Mouse Sphingosine Kinase Isoforms SPHK1a and SPHK1b Differ in Enzymatic Traits Including Stability, Localization, Modification, and Oligomerization*

Sphingosine kinases catalyze the production of the bioactive lipid molecule sphingosine 1-phosphate. Mice have two isoforms of sphingosine kinase type 1, SPHK1a and SPHK1b. In addition to the previously reported difference in their enzyme activities, we have found that these isoforms differ in several enzymatic characteristics. First, SPHK1b is unstable, whereas SPHK1a is highly stable. Degradation of SPHK1b occurs at the membrane and is inhibited by a proteasome inhibitor. Second, only SPHK1b exhibits abnormal mobility on SDS-PAGE, probably due to its SDS-resistant structure. Third, SPHK1a and SPHK1b are predominantly detected in the soluble and membrane fractions, respectively, when their degradation is inhibited. Fourth, only SPHK1b is modified with lipid, on its unique Cys residues (Cys-4 and Cys-5). Site-directed mutagenesis at these Cys residues resulted in increased sphingosine kinase activity, suggesting that the modification is inhibitory to the enzyme. Finally, SPHK1b tends to form homo-oligomers, whereas most SPHK1a is presented as monomers. We have also determined that the lipid modification of SPHK1b is involved in its homo-oligomerization. Thus, although these two proteins differ only in a few N-terminal amino acid residues, their enzymatic traits are extremely different.

Sphingosine kinases catalyze the production of the bioactive lipid molecule sphingosine 1-phosphate. Mice have two isoforms of sphingosine kinase type 1, SPHK1a and SPHK1b. In addition to the previously reported difference in their enzyme activities, we have found that these isoforms differ in several enzymatic characteristics. First, SPHK1b is unstable, whereas SPHK1a is highly stable. Degradation of SPHK1b occurs at the membrane and is inhibited by a proteasome inhibitor. Second, only SPHK1b exhibits abnormal mobility on SDS-PAGE, probably due to its SDS-resistant structure. Third, SPHK1a and SPHK1b are predominantly detected in the soluble and membrane fractions, respectively, when their degradation is inhibited. Fourth, only SPHK1b is modified with lipid, on its unique Cys residues (Cys-4 and Cys-5). Site-directed mutagenesis at these Cys residues resulted in increased sphingosine kinase activity, suggesting that the modification is inhibitory to the enzyme. Finally, SPHK1b tends to form homo-oligomers, whereas most SPHK1a is presented as monomers. We have also determined that the lipid modification of SPHK1b is involved in its homo-oligomerization. Thus, although these two proteins differ only in a few N-terminal amino acid residues, their enzymatic traits are extremely different.
The bioactive lipid molecule sphingosine 1-phosphate (S1P) 2 regulates several cellular processes such as cell proliferation, cell migration, and differentiation through binding to its cell surface receptors, which are S1P/Edg family members (1)(2)(3). In addition to its extracellular action, S1P is presumed to act intracellularly in Ca 2ϩ mobilization, cell proliferation, and apoptosis inhibition (1,2). S1P is abundant in blood (4) and is physiologically important, especially in the vascular and immune systems (5). Its importance in the vascular system is evident in S1P 1 /Edg1-null mice, which die in utero with severe hemorrhage resulting from impaired vessel integrity due to a deficiency in smooth muscle cell recruitment (6). Its influence in the immune system has been demonstrated using the synthetic immunosuppressant FTY720, which is phosphorylated in vivo and binds to S1P receptors (7,8). S1P and the S1P 1 receptor also have important functions in the egress of lymphocytes from lymphoid organs, and phosphorylated FTY720 induces the down-regulation of S1P 1 on lymphocytes and inhibits their recirculation (9,10).
Sphingosine kinases catalyze the production of S1P from sphingosine and are also responsible for the phosphorylation of FTY720 (11,12). Although sphingosine kinases possess no apparent enzymatic motif or transmembrane domain, there are five regions, termed C1 to C5, that are conserved among sphingosine kinases (13). The C1 to C3 regions bind to Mg 2ϩ -ATP (14), whereas the C4 is involved in sphingosine binding (15). Two mammalian sphingosine kinases are known, SPHK1, which was identified by purification of the enzyme and subsequent sequence determination (13), and SPHK2, which was cloned based on its homology to SPHK1 (16). Both sphingosine kinases are expressed ubiquitously among tissues, although their tissue-specific patterns differ (16). The enzymes may share redundant functions, since the single knock out of either kinase confers no apparent phenotype (12,17).
Intracellular S1P levels are regulated by a balance between synthesis by sphingosine kinase and degradation by either phosphohydrolase or lyase. Sphingosine kinases are activated by stimuli such as treatment with platelet-derived growth factor (18), tumor necrosis factor ␣ (19), or phorbol ester (20) as well as the cross-linking of Fc␥R1 (21) or Fc⑀R1 (22). Additionally, SPHK1 is known to be regulated by protein-protein interactions (23)(24)(25)(26)(27), phosphorylation (28,29), and translocation to the plasma membrane (28,29). Although such regulation is involved in the stimuli-dependent activation of SPHK1, the precise molecular mechanisms that link the stimuli and the activation still remain largely unknown in most cases.
Two isoforms for mouse SPHK1 have been reported, SPHK1a and SPHK1b (13). Interestingly, the activity of SPHK1b is significantly lower (30 -200-fold) than that of SPHK1a (13), although these proteins differ only in a few amino acid residues in their N termini. In the present study we have demonstrated that SPHK1b does not accumulate within the cell due to its low stability. We also found that SPHK1b differs from SPHK1a in several enzymatic characteristics. These include structural resistance toward SDS, membrane localization, protein stability, post-translational modification, and oligomer formation. Thus, the N terminus plays an important role in the determination of the enzymatic properties of SPHK1.

MATERIALS AND METHODS
Cell Culture and Transfection-Mouse F9 embryonal carcinoma cells and human embryonic kidney (HEK) 293T cells were grown in Dulbecco's modified Eagle's medium (D6429; Sigma) containing 10% fetal calf serum and supplemented with 100 units/ml penicillin and 100 g/ml streptomycin in 0.1% gelatin-and 0.3% collagen-coated dishes, respectively. Transfections were performed using Lipofectamine TM 2000 reagent (Invitrogen) for F9 cells and Lipofectamine Plus TM reagent (Invitrogen) for HEK 293T cells.
The pCE-puro 3ϫFLAG-Ub plasmid, which encodes ubiquitin tagged with a 3ϫFLAG epitope at the N terminus, was constructed as follows. The ubiquitin cDNA was amplified from 17-day mouse embryo cDNA (BD Biosciences Clontech, Palo Alto, CA) using primers 5Ј-AG-GATCCATGCAGATCTTTGTGAAGACCCTG-3Ј and 5Ј-TTTAG-CCACCTCTGAGGCGAAGGACCAGG-3Ј. The amplified fragment was cloned into the pGEM-T Easy vector, creating the pGEM Ub plasmid. The pCE-puro 3ϫFLAG-Ub plasmid was constructed by cloning the 0.26-kilobase BamHI-NotI fragment of the pGEM Ub plasmid into the BamHI-NotI site of the pCE-puro 3ϫFLAG-1 vector, which had been designed to produce an N-terminal 3ϫFLAG tagged protein.
Pulse-Chase and Immunoprecipitation -HEK 293T cells were transfected with the pCE-puro SPHK1a, pCE-puro SPHK1b, or pCE-puro SPHK1b-C1C2 plasmid. Eighteen h after transfection, the cell layer was disrupted by a trypsin/EDTA solution, and the cell suspension was divided into 4 or 5 aliquots. Each aliquot was added to a new 30-mm culture dish and incubated at 37°C for 24 h. Culture medium was then changed to 1 ml of Dulbecco's modified Eagle's medium without Met/ Cys (D0422, Sigma) and incubated at 37°C for 1 h. Cells were pulselabeled with [ 35 S]Met/[ 35 S]Cys (22 Ci/dish EXPRESS TM protein labeling mix; PerkinElmer Life Sciences) for 20 min and chased with unlabeled Met (final concentration 0.5 mg/ml) and Cys (final concentration 0.1 mg/ml) in 1 ml of Dulbecco's modified Eagle's medium (D6429) containing 10% fetal calf serum. At predetermined times cells were washed with phosphate-buffered saline (PBS), treated with 1 ml of radioimmune precipitation assay buffer (50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% SDS, 0.1% sodium deoxycholate, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1ϫ protease inhibitor mixture (Complete TM EDTA free; Roche Diagnostics)), and kept on ice. Cells were disrupted by 5 passages through a 21-gauge needle, and debris was removed by centrifugation at 20,000 ϫ g for 5 min at 4°C. Cell lysates with equal radioactivity were incubated with affinity-purified anti-SPHK1 antibodies and protein A-Sepharose (Amersham Biosciences) at 4°C for 14 h. After 2 washes with 1 ml of radioimmune precipitation assay buffer and 1 with 1 ml of 10 mM Tris-HCl (pH 8.0), beads were suspended in 2ϫ SDS sample buffer (125 mM Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, and a trace amount of bromphenol blue) containing 10% 2-mercaptoethanol and boiled for 3 min. The precipitates were then separated by SDS-PAGE. Radioactivities associated with SPHK1 were quantified using a Bio-Imaging Analyzer BAS2500 (Fuji Photo Film, Tokyo, Japan).
For detection of SPHK1-ubiquitin conjugations, HEK 293T cells were transfected with pCE-puro 3ϫFLAG-Ub and either pCE-puro SPHK1a or pCE-puro SPHK1b, then incubated for 40 h at 37°C. Preparation of cell lysates and immunoprecipitation with anti-SPHK1 antibodies were performed as described above. Precipitates were separated by SDS-PAGE and subjected to immunoblotting using an anti-FLAG M2 antibody.
To detect homo-oligomer formation of SPHK1, HEK 293T cells were transfected with two plasmids, one carrying 3ϫFLAG-tagged SPHK1 genes and the other harboring Myc-tagged SPHK1 genes. Twenty-two hours after transfection, 20 M MG132 was added to inhibit the proteasome-dependent SPHK1 degradation, and the cells were incubated for 4 h at 37°C. The cells were washed twice with PBS, suspended in buffer A (PBS, 1 mM dithiothreitol, 1ϫ protease inhibitor mixture, and 1 mM phenylmethylsulfonyl fluoride), and sonicated. After a centrifugation at 300 ϫ g for 3 min at 4°C, the resulting supernatant was treated with 1% Triton X-100 for 30 min at 4°C. Samples were centrifuged at 100,000 ϫ g for 30 min at 4°C, and the supernatant was incubated with an M2 affinity gel (Sigma) for 2 h at 4°C. The gel was washed 3 times with PBS containing 0.1% Triton X-100, suspended in 2ϫ SDS sample buffer, and boiled. The obtained precipitates were treated with 10% 2-mercaptoethanol, boiled for 3 min, separated by SDS-PAGE, and subjected to immunoblotting with an anti-FLAG M2 antibody or an anti-Myc PL14 antibody.
In Vivo [ 3 H]Palmitic Acid Labeling-HEK 293T cells grown on a 30-mm dish were transfected with plasmids and incubated for 23 h at 37°C. Culture medium was then changed to 1.5 ml of serum-free Dulbecco's modified Eagle's medium. After a 30-min incubation at 37°C, 3 g/ml cerulenin, which inhibits fatty acid synthesis, and 20 M MG132 were added to the medium, and the cells were incubated for 30 min at 37°C. Cells were labeled with 0.2 mCi of [ 3 H]palmitic acid (60 Ci/mmol; American Radiolabeled Chemical, St. Louis, MO) at 37°C for 3 h. After washing with PBS, cells were treated with 1 ml radioimmune precipitation assay buffer. SPHK1 was immunoprecipitated using anti-SPHK1 antibodies and protein A-Sepharose as described above. Immunoprecipitates suspended in 2ϫ SDS sample buffer containing 10 mM 2-mercaptoethanol were boiled for 3 min, separated by SDS-PAGE, and visualized by autoradiography using the fluorographic reagent EN 3 HANCE TM (PerkinElmer Life Sciences).
Preparation of Soluble and Membrane Fractions-HEK 293T cells transfected with plasmids were washed twice with PBS, suspended in buffer A, and sonicated. After removal of cell debris by centrifugation at 300 ϫ g for 3 min at 4°C, cell lysates were centrifuged at 100,000 ϫ g for 1 h at 4°C. The resulting supernatant and pellet were used as soluble and membrane fractions, respectively.

RESULTS
Tissue-specific Expression of SPHK1 Isoforms-There are variations in mouse SPHK1 mRNAs due to the differences in transcriptional initiation sites, and as a result, two different polypeptides, SPHK1a and SPHK1b, are produced (34). At least four types of mRNAs, differing in the first exons, encode the same SPHK1a protein, whereas only one mRNA is known for SPHK1b (34). Translation of the SPHK1a protein is initiated near the end of the common second exon (Fig. 1). By searching EST clone databases, we found that certain SPHK1a-encoding mRNAs contain a 3-bp insertion in the coding region. This newly identified variation, named here SPHK1a 2 , is produced by an altered junction between the second and third exons. Consequently, the SPHK1a 2 protein contains an extra Val residue. In contrast, transcription of SPHK1b mRNA is initiated within the intron region between the second and third exons of SPHK1a/SPHK1a 2 , so the 3Ј-half of the first exon of the SPHK1b gene is common to the third exon of SPHK1a/SPHK1a 2 (34). Translation of the SPHK1b protein is initiated from 28 bp upstream of the 5Ј-terminal end of the third exon of SPHK1a 2 (Fig. 1). Thus, there are three types of SPHK1 proteins, which differ only in their N termini. The N-terminal sequences specific to each are: SPHK1a, Met-Glu-Pro; SPHK1a 2 , Met-Glu-Pro-Val; SPHK1b, Met-Trp-Trp-Cys-Cys-Val-Leu-Phe-Val-Val. Of the 23 EST clones registered, we found that 8 represented SPHK1a, 12 represented SPHK1a 2 , and 3 represented SPHK1b, suggesting that SPHK1a 2 may be the most abundant isoform, and SPHK1b may be only a minor isoform.
We examined tissue-and embryonic development-specific expression patterns of SPHK1a/a 2 and SPHK1b, although we were unable to prepare a specific primer that would distinguish SPHK1a mRNA from SPHK1a 2 mRNA. Real-time quantitative PCR analysis revealed that SPHK1a/a 2 mRNA is ubiquitously expressed, although levels vary among tissues and embryonic stages (Table 1). However, very little expression of SPHK1a/a 2 mRNA was detectable in skeletal muscle, consistent with the extremely low sphingosine kinase activity in this tissue (31). The expression of SPHK1b mRNA was more restrictive than that of SPHK1a (Table 1). Little SPHK1b mRNA expression was detected in heart, brain, liver, and skeletal muscle. The expression levels of SPHK1b mRNA were observed to be lower than those of SPHK1a/a 2 mRNA regardless of the tissue, which is consistent with the small number of the EST clones carrying SPHK1b. The amount of SPHK1b mRNA was 10 -20% that observed for SPHK1a/a 2 mRNA in spleen, lung, kidney, and testis. In embryonic stages, days 11-15, the discrepancy was somewhat less, with SPHK1b mRNA levels at 67 and 37% of SPHK1a/a 2 mRNA levels. Thus, it would appear that the expression of SPHK1a/a 2 mRNA and that of SPHK1b mRNA are regulated differently among tissues and embryonic stages.
Abnormal Mobility of SPHK1b on SDS-PAGE-SPHK1b reportedly has a much lower activity than SPHK1a (13), which we also observed (data not shown), yet the reason was unclear. To study this phenomenon, we cloned each of the three isoforms and expressed them in mouse F9 cells. In immunoblots using anti-SPHK1 antibodies, SPHK1a and SPHK1a 2 were detected at ϳ43 kDa ( Fig. 2A), in accordance with their predicted molecular masses (both 42.4 kDa). We found no differences in enzymatic characteristics (activity, gel mobility, membrane localization, stability, and oligomerization) between SPHK1a and SPHK1a 2 ( Fig. 2A and data not shown), so we described hereafter only SPHK1a.
In contrast, the mobility of SPHK1b deviated from its predicted molecular mass (43.3 kDa) and was instead detected at 34 kDa ( Fig. 2A). This mobility was not cell-specific, since it was also observed for HEK 293T cells (Figs. 3-6). An additional band at 68 kDa, which may represent a SDS-resistant dimer (see Fig. 6), was also detected, although its level varied with the experimental conditions. A similar upper band was often observed in SPHK1a blots (Figs. 3B, 4A, and 6A); however, the intensity of the band was always lower than that observed for SPHK1b blots.
To examine the possibility that the fast mobility of SPHK1b was due to the proteolytic removal of either the N or C terminus, we prepared SPHK1a and SPHK1b constructs each tagged with 3ϫFLAG at the N terminus and Myc at the C terminus (3ϫFLAG-SPHK1a-Myc and 3ϫFLAG-SPHK1b-Myc). In immunoblots using either an anti-FLAG or anti-Myc antibody, 3ϫFLAG-SPHK1a-Myc was detected at 50 kDa (Fig. 2B), which was well accordant with the predicted molecular mass (49.2 kDa). Again, however, the apparent molecular mass of 3ϫFLAG-SPHK1b-Myc by immunoblot was about 41 kDa, 9 kDa lower than the predicted mass. Importantly, both antibodies detected the 41-kDa band, indicating that SPHK1b was not degraded at either the N terminus or the C terminus; therefore, the observed band represented the full-length protein. Thus, the cause of the abnormally fast gel mobility may be structural as is often observed for multi-span membrane proteins (32,35,36). The possibility of an intracellular disulfide bond, however, can be discounted, since we always performed SDS-PAGE under reducing conditions. Although most polypeptides are denatured by SDS and are

TABLE 1 Expression levels of SPHK1a/a 2 and SPHK1b in various tissues and during embryonic development
The mRNA expression of SPHK1a/a 2 and SPHK1b was investigated in the indicated tissues and embryonic stages by real-time quantitative PCR as detailed under "Materials and Methods." Values provided are relative to the GAPDH mRNA levels in the respective tissues or embryonic stages and represent the mean Ϯ S.D. from three independent experiments.    FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7

Different Traits between Two SPHK1 Isoforms
in their extended form in SDS-containing solutions, some proteins, including SPHK1b, may be resistant to SDS and, therefore, remain compact.
Degradation of Membrane-bound Forms of SPHK1s-The total amount of SPHK1b expressed in the F9 cells was much lower than that of SPHK1a/SPHK1a 2 ( Fig. 2A), suggesting that the SPHK1b protein is unstable. Therefore, we performed pulse-chase experiments using [ 35 S]Met/Cys. As shown in Fig. 3A, SPHK1a was highly stable, and only a slight decrease was detected at the 7-h chase point. In contrast, about half of the SPHK1b was already degraded after a 3-h incubation, indicating that SPHK1b is unstable.
We next examined the effect of MG132, a proteasome inhibitor, on the total amount of each SPHK1. The amount of SPHK1a slightly increased after treatment with MG132 (Fig. 3B). MG132 also caused an increase in the SPHK1b levels, and in this case the effect was much more prominent. These results suggest that the proteasome is involved in the degradation of SPHK1s. To further investigate this finding, the kinases were fractionated by centrifugation into soluble or membrane-bound forms. In the absence of MG132, SPHK1a was only recovered in the soluble fraction (Fig. 3C). Treatment with MG132 resulted in the appearance of membrane-bound SPHK1a with almost no change in the amount of soluble SPHK1a. Similar results were obtained for SPHK1b; MG132 caused a large increase only in the membrane fraction (Fig. 3C). These results suggest that only membrane-bound SPHK1s are susceptible to degradation by the proteasome. Furthermore, in the presence of MG132 the greater share of SPHK1a (ϳ80%) was found in the soluble fraction, whereas ϳ90% of SPHK1b was in the membrane fraction; thus, a greater tendency to associate with the membrane may account for the lower stability of SPHK1b. The membrane-bound forms of both SPHK1a and SPHK1b were found to be localized in the plasma membrane by indirect immunofluorescence microscopy. Both proteins were detected in the cytosol and the plasma membrane (Fig. 3D).
Proteins are generally modified by ubiquitin before degradation by the proteasome. To examine the degree of its ubiquitination, each SPHK1 was expressed in HEK 293T cells together with 3ϫFLAG-ubiquitin, immunoprecipitated with anti-SPHK1 antibodies, and subjected to immunoblotting with an anti-FLAG antibody. Bands were detected on the immunoblots of cells expressing either SPHK1a or SPHK1b, indicating ubiquitination (Fig. 4B). Considering the molecular mass of 3ϫFLAG-ubiquitin (11.8 kDa), it is conceivable that the broad upper bands represent polyubiquitinated SPHK1s, and the bands at the lower end monoubiquitinated proteins. Although the total amount of SPHK1b was much lower than that of SPHK1a (Fig. 4A), the ubiquitinated bands were more prominent (Fig. 4B), consistent with the high susceptibility of SPHK1b to proteasome-dependent degradation.
Lipid Modification of SPHK1b within Its Unique N-terminal Sequence-We recently demonstrated that the yeast long-chain base kinase Lcb4p is palmitoylated (37). Although there are no exact motifs for palmitoylation, FIGURE 4. Ubiquitination of SPHK1. HEK 293T cells were transfected with a plasmid encoding 3ϫFLAG-ubiquitin together with one encoding no protein, SPHK1a, or SPHK1b. A, total cell lysates were prepared, separated by SDS-PAGE, and subjected to immunoblotting with anti-SPHK1 antibodies. B, SPHK1 and any associated proteins were immunoprecipitated with anti-SPHK1 antibodies. The precipitates were separated by SDS-PAGE followed by immunoblotting with an anti-FLAG antibody. FIGURE 5. Lipid modification of SPHK1b at the N-terminal two Cys residues. HEK 293T cells were transfected with plasmid encoding SPHK1b, SPHK1b-C1, SPHK1b-C2, or SPHK1b-C1C2. A, total lysates (2 g) were separated by SDS-PAGE followed by immunoblotting with anti-SPHK1 antibodies. WT, wild-type. B, cells were labeled with [ 3 H]palmitic acid at 37°C for 3 h. Total cell lysates were prepared, and SPHK1 was immunoprecipitated with anti-SPHK1 antibodies. The precipitates were separated by SDS-PAGE and detected by autoradiography (upper panel) or by immunoblotting with anti-SPHK1 antibodies (lower panel). pal-SPHK1b indicates palmitoylated SPHK1b. C, total cell lysates (10 g) were incubated with [␥-32 P]ATP, 1 mM ATP, and 50 M sphingosine for 15 min at 37°C. Lipids were extracted and separated by TLC. Radioactivities associated with S1P were quantified using a Bio-Imaging Analyzer BAS2500. Values shown represent the mean Ϯ S.D. from three independent experiments. D, after cells were treated with 15 M MG132 for 12 h, total cell lysates were prepared and separated into membrane (M) and soluble (S) fractions by centrifugation at 100,000 ϫ g for 1 h. Each fraction was separated by SDS-PAGE followed by immunoblotting with anti-SPHK1, anti-calnexin, and anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibodies. E, cells were pulse-labeled with [ 35 S]Met/[ 35 S]Cys for 20 min, then incubated with excess cold Met/Cys for 0.5, 1.5, 3, or 6 h. Total lysates were prepared and subjected to immunoprecipitation with anti-SPHK1 antibodies, and the precipitates were separated by SDS-PAGE. Radioactivities associated with SPHK1 were quantified using a Bio-Imaging Analyzer BAS2500 and are expressed as a percentage of those at the 0.5-h chase point. many proteins are palmitoylated at Cys residues (often in clusters) near the N terminus (38,39), and in fact Lcb4p is palmitoylated at Cys-43 and Cys-46 residues (37). The N-terminal sequence that is unique to SPHK1b (and not found in SPHK1a) contains two Cys residues (Cys-4 and Cys-5); therefore we investigated the possibility that SPHK1b is modified by palmitoylation on these Cys residues. We created Cys-to-Ser mutants for Cys-4 and Cys-5 each and for both Cys-4 and Cys-5 and named these C1, C2, and C1C2, respectively. When separated by SDS-PAGE, the C1 and C2 mutants aligned with wild-type SPHK1b, whereas the C1C2 mutant exhibited an intermediate mobility between that of wild-type SPHK1a and SPHK1b (Fig.  5A). In vivo [ 3 H]palmitic acid labeling revealed that wild-type SPHK1b is labeled with [ 3 H]palmitic acid (Fig. 5B), as suspected; however, the C1 and C2 mutants were found to be less radioactive than the wild-type protein, and almost no labeling was observed for the C1C2 mutant. These results suggest that SPHK1b is modified with palmitic acid (or its metabolite) at Cys-4 and Cys-5 in the N-terminal region.
We also investigated the in vitro sphingosine kinase activities of these mutant forms of SPHK1b. Both C1 and C2 mutations caused an increase in the activity of SPHK1b, and the C1C2 double mutation further enhanced the activity (ϳ3-fold compared with wild-type SPHK1b; Fig.  5C). These results suggest that the lipid modification inhibits the enzyme activity of SPHK1b.
Because palmitoylation often anchors soluble proteins to the membrane, we investigated the membrane localization of the C1C2 mutant. Most of the C1C2 localized to the membrane in the presence of MG132, similar to wild-type SPHK1b (Fig. 5D). Moreover, the kinetics of C1C2 degradation were similar to those observed for wild-type SPHK1b (Fig.   5E). This is consistent with the idea that a tendency to associate with the membrane corresponds to the instability of the enzyme.
Homo-oligomerization of SPHK1s-As described above, two bands were often detected in blots of both SPHK1a and SPHK1b. The upper bands did not represent ubiquitinated proteins, since they were not detected by an anti-FLAG antibody when co-expressed with 3ϫFLAGubiquitin (Fig. 4B). The apparent molecular mass of each upper band was nearly double the mass of its respective monomer band. Until now, many examples illustrating SDS-resistant oligomers have been reported (35, 36, 40 -42). Therefore, we investigated whether SPHK1s could form oligomers. For this purpose, SPHK1a-Myc and SPHK1b-Myc were each expressed together with their counterpart SPHK1-3ϫFLAG protein and subjected to immunoprecipitation with an anti-FLAG antibody. Both SPHK1a-3ϫFLAG and SPHK1b-3ϫFLAG were effectively immunoprecipitated with the anti-FLAG antibody (Fig. 6A, left panel). Additionally, a significant amount of SPHK1b-Myc was detected with the immunoprecipitated SPHK1b-3ϫFLAG (Fig. 6A, right panel), accounting for ϳ45% of the total SPHK1b-Myc (Fig. 6B). In contrast, only a small amount (ϳ5%) of SPHK1a-Myc was precipitated with the SPHK1a-3ϫFLAG (Fig. 6A); most was detected in the supernatant (data not shown). The upper band of SPHK1a-Myc was only a minor species in the input fraction (data not shown), whereas its level was equivalent to the lower (monomer) band in the immunoprecipitated fraction (Fig.  6A, right panel). This enrichment of the upper band in the precipitates supports the idea that the upper band represents a dimer. The putative dimer band was also detected for SPHK1b-Myc in the precipitates (Fig.  6A, right panel). In addition, we found additional upper bands, suggesting that SPHK1b forms multimers. These results indicate that both SPHK1a and SPHK1b can form oligomers, although their oligomerization tendencies differ.
The putative dimer bands always appeared weaker in blots of the C1, C2, and C1C2 SPHK1b mutants compared with wild-type SPHK1b (for example, see Fig. 5A). To investigate the oligomer status of the C1C2 mutant, we expressed SPHK1b-C1C2-3ϫFLAG and SPHK1b-C1C2-Myc in HEK 293T cells and subjected cell lysates to co-immunoprecipitation analysis. As shown in Fig. 6B, recovery of SPHK1b-C1C2-Myc in the precipitated fraction was greatly reduced (ϳ15%) compared with wild-type SPHK1b-Myc (ϳ45%). These results suggest that the lipid modification is required for efficient oligomer formation of SPHK1b.

DISCUSSION
In the present study we have revealed that two isoforms of SPHK1, SPHK1a and SPHK1b, possess completely different molecular characteristics such as mobility on SDS-PAGE, stability, membrane localization, and oligomerization. Moreover, we found two regulatory mechanisms that are common to both SPHK1s. First, both SPHK1s are degraded after associating with membrane; however, a difference in the abilities of SPHK1a and SPHK1b to interact with the membrane may account for the difference in their stability. Second, both SPHK1s can form oligomers, although SPHK1b exhibits a much higher tendency to oligomerize. This is the first report describing the oligomerization of sphingosine kinases, although homo-oligomerization of a related lipid kinase, diacylglycerol kinase, has been reported (43).
SPHK1b exhibited an aberrant, fast mobility on SDS-PAGE. The principle that proteins of the same size migrate equally on SDS-PAGE relies on the supposition that the native structures of proteins are unfolded by SDS, which binds to the hydrophobic regions of proteins. Because proteins of the same size generally contain similar numbers of such regions, they are thought to bind nearly equal amounts of SDS molecules at levels so great that the overall negative charge of the SDS  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 overwhelms the intrinsic charges of the proteins, making the proteins roughly equal in charge. Considering this premise, three possible reasons are conceivable for the aberrant mobility of SPHK1b; they are an extraordinarily high intrinsic charge that cannot be overwhelmed by SDS, excess binding of SDS, and an SDS-resistant structure. The first possibility can be ruled out since SPHK1b differs from normally migrating SPHK1 in only a few amino acid residues, which carry no charge. The second possibility is also unlikely, since the sequence unique to SPHK1b (a nona-peptide) is too small to be covered by excess SDS. One might imagine that the palmitic acid (or its derivative) that modifies SPHK1b would bind a large amount of SDS and cause faster mobility; however, palmitoylation usually causes slower migration compared with the unmodified protein because it increases the molecular mass (44,45). Indeed, we found no examples of palmitoylation causing faster mobility. Thus, we conclude that the last possibility, an SDS-resistant structure, is the most likely. SDS does not always disrupt non-covalent interactions, and many reports describe SDS-resistant oligomers (35, 36, 40 -42). Importantly, as with the results described here for SPHK1b, SDS-resistant oligomers are sometimes accompanied by monomers exhibiting aberrant, fast mobility, as is especially apparent for highly hydrophobic proteins such as multi-span membrane proteins (35,36). Thus, the SDS resistance of the oligomers and the aberrant mobility of the monomers may be caused by a similar mechanism, probably certain hydrophobic interactions. Consistent with this notion, the putative SDS-resistant dimer of SPHK1b was reduced in the C1C2 mutant, which is not lipid-modified, and the aberrant mobility was also partially restored (Fig. 5A).

Different Traits between Two SPHK1 Isoforms
We recently reported that the yeast long-chain base kinase Lcb4p is anchored to the membrane through palmitoylation (37). In the present study we demonstrated that SPHK1b, but not SPHK1a, is modified with palmitic acid (or its metabolite) ( Fig. 5B and data not shown). However, even the C1C2 mutant of SPHK1b still associated with the membrane, suggesting that the lipid modification is not required (Fig. 5D). The N-terminal sequence unique to SPHK1b is enriched in hydrophobic amino acids such as Leu, Phe, Val, and Trp. Therefore, this region may be able to associate with the membrane without the modification. Conforming to the idea that the membrane-bound forms of SPHK1s are subject to degradation, the SPHK1b C1C2 mutant was as unstable as the wild-type SPHK1b protein (Fig. 5E).
Here we have also identified a novel isoform of SPHK1, SPHK1a 2 . Because we could not find any differences in enzymatic characteristics between SPHK1a and SPHK1a 2 , any physiological importance of the alternation in one amino acid residue is unclear. The second intron of SPHK1a is terminated by the sequence 5Ј-. . . agtag-3Ј, so two ag sequences exist. Most introns begin with gt and end with ag. Therefore, wobble seems to exist for the usage of these two ag sequences. SPHK1a 2 is produced by using the first ag, whereas usage of the second ag creates SPHK1a.
SPHK1a and SPHK1b mRNAs are produced due to differences in the transcriptional initiation sites. At least four alternate first exons direct production of SPHK1a protein (34). A so-called CpG island, in which CG sequences are enriched, exists in the SPHK1 gene (34). The methylation status of the cytosine residues within the CpG island seems to regulate the transcription initiation site of the SPHK1 gene as well as the expression level of each transcription variant. Interestingly, the methylation status differs among tissues and changes during the embryonic stages (46). Thus, it is possible that expression of SPHK1b is specifically induced by certain stimuli or conditions through changes in the methylation status of the CpG island.
A previous study reported that the activity of SPHK1b was much lower than that of SPHK1a (13), although the reason was unclear. In the present study we demonstrated that two reasons are responsible for the low activity of SPHK1b. First, SPHK1b cannot accumulate in the cells due to its instability (Fig. 3A). Second, the N-terminal palmitoylation inhibits the enzyme activity (Fig. 5C). Although information regarding the three dimensional structure of sphingosine kinase is not available, it is likely that the N terminus is located adjacent to the catalytic site and regulates the activity. Human SPHK1 also has three isoforms differing in the length of their N termini (11). Therefore, it is possible that their N termini differentially regulate their activities. Future studies using x-ray crystallography or NMR analysis will be required to reveal the structure of SPHK1 and the relationship between structure and regulation.