Chlamydia trachomatis Relies on Autonomous Phospholipid Synthesis for Membrane Biogenesis*♦

Background: C. trachomatis has a reduced genome and was thought to obtain phospholipids as well as other nutrients from the host. Results: The new phospholipid molecular species that appear in infected cells are produced by C. trachomatis. Conclusion: C. trachomatis needs only glucose, isoleucine, and serine to synthesize its own membrane phospholipids. Significance: C. trachomatis relies on autonomous phospholipid synthesis. The obligate intracellular parasite Chlamydia trachomatis has a reduced genome and is thought to rely on its mammalian host cell for nutrients. Although several lines of evidence suggest C. trachomatis utilizes host phospholipids, the bacterium encodes all the genes necessary for fatty acid and phospholipid synthesis found in free living Gram-negative bacteria. Bacterially derived phospholipids significantly increased in infected HeLa cell cultures. These new phospholipids had a distinct molecular species composition consisting of saturated and branched-chain fatty acids. Biochemical analysis established the role of C. trachomatis-encoded acyltransferases in producing the new disaturated molecular species. There was no evidence for the remodeling of host phospholipids and no change in the size or molecular species composition of the phosphatidylcholine pool in infected HeLa cells. Host sphingomyelin was associated with C. trachomatis isolated by detergent extraction, but it may represent contamination with detergent-insoluble host lipids rather than being an integral bacterial membrane component. C. trachomatis assembles its membrane systems from the unique phospholipid molecular species produced by its own fatty acid and phospholipid biosynthetic machinery utilizing glucose, isoleucine, and serine.

(CT775 ϩ CT776) that function to re-cycle PE that is degraded during lipoprotein maturation (9,10). The genes in Fig. 1 are all expressed by 8 h post-infection when the RBs replication begins (11,12). The exceptions are the acc genes, which are detected at 3 h, and the cls gene rises at 16 h. Thus, the genes required to initiate FASII are induced when the RBs are replicating, and the CL synthesis gene is elevated as the cells differentiate into EB. There are no missing pieces that would suggest C. trachomatis would need to import host phospholipids to assemble its inner or outer membrane systems. The bioinformatic model predicts that the only phospholipid precursors that would be required from the host are as follows: isoleucine, to initiate branched-chain fatty acid synthesis; glucose, to supply acetyl-CoA and Gro-3-P; and serine, for the synthesis of PE.
Although it is widely understood that C. trachomatis appears capable of synthesizing its own lipids, two main lines of evidence have led to the conclusion that C. trachomatis depends on the acquisition of host phospholipids for membrane biogenesis and proliferation (13,14). First, isolated C. trachomatis have a phospholipid class composition that reflects the composition of the host cell (15)(16)(17)(18)(19). One distinct difference is that branched-chain fatty acids (i.e. anteiso-15:0; number of carbons/number of double bonds) are found in the 2-position of phospholipids in C. trachomatis-infected HeLa cells (20). Mammalian cells do not make this type of fatty acid, but radioactivity from the branched-chain fatty acid precursor isoleucine is incorporated into host phosphatidylcholine (PC) leading to the idea that host phospholipids were remodeled with branched-chain fatty acids by a deacylation-reacylation pathway (18). Host cPLA 2 is activated during infection and is proposed to initiate the remodeling pathway by hydrolyzing fatty acids from the 2-position of host phospholipids (21). Recently, a C. trachomatis acyltransferase was reported to reacylate lyso-PC to thereby complete the remodeling process (22). Second, C. trachomatis co-opts host trafficking pathways to mobilize host cell membrane lipids (PC, cholesterol, and sphingomyelin (SM)) from cellular organelles to assemble the inclusion and bacterial membrane systems. SM and cholesterol are trafficked from the Golgi (15,16,19,(23)(24)(25), and components of the host high density lipoprotein biogenesis machinery are recruited to the inclusion membrane to facilitate the translocation of PC (26). Lipid transfer is facilitated by the close association of the C. trachomatis inclusion with host organelles such as the endoplasmic reticulum (23,(27)(28)(29)(30) and multivesicular bodies (31)(32)(33). C. trachomatis is also proposed to utilize host lipid droplets and fatty acid-binding proteins to assimilate lipids (24,34,35). A functioning inclusion membrane is critical to C. trachomatis proliferation, so it is not clear whether the lower C. trachomatis titers observed when lipid trafficking is inhibited arise from insufficient lipid for C. trachomatis and/or inclusion membrane formation. This has been a particularly difficult area for study because the inclusion membrane has not been isolated in vitro, and there are tight connections between C. trachomatis and the host-derived inclusion membrane through a structure recently described as the "pathogen synapse" (3,36,37). These tight connections suggest that it may be difficult to isolate bacteria devoid of host membrane contamination.
The goal of this study is to determine the contributions of C. trachomatis phospholipid synthesis and host phospholipid modification. We find that de novo C. trachomatis lipid synthesis is responsible for the increase in abundance of PE, PG, and CL in C. trachomatis-infected cultures. Lipidomic analyses show that there is no modification of host phospholipids with bacterially derived fatty acids and that C. trachomatis produces a unique constellation of disaturated phospholipid molecular The genes were easily identified based on comparisons with well characterized homologues in E. coli (6) with the exception of the Gro-3-P acyltransferase. The gene predicted to encode this activity (plsE) is most closely related to the soluble acyltransferase of plant plastids (7). C. trachomatis lacks the genes for unsaturated fatty acid biosynthesis (65), but like many bacteria, C. trachomatis produces branched-chain fatty acids as a substitute for unsaturated fatty acids. C. trachomatis has a branched-chain ketoacid dehydrogenase (bkd) to supply the primer for branched-chain fatty acid biosynthesis, and gpsA (CT714) produces the Gro-3-P for the acyltransferases. The C. trachomatis cls gene was related to all three E. coli cls genes (e-value of 1e Ϫ6 to cls1, 6e Ϫ9 to cls2, and 5e Ϫ5 to cls3). The multiple enzymes that dephosphorylate phosphatidylglycerolphosphate were difficult to identify in E. coli (64), and the phosphatase in C. trachomatis was not identified with confidence. species that are not found in the host. These results establish that C. trachomatis uses its own biosynthetic pathways ( Fig. 1) to synthesize unique molecular species of PE, PG, and CL that are essential for the biogenesis of its membrane systems.

Experimental Procedures
Materials-Unless otherwise stated, chemicals were obtained from either Sigma or Fisher. Radiolabeled chemicals were from American Radiolabeled Chemicals, and other isotopic labeled chemicals were from Cambridge Isotope Laboratories. Phospholipids were obtained from Avanti Polar Lipids. Cell culture media and supplies were from Invitrogen.
Chlamydia Strain and Propagation-C. trachomatis serovar L2 (strain 434/Bu) was propagated by standard infection through centrifugation protocol in HeLa cells grown in DMEM containing 10% fetal bovine serum (38) in 6-well tissue culturetreated plates (Corning). Mock infection controls for HeLa cells were performed by going through the same protocol but without C. trachomatis.
Phospholipid NMR-Eight plates of mocked infected or C. trachomatis-infected HeLa cells grown for 48 h after infection were collected for each NMR sample. The cells were scraped from the plate and washed twice with phosphate-buffered saline. The lipids were extracted using the Bligh and Dyer method (39). The NMR analysis was performed as described (40). Briefly, the lipid extracts were suspended into 500 l of the cholate detergent system through vortexing and dispersion in a sonicating bath at 60°C for 30 min. The cholate detergent system consisted of sodium cholate (10% w/v), EDTA(10% w/v), phosphonomethylglycine for internal standard (0.3 g/liter), and 20% D 2 O for deuterium field frequency locking. The pH was adjusted to 7.3 using sodium hydroxide. Continuous protondecoupled 31 P NMR spectra were recorded on a Bruker Ultrashield 400 Plus instrument at a probe temperature of 300 K, tuned to 161.98 MHz, using 5-mm high resolution NMR tubes. Spectra were obtained with a spectral width of 64102.56 Hz, 2.0 s relaxation delay, and 32,768 complex points in the time domain using simultaneous detection of real and imaginary components. Ten thousand scans were taken per sample. Chemical shifts are reported relative to the internal sample of 0.3 g/liter phosphonomethylglycine (Ϫ7.33 ppm). The phospholipids were identified by comparing the chemical shifts to phospholipid standards run under the same conditions. Experiments were performed in triplicate with average and standard deviations reported.
Chlamydia The lipids were extracted from the harvested cells, and the amount of label incorporated was quantified by liquid scintillation counting. The distribution of [ 14 C]glucose label in different lipid classes was analyzed via thin layer chromatography as described above. HeLa cells were plated at 12.5% confluence on 10-cm cell culture plates in DMEM with 10% FBS (10 ml) and 5 Ci of [ 3 H]arachidonic acid per plate. The cells were allowed to grow to confluence over 3 days, and then cells from each 10-cm plate were split into 2 ϫ 6-well plates. The cells in the 6-well plates were infected (or mock-infected) with C. trachomatis and grown for 48 wells. Then, the media were collected, and the cells were resuspended into HBSS and harvested. The radioactivity in the media was counted by liquid scintillation counting to determine the amount of [ 3 H]arachidonic acid released into the media. The harvested cells were extracted, and the amount of radioactivity remaining in the cells was determined by liquid scintillation counting. The extracts were also chromatographed on Silica Gel H layers developed with 50:25:8:4 chloroform/ methanol/acetic acid/water, and percentage of PE radioactivity in the phospholipid fraction was determined. The experiment was conducted in quadruplicate with each replicate consisting of three wells of material. For ethanol-1,1,2,2-d 4 -amine (D4-Etn) labeling experiments, C. trachomatis-infected HeLa cells were incubated in DMEM supplemented with 100 M D4-Etn at 18 h post-infection for 30 h. Cells were collected, and the lipids were extracted and analyzed by mass spectrometry.
Mass Spectrometry-Phospholipid molecular species fingerprints were determined using direct infusion electrospray ionization-mass spectrometry technology (41,42). Mass spectrometry analysis was performed using a Finnigan TM TSQ Quantum (Thermo Electron, San Jose, CA) triple quadruple mass spectrometer. The lipid extracts were resuspended in 50:50 (v/v) chloroform/methanol with 1% formic acid. The instrument was operated in positive ion mode for PC and PE analysis and negative ion mode for PG and CL analysis. Ion source parameters were spray voltage 3500 V, capillary temperature 270°C, and capillary offset 35 V, and the tube lens offset was set by infusion of the polytyrosine tuning and calibration solution (Thermo Electron, San Jose, CA) in electrospray mode. Parameters for the analysis of PC are as follows: scan range, 600 -900 m/z; scan time, 0.3 s; product mass, 184.1 m/z; collision energy, 40 V; peak width, Q1 and Q3 0.7 full width at half-maximum (FWHM); and Q2 CID gas, 0.5 mtorr. Parame-ters for the analysis of PE are as follows: scan range, 600 -900 m/z; scan time, 0.3 s; neutral loss mass, 141.0 m/z; collision energy, 30 V; peak width, Q1 and Q3 0.7 FWHM; and C2 CID gas, 0.5 mtorr. Instrument control and data acquisition were performed using the Finnigan TM Xcalibur software (Thermo Electron, San Jose, CA). Acyl chain lengths were assigned from the mass based on predications from LipidMaps (41).
For fatty acids scans, lipids from C. trachomatis-infected HeLa cells were separated using Silica Gel G layers developed with chloroform/methanol/acetic acid/water (50:20:8:3, v/v). The area of silica where PE migrated was scraped off the plate, and PE was extracted using the Bligh and Dyer method. The instrument was operated in negative ion mode. Parameters for fatty acid scans are as follows: scan range, 600 -900 m/z; scan time, 0.3 s; collision energy, 45 V; peak width, Q1 and Q3 0.7 FWHM; and Q2 CID gas, 0.5 mtorr. The parent scans for each fatty acid with their m/z are as follows: ai15:0, 257. C. trachomatis Isolation Methods-The standard Renografin protocol was used to isolate the EB of C. trachomatis (44). Briefly, infected HeLa cells were collected by cell scraping, washed twice with HBSS, and lysed with gentle sonication (three time for 20 s at 8 watts) in HBSS on ice. The resulting solution was centrifuged at 500 ϫ g for 15 min, and the supernatant was pelleted via centrifugation over a 40% sucrose cushion (0.5 ml of 40% sucrose for 1 ml of supernatant) for 30 min at 20,000 ϫ g. The resulting pellet was resuspended in sucrose/ phosphate/glutamic acid buffer (10 mM sodium phosphate (8 mM Na 2 HPO 4 ⅐2 mM NaH 2 PO 4 ), 220 mM sucrose, 0.50 mM L-glutamic acid) and centrifuged over a Renografin (Reno-60 Bracco Diagnostics) gradient. Briefly, a 4-ml gradient was composed of layers of 0.75 ml of 54% Reno-60, 1.75 ml of 38% Reno-60, and 1.25 ml of 30% Reno-60 in sucrose/phosphate/glutamic acid. The cells collected from four plates were loaded onto a 4-ml gradient via centrifugation at 150,000 ϫ g for 45 min at 4°C. The EB banded at the 38 and 54% Reno-60 interface and was extracted via syringe. The EB band was diluted 10-fold with sucrose/phosphate/glutamic acid buffer and centrifuged at 80,000 ϫ g for 30 min at 4°C to pellet and collect the isolated EB.
The Nonidet P-40 isolation method was based on two previous studies (45,46). Infected HeLa cells were collected by cell scraping, washed twice with HBSS, resuspended in 0.5% Nonidet P-40 (octyl phenoxypolyethoxylethanol, Sigma IGEPAL CA-630), and lysed with gentle sonication (three times for 20 s at 8 watts). The resulting solution was centrifuged at 500 ϫ g for 15 min, and the supernatant was pelleted over a 40% sucrose cushion via centrifugation for 30 min at 20,000 ϫ g. The pellet was resuspended in 0.1% Nonidet P-40 and centrifuged over 40% sucrose cushion two more times to collected the EB.
Molecular Biology-The plsE (CT807) and plsC (CT453) genes of C. trachomatis strain D/UW-3/Cx (NCBI Microbial Genomes Database) was optimized for expression in E. coli through GeneArt Gene Synthesis Technology (Life Technologies, Inc.). For the plsC gene, an NdeI restriction site was engineered at the 5Ј of the gene with start codon in the NdeI site, and a His 6 tag, stop codon, and an EcoRI restriction site were sequentially engineered at the 3Ј of the gene. The plsC gene was cloned into the pPJ131 plasmid (a modified version of the pBlueScript plasmid with the multiple cloning site from pET28a) via the NdeI and EcoRI (New England Biolabs) restriction sites (5). A BamHI restriction site was engineered at the 5Ј of the plsE gene, and a His 6 tag, stop codon, and an EcoRI restriction site were sequentially engineered at the 3Ј of the gene. The plsE gene was cloned into pFastBac1 plasmid (Life Technologies, Inc.) via the BamHI and EcoRI restriction sites.
PlsE Characterization and Assay-The pFastBac1-CtplsE vector was expressed in insect cells following protocols of the Invitrogen Bac-to-Bac Expression System and then purified via standard nickel chelation chromatography. The fractions containing protein, as determined by the Bradford reagent (47), were collected and dialyzed against 20 mM Tris, pH 8.0, 10 mM EDTA, and 150 mM NaCl at 4°C overnight. A pure protein (Ͼ95%) running at ϳ38 kDa (theoretical average mass of 38.7 kDa with an N-terminal methionine) was observed on a NuPAGE 10% bis-Tris gel (Life Technologies, Inc.). Approximately 5-10 mg of purified PlsE was purified per liter of culture. Synthesis of 16:0-ACP-E. coli ACP was used for the PlsE assay. Apo-ACP was purified as described previously (48) and converted into 16:0-ACP in a two-step reaction. Apo-ACP (100 M), dithiothreitol (5 mM), MgCl 2 (10 mM), coenzyme A (1 mM), and 10 g per ml of Streptococcus pneumoniae ACP synthase was incubated in 100 mM Tris-HCl, pH 8.0, at 37°C for 2 h. Then, a final concentration of ATP (10 mM), palmitic acid (16:0, 150 M) in DMSO (1% final volume), Vibrio harveyi acyl-ACP synthetase (10 g per ml) was added, and the reaction was incubated at 37°C for an additional 4 h. Reaction completion was monitored by analyzing the reaction in a 2.5 M, 15% acrylamide urea gel to detect acyl-ACP formation (49). The reaction mixture was adsorbed to a small nickel-nitrilotriacetic acid column (10 l of resin per 1-ml reaction) to remove the ACP synthase and acyl-ACP synthetase and dialyzed overnight in 20 mM Tris, pH 6.0, at 4°C.
PlsC Characterization-The pPJ131-CtplsC plasmid was transformed into the plsC temperature-sensitive E. coli strain SM2-1 to determine complementation. Strain SM2-1 was viable at 30°C but nonviable at 42°C without plsC gene complementation. The SM2-1 cells were transformed with the pPJ131-CtplsC plasmid, the pPJ131 parent plasmid, and the pBluescript plasmids expressing plsC from E. coli and Staphylococcus aureus (50) and then plated on Luria Bertani (LB) plates at 30°C with 100 g/ml carbenicillin. The transformed cells were restreaked onto LB plates with 100 g/ml carbenicillin and grown at 30 or 42°C to determine whether the C. trachomatis plsC complements the E. coli plsC activity. For molecular species profiling, strain SM2-1 expressing the three different plsC genes were grown in LB with 100 g/ml carbenicillin overnight at 30°C. The cultures were diluted to A 600 ϭ 0.05, and grown in 10 ml of LB at 42°C until A 600 ϭ 1. The cells were collected and the lipids extracted via the Bligh and Dyer method, and the PE molecular species were analyzed by mass spectrometry.

Phospholipid Content and Synthesis in Ctr-infected HeLa
Cells-The phospholipid compositions of uninfected and C. trachomatis-infected HeLa cells were quantified and compared using 31 P NMR spectroscopy. In uninfected HeLa cells, PC was the predominant species, followed by PE, SM, phosphatidylserine (PS), phosphatidylinositol, and CL ( Fig. 2A). PG was below the level of detection. In C. trachomatis-infected HeLa cells, there was a significant increase in three phospholipid classes, PE, PG, and CL (Fig. 2B). These three phospholipids were predicted by the bioinformatic model to be synthesized by C. trachomatis (Fig. 1). PC and SM did not increase and PS decreased (Fig. 2B). Thus, the phospholipid class analysis suggested new membrane systems with a composition that was characteristic of Gram-negative bacteria were being produced in C. trachomatis-infected HeLa cultures.
Metabolic labeling experiments were conducted to determine whether the metabolism of PC and PE was altered between 18 and 26 h post-infection. Glucose was selected as a metabolic precursor because it should gain entry to both host and bacterial intermediary metabolism and be converted to Gro-3-P and acetyl-CoA for the synthesis of host and bacterial phospholipids. [ 14 C]Glucose incorporation into total phospholipids doubled in the C. trachomatis-infected HeLa cultures compared with the uninfected cultures (Fig. 2C). The distribution of label between the PC and PE phospholipid classes was determined by thin layer chromatography. There was a relative decrease in the amount of [ 14 C]glucose incorporated into PC in the C. trachomatis-infected cells (Fig. 2D) that was reflected by a relative increase in glucose labeling of PE (Fig. 2E). PC is produced by the CDP-choline pathway in HeLa cells, and the activity of this pathway was assessed by labeling with [ 14 C]choline (51). The amount of [ 14 C]choline incorporated into PC was the same in uninfected and infected HeLa cultures (Fig. 2F) indicating that C. trachomatis infection had little impact on the activity of the CDP-choline pathway. Mammalian cells synthesize PE by both the CDP-ethanolamine and PS decarboxylase pathways (52), whereas C. trachomatis was predicted to use only the bacterial PS decarboxylase pathway (Fig. 1) (Fig. 2G). In contrast, the incorporation of [ 14 C]serine increased by 50% in the C. trachomatis-infected HeLa cultures indicating that the PS decarboxylase pathway activity was elevated in infected cells (Fig. 2H). The compositional and metabolic labeling data were consistent with C. trachomatis autonomous phospholipid synthesis driving the significant changes in phospholipid class composition during infection.
Phospholipid Molecular Species in C. trachomatis-infected Cells-Previous work reported the appearance of bacterially produced branched-chain fatty acids (notably anteiso-15:0) in infected HeLa cells (17,18,20). One idea was that branched-chain fatty acids were incorporated into host phospholipids by a deacylation-reacylation cycle (18,21). Alternatively, the bioinformatic model suggested that C. trachomatis would incorporate branched-chain fatty acid products of FASII using its own Gro-3-P acyltransferase system (Fig. 1). Molecular species analysis was performed on the lipid extracts of uninfected or C. trachomatis-infected HeLa cells to determine which phospholipids contained branched-chain fatty acids and how the structures of PC, PE, PG, and CL were altered in C. trachomatis-infected cells. There were no new PC molecular species detected in C. trachomatis-infected HeLa cells com- pared with uninfected cells (Fig. 3, A and B). In both cases, the most abundant PC species were 16:0/18:1, 16:0/16:1, 18:1/18:1, and 14:0/16:0. There was no evidence for branched-chain fatty acids in the PC molecular species, ruling out their incorporation into PC via a deacylation/reacylation pathway. In contrast, the PE molecular species profile was significantly different in the C. trachomatis-infected cultures (Fig. 3, C and D). Abundant PE species observed in uninfected HeLa cells contained polyunsaturated fatty acids (18:0/20:4 and 18:1/22:6). These major PE molecular species in uninfected cells became minor species in the C. trachomatis-infected HeLa cells. In the C. trachomatis infected HeLa cells, odd-number disaturated PE molecular species containing branched-chain fatty acids became predominant. The most abundant new PE species were 16:0/15:0 and 17:0/15:0. These data showed that bacterially derived branched-chain fatty acids were only detected in phospholipids predicted to be bacterially synthesized, but they were absent from host phospholipids. These data also showed that the contribution of C. trachomatis-derived PE to the total PE in the infected cultures was larger than the 31 P NMR experiments implied (Fig. 2, A and B) because major HeLa cell PE molecular species containing polyunsaturated fatty acids were lost from the PE pool as a consequence of C. trachomatis infection.
PG was a phospholipid that was observed in infected HeLa cells that was not detected in uninfected HeLa cells (Fig. 2, A  and B). The bioinformatic model posits that PG and PE would arise from a common PA precursor, and therefore these two phospholipids would be expected to have a similar molecular species profile. The major PG molecular species from uninfected HeLa cells contained unsaturated fatty acids (primarily 16:0/18:1 and 18:1/18:1) (Fig. 3E). The molecular species of PG isolated from C. trachomatis-infected HeLa cells (Fig. 3F) were dominated by the same saturated branched-chain fatty acid combinations that characterized the new PE molecular species appearing in C. trachomatis-infected cells (Fig. 3D). Host CL molecular species contained primarily unsaturated fatty acids (Fig. 3G). In contrast, the most abundant CL species in C. trachomatis-infected HeLa cells were saturated and contained between 60 and 65 total carbons (Fig. 3H). The most abundant CL species containing 62 carbons corresponded to the condensation of the most abundant PG molecular species (16:0/15:0; Fig. 3F). The data showed that the new PE, PG, and CL have molecular species profiles consistent with the bioinformatic model that predicts they arise from the same PA pool.
Host Phospholipids Were Not Remodeled during C. trachomatis Infection-C. trachomatis infection activates cPLA 2 , and the degraded phospholipids may be subsequently modified by putative C. trachomatis acyltransferases to create new molecular species (21,53). The reduction in host PE molecular species containing polyunsaturated fatty acids in C. trachomatis-infected cultures was consistent with the degradation and/or remodeling of selected PE molecular species. Metabolic labeling experiments with D4-Etn were performed to determine whether the new PE molecular species in C. trachomatis-infected cultures were derived from remodeling of host PE. Host PE arises from either the CDP-Etn or PS decarboxylase pathways, whereas the bioinformatics model predicted that C. trachomatis only uses the PS decarboxylase route (Fig. 1). Thus, host PE was selectively labeled via the CDP-Etn pathway by growing C. trachomatis-infected HeLa cells in the presence of 100 M D4-Etn. The PE molecular species containing normal and D4-Etn were identified by mass spectrometry. The molecular species profile of PE containing normal Etn (Fig. 4A) was the same as previously determined for C. trachomatis-infected cells (Fig. 3D). PE molecular species containing heavy D4-Etn consisted only of unsaturated host PE molecular species (Fig.  4B). There was no evidence for D4-Etn incorporation into the new saturated PE molecular species that arose in C. trachomatis-infected cell cultures (Fig. 4B) (Fig. 4B) meaning that the PE backbone was of bacterial origin. Mass spectrometry was used to identify phospholipids containing a branched-chain fatty acid (15:0) produced by C. trachomatis FASII. The branched-chain fatty acid was only detected in the saturated PE molecular species arising in C. trachomatis-infected cells (Fig.  4C). We also detected a PG molecular species containing 15:0 (Fig. 4C). Significantly, PC did not contain 15:0. These data were consistent with the conclusion that the new PE molecular species found in C. trachomatis-infected cells arose from the de novo C. trachomatis biosynthetic pathway (Fig. 1) and not from the modification of pre-existing host PE.
The analysis of C. trachomatis-infected cells indicated that PE molecular species containing polyunsaturated fatty acids were selectively reduced during the infection. This point was investigated by pre-labeling HeLa cells with [ 3 H]arachidonic acid and, after removing the label from the cells, comparing uninfected and C. trachomatis-infected HeLa cells to determine whether infection altered the metabolism of PE-containing [ 3 H]arachidonic acid. There was a significant increase in the release of [ 3 H]arachidonic acid into the medium and a concomitant decrease in [ 3 H]arachidonic acid in the PE of C. trachomatis-infected cultures (Fig. 4D). These data indicated that the selective degradation of host polyunsaturated PE molecular species accounted for the alteration in the host PE molecular species profile in C. trachomatis-infected cells (Fig. 3, C and D).
C. trachomatis Acyltransferases and PA Synthesis-Gro-3-P acyltransferase (PlsE) and 1-acyl-Gro-3-P acyltransferase (PlsC) encoded by C. trachomatis were predicted to catalyze the successive acylation of Gro-3-P to make PA, the precursor to all bacterial phospholipid species (Fig. 1). Therefore, the substrate specificities of the acyltransferases will be reflected in all derived phospholipid classes and thereby explain the similarity in the molecular species profiles of PE, PG, and CL arising in C. trachomatis-infected cultures (Fig. 3). There are two characterized integral membrane bacterial Gro-3-P acyltransferases, PlsB and PlsY (8). PlsB is found in ␥-proteobacteria and mammals and utilizes either acyl-ACP or acyl-CoA as the acyl donor. PlsY is the most widely distributed integral membrane acyltransferase that uses acyl-phosphate exclusively as the acyl donor. C. trachomatis expresses neither of these proteins, but rather it has a gene predicted to encode a soluble Gro-3-P acyltransferase resembling the enzyme found in plant plastids. The C. trachomatis plsE gene encodes a soluble Gro-3-P acyltransferase related to that found in plant plastids (38% identity, 58% similarity, and an e-value of 2e Ϫ61 to the plastidial Gro-3-P acyltransferase of Cucurbita moschata). This gene was designated plsE to distinguish it from the typical bacterial Gro-3-P acyltransferases. PlsE was expressed using a baculovirus system and purified (Fig. 5A). PlsE was a soluble protein that was a monomer based on its elution position in gel filtration chromatography (Fig. 5B). Purified PlsE catalyzed the conversion of [ 14 C]Gro-3-P into lyso-PA using 16:0-ACP as the acyl donor, but unlike the plant acyltransferase (7), PlsE did not use 16:0-CoA as an acyl donor (Fig. 5C). These data showed that C. trachomatis-specific phospholipid synthesis was initiated by a soluble, acyl-ACP-specific Gro-3-P acyltransferase.
C. trachomatis PlsC contained the HX 4 D motif found in other PlsC proteins and has homology to the S. aureus PlsC, which is also selective for anteiso-15:0 fatty acids (23% identity, 43% similarity, and e-value of 6e Ϫ13 ). The predicted function of C. trachomatis PlsC in PA synthesis was validated by complementation experiments using temperature-sensitive E. coli strain SM2-1 (plsC(Ts)) (54) to analyze PlsC function. The plasmids used were the empty vector and vectors expressing either C. trachomatis, E. coli, or S. aureus plsC. All three plsC genes restored the growth of strain SM2-1 at the nonpermissive temperature, but the empty plasmid did not (Fig. 5, D and E). While this work was underway, the CT807 gene was shown to complement strain SM2-1 (22). These results showed that the C. trachomatis PlsC was a 1-acyl-Gro-3-P acyltransferase that utilized acyl-ACP. PlsC substrate specificity can also be inferred from these experiments. E. coli PlsC selectively utilizes 16:1, an unsaturated fatty acid. Thus, the molecular species of PE produced in strain SM2-1 complemented with E. coli PlsC was 16:0/16:1 (Fig. 5F). In S. aureus, anteiso-15:0 is the primary fatty acid found in the 2-position (50), and complementation of strain SM2-1 with S. aureus PlsC The amounts of radioactivity released into the media and retained in cellular PE were determined by scintillation counting and thin layer chromatography. Data were from quadruplicate biological replicates, and the significance was determined using Student's t test: ***, p Ͻ 0.001. resulted in the formation of 16:0/14:0 PE molecular species (Fig. 5G). S. aureus PlsC incorporated a 14:0 fatty acid into the 2-position because it was the closest structural analogue to anteiso-15:0 (14-carbon chain with a methyl branch) synthesized by E. coli FASII (Fig. 5G). In strain SM2-1 complemented with the C. trachomatis PlsC, the molecular species analysis showed the incorporation of 14:0 as did the strain complemented with S. aureus PlsC (Fig. 5H), indicating that the substrate specificity of C. trachomatis PlsC was similar to S. aureus PlsC. These data were consistent with the predicted role for C. trachomatis PlsC in placing anteiso-15:0 fatty acids into the 2-position during the synthesis of PA.
Phospholipids Associated with Isolated C. trachomatis EB-We compared two established methods for C. trachomatis isolation to address the host lipid contribution to C. trachomatis cell membranes. The first method used a Renografin density gradient centrifugation to isolate C. trachomatis EB (44), and the second employed a Nonidet P-40 detergent extraction step. The detergent extraction was developed to remove extraneous host cellular components, lyse immature RBs, and solubilize connected host membrane systems (46). Although the detergent step removed some C. trachomatis outer membrane proteins, electron microscopy shows that detergent-extracted C. trachomatis retained their intracellular contents and have intact inner and outer membrane systems (45,55,56). Uninfected and C. trachomatis-infected HeLa cell cultures were labeled with [ 14 C]choline to quantify the proportion of host choline-containing phospholipids that were associated with Renografin-and Nonidet P-40-isolated C. trachomatis. The amount of host PC recovered in the C. trachomatis fraction was 1.33 Ϯ 0.08% of the total [ 14 C]PC applied to the Renografin gradient (Fig. 6A). A similar experiment performed with lysates derived from [ 14 C]choline-labeled uninfected HeLa cells showed that 0.24 Ϯ 0.01% of the [ 14 C]PC was located in the gradient fraction that would have contained C. trachomatis. Contamination of the Renografin C. trachomatis preparation with host PC was confirmed by these experiments. C. trachomatis from [ 14 C]choline-labeled C. trachomatis-infected HeLa cells were also isolated by the Nonidet P-40 method. In these experiments, 0.27% of the total [ 14 C]PC was found in the EB fraction, and a much smaller amount of label was recovered from the uninfected controls (Fig. 6A). These data indicated that Nonidet P-40 detergent extraction was more efficient at removing host lipids from C. trachomatis.
The PC molecular species profile from Renografin-isolated C. trachomatis (Fig. 6B) was similar to the profile of host PC (Fig. 3A). In contrast, the phosphocholine-containing molecular species detected in the Nonidet P-40-isolated C. trachomatis were distinctly different. The SM in HeLa cells that was a minor species detected in the scan of total cellular PC (Fig. 3A) was highly enriched in the Nonidet P-40-isolated bacteria (Fig. 6C). The PE molecular species profiles were similar in both the Renografin and Nonidet P-40 isolation methods (Fig. 6, D and  E). The amount of host PC/SM compared with bacterial PE/PG/CL was estimated by labeling infected HeLa cells with [ 14 C]glucose followed by the isolation of C. trachomatis by the Nonidet P-40 method. A thin layer chromatography system was developed to clearly separate these two groups of phospholipids, and the distribution of radioactivity was determined (Fig.  6F). Approximately 15% of the labeled phospholipids associated with isolated C. trachomatis were SM/PC, and 85% were of bacterial origin. One interpretation of these data could be that SM was an integral component of C. trachomatis membranes. However, further corroborating evidence will be required to establish SM as an integral bacterial membrane component because it is possible we were detecting contaminating detergent-insoluble host SM in the detergent-washed C. trachomatis preparation.

Discussion
Our results show C. trachomatis produces the phospholipids required for the formation of its membrane systems using its own biosynthetic pathways described in Fig. 1. The membrane structures of C. trachomatis are morphologically similar to free living Gram-negative bacteria with clearly defined inner and outer membrane systems (3,36,37,57), and C. trachomatis cells possess the gene set found in free-living bacteria like E. coli to generate the phospholipid components. Two differences are the absence of genes for unsaturated fatty acid synthesis and the use of an acyl-ACP-specific plant-like Gro-3-P acyltransferase. As in many bacteria, C. trachomatis encodes a branched-chain ketoacid dehydrogenase and produces a mixture of saturated and branched-chain fatty acids (58). Thus, C. trachomatis FASII is designed to supply the fatty acids detected in the new PE, PG, and CL molecular species produced in C. trachomatisinfected cultures. The removal of host phospholipids may be difficult because C. trachomatis cells are tightly associated with the host inclusion membrane by a structure termed the pathogen synapse (3,36,37). The existence of these intimate C. trachomatis-inclusion membrane interactions may account for the higher level of host lipid associated with C. trachomatis EB in the Renografin density gradient centrifugation isolation method than in the isolation method using a detergent extraction step. Regardless of whether there is a small contribution of host phospholipids to C. trachomatis membranes, it is clear that C. trachomatis uses its own biosynthetic apparatus to produce the bulk of the fatty acids and glycerophospholipids used to construct its membrane systems. Building blocks that are required to assemble these phospholipids (glucose, isoleucine and serine) are readily available from the medium via the host cell.
Our experiments provide analytical data that address the contributions of bacterial and host-derived phospholipids to C. trachomatis membranes. Many experiments with fluorescent probes summarized in recent reviews (13,14) show how host membrane trafficking pathways are hijacked to assemble the inclusion membrane, and the association of these probes with C. trachomatis or the inclusion lumen led to the idea that host phospholipids are used to construct C. trachomatis membranes. These experiments did not quantify the contribution of host lipids to the total and rely on the probes reflecting the behavior of host phospholipid molecular species. SM is the pre- The two cell cultures were lysed, and the C. trachomatis fractions were isolated either using the Renografin density gradient centrifugation or the Nonidet P-40 extraction methods. The amount of [ 14 C]choline that was associated with the C. trachomatis fraction in uninfected cells (blue) compared with C. trachomatisinfected cells (red) was measured by scintillation counting. Data were derived from triplicate biological experiments. B, PC molecular species profile of the C. trachomatis fraction isolated from C. trachomatis-infected cells by Renografin density gradient centrifugation. C, PC molecular species profile of the C. trachomatis fraction isolated from C. trachomatis-infected cells by the Nonidet P-40 method. D, PE molecular species profile of Renografin-purified C. trachomatis. E, PE molecular species profile of C. trachomatis isolated by the Nonidet P-40 method. The spectra shown are representative of duplicate biological experiments. The major molecular species are labeled in the figure panels. F, C. trachomatis-infected HeLa cells were labeled with [ 14 C]glucose for 48 h, and the C. trachomatis was isolated using the Nonidet P-40 method. Lipids were extracted from the C. trachomatis-containing fraction, and the PC/SM fractions were separated from the PE/PG/CL fraction by thin layer chromatography and quantified using the Bioscan Imaging Detector. A representative chromatogram is shown. The inset shows the quantification of the chromatograms from biological triplicates. dominant host phospholipid detected in detergent-extracted C. trachomatis. The detection of SM in the detergent-extracted bacteria coupled with the association of fluorescent ceramide probes with individual C. trachomatis in vivo suggests that SM may be an integral component of C. trachomatis membranes. However, our analytical experiments are perhaps inconclusive because SM is enriched in detergent-extracted mammalian membranes (59 -61), and the enrichment in SM detected in Nonidet P-40-isolated C. trachomatis may arise from contaminating host membranes containing SM that is refractory to detergent extraction. Alternatively, the analytical work may support the incorporation of host SM into C. trachomatis membranes, which may have an unappreciated role in its complex life cycle. Additional research will be needed to evaluate these alternatives.
The idea that C. trachomatis modifies host phospholipids with branched-chain fatty acids by a deacylation-reacylation pathway (17,18,21) is definitively ruled out by our experiments. Mass spectrometry of C. trachomatis-infected cultures shows that neither host PC nor PE is modified with bacterially derived branched-chain fatty acids. The recent report identifying a C. trachomatis gene as a lyso-PC acyltransferase that functions with a host acyl-CoA-binding protein to remodel PC following C. trachomatis infection (22) is also incompatible with the analytical data. The selective degradation of polyunsaturated PE molecular species and the release of arachidonic acid into the medium are consistent with the activation of cPLA 2 observed by others (21,53,62). However, cPLA 2 activation appears to be related to the cellular immunity response to chlamydial infection rather than participating in bacterial phospholipid synthesis. The consistent detection of a minor 18:1/15:0 PE molecular species in the C. trachomatis-infected cultures that does not arise from remodeling host PE indicates that host fatty acids are able to access the C. trachomatis acyltransferases. The activation and utilization of host fatty acids may occur by the acyl-ACP synthetase/2-acylglycerophosphoethanolamine acyltransferase system (CT775 ϩ CT776 ϭ aas). This enzyme system in E. coli activates fatty acids for their transfer to the 1-position of 2-acyl-lyso-PE via an ACP intermediate (9,10), but it does not release acyl-ACP for use by other acyltransferases. Further work will be required to determine whether the C. trachomatis genes produce proteins with the same functions.
There is considerable re-organization of host intracellular membranes and lipid trafficking following C. trachomatis infection to construct the expanding inclusion membrane that grows as C. trachomatis numbers increase and almost reaches the size of the entire cell as the infectious cycle nears completion (63). PC, cholesterol, and SM are trafficked to the chlamydial inclusion to construct this specialized membrane (19,23,24,26). The lower yields of infectious C. trachomatis EB observed when host lipid synthesis/trafficking are inhibited illustrate the importance of host lipid metabolism to maximizing the yield of C. trachomatis (17,19,23,26). The effect of SM inhibitors shows that blocking SM synthesis does not stop C. trachomatis replication but leads to the premature rupture of the inclusion membrane releasing C. trachomatis into the host cytoplasm (23). These data argue that host lipid trafficking is required to maximize the number of infectious EB produced per cell, but it does not impact the fitness of individual C. trachomatis. In contrast, autonomous C. trachomatis phospholipid synthesis is vital to the differentiation, replication, and survival of individual C. trachomatis. The targeted inhibition of the FabI component of FASII arrests C. trachomatis in an RB-like state and prevents replication and the differentiation into infectious EB, illustrating the effectiveness of this approach (5). Thus, the essentiality of autonomous fatty acid and phospholipid synthesis opens the door to the development of new anti-FASII therapeutics to treat C. trachomatis.
Author Contributions-J. Y. and C. O. R. designed the experiments. M. W. F. performed the mass spectrometry, and P.T.C. performed the NMR. J. Y. wrote the first draft, and all authors contributed to the final version of the paper. All authors approved the final version.