The specific acylation of glycerol 3-phosphate to monoacylglycerol 3-phosphate in Escherichia coli. Evidence for a single enzyme conferring this specificity.

Abstract Pronounced positional specificity during the acylation of glycerol 3-phosphate to form monoacylglycerol 3-phosphate has been shown with a particulate enzyme preparation from Escherichia coli. Palmitic acid is found to be esterified exclusively to position 1 while unsaturated fatty acids are predominantly esterified to position 2. Evidence for a single enzyme being involved in this specific acylation is presented. This evidence is based on studies of single site mutants of E. coli possessing a glycerol 3-phosphate acyltransferase of greatly increased thermolability and on chemical modification of the enzyme. Additional experiments show that the acylation of 1-acylglycerol-3-P involves an enzyme activity or activities separate from that which acylates glycerol-3-P.

to form monoacylglycerol d-phosphate has been shown with a particulate enzyme preparation from Escherichiu coZi. Pahnitic acid is found to be esterified exclusively to position 1 while unsaturated fatty acids are predominantly esterified to position 2. Evidence for a single enzyme being involved in this specific acylation is presented. This evidence is based on studies of single site mutants of E.
coli possessing a glycerol 3-phosphate acyltransferase of greatly increased thermolability and on chemical modification of the enzyme.
Additional experiments show that the acylation of 1-acylglycerol-3-P involves an enzyme activity or activities separate from that which acylates glycerol-3-P.

Phosphoglycerides
play an important role in the structure and function of all biological membranes (l-5).
In most naturally occurring phosphoglycerides unsaturated fatty acids are preferentially esterified at position 2 and saturated fatty acids at position 1 of the glycerol molecule (2). This asymmetric fatty acid distribution is thought to be of major importance in the functional and structural role of phosphoglycerides in membrane processes. Perturbations in the asymmetric distribution of the fatty acids of membrane phosphoglycerides can produce drastic disturbances in cellular physiology (6-13).
Thus the enzymatic mechanisms conferring such positional specificity of fatty acid residues are of central importance to the cell. A likely origin of this asymmetric distribution of fatty acids would be the esterification of glycerol a-phosphate to form phosphatidic acid, a key intermediate in the biosynthesis de 7u)uo of phospho-* This investigation was supported in part by Grant GB-5142X from the National Science Foundation and Grant l-ROl-HE-19496 from the National Institutes of Health.

glycerides.
Early studies on the positional substrate specificity of glycerol a-phosphate acyltransferase in liver microsomal systems (14-16) indicated that glycerol a-phosphate was acylated in a nearly random manner, although marked positional specificity was exhibited in the esterification of l-and 2-acylglycerol a-phosphates (17) or at least in the esterification of l-acylglyc-erol3-P (18). However, several subsequent investigations with the enzyme from rat liver have shown that glycerol-3-P acylation proceeds in a nonrandom, asymmetric manner, so that unsaturated fatty acids are preferentially esteritied at position 2 of the phosphatidic acid (19)(20)(21). Some of these studies have reported monoacylglycerol-3-P as an intermediate in this reaction (22-25).
A recent report of van den Bosch and Vagelos (25) has shown that acylation of glycerol-3-P by a particulate enzyme system in Escheriehia coli proceeds with similar preferential positioning of saturated and unsaturated fatty acids.
In the present study, we have used our collection of E. coli strains possessing a mutation in glycerol-3-P acyltransferase (26) to produce evidence that a single enzyme catalyzes the positionally specific acylation of glycerol-3-P by either saturated or unsaturated fatty acyl coenzyme A substrates to form monoacylglycerol-3-P.
We also present evidence that this enzyme is distinct from that activity which acylates I-acylglycerol-3-P to form phosphatidic acid.

MATERIALS AND METHODS
Muter&--Glycerol a-phosphate labeled with either aH or 14C was synthesized by the reaction of uniformly labeled glycerol with glycerokinase followed by purification as described by Chang and Kennedy (27). The resulting products were radiochemically pure in three different paper chromatographic systems and were assayed by the reaction with glycerol a-phosphate dehydrogenase (28). 3H-Oleyl-CoA was prepared from 9, lo-aH-oleic acid by the method of Goldman and Vagelos (29) followed by purification on DEAE-cellulose with a linear gradient from 0 to 2 M LiCl (30). Oleyl-CoA and palmityl-CoA were the products of P-L Biochemicals and Sigma, respectively. The other acyl-CoA's were synthesized as described by Al-Arif and Blecher (31) except that the N-hydroxysuccinimide esters of the unsaturated fatty acids were crystallized from methanol-water mixtures.

I-Acylglycerol
Bovine serum albumin (fatty acid free), DTNB,l and glycerol 3-phosphate were the products of Pentex Biochemicals, Kankakee, Illinois, Sigma, and Calbiochem, respectively. Unsaturated fatty acids were from the Hormel Institute, Austin, Minnesota. All radioactive compounds were purchased from New England Nuclear.
Growth of Bacteria and Enzyme Preparations-Mutant CV15 and its parent strain 8 have been described in a previous publication (26). Mutants CV2 and CV31 were derived from strain 8 by the radioactive suicide selection procedure similar to that described for CV15. These mutants grow normally at 25" but immediately cease growth and rapidly lose greater than 90% of their phospholipid biosynthetic capacity in viva at 37". These mutants possess glycerol-3-P acyltransferase activities much more thermolabile than that of their parent, strain 8. Genetic criteria indicate that this thermolability is caused by a single lesion probably in the structural gene for the acyltransferase.
The medium and conditions of growth for the bacterial strains were described previously as was the preparation of the particulate enzyme fraction (26).
Enzyme Assays-The spectrophotometric assay procedure for glycerol-3-P acyltransferase activity was described previously (26). The reaction mixture consisted of Tris-HCl, pH 8.5, 0.1 M; MgC12, 5 ITIM; glycerol-3-P, 1.5 mu; DTNB, 1 mu; acyl-CoA, 30 pM; bovine serum albumin, fatty acid free, 1 mg per ml; and particulate enzyme protein, 0 to 1 mg per ml. The acylation of I-acylglycerol-3-P was assayed at 25" by a similar spectrophotometric procedure. A typical incubation mixture consisted of Tris-HCl buffer, pH 9.0, 0.1 M; MgCL, 0.5 mM; bovine serum albumin, fatty acid free, 1 mg per ml; DTNB, 1 IIIM; acyl-CoA, 15 PM; I-acylglycerol-3-P, 50 PM; and particulate enzyme protein, 0 to 1 mg per ml. DTNB was added as a 0.01 M solution in 0.1 M potassium phosphate buffer, pH 7.0. The release of CoA was measured by the reaction of the thiol with DTNB to give an increase of absorbance at 413 nm.
Protein concentrations were determined by a microbiuret procedure (34).
Chemical Inactivation of A yltransferase Activities-The particulate enzyme preparation from strain 8 was treated with maleic anhydride essentially according to the method described by Sia and Horecker (35). Various amounts of an acetone solution of maleic anhydride, e.g. 0.075 to 0.5 mg of maleic anhydride per mg of particulate enzyme protein, were added to the enzyme preparations in a total reaction mixture volume of 1.5 ml. The pH of the mixture was kept at about 7.0 with the addition of 1 M NaOH.
After 10 min the particles were centrifuged and resuspended in 0.02 M potassium phosphate buffer, pH 7.0, and were assayed for the glycerol-3-P and l-acylglycerol-3-P acyltransferase activities.
Analysis of Reaction Products-All lipid reaction products were extracted from the reaction mixtures by a modification of the method of Bligh and Dyer (36). Boric acid (0.1 M) solution was substituted for water and the chloroform was saturated with boric acid solution before use. The lipids were separated into 'The abbreviation used is : DTNB, 5,5'-dithiobis (2-nitrobenzoic acid). phosphatidic acid, monoacylglycerol-3-P, and monoglyceride on silica gel thin layer plates impregnated with Na&Oz according to the method of Hajra and Agranoff (37) with ohloroformmethanol-acetone-acetic acid-water (200 : 80 : 30 : 40 : 20 v/v) as the developing solvent.
The identity of the various lipids was ascertained by cochromatography with authentic standards. The lipids were located by iodine vapor and the appropriate areas of silica gel were scraped from the plate. The silica gel was then eluted with 10 ml of chloroform-methanol (1:4 v/v) for further analysis, and aliquots were counted in Bray's solution (38) after evaporation to dryness under a stream of nitrogen. Alternatively, the radioactivity was measured by suspending the scrapings from the silica gel plate into toluene scintillation solution containing 3.5% thixotrophic gel (Cab-0-Sil, Packard Instruments, Downers Grove, Illinois). All counting was done in a Packard 3380 scintillation counter. Analysis of Monoaylglycerol-S-P-Monoacylglycerol-3-P was at least 80 to 85% degraded to monoglyceride by phosphatidic acid phosphatase prepared from rat liver and utilized as described by Wilgram and Kennedy (39), except that the reaction mixture contained additional 0.05 M borate buffer to prevent migration of acyl groups. The monoglycerides thus formed were fractionated into the I-and a-isomers on boric acid-impregnated silica gel thin layer plates with Solvent Systems A and F of Thomas, Scharoun, and Ralston (40) and quantitatively determined by counting the appropriate areas of silica gel in Bray's solution.
A number of saturated and unsaturated monoglycerides were tested in these systems to ascertain that the degree of saturation of the acyl portion of the monoglyceride has no effect on the separation.
The monoglycerides formed were further characterized by elution from silica gel and treatment (41) with porcine pancreatic lipase from Worthington.
Free fatty acids and monoglycerides from the lipase digestions were separated on silica gel plates according to the method of Freeman and West (42) and assayed for radioactivity.

RESULTS
Specijkity in A ylattin of Glycerol-3-P to Monoacylglycerol-3-P - Fig.  1 shows the time course of the acylation of glycerol-3-P with either palmityl-CoA or oleyl-CoA as the acyl donor. The total of the three products of the acylation reaction, monoglyceride, monoacylglycerol-3-P, and phosphatidic acid, is stoichiometric with the release of free CoA as measured in the spectrophotometric assay. Previous reports from this laboratory (22, 25) have shown that the relative amounts of the three products formed in vitro is dependent on the experimental conditions employed.
The pH of the incubation mixture is especially important since the acylation reaction which produces monoacylglycerol-3-P, the reaction which acylates l-acylglycerol-3-P, and the phosphatase which dephosphorylates monoacylglycerol-3-P to monoglyceride, all have distinctly different pH optima.
The acylation reaction which forms monoacylglycerol-3-P is linear for at least 10 min as was the release of free CoA as determined spectrophotometrically.
However the formation of monoglyceride and phosphatidic acid cease after about 4 to 6 min of reaction.
Thus with either of these substrates, we can obtain conditions under which monoacylglycerol-3-P is the major product of the reactions. Preliminary results of van den Bosch and Vagelos (25) have shown that the acylation of glycerol-3-P with palmityl-CoA results in a monoacyl product in which position 1 of the glycerol was exclusively acylated.
In contrast, when oleyl-CoA was used as the acyl donor in an identical system, acylation of position 2 predominated.
These data suggested that the acylating system possessed the ability to discriminate between saturated and unsaturated fatty acids and also between the two isomeric hydroxyl groups of the glycerol moiety available for acylation. We have extended these observations with conditions designed to maximize the accumulation of monoacylglycerol3-P and to miniiize the migration of the acyl moieties between the two hydroxyl groups of the monoacyl lipids during extraction and chromatography.
Thus we have been able to make a more rigorous examination of this specificity. The species of monoacylglycerol-3-P formed by incubation of radioactive glycerol- The standard assay mixture for glycerol-3-P acylation wasused except that uniformly labeled W-glycerol-3-P (final specific activity 1 mCi per mmole) was added and the various acyl-CoA's were added at the indicated concentrations, which were found to give maximal activity in the spectrophotometric assay. After incubation for 10 min at 25" the reaction mixtures were extracted and the monoacylglycerol-3-P fractions were isolated, dephosphorylated with phosphatidic acid phosphatase, and the resulting isomeric monoglycerides separated by borate-impregnated silica gel thin layer chromatography (See "Materials and Methods").
-i- monoglyceride fractions Glycerol-3-P acylation reaction mixtures were incubated with an enzyme preparation from strain 8 as in Table I with Wlabeled glycerol-3-P and either palmityl-CoA or oleyl-CoA as substrates.
The monoglycerides were obt,ained by dephosphorylation of the monoacylglycerol-3-P fraction resulting from the incubation with phosphatidic acid phosphatase and chromatography on a borate-impregnated silica gel thin layer plate with solvent System A of Thomas et aZ. (40). The fractions designated 1-monoglyceride and 2-monoglyceride were identified by cochromatography with authentic standards. The 1-monoglyceride fraction resulting from the palmityl-CoA incubation and the 2-monoglyceride fraction resulting from the oleyl-CoA incubation were then eluted from the silica gel with chloroform.
The chloroform was evaporated under nitrogen and the monoglyceride fractions were incubated with pancreatic lipase. After incubation the mixture was fractionated into chloroform-and water-soluble fraction by the method of Bligh and Dyer (36). The degree of hydrolysis was ascertained by counting both the chloroform-and the water-soluble products (see "Materials and Methods" for further details). 3-P with various acyl-CoA's were treated with phosphatidic acid phosphatase and the resulting monoglycerides were separated into their l-and a-isomers by thin layer chromatography.
As seen in Table I all unsaturated fatty acids tested are predominantly esterified to position 2 while palmitic acid is almost completely ester&xl to position 1. Myristic and stearic acids, the other saturated acyl-CoA's tested, are found to be distributed between both positions with position 2 predominating.
The identity of the isomeric monoglycerides isolated and identified by thin layer chromatography was confirmed by treatment with pancreatic lipase. Pancreatic lipase is known to have an absolute specificity for position 1 since it hydrolyzes l-monoglyceride completely but has no effect on 2-monoglyceride (43). As shown in Table II 2. Thermolability of the glycerol-3-P acyltransferase activity from Mutants CV15 and CV31 with palmityl-CoA or oleyl-CoA as acyl donors. Enzyme preparations from the mutants were incubated for various time periods at 37", chilled to O', and assayed spectrophotometrically for 5 min at 25" with either palmityl-CoA (30 PM) or oleyl-CoA (30 PM). The data are expressed relative to an unheated sample of the same extract (= 100%). Typical specific activities (nanomoles per min per mg of protein) of unheated enzyme preparations of CV15 and CV31 were 2.6 and 2.2, respectively, when palmityl-CoA was used as acyl donor. The correspondmg activities of CV15 and CV31 were 1.3 and 1.5, respectively, when oleyl-CoA was used as acyl donor. Symbols: CV15, palmityl-CoA drolyzed by pancreatic lipase. Thus it was apparent that this particulate enzyme system was able to acylate glycerol-3-P to monoacylglycerol-3-P in a highly specific asymmetric manner.
We next attempted to understand the enzymatic basis of this specificity, i.e. does this specific acylation require more than one enzyme?
Evidence for Single Enzyme Involved in SpeciIc Acylation of Glycerol-S-P-We have previously described the isolation of temperature-sensitive strains of E. coli possessing a mutant glycerol-3-P acyltransferase of greatly increased thermolability (26). Several other mutants were subsequently isolated.
The first mutant described was called CV15; two other mutants, CV2 and CV31, have also been studied in detail.
These mutants were all independently isolated and their reversion frequencies to temperature resistance are consistent (44) with their phenotype being the result of single mutations in the structural gene for the acyltransferase.
In addition the transduction frequency to temperature resistance for CV15 obtained with Phage PlKc grown on strain 8 is consistent with the phenotype of CV15 being caused by a single mutation (44).
Figs. 2 and 3 show the kinetics of the thermal inactivation of glycerol-3-P acyltransferase activities of particulate preparations derived from these mutants and their parent, strain 8, with either saturated or unsaturated acyl-CoA's as acyl donors. It is apparent in Fig. 2, where glycerol-3-P acyltransferase activities of Mutants CV15 and CV31 were studied, that the rate of thermal inactivation of the enzyme is similar in these two mutants (half-life of 17 min for CV15,18+ min for CV31).
More important, however, is the finding that the thermolability of the activity with either palmityl-CoA or oleyl-CoA as acyl donor is identical in enzyme preparations from a given mutant. The maleylated preparations were then assayed spectrophotometrically for both glycerol-3-P acyltransferase and 1-acylglycerol-3-P acyltransferase activities with either palmityl-CoA or oleyl-CoA. The data are expressed relative to an untreated sample of the preparations (= 100%). Typical specific activities of 1-acylglycerol-3-P acyltransferase of the unheated sample of strain 8 are 4.2 and 6.7 with palmityl-CoA and oleyl-CoA, respectively, as acyl donor. l and n denote glycerol-3-P acyltransferase activity; 0 and [7 denote l-acylglycerol-3-P acyltransferase activity. Activities with either palmityl-CoA (O and 0) or oleyl-CoA (m and 0) are shown.
strain 8, with several additional acyl-CoA's. Fig. 3 indicates that the enzyme preparation of CVZ (half-life 9 min) is much more thermolabile than either the parent, strain 8, or Mutants CVl5 and CV31 (see Fig. 2). Again it is noted that the rate of enzyme inactivation is the same in either of these two strains whether the saturated acyl donors, stearyl-CoA and palmityl-CoA, or unsaturated acyl donors, oleyl-CoA, cis-vaccenyl-CoA and palmitoleyl-CoA, are used as substrates.
In addition the experiments of Fig. 3 show that the rate of thermal inactivation is independent of chain length since the rates are the same with acyl groups of 18 (Fig. 3A) and 16 (Fig. 3B) carbon atoms. It is known that under the conditions of these acyltransferase reactions all the unsaturated fatty acids tested are predominantly transferred to position 2 of glycerol-3-P whereas palmitate is transferred almost exclusively to position 1 (Table I). Thus the finding that enzyme activity is heat inactivated at the same rate when tested with all these acyl donors suggests that a single enzyme catalyzes acyl transfer to position 2 or to position 1 of glycerol-3-P.
These results, coupled with the genetic evidence, FIG. 5. Comparative thermolabilities of the glycerol-3-P acyltransferase and the 1-acylglycerol-3-P acyltransferase activities of Mutant CV15. A particulate enzyme preparation from Mutant CV15 was incubated for various time intervals at 37" before assay as described in Fig. 2. The preparation was then assayed for both glycerol-3-P acyltransferase (O-O) and l-acylglycerol-3-P acyltransferase (O-O) activities with palmityl-CoA as the substrate.
Typical specific activity of 1-acylglycerol-3-P acyltransferase of the unheated enzyme preparation of CV15 is 10.4 with palmityl-CoA as acyl donor.
indicate that the product of a single gene is involved in the specific acylation of glycerol-3-P to monoacylglycerol-3-P. This product may be either a single enzyme or a single protein or lipid essential to the activity of two or more acylating enzymes.
Further support for the proposal that a single enzyme catalyzes the acylation of glycerol-3-P was obtained by studying the chemical inactivation of an enzyme preparation derived from strain 8. Fig. 4 indicates the effect of treatment of the particles with increasing concentrations of maleic anhydride for 10 min. It is apparent that glycerol-3-P acyltransferase activity is severely inhibited by this treatment and that the decline of activity with increased concentration of inhibitor is the same whether oleyl-CoA or palmityl-CoA is used as acyl donor in the assay. Also shown in Fig. 4 is the effect of maleic anhydride treatment on 1-acylglycerol-3-P acyltransferase activity. It is noted that this activity, tested with either oleyl-CoA or palmityl-CoA is inhibited to a different extent than the glycerol-3-P acyltransferase.
Evidence that A ylattin of Glycerol-3-P or of l-Monoacylglycerol-S-P Involves Separate Enzymatic Activities-The maleylation experiment shown in Fig. 4 indicates that the enzyme or enzymes involved in the acylation of 1-acylglycerol-3-P is diierent from that which acylates glycerol-3-P. This is consistent with the differing pH and magnesium ion optima of the two activities (25). More rigorous evidence is presented in the experiments shown in Figs. 5 and 6. Fig. 5 shows that the particulate fraction derived from Mutant CV15 possesses a very thermolabile glycerol-3-P acyltransferase activity while the 1-acylglycerol-3-P acyltransferase activity is not significantly affected by heating at 37". This is also true of glycerol-3-P acyltransferase Mutants CV2 and CV31 as shown in Fig. 6. The I-acylglycerol&P acyltransferase activity of these mutants has the same thermolability as that of their parent, strain 8. These experiments show that 1-acylglycerol-3-P aoyltransferase activity is no more thermolabile than the wild type enzyme in B  I  1  I  I  1  I.1   *  O  IO  20  30  MlNUTES  AT 37" FIG. 6. Comparative thermolabilities of the 1-acylglycerol-3-P acyltransferase activity of strain 8 and Mutants CV2 and CV31 as assayed with either palmityl-CoA (0-C) or oleyl-CoA CR-q ) as substrate. Typical specific activities of l-acylglycerol-3-P acyltransferase of unheated samples of CV2 and CV31 are 5.9 and 4.9, respectively, with palmityl-CoA; 6.1 and 4.4, respectively, with oleyl-CoA (see Fig. 2 and "Materials and Methods" for experimental details) .   TABLE   III   Synthesis of phosphatidic acid containing both saturated and unsaturated fatty acids A mixture of "C-palmityl-CoA, 20 PM (1.12 mCi per mmole), and 8H-oleyl-CoA, 30 PM (5.7 mCi per mmole), were incubated in the standard glycerol-3-P acyltransferase assay mixture with unlabeled glycerol-3-P.
After incubation for 10 min the resulting lipids were chromatographed as described under "Materials and Methods." The monoacylglycerol-3-P and phosphatidic acid fractions were eluted from the silica gel and counted. Thus these experiments show that diierent enzymes catalyze acyl transfer to glycerol-3-P or to l-acylglycerol-3-P.
It should be noted that the rate of heat inactivation of 1-acylglycerol-3-P acyltransferase is the same with either palmityl-CoA or oleyl-CoA as a substrate. It was of interest to determine whether the acyltransferase preparation in vitro could produce phosphatidic acid with a fatty acid distribution consistent with that found in the cellular phospholipids.
A mixture of 8H-oleyl-CoA and 14C-palmityl-CoA was incubated with glycerol-3-P and the distribution of the fatty acids was determined in the isolated monoacylglycerol-3-P and phosphatidic acid fractions by scintillation counting.
As seen in Table III, phosphatidic acid containing both fatty acids, monopalmitylglycerol-3-P and monooleylglycerol-3-P are all formed.
Thus the system in vitro is able to produce phosphatidic acid of the required specificity.

DISCUSSION
The specificity of the system in vitro used in the study of the acylation of glycerol-3-P in this paper is remarkably consistent with the distribution of the fatty acid moieties of the phospholipids found in the bacterial cell (13, 45,46). Palmitic acid is the predominant saturated fatty acid found in E. coli and it is found almost exclusively in the 1-acyl position of the cellular phospholipids.
Myristic acid, the other major saturated fatty acid of E. coli, is found to be distributed about equally between positions 1 and 2 although position 2 predominates when an unsaturated fatty acid auxotroph is deprived of fatty acid (13).
Stearic acid is only a trace component of E. coli phospholipids and the data in the literature suggest that it is randomly distributed between the two positions although the amounts have not been accurately quantitated.
Palmitoleic and cis-vaccenic acids comprise the normal unsaturated fatty acid complement of E. coli and chiefly occupy position 2 of the cellular phospholipids Oleic acid is not a normal component of E. coli (47) but can be efficiently incorporated when an unsaturated fatty acid auxotroph is supplemented with oleic acid (48). Under these conditions oleic acid is found exclusively in position 2 (13).
The results presented strongly indicate that a single gene product is involved in the specific acylation of glycerol-3-P to monoacylglycerol-3-P.
These reactions have been studied extensively in mammalian systems with little information as to the detailed enzymatic mechanisms involved in the specific acylation reactions.
The present studies show that the acylation reactions might be rather simple in regard to the number of enzymes required.
We have shown that the specific acylation of glycerol-3-P either requires only a single enzyme or is catalyzed by a number of enzymes sharing a common essential component. Further proof that a single enzyme catalyzes acyl transfer to positions 1 or 2 of glycerol-3-P must await protein solubilization and purification.
However, if one protein catalyzes these reactions, then that protein must possess remarkable recognition abilities since it must catalyze palmityl transfer to position 1 or unsaturated acyl group transfer to position 2.
The complexity of the curves for the inactivation of l-acylglycerol-3-P acyltransferase by maleylation is consistent with either of two possibilities.
The curves can be explained if there are two enzymes which catalyze this reaction or if the completely maleylated enzyme retains about one-half the enzyme activity found in the untreated enzyme. If the former possi-