Regulation of Diacylglycerol Kinase Biosynthesis in Escherichia coli A TRANS-ACTING dgkR MUTATION INCREASES TRANSCRIPTION OF THE STRUCTURAL GENE*

The mechanism of a trans-acting mutation, dgkR1, which causes a 7-fold elevation of diacylglycerol kinase activity in membranes (Raetz, C. R. H., Kantor, G. D., Nishijima, M., and Jones, M. L. (1981) J. Biol. Chem. 256, 2109-2112) was investigated by direct measurement of diacylglycerol kinase polypeptide by high performance liquid chromatography and by con- struction of fusions of the dgkA promoter to 8-galac-tosidase and galactokinase. The dgkR1 mutation was demonstrated to act by increasing the transcription of the structural gene for diacylglycerol kinase, dgkA. Additionally, sn-glycerol-3-phosphate acyltransferase activities were shown to be decreased 30-50% in membranes from dgkR1 mutant strains. Increased diacyl- glycerol levels occurred when cells were grown on low osmolarity media. This did not affect dgkA expression. In a dgkR+ background, enhanced expression of sn-1,2-diacylglycerol kinase activity in cells containing a high copy number plasmid bearing dgkA decreased sn-1,2-diacylglycerol levels. However, overproduction of diacylglycerol kinase in a dgkRl genetic background did not affect diacylglycerol levels, suggesting that the dgkRl mutation affects diacylglycerol

1,2-diacylglycerol kinase activity in cells containing a high copy number plasmid bearing dgkA decreased sn-1,2-diacylglycerol levels. However, overproduction of diacylglycerol kinase in a dgkRl genetic background did not affect diacylglycerol levels, suggesting that the dgkRl mutation affects diacylglycerol metabolism by mechanisms additional to enhancement of dgkA transcription.
In Escherichia coli, diacylglycerol kinase functions to recycle sn-l,2-diacylglycerol generated from the turnover of membrane phospholipids back into the main pathway of phospholipid biosynthesis (1). During growth on media of low osmolarity, the major source of diacylglycerol is the transfer of phospholipid head groups to periplasmic membrane derived oligosaccharides (2)(3)(4)(5). Some additional phospholipid turnover and diacylglycerol formation occurs which is not dependent on osmolarity (4, 5). This latter turnover may be related to the transfer of phosphorylethanolamine groups to outer membrane lipopolysaccharide (6). Phospholipase C (7,8) and phosphatidic acid phosphatase (9, 10) activities have been reported in E. coli, but their physiologic significance is uncertain.
Raetz and co-workers (11) have described a mutation, dgkR1, which increases diacylglycerol kinase activity in isolated membranes 7-fold. The location of this mutation was shown to be near the melA gene several minutes clockwise from the diacylglycerol kinase structural gene, dgkA (11). These workers have also described a mutation, pssR1, which * This research was supported by Grants GM 20015 and GM 33687 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "duertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed.
increases the activity of the lipid biosynthetic enzyme, phosphatidylserine synthetase, in a trans-acting manner (12). In the case of the pssR mutation, it was possible to demonstrate that the elevated enzyme activity was associated with an increased level of the phosphatidylserine synthetase polypeptide in the cell membrane. The mechanism of the dgkR mutation, however, has not been elucidated. Phospholipid biosynthesis in E. coli is clearly under regulatory controls. Phospholipids are maintained at a constant fraction of membrane mass (13) and pkB mutants, with a greatly reduced in uitro glycerol-3-phosphate acyltransferase activity, require the presence of a second mutation, pkX, to exhibit a growth phenotype (14). The pkX mutation presumably prevents the cellular regulatory machinery from adapting to the mutant pkB activity (14). Regulation is also exerted on the distribution of major phospholipid molecular species. The proportions of these species do not normally change except for a shift of the anionic lipid pool from phosphatidylglycerol to cardiolipin as cell cultures reach the stationary phase of growth (1). The enzymes at the cytidine diphosphatediacylglycerol branchpoint of E. coli lipid biosynthesis have been overproduced both by recombinant DNA methods (15,16) and, in the case of phosphatidylserine synthetase, by the presence of a trans-acting mutation (12). The changes in the cellular phospholipid composition induced by these manipulations have always been small in comparison to the changes in in uitro enzyme activity (12, 15, 16). Moreover, mutant strains of E. coli have been isolated in which the activity of one of the cytidine diphosphate-diacylglycerol branchpoint enzymes was markedly reduced, but the cellular phospholipid composition was unaltered (17, 18). These results indicate the functioning of enzymatic mechanisms in the regulation of phospholipid composition. The dgkR and pssR mutations indicate that genetic regulation of phospholipid metabolism is also occurring. Information on the molecular basis of these regulatory mechanisms and their integration in the overall control of phospholipid biogenesis is, however, lacking. In this paper we report studies on the mechanism of action of the trans-acting dgkRl mutation on diacylglycerol kinase activity and structural gene expression, the effect of elevated cellular diacylglycerol' levels on dgkA expression, the effect of dgkA expression on diacylglycerol levels, and the effect of dgkR1 on diacylglycerol levels. The effect of the dgkRl mutation on the activity of glycerol-]? acyltransferase, which is transcribed divergently from diacylglycerol kinase off of a 170-base pair segment of DNA, was also investigated. The dgkRl mutation was shown to act on dgkA expression at the transcriptional level. Increased cellular diacylglycerol production, induced by growth on a low osmolarity medium, did not affect dgkA expression. Increased diacylglycerol kinase activ-'Throughout this paper, the term "diacylglycerol" refers exclusively to sn-1,2-diacylglycerol. ity due to amplification of the enzyme's structural gene on a high copy number plasmid decreased the cellular diacylglycerol level. However, this effect was suppressed by the dgkRl mutation. Finally, we demonstrated that glycerol-3-phosphate acyltransferase activity was reduced in dgkRl mutants, indicating that the genes encoding this enzyme and diacylglycerol kinase may be regulated reciprocally.
DNA-All plasmids used in this work are listed in Table I. Plasmid DNA was prepared by alkaline lysis of 1.5 ml overnight cultures as described in Maniatis et al. (28) through the ethanol precipitation step. Eppendorf microcentrifugation was for 10 min in all cases. The ethanol pellets were resuspended in 100 pl of 10 mM Tris, 1 mM EDTA, pH 8.0 (TE). Ammonium acetate (50 pl of 7.5 M) was then added and a precipitate allowed to form at room temperature for 5 min. After centrifugation at room temperature, the supernatants were transferred to fresh tubes, 0.3 ml of ethanol was added, and a precipitate allowed to form for 30 min at -20 "C. After centrifugation at 4 "C, the pellets were dried for a few minutes in uacuo. The dried pellets were resuspended in 50 pl of TE containing 50 Ng/ml ribonuclease A and incubated at 37 "C for 1 h. Plasmid DNA was then precipitated by adding 50 pl of 20 mM spermine, 200 mM NaCl and keeping the tubes on ice for 15 min (30). After centrifugation at 4 "C, the supernatants were discarded. The spermine pellets were resuspended in 100 pl of 10 mM Tris, 100 mM NaC1, 1 mM EDTA, pH 8.0, 250 pl of ethanol added, and the DNA allowed to precipitate for at least 30 min at -70 "C. The supernatants were discarded after centrifugation at 4 "C and the pellets dried for a few minutes in uacuo. The dried pellets were resuspended in 50 p1 of TE and stored at 4 "C until used. Restriction endonuclease digestions were performed as recommended by the suppliers. Other nucleic acid methods, including agarose-ethidium bromide electrophoresis, DNA ligations, and attachment of oligonucleotide linkers were according to standard procedures (28,31).
Preparation of Membranes and Cell Free Supermtants-"embranes were prepared from 100 ml cultures of cells grown in LB medium to an optical density at 600 nm of 0.6. Cells were harvested by centrifugation at 7,000 X g for 10 min and resuspended in 25 ml of 0.1 M sodium phosphate, pH 7.0, 5 mM 2-mercaptoethanol, 1 mM diethylenetriaminepentaacetic acid. The cells were lysed by one passage through a French pressure cell at 18,000 psi. Unbroken cells and large debris were removed by centrifugation at 5,000 X g for 10 min. Membranes were then pelleted by centrifugation at 160,000 X g for 90 min and resuspended. in 1 ml of 20% glycerol, 25 mM sodium phosphate, pH 7.0, 5 mM 2-mercaptoethanol, 1 mM diethylenetriaminepentaacetic acid. These membrane suspensions were stored at -70 "C until used.
Cell free supernatants used in assays of soluble enzymes were prepared from 12.5 ml cultures of cells grown to an optical density at 600 nm of 0.6. Cells were harvested by centrifugation at 7500 X g for 10 min and washed in 5 ml of the appropriate assay buffer. The washed cell pellets were resuspended in 2 ml of assay buffer and the cells broken by two 15-8 bursts from a Branson sonifier at a microtip setting of 3 (40 W) on ice. The broken cell suspensions were centrifuged for 15 min at 4 "C in an Eppendorf microcentrifuge and the supernatants saved for assay.
Enzyme Assays-Diacylglycerol kinase (EC 2.7.1.107) was assayed as described previously (21). From 0.1 to 4 pg of membrane protein was used, depending on the level of diacylglycerol kinase expression.
Glycerol-3-phosphate acyltransferase (EC 2.3.1.15) was assayed by a modification4 of a previously described method (32). Membranes (1-2 mg of protein/ml) were first diluted to 0.35 mg protein/ml with an appropriate volume of 25 mM Tris, pH 8.4, 5 m M 2-mercaptoethanol. Triton X-100 was added to a final concentration of 0.5% and the membranes allowed to solubilize on ice for 10 min. Then 20 pl of this mixture was combined with 180 pl of reconstitution buffer (described below) and allowed to stand on ice for an additional 10 min. Assays were performed by combining 2-20 pl of reconstituted enzyme, 50 pl of 2X GPAT buffer (0. press. coenzyme A, 5 pl of 20 mg/ml bovine albumin, and water to give a total of 80 pl. The reaction was initiated by addition of 20 pl of 10 mM sn-[3H]glycerol-3-phosphate and terminated after 10 min at 25 "C by addition of 0.6 ml of 1% HC10, and 3 ml of methanol/chloroform (2:1, v/v). The [3H]lysophosphatidic acid product was extracted into 1.8 ml of chloroform as described previously (21) and 1 ml of this chloroform dried in a hot water bath and counted for 3H in 4 ml of Aquasol-2. The concentration of the palmitoyl coenzyme A stock solution was adjusted spectrophotometrically using an extinction coefficient at 260 nm of 15.4 cm" mM" (33). Reconstitution buffer was prepared by combining equal volumes of 2 X GPAT buffer (described above), water, and 10 mg/ml E. coli phospholipid. The phospholipid was prepared by drying down a chloroform solution of E.
coli phospholipid under a stream of N2 and then suspending it in an appropriate volume of 150 mM Tris, pH 8.4, by three 20-s bursts from a Branson sonifier at a microtip setting of 4 (50 W) on ice.
D-Lactate dehydrogenase (EC 1.1.2.4) was assayed by a modification of a published procedure (34). Membranes (2-10 pg of protein) were combined with assay buffer (0.1 M potassium phosphate, pH 7.5) to give a total volume of 0.69 ml. N-Ethylmaleimide (10 pl of 100 mM) and Triton X-100 (0.1 ml of a 1% solution in assay buffer) were added. After a few minutes at 25 "C, the reaction was initiated by adding 0.1 ml of 0.1 M lithium 0-lactate and 0.1 ml of 0.8 mM 2,6dichlorophenol-indophenol, both in assay buffer. The absorbance at 600 nm was monitored for 10 min at 25 "C in a Gilford recording spectrophotometer. Specific activities were calculated using an extinction coefficient for dichlorophenol-indophenol of 20.6 cm" mM" (35). These activities were corrected for a small dichlorophenolindophenol reductase activity observed in the absence of lactate (1-3% of the lactate-dependent activity).
Galactokinase (EC 2.7.1.6) was assayed by a modification of previous methods (27,36,37). Cell free supernatants were diluted to 2-200 pg of protein/ml in 100 mM Tris, pH 7.8, 1 mM EGTA, 10 mM dithiothreitol. Fifty pl of this diluted supernatant was combined with 100 pl of 200 mM Tris.HC1, pH 7.8, 20 mM NaF, 2 mM EGTA, and 50 pl of 7.5 mM Na2ATP, 20 mM MgCl2, pH 6.5, and incubated for several minutes at 30 "C. The reaction was initiated by adding 50 pl of 5 mM D-['4C]galaCtOSe (0.4 Ci/mol), 0.1 mM diethylenetriaminepentaacetic acid, and 100 pl of the mixture was immediately loaded onto a 1 ml Pasteur pipet column of AG 1-X8 for background counts and washed onto the column with 0.5 ml of water. After 15 min at 30 "C, another 100 p1 of the reaction mixture was similarly loaded onto a second AG 1-X8 column. The columns were washed with 5 ml of water and then eluted into scintillation vials with 4 ml of 1 N HC1. Eight ml of Aquasol-2 was added and the vials counted for "C.
8-Galactosidase (EC 3.2.1.23) was assayed by a modification of previous methods (38, 39). Cell free supernatants were prepared and assayed in 0.1 M sodium phosphate, pH 7.25, 1.25 mM MgSOr, 12.5 mM 2-mercaptoethanol. In a typical assay, 10-200 pl of supernatant was brought to a total volume of 0.8 ml with assay buffer and the sample incubated a few minutes at 30 "C. The reaction was initiated by adding 0.2 ml of 60 mM o-nitrophenyl-8-D-galactopyranoside in 0.1 M sodium phosphate, pH 7.25, and allowed to proceed at 30 "C until a light yellow color developed. The reaction was then terminated by addition of 0.5 ml of 1.0 M NaZC03 and the absorbance at 416 nm determined. Activities were calculated using an extinction coefficient for o-nitrophenolate of 4.60 cm-I mM" (40).
8-Lactamase (EC 3.5.2.6) activity was also determined by a modification of previous methods (41,42). Cell free supernatants were prepared and assayed in 0.1 M sodium phosphate, pH 7.0, 1 mM EDTA. In a typical assay, supernatant (0.2-1.0 pg of protein from ampicillin-resistant cells, 10-40 pg from sensitive cells) was diluted with assay buffer to give a total volume of 0.79 ml. N-Ethylmaleimide (10 pl of 100 mM) was added and the samples incubated a few minutes at 30 "C. The reaction was initiated by addition of 0.2 ml of 2.5 mM nitrocefin, 5% dimethyl sulfoxide, 0.1 M sodium phosphate, 1 mM EDTA, pH 7.0, and the absorbance at 495 nm monitored in a recording spectrophotometer at 30 "C. The nitrocefin stock solution was prepared by adding 1 part of 50 mM nitrocefin in dimethyl sulfoxide to 19 parts of assay buffer. Activities were calculated assuming that cleavage of the nitrocefin 0-lactam ring caused a change in absorbance at 495 nm of 14.1 cm" mM" (42).
High Performance Liquid Chromatography-Preparation of E. coli inner membranes and subsequent high performance liquid chromatography analysis of their content of diacylglycerol kinase polypeptide were performed as described previously (20).
Analysis of Total Cellular Diacylglyceml-Total E. coli lipids were extracted as described by Rotering and Raetz (4) except that the extracting solvent was not acidified in order to avoid acyl chain migration of sn-1,2-diacylglycerols (43). Cultures (100 ml) were grown to an optical density at 600 nm of 0.6 and harvested by centrifugation. The cells were washed in 0.2 M sodium phosphate, pH 7.0, 1 mM EDTA and resuspended in 1 ml of 1.0 M NaCl. Four ml of methanol and 2 ml of chloroform were added. After 1 h the cell debris was removed by centrifugation and the supernatants transferred to fresh tubes. Phases were broken by addition of 2 ml of chloroform and 2 ml of 1.0 M NaCl and separated by centrifugation. The lower chloroform phase was washed twice with 4 ml of 1.0 M NaCl and saved for analysis. Diacylglycerol content was determined by quantitative phosphorylation with [ Y -~~P I A T P using E. coli diacylglycerol kinase as described elsewhere (44). The radioactive product from every experiment was analyzed by thin layer chromatography and, in all cases, comigrated as a single spot with the phosphatidic acid standard.
No additional 32P products were observed. Total lipid phosphorous was determined by the method of Ames and Dubin (45). If the insoluble material from the organic extraction was re-extracted with acidified solvent (41, no diacylglycerol and only 2% of the initial phosphorous were recovered. Other Methods-Incorporation of [3H]glycerol into membranederived oligosaccharides and total cellular lipid was determined as described by Rotering and Raetz (4). Protein concentrations were determined by the Peterson modification of the method of Lowry (46). Thin layer chromatography of 32P-phosphorylated E. coli lipid extracts was performed on 0.25-mm silica gel 60 plates (Merck). The plates were developed in chloroform/methanol/acetic acid (65255, V/V).

RESULTS AND DISCUSSION
Overproduction of the Diacylglycerol Kinase Polypeptide in dgkR Bearing E. coli Cells-The dgkRl mutation results in a 7-fold elevation of diacylglycerol kinase activity in membranes (11). To determine whether this mutation caused an increase in the diacylglycerol kinase polypeptide or enhanced activity by some other mechanism, membranes from dgkRl and dgkR' E. coli strains were extracted with trifluoroacetic acid/heptane/2-propanol (0.025:1:4, v/v) and the extracted peptides subjected to high performance liquid chromatography analysis. Prior work demonstrated that the diacylglycerol kinase polypeptide was quantitatively extracted and migrated as a single peak (20). The strains used were transformed with pJW1, a high copy number plasmid which contains the diacylglycerol kinase structural gene (20), to increase the diacylglycerol kinase polypeptide signal. The high performance liquid chromatography profile of the dgkRl strain, GK1/ pJW1, had a prominent diacylglycerol kinase polypeptide peak which was absent in the dgkR+ strain, R477/pJW1 (Fig.  1). This demonstrated that the dgkR1 mutation increased the level of diacylglycerol kinase polypeptide and ruled out a posttranslational modification mechanism for the increased diacylglycerol kinase activity. However, this experiment did not establish the mechanism causing the elevated diacylglycerol kinase polypeptide, which could reflect changes in transcription, translation, or enzyme degradation. Hence, additional studies on the mechanism of the dgkRl mutation were undertaken.
Construction of dgkA Transcriptional and Translational Fusion Plasmids-To ascertain whether the dgkRl mutation affects diacylglycerol kinase transcription, translation, or both, fusion plasmids were constructed with the dgkA promoter controlling the synthesis of /%galactosidase and galactokinase. The structures of these plasmids are shown in Fig.   2, and the details of their construction are given in the figure legend. Plasmid pJW27 encodes a translational fusion of the first two amino acids of diacylglycerol kinase to a functional @-galactosidase. The absence in the fusion gene product of diacylglycerol kinase sequences involved in membrane binding was considered important inasmuch as binding of @galactosidase to the membrane might interfere with its activ- and GKl/pJWl (dgkR1) were extracted into organic solvent (heptane/2-propanol/trifluoroacetic acid, 1:4:0.025, v/v) and analyzed by HPLC as described (20). Absorbance at 280 nm was monitored. The majority of extracted protein eluted as a single peak at 4.4 min while diacylglycerol kinase eluted at 5.4 min. The peaks at 6.6 and 7.2 min are due to phospholipid while the peak at 7.8 min was due to 2mercaptoethanol (20). Tracing A is a profile of the organic extract (12.5 pg of protein) from membranes of R477/pJW1 while Tracing B is the same amount of GKl/pJWl extract. Specific activities of the membrane preparations used were 0.39 and 4.64 pmol/min/mg of protein, respectively. An arrow indicates the prominent diacylglycerol kinase peak in GKl/pJW1. Near quantitative (greater than 90%) recovery of diacylglycerol kinase in the 5.4-min peak was demonstrated previously using 3H-labeled diacylglycerol kinase prepared in maxicells (20). Additional details of the methods used are given elsewhere (20).
ity (47). Plasmid pJW33 encodes a messenger RNA which is transcribed under control of the dgkA promoter. The galactokinase synthesized by this mRNA is under the control of its own translational initiation signals (27). A short peptide which begins with the first two amino acids of diacylglycerol kinase and terminates 74 base pairs prior to the galactokinase start codon is also predicted from the DNA sequence. The use of both of these plasmids should effectively distinguish transcriptional and translational effects on dgkA expression.
The dgkRl Mutation Increases dgkA Transcription-To investigate the mechanism of the dgkR1 mutation, diacylglycerol kinase, p-galactosidase, and galactokinase activities were determined in E. coli strains R477 (dgkR+) and GK1 (dgkR1) bearing the dgkA fusion plasmids. To exclude any plasmid copy number effects, p-lactamase activities were also determined. The @-lactamase activity expressed by these plasmids has been shown to be proportional with gene dosage (48)(49)(50).
Inasmuch as E. coli membrane preparations can be contaminated with a variable amount of outer membrane, the membrane bound D-lactate dehydrogenase activity was assayed as an inner membrane marker (34). The specific activities observed are shown in Table 11. Diacylglycerol kinase activity was increased 6.2-8.2-fold in strain GK1 as compared to R477 with all of the plasmids examined with the exception of pJW1. When these activities were corrected for small variations observed in the marker lactate dehydrogenase activity, the range of activity ratios narrowed to 6.6-7.3. The anomalously Plasmid pJW27 was constructed by inserting a 2700-base pair EcoRI linkered fragment which included the entire plsB gene and the first two amino acids of dgkA into the translational fusion vector, pMLB1034 (26). Recombinants were selected by transformation of strain VL3 to ampicillin resistance and glycerol-3-phosphate prototrophy. Several tetracycline-sensitive recombinants were identified which gave a dark blue reaction on XGAL plates and one of them, pJW27, was verified by restriction mapping to have the structure shown. To construct plasmid pJW33, the same 2700-base pair EcoRI fragment was inserted into the transcriptional fusion vector, pJW18, by selection in strain VL3 as described above. Recombinants with the insert in the correct orientation were identified by restriction mapping. Plasmid pJWl8 was derived from pKOl (27)

by insertion of a d(GGAATTCC) EcoRI linker into the SmaI site and concomitant
deletion of the small EcoRI-SmI fragment. The construction of plasmid pJW20, the source of the 2700-base pair, plsB-containing DNA fragment used in these constructions, is detailed elsewhere? high diacylglycerol kinase induction observed with plasmid pJWl was probably due to an increased plasmid copy number in the dgkRl strain as indicated by the @-lactamase a~tivities.~ The presence of promoter DNA on a high copy number plasmid can lead to an anomalous induction or repression of its activity due to titration of DNA binding regulatory molecules (51-53). The excellent agreement of the activity ratios in cells bearing the fusion plasmids with the ratios observed in cells not bearing a plasmid indicates, however, that such an effect is not occurring with dgkA. When p-galactosidase and galactokinase activities were determined in these strains bearing the fusion plasmids, activity inductions of 9.1-and 8.3-fold, respectively, were obtained. When these numbers were corrected for differences in plasmid copy number, as reflected by the @-lactamase activities, the ratios became 7.0and 7.5-fold, respectively. Background activities, observed in these strains without the fusion plasmids, were negligible. 6 The excellent agreement of the ratios for native diacylglycerol kinase activity with those for the fusion plasmid activities indicates that the effect of the dgkRl mutation is exerted solely at the level of transcription.
Effect of the dgkR Mutation on plsB Expression-In E. coli the pkB and dgkA genes are transcribed divergently from a 170-base pair region of DNA (54). It is, therefore, possible When the diacylglycerol kinase activities with plasmid pJWl are corrected for both the marker lactate dehydrogenase and the difference in 8-lactamase activities, the activities ratio becomes 5.8. An plasmid pCV1, the parent vector of pJWl (20). 8-Lactamase activities increased plasmid copy number in strain GK1 was also observed with in strains R477/pCV1 and GKl/pCV1 were 11.8 and 19.6 pmol/min/ mg of protein, respectively. This effect is therefore not due to the presence of cloned dgkA DNA.
'jPlasmids pMLB1034 and pKO1, the parent vectors for pJW27 and pJW33, do not encode @-galactosidase or galactokinase activities (25,26). In strain R477, the background 8-galactosidase and galactokinase activities were 0.003 and 0.01 pmol/min/mg of protein, respectively. Identical background activities were observed in strain GK1. that factors affecting the expression of dgkA could also influence the expression of plsB. The activity of sn-glycerol-3phosphate acyltransferase, the product of the pkB gene, was therefore determined in the dgkR mutant strain, GK1, and its dgkR+ parent, R477, with and without plasmids containing p k B DNA (Table 111). In all cases, the activities were reduced 30-50% in the dgkR mutant as compared to wild type. The same result was obtained if the membranes were assayed directly as described previously (32) without the prior solubilization and reconstitution (data not shown). The observed decreases in activity were not affected by correction for small differences in the marker lactate dehydrogenase (data not shown). This result suggests that the dgkA and pkB genes may be reciprocally regulated. Additional studies will be needed, however, before a firm conclusion can be drawn.
Effect of Growth Medium Osmolarity on Diacylglycerol Kinase Gene Expression-During growth on media of low osmolarity, E. coli synthesizes large amounts of membranederived oligosaccharide, generating diacylglycerol as a byproduct. When maximally induced, MDO biosynthesis is the major source of cellular diacylglycerol (4, 5). Expression of dgkA may thus be regulated by growth medium osmolarity.
T o examine this possibility, diacylglycerol kinase activities were determined in membranes from strains GK1 (dgkR1) and R477 (dgkR+) grown on media of low osmolarity. The results are shown in Table IV. These activities, after correction for small differences in the marker lactate dehydrogenase, differed by less than 10% from those observed in NaCl grown cells (Table 11). Examination of the activities encoded by the fusion plasmids (Tables I1 and IV) also failed to reveal an effect of osmolarity on dgkA expression in either train.^ Diacylglycerol kinase activities were also determined in membranes from cells grown on high and low osmolarity peptone medium. Again, medium osmolarity was without effect in either strain (data not shown). The ability of the dgkRl mutant to induce MDO biosynthesis in response to low osmolarity was also examined (Table V). Total MDO biosynthesis, expressed as the ratio of [3H]glycerol counts incorporated into MDO to counts incorporated into total lipid, was unaffected by this mutation. These results indicate that the induction of MDO biosynthesis by low osmolarity is not The small decreases in the fusion plasmid-encoded activities on low osmolarity media were accompanied by corresponding decreases in the P-lactamase activities, indicating a plasmid copy number effect. A similar decrease in 0-lactamase activity was observed in strains bearing plasmid pBR322 grown on low osmolarity medium. 8-Lactamase activities in strains R477/pBR322 and GKl/pBR322 were, respectively, 15.0 and 14.9 pmol/min/mg of protein on LB with 0.1 M NaCl and 12.4 and 12.8 pmol/min/mg of protein on LB without dgkA DNA.
the NaC1. This effect is therefore not due to the presence of cloned (dgk+) and GKl (dgkR1) bearing various phmirki Activities were determined as described in the "Experimental Procedures" section. Activities with pJW27 and pJW33 are higher because of the presence on these plasmids of the glycerol-P acyltransferase structural gene, pkB. Glycerol-P, sn-glvcerol-3-phosphate.  accompanied by an increase in dgkA expression and is not affected by the dgkR locus. Effects of Growth Medium Osmolarity and Diacylglycerol Kinase Overproduction on Cell Diacylglycerol Levels-The observations that diacylglycerol kinase expression and activity were unaffected by growth medium osmolarity suggested that cellular diacylglycerol levels may vary in response to osmolarity. Diacylglycerol levels were, therefore, determined in strains R477 and GK1 grown on media of high and low osmolarity (Table VI). Growth on a low osmolarity medium caused a 3to %fold increase in cell diacylglycerol. The mole percent increase was similar for all strains on a given medium, about 0.3% for cells grown on LB and 0.5% on peptone. This suggested that removal of diacylglycerol generated during MDO biogenesis was not limited by diacylglycerol kinase  activity. In dgkR+ cells grown on high osmolarity media, enhanced expression of diacylglycerol kinase by amplification of its structural gene on the high copy number plasmid, pJW1, resulted in a 2.5-to 4-fold reduction of the diacylglycerol content, However, in the dgkRl genetic background, overproduction of diacylglycerol kinase activity up to 100-fold was without effect on the diacylglycerol levels. The dgkR mutant strain, GK1, apparently has an alteration in its diacylglycerol metabolism which is additional to the increased dgkA transcription. Diacylglycerol levels in cells grown on the high osmolarity peptone medium were generally higher than those grown on the high osmolarity LB. This probably reflects a difference in the rate of diacylglycerol generation on these two media, inasmuch as a similar increase was observed in strain RZ6, which lacks diacylglycerol kinase activity.

CONCLUDING DISCUSSION
The present work demonstrates that the E. coli dgkRl mutation acts by increasing the transcription of the dgkA gene. Additionally, the observation of decreased glycerol-3phosphate acyltransferase activities in the dgkR mutant strain suggests that the structural genes for this enzyme and diacylglycerol kinase are reciprocally regulated. The biological significance of dglzR, however, remains unknown.
Given that transcription of dgkA is regulated in E. coli, the question arises as to the mechanism operating on its promoter. The observation that expression is proportional with gene dosage in high copy number plasmid vectors argues against the involvement of a DNA-binding molecule present at only a few copies per cell, such as a repressor (51-53). This result is consistent with previous studies in which genes for lipid biosynthetic enzymes were shown to be regulated in a gene dose dependent manner (15, 16). The availability of the dgkA'-lac'Z fusion plasmid should facilitate the application of E. coli lactose operon methodology (26) to further investigation of the dgkA regulatory mechanism. Induction of diacylglycerol production by growth on a medium of low osmolarity was without effect on dgkA expression, indicating that low osmolarity, by itself, was insufficient to regulate dgkA. This observation is consistent with a previous report that phosphoglycerol transferase I, an MDO biosynthetic enzyme, was synthesized constitutively and not induced by low osmolarity (55). The increase in the steady state membrane diacylglycerol content caused by growth on low osmolarity media was not affected by the amount of diacylglycerol kinase present in the membrane. This may reflect compartmentalization of this diacylglycerol pool. Diacylglycerol generated during MDO biosynthesis is localized to the cytoplasmic membrane (5) and transfer of phospholipid head groups to MDO's occurs in the periplasmic space (56). Therefore, translocation of diacylglycerol generated during MDO synthesis from the outer to the inner leaflet of the cytoplasmic membrane must occur prior to its phosphorylation by diacylglycerol kinase. If this diacylglycerol translocation was rate limiting for its rephosphorylation, then the diacylglycerol in the outer leaflet of the cytoplasmic membrane could be the low osmolarity pool inferred above. However, data from other membrane systems indicates that transmembrane movement of diacylglycerol is an extremely rapid process (57, 58) and other explanations for this result should be considered.
The finding that diacylglycerol levels were decreased by diacylglycerol kinase overproduction only in the dgkR+ genetic background was unexpected. This indicated that the dgkRl mutant strain was additionally altered in its diacylglycerol metabolism in some way not attributable to the enhanced diacylglycerol kinase activity. It is possible that the increased dgkA transcription in this strain is secondary to this other defect. Overall, these results indicate that diacylglycerol metabolism in E. coli is more complicated than previously thought. Given that diacylglycerols function as second messengers in eukaryotic cells (59, 60), the possibility that they function as signal molecules in E. coli as well is worthy of consideration. A recent report that transfer of phosphoglycerol headgroups to MDO, the diacylglycerol producing reaction, was curtailed abruptly by increasing the osmolarity of the growth medium is consistent with such a regulatory role (55). The increased diacylglycerol level produced from MDO biosynthesis may thus be involved in eliciting the numerous cellular responses mounted by E. coli to a low osmolarity environment, including alterations in the outer membrane protein composition (61), induction of K+ transport (62), accumulation of polyamines (63), and accumulation of glycine betaine (64).