Glycosylation of Trypanosoma cruzi TcI antigen reveals recognition by chagasic sera

Chagas disease is considered the most important parasitic disease in Latin America. The protozoan agent, Trypanosoma cruzi, comprises six genetic lineages, TcI-TcVI. Genotyping to link lineage(s) to severity of cardiomyopathy and gastrointestinal pathology is impeded by the sequestration and replication of T. cruzi in host tissues. We describe serology specific for TcI, the predominant lineage north of the Amazon, based on expression of recombinant trypomastigote small surface antigen (gTSSA-I) in the eukaryote Leishmania tarentolae, to allow realistic glycosylation and structure of the antigen. Sera from TcI-endemic regions recognised gTSSA-I (74/146; 50.7%), with no cross reaction with common components of gTSSA-II/V/VI recombinant antigen. Antigenicity was abolished by chemical (periodate) oxidation of gTSSA-I glycosylation but retained after heat-denaturation of conformation. Conversely, non-specific recognition of gTSSA-I by non-endemic malaria sera was abolished by heat-denaturation. TcI-specific serology facilitates investigation between lineage and diverse clinical presentations. Glycosylation cannot be ignored in the search for immunogenic antigens.

www.nature.com/scientificreports/ TcIV a secondary cause in Venezuela 11 . TcII, TcV and TcVI are prevalent among cases in the Southern Cone countries of South America (Argentina, Bolivia, Brazil, Chile, Paraguay and Uruguay); TcIII is uncommonly found in human infections 1 .
In 1981 12 it was proposed that the different geographical distributions of the T. cruzi lineages may contribute to the disparate clinical presentations of Chagas disease in the Southern Cone countries, where megasyndromes are found, compared to northern South America, where they are not reported 1 . However, it is complex to prove such an association by parasite genotyping, because T. cruzi blood parasitaemia is scanty in chronic Chagas disease, does not necessarily represent lineages sequestered in the internal organs [13][14][15][16] , and growth rate competition occurs between isolates grown in vitro.
One approach to surveillance of clinical, geographical and ecological distributions of the T. cruzi lineages is to develop lineage-specific serology, originally proposed by Di Noia et al. 17 . Specific epitopes of the T. cruzi trypomastigote small surface antigen (TSSA), a cell surface mucin, have been identified for all six genetic lineages, with the hybrid lineages TcV and TcVI having two epitopes encoded at the heterozygous locus, one of which is shared with TcII, as shown by Bhattacharyya et al. 18 . Lineage-specific serology with synthetic peptides representing the TcII/V/VI and TcV/VI epitopes enabled surveillance of chagasic patients 19 , and the discovery of reservoir hosts 20,21 . Furthermore, TcII/V/VI serology, adaptable to rapid diagnostic test (RDT) format, demonstrated that among Bolivian patients stratified by severity of cardiomyopathy, TcII/V/VI seropositives were five-fold more prevalent in the severe versus no evidence of cardiomyopathy groups 22 . RDTs also identified TcII/V/VI seropositive sympatric humans and dogs in the Argentine Chaco 23 .
A long-standing research objective is the validation of a robust and sensitive TcI-specific antigen that would enable the enigma of link between infective lineage and clinical prognosis to be more comprehensively investigated. Furthermore, this would enable systematic low-cost analysis of T. cruzi transmission cycles and evaluation of the risk of emergence of sylvatic lineages into the domestic environment. However, repeated attempts have failed to develop a lineage-specific serological test for the TcI specific epitope, either using an E. coli-expressed recombinant protein or a synthetic peptide 19,[24][25][26] .
Here, we have expressed the TSSA-I epitope within a related trypanosomatid, Leishmania tarentolae, which enables O-linked and N-linked glycosylation 27,28 to determine whether glycosylation and/or structural integrity impart serological recognition of this TcI antigen.

Methods
Ethics. All human sera used here were archived, with consent for research, were anonymised, coded, and did not reveal patient identities. Informed consent was obtained from all subjects or, if subjects are under 18, consent was provided by a parent and/or legal guardian. No samples were collected specifically for this work.
Colombian (Bogotá), Venezuelan and Ecuadorean samples: these were collected as part of routine diagnostic examination, with local institutional ethical approvals Universidad de los Andes, Bogotá, Colombia; ( Gambian malaria sera were provided, with consent for further research on diagnostics, from London School of Hygiene and Tropical Medicine archives.
Non-endemic control sera were provided at the London School of Hygiene and Tropical Medicine and used with consent for further research on diagnostics.

Assessing serological recognition of synthetic peptides by ELISA. Colombian (Medellín) sera
were assayed by ELISA with synthetic peptides TSSApep-I, -II/V/VI, -III, -IV, -V/VI, according to protocols described previously 19 . Colombian (Bogotá), Ecuadorean and Venezuelan sera were previously assayed with synthetic peptides TSSApep-I, -II/V/VI, -III, -IV, -V/VI by ELISA; all were negative with TSSApep-I 19 ; Peruvian and Bolivian sera were previously assayed by TSSApep-II/V/VI RDT 22 . Prediction of glycosylation sites within gTSSA-I recombinant. In order to determine the presence of O and N glycosylation sites on the TSSA-I epitope, the amino acid sequence was submitted to NetOGlyc 4.0 (www.cbs.dtu.dk/services/NetOGlyc/) and NetNGlyc 1.0 (www.cbs.dtu.dk/servi ces/NetNG lyc/) online servers. Prediction included coverage of the SUMO component of gTSSA-I. L. tarentolae production of recombinant antigens gTSSA-I and gTSSA-II/V/VI. The TSSAspecific sequences were cloned into separate expression plasmids pLEXSY_I-blecherry3 with an upstream His tag and SUMO fusion partner to aid solubility of the resulting recombinant proteins, hereafter called gTSSA-I or gTSSA-II/V/VI, expressed in the L. tarentolae system (Jena Biosciences, Germany). Figure 1a,b depict the sequences of these recombinant antigens, with N-terminal histidine tag (blue), SUMO sequence (green) and TSSA-I sequence (red).
Production of recombinant protein, based on the methodology of Rooney et al. 29 was carried out in 1 L baffled Erlenmeyer flasks in BHI medium (supplemented with antibiotics and hemin) and the medium was harvested when the OD 600 reached 4 (approx. 70 h post inoculation, 10 8 cells/ml). All media components were from Jena Bioscience. Clarified medium was concentrated 20-fold on a Pellicon XL 50 Ultrafiltration cassette (10 kDa MWCO) and diluted four times in binding buffer (20 mM Phosphate, 500 mM NaCl, 10 mM Imidazole) before addition to an equilibrated His Trap (GE Healthcare) FPLC column. Bound proteins were eluted using an increasing Imidazole gradient (20 mM Phosphate, 500 mM NaCl, 500 mM Imidazole). Peak protein containing fractions, as determined by A 280 nm measurement, were combined, desalted and concentrated by centrifugation in Amicon Ultra-15 device (5 kDa MWCO). Purified recombinant gTSSA-I was run on Coomassie blue-stained pre-cast 4-12% BisTris gradient SDS-PAGE gel (Novex) using the MOPS running system, along with Precision Plus Protein Unstained Standards (BioRad). The final protein was stored at 1 mg/ml in solution in PBS, 15% glycerol at -20 °C.
Glycoproteomics. For glycoproteomics 20 µg of the glycoprotein with the gTSSA-I epitope were denaturated in 10 µL 8 M guanidine hydrochloride solution and reduced with 1 µL of 10 mM DTT (Thermo Fisher Scientific) in 50 mM ammonium bicarbonate buffer (pH 8.4) for 30 min at 56 °C. 1 µL of 55 mM iodoacetamide (Thermo Fisher Scientific) was added and incubated at room temperature for 30 min in the dark. Finally, the digest was diluted with 30 µl of 50 mM ammonium bicarbonate buffer prior to adding 1 µL of trypsin (1:20 w/w enzyme:protein ratio).
The online ESI-LC-MS data were recorded in MSE mode for 60 min using a Waters Synapt G2 mass spectrometer (Waters, Milford, MA). Separations were achieved on C18 Acquity UPLC M-Class column (HSS T3 1.  www.nature.com/scientificreports/ performed manually by previously reported methods based on amino acid and sugar masses together with the known fragmentation of glycopeptides 30 , with the assistance of MassLynxV4.1. (Waters, Milford, MA).
Assessing serological efficacy of the recombinant antigen gTSSA-I. Separate wells of a 96-well flat bottomed ELISA plates (735-0465: Immulon 4HBX, VWR) were coated with 50 μl/well of 1 µg/ml L. tarentolae-expressed gTSSA-I, and with 100 μl/well of 2 µg/ml lysate of T. cruzi TcII strains IINF/PY/00/Chaco23 or MHOM/BR/00/Y in coating buffer (15 mM Na 2 CO 3 , 34 mM NaHCO 3 , pH 9.6). Lysates were prepared as described previousy 19  Assessing the contribution of glycosylation and secondary structure to antigenicity of gTSSA-I. Contribution of glycosylation to antigenicity was assessed using an assay based on the protocol of Woodward et al. 31 , and employed subsequently 32,33 , which described the oxidative cleavage of carbohydrate vicinal -OH groups and subsequent reduction of generated aldehyde groups to prevent non-specific antibody binding. After the blocking and washing steps of the ELISA, described above, all wells were rinsed in periodate buffer (50 mM sodium acetate buffer, pH 4.5), and the wells that had been coated with gTSSA-I received 5 mM freshly-made sodium (meta)periodate (71859: Sigma Aldrich) in periodate buffer; the remaining wells received periodate buffer only. Plates were incubated in the dark at room temperature for 1 h, followed by rinsing of all wells with periodate buffer. The periodate-treated wells were then reduced with freshly-made 50 mM sodium borohydride (71320: Sigma Aldrich) in PBS for 30 min; the remaining wells received PBS only. Following this step, wells were washed three times with PBST before addition of sera and subsequent processing, as described above.
To investigate the contribution of secondary structure to the antigenicity of gTSSA-I, an aliquot of gTSSA-I in coating buffer was heated > 95 °C for up to 10 min, prior to coating onto the plate and performance of the serological efficacy was assessed as described above.
Statistical analysis. Replica ELISA plates were run in duplicate simultaneously. Cut-off values were determined by first subtracting the background absorbance values (i.e., mean of wells with coating buffer only, no antigen) from the mean reading for each sample; samples that were then greater than three standard deviations above the mean of seronegative non-endemic controls were considered positive. P values were determined by performing two-sample T test either unpaired (gTSSA-I against gTSSA-II/V/VI) or paired (unmodified gTSSA-I against oxidised or denatured) with GraphPad Prism (GraphPad Software, San Diego, USA).

Results
Synthetic peptides were not recognised by TcI endemic chagasic sera. All Colombian (Medellín and Bogotá), Ecuadorean, Venezuelan, and northern Peruvian sera samples (n = 146) were from regions considered to be endemic for TcI principally 11 , and were predominantly from chronic cases of Chagas disease in rural locations 1,34 . All sera were seropositive with T. cruzi lysate; (TcII and TcI lysate antigens do not discriminative between lineage infections). Apart from Colombia (Medellín) and Peruvian samples, all had been previously assayed with synthetic peptide TSSApep-I, and no reaction had been identified 19 . Here, there was no TSSApep-I recognition by the Colombian sera (Medellín) and no indication of the presence of infection with any of the other synthetic peptides.

Absence of SUMO cross-reactivity between gTSSA-I and gTSSA-II/V/VI. A subset of Colombian
(Medellín) sera positive for gTSSA-I (n = 34) were assayed by ELISA against gTSSA-II/V/VI, to assess levels of potential cross-reactivity to the SUMO component of gTSSA-I. No recognition of gTSSA-II/V/VI was observed with these sera (Fig. 2a). Grouping these data, there was a significant difference in the absorbance values between these two recombinant antigens (P < 0.0001) (Fig. 2b). Thus, there was no cross reactivity between these two recombinant antigens attributable to antibody recognition of the SUMO component. Bolivian sera shown previously to be reactive with synthetic peptide TSSApep-II/V/VI (n = 5) reacted with gTSSA-II/V/VI; two sera also recognising gTSSA-I (Fig. 2c), suggesting TcII/V/VI and TcI co-infection.

gTSSA-I antigenicity is principally dependent on glycosylation, not structure. A subset of
Colombian (Medellín) samples that recognised gTSSA-I (n = 12) were assayed against periodate-treated, heat denatured and unmodified gTSSA-I on the same ELISA plates. Recognition of gTSSA-I decreased substantially after periodate treatment, compared to the heat denatured antigen, for all samples except sample 2 (Fig. 3a).
The absorbance values were significantly lower (P = 0.0002) after periodate treatment, with only a single sample retaining the same level of recognition as with the unmodified antigen. After heat denaturation the absorbance values of all 12 samples were reduced to a much lesser extent, albeit with a significant difference (P < 0.0001) (Fig. 3a).
gTSSA-I recognition by some Gambian malaria sera is dependent on structure. Unexpectedly, recognition of gTSSA-I was observed with 13/24 Gambian malaria sera samples. Eight of these reactive samples were assayed against periodate-treated (oxidised) and heat denatured gTSSA-I. Oxidation had little effect on recognition of gTSSA-I, with no significant difference between absorbance values (P = 0.5732) and recognition of gTSSA-I was retained by all 8 samples. However, this was abolished by heat denaturation of this antigen (P = 0.0089). One sample remained above the cut-off, though the absorbance value had been greatly reduced (Fig. 3b,c). Any recognition of gTSSA-I by non-endemic healthy controls (NEHC) was also abrogated by heatdenaturation.

Discussion
Lineage-specific serology for T. cruzi infections provides a powerful tool for understanding the epidemiology and ecology of Chagas disease. Here, we have investigated lineage-specific epitopes to determine the importance of glycosylation in antigen recognition, particularly in relation to TcI. The L. tarentolae expression system has been used previously to produce Trypanosoma brucei gambiense surface antigens, as a low cost alternative to harvesting diagnostic antigens from T. b. gambiense grown in vitro 29 . Expression in L. tarentolae has also been applied in attempts to improve diagnostic antigens for the trypanosomatids Leishmania braziliensis 35 and Leishmania donovani 36 .
Here, we applied L. tarentolae expression to produce recombinant T. cruzi gTSSA-I, and to determine whether consequent glycosylation or more bona fide structural conformation of the protein conferred serological recognition upon the antigen.
All Colombian, Venezuelan, and Ecuadorean sera that were positive here with gTSSA-I were previously seronegative with the synthetic peptide TSSApep-I. The Medellín samples showed by far the highest proportion that recognised gTSSA-I; 85.5%. Reasons for this are unclear, but may be related to lower levels of anti-T. cruzi IgG in the other sera; in some localities there are low IgG antibody levels in T. cruzi infections 37 . Furthermore, the gTSSA-I positive samples were not seropositive with gTSSA-II/V/VI, demonstrating that there was no serological cross reaction with the SUMO component of the recombinant antigens. Thus, we have demonstrated www.nature.com/scientificreports/ clear, robust TcI lineage-specific serology with sera originating from countries where TcI has been identified by genotyping as the principal cause of Chagas disease.
In comparison with synthetic peptides or production of recombinants in the bacterium E. coli, heterologous expression in the eukaryote L. tarentolae enables glycosylation of the trypanosomatid proteins and adoption of a more natural conformation. O-and N-linked glycosylation of recombinant proteins in L. tarentolae has been described, and the wild type pattern in T. cruzi also shown to be O-linked (GlcNAc) 28,38 . Assays with periodatetreated antigen showed that the protein glycosylation (predicted by the online algorithms and demonstrated by glycoproteomics), and not the protein conformation, was important for serological recognition of gTSSA-I. Thus, expression in L. tarentolae revealed antigenic properties of TSSA-I that are not evident in synthetic peptides or E. coli-expressed recombinant proteins.
The specificity of gTSSA-I was questioned, due to some cross-reactivity with non-endemic (Gambian) malaria control sera. The sera from malaria were included due to breadth of immune response associated with antigenic variation, and recent research interest in whether antibodies to Plasmodium may recognise glycosylated antigens 39 . Two of these controls were also seropositive with the lateral flow diagnostic test specifically to detect exposure to T. b. gambiense infection (data not shown), suggesting an explanation for their cross reactivity. However, heating gTSSA-I abolished recognition by all Gambian malaria sera, demonstrating that this was due to analogous protein structure of gTSSA-I, not glycosylation. Since this is contrary to gTSSA-I recognition by chagasic sera, heating of gTSSA-I prior to performance of diagnostic ELISA or to incorporation into RDTs provides TcI-specific diagnosis via recognition of glycosylation antigenicity.
We have previously demonstrated a link between serological recognition of TSSApep-II/V/VI and severity of chagasic cardiac symptoms 19,22 . Availability of gTSSA-I serology enables parallel investigation of clinical status associated with TcI infection, and serological detection of sporadic TcII/V/VI and TcI co-infections, which occur in some Bolivian and Brazilian endemic foci 40,41 . As with application of TSSA-II/V/VI serology to sylvatic Overall absorbance values for Colombian (Medellín) sera against gTSSA-I (blue) and gTSSA-II/V/VI (orange); gTSSA-I is recognised, whereas gTSSA-II/V/VI is not (P < 0.0001). (c) ELISA plate illustrating the recognition of gTSSA-I but not of gTSSA-II/V/VI; all samples were seropositive with lysate, and coating buffer controls were negative. Sample numbers correspond with (a). Positive control: serum from a Bolivian patient previously shown to be reactive with synthetic peptide TSSApep-II/V/VI and also seropositive with gTSSA-I, indicating co-infection (see text). www.nature.com/scientificreports/