Ceramide Glycosylation by Glucosylceramide SynthaseSelectively Maintains the Properties of Breast Cancer StemCells*□S

Received for publication, June 29, 2012, and in revised form, August 28, 2012 Published, JBC Papers in Press, August 30, 2012, DOI 10.1074/jbc.M112.396390

Vineet Gupta‡, Kaustubh N. Bhinge‡, Salman B. Hosain‡, Katherine Xiong‡, Xin Gu§, Runhua Shi¶, Ming-Yi Ho�,Kay-Hooi Khoo¶, Su-Chen Li**, Yu-Teh Li**, Suresh V. Ambudkar‡‡1, S. Michal Jazwinski§§, and Yong-Yu Liu‡2

From the ‡Department of Basic Pharmaceutical Sciences, University of Louisiana at Monroe, Monroe, Louisiana 71209, the§Department of Pathology and the ¶Feist-Weiller Cancer Center, Louisiana State University Health Sciences Center, Shreveport,Louisiana 71130, the �Institute of Biological Chemistry, Academia Sinica, Nankang, Taipei 115, Taiwan, the **Department ofBiochemistry and Molecular Biology and §§Department of Medicine and Tulane Center for Aging, Tulane University School ofMedicine, New Orleans, Louisiana 70112, and the ‡‡Laboratory of Cell Biology, National Cancer Institute,Bethesda, Maryland 20892

Background:Glucosylceramide synthase catalyzes ceramide glycosylation that regulates the synthesis of glycosphingolipids.Results: Increased globo-series glycosphingolipids in breast cancer stem cells activate c-Src signaling and �-catenin-mediatedtranscription up-regulating stem cell factors.Conclusion: Ceramide glycosylation maintains the stemness of cancer stem cells.Significance: Glycosphingolipids in cell membrane actively participate in maintaining cancer stem cells.

Cancer stem cells are distinguished from normal adult stemcells by their stemness without tissue homeostasis control. Gly-cosphingolipids (GSLs), particularly globo-series GSLs, areimportant markers of undifferentiated embryonic stem cells,but little is known about whether or not ceramide glycosylation,which controls glycosphingolipid synthesis, plays a role inmod-ulating stem cells. Here, we report that ceramide glycosylationcatalyzed by glucosylceramide synthase, which is enhanced inbreast cancer stem cells (BCSCs) but not in normal mammaryepithelial stem cells, maintains tumorous pluripotency ofBCSCs. Enhanced ceramide glycosylation and globotriosylcer-amide (Gb3) correlate well with the numbers of BCSCs in breastcancer cell lines. In BCSCs sorted with CD44�/ESA�/CD24�

markers, Gb3 activates c-Src/�-catenin signaling and up-regu-lates the expression of FGF-2, CD44, and Oct-4 enrichingtumorigenesis.Conversely, silencing glucosylceramide synthaseexpression disrupts Gb3 synthesis and selectively kills BCSCsthrough deactivation of c-Src/�-catenin signaling. These find-ings highlight the unexploited role of ceramide glycosylation inselectively maintaining the tumorous pluripotency of cancerstem cells. It speculates that disruption of ceramide glycosyla-tion or globo-series GSL is a useful approach to specifically tar-get BCSCs specifically.

Cancer stem cells (CSCs),3 which have been characterizedwith surface markers in tumors, possess the malignant capaci-ties of self-renewal and pluripotency, thus initiating tumorigen-esis and driving tumor progression (1–4). In human breast can-cer, the CD44�/ESA�/CD24�/low cells have been tested asbreast cancer stem cells (BCSCs), because they are able to dif-ferentiate into cells with diverse phenotypes and have tumor-ous pluripotency to generate mammary tumors andmetastasesin vivo (3, 5). BCSCs, like other CSCs, give rise to tumor resist-ance to chemotherapy and radiation (6–8). As a cause of tumormetastasis and recurrence, CSC is an adverse prognostic factorfor cancer patients (9–11) and a critical target to eradicate can-cers (12–15). CSCs are distinguished by loss of tissue homeo-stasis control and often display unlimited proliferation andgrowth from normal stem cells, even though both share thesimilar properties in self-renewal, pluripotency, and resistanceto cytotoxins (2, 16). Targeting CSCs pharmacologically fortherapeutic purpose requires understanding by which mecha-nisms CSCs maintain their tumor behaviors.Several cellular signals includingWnt, Notch, andHedgehog

have been reported to be implicated in the self-renewal abilityand pluripotency of normal stem cells in mammary glanddevelopment or remodeling and of BCSCs in cancer pathogen-esis (5, 16–19). It is not clear how BCSCs, like other CSCs,maintain tumor pluripotency without tissue homeostasis con-trol. Glycosphingolipids (GSLs), particular globo-series GSLs,are common markers of undifferentiated embryonic stem cells(ESCs) (20–22). As essential components of lipid rafts, partic-ularly GSL-enriched microdomains in plasma membrane,

* This work was supported, in whole or in part, by National Institutes of HealthGrants 5P20RR016456-11 from the National Center for Research Resourcesand 8 P20 GM103424-11 from the National Institute of General MedicalSciences (to Y.-Y. L.). This work was also supported by funds from the Mizu-tani Foundation for Glycoscience, Japan (to Y.-Y. L.).

□S This article contains supplemental Fig. S1.1 Supported by the Intramural Research Program, Center for Cancer Research,

NCI, National Institutes of Health.2 To whom correspondence should be addressed: Dept. of Basic Pharmaceu-

tical Sciences, University of Louisiana at Monroe, 700 University Ave., Mon-roe, LA 71209. Tel.: 318-342-1709; Fax: 318-342-1737; E-mail: yliu@ulm.edu.

3 The abbreviations used are: CSC, cancer stem cell; BCSC, breast CSC; GSL,glycosphingolipid; ESC, embryonic stem cell; GCS, glucosylceramide syn-thase; MBO, mixed backbone oligonucleotide; HPTLC, high performancethin layer chromatography; Gb3, globotriosylceramide; PP2, pyrazolopyrimidine; Gb5, globopentosylceramide; MSGb5, monosialyl globo-pentosylceramide; NBD, 7-nitrobenz-2-oxa-1,3-diazole.

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 287, NO. 44, pp. 37195–37205, October 26, 2012Published in the U.S.A.

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37195

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

http://www.jbc.org/content/suppl/2012/08/30/M112.396390.DC1.html Supplemental Material can be found at:

GSLs actively mediate cellular signals, gene regulation, andfunctions (23–26). GSLs may play a vital role in maintainingESCs, because deletion of GSLs induces apoptosis of ESCs andstops embryo development in ugcg (encoding glucosylceramidesynthase, GCS) knock-out mouse (27). GCS catalyzes ceramideglycosylation and is a limiting enzyme controlling the synthesisofGSLs (28, 29). Enhanced ceramide glycosylation byGCS con-verts ceramide toGSLs, conferringmultidrug resistance of can-cer cells (30, 31). It has been reported thatGCS is overexpressedin metastatic breast tumors (32, 33) and inhibition of GCSdecreases lung tumor metastasis (34). These studies suggestthat ceramide glycosylation by GCS is associated with CSCbehaviors. We examined the correlation of ceramide glycosyl-ation with BCSCs in drug resistance and tumorigenesis.

EXPERIMENTAL PROCEDURES

Cell Culture and Treatments—HumanMCF-7 breast cancercells and the doxorubicin-selected subline MCF-7/Dox werekindly provided by Dr. Kapil Mehta (M. D. Anderson CancerCenter, Houston, TX). MCF-7/Dox cells were derived fromMCF-7 cells by stepwise culture with doxorubicin (35). HumanMCF-12A mammary epithelial cells were purchased fromAmerican Type Culture Collection (Manassas, VA). TheMCF-7 and MCF-7/Dox were cultured in RPMI 1640 mediumcontaining 10% FBS, 100 units/ml penicillin, 100 �g/ml strep-tomycin, and 584 mg/liter L-glutamine. MCF-12A cells werecultured in Dulbecco’s modified Eagle’s medium/F-12 (1:1)with 5% horse serum, insulin (5 �g/ml), hydrocortisone (500ng/ml), human epidermal growth factor (20 ng/ml), and chol-era toxin (100 ng/ml). All of the cells were maintained in anincubator humidified with 95% air and 5% CO2 at 37 °C. Celllines were authenticated in November 2010 at the John Hop-kins University Fragment Analysis Facility (Baltimore, MD) byprofiling short tandem repeats for breast cells. FBS was pur-chased from HyClone (Logan, UT), medium was from Invitro-gen, and other reagents from Sigma-Aldrich.A mixed backbone oligonucleotide (MBO) was designed to

target the ORF 18–37 of human GCS and designated as MBO-asGCS (36, 37). ForMBO-asGCS treatment, cells (2� 106 cells/100-mm dish) were grown in 10% FBS RPMI 1640 mediumovernight, and MBO-asGCS (100 nM) were introduced intocells using Lipofectamine 2000 (Invitrogen). The cells were cul-tured with MBO-asGCS in 10% FBS RPMI 1640 medium for 7days. An additional transfection of MBO-asGCS under thesame conditions was conducted on day 4. MBO-asGCS wassynthesized and purified by reverse phase HPLC and desaltingin Integrated DNA Technologies (Coralville, IA).Cell Viability Assay—Cell viability was analyzed by quantifi-

cation of ATP, an indicator of metabolically active cells usingthe CellTiter-Glo luminescent cell viability assay (Promega,Madison,WI), as described previously (38). Briefly, cells (4,000cells/well) were grown in 96-well plates with 10% FBS RPMI1640 medium for 24 h. MBO-asGCS was introduced intocells by Lipofectamine 2000 (vehicle control) in Opti-MEMreduced serum medium for 4 h of incubation. The cells werethen incubated with increasing concentrations of doxorubi-cin in 5% FBS medium for another 72 h. Cell viability wasdetermined by the measurement of luminescent ATP in a

Synergy HT microplate reader (BioTek, Winooski, VT), fol-lowing incubation with CellTiter-Glo reagent (Promega).Colony Formation Assay—The cells (10,000 cells/800 �l)

were suspended in 10% FBS RPMI 1640 medium containing0.2% agarose and overlaid onto 24-well plates containing asolidified bottom layer of 0.3% agarose (600 �l) in 10% FBSRPMI 1640 medium. Once the top layer solidified, 600 �l of10% FBS RPMI 1640mediumwas placed on the top to keep thewells moist. For MBO-asGCS treatment, the cells were pre-treated with MBO-asGCS as described above. In addition, theMBO-asGCS (200 nM) was added with the 0.2% agarosemedium (top layer). The plates were incubated for 1 week untilthe colonies were visible. The cell numbers of colonies weredetermined by the measurement of luminescent ATP in aSynergy HT microplate reader, following incubation withCellTiter-Glo reagent for 30 min, at room temperature withgentle shaking.Tumor Sphere Assay—It was performed as described previ-

ously with minor modification (39, 40). Briefly, after sorting,BCSCs or CD44�/CD24� cells (5,000 cells/well) were plated inultralow attachment 24-well plates (Corning, Lowell, MA) withDMEM/F-12 (1:1) medium containing insulin (5 �g/ml),human basic fibroblast growth factor (10 ng/ml), human epi-dermal growth factor (20 ng/ml), and 0.4% BSA. The BCSCswere treated with MBO-asGCS (100 nM) for 7 days, and themedium was refreshed on day 4. The tumor spheres wereobserved and photomicrographed.Cell Sorting—Sorting was performed using magnetic beads

(Dynabeads) as described in the manufacturer’s protocol.Briefly, after harvest with trypsin-EDTA and PBS pH 7.4 wash,cells (1 � 107 cells) were incubated with mouse anti-humanCD24 monoclonal antibody (25 �l; Santa Cruz Biotechnology,Santa Cruz, CA) in 1 ml of cells sorting buffer (PBS containing0.1% BSA and 2 mM EDTA, pH 7.4) at 4 °C for 10 min. Afterremoval of unbound antibody by centrifugation and PBS wash-ing, the cells were incubated with goat anti-mouse IgG anti-body-conjugated Dynabeads M-280 (1 � 107 beads/ml; Invit-rogen) at 4 °C for 30 min and placed in a MagCellect magnet(R & D Systems, Minneapolis, MN) for 1 min. An unboundfraction of cells (CD24�) was collected for flow cytometry anal-ysis and immunohistochemistry. To separate BCSC and othersubpopulations, the CD24� fraction from MCF-7/Dox cells(1 � 107 cells) was further incubated with 25 �l of mouse anti-human CD44 monoclonal antibody (Sigma) at 4 °C for 10 minand with goat anti-mouse IgG antibody-conjugated DynabeadsM-280 (1 � 107 beads/ml) for 20 min following removal ofunbound antibody. TheCD44� fraction (CD44�/CD24�) wereused for further analysis and CD44� fraction (CD44�/CD24�)was cultured in 10% FBS RPMI 1640 medium for 48 h to disso-ciate antibody Dynabeads. After incubation with mouse anti-ESA/flotillin-2 antibody (25�l/10� 107 cells; fromSantaCruz)and then anti-mouse IgG antibody-conjugated DynabeadsM-280 (1 � 107 beads/ml), the ESA� fraction (ESA�/CD44�/CD24�) and ESA� fraction (ESA�/CD44�/CD24�) were har-vested after magnetic separation and cultured for furtherexperiments. The MBO-asGCS (100 nM) treatment of MCF-7/Dox cells was performed as described above.

Ceramide Glycosylation Maintains Cancer Stem Cells

37196 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 44 • OCTOBER 26, 2012

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

Flow Cytometry—Flow cytometry was performed asdescribed previously (15). For analysis of BCSCs, the CD24�

cells sorted from each cell line were incubated with AlexaFluor� 488 anti-CD44 mouse antibody (5 �l/106 cells; BioLeg-end, SanDiego, CA) andAlexa Fluor� 647 anti-ESA antibody (5�l/106 cells; BioLegend) in blocking buffer (PBS containing 5%serum) at 4 °C for 30 min. After washing off unbound antibod-ies, the cells were resuspended in 1 ml of PBS and immediatelyanalyzed on a BD FACSCalibur with the BDCellQuest Pro pro-gram (BD Bioscience, San Jose, CA) and FlowJo v10 (Tree Star,Ashland, OR). To identify CD44�/ESA� cells, each sample wasincubated in RPMImedium containing serum to determine theautofluorescence, as negative control. To analyze BCSCs intumors, cell suspensions were prepared immediately afterresection (�10 min). Tumor tissues (�60 mg) were cut intosmall pieces (�1 mm) and incubated with collagenase IV (500units/ml; purchased from Sigma) in RPMI 1640 and incubatedat 37 °C for 2 h with shaking (100 rpm). Further, the cells werepassed through a 70-�m cell strainer and washed twice withPBS.Cellular Ceramide Glycosylation and GSL Analysis—Cer-

amide glycosylation that is catalyzed in cells was analyzed asdescribed previously (41). Cells were grown 24 h in 35-mmdishes (5 � 106 cells/dish) in 10% FBS RPMI 1640 medium andMBO-asGCS (100 nM) was introduced as described above.After 12 h of growth in 10% FBS RPMI 1640 medium, the cellswere switched to 1% BSA RPMI 1640 medium containing 5 �M

NBD C6-ceramide complexed to BSA (Invitrogen). After 2 h ofincubation at 37 °C, the lipids were extracted and resolved onPartisil high performance thin layer chromatography (HPTLC)plateswith fluorescent indicator (number 4806-410;Whatman,Florham Park, NJ) in a solvent system containing chloroform,methanol, and 3.5 N ammonium hydroxide (85:15:1, v/v/v) asdescribed previously (41). NBD C6-glucosylceramide and NBDC6-ceramide were identified using AlphaImager HP imagingsystem (Alpha Innotech, San Leandro, CA) and quantitated ona Synergy HT multi-detection microplate reader. For quantifi-cation, calibration curves were established after TLC separa-tion ofNBDC6-ceramide (Invitrogen) andNBDC6-glucosylce-ramide (N-hexanol-NBD-glucosylceramide; Matreya, PleasantGap, PA).The endogenous GSL were extracted and analyzed as

described previously (25, 42). Briefly, cellular lipids wereextracted with chloroform/methanol/water (1:1:1, v/v/v) fromeach cell line or subpopulation after vehicle or MBO-asGCStreatments. Extracted lipids were resuspended in chloroform/methanol (1:1, v/v) and applied to Partisil HPTLC plates withfluorescent indicator. Lipids were resolved using the solventsystems of chloroform/methanol/water (65:35:8, v/v/v) forGSLs. HPTLC plates were dipped for 10 s in 0.02% primuline(w/v; purchased from Sigma) in acetone/water (4:1, v/v). Fluo-rescence TLC profile graph was visualized under long wave UVlight (360 nm) and captured using AlphaImager HP system(Alpha Innotech). Neutral GSL Qualimix (Matreya) were usedas TLC standards. Globotriosylceramide (Gb3) of each samplewas quantified by fluorescence intensity against the standardcurve established using Gb3 lipids (Matreya) and normalizedagainst cellular proteins.

MALDI-MSandMSAnalysis ofGSLs—MALDI-MSprofilingof permethylated GSLs was performed as described previously(43). Briefly, the extracted GSLs (from 7.5 � 105 cells) wereincubated with leech ceramide glycanase (50 milliunits/100 �l)(44) in 50 mM sodium acetate, pH 5.5, containing 0.1% sodiumcholate at 37 °C for 16 h. The released glycans were separatedfrom the ceramides anddetergent by passing through a Sep-PakC18 cartridge (Waters, Milford, MA), and further desalted by aporous graphitized carbon solid phase extraction cartridge.Permethylation was carried out by the NaOH/Me2SO slurrymethod as described previously (45). MALDI-MS analysis ofthe derivatives were carried out in positive ion mode on a ABI4700 MALDI-TOF/TOF mass spectrometer (Applied Biosys-tems, Foster City, CA) using 2�,4�,6�-trihydroxyacetophenoneas the MALDI matrix. Gb3 mass intensity was normalizedagainst cellular proteins of each sample.Western Blot Analysis—After treatment, cells or tissue

homogenates were lysed using Nonidet P-40 cell lysis buffer(BIOSOURCE, Camarillo, CA). The amount of total proteinswas measured using a BCA protein assay kit (Pierce). Equalamounts of detergent-soluble proteins (50 �g/lane) wereresolved using 4–20% gradient SDS-PAGE (Invitrogen).After transferring, the blots were blocked in 5% fat-free milkin PBS for 1 h at 26 °C. The blots were then incubated withspecific primary antibodies against GCS, FGF-2, Oct-4,CD44, c-Src, phosphorylated c-Src, �-catenin, and phos-phorylated �-catenin (1:200 to 1:5000 dilution) overnight at4 °C and then with respective horseradish peroxide-conjugatedsecondary antibodies (1:2500 dilution) for 1 h at 26 °C afterwashing. The proteins were detected using enzyme-linkedSuperSignal� West Femto Maximum Sensitivity Substrate(Thermo Scientific, Rockford, IL) as described previously (25,30). EndogenousGAPDHwas used as a loading control for eachsample.Tumor-bearing Mice and Treatments—All of the animal

experiments were approved by the Institutional Animal Careand Use Committee of the University of Louisiana at Monroeand were handled in strict accordance with good animal prac-tice as defined by National Institutes of Health guidelines.Athymic nude mice (Foxn1nu/Foxn1�, 4–5 weeks, female)were purchased fromHarlan (Indianapolis, IN) andmaintainedin the vivarium at University of Louisiana at Monroe. Animalstudies were conducted as described previously (25, 36). Briefly,cells of each subpopulation (BCSC, CD44�/CD24�) were cul-tured 24 h after sorting and resuspended in serum-free RPMI1640 medium (20 �l). After anesthesia, cell suspensions (2.5 �104 cells/20 �l or 1 � 105 cells/20 �l) were inoculated into thesecond mammary gland fat pad just beneath the nipple of eachmouse implanted with 17�-estradiol tablets (0.72 mg, 90 daysrelease; Innovative Research of America, Sarasota, FL) to gen-erate orthotopic breast tumors. The mice were monitored bymeasuring tumor volume, body weight, and clinical observa-tion. Tumor volume (V) was calculated byV�L/2�W2, whereL was the length, andW was the width of tumors. Tumors andmetastasis of mice were examined by a pathologist with hema-toxylin and eosin stain of tissues sections at Louisiana StateUniversity Health Science Center (Shreveport, LA).

Ceramide Glycosylation Maintains Cancer Stem Cells

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37197

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

Statistical Analysis—All of the experiments were conductedin triplicate and repeated two times. The data were analyzed byusing Prism version 4 (GraphPad software, San Diego, CA) orSAS 9.2 (SAS Institute, Gary, NC) and presented as themeans � S.D. Two-tailed Student’s t tests were used to com-pare the continuous variables between groups, and Fisher’sextract test was used to compare the proportion betweengroups. All p � 0.05 was considered statistically significant.

RESULTS

GCS Is Associated with BCSCs in Drug Resistance andTumorigenicity of Cancer Cells—Different cancer cell lines dis-play their own behavior in response to anticancer drugs and intumor formation. In assessment of cell response to doxorubi-cin, we found that the half-maximal inhibitory concentration(IC50) value for doxorubicin in MCF-7/Dox cells was 30-fold(9.8 �M versus 0.33 �M; p � 0.001) higher than the MCF-7breast adenocarcinoma cells or human MCF-12A mammaryepithelial cells (Fig. 1A). These results are consistent with pre-vious reports, because MCF-7/Dox cells, which have beenselected with doxorubicin culture from MCF-7 cells, are resis-tant to doxorubicin (35). Furthermore,MBO-asGCS treatment(100 nM) to silence GCS expression (36) in the present studysignificantly decreased the IC50 for doxorubicin by 13-fold (9.8�M versus 0.75 �M; p � 0.001) in MCF-7/Dox cells (Fig. 1A).Examining colony formation in soft agar, we found that MCF-7/Dox cells formed160%more colonies (9,325 cells versus 5,715cells;p� 0.001) thanMCF-7 cells orMCF-12Acells (Fig. 1B); incontrast, MBO-asGCS treatment reduced the colonies by2-fold (9,315 cells versus 4,702 cells; p � 0.001) in MCF-7/Doxcells (Fig. 1B). These results show that GCS expression is asso-ciated with drug resistance and tumorigenicity, which are char-acteristics of CSCs in these cancer cell lines.The cells with CD44�/ESA�/CD24� phenotype have been

characterized as BCSCs in tumors and in cancer cell lines (3, 6,46). To test whether BCSC is a cause of drug resistance andtumorigenicity presenting in MCF-7/Dox cells, we analyzed

BCSCs with CD44�/ESA�/CD24� phenotype in these celllines. In CD24 separation, �36% ofMCF-7 and 54% ofMCF-7/Dox were CD24� cells. Different from these cancer cell lines,only 28% of MCF-12A cells were CD24�. It was found that thenumber of BCSCs (CD44�/ESA�/CD24�) inMCF-7/Dox cellswas 6-fold (3.1% versus 0.5% of cancer cells; p � 0.001) morethan MCF-7 and 10-fold (3.1% versus 0.3% of cells; p � 0.001)more than MCF-12A cells (Fig. 2). Silencing of GCS usingMBO-asGCS significantly reduced the BCSCs to 42% (1.3% ver-sus 3.1% of cancer cells; p � 0.001) inMCF-7/Dox cells (Fig. 2).Ceramide Glycosylation Catalyzed by GCS Is Crucial for

Tumor Formation—To determine whether ceramide glycosyl-ation plays a role in modulating BCSCs, we examined the cor-relation of GCS expression, ceramide glycosylation, and endog-enous GSLs with BCSCs. Western blotting showed that GCSlevels inMCF-7/Dox cells were three times higher thanMCF-7or MCF-12A cells; silencing of GCS using MBO-asGCS sub-stantially reduced GCS to 40% in MCF-7/Dox cells (Fig. 3A).Interestingly, the expression of FGF-2 and Oct-4, which arerequired for the proliferation of normal stem cells and CSCs (2,47, 48), was substantially increasedwith levels ofGCSprotein inthese cells (Fig. 3A). Consistent with protein level, GCS activityin MCF-7/Dox cells was 2-fold (26 pmol/mg versus 13 pmol/mg) higher than MCF-7 cells or MCF-12A cells (Fig. 3B). TheGSL profiling analyzed by HPTLC indicated that GSLs, partic-ularly of Gb3 in MCF-7/Dox cells were 2-fold (32.6 versus 15.4ng/mg; p � 0.001) greater than MCF-7 and 4-fold (32.6 versus8.2 ng/mg; p � 0.001) greater than MCF-12A cells (Fig. 3C).Silencing of GCS with MBO-asGCS decreased GSLs, and theGb3was reduced to 28% (9.1 ng versus 32.6 ng/mg;p� 0.001) inMCF-7/Dox cells (Fig. 3C). To verify these findings, GSLs fur-ther profiled using MALDI-MS and MS/MS analysis (43). TheGb3 levels in MCF-/Dox cells was 250% higher than in MCF-7cells (10,179 versus 4,139 in absolute height; p � 0.001) and347% higher than in MCF-12A cells (10,179 versus 2,833 inabsolute height; p � 0.001) (Fig. 3, D–F). MBO-asGCS treat-

FIGURE 1. Cell response to doxorubicin and colony formation of humanbreast cancer cells. A, cell response to doxorubicin. The cells were exposedto increasing concentrations of doxorubicin (0 –20 �M) in 5% FBS medium for72 h. The half-maximal inhibitory concentration (IC50) for cell viability wascalculated using Prism software after measurement. Dox/asGCS, MCF-7/Doxcells treated with MBO-asGCS (100 nM). *, p � 0.001compared with MCF-7cells; **, p � 0.001 compared with vehicle control of MCF-7/Dox cells. B, col-ony formation in soft agar. The colonies (cell numbers) formed in agarose(0.2%, 7 days) were measured using CellTiter-Glo reagent.

FIGURE 2. Breast cancer stem cells in human breast cancer cell lines.A, flow cytometry histograms of CD24� cells. The CD44�/ESA� cells (BCSCs)enclosed in ellipses (upper right) were compared with negative controls(CD44�, ESA�) of the CD24� fraction following magnetic separation of eachcell line. B, BCSCs. After magnetic separation (CD24) and flow cytometry anal-ysis (CD44/ESA), BCSCs were determined and presented as the percentagesof total cells. Dox/asGCS, MCF-7/Dox cells treated with 100 nM MBO-asGCS for7 days. *, p � 0.001 compared with MCF-7 cells; **, p � 0.001 compared withvehicle control of MCF-7/Dox cells.

Ceramide Glycosylation Maintains Cancer Stem Cells

37198 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 44 • OCTOBER 26, 2012

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

ment reduced the Gb3 to 33% (3349 versus 10179 in absoluteheight; p � 0.001) in MCF-7/Dox cells (Fig. 3D). We furtherexamined whether or not GCS itself is able to induce BCSCsin MCF-7 cells. As shown in supplemental Fig. S1, GCStransfection did not significantly enhance either BCSC num-

ber or protein levels of FGF-2 and Oct-4 in MCF-7/GCScells, although these cells overexpressed GCS. These resultsindicate that GCS and Gb3 modulate the maintenance ofBCSCs, possibly through stem cell mediators like FGF-2 andOct-4.

FIGURE 3. GCS expression and GSLs in human breast cancer cell lines. Dox/asGCS, MCF-7/Dox cells treated with MBO-asGCS (100 nM) for 7 days. A, Westernblot. Detergent-soluble proteins (50 �g/lane) were resolved by 4 –20% SDS-PAGE and immunoblotted with anti-GCS, FGF-2, and Oct-4 antibodies, respectively.B, fluorescence chromatogram of GCS activity. The cells were incubated with NBD C6-ceramide (50 �M, 2 h) in 1% BSA RPMI 1640 medium to measure cellularGCS activity. An equal amount of lipids (5,000 FI) was spotted on an HPTLC plate. Cer, NBD C6-ceramide; GlcCer, NBD C6-glucosylceramide synthase. GCS activityis presented as GlcCer generated per cellular proteins (pmol/mg). *, p � 0.001, compared with MCF-7 or MCF-12A cells; **, p � 0.001, compared with vehiclecontrol of MCF-7/Dox cells. C, GSL profiling. Equal amounts of cellular lipids (equal to 100 �g of proteins) were separated by HPTLC and detected by stainingwith 0.02% primuline and followed by fluorescence imaging. Std, standards of neutral glycosphingolipid Qualimix; GlcCer, glucosylceramide; LC, lactosylcer-amide. Endogenous Gb3 was quantitatively measured by comparison of the fluorescence intensity to a standard curve and normalized against proteins. *, p �0.001, compared with MCF-7 or MCF-12A cells; **, p � 0.001, compared with vehicle control of MCF-7/Dox cells. D, Gb3 MS intensities of cancer cells. GSLs wereanalyzed by using MALDI-MS analysis, and the Gb3 MS intensity was normalized against cellular proteins of each sample. E, MALDI-MS profiles of permethy-lated GSLs of normal mammary epithelial MCF-12A cells. F, MALDI-MS profiles of permethylated GSLs of MCF-7/Dox breast cancer cells. Lc, lactosylceramide;*, unidentified peaks; Gb3 abs int., absolute MS intensity of Gb3, as calculated by the percentage of Gb3 peak � MS response (the count on the right y axis).

Ceramide Glycosylation Maintains Cancer Stem Cells

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37199

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

Furthermore, we observed tumor formation of MCF-12A,MCF-7, andMCF-7/Dox cell lines in vivo. After 35 days inocu-lation, there were no tumors detected inMCF-12A group (zeroof five mice). The tumor volume in MCF-7/Dox was 7-fold(109.4mm3 versus14.4mm3; p� 0.001; five of fivemice) greaterthan MCF-7 group (four of five mice). Silencing of GCS withMBO-asGCS (4 mg/kg for 2 weeks) retarded the tumor growthinMCF-7/Dox, because the tumor volume was reduced to 11%(12.4 mm3 versus 109.4 mm3; p� 0.001). Assessment of BCSCs(CD44�/ESA�/CD24�) showed that BCSCs fromMCF-7/Doxtumors were 4-fold (4.6% versus 1.1% of tumor cells; p � 0.001)larger than MCF-7 tumors. MBO-asGCS treatment decreasedBCSCs to 43% (2.0% versus 4.6% tumor cells; p � 0.001) (Fig.4B). These results indicate BCSCs that highly express GCS arethe cause of tumorigenicity of these cell lines in vivo.Ceramide Glycosylation Determines Stem Properties of

BCSCs in Tumor Formation and Metastasis—To determinewhether or not BCSCs rely on ceramide glycosylation, wesorted BCSCs and other subpopulations fromMCF-7/Dox cellsand examined their tumorigenicity and ceramide glycosylation.Consistent with previous reports (6), it was found that theclonogenicity of BCSCs (CD44�/ESA�/CD24�) was 3-foldgreater (19,540 cells versus 5,717 cells; p � 0.001) than theCD44�/CD24� subset and significantly higher than other non-stem cell subsets (CD44�/CD24�/ESA�, CD44�/CD24�, andCD44�/CD24�) (data not shown here). Conversely, silencingGCS by MBO-asGCS reduced the number of colonies to 45%(8,724 cells versus 19,540 cells; p � 0.001) in BCSCs (Fig. 5A).These results further were confirmed by tumor sphere assay(Fig. 5A, bottom panels). Furthermore, the tumorigenicity ofBCSCs and CD44�/CD24� cells were examined in athymic

nude mice. It was found that BCSCs formed significantly moretumors, and the tumors were bigger than the non-stem cellsubset (CD44�/CD24�) (Table 1 and Fig. 5B). The tumor inci-dent detected in the BCSC group (1 � 105 cells/mouse inocu-lation; 12 of 12 mice) is statistically significant, 6-fold higher(p � 0.0071) than CD44�/CD24� group (one of six mice).Tumors from BCSCs grew aggressively and were 6-fold largerthan the non-stemcell subset (1,858 versus 274mm3;p� 0.001)(Fig. 5B). Moreover, lung metastases were detected in miceinjected with BCSCs (two of 12 mice), but not in mice inocu-latedwithCD44�/CD24� cells, as indicated in the hematoxylinand eosin staining of lung section (Fig. 5B, bottom panels).However, the difference of metastasis incidence was not signif-icant (p � 0.99) between the two groups, because of small sam-ple size.It has been reported lately that CD44�/CD24� subset cells

overexpress the mRNA of ugcg and another 15 genes (49). WeexaminedGCS activities in BCSCs (CD44�/ESA�/CD24�) andother subsets in this study. It was found that GCS enzyme activ-ity in BCSCs was 2-fold higher (33 versus 16 pmol/mg; p �0.001) than in the CD44�/CD24� subset (Fig. 6A) and signifi-cantly higher than other non-stem cell subsets (CD44�/

FIGURE 5. Colony formation and tumorigenesis of BCSCs. BCSC andCD44�/CD24� subsets were sorted from MCF-7/Dox cells. BCSC/asGCS,BCSCs were treated with MBO-asGCS (100 nM, 7 days). A, colony formation ofBCSCs. BCSC and other cell subsets were sorted from MCF-7/Dox cells andplated for colony formation. Bottom panels, tumor spheres were photomicro-graphed (200 � magnification). B, tumor formation of BCSCs. The cells (1 �105 cells/mouse) of BCSC (CD44�/ESA�/CD24�) and of the non-stem cell sub-set (CD44�/CD24�) were inoculated in the mammary gland of athymic nudemice, and the tumor growth was observed for 84 days. *, only one benignxenograft was found (one of 12) in this group. Bottom panels, hematoxylinand eosin (H&E) staining of lung sections.

TABLE 1Breast tumor formation by BCSC and CD44�/CD24� cells in athymicnude mice

Cells injectedTumors formed

in miceLung

metastasis

BCSC100,000cells/mouse

12/12a 2/12

25,000cells/mouse

2/4 0/4

CD44�/CD24�

100,000cells/mouse

1/4 0/4

25,000cells/mouse

0/4 0/4

a The p value of tumor volumes is 0.0071 compared with CD44�/CD24� cells.

FIGURE 4. BCSCs in tumors. A, flow cytometry histograms of CD44 and ESAcells from tumors. CD24� cells were separated from tumors from each groupand analyzed using flow cytometry. The CD44�/ESA� cells (BCSCs) enclosedin the ellipses (upper right) were compared with negative controls (CD44�,ESA�) of the CD24� fraction. B, tumor BCSCs. After magnetic separation(CD24) and flow cytometry analysis (CD44/ESA), BCSCs are presented as thepercentage of total tumor cells. Dox/asGCS, MCF-7/Dox tumors were treatedwith MBO-asGCS (4 mg/kg, intraperitoneal, 14 days). *, p � 0.001 comparedwith MCF-7 tumors; **, p � 0.001, compared with vehicle treatment of MCF-7/Dox tumors.

Ceramide Glycosylation Maintains Cancer Stem Cells

37200 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 44 • OCTOBER 26, 2012

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

CD24�/ESA�, CD44�/CD24�, and CD44�/CD24�) (data notshown here). Silencing GCS with MBO-asGCS reduced theenzyme activity of BCSCs to 30% (33 versus 10 pmol/mg; p �0.001) (Fig. 6A). Profiling of GSLs using HPTLC indicated thatthe levels of GSLs were consistent with the alteration of GCSactivities in BCSCs, non-stem cell subset, and BCSCs treatedwith MBO-asGCS (Fig. 6, A and B). Markedly, Gb3 levels inBCSCs were significantly higher than in non-stem cell subset(CD44�/CD24�), and MBO-asGCS treatment reduced Gb3 to37% (46 versus 17 ng/mg; p � 0.001) in BCSCs (Fig. 6B). These

results were further confirmed by findings using MALDI-MSand MS/MS analysis. As shown in Fig. 6 (C and D), MBO-as-GCS treatment significantly reducedGb3 level in BCSCs to 50%(15816 versus 31430 in absolute height; p � 0.001). Together,these findings indicate enhanced ceramide glycosylation is aunique characteristic of BCSCs, and it plays a crucial role inmaintaining the stemness of BCSCs.c-Src/�-Catenin Signaling Modulated by GCS Is Essential for

BCSCs—It has been reported that Wnt/�-catenin signaling iscritical for the self-renewal of normalmammary epithelial stem

FIGURE 6. GCS activity and GSLs of BCSCs. BCSC and non-stem cell subset were sorted from MCF-7/Dox cells. BCSC/asGCS, BCSCs were treated withMBO-asGCS (100 nM, 7 days). A, fluorescence chromatogram of GCS activity. Cer, NBD C6-ceramide; GlcCer, NBD C6-glucosylceramide; Std, standards of Cer andGlcCer. GCS activity is presented as GlcCer generated per cellular proteins (pmol/mg). *, p � 0.001 compared with non-stem cell subset CD44�/CD24�; **, p �0.001, compared with vehicle control of BCSC. *, p � 0.001 compared with non-stem cell subset, CD44�/CD24�. BCSC/asGCS, BCSCs were treated withMBO-asGCS (100 nM). **, p � 0.001 compared with vehicle control of BCSC. B, GSL profiles. GlcCer, glucosylceramide; LC, lactosylceramide; Std, standards ofneutral glycosphingolipid Qualimix. *, p � 0.001 compared with the CD44�/CD24� subset; **, p � 0.001 compared with vehicle control of BCSC. C, MALDI-MSprofiles of permethylated GSLs of BCSC. D, MALDI-MS profiles of permethylated GSLs of BCSC treated with MBO-asGCS. Lc, lactosylceramide; *, unidentifiedpeaks; Gb3 abs. int., absolute intensity of Gb3.

Ceramide Glycosylation Maintains Cancer Stem Cells

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37201

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

cells andBCSCs (17, 18, 50, 51). Globo-series ofGSLs canmedi-ate Wnt/�-catenin signaling and cell functions (25, 52, 53). Toelucidate how GSLs modulate BCSCs, a stepwise inhibitionapproach was employed to examine the effects of ceramide gly-cosylation on BCSC stemness. As shown in Fig. 7, the proteinlevels of CD44, FGF-2, Oct-4, phosphorylated c-Src, and�-catenin (active form) were substantially increased in BCSCs,as compared with non-stem cells (CD44�/CD24�); suppres-sion of ceramide glycosylation by MBO-asGCS (100 nM) sub-stantially repressed the protein expression of CD44, FGF-2, andOct-4, accompanied by a decrease of phosphorylated c-Src and�-catenin (active form) in BCSCs (CD44�/ESA�/CD24�) (Fig.7, A and B). Consequently, MBO-asGCS decreased BCSCs to30% in MCF-7/Dox cells (Fig. 7C). FH535 can specificallyinhibit �-catenin recruitment of transcription factors and reg-ulation of gene expression (25, 54, 55). We found that FH535treatments (5 �M) repressed the protein expression of CD44,FGF-2, and Oct-4 by �3-fold in BCSCs, and thus decreasedBCSCs to 41% inMCF-7/Dox cells (Fig. 7,A andC). It has beenreported that pyrazolo pyrimidine (PP2) can inhibit c-Src phos-phorylation (25, 56, 57). In this study, PP2 treatments (5 �M)significantly decreased the levels of phosphorylated c-Src and

reduced the level of �-catenin to 7% (0.9 versus 13.0 �-catenin/p�-catenin; p � 0.001) in BCSCs (Fig. 7B). Consistently, PP2treatment decreased BCSCs to 33% in MCF-7/Dox cells (Fig.7C). Together, these findings strongly suggest that ceramideglycosylation activates c-Src and �-catenin signaling pathwayand mediates the expression of CD44, FGF-2, and Oct-4 tomaintain BCSCs.

DISCUSSION

Our results clearly indicate that ceramide glycosylation cat-alyzed by GCS is important for CSCs in drug resistance andtumorigenesis. BCSCs have been identified with surface mark-ers, such as CD44�/ESA�/CD24�, in established breast cancercell lines and in tumor specimens (3, 5, 6, 46). In addition totumorigenesis, the accumulated BCSCs in tumors drive tumorprogression to disseminated metastasis and poor response tochemotherapy (8, 10, 58). Among human breast cancer celllines, the tumorigenesis and drug resistance are reported highlyassociated with the numbers of BCSCs (5, 6, 49). In the presentstudy,MCF-7/Dox cells that have 3-foldmore BCSCs displayedmarked resistance to doxorubicin and were aggressive intumorigenesis (Figs. 1, 2, and 4). DespiteGCSoverexpression inmetastatic breast tumors and causing drug resistance (31, 33),there is no clear experimental evidence linking these to CSCs.Ceramide glycosylation by GCS found in the present study cor-related well with BCSCs, as well as their tumorous property indrug resistance and tumorigenicity. MCF-7/Dox cells overex-pressed GCS protein and produced 2-fold more Gb3 followingenhanced ceramide glycosylation, as compared with MCF-7breast cancer cells. In contrast, the levels of GCS protein andGb3, which were consistent with the BCSC numbers, were sub-stantially lower in normal MCF-12A mammary epithelial cells(Fig. 3). Disruption of ceramide glycosylation by silencing ofGCS decreased BCSCs in cultured cells and in tumors gener-ated from MCF-7/Dox cells (Figs. 2 and 4); consequently,silencing of GCS reversed drug resistance and eliminatedtumorigenesis. These results together indicate that enhancedceramide glycosylation caused by GCS overexpression is essen-tial for BCSCs retaining drug resistance and tumorigenicity.With direct evidence, the present study demonstrated that

globo-series GSLs play a crucial role in maintaining CSCs.Among GSLs synthesized after ceramide glycosylation, globo-pentosylceramide (Gb5, Gal�3GalNAc�Gal�4Glc�1Cer) andmonosialyl globopentosylceramide (MSGb5, Sia�2,3Gal�3GalNAc�Gal�4Glc�1Cer), which are generated from Gb3 andrecognized as stage specific embryonic antigen-3 and -4, arebroadly used as markers to characterize undifferentiatedhuman ESCs including isolated, derived, and induced pluripo-tent stem cells (20–22, 60, 61). Gb5 and other globo-seriesGSLs are synthesized by a serial glycosylation of ceramide cat-alyzed by GCS, lactosylceramide synthase, Gb3 synthase, andGb5 synthase, butGCS is the first rate-limiting enzyme for theirsynthesis in mammalian cells (62, 63). GSLs play importantroles in ESCs, because homozygous knock-out of ugcg inmouseprevents the development and differentiation of the gastrulat-ing embryo bymassive apoptosis largely in the ectodermal layer(27). Deletion of GSLs by using GCS inhibitor (1-phenyl-2-de-canoylamino-3-morpholino-1-propanol), previous study indi-

FIGURE 7. GCS increases c-Src/�-catenin signaling for BCSC maintenance.The BCSC and CD44�/CD24� subsets were sorted from MCF-7/Dox cells.BCSC/asGCS, BCSCs sorted from MCF-7/Dox cells after MBO-asGCS treatment(100 nM for 7 days). A, �-catenin is crucial for GCS up-regulating FGF-2, Oct-4,and CD44 in BCSC. Equal amounts of detergent-soluble proteins (50 �g/lane)were resolved by 4 –20% SDS-PAGE and immunoblotted with anti-CD44,FGF-2, Oct-4, and GAPDH antibodies. BCSC�FH535, BCSCs sorted from MCF-7/Dox cells treated with FH535 (�-catenin inhibitor, 5 �M for 24 h). *, p � 0.001compared with the FGF-2/GAPDH ratio of CD44�/CD24� subset; **, p � 0.001compared with the FGF-2/GAPDH ratio of BCSC. B, c-Src phosphorylationmediates GCS effect on enhancing �-catenin in BCSCs. Equal amounts ofdetergent-soluble proteins (50 �g/lane) were resolved by 4 –20% SDS-PAGEand immunoblotted with anti-GCS, c-Src, p-c-Src, p-�-catenin, �-catenin, andGAPDH antibodies. BCSC�PP2, BCSCs sorted from MCF-7/Dox cells treatedwith PP2 (c-Src kinase inhibitor, 5 �M for 24 h); p-cSrc, phosphorylated c-Src;p�-catenin, phosphorylated �-catenin. *, p � 0.001 compared with the�-catenin/GAPDH ratio of CD44�/CD24� subset; **, p � 0.001 compared withthe �-catenin/GAPDH ratio of BCSCs. C, effects of inhibitions of GCS, c-Srckinases, and �-catenin on BCSCs. BCSCs were analyzed following MCF-7/Doxcells treated with MBO-asGCS (100 nM, 7 days; Dox/asGCS), PP2 (5 �M, 24 h;Dox/PP2), and FH535 (5 �M, 24 h; Dox/FH535). *, p � 0.001 compared withvehicle control of MCF-7/Dox cells.

Ceramide Glycosylation Maintains Cancer Stem Cells

37202 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 44 • OCTOBER 26, 2012

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

cates that Gb5 and MSGb5 do not play critical functional rolesin maintaining the pluripotency of human ESCs, even thoughtheir expression clearly marks undifferentiated ESCs (64). Gb5and Glob H (Fuc�1–2Gal�1–3GalNac�1–3Gal�1–4Gal�1–4Glc�1) generated from Gb5 by fucosyltransferase are overex-pressed on breast cancers and other epithelial cell tumors, suchas colon, ovarian, gastric pancreatic, lung, and prostate cancers(28, 65, 66).However,Gb5 andGloboH,which are examined byflow cytometry analysis are expressedwith no significant differ-ence between BCSCs (CD45�/CD24�/CD44�) and non-BC-SCs (CD45�/CD24�/CD44� and CD45�/CD24�/CD44�)sorted from tumor specimens of breast cancer patients (66). Toclarify the role ofGSLs inCSCs,we employeddirect approachesto assess GCS protein, ceramide glycosylation, Gb3 by massspectrometry, BCSC numbers, and their tumorous propertiesof BCSCs at the present of GCS silencing. Non-stem cell subset(CD44�/CD24�) presented much lower tumorigenicity andGCS activity than in any other populations (Fig. 5). BCSCs(CD44�/ESA�/CD24�) overexpressed GCS and producedmore Gb3. Importantly, disruption of ceramide glycosylationby silencing of GCS inhibited Gb3 in BCSCs (CD44�/ESA�/CD24�), and consequently it eradicated BCSCs, displaying intumor formation and tumor progression (Figs. 5 and 6).Ceramide glycosylation by GCS is a unique process by which

CSCs maintain their tumorous stemness. Ceramide glycosyla-tion, which is known for reducing cellular ceramide biochemi-cally and protecting cell from ceramide-induced apoptosis, canendow CSC resistance to cytotoxins and anticancer drugs. Ithas been reported that CSCs (CD55hi) have high tolerance toceramide-induced apoptosis, and this resistance is attributed toabundance of sphingomyelin synthase 1 that can converts cer-amide to sphingomyelin (67). Because GCS level is higher inBCSCs versus lower in bone marrow stem cells, doxorubicincan enhance BCSC number and decrease the number of bonemarrow stem cells (68). Numerous studies show that GSLs,such asGb5 andMSGb5, aremarkers of stem cells; however, nocutting evidence links their expression to a role in pluripotency,and their molecular functions remain unclear. Gb5 andMSGb5, as markers of human ESCs, disappear with differenti-ation (64, 69). Mass spectrometry analysis that can distinguishglycans shows a switching of the core structures of glycosphin-golipids from globo-series and lacto-series to ganglio-serieswhen human ESCs differentiate into embryoid body outgrowthwith three germ layers (69). Recent study shows that ganglio-sideGD2 can be used as amarker to identifyCSCs (GD2�) frombreast cancer cell lines and patient samples, and interferencewith GD3 synthase can reduce CSC population and CSC-asso-ciated properties (70). The present study found GCS and Gb3overproduced in BCSCs, not normal adult stem cells, such asmammary epithelial stem cells (Fig. 3) and bone marrow stemcells reported by another study (68). It is not clear by whichmolecularmechanismCSCs regain ability in ceramide glycosyl-ation that exists in ESCs. We now know that ceramide gener-ated in cells exposed to doxorubicin can activateGCS promoterand up-regulate its expression (38).Amajor contribution of the presentwork is thatwe identified

the role of Gb3 inmaintaining stem properties of BCSCs. It hasbeen reported that Gb3 (CD77), a precursor of Gb5 is associ-

ated with Src family Yes kinase on GSL-enriched microdo-mains (25, 71). Gb3 canmodulate c-Src kinase inGSL-enrichedmicrodomains and up-regulate mdr1 expression through�-catenin signaling (25). Wnt/�-catenin signaling maintainshematopoietic stem cells and neuronal stem cells and the epi-thelial to mesenchymal transition for BCSCs (72); Wnt inhibi-tors can reduce the size of tumor spheres and self-renewal inprostate cancer stem cells (73). FGF-2 is critical for the self-renewal and maintenance of normal and cancer stem cells (47,48), and Oct-4 is a transcription factor that is involved inmain-taining the pluripotency of ESCs (74) and BCSCs (59, 75). Thisstudy has found that the levels of GCS, Gb3, phosphorylatedc-Src, �-catenin, FGF-2, and Oct-4 are substantially higher inBCSCs (Fig. 7). Stepwise inhibition of GCS, c-Src kinases, and�-catenin indicates that these molecules are critical for themaintenance of BCSCs (Fig. 7). Although stem cell factorsincluding �-catenin, FGF-2, Oct-4, and CD44 are involved inregulating stemness of either normal stem cells or CSCs, thisstudy demonstrates that GSLs in BCSCs activate c-Src/�-catenin signaling and up-regulate these factors to maintainBCSCs selectively. In addition to insight regulating stemnessof stem cells, this finding opens the possibility of disruptingceramide glycosylation or GSL synthesis to target CSCsspecifically.

Acknowledgments—We thank Dr. Robert K. Yu (Institute of Molecu-lar Medicine and Genetics and Institute of Neuroscience, MedicalCollege of Georgia) andDr. BrianRowan (TulaneUniversity School ofMedicine, New Orleans, LA) for critical comments, Dr. Kapil Mehta(M. D. Anderson Cancer Center, Houston, TX) for cell lines of MCF-7and MCF-7/Dox.

REFERENCES1. Carney, D. N., Gazdar, A. F., Bunn, P. A., Jr., and Guccion, J. G. (1982)

Demonstration of the stem cell nature of clonogenic tumor cells from lungcancer patients. Stem Cells 1, 149–164

2. Reya, T., Morrison, S. J., Clarke, M. F., and Weissman, I. L. (2001) Stemcells, cancer, and cancer stem cells. Nature 414, 105–111

3. Al-Hajj, M., Wicha, M. S., Benito-Hernandez, A., Morrison, S. J., andClarke, M. F. (2003) Prospective identification of tumorigenic breast can-cer cells. Proc. Natl. Acad. Sci. U.S.A. 100, 3983–3988

4. Lee, C. J., Dosch, J., and Simeone, D. M. (2008) Pancreatic cancer stemcells. J. Clin. Oncol. 26, 2806–2812

5. Harrison, H., Farnie, G., Howell, S. J., Rock, R. E., Stylianou, S., Brennan,K. R., Bundred, N. J., and Clarke, R. B. (2010) Regulation of breast cancerstem cell activity by signaling through the Notch4 receptor. Cancer Res.70, 709–718

6. Fillmore, C. M., and Kuperwasser, C. (2008) Human breast cancer celllines contain stem-like cells that self-renew, give rise to phenotypicallydiverse progeny and survive chemotherapy. Breast Cancer Res. 10, R25

7. Phillips, T. M., McBride, W. H., and Pajonk, F. (2006) The response ofCD24�/low/CD44� breast cancer-initiating cells to radiation. J. Natl. Can-cer Inst. 98, 1777–1785

8. Tanei, T., Morimoto, K., Shimazu, K., Kim, S. J., Tanji, Y., Taguchi, T.,Tamaki, Y., and Noguchi, S. (2009) Association of breast cancer stem cellsidentified by aldehyde dehydrogenase 1 expression with resistance to se-quential Paclitaxel and epirubicin-based chemotherapy for breast cancers.Clin. Cancer Res. 15, 4234–4241

9. Balic, M., Lin, H., Young, L., Hawes, D., Giuliano, A., McNamara, G.,Datar, R. H., and Cote, R. J. (2006) Most early disseminated cancer cellsdetected in bone marrow of breast cancer patients have a putative breastcancer stem cell phenotype. Clin. Cancer Res. 12, 5615–5621

Ceramide Glycosylation Maintains Cancer Stem Cells

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37203

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

10. Lee, H. E., Kim, J. H., Kim, Y. J., Choi, S. Y., Kim, S. W., Kang, E., Chung,I. Y., Kim, I. A., Kim, E. J., Choi, Y., Ryu, H. S., and Park, S. Y. (2011) Anincrease in cancer stem cell population after primary systemic therapy is apoor prognostic factor in breast cancer. Br. J. Cancer 104, 1730–1738

11. Giatromanolaki, A., Sivridis, E., Fiska, A., and Koukourakis, M. I. (2011)The CD44�/CD24� phenotype relates to ’triple-negative’ state and unfa-vorable prognosis in breast cancer patients.Med. Oncol. 28, 745–752

12. Korkaya, H., and Wicha, M. S. (2009) HER-2, notch, and breast cancerstem cells. Targeting an axis of evil. Clin. Cancer Res. 15, 1845–1847

13. Tang, C., Ang, B. T., and Pervaiz, S. (2007) Cancer stem cell. Target foranti-cancer therapy. FASEB J. 21, 3777–3785

14. Frank, N. Y., Schatton, T., and Frank,M. H. (2010) The therapeutic prom-ise of the cancer stem cell concept. J. Clin. Invest. 120, 41–50

15. Gupta, V., Zhang, Q. J., and Liu, Y. Y. (2011) Evaluation of anticanceragents using flow cytometry analysis of cancer stem cells. Methods Mol.Biol. 716, 179–191

16. Takebe, N., Harris, P. J., Warren, R. Q., and Ivy, S. P. (2011) Targetingcancer stem cells by inhibitingWnt, Notch, andHedgehog pathways.Nat.Rev. Clin. Oncol. 8, 97–106

17. Korkaya, H., Paulson, A., Charafe-Jauffret, E., Ginestier, C., Brown, M.,Dutcher, J., Clouthier, S. G., andWicha, M. S. (2009) Regulation of mam-mary stem/progenitor cells by PTEN/Akt/�-catenin signaling. PLoS Biol.7, e1000121

18. Liu, B. Y., McDermott, S. P., Khwaja, S. S., and Alexander, C. M. (2004)The transforming activity of Wnt effectors correlates with their ability toinduce the accumulation of mammary progenitor cells. Proc. Natl. Acad.Sci. U.S.A. 101, 4158–4163

19. Sansone, P., Storci, G., Giovannini, C., Pandolfi, S., Pianetti, S., Taffurelli,M., Santini, D., Ceccarelli, C., Chieco, P., and Bonafé, M. (2007) p66Shc/Notch-3 interplay controls self-renewal and hypoxia survival in humanstem/progenitor cells of the mammary gland expanded in vitro as mam-mospheres. Stem Cells 25, 807–815

20. Kannagi, R., Cochran, N. A., Ishigami, F., Hakomori, S., Andrews, P. W.,Knowles, B. B., and Solter, D. (1983) Stage-specific embryonic antigens(SSEA-3 and -4) are epitopes of a unique globo-series ganglioside isolatedfrom human teratocarcinoma cells. EMBO J. 2, 2355–2361

21. Klimanskaya, I., Chung, Y., Becker, S., Lu, S. J., and Lanza, R. (2006) Hu-man embryonic stem cell lines derived from single blastomeres. Nature444, 481–485

22. Yu, J., Vodyanik, M. A., Smuga-Otto, K., Antosiewicz-Bourget, J., Frane,J. L., Tian, S., Nie, J., Jonsdottir, G. A., Ruotti, V., Stewart, R., Slukvin, I. I.,and Thomson, J. A. (2007) Induced pluripotent stem cell lines derivedfrom human somatic cells. Science 318, 1917–1920

23. Hakomori, S. I. (2010) Glycosynaptic microdomains controlling tumorcell phenotype through alteration of cell growth, adhesion, and motility.FEBS Lett. 584, 1901–1906

24. Ngamukote, S., Yanagisawa, M., Ariga, T., Ando, S., and Yu, R. K. (2007)Developmental changes of glycosphingolipids and expression of glyco-genes in mouse brains. J. Neurochem. 103, 2327–2341

25. Liu, Y. Y., Gupta, V., Patwardhan, G. A., Bhinge, K., Zhao, Y., Bao, J.,Mehendale, H., Cabot, M. C., Li, Y. T., and Jazwinski, S. M. (2010) Gluco-sylceramide synthase upregulates MDR1 expression in the regulation ofcancer drug resistance through c-Src and beta-catenin signaling. Mol.Cancer 9, 145

26. Patwardhan, G. A., and Liu, Y. Y. (2011) Sphingolipids and expressionregulation of genes in cancer. Prog Lipid Res. 50, 104–114

27. Yamashita, T., Wada, R., Sasaki, T., Deng, C., Bierfreund, U., Sandhoff, K.,and Proia, R. L. (1999) A vital role for glycosphingolipid synthesis duringdevelopment and differentiation. Proc. Natl. Acad. Sci. U.S.A. 96,9142–9147

28. Hakomori, S. I. (2008) Structure and function of glycosphingolipids andsphingolipids. Recollections and future trends. Biochim. Biophys. Acta1780, 325–346

29. Basu, S., Kaufman, B., and Roseman, S. (1968) Enzymatic synthesis ofceramide-glucose and ceramide-lactose by glycosyltransferases from em-bryonic chicken brain. J. Biol. Chem. 243, 5802–5804

30. Liu, Y. Y., Han, T. Y., Giuliano, A. E., and Cabot, M. C. (1999) Expressionof glucosylceramide synthase, converting ceramide to glucosylceramide,

confers adriamycin resistance in human breast cancer cells. J. Biol. Chem.274, 1140–1146

31. Liu, Y. Y., Han, T. Y., Giuliano, A. E., and Cabot, M. C. (2001) Ceramideglycosylation potentiates cellular multidrug resistance. FASEB J. 15,719–730

32. Liu, Y. Y., Patwardhan, G. A., Xie, P., Gu, X., Giuliano, A. E., and Cabot,M. C. (2011) Glucosylceramide synthase, a factor in modulating drug re-sistance, is overexpressed inmetastatic breast carcinoma. Int. J. Oncol. 39,425–431

33. Juul, N., Szallasi, Z., Eklund, A. C., Li, Q., Burrell, R. A., Gerlinger, M.,Valero, V., Andreopoulou, E., Esteva, F. J., Symmans, W. F., Desmedt, C.,Haibe-Kains, B., Sotiriou, C., Pusztai, L., and Swanton, C. (2010) Assess-ment of an RNA interference screen-derived mitotic and ceramide path-way metagene as a predictor of response to neoadjuvant paclitaxel forprimary triple-negative breast cancer. A retrospective analysis of five clin-ical trials. Lancet Oncol. 11, 358–365

34. Inokuchi, J., Jimbo, M., Momosaki, K., Shimeno, H., Nagamatsu, A., andRadin, N. S. (1990) Inhibition of experimental metastasis of murine Lewislung carcinoma by an inhibitor of glucosylceramide synthase and its pos-sible mechanism of action. Cancer Res. 50, 6731–6737

35. Mehta, K. (1994) High levels of transglutaminase expression in doxorubi-cin-resistant human breast carcinoma cells. Int. J. Cancer 58, 400–406

36. Patwardhan, G. A., Zhang, Q. J., Yin, D., Gupta, V., Bao, J., Senkal, C. E.,Ogretmen, B., Cabot, M. C., Shah, G. V., Sylvester, P. W., Jazwinski, S. M.,and Liu, Y. Y. (2009) A new mixed-backbone oligonucleotide against glu-cosylceramide synthase sensitizes multidrug-resistant tumors to apopto-sis. PLoS One 4, e6938

37. Liu, Y. Y., Han, T. Y., Yu, J. Y., Bitterman, A., Le, A., Giuliano, A. E., andCabot,M. C. (2004)Oligonucleotides blocking glucosylceramide synthaseexpression selectively reverse drug resistance in cancer cells. J. Lipid Res.45, 933–940

38. Liu, Y. Y., Yu, J. Y., Yin, D., Patwardhan, G. A., Gupta, V., Hirabayashi, Y.,Holleran, W. M., Giuliano, A. E., Jazwinski, S. M., Gouaze-Andersson, V.,Consoli, D. P., and Cabot, M. C. (2008) A role for ceramide in drivingcancer cell resistance to doxorubicin. FASEB J. 22, 2541–2551

39. Shiota, M., Heike, T., Haruyama, M., Baba, S., Tsuchiya, A., Fujino, H.,Kobayashi, H., Kato, T., Umeda, K., Yoshimoto, M., and Nakahata, T.(2007) Isolation and characterization of bonemarrow-derivedmesenchy-mal progenitor cells withmyogenic and neuronal properties.Exp. Cell Res.313, 1008–1023

40. Giarratana, M. C., Kobari, L., Lapillonne, H., Chalmers, D., Kiger, L.,Cynober, T., Marden, M. C., Wajcman, H., and Douay, L. (2005) Ex vivogeneration of fully mature human red blood cells from hematopoieticstem cells. Nat. Biotechnol. 23, 69–74

41. Gupta, V., Patwardhan, G. A., Zhang, Q. J., Cabot, M. C., Jazwinski, S. M.,and Liu, Y. Y. (2010) Direct quantitative determination of ceramide gly-cosylation in vivo. A new approach to evaluate cellular enzyme activity ofglucosylceramide synthase. J. Lipid Res. 51, 866–874

42. Liu, Y. Y., Patwardhan, G. A., Bhinge, K., Gupta, V., Gu, X., and Jazwinski,S. M. (2011) Suppression of glucosylceramide synthase restores p53-de-pendent apoptosis in mutant p53 cancer cells.Cancer Res. 71, 2276–2285

43. Lin, C. H., Fan, Y. Y., Chen, Y. Y., Wang, S. H., Chen, C. I., Yu, L. C., andKhoo, K. H. (2009) Enhanced expression of �3-galactosyltransferase 5activity is sufficient to induce in vivo synthesis of extended type 1 chainson lactosylceramides of selected human colonic carcinoma cell lines.Gly-cobiology 19, 418–427

44. Zhou, B., Li, S. C., Laine, R. A., Huang, R. T., and Li, Y. T. (1989) Isolationand characterization of ceramide glycanase from the leech, Macrobdelladecora. J. Biol. Chem. 264, 12272–12277

45. Dell, A., Reason, A. J., Khoo, K. H., Panico, M., McDowell, R. A., andMorris, H. R. (1994) Mass spectrometry of carbohydrate-containingbiopolymers.Methods Enzymol. 230, 108–132

46. Pandey, P. R., Okuda, H., Watabe, M., Pai, S. K., Liu, W., Kobayashi, A.,Xing, F., Fukuda, K., Hirota, S., Sugai, T., Wakabayashi, G., Koeda, K.,Kashiwaba, M., Suzuki, K., Chiba, T., Endo, M., Fujioka, T., Tanji, S., Mo,Y. Y., Cao, D.,Wilber, A. C., andWatabe, K. (2011) Resveratrol suppressesgrowth of cancer stem-like cells by inhibiting fatty acid synthase. BreastCancer Res. Treat. 130, 387–398

Ceramide Glycosylation Maintains Cancer Stem Cells

37204 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 287 • NUMBER 44 • OCTOBER 26, 2012

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from

47. Greber, B., Lehrach, H., and Adjaye, J. (2007) Fibroblast growth factor 2modulates transforming growth factor beta signaling inmouse embryonicfibroblasts and human ESCs (hESCs) to support hESC self-renewal. StemCells 25, 455–464

48. Kondo, T., Setoguchi, T., and Taga, T. (2004) Persistence of a small sub-population of cancer stem-like cells in the C6 glioma cell line. Proc. Natl.Acad. Sci. U.S.A. 101, 781–786

49. Calcagno,A.M., Salcido, C.D., Gillet, J. P.,Wu,C. P., Fostel, J.M.,Mumau,M. D., Gottesman, M. M., Varticovski, L., and Ambudkar, S. V. (2010)Prolonged drug selection of breast cancer cells and enrichment of cancerstem cell characteristics. J. Natl Cancer Inst. 102, 1637–1652

50. Li, Y., Welm, B., Podsypanina, K., Huang, S., Chamorro, M., Zhang, X.,Rowlands, T., Egeblad, M., Cowin, P., Werb, Z., Tan, L. K., Rosen, J. M.,and Varmus, H. E. (2003) Evidence that transgenes encoding componentsof the Wnt signaling pathway preferentially induce mammary cancersfrom progenitor cells. Proc. Natl. Acad. Sci. U.S.A. 100, 15853–15858

51. Woodward,W. A., Chen, M. S., Behbod, F., Alfaro, M. P., Buchholz, T. A.,and Rosen, J. M. (2007) WNT/�-catenin mediates radiation resistance ofmouse mammary progenitor cells. Proc. Natl. Acad. Sci. U.S.A. 104,618–623

52. Booker-Dwyer, T., Hirsh, S., and Zhao, H. (2008) A unique cell populationin themouse olfactory bulb displays nuclear beta-catenin signaling duringdevelopment and olfactory sensory neuron regeneration. Dev. Neurobiol.68, 859–869

53. Steelant, W. F., Kawakami, Y., Ito, A., Handa, K., Bruyneel, E. A., Mareel,M., and Hakomori, S. (2002) Monosialyl-Gb5 organized with c-Src andFAK in GEM of human breast carcinoma MCF-7 cells defines their inva-sive properties. FEBS Lett. 531, 93–98

54. Handeli, S., and Simon, J. A. (2008) A small-molecule inhibitor of Tcf/�-catenin signaling down-regulates PPAR� and PPAR� activities.Mol. Can-cer Ther. 7, 521–529

55. Lim, S. K., Orhant-Prioux, M., Toy, W., Tan, K. Y., and Lim, Y. P. (2011)Tyrosine phosphorylation of transcriptional coactivator WW-domainbinding protein 2 regulates estrogen receptor � function in breast cancervia the Wnt pathway. FASEB J. 25, 3004–3018

56. Maupas-Schwalm, F., Robinet, C., Augé, N., Thiers, J. C., Garcia, V., Cam-bus, J. P., Salvayre, R., and Nègre-Salvayre, A. (2005) Activation of the�-catenin/T-cell-specific transcription factor/lymphoid enhancer fac-tor-1 pathway by plasminogen activators in ECV304 carcinoma cells.Cancer Res. 65, 526–532

57. Yokoyama, N., and Malbon, C. C. (2009) Dishevelled-2 docks and acti-vates Src in a Wnt-dependent manner. J. Cell Sci. 122, 4439–4451

58. Ginestier, C., Hur, M. H., Charafe-Jauffret, E., Monville, F., Dutcher, J.,Brown,M., Jacquemier, J., Viens, P., Kleer, C. G., Liu, S., Schott, A., Hayes,D., Birnbaum, D., Wicha, M. S., and Dontu, G. (2007) ALDH1 is a markerof normal and malignant human mammary stem cells and a predictor ofpoor clinical outcome. Cell Stem Cell 1, 555–567

59. Bussolati, B., Grange, C., Sapino, A., and Camussi, G. (2009) Endothelialcell differentiation of human breast tumor stem/progenitor cells. J. CellMol. Med. 13, 309–319

60. Thomson, J. A., Kalishman, J., Golos, T. G., Durning, M., Harris, C. P.,Becker, R. A., andHearn, J. P. (1995) Isolation of a primate embryonic stemcell line. Proc. Natl. Acad. Sci. U.S.A. 92, 7844–7848

61. Suila, H., Pitkänen, V., Hirvonen, T., Heiskanen, A., Anderson, H., Laiti-nen, A., Natunen, S., Miller-Podraza, H., Satomaa, T., Natunen, J., Laiti-nen, S., and Valmu, L. (2011) Are globoseries glycosphingolipids SSEA-3and -4markers for stem cells derived from human umbilical cord blood? J.

Mol. Cell. Biol. 3, 99–10762. Radin, N. S. (1994) Rationales for cancer chemotherapy with PDMP, a

specific inhibitor of glucosylceramide synthase.Mol. Chem. Neuropathol.21, 111–127

63. Merrill, A. H., Jr. (2011) Sphingolipid and glycosphingolipid metabolicpathways in the era of sphingolipidomics. Chem. Rev. 111, 6387–6422

64. Brimble, S. N., Sherrer, E. S., Uhl, E. W., Wang, E., Kelly, S., Merrill, A. H.,Jr., Robins, A. J., and Schulz, T. C. (2007) The cell surface glycosphingo-lipids SSEA-3 and SSEA-4 are not essential for human ESC pluripotency.Stem Cells 25, 54–62

65. Dube, D. H., and Bertozzi, C. R. (2005) Glycans in cancer and inflamma-tion. Potential for therapeutics and diagnostics. Nat. Rev. Drug Discov. 4,477–488

66. Chang, W. W., Lee, C. H., Lee, P., Lin, J., Hsu, C. W., Hung, J. T., Lin, J. J.,Yu, J. C., Shao, L. E., Yu, J.,Wong, C. H., and Yu, A. L. (2008) Expression ofGlobo H and SSEA3 in breast cancer stem cells and the involvement offucosyl transferases 1 and 2 in Globo H synthesis. Proc. Natl. Acad. Sci.U.S.A. 105, 11667–11672

67. Xu, J. X., Morii, E., Liu, Y., Nakamichi, N., Ikeda, J., Kimura, H., and Ao-zasa, K. (2007) High tolerance to apoptotic stimuli induced by serumdepletion and ceramide in side-population cells. High expression of CD55as a novel character for side-population. Exp. Cell Res. 313, 1877–1885

68. Bhinge, K. N., Gupta, V., Hosain, S. B., Satyanarayanajois, S. D., Meyer,S. A., Blaylock, B., Zhang,Q. J., and Liu, Y. Y. (2012) The opposite effects ofdoxorubicin on bone marrow stem cells versus breast cancer stem cellsdepend on glucosylceramide synthase. Int. J. Biochem. Cell Biol. 44,1770–1778

69. Liang, Y. J., Kuo,H.H., Lin, C.H., Chen, Y. Y., Yang, B. C., Cheng, Y. Y., Yu,A. L., Khoo, K. H., and Yu, J. (2010) Switching of the core structures ofglycosphingolipids from globo- and lacto- to ganglio-series upon humanembryonic stem cell differentiation. Proc. Natl. Acad. Sci. U.S.A. 107,22564–22569

70. Battula, V. L., Shi, Y., Evans, K. W., Wang, R. Y., Spaeth, E. L., Jacamo,R. O., Guerra, R., Sahin, A. A., Marini, F. C., Hortobagyi, G., Mani, S. A.,and Andreeff, M. (2012) Ganglioside GD2 identifies breast cancer stemcells and promotes tumorigenesis. J. Clin. Invest. 122, 2066–2078

71. Katagiri, Y. U., Mori, T., Nakajima, H., Katagiri, C., Taguchi, T., Takeda,T., Kiyokawa, N., and Fujimoto, J. (1999) Activation of Src family kinaseyes induced by Shiga toxin binding to globotriaosyl ceramide (Gb3/CD77)in low density, detergent-insoluble microdomains. J. Biol. Chem. 274,35278–35282

72. Yook, J. I., Li, X. Y., Ota, I., Hu, C., Kim, H. S., Kim, N. H., Cha, S. Y., Ryu,J. K., Choi, Y. J., Kim, J., Fearon, E. R., andWeiss, S. J. (2006) AWnt-Axin2-GSK3� cascade regulates Snail1 activity in breast cancer cells. Nat. CellBiol. 8, 1398–1406

73. Bisson, I., and Prowse, D.M. (2009)WNT signaling regulates self-renewaland differentiation of prostate cancer cells with stem cell characteristics.Cell Res. 19, 683–697

74. Guo, Y., Mantel, C., Hromas, R. A., and Broxmeyer, H. E. (2008) Oct-4 iscritical for survival/antiapoptosis of murine embryonic stem cells sub-jected to stress. Effects associated with Stat3/survivin. Stem Cells 26,30–34

75. Ponti, D., Costa, A., Zaffaroni, N., Pratesi, G., Petrangolini, G., Coradini,D., Pilotti, S., Pierotti, M. A., and Daidone, M. G. (2005) Isolation and invitro propagation of tumorigenic breast cancer cells with stem/progenitorcell properties. Cancer Res. 65, 5506–5511

Ceramide Glycosylation Maintains Cancer Stem Cells

OCTOBER 26, 2012 • VOLUME 287 • NUMBER 44 JOURNAL OF BIOLOGICAL CHEMISTRY 37205

at Tulane U

niv Serials Library, on M

arch 28, 2013w

ww

.jbc.orgD

ownloaded from