The G3 Domain of Versican Enhances Cell Proliferation via Epidermial Growth Factor-like Motifs*

Versican is a member of the large aggregating chondroitin sulfate proteoglycan family. We have expressed in NIH3T3 fibroblasts a recombinant versican mini-gene comprising the G1 and G3 domains and 15% of the CS domain. We observed that expression of the mini-versican gene stimulated cell proliferation as determined by cell counting and cell cycle analysis. Addition of exogenous mini-versican protein to cultured cells produced the same result. The effects of the mini-versican were greatly reduced when the G3 domain was deleted. Expression of the G3 domain alone promotes cell proliferation, and addition of purified G3 gene products to NIH3T3 fibroblasts and cultured chicken fibroblasts enhances cell growth. Further, deletion of the epidermal growth factor (EGF)-like motifs in the versican G3 domain reduced the effects of the mini-versican on cell proliferation. In the presence of the purified mini-versican protein, antisense oligonucleotides to the EGF receptor inhibited proliferation of NIH3T3 fibroblasts, compared with control sense oligonucleotides. Taken together, these results imply that versican enhances cell proliferation, and this effect is mediated, at least in part, by the action of versican EGF-like motifs on endogenous EGF receptor.

subdomains. In this respect, G3 is similar to the selectin family (8) except that the G3 domain contains only one complementbinding protein-like subdomain while selectins contain at least two. Functionally, selectins play a role in mediating cell adhesion, but no such role has been identified for the G3 domain.
The versican gene (named as CSPG2) (9,10) is expressed in a variety of tissues. As a result of alternative splicing, the protein has many isoforms, which may express different functions. Versican is suspected to play a role in cellular attachment, migration, and proliferation by interacting with cell surfaces and extracellular matrix molecules (2). It has been reported that versican interferes with the attachment of cells to various extracellular matrix components such as collagen I, fibronectin, and laminin (11) and inhibits cell adhesion (12,13). It may also be involved in the formation of tissues that act as barriers to migratory neural crest cells and outgrowing axons during embryonic development (14). Furthermore, versican appears to regulate cell differentiation in hair follicles (15).
The tissues that contain high concentrations of versican include embryonic tissues such as human embryonic lung fibroblasts (2), the mesenchymal cell condensation area of limb buds (16,17), the perinotochordal mesenchyme between the notochord and neural tube, and basement membranes facing the neuroepithelial cells of chicks (18). It is also distributed in embryonic aorta, lung, cornea, and skeletal muscle (18). In adult tissues, versican is detected in the loose connective tissue of various organs including the central and peripheral nervous system, the luminal surface of glandular epithelia (19), blood vessels (20), vessels of brain tumors (21), dermis, and in the proliferative zone of the epidermis (22). Since versican is highly expressed in fast growing tissues and cells, it has been suggested that versican plays a role in cell proliferation. Using molecular biological techniques, we show in this report that the expression of a truncated versican gene, or exogenous addition of a truncated versican gene product, promotes cell proliferation. Expression of the versican G3 domain alone achieves the same effect, but deletion of the EGF-like motifs greatly reduces the ability of versican to stimulate growth. Antisense oligonucleotides to EGF receptor (EGFR) inhibited cell proliferation, and the effect was greater in the recombinant versican-transfected cell line than in the vector-transfected cell line. These results suggest that the EGF-like motifs in versican G3 domain may promote cell proliferation through a direct or indirect interaction with EGFR.

EXPERIMENTAL PROCEDURES
Materials-The reverse transcription-PCR kit was from CLON-TECH. Taq DNA polymerase, T4 DNA ligase, and restriction endonucleases were from Boehringer Mannheim. Bacterial growth medium was from Difco. Prep-A-Gene DNA purification kit, prestained protein markers, and the protein assay kit were from Bio-Rad. DNA mini-prep * This work was supported by Grant MT-13730 from Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Postdoctoral Fellow of Sunnybrook Trust. § Scholar of the Arthritis Society of Canada. To whom correspondence should be addressed: Sunnybrook Health Science Centre, University of Toronto, 2075 Bayview Ave., Toronto, Ontario M4N 3M5 Canada. Tel.: 416-480-5874; Fax: 416-480-5737; E-mail: byang@srcl.sunnybrook. utoronto.ca. 1 The abbreviations used are: EGF, epidermal growth factor; EGFR, epidermal growth factor receptor; PCR, polymerase chain reaction; DMEM, Dulbecco's modified Eagle's medium; PAGE, polyacrylamide gel electrophoresis; CRD, carbohydrate recognition domain; NTA, nitrilotriacetic acid; ELISA, enzyme-linked immunosorbent assay. of the CS domain, and the G3 domain were individually cloned into pUC19. Briefly, a HindIII site and an MluI site were added at the 5Ј and 3Ј ends of the G1 domain using G1NHindIII (AAA AAG CTT GCC GCC ACC ATG TTG TTA AAC ATA AAA AGC) and G1CMluI (AAA ACG CGT TTC GTA GCA GTA GGC ATC AAA) as primers in a reverse transcription-PCR reaction. These primers were used to amplify nucleotides 145-1182 in the versican cDNA sequence published by Shinomura et al. (1). Total RNA was extracted from the sterna of 14-day-oldchicken embryos and used as template for reverse transcription-PCR. The PCR product was purified from agarose gel following electrophoresis using the Prep-A-Gene kit. This generated a cDNA of 1038 base pairs corresponding to the G1 domain of the published sequence (nucleotides 145-1182) (1). The G3 fragment was prepared in a similar manner: the restriction sites for XhoI and SphI were created using primers G3NXhoI (5Ј-AAA CTC GAG GGA CAG GAT CCA TGC AAA AGT) and G3CSphI (5Ј-AAA GCA TGC GCG CCT TGA GTC CTG CCA CGT). These primers are located in the nucleotide sequences 9904 -9924 and 10810 -10830 and amplify a 927-base pair cDNA corresponding to the G3 domain of versican (1). To clone a section of the CS sequence, chicken genomic DNA was used as template since the CS sequence is in a large exon. CSNMluI (5Ј-AAA ACG CGT C 1183 GT AAA AAA ATT GTA TCA GAG) and CNCXhoI (5Ј-AAA CTC GAG A 2424 CT CAT TTC TGG CTC CTT TGT) were used as primers to create MluI and XhoI sites at the 5Ј and 3Ј ends of the CS sequence. These primers generate a cDNA of 1242 base pairs, containing nucleotides 1183-2424 of versican. This represents the portion of the CS sequence immediately 3Ј of the G1 domain, and contains a high proportion of potential sites for glycosaminoglycan chain attachment. We chose not to clone the whole 8-kilobase pair CS sequence because we are primarily interested in studying the functions of the G1 and G3 domains. All constructs obtained by PCR in the study were sequenced to confirm their identity. Finally, the three DNA fragments prepared above were ligated together, as shown in Fig. 1A, into pcDNA1 (Invitrogen). The recombinant versican gene (G1CSG3) was 3.2 kilobase pairs, and was expected to yield a core protein of 150 kDa. With the attachment of glycosaminoglycan chains, we expected that the recombinant proteoglycan would migrate on SDS-PAGE gel as a smear, at around 200 kDa or higher.
In generating the G3 construct, the leading peptide of link protein (nucleotides 1-180) (23) was joined with the versican G3 domain in order to allow secretion of the G3 gene product. The leading peptide of link protein was amplified with the primers 5Ј-AAA GAA TTC GCC GCC ACC ATG GCA AGT CTA CTC TTT CTG and 5Ј-AAA CTC GAG AGG CAG TGT GAC GTT GCC in a PCR reaction. The PCR product was purified and digested with EcoRI and XhoI. The G3 domain was derived from the G1CSG3 construct using the restriction enzymes XhoI and SphI. The G3 fragment and link protein leading peptide were ligated into the pcDNA1 vector digested with EcoRI and SphI. As well, a MRGS⅐ His (Qiagen) epitope (MRGSHHHHHH) was generated at the C terminus of the G3 construct for staining and purification with the monoclonal antibody MRGS⅐ His. This was obtained using a primer (AAA TCT AGA GTG ATG GTG ATG GTG ATG AGA TCC TCT CAT TGA GTC CTG CCA CGT CCT) (1), which is complementary to the 3Ј end of the G3 domain and contains a sequence encoding the MRGS⅐ His epitope, and a second primer located in the 5Ј terminus of the G3 domain, in a PCR reaction. The reaction generated the G3 construct containing the MRGS⅐ His epitope.
The G1CS construct shown in Fig. 1A was generated by insertion of the link protein leading peptide and G1CS sequence between the EcoRI and XhoI sites of pcDNA1. The leading peptide contains an EcoRI site at its 5Ј end and a BamHI site at its 3Ј end, while the G1CS sequence contains a BamHI site at its 5Ј end (nucleotides 244 -249) (1) and a XhoI site at its 3Ј end created as described above.
To delete the EGF-like motifs from the G3 domain of versican, a site for the restriction enzyme XhoI was engineered at the 5Ј end of the CRD. This was performed by using CRDNXhoI (AAA CTC GAG C 10129 AA GAC ACA GAG ACT) and G3CSphI (5Ј-AAA GCA TGC G 10830 CG CCT TGA GTC CTG CCA CGT) as primers in a PCR reaction. The product, a G3 fragment lacking the EGF-like motifs (G3⌬EGF), was digested with XhoI and SphI and ligated into the recombinant mini-versican from which the full-length G3 domain had been removed by digestion with XhoI and SphI. The pcDNA1-G1CSG3 construct (mini-versican) was transiently transfected into COS-7 cells (American Type Culture Collection) using Lipofectin (Life Technologies, Inc.) as originally described by Felgner et al. (24). Growth medium and cells were harvested separately after 3 days of transfection. Expression of genes was analyzed by Western blot, using anti-G3 antibody, as described below. We confirmed that glycosaminoglycan chains were attached to the expressed core protein, resulting in a characteristic proteoglycan smear on SDS-PAGE and Western blot, and also that gene products were secreted and present in the growth medium.
Once we had verified that the transiently-expressed products were properly processed and secreted, the mini-versican construct was subcloned into the pcDNA3 vector (Invitrogen) and stable transfection of NIH3T3 cells was performed. Three days after transfection, Geneticin was added to the growth medium at a concentration of 1 mg/ml, and the cells were maintained in this medium until individual colonies were large enough for cloning. Chemically selected cell lines were maintained in medium containing 0.5 mg/ml Geneticin or stored in liquid nitrogen. PCR technique was used to confirm that the selected cell lines had incorporated the genes of interest. Briefly, genomic DNA was prepared from cell lysate (25) and used as template in a PCR reaction with primers (5Ј-GAG CAA GAC ACA GAG ACT and 5Ј-GCG CCT TGA GTC CTG CCA) which amplify nucleotides 10125-10830 of chicken versican (1).
Western Blot Assays for Proteoglycans-Electrophoresis of recombinant proteoglycans was performed in SDS-PAGE-Western blot assay. Cell lysate and growth medium that contained recombinant gene products were subjected to SDS-PAGE electrophoresis. The stacking gel contained 4% polyacrylamide. A 5% separating gel was used for the mini-versican product and a 12% separating gel for the G3 domain. The buffer system is 1x TG (Tris-glycine buffer, Amresco product) containing 1% SDS. Proteins separated on SDS-PAGE were transblotted onto a nitrocellulose membrane (Bio-Rad) in 1x TG buffer containing 20% methanol. The membrane was blocked in TBST (10 mM Tris-Cl, pH 8.0, 150 mM NaCl, 0.05% Tween 20) containing 10% non-fat dry milk powder (TBSTM) for 1 h at room temperature, and then incubated at 4°C overnight with anti-G3 polyclonal antibody (1:1000 dilution) in TBSTM. The membranes were washed with TBST (three times for 30 min each) and then incubated for 1 h with goat anti-rabbit antibody conjugated to horseradish peroxidase in TBSTM. After washing as above, the bound antibody was visualized with an ECL kit according to the manufacturer's instructions (Amersham).
Electrophoresis of endogenous versican was performed in agarose gel (agarose-Western blot assay). The agarose gel (4-cm height containing 1.5% agarose in a barbital buffer containing 0.124 M Tris-Cl, 27 mM barbituric acid, 1 mM EDTA, pH 8.7) was made on top of a 1-cm conventional 10% polyacrylamide gel. The polyacrylamide gel was used only to seal the bottom of the casting stand and acted as a base for the agarose separating gel. The barbital buffer was also used as a running buffer and the electrophoresis took place at 40 V for 5 h at room temperature. Molecules, up to 2 million Da in size (blue dextran 2000), are able to enter the agarose gel. Growth media from chicken mesenchymal culture and NIH3T3 fibroblast culture of equal protein concentrations were analyzed in the gel. To allow transfer of such large molecules onto the nitrocellulose membrane, the blotting took place in TG buffer at 20 V overnight at 4°C. The rest of the procedure was similar to the Western blot protocol described above.
Digestion of Glycosaminoglycan Chains with Chondroitinase ABC-The conditioned medium collected from the mini-versican-transfected cells was dialyzed overnight against 10 volumes of reaction buffer (100 mM Tris-HCl, pH 8.0, 30 mM sodium acetate). After a buffer change, the mixture was further dialyzed for 1 h. Meanwhile, chondroitinase ABC (Sigma, catalog no. C2905) was dissolved in the reaction buffer at a concentration of 1.33 units/ml. This enzyme was added to the dialyzed versican solution (1:1), and the mixture was incubated at 37°C overnight to digest the glycosaminoglycan chains. In some samples, chondroitinase ABC was replaced by hyaluronidase, which was used as a control. The enzyme-digested products were subjected to SDS-PAGE and analyzed on Western blot.
Antigen Preparation and Antibody Generation-Strategy for inser-tion of the G3 domain into pQE30 (Qiagen Inc., catalog no. 32149) is shown in Fig. 1A. pQE30, a bacterial expression vector, contains an epitope (MRGSHHHHHH) at its amino terminus recognized by the monoclonal antibody anti-MRGS⅐ His. The six histidines in this epitope bind nickel and allow purification of fusion proteins on a Ni-NTA affinity column. The epitope is followed by multiple cloning sites. The G3 domain of the recombinant versican was ligated into the first cloning site, BamHI, producing an epitope-tagged G3 gene product. The G3 domain was generated using two primers G3NBamHI (AAA GGA TCC GGA CAG GAT CCA TGC AAA) and G3CSphI (as above) in a PCR reaction, as above, and ligated into the bacterial expression vector pQE30. The resulting construct contained an N-terminal MRGS⅐ His tag and was expressed in Escherichia coli strain M15. Gene products produced by bacteria were purified on a Ni-NTA affinity column (Qiagen, catalog no. 30230) according to the manufacturer's instructions. To eliminate trace amounts of contamination that may occur during column purification, the column-purified product was subjected to preparative SDS-PAGE. To produce polyclonal anti-G3 antibody, adult New Zealand White rabbits were immunized with the SDS-PAGE purified G3 domain peptide, according to methods previously described (26). The antiserum obtained was used to monitor the expression of the recombinant genes on Western blot. Proliferation Assays-1 ϫ 10 4 fibroblasts were seeded into each well of 24-well plates and cultured with standard medium (DMEM supplemented with 5% fetal bovine serum). Cell numbers were determined at day 1, 2, and 3 using a cytometer.
In the chicken fibroblast proliferation assay, the fibroblasts were isolated from 14-day-old chicken embryos. Chicken skin was removed and rinsed with phosphate-buffered saline. The skin was then cut into pieces and incubated at 37°C for 10 min in dissociation medium containing 0.1% trypsin and 0.3% collagenase (Sigma, catalog no. C-6885) dissolved in Hanks' balanced salt solution. The dissociation medium was changed, and the tissue was incubated for a further 2 h. An equal volume of growth medium was added to stop the enzyme reaction. Cells were collected by passing the mixture through a filter (70-m nylon) and centrifuging the filtrate at 1100 rpm. The cells were then resuspended in growth medium and plated in tissue culture dishes. The cells could be used for assay after 1 week of culture. Proliferation of chicken fibroblasts was assayed by counting as above.
Cell Cycle Analysis-Flow cytometry of labeled DNA was used to analyze cell cycles. The mini-versican-transfected and vector-transfected stable cell lines were cultured in DMEM supplemented with 5% fetal bovine serum for 1-3 days. Cells were harvested with 10 mM EDTA dissolved in DMEM and collected by centrifugation at 200 ϫ g. The cell pellets were resuspended gently in 1 ml of hypotonic propidium iodide (Sigma) solution (50 g/ml) prepared in 0.1% sodium citrate plus 0.1% Triton X-100 for DNA staining. The samples were analyzed with a FACScan (Becton Dickinson) as described previously (27).
Purification of G3 and Mini-versican Gene Products-To study the effect of G3 gene products on cell proliferation, G3 was purified from the G3-transfected COS-7 cells and G3-expressing bacteria. Briefly, conditioned medium from COS-7 cells that had been transfected with G3 construct, and cell lysate from G3-transformed bacteria, were dialyzed against equilibration buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10 mM imidazole). This buffer was also used to equilibrate a Ni-NTA column. The dialyzed samples were incubated with the Ni-NTA column and washed extensively with washing buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 20 mM imidazole, 1 mM phenylmethylsulfonyl fluoride). G3 peptide was eluted with elution buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 250 mM imidazole). The volume of elution buffer used in each sample was the same as the volume of the original cultures. The purified products were dialyzed against phosphate-buffered saline, and then analyzed on SDS-PAGE and visualized with Coomassie Blue dye or silver staining.
Purification of the mini-versican products obtained from mini-versican-transfected COS-7 cells was performed as above. Lysate from vector-transfected COS-7 cells was subjected to the same purification procedure to serve as a control in order to estimate the effect of any contaminants that were bound to the column and eluted by the elution buffer. The elution buffer containing purified mini-versican and the control elution buffer were dialyzed against phosphate-buffered saline overnight with two changes of phosphate-buffered saline.
Normalization of Gene Products in Conditioned Media-To evaluate the concentration of gene product in the growth media, both Western blot and ELISA were used. Purified link protein was used as a standard in Western blots. Link protein contains an epitope recognized by a monoclonal antibody (4B6) (28) in its leading peptide, and we added this leading peptide to the G3 construct. This served a dual purpose; it allowed secretion of the G3 gene product and allowed detection by the 4B6 antibody. We transiently transfected COS-7 cells with the G3 construct and the control vector alone in order to produce the G3 gene product. Expression of the G3 construct was demonstrated with Western blot assay, using cell growth medium. Briefly, 10 l of G3-containing growth medium were run on SDS-PAGE alongside a series of link protein dilutions (1-20 ng), electrotransferred, and probed with the 4B6 antibody. Since the antibody recognizes both link protein and the G3 construct product with the same affinity, and since both link protein and G3 proteins are approximately the same size, similar protein concentrations would produce similar band intensities after ECL development. This technique provides a reasonable estimate of gene product concentration.
For evaluation of the concentration of versican in growth medium, Western blot was not appropriate because versican migrates as a smear in SDS-PAGE, making determination of band density difficult. The ELISA assay provided a useful means of evaluating versican's concentration. In ELISA, growth medium (100 l) from vector-and recombinant versican-transfected chicken fibroblasts was coated onto each well of a 96-well plate and incubated at 4°C overnight. The medium was then removed and the plate was blocked with TBSTM (200 l/well) at room temperature for 1 h. Washes and incubation with primary and secondary antibody were similar to Western blot assay. The primary antibody was rabbit anti-G3 antibody, and the secondary antibody was peroxidase-conjugated goat anti-rabbit antibody. TMB (Amersham, code RPN 2718) was used for color development. Optical density was determined using an ELISA reader set to 570 nm according to the manufacturer's instructions. The optical density value in the medium from the vector-transfected cells was assigned a value of 100%.
Cell Proliferation Affected by Antisense Oligonucleotides-To test the possible role of the EGFR in versican's effects on cell proliferation, 200 l of NIH3T3 fibroblast suspension at a concentration of 3 ϫ 10 4 cells/ml were inoculated into each well of 96-well tissue culture plates followed by addition of 50 l of purified mini-versican solution at an estimated concentration of approximately 0.4 ng/l. The control solution obtained from the purification was used as a negative control. Human EGF (Sigma, catalog no. E-9644) at a final concentration of 15 ng/well was used as a positive control. As well, 15 l of sense (ATG CGA CCC TCA GGG ACC GCG, located in the 5Ј end of the coding sequence of EGFR) (29) or antisense (antisense 1, CGC GGT CCC TGA GGG TCG CAT, 5Ј end of EGFR; and antisense 2, TGC TCC AAT AAA CTC ACT GCT, 3Ј end of EGFR) oligonucleotides were added to the cultures at a final concentration of 3.2 M. Cells without oligonucleotide treatment served as controls. Cultures were incubated at 37°C in a tissue culture incubator. After 3 days, the fibroblasts were harvested by incubating the cells with 10 mM EDTA for 10 min. Cell number was determined by cell counting as described above. Each treatment was done in quadruplicate and analyzed statistically.

Expression of a Mini-versican Gene Stimulated Cell Proliferation-
The mini-versican gene, containing the G1 domain, a portion of the central sequence, and the G3 domain, was constructed ( Fig. 1A) in pcDNA1 for transient expression and in pcDNA3 for stable expression. To monitor the expression of the recombinant constructs, we raised antibody against the G3 domain. The G3 construct was expressed in E. coli M15 using the vector pQE30 (Fig. 1A) and purified on a Ni-NTA affinity column. The antibody specifically recognizes versican and the G3 construct (see below).
We have expressed the recombinant versican in NIH3T3 fibroblasts. For stable expression, cell lines were selected with Geneticin (G418) at a concentration of 1 mg/ml. Expression of the mini-versican in the selected cell lines was verified on a Western blot probed with antibody to the G3 domain. Protein bands observed in culture medium corresponded to the expected size of recombinant glycosylated versican with attached glycosaminoglycan chains (190 -250 kDa; Fig. 1B, versican). Endogenous versican was not observed in the Western blot, perhaps because it did not migrate into the separating gel or was not transferred onto the membrane. After deglycosylation with chondroitinase ABC, a major band of 150 kDa, corresponding to the expected size of the recombinant versican core protein, was observed (Fig. 1B, chondroitinase ABC).
We observed that cell lines expressing the recombinant versican had a higher growth rate and reached confluence faster than the vector-transfected cell lines. This observation is extremely important in light of the fact that versican is highly expressed in mesenchymal cells in the early stage of chicken development (16,17) and restricted to the zone of keratinocyte proliferation in the epidermis (22). To investigate the function of versican in detail, we performed different growth assays on one transfected cell line to confirm this observation. One of the cell lines was monitored by cell counting for three days (Fig. 2). As well, cell proliferation was confirmed after 3 days of culture in three cell lines transfected with the mini-versican gene and three cell lines transfected with the control vector (Table I). We observed that the mini-versican-transfected cell lines exhibited a 2-3-fold increase in proliferation rate as compared with the vector-transfected cell lines. Assays of protein concentration and thymidine incorporation also indicated that growth of mini-versican-transfected cells was enhanced, compared with controls (data not shown). The G1 domain includes nucleotides 145-1182 of mature versican, the truncated CS region, nucleotides 1183-2424; and the G3 domain, nucleotides 9904 -10830. SP, signal peptide; IgG, immunoglobular domain; TR, tandem repeat; EGF, epithelial growth factor-like motif, CBP, complementary binding protein; *His 6 tag. Numbers above schematic correspond to nucleotides in the sequence of full-length versican. B, the culture medium from one vector-and one mini-versican-transfected cell line were separated on SDS-PAGE (5% gel) and analyzed on Western blot probed with antibody to the G3 domain. The blot confirmed that versican was synthesized and secreted into the growth medium and migrated as a smear from 200 to 250 kDa. The conditioned medium collected from the vector-and versican-transfected cell lines was dialyzed and incubated with chondroitinase ABC to digest the glycosaminoglycan chains in the proteoglycan. The digested products were analyzed on Western blot (right, chondroitinase ABC). One major band, migrating at the expected mini-versican core protein size of 150 kDa, and some incompletely digested proteoglycan were observed.

TABLE II Expression of the mini-versican gene altered the patterns of cell cycles
Cell cycle assays were performed to analyze the growth rate of the mini-versican-and vector-transfected cell lines. 1 ϫ 10 6 cells were stained with propidium iodide solution, and cell cycles were analyzed with FACScan for 3 days after cell inoculation. Note that at all time points the mini-versican-transfected cell lines have more cells in S phase than do the vector-transfected cell lines. To further investigate the enhanced cell proliferative activity, cell cycle patterns were analyzed by flow cytometry. During the first 3 days, nearly twice as many mini-versican-transfected cells were found in S phase compared with the vectortransfected cells (Table II). At all time points tested, fewer cells expressing mini-versican were observed in G1, compared with control. This again suggests that the mini-versican-transfected cells exhibit a higher growth rate than vector-transfected cells.
To determine whether the addition of exogenous mini-versican gene products was sufficient for stimulation of cell proliferation, we collected conditioned medium from the mini-versican-and vector-transfected NIH3T3 cell lines. Media from different sources were added to fibroblast cultures to assess their effects on cell proliferation. Exogenous addition of the mini-versican product from versican-transfected NIH3T3 fibroblast cell lines (Fig. 3A) significantly stimulated cell proliferation. Similar stimulation was observed using growth media from the mini-versican-transfected COS-7 cells (data not shown). Thus, these results further indicate that versican can stimulate cell growth and plays an important role in cell proliferation.
Because fibroblasts are known to express versican (18), it was necessary to evaluate whether total versican expression in fibroblasts is increased by transient transfection of the miniversican gene. Growth medium was collected from the miniversican-and vector-transfected fibroblasts. The ELISA procedure was used to assess versican levels with anti-G3 antibody, as described under "Experimental Procedures." Recognition of endogenous versican with the anti-G3 antibody was confirmed by agarose-Western blot assay (Fig. 3B). According to our ELISA results, transfection of the mini-versican resulted in a 3-fold increase in total versican expression compared with the fibroblasts transfected with vector alone (Fig. 3C). We also noted that the immunoreactivity of the purified recombinant versican with the anti-G3 antibody was dose-dependent, varying with the concentration of recombinant versican (data not shown).
Versican G3 Domain Promotes Fibroblast Proliferation-Versican consists of a G1 domain, a G3 domain, and a central region for glycosaminoglycan chain attachment. We suspected that the G3 domain is important in stimulating cell proliferation since it has been shown that the EGF motifs in laminin and thrombomodulin are responsible for cell proliferation (30,31), and versican's G3 domain contains two EGFlike motifs.
To determine whether recombinant G3 peptide alone was sufficient to stimulate cell proliferation, COS-7 cells were transiently transfected with the G3 construct. Expression of the G3 construct was evaluated using link protein as a standard be- FIG. 3. Addition of exogenous recombinant versican also stimulates cell proliferation. A, growth medium was collected from an NIH3T3 fibroblast cell line stably expressing the recombinant versican. Medium collected from the vector-transfected cell line was used as a control. NIH3T3 cells were seeded in 96-well plate (2.5 ϫ 10 3 cells/well, 0.5 ml/well). Growth media were introduced into the cell cultures (0.5 ml/well), and the cultures were incubated for 3 days in an incubator. The medium from the mini-versican-transfected cells promoted cell proliferation significantly, compared with the control (n ϭ 4; **p Ͻ 0.01). B, growth media collected from chicken mesenchymal culture and NIH3T3 fibroblast culture were subjected to agarose gel electrophoresis as described under "Experimental Procedures." The gel was transblotted onto a nitrocellulose membrane and probed with anti-G3 antibody. The antibody recognized versican from both sources. C, fibroblasts isolated from chicken embryos were transiently transfected with vector and the mini-versican gene. Growth media (100 l) were collected after 3 days of transfection and coated onto each well of a 96-well plate. The plate was incubated with polyclonal antibody to the G3 domain of versican. The bound antibody was detected by peroxidase-conjugated secondary antibody followed by color development. Optical density was determined using an ELISA reader. Transfection of fibroblasts with mini-versican increases versican staining 3-fold compared with untransfected cells (n ϭ 4; **p Ͻ 0.01). cause both the G3 and link protein contain the same epitope (4B6). The assay showed that a band produced by 10 l of the growth medium has a density similar to 10 ng of link protein (Fig. 4A), indicating that the growth medium from G3-transfected cells contained approximately 1 ng of G3 protein per microliter of medium. Growth medium from G3-transfected COS-7 cells was mixed with fresh growth medium in different proportions, and the mixture was introduced into fibroblast cultures to test its effect on cell proliferation. The assay showed that medium containing the G3 gene product promoted cell proliferation in a dose-dependent manner, but medium from vector-transfected cells did not (Fig. 4B).
To ascertain that the cell growth was stimulated by G3 itself rather than other factors in the conditioned medium, we purified the G3 products from COS-7 cells and bacteria. G3 from both sources was purified on Ni-NTA affinity columns and eluted into one volume of elution buffer. The purified G3 gene products were analyzed on Coomassie Blue-or silver-stained SDS-PAGE and assayed by Western blot. 20 l of purified G3 from bacterial culture were stained as a single band by Coomassie Blue dye, while 20 l of G3 solution from COS-7 culture produced a band barely detectable with silver staining, but clearly detectable on Western blot (data not shown). Since 1 l of purified G3 from bacterial culture exhibited a level of immunoreactivity similar to 20 l of purified G3 from COS-7 cell culture (Fig. 5A), we estimated that the G3 obtained from bacterial cultures was approximately 20 times more concentrated than that obtained from COS-7 cells. Purified G3 products (50 l) from COS-7 culture or bacterial culture were applied to NIH3T3 fibroblasts and chicken fibroblasts to test their effects on cell proliferation. Addition of purified G3 gene products, from either COS-7 culture or bacterial culture, resulted in a 1-fold increase in the growth of NIH3T3 fibroblasts (Fig. 5B). Similar stimulation was observed in chicken fibroblasts (data FIG. 4. The effect of G3 on cell proliferation is dose-dependent. A, COS-7 cells was transiently transfected with the G3 construct. Expression of G3 was analyzed on Western blot. Growth medium (10 l) was run on SDS-PAGE alongside link protein (10 l at concentrations of 0.5, 1.0, and 1.5 ng/l). The resulting blot was probed with monoclonal antibody (4B6) and visualized as described under "Experimental Procedures." The G3 protein band in 10 l of growth medium had a density comparable to the 10-ng link protein, indicating that each milliliter of growth medium contained 1 g of G3 protein. B, NIH3T3 fibroblasts were seeded in 24-well tissue culture plates at a concentration of 1.25 ϫ 10 4 cells/well for the cell proliferation assay as described under "Experimental Procedures." Growth medium collected from COS-7 cells transiently transfected with either G3 construct or control vector was added to fresh medium in different proportions as shown. The mixed medium was introduced into the fibroblast cultures, and the cultures were maintained at 37°C. Cells were counted after 2 days. The growth medium from G3-transfected cells enhanced cell proliferation significantly (n ϭ 4; p Ͻ 0.01).
FIG. 5. Cell proliferation stimulated by purified G3 gene products. A, the G3 gene products were purified from bacterial lysate and growth medium from G3-transfected COS-7 cells using a Ni-NTA affinity column (Qiagen) according to the manufacturer's instructions. Growth medium from vector-transfected COS-7 cells was used as a control. The purified products from COS-7 cells (G3c, 20 l) and bacterial culture (G3b, 1 l) are shown on a Western blot stained with a monoclonal antibody to an epitope added to the G3 construct. The purified G3 protein band is indicated. B, The effect of G3 gene products on fibroblast growth was tested in cultures seeded at 10 3 cells/well in 96-well plates. Each well contained 200 l of growth medium, to which 50 l of purified gene products were added (G3c or G3b). Cell number was determined after 3 days. These experiments demonstrated that gene products produced by COS-7 cells and bacteria enhanced 3T3 fibroblast growth significantly (n ϭ 4, p Ͻ 0.001). not shown). G3 produced in COS-7 cells had a greater effect on fibroblasts than did the prokaryotic G3 product. Since Western blot had indicated that the COS-7 product was 20-fold more dilute, the difference may indicate that the eukaryotic G3 product is more efficient in promoting proliferation, presumably because it is properly processed and folded.
To further investigate the role of the G3 domain in cell proliferation, we stably expressed the G3 construct in NIH3T3 fibroblasts. Cell lines were selected with Geneticin (G418). Expression of the G3 construct was confirmed by Western blot as above (data not shown). Proliferation of the four cell lines was measured using three vector-transfected cell lines as controls. All G3-transfected cell lines had increased proliferative activity (Fig. 6A). Cell cycle analysis demonstrated that the number of cells in S phase in all G3-transfected cell lines was nearly twice that found in vector-transfected cell lines (Fig.  6B), indicating a high rate of proliferation in the G3-transfected cells.
Deletion of the G3 Domain Suppresses the Effects of Versican on Cell Proliferation-We removed the G3 domain from our mini-versican gene making a new construct, G1CS, in pcDNA3. COS-7 cells were transiently transfected with the mini-versican, G3 and G1CS constructs, and the vector, under the same conditions. Growth media from different transfections were examined for their effects on cell proliferation. Mini-versican and G3 alone stimulated cell proliferation to a similar degree. However, the G1CS product showed a significant reduction of cell proliferation when compared with the effects of versican (Fig. 7). These results further demonstrated that the effects of the mini-versican on cell proliferation are mediated, at least in part, by the activity of the G3 domain.
Deletion of the EGF-like Motifs Greatly Reduced the Ability of Versican to Stimulate Cell Proliferation-Given the importance of the versican G3 domain in the observed proliferation effects, we sought to characterize the effects of this domain in detail. G3 contains two EGF-like motifs, and to investigate their role in this phenomenon, we deleted them from the mini-versican gene to obtain the versican⌬EGF construct as described under "Experimental Procedures" and shown in Fig. 1A. Fibroblasts were transfected with the versican⌬EGF construct and cell lines stably expressing the versican⌬EGF construct were isolated. Growth medium collected from these cell lines was analyzed for versican⌬EGF expression and secretion, while growth medium from the mini-versican-transfected cells was used as control. We verified that deletion of the EGF-like motifs resulted in smaller bands which migrated slightly faster than mini-versican on Western blot (Fig. 8 A).
In a cell proliferation assay, cell lines stably transfected with versican⌬EGF, the mini-versican, and vector alone were seeded into 96-well tissue culture plates and allowed to grow for 2 days. Cell density was determined. We found that deletion of EGF-like motifs from mini-versican greatly, but not completely, abolished the ability of versican to stimulate cell proliferation (Fig. 8B).
Since the EGF-like motifs are apparently involved in the effects of versican, we next investigated if the effect was specifically mediated by the EGFR. To do so, we used an antisense approach to repress endogenous EGFR expression in our versican-transfected fibroblast cell lines. Fibroblasts were seeded into 96-well tissue culture plates, and the cultures were incubated for 3 days with antisense and sense oligonucleotides of mouse EGFR, in the presence or absence of purified miniversican products. (Cells not treated with purified mini-versican were incubated in a column elute from vector-transfected cells only.) The antisense oligonucleotides are complementary to sequences located at the 5Ј and 3Ј of the gene (29). The growth of both mini-versican-treated and the control NIH3T3 fibroblasts was inhibited in the presence of antisense oligonucleotides, presumably because EGFR is involved in a wide range of cellular processes. However, the extent of inhibition in the mini-versican-treated cells was significantly greater than that of the controls (Table III), consistent with the involvement of EGFR in versican-induced cell proliferation. Cells serving as positive controls were treated with human EGF, and these cells were also subjected to high levels of inhibition in the presence of EGFR antisense oligonucleotides, as compared with the negative control (Table III). DISCUSSION The large chondroitin sulfate proteoglycan versican is highly expressed in various tissues where the cells are actively metabolizing and proliferating such as in mesenchymal cells. In epidermis, versican is restricted to the proliferating zone. In cell culture, versican is highly expressed in both dermal fibro-blasts and keratinocytes while the cells are actively proliferating. Once confluent, the growth rate of these cells slows and the expression of versican is turned off (22). Therefore, it is thought that versican plays a role in cell proliferation.
Chicken mesenchymal cells provide an ideal model for the study of cell proliferation and differentiation. We have previ- FIG. 7. Deletion of the G3 domain of versican reduced versican's effect on cell proliferation. Growth media from cells transfected transiently with vector, G1CS, G3, or mini-versican constructs were collected and normalized for equal concentration of gene products by addition of growth medium from vector-transfected cells to media containing higher concentrations of gene products. The normalized growth media were then mixed in a 1:1 ratio with standard medium (DMEM containing 5% fetal bovine serum) and introduced into NIH3T3 cultures which had been seeded into a 96-well tissue culture plate at a concentration of 10 3 cells/well. Cell numbers were determined at day 3. Compared with G3 or versican, the G1CS gene product showed decreased ability to stimulate cell proliferation (n ϭ 4, p Ͻ 0.01).
FIG. 8. The role the EGF-like motifs in cell proliferation. A, growth medium collected from cells stably transfected with either vector, the mini-versican, or versican⌬EGF genes was analyzed on Western blot probed with 4B6. (The leading peptide in the mini-versican and versican⌬EGF constructs was replaced with that from the G1CS construct to allow staining by 4B6.) Deletion of the EGF-like motifs resulted in a smaller core protein, and this proteoglycan migrated slightly faster than the recombinant mini-versican. B, the versican⌬EGF construct was stably expressed in NIH3T3 fibroblasts. Cell proliferation was tested as described under "Experimental Procedures." A recombinant versican-transfected cell line was used as a positive control and a vector-transfected cell line as a negative control for cell proliferation. The versican⌬EGF construct showed a decreased ability to stimulate cell proliferation, compared with mini-versican. Cells transfected with vector showed only low levels of proliferation (n ϭ 4; *p Ͻ 0.05; **p Ͻ 0.01).
ously established an in vitro model of chicken chondrogenesis 2 in order to study the role of versican in chicken mesenchymal condensation and chondrogenesis. However, stable transfection of genes into chicken mesenchymal cells is not possible since chicken mesenchymal cells cannot be maintained in culture. We therefore used NIH3T3 fibroblasts as a model to investigate the role of versican. It has been well documented that integration of transfected genes into a genome is a random event (32,33). We reported that cell lines transfected with a recombinant mini-versican grew faster, incorporated more [ 3 H]thymidine into DNA, and exhibited faster cell-cycle turnover. Our experiments performed on at least three stably transfected fibroblast cell lines demonstrated that the enhanced cell growth was a result of expression of the mini-versican gene rather than a disruption of genes controlling cell proliferation in the genome. Furthermore, we found that the exogenous addition of recombinant gene product had the same effects on cell proliferation. As versican is a component of extracellular matrix, addition of versican into cell culture may mimic the effect of this molecule when it is secreted by the cells. This indicates that we can study the role of versican in chondrogenesis by introducing gene products generated in NIH3T3 fibroblasts into chicken mesenchymal cells. In other studies, we demonstrated that addition of matrix molecules to the cultures plays a role in mediating cell activities (34). Such an approach is useful in the investigation of the role of matrix molecules.
Versican is a large molecule and its domains exhibit characteristic properties. For example, the G1 domain, which resides at the N terminus of versican, demonstrates hyaluronan binding activity (7). The hyaluronan-binding property of the gene may be important in cell-cell and cell-matrix interactions such as those that involve CD44 (35). The G1 domain may also be involved in cell migration, since receptor for hyaluronic acidmediated motility, another hyaluronan-binding receptor, is known to regulate cell migration (36). The central sequence is a large region situated between the G1 and G3 domains for the attachment of chondroitin sulfate side chains. It has been reported that the chondroitin sulfate chains are involved in cell adhesion (11). The lectin-like motif in the G3 domain binds tenascin-R (37,38).
In this study, we demonstrated that the ability of a miniversican to stimulate cell proliferation was due to the G3 domain. The G3 domain is composed of a lectin-like subdomain, a complement binding protein-like motif (2), and one or two EGFlike repeats (depending on splicing). Specifically, we demonstrated that deletion of the G3 domain significantly abolished the effect of the mini-versican on cell proliferation and that the G3 domain alone was as effective as the intact mini-versican.
The G3 domain of versican we studied contains two EGF-like repeats. The EGF-like unit is a sequence of approximately 40 amino acid residues which has a significant homology to epidermal growth factor. It contains six conserved cysteine residues which form three disulfide bonds, folding the sequence into its active form. Other amino acids in this motif are also highly conserved. Many proteins contain EGF-like regions in single or multiple copies, including urokinase; laminin B1; low density lipoprotein receptor; tissue plasminogen activator; coagulation factors IIV, IX, X, and XII; protein C, S, and Z; fibronectin; thrombomodulin; transforming growth factor; and other proteins (39). All of these proteins are known or thought to take part in protein-protein or protein-cell interactions. For example, transforming growth factor, fibronectin, laminin, and thrombomodulin are able to stimulate cell proliferation (30,31,40). A fragment containing the inner rodlike segments from the short arm of laminin, which are composed of EGF-like repeats, stimulated thymidine incorporation in cultured cells possessing EGF receptors but had no effect on a cell line lacking this receptor. The recombinant thrombomodulin peptide, which is composed of six EGF-like structures, induced proliferation of Swiss 3T3 cells and accelerated [ 3 H]thymidine uptake into their DNA (30).
Based on these observations, we hypothesized that the effects of G3 on cell proliferation were mediated by the EGF-like motifs. To investigate this, we deleted the EGF-like motifs from the mini-versican and observed that deletion of the EGF-like motifs greatly, but not completely, reduced the effect of versican on cell proliferation.
It has been reported that NIH3T3 fibroblasts express low levels of functional EGF receptors and respond to EGF stimulation by exhibiting denser growth in monolayer cultures and increased DNA synthesis (41). Overexpression of EGFR conferred on these and other EGFR-negative cells heightened the response to EGF stimulation. In our study, deletion of the EGF-like motifs greatly reduced versican's ability to stimulate NIH3T3 cell proliferation, suggesting that the EGF-like motifs play a role in versican's enhancement of cell growth, perhaps by interacting with the EGFR on the cell surface. Evidence for this was provided by antisense studies in which the growth of NIH3T3 fibroblasts was inhibited by treatment with antisense oligonucleotides to mouse EGFR. However, the levels of growth inhibition of cells treated with purified mini-versican were greater as compared with the control. This suggests that EGFR is involved in the mini-versican-induced promotion of cell proliferation. The growth of control cells was also reduced after the treatment with antisense oligonucleotides, indicating that antisense treatment also interfered with basal proliferation levels generated by endogenous versican and EGFR-binding growth factors. Since deletion of EGF-like motifs did not completely abolish versican's ability to stimulate cell proliferation, it TABLE III Cell proliferation affected by antisense oligonucleotides complementary to EGFR NIH3T3 fibroblasts were cultured in 96-well tissue culture plates at a concentration of 3 ϫ 10 4 cells/ml. Three different reagents, human EGF, purified mini-versican, and control solution, were added to the fibroblast cultures followed by addition of oligonucleotides as described under "Experimental Procedures" to a final concentration of 3.2 M. The cultures were maintained at 37°C for 3 days, and the cell number was determined. The antisense oligonucleotides were found to inhibit cell proliferation under all conditions. This inhibition was significantly greater in the mini-versican-treated cells and in EGF-treated cells than in the control cells (n ϭ 4, p Ͻ 0.01). Effects of antisense oligonucleotides on different treatments (control, mini-versican, and EGF) are expressed as a percentage inhibition, compared to cells treated with sense oligonucleotides, as follows: inhibition ϭ ((number of cells following sense treatment) Ϫ (number of cells following antisense treatment))/(number of cells following sense treatment) ϫ 100%. seems that this effect is not mediated entirely by these motifs. Further studies will likely lead to the identification of other elements in the versican molecule which play a role in enhancing cell proliferation.