RANK-Fc: A therapeutic antagonist for RANK-L in myeloma

Authors

  • Emilia Mia Sordillo M.D., Ph.D.,

    1. Departments of Medicine and Pathology and Laboratory Medicine, St. Luke's-Roosevelt Hospital Center, College of Physicians and Surgeons, Columbia University, New York, New York
    Search for more papers by this author
  • Roger Niles Pearse M.D., Ph.D.

    Corresponding author
    1. Myeloma Research Center, Department of Medicine, New York-Presbyterian Hospital, Cornell University Medical College, New York, New York
    • Myeloma Research Center, Department of Medicine, New York-Presbyterian Hospital, Cornell University Medical College, 1300 York Avenue, New York, NY 10021
    Search for more papers by this author
    • Fax: (212) 746-7888


Abstract

BACKGROUND

Severe bone destruction due to inappropriate osteoclastogenesis is a prominent feature of multiple myeloma (MM). MM increases bone loss by disrupting the checks that normally control signaling by receptor activator of nuclear factor κB ligand (RANK-L, also called TRANCE [tumor necrosis factor-related, activation-induced cytokine], osteoprotegerin ligand [OPG-L], osteoclast differentiation factor [ODF], and tumor necrosis factor superfamily member 11 [TNFSF11]), a TNF-family cytokine required for osteoclast differentiation and activation. RANK-L binds to its functional receptor RANK (TNF receptor superfamily member 11a [TNF RSF11a]) to stimulate osteoclastogenesis. Osteotropic cytokines regulate this process by controlling bone marrow stromal expression of RANK-L. Further control over osteoclastogenesis is maintained by regulated expression of osteoprotegerin (OPG, also called osteoclastogenesis inhibitory factor and TNFRSF11b), a soluble decoy receptor for RANK-L. In normal bone marrow, abundant stores of OPG in stroma, megakaryocytes, and myeloid cells provide a natural buffer against increased RANK-L. MM disrupts these controls by increasing expression of RANK-L and decreasing expression of OPG. Concurrent deregulation of RANK-L and OPG expression is found in bone marrow biopsies from patients with MM but not in specimens from patients with non-MM hematologic malignancies.

METHODS

RANK-Fc is a recombinant RANK-L antagonist that is formed by fusing the extracellular domain of RANK to the Fc portion of human immunoglobulin G1 (hIgG1). In vitro, addition of RANK-Fc virtually eliminates the formation of osteoclasts in cocultures of MM with bone marrow and osteoblast/stromal cells. The severe combined immunodeficiency (SCID)/ARH77 mouse model and the SCID-hu-MM mouse model of human MM were used to assess the ability of RANK-Fc to block the development of MM-induced bone disease in vivo. Mice received either RANK-Fc or hIgG1 200 μg intravenously three times per week.

RESULTS

RANK-Fc limited bone destruction in both the SCID/ARH-77 model and the SCID-hu-MM model. Administration of RANK-Fc also caused a marked reduction in tumor burden and serum paraprotein in SCID-hu-MM mice that was associated with the restoration of OPG and a reduction in RANK-L expression in the xenograft.

CONCLUSIONS

MM-induced bone destruction requires increased RANK-L expression and is facilitated by a concurrent reduction in OPG, a natural decoy receptor for RANK-L. Administration of the RANK-L antagonist RANK-Fc limits MM-induced osteoclastogenesis, development of bone disease, and MM tumor progression. Cancer 2003;97(3 Suppl):802–12. © 2003 American Cancer Society.

DOI 10.1002/cncr.11134

The diagnosis of multiple myeloma (MM) is virtually synonymous with bone disease.1, 2 Although small numbers of MM patients develop sclerotic bone lesions in the context of a syndrome that includes polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and sclerodactyly (POEMS syndrome),3, 4 most patients with MM develop lytic lesions or diffuse bone loss as a result of heightened resorption compounded by suppressed bone formation.5 MM-associated bone resorption, which is the consequence of increased production and activation of osteoclasts, has been linked consistently with the degree of plasma cell infiltration.6

The production of osteoclast activating factors (OAFs) by MM has been recognized for over 25 years.7 A number of cytokines, including interleukin 1 (IL-1), IL-6, IL-11, tumor necrosis factor α (TNFα), and transforming growth factor β (TGF-β), can activate osteoclasts in vitro and are produced by MM cell lines or are increased in some patients. However, none of these cytokines are increased universally in patients with MM who have lytic bone disease.8 By contrast, recent evidence from several laboratories suggests that the receptor activator of nuclear factor κB ligand (RANK-L, also called tumor necrosis factor-related, activation-induced cytokine [TRANCE], osteoprotegerin ligand [OPG-L], osteoclast differentiation factor [ODF], and tumor necrosis factor superfamily member 11 [TNFSF11]) is the common mediator of MM-produced OAFs and that MM triggers osteoclastogenesis by disrupting the balance between RANK-L and its natural inhibitor, osteoprotegerin (OPG; also called osteoclastogenesis inhibitory factor and TNFRSF11b).9–13

In normal bone homeostasis, RANK-L and OPG participate in a cytokine axis that tightly controls the generation of osteoclasts from monocyte precursors (Fig. 1A). RANK-L, a novel member of the TNF superfamily that is expressed by osteoblasts and bone marrow stromal cells, binds to its functional receptor, RANK (TNF receptor superfamily member 11a [TNFRSF11a]), to stimulate osteoclastogenesis. OPG, which is expressed by osteoblasts, stromal cells, dendritic cells, and megakaryocytes, limits this process by acting as a soluble decoy receptor for RANK-L. Osteotropic hormones, such as estrogen and vitamin D3 (VitD3), also regulate the generation of osteoclasts by modulating the expression of RANK-L and OPG (Table 1). In addition, osteoclastogenesis is controlled by feedback loops that are initiated by the action of osteoclasts on bone. Bone resorption elevates the local concentration of calcium, which can stimulate osteoblast and stromal expression of OPG.14 TGF-β deposited in bone matrix by osteoblasts during bone formation is released during bone resorption. In turn, TGF-β limits osteoclastogenesis by decreasing stromal expression of RANK-L and increasing stromal expression of OPG.15 During bone resorption, osteoclasts eliminate osteocytes, which are the osteoblast-derived cells within bone that act as sensors for bone damage. Osteocytes communicate directly with osteoblasts and osteoblast-derived lining cells by projections through lacunae; severing this communication also may result in osteoblast expression of OPG.

Figure 1.

(A) Normal bone homeostasis is maintained by coordinate expression of RANK-L and OPG. RANK-L binds to its functional receptor, RANK, to stimulate the development of osteoclasts from monocytic precursors. Osteotropic hormones, such as estrogens and vitamin D3 (VITD3) control osteoclast development by regulating stromal cell expression of RANK-L and the RANK-L decoy receptor, OPG. In the adult, bone homeostasis is maintained by OPG levels in excess of RANK-L, resulting in a check on osteoclast generation. This check is reinforced by transforming growth factor β (TGF-β), which is released from bone during resorption and acts as feedback to decrease RANK-L and to increase OPG expression. (B) Multiple myeloma (MM) disrupts the RANK-L/OPG balance to trigger osteoclastogenesis and bone destruction. MM-associated osteoclast activating factors (OAFs) increase RANK-L and decrease OPG expression by stromal cells. MM also inhibits TGF-β stimulation of OPG expression. Together, these effects lead to enhanced RANK-L activity, resulting in increased osteoclast generation. HGF: hepatocyte growth factor; IL-1β: interleukin 1β; MIP-1α: macrophage inflammatory protein-1α.

Table 1. Regulation of RANK-L and OPG Expression
StimulusExpression (model)a
RANK-LOPG
  • RANK-L: receptor activator of nuclear factor κB ligand; OPG: osteoprotegerin; IL: interleukin; mOB: primary murine osteoblasts; PGE2: prostaglandin E2: PTH: parathyroid hormone; mMS: primary murine bone marrow stroma; hFOB: human fetal osteoblasts; hOB: human osteoblasts; hMS: primary human bone marrow stroma; ST-2: murine bone marrow stromal cell line; MG-63: human osteosarcoma cell line; TNF-α: tumor necrosis factor α; FGF; fibroblast growth factor; VitD3: vitamin D3; TGF-β: transforming growth factor β; MC3T3: murine osteoblast cell line; IMR-90: lung fibroblast cell line; CD40L: CD40 ligand; DC: bone marrow-derived dendritic cells; BMP-2: bone morphogenic protein 2.

  • a

    Superscript numbers refer to the list of references.

IL-17Increased (mOB)40
PGE2Increased (mOB)18, 38Decreased (mOB)95, 96
PTHIncreased (mOB)18, 39Decreased (mOB)95
CortisolIncreased (mMS, hFOB)97Decreased (hFOB, hOB, hMS, ST-2)95, 98, 99
IL-1βIncreased (hMS, MG-63)41Increased (hFOB)32, 61
IL-11Increased (mOB)18, 39Increased (mOB)39
TNF-αIncreased (hMS, MG-63)41Increased (hFOB, MG-63)32, 61, 100
Basic FGFIncreased (mOB),42 decreased (ST-2, mOB)101Decreased (mOB)42
VitD3Increased mOB, ST-2)18, 39Increased (hFOB, hOB),61 decreased (ST-2, mOB)39, 95
IL-1αIncreased (MG-63, hOB),102 decreased (mOB)95
TGF-βDecreased (ST-2)15Increased (MC3T3, ST-2, mOB)15, 95
EstradiolUnchanged (hFOB, hOB)41Increased (hFOB, hOB)41
CaCl2Increased (IMR-90, ST-2, MC3T3)14
TNF-βIncreased (MG-63)100
CD40LIncreased (DC)34
BMP-2Increased (hFOB)61

RANK-L

RANK-L was cloned initially from activated T-cells as a TNF-related survival factor for thymocytes and dendritic cells.16, 17 A role for RANK-L in osteoclastogenesis became evident when it was identified as a ligand for OPG.18, 19 The absolute requirement for RANK-L in osteoclast generation and survival was confirmed by the absence of osteoclasts in RANK-L −/− mice.20–22RANK-L −/− mice also demonstrate defective lymph node and breast alveolar development as well as abnormal development of B lymphocytes and CD25+ regulatory T lymphocytes, suggesting critical roles for RANK-L, RANK, and OPG in these processes.20, 23–25

RANK-L is encoded by a single gene at human chromosome 13q14. Alternative splicing of RANK-L mRNA allows expression as a type II transmembrane glycoprotein of either 316 or 270 amino acids or as a soluble ligand of 243 amino acids.20, 26 In addition, RANK-L can be released from its membrane bound state by metalloproteinases, including TNFα convertase.27, 28 All four isoforms of RANK-L associate into trimeric molecules capable of triggering osteoclastogenesis. RANK-L has limited sequence homology to TNF-related apoptosis inducing ligand [TRAIL] (34%), CD40 ligand (28%), and FAS ligand (19%), although it displays remarkably similar topology to these TNFSF members. Like all TNFSF members, RANK-L assumes a structure based on 10 β-pleated strands arranged into inner and outer sheets. The inner β-pleated strands are responsible for homotrimerization and share the greatest homology to other TNFSF members. The greatest divergence is found in four surface loops that connect the outer strands and dictate specificity for its receptor, RANK.29, 30

RANK-L is expressed by activated CD4+ and CD8+ T lymphocytes, double negative thymocytes, immature B lymphocytes, osteoblasts, bone marrow stroma, vascular endothelia, developing lymph node anlage, and developing breast epithelia.16, 17, 19–21, 23 RANK-L mRNA is present in bone marrow, peripheral lymph nodes, spleen, thymus, Peyer's patches, fetal liver, heart, skeletal muscle, lung, stomach, intestine, placenta, thyroid gland, peripheral blood leukocytes, and murine cerebellum.16–19, 23, 31–37 Expression of RANK-L is increased by glucocorticoids, IL-1β, IL-11, parathyroid hormone (PTH), TNF-α, prostaglandin E2, and VitD3 and is decreased by TGF-β (Table 1).15, 18, 38, 38–42 Expression of RANK-L, thus, is regulated by many cytokines previously identified as OAFs.

RANK

The TNFRSF member RANK probably is the only functional receptor for RANK-L, because RANK −/− and RANK-L −/− mice display similar phenotypes.43, 44 RANK is a type I transmembrane glycoprotein encoded on human chromosome 18q22.1 and is expressed on the surface of osteoclasts and osteoclast precursors as well as bone marrow-derived dendritic cells, activated T-cells, vascular endothelia, chondrocytes, bone marrow fibroblasts, and mammary gland epithelia. RANK expression is increased on dendritic cells by CD40 ligand; on activated T-cells by IL-4 and TGF-β; on osteoclast precursors by CSF-1 and TGF-β; and on bone marrow fibroblasts by IL-1, IL-6, IL-11, and TNFα. RANK mRNA also is found in skeletal muscle, adrenal gland, intestine, and liver.16–19, 23, 31–37

The ectodomain of RANK is formed by 4 cysteine-rich pseudorepeats with 40% overall homology to CD40 (Fig. 2).16 All TNFRSF members possess ectodomains comprised of from one to five cysteine-rich pseudorepeats.45 Each pseudorepeat is characteristically 40 amino acids in length and contains 6 conserved cysteine residues that form 3 intrachain disulfide bonds. Based on structural analysis of other TNFR, the four pseudorepeats of RANK are projected to form an elongated structure capable of binding to one of the three grooves formed by the trimerization of RANK-L.29 Thus, each RANK-L trimer engages three molecules of RANK.46

Figure 2.

Receptors for RANK-L. RANK, the functional receptor for RANK-L, is a type 1 transmembrane glycoprotein. The ectodomain of RANK (ECTO) contains four cysteine-rich domains (CRDs) that are responsible for RANK-L binding. RANK-Fc is a chimeric protein formed by fusing the four CRDs of RANK with the Fc portion of human immunoglobulin G1 (IgG1). The two disulfide bonds of the Fc portion allow homodimerization, which may contribute to avidity for RANK-L. osteoprotegerin (OPG), the natural decoy receptor for RANK-L, contains four CRDs with which it binds RANK-L and TRAIL, as well as two death domains (DD), a heparin binding domain (HBD), and a cysteine responsible for homodimerization. Avidity for RANK-L may be increased by dimerization as well as intermolecular interactions mediated by the DD or HBD. OPG-Fc is a chimeric protein formed by fusing OPG with the Fc portion of human IgG1.

The tripartite symmetry of TNF signaling continues inside the cell. Trimerization triggers a conformational change in the cytoplasmic domain of RANK that allows recruitment of TNFR-associated factors (TRAFs). Each cytoplasmic tail of the RANK trimer binds a molecule of TRAF, resulting in TRAF trimerization and the initiation of intracellular signaling cascades. TNFRSF members fall into two broad groups: Fas, TNFR1, DR3, DR4, DR5, and DR6 possess cytoplasmic death domains through which they indirectly recruit TRAFs through the adapter molecules TRADD and FADD. In contrast, RANK, TNFR2, CD40, and CD30 do not contain death domains but, rather, bind TRAFs directly. The cytoplasmic domain of RANK contains docking sites for TRAF1, TRAF2, TRAF3, TRAF5, and TRAF6. However, TRAF2 and TRAF6 are most critical for RANK signaling.47–49 TRAF2 mediates activation of AP1 in concert with ASK1.50, 51 TRAF6 complexes with c-Src and c-Cbl to activate PI3K, leading to PKB activation and cytoskeletal reorganization.52–54 In addition, TRAF6 activates microphthalmia transcription factor (MITF) by activating the p38 microtubule-associated protein kinase pathway through TAB2 and TAK1.55 It is also likely that TRAF6, by recruiting TAB1, TAB2, and TAK1, activates IKKα, and thus NF-κB. All of these signaling events appear to be necessary for RANK signaling, because animals that lack RANK, TRAF6, c-src, p38α, IKKα, NF-κB, c-fos, or MITF fail to generate activated osteoclasts.43, 44, 56, 57 Cytokines and growth factors that influence the components of these pathways potentially may modulate RANK-L signaling. For example, interferon γ (IFN-γ) targets TRAF6 for ubiquinization, and IFN-β inhibits the ability of RANK-L to activate FOS.58, 59 The importance of such modulation is confirmed by the appearance of osteoporosis in mice deficient in IFN-β or IFN-γ signaling.

OPG

OPG is a soluble, 110-kDa, disulfide-linked, homodimeric glycoprotein that functions as a decoy receptor for RANK-L. Thus, OPG modulates osteoclast formation by inhibiting RANK activation.19 The ability of OPG to inhibit osteoclastogenesis was demonstrated first using transgenic mice that overexpressed OPG under control of the liver specific apolipoprotein E gene promoter and associated enhancer.31 These animals develop severe osteopetrosis as the result of arrested osteoclastogenesis. In contrast, targeted deletion of OPG leads to severe osteoporosis due to increased osteoclastogenesis.60 In addition, OPG −/− mice develop arterial calcification. Although arterial calcification may reflect the hypercalcemia that accompanies increased bone turnover, it also is possible that OPG plays a role in the maintenance of vascular smooth muscle homeostasis.

OPG possesses four cysteine-rich pseudorepeats, two death domains, a heparin-binding domain, and five potential N-glycosylation sites (Fig. 2). Based on structural analysis of other TNFRs, the four pseudorepeats of OPG are projected to form an elongated structure capable of binding to one of the three grooves formed by the trimerization of RANK-L.29 A cysteine at the carboxy terminus of OPG dictates dimerization, which probably is necessary to allow sufficient avidity for RANK-L in solution. The death domains, which are a unique feature for a secreted protein, may contribute to this avidity by providing additional oligomerization. OPG mRNA has been detected in B cells, bone marrow-derived and follicular dendritic cells, vascular endothelia, arterial smooth muscle, heart, lung, kidney, bone, stomach, intestine, placenta, liver, thyroid, skin, spinal cord, and brain.16–19, 23, 31–37 Expression in osteoblasts and bone marrow fibroblasts is increased by TGF-β, 17β-estradiol, VitD3, CaCl2, and TNF-α and is decreased by IL-1β, glucocorticoids, VitD3, and PTH (Table 1).14, 15, 61 OPG is encoded by a single gene on chromosome 8q24 comprised of 5 exons spanning 29 kb.62 A single intron between the two exons encoding the four cysteine-rich domains suggests divergence from other TNFRs.

OPG also can bind the TNFSF member TRAIL, and it has been found that OPG inhibits TRAIL-induced apoptosis of Jurkat and LNCaP cells in culture.63, 64 In addition, it has been found that OPG acts as a survival factor for serum-deprived endothelial cells isolated from rat aorta.65 Using surface plasmon resonance, the affinities of OPG for RANK-L and TRAIL at 37 °C have been estimated at 6.7 nM and 400 nM, respectively.66 In contrast, the affinity of DR5 for TRAIL has been estimated at 2 nM. Although OPG has relatively low affinity for TRAIL, this may be offset by the abundance of OPG expression in some tissues. Currently, the biologic significance of TRAIL binding by OPG remains unclear.

The RANK-L/RANK/OPG Cytokine Axis in MM

RANK-L expression is increased in MM

Expression of RANK-L mRNA and protein, detected by in situ hybridization and immunohistochemistry, is increased dramatically in bone marrow biopsies from patients who have MM compared with specimens from patients who have MGUS or other hematologic malignancies.9, 11, 12 Evaluation of serial sections suggests that RANK-L is expressed by stromal cells, osteoblasts, and activated (CD30+) T cells in areas infiltrated by MM. RANK-L-expressing cells are rare in bone marrow that has been replaced completely by MM and in plasmacytoma specimens, stressing the importance of stroma in deregulated expression of RANK-L.

RANK-L also may be produced by malignant plasma cells in some patients with MM. Evidence presented by Oyajobi et al.67 and Croucher et al.68 suggests that RANK-L is expressed by the murine MM cell line 5T2 upon coculture with bone marrow stroma or upon infiltration into syngeneic bone marrow.68 In addition, flow cytometric evidence has suggested that CD138+ cells from at least a subset of patients with MM may be stained with antibody raised against RANK-L.10

However, in coculture, MM induced osteoclast development from murine bone marrow precursors only when it was supported by stroma/osteoblasts from wild type mice and not when it was supported by stroma from RANK-L −/− mice, indicating that these MM cell lines do not produce RANK-L directly, even after coculture with murine stromal cells/osteoblasts.9

OPG expression is decreased in MM

OPG stores are high in normal bone marrow, providing a natural buffer against aberrantly expressed RANK-L. Immunohistochemistry of bone marrow specimens from normal individuals demonstrates intense staining by anti-OPG antibodies in stromal cells and, particularly, in megakaryocytes. In contrast, in MM-infiltrated bone marrow, staining by anti-OPG antibodies is reduced markedly, and many megakaryocytes are negative or are only trace positive for OPG expression.9, 11 A profound decrease in OPG stores is supported by the report of decreased serum OPG levels in patients with MM comparison with age-matched controls.13

Coculture studies indicate that MM cell lines not only reduce constitutive OPG expression by human stromal cell lines but also interfere with OPG up-regulation in response to TGF-β, in part through the induction of SMAD7.9 Furthermore, addition of MM cell lines counteracts the ability of exogenous OPG to limit RANK-L-induced osteoclastogenesis from murine bone marrow precursors. The mechanisms responsible for this subversion of OPG function are unknown, but they may involve the ability of syndecan-1, which is expressed at high levels on the surface of malignant and nonmalignant plasma cells, to bind the heparin-binding domain of OPG.69, 70 These results suggest that subversion of OPG is an important element in disruption of the RANK-L/RANK/OPG cytokine axis by MM, resulting in increased osteoclastogenesis (Fig. 1B).

Use of RANK-Fc to control MM-associated bone destruction

The importance of RANK-L/OPG deregulation to the development of MM-associated bone destruction can be demonstrated using RANK-Fc, a recombinant RANK-L antagonist. RANK-Fc is a fusion of the extracellular domain of RANK (amino acids 22–201) with the constant region of human immunoglobulin G1 (IgG1) and is prepared using a baculovirus expression system in Sf9 cells (Fig. 2).71 RANK-Fc has the potential advantage over OPG of greater specificity for RANK-L. This may be significant, because it has been shown that MM undergoes apoptosis in response to TRAIL.72, 73 Fusion of RANK to the constant region of hIgG1 dictates homodimerization, which probably increases its avidity for RANK-L.

RANK-Fc Blocks MM-Induced Osteoclastogenesis in Vitro

Once it is bound by RANK-Fc, RANK-L is unable to bind to cell-expressed RANK. In vitro, osteoclasts develop in cocultures of CSF-1-treated bone marrow with MM in the presence of wild type stroma/osteoblasts. The addition of RANK-Fc to these cocultures results in dose dependent inhibition of osteoclast formation.9

Activity of RANK-Fc in Animal Models of Human MM

The requirement for RANK-L in the development of MM-induced bone destruction and the ability of RANK-Fc to prevent this complication can be demonstrated in vivo in established murine models.

Model 1: SCID/ARH-77 mice

Mice with severe combined immunodeficiency (SCID) that are injected intravenously with the human plasma cell leukemia-derived ARH-77 cell line develop osteolytic lesions (100% of mice) and hind limb paralysis (80% of mice).74 We injected SCID/ARH-77 mice with either RANK-Fc or hIgG1 (λ) (200 μg intravenously three times per week) starting the day after injection of ARH-77. At 6 weeks, RANK-Fc-treated mice exhibited significantly less bone turnover, as measured radiographically or by urinary excretion of cross-linked deoxypyridinoline (P < 0.01), and did not develop hind limb paralysis (P < 0.01). In contrast to the vertebral bodies of hIgG1-treated mice, which were infiltrated at autopsy by ARH-77 cells and showed macroscopic evidence of bony destruction and tumor growth, vertebral bodies from the RANK-Fc-treated animals appeared intact. Both groups had comparable titers of hIgG1 (κ), the antibody produced by the ARH-77 cell line, and comparable extraosseous growth of ARH-77 cells.

Model 2: SCID-hu-MM mice

In this model of primary human MM, SCID mice implanted with human fetal bone xenografts are injected with bone marrow from patients with MM.75, 76 The human xenograft becomes eroded by osteoclasts as the MM cells engraft, demonstrating pathology similar to that seen in MM patients. Pairs of SCID-hu-MM mice were generated, with each pair receiving aliquots of bone marrow from a single individual. Treatment with either RANK-Fc or hIgG1 (200 μg intravenously three times per week) began when titers of the MM paraprotein were detected in both mice of a pair, the mouse with the higher titer receiving RANK-Fc. RANK-Fc treatment protected against osteolysis of the xenograft, which was evident radiographically as severe osteopenia in all three hIgG1-treated mice but not in the RANK-Fc-treated mice.9, 77 When the xenografts were removed and examined histologically, significantly fewer osteoclasts were seen in xenografts from mice that were treated with RANK-Fc compared with the mice that received IgG1 (P < 0.001). Examination by immunohistochemistry indicates that RANK-Fc administration also reduced RANK-L and restored OPG expression in the xenografts. Unlike the SCID/ARH-77 mice, administration of RANK-Fc to SCID-hu-MM mice caused a marked reduction in serum paraprotein and a qualitative reduction in tumor burden.

The effects of RANK-Fc treatment in both models clearly demonstrate the potential of RANK-L antagonism to prevent MM-associated bone destruction in vivo. More importantly, the ability of RANK-Fc treatment to affect serum paraprotein and reduce tumor burden in SCID-hu-MM mice, but not in SCID/ARH-77 mice, reflects the observed dependence of primary MM, but not plasma cell leukemia, on bone and bone marrow microenvironments for growth and survival.

Does RANK-Fc Directly Inhibit MM Progression?

RANK-L promotes the survival of osteoclasts and dendritic cells, in part, through the activation of NF-κB and AKT.52 In addition, it has been found that RANK-L provides an autocrine survival signal in T-cell leukemia.78 We have found that RANK expression can be detected by Western blot analysis in only 1 of 11 MM cell lines (KMS11). Signaling through this receptor appears intact, because engagement of RANK-L results in phosphorylation of IKB and AKT. Nonetheless, the addition of either RANK-L or RANK-Fc does not affect the survival in vitro of primary MM isolated by CD138 selection or the growth or survival of MM cell lines (including KMS11) exposed to dexamethasone or vincristine.

Thus, the effect of RANK-Fc on MM tumor burden in the SCID-hu-MM model does not appear to be explained by a direct effect of RANK-L or RANK-Fc on plasma cell survival. However, MM cells may be dependent on a RANK-Fc-sensitive cell population. RANK-L was described originally as a survival factor for bone marrow-derived dendritic cells, and Sze et al. have found that long-lived plasma cells of the spleen are dependent on CD11c+ dendritic cells.79, 80 Although overall dendritic cell development appears normal in RANK-L −/− and RANK −/− mice, it is possible that a subset of dendritic cells is RANK-L dependent.43, 44 Osteoclasts, which clearly are RANK-L dependent, also may support MM through the production of cytokines, including IL-6.

Does RANK-Fc Inhibit MM-Triggered Neovascularization?

Although no vascular phenotype has been described for RANK-L −/− or RANK −/− mice, it has been shown that RANK-L acts in synergy with vascular endothelial growth factor to stimulate proliferation of endothelial cells.81 In addition, RANK-L triggers neovascularization in either subcutaneously implanted Matrigel plugs or chick chorioallantoic membranes. Because endothelial cells can express RANK and RANK-L, it is possible that administration of RANK-Fc blocks autocrine stimulation of endothelial growth. However, the inability of RANK-Fc to impact extraosseous growth of ARH-77 in SCID mice or plasma cell leukemia in SCID-hu-MM mice argues against either a direct effect on MM or an effect on MM-induced neovascularization.69, 77

How Else Has RANK-Fc Been Used?

The efficacy of RANK-Fc also has been evaluated in a xenograft model of hypercalcemia of malignancy. In that model, SCID mice were implanted subcutaneously with the human cell line RWGT2, which was derived from a patient with squamous cell carcinoma of the lung.67 It was found that treatment with RANK-Fc ameliorated tumor-associated hypercalcemia but not tumor growth.

Is MM Dependent on Bone Turnover?

Given the critical role of RANK-L in osteoclastogenesis, it is likely that inhibition of MM progression by RANK-Fc is a reflection at least in part of its ability to inhibit bone resorption. This conclusion is supported by the ability of RANK-Fc to limit the growth of intramedullary MM but not extramedullary MM. Dependence of MM on bone turnover also is suggested by the ability of bisphosphonates to limit intramedullary growth of MM in SCID-hu mice.77

A link between bone resorption and MM growth is consistent with the established requirement for bone resorption to support B-cell development.82 Tagaya et al. have shown that IL-7-responsive pre-B cells are decreased in bone marrow of mice that are osteopetrotic as the result of deletions of csf-1, MITF, or c-fos.83 A complementary observation is that OPG −/− mice, which demonstrate heightened bone turnover, possess pro-B cells that are hyperresponsive to IL-7.84 Similarities between pro-B cells and long-lived plasma cells suggest that common mechanisms may underlie this apparent dependence on bone resorption: Both populations are dependent on the bone marrow environment, express little Pax-5, respond to SDF-1, are inhibited by estrogen, express syndecan-1, and may be dependent on wnt signaling in association with syndecan-1 expression.85, 86

Does OPG-Fc Have Similar Effects on MM?

The potential role of OPG-Fc in MM has been studied in the 5T2MM model.68 In that model, the 5T2MM cell line was injected into C57BL/KaLwRijHsd mice. An important difference between the effect of OPG-Fc in the 5T2MM model and our findings is that, although OPG-Fc also inhibited the development of bone disease, the decrease in mean serum paraprotein levels was only 25% and was not statistically significant. One critical difference between the 5T2MM cells and the primary human MM and human MM cell lines in our studies is the expression of RANK-L by 5T2MM cells. Although this may reflect a difference between murine MM and human MM and, thus, a limitation of the murine model, it ultimately may be important for prediction of the clinical efficacy of RANK-L antagonists if plasma cell RANK-L expression is detected in some patients with MM.

Do Bisphosphonates Affect the RANK-L/RANK/OPG Axis?

It has been reported that bisphosphonates induce apoptosis of MM in vitro and limit tumor growth in some animal models of MM.77, 87 Thus, although it has been difficult to demonstrate in clinical trial, bisphosphonates may be expected to limit intramedullary tumor growth in patients with MM.88, 89 Control of tumor growth by bisphosphonates, then, would be expected to restore homeostasis to the expression of RANK-L and OPG in patients with MM.

Several observations suggest that bisphosphonates also may have a direct effect on the RANK-L/OPG cytokine axis. Bisphosphonates promote osteoblast and osteocyte survival by inhibiting apoptosis of these cells.90, 91 In vitro, the bisphosphonate ibandronate stimulates osteoblast cell lines to secrete a heat-labile, proteinase K-inactivated soluble factor that can inhibit pit formation by rat osteoclasts.92 Although the authors did not characterize further the osteoclast-inhibitory activity, it has been shown recently that pamidronate and zoledronate stimulate the secretion of OPG by primary human osteoblasts93 and, thus, potentially may counteract the dramatic down-regulation of OPG observed in the bone marrow of patients with MM.

CONCLUSION

Despite its malignant nature, MM must interact with the bone marrow microenvironment to survive and proliferate.94 One consequence of this interaction is increased bone turnover, which is the result of heightened osteoclastogenesis due to enhanced stromal expression of RANK-L accompanied by reduced expression of OPG. The ability of RANK-Fc to inhibit MM progression in SCID-hu mice illustrates the importance of bone resorption to the survival of patients with MM.

For osteopenia or osteoporosis to develop, active destruction must exceed bone formation (uncoupling of bone homeostasis). It may be relevant that MM is exceedingly rare during active growth, when the balance strongly favors bone formation. For example, a review of the SEER data bases (http://canques.seer.cancer.gov) indicates that, in 1998, no diagnoses of MM were reported in persons age < 20 years. Furthermore, in patients aged 20–24 years, MM diagnoses were reported only among females; MM first appeared in males in the group aged 25–29 years. Because MM is more common in men in other age groups until old age, when survival may be a factor, the excess of women in the group aged 20–24 years is even more startling. One well-recognized difference in this age group is that young women complete bone growth and achieve adult bone mass 2–3 years earlier than do young men.

It is intriguing that, although there is a sizable plasma cell mass in gut-associated lymphoid tissue and spleen, MM rarely occurs at these sites. One explanation may be that the interaction of these cells with their stromal environment does not support the malignant process.

Ancillary