p38 MAPK in Myeloma Cells Regulates Osteoclast and ...cancerres.aacrjournals.org/content/canres/72/24/6393.full.pdfMicroenvironment and Immunology p38 MAPK in Myeloma Cells Regulates

Microenvironment and Immunology

p38 MAPK in Myeloma Cells Regulates Osteoclast andOsteoblast Activity and Induces Bone Destruction

Jin He1, Zhiqiang Liu1, Yuhuan Zheng1, Jianfei Qian1, Haiyan Li1, Yong Lu1, Jingda Xu1, Bangxing Hong1,Mingjun Zhang1, Pei Lin2, Zhen Cai3, Robert Z. Orlowski1, Larry W. Kwak1, Qing Yi1, and Jing Yang1

Abstractp38 mitogen-activated protein kinase (MAPK), which is constitutively activated in human myeloma, has

been implicated in bone destruction by this cancer, but the processes it recruits are obscure. In this study, weshow that p38 activity in myeloma inhibits osteoblast differentiation and bone formation, but also enhancesosteoclast maturation and bone resorption. p38 regulated the expression and secretion of the Wnt pathwayantagonist DKK-1 and the monocyte chemoattractant MCP-1. Attenuating p38, DKK-1, or MCP-1 were eachsufficient to reduce bone lesions in vivo. Although it is well known that DKK-1 inhibits osteoblastdifferentiation, we found that together with MCP-1, it could also promote osteoclast differentiation andbone resorption. The latter effects were mediated by enhancing expression of RANK in osteoclast progenitorcells and by upregulating secretion of its ligand RANKL from stromal cells and mature osteoblasts. Insummary, our study defined the mechanisms by which p38 signaling in myeloma cells regulates osteo-blastogenesis, osteoclastogenesis, and bone destruction. Our findings, which may have implications for boneinvasion by other cancers where p38 is elevated, strongly suggests that targeting p38 for inhibition mayoffer an effective therapeutic approach to treat osteolytic bone lesions in patients with myeloma. Cancer Res;72(24); 6393–402. �2012 AACR.

IntroductionBone destruction is a hallmark of multiple myeloma. More

than 80% of myeloma patients have osteolysis, which is char-acterized by pathologic fractures, severe bone pain, spinal cordcompression, and hypercalcemia. These symptoms can severe-ly compromise a patient's quality of life and performancestatus (1, 2). It has been proposed that myeloma cells activateosteoclast-mediated bone resorption and inhibit osteoblast-mediated bone formation (3–5), but themechanismunderlyingthe association of myeloma cells with bone lesions remainspoorly elucidated.Constitutive activation of p38 mitogen-activated protein

kinase (p38 MAPK) has been found in benign bone diseases

and malignant osteolytic tumors, including multiple myeloma(6–8). We recently discovered that p38 activity in myelomacells is a master contributor to osteolysis in multiple myeloma(9). Our results show that the majority of established myelomacell lines and primary myeloma cells from patients havehigh levels of phosphorylated p38 (p-p38). Injection of myelo-ma cells with high or detectable p38 levels into severecombined immunodeficient mice (SCID) and SCID-hu micenot only established myeloma but also caused severe lyticlesions in the murine and human bones; in contrast, injectionof myeloma cells with no detectable p38 activity only estab-lished myeloma. Furthermore, disruption of p38 activity inmyeloma cells by specific p38 shRNAs or inhibitors abrogatedmyeloma-induced bone lesions in mice, without affectingtumor growth, survival, or ability to home to the bones. Inthis study, we investigated the roles and mechanisms ofactivated tumor cell p38 in myeloma-mediated osteoblasto-genesis and osteoclastogenesis.

Our results show that constitutive activation of p38 inmyeloma cells leads to monocyte chemotactic protein-1(MCP-1) and dickkopf-1 (DKK-1) expression and secretion.The p38-upregulated DKK-1 inhibits osteoblastogenesis,whereas p38-upregulated DKK-1 and MCP-1 promote osteo-clast maturation and function via enhancing RANK/RANKLexpression and activating NF-kB, p38, and ERK signalingpathways in their progenitor cells. These studies elucidate anovel mechanism ofmyeloma cell p38-induced osteolytic bonelesions and provide a strong rationale for developing newstrategies targetingmyeloma cell p38 activity for the treatmentor prevention of myeloma bone disease.

Authors' Affiliations: 1Department of Lymphoma/Myeloma, Division ofCancer Medicine and Center for Cancer Immunology Research and2Department of Hematopathology, Division of Pathology and LaboratoryMedicine, The University of Texas MD Anderson Cancer Center, Houston,Texas77030,USA; and 3BoneMarrowTransplantationCenter, Departmentof Hematology, The First Affiliated Hospital, School of Medicine, ZhejiangUniversity, Hangzhou, China.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Q. Yi and J. Yang contributed equally to this work.

Corresponding Authors: Jing Yang, Department of Lymphoma/Mye-loma, The University of Texas MD Anderson Cancer Center, 1515Holcombe Boulevard, Unit 0903, Houston, TX 77030. Phone: 713-563-0357; Fax: 713-563-9241; E-mail: [email protected]; andQing Yi, E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-12-2664

�2012 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 6393

on June 10, 2018. © 2012 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 11, 2012; DOI: 10.1158/0008-5472.CAN-12-2664

http://cancerres.aacrjournals.org/

Materials and MethodsTumor cell lines and primary myeloma cells

The myeloma cell lines ARP-1 and MM.1S have beendescribed previously (10). Other myeloma cell lines werepurchased from American Type Culture Collection (ATCC).These cell lines were authenticated by short tandem repeatprofiling and by matching with the profile published inATCC. All myeloma cell lines were cultured in RPMI-1640supplemented with 10% FBS (Invitrogen). Primary myelomacells were isolated from bone marrow aspirates obtainedfrom patients during routine clinic visits by magnetic beadsorting for CD138þ cells (Miltenyi Biotec GmbH). Thestudy was approved by the Institutional Review Board atThe University of Texas MD Anderson Cancer Center(Houston, TX).

Plasmids and reagentsShort hairpin RNAs (shRNA) for p38 3 isoforms including

a, b, and g were purchased from Santa Cruz Biotechnologyand packed into the retroviral vector pSIREN-RetroQ (BDBiosciences, Clontech). Retroviral infections were conductedaccording to the manufacturer's instructions. Retroviralvector supernatants of the p38 shRNAs were pooled andused to infect myeloma cells at 1:4 dilution. Stable cell lineswere established in the presence of 1 mg/mL puromycin. Inaddition, siRNAs specific for p38 a, b, and g were purchasedfrom Santa Cruz Biotechnology. In the experiments, cellswere harvested, plated on a 24-well plate at a concentrationof 2 � 105 cells per well, and transiently transfected withpooled siRNAs or nonspecific/control siRNA at differentdoses using the Oligofectamine transfection reagent (Mirus)according to the manufacturer's instructions. The p38MAPK-specific inhibitors were purchased from Axon Med-chem BV. Recombinant DKK-1 and MCP-1 were purchasedfrom R&D Systems.

Mouse model, antibody treatment, and detection ofosteolytic bone lesions by radiography

CB.17 SCID mice were purchased from Harlan Labora-tories. All mice were maintained in the American Associ-ation of Laboratory Animal Care-accredited facilities, andthe studies were approved by the Institutional Animal Careand Use Committee of MD Anderson Cancer Center. Six- to8-week-old mice were inoculated intravenously with 1 � 106

myeloma cells. Serum was collected from the mice dailyduring the treatment and tested for myeloma-secretedM-proteins [human immunoglobulin (Ig)] or their lightchains by using ELISA. For antibody treatment, the micewere injected intraperitoneally with neutralizing antibodiesspecific for MCP-1 or DKK-1 or with control IgG twice perweek for 20 days. To measure lytic bone lesions, radio-graphs were obtained using a Faxitron X-ray cabinet (Fax-itron X-ray).

For immunohistochemical analysis, Western blotting anal-ysis, real-time PCR, identification of myeloma-derived cyto-kines in culture medium by cytokine array, measurements ofsoluble factors in culture medium and serum by ELISA, in vitroosteoclast formation and bone pit analysis, in vitro osteoblast

formation and function assays, and statistical analysis, seeSupplementary Experimental Procedures.

ResultsMyeloma cell p38 upregulates myeloma cell productionof osteolytic mediators DKK-1 and MCP-1

Recent studies have shown that myeloma cells releasemultiple cytokines or chemokines to the bone marrow micro-environment and regulate the process of bone remodeling (3).To elucidate the mechanisms underlying myeloma p38-medi-ated bone destruction, we specifically and stably knockeddown p38 using shRNAs (9) and examined cytokines andchemokines that are regulated by p38 and play a role in boneremodeling in myeloma cells (ARP-1 and MM.1S). A cytokinearray containing specific antibodies for 48 inflammatory cyto-kines and chemokines, the majority of which are knownregulators of bone remodeling, was used. As shown in Fig.1A, the levels of 10 osteoclastogenesis-associated cytokineswere lower in p38 shRNA-treated ARP-1 (p38 shRNA-ARP-1)cells than in vector control-treated ARP-1 (vector-ARP-1) cells.Among these cytokines, MCP-1, which has recently beenshown to induce osteoclast differentiation in prostate cancer(11), was highly expressed in vector-ARP-1 cells but signifi-cantly downregulated in p38 shRNA-ARP-1 cells (Fig. 1B; P <0.01). Real-time PCR (Fig. 1C) and ELISA (Fig. 1D) confirmedthe results (P < 0.001). Similar results were obtained withMM.1S cells (data not shown). Because DKK-1 and RANKL,2 important cytokines involved in bone resorption, were notincluded in the cytokine array, we examined their expression inthemyeloma cells. Both real-time PCR (Fig. 1C) and ELISA (Fig.1E) showed that the expression and secretion of DKK-1 aredownregulated in p38 shRNA-ARP-1 and p38 shRNA-MM.1Scells (P < 0.001, compared with controls). RANKL (data notshown) was not expressed by the cells. Consistent with theseresults, DKK-1 and MCP-1 were the main cytokines that werehighly secreted by primary myeloma cells with high or detect-able pp38 as compared with primary myeloma cells withundetectable pp38 (Fig. 1F; P < 0.01). These results clearlysuggest that DKK-1 and MCP-1 are the products of p38 inmyeloma cells and might be involved in the formation of bonelesions.

On the basis of these results, we focused onDKK-1 andMCP-1 in subsequent studies. First, to verify the relationshipbetween p38 and the expression of DKK-1 and MCP-1 inmyeloma cells, the p38-specific inhibitor SB203580 (SB20) orsiRNAs were used to inhibit or knockdown p38 in myelomacells. Western blot analysis showed that treatment of ARP-1 orMM.1S (data not shown) cells with SB20 significantly down-regulated, in a dose-dependent manner, the levels of phos-phorylated MAPKPK-2 (pMK2), a downstream kinase of p38,DKK-1, andMCP-1, but not p-p38 (Fig. 2A). Because SB20 doesnot inhibit the phosphorylation of p38 but inhibits phosphor-ylation of its downstream kinase (12), slightly elevated levels ofpp38 were observed in SB20-treated cells. Moreover, siRNAsspecific for p38 (Fig. 2B) or the p38 upstream kinase MKK3/6(Fig. 2C), but not control siRNAs (data not shown), alsosignificantly inhibited, in a dose-dependent manner, theexpression of DKK-1, MCP-1, p-p38, and p38 or pMKK3/6 and

He et al.

Cancer Res; 72(24) December 15, 2012 Cancer Research6394




MKK3/6, respectively, in ARP-1 and MM.1S cells. ARP-1 andMM.1S secretion of DKK-1 (Fig. 2D) and MCP-1 (Fig. 2E) intoculture supernatant was also inhibited by treatment with SB20or siRNAs specific for p38 or MKK3/6 (P < 0.01 to P < 0.001).More importantly, SB20 treatment also significantly inhibitedsecretion of DKK-1 (Fig. 2F) and MCP-1 (Fig. 2G) by primarymyeloma cells (P < 0.01).To examine the role of these cytokines in myeloma cell-

induced bone lesions in vivo, we measured and comparedthe levels of circulating DKK-1 and MCP-1 in SCID miceinjected with the myeloma cells. Before tumor inoculation,the levels of human DKK-1 (Fig. 3A) and MCP-1 (Fig. 3B) inthe mouse sera were undetectable. Eight weeks after tumorinjection, high levels of human DKK-1 and MCP-1 weredetected in mice injected with control ARP-1 or MM.1S(data not shown) cells, and significantly lower levels of the 2cytokines were observed in mice injected with p38 shRNA-

ARP-1 cells (P < 0.001). Immunohistochemical stainingrevealed positive staining for DKK-1 and MCP-1 in vec-tor-ARP-1 but not p38 shRNA-ARP-1 cells in the bonemarrow of myeloma-bearing mice (Fig. 3C). Similar resultswere also obtained in ARP-1- or MM.1S-xenografted SCIDmice (data not shown) and in primary myeloma cell-xeno-grafted SCID-hu mice treated with or without the p38inhibitor SD-169 (Fig. 3D). However, knocking down p38activity in myeloma cells did not affect their growth in themice (9).

Neutralizing antibodies against DKK-1 or MCP-1 wereused to confirm their roles in the formation of bone lesionsin vivo. SCID mice were injected intravenously with theparental ARP-1 or MM.1S cells, and beginning at week 4(when tumors were established, without bone destruction)received intraperitoneal injections of anti-DKK-1 (10 mg/kg)and/or anti-MCP-1 (20 mg/kg) antibodies twice a week for a

Figure 1. Identification of osteolytic mediators DKK-1 and MCP-1 as the targets of myeloma cell p38 activity. Representative images (A) and densitometricdata (B) of a cytokine array showing the profile of cytokine expression in the conditioned medium of vector- and p38 shRNA-ARP-1 and MM.1S (datanot shown) cells. Detection of mRNA expression in ARP-1 andMM.1S cells (C) by using real-time PCR and secreted MCP-1 (D) and DKK-1 (E) in conditionedmedium of the cells by using ELISA. F, real-time PCR showing elevated mRNA levels of MCP-1 and DKK-1, but not other cytokines, in primary tumorcells isolated from patients with myeloma with high or detectable p38 activity (p-p38þ; n ¼ 8) compared with primary tumor cells with undetectable p38activity (p-p38�; n ¼ 6). Representative results of 3 experiments are shown. ��, P � 0.01; ��, P � 0.001.

Myeloma p38 Regulates Osteoclasts and Osteoblasts

www.aacrjournals.org Cancer Res; 72(24) December 15, 2012 6395




total of 8 injections. Injections of equal amounts of IgG orPBS served as controls. Our results showed that treatmentwith antibodies against MCP-1 or DKK-1, but not controlIgG, significantly reduced lytic bone lesions (Fig. 3E; P < 0.01)without affecting tumor burdens in the mice (data notshown). Interestingly, treatment with antibodies againstboth DKK-1 and MCP-1 almost prevented the developmentof osteolytic bone lesions in most (8/10) of the myeloma-bearing mice. The neutralizing antibodies were also used inprimary myeloma-bearing SCID-hu mice, with similarresults (data not shown). Taken together, these findingsclearly indicate that myeloma cell-derived DKK-1 andMCP-1 play a critical role in mediating the development ofosteolytic bone lesions in vivo.

Myeloma cell p38 inhibits osteoblast differentiation andfunction via DKK-1

It is well known that bone remodeling is regulated byosteoblast-mediated bone formation and osteoclast-induced bone resorption (13). We have shown that myelomacell p38 inhibits osteoblastogenesis and enhances osteo-clastogenesis in vivo in myeloma-bearing SCID and SCID-humouse models. In this study, we examined the ability ofand mechanism by which tumor cell p38 downregulatesosteoblast differentiation and bone formation activity viaDKK-1 and/or MCP-1. The differentiation and maturityof osteoblasts can be monitored by their production ofalkaline phosphatase (ALP) in culture medium and their

expression of bone morphogenetic protein-2 (BMP-2) andosteocalcin.

The results showed that vector-ARP-1 cells (data not shown)or their conditioned medium suppressed the differentiation ofosteoblasts, as evidenced by large numbers of poorly differen-tiated cells that were BMP-2- and osteocalcin-negative (Fig.4A), low production of ALP (Fig. 4B; P < 0.001), and compro-mised bone formation activity, as determined by Alizarin Red-S staining assay (Fig. 4C; P < 0.01). In contrast, culturescontaining conditioned medium from p38 shRNA-ARP-1 cellsgenerated BMP-2- and osteocalcin-positive osteoblasts, withproduction of high levels of ALP and bone formation activity.Similarly, conditioned medium of MM.1S cells (data notshown) or primary myeloma cells (Fig. 4D; P < 0.001) withhigh or detectable p38 activity (Pt 9) suppressed osteoblastdifferentiation, as evident by the low production of ALP,whereas primary myeloma cells with low or undetectablep38 activity (Pt 10) had no such effects (Fig. 4E). Furtherstudies showed that DKK-1, but not MCP-1, secreted bymyeloma cells is mainly responsible for inhibited osteoblas-togenesis, because the addition of neutralizing antibodiesspecific for DKK-1, but not MCP-1 or control IgG (data notshown), to the conditionedmedium of vector-ARP-1 or MM.1S(data not shown) cells restored the generation of functionalosteoblasts (Fig. 4F; P < 0.001). Similarly, the addition ofexogenous DKK-1, but not MCP-1, to conditioned mediumfrom p38 shRNA-ARP-1 cells also inhibited osteoblastogenesis(Fig. 4G; P < 0.001). Taken together, these results clearly

Figure 2. p38 upregulates myeloma cell expression and secretion of DKK-1 and MCP-1. Western blot analysis showing protein expression of DKK-1,MCP-1, p-p38, p38, pMK2, pMKK3/6, or MKK3/6 in p38 inhibitor SB20-treated (A), p38 siRNA-treated (B), and MKK3/6 siRNA-treated ARP-1 cells (C).ELISA showing the levels of secreted DKK-1 (D) or MCP-1 (E) in p38 inhibitor SB20-treated, p38 siRNA-treated, and MKK3/6 siRNA-treated ARP-1and MM.1S cells. Shown are dose-dependent protein expression and secretion at 48 hours after the treatments. Secretion of DKK-1 (F) or MCP-1 (G)by primary cells (from 6 myeloma patients) in culture with PBS or with p38 inhibitor SB20. Representative results of 3 independent experiments are shown.��, P � 0.01; ��, P � 0.001.

He et al.





indicate that DKK-1 was responsible for myeloma p38-medi-ated inhibition of osteoblast differentiation and activity.

Myeloma cell p38 activates osteoclast differentiation andbone resorption activity via DKK-1 and MCP-1Next, we investigated the impact and mechanism of tumor

cell p38 activity on osteoclast differentiation and bone resorp-tion in vitro. To mimic the in vivo bone marrow microenvi-ronment, a Transwell coculture system was used in which1 � 105 osteoclast precursors/mL were seeded on the bottomof the wells and 10-fold numbers of mesenchymal stem cells(MSC) were planted in the Transwell inserts. The cells werecocultured in osteoclast medium, with or without RANKL. Insome experiments, conditioned media of different ARP-1 orMM.1S cells were added to the cocultures. Mature multinu-clear osteoclasts were generated in cocultures with addition ofRANKL and in cocultures without RANKL in the presence ofconditioned medium of vector-ARP-1 cells but not in thepresence of p38 shRNA-ARP-1 cell-conditioned medium (Fig.5A and B; P < 0.001). As TRAP5b is the active component ofTRAP and is secreted only bymature osteoclasts, wemeasuredthe levels of TRAP5b in media from the cocultures. Theaddition of conditioned medium from vector-ARP-1, but notfrom p38 shRNA-ARP-1 cells upregulated TRAP5b secretion in

a dose-dependent manner (Fig. 5C; P < 0.01 to P < 0.001).Similar results were obtained from cocultures of MM.1S (datanot shown) or primary myeloma cells with high, but not withlow, p38 activity (Fig. 5D; P < 0.001).

Furthermore, osteoclasts generated in the cocultures withconditioned medium of ARP-1 cells resorbed efficiently miner-alized matrices (Fig. 6A; top) and bone slices (bottom). Analysisof the resorbed area indicates that the osteoclasts formedlarger resorption areas (Fig. 6B; P < 0.001) and deeper resorbedbone pits (Fig. 6C; P < 0.001) and secreted more soluble CTX-1(Fig. 6D; P < 0.001) in comparison with cells generated incocultures in medium devoid of RANKL or with the additionof conditioned medium from p38 shRNA-ARP-1 cells.

Because myeloma cells secreted DKK-1 and MCP-1, weexamined whether these cytokines were involved in osteoclas-togenesis. Antibodies specific for DKK-1 and/or MCP-1 wereadded to the coculture of osteoclast precursors and MSCs inthe presence or absence of conditioned medium of parentalARP-1 cells. After a 7-day culture, the telomeric repeat ampli-fication protocol (TRAP)-positive, multinuclear osteoclastswere counted and the levels of TRAP5b in the medium weremeasured by ELISA. As shown in Fig. 6E, the addition of eitheranti-DKK-1 or anti-MCP-1 antibodies partially inhibited thegeneration of mature osteoclasts, and the addition of both

Figure 3. Myeloma cell p38 activityleads to production of DKK-1 andMCP-1 in myeloma-bearing mice.Levels of circulating DKK-1 (A) orMCP-1 (B) in serum at 0 week(0 wks) and 8 weeks (8 wks). C,immunohistochemical staining forDKK-1 (top) and MCP-1 (bottom)expression in established myelomacells in mice injected with vector- orp38 shRNA-ARP-1 cells at 8 weeksafter tumor inoculation. D,immunohistochemical staining forDKK-1 (top) and MCP-1 (bottom)expression in established primarymyeloma cells in SCID-hu micetreated without (vehicle) or with thep38-specific inhibitor SD-169 at 8weeks after tumor injection. Scalebar, 20 mm; scale bar for insets, 100mm. E, representative radiographsshowing reduced osteolytic bonelesions in the femurs of ARP-1-bearing mice (5 per group) treatedwith antibodies against MCP-1and/or DKK-1 comparedwith controlIgG. Representative results of3 independent experiments areshown. ��, P � 0.01; ��, P � 0.001.






antibodies had a greater effect (P < 0.001). Alternatively, theaddition of exogenous DKK-1, MCP-1, and especially both, tothe conditioned medium of p38 shRNA-ARP-1 cells restoredthe generation of mature osteoclasts in the cocultures devoidof RANKL (Fig. 6F; P < 0.001). Taken together, these resultsclearly indicate that DKK-1 and MCP-1 are responsible formyeloma p38-mediated activation of osteoclast differentiationand activity in vitro and in vivo.

Myeloma cells activate RANK/RANKL signaling inosteoclast precursors via p38-upregulated DKK-1 andMCP-1

Next, we examined signaling pathways in osteoclast activa-tion in relation to tumor cell p38 activity. We focused on NF-kB, PI3K/Akt, and MAPKs, such as ERK, p38, and c-jun-NH2-

kinase (JNK), because these pathways have been reported to beinvolved in the regulation of osteoclast differentiation (14). Inthe experiments, osteoclast precursors were treated withconditioned medium of ARP-1 or MM.1S (data not shown)cells for 3 days. The cells were then collected, and phosphor-ylation of IkBa, Akt, ERK, p38, and JNK was detected by usingWestern blot analysis. As shown in Fig. 7A, the addition oftumor conditioned medium from vector-ARP-1, but not p38shRNA-ARP-1 cells, upregulated the phosphorylation of IkBa,p38, ERK1/2 and Akt, and slightly decreased the phosphory-lation of JNK during osteoclast differentiation. The levels ofnonphosphorylated kinases (data not shown) and b-actin wereunchanged. These signaling pathways were essential for oste-oclast differentiation, because the addition of kinase inhibitorsspecific for IkBa, ERK, or p38 to the cultures significantly

Figure 4. Activated tumor cell p38 inhibits osteoblast differentiation and bone formation. Generation of osteoblasts (OB) from normal MSCs in osteoblastmediumwith addition of conditioned medium (CM) of vector-ARP-1 (V) or p38 shRNA-ARP-1 (SH) cells. Mature osteoblasts are identified as BMP-2þ (A, top)and osteocalcinþ (A, bottom) cells that produce large amounts of ALP (B) and display osteoblast bone formation activity, with strong staining of AlizarinRed-S (C). D, similarly, addition of conditioned medium from primary myeloma cells with high/detectable p38 activity, but not primary myeloma cellswith low/undetectable p38 activity, inhibited osteoblast differentiation as evidenced by inhibited ALP production. E, Western blotting showing the levelsof p-p38, p38, and b-actin in patient 9 (Pt 9; high p-p38) and patient 10 (Pt 10; low p-p38). F, effects of DKK-1- and MCP-1–neutralizing antibodieson the differentiation of mature osteoblasts in culture with addition of conditioned medium of vector-ARP-1 cells (CM-V). G, effects of exogenous DKK-1and MCP-1 on the differentiation of mature osteoblasts in culture with addition of conditioned medium from p38 shRNA-ARP-1 cells (CM-p38h). The levelsof ALP were indicators of osteoblast differentiation and maturation. Representative results from 4 independent experiments are shown. ��, P � 0.01;��, P � 0.001.

He et al.





abrogated the generation of osteoclasts induced by the tumor-conditioned medium (Fig. 7B; P < 0.001).Recent studies have shown that IkBa, ERK1/2, and p38

signaling pathways are activated by RANKL, an osteoclastactivator, which binds to its receptor RANK on osteoclastprecursors. However, as myeloma cells express low levels ofRANKL (data not shown), we hypothesized that myeloma cellsactivate RANK/RANKL signaling pathways in osteoclast pre-cursors via secreted DKK-1 and MCP-1. By using Western blotanalysis, we examined the expression of RANK in CD14þ

monocytes and in M-CSF–induced osteoclast precursors trea-ted with or without conditioned medium of vector- or p38shRNA-ARP-1 or MM.1S (data not shown) cells. The condi-tioned medium of vector-ARP-1 cells significantly increasedRANK expression in both monocytes and osteoclast precur-sors, whereas the medium of p38 shRNA-ARP-1 cells onlyslightly upregulated RANK expression (Fig. 7C and 7D). Ourstudies also showed that conditionedmediumof vector-ARP-1,but not p38 shRNA-ARP-1, cells significantly upregulated theexpression (Fig. 7E) and secretion (Fig. 7F; P < 0.001) of RANKLby MSCs and mature osteoblasts. Hence, these results supportour hypothesis that myeloma cells modulate the expression ofRANK and RANKL in osteoclasts and marrow stromal cells.To determine whether DKK-1 and MCP-1 mediate the

myeloma p38-induced increase of RANK/RANKL expression,neutralizing antibodies against these cytokines were used. Asshown in Fig. 7G, antibodies against MCP-1, but not DKK-1,significantly abrogated conditioned medium-induced expres-sion of surface RANK on CD14þ monocytes (P < 0.01). On theother hand, anti-DKK-1 antibody remarkably inhibited tumor-

induced RANKL secretion in MSCs (Fig. 7H; P < 0.001). Theseresults indicate that tumor-derived DKK-1 upregulates RANKLexpression and secretion from stromal cells, and tumor-derived MCP-1 enhances RANK expression in the progenitorsof osteoclasts, both of which activate RANK/RANKL-mediatedsignaling pathways in the progenitor cells, leading to osteoclastdifferentiation and activation.

DiscussionThe p38 MAPK is an important signaling pathway involved

in cell growth, survival, differentiation and inflammatoryresponse, and can be activated by a variety of cellular stresses,including inflammatory cytokines, endotoxin, ultraviolet light,and growth factors (6–8). Functionally, p38 mediates theproduction of proinflammatory cytokines and other factorssuch as TNF-a (15, 16) and soluble RANKL (17, 18). Preclinicalstudies indicate that inhibition of p38 suppresses the inductionof inflammatory, angiogenic, and proapoptotic cytokines frombone marrow stromal cells (19). Moreover, p38 signaling playsan important role in the pathogenesis of multiple myeloma,including myeloma growth and survival and drug resistance(20, 21). Our previous study has shown that activated p38signaling in myeloma cells induces osteolytic bone lesions inmyeloma by inhibiting osteoblastogenesis and enhancingosteoclastogenesis (9). The current study further elucidatesthat myeloma cells express and secrete DKK-1 and MCP-1 as aresult of constitutive activation of myeloma p38 signaling.Consequently, p38-upregulated DKK-1 and MCP-1 enhanceRANK/RANKL expression and activate NF-kB, p38, and ERKsignaling pathways in the progenitors of osteoclasts, leading to

Figure 5. Activated tumor cellp38 enhances osteoclastdifferentiation. Shown is thegeneration of osteoclasts incocultures of osteoclast precursorswith MSCs in osteoclast mediumwithout (medium) or with RANKL, orin osteoclast medium in the absenceof RANKL with addition ofconditioned medium (CM) of vector-or p38 shRNA-ARP-1 cells. Matureosteoclasts are defined bymorphology (A) and high numbers ofTRAPþ multinuclear cells (B), withproduction of high levels of solubleTRAP5b (C). D, similarly, the additionof conditioned medium of primarymyeloma cells with high/detectablep38 (Pt 9) activity, but not that ofprimary myeloma cells withlow/undetectable p38 (Pt 10) activity,promoted the differentiation ofmature osteoclasts as evidencedby the production of high levels ofTRAP5b. ��, P� 0.01; ��, P� 0.001.






osteoclast activation and bone resorption, whereas p38-upre-gulated DKK-1 inhibits osteoblast differentiation and boneformation, all of which result in bone destruction (Supplemen-tary Fig. S1).

DKK-1 is a well-known inhibitor of osteoblast differenti-ation. Elevated levels of DKK-1 have been reported to inducebone loss in patients with inflammatory rheumatoid arthri-tis (21) and myeloma (13, 26). Previous studies have shownthat DKK-1, produced by myeloma cells, inhibits osteoblastdifferentiation (22–24) and may enhance osteoclast differ-entiation (25) and bone lesions (26) by upregulating RANKLexpression on stromal cells. However, the mechanism bywhich myeloma cells produce DKK-1 is unclear. By theanalysis of Western blotting, real-time PCR and ELISA, wefor the first time show that constitutively activated myelomacell p38 signaling upregulates DKK-1 expression and secre-

tion. Inhibition or shRNA knockdown of myeloma p38significantly downregulates DKK-1 secretion in vitro and inmyeloma-bearing mice. Furthermore, we also observe thatmyeloma cell p38 signaling induces the expression andsecretion of MCP-1, a chemokine that recruits monocytes,T cells, and dendritic cells to sites of infection or tissueinjury. MCP-1 is upregulated during osteoclastogenesis,promotes the fusion of hematopoietic precursors to matureosteoclasts (27), and determines osteoclast behavior ininflammatory osteoporosis via binding to receptor CCR2(28). MCP-1 is mainly released from bone marrow stromalcells (29). However, our results show that myeloma cells alsosecrete MCP-1 in vitro and in myeloma-bearing mice, as aresult of constitutive activation of myeloma cell p38. Knock-ing down myeloma cell p38 reduces MCP-1 expression bymyeloma cells.

Figure 6. Activated tumor cell p38 enhances osteoclast bone resorption. Osteoclasts were generated in cocultures of osteoclast precursors with MSCs inosteoclast medium without (medium) or with RANKL, or in osteoclast medium in the absence of RANKL with addition of conditioned medium (CM) ofvector- or p38 shRNA-ARP-1 cells. A, resorption pit analysis showing pit formation on mineralized matrices (top) or dentin slices (bottom) by matureosteoclasts generated in osteoclast medium devoid of RANKL with the addition of conditioned medium of vector-ARP-1 (Vector), but not of p38shRNA-ARP-1 (p38 shRNA) cells. Color contours represent the depth of the resorption pits on dentin slices. Summarized data of resorption area perosteoclasts on mineralized matrices (B), average pit depth on dentin slices (C), and the levels of bone degradation product CTX-1 in culture medium (D).Representative results of 5 experiments are shown. E, effects of DKK-1- and MCP-1–neutralizing antibodies on the differentiation of matureosteoclasts in the cocultures with addition of conditioned medium from vector-ARP-1 cells (CM-V), and effects of exogenous DKK-1 and MCP-1 on thedifferentiation of mature osteoclasts in cocultures with addition of conditioned medium from p38 shRNA-ARP-1 cells (CM-p38sh; F). The levels of secretedTRAP5bwere used as indicators of osteoclast differentiation andmaturation. Representative results from 4 independent experiments are shown. ��,P� 0.01;��, P � 0.001.

He et al.





In addition, our results for the first time observe that p38-increased DKK-1 not only inhibits osteoblastogenesis, but alsoinduces osteoclast differentiation and bone resorption byenhancing RANKL secretion from bone marrow stromal cells.On the other hand, p38-increased MCP-1 upregulates RANKexpression in the progenitors of osteoclasts and activatedRANK/RANKL signaling pathways such as NF-kB, ERK1/2,and p38 in the cells, leading to osteoclast activation and boneresorption. Neutralizing both DKK-1 and MCP-1 reduces mye-loma-induced osteoclast activation in vitro and bone lesions invivo. These results provide an alternative explanation ofRANKL/RANK perturbation in myeloma.Taken together, our findings indicate that the development

of myeloma-induced bone lesions is initiated and regulated byactivated p38 signaling in myeloma cells, but is sustainedthrough the interactions between myeloma cells and sur-rounding bone marrow cells via p38 upregulating cytokinesor chemokines by myeloma cells. These results shed light onthe regulation of osteolytic cytokine (such as DKK-1 andMCP-1) production by myeloma cell p38 activity and identify novel

targets for treating myeloma-induced osteolytic bone lesionsin patients.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: Z. Liu, Z. Cai, Q. Yi, J. YangDevelopment of methodology: Z. Liu, Y. Zheng, M. Zhang, Z. CaiAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): J. He, Z. Liu, J. Qian, Y. Lu, J. Xu, B. Hong, P. Lin,Z. Cai, R.Z. OrlowskiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): Z. Liu, B. Hong, Z. CaiWriting, review, and/or revision of themanuscript: Z. Cai, R.Z. Orlowski, L.W. Kwak, Q. Yi, J. YangAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): J. Qian, H. Li, Z. Cai, L.W. KwakStudy supervision: Z. Cai, Q. Yi, J. Yang

AcknowledgmentsThe authors thank Dr. Tomasz Zal for technical support with confocal

microscopy and analysis, our Departmental Myeloma Tissue Bank for patientsamples, and Ms. Alison Woo and Dawn Chalaire for providing editorialassistance.

Figure 7. Myeloma cell p38 activates RANK/RANKL signaling in osteoclast precursors via secreted DKK-1 and MCP-1. A, Western blot analysisshowing upregulated levels of phosphorylated (p) IkBa, p38, and ERK, but not JNK and Akt, in osteoclast precursors treated with conditioned medium ofvector-ARP-1 (V) or p38 shRNA-ARP-1 (p38sh) cells. Cells cultured in medium (Med) alone served as control. B, effects of inhibitors specific for IkBa (IkBinh,15 mmol/L), p38 (p38inh, SB202190, 50 nmol/L), or ERK (ERKinh, PD98059, 50 mmol/L) on the generation ofmature osteoclasts in cocultures with the additionof conditioned medium (CM) of vector-ARP-1 as detected by the levels of secreted TRAP5b. Effects of conditioned medium of vector- ARP-1 (V) orp38 shRNA-ARP-1 (p38sh) cells on the expression of RANK on monocytes (Mo) and osteoclast precursors (PreOC) detected by flow cytometry(monocytes only; C) or Western blot analyses (D) and on the expression of RANKL (E) or secretion of soluble RANKL (sRANKL; F) by MSCs and matureosteoblasts (OB). G, effects of MCP-1- and DKK-1–neutralizing antibodies on RANK expression detected by flow cytometry on monocytes or on thesecretion of soluble RANKL by MSCs, measured by ELISA (H), induced by conditioned medium of vector- ARP-1 (CMV) or p38 shRNA-ARP-1 (CM-p38sh)cells. Representative results of 5 independent experiments are shown. ��, P � 0.01; ��, P � 0.001.






Grant SupportThis work was supported by National Cancer Institute K99/R00 CA137158 (J.

Yang), American Society of Hematology (J. Yang), National Cancer InstituteR01s CA138402 and CA138398 and P50 CA142509 (Q. Yi), the Leukemia andLymphoma Society Translational Research Grants, Multiple Myeloma ResearchFoundation (Q. Yi), Commonwealth Foundation for Cancer Research (Q. Yi), andby funds from the University Cancer Foundation and the Center for TargetedTherapy of The University of Texas MD Anderson Cancer Center (Q. Yi).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact.

Received July 5, 2012; revised September 21, 2012; accepted October 6, 2012;published OnlineFirst October 11, 2012.

References1. Roodman GD. Pathogenesis of myeloma bone disease. Leukemia

2009;23:435–41.2. Roodman GD. Novel targets for myeloma bone disease. Expert Opin

Ther Targets 2008;12:1377–87.3. Roodman GD. Bone-breaking cancer treatment. Nat Med 2007;13:

25–6.4. Roodman GD. Role of the bone marrow microenvironment in multiple

myeloma. J Bone Miner Res 2002;17:1921–5.5. Roodman GD. Mechanisms of bone resorption in myeloma. J Muscu-

loskelet Neuronal Interact 2003;3:271–2; discussion 92–4.6. Wang S, Hong S, Yang J, Qian J, Zhang X, Shpall E, et al. Optimizing

immunotherapy in multiple myeloma: Restoring the function ofpatients' monocyte-derived dendritic cells by inhibiting p38 or acti-vating MEK/ERK MAPK and neutralizing interleukin-6 in progenitorcells. Blood 2006;108:4071–7.

7. WangS,Yang J,Qian J,WezemanM,KwakLW,YiQ. Tumor evasionofthe immune system: inhibiting p38 MAPK signaling restores thefunction of dendritic cells in multiple myeloma. Blood 2006;107:2432–9.

8. Xie J,Qian J, Yang J,WangS, FreemanME III, YiQ.Critical roles ofRaf/MEK/ERK and PI3K/AKT signaling and inactivation of p38MAP kinasein the differentiation and survival of monocyte-derived immature den-dritic cells. Exp Hematol 2005;33:564–72.

9. Yang J, He J, Wang J, Cao Y, Ling J, Qian J, et al. Constitutiveactivation of p38 MAPK in tumor cells contributes to osteolytic bonelesions in multiple myeloma. Leukemia 2012;26:2114–23.

10. Yang J, Zhang X, Wang J, Qian J, Zhang L, Wang M, et al. Anti beta2-microglobulin monoclonal antibodies induce apoptosis in myelomacells by recruiting MHC class I to and excluding growth and survivalcytokine receptors from lipid rafts. Blood 2007;110:3028–35.

11. Lu Y, Cai Z, Xiao G, Keller ET, Mizokami A, Yao Z, et al. Monocytechemotactic protein-1 mediates prostate cancer-induced boneresorption. Cancer Res 2007;67:3646–53.

12. TongL,PavS,WhiteDM,RogersS,CraneKM,CywinCL, et al. A highlyspecific inhibitor of human p38 MAP kinase binds in the ATP pocket.Nat Struct Biol 1997;4:311–6.

13. Roodman GD. Treatment strategies for bone disease. Bone MarrowTransplant 2007;40:1139–46.

14. Lee ZH, Kim HH. Signal transduction by receptor activator of nuclearfactor kappa B in osteoclasts. Biochem Biophys Res Commun2003;305:211–4.

15. Rossa C, Ehmann K, Liu M, Patil C, Kirkwood KL. MKK3/6-p38 MAPKsignaling is required for IL-1beta and TNF-alpha-induced RANKLexpression in bone marrow stromal cells. J Interferon Cytokine Res2006;26:719–29.

16. Zwerina J, Hayer S, Redlich K, Bobacz K, Kollias G, Smolen JS, et al.Activation of p38 MAPK is a key step in tumor necrosis factor-medi-ated inflammatory bone destruction. Arthritis Rheum 2006;54:463–72.

17. Huang H, Ryu J, Ha J, Chang EJ, Kim HJ, Kim HM, et al. Osteoclastdifferentiation requires TAK1 andMKK6 for NFATc1 induction andNF-kappaB transactivation byRANKL.Cell DeathDiffer 2006;13:1879–91.

18. SundaramK,Nishimura R, Senn J, Youssef RF, LondonSD, ReddySV.RANK ligand signalingmodulates thematrixmetalloproteinase-9 geneexpression during osteoclast differentiation. Exp Cell Res 2007;313:168–78.

19. Navas T, Zhou L, Estes M, Haghnazari E, Nguyen AN, Mo Y, et al.Inhibition of p38alpha MAPK disrupts the pathological loop of proin-flammatory factor production in the myelodysplastic syndrome bonemarrow microenvironment. Leuk Lymphoma 2008;49:1963–75.

20. Nguyen AN, Stebbins EG, Henson M, O'Young G, Choi SJ, Quon D,et al. Normalizing the bone marrow microenvironment with p38 inhib-itor reduces multiple myeloma cell proliferation and adhesion andsuppresses osteoclast formation. Exp Cell Res 2006;312:1909–23.

21. Hideshima T, Akiyama M, Hayashi T, Richardson P, Schlossman R,Chauhan D, et al. Targeting p38 MAPK inhibits multiple myeloma cellgrowth in the bone marrow milieu. Blood 2003;101:703–5.

22. Qiang YW, Chen Y, Stephens O, Brown N, Chen B, Epstein J, et al.Myeloma-derived Dickkopf-1 disrupts Wnt-regulated osteoprotegerinand RANKL production by osteoblasts: a potential mechanism under-lying osteolytic bone lesions in multiple myeloma. Blood 2008;112:196–207.

23. TianE, ZhanF,WalkerR, RasmussenE,MaY,Barlogie B, et al. The roleof theWnt-signaling antagonist DKK1 in the development of osteolyticlesions in multiple myeloma. N Engl J Med 2003;349:2483–94.

24. Qiang YW, Barlogie B, Rudikoff S, Shaughnessy JD Jr. Dkk1-inducedinhibition of Wnt signaling in osteoblast differentiation is an underlyingmechanism of bone loss in multiple myeloma. Bone 2008;42:669–80.

25. Qiang YW, Chen Y, Brown N, Hu B, Epstein J, Barlogie B, et al.Characterization of Wnt/beta-catenin signalling in osteoclasts in mul-tiple myeloma. Br J Haematol 2010;148:726–38.

26. Heath DJ, Chantry AD, Buckle CH, Coulton L, Shaughnessy JD Jr,Evans HR, et al. Inhibiting Dickkopf-1 (Dkk1) removes suppression ofbone formation and prevents the development of osteolytic bonedisease in multiple myeloma. J Bone Miner Res 2009;24:425–36.

27. Kim MS, Day CJ, Morrison NA. MCP-1 is induced by receptoractivator of nuclear factor-{kappa}B ligand, promotes human oste-oclast fusion, and rescues granulocyte macrophage colony-stim-ulating factor suppression of osteoclast formation. J Biol Chem2005;280:16163–9.

28. Binder NB, Niederreiter B, Hoffmann O, Stange R, Pap T, Stulnig TM,et al. Estrogen-dependent and C-C chemokine receptor-2-dependentpathways determine osteoclast behavior in osteoporosis. Nat Med2009;15:417–24.

29. Lee SK, Choi BK, Kang WJ, Kim YH, Park HY, Kim KH, et al. MCP-1derived from stromal keratocyte induces corneal infiltration of CD4þ Tcells in herpetic stromal keratitis. Mol Cells 2008;26:67–73.

He et al.





2012;72:6393-6402. Published OnlineFirst October 11, 2012.Cancer Res Jin He, Zhiqiang Liu, Yuhuan Zheng, et al. Activity and Induces Bone Destructionp38 MAPK in Myeloma Cells Regulates Osteoclast and Osteoblast

Updated version

10.1158/0008-5472.CAN-12-2664doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2012/10/11/0008-5472.CAN-12-2664.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/72/24/6393.full#ref-list-1

This article cites 29 articles, 7 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/72/24/6393.full#related-urls

This article has been cited by 5 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/72/24/6393To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-12-2664

http://cancerres.aacrjournals.org/content/suppl/2012/10/11/0008-5472.CAN-12-2664.DC1

http://cancerres.aacrjournals.org/content/72/24/6393.full#ref-list-1

http://cancerres.aacrjournals.org/content/72/24/6393.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/72/24/6393


Documents

p38 MAPK in Myeloma Cells Regulates Osteoclast and ...cancerres.aacrjournals.org/content/canres/72/24/6393.full.pdfMicroenvironment and Immunology p38 MAPK in Myeloma Cells Regulates