Experimental Hematology 32 (2004) 685–691

030doi

New insight in the mechanism ofosteoclast activation and formation in multiple

myeloma: Focus on the receptor activator of NF-κB ligand (RANKL)

Nicola Giuliani, Simona Colla, and Vittorio RizzoliHematology and BMT Unit, University of Parma, Parma, Italy

(Received 11 March 2004; revised 26 March 2004; accepted 29 March 2004)

The increase of osteoclast activation and formation is mainly involved in the development ofthe osteolytic bone lesions that characterize multiple myeloma (MM) patients. The mechanismsby which myeloma cells induce bone resorption have not been clear for many years. Recently,new evidence has elucidated which factors are critically involved in the activation of osteoclasticcells in MM. The potential role of the critical osteoclastogenic factor, the receptor activatorof NF-kB ligand (RANKL), and its soluble antagonist osteoprotegerin (OPG) in the activation ofbone resorption in MM is summarized in this review. It has been demonstrated that humanMM cells induce an imbalance in the bone marrow environment of the RANKL/OPG ratioin favor of RANKL that triggers the osteoclast formation and activation leading to bonedestruction. The direct production of the chemokine macrophage inflammatory protein-1 a(MIP-1a) by myeloma cells, in combination with the RANKL induction in BM stromalcells in response to myeloma cells, are critical in osteoclast activation and osteoclastogenesis.

� 2004 International Society for Experimental Hematology. Published by Elsevier Inc.

Multiple myeloma (MM) is a plasma cell malignancy charac-terized by the high capacity to induce bone destruction[1,2]. Almost all patients with MM have osteolytic bonelesions more frequently localized at the spine, ribs, skull,and pelvis even though any skeletal site can be involved.Osteolytic bone lesions in MM patients mainly result froman increased bone resorption related to the stimulation ofosteoclast recruitment and activity [2–4].

The histomorphometric studies performed in MM patientshave demonstrated that the increase of osteoclastogenesis andosteclast activity is an early event that occurs in close contactwith myeloma cells [2,5,6], suggesting that local factorsrather than systemic mechanisms are involved in the patho-genesis of osteolytic bone lesions. An increase of os-teoblastogenesis has been also observed in the early phaseof disease or in patients with a low myeloma cell burden.On the contrary, MM patients with high plasma cell infiltrateor active disease are characterized by a lower number ofosteoblasts and a decreased bone formation that contribute,together with the increased osteoclast activity, in the devel-opment of bone lesions [2,5].

Offprint requests to: Nicola Giuliani, M.D., Ph.D., Hematology and BMTUnit, University of Parma, via Gramsci 14, 43100 Parma, Italy; E-mail:nicola.giuliani@unipr.it / ematopr@unipr.it

1-472X/04 $–see front matter. Copyright � 2004 International Society for: 10.1016/ j .exphem.2004.03.015

Although it is known that myeloma cells induce osteo-clastic bone resorption, the biological mechanisms involvedin the pathophysiology of MM-induced bone resorption havebeen unclear for several years.

Osteoclast activating factorsFirst, Mundy et al. [7] demonstrated that the conditionedmedia of human myeloma cells stimulated osteoclast activitybut, despite many suggested candidates, the critical osteo-clast activating factors (OAFs) involved have resisted identi-fication. It has been suggested that myeloma cells are ableto produce in vitro several osteoclastogenic cytokines suchas IL-6, IL-1β, TNF-α, HGF, and PTHrP but most of thestudies on MM samples in vivo have been inconclusivebecause of the presence of other contaminating cells (stromalcells or lympho-monocytes) as demonstrated by the lack ofIL-6 and IL-1β production or the rarity of PTHrP expressionby highly purified myeloma cells [3,4,8,9]. It has alsobeen reported that myeloma cells induce the release of os-teoclastogenic cytokines such as IL-6 or IL-11 by stromal/osteoblastic cells, underlining the potential role of the micro-environment in the activation of osteoclastic cells [9–12].

However, none of these cytokines has been demonstratedto be critical in the induction of the bone destruction in vivo

Experimental Hematology. Published by Elsevier Inc.

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691686

and in MM patients or correlated with the extension of theMM bone disease.

Osteoprotegerin (OPG)/RANKL/RANK systemRecently, two molecules belonging to the TNF receptor-ligand superfamily, the osteoprotegerin (OPG) and its ligandOPGL, namely the receptor activator of NF-κB ligand(RANKL) also known as TNF-related activation-induced cy-tokine (TRANCE), have been identified as critical in theregulation of osteoclast activity, leading to a new paradigmin the bone biology [13,14]. Extensive studies have shownthat OPG and RANKL exert a coupled control of boneresorption [13–22].

RANKL is a polypeptide of 217 amino acids that exertsits biological activity both in a trans-membrane form ofabout 40 to 45 kDa and in soluble form of 31 kDa. It hasbeen demonstrated that stromal/osteoblastic cells expressRANKL in response either to systemic factors such as PTH,dexamethasone, and vitamin D3 or local osteoclastogeniccytokines IL-1, TNF, and IL-11 [13–16]. RANKL directlyinduces osteoclastogenesis together with M-CSF and inhibitsosteoclast apoptosis by binding to its specific receptor(RANK) present on osteoclast progenitors and mature osteo-clasts [15,16,18,19]. More recently, it has been suggestedthat activated T lymphocytes, other than stromal/osteoblasticcells, produce RANKL and may maintain bone homeostasisthrough the cross-talk between RANKL production andinterferon-γ (IFN-γ) secretion [23]. In pathophysiologicalconditions, such as arthritis, activated T cells are capable ofregulating bone loss through the expression of RANKL[24,25].

OPG is a soluble decoy receptor of about 100 to 110kDa, produced by stromal/osteoblastic cells, that antagonizesthe effects of RANKL on osteoclastic cells inhibiting boneresorption [13,14,17]. It has been shown that OPG bindsRANKL and prevents the interaction between RANKL andits receptor RANK, blocking the osteoclast formation[13,14,17]. The critical role of OPG/RANKL system in theregulation of bone resorption has been confirmed inmouse models [13–17,19–22]. The presence of severe osteo-porosis with fractures has been reported in OPG knockoutmice [17,20] and the development of osteopetrosis is ob-served in transgenic mice overexpressing OPG [16] or inRANKL and RANK knockout mice [13–15,21,22]. More-over, RANKL administration in mice induces a dose-depen-dent hypercalcemia and consistently OPG is able to blockmalignant hypercalcemia in mouse models [26].

Role of RANKL/OPG in thepathophysiology of MM-induced bone diseaseBecause RANKL and its soluble antagonist OPG are likelyto play a critical role in the regulation of bone resorption

and the OPG/RANKL ratio in stromal/osteoblastic cells de-termines the level of osteoclast formation, we and othershave investigated the potential involvement of this systemin MM-induced bone destruction.

Do human myeloma cells express or produce RANKL?The direct RANKL expression by human myeloma cells iscontroversial. RANKL mRNA and protein have not beenfound in several human myeloma cell lines (HMCLs). Pearseet al. [27] tested ARP-1, U266, RPMI-8226, H929, andEBV-positive ARH-77 cell line and failed to find RANKLproduction or mRNA expression. Similarly, Giuliani et al.[28] found that IL-6-dependent HMCLs XG-1, XG-6, andMDN, or IL-6-independent HMCLs U266, RPMI-8226,OPM-2, LP-1, and JJN-3, did not express RANKL. Consis-tently, highly purified CD138� MM cells obtained from MMpatients failed to express RANKL mRNA. RANKL was alsonot detected in the conditioned medium of fresh MM cellsand HMCLs [29]. Moreover, RANKL immunostaining, per-formed by different groups on bone marrow (BM) biopsies ofMM patients using a specific anti-RANKL mAb, has shownthat myeloma cells are negative for RANKL expression[27,28,30]. More recently, Shaughnessy et al. [31], usingthe microarray technology, confirmed that RANKL was notdetected in any MM cells of either 83 osteolytic or 87 nonos-teolytic MM patients as well as in normal BM plasmacells. All these experimental observations clearly demon-strate that human myeloma cells do not express or directlyproduce the critical osteoclastogenic factor RANKL. In con-trast, other authors found that human myeloma cells produceRANKL [32–35]. In particular, Farrugia et al. [32] haveshown that the sorted CD38��� subpopulation expressedRANKL protein by flow cytometry and RANKL mRNA byreverse transcriptase polymerase chain reaction (RT-PCR).However, they also found that CD38� cells that are notmyeloma cells express RANKL, suggesting that the directRANKL production by myeloma is not a critical determi-nant in myeloma-induced osteoclast formation. Furthermore,the authors have used a more sensitive PCR technology,supporting the hypothesis that RANKL expression inCD38��� cells is very low.

Myeloma cells upregulate RANKLand inhibit OPG in stromal/osteoblastic cellsStrong evidence suggests that human MM cells induceRANKL expression in stromal cells and they decrease OPGexpression and secretion by osteoblastic cells, inducing animbalance of OPG/RANKL ratio in favor of RANKL[27,28,30]. RANKL upregulation has been observed at bothmRNA and protein level in a coculture system with HMCLsand either human BM stromal cells (BMSC)/preosteoblasticcells [28] or primary murine stromal cells [27]. MM cellsalso inhibit OPG expression and secretion by osteoblasticcells in a coculture system [27,28]. The cell-to-cell contactis critical in the induction of RANKL in BMSC by MMcells, as demonstrated by the lack of effect on RANKL

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691 687

expression in transwell system without cellular contact. Inparticular, the cell adhesion between MM cells and BMSCis mediated by VLA-4/VCAM-1 integrin system [9,10]. Ithas been reported that blocking antibody anti-VLA-4 com-pletely blunted the induction of RANKL by MM in humanBMSC [28]. The role of cell adhesion and VLA-4/VCAM-1interaction in the activation of osteoclastic cells by myelomacells has also been reported in a murine model of MM-induced bone disease, showing that the blocking of VLA-4 binding of myeloma cells to VCAM-1 on stromal cellsdecreases the release of bone resorbing factors by stromalcells and suppresses the development of osteolytic bonelesions [36].

RANKL induction in BMSC by MM cells is critical inthe formation of osteoclastic cells, as demonstrated by thecapacity of the RANK-Fc, a molecule made by fusing the Fcportion of immunoglobulin to a soluble form of the RANKreceptor that antagonizes RANKL/RANK interaction, toblock in vitro osteoclastogenesis in a coculture system withmurine stromal cells and MM cells [27].

Themechanisms by which myeloma cells induce RANKL/OPG imbalance in BMSC/osteoblastic cells are summarizedin Figure 1.

RANKL upregulation inT cells by myeloma cells: role of IL-7Growing evidence suggests that T lymphocytes may alsoregulate bone resorption and maintain bone homeostasisthrough the cross-talk between RANKL and IFN-γ, a cytok-ine that strongly suppresses osteoclastogenesis [23]. In phys-iopathological conditions, such as arthritis, activated Tcells are capable of regulating bone loss through the expres-sion of RANKL [24,25]. In addition, recent data highlightthe involvement of RANKL expressed by T lymphocytesin the mechanism of hypercalcemia in adult T-cell leukemia

Figure 1. MM-induced imbalance of RANKL and OPG in BM stromal/osteoblastic cells. MM cells upregulate RANKL and downregulate OPGin stromal/osteoblastic cells through the cell-to-cell contact. The RANKL/OPG imbalance in favor of RANKL induces osteoclast activation andbone destruction.

[37]. These observations prompted us to investigate whetherMM cells could also affect RANKL expression by T lympho-cytes. We found that HMCLs upregulate RANKL expressionand secretion in activated T cells in a transwell coculturesystem and similarly fresh purified MM cells induce RANKLin autologous T cells [29]. In the same system a downregula-tion of IFN-γ was also observed. The upregulation ofRANKL in T lymphocytes by MM cells seems to be medi-ated by the release of soluble factors. Among the moleculesthat could be responsible for the stimulation of RANKL,we focused our attention on IL-7. Recently it has been postu-lated that IL-7 might be involved in osteoclast activationbecause IL-7 stimulates RANKL production by T cellsin vitro [38] and induces bone loss in vivo by induction ofRANKL [39]. We found that HMCLs and fresh MM cellssecrete IL-7 in the presence of IL-6 [29]. The role of IL-7on RANKL stimulation in T lymphocytes by myeloma cellswas confirmed by the inhibitory effect exerted by an antibodyanti-IL-7 in the cocultures; furthermore, we found that IL-7neutralization inhibited MM-induced in vitro osteoclasto-genesis [29]. An inhibitory effect on RANKL in the cocul-tures was also observed in the presence of anti-IL-6 mAb.It is likely that IL-6 is both indirectly and directly involved inthe mechanism underlying RANKL stimulation by HMCLs.There are not data indicating that IL-6 stimulates RANKLin T lymphocytes or human osteoblasts [40]; however, wecannot exclude a direct effect of IL-6, since it has beendemonstrated that IL-6 is a potent inducer of RANKL pro-duction in murine systems [41,42].

A vicious loop between IL-6 and IL-7 in MM can behypothesized because it has been recently demonstrated thatIL-7 stimulates IL-6 secretion by BM stromal cells [43] andour data indicate that IL-6 induces IL-7 in MM cells. Thus,high levels of IL-6 in the BM environment could induce IL-7production by MM that, in turn, contributes to maintain highIL-6 levels and to stimulate RANKL in T cells. The potentialinvolvement of IL-7 is supported by the “in vivo” findingof higher IL-7 levels in peripheral serum and BM plasmaof MM patients than in normal subjects [29].

OPG/RANKL imbalancein the BM environment of MM patientsDifferent groups have shown that MM patients have animbalance in the OPG/RANKL expression in the BM envi-ronment, confirming the “in vitro” experimental data. Giu-liani et al. [28] have shown that ex vivo BM specimensobtained from MM patients overexpress RANKL mRNA incomparison with those from healthy donors (Fig. 2). More-over, immunostaining on BM biopsies has demonstrated anincrease of the number of RANKL� stromal cells togetherwith a reduction of OPG expression in trabecular osteoblastsof MM patients with osteolytic lesions as compared tohealthy subjects. Roux et al. [30] tested RANK and RANKLexpression by immunohistochemistry in 15 MM patients,6 patients with monoclonal gammopathy of undetermined

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691688

Figure 2. RANKL/OPG imbalance in MM patients. Immunohistochemical staining of RANKL and OPG performed on bone marrow biopsies of MM patients.

significance, and 10 normal subjects. They found thatRANKL is expressed by endosteal bone surface, vessels,and vimentin� stromal cells but not by plasma cells. More-over, they confirmed that stromal cells of MM patientspresented significantly higher levels of RANKL expressionas compared to normal and MGUS subjects. Finally, Pearseet al. [27] confirmed with both immunohistochemistry andin situ hybridization that MM patients have an increasedRANKL staining with a decreased OPG expression in theBM marrow as compared to healthy subjects or non-MMB-cell malignancies.

They have also found that, other than stromal cells,CD3�-activated T cells in the BM biopsies express RANKL[27], indicating that RANKL expression is increased in MM-infiltrated environment by the interaction of malignantplasma cells with both stromal cells and activated T cells.In line with this observation, Giuliani et al. [29] found thatpurified CD3� T lymphocytes obtained from MM patientswith extensive skeletal destruction are activated and theyexpress RANKL mRNA in contrast to those without bonelesions [29]. The presence of activated T cells in MM patients

has been previously demonstrated [44], confirming the poten-tial involvement of T lymphocytes in MM-induced bonedisease.

The imbalance of RANKL/OPG system observed in theBM environment has been confirmed by the finding of highRANKL serum levels and reduced OPG levels in MM patientsas compared to normal subjects [28,45,46]. Seidel et al.[45] showed that OPG levels were decreased to a greaterextent in patients with osteolytic lesions as compared topatients without bone disease. More recently, Terpos et al.[46] have measured soluble RANKL and OPG in 121 newlydiagnosed MM patients, showing that serum RANKL levelsand RANKL/OPG ratio were elevated in MM and correlatedwith bone disease and with markers of bone resorption.Moreover, RANKL/OPG ratio, together with β2-microglob-ulin and C-reactive protein, were independent prognosticfactors predicting survival in MM patients.

Effect of RANKL system inhibitionin murine models of human MM bone diseaseThe critical role of RANKL in MM-induced bone diseasehas been further confirmed in murine models of human MM

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691 689

bone using RANKL-specific inhibitors OPG or RANK-Fc.Pearse et al. [27] found that RANK-Fc completely blocksbone destruction, reducing the number of osteoclastic cells insevere combined immunodeficiency (SCID)/ARH-77 mousemodel and in the SCID-hu-MM mouse model injected withprimary MM cells. Administration of RANK-Fc also causeda marked reduction of tumor burden assessed histologicallyand of serum paraprotein in the SCID-hu-MM mice accom-panied by restoration of OPG and RANKL expression withinthe human xenograft, further supporting the causal associa-tion between MM bone destruction and RANKL/OPG dereg-ulation. Other authors have demonstrated that OPG alsoinhibits the development of osteolytic lesions in 5T2 MMmurine models [35] and decreases both osteolytic lesionsand tumor burden in 5T33 MM murine models [47]. Allthese data suggest that blocking bone resorption induced byRANKL, tumor burden as well as bone destruction in MMpatients may be decreased, supporting the critical role ofthe environment in MM cell growth.

Role of MIP-1a in MM-induced bone destructionMIP-1α is a chemokine recently suggested as a new OAFin MM-induced bone disease. MIP-1α is chemoattractantfor human osteoclasts [48] and induces osteoclast formationin vitro in rat marrow cultures [49]. It has been demonstratedthat MIP-1α is produced directly by several HMCLs and freshMM cells in the majority of MM patients [50–52]. HigherMIP-1α protein and mRNA levels have been reported in MMpatients as compared to normal subjects [50]. Moreover,blocking anti-MIP-1α or its receptor CCR5 Abs reducesMM-induced in vitro osteoclast formation [50,52]. Antisenseanti-MIP-1α is also able to block bone destruction in SCID/ARH-77 mouse model [53]. A strong association betweenMIP-1α mRNA expression by purified MM cells and activebone disease has been demonstrated in MM patients bymicroarray technology [54]. All these data suggest thatMIP-1α is a potential candidate as OAF involved in theosteoclast activation in MM patients, even if MIP-1α is nota critical factor for the osteoclast formation and activationsuch as RANKL. The potential relationship between MIP-1α and RANKL has been evaluated in human and murinesystems, with controversial results. It has been demonstratedthat MIP-1α induces RANKL in murine stromal cells [52]and it requires RANKL for its effects on osteoclastic cellsas demonstrated by the lack of the effect of MIP-1α inknockout RANK�/� mice [55]. These data suggest that theRANK/RANKL signaling pathway seems to be essential inmice for the osteoclastogenic effect of MIP-1α in vivo,leading further to the notion that RANKL is the final criticalfactor in the osteoclast activation. On the contrary, Han etal. have demonstrated that the osteoclastogenic effect ofMIP-1α is independent of RANKL because MIP-1α directlystimulates the osteoclast progenitors without increasing the

expression of RANKL in human nonadherent BM mononu-clear cells [56]. This notion supports the hypothesis thatMIP-1α, produced by myeloma cells, independently and incombination with RANKL enhances the osteoclast forma-tion in MM patients.

ConclusionsIn conclusion, we can now propose a new pathophysiologicalmodel of the MM bone destruction that highlights thecritical role of the OPG/RANKL system. Myeloma cellsinduce an imbalance in the OPG/RANKL ratio in stromal/osteoblastic cells through cell-to-cell contact. In addition,myeloma cells can stimulate RANKL and downregulate IFN-γ secretion by T cells at least in part through the direct releaseof IL-7 or the indirect involvement of the high IL-6 levelsinduced by myeloma cells in bone environment (Fig. 3).

The high BM expression and level of the critical osteo-clast activating factor RANKL associated with lower levelsof the osteoclastogenic inhibitor OPG induce osteoclast acti-vation and bone destruction in MM patients.

Other OAFs produced directly by MM cells, in particularthe chemokine MIP-1α, contribute in the activation of osteo-clastic cells through the final common mediator RANKL.

The recognition of the critical role of the RANKL/OPGsystem in the pathophysiology of MM-induced osteoclasto-genesis gives the rational design for future new therapeuticapproaches using RANKL-specific inhibitors OPG orRANK-Fc for the treatment of MM-induced bone disease,as suggested by a recent phase I study with recombinantOPG in MM patients [57].

Figure 3. Model for MM-induced osteoclastogenesis through RANKLinduction. Myeloma cells induce an imbalance in the OPG/RANKL ratioin stromal/osteoblastic cells through the cell-to-cell contact. In addition,myeloma cells stimulate RANKL and downregulate IFN-γ secretion by Tcells at least in part through the direct release of IL-7 or indirect involvementof the high IL-6 levels induced by myeloma cells in bone environment. Thehigh BM expression and level of the critical osteoclastogenic factor RANKLassociated with lower levels of OPG induce the activation of osteoclastsand trigger the bone destruction in MM patients.

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691690

References1. Bataille R, Harousseau JL. Multiple myeloma. N Engl J Med. 1997;

336:1657–1664.2. Bataille R, Manolagas SC, Berenson JR. Pathogenesis and management

of bone lesions in multiple myeloma. Hematol Oncol Clin North Am.1997;11:349–361.

3. Croucher PI, Apperley JF. Bone disease in multiple myeloma. Br JHaematol. 1998;103:902–910.

4. Callander NS, Roodman GD. Myeloma bone disease. Semin Hematol.2001;38:276–285.

5. Bataille R, Chappard D, Marcelli C, et al. Recruitment of new osteo-blasts and osteoclasts is the earliest critical event in the pathogenesisof human multiple myeloma. J Clin Invest. 1991;88:62–66.

6. Bataille R, Chappard D, Basle M. Quantifiable excess of bone resorp-tion in monoclonal gammopathy is an early symptom of malignancy:a prospective study of 87 bone biopsies. Blood. 1996;87:4762–4769.

7. Mundy GR, Raisz LG, Cooper RA, et al. Evidence for the secretionof an osteoclast stimulating factor in myeloma. N Engl J Med. 1974;291:1041–1046.

8. Roodman D. Biology of myeloma bone disease. Hematology (Am SocHematol Educ Program). 2002;227–232.

9. Chauhan D, Uchiyama H, Akbarali Y, et al. Multiple myeloma celladhesion-induced interleukin-6 expression in bone marrow stromal cellsinvolves activation of NF-κ B. Blood. 1996;87:1104–1112.

10. Barille S, Collette M, Bataille R, et al. Myeloma cells upregulateinterleukin-6 secretion in osteoblastic cells through cell-to-cell contactbut down-regulate osteocalcin. Blood. 1995;86:3151–3159.

11. Hjertner O, Torgersen ML, Seidel C, et al. Hepatocyte growth factor(HGF) induces interleukin-11 secretion from osteoblasts: a possiblerole for HGF in myeloma-associated osteolytic bone disease. Blood.1999;94:3883–3888.

12. Hofbauer LC, Khosla S, Dunstan CR, et al. The roles of osteoprotegerinand osteoprotegerin ligand in the paracrine regulation of bone resorp-tion. J Bone Miner Res. 2000;15:2–12.

13. Khosla S. Minireview: the OPG/RANKL/RANK system. Endocrinol-ogy. 2001;142:5050–5055.

14. Lacey DL, Timms E, Tan HL, et al. Osteoprotegerin ligand is acytokine that regulates osteoclast differentiation and activation. Cell.1998;93:165–176.

15. Kong YY, Yoshida H, Sarosi I, et al. OPGL is a key regulator ofosteoclastogenesis, lymphocyte development and lymph-node organo-genesis. Nature. 1999;397:315–323.

16. Simonet WS, Lacey DL, Dunstan CR, et al. Osteoprotegerin: a novelsecreted protein involved in the regulation of bone density. Cell. 1997;89:309–319.

17. Yasuda H, Shima N, Nakagawa N, et al. Osteoclast differentiation factoris a ligand for osteoprotegerin/osteoclastogenesis-inhibitory factor andis identical to TRANCE/RANKL. Proc Natl Acad Sci U S A. 1998;95:3597–3602.

18. Burgess TL, Qian Y, Kaufman S, et al. The ligand for osteoprotegerin(OPGL) directly activates mature osteoclasts. J Cell Biol. 1999;145:527–538.

19. Bucay N, Sarosi I, Dunstan CR, et al. Osteoprotegerin-deficient micedevelop early onset osteoporosis and arterial calcification. Genes Dev.1998;12:1260–1268.

20. Min H, Morony S, Sarosi I, Dunstan CR, et al. Osteoprotegerin reversesosteoporosis by inhibiting endosteal osteoclasts and prevents vascularcalcification by blocking a process resembling osteoclastogenesis. JExp Med. 2000;192:463–474.

21. Li J, Sarosi I, Yan XQ, et al. RANK is the intrinsic hematopoieticcell surface receptor that controls osteoclastogenesis and regulationof bone mass and calcium metabolism. Proc Natl Acad Sci U S A.2000;97:1566–1571.

22. Dougall WC, Glaccum M, Charrier K, et al. RANK is essential forosteoclast and lymph node development. Genes Dev. 1999;13:2412–2424.

23. Takayanagi H, Ogasawara K, Hida S, et al. T-cell-mediated regulationof osteoclastogenesis by signalling cross-talk between RANKL andIFN-γ. Nature. 2000;408:600–605.

24. Kong YY, Feige U, Sarosi I, et al. Activated T cells regulate boneloss and joint destruction in adjuvant arthritis through osteoprotegerinligand. Nature. 1999;402:304–309.

25. Kotake S, Udagawa N, Hakoda M, et al. Activated human T cellsdirectly induce osteoclastogenesis from human monocytes: possiblerole of T cells in bone destruction in rheumatoid arthritis patients.Arthritis Rheum. 2001;44:1003–1012.

26. Capparelli C, Kostenuik PJ, Morony S, et al. Osteoprotegerin preventsand reverses hypercalcemia in a murine model of humoral hypercalce-mia of malignancy. Cancer Res. 2000;60:783–787.

27. Pearse RN, Sordillo EM, Yaccoby S, et al. Multiple myeloma disruptsthe TRANCE/osteoprotegerin cytokine axis to trigger bone destructionand promote tumor progression. Proc Natl Acad Sci U S A. 2001;98:11581–11586.

28. Giuliani N, Bataille R, Mancini C, et al. Myeloma cells induce imbal-ance in the osteoprotegerin/osteoprotegerin ligand system in the humanbone marrow environment. Blood. 2001;98:3527–3533.

29. Giuliani N, Colla S, Sala R, et al. Human myeloma cells stimulatethe receptor activator of nuclear factor-κ B ligand (RANKL) in Tlymphocytes: a potential role in multiple myeloma bone disease. Blood.2002;100:4615–4621.

30. Roux S, Meignin V, Quillard J, et al. RANK (receptor activator ofnuclear factor-κB) and RANKL expression in multiple myeloma. BrJ Haematol. 2002;117:86–92.

31. Shaughnessy JD, Barlogie B. Interpreting the molecular biology andclinical behavior of multiple myeloma in the context of global geneexpression profiling. Immunol Rev. 2003;194:140–163.

32. Farrugia AN, Atkins GJ, To LB, et al. Receptor activator of nuclearfactor-κB ligand expression by human myeloma cells mediates osteo-clast formation in vitro and correlates with bone destruction in vivo.Cancer Res. 2003;63:5438–5445.

33. Heider U, Langelotz C, Jakob C, et al. Expression of receptor activatorof nuclear factor κB ligand on bone marrow plasma cells correlateswith osteolytic bone disease in patients with multiple myeloma. ClinCancer Res. 2003;9:1436–1440.

34. Sezer O, Heider U, Jakob C, et al. Human bone marrow myelomacells express RANKL. J Clin Oncol. 2002;20:353–354.

35. Croucher PI, Shipman CM, Lippitt J, et al. Osteoprotegerin inhibitsthe development of osteolytic bone disease in multiple myeloma. Blood.2001;98:3534–3540.

36. Michigami T, Shimizu N, Williams PJ, et al. Cell-cell contact betweenmarrow stromal cells and myeloma cells via VCAM-1 and α(4)β(1)-integrin enhances production of osteoclast-stimulating activity. Blood.2000;96:1953–1960.

37. Nosaka K, Miyamoto T, Sakai T, et al. Mechanism of hypercalcemiain adult T-cell leukemia: overexpression of receptor activator of nuclearfactor κB ligand on adult T-cell leukemia cells. Blood. 2002;99:634–640.

38. Weitzmann MN, Cenci S, Rifas L, et al. Interleukin-7 stimulates osteo-clast formation by up-regulating the T-cell production of soluble os-teoclastogenic cytokines. Blood. 2000;96:1873–1878.

39. Toraldo G, Roggia C, Qian WP, et al. IL-7 induces bone loss in vivoby induction of receptor activator of nuclear factor κ B ligand andtumor necrosis factor α from T cells. Proc Natl Acad Sci U S A.2003;100:125–130.

40. Hofbauer LC, Lacey DL, Dunstan CR, et al. Interleukin-1β and tumornecrosis factor-α, but not interleukin-6, stimulate osteoprotegerinligand gene expression in human osteoblastic cells. Bone. 1999;25:255–259.

41. Palmqvist P, Persson E, Conaway HH, Lerner UH. IL-6, leukemiainhibitory factor, and oncostatin M stimulate bone resorption and regu-late the expression of receptor activator of NF-κ B ligand, osteoproteg-erin, and receptor activator of NF-κ B in mouse calvariae. J Immunol.2002;169:3353–3362.

N. Giuliani et al. /Experimental Hematology 32 (2004) 685–691 691

42. O’Brien CA, Gubrij I, Lin SC, Saylors RL, Manolagas SC. STAT3activation in stromal/osteoblastic cells is required for induction ofthereceptor activator of NF-κB ligand and stimulation of osteoclasto-genesis by gp130-utilizing cytokines or interleukin-1 but not 1,25-dihydroxyvitamin D3 or parathyroid hormone. J Biol Chem. 1999 Jul2;274(27):19301–19308.

43. Iwata M, Graf L, Awaya N, Torok-Storb B. Functional interleukin-7receptors (IL-7Rs) are expressed by marrow stromal cells: binding ofIL-7 increases levels of IL-6 mRNA and secreted protein. Blood. 2002;100:1318–1325.

44. Dianzani U, Pileri A, Boccadoro M, et al. Activated idiotype-reactivecells in suppressor/cytotoxic subpopulations of monoclonal gammopa-thies: correlation with diagnosis and disease status. Blood. 1988;72:1064–1068.

45. Seidel C, Hjertner O, Abildgaard N, et al., Nordic Myeloma StudyGroup. Nordic Myeloma Study Group. Serum osteoprotegerin levelsare reduced in patients with multiple myeloma with lytic bone disease.Blood. 2001;98:2269–2271.

46. Terpos E, Szydlo R, Apperley JF, et al. Soluble receptor activatorof nuclear factor κB ligand–osteoprotegerin ratio predicts survival inmultiple myeloma: proposal for a novel prognostic index. Blood. 2003;102:1064–1069.

47. Vanderkerken K, De Leenheer E, Shipman C, et al. Recombinant osteo-protegerin decreases tumor burden and increases survival in a murinemodel of multiple myeloma. Cancer Res. 2003;63:287–289.

48. Kukita T, Nomiyama H, Ohmoto Y, et al. Macrophage inflammatoryprotein-1 α (LD78) expressed in human bone marrow: its role inregulation of hematopoiesis and osteoclast recruitment. Lab Invest.1997;76:399–406.

49. Scheven BA, Milne JS, Hunter I, et al. Macrophage-inflammatoryprotein-1α regulates preosteoclast differentiation in vitro. BiochemBiophys Res Commun. 1999;254:773–778.

50. Choi SJ, Cruz JC, Craig F, et al. Macrophage inflammatory protein-1α is a potential osteoclast stimulatory factor in multiple myeloma. Blood.2000;96:671–675.

51. Uneda S, Hata H, Matsuno F, et al. Macrophage inflammatory protein-1α is produced by human multiple myeloma (MM) cells and its expres-sion correlates with bone lesions in patients with MM. Br J Haematol.2003;120:53–55.

52. Abe M, Hiura K, Wilde J, et al. Role for macrophage inflammatoryprotein (MIP)-1α and MIP-1β in the development of osteolytic lesionsin multiple myeloma. Blood. 2002;100:2195–2202.

53. Choi SJ, Oba Y, Gazitt Y, et al. Antisense inhibition of macrophageinflammatory protein 1-α blocks bone destruction in a model of my-eloma bone disease. J Clin Invest. 2001;108:1833–1841.

54. Magrangeas F, Nasser V, Avet-Loiseau H, et al. Gene expression profil-ing of multiple myeloma reveals molecular portraits in relation to thepathogenesis of the disease. Blood. 2003;101:4998–5006.

55. Oyajobi BO, Franchin G, Williams PJ, et al. Dual effects of macrophageinflammatory protein-1α on osteolysis and tumor burden in themurine 5TGM1 model of myeloma bone disease. Blood. 2003;102:311–319.

56. Han JH, Choi SJ, Kurihara N, Koide M, Oba Y, Roodman GD. Macro-phage inflammatory protein-1α is an osteoclastogenic factor in my-eloma that is independent of receptor activator of nuclear factor κBligand. Blood. 2001;97:3349–3353.

57. Body JJ, Greipp P, Coleman RE, et al. A phase I study of AMGN-0007, a recombinant osteoprotegerin construct, in patients with multiplemyeloma or breast carcinoma related bone metastases. Cancer. 2003;97:887–892.