Human osteosarcoma expresses specific ephrin profiles
Implications for tumorigenicity and prognosis
Article first published online: 31 JUL 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 4, pages 862–869, 15 August 2002
How to Cite
Varelias, A., Koblar, S. A., Cowled, P. A., Carter, C. D. and Clayer, M. (2002), Human osteosarcoma expresses specific ephrin profiles. Cancer, 95: 862–869. doi: 10.1002/cncr.10749
- Issue published online: 31 JUL 2002
- Article first published online: 31 JUL 2002
- Manuscript Accepted: 21 MAR 2002
- Manuscript Revised: 5 MAR 2002
- Manuscript Received: 6 DEC 2001
- reverse transcription-polymerase chain reaction analysis;
The molecular mechanisms underlying malignancy of osteosarcoma are unknown. It has been reported that eph receptor protein tyrosine kinases and their ligands, ephrins, are associated with increased tumorigenicity in patients with breast carcinoma and melanoma. The expression and role of eph/ephrins in human osteosarcoma has not yet been characterized.
Ephrin-A1, ephrin-A3, ephrin-A4, ephrin-A5, ephrin-B1, ephrin-B2, and ephrin-B3 mRNA expression was examined by reverse transcription polymerase chain reaction analysis in nine specimens of human osteosarcoma tissue and five human osteosarcoma cell lines. Ephrin-B1 protein expression was detected immunohistochemically in human osteosarcoma tissue. Clinicopathologic correlation was made between the osteosarcoma specimens and their ephrin expression profiles.
Normal bone specimens, osteosarcoma tissue specimens, and osteosarcoma cell lines expressed a distinct mRNA profile of ephrin-A1, ephrin-A4, and ephrin-B2. A second mRNA profile that included ephrin-A3, ephrin-A5, and ephrin-B1 was expressed by a subset of tumors. The expression of ephrin-B1 was correlated with a poorer clinical prognosis. Ephrin-B1 protein was expressed by osteosarcoma cells and blood vessels.
The results of this study suggest that ephrin-B1 expressed by osteosarcoma may be a poor prognostic marker through increased tumorigenicity. Cancer 2002;95:862–9. © 2002 American Cancer Society.
Osteosarcoma is a rare, malignant neoplasm of bone. It may exist as a low-grade lesion with low malignant potential for metastasis through to a high-grade lesion with a high potential for distant metastasis. Variability in tumor grade and histologic pattern occurs, with only the production of malignant osteoid remaining a consistent feature. The majority of osteosarcomas are high grade; however, even within this group of patients, there is much variation with respect to treatment and prognosis. The response of osteosarcoma to chemotherapy is currently the most reliable prognostic indicator.1, 2 Little is known about the molecular mechanisms that control the tumorigenicity of osteosarcoma.
Receptor protein tyrosine kinases (RPTKs) transduce extracellular signals inside the cell and play critical roles in cell growth, differentiation, and migration.3 Eph receptors are the largest known subfamily of the RPTKs, consisting of 14 receptors and 9 interacting ligands, the ephrins. Ephrins are membrane-bound proteins attached by either a glycosylphosphatidylinositol linkage (ephrin-A subclass) or through a transmembrane domain (ephrin-B subclass). The subdivision of ephrins corresponds to their binding preferences, with ephrin-A ligands binding to EphA receptors and ephrin-B ligands binding to EphB receptors in the main. The exception is EphA4, which is capable of binding to ephrins from either subclass.4, 5 The functional role of this subfamily has been investigated predominantly during embryonic development and was found to coordinate processes of cell migration and membrane motility in axon guidance of neurons.6–9
Isolation of the first eph, EphA1, was made from a hepatocellular carcinoma cell line10; since then, it has been found that many other carcinomas express members of this RPTK subfamily.11–18 The up-regulation of ephrin-A1 mRNA and the receptor to which it can bind, EphA2, in human melanoma has been reported.12, 13 The over-expression of ephrin-B2 mRNA was reported in patients with malignant melanoma and was correlated with increased tumorigenicity and metastatic potential of malignant melanoma.17 EphB2 has been identified in cell lines of gastric, esophageal, and colonic carcinoma but rarely in sarcomas.11 Ephrin expression has not been reported in sarcomas, although their cognate receptors, Eph RPTKs, have been identified in one osteosarcoma cell line, HuO-3N1.11 We investigated the mRNA expression profile of ephrins in human osteosarcoma tissue specimens and cell lines, and we examined ephrin-B1 protein expression using immunohistochemistry. The clinical course for individual patients was correlated with ephrin-B1 expression to analyze its potential as a prognostic marker of tumorigenicity.
MATERIALS AND METHODS
Patient Samples and Cell Lines
Tumor specimens were collected at the time of surgery, snap frozen in liquid nitrogen, and stored at −70 °C. RNA samples from tumor specimens T4 and T5 were kindly provided by Dr. D. Findlay (Royal Adelaide Hospital, South Australia, Australia). The 143B, G-292, MG-63, SAOS-2, and SJSA-1 osteosarcoma cell lines were a generous gift from Dr. R. Reddell (Children's Medical Research Institute, New South Wales, Australia). The cell lines were cultured in Dulbecco modified Eagle medium supplemented with 10% fetal calf serum and 2 mM L-glutamine and maintained in 5% CO2 in a humidified incubator at 37 °C.
Total RNA Extraction
Total RNA was extracted from tissue using TRIzol® reagent (Life Technologies Inc., Gaithersburg, MD). The RNA was quantitated by measuring the absorbance at 260 nm using a Beckman DU650 spectophotometer (Beckman Instruments, Palo Alto, CA). Any genomic DNA contamination of RNA preparations was removed by digestion with DNase I (Life Technologies Inc.). Briefly, 1 μg total RNA was incubated with 1 U DNase I for 15 minutes at room temperature. DNase I was inactivated by the addition of 1 μL of 25 mM ethylenediamine tetraacetic acid and heated at 65 °C for 10 minutes. RNA samples were ready for use in reverse transcription (RT).
First-strand complementary DNA (cDNA) was synthesized by RT. Briefly, 1 μg total RNA (DNase I treated) was mixed with 0.5 μg oligo dT in a final volume of 12 μL with sterile RNase free water. The RNA was denatured by heating at 70 °C for 10 minutes and then cooled immediately by transferring to ice. To this mixture, 4 μL 5 × First-strand buffer consisting of 2 μL 0.1 M dithiothreitol and 1 μL 10 mM dNTP mix (10 mM of each deoxynucleotide) was added and incubated at 42 °C for 2 minutes. Two hundred units (1 μL) of Superscript II (Life Technologies Inc.) were added, and the reaction was incubated at 42 °C for 50 minutes. The enzyme was inactivated by heating at 70 °C for 15 minutes.
cDNA (10% of the RT volume) was added to a mixture of 1 × PCR buffer (Life Technologies Inc.), 1.5 mM MgCl2 (Life Technologies Inc.), 200 μm of each dNTP (Promega, Madison, WI), 1 mM of each oligonucleotide primer, and 1 U Taq DNA polymerase (Promega) in sterile, RNase free water to a final volume of 25 μL. To amplify ephrin cDNA, the PCR reaction was heated to 94 °C for 5 minutes before 35 cycles at 94 °C for 1 minute, 58 °C for 1 minute, and 72 °C for 1 minute followed by a final extension at 72 °C for 7 minutes. To amplify β-actin cDNA, the PCR reaction was heated to 94 °C for 5 minutes before 23 cycles at 94 °C for 1 minute, 55 °C for 30 seconds, and 72 °C for 30 seconds followed by a final extension at 72 °C for 7 minutes. The PCR cycle numbers for amplification of ephrin and β-actin cDNA were within the linear range of PCR (data not shown). The primer sequences (5′ to 3′) that were used for the PCR amplification of ephrin cDNA were as follows: ephrin-A1 (accession no. 004428), A1F (ctgctgatcgccacaccgtcttc; nucleotides [nt] 123–145) and A1R (gcactgtgaccgatgctatgtag; nt 599–621), yielding a 501-base pair (bp) PCR product; ephrin-A3 (accession no. 004952), A3F (ctatctggatatttactgcccg; nt 171–192) and A3R (cgcagcagacgaacaccttcat; nt 481–502), yielding a 334-bp PCR product; ephrin-A4 (accession no. 005227), A4F (ctcaacgattacctagacattg; nt 175–196) and A4R (gcagtaatagcaagagacag; nt 579– 598), yielding a 426-bp PCR product; ephrin-A5 (accession no. 001962), A5F (ctacgctctctactggaacagc; nt 375–396) and A5R (gactcatgtacggtgtcatctgc; nt 844–866), yielding a 494-bp PCR product; ephrin-B1 (accession no. 004429), B1F (ctcggcaagtggcttgtggcga; nt 726–747) and B1R (gctgctcaggcgtcacagcattg; nt 1205–1227), yielding a 504-bp PCR product; ephrin-B2 (accession no. 004093), B2F (ctctcctcaactgtgccaaacc; nt 297–318) and B2R (gatgaagatgatgcatcctgaag; nt 702–724), yielding a 430-bp PCR product; and ephrin-B3 (accession no. 001406), B3F (ctccttctcacttgtgatcgcc; nt 911–932) and B3R (cagccactgcaggcatgctggg; nt 1280–1301), yielding a 393-bp PCR product.
The primer sequences (5′ to 3 ′) that were used to amplify β-actin cDNA were as follows: actin F (ctgactgactacctcatgaag) and actin R (gaggagcaatgatcttgatct), yielding a 447-bp PCR product. All PCR reactions were performed on a PTC-200 DNA Engine thermal cycler (MJ Research Inc., Cambridge, MA). PCR products were electrophoresed through 1.2% agarose gels, stained with ethidium bromide (1 μg/mL), and analyzed using the Kodak Digital Science® Electrophoresis Documentation and Analysis System 120 (Eastman Kodak, Rochester, NY).
Antigen retrieval was undertaken using the technique of Shi et al.19 Briefly, 3–4 μm sections were cut from formalin fixed, paraffin embedded human osteosarcoma specimens. The sections were mounted on 3-aminopropyltriethoxysilane (silane)-coated slides and treated in a microwave in 0.01 M sodium citrate buffer, pH 6.0, for 10 minutes. The sections were washed with phosphate-buffered saline (PBS) followed by distilled water before endogenous peroxidase activity was blocked with 3% hydrogen peroxide. The slides were washed with PBS and incubated with anti-ephrin B1 (A-20) polyclonal antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA) in a dilution of 1:125 overnight at 4 °C. The sections were washed with PBS, and bound primary antibody was labeled using EnVision™ horse radish peroxidase (Dako Corporation, Carpinteria, CA) for 30 minutes at 4 °C. The sections were washed with PBS, and the tissue bound peroxidase activity was detected using the chromagen 3′3′-diaminobenzidine tetrahydrochloride (Dako Corporation). Sections were counterstained with hematoxylin, then washed in tap water, dehydrated in alcohol, and mounted in DePeX. Negative controls were provided by substituting the ephrin B1 antibody with a murine immunoglobulin G1 monoclonal antibody reactive against 2,4,6-trinitrophenyl (TIB-191).
Human Osteosarcoma Expresses a Unique mRNA Profile of Ephrin-A3, Ephrin-A5, and Ephrin-B1
We examined the mRNA expression profile of ephrin ligands from nine specimens of human osteosarcoma (Table 1). Seven human osteosarcoma specimens were obtained from the primary bone site, and two specimens were obtained from sites of recurrence (specimens T3 and T8). One patient (Patient B) had the initial osteosarcoma specimen (specimen T2) taken from the primary bone site and had a second specimen taken from a recurrence in the abdomen (specimen T3), thus providing a unique opportunity to analyze change in ephrin expression with advance of the malignancy.
|Tumor specimen||Patient||Age (yrs)||Gender||Tumor type||Grade||Source||Site||Metastatic spread|
RT-PCR analysis was used due to the limited amount of the tissue available from surgical specimens and the low yield of mRNA extracted from bone. We did not examine ephrin-A2 and ephrin-A6 mRNA expression due to the lack of appropriate human cDNA positive controls. We found a specific mRNA profile of three ephrin ligands expressed in normal bone (n = 2 normal bone specimens) and in all osteosarcoma specimens (T1–T9): ephrin-A1, ephrin-A4, and ephrin-B2 (Fig. 1). We were unable to amplify an ephrin-B3 PCR product from either normal bone or osteosarcoma, even though β-actin was amplified successfully from the same cDNA template, and an ephrin-B3 PCR product was amplified from an ephrin-B3-containing plasmid. Therefore, we concluded that ephrin-B3 is not expressed in these tissues.
After obtaining this result, we were particularly interested to determine whether a unique mRNA profile of ephrin expression in human osteosarcoma specimens existed that was not present in normal bone. We found a unique expression profile of three ephrin ligands in osteosarcoma: ephrin-A3, ephrin-A5, and ephrin-B1 (Fig. 1). Of these three ligands, ephrin-B1 appeared to be the most interesting, with clear expression in four human osteosarcoma specimens (T3, T4, T5, and T8). Two of these specimens (T3 and T8) were taken from patients with clinically advanced osteosarcomas that had recurred with metastasis to distant sites from the original primary bone site. Osteosarcoma specimens T2 and T3 from Patient B demonstrated the acquisition of ephrin-B1 mRNA expression during the progression of this malignancy. Specimen T2 was from a patient with the primary bone site of osteosarcoma in the pelvis, and specimen T3 was taken 1 year later when a local recurrence was detected in the abdominal wall. Osteosarcoma specimen T4, which was identified as a Paget sarcoma, was a clinically more aggressive malignancy. Therefore, all specimens that expressed ephrin-B1 were correlated clinically with an osteosarcoma of greater malignancy. The expression of ephrin-A3 mRNA (1 of 9 specimens) and ephrin-A5 mRNA (4 of 9 specimens) in the osteosarcoma specimens did not have a clear clinical correlation with malignancy like that found with ephrin-B1 expression.
Human Osteosarcoma Cell Lines Express Ephrin-A1, Ephrin-A4, and Ephrin-B2 mRNA
We observed two different ephrin expression profiles from nine human tissue specimens of osteosarcoma: 1) Consistent with normal bone, osteosarcoma specimens expressed ephrin-A1, ephrin-A4, and ephrin-B2 mRNA. 2) A unique profile of ephrin-A3, ephrin-A5, and ephrin-B1 mRNA was observed in osteosarcoma specimens that was not present in normal bone. This led us to examine the ephrin profile expressed by human osteosarcoma cell lines.
Five different human osteosarcoma cell lines were obtained, but detailed clinical data were unavailable. We found that the ephrin-A1, ephrin-A4, and ephrin-B2 profile was expressed by almost all of the human osteosarcoma cell lines (Fig. 2). This was consistent with our findings from the human tissue surgical specimens.
All five osteosarcoma cell lines expressed ephrin-A4 and ephrin-B2 transcripts (Fig. 2). Ephrin-A1 displayed a variable expression pattern, in that only three of five osteosarcoma cell lines (143B, SAOS-2, and SJSA-1) expressed this transcript. We also amplified a 450-bp product with ephrin-A1 primers in the 143B cell line that was the same size as a second product amplified in three of the human osteosarcoma specimens (see Fig. 1, specimens T3, T7, and T8). Direct sequencing of this smaller PCR product confirmed its identity as an alternatively spliced version of ephrin-A1 with a putative 66-bp exon omitted (data not shown). Because the genomic sequence for ephrin-A1 is not yet available, the identity of the exon remains unclear. Spliced forms of members of the Eph family have been reported previously, namely, for ephrin-A420 and EphB221.
The ephrin-A3, ephrin-A5, and ephrin-B1 mRNA expression profile was not observed in any osteosarcoma cell line. Finally, ephrin-B3 was not expressed in any of the osteosarcoma cell lines, which was consistent with its absence from the human osteosarcoma tissue specimens.
Ephrin-B1 Protein Expressed by Osteosarcoma Cells and Blood Vessels
In the T3, T4, T5, and T8 human osteosarcoma specimens, we determined that ephrin-B1 mRNA was expressed and correlated with a more aggressive malignancy clinically. Using a polyclonal antibody to human ephrin-B1, we confirmed this finding by demonstrating that ephrin-B1 protein was expressed by the T8 human osteosarcoma specimen (Fig. 3).
We found that osteosarcoma cells expressed ephrin-B1 protein with strong staining in the cytoplasm (Fig. 3A). Another pattern was observed with ephrin-B1 immunoreactivity localized to endothelial cells lining the lumen of blood vessels in the tumor (Fig. 3B). We have used this anti-ephrin-B1 antibody in Western blot analysis and immunohistochemical reactions to other tissues and have found that it is specifically immunoreactive to ephrin-B1 (data not shown). In the T8 human osteosarcoma specimen, no background immunoreactivity was evident using a negative control antibody (Fig. 3C).
The major finding of this study is that ephrin-B1 mRNA and protein are expressed in a subset of human osteosarcoma tumors and are correlated with clinical data, suggesting that ephrin-B1 expression may indicate a more aggressive malignancy. The clinicopathologic relation between ephrin-B1 expression and a poorer clinical prognosis requires statistical support by analysis of larger numbers of patients with osteosarcomas. However, the finding that ephrin-B1 protein is expressed by osteosarcoma cells and blood vessels provides histologic evidence that interactions of this ephrin ligand with a cognate Eph receptor or receptors may control mechanisms of increased tumor growth or metastasis. Similar assertions relating to tumorigenicity and metastatic potential have been reported for other ephrin ligands expressed by human tumors, namely, ephrin-B2 and ephrin-A1 in melanoma13, 17 and ephrin-A1 in breast carcinoma and Kaposi sarcoma.14
We found that ephrin-B1 mRNA was not expressed by normal bone or in osteosarcoma of low-grade malignancy (specimen T1). In one patient, the osteosarcoma acquired expression of ephrin-B1 with local recurrence. Three other patients with osteosarcomas that expressed ephrin-B1 developed distant metastases in subsequent clinical follow-up. To date, only one of the remaining four patients with osteosarcoma that was negative for ephrin-B1 has developed metastatic disease. Ephrin-B1 was expressed by two of the four osteosarcomas (specimens T3 and T8), which were recurrent tumors in patients who had been treated with chemotherapy. This suggests another explanation why a subset of osteosarcoma expressed ephrin-B1: Chemotherapy may have induced ephrin-B1 expression in these tumors or may have selected for a population of ephrin-B1-expressing tumor cells. However, this would not explain why two osteosarcoma specimens that had not been exposed previously to chemotherapy expressed ephrin-B1. Although this hypothesis seems less tenable than the correlation of ephrin-B1 expression with increased tumorigenicity, further investigation of osteosarcoma specimens from nontreated and postchemotherapy treated patients is required.
The expression of ephrin-B1 protein by endothelial cells of blood vessels within osteosarcoma had a pattern similar to that found for ephrin-A1 in human breast carcinoma.14 The clinical correlation that ephrin-B1 positivity in osteosarcoma may indicate a more aggressive malignancy and, ultimately, a worse prognosis suggests that ephrin-B1 expression in blood vessels may have a role in tumorigenicity. Whether this is due to increased tumor growth from neovascularization, metastatic spread through blood vessels, or another mechanism remains to be determined. Evidence from ephrin-A1 and EphA2 expression studies in human breast carcinoma and in in vitro angiogenesis experiments suggest that ephrin-A1 interacting with its cognate EphA2 receptor may influence tumor neovascularization.14 The role of B-subclass ephrin ligands and Eph receptors in tumor neovascularization has not been reported to our knowledge. However, there is strong evidence that these molecules play a significant role in angiogenesis during embryogenic development.22, 23
The possible role of ephrin-B1 expression in the cytoplasm of osteosarcoma cells is more difficult to understand in this process of tumor neovascularization. The finding that osteosarcoma cells clearly express ephrin-B1 and that this may indicate a worse clinical prognosis raises several possibilities for its function: a role in osteosarcoma cell growth, survival, or metastatic spread. There are conflicting reports of clinical correlations between increased ephrin ligand expression in other tumor types and prognosis. The up-regulation of ephrin-A113 and ephrin-B217 expression by melanoma cells reportedly is associated with a poor prognosis, whereas patients with ephrin-B2 expression in neuroblastoma had a more favorable clinical outcome.15, 24
The specific expression profile of ephrin-A1, ephrin-A4, and ephrin-B2 that was found in almost all specimens of normal bone, osteosarcoma tissue, and osteosarcoma cell lines suggests that there may be a constitutive level of mRNA expressed by bone that has a role independent of tumorigenicity. This role can be clarified by determining the cell type or types that express these ligands. Their presence in osteosarcoma cell lines would suggest that ephrin-A1, ephrin-A4, and ephrin-B2 may be found in a specific bone cell type, the osteoblast. Functional evidence from studies in embryonic development have shown that eph/ephrins coordinate cell migration and other processes involving cell-to-cell contact8; thus, ephrin-A1, ephrin-A4, and ephrin-B2 expression may be involved in similar processes in bone.
The converse may exist for the expression profile ephrin-A3, ephrin-A5, and ephrin-B1. Why the five osteosarcoma cell lines examined lacked this expression profile is unknown. A possible explanation may be gene silencing as a result of DNA methylation of the promoter region of the ephrin-A3, ephrin-A5, and ephrin-B1 genes, as reported previously for EphA3 in human leukemia.25 All osteosarcomas are malignant, and we have found that a subset of patients with tumors that express the ephrin-A3, ephrin-A5, and ephrin-B1 profile may have a worse prognosis. It would be of interest to generate an osteosarcoma cell line that expresses this specific ephrin profile and to examine the tumorigenic characteristics of such a cell line compared with the five cell lines we have investigated that were negative.
The need for further investigation of the role of ephrin ligands in osteosarcoma is well supported by the findings from this study. In particular, the role of ephrin-B1 is of most interest, in that it may be an attractive candidate as a poor prognostic marker or a possible target for therapeutic intervention.
The authors thank Dr. D. Cerretti from Immunex Research and Development Corporation, Seattle, Washington, for the plasmid constructs containing human ephrin cDNA. They thank Dr. R. Reddell from the Children's Medical Research Institute, Sydney, Australia, for the generous gift of osteosarcoma cell lines. They thank Dr. D. Findlay and Dr. A. Evdokiou from the Department of Orthopedics and Trauma, The Royal Adelaide Hospital, Adelaide, Australia, for providing RNA from two patients with osteosarcoma. Finally, the authors thank Mr. David Taylor from Drs. King & Mower, Histopathology and Cytopathology, Adelaide, Australia, for the immunohistochemical staining of osteosarcoma specimens.