Structural Requirements for Catalysis and Membrane Targeting of Mammalian Enzymes with Neutral Sphingomyelinase and Lysophospholipid Phospholipase C Activities Analysis by Chemical Modification and Site-directed Mutagenesis

The sequence similarity with bacterial neutral sphingomyelinase (NSM) resulted in the isolation of putative mammalian counterparts and, subsequently, identification of similar molecules in a number of other eukaryotic organisms. Based on sequence similarities and previous characterization of the mammalian enzymes, we have chemically modified specific residues and performed site directed mutagenesis in order to identify critical catalytic residues and determinants for membrane localization. Modification of histidine residues and the substrate protection experiments demonstrated the presence of reactive histidine residues within the active site. Site directed mutagenesis suggested an essential role in catalysis for two histidine residues (His136 and His272) which are conserved in all sequences. Mutations of two additional histidines (His138 and His151), conserved only in eukaryotes, resulted in reduced NSM activity. In addition to sphingomyelin, the enzyme also hydrolyzed lysophosphatidylcholine. Exposure to an oxidizing environment or modification of cysteine residues using several specific compounds also inactivated the enzyme. Site-directed mutagenesis of eight cysteine residues and gel-shift analysis demonstrated that these residues did not participate in the catalytic reaction and suggested the involvement of cysteines in the formation/breakage of disulfide bonds which could underlie the reversible inactivation by the oxidizing compounds. Cellular localization studies of a series of deletion mutants expressed as GFP-fusion proteins, demonstrated that the transmembrane region contains determinants for the ER localization. were subjected to analysis by steady-state kinetics. Reactions were performed under the standard assay conditions with initial velocities measured at SM concentrations ranging from 22 to 350 m M (1.4-22 mol%). The kinetic data were fitted to the Michaelis-Menten equation and kinetic parameters (K m , V m , k cat , k cat /K m ) were determined from secondary Linewear-Burk plots. further studies of defining the role for these enzymes in mammalian cells.


Introduction
Hydrolysis of sphingomyelin (SM) by sphingomyelinases (SMases), with the subsequent generation of ceramide, is a signalling pathway implicated in a number of cellular responses (1)(2)(3). Ceramide has been suggested to play important roles in cell cycle arrest, apoptosis, inflammation and the regulation of the eukaryotic stress response. Although ceramide can be generated by de novo synthesis through ceramide-synthase, for the majority of cellular responses it is generated from sphingomyelin by the action of neutral or acidic sphingomyelinases (1)(2)(3). These enzymes are sphingomyelin-specific phosphodiesterases that hydrolyze the phosphodiester bond of sphingomyelin yielding ceramide and phosphocholine. gluthatione and reactive oxygen species (8). However, the function of this enzyme in signalling and its activity towards cellular SM has not been clearly demonstrated and would require further studies (9,10). Nonetheless, using an antisense strategy, Tonnetti et al. (11) showed that the cloned enzyme could be involved in ceramide-mediated apoptosis triggered by TCR activation.
Recently, it has also been shown that the enzyme has phospholipase C activity towards specific lysophospholipids (9). Based on activities detected in vitro, we refer to the mouse and human clones as mammalian enzymes with neutral sphingomyelinase and lysophospholipid phospholipase C activities (NSM/LysoPLC).
Although some insights into the properties of the cloned enzyme have been described, the functional significance of sequence similarities and differences between bacterial and mammalian proteins has not been investigated. In this study we aimed to identify residues that are essential for catalysis and those involved in reversible inhibition by reactive oxygen species.
Determinants for localization to the ER, a specific property of the eukaryotic enzymes, have also been analyzed. Our data suggest that bacterial and mammalian enzymes have a common catalytic mechanism involving conserved histidine residues. A property that is not shared with the bacterial enzymes is the redox state dependent reversible regulation of activity which could involve the formation and breakage of S-S bonds between cysteine residues, while the transmembrane region contains the main determinants for the ER localization. by guest on April 26, 2019 http://www.jbc.org/ Downloaded from A series of deletion mutants were also made by a PCR approach using pEGFP-C1 as a vector, generating N-terminally GFP-tagged proteins. The oligonucleotides are listed in Table IB.
Heterologous Expression and Purification of Proteins-Transformed bacteria were induced with 0.2 mM isopropyl-1-thio-β-D-galactopyranoside and grown at 18 o C for 18 hours. GST-fusion proteins were prepared as described previously (8). Eluted GST-fusion protein was buffer exchanged into 25 mM Tris-HCl, pH 7.4, 1 mM DTT, 0.1% Triton X-100, 10% glycerol and stored at -20 o C. Purity of the preparation was more than 90% and only minor protein contaminants could be detected.
PolyHis-NSM/LysoPLC protein was prepared essentially as described for GST-NSM/LysoPLC but no DTT was included in the buffers. The supernatant was added to 2 ml of nickel resin incubated with rotation at 4 o C for 90 minutes. Subsequently, the resin was washed with 20 mM sodium phosphate, 500 mM NaCl, 0.2% Triton X-100, pH 6 and proteins eluted with 250 mM imidazole in the same buffer. The eluate was buffer exchanged into 25 mM Tris-HCl, pH 7.4, 0.1% Triton X-100, 10% glycerol and stored at -20 o C.
Myc-tagged proteins were expressed in Cos cells as previously described (8). A postnuclear supernatant was prepared and used in the disulfide bond-sensitive electrophoretic shift assay described below.
Analysis of subcellular localization-Constructs of various deletion mutants with the GFP tag at the N-terminus, were transfected into Cos cells as previously described (8). Twenty four hours after transfection, the cells were fixed with 4% formaldehyde for 10 minutes, washed in PBS and mounted prior to visualisation using confocal microscopy.
For expression studies (Western blotting) and activity measurements of the deletion mutants, the post-nuclear supernatant (PNS) was prepared (8) and analysed as described below. were measured using radiolabelled substrate as previously described (8

Results
Sequence Alignments of Eukaryotic Enzymes with Bacterial NSM. In addition to the mammalian enzymes cloned according to sequence similarity to bacterial NSM, other eukaryotes including yeast, nematodes, fruit fly and silk worm, have homologous sequences. Several of these sequences were aligned in Fig. 1 showing that the mouse enzyme, used in our studies, shared 20%, 28%, 42% and 81% identity with Bacillus cereus, yeast (S.pombe and S.cerevisiae), C.elegans, and human enzymes, respectively. The regions of strongest conservation (shaded in Fig. 1) are also found in other phosphodiesterases although overall similarity was very low. is not carried out by a His in a similar position in the mammalian sequence but by a residue unique to eukaryotic enzymes (7). The comparison of NSM sequences reveals the presence of two His residues (His138 and 151) in the vicinity of His 136, which are only present in eukaryotic sequences (Fig. 1).
In addition to the regions of similarity with bacterial NSM, all eukaryotic sequences have a C-terminal extension (Fig. 1). This region is predicted to incorporate two transmembrane domains (residues 325-346 and 353-375 in mouse sequence); this is consistent with mammalian enzymes being integral membrane proteins (8,10). Despite some similarities, there are other specific features of different sequences. For example, murine and human enzymes are characterized by a high content of Cys residues, eight of which are conserved between the two sequences ( Fig. 1). Our previous study has demonstrated that some Cys residues are highly reactive and that the reduced state is essential for activity (8). These properties are not shared with the bacterial enzymes and in the case of B. cereus (23) formation of one disulfide bridge is required for the enzyme activity while the addition of reducing agents has an inhibitory effect.
To examine the functional significance of structural similarities and differences described above, we analysed the murine enzyme using chemical modification of specific residues and mutagenesis. In particular, we focused on the function of His and Cys residues and the C-terminal extension present in eukaryotic enzymes.

Chemical Modification of Histidine and Cysteine Residues.
For the chemical modification studies, mouse enzyme expressed as a GST fusion protein was used. This recombinant enzyme was produced and purified from bacteria and has been shown to have properties identical to protein expressed in either insect or mammalian cells (8). For the specific modification of His residues, DEPC is the most widely used reagent (24). To determine whether His residues are important for the enzyme activity towards SM, recombinant enzyme was preincubated with increasing concentrations of DEPC and residual NSM activity measured. As shown in Fig Cys residues conserved between mouse and human sequences were mutated since the activity of the human enzyme is also redox sensitive and inhibited by sulfhydryl reagents (data not shown), as described for the mouse enzyme (8).
Each single mutant was constructed by three step PCR using mutated primers (Table IA) Table II, the specific activity of His136Ala and His272Ala mutants were 0.5% that of WT enzyme, while the His138Ala and His151Ala mutants were 47% and 11%, respectively. In contrast to the His mutants, the specific activity of each Cys mutant was similar to the WT enzyme with specific activities ranging from 100% to 115%.
These results showed that the two highly conserved residues His136 and His272, and to a lesser extent His151, are important for the enzyme activity. However, the eight conserved Cys residues have no catalytic role since their replacement by alanines did not impair the catalytic activity.
Given that the mouse enzyme contains 17 Cys residues, it is likely that inhibition of the enzyme by sulfhydryl reagents is due to steric hindrance resulting from the alkylation of Cys residues close to the active site.
Further Characterization of Histidine Residues. In addition to the analysis of levels of expression and mobility (Fig 3A), a comparison of the fluorescent spectrum of the WT enzyme and His mutants (Fig. 3C) showed that replacements by Ala did not cause detectable conformational changes to these proteins. These data rule out that mutagenesis-induced gross structural changes could underlie inactivation of the His mutant enzymes.
To confirm that essential catalytic His residues are within the active site of the mouse enzyme, the protein was preincubated with DEPC in the presence of increasing concentrations of SM, PC or MgCl 2 . In the presence of SM substrate protection of the enzyme against DEPCinactivation was observed (Fig. 4). Under the same conditions, no protection was obtained with PC ( Fig. 4), which is not a substrate for this enzyme (8), or with MgCl 2 (data not shown). These results demonstrated that essential DEPC-sensitive His residues are present in the active site.
This data also suggested that these His residues were not involved in the chelation of the magnesium cation.
We also performed steady-state kinetics and characterization of His138Ala and His151Ala mutants that still possessed significant enzyme activity. Steady-state kinetics parameters (V m , K m , k cat and k cat /K m ) were determined from Linewear-Burk plots of standard  5) show that despite their reduced activity compared to the WT enzyme, these two mutants can be totally inhibited by DEPC. These data demonstrate that full inhibition of these mutants requires the inactivation of the other essential His residues in the active site.
As described earlier, in addition to SM, the WT enzyme was previously shown to also hydrolyze lysophospospholipids with the choline headgroup, such as lysoPC (9 for NSM activity and also in a "disulfide-sensitive mobility shift" assay used in similar studies Mapping of determinants for the ER localization -Although cloned mammalian enzymes do not have recognized ER retention signals (29), they were found in this compartment (8,10). Therefore, to determine the region of mouse NSM/LysoPLC required for the ER localization, a series of deletion mutants from the N-and C-terminus were made (Fig. 7A). The expression of all proteins, containing the GFP tag, was confirmed by Western blotting (Fig.7B). The wild type and all of the mutants, except those consisting solely of the transmembrane domains, were assessed for NSM activity. While the expression of the wild type protein resulted in 100-150 fold increase of the enzyme activity in COS cell extracts, the activity of all mutants was identical to background levels (data not shown). Even the mutant that lacks the two transmembrane domains but includes entire region of similarity with the bacterial enzyme (residues 1-287) had no detectable enzyme activity.
When the localization of the mutants was examined, it was found that deletions from the N-terminus and from the C-terminus leaving the first transmembrane region (TM1) intact (e.g. protein containing residues 110-350) had the same localization to the ER as the wild type. The removal of both TM regions, however, in the 1-287 deletion mutant resulted in a loss of the ER localization (Fig. 7C). The importance of the TM1 region for the ER localization was further demonstrated by the study of GFP-fusion protein incorporating only residues within the TM1 region (residues 320-346). As shown in Fig. 7C

Discussion
The isolation of mammalian proteins with sequence similarity to bacterial NSM (6) opened a possibility that these proteins could be involved in the regulated generation of ceramide known to be important for a number of cellular functions (1)(2)(3). However, related bacterial and mammalian lipid-hydrolyzing enzymes (e.g. PI-PLC (7)) could have a number of different properties including critical catalytic residues, substrate specificities, regulatory mechanisms and determinants of cellular localization. Characterization of these properties is important for the understanding of their cellular functions.
The comparison of eukaryotic sequences with bacterial NSM, together with the secondary structure prediction, suggested that the eukaryotic enzymes contain a domain involved in catalysis (adopting the DNAse I-like structure) and a unique transmembrane domain incorporating two membrane spanning regions (Fig. 1). Based on this comparison, it is also likely that all enzymes share the same catalytic mechanism i.e. general acid/base mechanism involving two histidine residues (4). Chemical modification of histidine residues and the substrate protection experiments demonstrated that the murine enzyme contains essential histidines which are present within the active site ( Fig. 2 and 4). Subsequent site-directed mutagenesis ( Table II) has shown that mutations His136Ala and His272Ala at positions corresponding to general base and acid in other phosphodiesterases, resulted in a great reduction of the enzyme activity consistent with their proposed function. Mutational analysis of the corresponding residues in bacterial NSM and DNAse I had a similar impact on activity of these enzymes (4). It has also been reported (10) that the His272Asn mutation in the human enzyme resulted in a loss of NSM activity, as determined in transiently transfected HEK 293 cells.
However, in those experiments effects on folding and stability could not be ruled out. In our studies, purified protein was used and the analysis of fluorescence spectra excluded the possibility of large conformational changes. In addition to these two histidines, the replacement of two other His residues (His138 and His151), present only in eukaryotic sequences, had somewhat smaller effects on the enzyme activity. Although their role is not clear, kinetic analysis and fluorescence spectra of purified proteins suggest that the mutations affected the catalytic rate rather than substrate binding or overall protein folding (Fig. 3, Tables II and III). to membrane phospholipids (8,22). However, hydrolysis of lysophopholipids such as lysoPC at a much lower rate (0.5-5%) compared to SM, has been reported for bacterial NSMs (22,30).
Lysophospholipids are also hydrolyzed by mammalian ASM (31) and the cloned mammalian enzymes similar to bacterial NSM (9) consequently designated as NSM/lysoPLC. As described in Results, we have shown that the mammalian enzyme had a lower ratio of SM/lysoPC hydrolysis when compared directly to the B. cereus NSM. Furthermore, in cells stably expressing the human clone, accumulation of the product of lysoPAF and not of SM hydrolysis have been detected (9). These data support the possibility that in cells lysoPAF rather than SM could be used as a substrate.
Data described in our previous studies (8) and the data presented here, demonstrate that both mammalian sequences (mouse and human) contain highly reactive Cys residues and that the reduced state is essential for the activity. This is further supported by the chemical modifications of these residues (Fig. 2). However, mutational analysis of Cys residues conserved between the mouse and human enzyme (Table II)  Since NSM/lysoPLC resides in the ER (8,10), it is difficult to consider the redox environment and its possible changes without knowing the membrane topology of the enzyme. Even if the catalytic part is luminally oriented and exposed to the oxidizing environment, the protein could still be kept in the reduced state as described, for example, for the cholera toxin (33).
Originally, finding that the mammalian enzymes localize to the ER (8,10) was surprising since they were considered as candidates for signalling NSM and expected to be present at the plasma membrane where agonist-induced SM hydrolysis had been suggested to take place (34). (35,36), demonstrating that the ER has a separate signalling machinery responding to signals known to induce stress in this organelle, do not preclude a signalling role for these enzymes. However, very low abundance of SM in the ER (37) may imply that other substrates (e.g. LysoPAF) could be hydrolyzed in this compartment. Our studies have demonstrated (Fig. 7) that the cloned enzyme requires only one of the two transmembrane domains for the ER localization and that this region was sufficient for this specific interaction.

Recent findings
Thus, the localization is not determined by specific retention sequences such as KDEL or dilysine/di-arginine motifs found in some ER proteins (29) and absent in NSM/lysoPLC, but is likely to be related to the properties of the TM regions. Analysis of different ER proteins suggested that both, the length and composition (within and in the proximity) of the TM regions could be involved in determining the ER localization (38-40). In addition, and in agreement with a recent study (10), we found that removal of transmembrane helices resulted in a loss of the NSM activity. It is therefore possible that interaction surfaces are formed between the catalytic part and the C-terminal part containing the TM regions, which could be important for the formation of the functional protein.
In summary, our studies provide further insights into the properties of the mammalian enzymes and their relationship with the bacterial NSMs. These related phosphodiesterases are likely to share a common catalytic mechanism but could have overlapping substrate specificity; in addition, the mammalian enzymes have unique properties related to possible regulatory mechanisms and the cellular localization. Our data also suggest mutations that could generate potential dominant negative (removal of catalytic histidine residues) or constitutively active (removal of cysteine residues) molecules that could help further studies of defining the role for these enzymes in mammalian cells.   Blanks were determined with DEPC alone and with enzyme alone.
B. As above, purified protein was preincubated without (control) or with specified final concentrations of different sulfhydryl-specific reagents for 15 min at room temperature.