Purification, Characterization, and Identification of a Sphingomyelin Synthase from Pseudomonas aeruginosa

Sphingomyelin synthase is the enzyme that synthesizes sphingomyelin (SM) in mammalian cells by transferring a phosphorylcholine moiety from phosphatidylcholine to ceramide. Despite its importance, the gene and/or the protein responsible for this activity has not yet been identified. Here we report the purification, identification, and biochemical characterization of an enzymatic activity that synthesizes SM in Pseudomonas aeruginosa. SM synthase-like activity was found secreted in the culture medium of P. aeruginosa, strains PA01 and PAK, whereas it could not be detected in cultures of Escherichia coli. From the medium of PAK cultures, SM synthase was purified through sequential chromatographic columns. After separation on polyacrylamide-SDS gels and visualization by silver staining, the purified enzyme showed two bands, one of ∼75 kDa and one of 30–35 kDa. Interestingly, the highly purified SM synthase preparation also showed neutral sphingomyelinase activity. We therefore investigated whether the protein we purified as SM synthase could actually be the previously identified PlcH, a 78-kDa phospholipase C known to hydrolyze phosphatidylcholine and SM in P. aeruginosa. First, the purified SM synthase preparation contained a 78-kDa protein that reacted with monoclonal antibodies raised against purified PlcH. Second, purified PlcH showed SM synthase activity. Third, using different knockout mutant strains for the PlcH operon, PlcH was found to be necessary for SM synthase activity in P. aeruginosa. Interestingly, SM synthase activity was specific to the Pseudomonas PlcH as other bacterial phospholipases did not display SM synthase activity. Biochemical studies on the Pseudomonas SM synthase confirmed that it is a transferase, similar to the mammalian enzyme, that specifically recognizes the choline head-group and the primary hydroxyl on ceramide. This SM synthase did not have reverse transferase activity. In conclusion, the Pseudomonas PlcH also exerts SM synthase activity; therefore, for the first time, we have identified a structural gene for a SM synthase.

Sphingomyelin (SM) 1 is a ubiquitous structural component of mammalian cell membranes, and its cellular levels are regulated by both synthetic and catabolic pathways. In particular, the biochemical synthesis of SM occurs via the action of a phosphatidylcholine:ceramide choline phosphotransferase (sphingomyelin synthase; SM synthase), which transfers the phosphorylcholine moiety from phosphatidylcholine (PC) onto the primary hydroxyl of ceramide, thus producing SM and diacylglycerol (DAG) (1,2). Important biological roles have been clearly established for ceramide and DAG (substrate and product of SM synthase) in the regulation of fundamental cellular functions such as proliferation and apoptosis (for reviews, see Refs. [3][4][5][6][7][8]. Therefore, it has been hypothesized that the cellular role of SM synthase goes beyond the production of SM. In fact, SM synthase could represent a key mechanism in the control of the cellular levels of ceramide and DAG and therefore would influence functions mediated by these bioactive lipids. Indeed, a number of studies have documented regulation of SM synthase activity. It has been shown, for instance, that activity of SM synthase is enhanced in conditions of increased proliferation such as regenerating rat liver (9), SV-40 transformation of human fibroblasts (10), highly malignant hepatoma (10,11), and treatment of astrocytes with bFGF (12). Additionally, it has been shown that SM synthase activity is inhibited by tumor necrosis factor ␣ in Kym-1 rhabdomyosarcoma cells before the onset of tumor necrosis factor-induced apoptosis and that this inhibition is a caspase-dependent event (13). Finally, DAG formed through activation of SM synthase has been associated with activation of NF-B in SV-40-transformed human fibroblasts (14). Despite the biological importance of SM synthase, understanding of the molecular mechanisms of its regulation is limited by the fact that no successful purification of the protein has been achieved and the gene(s) encoding for this activity remain elusive.
Although sphingolipids are primarily present in eukaryotes, some prokaryotic cells express and secrete enzymes with sphingolipid-metabolizing capability. For example, the bacteria Bacillus cereus and Staphylococcus aureus produce and secrete a phospholipase C that recognizes SM as substrate at neutral pH, and this enzyme is responsible for their hemolytic activity (15,16). Importantly, analysis of the nucleotide sequence of the neutral sphingomyelinase (N-SMase) from B. cereus has led to the identification of at least three homologous genes: two in mammalian cells (nSMase1/lysoPAF-PLC and nSMase2) (17,18) and one in yeast (ISC1) (19). The products of these genes have all proven to hydrolyze SM in vitro at neutral pH and therefore to have neutral SMase activity.
Other bacteria known to express and secrete sphingolipidmetabolizing enzymes are Pseudomonas species. SMases have been isolated from Pseudomonas aeruginosa (20), Pseudomonas fluorescens (21), and Pseudomonas cepacia (22), although sequence and substrate specificity might be different. The SMase from P. aeruginosa (PlcH) is a phospholipase C that recognizes both SM and PC as substrates, whereas the SMase from P. fluorescens only recognizes SM. In addition, an alkaline ceramidase (CDase) has been purified, characterized and identified from the culture media of P. aeruginosa AN17 (23). Importantly this bacterial CDase shares significant homology to the mammalian mitochondrial CDase (24,25).
In this study we report the existence of SM synthase activity in the culture medium of P. aeruginosa. We have purified and characterized this bacterial SM synthase activity and have found it to correspond to the previously identified PlcH.

EXPERIMENTAL PROCEDURES
Materials-All columns used for protein purification were from Amersham Biosciences. BCA protein assay kit and Triton X-100 were from Pierce. Phosphatidylcholine (brain) was purchased from Avanti Polar Lipids and NBD-C 6 -ceramide from Molecular Probes. Thin layer chromatography silica plates (60 Å) were from Whatman. Sample concentrators Microcon Centrifugal Filter Devices (YM-10) were from Millipore Corp. (Bedford, MA). Silver staining kit was purchased from OWL Separations Systems (Portsmouth, NH). Lennox L Broth Base was from Invitrogen (Paisley, Scotland, United Kingdom). Ammonium sulfate, MES monohydrate, EDTA, and phenylmethylsulfonyl fluoride were from Sigma.
Cultures of Pseudomonas for Purification-A starting culture of P. aeruginosa cells, PAK strain (kind gift from Dr. Joan Olson, Pathology and Laboratory Medicine, Medical University of South Carolina, Charleston, SC), from a frozen stock was incubated with vigorous shaking at 30°C in LB for 14 h. A 1:500 dilution of the starting culture was incubated at 30°C overnight in 6 liter of synthetic medium containing 8.2 mM NH 4 Cl, 2.5 mM K 2 HPO 4 , 74.9 mM NaCl, 21.8 mM glucose, 2.5 g/liter LB, pH 7.2. The culture was centrifuged at 6500 ϫ g for 25 min at 4°C. The supernatant was then precipitated with 60% ammonium sulfate for 3-4 h at 4°C with gentle swirling, and subsequently centrifuged at 2500 ϫ g for 30 min at 4°C. The pellet was resuspended in buffer A (20 mM MES, pH 6.5, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.005% Triton X-100). The suspension was clarified through a 0.45-m syringe filter.
Butyl-Sepharose-The filtered solution was adjusted to 1 M ammonium sulfate and applied to a 20-ml HiTrap butyl-Sepharose column (1.2 mg of proteins/ml of column) previously equilibrated with buffer B (buffer A ϩ 1 M ammonium sulfate) at a flow rate of 0.5 ml/min. The bound proteins were washed with 5 column volumes of buffer B at 1 ml/min flow rate. Proteins were eluted by decreasing the ammonium sulfate concentration from 1 M to zero applying a 6-column volume linear gradient of buffer A. Fractions of 3 ml were collected, and SM synthase activity (50 l/fraction) and protein assay were performed. The butyl-Sepharose column was then washed with 5 column volumes of double distilled water (ddH 2 O) at a rate of 2 ml/min, and fractions of 5 ml were collected. Water wash fractions were also tested for SM synthase activity and protein concentration, and fractions with peak activity were combined with fractions with peak activity from the elution step of the butyl-Sepharose.
Octyl-Sepharose-Pooled fractions with peak activity from the butyl-Sepharose column were adjusted to 1 M ammonium sulfate and loaded on a 10-ml octyl-Sepharose column (0.7 mg of proteins/ml of column) previously equilibrated with buffer B. The loosely associated proteins were washed at 0.5 ml/min for 3 column volumes of buffer B, followed by a 2.5-column volume step elution at 55% buffer B. Proteins were further eluted at a flow rate of 1 ml/min with a 1.5-column volume linear gradient of buffer B (from 55 to 0%), and the column was further washed with 2.4 column volumes. Fractions of 1 ml were collected, and SM synthase activity (50 l) and protein concentration were determined.
Blue Sepharose-Fractions with peak activity from the octyl-Sepharose column were pooled, adjusted to pH 6.0, and loaded (0.3 mg of proteins/ml of column) at 0.5 ml/min on a 2-ml Blue Sepharose high performance column equilibrated with buffer C (20 mM MES, pH 6.0, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.005% Triton X-100). A step elution with 7.5 column volumes of 12% buffer D (buffer C ϩ 1.5 M NaCl) was followed by an increasing linear gradient of NaCl (from 12 to 75% buffer D) and a column wash with 75% buffer D of 11 column volumes each. One-ml fractions were collected, and the SM synthase activity (50 l) and protein concentration were determined. The fractions with peak activity were then pooled, desalted, and bufferexchanged with buffer E (20 mM Tris, pH 7.4, 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 0.005% Triton X-100) on PD-10 desalting columns.
Mono P-The desalted, buffer-exchanged fractions were applied to a 1-ml Mono P column equilibrated with Buffer E. The column was washed at 0.5 ml/min with 7 column volumes of Buffer E, and then eluted by decreasing the pH with Buffer F (1:8 dilution of Polybuffer 74, pH 4.0). Fractions of 0.5 ml were collected and neutralized with concentrated Tris buffer, pH 7.4 -9.0. SM synthase activity (50 l) was determined.
Mono S-The fractions with peak activity from the Mono P column were combined and adjusted to 50 mM sodium acetate, pH 4.0, then loaded onto a 1 ml Mono S column previously equilibrated with Buffer G (50 mM sodium acetate, pH 4.0, 0.2 mM phenylmethylsulfonyl fluoride, 0.005% Triton X-100). A column wash at 0.5 ml/min with 7 column volumes of Buffer G preceded the elution step with Buffer H (Buffer G ϩ 1.5 M NaCl) consisting of a linear gradient (0 -0.75 M NaCl) of 10 column volumes and 20 column volumes of 1.5 M NaCl. One-ml fractions were collected in tubes containing 0.1 ml of 1.5 M Tris, pH 9.0. SM synthase assay was performed on 100 l/fraction, and active fractions were concentrated on YM-10 Microcon centrifugal filters, separated on an 8% Tris-glycine gel, and visualized by silver staining.
SM Synthase Assay-The SM synthase assay was performed using aliquots (50 -100 l depending on the column) from the fractions of interest. The substrate was prepared as a 2ϫ mixture of 40 M NBD-C 6 -ceramide and 240 M PC resuspended in 100 mM Tris, pH 7.4, 50 mM KCl, 1 mM EDTA by sonication and vortexing until clear. The substrate was diluted 1:1 with the protein, and the incubation was carried out for 1 h in the dark at 37°C. The reaction was stopped on ice by addition of a volume of chloroform:methanol (1:1, v/v) 3 times the assay volume. After vortexing, the phases were clarified by centrifugation at 2500 ϫ g for 5 min. The lower phase was transferred to new tubes, dried down, and lipids were resuspended with 30 l of chloroform:methanol (2:1, v/v). NBD-C 6 -sphingomyelin was separated from the substrate by thin layer chromatography in chloroform, methanol, and 15 mM CaCl 2 (60: 35:8). Fluorescence was measured by using a Storm 860 Imaging Analysis System from Amersham Biosciences. Data were analyzed using ImageQuant software from Amersham Biosciences.
Characterization Experiments-PlcHR 2 was purified as previously described (26) and diluted to 0.1 g/ml in 50 mM Tris, pH 7.4 for the SM synthase assay. The reaction mixture contained 16.7 pg/l of purified protein, 75 mM Tris, pH 7.4, 100 M NBD-C 6 -ceramide, and 120 M PC in a total volume of 100 l and was carried out for 15 and 30 min at 30°C. Formation of NBD-C 6 -SM was determined as previously indicated. In some experiments, radioactive PC was used instead of fluorescent C 6 -ceramide as the labeled substrate. Briefly, a 2ϫ substrate mixture containing 200 M C 6 -ceramide and 240 M [ 14 C-choline]PC (specific activity of ϳ45 nCi/nmol) was resuspended in 0.1 M Tris, pH 7.4. The substrate was diluted 1:1 with the protein (200 l final reaction volume) and incubated for 30 min at 30°C in the dark. The reaction was stopped with 1.5 ml of chloroform:methanol (2:1, v/v) and 0.2 ml of ddH 2 O. After vortexing, the samples were centrifuged at 2500 ϫ g for 5 min to separate the phases. The lower phase was collected and dried down, then resuspended with 40 l of chloroform:methanol (2:1, v/v), and separated by thin layer chromatography in chloroform, methanol, and 15 mM CaCl 2 (60: 35:8). The plate was analyzed by autoradiography, the [ 14 C-choline]C 6 -SM band was identified by comparison with an authentic standard, and the radioactivity associated with the band of interest was quantified by scintillation counting.
Neutral SMase Activity-Neutral SMase activity was determined using radiolabeled substrate delivered in a mixed micelle system. Briefly, the reaction mixture contained 0.1 M Tris, pH 7.4, 2.5 mM MgCl 2 , 0.1% Triton X-100, 10 nmol of sphingomyelin, 100,000 cpm of 14 C-labeled sphingomyelin and an aliquot from the fractions to be tested (10 -50 l) in a final volume of 200 l. The incubation was carried out at 37°C for 30 min in a water bath. The reaction was stopped by addition of 1.5 ml of chloroform:methanol (2:1, v/v) and 0.2 ml of water.
After vortexing and centrifugation at 1500 ϫ g for 5 min, 400 l from the upper phase were counted in 4 ml of scintillation fluid.
PC-PLC Activity-Purified PlcHR 2 (40 -140 pg/l) was incubated with 20 mM p-nitrophenylphosphorylcholine (p-NPPC) in 100 mM Tris, pH 7.4, and 25% glycerol, in a total volume of 200 l. Incubation was carried out at 37°C for 1 h in a 96-well plate, and production of the chromogenic product p-nitrophenyl was measured at 405 nm (26).
BCA Protein Determination-Protein concentration was determined according to the instructions from the manufacturer (Pierce).
Western Blotting-The Western blot analysis for PlcHR 2 was performed as previously described (26). Proteins were separated on a 7% PAGE-SDS gel and transferred onto nitrocellulose at 80 V for 1 h. The membranes were blocked with 5% milk in phosphate-buffered saline containing 0.05% Tween, and blotted with monoclonal antibodies (H952, 1:250 dilution) in the blocking solution. After extensive washing, the membranes were blotted with secondary anti-mouse horseradish peroxidase-conjugated antibodies (1:4000 dilution) in blocking medium for 1 h. The signal was visualized by enhanced chemiluminescence (Amersham Biosciences).

RESULTS
Purification of SM Synthase from P. aeruginosa-Given that P. aeruginosa produces and secretes enzymes with sphingolipid-metabolizing activities such as neutral sphingomyelinase and neutral/alkaline ceramidase, we wondered whether this bacterium also produced SM synthase activity. Indeed we found that two different strains of P. aeruginosa, PA01 and PAK (but not other bacteria such as Escherichia coli) produced and secreted in the medium an enzymatic activity able to synthesize SM using phosphatidylcholine and ceramide as substrates (data not shown). The activity found in the medium from PAK cultures was greater than that measured from PA01; therefore, the protocol was optimized for the purification of the secreted SM synthase activity from Pseudomonas using the PAK strain.
SM synthase present in the medium of an overnight culture was concentrated by incubation with 60% ammonium sulfate, and then the precipitated proteins were resuspended in 20 mM MES, pH 6.5. The suspension was filtered, adjusted to a final concentration of 1 M ammonium sulfate, and subjected to hydrophobic interaction chromatography using a butyl-Sepharose column previously equilibrated with the same buffer containing 1 M ammonium sulfate (buffer B) (Fig. 1A). In these conditions, most of the activity was found to bind to the matrix whereas the bulk of the proteins (ϳ84% of the total proteins), represented mainly by small size proteins and/or peptides, were found in the flow-through. Elution of the activity was observed in two distinct peaks upon decrease of the ammonium sulfate concentration, a first peak at ϳ0.43 M ammonium sulfate and a second peak at ϳ100 mM ammonium sulfate. Moreover, when the column was regenerated with a wash of double distilled water, ϳ30% of the initial SM synthase activity and 10% of proteins were eluted from the column. The fractions containing the two peaks of activity were combined with the activity recovered in the water wash, and the buffer and salt concentration were adjusted to 20 mM MES, pH 6.5, and 1 M ammonium sulfate. The total volume was applied to an octyl-Sepharose column (Fig. 1B) previously equilibrated with the same buffer. The SM synthase activity was found to bind to the column, as did most of the other proteins. Elution of SM synthase was then achieved by decreasing the ammonium sulfate concentration similar to what was observed with the butyl-Sepharose column. The bulk of the proteins eluted at ϳ50 mM ammonium sulfate (although this value has been at times lower), whereas detachment of SM synthase from the column was mostly obtained at zero salt. This column therefore almost completely desalted the preparation obtained from the previous column and allowed an additional step in the purification process (2-3-fold). Also in this case, ϳ20% of the loaded activity and proteins were still bound to the matrix after the elution step.
The water-eluted enzyme was not purified further, as it behaved differently in the next column from the one eluted with 20 mM MES, pH 6.5, and no salt, underscoring the possibility of the existence of SM synthase in different isoforms or at different stages of denaturation and/or aggregation.
Fractions with peak activity were pooled (typically fractions 228 -238; Fig. 1B), the pH was adjusted to 6.0, and the total volume was applied to a Blue Sepharose affinity column previously equilibrated with 20 mM MES, pH 6.0 (Fig. 1C). The bulk of the proteins did not bind to the column, whereas SM synthase did, and it was eluted at 0.4 M NaCl as a single peak by application of an increasing linear gradient of salt (from 0.18 M to 1.13 M NaCl). To elute the SM synthase activity in the middle of the gradient as a defined peak without too much tailing, it was important that the sample that was loaded had a conductivity between 0.6 and 3 millisiemens. In this step of purification, SM synthase was enriched ϳ20 -25-fold.
Next, fractions containing peak activity were pooled, desalted, and buffer-exchanged with 20 mM Tris, pH 7.5, using PD-10 columns. The enzyme was further purified by application to a Mono P chromatofocusing column (Fig. 1D) previously equilibrated with 20 mM Tris, pH 7.5. Proteins were eluted by decreasing the pH with an increasing linear gradient of Polybuffer 74 adjusted to pH 4.0. The isoelectric point of the protein was determined to be ϳ4.7-4.9. Because the low amount of protein applied to this and to the next column (Mono S) did not allow an accurate protein determination in the fractions, the efficacy of the columns was determined by silver staining.
Fractions at pH 4.7-4.9 were pooled together (typically a total of 2 ml), adjusted to pH 4.0 with sodium acetate (at a final concentration of 20 mM), and applied to a cation exchange Mono S column (Fig. 1E). In these conditions, SM synthase bound to the column, and it was eluted with 0.34 M NaCl, applied as an increasing linear gradient. By analysis with silver staining of the proteins separated on a SDS gel, fractions with peak SM synthase activity contained only two bands, one at ϳ75 kDa and one at ϳ30 kDa, and both correlated with the activity (data not shown).
Purified SM Synthase Also Shows N-SMase Activity-P. aeruginosa PA01 is known to express and secrete a N-SMase activity (encoded by the plcH gene) for which the theoretical isoelectric point and molecular mass are ϳ5.0 and 78 kDa, respectively (20), similar to what was observed with the purified SM synthase. Because SM synthase activity was observed in both PAK and PA01 strains, and the gene encoding for N-SMase activity in PAK is identical to the plcH gene in PA01, 2 we wondered whether the purified preparation of SM synthase from PAK also contained N-SMase activity. To this end, N-SMase activity was measured in the fractions collected from the Mono P chromatofocusing column ( Fig. 2A). As shown in the figure, not only was N-SMase activity detected in the fractions but this activity mimicked the profile of the SM synthase activity (Fig. 1D). In contrast, no significant acid SMase activity was detected (data not shown). N-SMase activity was then measured in fractions collected from the last column of the purification protocol, the cation exchange chromatography (Fig. 2B). Also in this case, N-SMase activity paralleled exactly SM synthase activity.
To confirm the existence of a N-SMase in the highly purified SM synthase preparation, a Western blot analysis was performed on the fractions with peak activity for SM synthase from a Blue Sepharose column using monoclonal antibodies raised against purified PlcH from P. aeruginosa, strain PA01 (26) (Fig. 3). As shown in the figure, fractions from the elution step with peak SM synthase activity (fractions 42-46, Fig. 3A) tested positive for the presence of a 78-kDa protein, which reacted with the specific antibodies against PlcH and that run similarly to the purified PlcH from PA01 (Fig. 3B). These results show that the preparation of SM synthase purified from PAK indeed contained also a protein homologous to the PlcH isolated and purified from strain PA01. Therefore, these results raised the question of whether the SM synthase from PAK is identical to this N-SMase.
PlcHR 2 Is Sufficient and Necessary for SM Synthase Activi-ty-Given the high homology between N-SMase from PAK and PlcH from PA01, we utilized purified PlcH from PA01 to determine whether this enzyme also had SM synthase activity in vitro (Fig. 4). The protein responsible for the N-SMase activity in PA01 (PlcH) is secreted as a complex with two accessory proteins (PlcR 1 and PlcR 2 ), believed to regulate PlcH secretion (26,27). In the purification process, however, only PlcR 2 maintains its association with PlcH, forming the PlcHR 2 complex. Therefore, purified PlcHR 2 was incubated with vesicles containing phosphatidylcholine and NBD-C 6 -ceramide, and the FIG. 1. Purification of SM synthase. A, the ammonium-precipitated proteins from the medium of P. aeruginosa culture PAK strain were applied to a butyl-Sepharose column equilibrated with Buffer B. After washing the column, SM synthase activity was eluted by decreasing the salt concentration from 1 M ammonium sulfate to 0. Fractions of 3 ml were collected. B, the active fractions obtained from the butyl-Sepharose column were pooled, adjusted to 1 M ammonium sulfate, and loaded on an octyl-Sepharose column. After washing the column, SM synthase activity was eluted by decreasing the concentration of ammonium sulfate from 0.55 M to 0. Fractions of 1 ml were collected. C, the active fractions from the octyl-Sepharose were pooled and applied directly to a Blue Sepharose column equilibrated with buffer C. After washing the column with buffer C, SM synthase was eluted with a linear gradient from 0.18 to 1.13 M NaCl. Fractions of 1 ml were collected. D, active fractions were desalted, buffer-exchanged with buffer E and applied on a Mono P column previously equilibrated with the same buffer. After washing the column with buffer E, SM synthase was eluted by decreasing the pH with buffer F. Fractions of 0.5 ml were collected and neutralized. E, active fractions were pooled, adjusted to 50 mM sodium acetate, pH 4.0, and applied to a Mono S column equilibrated with buffer G. After washing the column, SM synthase activity was eluted with a linear gradient of NaCl from 0 to 0.75 M. Fractions of 1 ml were collected. formation of NBD-C 6 -sphingomyelin was monitored as described under "Experimental Procedures." As shown in Fig. 4, purified PlcHR 2 was able to synthesize SM in a protein-dependent (Fig. 4A) and time-dependent (Fig. 4B) manner. Therefore, these results clearly show that PlcHR 2 is sufficient for SM synthase activity.
Next, we asked whether PlcH was necessary for SM synthase activity in P. aeruginosa, PA01. To this end, SM synthase activity was measured in different cultures of mutant strains lacking the genes responsible for the synthesis and secretion of PlcH. Because the plcR 1,2 genes are located 3Ј to plcH and are expressed from in-phase overlapping genes, the single deletion of plcH causes a block of transcription of the plcR 1,2 , thus delivering in fact a plcH/plcR double knock-out. In the following experiments, strains were used that were lacking in either plcR 1,2 (⌬R) or plcH and plcR 1,2 (⌬HR). As a control we employed a strain that was deficient in the neutral/alkaline ceramidase (⌬CDase) gene, which is adjacent to the PlcHR operon (Fig. 5). As expected, the ⌬R strain secreted significantly less N-SMase activity in the media compared with the parental strain (PA01) (left top panel) retaining the activity in the cells (left bottom panel). The lack of the PlcHR operon (⌬HR) caused an almost total disappearance of N-SMase activity in both medium and cells. On the other hand, absence of the CDase gene did not affect significantly N-SMase activity in either medium or cells, demonstrating that the observed effects on N-SMase activity are specific to the PlcH/R 1,2 genes. When SM synthase activity was measured in these same cultures, an almost identical profile was observed; the ⌬R strain retained the activity in the cells, and lack of the PlcHR operon caused an almost total disappearance of SM synthase in both cells and medium. Moreover, no significant difference between the parental strain and the ⌬CDase in both cells and medium was observed when measuring the SM synthase activity, thus confirming also in this case the specificity of the results (Fig. 5). These results therefore support the necessity of the expression of the PlcH for SM synthase activity in P. aeruginosa, PA01.
Free PlcR 2 Enhances SM Synthase Activity of Free PlcH-As mentioned, PlcR 1,2 chaperones PlcH out of the cells where their interaction is maintained. More recently it has also been shown that the interaction of free PlcR 2 with free PlcH increased the phospholipase activity of PlcH (26). Therefore we investigated the effect of free PlcR 2 on SM synthase activity of free PlcH (Fig. 6). As shown in the figure, addition of increasing amount of free PlcR 2 to a constant amount of free PlcH significantly augmented PlcH ability to synthesize NBD-C 6 -sphingomyelin. No significant SM synthase activity was detected from free PlcR 2 alone (data not shown). These results suggest that in the case of SM synthase, as for PC-PLC, PlcH may represent the catalytic subunit of the PlcHR 2 complex whereas PlcR 2 could represent a possible regulatory subunit.
Characterization of pH Dependence and Effects of Cations on SM Synthase Activity of PlcHR 2 -To characterize and further optimize the conditions for the SM synthase activity of PlcHR 2 , PlcH is present in fractions containing highly purified SM synthase. Fractions with SM synthase peak activity from the elution step of a Blue Sepharose column (A) were concentrated using YM-10 Centricon microconcentrators and subsequently run in a 7% SDS-polyacrylamide electrophoretic gel along with an aliquot of purified PlcH from P. aeruginosa PA01. After transfer on nitrocellulose membrane, the proteins were blotted using monoclonal antibodies against purified PlcH from P. aeruginosa PA01 (B). After reaction with the appropriate secondary antibodies, the signal was visualized by enhanced chemiluminescence. FT, flow-through from the Blue Sepharose column. the pH optimum of the reaction was determined using purified PlcHR 2 (Fig. 7A). As shown in the figure, purified PlcHR 2 showed significant activity over a wide range of pH (from 7 to 9.5) with peak activity observed between 7.5 and 8.5. At pH lower than 6.0 and higher than 9.5, activity was almost undetectable.
Next, the effects of cations were studied. It has been described that some divalent cations significantly affect PC-PLC activity of PlcHR 2 (26); therefore, the effects of these cations on SM synthase activity were investigated (Fig. 7B). As shown in the figure, SM synthase activity was effectively inhibited by zinc and to a lesser extent by nickel, thus demonstrating effects similar to those observed on PC-PLC activity. Manganese significantly inhibited SM synthase activity but at higher concen-trations than zinc and nickel, also reproducing the effects observed with PC-PLC activity. Calcium, magnesium, and low concentrations of manganese slightly but reproducibly activated SM synthase, whereas they did not significantly affect PC-PLC, suggesting a possible difference in the regulation of these activities.
SM Synthase from P. aeruginosa Is a Transferase-PlcH has been previously described as a phospholipase C that hydrolyzes both PC and SM, releasing phosphorylcholine (20). In mammalian cells, SM synthesis occurs upon transfer of a phosphorylcholine (choline-P) moiety onto the primary hydroxyl of ceramide through a transferase-like reaction (2). Therefore, it became important to determine the biochemical nature of the bacterial SM synthase activity. A screen with different donors of choline-P was set up, and the possibility of each compound to be used as substrate for the synthesis of SM was tested (Fig. 8A). As shown in the figure, free phosphorylcholine (Ch-P) was not a substrate for SM synthesis (as compared with PC). Moreover, neither CDP-choline (CDP-Ch) nor p-NPPC (often used as substrate for the phospholipase C activity in vitro) was used as substrate for SM synthesis. On the other hand, PC, platelet activating factor, and their respective lyso-forms were efficiently recognized as substrates. SM itself and its lyso-counterpart were also used as donors of the phosphorylcholine group but with significantly less efficiency. These results suggest that the bacterial SM synthase is a transferase that requires at least a fatty acyl chain on the choline-P donor molecule to be recognized as a substrate. Next, the possibility that the bacterial SM synthase would use phospholipids other than PC as substrates was investigated. As shown in Fig. 8B, none of the phospholipids tested (phosphatidylethanolamine, phosphatidylinositol, phosphatidylserine, phosphatidic acid, phosphatidylglycerol) other than PC induced formation of SM. These results suggest that SM synthase specifically recognizes the choline head group on its substrate.
On the other hand, we then determined the specificity of the phosphorylcholine-acceptor molecule (i.e. ceramide). First, the stereospecificity of the transferase reaction when ceramide was used as substrate was evaluated by directly measuring the conversion of 14 C-fatty acid-labeled C 6 -ceramide stereoisomers to their respective short chain sphingomyelins. As shown in Fig. 9A, the enzyme best operated when the erythro-C 6 -ceramide isomers were used as substrates, whereas the threo isomers showed lower activity. Importantly, these results suggest that the transferase reaction is stereoselective. Then the ability of C 6 -dihydroceramide to be recognized as substrate was evaluated as compared with the ceramide isomers (Fig. 9A), and it was found to be significantly lower than the erythro isomers and similar to the threo isomers. This observation might implicate the relevance of the presence of the 4 -5 double bond on the ceramide molecule to be efficiently recognized as a substrate. Next, the specificity of the reaction was further tested using short chain diacylglycerol, a molecule biochemically similar to short chain ceramide, and in particular having an accessible hydroxy-group in the sn-3 position. As shown in Fig. 9A, it is apparent that the bacterial enzyme does not recognize diacylglycerol as acceptor molecule for the transferase reaction. This result would suggest that the transferase reaction discriminates among acceptor hydroxyls. To further verify this hypothesis, other alcohols were tested as substrates for the bacterial enzyme (Fig. 9B) in competition with NBD-C 6 -ceramide (30 M; K m ϭ ϳ20 M). In this experiment, L-erythro-and L-threo-C 6ceramides were used as controls. As expected from the results in Fig. 9A, L-erythro-C 6 -ceramide, being a good substrate for the transferase activity, efficiently competed with the NDB-C 6ceramide for the formation of NBD-C 6 -SM (ϳ80 and 90% inhi- bition at 60 and 90 M, respectively) whereas L-threo-C 6 -ceramide was not as effective (ϳ20 and 50% inhibition at 60 and 90 M, respectively). Importantly, none of the alcohols used in the assay, namely ethyl, lauryl, or stearyl alcohol, were able to compete with the NBD-C 6 -ceramide. Thus, they were neither inhibitors nor substrates because, according to this approach, a substrate would act as a substrate competitor of NBD-C 6 -ceramide. These results therefore strengthen the hypothesis that the transferase reaction discriminates among acceptor hydroxyl groups.
Altogether, these results show that the transferase reaction is selective in the choice of the phosphorylcholine acceptor molecule at different levels; the reaction is stereoselective, it favors the presence of the 4 -5 double bond on the ceramide, and it exchanges only with the ceramide primary hydroxyl and not other hydrophobic alcohols.
SM Synthase from P. aeruginosa Has No Reverse Activity-It has been suggested that the enzyme responsible for SM synthesis in Madin-Darby canine kidney cells also has reverse activity (28). Therefore, we investigated whether this was also the case for the bacterial SM synthase. By using radiolabeled sphingomyelin in the choline moiety and diacylglycerol as substrates, formation of radioactive phosphatidylcholine was monitored. To determine the production of phosphatidylcholine, the lipids were extracted with 1.5 ml of chloroform:methanol (2:1, v/v) and 200 l of ddH 2 O, and then processed with two alternative protocols. First, the lower phase was base-hydrolyzed (causing liberation of radioactive choline-P from the newly synthesized PC), and, after a second lipid extraction, aliquots of the aqueous phase were counted (Table I). Alternatively, the whole lower phase from the first extraction was dried down and lipids separated by thin layer chromatography (TLC) (Fig. 10A). As shown in the table and in the figure, no formation of radioactive PC and therefore transferase activity was detected by either protocol, even though hydrolysis of SM occurred (Table I).
The optimum pH for SM synthesis by PlcHR 2 is between 7.5 and 8.5 (Fig. 7A); however, the reverse reaction may demonstrate a different pH optimum (a condition seen with neutral ceramidase; Ref. 29). Therefore, PC synthesis was examined at different pH by TLC analysis (Fig. 10B). Also in this case, no FIG. 5. PlcH is necessary for SM synthase activity in P. aeruginosa, PA01. Different P. aeruginosa PA01 strains (parental PA01 strain and knock-out (⌬) strains for PlcR, PlcHR 1,2 , and CDase) were grown overnight in 100% LB medium. The appropriate amount was then transferred in synthetic medium (8.2 mM NH 4 Cl, 2.5 mM K 2 HpO4, 74.9 mM NaCl, 21.8 mM glucose, and 2.5 g/liter LB) and incubated at 30°C for 12 h. At this time all the cultures reached the same optical density readings at 600 nm. The cultures were centrifuged at 10,000 ϫ g for 30 min at 4°C. The cells were separated from the media, resuspended in 50 mM Tris, pH 7.4, containing 1 mM phenylmethylsulfonyl fluoride, and lysed by sonication (two cycles of 20 s each at 10% power with intervals of 1 min on ice). The bacterial lysates were clarified by centrifugation at 5500 ϫ g for 15 min at 4°C, and the supernatants were checked for protein concentration and used for SM synthase determination. Proteins from clarified cell lysates and media were tested for neutral SMase (left panels) and SM synthase (right panels) activities. N-SMase activity was determined using 100 l from media (upper panel) and 40 l from clarified lysates (lower panel). The incubation was carried out as described under "Experimental Procedures." SM synthase activity was determined on 100 l from media (upper panel) and 50 l from clarified lysates (lower panel) using 50 M NDB-C 6ceramide and 120 M PC. After 30 min of incubation at 30°C in the dark, the reaction mixtures were extracted according to the procedure described under "Experimental Procedures." formation of PC from SM could be observed. Altogether, these results suggest that PlcHR 2 performs only the synthesis of SM and not the reverse transferase reaction.
SM Synthase Activity Is Specific to the Phospholipase C of P. aeruginosa-Other bacterial phospholipases C have been identified, such as sphingomyelinase and PC-specific phospholipase C from B. cereus (15,30) or S. aureus (16). These enzymes show similar substrate preferences (PC or SM) to the P. aeruginosa PlcH, and, therefore, they were evaluated for SM synthase activity. Purified sphingomyelinase (bSMase) or PC-PLC (bPC-PLC) from B. cereus, which gave comparable N-SMase (cpm) or PC-PLC activity to the PlcHR 2 from Pseudomonas were tested for SM synthase activity. As shown in Table  II, neither bSMase nor bPC-PLC showed detectable SM synthase activity. Additionally, several other dilutions of the two enzymes were tested for SM synthase and found to have no activity under any condition (data not shown). It has been shown that magnesium activates bSMase and upon magnesium activation bSMase would also act as a PC-PLC. Therefore, the effects of magnesium on possible SM synthase activity of bSMase were examined (Table II). Whereas PC-PLC activity could be clearly measured under these conditions (Table II), there was no detectable SM synthase activity. These results clearly show that the ability to synthesize SM is specific to the Pseudomonas PlcH.
SM Synthase Activity of P. aeruginosa Is Inhibited by D609 -Mammalian SM synthase has been previously shown to be inhibited by an inhibitor of the bacterial PC-PLC, D609 (10,12). Therefore, the effect of D609 on the Pseudomonas SM synthase was also tested. As shown in Fig. 11, preincubation of the enzyme with different concentrations of D609 significantly inhibited the reaction in a dose dependent manner. The extent of inhibition is comparable with that observed with mammalian SM synthase and PC-PLC. 3 These results provide an additional level of similarity between mammalian and bacterial SM synthase. DISCUSSION In this study we have purified and identified a protein able to synthesize sphingomyelin from the culture supernatant of P. aeruginosa. The synthesis of sphingomyelin by this bacterial SM synthase occurs through the transfer of the phosphorylcholine moiety from phosphatidylcholine to ceramide. The bacte-rial enzyme therefore acts as a transferase similar to the mammalian counterpart, and selectively recognizes as substrate glycerophospholipids carrying a choline head-group and at least one fatty acyl chain (1,2). Moreover, the bacterial transferase reaction is stereoselective as it favors erythro ceramides versus threo as substrates, and it specifically recognizes the primary hydroxyl on ceramide as other alcohols are not used as substrates. As with the mammalian reaction, synthesis of SM by the bacterial enzyme proceeds best at neutral to alkaline pH and does not appear to require a particular cation, although activity of mammalian SM synthase has been shown at times to be favored by manganese, calcium and magnesium (31). Similarly to the mammalian SM synthase, the bacterial enzyme is inhibited by D609.
On the other hand, the bacterial SM synthase is a soluble protein, whereas the mammalian enzyme has been described as a tightly membrane-bound enzyme mainly residing in the Golgi apparatus (32-37), but also present in the plasma membrane (28, 31, 32, 38 -41) and in the nucleus (42,43). Interestingly, however, during the purification protocol, the bacterial SM synthase was found to bind strongly to butyl-and octyl-Sepharose matrices, highlighting an important hydrophobic component of this soluble protein. This aspect is not surprising given the lipid-metabolizing activity of the enzyme, and thus the necessity for a lipid/protein interaction exemplified by the requirement of at least one fatty acyl chain in the donor molecule for SM synthesis.
The difference in the solubility between sphingolipidmetabolizing enzymes from Pseudomonas and the mammalian counterparts is not unique to the SM synthase protein. In fact, a similar situation is observed in the case of both neutral/alkaline ceramidase and neutral SMase. Both are secreted proteins from P. aeruginosa (23) and P. fluorescens (21), respectively, whereas their mammalian counterparts, the mitochondrial/neutral ceramidase and the nSMase2, are associated with membranes. Despite this difference, a significant homology between the amino acidic sequences can be observed. For instance, the Pseudomonas ceramidase and the mammalian mitochondrial ceramidase share 35% identity and 53% similarity. Importantly, a similar correspondence could support the possibility of finding a mammalian homologue to the bacterial SM synthase.
The bacterial SM synthase was found to be the previously isolated PlcH, a phospholipase C known to hydrolyze phosphatidylcholine and sphingomyelin, and responsible for the hemolytic activity of P. aeruginosa. The results from this work, by using a combination of biochemical and genetic approaches, demonstrate that PlcH is sufficient and necessary for SM syn-thase activity. First, the purified protein from PAK strain showed neutral SM hydrolytic activity and similar molecular weight and isoelectric point to the PlcH from PA01 strain. hexanoyl]C 6 -ceramide stereoisomers or dihydroceramide for 30 min at 30°C. At this protein concentration, K m for De-C 6 -ceramide was found to be ϳ80 M. The reaction was terminated by addition of three volumes of a mixture of chloroform:methanol (1:1), and the organic phase was dried and lipids separated by TLC using a mixture of chloroform, methanol, and 15 mM CaCl 2 (60: 35:8). The plate was analyzed by autoradiography, and the radioactivity associated with the short chain sphingomyelins was determined. In the case of short chain DAG (1oleoyl-2-acetoyl-sn-glycerol, used at 100 M), 120 M [ 14 C-choline]PC was used as substrate, and the formation of short chain [ 14 C-choline]PC was evaluated as indicated for short chain sphingomyelins. B, PlcHR 2 (2.1 pg/l) was incubated in the presence of vesicles loaded with 120 M PC, 30 M NDB-C 6 -ceramide, and the lipids and alcohols indicated in the figure for 30 min at 30°C. At this protein concentration, K m for NDB-C 6 -ceramide was found to be ϳ20 M. The reaction was terminated by addition of three volumes of a mixture of chloroform:methanol (1:1) and processed as described under "Experimental Procedures." CT, control; De, D-erythro-C 6 -ceramide; Le, L-erythro-C 6 -ceramide; Dt, D-threo-C 6 -ceramide; Lt, L-threo-C 6 -ceramide; DH, dihydro-C 6 -ceramide; DAG, 1-oleoyl-2-acetoyl-sn-glycerol.  (20). This observation brings up the possibility that the ability of PlcH to recognize SM might be a requirement for SM synthase activity.
The ability of PlcH to hydrolyze PC (PC-PLC) and at the same time to synthesize SM (SM synthase) raises the interesting issue of the potential identity of the mammalian PC-PLC/SM synthase. Based on the biochemical similarities between the two reactions (hydrolysis of PC and formation of DAG, and inhibition by D609) and possible biological phenotypes (increased activity in transformation), we previously proposed that the PC-PLC and SM synthase reactions could be regulated by the same enzyme (10). The results presented in this study importantly confirm the validity of this hypothesis, at least for the Pseudomonas enzyme.
The possibility of PlcH to catalyze different reactions (PC-PLC, SMase, and SM synthase) raises the interesting question on how these reactions might be regulated. Interestingly, several ceramidases (yeast, mammalian, and bacterial) have been reported to have dual activity, hydrolytic (ceramidase) and synthetic (ceramide synthase) (23,29,44,45). Biochemical characterization has shown that regulation of these reactions might involve different mechanisms such as different pH optima, product inhibition (sphingosine on ceramidase), and differential effects of lipid cofactors (cardiolipin and phosphatidyl-  A and B), MES, pH 6.5, or sodium acetate (Na Acetate), pH 5.5 (B). After a 30-min incubation at 30°C, the 200-l reaction mixtures were neutralized and then extracted by addition of 1.5 ml of chloroform:methanol (2:1, v/v) and 200 l of ddH 2 O. After clarification of the phases by centrifugation, the organic phases were dried down and analyzed by TLC using a mixture of chloroform, methanol, and 15 mM CaCl 2 (60: 35:8). The plates were then subjected to autoradiography.

TABLE II
SM synthase activity is specific to PlcHR 2 of P. aeruginosa PlcHR 2 , sphingomyelinase (bSMase), and PC-phospholipase C (bPC-PLC) from B. cereus were assayed for N-SMase, PC-PLC, and SM synthase activities as described under "Experimental Procedures." Magnesium was included in the reaction at a final concentration of 20 mM. *, B. cereus SMase was also assayed for PC-PLC activity in the presence of magnesium; in this case PC-PLC activity was detected (A* ϭ 0.513) whereas still no SM synthase activity could be measured. ND, not determined.  11. D609 inhibits Pseudomonas SM synthase activity. After a 15-min preincubation at 30°C with different concentrations of D609 (Biomol), PlcHR 2 (16.7 pg/l) was incubated in the presence of vesicles loaded with 120 M PC and 100 M [ 14 C-hexanoyl]De-C 6 -ceramide for additional 30 min. The reaction was terminated by addition of three volumes of a mixture of chloroform:methanol (1:1), and the organic phase was dried and lipids separated by TLC using a mixture of chloroform, methanol, and 15 mM CaCl 2 (60: 35:8). The plate was analyzed by autoradiography, and the radioactivity associated with the short chain sphingomyelin was determined. The preincubation with D609 was found to be important to obtain maximal inhibition of SM synthase activity. glycerol promote ceramidase activity and inhibit the ceramide synthase). In the case of PlcH, hydrolytic and synthetic reactions occur at similar pH optima excluding a differential regulation at different pH. Interestingly, it has been previously reported that PlcH and its chaperone PlcR 2 bind to calcium and that the latter predicted to contain four EF-hands (26). Therefore, one possible regulatory mechanism might occur through the interaction of specific cations, in particular calcium. Although no effect of calcium has been reported on PC-PLC activity and a slight but reproducible activation has been now shown with SM synthase, we found that the neutral SMase activity of PlcH was significantly inhibited by calcium in the absence of detergent. 3 This observation might suggest that PlcH hydrolyzes PC and SM through a different catalytic mechanism/site. On the other hand, it has not yet been determined whether PC-PLC and SM synthase reactions share a common hydrolytic site for PC.
Importantly, bacterial SM synthase activity is specific to P. aeruginosa as it has been found neither in other bacteria such as E. coli nor in other purified phospholipases C with similar substrate preference to PlcHR 2 . This observation is not surprising, given the lack of sequence homology between, for instance, PC-PLC and N-SMase from B. cereus and S. aureus and PlcH, which appears to belong to a distinct family of enzymes. Also this observation would lead to the prediction that the mammalian phospholipases homologue to these bacterial enzymes (such as nSMase1) should not exhibit SM synthase activity.
In conclusion, this study identifies the first structural gene for a SM synthase, providing the first molecular tool for the study of this class of enzymes. Importantly, it may potentially lead to the identification of the elusive mammalian SM synthase.