrosine kinase signaling pathway (27).Therefore, the MMP-dependent regulation of VEGF expression observed in this study could be the consequence of increased membrane-type 1 MMP in EMMPRIN-overexpressing tumor cells (17).Alternatively, tumor cell-derived MMP-2 or MMP-9 could also elaborate soluble VEGF from the extracellular matrix (4,5).Therefore, it is equally plausible that in vivo MMP expression induced by EMMPRIN in both tumor and stromal compartments in turn release biologically active angiogenic growth factors from matrix-bound complexes.The EMMPRIN-MMP-VEGF relationship we observed in this study recapitulates clinical findings in which MMP expression in tumor tissues often correlates with VEGF expression and tumor angiogenesis (2, 3), suggesting the involvement of this system in cancer disease progression.
The expression pattern of the EMMPRIN-MMP system in vivo is of particular interest.When EMMPRIN was overexpressed in tumor cells, a coordinated regulation of different components in the EMMPRIN-MMP system was detected not only in tumor cells, but more noticeably in host cells of the stromal compartment.For example, in addition to increased MMP-2 and MMP-9 levels derived from tumor cells, a substantial increase in MMP-2, MMP-9, and EMMPRIN also occurred in stromal cells infiltrated into or in close proximity to tumor tissue.These results, for the first time, furnish evidence verifying the involvement of the stromal compartment in EMMPRIN expression in tumor tissue.The coexpression of EMMPRIN and MMPs are commonly seen in various cancers, where EMMPRIN expression is usually associated with cancer cells, and MMPs are detected in the stroma (11).However, recent findings of EMMPRIN mRNA expression in peritumoral fibroblasts in ovarian carcinoma (28) support our findings of EMMPRIN expression in stromal cells.

Conclusion
We have identified an EMMPRIN-MMP-VEGF system in which tumor cell-associated EMMPRIN stimulates tumor angiogenesis by elevating MMP and VEGF levels in both tumor and stromal compartments.These findings extend the role of EMMPRIN from an MMP stimulator to an angiogenic promoter, highlight the importance of tumor-stroma interactions in cancer, and suggest a novel tumor angiogenesis mechanism driven by tumor cellassociated EMMPRIN expression (Supplementary Figure 1).It is of further interest to investigate the relationship of EMMPRIN, MMP, and VEGF expression and tumor angiogenesis in clinical situations, and determine if targeting EMMPRIN may represent a feasible approach to managing or treating cancer.

Figure 1 .
Figure 1.Functional characterization of tumor cells expressing different levels of EMMPRIN.MDA-MB-231 tumor cells engineered to express different levels of EMMPRIN were analyzed in coculture assays mimicking in vivo tumor-stroma interactions.Tumor cells (10 5 ) were either cultured alone or together with normal human dermal fibroblasts cells (2 Â 10 5 ).Serum-free conditioned media containing 10 Ag of total protein were analyzed in gelatin zymography.Representative results from three separate experiments are shown.WT, wild type cells; S1-3, cells stably transduced with sense EMMPRIN constructs and overexpress EMMPRIN; AS1-5 and AS2-5, cells transduced with antisense EMMPRIN constructs and underexpress EMMPRIN.

Figure 2 .
Figure 2. EMMPRIN modulates VEGF expression.A, angiogenic growth factor expression profile of tumor cells expressing different levels of EMMPRIN.Serum-free conditioned media were collected from MDA-MB-231 wild-type, vector control-, sense-and antisense-transduced tumor cells.Conditioned media containing 100 Ag of total protein were incubated with antibody arrays.The circular outline depicts the location of two duplicate VEGF spots.Four spots in the top left-and two in the bottom right-corners are positive controls; B, ELISA determination of VEGF production.Soluble VEGF level was quantitatively determined using the same conditioned media from (A ).Bars, SD; *, P < 0.02 compared with VEGF level in wild-type tumor cells; C, quantitative reverse transcription-PCR analysis of VEGF gene expression in tumor cells expressing different levels of EMMPRIN.*, P < 0.01 compared with VEGF mRNA level in wild-type tumor cells; D, determination of endothelial cell chemotactic activity contained in serum-free medium conditioned by tumor cells.Endothelial cell migration was quantitatively assessed using image analysis.Cell migration induced by conditioned medium derived from wild-type tumor cells was assigned as 100%.Bars, SD; *P < 0.01 compared with cell migration induced by media conditioned by wild-type tumor cells; E, inhibition of endothelial cell migration by neutralizing antibodies to VEGF.Endothelial cell migration stimulated by conditioned medium from S1-3 tumor cells, assigned as 100%, was dose-dependently inhibited by anti-VEGF mAb.Bars, SD; *, P < 0.01 compared with cell migration in the absence of the anti-VEGF mAb; F, VEGF level in coculture of fibroblasts with MDA-MB-231 tumor cells expressing different levels of EMMPRIN.Conditioned medium containing equal amount of total protein were analyzed.Bars, SD; *, P < 0.01 compared with VEGF levels in coculture of wild-type tumor cells with fibroblasts; G, VEGF production in coculture is dependent on MMP and EMMPRIN activity.For studies with inhibitors, VEGF levels were measured when MMP activity was inhibited with 1,10-phenanthroline, or when EMMPRIN was inhibited with RDI-CD147, a neutralizing anti-EMMPRIN mAb.Data are representative of two independent experiments.Bars, SD; *, P < 0.01 compared with VEGF levels in coculture of wild-type tumor cells with fibroblasts.

Figure 3 .
Figure 3. Tumor cell-associated EMMPRIN regulates tumor growth and angiogenesis by modulating VEGF expression.A, final tumor weight of xenograft tumors derived from wild-type, vector control, S1-3, AS1-5, and AS2-5 tumor cells.Each group contained eight animals.Bars, SD; *, P < 0.05 compared with wild-type tumors.Similar results were obtained from two individual experiments; B, CD31 staining of blood vessels (red) in xenograft tumors derived from wild-type and S1-3 tumor cells (Â200); C and D, modulation of VEGF expression by EMMPRIN in both tumor and host compartments.Human and mouse VEGF levels derived from xenograft tumor or host stromal cells, respectively, were quantitatively determined by ELISA analysis.Bars, SD; *, P < 0.02 compared with wild-type tumors.Similar data were obtained from two independent in vivo studies.Due to the small size of antisense tumors, AS1-5 and AS2-5 tumors were combined for this study and denoted as MDA-MB-231 AS.

Figure 4 .
Figure 4. EMMPRIN-MMP enzymatic system in xenograft tumors.A, EMMPRIN expression in tissue extracts from xenograft tumors derived from wild-type, vector control, EMMPRIN sense, and antisense tumor cells; B, human MMP-2 levels in xenograft tumors; C and D, human and mouse MMP-9 levels in xenograft tumors.Bars, SD; *, P < 0.01 compared with wild-type tumors.Similar results were obtained from two independent in vivo studies.

Figure 5 .
Figure 5. Immunohistochemical staining for MMP-9, EMMPRIN, and blood vessels in tumors expressing different levels of EMMPRIN.A and B, H&E staining of MDA-MB-231 xenograft tumors; C and D, mouse MMP-9 staining; E and F, mouse EMMPRIN staining; G and H, blood vessel staining with anti-mouse CD31 antibodies.Control tumors (A, C , E, G); tumors overexpressing EMMPRIN (B, D , F , H ). Dotted lines denote the boundary between tumor (T ) and stromal (S ) tissues (Â100), (inserts in D and F , Â400).