Mesenchymal Stem Cells from Multiple Myeloma Patients Display Distinct Genomic Proﬁle As Compared with Those from Normal Donors

Mesenchymal stem cells from multiple myeloma patients display distinct genomic proﬁle as compared with those from normal donors

M Garayoa1,5,6, JL Garcia1,2,6, C Santamaria3, A Garcia-Gomez1, JF Blanco4, A Pandiella1, JM Herna´ndez3, FM Sanchez-Guijo3,5, M-C del Can˜izo3,5, NC Gutie´rrez3, and JF San Miguel1,3,5

1Centro de Investigacioń del Cańcer, Instituto de Biologıá Molecular y Celular del Cańcer, Universidad de Salamanca-CSIC, Salamanca, Spain; 2Unidad de Investigacioń, Instituto de Estudio de Ciencias de la Salud de Castilla y Leoń (IECSCYL) – Hospital Universitario de Salamanca. Salamanca, Spain; 3Servicio de Hematologıá. Hospital Universitario de Salamanca. Salamanca, Spain; 4Servicio de Traumatologıá. Hospital Universitario de Salamanca. Salamanca, Spain and 5Centro en Red de Medicina Regenerativa y Terapia Celular de Castilla y Leoń, Salamanca, Spain

It is an open question whether in multiple myeloma (MM) bone directly by interactions of myeloma cells with BM stromal cells marrow stromal cells contain genomic alterations, which may and extracellular matrix proteins or indirectly by secretion of contribute to the pathogenesis of the disease. We conducted an array-based comparative genomic hybridization (array-CGH) soluble cytokine and growth factors by myeloma cells and/or analysis to compare the extent of unbalanced genomic altera- stromal cells. These interactions and growth factor circuits tions in mesenchymal stem cells from 21 myeloma patients ultimately lead to the activation of pleiotrophic signalling (MM-MSCs) and 12 normal donors (ND-MSCs) after in vitro cascades, which promote proliferation, cell survival, anti- culture expansion. Whereas ND-MSCs were devoid of genomic apoptotic signalling, drug resistance and migration of MM imbalances, several non-recurrent chromosomal gains and cells.4,5,7 Furthermore, crosstalk of myeloma cells with other losses (41 Mb size) were detected in MM-MSCs. Using real- time reverse transcription PCR, we found correlative deregu- cells in the BM milieu, such as osteoblasts and osteoclasts, lated expression for five genes encoded in regions for which results in inhibition of the former and activation of the latter, genomic imbalances were detected using array-CGH. In addi- which produces an unbalanced bone remodelling responsible tion, only MM-MSCs showed a specific pattern of ‘hot-spot’ for the development of bone lesions in MM patients.6,8,9 regions with discrete (o1 Mb) genomic alterations, some of Bone marrow mesenchymal stem cells (MSCs) are an essential which were confirmed using fluorescence in situ hybridization cell type in the formation and function of the BM micro- (FISH). Within MM-MSC samples, unsupervised cluster analysis did not correlate with particular clinicobiological features of MM environment, being the progenitor cells of osteoblasts and patients. We also explored whether cytogenetic abnormalities the haemopoietic-supporting stroma components of the 10,11 present in myelomatous plasma cells (PCs) were shared by marrow. In fact, differences between MSCs derived from matching MSCs from the same patients using FISH. All MM- MM patients (MM-MSCs) and those from healthy donors MSCs were cytogenetically normal for the tested genomic (ND-MSCs) have been reported.12–16 When compared with alterations. Therefore we cannot support a common progenitor their normal counterparts, MM-MSCs differ in spontaneous and for myeloma PCs and MSCs. Leukemia (2009) 23, 1515–1527; doi:10.1038/leu.2009.65; myeloma cell-induced production of cytokines, exhibit a published online 9 April 2009 decreased proliferative capacity and present reduced efficiency Keywords: array-CGH; mesenchymal stem cells; multiple to inhibit T-cell proliferation, and the osteoblasts derived from myeloma; genetic alteration MM-MSCs show a diminished matrix mineralization potential when compared with their normal counterparts.12–16 Moreover, Corre et al.13 have observed a distinctive gene expression profile for MM-MSCs and ND-MSCs upon using microarray analysis, with differential expression of genes coding for growth and Introduction angiogenic factors, as well as for factors related to bone differentiation.13 All these differences were detected after MSCs Multiple myeloma (MM) is a B-cell neoplasia characterized by isolation and expansion in culture. Although it has been accumulation of clonal plasma cells (PC) in the bone marrow, suggested that these differences between normal and myeloma- the presence of monoclonal immunoglobulin in blood and/or tous MSCs could be attributed to the presence of genomic urine, and the existence of bone lesions. The genetic basis of the alterations in MM-MSCs,12–14 this is still an issue to be clarified. disease includes recurrent and complex genetic abnormalities in On the other hand, the clinical observation that bone lesions myelomatous cells, which besides contributing to the pathogen- from MM patients do not heal even after response to therapy9,17 esis of MM also influences the heterogeneous clinical course of also seems to support the idea of a permanent defect in the 1–3 the disease. In addition, the involvement of the bone marrow capacity of MM-MSCs to generate fully functional bone- (BM) microenvironment in the pathophysiology of the disease is generating osteoblasts. This last issue could also be explained 4–7 nowadays well-accepted. This contribution is mediated either by genomic alterations present in MM-MSCs, which would remain in the absence of myeloma cells. Correspondence: Dr M Garayoa, Centro de Investigacioń del Cańcer - An additional controversial issue in the field of MSCs is the CSIC, Campus Miguel de Unamuno, Avda. Coimbra s/n. 37007, idea of a common haematopoietic and mesenchymal progeni- Salamanca, Spain. tor, with occasional reports favouring this possibility.18–20 E-mail: [email protected] 6These two authors contributed equally to this work. Theoretically, if MSCs from myeloma patients would encompass Received 30 September 2008; revised 25 February 2009; accepted 27 cytogenetic markers present in myelomatous PCs, the genomic February 2009; published online 9 April 2009 events leading to those genomic aberrations should have Genomic profile of multiple myeloma MSCs M Garayoa et al 1516 occurred earlier to a mesenchymal or haematopoietic commit- Healthcare, Uppsala, Sweden) and cultured in low-glucose ment, thus supporting the idea of a common precursor for both Dubecco’s modified Eagle’s medium supplemented with 10% types of cells. heat-inactivated fetal bovine serum, 100 U/ml penicillin, To gain insight into these questions, we carried out a genome- 100 mg/ml streptomycin and 1% L-glutamine. After 3 or 4 days wide scan on MSCs derived from MM patients and healthy in culture, non-adherent cells were removed, whereas MSCs donors using array-based comparative genomic hybridization were selected by their adherence to the plasticware (this was (array-CGH). This technique has already been successfully used considered as passage 0, P0).10 For five selected donors, the to explore the presence of gains and losses of genetic material in non-adherent fraction was not discarded, but harvested to several types of cancer21–25 and also to discriminate, within the obtain a normal control DNA for array-CGH hybridization. The genomic alteration, the contribution of the stromal vs epithelial culture medium was replaced twice weekly until MSC cultures compartments.26 Besides, we have also explored the presence of were approximately 90% confluent or had remained a maxi- genomic alterations of myelomatous PCs in the matching MSCs mum of 21 days in culture; at this point, cells were trypsinized from the same patients using fluorescence in situ hybridization (0.05% trypsin-EDTA) and expanded in a 1:3 ratio (P1). The (FISH). MSCs were collected near confluency at P3 for subsequent genomic DNA isolation or FISH studies. No apparent haema- tological cell contamination was observed under the micro- Materials and methods scope at the time of MSC harvest. The cultures were maintained at 37 1C and 5% CO2. All the cell culture media and reagents Participants were purchased from Invitrogen (Paisley, UK). A total of 26 patients with newly diagnosed MM were included Selected MSCs from both MM patients (n ¼ 4) and healthy in this study (median age was 68 years, ranging from 28 to 89 donors (n ¼ 4) at P3 were also tested to meet the minimal criteria years). Characteristics of the patients are listed in Table 1. A total as defined by the International Society for Cellular Therapy for of 12 healthy controls of BM samples were obtained from multipotent mesenchymal stromal cells.27 These criteria in- participants undergoing orthopaedic surgery, with a similar age cluded specific cell surface antigen expression and trilineage range to MM patients (median age 58 years, range 26–88 years). mesenchymal differentiation potential. Combinations of mono- Every sample was obtained after receiving informed written clonal antibodies anti-CD34-APC (allophycocyanine), -CD19- consent of patients and donor volunteers and following approval APC, -CD45-PerCPCy5.5, -HLA-DR-PerCPCyC5.5, -CD14-FITC from the Ethical Committee of our Institution. (fluorescein isothiocyanate) (Becton-Dickinson Biosciences, San Jose´, CA, USA), -CD90-FITC (allophycocyanine), -CD73- PE, -CD166-PE, -CD106-PE (BD Biosciences Pharmingen, San MSC harvest, culture conditions and characterization Jose´, CA, USA) and anti-CD105-FITC (ImmunoStep, Salamanca, Mononuclear cells from BM samples were obtained after density Spain) were used to label 5 Â 105 cells at dilutions recom- gradient centrifugation using Ficoll-Paque Premium 1.073 (GE mended by the manufacturers. The appropriate isotopic control

Table 1 Multiple myeloma patient characteristics

Sample Gender Age (years) MM type PCBM MM-PC cytogenetic alterations (FISH)a Bone lesions

1 M 66 IgA 11 Normal IGH Tx, RB and P53 +++ 2 F 81 IgG k 6,2 Normal IGH Tx; del RB: 85%; normal P53 + 3 M 83 IgA k 1 Normal IGH Tx; del RB: 83% À 4 F 70 Bence Jones k 38 Normal IGH Tx; del RB: 88%; normal P53 +++ 5 F 72 UK 12 del RB: 25% À 6 M 86 IgA k 14 t(4,14): 87%; del RB: 88% À (Paget) 7 M 82 IgA k 14,5 Normal IGH Tx; del RB: 98%; normal P53 À 8 F 67 Bence Jones k 2,1 t(11,14): 57%; del RB:11%; normal P53 À? 9 F 44 IgG k 31 t(11,14): 56%; normal RB + (hyperCa) 10 F 67 UK 24 Normal IGH Tx, RB and P53 UK 11 M 89 IgG k 43 t(4,14): 80%; normal RB and P53 À 12 F 58 Bence Jones, plasmacytoma 38,3 Normal IGH Tx; del RB: 67%; normal P53 +++ 13 F 75 IgG k 23 Normal IGH Tx; del RB: 80%; normal P53 ++ 14 M 61 IgG k 8,3 t(4,14): 89%; del P53: 90% À 15 F 79 IgA k, oligosecretory 12 t(4,14): 65%; del RB: 65% +++ 16 F 63 IgA k 35,6 Normal IGH Tx, RB and P53 À 17 F 59 Bence Jones k 19 t(11,14): 61%; normal RB and P53 +++ 18 M 55 Bence Jones k 72,5 t(11,14): 98%; del RB: 90%; normal P53 + 19 F 28 Bence Jones k 7,3 Normal IGH Tx; del RB: 23%; del P53: 25% ++ 20 M 77 IgA l 11 Normal IGH Tx; del RB: 40% À 21 M 72 IgA k 14 t(11,14): 90%; normal RB and P53 À 22 F 57 Bence Jones k 22 IGH Tx: 86%; normal RB and P53 + 23 M 61 IgA k 25 t(4,14): 95%; del RB: 90%; normal P53 + 24 F 75 Bence Jones k 15 Normal IGH Tx; del RB: 77%; normal P53 + 25 M 65 UK 20 t(4,14): 86%; del RB: 71%; del P53: 55% UK 26 M 63 IgG k 28 Normal IGH Tx, RB and P53 + Abbreviations: F, female; FISH, fluorescence in situ hybridization; hyperCa: hypercalcemia; IgG, immunoglobulin G; M, male; MM, multiple myeloma; Paget: Paget’s disease; PC, plasma cell; PCBM, myelomatous PCs infiltration in bone marrow as assessed by flow cytomety; UK, unknown. aCytogenetic markers tested are stated for each patient.

Leukemia Genomic profile of multiple myeloma MSCs M Garayoa et al 1517 was also included. Labelled cells were acquired using a and labelled with Cy3-dCTP or Cy5-dCTP. Paired hybridization FACSCalibur flow cytometer (Becton Dickinson Biosciences) samples were then cohybridized to the arrays in a HS 4800 Pro- using the CellQuest program (Becton Dickinson Biosciences); Tecan (Tecan, Ma¨nnedorf, Switzerland) hybridizer for 48 h at percentages of antigen expression were obtained using the 42 1C according to the manufacturer’s recommended protocol. Paint-A-Gate program (Becton Dickinson Biosciences). The For each microarray, data from two-color hybridizations were purity of MM-MSCs after tissue culture expansion at P3 was normalized using print-tip loess method with the Diagnosis and estimated to be above 98% in all the cases. At this point, both Normalization Module Arrays Data Tool of the GEPAS soft- MM- and ND-MSCs were consistently devoid of haematopoietic ware.29 Many threshold values to identify gains or losses in cells, lacking expression of haematopoietic antigens (CD34, array-CGH analyses have been reported in the literature CD19, CD45, CD14, HLA-DR), whereas being positive for depending on the type of study.30 We carried out array-CGH CD73, CD90, CD166, CD105 and CD106. analyses of five ‘normal to normal hybridizations’ (genomic In addition, osteogenic, adipogenic and chondrogenic differ- DNA from the mononuclear fraction of five selected donors and entiation was also confirmed in four MM-MSC and four genomic DNA from placenta) and identified the median±3s.d. ND-MSC samples at P3 using standard in vitro tissue culture- as the cutoff level to use in our study. Owing to the resolution of differentiating conditions. In brief, for osteoblastic and adipo- our array-CGH (75–100 kb length for each clone and E1Mb cytic differentiations, 80% confluent MSCs were grown for 2 or spacing/intervals between the clones), at least two consecutive 3 weeks in a specific differentiation medium changing the clones with log2 ratios above or below our threshold value were medium on every fourth day (a minimal essential medium needed to be considered as genomic gains or losses, respec- supplemented with 10% fetal bovine serum, 10 mM b-glycerol tively. Data sets were carefully reviewed for frequently affected phosphate (Sigma-Aldrich, St Louis, MO, USA), 50 mg/ml chromosomal sites of physiological copy number polymor- ascorbic acid and 10nM dexamethasone for osteoblasts, or NH phisms (CNP: http://www.project.tcag.ca/variation and DECIPHER: AdipoDiff medium (Macs Media, Miltenyi Biotec, Bergisch http://www.decipher.sanger.ac.uk/perl/application/).31 Detailed Gladbach, Germany) for adipocytes). To induce chondrocytic information of array-CGH production, hybridization and ana- differentiation, a MSC pellet was maintained in a conical lysis is available in the Supplementary material 1. polypropylene tube with NH ChondroDiff medium (Mitenyi Biotec) for 24 days changing the medium on every third day. Evaluation of osteocytic, adipocytic or chondrocytic differen- Array-CGH on sorted MSCs tiations of MSCs was achieved using alkaline phosphatase, Genomic DNA from sorted MSCs was extracted using the Oil-Red-O, and Masson’s trichrome stainings, respectively. Qiamp DNA Mini kit (Qiagen GmbH) and then subsequently amplified with the GenomiPhi DNA amplification kit (Amersham Biosciences, Piscataway, NJ, USA) following the Sorting of MSCs manufacturer’s instructions. In brief, genomic DNA was Isolation of MSCs directly from BM aspirates was carried denatured at 95 1C and then amplified using the Phi29 DNA out in five MM patients (No. 22–26 in Table 1), using a polymerase at 30 1C for 16 h. After heat inactivation of the FACSAria flow cytometer equipped with the FACSDiva software enzyme, post-amplification cleanup was achieved by standard (Becton Dickinson Biosciences) as described earlier.28 In brief, ethanol precipitation. The absence of human-specific sequences cells were stained with a single four-color combination of in non-specific amplification products was verified using PCR. monoclonal antibodies: anti-CD45-FITC (Miltenyi Biotec)/anti- Array-CGH was carried out as described for MSCs at P3 with CD73-PE (BD Biosciences Pharmingen, San Jose´, CA, USA)/ minor modifications. Essentially, amplified genomic DNAs from antiCD34-PeCy5 (Immunotech Coulter Company, Marseille, sorted MSCs were used as test DNAs. For each patient, genomic France)/anti-CD271-APC (Miltenyi Biotec). Isolation of MSCs DNA from ammonium chloride-lysed peripheral blood, or was achieved by sorting of BM cells with the CD45À/CD73 þþ/ otherwise, genomic DNA from CD138À/CD271À/CD73À BM- CD34À/CD271 þþ phenotype. Owing to the low proportion of sorted cells were similarly isolated and amplified, and used as MSCs in the sample, acquisition of MSCs was carried out reference genomic DNAs. Both test and reference DNAs were through a ‘live gate’ drawn on the CD73 þ /CD271 þ region of amplified using the same amount of starting DNA. the total nucleated cells present in the sample. The purity of the isolated MSC populations was 99%. Fluorescence in situ hybridization analyses For each patient, MM-MSCs at P3 were grown on sterilized Array-based comparative genomic hybridization of microscope slides and were used to examine the presence of MSCs at passage 3 specific genomic alterations found in the matching myeloma The DNA from MSCs was isolated after trypsinization of cells PCs using interphase FISH and standard protocols.32 MSC reaching 90% confluency at P3, using the Qiamp DNA Mini kit pretreatment and probes used for detection of IGH transloca- (Qiagen GmbH, Hilden, Germany). Normal DNA was prepared tion, specific IGH translocations, and 13q and 17p deletions can from human placenta of healthy donors using standard methods. be consulted in Supplementary material 2. A total of 200 Genomic-wide survey of DNA copy number changes from 21 interphase nuclei were analysed for each probe in this study. cases of MM-MSCs and 12 ND-MSCs was carried out using ND-MSC interphases were used as controls. array-CGH. Slides containing 3528 bacterial artificial chromo- To confirm selected genomic imbalances identified by array- somes (BACs) spanning the full genome at E1-Mb density were CGH, FISH analysis was also carried out on MM-MSCs using produced at our institution as distributed by the Welcome Trust DNA from correspondent BAC clones as specific probes (see Sanger Institute (Cambridge, United Kingdom; clone content is Table 2). In brief, DNA was isolated and directly labelled with available in the ‘Cytoview’ windows of the Sanger Institute either digoxigenin-dUTP or biotin-dUTP using nick translation mapping database site (http://www.ensembl.org/). In brief, (Abbot Molecular Inc., Des Plaines, IL USA). After DNAse reference genomic DNA (DNA of placenta) and non-amplified I-controlled digestion and further purification of the probe, FISH genomic test DNA (DNA from MSCs) were separately digested was carried out on correspondent MM-MSC slides using TRITC

Leukemia Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1518 Table 2 FISH probes used for array-CGH conﬁrmation and FISH results

BAC clone Clone Clone size Putative gene encoded Patient no. (% cells location carrying the alteration)a

Losses RP11-304A15b 2p13.3 164 kb Rho GTPase activating protein 25 (ARHGAP25)//Bone 8 (48) morphogenetic protein 10 (BMP10) RP11-245N4b 2p13.3 140 kb Small nuclear ribonucleoprotein 10 (snRNP G) 8 (50) RP11-25F15c 3q13.12 149 kb Leukocyte surface CD47 antigen precursor (CD47)// 19 (68) Intraﬂagellar transport 57 homolog (IFT57) RP11-93N14c 3q13.13 154 kb Guanylate cyclase activator 1C (GUCA1C)//MORC family 13 (80), 19 (76) CW-type zinc ﬁnger protein 1 (MORC1)

Gains RP4-722L13 1p22.3 95 kb HIT-type zinc finger-containing protein C1orf181 (Serologically 4 (65), 10 (73), 15 (49) defined breast cancer antigen NY-BR-75) RP11-353N19d 14q32.12 172 kb EGF-like protein fibulin 5 (FBLN5)//Thyroid hormone receptor 17 (65) interactor 11 (TRIP11) RP11-371E8d 14q32.12 193 kb Inositol 1,3,4-triphosphate 5/6 kinase (ITPK1) 17 (64) RP5-1075G21 20q13.2 110 kb Cytochrome P450, family 24, subfamily A (CYP24A1) 15 (45) Abbreviations: array-CGH, array-based comparative genomic hybridization; BAC, bacterial artificial chromosome; FISH, fluorescence in situ hybridization. aAs assessed by FISH. b–b, c–c, d–d Contiguous BAC clones.

Table 3 Primer sequences of genes evaluated by real-time PCR (copy number variation) (from 50 to 30)

Gene Forward primer Reverse primer

BCMO1 CGTGCCTCTGTTGATGTCGAT CTTGTCGCCCTGTCCCG CaM1 GCAGTAACTTGAGAGGGCATAGC TAACCATTGCCATCCTGTGAAC EPBH1 CCCTTAGCCCCAGGATTCTC GTAGGTGCGGATGGTGTTCA KCTD8 TTGGATCGCCCCTCTAAAAA CCCGGGACTTGCTGATGA C1orf181 ATCAGTGGCTCCACATTGTTTG ATGTGATGAAAGGCACAGAATTCA Albumina TGAAACATACGTTCCCAAAGAGTTT CTCTCCTTCTCAGAAAGTGTGCATAT RNAse Pa AACCAAACATTGTCGTTCAGAAGA TTACTTGCTTTGCTTTCATCTACCA aControl genes for copy number variation analysis.

(tetramethyl rhodamine isothiocyanate)- or FITC-labelled Step One Plus Real-Time PCR System with TaqMan Fast secondary antibodies. A minimum of 100 interphase nuclei Universal PCR Master Mix and TaqMan Gene Expression Assays were scored. (Applied Biosystems) according to the manufacturer’s instructions. Experiments were carried out in duplicate for both the target and the endogenous gene (GAPDH). Relative quantifica- Real-time PCR-based analyses tion of the target gene expression was calculated using the To check the most frequent single targets from array-CGH results, comparative threshold cycle (Ct) method. The expression of a we carried out a relative quantification of DNA copy number of target gene was normalized to the expression of an endogenous several genes using the Step One Plus Real-Time PCR System and control (GADPH), and presented relative to the expression in an the Fast SYBR-Green Master Mix according to the manufacturer’s experiment-specific calibrator (control patient with no genomic instructions (Applied Biosystems,FosterCity,CA).Primersequences imbalances for the target gene). The value of the relative for these genes and for the reference genes (albumin and RNAse P) expression of the target gene was calculated using the equation are provided in Table 3. Five ND-MSC DNA samples were tested as a calibrator. The arithmetic mean of replicated C values for DD t 2À Ct each gene was transformed to a relative quantity, using the equation ÀDDCt ð1 þ EÞ where E is the amplification efficiency, DCt ¼ where DCt ¼ Ct À Ct and DDCt ¼ DCt À 33 Target gene GAPDH sample for target CtGene À CtControl genes ðmeanÞ and DDCt ¼ DCtpatient À DCtcalibrator ðmeanÞ . Addi- DCtcalibrator for target ðmeanÞ . Results in control patients were normalized tional aspects of primer design and real-time PCR procedure for by setting the corresponding value as 1. The assay IDs were: copy number variation are explained in Supplementary material 3. FBLN5, Hs00197064_m1; ZNF217, Hs00232417_m1; ITPK1, The expression level of certain genes located in array-CGH- Hs356546_m1; KCTD8, Hs00957322_s1, and BMP10, detected gains (FBLN5, ZNF21, ITPK1) or losses (KCTD8, Hs00205566_m1. BMP10) of genomic DNA were also quantified using real-time RT-PCR. Total RNA was isolated from MSCs at passage 3 using the RNeasy Mini kit (Qiagen GmbH). Reverse transcription was Data analyses carried out with 1.0 mg RNA in the presence of random For unsupervised cluster analysis of array-CGH data, we hexamers and 100 U of SuperScript II reverse transcriptase converted the relative ratio value for each BAC clone to a score (Invitrogen, Carlsbad, CA, USA). For PCR reactions, we used the of 1 (gain/amplified), 0 (no change), or À1 (loss) based on the

Leukemia Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1519 Table 4 Losses of genomic material deﬁned by two contiguous BAC clones

Contiguous Chromosome Putative affected gene(s) Frequency of BACs location losses

RP11-385M4 1q32.2 1/21 RP5-879K22 1q32.2 Hedgehog acyltransferase (HHAT) RP11-304A15 2p13.3 Rho GTPase activating protein 25 (ARHGAP25)//Bone morphogenetic protein 10 (BMP10) 1/21 RP11-245 N4 2p13.3 Small nuclear ribonucleoprotein 10 (snRNP G) RP11-115B22 3q13.11 1/21 RP11-91B3 3q13.11 E3 ubiquitin-protein ligase CBL-B (CBLB) RP11-91B3 3q13.11 E3 ubiquitin-protein ligase CBL-B (CBLB) 1/21 RP11-20N7 3q13.11 RP11-25F15 3q13.12 Leukocyte surface CD47antigen precursor (CD47)//Intraflagellar transport 57 homolog 1/21 (IFT57) RP11-93N14 3q13.12 Guanylate cyclase activator 1C (GUCA1C)//MORC family CW-type zinc finger protein 1 (MORC1) RP11-93N14 3q13.13 Guanylate cyclase activator 1C (GUCA1C)//MORC family CW-type zinc finger protein 1 2/21 (MORC1) RP11-286P15 3q13.13 RP11-129J11 3q21.3 1/21 RP11-21N8 3q21.3 Phosphoinositide 3-kinase regulatory subunit 4 (PIK3R4) RP11-91K8 3q22.1 Phakinin, beaded filament structural protein 2 (BFSP2) 1/21 RP11-22E12 3q22.1 Kyphoscoliosis peptidase (KY)//Ephrin type-B receptor 1 (EPHB1) RP11-103J17 4p14 1/21 RP11-36B15 4p14 Receptor expressed in lymphoid tissues like 1 (RELL1) RP11-392K14 4p14 1/21 RP11-213G21 4p14 Kruppel-like factor 3 (basic) (KLF3) RP11-542I3 4p14 Potassium channel tetramerization domain 8 (KCTD8) 3/21 RP11-55C6 4p13 RP11-55C6 4p13 1/21 RP11-416A5 4p13 Gamma-aminobutyric acid (GABA) A receptor, gamma 1 (GABRG1) RP11-336A10 10p15.1 Ankyrin repeat and SOCS box-containing 13 (ASB13) 1/21 RP11-298K24 10p15.1 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 3 (PFKFB3)//Renal carcinoma antigen NY-REN-56 RP11-189M8 10p14 1/21 RP11-251M9 10p14 RP11-90E5 15q26.2 ADAM metallopeptidase with thrombospondin type 1 motif, 17 (ADAMTS17) 1/21 RP11-168G16 15q26.2 LAG1 longevity assurance homolog 3 (LASS3) Abbreviation: BAC, bacterial artificial chromosome.

median±3s.d. cutoff for each microarray, and analysed data Presence of ‘hot-spot’ regions in cultured MM-MSC with Cluster and TreeView (Multi Experiment Viewer 4.0) based When comparing along the genomic alterations of MSCs from on the average linkage method with the Pearson Uncentered both normal donors and MM patients, we observed that in the metric correlation. latter, there were specific regions of variable length in the Statistical evaluation was carried out using the SPSS 15.0 genome, in which patients showed discrete alterations (‘one- statistical software. All P-values reported were two-sided and target BAC alterations’, o1 Mb in size) (Figure 2). We have statistical significance was defined as P-values o0.05. named these regions as ‘hot-spots’ in MM-MSCs, and their specific location along the genome is depicted in Figure 3. We found a predominance of hot-spots with discrete gains of genomic material over that of hot-spots with genomic losses. It Results should be noted that these abnormalities were present at a higher frequency than when two contiguous BACs were Genomic changes in cultured ND-MSCs and MM-MSCs considered (see Supplementary Table 1). Although some may When we looked for gene copy number variation on arrays from argue against the specificity of these alterations, it is of interest ND-MSCs, no evidence was found for genomic imbalances. By that these regions of one-target BAC alterations were constantly contrast, several chromosomal gains and losses were detected in absent in ND-MSCs (Figure 2). MM-MSCs as defined by two consecutive-target BACs (regions 41 Mb) (see Tables 4 and 5). These DNA copy number changes were not recurrent within the patients. Only a chromosomal loss Unsupervised analysis of genomic changes in cultured located at 4p14-4p13 was observed in 3 out of 21 MM-MSCs, MSCs and another loss at 3q13.13 was detected in MSCs from two When an unsupervised cluster analysis with array-CGH data patients. However, although these genomic alterations showed from both ND- and MM-MSCs at P3 was carried out, the almost no recurrence within patients, in many cases the affected dendogram discerned two major clusters with common MSC BACs were located within the same chromosomal region origin (Figure 4). One cluster contained all 12 ND-MSCs (encompassing a variable length of 6–15 Mb), so that in several samples with only four MM-MSCs interspersed between them; patients the genomic alterations partially overlapped or were these four samples from MM patients were found to have located nearby (see Figure 1). minimal genomic imbalances. The other cluster exclusively

Leukemia Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1520 Table 5 Gains of genomic material deﬁned by two (or more) contiguous BAC clones

Contiguous Chromosome Putative affected gene(s) Frequency BACs location of gains

RP11-463C8 14q24.3 Transmembrane protein 63C (TMEM63C) 1/21 RP11-61F4 14q24.3 SRA stem-loop-interacting RNA-binding protein, mitochondrial precursor (C14orf156)// SNW domain containing 1 (SNW1)//aarF domain containing kinase 1 (ADCK1) RP11-61F4 14q24.3 SRA stem-loop-interacting RNA-binding protein, mitochondrial, precursor (C14orf156)// 1/21 SNW domain containing 1 (SNW1)//aarF domain containing kinase 1 (ADCK1) RP11-285P21 14q24.3 RP11-526N18 14q31.1 Neurexin 3 (NRXN3) 1/21 RP11-114N19 14q31.1 Thyroid stimulating hormone receptor (TSHR) RP11-114N19 14q31.1 Thyroid stimulating hormone receptor (TSHR) 1/21 RP11-226P1 14q31.1 RP11-353N19 14q32.12 Fibulin 5, EGF-like protein (FBLN5)//Thyroid hormone receptor interactor 11 (TRIP11) 1/21 RP11-371E8 14q32.12 Inositol 1,3,4-triphosphate 5/6 kinase (ITPK1) RP11-245A17 18p11.21 1/21 RP11-92E1 18p11.21 RP11-78A19 18p11.21 Guanine nucleotide-binding protein G (GNAL) RP4-715N11 20q13.2 1/21 RP4-724E16 20q13.2 Zinc finger protein 217 (ZNF217) RP4-724E16 20q13.2 Zinc finger protein 217 (ZNF217) 1/21 RP5-1075G21 20q13.2 Cytochrome P450, family 24, subfamily A, polypeptide 1 (CYP24A1) Abbreviations: BAC, bacterial artificial chromosome; EGF, epidermal growth factor.

contained MSCs derived from the remaining MM patients. One calculated as the ratio between its expression on MM-MSC sample of a MM patient segregated alone. samples at P3 from patients carrying the genomic imbalance, We also carried out unsupervised cluster analysis on array- and the mean expression of the same target gene on MSCs from CGH data only from MSCs obtained from myeloma patients. patients with normal copy number for the affected BAC (control Patient segregation did not seem to correlate with clinico- patient). As can be seen in Figure 6b, a downregulated biological features from patients (age, MM-type, MM stage at expression of KCTD8 and BMP10 was observed in MSCs of diagnosis, PC inﬁltration, PC cytogenetics determined using patients with correspondent lost BACs; on the contrary, three out FISH, presence of bone lesions and patient response to of four patients with a genomic gain at array-CGH level treatment). presented increased RNA levels for FBLN5, ZNF217 and ITPK1.

Array-CGH confirmatory studies using FISH and Array-CGH on sorted MM-MSCs real-time PCR-based analyses In five MM cases, array-CGH analysis was conducted after To confirm some of the results obtained using array-CGH, FISH direct sorting of MSCs from BM aspirates. Although discrete studies were carried out on MSCs derived from MM patients at gains or losses of genomic material (o1Mb) were observed in P3 and using DNA from selected BAC clones as specific probes every sample, only one MM-MSC sample presented genomic (see Table 2). Four gains and four losses of genomic material gains 41 Mb in size (see Table 6). were tested and confirmed using FISH (Figure 5). It should be noted that only a percentage (ranging from 45 to 80%) of the scored MSCs were found to present a duplication or deletion of MM-MSCs do not share cytogenetic alterations found in the correspondent DNA clone, whereas the rest of the cells plasma cells remained diploid (Figure 5 and Table 2). We also tested the presence of the cytogenetic aberrations of We also attempted to confirm most recurrent gains or losses myelomatous cells in MSCs derived from the same patient using from the hot-spot regions using a real-time PCR assay for gene interphase FISH studies. From a total of 21 patients, PCs from 18 copy number variation and primers located at candidate genes patients showed IGH translocations, or deletions of RB or P53, within the gained or lost BACs (see Table 3). In this type of assay, as assessed using FISH (see Table 1). In none of these 18 a relative value of 1 should be expected for normal diploid copy patients, MM-MSCs displayed the genetic abnormalities present number and, subsequently, a value of p0.5 would be associated in the correspondent myelomatous PCs (data not shown). with one allele deletion and a value X1.5 would be associated with one allele duplication.33 Except for one case (BCMO1), real-time PCR was unable to detect DNA copy number Discussion variations (either gains or losses) observed using array-CGH (Figure 6a). This discrepancy is probably related to the fact that Although the implication of the BM microenvironment in the genomic alterations were present only in a fraction of MSCs, MM4,5,7 (and that of the surrounding stroma in solid as shown above in FISH analyses. tumours34–36) as key determinants in the malignant progression In addition, we evaluated the RNA expression levels of five of cancer is nowadays well-accepted, it is not clear whether this putatively affected genes using real-time RT-PCR, which were role is in turn supported by genomic alterations of stromal cells, located within losses (KCTD8, BMP10; see Table 4) or gains which can coevolve in the tumourigenesis process.26 In fact, (FBLN5, ZNF217, ITPK1; Table 5) of genomic material using several independent groups have documented the existence of array-CGH analysis. The expression of each target was genomic alterations (including copy number variation, somatic

Leukemia Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1521 mutations and/or loss of heterozygosity) in the stroma of several types of solid tumours37–44 Subsequently, genetically altered stromal cells would undergo clonal selection for stromal cells with the ability to modify tumour–stromal interactions and promote tumour growth.26,37 Even more, in a mouse model of prostate carcinoma, selection of genetic changes in the stroma have been found to occur from paracrine pressures imposed by oncogenic stress in the epithelium.41 In this study, we conducted array-CGH to explore genomic changes in MSCs (as progenitor cells of stromal cells and osteoblasts in the BM) derived from MM patients, and to compare them with those in MSCs derived from normal donors. p16 p17 p18 p12 p21 p20 normal donors (ND-MSC; d1-d12) and Besides, for each patient, the possibility of MSCs sharing the

p19 cytogenetic aberrations found in myelomatous PCs was also

) and were found only in MM-MSCs. Some of evaluated using FISH. p14 Our results from array-CGH show that the genomic proﬁles of he same region. p13 ND-MSCs and of MM-MSCs after culture are clearly distinctive.

p15 In fact, ND-MSCs, as studied in P3, do not show genomic imbalances using array-CGH. Our ﬁnding is in agreement with p11 that of other authors who have genetically characterized BM

p10 MSCs because of their potential use in cell therapy and tissue

p9 engineering approaches. These authors did not show evidence

p8 of chromosomal abnormalities in MSCs using cytogenetic

p4 analyses and array-CGH at time of collection, at passage 3 or even after prolonged in vitro culture.45–48 p3 On the contrary, we have shown that in vitro expanded MSCs p6 derived from MM patients present several gains and losses of genomic material, as deﬁned by two contiguous-target BAC clones (41 Mb). In general, these alterations were not recurrent

p2 p7 p5 within the patients, as the highest frequency observed for a

p1 genomic deletion was 3 out of 21 MM-MSC samples (at 4p14- 4p13), and another deletion occurring in two of MM-MSC samples (3q13.13); by contrast, no recurrence was observed for the rest of the deletions or at any of the genomic gains (see Tables 4 and 5). Although some of the implicated BAC clones did not encompass any known genes, other BACs harboured genes with potential biological interest in MSC physiology and/ or myeloma–stroma interactions (for example, bone morpho-

d9 d5 d8 d10 d11 d12 genetic protein 10 (BMP10), ephrin type-B receptor 1 (EPHB1),

d7 fibulin 5 EGF-like protein (FBLN5), receptor expressed in lymphoid tissues like 1 (RELL1), ADAM metallopeptidase with thrombospondin type 1 motif 17 (ADAMTS17)). Interestingly, for most affected patients, we found correlative deregulated expression for five genes encoded in the regions detected as d2 d3 d4 d6 genomic gains or losses using array-CGH (see Figure 6b). These d1 altered expressions may have an impact on the pathogenesis or the course of the disease. For example, a downregulated 4p15.1 4p15.1 4p15.1 4p14 4p14 4p14 4p13 4p13 4p13 4p13 4p13 4p13 4p12 4p12 4p14 expression of BMP10 in MSCs may affect their osteoblastic differentiation capacity; the augmented levels of FBLN5, being a protein similar to epidermal growth factor, may promote the growth of myelomatous cells. The non-recurrent character of our results suggests that different stromal genomic alteration 26,37,44 RP11-55C6 RP11-213G21 RP11-227F19 RP11-109E24 RP11-542I3 RP11-416A5 RP11-100N21 RP11-19F13 RP11-79E3 RP11-24H13 RP11-135M12 RP11-103J17 RP11-36B15 RP11-392K14 RP11-415L23 pathways might be effective in creating a microenvironment permissive for the growth of MM cells. Besides, we also identified a specific pattern of one-target BAC imbalances in cultured MM-MSCs at certain regions of

33,60Mb 36,76Mb 42,40 Mb 45,70Mb 48,58Mb the genome, which we termed ‘hot-spot’ regions (see Figure 3). These regions included both losses and gains of genomic material, although the latter predominated. As these discrete

4 genomic alterations (o1 Mb) are under the resolution limit of our array-CGH, we cannot exclude the presence of false positive Distribution of unbalanced genomic alterations detected by array-based comparative genomic hybridisation in mesenchymal stem cells derived from results and validation for each case is needed. Nevertheless, the lack of one-target BAC imbalances in ND-MSCs as well as the high frequency observed for some of these genomic alterations

Figure 1 from multiple myeloma patients (MM-MSCs;the p1-p21) at losses 4p14-4p13, were next coincident to the within centromere. patients, Most whereas of the others genomic partially alterations overlapped were deletions or (shown were in located grey may at nearby bacterial artiﬁcial chromosomes within t reﬂect a true recurrent aberration in MSCs (see

Leukemia 1522 Leukemia

18 d1 d2 d3 d4 d6 d7 d9 d5 d8 d10 d11 d12 p1 p2 p7 p5 p6 p3 p4 p8 p9 p10 p11 p15 p13 p14 p19 p16 p17 p18 p12 p21 p20 56 Mb RP11-396N11 18q21.32 RP11-520K18 18q21.32 RP11-13L22 18q21.32 RP11-215A20 18q21.33 eoi rﬁeo utpemeoaMSCs myeloma multiple of proﬁle Genomic RP11-28F1 18q21.33 RP11-233O10 18q22.1 RP11-23N5 18q22.1 RP11-217N19 18q22.1 RP11-389J22 18q22.1 63Mb RP11-526H11 18q22.1 RP11-21L20 18q22.1 RP11-342G3 18q22.2 RP11-484N16 18q22.2

66,84Mb RP11-430H7 18q22.2 Garayoa M RP11-45A1 18q22.3 RP11-169F17 18q22.3 70,88Mb RP11-556L15 18q22.3 tal et

d1 d2 d3 d4 d6 d7d9d5 d8 d10 d11 d12 p1 p2 p7 p5 p6 p3 p4 p8 p9 p10 p11 p15 p13 p14 p19 p16 p17 p18 p12 p21 p20 32,90 Mb RP11-48O9 11p13 RP1-85M6 11p13 RP1-316D7 11p13 RP11-202M19 11p13 RP11-31I23 11p13 RP4-607I7 11p13 32,90 Mb RP1-136N16 11p13

Figure 2 Schematic representations of genomic imbalances in mesenchymal stem cells derived from normal donors (ND-MSCs; d1-d12) and from multiple myeloma (MM-MSCs; p1-p21) patients detected by array-based comparative genomic hybridisation. (a) Hot-spot region at the q arm of chromosome 18 (only present in MM-MSCs), with predominating single-bacterial artiﬁcial chromosomes (BAC) gains of genomic material shown in black, and with low recurrence within patients. (b) Smaller hot-spot region in the p arm of chromosome 11, with predominance of single-BAC losses of genomic material shown in grey. Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1523 36.3 36.2 36.1 25 35 34.2 24 33 23 32 22 21 26 31 16 25 24 16 14 22 15.3 13 22 15.3 25 12 15.1 15.1 24 21 21 11.2 14 14 23 22 13 13 13 22 21 11.2 14 12 22.3 12 12 12 21.3 23 22.2 12 15 24 15 22.1 13 13 11.2 22 14.1 13 21.2 14 23 14 15 12 12 12 13 21 13 21 14.3 12 21 21 13 12 12 14 11.4 21 21 12 11.2 12 11.2 13 11.2 13 22 22 22 14 13 12 12 11.2 13.1 14 11.2 23 23 24 15 11.2 11.2 11.2 24 15 12 24 13.3 12 25 26 21 16 21 13 12 13 21 22 13 21 31 27 21.2 31 22 21 21 13 23 23 22 21 32.1 28 21.3 22 24 22 14 22 32.3 31.1 22 32 33 25 31 31 22 23 21 23 31.3 23 34 26.1 31 24 22 41 32 32 24 32 23 25 35 26.3 33 33 32 33 33 25 23 26 42 36 27 34 25 35 24.1 28 34 26 27 43 37 36 34 26 24 44 29 35 35 27 24.3 25 28 1 2 3 4 5 6 7 89 10 11 X

13 12 13 13 11.2 12 12 13 11.2 11.2 12 11.2 12 13.3 12 11.2 13 13 12 13.1 13 12 12 11.3 14 13 14 12 13.3 14 15 11.2 11.2 15 21 11.2 13.2 13 11.3 21 11.2 13.1 13 13 21 21 12 11.2 12 12 12 11.2 22 11.2 12 23 22 21 12 11.2 11.2 11.2 22 22 12.1 12 24 23 13 11.2 11.2 11.2 23 31 13 22 21 13.1 11.2 24 23 21 12 24.1 32 31 25 22 13.2 12 24 22 13.1 12 24.2 33 23 13.3 22 13 24.3 34 32 26 24 25 23 13.4 13.2 12 13 14 15 16 17 18 19 20 21 22 Y

Figure 3 Distribution of hot-spot regions found in mesenchymal stem cells derived from multiple myeloma patients by array-based comparative genomic hybridisation in our study. The bars represent hot-spot regions with predominant gains (depicted in black, right-hand side of the chromosome), or losses of genomic material (shown in grey, left-hand side of each chromosome). p1 d3 d2 d4 d1 d7 d5 d6 d8 d9 p6 p2 p7 p5 p9 p8 p3 p4 p18 d12 p11 d10 d11 p14 p21 p17 p15 p13 p10 p19 p12 p16 p20

Figure 4 Unsupervised hierarchical cluster analysis showing major separation of mesenchymal stem cells from multiple myeloma (MM) patients (right side of the dendogram) and from normal donors (left side of the dendogram). Four mesenchymal stem cells derived from multiple myeloma patients samples (with less genomic imbalances), were intermingled within the normal donor samples, and another MM patient’s sample segregated alone.

Supplementary Table 1, with more frequent alterations in 6–3/ partial gain of 17q material at a similar amplicon has also been 21 patients, which supposes a 28–14% of MM patients in this reported in human embryonic stem cells in culture,49 and in study). neuroblastoma,50 breast51,52 and other types of cancers. These One aspect to be considered is the putative biological evidences are indicative that this region may house a number of signiﬁcance of these multiple and small genomic alterations in genes that are important to malignant cells.53 This could also be the hot-spot regions. We can hypothesize that these zones might the case for MM-MSCs, and an increased genomic dosage of be more genetically unstable in MM-MSCs and may contain some of the affected 17q genes in the MSCs of certain patients genes of importance in cancer progression; if that is the case, may confer a stromal microenvironment advantageous to the random ampliﬁcation of discrete regions within these zones in growth of myeloma cells. Alternatively, the presence of these MM-MSCs may affect expression of genes of importance to hot-spot regions with o1 Mb genomic alterations, although myeloma progression. In this sense, one of the detected hot-spot irrelevant per se, may predispose or may be a prerequisite for regions in our study is located at the terminal end of 17q the occurrence of other genomic alterations of biological (17q23.2-17q25.3), and contains 15 single-BAC gains of importance for neoplastic progression. genomic material, with almost no recurrence within the Validation of our array-CGH data was carried out for selected patients, along a region encompassing E27 Mb. Recurrent gains and losses of genomic material (including both X2

Leukemia Genomic profile of multiple myeloma MSCs M Garayoa et al 1524 In view of our array-CGH results on in vitro expanded MSCs, it could be argued that the unbalanced genomic alterations observed in MM-MSCs at passage 3 could be a consequence of changes associated with the cell expansion process and adaptation to tissue culture conditions. To check whether uncultured MM-MSCs (passage 0) already harbour genomic abnormalities, freshly isolated BM stromal cells were sorted by expression of the specific phenotype CD45À/CD73 þþ/CD34À/ CD271 þþ;54 genomic DNA was then isolated from these cells, subjected to amplification and analysed using array-CGH in a similar manner to array-CGH from cultured MSCs. This approach has already been validated by our group to obtain uncultured MSCs from BM aspirates of patients suffering from myelodysplastic syndrome28 and yielded MSCs of 499% purity. As shown in Table 6, genomic imbalances (41 Mb) were present in one out of five sorted MM-MSCs. We cannot rule out that the expansion process and adaptation to culture conditions may increase the genomic changes observed in the sorted population. Alternatively, although some MM-MSCs may Figure 5 Confirmation of array-based comparative genomic hybridisation genomic imbalances by fluorescence in situ hybridization. already contain genomic alterations in the BM microenviron- (a) Patient 8, loss of RP11-304A15; 45% of the cells carried the ment, these alterations may not be present in a sufficient number deletion. (b) Patient 19, loss of RP11-93N14; 68% of the cells carried of MSCs, as to be detected using array-CGH analysis on sorted- the deletion. (c) Patient 17, gain of RP11-353N19; 65% of the cells uncultured MSCs, because of sensitivity limitations of this carried the amplification. (d) Patient 17, gain of RP11-371E8; 69% of technique. It can be speculated that in vitro culture conditions the cells carried the amplification. may allow a positive clonal selection of those MM-MSCs that carry genomic alterations conferring an advantageous phenotype within the MSC population. The fact that our FISH flanking targets or single-target imbalances) using interphase confirmatory studies showed that the genetic changes were FISH, using labelled BACs as probes. Four cases of amplifica- restricted to a fraction of MM-MSCs suggests that clonal tions and four deletions were tested on the correspondent MM- selection processes are effectively occurring under culture. It MSCs cultured on slides (Figure 5; Table 2). Although FISH is likely that a combination of both processes (adaptation of analysis confirmed these genomic alterations, genomic mosai- MSCs to culture conditions plus clonal selection) might be cism was observed in the MSC populations, with a variable responsible for the higher ratio of patients with genomic number of cells harbouring the tested genetic alteration (45– aberrations in MM-MSCs at passage 3 as compared with sorted 80% of the cells were counted positive for the alteration; see MSCs. Nevertheless, the absence of detectable genomic Table 2). Likewise, we also attempted to confirm some of the imbalances (41 Mb) and ‘hot-spot regions’ using array-CGH more recurrent genomic alterations defined by single-target in ND-MSCs as compared with those in MM-MSCs after BACs (o1 Mb) using real-time PCR (gene copy number variation expansion, clearly establishes genomic differences for both assay). In contrast to FISH validation, we were only able to types of MSCs. validate one out of five single-target BAC alterations observed Finally, we have tested the possibility that cytogenetic using array-CGH. This discrepancy could be explained by the aberrations present in myeloma cells could also be present in relative low proportion of MSCs bearing the copy number the matching MM-MSCs. None of the MM-MSCs display variation. The theoretical difference between normal diploid the IGH translocations, nor RB or P53 deletions of myeloma copy and one allele deletion or duplication using real-time PCR cells. Our findings are in agreement with earlier studies should be 1 Ct as long as 100% of the cells harboured such in which MSCs from myeloma patients also failed to exhibit alteration. As the genomic alterations in this study were present the chromosomal abnormalities detected in myeloma PCs.12 only in a fraction of MSCs, the observed differences between Moreover, taking together that MM-MSCs do not contain normal and altered samples might be overlapped by the highest myeloma-specific abnormalities and the fact that array-CGH allowed deviation in real-time PCR replicates. profiles of myeloma cells 55 and those of MM-MSCs (present Regardless of the impact that the genomic alterations found in data) are clearly different, it should be concluded that genetic MM-MSCs may have in the pathophysiology or in the course of alterations of myeloma cells and BM stromal cells occur as the disease, considering the non-recurrent character of the independent events. Accordingly, a different origin for the genomic alterations together with the nature of the putative MM-MSCs and for the neoplastic malignant clone should be genes encoded in those genomic imbalances, it is our postulated. interpretation that these genomic alterations do not seem to be In conclusion, we have observed that cultured MSCs derived responsible, at least directly, for the functional and gene from MM patients present a distinctive array-CGH profile from expression differences reported upon comparing MSCs derived that observed in their normal counterparts; these profiles are from MM patients and those from normal donors.12–14 Other characterized by the presence of several non-recurrent gains genomic alterations, such as point mutations, loss of hetero- and losses (41 Mb) of genetic material together with a specific zygosity/allelic imbalance or epigenetic changes, or even a pattern of ‘hot-spot’ regions with discrete (o1 Mb) genomic distinctive miRNA (microRNA) profile might be responsible for alterations. To which extent these structural aberrations in MM- the commented differences. Alternatively, these gene expression MSCs may have an impact on stromal cell function and, thus, on differences in MSCs may rely on MM–stromal cell interactions, the progression/relapse of the disease still remains to be and subsequently remain present in a permanent fashion in determined. In fact, functional and gene expression differences in vitro MSC cultures in the absence of myeloma cells.14 between MM- and ND-MSCs (in the absence of myeloma cell

Leukemia Genomic proﬁle of multiple myeloma MSCs M Garayoa et al 1525 Gene copy number variation-DUPLICATION 2

1,3537 1.5

1,0321 1,0220 1

Relative quantity 0.5

0 BCMO1

RNA relative expression 7

6 Control patient KCTD8 5 BMP10 4 FBLN5 ITPK1 3 ZNF217 2

1 RNA, X-fold of control patient

Pat #8 Pat #8 Pat #10Pat #13Pat #15 Pat #17 Pat #17Pat #15 Control Pat Control Pat Control Pat Control Pat Control Pat

Figure 6 Confirmation of array-based comparative genomic hybridisation data by real-time PCR-based analyses. (a) Copy number variation assay. Results for two patients (white bars), one of them with a duplication seen in 16q23.2 (bacterial artificial chromosome (BAC) RP11-446C5), which includes the BCMO1 locus. The mean of five normal samples (grey bar) is shown as a reference. (b) Gene expression assay for KCTD8 and BMP10 genes (located in lost DNA regions) and FBLN5, ZNF21, and ITPK1 genes (located at the regions of genomic gain). The expression levels for each target gene were quantified by real-time PCR and presented as the fold increase/decrease in affected patients relative to control patients. Results from control patients were normalized by setting the corresponding value as 1.

Table 6 Genomic imbalances (gains) found on sorted-uncultured MSCs deﬁned by two (or more) contiguous BAC clones

Contiguous Chromosome Putative affected gene(s) Frequency BACs location of gains

RP11-243K18 16p13.3 Rho GDP dissociation inhibitor (GDI) gamma (ARHGDIG)//Protein disulﬁde isomerase family 1/5 A, member 2 (PDIA2)//Axin 1 (AXIN1)//RAB11 family-Interacting protein 3 (RAB11FIP3) RP11-161M6 16p13.3 SRY (sex determining region Y)-box 8 (SRY-BOX 8)//Somatostatin receptor 5 (SSTR5)// Complement C1q tumor necrosis factor-related protein 8 precursor (C1QTNF8) RP11-304L19 16p13.3 Polycystic kidney disease 1(PKD1)//Ras-associated protein RAB-26 (RAB26)//TNF receptor-associated factor 7 (TRAF7)//CASK-interacting protein-1 (CASKIN1)// phosphoglycolate phosphatase (PGP)//Transcription factor E4F (E4F1)//deoxyribonuclease 1-Like 2 (DNASE1L2)//dodecenoyl-CoA-delta isomerase (DCI)//RNA-binding protein S1 (RNPS1) Abbreviations: BAC, bacterial artiﬁcial chromosome; MSC, mesenchymal stem cell.

interactions) do not seem to be directly sustained by the genetic Acknowledgements imbalances found in our study. In addition, we found that none of the MM-MSCs showed the cytogenetic abnormalities present We are thankful to Ms Montserrat Martıń, Sara Gonza´lez, Irene in the myelomatous cells from the same patient. Rodrı´guez, Isabel Isidro, Pilar Hernańdez, Almudena Martıń,

Leukemia Genomic profile of multiple myeloma MSCs M Garayoa et al 1526 Montserrat Hernańdez, Sandra Muntioń and Irene Real for their and normalises indices of bone remodelling in patients with relapsed excellent technical work and assistance. We also thank the multiple myeloma. Br J Haematol 2006; 135: 688–692. Genomic Unit (Centro de Investigacioń del Cańcer, Universidad 18 Huss R, Hong DS, McSweeney PA, Hoy CA, Deeg HJ. de Salamanca-CSIC) for spotting, hybridization and scanning of Differentiation of canine bone marrow cells with hemopoietic characteristics from an adherent stromal cell precursor. Proc Natl CGH arrays. This work was partially supported by grants from Acad Sci USA 1995; 92: 748–752. ‘Proyecto Centro en Red de Medicina Regenerativa y Terapia 19 Huss R, Moosmann S. The co-expression of CD117 (c-kit) and Celular de Castilla y Leoń (Consejerıá de Sanidad JCyL–ISCIII)’, osteocalcin in activated bone marrow stem cells in different the Spanish Myeloma Network Program (RD06/0020/0006), and diseases. Br J Haematol 2002; 118: 305–312. ‘Grupos de Excelencia de Castilla y Leoń’ (Ref. GR33). MG was 20 Zhang W, Knieling G, Vohwinkel G, Martinez T, Kuse R, Hossfeld supported by the ‘Plan Nacional de Investigacioń Cientı´fica, DK et al. Origin of stroma cells in long-term bone marrow cultures Desarrollo e Innovacioń Tecnolo´gica (I þ D þ I)’ and the ISCIII-FIS from patients with acute myeloid leukemia. Ann Hematol 1999; 78: 305–314. (CP 05/0279). AG-G was supported by the ‘Proyecto Centro en 21 Chen W, Houldsworth J, Olshen AB, Nanjangud G, Chaganti S, Red de Medicina Regenerativa y Terapia Celular de Castilla y Venkatraman ES et al. Array comparative genomic hybridization Leoń (Consejerıá de Sanidad JCyL–ISCIII)’. reveals genomic copy number changes associated with outcome in diffuse large B-cell lymphomas. Blood 2006; 107: 2477–2485. 22 Nakao K, Mehta KR, Fridlyand J, Moore DH, Jain AN, Lafuente A References et al. High-resolution analysis of DNA copy number alterations in colorectal cancer by array-based comparative genomic hybridiza- 1 Hideshima T, Bergsagel PL, Kuehl WM, Anderson KC. Advances in tion. Carcinogenesis 2004; 25: 1345–1357. biology of multiple myeloma: clinical applications. Blood 2004; 23 Pollack JR, Sorlie T, Perou CM, Rees CA, Jeffrey SS, Lonning PE 104: 607–618. et al. Microarray analysis reveals a major direct role of 2 San Miguel JF, Mateos MV, Pandiella A. Novel drugs for multiple DNA copy number alteration in the transcriptional program myeloma. Hematology (EHA Educ Program) 2006; 2: 205–211. of human breast tumors. Proc Natl Acad Sci USA 2002; 99: 3 Zhan F, Huang Y, Colla S, Stewart JP, Hanamura I, Gupta S et al. 12963–12968. The molecular classification of multiple myeloma. Blood 2006; 24 Snijders AM, Schmidt BL, Fridlyand J, Dekker N, Pinkel D, Jordan 108: 2020–2028. RC et al. Rare amplicons implicate frequent deregulation of cell 4 Hideshima T, Mitsiades C, Tonon G, Richardson PG, Anderson KC. fate specification pathways in oral squamous cell carcinoma. Understanding multiple myeloma pathogenesis in the bone Oncogene 2005; 24: 4232–4242. marrow to identify new therapeutic targets. Nat Rev Cancer 25 Weiss MM, Kuipers EJ, Postma C, Snijders AM, Siccama I, Pinkel D 2007; 7: 585–598. et al. Genomic profiling of gastric cancer predicts lymph node 5 Mitsiades CS, McMillin DW, Klippel S, Hideshima T, Chauhan D, status and survival. Oncogene 2003; 22: 1872–1879. Richardson PG et al. The role of the bone marrow microenviron- 26 Pelham RJ, Rodgers L, Hall I, Lucito R, Nguyen KC, Navin N et al. ment in the pathophysiology of myeloma and its significance in Identification of alterations in DNA copy number in host stromal the development of more effective therapies. Hematol Oncol Clin cells during tumor progression. Proc Natl Acad Sci USA 2006; 103: North Am 2007; 21: 1007–1034, vii–viii. 19848–19853. 6 Podar K, Richardson PG, Hideshima T, Chauhan D, Anderson KC. 27 Dominici M, Le Blanc K, Mueller I, Slaper-Cortenbach I, Marini F, The malignant clone and the bone-marrow environment. Best Krause D et al. Minimal criteria for defining multipotent Pract Res Clin Haematol 2007; 20: 597–612. mesenchymal stromal cells. The International Society for Cellular 7 Yasui H, Hideshima T, Richardson PG, Anderson KC. Novel Therapy position statement. Cytotherapy 2006; 8: 315–317. therapeutic strategies targeting growth factor signalling cascades in 28 Lopez-Villar O, Garcia JL, Sanchez-Guijo FM, Robledo C, Villaron multiple myeloma. Br J Haematol 2006; 132: 385–397. EM, Hernańdez-Campo P et al. Both expanded and uncultured 8 Esteve FR, Roodman GD. Pathophysiology of myeloma bone mesenchymal stem cells from MDS patients are genomically disease. Best Pract Res Clin Haematol 2007; 20: 613–624. abnormal, showing a specific genetic profile for the 5q- syndrome. 9 Barille-Nion S, Barlogie B, Bataille R, Bergsagel PL, Epstein J, Leukemia 2009, doi:010.1038/leu.2008.1361 [E-pub ahead of Fenton RG et al. Advances in biology and therapy of multiple print]. myeloma. Hematology Am Soc Hematol Educ Program 2003, 29 Herrero J, Al-Shahrour F, Diaz-Uriarte R, Mateos A, Vaquerizas 248–278. JM, Santoyo J et al. GEPAS: a web-based resource for microarray 10 Minguell JJ, Erices A, Conget P. Mesenchymal stem cells. Exp Biol gene expression data analysis. Nucleic Acids Res 2003; 31: Med (Maywood) 2001; 226: 507–520. 3461–3467. 11 Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca 30 Ng G, Huang J, Roberts I, Coleman N. Defining ploidy-specific JD et al. Multilineage potential of adult human mesenchymal stem thresholds in array comparative genomic hybridization to improve cells. Science 1999; 284: 143–147. 12 Arnulf B, Lecourt S, Soulier J, Ternaux B, Lacassagne MN, the sensitivity of detection of single copy alterations in cell lines. Crinquette A et al. Phenotypic and functional characterization of J Mol Diagn 2006; 8: 449–458. bone marrow mesenchymal stem cells derived from patients with 31 Iafrate AJ, Feuk L, Rivera MN, Listewnik ML, Donahoe PK, Qi Y multiple myeloma. Leukemia 2007; 21: 158–163. et al. Detection of large-scale variation in the human genome. Nat 13 Corre J, Mahtouk K, Attal M, Gadelorge M, Huynh A, Fleury- Genet 2004; 36: 949–951. Cappellesso S et al. Bone marrow mesenchymal stem cells are 32 Hernandez JM, Gonzalez MB, Granada I, Gutierrez N, Chillon C, abnormal in multiple myeloma. Leukemia 2007; 21: 1079–1088. Ramos F et al. Detection of inv(16) and t(16;16) by fluorescence in 14 Garderet L, Mazurier C, Chapel A, Ernou I, Boutin L, Holy X et al. situ hybridization in acute myeloid leukemia M4Eo. Haematologica Mesenchymal stem cell abnormalities in patients with multiple 2000; 85: 481–485. myeloma. Leuk Lymphoma 2007; 48: 2032–2041. 33 De Preter K, Speleman F, Combaret V, Lunec J, Laureys G, Eussen 15 Wallace SR, Oken MM, Lunetta KL, Panoskaltsis-Mortari A, BH et al. Quantification of MYCN, DDX1, and NAG gene copy Masellis AM. Abnormalities of bone marrow mesenchymal cells number in neuroblastoma using a real-time quantitative PCR assay. in multiple myeloma patients. Cancer 2001; 91: 1219–1230. Mod Pathol 2002; 15: 159–166. 16 Zdzisinska B, Bojarska-Junak A, Dmoszynska A, Kandefer-Szerszen 34 Bhowmick NA, Chytil A, Plieth D, Gorska AE, Dumont N, M. Abnormal cytokine production by bone marrow stromal cells of Shappell S et al. TGF-beta signaling in fibroblasts modulates the multiple myeloma patients in response to RPMI8226 myeloma oncogenic potential of adjacent epithelia. Science 2004; 303: cells. Arch Immunol Ther Exp (Warsz) 2008; 56: 848–851. 207–221. 35 Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer 2006; 17 Terpos E, Heath DJ, Rahemtulla A, Zervas K, Chantry A, 6: 392–401. Anagnostopoulos A et al. Bortezomib reduces serum dickkopf-1 36 Tlsty TD, Coussens LM. Tumor stroma and regulation of cancer and receptor activator of nuclear factor-kappaB ligand concentrations development. Annu Rev Pathol 2006; 1: 119–150.

Leukemia Genomic profile of multiple myeloma MSCs M Garayoa et al 1527 37 Fukino K, Shen L, Matsumoto S, Morrison CD, Mutter GL, Eng C. cells do not undergo transformation after long-term in vitro culture Combined total genome loss of heterozygosity scan of breast and do not exhibit telomere maintenance mechanisms. Cancer Res cancer stroma and epithelium reveals multiplicity of stromal 2007; 67: 9142–9149. targets. Cancer Res 2004; 64: 7231–7236. 47 Zhang ZX, Guan LX, Zhang K, Wang S, Cao PC, Wang YH et al. 38 Weber F, Shen L, Fukino K, Patocs A, Mutter GL, Caldes T et al. Cytogenetic analysis of human bone marrow-derived mesenchy- Total-genome analysis of BRCA1/2-related invasive carcinomas of mal stem cells passaged in vitro. Cell Biol Int 2007; 31: 645–648. the breast identifies tumor stroma as potential landscaper for 48 Muller I, Vaegler M, Holzwarth C, Tzaribatchev N, Pfister SM, neoplastic initiation. Am J Hum Genet 2006; 78: 961–972. Schutt B et al. Secretion of angiogenic proteins by human 39 Moinfar F, Man YG, Arnould L, Bratthauer GL, Ratschek M, multipotent mesenchymal stromal cells and their clinical potential Tavassoli FA. Concurrent and independent genetic alterations in in the treatment of avascular osteonecrosis. Leukemia 2008; 22: the stromal and epithelial cells of mammary carcinoma: implica- 2054–2061. tions for tumorigenesis. Cancer Res 2000; 60: 2562–2566. 49 Baker DE, Harrison NJ, Maltby E, Smith K, Moore HD, Shaw PJ 40 Condon MS. The role of the stromal microenvironment in prostate et al. Adaptation to culture of human embryonic stem cells and cancer. Semin Cancer Biol 2005; 15: 132–137. oncogenesis in vivo. Nat Biotechnol 2007; 25: 207–215. 41 Hill R, Song Y, Cardiff RD, Van Dyke T. Selective evolution of 50 Lastowska M, Cotterill S, Bown N, Cullinane C, Variend S, Lunec J stromal mesenchyme with p53 loss in response to epithelial et al. Breakpoint position on 17q identifies the most aggressive tumorigenesis. Cell 2005; 123: 1001–1011. neuroblastoma tumors. Genes Chromosomes Cancer 2002; 34: 42 Tuhkanen H, Anttila M, Kosma VM, Heinonen S, Juhola M, 428–436. Helisalmi S et al. Frequent gene dosage alterations in stromal 51 Barlund M, Tirkkonen M, Forozan F, Tanner MM, Kallioniemi O, cells of epithelial ovarian carcinomas. Int J Cancer 2006; 119: Kallioniemi A. Increased copy number at 17q22-q24 by CGH in 1345–1353. breast cancer is due to high-level amplification of two separate 43 Ricci F, Kern SE, Hruban RH, Iacobuzio-Donahue CA. Stromal regions. Genes Chromosomes Cancer 1997; 20: 372–376. responses to carcinomas of the pancreas: juxtatumoral gene 52 Tirkkonen M, Tanner M, Karhu R, Kallioniemi A, Isola J, expression conforms to the infiltrating pattern and not the biologic Kallioniemi OP. Molecular cytogenetics of primary breast cancer subtype. Cancer Biol Ther 2005; 4: 302–307. by CGH. Genes Chromosomes Cancer 1998; 21: 177–184. 44 Weber F, Xu Y, Zhang L, Patocs A, Shen L, Platzer P et al. 53 Kalikin LM, George RA, Keller MP, Bort S, Bowler NS, Law DJ Microenvironmental genomic alterations and clinicopathological et al. An integrated physical and gene map of human distal behavior in head and neck squamous cell carcinoma. JAMA 2007; chromosome 17q24-proximal 17q25 encompassing multiple 297: 187–195. disease loci. Genomics 1999; 57: 36–42. 45 Bernardo ME, Avanzini MA, Perotti C, Cometa AM, Moretta A, 54 Buhring HJ, Battula VL, Treml S, Schewe B, Kanz L, Vogel W. Lenta E et al. Optimization of in vitro expansion of human Novel markers for the prospective isolation of human MSC. multipotent mesenchymal stromal cells for cell-therapy Ann NY Acad Sci 2007; 1106: 262–271. approaches: further insights in the search for a fetal calf serum 55 Carrasco DR, Tonon G, Huang Y, Zhang Y, Sinha R, Feng B et al. substitute. J Cell Physiol 2007; 211: 121–130. High-resolution genomic profiles define distinct clinico-patho- 46 Bernardo ME, Zaffaroni N, Novara F, Cometa AM, Avanzini MA, genetic subgroups of multiple myeloma patients. Cancer Cell Moretta A et al. Human bone marrow derived mesenchymal stem 2006; 9: 313–325.

Supplementary Information accompanies the paper on the Leukemia website (http://www.nature.com/leu)

Leukemia