mesenchymal stromal cells promote tumor growth through the

INTRODUCTIONGrowth of solid tumors requires for-

mation of the tumor stroma, which sup-plies oxygen and nutrients to tumor cells(1). The tumor stroma is composed of ex-tracellular matrix and various mesenchy-mal cell types, including macrophages,endothelial cells, lymphocytes, pericytes,fibroblasts and myofibroblasts (2). Thesestromal cells communicate with tumorcells both through direct contact and

through paracrine signaling mechanisms,mediated by secretion of soluble factors,including cytokines, chemokines andgrowth factors (3–7). Interactions be-tween tumor cells and stromal cells regu-late tumor growth, invasion, metastasisand angiogenesis (3–7). Among stromalcells, tumor-associated fibroblasts havebeen shown to be associated with in-creases in tumor growth and metastaticpotential, leading to a poor prognosis

(8,9). Tumor-associated fibroblasts andmyofibroblasts originate from multiplesources and range from migratory neigh-boring cells to distant invading cells (10).Data from human tumors and mousetumor models suggest that at least a por-tion of the stromal cells are derived fromthe bone marrow (11–13).

Mesenchymal stromal cells (MSCs),also called mesenchymal stem cells, arepluripotent progenitor cells that have thecapability to differentiate into chondro-cytes, adipocytes and osteoblasts, amongother types of cells (14). Although MSCsprimarily reside in the bone marrow (15),they are also found in adipose tissue, inthe lungs and in many other organs,where they are involved in maintenance

M O L M E D 1 7 ( 7 - 8 ) 5 7 9 - 5 8 7 , J U L Y - A U G U S T 2 0 1 1 | S U Z U K I E T A L . | 5 7 9

Mesenchymal Stromal Cells Promote Tumor Growth throughthe Enhancement of Neovascularization

Kazuhiro Suzuki,1,2 Ruowen Sun,1,3 Makoto Origuchi,4 Masahiko Kanehira,4 Takenori Takahata,1 Jugoh Itoh,1

Akihiro Umezawa,5 Hiroshi Kijima,6 Shinsaku Fukuda,2 and Yasuo Saijo1

1Department of Medical Oncology, Hirosaki University Graduate School of Medicine, Hirosaki, Japan; the 2Department ofGastroenterology and Hematology, Hirosaki University Graduate School of Medicine, Hirosaki, Japan; the 3Department ofRheumatology and Immunology, Shengjing Hospital of China Medical University, Shenyang, China; the 4Department of MolecularMedicine, Tohoku University Graduate School of Medicine, Sendai, Japan; the 5Department of Reproductive Biology, NationalInstitute for Child Health and Development, Tokyo, Japan; and the 6Department of Pathology and Bioscience, Hirosaki UniversityGraduate School of Medicine, Hirosaki, Japan

Mesenchymal stromal cells (MSCs), also called mesenchymal stem cells, migrate and function as stromal cells in tumor tissues.The effects of MSCs on tumor growth are controversial. In this study, we showed that MSCs increase proliferation of tumor cells invitro and promote tumor growth in vivo. We also further analyzed the mechanisms that underlie these effects. For use in in vitroand in vivo experiments, we established a bone marrow–derived mesenchymal stromal cell line from cells isolated in C57BL/6mice. Effects of murine MSCs on tumor cell proliferation in vitro were analyzed in a coculture model with B16-LacZ cells. Both co-culture with MSCs and treatment with MSC-conditioned media led to enhanced growth of B16-LacZ cells, although the magni-tude of growth stimulation in cocultured cells was greater than that of cells treated with conditioned media. Co-injection of B16-LacZ cells and MSCs into syngeneic mice led to increased tumor size compared with injection of B16-LacZ cells alone. Identicalexperiments using Lewis lung carcinoma (LLC) cells instead of B16-LacZ cells yielded similar results. Consistent with a role for neo-vascularization in MSC-mediated tumor growth, tumor vessel area was greater in tumors resulting from co-injection of B16-LacZcells or LLCs with MSCs than in tumors induced by injection of cancer cells alone. Co-injected MSCs directly supported the tumorvasculature by localizing close to vascular walls and by expressing an endothelial marker. Furthermore, secretion of leukemia in-hibitory factor, macrophage colony-stimulating factor, macrophage inflammatory protein-2 and vascular endothelial growth fac-tor was increased in cocultures of MSCs and B16-LacZ cells compared with B16-LacZ cells alone. Together, these results indicatethat MSCs promote tumor growth both in vitro and in vivo and suggest that tumor promotion in vivo may be attributable in partto enhanced angiogenesis.© 2011 The Feinstein Institute for Medical Research, www.feinsteininstitute.orgOnline address: http://www.molmed.orgdoi: 10.2119/molmed.2010.00157

Address correspondence and reprint requests to Yasuo Saijo, Department of Medical On-

cology, Hirosaki University Graduate School of Medicine, 5 Zaifumacho, Hirosaki 036-8562,

Japan. Phone: +81-172-39-5345; Fax: +81-172-39-5347; E-mail: [email protected].

Submitted August 20, 2010; Accepted for publication March 10, 2011; Epub

(www.molmed.org) ahead of print March 11, 2011.

and regeneration of connective tissues(16). MSCs are known to migrate to tis-sues as a result of inflammation or injury,where they contribute to regeneration ofthe damaged tissues (17). For these rea-sons, MSCs have considerable therapeu-tic potential in tissue regeneration (18,19).

Recent results from both animal mod-els and human tumors have suggestedthat MSCs also migrate to tumor tissues,where they incorporate into the tumorstroma (20,21). This tropism of MSCs fortumors is reportedly due to the presenceof soluble factors secreted by tumorcells, similar to inflammatory responses(22,23). These findings have led to in-creased interest in understanding the ef-fects of MSCs in the tumor microenvi-ronment. Several studies havesuggested that MSCs promote tumorgrowth and the metastatic potential oftumor cells (3,24). MSCs can differenti-ate into fibro blasts, myofibroblasts orpericyte-like cells and induce neoangio-genesis, resulting in the promotion oftumor growth in vivo (24,25). In con-trast, few studies have indicated thatMSCs inhibit tumor growth (26). Severalstudies have used human MSCs, as op-posed to murine MSCs, to assess the ef-fects of this cell type on tumor growthin mouse models because of the ease ofexpansion of human MSCs (3,24). Inthese tumor xenograft models, thetumor stroma consists of mouse cells,but the tumoral cells and MSCs are ofhuman origin. Thus, because of themixed lineages of these cells, the effectsof MSCs on tumor growth may be af-fected by unknown interactions. On thebasis of these previous studies, weelected to use only murine cellsthroughout the present study for thepurpose of clearly interpreting the resulting findings.

In this study, we developed a quantita-tive assay for tumor growth in vitrousing coculture models with MSCs andB16 melanoma cells expressing LacZ(B16-LacZ). We demonstrated that bothdirect contact with MSCs and release ofsoluble factors from MSCs promote B16-LacZ cancer cell proliferation in vitro.

Furthermore, our results suggest that co-injection of MSCs with B16-LacZ cellspromotes tumor formation in micethrough enhanced angiogenesis, inducedby secretion of proangiogenic factorsfrom MSCs.

MATERIALS AND METHODS

Cell Culture and AnimalsB16-LacZ, a mouse melanoma cell line

expressing β-galactosidase, and TSt-4, amouse MSC cell line derived from fetalthymus tissue (27), were obtained fromthe RIKEN BioResource Center (Tsukuba,Japan). B16-LacZ cells were cultured inDulbecco’s modified Eagle’s medium(DMEM) supplemented with 10% fetalbovine serum (FBS; DMEM/10% FBS).Lewis lung carcinoma (LLC) cells wereobtained from the Cell Resource Centerfor Biomedical Research, Tohoku Univer-sity (Sendai, Japan). LLC cells were prop-agated in RPMI-1640 containing 10% FBS.A mouse bone marrow–derived mes-enchymal cell line (MSC) was establishedfrom bone marrow cells isolated fromC57BL/6 mice, as described previously(28). MSCs were cultured in DMEM lowglucose 1× medium (DMEM + Gluta-MAX; Invitrogen, San Diego, CA, USA)containing 10% FBS (Invitrogen). FemaleC57BL/6 mice, 6–8 wks of age, were pur-chased from Japan Charles River (Atsugi,Japan). Female C57BL/6-Tg (cytomegalo -virus immediate early enhancer/beta-actin promoter–enhanced green fluores-cent protein [CAG-EGFP]) mice werepurchased from Japan SLC (Hamamatsu,Japan). For primary culture of MSCsfrom green fluorescent protein (GFP)mice, the bone marrow suspension wascultured in DMEM + GlutaMAX with10% FBS. When adherent cells reached70–80% confluence, cells were harvestedand expanded. When a homogeneouscell population was obtained, after 3 to 5passages, these cells were used for subse-quent experiments.

Analysis of MSC Cell Surface MarkersCell surface antigens were analyzed by

flow cytometry in cultured murine

MSCs. Briefly, 1 × 105 cells were incu-bated with the following fluorescence-conjugated rat monoclonal antibodies:antimouse Sca-1 (Ly-6A/E; BD Pharmin-gen, San Diego, CA, USA), antimouseCD44 (Pgp-1/Ly-24; eBioscience, SanDiego, CA, USA), antimouse CD34(Beckman Coulter, Fullerton, CA, USA),antimouse CD45 (leukocyte commonantigen) (Beckman Coulter) and anti-mouse CD90 (Thy-1; Beckman Coulter).Nonspecific fluorescence was assessedby incubation of cells with isotype-matched rat monoclonal antibodies (BDPharmingen). Data were analyzed by col-lecting 20,000 events on a Cell LabQuanta SC (Beckman Coulter).

In Vitro Cell Proliferation AssaysFor proliferation assays using cocul-

tured cells, MSCs were seeded at 5.0 ×103 cells/well in 96-well plates in DMEMcontaining 1% FBS (DMEM/1% FBS).After 12 h, B16-LacZ cells were added(5.0 × 103 cells/well) to cultured MSCs.After an additional 24 h, cells were fixedby incubation in phosphate-bufferedsaline (PBS) containing 5.4% formalde-hyde and 0.8% glutaraldehyde at 12-htime points. After two washes with PBS,100 μL 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) solution(2 mg/mL) was added to each well. Cellswere then incubated at 37°C in a humidi-fied atmosphere containing 5% CO2 inthe dark for 12 h. Absorbance at 595 nmwas measured using an E-Max precisionmicroplate reader (Molecular Devices,Menlo Park, CA, USA).

For proliferation assays performed inthe absence of direct contact betweenMSCs and B16-LacZ cells, MSCs wereseeded at a density of 5.0 × 104 cells/wellin 24-well plates in DMEM/1% FBS.After a 12-h incubation, wells were cov-ered with Cell Disks (Sumitomo Bakelite,Tokyo, Japan) that serve as a bulkhead,and B16-LacZ cells (5.0 × 104 cells/well)were added to the cell disks. After an ad-ditional 48 h, cells were lysed by the ad-dition of lysis buffer (0.5% Triton X-100,2 mol/L NaCl in PBS) at 12-h timepoints, and 100 μL 2 mg/mL X-Gal solu-

5 8 0 | S U Z U K I E T A L . | M O L M E D 1 7 ( 7 - 8 ) 5 7 9 - 5 8 7 , J U L Y - A U G U S T 2 0 1 1

T U M O R P R O M O T I O N B Y M S C s

tion was combined with 15 μL of the cellsuspension. The absorbance at 595 nm ofthe resulting solution was then measuredas described above.

To assess cell proliferation in the pres-ence of conditioned media from MSCs,media were collected from MSCs (2 × 106

cells) cultured in 10 mL DMEM/1% FBSin a culture dish (10 cm in diameter) for48 h. The media were clarified by cen-trifugation (1,000g, 5 min), and the re-sulting supernatant was used as condi-tioned media. B16-LacZ cells wereseeded at a density of 5.0 × 103 cells/wellin 96-well plates and cultured inDMEM/ 10% FBS for 12 h. Subsequently,the media were replaced with either con-ditioned media or DMEM/1% FBS. Next,10 μL Alamar Blue assay solution(Biosource International, Camarillo, CA,USA) was added to the wells at 12-htime points, and the plates were incu-bated at 37°C. Fluorescence was mea-sured using a Fluoroskan Ascent CF ap-paratus (Labsystems, Helsinki, Finland)with excitation set to 544 nm and emis-sion set to 590 nm.

Analysis of In Vivo Tumor GrowthAll animal experiments were reviewed

and approved by the Institutional Ani-mal Care and Use Committee of HirosakiUniversity.

All mice (n = 8 for each group) weredivided into groups that received subcu-taneous injections of either (a) B16-LacZcells alone (5.0 × 105 cells), (b) B16-LacZcells (5.0 × 105 cells) with MSCs (1 × 105

cells) at a 1:0.2 ratio, (c) B16-LacZ cells(5.0 × 105 cells) with MSCs (5.0 × 105 cells)at a 1:1 ratio, (d) B16-LacZ (5.0 × 105

cells) with MSCs (2.5 × 106 cells) at a 1:5ratio or (e) MSCs alone (2.5 × 106 cells).All cell suspensions were delivered in afinal volume of 200 μL and injected sub-cutaneously into the right side of the ab-domen. LLC cells were transplanted in-stead of B16-LacZ cells under identicalconditions. Beginning 5 d after cell injec-tions, the tumor volume was calculatedevery 2 d using the following formula:tumor volume (mm3) = 0.52 × width(mm)2 × length (mm).

Immunohistochemistry of the TumorsB16-LacZ tumors on day 21 were re-

moved, fixed in 10% buffered formalin for24 h and then stained with hematoxylinand eosin for histological examination.For immunohistochemical staining forKi-67, sections were deparaffinized andantigen retrieval was conducted using anantigen retrieval solution (415211;Nichirei, Tokyo Japan). After blocking ofendogenous peroxidase activity with aperoxidase blocking reagent (S2001; Dako,San Antonio, TX, USA), tissue sectionswere incubated with rat antimouse Ki-67(1:25; Dako) and the appropriate second-ary antibody. Color development was per-formed using the peroxidase substrate3-amino-9-ethylcarbazole.

To characterize the effects of MSCs onmicrovessel area in tumor tissues, tumorswere removed when the tumor sizereached around 1 cm3 or at day 21.Cryosections were fixed with 4%paraformaldehyde and stained with ratantimouse CD31 (1:100; BD Pharmingen).Sections were then incubated with la-beled polymer (N-Histofine Simple Stain

Mouse MAX PO [Rat]; Nichirei, Tokyo,Japan). Color development was per-formed using 3-amino-9-ethylcarbazole.Sections were counterstained with hema-toxylin. Microvessel area in cryosectionsfrom each tumor was quantified using 10images from each of 10 different tumorsper group under 200× magnification.Vessel area was measured using the Im-ageJ software (http:// rsbweb.nih.gov/ij).

ImmunofluorescenceOn day 21, B16-LacZ tumors with

MSCs expressing GFP (GFP-MSCs) werefixed with 4% paraformaldehyde and in-creasing concentrations of sucrose buffer(12%, 15% and 18%) over 12 h. Frozensections were blocked with Protein BlockSerum-Free (X0909; Dako) and were in-cubated with rat antimouse CD31 (1:100;BD Pharmingen) or rabbit antimouseα smooth muscle actin (α-SMA, 1:50,ab5694; Abcam, Cambridge, U.K.). RatIgG2a and rabbit IgG were used as iso-type controls. Secondary antibodies usedwere donkey antirat IgG-Alexa Fluor 594(1:200; Invitrogen) or goat antirabbit IgG-

R E S E A R C H A R T I C L E


Figure 1. Flow cytometric and morphological analysis of MSCs. (A) Mouse bonemarrow–derived MSCs established from cells obtained in C57BL/6 mice were stained withantibodies directed against Sca-1, CD90, CD44, CD34 and CD45 and were analyzed byflow cytometry. (B) Morphology of MSCs in culture.

Alexa Fluor 594 (1:200; Invitrogen), re-spectively. To improve primary antibodypenetration, sections for α-SMA stainingwere incubated in PBS/0.02% TritonX-100 for 30 min at room temperature be-fore primary antibody incubation. Allprocedures were protected from light.Sections were examined with a confocallaser scanning microscope.

Analysis of Angiogenic Factor Levelsin Supernatants from B16-LacZ Cellsand MSCs

To measure the levels of angiogenic fac-tors secreted by B16-LacZ cells and MSCs,B16-LacZ cells alone (1.0 × 106 cells), MSCsalone (1.0 × 106 cells) or both B16-LacZcells and MSCs (1.0 × 106 cells each) werecultured for 24 h in 10 mL DMEM/10%FBS media. The media were then collectedand clarified by centrifugation (1,000g, 5 min), and the resulting supernatantswere used for analysis. Concentrations ofvascular endothelial growth factor (VEGF),macrophage inflammatory protein-2 (MIP-2; functional homolog of human interleukin [IL]-8), macrophage colony-stimulating factor (M-CSF), leukemia in-hibitory factor (LIF), IL-15, IL-18, basic fi-broblast growth factor (bFGF), monokineinduced by interferon γ (MIG) and platelet- derived growth factor (PDGF) were deter-mined with the Bio-Plex cytokine assay(Bio-Rad, Hercules, CA, USA) using a Luminex 200 (Luminex, Austin, TX, USA).

Statistical AnalysisResults are expressed as the mean ±

standard deviation (SD). Comparisonsbetween groups were performed using atwo-tailed Student t test. Linear regres-sion curves were produced using thePearson correlation. P values <0.05 indi-cated statistical significance.

RESULTS

In Vitro–Cultured Mesenchymal CellsExpress MSC Markers

Mouse bone marrow–derived MSCcultures were established from bonemarrow cells isolated in C57BL/6 mice.Phenotypically, MSCs were characterized

by being negative for expression of theCD34 and CD45 hematopoietic cellmarkers and positive for Sca-1, CD90and CD44. Flow cytometric analysis re-vealed that cultured MSCs were positivefor Sca-1, CD90 and low expression ofCD44, but negative for CD34 and CD45(Figure 1A), as expected. The MSCs ex-hibited a spindle-shaped morphology inculture (Figure 1B) and have been re-ported to differentiate into adipocytesand osteocytes (28). These results wereconsistent with previous reports and in-dicated that the established cell line in-deed consisted of MSCs (29).

Coculture with MSCs PromotesProliferation of B16-LacZ Cells In Vitro

We next analyzed whether MSCs couldpromote proliferation of the murine mela-noma cell line B16-LacZ in an in vitro co-

culture model. Initially, a linear regressionstandard curve was generated to correlatethe number of B16-LacZ cells with ab-sorbance at 595 nm (OD595). The assaywas selective for B16-LacZ cells, becauseonly this cell line expressed β-galactosi-dase (Figure 2A). Results from this analy-sis revealed that coculture of B16-LacZcells with MSCs led to a marked increasein proliferation of B16-LacZ cells com-pared with B16-LacZ cells cultured alone(Figure 2B). After 48 h, the number ofcells grown under coculture conditionswas 3.6-fold greater than that of B16-LacZcells cultured alone. On the basis of theseresults, we next analyzed the dose- response effect of MSCs on B16-LacZ pro-liferation (Figure 2C). These resultsshowed that proliferation of B16-LacZcells increased in accordance with thenumber of MSCs present in the coculture.



Figure 2. Proliferation of B16-LacZ cells cocultured with mesenchymal stromal cells (MSCs).(A) Standard curve correlating LacZ expression with cell numbers. B16-LacZ cells wereseeded on 96-well plates at different cell numbers. After 24 h, β-galactosidase protein lev-els were analyzed in each well by measuring the absorbance at 595 nm using a micro -plate reader, as described in Materials and Methods. (B) B16-LacZ cells were cultured withMSCs at a 1:1 ratio. β-Galactosidase protein expression was measured at the indicatedtime points (n = 5; **P < 0.01). (C) B16-LacZ cells were incubated with MSCs at different ra-tios. β-Galactosidase protein expression was measured after 48 h (n = 8; **P < 0.01). OD595, optimal density absorbance at 595 nm.

MSCs Promote B16-LacZ CellProliferation in the Absence of DirectContact

To determine whether direct contactbetween B16-LacZ cells and MSCs is re-quired for stimulation of B16-LacZ cellproliferation, cell growth in the absenceof direct contact was investigated usingcell disks and conditioned medium. Al-though the magnitude of stimulation ofcell proliferation cannot be directlycompared between cocultured cells andcells cultured in the absence of directcontact because of differences in growthconditions, a 2.3-fold increase in prolif-eration of B16-LacZ cells occurred inthe presence of MSCs cultured on celldisks (P < 0.01; Figure 3A), and a 1.4-fold increase occurred in the presenceof conditioned media (P < 0.01; Figure

3B) at 48 h. These values were slightlylower than the 6.4-fold increase in pro-liferation that occurred under condi-tions of 60-h direct contact in cocultures(see Figure 2B). To characterize themechanisms underlying these changesin cell proliferation, we performed cellcycle analysis using flow cytometry.However, no obvious change in the cellcycle distributions of the cancer cellswas apparent at the time points ana-lyzed (data not shown).

MSCs Derived from Bone MarrowPromote Tumor Growth In Vivo

B16-LacZ cells and LLC cells, a lungcarcinoma cell line, were used to assessthe effects of MSCs on tumor promo-tion and growth in an in vivo model.Mixtures containing each of these celltypes, along with MSCs, at ratios of1:0.2, 1:1 and 1:5, were subcutaneouslyinjected into syngeneic C57BL/6 mice,and tumor formation and growth wereassayed. At day 21 after tumor inocula-

tion, mice injected with B16-LacZ cellsand MSCs at 1:1 and 1:5 ratios exhibited2.3-fold (P < 0.01) and 4.3-fold (P <0.01) greater tumor volumes, respec-tively, than mice injected with B16-LacZcells alone (Figure 4A, B). However,tumor ratio of 1:0.2 did not increasetumor size compared with B16-LacZcells alone (see Figure 4B). At day 13after tumor inoculation, mice injectedwith LLCs and MSCs at 1:1 and 1:5 ra-tios exhibited 2.1-fold (P < 0.05) and2.6-fold (P < 0.01) greater tumor vol-umes, respectively, compared with miceinjected with LLCs alone (Figure 4A,C). Injection of MSCs alone did not re-sult in tumor formation. Mixtures ofB16-LacZ cells and another MSC cellline, TSt-4, derived from fetal thymustissue, were also injected into C57BL/6mice to assess the tumor growth effectof MSCs from a different origin. How-ever, we observed no stimulatory or in-hibitory effect on B16-LacZ tumorgrowth by TSt-4 MSCs at 1:1 and 1:5 ra-



Figure 3. Effects of MSCs on B16-LacZ cellproliferation in the absence of direct con-tact. (A) B16-LacZ cells and MSCs were cul-tured together but were separated by abulkhead consisting of a cell disk. Ab-sorbance at 595 nm was determined at theindicated time points (n = 7; *P < 0.05; **P <0.01). (B) B16-LacZ cells were cultured inconditioned medium from MSCs. Cell prolif-eration was measured using the AlamarBlue assay as described in the Materialsand Methods (n = 7; *P < 0.05; **P < 0.01).

Figure 4. Analysis of tumors derived from xenografts of B16-LacZ cells or LLCs co-injectedwith MSCs. (A) Representative photographs of B16-LacZ tumors and LLC tumors. (B) B16-LacZ cells and MSCs were co-injected into the right side of the abdomen of C57BL/6mice at a ratio of 1:0.2, 1:1 or 1:5. Tumor volume was calculated at 2-day intervals (n = 7;*P < 0.05; **P < 0.01). (C) LLCs and MSCs were co-injected as described for B16-LacZ cells,and tumor volume was calculated at 2-day intervals (n = 7; *P < 0.05, **P < 0.01).

tios (data not shown). To exclude thepossibility that increased tumor growthwas due to immunosuppression causedby MSCs, B16-LacZ cells and/or MSCswere subcutaneously injected intoBALB/c nude mice. Similar to the re-sults described above for syngeneicmice, co-injection of B16-LacZ cells andMSCs caused a significant increase intumor volume compared with injectionof B16-LacZ cells alone in nude mice (P < 0.05; data not shown).

MSCs Promote Tumor Growth In Vivoby Increasing Angiogenesis

To investigate mechanisms of tumorpromotion by MSCs, we analyzed B16-LacZ + MSC tumors on day 21 (Fig-ure 5). Hematoxylin and eosin stainingrevealed a massive necrotic area in B16-LacZ alone tumors, but not in B16-LacZ +MSC tumors (Figure 5A). In vivo analy-sis of cell proliferation using Ki-67 label-ing was compared between B16-LacZalone and B16-LacZ + MSC tumors (seeFigure 5A). Percentages of Ki-67–positivecells in B16-LacZ alone tumors, B16-LacZ + MSC tumors at a 1:1 ratio andB16-LacZ + MSC tumors at a 1:5 ratiowere 44.3 ± 5.1, 49.1 ± 3.4 and 42.4 ± 7.8,respectively, and were not significantlydifferent. Thus, we hypothesized thatMSCs may promote tumor growth byincreasing angiogenesis. To assesswhether MSCs promote angiogenesis,cryosections from tumors were stainedwith an antibody directed againstCD31 to visualize blood vessels (seeFigure 5A). Blood vessels were richer inB16-LacZ + MSC tumors (at ratios of 1:1and 1:5) than in B16-LacZ alone tumorson day 21.

For optimal quantitative analysis ofangiogenesis, we stained smaller tumors(around 1 cm3) with CD31 antibody in-stead of large tumors on day 21. Bloodvessel density was then analyzed byquantification of CD31+ areas. Resultsfrom this analysis revealed that vesselarea was increased in mice co-injectedwith B16-LacZ cells and MSCs (at ratiosof 1:1 and 1:5) compared with mice in-jected with B16-LacZ cells alone (P <



Figure 5. Analysis of B16-LacZ tumors in xenograft tumor models in the presence and ab-sence of co-injected MSCs. (A) Photographs of tumors generated by injection of B16-LacZalone or by co-injection of B16-LacZ cells with MSCs. On day 21, tumors were stained withhematoxylin and eosin, a Ki-67 antibody and a CD31 antibody. (B–D) Quantitative analysesof tumor angiogenesis. When tumors reached approximately 1 cm3, they were removed,fixed and stained with a CD31 antibody (B). Microvessel density in tumor cryosections wasanalyzed by visualization of CD31 staining at 200× magnification. Microvessel density wasquantified using the ImageJ software (n = 10; *P < 0.05; **P < 0.01). (C) B16-LacZ. (D) LLC.

0.01; Figure 5B, C). Similar results wereobserved when LLCs were used in placeof B16-LacZ cells (P < 0.01). However, nosignificant difference was observed inmice co-injected with different ratios ofcancer cells to MSCs (1:1 and 1:5, respec-tively; Figure 5B, D).

Differentiation of MSCs in the TumorsTo assess the role of MSCs in tumor

promotion, MSCs from GFP mice (GFP-MSCs) were co-injected with B16-LacZcells at a ratio of 1:1 into mice. AlthoughGFP-MSCs presented in tumor tissues

on day 21 (Figure 6A), the number ofMSCs was quite low and MSCs wererandomly distributed in tumor tissues.Some MSCs were closely presented atvessel structures. To assess differentia-tion and its association with tumor ves-sels, tumor tissues were stained with aCD31 antibody, as an endothelialmarker (Figure 6B), or an α-SMA anti-body, as a pericyte marker. Confocal mi-croscopy revealed that some MSCs thatwere closely presented in the tumor vasculature expressed CD31 but notα-SMA.

MSCs Secrete Several AngiogenicCytokines

On the basis of the observation thatMSCs promoted angiogenesis in vivo,we next assessed the levels of severalsecreted proangiogenic factors in mediafrom MSCs and B16-LacZ cells. Levelsof VEGF, MIP-2, M-CSF, LIF, IL-15, IL-18, bFGF, MIG and PDGF secretedby B16-LacZ cells (1.0 × 106 cells) alone,MSCs (1.0 × 106 cells) alone and cocul-tures of B16-LacZ cells and MSCs (1.0 ×106 cells each) were analyzed. Resultsfrom this analysis showed that MSCssecreted higher levels of LIF, M-CSF,MIP-2 and VEGF than tumor cells (Fig-ure 7, upper panel). Both B16-LacZ cellsand MSCs secreted little IL-15, IL-18,bFGF, MIG or PDGF (Figure 7, lowerpanel).

DISCUSSIONIn this study, we demonstrated that

MSCs promote tumor cell proliferation invitro and tumor growth in vivo. In both



Figure 6. Presence and differentiation of MSCs in the tumors. (A) GFP-MSCs present andrandomly distributed in the tumors on day 21 generated by co-injection of B16-LacZ cellsand GFP-expressing MSCs. (B) GFP-MSCs closely localized to tumor vessels coexpressedthe endothelial marker CD31, but did not coexpress the pericyte marker α-SMA.

Figure 7. Expression of proangiogenic fac-tors by MSCs. Secreted levels of LIF, M-CSF,MIP-2, VEGF (upper panel), IL-15, IL-18,bFGF and PDGF (lower panel) were ana-lyzed in media from B16-LacZ cells (1.0 ×105 cells/mL), MSCs (1.0 × 105 cells/mL) orcocultures of B16-LacZ cells and MSCs(1.0 × 105 cells/mL each). Concentrationswere analyzed using a Luminex 200 (n = 3;**P < 0.01).

the presence and absence of direct con-tact, MSCs stimulated proliferation ofB16-LacZ cells in vitro. Furthermore,combined administration of MSCs andtumor cells (B16-LacZ cells or LLCs) pro-moted tumor growth by enhancing an-giogenesis in syngeneic tumor models.This enhanced neovascularization canlikely be attributed to direct support ofneovascularization by MSCs and to se-cretion of angiogenic factors, includingVEGF and others, by MSCs.

Our results suggest that both directcell–cell contact and soluble factors likelyplay important roles in MSC-mediatedstimulation of tumor cell proliferation invitro, although we were not able to deter-mine the molecules responsible for thisphenotype. To date, few in vitro studieshave assessed the effects of MSCs ontumor cell proliferation, because of thedifficulty in distinguishing tumor cellproliferation from that of MSCs undercoculture conditions. Sasser et al. (30)demonstrated that MSCs enhance prolif-eration of human breast cancer cellsthrough the use of tumor cells that ex-pressed stable red fluorescence. Similarto our β-galactosidase–based system,measurement of fluorescence intensitycould distinguish between proliferationof fluorescent tumor cells and humanMSCs in cocultures (30). Results fromthis study indicated that stimulation oftumor cell proliferation by human MSCswas solely due to secreted soluble fac-tors, in contrast to our results suggestingthat both direct contact and soluble fac-tors play a role in stimulating prolifera-tion. Another study demonstrated thattreatment of pancreatic tumor cells withconditioned media from cancer-associ-ated stromal fibroblasts enhanced cellproliferation in vitro; however, effects incocultures containing both fibroblastsand tumor cells were not examined (4).Secreted IL-6 from fibroblasts was re-ported to enhance the rate of cancer cellproliferation in a previous study (5).Thus, IL-6 may represent one of the solu-ble factors partially responsible for stim-ulation of cell proliferation in B16-LacZcells. Although these studies demon-

strated MSC-mediated stimulation oftumor cell proliferation in vitro, none ofthem analyzed changes in the cell cycle.We did not observe any change of cellcycle distribution. This negative observa-tion may have been because of usingnon-synchronized cells. Nevertheless,further investigation is needed to iden-tify the precise molecules that mediatethis effect.

Subcutaneous co-injection of tumorcells and MSCs resulted in more rapidtumor growth in mice compared with in-jection with tumor cells alone in assaysusing either B16-LacZ or LLC cells. In-creased tumor growth was independentof the mouse strain background; indeed,similar effects were observed in bothsyngeneic mice and nude mice. The MSCcell line TSt-4 derived from fetal thymustissue did not affect B16-LacZ tumors invivo. The reported effects of MSCs onprogression of primary tumors havebeen both pro- and antitumorigenic, butvariability in these results, includingours, could be attributable to differencesin the sources of the MSCs and the typeof tumor model used for analysis. For ex-ample, intravenous injection of humanMSCs into Kaposi’s sarcoma-bearingnude mice inhibited tumor growth (26).MSCs also decreased proliferation of Ka-posi’s sarcoma cells in vitro under condi-tions of direct contact through inhibitionof Akt signaling. Human MSCs werefound to inhibit proliferation of aleukemia cell line and a small cell–lungcancer cell line in vitro, whereas tumorcells grew significantly faster when co-injected with MSCs into nonobese dia-betic severe combined immunodeficientmice compared with injection of tumorcells alone (31). However, MSC-mediatedinhibitory effects have only been ob-served in a few models, and most studiesreported protumorigenic effects, consis-tent with our findings. Note also that im-munosuppressive effects caused byMSCs were shown to promote B16 tumorgrowth in an allogeneic mouse model(32). However, this was not the case inour study, because we observed no dif-ference in the effect of MSCs on tumor

growth between nude and syngeneicmice.

We observed neither a change in theKi-67 labeling index in the B16-LacZ +MSC tumors nor a high number of MSCsin the tumor tissues. These observationssuggest that tumor promotion by MSCswas not due to stimulated tumor cellgrowth or an increased number of MSCs.We demonstrated that MSCs stronglystimulate tumor angiogenesis, likelythrough secretion of several angiogenicfactors. The effects of MSCs on the tumorvasculature remain unclear and are likelyto be complex (33,34). We hypothesizethat MSCs exert paracrine effects on en-dothelial cells via secreted growth factorsand cytokines, directly contributing toblood vessel formation in the tumor mi-croenvironment. MSCs secrete manyproangiogenic factors, including VEGF,IL-6, transforming growth factor-β andIL-8 (24,34–36), and secretion of theseproangiogenic factors was shown to besignificantly enhanced by treatment withconditioned media from tumor cells (24).However, we did not observe enhancedsecretion of these factors from MSCs as aresult of coculture with tumor cells, al-though MSCs did secrete higher amountsof proangiogenic factors than tumorcells. VEGF expression in MSCs can beenhanced by hypoxia, a common phe-nomenon in tumor tissues (37). Together,these data and our findings suggest thatMSCs play a key role in tumor neovascularization.

Another contribution of MSCs to thetumor vasculature occurs through thesupport of vascular formation in tumortissues. In the present study, we at-tempted to investigate the differentiationstate of MSCs after co-injection withtumor cells using GFP-MSCs. SomeMSCs closely presented to the tumorvasculature and expressed the endothe-lial cell marker CD31, but not the peri-cyte marker α-SMA. Grafted MSCs wereshown to integrate into tumor vesselwalls and express pericyte markers, butnot endothelial markers, suggesting thatMSCs in tumor tissues differentiate intopericytes and contribute to the tumor



vasculature (25). Although this reportwas different from ours, these studiessuggest that MSCs directly support thetumor vasculature by differentiating intoendothelial cells, pericytes or other typesof cells. Additionally, grafted MSCs werereported to differentiate into tumor- associated fibroblasts and to contributeto tumor progression (38).

In conclusion, results from this studydemonstrate that MSCs stimulate tumorcell proliferation in vitro and tumorgrowth in vivo. Enhanced tumor growthin syngeneic mouse models by MSCsmay be, in part, due to promotion oftumor neovascularization by directlysupporting the tumor vasculature andsecreting proangiogenic factors. These re-sults suggest that MSCs play an impor-tant role in tumor progression.

ACKNOWLEDGMENTSThis work was supported in part by

grants-in-aid for scientific research fromthe Ministry of Education, Science,Sports, Culture and Technology, Japan(numbers 16022206 and 20390229).

DISCLOSUREThe authors declare that they have no

competing interests as defined by Molec-ular Medicine, or other interests thatmight be perceived to influence the re-sults and discussion reported in thispaper.

REFERENCES1. Dvorak HF. (1986). Tumors: wounds that do not

heal. Similarities between tumor stroma generationand wound healing. N. Engl. J. Med. 315:1650–9.

2. Bissell MJ, Radisky D. (2001). Putting tumours incontext. Nat. Rev. Cancer. 1:46–54.

3. Karnoub AE, et al. (2007). Mesenchymal stemcells within tumour stroma promote breast can-cer metastasis. Nature. 449:557–63.

4. Hwang RF, et al. (2008). Cancer-associated stro-mal fibroblasts promote pancreatic tumor pro-gression. Cancer Res. 68:918–26.

5. Studebaker AW, et al. (2008). Fibroblasts isolatedfrom common sites of breast cancer metastasisenhance cancer cell growth rates and invasive-ness in an interleukin-6-dependent manner. Can-cer Res. 68:9087–95.

6. Kim S, et al. (2009). Carcinoma-produced factorsactivate myeloid cells through TLR2 to stimulatemetastasis. Nature. 457:102–6.

7. Grugan KD, et al. (2010). Fibroblast-secreted he-patocyte growth factor plays a functional role inesophageal squamous cell carcinoma invasion.Proc. Natl. Acad. Sci. U. S. A. 107:11026–31.

8. Paulsson J, et al. (2009). Prognostic significance ofstromal platelet-derived growth factor beta-re-ceptor expression in human breast cancer. Am. J.Pathol. 175:334–41.

9. Utispan K, et al. (2010). Gene expression profilingof cholangiocarcinoma-derived fibroblast revealsalterations related to tumor progression and indi-cates periostin as a poor prognostic marker. Mol.Cancer. 9:13.

10. Wels J, Kaplan RN, Rafii S, Lyden D. (2008). Mi-gratory neighbors and distant invaders: tumor-associated niche cells. Genes. Dev. 22:559–74.

11. Direkze NC, et al. (2004). Bone marrow contribu-tion to tumor-associated myofibroblasts and fi-broblasts. Cancer Res. 64:8492–5.

12. Udagawa T, Puder M, Wood M, Schaefer BC,D’Amato RJ. (2006). Analysis of tumor-associatedstromal cells using SCID GFP transgenic mice:contribution of local and bone marrow-derivedhost cells. FASEB J. 20:95–102.

13. Worthley DL, et al. (2009). Human gastrointesti-nal neoplasia-associated myofibroblasts can de-velop from bone marrow-derived cells followingallogeneic stem cell transplantation. Stem Cells.27:1463–8.

14. Pittenger MF, et al. (1999). Multilineage potentialof adult human mesenchymal stem cells. Science.284:143–7.

15. Prockop DJ. (1997). Marrow stromal cells as stemcells for nonhematopoietic tissues. Science.276:71–4.

16. Campagnoli C, et al. (2001). Identification of mes-enchymal stem/progenitor cells in human first-trimester fetal blood, liver, and bone marrow.Blood. 98:2396–402.

17. Karp JM, Leng Teo GS. (2009). Mesenchymalstem cell homing: the devil is in the details. CellStem Cell. 4:206–16.

18. Ortiz LA, et al. (2003). Mesenchymal stem cell en-graftment in lung is enhanced in response tobleomycin exposure and ameliorates its fibroticeffects. Proc. Natl. Acad. Sci. U. S. A. 100:8407–11.

19. Lian Q, et al. (2010). Functional mesenchymalstem cells derived from human induced pluripo-tent stem cells attenuate limb ischemia in mice.Circulation. 121:1113–23.

20. Studeny M, et al. (2004). Mesenchymal stem cells:potential precursors for tumor stroma and tar-geted-delivery vehicles for anticancer agent. J.Natl. Cancer Inst. 96:1593–603.

21. Kidd S, et al. (2009). Direct evidence of mes-enchymal stem cell tropism for tumor andwounding microenvironments using in vivo bio-luminescent imaging. Stem Cells. 27:2614–23.

22. Klopp AH, et al. (2007). Tumor irradiation in-creases the recruitment of circulating mesenchy-mal stem cells into the tumor microenvironment.Cancer Res. 67:11687–95.

23. Coffelt SB, et al. (2009). The pro-inflammatorypeptide LL-37 promotes ovarian tumor progres-

sion through recruitment of multipotent mes-enchymal stromal cells. Proc. Natl. Acad. Sci.U. S. A. 106:3806–11.

24. Spaeth EL, et al. (2009). Mesenchymal stem celltransition to tumor-associated fibroblasts con-tributes to fibrovascular network expansion andtumor progression. PLoS One. 4:e4992.

25. Bexell D, et al. (2009). Bone marrow multipotentmesenchymal stoma cells act as pericyte-like mi-gratory vehicles in experimental gliomas. Mol.Ther. 17:183–90.

26. Khakoo AY, et al. (2006). Human mesenchymalstem cells exert potent antitumorigenic effects ina model of Kaposi’s sarcoma. J. Exp. Med.203:1235–47.

27. Watanabe Y, et al. (1992) A murine thymic stro-mal cell line which may support the differentia-tion of CD4–8- thymocytes into CD4+8- alphabeta T cell receptor positive T cells. Cell. Immunol.142:385–97.

28. Umezawa A, et al. (1992). Multipotent marrowstromal cell line is able to induce hematopoiesis invivo. J. Cell Physiol. 151:197–205.

29. Annabi B, et al. (2003). Hypoxia promotes murinebone-marrow-derived stromal cell migration andtube formation. Stem Cells. 21:337–47.

30. Sasser AK, et al. (2007). Human bone marrowstromal cells enhance breast cancer cell growthrates in a cell line-dependent manner when eval-uated in 3D tumor environments. Cancer Lett.254:255–64.

31. Ramasamy R, et al. (2007). Mesenchymal stemcells inhibit proliferation and apoptosis of tumorcells: impact on in vivo tumor growth. Leukemia.21:304–10.

32. Djouad F, et al. (2003). Immunosuppressive effectof mesenchymal stem cells favors tumor growthin allogeneic animals. Blood. 102:3837–44.

33. Beckermann BM, et al. (2008). VEGF expressionby mesenchymal stem cells contributes to angio-genesis in pancreatic carcinoma. Br. J. Cancer.99:622–31.

34. Otsu K, et al. (2009). Concentration-dependent inhi-bition of angiogenesis by mesenchymal stem cells.Blood. 113:4197–205.

35. Gunn WG, et al. (2006). A crosstalk between mye-loma cells and marrow stromal cells stimulatesproduction of DKK1 and interleukin-6: a poten-tial role in the development of lytic bone diseaseand tumor progression in multiple myeloma.Stem Cells. 24:986–91.

36. Romieu-Mourez R, et al. (2009). Cytokine modu-lation of TLR expression and activation in mes-enchymal stromal cells leads to a proinflamma-tory phenotype. J. Immunol. 182:7963–73.

37. Potier E, et al. (2007). Hypoxia affects mesenchy-mal stromal cell osteogenic differentiation andangiogenic factor expression. Bone. 40:1078–87.

38. Mishra PJ, et al. (2008). Carcinoma-associated fibroblast-like differentiation of human mes-enchymal stem cells. Cancer Res. 68:4331–9.



mesenchymal stromal cells promote tumor growth through the

Documents