Molecular Characterization of Membrane-Associated Soluble Serine Palmitoyltransferases from Sphingobacterium multivorum and Bdellovibrio stolpii

Chemicals. l-Serine and the other natural l-amino acids were obtained from Nacalai Tesque (Kyoto, Japan). Palmitoyl CoA and lauroyl CoA were from Funakoshi (Tokyo, Japan). Isopropyl 1-thio-β-d-galactoside, myristoyl CoA, n-heptadecanoyl CoA, stearoyl CoA, arachidoyl CoA, palmitoleoyl CoA, and oleoyl CoA were from Sigma. The LMW gel filtration calibration kit, gel filtration calibration kit PD-10, and Sephacryl S-200 HR were from Amersham Bioscience/GE Healthcare. DEAE-Toyopearl 650 M and butyl Toyopearl 650 M were from Tosoh (Tokyo, Japan). The competent E. coli JM109 was purchased from Nippon Gene (Tokyo, Japan). E. coli BL21(DE3)(pLysS) and plasmids pET21b and pET28b were from Novagen. Plasmid pUC118 was from Takara Bio (Kyoto, Japan). All other chemicals were of the highest grade available from commercial sources.

Bacterial strains and growth conditions. Sphingomonas paucimobilis EY2395^T and S. spiritivorum EY3101^T were gifts from Eiko Yabuuchi (Gifu University School of Medicine, Gifu, Japan). S. multivorum GTC97 was from the Gifu Type Culture Collection of Gifu University, Japan. These three strains were aerobically grown in Bacto heart infusion broth (Difco, Becton Dickinson) at 30°C. B. stolpii (ATCC 27052) was purchased twice from ATCC, but it did not grow from the freeze-dried stock. A living culture was a gift from Yoko Watanabe (Niigata University, Niigata, Japan). This strain was grown in culture medium containing 1% Bacto tryptone and 0.3% Bacto yeast extract (Difco, Becton Dickinson, MD) at 30°C. The cells were collected by centrifugation and stored at −20°C before use.

Isolation and sequencing of genomic DNA clones encoding S. multivorum SPT.The genomic DNA of S. multivorum was prepared according to a standard method (5). Based on the amino acid sequences of the Sphingomonas SPT and eukaryotic LCB1/LCB2 proteins, we synthesized degenerate oligonucleotides to obtain partial DNA fragments encoding the S. multivorum SPT gene by PCR with genomic DNA from S. multivorum. The oligonucleotides 5′-GG(TCAG) (TA)(CG)(TCAG)TA(TC)AA(TC)TA(TC)(TA)T(AG)GG(TCAG)(TA)T-3′ and 5′-CC(TCAG)AT(TCAG)G(TA)(AG)TG(TCAG)GC(TC)TC(AG)TC-3′ corresponded to the amino acid sequences GSYNYLMGF and DEAHSMG, respectively, of the Sphingomonas SPT. PCR was performed using the LA Taq DNA polymerase (Takara Bio, Kyoto, Japan) under the following conditions: 30 cycles of 94°C for 30 s, 40°C for 30 s, and 72°C for 1 min and then 72°C for 10 min. The PCR product was directly cloned into a pCRII vector (Invitrogen) and sequenced using a DYEnamic ET Dye Terminator cycle sequencing kit (Amersham Biosciences) and an ABI 310 DNA sequencer (Perkin-Elmer). To obtain the full-length SPT gene, a genomic DNA library (3 × 10⁴ recombinants) was screened with the ³²P-labeled PCR product (500 bp) as a probe. The library was constructed as follows: genomic DNA from S. multivorum was partially digested with Sau3AI, and fragments of between 2.5 and 3.5 kb were purified by agarose gel electrophoresis and ligated into the BamHI-digested pUC118; these constructs were used to transform the E. coli JM109. Labeling of the probe and detection of the hybridizing fragments were performed using the BcaBEST labeling kit (Takara Bio, Kyoto, Japan) and Quick-Hyb hybridization solution (Stratagene), respectively. Three positive clones were isolated in the first screening, and the complete DNA sequence was determined for both strands of all three clones.

Isolation and sequencing of genomic DNA clones encoding the S. spiritivorum SPT.Genomic DNA from S. spiritivorum was prepared by ISOPLANT (Wako, Osaka, Japan) according to the manufacturer's specifications. Based on the amino acid sequences of Sphingomonas paucimobilis and S. multivorum SPTs, we synthesized degenerate oligonucleotides to obtain partial DNA fragments encoding the S. spiritivorum SPT gene by PCR with genomic DNA from S. spiritivorum. The oligonucleotides, 5′-CCA(TC)GC(TCAG)TC(AG)AT(TC)(TA)(TA)(TC)GA(TC)G-3′ and 5′-CC(AG)CC(GT)A(TCAG)(TA)G(TA)(GT)(CG)C(TCAG)A(CA)CGA(TC)TT-3′, corresponded to the amino acid sequences HASIID and KSLASLGG, respectively, of the S. multivorum SPT. PCR was performed using the LA Taq DNA polymerase under the following conditions: 30 cycles of 94°C for 30 s, (40 + t)°C for 30 s, and 72°C for 1 min and then 72°C for 10 min, where t denotes that the annealing temperature was successively increased by 0.25°C for each cycle. The PCR product was cloned and sequenced. To obtain the full-length SPT gene, a genomic DNA library (4 × 10⁴ recombinants) was screened with the ³²P-labeled PCR product (342 bp) as a probe. The construction of the genomic DNA library, labeling of the probe, and detection of the hybridizing fragments were carried out in the same way as for the S. multivorum. Two positive clones were isolated in the first screening, and the complete DNA sequences were determined for both strands of both clones.

Isolation and sequencing of genomic DNA clones encoding the B. stolpii SPT.Genomic DNA from B. stolpii was prepared by using ISOPLANT according to the manufacturer's specifications. Based on the amino acid sequences of the bacterial SPTs, we synthesized degenerate oligonucleotides to obtain partial DNA fragments encoding the B. stolpii SPT gene by PCR with genomic DNA from B. stolpii. The oligonucleotides, 5′-TGG(CA)TCACG(GT)(AT)T(CG)(TC)T(CA)AACGG(TC)AC(CG)TT-3′ and 5′-CGAC(AG)AA(GT)CC(AG)CC (GT)A(CA)TG(AT)(GT)(CG)C(GT)A(CA)CGATTTTGA-3′, corresponded to the amino acid sequences GSRFLNGTLD and SKSLASLGGFVA, respectively, of the S. multivorum SPT. The PCR conditions were the same as those for the S. spiritivorum SPT gene. To obtain the full-length SPT gene, a genomic DNA library (1 × 10⁵ recombinants) was screened with the ³²P-labeled PCR product (536 bp) as a probe. Other methods to isolate the SPT gene were the same as described above. Twenty-eight positive clones were isolated in the first screening, and the complete DNA sequences were determined for both strands of the three longest clones.

Expression of S. multivorum, S. spiritivorum, and B. stolpii SPT genes in E. coli.In order to construct an expression system for the bacterial SPTs in E. coli, new restriction sites, NdeI and EcoRI (for the S. multivorum SPT) or NdeI and SalI (for the S. spiritivorum and B. stolpii SPTs) were introduced into each SPT gene at the translation initiation and termination sites, respectively, by PCR. The internal NdeI restriction site (⁸⁸⁹CATATG) of the S. spiritivorum SPT gene was changed to CACATG without changing the codons by site-directed mutagenesis using the QuikChange site-directed mutagenesis kit (Stratagene). Each modified DNA fragment was ligated into the pET21b or pET28a vector using the LigaFast Rapid DNA ligation system (Promega), and each recombinant plasmid was used to transform the E. coli BL21(DE3)(pLysS) cells. Protein expression was induced with 0.1 mM isopropyl 1-thio-β-d-galactoside and continued for 6 h at 37°C.

Purification of recombinant SPTs.All purification procedures were performed at 4°C. All buffers used for the purification contained 0.1 mM EDTA, 5 mM dithiothreitol, and 20 μM PLP unless otherwise indicated. For the purification of B. stolpii SPT, 20% (wt/vol) glycerol was added to the purification buffers. The cells (10 to 25 g [wet weight]) were suspended in 150 ml of 20 mM potassium phosphate buffer (pH 7.6) and disrupted by sonication (Branson Sonic Power, Sonifier model 450) at 20 kHz for 3 min, three times. The intact cells and debris were removed by centrifugation (100,000 × g, 60 min). The supernatant solution was applied to a DEAE-Toyopearl 650 M column (2.5 by 20 cm) equilibrated with the same buffer. The proteins were eluted with a linear gradient of 0 to 500 mM NaCl in 1 liter of 20 mM potassium phosphate. The fractions containing the SPT were collected. (NH₄)₂SO₄ was added to 30% saturation, and the solution was applied onto a Butyl-Toyopearl 650 M column (2.5 by 20 cm) equilibrated with the same buffer containing 30%-saturated (NH₄)₂SO₄. For S. spiritivorum SPT, the condition of 20%-saturated (NH₄)₂SO₄ was adapted. SPT was eluted with a decreasing linear gradient of (NH₄)₂SO₄ concentrations (30 to 0% or 20 to 0%) in 1 liter of 20 mM potassium phosphate. The pooled fractions were concentrated and then applied to a hydroxyapatite column (1.6 by 20 cm) equilibrated with 10 mM potassium phosphate buffer (pH 7.6). The proteins were eluted with a linear gradient of 10 to 250 mM potassium phosphate in 1 liter. The SPT fractions were concentrated and then applied to a Sephacryl S-200 HR column (1.6 by 80 cm) equilibrated with 50 mM potassium phosphate buffer (pH 7.6) containing 0.1 mM EDTA and 150 mM NaCl. The active fractions were combined, concentrated to 2 to 5 ml, filtered, and stored at 4°C.

Mass spectrometric analyses of reaction products.To identify the reaction products of the bacterial SPT, the reaction was carried out in the presence of 1.6 mg purified SPT, 20 mM l-serine, 5 mM acyl CoA, 50 mM EDTA, 50 mM HEPES, and 0.15 M KCl (pH 7.5) in a final volume of 0.1 ml. The reaction was stopped by the addition of 0.1 ml of ∼2 M ammonia, and then the total lipids were extracted by the method of Bligh and Dyer (9), followed by hexane-2-propanol (3:2 [vol/vol]) and a salt solution partition (45). The lipid products were identified by electrospray ionization (ESI)/ion-trap mass spectrometry connected online to the normal-phase high-performance liquid chromatography (HPLC) (30, 58). LCBs can be analyzed by a method similar to the ceramide analysis previously reported (30), with minor modifications. A trap column (1 by 20 mm) fitted to a switching valve (Valco Instruments Co., Houston, Texas) was preequilibrated with solvent A (hexane containing 0.1% formic acid). An aliquot of the extracted lipids was applied onto the trap and then washed with the same solvent. Immediately after connecting the trap by valve switching to a separation silica column (1 by 150 mm; OmniSeparo-TJ, Hyogo, JAPAN) equilibrated with solvent A, the LCBs bound to the trap were eluted with the following gradient sequence: from 100% solvent A to 70% solvent A and 30% solvent B (hexane:2-propanol, 4:6 by volume) in 6 min, to 95% solvent B and 5% solvent C (hexane:2-propanol:1 M ammonium formate:water, 40:60:2.24:9.76 by volume) in 5 min, and then to 50% solvent B and 50% solvent C in 15 min. The effluent was monitored by a ThermoElectron LCQdeca mass spectrometer equipped with an ESI tip, FortisTip (20-mm inside diameter and 150-mm outside diameter; OmniSeparo-TJ, Hyogo, JAPAN) on an xyz stage (AMR, Tokyo, Japan) in positive-ion full scan and data-dependent positive-ion tandem mass spectrometry (MS/MS) modes on a single run. An ESI voltage of 1.6 kV was used.

Spectrophotometric measurements.The absorption spectra of the SPTs were recorded by a Hitachi U-3300 spectrophotometer at 25°C. The circular dichroism (CD) spectra of the SPTs were recorded by a Jasco J720-WI spectropolarimeter at 25°C. The buffer solution for the spectrometric measurements contained 50 mM HEPES-NaOH (pH 7.5), 150 mM KCl, and 0.1 mM EDTA. The purified enzyme was equilibrated with this buffer by gel filtration using a PD-10 (Sephadex G-25) column prior to the measurements.

Antibody preparation against each bacterial SPT.The antiserum against Sphingomonas paucimobilis, S. multivorum, or B. stolpii SPT was prepared by immunization of rabbits with the purified SPT proteins. Each anti-SPT immunoglobulin G (IgG) was affinity purified with the corresponding SPT-immobilized Sepharose 4B from the total IgG fraction of the rabbit serum by a standard method.

Immunoblot analysis.sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis was performed as described by Laemmli (35) with the SDS-Tris system using a 3% (wt/vol) stacking and a 10% (wt/vol) separating gel. The gels were blotted onto Immobilon-P^SQ polyvinylidene difluoride membranes (Millipore, MA) using a semidry blotting method. The membranes were blocked at room temperature for 2 h in phosphate-buffered saline (PBS) containing 1.5% (wt/vol) bovine serum albumin (BSA), followed by incubation with diluted anti-SPT rabbit antibodies at room temperature for 3 h. The membranes were washed and incubated with a 1:5,000 dilution of the horseradish peroxidase-conjugated goat antirabbit antibody solution at room temperature for 3 h. Visualization of the immunoreactive bands was performed using chemical luminescence (ECL detection kit; Amersham Biosciences Inc., Piscataway, NJ).

Electron microscopic analysis by the thin-section (postembedding) method.The cells were first fixed in a 0.1 M sodium cacodylate buffer (pH 7.4) containing 4% paraformaldehyde and 0.1% glutaraldehyde for 2 h at 4°C, dehydrated in a graded series of ethanol solutions (50% to 100%), and embedded in LR White resin (London Resin Co., Ltd., Hampshire, United Kingdom) overnight at 60°C. Ultrathin sections (90 nm in thickness) were prepared by an ultramicrotome (2088-V, LKB; Bromma, Sweden). The sections were transferred onto fine nickel grids and incubated in PBS containing 1% BSA and 1.5% normal goat serum for 15 min at room temperature and in PBS containing 1% BSA, 1.5% normal goat serum, and anti-SPT IgG overnight at 4°C. Subsequently, the grid was incubated in PBS containing 1% BSA, 1.5% normal goat serum, and goat anti-rabbit IgG conjugated to 10-nm gold particles for 30 min at room temperature. All sections were finally stained with uranyl acetate and lead citrate. The preparation was examined using an electron microscope, JEM2000EX (JEOL, Tokyo, Japan).

Other methods.SPT activity was measured according to previously described methods (27). The protein concentration during the purification procedure was determined using a BCA protein assay kit (Pierce Chemical) with bovine serum albumin as the standard. The protein concentration of the purified SPT was spectrophotometrically determined using the following molar extinction coefficients at 280 nm for the PLP form of each enzyme: 2.83 × 10⁴ M⁻¹·cm⁻¹ for S. paucimobilis SPT (27); 2.68 × 10⁴ M⁻¹·cm⁻¹ for S. multivorum SPT; 2.68 × 10⁴ M⁻¹·cm⁻¹ for S. spiritivorum SPT; and 2.68 × 10⁴ M⁻¹·cm⁻¹ for B. stolpii SPT.

Nucleotide sequence accession numbers.The nucleotide and protein sequences of the spt genes have been submitted to the GenBank database (S. multivorum, AB259214; S. spiritivorum, AB259215; B. stolpii, AB259216).

SPT activity in S. multivorum and B. stolpii.We previously reported that both Sphingomonas paucimobilis SPT and S. spiritivorum SPT are water-soluble enzymes (27). The SPT activities in S. multivorum and B. stolpii were examined in a similar way (27), and the results are presented in Fig. 1 together with those for Sphingomonas paucimobilis, S. spiritivorum, and the mouse liver microsome. For all of the bacterial strains, SPT activity was found in both the supernatant and precipitate fractions, which were prepared by ultracentrifugation. The distribution profile of the activity in the supernatant and the precipitate seems to vary depending on the species. However, since the precipitate fraction contains a nonnegligible amount of unlysed cells, it is hard to precisely estimate how much of the SPT activity is associated with the membrane. This issue was morphologically assessed using immunoelectron microscopy, as described later in detail.

• Open in new tab
• Download powerpoint

FIG. 1.

Thin-layer chromatography of radiolabeled products obtained by SPT assay reactions of mouse liver microsomes, S. paucimobilis, S. multivorum, S. spiritivorum, and B. stolpii. The assay reactions and thin-layer chromatography were carried out as described in reference 27. Reaction products formed in the presence of the crude extract, the precipitate (resuspended), or the supernatant were spotted. The volume of the crude extract used for the reaction was 100 μl (140 mg protein/ml), and the amounts of the precipitate and the supernatant were those obtained from the same volume (100 μl) of the crude extract. Lanes 1 and 8, mouse liver microsomes as a reference; lanes 2, 5, 9, and 12, crude extracts after sonication of S. paucimobilis (lane 2), S. multivorum (lane 5), S. spiritivorum (lane 9), and B. stolpii (lane 12); lanes 3, 6, 10, and 13, the precipitates after centrifugation at 1,000,000 × g of the crude extracts of S. paucimobilis (lane 3), S. multivorum (lane 6), S. spiritivorum (lane 10), and B. stolpii (lane 13); lanes 4, 7, 11, and 14, the supernatants after centrifugation at 100,000 × g of the crude extracts of S. paucimobilis (lane 4), S. multivorum (lane 7), S. spiritivorum (lane 11), and B. stolpii (lane 14).

Cloning of SPT genes from S. multivorum, S. spiritivorum, and B. stolpii.The three SPT genes (spt) from S. multivorum, S. spiritivorum, and B. stolpii were cloned by degenerate PCR and genome library screening, as described in Materials and Methods. Properties of the gene products of the bacterial SPTs are summarized in Table 1.

Sequence comparisons.The amino acid sequence alignment of the human SPT subunits, SPTLC1/SPTLC2/SPTLC3, products of isolated bacterial SPT genes and the putative SPT gene of Zymomonas mobilis ZM4 (50), and two other enzymes of the α-oxamine synthase family, i.e., E. coli 8-amino-7-oxonanoate synthase (AONS) and human 5-aminolevulinate synthase (ALAS), is shown in Fig. 2A. AONS and ALAS were selected as the proteins most and second most homologous to SPT, respectively. An overall sequence similarity was found between these proteins. Bacterial SPTs are highly similar to each other, and conserved amino acids are distributed throughout the entire sequences of the polypeptides. The S. multivorum SPT shared the highest amino acid sequence identity (87.7%) with the S. spiritivorum SPT. The Z. mobilis SPT also showed a high identity (73.1%) with the Sphingomonas SPT. The amino acid sequence identities among other bacterial SPTs were 35.0 to 48.2%. The conservation of the amino acid sequences between each bacterial enzyme was higher than that between the bacterial SPTs and human SPTLC proteins. S. multivorum SPT is 25%, 31%, and 32% identical, S. spiritivorum SPT is 27%, 32%, and 33% identical, and B. stolpii SPT is 24%, 33%, and 34% identical to human SPTLC1, SPTLC2, and SPTLC3, respectively. Several small hydrophobic stretches of amino acids were distributed throughout the bacterial SPT protein; however, no obvious transmembrane region(s) were found. The SPT-specific PLP-binding motif (GTFSKSXXXXGG) is completely conserved among all the bacterial SPTs.

• Open in new tab
• Download powerpoint

FIG. 2.

Sequence alignment and molecular phylogenetic tree of SPTs. (A) Aligned sequences of bacterial SPTs, human SPT subunits (SPTLC1, SPTLC2, and SPTLC3), E. coli AONS, and human ALAS2. The deduced amino acid sequences of the SPTs and other proteins were aligned using the CLUSTALX version 1.83 program (59). The gap opening and extension parameters were set to 10 and 0.2, respectively. Zymomonas SPT, Zymomonas mobilis SPT; Sphingomonas SPT, Sphingomonas paucimobilis SPT; S. multivorum SPT, Sphingobacterium multivorum SPT; S. spiritivorum SPT, Sphingobacterium spiritivorum SPT; Bdellovibrio SPT, Bdellovibrio stolpii SPT. Residues identical among all the proteins are in light blue, and those conservatively substituted are in green. Residues indicated by the reverse triangle are active-site residues that are considered to be important for catalysis. The red-boxed sequences are the SPT-specific PLP-binding motif (GTFSKSXXXXGG). (B) Molecular phylogenetic tree of SPTs from various sources. The phylogenetic tree was constructed by the neighbor-joining method using the E. coli AONS protein as an outgroup. The number at each node represents the bootstrap value as a percentage of 1,000 replications.

A phylogenetic tree of the SPTs was constructed by the neighbor-joining method using the E. coli AONS protein as an outgroup (Fig. 2B). The selection of this protein as the outgroup was done because E. coli AONS is the protein apparently distinct from SPT but has the highest similarity to SPT among the α-oxamine synthase family enzymes. The bootstrap values at the nodes, except for the leftmost two nodes, were all 100%. The reason for the relatively low values of 74% and 82% of the two nodes is not clear, but it may be partially attributed to the use of the evolutionarily distant (i.e., functionally different) outgroup. The divergence of the bacterial SPTs reflects the phylogeny of the bacteria; Sphingomonas paucimobilis and Z. mobilis belong to the same family of bacteria, called Sphingomonadaceae, and S. multivorum and S. spiritivorum belong to the family Sphingobacteriaceae (50, 69). The branch lengths of the LCB1 proteins are significantly greater than those of other proteins, including the LCB2, SPTLC3, and bacterial SPTs, suggesting relatively higher evolution rates of the LCB1 proteins. S. spiritivorum SPT is the nearest relative to the mammalian LCB2 proteins, followed by the SPTs from S. multivorum, B. stolpii, Z. mobilis, and Sphingomonas paucimobilis.

Overproduction and purification of recombinant SPTs.Each bacterial SPT has been stably overexpressed as a soluble protein in E. coli. The expression levels of the recombinant proteins reached approximately 10 to 20% of the total protein of E. coli cells without growth inhibition of the expression host. While the B. stolpii SPT was most abundantly expressed, half of the expressed protein formed inclusion bodies. In order to increase the solubility of the recombinant enzyme, we made a deletion variant of the B. stolpii SPT lacking the N-terminal 13 amino acid residues. The addition of 20% (wt/vol) glycerol was necessary to prevent the precipitation of the B. stolpii SPT during purification and storage. All of the recombinant enzymes were purified to homogeneity by column chromatography in three steps, and the purified SPTs showed a single protein band with an apparent M _r of approximately 45,000 for the S. multivorum and S. spiritivorum SPTs and 50,000 for the B. stolpii SPT, respectively, by SDS-polyacrylamide gel electrophoresis (data not shown). About 20 mg of purified enzyme was routinely obtained from 1-liter cultures in each case and could be stored at 4°C for more than 6 months.

Physicochemical characterizations.The M _r values of all three bacterial SPTs were estimated to be 90,000 by gel filtration. Matrix-assisted laser desorption ionization-time-of-flight MS analyses gave a signal at m/z 43,645 for the S. multivorum SPT, 43,780 for the S. spiritivorum SPT, and 44,397 for the B. stolpii SPT lacking the N-terminal 13 amino acid residues. These values were in good agreement with the values of 43,640, 43,797 and 44,522, which were calculated from the deduced amino acid sequences of each recombinant enzyme without the first methionine within experimental error. These results show that these bacterial SPTs have dimeric structures composed of two identical subunits.

The pH stability of the purified SPTs was investigated (data not shown). This information is necessary for both crystallization and storage. Enzymes from both S. spiritivorum and B. stolpii SPTs showed an above-90% activity in the pH range of 6.8 to 8.5, with an optimum pH at 7.0 to 8.0. Only the S. multivorum SPT was denatured below pH 7.2, and its optimum pH was 7.4 to 8.0.

All purified recombinant SPTs showed characteristic absorption spectra of PLP-dependent enzymes (Fig. 3). For all of the bacterial SPTs, the intensities of the absorption peaks were not changed by the pH. The shapes of the spectra of the two Sphingobacterium SPTs were different from that of the Sphingomonas enzyme (Fig. 3A, B, and C). They had only a single peak at 426 nm in addition to the protein absorption peak at 278 nm. On the other hand, the B. stolpii SPT showed two peaks, at 338 and 426 nm (Fig. 3D), and in this respect was similar to the Sphingomonas enzyme, although the relative intensities of the two peaks are different. These absorption peaks, respectively, correspond to the enolimine and ketoenamine forms of the internal Schiff base (aldimine formed between the aldehyde group of PLP and the ε-amino group of a lysine residue in the active site) of SPT. The addition of l-serine to the purified SPTs gave rise to an intense absorption band at 426 nm for all of the SPTs and a weak band at 338 nm for the Sphingomonas and B. stolpii SPTs. These spectral changes showed hyperbolic dependencies on the concentrations of l-serine, and the apparent dissociation constants (K_D ) for l-serine were calculated to be 0.47, 1.20, and 2.55 mM, respectively, for the S. multivorum, S. spiritivorum, and B. stolpii SPTs (Table 2). The CD spectra of these bacterial SPTs showed positive bands at 426 nm (and additionally at 338 nm for B. stolpii SPT), corresponding to the absorption spectra of each enzyme (data not shown). The CD spectra in the presence of a saturating amount of l-serine showed a negative band at 426 nm (data not shown). These results indicate that the added l-serine formed the external Schiff base (aldimine formed between PLP and extraneously added amino acid) with PLP. The addition of the second substrate, palmitoyl CoA, to the B. stolpii SPT, which is saturated with l-serine, resulted in a slight decrease in the 426-nm peak and the appearance of an absorption band at 515 nm (Fig. 3C). The formation of the 515-nm peak was transient, and the peak vanished within a few minutes. Such spectral changes were not observed for the Sphingobacterium SPTs or the Sphingomonas SPT.

• Open in new tab
• Download powerpoint

FIG. 3.

Absorption spectra of purified SPTs. The purified SPTs (10 μM) were dissolved in 50 mM HEPES-NaOH (pH 7.5) containing 150 mM KCl and 0.1 mM EDTA, and their absorption spectra were measured at 25°C. (A to C) Absorption spectra of the Sphingomonas paucimobilis, S. multivorum, and S. spiritivorum SPTs in the absence (solid line) or presence (dashed line) of 45 mM l-serine. (D) Absorption spectra of the B. stolpii SPT in the absence (solid line) or presence (dashed line) of 100 mM l-serine. The dotted line shows the spectrum in the presence of 100 mM l-serine and 90 μM palmitoyl CoA.

Identification of reaction products of bacterial SPTs.The formation of KDS by bacterial enzymes was confirmed by HPLC/ESI-ion-trap mass spectrometry (Fig. 4). Figure 4A shows the ion chromatograms (m/z 300) of the SPT reaction product in the positive-ion mode. The most abundant ions at m/z 300 (Fig. 4C) corresponded to the protonated molecular ions [M + H]⁺ of C18:0 KDS formed from l-serine and palmitoyl CoA, the structure of which is shown in Fig. 4B. This LCB structure was confirmed by on-line MS/MS. A positive-ion MS/MS spectrum of the m/z-300 ion showed the presence of ions of m/z 282 and 270, arising from the neutral loss of H₂O and HCHO, respectively (Fig. 4B and D).

• Open in new tab
• Download powerpoint

FIG. 4.

HPLC/ESI-ion-trap mass spectroscopy of reaction products of bacterial SPTs. (A) Ion chromatography of the reaction products of bacterial SPTs. (B) The structure of KDS, position of fragmentation, and the size of the fragment ion are indicated. (C) MS data for the reaction product of the S. multivorum SPT. (D) MS-MS data for the reaction product of the S. multivorum SPT.

Catalytic properties of bacterial SPTs.Like eukaryotic enzymes, all of the bacterial SPTs used only the l configuration of serine as an amino acid substrate. The bacterial enzymes exhibited a broad substrate specificity concerning the chain length and the degree of unsaturation of acyl CoA (Fig. 5). Figure 5A shows ion chromatograms of the reaction product formed by the S. multivorum SPT. In each chromatogram, a single ion peak of the reaction product with a molecular weight corresponding to the value expected from the chain length of the acyl CoAs was detected. The longer chain length of KDS gave a slightly shorter retention time. Figure 5B shows the MS data in which the molecular ion at m/z 356 corresponded to [M + H]⁺ of C22:0 KDS formed from l-serine and arachidoyl CoA. As shown in Fig. 5C, the MS/MS spectrum of the m/z-356 ion indicated the presence of ions of m/z 338, 326, and 320, arising from the neutral loss of H₂O, HCHO, and 2H₂O, respectively. The structures of the other KDS with different chain lengths were also confirmed in the same way.

• Open in new tab
• Download powerpoint

FIG. 5.

HPLC/ESI-ion-trap mass spectroscopy of reaction products with various chain lengths formed by the S. multivorum SPT. (A) Ion chromatography of the reaction products with various chain lengths formed by the S. multivorum SPT. (B) MS data for the reaction product of the S. multivorum SPT from l-serine and arachidoyl CoA (C20:0). (D) MS-MS data for the reaction product of the S. multivorum SPT from l-serine and arachidoyl CoA (C20:0).

Figure 6 shows the substrate concentration dependency of the reaction rate of the bacterial SPTs. It has been reported that eukaryotic SPT is inhibited by palmitoyl CoA concentrations greater than 50 μM (20, 53). A similar phenomenon was observed with the B. stolpii SPT, which was significantly inhibited by palmitoyl CoA concentrations greater than 100 μM (Fig. 6C). The SPTs of S. multivorum and S. spiritivorum were not inhibited by the high concentrations of palmitoyl CoA. For the last two enzymes, we could analyze the experimental data under steady-state conditions according to the ordered Bi-Bi mechanism (Fig. 6A and B) (49), and the kinetic parameters were obtained (Table 2). For all the enzymes, the K_m values for l-serine were in the range of 3 to 5 mM. The k _cat values of the S. multivorum and S. spiritivorum enzymes are apparently lower than that of the S. paucimobilis enzyme. The K_m values for palmitoyl CoA were 0.1 to 0.7 mM, except for the B. stolpii enzyme. For the B. stolpii enzyme, the K_m value for palmitoyl CoA and k _cat value could not be obtained due to the substrate inhibition by palmitoyl CoA. For comparison with other data, the v/[E_t] value in the presence of 100 μM palmitoyl CoA, where the maximum activity is obtained, is shown.

• Open in new tab
• Download powerpoint

FIG. 6.

Kinetic characterization of recombinant SPTs. The enzyme assay was performed as previously described (27). The apparent rate constants (k _app = v/[E_t]) for the KDS formation were plotted as a function of the palmitoyl CoA concentration at fixed concentrations of serine: 4 mM (open circle), 10 mM (open triangle), 20 mM (inverted open triangle), or 40 mM (open square). Each solid line represents the theoretical curve according to the initial velocity kinetics for the ordered Bi-Bi mechanism using the kinetic parameters summarized in Table 2. (A) S. multivorum SPT; (B) S. spiritivorum SPT; (C) B. stolpii SPT.

Intracellular localization of bacterial SPTs (immunolocalization).In order to examine the localization of the SPTs in the intact cell, specific polyclonal antibodies were prepared using the recombinant enzymes. As shown in Fig. 7, each antibody specifically recognized its corresponding SPT. There was no signal in the lanes containing the E. coli lysate transformed with the empty vector (Fig. 7, lanes 1, 4, and 7). The anti-S. multivorum SPT antibody cross-reacted with the S. spiritivorum SPT (data not shown). Considering the phylogenetic similarity of S. spiritivorum to S. multivorum, which might yield similar morphological results, and the technical constraints on the morphological analysis of this weak pathogenic species, S. spiritivorum was excluded from further analyses. For the B. stolpii lysate, the antibody recognized two bands, of approximately 50 kDa and 48 kDa, the latter being the same size as the recombinant SPT lacking the N-terminal 13 amino acid residues (Fig. 7, lanes 8 and 9). The thin sectioned profiles of the whole bacterial cells of Sphingomonas paucimobilis, S. multivorum, and B. stolpii are presented in Fig. 8A, B, and C, respectively. Sphingomonas paucimobilis cells have a rod shape, and the average size (width) of the cells was 0.8 μm (Fig. 8A). The cytoplasm of the Sphingomonas paucimobilis cells was characterized by the presence of clearly identifiable ribosome particles and a nonnucleoid electron-dense area. The cell envelope consisted of a one-electron-dense bilayer as an outer membrane and one additional bilayer structure as an inner membrane. The ribosome particles showed a condensed distribution near the inner membrane or the nonnucleoid electron-dense area. The shape of the S. multivorum cells was also rod type, and the average size (width) was 0.45 μm (Fig. 8B). The S. multivorum cells were slightly shorter than the Sphingomonas paucimobilis cells and had no flagella. At the inside of the cell wall, a multilayered inner membrane structure was observed. The B. stolpii cells had the shape of curved rods or spheres with a rugged cell wall (Fig. 8C). The average size (width) of the cells was 0.36 μm. As with Sphingomonas paucimobilis and S. multivorum, a multilayered cell membrane structure was seen inside of the cell wall of the B. stolpii cells. The ribosome particles were widely distributed in the electron-dense cytoplasm. In the middle of the cytoplasm, an organelle-like multilayered membrane structure was observed. The intracellular localization of SPT was analyzed for each bacterium using immunoelectron microscopy. The immunogold-labeled SPT was readily detectable as a spot-like distribution throughout the cytoplasm in the Sphingomonas paucimobilis cells (Fig. 8D). The nonnucleoid electron-dense area observed in these bacterial cells was more intensely immunostained. Some of the immunogold clusters seemed to localize near the inner membrane of the cell. Sparse immunogold particles were also detected on the outside of the cell. In the S. multivorum cells, about 88% of the immunogold-labeled SPT was distributed near the inner membrane of the cell, and the remaining immunogold particles were detected in the center of the cell (cytoplasm) (Fig. 8E). In the B. stolpii cells, the immunogold-labeled SPT was detected in a spot-like pattern predominantly concentrated in a limited region near the inner membrane or organelle-like multilayered membrane structure (Fig. 8F). When each primary antibody had been preabsorbed with the corresponding SPT protein, the signals disappeared or at least became very faint (Fig. 8G to I).

• Open in new tab
• Download powerpoint

FIG. 7.

Specificities of anti-SPT antibodies. The specificities of the antibodies against each bacterial SPT were confirmed by Western blot analysis with cell lysates: lanes 1, 4, and 7, E. coli BL21(DE3)(pLysS) (pET21) (empty vector); lane 2, E. coli BL21(DE3)(pLysS)(pET21-aSPT) (expressing the Sphingomonas SPT); lane 3, Sphingomonas paucimobilis; lane 5, E. coli BL21(DE3)(pLysS)(pET21-bSPT) (expressing the S. multivorum SPT); lane 6, S. multivorum; lane 8, E. coli BL21(DE3)(pLysS) (pET21-dSPT) (expressing the B. stolpii SPT); lane 9, B. stolpii.

• Open in new tab
• Download powerpoint

FIG. 8.

Electron microscopy of sphingolipid-containing bacteria. Upper panels, morphological examination of sphingolipid-containing bacteria. Ultrathin sections of the bacterial cells were negatively stained and examined by electron microscopy. (A) Sphingomonas paucimobilis; (B) S. multivorum; (C) B. stolpii. Middle panels, the intracellular localization of SPT. The localization of SPT was analyzed by the postembedding immunogold-labeling method of electron microscopy. Ultrathin sections of bacterial cells were treated with anti-SPT antibody and stained with gold-labeled secondary antibodies. (D) Sphingomonas paucimobilis; (E) S. multivorum; (F) B. stolpii. Arrows indicate the condensed distribution of immunogold particles. Lower panels, the negative control section. Ultrathin sections of bacterial cells were treated with a preabsorbed primary antibody solution and stained with gold-labeled secondary antibodies. An excess of the purified SPT protein was added to the primary antibody working solution, and the solution was incubated at 4°C for 48 h before it was added to each section. (G) Sphingomonas paucimobilis; (H) S. multivorum; (I) B. stolpii. Size markers are indicated.

Catalytically important amino acid residues are conserved in bacterial SPTs.SPT belongs to the α-oxamine synthase family of the PLP-dependent enzymes, which includes ALAS, 2-amino-3-ketobutyrate ligase (KBL), and AONS (1, 2, 4, 32, 48, 63). Bacterial SPTs show about a 30% identity with other members of this family. Previous X-ray crystallography on AONS and KBL from E. coli suggested catalytically important active-site residues that interact with PLP and are completely conserved in all of the bacterial SPTs. These conserved residues include (in S. multivorum numbering) Lys244, which forms a Schiff base linkage with PLP, Asp210, which forms a salt bridge/hydrogen bond with the pyridine N of PLP, His213, which hydrogen bonds to 3-O of PLP, and His138, which stacks with the pyridine ring of PLP (Fig. 2A). The structures of the complexes of AONS and KBL with substrate analogues suggested that Asn52 and Arg367 are the potential hydrogen-bonding partners of the carboxylate group of the substrate l-serine in the external aldimine complex of SPT. Arg367 is indeed conserved in all of the SPTs, but Asn52 is partially conserved; it shifts by one residue in the Sphingomonas and Z. mobilis SPTs.

The sequence similarities between the bacterial SPTs and human LCB2 are higher than those between the bacterial enzymes and human LCB1. The active-site residues described above are not conserved in the human LCB1; Lys244, His213, and His138 are replaced by Asn, Leu, and Cys, respectively. These residues in LCB1 cannot functionally substitute for the corresponding residues of the other α-oxamine family enzymes, including LCB2. Apparently, LCB1 does not have a catalytic function. This is consistent with the longer branch length of LCB1, because the lack of functional constraints on the protein is considered to accelerate the evolution rate.

Human hereditary sensory neuropathy type I-related mutation site of bacterial SPT.The single mutations (C133Y, C133W, V144D, or G387A) in human LCB1 cause HSN1, which is the most common hereditary disorder of peripheral sensory neurons (6, 11, 61). It remains elusive how the LCB1 mutations cause changes in the SPT activity of the heterodimer and how these changes participate in the neurodegenerative symptoms in HSN1. The dominant-negative inhibition of the SPT activities by overexpression of the HSN1-related LCB1 mutants in yeast, CHO cells, and transgenic mouse strongly supports the idea that the neurodegeneration in HSN1 is related to a decrease in the sphingolipid synthesis (7, 17, 40). However, there is a claim that the remaining SPT activity is sufficient for the normal sphingolipid metabolism and viability of the HSN1 patient cells (12). An alternative mechanism is that the HSN1 mutations of LCB1 accelerate the aggregation of the SPT protein induced by hypoxia in the human lymphocytes, which leads to nonapoptotic death (13). Contrary to these observations, de novo glucosyl ceramide synthesis increased in the lymphoblast cell lines from HSN1 patients (11), suggesting that the neural degeneration in HSN1 is due to the overproduced ceramide or the abnormal cellular lipid composition. Cys133 and Val144 of human LCB1 correspond to Cys78 and Ile90, respectively, of S. multivorum SPT. These residues are also conserved in the SPTs of S. spiritivorum and B. stolpii but not in Sphingomonas paucimobilis and Z. mobilis sequences, in which Cys133 and Val144 correspond to Thr and Asp/Gly, respectively. It should be noted that Sphingomonas SPT, which carries an amino acid substitution equivalent to the HSN1-type mutation (Asp at the position of Val144 in human Lcb1p), has a much higher activity than other bacterial enzymes carrying a hydrophobic amino acid (Ile) at that position. These findings lead us to speculate on the possibility that the HSN1-type mutations in human LCB1 may result in some toxic gain of function, such as increased activity toward normal or abnormal acyl CoAs.

Immunolocalization of bacterial SPT as a peripheral membrane protein.All of the bacterial SPTs examined so far are water-soluble homodimeric enzymes. Their water-soluble character is the most different aspect in relation to the membrane-bound enzymes of the eukaryotes. The reaction product of SPT, KDS, is a very hydrophobic sphingolipid intermediate that is easily incorporated into membranes. While membrane localization of the eukaryotic SPT complex seems very reasonable, the characteristics of the bacterial SPT as a water-soluble protein in vitro raised questions about the mechanism of the product release or the interaction with the cellular membrane in vivo. Therefore, we further analyzed the intracellular localization of the bacterial SPT by immunoelectron microscopy (Fig. 8). These results demonstrated the limited distribution of SPT molecules in bacterial cells in a spot-like pattern (Sphingomonas paucimobilis) or a clearly condensed pattern near the inner membrane of the cells (S. multivorum and B. stolpii). These results are different from the homogenous distribution pattern of the general soluble proteins. S. multivorum and B. stolpii SPT may loosely bind to the inner membrane of bacterial cells like a peripheral membrane protein or may indirectly interact with the membrane via some anchor protein in vivo. The KDS released from these SPTs may directly enter the inner membrane and then be efficiently converted into various glycosphingolipids as the final products by other modification enzymes, which may be present in the bacterial cell membrane.

Most eukaryotic-like SPT from B. stolpii.The SPT from B. stolpii is different from the other bacterial SPTs in the following ways. First, during purification, the recombinant enzyme required the addition of 20% glycerol, which is the most general stabilizer for membrane proteins. Second, B. stolpii has an organelle-like multilayered membrane structure within the cell, and the immunoreaction to the native SPT was detected on the cytosolic face of the organelle-like structure. Third, the reaction rate is the lowest among the bacterial SPTs examined and is on the same order as the eukaryotic enzymes. Fourth, when both l-serine and palmitoyl CoA were added to the enzyme, a transient accumulation of the quinonoid intermediate was spectroscopically detected, suggesting a slow catalytic turnover of the enzyme. Finally, inhibition by palmitoyl CoA as seen in the mammalian SPTs was observed. These data indicate that the B. stolpii SPT is the most eukaryotic-like enzyme among the bacterial STPs, although it is a bacterial homodimeric SPT and in this respect different from the heterodimeric enzymes of the eukaryotes.

The physiological function of the sphingolipids in bacteria is unknown except for their role as a main component of the bacterial cell membrane. The broad acyl CoA specificity of bacterial SPT might be advantageous to bacteria in that they can escape from the influence of the environmental changes on bacterial envelope properties that affect their survival. If the fatty acids available to the bacterial cells are changed, their SPTs can utilize other acyl CoAs with different chain lengths for LCB synthesis. On the other hand, there is no current report that these organisms produce sphingolipids of various LCB lengths depending on the substrate fatty acids to which they are exposed. Considering that the ¹⁴C-labeled palmitic acid in the culture medium is selectively incorporated into the LCB of Sphingomonas under normal conditions (70), it seems more probable that these organisms make essentially the same sphingolipid LCBs regardless of the fatty acid composition in the environment. This is a future research topic.

There is an interesting report demonstrating that the predatory bacterium Bdellovibrio bacteriovorus UKi2 contains phosphosphingolipids, and the mutant strain UKi1, lacking sphingolipids, loses its parasitic ability (56). The sphingolipids of Bdellovibrio may be associated in some way with the ability of this bacterium to attack and grow on suitable bacterial hosts. It has been reported that the glycosphingolipids from Sphingomonas sp. and Ehrlichia muris were recognized by the CD1d-restricted NKT cells in the mouse and human, which provide an innate-type immune response (34, 39, 55, 66). We have found putative SPT genes among the genome databases of some pathogenic bacteria and have already examined the SPT activities of these gene products. These pathogenic bacteria may also have cell walls containing glycosphingolipids that serve as direct targets for the NKT cells. Another example of the possible involvement of sphingolipids in pathogenesis is shown by a recent finding that the marine planktonic pathogen Coccolithovirus EhV-86 has a cluster of genes highly homologous to the enzymes of the sphingolipid metabolism (65) and these genes are coordinately expressed within 2 h of infection (3). This viral SPT is a unique monomeric enzyme, which is composed of two separable domains, suggesting a fused heterodimer corresponding to the eukaryotic SPTs (18). This viral SPT expressed in the yeast cell was localized to the endoplasmic reticulum and preferred myristoyl CoA (C14) to palmitoyl CoA (C16) as the substrate. It was suggested that the viral SPT may alter the sphingolipid metabolism of the host during pathogen infection.

We obtained novel SPT molecules with various characteristics, from the highly stable and water-soluble type to the eukaryotic-like and loosely membrane-bound type. Recently the recombinant SPT from S. multivorum yielded crystals of sufficiently good quality for X-ray crystallographic analysis. Further structural analyses of the SPT complex with the substrate, product, or analogues are now under way. Not only the S. multivorum SPT but also other bacterial enzymes will be useful because they have the characteristic spectroscopic features reflecting the intermediate accumulation of each step in the catalytic cycle. The reaction mechanism of SPT could be clarified in the context of the three-dimensional structure of the bacterial SPT. Information obtained from the bacterial enzymes will provide clues to the reaction mechanism of the more complex eukaryotic homologue.