Ether Phospholipids and Glycosylinositolphospholipids Are Not Required for Amastigote Virulence or for Inhibition of Macrophage Activation by Leishmania major*

Ether phospholipids are major components of the membranes of humans and Leishmania. In protozoan parasites they occur separately or as part of the glycosylphosphatidylinositol (GPI) anchor of molecules implicated in virulence, such as lipophosphoglycan (LPG), smaller glycosylinositolphospholipids (GIPLs), and GPI-anchored proteins. We generated null mutants of the Leishmania major alkyldihydroxyacetonephosphate synthase (ADS), the first committed step of ether lipid synthesis. Enzymatic analysis and comprehensive mass spectrometric analysis showed that ads1- knock-outs lacked all ether phospholipids, including plasmalogens, LPG, and GIPLs. Leishmania ads1- thus represents the first ether lipid-synthesizing eukaryote for which a completely null mutant could be obtained. Remarkably ads1- grew well and maintained lipid rafts (detergent-resistant membranes). In virulence tests it closely resembled LPG-deficient L. major, including sensitivity to complement and an inability to survive the initial phase of macrophage infection. Likewise it retained the ability to inhibit host cell signaling and to form infectious amastigotes from the few parasites surviving the establishment defect. These findings counter current proposals that GIPLs are required for amastigote survival in the mammalian host or that parasite lyso-alkyl or alkylacyl-GPI anchors are solely responsible for inhibition of macrophage activation.

The surface of the Leishmania parasite is a major point of interaction with the host throughout the infectious cycle, which includes an extracellular promastigote form residing in the midgut of sand flies and an intracellular amastigote form adapted for survival within the phagolysosomes of vertebrate macrophages. Glycosylphosphatidylinositol (GPI) 1 -anchored molecules dominate the parasite surface, and many of these have been implicated in the ability of the parasite to survive in such hostile environments (1). Abundant surface molecules include lipophosphoglycan (LPG, containing 15-30 copies of a phosphoglycan (PG) repeating unit), GPI-anchored proteins such as membrane proteophosphoglycans, gp63 and gp46, and a heterogeneous group of small glycosylinositolphospholipids (GIPLS; for reviews, see Refs. 2 and 3).
LPG, GIPLs, and related molecules have been shown to inhibit activation and signaling when applied exogenously to macrophages (4 -7). The expression of LPG and GPI-anchored proteins decreases greatly in intracellular amastigotes, while GIPLs remain at high levels (2,8). These and other data have led to proposals that GIPLs are key molecules for survival of amastigotes within the macrophage phagolysosome (9 -11).
A shared structural motif of GPI-anchored glycoconjugates as well as other abundant phospholipids in Leishmania is the presence of ether lipids. The lipid moieties of GIPLs and GPIanchored proteins are sn-1-alkyl-2-acyl-PIs (for reviews, see Refs. [1][2][3] while that of LPG is sn-1-alkyl-2-lyso-PI where the alkyl group is C 24 -26 (12,13). The GPI anchors of GIPLs and LPG with its very long chain alkyl group have been implicated in down-regulation of host cell responses, for example in the inhibition of protein kinase C and nitric oxide (NO) production (5, 14 -16). At least 10% of the total membrane lipids in Leishmania consist of ether lipids within GPI anchors or in phospholipids including phosphatidylethanolamine (PE), phosphatidylcholine (PC), and phosphatidylserine (17)(18)(19). Ether lipid analogs have strong inhibitory activity against parasites exemplified by the recent introduction of miltefosine into clinical usage against Leishmania donovani (for a review, see Ref. 20). In mammals ether phospholipids constitute nearly 20% of total phospholipids, occurring most commonly as 1-O-alkenyl-2-alkyl-sn-glycero-3-phosphocholine or phosphoethanolamine plasmalogens (for a review, see Ref. 21). Plasmalogens have been implicated in signal transduction, membrane fusion and trafficking, oxidant resistance, and prostaglandin synthesis (for a review, see Ref. 21).
To focus on the question of the role of the lipid moiety of GPI anchors in the infective cycle, we took a genetic approach targeting alkyldihydroxyacetonephosphate synthase (ADS, EC 2.5.1.26), the first committed step in ether lipid synthesis. In Leishmania as in other organisms, this pathway begins with the acylation of dihydroxyacetonephosphate (DHAP) followed by replacement of the acyl group with an alkyl group and then reduction to give the ether lipid precursors 1-alkylglycerol-3phosphate (22,23). In mammals this pathway occurs in peroxisomes, while in trypanosomatids it occurs in glycosomes, a microbody variant of peroxisomes (for reviews, see Refs. 22 and 24). ADS activity has been characterized, and its gene has been identified in a number of species including Trypanosoma brucei (22,23,25,26). We used this information to identify the Leishmania gene ADS1 and to create null mutants to assess the role of ether lipids in parasite metabolism and virulence.
Molecular Biology-All enzyme restriction reactions, PCR, and DNA and RNA extractions and hybridizations (Southern blot, Northern blot, and colony hybridization) were performed by standard methods (29). DNA hybridization probes were made using a random priming kit (Roche Applied Science). The gene D/SHERP probe was obtained by PCR from genomic DNA with primers Y09088 137-156 (5Ј-GATCCGC-GCAGACCAAGATG) and Y09087 320 -300 (5Ј-CAGAGAACGGCGAA-GGGACTG). For reverse transcriptase PCR, cDNA was prepared from RNA isolated from logarithmic phase promastigotes using random primers for the reverse transcription reaction (Invitrogen) and used as template for PCR with an oligonucleotide specific for the miniexon (SMB936, 5Ј-ACCGCTATATAAAGTATCAGTTCTGTACTTTA) and one specific for ADS1 (either SMB1012, 5Ј-ACTAGTGCTGTCCTGTGTTT-TATCG, located in the 5Ј untranslated region, or SMB1017, 5Ј-ATCT-GCATCTGGACATCC, located within the ADS1 coding region). These products were directly sequenced by the chain termination method.
Cloning of ADS1 of L. major and Plasmid Construction-A 1.58-kb fragment was obtained following PCR with L. major genomic DNA template and the ADS1 degenerate primers SMB929 (5Ј-AAGTGGAAY-GGNTGGGG) and SMB931 (5Ј-CCSATNCCGTGGTGGTGNGT). This was inserted into pCR2.1 (Invitrogen) to give pCRII.DHAP-PCR (strain B3772) and sequenced. This PCR product was used to screen a L. major Friedlin strain V1 cLHYG cosmid library (30) by colony hybridization. Positive clones were further analyzed by restriction enzyme digestion followed by Southern blot analysis probed with the 1.58-kb PCR product. A 3.6-kb SphI fragment containing ADS1 was cloned into pUC19 yielding pUC.DHAP.Sp1 (B3793) and sequenced on both strands.
Enzyme Assays-ADS activity was measured in an enriched glycosome fraction using slight modifications of the protocol by Heise et al. (22). Cells were resuspended in 7 ml containing 0.25 M sucrose, 1 mM EDTA, and 25 mM Tris-HCl, pH 7.8 (STE buffer) and disrupted by nitrogen cavitation (1500 -2000 p.s.i. for 30 min at 4°C). Cellular debris were removed by centrifugation at 1000 ϫ g for 10 min, and the supernatant was briefly sonicated. The sample was mixed with 80% Percoll, 20% STE buffer to a final volume of 40% Percoll and centrifuged at 70,000 ϫ g in a Beckman NVI90 rotor for 30 min at 4°C. Glycosome fractions banding at a density of 1.09 -1.1 g/ml were taken. Protein concentration was determined according to the bicinchoninic acid assay with a bovine serum albumin standard. Alkyl-DHAP synthase activity was measured in triplicate as described previously with slight modifications (26). Briefly, 0.4 mg of glycosomal proteins were incubated in 100 l containing 50 mM potassium phosphate, pH 8.0, 100 M [ 14 C]hexadecanol, 90 M palmitoyl-DHAP, 1 mM dithiothreitol, 50 M NaF, 0.1% Triton X-100, 2 g/ml pepstatin, 2 g/ml leupeptin, and 10 g/ml trypsin inhibitor at 37°C for 40 min and extracted with chloroform:methanol (1:2, v/v) (35). The organic phase was dried under nitrogen, spotted onto silica gel 60 plates, and resolved in solvent chloroform, methanol, acetic acid, 5% sodium bisulfite (100:40:12:4, v/v/v/v); bands were visualized with Lester reagent (36), and the radioactivity in bands with ϳR F of 3.5 was measured by scintillation counting.
For secretory acid phosphatase assay cell supernatants were loaded on a non-denaturing polyacrylamide gel in the absence of SDS (37); separating and stacking gels were 6 and 3% acrylamide, respectively. Acid phosphatase was visualized as described previously (38).
For transmission electron microscopy, parasites were washed in 0.1 M cacodylate buffer, pH 7, and fixed in 2.5% glutaraldehyde (electron microscopy grade, Sigma) in cacodylate buffer containing 5 mM CaCl 2 , 5% sucrose, and 0.15% ruthenium red (Electron Microscopy Sciences, Ft. Washington, PA) for 1 h at 4°C. Following three washes in cacodylate buffer, parasites were postfixed in 1% osmium tetroxide (Polysciences Inc., Warrington, PA), 2 mM CaCl 2 , 0.15% ruthenium red in cacodylate buffer for 1 h at room temperature. The samples were then rinsed extensively in cacodylate buffer, 3% sucrose and deionized water prior to staining with 1% uranyl acetate. Following several rinses in deionized water, samples were dehydrated in a graded series of ethanol and embedded in Spurrs resin (Electron Microscopy Sciences). Sections of 70 -80 nm were cut, stained with lead citrate, and viewed on a Zeiss 902 transmission electron microscope.
Phospholipid, GIPL, and LPG Analysis-Bulk phospholipids and GIPLs were purified and analyzed by electrospray mass spectrometry (ES-MS) in positive and negative ion modes as described previously (10,42). Phospholipids were dissolved in chloroform:methanol (1:2, v/v) (approximately 2 ϫ 10 6 cell eq/l) and introduced into the electrospray source of the mass spectrometer (Quattro triple quadrupole, Micromass) at 5 l/min with a syringe pump. Parameters were as follows: skimmer/cone offset 5 V; capillary, high voltage lens, and cone voltages, 3 kV, 0.5 kV, and 25 V, respectively for positive ion mass spectra and 3 kV, 0.5 kV, and 45 V for negative ion mass spectra. CID spectra were achieved in a hexapole collision cell containing argon (2.5 ϫ 10 Ϫ3 torrs). Parent ion spectra were taken using the following accelerating voltages into the collision cell: parents of   6 h. PG repeat structures were determined from LPG (ϳ50 pmol in 5% propan-1-ol, 5 mM ammonium acetate) by ESI-MS as described previously (43).
Detergent-resistant Membrane (DRM) Preparation and Analysis-Crude membranes were prepared from 10 8 log phase parasites, and DRMs were isolated on a step gradient after extraction with 1% Triton X-100 at 4°C as described previously (44). Fractions were then taken from the top of the gradient and analyzed by Western blotting.
Mouse and Macrophage Infections-Mice were injected with promastigotes that had been in stationary phase 4 days (1 ϫ 10 6 ) or amastigotes (1 ϫ 10 5 ) subcutaneously into footpads, and lesion size and parasite numbers were measured (41,45). In vitro infection of peritoneal macrophages was performed as described previously (41,46). Metacyclic parasites were purified on a Ficoll gradient (28), and crude amastigotes were prepared from lesions (41). Briefly, excised lesions were placed in cold Dulbecco's modified Eagle's medium and dissociated with a Dounce homogenizer, tissue fragments and intact host cells were removed by a low speed centrifugation at 200 ϫ g, and amastigotes were pelleted at 1800 ϫ g. Amastigotes were washed twice in cold Dulbecco's modified Eagle's medium and then counted with a hemocytometer. For

RESULTS
Characterization of L. major ADS1-From known ADS sequences we designed degenerate oligonucleotide primers for amplification of the L. major gene and recovered the parasite gene by screening a L. major cosmid library. Southern blot analysis showed that the Leishmania ADS1 gene was single copy (Fig. 1A), which was confirmed by targeted gene replacement (described below, Fig. 1D). The sequence of the ADS1 gene predicted a protein of 621 amino acids (GenBank TM accession number AY328521) with strong homology to the ADS proteins of T. brucei (59% identity) and other eukaryotes including mammals, Caenorhabditis elegans, and Dictyostelium (30 -37% identity). ADS1 lacked obvious transmembrane domains but contained at its C terminus the sequence SHI, which resembled typical type 1 peroxisomal targeting signals (PTS1, Ref. 48). We created an N-terminal GFP-ADS1 fusion protein, which was shown to be enzymatically active following transfection into ads1 Ϫ null mutants (below and data not shown). Fluorescence microscopy showed that GFP-ADS1 fusion protein was correctly targeted to the Leishmania glycosome (analogous to peroxisomes of other creatures) and co-localized with the glycosomal marker hypoxanthine guanine phosphoribosyltransferase ( Fig. 1E and Ref. 39).
In trypanosomatids mRNAs bear a 39-nucleotide 5Ј-terminal exon added by trans-splicing (49). Reverse transcriptase PCR with miniexon and ADS1-specific primers mapped the splice acceptor site to nucleotide position Ϫ355 relative to the predicted ADS1 initiation codon. Northern blot analysis showed that ADS1 mRNA was expressed at similar levels in all developmental stages as two transcripts of 4.6 and 6.6 kb (Fig. 1B).
Generation of an ads1 Ϫ Knock-out-An ads1 Ϫ null mutant (⌬ADS1::SAT/⌬ADS1::HYG) was generated by two rounds of gene replacement since Leishmania is an asexual diploid (Ref. 31 and see Fig. 1, C and D, for predicted structures of replacements and Southern blot confirmation). These were recovered at typical frequencies, appeared morphologically to be normal, and grew well in culture with a doubling time of 8.1 versus 7 h for wild type (WT, Table I). As a control "add-back" line for subsequent studies, several ads1 Ϫ mutants were transfected with the ADS1-expressing plasmid pXG-ADS1, yielding lines termed ads1 Ϫ /ϩADS1. ads1 Ϫ differentiated normally to the infective metacyclic stage (Table I) where it expressed stagespecific markers such as SHERP (Fig. 2B).
ADS activity was measured from enriched glycosomal preparations. WT showed high levels of activity, while ads1 Ϫ showed only background levels, and ads1 Ϫ /ϩADS1 restored ADS activity to WT levels ( Fig. 2A). Thus ADS1 was responsible for cellular ADS activity.
In the ads1 Ϫ cells the positive ion spectrum was markedly different (Fig. 3A, middle panel). While the lyso-acyl-and diacyl-PC species remained, the alkylacyl-PEs were absent. These phospholipids were all restored in ads1 Ϫ /ϩADS1 (Fig. 3A,  lower panel), establishing their dependence on ADS1.
Negative ion ES-MS and ES-MS-CID-MS were used to identify PI species (Fig. 3B, upper panel) identity of the ion m/z 552 is unknown. The negative ion spectra of ads1 Ϫ phospholipids (Fig. 3B, middle panel) lacked the alkylacyl-PI ions (at m/z 849 and 933) but contained four unidentified (non-inositolphospholipid) ion clusters at m/z 479, 504, 554, and 604. The spectrum for the ads1 Ϫ /ϩADS1 phospholipids (Fig. 3B, lower panel) showed restoration of the alkylacyl-PI ions and disappearance of the unidentified ions. As seen in the positive ion mode, these data established that lack of ADS1 leads to complete loss of expression of ether phospholipids.
Interestingly the ads1 Ϫ /ϩADS1 GIPL fraction also showed ions at m/z 1156 and 1318, whose ES-MS-CID-MS spectra identified them as lyso-iM2 and lyso-GIPL-1 species, respectively, with C24:0 alkyl chains. These latter species were LPG precursors that are barely detectable in WT cells (51), suggesting that overexpression of ADS1 in the ads1 Ϫ /ϩADS1 line may lead to their overproduction.
The ads1 Ϫ Mutant Synthesizes Normal Amounts of GPIanchored Proteins-We examined the synthesis of the abundant alkylacyl-GPI-anchored proteins gp46/PSA-2 and gp63 by immunoblots and flow cytometry. In contrast to the results obtained with GIPLs and ether phospholipids, the WT, ads1 Ϫ , and add-back strains synthesized similar levels of gp46 (Fig.  5A). gp46 was localized on the plasma membrane in ads1 Ϫ where it showed enhanced fluorescence (Fig. 5B). This reflected the absence of LPG as seen previously (53) and as discussed below. Similar studies showed that gp63 steady state expression was not affected in the ads1 Ϫ mutant (data not shown). Efforts to determine the structure of the gp63 anchor in ads1 Ϫ were not successful (data not shown).
The ads1 Ϫ Mutant Lacks LPG-Metabolic labeling of WT, ads1 Ϫ and ads1 Ϫ /ϩADS1 parasites with [ 3 H]Gal followed by extraction and purification of the LPG fraction showed that ads1 Ϫ contained about 7% of WT levels (Fig. 6A). LPG was visualized by Western blot analysis with the anti-PG monoclonal antibody WIC79.3 (54). LPG from WT and ads1 Ϫ / ϩADS1 lines migrated as a smear with an apparent molecular mass of 30 -60 kDa both from cells and when shed into culture FIG. 6. Phosphoglycan expression and glycocalyx structure in WT and ads1 ؊ L. major. A, LPG was metabolically labeled with [ 3 H]Gal, purified, and quantified by liquid scintillation counting. Val-ues are given relative to WT, and the S.D. is shown. B, whole cell extracts (1 ϫ 10 6 cells) or cell supernatants (30 l/lane) of a mid-log (5 ϫ 10 6 /ml) culture from WT, ads1 Ϫ , and ads1 Ϫ /ϩADS1 cells were separated by SDS-PAGE and subjected to immunoblotting with the antiphosphoglycan antibody WIC79.3. Protein marker (kDa) is indicated. C, membrane proteophosphoglycans (mPPG) show decreased mobility in ads1 Ϫ consistent with increased phosphoglycosylation. Proteophosphoglycans were resolved in the stacking gel by discontinuous 4%/10% SDS-PAGE and subjected to Western blotting with WIC79.3; the arrow marks the border between the stacking and separating gel. D, phosphoglycosylation of an L. donovani sAP PG "reporter" is elevated in the ads1 Ϫ line. Culture supernatants from L. major WT, ads1 Ϫ , or ads1 Ϫ / ϩADS1 lines transfected with either pX63PAC (denoted by Ϫ) or pX63PAC-LdSAcP-1 (denoted by ϩ) were separated by native PAGE and stained for acid phosphatase activity. E, transmission electron microscopy of the surface glycocalyx of WT, ads1 Ϫ , and ads1 Ϫ /ϩADS1 cells stained with ruthenium red for carbohydrate. The bar corresponds to 0.1 m. supernatants (Fig. 6B). ads1 Ϫ contained significantly less WIC79.3-reactive material (termed LPG-like), which had reduced electrophoretic mobility (50 -100 kDa) and was not secreted into the medium (Fig. 6B). In contrast, Western blots showed similar levels of proteophosphoglycan in both WT and ads1 Ϫ (Fig. 6C), again with that of ads1 Ϫ showing reduced mobility. Analysis of the PG repeat structures by cone voltage fragmentation ES-MS (43) of LPG from WT and ads1 Ϫ /ϩADS1 parasites and the LPG-like fraction of ads1 Ϫ showed that they were identical (data not shown). Attempts to determine the structure of a possible PI lipid component in purified preparations of the ads1 Ϫ LPG-like material by negative ion ES-MS analysis were unsuccessful.
LPG comprises a major portion of the parasite surface glycocalyx (55). Correspondingly transmission electron microscope analysis of plasma membranes stained with ruthenium red showed the typical LPG-rich glycocalyx for WT and ads1 Ϫ / ϩADS1 parasites but relatively little staining in the ads1 Ϫ mutant (Fig. 6E). The absence of the surface glycocalyx in ads1 Ϫ conferred increased sensitivity to lysis by human complement (Table I) and labeling with antisera to GPI-anchored proteins (Fig. 5B).
Phosphoglycosylation Is More Extensive in ads1 Ϫ Mutants-The slower electrophoretic mobility of PG-containing molecules in ads1 Ϫ relative to WT suggested that they could be more extensively phosphoglycosylated (for a review, see Ref. 56). To test this, we used the L. donovani secretory acid phosphatase (sAP) gene SAcP-1 as a "phosphoglycosylation reporter" following transfection into L. major. sAP lacks a GPI membrane anchor, and L. major normally expresses very low levels (57). In situ enzymatic activity assays showed that WT and the ads1 Ϫ /ϩADS1 SAcP-1 transfectants expressed sAPs with identical electrophoretic mobility, while ads1 Ϫ transfectant sAPs migrated more slowly; as expected, vector transfectant controls lacked detectable sAP (Fig. 6D). This suggested that the degree of phosphoglycosylation was elevated in ads1 Ϫ .
Increased phosphoglycosylation occurs during Leishmania development with higher numbers of PG repeats found in the LPG of metacyclic promastigotes (8,58). However, metacyclogenesis and the stage-specific expression of the metacyclic marker SHERP were normal in ads1 Ϫ (Table I and Fig. 2B), suggesting that differentiation was not altered in this line.
ADS1 Is Not Required for Formation of Lipid Rafts (DRMs)-We asked whether the absence of ether lipids leads to alterations in the formation of "lipid rafts" as defined by buoyant DRM criteria. DRMs were prepared by standard procedures from log phase parasites and further separated on density gradients, and the distribution of LPG and gp63 into a buoyant fraction expected for lipid rafts was assessed. As previously observed, in WT L. major the GPI-anchored protein gp63 resides in a buoyant fraction (Fig. 7, upper left panel,  fraction 2), whereas the majority of cellular material was found in dense fractions toward the bottom of the gradient (data not shown, Refs. 44 and 59). Similar results were obtained with ads1 Ϫ (Fig. 7, upper right panel), showing that DRM formation was not altered. In contrast, LPG did not reside in buoyant DRMs in either WT or ads1 Ϫ parasites (Fig. 7, lower panels). This conflicted with results reported previously in WT parasites (44) where LPG did show enrichment in the buoyant DRM fraction. This discrepancy was shown to reflect differences in the methods used in the previous study, which examined pulselabeled cells treated with tunicamycin rather than steady state levels in the unperturbed cells as studied here. 2 Regardless the key finding was that DRMs did not differ between WT or ads1 Ϫ parasites.
ADS1 Is Important for Establishment of Infections in Mice and Macrophages-Following inoculation of stationary phase promastigotes into susceptible BALB/c mice, WT parasites formed lesions that appeared after ϳ15 days and progressed rapidly, while their appearance in ads1 Ϫ was delayed until 40 days and progressed somewhat more slowly thereafter (Fig.  8A). Lesion size correlated with the parasite burden (data not shown), and the ads1 Ϫ /ϩADS1 line induced lesions in a manner similar to WT (Fig. 8A). ads1 Ϫ amastigotes were recovered from mice showing lesions around day 70, allowed to differentiate back to promastigotes, and reinoculated into animals, yielding identical results (data not shown). This showed that the delayed lesion appearance was not due to the presence of revertants or contaminants.
Macrophage infections were performed with stationary stage promastigotes parasites opsonized with C5-deficient mouse serum. While ads1 Ϫ parasites were taken up into macrophages somewhat better than WT (as seen previously in other LPGdeficient lines, Refs. 41 and 60), nearly 95% perished within 2 days of infection versus 30 -50% for WT or the add-back strain (Fig. 8B, top panel). The extent of destruction was higher than seen previously with an LPG-deficient lpg1 Ϫ mutant (41,61). Quantitation of the few surviving ads1 Ϫ parasites showed that they went on to replicate albeit at about 50% the rate seen in the WT or ads1 Ϫ /ϩADS1 (Fig. 8B, lower panel).
ADS1 Is Not Required for Replication or Infection of Amastigotes-The promastigote infections of macrophages and mice suggested that while defective in establishment of infection those ads1 Ϫ parasites that escaped initial destruction during the "establishment phase" were able to survive and propagate as amastigotes. In this aspect ads1 Ϫ resembled the specifically LPG-deficient mutant lpg1 Ϫ (41,61) where this was expected since the amastigote stage normally lacks LPG. Extrapolation to ads1 Ϫ would argue that ether lipids are not essential for replication as amastigotes.
To test this, ads1 Ϫ amastigotes were purified from mouse lesions similar to those shown in Fig. 8A and then used to initiate macrophage infections directly. In contrast to the re-2 P. Denny and D. F. Smith, unpublished. Cold Triton X-100-extracted DRMs from WT and ads1 Ϫ (left and right panels, respectively) parasites were fractionated on a density gradient (fractions 1-6, low to high density) and subjected to Western blotting with antisera against gp63 (top panels) or LPG (WIC79.3, lower panels). The size of the protein makers is indicated in kDa.
sults obtained with stationary phase promastigote infections, ads1 Ϫ amastigotes entered and survived well in macrophages and went on to replicate well albeit again at about one-third the rate seen in WT or ads1 Ϫ /ϩADS1 ( Fig. 8D and Table I). Similar findings were obtained in mouse infections by amastigotes (data not shown).
ads1 Ϫ Retained the Ability to Inhibit Macrophage NO Production-Purified metacyclic parasites were used to infect two sets of peritoneal macrophages; after 6 h, interferon-␥ and lipopolysaccharide were added to one set of infected macrophages, and NO production was determined. As expected, infections with WT L. major did not induce NO synthesis, and these parasites inhibited macrophage NO production by 70% following treatment with activators (Fig. 8C); similar results arose with the control ads1 Ϫ /ϩADS1. Surprisingly ads1 Ϫ showed the same profile even though the overwhelming majority of these parasites were destroyed by macrophages (Fig. 8, B  and C). DISCUSSION In this work we used multiple approaches to establish that the Leishmania alkyl-DHAP synthase encoded by ADS1 is required for all cellular ether phospholipid synthesis and to explore its role in parasite biology. Despite the fact that more than 10% of Leishmania cellular lipids are comprised by ether phospholipids, the ads1 Ϫ null mutant was viable. Enzymatic studies showed that ADS activity was absent in ads1 Ϫ , and biochemical studies showed that all known ether phospholipid species were lacking. These included alkenylacyl-PEs as well as GPI-anchored molecules such as LPG and GIPLs. In this respect ads1 Ϫ differs from previously identified mammalian mutants defective in ether lipid biosynthesis that are typically "leaky" and express residual levels of plasmalogens (21). While GPI-anchored proteins were retained at the parasite surface at normal levels, it seems likely that they now contain an alternative lipid anchor, probably diacylglycerol (as seen in trypano- some variant surface glycoproteins), although efforts to confirm this were inconclusive. Since no other specific alterations in lipid composition were seen in ads1 Ϫ , we presume that that a modest up-regulation of remaining membrane lipid species compensated for the general lack of ether phospholipids.
Remarkably the complete loss of ether phospholipids, LPG, and GIPLs was accompanied by minimal changes in many aspects of parasite biology: for example, only a modest reduction in growth rate was observed in vitro as promastigotes or as amastigotes within macrophages in vivo. A secondary phenotype was increased phosphoglycosylation of endogenous or reporter proteins in the ads1 Ϫ mutant. Its basis was not sought; it may arise from small differences in membrane vesicular trafficking dependent upon ether lipids as seen in mammalian cells (62,63).
Previous studies have shown that ether lipids can associate and potentially contribute to the stability of membrane microdomains commonly termed lipid rafts, which are also rich in sterols and sphingolipids (64 -68). However, lipid rafts (as defined by DRM criteria) were maintained in ads1 Ϫ in the absence of the 10% of cellular lipids comprised by Leishmania ether phospholipids, possibly reflecting the ameliorating abundance of sphingolipids and especially ergosterol in the parasite membrane (17,19). It should be emphasized that retention of DRMs does not necessarily imply that the lipid rafts remaining in ads1 Ϫ are identical to those of WT; to address this, additional markers for parasite rafts will have to be identified and examined in the future.
The synthesis of the major classes of GPI-anchored molecules has been proposed to diverge from a common Man-GlcN-PI precursor (3,69) with LPG and type 2 GIPLs requiring the formation of Man(␣1-3)Man-GlcN-PI and protein GPI anchors requiring the formation of Man(␣1-6)Man-GlcN-PI. The loss of LPG/GIPL but not protein GPI anchor synthesis in the ads1 Ϫ line suggests that the GDP-Man:Man-GlcN-PI (␣1-3) mannosyltransferase may be dependent on Man-GlcN-PI acceptors that contain sn-1-alkyl-2-acyl-PI, whereas the dolicholphosphoryl-Man:Man-GlcN-PI (␣1-6) mannosyltransferase may also function with diacyl-PI-containing acceptors. Consistent with this model, [ 3 H]GlcN labeling studies have shown that GlcN-diacyl-PIs are synthesized by L. major promastigotes (70), providing potential non-ether lipid substrates for protein GPI anchor synthesis, and cell-free GPI biosynthesis studies have suggested that the GDP-Man:Man-GlcN-PI (␣1-3) mannosyltransferase may not act on Man-GlcN-diacyl-PI (3). There are precedents for lipid specificity/selectivity for certain mammalian GPI biosynthetic enzymes, although GPI anchor synthesis in T. brucei shows a relaxed specificity for the composition of the lipid anchor (71,72). Future studies may explore the lipid specificity of LPG/GIPL and protein GPI anchor synthesis in L. major in more detail.
The abundance of GIPLs in the intracellular amastigote stage (3,73) and the ability of purified GPI anchors and GIPLs to modulate key signaling pathways implicated in parasite survival in macrophages (5,7) led to proposals that these molecules play major roles in parasite virulence. However, genetic studies of GIPL function in Leishmania mexicana have yielded contradictory results possibly because they were based upon mutants with broad and complex effects beyond GIPL synthesis (9,11,74,75). Moreover L. mexicana and L. major differ greatly in their dependence upon LPG and phosphoglycans for virulence probably due to interactions with the host immune response (76). The L. major ads1 Ϫ mutant studied here affected a defined set of parasite ether phospholipids in a species in which for both LPG and PGs the general roles in virulence and the specific mechanisms by which these act have been defined genetically (41,61).
Remarkably the phenotype seen for L. major ads1 Ϫ was nearly indistinguishable from the LPG-deficient mutant lpg1 Ϫ : both showed increased sensitivity to lysis by complement due to disruption of the glycocalyx, normal metacyclogenesis, increased destruction following macrophage infection, and delayed lesion appearance (Table I and Figs. 2B and 8 and Refs.  41 and 61). That the ads1 Ϫ phenotype included the lpg1 Ϫ phenotype was not surprising given that it lacks LPG, but that the absence of both GIPLs and ether phospholipids conferred little additional effect was unexpected especially in the amastigote stage. In contrast, the globally PG-deficient L. major mutant lpg2 Ϫ was unable to establish macrophage infections at all and did not induce pathology in mouse infections (60). We conclude therefore that neither GIPL nor ether phospholipid synthesis is uniquely required for amastigote growth and survival in macrophages in vitro or in mouse infections in vivo.
Similarly we found that despite its attenuated ability to establish infections in macrophages, the ads1 Ϫ mutant did not induce NO following entry and inhibited the ability of macrophages to make NO following the strong induction signal of interferon-␥ plus lipopolysaccharide (Fig. 8C). This was remarkable given reports that the alkylglycerol anchor of LPG or the alkylacyl anchor of GIPLs can inhibit macrophage activation pathways leading to the activation of protein kinase C and/or the formation of NO or interleukin-12 (4 -7). There are a number of potential explanations. One is that the highly purified GPIs studied previously contained traces of an active, non-GPI species. Similar problems were encountered in assessing the antigenicity of purified LPG preparations (77). A second possibility is that differences in the amount, form, or delivery route of GPI-anchored molecules tested in vitro does not closely mimic what occurs in natural infection in vivo. A third possibility invokes redundancy of GIPL functions with those of other parasite molecules as it seems likely during evolution that selection for multiple mechanisms ensuring repression of macrophage activation would be advantageous. Reasonable candidates for this role might be other parasite glycolipids such as sphingolipids or protein-GPI anchor moieties remaining in ads1 Ϫ , and the possibility of molecules other than glycolipids cannot be excluded.