Lipid dependence and basic kinetics of the purified 1,2-diacylglycerol 3-glucosyltransferase from membranes of Acholeplasma laidlawii.

UDP-glucose: 1,2-diacylglycerol 3-glucosyltransferase (EC 2.4.1.157), catalyzes the transfer of glucose from UDP-glucose to diacylglycerol (DAG) to yield monoglucosyldiacylglycerol (MGlcDAG) and UDP. MGlcDAG is the first glucolipid along the glucolipid pathway, and a major (nonbilayer-prone) lipid in the single membrane of Acholeplasma laidlawii. MGlcDAG is further glucosylated to give the major diglucosyldiacylglycerol (DGlcDAG). The bilayer fractions of these lipids are crucial for the metabolic maintenance of phase equilibria close to a potential bilayer-nonbilayer transition and a nearly constant spontaneous curvature. The glucolipid syntheses are also balanced against the phosphatidylglycerol pathway, competing for the common minor precursor phosphatidic acid, to retain a constant lipid surface charge density. The 1,2-diacylglycerol 3-glucosyltransferase was purified to homogeneity from detergent-solubilized A. laidlawii cells by three column chromatography methods (enrichment ≈9 000 ×), and identified as a minor 40-kDa protein by using SDS-polyacrylamide gel electrophoresis. In CHAPS detergent, mixed micelles, a cooperative dependence on anionic lipids for activity was confirmed. Dependence of the enzyme on UDP-glucose followed Michaelis-Menten kinetics while the other hydrophobic substrate dioleoylglycerol stimulated the enzyme by an activating, potentially cooperative mechanism. Physiological concentrations of the activator lipid dioleoyl-phosphatidylglycerol influenced the turnover number of the enzyme but not the interaction with UDP-glucose, as inferred from variable and constant values of the apparent Vmax and Km, respectively. Dipalmitoylglycerol was a better substrate than the oleoyl species, supporting earlier in vivo and crude enzyme data. The responses of the purified 1,2-diacylglycerol 3-glucosyltransferase indicated that (i) the regulatory features of the MGlcDAG synthesis is held by the catalytic enzyme itself, and (ii) this strongly corroborates the “homeostasis” model for lipid bilayer properties in A. laidlawii proposed earlier.

MGlcDAG is the first glucolipid in the glucolipid pathway, and is further glucosylated by another enzyme to give diglucosyldiacylglycerol (DGlcDAG) (1)(2)(3), the precursor for the syntheses of two (usually minor) phosphoglucolipids. Acylated variants of the two glucolipids are also made under some conditions. The hydrophobic substrate DAG for 1,2-diacylglycerol 3-glucosyltransferase is synthesized from phosphatidic acid (PA) by the enzyme phosphatidic acid phosphatase. 2 PA is also the precursor for the separate and competing pathway leading to the major (anionic) phosphatide phosphatidylglycerol (PG). In A. laidlawii, these lipid syntheses are regulated in response to external or internal factors influencing the physical properties of the lipid bilayer to give a strongly varying but adapted lipid composition (4). The purpose is to maintain (i) phase equilibria close to a potential bilayer/nonbilayer transition (5), (ii) a nearly constant radius of spontaneous curvature (6), and (iii) a constant (anionic) surface charge density (7) for the membrane lipid bilayer mixture. Similar packing properties are also maintained in other bacteria by regulation especially at the acyl chain level, like Escherichia coli (8,9), Clostridium butyricum (10), and Bacillus megaterium (11), the latter two also by some extent of polar head group changes.
The regulation of the packing properties is mainly accomplished by altering the relative proportions of MGlcDAG and DGlcDAG, forming essentially nonlamellar and lamellar phases (5,(12)(13)(14). In vivo, parameters giving more nonlamellar properties to the bilayer, like higher growth temperature and more of longer and unsaturated acyl chains, result in an increase of the DGlcDAG fraction and a decrease of the MGlc-DAG fraction (14), as do the presence of various foreign, nonbilayer-promoting molecules (15). This follows the rules for the packing and phase equilibria of amphiphiles (16,17). In vitro, the activity of MGlcDAG synthase is critically dependent upon a lipid environment where a substantial fraction of anionic amphiphiles is present. The best activator seems to be 1,2dioleoyl-phosphatidylglycerol (DOPG), allowing the enzymatic reaction and stimulating it in a cooperative fashion at increasing mole fractions (18,19). This dependence of the enzyme is most likely involved in keeping the rate balance between the PG and glucolipid pathways in vivo, maintaining a constant surface charge density of the lipid membrane. The lamellar/ nonlamellar balance is more likely regulated by the consecutive synthesis of DGlcDAG since the activity of this step is strongly influenced by the presence of molecules increasing the bilayer chain order or the spontaneous curvature, e.g. like certain sterols, in a manner that correlates with the regulation in vivo (19,20).
A simplification of the crude systems used, and a purification of the proteins involved, would give valuable information about how a lipid-synthesizing enzyme is regulated by its amphiphilic environment. What properties of the environment, which are strongly altered by the nonlamellar-prone enzyme product, are sensed, and what are the regulatory responses of the enzyme. Are they in agreement with the regulation model observed in vivo and with the details accomplished from coarse experiments in vitro?
We report here the purification and identification of the MGlcDAG synthesizing enzyme 1,2-diacylglycerol 3-glucosyltransferase from A. laidlawii membranes. Confirmation of the regulatory properties of the enzyme, indicated from crude proteins in a vesicular system (18,19), and a characterization of its dependence on the DAG and UDP-Glc substrates in a pure micellar system, add knowledge to the regulation of lipid bilayer properties.

MATERIALS AND METHODS
Strain and Growth Conditions-A. laidlawii strain A-EF22 was cultivated in an oleic acid (C18:1)-supplemented medium and harvested, and the protein content of cells and membranes was determined, as described by Dahlqvist et al. (19).
Cell Solubilization-Harvested cell suspensions, ϳ25 mg of cell protein/ml, were mixed with a solubilization buffer to ϳ3 mg of protein/ml, 24 mM Tris maleate, pH 8, 30 mM CHAPS detergent, and 20% (v/v) glycerol, conditions as optimized by Karlsson et al. (18). After 60 min on ice, with gentle swirling every 15 min, an insoluble fraction containing no significant MGlcDAG synthesis activity was removed by centrifugation at 20,000 ϫ g for 30 min at 2°C.
Ion Exchange Chromatography (IEC)-The NaCl concentration in the solubilized cell preparation was raised to 0.4 M by adding 4 M NaCl-solution (plus glycerol to keep 20% (v/v) glycerol). A Pharmacia XK column packed with 30 ml of SP-Sepharose (Pharmacia) was equilibrated with at least 150 ml of 20 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 8, followed by 60 ml of 20 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 8, 20% (v/v) glycerol, 1 mg/ml bovine serum albumin (BSA), and 0.4 M NaCl. BSA increased the yield of enzyme, probably by blocking unspecific protein binding. Excess BSA was rinsed out by 150 ml of buffer without BSA. Prior to sample loading, the column was equilibrated with 45 ml of IEC-buffer: 20 mM NaH 2 PO 4 /Na 2 HPO 4 , pH 8, 20% (v/v) glycerol, 20 mM CHAPS, and 0.4 M NaCl. After loading of the solubilized cells, the column was washed with 76 ml of IEC-buffer. The MGlcDAG synthesizing enzyme (analyzed by activity) was eluted with a 150 ml (250 ml preparative scale) gradient of 0.4 to 1.0 M NaCl in IEC-buffer. Finally, the remaining proteins were eluted with 60 ml 1.0 M NaCl in IECbuffer. Fractions (5 ml) collected from the start of application were analyzed for protein concentration and enzyme activity and stored at Ϫ80°C.
Gel Filtration-One of the fractions with the largest MGlcDAG synthesis rate, from the ion exchange chromatography, were loaded on a 16-mm diameter column with 154 ml of Sephacryl S-100 HR (Pharmacia), which had been washed with degassed and temperature-equilibrated elution buffer (100 mM KH 2 PO 4 /K 2 HPO 4 , pH 8, 20% (v/v) glycerol, and 20 mM CHAPS). Proteins were eluted at 0.25 ml/min into 5-ml fractions, which were stored at Ϫ80°C and analyzed for protein concentration and enzyme activity. Since only one fraction at a time from the ion exchange chromatography could be loaded on the gel filtration column, the procedure was repeated for every one of the fractions that contained substantial amounts of enzyme activity. In a preparative version of the method, a larger column (95 o mm long and 26 mm in diameter) was used, which allowed 25 ml-loading volumes with retained elution profiles as shown in Fig. 1B.
Hydroxyapatite Chromatography (HAC)-A column (16 mm in diameter) packed with 6 ml of ceramic hydroxyapatite Macro-Prep (Bio-Rad) was equilibrated with HAC-buffer: 100 mM KH 2 PO 4 /K 2 HPO 4 , pH 8, 20% (v/v) glycerol, and 20 mM CHAPS. Fractions eluted before the major background protein peak in the gel filtration step were pooled and loaded on the HAC column. After washing with 6 ml of HAC-buffer, the enzyme was eluted with 50 ml of HAC-buffer consisting of a gradient of 200 -408 mM KH 2 PO 4 /K 2 HPO 4 . Proteins remaining in the column were then eluted with 38 ml of 465 mM KH 2 PO 4 /K 2 HPO 4 in HAC-buffer. Fractions of 2.5 ml were collected and analyzed with respect to protein content and enzyme activity.
For some cell batches cultivated in the presence of a 3 H-labeled amino acid mixture (19), protein concentration in the column fractions were determined by liquid scintillation counting. All solutions used throughout the chromatography steps were temperature equilibrated at 4°C, and all the procedures were performed at 4°C. The enzyme preparations were stable for several months when stored in aliquots at Ϫ80°C. In the preparative version of the methods no 3 H-labeling was used; instead, the BCA kit (Pierce) was used for protein determination. The protein composition in the various column fractions was analyzed with SDS-polyacrylamide electrophoresis, as described by Nyström et al. (21).
Solubilization of Substrate and Matrix Lipids by CHAPS-Lipid mixtures with 0.12 mM DOG, with or without 1.5 mM DOPG, DGlcDAG, DOPC, or sphingosine in 100 mM HEPES, pH 8, 20 mM MgCl 2 , and 0 -30 mM CHAPS, were prepared as the assay micellar solutions (see below). Solubilization was monitored as the decrease in absorbance at 350 nm in a 10-mm path length quartz cuvette.
Assay for 1,2-Diacylglycerol 3-Glucosyltransferase Activity-Mixedmicellar solutions were prepared by mixing lipids solubilized in 1,1,1trichloroethane to the final concentrations specified below. The solvent was evaporated under a stream of N 2 , and the lipids were dried under reduced pressure for at least 1 h. Lipid mixtures were then solubilized to homogeneity in CHAPS buffer (cf. above) by extensive vortexing, overnight incubation at 4°C, and bath sonication for 5 min. This caused a significant isomerization of DAG between its different chain stereoisomers, as revealed by thin-layer chromatography (TLC) analyses. The lipids were extracted from the micelle preparations into an equal volume of 1,1,1-trichloroethane and applied on Silica Gel 60 (Merck) TLC plates, which were developed in chloroform/acetone, 24:1 (v/v). Stock solutions of the diacylglycerols were used as references. Especially, 1,3-DAG was substantially isomerized into the sn-1,2 and sn-2,3 variants. Therefore, rac-1,2-DAG was used as substrate lipid in most experiments. The activity obtained with rac-1,2-DAG was approximately 50% of the activity with sn-1,2-DAG, indicating that the sn-2,3-DAG does not act as a substrate or as an inhibitor.
In the standard assay, 5 l of enzyme solution (50 ng of protein) was added to 85 l of micellar solution. The mixtures were kept for 25 min on ice and preincubated for 10 min at 28°C. Reactions were started by addition of 10 l of UDP-[ 14 C]glucose (37-74 GBq/mol) to give a concentration of 1 mM in a volume of 100 l in 100 mM HEPES, pH 8, 20 mM MgCl 2 , and 20 mM CHAPS buffer. Standard lipid concentration was 10 mM (0.23 mM DAG substrate, plus activator DOPG and matrix lipid). Activity from chromatography columns was analyzed in 50-l samples mixed with 40 l of micellar solution to give final concentrations of 1 mM DOG, 9 mM DOPG, 110 mM Tris maleate, pH 8, 20 mM MgCl 2 , and 20 mM CHAPS. Micellar solution preparation and assay were the same as described above. After 5 min of incubation (or 30 min for chromatography fractions) at 28°C, reactions were stopped with 375 l of meth-anol/chloroform, 2:1 (v/v), and the lipids, including newly synthesized MGlcDAG, were extracted and separated by TLC (18). MGlcDAG on the TLC plates was visualized and quantified by electronic autoradiography (Packard Instant Imager TM ). The MGlcDAG product migrated as a single spot with a unique R f -value, corresponding to synthetic or structurally determined MGlcDAG in two different TLC solvent systems (12,22). The synthesis rate for MGlcDAG (linear for at least 30 min) was expressed as nanomoles of lipid synthesized per hour, as calculated from the amount of radioactive glucose incorporated into MGlcDAG.

RESULTS
Enzyme Purification-Starting the enzyme purification by total solubilization of intact cells in 30 mM CHAPS by the method optimized by Karlsson et al. (18) (see "Materials and Methods"), instead of prepared membranes, substantially improved the final results and gain in activity. The stability of the enzyme was dependent on a high buffer concentration of glycerol (20% v/v) (cf. Ref. 18) and a low temperature (maximum of 4°C) during the entire purification procedure. After removal of insoluble particles by centrifugation, the micelle solution was loaded directly on an IEC column (SP-Sepharose).
The MGlcDAG synthase did bind (at pH 8 and 0.4 M NaCl) and was essentially separated from the major peak of retained proteins by elution with a concentration gradient of NaCl (0.4 -1.0 M) (Fig. 1A), giving a 190-fold enrichment. The total yield of MGlcDAG synthesis activity eluted was 47% of the activity loaded, but only 27% of the activity persisted in fractions usable for further purification (Table I). The endogenous lipids were most likely removed from the proteins by the excess of detergent present during the procedure. No lipid determination was performed, but the yellow color that is very characteristic for A. laidlawii lipid extracts was washed out of the column before the enzyme.
By gel filtration, the enzyme activity was eluted from a Sephadex S-100 column in two separate peaks, one before the major protein fractions and the other with them (Fig. 1B). The relative quantity of the two activity peaks varied substantially between runs, but the protein elution profiles were almost identical. This indicates an irregular behavior of the MGlcDAG synthase in this chromatography system, but there were no differences between the enzymes in the two peaks causing them to behave differently in the following hydroxyapatite chromatography procedure. However, only for enzyme from the first gel filtration peak was the specific activity improved, and these fractions, therefore, constituted the target for further purification. The activity-enrichment factor was 12, and the yield was 32% (Table I), but since the desalting and the change in buffer after the IEC gave a total yield of around 200% (for all fractions), it is likely that the actual enrichment of the enzyme by the gel filtration was lower.
The final step in the purification procedure was HAC. The best fractions from the gel filtration were pooled and directly loaded on the hydroxyapatite column. The enzyme bound firmly to the ceramic hydroxyapatite at 100 mM KH 2 PO 4 / K 2 HPO 4 , pH 8, and not until the concentration of KH 2 PO 4 / K 2 HPO 4 was raised to ϳ340 mM in a gradient was the enzyme eluted from the column in a sharp, specific peak (Fig. 1C). The enrichment factor of 4.2 is probably correct since the total yield was near 100%. The achieved yield was 76% of the activity loaded on the column. SDS-polyacrylamide gel electrophoresis of proteins in the eluted fractions revealed the occurrence of a protein of Ϸ40 kDa, the amounts of which correlated with the elution profile of the MGlcDAG-synthesis activity (as in Fig. 1C). The gel photo in Fig. 2 shows the single protein band from a high-activity HAC fraction. The N-terminal amino acid sequence of this protein was verified from two preparations (data not shown). An enrichment factor of Ϸ9,000 is of similar magnitude as for several purified eukaryotic and E. coli lipid-synthesizing enzymes, (e.g. Refs. 23 and 24). These conditions strongly indicate that the 40-kDa protein is the 1,2-diacylglycerol 3-glucosyltransferase and that it was purified to near homogeneity.
Stability and Dependence on Mg 2ϩ -The purified MGlcDAG synthase was thermo-unstable. During storage in HAC buffer, the activity half-life of the enzyme at 4°C was around 50 h, and at 20°C, it was only 1.8 h. At Ϫ20 and Ϫ50°C, the enzyme was stable for at least 48 days. Since chelating and reducing agents had only small effects on the stability of the impure enzyme, they were not tested with the purified. It has previously been shown that the MGlcDAG-synthesizing (crude) enzyme is dependent on the presence of ϳ20 mM Mg 2ϩ (1). A strong dependence of Mg 2ϩ was revealed for activity of the purified enzyme, with a maximum between 15 and 28 mM MgCl 2 added (data not shown). It is not possible to conclude if Mg 2ϩ is a cofactor bound to the enzyme or if it acts by altering the charge distribution on the enzyme or the surrounding lipid surface.
Properties of Micellar Environment-It has been reported that CHAPS is poor in solubilizing (hydrophobic) lipids with small polar headgroups like DAG, requiring the presence of large headgroup lipids (preferentially anionic) in order to obtain homogenous, mixed micelles (25). The different matrix and activator lipids were investigated for their ability to solubilize the DAG substrate in mixed CHAPS micelles. DOG (0.12 mM), applied as dry films on the tube wall, was hydrated as de- scribed under "Materials and Methods" with buffer containing 0 -30 mM CHAPS. More than 15 mM CHAPS were required for maximal solubilization. This was indicated by a decreased turbidity of the preparations (Fig. 3). Addition of 1.5 mM DOPG, DGlcDAG, DOPC, or sphingosine to the DOG significantly increased the solubility of the latter in CHAPS (Fig. 3). The endogenous matrix lipid DGlcDAG was less efficient than DOPC at CHAPS concentrations below 10 mM (Fig. 3). However, the difference was not important at enzyme assay conditions since the fraction of DOPG activator required was the same with either DGlcDAG or DOPC as matrix (Fig. 5). Sphingosine caused very low turbidities from low up to almost 20 mM concentrations of CHAPS (Fig. 3), most likely due to the detergent-like properties of sphingosine molecules.
Assay buffer with 20 mM CHAPS was used to solubilize varying concentrations of lipids with an intermediate (0.025) and a higher (0.081) fraction of DOG in DOPG. A final concentration of 10 mM lipids were fully solubilized in 20 mM CHAPS at both DOG concentrations tested (Fig. 3, inset). Hence, it can be concluded that all lipids tested helped to solubilize DOG in mixed CHAPS micelles at standard assay conditions.
Enzyme Activity Depends on Amphiphile Concentration and Fraction of DOPG-The effect of varying DOPG/CHAPS ratio was analyzed at different total amphiphile concentrations. A larger total DOPG fraction was needed for maximal activity at higher total amphiphile concentrations than at lower ones ( Fig.  4), indicating that it was the DOPG fraction in the micelles that was important for activity rather than the bulk concentration. The critical micellar concentration of CHAPS is varying with different lipid compositions, going as high as 13.7 mM (25). If an estimated CHAPS water-phase monomer concentration of 5 mM is subtracted before calculating the estimated DOPG/ CHAPS ratio (in the micelles) on the x-axis (Fig. 4), the two peaks will essentially coincide. Hence, the two curves in Fig. 4 both show the same dependence of the surface concentration of DOPG in the micelles. The decrease in activity at DOPG concentrations higher than the optima (Fig. 4) might be due to a transition from one kind of micellar structure to another (perhaps bilayer-like), caused by the high lipid to detergent ratio (see Ref. 26).
A DOPG/CHAPS ratio of 0.46 was set as standard in order to minimize the risk of exceeding the optimal fraction of DOPG at 30 mM total amphiphile. This amphiphile concentration should be safely higher than the CHAPS monomer concentration.
Activation by Anionic Lipids-Using the standard conditions (Figs. 3 and 4), the proportions of DOPG to other polar lipids in the CHAPS micelles were systematically varied. With the zwitterionic phospholipid DOPC as the other matrix lipid, the fraction of DOPG had to be above 0.3 before any activity was detectable, and with DOPG fractions higher than 0.6, the activity increased in a sigmoidal manner (Fig. 5). The endogenous, uncharged glucolipid DGlcDAG as matrix lipid yielded essentially the same results as for DOPC (Fig. 5).  The anionic phospholipid DOPS was able to activate the enzyme to the same extent as DOPG (Fig. 5), indicating that the anionic charge is more important than the specific polar head structure. However, when DOPG and DOPS were mixed in ratios between 0.8 and 0.9, an activity maximum appeared, which implies that the overall structure of the anionic charge distribution is influencing enzyme activity. It has been shown that the activation by anionic lipids can be substantially counteracted by the cationic lipid sphingosine as a matrix lipid (18). This was the case also for the pure enzyme in the micellar system, but the effect was quantitatively smaller. With sphingosine as the matrix, the activation did not start until the DOPG fraction was above 0.6, and the sigmoidal appearance was only evident at DOPG fractions over 0.7 (Fig. 5). All in all, these result with the pure enzyme corroborates previous findings gained from crude preparations regarding the activation of the MGlcDAG synthase by an anionic lipid environment.
Kinetic Characterization of the MGlcDAG Synthase-Activity of the enzyme correlated with the UDP-Glc substrate concentration in a hyperbolic manner when UDP-Glc was altered from 0 to 3.75 mM, with a sharp raise up to 0.8 mM (Fig. 6A). The data could be fitted to the Michaelis-Menten equation with high accuracy, and the parameters V max and K m could be calculated from the fitted curve (Table II). Hence, the enzyme follows Michaelis-Menten kinetics in its dependence on the soluble substrate UDP-Glc. The calculated V max is only valid for the concentration of DOG (0.8 mol %) used here, whereas the K m remained unaltered at varying DOG concentrations. To get the absolute V max , one has to either saturate the enzyme with DOG or repeat the experiment at various DOG concentrations and calculate the absolute V max by extrapolation. The former method is restricted by the solubility limit of DOG (cf. Fig. 7), and the latter method is only applicable if kinetics of both substrates can be described by known equations. Variation of experimental diacylglycerol concentration was complicated since several parameters were simultaneously altered. To keep the total amphiphile concentration constant, the activator/matrix lipid or the CHAPS concentration had to be lowered, both of which have substantial influence on the activity (Fig. 3 and 4) and alter the micellar structure. However, DOG concentrations between 0.25 and 5 mol % only shift the total amphiphile concentration from 29.3 to 30.8 mM, which has a  (Table II). Curves in B were calculated from the Hill-equation fitted to the data points. Since the data points are only covering a very small range of the calculated Hill curve, no estimation of kinetic parameters was given. smaller effect on the enzyme activity.
Variation of the DOG concentration between 0 and 5 mol % revealed a sigmoidal correlation for the enzyme activity rather than a hyperbolic one (Fig. 6B). The data could be fitted to the Hill equation, but the parameters V max and K m could not be determined since the range of experimental data only covered a limited part of the calculated curve. However, it was obvious that DOG had an activating effect on the MGlcDAG synthase that could not be explained by the small increase in total amphiphile concentration. In vivo, DOG fractions are very low in membranes with unsaturated acyl chains (Ͻ1 mol %) (14).
The substrate dependence of the enzyme basically followed the same pattern at three DOPG concentrations; this was evident for both substrates (Fig. 6, A and B). Calculated values of V max and K m for UDP-Glc (Table II) revealed that the catalytic rate of the enzyme was altered by the variation in DOPG concentration, but that K m remained essentially the same. From this, it can be assumed that DOPG does not influence the MGlcDAG synthase activity by altering the interaction between the enzyme and UDP-Glc, but instead is affecting the turnover number of the enzyme reaction.
Hence, the MGlcDAG synthase is activated, in a potentially cooperative fashion, by its lipid substrate, DAG, and it follows Michaelis-Menten kinetics with respect to its other soluble substrate, UDP-glucose. The activator DOPG does not affect the interaction of the enzyme with UDP-glucose but increases the turnover of the enzymatic catalysis in a cooperative manner.
The Type of Acyl Chains in Diacylglycerol Substrate Are Important-Enzyme activities with a racemic mixture of DOG were at all concentrations approximately half of the activity gained with a pure sn-1,2-diacylglycerol, which indicates that the sn-2,3-diacylglycerol is neither a substrate nor an inhibitor of the enzyme. Analyses with sn-1,3-DAG on thin layer chromatography showed that a substantial fraction of the 1,3 variant was isomerized into the other chain forms during the micelle preparation; this is well known to occur in aqueous systems (27). Consequently, the racemic-1,2-DAG used in most of the experiments here gave less variable assay results than the sn-1,2-diacylglycerol since it probably was in equilibrium with the sn-2,3-isomer from the start.
sn-1,2-Dipalmitoylglycerol (DPG) as substrate revealed twice as high enzyme activity as sn-1,2-dioleoylglycerol (DOG) at all concentrations up to 2.5 mol % concentration (Fig. 7). At this point, the MGlcDAG synthase activity decreased at increasing DPG concentrations. The breaking point was most likely due to segregation of DPG into separate domains or aggregates when the solubility limit of DPG in 20 mM CHAPS and 9.3 mM DOPG was exceeded. Such a breaking point also existed for DOG, but occurred first at 10 mol % (data not shown). Thus, for the activation of the enzyme by DAG, the upper accessible (experimental) concentration of the substrate has a limit that is far below the concentrations that should represent K m or give V max according to the calculated curves in Fig. 6B. The activity at the limiting concentration may be regarded as the absolute V max in this particular experimental system, and the highest specific activity recorded for the MGl-cDAG synthase in this system was 2 mmol/mg/h. Such a rate is, as estimated from the total yield of purified enzyme activity (Table I), in many-fold excess of what should be required for supplying fast-growing cells with enough MGlcDAG for the glucolipid pathway (given from Nyström et al. (21)).
These results are in agreement with results from the crude enzyme vesicular system where DPG was a better substrate at low concentration and DOG was the better one at high concentrations (19). Hence, the acyl chain preference for the DAG species is valid also for the purified MGlcDAG synthase.

DISCUSSION
In order to maintain a functional membrane, A. laidlawii adapts the lipid composition to shifting environmental and nutritional conditions. A constant surface charge density, similar phase equilibria, and a constant spontaneous curvature are thereby kept by the concerted actions of the lipid-synthesizing enzymes. However, the properties of the lipid bilayer that are sensed by such enzymes is less well known (e.g. Ref. 23). In A. laidlawii membranes, the major nonlamellar lipid MGlcDAG is synthesized by transfer of a glucose moiety from UDP-glucose to DAG, catalyzed by the enzyme 1,2-diacylglycerol 3-glucosyltransferase. MGlcDAG is also the first glucolipid in the glucolipid pathway. The variations in activity of this enzyme upon different properties of the amphiphilic environment is, therefore, of particular interest and has been studied earlier in crude cell preparations (2,18,19).
Enzyme Purification-The purification method used here consists of three column chromatography steps in a CHAPS micellar environment. Basically, it relies on the first cationic exchange chromatography, leading to a 190-fold increase (relative to solubilized cells) in the specific activity ( Fig. 1A and Table I). The identification of the enzyme in SDS-polyacrylamide gels as a minor 40-kDa protein was possible only after a several thousand-fold enrichment in activity and is mainly based on the correlation between protein gel-band intensity and enzyme activity in different fractions and by a single protein in the final high activity fractions. This is supported by a resistance against digestion by proteinase K for both the enzyme activity and this protein band. Resistance is dependent on the presence of anionic lipids like DOPG and PS, as is the enzyme activity, 3 corroborating the identification.
The CHAPS Micellar System-It was revealed that the activity of the purified MGlcDAG synthase required a certain fraction of DOPG in the CHAPS micelles (Fig. 4) as it does in the crude preparations (18). There seems to be a correlation between the packing properties of the lipids and their solubility in CHAPS. The hydrophobic, nonbilayer substrate lipid DAG, known to increase packing constraints and acyl chain order in lipid bilayers, had a poor solubility in CHAPS (Figs. 3 and 7) (see also Maréchal et al. (28)). However, if DAG were mixed with large mole fractions of lipids with larger polar headgroups, it could more easily be solubilized by CHAPS (Fig. 3). This was evident for all of the lipids used as activator or matrix lipids in this study. The slight differences between them in solubilization of DAG at low CHAPS concentrations (Fig. 3)  not affect the DAG solubility at assay conditions, as revealed by the similarity between the different curves in Fig. 5.
Lipid Dependence of the Enzyme-Many membrane-associated enzymes are dependent upon activator lipids, often anionic ones, for example the protein kinase C (29 -32) and the phosphocholine cytidylyltransferase (33,34). The MGlcDAG synthase seems to have a dependence on a critical fraction of anionic lipids (18,19). The lipid best suited to fulfill this requirement was DOPG, the major endogenous anionic phospholipid in membranes of A. laidlawii (18).
None of the uncharged (DGlcDAG), neutral-charged (DOPC), or positive-charged (sphingosine) lipids could replace DOPG as activator for the purified MGlcDAG synthase (Fig. 5). On the other hand, the anionic phospholipid PS could fully substitute for DOPG; DOPG mixtures with small fractions of PS were even better. The surface (or ) potential of the lipid aggregates must be efficiently quenched by the Mg 2ϩ present, cf.
McLaughlin (35). Together this indicates that it is the anioniccharged groups of the activator lipids that hold the activating property, as proposed earlier (18,19).
These results essentially confirm the earlier findings from both crude vesicular and mixed micellar preparations and strengthens the hypothesis that the MGlcDAG synthase demands a critical fraction of anionic lipid charges in its amphiphilic environment in order to be active. The sigmoidal appearance of the activation curves was also evident, indicating that the enzyme is activated by the same mechanism in both crude and pure systems.
Interpretation of the Kinetics and Comparison with Similar Systems-The variation in activity of the MGlcDAG synthase with altering bulk DAG concentration at fixed CHAPS and DOPG concentrations followed a sigmoidal path (Fig. 6B), indicating an activating ability of DAG. This is according to "the dual phospholipid model" of Hendrickson and Dennis (36), a strong indication that the reaction involves more than one type of interaction with DAG where one is a binding step to the micellar surface. However, the MGlcDAG synthase is behaving as a protein anchored to the amphiphilic matrix by hydrophobic interactions (3), 4 but DAG may, as well, promote interactions (i) between (an external) catalytic domain and the lipid surface, or (ii) between enzyme monomer subunits. Both alternatives are also compatible with a cooperative mechanism. The Hill equation could be more accurately fitted to the data than could the equation suggested by Hendrickson and Dennis (36), but since the data only represented a very small range of the calculated curves, none of the equations can be ruled out and no parameters could be calculated with certainty. DAG might also function as a noncooperative activator, as it does for the E. coli diacylglycerol kinase in the absence of activating phospholipids (37). Although the enzyme product MGlcDAG is a potent nonbilayer-prone lipid, the activity data from the various lipid supplements and the DPG and DOG substrates argues against any (major) influence of the spontaneous curvature on activity (or activation) of the MGlcDAG synthase at physiological DOPG concentrations. This is in accordance with crude-enzyme results (18 -20).
The dependence of the MGlcDAG synthase on its soluble substrate UDP-glucose was more easily interpreted since the data could be fitted to the Michaelis-Menten equation, and the concentration of UDP-Glc could be varied to give data points in all essential parts of the curve (Fig. 6A). A comparison of the apparent values of V max and K m at different concentrations of the activator DOPG (Table II) revealed that the K m was not altered. V max , on the other hand, was changed in accordance with the activation by DOPG, as shown in Fig. 5. The enzyme is thus following Michaelis-Menten kinetics with UDP-Glc, and the concentration of DOPG does not affect the interaction between the enzyme and UDP-Glc, but the turnover of the enzyme is substantially affected. The apparent V max was increased almost 6-fold by altering the DOPG fraction from 20 to 31 mol %, i.e. at physiological concentrations. An 11-mol % variation in DOPG at these conditions will only have minor effects on the charge density and the resulting surface () potential, as deduced from several other studies (7,35,38,39). This strongly indicates that the activation is not linearly dependent on the lipid surface charge concentration.
Importance for the Regulation Model of A. laidlawii-This study supports the regulation model for A. laidlawii lipid synthesis (3,19) where the MGlcDAG synthesis is enhanced by the negative surface charge brought by DOPG and two (usually minor) phosphoglucolipids. Since MGlcDAG is the first glucolipid along the glucolipid metabolic pathway, the regulation will act to maintain a balance to the phospholipid (i.e. PG) pathway, competing for the same PA precursor. The model is in accordance with the ability of the purified MGlcDAG synthase to be activated by a critical fraction of anionic lipids in the micellar system (Fig. 5). Interestingly, a dependence on anionic lipids for the in vivo synthesis of the major nonlamellar-prone lipid in E. coli has recently been suggested (40).
The higher efficiency of DPG than of DOG as a substrate is also in accordance with both the regulation model and the crude vesicular system (19). The amounts of MGlcDAG increases strongly with short acyl chains in vivo (14,41), and this must be governed by a substrate preference of the enzyme for such chains since the MGlcDAG synthase is only weakly regulated by properties affecting chain order and curvature (see above and Ref. 19).
From these observations on the purified MGlcDAG synthase, it can be concluded that the regulatory properties of the enzyme step are held by one single protein and are probably not part of a complex multiprotein regulatory mechanism. The interaction of this enzyme with its amphiphilic environment can now be investigated.