Purpose: Bone morphogenetic proteins (BMP), belonging to the transforming growth factor-β superfamily, are important regulators of cell growth, differentiation, and apoptosis. The biological effects of BMPs on malignant lymphoma, however, remain unknown. Promoter methylation of the BMP-6 gene in lymphomas was investigated.
Experimental Design: We investigated BMP-6 promoter methylation and its gene expression in various histologic types of 90 primary lymphomas and 30 lymphoma cell lines. The effect of BMP-6 promoter hypermethylation on clinical outcome was also evaluated.
Results:BMP-6 was epigenetically inactivated in subsets of lymphomas. The silencing occurred with high frequency in diffuse large B-cell lymphoma (DLBCL) and Burkitt's lymphoma in association with aberrant BMP-6 promoter methylation. The methylation was observed in 60% (21 of 35) of DLBCL cases and 100% (7 of 7) of DLBCL cell lines, and in 83% (5 of 6) of Burkitt's lymphoma cases and 86% (12 of 14) of Burkitt's lymphoma cell lines. In contrast, other histologic types of primary lymphomas studied had little or no detectable methylation (1 of 49; 2%). The presence of BMP-6 promoter hypermethylation in DLBCL statistically correlated with a decrease in disease-free survival (P = 0.014) and overall survival (P = 0.038). Multivariate analysis showed that the methylation profile was an independent prognostic factor in predicting disease-free survival (P = 0.022) and overall survival (P = 0. 046).
Conclusion:BMP-6 promoter was hypermethylated more often in aggressive types of lymphomas, and the hypermethylation is likely to be related to the histologic type of lymphomas. BMP-6 promoter methylation may be a potential new biomarker of risk prediction in DLBCL.
Epigenetic gene silencing represents an important mechanism of inhibition of tumor suppressor gene expression in various cancers (1, 2). In contrast to the DNA mutation that alters gene-encoding sequences, the epigenetic silencing affects primarily gene promoter. Aberrant methylation of the promoter DNA regions rich in CpG island is the key step in the epigenetic gene silencing. In malignant lymphoma, such methylation-dependent gene silencing has been described for a number of genes including the cyclin-dependent kinase inhibitors CDKN2A, CDKN2B, and KIP2, the TP53 homologue TP73, the death-associated protein kinase DAPK, the cellular retinol-binding protein 1 CRBP1, and other genes with putative tumor suppressor functions, many of which encode proteins with critical functions in cell growth control and apoptosis (3, 4). Finding of additional genes whose expressions are regulated by methylation in malignant lymphomas will further identify genes having important roles in lymphomagenesis as well as prognostic implications of the diseases.
Malignant lymphomas constitute heterogeneous groups of lymphoproliferative neoplasms with different biological and clinical features. Detailed biological mechanisms leading to the development of lymphomas have not been well delineated. There have been several reports suggesting a potential role of transforming growth factor-β (TGF-β) in the pathogenesis of B-cell lymphoproliferative disorders (5–8). The TGF-β signaling regulates tumorigenesis and its pathways are often modified during tumor initiation and progression. There is abundant evidence to show that TGF-β acts as a negative regulator of various types of cancers and induces apoptosis (9–11).
Bone morphogenetic proteins (BMP), belonging to the members of the TGF-β superfamily, were originally identified as molecules that induce bone and cartilage formations and are now considered multifunctional cytokines (12). To date, more than 20 BMPs have been characterized. Based on amino acid homology, they can be subdivided into several different classes such as the BMP-2/4 group and the OP-1 group including BMP-5, BMP-6, BMP-7, and BMP-8. In various human malignancies, as with TGF-β, BMPs modulate cell differentiation and proliferation frequently in an inhibitory manner and are thought to be involved in tumorigenesis. For example, loss of BMP-2 expression has been reported in cancers of the prostate, colon, and stomach (13–15), and inactivation of BMP-3b and BMP-6 has been suggested to promote development of lung cancer (16–18). However, despite the loss of BMP expression in these cancers, the underlying mechanism has not been well defined.
There has been a lack of unambiguous data in the literature on the specific roles of BMPs in the pathogenesis of malignant lymphoma. As the promoter region of the BMP-6 gene has rich CpG islands (19), we were interested in investigating the methylation status of the BMP-6 gene promoter in malignant lymphomas with various histologic types. In this study, we found frequent aberrant hypermethylation and the resultant loss of BMP-6 expression in certain lymphomas, especially in diffuse large B-cell lymphoma (DLBCL) and Burkitt's lymphoma. The prognostic relevance of the BMP-6 promoter hypermethylation was also evaluated.
Materials and Methods
Patients and cell lines. A total of 110 lymph node tissues were obtained by incisional biopsy at presentation from 82 patients with non-Hodgkin's lymphoma, 8 patients with Hodgkin's lymphoma, and 20 patients with reactive benign lymphadenopathy. The diagnosis of malignant lymphoma was made by morphology and immunohistochemical analysis according to the WHO classification. The histologic types for the non-Hodgkin's lymphoma samples were DLBCL (n = 35, denoted as cases DL1 to DL35), follicular lymphoma (n = 19), mantle cell lymphoma (n = 7), Burkitt's lymphoma (n = 6), peripheral T-cell lymphoma, unspecified (n = 8), and angioimmunoblastic T-cell lymphoma (n = 7). All 19 follicular lymphoma cases were diagnosed as low-grade follicular lymphoma (grade 1 or 2) according to the WHO classification, and mantle cell lymphoma cases were all histologically typical mantle cell lymphoma, not inclusive of aggressive (blastoid) variants. Peripheral blood mononuclear cells from five healthy donors were also tested for the presence of BMP-6 promoter methylation. After obtaining consents for sample drawing and storage, they were stored at −80°C until processing.
The following 30 lymphoma cell lines (20) were also examined in this study: seven DLBCL lines (Pal-1, OPL-1, OPL-2, OPL-3, OPL-4, OPL-5, and OPL-7); 14 Burkitt's lymphoma lines (Akata, Katata, Raji, Daudi, DG75, Wan, Rael, BL-16, BL-41, Salina, Mutu I, BJAB, Namalwa, and Ramos); a mantle cell lymphoma line (SP-53); a Hodgkin's lymphoma line (HD-70); three primary effusion lymphoma lines (KS-1, BCBL-1, and JSC-1); and three CD30-positive anaplastic large-cell lymphoma lines (DL-40, DL-95, and DL-110). Deglis was established from a polymorphic centroblastic lymphoma (Kiel classification) with dual B-cell and T-cell gene rearrangements (21).
Methylation analysis. Isolated DNA was treated with sodium bisulfite as previously described (22). Aliquots of the bisulfite-treated DNA were amplified in reaction mixtures containing 67 mmol/L Tris-HCl (pH 8.8), 16.6 mmol/L (NH4)2SO4, 6.7 mmol/L MgCl2, 10 mmol/L β-mercaptoethanol, 1.25 mmol/L each deoxynucleotide triphosphate mixture, 0.5 μmol/L of each primer, and 1 unit of Ex Taq Hot Start Version polymerase (Takara). The primers were 5′-GGGGTAAATTTTATGGTGGTT-3′ (sense primer, designated 2F) and 5′-ACCTACRCCCTAACTCCTA-3′ (antisense, designated 2R2), which give a PCR product of 205 bp, encompassing the BMP-6 promoter region −521 to −316 bp relative to the transcriptional start site. We also used another antisense primer 4R2 (5′-CCTCAATCCTTATCTCTCATA) to amplify a 387-bp fragment corresponding to the region −521 to −134 bp relative to the transcriptional start site. The PCR condition was 3 min at 94°C followed by 35 cycles of 94°C for 40 s, 56°C for 30 s, and 72°C for 30 s. Combined bisulfite restriction analysis (COBRA) was carried out by overnight digestion of the PCR product at 60°C with the restriction enzyme BstUI (New England BioLabs), which has the recognition sequence 5′-CGCG-3′. The resultant DNA fragments were electrophoresed on agarose gels and stained with ethidium bromide. The proportion of methylated (M) versus unmethylated (U) product (digested versus undigested) was quantitated by using a densitometer to determine the density of methylation. The percent methylation was calculated as follows: M / (M + U) × 100 (23).
Bisulfite genomic sequencing. The 387-bp PCR products were purified with the Gel Extraction kit (Bionex), cloned into pGEM-T Easy vector (Promega), and transformed into Escherichia coli. Plasmid DNA from isolated clones containing the insert was purified using the FlexiPrep kit (GE Healthcare Bio-Sciences). Seven to eight clones for each primary lymphoma sample and cell line were sequenced with the pUC/M13 primer. The DNA sequence was determined by using the BigDye Terminator Cycle Sequencing kit ver.1.1 (Applied Biosystems) and an ABI 3130 automated DNA sequencer.
Reverse transcription-PCR. Total RNA was isolated using Trizol reagent (Invitrogen) according to the manufacturer's instructions. Before reverse transcription, a total of 1 μg of RNA was treated with DNase I (Invitrogen) to remove any DNA contaminant. The DNase I–treated RNAs were subjected to reverse transcription with ThermoScript reverse transcriptase (Invitrogen), as previously described (24). The cDNA samples (equivalent to the cDNA amount from 50 ng of initial total RNA) were PCR amplified, and the products were electrophoresed on agarose gels followed by ethidium bromide staining and visualization under UV light for the presence of DNA bands of appropriate sizes. The sequences of the primers to amplify the BMP-6 gene by reverse transcription-PCR (RT-PCR) were 5′-CGACAACAGAGTCGTAATCG-3′ and 5′-GCATTCTCCATCACAGTAATTG-3′, yielding a 195-bp fragment. As a control for RNA integrity and PCR reactions, the gene encoding β-actin was amplified in parallel with the primers 5′-ACCTTCAACACCCCAGCCATG-3′ and 5′-GGCCATCTCTTGCTCGAAGTC-3′, giving a 309-bp fragment.
Western blot analysis. Proteins (corresponding to 2 × 104 cells) were subjected to SDS-PAGE and electrophoretically transferred onto a polyvinylidene difluoride membrane, as previously described (24). The filter was incubated with a 1:750 dilution of the BMP-6 monoclonal antibody, and the second-step reaction was done by incubation of the washed filter with a horseradish peroxidase–conjugated rabbit anti-mouse antibody. Reactive proteins were detected by incubation of the washed filter in the enhanced chemiluminescence system according to the manufacturer's instructions (GE Healthcare Bio-Sciences) followed by exposure to an autoradiographic film. The monoclonal antibody against β-actin (Sigma) was used in parallel to confirm protein integrity and immunoblot reactions. The BMP-6 monoclonal antibody recognized a molecular mass at 64 kDa under reduced conditions, consistent with a size previously reported (25).
5-Aza-2′-deoxycytidine treatment. Two lymphoma cell lines, Pal-1 and Rael, were incubated in RPMI 1640 supplemented with 10% FCS in the presence or absence of 3 μmol/L of the demethylating agent 5-aza-2′-deoxycytidine (5-aza-dC; Sigma) for 6 days. The cells were split on day 3 with the addition of fresh drug. After the drug treatment, cells were harvested for DNA and RNA extractions.
Statistical analysis. The statistical correlation between variables was analyzed by the two-sided Fisher's exact test. Disease-free and overall survival curves were estimated by the Kaplan-Meier method and were compared with the use of the long-rank test. Multivariate analysis was done with Cox's proportional hazards regression technique to define the prognostic significance of selected covariates, including methylation status. P < 0.05 was considered to be significant.
Methylation analysis of the BMP-6 gene in primary lymphoma samples and cell lines. Lymph node tissues from pretreated patients with various histologic types of malignant lymphomas as well as nonmalignant lymph nodes with reactive hyperplasia were tested for the presence of promoter methylation of the BMP-6 gene by COBRA with PCR products amplified using a primer set of 2F and 2R2. Because aberrant hypermethylation is likely to influence not only the number of methylated CpG but also the density of methylation, we quantitated the methylation density (percent methylation; ref. 23). We analyzed the methylation status of 20 nonmalignant lymph nodes with reactive lymphadenopathy and peripheral blood mononuclear cells from five healthy donors, none of which showed detectable methylation. Because the densitometric assay showed that percent methylation of these control samples did not exceed 10%, a threshold level of 10% methylation was chosen as evidence of the presence or absence of hypermethylation of the BMP-6 promoter. Figure 1A shows representative results of the BMP-6 promoter methylation in primary lymphoma samples, and Fig. 1B summarizes the distribution of methylation density for each histologic group. The promoter methylation of the BMP-6 gene was found in 21 of 35 (60%) DLBCL samples, in 5 of 6 (83%) Burkitt's lymphoma samples, and in 1 of 8 (13%) peripheral T-cell lymphoma, unspecified samples. The methylation was not detected in any samples from follicular lymphoma (n = 19), mantle cell lymphoma (n = 7), angioimmunoblastic T-cell lymphoma (n = 7), and Hodgkin's lymphoma (n = 8). These results indicated a significant higher frequency of BMP-6 promoter methylation in DLBCL (P < 0.001) and Burkitt's lymphoma (P < 0.001) compared with other histologic types of primary lymphomas studied (1 of 49; 2%). Unlike cell lines with the fully methylated BMP-6 gene (as stated below), in these clinical samples, there were always unmethylated bands on COBRA, indicating the presence of normal tissues in the lymph node samples.
To further investigate the BMP-6 promoter methylation in malignant lymphomas, we subjected 30 lymphoma cell lines to COBRA. Illustrative examples are shown in Fig. 1C, and Table 1 summarizes the results. The BMP-6 promoter was fully methylated in all DLBCL cell lines analyzed (Pal-1, OPL-1, OPL-2, OPL-3, OPL-4, OPL-5, and OPL-7). The methylation was found in 12 of 14 (86%) Burkitt's lymphoma lines (Akata, Katata, Raji, Daudi, DG75, Wan, Rael, BL-16, BL-41, Salina, Mutu I, and BJAB), in 3 of 3 (100%) primary effusion lymphoma lines (KS-1, BCBL-1, and JSC-1), and in 2 of 3 (67%) CD30-positive anaplastic large-cell lymphoma lines (DL-40 and DL-95), whereas neither mantle cell lymphoma line SP-53 nor Hodgkin's lymphoma line HD-70 had methylated BMP-6 promoter.
To confirm these results, we repeated COBRA with PCR products amplified with another primer set 2F and 4R2. Representative results are shown in Fig. 1D. In good concordance with the results of COBRA using the primer set 2F and 2R2, DLBCL and Burkitt's lymphoma cases showed promoter hypermethylayion of BMP-6, whereas other histologic types of lymphomas tested had little or no detectable methylation.
Bisulfite sequencing analysis. To confirm the methylation at the restriction enzyme (BstUI) cleavage sites along with the methylation of other neighboring CpG sites, five primary lymphoma samples including two DLBCL cases (DL15 and DL32), a Burkitt's lymphoma case (BL1), a follicular lymphoma case (FL1), and a Hodgkin's lymphoma case (HL1), as well as three lymphoma cell lines (Pal-1, Akata, and SP-53), were subjected to bisulfite sequencing. The 387-bp PCR products were cloned into a plasmid vector, and seven to eight independent clones were sequenced. As shown in Fig. 2 , bisulfite sequencing showed good concordance with COBRA. In lymphoma cases found to be methylated by COBRA (DL15, DL32, and BL1), not only the CpG sites at the BstUI cleavage sites but also other CpG sites were partially or fully methylated (Fig. 2A), whereas the unmethylated samples, FL1 and HL1, showed little methylation at CpG sites of the promoter region. As expected, extensive methylation across the entire CpG islands was seen in the methylated cell lines of DLBCL (Pal-1) and Burkitt's lymphoma (Akata; Fig. 2B). In contrast, the unmethylated mantle cell lymphoma line SP-53 showed no evidence of promoter hypermethylation of BMP-6.
Association of the BMP-6 promoter methylation with transcriptional gene silencing. To elucidate whether the aberrrant methylation of BMP-6 is associated with loss of BMP-6 expression, we analyzed expression of BMP-6 transcripts in primary lymphoma cells as well as lymphoma cell lines by RT-PCR with cycles that amplified the β-actin cDNA to plateau levels. Representative results are shown in Fig. 3A . Primary lymphoma samples lacking the BMP-6 promoter methylation expressed BMP-6 mRNA transcripts, whereas samples having the BMP-6 promoter methylation expressed little or no detectable BMP-6 mRNAs.
Similarly, the lymphoma cells lines that were methylated at the BMP-6 promoter (for example, Pal-1, OPL-5, OPL-7, Akata, Daudi, Rael, BJAB, BCBL-1, Deglis, and DL-40) did not show expression of the BMP-6 RNA transcripts (Fig. 3A; Table 1). In contrast, the cell lines that were unmethylated, such as Ramos, Namalwa, SP-53, and HD-70, expressed the BMP-6 transcripts. Ramos cells, which had been described to express BMP-6 mRNA, served as a positive control (26). These results indicated that the BMP-6 promoter hypermethylation correlated with loss of BMP-6 expression.
We next assessed the association between this epigenetic aberration and transcriptional inactivation of the BMP-6 gene at the protein levels. Representative results are shown in Fig. 3B. Western blot analysis showed that there was a good correlation between the BMP-6 promoter methylation and BMP-6 protein expression. Methylated primary DLBCL samples (e.g., cases DL32 and DL35) did not express the BMP-6 protein, whereas the exact opposite occurred in the unmethylated samples (e.g., cases DL13 and DL27). Similarly, the hypermethylated cell lines such as Pal-1, OPL-5, OPL-7, Akata, Rael, BJAB, and DL-40 lacked expression of the BMP-6 protein, whereas unmethylated cell lines such as Ramos, Namalwa, DL-110, and SP-53 expressed the protein (Fig. 3B; Table 1).
To confirm that this loss of expression was due to the BMP-6 promoter hypermethylation, two cell lines, Pal-1 and Rael, were incubated in the presence or absence of the 5-aza-dC, and methylation status and BMP-6 mRNA expression were analyzed by COBRA and RT-PCR, respectively. Both Pal-1 and Rael cells had extensive promoter methylation of the BMP-6 gene, but treatment with 5-aza-dC led to partial demethylation (Fig. 4A ). In parallel with the demethylation, there was reexpression of the BMP-6 transcripts (Fig. 4B), implying that the methylation pattern is associated with transcriptional silencing.
Clinical correlates of the BMP-6 promoter methylation in DLBCL patients. The 35 DLBCL were divided into two groups: those with methylated BMP-6 promoter (n = 21) and those with unmethylated BMP-6 promoter (n = 14). The association between the BMP-6 promoter methylation and the clinical characteristics at presentation was analyzed by the Fisher's exact two-tail test. The presence of BMP-6 promoter methylation was not associated with any difference in the age of onset, gender, clinical stage (Ann Arbor), performance status (Eastern Cooperative Oncology Group), serum lactate dehydrogenase levels, number of sites of extranodal disease, presence of B symptoms, or bone marrow involvement (Table 2 ). Most importantly, we found that the presence of BMP-6 promoter methylation was associated with a statistically significant decrease in overall survival (Kaplan-Meier, P = 0.038) and disease-free survival (Kaplan-Meier, P = 0.014; Fig. 5 ). Multivariate analysis showed that the presence of BMP-6 promoter methylation was an independent prognostic factor for predicting both disease-free survival and overall survival in the series (Table 3 ). These observations suggested that the BMP-6 promoter methylation is a likely predictor of poor outcome in patients with DLBCL.
BMP-6 is a member of the TGF-β superfamily of signaling molecules that are important negative cell proliferation regulators inducing apoptosis in various types of cells including B lymphocytes (12, 27). BMP-6, like the other BMP members, signals through ligation of type I and type II serine-threonine kinase receptors (BMPR), which subsequently propagate the signal downstream by phosphorylating receptor-activated Smad proteins (Smad1, Smad5, and Smad8; ref. 12). These Smads then form complexes with the common mediator Smad (Smad4) and are translocated into nucleus where they exert regulation of target genes specific for the BMP pathway. Thus, BMP signaling is similar to the paradigm established by TGF-β signaling. It is logical, therefore, to assume that any functional impairment by genetic alterations or epigenetic inactivation of genes involved in the BMP/TGF-β pathway may predispose to development of malignant diseases (9–11, 28). In fact, several recent findings showed the tumor suppressor function of BMPs. First, BMP-2 inhibited the growth of cancer cells of many origins including cells derived from colorectal, prostatic, and gastric cancers (14, 29, 30). BMP-6 inhibited proliferation of prostate cancer cells by up-regulation of several cyclin-dependent kinase inhibitors (31). Similarly, BMP-7 showed growth inhibitory effect on thyroid carcinoma cells by inducing cell cycle arrest via up-regulation of the cyclin-dependent kinase inhibitors (32). These inhibitory effects of BMPs are consistent with the general characteristics of TGF-β superfamily members. Second, decreased expression of BMPR type II correlated with resistance to the growth-inhibitory effect of BMP-6 in renal cell carcinoma, suggesting that loss of sensitivity to BMP-6 is necessary to achieve the malignant phenotypes (33). Third, loss of BMP-2 expression has been reported for several cancers, including prostatic, colorectal, and gastric cancers (13–15). Furthermore, epigenetic inactivation of BMP-3b and BMP-6 by gene promoter hypermethylation promoted lung tumor development (16–18). Thus, molecular alterations involved in the BMP signaling cascade have been shown to be associated with tumorigenesis and/or disease progression in several cancers. In the realm of hematologic malignancies, several studies showed that BMPs (BMP-2, BMP-4, BMP-5, BMP-6, and BMP-7) inhibited proliferation and induce apoptosis of myeloma cells (34–37), but little is known about the biological roles of BMPs in other hematologic malignancies including malignant lymphoma.
Of these BMPs, BMP-6 came into the focus of out interest because the BMP-6 promoter sequence was previously identified as a target for aberrant DNA methylation (19). In this study, we have analyzed methylation status of the BMP-6 promoter region in primary lymphoma cells and lymphoma cell lines with various histologic types. We found intensive promoter methylation with significant high frequency in DLBCL and Burkitt's lymphoma. On the contrary, other histologic types of lymphomas tested, including Hodgkin's lymphoma, follicular lymphoma (grade 1 or 2), mantle cell lymphoma, peripheral T-cell lymphoma, unspecified, and angioimmunoblastic T-cell lymphoma, showed little or no detectable methylation. The BMP-6 promoter methylation was not observed in any of the lymph node tissues of reactive lymphadenopathy and peripheral blood mononuclear cells from healthy donors. These findings suggested that the BMP-6 promoter methylation seems to be tumor specific and that there was a strong trend toward a higher BMP-6 hypermethylation frequency in subsets of aggressive non-Hodgkin's lymphomas such as DLBCL and Burkitt's lymphoma, compared with indolent lymphomas. We also showed methylation-dependent loss of BMP-6 expression at mRNA and protein levels. Furthermore, the transcriptional repression is reversible by treatment with the demethyalting agent 5-aza-dC. Thus, our findings implied a causal relationship between methylation of the BMP-6 promoter and transcriptional repression. Our data are, to the best of our knowledge, the first demonstration of epigenetic inactivation of a BMP family member in malignant lymphoma.
Importantly, this study reports that the BMP-6 promoter hypermethylation may provide a novel independent marker for the prognostic assessment of survival in patients with DLBCL. Because the number of samples analyzed is relatively small, larger randomized prospective studies are required to confirm the prognostic significance of BMP-6 promoter hypermethylation.
In summary, our study showed for the first time that promoter of the BMP-6 gene was methylated more often in aggressive types of non-Hodgkin's lymphomas such as DLBCL and Burkitt's lymphoma. The BMP-6 promoter hypermethyaltion is likely to be related to the histologic type of malignant lymphomas with more aggressive clinical behavior. The study reported here also showed that BMP-6 promoter methylation profile seems to be an important marker in predicting the clinical outcome in DLBCL. An understanding of the BMP signaling pathways in lymphoma cells will be valuable in elucidating the molecular mechanisms of the roles of BMPs in lymphomagenesis and for the establishment of novel strategies for their treatment.
Grant support: Japanese Ministry of Education, Culture, Sports, Science, and Technology.
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.
- Received November 21, 2006.
- Revision received March 8, 2007.
- Accepted March 22, 2007.