Macrophages Deficient in CTP:Phosphocholine Cytidylyltransferase-a Are Viable under Normal Culture Conditions but Are Highly Susceptible to Free Cholesterol-induced Death MOLECULAR GENETIC EVIDENCE THAT THE INDUCTION OF PHOSPHATIDYLCHOLINE BIOSYNTHESIS IN FREE CHOLESTEROL-LOADED MACROPHAGES IS AN ADAPTIVE RESPONSE*

Macrophages in atherosclerotic lesions accumulate excess free cholesterol (FC) and phospholipid. Because excess FC is toxic to macrophages, these observations may have relevance to macrophage death and necrosis in atheromata. Previous work by us showed that at early stages of FC loading, when macrophages are still healthy, there is activation of the phosphatidylcholine (PC) biosynthetic enzyme, CTP:phosphocholine cytidylyltransferase (CT), and accumulation of PC mass. We hypothesized that this is an adaptive response, albeit transient, that prevents the FC:PC ratio from reaching a toxic level. To test this hypothesis directly, we created mice with macrophage-targeted disruption of the major CT gene, CTa, using the Cre-lox system. Surprisingly, the number of peritoneal macrophages harvested from CTa-deficient mice and their overall health under normal culture conditions appeared normal. Moreover, CT activity and PC biosynthesis and in vitro CT activity were decreased by 70–90% but were not absent. As a likely explanation of this residual activity, we showed that CTb2, a form of CT that arises from another gene, is induced in CTa-deficient macrophages. To test our hypothesis that increased PC biosynthesis is an adaptive response to FC loading, the viability of wild-type versus CTa-deficient macrophages under control and FC-loading conditions was compared. After 5 h of FC loading, death increased from 0.7% to only 2.0% in wild-type macrophages but from 0.9% to 29.5% in CTa-deficient macrophages. These data offer the first molecular genetic evidence that activation of CTa and induction of PC biosynthesis in FC-loaded macrophages is an adaptive response. Furthermore, the data reveal that CTb2 in macrophages is induced in the absence of CTa and that a low level of residual CT activity, presumably due to CTb2, is enough to keep the cells viable in the peritoneum in vivo and under normal culture conditions. Cholesterol-loaded macrophages, or foam cells, play critical roles in atherogenesis, including the conversion of relatively benign early lesions into advanced lesions that lead to acute vascular occlusion (1, 2). An important event related to macrophage foam cells in advanced atherosclerosis is the death of these cells (3–6), which likely plays an important role in lesional necrosis and perhaps plaque rupture and acute thrombotic vascular occlusion (2, 3, 6). Because advanced lesional macrophages are known to accumulate large amounts of free cholesterol (FC) (7–10), and because excess cellular FC is known to lead to macrophage death (11, 12), FC-induced death in macrophages in advance atherosclerotic lesions may be a critically important event in advanced lesional complications (13, 14). In this context, our laboratory has studied cellular events that occur during FC loading of macrophages, and we discovered that the cells respond initially by an increase in PC biosynthesis and cellular PC mass, which also appears to occur in lesional macrophages in vivo (15, 16). The mechanism of this response is post-translational activation of CTP:phosphocholine cytidylyltransferase (CT; also known as CCT), which catalyzes the conversion of choline-phosphate to CDP-choline (15, 17). We have hypothesized that activation of CT in response to FC loading in macrophages is an adaptive response (12). Cellular FC excess is known to be toxic to cells primarily by altering the physical properties of cellular membranes, leading to dysfunction of integral membrane proteins (18). We therefore reasoned that the induction of PC biosynthesis in FCloaded cells would help keep the FC:phospholipid ratio in cellular membrane from reaching cytotoxic levels. Observations of macrophages after prolonged periods of FC loading, where cell death follows a decline in the PC biosynthetic response, provided evidence in support of this hypothesis, but direct molecular support has been lacking (12). To address this issue, we sought to create a mouse model whose macrophages had deficient CT activity and compromised PC biosynthetic capacity. Our hypothesis predicts that macrophages from such a mouse would be more susceptible to FC* This work was supported by NHLBI Grants HL-54591 (to I. T.) and GM45737 (to S. J.) from the National Institutes of Health, by Cancer Center (CORE) Support Grant CA21765 (to S. J.), and by the American and Lebanese Syrian Associated Charities (to S. J.). 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. i To whom correspondence and reprint requests should be addressed: Dept. of Medicine, Columbia University, 630 West 168th St., New York, NY 10032. Tel.: 212-305-9430; Fax: 212-305-4834; E-mail: iat1@ columbia.edu. 1 The abbreviations used are: FC, free cholesterol; acetyl-LDL, acetylated low-density lipoprotein; CT, CTP:phosphocholine cytidylyltransferase; CTaflox, CTa genomic construct containing flanking loxP sites; DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal bovine serum; LysMCre, Cre recombinase gene in the murine lysozyme locus; PBS, phosphate-buffered saline; PC, phosphatidylcholine; kb, kilobase pairs; RT-PCR, reverse transcriptase-polymerase chain reaction; bp, base pair. THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 275, No. 45, Issue of November 10, pp. 35368–35376, 2000 © 2000 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.

Cholesterol-loaded macrophages, or foam cells, play critical roles in atherogenesis, including the conversion of relatively benign early lesions into advanced lesions that lead to acute vascular occlusion (1,2). An important event related to macrophage foam cells in advanced atherosclerosis is the death of these cells (3)(4)(5)(6), which likely plays an important role in lesional necrosis and perhaps plaque rupture and acute thrombotic vascular occlusion (2,3,6). Because advanced lesional macrophages are known to accumulate large amounts of free cholesterol (FC) 1 (7)(8)(9)(10), and because excess cellular FC is known to lead to macrophage death (11,12), FC-induced death in macrophages in advance atherosclerotic lesions may be a critically important event in advanced lesional complications (13,14).
In this context, our laboratory has studied cellular events that occur during FC loading of macrophages, and we discovered that the cells respond initially by an increase in PC biosynthesis and cellular PC mass, which also appears to occur in lesional macrophages in vivo (15,16). The mechanism of this response is post-translational activation of CTP:phosphocholine cytidylyltransferase (CT; also known as CCT), which catalyzes the conversion of choline-phosphate to CDP-choline (15,17). We have hypothesized that activation of CT in response to FC loading in macrophages is an adaptive response (12). Cellular FC excess is known to be toxic to cells primarily by altering the physical properties of cellular membranes, leading to dysfunction of integral membrane proteins (18). We therefore reasoned that the induction of PC biosynthesis in FCloaded cells would help keep the FC:phospholipid ratio in cellular membrane from reaching cytotoxic levels. Observations of macrophages after prolonged periods of FC loading, where cell death follows a decline in the PC biosynthetic response, provided evidence in support of this hypothesis, but direct molecular support has been lacking (12).
To address this issue, we sought to create a mouse model whose macrophages had deficient CT activity and compromised PC biosynthetic capacity. Our hypothesis predicts that macrophages from such a mouse would be more susceptible to FC-induced cytotoxicity. Our strategy was to manipulate genetically CT in macrophages using the Cre-lox system (19). By way of background, at least two genes in mammalian cells encode CT enzymes as follows: one that gives rise to CT␣, which is ubiquitous and the most active form of CT, and another gene called CT␤, which gives rise to an enzyme called CT␤2 and, at least in humans, to a truncated form arising by alternative splicing called CT␤1 (20,21). CT␣ is subject to regulation by both lipid binding and phosphorylation (22,23). Both forms of CT␤ have catalytic and lipid-binding domains that are homologous to CT␣, and CT␤2 also has a homologous phosphorylation domain (21), but little is known about the regulation or function of either isoform of CT␤.
We now report the creation of a mouse in which the CT␣ gene has been disrupted in macrophages using the Cre-lox system. We show that peritoneal macrophages obtained from this mouse model have no detectable CT␣ protein and markedly decreased, although not absent, CT activity and PC biosynthesis. Interestingly, CT␤2 is induced in these macrophages, and this finding likely explains the residual CT activity in CT␣deficient macrophages, the normal number of peritoneal macrophages present in the mouse, and the ability of the CT␣deficient macrophages to remain healthy under normal culture conditions. Most importantly, CT␣-deficient macrophages are markedly more sensitive to FC-induced death, thus providing important molecular genetic evidence in support of the hypothesis described above.

EXPERIMENTAL PROCEDURES
Materials-The restriction endonucleases and other enzymes were purchased from either New England Biolabs (Beverly, MA) or Life Technologies, Inc. [␣-32 P]dCTP, [methyl-3 H]choline, and phospho [methyl-14 C]choline were from PerkinElmer Life Sciences. The DNA preparation kit was from Qiagen (Chatsworth, CA). The random primer labeling kit and synthetic primers were from Life Technologies, Inc. The pBluescript II KSϩ vector was from Stratagene (La Jolla, CA). Goat anti-rabbit IgG and the Super Signal-enhanced chemiluminescence immunoblot detection kit were purchased from Pierce. The Falcon tissue culture plasticware used in these studies was purchased from Fisher. Tissue culture media and other tissue culture reagents were obtained from Life Technologies, Inc. Fetal bovine serum (FBS) was obtained from HyClone Laboratories (Logan, UT) and was heat-inactivated for 1 h at 65°C (HI-FBS). Compound 58035 (3-[decyldimethylsilyl]-N-[2-(4-methylpheny)-1-phenylethyl] propanamide (24), an inhibitor of acyl-CoA:cholesterol acyltransferase, was generously provided by Dr. John Heider of Sandoz, Inc. (East Hanover, NJ); a 10 mg/ml stock solution was prepared in dimethyl sulfoxide, and the final dimethyl sulfoxide concentration in both treated and control cells was 0.05%. All other chemicals and reagents were from Sigma, and all organic solvents were from Fisher.
Construction of the CT␣ flox Replacement Vector-The CT␣ flox replacement vector was constructed by manipulation of a 12.5-kb fragment of the murine CT␣ gene containing exons 4 -8, the intervening introns, and parts of introns 3 and 8 (clone 4 in Ref. 25). As depicted in Fig. 1A, the "short arm" included a 5.5-kb BamHI fragment containing the 34-bp loxP sequence (5Ј-ATAACTTCGTATAGCATACATTATACGAAGTTAT-3Ј) inserted at the NsiI site, which is 330 bp upstream of exon 4. The "long arm" consisted of a 5.5-kb BamHI/NheI fragment that included exons 6 and 7. These two arms plus a 1.9-kb neomycin resistance cassette (Neo) flanked by loxP sites ("floxed" Neo) and a 1.8-kb thymidine kinase gene cassette were assembled in the order shown in Fig. 1A, and the final construct was embedded into pBluescript II KSϩ vector.
Generation and Identification of Gene-targeted ES Clones-The replacement vector was transfected into 129/Sv ES cells, and G-418resistant clones were selected and screened by Southern blot analysis using a 1-kb NheI/EcoRI exon 8-containing probe (Fig. 1A) as follows. Genomic DNA (3-5 g) was digested with EcoRI and separated in a 0.5% agarose gel. After transferring to the Nylon membrane, the hybridization was performed using Quickhyb solution (Stratagene) containing 1 ϫ 10 6 cpm/ml of 32 P-labeled probe for 2 h and exposed overnight. ES cells undergoing homologous recombination with the replacement vector would show a 6.5-kb band in addition to the 12.5-kb wild-type band (see Fig. 1A and "Results"). To demonstrate that the recombinant ES cells retained the first loxP site (i.e. 5Ј to exon 4), PCR was conducted using the CT382 primer (5Ј-TCTTTGCTTGCATCA-GAGCC-3Ј) and the loxP-5-16 primer (5Ј-CTTCGTATAGCA-3Ј), with an expected PCR product of 440 bp. Another PCR was conducted to exclude the possibility that a positive result of first PCR was due to the random insertion of the target vector in ES cell chromosomal DNA. This was done using the CT382 primer (above) and M13 reverse primer, which is located on pBluescript II KSϩ vector, upstream of the short arm. Homologous recombination would fail to produce a PCR product, whereas random insertion would yield a product of 3.2 kb (see Fig. 1A and "Results").
Generation and Identification of CT␣ flox Mice and CT␣ flox /LysMCre Mice-Cells from an ES cell colony containing homologously recombined CT␣ flox were injected into C57BL6/J host blastocysts, which were then implanted into pseudopregnant female mice by the Animal Core Facility of Rockefeller University. Male offspring with 75-90% agouti color, the coat color contributed by the ES cells, were bred with C57BL6/J females. Pups that were 100% agouti, indicating germ line transmission, were screened as follows. Tail clips were digested for 18 h at 55°C in a buffer containing 0.3 mg/ml proteinase K, 0.1 M NaCl, 10 mM Tris-HCl, pH 8.0, 2.5 mM EDTA, and 0.5% SDS; the DNA was extracted with phenol/chloroform, purified by ethanol precipitation, and dissolved at 65°C in 10 mM Tris-HCl, pH 7.4, containing 1 mM EDTA. The DNA was then subjected to Southern analysis using the exon 8-containing probe as described above. Heterozygous CT␣ flox mice were identified and sibmated, and homozygous CT␣ flox mice resulting from this mating (ϳ25% of the offspring) were identified by PCR using the CT382 primer (above) and a primer called CT3UL (5Ј-GAAGTAG-GCACTGAACTTAGG-3Ј) just upstream of the 3Ј loxP site (see Fig. 2A and "Results"). These homozygous CT␣ flox mice were bred with homozygous LysMCre mice (obtained from Dr. Irmgard Förster, Technical University of Munich), in which the Cre recombinase is driven by the lysozyme promoter via gene targeting into the lysozyme locus (26). The resulting pups, which were heterozygous for both CT␣ flox and LysMCre, were then bred with homozygous CT␣ flox mice. By PCR screening tail DNA from the pups for both CT␣ flox (above) and Cre (26), mice homozygous for CT␣ flox and heterozygous for LysMCre (i.e. mice having CT␣deficient macrophages) were identified. DNA samples from both tail clips and macrophages were further analyzed by the PCR assay described above (CT382-CT3UL) to distinguish the wild-type allele, the CT␣ flox allele, and Cre-mediated recombination (see "Results" and Fig. 2A).
Harvesting, Culturing, and Incubations of Mouse Peritoneal Macrophages-Macrophages were harvested from the peritoneum of mice 3 days after the intraperitoneal injection of 40 g of concanavalin A in 0.5 ml of PBS and then cultured as described previously (27). On the day of the experiment, the cells were washed three times with warm PBS and incubated for the indicated times in 0.2 ml of DMEM, 1% FBS (w/v) alone or containing 100 g of acetyl-LDL/ml plus 10 g of compound 58035/ml, or each compound separately, as described previously (15). At the end of the incubation period, the cells were assayed for cell death as described below.
Cell Death Assays-The binding of Alexa-488 annexin V was conducted using a modification (28) of the method of Vermes et al. (29). After the indicated incubations, the cells were incubated in 100 l of 1ϫ annexin-binding buffer (25 mM HEPES, 140 mM NaCl, 1 mM EDTA, pH 7.4, 0.1% BSA) containing 5 l of Alexa-488 annexin V and 1 l of 100 g/ml propidium iodide for 15 min at room temperature, according to the manufacturer's instructions (Molecular Probes). Cells were immediately viewed with a 40ϫ objective using an Olympus IX-70 inverted fluorescence microscope equipped with filters appropriate for fluorescein (Alexa-488 annexin) and rhodamine (propidium iodide). Five fields of cells for each condition (ϳ1000 cells) were counted.
Immunoblot Analysis-Cells were scraped on ice in PBS, and the total amount of cellular protein was determined by the method of Lowry et al. (30). The cellular homogenates were mixed with concentrated Laemmli sample buffer (31) to give a final protein concentration of 2 g/ml and then boiled for 5 min. Proteins were separated by 5-20% gradient SDS-polyacrylamide gel electrophoresis using 30 g of protein/ lane and then electrotransferred to nitrocellulose membranes. These blots were blocked in Tris-buffered saline containing 5% nonfat dry milk for 1 h at room temperature and then incubated with the indicated antibodies in buffer A (Tris-buffered saline containing 0.1% bovine serum albumin and 0.1% Tween 20 for 3 h at room temperature). The three rabbit polyclonal antibodies used in this study were as follows: an antibody directed against the N terminus of CT␣ that is specific for this form of CT (Ref. 32; kindly supplied by Dr. Claudia Kent, University of Michigan); an antibody directed against the B3 domain of CT␤2 that is specific for this form of CT (21); and an antibody directed against the B2 epitope of CT␤ that recognizes both CT␤1 and CT␤2 (21). After washing four times (10 min each) with buffer A, the blots were incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:15,000) for 1 h in buffer A at room temperature. The blots were then washed extensively with buffer A, incubated with SuperSignal chemiluminescence reagent for 7 min, and exposed to x-ray film for up to 10 min.
Detection of CT␤2 mRNA by RT-PCR-CT␤2 mRNA was detected by RT-PCR (21) using primers based on the sequence of rat CT␤, which is 98% identical in amino acid sequence to human CT␤2 (33). The forward primer was 5Ј-CCAAGGAGCTGAATGTCAGC-3Ј (corresponding to nucleotides 689 -709 of the rat sequence) and the reverse primer was 5Ј-GCACTTGTTAACCAGGCACC-3Ј (corresponding to nucleotides 1141-1121 of the rat sequence); the reverse primer should be specific for CT␤2 based on comparison with the human CT␤1 and CT␤2 sequences (21). The PCR product in rats is 452 base pairs, and a similar size product was found in mice (see "Results").
In Vitro CT Assay-CT activity was assayed in macrophage homogenates by measuring the incorporation of After sitting at room temperature for 30 min, the charcoal was pelleted by centrifugation (1000 ϫ g for 5 min); the supernatant was discarded, and the charcoal pellet was washed four times with water. Finally, the charcoal pellet was extracted with water, ethanol, 28% NH 4 OH (116: 188:11). The extract, which contained the CDP-[ 14 C]choline, was neutralized with glacial acetic acid and counted.

H]Choline Labeling of PC and Biosynthetic
Intermediates-Incorporation of [ 3 H]choline into phosphatidylcholine and biosynthetic intermediates ( Fig. 3) was determined in living macrophages as described by Shiratori et al. (15). Macrophages were pulse-labeled in medium containing 15 Ci of [ 3 H]choline for 15 min and then chased in the same medium without label for 180 min. The cells were washed with PBS and then scraped in methanol:water (5:4, v/v), and this material was then extracted by the method of Bligh and Dyer (35). The organic phase was subjected to TLC fractionation in a solvent system of chloroform:methanol:acetic acid:water (50:25:8:4, v/v), and the radioactivity in phosphatidylcholine was quantified. The aqueous phase was fractionated by TLC in a solvent system of methanol, 0.6% sodium chloride, ammonium hydroxide (10:10:1, v/v). Choline, choline-phosphate, and CDP-choline were visualized by iodine, scraped from the TLC plates, and quantified by scintillation counting. For the PC biosynthesis experiment in Fig.  4C, macrophages were incubated under control or FC-loading conditions for 5 h and then washed with PBS and incubated with 5 mCi/ml [ 3 H]choline for 1 h. The cells were then rinsed with PBS, extracted in hexane:isopropyl alcohol (3:2 v/v), and fractionated by TLC as described previously (15).
Statistics-Results are given as means Ϯ S.E. (n ϭ 3 for all experiments except the cell death experiment in Fig. 6, where n ϭ 5 fields of cells at ϳ200 cells per field); absent error bars in the figures signify S.E. values smaller than the graphic symbols.

RESULTS
Creation of Mice Whose Macrophages Are Depleted in CT␣-To obtain mouse peritoneal macrophages with decreased PC biosynthesis, the overall strategy was to create a mouse with the endogenous CT␣ locus replaced by a CT␣ gene with loxP sites flanking one or more exons critical for CT␣ function. This mouse would then be crossed with a previously described mouse in which the Cre recombinase is driven by the endogenous lysosome promoter (LysMCre) and thus is expressed primarily in differentiated macrophages (as well as in neutrophils) (26). The progeny from this cross should therefore have macrophages with a dysfunctional CT␣ gene (19,26).
To carry out this strategy, we created a replacement vector in which exons 4 and 5, which encode a major portion of the catalytic domain of CT␣ (25), were flanked by loxP sites (Fig.  1A). A Neo cassette was also included so that ES cell transfectants could be positively selected with G-418. Cre-mediated scission at the loxP sites would not only lead to deletion of exons 4 and 5, but the partially deleted gene would also encode a new stop codon at the site of scission. 129/Sv ES cells were transfected with the replacement vector, and Neo-containing clones (i.e. those able to grow in G-418) were isolated and expanded into individual colonies. To distinguish homologous recombination from random insertion, the colonies were subjected to Southern analysis using a probe containing exon 8, which lies external to the 5Ј end of the replacement vector (refer to Fig. 1A). As shown in Fig. 1B, mock-transfected colony 4 showed only the presence of the 12.5-kb wild-type EcoRI fragment (i.e. containing exon 8 but not the Neo cassette), whereas the other colonies shown in this figure demonstrated both the wild-type allele and the recombinant allele, as evidenced by the 6.6-kb EcoRI fragment that indicates the presence of both the Neo cassette and exon 8. To verify that the loxP site 5Ј to exon 4 was still present and in the correct location, the colonies were subjected to PCR as depicted in the bottom scheme in Fig. 1A. The data in Fig. 1C show that all four positive colonies and the vector control, but not the negative colony, yielded the predicted 440-bp PCR product. Finally, to rule out further the possibility of random vector insertion, the colonies were subjected to PCR in which the 5Ј primer was directed to an area of the vector that was external to the 5Ј portion of the CT␣ replacement vector (refer to middle scheme in Fig. 1A). As predicted, none of the colonies yielded the predicted PCR product, which is shown for the vector control (Fig. 1D). Thus, several ES cell transfectants show evidence of successful homologous recombination of the loxP-containing CT␣ replacement vector (CT␣ flox ).
Cells from colony 124 were injected into C57BL6/J blastocysts, which were then implanted into pseudopregnant female mice. Nine males with 75-90% agouti color were born and were eventually bred with C57 female mice. All litters except one were 100% agouti, indicating germ line transmission of the recombinant allele. These pups were screened by Southern analysis (above), and those showing both the 12.5-and 6.5-kb bands (ϳ50% of the pups), indicating heterozygosity for the CT␣ flox allele, were further bred. Some of these heterozygous mice were bred with each other, and the progeny showed the expected ratio of 1:2:1 wild-type:heterozygous:homozygous CT␣ flox mice. Furthermore, all groups of mice developed and gained weight equally and had similar lifespans, indicating no obvious adverse effects of the CT␣ flox genotype.
By using the breeding scheme described under "Experimental Procedures," wild-type, heterozygous, and homozygous CT␣ flox mice with no or one copy of LysMCre were generated, and DNA preparations from tail clips and from peritoneal macrophages of these mice were analyzed by PCR as shown in Fig. 2A. The presence of the loxP site generates a product slightly larger than that derived from the wild-type allele, and Cre-mediated scission results in the absence of a PCR product. Tail DNA from heterozygous and wild-type mice without LysMCre showed the PCR products expected in the absence of scission (lanes 1 and 2 in Fig. 2A), and tail DNA from wild-type and heterozygous CT␣ flox mice with LysMCre had the same respective pattern (lanes 3-5). Thus, in tail DNA, no scission occurred in the presence of LysMCre.
All of the altered mice, including homozygous CT␣ flox / LysMCre mice, yielded numbers of peritoneal macrophages that were similar to those of wild-type mice, and, under normal culture conditions, the cells appeared healthy (see below). Macrophage DNA from heterozygous and wild-type CT␣ flox mice without Cre generated the PCR product expected in the absence of scission (lanes 6 and 7), but macrophage DNA from heterozygous CT␣ flox mice with LysMCre generated no detectable PCR product from the CT␣ flox allele (lanes 9 and 10), indicating disruption of the loxP-exon 4 portion of this allele. Macrophage DNA from homozygous CT␣ flox mice without Cre generated the expected CT␣ flox product (lane 11), whereas macrophage DNA from homozygous CT␣ flox mice with LysMCre generated no detectable PCR products at all (not displayed). In summary, the CT␣ flox allele in macrophage DNA, but not tail DNA, was successfully disrupted in the presence of LysMCre but not in its absence.
To determine if disruption of the CT␣ gene resulted in decreased CT␣ protein expression, homogenates of macrophages from wild-type and homozygous CT␣ flox /LysMCre mice were subjected to immunoblot analysis using an antibody specific for CT␣. As demonstrated in Fig. 2B, the CT␣ flox /LysMCre macrophages expressed no detectable CT␣ in comparison with wildtype macrophages (lane 1) or macrophages with LysMCre alone (not shown). As expected from these data, the homogenates from the CT␣ flox /LysMCre macrophages also demonstrated very little CT activity (Fig. 2C). Thus, Cre-mediated disruption of the macrophage CT␣ gene results in no detectable CT␣ expression and a marked decrease in in vitro CT activity.
Living macrophages from wild-type and CT␣ flox /LysMCre mice were examined for their ability to convert [ 3 H]choline to PC and biosynthetic intermediates. The cells were pulse-la-beled with [ 3 H]choline and then chased without label for 3 h. The data in Fig. 3A show that the cellular content of [ 3 H]choline itself was relatively low in both types of macrophages (compare the y axes of A with those of the other three panels in Fig. 3), indicating that the majority of intracellular label was utilized for PC biosynthesis over the 3-h chase period. The cellular content of [ 3 H]choline-phosphate was approximately 2-fold higher in the CT␣ flox /LysMCre macrophages (Fig. 3B) and that of CDP-[ 3 H]choline was approximately 3-fold lower (Fig. 3C) in comparison with wild-type macrophages. These data are consistent with a partial block in the CT reaction. Finally, as shown in D, [ 3 H]PC content was decreased to a similar relative degree as CDP-[ 3 H]choline content in CT␣ flox / LysMCre macrophages. Thus, disruption of the CT␣ gene results in disruption of choline flux at the CT step, and this results in decreased PC synthesis. 2 CT␤2 Is Present in Macrophages and Induced in CT␣ flox / LysMCre Macrophages-Although CT␣ flox /LysMCre macrophages showed a marked decrease in CT activity and PC biosynthesis (above), these values were clearly not zero. Indeed, the residual CT activity in these macrophages undoubtedly 2 To avoid further manipulation of the ES cells, the "floxed" Neo cassette was not removed prior to blastocyst injection. Probably as a result of the residual presence of this element in the CT␣ gene (44), CT␣ expression and CT activity were decreased ϳ50% in macrophages from homozygous CT␣ flox mice without Cre (data not shown). Therefore, for the experiments in this project, we used macrophages obtained from either wild-type mice or from LysMCre mice as controls that express normal CT␣ protein and activity. accounts for the normal number of peritoneal macrophages and their overall good health under normal culture conditions (see below). In light of the recent findings that a second gene can give rise to another form of CT, called CT␤ (21), we sought evidence for the existence of this enzyme in the CT␣-deficient macrophages. Through alternative splicing, the CT␤ gene gives rise to two transcripts encoding CT␤1 and CT␤2. By using an antibody that specifically recognizes CT␤2 via its unique B3 epitope (21), we were able to detect this protein in homogenates of wild-type macrophages (not shown) and in control macrophages from mice that express LysMCre but not CT␣ flox (1st lane in Fig. 4A). Importantly, in CT␣-deficient macrophages (i.e. those obtained from homozygous CT␣ flox /LysMCre mice), there was increased expression of CT␤2 (Fig. 4A). By using PCR primers that are able to detect murine CT␤2 mRNA by RT-PCR, we compared RNA isolated from macrophages obtained from control mice (i.e. expressing only LysMCre) and from CT␣ flox /LysMCre mice. As shown in Fig. 4B, the CT␤2 RT-PCR product was below the limit of detection in the control macrophages (lane 1) but was clearly visible in the CT␣ flox / LysMCre macrophages (lane 2); for comparison, the PCR products for liver (lane 3) and brain (lane 4) from wild-type mice are also shown. In summary, the data in Fig. 4 indicate that CT␣deficient macrophages have an increased expression of CT␤2 mRNA and protein. PC biosynthesis is induced in wild-type macrophages by FC loading via post-translational activation of CT (15,17), which, as we have seen above, is almost entirely CT␣. The availability of macrophages with CT␤2 but no CT␣ afforded us the unique opportunity to determine whether such cells could also respond to FC loading with an increase in incorporation of [ 3 H]choline into [ 3 H]PC. 4 As shown in Fig. 4C, [ 3 H]choline incorporation in control macrophages was increased 2-fold after 5 h of FC loading. Although total [ 3 H]PC in CT␣-deficient macrophages was much lower than that in the control cells, it was clearly induced by FC loading. In fact, this value in the FC-loaded CT␣-deficient macrophages might be as much as ϳ30% higher given the substantial number of dead cells in this population (see below). These data suggest that the signaling involved in activation of CT␣ in control macrophages (17) may also work on CT␤2 in the CT␣-deficient macrophages. The data also show that the level of [ 3 H]choline incorporation into [ 3 H]PC after a 5-h period of FC loading, the precise conditions used in the cell death studies below, is substantially less in CT␣-deficient macrophages versus control macrophages.
CT␣-deficient Macrophages Are Markedly More Susceptible to FC-induced Death-We have hypothesized that the increase in PC biosynthesis and cellular PC mass that occurs in FCloaded macrophages is part of an adaptive response (12). We propose that this response helps protect cells, albeit transiently, from FC-induced death by preventing a lethal rise in the FC:PC ratio in critical cellular membranes (12). By having access to macrophages that have deficient CT activity and PC biosynthesis but that are able to remain healthy under normal culture conditions allowed us to test this hypothesis directly. Therefore, peritoneal macrophages from control mice (expressing only LysMCre) and from CT␣ flox /LysMCre mice were incubated under control or FC-loading conditions for 5 h. To detect apoptotic changes, the cells were stained with the phosphatidylserine-binding protein, annexin V, which in the assay used here is conjugated to Alexa-488 (green) (29). We have recently shown that after 9 h of FC loading, peritoneal macrophages undergo apoptotic changes, including externalization of phosphatidylserine (28). To detect more advanced cell death, including necrotic death, the cells were also exposed to the membrane-impermeable nucleic acid fluorophore, propidium iodide (red-orange). As shown in Fig. 5A, most of the control macro-CT␤2 (but not CT␣) via their common B2 epitope (21), we were unable to detect CT␤1 in either wild-type or CT␣-deficient macrophages or in mouse liver or brain; as a positive control, a prominent band was detected in Chinese hamster ovary cell transfectants expressing large amounts of human CT␤1 (data not shown; see Ref. 21). Moreover, neither human CT␤1 primers (21) nor primers designed from the rat sequence that theoretically should detect CT␤1 based on the human sequence yielded a specific product in mouse tissues (data not shown and M. Bakovic and D. Vance, personal communication). Thus, it is possible that the alternative splicing that generates CT␤1 from the CT␤ gene in humans does not occur or is a minor pathway in the mouse. 4 In a previous study (15)  phages under non-loading conditions showed no signs of death, and after 5 h of FC loading, the number of dead cells increased only slightly (Fig. 5B). CT␣-deficient macrophages under nonloading conditions also appeared healthy, and the vast majority demonstrated neither Alexa-488 annexin nor propidium iodide staining (Fig. 5C). In striking contrast, there were many an-nexin and/or propidium iodide-stained cells among the FCloaded CT␣-deficient macrophages (Fig. 5D).
Quantification of data from a large number of cells verified that 5 h of FC loading in CT␣-deficient macrophages resulted in much more cell death than that observed in similarly treated control macrophages (Fig. 6A). To rule out the possibility that CT␣ deficiency enhanced the effect of cell death inducers in general, we treated control and experimental macrophages with the protein kinase C inhibitor staurosporine, which is a known inducer of apoptotic death in mononuclear leukocytes (36). As shown in the first set of bars in Fig. 6B, we chose a concentration of staurosporine that causes only partial cell death in control macrophages so that any enhancing effect could be readily detected in the experimental macrophages. As shown by the data in Fig. 6B, however, CT␣-deficient macrophages show no significant increase in staurosporine-induced death compared with the control cells. In summary, the absence of CT␣, leading to a deficiency of PC biosynthesis, renders macrophages much more susceptible to FC-induced death. This finding strongly supports the hypothesis that the activation of CT␣ and the increase in PC biosynthesis seen in FCloaded macrophages is a cell-survival adaptive response. DISCUSSION A major impetus for the project described in this report was related to our previous finding that macrophages respond to FC loading by increasing CT activity, PC biosynthesis, and cellular PC mass (15). Several observations have led us to propose that these events represent an adaptive response to prevent, at least initially, the FC:phospholipid ratio from reaching cytotoxic levels (18). For example, after prolonged FC loading of macrophages, CT activity and PC biosynthesis decline, and shortly thereafter the cells begin to die (18). The most direct test of this hypothesis, however, required studying macrophages with defective PC biosynthesis via molecular genetic manipulation of CT. The data in this report clearly demonstrate that such macrophages are markedly more susceptible to FC-induced death, thus providing strong support to the hypothesis. We propose that in FC-loaded wild-type macrophages, the excess PC, which accumulates mostly in intracellular whorllike membrane structures (15), acts as a sink for the excess FC and thus prevents accumulation of FC in cellular membranes.
Based on the data of Yeagle (18) and others (14), this process would be expected to promote cell survival by protecting enzymes and transport proteins in the plasma membrane and other membranes from an abnormally rigid lipid environment. The eventual failure of this adaptive response even in wild-type macrophages (above) may be due to direct inactivation of CT itself by an overwhelming load of FC. Moreover, this process may be accelerated by apoptosis-induced inhibition of choline phosphotransferase (37). In FC-loaded CT␣-deficient macrophages, the failure of adaptation is accelerated by the low level of PC biosynthetic activity in these cells.
The unique model described in this report has allowed us to address another important issue related to macrophage PC metabolism, namely the overall response of macrophages to CT␣ deficiency. Our data suggest that an increased level of expression of CT␤2 in CT␣-deficient macrophages provides sufficient CT activity and PC biosynthesis to allow the cells to survive under normal conditions both in vivo, as evidenced by the overall health of the mice and the normal number of peritoneal macrophages, and in cell culture (Fig. 5C). The unmasking of a phenotype in these macrophages under FC-loading conditions, however, suggests that other "stressors" might also uncover interesting deficiencies in these cells. For example, macrophage proliferation (38), an event known to occur in atherosclerotic lesions (39 -41), as well as other cellular events associated with membrane remodeling, such as phagocytosis and exocytosis (42,43), might be deficient in CT␣-deficient macrophages.
The functions of CT␤ are still under investigation. In our study with murine macrophages, we were only able to detect the presence of CT␤2, which could be due to either suboptimal reagents for murine tissue (i.e. antibodies and PCR primers based on human or rat CT␤) or to the possibility that mice have no or very low levels of CT␤1. 3 CT␤2 lacks the N-terminal nuclear localization domain present in CT␣ but has regions FIG. 6. Quantitative analysis of cell death in control or CT␣-deficient macrophages exposed to FC loading or staurosporine. A, quantitative data from the experiment in Fig. 5 in which the percentage of total cells that were green, red-orange, or both in 5 fields of cells was determined (total cells counted for each condition ϭ 700 -1000); basal (cross-hatched bars) refers to macrophages not loaded with FC. B, an experiment similar to that in Figs. 5 and 6A was conducted in which the macrophages were incubated for 5 h in the absence (Basal) or presence of 50 nM staurosporine (instead of acetyl-LDL plus 59035).
homologous to the catalytic, membrane-binding, and C-terminal phosphorylation domains of CT␣ (21). Most importantly, CT␤ demonstrates catalytic activity, although its specific activity appears to be less than that of CT␣ when expressed in COS-7 cells (20). This latter point may account for our finding that PC biosynthesis was still markedly decreased in CT␣deficient macrophages even in the face of induced expression of CT␤2. Another interesting result from our studies was that [ 3 H]choline incorporation into [ 3 H]PC was increased by FC loading in CT␣-deficient macrophages (Fig. 4C). Pending further work in this area, we hypothesize that CT␤2, like CT␣ (15,17), can be activated by FC loading. In this context, we previously found that FC-induced CT activation in wild-type macrophages was associated with dephosphorylation of CT␣ (17), and so the presence of a phosphorylation domain in CT␤2 may be important in the increase in [ 3 H]PC biosynthesis in FC-loaded CT␣-deficient macrophages.
The current study has focused on cultured peritoneal macrophages obtained from the unique model described herein. However, in future studies we plan to address the role in vivo of PC biosynthesis in the function of macrophages as well as neutrophils, which are also targeted by LysMCre (26). To carry out these studies, it will first be necessary to remove the "floxed" Neo cassette from CT␣ flox gene in our mice (44). 2 Once this has been accomplished, we will be able to test specific hypotheses in vivo using the LysMCre system. For example, we and others (13,14,45) have proposed that FC-induced macrophage death is an important cause of necrosis in advanced atherosclerotic lesions. Based on this proposal and the cell culture data reported here, we would predict that lesional necrosis would be increased in CT␣ flox /LysMCre mice with an atherosclerotic background (e.g. after crossing into the apolipoprotein E knockout background). Because lesional necrosis is a likely contributor to acute vascular events (46), such a finding might have important implications related to the clinical complications of atherosclerotic vascular disease. Similarly, CT␣ flox /LysMCre mice can be used to study other PC-dependent functions of macrophages in vivo (e.g. in host defense), and the CT␣ flox mice can be crossed with mice expressing Cre in other tissues to address specific questions related to CT␣, CT␤, and PC biosynthesis in these tissues.