Overexpression of genes, through genomic amplification and other mechanisms, can critically affect the behavior of tumor cells. Genomic amplification of the 13q31-32 region is reported in many tumors, including rhabdomyosarcomas that are primarily pediatric sarcomas resembling developing skeletal muscle. The minimum overlapping region of amplification at 13q31-32 in rhabdomyosarcomas was defined as containing two genes: Glypican-5 (GPC5) encoding a cell surface proteoglycan and C13orf25 encompassing the miR-17-92 micro-RNA cluster. Genomic copy number and gene expression analyses of rhabdomyosarcomas indicated that GPC5 was the only gene consistently expressed and up-regulated in all cases with amplification. Constitutive overexpression and knockdown of GPC5 expression in rhabdomyosarcoma cell lines increased and decreased cell proliferation, respectively. A correlation between expression levels of nascent pre-rRNA and GPC5 (P = 0.001), but not a C13orf25 transcript containing miR-17-92, in primary samples supports an association of GPC5 with proliferative capacity in vivo. We show that GPC5 increases proliferation through potentiating the action of the growth factors fibroblast growth factor 2 (FGF2), hepatocyte growth factor (HGF), and Wnt1A. GPC5 enhanced the intracellular signaling of FGF2 and HGF and altered the cellular distribution of FGF2. The mesoderm-inducing effect of FGF2 and FGF4 in Xenopus blastocysts was also enhanced. Our data are consistent with a role of GPC5, in the context of sarcomagenesis, in enhancing FGF signaling that leads to mesodermal cell proliferation without induction of myogenic differentiation. Furthermore, the properties of GPC5 make it an attractive target for therapeutic intervention in rhabdomyosarcomas and other tumors that amplify and/or overexpress the gene. [Cancer Res 2007;67(1):57–65]
- Genomic amplification
- heparan sulfate proteoglycan
Genomic amplification is frequently observed in human tumors and is one mechanism leading to the up-regulation of genes that may affect cellular behavior and drive tumor progression. Determining the key genes involved in amplification events will lead to increased understanding of tumorigenic processes and may provide prognostic or diagnostic markers in addition to targets for novel therapeutic approaches. However, few genes from the many amplicons described have been unambiguously implicated in tumor development ( 1).
Rhabdomyosarcomas are predominantly pediatric sarcomas resembling developing skeletal muscle. They are broadly divided into two main subgroups on the basis of histology: alveolar and embryonal rhabdomyosarcomas. Alveolar rhabdomyosarcomas are generally associated with a poorer prognosis than the embryonal subtype and often contain either a t(2;13)(q35;q14) or t(1;13)(p36;q14) translocation, which produces a PAX3/FOX01A or PAX7/FOX01A gene fusion, respectively ( 2). In addition to these translocations, we and others have identified characteristic patterns of chromosomal imbalance in rhabdomyosarcomas 5 ( 3, 4), which include a region of amplification at 13q31-32 in ∼20% of the alveolar subtype ( 4). Using a technique that we developed to profile differential expression at the chromosomal level ( 5), we showed that gain of the 13q31-32 region in rhabdomyosarcomas frequently corresponds to overexpression ( 4, 5). Furthermore, overexpression from the 13q31-32 region was seen in rhabdomyosarcomas, including embryonal rhabdomyosarcomas, without evidence for gain. Therefore, overexpression of genes from this region seems a more widespread phenomenon than indicated by amplification events.
Amplification of 13q31-32 is frequently seen across a broad range of tumor types ( 6). Other sarcomas, in addition to rhabdomyosarcomas, display amplification of 13q31-32, including poor prognosis liposarcomas ( 7) and other tumor types described with amplification of this region include lymphomas ( 8), lung carcinomas ( 9), breast cancers ( 10), and neurologic tumors ( 11).
Here, we define the minimum overlapping region of amplification at 13q31-32 in primary rhabdomyosarcoma, which, combined with expression analyses, indicates the Glypican-5 (GPC5) gene as a target of the amplification event. The product of GPC5 is characterized as functionally relevant to the development of rhabdomyosarcoma and a potential target for therapeutic approaches.
Materials and Methods
Patient samples and cell lines. Rhabdomyosarcoma samples were collected and stored; their pathology was reviewed as previously described ( 12). The study of rhabdomyosarcoma had both local and national ethical approval (Local Research Ethics Committee No. 1836 and Multi-Regional Research Ethics Committee/98/4/023). The study of patient materials was conducted according to the principles expressed in the Declaration of Helsinki. Snap-frozen tissues were used to extract DNA and RNA and the latter was treated with DNAase (DNA Free, Ambion, Austin, TX). The rhabdomyosarcoma cell line RD and lymphoma cell line Namalwa were obtained from American Type Culture Collection (ATCC, Manassas, VA). The rhabdomyosarcoma cell line CW9019 was grown in DMEM with 10% FCS.
Fluorescence in situ hybridization. Bacterial artificial chromosome (BAC) clones spanning the 13q31-32 region were confirmed to map to the appropriate region by hybridization to normal metaphase chromosomes before use. Interphase fluorescence in situ hybridization (FISH) was done on touch imprints from alveolar rhabdomyosarcoma primary samples as previously described ( 13).
Quantitative real-time PCR. Six sets of primers and probes were designed to measure the genomic copy number and mRNA levels of GPC5, GPC6, and the C13orf25 transcript containing the miR-17-92 micro-RNA (miRNA) cluster ( 14). GJB2 was chosen as an endogenous control for copy number as it is on a region of chromosome 13 not frequently altered in rhabdomyosarcoma. To measure the amount of RNA for these genes, primers were designed across exon boundaries. Glyceraldehyde-3-phosphate dehydrogenase (Applied Biosystems, Foster City, CA) assay was used as an endogenous control. Twenty-five microliters of multiplex PCR reactions with 10 ng of DNA or cDNA were run using the primer and probes indicated in Supplementary Table S1. Samples were run in triplicate on an ABI7700 SDS (Applied Biosystems). Genomic levels were measured relative to normal genomic DNA, whereas expression levels were measured relative to a pool of RNAs from 11 normal muscle biopsy samples.
GPC5 construct and transfection. Image clone 5744533 containing the full coding region of human GPC5 was obtained from the ATCC. The coding region was cloned into pCMV-TnT (Promega, Madison, WI) and pCS2+ using standard methods. All sequences were verified and an in vitro translation (TnT, Promega) using biotinylated lysine was used to confirm production of appropriate protein. In vitro translation products were separated by SDS-PAGE and developed using streptavidin-alkaline phosphatase and appropriate colorimetric reagents. A single ∼63 kDa protein of the appropriate core size for GPC5 was produced. Plasmids were transfected into the rhabdomyosarcoma cell line CW9019 using FuGene6 transfection reagent (Roche, Basel, Switzerland). pCMV-TnT-GPC5 was cotransfected in a molar ratio of 10:1 with pTK-Hyg (Clontech, Mountain View, CA). A control transfection using empty pCMV-TnT vector was also done. Hygromycin (200 μg/mL) was added 48 h posttransfection to produce stable clones.
Polyclonal antibody production, purification, and Western blotting. A custom polyclonal antibody to GPC5 was raised to the epitope peptide H2N-CKSYTQRVVGNGIKAQ-COOH (close to the COOH terminus; done by Eurogentec, Seraing, Belgium). Final bleeds were affinity purified and antibody specificity was confirmed by Western blotting of in vitro translated GPC5 protein using a 1:10,000 anti–rabbit-horseradish peroxidase secondary antibody (Vector Laboratories, Burlingame, CA) and by using blocking epitopes to show that bands can be abolished and are thus likely to be epitope specific. In addition, a commercial monoclonal antibody for GPC5 (MAB2607) was available although the epitope to which this is raised is unknown (R&D Systems, Minneapolis, MN). Cell lysis buffer (Cell Signaling, Danvers, MA) was used to produce whole-cell lysates and 30 μg of protein were run in each lane. Many posttranslational modifications of glypicans, including GPC5, are reported ( 15). Heparitinase digestion (0.5 units/1 × 106 cells) was used in conjunction with detection using the commercial antibody. A band ∼14 kDa in size was detected with the polyclonal antibody raised, which corresponds to that previously reported by Veugelers et al. ( 15). The commercial monoclonal antibody detected a band ∼65 kDa corresponding to the core protein ( 15, 16).
Cell proliferation assay. Five thousand cells of each transfected clone were added to each well of a 24-well plate and cell numbers were measured at a given time point following a set of pilot experiments to determine the best seeding density to maximize early growth. Viable cell numbers were measured using a p-Nitrophenyl-N-acetyl-β-D-glucosaminide (Sigma, St. Louis, MO) colorimetric metabolic dye method as previously described ( 17). To control for any potential differences in plating efficiency, results were normalized to the first time point following overnight incubation. Log-normalized growth was calculated as the mean natural logarithm of absorbance of triplicates at 64 h minus the mean natural logarithm of absorbance of triplicates at 16 h.
GPC5 small interfering RNA treatment of RD cells. SmartPool oligonucleotides were supplied by Dharmacon (Lafayette, CO). RD cells were seeded at 4.5 × 104 per well in a 24-well plate were transfected with small interfering RNA (siRNA; 100 nmol/L) before assessing proliferative capacity. Forty-eight hours after siRNA treatment, cells were plated in fresh medium in 96-well plates. After 16 h, allowing cells to attach to the plates, a commercial 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (CellTiter 96 Aqueous, Promega) was used to measure proliferation over the following 72 h. Protein and RNA were extracted and analyzed by Western blot after 72 h.
Assessment of the effect of GPC5 on response to growth factors. Transfected cells from the clone expressing the highest levels of GPC5 and randomly selected empty vector controls were plated into 96-well plates and serum starved (DMEM and 0.2% FCS) overnight followed by exposure to extracellular growth factors for 72 h. Fibroblast growth factor 2 (FGF2; R&D Systems), hepatocyte growth factor (HGF; Peprotech, Rocky Hill, NJ), and Wnt1A (Peprotech) were used at a final concentration of 20, 50, and 20 ng/mL in DMEM + 0.2% FCS, respectively. Cell numbers were assessed using a commercial MTT assay (CellTiter 96 Aqueous, Promega), and data were normalized to untreated control.
Assessment of the effect of GPC5 on growth factor signaling. The alveolar rhabdomyosarcoma cell line RH30 was transfected with either a constitutive GPC5 expression vector or the equivalent empty control vector. These vectors were cotransfected with pSRE-Luc (Stratagene, La Jolla, CA) reporter vector and pRL-TK (Promega) for normalization. Twelve hours after transfection, cells were serum starved overnight (DMEM and 0.2% FCS) then treated with 10 and 20 ng/mL FGF2 (R&D Systems) and 50 ng/mL HGF (Peprotech) in DMEM + 0.2% FCS. Firefly luciferase units were measured 8 h posttreatment and were normalized by renilla luciferase units. The mean vector control value was subtracted from each result.
FGF2 immunohistochemistry. Immunohistochemistry was done using a 1:50 dilution of rabbit anti-FGF2 antibody (Santa Cruz Biotechnology, Santa Cruz, CA) and Alexa Fluor 488 anti-rabbit antibody (Invitrogen, Carlsbad, CA) for secondary detection. Cells were counterstained with TOPRO-3 and analyzed by fluorescence confocal microscopy using a Leica Sp1 with Alexa 488 and 633 lasers.
Microinjection of human GPC5 RNA into Xenopus laevis embryos. X. laevis embryos obtained by hormone-induced laying were in vitro fertilized, dejellied in 2% cysteine (pH 8.0), and washed in 0.1× modified Barth's solution (MBS). Capped RNAs were injected into embryos in 0.2× MBS supplemented with 4% Ficoll. Capped RNA was synthesized in vitro from linearized plasmid using the SP6 Message Machine kit (Ambion). Animal caps were dissected at stage 9, and explants were cultured in 0.7× MBS containing 0.1% bovine serum albumin and 25 μg/mL gentamicin. For growth factor stimulation, caps were incubated with 20 ng/mL of human FGF2 (basic fibroblast growth factor) and FGF4 (R&D Systems). Total RNA was isolated from Xenopus embryos or animal caps using Qiagen RNeasy mini kit according to the manufacturer's instructions. cDNA was synthesized using Superscript II (Invitrogen). Quantitative reverse transcription-PCR for determining expression levels of Xbra and Xmyod was done using Sybr green (Qiagen, Heiden, Germany) on a Corbett Research RoterGene PCR machine using ODC as a positive control.
Definition of 13q31-32 amplicon. The extent of amplification at 13q31-32 in two primary rhabdomyosarcomas was ∼3.5 Mb in length from ∼89,300K to ∼92,800K on the human genome map (Build 35) of chromosome 13 ( Fig. 1A ). The 2 Mb in the center of this region showed the highest copy number in the rhabdomyosarcoma samples (20–35 copies); this interval contains the GPC5 and C13orf25 genes. The other three annotated genes in this region are pseudogenes.
Quantification of GPC5, GPC6, and C13orf25 copy number and expression in rhabdomyosarcoma. Thirteen of 101 rhabdomyosarcomas (13%) showed gain of GPC5 copies >1.5-fold relative to normal DNA: by subtype, 7 of 45 alveolar rhabdomyosarcomas (16%), 6 of 51 embryonal rhabdomyosarcomas (12%), and 0 of 5 rhabdomyosarcomas not otherwise specified (0%). Results were consistent with previous molecular cytogenetic data for the 13q31-32 region where available ( 3). There was no significant difference between genomic copy number in alveolar and embryonal rhabdomyosarcomas, and no samples showed gain of GPC6 relative to the marker at 13q11. In 85 samples analyzed, expression of GPC5 was consistently greater than normal muscle ( Fig. 1B). Genomic copy number significantly correlated with expression levels (Spearman's ρ = 0.297, n = 80, P = 0.008). It is notable, however, that most variation in GPC5 expression is not due to copy number changes. There was no significant difference in GPC5 expression levels between embryonal and alveolar rhabdomyosarcomas: Alveolar rhabdomyosarcoma median is 80.5 times greater than normal muscle, whereas embryonal rhabdomyosarcoma median is 126 times greater than normal muscle. In cell lines, high RNA expression correlated with high protein levels, and low RNA levels correlated with low protein expression (data not shown). GPC6 expression was undetectable in all but one rhabdomyosarcoma sample tested.
C13orf25 expression was also quantified in the same rhabdomyosarcoma samples and, for validation purposes, in lymphoma cell lines previously investigated. The C13orf25 transcript containing the miR-17-92 cluster was quantified ( 14). Our TaqMan data for C13orf25 copy number and expression in the lymphoma cell line Namalwa was entirely consistent with the previously reported copy number increase and overexpression of C13orf25 and the miRNAs of the miR-17-92 cluster ( 18, 19). Although rhabdomyosarcoma samples with amplification frequently showed high expression of the C13orf25 transcript, two cases with gain of copy number notably had no detectable expression ( Fig. 1C). There was no significant correlation between copy number and expression (Spearman's ρ = 0.229, n = 58, P = 0.84) and no significant difference between expression levels in alveolar and embryonal rhabdomyosarcomas: Alveolar rhabdomyosarcoma median is 9 times greater than normal muscle, whereas embryonal rhabdomyosarcoma median is 10 times greater than normal muscle.
Proliferation in primary tumors. Cell proliferation in many tumors has been associated with increased ribosome biogenesis ( 20). As an indicator of this, we have previously measured nascent pre-rRNA expression levels in the primary rhabdomyosarcoma samples ( 21). A correlation between these and GPC5 expression levels was identified (Spearman's ρ = 0.42, P = 0.001, n = 63). No such correlation was found between levels of C13orf25 and pre-rRNA expression. These data are consistent with a role for GPC5 in enhancing proliferation in vivo ( Fig. 1D).
GPC5 overexpression causes increased cell proliferation in the CW9019 rhabdomyosarcoma cell line. To assess the phenotypic effects of GPC5 overexpression, we compared the proliferative capacity of cell lines manipulated to overexpress the gene with suitable controls. A construct to express GPC5 was tested by in vitro translation and was shown to produce the appropriate size protein (∼63 kDa). The rhabdomyosarcoma cell line CW9019, which expresses GPC5 at levels similar to that of normal muscle, was transfected, and colonies were selected with the GPC5 construct and an empty vector control. In five randomly selected clones, the levels of constitutive overexpression of GPC5 were found to be within the range found in primary samples with amplification, and the associated levels of protein in the clones are indicated in Fig. 2A . These five plus five empty vector control–transfected colonies were subjected to a cell proliferation assay. Log-normalized growth was calculated as the mean natural logarithm of absorbance of triplicates at 64 h minus the mean natural logarithm of absorbance of triplicates at 16 h. Overall, there was a significant increase in the log-normalized growth in culture of GPC5-overexpressing clones compared with control colonies (P = 0.013, t = 3.17, n = 10; Fig. 2B). The mean normalized growth curves of transfected and nontransfected clones are shown in Supplementary Fig. S1. The doubling time throughout the assay was decreased from a mean of 73 h in the control clones to a mean of 50 h for the GPC5-transfected clones.
GPC5 knockdown causes decreased proliferation. siRNA molecules were used to knockdown GPC5 expression in the rhabdomyosarcoma cell line RD, which was the rhabdomyosarcoma cell line with the highest GPC5 expression. Using a commercially designed mixture of four siRNAs at 100 nmol/L (Dharmacon SmartPool), expression of GPC5 was knocked down by ∼70% after 24 h compared with a commercial Non-Targeting Control siRNA (Dharmacon) with concomitant reduction in protein levels ( Fig. 3A and B ). A proliferation assay after siRNA transfection showed a significant concomitant reduction in the growth of RD cells after 72 h ( Fig. 3C).
Overexpression of GPC5 in CW9019 rhabdomyosarcoma cells increases their response to FGF2, HGF, and Wnt1A. Given that other members of the Glypican family are known to modulate the function of heparin-binding growth factors, we examined the effect of GPC5 overexpression on the action of three heparin-binding growth factors, FGF2, HGF, and Wnt1A. Following overnight serum starvation, cells were grown in low-serum medium with or without the addition of exogenous growth factor. Increase in cell numbers normalized to the respective untreated cells (i.e., increase in proliferation attributable to the addition of extracellular growth factors) was greater in the GPC5-transfected clone compared with the empty vector control for FGF2, HGF, and Wnt1A ( Fig. 4 ). The most marked increase after 24 h and the greatest absolute increase in proliferation occurred in response to FGF2. Based on these data, we conclude that overexpression of GPC5 is able to increase the proliferative effects of multiple growth factors. It was noted that the overexpression of GPC5 still conferred a proliferation advantage when grown in low-serum medium alone.
Overexpression of GPC5 increases growth factor signaling by FGF2 and HGF in rhabdomyosarcoma cell line RH30. As overexpression of GPC5 caused increased cell proliferation and more specifically enhanced FGF2-, HGF-, and Wnt1A-induced proliferation, we investigated whether GPC5 overexpression was associated with increased growth factor signaling. The promoters of many immediate-early genes, the transcription of which is rapidly and transiently induced after growth factor stimulation, contain a common regulatory element, the serum response element (SRE; ref. 22). Thus, transcriptional activity of the SRE was used as a measure of growth factor signaling. Following serum starvation, the overexpression of GPC5 increased the effect of exogenously added FGF2 on SRE transcriptional activity in a manner that increased with dose and was greater than the sum of transcriptional activity produced by either GPC5 overexpression or FGF2 treatment alone. HGF showed some increase in effect but this was more modest than FGF2 ( Fig. 5A ).
Overexpression of GPC5 increases the amount of cellular FGF2 and alters its distribution on the cell membrane. To examine how GPC5 may affect the localization of growth factors, transfected GPC5-overexpressing CW9019 colonies were examined by fluorescence confocal microscopy. Following overnight serum starvation, the GPC5-transfected cells showed a greater amount of FGF2 staining than the equivalent control without GPC5 ( Fig. 5B). Subsequent treatment of serum-starved cells with 20 ng/mL exogenous FGF2 for 45 min caused intracellular staining to be dramatically increased in both GPC5-transfected and control colonies. However, the GPC5-overexpressing cells showed distinct and discrete membrane staining that was absent in the control clone ( Fig. 5B). Furthermore, Western blotting of RD cells 48 h after transient transfection with GPC5 showed that cellular FGF2 is increased, particularly the high molecular weight 22 kDa isoform ( Fig. 5C). RD cells were investigated because they show moderate endogenous GPC5 and FGF2 expression and are readily transiently transfected.
Microinjection of human GPC5 RNA into X. laevis embryonic ectodermal explants synergistically increased the mesoderm-inducing properties of FGF2 and FGF4. To further investigate the effect of GPC5 in potentiating FGF signaling, we overexpressed GPC5 in a X. laevis ectodermal explant model. There are Xenopus homologues of human glypican genes ( 23). The most homologous transcript to human GPC5 was identified as GENESH18695 in the Ensembl Xenopus tropicalis genome database, 6 which has a 59% protein similarity to human GPC5. Both blastomeres of two-cell-stage embryos were injected with 0.3 ng human GPC5 RNA. At stage 9, animal caps (ectodermal precursors) were dissected and cultured until stage 11. Nine to 10 caps per treatment were collected and subjected to real-time quantitative PCR. Expression of the gene Xbra (a marker of early mesoderm induction) showed a slight increase over the uninjected control when 0 to 1 ng GPC5 RNA was injected alone and showed some increase in expression when FGF2 or FGF4 was added exogenously. Most striking was the synergistic effect on Xbra induction upon treating ectodermal explants with FGF2 and FGF4 in combination with GPC5: An increase in Xbra levels of ∼3× and ∼13×, respectively, over growth factor treatment alone ( Fig. 6A ). XmyoD induction in ectodermal explants following treatment with FGFs was also examined by real-time PCR but showed no difference ( Fig. 6B).
In this study, the genetic and functional data presented are consistent with a role for the GPC5 protein in the development of rhabdomyosarcoma. The minimum overlapping region involved in the 13q31-32 amplification event in rhabdomyosarcoma is shown to harbor the annotated genes GPC5 and C13orf25, which contains the miRNA cluster miR-17-92. Two rhabdomyosarcoma samples with amplification of the region had no detectable expression of the C13orf25 transcript. In all cases with amplification, as well as in cases without, high levels of GPC5 expression relative to normal skeletal muscle were found; expression levels in muscle are reported as similar to most normal nonfetal tissues ( 16). Broadly, protein levels correlated with levels of RNA expression. Moreover, by increasing or decreasing expression of GPC5 in rhabdomyosarcoma cell lines, we showed that GPC5 plays a role in increasing cell proliferation. Furthermore, as cell proliferation in many tumors has been associated with increased ribosome biogenesis ( 20), the correlation found between nascent pre-rRNA and GPC5 expression levels in primary tumors supports a role for GPC5 on enhancing proliferation in vivo. It is noteworthy that a correlation between the levels of ribosomal gene expression and the C13orf25 transcript containing the miR-17-92 cluster was not found, suggesting that this gene is not involved in increasing proliferative capacity.
The minimum region of amplification at 13q31-32 that we defined in rhabdomyosarcoma is similar to that identified in lymphomas in that both GPC5 and C13orf25 are reported to be amplified ( 8, 14, 24). However, one lymphoma cell line is described where the GPC5 gene was disrupted by the amplicon boundary and therefore excluded from involvement ( 14). C13orf25 is overexpressed in some lymphoma cell lines and the miR-17-92 miRNAs were shown to reduce the latency of leukemia development in irradiated myc transgenic mice ( 19). Overexpression of the miR-17-92 miRNA cluster has recently been reported in some lung cancer cell lines, particularly those derived from small-cell lung cancers, and this was occasionally associated with increased copy number ( 25). While there is a significant correlation with copy number, much of the variation in expression of GPC5 in our rhabdomyosarcoma data are not due to copy number change. For both GPC5 and C13orf25, our data are consistent with expression being regulated by mechanisms other than genomic amplification, a situation that has been reported for other genes in regions amplified in other tumor types ( 5, 26). A role in rhabdomyosarcoma for the miR-17-92 miRNAs, and potentially as yet uncharacterized genes in the region, in addition to GPC5, cannot be excluded and requires direct investigation. Similarly, direct investigation of GPC5 in other tumor types described with increased copy number of the 13q31-32 region is also required.
GPC5 encodes a cell surface heparan sulfate proteoglycan (HSPG) that has not been previously implicated in tumor development, although other members of the Glypican family have been shown to exhibit aberrant expression in tumors. GPC1 is overexpressed in human pancreatic ( 27) and breast cancers ( 28). GPC3, the family member that shows highest homology to GPC5, is overexpressed in neuroblastoma, Wilm's tumors ( 29), and melanoma ( 30). GPC3 was also shown to be overexpressed in hepatocellular carcinoma ( 31), and engineered GPC3 overexpression in hepatocellular carcinoma cell lines was associated with modulated proliferation ( 32). Missense mutations in GPC3 are found in Simpson-Golabi-Behemel syndrome, which is associated with overgrowth and a reported predisposition to develop pediatric tumors ( 33).
Although there are no direct reports for GPC5, HSPGs are generally thought to alter the binding kinetics of heparan sulfate–binding growth factors by facilitating the physical interaction of growth factor and receptor ( 34). There is evidence for GPC3 that this occurs in a context-specific manner; that is, certain HSPGs will enhance signaling of a given growth factor in one cell type and inhibit signaling in another ( 32, 34). This may or may not require the heparan sulfate side chains, and, indeed, recent evidence suggests that nonglycanated GPC3 core protein can form complexes with Wnt growth factors ( 35). We have shown here that GPC5 increases proliferation in rhabdomyosarcoma through potentiating the effects of FGF2, HGF, and Wnt1A. In the case of HGF, this was unusual insofar as GPC5 overexpression reversed the growth-suppressive effect of adding high levels of HGF to CW9019 cells. HGF is known to have antiproliferative activity in a number of cell lines ( 36), especially when applied at high concentrations. It is unclear by what mechanism GPC5 alters the proliferative effects of HGF; however, given that we have shown that GPC5 can increase downstream signaling, it is unlikely to be due to simply inhibiting an antiproliferative signal. The percentage increases in proliferation that we found are comparable with the levels of change seen for FGF2 in a model for GPC3 in hepatocarcinomas ( 29). It is likely that other heparan-binding growth factors are also associated with GPC5.
As FGF, HGF, and Wnt signaling pathways have been shown to play a role in myogenesis and in rhabdomyosarcoma tumorigenesis, our data are highly consistent with a role for GPC5 in the development of rhabdomyosarcoma ( 35, 37– 41). It is notable that in a turkey line bred for larger posthatch weight, glypican expression was elevated in the embryonic and posthatch pectoralis major muscle ( 42). Here, we show that GPC5 synergistically increased the mesoderm-inducing activity of FGFs in an in vivo Xenopus model. This is consistent with a role for GPC5 in maintaining or promoting a mesodermal phenotype in rhabdomyosarcoma cells. Expression of MyoD, a transcription factor associated with myogenic differentiation toward skeletal muscle, was not promoted by GPC5 in this model system and is in keeping with the generally undifferentiated nature of rhabdomyosarcoma cells. When tested, FGF4 showed a limited effect on proliferation and SRE transcription in the rhabdomyosarcoma cell lines studied (data not shown), suggesting that the interaction between FGF2 and GPC5 may be of more importance in these cell lines. Taken together, our data are consistent with a role for GPC5 in rhabdomyosarcomagenesis by increasing the effects of growth factor signaling and in particular enhancing FGF signaling that leads to mesodermal cell proliferation without induction of myogenic differentiation.
In addition to GPC5 promoting interactions between growth factors and their ligands, it is possible that GPC5 may influence growth factor signaling by retaining growth factors either at the cell surface or by facilitating internalization. Such a model for glypican interactions with growth factors has been proposed by Chu et al. ( 43). In support of this, we showed that a consequence of GPC5 overexpression was an increase in the cellular levels of FGF2 in situations where growth factor was scarce and also an increase in membrane binding of FGF2 in situations where growth factor was plentiful.
As decrease in expression of GPC5 in rhabdomyosarcoma cell lines reduced cell proliferation, it may be possible to exploit this property for therapeutic intervention. As a potential modulator of multiple growth factors, therapies that reduce the function of GPC5 could affect multiple tumorigenic pathways. Also, as GPC5 is a cell surface protein, it is physically accessible to a number of potential anti-GPC5 therapies (e.g., humanized antibodies and therapeutic peptides). Potential immunotherapeutic strategies have recently been described involving the use of GPC3 ( 44), and we propose that GPC5 represents a similarly appropriate target for this approach in rhabdomyosarcoma. We conclude that amplification of GPC5 in rhabdomyosarcoma and its expression plays a role in tumor development that we have begun to define, and that this may be exploited therapeutically in tumors that overexpress the protein.
Grant support: Joshua Gilbert Rhabdomyosarcoma Appeal and Cancer Research UK (CRUK) program grants C189, C309/A2187.
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.
We thank Jaclyn Biegel and Peter Houghton for providing the cell lines CW9019 and RH30, respectively, and the Children's Cancer and Leukaemia Group for their support in tumor collection.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
P. Workman is a Cancer Research UK Life Fellow.
- Received May 4, 2006.
- Revision received October 5, 2006.
- Accepted October 31, 2006.
- ©2007 American Association for Cancer Research.