Mutants of Saccharomyces cerevisiae Defective in sn-1,2-Diacylglycerol Cholinephosphotransferase ISOLATION, CHARACTERIZATION, AND CLONING OF THE CPTl GENE*

A colony autoradiographic assay for the sn-1,2-di- acylglycerol cholinephosphotransferase activity in Saccharomyces cerevisiae was developed. Twenty-two mutants defective in cholinephosphotransferase activity were isolated. Genetic analysis revealed that all of these mutations were recessive, and three complementation groups were identified. The cholinephospho- transferase activities in membranes prepared from cptl mutants were reduced 2-10-fold compared to wild-type activity. The cholinephosphotransferase activities of two cptl isolates differed from wild-type activity with respect to their apparent KM for CDP-choline. The residual cholinephosphotransferase activities of cptl isolates were more sensitive to inhibition by CMP than the wild-type activity. The CPTl gene was cloned by genetic complementation of cptl using a yeast genomic library. In strains transformed with the CPTl-bearing plasmid, a &fold overproduction of cholinephosphotransferase activity with wild-type kinetic properties was observed. The CPTl gene was localized to a 1.2-2.4-kilobase region of DNA by transposon Tn5 mutagenesis and deletion mapping. An insertional mutant of the CPTl gene was constructed and introduced into the chromosome by integrative transformation. The resulting cpt insertional mutant fell into the cptl complementation pmol unlabeled ethanolamine and the was purified by ion exchange chromatography above for choline phosphate. After lyophilization, the dissolved in 250 p1 of 1 M NaOH and with 15 rl of benzyloxycarbonyl chloride (100 pmol) for 30 at 0 "C (25) to convert the free amine to the benzyl carbamate. The ethanolamine phosphate applied to a 5-ml Bio-Rad AG 50W-X4 H+ column, and the fully protonated product was in five successive 1-ml water washes. After lyophiliza-tion, the residue was dissolved in 250 p1 of anhydrous N,N-dimethyl- formamide and reacted with 21 mg (30 rmol) of CMP-morpholidate for 12 h at 70 "C (26) to form CDP-ethanolamine benzyl carbamate. The product was deprotected by catalytic transfer hydrogenation in the presence of 0.5 M ammonium formate and 25 mg of 10% pallad- ium/carbon catalyst for 5 min at room temperature (27). The reaction mixture was brought to pH 9 with concentrated ammonium hydroxide and purified by ion exchange chromatography as described above for CDP-choline. After lyophilization, the purified CDP-ethanolamine was dissolved in water, and its concentration was determined by phosphate analysis The total radiochemical yield was lo%, and the purity was as by paper chromatography in endogenous diacylglycerol supported an activity representing 25% of the saturated activity, as previously observed (31). Diacylglycerol ethanolaminephosphotransferase activity was assayed in the same reaction mixture except that 0.25 mM [32P]CDP-ethanolamine was used. When the radioactive chloroform-soluble products were sub- jected to thin layer chromatography with 65:25:5 chloroform/metha-nol/acetic acid (solvent system 111) as developing solvent, 92% of the radioactivity co-migrated with authentic phosphatidylethanolamine and co-migrated with authentic phosphatidylcholine. There was no detectable mono- or dimethylated phosphatidylethanolamine. Both assays were linear with time and protein in the range employed.

A colony autoradiographic assay for the sn-1,2-diacylglycerol cholinephosphotransferase activity in Saccharomyces cerevisiae was developed. Twenty-two mutants defective in cholinephosphotransferase activity were isolated. Genetic analysis revealed that all of these mutations were recessive, and three complementation groups were identified. The cholinephosphotransferase activities in membranes prepared from cptl mutants were reduced 2-10-fold compared to wild-type activity. The cholinephosphotransferase activities of two cptl isolates differed from wild-type activity with respect to their apparent KM for CDPcholine. The residual cholinephosphotransferase activities of cptl isolates were more sensitive to inhibition by CMP than the wild-type activity. The CPTl gene was cloned by genetic complementation of cptl using a yeast genomic library. In strains transformed with the CPTl-bearing plasmid, a &fold overproduction of cholinephosphotransferase activity with wild-type kinetic properties was observed. The CPTl gene was localized to a 1.2-2.4-kilobase region of DNA by transposon Tn5 mutagenesis and deletion mapping. An insertional mutant of the CPTl gene was constructed and introduced into the chromosome by integrative transformation. The resulting cpt insertional mutant fell into the cptl complementation group. The cholinephosphotransferase activity in membranes prepared from the cptl insertional mutant was reduced 5fold and exhibited CMP sensitivity. The sn-l,a-diacylglycerol ethanolaminephosphotransferase activities in membranes from all of the cptl isolates including the insertional mutant were normal. The data indicate that the cloned CPTl gene represents the yeast cholinephosphotransferae structural gene, that the yeast choline-and ethanolaminephosphotransferase activities are encoded by different genes, and that the CPTl gene is nonessential for growth.
Phosphatidylcholine is the predominant membrane phospholipid in eukaryotic cells (1). Two major routes exist for the biosynthesis of phosphatidylcholine. In the Kennedy pathway (2), phosphatidylcholine is produced from sn-1,2diacylglycerols by the action of diacylglycerol cholinephosphotransferase (EC 2.7.8.2) which uses CDP-choline as substrate; phosphatidylethanolamine is analogously synthesized by the action of diacylglycerol ethanolaminephosphotransferase which uses CDP-ethanolamine as substrate (1). Alternatively, phosphatidylcholine is formed by sequential methyla-* This work was supported by Grant GM20015 from the National Institute of General Medical Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tion of phosphatidylethanolamine (1) which is derived from the Kennedy pathway and, in yeast, from the decarboxylation of phosphatidylserine (3). Thus, phosphatidylcholine synthesis is intimately related to the biosynthesis of other phospholipid classes and neutral acylglycerols, suggesting a requirement for complex regulation.
Obtaining an understanding of the molecular mechanisms regulating phosphatidylcholine synthesis in mammalian systems has been impeded by the difficulties inherent to the study of the integral membrane proteins involved. To overcome these difficulties and to gain fundamental insight into the regulation of phosphatidylcholine synthesis, we have chosen Saccharomyces cerevisiae as a model eukaryotic system in which to develop the molecular tools required for detailed analysis of diacylglycerol cholinephosphotransferase. This system offers the advantages of combined genetic and biochemical approaches. Recent progress in the study of other s. cerevisiae lipid biosynthetic enzymes (reviewed in Ref. 3) and previous work from our laboratory (4) have demonstrated the value of such an approach.
The pathways of phosphatidylcholine synthesis in yeast are similar to those in higher eukaryotes ( 5 ) . A significant body of information regarding the genetics, biochemistry, and complex regulation of the phosphatidylethanolamine methylation pathway of phosphatidylcholine synthesis has emerged in recent years (reviewed in Ref. 3). However, little is known about the role and regulation of the Kennedy pathway and its integration with the methylation pathway. A detailed knowledge of the structure, function, and regulation of the enzymes of the Kennedy pathway and their genes should prove instrumental in addressing these broader issues.
The present work reports the development of a genetic and biochemical system for the detailed analysis of diacylglycerol cholinephosphotransferase. Mutants defective in cholinephosphotransferase activity (cpt mutants) were isolated by a colony autoradiographic assay. These mutants. were genetically characterized. The CPTl gene has been cloned by genetic complementation. Enzymological and genetic evidence implicating the CPTl gene as the cholinephosphotransferase structural gene was obtained.

EXPERIMENTAL PROCEDURES
M~t e r i a l s -~~P~, [methyl-3H]choline chloride, [Y-~*P]ATP, Aquasol-2, and EN3HANCE fluorographic enhancer were obtained from New England Nuclear. Amino acids, antibiotics, choline kinase (grade II), phosphorylcholine chloride, CMP, ethyl methanesulfonate, CMPmorpholidate, ethanolamine, and bovine serum albumin (essentially fatty acid-free) were purchased from Sigma. Thionyl chloride, benzyloxycarbonyl chloride, 10% palladium on activated charcoal, and N,N-dimethylformamide (gold label) were from Aldrich. sn-1,2-Dioleoylglycerol was prepared by phospholipase C digestion of L-adioleoylphosphatidylcholine (6) obtained from Avanti Polar Lipids, Inc. Choline chloride was purchased from Eastman Kodak. ATP was 3909 3910 Yeast Mutants in Diacylglycerol Cholinephosphotransferase from Pharmacia P-L Biochemicals. Enzymes used in generating ATP were from Boehringer Mannheim. Materials for growth media were from Difco. Restriction enzymes, T4 DNA ligase, and cloning vector pUC18 were purchased from Bethesda Research Laboratories. All other reagents were the highest quality commercially available.
Where appropriate, canavanine sulfate was added at 80 pg/ml. Agar Escherichia coli strain HBlOl (7), obtained from Dr. P. Modrich (Duke University Medical Center), was used as a recipient for transformation and in the isolation of plasmid DNA. Strain LE392 (supE44, supF58, X-) (7) was from our laboratory stock and employed to prepare and titer bacteriophage X467. Strain R477 (RP477) (leuB6, Suo, is) (8) was from Dr. C. Raetz (University of Wisconsin-Madison) and used as a host for transposon Tn5 mutagenesis and for the xenogenetic selection of the yeast LEU2 gene. Bacteria were grown on LB or M9 medium prepared as described by Maniatis et al. (7). Where appropriate, ampicillin was included at 50 pg/ml and kanamycin at 20 pg/ml. Amino acids were provided at 50 pg/ml. Agar plates included 1.5% Difco agar.
The yeast genetic methods employed were adapted from standard protocols (9). DBY746 was mutagenized to 50% survival with ethyl methanesulfonate (lo), plated on YPD agar at 500-1000 colonies/ plate, and grown at 30 "C for screening by the colony autoradiographic assay described below.
Tn5 Mutagenesis-Bacteriophage X467 (b221, ren::Tn5, cI857, Oam29, PanBO) (11, 12) was obtained from Dr. R. Webster (Duke University Medical Center) and used as a source of transposon Tn5. It was propagated and titered as described by Miller (13). Plasmid pRHl was mutagenized with Tn5 according to the protocol of de-Bruijn and Lupski (12). E. coli strain R477 was transformed with pRHl and infected with X467 at a multiplicity of infection of 3, and transposition events were selected on LB plates containing ampicillin and kanamycin. The colonies from 10,000 independent transpositions were pooled, plasmid DNA was isolated, and E. coli strain HBlOl was transformed with the plasmid pool, selecting for ampicillin and kanamycin resistance. The insertions were mapped to low resolution by determining which EcoRI fragment of pRHl contained Tn5. Insertions within the insert region of pRHl were mapped to high resolution by determining the distance from the XhoI sites flanking Tn5 to the EcoRI sites of pRH1, with ambiguities being resolved by mapping the XhoI sites relative to a second restriction site outside of the EcoRI fragment containing Tn5.
tor YEpl3 (14-16) was provided by Dr. K. Nasmyth (Medical Re-Plasmids-A yeast genomic library constructed in the shuttle vecsearch Council Laboratories, Cambridge, England). Plasmid pUC18 (17) was used as a cloning vector. Plasmid pRH104 was constructed from pRHl::Tn5-3 ( Fig. 6) and pUC18. A complete EcoRVIKpnI digest of pRHl::Tn5-3 was mixed with pUC18 that had been digested first with SmaI and then with KpnI. The mixture was ligated and used to transform HBlOl to ampicillin and kanamycin resistance. One recombinant plasmid contained the desired 11.3-kb' EcoRV/ KpnI fragment joined at its KpnI end to a 0.5-kb EcoRVlKpnI fragment derived from the YEpl3 portion of pRHl::Tn5-3; the EcoRV ends of this concatameric fragment were inserted into the SmaI site of pUC18, indicating that the KpnI site in the multiple cloning site of pUC18 had resisted cleavage. XhoI sites within the Tn5 portion of pRH104 were then used to insert a 2.1-kb XhoI/SalI fragment derived from YEpl3 and carrying the yeast selectable marker LEU2. The ligation products of a mixture of pRH104 digested with XhoI and YEpl3 digested with XhoI and Sal1 were transformed into R477, selecting for ampicillin resistance. Transformants were scored for kanamycin sensitivity and leucine prototrophy to detect the insertion of the yeast LEU2 gene (by complementation of the E. coli leuB mutation (15)) into Tn5.
The resulting plasmid was designated pRH105 (see Fig. 7). Other plasmids isolated or constructed in this work are described in the text.
formed essentially as described by Beggs (18) with modifications described by Sherman et a1 (9). For transformation with the gene library, 1 pg of DNA was incubated with each 0.1-ml aliquot of spheroplasts. Rather than using regeneration agar, one-third of each aliquot was directly plated onto a minimal plate containing 1.0 M sorbitol and lacking leucine. Approximately 1000 transformants/plate were obtained. For integrative transformation, 2 pg of DNA was used per plate. Transformation of bacterial cells was according to the method of Kushner (19). Yeast plasmid DNA was isolated from 10 ml of cells grown on selective minimal medium by the protocol in Ref. 9. Plasmid DNA from bacteria was extracted by sodium dodecyl sulfate lysis of 500 ml of chloramphenicol-amplified cells (7) and purified by two successive cesium chloride/ethidium bromide equilibrium centrifugations. Alternatively, small quantities of plasmid DNA from bacteria were prepared by a rapid alkaline lysis procedure (7). Restriction endonuclease digestion of plasmid DNA, agarose electrophoresis, and recombinant DNA manipulations used in plasmid construction were performed using standard methods (7).
Isolation of Membranes-Yeast cultures (200 ml) grown on selective minimal media with supplemental amino acids to an Am of 1.0-2.0 were harvested at 1,000 X g for 10 min, washed once with 100 ml of water, and washed once with 25 ml of 20% glycerol, 50 mM MOPS/ NaOH (pH 7.5), and 1 mM EDTA (GME buffer). The cell pellet was resuspended in a total volume of 1.0 ml with GME buffer and disrupted by glass beads in a mini-beadbeater vial as previously described (4). The homogenate was transferred to a 1.5-ml Eppendorf vial and centrifuged at 16,000 X g for 10 min to remove unbroken cells. The supernatant was then diluted to 10 ml with GME buffer and centrifuged at 100,000 X g for 1 h. The pellet was resuspended with the aid of a Teflon homogenizer in 0.5 ml of GME buffer and frozen in aliquots at -70 "C. The preparation was maintained at 0-4 "C throughout all steps of the membrane isolation. Membrane protein was estimated by the method of Peterson (20) using bovine serum albumin as standard.
Synthesis of Radiolabeled CDP-ch~line-[~~P]-and Imeth~l-~H] phosphorylcholine were prepared enzymatically using choline kinase from S. cereuisiae as reported by Vance (21). The reaction mixture for the tritium derivative consisted of 0.22 mM [meth~l-~HIcholine chloride (5 mCi, 80 Ci/mmol), 10 mM MgC12, 36 mM Tris-HC1 (pH 8.0), 10 mM ATP, and 0.1 unit of choline kinase in a total volume of 280 pl. The 32P-labeled derivative was similarly prepared except that the reaction mixture contained 0.36 mM [ T -~~P ] A T P (1 mCi, 1 Ci/ mmol) and 1.1 mM choline chloride. The specific activity was reduced by addition of 15 pmol of unlabeled phosphorylcholine; the pH was adjusted to 9 with concentrated ammonium hydroxide; and the material was converted to its zwitterionic salt form by adsorbing it to a 5-ml column of Dowex 1-X2 formate washing with 5 volumes of water, and eluting with 3 volumes of 50 mM formic acid. The eluent was lyophilized and resuspended in 1 ml of water, and its concentration was determined by phosphate analysis according to the method of Ames and Dubin (22). To convert radiolabeled phosphorylcholine to CDP-choline, a modification of the direct condensation with cytidine 5'-monophosphate using the Vilsmeier-Haack reagent (23) was employed. To 10 pmol of phosphorylcholine dried thoroughly in uucuo was added 40 pl of freshly prepared 1.0 M thionyl chloride (40 pmol) in anhydrous N,N-dimethylformamide. After 5 min at 25 "C, 3.4 mg (10 pmol) of CMP was added. The reaction mixture was vortexed until clear and held at 25 "C for 1 h after which 200 p l of water was added. The quenched reaction mixture was held for 1 h at 25 "C, and then the pH was adjusted to 9 with concentrated ammonium hydroxide. The material was submitted to a 5-ml column of Dowex 1-X2 formate and washed with 5 volumes of water to remove the side product dicholine pyrophosphate and an unidentified product. The column was then eluted with a 0-50 mM gradient of formic acid in a total volume of 100 ml. The order of elution was: unreacted phosphorylcholine, CDP-choline, followed by unreacted cytidine monophosphate. The fractions containing CDP-choline were pooled, lyophilized, and dissolved in water. The concentration was determined by its ultraviolet absorbance at 280 nm assuming an extinction coefficient of 12.8 mM" cm" (23). The total radiochemical yield was 20%, and the radiochemical purity was >99% as assessed by paper chromatography on Whatman No. 3M paper using 3:2:1 isoamyl alcohol/formic acid/water (solvent system I) (23) as the developing solvent.
Synthesis of Radiolabeled CDP-ethanolamine-Ethanolamine ["PI phosphate was prepared from 25 mCi of 32Pi by including yeast choline kinase (0.1 unit) and ethanolamine (added as 100 mM ethanolamine hydrochloride (pH 8) to a final concentration of 2 mM) in a standard enzymatic [32P]ATP-generating system (24). After incubation at 37 "C for 1 h, 50 pmol of unlabeled ethanolamine phosphate was added, and the product was purified by ion exchange chromatography as described above for choline phosphate. After lyophilization, the ethan~lamine[~~P]phospbate was dissolved in 250 p1 of 1 M NaOH and reacted with 15 r l of benzyloxycarbonyl chloride (100 pmol) for 30 min at 0 "C (25) to convert the free amine to the benzyl carbamate. The protected ethanolamine phosphate was then applied to a 5-ml Bio-Rad AG 50W-X4 H+ column, and the fully protonated product was collected in five successive 1-ml water washes. After lyophilization, the residue was dissolved in 250 p1 of anhydrous N,N-dimethylformamide and reacted with 21 mg (30 rmol) of CMP-morpholidate for 12 h at 70 "C (26) to form CDP-ethanolamine benzyl carbamate. The product was deprotected by catalytic transfer hydrogenation in the presence of 0.5 M ammonium formate and 25 mg of 10% palladium/carbon catalyst for 5 min at room temperature (27). The reaction mixture was brought to pH 9 with concentrated ammonium hydroxide and purified by ion exchange chromatography as described above for CDP-choline. After lyophilization, the purified CDP-ethanolamine was dissolved in water, and its concentration was determined by phosphate analysis (22). The total radiochemical yield was lo%, and the radiochemical purity was >99% as assessed by paper chromatography in solvent system I.
Diacylglycerol Cholinephosphotransferase Colony Autoradiographic Assay-The colony autoradiographic assay (28, 29) described was modified from a similar assay for the yeast sn-glycerol-3-phosphate acyltransferase activity developed in our laboratory (4). Yeast colonies on agar plates were transferred to a Whatman No. 42 filter paper circle. The paper and adhered colonies were frozen at -70 "C for 1 h and then air-dried. This freeze-thaw-dry cycle was an essential step and served to permeabilize the colonies to substrate. The papers were then incubated for 1 h in Petri dishes containing 1 ml of reaction solution which consisted of 50 mM MOPS/NaOH (pH 7.5), 20 mM MgC12, 1.5 mg/ml bovine serum albumin, and 100 p~ radiolabeled CDP-choline.
[3H]CDP-choline was used at 25 mCi/mmol, and ["PI CDP-choline was used at 2-20 mCi/mmol. After a 1-h incubation at ambient temperature, the paper was transferred to a new Petri dish containing 1 ml of 10% trichloroacetic acid to quench the reaction and precipitate the phosphatidylcholine product. After 30 min in the 10% trichloroacetic acid, the paper was washed with 6 X 25-ml portions of ice-cold 2% trichloroacetic acid in a Buchner funnel and air-dried. The paper was then applied to Kodak XAR-5 film, with fluorographic enhancer being applied prior to exposure in the case of tritiated label. The film was exposed for 2-4 weeks (3H) or 4-12 h (32P) and developed. After autoradiography, the paper was stained for 5 min with 0.4% Coomassie Blue in ethanol and destained with ethanol for visualization and comparison to the autoradiograms. For analysis of radioactive products, cells on filter papers were extracted overnight at 4 "C in 2 ml of 1:l chloroform/methanol, followed by further extraction for 1 h at room temperature after adding 0.8 ml of 1 h ' hydrochloric acid and 2 ml of methanol. Phases were separated by adding 1 ml each of chloroform and 1 N hydrochloric acid. The organic phase was washed twice with 2 ml of 1 N hydrochloric acid. Portions of the chloroform phase were then analyzed by liquid scintillation counting or thin layer chromatography and scintillation counting.
Enzyme Assays-Diacylglycerol cholinephosphotransferase activity present in membrane preparations was determined as the amount of 3H incorporated from [3H]CDP-choline into phosphatidylcholine. The assay employed was a modification of that developed for the rat liver microsomal cholinephosphotransferase by Coleman and Bell (30) and was similar to that reported by Percy et al. (31) for the yeast enzyme. The assay mixture contained 50 mM MOPS/NaOH (pH 7.5), 20 mM MgClz, 1.5 mg/ml bovine serum albumin, 0.1 mM dioleoylglycerol, and 0.25 mM [3H]CDP-choline in a final volume of 0.2 ml. The assay was initiated by the successive addition of membranes in isolation buffer and 10 $1 of 2 mM dioleoylglycerol in absolute ethanol (5% (v/v) final concentration). After 10 min at 20 'C, 0.6 ml of 1% perchloric acid was added to terminate the reaction.
[3H]Phosphatidylcholine was extracted by the method of Bligh and Dyer (32). After washing the chloroform phase twice with 2 ml of 1% perchloric acid, 1 ml of the organic phase was placed in a scintillation vial, dried by heating, and counted in 4 ml of Aquasol-2 in an LKB 1217 Rackbeta liquid scintillation counter. Greater than 99% of the chloroformsoluble radioactivity co-migrated with authentic phosphatidylcholine upon thin layer chromatography on Silica Gel 60 plates (Merck) using 25:15:4:2 chloroform/methanol/water/acetic acid (solvent system 11) as developing solvent. The assay was optimized with respect to pH, MgCl, concentration, and amount of bovine serum albumin. Whereas the assay was saturable with respect to exogenous dioleoylglycerol, endogenous diacylglycerol supported an activity representing 25% of the saturated activity, as previously observed (31). Diacylglycerol ethanolaminephosphotransferase activity was assayed in the same reaction mixture except that 0.25 mM [32P]CDP-ethanolamine was used. When the radioactive chloroform-soluble products were subjected to thin layer chromatography with 65:25:5 chloroform/methanol/acetic acid (solvent system 111) as developing solvent, 92% of the radioactivity co-migrated with authentic phosphatidylethanolamine and 8% co-migrated with authentic phosphatidylcholine. There was no detectable mono-or dimethylated phosphatidylethanolamine. Both assays were linear with time and protein in the range employed.

RESULTS AND DISCUSSION
Filter Paper Assay and Mutant Selection-A colony autoradiographic assay (4, 28, 29) was devised to facilitate the rapid screening of a large number of mutagenized cells for defects in cholinephosphotransferase. The assay was based on the incorporation of radiolabeled CDP-choline into phosphatidylcholine in permeabilized yeast colonies on filter papers using endogenous diacylglycerol as the phosphorylcholine acceptor. The radioactive products of the in situ reaction were characterized by extraction of the papers followed by thin layer chromatography (see "Experimental Procedures"). Greater than 97% of the radioactivity remaining after washing was chloroform-extractable, and greater than 99% of this comigrated with authentic phosphatidylcholine during thin layer chromatography. The dependences of the assay on time and substrate concentration were also investigated. As shown in Fig. l A , incorporation into phosphatidylcholine was linear up to 90 min, demonstrating that endogenous diacylglycerol was not limiting in the assay. The CDP-choline concentration required for 50% maximal activity (Fig. 1B) was similar t o the apparent KM for CDP-choline measured in membrane preparations (60 ~L M , see Table 11). Thus, delivery of the radioactive substrate to the enzyme was efficient.

Yeast Mutants in
Diacylglycerol Cholinephosphotransferase screening in order to maximize incorporation while retaining sensitivity to mutants having an altered KM for CDP-choline.
Taken together, these experiments establish a colony autoradiographic assay for yeast cholinephosphotransferase. The colony autoradiographic assay was used to screen 15,000 colonies of DBY746 mutagenized to 50% survival with ethyl methanesulfonic acid. By comparing the autoradiographic patterns to the Coomassie Blue-stained filters, 22 isolates were identified which had reproducibly decreased in citu cholinephosphotransferase activities. Fig. 2 shows the ident.ification of one of these mutants, HJ110, as an example of the screening method. The extent of reduction of CDPcholine incorporation ranged from slight to nearly complete (see mutants shown in Fig. 3). Since the relative differences observed between various mutant strains were highly reproducible in separate experiments, a qualitative scale reflecting the colony autoradiographic phenotype was devised (see Table   I).
Complementation Analysis-Genetic complementation analysis of the mutants was undertaken as a first step in their characterization. The colony autoradiographic assay phenotype proved suitable for this purpose (see Fig. 3). The cholinephosphotransferase activities of diploids formed in crosses of each mutant to the wild-type strain AH22 were normal; thus, all of the cpt mutations were recessive. The mutant allele present in HJllO was designated cptl, and a strain bearing this allele and which was suitable for genetic crosses to the remaining mutants, H J l l l (a leu2-3 leu2-112 his4-519 canl cptl), was constructed by standard genetic methods. Seventeen of 21 of the remaining mutants failed to provide phenotypic complementation when crossed t o H J l l l and were thus designated cptl. Four mutants fully complemented the cptl mutation of HJ111. The mutant allele of one of these, HJ214, was designated cpt2, and an additional mater strain bearing it, HJ213 (a leu2-3 leu2-112 canl his4-519 cpt2), was constructed. The complementation data for the remaining mutants as well as HJllO against cpt2 is shown in Fig. 3. All of the remaining mutants complemented the cpt2 allele of HJ213. The mutant allele of HJ128 was designated cpt3. Since the remaining mutants, HJ120 and HJ244, were phenotypically weak, further complementation grouping was not pursued. Random spores of H J l l l X HJ214 and H J l l l X HJ128 exhibited wild-type cholinephosphotransferase activity by filter paper assay at a frequency sufficient to preclude intragenic complementation within the cptl locus (data not shown).
Flc. 2. Isolation of H J l l O ( c p t l ) by colony autoradiographic assay. An autoradiogram (right) of a group of mutagenized colonies of DBY746 assayed in situ for cholinephosphotransferase activity (see "Experimental Procedures") is shown next to the Coomassie Blue-stained filter paper it represents (left). In this field, a mutant colony (HJ110) is clearly identified (arrowhead) which lacks associated radioactivity. presented. The array shown is an autoradiogram of a colony autoradiographic assay in which each column displays the activity of a mutant (identified at bottom), a cross between mutant and HJ213 (cpt2), and a cross between mutant and AH22 (*, not shown here). Controls (DBY746 and HJ213) are included in each column. Since full complementation is seen for each mutant in crosses with both cpt2 (HJ213) and wild-type (AH22) strains, all mutants are recessive and in a complementation group distinct from cpt2. HJllO is defined as cptl strain, and HJ128 is designated cpt3. HJ120 and HJ244 are phenotypically weak and are either cpt3 or new complementation groups.

TABLE I Cholinephosphotransferase activities of mutant and
plasmid-bearing strains Membranes from each of the strains were assayed for cholinephosphotransferase activity as described under "Experimental Procedures" at 250 ~L M CDP-choline. The genetic designations were assigned as described in the text. The filter paper phenotype is a qualitative scale based on examining films such as in Fig. 3. The scale ranges from 5+ (wild-type) down to 1+ (HJ110).

Kinetic characterization of the cholinephosphotransferase activities of mutant and plasmid-bearing strains
Membranes prepared from each strain were assayed for cholinephosphotransferase activity as a function of CDP-choline concentration as described under "Experimental Procedures." The kinetic constants were obtained from double-reciprocal plots as illustrated in Fig. 4 under these conditions. The qualitative scale reflecting the extent of the filter paper phenotype correlated well with the cholinephosphotransferase activities assessed in membranes. Mutants exhibiting a strong phenotypic defect had 5-7-fold reductions in in vitro cholinephosphotransferase activities, whereas more moderately affected cptl mutants were associated with intermediate activities. In contrast, the cpt2 and cpt3 isolates exhibited near normal cholinephosphotransferase activities under these conditions despite their strong defects in the colony autoradiographic assay. This initial in vitro screening suggested that the cptl complementation group most likely represented the desired mutants in the cholinephosphotransferase structural gene. Therefore, the cpt2 and cpt3 isolates were not pursued further.* The finding of additional complementation groups strongly affected in the cholinephosphotransferase autoradiographic assay may have important implications in future studies of regulation.
Enzyrnological Analysis-Additional characterization of the cholinephosphotransferase activity of several cptl mutants was performed. Kinetic constants determined from Lineweaver-Burk plots of the CDP-choline dependences are shown in Table 11. The 8-fold reduction in V,,, and normal KM for CDP-choline shown for HJ135 was typical of other phenotypically strong cptl mutants (data not shown). In contrast, HJ125 showed a normal V,,, but a 4.5-fold increase in KM for CDP-choline (Fig. 4). Similarly, HJ115 exhibited a 2-fold higher KM value for CDP-choline than wild-type and a slightly decreased VmaX. These alterations in the kinetic properties of cholinephosphotransferase in cptl mutants strongly suggest that the cptl locus represents the cholinephosphotransferase structural gene. As discussed under "Experimental Procedures,'' the cholinephosphotransferase assay employed was only partially dependent on exogenous diacylglycerol. Kinetic interpretation of diacylglycerol dependences was therefore difficult. However, the yeast cholinephosphotransferase could be assayed in Triton X-100-mixed micelles where a complete dependence on exogenous diacylglycerol was o b s e~e d .~ T h e The possibility that the autoradiographic phenotype of cpt2 and cpt3 mutants resulted from altered levels of endogenous diacylglycerol was considered. When the cholinephosphotransferase activities of membranes isolated from cpt2 and cpt3 strains were assayed using endogenous diacylglycerols as substrates, normal cholinephosphotransferase activities were obtained as compared to control. Furthermore, the incorporation of radiolabeled CDP-choline into phosphatidylcholine in the colony autoradiographic assay was proportional with time using colonies of the cpt2 and cpt3 strains. Phosphatidicacid phosphatase activities were unaffected in membranes prepared from cpt2 and cpt3 strains relative to wild-type, precluding a block in the production of diacylglycerol via this enzyme. Taken together, these data suggest that the autoradiographic defect in cpt2 and cpt3 strains is not a consequence of reduced diacylglycerol levels.
R. H. Hjelmstad and R. M. Bell, unpublished data. as a function of CDPcholine concentration. Linear regression analysis of the data was used to construct the lines as shown and to determine the kinetic constants reported in Table 11. A, DBY746/YEp13; A, HJ125/YEp13. Membranes from each strain were assayed for cholinephosphotransferase activity as described under "Experimental Procedures" with the addition of the indicated concentrations of GMP. The activities are expressed as percent remaining relative to no addition. A, HJllO/YEpl3; A, DBY746/YEp13; 0, HJ110/ pRH1. apparent KM for sn-1,2-dioleoylglycerol determined for each of the strains in Table I1 using this assay was similar to wildtype (data not shown).
The yeast ethanolaminephosphotransferase activity has been reported to be more strongly inhibited by CMP than is the cholinephosphotransferase activity (31). Therefore, the effect of CMP on the residual cholinephosphotransferase activity of cptl mutants was measured to assess the possibility that the residual activity reflects a cross-specificity of ethanolaminephosphotransferase for CDP-choline. As shown in Fig. 5, the residual cholinephosphotransferase activity of HJllO was inhibited 75% in the presence of 1 mM CMP, whereas that of DBY746 was only inhibited 30%. Curves nearly identical to the inhibition profile of HJllO were obtained for the other strong cptl mutants (HJ112, HJ135, and HJ137) (data not shown). This property could reflect an alteration of the activity of cholinephosphotransferase as a consequence of mutation within its structural gene. Since the cholinephosphotransferase activities of four independent cptl mutants were found to possess this property, this explanation was considered unlikely. Alternatively, the residual cholinephosphotransferase activity could reflect the activity of a Yeast Mutants in Diacylglycerol Cholinephosphotransferase second enzyme which exhibits the observed inhibition by CMP. Ethanolaminephosphotransferase or a previously unidentified cholinephosphotransferase is a likely candidate (see "Concluding Discussion").
Cloning of the CPTl Gene-A yeast genomic library constructed in the vector YEpl3 was transformed into HJ110, and the resulting transformants were screened for complementation of cptl by the colony autoradiographic assay. Approximately 18,000 transformants were screened, and six plasmids were selected which reproducibly conferred increased cholinephosphotransferase activity relative to HJllO/YEp13.
Plasmid DNA was isolated from these strains and transformed into E. coli strain HBlOl from which quantities of plasmid sufficient for physical analysis were purified. Restriction endonuclease mapping revealed that four of the plasmids contained a common 5.0-kb region of insert DNA. A representative of this group, pRH1, was chosen for further characterization. The insert DNA of pRHl is shown in Fig. 6. The ability of pRHl to complement the cptl defect in a plasmidtransmissible fashion was shown by colony autoradiography. Co-segregation of phenotypic correction with the LEU2 gene carried on the YEpl3 portion of pRHl was demonstrated by growing HJllO/pRHl for 10 generations on YPD, plating single colonies onto YPD plates, replica printing to synthetic minimal plates lacking leucine, and performing cholinephosphotransferase colony autoradiography on the master print. In this experiment, 80% of the colonies had simultaneously become auxotrophic for leucine and defective in the filter paper assay; there was an exact co-segregation of the two phenotypes.
Enzymological characterization of strains transformed with pRHl was undertaken. As shown in Table I, the presence of pRHl in HJllO fully restored the cholinephosphotransferase activity and led to an approximately 30-and &fold overproduction of activity in mutant and wild-type backgrounds, respectively. This degree of overproduction is typical of other membrane-bound enzymes whose structural genes have been cloned into YEpl3 (33, 34) and is within the range expected based on the copy number of YEpl3 (35, 36). The overproduced cholinephosphotransferase activity expressed in strains transformed with pRHl was wild-type with respect to sub- The 6.9-kb insert region of pRHl (solid line) carrying the CPT gene is shown within the vector YEpl3 sequences (cross-hatched line). The insert lies within the BarnHI site of the tetracycline resistance gene of the pBR322 portion of YEp13; the ampicillin resistance gene is to the left. The indicated restriction endonuclease recognition sites were determined by standard mapping techniques relative to known restriction sites in YEpl3. As described in the text, 20 transposon Tn5 insertions into the insert region of pRHl were aligned to the physical map of pRH1. The capacity of each to provide functional complementation of the cptl defect of HJllO was assessed by cholinephosphotransferase colony autoradiography as described under "Experimental Procedures." The location and functional activity of the Tn5 insertions are shown: V, active; 7 , inactive. The location and boundaries of the CPT gene are also shown: -.-, 2.4-kb upper limit based on active insertions; ---, 1.2-kb lower limit based on inactive insertions and deletion mapping data (see text). *, location of the Tn5 insertion (pRHl::Tn5-3) used in the construction of pRH105 (Fig.   8).
strate dependences, thermal inactivation (data not shown), and CMP inhibition profile. As seen in Table 11, pRHl completely corrected the KM defect present in HJ125. The overexpressed cholinephosphotransferase activity in HJ110/ pRHl was wild-type with respect to inhibition by CMP (Fig.  5). The overproduction of cholinephosphotransferase activity directed by the cloned C P T gene which is indistinguishable from wild-type activity in several genetic backgrounds is strong evidence that the cloned gene represents the cholinephosphotransferase structural gene.
Sublocalization of the CPT Gene-In order to localize the coding region of the presumptive CPT gene within the pRHl insert DNA, pRHl was mutagenized with transposon Tn5 as described under "Experimental Procedures." Twenty Tn5 insertions were found within the insert region of pRH1. The locations of these were precisely determined and are shown in Fig. 6 (triangles).
Each of the 20 pRHl::Tn5 insertions was transformed into HJ110, and their functional ability in complementing the cholinephosphotransferase activity defect was evaluated by the colony autoradiographic assay. Four contiguous Tn5 insertions (Fig. 6, filled triangles) failed to provide phenotypic complementation, setting a lower limit to the size of the CPT gene of 0.8 kb. The Tn5 insertions which were fully active and thus within nonessential DNA set the upper limit of size at 2.4 kb. Additional localization data were obtained by deletion mapping. A complete HindIII digest of pRHl was recircularized to construct pRH103; this construction deleted the region from the leftmost HindIII site within the insert to the pBR322 Hind111 site 0.35 kb into the YEpl3 sequences to the right of the insert border. Upon transformation into HJ110, pRH103 failed to complement the cholinephosphotransferase filter paper assay defect, indicating that the CPT gene extended a t least up to the leftmost HindIII site from the inactive Tn5 insertions (Fig. 6). Thus, the minimal size of the C P T gene was determined to be 1.2 kb.
Construction of a cptl Insertional Mutant-To establish that the CPT gene cloned is the CPTl gene, a one-step gene disruption experiment (37) was performed. For this purpose, an integrative plasmid containing the CPT gene disrupted by a yeast selectable marker was required. Since pRHl contained no restriction sites within the known CPT coding region convenient for the placement of a yeast selectable marker, the disrupted copy of the CPT gene in pRHl::Tn5-3 (Fig. 6, asterisk) was subcloned into pUC18 so that the restriction sites within Tn5 could be used to insert the LEU2 fragment of YEpl3 (see "Experimental Procedures"). The structure of the plasmid constructed, pRH105, is shown in Fig. 7.
The disrupted CPT gene of pRH105 was liberated by digestion with BamHI and PstI and transformed into DBY746, selecting for leucine prototrophy to detect integrative events. Five out of 10 transformants had acquired a strong defect in cholinephosphotransferase activity when screened by the colony autoradiographic assay. The extent of the defects was similar to that of HJllO (see Fig. 8). One of these, designated HJ001, was selected for further study. Both the leu+ and cptphenotypes of HJOOl were mitotically stable on nonselective medium, indicating that the disrupted CPT gene was stably integrated into the chromosome.
Characterization of the cpt Insertional Mutant-Complementation analysis for HJOOl is shown in Fig. 8. The cpt insertional mutant complemented the cpt2 and cpt3 (data not shown) mutations but failed to complement the cptl mutation. Since no four-spore asci were generated upon sporulation of HJOOl X AH22, random spores were analyzed for cholinephosphotransferase activity and growth on leucine. In this was transformed to leucine prototrophy by pRHlO5 but had not acquired a cptl mutation (Table 111). The residual cholinephosphotransferase activity exhibited a wild-type KIM for CDP-choline (data not shown) and, unlike the wild-type activity, was strongly inhibited by CMP (Fig. 9). Since the integrated cptl gene in HJOOl was intentionally disrupted near the middle of the known functional region of the gene, it is assumed that the CPTl gene product would be nonfunctional. The data indicate that the CMP-sensitive residual activity does not represent an altered activity of the CPTl gene product (see "Concluding Discussion"). Thus, the ability of a haploid strain containing the cptl insertional mutation to grow shows that the CPTl gene is nonessential for growth.
EthanolaminephosphotransferaseActivities-To investigate the relationship between the yeast choline-and ethanolaminephosphotransferases, membranes possessing defective and overproduced cholinephosphotransferase activities were assayed for ethanolaminephosphotransferase activity. Tables  8. Complementation analysis of the cptl insertional mutant. The colony autoradiographic assay shows the activity of the cptl insertional mutant (HJ001) and crosses of HJOOl to cpt mater strains. The cpt insertional mutation is recessive to the wild-type gene ( X AH22), fully complements cpt2 ( X HJ213) and cpt3 (data not shown), and fails to complement cptl ( X HJ111). The activities of HJ213, HJ111, and DBY746 were included as controls. Yeast M u t a n t s in Diacylglycerol Cholinephosphotransferase I and IV show that the large reductions in cholinephosphotransferase activity in cptl mutants were associated with only slightly reduced ethanolaminephosphotransferase activities. Ethanolaminephosphotransferase activity decreased 5-15% as compared to the decline in cholinephosphotransferase activity. The converse was also observed. When the cholinephosphotransferase activity was elevated due to the presence of the CPTl gene of pRH1, the ethanolaminephosphotransferase activity increased to an extent approximately 8% of the increase in cholinephosphotransferase activity. Similar results were obtained in all mutant strains bearing pRHl that were examined. The isolation of mutants defective in cholinephosphotransferase but not ethanolaminephosphotransferase establishes genetically that the two enzymatic activities are encoded by different genes. The attenuated response of the ethanolaminephosphotransferase activities to variation in cholinephosphotransferase activities could be interpreted to reflect a specificity of the cholinephosphotransferase for CDP-ethanolamine; thus, cholinephosphotransferase appears to utilize CDP-ethanolamine as substrate at a rate approximately 10-fold less than CDP-choline.
Concluding Discussion-Enzymological characterization of the cptl mutants isolated in this work supports the conclusion that the cptl locus represents the structural gene for yeast cholinephosphotransferase. The cptl complementation group encompassed independent isolates which were defective in cholinephosphotransferase activity to varying extents, including one whose defect was completely attributable to an altered KM for CDP-choline. The overproduction of cholinephosphotransferase activity having wild-type properties in several genetic backgrounds bearing the cloned CPTl gene further supports this view. Since we have demonstrated a genetic link between the cloned CPTl gene and the cptl complementation group, we further conclude that the cloned gene is the cholinephosphotransferase structural gene.
The residual cholinephosphotransferase activity present in strong cptl mutants and the cptl insertional mutant clearly differed from the major wild-type activity in its CMP sensitivity; this strongly suggests the presence of a second cholinephosphotransferase activity in yeast. However, no conclusion regarding the identity of the enzyme(s) responsible for the residual activity is justified. As discussed earlier, the presence of the CMP-sensitive activity in all strong cptl mutants and the cptl insertional mutant argues against it reflecting an altered activity of the CPTl gene product. The second activity may be due to a weak activity of ethanolaminephosphotransferase using CDP-choline as substrate. In view of the evidence presented here that cholinephosphotransferase exhibits weak activity using CDP-ethanolamine, it would not be surprising if a similar cross-specificity is manifested by ethanolaminephosphotransferase which could quantitatively account for the residual cholinephosphotransferase. Ethanolaminephosphotransferase was found to be CMP-sensitive as previously reported (31). However, the residual cholinephosphotransferase of the cptl insertional mutant was somewhat high when compared to that observed in other cptl mutants and was not associated with a proportionately elevated ethanolaminephosphotransferase activity. In this regard, it is noteworthy that the residual activity of the insertional mutant was also surprisingly high given its dramatically reduced activity in the colony autoradiographic assay. Several factors may underlie these observations. Whereas alterations in the predominant cholinephosphotransferase activity were reflected in the colony autoradiographic assay, physical barriers such as subcellular compartmentalization may preclude the assay from detecting the minor residual activity. Moreover, physiological conditions and growth properties may influence the level of the second activity or its expression, explaining differences between activities observed in the colony autoradiographic assay and membranes. Finally, the residual activity or its expression could be regulated by the CPTl gene or its gene product, accounting for differences in its level in various cpt1 mutants, especially the insertional mutant. Thus, we conclude that the residual cholinephosphotransferase activity is due to a crossspecificity of ethanolaminephosphotransferase and/or the presence of a second cholinephosphotransferase. This possibility may bear on our observation that the CPTl gene is nonessential for cell growth. Our current efforts to obtain mutants in ethanolaminephosphotransferase will assist in the resolution of this problem. Our data support and extend previously reported enzymological evidence that the yeast cholinephosphotransferase and ethanolaminephosphotransferase activities are catalyzed by separate enzymes (31). Our demonstration of the genetic distinguishability of the two activities constitutes conclusive evidence for distinct enzymes.
The genetic foundation and molecular tools assembled in this work should greatly facilitate further studies of the structure, function, and regulation of yeast cholinephosphotransferase. We anticipate that future studies of the physiologic significance of cptl mutations and detailed analysis of the CPT gene and gene product will yield fundamental insight into the regulation of phosphatidylcholine biosynthesis.