Activation of RAS is a common feature in a wide spectrum of tumours. In multiple myeloma (MM), RAS mutations have been observed at variable incidences with most of them >40% and showing generally more common N- than K-RAS mutations in western studies (Neri et al, 1989; Portier et al, 1992; Corradini et al, 1993; Liu et al, 1996; Feinman et al, 1997; Bezieau et al, 2001). However, data on Chinese patients with MM have been lacking. In in vitro studies, the introduction of an activated H- and N-RAS into an Epstein-Barr virus-transformed B-lymphoblastoid cell line resulted in differentiation and transformation into malignant plasma cells (Seremetis et al, 1989). Mutant RAS expression also affects the growth/survival characteristics of the myeloma cell line ANBL6 by promoting interleukin-6 (IL-6) independent growth and resistance to apoptosis from IL-6 deprivation (Billadeau et al, 1995). In contrast to the growth enhancement effects, it has been shown that Ras-p21 proteins may also induce apoptosis (Vos et al, 2000). These anti-neoplastic effects may be equally important during RAS-dependent tumour development (Vos et al, 2000). RASSF1A, a tumour suppressor gene found on chromosome 3p21.3, has been shown epigenetically inactivated in multiple cancers (Dammann et al, 2000, 2001; Agathanggelou et al, 2001; Lo et al, 2001). It has been demonstrated that RASSF1 may serve as a novel RAS effector protein mediating the apoptotic effects of oncogenic RAS (Vos et al, 2000). B-Ras associated factor (BRAF) is a member of the RAF family, which codes for kinases that are regulated by binding Ras-p21 and mediates cellular response to growth signals (Davies et al, 2002; Rajagopalan et al, 2002). BRAF mutations have been observed in a high frequency in melanomas and recently, also in colorectal cancers (Davies et al, 2002; Rajagopalan et al, 2002). To study the involvement of RAS signalling pathway in MM pathogenesis, we investigated the methylation status, mutation and expression of RASSF1A, mutations of RAS and BRAF in primary MM samples from Chinese patients and MM-derived cell lines.
Summary. The methylation status, mutation and expression of RASSF1A, and mutations of RAS and BRAF were studied in 52 patients with multiple myeloma (MM), one plasma cell leukaemia (PCL) patient and four MM-derived cell lines. Aberrant methylation of RASSF1A was found in nine of 32 MM patients and in one of four MM cell lines (U266), where the associated loss of transcription was reversible by demethylation treatment. RASSF1A transcription was further investigated on anti-CD138-sorted plasma cell-enriched bone marrow samples from 10 MM, one PCL and three reactive plasmacytosis patients. While the wild-type RASSF1A transcript was detected in all three reactive plasmacytosis and the PCL samples, we found no detectable wild-type transcripts in six of 10 MM samples studied. In two MM samples, only the non-functional variant transcript was detected, whereas the other four showed loss of transcription. In great contrast to western data, RAS mutations were identified in only four of 31 (13%) MM patients. While no RASSF1A or BRAF mutation (V599E) was detected in any of the primary MM studied (n = 21), the latter was found in the U266 cell line. Taken together, these data indicate that alterations of RAS signalling are critical in MM pathogenesis. In our current studies of Chinese MM patients, these alterations involved frequent RASSF1A inactivation (60%) as a result of transcriptional silencing or expression of a non-functional variant transcript.
Patients and methods
Patients and samples.
With informed consent, a total of 52 primary MM and one plasma cell leukaemia (PCL) patients, diagnosed and treated at Prince of Wales Hospital or Pamela Youde Nethersole Eastern Hospital, Hong Kong (SAR), were recruited for the study. Diagnosis and staging were identified according to the criteria of Salmon and Durie (Durie & Salmon, 1975; Durie, 1998). In addition, bone marrow (BM) samples showing plasmacytosis from three patients presenting with pyrexia of unknown origin that had been shown not to harbour any malignancies were recruited to provide the source of reactive plasma cells (RP1–3). Unsorted BM samples from 32 MM patients [unique patient number (UPN) 3–34] were analysed for RASSF1A promoter hypermethylation and 21 of these 32 samples were also evaluated for mutation of RASSF1A, RAS and BRAF (Table I). However, BM materials from 20 MM, one PCL (UPN 42–52 and 61–70) and the three reactive plasmacytosis patients were subjected to immunomagentic cell sorting using anti-CD138. These plasma cell-enriched BM samples (>98% plasma cells) were then analysed for RASSF1A mRNA expression [10 MM, one PCL (UPN 42–52), RP1–3] and for RAS mutation detection [10 MM (UPN 61–70)] (Table I).
|UPN||A/S||Ig||Stage||PC (%)||Grade||Pma||Tx||Hb g/dl||WBC ×109/l||PLT ×109/l||Spec||Rass Met||Rass Mut||Rass Exp||Ras Mut||BRAF Mut|
|08||58/F||GK||IIIa||80||M||+||MP + RT||10·1||6·1||261||DX||U||−||ND||−||−|
|12||69/M||GK||IIIa||55||Mono||−||MP + RT||10·9||7·8||187||DX||M||ND||ND||ND||ND|
|16||61/M||AK||IIIa||75||I||−||MP + RT||9·3||6·7||155||DX||U||−||ND||K35 ACA → ATA||−|
|18||48/M||GK||IIIa||80||I||+||VAD + RT||5·9||8·4||136||DX||U||ND||ND||ND||ND|
|19||74/M||GL||IIIb||98||B||−||MP + RT||7||4·6||215||DX||U||ND||ND||ND||ND|
|K61 CAA → CACK72 ATG → ATAH81 GTG → ATG|
|24||47/F||AK||IIIb||47||I||−||VAD/CP/Pusle D||6·8||5·1||123||7 m||M||−||ND||−||−|
|25||62/F||AK||IIIa||70||I||−||MP/CEVAD/IFN + D||6·5||7·9||191||12 m||U||−||ND||−||−|
|26||51/M||GL||IIIa||38||M||+||MP/VAD + RT||12·4||4·2||204||11 m||U||−||ND||−||−|
|30||28/M||GK||IIIb||74||I||−||CEVAD/PBSCT/IFN||7·7||8·7||232||44 m||U||−||ND||K12 GGT → GAT||−|
|31||62/M||K||IIb||80||B||+||MP + RT||11·9||5·3||161||16 m||U||−||ND||−||−|
|33||56/F||K||IIa||85||M||+||VAD/MP + RT||11||7·7||219||5 m||U||−||ND||−||−|
|43||52/F||GK||IIIb||85||I||−||VAD + RT/MP||7·4||4·8||134||DX||ND||ND||W||ND||ND|
|48*||50/M||K||IIIb||80||B||−||Died before Tx||9·2||19·9||78||DX||ND||ND||W + V||ND||ND|
|49||64/M||AK||IIIa||26||I||−||VAD/Thalido + D||9·9||4·7||260||36 m||ND||ND||−||ND||ND|
|50||69/F||GL||Ia||38||I||−||MP||11·6||5·5||223||DX||ND||ND||W + V||ND||ND|
|51||73/F||GK||IIa||30||I||−||MP||8·7||2·5||85||DX||ND||ND||W + V||ND||ND|
|66||69/M||GK||IIIb||65||I||−||Died before Tx||8·1||6·6||88||DX||ND||ND||ND||−||ND|
|70||60/M||AK||IIIb||70||Rods||−||MP/VAD||6·9||7·5||332||DX||ND||ND||ND||N61 CAA → CAC||ND|
Multiple myeloma cell lines.
Four human MM-derived cell lines (NCI-H929, RPMI8226, U266 and ARH-77) purchased from the American Type Culture Collection were used for molecular studies of RASSF1A and mutation identification of BRAF. Cultures of MM cell lines were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco BRL, Life Technologies, Gaithersburg, MD, USA) supplemented with 15% fetal bovine serum (Gibco BRL).
Normal peripheral blood (PB) samples were obtained from 30 volunteer healthy staff and students from the Chinese University of Hong Kong; they served as the normal control group for the methylation studies.
Plasma cell enrichment by immunomagnetic cell sorting using anti-CD138.
Plasma cells were purified from the BM samples using the monoclonal anti-CD138 [magnetic-activated cell sorting (MACS); Miltenyi Biotec, Bergisch, Gladbach, Germany], immunomagnetic beads and columns (MACS) according to the manufacturer's instructions. CD138 is known as Syndecan-1 and is expressed on normal and malignant plasma cells, but not on circulating B cells, T cells and monocytes (Wijdenes et al, 1996). Cells were magnetically labelled with anti-CD138-conjugated microbeads and separated on a column, which was placed in the magnetic field of a separator. The magnetically labelled CD138+ cells were retained in the column while the unlabelled cells were eluted during cell washing. The unlabelled cells were depleted of CD138+ cells. After removal of the column from the magnetic field, the magnetically retained CD138+ cells were then eluted as a positively selected cell fraction. The purity of plasma cells in the CD138+ fraction was assessed morphologically on May–Grünwald–Giemsa stained cytospin slides and further sorting was to be performed if required.
Methylated-specific polymerase chain reaction.
Genomic DNA was extracted from the buffy coat fraction of BM aspirates using the standard phenol/chloroform/isoamyl alcohol method followed by ethanol precipitation. The methylation status of the genes in the samples was determined by methylated-specific polymerase chain reaction (MSP) as previously described (Herman et al, 1996). Briefly, the DNA samples were bisulphite-modified using the CpGenome DNA Modification kit (Intergen Co., Purchase, NY, USA) and the modified DNA was subjected to two polymerase chain reactions (PCRs). The one using M-primers is specific to the methylated sequence and the other, using U-primers, to the unmethylated sequence. The sequences of M- and U-primers have been previously reported (Lo et al, 2001). Hot-start PCR using AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA, USA) was used in all amplification reactions. The enzyme was activated by preheating the reaction mixtures at 95°C for 6 min prior to PCR. This protocol was chosen to minimize non-specific product amplification. The PCR programme was 40 cycles of 1 min at 94°C, 1 min at 55°C and 1 min at 72°C. The PCR products were electrophoresed onto 10% polyacrylamide gels and visualized by ethidium bromide staining under ultraviolet (UV) transillumination. Concordant results from at least two separate experiments were used for determination of methylation status by MSP.
Mutation screening and DNA sequencing.
For RASSF1A mutation analysis, DNA samples were amplified using primer pairs as previously described (Burbee et al, 2001; Lo et al, 2001). PCR products from MM samples mixed with wild-type PCR products were subjected to mutation screening using heteroduplex analysis on a WAVETM denaturing high performance liquid chromatography (DHPLC) instrument (Transgenomic Inc., San Jose, CA, USA) with details as described elsewhere (Lam et al, 2001). For K-, H-, N-RAS and BRAF mutation screening, we performed single strand conformational polymorphism (SSCP) analysis under two SSCP conditions (8% polyacrylamide gel with 5% glycerol at 2·8 W for 16 h at room temperature and 8% polyacrylamide gel at 30 W for 3·5 h at 4°C) (Fong et al, 1997). The primers used for the assays were as described (Sakamoto et al, 2001; Semczuk et al, 2001; Hiyama et al, 2002) and as listed (BRAF): F-5′-CCA CAA AAT GGA TCC AGA CA-3′ and R-5′-TGA AGA CCT CAC AGT AAA AAT AGG TG-3′, which were newly designed for the human BRAF exon 15. The shifted bands were eluted and re-amplified. Samples that were positive by WAVETM DHPLC or SSCP were subjected to DNA sequencing using the ABI PRISM BigDye Primer Cycle Sequencing kit and ABI PRISM 310 Genetic Analyzer (Applied Biosystems) according to the manufacturer's recommendation.
Demethylation treatment using 5′-Aza-2′deoxycytidine
To achieve demethylation, U266 MM cells were seeded at a density of 106 cells/ml medium in 25-cm2 culture flasks and exposed to 1, 3, 5 or 10 μmol/l 2′-deoxy-5′-azacytidine (Sigma Chemical Co., St Louis, MO, USA) for 4 d. The cells were fed with fresh medium supplemented with the drug, daily. At the end of the treatment period, the medium was removed, and total RNA from the treated cells was extracted for reverse transcription (RT)-PCR.
Reverse transcription polymerase chain reaction
Total RNA was extracted from the anti-CD138 immunosorted normal and malignant plasma cells using TRIZOL reagent (Invitrogen Co., Carlsbad, CA, USA) according to the manufacturer's recommendation. The quality of each total RNA sample was controlled by measurement of absorbance and the RNA integrity was determined by RT-PCR on the β-actin gene. One microgram of total cellular RNA was reverse transcribed using the GeneAmp RNA PCR kit (Perkin-Elmer, Norwalk, CT, USA), and 1·5 μl of the resultant cDNA were used in each PCR. The sequences of intron-spanning primers for RASSF1A transcript were previously described (Lo et al, 2001) and they recognized exon 1α and exon 4 of RASSF1A. The PCR products were electrophoresed onto 2% agarose gel with ethidium bromide and visualized under UV transillumination.
This study cohort showed a male to female ratio of 1·7:1 with a median age of 65 years (range 35–89) comprising 35 stage III, 12 stage II and six stage I disease according to the Salmon and Durie classification (Durie & Salmon, 1975; Durie, 1998). The immunoglobulin isotypes included 18 IgGκ, 13 IgGλ, 10 IgAκ, two IgAλ, two IgDλ, 7κ and 1λ. Histologically, 17 MMs were classified as mature, 22 as intermediate and 10 as blastic as described previously (Ng et al, 1997), whereas four others showed atypical morphological features (Table I). The mean percentage of plasma cell infiltration of the BM samples analysed was 64 ± 27 (range 13–95%). The complete blood count showed a mean haemoglobin of 8·5 ± 1·8 g/dl (range 6·1–12·7), white blood cell count of 6·3 ± 2 × 109/l (range 2·5–19·9) and platelet count of 199 ± 95 × 109/l (range 9–397).
Using MSP, we found RASSF1A methylation in nine of 32 (28%) primary MM samples and in one of four (25%) MM-derived cell lines (U266) (Table I), but in none of the 30 normal PB samples tested (Fig 1A). In all of the RASSF1A methylated primary MM cases, unmethylated DNA bands were also observed as a result of the presence of normal BM cellular elements in all these samples (Fig 1A). In contrast, in the RASSF1A methylated U266, a complete methylation pattern was observed when only a methylated DNA band was detected (Fig 1A).
Mutation screening and identification of RASSF1A, BRAF and RAS
No RASSF1A or BRAF mutations were identified in 21 primary MM samples and all MM cell lines except U266, which showed a polymorphism of RASSF1A at codon 133 and also a BRAF mutation of T→A at nt 1796 (substitution of valine by glutamic acid, V599E). Mutations of N-RAS [codon 61 (CAA-CAC)], K-RAS [codon 12 (GGT-GAT), 35 (ACA-ATA), 61 (CAA-CAC), 72 (ATG-ATA)] and H-RAS [codon 81 (GTG-ATG)] were identified in four of 31 (13%) primary MM analysed on unsorted (three of 21) and anti-CD138-sorted plasma cell-enriched (one of 10) BM samples, where concurrent K- and H-RAS mutations were observed in one patient (Table I).
Transcriptional silencing of RASSF1A by promoter methylation and restoration of expression after demethylation
To assess whether RASSF1A hypermethylation would lead to loss of transcription, the RASSF1A transcription was investigated by RT-PCR in all four MM cell lines (RPMI8226, NCI-H929, U266 and ARH77), and again in U266 after demethylation treatment using 5′-aza-2′-deoxycytidine. Before demethylation, loss of transcription was observed only in U266, which showed complete methylation of RASSF1A (Fig 1A,B), but not in the other three MM cell lines (RPMI8226, NCI-929 and ARH77), where no methylation was observed in the MSP evaluation (Fig 1A,B). The RASSF1A expression of U266 could be restored after demethylation treatment in a dose-dependent manner (Fig 1C). It should also be noted that a deletion variant transcript, generated from alternative splicing and coding for a truncated non-functional protein, was also present together with the wild-type transcript in the RASSF1A-expressing MM cell lines (RPMI-8226, NCI-929 and ARH77) (Fig 1B).
RASSF1A mRNA expression in primary MM
RASSF1A transcription was investigated by RT-PCR on plasma cell-enriched (>98% purity) BM samples from 10 primary MM, one PCL and three reactive plasmacytosis. Six of 10 (60%) primary MM samples showed no detectable wild-type RASSF1A transcript (Fig 1B). In two of these six MM samples, only the deletion variant transcript was detected, whereas the other four showed no transcription at all. RASSF1A wild-type transcript was detected in four of 10 MM patients and in two of these four, only the wild-type transcript was expressed whereas it was expressed together with the deletion variant in the other two patients. Both the wild-type and variant RASSF1A transcripts were detected in the PCL patient. In contrast, the wild-type RASSF1A transcript was detected in all the three reactive plasmacytosis samples, although RP3 also expressed the variant transcript (Fig 1B).
RASSF1A is a tumour suppressor gene on chromosome 3p21.3. RASSF1A has been associated with DNA repair protein XPA and plays a role in the RAS signalling pathway by serving as a novel RAS effector protein mediating the apoptotic effects of oncogenic RAS (Dammann et al, 2000; Vos et al, 2000). Promoter methylation of RASSF1A was common in lung, breast and nasopharyngeal carcinomas, while mutations of RASSF1A were rare among all tumours studied (Dammann et al, 2000, 2001; Agathanggelou et al, 2001; Lo et al, 2001). In the current study, we found RASSF1A methylation in nine of 32 (28%) MM patients and one of four (25%) MM-derived cell lines. Unlike other tumours demonstrating RASSF1A methylation, chromosome 3p deletion is uncommon in MM by G-banding and comparative genomic hybridization (CGH) (Avet-Loiseau et al, 1997; Feinman et al, 1997; Cigudosa et al, 1998; Ng et al, 2003). A similar picture of frequent hypermethylation-associated inactivation of p16/p15 (9p21) and death-associated protein kinase (9q34.1) without deletions of chromosome 9p or 9q respectively has also been observed in MM (Avet-Loiseau et al, 1997; Ng et al, 1997, 2001a, 2003; Cigudosa et al, 1998; Wong et al, 1998). By contrast, increased copy numbers of chromosome 9 as trisomies or tetrasomies are frequent in MM, suggesting that promoter methylation may be a unique mechanism of genetic inactivation in this particular setting (Ng et al, 1999a,b). This is further supported by the fact that methylations of multiple genes are common in MM (Leung et al, 2003). Although 3p deletion is uncommon in MM, it is interesting that we recently demonstrated consistent chromosomal rearrangements involving 3p21 in two IgD MM patients analysed by multicolor spectral karyotyping (SKY) (Ng et al, 2001b) and both cases (UPN 11 and 22) showed RASSF1A methylation in the current study.
Multiple myeloma has been reported to show relatively high incidences (>40%) of RAS mutations in most western studies (Neri et al, 1989; Portier et al, 1992; Corradini et al, 1993; Liu et al, 1996; Feinman et al, 1997; Bezieau et al, 2001). Nevertheless, data on RAS mutations in Chinese patients with MM have not been reported. In great contrast to the western data, which showed frequent RAS mutations in MM, only three of 21 of our MM cohort were identified to harbour RAS mutations despite the fact that most of our patients had stage III disease and most of the BM samples analysed contained >30% of MM cells. Epidemiology data indicate that the incidence of MM is much lower in the Chinese as compared with Caucasians (Ferlay et al, 2001). Apart from the disease incidence, we have also observed other differences between MM in Chinese and Caucasians. Distinctively higher incidences of IgD subtype (5–8%, our own unpublished observation), loss of chromosome 22/22q and structural aberrations of chromosome X (detected by combined G-banding and SKY), and chromosome 4q loss (detected by CGH on plasma cell-enriched BM samples) have been found in Chinese MM (Cheng et al, 2002; Ng et al, 2003). These observations suggest that MM in Chinese may have its own unique characteristics. Thus the finding of a low incidence of RAS mutations may represent a difference in tumour genetics with aetiological implications. Moreover, in contrast to other studies from the west, K-RAS (three of 31) was notably more common than N-RAS (one of 31) in our Chinese MM cohort.
Constitutive RAS activation has been observed in the U266 cell line despite the fact that it expresses wild-type RAS. Why this happens has been unknown (Chatterjee et al, 2002). In the current study, whereas no RASSF1A or BRAF mutations were found in primary MM, we first demonstrated the V599E mutation of BRAF in U266. This finding is of special interest as it helps explain the inconsistency of RAS activation in the absence of an underlying mutation in U266. The observation of frequent V599E mutation in colorectal cancers showing no K-RAS mutation lends further support to this explanation (Davies et al, 2002; Rajagopalan et al, 2002). The downstream signalling of RAS involves RAF-MEK-ERK-MAP kinase pathway mediating the cellular responses to growth signals. Thus, mutated BRAF proteins have elevated kinase activity thereby bypassing the need for RAS mutation activation in cancer growth.
To further evaluate whether RASSF1A is really inactivated and the incidence of this in primary MM cells, and whether this is truly a genetic event associated with neoplastic transformation and not a genetic change specific to plasma cell development, the RASSF1A transcription was investigated by RT-PCR on plasma cell-enriched (>98% purity) BM samples from 10 prospective primary MM, one PCL and three reactive plasmacytosis patients, and in all four MM-derived cell lines. Six of 10 (60%) MM samples showed no detectable wild-type RASSF1A transcripts. In two of these six, a deletion variant transcript generated from alternative splicing and coding for a truncated non-functional protein was detected, whereas the other four showed loss of transcription. By contrast, the wild-type RASSF1A transcript was detected in all the three reactive plasmacytosis samples, where RP3 also expressed the variant transcript (Fig 1B). These data indicate that inactivation of RASSF1A is a common and important event associated with MM transformation. The incidence of loss of transcription (four of 10) is closely comparable with the frequency of promoter methylation we observed in the other 32 MM patients. Concordant with other studies, the in vitro data also confirmed transcriptional silencing of RASSF1A associated with promoter methylation in MM. In addition, the alternative abnormal splicing may further contribute to allelic inactivation of RASSF1A. The observation of the presence of both the wild-type and non-functional variant transcripts in the reactive plasma cells (RP3), primary MM (UPN 51–52), the PCL patient (UPN 48) and MM cell lines (three of four) raises the possibility of genetic predisposition in those with haploinsufficiency of RASSF1A function. However, the significance of this needs further investigation.
It may be pivotal that inactivation of RASSF1A in human MM may shift the balance of RAS activities towards a growth promoting effect without the necessity of RAS activating mutations. This hypothesis has been supported by findings of rare RAS mutations (<1%), but frequent RASSF1A inactivation (by hypermethylation in 80%) in small cell lung cancers (Dammann et al, 2001). Taken together, our data indicate that alterations of RAS signalling are critical in MM pathogenesis. In Chinese MM, these alterations frequently involve RASSF1A inactivation (60%) as a result of transcriptional silencing or expression of a non-functional variant transcript.
We thank the clinical haematology colleagues and the haematology laboratory staff for their support of this study, particularly Ms Yonna Leung for her arrangements of sample materials. This work described was partially supported by a grant from the Research Council of the Hong Kong Special Administrative Region, China (project no. CUHK 4068/02M).