Purification of Phosphatidylethanolamine N-Methyltransferase from Rat Liver*

Phosphatidylethanolamine (PE) N-methyltransferase catalyzes the synthesis of phosphatidylcholine by the stepwise transfer of methyl groups from S-adenosylmethionine to the amino head group of PE. PE Nmethyltransferase was solubilized from a microsomal membrane fraction of rat liver using the nonionic detergent Triton X-100 and purified to apparent homogeneity. Specific activities of PE N-methyltransferase with  PE, phosphatidyl-N-monomethylethanolamine (PMME), and phosphatidyl-N,N-dimethylethanolamine (PDME) as substrates were 0.63, 8.59, and 3.75 pmol/min/mg protein, respectively. The purified enzyme was composed of a single subunit with a molecular mass of 18.3 kDa as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Methylation activities dependent on the presence of PE, PMME, and PDME and the 18.3-kDa protein co-eluted when purified PE N-methyltransferase was subjected to gel filtration on Sephacryl S-300 in the presence of 0.1% Triton X-100. All three methylation activities eluted with a Stokes radius 2.1 A greater than that determined for pure Triton micelles (molecular mass difference of 27.4 kDa). Two-dimensional analysis of PE Nmethyltransferase employing nonequilibrium pH gradient gel electrophoresis and sodium dodecyl sulfatepolyacrylamide gel electrophoresis indicates that the enzyme is composed of a single isoform. Analysis of enzyme activity using PE, PMME, and PDME at various Triton X-100 concentrations indicated the enzyme follows the “surface dilution’’ model proposed for other enzymes that act at the surface of mixed micelle substrates. Initial velocity data for all three lipid substrates (at fixed concentrations of Triton X-100) were highly cooperative in nature. Hill numbers for PMME and PDME ranged from 3 at 0.5 mM Triton to 6 at 2.0 mM Triton. All three methylation activities had a pH optimum of 10. These results provide evidence that a single membrane-bound enzyme catalyzes all three methylation steps for the conversion of PE to phosphatidylcholine.

Successive transfer of methyl groups from AdoMet to PE results in the synthesis of the mono-and dimethylated intermediates, PMME and PDME, and finally PC. PE N-methyltransferase (EC 2.1.1.17) accounts for 20-30% of PC synthesis in hepatocytes (3) and is localized almost exclusively in a hepatic microsomal fraction (4). P E N-methyltransferase is an integral membrane protein catalyzing a multistep reaction and as such there are inherent problems in assaying and purifying the enzyme. These problems are exemplified by the many disagreements over the number of enzymes involved in the PE methylation pathway. It has been reported that rat liver microsomes contain a methyltransferase that methylates PE to PMME and a second enzyme that converts PMME to PC (5). Similar conclusions based on kinetic data, pH, and M$+ dependence have been made for P E N-methyltransferase activity in erythrocyte membranes (6), rat brain synaptosomes (7), and bovine adrenal medulla (8). The theory that two enzymes convert P E to PC, drawn from kinetic data that did not take into account the steady state nature of intermediates in the methylation pathway, may be erroneous (9). In contrast, it has been reported that all three methylation activities in rat liver microsomes (9) have p H optima between 10 and 10.5 and similar K,,, values for AdoMet. Data on partially pure PE Nmethyltransferase also suggested the methylation of P E was catalyzed by a single enzyme (10,11).
A lack of consensus on the nature of the enzyme(s) for methylation of PE can be reconciled on the basis of tissue differences and interpretation of kinetic data. True characterization, however, awaited purification of the enzyme. Previous attempts to purify Triton X-100-solubilized P E Nmethyltransferase from mouse (12) and rat (10) liver were partially successful but met with a problem common to membrane enzyme purification: enzyme instability in detergents. P E N-methyltransferase has been purified from rat liver to a final specific activity of 0.27 prnollminlmg protein (13) and reported to consist of a 25-kDa monomer and 50-kDa dimer. The catalytic subunit of CAMP-dependent protein kinase (14) and protein kinase C (15) phosphorylate the 50-kDa protein with a resultant increase in P E N-methyltransferase activity. The 24-kDa protein was photolabeled with 8-azido AdoMet (11). We reported (in abstract form) the purification of P E N-methyltransferase to a similar specific activity reported by Pajares et al. (13) and the presence in that preparation of a 50-kDa protein (16). However, we were concerned that the final specific activity was too low for a homogenous prepara-tion of P E N-methyltransferase. The relative abundance of the 50-kDa protein in crude soluble microsomes also suggested that this was not the methyltransferase.
In this paper we report the copurification of PE-, PMME-, and PDME-dependent methylation activities from Triton X-100-solubilized microsomes to final specific activities of 0.63, 8.59, and 3.75 Fmol/min/mg protein, respectively. Purified PE N-methyltransferase was composed of a single protein of 18.3 kDa. P E N-methyltransferase activities were determined with each of the pure lipid substrates in a Triton X-100 mixed micelle assay. Kinetic analysis with regard to lipid substrates indicated an adherence to the "surface dilution" model proposed for enzymes that act on mixed micelle substrates (28,29).

EXPERIMENTAL PROCEDURES
Materials PMME and PDME were from Avanti Polar Lipids. AdoMet was purchased from Boehringer Mannheim, Canada. [meth~l-~HIAdoMet was from Amersham Corp. Triton X-100, DTT, and bovine serum albumin were from Sigma. Molecular mass standards for gel filtration and SDS-gel electrophoresis, octyl Sepharose, Sephacryl S-300, and PBE 94 polybuffer exchanger for chromatofocusing were purchased from Pharmacia LKB Biotechnology Inc. Preparative (2.0 mm) and analytical (0.2 mm) Silica Gel 60 thin layer plates were from Merck. All other materials were of reagent grade.

Purification of PE N-Methyltransferase
Isolation of Microsome-Microsomes were isolated from the livers of female Wistar rats (175-225 g) in the following manner. Rats were put to death by cervical dislocation, and the livers were immediately removed and placed in ice-cold 10 mM Tris-HC1 (pH 7.2) buffer containing 150 mM NaCI, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1 mM DTT. The liver was cut into small pieces, suspended at a final concentration of 25% (w/v) in the Tris-saline buffer, and homogenized using a motor driven Potter-Elvehjem apparatus. The homogenate was centrifuged at 12,000 X g for 10 min. The supernatant fraction was then subjected to centrifugation at 120,000 X g for 1 h. The cytosol was immediately decanted and the microsomal pellet resuspended in a 20 mM potassium phosphate buffer (pH 7.9) containing 1 mM EDTA, 1 mM DTT, and 250 mM sucrose using a hand-held glass Dounce homogenizer. Microsomes were also isolated from livers that had been perfused with 150 mM NaCl and 0.5 mM EGTA. Following perfusion the livers were treated in an identical manner as that described above. Microsomes prepared from perfused livers had a higher initial PE N-methyltransferase specific activity.
Preparation of Microsomal Membranes-Microsomes (20-30 mg/ ml protein) were suspended in 100 mM Na2C03, 5 mM DTT at a final protein concentration of 4 mg/ml and stirred at 4 "C for 30 min. The suspension was centrifuged at 120,000 X g for 1 h, and the membrane pellet was collected and resuspended in 20 mM potassium phosphate buffer (pH 7.9) containing 10% (v/v) glycerol and 5 mM DTT. This buffer was used in all subsequent purification steps and will be referred to as buffer A.
Solubilization of Microsomal Membranes-Microsomal membranes were suspended in buffer A containing 0.7% (w/v) Triton X-100 to a final protein concentration of 4 mg/ml and stirred for 1 h. The mixture was centrifuged at 120,000 X g for 1 h, and the supernatant was collected and used as a source of soluble enzyme.
Chromatography on Whatman DE52 Cellulose-Soluble PE Nmethyltransferase was passed through a column of DE52 cellulose (30 X 2.5 cm) previously equilibrated in buffer A plus 0.7% Triton X-100 at a flow rate of 1.0 ml/min. PE N-methyltransferase activity was recovered in the unbound fractions.
Chromatography on Whatman P-11 Phosphcellulose-A column of Whatman P-11 phosphocellulose (16 X 1.6 cm) was equilibrated in buffer A containing 0.7% (w/v) Triton X-100. The pooled fractions from the previous step were applied to the column at a flow rate of 0.25 ml/min. When loading was complete the column was flushed in succession with 100 ml of buffer A containing 0.7% (w/v) Triton X-100 and 100 ml containing 0.25% (w/v) Triton X-100. PE N-methyltransferase activity was eluted from the column with a linear gradient of NaCl from 0 to 0.8 M in 0.25% (w/v) Triton X-100. PE N-methyltransferase activity eluted in a broad peak from 0.2 to 0.6 M. We found it necessary to use P-11 phosphocellulose that had been used for several purification attempts (and not regenerated according to the manufacturer's specifications) in order to achieve good recoveries from this as well as from the following step. The phosphocellulose could also be pretreated with 0.2% (w/v) bovine serum albumin in buffer A containing 0.7% Triton X-100 followed by elution with 2.0 M NaCl in the same buffer. This procedure seemed to block high affinity sites on the resin that would otherwise reduce recovery of PE N-methyltransferase and change its chromatographic characteristics on octyl Sepharose.
Chromatography on Octyl Sephurose CL-4B-Active fractions from the phosphocellulose column were pooled and diluted with buffer A (no Triton X-100) to a final Triton X-100 concentration of 0.05% (w/v). The enzyme solution was pumped (0.5 ml/min) onto a column of octyl Sepharose (17 X 1.6 cm) equilibrated in buffer A containing 0.05% (w/v) Triton X-100. The column was flushed with 100 ml of buffer A containing 0.05% Triton X-100. It is critical to the final purity of the enzyme that this column be eluted so that the main PE N-methyltransferase peak is well separated from the major protein peak that precedes it (Fig. 1). The best results were obtained by eluting the column with 100 ml of buffer A containing 0.15% (w/v) Triton X-100 followed by 250 ml of a linear gradient of Triton X-100 from 0.15 to 0.5% (w/v) in the same buffer.
Chromatography on PBE 94-The final step in the purification takes advantage of the very basic nature of PE N-methyltransferase. The pooled fractions from the previous step, adjusted to pH 9.4 with 250 mM ethanolamine, were applied to a column of PBE 94 (15 X 1.6 cm) equilibrated in 25 mM ethanolamine (pH 9.4), 5 mM DTT, 10% (v/v) glycerol, and 0.1% (w/v) Triton X-100. PE N-methyltransferase was loaded at a flow rate of 0.2 ml/min. Similar to the DE52 cellulose step, PE N-methyltransferase activity was not bound to this anion exchange resin. Concentration of the dilute purified enzyme can be achieved by reapplying the enzyme solution to a 3-ml column of hydroxylapatite and eluting the activity with 0.5 M K,HPO, (pH 7.9) in 0.1% Triton X-100.

PE N-Methyltransferase Assay
The presence of microsomal phospholipids in partially purified fractions (steps 1-4) necessitated the use of higher concentrations of Triton X-100 and exogenous phospholipid substrates to achieve maximal expression of PE N-methyltransferase activity. The complete removal of endogenous phospholipids after phosphocellulose chromatography (Table I) allowed the use of lower Triton X-100 and exogenous lipid substrate concentrations. All assays were in 125 mM Tris-HC1 (pH 9.2) and 5 mM DT" with a final assay volume of 150 pl. Samples from steps 1-4 contained 1.0 mM Triton X-100 and either no addition, 2.0 mM PE, 0.4 mM PMME, or 0.4 mM PDME. No more than 25 pg of protein was assayed in these first 4 steps. Purification steps 5-7 were assayed in the presence of 0.5 mM Triton X-100 and 2.0 mM PE, 0.25 mM PMME, or 0.45 mM PDME. Dilution of the enzyme source to 0.5 mM Triton X-100 was the major factor in deciding the volume of enzyme to be assayed in the final three purification steps. PE, PMME, and PDME were added to the assay as vesicles prepared in the following manner. Lipids were dried under a stream of nitrogen and further dried under high vacuum for 30 min. The dry lipids were resuspended in 20 mM Tris-HC1 (pH 9.2), 0.01% (w/v) EDTA, and 0.02% (w/v) Triton X-100 by vortexing for 1 min and immediately sonicated at 37 "C for 3 min. Following the addition of these assay components the mixture was placed on ice and the enzyme source was added and allowed to equilibrate for 10 min.
[meth~l-~HIAdoMet (21 pCi/pmol) was added to a final concentration of 200 p~, and the mixture was incubated at 37 "C for 10 min to assay PMME-and PDME-dependent activity or 30 min to assay PEdependent activity unless otherwise specified. PE, PMME, and PDME methylation was linear for 40 min.
The assay was stopped by the addition of 2 ml of chloroform:methanol (2:1, v/v), and methylated phospholipids were extracted as previously described (9). Total ch1oroform:methanol (2:1, v/v) soluble counts were determined directly and expressed as mass of CH3 groups transferred to phospholipid/min/mg protein or products of the assay were separated by thin layer chromatography in a solvent system of ch1oroform:methanol:acetic acid:water (50305:2, v/v/v/v). Carrier PE, PC, PMME, and PDME were added to the extracts prior to separation. Phospholipids were visualized by exposure to I, vapors, scraped into scintillation vials containing 250 pl of water, 5 ml of ACS scintillation fluid was added, and radioactivity was measured after 24 h. SDS and Two-dimensional Gel Electrophoresis-Purified PE Nmethyltransferase and partially purified fractions were precipitated with 10% trichloroacetic acid in 0.15% Triton X-100. Precipitates were set on ice for 30 min and subsequently centrifuged at 10,000 rpm for 5 min in a bench-top Eppendorf microcentrifuge. Precipitates were washed twice with 1 volume of acetone at -20 "C, air dried and dissolved in a 2% SDS, 5% 2-mercaptoethanol, 20% glycerol, and 0.25 M Tris-HC1 buffer (pH 6.8). Samples that did not require concentration were diluted 1:1 with this SDS buffer. Electrophoresis was done in 10% acrylamide gels containing 0.1% SDS essentially as described by Laemmli (17).
Purification of Microsomal PE-Microsomes (400 mg of microsomal protein, 25 mg/ml) were mixed with 8 volumes of chloroform:methanol (l:l, v/v) and stirred at 20 "C for 30 min. The mixture was filtered through a plug of glass wool, total lipids were extracted by the method of Folch et al. (20), and the solvent flash evaporated. The lipids were dissolved in a small volume of chloroform and applied to preparative thin layer plates (Silica Gel 60, 2.0 mm). Plates were developed and the PE zone identified and eluted according to the method of Arvidson (21). Purified microsomal PE was stored in ch1oroform:methanol (2:1, v/v) under nitrogen at -20 "C.
Other Methods-Protein was determined by the method of Lowry et al. (22) modified to contain deoxycholate at a final concentration of 0.04% (w/v) or by a sensitive silver binding assay (23). Both protein assays used bovine serum albumin as a standard. The very dilute protein concentrations in the final three steps of the purification scheme required the use of a sensitive protein assay to avoid using most of the pure sample to obtain accurate protein determinations, In regions of the purification scheme where protein concentrations could be determined accurately by both methods, the silver binding assay gave values that were 10-20% lower than Lowry determinations.
Lipid phosphorous was determined by the method of Rouser et al. (24). AdoMet and [meth~l-~HIAdoMet were repurified on Dowex 1-X8 and Cellex-P, respectively, as previously described (25,26). AdoMet concentrations were determined by absorbance measurements at 257 nm (molar absorption coefficient = 15,000 M" cm"). Table I shows a typical purification of PE N-methyltransferase from a crude microsomal fraction of rat liver. Activities for methylation of PE, PMME, and PDME copurify but do not show the same degree of purification. The activities for the methylation of PMME and PDME purify to the same degree to the phospho-cellulose step, after which the PMME-dependent activity shows a substantially higher fold purification. Microsomal PE-dependent methylation activity was found to purify 429fold. These results can be rationalized by considering that the individual lipid substrates are also providing the enzyme with an unique phospholipid environment in each case. Differences in phospholipid head group and acyl composition would affect the properties of the mixed micelle substrate thereby influencing PE N-methyltransferase activity. Unlike previous purification schemes that relied on endogenous PE in order to assay PE N-methyltransferase (ll), it was found that PE Nmethyltransferase activity was completely dependent on exogenous substrate following the phosphocellulose step. This is consistent with the observation that all measurable lipid phosphorous was removed following this step (Table I). were determined by relating the absorbance of unknown samples to those of standards at 257 nm. Total activity, recovery, and -fold purification were calculated for PMME-dependent methylation.

PE N-Methyltransferase Purification-
The hydrophobic character of PE N-methyltransferase is illustrated by the high Triton X-100 concentration necessary to release it from microsomal membranes and by its high affinity for octyl Sepharose CL-4B (Fig. 1). This step offered the most substantial purification (100-fold) and was critical with regard to final enzyme purity. As illustrated in Fig. 1, PE N-methyltransferase elutes in a broad peak following single step elution with 0.15% Triton X-100. The gradient was necessary to keep the volume of the pooled fractions to a minimum.
The distribution of methylated products at each step of the purification and for all 3 lipid substrates is shown in Table  11. As previously reported (9), PMME and PDME methylation resulted in the formation of predominately PDME and PC, respectively. Methylation of these two substrates using the Triton X-100 mixed micelle assay showed the same trend with 95-99% of the radioactivity found in these two products. This trend was observed throughout the entire purification. PC (the major product of PE methylation) showed a 10-15% increase in content with a comcomitant decrease in PMME and PDME, as the enzyme was purified.
Molecular Mass of PE N-Methyltransferme-Analysis of the purified enzyme by SDS-PAGE in 10% acrylamide gels indicated that PE N-methyltransferase was composed of a single 18.3 2 0.7 (n = 3)-kDa subunit (Fig. 2, lane 1). It is apparent that during the purification of PE N-methyltrans-

TABLE I1
Distribution of methylated phsphlipids Distribution of the individual phospholipids is expressed as the percentage of the total counts in PC, PMME, and PDME. The recovery of applied radioactivity was 60-70% with >95% of the recovered counts in the three methylated products. Results are expressed as the average f S.D. for three determinations unless indicated bv numbers in Darentheses. ferase from microsomes there is an increase in the amount of an 18.3-kDa prutein. The 50-kDa protein previously thought to be PE N-methyltransferase (11, 13-15) steadily decreased in content (lanes 7-2) and was absent from the pure enzyme.

Product distribution
Molecular mass analysis of the native enzyme in Triton X-100 by gel filtration on Sephacryl S-300 showed that PE-, PMME-, and PDME-dependent activities co-chromatographed with a Stokes radius of 55.2 A (n = 2, Fig. 3A). Pure Triton micelles were found to have a Stokes radius of 53.1 A when chromatographed on the same column. The apparent molecular mass difference between PE N-methyltransferase and pure micelles was determined to be 24.7 kDa. These results indicated that there was a single subunit/Triton micelle. Indeed, attempts to cross-link the enzyme subunits with dimethylsuberimidate were negative. Analysis of the elution profile of the 18.3-kDa protein on Sephacryl S-300 is shown in Fig. 3B. SDS-PAGE and silver staining of concentrated column fractions revealed that the putative methyltransferase protein (inset) co-chromatographed with all three methylation activities. This is strong evidence to suggest that the 18.3-kDa protein is indeed PE N-methyltransferase and that a single enzyme performs all three methylation steps.
During purification it became apparent that PE N-methyltransferase possessed an extremely basic PI. This observation was corroborated by the finding that PE N-methyltransferase (solubilized in urea and Triton X-100) would not enter conventional isoelectric focusing gels (pH 3-10). A two-dimensional electrophoresis system employing NEPHGE (the method of choice for resolving basic proteins) in the first dimension and SDS-PAGE in the second revealed that PE N-methyltransferase is composed of a single isoprotein (Fig.   4).
Kinetic Analysis of PE N-Methyltransferme-We have developed a simple Triton X-100 mixed micelle assay to individually determine PE-, PMME-, and PDME-dependent PE Nmethyltransferase activity. The purified enzyme was found to FIG. 4. Two-dimensional gel electrophoresis of P E N-methyltransferase. Two wg of purified PE N-methyltransferase was resolved in a NEPHGE system containing pH 3-10 ampholytes followed by SDS-PAGE in the second dimension. The gel was fixed in 30% (v/v) methanol, 10% (w/v) trichloroacetic acid, 3.5% (w/v) sulfosalicyclic acid for 2 h prior to silver staining. be maximally active when assayed in the presence of 0.5 mM Triton X-100 and 2.0 mM (88 mol %) PE, 0.25 mM (49 mol %) PMME, or 0.4 mM (61 mol %) PDME. The mol % of Triton X-100 at which maximum PMME and PDME methylation occurred is in the region where a homogenous population of micelles exists (27). The 88 mol % of PE required for maximal activity is well above 68 mol % (Triton X-100 mol fraction of 0.32) sphingomyelin, which is the effective limit for a monodisperse population of mixed micelles (27). Also, freeze fracture analysis of 91 mol % PE revealed the presence of large multilamellar vesicles (43).
The role of Triton X-100 in this mixed micelle assay was investigated to determine if PE N-methyltransferase had kinetic properties similar to those described for other membrane-bound enzymes. Fig. 5 illustrates that all three methylation activities are subject to "surface dilution" inhibition, a result often seen for enzymes that act on mixed micelle substrates (28, 29). All three methylation activities showed a definite peak of activity between 0.5 and 2.5 mM Triton X-100. However, PE and PDME methylation activities were not inhibited as much as PMME methylation by low surfactant concentrations. The inhibition of PE, PMME, and PDME methylation was not the result of increasing micelle concen- tration (inset). Optimal enzyme activity was essentially linear over a range of Triton X-100 concentrations and at fixed PE, PMME, and PDME of 88,49, and 61 mol %, respectively.
Double reciprocal plots of initial velocity uersus lipid substrate concentration (at fixed Triton X-100) were found to be nonlinear and highly cooperative. Plots of PE-dependent activity a t 0.5 mM Triton X-100 (Fig. 6) gave a Hill number of 3.7. Similar behavior was noted for PMME and PDME. In the case of PMME, there was a marked increase in the Hill number with higher concentrations of detergent (Fig. 7). Hill numbers for 0.5, 1.0, and 2.0 mM Triton X-100 were 3.1, 4.7, and 6.4, respectively. Similarly, Hill numbers of 2.5, 2.8, 6.1, and 12.8 were determined for PDME methylation at 0.5, 1.0, 2.0 and 3.0 mM Triton, respectively (Fig. 8). This type of cooperative kinetic behavior has been described by Deems et al. (31) using phospholipase A, and has been attributed to changes in the size and shape of the mixed micelle as the concentration of phospholipid in the mixed micelle increases.
As illustrated in Fig. 9, all three methylation activities of the purified enzyme have a pH optimum at 10. The alkaline pH optima of the three methylation activities probably reflects the pK, values of the substrate's amino group. It is feasible that protonation of the substrate's amino group results in poor binding to the active site of PE N-methyltransferase. PMME-and PDME-dependent activities were found to have K,,, values for AdoMet of 40.8 and 23.7 p~, respectively. The K, for AdoMet for the complete methylation of PE to PC was 36.7 pmol. These pH optima and K , values are similar to that reported for microsomes (9).
Methylation of PE, PMME, and PDME was found to be linear for up to 30 min. As previously reported (9), the major  product of PDME methylation is PC. When PMME methylation was examined the major product formed was PDME (Fig. 10); however, with increasing time there was a proportional increase in counts in PC and a concomitant decrease in counts in PDME. Total methylation remained constant for 30 min (inset). This result would indicate that as the proportion of PDME in the mixed micelle increases it can compete effectively for methylation with PMME. The linearity of the methylation reactions is a good indication that the rate of exchange of phospholipid substrate between micelles is not rate-limiting.

DISCUSSION
PE methylation is considered to be a minor route for the synthesis of PC in liver but may be more important in situations where PC synthesis via the CDP-choline pathway is compromised (32, 33). Alternatively, it has been demonstrated that factors that stimulate the CDP-choline pathway inhibit P E methylation (34-36). As a prelude to understand better the factors that regulate hepatic PC synthesis, we have purified PE N-methyltransferase to apparent homogeneity from rat liver microsomes. The 7-step purification scheme we employed illustrates some properties of this enzyme. Contrary to previous reports from this laboratory (lo), we were successful in using the nonionic detergent Triton X-100 as a solubilizing agent after substituting 20 mM potassium phosphate with 10% glycerol for the standard Tris-HC1 buffers used in other purification attempts. Purified PE N-methyltransferase showed no perceivable loss in activity for at least 2 months when stored in buffer A plus 0.1% Triton X-100 at 4 "C. This is in sharp contrast to the 80% loss of activity reported for the partially pure enzyme after 16 h at 2 "C (10). A second property that became apparent during our purification efforts was that PE N-methyltransferase is a very basic protein. P E N-methyltransferase passed unretained through an anion exchange resin (PBE 94, step 7) a t p H 9.4 and can only be resolved into a single isoprotein when electrophoresed toward the anode in a NEPHGE system. What relation this alkaline PI has to enzyme function is yet unknown.
The subunit molecular mass (18.3 kDa) of PE N-methyltransferase is unusually small considering the fairly complex series of methylation reactions it catalyzes. Analysis of the pure enzyme in Triton X-100 micelles by gel filtration indicated that a single subunit is present per micelle. No information is yet available concerning the subunit structure in phospholipid membranes. Low molecular masses have been reported for several other phospholipid biosynthetic enzymes: the 34-kDa phosphatidylinositol synthase (37) and 23-kDa phosphatidylserine synthase (29) purified from Saccharomyces cerevisiue and the 13.2-kDa diacylglycerol kinase (38) and 27-kDa CDP-diacylglycerol synthetase (30) from Escherichia coli. P E N-methyltransferase, purified to homogeneity from Zymomonm mobilis, has a reported molecular mass of 42 kDa (39).
Two reports by Pajares et al. (11,13) have claimed purification of P E N-methyltransferase from rat liver. These researchers have co-purified P E N-methyltransferase activity with a 50-kDa protein. The specific activity of this preparation (assayed in the presence of a mixture of PE, PMME, and PDME) was 0.27 Fmol/min/mg protein (13). Based on the data we have presented, two lines of argument would indicate that the 50-kDa protein bears no relation to PE N-methyltransferase. First, we have achieved final specific activities that are 2.3-, 32-, and 14-fold higher than that reported by Pajares et al. (13), using PE, PMME, and PDME as sub- Time course for activity of PE N-methyltransferase with PMME as substrate. Sixty ng of purified PE N-methyltransferase was assayed in the presence of 0.5 mM Triton X-100 and 0.25 mM PMME for the indicated periods of time. PDME (+) and PC (El) were separated by thin layer chromatography, counted, and expressed as total disintegrations/min recovered. 65-70% of the applied counts were recovered for all time points. The formation of total methylated phospholipid was linear for 30 min (inset).
strates. Although these authors assay P E N-methyltransferase activity at 20 "C, pH 8.35, and at 100 wmol of AdoMet, all of which are suboptimal assay conditions in our hands, direct comparison would indicate a substantially greater purification in our case. Second, examination of Fig. 2 would indicate that the 50-kDa protein is the major protein in partially purified fractions (steps 2-5) but is less abundant in step 6 relative to the 18.3-kDa protein and completely absent in the pure fraction. These two points taken together indicate that the 50-kDa protein is only a persistent contaminant and that a microsomal protein of 18.3 kDa is PE N-methyltransferase.
A common problem encountered in the assay of PE Nmethyltransferase has been the efficient delivery of lipid substrates to the enzyme. Previously, exogenous lipid substrates have been added to assay mixtures as vesicle preparations with no added detergent. Under these conditions enzymatic activity would be limited by the rate of exchange of vesicle phospholipids into microsomes. P E N-methyltransferase assayed in the presence of Triton X-100 has been reported to be activated by PMME and PDME (12). Tanaka et al. (12) also reported PE-dependent activity in Triton X-100-solubilized microsomes. A simple Triton X-100 mixed micelle assay has been developed that can be used to assay all three methylation activities. The addition of increasing concentrations of surfactant inhibited P E N-methyltransferase methylation of PE, PMME, and PDME by dilution of phospholipid on the micelle surface. This inhibition was found to be independent of the concentration of mixed micelles when a fixed mol % of phospholipid substrate was used. Also consistent with the model proposed for enzymes acting on micellar substrates (31), reciprocal plots of initial velocity at fixed Triton X-100 were found to be highly nonlinear. This behavior can be explained by considering that the surface curvature (and thus the surface area/phospholipid and Triton X-100 molecule) decreases since Triton X-100 mixed micelles become more ellipsoidal as the mol % of phospholipid increases (31). It is more likely, however, that PE, PMME, and PDME are activators as well as substrates for P E N-methyltransferase. P E N-methyltransferase may require a structural feature of the phospholipid-Triton X-100 micellar substrate for full reconstitution of activity since, unlike phospholipase A, for which the kinetic scheme was formulated (31), PE N-methyltransferase is an integral membrane protein with hydrophobic domains contiguous with phospholipid or detergent. Also, because of its integral nature, P E N-methyltransferase has no true dissociation constant for binding to micelles. Kinetic analyses of E. coli diacylglycerol kinase, a hydrophobic 13.2-kDa enzyme, have shown that 1,2-diacylglycerol-dependent activity is nonlinear and cooperative (40). Addition of nonsubstrate lipid activates the kinase and linearizes the diacylglycerol-dependent activity. An analogous situation may occur for P E N-methyltransferase if nonsubstrate lipids were added to the mixed micelles.
Optimal P E methylation, unlike PMME or PDME, occurs at a mol % where nonmicellar structures no doubt exist. It is conceivable that pure PE, rather than PE in a micellar form, is more readily methylated by P E N-methyltransferase. Pure P E in solution is proposed to exist in a hexagonal I1 array (41). This is probably not the case in our assay where the pH is >9.0 and 12 mol % of Triton X-100 is present. Both pH >9.0 (42) and a Triton X-100/PE ratio of 0.1 or 9 mol % Triton X-100 (43) favor the formation of bilayer structure. Robinson and Waite (44) have reported a lysosomal phospholipase A2 that preferentially hydrolyzes P E in what appears to be a hexagonal I1 phase. PC, phosphatidylinositol, and phosphatidylglycerol underwent maximal hydrolysis when in mixed micelles. The propensity of PE N-methyltransferase for P E bilayers uersus micellar structures is quite intriguing and clearly requires more study in a reconstituted system free of detergent.
There is now strong evidence that PE methylation is catalyzed by a single enzyme. It is possible, however, that there are multiple active sites for the lipid substrates. We feel this is unlikely for several reasons. The major product of PMME methylation is PDME (9), which suggests that excess PMME is competing with and releasing PDME from the enzyme. Further evidence for competition between PMME and PDME can be seen in Fig. 10. As the concentration of newly formed PDME approaches 0.02 mM in the assay it competes efficiently with PMME for methylation. These results would seem to indicate that there is competition between PMME and PDME for methylation by P E N-methyltransferase. However, other explanations for Fig. 10 such as micelle modification and specific time-dependent inactivation of an active site cannot as yet be eliminated.
Analysis of the products formed by microsomal PE N-methyltransferase, using PMME as a substrate, indicated almost complete inhibition of PC formation (>95%) from endogenous P E (Table 11). It seems likely that PMME is competing with P E for an active site on the enzyme. It is interesting to note that like methylation of microsomal PE, methylation of pure microsomal P E by purified P E N-methyltransferase results in the formation of PC (92%). Clearly P E methylation is the rate-limiting step in the reaction sequence and PE does not compete with and release the two partially methylated intermediates.
Controversy concerning the molecular nature of the methylation system has also extended to possible modes of regulation. P E N-methyltransferase is regulated by the cellular levels of AdoMet and S-adenosyl-L-homocysteine (34, 451, fatty acids (36), and PE/PC ratios (46). There is a growing body of evidence that suggests P E N-methyltransferase is regulated by phosphorylation. Glucagon and cAMP analogues caused a 2-fold activation of PE N-methyltransferase in hepatocytes in suspension and in primary culture (47,48). Oddly, glucagon and cAMP analogues were found to inhibit PE to PC conversion in cultured hepatocytes (48,49). Recent reports (14,15) have suggested that purified PE N-methyltransferase is phosphorylated and activated by CAMP-dependent protein kinase and protein kinase C. The 50-kDa protein, claimed to be P E N-methyltransferase, is phosphorylated on serine in both cases. In light of the findings presented in this paper that indicate that the 50-kDa protein is a persistent contaminant in partially pure P E N-methyltransferase, the role of phosphorylation in P E N-methyltransferase regulation should be re-evaluated. Preliminary evidence' has shown that the 18.3-kDa protein is phosphorylated by the catalytic subunit of CAMP-dependent protein kinase. No information is yet available on the effects on enzyme activity. With the availability of pure PE N-methyltransferase we should presently be able to assess accurately what effect phosphorylation and other putative regulatory mechanisms have on P E Nmethyltransferase i n vivo and i n vitro.