Inhibition of Nucleotide Sugar Transport in Trypanosoma brucei Alters Surface Glycosylation*

Background: Nucleotide sugar transporters (NSTs) provide availability of nucleotide sugars for glycosylation in the lumen of the Golgi apparatus. Results: The substrate specificities of four Trypanosoma brucei NSTs were characterized, and inhibition of TbNST4 resulted in glycosylation defects. Conclusion: TbNST4 plays an essential role in biosynthesis of glycoproteins in T. brucei. Significance: To understand the roles of TbNST(s) in the growth and pathogenesis of T. brucei. Nucleotide sugar transporters (NSTs) are indispensible for the biosynthesis of glycoproteins by providing the nucleotide sugars needed for glycosylation in the lumen of the Golgi apparatus. Mutations in NST genes cause human and cattle diseases and impaired cell walls of yeast and fungi. Information regarding their function in the protozoan parasite, Trypanosoma brucei, a causative agent of African trypanosomiasis, is unknown. Here, we characterized the substrate specificities of four NSTs, TbNST1–4, which are expressed in both the insect procyclic form (PCF) and mammalian bloodstream form (BSF) stages. TbNST1/2 transports UDP-Gal/UDP-GlcNAc, TbNST3 transports GDP-Man, and TbNST4 transports UDP-GlcNAc, UDP-GalNAc, and GDP-Man. TbNST4 is the first NST shown to transport both pyrimidine and purine nucleotide sugars and is demonstrated here to be localized at the Golgi apparatus. RNAi-mediated silencing of TbNST4 in the procyclic form caused underglycosylated surface glycoprotein EP-procyclin. Similarly, defective glycosylation of the variant surface glycoprotein (VSG221) as well as the lysosomal membrane protein p67 was observed in Δtbnst4 BSF T. brucei. Relative infectivity analysis showed that defects in glycosylation of the surface coat resulting from tbnst4 deletion were insufficient to impact the ability of this parasite to infect mice. Notably, the fact that inactivation of a single NST gene results in measurable defects in surface glycoproteins in different life cycle stages of the parasite highlights the essential role of NST(s) in glycosylation of T. brucei. Thus, results presented in this study provide a framework for conducting functional analyses of other NSTs identified in T. brucei.


Nucleotide sugar transporters (NSTs) are indispensible for the biosynthesis of glycoproteins by providing the nucleotide sugars needed for glycosylation in the lumen of the Golgi apparatus. Mutations in NST genes cause human and cattle diseases and impaired cell walls of yeast and fungi. Information regarding their function in the protozoan parasite, Trypanosoma brucei, a causative agent of African trypanosomiasis, is unknown.
Here, we characterized the substrate specificities of four NSTs, TbNST1-4, which are expressed in both the insect procyclic form (PCF) and mammalian bloodstream form (BSF) stages. TbNST1/2 transports UDP-Gal/UDP-GlcNAc, TbNST3 transports GDP-Man, and TbNST4 transports UDP-GlcNAc, UDP-GalNAc, and GDP-Man. TbNST4 is the first NST shown to transport both pyrimidine and purine nucleotide sugars and is demonstrated here to be localized at the Golgi apparatus. RNAimediated silencing of TbNST4 in the procyclic form caused underglycosylated surface glycoprotein EP-procyclin. Similarly, defective glycosylation of the variant surface glycoprotein (VSG221) as well as the lysosomal membrane protein p67 was observed in ⌬tbnst4 BSF T. brucei. Relative infectivity analysis showed that defects in glycosylation of the surface coat resulting from tbnst4 deletion were insufficient to impact the ability of this parasite to infect mice. Notably, the fact that inactivation of a single NST gene results in measurable defects in surface glycoproteins in different life cycle stages of the parasite highlights the essential role of NST(s) in glycosylation of T. brucei. Thus, results presented in this study provide a framework for conducting functional analyses of other NSTs identified in T. brucei.
Nucleotide sugar transporters (NSTs) 3 play a key role in the biosynthesis of glycoproteins, as they translocate nucleotide sugars, which serve as sugar donors for glycosyltransferases in glycosylation reactions, from the cytosol into the lumen of the Golgi apparatus (1,2). Given their essential function in glycosylation, NSTs have been shown to play critical roles in development and organogenesis of mammals and higher multicellular eukaryotes as well as in the formation of cell wall or surface coat of lower unicellular eukaryotes (3,4). Biochemical and structural studies have shown that mutations or deletions of specific NSTs resulted in defects in the particular glycoconjugates containing the sugars whose corresponding nucleotide sugar transport was compromised (3)(4)(5). More recent studies have shown that siRNA-mediated silencing of NSTs in cultured mammalian cells resulted not only in defective glycosylation of glycoproteins, but also a more global defect in the synthesis and secretion of nonglycoproteins (6). Furthermore, instances in which cell surface glycoproteins represent the main interface between a pathogen and its host, carbohydrate composition is likely to play an important role in pathogenesis.
The present study addresses the function of NSTs in African trypanosomes (Trypanosoma brucei ssp.), which are parasitic protozoa. These parasites cause human African trypanosomiasis, e.g. sleeping sickness and veterinary disease in cattle (Nagana). Infection is fatal without treatment and consequently human African trypanosomiasis represents a major health problem in sub-Saharan Africa wherever the insect vector (tsetse fly, genus Glossina) is found. T. brucei has two life cycle stages that are amenable to biochemical and biological studies: a procyclic form (PCF) found in the midgut of the tsetse fly and a pathogenic bloodstream form (BSF) in the mammalian host. Each has a glycoprotein coat composed of a stage-specific major surface glycoprotein: procyclins in the PCF stage (7) and variant surface glycoproteins (VSGs) in the BSF stage (8). Both VSGs and procyclins play pivotal roles in pathogenesis, VSG as the lynchpin of antigenic variation in the mammalian bloodstream (9) and procyclin as a critical component facilitating colonization in the tsetse midgut (10). In addition, there are many less abundant surface glycoproteins including invariant surface antigens, transferrin receptor, and other nutrient transporters that are critical to the success of these important human pathogens (11).
Due to their relative abundance (5-10% of total cellular proteins), procyclins and VSGs have been the primary focus of studies on the glycobiology of trypanosomes. Both are glycosylphosphatidylinositol (GPI) anchored and N-glycosylated, and there are distinct stage-specific variations for each class of glycomodification (12). In PCF cells only oligomannose N-glycans (Man 9 GlcNAc 2 ) are transferred to nascent glycoproteins, and these are typically trimmed to small triantennary (Man 5 GlcNAc 2 ) structures (13,14); subsequent extension by addition of other hexose units in the Golgi apparatus has not been found. In addition to this kind of N-glycosylation, BSF cells are also capable of transferring abbreviated biantennary paucimannose structures (Man 5 GlcNAc 2 ), and these can be trimmed and extended in the Golgi apparatus by addition of N-acetyllactosamine (LacNAc, Gal␤1-4GlcNAc) units (15,16). In extreme cases, such N-glycans can be hypermodified with poly-N-acetyllactosamine structures with 50 or more LacNAc units (17,18). In some instances shorter LacNAc chains can be capped with ␣1-3Gal residues (19). These differences in N-glycosylation between PCF and BSF cells are due to differential stage-specific expression of two distinct ER oligosaccharyltransferses, TbSTT3A and TbSTT3B, with unique donor/acceptor substrate specificities (20).
Likewise GPI anchors are modified in a stage-specific manner. In PCF the GPI core is decorated with a branched LacNAc 9 (average) structure, which serves as an acceptor for terminal sialylation by cell surface trans-sialidases (12,21). In the context of this report, it is important to note that addition of sialic acid occurs by transfer from existing host serum sialoglycoproteins and is not dependent on biosynthetic transfer from the biosynthetic donor CMP-sialic acid, which trypanosomes do not make. In contrast, in BSF the GPI core is less extensively modified by addition of branched galactose structures as VSG passes through the Golgi apparatus (15,22).
Collectively, these structural studies indicate that both PCF and BSF cells are capable of glycan processing as secretory and membrane proteins transit the Golgi apparatus. This requires an extensive repertoire of glycosyltransferases (23) as well as specific NSTs to import the requisite nucleotide sugar donors (such as UDP-Gal and UDP-GlcNAc) into the lumen of the Golgi apparatus.
In this study, we characterized the substrate specificities of four NSTs, TbNST1-4, of putative eight identified in the T. brucei genomic database. We found that TbNST1/2 transports UDP-Gal/UDP-GlcNAc, TbNST3 transports GDP-Man, and TbNST4 transports UDP-GlcNAc, UDP-GalNAc, and GDP-Man. TbNST4 is the first NST shown genetically and biochemically to transport both pyrimidine and purine nucleotide sugars and is demonstrated here to be localized at the Golgi apparatus. TbNST1-4 are expressed in different life cycle stages (PCF and BSF). Because of its unique substrate specificity, TbNST4 was chosen for further functional analyses. RNAimediated silencing of TbNST4 in PCF caused underglycosylated surface glycoprotein EP-procyclin. Similarly, defective glycosylation of VSG221 as well as the lysosomal membrane protein, p67, was observed in ⌬tbnst4 BSF T. brucei. Relative infectivity analysis shows that defects in glycosylation of the surface coat resulting from tbnst4 deletion were insufficient to impact the ability of this parasite to infect mice, likely due to functional redundancy of NSTs. Overall, we demonstrate that inactivation of a single NST gene in T. brucei results in defects in glycosylation of surface proteins in different life cycle stages of the parasite, highlighting the essential role of NST(s) in glycosylation in T. brucei. Thus, results presented in this study provide a framework for conducting functional analyses of other identified TbNSTs, individually and combinatorially, to better understand their biological impact on this parasite.

Total RNA Isolation and Reverse Transcription PCR
Total RNA extraction was achieved with the RNeasy kit with on column DNase digestion (RNase-free DNase, Qiagen) or with TRIzol (Invitrogen) followed by DNase I treatment according to the manufacturer's instructions. cDNA was obtained using the SuperScript first-strand synthesis system (Invitrogen) and RT-PCR amplification was carried out with BIO-X-ACT TM Short MiX containing DNA polymerase (Bioline). A 446-bp PCR product from nt 1 to 446 of the open reading frame was obtained for TbNST1 using TbNST1-5(F)/ TbNST1-6(R) primers. A ϳ900-bp PCR product from nt 1 of the spliced leader to nt 600 of the open reading frame was obtained for TbNST2 using TbSLRNA-1(F)/TbNST2-2(R) primers. A ϳ1000-bp PCR product from nt 1 of the spliced leader to nt 781 of the open reading frame was obtained for TbNST3 using TbSLRNA-1(F)/TbNST3-6(R). A ϳ1220-bp PCR product from nt 1 of the spliced leader to nt 1002 was obtained for TbNST4 using TbSLRNA-1(F)/kTbNST4-B(R). Note that all trypanosome mRNAs have a 5Ј spliced leader (SL) sequence as a result of trans-splicing. All primer sequences are detailed in supplemental Table S1.

Generation of DNA Constructs and Transgenic Trypanosome Cell Lines
TbNST4-RNAi PCF Cell Line-A construct producing inducible TbNST4 dsRNA in the form of a stem-loop structure was created as previously described in Ref. 26 using pJM325 and pLew100 vectors (gifts from Dr. Paul Englund, Johns Hopkins University). The stem sequences were from a 608-bp fragment containing the TbNST4 coding sequence with opposite orientations. The above plasmids were linearized with EcoRV and transfected into strain 29-13 (27). Induction of TbNST4 dsRNA was achieved with 1 g/ml of tetracycline.
tbnst4-null BSF Cell Line-A tbnst4 homozygous knock-out (KO) was created using vectors pLew13-NEO and pLew90-HYG. To generate the first allele KO construct (pSKO-TbNST4), the 5Ј and 3Ј UTRs of tbnst4 were PCR amplified from BSF genomic DNA. PCR products of a 304-bp fragment containing the 5Ј UTR and a 327-bp fragment containing the 3Ј UTR were inserted into the NotI/MluI and StuI/XbaI sites of pLew13, respectively. To generate the second allele KO construct (pDKO-TbNST4), the SwaI/XhoI fragment containing a neomycin marker and the gene of T7 RNA polymerase of pSKO-TbNST4 was replaced with the 2491-bp StuI/XhoI fragment containing the hygromycin marker and tetracycline repressor gene from pLew90. The stable tbnst4-null cell line was obtained by transfection with NotI-linearized pSKO-TbNST4 and selection with G418. Subsequently, the cell line carrying pSKO-TbNST4 was transfected with NotI-linearized pDKO-TbNST4 creating tbnst4-null mutants under the selection of 2.5 g/ml of G418 and 5 g/ml of hygromycin B. Introduction of T7 RNA polymerase and tetracycline genes into the genome of the double KO strain will facilitate RNAi silencing of other transporters in the future studies.
tbnst4 in Situ Tagging in BSF Cells-A C-terminal tagging PCR fragment was obtained using vector pMOTag3H as previously described in Ref. 28. The transgenic cell line expressing tagged tbnst4 was generated by transfection with a 2527-bp PCR product containing the 150-bp TbNST4 coding sequence upstream to stop codon and the 150-bp TbNST4 3Ј UTR sequence downstream to the stop codon under the selection of hygromycin B. Single marker BSF cells (29) with Ty-tagged Golgi putative glycosyltransferase (SM-BS TbGT15:15Ty) (30) was used for the colocalization experiment.
After washing with 0.3% BSA in PBS, cells were resuspended in 350 l of PBS and analyzed in a flow cytometer (BD Biosciences FACScan) using detector FL1 (530 Ϯ 30 nm). Histograms were generated with FlowJo 9.5.2 software.

Immunofluorescence Analysis
The procedure of immunostaining and microscope imaging for both PCF and BSF was adapted from that previously described in Ref. 31 with changes in fixation for BSF cells. Specifically, 50 l of BSF cell suspension was pipetted onto coverslips in a 6-well plate and placed on ice for 0.5 h for the cell attachment. 4% Paraformaldehyde was used for fixation on ice for 0.5 h. Both mouse anti-EP (Cedarlane) and anti-HA (Covance) antibodies were diluted at 1:500 for the staining. Cell images were captured with a Nikon wide-field epifluorescence microscope. All images were processed using ImageJ software (rsb.info.nih.gov/ij/).

Soluble Form VSG (sVSG) Isolation
sVSG was isolated from 1 ϫ 10 8 BSF cells (in 100 ml of culture) as previously described in Refs. 20 and 32, except that the sVSG-containing fraction was not applied to an anion exchange column (DE52) but was directly applied to Western blotting.

Antibody and Lectin Blotting
T. brucei cells were lysed with RIPA buffer (1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, 50 mM Tris-HCl, pH 7.5, and 150 mM NaCl) with Complete protease inhibitor mixture (Roche Applied Science). The resulting cell lysates were subjected to SDS-PAGE, transferred to a nitrocellulose membrane, blocked, and probed with mouse anti-HA (Covance), mouse anti-␣-tubulin (Sigma), rabbit anti-EP, or rabbit anti-VSG221 antibodies. After washing, the bound primary antibodies were incubated with goat anti-mouse or goat anti-rabbit IgG conjugated to horseradish peroxidase at RT for 0.5 h (Sigma). Appropriate dilutions were used for all antibodies. Lectin blotting was done as described in Ref. 20. Membranes were blocked and incubated with biotinylated ECL (5 g/ml), RCA I (5 g/ml), ConA (1 g/ml), and TL (0.33 g/ml) lectins (Vector Laboratories), with or without corresponding sugar inhibitors. Bound biotinylated lectins were detected with streptavidinperoxidase (Kirkegaad & Perry Laboratory). All membranes were developed after treatment with ECL chemiluminescent reagent (GE Healthcare).

Metabolic Radiolabeling and Immunoprecipitation
Pulse-chase metabolic radiolabeling and subsequent immunoprecipitation of radiolabeled proteins from cell lysates were performed as previously described (33,34). Briefly, BSF WT (VSG221) and tbnst4-null cells were pulse labeled with 200 Ci of [ 35 S]Met/Cys for 15 min and then chased at 1:10 dilution with complete HMI-9 for 2 h in the presence of FMK024 (20 M) (MP Biomedicals, Aurora, OH), a selective thiol protease inhibitor. Immunoprecipitated proteins were analyzed by 10% SDS-PAGE and fluorography.

Mouse Infection
Parental BSF T. brucei (VSG221) and tbnst4-null cells were grown to a density of 1 ϫ 10 6 without selection drugs (hygro-mycin B and G418) for 24 h. C57BL6 mice (5 per group) were infected intraperitoneally with 10 2 or 10 3 parasites in 0.2 ml of HMI-9 medium. Relative infectivity was assessed by determining parasitemia in tail vein blood using a Neubauer hemocytometer.

Generation of DNA Constructs Expressing TbNSTs in S. cerevisiae and K. lactis
To generate the constructs expressing TbNSTs in S. cerevisiae, the open reading frame sequences of the TbNSTs were PCR amplified from PCF genomic DNA. The 1113-, 1359-, 1056-, and 1002-bp of PCR products were inserted into vector p426GPD at the BamHI/EcoRI sites creating p426GPD-TbNST1 and p426GPD-TbNST3, at the SpeI/XhoI sites creating p426GPD-TbNST2, and at the BamHI/XhoI sites creating p426GPD-TbNST4. The reverse primers introduce a VSV tag sequence. To generate the constructs expressing TbNSTs in K. lactis, PCR products containing the respective open reading frames of TbNST1-4 were inserted into vector pE4 at the XhoI/EcoRI sites creating pE4-TbNST1 and pE4-TbNST4, at the BglII/XhoI sites creating pE4-TbNST2 and at the BglII/ EcoRI sites creating pE4-TbNST3.

Subcellular Fractionation and Nucleotide Sugar Transport Assay
S. cerevisiae cells transformed with empty vector p426GPD or p426GPD-TbNST1-4 were individually grown to an A 600 of ϳ2. Spheroplasts were prepared as previously described (39) using zymolase 100 T (Seikagaku America, Rockville, MD) and Golgi-enriched vesicles were obtained by fractionation of spheroplast suspension with successive centrifugation (40). Nucleotide sugar transport assays were carried out as previously described (39). Radiolabeled substrates used were UDP-
Griffonia simplicifolia II (GSII)-FITC Binding-K. lactis WT (KL8) and mutant (KL3, defective in UDP-GlcNAc transport) (35) cells carrying empty vector pE4 and pE4-TbNST1-4, each individually, were washed with 0.9% NaCl, 0.5 mM CaCl 2 and labeled with GSII-FITC (10 g/ml) (EY Laboratories) at RT for 1 h and then washed 5 times with the same buffer. Absorbance of cell aliquots was determined with a photocytometer at 600 nm and the relative fluorescence was measured with a tecan microplate reader at 535 nm. The ratio of GSII-FITC/absorbance was calculated as relative fluorescence intensity.
Congo Red Assay-S. cerevisiae WT and vrg4 -2 mutant (defective in GDP-Man transport) cells (37), transfected with empty vector p426GPD and p426GPD-TbNST1-4, each individually, were grown to mid-logarithmic phase in the synthetic complete medium minus uracil, diluted to absorbance of ϳ0.2 at 600 nm, and equal volumes were applied onto uracil minus plates containing 0 -50 g of Congo red. Plates were scored and photographed after 3 days at 30°C.

Identification of T. brucei Nucleotide Sugar Transporters-
To identify TbNSTs, we used BLASTp to search the T. brucei predicted protein database (tritrypdb.org) for homologs with well characterized NSTs from different organisms such as L. major, C. elegans, and Homo sapiens. Eight proteins were revealed to be putative NSTs, referred to as TbNST1-8. The evolutionary distances of the amino acid divergence between or among the putative TbNSTs and known NSTs from different organisms are depicted by a phylogenetic tree (Fig. 1). TbNST1 and TbNST2 share the most conserved amino acid identity with known UDP-sugar transporters such as for UDP-Gal and UDP-GlcNAc (Fig. 2), whereas they show little sequence relationship with GDP-sugar transporters. TbNST3 and TbNST4 share the most conserved amino acid identity with known GDP-Man transporters and also show the similarity to the GDP-Fuc transporters (Fig. 3). Of the eight candidate NSTs, TbNST1-4 were selected for further experimental evaluation, based on their high degree of amino acid sequence homology with known transporters.
It is known that protein expression in T. brucei during the different life cycle stages is highly dynamic and stage specific. To examine the expression of TbNST1-4, total RNA from both PCF and BSF cells was isolated, reverse transcribed, and the mRNA transcripts were examined. Fig. 4 shows that TbNST1-4 are all expressed in PCF (left panel) and BSF (right panel) T. brucei.
Substrate Specificity of TbNST1-4-Previous studies from our and other laboratories showed that the substrate specificity of NSTs cannot be simply predicted from primary sequence comparisons (1,2). NSTs sharing 50 -60% of amino acid sequence identity may have different substrate specificities, whereas NSTs sharing as little as 20% of amino acid sequence identity may have the same substrate specificity (3,5). Therefore, the substrate specificity of putative TbNSTs must be determined experimentally. To this end, we have used two approaches: (a) in vitro biochemical transport assays using yeast S. cerevisiae Golgi-enriched vesicles expressing the tagged TbNSTs and (b) in vivo genetic complementation of previously well characterized yeast or mammalian mutants, specifically defective in transport of the particular nucleotide sugars.
S. cerevisiae has been a powerful system to determine the substrate specificity of putative NSTs, because it contains only two endogenous nucleotide sugar transport activities, those for UDP-Glc and GDP-Man. To evaluate the transport activities of putative TbNSTs for nucleotide sugars except the above mentioned two endogenous substrates, we transformed WT S. cerevisiae cells (PRY225) with empty vector p426GDP or p426GDP containing TbNST(1-4)-VSV separately. The Golgienriched vesicles expressing TbNSTs were isolated and assayed for transport of radiolabeled UDP-Gal, UDP-GalNAc, UDP-GlcNAc, and GDP-Fuc as previously described (39). As shown in Fig. 5, A and B, the transport activity of vesicles expressing TbNST1 and TbNST2 for UDP-Gal and UDP-GlcNAc is at least 2-fold higher than that of the empty vector; no transport activity for GDP-Fuc or UDP-GalNAc was detected. Transport activity of vesicles expressing TbNST4 for UDP-GlcNAc and UDP-GalNAc is ϳ3.5-4-fold higher than that of the empty vector; minor or no transport activity was detected for UDP-Gal or GDP-Fuc, respectively (Fig. 5D). Notably, one of the characteristics of NSTs is their rate-limiting transport of substrates with an apparent K m in the 1-30 M range. As shown in Fig. 5E, transport of UDP-GlcNAc by TbNST4 was saturable with a K m of 14.0 M. No transport activity for UDP-Gal, UDP-GlcNAc, UDP-GalNAc, or GDP-Fuc was detected for TbNST3 (Fig. 5C).
To further evaluate the UDP-GlcNAc substrate specificity for TbNST1, -2, and -4, we used a K. lactis mutant strain, KL3, which was previously shown to be specifically defective in UDP-GlcNAc transport (35). This mutant lacks terminal GlcNAc in its glycoproteins at the cell surface and therefore, has a surface binding defect to the GSII lectin, which recognizes terminal ␣or ␤-linked GlcNAc. Thus, TbNST1-4 were individually expressed in the KL3 mutant strain and phenotypic correction was measured by restoration of cell surface binding of fluorescently labeled GSII lectin. As shown in Fig. 6A, expression of TbNST4 restored GSII binding to the cell surface of KL3 mutant cells to amounts comparable with that of WT KL8 cells; a modest restoration was also detected for TbNST1-3 as compared with that of KL3 transformants with the empty vector.
To further determine whether TbNST1 and TbNST2 transport UDP-Gal, we used a mutant Madin-Darby canine kidney cell line, RCAr, which was previously shown to be specifically defective in UDP-Gal transport (41). Lack of galactose in glycoproteins at the cell surface renders this mutant strain defective in surface binding of the MAA lectin, which recognizes sialic acid (␣2,3)-galactose. We expressed TbNST1-4 individually in RCAr mutant cells and measured phenotypic correction by restoration of cell surface binding of fluorescently labeled MAA lectin. Expression of TbNST1 or TbNST2 significantly restored MAA binding to the cell surface of RCAr mutant cells, whereas TbNST3 and TbNST4 did not (Fig. 6B).
Based on sequence comparisons, TbNST3 and TbNST4 most likely transport GDP-Man. It is known that WT S. cerevisiae has a high endogenous GDP-Man transport activity. To test whether these two transporters transport GDP-Man, we used a mutant S. cerevisiae strain, vrg4 -2, which was previously shown to have a significant defect in transport of GDP-Man (37,42), therefore, resulting in much lower endogenous GDP-Man transport activity than WT. Thus, vrg4 -2 mutant cells were transformed with empty vector p426GDP or p426GDP-TbNST1-4 individually. Following isolation of Golgi-enriched vesicles expressing the TbNSTs, transport assays were performed as described for WT S. cerevisiae. Fig. 6C shows that expression of TbNST3 or TbNST4 increased transport activity of vrg4 -2 mutant cells by ϳ5or ϳ3-fold, respectively, as compared with that of vrg4 -2 transformed with the empty vector. No restoration of GDP-Man transport activity for vrg4 -2 mutant cells was detected for TbNST1 or TbNST2 (data not shown). To further validate the above in vitro results, we performed an in vivo genetic complementation experiment using the same vrg4 -2 mutant. The cell wall of vrg4 -2 mutants is known to be profoundly compromised due to a lack of surface mannoproteins (43), which is compensated by an increase of cell surface chitin. This results in hypersensitivity of vrg4 -2 mutant cells to Congo red, an anionic dye that binds chitin and kills cells by disrupting cell wall formation. Vrg4 -2 mutant cells expressing TbNST1-4, respectively, and transformants with the empty vector were grown on YPDA agar plates in the presence of different concentrations of Congo red. As shown in Fig.  6D, both TbNST3 and TbNST4 restored Congo red resistance of vrg4 -2 mutant cells, whereas TbNST1 and -2 did not (data not shown).
Thus, based on these collective findings, we conclude that TbNST1 and TbNST2 transport UDP-Gal and UDP-GlcNAc, TbNST3 transports GDP-Man, and TbNST4 transports UDP-GlcNAc, UDP-GalNAc, and GDP-Man as summarized in Table 1. To our knowledge, this is the first NST characterized biochemically and genetically to transport both pyrimidine and purine nucleotide sugars.
TbNST4 Is Localized at the Golgi Apparatus in T. brucei-In other eukaryotes, most nucleotide sugars are synthesized in the cytosol, and translocated by NSTs into the lumen of the Golgi apparatus. In T. brucei, synthesis of nucleotide sugars takes place in glycosomes (44 -46). This raises the question of whether these NSTs identified so far reside in membranes of glycosomes, the Golgi apparatus, or both. To determine the subcellular localization of TbNSTs, we used one-step PCR in situ tagging methodology to tag the endogenous tbnst4 with 3ϫ HA coding sequence at its C terminus (28). This method minimizes the likelihood of mislocalization of tagged proteins from overexpression. A 1842-bp PCR product containing a tagging cassette, which is specific to target the C-terminal locus of the tbnst4 gene, was transfected into the single marker BSF strain (29) with Ty-tagged Golgi putative glycosyltransferase (SM-BS TbGT15:Ty) (30). Following the establishment of a stable transgenic cell line, the in situ tagging locus was confirmed by genomic PCR. T. brucei is a diploid organism and this strategy resulted in the tagging of only one target allele. By PCR analysis, a 987-bp fragment was detected for both the untagged and tagged loci (Fig. 7A, lanes 1 and 2) and as expected, a 2527-bp fragment was detected only in cells expressing the tagged gene (Fig. 7A, lane 2, but not lane 1). Expression of the tagged TbNST4 protein was confirmed by Western blotting with anti-HA antibody (Fig. 7B, lane 2).
The localization of the tagged TbNST4 in BSF cells was visualized by immunofluorescence. As shown in Fig. 7C, HA-tagged TbNST4 (green) was confined to a single dot in the postnuclear region in close association with the flagellar attachment zone, as is typical for the Golgi in trypanosomes (47), rather than multiple glycosomal microbodies. It colocalized with a Ty-tagged Golgi marker, glycosyltransferase (red, upper panel), but not with p67, a lysosomal membrane protein (red, lower panel). The above results therefore strongly suggest that TbNST4 is localized at the Golgi apparatus.
RNAi-mediated Silencing of TbNST4 in PCF T. brucei and Its Effect on Cell Growth-To investigate the role of TbNST4 in PCF cells, we generated a conditional RNAi cell line carrying the construct to produce the stem-loop dsRNA specific for TbNST4 mRNA. Gene expression of TbNST4 can be knocked down by induction with tetracycline (26). To examine the efficiency and specificity of RNAi silencing, total RNA was isolated from uninduced and induced cells (on day 4), and RT-PCR was performed for detection of RNA transcripts of TbNST4 and control genes. As shown in Fig. 8A, the mRNA level of TbNST4 in induced cells was reduced by above 90% (lane 3) as compared with that of WT and uninduced cells (lanes 1 and 2). This reduction was TbNST4-specific, because mRNA levels of TbNST2 (Fig. 8A, right panel, lane 3) and TbNST3 (Fig. 8A, left  panel, lane 3) were not affected.
Silencing of TbNST4 causes a modest growth defect (Fig.  8B); the doubling time of induced cells was delayed by 3 h as compared with that of uninduced cells (21 versus 18 h). After induction of TbNST4 dsRNA for 4 days, no apparent morphological changes were visualized for TbNST4 silenced cells via light microscope (data not shown).
TbNST4 Is Essential for the Maturation of the Major Surface Glycoprotein, EP-Procyclin-To investigate the effect of TbNST4 silencing on the glycosylation in PCF cells, we examined the maturation of EP-procyclins, one of the major surface GPI-anchored glycoproteins, which is characterized by an internal dipeptide, ϳ20 -30 of glutamate-proline (EP) repeat. Two of the three EP procyclins have a single high mannose N-glycan that is not pro-  Total RNA was isolated from BSF and PCF cells. Reverse transcription was performed with (ϩ) or without (Ϫ) reverse transcriptase (RT). cDNA was amplified using a forward primer from the 5Ј SL sequence and reverse primers, which are specific to TbNST1-4 coding sequences, respectively. RT-PCR products were analyzed on a 1.5% agarose gel. A 100-bp ladder was used to indicate the sizes of the PCR products.
cessed in the Golgi apparatus (7,13,48). However, GPI anchors of procyclins are heavily modified with the large branched poly-Lac-NAc (Gal ␤1-4GlcNAc)-containing side chains with an average of about 8 -12 repeats (49). EP-procyclins from total cell lysates of uninduced and induced cells (for 4 days) were subjected to Western blotting and detected with anti-EP antibody. The results showed that EP-procyclins from TbNST4 silenced cells migrated faster than that of WT cells (Fig. 9A, comparison of lane 2, Tetϩ with lane 1, TetϪ) indicating a reduction in the apparent molecular weights of EP-procyclins. This strongly suggests that silencing TbNST4 gene expression causes underglycosylated EP-procyclins, whereas there is no apparent decrease in total EP. We next asked whether underglycosylated EP-procyclin affects its transport to the cell surface. If this is the case, a reduction of surface EP-procyclins would be expected in TbNST4silenced cells. To examine surface EP-procyclins, live cells uninduced and induced on day 4 after addition of tetracycline were incubated with anti-EP antibody and subjected to flow cytometric analysis. Fig. 9B shows that EP-antibody surface binding to induced cells (reflected by the intensity of fluorescence) was reduced by 48% as compared with that of uninduced cells suggesting a transport defect.
To further examine the distribution and localization of EP-procyclins at the surface and within TbNST4-silenced cells, immunofluorescence analyses were performed with anti-EP antibody with or without cell permeabilization. As shown in Fig. 9C (left panel), in uninduced, permeabilized cells, most EP-procyclins were on the cell surface, whereas in induced cells, significantly less EP-procyclins were seen on the surface, with a compensatory increase in internal staining in a location consistent with the Golgi apparatus in PCF cells (47). In Fig. 9C (right panel), in uninduced, nonpermeabilized cells, EP-procyclins were homogenously distributed at the cell surface, whereas considerably less staining can be seen on the surface of the induced cells. These results are in good agreement with the above flow cytometric analysis. Overall, these results strongly suggest that silencing of TbNST4 results in a glycosylation defect of EP-procyclin and further affected its translocation to the cell surface.
Knock-out of TbNST4 Gene in BSF T. brucei and Its Effect on Cell Growth-As shown in Fig. 4, TbNST4 is expressed in both PCF and BSF life cycle stages of T. brucei. Because BSF T. brucei infects mammals, we decided to examine the role of TbNST4 in growth, glycosylation, and infectivity. As described above, T. brucei is a diploid; we knocked out two alleles of the TbNST4 gene by successive two-step transfections. The first step transfection introduced a gene expressing a neomycin selection marker (NEO) into the locus of the one allele of the TbNST4 gene to generate a single KO (sKO) cell line. The second step transfection introduced a gene expressing a hygromycin selection marker (HYG) into the locus of the other allele of the TbNST4 gene to generate a double KO (dKO) cell line. The tbnst4-null mutants (⌬tbnst4::neo/⌬tbnst4::hyg) were selected in the presence of neomycin and hygromycin B.
Replacement of the two alleles was confirmed by genomic PCR amplification of the tbnst4 locus using two pairs of primers: one pair with a forward primer complementary to 5Ј UTR outside the recombination sequence and a reverse primer com-  Tb927.6.3960 Ϫ b ϩ ϩ ϩ a Although TbNST3 showed a modest restoration of GSII binding to K. lactis mutant cells defective in UDP-GlcNAc transport, the in vitro transport assay does not support this. Therefore, we scored the substrate specificity of UDP-GlcNAc for TbNST3 as ϩ/Ϫ. b The p value obtained from transport assay (TbNST4 versus vector) for UDP-Gal is 0.03 (statistically significant), but the transport activity is low and genetic complementation is negative. Therefore, we scored the substrate specificity of UDP-Gal for TbNST4 as Ϫ.
plementary to the coding sequence (Fig. 10A, left panel); the other pair with a forward primer complementary to the coding sequence and a reverse primer complementary to the 3Ј UTR outside the recombination sequence (Fig. 10A, right panel). Specific PCR products were not detected in the double alleledeleted (dKO) clonal cell lines, A6 and B1 (Fig. 10A, lanes 3 and  4), whereas they were only detected in WT and the single alleledeleted strain (sKO) (Fig. 10A, lanes 1 and 2). This suggests that the endogenous TbNST4 gene was deleted from its genomic locus. To further confirm inactivation of TbNST4 gene expression in dKO cells, RT-PCR was performed using both forward and reverse primers complementary to coding sequences. The results showed that mRNA was not detected in two dKO clonal cell lines, A6 and B1 (Fig. 10B, lanes 3 and 4); they were only detected in WT and the sKO strain (Fig. 10B, lanes 1 and 2) suggesting that expression of the TbNST4 gene was completely ablated in tbnst4-null cells. tbnst4-dKO cells showed a slower growth rate as compared with WT and sKO cells (Fig. 10C). The doubling time of dKO cells were delayed by 2 h as compared with that of WT and sKO cells (6 versus 8 h). This result suggests that, similar to PCF, TbNST4 is important, but not essential for BSF T. brucei growth.
Knock-out of the TbNST4 Gene Causes Glycosylation Defects in BSF T. brucei-Earlier biochemical studies showed that glycoconjugates were deficient in the particular sugar as well as those sugars linked to the particular sugar for which the corresponding nucleotide sugar transport was defective (3,5). We therefore expected defects in glycoconjugates containing GlcNAc and mannose in a tbnst4-null mutant, as TbNST4 transports UDP-GlcNAc and GDP-Man. To assess the glycosylation phenotype in tbnst4-dKO cells, we took advantage of the commercial availability of a variety of biotinylated or fluorescently labeled lectins, which recognize the above two sugarcontaining carbohydrates. Those include (a) ECL lectin, which recognizes a single LacNAc, (b) ConA lectin, which recognizes ␣-linked mannose, (c) RCA I lectin, which recognizes the terminal galactose preferentially linked to GlcNAc, and (d) TL, which recognizes poly-LacNAc. We quantitatively measured the surface binding by flow cytometry of the above four fluorescein lectins to WT and two tbnst4-dKO clonal cell lines. As shown in Fig. 11, surface lectin binding to tbnst4-dKO cells (reflected by the intensity of fluorescence), compared with WT, was reduced by ϳ52 (ECL, Fig. 11A), ϳ49 (RCA1, Fig. 11B), and ϳ55% (TL, Fig. 11C). Interestingly, the ConA lectin binding of two tbnst4-dKO clonal cell lines slightly was increased as compared with that of WT (Fig. 11D) suggesting perhaps, a compensatory effect. Because the three lectins, which show a binding defect in tbnst4-dKO cells, share LacNAc (either single or  poly) as a common recognition carbohydrate structure, the results tentatively suggest that dKO of tbnst4 caused a glycosylation defect in the LacNAc structure.
VSGs are critical for BSF growth and infectivity. The sugar modifications of VSG221 used in this study have been characterized as two N-glycans: one, an oligomannose and the other, a complex glycan containing a LacNAc structure (23,50). Based on the results of the flow cytometric analysis above, a defect in the formation of the complex N-glycan of VSG221 would be expected in tbnst4-dKO cells. To test this, sVSGs and total cell lysates depleted with sVSG were subjected to Western blotting with biotinylated lectins. A significant reduction in binding was detected with ECL for both sVSG (Fig. 12A, left panel) and sVSG-depleted cell lysates (Fig. 12C, left panel) in tbnst4-dKO cells as compared with that of WT (comparison of lanes 2 and 3 with lane 1). No significant differences between WT and dKO were detected with ConA lectin for either sVSG (Fig. 12B, left  panel) or sVSG-depleted cell lysates (Fig. 12D, left panel). Competitive inhibition of lectin binding to blots with the appropriate sugars confirmed specificity (Fig. 12, A-D, right panels).
Similar to the ECL results above, a minor decrease in binding was detected with RCA I lectin for both sVSG (Fig. 13A, left  panel) and sVSG-depleted cell lysates (Fig. 13C, left panel). As predicted, TL did not bind to sVSG, as it lacks poly-LacNAc; this serves as a negative control (Fig. 13B). A slight decrease in binding was detected with TL for sVSG-depleted cell lysates in tbnst4-dKO cells (Fig. 13D, left panel). Again, the ability of appropriate sugars to compete for lectin binding on blots demonstrates specificity (Fig. 13, A, C, and D, right panels). All these biochemical data are consistent with the surface lectin binding results obtained from flow cytometry. Importantly, similar lectin binding results were also obtained from RNAi-mediated TbNST4 knockdown of BSF cells (data not shown) suggesting phenotypic glycosylation specificity. In summary, KO of tbnst4 causes a defect in LacNAc-N-linked glycoproteins and had no apparent effect on high mannose oligosaccharides.
It was important to assess the effect of tbnst4 KO on glycosylation of other proteins in BSF cells; therefore, we examined the synthesis and maturation of p67 and the cathepsin L orthologue (TbCatL) (51,52). Both are well characterized lysosomal glycoproteins. p67 is synthesized as a 100-kDa glycoform in the ER and further modified with poly-LacNAc in the Golgi apparatus resulting in a 150-kDa glycoform (34), whereas TbCatL is synthesized in the ER as a proprotein with 2 oligomannose N-glycans without further processing during transport to the lysosome (53). Because p67 contains the poly-LacNAc structure, we speculated, based on the above lectin binding results, that synthesis of p67 would be affected in the tbnst4-dKO cells. Nascent polypeptides from WT or dKO cells were pulse-labeled with 35 S for 15 min and chased for 2 h. EMK024, a lysosomal cysteine protease inhibitor (54), was used to block turnover of p67 and TbCatL. The extracts were immunoprecipitated with anti-p67 or anti-TbCatL antibodies, respectively. Mature p67 from tbnst4-dKO cells exhibited a faster mobility as compared with that of WT (Fig. 14, comparison of lane 4 with lane 2 in top panel). No difference in mobility between WT and dKO cells was detected for TbCatL (Fig. 14, bottom panel). These results are consistent with the above conclusion, namely, KO of tbnst4 results in a defect in synthesis of LacNAc and has no effect on the synthesis of high mannose oligosaccharides.
The Effect of tbnst4 Knock-out in BSF T. brucei on Growth and Infectivity in Mice-To determine whether loss of tbnst4 expression and the concomitant effect on the glycosylation of VSG impacts the ability of T. brucei BSF to survive in the mammalian bloodstream, 8-week-old C587BL6 mice (five in each group) were inoculated intraperitoneally with 10 2 or 10 3 WT or tbnst4-dKO parasites and parasitemia were monitored daily. Lethal parasitemias (ϳ10 9 /ml) were reached within 5 days with an inoculum of 10 3 WT (Fig. 15A, right panel) and 6 days for 10 2 parasites (Fig. 15A, left panel). With an inoculum of 10 2 WT parasites, all mice died before day 6 post-infection (Fig. 15B); however, with the same inocula of tbnst4-dKO parasites, a clearance of circulating parasites occurred in 2 out of 4 mice between days 5 and 7 with a relapse afterward (Fig. 15C). Thus, tbnst4 deletion in T. brucei resulted in a slight delay in the onset of parasitemia and mortality, possibly related to the slight decrease in the doubling time of these mutants (data not shown), however, overall virulence was not affected. Histograms were generated with FlowJo 9.5.2 software. Unstained (red) uninduced (blue) and induced (green) are shown. C, immunofluorescence analysis of EP-procyclins following TbNST4 RNAi silencing. Uninduced (TetϪ) and induced (Tetϩ) cells (on day 4) were fixed, with (left panel) or without (right panel) permeabilization, and incubated with anti-EP monoclonal antibody (Cedarlane). Cells were detected with anti-mouse IgG (FITC) for EP (green). The images were captured as described in the legend to Fig. 7. DAPI staining for nuclei and kinetoplasts is shown in blue.

DISCUSSION
As a first step toward understanding the roles of TbNSTs in glycosylation of surface glycoconjugates and T. brucei infectivity, we identified eight putative TbNST1-8 from the predicted protein database by comparison of sequence homology with known NSTs. The substrate specificities of TbNST1-4 were experimentally determined using in vitro biochemical transport assays and by in vivo genetic complementation of previously well defined yeast and mammalian mutants, defective in the transport of particular nucleotide sugars as summarized in Table 1. T. brucei is one of the earliest diverging unicellular organisms in eukaryotic evolution; many of the biochemical features of its glycosylation pathways differ from those in its mammalian hosts (49,55). These features include: (a) the modification of most of its membrane and secretory proteins with N-linked glycans (56), whereas O-linked GalNAc, the major type of O-linked glycans of mammalian cells, has not been reported in T. brucei. No UDP-GalNAc was found in a nucleo-  tide sugar pool of T. brucei, where GDP-Man, GDP-Fuc, UDP-Glc, UDP-Gal, and UDP-GlcNAc were identified (57). Unlike the human epimerase (58), the T. brucei homolog does not have epimerase activity to convert UDP-GlcNAc to UDP-GalNAc (59,60). This may explain why GalNAc has yet to be found in any glycoconjugates of this parasite. (b) All the enzymes identified in the biosynthesis of the above five nucleotide sugars were localized to the glycosomes (44,46,59,(61)(62)(63), which are microbodies evolutionarily related to peroxisomes in higher eukaryotes and their function is essential in carbohydrate metabolism in T. brucei. However, in its mammalian hosts, most nucleotide sugars are synthesized in the cytosol. T. brucei nucleotide sugars following their synthesis in glycosomes must be translocated into the cytosol via unidentified membrane transporters. This motivated us to determine the location of TbNST4. By analogy to most known NSTs in eukaryotes, TbNST4 was localized to the Golgi apparatus. (c) In this study we characterized two GDP-Man transporters in T. brucei. GDP-Man transporters have been identified in a variety of lower eukaryotes such as Leishmania, yeast, and fungi and have been shown to be essential for virulence of some Leishmania strains (64, 65) and for cell growth of yeast (66,67). Mammals, however, do not use GDP-Man transport and lack a GDP-Man transporter gene (68). Instead, they exclusively use a lipid donor, dolichol-phosphate-mannose, which is translocated into the ER by lipid flipping (69). The possibility that T. brucei has glycoconjugates with mannose being added in the Golgi apparatus cannot be ruled out. Studies of tbnst3 KO biochemical and biological phenotypes may shed light on this issue. The differences in the requirement for Golgi GDP-Man transport between the lower eukaryotic pathogens and their mammalian hosts makes this transporter a potential target for possible chemotherapeutic intervention.
Although T. brucei exhibits unique characteristics in glycosylation pathways, the identified TbNSTs are conserved in many aspects with known NSTs from other organisms, e.g. their primary amino acid sequence, molecular weight, topology of the predicted transmembrane domains, and biochemical transport characteristics such as K m values in the lower micromolar range. Specifically, TbNST3 and TbNST4 contain a consensus motif "VGXLNK" shared among GDP-Man transporters of Leishmania Lpg2p, fungi Gmt1, and yeast Vrg4p, which has been shown to be involved in affinity and transport in S. cerevisiae (42).
Unique to TbNST4 is its capacity to transport both pyrimidine and purine nucleotide sugars (UDP-GlcNAc, UDP-GalNAc, and GDP-Man). In previous studies, NSTs were found to transport either pyrimidine or purine nucleotide sugars  1 and 4) and two dKO clonal cell lines (A6 and B1, lanes 2 and 3 or 5 and 6) were separated by SDS-PAGE and subjected to Western blotting, which was detected with biotinylated ECL (A) or ConA (B) lectins with (right panels) or without (left panels) the corresponding sugar inhibitors for each of lectins and subsequently, reacted with HRP-conjugated streptavidin. C and D, the same procedure as A and B was applied except that sVSG-depleted cell lysates were used for Western blotting. ECL (C) and Con A (D) blots are indicated. 1 ϫ 10 6 and 7 ϫ 10 6 cell equivalents per lane were loaded for sVSG and sVSG-depleted cell lysates, respectively. VSG was used as loading controls.  1 and 4) and two-allele deleted clonal cell lines (A6 and B1, lanes 2 and 3 or 5 and 6) were separated by SDS-PAGE and subjected to Western blots, which were detected with biotinylated RCA I lectin (A) or tomato lectin (B), with (right panels) or without (left panels) the corresponding sugar inhibitors for each lectins and subsequently reacted with HRPconjugated streptavidin. C and D, the same procedure as A and B was applied except that sVSG-depleted cell lysates were used for Western analyses. 1 ϫ 10 6 and 7 ϫ 10 6 cell equivalents per lane were loaded for sVSG and sVSG-depleted cell lysates, respectively. VSG was used as loading controls.
exclusively. Although a few previous studies suggested that specific NSTs transport both types of substrates, the evidence was not compelling as only a biochemical approach was used with signals below twice the background levels and no genetic complementation analyses were performed (70). The fact that TbNST4 transports both types of nucleotide sugars and also transports UDP-GalNAc, a nucleotide sugar, which has not been reported to date in T. brucei and this organism lacks Gal-NAc in its glycoconjugates, suggests that the recognition and affinity of NSTs for their nucleotide sugar substrates may be less strict than previously found in other organisms.
The important roles of TbNST4 in two major surface glycoproteins, EP-procyclin in PCF and VSG221 in BSF cells, have been demonstrated by specific silencing in PCF and gene disruption in BSF cells, respectively. TbNST4-silenced PCF cells expressed underglycosylated surface EP-procyclins with the lower apparent molecular weights (faster mobilities on SDS-PAGE). We hypothesize that this defect results from impaired synthesis of LacNAc-containing GPI anchor side chains rather than N-glycosylation defects. This is based on the following findings. (a) The procyclic GPI anchor precursors are assembled in the ER and further modified with large heterogeneous glycan side chains by addition of poly-LacNAc structures in the Golgi apparatus (12,15), where TbNST4 resides. (b) Two of the three EP-procyclins have only a single high mannose N-glycan, which is added to the nascent proteins in the ER and is not further processed in the Golgi apparatus. (c) A similar smaller size pattern of procyclins was also observed in null mutants of tbgt8 (encoding a UDP-GlcNAc:␤-Gal-GPI␤1-3GlcNAc transferase) as a result of changes in the procyclin GPI anchor side chains (23).
Silencing of TbNST4 causes a reduction of EP-procyclins at the cell surface suggesting a defect in transport, which may also be the result of an impaired GPI anchor structure. The GPI anchor has been shown to play an important role in forward secretory trafficking (71). Although significant glycosylation defects of EP-procyclins were observed in the TbNST4-silenced cells, this had little impact on cell viability. These results are consistent with previous studies showing that depletion of procyclins on the cell surface, either by impairing GPI anchor synthesis or KO of total procyclin genes (gpeet and ep1-3), had little or no effect on cell growth. Further studies revealed that the absence of cell surface procyclin is compensated by an increase of glycosylinositolphospholipids (72)(73)(74). Importantly, parasites lacking procyclins on their surface were unable to infect their insect vector, tsetse flies (73,75). Based on these studies, we speculate that PCF tbnst4-dKO cells might be impaired in their ability to infect tsetse flies.
Likewise deletion of tbnst4 in BSF cells also caused glycosylation defects in synthesis of LacNAc-containing proteins. This was demonstrated by a decrease in binding of ECL and TL lectins to live cells by flow cytometry (Fig. 11, A and C) and in binding of ECL to sVSG-depleted lysate proteins by lectin blots (Fig. 12C). We expected a glycosylation defect in VSG221 and p67 in tbnst4-dKO cells, because both are modified with complex glycans, which contain either single or poly-LacNAc structures (19,34,76). Indeed, a reduction in the amount of ECL lectin binding was observed for sVSG221 ( Fig. 12A) and an impaired biosynthesis of mature glycoprotein was detected for p67 (Fig. 14). However, no changes in ConA binding to either sVSG or sVSG-depleted total cell lysates were detected in tbnst4-dKO cells (Fig. 12, B and D) suggesting that the oligomannose N-glycans were not affected by KO of tbnst4. Consistent with these results, synthesis and maturation of TbCatL, which is modified only by two oligomannose N-glycans, was not affected by KO of tbnst4. This is most likely because addition of oligomannose N-glycans to nascent proteins in T. brucei occurs in the ER. However, we cannot completely rule out that the GDP-Man transport activity reflects a relaxed substrate specificity by TbNST4 and TbNST3 transporters.
In PCF cells, silencing of TbNST4 causes a drastic reduction in the EP-procyclin molecular weight, whereas, in BSF tbnst4-dKO cells, changes in the apparent molecular weight of VSG221 were not observed. This could be explained by the major differences of processing of GPI anchors in these two different life cycle stages. In PCF cells, GPI anchors of EPprocyclins are extensively modified by large branched poly-LacNAc side chains (12,21). A reduction in EP-procyclin molecular weights indicates a significant decrease in total poly-LacNAc synthesis, whereas in BSF cells, the GPI cores are only added by several terminal or branched galactoses (up to 8) (15,22). KO of tbnst4 in BSF cells appears to only affect the complex sugar structure on one of the two N-glycans, which may not be sufficient to cause a change in electrophoretic mobility, whereas showing a reduction in binding of ECL lectin.
Even though KO of tbnst4 resulted in underglycosylated surface VSGs, no major effect on cell growth or parasite infectivity was observed. We speculate that this may be the result of functional redundancy of NSTs. Several lines of evidence support this hypothesis. (a) Several TbNSTs transport the same nucleotide sugars, e.g. TbNST1, -2, and -4, all transport UDP-GlcNAc and both TbNST3 and TbNST4 transport GDP-Man. Additionally, substrates for the other four putative TbNSTs remain to be identified. (b) Functional redundancy has previously been observed for C. elegans NSTs, e.g. srf-3, a UDP-Gal/UDP-GlcNAc transporter mutant alone did not show any morphological change, but silencing of another NST, C03H5.2 (which transports UDP-GlcNAc/UDP-GalNAc) in srf-3 mutant cells, caused major developmental defects (40,77). (c) We found that in mice inoculated with tbnst4-KO cells, parasitemia was attenuated in the early stage of infection, probably due to a slower cell growth, but thereafter parasites were able to survive, replicate, and induce disease normally as WT. This suggests that other alternate pathways, leading to compensate the function of TbNST4, perhaps, other transporters, bypassed the defect of tbnst4-null mutants. This may be sufficient for cell survival and for having the ability to synthesize glycoconjugates and/or other molecules to fulfill the role of TbNST4 in its absence.
Previous studies have shown that enzymes involved in nucleotide sugar biosynthesis in T. brucei such as UDP-Gal (59,61,78), GDP-Fuc (62), GlcNAc (63), and GDP-Man are essential for cell growth, but none of them were shown to cause pathogenicity changes in BSF T. brucei. We speculate that the function of the totality of TbNSTs is essential for cell growth and that inhibition of a specific or multiple tbnst involved in glycosylation of essential proteins for virulence will change the parasites' infectivity. We will next focus on the KO of other identified TbNSTs, individually and combinatorially to determine their biological roles in this parasite.