INTRODUCTION

Polysaccharides located on the bacterial surface or embedded in the outer envelope, including peptidoglycan, teichoic acid, lipoteichoic acids, teichuronic acids, cell wall polysaccharides, lipopolysaccharides, and capsular polysaccharides, play an important role in the integrity of the bacterial cell. The biosynthesis of these polysaccharides requires a special group of enzymes: the glycosyltransferases, which are usually located in the cytoplasmic membrane and are capable of adding specific sugar(s) onto a polysaccharide chain. The activity of such membrane-associated enzymes was initially reported many years ago (1, 2), but their purification has not been easy. Because of the importance of bacterial surface polysaccharides in virulence and immunity (3, 4, 5, 6, 7, 8, 9), much research has been directed in recent years to identification of the genes and gene products responsible for the biosynthesis of these structures. Several relevant genes have been characterized in both Gram-negative and Gram-positive bacteria (10, 11, 12, 13, 14, 15, 16). Many of the glycosyltransferase genes have been characterized through assays demonstrating the ability of the cloned gene to complement defective production of polysaccharides. However, the gene product itself is very hard to recover and analyze for enzyme activity. This difficulty may be attributable to the secondary modifications of gene products and the complexity of polysaccharide biosynthesis. In this study, through a delicate enzymological approach, we purified to near homogeneity the membrane-associated enzyme glucosyltransferase from the Gram-positive bacterium Micrococcus luteus.

The teichuronic acid of the cell wall of M. luteus is a polysaccharide consisting of D-glucose and N-acetyl-D-mannosaminuronic acid (ManNAcU)¹ residues in which the repeating unit is [4)b-D-ManNAcUp-(16)a-D-Glcp-(1]_n (17, 18). In vitro synthesis of teichuronic acid by cytoplasmic membrane fragments requires UDP-GlcNAc, UDP-ManNAcU, and UDP-D-glucose (19, 20). During biosynthesis, UDP-GlcNAc is the donor of the initial residue, or so-called linker, that becomes the potential reducing terminal residue of native teichuronic acid, which is linked to peptidoglycan through a phosphodiester bond (21, 22). However, cell wall-membrane preparations require only UDP-glucose and UDP-ManNAcU to effect elongation of teichuronic acid already present in wall fragments (Fig. 1) (23). If exogenous soluble teichuronic acid is supplied as acceptor, UDP-glucose and UDP-ManNAcU are codependently required by detergent-solubilized extracts of cytoplasmic membrane fragments to effect its significant elongation (24). Transfer of glucosyl residues from UDP-glucose to teichuronic acid is catalyzed by glucosyltransferase, and the addition of N-acetylmannosaminuronic acid residues is catalyzed by the ManNAcU-transferase. The detergent Thesit (dodecyl alcohol polyoxyethylene ether) effects solubilization of the cytoplasmic membrane fragments, after which the glucosyltransferase is partially purified (25). Herein, we report the use of a combination of Thesit and another detergent, CHAPS, for the solubilization and subsequent purification of the M. luteus glucosyltransferase, which has been found to consist of four copies each of two types of subunit.

EXPERIMENTAL PROCEDURES

Cultures of M. luteus (ATCC 4698) were grown to mid-logarithmic phase at 30 °C with vigorous shaking in a liquid medium containing peptone (10 g/liter) and NaCl (5 g/liter). The cells were harvested by centrifugation at 3,000 × g at 4 °C for 10 min. The pellet of wet packed cells was washed in TMM buffer (50 mM Tris-HCl buffer, pH 8.8, 0.1 mM magnesium acetate, and 2 mM 2-mercaptoethanol); sedimented again; and then suspended at a concentration of 1 g/ml in 10-fold diluted TMM buffer to which lysozyme (1 mg/g of cells), DNase (3 mg/g), and RNase (30 mg/g) had been added. After incubation at 25 °C for 120 min to effect lysis, the suspension was homogenized and fractionated three times by centrifugation at 4 °C for 10 min. The particulate enzyme fraction was defined as the fraction sedimenting at 3,000-12,000 × g and containing cytoplasmic membrane fragments. This fraction was resuspended in TMM buffer to a final protein concentration of 5-15 mg/ml and stored at 15 °C.

The cytoplasmic membrane fragments were solubilized by thorough mixing of 2 volumes of buffer D, which consisted of 40 mM Tris-HCl, pH 8.2, 1.6 M glycerol, 0.5 mM EDTA, 0.5 mM magnesium acetate, 1 mM 2-mercaptoethanol, 0.3 mM dithiothreitol, 0.5 mg/ml CHAPS, and 1.0 mg/ml Thesit. The insoluble residue was removed by centrifugation at 20,000 × g for 15 min and discarded. The detergent-solubilized enzyme extract was stored at either 4 or 15 °C.

The detergent-solubilized enzyme samples were applied to 10-ml adsorbent columns packed with Bio-Beads SM-2 (Bio-Rad) and were eluted at room temperature with buffer D, pH 7.3, at a flow rate of 0.3-0.5 ml/min. The nonadsorbed material that passed directly through the column was collected and assayed. Fractions containing glucosyltransferase activity were pooled.

Pooled fractions containing enzyme activity from the adsorbent column were applied to a column of DEAE-cellulose (4.5 × 20 cm) and eluted first with 3 bed volumes of buffer D, pH 7.3, and then with a 1000-ml linear gradient of 0.13-0.55 M NaCl in the same buffer at a flow rate of 1 ml/min. Fractions (4.3 ± 0.2 ml) were collected and assayed for enzyme activity and protein concentration. Dilute fractions were concentrated by loading of a 2-ml volume of sample on a Centricon 10 microconcentrator or larger volumes on a Centriflo CF25 membrane cone (Amicon, Inc., Beverly, Massachusetts) and centrifugation at 3,000 × g for 1-2 h at 4 °C. The material retained by the 10- and 25-kDa cutoff filters was collected.

Concentrated active fractions from the DEAE-cellulose column were applied to a Bio-Gel P-300 gel-filtration column (1.5 × 50 cm). Prestained protein standards (1.4 mg; Bio-Rad) were included with the enzyme sample. The column was eluted with buffer D, pH 7.3, containing 0.1 M NaCl at a flow rate of 0.2 ml/min.

Partially purified protein samples were analyzed by electrophoresis on 3-12% (18 × 12 × 0.15 cm) or 4-15% (9 × 7 × 0.1 cm) nondenaturing gradient polyacrylamide slab gels in the absence of SDS. The former gels were purchased from Integrated Separation Systems (Natick, MA), and the latter gels were from Bio-Rad. The sample buffer was 75 mM Tris-HCl, pH 6.8, containing 500 mg/ml glycerol, 1 mg/ml CHAPS, 1.0 mM 2-mercaptoethanol, and 2.5 mg/ml bromphenol blue. One volume of sample buffer was mixed with 2 volumes of protein sample. Electrophoresis was performed on 3-12% gels run for 3 h at 200 V of constant voltage in a running buffer consisting of 24.8 mM Tris and 192 mM glycine, pH 8.3, containing 0.2 mg/ml CHAPS (detergent in upper electrode buffer solution only) and 0.1 mM 2-mercaptoethanol. Electrophoresis on 4-15% gels was performed for 45 min at 200 V of constant voltage in the same running buffer. Proteins were visualized by staining with Coomassie Brilliant Blue, and glucosyltransferase activity was assayed as described below.

The molecular mass of glucosyltransferase was estimated by comparison of the mobility of the enzyme's activity with that of nondenatured protein molecular mass markers on 3-12% native gradient PAGE. The relationship between molecular mass and migration of the standards was calculated by best fit logarithmic regression with the following standard proteins: jack bean urease hexamer (545 kDa), jack bean urease trimer (272 kDa), bovine serum albumin dimer (132 kDa), and bovine serum albumin monomer (66 kDa).

Native gradient PAGE was also used as the final preparative step for purification of glucosyltransferase. For this procedure, 3-12% nondenaturing gradient gels (18 × 12 × 0.15 cm) in Tris/glycine buffer (in the absence of SDS) were used, and 3% stacking gel with sample wells was freshly made before the electrophoresis. A custom-made comb was used to divide the gel into two large sample wells (6 × 2 × 0.15 cm), each holding ~1 ml or 2 mg of protein sample, and three small wells (0.7 × 2 × 0.15 cm) for the protein standards and control samples that were used to monitor the run and to localize the proteins. After electrophoresis under the standard conditions described above, the band at 4.25 ± 0.1 cm was excised with a razor blade or a gel cutter. The gel slices either were soaked in buffer D and stored at 20 °C for further study or were electroeluted in the same Tris/glycine/CHAPS running buffer for 2.5 h at 10 mA of constant current/glass tube at room temperature in a Bio-Rad electroeluter. Eight such slab gels were able to separate 26 mg of protein sample after gel-filtration column chromatography.

Samples from each step of purification were analyzed by SDS-PAGE (26). Proteins were subjected to staining with Coomassie Brilliant Blue R-250 (10 mg/ml) or to double staining with silver/Coomassie Blue (27).

To determine the molecular masses of the subunits of glucosyltransferase, the single protein band with glucosyltransferase activity that migrated onto a 7.2% or 4.25-cm nondenaturing gradient gel was excised. The excised gel slices were electroeluted as described above, mixed with SDS-PAGE sample buffer, and then subjected to SDS-PAGE. Molecular mass was calculated on the basis of a calibration curve prepared with the following standard proteins: myosin (200 kDa), Escherichia coli -galactosidase (116.25 kDa), rabbit muscle phosphorylase b (97.4 kDa), bovine serum albumin (66.2 kDa), hen egg white ovalbumin (42.7 kDa), bovine carbonic anhydrase (31 kDa), and soybean trypsin inhibitor (21.5 kDa).

Isoelectric focusing was performed at 10 °C with Ampholine polyacrylamide gel plates (pH 3.0-9.5 and 4-6.5; Pharmacia Biotech Inc.) on an FBE-3000 flatbed apparatus (Pharmacia Biotech Inc.). The anode solutions were 1 M H₃PO₄ and 0.1 M glutamic acid in 0.5 M H₃PO₄ for the pH 3.0-9.5 and 4-6.5 isoelectric focusing gels, respectively; the respective cathode solutions were 1 M NaOH and 0.1 M -alanine. Isoelectric focusing standards from Bio-Rad were used. The conditions used for focusing were 1,500-2,000 V, 50-55 mA, and 25-30 watts for ~3 h, by which time the colored standards were sharply focused. The gel pH was measured immediately after completion of the run. Proteins were stained with 1% Coomassie Brilliant Blue R-250 or double-stained with silver/Coomassie Blue, and glucosyltransferase activity was assayed as described below.

For determination of the isoelectric point of glucosyltransferase by isoelectric precipitation, aliquots of detergent-solubilized enzyme extract were suspended in a series of buffers with pH values ranging from 4.0 to 6.5, incubated for 10 min at 25 °C, and centrifuged at 14,000 × g at 4 °C for 10 min. After the sedimented materials were quickly resuspended in buffer D, pH 8.8, glucosyltransferase activity and protein concentration were determined. The precipitation of glucosyltransferase at its isoelectric point was also used for partial purification of M. luteus glucosyltransferase.

Protein composition was determined spectrophotometrically by the bicinchoninic acid method at 562 nm (28). Bovine serum albumin served as the reference standard.

Glucosyltransferase activity (24) was assessed by incubation of 50 ml of enzyme sample in a reaction mixture (final volume, 100 ml) containing 0.7 mM UDP-[¹⁴C]glucose (2 mCi/mmol), 50 mM HEPES buffer, pH 8.2, 20 mM magnesium acetate, 5 mM 2-mercaptoethanol, and 30 mg/ml teichuronic acid (the glucose acceptor). After 60 min of incubation at 25 °C, the reaction was stopped by the addition of 20-30 ml of isobutyric acid. Residual substrate was separated from the reaction product by descending paper chromatography in isobutyric acid and 1 M NH₄OH (5:3, v/v). The product remained at the origin of the chromatogram and was quantified by liquid scintillation counting. This product at the origin had previously been analyzed by mass spectrometry, NMR imaging, and carbohydrate PAGE and had been confirmed to be teichuronic acid (18, 24).

Glucosyltransferase activity on native polyacrylamide gels or on isoelectric focusing gels was determined by cutting the gel into 2-mm slices with a razor blade, soaking each gel slice in 50 ml of buffer D, and incubating these gel slices in the glucosyltransferase assay reaction mixture (final volume, 100 ml) for 120 min at 25 °C. A standard portion of the reaction mixture was removed for paper chromatography and subsequent determination of the amount of glucose added to the acceptor. The sensitivity of glucosyltransferase was tested by incubation of different concentrations of antibiotics (novobiocin, tunicamycin, bacitracin, and tetracycline) with the enzyme 1 h prior to the standard assay for glucosyltransferase activity.

RESULTS

The purification of glucosyltransferase was begun with 120 g (wet weight) of M. luteus cells harvested in the mid-logarithmic phase of growth, when the enzyme's activity was maximal. A suspension of the cells in hypotonic medium was digested with lysozyme to solubilize peptidoglycan and to lyse resulting spheroplasts. The crude cell extract was treated with DNase and RNase to reduce viscosity and thereby facilitate subsequent purification procedures. These digestion steps were normally conducted at room temperature since it had previously been observed that incubation at or above 37 °C caused the loss of glucosyltransferase activity.

Cytoplasmic membrane fragments were recovered from the crude extract by centrifugation between 3,000 and 12,000 × g. This fraction was resuspended and recovered twice more by centrifugation. Sedimentation and washing of the membrane fragments resulted in an increase in specific activity, but only with a considerable loss of total activity, much of which remained in the supernatant solutions. SDS-PAGE of the membrane fraction showed a substantial decrease in the diversity of contaminating proteins.

Since glucosyltransferase is associated with cytoplasmic membrane fragments, a significant step in its isolation and purification is its release from the membrane with a minimal loss of activity and in a form amenable to subsequent purification. Detergents are routinely used for this purpose. Although some success has accompanied the use of Thesit (24, 25), electrophoresis of samples containing substantial amounts of Thesit has been problematic. Hence, other detergents, such as CHAPS, Triton X-100, and Tween 20, were evaluated as alternatives. Of these, CHAPS was the most promising. The effect of treatment of representative crude cell fractions with a combination of Thesit and CHAPS was examined by measurement of the glucosyltransferase activity released to the supernatant fraction. Table I shows that detergent treatment approximately doubled the glucosyltransferase specific activity detected in spent culture supernatant and in a suspension of whole cells, whereas it more than tripled that released from cytoplasmic membrane fragments. Presumably, the increase in specific activity reflected the solubilization or dispersal of the cytoplasmic membrane or similar lipid-based micelles, with the concomitant solubilization or release of enzyme previously sequestered inside vesicles and thus unavailable to the assay system. In light of these observations, membrane components were solubilized by treatment with both Thesit and CHAPS. Glycerol and magnesium ion were added to stabilize the solubilized enzyme.

Table I. Detergent release of M. luteus glucosyltransferase activity Fraction, detergent treatmenta Protein conc Activityb Specific activity mg/ml units/ml units/mg Culture supernatant   No 2.13 8.0 3.76   Yes 1.90 15.1 7.95 Cell suspension   No 4.78 24.3 5.08   Yes 5.10 58.9 11.5 Membrane fragments   No 2.69 53.2 19.8   Yes 2.31 168.0 72.7 a  The fraction was suspended in buffer lacking detergents or in buffer containing Thesit (1 mg/ml) and CHAPS (0.5 mg/ml). b  One unit is equal to 1 nmol of [14C]glucose incorporated in 60 min in a standard assay.

Detergent-solubilized extract was passed through an adsorbent column (Bio-Beads SM-2) for the removal of nonpolar substances. Both protein and glucosyltransferase activity came directly through the column, with a slight enhancement of specific activity. The material that passed through the column was applied directly to a DEAE-cellulose column and then eluted with a gradient of NaCl (Fig. 2). About 25% of the protein and most of the carotenoids from the membrane were washed directly through the column. Several protein peaks were eluted with salt, but only those fractions that eluted at an NaCl concentration between 0.25 and 0.30 M contained glucosyltransferase. A faint yellow color was characteristic of the fractions containing the enzyme and could be used to predict which fractions would test positive for its activity.

Glucosyltransferase recovered from DEAE-cellulose was concentrated and applied to a Bio-Gel P-300 column, which was eluted with buffer D supplemented with 0.1 M NaCl. Glucosyltransferase eluted in the void volume (Fig. 3). The presence of salt in the developing buffer improved the recovery of enzyme activity. It was not determined whether the enzyme bound to the gel matrix or was dissociated in the absence of salt. In contrast, attempts to use gel-filtration matrices of dextran were unsuccessful: no activity was eluted from the column, nor was any matrix-bound enzyme detected. Dextran-based matrices were eliminated from further consideration after it was determined that the addition of dextran to any preparation containing glucosyltransferase caused a loss of all activity.

Glucosyltransferase recovered from the Bio-Gel P-300 column was subjected to electrophoresis on a native gradient polyacrylamide gel (Fig. 4). Enzyme activity was determined from the sample in one lane (Fig. 4, right panel), and protein composition was revealed by staining of an adjacent duplicate lane (Fig. 4, left panel, lane 2). Although several protein bands (migrating from 3 to 10.5 cm) were detected by native gradient PAGE, only the band migrating at 4.25 cm displayed glucosyltransferase activity. A duplicate gel lane containing glucosyltransferase was incubated with UDP-[¹⁴C]Glc in situ. Incubation was followed by exhaustive washing of the gel to remove unbound substrate. An autoradiogram of this gel displayed only one band, whose location corresponded to that of glucosyltransferase activity (data not shown). These results provide evidence that all other protein bands (~90% in mass) detected by native PAGE were either contaminants or the dissociated subunits of glucosyltransferase that were no longer active. Therefore, in a further study, native gradient PAGE was used as the final preparative step for purification of glucosyltransferase. The protein in the single band migrating at 4.25 cm was isolated either by excision or by electroelution and was analyzed as purified glucosyltransferase. Results of the purification steps described above are summarized in Table II. The purification resulted in a 150-fold increase in specific activity, with an overall yield of 13%. Electroelution is a powerful method for solubilizing purified protein and for providing sample for further analysis of the subunits of glucosyltransferase; however, prolonged electroelution caused a partial (30-70%) loss of the specific activity of glucosyltransferase (see "Discussion"). In contrast, the enzyme remaining in the excised gel slices from native PAGE without Coomassie Blue staining was relatively stable in buffer D at 20 °C. Those samples were later used for enzymological studies of glucosyltransferase.

Table II. Purification of the glucosyltransferase of M. luteus Purification step Total protein Total activitya Yield Specific activity Purification mg units % units/mg -fold Lysozyme digestion 3,000 19,800 100 6.6 1.0 Membrane fraction 570 10,800 55 19 2.9 Detergent extraction 162 11,200 57 69 10 Adsorbent column 160 12,800 65 80 12 DEAE-cellulose column 79 8,980 45 115 17 Gel-filtration column (P-300) 26 7,710 39 295 45 Protein band isolated at 4.25 cm on 12-cm native gradient PAGE 2.6 2,530 13 970 150 a  One unit is equal to 1 nmol of [14C]glucose incorporated in 60 min in a standard assay.

In our early attempt to purify glucosyltransferase, the active samples obtained by adsorbent column and DEAE-cellulose column chromatography were isoelectrically precipitated at pH 5.0 by centrifugation at 14,000 × g at 4 °C. The sedimented materials were resuspended in buffer D, pH 8.8, and centrifuged again to remove the insoluble precipitates. By this method, the specific activity of glucosyltransferase in the supernatant was significantly increased. The purification resulted in a >2-fold increase in specific activity (~250 units/mg). Significantly fewer protein bands were obtained by SDS-PAGE compared with the previously used purification process (DEAE-cellulose column chromatography) (Fig. 5). SDS-PAGE also yielded two major bands (54 and 52.5 kDa), with many minor protein bands in the background indicative of only partial purification (Fig. 5, lane 5). Two problems were later found with regard to this purification step. First, purification resulted in low yield (20%), with more than half of glucosyltransferase irreversibly denatured. The solubilized fraction in the supernatant contained some protein aggregates that interfered with further purification. Second, similar problems arose when isoelectric focusing PAGE was used as the purification method. Since isoelectric precipitation is not practical for the final purification of glucosyltransferase, we eliminated this step in our later purification effort.

In the native gradient PAGE shown in Fig. 4, glucosyltransferase activity was found in a band that migrated at 4.25 cm, only slightly faster than the molecular mass standard jack bean urease hexamer (545 kDa). A plot of distance migrated versus log molecular mass was constructed by computer. The curve in the plot is the best fit log regression line based on the migration of the molecular mass markers. According to the location of the band with enzyme activity, the molecular mass of glucosyltransferase was estimated to be 440 kDa.

The protein band displaying glucosyltransferase activity was recovered from a duplicate native gradient gel and subjected to SDS-PAGE. Silver staining revealed two major bands, presumably subunits of glucosyltransferase (Fig. 6). The two bands took on a purplish brown color after double staining, a result suggesting the presence of a small amount of carbohydrate or lipid. The approximate molecular masses for these bands (52.5 and 54 kDa) were determined from another plot on the basis of molecular mass standards. (Faint minor bands with mobilities corresponding to 23, 58, 70, and 75 kDa were occasionally observed; however, their quantity was insignificant.) The results of SDS-PAGE of partially purified glucosyltransferase fractions are shown in Fig. 5. Although many minor bands can be detected in the last fraction (lane 5), it is obvious that the 54- and 52.5-kDa bands have changed from minor bands to major bands as a result of the purification process. These results prove that the increase in the specific activity of glucosyltransferase during purification is associated with the increase in the intensity of the bands of 52.5 and 54 kDa (Fig. 5).

In addition, we observed that the running condition of native gradient PAGE is critical. When the partially purified glucosyltransferase was exposed to prolonged electrophoresis, a waving band, which is an aggregate of dissociated glucosyltransferase components with similar 54- and 52.5-kDa subunits, was observed below the 4.25-cm glucosyltransferase band in native PAGE. It seems that the longer the period of electrophoresis, the farther the waving band migrated.

Glucosyltransferase was stable for 3-6 months when stored at 15 °C in buffer containing 2 M glycerol, 20 mM magnesium acetate, 2 mM 2-mercaptoethanol, 1 mM dithiothreitol, and 0.1-3% Thesit and CHAPS. However, repeated freezing (to 15 °C) and thawing caused a gradual loss of activity. Activity was enhanced by 30% after 24 h of incubation at 25 °C. However, at higher temperatures, the enzyme was susceptible to inactivation. Incubation at 37 °C for 5 and 24 h resulted in the loss of 70 and 90% of the enzyme's activity, respectively. Glucosyltransferase was stable when the pH of the above solution was maintained between 5.5 and 9.0 at 4 °C, but at its isoelectric point, it underwent inactivation.

The isoelectric point of the purified glucosyltransferase, as determined by isoelectric focusing on polyacrylamide gels, was between 4.5 and 4.9. Isoelectric precipitation of the enzyme yielded a pI value of 5.0 (Fig. 7). The diminished recovery of glucosyltransferase in the middle of the isoelectric precipitation plot is an indication of rapid loss of activity at the isoelectric point. The enzymatic reaction followed simple Michaelis-Menten kinetics. The apparent K_m value of glucosyltransferase for its substrate, UDP-Glc, is ~300 µM.

Purified glucosyltransferase was assayed for its sensitivity to inhibition by several antibiotics (novobiocin, tunicamycin, bacitracin, and tetracycline) at different concentrations. Of the antibiotics tested, only novobiocin substantially inhibited the enzyme. Specifically, 10 µM inhibited ~60% of glucosyltransferase activity, while a relatively high concentration of novobiocin (100 µM) was required for nearly complete inhibition. In contrast, 100 µM tunicamycin inhibited glucosyltransferase activity by only 50%, and bacitracin and tetracycline had no inhibitory effect.

DISCUSSION

The glucosyltransferase of M. luteus is one of a pair of enzymes that act together to effect the synthesis of teichuronic acid, in which the alternating residues of the polymer are glucose and ManNAcU (Fig. 1) (23, 24, 29). Glucosyltransferase catalyzes the transfer of a glucosyl residue from UDP-glucose to an acceptor teichuronic acid that has a nonreducing terminal ManNAcU residue. The other enzyme, ManNAcU-transferase, catalyzes the addition of a ManNAcU residue from UDP-ManNAcU to an acceptor teichuronic acid that has a nonreducing terminal glucose residue (the product of the glucosyltransferase reaction). As part of our ongoing investigation of teichuronic acid synthesis (20, 24), we sought to purify these enzymes so that we could study the mechanism of bacterial polysaccharide biosynthesis and examine why D-glucosyl residues are incorporated as -glycosides, whereas D-ManNAcU residues are incorporated as the -anomer (18). We chose to purify glucosyltransferase first since UDP-[¹⁴C]glucose is more readily available than UDP-[¹⁴C]ManNAcU. Knowledge gained from the solubilization and purification of glucosyltransferase should be applicable to the purification of ManNAcU-transferase. The purified enzymes should prove useful in generating antibodies for studies of the enzymes' cellular localization.

Enzymes involved in the biosynthesis of bacterial cell wall polymers have been detected mainly in cytoplasmic membrane fragments. Indeed, all prior investigations of teichuronic acid biosynthesis had made use of either cytoplasmic membrane fragments (also termed the particulate enzyme fraction) (19, 29) or a wall-plus-membrane complex (23, 30). Thus, the first major step toward purification of glucosyltransferase was the solubilization of protein components of the membrane fragments. Several detergents were tested for their ability to solubilize membranes while maintaining the activity of the glucosyltransferase. Thesit proved to be suitable for solubilization (24, 25), but later was found to interfere with subsequent purification steps as well as with PAGE analysis. For example, Thesit-solubilized enzyme fractions contain many other membrane proteins, such as ATPase (31), and membrane lipids, in particular, the typical membrane pigments of M. luteus, carotenoids (32). Thesit did not fully solubilize the enzyme from the pieces of membrane fragment. CHAPS is a suitable detergent for solubilization, but is quite expensive. Because of the adverse effects of one and the cost of the other, the use of these two detergents in combination appeared to be superior to the use of either by itself.

The purification of glucosyltransferase was monitored routinely by nondenaturing PAGE and SDS-PAGE. Samples always displayed several protein bands, suggesting the presence of substantial amounts of contaminant protein. Furthermore, the use of numerous purification steps in various combinations consistently failed to increase the level of specific activity above 1000 units/mg. It is now apparent that glucosyltransferase is a multisubunit enzyme whose disaggregation upon prolonged electrophoresis or electroelution on polyacrylamide gel permits the detection of subunits. This finding is consistent with the observation that glucosyltransferase elutes in the exclusion volume of a Bio-Gel P-300 column (Fig. 3) and has a mobility in native gradient PAGE that is in line with a molecular mass of 440 kDa (Fig. 4). This interpretation was verified by elution of the glucosyltransferase band from native gradient PAGE and by the result of subsequent analysis by SDS-PAGE, in which the usual pattern of protein bands was generated. The bands at 52.5 and 54 kDa are always predominant and of equal intensity. The stoichiometry of the subunits has not been accurately determined; however, if the 54- and 52.5-kDa subunits are considered to be present with a stoichiometry of four copies each per aggregate, then the masses of the subunits add up to 426 kDa, a figure in reasonable agreement with the observed mass of 440 kDa for the aggregate. Glucosyltransferase strictly requires four pairs of subunits to maintain catalytic function. Since glucosyltransferase is a multisubunit protein, it is easy to understand why early attempts to purify a conventional-sized protein (200 kDa) led to a loss of enzyme activity. Furthermore, the glucosyltransferase subunits did not display the well defined mobilities and staining properties associated with many standard proteins. This difference may be a consequence of hydrophobicity or of special modifications of this membrane-derived protein or may be due to an unknown factor.

Reviewing the purification processes, we found that the most critical step in the final purification of this "fragile" membrane enzyme was the application of native gradient PAGE. Like SDS-PAGE, this procedure separates native proteins according to their molecular sizes; unlike SDS-PAGE, it causes no enzyme denaturations. The availability of a commercial gradient gel made this purification possible. The advantages offered by such gels are that they are easy to use and the reproducibility of results obtained with them (especially for migration distance) and the multiple subsequent uses of the purified protein band for enzyme assays, subunit analysis, and preparative purifications. The disadvantage of this method is the difficulty of scaling up purification.

The glucosyltransferase of M. luteus differs substantially from other glycosyltransferases that have been described. Unlike the glycosyltransferases involved in the biosynthesis of cell envelope polysaccharides of other bacteria, the glucosyltransferase of M. luteus is capable of elongating the teichuronic acid polysaccharide chain without a lipid carrier (such as glycosylpolyisoprenyl phosphate) as the glycosyl donor. The glucosyltransferases involved in the synthesis of membrane-derived oligosaccharides of E. coli (33) and the cyclic glucans of Bradyrhizobium japonicum (34) both use UDP-glucose as a glucosyl donor. However, membrane-derived oligosaccharide synthesis also requires both polyisoprenyl phosphate (33) and acyl carrier protein (35). Several other properties of the M. luteus glucosyltransferase were also characterized in this study, such as its thermostability, its isoelectric point, its K_m, and its sensitivity to antibiotics. Further studies involving molecular cloning as well as structural and functional characterization will help to elucidate the catalytic mechanism of this molecule and will eventually establish the precise relationship between the two enzymes (glucosyltransferase and ManNAcU-transferase) that are responsible for the synthesis of the entire cell wall polymer teichuronic acid.