Xg Expression in Ewing's Sarcoma Is of Prognostic Value and Contributes to Tumor Invasiveness

Ewing's sarcoma (EWS) is a highly aggressive malignant tumor that develops in bones and soft tissues and occurs preferentially in children and young adults. Survival rates reach approximately 60% to 70% for patients with localized tumors; however, the presence of metastases at diagnosis defines worse prognosis (survival rates approximately 20–40%). Bone metastases confer a poorer outcome than lung/pleura metastases (<20% compared with 20–40% 5-year survival; ref. 1). Two biological features characterize EWS and help in diagnosis: chromosomal translocation between the 5′ portion of the EWS gene with the 3′ end of the ETS family genes (FLI1, ERG, ETV1, E1AF, or FEV; refs. 2, 3). The most frequent aberrant transcription factor generated is the EWS-FLI1 chimeric protein (85% of cases), which is the result of the t(11;22)(q14;q12) translocation. The oncogenic role of this aberrant transcription factor has been described in fibroblasts, murine primary bone-derived cells, and primary bone marrow–derived mesenchymal progenitor cells (4–6). It has also been shown recently that EWS can originate from mesenchymal stem cells (7). The other feature is the large amount of CD99 molecules in almost all EWS cases (8, 9). CD99 is also displayed at high density on T-cell acute lymphoblastic leukemia (10). CD99 is a 32-kDa transmembrane, highly glycosylated and sialylated protein, which is ubiquitously expressed on hematopoëitic cells. CD99 has two isoforms a long form of 32 kDa and a short one of 28 kDa that possess adverse functional properties (11–13). From a functional viewpoint, we recently described CD99 as a regulator of HLA class I expression on the cell surface (14). Moreover, CD99 is involved in cell adhesion and motility (11), including intercellular adhesion of lymphocyte endothelial cells (15) and transendothelial migration of monocytes (16), neutrophils (17), and CD34⁺ cells (18). In addition, it can induce immediate lymphocyte arrest on an inflammatory endothelium (15). In a pathologic context, CD99 can exert an oncosuppressor effect on osteosarcoma (19) and, in contrast, contributes to the malignant properties of the EWS family of tumors (13, 20).

CD99 exhibits 48% and 47% of sequence homology with two other proteins, Xg and mouse CD99L2, respectively, whose genes are located on the X chromosome (21, 22). Xg is a 26-kDa molecule, whose gene which is located on/at the pseudoautosomal boundary of the short arm of the X chromosome (Xp22), in the vicinity of the CD99 gene (21, 23). The gene of CD99L2 (40 kDa) is located at the end of the long arm of the X chromosome (Xq28; ref. 22). From a structural viewpoint, CD99, Xg, and CD99L2 present four highly conserved regions including the FXLXDAL motif (in which X is any amino acid) and a glycine-rich region on the extracytoplasmic domain, the transmembrane domain. The fourth domain is the juxtamembrane region that is composed of positively charged lysine and arginine residues (22). CD99 and Xg also present a supplemental highly conserved domain that is a proline-rich region. From a functional viewpoint, CD99L2 has also been characterized as an adhesion molecule required for neutrophil extravasation in mice (23).

The Xg protein was initially characterized on RBC and defines a blood group (24). Studies into families revealed a difference in the Xg(+) frequency according to gender. In fact, 89% of females and 62% of males are Xg(a+) (ref. 24). The CD99 and Xg antigens present a quantitative polymorphism on RBC (25, 26). However, the exact tissue distribution of the Xg protein, initially described as limited to RBC, and its biological function(s) remain to be characterized.

Here, we found the Xg protein to be expressed on cancer cells in 24% of EWS cases. Moreover, we found that the expression of Xg is associated with worse prognosis of EWS patients. To study the role of the Xg protein in the malignancy of EWS cells, we chose two approaches based on forced and silenced expression of Xg in two EWS cell lines. Furthermore, we show that Xg contributes to the metastatic properties of EWS cells by regulating migration in vitro and in vivo.

EWS and osteosarcoma cell lines (IOR/BRZ, IOR/CAR/, IOR/BER, IOR/CLB, LAP35, U-2OS, OS9, SARG, and MOS) were generated at the Istituto Ortopedici Rizzoli. EWS cell lines SK-ES-1, SK-N-MC, and RDES were obtained from the American Type Culture Collection. TC-71 and 6647 were a generous gift from TJ Triche (Childrens Hospital, Los Angeles, CA). WE-68, NT-68, TC-83, and VH-64 were a generous gift of Dr. Frans van Valen (Laboratory for Experimental Orthopaedic Research, University Hospital of Munster, Germany) (27). EWS cells were cultured in DMEM (Life Technologies; Invitrogen) supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin, 1 mmol/L sodium pyruvate, and 10% inactivated fetal bovine serum (FBS; Perbio). Cells were maintained at 37°C in a humidified 5% CO₂ atmosphere.

The Xg expression vector (pcDNA3.1) was a generous gift from JP Cartron [Institut National de la Sante et de la Recherche Medicale (INSERM) U76, Biologie des groupes sanguins humains, Paris, France; ref. 28]. Stably transfected IOR/BRZ cells expressing Xg were obtained using a calcium phosphate transfection (Invitrogen). Cells transfected with empty pcDNA3.1 (+) were used as a negative control. Transfectants were selected with 1 mg/mL neomycin and analyzed by phycoerythrin (PE)-conjugated Xg staining using FACScan (Becton Dickinson). The NBL-1 hybridoma (Anti-Xg) was generated and validated by Ellis and colleagues (24). This clone has also been validated by Fouchet and colleagues (28, 29). Purified NBL-1 was coupled to PE (Proteogenix).

Using the calcium phosphate reagent (Invitrogen), the TC-71 EWS cell line was stably transfected with a mix of two short hairpin RNA (shRNA) targeting Xg expression plasmids (pSilencer4.1-CMV puro, Ambion). The target sequences used were 5′-GCCGAGGTCA AAGAGACTTTGATTTCAAGAGAATCAAAGTCTCTTTGACCTCGGTTTTTT-3′ and 5′-GGGAGGTTACTTCAATGATGTGGATCAAGAGTCCACATCATTGAAGTAACCTC CTTTTTT-3′. The empty vector was used as a negative control. Xg-silenced cells were selected with 0.2 μg/mL puromycin.

Messenger RNA was extracted using an Oligotex direct mRNA Micro kit (Qiagen) and reverse transcribed to cDNA using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems). Taq platinium (Invitrogen) was used for PCRs performed using an ABI PRISM 7900 Sequence Detection System (Applied Biosystems). The relative mRNA expression of Xg was determined by the comparative cycle threshold method using the mean of four housekeeping genes: Human Actin, GAPDH, HPRT, and ubiquitin. The Xg primers used for Xg PCR amplification were 5′-GATGTGGACCGTGATGACGGAC-3′ (forward) and 3′-TACCATATTGCCTTCTGGATTG CC-5′ (reverse).

Xg expression was analyzed on the primary tumors of 97 patients suffering from EWS collected from the tissue bank of the Laboratory of Experimental Oncology, Rizzoli Institute. Seventy-five patients presented with a primary localized tumor at diagnosis, whereas 22 showed metastases at diagnosis (median follow-up, 60 mo; range, 13–60 mo). Chemotherapy was administered according to consecutive protocols sequentially activated EW-REN1-3 and ISG/SSGIII-IV (30–32). A panel of pathologists reviewed histology. A standard follow-up surveillance protocol was adopted for up to 5 years and clinical data were updated. Adverse events were defined as metastasis at any site or death during remission and event-free survival was calculated from the date of initial diagnosis. Clinical pathologic features of the different series of EWS patients included in the study are summarized in Table 1. Sample sets were handled in a coded fashion. The ethical committee of the Rizzoli Institute approved the studies and informed consent was obtained from all subjects involved. Kaplan-Meier and log-rank methods were used, respectively, to draw and evaluate the significance of survival curves in EWS patients. Cox's proportional hazards regression analysis was used for multivariate analysis.

Duplicate 1-mm cores were taken from representative areas of tumors and inserted into a recipient paraffin block to create a tissue microarray containing a total of 120 cores, including six control cores of nontumor tissue (normal kidney or liver). In all cases, CD99 positivity was confirmed. Immunohistochemistry was performed on tumor cores using the NBL-1 monoclonal antibody (mAb; Anti-Xg) as the primary antibody (1:100 dilution) followed by a horseradish peroxidase secondary antibody (1:300 dilution; VECTASTAIN ABC KIT; ref. 24). Analysis of Xg expression in normal tumors was made on tissue microarrays purchased from Ambion, Inc., Cambridgeshire UK. Each tissue sample was obtained in duplicate from four individual donors, except endometrium samples that came from two individual donors.

Six-week old female athymic Nude-Foxn1 nu/nu mice were purchased from Harlan. Metastasis was determined after i.v. injection of 1 × 2.10⁶ cells. The number of pulmonary metastases was assessed 2 months after cell inoculation by counting after staining with black India ink. Twelve mice were used for each experimental group. All mice were cared for in accordance with the INSERM guidelines. Experimental procedures were approved by the local ethical committee.

Cells (1 × 5.10⁶) were stained with a PE-conjugated NBL-1 mAb (Anti-Xg) in PBS for 20 minutes at 4°C (24). After washing with PBS, cells were fixed with 1% paraformaldehyde and analyzed using a FACScan (Becton Dickinson).

Cells were lysed in radioimmunoprecipitation assay buffer containing Tris-HCl (pH 7.4), 1 mmol/L NaCl, 1% SDS, 1% Triton X-100, 1% deoxycholate and protease inhibitors mmol/L phenylmethylsulfonyl fluoride, 10 μg/mL aprotinin, and 10 μg/mL leupeptin. The protein concentration of the samples was determined by the Bio-Rad protein assay (Bio-Rad Laboratories). Fifty micrograms of total cell lysates were separated by 10% SDS-PAGE and transferred onto a nitrocellulose membrane (Amersham). The membrane was blocked with PBS containing 0.4% Tween and 5% dried nonfat milk for 1 h at room temperature and probed with the NBL-1 mAb (Anti-Xg) overnight at 4°C with gentle rocking (24). Proteins were revealed using a horseradish peroxydase–linked anti-mouse antibody (1:3000; CST) and visualized with enhanced chemiluminescence reagents (Pierce). Erk2 (TEBU) was used as a loading control.

Motility assays were performed using Transwell Permeable Supports chambers (Corning Incorporated). Cells (1 × 1.10⁶) in serum-free culture medium were seeded in the upper compartment, whereas the lower compartment contained 600 μL of medium with 10% FBS used as a source of chemoattractant. Cells were allowed to migrate for 20 h at 37°C. EWS cells were harvested with 2 mmol/L EDTA PBS. Migrated cells were counted using 25 μL of Flow-Count Fluorospheres (Beckman Coulter) and were analyzed by flow cytometry. For the invasion assay, 1 mg/mL growth factor–reduced BD Matrigel matrix (BD Biosciences) was added onto the insert and incubated for 30 minutes at 37°C to allow polymerization. Experiments were performed in triplicate.

As CD99 and Xg belong to the same family of proteins, we looked for Xg expression in EWS, in which almost all cases possess the salient feature of CD99 expression (8, 9). Quantitative real-time PCR analysis revealed that Xg transcripts were always detected in EWS cell lines, in contrast to osteosarcoma cell lines that were Xg negative (Fig. 1A). Figure 1B shows that the expression level of the Xg protein is variable in EWS cell lines, whereas it was constantly negative in osteosarcoma cell lines (Supplementary Fig. S1). In accordance with the Western blotting data, the presence of Xg on the cell surface was detected in 13 of 15 cell lines of the EWS family with widespread median cytofluorimetric fluorescence intensity (Fig. 1C). Although it was reported that Xg positivity and the CD99 level of expression are coregulated on erythrocytes (28), we found no significant correlation (P = 0.8439) for Xg and CD99 expression when evaluated by immunohistochemistry on tissues from EWS patients (data not shown). Moreover, Xg expression in EWS tumors was independent of the gender of the patients, despite the fact that the Xg molecule has X-linked expression on RBC (24). It must be noted that we examined a hundred samples of leukemia and hematosarcoma cells for which we could not find a detectable level of the Xg molecule (data not shown).

To obtain a better appraisal of Xg expression in primary EWS cells, we performed immunohistochemical analysis of Xg on paraffin-embedded tissue sections from 97 primary EWS tumors (Table 1). Patients ages ≤40 years with biopsy-proven histologic diagnosis of Ewing family of tumors (EFT) were eligible for the study. Histology was reviewed by a panel of pathologists. The patients included in this study represent ∼40% of all patients treated at the Rizzoli Institute in the period between 1995 and 2005. We selected the cases for which representative tissue material was available for the construction of tissue array and the clinical data and the follow up could be completed. This cohort included 75 patients with localized tumors and 22 patients presenting with detectable metastasis at diagnosis. Therefore, the series of patients investigated was representative of a typical EWS patient population (33). Figure 2A displays typical positive staining of EWS tissue sections, which can be compared with the Xg negativity observed in our series of EWS patients. The percentages of positivity were 24% (18 of 75) in localized tumors and 23% (5 of 22) in disseminated tumors. We evaluated the prognostic relevance of Xg expression either considering all the patients or considering only those with localized tumors. Taking into account clinical parameters, patients who showed total necrosis after preoperative chemotherapy had a favorable prognosis (Table 1). Xg expression was instead associated with adverse prognosis. This association is clearer when considering only the localized tumors and seemed to be significant when overall survival (P = 0.047) but not event-free survival (P = 0.1) was determined (Fig. 2B). Considering that the final patient outcome depends on both tumor intrinsic malignancy and response to treatment, which very likely determines overall and event-free survival, respectively, this information may indicate that Xg expression correlates with EWS aggressiveness rather than chemosensitivity. Multivariate analyses confirmed that Xg expression and the percentage of histologic tumor response to neoadiuvant chemotherapy are independent variables and confirmed a significant association with prognosis for both the percentage of necrosis and Xg expression. Cox proportional hazards regression analysis for overall survival in 75 EWS patients details were as follows: adjusted risk rate ratio of 5.67 and 2.9 for necrosis and Xg expression, respectively, 95% confidence interval of 1.27 to 25.32 and 1.12 to 7.52 for necrosis and Xg expression, respectively, P value of 0.02 and 0.03 for necrosis and Xg expression, respectively.

To investigate the effect of Xg positivity in Ewing's tumors, we transfected the naturally Xg-negative IOR/BRZ EWS cell line with an expression vector carrying Xg or with an empty vector. Using flow cytometry, two stable transfectants (Xg1 and Xg2) were selected for Xg expression at a similar level to that of EWS cell lines presented in Fig. 1C (Fig. 3A).

The data described above indicate that Xg expression correlated with the survival of EWS patients. As metastatic dissemination is a determining factor of prognosis of EWS patients, we examined the functional role for Xg in the occurrence of metastases. We assessed the in vitro migration and invasion of two Xg transfectants (Xg1 and Xg2) compared with empty vector–transfected and wild-type (WT) IOR/BRZ cells. Experiments were performed over 20 hours using a Boyden chamber in the presence or absence of FBS, which was used as a chemoattractant. Forced expression of Xg significantly increased the number of migrating cells compared with that of empty vector–transfected and WT EWS cells (P < 0.05; Fig. 3B, top). In addition, Xg expression induced a significant increase in the capacity of EWS cells to invade Matrigel, which was used to mimic the extracellular matrix (P < 0.05; Fig. 3B, bottom). These data suggest a role for Xg in the invasiveness of EWS sarcoma.

To confirm the functional role(s) of Xg in EWS cells, Xg expression was silenced in the naturally Xg-positive TC-71 EWS cell line. As described in Materials and Methods section, two distinct transfectants obtained with a mix of two anti-Xg shRNA were selected for Xg negativity (Fig. 3C). Figure 3D shows that the silencing of Xg expression significantly decreased (P < 0.05) the migratory and invasive ability of TC-71 EWS cells. Thus, Xg regulates the invasive properties of EWS cells in vitro, suggesting an important role for Xg in metastatic progression of EWS.

To investigate the importance of Xg in EWS tumor progression in vivo, WT, empty vector– and Xg-transfected IOR/BRZ cells were subsequently injected into nude mice, as indicated in the Materials and Methods section. The incidence of lung metastasis was increased in mice injected with Xg-transfected cells compared with WT and empty vector–transfected cells (Fig. 4A). In the same way, the cumulative number of lung metastases was three times higher in the Xg-injected groups of mice (Fig. 4B). To confirm the function of Xg on metastatic formation, we performed xenotransplantation in immunodeficient mice with WT, vector-, shXg1-, and shXg2-transfected TC-71 EWS cells. Consistent with these data, the shRNA-mediated knockdown of Xg in TC-71 cells decreased the incidence and the absolute number of lung metastases (Fig. 4C and D). Yet, the differences we observed between all the groups we tested did not reach significance; this could be due to the heterogeneity in the number and occurrence of metastases between all groups. Together, our results indicate an important role for Xg in EWS metastatic formation. On the basis that the occurrence of metastasis is linked to progression of EWS and so to the prognosis of EWS patients (33), our data suggest that Xg expression could be linked to the survival of patients through its involvement in metastatic formation.

We report here the first characterization of a biological function of the Xg protein that is of clinical relevance to EWS patients. Xg, initially described as a RBC protein, was expressed on nearly all the EWS cell lines tested. Among the EWS primary tumors, we found that 24% of EWS cases expressed the Xg molecule, which was significantly associated with a shorter overall survival of EWS patients with localized tumors. Xg expression was indicative of strong tumor aggressiveness. As the invasiveness of cancer cells is dependent on migratory and invasive properties, we investigated the role of Xg in the motility of EWS cells. We observed a contribution of the Xg protein to tumor invasiveness and metastatic formation.

Thus far, Xg expression has been observed only on the RBC lineage. During erythropoïesis, Xg expression appears after band 3, i.e., from the orthochromatic normoblast stage up to mature erythrocytes. However, Northern blotting and reverse transcription-PCR showed that Xg could be expressed on other cell lineages (24, 29). The Xg mRNA has also been detected in fetal tissues (liver, spleen, and thymus). Using immunohistochemical staining on an array of normal tissues, we observed that Xg was particularly well expressed in the myometrium, kidney, lung, and skeletal muscle (summarized in the Table 2). Interestingly, the distribution of the Xg protein, although rather limited, is not restricted to the RBC lineage.

Because EWS behave as very aggressive tumors, they require intensive therapy including surgery, radiotherapy, and chemotherapy to abolish micrometastasis. EWS occurs in children and young adults with a long life expectancy; it seems particularly useful to identify high-risk patients to adapt the therapy regimen and avoid toxicity. Indeed, a series of prognostic laboratory markers have been described in EWS including lactate dehydrogenase, S-100 protein, CCN3, p53, and EWS FLI fusion transcripts (34–43). Given the almost constant and the particularly high level of CD99 expression in EWS, we looked for Xg expression in these tumors in comparison with the most frequent bone tumors in children, namely osteosarcoma. Remarkably, we detected Xg at the mRNA level in EWS but not in osteosarcoma cell lines. The expression of Xg at the cell surface was confirmed by flow cytometry on 13 EWS cell lines, whereas it could not be detected at the osteosarcoma cell surface. As for primary tumors, immunohistochemistry analysis revealed, strikingly, that although 100% of EWS carry CD99, only 24% of patients displayed Xg. The expression of Xg in primary tumors of EWS patients was less frequent, possibly reflecting selection events. Xg expression in EWS patients has two major features: (a) it is independent of the gender of the patients and (b) it is not significantly linked to the level of CD99 expression. As pointed out in the introduction, CD99 is present in two isoforms due to alternative splicing (11). This is not the case for Xg, which was always expressed as a short form. Thus far we could not detect a long Xg isoform. Comparison of Xg and CD99 sequences shows that Xg shares the best homology with the short isoform of CD99 (data not shown). It is noteworthy that the forced expression of the short isoform of CD99 reduces cell-cell adhesion and enhances the cell migratory capacity (11, 13). Thus it is tempting to speculate that Xg acts in a similar fashion to the CD99 short form. Although we could not detect physical links between both molecular species, we could not exclude functional interactions between CD99 and Xg, which would reinforce the migratory capacity of the carrying tumor cells.

Therefore, we investigated the prognostic relevance of Xg expression and performed retrospective analysis on a cohort of 97 patients: 75 patients with localized tumors and 22 presenting with metastases at diagnosis. We found that in localized patients, ectopic expression of Xg is significantly linked to a shorter overall survival at 5 years (P = 0.047), indicating a putative role of this molecule in regulating the aggressiveness and metastatic ability of EWS cells.

To assess the direct effect of Xg on EWS progression, we injected immunodeficient mice with an IOR/BRZ cell line displaying forced Xg expression. We observed that forced expression of Xg in EWS cells lead to the upregulation of cellular migration and invasion in vitro and metastatic formation in vivo. Accordingly, our results, obtained with Xg-silenced EWS cell lines, confirmed the biological function of Xg. The function exerted by Xg is in line with the function already described for the other members of the CD99 family. Each of these proteins has a potent global effect on cellular adhesion and migration. Indeed, CD99 was described as being involved in the migration of different cell types (16–18).

Taken together, our results show that Xg, a member of the CD99 family of molecules, is expressed in aggressive EWS tumors. This view, however, would not exclude the possibility that CD99 and Xg could act synergistically in EWS progression and that therapeutic strategies would benefit from a combined approach targeting both molecules.

No potential conflicts of interest were disclosed.

We thank J.P. Cartron for providing us with the Nbl1 hybridoma and the Xg expression vector.

Grant Support: INSERM, European project PROgnosis and THerapeutic targets in the Ewing family of TumorS, Ministère de la Recherche et des Technologies, Institut National du Cancer Cancéropôle PACA, Association pour la Recherche sur le Cancer, La Ligue contre le cancer, Fondation de France, Fondation de la Recherche Médicale, Agence Nationale de Biomédecine, Italian Association for Cancer Research, Associazione Italiana per la Ricerca sul Cancro (K. Scotlandi), and the Italian Ministry of Health (Strategico Oncologia RFPS-2006-3-340280; Alliance against Cancer).

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.