Biosynthesis of Mycobacterial Lipoarabinomannan*

The mycobacterial lipoglycans, lipomannan (LM) and lipoarabinomannan (LAM), are potent immunomodulators in tuberculosis and leprosy. Little is known of their biosynthesis, other than being based on phosphatidylinositol (PI), and they probably originate in the phosphatidylinositol mannosides (PIMs; PIMans). A novel form of cell-free incubation involving in vitro andin situ labeling with GDP-[14C]Man of the polyprenyl-P-mannoses (C35/C50-P-Man) and the simpler PIMs of mycobacterial membranes, reisolation of the [14C]Man-labeled membranes, and in situ chase demonstrated the synthesis of a novel α(1→6)-linked linear form of LM at the expense of the C35/C50-P-Man. There was little or no synthesis under these conditions of PIMan5with its terminal α(1→2)Man unit or the mature LM or LAM with copious α(1→2)Man branching. Synthesis of the linear LM, but not of the simpler PIMan2, was susceptible to amphomycin, a lipopeptide antibiotic that specifically inhibits polyprenyl-P-requiring translocases. A mixture of P[3H]I and P[3H]IMan2 was incorporated into the linear LM, supporting other evidence that, like the PIMs, LM and LAM, it is a lipid-linked mannooligosaccharide and a new member of the mycobacterial glycosylphosphatidylinositol lipoglycan/glycolipid class. Hence, the simpler PIMs originate in PI and GDP-Man, but further growth of the linear backbone emanates from C35-/C50-P-Man and is amphomycin-sensitive. The origin of the α(1→2)Man branches of mature PIMan5, LM, and LAM is not known at this time but is probably GDP-Man.

The cell wall of Mycobacterium spp. consists of a core composed of peptidoglycan linked to the heteropolysaccharide arabinogalactan which, in turn, is attached to the mycolic acids (1). Complementing the mycolyl residues is a variety of free lipids, and interspersed in this framework are lipoarabinomannan (LAM), 1 lipomannan (LM), and various proteins (1). LAM has emerged as a major modulator of the host immune response in the course of tuberculosis and leprosy, through gen-eralized suppression of immunity (2), induction of inflammatory cytokines (3), and the neutralization of potentially cytotoxic O 2 free radicals (4). LAM may also act as a key ligand in the phagocytosis of Mycobacterium tuberculosis (5).
The structural progression from PI to PIMan 2 to Ac 1 PIMan 2 to LM and LAM has suggested a similar biosynthetic order (8); however, to date, this was mere speculation. In this present study, we have defined the origins of the Man units of the PIMs and LM and, in the course of the work, identified an ␣136linked linear form of LM, the apparent precursor of mature LM and LAM.

Preparation of Enzymatically Active Membranes and Cell Envelope-
The transformable strain of Mycobacterium smegmatis, mc 2 155 (15), was grown in Bacto Nutrient broth (Difco Labs, Detroit, MI) to mid-log phase (about 24 h), harvested, washed with physiological saline, and stored at Ϫ20°C. M. tuberculosis H37Ra (ATCC 25177) was grown in a liquid medium containing glycerol, alanine, and salts (16) for 14 -16 days before harvesting. Mycobacteria (20 g, wet weight) were first washed and then resuspended at 4°C in a buffer (20 ml) containing 50 mM MOPS (adjusted to pH 7.9 with KOH), 5 mM ␤-mercaptoethanol, 10 mM MgCl 2 (buffer A), 150 g of DNase I (type IV, Sigma), and 250 g of RNase (microsomal nuclease (Sigma)) and subjected to five passes through a French pressure cell (Aminco, Silver Spring, MD) at 10,000 p.s.i. The preparation was centrifuged initially at 600 ϫ g to remove large particles and then at 27,000 ϫ g for 20 min at 4°C, and the membranes were obtained by centrifugation of the supernatant at 100,000 ϫ g for 1 h at 4°C. These (i.e. membranes arising from 20 g wet weight of cells) were resuspended in 1 ml of buffer A or buffer B (0.1 M Tris-HCl (pH 8.0), containing 0.25 M sucrose and 1 mM EDTA-Na 2 ) (17)) to yield a total of ϳ20 mg of protein that was frozen in small aliquots; the enzyme activity in membranes was slightly diminished (ϳ20%) by prolonged (2 months) storage at Ϫ20°C. The pellet from the 27,000 ϫ g centrifugation was resuspended in 7 ml of buffer A (20 mg protein/ml) or occasionally in the same volume of buffer B and used directly as an enzymatically active cell envelope fraction containing both cell walls and some membranes; the enzyme activity of this preparation was unaffected by prolonged storage (2 months) in small aliquots at Ϫ20°C. Alternatively, the 27,000 ϫ g pellet was resuspended in 40 ml of buffer A divided equally among four 40-ml centrifuge tubes to which were added 15 ml of Percoll (Pharmacia, Sweden) and centrifuged at 27,000 ϫ g for 60 min at 4°C (18). The particulate, upper diffuse band, containing both cell walls and membranes, was removed, collected by centrifugation, washed three times in buffer A, and finally resuspended in 4 ml of buffer A. The final concentration of this Percoll-60-purified cell envelope fraction (P-60) was 8 -10 mg of protein/ml. Over 80% of the enzyme activity was lost on freezing and thawing, and hence, this fraction was freshly prepared. 1.0 ml for the proceeds from 10 replicate reactions), and 100-l aliquots (2 mg of protein) were then redistributed into 10 fresh tubes, each containing 62.5 M ATP and usually 2 mg (200 l) of the Percollpurified cell envelope (P-60) in a total volume of 320 l and incubated further.
Pretreatment of Membranes with Amphomycin-Amphomycin (calcium salt) was a gift to C. J. Waechter from Bristol Laboratories and R. Bedensky, Case Western Reserve University, Cleveland, OH. The lipopeptide (up to 2 mg) was dissolved in 500 l of 0.1 N acetic acid, and the solution was adjusted to 0.05 M sodium acetate (pH 7.0) with 0.1 N NaOH for a final concentration of 2 mg/ml (19). Membranes with or without the in situ labeled [ 14 C]Man lipids were preincubated with amphomycin (10 g/100 l of reaction mixture) at 37°C for 10 min prior to various manipulations such as the addition of the P-60-purified cell envelope fraction and further incubation.

Extraction of [ 3 H]/[ 14 C]Man-labeled Products from Reaction
Mixtures-At the end of incubations, the reactions were terminated by the addition of CHCl 3 /CH 3 OH (2:1) (2.5 ml per 100 l of reaction mixture) followed by centrifugation to separate the pellet (17,20). The pellet was extracted once more with one-half the volume of CHCl 3 /CH 3 OH (2:1). The combined CHCl 3 /CH 3 OH (2:1) extracts were washed with 0.9% NaCl followed by CHCl 3 /CH 3  In an effort to determine whether the insoluble [ 14 C]Man-containing products associated with the final insoluble residue were LM/ LAM, it was successively extracted with refluxing 50% ethanol and 88% phenol, as described (7).

Preparation of Mycobacterial PIMan 2 s, P[ 3 H]I, and P[ 3 H]IMan 2 s-
Characterization of the various PIMs followed earlier work (9,11,21) and also by comparison with well defined, two-dimensional TLC maps of the full spectrum of PIMs (22). Thus, characterizations were based on fast atom bombardment mass spectroscopy analysis (9) and chromatographic patterns of the intact PIMs (23,24) and the deacylated forms (21), e.g. the deacylated PIMan 2 s produced a single product with an R f compared with deacylated PI of 0.8 on paper chromatograms in 2-propyl alcohol/NH 4 OH (2:1), whereas the deacylated PIMan 5 s produced a single product with a comparative R f of 0.30. To prepare P[ 3 H]I and P[ 3 H]IMan 2 s, M. smegmatis was grown in 100 ml of the glycerol/alanine/salts medium to mid-log phase (24 h) at which stage 250 Ci of [2-3 H]myo-inositol (21 Ci/mmol; NEN Life Science Products) was added to the medium and growth continued for another 12 h to late-log phase. Cells were harvested aseptically, and the recovered medium was reinoculated with fresh M. smegmatis and the cycle repeated. The combined harvested cells were washed several times with phosphate-buffered saline, lyophilized (184 mg), extracted several times with CHCl 3 / CH 3 OH (2:1), the extracts washed (25), dried (17 mg), and the phospholipids precipitated with acetone (8 mg; 2.1 ϫ 10 7 cpm). Twodimensional TLC, first in CHCl 3 /CH 3  s into the linear LM, the dry preparation containing them was suspended by sonication in buffer A, and 500,000 cpm aliquots were added to each of five standard reaction mixtures (320 l) containing membranes (8 mg), ATP, P-60 and buffer, which were incubated for 1 h, extracted with CHCl 3 /CH 3 OH (2:1) followed by CHCl 3 /CH 3  Selective Isolation of PIMan 5 s-This study resulted in an excellent protocol for the selective purification of the PIMan 5 s. We had observed that little of the PIMan 5 s (21) were present in the CHCl 3 /CH 3 OH (2:1) extracts of M. smegmatis. However, extraction of the resulting residue (from 40 g wet weight of cells) three times with CHCl 3 /CH 3 OH/H 2 O (10:10:3) resulted in 30.7 mg of material. Two-dimensional TLC, first in CHCl 3 /CH 3 OH/H 2 O (60:30:6) and then in CHCl 3 /CH 3 COOH/CH 3 OH/ H 2 O (40:25:3:6), followed by staining with an ␣-naphthol-containing reagent (26), demonstrated the presence only of the two PIMan 5 s, PIMan 5 and AcPIMan 5 (27).
Analytical Procedures-TLC was conducted in one-and two-dimensions on aluminum-backed plates of silica gel 60 F 254 (E. Merck, Darmstadt, Germany) in the solvents described in the text. An ␣-naphthol spray (26) and the molybdenum blue (28) dip reagent were used to detect carbohydrate and phosphorus in lipids, respectively. SDS-polyacrylamide gel electrophoresis and silver periodic acid-Schiff staining was performed as described (29). Autoradiograms were obtained by exposing chromatograms to Kodak X-Omat AR films at Ϫ70°C, usually for 4 -5 days. To locate 3 H-labeled products, EN 3 HANCE (NEN Life Science Products) was used according to instructions. Plates were also scanned for radioactivity using the Bio-Scan System 200 Imaging Scanner with the Autochanger 3000 (plates were read stepwise, using the 10-mm collimator, in 3-mm stops, each stop reading for 15 min).
Mild acid hydrolysis of CHCl 3 /CH 3 OH (2:1)-soluble lipids was conducted in 0.5 N HCl tetrahydrofuran (1:4) for 2 h at 50°C (30). Samples were neutralized with 200 l of 0.1 N NaOH in 0.02 N sodium phosphate, dried, resuspended in CHCl 3 /CH 3 OH/H 2 O (4:2:1), centrifuged, and the aqueous and organic phases separated and analyzed. For mild acid hydrolysis of the CHCl 3 /CH 3 OH/H 2 O (10:10:3)-soluble lipids, samples (ϳ100 g) were suspended in 50 l of 1-propyl alcohol, sonicated, treated with 100 l of 0.02 N HCl at 60°C for 30 min, neutralized with 15 l of 0.2 N NaOH (20,31), and subjected to gel filtration chromatography. For mild alkaline hydrolysis of both sets of lipids, samples were dissolved or suspended in 900 l of ethanol by sonication, followed by the addition of 100 l of 1 N NaOH and incubation at 40°C for 1 h. The NaOH was neutralized with ethyl formate (ϳ144 l), held at 40°C for 5 min, and dried. In the case of the CHCl 3 /CH 3 OH (2:1)-soluble lipids, the products were partitioned within CHCl 3 /CH 3 OH/H 2 O (4:2:1) and analyzed separately. The products from the CHCl 3 /CH 3 OH/H 2 O (10:10: 3)-soluble lipids were not partitioned but were applied directly to columns. Gel filtration chromatography was conducted on columns (1 ϫ 116 -175 cm) of various Bio-Gel P resins (Bio-Rad) in 0.1 M sodium acetate (pH 6.5). Repeated efforts to fully methylate the linear LM by the NaOH method (32) were unsuccessful. Successful methylation was accomplished by the method of Hakomori (33) with modifications, particularly the additions of fresh CH 3 I and 4.8 M dimethyl sulfinyl carbanion. The final reaction mixture was applied to a C 18 Sep-Pak cartridge (Waters, Milford, MA) and the permethylated linear LM recovered in the CH 3 CN eluant as determined by counting 1% of the eluants. The per-O-methylated linear LM was hydrolyzed in 2 M trifluoroacetic acid at 110°C for 2 h, the acid evaporated, reduced with NaBH 4 , per-O-acetylated, and the alditol acetate characterized by gas chromatography/mass spectrometry as described (34).

RESULTS AND DISCUSSION
The Specific Question-The immediate question was the manner of biosynthesis of LM with a view to ultimate LAM synthesis. Work in the late 1960s provided possible clues. Brennan and Ballou (23,24) and Ballou (35)  with the further addition of palmitate residues from palmitolyl-CoA to yield a mixture of PIMan 2 , the monoacyl (Ac 1 PIMan 2 ) and the diacyl (Ac 2 PIMan 2 ) derivatives. It was later recognized that LAM, in its various forms, i.e. ManLAM and AraLAM (22), and LM contain a mannan core linked to PI similar to that in the family of PIMs (PIMan 1-6 ) (6 -8). Specifically, it was shown that these lipoglycans contain the D-myo-inositol 2,6-bis-␣-Manp unit and thus to be based on PIMan 2 , and hence, it was assumed that PIMan 2 (and probably PIMan 3 ) was the immediate precursor of LM/LAM. Moreover, since Khoo et al. (9) demonstrated the presence of a palmitate substituent on C-6 of the Manp unit that is linked directly to C-2 of the Ino within LM/LAM, it was further assumed that Ac 1 PIMan 2 was the precise precursor. Throughout, it was thought that GDP-Man was the immediate donor of all of the Manp units of LM/LAM simply because it was the demonstrated precursor of the Manp units of the PIMan 2 family (23,24). However, shortly after this initial work, Takayama et al. (36,37) and Schultz and Elbein (38) described two alkali-stable mannophospholipids in M. tuberculosis and M. smegmatis, a mannosyl-1-phosphoryl-decaprenol (C 50 -P-Man) and a mannosyl-1-phosphoryl heptaprenol (C 35 -P-Man), which, in light of their group transfer potential and known role in mannolipid synthesis from other organisms (39), could be donors of polymerized Man in mycobacterial cell walls. Indeed, Schultz and co-workers (38,40) demonstrated indirectly that C 50 /C 35 -P-Man was the Man donor of undefined polymers. In addition, more recently, Yokoyama and Ballou (41) demonstrated that the ␤-mannosylphosphoryldecaprenol (C 50 -P-␤Man) was the direct Man donor of a series of ␣(136)linked oligosaccharides, clearly not LM or LAM with their copious ␣(132) branches. Hence, the questions posed at the outset of this work concerned the metabolic relationships among GDP-Man, C 35 /C 50 -P-Man, the PIMs, LM/LAM, and the connection, if any, between the ␣(136)-linked mannooligosaccharides and LM/LAM.

Incorporation of GDP-[ 3 H/ 14 C]Man by M. smegmatis Membranes and Cell
Wall-A variety of incubation conditions was applied as follows: those developed in the context of PIM biosynthesis (23,24); those (17) designed to examine the specificity of dolichyl-P-Man-dependent mannosyltransferase activity in mammalian glycoprotein synthesis; those for the incorporation of Man from GDP-Man and C 50 -P-␤Man into the ␣(136)linked oligosaccharides of M. smegmatis (41); and a more universal assay appropriate for the incorporation of [ 14 C]GlcNAc, [ 14 C]Rha, [ 14 C]Gal, and [ 14 C]Ara from their corresponding donors into the mycobacterial cell wall "core" (15,18). Incorporation in all cases was quantitatively and qualitatively comparable. However, replacement of 10 mM MgCl 2 (17) with 10 mM MnCl 2 or 10 mM CaCl 2 resulted in about 35% reduction of activity but no change in product profile; inclusion of 5 mM EDTA reduced incorporation by about 90%, pointing to the need for divalent cations for these mannosyltransferase activities.
Under all of these conditions, the majority (ϳ89%; ϳ0.4 ϫ 10 6 cpm/mg protein/reaction mixture) of the incorporated [ 3 H]Man was present in the CHCl 3 /CH 3 OH (2:1)-soluble membrane lipids with only about 10% in other material (Table I) First, there was a dramatic overall reduction in the synthesis of total membrane mannolipids (ϳ60%). Second, inhibition was specific for the C 35 /C 50 -P-Man pair of lipids, and synthesis of the PIMan 2 was unaffected ( Fig. 1). Therefore, the immediate donor of the Man residues of PIMan 2 is not C 35 /C 50 -P-Man, but GDP-Man, obviously reacting with PI as demonstrated previously (23,24).
Chase of in Situ Labeled C 35 /C 50 -P-[ 14 C]Man-To determine whether GDP-Man or C 35 /C 50 -P-Man was the donor for further mannosylation (e.g. in LM/LAM biosynthesis), a novel assay system was designed. Membranes were pulsed with GDP-[ 14 C]Man during a short (10 min) incubation period, but instead of extracting with CHCl 3 /CH 3 OH, the [ 14 C]Man-labeled membranes were re-harvested by centrifugation at 100,000 ϫ g, washed by suspending in MOPS buffer, and again harvested. The [ 14 C]Man-labeled membranes, shown to be devoid of GDP-[ 14 C]Man, were then further incubated for various times with or without the cell envelope (P-60) prior to extraction with CHCl 3 /CH 3 OH/H 2 O (4:2:1) to form a biphase and provide the CHCl 3 /CH 3 OH (2:1)-soluble lipids for TLC (Fig. 2). The results were striking. At the zero chase time in the presence of membranes only, practically all of the radioactivity was associated with the CHCl 3 /CH 3 OH (2:1)-soluble lipids (310,800 cpm/mg protein), i.e. the C 35 /C 55 -P-Man combination and Ac 1 PIMan 2 (Fig. 2, lane 1). Further incubation of membranes for 10 min resulted in little change in lipid radioactivity (Fig. 2, lane 2), whereas additional incubation for 60 min resulted in a significant loss of radioactivity (201,700 cpm/mg protein/reaction mixture) (Fig. 2, lane 3). However, the addition of the cell envelope (P-60) fraction to the reaction mixture in addition to the 60 min chase resulted in dramatic loss of lipid radioactivity (68,400 cpm/mg protein/reaction mixture), and TLC indicated

Properties of the End Products of the C 35 /C 50 -P-[ 14 C]Man
Chase-Efforts to identify the end products of the chase were based on this new assay in which these lipids were generated in situ through short (10 min) pulse labeling of membranes with GDP-[ 14 C]Man to preferentially generate C 35 /C 50 -P-[ 14 C]Man which were then further incubated for 1 h in the presence of the cell envelope (P-60) preparation. Twenty of these basic reactions were conducted, but this time, after extraction with CHCl 3 /CH 3 OH (2:1) and washing of the pellet to remove possible residual GDP-[ 14 C]Man, the pellet was further extracted with CHCl 3 /CH 3 Table II. A surprisingly large proportion of the incorporated radioactivity was solubilized by extraction with CHCl 3 /CH 3 OH/H 2 O (10:10:3).
The results of a comparison of the TLC radioautography profiles of the CHCl 3 /CH 3 OH (2:1) and the CHCl 3 /CH 3 OH/H 2 O (10:10:3)-soluble lipids (Fig. 3) demonstrated the occurrence in the latter fraction of a family of gradated highly polar [ 14 C]Man-containing glycolipids. This population was susceptible to alkali treatment but resistant to the mild acid conditions, indicating that they were possibly a family of PIMs intermediate in length between PIMan 3-5 and LM. Since the chromatographable components of the CHCl 3 /CH 3 OH/H 2 O (10: 10:3)-soluble lipids were minor, they were not further characterized. Clearly, the bulk of the CHCl 3 /CH 3 OH/H 2 O (10:10:3)soluble material synthesized at the expense of C 35 /C 50 -P-[ 14 C]Man remained at the origin in this solvent. Synthesis of this material also proved to be sensitive to amphomycin. When membranes were pretreated with amphomycin as described in  (27)) were the only higher PIMs appreciably synthesized (Fig.  4, A and B) by M. smegmatis. However, there was no synthesis of these PIMan 5 s under the cell-free C 35 /C 50 -P-Man chase conditions (Fig. 4C). The implications were that C 35 /C 50 -P-Man is not a precursor of the ␣(132)Man unit that differentiates PIMan 4 -6 from the PIMan 2 s and PIMan 3 (21). The results of the experiments described in Fig. 4 also demonstrated the presence of [ 3 H]Ins in the material at the origin (Fig. 4B), indicating for the first time that the major CHCl 3 /CH 3

Results of large scale chase of in situ labeled membranous C 35 /C 50 -P-[ 14 C]Man into polymerized materials
A large batch of membranes were prepared from 20 g of M. smegmatis. These were divided among 20 tubes in the basic assay format, and each was incubated for 10 min with GDP-[ 14 C]Man. The radiolabeled membranes were recovered by centrifugation at 100,000 ϫ g, washed, and resuspended in 2 ml of buffer A, and 100 l was added to each of 20 fresh tubes and further incubated for 1 h in the presence of 200 l of cell envelope fraction (P-60). The reaction mixtures were then extracted with CHCl 3 /CH 3 OH (2:1) and CHCl 3 /CH 3 OH/H 2 O (10:10:3). The majority of the residual radioactivity was not extractable with either 50% ethanol or 88% phenol, indicating that it was not LM/LAM.  3 COONa, demonstrated that all of the radioactivity was excluded (Fig. 5A), possibly more a reflection of micelle formation than size. The deacylated fraction was included in Bio-Gel P100 (Fig. 5B) and Bio-Gel P10 (Fig. 5C) and, barely, in Bio-Gel P-2, generally demonstrating the retention volumes of eicosaccharides. Clearly, its size was smaller than that of deacylated LM and deacylated LAM (Fig. 5, B and C).
A large scale, 14-fold reaction mixture was prepared in which the membranes from 20 g of cells were preincubated with 14 Ci of GDP-[ 14 C]Man, harvested, washed, and then further incubated with the cell envelope fraction from the same batch of cells for 1 h. The CHCl 3 /CH 3 OH/H 2 O (10:10:3)-soluble lipids were obtained after pre-extraction with CHCl 3 /CH 3 OH (2:1). About 1 ϫ 10 6 cpm were recovered. SDS-polyacrylamide gel electrophoresis and subsequent autoradiography of the dried gels showed that this material had mobility properties intermediate between that of the family of PIMs and LM, pointing again to a product intermediate between the higher PIMs and LM.
To directly demonstrate the presence of PI in the CHCl 3 / CH 3 OH/H 2 O-soluble polymer, the membranes containing the in situ labeled C 35 /C 50 -P-[ 14 C]Man (from five reaction mixtures) were further incubated with P[ 3 H]IMan 2 (5 ϫ 500,000 cpm) for 1 h and extracted with CHCl 3 /CH 3 OH (2:1) followed by CHCl 3 /CH 3 OH/H 2 O (10:10:3). The latter extract was treated with alkali and applied to Bio-Gel P-2 (Fig. 6) All of the evidence pointed to the synthesis of a new form of LM (linear LM) under the in vitro conditions. To examine the linkage pattern of Man in the newly synthesized product, the CHCl 3 /CH 3 OH/H 2 O (10:10:3)-soluble material was subjected to a number of methylation attempts. Only the dimethylsulfinium carbanion-catalyzed method (33) was successful, with about 50% conversion of the original 1 ϫ 10 6 cpm into a full permethylated product. This was purified on a small column of SepPak, hydrolyzed with 2 M CF 3 COOH, reduced with NaB[ 2 H] 4 , acetylated, and the alditol acetates analyzed by gas chromatography/mass spectrometry with simultaneous counting of radioactivity (15). Only two products were obtained, the terminal Manp derivative (1,5- mCi/mmol) in a total volume of 320 l/reaction mixture (total of 8 such reaction mixtures) and incubating for an additional 10 min. The reaction mixtures were combined and centrifuged at 100,000 ϫ g for 1 h, washed, and resuspended to 800 l. Eight new reaction mixtures were prepared containing 100 l of the radiolabeled membranes, 200 l of the cell envelope P-60 fraction (4 mg/protein), and 0.06 mM ATP in a total volume of 320 l. The reaction mixtures were incubated at 37°C for 1 h and then extracted with CHCl 3 /CH 3 OH (2:1) followed by CHCl 3 / CH 3  Conclusions-This present body of research solves several questions left in abeyance for many years but now of importance in light of the role of LAM in phagocytosis of M. tuberculosis and generalized immunosuppression. The structural relationship between LAM/LM and the PIMs was first recognized with the discovery of the presence of the basic PI (7) and PIMan 2 (8) units in the molecules. However, the biosynthetic origins of the PIMs had always been in doubt and that of LM/LAM had never been explored. Ballou and colleagues (23,24,35,46) had clearly demonstrated that GDP-Man was the source of the Man units of PIMan 2 and PI was a suitable acceptor of [ 14 C]Man from GDP-[ 14 C]Man. However, this evidence emerged prior to the recognition of the existence of a heptaprenyl (3,7,11,15,19,23,27-heptamethyl-2,6,10-octacosatriene-1-ol)-P-Man and a decaprenyl-P-Man (the structure of the decaprenol is probably similar to that in the decaprenyl-P-Araf (47)) in mycobacterial membranes (37) and their possible roles in polymer synthesis (38,40). The use, in this present work, of amphomycin, a member of the mureidomycin family known to inhibit a variety of translocases by chelating to a variety of polyprenol monophosphates (19), resulted in complete inhibition of the synthesis of the polyprenyl-P-Man, but allowed continuing synthesis of the PIMan 2 s, demonstrating that the pathway proposed by Brennan and Ballou (23) prevails and that suggested by others (48) is incorrect.
The first definitive evidence of the precursor role of a polyprenol-P-Man, the decaprenyl-P-Man, in mycobacterial mannan synthesis came from the work of Yokoyama and Ballou (41), definitive in that it involved the isolation of pure C 50 -P- [ 14 C]Man, its incubation with membrane preparations, and characterization of end products. Clearly, the newly synthesized products were ␣(1-6)-linked mannooligosaccharides, and there was no synthesis of (132)Man branches. However, the relationship of these mannooligosaccharides to the present linear LM, or LM, or LAM proper is not clear. Whether these mannooligosaccharides were lipid-linked or even reducing is also not clear. However, it would seem from earlier work (49) that at least one of these is an (␣1-6)Man-linked eicosaccharide, non-reducing, and hence with no apparent direct relationship to LM or LAM. These mannooligosaccharides may be natural autolysis products of linear LM; mycobacteria contain a non-lipidated, reducing mannan (50), apparently identical to the mannan within LM. Regardless of any relationship, the present body of work and that of Yokayama and Ballou (41) demonstrate that the ␣(136)-linked Man backbone of the mannooligosaccharides, the linear LM, and presumably mature LM and LAM arise in C 50 -P-Man, and the (␣1-2)-linked branches of PIMan 5 , LM, LAM, and other mannooligosaccharides arise elsewhere. Present tentative evidence indicates that GDP-Man is the source of these external Man residues as in the case of the PIMan 2 s. Further incubation of GDP-[ 14 C]Man-pulsed, washed and chased membranes with fresh GDP-[ 14 C]Man resulted in modest incorporation of [ 14 C]Man into products indistinguishable from LM and LAM (results not shown). Thus, the combined evidence from the past and the present points to the pathway outlined in Fig. 7 for the biosynthesis of the native branched LM. Separately, we have shown that the arabinan of LAM arises by the single transfer of Araf units from C 50 -P-Araf (47) to an endogenous acceptor (18). Whether or not this acceptor is endogenous LM or linear LM has not been established.