Establishment and exploitation of hyperdiploid and non-hyperdiploid human myeloma cell lines

Summary

The establishment of clinically relevant human myeloma cell lines is central for our understanding of myeloma pathogenesis and development of novel therapies for the disease. Unfortunately, most available lines were generated from extramedullary sites, harbored multiple genetic abnormalities and categorized as non-hyperdiploid. In contrast, hyperdiploid myeloma cell lines, which represent more than 50% of patients, are rare. We established procedures for establishment of stroma-dependent myeloma lines by passaging primary myeloma cells, in severe combined immunodeficient-human (SCID-hu) or SCID-rab mice followed by maintenance in co-culture with stromal cells. We described the establishment and characterization of two hyperdiploid (LD and CF) and two non-hyperdiploid (JB and BN) cell lines. Using our animal models, we also established bortezomib-sensitive and -resistant BN lines. These cell lines were cellularly, phenotypically and molecularly characterized using flow cytometry immunophenotyping, DNA content, G-band and multicolor spectral karyotyping (SKY) and global gene expression profiling. All four cell lines were infected with lentiviral-expressing luciferase for detection of tumour cells at high sensitivity level and for monitoring myeloma growth in co-cultures and in vivo by live animal imaging. These myeloma cell lines and the procedures used for their establishment provide essential tools for studying myeloma biology and therapy.

Recently, global gene expression profiling and the sensitive molecular techniques used to detect genetic abnormalities revealed that myeloma is a heterogeneous disease associated with multiple chromosomal abnormities (Hallek et al, 1998; Fonseca et al, 2004; Bergsagel et al, 2005; Bergsagel & Kuehl, 2005) and highly deviated molecular profiles (Zhan et al, 2002, 2006). Over the past two decades, many human myeloma cell lines have been established that have significantly contributed to our understanding of various aspects of myeloma pathogenesis, including the identification of common chromosomal translocations and genetic aberrations, mechanisms of drug resistance, and major signalling pathways associated with responses to specific growth factors (Bataille et al, 1989; Mazars et al, 1992; Abbaszadegan et al, 1996; Billadeau et al, 1997; Akiyama et al, 2002; Croonquist et al, 2003; Hazlehurst et al, 2003; Hideshima et al, 2003; Mitsiades et al, 2003; Tassone et al, 2005). Nevertheless, most of these cell lines, including stroma- and cytokine-dependent myeloma lines (Chesi et al, 1997; Bergsagel & Kuehl, 2001; De Vos et al, 2001; Verdelli et al, 2005), were derived from extramedullary sites of patients at the end stage of myeloma, are highly proliferative, and grow aggressively and independently of the bone marrow microenvironment in vitro and in animal models; their growth patterns do not reflect those seen clinically, where low-proliferating tumour cells restrictively reside in the haematopoietic marrow and are protected from spontaneous and drug-induced apoptosis. Furthermore, the majority of these lines are cytogenetically classified as non-hyperdiploid myeloma, which represents patients with poor prognosis (Smadja et al, 1998, 2001; bes-Marun et al, 2003; Fonseca et al, 2003).

While previously established cell lines have been useful, at this stage of myeloma research there is a need for clinically relevant hyperdiploid myeloma cell lines. Such lines will help to understand the key aspects of myeloma biology, including the myeloma ‘stem cell’ and its proliferative compartment, transformation from benign to active disease stages, and development of drug resistance; these lines will also facilitate validation of novel targeted therapies for myeloma and its related bone disease, based on interactions between myeloma cells and their microenvironment.

To date, the severe combined immunodeficient-human (SCID-hu) and SCID-rab murine models are the only systems in which primary myeloma cells taken from patients with medullary myeloma can potentially undergo sequential passages in vivo and ex vivo and become low-proliferating, bone marrow-dependent myeloma lines. Reproducible engraftment of primary human myeloma cells in SCID-hu (Yaccoby et al, 1998) and SCID-rab (Yata & Yaccoby, 2004) mice has fundamentally contributed to our understanding of crucial aspects of the clonogenic potential of mature myeloma plasma cells (Yaccoby & Epstein, 1999; Yata & Yaccoby, 2004) and the role of bone marrow microenvironment in disease progression (Yaccoby et al, 2002a,b, 2006, 2007). In these systems, myeloma cells from patients at various disease stages are successfully engrafted, grow slowly but exclusively in the implanted bone, and produce typical disease manifestations of myeloma. These models also enabled us to generate clinically relevant myeloma cell lines from primary patient samples of intramedullary disease. Here, we describe the establishment and characterization of four myeloma cell lines.

Materials and methods

Myeloma cell analyses

Multiple myeloma (MM) cells were obtained from heparinized bone marrow (BM) aspirates from four patients with active myeloma during scheduled clinic visits. Signed Institutional Review Board – approved informed consent forms are kept on record. Pertinent patient information is provided in Table I. The BM samples were separated by density centrifugation using Ficoll-Paque (specific gravity, 1.077 g/ml; Amersham Biosciences Corp., Piscataway, NJ, USA). Myeloma cells recovered from tumour-engrafted mice were analysed for the presence of Epstein–Barr virus (EBV) sequences and transcripts by polymerase chain reaction (PCR) and reverse transcription (RT)-PCR (Miyashita et al, 1995). All tested samples were negative. Flow cytometry immunophenotypic characterization of myeloma cells was performed using CD45, CD38, CD138 and CD56 monoclonal antibodies (Becton Dickenson, San Jose, CA, USA). Cytofluorimetric analysis of DNA content, conventional karyotyping and multicolor spectral karyotyping (SKY) were performed as previously described (Latreille et al, 1980; Sawyer et al, 2001). Gene expression profiling on CD138-selected myeloma cells was performed with the Affymetrix U133Plus2.0 microarray platform (Santa Clara, CA, USA) using methods previously described (Zhan et al, 2002, 2006).

Table I.

Characteristics of cell lines and patients.

	Cell line
Patient characteristics	LD	CF	JB	BN
Disease status	MM	MM	MM	MM
Stage	IIIA	IIIA	IIIB	IIIA
Isotype	IgAλ	IgGκ	Freeλ	IgGλ
DNA index*	1.35	1.16	1.11	1.04
Cell source	BM	BM	BM	BM
Cytogenetics	Hyperdiploid	Hyperdiploid	Pseudodiploid	Pseudodiploid
Microarray classification†	Proliferation	Proliferation	MMSET	MMSET

MM, multiple myeloma; BM, bone marrow.

^*DNA content detected according to Latreille et al, 1980.

^†According to Zhan et al, 2006.

Myelomatous SCID-hu and SCID-rab mice

Both SCID-hu and SCID-rab mice were constructed as previously described (Yaccoby et al, 1998; Yata & Yaccoby, 2004). Myeloma cells (0.1–5 × 10⁶ cells) in 50–100 μl of phosphate-buffered saline (PBS) were injected directly into the implanted bone in SCID-hu or SCID-rab mice. Mice were periodically bled from the tail vein and changes in levels of circulating human immunoglobulin (hIg) of the M-protein isotype were used as an indicator of MM growth. For testing the effect of bortezomib (Millennium Pharmaceuticals, Inc., Boston, MA, USA), myelomatous hosts were treated subcutaneously with saline or bortezomib (0.5 mg/kg body weight twice a week) (LeBlanc et al, 2002) for 5 weeks.

Determination of human hIg levels

Levels of human κ and λ light chains were determined by enzyme-linked immunosorbent assay (ELISA) as described (Yaccoby et al, 1998; Yaccoby & Epstein, 1999). At the end of each experiment, all samples were analysed in the same assay in order to preclude inter-assay variability.

Myeloma-stromal cells co-cultures

Mesenchymal stem cells (hereinafter, stromal cells) were prepared as previously described (Yaccoby, 2005; Ge et al, 2006). Briefly, bone marrow mononucleated cells (2 × 10⁶ cells/ml) from human fetal bone fragments (Advanced Bioscience Resources, Alameda, CA, USA) were cultured in low-glucose Dulbecco's modified Eagle's medium (DMEM-LG) supplemented with 10% fetal bovine serum (FBS) and antibiotics. One-half of the medium was replaced every 4–6 d, and adherent cells were allowed to reach 80% confluency before they were subcultured with trypsin–EDTA. The adherent stromal cells expressed CD166, but not CD45 or CD34, and were capable of differentiating into mesenchymal lineage, such as osteoblasts and adipocytes (Yaccoby et al, 2006). Myeloma cell lines were co-cultured with stromal cells in RPMI-1640 media supplemented with 10% FBS and antibiotics and one-half of the medium was replaced every 5–7 d. For passaging myeloma cells to new cultures, non-adherent myeloma cells were collected from co-cultures, spun to remove excessive media and then seeded in flasks-containing stromal cells.

Lentiviral vector, myeloma cell infection and live animal imaging

We constructed a lentiviral vector (pLenti6/V5/EGFPLuc) expressing both enhanced green fluorescent protein (EGFP) and luciferase. DNA sequences of EGFPLuc encodes a fusion of EGFP and luciferase from the firefly Photinus pyralis was cloned by PCR from the pEGFPLuc vector (Clontech, Mountain View, CA, USA) and inserted into pLenti6/V5-D-TOPO vector using pLenti/V5 Directional TOPO Cloning Kit (Invitrogen, Carlsbad, CA, USA). Lentiviral particles were generated by cotransfecting the expression vector pLenti6/V5/EGFPLuc and ViraPower Packaging Mix into 293FT cells according to the Invitrogen's ViraPower Lentiviral Expression Systems protocol. Myeloma cells were plated in 24-well plate (100 000 cells/well) and transduced by adding the lentiviral particles to cell suspensions at a multiplicity of infection (MOI) of 2 for 24–48 h. The medium containing lentivirus was removed, replaced with fresh RPMI-1640 medium supplemented with 10% FBS and antibiotics and infected cells were allowed to grow in co-culture with stromal cells (Yaccoby et al, 2006). EGFP/luciferase-expressing myeloma cells were then sorted by fluorescent-activated cell sorting (FACS) based on EGFP expression and further expanded in co-culture with stromal cells prior to their exploitation in vivo and in vitro. For live imaging, mice were anaesthetized with ketamine plus xylazine and injected intraperitoneally with d-luciferin firefly (150 mg/kg; Xenogen, Corp. Alameda, CA, USA). Luciferase activity was detected using IVIS 200 imaging system (Xenogen).

In vitro luciferase assay

Stromal cells were cultured in 96-well microplates with white clear bottom (Thermo Labsystems, Waltham, MA, USA). Indicated numbers of myeloma cells were co-cultured with MSCs in 100 μl of RPMI 1640 media supplemented with 10% FBS and antibiotics for indicated periods of time. In some experiments co-cultures were treated with various doses of bortezomib, dexamethasone and melphalan for 72 h. At the end of the incubation period, d-luciferin K+ salt was added to the cultures (150 μg/ml) and luciferase activity was determined after 5 min using LumiCount (PerkinElmer Life and Analytical Sciences, Shelton, CT, USA).

Results

Establishment of myeloma cells lines

The growth of primary myeloma cells in SCID-hu and SCID-rab mice have been intensively characterized (Yaccoby et al, 1998; Yaccoby & Epstein, 1999; Yaccoby et al, 2002a,b; Yata & Yaccoby, 2004; Yaccoby et al, 2007). Using more than 300 samples, we have found that myeloma cells from >80% of patients successfully engrafted and grew similarly in both models. In the majority of cases, myeloma growth was restricted to the implanted bones; however, in approximately 5–10% of cases, myeloma growth was also apparent on the outer surface of the implanted bone. This growth pattern is expected when myeloma cells are taken from patients with extramedullary disease (e.g. plasma cell leukaemia, plasmacytoma), but we also observed this growth pattern from cells of some tumours obtained from bone marrow of patients with medullary myeloma. Those cases provided the materials for establishing four clinically relevant myeloma cell lines. A large amount of these tumour cells were recovered from the implanted bone and passaged to freshly prepared SCID-hu or SCID-rab mice.

Cellular and molecular characterization of myeloma cell lines

Following passages in vivo our cell lines retained their dependence on bone marrow in vitro; they did not grow independently in culture but grew and maintained in co-culture with stromal cells as stroma-dependent myeloma lines. These cells also retained the slow growth characteristics of primary myeloma; in contrast to typical stroma independent lines whose doubling time is approximately 1–2 d, the doubling time for lines BN, JB and CF was approximately 4–6 d, while that of LD was approximately 10 d. Cell–cell contact with stromal cells was required for their growth, and single factors [e.g. interleukin (IL)-6 or insulin-like growth factor-1] were insufficient to stimulate their long-term growth in vitro (data not shown). When cells from our lines were injected back into implanted bones of newly constructed SCID-hu or SCID-rab mice, they grew and produced typical myeloma manifestations (e.g. bone disease) (Yaccoby et al, 2002b, 2007); however, these cells did not grow after intraperitoneal (i.p.), subcutaneous (s.c.), or intravenous (i.v.) injection into SCID mice.

All the four established cell lines were EBV-negative and produced the same M isotype as the original patients' cells; these cell lines expressed high levels of CD38 and CD138 and varying levels of CD45 and CD56 (Fig 1). Flow cytometric analysis of DNA content (Latreille et al, 1980) revealed that LD and CF lines were hyperdiploid and JB and BN were pseudodiploid (Table I).

Fig 1.

Immunophenotypic analysis of four myeloma cell lines. LD, CF, JB and BN cells were stained with CD45/CD38 and CD138/CD45 and analysed by flow cytometry. Note the high expression of CD38 and CD138, and various expression level of CD45 and CD56 by these cells.

Conventional G-band karyotype designations of the original patient's MM cells and the corresponding cell line designations indicated that the cell lines originated from the patient's tumour clone (Table II). Combined G-banding and SKY analysis (Sawyer et al, 2001) were used to characterize the multiple complex cytogenetic abnormalities in each of these lines (Fig 2). The two hyperdiploid, LD and CF lines were characterized by multiple trisomies; LD cells showed trisomies for chromosomes 5, 6, 8, 9 and 20, and structural aberrations involving extra copies of 1q translocated to 8(p23)×2. The CF cell line showed at least partial trisomies for chromosomes 1, 3, 5, 7, 9, 11, 18 and 19, and structural aberrations including dup(1)(q12∼23)×2. The JB and BN cell lines were both hypodiploid and were monosomic for chromosome 13. The JB cells showed structural aberrations including extra segments of 1q translocated to 2(p23) and 15(pter). The BN cell line showed two whole-arm translocations involving 1q, one involving 1q and 19p, and the other involving 1q and 17q.

Table II.

G-band karyotype designations of patient's tumour specimens and corresponding G-band cell line designations.

Patient MM cells: 49∼51,X,−Y,+6,+8,der(8)t(1;8)(q11∼12;p23) ×2,+9,dic(12;21)(q24.3;q22),+18,der(19) t(11;19)(q13;q13.4),+?add(21)(p11.1),+mar[cp20]

Cell Line LD: 51∼52,XY,+5,+6,+8,der(8)t(1;8)(q11∼12;p23) ×2,+9,+18,der(19)t(11;19)(q13;q13.4), +20,der(21)t(19;21)(p11;p11.1)[cp20]

Patient MM cells: 58∼61,XXY,−1,dup(1)(q21q31) ×2,−2,del(2)(p?21),der(3)(1;3)(q?21;p?13),−4,+5,del(5)(q31) ×2,−6,+add(7)(p15),−8,del(9)(q12q22) ×2,+add(9)(p11.2),−10,+11,−12,−13,−14,−15,−16,−16,−17,+18,−21,i(21)(q10) ×2,−22[cp20]

Cell line CF:56∼57,XY,dup(1)(q21q31),+der(1)dup(1)(q21q31)t(1;18)(q32;q?11.2), t(2;5)(p?13;q31),+der(3)del(3)(p21)del(3)(q11.2),+der(5)t(2;5)(p?13;q13), +7,der(9)t(9;20)(p13;q13.1),+del(9)(q12q22) ×2,+11,der(16)t(16;21)(p11.2;q?22),der(16)t(16;21)(q12.1;q?22),+der(16)t(7;16)(q11.2;q12.1),+18,+?18,+19,i(21)(q10), der(20)t(5;20)(p11.1;p13)[cp20]

Patient MM cells: 42∼44,X,der(X;1)(q10;q10),?t(1;2)(q21;p23),der(1;15)(q10;q10),der(2)t(1;2) (p?22;q?37),der(4)t(1;4)(q10;q10)del(1)(q32)add(4)(q?32),?add(5)(p?15.3), add(7)(p22),add(7)(q36),−13,−14,del(16)(q22),+del(17)(p12),add(18)(q?22), ?del(19)(p12)[cp7]

Cell line JB: 42∼44,X,der(X;4)(p10;q10)t(4;19)(q35;?p13.3), +der(X;1)(q10;q10)t(1;19)(q31;?p13.3),der(1)t(1;18)(q21;q21), der(1;15)(q10;q10), der(2)t(1;2)(q25;p21),der(2)t(1;2)(p?32;q37), der(5)t(4;5)(q34;p15.3),+der(7)t(1;7)(q44q25;p22),der(7)t(1;7)(q25;q32) ×2, −13,−14,der(18)t(2;18)(q3?3;q21),del(19)(p13.1)[cp14]/85∼88,idem ×2[cp6]

Patient MM cells: 44∼47,XY,der(1;17)(q10;q10),+der(1;19)(q10;p10),del(3)(p21),del(4)(p?12p?15.2), add(5)(q?14),i(11)(q10),−13,der(14)t(?8;14)(q22;q24),add(17)(q25),add(21)(p11.1) [cp10]/44∼47,idem,-del(3),+3,?t(4;8)(q?22;p?21)[cp7]

Cell line BN: 43∼44,XY,der(1;17)(q10;q10),+der(1;19)(q10;p10),der(3;20)(q10;q10), del(4)(p?12p?15.2),t(5;12)(q15;q15),der(9)t(3;9)(p21;q34), der(10)t(10;11) (q?25;?p15),der(11;21)(q10;q10), −13,+16,der(17)t(1;17)(p32;q25), +der(21)t(X;21)(q24;p11.1)[cp10]

Fig 2.

Representative G-band (top panel) and spectral karyotyping (SKY; bottom panel) karyotypes demonstrating different translocations and chromosome aneuploidies in LD (A), CF (B), JB (C) and BN (D) myeloma cell lines. The most common structural aberrations in the cell lines were duplications and translocations of 1q. Note that each cell line harbors basically the same karyotype aberrations found in the original tumour specimen of the patients.

To further characterize our lines at the molecular level, global gene expression profiling was performed and listed genes known to be associated with myeloma pathogenesis (Table III). Expression levels of these genes in each of the lines were compared with gene expression levels of newly diagnosed myeloma patients included in the TTII clinical trial at our institute (Zhan et al, 2006). Global gene expression profiling has been successfully used to predict major gene translocations involving the immunoglobulin heavy chain (IgH) locus and chromosomal deletion based on spiked genes (Zhan et al, 2002; Bergsagel et al, 2005; Zhan et al, 2006; Chng et al, 2007). According to the spiked expression of WHSC1 (also known as MMSET), FGFR3 and CCND2 in JB and BN cells belong to the MMSET subgroup (Zhan et al, 2006) which harbors the t(4;14) translocation (Zhan et al, 2002; Bergsagel et al, 2005). In contrast, LD and CF cells, which appear to have no IgH translocations expressed high level of DKK1, a Wnt signalling inhibitor that has been closely associated with hyperdiploid myeloma (Table III) (Tian et al, 2003; Zhan et al, 2006).

Table III.

Microarray signal of selected genes associated with myeloma pathogenesis in four myeloma cell lines.

	TTII patients*			Cell lines
Gene	Min	Mean	Max	LD	CF	JB	BN
PTPRC (CD45)	5	266	5028	3099	3263	49	620
NCAM1 (CD56)	147	3551	12 742	1619	181	3293	1679
CD28	6	842	9412	132	2390	644	1620
IL6	12	406	3894	720	114	354	2311
IL6R	59	3609	19 020	1150	570	2476	5912
IGF1	12	406	3894	993	888	398	2445
IGF1R	222	2669	9334	5	18	443	989
IL10RA	42	885	10 536	119	265	953	347
IL10RB	623	1627	4117	563	872	1549	1191
VEGF	168	2545	8798	1310	858	5587	5651
HGF	5	1913	22 055	5	199	528	348
PT53	10	1460	5140	2417	397	2227	817
MYC	37	6351	33 772	11 347	7849	6	12 180
WHSC1 (MMSET)	18	1298	23 600	1280	1706	6471	5230
FGFR3	196	736	3234	703	699	4110	576
MAF	7	494	8746	85	82	304	1155
MAFB	4	811	45 816	233	74	228	100
CCND1	44	3677	26 493	114	247	132	222
CCND2	28	5916	43 642	5553	13	13 861	14 097
CCND3	177	2477	35 631	719	1295	1998	1247
RB1	5	3112	12 237	2446	131	1600	2359
CXCR4	1112	18 257	77 508	2140	7563	154	3430
CCR1	13	581	5666	3063	1323	1423	3169
CCR2	145	11 244	50 891	6641	12 345	1702	2120
DKK1	171	4638	30 884	10 835	7089	1258	540
SFRP2	9	176	6619	46	80	63	99

^*Minimum (Min), mean and maximum (Max) signal intensity in newly diagnosed patients with myeloma enroled in TTII clinical trial in our institute

Exploitation of myeloma cells for studying therapeutic approaches

To improve the use of our cell lines for studying myeloma biology and therapy in vivo and in vitro, we infected these cells with lentiviral-expressing luciferase. As shown in Fig 3, the growth of myeloma cells in SCID-hu mice, as determined by live animal imaging highly, correlated with hIg levels in mice sera. Live animal imaging also confirmed our previous observation demonstrating that myeloma cell growth is restricted to the implanted bone. In additional studies, hosts engrafted with CF and BN cells and, upon establishment of myeloma growth, were treated with bortezomib for 5 weeks. As demonstrated in Fig 3D, BN but not CF cells responded to bortezomib, and BN-bearing mice showed no detection of luciferase at the given sensitivity level after treatment.

Fig 3.

Bioluminescent imaging of myeloma growth in vivo. Our myeloma cell lines were infected with luciferase-containing lentivirus as a marker for tracing myeloma growth in live SCID-hu and SCID-rab mice. Luciferase activity was detected using IVIS 200 imaging system. (A) BN myeloma cells were injected into mice and immediately analysed for bioluminescence intensity (photons/s/cm²/steradian). Note detection of 5000 or more BN cells by this system. (B, C) Bioluminescence intensity of BN cells (100 000 cells) was increased with time (B and C) and was highly correlated increased circulating hIg (B). (D) Treatment of BN- but not CF-bearing hosts with bortezomib (BOR, subcutaneously, 0.5 mg/kg/body weight, twice a week) for 5 weeks resulted in eradication of myeloma growth.

To further explore the clinical application of these cell lines, we established bortezomib-resistant and -sensitive BN cells by exposing SCID-rab mice bearing BN cells to bortezomib (n = 3). A representative experiment of one host is demonstrated in Fig 4A. Following 4 weeks of bortezomib treatment, mice achieved complete response, and treatment was stopped. Mice remained in remission for 14 weeks but then relapsed, and treatment was resumed (Fig 4A). At this point, mice did not respond to treatment, and myeloma cells recovered from one of these mice were also resistant to bortezomib following passages into newly constructed SCID-rab mice (Fig 4B) and in vitro (Fig 4C). We also compared gene expression profiles of bortezomib-resistant and -sensitive BN cells (Table IV). While some identified genes are directly associated with drug resistance (e.g. ABCB1, BCL2), other genes may be used as prognosis markers (e.g. CD9) associated with response to bortezomib.

Fig 4.

Establishment of a bortezomib-resistant myeloma cell line in vivo. SCID-hu mice engrafted with BN cells were treated with bortezomib (BOR) or left untreated. Note that hosts initially responded to BOR; however, when myeloma recurred, BN cells were resistant to the drug (A). Bortezomib-resistant BN cells (BN-R) recovered from mice (and were not continuously exposed to the drug), remained resistant to bortezomib following passage into newly constructed SCID-rab mice (B) and in co-culture with stromal cells (C), while bortezomib-sensitive BN cells (BN-S) responded to BOR in a dose–response manner (C).

Table IV.

Partial list of genes differentially expressed between bortezomib-resistant and -sensitive BN cells.

Overexpressed in bortezomib-resistant cells		Underexpressed in bortezomib-resistant cells
Gene names	Fold	Gene names	Fold
ABCB1	34	ABCA13	0.071
ABCB5	9	ABCC1	0.125
ABCB9	4.5	ABCD3	0.25
ABCB4	4.2	IL1RAP	0.2
KCNN3	16	CD9	0.002
ATP2A3	86	CD74	0.2
BCL2L1	8	CD58	0.25
BCLAF1	2.2	NUMA1	0.43
BAX	1.5	SULF2	0.02
BCL6B	4	MGST1	0.03
BAG1	1.4	WNT11	0.19
BCL2L11	3	SNX9	0.06
BCL3	2	AP4E1	0.06
BCL6	4	EPHA3	0.007
MAFB	15	CDH3	0.003
IL6ST	1.5	USP9Y	0.007
CXCR4	2	COL18A1	0.015
IL21R	2.6	SERPING1	0.015
SOCS3	5.7	SV2B	0.016
IL32	157	CXCL13	0.016
IL10RA	13	TP73L	0.016
HGF	13	CAMK1D	0.023
SOCS1	1.3	MAGEB1	0.023
PTPRC	50	CAMK1D	0.023
CD52	50	RHOU	0.12
CD180	60	TOM1L1	0.21
CD59	19	TLR	0.22
CD8B1	11	CCL3	0.19
CD79A	4	KCTD14	0.16
CD28	3	CARG15	0.11
RRBP1	5	TNFRSF10B	0.12
RAB15	5	GIMAP2	0.13
RAB4B	4	GPR157	0.13
HDAC6	1.5	CGREF1	0.14
ITGA6	3.7	RAI14	0.17
WNT5A	12	IFI44L	0.04
CSF2RB	6	ST14	0.01
SPP1	33	GPR56	0.03
MT1M	24	FCRLM1	0.03
CX3CR1	20	E2F8	0.03
MAFB	15	CREB5	0.04
FOS	15	IFITM3	0.04
HGF	13	MAGEB3	0.05
SIT1	10	TSPAN5	0.07
HCK	9	SPATA18	0.07
SLA	10	MAL2	0.08
HIST1H4J	8	TUSC3	0.08
TP53INP1	6	FABP6	0.08
FRZB	120	MAGEB2	0.09
CCL2	79	GJB2	0.1
PTPRCAP	50	TP73L	0.1
SPP1	33
GPR114	24
PTGDR	10
TNFRSF11B	5
TNFSF8	8
LEPR	6

Using our luciferase-expressing myeloma cell lines we established a growth assay for myeloma cells in co-culture with stromal cells. Myeloma cells often firmly adhere to stroma cells, preventing recovery of cells. We initially correlated between the number of myeloma cells seeded on stromal cells and level of luciferase activity by these cells. There was high correlation (r = 0.997) between myeloma cell number and luciferase activity in the co-cultures (Fig 5). This assay was exploited to study the effect of bortezomib, melphalan and dexamethasone on growth of BN myeloma cells (20 000 cells/well) in co-culture with the supporting stromal cells. Whereas bortezomib and melphalan inhibited BN cell growth in a dose-dependent manner (50% inhibitory concentration (IC₅₀) c. 5 nmol/l and c. 0.9 μmol/l for bortezomib and melphalan respectively), BN cells responded to dexamethasone only in doses greater than 200 μmol/l (Fig 5).

Fig 5.

Establishment of in vitro growth assay of myeloma cells in co-culture with stromal cells. (A) Luciferase (Luc) activity was highly correlated with number of tumour cells in co-cultures (r = 0.997). (B) A study demonstrating the effect of various doses of bortezomib, dexamethasone and melphalan (72 h, triplicates) on growth of BN cells in co-culture with stromal cells. Results are expressed as mean ± SEM.

Discussion

Myeloma has been classified into two broad genetic, hyperdiploid and non-hyperdiploid subtypes (Fonseca et al, 2004; Bergsagel & Kuehl, 2005). Hyperdiploid myeloma cells are characterized by an increased number of chromosomes (48–74) often reflecting multiple trisomies, lower prevalence of primary translocations involving the IgH locus at 14q32 and high dependency on the BM microenvironment. Patients with non-hyperdiploid myeloma are associated with the presence of primary IgH translocations such as t(4;14), t(11;14) and t(14;16) (Fonseca et al, 2003; Bergsagel et al, 2005). Progress in understanding the biology of hyperdiploid myeloma has been hampered mainly due to lack of appropriate available cell lines (Fonseca et al, 2004). This report describes the establishment of LD, CF, JB and BN human myeloma cell lines, two of which (LD and CF) were categorized as hyperdiploid myeloma. G-banding karyotyping analysis showed that the cell lines originated from the corresponding patient's clone. Although hyperdiploid myeloma patients are generally thought to have better prognosis than those with non-hyperdiploid myeloma, a subset of hyperdiploid myeloma with poor outcome has been recently identified by global gene expression profiling (Zhan et al, 2006; Chng et al, 2007). The molecular classification of the two established hyperdiploid lines in the proliferation group indicated that they originated from high-risk myeloma (Zhan et al, 2006). Using our animal models, we also developed in vivo procedures to establish drug-resistant variants and demonstrated the development of bortezomib-sensitive and -resistant BN cells.

Our myeloma cell lines retained slow growth as stroma-dependent cells in vitro and disseminated restrictively in the implanted human or rabbit bones in SCID-hu and SCID-rab mice respectively. These cells thus better reflect the growth of primary myeloma in the bone marrow. The two hyperdiploid myeloma cell lines, particularly LD cells, have lower growth rate in vitro and in vivo than the two non-hyperdiploid lines, and thus are favoured for studying the interaction of myeloma cells with the bone marrow microenvironment.

Using lentiviral-expressing luciferase we developed a highly sensitive assay to monitor tumour growth in co-culture of our myeloma cell lines with stromal cells and in SCID-hu/SCID-rab mice. Fewer than 500 myeloma cells could reliably be detected in vitro and 5000 cells in the implanted bone of SCID-hu/SCID-rab mice. The luciferase-expressing myeloma cells have been successfully exploited to test the effect of various drugs on myeloma cell growth in co-culture and in vivo.

Taken together, we have established two hyperdiploid and two non-hyperdiploid human stroma-depended myeloma lines. Their growth rate and patterns reflect primary myeloma and thus are biologically and clinically relevant. These cells have been exploited to monitor tumour growth using luciferase-expressing myeloma cells in ongoing co-cultures and in live animals, to study myeloma therapy and development of drug-resistance. These myeloma cell lines and the procedures used for their establishment provide essential tools for studying myeloma biology and therapy.