Isolation and Characterization of Escherichia coli Strains Defective in CDP-diglyceride Hydrolase*

CDP-diglyceride, an obligatory intermediate in the biosynthesis of the glycerophospholipids in Esche- richia coli, is clFaved in vitro to phosphatidic acid and CMP by a membrane-bound hydrolase. Previous work from our laboratory (Bulawa, C. E., Hermes, J. D., and Raetz, C. R. H. (1983) J. Biol. Chem. 258, 14974- 14980) has demonstrated that this enzyme also catalyzes the transfer of CMP from CDP-diglyceride to phosphate and numerous phosphomonoesters. We now report the isolation of E. coli mutants which are defective in CDP-diglyceride hydrolase. These mutations, designated cdh, map at minute 88 between pfkA and tpi. This information permitted the identification of a ColE1 hybrid plasmid, pLC16-4, which causes the overproduction of hydrolase activity. The isolation of deletion and TnlO insertion mutants at cdh suggests that the hydrolase is nonessential for cell growth. Hy- drolase mutants are defective in both CDP-diglyceride hydrolysis and CDP-diglyceride-dependent cytidyly- lation, indicating that both activities are encoded by the cdh gene. Although previously described as a ri- bospecific enzyme, we have found that incubation of the partially purified hydrolase with [a-’*P]dCDP-di-glyceride and phosphate yields two products, [“PI dCMP and [a-”PIdCDP. That of performed described under Procedures," with one exception. Due to the temperature sensitivity of thepss-21 mutation (15), the isogenic strains in part C were growth at 30 "C. A, CDP-diglyceride hydrolase activity in isogenic cdh::TnlO and rh0::TnlO strains. B, CDP-di-glyceride hydrolase activity in two deletion mutants. C, effect of pss- 21 on hydrolase activity.

ally lethal CDP-diglyceride synthetase mutants are defective in the formation of both phosphatidylethanolamine and phosphatidylglycerol under the nonpermissive conditions (4). Thus, a substantial body of biochemical and genetic evidence has established that CDP-&glyceride is the sole phosphatidyl donor for glycerolipid biosynthesis in E. coli.
The first indication that CDP-diglyceride had an additional metabolic fate came from the work of Raetz et ai. (5). These investigators found that CDP-&glyceride was cleaved in vitro to phosphatidic acid and CMP by a membrane-bound hydrolase. Partial purification of this activity (6) showed that it was distinct from the lipid biosynthetic enzymes. Although several riboliponucleotides, such as CDP-diglyceride, UDPdiglyceride, and CDP-ceramide, were rapidly cleaved by the hydrolase, dCDP-diglyceride was not utilized at an appreciable rate (6). In addition, AMP-and ADP-diglyceride were identified as potent inhibitors of CDP-diglyceride hydrolysis (5, 6). Recently, Rittenhouse et al. have reported a similar AMP-sensitive pyrophosphatase in guinea pig brain (7). In the latter case, both CDP-and dCDP-diglyceride were substrates for the enzyme.
Recent studies from our laboratory have provided new insights into the function of the CDP-diglyceride hydrolase. We have demonstrated that the partially purified hydrolase catalyzes the transfer of CMP from CDP-&glyceride to various phosphomonoester acceptors (8). To further explore the phenomenon of CDP-diglyceride-dependent cytidylylation, we have isolated E. coli mutants defective in CDP-diglyceride hydrolase (9). In the present study, we show that the hydrolase and cytidylyltransferase activities are encoded by a single locus, designated cdh. Contrary to previous reports, we found that dCDP-diglyceride was efficiently hydrolyzed in vitro, and furthermore, that it could serve as a dCMP donor. Analysis of hydrolase mutants revealed the accumulation of CDP-and dCDP-diglyceride, indicating that both liponucleotides are substrates for the hydrolase in vivo.

EXPERIMENTAL PROCEDURES
Materials-Chemicals were purchased from the following companies: ribo-and ~~O X~-[ C X -~* P ] C T P , Amersham Corp.; [5-3H]CTP and 32Pir New England Nuclear; tetrabutylammonium bromide, Aldrich. Other fine chemicals were from Sigma. Purified phosphatidylserine synthase (IO), used for the synthesis of radiolabeled CDP-and dCDPdiglyceride, was the gift of Dr. William Dowhan. Silica Gel 60 thin layer plates (0.25 mm) and PEP-cellulose F thin layer plates (0.1 mm) were the products of E. Merck, Darmstadt, Germany. Aquasol was purchased from New England Nuclear.
Bacterial Strains, Plasmids, and Growth Media-The strains of E. coli K12 used in this work are listed in Table I. Three different media were employed for the growth of bacteria: LB (16), G56 (17). and minimal A (16). Carbon sources were included at a final concentration The abbreviations used are: PEI, polyethyleneimine; Mes, 2-(N-morpho1ino)ethanesulfonic acid; HPLC, high pressure liquid chromatography.

in
Mutants of E. coli Preparation of Crude Extracts-Extracts were prepared from fresh overnight cultures unless otherwise indicated. Cells were grown at 37 "C in LB medium and then harvested at 4 "C by centrifugation.
After washing once with cold 10 mM potassium phosphate, pH 7.5, the cells were resuspended in the same buffer at a final concentration of about 5 mg/mI of protein. The cells were broken by passage through an Aminco pressure cell at 18,000 p.s.i. After removing intact cells by centrifugation, extracts were assayed as described below.
Purification of CDP-diglyceride Hydrolase-CDP-diglyceride hydrolase was partially purified from the overproducing strain JA200/ pLC16-4 through the DEAE-cellulose step (6). The final preparation (8) was purified 300-fold relative to a crude extract from a wild type strain and was free of phosphatidylglycerophosphate synthase and phosphatidylserine synthase activities.
Enzyme Assays-CDP-diglyceride hydrolase activity was determined at 37 "C as follows, unless otherwise indicated. The substrate was either [~x-~~PJCDP-diglyceride or [cr-32P]dCDP-diglyceride (1000 cpm/nmol) at a final concentration of 0.33 mM. In addition, the assay mixture contained 100 mM potassium phosphate, pH 7.5,0.1% Triton X-100, and 0.7 mg/rnl of bovine serum albumin in a total volume of 60 pl (6). Reactions were terminated by Bligh-Dyer extraction at pH 2 (solvents and volumes are given in the legend to Fig. 5), and 2 ml of the upper phase were transferred to a scintillation vial. After the addition of 10 ml of Aquasol, the radioactivity was quantitated by liquid scintillation spectrometry. We recently discovered that when phosphate is used as the buffer, the hydrolase catalyzes the formation of two water-soluble products, [3zP]CMP and [cY-~'P]CDP (8). Therefore, using this assay, CDP-diglyceride hydrolase activity is measured as the nanomoles of CMP plus CDP produced/min/mg of protein.
Cytidylyltransferase activity was determined in a 15-pl reaction mixture containing 100 mM Mes, pH 6.0, 1.3% octyl 8-D-glucoside, and 16.6 mM potassium phosphate. Either [a-3ZP)CDP-or [a-"P] dCDP-diglyceride (0.33 mM, 1 X lo' cpm/nmol) served as the cytidylyl donor, Following incubation at 37 'C for 30 min, the reaction was stopped by acidic Bligh-Dyer extraction, and a portion of the upper phase was analyzed by PEI-cellulose thin layer chromatography as described previously (8). Two different solvent systems were utilized. Solvent A (0.55 M NaCl in 0.2% formic acid) separated nucleotides of different net charge (20), while solvent B (2 M LiCl, 2% boric acid (12, v/v)) resolved ribo-and deoxynucleotides (21). After chromatography, the positions of the radioactive products were determined by in 10 ml of Patterson and Green scintillation fluid (22). Cytidylyl-autoradiography. The spots were scraped from the plate and counted transferase activity is defined as nanomoles of [cY-~'P]CDP formed/ min/mg of protein.
Isolation of a TnlO Insertion in cdh-For mutagenesis of W3110 with TnlO, cells were grown in h ym broth (1% Tryptone, 0.25% NaC1, 0.2% maltose, and 0.01% yeast extract) to early stationary phase, harvested, and resuspended in 0.05 volume of fresh medium.
After addition of hNK370 (b221 cI857 cIl7l::TnlO Ouga261) (N. Kleckner, Biological Laboratories, Harvard University, Cambridge, MA) at a multiplicity of infection of 0.2, the suspension was incubated at 37 'C for 45 min. Portions (1.5 X log cells/plate) were then spread on LB agar containing 15 Mg/ml of tetracycline and 0.0025 M sodium pyrophosphate. The plates were incubated at 42 "C, and approximately 8500 tet' colonies were obtained. To enrich for cdh::Tn20 mutants, the colonies were harvested from the plates, infected with Pluir, and the resulting lysate was used to transduce MA1000 to pfkA+tet'. Recombinants were selected at 37 "C on minimal A plates containing 15 pg/ml of tetracycline and 0.2% mannitol as the sole carbon source. CDP-diglyceride hydrolase activity was scored by colony autoradiography. Transductants were also scored on Mc-Conkey plates containing 1% rhamnose. A rha::TnlO mutant was isolated and used as a cdh+ter' control in experiments studying the cdh-4r:TnlO strain. Genetic manipulations using Pluir were performed as described by Miller (16).
Miscellaneous-Mutagenesis with N-methyl-N'-nitro-N-nitrosoguanidine, preparation of Pluir lysates, genetic crosses, and transductions were carried out as described by Miller (16). Protein concentration was determined by the method of Lowry et al. (23) using bovine serum albumin as the standard. For autoradiography of 3'P-containing compounds, Kodak XAR-5 film was used in conjunction with an intensifying screen (Dupont Lightning Plus). HPLC was performed with a Waters M6000 solvent delivery system, a model 720 system controller, a U6K universal liquid chromatograph injector, and a hmax 480 LC spectrophotometer.

RESULTS
Isolation of Mutants Defective in CDP-diglyceride Hydrolase-Cells of strain RA5 (Table I) were treated with Nmethyl-N'-nitro-N-nitrosoguanidine and plated on LB agar. After incubation at 30 "C for 24 h, the colonies were transferred to filter paper, lysed in situ, the assayed for CDPdiglyceride hydrolase as reported previously (9). In a screening of approximately 16,000 colonies, five putative mutants were identified. Cell-free extracts were prepared from these strains and assayed by quantitative methods. Three strains had reduced levels of CDP-diglyceride hydrolase (data not shown); the most defective mutant, RA5-2, retained approximately 15% of the wild type activity. The isolation of these mutants was reported in a previous communication (9).
Mapping of the CDP-diglyceride Hydrolase (cdh) Defect-Preliminary mapping of cdh-1 was conducted by conjugation of RA5-2 with various Hfr strains. These studies located cdh-1 between the points of origin of KL25 and HfrH (Fig. lA) based on the following observations. (i) In a mating of HfrH with RA5-1, no cdh' streptomycin-resistant exconjugants were obtained when leu+, trp+, or his' was the selected marker. (ii) However, approximately 64% of the argH+ streptomycin-resistant recombinants were cdh+ in a cross of KL25 and RA5-1 (data not shown).
The position of cdh-1 was more precisely determined by Plvir mapping of the minute 85 to 95 interval. Linkage to argH and pfkA (Table I1 and

TABLE I1
Linkage between cdh-1 and markers near minute 88 Transductants from the various crosses were scored for cdh-2 by colony autoradiography, except that the transductants from RA5-1 X AM1 were assayed for hydrolase activity in cell-free extracts. transduce E5482 (rha+ cdh-1 metB-) to methionine prototrophy. Of the 100 transductants scored for the unselected genes, none were simultaneously cdh-1 and rhu-(data not shown). Thus, cdh-1 lies between the rhu and metB loci (Fig. 1B). Like cdh-1, cdh-2 and cdh-3 were co-transducible with metB (data not shown), suggesting that these three mutations were allelic. Finally, these data demonstrate that cdh is distinct from the other enzymes of CDP-&glyceride metabolism, since these genes map far away from minute 88 (Fig. 1).
Overproduction of CDP-diglyceride Hydrolase in a Hybrid Plasmid-bearing Strain-The mapping of cdh-1 facilitated the Mutants of E. coli identification of a ColEl hybrid plasmid carrying the cdh gene. Clones of rhu and pfkA, two genes in close proximity to cdh, had been previously isolated from the Clarke and Carbon hybrid plasmid bank (14,25). Since these plasmids contain as much as 0.5 min of chromosomal DNA (26), it was possible that cdh was also carried on one of these rhu or pfkA clones. Because ColEl plasmids are retained in multiple copies per cell (26), the presence of cdh was expected to produce an elevation in enzyme activity. Cell-free extracts were prepared from a pfkA tpi clone (JA200/pLC16-4), a rha clone (JA200/ pLC5-5), and a control (JA200/pLC44-14), and assayed for CDP-diglyceride hydrolase activity. As shown in Table 111, pLC16-4 caused a 6-fold overproduction of the hydrolase, indicating that the cdh gene was present on this plasmid. In a complementation analysis, pLC16-4 corrected cdh-2 but not glpK(data not shown). Several conclusions can be drawn from these data. First, the overproduction of hydrolase activity by pLC16-4 suggests that cdh is the structural gene for the enzyme, although the possibility that it encodes a regulatory element cannot be excluded. Second, complementation of cdh-2 shows that this mutation is recessive to the wild type allele. Finally, cdh must be counterclockwise from glpK since the latter is not carried on pLC16-4.
Isolation of Deletion and TnlO Insertion Mutants at cdh-The most defective strain isolated in our initial screening still possessed 15% of the activity present in wild type strains. We attempted to obtain mutants completely defective in the hydrolase in two ways: 1) by insertion of TnlO within cdh and 2) by deletion of the cdh gene. 1) To accomplish the former, cells of W3110 were infected with X::TnlO, and random transposition into the chromosome produced a collection of tet' mutants. To enrich for insertions near minute 88, a Pluir lysate was prepared from these cells and used to transduce MA1000 (pfM tet8) to pfkA+tet'. Hydrolase activity was scored by colony autoradiography, and from approximately 500 pfM'tet' colonies, one putative cdh::TnlO mutant was obtained. When this strain was assayed by quantitative methods, it was found to have reduced but still measurable levels of enzymatic activity (Table IV,

experiment A). 2)
The isolation of a cdh deletion mutant was greatly facilitated by the availability of strains with chromosomal deletions in minute 88 region (27). We scored the presence of cdh in these strains by in vitro assay, and the results are presented in Table IV, experiment B. Removal of rha-pfkA had no effect on hydrolase activity, but extension of the deletion through tpi removed at least a portion of the cdh gene, as judged by the reduced levels of hydrolase in ET2036. Similar to the cdh::TnlO and cdh-2 mutants, the Arha-tpi strain retained approximately 15% of wild type activity.
The implications of this data for the mapping of the cdh locus are summarized in Fig. 1B. Since JFlOlO has normal levels of the hydrolase, cdh must be clockwise from the rha-pfkA deletion end point. The low activity in ET2036 confirms

Comparison of the residual activity in various cdh mutants
The growth of bacteria, the preparation of extracts, and the determination of CDP-diglyceride hydrolase activity were performed as described under "Experimental Procedures," with one exception. Due to the temperature sensitivity of thepss-21 mutation (15), the isogenic strains in part C were growth at 30 "C. A, CDP-diglyceride hydrolase activity in isogenic cdh::TnlO and rh0::TnlO strains. B, CDP-diglyceride hydrolase activity in two deletion mutants. C, effect of pss-21 on hydrolase activity. the counterclockwise placement of cdh relative to glpK but fails to orient cdh with respect to tpi. This remaining ambiguity has recently been resolved by the work of Silverman and eo-workers. In a Southern blot analysis of MW1104, the transposon, and, therefore, cdh, mapped between pfM and tpi.2 Some of the Residual Activity in cdh Mutants Is Due to Phosphatidylserine Synthase-As discussed above, we obtained mutants defective in CDP-diglyceride hydrolase by three methods: chemical mutagenesis, transposon insertion, and deletion. In each case, approximately 15% of the wild type activity remained. This constant amount of residual activity suggested the existence of one or more additional enzyme(s) capable of cleaving CDP-&glyceride in uitro. Of particular relevance was the previous observation that phosphatidylserine synthase will catalyze CDP-diglyceride hydrolysis in the absence of serine (19). To assess the contribution of this enzyme to the residual activity, a cdh-4::TnlO pss21 double mutant was constructed. Assay of a crude extract prepared from this strain indicated that phosphatidylserine synthase accounted for about 80% of the activity in the cdh-4::TnIO mutant (Table IV, experiment C).

Experiment
The activity remaining in RB421 (cdh-4::TnlO pss21), which is only 5% of the wild type value, may represent the minimal level of hydrolase able to support cell growth. Alternatively, the hydrolase may be nonessential, and nonspecific pyrophosphatases may account for the 5% residuum. Current mapping data support the latter. Since cdh is counterclockwise from tpi, the rhu-tpi deletion should remove the entire cdh gene. Recently, Southern blot analysis has confirmed the absence of cdh DNA in ET2036.2 Thus, CDP-diglyceride hydrolase appears to be a nonessential enzyme in E. coli.
Characterization of CDP-diglyceride Hydrolase Mutants-To determine whether or not there was a phenotype associated with the cdh lesion, we examined the growth of hydrolase mutants under a variety of conditions. None of the cdh mutants were temperatureor osmotically sensitive. Furthermore, exposure to numerous detergents, colicins, and antibiotics failed to reveal any difference between cdh-and cdh' strains.
Next, we analyzed the phospholipid composition of JB1105, JB1204, and JB1204/pLC16-4; these strains possess reduced, normal, and elevated levels of CDP-diglyceride hydrolase, respectively. After uniformly labeling cells with 32Pi, the phospholipids were extracted by the method of Bligh and Dyer (28) and separated by thin-layer chromatography on silicic acid. The results, presented in Table V, A, showed that neither the hydrolase mutant nor the overproducer had any alteration in the levels of the major phospholipids.
Since CDP-diglyceride is a substrate for the hydrolase in vitro, we looked for a perturbation in the cellular levels of this metabolite in hydrolase mutants and overproducers. To mea-

TABLE V
Comparison of the phospholipid composition and the CDP-diglyceride levels in cdh mutants and overproducers Cells were grown for several generations in G56 medium (17) containing 32Pi at approximately 150 gCi/ml. At an A m of 0.6-0.8, 0.7 ml of culture was added to 3 ml of ch1oroform:methanol (1:2, v/ v) containing 10 nmol of cold carrier dCDP-diglyceride. In addition, 10 nmol of [5-3H]CDP-diglyceride (1 X 10' cpm/nmol) was included as an internal standard. The sample was acidified with 100 p1 of 1 M HC1, then centrifuged to remove insoluble debris. After the addition of 1 ml of chloroform and 1 ml of 1 M NaCI, pH 2, the phases were separated by centrifugation. The chloroform layer was retained and washed 2 times with pre-equilibrated upper phase. A, to determine the phospholipid composition, a portion of the lower phase was appliedta two Silica Gel 60 thin-layer plates. One plate was developed in ch1oroform:methanol:acetic acid (652510, v/v) which separated phosphatidylglycerol, phosphatidylethanolamine, and cardiolipin. The other was run in ch1oroform:pyridine:formic acid (50307, v/v) which resolved phosphatidic acid from the major phospholipids. After chromatography, the spots were located by autoradiography, scraped into scintillation vials, and counted in 10 ml of Patterson and Green scintillation fluid (22). B, CDP-and dCDP-diglyceride pools were measured by a modification of the method of Snider (21). The remaining lower phase from part A above was dried under N2 and treated with purified phosphatidylserine synthase. In addition to 0.2 pg of enzyme, the reaction mixture contained 2 mM L-serine, 100 mM potassium phosphate, pH 7.4, 0.1% Triton X-100, and 1 mg/ml of bovine serum albumin in a total volume of 50 pl. After 90 min of incubation at 30 'C, the reaction was stopped with 300 p1 of chloroform:methanol (12, v/v). Next, the water-soluble products were extracted by the addition of 30 p l of water, 100 pl of 75 mM HCl, and 100 p1 of chloroform. For analysis by HPLC, the upper phase was dried under Nz and then redissolved in 100 pl of 0.2 mM CMP and dCMP, which served as chemical standards. A 75-pl portion was injected onto a Waters pBondapak C18 column and eluted isocratically with 5 mM potassium phosphate, pH 5.5,5 mM tetrabutylammonium bromide:rnethanol (9:1, v/v) at 2 ml/min. Fractions (1 ml) were collected, and 0.75-ml portions were analyzed for 32P and 3H in 10 ml of Aquasol. The results for JB1104 are depicted in Fig. 2 ' Values are corrected for the recovery of [5-3H]CDP-diglyceride, which was between 75-78% for the three determinations shown above. It was assumed that CDP-and dCDP-&glyceride were recovered with equal efficiency. sure CDP-and dCDP-diglyceride pools, cells were grown in medium containing 32Pi and then Bligh-Dyer extracted (28) under acidic conditions (see legend to Table VU). After the addition of [5-3H]CDP-diglyceride as an internal standard, the chloroform-soluble material was incubated with purified phosphatidylserine synthase and serine which generated the reaction, The water-soluble reaction products were isolated and analyzed by high pressure liquid chromatography. A typical elution profile is depicted in Fig. 2, and the quantitation of the results is given in Table V, B. Comparison of cdh' and cdhstrains revealed a 5-fold elevation in the level of CDP-diglyceride in the mutant, indicating that this compound is a physiologically relevant substrate for the hydrolase. Surprisingly, dCDP-diglyceride also accumulated, a finding which was not expected since earlier studies indicated that this liponucleotide was not utilized in vitro (6). Overproduction of the hydrolase had no effect on either ribo-or deoxy-CDPdiglyceride pools, suggesting that the excess enzyme is not active in vivo. dCDP-diglyceride Is a Substrate for the Hydrolase-In order to explain the elevation of dCDP-diglyceride in vivo, we reexamined the substrate specificity of the hydrolase. Contrary to previous studies, we found that the partially purified hydrolase converted [~~-~~P]dCDP-diglyceride to a water-soluble radioactive product which was identified as dCMP by PEIcellulose thin-layer chromatography (Fig. 3, lnne 5). While this reaction was linear with both time and protein (data not shown), the rate was 5-to 6-fold lower than the hydrolysis of the ribo-substrate (Table VI, A). Identical results were obtained using crude cell extracts (data not shown).
Importantly, utilization of dCDP-&glyceride by the partially purified hydrolase showed three characteristics previously observed with the ribo-substrate. 1) Analogous to CDPdiglyceride hydrolysis, cleavage of dCDP-diglyceride was optimal at an acidic pH. As shown in Table VI, A, changing the pH from 7.5 to 6.0 produced a 5-fold stimulation in dCDPdiglyceride hydrolysis. Since CDP-diglyceride hydrolysis increased by only 70%, the discrepancy between ribo-and deoxy-activities was less than 2-fold at pH 6.0. 2) AMP, a potent inhibitor of CDP-diglyceride hydrolysis (6), also re- duced dCDP-diglyceride utilization (Table VI, B). 3) Both CDP-and dCDP-diglyceride functioned as nucleotidyl donors (discussed further below). As shown in Fig. 3, lane 6, the hydrolase catalyzed the formation of dCDP in a reaction dependent on inorganic phosphate. In addition, nonradioactive CDP-diglyceride inhibited the hydrolysis of [a-"PI dCDP-diglyceride (Table VI, B), suggesting that a single enzyme was responsible for the cleavage of both liponucleotides. When the converse experiment was performed, the expected inhibition of [a-"PICDP-diglyceride hydrolysis by dCDP-diglyceride was observed (Table VI, B). However, higher levels of dCDP-diglyceride were required, perhaps indicating a lower affinity of the enzyme for the deoxy-compound. Further evidence supporting dCDP-diglyceride as a hydro-

TABLE VI Characterization of dCDP-diglyceride hydrolase activity
A, comparison of the rates of CDP-and dCDP-diglyceride hydrolysis a t pH 7.5 and 6.0. Partially purified CDP-diglyceride hydrolase (0.2 pg) was incubated with 0.33 mM of the indicated substrate in either 100 mM potassium phosphate, pH 7.5, or 100 mM potassium phosphate, 100 mM Mes, pH 6.0. In addition, the assay mixture contained 0.1% Triton X-100 and 0.7 mg/ml of bovine serum albumin. Other buffers could be substituted for Mes with similar results. Omission of the potassium phosphate decreased the specific activity with both substrates by eliminating CDP and dCDP synthesis. B, effect of various inhibitors on the hydrolysis of CDP-and dCDPdiglyceride. Assays were performed in 100 mM potassium phosphate, 100 mM Mes, pH 6.0, with either CDP-or dCDP-diglyceride and 0.2 pg of DEAE-purified enzyme. Inhibitors were present at the final concentrations indicated.  lase substrate was obtained by assaying cdh mutants and clones. As shown in Table VII, cdh::TnlO strains are defective in both CDP-and dCDP-diglyceride hydrolysis, while pLC16-4 causes a 9-fold overproduction of both activities. Thus, both compounds appear to be substrates for the cdh gene product.
Taken together, these data are consistent with the in vivo accumulation of CDP-diglyceride and dCDP-diglyceride and provide strong evidence for the utilization of both liponucleotides by CDP-diglyceride hydrolase.
Hydrolase Mutants Are Defective in Cytidylylution-We recently reported that a partially purified preparation of CDPdiglyceride hydrolase catalyzed the formation of CDP from CDP-diglyceride and inorganic phosphate (8). Since homogeneous CDP-diglyceride hydrolase was not employed, it was possible that the cytidylyltransferase activity was due to a different enzyme. To demonstrate that both the hydrolase and the transferase activities are encoded by the cdh locus, we assayed CDP synthesis in extracts of hydrolase mutants and clones. Analogous to CDP-diglyceride hydrolysis, the formation of CDP was absent in the mutant and elevated in the clone (Table VIII). Similar results were obtained when dCDP-diglyceride was used as the nucleotidyl donor (data not shown). Thus, the cdh gene product catalyzes the transfer of CMP or dCMP from the liponucleotide to water or inorganic phosphate.

DISCUSSION
We have isolated E. coli mutants defective in CDP-diglyceride hydrolase. This mutation, designated cdh, maps a t minute 88 and is carried on the ColEl hybrid plasmid (26,27), pLC16-4, as judged by the 6-fold overproduction of hydrolase activity. Using recombinant DNA techniques, further overproduction of the hydrolase should be possible, thereby simplifying its purification to homogeneity. The CDP-diglyceride hydrolase from E. coli was originally reported to specifically hydrolyze riboliponucleotides (6). Subsequently, Rittenhouse et al. discovered a similar pyrophosphatase in brain, except both CDP-and dCDP-diglyceride served as substrates (7). The present finding that dCDPdiglyceride is utilized (though less effectively) by the E. coli enzyme eliminates this discrepancy. In addition to the similarity in substrate specificity, both enzymes have an acidic pH optimum and both are inhibited by AMP-and ADPdiglyceride (6,7). In view of these common characteristics, the possibility that the prokaryotic and eukaryotic hydrolases perform the same cellular function deserves consideration.
The revision of the substrate specificity has important implications for considerations of hydrolase function. The original notion that ribospecific hydrolysis controls the intracellular ratio of CDP-to dCDP-diglyceride (6) is eliminated by the present work. Currently, no E. coli enzyme has been identified which distinguishes between CDP-and dCDPdiglyceride, and the physiological significance of the two liponucleotide forms, if any, remains obscure.
Interestingly, extracts prepared from cdh::TnlOpss-21 double mutants can catalyze CDP-diglyceride hydrolysis at a slow rate. One intriguing possibility is the existence of a second TABLE VI11 CDP-diglyceride-dependent cytidylylation in a cdh mutant and clone Extracts were prepared in 10 mM Mes, pH 6, and assayed for cytidylyltransferase activity in 100 mM Mes, pH 6.0, 1.3% octyl 0-Dglucoside, and 16.6 mM potassium phosphate. The water-soluble reaction products were separated by PEI-cellulose thin layer chromatography using solvent B. Following autoradiography, the spots corresponding to [32P]CMP and [a-32P]CDP were scraped from the plate, and the radioactivity was quantitated by liquid scintillation spectrometry. "Under the transferase assay conditions, there is no detectable hydrolysis of CDP-diglyceride in cdh-extracts. In separate experiments, we have shown that this loss of residual activity is due to the shift from pH 7.5 to 6.0 (data not shown). This observation is consistent with the previous finding that phosphatidylserine synthase is responsible for much of the residual activity in cdh mutants (see Table IV, C).
*These assay conditions are optimal for CDP formation, which appears to occur at the expense of CDP-diglyceride hydrolysis (see Table I  form of the hydrolase, analogous to the two E. coli isozymes of phosphofructokinase (29, 30). Further characterization of CDP-diglyceride hydrolysis in cdhpss-extracts may reveal yet another liponucleotide-specific hydrolase.
The recent discovery of CDP-diglyceride-dependent cytidylylation (8) suggests a biosynthetic role for the hydrolase. Previous enzymological studies from our laboratory showed that the partially purified hydrolase catalyzes the transfer of CMP from CDP-diglyceride to phosphate as well as to a variety of phosphomonoesters (8). In the present study, we have demonstrated that the hydrolase and cytidylyltransferase activities reside in a single enzyme, the product of the cdh gene.
In some respects, the hydrolase is similar to the phospholipases of E. coli, particularly the detergent-resistant phospholipase A of the outer membrane. Both enzymes appear to be nonessential for cell survival (31). In addition, Nishijima et al. have obtained evidence suggesting that the phospholipase A is involved in the formation of acylphosphatidylglycerol (32). Thus, it is possible that these apparently catabolic enzymes actually catalyze the synthesis of cell envelope components.
The accumulation of CDP-and dCDP-diglyceride in cdh mutants establishes that these two liponucleotides are physiological substrates for the hydrolase. The question of whether they are indeed cytidylyl donors in vivo can only be resolved by the elucidation of the "true" CMP acceptor. Currently we are analyzing cdh-and cdh' strains by radiochemical and chromatographic methods. The results of these studies should provide new insight into the nature of the physiological CMP acceptor.