Identification of the defect in lipophosphoglycan biosynthesis in a non-pathogenic strain of Leishmania major.

The major macromolecule on the surface of the protozoan parasite, Leishmania major, is a complex lipophosphoglycan (LPG), which is anchored to the plasma membrane by an inositol-containing phospholipid. A defect in LPG biosynthesis is thought to be responsible for the avirulence of the L. major strain LRC L119 in mice. In order to identify the nature of this defect we have characterized two truncated forms of LPG, which are accumulated in this strain, by one- and two-dimensional 500-MHz 1H NMR spectroscopy, two-dimensional heteronuclear 1H-31P NMR spectroscopy, methylation analysis, and exoglycosidase digestions. The structures of these glycoinositolphospholipids, termed GIPL-4 and -6, are as follows: [formula: see text] The glycan moieties of GIPL-4 and -6 are identical to the anchor region of LPG, which is also substituted with a Glc-1-PO4 residue in approximately 60% of the structures. However, instead of being capped with chains of phosphorylated oligosaccharide repeat units, both glycan moieties terminate in Man alpha 1-PO4, suggesting that the defect in LPG biosynthesis is in the transfer of galactose to this residue to form the disaccharide backbone of the first repeat unit. These results indicate that the phosphoglycan moiety of LPG is essential for intracellular survival of the parasite and have implications for LPG biosynthesis.

The glycan moieties of GIPL-4 and -6 are identical to the anchor region of LPG, which is also substituted with a Glc-1-P04 residue in approximately 60% of the structures. However, instead of being capped with chains of phosphorylated oligosaccharide repeat units, both glycan moieties terminate in Manal-POr, suggesting that the defect in LPG biosynthesis is in the transfer of galactose to this residue to form the disaccharide backbone of the first repeat unit. These results indicate that the phosphoglycan moiety of LPG is essential for intracellular survival of the parasite and have implications for LPG biosynthesis.
The etiological agent of human cutaneous leishmaniasis, Leishmania mujor, is a parasitic protozoan that alternates between an extracellular, flagellated promastigote stage in the digestive tract of its sandfly vector and an intracellular amastigote stage within the phagolysosome compartment of mammalian macrophages. The cell surface of the promastigote stage is coated by a complex glycocalyx which contains two novel classes of glycoconjugate; the heterogeneous lipophos- phoglycans (LPGs)' (Handman and Goding, 1985;Mc-Conville et al., 1987, 1990a and the low molecular weight glycoinositol-phospholipids (GIPLs) (McConville and Bacic, 1989;McConville et al., 1990b). LPG consists of a polymer of repeating phosphorylated oligosaccharides, containing the backbone sequence PO4-6Gal~l-4Mancul-, where the 3-position of the Gal residue may be substituted with side chains of Gal, Glc, and Ara. These chains are attached to the plasma membrane by a complex glycolipid which contains a hexasaccharide core and a lyso-alkylphosphatidylinositol (lyso-alkyl-PI) lipid moiety (McConville et al., , 1990a. The GIPLs are structurally similar to the LPG anchors in having the same glycan core sequence, but differ from these anchors in containing predominantly alkylacyl-PI lipid moieties (Mc-Conville and Bacic, 1989;McConville et al., 199Ob).
Studies with LPG-deficient strains of Leishmania have provided evidence that LPG is essential for promastigote infectivity in the mammalian host (Handman et al., 1986;Elhay et al., 1990;McNeely and Turco, 1990). These strains are non-pathogenic in mice and are rapidly killed in the phagolysosome of in vitro infected macrophages. Importantly, intracellular survival of these strains can be prolonged if exogenous LPG is inserted into the promastigote plasma membrane (Handman et al., 1986;McNeely and Turco, 1990). In this regard, LPG is thought to prevent complement-mediated lysis of promastigote in the bloodstream of the host (Puentes et al., 1988), to be involved in mediating the initial attachment of promastigotes to the macrophage (Handman and Goding, 1985;Puentes et al., 1988;da Silva et al., 1989;Talamas-Rohana et al., 1990), and to protect the parasite membrane from hydrolytic enzymes and the oxidative burst in the phagolysosome (El-On et aL, 1980;Chan et al., 1989;McNeely and Turco, 1990). We have previously shown that the LPG-deficient strain of L. major, LRC-L119 (L119) accumulates a number of polar GIPLs which may be truncated forms of LPG (McConville and Bacic, 1989Bacic, , 1990. We have now characterized these glycolipids in order to identify the defect in LPG biosynthesis and to determine which portions of LPG are still expressed in this strain. These studies indicate that L119 has a mutation effecting the galactosyltransferase which is involved in forming the backbone sequence of the repeat units. They also refine the structure of the L. major LPG anchor and have implications for LPG biosynthesis.

EXPERIMENTAL PROCEDURES AND RESULTS~
The GIPL profiles of the virulent, LPG-producing L. major strain V121 and the avirulent, LPG-deficient strain L119 are shown in Fig. 1. As reported earlier (McConville and Bacic, 1990;McConville et al., 1990a), both strains contain similar levels of GIPL-1, -2, and -3. However, L119 differs notably from V121 in containing three additional species of GIPL (GIPL-4, -5, and -6) which have a slow HPTLC mobility. These species account for approximately 50% of the L119 GIPLs and were not detectable in the V121 profiles. Two of these species (GIPL-4 and -6) were purified by a combination of octyl-Sepharose chromatography and HPTLC. From the yields of myo-inositol in each fraction, and assuming 1 inositol residue/molecule, there are approximately 5 X lo6 GIPL-6 and 2 X lo6 GIPL-4 molecules per cell. The glycan head groups of these GIPLs were characterized as described below.
Structure of the GIPL-6 Glycan-The glycan moiety of GIPL-6 was released with PI-specific phospholipase C and analyzed by one-dimensional 'H 500-MHz NMR and by twodimensional 'H-'H correlated spectroscopy (COSY). Eight anomeric protons of unit intensity were observed in the low field region of the spectrum (Fig. 2a), indicating the presence of eight monosaccharide residues in GIPLB, consistent with the monosaccharide analysis ( Table I). Six of these proton resonances bear a strong resemblance in both splittings and shifts (Table 11) to those observed in the 'H NMR spectrum of the hexasaccharide-inositol phosphate moiety derived from GIPL-3 (McConville et al., 1990b). In particular, the throughbond connectivities in the 'H COSY spectrum of GIPL-6 ( Fig.  3), together with knowledge of spin-coupling constants between endocyclic protons measurable from cross-peak multiplicities, suggest the presence of two Manp, two Galp, one Galf , one GlcNp, and one myo-inositol residue. The sequential arrangement of these residues and myo-inositol was obtained from 'H NOESY measurements (Fig. 4) as described previously (McConville et al., 1990b). By use of inter-residue through-space NOE connectivities across the glycosidic linkages, together with methylation analysis of the neutral, deam-* Portions of this paper (including "Experimental Procedures," part of "Results," Tables 1-111, and Figs. 5, 7, and 8) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press. inated glycan core (Table 111), the following partial sequence could be unambiguously defined Galpal-6Galpal-3Galf~l-3Manpal-3Manpa1-4GlcNpal-6-myo-inosito1, where the anomeric configurations were defined from the magnitude of the scalar coupling (J1,*) between the C-1 and C-2 protons of each residue, with the exception of the Galf residue. The very small splitting (below the linewidth) of the C-1 proton of the latter did not reliably reflect the anomeric configuration. However, the near identity of the chemical shifts of the C-1 proton of Galf in GIPL-6 (Table 11) and GIPL-3 (McConville et dl., 1990a, 199Ob) strongly suggested that this residue was in the 8-anomeric configuration in GIPL-6.

GI$al
The remaining two proton resonances in Fig. 2a were composed of a doublet of doublets, and had a large spin-coupling constant , suggesting that they corresponded to the anomeric protons of two hexose-l-P04 residues. To confirm this postulate, a one-dimensional 'H NMR spectrum of GIPL-6 was recorded with broadband 31P decoupling. Under these conditions, each doublet of doublets collapsed into a single doublet (Fig. 2b). From the spin coupling constants and the chemical shifts measured in the 'H COSY experiment, the two residues phosphorylated at C-1 were identified as aManp and aGlcp. Furthermore, mild acid hydrolysis of native GIPL-6 released the equivalent of 1 nmol each of Man and Glc/ nmol of myo-inositol (see Fig. 5 in the Miniprint Section), consistent with these residues being linked to the hexasaccharide core via phosphodiester bonds. The site of attachment of these residues to the core was determined by a two-step heteronuclear relayed corrrelation experiment (H-P-H). From the spectrum in Fig. 6 it can be seen that the C-1 protons of both the Glca-l-P04 and Mana-1-PO4 residues exhibit relayed connectivities via phosphorus to a pair of resonances. FIG. 3. 'H-'H COSY spectrum (region 3.26-4.46 ppm) of the GIPL-6 glycan. The resonance assignment pathway is illustrated for Man-2. Assignment of Cl-C4 proton resonances is straightforward from this spectrum. Assignment of the C-6 proton can be made from the 1H-31P-1H relay experiment (see Fig. 6). A weak cross-peak can be observed between these proton resonances in the COSY spectrum, together with a cross-peak correlating one of the C-6 proton resonances with the C-5 proton resonance (see inset). The latter has a chemical shift virtually identical to the C-3 proton resonance. These resonances were readily assigned to C-6 protons from the large observable geminal splittings (--12 Hz) in the cross-peaks. From these data, it was concluded that the Manal-po, was attached to the 6-position of the terminal Gal residue in the hexasaccharide core, since the C-6 protons of the Gal residue could be determined unambiguously from the COSY specrum. The assignment of the Glca-1-PO, linkage position to the 6-position of the Man residue distal to the glucosamine was similarly obtained, although the cross-peaks to, from, and between the C-6 protons of this latter residue were of lower intensity in the COSY spectrum (Fig. 3). The site of attachment of the Man-1-PO4 and the Glc-1-P04 residues to the core was also deduced by assessing the susceptibility of the phosphate residues to alkaline phosphatase before and after digestion of the deaminated glycan with jack bean a-mannosidase (see Miniprint Section, Fig. 7). Taken together, these data indicated that the structure of the GIPL-6 glycan was as shown in Structure 1, where the myo-inositol residue formed part of the lyso-1-0-alkyl-PI lipid moiety which contained predominantly Czk0 and Cza0 alkyl chains (McConville and Bacic, 1989). Reporter region of the 1H-31P-'H relayed correlation spectrum of the GIPL-6 glycan. This spectrum displays relayed connectivities via phosphorous between the anomeric protons of Glcal-PO4 and Manal-P04 to the C-6 protons of the corresponding aglycon. These connectivities were labeled using the same notation as in Fig. 2a.

Glcpal-PO,
Structure of the GIPL-4 Glycan-The PI-specific phospholipase C-released glycan of GIPL-4 contained seven anomeric protons of unit intensity in the low field region of the 'H NMR spectrum (Fig. 2c). These resonances were present in essentially identical positions to the anomeric resonances in the GIPL-6 spectrum ( Table 11), suggesting that GIPL-4 had the same sequence as GIPL-6, minus the Glca-l-P04 residue. This was confirmed from the 'H NOESY spectrum of GIPL-4 (not shown) which gave identical through-bond connectivi-Leishmania Lipophosphoglycan Biosynthesis ties to GIPL-6 and by methylation analysis which indicated that both GIPLs contained the same glycan core (Table 111). The location of the Manal-PO4 residue on the terminal Gal residue of the core was inferred from the near identity of the chemical shifts of the C-1 proton of this residue in GIPL-4 and in GIPL-6, together with the near identity of the shifts of the C-6 protons of the terminal Galp residue. This postulate was confirmed by Dionex HPLC analysis of the products of sequential jack bean a-mannosidase and alkaline phosphatase digestion (see Miniprint Section, Fig. 8). Taken together, these data suggest that the glycan moiety of GIPL-4 is as shown in Structure 2.
Location of Glc-1-PO4 on the Inner Mannose of the L. major LPG Anchor-We have previously shown that the oligosaccharide repeat units of L. major LPG are attached via phosphodiester linkage to the 6-position of the terminal Gal residue of the core (McConville et al., 1991a). In addition, approximately 60% of the LPG molecules contain a second phosphate residue on the mannose residue in the core which is distal to the glucosamine. This phosphate was partially resistant to alkaline phosphatase suggesting that it was also substituted, although the nature of this substituent was not determined. Mild acid hydrolysis of intact L. major LPG released approximately 0.6 nmol of Glc/nmol of inositol (Fig.   5 , panel C ) , raising the possibility that this mannose residue was substituted with Glc-1-P04 in the native molecule. This was confirmed by analysis of the one dimensional 'H NMR spectrum of L. major LPG (not shown) which exhibited a pair of doublets corresponding to the C-1 proton resonance of the Glcal-PO4 moiety, at essentially identical chemical shifts to that found in the GIPL-6 glycan.

DISCUSSION
The highly polar GIPLs of L. major L119 appear to be truncated forms of LPG which are accumulated due to a defect in LPG biosynthesis in this strain. This is suggested by the finding that 1) these glycolipids are not detectable in wild type strains of L. major which express LPG, 2) the levels of expression of these glycolipids in L119 are comparable with the levels of expression of LPG in virulent strains of L. major (McConville and Bacic, 1990) and 3) they show a remarkable degree of structural homology to the anchor region of LPG. The structures of these glycolipids were determined by twodimensional 'H and 1H-31P 500 MHz NMR, methylation analysis and chemical and enzymatic digestions and are summarized in Fig. 9. The structure of the GIPL-5 glycan was not determined in this study, but previous results suggest that this species is the lyso-derivative of GIPL-4 (McConville and Bacic, 1989). All these glycolipids contain a hexasaccharide core with a terminal Galp residue. In LPG, this residue is the site of attachment for the linear chains of phosphorylated oligosaccharide repeat units which have the backbone structure -6Gal/31-4Mana-l-P04 (McConville et al., 1991a, see Fig. 9). A striking feature of the polar L119 GIPLs is that they all terminate with Manal-P04, suggesting that the biosynthetic defect is in the transfer of galactose to this residue to form the first repeat unit. The identical location of the unusual Trivial name S I W r e GIPL-1 GaIfj31-3Manal"ana1-4G!cNalbPI

GIPL-6
Manal- PO~-6Galal-6Galal-3Galf~1-3~nal-3Manal-4GkNal-61).soPI Glcal-P04 residue on both GIPL-6 and on more than 60% of the LPG molecules further demonstrates the similarity between the polar GIPLs and LPG. The presence of this substituent on the LPG core was previously indicated by the apparent resistance of the core Man-6-PO4 to alkaline phosphatase treatment (McConville et al., 1990a), although the nature of the substituent was not identified. Subsequent analysis of the LPGs from L. donovani and L. menicana has revealed that nearly 100% of these molecules are substituted on the core with G l~a -l -P 0~.~~~ This substitution appears to be unique to LPG in wild type strains, as none of the GIPLs are similarly glucosylated (Fig. 9). GIPL-5 and -6 are also homologous to the LPG anchor in having the same lyso-alkyl-PI lipid moiety with predominantly Czk0 and C2e0 alkyl chains. By contrast, GIPL-4 has an alkylacyl-PI which is identical to the lipid moieties of GIPLs 1-3 in containing and Czz:fl as well as Czk0 and Cze0 alkyl chains (McConville and Bacic, 1989;McConville et al., 1990b).

4-
The most likely explanation to account for the accumulation of these glycolipids is that L119 has a mutation effecting the /3l-4 galactosyl transferase which is involved in forming the first repeat unit of LPG. Alternative possibilities, involving an impairment in intracellular vesicular transport or a defect in UDP-Gal synthesis or transport are unlikely, as all these Gal-containing GIPLs are still expressed in high copy number at the cell surface (McConville and Bacic, 1989).
Several other LPG-deficient clones of L. major have been prepared by mutagenesis of the LPG-producing strain V121 (Elhay et al., 1990). Analysis of the GIPL profiles of these clones suggests that a similar defect in the enzyme(s) involved in the assembly of the repeat units is responsible for the failure of these clones to synthesize mature LPG.' It is of interest that none of these mutants have a defect in GIPL biosynthesis, raising the possibility that such mutations may be lethal.
This study defines precisely which portion of the LPG has been deleted in L119 and supports the notion that the phosphoglycan moiety of LPG is essential for intracellular survival of the promastigote in macrophages. These phosphoglycan chains may be involved in mediating the phagocytosis of promastigotes by binding to the appropriate macrophage receptor (Handman and Goding, 1984;Russell and Wright, 1988;da Silva et al,, 1989;Talamas-Rohana et al., 1990) and in protecting the surface of the promastigote from hydrolytic enzymes and oxygen radicals in the phagolysosome (El-On et al., 1980;Chan et al., 1989). Less is known about the role of the anchor domain of LPG. This portion of the LPG provides the sole mechanism of attachment for the phosphoglycan chains to the plasma membrane in L. major. It may also prevent the induction of the macrophage oxidative burst by inhibiting the protein kinase C of the host cell (McNeely and Turco, 1987;McNeely and Turco, 1990). However, GIPL-6 is an equally effective inhibitor of purified protein kinase C6 and is also expressed in high levels in L119, suggesting that this latter property of LPG is not sufficient, by itself, to protect the intracellular promastigotes.
These results have several implications for LPG biosynthesis. First, the finding that, in L119, only the Mana-l-PO4 residue is transferred to the glycan core strongly suggests that the repeat units of LPG are built up by the sequential addition of monosaccharide units, rather than by the e n bloc transfer of preformed repeat units from a lipid carrier, as occurs in the synthesis of many prokaryote glycoconjugates (reviewed by Sutherland (1985)). This proposal is consistent with the results of Carver and Turco (1991) on the cell-free synthesis of L. donouani LPG, which also indicated that lipid-linked precursors were not involved in the formation of the repeat units. Second, GIPL-4 has the same alkylacyl-PI lipid moiety as GIPL-3, suggesting that LPG biosynthesis may be initiated by the transfer of a Man-1-PO4 residue to GIPL-3, without the prior deacylation of this species to lyso GIPL-3. It is not known whether the GIPL-3 lyso-derivatives, which are present in significant levels in some strains ( Fig. 1) (McConville et al., 1990b), are also LPG precursors. Third, in addition to the transfer of Man-1-PO4, it is likely that there are two other early steps in LPG biosynthesis, involving the deacylation of the PI moiety and the glucosylation of the glycan core. It is intriguing that, while the alkylacyl-PI lipid moiety of GIPL-4 has the same heterogeneous alkyl chain composition as GIPL-3, the deacylated PI lipid moiety of GIPL-6 is highly M. J. McConville and P. Robinson, unpublished results. enriched for Czk0 and C2e0 alkyl chains, as found in the LPG lipid moiety. These results indicate either that the alkyl chains of these glycolipids are being remodeled, or alternatively, that one or more of the enzymes involved in the deacylation and glucosylation steps of the LPG anchor preferentially recognize molecular species of GIPL-3 that contain longer alkyl chains. This enrichment for long alkyl chains is evident in the LPGs from other species of Leishmania (Orlandi and Turco, 1987)4 and may be necessary for the stable association of LPG molecules with the plasma membrane.