Sphingosine Kinase 1 Induces Tolerance to Human Epidermal Growth Factor Receptor 2 and Prevents Formation of a Migratory Phenotype in Response to Sphingosine 1-Phosphate in Estrogen

MOLECULAR AND CELLULAR BIOLOGY, Aug. 2010, p. 3827–3841 Vol. 30, No. 150270-7306/10/$12.00 doi:10.1128/MCB.01133-09Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Sphingosine Kinase 1 Induces Tolerance to Human Epidermal GrowthFactor Receptor 2 and Prevents Formation of a Migratory Phenotype

in Response to Sphingosine 1-Phosphate in EstrogenReceptor-Positive Breast Cancer Cells�

Jaclyn S. Long,1 Joanne Edwards,2 Carol Watson,2 Sian Tovey,2 Kirsty M. Mair,1 Rachel Schiff,3Viswanathan Natarajan,4 Nigel J. Pyne,1* and Susan Pyne1*

Cell Biology Group, Strathclyde Institute of Pharmacy and Biomedical Sciences, University of Strathclyde, 27 Taylor St., Glasgow G4 0NR,United Kingdom1; Section of Surgery, Division of Cancer Sciences and Molecular Pathology, Faculty of Medicine,

Glasgow Royal Infirmary, University of Glasgow, Glasgow G31 2ER, United Kingdom2; Breast Center, MS:BCM 600,Baylor College of Medicine, 1 Baylor Plaza, Houston, Texas 770303; and Department of Pharmacology,

University of Illinois at Chicago, South Wolcott Ave., Chicago, Illinois 606124

Received 20 August 2009/Returned for modification 5 November 2009/Accepted 13 May 2010

We demonstrate here a new concept termed “oncogene tolerance” whereby human EGF receptor 2 (HER2)increases sphingosine kinase 1 (SK1) expression in estrogen receptor-positive (ER�) MCF-7 HER2 cells andSK1, in turn, limits HER2 expression in a negative-feedback manner. The HER2-dependent increase in SK1expression also limits p21-activated protein kinase 1 (p65 PAK1) and extracellular signal regulated kinase 1/2(ERK-1/2) signaling. Sphingosine 1-phosphate signaling via S1P3 is also altered in MCF-7 HER2 cells. In thisregard, S1P binding to S1P3 induces a migratory phenotype via an SK1-dependent mechanism in ER� MCF-7Neo cells, which lack HER2. This involves the S1P stimulated accumulation of phosphorylated ERK-1/2 andactin into membrane ruffles/lamellipodia and migration. In contrast, S1P failed to promote redistribution ofphosphorylated ERK-1/2 and actin into membrane ruffles/lamellipodia or migration of MCF-7 HER2 cells.However, a migratory phenotype in these cells could be induced in response to S1P when SK1 expression hadbeen knocked down with a specific siRNA or when recombinant PAK1 was ectopically overexpressed. Thus, theHER2-dependent increase in SK1 expression functions to desensitize the S1P-induced formation of a migratoryphenotype. This is correlated with improved prognosis in patients who have a low HER1-3/SK1 expression ratioin their ER� breast cancer tumors compared to patients that have a high HER1-3/SK1 expression ratio.

“Oncogene addiction” is a term that has been used to de-scribe the reliance of cancer cells on the continued expressionof oncogenes in order to maintain the diseased phenotype,progression, and metastasis of the cancer cell (39). Oncogeneaddiction is intrinsically susceptible to cross talk and feedbackregulation that reflects abnormal signaling wiring in the cancercell and which potentially makes these cells more susceptibleto drug intervention at the level of the oncogene than normalcells. HER2 is a well-established oncogene that has an impor-tant role in enhancing breast cancer progression (2). The im-portance of its functional role as an addictive oncogene isexemplified by the fact that targeting HER2 with antibodymediated therapies, such as herceptin, demonstrates significantclinical efficacy (40). Indeed, this approach is in line with theconcept that oncogene addiction is the “Achilles heel” of thecancer cell.

There are four members of the human epidermal growthfactor (EGF) receptor-related family, termed HER1 to HER4.The HER2/neu (c-erbB-2) gene encodes a 185-kDa transmem-

brane receptor tyrosine kinase, which is similar in amino acidsequence to other members of the EGF receptor family (30).Moreover, the overexpression of HER2/neu and gene ampli-fication is found in up to 30% of primary breast cancers, and itsexpression is correlated with increased tumor invasion, poorprognosis, and therapeutic resistance (2). Overexpression ofHER2 is also associated with downregulation of the estrogenreceptor (ER) but not necessarily the complete elimination ofthis receptor. A soluble ligand for HER2 has not been identified,although biochemical evidence suggests that HER2 operates as ashared receptor subunit of other ErbBs. In this regard, HER2 isthe preferred and potent heterodimerization partner of HER1,also known as the EGF receptor (13). HER2 delays EGF disso-ciation from its receptor, enhancing coupling of EGF receptor tothe mitogen-activated protein kinase (MAPK) pathways and im-peding the rate of EGF receptor downregulation. Thus, HER2is a master regulator of a signaling network that drives epithe-lial cell proliferation. For instance, the ectopic expression ofHER2/neu in MCF7 (ER�/HER2�) cells activates the phos-phatidylinositol 3-kinase (PI3K)/Akt pathway, which results indownregulation of p53 (45), which is purported to increase thesurvival of these cells. Indeed, inhibition of PI3K essentiallyreverses resistance to gamma irradiation-induced cell death,suggesting a key role for the HER2/PI3K/p53 signaling modulein regulating cell survival and promoting the proliferation ofthese cells (45).

* Corresponding author. Mailing address: Cell Biology Group, Strath-clyde Institute of Pharmacy and Biomedical Sciences, University ofStrathclyde, 27 Taylor St., Glasgow G4 0NR, United Kingdom. Phone forSusan Pyne: 44-141 548 2012. Fax: 44-141 552 2562. E-mail: [email protected]. Phone for Nigel J. Pyne: 44-141 548 2659. Fax: 44-141 5522562. E-mail: [email protected].

� Published ahead of print on 1 June 2010.

3827

Sphingosine 1-phosphate (S1P) is a bioactive lipid that hasbeen implicated as having an important role in regulating thegrowth, survival, and migration of mammalian cells. S1Pbinds to a family of five G-protein-coupled receptors termedS1Pn (where n � 1 to 5), that are differentially coupled toheterotrimeric G-proteins to regulate various effectors, suchas MAPKs, linked to diverse cellular processes (6). S1P isproduced by sphingosine kinase (two isoforms termed SK1 andSK2) that catalyze the phosphorylation of sphingosine to pro-duce S1P (28). SK1 is regulated by phosphorylation, catalyzedby extracellular signal regulated kinase 1/2 (ERK-1/2) (27) andhas been demonstrated to increase Ras-dependent transfor-mation of cancer cells (41), possibly via ERK-1/2-dependenttargeting of SK1 to the plasma membrane of cells, and increasethereof of its so-called oncogenic function. There are severalreports demonstrating that SK1 has an important role in breastcancer cells (12, 24, 29). For instance, ectopic expression ofSK1 increased S1P levels, estrogen-dependent tumorigenesis,and blocked apoptosis of MCF-7 cells induced by anticancerdrugs, sphingosine, and tumor necrosis factor alpha (24). SK1is also required for EGF-induced MCF-7 migration, prolifer-ation, and cell survival (29), and S1P stimulates breast cancercell growth through activation of the serum response elementand indirectly by enhancing IGF-II synthesis and function (12).There is no evidence that mutations occur in the SK1 genelinked to cancer, and therefore in strict terms SK1 cannot bedefined as an oncogene. However, the term “nononcogeneaddiction” has been used to describe the role of SK1 in cancerprogression (35).

We describe here a novel pathway in which SK1 inducestolerance to HER2 by modulating the expression of HER2 andattendant signaling components such as PAK1 that are nor-mally required to produce a migratory phenotype in responseto S1P. Moreover, this is correlated with improved prognosis inpatients who have a low HER1-3/SK1 expression ratio in theirER� breast cancer tumors.

MATERIALS AND METHODS

Materials. All general biochemicals were from Sigma (Poole, United King-dom). High-glucose Dulbecco modified Eagle medium (DMEM), Europeanfetal calf serum (EFCS), penicillin-streptomycin, oligo(dT)12-18, deoxynucleosidetriphosphate mix, Superscript II RT, and Lipofectamine 2000 were from Invitro-gen (Paisley, United Kingdom). SK1 small interfering RNA (siRNA) and Dhar-maFECT 2 reagent were from Dharmacon (Cromlington, United Kingdom).S1P3 siRNA and anti-phosphorylated ERK1/2, anti-PAK1, and anti-myc tagantibodies were obtained from Santa Cruz (Santa Cruz, CA). Anti-ERK2 anti-body was from BD Transduction Laboratories (Oxford, United Kingdom), andanti-SK1 antibody for cell biochemical studies was kindly provided by AndreaHuwiler (University of Bern, Bern, Switzerland) (15). Anti-SK1 antibody forimmunohistochemical analyses was from Abgent (Abingdon, United Kingdom).Anti-ER� antibody was from Dako (Cambridgeshire, United Kingdom). Anti-HER2 and anti-phospho-PAK1/2 (Thr423) were from Cell Signaling/New En-gland Biolabs, Ltd. (Hitchin, United Kingdom). Anti-FLAG M2 antibody wasfrom Stratagene (La Jolla, CA). Antibodies were used for both Western blottingand cell immunofluorescence. S1P was from Avanti-Polar Lipids (Alabaster,AL), EGF was from Sigma (Poole, United Kingdom), JTE013 was from TocrisBiosciences (Bristol, United Kingdom), and CAY10444 was from CaymanChemicals (Tallinn, Estonia). SKi was from Merck Biosciences (Nottingham,United Kingdom). The Myc-PAK1 plasmid (Addgene plasmid 12209) constructwas from Addgene, Inc. (Cambridge, MA) (31).

Cell culture. MCF-7 (Neo and HER2) breast cancer cells were grown in amonolayer culture in high-glucose DMEM with 10% EFCS and 1% Pen-Strep(penicillin G sodium [10,000 U/ml], streptomycin sulfate [10,000 �g/ml]), 0.4%

Geneticin, and 15 �g of insulin/ml at 37°C with 5% CO2. BT474 cells were grownin the same medium but without Geneticin.

PAK1 and SK1 transfection. MCF-7 HER2 cells were transfected with myc-tagged PAK1 plasmid construct (1 �g per well), while BT474 cells were transfectedwith wild-type hSK1 or G82D hSK1 plasmid construct (42) using Lipofectamine2000 reagent according to the manufacturer’s instructions. Transfection was per-formed for 24 h at 37°C before serum starvation for a further 24 h prior to cellstimulation as appropriate.

siRNA treatment. SK1 and S1P3 downregulation was achieved using sequence-specific SK1 or S1P3 siRNA. siRNA transfection was performed according to theprotocol detailed by Dharmacon. Briefly, cells grown on 24-well plates weretransfected with 100 nM siRNA or scrambled siRNA prepared in a mix withDharmaFECT 2 reagent and DMEM containing 10% serum. The cells were thencultured for 48 h before serum starvation for 24 h prior to stimulation or RNApurification.

cDNA synthesis using Superscript II RT. In a 20-�l reaction, 1 �l ofoligo(dT)12-18 (500 �g/ml), 0.5 �l of deoxynucleoside triphosphate mix (20 mM),and 1 �g of total RNA were mixed in a nuclease-free microcentrifuge tube withthe appropriate amount of nuclease-free water to a total volume of 13 �l. Themixture was heated at 65°C for 5 min and chilled quickly on ice before the ad-dition of 4 �l of 5� first-strand buffer and 2 �l of 0.1 M dithiothreitol. Thecontents of the tube were then incubated at 42°C for 2 min. Finally, 1 �l (200 U)of Superscript II reverse transcriptase (RT) was added to the mixture, which wasincubated at 42°C for a further 50 min before inactivation at 70°C for 15 min.Control samples (RT�) were prepared using 1 �l of nuclease-free water insteadof Superscript II RT. The cDNA produced was subsequently used as a templatefor amplification in PCR.

PCR. PCR was performed to identify the mRNA transcript for PAK1, S1P2,S1P3, and SK1. The primers used for the reactions were as follows: S1P2, forward(CACTCGGCAATGTACCTGTTTC) and reverse (GACGCCTAGCACGATGGTGAC); S1P3, forward (GACTGCTCTACCATCCTGCCC) and reverse (GTAGATGACCGGGTTCATGGC); SK1, forward (CTGTCACCCATGAACCTGCTGTC) and reverse (CATGGCCAGGAAGAGGCGCAGC); G3PDH,forward (TGAAGGTCGGTGTCAACGGATTTGGC) and reverse (CATGTAGGCCATGAGGTCCACCAC); and PAK1, forward (CACTCCACCAGATGCTTTGA) and reverse (CAGGGACCAGATGTCAACCT).

The PCR conditions were as follows: 1 cycle of initial denaturation at 94°C for2 min; followed by 30 cycles of amplification at 94°C for 1 min 30 s, then either52°C (G3PDH), 56°C (S1P2 and S1P3), 55°C (SK1), or 54°C (PAK1) for 30 s, andthen 72°C for 1 min 40 s; followed by a final extension at 72°C for 5 min.

Western blotting. Analysis of proteins by SDS-PAGE and Western blotting wasperformed as previously described by us (1) using anti-phosphorylated ERK1/2,anti-ERK2, anti-SK1, anti-PAK1, anti-HER2, and anti-ER� antibodies.

Immunofluorescence microscopy. Cells were plated onto autoclaved 13-mmglass coverslips and grown to 60% confluence before serum starvation for 48 hprior to stimulation. Cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS) for 10 min, permeabilized with 0.1% Triton X-100 in PBSfor 1 min, and then incubated in blocking solution (5% EFCS and 1% bovineserum albumin in PBS) for 30 min at room temperature. Coverslips were thenincubated with primary antibody (1:100 dilution in blocking solution), as re-quired, overnight at 4°C. The coverslips were washed with PBS and incubatedwith fluorescein isothiocyanate (FITC)- or TRITC (tetramethyl rhodamine iso-thiocyanate)-conjugated anti-mouse IgG or anti-rabbit IgG secondary antibody(1:100 dilution in blocking solution), as appropriate, at room temperature for 1 h.The coverslips were washed in PBS, the excess moisture was removed with atissue, and the coverslips were then mounted on glass slides using Vectashieldhard set mounting medium with DAPI (4�,6�-diamidino-2-phenylindole). Immu-nofluorescence was visualized by using a Nikon (Surrey, United Kingdom) E600epifluorescence microscope.

Scratch assay. MCF-7 Neo and MCF-7 HER2 cells were grown to 95%confluence in 24-well plates and serum starved for 48 h prior to experimentation.The cell monolayer was subsequently manually scratched with a sterile 1-mlpipette tip before treatment with or without S1P. The cells were then left at 37°Cfor 24 h to allow the migration to proceed. The scratched area was photographedimmediately after scratching (0 h) and after 24 h at the same area to compare theextent of closure as a representation of cell migration. The percentage of mi-gration was determined as the ratio of the difference between the scratch widthat 0 h and that at 24 h. Statistical analysis was performed with the GraphPadPrism software, using one-way analysis of variance (ANOVA), followed by theNewman-Keuls post-hoc test.

BT474 cells were grown to 70% confluence in 12-well plates, transfected withvector or G82D hSK1 plasmid construct for 24 h, and then serum starved for24 h. The cells were scratched and, after 24 h, photographed. The number of

3828 LONG ET AL. MOL. CELL. BIOL.

migrating cells was counted, and the results are expressed as the average �

standard deviation for vector- versus G82D hSK1-overexpressing cells. Statisticalanalysis was performed using GraphPad Prism software and the Student t test.

SK1 assay. Assays were performed according to the method of Delon et al. (7).Densitometry. Densitometric quantification of Western blots was performed

using the Molecular Analyst Software (Bio-Rad Laboratories, Ltd.). Statisticalanalysis was performed using GraphPad Prism software and one-way ANOVA,followed by the Newman-Keuls post-hoc test. A P value of 0.05 was consideredsignificant.

Patient samples. The local Ethics Committee granted ethical approval for thestudy. The patient cohort consisted of 304 ER�-positive breast cancer patientswho had been diagnosed with operable disease between 1980 and 1999 at theGlasgow Royal Infirmary. Patient follow-up details included information onclinical attendances, recurrence and metastasis, and date and cause of death, aswell as adjuvant therapy details. All patients were treated with tamoxifen for 5years. In addition to receiving tamoxifen, 95 patients (31%) received adjuvantchemotherapy (3 unknown), and 98 (32%) received adjuvant radiotherapy (3unknown).

Immunohistochemistry. The breast tumor specimens were each formalin fixedand paraffin embedded, and then 3- to 4-�m sections were cut and mounted onslides. Immunohistochemistry was performed using tissue microarray technology.Staining for ER and HER family members has been previously performed for thecohort (34). The tissue microarray slides were first dewaxed and rehydratedthrough a series of xylene and alcohol washes. Antigen retrieval was performedby microwaving the slides under pressure in a citrate buffer for 5 min (pH 6.0).Endogenous peroxidase was blocked by using 3% hydrogen peroxide for 20 min,and nonspecific background staining was reduced by blocking with a 1:10 con-centration of casein diluted 10-fold in Tris-buffered saline (pH 7) for 20 min. Thesections were incubated with the primary antibody. Anti-SK1 antibody (Abgent)was incubated at a dilution of 1:50 and at temperatures of 25°C in a humidifiedincubator for 60 min. EnVision-horseradish peroxidase conjugate was used forsignal amplification, and positive staining was identified by using 3,3�-diamino-benzidine chromogen (Vector Laboratories). The slides were then counter-stained with hematoxylin and Scott’s tap water substitute before dehydration andmounting.

Scoring method. SK1 protein expression was evaluated by two independentobservers using a semiquantitative weighted histoscore method (18). The histo-score was calculated by examining the intensity of staining for each specified celllocation (membrane, cytoplasm, and nuclear) using a light microscope. Theintensity of staining was categorized as follows: 0 for negative, 1 for weakstaining, 2 for moderate staining, and 3 for strong staining. The percentage oftumor cells within each location was then estimated and the histoscore calculatedusing the following expression: (0 � % negative) � (1 � % weak) � (2 � %moderate) � (3 � % strong). The final histoscore ranged from between 0 to amaximum of 300. Each tissue microarray result was triple scored and an averagehistoscore for each location was calculated. The two independent observer’sscores were compared and the interclass correlation coefficient was calculated foreach cellular location. A value above 0.7 is categorized as excellent, and thevalues in the present study for membrane, nuclear, and cytoplasmic were 0.861,0.943, 0.910, respectively. The weighed histoscore method provides a numericalvalue ranging from 0 to 300 for each protein in each cellular location for eachsample. We used the widely utilized method of categorizing tumors by weightedhistoscore, which uses the median histoscore as the cutoff (3, 4, 20, 21). There-fore, in the present study we used the median weighted histoscore as the cutofffor categorizing tumors into high and low expressions of SK1. Expression levelsof less than the median histoscore were categorized as the low-expression cohort,and levels greater than the median were categorized as the high-expressioncohort. HER1-3 status was assessed by using IHC. If the intensity of the stainingwas categorized as 3�, then the tumor was considered positive; if it was 1�, thenthe tumor was considered negative; and if it was 2�, then the sample wasevaluated by fluorescence in situ hybridization to determine whether HER2 wasamplified or not.

Statistical analysis. Standard clinical follow-up for breast cancer patients is 10years; therefore, all clinical data were measured and calculated with a 10-yearcutoff. The endpoints used in the analysis of breast cancer survival were disease-free survival, overall survival, and recurrence while on tamoxifen therapy. Thestatistical software package used for the analysis of data was SPSS (version 15.0for Microsoft Windows). Histoscores were assessed, and expression levels weregiven as means and interquartile ranges. A Pearson chi-square test was used tocorrelate expression levels with survival analysis.

RESULTS

HER2 and SK1 regulate each others expression in MCF-7cells. It has been previously established that EGF increasesthe expression of SK1 in MCF-7 cells (8). We have thereforeinvestigated whether a related EGF receptor family member,HER2, also regulates SK1 expression. HER2 was chosen be-cause of its oncogenic function and defined role in breastcancer progression. For the present study we used MCF-7 cellsstably transfected with a HER2 (MCF-7 HER218 clone) plas-mid construct. Comparison was made with MCF-7 Neo cells(stably transfected with Neo vector). MCF-7 HER2 cells ex-hibit ER� expression (Fig. 1A, which is reduced compared toMCF-7 Neo vector transfected cells) and stably overexpressHER2 (185 kDa) (Fig. 1B). Both MCF-7 HER2 and Neo cellsexpress SK1 (42 kDa) (Fig. 2).

A major finding in terms of defining a functional interactionbetween SK1 and HER2 was the observation that SK1 expres-sion was increased in MCF-7 HER2 cells compared to MCF-7Neo cells (Fig. 2). This was established by increased mRNAtranscript, SK1 protein, and SK1 activity levels in MCF-7HER2 cells compared to their Neo counterparts (Fig. 2). Toinvestigate whether the increased expression of SK1 has anyeffect on HER2, we used a specific siRNA to knock down SK1expression (comparison was made with scrambled siRNA).siRNA knockdown of SK1 (Fig. 3A) had no effect on ERK-2levels (Fig. 3F), the latter being used to establish equal proteinloading and to ascertain that cell viability was not affected bythe siRNA. The siRNA knockdown of SK1 (42 kDa) in MCF-7HER2 cells increased the expression of HER2 (Fig. 3B). Wealso focused on p21-activated protein kinase 1 (p65 PAK1)because this kinase functions downstream of HER2 (37), has acritical role in regulating migration by promoting actin poly-merization in membrane ruffles/lamellipodia, and acts as ascaffold for Raf/MEK1/ERK-1/2 (10). Indeed, ERK-1/2 alsohas an important role in malignant cancer (14, 36).

MCF-7 HER2 cells express two proteins of 59 and 70 kDathat immunoreact with the anti-PAK1 antibody (Fig. 3C). Theconventional PAK1, which has a molecular mass of 65 kDa, isnot present. However, siRNA knockdown of SK1 in MCF-7HER2 cells increased p65 PAK1 and p70 expression and de-creased the amount of p59 (Fig. 3C). These changes are re-ciprocal and stoichiometric, suggesting that p59, p65 PAK1,and p70 might be interconvertible forms of PAK1. p59, but notp70 is phosphorylated (detected with anti-phospho [Thr423]PAK1 antibody) in scrambled siRNA-treated MCF-7 HER2cells (Fig. 3D). However, siRNA knockdown of SK1 increased

FIG. 1. Expression of ER� and HER2 in MCF-7 cells. MCF-7 Neoand MCF-7 HER2 cell lysates were subjected to Western blot analysiswith specific antibodies to show ER� expression (A) and (B) expres-sion of HER2. Equal loading of protein in panels A and B was estab-lished by Western blotting with anti-ERK-1/2 antibody.

VOL. 30, 2010 HER2 AND SPHINGOSINE KINASE 1 IN BREAST CANCER CELLS 3829

the amount of phosphorylated Thr423 p65 PAK1 and p70 (Fig.3D), which correlates with increased formation of p65 PAK1and p70 protein. This contrasts with MCF-7 Neo cells, whereonly p65 PAK1 is phosphorylated and which was increased bystimulation of cells with S1P and unaffected by siRNA knock-down of SK1 (Fig. 3D; see Fig. 7A for the knockdown of SK1expression in MCF-7 Neo cells).

The increase in p65 PAK1 expression in MCF-7 HER2treated with SK1 siRNA was not due to changes in transcrip-tional regulation of the PAK1 gene since PAK1 mRNA levelswere unaltered (Fig. 3E), thereby providing additional supportfor a posttranslational mode of regulation. siRNA knock-down of SK1 also increased basal ERK-1/2 phosphorylationin MCF-7 HER2 cells (Fig. 3F).

S1P signaling. We next investigated the effect of the HER2-dependent increase in SK1 expression on S1P signaling. Therationale for investigating this is based on the fact that S1P canregulate PAK1, which regulates actin cytoskeleton dynamicsand cell migration (9, 19, 26). Therefore, the ability of SK1 tosuppress p65 PAK1 expression in MCF-7 HER2 cells promptedus to investigate whether there are differences in S1P signaling inMCF-7 Neo and MCF-7 HER2 cells.

MCF-7 Neo and MCF-7 HER2 cells express abundantmRNA transcript for S1P3 (product size, 345 bp) and lowlevels of S1P2 (product size, 528 bp) (Fig. 4A). S1P1 was notexpressed in MCF-7 Neo or MCF-7 HER2 cells (data notshown). The identities of the S1P2/3 mRNA transcripts wereconfirmed by nucleotide sequencing. Parental MCF-7 cells ex-hibit an identical S1P receptor expression profile (data notshown). The treatment of MCF-7 Neo cells with S1P stimu-lated ERK-1/2 activation (Fig. 4B). To define which receptortype mediates the effect of S1P on ERK-1/2, we used S1P3-specific siRNA. siRNA knockdown of S1P3 decreased the stim-ulation of ERK-1/2 activation by S1P (Fig. 4B, ratios of phos-phorylated ERK-1/2 to ERK-2: control/scrambled siRNA,0.175 � 0.07; control/S1P3 siRNA, 0.35 � 0.14; S1P/scrambledsiRNA, 2.8 � 0.42; S1P/S1P3 siRNA, 0.77 � 0.18; n � 3, P

0.01 for S1P/S1P3 siRNA versus S1P/scrambled siRNA), indi-cating that S1P induces activation of the ERK-1/2 pathway viaan S1P3-dependent mechanism. As a test of the specificity, wedemonstrate here that siRNA knockdown of S1P3 had noeffect on the EGF-induced activation of ERK-1/2 (Fig. 4B, n �2). Additional evidence to support a role for S1P3 was obtainedby using a pharmacological approach, where CAY10444 (anS1P3 antagonist [17]) reduced the S1P stimulation of ERK-1/2(Fig. 4C, ratios of phosphorylated ERK-1/2 to ERK-2: control,0.22 � 0.07; CAY10444, 0.22 � 0.04; S1P, 2.7 � 0.58; S1P/CAY10444, 0.95 � 0.07; n � 3, P 0.01 for S1P/CAY10444versus S1P), while JTE013 (an S1P2 antagonist [25]) at 1 to 10�M had no significant effect (data not shown). We also foundthat the treatment of MCF-7 Neo cells with S1P stimulated thetranslocation of phosphorylated ERK-1/2 to the nucleus andpromoted the accumulation of phosphorylated ERK-1/2 intomembrane ruffles in lamellipodia (Fig. 4D, the percentages ofcells exhibiting nuclear and membrane ruffle phosphorylatedERK-1/2 were as follows: control, 0; and S1P, 71% � 3.8%).The results are for five to seven fields of view from threeseparate experiments (P 0.001 for S1P versus the control).An identical redistribution of phosphorylated ERK-1/2 in re-sponse to S1P was observed in MCF-7 parental cells (data notshown). The redistribution of phosphorylated ERK-1/2 intomembrane ruffles/lamellipodia and the nucleus in MCF-7 Neocells is characteristic of cells that are in the process of migrat-ing and, indeed, S1P stimulated MCF-7 Neo cell migration, asassessed by using a scratch assay (Fig. 5A). The redistributedphosphorylated ERK-1/2 localization in response to S1P istherefore prognostic of a migratory phenotype.

S1P also stimulated ERK-1/2 activation via an S1P3-depen-dent mechanism in MCF-7 HER2 cells, as evidenced by siRNAknockdown of S1P3, which ablated activation of ERK-1/2 byS1P (Fig. 6A, ratios of phosphorylated ERK-1/2 to ERK-2:control/scrambled siRNA, 1.36 � 0.34; control/S1P3 siRNA,1.8 � 0.1; S1P/scrambled siRNA, 3.16 � 0.29; S1P/S1P3

siRNA, 2.24 � 0.24; n � 3, P 0.05 for S1P/S1P3 siRNA

FIG. 2. Effect of HER2 on SK1 expression in MCF-7 HER2 cells. The data show the comparison of SK1 transcript, protein expression, andactivity in MCF-7 Neo versus MCF-7 HER2 cells. RT-PCR analysis of G3PDH mRNA transcript expression confirmed the use of equal quantitiesof RNA between samples. Western blots were probed with anti-SK1 antibody and anti-ERK-2 antibody (for equal protein loading control). Theidentity of the SK1 mRNA transcript was confirmed by sequence analysis.


versus S1P/scrambled siRNA). This was supported by evidenceobtained using a pharmacological approach where CAY10444abolished the S1P-stimulation of ERK-1/2 (Fig. 6B, ratios ofphosphorylated ERK-1/2 to ERK-2: control, 1.72 � 0.12;CAY10444, 1.35 � 0.43; S1P, 3.19 � 0.15; S1P/CAY10444,1.66 � 0.37; n � 3, P 0.05 for S1P/CAY10444 versus S1P),while JTE013 was without effect (Fig. 6B, ratios of phosphor-ylated ERK-1/2 to ERK-2: control, 1.3 � 0.32; JTE013, 1.43 �0.23; S1P, 2.72 � 0.51; S1P/JTE013, 3.04 � 0.37; n � 3).However, in contrast to MCF-7 Neo cells, phosphorylatedERK-1/2 was localized to the cytoplasm of S1P-stimulatedMCF-7 HER2 cells (Fig. 6C). There was no nuclear translo-cation or accumulation of phosphorylated ERK-1/2 in mem-

brane ruffles/lamellipodia (Fig. 6C, the percentages of cellswith cytoplasmic phosphorylated ERK-1/2 were as follows:control, 54% � 2%; and S1P, 89% � 7%). The results are forfive to eight fields of view from four separate experiments (P 0.01 for S1P versus control). Moreover, MCF-7 HER2 cellsfailed to migrate or spread in response to S1P (Fig. 5B). There-fore, S1P/S1P3 signaling is impaired and correlated with thereduced HER2 and ablated PAK1 expression induced by SK1in MCF-7 HER2 cells (Fig. 3B and C).

SK1 regulates formation of a migratory phenotype in MCF-7cells. Our data suggest that the HER2-dependent increase inSK1 expression might result in desensitization of S1P-inducedmigration. Implicit in this model is that under conditions whereSK1 expression is not elevated to levels achieved in MCF-7HER2 cells, SK1 might be involved in promoting formation ofa migratory phenotype. To assess this, we investigated the roleof SK1 in regulating actin cytoskeletal rearrangement, which isrequired for migration of MCF-7 Neo cells. MCF-7 Neo cellsexhibit a phenotype in which actin is clustered into adhesionfoci that are concentrated at the cell periphery in scrambledand SK1 siRNA-treated control cells (Fig. 7A, upper left andright panels, arrows denote actin adhesion foci). S1P stimula-tion of scrambled siRNA-treated MCF-7 Neo cells produced amarked rearrangement of actin, which spread into membraneruffles/lamellipodia (Fig. 7A, lower left panel, arrows denoteactin membrane ruffles). siRNA knockdown of SK1 expressionreduced S1P-stimulated formation of actin containing mem-brane ruffles/lamellipodia and restored formation of adhesionfoci (Fig. 7A, lower right panel, arrows denote actin adhesionfoci; the percentages of cells with actin rearrangement intomembrane ruffles were as follows: control/scrambled siRNA,2% � 1.4%; S1P/scrambled siRNA, 63.6% � 8.5%; control/SK1 siRNA, 8.5% � 0.7%; S1P/SK1 siRNA, 11.3% � 4.2%).The results are for 5 to 14 fields of view from three separateexperiments (P 0.01 for S1P/SK1 siRNA versus S1P/scram-bled siRNA), demonstrating a key role for SK1 in regulatingthe transition from adherence to migration.

Functional interaction between SK1 and S1P3 was demon-strated by using anti-SK1 antibody and immunofluorescenceanalysis of MCF-7 Neo cells. SK1 was localized in the nucleusand cytoplasm of MCF-7 Neo cells (Fig. 7B). The nuclearlocalization of SK1 is consistent with the fact that SK1 has twofunctional nuclear export sequences and can transit throughthe nucleus (16). The stimulation of MCF-7 Neo cells with S1Ppromoted the relocalization of SK1 into membrane ruffles/lamellipodia (Fig. 7B), and this was blocked by pretreating thecells with the S1P3 antagonist CAY10444 (Fig. 7B, the per-centages of cells exhibiting plasma membrane SK1 were asfollows: control, 0%; S1P, 61.6% � 8.4%; CAY10444, 1.4% �0.07%; and S1P/CAY10444, 7% � 1.4%). The results are for 8to 12 fields of view from three to five separate experiments(P 0.001 for S1P versus S1P/CAY10444).

A role for S1P3 and SK1 in regulating S1P-stimulated mi-gration of MCF-7 Neo was assessed by using CAY10444 or SKi[2-(p-hydroxyanilino)-4-(p-chlorophenyl)thiazole] (11). Weconfirmed that SKi inhibits recombinant SK1 in vitro (data notshown). S1P3 and SK1 are involved in regulating MCF-7 Neocell motility based on results showing that SKi and CAY10444reduced S1P-stimulated cell migration in the scratch assay

FIG. 3. Effect of SK1 siRNA on HER2, PAK1, and ERK-1/2 sig-naling. MCF-7 HER2 cells were pretreated with scrambled or SK1siRNA (100 nM, 48 h). (A) A Western blot shows the knockdown ofSK1 (42 kDa) with SK1 siRNA. (B and C) Western blots show theeffects of siRNA knock down of SK1 on HER2 expression (185 kDa)(B) and PAK1 expression (65 kDa) (C). A Western blot also demon-strates the possible interconversion of p65 PAK1, p70, and p59 inMCF-7 HER2 cells. MCF-7 HER2 cells were stimulated with or with-out S1P (1 �M) for 5 min. (D) Phosphorylated p65 PAK1 (detectedwith anti-phospho PAK1 antibody) in scrambled siRNA versus SK1siRNA-treated MCF-7 Neo cells and phosphorylated p65 PAK1, p70,and p59 in scrambled siRNA versus SK1 siRNA-treated MCF-7 HER2cells. Both MCF-7 Neo and MCF-7 HER2 cells were treated with orwithout S1P (1 �M) for 5 min. (E) RT-PCR analysis with PAK1specific primers shows that siRNA knockdown of SK1 does not affectPAK1 mRNA transcript levels in MCF-7 HER2 cells (G3PDH wasused as a control for equal RNA loading). (F) Western blotting showsthe effect of SK1 siRNA on ERK-1/2 activation in MCF-7 HER2 cells(detected with anti-phosphorylated ERK-1/2 antibody; blots were alsoprobed with anti-ERK-2 antibody to ensure equal protein loading).



(Fig. 7C and D). A role for phosphorylated ERK-1/2 in regu-lating migration was evidenced by data showing that U0126(MEK inhibitor) reduced S1P-stimulated migration in scratchassays (Fig. 7C).

SK1 regulates actin rearrangement in MCF-7 HER2 cells.We next investigated whether the regulation of actin rear-rangement is impaired in MCF-7 HER2 cells in order to ex-plain the failure of these cells to migrate to S1P (Fig. 5B). Wealso assessed the role of SK1 in regulating actin reorganizationin these cells. Actin was clustered into adhesion foci that wereconcentrated at the cell periphery in control MCF-7 HER2cells treated with scrambled or specific SK1 siRNA (Fig. 8A,upper left and right panels, arrows denote actin adhesion foci).In contrast to MCF-7 Neo cells, S1P stimulation of MCF-7HER2 cells failed to induce actin containing membrane ruffleformation (Fig. 8A, lower left panel, arrows denote actin ad-hesion foci), a finding consistent with the lack of effect of S1Pon migration (Fig. 5A). However, actin spreads into membraneruffles/lamellipodia upon S1P stimulation of MCF-7 HER2cells when SK1 was knocked down with siRNA (Fig. 8A, lowerright panel, arrows denote actin membrane ruffles; the per-centages of cells with actin rearrangement into membrane ruf-fles were as follows: control/scrambled siRNA, 2.2% � 1.45%;S1P/scrambled siRNA; 6.5% � 3.7%; control/SK1 siRNA,7.3% � 4.4%; and S1P/SK1 siRNA, 35.6% � 5.1%). Theresults are for 5 to 14 fields of view from three separate ex-periments (P 0.01 for S1P/SK1 siRNA versus S1P/scrambledsiRNA). These data are therefore consistent with a model in

which the siRNA knockdown of SK1 converts MCF-7 HER2cells to a more aggressive migratory phenotype. Indeed, siRNAknockdown of SK1 restored responsiveness to S1P in termsof the stimulation of migration of MCF-7 HER2 cells (Fig. 8B).The increase in S1P-stimulated migration represents 50% ofthe migratory response to S1P observed with MCF-7 Neo cells(Fig. 5A). A migratory actin membrane ruffle/lamellipodiaphenotype could also be induced with S1P when MCF-7 HER2cells were enforced to overexpress myc-tagged p65 PAK1 (Fig.8C, arrows denote actin-containing membrane ruffles in myc-tagged p65 PAK1-overexpressing cells stimulated with S1P[bottom left panel, the percentages of myc-tagged p65 PAK1overexpressing MCF-7 HER2 cells in which actin is rearrangedinto membrane ruffles were as follows: control, 7.5% � 0.6%;and S1P, 53.8% � 8.1%]). The results are for 5 to 14 fields ofview from three separate experiments (P 0.01 for S1P versuscontrol), thereby clearly demonstrating that S1P can use p65PAK1 to induce formation of the migratory phenotype. More-over, enforced expression of myc-tagged p65 PAK1 in MCF-7HER2 increased basal and S1P-stimulated activation of ERK-1/2 (Fig. 8D).

Role of SK1 in regulating S1P3 receptor expression. SincesiRNA knockdown of SK1 markedly reduced the S1P-stimu-lated formation of the migratory phenotype in MCF-7 Neocells, we assessed whether SK1 regulates S1P3 expression inthese cells. Using RT-PCR to measure S1P3 transcript levels,we found that the siRNA knockdown of SK1 reduced S1P3

mRNA expression by 50 to 75% (n � 3 separate experi-ments) in MCF-7 Neo cells (Fig. 9A). The reduction in S1P3

expression appears to have a functional effect on the S1P/S1P3-dependent regulation of ERK-1/2 in MCF-7 Neo cells. Thiswas evidenced by results showing that the S1P stimulation ofERK-1/2 in MCF-7 Neo cells was substantially reduced whenSK1 expression was knocked down with siRNA (Fig. 9B, ratiosof phosphorylated ERK-1/2 to ERK-2: control/scrambledsiRNA, 0.21 � 0.7; control/SK1 siRNA, 0.14 � 0.04; S1P/scrambled siRNA, 2.6 � 0.56; S1P/SK1 siRNA, 0.91 � 0.07;n � 3, P 0.01 for S1P/SK1 siRNA versus S1P/scrambledsiRNA). In contrast, the siRNA knockdown of SK1 increasedbasal ERK-1/2 activation in MCF-7 HER2 cells (Fig. 3F) andan additive stimulation of ERK-1/2 was obtained when thesecells were treated with S1P (ratios of phosphorylated ERK-1/2to ERK-2: control scrambled siRNA treated, 1.8 � 0.16; con-trol SK1 siRNA treated, 2.98 � 0.28; S1P [1 �M, 5 min]stimulated scrambled siRNA-treated, 3.54 � 0.79; S1P [1 �M,5 min] stimulated SK1 siRNA treated, 4.7 � 0.33; n � 3

FIG. 4. S1P-induced regulation of ERK-1/2 in MCF-7 Neo cells. (A) RT-PCR analysis with gene-specific primers showing the expression ofS1P2 and S1P3 mRNA transcript in MCF-7 Neo and HER2 cells. (B and C) MCF-7 Neo cells were treated with or without S1P3 siRNA (100 nM,48 h) or CAY10444 (10 �M) for 30 min and then stimulated with or without S1P (1 �M) or EGF (25 ng/ml) for 5 min. In panel B, a Western blotshows the effect of siRNA knockdown of S1P3 on S1P and EGF stimulation of ERK-1/2 in MCF-7 Neo cells. An inset shows RT-PCRdemonstrating that S1P3 siRNA reduced S1P3 mRNA transcript but not G3PDH mRNA (housekeeping gene) transcript. In panel C, a Westernblot shows the effect of CAY10444 on S1P-stimulated ERK-1/2 activation in MCF-7 Neo cells. In panels B and C, phosphorylated ERK-1/2 wasdetected on Western blots probed with anti-phosphorylated ERK-1/2 antibody. The blots were also probed with anti-ERK-2 antibody to ensureequal protein loading. (D) Immunofluorescence microscopy images showing the subcellular distribution of phosphorylated ERK-1/2 in MCF-7 Neocells stimulated with S1P (1 �M, 5 min) or control (Con) cells. Phosphorylated ERK-1/2 was detected with anti-phosphorylated ERK-1/2 antibody.There was no significant immunostaining with TRITC-linked secondary antibody alone (data not shown). Arrows denote areas of accumulationof phosphorylated ERK-1/2 in membrane ruffles/lamellipodia and the nucleus (costained with DAPI). Bar, 50 �m.

FIG. 5. Migration MCF-7 Neo and HER2 cells. MCF-7 Neo(A) and MCF-7 HER2 (B) cells were subjected to a scratch assay andthen treated with or without S1P (1 �M) for 24 h. For each celltreatment, 6 to 12 random views of three to six scratches were cap-tured. The histogram shows the effect of S1P on migration in the twocell types. The percentage of migration is presented as the ratio of thedifference between scratch widths at 0 and 24 h.


FIG. 6. S1P-induced regulation of ERK-1/2 in MCF-7 HER2 cells. MCF-7 HER2 cells were treated with or without S1P3 siRNA (100 nM, 48 h)or CAY10444 (10 �M) or JTE013 (5 �M) for 30 min and then treated with or without S1P (1 �M) for 5 min. (A) A Western blot shows the effectof siRNA knockdown of S1P3 on S1P stimulation of ERK-1/2. In the inset, RT-PCR shows that S1P3 siRNA reduced S1P3 mRNA transcript butnot G3PDH mRNA (the housekeeping gene) transcript. (B) A Western blot shows the effect of CAY10444 and JTE013 on S1P-stimulatedERK-1/2 activation. In both panels A and B, phosphorylated ERK-1/2 was detected on Western blots probed with anti-phosphorylated ERK-1/2antibody. The blots were also probed with anti-ERK-2 antibody to ensure equal protein loading. (C) Immunofluorescence images showing thesubcellular distribution of phosphorylated ERK-1/2 in MCF-7 HER2 cells stimulated with S1P (1 �M, 5 min) or control (C) cells. PhosphorylatedERK-1/2 was detected with anti-phosphorylated ERK-1/2 antibody. DAPI staining was for identification of the nucleus. Bar, 50 �m.

3834

separate experiments, P 0.05 for S1P [1 �M, 5 min] stimu-lated and SK1 siRNA treated versus each other treatment).

EGF stimulation of the ERK-1/2 pathway does not involveS1P3 in MCF-7 Neo cells based on the finding that siRNAknockdown of S1P3 had no effect (Fig. 4B). The lack of in-volvement of S1P3 is therefore consistent with the finding thatthe EGF-stimulated activation of ERK-1/2 was not reduced bythe siRNA knockdown of SK1 (data not shown). We alsofound that phorbol ester (phorbol myristate acetate)-inducedactivation of ERK-1/2 (receptor independent) was not affectedby siRNA knockdown of SK1 (data not shown).

We also evaluated whether the stimulation of ERK-1/2 byexogenously added S1P involves regulation by so-called “in-side-out” signaling, where intracellular S1P, produced by SK1,is released into the extracellular milieu or partitions into lipidmicroenvironments in close proximity to S1P receptors. Thismight allow privileged and efficient binding of S1P to S1P recep-tors. To test this possibility, we pretreated MCF-7 Neo cells withSKi for 15 min to inhibit SK1 prior to agonist stimulation. How-ever, SKi had no effect on S1P- or EGF-stimulated activation ofERK-1/2 (Fig. 9C), suggesting that “inside-out” signaling does notoperate for either EGF or S1P in MCF-7 cells.

FIG. 7. Role of SK1 in regulating S1P-stimulated actin arrangement and migration in MCF-7 Neo cells. MCF-7 Neo cells were pretreatedwith scrambled or SK1 siRNA (100 nM, 48 h) or SKi (10 �M), CAY10444 (10 �M), or U0126 (5 �M) for 15 min and then stimulated ornot stimulated (control [C]) with S1P (1 �M) for 5 min (A and B) or 24 h (C and D). (A) The images show the S1P-stimulated redistributionof actin in MCF-7 Neo cells that were treated with scrambled or SK1 siRNA. Also shown is a Western blot probed with anti-SK1 antibodyand demonstrating the knockdown of SK1 with siRNA. Actin was stained with phalloidin red. Arrows denote actin-containing focaladhesions in scrambled siRNA-treated control cells (upper left panel), SK1 siRNA-treated control cells (upper right panel), and SK1siRNA-treated S1P-stimulated cells (lower right panel) and actin containing membrane ruffles in scrambled siRNA treated S1P stimulatedcells (lower left panel). Bar, 50 �m. (B) Immunofluorescence microscopy images showing the subcellular distribution of SK1 in control,CAY10444-, S1P-, and S1P/CAY10444-treated cells, using anti-SK1 antibody. Arrows denote areas of accumulation of SK1 in membraneruffles/lamellipodia. FITC-labeled secondary antibody alone staining is also shown. Bar, 50 �m. (C) Histogram showing the effect of SKi andU0126 on S1P-stimulated cell migration. (D) Histogram showing the effect of CAY10444 on S1P-stimulated cell migration. In panels C andD, 12 to 18 random views of four to six scratches were captured for each treatment.


Tissue microarray analysis. Therefore, our findings suggestthat, dependent on the phenotype of the cancer cell and spe-cifically the expression of HER2, SK1 can acquire differentfunctionality. In order to establish whether a similar relation-ship of SK1 holds in primary breast cancer, we analyzed thetumors of 304 patients with ER� breast cancer who subse-quently received tamoxifen therapy for a median period of 5years. An example of tissue microarray analysis with anti-SK1antibody demonstrating low and high expression of SK1 in theER� patient cohort is shown (Fig. 10A). Immunostaining withanti-SK1 antibody was abolished by inclusion of the immuniz-ing peptide in the tissue microarray incubation (Fig. 10A),thereby confirming the specificity of the anti-SK1 antibody.

High cytoplasmic SK1 expression in patient tumors was signif-icantly associated with a shorter mean time to recurrence ontamoxifen (increased resistance) and a reduced mean disease-specific survival time compared to patients whose tumors ex-press low SK1 levels (Fig. 10B). Moreover, we have previouslydemonstrated that phosphorylated cytoplasmic ERK-1/2 (andits upstream regulator phosphorylated Raf-1), in ER� breastcancer tumors, is also associated with reduced patient survival(22). Therefore, our evidence supports the existence of anactive SK1/ERK-1/2 pathway in human ER� breast cancer(that is associated with poor prognosis) and which recapitu-lates the biochemical pathways identified by us in culturedbreast cancer MCF-7 cells, where we have shown that S1P/

FIG. 8. Effect of siRNA knockdown of SK1 and myc-tagged PAK1 overexpression on actin arrangement in MCF-7 HER2 cells. (A) Imagesshow the subcellular distribution of actin in MCF-7 HER2 cells pretreated with scrambled or SK1 siRNA (100 nM, 48 h) and then stimulated ornot stimulated (control [C]) with S1P (1 �M) for 5 min. Arrows denote actin-containing focal adhesions in scrambled siRNA-treated control cells(upper left panel) and in siRNA SK1-treated control cells (upper right panel) and in scrambled siRNA-treated S1P-stimulated cells (lower leftpanel) and actin-containing membrane ruffles in siRNA SK1-treated S1P-stimulated cells (lower right panel). Actin was stained with phalloidin red.Bar, 50 �m. (B) Histogram showing migration of MCF-7 HER2 cells in response to S1P after siRNA knockdown of SK1. These results are from8 to 10 views from two to four scratches. (C) Images show the subcellular distribution of actin in MCF-7 HER2 cells overexpressing recombinantmyc-tagged p65 PAK1 (detected with anti-myc tag antibody [green]) and then stimulated with S1P (1 �M) for 5 min. As a control, FITC-labeledsecondary antibody incubation alone with cells is also shown. Arrows denote actin-containing membrane ruffles in myc-tagged p65 PAK1-overexpressing cells stimulated with S1P (bottom left panel). Bar, 50 �m. (D) Western blot showing the effect of enforced expression of myc-taggedp65 PAK1 on basal and S1P (1 �M, 5 min) stimulation of ERK-1/2 in MCF-7 HER2 cells. Phosphorylated ERK-1/2 was detected on Western blotsprobed with anti-phosphorylated ERK-1/2 antibody. Blots were also probed with anti-ERK-2 antibody to ensure equal protein loading.


S1P3/SK1/ERK-1/2 regulates the formation of a migratory phe-notype.

We stratified the data according to the HER1-3 status ofpatient tumors. An example of tissue microarray analysis withanti-SK1 antibody demonstrating low and high expression ofSK1 in the HER1�/HER2�/HER3� patient cohort is shown(Fig. 10C). We found that high cytoplasmic SK1 expression inHER1�/HER2�/HER3� tumors was associated with an in-crease in the mean disease-specific patient survival time (Fig.10C). When the patient group was stratified for HER2� aloneand high versus low SK1 expression was compared, a protec-tive effect of high SK1 was also observed on patient survival.The HER2�/low-SK1 group had reduced mean survival timecompared to the HER2�/low-SK1 group (P � 0.03), whereasthe HER2�/high-SK1 group mean survival time was not sig-nificantly different compared to the HER2�/high-SK1 group(P � 0.473). High cytoplasmic SK1 expression was also asso-ciated with a significant increase in the mean time to recur-

rence during tamoxifen treatment in the HER1�/HER2�

group (Fig. 10D).The findings of our in vitro studies demonstrated a complex

interplay between HER2 and SK1 expression, while anotherstudy demonstrated a relationship between HER1 and SK1 inMCF-7 cells (8). Therefore, for each individual patient, we ex-pressed the data as a HER1-3/SK1 ratio. Patients with a lowHER1-3/SK1 ratio had increased mean disease-specific survivaltime and time to recurrence on tamoxifen compared to patientswith a high HER1-3/SK1 ratio in their tumors (Fig. 10E).

BT474 cells and interaction between SK1 and HER2 signal-ing. Finally, we investigated whether the ability of SK1 toinduce tolerance to HER2 could be extended to other breastcancer cell types. We used BT474 cells because they expressSK1 (42 kDa), confirmed by knockdown with SK1 siRNA (datanot shown), ER�, and HER2 (Fig. 11). This is not a simplecomparison, because the behavior of ER� breast cancer cellstypical of that observed in MCF-7 HER2 cells is likely governedby a very precise ratio of SK1 to HER2. In order to increase theSK1/HER2 ratio, we transiently transfected BT474 cells withplasmid construct encoding FLAG-tagged hSK1, and com-pared this with BT474 cells transfected with vector. Overex-pression of hSK1 reduced the expression of HER2 and p65PAK1 (and p70 [p59 was absent]) and reduced the phosphor-ylation of ERK-1/2 in BT474 cells (Fig. 11A). Moreover, over-expression of the dominant-negative G82D kinase-dead hSK1mutant in BT474 cells results in an increase in HER2 and p65PAK1 (and p70) expression and ERK-1/2 phosphorylation (Fig.11A). These findings demonstrate that changes in the HER2-p65PAK1-phosphorylated ERK-1/2 network are dependent on SK1activity and are similar to the functional interaction between SK1and HER2/PAK1 and ERK-1/2 signaling in MCF-7 HER2 cells.Moreover, G82D kinase-dead SK1-overexpressing BT474 cellsdemonstrate increased migration in the scratch injury assay com-pared to vector-transfected BT474 cells (Fig. 11B). This is pre-dicted for G82D kinase-dead hSK1-overexpressing cells becauseof the increased HER2 and PAK1 expression and enhanced basalERK-1/2. The results are also in line with the clinical data show-ing that SK1 is protective (possibly because of reduced breastcancer cell motility) against mortality in ER�/HER2� breast can-cer patients (Fig. 10C).

DISCUSSION

Evidence for a key role of S1P3/SK1 in regulating the for-mation of a migratory phenotype was obtained from our invitro findings using MCF-7 Neo cells. We demonstrated thatS1P binding to S1P3 stimulates the accumulation of phosphor-ylated ERK-1/2 into membrane ruffles/lamellipodia and thenucleus and promotes MCF-7 Neo cell migration. Evidence fora role for S1P3 in this biochemical mechanism is based onresults showing that siRNA knockdown of S1P3 or using apharmacological S1P3 antagonist, CAY10444, reduced theS1P-stimulated activation of ERK-1/2. The accumulation ofphosphorylated ERK-1/2 in membrane ruffles/lamellipodiaand the nucleus of MCF-7 Neo cells is therefore indicative ofthe formation of a migratory phenotype. Nuclear phosphory-lated ERK-1/2 has been demonstrated to induce metallopro-teinase genes, which enables proteolytic degradation of the cellmatrix, required to free cells so they can move, while phosphor-

FIG. 9. Effect of siRNA knockdown of SK1 on S1P3 expression andevaluation of “inside-out” signaling. (A) RT-PCR analysis with gene-specific primers showing the expression of S1P3 mRNA transcript inMCF-7 Neo-treated cells with scrambled or SK1 siRNA (100 nM,48 h). G3PDH mRNA was measured to ensure equal RNA template.(B) MCF-7 Neo cells were pretreated with scrambled siRNA or SK1siRNA (100 nM, 48 h) and then stimulated with S1P (1 �M) for 5 min.Western blotting shows the effect of siRNA knockdown of SK1 onS1P-stimulated ERK-1/2 activation in MCF-7 Neo cells. (C) MCF-7Neo cells were pretreated with SKi (10 �M) for 15 min prior tostimulation with S1P (1 �M) or EGF (25 ng/ml) for 5 min. TheWestern blots show the lack of effect of SKi on S1P- or EGF-stimu-lated activation of ERK-1/2. Phosphorylated ERK-1/2 was detectedwith anti-phosphorylated ERK-1/2 antibody. Blots were also probedwith anti-ERK-2 antibody to ensure equal protein loading.


ylated ERK-1/2 localization in membrane ruffles regulates theactomyosin contractility required for cell migration (38). In-deed, phosphorylated ERK-1/2 has an important role in regu-lating the migratory phenotype of MCF-7 Neo cells sinceU0126 (MEK-1 inhibitor) ablated the migration of these cellsin response to S1P. In addition, S1P-stimulated formation ofmembrane ruffles (data not shown) and migration are blockedby the S1P3 antagonist, CAY10444. Thus, the stimulation ofERK-1/2 by S1P is a major pathway regulating the formation ofa migratory phenotype. S1P3 has previously been shown to bea promigratory receptor in a number of cell types (32, 43), andthis is consistent with a role proposed here for S1P3 in breastcancer cells.

We have also shown that the siRNA knockdown of SK1

reduces S1P3 mRNA transcript in MCF-7 Neo cells. The func-tional consequence of this appears to be to decrease the S1P/S1P3-stimulated activation of ERK-1/2 and thereby abrogateformation of the migratory phenotype. The mechanism by whichSK1 can regulate S1P3 expression is under investigation. In thisregard, the S1P3 5� untranslated region (using TRANSFAC atBiobase) contains predicted ER�, Sp1, and c-Jun binding sites.Sp1 and c-Jun are regulated by ERK-1/2 (33, 44), and there-fore ERK-1/2 may play a role in a positive-feedback loop thatdefines the efficacy with which S1P induces a migratory phe-notype in MCF-7 cells. In contrast, in ER�/HER2� MDA-MB-453 breast cancer cells, the predominant functional S1Preceptor is S1P4 and S1P, binding to this receptor transacti-vates HER2 to regulate the ERK-1/2 pathway (J. S. Long et al.,

FIG. 10. Effect of SK1 on survival and/or recurrence in ER� patients. (A) Tissue microarray analysis with anti-SK1 antibody showing low andhigh expression of SK1 in tumors from the ER� patient cohort and that immunizing peptide reduces SK1 immunoreactivity. (B) Kaplan-Meiersurvival curves demonstrate cumulative disease-free survival differences (endpoint of breast cancer recurrence) and cumulative disease-specificsurvival differences (endpoint of death due to breast cancer) according to high or low cytoplasmic SK1 expression. (C) Tissue microarray analysiswith anti-SK1 antibody showing low and high expression of SK1 in breast cancer tumors from the HER1�/HER2�/HER3�/ER� patient cohortand Kaplan-Meier survival curves demonstrating cumulative disease-specific survival differences (endpoint of death due to breast cancer) accordingto HER family expression when the cohort has been subdivided into patients with tumors with high or low expression of SK1 and comparingHER1�/HER2�/HER3� tumors versus HER1�/HER2�/HER3� tumors. (D) Kaplan-Meier survival curves demonstrating cumulative disease-free survival differences (endpoint of breast cancer recurrence) according to HER1�/HER2� versus HER1�/HER2� expression when the cohorthas been subdivided into patients with tumors expressing high or low cytoplasmic SK1 expression. (E) Kaplan-Meier survival curves demonstratingcumulative disease-specific survival differences (endpoint of death due to breast cancer) and cumulative disease-free survival differences (endpointof breast cancer recurrence) according to a high versus low HER1-3/SK1 ratio for individual patients. P values represent log-rank testing of thedifference in cumulative disease-specific survival.


unpublished data). Indeed, we have found that SK1 inhibitoror overexpression of hSK1 has no effect on basal, S1P- orEGF-stimulated activation of the ERK-1/2 pathway in theseER� breast cancer cells.

Additional evidence supporting an important role for SK1 in

regulating formation of a migratory phenotype was demon-strated by results showing that siRNA knockdown of SK1 ab-lated the S1P-stimulated formation of actin containing mem-brane ruffles/lamellipodia and restored the formation of focalactin adhesion contacts. The role of S1P3/SK1 in terms ofregulating ERK-1/2 activation, redistribution of actin intomembrane ruffles/lamellipodia/nucleus, and the formation of amigratory phenotype in MCF-7 Neo cells is summarized inFig. 12.

Oncogene tolerance is associated with a HER2-dependentincrease in the expression and activity of SK1 in MCF-7 HER2cells. The promoter of SK1 contains Sp1 and AP2 transcrip-tional sites (23). Sp1 is constitutively associated with the SK1promoter and interacts cooperatively with AP2 that is activatedby agents acting via ERK-1/2 (23, 33). Direct evidence thatSK1 can induce tolerance to HER2 was the finding that thesiRNA knockdown of SK1 resulted in increased expression ofHER2. Thus, HER2 increases SK1 expression and SK1 limitsHER2 expression. The mechanism by which HER2 and SK1regulate each other’s expression is currently being studied inour laboratory.

We also demonstrated that siRNA knockdown of SK1 in-creased p65 PAK1 expression and basal ERK-1/2 activation inMCF-7 HER2 cells. p65 PAK1 is an upstream regulator ofERK-1/2 (5), and this explains how the increased p65 PAK1expression in cells in which SK1 is knocked down with siRNAleads to higher basal ERK-1/2 activation. Indeed, enforcedoverexpression of p65 PAK1 increased basal and S1P-stimu-lated activation of ERK-1/2 in MCF-7 HER2 cells. Therefore,by reducing p65 PAK1 expression, SK1 prevents an MCF-7HER2 cell migratory phenotype from being formed in re-sponse to S1P. Indeed, the migratory phenotype can be res-cued by siRNA knockdown of SK1 expression. The importantquestion is then “what is the major biochemical change interms of producing a migratory phenotype when SK1 isknocked down in MCF-7 HER2 cells?” We consider that thekey finding here is that p65 PAK1 expression is very substan-tially increased by siRNA knockdown of SK1, and this restores

FIG. 11. Effect of SK1 overexpression on HER2, p65 PAK1, andphosphorylated ERK-1/2 levels and migration of BT474 cells. BT474 cellswere transfected with plasmid construct encoding hSK1 or hG82D SK1 orvector. (A) Western blots showing the effect of hSK1 or hG82D SK1 onthe expression of HER2 (185 kDa and a smaller immunoreactive 175-kDafragment), p65 PAK1 (and p70), and phosphorylated ERK-1/2. Alsoshown are Western blots probed with anti-FLAG tag antibody showingthe expression of FLAG-tagged hSK1 and FLAG-tagged G82D hSK1.Anti-ERK-2 antibody was used to establish equal protein loading. Eachlane of the Western blots represents a separate cell sample, some of whichhave been run on the two separate Western blots, such that ERK-2loading in all of the lanes is equal. (B) Histogram showing the effect ofG82D hSK1 compared to vector on the migration of BT474 cells inscratch injury assays. The results are from two random views of threescratches for each treatment.

FIG. 12. Role of S1P3 and SK1 in regulating redistribution of phos-phorylated ERK-1/2 and actin into membrane ruffles/lamellipodia toproduce a migratory phenotype in MCF-7 Neo cells.


sensitivity to S1P in terms of producing a migratory phenotypein MCF-7 HER2 cells. This is supported by results showingthat the enforced overexpression of recombinant p65 PAK1rescues the actin migratory phenotype of MCF-7 HER2 cells inresponse to S1P.

We have also demonstrated that SK1 might regulate theinterconversion of p65 PAK1 to p70 and p59 forms in MCF-7HER2 cells. The p70 form might be derived from p65 PAK1 byan SK1-induced hyperactivation of p65 PAK1 which results inits mobility shift on SDS-PAGE. In addition, the increase inHER2 expression in MCF-7 HER2 cells in which SK1 has beenknocked down with siRNA might contribute to the hyperacti-vation of p65 PAK1 via a phosphorylation-dependent mecha-nism. Recently, Spiegel and coworkers showed that S1P canactivate PAK1 directly (19). Similar binding of lipids such asdiaclyglycerol to protein kinase C (PKC) can lead to an imme-diate activation of PKC, followed by its proteolytic degrada-tion. Therefore, it is possible that intracellular S1P, formed bySK1, binds to and hyperactivates p65 PAK1 to form p70, fol-lowed by the degradation of p70 to p59, and that this occurs inMCF-7 HER2 cells because of the HER2-dependent increasein SK1 expression. SK1 also functions in a metabolic pathwaythat is linked to the production of other bioactive lipids, andthe role of these in the posttranslational regulation of HER2/p65 PAK1/ERK-1/2 therefore requires further investigation.The SK1-dependent regulation of HER2/p65 PAK1/ERK-1/2observed in MCF-7 HER2 cells was also recapitulated inBT474 cells. We were also able to detect p70 in BT474 cellswhich, in common with MCF-7 HER2 cells, is posttranslation-ally regulated by SK1. However, we were unable to detect p59in these cells, suggesting that this form might be further pro-cessed and eliminated in BT474 cells.

Notably, basal ERK-1/2 activation is also increased bysiRNA knockdown of SK1 and additive stimulation with S1P isproduced in MCF-7 HER2 cells. The higher levels of ERK-1/2activation might therefore contribute to the S1P-stimulatedformation of the migratory phenotype in SK1-depleted MCF-7HER2 cells. The role of SK1 in inducing a refractory migratoryphenotype in response to S1P in MCF-7 HER2 cells is shownin Fig. 13. Therefore, whereas SK1 is normally required forformation of a migratory phenotype in MCF-7 Neo cell, theHER2-dependent increase in SK1 expression appears to de-sensitize the formation of a migratory phenotype in responseto S1P in MCF-7 HER2 cells.

The predicted significance of this novel mechanistic pathwayinvolving SK1 is demonstrated by the finding that highly ex-pressed SK1 in primary ER� breast tumors is associated withreduced breast cancer-specific patient survival and increasedtamoxifen resistance. However, SK1 functionality is altered inER�/HER2� breast cancer patients, such that SK1 is protec-tive against mortality and reduces tamoxifen resistance. Thecomplex interplay between HER2/SK1 demonstrated in thepresent study and HER1/SK1 expression in MCF-7 cells sug-gests that the key factor in inducing tolerance and protectionagainst HER-dependent malignancy is the ratio of HER1-3 toSK1. Indeed, we have also demonstrated that patients with alow HER1-3/SK1 ratio, have a better prognosis than thosepatients with a high HER1-3/SK1 ratio. The implication ofthese findings is that patients with tumors that have relativelylow HER1-3 expression and lower SK1 expression might have

a poorer prognosis compared to patients with relatively highHER1-3 and higher SK1 expression. Therefore, the dual func-tion of SK1 in breast cancer might be of significant therapeuticimportance in terms of defining new strategies for targetingSK1 in the treatment of ER�/HER2� breast cancer.

ACKNOWLEDGMENTS

This study was funded by a CRUK grant (C23158/A7536) to S.P. andN.J.P., the Royal College of Surgeons (joint Glasgow and Edinburgh)and GRI endowment fund (to J.E.), and National Institutes of Healthgrant RO1 HL79396 to V.N.

REFERENCES

1. Alderton, F., S. Rakhit, K. C. Kong, T. Palmer, B. Sambi, S. Pyne, and N. J.Pyne. 2001. Tethering of the platelet-derived growth factor beta receptor toG-protein-coupled receptors: a novel platform for integrative signaling bythese receptor classes in mammalian cells. J. Biol. Chem. 276:28578–28585.

2. Borg, A., A. K. Tandon, and H. Sigurdsson. 1990. HER2/neu amplificationpredicts poor survival in node-positive breast cancer. Cancer Res. 50:4332–4337.

3. Campbell, E., E. McDuff, O. Tatarov, S. Tovey, V. Brunton, T. G. Cooke, andJ. Edward. 2008. Phosphorylated c-Src in the nucleus is associated withimproved patient outcome in ER positive breast cancer. Br. J. Cancer 99:1769–1774.

4. Cannings, E., T. Kirkegaard, S. M. Tovey, B. Dunne, T. G. Cooke, and J. M.Bartlett. 2007. Bad expression predicts outcome in patients treated withtamoxifen. Cancer Res. Treat. 102:173–179.

5. Cheng, C., X. Kong, H. Wang, H. Gan, Y. Hao, W. Zou, J. Wu, Y. Chi, J.Yang, Y. Hong, K. Chen, and J. Gu. 2009. Trihydrophobin 1 interacts withPAK1 and regulates ERK/MAPK activation and cell migration. J. Biol.Chem. 284:8786–8796.

6. Chun, J., E. G. Goetzl, T. Hla, Y. Igarashi, K. R. Lynch, M. Moolenaar, S.Pyne, and G. Tigyi. 2002. IUPHAR XXXIV lysophospholipid receptor no-menclature. Pharmacol. Rev. 54:265–269.

7. Delon, C., M. Manifava, E. Wood, D. Thompson, S. Krugmann, S. Pyne, andN. T. Ktistakis. 2004. Sphingosine kinase 1 is an intracellular effector ofphosphatidic acid. J. Biol. Chem. 279:44763–44774.

8. Doll, F., J. Pfeilschifter, and A. Huwiler. 2005. The epidermal growth factorstimulates sphingosine kinase-1 expression and activity in the human mam-mary carcinoma cell MCF-7. Biochim. Biophys. Acta 1738:72–81.

9. Edwards, D. C., L. C. Sanders, G. M. Bokoch, and G. N. Gill. 1999. Activa-tion of LIM-kinase by PAK1 couples Rac/Cdc42 to actin cytoskeleton dy-namics. Nat. Cell Biol. 1:253–259.

10. Eswaran, J., M. Soundararajan, and S. Knapp. 2009. Targeting group IIPAKs in cancer and metastasis. Cancer Metastasis Rev. 28:209–217.

11. French, K. J., R. S. Schrecengost, B. D. Lee, Y. Zhuang, S. N. Smith, J. L. M.Eberly, J. K. Jun, and C. D. Smith. 2003. Discovery and evaluation ofinhibitors of human sphingosine kinase. Cancer Res. 63:5962–5969.

12. Goetzl, E. J., H. Dolezalova, Y. Kong, and L. Zeng. 1999. Dual mechanism

FIG. 13. Role of SK1 in inducing a refractory migratory phenotypein response to S1P in MCF-7 HER2 cells.


for lysophospholipid induction of proliferation of human breast cancer cells.Cancer Res. 59:4732–4737.

13. Graus-Porta, D., R. R. Beerli, J. M. Daly, and N. E. Hynes. 1997. ErbB-2, thepreferred heterodimerization partner of all ErbB receptors, is a mediator oflateral signaling. EMBO J. 16:1647–1655.

14. Huang, C., K. Jacobson, and M. D. Schaller. 2004. MAP kinases and cellmigration. J. Cell Sci. 117:4619–4628.

15. Huwiler, A., F. Doll, S. Ren, S. Klawitter, A. Greening, I. Romer, S. Bubnova,L. Reinsberg, and J. Pfeilschifter. 2006. Histamine increases sphingosinekinase-1 expression and activity in the human arterial endothelial cell lineEA.hy 926 by a PKC�-dependent mechanism. Biochim. Biophys. Acta 1761:367–376.

16. Inagaki, Y., P. Y. Li, A. Wada, S. Mitsutake, and Y. Igarashi. 2003. Identi-fication of functional nuclear export sequences in human sphingosine kinase1. Biochem. Biophys. Res. Commun. 311:168–173.

17. Koide, Y., T. Hasegawa, A. Takahashi, A. Endo, N. Mochizuki, M. Naka-gawa, and A. Nishida. 2002. Development of novel EDG3 antagonists usinga 3D database search and their structure-activity relationships. J. Med.Chem. 45:4629–4638.

18. Kirkegaard, T., J. Edwards, S. Tovey, L. M. McGlynn, S. N. Krishna, R.Mukherjee, L. Tam, A. F. Munro, D. Dunne, and J. M. Bartlett. 2006.Observer variation in immunohistochemical analysis of protein expression,time for a change? Histopathology 48:787–794.

19. Maceyka, M., S. E. Alvarez, S. Milstein, and S. Spiegel. 2008. Filamin A linkssphingosine kinase 1 and sphingosine-1-phosphate receptor 1 at lamellipodiato orchestrate cell migration. Mol. Cell. Biol. 28:5687–5697.

20. McCall, P., C. J. Witton, K. V. Nielsen, and J. Edwards. 2008. Is PTEN lossassociated with clinical outcome measures in human prostate cancer? Br. J.Cancer 99:1296–1301.

21. McCall, P., L. K. Gemmell, R. Mukherjee, J. M. S. Bartlett, and J. Edwards.2008. Phosphorylation of the androgen receptor is associated with reducedsurvival in hormone refractory prostate cancer patients. Br. J. Cancer 98:1094–1101.

22. McGlynn, L. M., T. Kirkegaard, T. J. Edwards, S. Tovey, D. Cameron, C.Twelves, J. M. Bartlett, and T. G. Cooke. 2009. Ras/Raf-1/MAPK pathwaymediates response to tamoxifen but not chemotherapy in breast cancerpatients. Clin. Cancer Res. 15:1487–1495.

23. Nakade, Y., Y. Banno, K. T. Koizumi, K. Hagiwara, S. Sobue, M. Koda, M.Suzuki, T. Kojima, A. Takagi, H. Asano, Y. Nozawa, and T. Murate. 2003.Regulation of sphingosine kinase 1 gene expression by protein kinase C inhuman leukaemia cell line, MEG-O1. Biochim. Biophys. Acta 1635:104–116.

24. Nava, V. E., P. Hobson, S. Murthy, S. Milstien, and S. Spiegel. 2002. Sphin-gosine kinase type 1 promotes estrogen-dependent tumorigenesis of breastcancer MCF-7 cells. Exp. Cell Res. 281:115–127.

25. Ohmori, T., Y. Yatomi, M. Osado, F. Kazama, T. Takafuta, H. Ikeda, and Y.Ozaki. 2003. Sphingosine 1-phosphate induces contraction of coronary arterysmooth muscle cells via S1P2. Cardiovasc. Res. 58:170–177.

26. Okamoto, H., N. Takuwa, T. Yokomizo, N. Sugimoto, S. Sakurada, H. Shige-matsu, and Y. Takuwa. 2000. Inhibitory regulation of Rac activation, mem-brane ruffling and cell migration by the G-protein coupled S1P receptorEDG5 but not EDG1 or EDG3. Mol. Cell. Biol. 20:9247–9261.

27. Pitson, S. M., P. A. Moretti, J. R. Zebol, H. E. Lynn, P. Xia, M. A. Vadas, andB. W. Wattenberg. 2003. Activation of sphingosine kinase 1 by ERK1/2-mediated phosphorylation. EMBO J. 22:5491–5500.

28. Pyne, S., S.-C. Lee, J. S. Long, and N. J. Pyne. 2009. Role of sphingosinekinases and lipid phosphate phosphatases in regulating spatial sphingosine1-phosphate signaling in health and disease. Cell. Signal. 21:14–21.

29. Sarkar, S., M. Maceyka, N. C. Hait, S. W. Paugh, H. Sankala, S. Milstien,

and S. Spiegel. 2005. Sphingosine kinase 1 is required for migration, prolif-eration and survival of MCF-7 human breast cancer cells. FEBS Lett. 579:5313–5317.

30. Schechter, A. L., M. C. Hung, L. Vaidyanathan, R. A. Weinberg, T. L.Yang-Feng, U. Francke, A. Ullrich, and L. Coussens. 1985. The neu gene:and erbB homologues gene distinct from an unlinked from the gene encod-ing the EGF receptor. Science 4717:976–978.

31. Sells, M. A., U. G. Knaus, S. Bagrodia, D. M. Ambrose, and G. M. Bokoch.1997. Human p21 activated protein kinase 1 (PAK1) regulates actin organi-sation in mammalian cells. Curr. Biol. 7:202–210.

32. Takuwa, Y. 2002. Subtype-specific differential regulation of Rho family Gproteins and cell migration by the Edg family sphingosine-1-phosphate re-ceptors. Biochim. Biophys. Acta 1582:112–120.

33. Trisciuoglio, D., A. Iervolino, A. Candiloro, G. Fibbi, M. Fanciulli, and U.Zangemeister-Wittke. 2004. bcl-2 induction of urokinase plasminogen acti-vator receptor expression in human cancer cells through Sp1 activation:involvement of ERK1/ERK2 activity. J. Biol. Chem. 279:6737–6745.

34. Tovey, S., B. Dunne, C. J. Witton, A. Forsyth, T. G. Cooke, and J. M. Bartlett.2005. Can. molecular markers predict when to implement treatment witharomatase inhibitors in invasive breast cancer? Clin. Cancer Res. 11:4835–4842.

35. Vadas, M., P. Xia, G. McCaughan, and J. Gamble. 2008. The role of sphin-gosine kinase-1 in cancer: oncogene or non-oncogene addiction. Biochim.Biophys. Acta 1781:442–447.

36. Viala, E., and J. Pouyssegur. 2004. Regulation of tumour cell motility byERK-mitogen activated protein kinases. Ann. N. Y. Acad. Sci. 1030:208–218.

37. Wang, S. E., I. Shin, F. Y. Wu, D. B. Friedman, and C. L. Arteaga. 2006.HER2/Neu (ErbB2) signaling to Rac1-PAK1 is temporally and spatiallymodulated by transforming growth factor beta. Cancer Res. 66:9591–9600.

38. Waters, C., J. Long, I. Gorshkova, Y. Fujiwara, M. Connell, K. E. Belmonte,G. Tigyi, V. Natarajan, S. Pyne, and N. J. Pyne. 2006. Cell migration acti-vated by platelet-derived growth factor receptor is blocked by an inverseagonist of the sphingosine 1-phosphate receptor-1. FASEB. J. 20:509–511.

39. Weinstein, I. B. 2002. Cancer: addiction to oncogenes—the Achilles heel ofcancer. Science 297:63–64.

40. Whenham, N., V. D’Hondt, and M. J. Piccart. 2008. HER2 positive breastcancer: from trastuzumab to innovatory anti-HER2 strategies. Clin. BreastCancer 8:38–49.

41. Xia, P., J. R. Gamble, L. Wang, S. M. Pitson, P. A. Moretti, B. W. Watten-berg, R. J. D’Andrea, and M. A. Vadas. 2000. An oncogenic role of sphin-gosine kinase. Curr. Biol. 10:1527–1530.

42. Xia, P., L. Wang, P. A. Moretti, N. Albanese, F. Chai, S. M. Pitson, R. J.D’Andrea, J. R. Gamble, and M. A. Vadas. 2002. Sphingosine kinase inter-acts with TRAF2 and dissects tumor necrosis factor-alpha signaling. J. Biol.Chem. 277:7996–8003.

43. Yamashita, H., J. Kitayama, D. Shida, H. Yamaguchi, K. Mori, M. Osada, S.Aoki, Y. Yatomi,. Y. Takuwa, and H. Nagawa. 2006. Sphingosine 1-phosphatereceptor expression profile in human gastric cancer cells: differential regu-lation on the migration and proliferation. J. Surg. Res. 130:80–87.

44. Yen, L., N. Benlimame, Z. R. Nie, D. Xiao, T. Wang, A. E. Al Moustafa, H.Esumi, J. Milanini, N. E. Hynes, G. Pages, and M. A. Alaoui-Jamali. 2002.Differential regulation of tumour angiogenesis by distinct ErbB homo- andheterodimers. Mol. Biol. Cell 13:4029–4044.

45. Zheng, L., J. Q. Ren, H. Li, Z. L. Kong, and H. G. Zhu. 2004. Down-regulation of wild-type p53 protein by HER2/neu mediated PI3K pathwayactivation in human breast cancer cells: its effect on cell proliferation andimplication for therapy. Cell Res. 14:497–506.


Sphingosine Kinase 1 Induces Tolerance to Human Epidermal Growth Factor Receptor 2 and Prevents Formation of a Migratory Phenotype in Response to Sphingosine 1-Phosphate in Estrogen

Documents