The lipid-linked oligosaccharide donor specificities of Trypanosoma brucei oligosaccharyltransferases.

We recently presented a model for site-specific protein N-glycosylation in Trypanosoma brucei whereby the TbSTT3A oligosaccharyltransferase (OST) first selectively transfers biantennary Man(5)GlcNAc(2) from the lipid-linked oligosaccharide (LLO) donor Man(5)GlcNAc(2)-PP-Dol to N-glycosylation sequons in acidic to neutral peptide sequences and TbSTT3B selectively transfers triantennary Man(9)GlcNAc(2) to any remaining sequons. In this paper, we investigate the specificities of the two OSTs for their preferred LLO donors by glycotyping the variant surface glycoprotein (VSG) synthesized by bloodstream-form T. brucei TbALG12 null mutants. The TbALG12 gene encodes the α1-6-mannosyltransferase that converts Man(7)GlcNAc(2)-PP-Dol to Man(8)GlcNAc(2)-PP-Dol. The VSG synthesized by the TbALG12 null mutant in the presence and the absence of α-mannosidase inhibitors was characterized by electrospray mass spectrometry both intact and as pronase glycopetides. The results show that TbSTT3A is able to transfer Man(7)GlcNAc(2) as well as Man(5)GlcNAc(2) to its preferred acidic glycosylation site at Asn263 and that, in the absence of Man(9)GlcNAc(2)-PP-Dol, TbSTT3B transfers both Man(7)GlcNAc(2) and Man(5)GlcNAc(2) to the remaining site at Asn428, albeit with low efficiency. These data suggest that the preferences of TbSTT3A and TbSTT3B for their LLO donors are based on the c-branch of the Man(9)GlcNAc(2) oligosaccharide, such that the presence of the c-branch prevents recognition and/or transfer by TbSTT3A, whereas the presence of the c-branch enhances recognition and/or transfer by TbSTT3B.

Introduction

The African trypanosomes are tsetse fly-transmitted protozoan parasites. Two lifecycle stages are most amenable to laboratory cultivation: the procyclic form that normally grows in the tsetse fly midgut and the bloodstream form that causes African sleeping sickness in humans and nagana in cattle. The procyclic form of the parasite presents a coat that includes a set of glycosylphosphatidylinositol (GPI)-anchored glycoproteins known as procyclins, characterized by internal dipeptide (EP) or pentapeptide (GPEET) repeats (Roditi et al. 1989; Treumann et al. 1997), free GPI structures (Lillico et al. 2003; Vassella et al. 2003; Nagamune et al. 2004; Roper et al. 2005) and a high-molecular-weight glycoprotein complex (Guther et al. 2009). Most EP procyclins contain a single N-glycosylation site, occupied exclusively by a conventional triantennary Man5GlcNAc2 oligomannose oligosaccharide at the N-terminal side of the EP-repeat domain (Acosta-Serrano et al. 1999). The bloodstream form of the parasite is covered in a densely packed layer of 5 × 106 GPI-anchored variant surface glycoprotein (VSG) dimers. This coat protects the parasites from the alternative pathway of complement-mediated lysis, shields other cell surface proteins from the host immune system and, by the process of antigenic variation, allows the parasites to persist indefinitely in the host bloodstream (Cross 1996; Pays and Nolan 1998). The trypanosome genome contains several hundred silent VSG genes, some of which are pseudogenes, and antigenic variation involves the switching of the expression of these genes, which encode immunologically distinct GPI-anchored glycoproteins with 1–3 N-glycosylation sites (Mehlert et al. 1998; Hutchinson et al. 2003; Marcello and Barry 2007). The bloodstream-form parasite also expresses other less abundant GPI-anchored and/or N-glycosylated glycoproteins that are arranged either randomly in the VSG coat, like the invariant glycoproteins ISG65 and ISG75 (Ziegelbauer and Overath 1992), or in specific surface locations such as Fla1 which is localized to the flagellar adhesion zone (Nozaki et al. 1996) and the transferrin receptor which is localized to the flagellar pocket (Steverding et al. 1994). Other glycoproteins are located primarily in intracellular sites, like lysosomal p67 (Kelley et al. 1999), Golgi and lysosomal tGLP1 (Lingnau et al. 1999), endoplasmic reticulum (ER) GPIdeAc (Guther et al. 2003) and endosomal TbMBAP1 (Engstler et al. 2005). Although the procyclic form shares some of these glycoproteins, like p67, tGLP1 and Fla1, others are bloodstream-form specific, such as ISG65, ISG75, TbBMAP1 and the transferrin receptor.

Lipid-linked oligosaccharide (LLO) donors for protein N-glycosylation are assembled in the membrane of the ER. Each of the sugars is added to the growing LLO by asparagine-linked glycosylation (ALG) glycosyltransferases, which are numbered according to the order of their discovery rather than by the sequence of enzymatic steps (Burda and Aebi 1999). Genomic and experimental comparisons have shown that some lower eukaryotes do not possess all the ALG genes needed to make the canonical mature Glc3Man9GlcNAc2-PP-Dol LLO structure typical of most eukaryotes (Parodi 1993; Samuelson et al. 2005). Trypanosoma brucei, for example, lacks the ALG8 and ALG10 glucosyltransferase genes (Jones et al. 2005; Samuelson et al. 2005) such that Man9GlcNAc2-PP-Dol is the largest LLO it can synthesize. Interestingly, the Man5GlcNAc2-PP-Dol intermediate is by far the most abundant LLO in procyclic and bloodstream-form parasites (Low et al. 1991; Acosta-Serrano et al. 2004; Manthri et al. 2008). Uniquely, T. brucei uses both Man5GlcNAc2-PP-Dol and Man9GlcNAc2-PP-Dol to N-glycosylate its glycoproteins (Jones et al. 2005; Manthri et al. 2008; Izquierdo, Schulz, et al. 2009) and this explains why endoglycosidase-H-resistant N-glycans can appear on T. brucei glycoproteins co-translationally, as first noted and discussed by Bangs et al. (1988).

The enzymes that transfer oligosaccharides from the LLOs are called oligosaccharyltransferases (OSTs). These are typically multi-subunit integral membrane protein complexes that mediates the en bloc transfer of the preassembled oligosaccharide onto asparagine in glycosylation sequons (mostly Asn-Xaa-Thr/Ser; Xaa ≠ Pro) of nascent polypeptides entering the lumen of the ER (Yan and Lennarz 2002, 2005; Kelleher et al. 2003; Kelleher and Gilmore 2006). However, prokaryotic (Lizak et al. 2011) and kinetoplastid OSTs consist of a single-subunit homologous to the STT3 catalytic subunits of OST complexes (Kelleher and Gilmore 2006). In the case T. brucei, OST activity is catalyzed by two single-subunit enzymes that selectively transfer distinct oligosaccharide donors to specific glycosylation sites (Jones et al. 2005; Manthri et al. 2008; Izquierdo, Schulz, et al. 2009). Thus, TbSTT3A first transfers biantennary Man5GlcNAc2 from Man5GlcNAc2-PP-Dol to glycosylation sites in acidic to neutral regions of polypeptides and TbSTT3B transfers triantennary Man9GlcNAc2 from Man9GlcNAc2-PP-Dol to remaining glycosylation sites (Izquierdo, Schulz, et al. 2009). The selective recruitment of triantennary Man9GlcNAc2 or biantennary Man5GlcNAc2 to specific glycosylation sites in this way is highly significant because it predetermines the kind of processing, and therefore the repertoire of final glycan structures, that can be generated at that site. Specifically, triantennary Man9GlcNAc2 can only be processed as far as triantennary Man5GlcNAc2 by T. brucei, which lacks a Golgi α-mannosidase II gene, thus strictly limiting the glycoforms at such sites to the oligomannose series. In contrast, biantennary Man5GlcNAc2 is the only route to paucimannose and complex N-glycans in the parasite (Manthri et al. 2008), including the novel giant poly-N-acetyllactosamine structures found in bloodstream-form T. brucei (Atrih et al. 2005). This model of sequon pI controlling the oligomannose vs. paucimannose/complex N-glycosylation fate at particular N-glycosylation correctly predicted the N-glycan type attached to a previously uncharacterized T. brucei VSG (Mehlert et al. 2010).

The T. brucei ALG12 gene (TbALG12) encodes a Dol-P-Man:Man7GlcNAc2-PP-Dol -α1-6-mannosyltransferase. In the cultured form of the normally insect-dwelling (procyclic) life-cycle stage of the parasite, deletion of this gene reduced the maximum LLO size from Man9GlcNAc2-PP-Dol to Man7GlcNAc2-PP-Dol, as expected (Acosta-Serrano et al. 2004). The ALG12−/− cells proved to be resistant to the cytotoxic action of concanavalin A because they express procyclins with altered N-glycans, predominantly shorter paucimannose glycans (Man4GlcNAc2) with or without a terminal N-acetyllactosamine unit (Leal et al. 2004).

This paper describes the glycotyping of the VSG from bloodstream-form ALG12−/− mutants. The variant 221 (also known as MITat1.2) VSG coat glycoprotein is an excellent reporter because it contains two N-glycosylation sites: one at Asn263 occupied by small, biantennary paucimannose and complex structures, which originate from the transfer of Man5GlcNAc2 by TbSTT3A from the Man5GlcNAc2-PP-Dol precursor, and one at Asn428 occupied by conventional triantennary oligomannose structures, which originate from the transfer of Man9GlcNAc2 by TbSTT3B from Man9GlcNAc2-PP-Dol (Jones et al. 2005; Manthri et al. 2008; Izquierdo, Schulz, et al. 2009). Thus, analysis of VSG221 N-glycosylation allows us to simultaneously assess the effects of genetic or chemical perturbations on both mechanisms of protein N-glycosylation in this organism (Jones et al. 2005; Urbaniak et al. 2006; Manthri et al. 2008; Stokes et al. 2008; Izquierdo, Atrih, et al. 2009; Izquierdo, Schulz, et al. 2009).

Results

Analysis of intact soluble-form VSG221 synthesized by TbALG12 null mutant cells

Wild-type and TbALG12−/− mutant cell lines were grown in vitro, the latter in the presence and the absence of a cocktail of α-mannosidase inhibitors (MI), and samples of 2 × 108 cells from 100 mL of culture were processed to yield ∼100 μg of purified soluble-form VSG221 (sVSG221). sVSG arises from the cleavage of diacylglycerol from the GPI anchor by the action of an endogenous GPI-specific phospholipase C (Cardoso de Almeida and Turner 1983; Cross 1984; Ferguson et al. 1985). The sVSG221 synthesized by the ALG12−/− parasites showed different glycosylation patterns to that synthesized by wild-type cells, as judged by SDS–PAGE (Figure 1, inset). Aliquots of the purified sVSGs were also analyzed by positive-ion electrospray-mass spectrometry (ES-MS). The deconvolved mass spectrum of intact wild-type sVSG221 (Figure 1A; Table I) showed the range of expected glycoforms that arise from the known heterogeneity in the GPI anchor and N-glycan sites (Zamze et al. 1991; Mehlert et al. 1998; Jones et al. 2005; Manthri et al. 2008; Izquierdo, Schulz, et al. 2009). In the case of sVSG221 synthesized by the ALG12−/− mutant in the absence of MI, the analysis revealed a group of glycoforms centered around 49,293 Da similar to the underglycosylated sVSG221 glycoforms found in the TbALG3−/− mutant and in the UDP-GlcNAc pyrophosphorylase conditional null mutant under non-permissive conditions, i.e. with masses consistent with the C-terminal Asn428 N-glycosylation site being unoccupied (Urbaniak et al. 2006; Manthri et al. 2008; Stokes et al. 2008). The higher molecular weight group of glycoforms, centered around 50,673 Da, with both N-glycosylation sites occupied, showed a wider range of glycoforms than the wild-type profile, including glycoforms up to three hexose units smaller and some with significantly higher N-acetylhexosamine to hexose ratios than the wild-type glycoforms (Table I). This suggests that the latter group of glycoforms may be enriched in N-glycans with LacNac structures at one or both sites. The effect of MI on the ALG12−/− null mutant sVSG221 showed a general increase in size for both groups of glycoforms, and the significance of this is described in the end of the results section.

Fig. 1.

SDS–PAGE and mass spectrometric analyses of intact sVSG221 from wild-type and ALG12−/− null trypanosomes. The “inset” shows a Coomassie stained SDS–PAGE gel showing the migration of sVSG221 from wild-type cells (lane 1), the ALG12−/− null mutant (lane 2) and the ALG12−/− null mutant grown in the presence of MI (lane 3). The panels show the mass spectrometric analysis of sVSG221 from wild-type cells (A), the ALG12−/− null mutant (B) and the ALG12−/− null mutant grown in presence of MI (C).

Table I.

Isobaric glycoforms of sVSG221 detected by ES-MS

Proteina	I-cPb	GlcNb	EtNPb	HexNAc	Hex	Molecular mass (Da) WT/ALG12−/−/ALG12−/− MI (Theo.)	WT	ALG12−/−	ALG12−/− MI
1	1	1	1	2	10	NA/48809/NA(48816)	−	+	−
1	1	1	1	2	11	NA/48971/NA(48978)	−	+	−
1	1	1	1	2	12	NA/49133/49135(49140)	−	+	Traces
1	1	1	1	3	11	NA/49175/49176(49181)	−	+	Traces
1	1	1	1	2	13	NA/49293/49296(49302)	−	++	+
1	1	1	1	3	12	NA/49336/49337(49343)	−	+	+
1	1	1	1	2	14	NA/49455/49458(49464)	−	++	++
1	1	1	1	3	13	NA/49497/49500(49505)	−	+	+
1	1	1	1	2	15	NA/49616/49620(49626)	−	+	+
1	1	1	1	3	14	NA/49659/49661(49667)	−	+	+
1	1	1	1	2	16	NA/49778/49782(49788)	−	+	+
1	1	1	1	3	15	NA/NA/49823(49829)	−	−	+
1	1	1	1	4	15	NA/50026/NA(50032)	−	+	−
1	1	1	1	4	16	NA/50186/NA(50194)	−	++	−
1	1	1	1	5	15	NA/50229/NA(50235)	−	+	−
1	1	1	1	4	17	NA/50348/50351(50356)	−	+++	+
1	1	1	1	5	16	NA/50389/50392(50397)	−	++	Traces
1	1	1	1	4	18	50519/50510/50513(50518)	Traces	+++	+
1	1	1	1	5	17	NA/50551/50555(50559)	−	+++	+
1	1	1	1	6	16	NA/50597/NA(50600)	−	+	−
1	1	1	1	4	19	50681/50673/50676(50680)	++	+++	++
1	1	1	1	5	18	50717/50714/50716(50721)	Traces	++	+
1	1	1	1	6	17	50769/50759/50756(50762)	Traces	+	Traces
1	1	1	1	4	20	50841/50835/50838(50842c)	+++	+++	++
1	1	1	1	5	19	50881/50876/50879(50883)	+	++	+++
1	1	1	1	6	18	NA/50919/50918(50924)	−	+	+
1	1	1	1	4	21	50998/NA/51000(51004)	++	+	+++
1	1	1	1	5	20	51040/51032/51041(51045)	++	++	+++
1	1	1	1	6	19	NA/51081/51081(51086)	−	+	+
1	1	1	1	4	22	51157/NA/51161(51166)	+	+	++
1	1	1	1	5	21	51205/51201/51204(51207)	++	+	+++
1	1	1	1	6	20	51247/51244/51243(51248)	Traces	Traces	+
1	1	1	1	4	23	51320/NA/51324(51328)	+	+	++
1	1	1	1	5	22	51365/51364/51365(51369)	+	+	++
1	1	1	1	8	19	51487/51485/51486(51492)	Traces	+	+
1	1	1	1	7	21	NA/NA/51610(51613)	−	−	+
1	1	1	1	8	20	NA/51647/51651(51654)	−	+	+
1	1	1	1	7	22	NA/NA/51772(51775)	−	−	+
1	1	1	1	8	21	NA/51809/51812(51816)	−	Traces	+
1	1	1	1	7	23	NA/NA/51933(51937)	−	−	+
1	1	1	1	8	22	NA/NA/51973(51978)	−	−	+
1	1	1	1	7	24	NA/NA/52095(52099)	−	Traces	+

The molecular weights of different glycoforms of sVSG221 were calculated according to the indicated compositions. The −, traces, +, ++ and +++ scores indicate the relative abundances of those glycoforms observed in sVSG preparations from the different cell lines, i.e. wild-type cells (WT), ALG12−/− null mutant cells (ALG12−/−) and ALG12−/− null mutant cells growth with mannosidase inhibitors (ALG12−/− MI).

aProtein Mr is based on the amino acid sequence of the VSG221 precursor (accession no. P26332) minus residues 1–27 (signal peptide) and 460–476 (GPI attachment signal peptide) and allows for four disulfide bonds (Mr = 46,284).

bComponents specific to the GPI anchor and common to all glycoforms; I-cP myo-inositol-1,2 cyclic phosphate; EtNP, ethanolamine phosphate.

cThe most abundant glycoform of wild-type sVSG221 is expected to contain a GPI anchor of composition of Man3Gal5, a C-terminal N-linked glycan of Man9GlcNAc2 and an internal N-linked glycan of Man3GlcNAc2; i.e. HexNAc (GlcNAc) = 4 and Hex (Man + Gal) = 20.

Analysis of pronase glycopeptides from sVSG221 synthesized by TbALG12 null mutant cells

To further probe the nature of the changes in sVSG221 glycosylation induced by the deletion of the ALG12 gene, aliquots of wild-type and TbALG12 null mutant sVSG221 samples (50 μg) were digested with pronase and the resulting glycopeptides were enriched and analyzed by ES-MS and ES-MS/MS in a positive-ion mode (Manthri et al. 2008). As expected, the data showed no changes in the masses of the VSG GPI-peptide fragments in ALG12−/− cells or in these same cells treated with MI (Figure 2; Supplementary data, Table SI). However, the presence of glycopeptides like Hex4HexNAc4-NTT and Hex7HexNAc6-NTT are noteworthy (at 1957.76 and 2688.12 Da, respectively), since they indicate that, in the absence of its preferred Man9GlcNAc2-PP-Dol precursor, TbSTT3B is able to transfer biantennary Man5GlcNAc2 and/or Man7GlcNAc2 from their respective LLO precursors to the C-terminal Asn428 site, albeit with low efficiency as indicated by the underglycosylation apparent in Figure 1B and C. Evidence that both Man5GlcNAc2 and Man7GlcNAc2 can be transferred to Asn428 comes from the analysis of the pronase glycopeptides made from sVSG221 synthesized by in ALG12−/− cells in the presence of MI, where both Hex7HexNAc2-GNTNT at 2068.74 Da and Hex5HexNAc2-GNTNT at 1744.62 Da are apparent (Figure 2C).

Fig. 2.

Mass spectrometric analyses of pronase N-glycopeptides and GPI peptides. Aliquots of sVSG221 were digested with pronase and the glycopeptides were purified and analyzed by ES-MS in the positive-ion mode. The ES-MS spectra were processed using the Bayesian peptide reconstruction program in the ABI Analyst Software. The identities of the glycopeptides from wild-type (A), the ALG12−/− null mutant (B) and the ALG12−/− null mutant grown in presence of MI (C) are indicated (see also Supplementary data, Table SI) and were confirmed by MS/MS of the corresponding doubly charged ions (data not shown). GPI glycopeptides (which do not change between the different cell lines) are labeled in italics.

The presence of glycopeptides like Hex7HexNAc2-RNET (2058.88 Da) and Hex8HexNAc2-RNET (2220.92 Da) in the ALG12−/− sVSG221 sample suggests that TbSTT3A can utilize biantennary Man7GlcNAc2-PP-Dol as well as Man5GlcNAc2-PP-Dol (Figure 2B). We presume that the Hex8HexNAc2-containing glycopeptides are due to glucosylation of the a-branch of Man7GlcNAc2 by UGGT (Izquierdo, Atrih, et al. 2009), an assumption that is supported by their significantly greater intensity in the ALG12−/− sVSG221 from MI-treated cells, where the a-branch is protected from digestion by ER mannosidases.

Discussion

Taken together, the data presented here and summarized in Figure 3 indicate that TbSTT3A can use Man7GlcNAc2-PP-Dol with about equal efficiency to Man5GlcNAc2-PP-Dol (the main substrate used in wild-type cells) and that TbSTT3B can also use Man7GlcNAc2-PP-Dol with about equal efficiency to Man5GlcNAc2-PP-Dol. However, in the case of TbSTT3B, the underglycosylation of sVSG221 in TbALG3−/− (Manthri et al. 2008) and TbALG12−/− cells (this paper) suggests that neither are the preferred LLO donor for TbSTT3B, which appears to have a strong preference for Man9GlcNAc2-PP-Dol. This, in turn, suggests that TbSTT3B requires the c-branch of the LLO oligosaccharide for efficient LLO recognition and/or transfer, whereas TbSTT3A does not; indeed, its presence may impede LLO recognition and/or transfer by TbSTT3A. This model of LLO selection by TbSTT3s would be in agreement to the mechanisms suggested for OSTs in other protists, such as Trypanosoma cruzi, Entamoeba histolytica and Trichomonas vaginalis (Kelleher et al. 2007). In these examples, a terminal α1,2-linked mannose residue on the b- or c-branch of Man9GlcNAc2-PP-Dol is a positive determinant for substrate selection by the T. cruzi OST, whereas E. histolytica and T. vaginalis OSTs select donors with a non-glucosylated a-branch in the LLO but do not discriminate between Man5GlcNAc2-PP-Dol and Man9GlcNAc2-PP-Dol (Kelleher et al. 2007).

Fig. 3.

A model of N-glycan processing in ALG12−/− null T. brucei cells. The solid long arrows illustrate the transfer of Man5GlcNAc2 and Man7GlcNAc2 from their respective LLOs to each of the two N-glycosylation sites of VSG221 in ALG12−/− null mutant trypanosomes. The +/− symbol indicates inefficient oligosaccharide transfer of Man7GlcNAc2 and Man5GlcNAc2 by TbSTT3B that normally transfers Man9GlcNAc2. The short solid arrows indicate the processing of Man5GlcNAc2 and Man7GlcNAc2 once attached to each of the VSG221 N-glycosylation sites. The dashed short arrows indicate the processing of atypical Man4GlcNAc2 glycans to Man5GlcNAc3 glycans as described in the analysis of concanavalin A-resistant procyclic mutants (Hwa and Khoo 2000).

In summary, the underlying mechanism for site-specific N-glycosylation in T. brucei appears to be defined by the selectivity of TbSTT3B for LLO donors containing the c-branch and of TbSTT3A for LLO donors lacking the c-branch, coupled with the co-translational action of TbSTT3A and its selectivity for sequons in relatively acidic environments and the post-translational action of TbSTT3B with broad specificity for sequon environment (Izquierdo, Schulz, et al. 2009). Alternatively, it could be argued that TbSTT3B absolutely requires the complete Man9GlcNAc2 donor for activity and that the partial glycosylation observed in Asn428 is catalyzed entirely by TbSTT3A using either Man7GlcNAc2-PP-Dol or Man5GlcNAc2-PP-Dol, albeit inefficiently due to the absence of an optimal acidic-neutral polypeptide region. However, this latter model seems unlikely since it is known that, in the absence of TbSTT3B, TbSTT3A barely modifies Asn428 at all (Izquierdo, Schulz, et al. 2009) and, in contrast, there is a significant amount (>70%) of VSG with both N-glycosylation sites occupied in the ALG12−/− mutant cells (Figure 1B). Our results, therefore, support the notion that protists that cannot synthesize the canonical Glc3Man9GlcNAc2-PP-Dol LLO, such as T. cruzi and E. histolytica, have less stringent LLO donor specificity than organisms that do synthesize Glc3Man9GlcNAc2-PP-Dol, such as Saccharomyces cerevisiae and mammals (Kelleher et al. 2007).

The processing of the biantennary Man7GlcNAc2 structure in bloodstream-form trypanosomes described in this paper, together with those previously reported for procyclic trypanosomes (Hwa et al. 1999; Hwa and Khoo 2000; Leal et al. 2004) and for the processing of biantennary Man5GlcNAc2 in bloodstream-form and procyclic-form trypanosomes (Manthri et al. 2008), provides some insights into the specificities of the T. brucei UDP-GlcNAc:glycoprotein GlcNAc transferase type I (TbGnTI) and type II (TbGnTII) activities. These activities add βGlcNAc residues in 1-2-linkage to the 3- and 6-arm, respectively, of the Manα1-3(Manα1-6)Manβ1-4GlcNAcβ1-4GlcNAc (Man3GlcNAc2) core and can therefore be considered equivalent to the GnTI and GnTII activities found in other eukaryotes. However, the T. brucei genome does not contain any obvious candidate genes for these activities, although there are some 21 genes encoding putative UDP-GlcNAc/UDP-Gal-dependent glycosyltransferases of unknown function in the genome (Izquierdo, Nakanishi, et al. 2009). Given the apparent sequence disparity between general eukaryotic GnTI and GnTII enzymes and the parasite equivalents, one might also expect specificity differences, and this does appear to be the case: first, whereas GnTI from higher eukaryotes acts on Manα1-3(R-Manα1-6)Manβ1-4GlcNAcβ1-4GlcNAc, where R = H, Manα1-3 or Manα1-3(Manα1-6) (Kornfeld and Kornfeld 1985), TbGnTI does not appear to able to work on substrates containing the underlined α1-6-linked Man residue. This means that T. brucei is incapable of making conventional hybrid N-glycans. Second, whereas the prior action of higher eukaryote GnTI is required for the subsequent action of GnTII, this is clearly not the case for TbGnTII. Indeed, the addition of βGlcNAc to the 6-arm of the Manα1-3(Manα1-6)Manβ1-4GlcNAcβ1-4GlcNAc core by TbGnTII appears to be completely unaffected by the status of the 3-arm of the core, which can be unsubstituted or even substituted with mannobiose (i.e. Manα1-2Manα1-2Manα1-3(Manα1-6)Manβ1-4GlcNAcβ1-4GlcNAc).

The aforementioned peculiarities in protein N-glycosylation and N-glycan processing in T. brucei have prompted us to consider protein N-glycosylation as a potential therapeutic target against African trypanosomes. This is further supported by the fact that both the TbSTT3A and TbSTT3B catalyzed branches of N-glycosylation to paucimannose/complex and oligomannose glycans, respectively, are essential for parasite growth in animals (Izquierdo, Schulz, et al. 2009).

Materials and methods

Cultivation of trypanosomes

Bloodstream-form T. brucei genetically modified to express T7 polymerase and the tetracycline repressor protein were cultivated in HMI-9 medium containing 2.5 µg/mL of G418 at 37°C in a 5% CO2 incubator as described in Wirtz et al. (1999). The ALG12 null mutant strain was a generous gift from G. A. M. Cross lab (Leal et al. 2004).

Small-scale sVSG isolation

sVSG was isolated from 100 mL cultures containing ∼2 × 108 bloodstream-form T. brucei. The cultures were chilled in ice-water and centrifuged at 2500 × g for 10 min. The pellet was washed twice in trypanosome dilution buffer (Cross 1975) and transferred to a 1.5-mL Eppendorf tube. The pellet was resuspended in 300 µL of lysis buffer (10 mM NaH2PO4 buffer, pH 8.0, containing 0.1 mM 1-chloro-3-tosylamido-7-amino-2-heptanone, 1 µg/mL leupeptin and 1 µg/mL aprotinin) prewarmed to 37°C and incubated for 5 min at the same temperature. The sample was centrifuged at 14,000 × g for 5 min, and the supernatant was applied to a 200-µL DE52 anion exchange column pre-equilibrated in lysis buffer. Fresh lysis buffer (800 µL without protease inhibitors) was applied in four stages, and the pooled column eluate was concentrated and diafiltered with water on an YM-10 spin concentrator (Microcon, millipore, Watford, UK). The final sample of 50–100 µg of sVSG221 was recovered in a volume of 100 µL water.

ES-MS analysis of intact VSG

Samples of the sVSG preparations were diluted to ∼0.07µg/µL in 50% acetonitrile, 1% formic acid, loaded into nanotips (Micromass-type F) and analyzed by positive-ion ES-MS on a Q-Star XL instrument (Applied Biosystems, Paisley, UK). Data were collected and processed using the Bayesian protein reconstruction algorithm of Analyst software.

ES-MS and ES-MS/MS analysis of pronase glycopeptides

Aliquots of sVSG (∼50 µg in 50 µL water) were mixed with 5 µL of 1 M ammonium bicarbonate and 10 µL of 1 mg/mL pronase in 5 mM calcium acetate and incubated at 37°C for 36 h. The pronase glycopeptides were purified on Envicarb graphitized carbon microcolumns, which were prepared as follows: the contents of an Envicarb cartridge (Supelco, Poole, UK) were suspended in methanol and a bed of ∼20 µL of graphitized carbon was packed into a 100-µL C4 OMIX (Varian, Oxford, UK) pipette tip. The microcolumns were prepared by attaching them to a Gilson pipette, set at 100 µL, and pipetting up and down 10 times with 80% methanol, 1% formic acid; 60% methanol, 1% formic acid; and 1% methanol, 1% formic acid. The sample (10 µL of pronase digest) was mixed with 90 µL of 1% methanol, 1% formic acid and applied to the microcolumn by pipetting up and down 20 times. The microcolumns were washed by pipetting up and down 20 times with 1% methanol, 1% formic acid. The pipette was reset to 50 µL and the glycopeptides eluted by pipetting up and down 20 times with 50 µL of 60% methanol, 1% formic acid. Aliquots of these samples were loaded into nanotips (Micromass-type F) and analyzed by ES-MS and ES-MS/MS in the positive-ion mode on an ABI Q-Star-XL instrument with tip and declustering potentials of 900 and 60 V, respectively. The product ion spectra of selected ions were collected using collision energies of 30–60 V. The ES-MS spectra were processed using the Bayesian peptide reconstruction program in the ABI Analyst software.

Supplementary data

Supplementary data for this article is available online at http://glycob.oxfordjournals.org/.

Funding

This work was supported by a Wellcome Trust programme grant (085622) to MAJF and a Wellcome Trust Strategic Award (083481). LI was supported by a Marie Curie and an EMBO Fellowship.

Conflict of interest

None declared.

Abbreviations

ALG, asparagine-linked glycosylation; ER, endoplasmic reticulum; ES-MS, electrospray-mass spectrometry; GPI, glycosylphosphatidylinositol; LLO, lipid-linked oligosaccharide; MI, α-mannosidase inhibitors; OST, oligosaccharyltransferase; sVSG, soluble-form VSG; TbGnTI, T. brucei UDP-GlcNAc:glycoprotein GlcNAc transferase type I; VSG, variant surface glycoprotein.