Identification of the Motif in Versican G3 Domain That Plays a Dominant-negative Effect on Astrocytoma Cell Proliferation through Inhibiting Versican Secretion and Binding*

This study was designed to investigate the mechanisms by which mutant versican constructs play a dominant-negative effect on astrocytoma cell proliferation. Although a mini-versican or a versican G3 construct promoted growth of U87 astrocytoma cells, a mini-versican lacking epidermal growth factor (EGF) motifs (versicanΔEGF) and a G3 mutant (G3ΔEGF) exerted a dominant-negative effect on cell proliferation. G3ΔEGF-transfected cells formed smaller colonies, arrested cell cycle at G1 phase, inhibited expression of cell cycle proteins cdk4 and cyclin D1, and contained multiple nucleoli. In cell surface binding assays, G3 products expressed in COS-7 cells and bacteria bound to U87 cell surface. G3ΔEGF products exhibited decreased binding activity, but higher levels of G3ΔEGF products were able to inhibit the binding of G3 to the cell surface. G3ΔEGF expression inhibited secretion of endogenous versican in astrocytoma cells and also inhibited the secretion of mini-versican in COS-7 cells co-transfected with the mini-versican and G3ΔEGF constructs. The effect seems to depend on the expression efficiency of G3ΔEGF, and it occurred via the carbohydrate recognition domain.

Versican, a member of the large aggregating chondroitin sulfate proteoglycan family, was initially detected in the limb bud of chick embryo (1) and later cloned in human fibroblasts and chick embryo (2)(3)(4)(5). It is also expressed in normal human central nervous system and brain tumors (6). RT-PCR 1 reveals that transcripts of versican isoforms are present in astrocytomas, oligodendrocytomas, medulloblastomas, schwannomas, and meningiomas (6). Versican expression levels are low, however, because neuronal immunostaining for versican appears only pericellularly. Versican is highly expressed in the tissues flanking the regions where neural crest cells migrate in chick embryos, but it is absent from the actual migration pathways. Similar findings are noted for the outgrowing sensory and motor axons of chick embryos, because versican is notably absent in regions invaded by these axons (7). Versican is known to associate with a number of molecules in the extracellular matrix such as hyaluronan, tenascin, and fibronectin (8 -10). In the central nervous system, versican has been observed to co-localize with tenascin and hyaluronan (9). Tenascin binds at the C-type lectin unit of versican (10,11).
Structurally, versican is made up of an N-terminal G1 domain, a glycosaminoglycan attachment region, and a C terminus containing a selectin-like (or G3) domain. The latter contains two epidermal growth factor (EGF)-like repeats, a lectin-like motif (also known as carbohydrate recognition domain or CRD), and a complement binding protein (CBP)-like motif (3,12,13). Alternative splicing generates at least four versican isoforms (14 -16), and some of these are highly expressed in brain tumors (6). The role of versican in brain tumor formation and progression is not clear.
We have previously demonstrated that a mini-versican construct promoted NIH 3T3 fibroblast proliferation through the G3 domain, and two EGF-like motifs in the G3 domain are involved in this effect (17). Deletion of the EGF-like motifs from the mini-versican construct significantly reduced the effect of the mini-versican on cell proliferation and differentiation (17,18). Here we demonstrate that the mini-versican construct promotes astrocytoma cell proliferation through the G3 domain. To our surprise, deletion of these EGF-like motifs produced a dominant-negative effect on astrocytoma cell proliferation. We designed assays to uncover the mechanism associated with this dominant-negative effect. A G3 construct lacking the EGF-like motifs (G3⌬EGF) binds to the astrocytoma cell surface. This may have competed with endogenous versican for binding sites on the cell surface and blocked the function of versican in cell growth. Furthermore, we demonstrate that the mutant construct inhibited secretion of endogenous versican in glioma cells and the mini-versican in COS-7 cells. These two mechanisms may account for the dominant-negative effect of the mutant on cell growth. It appears that the effects of the G3⌬EGF on cell growth are concentration-dependent and occurred via the CRD motif. gies, Inc. ECL Western blot detection kit was from Amersham Pharmacia Biotech. Horseradish peroxidase-conjugated goat anti-mouse IgG and goat anti-rabbit secondary antibodies were from Sigma. Tissue culture plates were from Nunc Inc. Oligonucleotides were synthesized by BioBasic Inc. (Scarborough, Canada). All chemicals were from Sigma. Astrocytoma cell line U87 and COS-7 cells were obtained from the American Type Culture Collection (Manassas, VA). The cells were cultured in DMEM supplemented with 10% (U87) or 5% (COS-7) FBS at 37°C in a humidified incubator containing 5% CO 2 .
Construction of Recombinant Genes-A mini-versican gene consisting of a complete G1 domain, a partial CS domain (15% in size of the entire sequence) and a complete G3 domain was constructed and expressed in COS-7 cells. Briefly, the G1 domain encompasses nucleotides 145-1182 of versican. The CS sequence is a 1242-base pair cDNA (nucleotides 1183-2424 of versican). The G3 domain contains a 927base pair cDNA corresponding to nucleotides 9904 -10830 (3). The recombinant mini-versican gene is 3.2 kilobases, which yields a core protein of ϳ150 kDa. With the attachment of glycosaminoglycan chains, the recombinant proteoglycan migrated on SDS-PAGE gel as a smear, at around 200 kDa or higher. The preparation of three constructs, a mini-versican lacking the EGF-like motifs (versican⌬EGF), a G3 construct, and a G3 lacking the EGF-like motifs (G3⌬EGF), was described in detail earlier (17,18). In all constructs used in this study, the link protein leading peptide was attached at the N terminus to allow product secretion (17,19). This leading peptide contains an epitope recognized by the monoclonal antibody 4B6 (20). To generate CBP construct, CBP was synthesized using CBPN (5Ј-aaactcgaggttgcctgtggtcaacct) and G3C (5Ј-aaaaaagcatgcgcgccttgagtcctgccacgtcct) as primers in a PCR, and the product was digested with XhoI and SphI. The leading peptide was synthesized with LPN (5Ј-aaaaaagaattcctaagtctactctttctggtgctg) and LP60b (5Ј-aaactcgagaggcagtgtgacgttgcc) to generate an EcoRI restriction site at the N terminus and an XhoI site at the C terminus of the leading peptide. Thus, the leading peptide and the CBP fragment were inserted into EcoRI-and SphI-digested pcDNA1 plasmid. To produce the CRD construct, CRD was synthesized using CRDN (5Ј-aaactcgagcaagacacagagact) and CRDC (5Ј-aaatctagatgttcctttcttgcaggt) as primers in a PCR. The PCR products were digested with XhoI and XbaI and purified. The purified products were inserted into XhoI-and XbaIdigested CBP construct, in which the XhoI site was situated between the leading peptide and the CBP fragment, whereas the XbaI site was located at 3Ј of the SphI site.
Gene Expression-The pcDNA1-mini-versican construct was transiently expressed in COS-7 cells using Lipofectin as originally described by Felgner et al. (21). Growth medium and cells were harvested separately 3 days after transfection. Expression of constructs was analyzed on Western blot probed with 4B6 (20). To obtain stable expression, glioma cells were transfected with versican⌬EGF as previously described (22)(23)(24). Geneticin was introduced into the growth medium (0.5 or 1.5 mg/ml) 24 h after transfection, and the cells were maintained in this medium until individual colonies were large enough for cloning. The selected cell lines were stored in liquid nitrogen or maintained in growth medium containing 0.5 mg Geneticin/ml for subsequent gene expression assays and functional studies. Cell lines were monitored to ensure expression of the transgene for the duration of functional studies. In co-transfection assays, equal amounts of the two constructs were mixed and used for the transfection.
Analysis of Proteoglycans on Western Blot-Cell lysate and growth medium that contained recombinant gene products were each subjected to SDS-PAGE electrophoresis and immunoblotting as described previously (17,18). Primary antibodies were used at 1:1000 dilution, unless otherwise stated, and bound antibodies were visualized using an ECL kit according to the manufacturer's instructions. Because of its large size, electrophoresis of endogenous versican was performed in an agarose gel (agarose-Western blot assay). The agarose gel (4 cm height containing 1.5% agarose in a buffer containing 0.124 M Tris-Cl, 27 mM barbituric acid, 1 mM EDTA, pH 8.7) was poured on top of a 1-cm conventional 10% polyacrylamide gel, which served to seal the bottom of the casting apparatus. This buffer was also used as a running buffer, and the electrophoresis was carried out at 40 V for 5 h at room temperature. Molecules (up to 2 million daltons in size) were able to enter the agarose gel, as shown by use of blue dextran 2000 as an internal control. Growth media and lysate from the U87 cell line and human brain tumor tissues 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 Tris-glycine buffer at 20 V overnight at 4°C. Western blotting was performed as above.
Briefly, the G3 domain was expressed in Escherichia coli strain M15 using the bacterial expression vector pQE30 (Qiagen Inc., Chatsworth, CA; catalog number 32149) as shown in Fig. 1A. The G3 domain was amplified in a PCR using two primers, 5Ј-aaaggatccggacaggatccatgcaaa and 5Ј-aaagcatgcgcgccttgagtcctgccacgt. The product was subcloned into pQE30, and the resulting construct contained an N-terminal MRGS His tag. Peptides were purified on a Ni-NTA affinity column (Qiagen, catalog number 30230) according to the manufacturer's instructions.
Proliferation Assays-Growth media from COS-7 cells transfected with different recombinant constructs were mixed in a 1:1 ratio with native culture medium (DMEM supplemented with 2.5% FBS), and the mixture was introduced into glioma cells cultured in 96-well tissue culture plates at a density of 2 ϫ 10 3 cells/well. The cultures were maintained in an incubator for 3 days, and cell number was counted using a cytometer. To test the effects of the purified products on cell proliferation, glioma cells were plated on 96-well dishes at a density of 2-4 ϫ 10 3 cells/well, 200 l/well. Purified products were added into each well (50 l of column eluate/well). Lysate from vector-transfected bacteria was loaded onto Ni-NTA purification columns, and the eluate from these columns was used as a control. Cell proliferation was determined after 3 days of incubation. To test the effects of recombinant constructs on cell growth, glioma cells were transiently transfected with the constructs using the method described above. Four days after transfection, cells were counted. Cell lysate and culture medium were harvested for analysis of gene expression on Western blot. Cell proliferation was also tested in glioma cell lines stably transfected with versican⌬EGF construct and the control vector. Briefly, cells were seeded to 96-well tissue culture plates at a density of 2 ϫ 10 3 cells/well in DMEM containing 10% FBS. The cultures were maintained in an incubator at 37°C for 3 days, and cell number was counted as above.
Cell Cycle Analysis-U87 cells were plated on 6-well tissue culture plates at a density of 2 ϫ 10 5 cells/well in DMEM containing 10% FBS at 37°C for 2 days. The cells were analyzed by flow cytometry. Briefly, the cells were collected with trypsin/EDTA, pelleted by centrifugation, and resuspended in 1 ml of hypotonic propidium iodide solution (50 g/ml) dissolved in 0.1% sodium citrate plus 0.1% Triton X-100. The cells were analyzed using a FACScan (Becton Dickinson).
Colony Formation Assay-Glioma cells were seeded to six-well plates at a density of 2 ϫ 10 5 cells/well. The cells were allowed to attach and grow overnight in DMEM supplemented with 10% FBS to reach 70% confluence. Cultures in each well were transfected with 0.5-1 g of G3⌬EGF plasmid or a control vector (pcDNA3) accompanied by 8 l of Lipofectin as described above. Two days after transfection, Geneticin was added to the culture media at a final concentration of 1.5 mg/ml. The media were changed every 5 days or earlier if necessary. Colonies of transfected cells were observed after 2 weeks.
RT-PCR Assay-Astrocytoma cells transfected with the mutant G3 construct or the control vector (2.5 ϫ 10 6 cells) were harvested, and total RNA were extracted with Qiagen RNeasy mini kit. RT-PCR assays were performed as previously described (18). Briefly, 2 g of total RNA was used to synthesize cDNA, a portion of which (equal to 0.2 g of RNA) was used in a PCR with two appropriate primers. PCR products were analyzed in agarose gel electrophoresis and detected using ethidium bromide staining. The primers for endogenous versican were 5Ј-ccagccccctgttgtagaaaa and 5Ј-gcgcctcgactcctgccacct (producing a product of 297 base pairs). The primers for G3⌬EGF transgene were 5Ј-aaactcgaggttgcctgtggtcaacct and 5Ј-aaatctagagcgccttgagtcctgcca (complementary to nucleotides 10519 -10831 of versican encoding the CBP motif and the tail). The control primers were 5Ј-ccagagcaagagaggcatcc and 5Ј-ccgtggtggtgaagctgtag (complementary to ␤-actin nucleotides 247-683).
Cell Surface Binding and Competition Assays-Growth medium from COS-7 cells transfected with G3 or G3⌬EGF was incubated with glioma cells, which had been pretreated with hyaluronidase (165 units/ml) at 37°C for 1 h. The cells were incubated at 37°C with gentle shaking for 3 h. The medium was removed after centrifugation (at 1000 ϫ g), and the cells were washed with 10 ml of PBS with gentle shaking to prevent nonspecific interaction. The cells were collected and lysed in lysis buffer. Cell extract was analyzed on Western blot probed with 4B6 to detect the binding of G3 product and G3⌬EGF product to glioma cells. To test whether one product could compete with another for binding to glioma cells, 50 l of peptides purified from bacteria was mixed with different amounts of competing medium (from G3-or G3⌬EGF-transfected COS-7 cells). Culture medium from vector-transfected COS-7 cells was used to bring the final volume to 2 ml. The mixture was incubated with glioma cells at 37°C for 2 h. The cells were washed as above, and cell lysate was prepared. Equal amounts of proteins from each treatment were analyzed on Western blot to estimate the binding of the above products to glioma cells. As well, culture media from COS-7 cells expressing G3 (200 l) were mixed with 0, 500, or 1500 l of culture media from G3⌬EGF-transfected COS-7 cells, and the competition assay was performed as above.
Immunostaining of Versican in Astrocytoma Cell Lines-Astrocytoma cells U87 were cultured on glass slips to 80% confluence. The cells were fixed with 4% paraformaldehyde and stained with rabbit antiversican polyclonal antibody, which we generated to recognize the CS sequence of the construct (17). The secondary antibody was goat antirabbit IgG antibody conjugated with horseradish peroxidase. 3Ј-Amino-9-ethylcarbazole (Sigma) was used for color development according to the manufacturer's instructions. The stained cells were examined with a light microscope and photographed.

Dominant-negative Effect on Cell Growth by Deletion of EGF-
like Motifs-To examine the role of versican in glioma cell growth, we generated a mini-versican gene and a number of mutants (Fig. 1A). We first confirmed the expression of versican in human glioma sample using Western blot. Our purified polyclonal antibody, originally raised against chicken versican, recognized a proteoglycan migrating as a large smear in agarose gel, characteristic of large aggregating chondroitin sulfate proteoglycans (Fig. 1B). Versican expression was also detected in the glioma cell line U87. The mini-versican gene was expressed in COS-7 cells, and its expression and secretion were confirmed by Western blot. Growth media from the mini-versican-, versican⌬EGF-, and vector-transfected COS-7 cells were collected and introduced into U87 glioma cell cultures. After 3 days, cell count indicated that the mini-versican enhanced cell proliferation compared with the control, whereas A, recombinant constructs used in this study are as follows: mini-versican (containing a complete G1 domain, an abbreviated CS domain, and a complete G3 domain), versican⌬EGF (a mini-versican lacking EGF-like motifs), G3, G3⌬EGF (the G3 construct lacking the EGF-like motifs), and pQE30 (G3 domain subcloned in bacterial expression vector). The leading peptide from link protein was added to the N terminus of each mammalian expression construct for product secretion. IgG, immunoglobular domain; TR, tandem repeat. Numbers above schematic correspond to nucleotides in the sequence of full-length versican. B, human brain tumor tissue and U87 cell homogenate were analyzed on Western blot probed with polyclonal antibodies against versican. The smear is a characteristic of proteoglycans. Growth medium collected from COS-7 cells transiently transfected with versican⌬EGF, control vector, or the mini-versican construct was also analyzed on Western blot probed with 4B6. Deletion of the EGF-like motifs resulted in a smaller core protein, and this proteoglycan migrated slightly faster than the recombinant mini-versican. C, growth medium collected from COS-7 cells transfected with the versican⌬EGF construct was mixed with DMEM containing 2.5% FBS, and the mixture was introduced into glioma cultures that had been seeded into 96-well tissue culture plates at a cell density of 2 ϫ 10 3 cells/well, 200 l/well. Media collected from the vector-transfected cells were used as controls. After 3 days, cells were counted. Data represent the mean Ϯ S.D. of four separate experiments (n ϭ 4; *, p Ͻ 0.05). D, versican⌬EGF was stably expressed in glioma cells. Three such cell lines and three cell lines stably transfected with a control vector were seeded in tissue culture plates for cell proliferation assay. Data represent the means Ϯ S.D. of four separate experiments (n ϭ 4; **, p Ͻ 0.01). E, cells stably transfected with versican⌬EGF or the control vector were seeded in 96-well tissue culture plates. Growth media from COS-7 cells transfected with the mini-versican construct or the control vector were introduced into the glioma cultures as indicated. The cultures were maintained in an incubator for 3 days, and cell number was determined. The dominant-negative effect of the versican⌬EGF construct on cell proliferation was significantly reduced by addition of growth medium containing mini-versican products. Data represent the means Ϯ S.D. of four separate experiments (n ϭ 4; **, p Ͻ 0.01).
growth medium from versican⌬EGF-transfected cells exerted a dominant-negative effect, producing inhibition of glioma cell growth compared with the control (Fig. 1C). The dominantnegative effect of the mutant was further confirmed in cell lines stably transfected with versican⌬EGF; three cell lines expressing the versican⌬EGF construct exhibited a dominant-negative effect on cell growth compared with the vector control (Fig. 1D). One of the cell lines was cultured and incubated with growth medium from the mini-versican-and vector-transfected COS-7 cells. Addition of exogenous growth medium from the miniversican-transfected cells reversed this effect somewhat but not completely (Fig. 1E).
Having determined that the effect of versican on glioma cell growth is mediated, at least in part, by its G3 domain, we sought to further characterize the molecular determinant(s) of the effect. Specifically, we tested whether the EGF-like motifs in G3 might play a role. We tested our hypothesis by using a G3 domain from which the EGF-like motifs had been removed (G3⌬EGF). G3⌬EGF construct was expressed in COS-7 cells (Fig. 2A), and culture medium containing G3⌬EGF products was shown to have a weak inhibitory effect on cell growth (Fig.  2B), whereas purified G3⌬EGF product produced a significant inhibitory effect on cell growth (Fig. 2C). The effect of G3⌬EGF on cell proliferation was also obtained from colony formation assays. U87 cells transfected with G3⌬EGF and a control vector were treated with Geneticin (1.5 mg/ml), and the cultures were maintained in this medium until individual colonies were formed. G3⌬EGF transfection resulted in the formation of smaller colonies (Fig. 2E) than did transfection with control vector (Fig. 2D).
The results obtained from cell proliferation assays were confirmed by analyzing cell cycle progression. Overexpression of G3⌬EGF caused arrest of a greater number of cells in G 1 phase. A typical G3⌬EGF-transfected cell line and a vector-transfected cell line are shown in Fig. 3A, in which 88% of G3⌬EGFtransfected cells were arrested in G 1 phase (7% in G 2 phase and 4.7% in S phase). Only 65.5% of control vector-transfected cells were detected in G 1 phase (24.8% in G 2 phase and 9.7% in S phase). The effects of G3⌬EGF expression on two cell cycle proteins are shown in Fig. 3B. Cell lysate harvested from FIG. 2. The effects of G3⌬EGF on cell proliferation and colony formation. A, culture medium from COS-7 cells transfected with G3, G3⌬EGF, or the vector was analyzed on Western blot. Deletion of the EGF-like motifs resulted in a smaller band compared with the G3 product. The G3⌬EGF product was also purified over a Ni-NTA affinity column, and the purified product is shown. B, growth media from COS-7 cells transfected with G3⌬EGF and the vector were incubated for 3 days with glioma cell cultures that had been seeded on 96-well tissue culture plates at a cell density of 2 ϫ 10 3 cells/well. The growth medium containing the G3⌬EGF products exerted a dominant-negative effect on cell proliferation compared with the control (n ϭ 4). C, the purified G3⌬EGF products and the control elute were added to the glioma cultures as above. The purified G3⌬EGF products also exerted a dominant-negative effect on cell growth compared with the control elute (n ϭ 3; *, p Ͻ 0.05). Glioma cells in 6-well tissue culture plates at 70% confluence were also transfected with G3⌬EGF or the vector and selected with Geneticin (1.5 mg/ml). Formation of colonies was monitored under a light microscope. Typical examples of a vector-transfected colony (D) and a mutant G3-transfected colony (E) are shown. The entire colony of the vector-transfected cells was not included in the picture. The mutant G3-transfected colony was smaller than the vectortransfected colony, and the cells exhibited a shortened morphology.

FIG. 3. Expression of versican⌬EGF alters cell cycle and expression of cell cycle proteins and nucleoi. Cell cycle was analyzed in cell lines stably transfected with versican⌬EGF and the vector (A).
The proportion of cells in each phase, determined using a FACScan, is indicated in each figure. The G3⌬EGF-transfected cell line exhibited an altered cell cycle pattern. Its cycle was arrested in G 1 phase (88% of total cells). In the vector-transfected cell line, only 65.5% of cells were detected in G 1 phase. Cell lysate from each cell line was analyzed on Western blot probed with antibodies against cyclin D1 and cdk4 (B). The levels of cyclin D1 and cdk4 were reduced in the G3⌬EGF-transfected cells as compared with the control. The structure of nuclei from cells transfected with the vector (C) and G3⌬EGF (D) was examined. G3⌬EGF-transfected cells contained multiple nucleoli within each nucleus, whereas vector-transfected cells contained only one or two. Each insert is the enlargement of one nucleus. transfected cells was analyzed on Western blots probed with antibodies against cdk4 and cyclin D1 (Santa Cruz). Levels of cyclin D1 and cdk4 decreased dramatically in G3⌬EGF-transfected cell lines. The structure of nuclei was then examined, and it was observed that each nucleus of the vector-transfected cells contained one or two nucleoli (Fig. 3C), whereas each nucleus of the G3⌬EGF-transfected cells contained multiple nucleoli (Fig. 3D).
Interaction of G3 Domain with Glioma Cell Surface-There are a number of mechanisms by which the versican⌬EGF and G3⌬EGF constructs could exert a dominant-negative effect on cell growth. The simplest explanation is that these mutant gene products are able to bind to sites on the glioma cell surface and thus successfully block the proliferative effects of endogenous versican. Because the G3⌬EGF construct was alone sufficient to exert this effect, we used it in these studies to minimize complications arising from potential cell surface binding sites present in other versican domains (e.g. G1). We first demonstrated that the full-length G3 products were able to bind to the cell surface. Glioma cells were incubated in growth media from G3-and vector-transfected COS-7 cells. After extensive washing, cell lysate was harvested and analyzed on Western blot. G3 bound to the glioma cell surface, resulting in detection of a G3 band in the cell lysate (Fig. 4A). Using the same methods, we demonstrated that G3 produced by bacteria and added exogenously to U87 cells was also able to bind to the cell surface (Fig. 4A).
To test whether G3⌬EGF products were able to bind to glioma cells, U87 cells were incubated with growth media from COS-7 cells transfected with either G3⌬EGF or G3 construct, both of which were well expressed (Fig. 4B). The cells were washed extensively and lysed for Western blot analysis. The G3 signal was significantly more intense than that of G3⌬EGF, indicating that G3 had a higher affinity for the glioma cell surface (Fig. 4B). The media from G3-and G3⌬EGF-transfected cells were mixed with the cell lysate of bacteria expressing G3. The mixture was incubated with glioma cells, and the amount of His-tagged bacterial G3 product remaining on the cells was assessed. G3 and G3⌬EGF from COS-7 cells inhibited the binding of bacterial G3 product in a dose-dependent manner (Fig. 4C).
The finding that deletion of the EGF-like motifs from the mini-versican and the G3 construct has a dominant-negative effect on cell growth suggests that, at high concentrations, the products of the versican⌬EGF and G3⌬EGF constructs were able to compete with endogenous versican for binding sites on glioma cells, although their binding is apparently weaker. We tested whether G3⌬EGF could compete with G3 to bind to glioma cell surface. G3 was mixed, at a fixed concentration, with varying amounts of G3⌬EGF, and the mixtures were incubated with glioma cells. High concentration of G3⌬EGF inhibited G3 binding to glioma cell surface (Fig. 4D).
Inhibition of Versican Secretion by the G3⌬EGF Construct-To further characterize the mechanism of dominantnegative effect of G3⌬EGF on cell growth, we examined the expression of endogenous versican in cell lines stably transfected with the G3⌬EGF construct and the control vector. Interestingly, we observed that cells stably transfected with G3⌬EGF had a higher level of versican in their cytoplasm, as revealed by labeling with polyclonal anti-versican antibody (Fig. 5A) as compared with cells transfected with the control vector (Fig. 5B). This finding raised the possibility that the G3⌬EGF construct had no effect on the transcription and translation of endogenous versican, but in fact inhibited its post-translational processing. To test this, we analyzed culture media from cells stably transfected with G3⌬EGF or the vector on Western blot. Cells transfected with the vector did indeed secrete higher levels of versican into the growth medium as compared with the G3⌬EGF-transfected cells (Fig. 5C). It was then necessary to examine whether expression of G3⌬EGF had any effect on transcriptional regulation of the endogenous versican gene. RT-PCR was performed using RNA from astrocytoma cell lines transfected with G3⌬EGF or the vector. Levels of RT-PCR products were similar in both cell lines, implying FIG. 4. The interaction of G3 and G3⌬EGF with glioma cells. A, growth medium from G3-transfected COS-7 cells and cell lysate from G3-pQE-transformed bacteria were incubated with glioma cells for 2 h. The products that were retained by the cell surface were analyzed on Western blot. G3 and G3-pQE interacted with glioma cells resulting in the detection of G3 from both sources. B, growth medium from COS-7 cells transfected with G3, G3⌬EGF or the control vector was analyzed on Western blot (medium). The medium was then incubated with glioma cells to test the binding activities of these products to glioma cells. The protein band of G3 was significantly stronger than that of G3⌬EGF. C, cell lysate from G3-pQE30-expressing bacteria (50 l) was mixed with medium from G3-transfected COS-7 cells (upper panel) or from G3⌬EGF-transfected COS-7 cells (lower panel) in various amounts (0, 25, 250, 500, and 1500 l), and the mixture was incubated with glioma cells for binding assays. The interaction of G3-pQE with glioma cells was inhibited by G3 and G3⌬EGF in a dose-dependent manner. At high concentrations, some degraded G3-pQE migrated slightly faster. D, growth medium (200 l) from G3-transfected COS-7 cells was mixed with growth medium from G3⌬EGFtransfected COS-7 cells (0, 500, and 1500 l), and the mixture was incubated with glioma cells to test the binding of these products to glioma cells. High concentrations of G3⌬EGF reduced the binding of G3 to the cells. that G3⌬EGF had no effect on versican transcription (Fig. 5D).
To further confirm the effect of G3⌬EGF on proteoglycan secretion, we co-transfected COS-7 cells with a mini-versican construct and one of the following three constructs: G3⌬EGF construct, CD44, or a control vector. Cell lysate and culture media were analyzed on Western blot probed with 4B6, which recognizes an epitope present in the G3⌬EGF and mini-versican constructs. Cells co-transfected with the mini-versican/ CD44 or mini-versican/vector secreted much higher levels of mini-versican than did those co-transfected with mini-versican/ G3⌬EGF (Fig. 6A). However, cells expressing mini-versican and G3⌬EGF had higher levels of mini-versican staining in their cell lysate (Fig. 6A). Secretion of the G3⌬EGF product was unaffected because it was detected in the culture medium (Fig. 6B). Expression of CD44 was confirmed in cell lysate probed with a monoclonal antibody against CD44 (Fig. 6C). These findings indicate that co-transfection of mini-versican and G3⌬EGF results in a specific retention of mini-versican in the cell.
The above G3⌬EGF-transfceted cell lines were selected with high level of Geneticin (1.5 mg/ml), and we only used those cell lines expressing high levels of G3⌬EGF. It is obvious that the effect of G3⌬EGF on cell proliferation depends on the levels of G3⌬EGF expression. To further confirm this, cell lines expressing low levels of G3⌬EGF were selected with low levels of Geneticin (0.5 mg/ml). Most of these cell lines expressed low levels of G3⌬EGF. Three cell lines expressing low levels of G3⌬EGF and three cell lines expressing high levels of G3⌬EGF (shown in Fig. 7A with Western blot assay and Fig. 7B with RT-PCR) were used for cell proliferation assay. Cell lines expressing high levels of G3⌬EGF had higher levels of inhibitory effect on cell proliferation as compared with the vector control, whereas cell lines expressing low level of G3⌬EGF had moderate inhibitory effect on proliferation (Fig. 7C). The former also had a significant inhibitory effect on cell elongation compared with the control, whereas the latter had a median effect (Fig. 7D). The Effect of CRD and CBP Expression on Cell Proliferation-To dissect the motif in the G3⌬EGF construct that inhibited cell proliferation, constructs containing either CRD motif or CBP motif were produced as shown in Fig. 1A. Cell lines expressing CRD and CBP were selected, and their effects on cell proliferation were examined. Expression of CRD and CBP constructs were tested on Western blot probed with 4B6 (Fig.  8A). The cell lines expressing CRD had an inhibitory effect on cell proliferation as compared with the vector control, whereas expression of CBP had little effect on proliferation (Fig. 8C). Similarly, only the cell lines expressing CRD had a moderate effect on the alteration of cell morphology (Fig. 8D). DISCUSSION Versican is highly expressed in various tissues where the cells are metabolically active and proliferating, such as in the mesenchymal tissues. In epidermis, versican is found only in the proliferating zone (26,27). In cultured cells, versican is expressed only when cells are actively proliferating; once cells reach confluence, versican expression decreases (27). Therefore, it has long been suspected that versican is associated with the process of cell proliferation. Immunohistochemical studies have revealed that versican is expressed in brain tumors. This  6. Secretion of the mini-versican products was inhibited by co-transfection of mini-versican and G3⌬EGF. A, COS-7 cells were co-transfected with a mini-versican construct and one of the following: G3⌬EGF, CD44, or a control vector using equal amounts of plasmid DNA. Cell lysate and culture media were analyzed on Western blot (5% gel of SDS-PAGE) probed with 4B6. The levels of mini-versican in the culture media from cells co-transfected with the mini-versican/ CD44 or mini-versican/control vector were much higher than that in cells co-transfected with mini-versican/G3⌬EGF. However, the levels of mini-versican in the cell lysate were opposite: cells co-transfected with the mini-versican and G3⌬EGF had higher level of mini-versican staining than the others. (Because of its small size, G3⌬EGF had run out of the gel.) B, expression and secretion of the G3⌬EGF product to the culture medium was confirmed on Western blot probed with 4B6. C, expression of CD44 was confirmed in cell lysate probed with a monoclonal antibody against CD44. study was designed to investigate the role of versican in tumor cell growth.
In the central nervous system, chondroitin sulfate proteoglycans constitute the major proteoglycan component in the extracellular matrix. Versican is known to associate with a number of molecules in the extracellular matrix such as hyaluronan (8), tenascin (9 -11), fibronectin (29), fibulin (30), and CD44 (31). Versican is excluded from focal contact and its distribution is similar to hyaluronan, CD44, and tenascin. Interestingly, this same study demonstrated that tracks left by migra-tory fibroblasts on culture plates exhibited versican immunoreactivity (32). Thus, versican may be involved in cell invasion. In astrocytomas, however, the expression of versican is not restricted to the invasive borders but is present in all grades of astrocytomas, indicating that all the tumors examined possess some invasive potential. Whereas low grade tumors contain scattered individual cells with versican expression, high grade tumors have large clusters of versicanimmunoreactive cells, probably the result of clonal expansion of cells with invasive potential. Previous studies (33) of astrocy- toma cell lines have not been able to correlate the invasiveness of astrocytomas with their proliferative activity. But, clinically, grade III and IV astrocytomas with high proliferative activity are definitely highly invasive tumors that can infiltrate proximal tissue as well as the distant regions such as brain stem, the leptomeninges, and even the contralateral cerebral hemisphere through structures such as the corpus callosum. Our study suggests that there may be a correlation between astrocytoma proliferative activity and density of versican expression within these tumors, because we showed that versican can stimulate glioma cell growth. High histological grades or MIB-1 labeling index would be associated with a greater propensity to invade. One other finding is the expression of versican in areas surrounding tumor necrosis. This suggests that the up-regulation of versican expression could be either associated with factors produced by necrotic tumor tissue such as cytokines (34) or linked to tumor necrogenesis.
Our studies made use of an in vitro system by using glioma cells transfected with a mini-versican construct. We demonstrated that the mini-versican construct promoted glioma cell proliferation and that this occurred through the G3 domain in the mini-versican gene. We have previously demonstrated that two EGF-like motifs in the mini-versican construct were involved in enhancing NIH 3T3 fibroblast proliferation. In this report, we found that deletion of the EGF-like motifs not only abolished the effect of mini-versican on cell growth, but the resultant mutant also exerted a dominant effect on glioma cell proliferation. The G3⌬EGF construct had the same effect. This indicated that motifs in G3 other than the EGF-like motifs were important in producing this effect. This result allowed us to use mutant G3 construct (G3⌬EGF) to investigate the mechanisms by which deletion of the EGF-like motifs from the mini-versican generated a dominant-negative effect on glioma cell proliferation. The products of the G3⌬EGF construct was small enough to analyze even trace amount of G3⌬EGF product in Western blot assay. This has made the cell surface binding assay possible because the levels of protein bound to the cell surface was very low in some cases. Low levels of mini-versican, which migrated as a smear in the gel of SDS-PAGE, were impossible to be detected on Western blot. As well, in the assay that the G3⌬EGF construct inhibited endogenous versican secretion, the small G3⌬EGF products were easily separated from the endogenous versican. Otherwise, we would see the versican⌬EGF products overlapped by the endogenous versican. In the COS-7 cell transfection assays, we also benefited from the fact that the small mutant construct G3⌬EGF was able to inhibit the secretion of the mini-versican products.
We have previously demonstrated that a mini-versican construct also enhanced the growth of NIH 3T3 cells (17) and chicken chondrocytes (35). Deletion of two EGF-like motifs from the mini-versican significantly reduced the effect of the mini-versican on cell proliferation but did not completely abolish this effect. A dominant-negative effect on NIH 3T3 cell proliferation was not seen. Perhaps in NIH 3T3 cells, the endogenous versican has only a minimal effect on cell growth. Consequently, expression of exogenous mini-versican enhanced cell proliferation, but deletion of the EGF-like motifs did not result in a dominant-negative effect. In the studies reported here, we demonstrated that deletion of the EGF-like motifs from the mini-versican or the G3 construct produced a dominant-negative effect on glioma cell proliferation. Thus, this effect is apparently specific to the glioma U87 cell line. In glioma cells, endogenous versican probably plays an important functional role in enhancing cell proliferation, and the mutant constructs likely interfere with a process that is crucial for cell proliferation. In other cell types this process may be less crucial or nonexistent, and so the effect of the mutant versican is less profound.
In cell surface binding assays, we demonstrated that the products of mutant G3 construct (G3⌬EGF) were bound to the glioma cell surface. High levels of G3⌬EGF were able to inhibit the interactions of native G3 products with the cell. Thus, the G3⌬EGF mutant may suppress the role of endogenous versican in enhancing cell proliferation by hindering its interaction with the cell surface. Other G3 motifs such as CRD and/or CBP may bind to the cell surface, and this binding may promote interaction of the EGF-like motifs with molecules on the cell surface such as signal transduction molecules (e.g. the EGF receptor EGFR). The mutant G3⌬EGF, which still contains CRD and CBP regions, would retain a binding ability but be inactive, because it lacks EGF-like motifs. G3⌬EGF would thus compete with endogenous versican for binding. This represents a potential molecular mechanism to account for the dominant-negative effect of G3⌬EGF.
Another possible explanation for the dominant-negative effect is that the G3⌬EGF products might inhibit the production of endogenous versican. To test this, we analyzed the secretion of endogenous versican and observed that less versican was secreted from cells transfected with G3⌬EGF, compared with control. Immunostaining revealed that the endogenous versican was synthesized at similar levels in both types of cells.
Thus, it appears that G3⌬EGF expression does not inhibit the synthesis of endogenous versican but does suppress its secretion. This was further confirmed in co-transfection studies. In COS-7 cells co-transfected with G3⌬EGF and mini-versican, secretion of the mini-versican was inhibited, but synthesis was not affected. These results strongly suggested that the mutant G3⌬EGF construct plays a dominant-negative effect on cell proliferation through suppressing the secretion of endogenous versican, and this represents a second mechanism for dominant-negative effect of the mutant G3⌬EGF. This was further confirmed in our study that only those cell lines expressing high levels of G3⌬EGF had a lower rate of proliferation and shortened cell morphology. On the other hand, the cell lines expressing low levels of G3⌬EGF had little effect on cell proliferation and morphology. Because the major motifs in the G3⌬EGF construct are CRD and CBP, their effect on cell proliferation was investigated, and our studies suggested that the inhibitory effects of G3⌬EGF on cell proliferation and morphology occurred via the CRD motif. The effect of CBP motif on cell activity was not clear. Our previous study indicated that CBP plays a role in glycosaminoglycan chain attachment and product secretion (28,36).
Our studies have demonstrated two possible mechanisms that may underlie the dominant-negative effect of G3⌬EGF on glioma cell proliferation: competition from cell surface binding sites and suppression of secretion of endogenous versican. We cannot exclude a third possibility: that G3⌬EGF binds to tenascin-C, a molecule that is believed to play a role in cell proliferation. It has been shown that the CRD motif can bind to tenascin-C (10,11). The effect of this binding on cell proliferation awaits further investigation.