Sphingosine-1-phosphate receptor type 1 regulates glioma cell proliferation and correlates with patient survival

Sphingosine-1-phosphate receptor type 1 regulates glioma cellproliferation and correlates with patient survival

Yuya Yoshida1, Mitsutoshi Nakada1, Naotoshi Sugimoto2, Tomoya Harada1, Yasuhiko Hayashi1, Daisuke Kita1,

Naoyuki Uchiyama1, Yutaka Hayashi1, Akihiro Yachie3, Yoh Takuwa2 and Jun-ichiro Hamada1

1 Department of Neurosurgery, Graduate School of Medical Science, Kanazawa University, Kanazawa, Japan2 Department of Physiology, Graduate School of Medical Science, Kanazawa University, Kanazawa, Japan3 Department of Pediatrics, Graduate School of Medical Science, Kanazawa University, Kanazawa, Japan

Sphingosine-1-phosphate (S1P) is a bioactive lipid that signals through a family of G protein-coupled receptors consisting of

5 members termed S1P1–5, and it regulates cellular proliferation, migration and survival. We investigated the expression and

role of S1P receptors in glioma. Human glioma expressed S1P1, S1P2, S1P3, and S1P5 by quantitative real-time PCR analysis.

Expression of the S1P1 was significantly lower in glioblastoma than in the normal brain (p < 0.01) and diffuse astrocytoma

(p < 0.05). Immunoblotting showed that normal brain expressed more S1P1 protein than did glioblastoma.

Immunohistochemistry showed that S1P1 was localized predominantly in the astrocytes in the normal brain, but no staining

was observed in glioblastoma. Downregulation of S1P1 expression correlated with poor survival of patients with glioblastoma

(p < 0.05). S1P1 small interfering RNA promoted cell proliferation in high-expressor glioma cell lines (T98G, G112). Cell

proliferation was promoted by the pertussis toxin, which deactivates Gi/o type of G proteins; the S1P1 is exclusively coupled

to these proteins. Forced expression of the S1P1 in low-expressor cell lines (U87, U251) resulted in decreased cell growth and

led to suppressed tumor growth in transplanted gliomas in vivo. Furthermore, we found a significant association between the

S1P1 expression and early growth response-1, a transcriptional factor that exhibits tumor suppression in glioblastoma cells

(p < 0.05). These data indicate that the downregulation of S1P1 expression enhances the malignancy of glioblastoma by

increasing cell proliferation and correlates with the shorter survival of patients with glioblastoma.

Glioblastoma is the most common type of brain tumor andis a highly malignant tumor exhibiting aggressive invasive

growth,1 leading to a median life expectancy of only 10–12months after diagnosis.2 It is essential to understand the mo-lecular regulation of glioblastoma cell growth and invasion todevelop effective molecular-based therapies.

Sphingosine-1-phosphate (S1P) is a bioactive lipid thatregulates cellular proliferation, migration and survival in awide variety of tissues and cell types.3,4 S1P signals bothintracellularly as a second messenger5 and through 5 G pro-tein-coupled receptors (GPCRs) at the cell surface, which aretermed S1P1–5.

6,7 Each S1P receptor displays a unique tissueexpression pattern and is coupled to a distinct set of G pro-teins (Gi/o, Gq and G12/13), leading to the activation of recep-tor-specific intracellular signaling pathways.6 S1P affects theproliferation, motility and invasiveness of malignant cells andcan mediate proliferative8,9 or antiproliferative10,11 behaviorin cancer cells. The various effects of S1P may be attributedto the diversity of its receptors. Some studies have shownthat S1P, whose level is high in brain12 and glioma cells,13

plays a role in the growth and invasiveness of glioblastomacells,14–16 and these effects are at least partially mediatedthrough its GPCRs because both responses are sensitive topertussis toxin (PTX),14 which specifically inhibits signalingthrough Gi/o, which is a major family of G proteins. Glioblas-toma cell lines15 and tissues17 commonly express 3 membersof the S1P receptor family: S1P1, S1P2 and S1P3. S1P5 isexpressed in oligodendrocytes.18 S1P4, which is primarilyexpressed in cells of hematopoietic origin,19 has not been

Key words: glioblastoma, S1P receptors, S1P1 receptor, proliferation,

survival

Abbreviations: Egr-1: early growth response-1; GPCRs: G protein-

coupled receptors; OS: overall survival; PFS: progression-free

survival; PI3K: phosphatidylinositol 30 kinase; PTEN: phosphataseand tensin homologue deleted in chromosome 10; PTX: pertussis

toxin; QRT-PCR: quantitative real-time PCR; siRNA: small

interfering RNA; S1P: sphingosine-1-phosphate

Additional Supporting Information may be found in the online

version of this article

Grant sponsors: The Japanese Ministry of Education, Science,

Sports, Technology and Culture (Grants-in-aid for young scientists

research); Grant number: B-19790992; Grant sponsors: Japan Brain

Foundation, Foundation for Promotion of Cancer Research,

Hokkoku Cancer Research Foundation

DOI: 10.1002/ijc.24933

History: Received 30 Jun 2009; Accepted 17 Sep 2009; Online 6 Oct

2009

Correspondence to: Mitsutoshi Nakada, Department of

Neurosurgery, Graduate School of Medical Science, Kanazawa

University, 13-1 Takara-machi, Kanazawa 920-8641, Japan, Tel:

þ81-76-265-2384, Fax: þ81-76-234-4262,

E-mail: [email protected]

Can

cerCellBiology

Int. J. Cancer: 126, 2341–2352 (2010) VC 2009 UICC

International Journal of Cancer

IJC

detected in glioma cells. Therefore, the individual influenceof each receptor subtype on the S1P regulation of glioblas-toma cells behavior may depend on the amounts of the indi-vidual S1P receptors expressed. Currently, however, little in-formation is available concerning the roles played by the S1Preceptors in this disease.

In this study, we analyzed the expression of S1P receptors inhuman specimens at various histological grades of glioma andthe role of the S1P1 receptor in glioma cell lines. S1P1 receptorexpression is downregulated in glioblastoma, particularly inthose patients with a poor survival. S1P1 receptor signaling nega-tively controls cell proliferation either by knockdown or by over-expression of the S1P1 receptor in glioma cell lines. Further-more, expression of the S1P1 receptor correlated with that ofearly growth response-1 (Egr-1), which is an essential transcrip-tional factor considered to be a central regulator in tumor cellproliferation20–24 through phosphatase and tensin homologuedeleted in chromosome 10 (PTEN) in glioblastoma. Theseresults suggest that S1P1 receptor signaling plays an importantrole in the malignant behavior of human glioma.

Material and MethodsClinical samples and histology

Following an institutional review board-approved protocol,fresh human brain tumor tissues were obtained from 58patients who underwent therapeutic removal of gliomas.Non-neoplastic control brain tissues were identified from themargins of the tumors. The histoanalyses are based on therevised World Health Organization criteria.25 The 58 gliomasconsisted of 15 diffuse astrocytomas, 8 anaplastic astrocyto-mas, and 35 glioblastomas. All tumor tissues were obtainedat primary resection, and none of the patients had been sub-jected to chemotherapy or radiation therapy before resection.For assessment of the overall survival (OS) and progression-free survival (PFS),26 we made attempts at obtaining informa-tion of every 35 patient with glioblastoma, and data of theOS (35/35) and PFS (29/35) were collected.

Quantitative real-time PCR

Quantitative real-time PCR (QRT-PCR) was performed in aLightCycler (Roche Diagnostics, Indianapolis, IN) asdescribed previously.27 PCR was done with the following pri-mers: S1P1 receptor (NM_001400): sense, 50-AAATTCCACCGACCCATGTA-30; antisense, 50-AGTTATTGCTCCCGTTGTGG-30 (amplicon size, 250 bp); Another primer set ofS1P1 receptor was also used: sense, 50-TGCTCTCCATCGTCATTCTG-30; antisense, 50-CCAGGAAGTACTCCGCTCTG-30 (amplicon size, 252 bp); Egr-1 (NM_001964): sense, 50-TGACCGCAGAGTCTTTTCCT-30; antisense, 50-TGGGTTGGTCATGCTCACTA-30 (amplicon size, 203 bp); and b-actin(NM_001101): sense, 50-CTACAATGAGCTGCGTGTGGC-30; antisense, 50-CAGGTCCAGACGCAGGATGGC-30 (ampli-con size, 271 bp). The LightCycler analysis software was usedto analyze the PCR data as described previously.27

Antibodies and reagents

Anti-S1P1 receptor polyclonal antibody was purchased fromAffinity BioReagents (Golden, CO). Antiphosphorylatedextracellular signal-regulated kinase (ERK) 1/2 mitogen-acti-vated protein kinase (MAPK) (Thr 202/Tyr 204) antibody,anti-PTEN monoclonal antibody, anti-Akt polyclonal anti-body and antiphosphorylated Akt polyclonal antibody wereobtained from Cell Signaling Technology (Beverly, MA).Anti-b-actin monoclonal antibody and PTX were fromSigma-Aldrich (St. Louis, MO). S1P was purchased from Bio-mol (Plymouth Meeting, PA).

Western blot analysis

Western blot analyses were done as described previously.28

All antibodies were used at a dilution of 1:1000.

Immunohistochemistry

Immunohistochemistry was done using avidin–biotin immu-noperoxidase technique as described previously.28 Anti-S1P1receptor antibody was used at a dilution of 1:200. Nonim-mune rabbit IgG was used as negative controls.

Cell culture

Human astrocytoma cell lines U87, U251, T98G (AmericanType Culture Collection, Manassas, VA) and G11229 weremaintained in DME supplemented with 10% foetal bovine se-rum (FBS) at 37�C.

Expression plasmid and cell transfection

An expression plasmid for S1P1 receptor was described previ-ously.30 Transient transfection into U87 and U251 cells wasdone using Effectene (Qiagen, Valencia, CA). The overex-pression by transient transfection using Effectene lasted for atleast 96 hr as described previously.28 Stable transfectants ofU87 cells were selected after 1–2 wk in a medium containingG418 (800 lg/ml) and were used only for intracranial trans-plantation into mice. Cells transfected with empty plasmidvector were used as controls.

Silencing of endogenous S1P1 receptor with small

interfering RNA

Purified, duplexed small interfering RNAs (siRNAs) for S1P1receptor and control luciferase were purchased from Qiagen.The 2 target sequences of human S1P1 receptor (Genbankaccession number NM_001400) were (2055-2075 bp) 50-ATGATCGATCATCTATAGCAA-30 and (406-426 bp) 50-CTCGGTCTCTGACTACGTCAA-30. The sequences weredesigned to be unique relative to the sequences of other S1Preceptor members. Twenty nanomolar siRNA was transfectedinto cells cultured in 60-mm diameter dishes using Lipofect-amine 2000 (Invitrogen, Carlsbad, CA). Transfected cellswere cultured for 48 hr before use. The effects of siRNAlasted for at least 96 hr as described previously.28

Can

cerCellBiology

2342 S1P1 receptor in glioma proliferation


Cell proliferation assay

The Alamar Blue assay (Biosource, Camarillo, CA) was usedto assess proliferation as described previously.28 In brief,1000 cells of each population were seeded in wells of 96-wellplastic plates in 200 ll of culture medium supplemented with0.5% FBS. The plates were incubated for 4 hr at 37�C, andAlamar Blue was added in a volume of 20 ll (10% of totalvolume) to the cells and incubated. The plate was read on afluorescence plate reader (excitation, 530 nm; emission, 590nm) at 24, 48, 72, and 96 hr. Averages of the absorbance val-ues were calculated and plotted. To investigate the influenceof S1P on cell proliferation, cells were serum starved for 24hr and seeded in the proliferation assay format. The cellswere then treated with various concentrations of S1P. S1Pand PTX was readded every 24 hr during the course of theexperiment.

Cell migration and invasion assay

Chemotactic migration and invasion of cells were measuredin a modified Boyden chamber (Neuroprobe, Gaithersburg,MD) as described previously.31

In vivo tumor formation assay

Following an institutional review board-approved protocol,intracranial transplantation of glioma cells into mice wasdone. Female 7-week-old mice nonobese diabetic/severe com-bined immunodeficiency disease (NOD/SCID) were used asdescribed previously.32 All mice were purchased from CharlesRiver Lavoratories, Osaka, Japan. Mice were anaesthetizedwith an intraperitoneal injection of pentobarbital (60-70 mg/kg body weight). A burr hole was made in the skull 3 mmlateral to the bregma using a drill, and 1 � 105 U87 cells sta-bly transfected with empty plasmid vector or S1P1 receptorvector in 2 ll of PBS were stereotactically injected over 4min at a depth of 3 mm below the dura mater. This proce-dure reproducibly results in tumor growth and obvious tu-mor-related symptoms about 3 wk after intracerebral injec-tion. Five mice in each group were euthanized on day 21after tumor injection. The brain tissue was embedded in par-affin and then cut into 5-lm serial coronal sections. Tissuesections were stained by the standard hematoxylin and eosintechnique. The volume of the tumor was evaluated by meas-uring the long (a) and short (b) axes in the coronal sectionshowing the maximal area of each tumor. The approximatevolume of the tumor (V) was calculated according to theformula, V (mm3) ¼ a � b2/2.

Statistics

Statistical analyses were done using the v2 test and the two-tailed Mann-Whitney U test. The Kaplan-Meier method wasused for survival analysis. Glioblastoma cases were dividedinto the lower half vs. the upper half of S1P1 receptor expres-sion level as determined by QRT-PCR, as described previ-ously.17 p < 0.05 was considered significant.

ResultsDownregulation of S1P1 receptor expression in

glioblastoma cells

To evaluate the potential role of S1P receptors in the malig-nant behavior of human gliomas, the expression level of S1Preceptors was evaluated as a function of tumor grade in 58surgical specimens. The mRNA levels of S1P receptors inhuman normal brain and gliomas were evaluated by QRT-PCR using b-actin mRNA as the internal reference for nor-malization. QRT-PCR analysis demonstrated that thesetumors express S1P1, S1P2, S1P3 and S1P5 but not S1P4 (datanot shown). The S1P1 receptor mRNA levels (S1P1 receptormRNA: b-actin mRNA ratios) were significantly lower in glio-blastoma tissues (mean 6 SD, 0.124 6 0.121; n ¼ 35) than innormal brain tissues (0.329 6 0.295; p < 0.01; n ¼ 13) anddiffuse astrocytoma tissues (0.217 6 0.175; p < 0.05; n ¼ 15;Fig. 1a). Furthermore, the expression levels of the S1P1 recep-tor decreased in glioblastoma tissues obtained far from the ne-crosis of the tumors (0.42- to 0.60-fold) relative to those innormal brain tissues resected from the margins of the tumorsin 4 cases of glioblastoma (Fig. 1b). These results were consist-ent with experimental results obtained using another primerset for the S1P1 receptor. There was no significant correlationbetween the expression of S1P2, S1P3 and S1P5 and the histo-logical grade of the human gliomas (data not shown).

Immunoblotting with a specific antibody against S1P1 re-ceptor was performed using 24 surgical specimens. Anapproximately 42-kDa multiple band of S1P1 receptor, asdescribed previously,33,34 was detected in normal brain andin cases of diffuse astrocytoma, but in cases of glioblastoma,only faint bands were observed (Fig. 1c). These results wereconsistent with those of the QRT-PCR analysis.

Cells expressing the S1P1 receptor in normal brain and glio-blastoma specimens were identified using immunohistochemis-try. The S1P1 receptor was predominantly immunolocalized tothe astrocytes, neurons, and endothelial cells in normal braintissue (Fig. 2). Astrocytes were confirmed by immunopositivityfor glial fibrillary acidic protein staining (data not shown). Nostaining was observed in neoplastic astrocytes far from thenecrosis of the tumors in glioblastoma specimens (Fig. 2).

Expression levels of the S1P1 receptor were highly variableamong individual samples of the normal brains and tumorsby QRT-PCR analysis. Gliomas are characterized by necrosis,and thus there are many dead cells present in the center ofthe tumor. In addition, bigger tumors have more vessels. Insuch a case, it is possible that the gliomas have data spreadof the expression levels of the S1P1 receptor. However, theexpression of S1P1 receptor was significantly lower in glio-blastoma than in the normal brain, using all of QRT-PCR,immunoblotting and immunohistochemistry.

S1P1 receptor expression correlates with the survival

of patients with glioblastoma

We were also interested in examining the correlation of theexpression levels of S1P receptors with the clinical outcome

Can

cerCellBiology


Yoshida et al. 2343

Figure 1. Expression of S1P1 receptor in various human gliomas. (a) Relative mRNA expression levels of S1P1 receptor gene (S1P1 receptor

mRNA:b-actin mRNA ratios) in normal brains (NB), diffuse astrocytomas (DA), anaplastic astrocytomas (AA) and glioblastomas (GB) were

analyzed by QRT-PCR. Each mRNA level is expressed as a proportion of the highest mRNA level of S1P1 receptor, which was given a value

of 1. Horizontal bars indicate mean values. *p < 0.05, **p < 0.01. (b) S1P1 receptor expression in glioblastomas (GB) and normal brains

(NB) resected from the margins of the tumors in 4 glioblastoma cases (GB1 to GB4). (c) The total cell lysate from normal brains (Lanes

1–3, 9–11, 17–19, NB), diffuse astrocytomas (Lane 4, 12, 20, DA), anaplastic astrocytomas (Lane 5, 13, 21, AA) and glioblastomas (Lanes

6–8, 14–16, 22–24, GB) were immunoblotted with anti-S1P1 receptor antibody. The membrane was stripped and reprobed with anti-b-actin

antibody. Twenty-four representative samples.

Can

cerCellBiology



in glioblastoma cases. Interestingly, survival analysis of 35patients with glioblastoma showed a significant correlationbetween S1P1 receptor expression and PFS and OS (Fig. 3).PFS and OS analyses revealed that S1P1 receptor downregula-tion correlated with poor survival in patients with glioblas-toma (p ¼ 0.03; n ¼ 29 and p ¼ 0.02; n ¼ 35, respectively).Patients with glioblastoma whose tumors expressed high lev-els of S1P1 receptors had a PFS rate of 42.8% and an OS rateof 77.7% at 1 year, whereas those with low S1P1 receptorexpression levels had PFS and OS rates of only 13.3 and36.8%, respectively, at 1 year. There was no significant corre-lation between the expressions of the other S1P receptorswith survival in patients with glioblastoma (data not shown).

Inhibition of the S1P1 receptor by siRNA promotes glioma

cell proliferation

S1P1 receptor expression decreased significantly as the tumorgrade increased, and its downregulation correlated with thepoor survival of patients with glioblastoma. Therefore, weinvestigated the functional effects of altered S1P1 receptorexpression as a possible negative stimulant of glioma cell pro-liferation, migration, and invasion. We used 2 differentsiRNA sequences specifically to silence endogenous S1P1 re-ceptor expression in T98G and G112 cells, which exhibit

high S1P1 receptor expression. No obvious phenotypicchanges were observed by manipulating the S1P1 receptorgene in glioma cell lines (data not shown). Inhibition of S1P1receptor mRNA by both S1P1 receptor-1 and receptor-2siRNA was approximately 90%; this did not affect the expres-sion levels of other S1P receptor family members (data notshown). Figure 4a illustrates the reduction of S1P1 receptorprotein expression in S1P1 receptor-siRNA-transfected T98Gand G112 cells compared with its expression in untransfectedand luciferase-siRNA-transfected controls. First, to examinethe functional effects of S1P1 receptor in cell proliferation, weassessed the changes in cell number throughout time by theAlamar Blue assay. T98G and G112 cells transfected withS1P1 receptor siRNA significantly promoted cell proliferation(Fig. 4b).

Inhibition of the Gi/o protein by PTX affects glioma cell

proliferation

Furthermore, proliferation assays were performed to confirmthe effects of S1P1 receptor signaling on proliferation of gli-oma cells by using PTX. S1P1 receptor is coupled exclusivelyto the Gi protein and activates a variety of signaling path-ways.30 T98G and G112 cells were pretreated with S1P withPTX or with S1P without PTX or in the absence of S1P and

Figure 2. Immunolocalization of S1P1 receptor in glioblastoma and normal brain. Paraffin sections were immunostained with antibody

against S1P1 receptor (Panels a, b and c) or nonimmune rabbit IgG (Panel d). Note that S1P1 receptor is immunolocalized in the astrocytes

(Panel b, arrows), neurons (Panel c, arrows), and endothelial cells (Panel c, arrow head) in the normal brain, whereas no staining is

observed in the glioblastoma cells with S1P1 receptor (Panel a) and normal brain with nonimmune IgG (Panel d). Hematoxylin counterstain.

Scale bars: 50 lm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Can

cerCellBiology


Yoshida et al. 2345

PTX for 24 hr before the start of the experiments. Activationof ERK by agents that act through GPCRs is often mediatedby Gi protein and is therefore sensitive to PTX.30,35–37 Toconfirm the effects of PTX on glioma cells, we demonstratedthat ERK activation was decreased by PTX (Fig. 4c). Consist-ent with our data (Fig. 4b), in the presence of S1P ligand, gli-oma cells pretreated with PTX showed increased cell prolifer-ation compared with cells that were not pretreated (Fig. 4d).These data suggest that the Gi protein is strongly associatedwith glioma cell proliferation and that S1P1 receptor nega-tively regulates glioma cell proliferation through a Gi protein.

Overexpression of S1P1 receptor suppresses glioma cell

proliferation

Furthermore, to verify that S1P1 receptor could provide apotential therapeutic target, it was transiently transfected intoU87 and U251 cells, which express this receptor at low levels(Fig. 5a). No obvious phenotypic change was observed bymanipulating the S1P1 receptor gene in glioma cell lines(data not shown). Expectedly, S1P1 receptor overexpressionin U87 and U251 cells by transient transfection resulted indecreased cell proliferation compared with that in mocktransfectants (Fig. 5b). Decreased cell proliferation was fur-ther observed in S1P1 receptor-overexpressing U87 and U251cells in the presence of high concentrations of S1P ligand.These data support the specific role of S1P1 receptor signal-ing in the inhibition of cell proliferation.

Functional assays were also performed to confirm theeffects of altered S1P1 receptor expression on the migrationand invasion of glioma cells. In these assays, no significantchange was observed by either overexpression or knockdown

of the S1P1 receptor in glioma cell lines (SupplementaryData).

Overexpression of S1P1 receptor suppresses glioma

growth in vivo

Furthermore, to evaluate whether altered S1P1 receptorexpression could affect tumor growth in vivo, we used immu-nocompromised mice transplanted intracerebrally with U87cells stably transfected with empty plasmid vector or S1P1 re-ceptor vector. Tumor growth was significantly delayed in theanimals that were transplanted with S1P1 receptor-overex-pressing cells, whereas relatively larger tumors were observedin the other group (mean tumor volume 6 SD, 3.2 6 2.2mm3 vs. 12.0 6 5.8 mm3, respectively; p < 0.05) (Fig. 5d).No obvious change in glioma invasion in vivo was observedat the border of the tumor. The results demonstrated thatS1P1 receptor signaling could affect tumor growth in intrace-rebral gliomas.

S1P1 receptor expression correlates with Egr-1 expressionand is associated with the phosphatidylinositol 30 kinase(PI3K) and PTEN signaling pathways in glioblastoma cells

As a possible mechanism that S1P1 receptor regulates gli-oma cell proliferation, we investigated the correlationbetween the expressions of S1P1 receptor and Egr-1, whoseexpression was mediated by S1P1 receptor signaling in rat C6glioma cells.38,39 In those cells, S1P induced expression ofEgr-1,38 and S1P1 receptor may be the main receptor media-ting the stimulation of Egr-1.39 The levels of Egr-1 mRNA inglioma cell lines and glioblastoma tissue samples were exam-ined by QRT-PCR, and the association of Egr-1 mRNA levelswith the S1P1 receptor mRNA levels was analyzed. In S1P1

Figure 3. Kaplan–Meier survival curves for patients with glioblastoma. Tumors were divided into the lower half vs. the upper half of S1P1

receptor expression level as determined by QRT-PCR. The difference is statistically significant according to the PFS (p ¼ 0.03, n ¼ 29) and

the OS analysis (p ¼ 0.02, n ¼ 35).

Can

cerCellBiology



Figure 4. Growth curve analysis of glioma cell lines knocking down S1P1 receptor and pretreated or nonpretreated by PTX. (a) Extracts of

T98G or G112 cells treated by siRNA for either S1P1 receptor-1 (S1P1-1) or S1P1 receptor-2 (S1P1-2) or control luciferase (ctrl) were

immunoblotted with anti-S1P1 receptor or anti-b-actin antibodies. (b) T98G or G112 cells treated by siRNA for S1P1 receptor were grown

using 8 wells for each cell in the presence of 100 nM S1P. S1P was readded every 24 hr during the course of the experiment. (c) Extracts

of T98G or G112 cells pretreated with or without PTX (100 ng/ml) in the presence of 100 nM S1P for 24 hr before the experiments were

immunoblotted with antiphosphorylated ERK or anti-b-actin antibodies. (d) T98G or G112 cells pretreated or nonpretreated by PTX were

grown in the presence or absence of 100 nM S1P. S1P and PTX were readded every 24 hr during the course of the experiment. The plate

was read on a fluorescence plate reader at the indicated time points. Bars, SD. *p < 0.05 vs. control luciferase (b) or S1P without PTX (d).

Results shown in (b and d) are typical of at least 2 replicate experiments.

Can

cerCellBiology


Yoshida et al. 2347

receptor-overexpressing U87 and U251 cells, Egr-1 expressionlevels were 1.64-fold and 1.99-fold higher, respectively, com-pared with those in the control vector-transfected cells in thepresence of S1P ligand (Fig. 6a). Thus, Egr-1 transcriptionwas induced by S1P1 receptor signaling enhanced by S1Pligands. Furthermore, the expression levels of Egr-1 plotted

against those of S1P1 receptor in each glioblastoma sample (n¼ 35) showed a significant correlation (r ¼ 0.546, p < 0.05)(Fig. 6b). Furthermore, there is increasing evidence that Egr-1 regulates PTEN phosphate and expression.40 Therefore, wehypothesized that the S1P1 receptor expressed in glioblastomacells would regulate the PI3K and PTEN signaling pathways.

Figure 5. Growth curve analysis of glioma cell lines overexpressing S1P1 receptor and S1P1 receptor-overexpressing intracerebral U87

glioma growth of mice. (a and c) Extracts of U87 or U251 cells transiently transfected with S1P1 receptor vector (S1P1) or empty plasmid

vector (Mock) (a) or stably transfected U87 cells (c) were immunoblotted with anti-S1P1 receptor or anti-b-actin antibodies. (b) U87 or U251

cells transiently transfected with S1P1 receptor vector or empty plasmid vector were grown using 8 wells for each cell in the presence of

the indicated concentration of S1P. S1P was readded every 24 hr during the course of the experiment. The plate was read on a

fluorescence plate reader at the indicated time points. Bars, SD. *p < 0.05, **p < 0.01 vs. mock. Results shown in (b) are typical of at

least 2 replicate experiments. (d) Gross images and coronal sections from representative brains of immunocompromised mice implanted

with U87 cells stably transfected with S1P1 receptor vector (S1P1) or empty plasmid vector (Mock) are displayed. (e) The mean tumor

volume 6 SD in each group are shown. Columns, the mean tumor volume on day 21 after tumor injection; bars, SD. *p < 0.05. [Color

figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Can

cerCellBiology



Figure 6. Expression of S1P1 receptor and its correlation with expression of Egr-1, PTEN and phosphorylation of Akt in glioblastoma (GB) tissues

and glioma cell lines. (a) U87 or U251 cells transfected with S1P1 receptor vector (S1P1) or empty plasmid vector (Mock) were incubated with

100 nM S1P for 24 hr. Then, S1P1 receptor was detected by Western blotting and Egr-1 expression was determined by QRT-PCR. Each mRNA

level is expressed as a proportion of the highest mRNA level of Egr-1, which was given a value of 1. (b) S1P1 receptor and Egr-1 expressions

were determined by QRT-PCR in glioblastomas. Each mRNA level is expressed as a proportion of the highest mRNA level of S1P1 receptor or Egr-

1, which was given a value of 1. Direct correlations are observed with correlation coefficients of r ¼ 0.55 (p < 0.05). (c) The total cell lysates

from glioblastomas were immunoblotted with anti-PTEN or antiphosphorylated Akt or anti-b-actin antibodies. S1P1 receptor expression was

determined by QRT-PCR. Ten representative samples. (d) Correlation of expression of S1P1 receptor with expression of PTEN and phosphorylation

ratio of Akt in glioblastomas. Signals were quantified by densitometry. Relative protein expression levels of PTEN (PTEN protein/b-actin protein

ratios) and phosphorylation ratio levels of Akt (phosphorylated Akt protein:total-Akt protein ratios) were determined. Each protein level and

phosphorylation ratio level is expressed as a proportion of the highest protein level of PTEN or phosphorylation ratio level of Akt, which was

given a value of 1. Horizontal bars, mean values. *p < 0.05. The differences are statistically significant (p < 0.05, n ¼ 35).

Can

cerCellBiology

To confirm this, we analyzed PTEN, Akt and phosphorylatedAkt at the protein levels in glioblastoma tissues. Westernblotting by a specific antibody against PTEN was performed.The membranes were stripped, and a phosphorylated Aktantibody and a total Akt antibody were applied. The PTENband was identified at 54 kDa and phosphorylated Akt, at 60kDa (Fig. 6c). Signals were quantified by densitometry, andlevels of PTEN were evaluated using b-actin as the internalcontrol for normalization (PTEN/b-actin protein ratios). Inaddition, the Akt phosphorylation ratio (phosphorylatedAkt:total Akt ratio) was determined. We found a strong asso-ciation between S1P1 receptor and PTEN expression and Aktphosphorylation in glioblastoma tissues. The protein level ofPTEN directly correlated with the level of S1P1 receptor (p <

0.05; n ¼ 35; Fig. 6d), and low S1P1 receptor expression cor-related with the high phosphorylation ratio of Akt (p < 0.05;n ¼ 35; Fig. 6d). These results suggest that S1P1 receptor sig-naling through the Egr-1, PI3K and PTEN signaling path-ways plays an important role in the malignant behavior ofhuman gliomas.

DiscussionThe expression of S1P1 receptor was downregulated in glio-blastoma cells, particularly in those patients with a poor sur-vival. S1P1 receptor siRNA promoted cell proliferation inhigh-expressor glioma cell lines (T98G and G112). Cell pro-liferation was promoted by PTX, which inhibits Gi/o proteins.Forced expression of S1P1 receptor in low-expressor cell lines(U87 and U251) resulted in decreased cell growth concomi-tant with the activation of S1P1 receptor, and in vivo, forcedexpression resulted in suppressed tumor growth in trans-planted gliomas. Furthermore, expression of the S1P1 recep-tor was correlated with that of Egr-1 and PTEN and Aktphosphorylation in glioblastomas. The evidence presentedhere supports the role of S1P1 receptors in glioma cell prolif-eration in vitro and in vivo.

S1P is a phospholipid mediator present in high levels inthe brain, and it can induce diverse cellular responses such asproliferation, differentiation and migration in various celltypes.3,4,12 S1P acts as an important intercellular agonistfunctioning through S1P receptors. Currently, the roles ofS1P signaling through S1P receptors in mammalian develop-ment and normal tissues are being progressively determined.It would seem that S1P1 but not S1P2 or S1P3 is essential fornumerous normal physiological functions.41,42 In our study,immunohistochemistry showed that the S1P1 receptor wasimmunolocalized predominantly to astrocytes, neurons andendothelial cells in the normal brain, suggesting that S1P1 isessential for normal physiological functions in the normalbrain.

S1P has been reported to be a common mitogen for gli-oma cells.14 In contrast to the case of normal tissues, manymalignant cells have been shown to express S1P2 or S1P3 butnot S1P1.

9–11,43 Glioma cells were also observed to expressS1P2 and S1P3 receptors with considerably lower levels of

S1P1 receptor.15 In our study, QRT-PCR analysis and immu-noblotting showed that the expression levels of S1P1 receptordecreased significantly in glioblastoma and that S1P1 receptordownregulation correlated with the poor survival of patientswith glioblastoma. Dysregulation of S1P1 receptor expressionor function may play an important role in the malignantbehavior of human gliomas.

The increased cell proliferation in response to knockdownof the S1P1 receptor and the decreased cell proliferation inresponse to the forced expression of the S1P1 receptor wereunexpected because a recent study has demonstrated thatS1P1 promotes glioma cell proliferation.16 In this study byYoung et al., experiments manipulating gene expression dem-onstrated that S1P1, S1P2 and S1P3, which are commonlyexpressed in glioma cells, positively contributed to cell prolif-eration. The reason for this discrepancy between our resultsand those of Young et al. is unclear. However, differencesbetween the cell lines used and method for cell proliferationassays could be a probable cause for this discrepancy. Previ-ous studies have shown that S1P1 receptor negatively regu-lates cell proliferation in pheochromocytoma cells44 and Tcells45; these results are consistent with our data. In ourstudy, S1P1 receptor expression was downregulated in glio-blastoma, particularly patients with glioblastoma with poorsurvival. Thus, because there is a correlation between S1P1receptor downregulation and the in vitro and in vivo clinicaldata, the signaling pathway via S1P1 may contribute nega-tively to glioma cell proliferation.

Previous studies have shown that the S1P1 receptor cou-ples with Gi/o proteins and is linked to the concurrent activa-tion of multiple effector pathways, including Ras-ERK, Ca2þ

mobilization with the activation of phospholipase C, and in-hibition of adenylate cyclase.30 The ERK cascade is wellknown to promote glioma cell proliferation.46 However, ourstudy indicated that PTX increased glioma cell proliferation,concomitant with the inactivation of ERK. The ERK inactiva-tion with other downstream signaling cascades of Gi couldpossibly result in subtle changes in the proliferative signal incells. A recent study demonstrated that Egr-1 expression wassignificantly associated with enhanced patient survival andwas an independent good prognostic factor in multivariateanalysis in high-grade astrocytoma.20 In our study, the levelof S1P1 receptor significantly correlated with the level of Egr-1 in glioblastoma tissues, and these expressions were con-comitantly downregulated in glioblastoma patients with poorsurvival. Furthermore, the expression of S1P1 was correlatedwith that of PTEN and the downregulation of Akt phospho-rylation. These results suggest that dysregulation of Egr-1expression through S1P1 receptor signaling and the PTENsignaling pathway are the underlying factors influencing gli-oma proliferation regardless of ERK inactivation.

In our migration and invasion assays, no significantchange was observed by either knockdown or overexpressionof the S1P1 receptor in glioma cell lines, whereas a previousstudy has demonstrated that S1P1 receptor exerts a

Can

cerCellBiology



stimulatory effect for migration in glioma cell lines.16 How-ever, a previous study has shown that S1P1 receptor stimula-tion did not induce migration in thyroid cancer cells.10 Wehave previously shown that S1P2 prevents S1P-stimulatedmotility31,47,48 and that glioma cell lines expressing S1P2 re-ceptor in abundance did not exhibit enhanced motility inresponse to S1P.15 It is possible that the signaling of S1P2 butnot S1P1 is involved in cell migration and invasion.

Because rat glioma cells have been shown to make largeamounts of S1P and secrete it extracellularly,13 it is possiblethat autocrine signaling by S1P affects glioma cell biology.Our human gliomas of various histological grades alsoexpressed the enzymes that form S1P, sphingosine kinase-1and -2 (data not shown); however, no significant correlationwith histological grade or with patient survival was found(data not shown).

In conclusion, we demonstrated that the expression levelsof S1P1 receptor decrease significantly with an increase in the

histological tumor grade in human glioma tissues. We alsodemonstrated that S1P1 receptor downregulation enhancesthe malignancy of these tumors by increasing cell prolifera-tion and that it strongly correlates with the shorter survivalof patients with glioblastoma. Further studies are necessary toverify our statistical model and to determine the role of S1P1receptors in tumor growth. Measurement of the S1P1 recep-tor expression levels may be useful as a prognostic indicator.Understanding the function and regulation of S1P1 receptorsignaling could aid in the development of effective therapiesfor patients with glioma.

AcknowledgementsGrants-in-aid for young scientists research from the Japanese Ministry ofEducation, Science, Sports, Technology and Culture (B-19790992 to M.N.),a grant from Japan Brain Foundation (to M.N.), Foundation for Promotionof Cancer Research (to M.N.) and Hokkoku Cancer Research Foundation(to M.N., D.K., Y.H. J-i. H.).

References

1. Lefranc F, Sadeghi N, Camby I, Metens T,Dewitte O, Kiss R. Present and potentialfuture issues in glioblastoma treatment.Expert Rev Anticancer Ther 2006;6:719–32.

2. Sathornsumetee S, Rich JN. New treatmentstrategies for malignant gliomas. ExpertRev Anticancer Ther 2006;6:1087–104.

3. Spiegel S, Milstien S. Sphingosine-1-phosphate: an enigmatic signalling lipid.Nat Rev Mol Cell Biol 2003;4:397–407.

4. Hla T. Physiological and pathologicalactions of sphingosine 1-phosphate. SeminCell Dev Biol 2004;15:513–20.

5. Spiegel S, Milstien S. Exogenous andintracellularly generated sphingosine 1-phosphate can regulate cellular processesby divergent pathways. Biochem Soc Trans2003;31:1216–19.

6. Takuwa Y, Takuwa N, Sugimoto N. TheEdg family G protein-coupled receptorsfor lysophospholipids: their signalingproperties and biological activities.J Biochem 2002;131:767–71.

7. Young N, Van Brocklyn JR. Signaltransduction of sphingosine-1-phosphate Gprotein-coupled receptors. ScientificWorldJournal 2006;6:946–66.

8. Wang F, Nohara K, Olivera A, ThompsonEW, Spiegel S. Involvement of focaladhesion kinase in inhibition of motility ofhuman breast cancer cells by sphingosine1-phosphate. Exp Cell Res 1999;247:17–28.

9. Goetzl EJ, Dolezalova H, Kong Y, Zeng L.Dual mechanisms for lysophospholipidinduction of proliferation of human breastcarcinoma cells. Cancer Res 1999;59:4732–7.

10. Balthasar S, Samulin J, Ahlgren H, BergelinN, Lundqvist M, Toescu EC, Eggo MC,Tornquist K. Sphingosine 1-phosphatereceptor expression profile and regulation

of migration in human thyroid cancercells. Biochem J 2006;398:547–56.

11. Yamashita H, Kitayama J, Shida D,Yamaguchi H, Mori K, Osada M, Aoki S,Yatomi Y, Takuwa Y, Nagawa H.Sphingosine 1-phosphate receptorexpression profile in human gastric cancercells: differential regulation on themigration and proliferation. J Surg Res2006;130:80–7.

12. Edsall LC, Spiegel S. Enzymaticmeasurement of sphingosine-1-phosphate.Anal Biochem 1999;272:80–6.

13. Edsall LC, Cuvillier O, Twitty S, Spiegel S,Milstien S. Sphingosine kinase expressionregulates apoptosis and caspase activationin PC12 cells. J Neurochem 2001;76:1573–84.

14. Van Brocklyn JR, Letterle C, Snyder P,Prior T. Sphingosine-1-phosphatestimulates human glioma cell proliferationthrough Gi-coupled receptors: role of ERKMAP kinase and phosphatidylinositol 3-kinase beta. Cancer Lett 2002;181:195–204.

15. Van Brocklyn JR, Young N, Roof R.Sphingosine-1-phosphate stimulatesmotility and invasiveness of humanglioblastoma multiforme cells. Cancer Lett2003;199:53–60.

16. Young N, Van Brocklyn JR. Roles ofsphingosine-1-phosphate (S1P) receptors inmalignant behavior of glioma cells.Differential effects of. S1P2 on cellmigration and invasivenes. Exp Cell Res2007;313:1615–27.

17. Van Brocklyn JR, Jackson CA, Pearl DK,Kotur MS, Snyder PJ, Prior TW.Sphingosine kinase-1 expression correlateswith poor survival of patients withglioblastoma multiforme. Roles ofsphingosine kinase isoforms in growth of

glioblastoma cell lines. J Neuropathol ExpNeurol 2005;64:695–705.

18. Jaillard C, Harrison S, Stankoff B, AigrotMS, Calver AR, Duddy G, Walsh FS,Pangalos MN, Arimura N, Kaibuchi K,Zalc B, Lubetzki C. Edg8/S1P5: anoligodendroglial receptor with dualfunction on process retraction and cellsurvival. J Neurosci 2005;25:1459–69.

19. Graler MH, Bernhardt G, Lipp M. EDG6, anovel G-protein-coupled receptor related toreceptors for bioactive lysophospholipids, isspecifically expressed in lymphoid tissue.Genomics 1998;53:164–9.

20. Mittelbronn M, Harter P, Warth A,Lupescu A, Schilbach K, Vollmann H,Capper D, Goeppert B, Frei K, BertalanffyH, Weller M, Meyermann R, et al. EGR-1is regulated by N-Methyl-D-Aspartate-receptor stimulation and associated withpatient survival in human high gradeAstrocytomas. Brain Pathol 2009;19:195–204.

21. Levin WJ, Press MF, Gaynor RB,Sukhatme VP, Boone TC, Reissmann PT,Figlin RA, Holmes EC, Souza LM, SlamonDJ. Expression patterns of immediate earlytranscription factors in human non-smallcell lung cancer. The Lung Cancer StudyGroup. Oncogene 1995;11:1261–9.

22. Huang RP, Fan Y, de Belle I, Niemeyer C,Gottardis MM, Mercola D, Adamson ED.Decreased Egr-1 expression in human,mouse and rat mammary cells and tissuescorrelates with tumor formation. Int JCancer 1997;72:102–9.

23. Liu C, Yao J, Mercola D, Adamson E. Thetranscription factor EGR-1 directlytransactivates the fibronectin gene andenhances attachment of human

Can

cerCellBiology


Yoshida et al. 2351

glioblastoma cell line U251. J Biol Chem2000;275:20315–23.

24. Calogero A, Arcella A, De Gregorio G,Porcellini A, Mercola D, Liu C, Lombari V,Zani M, Giannini G, Gagliardi FM, CarusoR, Gulino A, et al. The early growthresponse gene EGR-1 behaves as asuppressor gene that is down-regulatedindependent of ARF/Mdm2 but not p53alterations in fresh human gliomas. ClinCancer Res 2001;7:2788–96.

25. Louis DN, Ohgaki H, Wiestler OD,Cavenee WK, Burger PC, Jouvet A,Scheithauer BW, Kleihues P. The 2007WHO classification of tumours of thecentral nervous system. Acta Neuropathol2007;114:97–109.

26. Narayana A, Kelly P, Golfinos J, Parker E,Johnson G, Knopp E, Zagzag D, Fischer I,Raza S, Medabalmi P, Eagan P, GruberML. Antiangiogenic therapy usingbevacizumab in recurrent high-gradeglioma: impact on local control and patientsurvival. J Neurosurg 2009;110:173–80.

27. Nakada M, Drake KL, Nakada S, Niska JA,Berens ME. Ephrin-B3 ligand promotesglioma invasion through activation ofRac1. Cancer Res 2006;66:8492–500.

28. Nakada M, Niska JA, Tran NL,McDonough WS, Berens ME. EphB2/R-Rassignaling regulates glioma cell adhesion,growth, and invasion. Am J Pathol 2005;167:565–76.

29. Giese A, Rief MD, Loo MA, Berens ME.Determinants of human astrocytomamigration. Cancer Res 1994;54:3897–904.

30. Okamoto H, Takuwa N, Gonda K, OkazakiH, Chang K, Yatomi Y, Shigematsu H,Takuwa Y. EDG1 is a functionalsphingosine-1-phosphate receptor that islinked via a Gi/o to multiple signalingpathways, including phospholipase Cactivation, Ca2þ mobilization, Ras-mitogen-activated protein kinase activation,and adenylate cyclase inhibition. J BiolChem 1998;273:27104–10.

31. Okamoto H, Takuwa N, Yokomizo T,Sugimoto N, Sakurada S, Shigematsu H,Takuwa Y. Inhibitory regulation of Racactivation, membrane ruffling, and cellmigration by the G protein-coupledsphingosine-1-phosphate receptor EDG5

but not EDG1 or EDG3. Mol Cell Biol2000;20:9247–61.

32. Natsume A, Wakabayashi T, Tsujimura K,Shimato S, Ito M, Kuzushima K, Kondo Y,Sekido Y, Kawatsura H, Narita Y, YoshidaJ. The DNA demethylating agent 5-aza-20-deoxycytidine activates NY-ESO-1antigenicity in orthotopic human glioma.Int J Cancer 2008;122:2542–53.

33. Igarashi J, Erwin PA, Dantas AP, Chen H,Michel T. VEGF induces S1P1 receptors inendothelial cells: implications for cross-talkbetween sphingolipid and growth factorreceptors. Proc Natl Acad Sci U S A 2003;100:10664–9.

34. Hu W, Huang J, Mahavadi S, Li F, MurthyKS. Lentiviral siRNA silencing ofsphingosine-1-phosphate receptors S1P1and S1P2 in smooth muscle. BiochemBiophys Res Commun 2006;343:1038–44.

35. Lee MJ, Van Brocklyn JR, Thangada S, LiuCH, Hand AR, Menzeleev R, Spiegel S, HlaT. Sphingosine-1-phosphateas a ligand forthe G protein-coupled receptor EDG-1.Science 1998;279:1552–5.

36. Van Brocklyn JR, Graler MH, BernhardtG, Hobson JP, Lipp M, Spiegel S.Sphingosine-1-phosphate is a ligand for theG protein-coupled receptor EDG-6. Blood2000;95:2624–9.

37. Gonda K, Okamoto H, Takuwa N, YatomiY, Okazaki H, Sakurai T, Kimura S, SillardR, Harii K, Takuwa Y. The novelsphingosine-1-phosphate receptor AGR16is coupled via pertussis toxin-sensitive and-insensitive G-proteins to multiplesignalling pathways. Biochem J 1999;337:67–75.

38. Sato K, Ishikawa K, Ui M, Okajima F.Sphingosine 1-phosphate inducesexpression of early growth response-1 andfibroblast growth factor-2 throughmechanism involving extracellular signal-regulated kinase in astroglial cells. BrainRes Mol Brain Res 1999;74:182–9.

39. Sato K, Ui M, Okajima F. Differential rolesof Edg-1 and Edg-5, sphingosine 1-phosphate receptors, in the signalingpathways in C6 glioma cells. Brain Res MolBrain Res 2000;85:151–60.

40. Virolle T, Adamson ED, Baron V, Birle D,Mercola D, Mustelin T, de Belle I. The

Egr-1 transcription factor directly activatesPTEN during irradiation-inducedsignalling. Nat Cell Biol 2001;3:1124–8.

41. Liu Y, Wada R, Yamashita T, Mi Y, DengCX, Hobson JP, Rosenfeldt HM, Nava VE,Chae SS, Lee MJ, Liu CH, Hla T, et al.Edg-1, the G proteincoupled receptor forsphingosine-1-phosphate, is essential forvascular maturation. J Clin Invest 2000;106:951–61.

42. Ishii I, Friedman B, Ye X, Kawamura S,McGiffert C, Contos JJ, Kingsbury MA,Zhang G, Brown JH, Chun J. Selective lossof sphingosine 1-phosphate signaling withno obvious phenotypic abnormality inmice lacking its G protein-coupledreceptor, LP (B3)/EDG-3. J Biol Chem2001;276:33697–704.

43. Yamaguchi H, Kitayama J, Takuwa N,Arikawa K, Inoki I, Takehara K, NagawaH, Takuwa Y. Sphingosine-1-phosphatereceptor subtype-specific positive andnegative regulation of Rac andhaematogenous metastasis of melanomacells. Biochem J 2003;374:715–22.

44. Molderings GJ, Gothert M, von KugelgenI. Characterization of an antiproliferativeeffect of imidazoline receptor ligands onPC12 cells. Pharmacol Rep 2007;59:789–94.

45. Jin Y, Knudsen E, Wang L, Bryceson Y,Damaj B, Gessani S, Maghazachi AA.Sphingosine 1-phosphate is a novelinhibitor of T-cell proliferation. Blood2003;101:4909–15.

46. Belsey MJ, Davies AR, Witchel HJ,Kozlowski RZ. Inhibition of ERK and JNKdecreases both osmosensitive taurinerelease and cell proliferation in gliomacells. Neurochem Res 2007;32:1940–9.

47. Takuwa Y. Subtype-specific differentialregulation of Rho family G proteins andcell migration by the Edg familysphingosine-1-phosphate receptors. BiophysBiochim Acta 2002;l1582:112–20.

48. Sugimoto N, Takuwa N, Okamoto H,Sakurada S, Takuwa Y. Inhibitory andstimulatory regulation of Rac and cellmotility by the G12/13-Rho and Gipathways integrated downstream of asingle G protein-coupled sphingosine-1-phosphate receptor isoform. Mol Cell Biol2003;23:1534–45.

Can

cerCellBiology



Documents

Sphingosine-1-phosphate receptor type 1 regulates glioma cell proliferation and correlates with patient survival