Sugar Nucleotide Pools of Trypanosoma brucei, Trypanosoma cruzi, and Leishmania major

Abstract

The cell surface glycoconjugates of trypanosomatid parasites are intimately involved in parasite survival, infectivity, and virulence in their insect vectors and mammalian hosts. Although there is a considerable body of work describing their structure, biosynthesis, and function, little is known about the sugar nucleotide pools that fuel their biosynthesis. In order to identify and quantify parasite sugar nucleotides, we developed an analytical method based on liquid chromatography-electrospray ionization-tandem mass spectrometry using multiple reaction monitoring. This method was applied to the bloodstream and procyclic forms of Trypanosoma brucei, the epimastigote form of T. cruzi, and the promastigote form of Leishmania major. Five sugar nucleotides, GDP-α-d-mannose, UDP-α-d-N-acetylglucosamine, UDP-α-d-glucose, UDP-α-galactopyranose, and GDP-β-l-fucose, were common to all three species; one, UDP-α-d-galactofuranose, was common to T. cruzi and L. major; three, UDP-β-l-rhamnopyranose, UDP-α-d-xylose, and UDP-α-d-glucuronic acid, were found only in T. cruzi; and one, GDP-α-d-arabinopyranose, was found only in L. major. The estimated demands for each monosaccharide suggest that sugar nucleotide pools are turned over at very different rates, from seconds to hours. The sugar nucleotide survey, together with a review of the literature, was used to define the routes to these important metabolites and to annotate relevant genes in the trypanosomatid genomes.

The trypanosomatids are protozoan parasites that impose a major health burden on many countries in the developing world, causing a wide range of diseases over three continents. Like most eukaryotes, the cell surfaces and endosomal/lysosomal systems of these organisms are rich in glycoconjugates, some of which play essential roles in their survival, infectivity, or virulence. The glycoconjugate repertoires of the three trypanosomatids studied here, Trypanosoma brucei, T. cruzi, and Leishmania major, are fundamentally different, reflecting their disparate life cycles, modes of infection, and disease pathologies (see references 5, 27, 28, 38, 47, 65, and 69 and references therein). The monosaccharides that make up these glycoconjugates vary between the three trypanosomatid species. For example, all three contain d-mannose (Man), d-N-acetylglucosamine (GlcNAc), d-glucosamine (GlcN), d-glucose (Glc), and d-galactopyranose (Galp), while only T. cruzi and L. major contain d-galactofuranose (Galf), only T. cruzi contains d-xylose (Xyl), l-rhamnopyranose (Rha), and l-fucose (Fuc), and only L. major contains d-arabinopyranose (Ara) (Table 1). However, a unifying theme of their glycobiology is an abundance of cell surface glycosylphosphatidylinositol (GPI)-anchored glycoproteins and/or non-protein-linked GPI structures.

TABLE 1.

Occurrence of sugars in trypanosomatid glycoconjugates^a

Sugar	T. brucei bloodstream form occurrence of:					T. brucei procyclic form occurrence of:				T. cruzi epimastigote form occurrence of:							L. major promastigote form occurrence of:								Known or presumed donorb
	All	GPI	GIPL	N-glycan	Base J	All	GPI	GIPL	N-glycan	All	GPI	GIPL	N-glycan	O-glycan	P-glycan	Base J	All	GPI	GIPL	LPG	N-glycan	P-glycan	β-Mannan	Base J
α-d-Manp	+	+	+	+	−	+	+	+	+	+	+	+	+	−	+	−	+	+	+	+	+	+	−	−	GDP-Man+Dol- P-Man*
β-d-Manp	+	−	−	+	−	+	−	−	+	+	−	−	+	−	−	−	+	−	−	−	+	−	+	−	GDP-Man
α-d-GlcNAcp	−	−	−	−	−	−	−	−	−	+	−	−	−	+	+	−	−	−	−	−	−	−	−	−	UDP-GlcNAc
β-d-GlcNAcp	+	−	−	+	−	+	+	+	+	+	−	−	+	−	−	−	+	−	−	−	+	−	−	−	UDP-GlcNAc
α-d-GlcNp	+	+	+	−	−	+	+	+	−	+	+	+	−	−	−	−	+	+	+	+	−	−	−	−	UDP-GlcNAc*
α-d-Glcp	+	−	−	+	−	+	−	−	+	+	−	−	+	−	−	−	+	−	−	+	+	−	−	−	UDP-Glc
β-d-Glcp	+	−	−	−	+	−	−	−	−	+	−	−	−	−	−	+	+	−	−	−	−	−	−	+	UDP-Glc
α-d-Galp	+	+	+	+	−	+	+	+	−	+	−	−	−	−	+	−	+	+	+	+	−	−	−	−	UDP-Galp
β-d-Galp	+	+	+	+	−	+	+	+	−	+	−	+	+	+	+	−	+	−	−	+	−	+	−	−	UDP-Galp
β-d-Galf	−	−	−	−	−	−	−	−	−	+	−	+	+	−	+	−	+	+	+	+	−	−	−	−	UDP-Galf
β-d-Xylp	−	−	−	−	−	−	−	−	−	+	−	−	−	−	+	−	−	−	−	−	−	−	−	−	UDP-Xyl
α-l-Rhap	−	−	−	−	−	−	−	−	−	+	−	−	−	−	+	−	−	−	−	−	−	−	−	−	UDP-Rha
β-d-Arap	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	−	+	−	−	+	−	+	−	−	GDP-Ara
α-l-Fucp	−	−	−	−	−	−	−	−	−	+	−	−	+	−	+	−	−	−	−	−	−	−	−	−	GDP-Fuc
α-d-NANA	−	−	−	−	−	+	+	−	−	+	−	−	−	+	−	−	−	−	−	−	−	−	−	−	α2-3 sialosides*

^aThe presence (+) or absence (−) of a particular sugar in a given organism is indicated in the “All” column for that organism; the “All” columns refer to whether or not that sugar appears in any or all of the following specific structural categories, as indicated: protein GPI anchors, free GIPLs, protein N-linked glycans, protein O-linked glycans, protein phosphodiester-linked glycans (P-glycan), and base J.

^b*, known.

The protein-linked GPI glycans of the leishmania are the simplest in structure, consisting of the conserved Manα1-2Manα1-6Manα1-4GlcN core. Those of T. cruzi are principally Manα1-2Manα1-2Manα1-6Manα1-4GlcN, and those of T. brucei are the most complex, with α-d-Galp and β-d-Galp side chains in the bloodstream form and sialylated poly-N-acetyl-lactosamine (poly-LacNAc) side chains in the procyclic form (27).

The non-protein-linked GPI structures include the so-called glycoinositolphospholipids, or GIPLs, that are most abundant in the leishmania and T. cruzi species (58, 64, 90) but are also found in procyclic-form T. brucei (60, 73, 122) and bloodstream form T. brucei (39, 63). Non-protein-linked GPIs also include the more exotic leishmania-specific lipophosphoglycan (LPG) structures (38, 47, 64, 65). The leishmania LPGs contain characteristic phosphosaccharide repeats of Galpβ1-4Manα1-P, which in L. major can have β-d-Galp and α-d-Arap side chains (66, 67). These repeats are also found in the secreted filamentous and GPI-anchored membrane proteophosphoglycans (PPGs) of L. major (47). The phosphosaccharide repeat chains of leishmania PPGs are attached to protein via Manα1-P-Ser linkages (49). Although considerably less abundant, P-linked glycans have also been found in T. cruzi, in gp72 (40) and the NETNES glycoprotein (61), and in T. congolense GARP (113). They may also exist in the T. brucei gp72 homologue Fla-1 (77). The occurrence of P-linked glycans in nature is highly restricted and, outside the trypanosomatids, has been reported as occurring only in Dictyostelium (108) and Entamoeba (70). Whereas the phosphosaccharides of NETNES are composed of mannose chains only, those of gp72 are extremely complex, containing α-l-Fuc, α-l-Rha, and β-d-Xyl as well as GlcNAc, Galp, and Galf (3, 31, 40). Galf does not occur in mammals but is abundant in T. cruzi and the leishmania, where it occurs in the LPG cores and some GIPLs. On the other hand, for the three trypanosomatids, Fuc, Rha, and Xyl have been found only in T. cruzi. T. cruzi is also unique in expressing abundant O-linked glycans with GlcNAcα1-O-Thr linkages (89). These are found on all life cycle stages in GPI-anchored mucin-like molecules and contain mostly βGalp, which can be α2-3 sialylated by the action of cell surface transialidases (1, 14, 88).

All three of the trypanosomatids express N-linked glycans. The dolichol (Dol) cycles of N-glycan precursor biosynthesis are atypical in trypanosomatids in that they lack the glucosyltransferases (Alg6, Alg8, and Alg10) to make the Glc₃Man₉GlcNAc₂-PP-Dol precursor common to most other eukaryotes (99). The mature Dol-PP oligosaccharide species used for transfer to protein vary according to trypanosomatid species. For example, L. mexicana utilizes biantennary Man₆GlcNAc₂-PP-Dol, whereas T. cruzi utilizes Man₉GlcNAc₂-PP-Dol during most of its life cycle but uses both Man₉GlcNAc₂-PP-Dol and Man₇GlcNAc₂-PP-Dol in its bloodstream trypomastigote stage (82). Procyclic-form T. brucei uses Man₉GlcNAc₂-PP-Dol (2), while bloodstream form T. brucei uses either Man₉GlcNAc₂-PP-Dol or Man₅GlcNAc₂-PP-Dol, depending on the glycosylation site (51). The final, fully processed, mature N-linked glycans of the leishmania are atypical biantennary oligomannose structures, some bearing an αGlc residue (80), while those of T. cruzi tend to be conventional oligomannose- and complex-type structures (83), though some also carry Galf (19) while others are sulfated, and a small fraction contain Fuc (8). The T. brucei structures are predominantly conventional oligomannose-type structures in the procyclic form (116) and a mixture of conventional oligomannose- and complex-type structures (125, 126) and unique giant poly-LacNAc-containing structures in the bloodstream form (5). The latter are located primarily in the flagellar pocket and endosomal/lysosomal system of the parasite.

The information above, summarized in Table 1, defines the glycosyl donor requirements of each organism. Based on precedent, it is likely that most of the glycosyl donors are conventional sugar nucleotides (e.g., GDP-mannose [GDP-Man], GDP-fucose [GDP-Fuc], UDP-N-acetylglucosamine [UDP-GlcNAc], UDP-glucose [UDP-Glc], UDP-galactopyranose [UDP-Galp], UDP-galactofuranose [UDP-Galf], and UDP-xylose [UDP-Xyl]), while our previous work has shown that the donor of the relatively rare sugar Ara is GDP-α-d-Ara (GDP-Ara) (100). The identity of the l-rhamnopyranose donor was unknown, but as a result of this study, we have identified it as UDP-rhamnose (UDP-Rha).

In this paper, we describe a new liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ES-MS/MS) method to identify and quantify sugar nucleotides in cell extracts and its application to three trypanosomatid parasites. The results are presented in the context of our knowledge about the glycobiology of these organisms and their recently completed genomes (11, 25, 50).

MATERIALS AND METHODS

Chemicals and reagents.

Commercially available sugar nucleotide standards UDP-Galp, UDP-Glc, UDP-GlcNAc, GDP-Man, GDP-Fuc, and GDP-Glc were purchased from Sigma-Aldrich. UDP-Galf was a kind gift of Andrei Nikolaev (University of Dundee), and UDP-Rha and UDP-Xyl were kind gifts of Ben Davies (University of Oxford). High-performance liquid chromatography (HPLC)-grade solvents were from VWR, and sequanal-grade triethylamine was from Pierce. High-grade (40%) sodium hydroxide was from BDH, and “Suprapur” sodium acetate was from Merck. A 200 mM stock solution of triethylammonium acetate (TEAA) buffer was made by adjusting the pH of triethylamine to pH 6 using “Aristar” acetic acid (BDH). The TEAA stock solution was frozen in aliquots at −20°C until use.

Cell culture.

T. brucei bloodstream form parasites (strain 427, variant MITat 1.2) were cultured in HMI-9 medium with 10% fetal calf serum (FCS; PAA Laboratories) (43) at 37°C with 5% CO₂. For extraction of sugar nucleotides, cells were harvested at ∼1 × 10⁶/ml. Total cell volume was assumed to be 52 μl per 10⁹ cells (37). T. brucei procyclic-form parasites were cultured in SDM-79 medium with 10% FCS (13) at 28°C. For extraction of sugar nucleotides, cells were harvested at ∼1 × 10⁷/ml. Total cell volume was assumed to be 48 μl per 10⁹ cells (21). L. major promastigotes (Friedlin strain) were grown in M-199 medium with 10% FCS (16) at 28°C in a shaking incubator. Cells were harvested at 5 × 10⁶/ml for sugar nucleotide analysis. Total cell volume was assumed to be 35 μl per 10⁹ cells (M. McConville, personal communication). T. cruzi epimastigotes (CL Brener strain) were grown in RTH medium (45, 86) supplemented with 10% FCS at 28°C and were harvested at 4 × 10⁶/ml for sugar nucleotide analysis. Total cell volume was assumed to be 55 μl per 10⁹ cells (59). Wild-type CHO (Chinese hamster ovary) cells (a kind gift of Paul Crocker) were grown in Ham's F-12 medium plus 10% FCS. Cells were grown at 37°C, 5% CO₂. Cells were harvested at approximately 50% confluence for sugar nucleotide analysis. Cell volume was estimated as 12 μl per 10⁷ cells (46).

Sugar nucleotide extraction.

Cells were pelleted by centrifugation, washed once in ice-cold phosphate-buffered saline, and then lysed with cold (−20°C) 70% ethanol on ice in the presence of a 20 pmol GDP-glucose internal standard (Sigma). The typical numbers of cells used per analysis were 1 × 10⁸ for bloodstream form T. brucei, 5 × 10⁷ for procyclic-form T. brucei and promastigote form L. major, 1 × 10⁷ for epimastigote form T. cruzi, and 1 × 10⁶ for CHO cells. The lysate was centrifuged at 16,000 × g for 10 min at 4°C to remove insoluble material, and the supernatant was dried under nitrogen. The samples were dissolved in 200 μl of 9% butan-1-ol and extracted three times with 400 μl of 90% butan-1-ol to remove lipids. The resulting aqueous phase was dried under nitrogen and resuspended in 5 mM ammonium bicarbonate, and sugar nucleotides were extracted using EnviCarb graphitized carbon columns (Supleco) as previously described (92). Briefly, columns were prepared by washing with 3 ml of 80% acetonitrile, 0.1% trifluoroacetic acid followed by 2 ml water. The sample was loaded in 5 mM ammonium bicarbonate, and the column was washed with 2 ml water, 2 ml of 25% acetonitrile, and 2 ml 50 mM TEAA buffer and eluted with 2 ml of 25% acetonitrile, 50 mM TEAA buffer. The eluate was freeze-dried and stored at −80°C prior to analysis.

Reverse-phase HPLC of sugar nucleotides.

HPLC conditions were adapted from reference 92. A 1- by 250-mm C₁₈ column (HiChrom) was used. The initial conditions were 0.5% acetonitrile, 20 mM TEAA buffer (pH 6) for 15 min at a flow rate of 20 μl/min followed by a linear gradient from 0.5 to 4% acetonitrile in 20 mM TEAA buffer over 20 min at a flow rate of 25 μl/min. The column was then washed for 20 min in 4% acetonitrile, 20 mM TEAA buffer before reequilibration into 0.5% acetonitrile, 20 mM TEAA buffer before running the next sample.

ES-MS of sugar nucleotides.

ES-MS was performed on a Quattro Ultima triple quadrupole mass spectrometer (Waters) in negative ion mode using a megaflow source with a capillary voltage of 3 kV and a cone voltage of 70 V. To establish mass transitions for sugar nucleotide standards, samples were introduced at 10 μl per minute using a Harvard syringe pump. Spectra were collected, analyzed, and processed using MassLynx software (Waters). For ES-MS/MS product ion scanning, the argon collision gas pressure was set to 3 × 10⁻³ torr and the collision voltage was 28 V. For LC-MS/MS runs, the flow from the HPLC was made up to 50 μl/min with acetonitrile delivered to a mixing T piece by a Harvard syringe pump. The mass spectrometer was operated in multiple reaction monitoring (MRM) mode for a number of predetermined mass transitions. To accommodate the higher flow rate, the source temperature was raised to 110°C and the desolvation temperature set to 375°C. Chromatograms were recorded for 40 min, and peak areas were integrated automatically using the MassLynx software.

HPAEC of sugar nucleotides.

Sugar nucleotides were separated by high-pH anion-exchange chromatography (HPAEC) using a Dionex LC system and Chromeleon software. HPLC conditions were based upon reference 115, and a 2- by 250-mm Carbopac PA1 column (Dionex) was used at a flow rate of 0.25 ml/min. Sugar nucleotides were detected by UV absorbance at 260 nm. The chromatograms were used to estimate the purity of standard solutions and to determine the relative ratios of sugar nucleotide standards in standard mixtures used to calibrate the LC-MS/MS system.

RESULTS

Establishing an LC-MS/MS method for the identification and quantification of sugar nucleotides.

In order to sensitively and quantitatively survey the sugar nucleotide pools of trypanosomatid parasites, we devised an appropriate extraction and analysis protocol. The criteria were that the method should be (i) sensitive and quantitative with a wide dynamic range (to allow the simultaneous detection and quantification of abundant and trace sugar nucleotides), (ii) insensitive to non-sugar nucleotide contaminants (which often confound UV detection methods), and (iii) capable of the unambiguous identification of 10 sugar nucleotides (UDP-Glc, UDP-Galp, UDP-Galf, UDP-GlcNAc, UDP-Rha, UDP-Xyl, GDP-Glc, GDP-Man, GDP-Fuc, and UDP-GlcA). To satisfy the first two criteria, we elected to use an LC-ES-MS/MS system in MRM mode. MRM methods are highly selective for the detection of compounds of interest in complex mixtures, where they may be relatively minor components. In essence, a compound is detected based upon both the mass of the whole molecule and the mass of one or more of its fragments. Thus, only when a molecular ion gives rise to a specific fragment ion is the ion recorded and assigned. To establish the so-called MRM transitions for each sugar nucleotide, sugar nucleotide standards were analyzed by negative ion ES-MS to find the [M-H]⁻ precursor ion, and that ion was then selected and subjected to collision-induced dissociation, yielding an ES-MS/MS product ion spectrum that was used to choose a prominent product ion. For example, GDP-Fuc is observed as the [M-H]⁻ ion at m/z 588 in the ES-MS spectrum that, when fragmented, produces a major product ion at m/z 442 corresponding to a [GDP-H]⁻ ion (see Fig. S1 in the supplemental material). This defines the MRM transition for GDP-Fuc as m/z 588 to m/z 442. The MRM transitions for the sugar nucleotides analyzed in this work are shown in Table 2. In all cases, the major product ion was a nucleotide, indicating that fragmentation took place around the chemically unstable sugar-1-phosphate or pyrophosphate bonds. There is a clear trend for sugar nucleotides with axial sugar C-2 hydroxyl groups to give rise to nucleotide diphosphate (NDP) fragments (probably assisted by the formation of sugar-1,2-epoxides and the elimination of the NDP-leaving group) and for sugar nucleotides with equatorial sugar C-2 hydroxyl groups to give rise to nucleotide monophosphate fragments (probably assisted by the formation of sugar-1,2-cyclic phosphates and the consequent elimination of nucleotide monophosphate). In the case of UDP-GlcNAc, which does not have a sugar C-2 hydroxyl group, the elimination of UDP is probably assisted by the formation of an oxazoline ion between the C-2 N-acetamido group and the C-1 position of the sugar.

TABLE 2.

MRM transitions for the sugar nucleotides analyzed in this work

Sugar nucleotide	MRM transition	Major production
UDP-Glc	565 → 323	UMP-H
UDP-Galp	565 → 323	UMP-H
UDP-GlcNAc	606 → 385	UDP-H-water
GDP-Man	604 → 424	GDP-H-water
GDP-Fuc	588 → 442	GDP-H
UDP-Galf	565 → 323	UMP-H
GDP-Ara	574 → 442	GDP-H
UDP-Rha	549 → 323	UMP-H
UDP-Xyl	535 → 323	UMP-H
GDP-Glc	604 → 362	GMP-H
UDP-GlcUA	579 → 403	UDP-H

Our requirement for an HPLC separation of sugar nucleotides that was compatible with MS led us to adapt an existing method that uses reverse-phase HPLC with the volatile ion-pairing agent TEAA (92). Additional solvent was added postcolumn to produce a solvent content compatible with stable ES ionization (approximately 50% acetonitrile).

In order to make our analyses quantitative, we added a known amount of an internal standard (GDP-Glc, a sugar nucleotide that is not present in trypanosomatid sugar nucleotide extracts) at the point of cell extraction. The molar relative response factors (MRRFs) relative to this internal standard can vary slightly from day to day due to ion source and collision cell conditions. Therefore, we determined the MRRF values for each batch of analyses by running a mixture of commercially available sugar nucleotides (see Fig. S2A in the supplemental material). The absolute proportions of sugar nucleotides in the mixture were determined by HPAEC using UV absorbance at 260 nm for detection (see Fig. S2C in the supplemental material). Sugar nucleotide levels in cell extracts were then calculated using the following formula: quantity of x = peak area of x/peak area of GDP-Glc × pmoles of GDP-Glc added to extract/MRRF.

To assess the linearity of the method, different amounts of the sugar nucleotide mixture were analyzed by LC-ES-MS/MS (see Fig. S3 in the supplemental material). The quantification of GDP-Man, GDP-Fuc, and UDP-GlcNAc showed excellent linearity and UDP-Gal and UDP-Glc showed only slight deviations from linearity, due to the fact that these two closely related sugar nucleotides are not resolved to baseline by the HPLC method. Operationally, the LC-ES-MS/MS method had a limit of detection for a given sugar nucleotide of about 1 pmol and was linear over a range of 10 to 1,000 pmol per injection.

We validated the method by analyzing the sugar nucleotide levels in a CHO mammalian cell line and comparing our results with those previously obtained by HPAEC and UV detection (115) (see Table S1 in the supplemental material). Our values are very similar to those previously reported, with the exceptions of our identification of UDP-Xyl, which has not been previously detected in CHO cell sugar nucleotide analyses, and of the figure for GDP-Man, where our measurement is fourfold lower that that previously obtained. The reason for the GDP-Man discrepancy is unclear but may be due to confounding UV-absorbing material coeluting with this low-abundance sugar nucleotide in the previously published HPAEC system.

Measurement of sugar nucleotides in trypanosomatid parasites.

There is evidence that sugar nucleotide concentration can change with the growth phase of cells (55). Consequently, only cells that were in the mid- to late log phase were used for sugar nucleotide analysis. For each cell type, at least four separate samples were analyzed, and the mean ± standard deviation for each sugar nucleotide was recorded (Table 3). Since the variance for the analytical method itself is small (the standard deviation for triplicate analyses is typically ±1%), the deviations from the means presumably represent errors introduced by the extraction part of the method.

TABLE 3.

Comparison of sugar nucleotide levels, concentrations, demands, and minimum turnover rates in the three trypanosomatids^a

Sugar nucleotide	T. brucei bloodstream form				T. brucei procyclic form				T. cruzi epimastigotes				L. major promastigotes
	Pool size		Demand and turnover		Pool size		Demand and turnover		Pool size		Demand and turnover		Pool size		Demand and turnover
	Conc (μM)	Amt (pmoles/107 cells)	Demand (pmoles/107 cells)	Time to turn over pool	Conc (μM)	Amt (pmoles/107 cells)	Demand (pmoles/107 cells)	Time to turn over pool	Conc (μM)	Amt (pmoles/107 cells)	Demand (pmoles/107 cells)	Time to turn over pool	Conc (μM)	Amt (pmoles/107 cells)	Demand (pmoles/107 cells)	Time to turn over pool
UDP-Glc	237 ± 33	123 ± 17	?	?	231 ± 26	111 ± 13	?	?	979 ± 230	538 ± 126	?	?	390 ± 71	136 ± 25	?	?
UDP-Galp	105 ± 25	55 ± 13	833	20 min	104 ± 17	50 ± 8	310	48 min	223 ± 40	123 ± 22	600	1 h	198 ± 29	69 ± 10	5830	3.6 min
UDP-GlcNAc	154 ± 39	80 ± 20	917	26 min	341 ± 30	164 ± 14	350	2.3 h	415 ± 16	229 ± 9	516	2.2 h	245 ± 33	86 ± 11	330	1.3 h
GDP-Man	11 ± 7	5.6 ± 3.9	2500	40 s	8 ± 1	3.6 ± 0.5	450	2.4 min	44 ± 4	24 ± 2.4	966	7.5 min	55 ± 8	19 ± 3	11300	32 sec
GDP-Fuc	0.3 ± 0.08	0.15 ± 0.04	?	?	0.3 ± 0.06	0.15 ± 0.03	?	?	1 ± 0.3	0.6 ± 0.15	18	10 min	1.3 ± 0.15	0.46 ± 0.05	?	?
UDP-Galf									14 ± 2.5	8 ± 1.4	166	14 min	15 ± 1.7	5.3 ± 0.6	330	4.8 min
GDP-Ara													3 ± 0.3	1 ± 0.1	250	1.2 min
UDP-Rha									35 ± 1	19 ± 0.6	18	5.4 h
UDP-Xyl									5 ± 0.3	3 ± 0.15	18	48 min
UDP-GlcA									1.1 ± 0.1	0.61 ± 0.05	?	?

^aThe concentrations of the sugar nucleotides were estimated by assuming that the metabolites are evenly distributed throughout the volume of the cell. The demand for each sugar nucleotide is a minimum value (that does not take into account macromolecular turnover) based on the estimated cellular content of the relevant sugar in the major cell glycoconjugates and assuming a mean doubling time for all cell types of 5 h. The contributions of low-abundance glycoconjugates are not taken into account where these would have little effect on the estimates. The assumptions made are as follows. For bloodstream form T. brucei, the figures are based on 10⁷ VSG molecules per cell with an average composition (based on VSG variant 221) of Man₁₅GlcNAc/GlcN_5.5Galp₅ (reference 51 and references therein). For procyclic-form T. brucei, the figures are based on 3 × 10⁶ EP-procyclin molecules per cell, with an average composition of Man₈GlcNAc/GlcN₁₀,Galp₉ (116), and 1 × 10⁶ GIPLs per cell, with an average composition of Man₃GlcNAc/GlcN₅Galp₄ (reference 96 and references therein). For CL Brener strain epimastigote form T. cruzi, the figures are based on 3.8 × 10⁶ mucin molecules per cell, with an average composition of Man₆GlcNAc/GlcN₇Galp₁₂, 1 × 10⁷ GIPLs, with an average composition of Man₄GlcNAc/GlcN₁Galf₁ (reference 62 and references therein), and 35,000 gp72 molecules per cell (106), with an average composition of Xyl₃₀Rha₃₀Fuc₃₀ (3, 29, 31). For procyclic promastigote form L. major, the figures are based on 1 × 10⁷ LPG molecules (half cell associated and half shed into the medium) per cell, with an average composition of Man₁₈GlcNAc/GlcN₁Galp₃₄Galf₁Ara_1.5, 1 × 10⁷ GIPLs, with an average composition of Man₂GlcNAc/GlcN₁Galp₁Galf₁ (64), and 8.0 nmol mannose in the form of accumulated β-mannan per cell (J. E. Ralton and M. McConville, personal communication). ?, unknown.

Representative MRM traces for bloodstream form T. brucei (Fig. 1A), promastigote form L. major (Fig. 1B) and epimastigote form T. cruzi (Fig. 1C) cells show the sugar nucleotides that we were able to detect in these organisms. Five sugar nucleotides were detected in all the organisms analyzed (UDP-Glc, UDP-Galp, GDP-Man, UDP-GlcNAc, and GDP-Fuc) together with the UDP-Galf and a GDP-pentose (assumed to be GDP-Ara) in L. major and UDP-Rha, UDP-Xyl, and UDP-Galf in T. cruzi. The yield of each sugar nucleotide (pmoles per 10⁷ cells) and the estimated cellular concentration, based on cell volume and assuming free distribution of the metabolites, are shown in Table 3.

FIG. 1.

Representative LC-ES-MS/MS MRM chromatograms of sugar nucleotide extracts. Sugar nucleotides were obtained from bloodstream form T. brucei (A), promastigote form L. major (B), and epimastigote form T. cruzi (C). The UDP-Galf peaks are amplified 15-fold. The GDP-Glc peaks are the internal standard added to allow quantification. Two peaks appear in the GDP-glucose channel due to “bleed-through” of GDP-Man, which undergoes the same mass transition.

DISCUSSION

A quantitative method for sugar nucleotide analysis.

There are several methods available in the literature to extract and analyze sugar nucleotides from cells (55, 92, 94, 107, 115). These methods use a variety of different techniques for the separation, detection, and characterization of sugar nucleotides. In some of the methods, sugar nucleotides are detected using UV absorbance at 260 nm and quantified based upon standard curves. In other methods, MS and nuclear magnetic resonance are used either online or offline to structurally identify and characterize sugar nucleotides or biosynthetic intermediates. However, none of the existing methods combines the power of MS for definitive metabolite identification with a quantitative analysis. For these reasons, we felt that there was scope for producing an LC-ES-MS/MS method that would be able to quantify multiple sugar nucleotides extracted from different organisms. A major benefit of the MRM-based approach that we took is that it can provide identification and quantification of coeluting sugar nucleotides as long as they have different masses. This is important because many of the existing methods that quantify sugar nucleotides are hampered by a lack of HPLC separation between these closely related metabolites. The MRM-based approach is also very sensitive because the mass spectrometer spends time looking for ions of interest rather than scanning an entire mass range. For these reasons, we believe that the MRM method is a good tool for detecting and quantifying low-abundance sugar nucleotides, such as UDP-Galf, UDP-Xyl, GDP-Ara, and GDP-Fuc, from relatively small culture volumes (10 to 100 ml). The method has already proved useful in characterizing sugar nucleotide levels for two UDP-Gal 4′-epimerase-deficient cell lines (62, 120) and will be applied to other trypanosome mutants to monitor the effects on sugar nucleotide levels.

Overview of the results.

Qualitatively, the sugar nucleotide contents of the three trypanosomatids obtained with the LC-ES-MS/MS method agree well with the monosaccharide contents of known trypanosomatid glycoconjugates (Table 1) and/or the presence of known or candidate sugar nucleotide biosynthesis enzymes (Fig. 2 and Table 4). In some cases, the metabolite identification data enable us to improve the annotations of genes likely to be involved in sugar nucleotide biosynthesis (Table 4). The concentrations of the various sugar nucleotides measured in this study reflect their steady-state pool sizes and do not appear to reflect the demand for these sugars (Table 3). In the following sections, we review the data for each sugar nucleotide pool obtained in this study and integrate it with our current knowledge of glycobiology and genomes of T. brucei, T. cruzi, and L. major.

FIG. 2.

Summary of the sugar nucleotide biosynthesis pathways in trypanosomatid parasites. The sugar nucleotides used for glycoconjugate biosynthesis are in bold. Sugar nucleotide biosynthetic intermediates are in parentheses. Salvage pathways are in italics. The numbers refer to the enzymes and known or candidate genes described in Table 4. The colors of the arrows and underlining indicate the presence of the corresponding enzyme and sugar nucleotide, respectively, in T. brucei (orange), T. cruzi (green), and L. major (red).

TABLE 4.

Enzymes and genes involved in trypanosomatid sugar nucleotide biosynthesis

Step no.a	Enzyme name	Enzyme no.	Putative/characterized trypanosomatid homologuesb
1	Hexokinase	EC 2.7.1.1	Tb10.70.5820 Tb10.70.5800 LmjF21.0240 LmjF21.0250 Tc00.1047053510121.20 Tc00.1047053508951.20
2	Glucose-6-phosphate isomerase	EC 5.3.1.9	Tb927.1.3830 LmjF12.0530 Tc00.1047053506529.508 Tc00.1047053510889.221
3	Glucosamine-fructose-6-phosphate aminotransferase	EC 2.6.1.16	Tb927.7.5560 LmjF06.0950 Tc00.1047053510303.200
4	Glucosamine-phosphate N-acetyltransferase	EC 2.3.1.4	Tb11.01.2886 LmjF28.3005 Tc00.1047053508831.120 Tc00.1047053511671.70
5	Phosphoacetylglucosamine mutase	EC 5.4.2.3	Tb927.8.980 LmjF07.0805 Tc00.1047053503733.70Tc00.1047053508569.80
6	UDP-N-acetylglucosamine pyrophosphorylase	EC 2.7.7.23	Tb11.02.0120mjF33.2520 Tc00.1047053504153.180
7	Phosphomannose isomerase	EC 5.3.1.8	Tb11.01.6410 LmjF32.1580 Tc00.1047053503677.10Tc00.1047053511717.90
8	Phosphomannomutase	EC 5.4.2.8	Tb10.70.0370 Tb11.01.6410LmjF36.1960Tc00.1047053510187.480
9	Mannose-1-phosphate guanyltransferase	EC 2.7.7.13	Tb927.8.2050LmjF23.0110Tc00.1047053510947.10Tc00.1047053503589.10
10	GDP-mannose 4,6-dehydratase	EC 4.2.1.46	Tb10.61.0880c00.1047053510027.10 Tc00.1047053508897.10 (middle fragment) Tc00.1047053437575.9 (N-terminal fragment) Tc00.1047053471901.5 (N-terminal fragment)
11	GDP-l-fucose synthetase	EC 1.1.1.271	Tb11.01.5560Tc00.1047053511727.300
12	Phosphoglucomutase	EC 5.4.2.2	Activity known in T. brucei; no gene identified LmjF21.0640Tc00.1047053508637.90 Tc00.1047053511911.130
13	UTP glucose-1-phosphate uridylyltransferase	EC 2.7.7.9	Tb10.389.0330 LmjF18.0990 Tc00.1047053503521.50 Tc00.1047053506359.60
14	UDP-galactose 4-epimerase	EC 5.1.3.2	Tb11.02.0330LmjF33.2300Tc00.1047053506811.190Tc00.1047053506303.50
15	UDP-galactopyranose mutase	EC 5.4.99.9	LmjF18.0200Tc00.1047053507993.160Tc00.1047053511277.600
16	Galactokinase	EC 2.7.1.6	LmjF35.2740 Tc00.1047053507001.110 Tc00.1047053507009.40Tc00.1047053508465.90 Tc00.1047053510667.120
17	UTP hexose-1-phosphate uridyltransferase or UDP-glucose hexose-1-phosphate uridylyltransferase	EC 2.7.7.10 or EC 2.7.7.12	No genes identified; however, L. major is thought to process Gal-1-P to UDP-Gal, based on the radiolabeling of Leishmania with [3H]Gal (118)
18	UDP-glucose 4,6-dehydratase	EC 4.2.1.46	Tc00.1047053503881.20 Tc00.1047053508119.120 (assigned as GDP-mannose 4,6-dehydratase) Tc00.1047053511019.84 (N-terminal fragment) LmjF26.2230 (assigned as GDP-mannose 4,6-dehydratase)
19	UDP-4-keto-6-deoxyglucose reductase	EC 1.1.1.133	Tc00.1047053503925.40 Tc00.1047053510349.10
20	UDP-glucose 6-dehydrogenase	EC 1.1.1.22	Tc00.1047053509715.90
21	UDP-glucuronate decarboxylase	EC 4.1.1.35	Tc00.1047053509965.380 Tc00.1047053509051.30 (assigned as dTDP-glucose dehydratase)
22	Pathway unknown		GDP-arabinose can be formed by labeling with [3H]Glc, evidence for decarboxylation at C-1 (100, 101)
23	Fucose (arabinose) kinase	EC 2.7.1.52 or EC 2.7.1.54	LmjF16.0440 LmjF16.0480 Tc00.1047053508379.10 Tc00.1047053508377.9 Tc00.1047053509111.28 Tc00.1047053507091.4 Tc00.1047053508375.89 Tc00.1047053511641.5
24	Arabinose-1-phosphate guanyltransferase	Not assigned	No gene identified; however, L. major is thought to posses this activity based on the Ara → Ara-1-P step (100, 101)

^aStep numbers refer to those shown in Fig. 2.

^bGene numbers in boldface have been functionally characterized, those underlined are as assigned in GeneDB, and those in boldface and underlined are assigned in this paper through BLAST searches with functionally characterized orthologues from other organisms.

GDP-Man biosynthesis and metabolism.

The presence of GDP-Man in T. brucei, T. cruzi, and L. major is consistent with the presence of α-d-Man and β-d-Man in glycoprotein N-linked glycans in all three species, the presence of α-d-Man in free and/or protein-linked GPIs in all three species, and the presence of α-d-Man in LPG and the P-linked glycans of the PPGs of L. major and NETNES of T. cruzi. GDP-Man is the direct donor substrate for many mannosyltransferases (e.g., for the synthesis of the Man₅GlcNAc₂-PP-Dol intermediate of protein N glycosylation), whereas for others it is an obligate intermediate, via Dol-P-Man synthetase, to the Dol-P-Man donor (e.g., for GPI anchor biosynthesis and the conversion of Man₅GlcNAc₂-PP-Dol to Man₉GlcNAc₂-PP-Dol in T. brucei and T. cruzi). GDP-Man is also the precursor of GDP-Fuc (see below).

The de novo pathway of GDP-Man biosynthesis is from glucose-6-phosphate, i.e., Glc-6-P → Fru-6-P → Man-6-P → Man-1-P → GDP-Man, and candidate or characterized genes for all of these steps can be found in the three genomes. However, the ability to biosynthetically radiolabel T. brucei, T. cruzi, and L. major glycoconjugates with [³H]Man (84, 111, 118) illustrates that a salvage pathway, presumably via the action hexokinase (Man → Man-6-P), also exists.

Despite the huge demand for Man by all three parasites (Table 3), the steady-state levels of GDP-Man are low (8 to 55 μM) and the minimum rates of turnover of the GDP-Man pools during logarithmic growth, based on the estimated copy numbers of Man residues per parasite, are in the orders of seconds for bloodstream form T. brucei and promastigote form L. major and minutes for procyclic-form T. brucei and epimastigote form T. cruzi (Table 3). The low steady-state concentrations of the cellular GDP-Man pools and the very high rates of flux through them suggests that inhibition of GDP-Man biosynthesis would have rapid and profound effects on glycoconjugate biosynthesis, and consequently on viability and/or virulence, in these organisms. Certainly, GDP-Man and Dol-P-Man biosynthesis may be assumed to be essential in bloodstream form T. brucei, where disrupting GPI anchor biosynthesis (which is dependent on Dol-P-Man made from GDP-Man) is known to be lethal for the parasite (15, 26, 60, 74). Since Man is present in every known type of trypanosomatid glycoconjugate carbohydrate chain, with the exception of the O-linked side chains of the T. cruzi mucins, GDP-Man biosynthesis has potential as a therapeutic target against all three parasites, so long as parasite-selective inhibitors can be found.

Mannose metabolism has been studied in any detail only for the leishmania. Surprisingly, the promastigote form of the organism can survive in culture in the complete absence of mannose metabolism. Thus, promastigote mutants null for phosphomannose isomerase (Fru-6-P → Man-6-P), phosphomannose mutase (PMM; Man-6-P → Man-1-P), GDP-Man pyrophosphorylase (GDPMP; Man-1-P → GDP-Man), Dol-P-Man synthetase (GDP-Man → Dol-P-Man), and a Golgi GDP-Man transporter (LPG2) were all viable in culture despite the devastating effects these mutations had on the mutants’ cell surface molecular architecture (32-34, 48, 110). Whereas the phosphomannose isomerase and Dol-P-Man synthetase mutants retained the ability to infect macrophages and mice, albeit with reduced virulence, the PMM and GDPMP mutants could not infect macrophages in vitro or in vivo. These results showed that GDP-Man is essential for reasons other than just supplying Dol-P-Man, upon which N glycosylation and GPI anchor, GIPL, and LPG biosynthesis are dependent. The recently characterized β-mannan storage polysaccharide and/or its biosynthetic intermediates appear be the essential element(s) of mannose metabolism for parasite virulence (93, 104). The extremely rapid turnover of GDP-Man pools in L. major promastigotes (Table 3) may be even more acute in intracellular amastigotes, where glucose is limiting and gluconeogenesis from amino acids is necessary to feed GDP-Man synthesis (72). Under such conditions, a mannose storage polysaccharide to buffer metabolic demands for this sugar may be essential. In any case, the essentiality of PMM and GDPMP for parasite virulence in the host suggests that these may be good drug targets against the leishmania. However, the high-resolution crystal structure of L. mexicana PMM shows that the active site is almost indistinguishable from that of the human enzyme, and this is likely to make the search for parasite-selective drug leads difficult (53). Interestingly, despite its structural and enzymatic similarities to L. major PMM, human PMM cannot complement an L. major PMM null mutant for virulence. However, the L. major PMM has a C-terminal PTS1 signal sequence (SKL) that may direct it into the glycosome (81), and mislocation of the human PMM in L. major might be the reason. The GDPMP has been expressed and biochemically characterized (18), but no crystal structure is yet available.

UDP-GlcNAc biosynthesis and metabolism.

The presence of UDP-GlcNAc in T. brucei, T. cruzi, and L. major is consistent with the presence of GlcNAc in glycoprotein N-linked glycans in all species and in O-linked glycans in T. cruzi. In addition, GlcN, derived from GlcNAc by de-N-acetylation (24), is present in all species in protein-linked and free GPI structures, including L. major LPG. In T. brucei, GlcNAc is found in LacNAc repeats of Galpβ1-4GlcNAc. These poly-LacNAc structures are found as part of giant N-linked glycans throughout the flagellar pocket and endosomal/lysosomal system of the bloodstream form (5, 76) and as side chains of the procyclin GPI anchor and GIPLs in the procyclic form (73, 96, 116) of the organism.

Bioinformatic analysis suggests that UDP-GlcNAc, the direct donor for all GlcNAc transferases, is made by the conventional eukaryotic de novo route from glucose-6-phosphate, i.e., Glc-6-P → Fru-6-P → GlcN-6-P → GlcNAc-6-P → GlcNAc-1-P → UDP-GlcNAc. However, the ability to biosynthetically radiolabel T. brucei and L. major glycoconjugates with [³H]GlcN (30, 91) shows that a salvage pathway, presumably via the action of hexokinase (GlcN → GlcN-6-P), also exists. Most likely, the former pathway is the most important in vivo, since free GlcN is not an abundant sugar in either mammals or insects and, at least in T. brucei, its N-acetyl derivative (GlcNAc) is not taken up (6). Since GlcNAc and/or GlcN is present in every known type of trypanosomatid glycoconjugate carbohydrate chain, with the exception of the side chains of leishmania and T. cruzi PPGs, UDP-GlcNAc biosynthesis has potential as a therapeutic target against all three parasites, so long as parasite-selective inhibitors can be found. The essential nature of UDP-GlcNAc pyrophosphorylase to bloodstream form T. brucei has recently been demonstrated by construction of a conditional null mutant (M. Stokes and M. A. J. Ferguson, unpublished data).

UDP-GlcNAc is relatively abundant in all three parasites (ranging from 150 μM in bloodstream form T. brucei to 410 μM in T. cruzi epimastigotes), and the minimum rates of turnover of the UDP-GlcNAc pools during logarithmic growth, based on the estimated copy numbers of GlcNAc residues per parasite, are in the orders of tens of minutes to hours (Table 3). The steady-state levels of UDP-GlcNAc between the parasites do not appear to reflect the demand for GlcNAc between the organisms.

UDP-Glc biosynthesis and metabolism.

UDP-Glc is also found in all three parasites and is the glucosyl donor for the unfolded glycoprotein glucosyltransferase (UGGT) involved in glycoprotein quality control in the endoplasmic reticulum (42, 117). UGGT is known to function in all three parasites (17, 51, 68, 80). UDP-Glc is also the presumed donor for the synthesis of base J (β-d-glucosylhydroxymethyluracil), a rare deoxynucleotide found in telomere-proximal DNA in the bloodstream form of T. brucei and in the leishmania and T. cruzi (121). UDP-Glc is also presumably the precursor of UDP-Rha and UDP-Xyl in T. cruzi and is the obligate precursor of UDP-Galp in T. brucei and T. cruzi (see below). The UDP-Glc pyrophosphorylase from L. major has been cloned, expressed, and enzymatically and structurally characterized (57, 109).

The demand for UDP-Glc must large, because it includes the flux through this pool to UDP-Galp via UDP-Glc 4′-epimerase (see below). However, since we know neither the average numbers of UGGT/glucosidase II-mediated cycles of transient glucosylation/deglucosylation involved in endoplasmic reticulum glycoprotein quality control in these organisms nor the rates of turnover of base J, is difficult to asses the total demand. Nevertheless, the steady-state UDP-Glc pool sizes are relatively large (110 to 540 μM) for all of the cells analyzed, probably because the substrates for the UDP-Glc pyrophosphorylase (i.e., Glc-1-P and UTP) are relatively abundant metabolites.

UDP-Galp biosynthesis and metabolism.

The presence of UDP-Galp in T. brucei, T. cruzi, and L. major is consistent with the presence of Galp in many of their glycoconjugates (Table 1). Furthermore, UDP-Galp is also the obligate precursor for UDP-Galf in T. cruzi and L. major (see below). The only route to UDP-Galp in bloodstream and procyclic-form T. brucei and epimastigote form T. cruzi is via the epimerization of UDP-Glc, because their hexose transporters are unable to transport Gal (9, 112). The UDP-Glc 4′-epimerase (GalE) responsible for this interconversion is essential in both life cycle stages of T. brucei (96, 97, 120) and appears to be essential for T. cruzi epimastigotes (62), suggesting that this enzyme may be a good therapeutic target against African trypanosomiasis and possibly Chagas’ disease.

With respect to therapeutic potential, the need for selective inhibition of trypanosome UDP-Glc 4′-epimerase over the human enzyme may be less acute for than for, say, enzymes of GDP-Man, UDP-GlcNAc, and UDP-Glc biosynthesis. This is because (i) human cells have salvage pathways for UDP-Galp biosynthesis (indeed, Gal-1-P and UDP-Galp accumulate in GalE null mammalian cells [103]) and (ii) although defects in human GalE lead to type III galactosemia (114), inhibition of human GalE over a period of days may be tolerable. In any case, three lines of evidence suggest that some selectivity against trypanosome GalE enzymes should be possible. First, the substrate specificities of the T. brucei and T. cruzi enzymes differ from that of the human enzyme. The latter can interconvert UDP-GlcNAc and UDP-GalNAc, whereas the trypanosome enzymes cannot (95, 97). Second, high-resolution crystal structures of the T. brucei epimerase, and comparison with those of the human counterpart (44), suggest that there is potential for selective inhibition of the parasite enzyme (4, 105). Third, a recent small-scale screen of 880 drug-like molecules revealed three compounds with 50% inhibitory concentration values for T. brucei GalE four times lower than those for human GalE (119).

In procyclic T. brucei and epimastigote T. cruzi, gene replacement of a single GalE allele results in partial galactose starvation, suggesting that GalE levels are close to limiting in these cells (62, 96). Further support for this notion as it pertains to T. cruzi comes from the measured UDP-Galp-to-UDP-Glc ratios. Whereas in T. brucei and L. major, the ratio of UDP-Galp to UDP-Glc is about 1:2.2, reflecting the equilibrium constant of the GalE epimerase, in T. cruzi the ratio is 1:4.3 (Table 3), suggesting that conversion of UDP-Glc to UDP-Galp is rate limiting in this organism. This may explain the profound cell surface glycoconjugate and morphological changes that occur when a single allele of T. cruzi GalE is replaced (62).

UDP-Glc is probably also a major route to UDP-Galp in L. major, but this organism is unique among the three trypanosomatids in being able to both take up and incorporate [³H]Gal into its glycoconjugates (118), demonstrating that a salvage pathway to UDP-Galp exists in L. major. This suggests that the putative L. major galactokinase gene is functional, although we were unable to find an obvious galactose-1-phosphate uridylyltransferase (EC 2.7.7.10) or UDP-Glc hexose-1-phosphate uridylyltransferase (EC 2.7.7.12) that might complete a UDP-Gal salvage pathway in this organism. Interestingly, T. brucei has a galactokinase pseudogene that might have degenerated when the organism lost its ability to take up Gal. Putative galactokinase genes also exist in T. cruzi, even though this organism also cannot take up Gal. It is possible that one or more of these may be involved in salvage pathways for other sugars in this organism.

UDP-Galf biosynthesis and metabolism.

Both L. major and T. cruzi express abundant Galf-containing molecules, and UDP-Galf is found in both organisms. UDP-Galf is made from UDP-Galp by the action of UDP-Gal mutase, and L. major and T. cruzi contain functional and putative UDP-Gal mutase genes, respectively (7, 12). T. brucei lacks a putative UDP-Gal mutase gene and contains neither UDP-Galf nor Galf-containing glycoconjugates.

The steady-state pool sizes of UDP-Galf in L. major and T. cruzi are extremely small, but they are in line with the reported equilibrium constant of the Escherichia coli UDP-Gal mutase (75). Thus, the ratio of UDP-Galp to UDP-Galf in both organisms is close to 14:1. Given the high levels of Galf in the glycoconjugates of both organisms, the turnover rate of the UDP-Galf pool must be in the order of minutes. Interestingly, when one allele of GalE is replaced in epimastigote form T. cruzi, the cells no longer express significant amounts of the Galp-rich GPI anchored mucins but retain most of their Galf-containing GIPLs (62). In order to achieve this preferential retention of Galf metabolism, the pool size of UDP-Galf (and therefore that of UDP-Galp) is not reduced significantly. Presumably, the flux through the UDP-Galp pool to UDP-Galp transporters/Galp transferases is reduced to maintain flux through the UDP-Galf pool.

The lack of Galf in mammals has led to the suggestion that the biosynthesis and metabolism of UDP-Galf might be potential therapeutic targets against the leishmania and T. cruzi, as well as other microbial pathogens (22, 85). However, a recent report by Kleczka and colleagues shows that L. major UDP-Galf mutase null mutants are still infectious to mice, albeit with reduced virulence (54). In the case of T. cruzi, anti-Galf antibodies have been reported to inhibit the invasion of host cells (19), but the relative importance of Galf, which is most abundant in the insect-dwelling epimastigote stage, in parasite survival and/or propagation in the mammalian host is unknown.

UDP-Xyl and UDP-Rha biosynthesis and metabolism.

Of the three trypanosomatids, only T. cruzi is known to express glycoproteins that contain Xyl and Rha and, consistent with this distribution, UDP-Xyl and UDP-Rha appear only in T. cruzi. The sugars are found as β-d-Xyl and α-l-Rhap in the novel low-abundance Ser/Thr-phosphate-linked carbohydrate chains recognized by the T. cruzi-specific monoclonal antibody WIC29.26 (3, 29, 40), which are found principally, but not exclusively, on gp72, a glycoprotein associated with flagellar attachment to the cell body (20). UDP-Xyl is the only known xylose sugar nucleotide in nature, and its biosynthesis goes from UDP-Glc via UDP-glucuronic acid (UDP-GlcUA). Genes for both of these steps can be found in the T. cruzi genome, and the intermediate sugar nucleotide (UDP-GlcUA) was also found in our analyses (Table 3). Whether UDP-GlcUA has any function other than as an intermediate on the pathway to UDP-Xyl is unclear at the moment, but the T. cruzi genome does not contain any identifiable putative glucuronosyltransferases.

l-Rhamnose is found both as UDP-Rha, described for plants and made from UDP-Glc (52, 79, 123), and as dTDP-Rha, described for bacteria and made from dTDP-Glc (35). The results reported here show that T. cruzi makes UDP-Rha, presumably from UDP-Glc, rather than dTDP-Rha from dTDP-Glc. We therefore suggest corrections to the GeneDB annotations, which suggest a dTDP-Rha pathway (Table 4). This is a common problem in the genome databases, where eukaryotic genes encoding dehydratases and epimerase/reductases likely to be involved in GDP-Fuc or UDP-Rha biosynthesis have, through automated annotation, been assigned to prokaryote-specific dTDP-Rha biosynthesis. In the plant Arabidopsis thaliana, the three enzymatic activities of UDP-Rha biosynthesis are located in a single polypeptide (RHM2/MUM4), with the UDP-Glc 4,6-dehydratase activity located in the N-terminal domain (residues 1 to 370) and the UDP-4-keto-6-deoxy-Glc epimerase and UDP-4-keto-Rha reductase activities located in the C-terminal domain (residues 371 to 667) (79, 123). In T. cruzi, these domains appear to belong to separate polypeptides. Presumably, one of the several putative T. cruzi NDP-sugar 4,6-dehydratases performs the first step, and the products of one or both of two closely related genes that show significant homology to the C-terminal domain of RHM2/MUM4 perform the epimerase and reduction steps (Table 4).

The presence and maintenance of multistep de novo biosynthetic pathways to UDP-Xyl and UDP-Rha in T. cruzi suggest that these sugars may play an important role in the biology of this organism. It is known that gp72 is important for flagellar adhesion in epimastigote form T. cruzi (20) and that gp72 null parasites have reduced virulence in both the insect vector and the mammalian host (10). However, our knowledge of gp72 carbohydrate structure does not extend beyond the epimastigote form, and it is clear that gp72 glycosylation and the expression of the WIC29.26 epitope is stage specific (40, 41). Thus, at the moment it is difficult to deduce the precise function(s) of Xyl and Rha metabolism in T. cruzi.

GDP-Ara biosynthesis and metabolism.

Only L. major and Crithidia fasciculata are known to express arabinose in the form of α-d-Arap. This sugar is found in L. major LPG and in the related C. fasciculata lipoarabinogalactan (66, 102). Correspondingly, GDP-Ara was found only in L. major in this study.

The source of GDP-α-d-Arap in L. major is not fully understood. It is known that L. major and C. fasciculata can convert d-Glc to d-Ara via the loss of the Glc C-1 carbon atom and that they can also take up exogenous Ara (100, 101). A 125-kDa C. fasciculata d-arabino-1-kinase activity has been characterized but not cloned (100), suggesting that the route is Glc → Ara → Ara1-P → GDP-Ara (Fig. 2). The C. fasciculata d-arabino-1-kinase also has fucose-1-kinase activity (d-Arap and l-Fucp differ by only a methyl group), and we suggest that the two genes annotated as putative L. major fucose kinase genes in GeneDB should be reclassified as putative arabinose/fucose kinase genes (Table 4). The proposed route to GDP-Ara requires a GDP-Ara pyrophosphorylase activity, but no candidate genes (e.g., homologues of mammalian GDP-Fuc pyrophosphorylase genes) are apparent in the L. major genome.

In L. major promastigotes, the proportion of phosphosaccharide repeats with β-d-Arap-terminating side chains is under developmental control, such that levels of Ara are low in logarithmically growing procyclic promastigotes and significantly higher in metacyclic promastigotes (67, 87). Studies on the arabinosyltransferases in L. major have led to the suggestion that the control of arabinosylation may be at the level of GDP-Ara availability rather than at that of arabinosyltransferase expression (23). Certainly, GDP-Ara levels were low in the procyclic promastigotes studied here.

GDP-Fuc biosynthesis and metabolism.

The fucose donor GDP-Fuc was found in all three trypanosomatids, even though Fuc has been identified thus far only in the aforementioned novel low-abundance Ser/Thr-phosphate-linked carbohydrate chains of T. cruzi gp72 (3, 29, 40). On the other hand, the presence of GDP-Fuc in T. brucei is not so surprising, given that the genomes of T. cruzi and T. brucei both contain homologues of the two enzymes, GDP-Man 4,6-dehydratase and the combined GDP-4-keto-6-deoxy-Man 3,6-epimerase and GDP-4-keto-Fuc reductase, needed for de novo synthesis of GDP-Fuc from GDP-Man (Fig. 2). T. brucei also contains homologues (Fla-1 and Fla-2) of the fucosylated T. cruzi glycoprotein gp72. Fla-1 and Fla-2 are also involved in flagellar attachment (56, 71, 77), although they do not react with WIC29.26, possibly because of the lack of Xyl and Rha in this organism (see above). Thus, the location and function of fucose in T. brucei remain enigmatic. The de novo pathway is probably the only source of GDP-Fuc in T. brucei, which lacks obvious fucose kinase and GDP-Fuc pyrophosphorylase salvage pathway genes.

In the case of L. major, there are no candidate genes for the de novo synthesis of GDP-Fuc from GDP-Man, and Fuc has not been directly identified for any leishmania glycoconjugate isolated from normally cultured cells. However, it is known that L. major promastigotes cultured in the presence of relatively high concentrations (50 mM) of Fuc can substitute l-Fuc for d-Arap in LPG side chains (124) and that [³H]GDP-β-l-Fuc can be isolated from L. major labeled with [6-³H]Fuc (78). It seems likely, therefore, that the d-arbino-1-kinase and putative GDP-Ara pyrophosphorylase that synthesize GDP-Ara in L. major are also capable of making GDP-Fuc from fucose taken up from the medium. The source of fucose is probably the 10% FCS used to culture L. major promastigotes. Mammalian sera generally contain low levels (<30 μM) of free fucose from the degradation of abundant fucose-containing serum glycoproteins (98). It is noteworthy that although the GDP-Fuc levels are only half of the GDP-Ara levels in L. major promastigotes (Table 2), only Ara is found in LPG under these culture conditions (66, 67, 124). Presumably, the GDP-Ara transporter and/or the GDP-Ara:LPG β1-2 arabinosyltransferases (23, 36) show a significant preference for GDP-Ara over GDP-Fuc, and the presence of GDP-Fuc in L. major promastigotes may have no physiological significance.