Urokinase Plasminogen Activator and Urokinase PlasminogenActivator Receptor Mediate Human Stem Cell Tropism toMalignant Solid Tumors
MARGARITA GUTOVA,a JOSEPH NAJBAUER,a RICHARD T. FRANK,a STEPHEN EDWARD KENDALL,b
ANNA GEVORGYAN,a MARIANNE Z. METZ,a MARK GUEVORKIAN,a MARISSA EDMISTON,a DONGHONG ZHAO,a
CARLOTTA A. GLACKIN,b SEUNG U. KIM,c,d KAREN S. ABOODYa,e
Divisions of aHematology/Hematopoietic Cell Transplantation, bMolecular Medicine, and eNeurosciences, City ofHope National Medical Center and Beckman Research Institute, Duarte, California, USA; cDepartment of Medicine,University of British Columbia Hospital, Vancouver, British Columbia, Canada; dGachon University School ofMedicine, Inchon, Korea
Human neural and mesenchymal stem cells have been iden-tified for cell-based therapies in regenerative medicine andas vehicles for delivering therapeutic agents to areas ofinjury and tumors. However, the signals required for hom-ing and recruitment of stem cells to these sites are not wellunderstood. Urokinase plasminogen activator (uPA) andurokinase plasminogen activator receptor (uPAR) are in-volved in chemotaxis and cell guidance during normal de-velopment and are upregulated in invasive tumors. Herewe provided evidence that activation of uPA and uPAR inmalignant solid tumors (brain, lung, prostate, and breast)augments neural and mesenchymal stem cell tropism.Expression levels of uPAR on human solid tumor cell linescorrelated with levels of uPA and soluble uPAR in tumorcell-conditioned media. Cytokine expression profiles ofthese tumor-conditioned media were determined by pro-
tein arrays. Among 79 cytokines investigated, interleukin(IL)-6, IL-8, and monocyte chemoattractant protein-1were the most highly expressed cytokines in uPAR-posi-tive tumors. We provided evidence that human recombi-nant uPA induced stem cell migration, whereas depletionof uPA from PC-3 prostate cancer cell-conditioned me-dium blocked stem cell migration. Furthermore, retrovi-rus-mediated overexpression of uPA and uPAR in neuro-blastoma (NB1691) cells induced robust migration of stemcells toward NB1691 cell-conditioned media, comparedwith media derived from wild-type NB1691 cells. Weconclude that expression of uPA and uPAR in cancer cellsunderlies a novel mechanism of stem cell tropism to ma-lignant solid tumors, which may be important for devel-opment of optimal stem cell-based therapies. STEM CELLS2008;26:1406 –1413
Disclosure of potential conflicts of interest is found at the end of this article.
INTRODUCTION
Migration of endogenous and exogenous neural stem cells(NSCs) and mesenchymal stem cells (MSCs) to areas of pathol-ogy is a critical step in tissue regeneration [1]. NSCs and MSCshave been shown to home to areas of brain pathology such asischemic and neoplastic lesions [2, 3]. This inherent homingability of stem cells makes them useful for regeneration of
damaged tissues, as well as for targeted delivery of therapeuticsubstances to sites of pathology [4].
Directed cell migration is initiated in response to chemoat-tractants. Numerous cytokines, growth factors, and their recep-tors have been shown to affect stem cell migration under normaland pathological conditions. Such cytokine/receptor pairs in-clude stromal cell-derived factor SDF-1/CXCR4 [5, 6], stemcell factor (SCF)/c-Kit [7], hepatocyte growth factor (HGF)/c-
Author contributions: M.G.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing, finalapproval of manuscript; J.N.: data analysis and interpretation, collection and assembly of data, manuscript writing, final approval ofmanuscript; R.T.F.: collection and assembly of data, data analysis and interpretation; S.E.K., M.G., M.E., D.Z.: collection and assembly ofdata; A.G.: conception and design; M.Z.M.: collection of data, data analysis; C.A.G.: data analysis and interpretation; S.U.K.: provisionof study material or patients; K.S.A.: financial support, administrative support, provision of study material or patients, data analysis andinterpretation, manuscript writing, final approval of manuscript.
Met [8, 9], vascular endothelial growth factor (VEGF)/VEGFreceptor (VEGFR) [10], monocyte chemoattractant protein-1(MCP-1)/CCR2 [11], and high mobility group box 1 (HMGB1)/RAGE [12, 13]. Among adhesion molecules, �1- and �2-inte-grins and L-selectin play a significant role in the mobilizationand homing of stem cells [14–16]. Extracellular matrix proteinshave also been associated with NSC-glioma tropism [17]. How-ever, the molecular mechanisms of NSC and MSC migration totumors of various origins and phenotypes are not well defined.
Urokinase plasminogen activator receptor (uPAR), alsoknown as CD87, was first cloned in 1985 [18]. It was believedthat the only role of uPAR was binding its ligand, urokinaseplasminogen activator (uPA), leading to plasminogen activationand degradation of the extracellular matrix. However, recentstudies have shown that interaction of uPAR with receptors ofthe integrin family, G-protein-coupled receptors, and vitronectininitiates activation of several intracellular signal transductionpathways involved in cell migration, adhesion, proliferation, andapoptosis [19–21]. uPA and uPAR also play a crucial role inearly stages of interneuron development, which has been shownto be impaired by the absence of uPAR-mediated signaling inuPAR(�/�) knockout mice [22].
uPA and uPAR are upregulated in tumors of various origins,where they play a critical role in the development of invasiveand chemoresistant cancer phenotypes [23–25]. Induction ofbiosynthesis of uPAR and its shedding as soluble urokinaseplasminogen activator receptor (suPAR) have also been ob-served in acute and chronic inflammatory conditions [26]. su-PAR is involved in the mobilization and migration of stem andinflammatory cells from the bone marrow to sites of injury,following granulocyte colony-stimulating factor (G-CSF) stim-ulation [16, 27].
Given the pleiotropic role of uPA and uPAR in cancer celland stem cell mobility, we hypothesized that expression ofuPAR on tumor cells may determine the degree of tropism ofNSCs and MSCs to solid tumors of various origins. Here wereport that expression of uPAR in human cancer cell lines leadsto release of uPA and suPAR into tumor cell-conditioned media.Cell migration assays revealed that chemoattraction of NSCsand MSCs to cancer cells strongly correlated with uPAR ex-pression levels on tumor cells. We determined the cytokineprofiles of conditioned media derived from high and low uPAR-expressing tumor cells. Expression of uPAR on cancer cells wasassociated with elevation of numerous cytokines, including in-terleukin (IL)-6, IL-8, MCP-1, and HGF, in the tumor-condi-tioned media. In summary, our results demonstrate that uPA anduPAR are significant mediators of stem cell tropism to tumors.
MATERIALS AND METHODS
Immunocytochemistry and Flow Cytometry AnalysisHuman cancer cell lines derived from brain (U251), neuroblastoma(SK-N-AS, NB1691), lung (H1415, H211, H1915), prostate (PC-3),colon (COLO-320), and breast (MDA-MB-231, MCF-7) were ob-tained from American Type Culture Collection ([ATCC] Manassas,VA, http://www.atcc.org). U251, SK-N-AS, NB1691 MDA-MB-231, and MCF-7 cells were grown in Dulbecco’s modified Eagle’smedium (DMEM; Irvine Scientific, Santa Ana, CA, http://www.irvinesci.com), and H1415, H211, COLO-320, and H1915 weregrown in RPMI 1640 modified medium (ATCC), both supple-mented with 10% heat-inactivated fetal bovine serum (FBS) and100 units/ml penicillin, 100 �g/ml streptomycin at 37°C in 6% CO2.The PC-3 cell line was grown in Ham’s F-12K medium with 2 mML-glutamine supplemented with 10% FBS (ATCC). For flow cy-tometry analysis, each cell line was detached by trypsinization andresuspended in staining buffer (SB) (Hanks’ balanced saline solu-
tion [HBSS]; Irvine Scientific, Santa Ana, CA, http://www.irvinesci.com) supplemented with 2% FBS and 10 mM HEPES at adensity of 5 � 106 cells per milliliter. Fifty microliters of cells(2.5 � 105 cells) was added to each well of a 96-well V-shapedplate. Antibodies (fluorescein isothiocyanate [FITC]-conjugateduPAR) were added in individually titrated concentrations (10 �l per106 cells). The 96-well plates were placed on ice, and cells wereincubated with antibodies for 30 minutes in the dark. Then, 150�l/well of wash buffer (HBSS, supplemented with 15% FBS and 10mM HEPES) was added, and the plates were centrifuged at 500g for5 minutes at 4°C. The cell pellets were resuspended in SB supple-mented with propidium iodide (1 �g/ml) to exclude nonviable cellsand subjected to flow cytometric analysis. Alternatively, cells werecultured on glass slides with a Teflon barrier (Electron MicroscopySciences, Hatfield, PA, http://www.emsdiasum.com) overnight,fixed in 4% paraformaldehyde, washed in phosphate-buffered sa-line, and incubated in blocking solution (DakoCytomation,Glostrup, Denmark, http://www.dakocytomation.com) for 1 hour atroom temperature. Cells were then stained with FITC-conjugatedanti-uPAR monoclonal antibodies (American Diagnostica Inc.,Stamford, CT, http://www.americandiagnostica.com) at a 1:100 di-lution in antibody diluent (DakoCytomation) overnight at 4°C.Slides were washed and incubated with secondary antibody (bio-tinylated anti-mouse IgG at 1:250 dilution) for 1 hour at roomtemperature. Slides were washed, stained with avidin-FITC (VectorLaboratories, Burlingame, CA, http://www.vectorlabs.com) at a1:1,000 dilution, and incubated for 1 hour. Thereafter, the slideswere counterstained with 4,6-diamidino-2-phenylindole and exam-ined using fluorescent microscopy. In all staining experiments,negative controls included omission of the primary antibodiesand staining with isotype-matched mouse IgG.
Real-Time Polymerase Chain ReactionTotal RNA was isolated from cancer cell lines by using the Qiagen(Hilden, Germany, http://www1.qiagen.com) RNeasy kit accordingto the manufacturer’s recommendations. Standard real-time poly-merase chain reaction (PCR) was performed using the followingprimers: �-actin, 5�-GCC GAT CCA CAC GGA GTA CT-3� (for-ward), 5�-CTG GCA CCC AGC ACA ATG-3� (reverse); uPAR,5-GCC CAA TCC TGG AGC TTG A-3� (forward), 5�-TCC CCTTGC AGC TGT AAC ACT-3� (reverse); and uPA, 5�-TTG CTCACC ACA ACG ACA TT-3� (forward), 5�-GGC AGG CAG ATGGTC TGT AT-3� (reverse).
Culturing of HB1.F3 Neural Stem CellsThe HB1.F3 cell line is a multipotent, cloned cell line that wasgenerated by immortalizing cells obtained from the telencephalon ofa human fetus of 15 weeks’ gestation, using a retrovirus encodingthe v-myc gene. The primary cells were obtained in accordance withthe guidelines of the Anatomical Pathology Department of Vancou-ver General Hospital, with permission to use fetal tissue granted bythe Clinical Research Screening Committee Involving Human Sub-jects of the University of British Columbia. HB1.F3 is an estab-lished, well-characterized, stable cell line [28, 29]. HB1.F3 cells arenontumorigenic and multipotent and can be induced to differentiateinto neurons, oligodendrocytes, and astrocytes. The use of this cellline circumvented the significant problem of limited availability oflarge quantities of primary cells and maximized reproducibilityamong experiments. Human NSCs (HB1.F3 and HB1.F5; two im-mortalized clonal lines, derived from the same human fetal telen-cephalon) were grown in DMEM supplemented with 10% FBS at37°C in 6% CO2. Multiple migration assays confirmed that bothHB1.F3 and HB1.F5 NSCs display similar migration properties.
Culturing of Mesenchymal Stem CellsHuman fetal bones were dissected from 18–24-week-old fetusesobtained by elective abortion (Advanced Bioscience Resources,Alameda, CA). Tissues were obtained with approved consent ofInstitutional Review Board (IRB number 94014). Mononuclear cellswere isolated from fetal bone marrow using Ficoll gradient centrif-ugation. Briefly, femurs of elective abortion fetuses were rinsedwith HBSS, and bones were trimmed of muscle and fat tissue. The
1407Gutova, Najbauer, Frank et al.
www.StemCells.com
center of the marrow was scraped with blunt forceps, collected,washed, and passed through nylon mesh. The collected bone mar-row cells were purified for mononuclear cells via Ficoll densitygradient fractionation. MSCs were cultured in growth mediumconsisting of DMEM 1% glucose, 15% heat-inactivated FBS, 100units/ml penicillin, and 100 �g/ml streptomycin at 37°C in 6% CO2.After the third passage, the resulting cells were analyzed and foundto react with SH2, SH3, and SH4 antibodies, which detect CD105(endoglyn) and CD73, two antigens coexpressed on MSCs. Thecells were then further characterized for MSC phenotype.
Migration AssayIn vitro chemotaxis assays were conducted using 24-well cell cul-ture plates with polycarbonate inserts (Millipore, Billerica, MA,http://www.millipore.com) with pore diameters of 8 or 10 �m forNSCs and MSCs, respectively. In brief, conditioned media wereprepared by addition of serum-free media to cultured cells (�75%confluence), followed by incubation at 37°C, 6% CO2 for 48 hours.Conditioned media from tumor lines were collected and added tothe lower chamber of 24-well plates (600 �l). Inserts were placedinto wells, and a suspension of stem cells was added in the upperchamber (105 cells per 400 �l in DMEM supplemented with 2%bovine serum albumin [BSA]). After incubation of the plates for 4hours at 37°C, the cells that did not migrate were removed from theinner surface of the filter, whereas migrated cells were detachedfrom the lower surface of the insert by trypsinization. Detached cellswere centrifuged for 5 minutes and counted using the Guava Via-Count assay (Guava Technologies, Hayward, CA, http://www.guavatechnologies.com). Only viable cells were included in ourdata analysis. The Guava ViaCount assay distinguishes betweenviable and nonviable cells on the basis of the differential perme-ability of DNA-binding dyes in the ViaCount reagent, and therefore,fluorescence of the dyes allows the quantitative assessment of bothviable and nonviable cells in suspension. Migration assay controlswere as follows: as a negative control, stem cells resuspended in 5%BSA were added to the upper chamber, with 5% BSA in the bottomchamber; as a positive control, 10% FBS was added to the lowerchamber as a chemoattractant.
Cells were allowed to migrate to human recombinant uPA(catalog no. 124 HMW-uPA; American Diagnostica) after additionof 0.25–100 ng/ml of uPA to the lower chamber in serum-freemedia, and the migration assay was performed under same condi-tions as described above. Inhibition of HB1.F3 neural stem cellmigration was achieved by preincubation of stem cells with uPARfunction-inhibiting antibodies (1 hour, 1 �g per 105 cells, 37°C),and cells were applied to migration assay. Small interfering RNAs(siRNAs) were designed to inhibit expression of uPAR in PC-3 cellsusing TriFECTa Dicer-Substrate RNAi system (IDT Inc., Cor-alville, IA, http://www.idtdna.com). Three siRNAs against humanuPAR were designed using IDT RNAi Design Software and pur-chased from IDT (NM_002659.3.1, NM_002659.3.4, NM_002659.3.8). Briefly, PC-3 cells were plated at a density of 2 � 104
cells per well in a 24-well plate. After 24 hours, cells were trans-fected using 10, 1, or 0.1 nM siRNA duplexes. Hypoxanthineguanine phosphoribosyltransferase 1 dicer-substrate (HPRTS1 DS)positive control and scrambled negative control duplexes were usedat concentration of 10 nM. X-tremeGENE siRNA transfection re-agent was used for all transfections (catalog no. 04476093001;Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com). Optimal inhibition of uPAR expression was achievedby using transfection of 10 nM plasminogen activator urokinasereceptor-siRNA. Cells were analyzed for uPAR expression by flowcytometry and reverse transcription (RT)-PCR after 48 hours oftransfection. Cell numbers were normalized, and cells were platedfor collection of conditioned media at 48 hours; these media wereused for migration assays.
Depletion of uPA from PC-3 cell-conditioned media wasachieved by incubation of media with anti-uPA antibody (isotype-matched mouse IgG as control) bound to protein A-coated Sepha-rose beads (18 hours, 4°C). Stem cell migration assay was per-formed using uPA-depleted conditioned media.
uPA and uPAR Enzyme-Linked ImmunosorbentAssayImmunobind uPA and uPAR enzyme-linked immunosorbent assays(ELISAs) were performed according to the manufacturer’s protocol(American Diagnostica). In brief, tumor cell-conditioned mediawere incubated in microtest wells that were precoated with uPA orsuPAR. uPA ELISA recognizes uPA (single chain and pro-uPA)and uPA/plasminogen activator inhibitor-1 (PAI-1) complexes,whereas suPAR ELISA recognizes suPAR and suPAR/PAI-1 com-plexes. Biotinylated secondary antibodies that recognized the pri-mary antibodies bound to uPA and uPAR molecules were thenadded. Addition of streptavidin-conjugated horseradish peroxidase(HRP) completed the formation of antibody-enzyme complexes.3.3�, 5.5�-tetramethylbenzidine substrate was added to create a colorreaction. The reaction was stopped by addition of 0.5 M sulfuricacid, and solution absorbance was measured at 450 nm. Absorbancevalues were compared with those of a standard curve.
Cytokine Profiling of Tumor-Conditioned MediaCytokine profiles of tumor cell-conditioned media were detected bythe Cytokine Antibody Array V (RayBiotech, Norcross, GA, http://www.raybiotech.com) according to the manufacturer’s instructions.Briefly, nitrocellulose blots were blocked for 1 hour and thenincubated overnight at 4°C with undiluted conditioned mediumderived from tumor cells. Blots were incubated at room temperaturewith a 1:500 dilution of biotin-conjugated antibodies for 2 hours,washed, and incubated for 1 hour at room temperature with a1:10,000 dilution of HRP-conjugated streptavidin. Chemilumines-cent detection of captured cytokines was quantified with the EPIChemi II kit (RayBiotech), and films were analyzed using NIHImageJ software. Signal intensity at each pixel was determined, andthe relative intensity of each spot on the array was quantified bysumming the intensities of each pixel within the spot. For eacharray, negative control intensities were used to determine back-ground, which was subsequently subtracted. Resulting data werenormalized to the average of four positive control spots to yield afinal normalized value for each cytokine. Inducers and noninducersof stem cell migration were grouped and averaged together forstatistical comparison. Relative expression of the cytokine levels inthe conditioned media from the cell lines was graphically repre-sented.
Overexpression of uPA and uPAR in NB1691 CellsThe uPA-pBABE and uPAR-pBABE constructs were obtained fromCell Biolabs, Inc. (San Diego, http://www.cellbiolabs.com). Retro-viruses containing uPA and uPAR were generated using the Pan-tropic Retroviral Expression System (Clontech, Palo Alto, CA,http://www.clontech.com). Briefly, GP2-293 packaging cells wereplated at a density of 4 � 106 cells per 150-mm cell culture plateovernight and then transfected with the constructs of interest. Theculture medium was collected after 48 hours and filtered through a0.45-�m syringe-mounted filter (Fisher, Hampton, NH, http://www.fishersci.com). The medium was then concentrated using a Sorvall90SE ultracentrifuge at 25,000 rpm at 4°C for 1.5 hours. Concen-trated retroviral supernatant was used to transduce 0.5 � 106
NB1691 cells per 35-mm plate over a 24-hour period. Stable lineswere then selected using 2 �g/ml puromycin and/or 100 �g/mlhygromycin. Colonies were selected and screened for transgeneexpression using real-time PCR.
RESULTS
uPAR Expression in Tumor Cell LinesWe used flow cytometric analysis and immunocytochemistry todetect uPAR expression in cell lines derived from human pros-tate, lung, breast, colon, and brain cancers (Fig. 1A, 1C). On thebasis of the expression levels of uPAR, the cell lines weregrouped as follows: (a) cells expressing high levels of uPAR(�20% of cells uPAR-positive), or (b) cells expressing low
1408 Stem Cell Tropism to Tumors
levels of uPAR (�5% of cells uPAR-positive). The cell lineswith high uPAR expression were derived from invasive high-grade human tumors such as glioblastoma multiforme (U251),neuroblastoma (SK-N-AS), breast carcinoma (MDA-MB-231),prostate cancer (PC-3), and non-small-cell lung cancer (H1915).The cell lines expressing low levels of uPAR were derived fromless aggressive tumors of colon (COLO 320), breast (MCF-7),metastatic neuroblastoma (NB1691), and small-cell lung(H1415, H211) cancer. The uPA and uPAR expression levelswere confirmed by quantitative real-time RT-PCR (Fig. 1B).We also performed immunocytochemical analysis of uPAR incultured cells, which indicated expression of uPAR on the cellsurface and in the cytoplasm of tumor cells, validating the flowcytometry data (Fig. 1C).
Attraction of Stem Cells to uPAR-ExpressingTumors In VitroTo investigate the role of uPA and uPAR in the directed migra-tion of stem cells, we used immortalized human NSCs (HB1.F3)and fetal human bone marrow-derived primary mesenchymalstem cells (hfMSCs) for in vitro cell migration assays. NSCsshowed a robust migration to tumor-conditioned media fromcell lines that expressed high levels of uPAR (SK-N-AS, U251,PC-3, MDA-MB-231, H1915) (Fig. 2A), whereas we detectedsignificantly lower levels of NSC migration to tumor-condi-tioned media from cell lines with low uPAR expression (COLO-320, NB1691, H1415, H211, MCF-7) (Fig. 2A). HB1.F3 NSCsand hfMSCs displayed tropism similar to that of tumor cell-conditioned media (Fig. 2A, 2B), although more NSCs (50%–100% of total number of cells) than MSCs (40%–75% of totalnumber of cells) migrated.
Detection of uPA and suPAR in Conditioned MediaTumor cell-conditioned media were then analyzed by ELISA forthe presence of uPA and suPAR. Cleavage of uPAR at its D1 orD2 extracellular domain generates suPAR. We found high levelsof uPA in tumor cell-conditioned media derived from cell lineswith high uPAR levels (U251, PC-3, MDA-MB-231, H1915,but not SK-N-AS cells), whereas low uPA levels were detectedin conditioned media from cell lines with low uPAR expression
(COLO-320, NB1691, H1415, H211, MCF-7) (Fig. 3A). Ex-pression of uPAR on the SK-N-AS tumor cells was not asso-ciated with uPA activation. However, uPAR expression wasassociated with secretion of other cytokines in the SK-N-AS-conditioned media, such as HGF and MCP-1 (Fig. 4A), both ofwhich are known chemoattractants for stem and inflammatorycells. Upregulation of uPAR on tumor cells was associated withgreater uPAR shedding in the form of suPAR into the condi-tioned media (Fig. 3B). A putative function of suPAR in uPA-dependent stem cell migration might be associated with pres-ence of conserved chemotactic sequence motifs, which arepresent in the two- or three-domain versions of suPAR [30].Alternatively, by direct binding to the membrane of stem cells,suPAR may facilitate the uPA-dependent cell adhesion andmigration [31].
Correlation of Cytokine Profile and Stem CellAttractionTo identify other cytokines involved in stem cell migration, weused cytokine arrays to screen conditioned media derived fromhigh (PC-3, U251, SK-N-AS, H1915, MDA-MD-231) and low(MCF-7, H211) uPAR-expressing cells. Conditioned mediumderived from PC-3 cells showed high levels of IL-6, IL-8, andMCP-1, and U251 glioma-conditioned media showed high lev-els of IL-8, MCP-1, and VEGF, whereas SK-N-AS-conditionedmedia displayed high levels of MCP-1 and HGF (Fig. 4A, upperpanel; Fig. 4B, array key). Breast cancer cell line MDA-MD-231 showed induction of IL-6 and IL-8, whereas MCF-7-de-rived conditioned media had low levels of these cytokines (Fig.4A). Among lung cancer cell lines, H1915-conditioned mediarevealed high expression of IL-8 and tissue inhibitor of metal-loproteinases-2 (TIMP-2) compared with media derived fromthe H211 cell line (Fig. 4A). Quantitative analysis of cytokinearrays confirmed greater levels of expression of the followingcytokines in conditioned media from tumor cells with highversus low uPAR expression: IL-6, IL-8, tumor necrosis fac-tor-� (TNF-�), TNF-�, MCP-1, MCP-2, MCP-3, SDF-1,VEGF, and HGF (Fig. 4B). These results suggest that tumorcells with high uPAR expression (PC-3, U251, SK-N-AS,H1915, MDA-MD-231) display similar cytokine expression
Figure 1. uPAR expression in cell linesderived from prostate, lung, breast, colon,and brain cancers using flow cytometry anal-ysis (A), real-time quantitative reverse tran-scription-polymerase chain reaction (B), andimmunofluorescent staining (C). On the ba-sis of expression levels of uPAR, the celllines were grouped as (a) cell lines express-ing high levels of uPAR (�20%), or (b) celllines expressing low levels of uPAR (�5%).Green staining represents uPAR immunoflu-orescence; blue staining is 4,6-diamidino-2-phenylindole nuclear stain (C). Abbrevia-tions: uPA, urokinase plasminogen activator;uPAR, urokinase plasminogen activator receptor.
1409Gutova, Najbauer, Frank et al.
www.StemCells.com
profiles. The cytokine expression patterns of cells with lowuPAR (H211, MCF-7) were similar to each other but differentfrom the cytokine profiles of high uPAR-expressing cells (Fig.4A, lower panel).
Evidence for Direct Involvement of uPA and uPARin Stem Cell MigrationTo show a direct cause-and-effect relationship between uPARexpression on the tumor cells and uPA in the tumor-conditionedmedia, we focused on the PC-3 prostate cancer cell line and used
function-inhibiting anti-uPAR and anti-uPA antibodies andsiRNA-mediated gene silencing. To show a direct effect of uPAon stem cell migration, we used purified human recombinantuPA as a chemoattractant for hfMSCs and HB1.F3 stem cells(Fig. 5A). Migration of both mesenchymal and neural stem cellswas induced by addition of uPA to serum-free medium, and weobserved highest migration at 0.25 ng/ml of uPA. At all con-centrations of uPA tested (0.1–100 ng/ml), the NSCs showedhigher migration than MSCs, which is consistent with data frommigration assays using tumor cell-conditioned media.
To demonstrate an important role of uPAR expression onHB1.F3 neural stem cells and the migration of these cells totumor-conditioned media (SK-N-AS, U251, PC-3, MDA-MB-231, H1915), we preincubated the HB1.F3 cells with function-inhibiting anti-uPAR antibodies. As shown Figure 5B, migra-tion of HB1.F3 cells was inhibited by 50%–90%. In our controlexperiments, we preincubated the HB1.F3 cells with theserine protease inhibitor bikunin, a natural inhibitor of uPAR,which resulted in inhibition of NSC migration similar to thatobserved with anti-uPAR antibodies (data not shown). Wehave previously determined by fluorescence-activated cellsorting analysis that uPAR is expressed on both NSCs andMSCs (more than 50% of cells were uPAR-positive) (data notshown).
Since our data showed that expression of uPAR on tumorcells correlated with stem cell tropism (Fig. 2A), we reasonedthat siRNA-mediated inhibition of uPAR expression on PC-3cancer cells would result in inhibition of HB1.F3 neural stemcell migration. As shown in Figure 5C, conditioned mediaderived from PC-3-uPAR-siRNA cells, compared with condi-tioned media from PC-3 control cells, displayed inhibition ofattraction of HB1.F3 cells by 70%–90%. Furthermore, HB1.F3cell migration to PC-3-conditioned medium was inhibited bydepletion of uPA from PC-3-conditioned medium using anti-PAantibodies bound to protein A-Sepharose beads (�90% of inhi-bition; Fig. 5D).
Transduction of Tumor Cells with uPA and uPARGenes Induces Stem Cell Attraction In VitroTo further investigate the role of uPA and uPAR expression instem cell tropism to tumors, we transduced NB1691 neuroblas-toma cells (a low uPA and uPAR-expressing line) withpBABE.uPA and pBABE.uPAR retroviral vectors. Overexpres-sion of uPA and uPAR was confirmed by quantitative real-timeRT-PCR analysis (data not shown). We used in vitro cell mi-gration assays to test whether NSCs (HB1.F3 and HB1.F5) andMSCs (hfMSCs) would migrate toward conditioned media de-rived from the transduced NB1691 cells (Fig. 6A, 6B). Retro-viral vector-mediated overexpression of uPAR alone onNB1691 neuroblastoma cells leads to threefold and twofoldgreater HB1.F3 and hfMSC migration, respectively. uPA over-expression in NB1691 cells, however, caused 15-fold and 6-fold
Figure 2. In vitro migration assay of a human neural stem cell line(HB1.F3) and hfMSCs. Neural stem cells (NSCs) showed a robustmigration to conditioned media derived from tumor cells with highlevels of urokinase plasminogen activator receptor (uPAR) (SK-N-AS,U251, PC-3, MDA-MB-231, and H1915) (A), whereas we detected littlemigration to conditioned media from tumor cells with low uPAR ex-pression (COLO-320, NB1691, H1415, H211, and MCF-7) (A). hfM-SCs showed a similar migration pattern compared with NSCs (B). Dataare expressed as mean � SD of triplicate measurements. The figurerepresents the pooled data from three independent experiments. Abbre-viations: BSA, bovine serum albumin; FBS, fetal bovine serum; hfMSC,fetal human bone marrow-derived primary mesenchymal stem cell.
Figure 3. Upregulation of uPA receptor ontumor cell lines is associated with secretionof high levels of uPA (A) and increased uPAreceptor shedding (suPAR) into the condi-tioned media (B). Abbreviations: suPAR,soluble urokinase plasminogen activatorreceptor; uPA, urokinase plasminogen ac-tivator.
1410 Stem Cell Tropism to Tumors
greater NSC and MSC migration, respectively, compared withthe NB1691 wild-type cells. Coexpression of uPA and uPAR inNB1691 cells resulted in stem cell migration comparable to thattoward PC-3 conditioned media, which induced the greatestmigration of stem cells in previous experiments (Fig. 2).
DISCUSSION
Our data demonstrate that uPAR and its ligand, uPA, play amajor role in chemoattraction of stem cells to tumors. Cellularmigration and invasion are important during normal develop-ment and tissue remodeling and in several pathological condi-tions, including cancer. During brain development, uPAR andHGF (scatter factor [SF]) play a crucial role in guiding themigration of interneurons to the neocortex, as HGF/SF bioac-tivity is impaired in uPAR knockout mice [32]. uPAR andmatrix metalloproteinases are essential for extracellular matrixdegradation and tissue remodeling [33]. uPA and uPAR werefirst shown to be involved in stem cell migration during G-CSF-induced mobilization of CD34-positive hematopoietic stem cellsfrom bone marrow to the peripheral blood. G-CSF-induced stemcell mobilization is mediated by upregulation of uPAR oncirculating CD33 and CD14-positive cells, which then leads toincreased uPAR shedding and formation of soluble uPAR [27].
Here we report that both NSCs and MSCs show signifi-cantly greater migration toward cancer cells that express highlevels of uPA and uPAR, compared with tumor cells with lowexpression of uPA and uPAR. Activation of the uPA/uPARsystem on cancer cells caused the release of uPA and suPAR
into the conditioned media. Furthermore, expression of uPA anduPAR on cancer cells was associated with secretion of severalcytokines, including HGF, IL-6, IL-8, and MCP-1 (CCL2).Comparative analysis of cytokine profiles of conditioned mediashowed that IL-6 and IL-8 were highly expressed in MDA-MB-231 cells, compared with MCF-7 breast cancer cell line. Highlevels of IL-8 and TIMP-2 were secreted into conditioned mediafrom H1915 cells, compared with H1415 lung cancer cells,whereas conditioned media from SK-N-AS neuroblastoma cellshad elevated MCP-1 and HGF. The PC-3 prostate cancer cellline, which displayed the highest uPAR expression and che-moattraction of stem cells, had elevated levels of all the afore-mentioned cytokines secreted into the conditioned media. Thesedata suggest that NSCs and MSCs can use multiple cytokinesfor tropism to tumors but that a common feature of tumors thatattract stem cells is that they all express uPA and uPAR. Weobserved induction of stem cell migration to human recombi-nant uPA, whereas significant inhibition of stem cell migrationwas observed after blocking uPAR on the stem cells and tumorsor after depletion of uPA from tumor cell-conditioned media.Overexpression of uPA and uPAR alone or in combinationresulted in increased migration of NSCs and MSCs to condi-tioned media from NB1691 neuroblastoma cells. The chemoat-tracting activity of uPA may occur directly and/or indirectly bycausing the induction of other cytokines, such as IL-6, IL-8, andMCP-1. There is evidence that cancer cell migration and inva-sion, which are modulated by uPAR, involve HGF, insulin-likegrowth factor, matrix metalloproteinase (MMPs), IL-6, IL-8,and MCP-1, which suggests that cancer cells and stem cells mayuse similar signaling pathways during migration [34]. Earlierreports suggest that nuclear translocation of �-catenin leads to
Figure 4. Cytokine expression profile of the CM derived from tumor cell lines. (A): Cytokine array of CM derived from various cancer cell lines. (B):Key for the cytokine array. (C): Graphical representation of the relative expression of cytokines in the CM derived from cancer cell lines. Films were scannedand analyzed using ImageJ software. The average values for migrating versus nonmigrating cell lines were calculated and used for statistical comparison.Paired t test for each cytokine was performed to compare the mean values of cytokine expression in CM from the group of uPAR(�) versus uPAR(�) cancercell lines. The group of uPAR(�) cell lines showed overall higher expression of cytokines than the group of uPAR(�) cell lines, which was statisticallysignificant (�, p .004; t 3.556; df 11). Abbreviations: BDNF, brain-derived neurotrophic factor; CM, conditioned media; GCSF, granulocytecolony-stimulating factor; HGF, hepatocyte growth factor; IL, interleukin; LIF, leukemia inhibitory factor; MCP, monocyte chemoattractant protein; MCSF,macrophage colony-stimulating factor; Neg, negative; PDGF, platelet-derived growth factor; Pos, positive; TGF, transforming growth factor; TIMP, tissueinhibitor of metalloproteinases; TNF, tumor necrosis factor; uPAR, urokinase plasminogen activator receptor.
1411Gutova, Najbauer, Frank et al.
www.StemCells.com
simultaneous activation of target genes, such as uPAR and IL-8[35]. Furthermore, activation of the IL-6 and IL-8 genes leads toincreased expression of uPAR, MMP2, and MMP9 in solidtumors [36]. TIMPs are endogenous inhibitors of MMPs thathave recently been identified as signaling molecules, and theyact via the MEK1-Erk signaling network, similarly to uPA anduPAR [37]. In summary, our data suggest that uPA and uPARunderlie a novel mechanism of stem cell tropism to invasivesolid tumors. Identification and characterization of cancer-asso-
ciated cytokines, specific for various phenotypes and tumororigins, will be important in the development of effective stemcell-based targeted cancer therapies.
ACKNOWLEDGMENTS
We thank Dr. Kristine A. Justus and Dr. Keely Walker forexpert editing of the manuscript. We also thank Dr. Chu-Chih
Figure 5. Induction of stem cell migration by recombinant uPA and inhibition by uPA depletion and uPAR function-inhibiting antibodies or siRNAknockdown. (A): Migration of HB1.F3 cells and hfMSCs to human recombinant uPA protein. (B): Inhibition of HB1.F3 neuronal stem cell migration totumor-derived conditioned media after preincubation with anti-uPAR antibodies for 2 hours prior to cell migration assay. (C): Migration of HB1.F3 cellsto conditioned media derived from control PC-3 cells or siRNA knockdown of uPAR. Inset, uPAR expression in PC-3 cell was confirmed by flow cytometryanalysis after 48 hours of siRNA treatment. (D): Migration of HB1.F3 cells to PC-3 cell-derived conditioned media was inhibited by using anti-PA antibodiesbound to protein A-Sepharose beads (isotype-matched IgG was used as control). Five percent BSA and 10% FBS were used as negative and positive controlsfor migration assays, respectively. Data are expressed as mean � SD of triplicate measurements. Three independent experiments were performed.Abbreviations: BSA, bovine serum albumin; FBS, fetal bovine serum; hfMSC, fetal human bone marrow-derived primary mesenchymal stem cell; siRNA,small interfering RNA; uPA, urokinase plasminogen activator; uPAR, urokinase plasminogen activator receptor.
Figure 6. NSC (HB1.F3) and hfMSC migra-tion toward NB1691 neuroblastoma cells over-expressing uPA and uPAR using pantropicretroviral expression system. Transgene ex-pression was confirmed using real-time poly-merase chain reaction (data not shown). Mi-gration of NSCs (A) and mesenchymal stemcells (B) to NB1691.wt, NB1691.uPA, andNB1691.uPAR-derived conditioned media.FBS (10%) and PC-3 conditioned media wereused as positive control, whereas BSA (2%)served as negative control. Error bars indicateSD of triplicate measurements from two inde-pendent cell migration assays. Abbreviations:BSA, bovine serum albumin; FBS, fetal bo-vine serum; hfMSC, fetal human bone mar-row-derived primary mesenchymal stem cell;NSC, neural stem cell; uPA, urokinase plas-minogen activator; uPAR, urokinase plasmin-ogen activator receptor.
1412 Stem Cell Tropism to Tumors
Shih for kindly providing mesenchymal stem cells. This workwas supported by National Institutes of Health/National CancerInstitute (CA113446), The Stop Cancer Foundation, JosephDrown Foundation, The Rosalinde and Arthur Gilbert Founda-tion, Neidorf Family Foundation, H.L. Snyder Foundation, andZiman Family Foundation.
DISCLOSURE OF POTENTIAL CONFLICTS
OF INTEREST
The authors indicate no potential conflicts of interest.
REFERENCES
1 Goldman S. Stem and progenitor cell-based therapy of the human centralnervous system. Nat Biotechnol 2005;23:862–871.
2 Aboody KS, Brown A, Rainov NG et al. Neural stem cells displayextensive tropism for pathology in adult brain: Evidence from intracra-nial gliomas. Proc Natl Acad Sci U S A 2000;97:12846–12851.
3 Kim SU. Genetically engineered human neural stem cells for brain repairin neurological diseases. Brain Dev 2007;29:193–201.
5 Imitola J, Raddassi K, Park KI et al. Directed migration of neural stemcells to sites of CNS injury by the stromal cell-derived factor 1alpha/CXC chemokine receptor 4 pathway. Proc Natl Acad Sci U S A 2004;101:18117–18122.
6 Kucia M, Reca R, Miekus K et al. Trafficking of normal stem cells andmetastasis of cancer stem cells involve similar mechanisms: Pivotal roleof the SDF-1-CXCR4 axis. STEM CELLS 2005;23:879–894.
7 Erlandsson A, Larsson J, Forsberg-Nilsson K. Stem cell factor is achemoattractant and a survival factor for CNS stem cells. Exp Cell Res2004;301:201–210.
8 Heese O, Disko A, Zirkel D et al. Neural stem cell migration towardgliomas in vitro. Neuro-oncol 2005;7:476–484.
9 Kendall SE, Najbauer J, Johnston HF et al. Neural stem cell targeting ofglioma is dependent on PI3K signaling. STEM CELLS 2008 [Epub aheadof print].
10 Schmidt NO, Przylecki W, Yang W et al. Brain tumor tropism oftransplanted human neural stem cells is induced by vascular endothelialgrowth factor. Neoplasia 2005;7:623–629.
12 Palumbo R, Bianchi ME. High mobility group box 1 protein, a cue forstem cell recruitment. Biochem Pharmacol 2004;68:1165–1170.
13 Palumbo R, Galvez BG, Pusterla T et al. Cells migrating to sites of tissuedamage in response to the danger signal HMGB1 require NF-kappaBactivation. J Cell Biol 2007;179:33–40.
14 Rafii S, Lyden D. Therapeutic stem and progenitor cell transplantationfor organ vascularization and regeneration. Nat Med 2003;9:702–712.
15 Son BR, Marquez-Curtis LA, Kucia M et al. Migration of bone marrowand cord blood mesenchymal stem cells in vitro is regulated by stromal-derived factor-1-CXCR4 and hepatocyte growth factor-c-met axes andinvolves matrix metalloproteinases. STEM CELLS 2006;24:1254–1264.
16 Wysoczynski M, Reca R, Ratajczak J et al. Incorporation of CXCR4 intomembrane lipid rafts primes homing-related responses of hematopoieticstem/progenitor cells to an SDF-1 gradient. Blood 2005;105:40–48.
17 Ziu M, Schmidt NO, Cargioli TG et al. Glioma-produced extracellularmatrix influences brain tumor tropism of human neural stem cells.J Neurooncol 2006;79:125–133.
18 Nagai M, Hiramatsu R, Kaneda T et al. Molecular cloning of cDNAcoding for human preprourokinase. Gene 1985;36:183–188.
19 Aguirre Ghiso JA, Alonso DF, Farias EF et al. Deregulation of thesignaling pathways controlling urokinase production. Its relationshipwith the invasive phenotype. Eur J Biochem 1999;263:295–304.
20 Alfano D, Franco P, Vocca I et al. The urokinase plasminogen activator
and its receptor: Role in cell growth and apoptosis. Thromb Haemost2005;93:205–211.
21 Blasi F, Carmeliet P. uPAR: A versatile signalling orchestrator. Nat RevMol Cell Biol 2002;3:932–943.
22 Eagleson KL, Bonnin A, Levitt P. Region- and age-specific deficits ingamma-aminobutyric acidergic neuron development in the telencephalonof the uPAR(-/-) mouse. J Comp Neurol 2005;489:449–466.
23 Gutova M, Najbauer J, Gevorgyan A et al. Identification of uPAR-positive chemoresistant cells in small cell lung cancer. PLoS ONE2007;2:e243.
24 Rao JS, Gondi C, Chetty C et al. Inhibition of invasion, angiogenesis,tumor growth, and metastasis by adenovirus-mediated transfer of anti-sense uPAR and MMP-9 in non-small cell lung cancer cells. Mol CancerTher 2005;4:1399–1408.
25 Rigolin GM, Tieghi A, Ciccone M et al. Soluble urokinase-type plas-minogen activator receptor (suPAR) as an independent factor predictingworse prognosis and extra-bone marrow involvement in multiple my-eloma patients. Br J Haematol 2003;120:953–959.
26 Zhou HQ, Chen G, Lu XG et al. Detection of plasma urokinase-typeplasminogen activator and its receptor in acute or chronic hepatitis Bpatients and its clinical significance [in Chinese]. Zhonghua Shi Yan HeLin Chuang Bing Du Xue Za Zhi 2006;20:238–240.
27 Selleri C, Montuori N, Ricci P et al. Involvement of the urokinase-typeplasminogen activator receptor in hematopoietic stem cell mobilization.Blood 2005;105:2198–2205.
28 Kim SU. Human neural stem cells genetically modified for brain repairin neurological disorders. Neuropathology 2004;24:159–171.
29 Kim SK, Kim SU, Park IH et al. Human neural stem cells targetexperimental intracranial medulloblastoma and deliver a therapeutic geneleading to tumor regression. Clin Cancer Res 2006;12:5550–5556.
30 Trigwell S, Wood L, Jones P. Soluble urokinase receptor promotes celladhesion and requires tyrosine-92 for activation of p56/59(hck). Bio-chem Biophys Res Commun 2000;278:440–446.
31 Mizukami IF, Todd RF 3rd. A soluble form of the urokinase plasminogenactivator receptor (suPAR) can bind to hematopoietic cells. J LeukocBiol 1998;64:203–213.
32 Levitt P. Disruption of interneuron development. Epilepsia 2005;46(suppl 7):22–28.
33 Zhang Y, Zhou ZH, Bugge TH et al. Urokinase-type plasminogenactivator stimulation of monocyte matrix metalloproteinase-1 productionis mediated by plasmin-dependent signaling through annexin A2 andinhibited by inactive plasmin. J Immunol 2007;179:3297–3304.
34 Gondi CS, Kandhukuri N, Dinh DH et al. Down-regulation of uPAR anduPA activates caspase-mediated apoptosis and inhibits the PI3K/AKTpathway. Int J Oncol 2007;31:19–27.
35 Levy L, Neuveut C, Renard CA et al. Transcriptional activationof interleukin-8 by beta-catenin-Tcf4. J Biol Chem 2002;277:42386 – 42393.
36 Rolhion C, Penault-Llorca F, Kemeny JL et al. Interleukin-6 overexpres-sion as a marker of malignancy in human gliomas. J Neurosurg 2001;94:97–101.
37 Biswas MH, Hasegawa HH, Rahman MA et al. SHP-2-Erk signalingregulates concanavalin A-dependent production of TIMP-2. BiochemBiophys Res Commun 2006;348:1145–1149.
