METHYLATION STATUS OF THE RETINOBLASTOMA GENE (RB1) IN OSTEOSARCOMA: No Evidence for Hypermethylation

Abstract

Alterations of the retinoblastoma (RB1) tumor suppressor gene are not only associated with retinoblastoma but also with several other malignancies including osteosarcoma. Besides direct sequence alterations, hypermethylation of a CpG island in the promoter region can cause inactivation of the RB1 gene as it has been shown in retinoblastomas. We examined the methylation status of the RB1 gene in 25 osteosarcoma specimens by using the methylation-sensitive restriction enzymes SacII and SmaI. The restriction fragments were hybridized with clone p123, which is a 1.8-kb genomic subclone that spans the RB1 CpG island including the promoter region and exon 1. Whereas we reconfirmed hypermethylation of the RB1 gene in a sporadic retinoblastoma, no hypermethylation could be detected in the 25 osteosarcoma specimens, suggesting that hypermethylation of the RB1 promoter is not of major importance during osteosarcoma genesis.

Keywords:

• CpG-rich islands
• DNA methylation
• osteosarcoma
• promoter
• retinoblastoma gene

The retinoblastoma gene (RB1) is a prototype of tumor suppressor genes [1]. The precise function of the RB1 protein still is part of recent research. Many studies suggest that the role of this protein is to control the cell cycle through regulation of transcriptional factors such as E2F [1]. Although the gene was named because of its prominent role in the genesis of retinoblastoma, abnormalities of the RB1 gene can also be observed in a variety of other tumors, such as osteosarcomas [2–4]. Inactivation of the RB1 gene is usually caused by mutations affecting the coding region, but mutations confined to the promoter region have also been reported [5].

In addition to genetic changes such as sequence alterations or deletions, epigenetic changes such as CpG hypermethylation of a promoter which normally is unmethylated can inactivate a tumour suppressor gene [6, 7]. The 5′-end of the retinoblastoma gene has a high cytosine and guanidine content and an apparent lack of CpG suppression, consistent with a “CpG island” [8]. Hypermethylation within the promoter of the retinoblastoma gene in sporadic retinoblastoma has been described by several authors [8–10]. Moreover, the RB1 promoter can be inactivated by CpG hypermethylation in vitro [11].

Osteosarcoma is the most frequent primary malignant bone tumour. It usually occurs in the extremities of young adolescents [12]. Because patients suffering from hereditary retinoblastoma have an increased risk to develop an osteosarcoma as secondary malignancy, involvement of RB1 gene alterations in the tumor genesis of osteosarcoma was suspected, and several reports have shown alterations of the RB1 gene in osteosarcoma [2–4].

Despite its significance as a potential mechanism for gene inactivation, hypermethylation of the RB1 promoter have been reported in addition to retinoblastoma only in glioblastomas and pituitary adenomas [13, 14]. To our knowledge no study exists concerning the methylation status of the RB1 promoter in osteosarcomas.

In accordance with the reports showing hypermethylation of the promoter CpG island in retinoblastomas, we used methylation-sensitive restriction enzymes to examine the methylation status of the promoter region of the RB1 gene in three osteosarcoma cell lines and 22 osteosarcomas.

MATERIALS AND METHODS

Clinical Data

Clinical characteristics of the examined 22 osteosarcomas were taken from the Cooperative Osteosarcoma Study (COSS). Two patients developed a central low-grade osteosarcoma, 2 patients a periosteal osteosarcoma, and 18 patients a conventional high-grade osteosarcoma. (One high-grade osteosarcoma developed as secondary malignancy in a patient with a history of hereditary retinoblastoma, one periosteal osteosarcoma occurred in a patient who had suffered from several malignancies.) Most patients received chemotherapy according to the COSS protocols (COSS 86, COSS 89, COSS 92 and COSS 96). Response to chemotherapy was defined according to Salzer-Kuntschik et al. [15].

Tissue Samples

Osteosarcoma specimens were evaluated for sufficient content of vital tumor cells by microscopic examination. Tissue samples of the 22 patients were available as follows: 14 samples from primary osteosarcomas (8 untreated specimens and 6 resections after chemotherapy revealing no or only poor response to chemotherapy: grades 4–6), 1 untreated local failure, 6 specimens from metastatic lung lesions (2 prior to a second chemotherapy, 4 after a second chemotherapy which did not result in regression of tumour), and 1 untreated skin metastasis. Tumor tissues were frozen in liquid nitrogen immediately after surgical removal and stored at −80°C before analysis. Additionally, three osteosarcoma cell lines (KHOS, TE 85 and U2OS) were studied as well.

As positive control we used DNA of a sporadic retinoblastoma where the hypermethylation of the RB1 gene had been proved by methylation-sensitive Southern approaches using restriction enzymes [8].

Detection of DNA Methylation

Genomic DNA was purified from tumors and osteosarcoma cell lines according to standard methods [16]. Whereas digestion with SacI was monitored by running an aliquot in a 0.8% agarose “checking gel” before starting the methylation-sensitive reaction, complete digestion of unmethylated DNA by SacII or SmaI restriction enzymes becomes visible only after hybridization as indicated by the complete absence of the 6.1-kb fragment created by SacI activity and the appearance of 4.1- and 1.7-kb fragments (SacII) or 4.0-kb and 1.6-kb (SmaI), respectively. Therefore, we tested different conditions for complete digestion of DNA to avoid false positive results. It turns out that 4 μg of DNA in a reaction volume of 60 μL were completely digested by overnight incubation with 60 Units of SacI followed by a second overnight incubation with 60 Units of the methylation-sensitive restriction enzymes.

The restriction fragments were separated by 0.8% agarose gel electrophoresis, transferred to nylon membranes, and hybridized with ³²P-labeled DNA of clone p123 as described by Greger et al. [17]. Clone p123 is a 1.8-kb genomic subclone that spans the RB1 CpG island, including the promoter region and exon 1 [18]. Because of its high guanidine and cytosine content, it cross-hybridises to a 28S rDNA fragment [19].

RESULTS

The CpG-rich island at the 5′-end of the RB1-gene in 22 osteosarcomas and three osteosarcoma cell lines was analyzed for hypermethylation by a Southern approach using restriction enzymes. The DNA samples were digested separately with SacI combined with one methylation-sensitive restriction enzyme, either SacII or SmaI. To avoid false-positive results due to incomplete digestion by the enzymes, we first chose the appropriate incubation conditions resulting in a complete digestion of their restriction sites. The digested fragments were hybridized with a radioactive labeled DNA probe derived from the 5′-region of the RB1 gene, resulting in very prominent and clear cut signals.

The complete sequence of the human RB1 gene has been reported [20], and based on these findings the precise locations of the recognition-sequences of these methylation-sensitive restriction enzymes can be determined. The 6.1-kb fragment SacI fragment of the 5′-region of the RB1 gene contains four SacII sites, two of which only separated by 23 bp, and three SmaI sites (Figure 1).

1 Retinoblastoma gene (RB1) containing 27 exons and the 6.1-kb SacI fragment with three (distinguishable) SacII restriction enzyme cutting sites, three SmaI cutting sites, and probe p123 that spans these restriction enzyme sites. The leftmost SacII circle encloses two SacII sites, which are separated by 23 bp only. Circles = SacII sites; triangle = SmaI sites.

The ability of SacII and SmaI to digest their restriction sites is dependent on an unmethylated state of the cytosines within them. When digested with SacII, the 6.1-kb SacI fragment from the 5′-end of the Rb1 gene is cleaved into a 4.1-kb fragment, a 1.7-kb fragment, and three fragments of less than 200 bp that cannot be detected by Southern blot hybridization. Figure 2a shows the Southern blot analysis from 7 DNA samples after digestion with the restriction endonucleases SacI and SacII. All tumor samples show signals corresponding to a 4.1-kb and 1.7-kb fragment, consistent with a non-methylated status of the SacII restriction sites.

2 Southern blot analysis of tumor DNA with methylation-sensitive restriction enzymes SacII (Figure 2a) and SmaI (Figure 2b). DNA was digested with the indicated enzymes and analyzed by Southern Blot hybridization with p123. The probe identifies restriction fragments from the 5′end of the RB1 gene and a 28S rDNA fragment. This figure shows the results of digesting DNA of 7 tumors with SacI and SacII respectively SacI and SmaI. The osteosarcoma DNA show fragments as expected if SacII respectively SmaI restriction sites are not methylated. Kb = kilobases. (a) 1 = Placenta DNA as normal control only with SacI; 2 = placenta DNA with SacI and SacII; 3–9 = osteosarcoma DNA with SacI and SacII. (b) 1 = Placenta DNA only with SacI; 2 = placenta DNA with SacI and SmaI; 3–10 = osteosarcoma DNA with SacI and SmaI.

After digestion with SmaI, the SacI fragment is cleaved into a 4.0-kb fragment and a 1.6-kb fragment if the recognition sequence is unmethylated. Figure 2b demonstrates the cleavage pattern of the same osteosarcoma DNA as in Figure 2a corresponding to an unmethylated status of the SmaI restriction sites. Examining the three osteosarcoma cell lines U2OS, TE 85, and KHOS, we received equal results (not shown).

In order to support the reliability of the approach, we used DNA derived from a retinoblastoma that was proved to harbor a methylated RB1 gene as positive control [8]. Figure 3 shows the different cleavage pattern for the methylated RB1 gene from the retinoblastoma and the unmethylated RB1 gene derived from an osteosarcoma. The methylated control DNA is completely resistant to the digestion with SacII and a heterogenous methylation pattern was observed when analysed with SmaI restriction enzyme, whereas the osteosarcoma DNA shows the expected fragments if the recognition sites are not methylated.

3 Hypermethylated RB1 gene derived from a sporadic retinoblastoma as positive control. 1 = Hypermethylated RB1 gene derived from a sporadic retinoblastoma digested with SacI and SacII. The 6.1-kb fragment persists indicating methylated restriction sites. 2 = Unmethylated RB1 gene derived from an osteosarcoma digested with SacI and SacII. 3 = Hypermethylated RB1 gene derived from a sporadic retinoblastoma digested with SacI and SmaI. 6.1-kb, 4.0-kb, and 2.0-kb fragments due to heterogenous methylation (Greger et al. 1989). 4 = Unmethylated RB1 gene derived from an osteosarcoma digested with SacI and SmaI.

In summary, none of the 25 osteosarcomas investigated showed hypermethylation of any of the CpG sites located within the targeted (six distinguishable by our approach) methylation-sensitive restriction sites.

DISCUSSION

A growing list of (typical) cancer related key genes which can be silenced by hypermethylation of CpG-rich islands in their promoter region have been documented in various malignancies [21]. Although the RB1 gene is commonly mutated in a variety of human cancer, hypermethylation and consecutive inactivation of the gene has been reported first of all only in retinoblastoma and later also in glioblastomas and pituitary adenomas, as far as we know [13, 14].

The hypermethylated region investigated in retinoblastomas includes the RB1 promoter and exon 1. Today, 241 retinoblastoma have been examined in order to detect hypermethylation in this region. In summary, 13 of 140 unilateral tumours and 1 of 101 hereditary retinoblastomas revealed hypermethylation [8–10].

Since osteosarcoma often showed mutations of the RB1 gene similar to retinoblastoma, we were interested to find out whether hypermethylation of the RB1 gene also contributes to osteosarcoma development. It was discussed that RB1 mutations in osteosarcoma may correlate with a poor prognosis of the disease and/or that inhibition of this tumour suppressor may belong to the genetic necessities of osteosarcoma development [3, 22]. Therefore, it is of biological interest, whether in addition (or alternatively) to direct mutations of the RB1 gene, hypermethylation of the promoter take part during osteosarcoma development as well. Moreover, if there were osteosarcomas with inactivated RB1 function due to hypermethylation of the 5′ part of this tumour suppressor gene, these patients might benefit from treatment with DNA demethylating agents.

In this study we examined the methylation status of the RB1 promoter and exon 1 in 22 osteosarcoma obtained from the German Cooperative Osteosarcoma Study (COSS) and three osteosarcoma cell lines. Using the methylation-sensitive restriction enzymes SacII and SmaI, we failed to demonstrate hypermethylation in any of the 25 osteosarcoma specimens examined, whereas the positive control revealed hypermethylation of the CpG island, supporting the reliability of the method applied. Moreover, all reported hypermethylated retinoblastoma samples have been detected by the same approach using restriction enzymes and Southern Blot analysis. At least two of the three SacII sites are hypermethylated in 13 of 14 tumors that have been identified in previous studies [8–10]. Furthermore, Stirzacker et al. showed that the entire RB1 CpG island was extensively methylated and that hypermethylation in sporadic retinoblastomas was not confined to specific CpG sites but occurs throughout the CpG island [23]. Thus, it is very improbable that by chance the CpG sites within the seven targeted recognition sites remain unmodified while the other sites located in the CpG island become methylated. Nevertheless, we have to keep in mind that our approach may not be suitable for detecting a weak methylation in a sample where all but very few cells harbor an unmethylated RB1 promoter. Here, a methylation-specific PCR resulting in amplification of rare methylated molecules might be superior [24].

Today, the mechanisms of hypermethylation of DNA sequences are not completely understood. However, a widely accepted hypothesis is that over expression of the DNA methyltransferase (DNA MTase) results in hypermethylation [6, 26–28]. Furthermore, the activity of DNA MTase is upregulated by the Ras oncogenic signalling pathway [6], suggesting that hypermethylation is a downstream component of an oncogenic program. In this context, it is of interest that Antillon-Klussman et al. [29] did not find any mutation of the Ras gene in 96 osteosarcoma specimens.

Our investigation give a first direct hint that hypermethylation of the 5′ part of the RB1 gene is not of major importance for the development of osteosarcoma. Since we also analyzed osteosarcoma relapses in addition to primary tumors, RB1 promoter methylation seems neither to contribute frequently to the early development of osteosarcoma nor to its progression. Nevertheless, a possible participation of RB1 inhibition by hypermethylation of its promoter region in rare cases cannot be excluded due to the limited number of osteosarcomas investigated by us.

We are very grateful to Prof. Dr. Bernhard Horsthemke for providing the RB1 gene-specific cDNA probe p123 and DNA harboring a hypermethylated RB1 gene derived from a sporadic retinoblastoma. Furthermore, we thank the members of the Cooperative Osteosarcoma Study (COSS) for providing tumor material and clinical data. We also thank the Deutsche Krebshilfe, Hamburger Landesverband für Krebsforschung, the Fördergemeinschaft zur Erforschung und Heilung von Krebskrankheiten e.V. Hamburg, and the Werner Otto Stiftung for financial support.