Biosynthesis of Tunicamycin and Metabolic Origin of the 11-Carbon Dialdose Sugar, Tunicamine*

Tunicamycin is a reversible inhibitor of polyprenol-phosphate: N-acetylhexosamine-1-phosphate translocases and is produced by several Streptomycesspecies. We have examined tunicamycin biosynthesis, an important but poorly characterized biosynthetic pathway. Biosynthetic precursors have been identified by incorporating radioactive and stable isotopes, and by determining the labeling pattern using electrospray ionization-collision induced dissociation-mass spectrometry (ESI-CID-MS), and proton, deuterium, and C-13 nuclear magnetic resonance (NMR) spectroscopy. Preparation and analysis of [uracil-5-2H]-labeled tunicamycin established the complete ESI-CID-MS fragmentation pathway for the major components of the tunicamycin complex. Competitive metabolic experiments indicate that 7 deuteriums incorporate into tunicamycin from [6,6′-2H,2H]-labeled d-glucose, 6 of which arise from d-GlcNAc and 1 from uridine and/ord-ribose. Inverse correlation NMR experiments (heteronuclear single-quantum coherence (HSQC)) of13C-labeled tunicamycin enriched fromd-[1-13C]glucose suggest that the unique tunicamine 11-carbon dialdose sugar backbone arises from a 5-carbon furanose precursor derived from uridine and a 6-carbonN-acetylamino-pyranose precursor derived from UDP-d-N-acetylglucosamine. The equivalent incorporation of 13C into both the α-1′′ and β-11′ anomeric carbons of tunicamycin supports a direct biosynthesis via 6-carbon metabolism. It also indicates that the tunicamine motif and the α-1′′-linked GlcNAc residue are both derived from the same metabolic pool of UDP-GlcNAc, without significant differential metabolic processing. A biosynthetic pathway is therefore proposed for tunicamycin for the first time: an initial formation of the 11-carbon tunicamine sugar motif from uridine and UDP-GlcNAc via uridine-5′-aldehyde and UDP-4-keto-6-ene-N-acetylhexosamine, respectively, and subsequent formation of the anomeric-to-anomeric α, β-1′′,11′-glycosidic bond.

11-carbon 2-aminodialdose sugar called tunicamine, and an amide-linked fatty acid. The ␣␤1,1Ј-glycosidic linkage between tunicamine and the GlcNAc substituent is also unique to the tunicamycin family of compounds. Tunicamycin structural variants occur that differ only in the nature of the N-linked acyl chain. We have recently introduced a structure-based naming system that identifies each tunicamycin by its signature fatty acid, i.e. Tun 13:1-Tun 18:1 (4).
Although a great deal is known about tunicamycin structure and function, no previous analysis of tunicamycin biosynthesis has been reported. The key to understanding the biosynthesis of tunicamycin is the origin of the 11-carbon tunicamine dialdose sugar and the kinetics for the formation of the ␣,␤-1ЈЈ,11Јglycosidic bond. A large number of natural products of Streptomyces origin are synthesized from 2-carbon units via a polyketide-type reaction sequence (5). However, other long chain sugars such as sialic acids, ketodeoxyoctulosonate (KDO) and ketodeoxyheptulosonate are synthesized from aldol condensation of lower sugars with phosphoenolpyruvate (PEP) 1 (6). In addition, the biosynthesis of similar nucleoside antibiotics, polyoxins and nikkomycins, occurs by ligation of PEP and uridine-5-aldehyde, generating 8-carbon octofuranuloseuronic acid nucleoside as an intermediate (7,8,9).
Here, metabolic radiolabeling experiments and stable isotope incorporations have been applied to unravel the metabolic origin of the 11-carbon dialdose sugar, tunicamine. We report that [2-14 C]uridine and [1-14 C]glucosamine are efficiently incorporated into tunicamycin by resting cells of Streptomyces chartreusis and that the [1-14 C]glucosamine feeds into both the 11-carbon tunicamine and the attached ␣-1ЈЈ-GlcNAc residue. Stable isotope incorporations using 2 H-or 13 C-labeled glucose and competitive metabolic experiments were monitored by LC-ESI-CID-MS and NMR (H-1, C-13, and HSQC) spectroscopy. The isotopic labeling patterns were consistent with carboncarbon bond formation between a 5-carbon precursor derived from uridine and a 6-carbon hexose intermediate, the latter most probably derived from UDP-GlcNAc. Heteronuclear C-13/ H-1 NMR correlations showed an equal incorporation of 13 C label from [1-13 C]glucose into both the ␤-11Ј and ␣-1ЈЈ anomeric carbons, indicating that both arise from a common precursor pool. Hence, both the pseudo-aminogalactopyranosyl (pseu-doGalN) ring of tunicamine and the ␣-1ЈЈ-linked GlcNAc residue are initially derived from the sugar nucleotide UDP-Glc-NAc. Based on the results of these experiments a biosynthetic pathway is proposed for tunicamycin for the first time.
Isolation of Tunicamycin-After growing S. chartreusis, NRRL 3882 in liquid TYD for 5 days, acid insoluble tunicamycin complex was precipitated by acidifying the cultures (0.2 M HCl). Mycelia and the acid precipitate were harvested by centrifugation (4000 ϫ g, 10 min), washed twice by resuspending in dilute HCl (0.2 M), and extracted by vortexing with methanol. Following centrifugation to remove cell debris, the methanolic supernatant was evaporated to dryness (40°C) and then redissolved in fresh methanol (200 l). Preparative thin layer chromatography (Silica 60 plates with a fluorescent indicator) was used to purify the tunicamycin complex (butanol:ethanol:water, 5:2:3 by volume). Where necessary, further purification was achieved by reversed-phase HPLC as previously described (4).
Metabolic Labeling Studies-Streptomycetes were cultured in liquid TYD medium for 5 days prior to the addition of radiolabeled precursors. The labeled tunicamycin complex was acid precipitated after another 2 days of growth and analyzed by thin layer chromatography-autoradiography. D-[1-14 C]glucosamine-labeled tunicamycin was selectively de-Oglycosylated and deacylated with 2 M HCl at 100°C for 3 h, as described by Tamura (1).
For the stable isotope experiments, 2 H-or 13 C-labeled glucose was substituted for the 0.6% glucose in TYD medium in 100-and 50-ml cultures, respectively. The tunicamycins were acid precipitated, and analyzed by LC-ESI-MS and NMR as described below. In some experiments the 99.9% deuterated glucose was diluted to 50% isotopic purity in order to better assess NMR proton-proton couplings.
Competitive Metabolic Experiments-S. chartreusis, NRRL 3882 cells were grown in 0.6% D-[6,6Ј-2 H, 2 H]glucose-substituted TYD medium in 10-ml shaken cultures. The unlabeled competitive metabolite, uridine, GlcNAc, ribose, glycerol, or succinate was added to the culture medium at the start of the growth period in a 3 molar excess of the deuterated glucose. The cultures were grown for 7 days at 28°C, except for the ribose-competed culture that took 14 days to grow to completion. Tunicamycin complex was recovered and purified as above. Results of the isotopic enrichment were assessed by LC-ESI-CID-MS.
Preparation of [Uracil-5-2 H]Tunicamycin-Tunicamycin (0.5 mg) was exchanged in 0.5 ml D 2 O at 60°C for 7 days in the presence of 0.4 M triethylamine. After evaporation to dryness, the hydroxyl protons were twice exchanged by lyophilization from deuterated water. Proton NMR analysis showed a 100% non-reversible exchange of the proton at position 5 in the uracil moiety of tunicamycin.
NMR and HSQC Spectroscopy-Standard and metabolically labeled tunicamycins were deuterium exchanged by lyophilization from D 2 O and redissolved in deuterated methanol. Spectra were recorded on a Bruker Avance 600 MHz instrument using pulse sequences supplied by Bruker. HSQC spectra were obtained using Echo/Antiecho-TPPI utilizing gradient selection with decoupling during acquisition.

RESULTS
S. lysosuperificus, S. clavuligerus, and S. chartreusis strains NRRL 3882 and 12338 were initially screened for tunicamycin production by metabolic incorporation of [2-14 C]uridine. The [2-14 C]tunicamycin was precipitated from cultures with mild acid (tunicamycin is insoluble under these conditions), waterwashed, and redissolved in methanol prior to analysis by thin layer chromatography-autoradiography (Fig. 1). The [2-14 C]tunicamycins eluted as a single spot and were readily separated from uridine by twice eluting with butanol:ethanol:water. The [2-14 C]uridine was taken up by S. chartreusis strains 3882 and 12338 in liquid TYD culture containing 0.6% glucose and incorporated into tunicamycin with high efficiency (2% incorporation of radiolabel) (Fig. 1A). This was particularly efficient in S. chartreusis, NRRL 3882, allowing incorporation of [2-14 C]uridine and subsequent isolation of [2-14 C]tunicamycin on a preparative (nCi) scale (Fig. 1B). Mycelium extracts and acid-precipitated culture supernatants from the S. chartreusis strains were also analyzed by reversed-phase HPLC and showed identical profiles of tunicamycin N-acyl components (Fig. 1C). These co-eluted with tunicamycin standards and were structurally confirmed by LC-ESI-MS (data not shown). As reported earlier, neither S. lysosuperificus nor S. clavuligerus produced detectable amounts of tunicamycin when grown in liquid TYD culture (4).
Stable Isotope Studies and Electrospray Mass Spectrometry-We have previously developed a reverse-phase, positiveion detected LC-ESI-MS assay for tunicamycins (4). Here, collision-induced dissociations (CID) were used to promote further MS fragmentations (LC-ESI-CID-MS) and were applied to localize the metabolic incorporation of stable isotopes into the tunicamycins (Fig. 2). To establish fragmentation pathways [uracil-5-2 H]tunicamycins were prepared by a novel application of Heller's deuterium exchange procedure (10). Tunicamycin complex (0.5 mg) was isotopically exchanged in D 2 O in the presence of triethylamine as a non-reactive volatile base. After incubation (7 days, 60°C) 100% exchange of deuterium at position 5 in the uracil ring was confirmed by LC-MS (  (Fig. 1B). The uracil moiety is subsequently lost as a neutral fragment, and the remaining fragment ions are unaffected by the 5-deuterium label, confirming the selective incorporation. The uracil neutral loss is also evident as a protonated ion at m/z 113 and 114 for the [uracil-5-2 H]tunicamycin.
The ESI-CID-MS fragmentation pathway for tunicamycin Tun 16:1 A is shown in Fig. 2 ion is resonance stabilized and consequently was often observed as the base peak. An acyloxy ion is then formed by neutral loss of [C 4 H 5 ON] from the C-10Ј amide bond. Importantly, corresponding fragments differing by 14 amu (i.e. one CH 2 ) are seen for tunicamycins with other N-acyl chains, i.e. these fragments retain their attached N-acyl group after dissociation of the uracil moiety and are therefore characteristic for each parent tunicamycin species.
This fragmentation pattern was confirmed by ESI-CID-MS of tunicamycins metabolically labeled from [1-13 C]glucose ( Fig.  3). An overall incorporation of two 13 C carbons was observed (i.e. m/z 869 for Tun 16:1, rather than m/z 867), and these were localized to the anomeric carbons C-1ЈЈ and C-11Ј. This is supported by m/z 625 rather than m/z 624 for the N-acyltunicamine-uracil oxonium fragment and by m/z 205 instead of m/z 204 for the GlcNAc oxonium ion. No major incorporations were observed in either the uracil ring (m/z 113) or the N-acyl chain (m/z 237). Fragments arising from the N-acyltunicamine-uracil parent ions (e.g. m/z 588, 512, 476, 392, and 320) are also all observed 1 amu higher in the 13 C-labeled tunicamycins (Fig. 3). The subsequent neutral loss of 84 amu, rather than 83 amu, from [C 4 H 5 ONAcyl] ϩ further indicates the position of the C-11Ј anomeric carbon.
Stable Isotope and Competitive Metabolic Experiments-Radiolabeling experiments indicated that S. chartreusis efficiently metabolizes [1-14 C]glucosamine (GlcN) into tunicamycin (Fig. 4A). Uptake and incorporation of the labeled GlcN by resting cells in TYD were apparently little affected by glucose catabolite repression. Radiolabeled tunicamycin was isolated by mild acid precipitation and preparative TLC and subsequently acid hydrolyzed to cleave the 1ЈЈ,11Ј-glycosidic bond. Analysis of the radiolabeled products by TLC indicated that [1-14 C]GlcN fed into both the 11-carbon tunicamine moiety and the attached GlcNAc residue. Two radiolabeled spots were observed that co-migrated with free glucosamine and tunicamineuracil (Fig. 4A). The latter slow-moving spot quenched the fluorescent indicator in the TLC plates because of the attached uracil group and co-migrated with a tunicamine-uracil standard. Hence, it is clear that tunicamycin can be metabolically radiolabeled from GlcN and that the GlcN is incorporated into both the ␣␤-1,1-linked GlcNAc residue and the tunicamineuracil core. These data suggest that tunicamine is derived from GlcN at the 6-carbon level rather than by a polyketide extension mechanism.
Establishment of the tunicamycin MS fragmentation pathways permitted the localization of stable isotope incorporations by LC-ESI-CID-MS. Culturing S. chartreusis NRRL 3882 on TYD containing 0.6% [6,6Ј-2 H, 2 H]glucose and isolation of the labeled tunicamycin fraction by mild acid precipitation resulted in the formation of deuterium-enriched tunicamycins. MS fragmentation analysis showed molecular ions (e.g. m/z 874 in Fig. 4, B.2) increased by 7 mass units relative to the corresponding unlabeled tunicamycins (m/z 867. Fig. 4 The mycelia were filtered off, acid-washed, and extracted with methanol. The culture supernatants were acidified to precipitate tunicamycins that were recovered by filtration. After redissolving in methanol, the extracts were assayed by HPLC on a RP-18 column eluted with MeCN:1% aq. acetic acid (40:60 v/v.). Eluents were monitored with a diode array detector at 360 nm, and assignments were confirmed by electrospray-MS (data not shown).
Competitive Metabolic Experiments-To further assign the localizations described above, competitive metabolic experiments were conducted. S. chartreusis cells were grown on TYD containing [6,6-2 H, 2 H]glucose, as described, but were supplemented with a 3-fold equivalent excess of an unlabeled competitive metabolite: GlcNAc, uridine, ribose, glycerol, or succinate ( Fig. 4B.3, Table I). These unlabeled precursors compete metabolically for the incorporation of labeled [6,6-2 H, 2 H]glucose into the tunicamycins, which resulted in selective editing of deuterium atoms. The competed [6,6-2 H, 2 H]glucose-labeled tunicamycins were analyzed by LC-CID-ESI-MS as described above. Unlabeled GlcNAc competed with the [6,6-2 H, 2 H]glucose label so that the observed tunicamycin molecular ions are only 1 mass unit greater than corresponding unlabeled controls, and 6 mass units less than a labeled control experiment (Fig. 4B). Fragmentation analysis indicates that this remaining deuterium was located in the tunicaminyl moiety rather than in the ␣-1ЈЈ-GlcNAc group. Hence, unlabeled GlcNAc competes for the incorporation of three deuteriums into the ␣-1ЈЈ-GlcNAc and three into tunicamine. These data support the concept that the tunicamycin pseudoGalN pyranose ring is metabolically derived from GlcNAc (presumably via UDP-GlcNAc).
Unlabeled uridine or ribose competed for the incorporation of [6,6-2 H, 2 H]glucose and gave tunicamycins with six deuteriums (Table I). MS fragmentation analysis indicated that the deuterium whose incorporation was blocked, D7, was located in the pseudoribosyl moiety. Hence, the D7 label was predominantly lost on the 72 amu neutral fragment (Fig. 4B.3), suggesting that it is located at the tunicamine C-5Ј. This was also deduced from the fact that three deuteriums were located in the ␣-1ЈЈ-GlcNAc residue and three others in the tunicaminyl core, again with no incorporation into the uracil ring. Hence, these data support the hypothesis that the tunicamine-uracil core is derived from 5,6-ligation of uridine and UDP-GlcNAc and confirm the earlier radiolabeling experiments showing that [ 14 C]glucosamine is efficiently incorporated into the GlcNAc and tunicaminyl moieties by S. chartreusis.
The deuterium label was also sometimes observed in the [C 4 H 5 ONAcyl] ϩ fragments and the acyloxy ions (Table I). This was confirmed by gas chromatography-mass spectrometry (GC-MS) analysis of the fatty acid component following methanolysis (data not shown) and by NMR analysis (see later). This is presumably because D7 occupies two locations (D7a and D7b) that average out as 1 mass unit.
To address whether biosynthesis of the tunicamine backbone might occur by 2-or 3-carbon extensions, the [6,6-2 H, 2 H]glucose label was metabolically competed with glycerol or succinate (Table I). Glycerol or succinate suppress the catabolism of the deuterated glucose into 3-carbon compounds so that incorporations occur predominantly via a 6-carbon metabolism. The presence of glycerol or succinate competed five deuteriums from tunicamycin overall, two from the ␣-1ЈЈ-GlcNAc and three from the N-acyltunicamine-uracil. Thus, the geminal deuteriums D1 and D2 are retained in the GlcNAc moiety, and similarly D4 and D5 are retained in the tunicamine pseudoGal-NAc ring because they arise directly from a 6-carbon metabolism. In contrast, D3 and D6 are competed out of the tunicamycins, presumably because glycerol or succinate block the conversion of [6,6-2 H, 2 H]glucose to 1-deuteroglucose via 3-carbon intermediates.
The remaining deuterium, D7, is competed from the N-acyl chains (i.e. at D7b), but the label is retained at the tunicamine C-5Ј (i.e. at D7a). This is evident from the neutral loss of [C 3 (Fig. 6) calls for an enzyme-catalyzed oxidation of uridine to uridine-5-aldehyde prior to 5,6-bond formation. Hence, assuming stereoselectivity, only one deuterium should be lost from [5,5-2 H, 2 H]uridine prior to incorporation into tunicamycins.
Stable Isotope Studies and NMR Spectroscopy-NMR provides a tool to comprehensively assign the localization of stable isotopes into tunicamycins. Key features are the 6Ј-CH 2 and 6ЈЈ-CH 2 groups of the tunicamine and ␣-1ЈЈ-GlcNAc, respectively; the origin of the C-5Јpseudo-ribosyl bridge; and the ␤-1Ј-N-, ␤-11Ј-O-, and ␣-1ЈЈ-Oanomeric assignments. Isotopically-labeled tunicamycins were isolated from cultures of S. chartreusis grown on TYD enriched with [1-13 C]glucose. Following purification by preparative TLC, the isolates were analyzed by H-1 NMR, C-13 NMR, and HSQC 13 C-1 H correlation spectroscopy. The methylene groups were assigned as geminal pairs from a DEPT-HSQC correlation experiment (Fig.  5), and anomeric assignments were made from COSY and total correlation spectroscopy (TOCSY) spectra (data not shown) and HSQC data (Fig. 5). Quantitative isotopic enrichments were deduced by comparing the normalized integrated HSQC signals of the 13 C-enriched tunicamycin with those of the unlabeled control. Significant enhancement of 16 HSQC signals was observed in the HSQC experiment ( Fig. 5 and Table II). The largest isotopic enrichments were at the anomeric carbons C-11Ј and C-1ЈЈ, 20-and 22-fold respectively, which were ϳ20% of the maximum theoretic isotopic incorporation. Significantly, both anomeric carbons are isotopically enhanced from [1-13 C]glucose to an equal extent, indicating a direct incorporation at the hexose level rather than prior metabolism to 2-or 3-carbon units. Hence, the tunicaminyl pseudoGalN moiety and the ␣␤-1Ј11ЈЈ-GlcNAc residue are apparently derived directly from glucose, from the same metabolic pool of GlcNAc or UDP-GlcNAc. In addition, isotopic enrichment was observed at C-6 in both the tunicamine pseudoGalN and the ␣-1ЈЈ-GlcNAc rings, 4-fold for the tunicamine and 6-fold for the GlcNAc signals. Incorporation at the C-6 positions results from the metabolism of [1-13 C]glucose via glycolysis to yield labeled 3-carbon compounds and their subsequent re-incorporation into [6-13 C]glucose via gluconeogenesis. This is consistent with Embden-Meyerhoff-Parnas catabolism in these bacteria rather than the Entner-Doudoroff pathway (13, 14, 15). Importantly, 7-and 4-fold enrichment at C-1Ј and C-5Ј of the pseudo-ribosyl motif is also observed, confirming that this part of the tunicamine backbone originates from a ribose carbon skeleton. Known sugar metabolism supports the formation of [1, 5-13 C]ribose from [1-13 C]glucose via the pentose phosphate pathway (8,16,17). Isotopic enrichment of the N-acetyl methyl carbon of the ␣-1ЈЈ-GlcNAc and the C-2ЈЈЈ and C-4ЈЈЈ carbons (but not C-3ЈЈЈ)

TABLE I
ESI-CID-MS analysis of competitive metabolic experiments S. chartreusis, NRRL 3882 was cultured in TYD media containing isotopically-labeled glucose as the metabolic precursor and a 3-fold molar excess of an unlabeled competitive metabolite: GlcNAc, uridine, succinate, glycerol, or ribose as indicated. Labeled tunicamycins were acid precipitated and methanol extracted after 5 days and analyzed by LC-ESI-CID-MS. In each case, the MS fragment ion masses shown are for tunicamycin Tun 15:1 A and are analogous to those deduced for Tun 16:1 A in Fig. 2. Numbers in parentheses indicate the mass increases relative to control unlabeled tunicamycin of the N-acyl chains occurs by 3-carbon metabolism, presumably via acetyl-CoA. Further isotopic incorporation into the -methyl groups of the fatty acid chains presumably originates from valine, a precursor in the branched-chain fatty acids biosynthesis (18,19). Minor enrichment at C-5 and C-6 in the uracil ring of tunicamycin can be ascribed to the following pathway.

DISCUSSION
The dialdose tunicamine has a configuration that resembles a D-ribofuranose 5-membered ring 5,6-linked to a D-aminogalactopyranosyl 6-membered ring. Although other hemi-acetal cyclic forms are possible, in practice they are not observed, 2 suggesting that tunicamine is formed by ligation of cyclic ribofuranose and aminogalactopyranose structures that are already locked into these configurations as their closed cyclic glycosides (i.e. as ribofuranosyl-uracil N-glycoside and Nacetylgalacto(gluco)samine-UDP O-glycoside, respectively). Hence, a biosynthetic ligation of two 5-and 6-membered cyclic sugars is favored. The alternative is that tunicamine biosynthesis occurs by sequential addition of smaller units, either via a polyketide-type mechanism or by ligation of uridine-5Ј-aldehyde to phosphoenolpyruvate (PEP).
The latter reaction has been implicated in the biosynthesis of analogous nucleotide antibiotics. Incorporation studies revealed that the C-6Ј of polyoxin and nikkomycin nucleosides arise from C-3 of PEP, by condensation of PEP with uridine-5Ј-aldehyde to give octofuranuloseuronic acid nucleoside as an intermediate (7,8,20). A putative enolpyruvate transferase gene (nikO) with homology to UDP-GlcNAc enolpyruvate  Table II.  [1-13 C]glucose into tunicamycin as analyzed by HSQC inverse-correlation spectroscopy HSQC signal intensities of the metabolically [1-13 C]glucose-labeled tunicamycin relative to unlabeled control. The signals were integrated and internally normalized against non-enriched baseline signals from the tunicamine (C-4Ј/H-4Ј and C-10Ј/H-10Ј) and ␣-1ЈЈ-GlcNAc (C-2ЈЈ/H-2ЈЈ) motifs (Fig. 2). 13 C-enrichments were calculated from the normalized integrals from the 13 C-labeled tunicamycin divided by the corresponding unlabeled control integrals. Only signals that are fully resolved in the HSQC experiment are included. transferases (MurA) and 5-enolpyruvyl shikimate 3-phosphate synthases (EPSP), has been implicated in this biosynthesis (20). MurA and EPSP catalyze the ligation of PEP to the 3Ј-OH of UDP-GlcNAc in peptidoglycan biosynthesis or to the 5Ј-OH of shikimate-3-phosphate, respectively. In an analogous aldoltype reaction, NikO is suggested to catalyze the condensation of PEP with uridine-5Ј-aldehyde or ribofuranosyl-4-formyl-4-imidazolone-5Ј-aldehyde. This reaction is also similar to those catalyzed by the deoxyheptulose-7-phosphate, deoxyoctulosonate-8-phosphate, and deoxynonulosonate-9-phosphate synthases involved in the biosynthesis of 7-, 8-, and 9-carbon long-chain sugars, respectively (6). A similar mechanism may be involved in carbon-chain extension of the tunicamine 11-carbon backbone. However, this does not seem to occur via PEP. If the pseudoGalN part of tunicamine (i.e. C6Ј-C11Ј) arose from PEP, the major incorporation of label from [1-13 C]glucose would be expected to occur at C-6Ј and would certainly be significantly greater that at C-11Ј. This is because [1-13 C]glucose is predominantly metabolized to [3-13 C]PEP via glycolysis, which if condensed with uridine-5Јaldehyde, the major incorporation into tunicamine should be expected at C-6Ј. The evidence indicates the reverse; incorporation of label at C-11Ј is 5-fold greater than at C-6Ј (Table II). Hence, [1-13 C]glucose incorporates into the pseudoGalN ring mainly by 6-carbon metabolism, so the label ends up mainly in the anomeric position, i.e. C-11Ј. In addition, label from [6,6-2 H, 2 H]glucose is incorporated into the 6Ј-position of the tunicamine pseudoGalN ring. If this enrichment occurred via PEP, it should be competed out by glycerol in the competitive metabolic experiments. The finding that this does not occur demonstrates that [6,6-2 H, 2 H]glucose is incorporated into tunicamine as a 6-carbon unit, not via PEP. Moreover, any alternative 2-carbon extension mechanism is also excluded, because incorporation would be expected at every second position rather than the selective incorporations observed.
The equivalent incorporation of 13 C-label from [1-13 C]glucose into both the ␣-C1ЈЈ of the GlcNAc residue and the ␤-C11Ј anomeric carbon of the pseudoGalN ring indicates that both are derived from the same metabolic pool without significant differential metabolic processing. Hence, the biosynthesis of both of these residues likely occurs from the same precursor, either UDP-GlcNAc or GlcNAc. This might implicate GlcNAc-␣,␤-1,1Ј-GlcNAc disaccharide as an intermediate during tunicamycin biosynthesis, analogous to trehalose or sucrose biosynthesis. The ␣,␤-1,1Ј-linked disaccharide could be selectively 4-epimerized and 5,6-dehydrated on the ␤-linked residue prior to coupling to the uridine/ribose moiety to generate the 11carbon sugar. In this case, however, one GlcNAc residue must first be activated as a sugar nucleotide, which would lead to non-equivalent incorporation of 13 C label. This is contrary to what was actually observed, and it is therefore more likely that the tunicamycin ␣-1ЈЈ-GlcNAc and the tunicamine are both derived from the same metabolic pool of UDP-GlcNAc. One possibility is that UDP-GlcNAc is converted by 4-epimerase, and 5,6-dehydratase activities to UDP-4-keto-5,6-ene-N-acetylhexosamine, similar to intermediates formed during 6-deoxysugar nucleotide metabolism (21).
The proposed condensation of UDP-4-keto-GlcNAc-5,6-ene intermediate with uridine-5Ј-aldehyde requires that it be first converted to a nucleophile. This may occur by hydride addition at C5 to form a carbanion at C6. Alternatively, pyranose ring opening may occur, generating a 5,6-enolate ion resonancestabilized with the 5-keto-6-carbanion. Nucleophilic attack can then occur on uridine-5Ј-aldehyde to generate the new carboncarbon bond. Pseudopyranosyl ring closure can only occur after reduction of the de novo C7 keto group to a 7-OH group. A subsequent reductive epimerization of the 4-keto gives the galactopyranosyl configuration. Thus, the 5,6-ene sugar nucleotide intermediate potentially undergoes an aldolase-catalyzed attack on uridine-5Ј-aldehyde to form the tunicamine-uracil core. A second equivalent of UDP-GlcNAc is then used to add the ␣-1ЈЈ-linked GlcNAc residue (Fig. 6). This sequence of events indicates that N-acetyltunicamine-uracil acts as the glycosidic acceptor and UDP-GlcNAc as the glycosidic donor during biosynthesis of the unique head-to-head glycosidic bond in a step catalyzed by an ␣␤-1,1 specific glycosyltransferase. Further experiments to establish the biosynthesis of tunicamycin at the enzymatic level are presently in progress.