Biochimica et Biophysica Acta 1833 (2013) 1006–1016

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbamcr

Sphingosine kinase 1 expression is downregulated duringdifferentiation of Friend cells due to decreased c-MYB

N. Mizutani a,1, M. Kobayashi a,1, S. Sobue b, M. Ichihara b, H. Ito a, K. Tanaka c, S. Iwaki c, S. Fujii c, Y. Ito d,K. Tamiya-Koizumi a, A. Takagi a, T. Kojima a, T. Naoe e, M. Suzuki f, M. Nakamura g, Y. Banno h,Y. Nozawa i, T. Murate a,⁎a Department of Pathophysiological Laboratory Science, Nagoya University Graduate School of Medicine, Nagoya, Japanb College of Life and Health Sciences, Chubu University, Kasugai, Japanc Department of Molecular and Cellular Pathophysiology and Therapeutics, Graduate School of Pharmaceutical Science, Nagoya City University, Nagoya, Japand Equipment Center for Research and Education, Nagoya University Graduate School of Medicine, Nagoya, Japane Department of Hematology and Oncology, Nagoya University Graduate School of Medicine, Nagoya, Japanf Division of Molecular Carcinogenesis, Nagoya University Graduate School of Medicine, Nagoya, Japang Department of Drug Information, Gifu Pharmaceutical University, Gifu, Japanh Department of Dermatology, Gifu Graduate School of Medicine, Gifu, Japani Tokai Gakuin University, Kakamigahara, Japan

Abbreviations: SPHK1, Sphingosine kinase 1; SPL, Spphosphate; HMBA, Hexamethylene bisacetamide; WT, Wtive; EMSA, Electrophoresis mobility shift assay; ChIP, Ch⁎ Corresponding author at: Department of Pathophy

Nagoya University Graduate School of Medicine, DaikNagoya, Aichi 461-8673, Japan. Tel./fax: +81 52 719 1

E-mail address: murate@met.nagoya-u.ac.jp (T. Mur1 These two contributed equally.

0167-4889/$ – see front matter © 2013 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.bbamcr.2013.01.001

a b s t r a c t

a r t i c l e i n f o

Article history:Received 7 April 2012Received in revised form 23 December 2012Accepted 2 January 2013Available online 14 January 2013

Keywords:Friend cellSPHK1c-MYBPromoter analysisChIPHMBA-induced differentiation

Sphingosine kinase 1 (SPHK1) overexpression in malignant cells has been reported. Mouse Friend cellsshowed higher SPHK1 but not SPHK2 expression compared with other mouse cell lines. A Sphk1 promoteranalysis demonstrated the region between −53 bp and the first exon as the minimal promoter. Further pro-moter truncation revealed the importance of a MYB-binding site. EMSA using this region as the probe demon-strated one band containing c-MYB protein, and its intensity decreased during erythroid differentiation withhexamethylane bisacetamide (HMBA), a potent inducer of erythroid differentiation of Friend cells. ChIP assayalso revealed in vivo binding of c-MYB. c-MYB overexpression and siRNA for c-Myb affected SPHK1 expression,confirming the important regulatory role of c-MYB in SPHK1 expression. HMBA reduced c-MYB expressionrapidly. Induced differentiation by HMBA caused a marked and rapid reduction of SPHK1 mRNA, proteinand enzyme activity leading to the rapid decrease of cellular sphingosine 1-phosphate level. Moreover, termi-nally differentiated cells did not resume SPHK1 expression. Compared with original Friend cells, stableoverexpression of wild-type SPHK1 showed higher cell proliferation, resistance to cell death by serum deple-tion. Interestingly, HMBA-induced differentiation of these cells was delayed but not completely suppressed. Incontrast, SPHK inhibitor and its siRNA inhibited cell growth and enhanced HMBA-induced differentiation sig-nificantly, suggesting that SPHK1 delayed HMBA-induced differentiation by its cell proliferation-promotingactivity. Effects of pertussis toxin, a G-protein-coupled receptor inhibitor, and S1P receptor antagonist on Friendcell growth and differentiationwere negligible, suggesting the importance of the intracellular SPHK1/S1P signal-ing in Friend cells.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Sphingolipids (SPLs) are complex lipids containing an amide-linkedfatty acid and a long-chain amino alcohol, sphingoid base. Recently,SPLs such as ceramide, sphingosine, sphingosine 1-phosphate (S1P),

hingolipid; S1P, Sphingosine 1-ild-type; DN, Dominant nega-

romatin immunoprecipitationsiological Laboratory Science,o-minami, 1-1-20 Higashi-ku,186.ate).

l rights reserved.

and ceramide 1-phosphate have been considered as intra- or inter-cellular signaling molecules in numerous biological processes [1].Spiegel and colleagues proposed a sphingolipid rheostat model, wherethe balance between ceramide and S1P determines the cell's fate [2,3].Ceramide acts as a signaling molecule, leading to cell responses suchas apoptosis, growth arrest, cell differentiation and inflammation,whereas S1P mostly works as an inhibitor of apoptosis and as a stimu-lator of cell proliferation.

S1P is produced by sphingosine kinases (SPHK) 1 and 2. Discovery ofS1P receptors in cell membrane helped elucidate the signaling pathwaydownstream of S1P/S1P receptors [4,5]. SPHK1 expression stimulatesproliferation and protects against apoptosis, and its overexpression isoncogenic in NIH3T3 cells [6]. On the contrary, overexpression of

1007N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

SPHK2 suppresses cell growth and induces apoptosis [7]. Thus, SPHKsplay an important role in cancer growth and survival [8]. Previous re-ports including ours have shown increased SPHK1 expression in variouscancer and leukemia cells [9–13]. Considering the role of sphingolipidmetabolic enzymes as a chemotherapy sensor [10,14], the elucidationand regulation of SPHK1 expression is clinically important.

The mouse erythroleukemia cell line caused by the Friend virus(Friend cells) has been extensively studied because of its suitabilityas a virus oncogenesis model [15,16] and also because of its prominentdifferentiation toward erythroid lineage by chemical inducers suchas DMSO or hexamethylene bisacetamide (HMBA) [17]. With HMBAtreatment, almost all Friend cells gradually ceased to proliferate andexhibited erythroid phenotypes such as high cellular hemoglobinwithin 4 or 5 days, although the rate of erythroid differentiation variesamong sublines.

A Friend virus complex contains a replication-defective spleenfocus-forming virus (SFFV) and a replication-competent Friend mu-rine leukemia virus (F-MuLV). gp55, a product of the SFFV env gene,interacts with the erythropoietin receptor and constitutively acti-vates its downstream signaling pathways, allowing proliferation ofvirus infected cells in the absence of erythropoietin [18]. The Friendvirus is usually found upstream of the PU.1 gene (Sfpi1), leading tooverexpression of the wild-type transcription factor, PU.1 [18,19].Similar to Friend cells, Spi-1/PU.1 transgenic mice develop multisteperythroleukemia [20]. Le Scolan et al. have reported that overex-pression of SPHK1 is an oncogenic event in mouse erythroleukemiacell lines developed from PU.1-transgenic mouse [21].

The sequential changes of proto-oncogenes (c-Myb, Fli-1, andSfpi1) and cell cycle regulators (CDK2, 4 and 6) during the chemicallyinduced differentiation process have been reported [22–25]. Thus,Friend cells have been utilized as an excellent model of leukemiadifferentiation therapy. However, it remains to be determined howSPHK1 overexpression is induced in Friend cells, how SPHK1 is regu-lated during this differentiation process, and whether SPHK level af-fects HMBA-induced differentiation. We approached these issues by(1) the analysis of SPHK1 transcriptional mechanism in Friend cells;(2) the analysis of the stably SPHK1- and SPHK2-overexpressingFriend cells; and (3) experiments using a chemical SPHK inhibitor,siRNA for Sphk1, and S1P receptor antagonists.

The present analysis indicated (1) c-MYB is themajor transcriptionfactor of the SPHK1 of Friend cells, (2) SPHK1 is rapidly decreased dur-ing HMBA-induced differentiation through reduced c-MYB expres-sion, and (3) SPHK1 expression level affects the HMBA-induceddifferentiation of Friend cells. The pathophysiological significance ofthese findings was discussed.

2. Materials and methods

2.1. Vectors and reagents

Wild-type (WT) expression vectors of mouse SPHK1 [21] andSPHK2 (as FLAG-tagged form) [26] were generous gifts of Dr. F.Moreau-Gachelin (Inserm U528, Institut Curie, Paris, France) andDr. T. Yoshimoto (Tokyo Medical University, Tokyo, Japan), respec-tively. WT- -Sphk1 cDNA inserts were obtained by PCR amplificationand reinserted into a MSCV-IRES-GFP vector, which was derivedfrom Dr. A. Katsumi (Hamamatsu University School of Medicine,Hamamatsu, Japan). The c-MYB expression vector and GST-c-MYBexpression vector [27] were the gifts of Dr. A. Tomita (Nagoya Uni-versity Graduate School of Medicine, Nagoya, Japan). c-Myb cDNAwas reinserted in pcDNA3.1 (Invitrogen, Carlsbad, CA, USA) and wasused for transfection. HMBA was purchased from Sigma (St. Louis,MO, USA). Sphingosine kinase inhibitor was obtained from Calbiochem(San Diego, CA). Pertussis toxin, VPC23019, and CAY10444 were fromBiomol (Plymouth Meeting, PA, USA).

2.2. Cell lines

Original Friend cells were described previously [28,29]. MouseIL-3 dependent cell lines, BaF/3 and 32D, were maintained in RPMI1640 medium supplemented with 10% FCS and 3% conditioned me-dium from WEHI 3B cells (WEHI-CM) as the source of IL-3. A mousemyoblastoid cell line, C2C12, and a mouse fibroblastic cell line,NIH3T3, were cultured in DMEM medium containing 5% FCS.

2.3. Establishment of stable transfectants

Stable transfectants of SPHK1 were established by sorting GFP-positive cells with the flow cytometer after retrovirus vector transfec-tion. After sorting, suitable clones after a single cell culture wereselected with Western blotting using our anti-SPHK1 antibody. Thisanti-mouse SPHK1 antibody was prepared by immunizing rabbitwith the C-terminal peptide of mouse SPHK1 we described earlier[30]. It was previously used in other mouse erythroleukemia celllines [21]. Stable SPHK2 transfectants were also established afterG418 selection. After limiting dilution, suitable clones were selectedbyWestern blotting using anti-FLAG antibody (Santa Cruz Biotechnol-ogy Inc., Santa Cruz, CA, USA), and then SPHK2 expression was alsoconfirmed using anti-SPHK2 antibody (Santa Cruz) as shown in Sup-plementary Fig. 5a. To establish c-MYB-transfectants, Friend cellswere transfected with the c-MYB expression vector, and selectedwith G418. Suitable clones were chosen with Western blotting usinganti-c-MYB antibody (Santa Cruz).

2.4. Hemoglobin staining, viable cell counting and commitment assay

Hemoglobin-positive cells were counted manually with liquid ben-zidine staining [28]. A commitment assay to detect the percentage ofcommitted cells to terminal erythroid differentiation was performedin triplicate as described previously [28]. The percentagewas calculatedas the mean±SD. Viable cells were counted in triplicate using trypan-blue dye exclusion method.

2.5. Quantitative RT-PCR

QRT-PCRwas performed according to the method described previ-ously [31] Primer sequences were as follows: mSPHK1; forward 5′-GGAGGAGGCAGAGATAAC-3′, reverse 5′-TTAGCCCATTCACCACTTCA-3′,mSPHK2; forward 5′-GTGCCCTCACGTGTCTCCT-3′, reverse 5′-GCATGGTTCTGTCGTTCTGTC-3′, and mouse α1-globin; forward 5′-GAGCTGAAGCCCTGGAAAGGATG-3′, reverse 5′-GACCTTCTTGCCGTGACCCTTG-3′, mouse β1-globin; forward 5′-CAACTTCAGAAACAGACATCATGG-′3, reverse 5′-TAGACAACCAGCCTGCC-′3, β-actin; forward 5′-CCGTGAAAAGATGACCCAGA-3′, reverse 5′-GTCTCCGGAGTCCATCACAA-3′.PCR conditions were 30 s initial denaturation followed by 40 cyclesof 96 °C for 5 s, and 60 °C for 31 s. A dissociation profile was generatedfor each run to verify specificity of the amplification. The PCR condi-tion for β-actin was the same as that of SPHK except that the appliedvolume was 10 times lower than that of SPHK1. The relative expres-sion of the SPHK and globin message was evaluated by calculatingthe ratio of SPHK or globin/β-actin.

2.6. SPHK enzyme activity

SPHK1 and SPHK2 enzyme activities were measured separately asdescribed before [32]. Briefly, SPHK1 activity was determined eitherin the presence of 0.25% Triton X-100 (SPHK1) or in 1 M KCl (SPHK2).SPHK1 activity was examined in the presence of 50 μM sphingosineand [γ-32P] ATP (10 μCi, 1 mM) containing 10 mMMgCl2 in 0.25% Tri-ton X-100, which inhibits SPHK2. SPHK2 activity was determined withsphingosine added as a complex with 4 mg/ml bovine serum albuminand [γ-32P] ATP containing 10 mM MgCl2 in the presence of 1 M KCl,

1008 N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

conditions in which SPHK2 activity is optimal and SPHK1 activity wasinhibited. Labeled S1P was extracted and separated by thin layer chro-matography on silica gel G60 with 1-butanol/acetic acid/H2O (3:1:1)as the solvent. The spot of S1P on the thin layer chromatography platewas visualized by autography and quantified using a densitometer.

2.7. Western blotting

Anti-FLAG, anti-Fli-1, anti-PU.1, anti-c-MYB, anti-GATA-1, anti-GATA-2, and anti-SPHK2 antibodies were purchased from Santa Cruz. In addi-tion to our anti-SPHK1 antibody, we used anti-SPHK1 antibody (#3297)from Cell Signaling Technology (Beverly,MA. USA) in the supplementa-ry experiments. Anti-β-actin antibody was purchased from Cytoskel-eton Inc. (Denver, CO, USA). Western blotting was performed usingImmobilon Western Chemiluminescent HRP Substrate (Millipore,Billerica, MA) as described by Sobue et al. [33].

2.8. Cellular sphingosine 1-phosphate measurement

Cellular sphingosine 1-phosphate (S1P) was measured accordingto liquid chromatography followed by tandem mass spectrometry(LC-MS/MS) as described before with minor modifications [12].

2.9. RNA ligase-mediated rapid amplification of 5′-cDNA ends (5′RACE)and promoter cloning of mouse Sphk1

Sphk1 transcription start site was determined by 5′RACE method(Invitrogen) using a primer set shown in Supplementary Table 1.For promoter cloning, we obtained approximately 1.7 kb of frag-ment covering the 5′ region of exon 1 of mouse Sphk1. This fragmentwas inserted into the BglII and HindIII sites of the pGL3 basic vec-tor (Promega, Madison, WI, USA) and named −1.7 kb/luc. We uti-lized the endogenous BglII site located around −1.7 kb from thefirst exon. Eight deletion mutants (named −322, −265, -241, −234,−164, −53, −21, and −4 bp/luc) and 4 mutated promoter vectors(−53 mut1, mut2, mut3, and −21 mut/luc) were constructed witheither restriction enzyme digestion or a PCR-based method. Primersets used for PCR amplification are shown in Supplementary Table 1.Vectors of −53 bp/luc or −4 bp/luc were constructed by digestionwith SmaI or StyI, respectively, followed by blunting and self-ligation.The sequence of truncated or mutated luciferase vectors was con-firmed. Promoter analysis was performed using these various lengthsof luciferase vectors. Six microliters of Fugene 6 (Roche, Mannheim,Germany), 2 μg of luciferase vector, and 1 μg of β-gal expression vectorwere incubated for 15 min in serum free medium (total volume of100 μl) according to the manufacturer's manual. Then, the mixturewas directly added to Friend cells (2.5×105/ml) cultured in serumcontaining culture medium. Assay was performed in triplicate. After24 h, cells were collected and luciferase activity and β-gal activitywere measured as described previously with minor modification [33].

2.10. Preparation of c-MYB protein

c-MYB proteinwas purified by digesting GST-c-MYB fusion proteinby Factor Xa according to the manufacturer's manual (AmershamBiosciences, Little Chalfont, Buckinghamshire, UK). c-Myb cDNA wasinserted into a pGEX-5X-2 expression vector (Amersham Biosci-ences). Fusion proteins were collected after stimulating GST-c-MYBprotein production in E. Coli, BL21, by 0.5 mM IPTG. Bacteria werelysed in modified NETN buffer (20 mM Tris–HCl (pH 8.0), 100 mMNaCl, 1 mM EDTA and 1% Triton X), then were sonicated. After centri-fugation, supernatants were incubated with MagneGST glutathioneparticles (MagneGST™ Pull-Down System, Promega) overnight at4 °C in NETN buffer (20 mM Tris–HCl (pH 8.0), 100 mM NaCl, 1 mMEDTA and 0.5% NP-40). GST-MYB proteinwas cleaved by bovine FactorXa (Funakoshi Co., Tokyo, Japan) and c-MYB was isolated.

2.11. Electrophoresis mobility shift assay (EMSA)

Nuclear extract was prepared from original Friend cells with orwithout HMBA treatment for 48 hours. EMSA was performed as de-scribed previously [31]. The labeled probe used was illustrated inFig. 2a. A cold competitor was used 10 or 100 times as much as the la-beled probe. Anti-c-MYB antibody was used at 2 or 6 μg/each sample.Isolated c-MYB fromGST-c-MYBproteinwas used in someexperiments.

2.12. Chromatin immunoprecipitation (ChIP) assay

ChIP assaywas performed as described previously [34]. Friend cellswere used for cross-linking with formaldehyde. For the immunopre-cipitation, normal rabbit IgG or anti-c-MYB antibody (500 ng/eachsample) was added to the same amount of nuclear protein and incu-bated at 4 °C overnight. Immunocomplexes were extracted, and thecross-linking was reversed by heating elutes at 65 °C overnight. Eluteswere then digested with proteinase K at 50 °C for 2 hours and ex-tracted with phenol/chloroform/isoamyl alcohol. DNA was purifiedby ethanol. The Sphk1 promoter region was PCR-amplified by primersas described in Supplementary Table 1.

2.13. Transient transfection of siRNA

Friend cells were transfected with siRNA using LipofectamineRNAiMAX (Invitrogen). siRNAs for cMyb and mouse Sphk1 were pur-chased from Sigma Genosys (Hokkaido, Japan), and their sequenceswere based on the previous report [31,35]. Nonspecific siRNAwas pur-chased from QIAGEN (Hilden, Germany).

2.14. Statistical analysis

Statistical significances were analyzed using Student's t-test orone-way factorial analysis of variance and multiple comparison test(Fisher's method) using Excel software (Microsoft).

3. Results

3.1. SPHK1 and SPHK2 expression of various mouse cell lines

We examined SPHK1 and SPHK2 protein levels of Friend cells aswell as several mouse cell lines (Supplementary Fig. 1a). SPHK1 pro-tein level of Friend cells was much higher than those of other mousecell lines analyzed, whereas SPHK2 protein level was not. This tenden-cy was also demonstrated in mRNA level by the quantitative RT-PCR.For example, SPHK1 mRNA level of Friend cells was about six timeshigher than that of BaF3 cells, whereas SPHK2 mRNA level of Friendcells was about a half compared to BaF3 cells (data not shown). InWestern blotting, two molecular sizes of SPHK1, 43 kDa and about28 kDa were observed in cell lines with higher SPHK1 expression.A molecule of 43 kDa is consistent with the estimated molecularweight from mouse Sphk1a cDNA. Actually, Le Scolan et al. [21] re-ported overexpression of two SPHK1 proteins in the mouse erythro-leukemia cell lines they established. We reexamined the SPHK1protein levels using the anti-SPHK1 antibody fromCell Signaling Tech-nology, which was reported to detect mouse SPHK1. In Supplementa-ry Fig. 1b(B), these two bands (43 kDa and 28 kDa) as well as manynon-specific bands were also detected in Friend cells.

siRNA for mouse Sphk1 reduced the intensity of these two bands ofFriend cells (Supplementary Fig. 1c). Moreover, transient transfectionof mouse SPHK1a into NIH3T3 cells produced two protein bandswith 43 and 28 kDa (Supplementary Fig1d). Thus, our results suggestthat this 28 kDa protein was derived from mouse SPHK1a protein.Concerning the appearance of low molecular weight band, Kihara etal. [36] reported the presence of small SPHK1, and hypothesized thatthe small SPHK1 molecule might be derived from mouse SPHK1b but

1009N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

not from SPHK1a. However, our 5′RACE experiments of Friend cellsdid not reveal the presence of a mouse Sphk1b transcription startsite, and this 28 kDa band was also observed in Friend cells over-expressing mouse SPHK1a (Fig. 5a).

We then tried to characterize this 28 kDa protein. In Supplementa-ry Fig. 2a and b, the fractionation pattern of Friend cell lysatewas illus-trated by the preparatory Superose 6 column (a) and the finalSuperdex 75 column (b) chromatography connected with SMARTHPLC system (Pharmacia LKB Biotechnology, Uppsala, Sweden).Three fractions (fraction 19–fraction 21 by Superose 6) were collectedand separated by Superdex 75. Although we could not completelyseparate these two proteins under our experimental condition evenusing other columns such as monoQ (data not shown), the fraction22 separated by the Superdex 75 column (b) contained 43 kDa mole-cule much more than 28 kDa, whereas fraction 25 contained more28 kDa molecule than 43 kDa. Furthermore, we compared the profileof SPHK1 enzyme activity of each fraction. Interestingly, the SPHK1enzyme profile paralleled those of 43 kDa SPHK1 but not those of28 kDa SPHK1, suggesting that 28 kDa SPHK1 protein had little or noenzyme activity. The combination of fraction 22 and fraction 25 didnot reduce the total SPHK1 enzyme activity (Supplementary Fig. 2c),indicating that 28 kDa did not play a dominant negative role.

3.2. 5′-promoter analysis

Based on our 5′ RACE experiments, we determined the transcrip-tion start point of Sphk1 in Friend cells. Our identified start site is78 bp upstream of Sphk1a type based on the database available online (NM_025367) (data not shown). We could not observe thetranscription start site of mouse Sphk1b type under our 5′RACE exper-imental condition. Promoter analysis using various lengths of reporter

Fig. 1. Promoter analysis of mouse Sphk1 in Friend cells. (a) 5′-promoter of Sphk1 was clonforms of 5′-promoter reporter vectors were prepared with −1.7 kb/luc as starting materialconstruct was illustrated at left. (b) Promoter analysis was performed using various reportenotes c-MYB-binding site. Squares show v-MYB-binding site, whereas circles illustrate Ets-Promoter assay was performed as described in Materials and methods.

vectors revealed that the region between −53 and −4 bp from thefirst exon is sufficient for the promoter activity of Friend cells(Fig. 1a). We also examined HMBA-treated Friend cells, but DNAtransfection efficiency was too low to obtain reliable results. Thus,we further analyzed this region by inserting mutations into the puta-tive transcription factor binding sites as illustrated in Fig. 1b left. Itshows that the proximal MYB-binding site (Area II) was the most im-portant part within this area, though the other two binding sites(Areas I and III) also demonstrate some promoter activity. Thus, we fo-cused on this Area II and c-MYB protein.

3.3. EMSA

The promoter region containing three MYB-binding sites and oneEts-binding site (shown in Fig. 2a containing Areas II and III of Fig. 1b)was used as the biotin-labeled probe for EMSA. Nuclear extracts ofFriend cells produced three bands, 1, 2 and 3, although bands 1 and2 were very weak (Fig. 2b). Bands 2 and 3 competed with the samenon-labeled probe, suggesting the specificity of these band forma-tions. Anti-c-MYB antibody decreased band 3, although not complete-ly (Fig. 2c). Therefore, we produced c-MYB protein from GST-c-MYBfusion protein as described in Materials and methods. It was shownthat band 3 contained c-MYB protein (Fig. 2d). HMBA-treated Friendcells demonstrated the reduction of band 3, which paralleled the cel-lular c-MYB protein level (Fig. 4a). Moreover, band 1 increased its in-tensity in HMBA-treated Friend cells (Fig. 2e). The complex occupancyof this promoter region by transcription factors except for c-MYB inHMBA-treated cells suggests that these changes are the possible inhibi-tion mechanism of Sphk1 transcription by HMBA. Further analysis isneeded to identify suppressive transcription factors involved.

ed by PCR method as described in the Materials and methods. Truncated and mutateds. Promoter analysis was performed as described in Materials and methods. Respectiver vectors with mutation. Three areas (I, II and III) are tentatively named. Diamond de-binding site. Open symbols represent wild-type and solid figures denote mutated one.

Fig. 2. EMSA. EMSA was performed with biotin labeled probe containing region II and III. (a) Location of the labeled oligoprobe was illustrated. (b) Cold denotes cold competitor(10- or 100-fold excess). Each band was named bands 1, 2, or 3. (c) Supershift experiment was performed using anti-c-MYB antibody (2 or 6 μg/each sample). Longer exposure timewas used to detect weak band 1 and 2 clearly. (d) EMSA using nuclear extract and purified c-MYB protein. EMSA was performed using Friend cell nuclear extract (designated as NP)or isolated c-MYB protein. Molecular sizes of GST-MYB and c-MYB isolated by Factor Xa digestion are shown in the lower part. (e) Nuclear extracts were prepared from Friend cellswith or without HMBA treatment for 48 h.

1010 N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

3.4. ChIP assay and effects of c-Myb siRNA and c-MYB overexpression onSPHK1 expression

Fig. 3a illustrates the DNA region amplified in the ChIP assay. In vivobinding of c-MYBwas clearly shown (Fig. 3b).Moreover, c-Myb silencing

Fig. 3. ChIP assay and modulation of c-MYB expression level by siRNA for c-Myb or c-MYperformed as described in Materials and methods. (c) siRNA for c-Myb or nonspecific (cwere collected 72 h after transfection and Western blotting was performed using antmock-vector were transfected to Friend cells according to Materials and methods. Westernwas different so as to allow detection of the differences between control and suppressed c-control and c-MYB-overexpressed Friend cells was measured according to Materials and meinal cells and c-MYB-overexpressed cells.

by its siRNA and stable c-MYB overexpression proved that the c-MYBlevel significantly affected SPHK1 expression (Fig. 3c and d). Increasedcellular S1P level was confirmed in c-MYB-overexpressed Friend cells(Fig. 3e). We also confirmed c-MYB-induced SPHK1 overexpression inNIH3T3 cells (Supplementary Fig. 1e).

B expression vector. (a) Primers used for ChIP assay were shown. (b) ChIP assay wasontrol) siRNA was transfected transiently according to Materials and methods. Cellsi-SPHK1, anti-c-MYB, and anti β-actin antibody. (d) c-MYB expression vector andblotting was performed as shown in Fig. 3c. Exposure time of c-MYB Western blottingMYB (Fig. 3c) or between mock and overexpressed c-MYB (Fig. 3d). (e) Cellular S1P ofthods. The asterisk means the statistically significant difference, pb0.001, between orig-

1011N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

3.5. Change of SPHK1 expression during HMBA-induced erythroiddifferentiation of Friend cells

We next examined HMBA-induced differentiation of the originalFriend cells. The commitment to erythroid differentiation reportedlyprecedes the expression of erythroid lineage-specific gene expres-sion as well as their protein expression [37]. The increase of thecommitted cells preceded the increase of hemoglobin-positive cellsas well as mouse globin mRNAs by about 16–24 h, and the increaseof hemoglobin-positive cells and mouse globin mRNAwas well corre-lated. (Supplementary Fig. 3a and b). During HMBA treatment, Friendcells rapidly decreased SPHK1 protein expression, whereas SPHK2protein did not decrease by HMBA but rather increased slightly(Fig. 4a). Moreover, removal of HMBA from the culture medium onday 3 did not restore SPHK1 expression (Fig. 4b), suggesting thatfully committed Friend cells by HMBA did not recover SPHK1 expres-sion as well as cell proliferation capacity. In contrast, SPHK2 did notshow such significant changes with this treatment (data not shown).Sphk1 mRNA and SPHK1 enzyme activity also rapidly decreased byHMBA (Fig. 4c and d). Cellular S1P produced by SPHKwas also rapidlydecreased during HMBA-induced differentiation (Fig. 4e), suggestingthe significant role of SPHK1 but not SPHK2 in Friend cell differentia-tion. We could not measure S1P in culture medium, because Friendcells could not maintain cell viability in serum-depleted or reducedmedium for a long period of time, which is a necessary step to excludeserum S1P for our S1P measurement by LC/MS/MS.

To prove the effects of proto-oncogenes reported previously con-cerning Friend cells [22] on SPHK1 expression, we examined FLI-1 andPU.1 proteins, which are reported to decrease during HMBA-induced

Fig. 4. SPHK1 expression during HMBA-induced differentiation of Friend cells. (a) Friend celected sequentially. Antibodies used are described in Materials and methods. (b) Friend cellscultures were continued for another 3 days. Western blotting of SPHK1 and β-actin was pequantitative RT-PCR according to Materials and methods. Open column means control Friendregarded as 1.0. (d) Friend cells were cultured with 5 mM of HMBA in triplicate. SPHK1 enzwas measured by the LC-MS/MS according to Materials and methods. Friend cells were culttriplicate. Solid column shows HMBA-treated cells (Day 1 and Day 3), while open column iscells.

differentiation [23]. c-MYB expression was also examined based onthe data shown in Figs. 1, 2, and 3. HMBA rapidly decreased c-MYB,FLI-1 and PU.1 proteins, whereas GATA-1, which regulates erythroid-lineage specific phenotype, was gradually increased by HMBA fromday 1 (Fig. 4a).

3.6. Effects of SPHK1 and SPHK2 overexpression

Because rapid decrease of SPHK1 was observed during HMBA-induced differentiation, we established SPHK1-overexpressing Friendcells to examine whether stable overexpression of SPHK1 affectscell proliferation, survival and differentiation. Fig. 5a shows thatestablished SPHK1 subclones exhibited high GFP and SPHK1 proteinlevels, because we used MSCV-IRES-GFP retrovirus vector, whichexpresses GFP and SPHK1, separately and simultaneously. Thesesubclones possessed high SPHK1 enzyme activity as well as high cellu-lar S1P levels (Fig. 5b). SPHK1-WT sc2 exhibited higher cell prolifera-tion than original Friend cells, and showed resistance against serumdepletion-induced cell death (Fig. 5c left and middle). Furthermore,erythroid differentiation of SPHK1-WTwas delayed but not complete-ly suppressed compared with the original Friend cells (Fig. 5c right).Other clones (SPHK1-WT sc3) exhibited similar results (data notshown). We also examined sequential changes of PU.1, FLI-1, andc-MYB protein as well as the cellular S1P level of HMBA-treatedSPHK1-overexpressed sc2. PU.1, FLI-1, and c-MYB exhibited the simi-lar rapid decrease as those observed in original Friend cells (Supple-mentary Fig. 4a). Interestingly and unexpectedly, the cellular S1Plevels also decreased gradually in spite of stable SPHK1 expression(Supplementary Fig. 4b), suggesting the possibility that HMBA affects

lls were treated with 5 mM of HMBA. Western blotting was performed using cells col-were cultured with HMBA for 3 days, then HMBA was removed from the medium andrformed using cells collected sequentially. (c) Sphk1 mRNA level was measured by thecells. Solid columns show HMBA-treated cells. Sphk1mRNA of control Friend cells wasyme activity was measured according to Materials and methods. (e) Cellular S1P levelured with 5 mM of HMBA. Cells were collected and their cellular S1P was measured incontrol Friend cells. The asterisk denotes pb0.001 between control and HMBA-treated

Fig. 5. Establishment of SPHK1-overexpressing Friend cells. Wild-type (WT) -SPHK1 transfectants were established as described in Materials and methods. (a) Left: Flow cytometryof cellular GFP was shown. The horizontal axis denotes fluorescence intensity. The vertical axis shows the population. Right: Western blotting of SPHK1 was shown. (b) Left: SPHK1enzyme activities of original Friend cells and SPHK1-WT were measured according to Materials and methods. Relative activities were calculated with that of original Friend cells as1.0. Mean±SD was demonstrated. Right: Cellular S1P level of original and SPHK1-overexpresssed sc2 was measured according to Materials and methods. (c) Left and middle partsillustrate cell growth curve under conventional (10%) or low (0.25%) serum containing culture. The initial cell concentration of the cell growth curve cultured in 10% FCS (left) was1×104/ml, whereas that of viable cell number analysis (middle) was 5×105/ml. Viable cell numbers were counted with trypan blue dye exclusion. Right: hemoglobin-positive cellstreated 5 mM of HMBA in 10% FCS (+) medium were counted as described in Materials and methods. Cultures were performed in triplicate and the mean±SD was illustrated.Experiments were repeated at least three times with similar results. Open triangle shows WT-sc2. Solid square means original Friend cells. The asterisk means statistically signif-icant difference, pb0.001, between original Friend cells and WT-SPHK1-overexpressed sc2.

1012 N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

other sphingolipid metabolic enzymes than SPHK1. Similarly, we alsoexamined cell proliferation and differentiation of c-MYB overex-pressed cell lines. Enhanced cell proliferation and delayed differentia-tion by HMBA were also observed (Fig. 6).

Because SPHK2 slightly increased during differentiation (Fig. 4a), wealso examined SPHK2-WT (sc3 and 4) transfectants (Supplementary

Fig. 6. Characterization of c-MYB-overexpressed Friend cells. Viable cell number of original aand middle). Cells were cultured in triplicate, and the mean±SD was calculated. Sequeoverexpressed cells (open triangle) was illustrated (right). The asterisk shows statistically scells.

Fig. 5 and data not shown). A moderate increase of SPHK2 activitywas observed, but unexpectedly, cellular S1P level was not signifi-cantly increased in our stably SPHK2-overexpressed cells. Our strat-egy of G418 selection might not be suitable for producing higherSPHK2 expressing cells. Cell proliferation as well as differentia-tion was delayed in SPHK2-WT sc3. (Supplementary Fig. 5c). Other

nd c-MYB-overexpressed cells was counted in culture with 10% FCS or 0.25% of FCS (leftntial change of HMBA-induced differentiation of original (solid square) and c-MYB-ignificant difference, pb0.001, between original Friend cells and c-MYB overexpressed

1013N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

SPHK2-WT sc4 cells showed a similar tendency (data not shown).The delayed differentiation observed in SPHK2-overexpressed cellsremains to be elucidated. Nuclear S1P produced by SPHK2 but nottotal cellular S1P might be important for cellular functions inducedwith activated SPHK2. However, at the moment, the contribution ofSPHK2 in HMBA-differentiation may well be small compared to therapid decrease of SPHK1 and cellular S1P during the differentiationprocess (Fig. 4).

3.7. Effects of SPHK inhibitor and S1P receptor antagonists

We analyzed the effects of S1P receptor antagonists (pertussistoxin, VPC23109 for S1P1/3 and CAY10444 for S1P3) to examinewhether Friend cell proliferation and differentiation were due to theextracellular or intracellular S1P action (Fig. 7a).We used the effectiveconcentrations of these inhibitors confirmed in previous experiments[38]. These reagents did not affect cell growth and HMBA-induced dif-ferentiation of original Friend cells, suggesting that the involvement ofS1P receptor was less likely and that S1P produced by SPHK1 mightfunction in an intracellular fashion. Fig. 7b left shows that chemicalSPHK inhibitor inhibited cell growth of SPHK1-overexpressed Friendcells (sc2) dose-dependently. Interestingly, SPHK inhibitor enhanceddifferentiation significantly (Fig. 7b right). We also examined the ef-fect of Sphk1 siRNA. Although Sphk1 siRNA did not suppress overex-pressed SPHK1 protein completely and stably (Fig. 7c), it was shownthat cell growth on day 3 was inhibited by 35% and that erythroid dif-ferentiation was mildly but significantly enhanced (Fig. 7c middle andright).

4. Discussion

The mouse Friend cell line exhibits terminal erythroid differentia-tion rapidly and with high efficiency by HMBA or DMSO [37]. Differ-entiated cells lose their proliferation capacity. Similar differentiationmodels analyzed are all-trans retinoic acid-induced human acutepromyelocytic leukemia cell lines such as NB4 or HL60 cells [39].Chemical differentiation induction of leukemia cells is important be-cause of its potent activity to eradicate leukemia cells via terminal dif-ferentiation, leading to the search for therapeutic drugs. Friend virusinfection is thought to shift erythropoietin (EPO)-dependent targetcells to EPO-independent ones after virus infection [40,41]. Moreover,PKC has been considered the cause of EPO-independent growth oferythroid cells infected with Friend spleen focus-forming virus [42].However, the mechanism of HMBA-induced differentiation of Friendcells remains to be fully elucidated.

Based on the previous report by others [21], we examined themechanism of SPHK1 overexpression in Friend cells and the role ofSPHK1 during the erythroid differentiation process. Our results con-firmed higher SPHK1 expression of Friend cells than various othermouse cell lines (Supplementary Fig. 1a). SPHK2 of Friend cells didnot show a similar overexpression as SPHK1, suggesting differentregulation of SPHK1 and SPHK2 expression. Consistent with these re-sults, we previously reported higher expressions of SPHK1 but notSPHK2 in the pre-leukemic state, myelodysplastic syndrome, andacute leukemia and also in a human erythroleukemia cell line, K562cells [11,12], suggesting a positive relationship between SPHK1 andleukemogenesis.

We experienced the presence of two molecules of mouse SPHK1(43 kDa and 28 kDa) in cellswith higher SPHK1 expression, as reportedbefore by others [21]. Because the smallermoleculewas not observed incells with low SPHK1 expression, it is less likely that this band isnon-specific. Further analysis using the antibody from Cell Signaling inaddition to our antibody detected similar two bands. siRNA for Sphk1and Sphk1 cDNA overexpression experiments strongly suggest thatthis lowmolecular band is derived frommouse SPHK1 (SupplementaryFig. 1). SPHK1 is evolutionally highly conserved andmouse Sphk1 cDNA

and human SPHK1 cDNA exhibit a high homology. Five important con-served domains (C1–C5) are also observed in mouse Sphk1 cDNA.Among these five domains, C1–C3 are thought to be catalytic domain,and are located at the N-terminal part [43].

Although we were unable to completely separate 43 kDa and28 kDa molecules by HPLC, (Supplementary Fig. 2b), it was revealedthat the profile of SPHK1 enzyme activity paralleled that of 43 kDacontent but not that of 28 kDa. Interestingly, the heterogeneousSPHK1 proteins has been reported in human platelets [44] and it hasalso been reported that endogenous cathepsin B digests humanSPHK1 producing several smaller fragments including 30–31 kDaand that its major cleavage site is located at H122 [45,46]. Our ob-served 28 kDa form of mouse SPHK1 is very close to the molecularsize of 31 kDa molecule and the cleavage site by cathepsin B is alsoconserved in mouse SPHK1.

This 28 kDa molecule was recognized by our antibody preparedwith the C-terminal 18 peptides immunization, suggesting that the28 kDa molecule possesses the C-terminal sided-portion. Concerningthe relationship between molecular structure and enzyme activity, ithas been reported that deletion of only 69 amino acids from the wildtype human SPHK1 severely reduced its enzyme activity [47]. More-over, it has also been reported that deletion of various regions fromSPHK1 mostly causes enzyme inactivation [48]. Thus, it is likely that28 kDamoleculewas cleaved from43 kDa SPHK1 probably by cathep-sin B or other proteases and that derived 28 kDa lacks SPHK1 enzymeactivity. The combination experiment of 43 kDa- and 28 kDa-richfractions did not indicate a dominant negative effect of 28 kDa mole-cule (Supplementary Fig. 2c).

Human and rat SPHK1 promoter analyses have been reported[33,49,50], and it was shown that Ap-2 and Sp1 sites within −0.2 kbof the 5′ SPHK1 promoter are important for both endogenous andNGF- or GDNF-induced transcription. However, mouse Sphk1 5′-pro-moter has not been examined in detail [31]. Our present analysis dem-onstrated that the short region between −53 bp and the first exon isimportant for mouse endogenous Sphk1 transcription of Friend cells(Fig. 1). Interestingly, this region contains two putative Ets bindingdomain and several MYB binding domains. FLI-1 and PU.1 belongto the Ets transcription factor family. Within this region (between−53 bp and −4 bp from the first exon), Area II MYB-binding sitewas found to be the most potent. Thus, Sphk1 transcription of Friendcells is coordinately regulated by these proto-oncogenes/transcriptionfactors.

Considering our present data and previous reports together, min-imal promoters involved in Sphk1 transcription in various cancer andleukemia cells are not homogenous, and different cellular contexts orsignaling pathways determine how Sphk1 promoter region is stimu-lated in its transcription. EMSA using this region as the labeledprobe revealed that c-MYB could bind this 5′-Sphk1 promoter regionand that band 3 is due to c-MYB protein binding (Fig. 2 b, c, and d).This band decreased in HMBA-treated Friend cells (Fig. 2e), sug-gesting the important role of MYB-induced SPHK1 in the HMBA sen-sitivity of Friend cells.

Increased intensity of band 1 after HMBA treatment was unex-pected and interesting. Conceivably, some proteins in Friend cellsbound with this region in competition with c-MYB, or independentlyto suppress Sphk1 transcription. Unidentified transcription factorsand accessory factors induced by HMBA are such candidates. Furtheranalysis is needed to identify all the components involved in de-creased Sphk1 transcription during HMBA-induced differentiation.At the moment, complex occupancy of the 5′-promoter of Sphk1 byheterogeneous factors during HMBA-induced differentiation revealedthat HMBA-induced modulation of Sphk1 transcription was morecomplex than expected.

ChIP assay confirmed the direct binding of c-MYB protein in vivo(Fig. 3b). Moreover, siRNA against c-Myb and overexpression ofc-MYB confirmed our results described above, suggesting c-MYB as

Fig. 7. Effects of SPHK inhibitor and S1P receptor antagonist. (a) SPHK1 overexpressed sc2 Friend cells were cultured in triplicate with or without 5 mM of HMBA in combination ofpertussis toxin (100 ng/ml), and VPC23019 (10 nM), and CAY10444 (10 nM). Cell proliferation (left) and the percentage of hemoglobin positive cells (right) on day 3 after HMBAtreatment were counted. (b) and (c) Similarly, effects of SPHK inhibitor and siRNA for Sphk1 were examined by counting cell proliferation and the percentage ofhemoglobin-positive cells of SPHK1 overexpresssed sc2. Effect of Sphk1 siRNA was shown in the Western blotting (7c left). In order to exhibit its inhibitory effect on two SPHK1molecules, short exposure time was chosen to evaluate 28 kDa protein. Hemoglobin-positive cells were counted on day 2 after HMBA treatment. The mean±SD was shown andthe statistical significance was examined. The asterisk denotes statistically significant difference, pb0.001, between control and respective treated cells.

1014 N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

the potentmodulator of SPHK1 expression (Fig. 3c andd). The increasedcellular S1P and SPHK1 overexpression in c-MYB-overexpressed cellsare consistent.

Transcription factors and oncoproteins involved in Friend cell leu-kemogenesis, c-MYB, PU.1, and FLI-1, exhibited a sequential decreaseby HMBA (Fig. 4a). These three molecules might function in concertin both leukemogenesis and the differentiation process. However,we focused on c-MYB based on the promoter analysis (Fig. 1) andthe fact that c-MYB is located upstream of SPHK1 (Fig. 2c and d). Dur-ing HMBA-induced differentiation of Friend cells, SPHK1 but notSPHK2 decreased very rapidly (Fig. 4a), suggesting a major role ofSPHK1 in determining the differentiation sensitivity of Friend cells.It is of note that fully-committed Friend cells did not restore SPHK1expression as well as cell proliferation even after removal of HMBA

(Fig. 4b), suggesting the positive link between SPHK1 level and theterminal erythroid differentiation of Friend cells.

Cellular S1P level was consistent with the rapid change of SPHK1enzyme activity and protein levels. Okazaki et al. reported increasedceramide level in Vitamin D3-induced HL-60 cells differentiation[51]. However, the analysis of SPKH1/S1P signaling in the differentia-tion process has not been extensively studied, and most studiesfocused on the oncogenic (cell proliferation promoting) activity ofSPHK1 [21,52,53]. Considering the sphingolipid biostat model, it issuggested that the increase of ceramide/S1P ratio due to the remark-able decrease of S1P plays a significant role in the cell cycle arrest ofdifferentiated Friend cells.

Friend cells exhibited rapid SPHK1 decrease during the HMBA-induced differentiation. Thus, we examined the effects of stable

1015N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

overexpression of SPHK1, which was not influenced by HMBA treat-ment. Our results of stably SPHK1 wild-type (WT) overexpressingFriend cells are consistent with the concept that SPHK1 overexpressionenhances cell survival under cellular stress, and stimulates cell prolifer-ation (Fig. 5b). The increase in cellular S1P of SPHK1 overexpressed cellssupports this idea. However, differentiation of these cells was delayedsignificantly but not completely suppressed. Actually some decrease incellular S1P was observed in SPHK1-overexpressed cells after HMBA-treatment (Supplementary Fig. 4b). Cellular PU.1, FLI-I, and c-MYB ofSPHK1-overexpresssed cells treated with HMBA also exhibited thesimilar rapid decrease (Supplementary Fig. 4a), suggesting that SPHK1does not affect upstream transcription factors whose modulation isthe step necessary for the differentiation induction and that SPHK1might not be the major regulator of the differentiation machinery inFriend cells.

The c-MYB/SPHK1 axis supported the increased cellular S1P, en-hanced cell proliferation and delayed differentiation induction ofc-MYB-overexpressed cells (Fig. 6). The significance of SPHK2 inFriend cells is of secondary importance, because cellular S1P rapidlydecreased regardless of non-fluctuating SPHK2 levels (Fig. 4). Thus,we did not discuss the role of SPHK2 in Friend cells further.

The involvement of cell surface S1P receptor in Friend cell prolifer-ation was not significant according to the results using S1P receptorantagonists and pertussis toxin (Fig. 7a). The intracellular role forSPHK1 has been reported in intestinal adenoma cell proliferation[52]. SPHK inhibitor inhibited cell proliferation of Friend cells in adose-dependent manner (Fig. 7b) and interestingly, enhanced ery-throid differentiation. It means that cell cycle arrest and/or SPHK1 in-hibition enhanced HMBA-induced differentiation. However, becauseSPHK1 inhibitor used is not free from other inhibitory effects againstother signaling molecules, we tried siRNA of Sphk1 (Fig. 7c). AlthoughsiRNA of Sphk1 exhibited the incomplete and short duration of SPHK1inhibition under our experimental condition, a mild and signifi-cant suppression of cell proliferation and enhanced differentiationwere observed compared with control siRNA at least in the early peri-od of differentiation induction. Thus, although we did not explaincompletely that high SPHK1 expression is inhibitory to HMBA-induced differentiation because of its cell proliferation promoting ac-tivity, our findings are strongly suggestive.

Taken together, our study revealed that the complex regulatorymechanism is operated in the HMBA-induced erythroid differentiationof Friend cells. We found (1) in Friend cells, c-MYB plays a major rolein Sphk1 transcription, (2) HMBA suppresses SPHK1 expression throughthe reduced c-MYB expression, and (3) SPHK1 level affects theHMBA-induced erythroid differentiation of Friend cells. Consideringthe potent growth-suppressive effect of sphingosine kinase inhibitor(Fig. 7) as well as the rapid decrease of SPHK1 of Friend cells duringHMBA-induced differentiation, our results strongly supports the ratio-nale behind the therapeutics combining SPHK inhibitor with appro-priate chemotherapeutic drugs or chemical differentiation induceragainst leukemia cells which are chemotherapy-resistant.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.bbamcr.2013.01.001.

Conflict of interest statement

The authors have no conflict of interest concerning this work.

Acknowledgments

The authors are grateful to Dr. F. Moreau-Gachelin (Inserm U528,Paris, France), Dr. T. Yoshimoto (Tokyo Medical University, Tokyo,Japan), Dr. A. Katsumi (Hamamatsu University School of Medicine,Hamamatsu, Japan), and Dr. A. Tomita (Nagoya University GraduateSchool of Medicine, Nagoya, Japan) for their vectors. This work wassupported in part by JSPS Kakenhi 23590667 and the Medical Science

Promotion Fund. Supplementary information is available at BBA Mo-lecular Cell Research's website.

References

[1] Y.A. Hannun, L.M. Obeid, Principles of bioactive lipid signalling: lessons fromsphingolipids, Nat. Rev. Mol. Cell Biol. 9 (2008) 139–150.

[2] M. Maceyka, S.G. Payne, S. Milstien, S. Spiegel, Sphingosine kinase, sphingosine-1-phosphate, and apoptosis, Biochim. Biophys. Acta 1585 (2002) 193–201.

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

[4] D. Verzijl, S.L. Peters, A.E. Alewijnse, Sphingosine-1-phosphate receptors: zoomingin on ligand-induced intracellular trafficking and its functional implications, Mol.Cells 29 (2010) 99–104.

[5] K. Takabe, S.W. Paugh, S.Milstien, S. Spiegel, “Inside-out” signaling of sphingosine-1-phosphate: therapeutic targets, Pharmacol. Rev. 60 (2008) 181–195.

[6] P. Xia, J.R. Gamble, L. Wang, S.M. Pitson, P.A. Moretti, B.W. Wattenberg, R.J.D'Andrea, M.A. Vadas, An oncogenic role of sphingosine kinase, Curr. Biol. 10(2000) 1527–1530.

[7] N. Igarashi, T. Okada, S. Hayashi, T. Fujita, S. Jahangeer, S. Nakamura, Sphingosinekinase 2 is a nuclear protein and inhibits DNA synthesis, J. Biol. Chem. 278 (2003)46832–46839.

[8] D. Shida, K. Takabe, D. Kapitonov, S. Milstien, S. Spiegel, Targeting SphK1 as a newstrategy against cancer, Curr. Drug Targets 9 (2008) 662–673.

[9] E. Ruckhaberle, A. Rody, K. Engels, R. Gaetje, G. von Minckwitz, S. Schiffmann, S.Grosch, G. Geisslinger, U. Holtrich, T. Karn, M. Kaufmann, Microarray analysis ofalatered sphingolipid metabolism reveals prognostic significance of sphingosinekianse 1 in breast cancer, Breast Cancer Res. Treat. 112 (2008) 41–52.

[10] D. Pchejetski, M. Golzio, E. Bonhoure, C. Calvet, N. Doumerc, V. Garcia, C.Mazerolles, P. Rischmann, J. Teissie, B. Malavaud, O. Cuvillier, Sphingosinekinase-1 as a chemotherapy sensor in prostate adenocarcinoma cell and mousemodels, Cancer Res. 65 (2005) 11667–11675.

[11] S. Sobue, T. Iwasaki, C. Sugisaki, K. Nagata, R. Kikuchi, M. Murakami, A. Takagi, T.Kojima, Y. Banno, Y. Akao, Y. Nozawa, R. Kannagi, M. Suzuki, A. Abe, T. Naoe, T.Murate, Quantitative RT-PCR analysis of sphingolipid metabolic enzymes inacute leukemia and myelodysplastic syndromes, Leukemia 20 (2006) 2042–2046.

[12] S. Sobue, S. Nemoto, M. Murakami, H. Ito, A. Kimura, S. Gao, A. Furuhata, A. Takagi,T. Kojima, M. Nakamura, Y. Ito, M. Suzuki, Y. Banno, Y. Nozawa, T. Murate, Impli-cations of sphingosine kinase 1 expression level for the cellular sphingolipidrheostat: relevance as a marker for daunorubicin sensitivity of leukemia cells,Int. J. Hematol. 87 (2008) 266–275.

[13] M. Vadas, P. Xia, G. McCaughan, J. Gamble, The role of sphingosine kinase 1 in cancer:oncogeneor non-oncogene addiction? Biochim. Biophys. Acta 1781 (2008) 442–447.

[14] J. Min, A.L. Stegner, H. Alexander, S. Alexander, Overexpression of sphingosine-1-phosphate lyase or inhibition of sphingosine kinase in Dictyostelium discoideumresults in a selective increase in sensitivity to platinum-based chemotherapydrugs, Eukaryot. Cell 3 (2004) 795–805.

[15] F. Moreau-Gachelin, Multi-stage Friend murine erythroleukemia: molecular in-sights into oncogenic cooperation, Retrovirology 5 (2008) 99.

[16] S.K. Ruscetti, Deregulation of erythropoiesis by the Friend spleen focus-formingvirus, Int. J. Biochem. Cell Biol. 31 (1999) 1089–1109.

[17] R.A. Rifkind, M. Sheffery, P.A. Marks, Induced erythroleukemia differentiation:cellular and molecular aspects, Blood Cells 13 (1987) 277–284.

[18] G. Juban,G. Giraud, B. Guyot, S. Belin, J.J. Diaz, J. Starck, C. Guillouf, F.Moreau-Gachelin,F. Morle, Spi-1 and Fli-1 directly activate common target genes involved in ribosomebiogenesis in Friend erythroleukemic cells, Mol. Cell. Biol. 29 (2009) 2852–2864.

[19] F. Moreau-Gachelin, Spi-1/PU.1: an oncogene of the Ets family, Biochim. Biophys.Acta 1198 (1994) 149–163.

[20] F. Moreau-Gachelin, F. Wendling, T. Molina, N. Denis, M. Titeux, G. Grimber, P.Briand,W. Vainchenker, A. Tavitian, Spi-1/PU.1 transgenic mice developmultisteperythroleukemias, Mol. Cell. Biol. 16 (1996) 2453–2463.

[21] E. Le Scolan, D. Pchejetski, Y. Banno, N. Denis, P. Mayeux, W. Vainchenker, T.Levade, F. Moreau-Gachelin, Overexpression of sphingosine kinase 1 is an onco-genic event in erythroleukemic progression, Blood 106 (2005) 1808–1816.

[22] R.G. Ramsay, K. Ikeda, R.A. Rifkind, P.A. Marks, Changes in gene expression asso-ciated with induced differentiation of erythroleukemia: protooncogenes, globingenes, and cell division, Proc. Natl. Acad. Sci. U. S. A. 83 (1986) 6849–6853.

[23] J. Starck, A. Doubeikovski, S. Sarrazin, C. Gonnet, G. Rao, A. Skoultchi, J. Godet, I.Dusanter-Fourt, F. Morle, Spi-1/PU.1 is a positive regulator of the Fli-1 gene in-volved in inhibition of erythroid differentiation in Friend erythroleukemic celllines, Mol. Cell. Biol. 19 (1999) 121–135.

[24] I. Matushansky, F. Radparvar, N. Rekhtman, A. Skoultchi, Reprogrammingerythroleukemia cells to terminal differentiation and terminal cell division, Front.Biosci. 5 (2000) D488–D492.

[25] J.W. Cui, L.M. Vecchiarelli-Federico, Y.J. Li, G.J. Wang, Y. Ben-David, ContinuousFli-1 expression plays an essential role in the proliferation and survival ofF-MuLV-induced erythroleukemia and human erythroleukemia, Leukemia 23(2009) 1311–1319.

[26] T. Yoshimoto,M. Furuhata, S. Kamiya, M. Hisada, H.Miyaji, Y.Magami, K. Yamamoto,H. Fujiwara, J. Mizuguchi, Positive modulation of IL-12 signaling by sphingosine ki-nase 2 associating with the IL-12 receptor beta 1 cytoplasmic region, J. Immunol.171 (2003) 1352–1359.

[27] A. Tomita, M. Towatari, S. Tsuzuki, F. Hayakawa, H. Kosugi, K. Tamai, T. Miyazaki,T. Kinoshita, H. Saito, c-Myb acetylation at the carboxyl-terminal conserved do-main by transcriptional co-activator p300, Oncogene 19 (2000) 444–451.

1016 N. Mizutani et al. / Biochimica et Biophysica Acta 1833 (2013) 1006–1016

[28] T. Murate, T. Kaneda, R.A. Rifkind, P.A. Marks, Inducer-mediated commitment ofmurine erythroleukemia cells to terminal cell division: the expression of commit-ment, Proc. Natl. Acad. Sci. U. S. A. 81 (1984) 3394–3398.

[29] T. Murate, T. Kaneda, Treatment of fetal bovine serum with activated charcoal en-hances spontaneous differentiation of murine erythroleukemia cells, Leuk. Res.13 (1989) 227–231.

[30] T. Murate, Y. Banno, K. Tamiya-Koizumi, K. Watanabe, N. Mori, A. Wada, Y.Igarashi, A. Takagi, T. Kojima, H. Asano, Y. Akao, S. Yoshida, H. Saito, Y. Nozawa,Cell type-specific localization of sphingosine kinase 1a in human tissues, J.Histochem. Cytochem. 49 (2001) 845–855.

[31] S. Sobue, M. Murakami, Y. Banno, H. Ito, A. Kimura, S. Gao, A. Furuhata, A. Takagi,T. Kojima, M. Suzuki, Y. Nozawa, T. Murate, v-Src oncogene product increasessphingosine kinase 1 expression through mRNA stabilization: alteration ofAU-rich element-binding proteins, Oncogene 27 (2008) 6023–6033.

[32] S. Nemoto, M. Nakamura, Y. Osawa, S. Kono, Y. Itoh, Y. Okano, T. Murate, A. Hara,H. Ueda, Y. Nozawa, Y. Banno, Sphingosine kinase isoforms regulate oxaliplatinsensitivity of human colon cancer cells through ceramide accumulation and Aktactivation, J. Biol. Chem. 284 (2009) 10422–10432.

[33] S. Sobue, K. Hagiwara, Y. Banno, K. Tamiya-Koizumi, M. Suzuki, A. Takagi, T.Kojima, H. Asano, Y. Nozawa, T. Murate, Transcription factor specificity protein1 (Sp1) is the main regulator of nerve growth factor-induced sphingosine kinase1 gene expression of the rat pheochromocytoma cell line, PC12, J. Neurochem. 95(2005) 940–949.

[34] H. Ito, K. Yoshida, M. Murakami, K. Hagiwara, N. Sasaki, M. Kobayashi, A. Takagi, T.Kojima, S. Sobue, M. Suzuki, K. Tamiya-Koizumi, M. Nakamura, Y. Banno, Y.Nozawa, T. Murate, Heterogeneous sphingosine-1-phosphate lyase gene expres-sion and its regulatory mechanism in human lung cancer cell lines, Biochim.Biophys. Acta 1811 (2011) 119–128.

[35] D. Ciznadija, R. Tothill, M.L. Waterman, L. Zhao, D. Huynh, R.M. Yu, M. Ernst, S.Ishii, T. Mantamadiotis, T.J. Gonda, R.G. Ramsay, J. Malaterre, Intestinal adenomaformation and MYC activation are regulated by cooperation between MYB andWnt signaling, Cell Death Differ. 16 (2009) 1530–1538.

[36] A. Kihara, Y. Anada, Y. Igarashi, Mouse sphingosine kinase isoforms SPHK1a andSPHK1b differ in enzymatic traits including stability, localization, modification,and oligomerization, J. Biol. Chem. 281 (2006) 4532–4539.

[37] P.A. Marks, R.A. Rifkind, Induced differentiation of erythroleukemia cells byhexamethylene bisacetamide: a model for cytodifferentiation of transformedcells, Environ. Health Perspect. 80 (1989) 181–188.

[38] M. Murakami, H. Ito, K. Hagiwara, M. Kobayashi, A. Hoshikawa, A. Takagi, T. Kojima,K. Tamiya-Koizumi, S. Sobue,M. Ichihara, M. Suzuki, Y. Banno, Y. Nozawa, T. Murate,Sphingosine kinase 1/S1P pathway involvement in the GDNF-induced GAP43 tran-scription, J. Cell. Biochem. 112 (2011) 3449–3458.

[39] J.W. Zhang, J.Y. Wang, S.J. Chen, Z. Chen, Mechanisms of all-trans retinoicacid-induced differentiation of acute promyelocytic leukemia cells, J. Biosci. 25(2000) 275–284.

[40] B. Zochodne, A.H. Truong, K. Stetler, R.R. Higgins, J. Howard, D. Dumont, S.A.Berger, Y. Ben-David, Epo regulates erythroid proliferation and differentiationthrough distinct signaling pathways: implication for erythropoiesis and Friendvirus-induced erythroleukemia, Oncogene 19 (2000) 2296–2304.

[41] J.P. Li, H.O. Hu, Q.T. Niu, C. Fang, Cell surface activation of the erythropoietin re-ceptor by Friend spleen focus-forming virus gp55, J. Virol. 69 (1995) 1714–1719.

[42] K.W. Muszynski, D. Thompson, C. Hanson, R. Lyons, A. Spadaccini, S.K. Ruscetti,Growth factor-independent proliferation of erythroid cells infected with Friendspleen focus-forming virus is protein kinase C dependent but does not requireRas-GTP, J. Virol. 74 (2000) 8444–8451.

[43] R. Alemany, C.J. van Koppen, K. Danneberg, M. Ter Braak, D. Meyer Zu Heringdorf,Regulation and functional roles of sphingosine kinases, Naunyn Schmiedeberg'sArch. Pharmacol. 374 (2007) 413–428.

[44] Y. Banno, M. Kato, A. Hara, Y. Nozawa, Evidence for the presence of multiple formsof Sph kinase in human platelets, Biochem. J. 335 (Pt 2) (1998) 301–304.

[45] T.A. Taha, M. El-Alwani, Y.A. Hannun, L.M. Obeid, Sphingosine kinase-1 is cleavedby cathepsin B in vitro: identification of the initial cleavage sites for the protease,FEBS Lett. 580 (2006) 6047–6054.

[46] T.A. Taha, K. Kitatani, J. Bielawski, W. Cho, Y.A. Hannun, L.M. Obeid, Tumor necro-sis factor induces the loss of sphingosine kinase-1 by a cathepsin B-dependentmechanism, J. Biol. Chem. 280 (2005) 17196–17202.

[47] J.A. Hengst, J.M. Guilford, E.J. Conroy, X. Wang, J.K. Yun, Enhancement of sphingo-sine kinase 1 catalytic activity by deletion of 21 amino acids from the COOH-terminus, Arch. Biochem. Biophys. 494 (2010) 23–31.

[48] S.M. Pitson, P.A. Moretti, J.R. Zebol, R. Zareie, C.K. Derian, A.L. Darrow, J. Qi, R.J.D'Andrea, C.J. Bagley, M.A. Vadas, B.W. Wattenberg, The nucleotide-binding siteof human sphingosine kinase 1, J. Biol. Chem. 277 (2002) 49545–49553.

[49] M. Murakami, M. Ichihara, S. Sobue, R. Kikuchi, H. Ito, A. Kimura, T. Iwasaki, A.Takagi, T. Kojima, M. Takahashi, M. Suzuki, Y. Banno, Y. Nozawa, T. Murate, RETsignaling-induced SPHK1 gene expression plays a role in both GDNF-induced dif-ferentiation and MEN2-type oncogenesis, J. Neurochem. 102 (2007) 1585–1594.

[50] Y. Nakade, Y. Banno, K. Tamiya-Koizumi, K. Hagiwara, S. Sobue, M. Koda, M.Suzuki, T. Kojima, A. Takagi, H. Asano, Y. Nozawa, T. Murate, Regulation of sphin-gosine kinase 1 gene expression by protein kinase C in a human leukemia cellline, MEG-O1, Biochim. Biophys. Acta 1635 (2003) 104–116.

[51] T. Okazaki, R.M. Bell, Y.A. Hannun, Sphingomyelin turnover induced by vitamin D3in HL-60 cells. Role in cell differentiation, J. Biol. Chem. 264 (1989) 19076–19080.

[52] M. Kohno, M. Momoi, M.L. Oo, J.H. Paik, Y.M. Lee, K. Venkataraman, Y. Ai, A.P.Ristimaki, H. Fyrst, H. Sano, D. Rosenberg, J.D. Saba, R.L. Proia, T. Hla, Intracellularrole for sphingosine kinase 1 in intestinal adenoma cell proliferation, Mol. Cell.Biol. 26 (2006) 7211–7223.

[53] Y. Kono, T. Nishiuma, Y. Nishimura, Y. Kotani, T. Okada, S. Nakamura, M.Yokoyama, Sphingosine kinase 1 regulates differentiation of human and mouselung fibroblasts mediated by TGF-beta1, Am. J. Respir. Cell Mol. Biol. 37 (2007)395–404.