The surface of the protozoan Trypanosoma cruzi is covered by a dense coat of mucin-type glycoconjugates, which make a pivotal contribution to parasite protection and host immune evasion. Their importance is further underscored by the presence of >1000 mucin-like genes in the parasite genome. In the present study we demonstrate that one such group of genes, termed TcSMUG L, codes for previously unrecognized mucin-type glycoconjugates anchored to and secreted from the surface of insect-dwelling epimastigotes. These features are supported by the in vivo tracing and characterization of endogenous TcSMUG L products and recombinant tagged molecules expressed by transfected parasites. Besides displaying substantial homology to TcSMUG S products, which provide the scaffold for the major Gp35/50 mucins also present in insect-dwelling stages of the T. cruzi lifecycle, TcSMUG L products display unique structural and functional features, including being completely refractory to sialylation by parasite trans-sialidases. Although quantitative real time-PCR and gene sequencing analyses indicate a high degree of genomic conservation across the T. cruzi species, TcSMUG L product expression and processing is quite variable among different parasite isolates.
- multigene family
- quantitative PCR
- Trypanosoma cruzi
The protozoan Trypanosoma cruzi is the etiological agent of Chagas' disease. In spite of recent successes in disrupting vector-borne transmission in the Southern Cone countries, Chagas' disease remains a major public health issue in Latin America . The variable course of the human disease, which may lead to severe cardiac and/or intestinal complications, has been attributed to the extensive genetic polymorphism from both the exposed human population and the infecting parasite. Indeed, T. cruzi typing schemes based on biochemical and genetic markers converged in the delineation of six major intra-species lineages termed TcI and TcIIa-e , which were recently renamed TcI–TcVI , with TcV and TcVI derived from independent hybridization events between TcII and TcIII stocks [4–6]. Importantly, this genetic heterogeneity underpins differences in biological parameters such as drug sensitivity, pathogenicity, tissue tropism and disease-associated symptoms .
Four major developmental stages can be observed in T. cruzi: bloodstream trypomastigotes and intracellular amastigotes present in the mammalian host, and epimastigotes and metacyclic trypomastigotes in the hematophagous triatomine vector. Transition between developmental forms is associated with a remodelling of the parasite surface coat. Mucins, surface glycoproteins in which the oligosaccharide chains are O-glycosidically linked to threonine or serine residues via NAcGlc units, are major components of this coat involved in parasite protection and host immune evasion [8,9]. They function as the main acceptors of sialic acid residues transferred from host glycoconjugates in a reaction catalysed by parasite-encoded TS (trans-sialidases) . Although T. cruzi is unable to synthesize sialic acid de novo, decoration of parasite surface molecules with this monosaccharide is essential to propagate within the mammalian host . Additional roles for T. cruzi mucins in subverting the vertebrate immune system have been suggested [11–13]. In addition to mucins and TS, the parasite coat is enriched in other glycoconjugates such as gp63 , mucin-associated surface proteins  and ceramide-containing GIPLs (glycoinositolphospholipids) [16,17].
T. cruzi mucins comprise a large gene repository that can be split into two gene families, termed TcMUC and TcSMUG, based on sequence comparisons [9,18]. TcMUC includes more than 1000 genes displaying extensive polymorphism, suggesting they are under significant positive selection favouring diversification of the encoded proteins . Since TcMUC expression seems to be restricted to the mammal-dwelling stages [20,21], the vertebrate immune system is likely the source of this selection pressure. The second mucin family, TcSMUG, displays significantly less diversity and encodes for very small open reading frames containing a putative signal peptide on the N-terminus and a GPI (glycosylphosphatidylinositol) anchor signal on the C-terminus. If these two regions were processed, the average predicted molecular mass for the mature apo-mucins would be ~7 kDa with threonine representing as much as 50% of the residues. In the reference clone CL Brener, TcSMUG is composed of two groups of genes, named L and S, organized in independent tandem arrays and differing in the structure of their genomic loci . Within the coding region, the deduced sequences of TcSMUG S and TcSMUG L display an average >80% identity and higher similarity values because of conservative changes.
Recent MS analyses of T. cruzi GPI-anchored molecules identified TcSMUG S products as the backbone for the 35/50 kDa mucins expressed on the surface of insect-dwelling stages (Gp35/50 mucins) . Gp35/50 mucins bind to target cells in a receptor-mediated manner and induce a bidirectional Ca2+ response which probably contributes to host cell invasion by metacyclic trypomastigotes . Gp35/50 mucins are routinely detected using mAbs [monoclonal Abs (antibodies)] directed towards their attached oligosaccharides. In particular, mAb 2B10, which reacts with Galp-containing epitopes present in Gp35/50 mucins from all parasite isolates and mAb 10D8, which recognizes Galf-containing epitopes restricted to Gp35/50 mucins from TcI isolates . As for TcSMUG L, none of the T. cruzi proteomic data sets revealed the presence of these products [23–25]. Levels of mRNA are differentially regulated among parasite developmental stages , being more abundant in replicative forms (amastigotes and epimastigotes). However, the complex post-transcriptional regulation driving gene expression in trypanosomatids and the presence of functional stage-specific AU-rich destabilizing motifs in the TcSMUG L 3′UTR (untranslated region)  strongly suggest that protein abundance might not correlate with steady-state mRNA levels. In the present study, we investigated the TcSMUG L group of genes across the T. cruzi species and undertook a detailed characterization of its encoded products.
The parasite stocks used were Sylvio X-10 c17 and Adriana (TcI), IVV cl4 (TcII), M6241 cl6 (TcIII), 92122102R (TcIV), SC43 cl92 (TcV), Tula cl2, RA and CL Brener (TcVI), and T. cruzi marinkellei clB7. Typing of T. cruzi evolutionary lineages was carried out by PCR  (results not shown). Parasite developmental forms were obtained and purified as described previously [19,20].
DNA extraction and gene amplifications
T. cruzi genomic DNA was purified by the phenol-chloroform method . Gene amplifications were obtained by PCR using 1–10 ng of DNA as template and Taq DNA Polymerase High Fidelity (Stratagene). The primers Fw 5′UTR and Rev TGA (see Supplementary Table S1 at http://www.BiochemJ.org/bj/438/bj4380303add.htm) were designed on TcSMUG conserved regions, thus leading to simultaneous amplification of both TcSMUG S and TcSMUG L genes. Amplicons were purified, cloned into the pGEM-T easy vector (Promega) and used to transform TOP10F cells (Invitrogen). DNA sequencing was carried out at our own facility.
Analysis of sequences
Multiple sequence alignments were calculated with T-Coffee . To calculate phylogenetic trees, each alignment was manually edited to remove the threonine-rich regions. Then a larger dataset (100 replicates with re-sampling and permutation) was generated using SEQBOOT, with all default options. Finally, trees were calculated for all of the re-sampled alignments using the maximum likelihood programs PROML (protein sequences) or DNAML (DNA sequences), and the majority rule consensus tree, along with the bootstrap support from re-sampling, was calculated using CONSENSE. SEQBOOT, PROML, DNAML and CONSENSE are part of the Phylogeny Inference Package . Visualizations of the resulting trees were generated by Archaeopteryx  and manually edited using Inkscape (http://inkscape.org/).
Real-time qPCR (quantitative PCR)
A 7500 Sequence Detection System device (Applied Biosystems) was used. The 20 μl reaction tube contained 1×SYBR Green Reaction Buffer (Applied Biosystems), DNA, 0.3 μM of each primer, 3 mM MgCl2, 1 mM of each dNTP, 0.01 unit/ml uracil-DNA glycosylase and 0.025 unit/ml Taq Gold DNA Polymerase. After 2 min of pre-incubation at 50 °C followed by 10 min at 95 °C (denaturing step), PCR amplification was carried out for 40 cycles (95 °C for 15 s and 60 °C for 60 s). qPCR primers (see Supplementary Table S1) were designed using Primer Express (Applied Biosystems). To verify the specificity, the reactions were (i) subjected to a heat dissociation protocol following the final cycle of the PCR  and (ii) fully sequenced. The analytical sensitivity of the qPCR was tested by using serial dilutions of DNA with each primer combination. The detection limits were ~3 pg of DNA with a dynamic range of ~0.01–10 ng (results not shown). Within this range, the efficiency of the qPCRs was estimated as ~100%, since 10-fold differences in DNA rendered variations in Ct (cycle threshold) values of ~3.32 (results not shown). The amount of DNA (0.3 ng) was finally chosen because the amplicons were detected at approximately Ct number 20, the mid-point of the qPCR dynamic range. Samples without T. cruzi DNA or primers were used as negative controls. Each qPCR experiment was performed in triplicate with at least 3 independent samples, and a mean value±S.D. was calculated for each experiment. The coefficients of Ct variation were less than 4% among different experiments. TcSMUG gene copy number was estimated by testing each DNA sample in parallel against the PDhα (pyruvate dehydrogenase α; Tc00.1047053507831.70) gene, a procedure for the correction of variations in DNA quality/quantity. Estimation of gene dosage was calculated according to the ΔΔCt method .
Total parasite RNA was size-fractionated, blotted and hybridized as described previously . The smugL probe was synthesized by PCR using oligonucleotides smugL 3′UTR-fw and smugL 3′UTR-rev (see Supplementary Table S1) and radiolabelled using [α-32P]dCTP (PerkinElmer).
Expression of recombinant TcSMUG L protein in bacteria and Ab development
The oligonucleotides smugLab-fw and smugLab-rev (see Supplementary Table S1) were annealed and cloned into the BamHI and SpeI restriction sites of a tailored version of the pGEX-1λT (GE Healthcare) vector in which the sequence 5′-CCCGGGACTAGTACGACTT-3′ had been previously inserted between the original BamHI and EcoRI restriction sites. The following schematic fusion protein was thus obtained upon expression: GST (glutathione transferase)–KAAGGDPKKNTSTT. GST–smugL protein was expressed and purified by affinity chromatography , and injected into animals as described previously . Abs were affinity-purified using the smugL peptide coupled to sulfo-link® resin (Pierce) and used for Western blot analysis at a 1:500 dilution. Other Abs directed against T. cruzi molecules used were rabbit anti-serine carboxypeptidase , rabbit antiglutamate dehydrogenase , rabbit anti-gp63 , rabbit anti-smugS , and mouse mAbs 10D8 and 2B10 to Gp35/50 mucins . For Western blotting, the first two Abs were used at a 1:3000 dilution and the remainder used at a 1:500 dilution.
Peptides were synthesized by Sigma-Genosys, acetylated on their N-termini and coupled through their C-terminal cysteine residue to maleimide-activated BSA (Pierce) . The sequences were AVFKAAGGDPKKNTTC (smugL) and VEAGEGQDQTC (smugS). Displacement assays were performed by pre-incubating the anti-smugL Abs with 0.5 μg/μl of the indicated BSA–peptide for 1 h at room temperature (25 °C).
Construction of epitope-tagged TcSMUG genes and T. cruzi transformation
A single FLAG tag was inserted within the threonine-rich region of the TcSMUG genes by PCR using the oligonucleotides SmugFLAG and T7 (see Supplementary Table S1). The templates used in each case were TcSMUG L and TcSMUG S genes isolated from the SC43 cl92 stock and previously cloned into the pGEM-T easy vector. These genes were chosen because they contain a single StyI site. Amplicons were digested with StyI and cloned into the corresponding template in which the fragment StyI–StyI (the second StyI was provided by the pGEM-T easy vector itself) has been excised. Several clones were sequenced and one displaying the proper orientation and another clone displaying the reverse orientation were subcloned into the pRIBOTEX vector  using the pGEM-T easy vector-encoded EcoRI restriction sites flanking the whole construct. Epimastigotes from the Adriana strain were transformed by electroporation, selected using 500 μg/ml G418 (Gibco) and used as populations after 45 days of selection . Transcription of the recombinant genes was analysed by reverse transcription–PCR , and protein products were identified by Western blot using either rabbit polyclonal anti-FLAG Abs (1:2000 dilution) or mouse anti-FLAG mAb (1:5000 dilution) (Sigma).
Gel electrophoresis and Western blots
Gel electrophoresis was performed using SDS/PAGE (12.5–15% gels). For Western blots, ~107 parasites were loaded into each lane, transferred on to PVDF membranes (GE Healthcare), reacted with the appropriate antiserum followed by HRP (horseradish peroxidase)-conjugated secondary Abs and developed using enhanced chemiluminescence (Pierce) . When comparing different developmental stages, 5 μg of proteins/lane (as determined using a Bradford assay) were loaded.
Parasites were harvested, washed in PBS and processed for IFAs (immunofluorescence assays) as described previously . Analysis was performed in a Nikon Eclipse E600 microscope coupled to a SPOT RT color camera (Diagnostic Instruments).
ConA (concanavalin A) fractionation and PNGase F treatment
Parasite pellets were homogenized in ConA buffer [50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 1% Nonidet P40, 0.1% sodium deoxycholate, 1 mM PMSF, 50 μM Tos-Lys-CH2Cl (‘TLCK’; tosyl-lysylchloromethane) and 1 mM DTT (dithiothreitol)] and fractionated on 200 μl of ConA–Sepharose (GE Healthcare) . Elution was carried out with 300 μl of ConA buffer with 0.5 M α-methylmannoside added (Sigma). Aliquots of GPI-anchored proteins (see below) were added with 0.15% SDS and 7.5 mM DTT, boiled for 5 min and treated with 5 units of PNGase F from Elizabethkingia meningoseptica (Sigma) .
Purification of GPI-anchored proteins and PI-PLC (phosphoinositolspecific phospholipase C) treatment
Pellets containing 1.5×108 parasites were homogenized in 2 ml of GPI buffer [10 mM Tris/HCl, pH 7.4, 150 mM NaCl, 2% Triton X-114, 1 mM PMSF and a protease inhibitor cocktail (Sigma)] on ice for 1 h . The homogenate was centrifuged at 8800 g for 10 min at 0 °C and the supernatant (S1) was stored at −20 °C for 24 h. The pellet (P1) was washed with 1 ml of buffer A (10 mM Tris/HCl, pH 7.4, 150 mM NaCl, 0.06% Triton X-114 and 1 mM PMSF) and stored. S1 was thawed and submitted to phase separation at 37 °C for 10 min followed by centrifugation at 3000 g for 3 min at room temperature. The upper phase (S2) was collected and the detergent-rich phase was re-extracted with 1 ml of buffer A. The upper phase (S3) was collected, and the detergent-rich phase was extracted with 1 ml of buffer A, homogenized, incubated for 30 min at 0 °C and centrifuged at 18000 g for 10 min at 0 °C. The pellet (P2) was washed with 1 ml of buffer A and stored, whereas the supernatant (S4) was submitted to a new phase separation. The upper phase (S5) was collected and the lower detergent-rich phase, enriched in GPI-anchored proteins, was taken as the GPI fraction (GPI). Before SDS/PAGE, P1 and P2 fractions were sonicated and resuspended in denaturing loading buffer containing 6 M urea. Aliquots of the S4 fractions were precipitated with cold acetone, resuspended in 200 μl of 10 mM Tris/HCl, pH 7.2, 0.1% sodium deoxycholate and treated for 4 h with 0.3 unit of PI-PLC from Bacillus cereus (Invitrogen) at 37 °C. Samples were then re-precipitated, resuspended in 200 μl of GPI buffer and submitted to phase separation. Both the upper aqueous phase and the detergent-rich phases were precipitated with cold acetone and resuspended in denaturing loading buffer containing 6 M urea.
Epimastigote forms (1–3×109) were delipidated by water/chloroform/butan-1-ol treatment as described previously . Briefly, the soluble fraction was evaporated under N2 stream and the insoluble material was re-extracted with 66% butan-1-ol in water at 4 °C. The butan-1-ol phase (F1) contains mainly lipids, phospholipids and GIPLs, whereas the aqueous phase (F2) is enriched in mucins . Both phases were further extracted with 9% butan-1-ol in water. Delipidated parasite pellets were also extracted with 9% butan-1-ol in water and the mucin-rich aqueous phase (F3) was stored. Final parasite pellets were resuspended in denaturing loading buffer containing 6 M urea and 100 μg/ml DNAse I (Sigma). Alternatively, parasite pellets were delipidated as above and extracted with heated 44% aqueous phenol as described previously .
Parasite pellets were resuspended (106/ml) in ice-cold immunoprecipitation buffer (150 mM NaCl, 50 mM Tris/HCl, pH 7.6, 1 mM EDTA, 0.1% Nonidet P40, 1% Triton X-100, 100 μM Tos-Lys-CH2Cl and 1 mM PMSF). After centrifugation, supernatants were fractionated into 25 μl of anti-FLAG–Sepharose (Sigma).
Surface labelling of epimastigotes
Epimastigote forms were extensively washed in ice-cold PBS and labelled in the presence of 30 ng of purified recombinant TS , 10 mM 2-deoxyglucose (Sigma) and 1 mM of the azido-sialyllactose analogue Neu5Azα2-3LacβOMe  for 30 min at room temperature. The reaction was heated to 65 °C to inactivate TS and labelled by the Staudinger method with 250 μM Phosphino-FLAG (Sigma) for 16 h at room temperature . Alternatively, epimastigote carbohydrates were labelled with 1 mg EZ-link Biotin-LC-hydrazide (Pierce), washed and fractionated into 25 μl of StreptAvidin-agarose (Sigma) .
Analysis of TcSMUG L genes and predicted proteins in T. cruzi
Genomic organization of the TcSMUG family of genes has been analysed in CL Brener stocks by Southern blot of cosmid libraries, yielding a complex hybridization pattern upon which the existence of ~80 genes per haploid genome was postulated . This number is compatible with that estimated using a shotgun coverage approach , but much larger than the final 19 copies of TcSMUG genes per haploid genome estimated after completion of the T. cruzi genome. To address this issue, and to look for possible intra-species variation in TcSMUG gene dosage, we estimated gene copy numbers in parasite stocks belonging to the six major T. cruzi lineages using a qPCR approach. Data was collected and normalized to the PDhα gene, which displays minimal intra-species dosage variability (Table 1). With the exception of the IVV cl4 stock, which presents an apparent significant reduction in TcSMUG S genes, these results show minor variations in the number of copies of TcSMUG genes across the species, where the number of TcSMUG L genes is consistently higher than those of TcSMUG S genes. Given that PDhα is a single-copy gene in T. cruzi [35,42], and considering that standard curve slopes indicate ~100% efficiency for every qPCR reaction (results not shown), the number of TcSMUG L and TcSMUG S genes can be roughly estimated as ~10–17 and ~5–12 per haploid genome respectively.
We then looked for sequence diversity in TcSMUG genes. For that purpose we amplified and sequenced several copies of TcSMUG genes from a panel of T. cruzi stocks representative of different evolutionary lineages (GenBank® accession numbers JN051897–JN052022). The amplification products span >95% of the coding sequence and include a ~70 bp region of the 5′ UTR. A total of 116 amplicons were analysed, representing coverage of 12–20 sequences for each T. cruzi group. Additional sequences from CL Brener, Y (TcII) and Sylvio X-10 stocks were retrieved from public databases and included in the analysis. In line with previous data , multiple sequence alignments at both the nucleotide and amino acid levels show an overall high degree of conservation at the N- and C-termini, whereas most of the inter- and intra-lineage differences accumulated at the threonine-rich core. These differences involve some insertions/deletions of threonine codons and/or entire repetitive units. An unrooted phylogenetic tree calculated from alignments containing only the more conserved terminal regions, i.e. devoid of the threonine-rich core, is shown in Supplementary Figure S1 (at http://www.BiochemJ.org/bj/438/bj4380303add.htm). The tree shows two main branches, supported by significant bootstrap values, clustering sequences from each TcSMUG group. One additional branch, also supported by significant bootstrap value, clusters sequences from TcSMUG-like genes obtained from the bat trypanosome T. cruzi marinkellei that were used as an outgroup. Within both T. cruzi branches, the tree also shows further clustering of sequences, although with lower bootstrap values, in groups that are logically consistent with the accepted phylogeny of the species. For instance, a group of TcSMUG S sequences from TcV/TcVI stocks clusters with TcIII-derived sequences, whereas another group, positioned far from the previous one, clusters with TcII-derived sequences. Assuming that TcV/TcVI stocks originated from hybridization events between TcII and TcIII stocks [4–6], it is likely that each group of TcV/TcVI sequences is derived from different allele(s) provided by either parental stock. The lower bootstrap values supporting all these additional grouping might be explained by the short length of the sequences used in our analysis, which makes the alignment very sensitive to the re-sampling with permutation technique used for bootstrapping. Notwithstanding these methodological issues, manual inspection of the multiple sequence alignment of TcSMUG complete proteins shows a number of prominent features that support our phylogenetic tree (Figure 1). First, a number of key substitutions including (i) the distal portion of the N-terminal region, (ii) the residues surrounding the aspartate residue that function as GPI acceptor and (iii) the repetitive unit core, clearly set apart members of TcSMUG S and TcSMUG L groups. Secondly, within both TcSMUG S and TcSMUG L products, there is a set of substitutions that, when considered as a whole, are diagnostic for each T. cruzi phylogenetic group. Hybrid strains (TcV/TcVI) indeed show two set of sequences differing at these positions, thus allowing for the unambiguous tracking of the parental allele from which they are likely derived.
Expression of TcSMUG L mRNAs and proteins in T. cruzi
As a first step toward identification of TcSMUG L products, we produced polyclonal Abs against an N-terminus-derived peptide. We chose this peptide because it is highly conserved among TcSMUG L sequences (Figure 1) and topology predictions indicate that, in contrast with the highly O-glycosylated threonine-rich region, it probably hovers above the parasite glycocalyx in its ‘naked’ form. A similar architecture, in which a threonine-rich region undergoing extensive glycosylation acts as a scaffold to display an outmost N-terminal peptide, has been proposed for TcMUC I products . To avoid cross-recognition of TcSMUG S products, the peptide immunogen was limited to the portion of the TcSMUG N-terminal region that differs between L and S products (Figure 2A). Although mice and rabbit antisera specifically recognized the smugL synthetic peptide (Figure 2B), they did not clearly label any developmental stage of the CL Brener clone of T. cruzi in IFA or Western blotting over total parasite lysates (results not shown). However, when we used a fraction enriched in glycoconjugates (ConA-extract), a single ~35 kDa band was detected in epimastigote forms by both mouse (Figure 2C) and rabbit antisera (Figure 2D). Recognition of this band was abolished when the antisera were pre-adsorbed with the cognate peptide, but not with a control peptide spanning the TcSMUG S region, thus reinforcing its specificity. In addition to this epimastigote-restricted band, and only when using the rabbit antiserum, we detected a high molecular mass smear (>150 kDa) in the amastigote forms. Reactivity of this smear was also abolished by addition of the smugL peptide.
When we probed ConA-extracts from epimastigotes from different parasite stocks, we observed a double ~38/48 kDa band in Sylvio X-10, whereas no signal was detected for other stocks such as RA and Adriana (Figure 2E). These results were normalized by re-probing the membrane with an antiserum directed against serine carboxypeptidase . Analyses of Northern blots probed with a region of the 3′UTR of a TcSMUG L gene  also revealed substantial intra-specific differences in the amounts of mRNA steady-state levels, being significantly more abundant in CL Brener (Figure 2F). Overall, these results indicate that TcSMUG L-encoded products can be detected in replicating stages (epimastigotes and amastigotes), although their pattern of mRNA and protein expression/processing seems quite variable among T. cruzi isolates/developmental forms.
TcSMUG L genes code for GPI-anchored, mucin-like surface proteins
To analyse if TcSMUG L products behave as mucin-type proteins, i.e. undergo extensive O-glycosylation, different approaches were followed. First, we purified mucins from epimastigotes following a standard butan-1-ol extraction protocol . As shown, CL Brener SMUG L products were detected in the F2 and F3 fractions (Figure 3A), which are highly enriched in mucin-type glycoconjugates as verified by 10D8 mAb reactivity (Figure 3B). SMUG L products from the Sylvio X-10 strain, on the other hand, were detected exclusively in the F3 fraction, suggesting an even more hydrophilic nature . The differential fractionation patterns in butan-1-ol further support strain-specific variations in the post-translational processing of TcSMUG L products from either parasite stock. Partial degradation, as revealed by the presence of additional, low molecular mass bands, can be inferred for both mucins preparations. As an additional control, we probed Sylvio X-10 fractions with an antiserum directed against gp63, a non-mucin type, GPI-anchored surface glycoprotein , which gave a faint signal exclusively in the delipidated parasite pellets (Figure 3B). TcSMUG L proteins, similarly to Gp35/50 mucins, were also recovered in the aqueous phase following an alternative phenol/water extraction procedure, whereas non-mucin-type control proteins partitioned to the phenolic phase  (see Supplementary Figure S2 at http://www.BiochemJ.org/bj/438/bj4380303add.htm). Taken together, these results strongly suggest that the extent of glycosylation present in SMUG L products is enough as to be considered mucin-type glycoproteins.
To verify the functionality of the predicted GPI-anchoring signal in TcSMUG L products, total GPI-anchored proteins were purified from CL Brener and Sylvio X-10 epimastigotes taking advantage of their preferential fractionation in Triton X-114 . Aliquots corresponding to the different purification steps were analysed by Western blotting. As observed in Figure 3(C), SMUG L reactivity is maximal in GPI fractions from both parasite stocks. Additional low molecular mass bands present in the Sylvio X-10 GPI fraction could be attributed to partial SMUG L degradation. More interestingly, and again in the case of Sylvio X-10 epimastigotes, a SMUG L-reactive ladder-like pattern is observed in the P2 fraction, which contains mostly membrane-associated molecules excluded from GPI- and sterol-rich micro-domains . Whether this signal corresponds to immature SMUG L products en route to the parasite surface or to a spurious cross-recognition is being currently investigated. Finally, when a similar analysis was carried out in CL Brener amastigote forms, the high molecular mass SMUG L-reactive smear partitioned exclusively to the S2 fraction (Supplementary Figure S2). Since this fraction contains mostly cytoplasmic molecules , this smear was not further analysed.
We next treated purified GPI-anchored molecules from CL Brener and Sylvio X-10 epimastigote stocks with PI-PLC, which specifically cleaves GPI anchors. Following this treatment, samples were submitted to a new Triton X-114 phase separation and both the aqueous phase containing the PI-PLC-solubilized molecules and the detergent-rich phase containing the lipid-anchored molecules were analysed by Western blotting. As shown, addition of PI-PLC caused the complete disappearance of SMUG L-reactive bands from the detergent-rich phases, although these bands were not recovered in the aqueous phases (Figure 3D). A similar biochemical behaviour, which supposedly relies on the inability of highly hydrophilic GPI-free mucin-type species to bind to membrane supports , was observed for Gp35/50 mucins and for TcMUC-encoded trypomastigote mucins , but not for other GPI-anchored glycoconjugates displayed on the T. cruzi surface such as Gp63. Altogether, these results are compatible with SMUG L products being anchored to the outer leaflet of the epimastigote membrane, most likely through a GPI moiety.
TcSMUG L products are not acceptors of sialic acids
One key functional aspect of T. cruzi mucins is their role as sialic acid acceptors on the parasite surface . To test whether SMUG L products may also function as sialic acid acceptors, we carried out parasite sialylation assays in the presence of exogenously added TS and Neu5Azα2-3LacβOMe  as a sialyl residue donor. Neu5Az incorporated in parasite glycoconjugates can be detected by the reaction of the azido group with a phosphine-tagged FLAG. As a result, these glycoconjugates would have a FLAG epitope covalently attached to the Neu5Az residue, which are then detected by immunochemical techniques. The exquisite chemical selectivity of the overall system has been recently demonstrated on T. cruzi cell-derived trypomastigotes .
Epimastigote forms from CL Brener, Sylvio X-10 and Adriana parasite stocks became strongly labelled upon addition of TS and Neu5Azα2-3LacβOMe, indicating the incorporation of the derivative sialyl residue at the cell surface (Figure 4A and results not shown). Control IFA carried out in the absence of either exogenous TS or Neu5Azα-3LacβOMe rendered negative results. Anti-FLAG immunoprecipitation assays on labelled epimastigotes followed by Western blotting show a pattern of FLAG-reactive bands compatible with the Gp35/50 mucins (Figure 4B). Furthermore, when these samples were probed by Western blotting with 10D8 or 2B10 mAb, part of the reactive species were found in the anti-FLAG immunoprecipitated fraction, indicating that Gp35/50 mucins indeed function as sialic acid acceptors in the conditions of the assay. On the contrary, smugL-reactive bands were observed exclusively in the flow-through fractions. Related experiments in which surface carbohydrates from CL Brener epimastigotes were labelled with a non-permeable biotin derivative demonstrate that TcSMUG L products, compared with cytoplasmic markers, are indeed accessible on the parasite surface (Figure 4C). Taken together, these findings indicate that, despite being exposed on the epimastigote surface, smugL products are not acceptors of sialic acid either because of steric hindrance or, more likely, because they do not bear terminal β-galactoses in the correct configuration. Attempts to immunoprecipitate smugL products to carry out an in-depth analysis of their glycan portions have been so far unsuccessful.
TcSMUG L product is expressed and properly processed in transfected T. cruzi epimastigotes
To evaluate whether TcSMUG L products can be expressed and properly processed in a ‘restrictive’ strain, i.e. a strain in which the endogenous TcSMUG L products cannot be detected, one representative member of the TcSMUG L gene group was FLAG-tagged (see Supplementary Figure S3A at http://www.BiochemJ.org/bj/438/bj4380303add.htm) and transfected into epimastigotes from the Adriana strain. As controls, we transfected parasites with a FLAG-tagged version of a TcSMUG S gene and with plasmids containing either construct in reverse orientation. Neither construct led to a noticeable growth phenotype on transfected parasites (see Supplementary Figure S3B). After 45 days of selection, parasites were analysed for the expression of FLAG-containing constructs. Reverse transcription–PCR assays revealed the expression of these constructs in transfected parasites (results not shown). Moreover, a mAb directed against the FLAG epitope detected a single ~13–15 kDa band in total lysates from TcSMUG S- and TcSMUG L-transfected parasites, but not from those transfected with control plasmids (Figure 5A).
Following Triton X-114 fractionation, FLAG-tagged molecules were highly concentrated in the final detergent-rich phase, indicating they are indeed GPI-anchored on the parasite surface. More importantly, additional ~27 and ~35 kDa FLAG-reactive bands, in both cases restricted to the GPI fraction, could be detected in TcSMUG L- and TcSMUG S-transfected parasites respectively (Figure 5B). Probing of GPI fractions with Abs directed either against the TcSMUG L or TcSMUG S N-terminal region indicated that the upper bands, but not the ~13–15 kDa ones, were reactive. Upon PI-PLC treatment, all FLAG-reactive bands showed the same biochemical behaviour described for the endogenous TcSMUG L products (Figure 5C). IFA studies on non-permeabilized epimastigotes allowed for the identification of TcSMUG L and TcSMUG S-transfected parasites showing peripheral FLAG reactivity (Figure 5D), further supporting membrane localization of these products. Additional FLAG-reactive vesicle-like structures could be observed in the vicinity of these parasites, particularly near the flagellum, suggesting active secretion of TcSMUG products. Altogether, these results are compatible with TcSMUG products being expressed and anchored to the surface of the transfected epimastigotes, most likely through a GPI moiety.
The mucin-type nature of FLAG-containing products was evaluated following butan-1-ol extraction of transgenic parasites. TcSMUG S–FLAG was detected in both F2 and F3 fractions (Figure 6A), which coincides with endogenous Gp35/50 mucins (Figure 3B). Interestingly, the TcSMUG L–FLAG product is detected exclusively in the F3 fraction, which coincides with the pattern verified for TcSMUG L endogenous products expressed in a different TcI strain (Sylvio X-10; Figure 3A). On the other hand, the FLAG-reactive, low molecular mass bands expressed by both transgenic parasites were detected in the F2 and, to a lesser extent, F3 fractions. This fractionation pattern suggest a more hydrophobic nature for these lower bands, and is reminiscent of GIPLs, free GPI species that are very prevalent on the T. cruzi surface . Upon fractionation of parasite membrane-rich preparations on ConA resin, these ~13–15 kDa FLAG-reactive products were recovered in the unbound fractions, whereas the upper FLAG-reactive bands were only eluted from the resin upon addition of α-methylmannoside. Taken together, these findings favour the idea of the ~13–15 kDa FLAG-reactive species being the truncated forms spanning a GPI-anchored version of the FLAG epitope, although other alternatives such as them being non-glycosylated or hypo-glycosylated forms might be considered.
Transgenic parasite lysates were fractionated using anti-FLAG affinity chromatography and analysed by Western blot. The TcSMUG S–FLAG full-length product, but neither the TcSMUG L–FLAG full-length product nor the ~13–15 kDa forms, is recognized by the 10D8 and 2B10 mAbs (Figure 6C and results not shown). This finding provides additional evidence supporting TcSMUG S products as the backbone for Gp35/50 mucins  and further highlights differences in the glycosylation pattern of TcSMUG L and TcSMUG S products. The latter aspect was also shown by the use of PNGase F. Despite bearing two consensus signals for the attachment of N-linked oligosaccharides (Supplementary Figure S3A), TcSMUG L–FLAG products are not modified upon PNGase F treatment (Figure 6D). Similar assays also suggest an absence of N-glycosylation in TcSMUG L endogenous products from both CL Brener and Sylvio X-10 isolates (results not shown). In contrast, the ~35 kDa TcSMUG S product, but not the ~13–15 kDa one, treated in parallel, slightly increased its electrophoretic mobility (Figure 6D), indicating it is indeed N-glycosylated in its unique consensus signal (Supplementary Figure S3A). A similar shift was observed for endogenous Gp35/50 mucins upon PNGase F treatment, as revealed by the 10D8 mAb. These results indicate that transgenic, TcSMUG ‘full-length’ products undergo complex post-translational modifications that correlate with those observed for the endogenous products.
TcSMUG-deduced products have typical features of insect stages apo-mucins, including lack of major sequence variability and the presence of a threonine-rich region in the central domain, which constitute an excellent substrate for the O-glycosylation initiating enzyme of T. cruzi . Indeed, TcSMUG S products were identified as the backbone for the Gp35/50 mucins expressed on the surface of insect-dwelling stages of the T. cruzi lifecycle . In the case of TcSMUG L products, however, both their mucin-type nature and expression profile were inferred based merely on genomic features: deduced amino acid composition, presence of numerous putative glycosylation sites and, chiefly, high level of structural similarity to TcSMUG S products. In the present study we present multiple evidence indicating that TcSMUG L genes indeed encode mucin-type glycoconjugates restricted to the epimastigote forms of T. cruzi. These features are supported by the in vivo tracing and characterization of both endogenous TcSMUG L products and recombinant tagged molecules expressed by transfected parasites.
Besides displaying substantial structural homologies, comparative analyses highlight certain differences between TcSMUG L and TcSMUG S products. On one hand, TcSMUG S products are N-glycosylated in vivo. Intriguingly, TcSMUG L products bear a putative N-glycosylation sequon which, although non-functional, is highly related to one that undergoes N-glycosylation in TcSMUG S products. Furthermore, depending on the evolutionary origin of the encoding allele, TcSMUG L products contain one or two additional N-glycosylation signals at the junction between the N-terminal region and the threonine-rich region, neither of which seems to be modified in vivo. Another major difference is that TcSMUG L products, unlike TcSMUG S ones, are not acceptors of sialic acid residues. The molecular basis underlying this observation is uncertain, although the fact that TcSMUG L carbohydrates are readily labelled with a non-permeable biotin derivative argues against a steric hindrance and/or accessibility phenomenon. However, absence of terminal β-Gal residues in the proper configuration in TcSMUG L glycans, which is further supported by their lack of reactivity with the 10D8 and 2B10 mAbs, seems a more appealing hypothesis. Indeed, molecules highly glycosylated at serine/threonine residues, but lacking terminal Gal residues, have been described in the epimastigote surface coat . In-depth glycan analysis of TcSMUG L endogenous/transgenic products will be instrumental to clarify this issue. From a biological standpoint, the absence of sialyl acceptor residues is consistent with the fact that expression of TcSMUG L products is restricted to the epimastigote stage, in which TS activity is minimal and sialic acid incorporation does not have a clear functional correlate [10,17]. Overall, structural and functional differences between TcSMUG L and TcSMUG S products, together with a distinct expression profiling along the T. cruzi life cycle, demonstrate that TcSMUG L products constitute a previously unrecognized kind of mucin-like molecule on the surface of the parasite.
In spite of the high degree of genomic conservation across the T. cruzi species, TcSMUG L expression and processing seems quite variable among different parasite isolates. In some cases, e.g. CL Brener compared with Sylvio X-10 stocks, these inter-strain variations are relatively minor and might be attributed to the expression of a distinct repertoire of genes differing in size and/or to the presence of a different set of glycosyltransferases acting on TcSMUG L products on either parasite stock. In other cases, however, these inter-strain variations are substantial and led to an apparent lack of TcSMUG L products in certain strains, which were thus referred to as ‘restrictive’ isolates. One possibility underlying these differences might be related to the detection technique. Since our Abs are directed towards a single peptide within these molecules that might be target of proteases, expression of TcSMUG L products on ‘restrictive’ strains and even on parasite developmental forms other than epimastigotes cannot be formally excluded. However, two arguments can be raised against this hypothesis. First, inter-strain TcSMUG L variations, although not exactly mirroring protein levels, were also observed at mRNA steady-state levels. Secondly, a tagged TcSMUG L product is readily expressed and detected on the parasite surface upon transfection into epimastigotes from a ‘restrictive’ strain. These findings, together with cumulative evidence indicating that control of gene expression is essentially driven by post-transcriptional mechanisms in this organism , support the idea that quantitative inter-strain variations in TcSMUG L expression are due to cis- and/or trans-acting factors operating on the UTRs of the mRNAs. Cloning and functional evaluation of TcSMUG L UTRs from different strains is currently underway.
Variations in the expression/processing of T. cruzi mucins have been reported. Indeed, differential recognition of Gp35/50 mucins by mAbs relies on the presence of strain-specific carbohydrate epitopes . These and other differences in their carbohydrate (and perhaps lipid) composition do have a correlate on the adhesiveness/signal transduction properties of these molecules, which may have been selected for in the host–parasite interplay . In a broader sense, structural and antigenic analyses led to the identification of quantitative and qualitative differences on multiple components of the T. cruzi surface coat across the species [7,10,46]. At variance with most of these findings, though, differences in the expression/processing of TcSMUG L products are unique in that they cannot be explained by a clear phylogenetic rationale.
Finally, the remarkable conservation of TcSMUG L deduced products along the predicted mature N-terminal peptide suggest they are under positive selection against diversification. Because TcSMUG L expression seems restricted to the parasite forms dwelling in an insect host, which has a non-specific immune response, it may be speculated that this is due to structural or functional constraints, or both. One interesting possibility would be that TcSMUG L products are involved in the attachment to the luminal midgut surface of the vector, as shown previously for GIPLs . Since TcSMUG L expression is turned off in the metacyclic trypomastigotes, which are detached from the midgut surface, a developmental regulation mediated by TcSMUG L might be proposed.
Carlos Buscaglia conceived and designed the experiments. Ivana Urban, Lucía Boiani Santurio, Agustina Chidichimo, Juan Mucci and Carlos Buscaglia performed the experiments. Ivana Urban, Fernán Agüero, Juan Mucci and Carlos Buscaglia analysed the data. Hai Yu and Xi Chen contributed reagents, materials and analysis tools. Fernán Agüero and Carlos Buscaglia wrote the paper.
This investigation received financial support from the UNICEF (United Nations Children's Fund)/UNDP (United Nations Development Programme)/World Bank/WHO (World Health Organization) Special Programme for Research and Training in Tropical Diseases (TDR), ANPCyT (Agencia Nacional de Promoción Científica y Tecnológica), Fundación Florencio Fiorini and UNSAM (Universidad Nacional De San Martín) (to C.A.B.). J.M., F.A. and C.A.B. are career investigators from the Argentinean Research Council (CONICET).
We thank Dr Patricio Diosque (Universidad Nacional de Salta, Argentina) for T. cruzi stocks, Leonardo Panunzi (IIB-INTECH) for molecular typing of T. cruzi stocks, Dr Juan José Cazzulo, Dr Alberto C. Frasch, Dr Oscar Campetella and Dr Daniel Sánchez (IIB-INTECH) for reagents, and Dr Nobuko Yoshida (UNSP, Brazil) for providing 10D8 and 2B10 mAbs. We also thank Liliana Sferco and Berta Franke de Cazzulo for culturing parasites and Fabio Fraga for animal care. Critical reading of the manuscript by Dr Igor C. Almeida (UTEP, TX, USA) prior to acceptance is appreciated.
Abbreviations: Ab, antibody; ConA, concanavalin A; Ct, cycle threshold; DTT, dithiothreitol; Gp35/50, mucins, mucins coating T. cruzi metacyclic trypomastigote and epimastigote forms; GIPL, glycoinositolphospholipid; GPI, glycosylphosphatidylinositol; GST, glutathione transferase; IFA, immunofluorescence assay; mAb, monoclonal antibody; PDhα, pyruvate dehydrogenase α; PI-PLC, phosphoinositol-specific phospholipase C; qPCR, quantitative PCR; Tos-Lys-CH2Cl, tosyl-lysylchloromethane; TS, trans-sialidase; UTR, untranslated region
- © The Authors Journal compilation © 2011 Biochemical Society