ORIGINAL ARTICLE

A new xenograft model of myeloma bone disease demonstrating the efficacy of humanmesenchymal stem cells expressing osteoprotegerin by lentiviral gene transfer

N Rabin1, C Kyriakou1, L Coulton2, OM Gallagher2, C Buckle2, R Benjamin1, N Singh3, J Glassford1, T Otsuki4, AC Nathwani1,PI Croucher2 and KL Yong1

1Department of Haematology, University College London, London, UK; 2Academic Unit of Bone Biology, University of Sheffield,Sheffield, UK; 3Department of Histopathology, University College London, London, UK and 4Department of Hygiene, KawasakiMedical School, Okayama, Japan

We describe a new model of myeloma bone disease in whichb2m NOD/SCID mice injected with KMS-12-BM cells developmedullary disease after tail vein administration. Micro-com-puted tomography analysis demonstrated significant bone lossin the tibiae and vertebrae of diseased animals compared tocontrols, with loss of cortical bone (Po0.01), as well astrabecular bone volume, thickness and number (Po0.05 forall). Bone marrow of diseased animals demonstrated anincrease in osteoclasts (Po0.01) and reduction in osteoblasts(Po0.01) compared to control animals. Both bone loss andosteoclast increase correlated with the degree of diseaseinvolvement. Mesenchymal stem cells (MSCs) were lentivirallytransduced to express human osteoprotegerin (hOPG). Sys-temic administration of OPG expressing MSC reduced osteo-clast activation (Po0.01) and trabecular bone loss in thevertebrae (Po0.05) and tibiae of diseased animals, to levelscomparable to non-diseased controls. Because of its predomi-nantly medullary involvement and quantifiable parameters ofbone disease, the KMS-12-BM xenogeneic model providesunique opportunities to test therapies targeted at the bonemarrow microenvironment.Leukemia (2007) 21, 2181–2191; doi:10.1038/sj.leu.2404814;published online 26 July 2007Keywords: myeloma bone disease; murine models;osteoprotegerin; mesenchymal stem cell

Introduction

Multiple myeloma (MM) is a B-cell neoplasm characterized bythe clonal expansion of plasma cells in the bone marrow. Bonedestruction, which is present in over 80% of patients,1 isresponsible for the most debilitating clinical features of thisdisease. Although the mechanisms that lead to the developmentof myeloma bone disease are not fully understood, histomorpho-metric studies have shown uncoupling of the normal process ofbone remodelling. Bone resorption is enhanced due to increasednumbers of osteoclasts, whereas bone formation is reduced.2,3

Increased osteoclast activity is generated through interactionsbetween osteoclast precursors, stromal/osteoblastic cells and themalignant plasma cells. These cellular interactions lead to theproduction of osteoclast-activating factors (OAFs) of which themajor ones are the receptor of nuclear factor-kB ligand (RANKL)and macrophage inflammatory protein 1-a (MIP-1a). RANKL hasbeen shown to play a crucial role in osteoclast development.4

RANKL is expressed by stromal cells and osteoblasts in the local

bone marrow microenvironment, and binds to RANK on thesurface of osteoclast precursors to stimulate osteoclast differ-entiation and activation.5 Osteoprotegerin (OPG) is a naturallyoccurring soluble decoy receptor for RANKL, inhibiting itsinteraction with RANK and preventing osteoclast formation.6

Evidence suggests that the increase in osteoclast activity inmyeloma is due to an imbalance in the ratio of RANKL to OPG.RANKL expression is increased in bone marrow stroma cells,7,8

and OPG is reduced in the bone marrow8 and serum9 of patientswith myeloma. MIP-1a is an OAF produced directly by MM cellsthat induces osteoclast formation in human bone marrowcultures and its expression is elevated in the bone marrow ofMM patients compared to healthy controls.10 In addition to theincrease in osteoclast activity, bone loss is exacerbated bysuppression of osteoblastic activity. Biological mechanismsinvolved in osteoblast inhibition by MM cells include modula-tion of Runx2/Cbfa1,11 inhibition of Wnt signalling12 andcytokines such as interleukin-3 (IL-3) and IL-7.13

The pathways that lead to increased osteoclast activity andosteoblast suppression present attractive therapeutic targets.RANKL blockade, achieved with recombinant OPG14 or RANK-Fc7 in murine models of osteolytic bone disease inhibitedosteoclastogenesis and prevented bone loss. Similarly, anantisense construct targeting MIP-1a reduced bone loss in axenogeneic model of osteolytic bone disease.15 These resultsprovide compelling evidence for the involvement of thesepathways in vivo. The successful development of such therapiestargeting mechanisms of bone disease requires reproducible andaccessible in vivo models of medullary myeloma, which arecurrently lacking. Syngeneic models, such as the 5T2 murinemodel, are useful models of medullary MM, but are not widelyavailable. Xenogeneic models employing human myeloma celllines cultured in vitro and injected systemically into immuno-deficient mice frequently develop extramedullary disease,16–19

which makes assessment of therapies targeted to the bonemarrow microenvironment difficult. The SCID/Hu model,20

where human fetal bones (or rabbit bones in the SCID-Rabmodel21) provide the bone marrow microenvironment forprimary myeloma cells, is clearly useful but does not reflectthe widespread anatomical nature of this disease or lend itself toinvestigations of tumour cell homing or migration.

With these issues in mind we aimed to set up and characterizea new model of medullary MM with associated bone disease,using immunohistochemistry, histomorphometry and micro-computed tomography (micro-CT). We have used this modelto test the efficacy of gene-modified human mesenchymal stemcells (hMSCs) overexpressing OPG. We have shown thatenforced expression of OPG in this xenogeneic model reversesosteoclast activation and trabecular bone loss.

Received 10 May 2007; accepted 15 May 2007; published online 26July 2007

Correspondence: Dr Kwee L Yong, Department of Haematology,University College London, 98 Chenies Mews, London, UK.E-mail: kwee.yong@ucl.ac.uk

Leukemia (2007) 21, 2181–2191& 2007 Nature Publishing Group All rights reserved 0887-6924/07 $30.00

www.nature.com/leu

Materials and methods

Cell linesKMS-12-BM was provided by Dr T Otsuki (Kawasaki MedicalSchool, Okoyama, Japan). All other cell lines were purchasedfrom (ATCC, Teddington, Middlesex, UK). KMS-12-BM weremaintained in culture in RPMI 1640 supplemented with 10%fetal calf serum (FCS), 100 U/ml penicillin, and 100 mg/mlstreptomycin. HeLa and 293 T-cell lines were maintained inculture in Dulbecco’s modified Eagle’s medium (DMEM)supplemented with 10% FCS, 100 U ml penicillin, and100 mg/ml streptomycin.

Xenogeneic myeloma modelAll procedures involving the use and care of animals wereperformed in accordance with the Animals Scientific ProceduresAct (1986), were approved by the Institutional Animal Care Useand licensed by the Home Office. All experiments involvedunconditioned 6- to 8-week-old b2m NOD/SCID mice, housedunder positive pressure in individually ventilated cages (Tecni-plast, Northamptonshire, UK).

Subcutaneous tumours were established after injection of1� 107 KMS-12-BM, and tumour measurements performed intwo dimensions with callipers at serial time points. Insubsequent experiments mice received either tumour (1� 106

or 1� 107 KMS-12-BM cells), or vehicle control (phosphate-buffered saline) by tail vein injection. Animals were killed whenthey developed hind limb paralysis or behavioural changesnecessitating killing. To assess the effect of gene-modified MSCin this tumour model, 1� 106 MSCOPG, MSCGPO, unmodifiedMSC or vehicle control, were injected by tail vein 2, 3 and 4weeks after administration of 1� 107 KMS-12-BM. These micewere killed at 6 weeks and the tibiae and lumbar vertebrae wereobtained for analysis.

Histological and histochemical analysisTibiae, lumbar vertebrae and visceral organs were fixed informalin. Tibiae and lumbar vertebrae were decalcified in EDTAand embedded in paraffin, and 3mm sections were cut andstained with haematoxylin and eosin. Immunohistochemistry todetect KMS-12-BM in vivo was performed using VS38C (Dako,Ely, Cambridgeshire, UK) on paraffin-embedded sections.VS38C is a mouse anti-human monoclonal antibody that detectsa transcription factor produced by normal and malignant plasmacells.22 Briefly, following blockade of endogenous peroxidase,VS38C primary antibody was applied (1 in 100 dilution),followed by secondary antibody visualized with HRP/DABsystem (Dako, UK). Tumour infiltration in the marrow space ofthe tibiae and vertebrae was graded as the proportion ofpositively stained cells to total number of cells as follows;greater than 70% involved (þ þ þ ), between 30 and 69%involved (þ þ ), and less than 30% (þ ) or not involved (�).Trabecular bone area (Tb.Ar) as a proportion of the total areawas determined in the tibia and lumbar vertebrae, in an area of0.525 mm2 and 0.2 mm from the growth plate, using anOsteomeasure analysis system (OsteoMetrics, Decatur, GA,USA). Sections were also stained for the presence of tartrate-resistant acid phosphatase (TRAP) to identify osteoclasts andcounterstained with Gill haematoxylin. Because bonemarrow infiltration by KMS-12-BM resulted in loss of trabecularbone, only osteoclasts lining the endocortical bone surfaceswere assessed. Endocortical surfaces were analysed for osteo-clastic surface (mm) over total endocortical bone surface (mm)

and expressed as a percentage, for 3 mm (tibia) or 2 mm(vertebrae) of endocortical surface beginning 0.2 mm from thegrowth plate. Osteoblasts were identified as cuboidal cells liningthe endocortical surface in sections of tibiae stained withhaematoxylin and eosin, and analysed as for osteoclasts (seeabove).

Micro-CTTibiae and lumbar vertebrae were fixed in formalin, andanalysed at a pixel size of 4.3 mm using the Skyscan model1172 and associated analysis software (Skyscan, Aartselaar,Belgium). During scanning, the samples were enclosed in atightly fitting rigid plastic tube to prevent movement. X-raysource was set to 50 kV, 100mA with a 0.5 aluminium filter andthe sample rotated through 1801 using a 0.71 rotation step.Regions of interest were defined as trabecular or cortical bonethat were 0.2 mm from the growth plate and extending a further1 mm. Thresholding was then applied to the images to segmentthe bone from the background, and the same threshold settingwas used for all samples. The following three-dimensional (3D)parameters were measured: volume, surface and surface-to-volume ratio. To calculate these, the software builds a 3Dcuboidal voxel model of bone. The following architecturalmeasurements were then made: trabecular thickness, numberand separation.

Isolation and characterization of human mesenchymalstem cellsHMSCs were obtained from the bone marrow of healthy donors,undergoing bone marrow harvest after informed consentaccording to the institutional guidelines approved by the localethics committee. Mononuclear cells obtained by centrifugationover Ficoll-Paque gradient (Amersham Biosciences, Piscataway,NJ, USA) were plated (2� 105 cells/cm2) in Mesencult CultureMedium (Stem Cell Technologies, Batiment, Grenoble, France).Non-adherent cells were removed after 48 h and adherent cellscultured until they reached confluence. Cells were trypsinized,and subcultured at densities of 1� 104/cm2. Adipocyticdifferentiation of MSC was induced by culture in DMEMsupplemented by 10% FCS, 1 mM dexamethasone, 10mg/mlinsulin, 100 nM indomethacin and 0.5 mM isobutylmethyl-xanthine (all from Sigma-Aldrich, Gillingham, Dorset, UK) for7–10 days, after which adipocyte differentiation was confirmedusing direct staining of lipid formation with Nile Red.Osteogenic differentiation of MSC was induced by culture inDMEM supplemented by 10% FCS, dexamethasone, b-glycer-ophosphate and ascorbic acid (all from Sigma-Aldrich, UK) for14–21 days. Osteogenic differentiation was confirmed bystaining for alkaline phosphatase.

Production of SIN Lentiviral vectorscDNA for human OPG (hOPG) was originally isolated fromMG63 cells, and its sequence corresponds to NM002546(National Center for Biotechnology Information; C Buckle,unpublished). The cDNA for hOPG was cloned into thepCL10.1 vector in either a forward (producing OPG) or reverse(control) direction upstream of the gene for enhanced greenfluorescent protein (eGFP), both under the direction of themurine stem cell virus promoter. Third generation self-inactivat-ing lentivirus were produced as described previously.23 Briefly,293 T cells were transfected with four plasmids containing (1)the lentiviral gag-pol genes, (2) the vesicular stomatitis virus

A new xenograft model of myeloma bone diseaseN Rabin et al

2182

Leukemia

glycoprotein envelope gene, (3) the reverse transcriptase, and (4)a bicistronic vector (pCL10.1) as described. Biological titres ofthe supernatant were determined by limiting dilution on HeLacells.

Lentiviral transduction of MSCMSCs at passages 2–4 were plated at a density of 5� 104 cellsper well in a six-well plate. For each of the following 5 days,cells were exposed to viral supernatant at a multiplicity ofinfection of 2 for 6 h in the presence of polybrene (8 mg/ml),generating MSCOPG (cloned in forward direction, expressinghOPG) or MSCGPO (control). Transduced MSCs were expandedand used for in vivo experiments. Transduction efficiency wasdetermined using fluorescence-activated cell sorting (FACS)analysis for GFP expression. To detect hOPG expression in vitro,conditioned media from MSC cultured in X-VIVO-10 for 72 hwas tested in an OPG enzyme-linked immunosorbent assay(ELISA; R&D Systems, Abingdon, Oxfordshire, UK), according tothe manufacturer’s protocol. In vivo expression of hOPG wasdetermined using a similar ELISA on mouse serum collected atvarious times after tail vein injection of MSCOPG or MSCGPO

(3� 106/animal) or vehicle control.

Analysis of short-term homing of MSC in vivoShort-term homing experiments using PKH26 (PKH)-labelledcells have previously been described.24 Briefly, MSCs werelabelled with 8mM PKH-26, injected at 2� 106 per animal bytail vein administration, and murine tissues analysed 24 h laterfor the presence of PKH-24-positive cells by FACS analysis ofsingle-cell suspensions. At least 100 000 events were collectedon the Epics-Elite flow cytometer (Beckman-Coulter, HighWycombe, UK).

Statistical analysisComparison between groups was performed by the Mann–Whitney U test.

Results

Engraftment of KMS-12-BM in b2m NOD/SCID miceThe MM cell line KMS-12-BM was originally isolated from thebone marrow of a patient with non-secretory MM.25 It isassociated with specific cytogenetic abnormalitiest(11:14)(q13:q32), which were confirmed (data not shown). Toinvestigate the in vivo tumorigenicity of KMS-12-BM cells and tooptimize the immunohistochemical detection of these cellsin vivo, 1� 107 cells in culture were injected subcutaneouslyinto the flank of unconditioned b2m NOD/SCID mice. Animalsdeveloped palpable solid tumours after 2 weeks. Immuno-histochemistry performed on these solid tumours revealed thatthey expressed CD38, CD138, Pax-5, Mum-1 and VS38C, butnot kappa, lambda or CD20 (data not shown). We subsequentlyused VS38C immunohistochemistry to assess tumour infiltrationin diseased animals.

To set up and characterize the xenogeneic myeloma modelusing KMS-12-BM, 1� 106 or 1� 107 cells were injected by tailvein into unconditioned b2m NOD/SCID mice. Animals werekilled between 5 and 12 weeks, with the onset of hind limbparalysis and behavioural changes such as reduced spontaneousactivity and changes in grooming. More rapid tumour take andshorter survival was observed in those mice treated with a highercell dose (Figure 1A). A cell dose of 1� 107 KMS-12-BM was

used in subsequent experiments. Immunohistochemistry withVS38C in diseased animals demonstrated tumour in the bonemarrow of the long bones (Figure 1C) and vertebrae (data notshown), although the extent of infiltration varied betweenanimals (Table 1). FACS analysis of bone marrow mononuclearcells collected from diseased animals demonstrated the pre-sence of CD138-positive cells (Figure 1B). Importantly extra-medullary disease was unusual, with liver being the only othersite of tumour infiltration (small foci of disease in two of fiveanimals), and no disease seen in any other organs, including thespleen and kidney (Figure 1C). Occasional tumour cells are seenin the lung, which reflects cells being trapped in the pulmonaryvasculature, as there was no evidence of tumour formation(Figure 1C).

To determine the kinetics of disease take, a series of animalswere killed at different time points following tail vein admin-istration of 1� 107 KMS-12-BM. A small focus of tumour wasfirst apparent in the tibiae after 3 weeks (data not shown), withincreasing bone marrow involvement thereafter, and almostcomplete effacement of the tibial bone marrow at 6 weeks. Thistime course of disease development was consistent betweenexperiments.

Cortical and trabecular bone is lost in the KMS-12-BMxenogeneic MM modelMicro-CT has allowed detailed assessment of bone disease inmalignant and non-malignant animal models of bone disease.We used this technique to examine bone loss in the tibiae andvertebrae 6 weeks after injection of 1� 107 KMS-12-BM incomparison with non-diseased controls. Cortical bone wasreduced in the tibiae of diseased animals (median corticalthickness 69.9mm, range 65.8–78.6 vs control median 84.5 mm,range 75.9–90.6, Po0.01) with evidence of lytic lesions(Figure 2a). Trabecular bone volume (TBV) was similarlyreduced in the tibiae of diseased animals (median %TBV 2.9,range 0.9–5.0, vs control median 6.4, range 3.5–10.8, Po0.05),as were trabecular thickness and number (Table 2). To confirmthat bone loss is caused by tumour infiltration, and becausetumour take is variable, we analysed separately those animalswith high (470%) and low (o70%) tumour infiltration based onVS38C immunohistochemistry. Loss of trabecular bone isrelated to the amount of tumour infiltration, with the moststriking loss seen in animals with heavy disease burden(Figure 2b). This suggests that the presence of tumour cells isdirectly related to bone loss. Interestingly, animals withintermediate disease burden also showed loss of trabecularbone.

Micro-CT analysis demonstrated extensive bone loss in thelumbar vertebrae. Trabecular bone volume was significantlyreduced in the vertebrae of diseased animals (median %TBV9.8, range 8.1–11.8, vs control median 13.1, range 11.1–14.1,Po0.01). In addition, the vertebrae of diseased animalsdemonstrate significant reduction of trabecular thickness andnumber (Table 3). As seen in the tibiae, the magnitude of boneloss varied directly with the degree of tumour infiltration(Figure 2c).

To further investigate bone loss in this model, histomorpho-metric analysis was performed. The tibiae of diseased animalsdemonstrated a reduction in trabecular bone area (Tb.Ar)(median %Tb.Ar 2.4, range 0.4–6.1 vs control median 4.6,range 1.5–8.9, Po0.05), with representative animals shown inFigure 2d. In keeping with micro-CT analysis, we found that thereduction in trabecular bone was most marked in tibiae withhigh tumour burden (470%) as revealed by VS38C immuno-

A new xenograft model of myeloma bone diseaseN Rabin et al

2183

Leukemia

histochemistry (median %Tb.Ar 0.7, range 0.1–1.1, Po0.01compared to control), whereas tibiae without significant tumourinvolvement had similar Tb.Ar to controls (median 5.1, range2.7–6.1, Figure 2e). There was no significant loss of trabecularbone area in the lumbar vertebrae of diseased animals byhistomorphometry (%Tb.Ar 10.3, range 7.3–12.5; vs 11.1, range7.3–15.3 in control mice, NS, Figure 2f), in contrast to themicro-CT results.

Bone disease in the KMS-12-BM xenogeneic MM modelis mediated by an increase in osteoclast numbers andreduced osteoblastsBone disease in MM is considered to be due to an uncoupling ofthe normal process of bone remodelling, with an increase in

osteoclasts and reduction in osteoblast numbers. To investigatethe cellular basis for bone loss in our model, we quantifiedosteoclast and osteoblast numbers in diseased animals and innon-diseased controls.

Diseased animals demonstrated a marked increase inosteoclasts in the lumbar vertebrae (median %OcPm 8.1, range3.1–15.9; vs control median 0.9, range 0.2–1.8, Po0.01,Figure 3a). This increase in osteoclasts was also apparent inthe tibiae (diseased animals median %OcPm 6.9, range 2.2–20.1; control median 0.6, range 0.0–2.1, Po0.05, Figure 3a). Inkeeping with trabecular bone loss, the increase in osteoclastperimeter was most marked in bones with heavy (470%)tumour infiltration (Figure 3b). There was a concomitantreduction in osteoblast numbers in diseased animals (median%ObPm 32.5, range 8.8–64.4; vs control median 77.8, range

Table 1 Infiltration of tibiae and vertebrae of b2m NOD/SCID mice injected with 1� 107 KMS-12-BM, and analysed after 6 weeks

Animal Right tibia(tumour)

Left tibia(tumour)

First lumbarvertebra (tumour)

Liver(tumour)

Spleen(tumour)

1 + + � � �2 +++ + + + �3 +++ +++ +++ � �4 +++ +++ +++ + �5 ++ + + � �

Variable infiltration of tumour cells in tibiae and vertebrae as assessed by VS38C immunohistochemistry. Grading of infiltration as follows; +++greater than 70% tumour; ++ between 30 and 69%; + less than 30%, � not involved. One representative experiment.

0.0 2.5 5.0 7.5 10.0 12.50

25

50

75

1001 x 107

1 x 106

Time (weeks)

Side scatter

CD

138p

e

Control Diseased

Per

cen

t su

rviv

al

A C

B

1000

10000.1

0.1

PE

1000

0.1

PE

SS10000.1

SS

F F

G: AB G: AB

a b

c d

e f

g h

Figure 1 Engraftment of KMS-12-BM cells in b2m NOD/SCID mice. (A) Survival of animals injected with 1� 106 (broken line) or 1� 107 cells(solid line) by tail vein. (B) FACS analysis of bone marrow mononuclear cells collected from control and diseased animals after staining for CD138.(C) Bone marrow stained with haematoxylin and eosin (Hþ E) or VS38C immunohistochemistry (�60) demonstrating infiltration at 6 weeks (c, d).Control bone marrow shown (a, b). Kidney (e) and spleen (f) show no disease infiltration (VS38C immunostaining, � 10). Small focus of tumourseen in liver (arrowed) (g), and occasional cells present in the lung (h) (VS38C immunostaining, � 60). FACS, fluorescence-activated cell sorting.

A new xenograft model of myeloma bone diseaseN Rabin et al

2184

Leukemia

Control Diseased

3D reconstruction

2D cross-section

0.0

2.5

5.0

7.5

10.0

0.0

0.5

1.0

1.5

2.0

30

40

50

60

70

80

% T

RA

BE

CU

LA

R B

ON

EV

OL

UM

E%

TR

AB

EC

UL

AR

BO

NE

VO

LU

ME

TR

AB

EC

UL

AR

TH

ICK

NE

SS

(MIC

RO

NS

)T

RA

BE

CU

LA

R T

HIC

KN

ES

S(M

ICR

ON

S)

TR

AB

EC

UL

AR

NU

MB

ER

(1/m

m)

TR

AB

EC

UL

AR

NU

MB

ER

(1/m

m)

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

a

b

NS NSp < 0.01

5.0

7.5

10.0

12.5

15.0

30

35

40

45

50

1.5

2.0

2.5

3.0

3.5

4.0

0.0

2.5

5.0

7.5

10.0

0

5

10

15

20

% T

RA

BE

CU

LA

R B

ON

EA

RE

A

% T

RA

BE

CU

LA

R B

ON

EA

RE

A

CONTROL

CONTROL

DISEASED

< 70

% TUMOUR

> 70

% TUMOUR

c

e f

Control

Diseased

d

Figure 2 KMS-12-BM model demonstrates loss of cortical and trabecular bone. (a) Micro-CT – 3D reconstruction and two-dimensional cross-section (200mm from growth plate) of control and diseased tibiae, demonstrating loss of cortical and trabecular bone. (b) Micro-CT analysis oftibiae – trabecular bone volume, thickness and number in animals with low (o70%) and high (470%) tumour infiltration. Control tibiae shown.(c) Micro-CT analysis of vertebrae – trabecular bone volume, thickness and number in animals with low (o70%) and high (470%) tumourinfiltration. Control vertebrae shown. (d) Hþ E stain of control and diseased tibiae (� 10) in representative animals. Presence of trabecular bone ina control animal (arrow) compared to its absence in a diseased animal. (e) Histomorphometric analysis of tibiae – loss of trabecular bone areaoccurs only in animals with 470% tumour infiltration on VS38C staining. Control tibiae shown. (f) Histomorphometric analysis of vertebrae doesnot demonstrate loss of trabecular bone. All graphs show data from individual animals with medians indicated. Micro-CT, micro-computedtomography.

A new xenograft model of myeloma bone diseaseN Rabin et al

2185

Leukemia

68.0–84.8, Po0.01, Figure 3c). Once again the most strikingloss of osteoblasts was seen in animals with significant tumourburden (Figure 3d). These observations suggest that theimbalanced increase in osteoclasts and reduction in osteoblastsis directly related to the extent of tumour infiltration.

Lentiviral transduction of hMSCNext, we investigated whether this model could be used to testnovel therapies that target myeloma bone disease. Human bonemarrow-derived MSCs were isolated by standard means andcharacterized by the ability to differentiate into osteoblasts andadipocytes, and the retention of a stable immunophenotype(SH2þ , SH4þ , CD13þ , CD49eþ , CD34�, CD45�) for up to14 passages. Using conditions we have previously optimized,26

early passage2–4 hMSCs were transduced with a bicistroniclentiviral construct containing cDNA encoding OPG and eGFP(MSCOPG, Figure 4a). Control MSCs were derived from the samebone marrow donor, and transduced with a bicistronic lentiviralconstruct with the cDNA for OPG cloned in reverse (MSCGPO,Figure 4a). Efficient transduction was demonstrated by high GFPexpression in transduced MSC (480%). Stable transgeneexpression was confirmed by detection of OPG in culturesupernatants of MSCOPG for up to 16 passages, as well as eGFPexpression by flow cytometry. There was no difference in OPGexpression detected in vitro between early and late passagegene-modified MSC (Figure 4b). MSCGPO expressed low levelsof hOPG in culture at similar levels to untransduced MSC (datanot shown). MSC transduced with lentiviral constructs retaintheir differentiating capacity and phenotype as reportedpreviously,26 with no significant differences seen betweenMSCOPG, MSCGPO and untransduced MSC (data not shown).

In vivo behaviour of gene-modified hMSCTo determine transgene expression in vivo, 3� 106 MSCOPG orMSCGPO were injected by tail vein into unconditioned b2mNOD/SCID mice. hOPG was detected in mouse serum collected

2 weeks after injection in the MSCOPG-treated group only(median 2 ng/ml, range 1.6–2.4) and levels persisted for up to 4weeks post-administration of gene-modified hMSC. hOPG wasconsistently undetectable in mice injected with MSCGPO or incontrol uninjected animals. We investigated the short-termhoming behaviour of gene-modified hMSC in b2m NOD/SCIDanimals and whether this was altered in the presence of tumour.We have previously demonstrated that transduction of MSCdoes not alter their in vivo short-term homing behaviour(Kyriakou et al., Blood 2006; 108: abstract 3182). To examinethe short-term localization of hMSC in diseased animals, 2� 106

PKH-labelled MSCs were administered by tail vein 6 weeks afterthe systemic administration of KMS-12-BM (107/animal). Ani-mals were killed 24 h later and tissues examined for PKH-positive cells by flow cytometry. There was a significant increasein hMSC homing to bone marrow of diseased animals (782positive events per 105 cells, range 476–1254) when comparedto non-diseased controls (417, range 238–818, Po0.05, Figures4c and d), whereas homing to spleen and lungs was similarbetween diseased and control animals (data not shown).

Gene-modified MSCs expressing OPG reverseosteoclast activation and reduce bone loss in theKMS-12-BM xenogeneic modelTo determine the effect of enforced OPG expression onosteoclast activation and bone loss in this model, 1� 106

MSCOPG, MSCGPO or unmodified MSCs were injected by tailvein into b2m NOD/SCID mice 2, 3 and 4 weeks afteradministration of 1� 107 KMS-12-BM (MM cells). Anothergroup of mice were injected with 1� 107 KMS-12-BM (that is,tumour alone) or vehicle control by tail vein injection. All micewere culled at 6 weeks. There was no significant difference intumour take between those groups that did and did not receiveMSC.

As demonstrated previously mice injected with MM cellsalone showed an increase in osteoclasts in the lumbar vertebrae(median %OcPm 8.4, range 3.2–9.5; control median 2.0, range

Table 2 KMS-12-BM model demonstrates loss of trabecular bone in the tibiae

Control (median, range) Diseased (median, range)

Bone volume (BV/TV) % 6.37 (3.45–10.80) 2.891 (0.89–5.03)Trabecular thickness mm 61.48 (53.76–67.72) 44.351 (32.28–58.07)Trabecular number 1/mm 1.03 (0.51–1.79) 0.621 (0.27–0.98)Trabecular separation mm 0.37 (0.31–0.41) 0.37 (0.31–0.49)

Diseased vs control.1Po0.05.Trabecular bone volume, thickness and number and trabecular separation in animals injected with KMS-12-BM and non-diseased control animals,analysed by micro-computed tomography. Data given for all animals, with median and ranges indicated.

Table 3 KMS-12-BM model demonstrates loss of trabecular bone in the vertebrae

Control (median, range) Diseased (median, range)

Bone volume (BV/TV) % 13.06 (11.12–14.08) 9.841 (8.07–11.79)Trabecular thickness mm 42.02 (40.09–44.61) 37.481 (35.52–40.88)Trabecular number 1/mm 3.12 (2.49–3.51) 2.611 (2.27–3.08)Trabecular separation mm 0.239 (0.22–0.26) 0.235 (0.22–0.24)

Diseased vs control.1Po0.05.Trabecular bone volume, thickness and number and trabecular separation in animals injected with KMS-12-BM and non-diseased control animals,analysed by micro-computed tomography. Data given for all animals, with median and ranges indicated.

A new xenograft model of myeloma bone diseaseN Rabin et al

2186

Leukemia

2.0–2.7, Po0.01, Figure 5a). The administration of MSCGPO orunmodified MSC (control MSC) had no effect on osteoclastnumbers in diseased animals (median %OcPm 6.2, range1.6–10.4, NS, Figure 5a), when compared to animals that hadreceived MM cells but no MSC. In contrast osteoclast numbersin the group of diseased animals that received MSCOPG weresimilar to those in non-diseased animals (median %OcPm 2.7,range 0.2–5.5) and significantly lower when compared with thecontrol groups of diseased animals (Po0.01, Figure 5a). Therewas a similar effect on osteoclast numbers in the tibiae wherethe administration of MSCOPG reduced osteoclast numbers(median 3.2, range 0.0–6.7), when compared with the otherdiseased animals (median 5.0, range 0.3–11.0) although theselevels remained higher than those in non-diseased controls(Figure 5b).

Next, we investigated whether reducing osteoclast numbersusing this strategy was able to prevent bone loss in diseasedanimals. As expected animals that received MM cells alone or incombination with control MSC demonstrated loss of TBV in thevertebrae (median %TBV 8.7, range 7.2–10.4 and 9.5, range

7.3–11.1, respectively) when compared to control non-diseasedanimals (median 11.1, range 9.7–12.9, Po0.01 and Po0.05,respectively, Figure 5c). Administration of control MSC had noeffect on TBV in diseased animals. In contrast animals thatreceived MSCOPG had TBV that were comparable to non-diseased animals, and significantly higher than the TBV in theother two groups of diseased animals (Po0.05). Similar effectswere seen on trabecular number (Figure 5e) and thickness(Figure 5g) in the vertebrae.

Similar effects on bone loss were seen in the tibiae of diseasedanimals. While animals injected with MM cells alone, or incombination with control MSC demonstrated significant loss ofTBV (Figure 5d), animals that received MM cells with MSCOPG

had TBV levels that were comparable to those of non-diseasedcontrols. Similar effects were seen for trabecular number(Figure 5f) and thickness (Figure 5h), although the latterremained below the levels in non-diseased controls.

Thus in vivo administration of hMSC expressing OPG bylentiviral gene transfer is able to reverse osteoclast activationand reduce bone loss in this KMS-12-BM xenogeneic model.

p < 0.01 p < 0.01 p < 0.01

Control

Diseased

Vertebrae Tibiae

Vertebrae Tibiae

0

5

10

15

20

0

5

10

15

20

25

0

25

50

75

100

a c

b

% O

ST

EO

CL

AS

T P

ER

IME

TE

R

% O

ST

EO

CL

AS

T P

ER

IME

TE

R

% O

ST

EO

CL

AS

T P

ER

IME

TE

R

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

CONTROL

< 70%

TUMOUR

> 70%

TUMOUR

d

Figure 3 Changes in osteoclast and osteoblast numbers in the KMS-12-BM model. (a) TRAP stain of control and diseased vertebrae and tibiaeshown (� 60), in representative animals. Increase in TRAP-positive cells (osteoclasts) in diseased animals (arrow) compared with control.(b) %OcPm in vertebrae and tibiae of diseased animals according to the degree of tumour infiltration. (c) Hþ E stain of control and diseased tibiaeshown (� 60), in representative animals. Cuboidal cells lining the endocortical surface (osteoblasts) are clearly visible in control tibia (arrow), andreduced in a diseased animal. (d) %ObPm in tibiae of diseased animals according to the degree of tumour infiltration. All graphs show data fromindividual animals with medians indicated. TRAP, tartrate-resistant acid phosphatase.

A new xenograft model of myeloma bone diseaseN Rabin et al

2187

Leukemia

Discussion

Here we have described an in vivo model of medullarymyeloma with associated bone disease that is consistentlydemonstrable and quantifiable by micro-CT and histology.There is clear evidence of osteoclast activation and osteoblastsuppression, which are directly related to the level of bonemarrow infiltration. Importantly, the use of OPG expressinghMSC was able to reverse osteoclast activation and restore bonevolume in diseased animals.

Current in vivo models of myeloma have contributed greatlyto our understanding of the cellular and molecular bases forbone loss, however, there is a lack of readily reproduciblemedullary models of human myeloma which can be used to teststrategies aimed at bone disease and the bone marrowmicroenvironment. Immunodeficient models employing myelo-ma cell lines cultured in vitro frequently develop extramedullarydisease,16–19 whereas the SCID-hu20 and SCID-rab21 models donot reflect the systemic nature of the disease. The myeloma cellline KMS-12-BM was established from a patient with non-

secretory MM and has not previously been described in ananimal model. These cells engraft in unconditioned b2m NOD/SCID mice with the development of local tumours whendelivered subcutaneously. Importantly, tail vein administrationleads to engraftment within the bone marrow, with little or noextramedullary involvement. The b2m NOD/SCID mouse strain,which lack mature lymphocytes, serum immunoglobulin andnatural killer (NK) cell activity allows engraftment of xenogeneiccells27 and KMS-12-BM cells engraft without need for con-ditioning. Engraftment occurs with reproducible kinetics over a6-week period (at a cell dose of 107/animal), thus allowingconsistent timing of interventions.

An important feature of the KMS-12-BM xenogeneic model isthe presence of osteolytic bone disease, manifested by loss ofcortical and trabecular bone in tibiae and trabecular bone in thevertebrae. Importantly, bone loss varies directly with extent ofdisease infiltration, and is associated both with an increase inosteoclasts, and a reduction in osteoblasts. The confinement ofdisease to medullary sites, even until the terminal phases,indicates that this tumour model is dependent upon the bone

- OPG in forward direction

- OPG in reverse direction

p < 0.05

Side scatter

PK

H

Control

Negative control Test

Diseased

Control Diseased

0

500

1000

1500

0.0

2.5

5.0

7.5

10.0

12.5

15.0

17.5MSC - OPGMSC - GPO

Passage number of hMSC

a

hu

man

OP

G (

ng

/mL

)

<10

<10

>10

>10

bP

KH

po

siti

ve c

ells

/ 105

cells

an

alys

ed

d

104

104

103

103

102

102

101

101100

100

PK

H L

og

104

103

102

101

100

PK

H L

og

104

103

102

101

100

PK

H L

og

104

103

102

101

100

PK

H L

og

SS Log

104103102101100

SS Log

104103102101100

SS Log

104103102101100

SS Log

Chimeric5' LTR

Chimeric5' LTR

Chimeric3' LTR

Chimeric3' LTR

∆gag

∆gag

∆pol

∆pol

RRE

RRE

MSCV

MSCV

IRES

IRES

eGFP

eGFP

OPG

GPO

cR3

R3 R3

R3

Figure 4 Lentiviral transduction of MSC to express OPG. (a) Structure of lentiviral construct in forward (MSCOPG) or reverse (MSCGPO) direction.(b) In vitro expression of hOPG by MSCOPG or MSCGPO. Supernatant collected from 5� 104 MSC in X-VIVO-10 over 72 h, and analysed by ELISA.Mean7s.e.m. shown. (c) Identification of PKH-stained MSC in the bone marrow 24 h post infusion. Cell suspensions from the bone marrow ofKMS-12-BM-injected animals (diseased) and control (non-diseased) animals were analysed by flow cytometry for the presence of PKH bright cells.Negative control animals were infused with saline only. Flow cytometric plots from a representative animal in each group. (d) Short-termlocalization of PKH-labelled MSC in bone marrow of control and diseased animals. Data shown is the number of PKH bright cells/105, from allanimals. MSC, mesenchymal stem cell; OPG, osteoprotegerin.

A new xenograft model of myeloma bone diseaseN Rabin et al

2188

Leukemia

marrow microenvironment and would be useful and appropriateto test strategies directed at this environment. This model alsoclosely resembles the clinical disease, where extensive bonedestruction is commonly associated with medullary forms ofmyeloma, in contrast to extramedullary myeloma where lyticbone disease is less frequent. Furthermore, bone loss isquantifiable, both by histomorphometry and by micro-CT. Useof micro-CT in this model will allow us to study the processesand mechanisms of bone loss at different stages of the disease.Interestingly, bone loss in the lumbar vertebrae was detected bymicro-CT but not by histomorphometry, suggesting that micro-CT maybe a more sensitive tool. Certainly, this model affords adetailed and comprehensive analysis of both cortical andtrabecular bone and will be a valuable tool to study theprocesses and mechanisms of bone loss in MM.

Blockade of RANKL with OPG or recombinant RANK hasbeen demonstrated in animal models of myeloma to reduce theassociated bone disease and, in some reports, to increasesurvival.7,14,28,29 In patients with myeloma and breast cancer-related bone metastases, bone resorption was reduced byrecombinant OPG treatment, an effect that was comparable toa single injection of a bisphosphonate.30 Recombinant OPGwhen administered systemically has a half-life of several days,thus to obtain a clinically significant response would requiremultiple injections over a period of time. In a xenogeneic modelof osteolytic bone disease, Doran et al.31 directly transduced theARH-77 cell line to express hOPG and demonstrated areduction in bone disease. Although providing proof ofprinciple, this strategy is not suitable for clinical application.Genetically modified MSC provide a novel way of delivering

OPG, with the potential for sustained expression. Our resultsconfirm the efficacy of this approach. In addition, the reductionin osteoclast numbers, and attenuation of bone loss achieved bythe use of MSCOPG provide direct evidence that the OPG/RANKL axis is important in this model of MM bone disease.

We timed our intervention to coincide with engraftment ofKMS-12-BM in the bone marrow, with weekly tail veininjections of gene-modified MSC from 2 weeks onward. Whileour results are encouraging and provide proof of principle, thereis scope for further optimization of the treatment schedule.Trabecular bone may have been lost (for example, from thetibiae) before the intervention was started, and earlier timing oftreatment may enhance the therapeutic effect. Additionallylevels of OPG may not be high enough to completely blockosteoclast activation. In an in vitro co-culture system,6 hOPGinhibited osteoclast differentiation by 50% at a concentration of1 ng/ml (similar to levels we have detected in vivo followinginjection of MSCOPG), however, complete suppression ofosteoclast differentiation required levels of 10–100 ng/ml, whichmay not have been achieved by the schedule of treatment usedhere. Furthermore the presence of syndecan-1 expressed onmyeloma cells (including KMS-12-BM) may bind and inactivateOPG, thus limiting its bioavailability32 in this model. Finallyenforced expression of OPG, while inhibiting osteoclastactivation, does not alter osteoblast suppression, which maybe required to produce a complete reversal of bone loss in ourmodel.

MSCs are a subset of non-haemopoietic progenitor cells thatare capable of differentiating into mesenchymal tissues includ-ing osteoblasts, adipocytes, chondrocytes and myocytes.33 Data

0.0

2.5

5.0

7.5

10.0

12.5

5.0

7.5

10.0

12.5

15.0

1.5

2.0

2.5

3.0

3.5

4.0

0.0

0.5

1.0

1.5

2.0

Vertebrae

p < 0.01

p < 0.01

p < 0.01

p < 0.01

p < 0.05 p < 0.05p < 0.05

p < 0.01

p < 0.01

p < 0.05

Tibiae

NS

NS NS

NS NS NS

0.0

2.5

5.0

7.5

10.0

12.5

15.0

20

40

60

80

100

30

35

45

40

50

% O

steo

clas

t p

erim

eter

0.0

2.5

5.0

7.5

10.0

12.5

15.0

% O

steo

clas

t p

erim

eter

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

non-dise

ased

MM only

MM + M

SC contro

l

MM + M

SC-OPG

% T

rab

ecu

lar

bo

ne

volu

me

% T

rab

ecu

lar

bo

ne

volu

me

Tra

bec

ula

r n

um

ber

(1/

mm

)

Tra

bec

ula

r n

um

ber

(m

icro

ns)

Tra

bec

ula

r n

um

ber

(1/

mm

)

Tra

bec

ula

r th

ickn

ess

(mic

ron

s)

Figure 5 Gene-modified MSCs expressing OPG reverse osteoclast activation and reduce bone loss in the KMS-12-BM model. Osteoclast numbers(expressed as %OcPm) in the vertebrae (a) and tibiae (b) of non-diseased controls, and diseased animals injected with KMS-12-BM cells alone (MMonly), with additional administration of control MSC (MMþMSC control), or MSCOPG (MMþMSCOPG). Data given for all animals, mediansindicated. Trabecular bone volume in vertebrae (c) and tibiae (d) with notations as for (a) and (b). Trabecular number (e) and (f) and thickness (g)and (h) in vertebrae and tibiae, respectively, labelled as above. Data for individual animals, medians indicated. MSC, mesenchymal stem cell; MM,multiple myeloma; OPG, osteoprotegerin.

A new xenograft model of myeloma bone diseaseN Rabin et al

2189

Leukemia

from animal experiments and a clinical trial suggest that theconditions characterized by increased cell turnover and tissueremodelling provide effective signals necessary for the survivaland proliferation of systemically delivered MSC.34,35 WhileMSCs have been shown to poorly engraft in healthy tissues,36

they have been shown to preferentially integrate in sites oftumours in animal models.37,38 Systemically delivered MSC,expressing therapeutic transgene led to tumour regression.38,39

In keeping with these reports, we observed that the presence oftumour cells in the bone marrow enhanced the short-termlocalization of infused MSC. Thus bone marrow-derived MSCsare attractive candidates to deliver therapeutic transgenes to thetumour environment in vivo, although clearly further work isneeded to optimize their efficacy. We have recently demon-strated that overexpression of CXCR4 improves short-termhoming of MSC to the bone marrow of irradiated b2m NOD/SCID mice (Kyriakou et al., Blood 2006; 108: abstract 3182).Another potential advantage of MSC is their ability todifferentiate into osteoblast progenitors. In support of thishypothesis is a recent report that MSC, when injected intotumour bearing bones, differentiated into osteoblasts, andreduced bone destruction in the SCID-Hu model.40

In myeloma, unopposed osteoclast activation resulting inbone destruction, fractures and deformity presents one of themost challenging aspects of this disease for both patients andtheir physicians. We describe a new xenogeneic model ofmyeloma bone disease, in which increased osteoclasts andreduced osteoblasts results in bone destruction which isquantifiable by histomorphometry and micro-CT. In addition,we have employed a novel strategy to deliver OPG in vivo anddemonstrated that this approach is able to reverse osteoclastactivation and restore bone volume. A full reversal of MMinduced bone disease may require the coordinate use ofstrategies to augment osteoblast function13 as well as inhibitingosteoclast activation.

References

1 Roodman GD. Mechanisms of bone lesions in multiple myelomaand lymphoma. Cancer 1997; 80 (8 Suppl): 1557–1563.

2 Bataille R, Chappard D, Marcelli C, Dessauw P, Baldet P, Sany Jet al. Recruitment of new osteoblasts and osteoclasts is the earliestcritical event in the pathogenesis of human multiple myeloma.J Clin Invest 1991; 88: 62–66.

3 Taube T, Beneton MN, McCloskey EV, Rogers S, Greaves M,Kanis JA. Abnormal bone remodelling in patients with myeloma-tosis and normal biochemical indices of bone resorption. Eur JHaematol 1992; 49: 192–198.

4 Kong YY, Yoshida H, Sarosi I, Tan HL, Timms E, Capparelli C et al.OPGL is a key regulator of osteoclastogenesis, lymphocytedevelopment and lymph-node organogenesis. Nature 1999; 397:315–323.

5 Anderson DM, Maraskovsky E, Billingsley WL, Dougall WC,Tometsko ME, Roux ER et al. A homologue of the TNF receptor andits ligand enhance T-cell growth and dendritic-cell function.Nature 1997; 390: 175–179.

6 Simonet WS, Lacey DL, Dunstan CR, Kelley M, Chang MS, Luthy Ret al. Osteoprotegerin: a novel secreted protein involved in theregulation of bone density. Cell 1997; 89: 309–319.

7 Pearse RN, Sordillo EM, Yaccoby S, Wong BR, Liau DF, Colman Net al. Multiple myeloma disrupts the TRANCE/osteoprotegerincytokine axis to trigger bone destruction and promote tumorprogression. Proc Natl Acad Sci USA 2001; 98: 11581–11586.

8 Giuliani N, Bataille R, Mancini C, Lazzaretti M, Barille S. Myelomacells induce imbalance in the osteoprotegerin/osteoprotegerinligand system in the human bone marrow environment. Blood2001; 98: 3527–3533.

9 Seidel C, Hjertner O, Abildgaard N, Heickendorff L, Hjorth M,Westin J et al. Serum osteoprotegerin levels are reduced in patients

with multiple myeloma with lytic bone disease. Blood 2001; 98:2269–2271.

10 Choi SJ, Cruz JC, Craig F, Chung H, Devlin RD, Roodman GDet al. Macrophage inflammatory protein 1-alpha is a potentialosteoclast stimulatory factor in multiple myeloma. Blood 2000; 96:671–675.

11 Giuliani N, Colla S, Morandi F, Lazzaretti M, Sala R, Bonomini Set al. Myeloma cells block RUNX2/CBFA1 activity in human bonemarrow osteoblast progenitors and inhibit osteoblast formationand differentiation. Blood 2005; 106: 2472–2483.

12 Tian E, Zhan F, Walker R, Rasmussen E, Ma Y, Barlogie B et al. Therole of the Wnt-signaling antagonist DKK1 in the development ofosteolytic lesions in multiple myeloma. N Engl J Med 2003; 349:2483–2494.

13 Giuliani N, Rizzoli V, Roodman GD. Multiple myeloma bonedisease: pathophysiology of osteoblast inhibition. Blood 2006;108: 3992–3996.

14 Croucher PI, Shipman CM, Lippitt J, Perry M, Asosingh K, Hijzen Aet al. Osteoprotegerin inhibits the development of osteolytic bonedisease in multiple myeloma. Blood 2001; 98: 3534–3540.

15 Choi SJ, Oba Y, Gazitt Y, Alsina M, Cruz J, Anderson J et al.Antisense inhibition of macrophage inflammatory protein 1-alphablocks bone destruction in a model of myeloma bone disease.J Clin Invest 2001; 108: 1833–1841.

16 Huang YW, Richardson JA, Tong AW, Zhang BQ, Stone MJ,Vitetta ES. Disseminated growth of a human multiple myeloma cellline in mice with severe combined immunodeficiency disease.Cancer Res 1993; 53: 1392–1396.

17 Alsina M, Boyce B, Devlin RD, Anderson JL, Craig F, Mundy GRet al. Development of an in vivo model of human multiplemyeloma bone disease. Blood 1996; 87: 1495–1501.

18 Tsunenari T, Koishihara Y, Nakamura A, Moriya M, Ohkawa H,Goto H et al. New xenograft model of multiple myeloma andefficacy of a humanized antibody against human interleukin-6receptor. Blood 1997; 90: 2437–2444.

19 Hjorth-Hansen H, Seifert MF, Borset M, Aarset H, Ostlie A, Sundan Aet al. Marked osteoblastopenia and reduced bone formation in amodel of multiple myeloma bone disease in severe combinedimmunodeficiency mice. J Bone Miner Res 1999; 14: 256–263.

20 Yaccoby S, Barlogie B, Epstein J. Primary myeloma cells growingin SCID-hu mice: a model for studying the biology and treatmentof myeloma and its manifestations. Blood 1998; 92: 2908–2913.

21 Yata K, Yaccoby S. The SCID-rab model: a novel in vivo system forprimary human myeloma demonstrating growth of CD138-expressing malignant cells. Leukemia 2004; 18: 1891–1897.

22 Turley H, Jones M, Erber W, Mayne K, de Waele M, Gatter K.VS38: a new monoclonal antibody for detecting plasma celldifferentiation in routine sections. J Clin Pathol 1994; 47: 418–422.

23 Hanawa H, Kelly PF, Nathwani AC, Persons DA, Vandergriff JA,Hargrove P et al. Comparison of various envelope proteins fortheir ability to pseudotype lentiviral vectors and transduceprimitive hematopoietic cells from human blood. Mol Ther2002; 5: 242–251.

24 Ahmed F, Ings SJ, Pizzey AR, Blundell MP, Thrasher AJ, Ye HTet al. Impaired bone marrow homing of cytokine-activated CD34+cells in the NOD/SCID model. Blood 2004; 103: 2079–2087.

25 Ohtsuki T, Yawata Y, Wada H, Sugihara T, Mori M, Namba M.Two human myeloma cell lines, amylase-producing KMS-12-PEand amylase-non-producing KMS-12-BM, were established from apatient, having the same chromosome marker, t(11;14)(q13;q32).Br J Haematol 1989; 73: 199–204.

26 Kyriakou CA, Yong KL, Benjamin R, Pizzey A, Dogan A, Singh Net al. Human mesenchymal stem cells (hMSCs) expressingtruncated soluble vascular endothelial growth factor receptor(tsFlk-1) following lentiviral-mediated gene transfer inhibit growthof Burkitt’s lymphoma in a murine model. J Gene Med 2006; 8:253–264.

27 Christianson SW, Greiner DL, Hesselton RA, Leif JH, Wagar EJ,Schweitzer IB et al. Enhanced human CD4+ T cell engraftment inbeta2-microglobulin-deficient NOD-scid mice. J Immunol 1997;158: 3578–3586.

28 Vanderkerken K, De Leenheer E, Shipman C, Asosingh K, Willems A,Van Camp B et al. Recombinant osteoprotegerin decreases tumorburden and increases survival in a murine model of multiplemyeloma. Cancer Res 2003; 63: 287–289.

A new xenograft model of myeloma bone diseaseN Rabin et al

2190

Leukemia

29 Sordillo EM, Pearse RN. RANK-Fc: a therapeutic antagonist forRANK-L in myeloma. Cancer 2003; 97 (3 Suppl): 802–812.

30 Body JJ, Greipp P, Coleman RE, Facon T, Geurs F, Fermand JP et al.A phase I study of AMGN-0007, a recombinant osteoprotegerinconstruct, in patients with multiple myeloma or breast carcinomarelated bone metastases. Cancer 2003; 97 (3 Suppl): 887–892.

31 Doran PM, Turner RT, Chen D, Facteau SM, Ludvigson JM, Khosla Set al. Native osteoprotegerin gene transfer inhibits the developmentof murine osteolytic bone disease induced by tumor xenografts.Exp Hematol 2004; 32: 351–359.

32 Tricot G. New insights into role of microenvironment in multiplemyeloma. Lancet 2000; 355: 248–250.

33 Pittenger MF, Mackay AM, Beck SC, Jaiswal RK, Douglas R, Mosca JDet al. Multilineage potential of adult human mesenchymal stem cells.Science 1999; 284: 143–147.

34 Liechty KW, MacKenzie TC, Shaaban AF, Radu A, Moseley AM,Deans R et al. Human mesenchymal stem cells engraft anddemonstrate site-specific differentiation after in utero transplanta-tion in sheep. Nat Med 2000; 6: 1282–1286.

35 Horwitz EM, Prockop DJ, Fitzpatrick LA, Koo WW, Gordon PL,Neel M et al. Transplantability and therapeutic effects of bone

marrow-derived mesenchymal cells in children with osteogenesisimperfecta. Nat Med 1999; 5: 309–313.

36 Koc ON, Peters C, Aubourg P, Raghavan S, Dyhouse S, DeGasperi Ret al. Bone marrow-derived mesenchymal stem cells remain host-derived despite successful hematopoietic engraftment after allogeneictransplantation in patients with lysosomal and peroxisomal storagediseases. Exp Hematol 1999; 27: 1675–1681.

37 Khakoo AY, Pati S, Anderson SA, Reid W, Elshal MF, Rovira II et al.Human mesenchymal stem cells exert potent antitumorigeniceffects in a model of Kaposi’s sarcoma. J Exp Med 2006; 203:1235–1247.

38 Studeny M, Marini FC, Champlin RE, Zompetta C, Fidler IJ,Andreeff M. Bone marrow-derived mesenchymal stem cells asvehicles for interferon-beta delivery into tumors. Cancer Res 2002;62: 3603–3608.

39 Nakamizo A, Marini F, Amano T, Khan A, Studeny M, Gumin Jet al. Human bone marrow-derived mesenchymal stem cells in thetreatment of gliomas. Cancer Res 2005; 65: 3307–3318.

40 Yaccoby S, Wezeman MJ, Zangari M, Walker R, Cottler-Fox M,Gaddy D et al. Inhibitory effects of osteoblasts and increased boneformation on myeloma in novel culture systems and a myeloma-tous mouse model. Haematologica 2006; 91: 192–199.

A new xenograft model of myeloma bone diseaseN Rabin et al

2191

Leukemia