In silico identification and characterization of a multigenic family of sialidases in T. vivax
Trypanosoma vivax genome (Y486 strain) is partly sequenced and available on the Genome Data Base TritrypDB (http://tritrypdb.org/tritrypdb/). We found seven genes encoding proteins displaying typical features of trypanosomal SA (Fig. 1, Fig. S1, Table S1) and annotated as putative T. vivax trans-sialidases. Of these genes, five which we named TvTS1–5 encode proteins of 849–898 amino acids (aa) which is similar to the size of other described trypanosomal SA and two encode shorter products (522 and 368 aa) (Fig. 1A). The two shorter forms have high identity with TvTS1 and TvTS5 but are likely to be the result of incomplete sequencing since the sequences contain stretches of undetermined nucleotides and frameshifts.
Figure 1. Multigenic trans-sialidases family of T. vivax.
A. In silico analysis of TvTS family. Protein domains were defined with InterProscan (IPR011040 neuraminidase, IPR013320 concanavalinA-like lectin, IPR008377 trypanosome SA) and ScanProsite (PS00082 extradiol ring-cleavage dioxygenases signature, PS51257 prokaryotic membrane lipoprotein lipid attachment site). The following softwares were used for prediction: SignalP 3.0 for signal peptide, TargetP 1.1 for subcellular localization (SP, secretory pathway; mTP, mitochondrial targeting peptide; Nd, not determined), GPI-SOM and PredGPI for GPI anchoring signal, DAS, HMMTOP and TopPred for transmembrane domains, CSS-Palm 2.0 for palmitoylation sites.
B. Comparison of TvTS1–5 amino acid (aa) sequences. Sequences were aligned with clustalw. The N-terminal catalytic domain and the C-terminal lectin-like domain are separated by a dotted line. FRIP box (double black line) and Asp boxes (single black line) are represented. The motifs VVDPTVVAK (A), ISRVIGNS (B), VPVMLITHP (C), LYCLHE (D), VTVxNYXLYNR (E) are underlined. Residues critical for enzyme function are marked with arrowheads and listed in Fig. 1.
C. Comparison of active sites in trypanosomal SA/TS. TbSA and TbTS: T. brucei SA and TS; TcTS: T. cruzi TS; TrSA: T. rangeli SA; TcoTS T. congolense TS. TcTS, TrSA, TbSA and TbTS analyses are based on previous studies (Tiralongo et al., 2003; Montagna et al., 2006). TvTS2 aa common to other TvTS are shown in black and differences in grey. Residues are illustrated in Fig. 1. Sia: sialic acids, SA: sialidase activity, TS: trans-sialidase activity.
See also Fig. S1 and Table S1.
Download figure to PowerPoint
Since TvTS1–5 contained all the SA motifs including the FRIP and Asp boxes and the residues critical for catalytic function, they were likely to encode active enzymes (Roggentin et al., 1993; Tiralongo et al., 2003) (Fig. 1 and C). The transferring abilities of trypanosomal TS also appeared to be conserved (Buschiazzo et al., 2002; Tiralongo et al., 2003; Montagna et al., 2006) (Fig. 1C). TvTS1–5 enzymes exhibited relatively high identity to each other: 50.9–66.3% (Table S1). Similar to T. brucei and T. congolense families, conservation was higher in the catalytic domain (63–90.1%) than in the lectin-like domain (42.8–50.3%) (Montagna et al., 2006; Coustou et al., 2012). Nevertheless, slight differences of crucial residues were evident. For example, the variability of residue 475 of TvTS2, important for the binding specificity, could reflect enzymatic diversity particularly concerning substrate specificity (Fig. 1C).
TvTS enzymes also showed 16.9–31.2% identity to the 8 SA/TS of T. brucei, 14.1–31.9% to the 11 SA/TS of T. congolense, and 25.5–31.9% to a representative T. cruzi TS (Table S1). But contrary to T. brucei and T. congolense, the degenerate TS-like subfamily was not detected in T. vivax. Furthermore, phylogenetic analysis showed that TvTS formed a separate cluster whereas TcoTS, TbTS and TbSA were closer to their orthologues than to members of their own family (Fig. S1).
In silico characterization of TvTS enzymes indicated other differences with trypanosomal SA/TS characterized thus far: (i) none of these proteins are predicted to be GPI-anchored nor possesses a trans-membrane domain, (ii) three of the TvTS1–5 proteins displayed a mitochondrial signalling peptide and only TvTS2 is clearly predicted to follow the secretory pathway, and (iii) TvTS1 and TvTS3 possess additional motifs, i.e. an extradiol ring-cleavage dioxygenase degenerate signature (some critical residues for dioxygenase activity are not conserved) and a prokaryotic membrane lipoprotein lipid attachment site profile respectively. In fact, most of the trypanosomal SA described so far were associated with cell membranes and/or secreted, here some of the TvTS enzymes could be mitochondrial signa-anchored proteins therefore associated with the outer membrane of the mitochondria. For now, the only known example of mitochondrial SA is the long isoform of human Neu4 SA predominantly expressed in the brain (Yamaguchi et al., 2005) and also associated with the outer membrane of the mitochondria. The biological significance of these differences is not obvious but suggest that T. vivax TS play additional or different roles during infections.
Detection and characterization of SA and TS activities and identification of TvTS enzymes expressed in T. vivax BSF
To investigate SA and TS activities presence in the mammalian stage of T. vivax (BSF), we used fluorimetric assays to quantify enzyme-mediated hydrolysis of sialic acid (SA assay) and transfer of sialic acid from a donor, α2,3-sialyllactose, to an acceptor, MU-β-d-galactopyranoside (TS assay). We previously characterized these activities in T. congolense IL3000 and demonstrated their absence in T. brucei 427 BSF (Coustou et al., 2012). Here, we compared SA and TS activities in crude BSF extracts of two T. congolense strains, two T. brucei strains and T. vivax Y486 (Fig. 2 and B). The results clearly demonstrated the presence of both activities in T. vivax BSF. While SA activity has previously been reported in literature, TS activity has never been described (Esievo et al., 1982; Buratai et al., 2006). The level of SA activity was comparable in T. congolense and T. vivax (∼ 12 μU per 109 cells versus ∼ 16) whereas TS activity was approximately four times higher in T. vivax (∼ 620 μU per 109 cells versus ∼ 2143). T. brucei displayed neither of these activities.
Figure 2. SA and TS activities in T. vivax Y486 BSF.
A and B. SA (A) and TS (B) activities were measured on crude extracts of T. brucei 427 and Antat 1.1, T. congolense IL3000 and STIB910 and T. vivax Y486 BSF. BSF were cultured in vitro or purified from mouse blood, results were the same regardless of origin. Each bar represents the SA or TS activity ± standard deviation (SD) of at least three independent experiments.
C. The proportion of TS activity, expressed as a percentage of total activity, was measured in the supernatant (soluble) or pellet (insoluble), following centrifugation of osmotically lysed cells (T. vivax Y486 BSF).
D. SA and TS activity in T. vivax Y486 BSF culture was monitored over 2 days. BSF were obtained from infected mice and cultivated for 2 days; during this short period cell death is minimal. Data are expressed as the mean of two independent experiments.
See also Figs S2 and S3.
Download figure to PowerPoint
Although T. brucei and T. congolense insect forms (procyclic, PCF and epimastigote, EMF) displayed very high SA/TS activities (Engstler et al., 1993; 1995; Coustou et al., 2012), neither of these activities were detected in T. vivax EMF. This could be explained by the fact that T. vivax BSF ingested during fly blood meal differentiate into EMF in the proboscis, pharynx and oesophagus of the fly (Hoare, 1972), whereas T. congolense and T. brucei differentiate into PCF which travel to the fly midgut. Therefore, protection from the digestive enzymes by trans-sialylation (Nagamune et al., 2004) may not be necessary for T. vivax EMF.
The TS/SA activities ratio suggested that TvTS enzymes expressed in BSF were predominantly TS and not strict hydrolases. Generally, TS exhibit low SA activity rather than high TS activity due to the catalytic properties of these enzymes (Demir and Roitberg, 2009). Furthermore we saw above that critical residues for transferring activity are conserved in TvTS1–5. The same situation is observed in T. congolense and could illustrate a common strategy for the virulence of these two parasites species.
To further characterize these activities, we determined the proportion of TS activity associated with insoluble material after hypotonic lysis in T. vivax BSF (Fig. 2C). We observed, as previously described for T. congolense (Coustou et al., 2012), that the major part of the activity was membrane-associated. We also monitored the culture medium of T. vivax BSF for released SA and TS activity and observed that significant activity accumulated in the medium during growth, a property also described for T. congolense (Coustou et al., 2012). Taken together, these data showed that part of the activity remained membrane-bound while part is secreted which correlates quite well with the in silico prediction as some TvTS enzymes were predicted to be secreted while others seem to be anchored in the outer membrane of the mitochondria.
Then to determine the expression pattern of TvTS enzymes, we produced recombinant proteins in heterologous systems to raise anti-TvTS-sera. Only the soluble TvTS2 was successfully produced in the yeast Pichia pastoris, while the others were produced as insoluble material in Escherichia coli. TvTS2 was active and its biochemical properties were analysed. TvTS2 hydrolysed MU-Neu5Ac, 3′sialyl-lactose, ganglioside GD3 or fetuin and was also able to transfer sialic acid to an acceptor, indicating TS activity. The basic kinetic parameters of TvTS2 were determined (Fig. S2). The results were very similar to those obtained from T. brucei or T. congolense TS (Tiralongo et al., 2003; Montagna et al., 2006; Koliwer-Brandl et al., 2011; Coustou et al., 2012).
Recombinant proteins were used to immunize rabbits or mice. A Western blot analysis with anti-sera gave a positive signal on T. vivax BSF crude extracts around 100 kDa, the expected size of TvTS enzymes (Fig. S3). But no significant signal was detected in EMF (data not shown). Immunofluorescence analyses were also realized and different kind of signals were obtained (Fig. 2E). Anti-TvTS2 serum showed typical mitochondrial labelling which was corroborated by colocalization with a mitochondrial marker, the heat-shock protein HSP60 (Bringaud et al., 1995), and also a less intense labelling of the whole cell, typical for cytosolic compartment. Anti-TvTS3 serum showed labelling of the whole cell surface with a dotted line along the cell body, which is typical for membrane protein (Field et al., 2010). Again these data are correlated with in silico prediction and also with activity as part of it is soluble and could be in the cytosol while another part is secreted or membrane-bound and could be either in the mitochondrial or in cell surface membrane.
The anti-sera were also used in immunoprecipitation assays on T. vivax BSF and EMF followed by mass spectrometry analysis but no TvTS peptides were obtained. As TS activity is mostly associated with insoluble material, membrane extracts of T. vivax Y486 EMF and BSF were analysed in mass spectrometry. Peptides from TvTS1, 3, 4 and 5 were detected in BSF (Table S2). The presence of TvTS1, 3 and 5-specific peptides allowed us to affirm that at least these three TvTS enzymes were expressed in T. vivax Y486 BSF. No TvTS peptides were identified in EMF. We also analysed the culture medium of T. vivax Y486 BSF as we showed that TS activity accumulated and we confirmed TvTS presence by Western blot analysis (data not shown). However, no TvTS peptides were detected by mass spectrometry analysis which could be explained by a quantity of TvTS enzymes too low to be detected or presence of other abundant proteins at this molecular weight masking TvTS (similar results were obtained with T. congolense).
In conclusion, we identified, for the first time, at least three TvTS family members expressed in T. vivax Y486 BSF, which are likely to be responsible for the SA/TS activities in the infected mammalian hosts. Furthermore, we confirmed that the absence of SA/TS activities in EMF insect stage is due to the absence of TvTS proteins.
SA and TS activities are present in the sera of T. vivax-infected mice
To investigate the potential erythrophagocytosis process, we defined a suitable experimental trypanosomiasis murine model. We chose CD1 mice where T. vivax infection is more subchronic than acute, which allowed a more thorough study of the development of anaemia and, almost significantly, closely resembles natural T. vivax infection in cattle (Chamond et al., 2010). We also defined the parasite inoculum and chose to inject 2·106 parasites of the rodent-adapted Y486 strain intraperitoneally. We obtained a subchronic infection profile with waves of parasitaemia and significant development of anaemia (Fig. 3A). Mice survival was ∼ 2–4 weeks. This type of infection mimicked the initial phase of T. vivax or T. congolense bovine infection with a rapid drop of PCV (up to ∼ 20%) and a high parasite load (up to ∼ 1·109 parasites per ml of blood). We monitored the reticulocyte level (erythropoietic precursor of erythrocytes) and observed a high level of reticulocyte production (up to nine times the basal level, ∼ 18% versus ∼ 2% of the RBC) in response to anaemia (Fig. 3A). This reticulogenesis followed the waves of parasitaemia and anaemia and clearly showed that haematopoiesis is not impaired during infection. This physiological compensation of anaemia had to be taken into account since the intact, freshly produced reticulocytes were included in the measures of RBC sialylation state at some points of the experiment.
Figure 3. Analysis of parasitaemia, anaemia and SA activity during experimental T. vivax infection in mice.
A. Parasitaemia, haematocrit and reticulocytes were monitored during T. vivax Y486 infection in CD1 mice. Haematocrit was expressed in PCV and reticulocytes as the percentage of RBC.
B. SA activity was monitored during infection and plotted with haematocrit and parasitaemia.
Two representative mice out of 10 are shown.
Download figure to PowerPoint
To determine the potential role of TvTS in anaemia, we also determined the level of SA activity in the serum and observed a significant SA activity that fluctuated with parasitaemia (Fig. 3B). Monitoring of TS activity gave identical profiles (data not shown). SA/TS activities probably resulted from active secretion with a correlation with parasite load in the blood but also from passive release after immune-mediated lysis which would explain the relatively high SA activity even in the absence of parasites between each wave of parasitaemia. Release of virulence factors after immune-mediated trypanosomes lysis is a well-known phenomenon (Antoine-Moussiaux et al., 2009). Moreover, these enzymes seem to be quite stable and could accumulate, the observed profiles have, therefore, resulted from a combination of all these infection traits. These results are comparable with T. congolense mouse infection (Coustou et al., 2012), and could again highlight a common mechanism of anaemia development.
Development of an ex vivo erythrophagocytosis assay
Bratosin et al. developed an elegant flow cytofluorimetric procedure using fluorescent erythrocytes to measure erythrophagocytosis of senescent RBC (Bratosin et al., 2002). This method is quantitative, sensitive and avoids the use of radioisotopes. To study the propensity of erythrocytes to be recognized and phagocytosed by host cells during T. vivax murine infection, we adopted a similar method to develop an ex vivo erythrophagocytosis assay. This assay is based on the use of erythrocytes, macrophages and serum from the same mouse to evaluate erythrophagocytosis by flow cytometry analysis. We improved erythrocyte labelling by using calcein AM (cell-permeant, converted by intracellular esterases into green-fluorescent calcein) instead of PKH-26 (irreversibly anchored into cell plasma membrane even in damaged cells) to decrease the background.
The assay procedure starts with collection of peritoneal macrophages from T. vivax Y486-infected mice or non-infected control mice that were put into culture plates to adhere (1–5·105 per well). At the same time, blood was collected from these mice, erythrocytes and serum were separated and erythrocytes were labelled with calcein AM. After 3 h, non-attached macrophages were eliminated, and erythrocytes (3·106) and mouse serum (10%) were added to the wells. Eight combinations of biological material were obtained depending on the source (infected, Y, or control, C) of the three components: macrophages, serum and erythrocytes (first, second and third letter of the combination respectively). After 4 and 24 h, erythrophagocytosis was measured by flow cytometry (Fig. 4). To differentiate between macrophages and residual external erythrocytes, macrophages were labelled with anti-CD16/32 (Fig. 4B), and the proportion of macrophages containing phagocytosed erythrocytes (calcein labelled) was determined (Fig. 4 and D).
Figure 4. Flow cytometry analysis of an ex vivo erythrophagocytosis assay.
A–D. Analytical method of flow cytometry data from an ex vivo erythrophagocytosis assay (CCY combination).
A. Dot plot representation of the mixed erythrocyte and macrophage populations. Delimitation of the two populations is shown and the macrophage population is marked as P1.
B. APC fluorescence histogram of population P1. APC is conjugated to anti-CD16/32 antibodies (macrophage antigen marker). The positive population is marked as P2.
C. FITC fluorescence histogram of population P2. FITC fluorescence reflects calcein-labelled erythrocytes. Positive population (P3) represents macrophages containing phagocytosed erythrocytes.
D. Dot plot representation of population P2. The negative APC population is not plotted. The proportions of APC-positive, FITC-negative macrophages (lower right quadrant) and APC-positive FITC-positive macrophages (upper right quadrant = P3) were determined.
E. Results of an ex vivo erythrophagocytosis assay. In infected mice, parasitaemia reached its maximum (5·108 parasites per ml) and haematocrit began to drop (35%).
Download figure to PowerPoint
An example was reported in Fig. 4 with a Y486-infected CD1 mouse 7 days after infection (top of the first peak of parasitaemia). The proportion of macrophages containing phagocytosed erythrocytes was dependent on the combination of the three serum components. The base levels of erythrophagocytosis were ∼ 2% (4 h of incubation) and ∼ 11% (24 h) under control conditions (CCC) and the highest level were ∼ 10% (3 h) and 47% (24 h) when all the components came from infected mice (YYY). To compare the results, the ratio of erythrophagocytosis between one combination and the base level (CCC) was the most relevant parameter i.e. erythrophagocytosis increased around five times from CCC compared with YYY. In this example, the ratio of erythrophagocytosis was high when erythrocytes from infected mice were used. This illustrates an increased sensitivity of infected-mouse erythrocytes to phagocytosis.
In conclusion, with this method, we were able to quantify ex vivo erythrophagocytosis which allowed extrapolation of the propensity of erythrocytes to be phagocytosed in vivo. Moreover, we evaluated the influence of each of the three components of the biological system. We therefore conducted a complete analysis of the different combinations throughout infection.
Erythrophagocytosis is a major cause of anaemia during T. vivax infection
To understand the significance of erythrophagocytosis during T. vivax infection, we defined four critical stages in Y486 experimental infection and examined the rate of erythrophagocytosis in each stage (Fig. 5A). Stages were defined as follows: stage 1, increase of parasites in the first wave of parasitaemia associated with slight anaemia; stage 2, peak of the first wave of parasitaemia and the beginning of the decrease associated with strong anaemia, and therefore the most interesting stage; stage 3, absence of parasites in the blood associated with haematocrit recovery; stage 4, high level of parasites in the blood causing the second wave of parasitaemia associated with strong anaemia.
Figure 5. Erythrophagocytosis throughout T. vivax experimental infection in mice.
A. Typical parasitaemia and haematocrit profiles of T. vivax Y486 experimental infection. Four important and representative stages of the infection were defined to serve as references.
B. Results of erythrophagocytosis assays expressed as ratio of macrophages containing phagocytosed erythrocytes in the indicated condition over the macrophages containing phagocytosed erythrocytes in control condition (CCC). Mean values and SD are shown when at least three assays were performed (assay number is indicated by n). A ratio lower than 1 (marked as 1) indicates that erythrophagocytosis was not different from the basal level. Mice collected for assays were classified into infection stages depending on infection parameters.
C–F. Histograms of erythrophagocytosis ratios through T. vivax infection for a given combination.
Download figure to PowerPoint
The ratio of erythrophagocytosis (Fig. 5B) clearly showed that erythrophagocytosis is greatly enhanced in stage 2 and 4 when erythrocytes from infected mice were used (YYY, CCY, CYY and YCY), whereas it is only slightly enhanced or equal in other conditions. On detailed examination of the results, the YYY combination (Fig. 5C) showed the most significant ratio of erythrophagocytosis, ∼ 8 (4 h) and ∼ 5 (24 h) in stage 2, and ∼ 3 in stage 4. In stage 1 and 3 the ratio is ∼ 2. This indicated that high levels of erythrophagocytosis occurred in infected mice particularly at the beginning of anaemia, which suggested that erythrophagocytosis is a major trigger for the development of anaemia during T. vivax infection. To analyse the role of each of the components, we analysed the combinations CCY (Fig. 4D), YCC (Fig. 4E) and CYC (Fig. 4F) which allowed examination of the involvement of erythrocytes, macrophages and serum respectively. Significantly, the ratio of erythrophagocytosis in CCY conditions showed approximately the same profile as YYY conditions with a lower ratio increase in stage 2 (∼ 4) and 1 (∼ 1). This confirmed the importance of erythrocytes for the stimulation of phagocytosis since even in the presence of control macrophages and serum, phagocytosis is four times higher at a time of infection where the level of anaemia is high. There was only a slight increase in erythrophagocytosis when infected macrophages and/or serum are used especially in stage 2 and their effect appeared to be synergistic. This demonstrated that even if macrophages and serum elements such as antibodies participate and amplify the phenomenon of erythrophagocytosis, their contribution is not definitive compared with erythrocytes.
We showed that this ex vivo erythrophagocytosis assay gave significant and reproducible results. We observed that infection-associated erythrophagocytosis was visible throughout infection and varied with parasitaemia and also established a direct correlation between the presence of parasites in the blood, the development of anaemia and ex vivo erythrophagocytosis. We propose that this reflects on many aspects the anaemia process encountered in vivo which highlighted the existence of erythrocyte modifications during infection which triggered their phagocytosis by monocytes and macrophages.
Erythrocyte sialylation profiles are modified during T. vivax infection
We hypothesized that TvTS enzymes released during infection desialylate erythrocytes rendering them prone to phagocytosis. To test our hypothesis, we used different techniques to examine erythrocyte glycosylation and sialylation profiles during infection.
The majority of erythrocyte glycoproteins are sialoglycoproteins named glycophorins. They constitute more than 80% of sialic acids bound to erythrocyte membrane glycoconjugates (Furthmayr et al., 1975). In mice, two glycophorin isoforms, gp2 and gp3, were identified (Sarris and Palade, 1982). Glycophorins tend to form multimers in SDS-PAGE resulting in complex migration profiles: gp2 and gp3 monomers are observed around 44 and 29 kDa respectively, gp3 dimers named gp1 around 66 kDa and other multimers around 94 kDa (Sarris and Palade, 1982). Moreover, experimental conditions (protein concentration, buffer composition) can modify this profile. Therefore, we determined glycosylation profiles in control mice with defined experimental conditions to map glycophorin bands which are likely to be preferential targets for erythrocyte desialylation.
Erythrocyte ghosts were submitted to SDS-PAGE and gels were silver stained directly or after periodic acid oxidation to enhance staining sensitivity which is typically greatly reduced by glycosylations (Fig. 6A) (Van-Seuningen and Davril, 1992). We obtained a typical profile and defined five glycoproteic bands (A–E), we assumed that band C corresponds to gp2, bands D and/or E to gp3, band B to gp1 and bands > 90 kDa named A to glycophorin multimers. Subsequently, Western blotting analysis with antibodies commonly used to detect glycophorins were performed (Fig. 6A). TER-119, a monoclonal antibody which recognizes a molecule associated with glycophorins and allows detection of all the glycophorin bands (Kina et al., 2000), as expected, recognized the glycoproteic bands A to E. Monoclonal antibodies anti-glycophorin A (GlyA, gp2 orthologue) and anti-glycophorin C (GlyC, gp3 orthologue) recognized band C and bands B and E respectively. Finally, peptides matching glycophorins were abundant and preferentially found in the bands A to E by mass spectrometry analyses (data not shown). Subsequently, we examined modifications of this glycophorin profile during infection with several tools. The results are presented for a single mouse which developed a rapid infection with a substantial decrease of haematocrit and a survival of 16 days to facilitate comprehension (Fig. 6B). However, it must be noted that the same results were obtained for three mice which developed similar infection patterns and for mice with more chronic infections, where glycophorin profile modification correlated with the degree of PCV variation (data not shown). First, we analysed general changes of the glycoprotein profile and observed significant modifications especially for the bands B, D and E which clearly increased as seen on the gel on the right showing only glycoproteins (Fig. 6C). Modifications were also visible on a total protein gel by decrease in size and transition from blurry regions to distinct bands, which is typical for glycosylation modification (stars in Fig. 6C). The increase of bands D and E signifies that glycophorin migration as monomers was more important (change in the ratio between multimer/monomer migrating forms) maybe as a consequence of glycophorins desialylation known to facilitate monomers on SDS-PAGE (Sarris and Palade, 1982).
Figure 6. Modification profiles of the mouse erythrocyte surface glycoproteins throughout T. vivax infection.
A. Identification of CD1 mouse erythrocyte membrane glycoproteins. After SDS-PAGE gels were silver stained directly (1) or after periodic acid oxidation to also reveal glycoproteins (2) or were analysed by Western blotting with monoclonal TER-119 (TER119), anti-glycophorin A (GlyA) and anti-glycophorin C (GlyC) antibodies.
B. Monitoring of parasitaemia and haematocrit in the Y486 T. vivax-infected CD1 mouse used to analyse erythrocyte membranes profiles (C–F).
C. Comparison of erythrocyte membranes proteins (left) and glycoproteins (right) throughout T. vivax infection on SDS-PAGE. Gels were silver stained directly (left) or after brief periodic acid oxidation to reveal only the glycoproteins (right).
D. Lectin blot analysis of erythrocyte membrane glycoproteins throughout T. vivax infection. Blots were probed with biotin-conjugated Sambucus nigra agglutinin (SNA). Black arrows (1–6) indicate lectin signals, grey arrows (7–9) indicate sialylated proteins appearing on day 15 post infection. Some arrows were replicated on related blots.
E. Western blot analysis on erythrocytes ghosts with TER119 throughout Y486 T. vivax infection.
F. Western blot analysis of glycophorins expression with GlyA or GlyC antibodies and of integrin β1 (β1) expression throughout Y486 T. vivax infection.
Molecular weights (kDa) are indicated on the left. A total of 107 erythrocytes ghosts were loaded per well. Days post infection are indicated at the top. See also Fig. S4.
Download figure to PowerPoint
Then, we used the Sambucus nigra agglutinin (SNA) to specifically see sialylation modifications. SNA binds preferentially to sialic acid attached to terminal galactose with α-2,6 linkage (highly represented in glycophorins), and to a lesser degree with α-2,3 linkage (Fig. 6D). The sialylation profile was strongly affected during the course of infection (signals 1–9). Signals 1–4 decreased but signal 4, which completely disappeared on day 8, reappeared from day 11–15. Bands of small molecular weights (signal 6) and signal 5 (∼ 130 kDa) appeared on day 8. Finally, signals 7–9 appeared at the end of the infection. These changes could correspond to desialylation, modification of the amount of sialylated proteins or modification of their migration as sialylation influences migration. Production of reticulocytes, observed during infection (Fig. 3A), is likely to participate to these changes as reticulocytes are strongly glycosylated and bear intact fresh glycosylations (that could explain some signal increases) and also possess membrane proteins absent from mature erythrocytes such as the transferrin receptor, or integrin β1 (Liu et al., 2009).
So, to explore the different possibilities, we used several antibodies: TER-119, GlyA and GlyC to examine the amount of the different glycophorin forms and anti-integrin β1 antibodies (β1), since this heavily sialylated protein is predicted to migrate at 130 kDa and constituted a good marker for reticulocytes (Fig. 6 and F). First, no decrease of glycophorins were observed showing that the decrease of SNA signal corresponded to desialylation and not to reduction of the amount of protein. In fact, as expected, the apparition of galactose residues on the erythrocyte surface in parallel with desialylation was observed using a specific lectin (Fig. S4). On the contrary, most of the glycophorin signals increased during infection and some new glycophorin bands appeared on TER-119 (corresponding to signals 7 and 9) and on GlyA (corresponding to signal 8) blots. β1 Western blot clearly demonstrated the increasing number of reticulocytes throughout the infection which is in accordance with the increase in the amount of glycophorins. To confirm these data, we used a bleeding induced-anaemia murine model where reticulogenesis increased as anaemia progressed (Fig. S4). We observed that gp2 (signal 4, GlyA) slightly increased with reticulogenesis and that integrin β1 is expressed concomitantly with reticulogenesis, demonstrating that some signal increase (like signal 8 or 5) results from reticulocytes production while others (like signals 1–4 decrease) are exclusively due to trypanosome-mediated erythrocyte desialylation. Again, all these data were corroborated by mass spectrometry analyses (data not shown).
Profiles of Wheat Germ Agglutinin and Maackia amurensis lectin which bind sialic acid in addition to other glycosidic residues, were more complex and, therefore, more difficult to interpret. However, all profiles showed a common trait: modifications appeared during infections and seemed to result either from desialylation or from increased reticulogenesis (data not shown).
To summarize, some glycoproteins, most of them being glycophorins, underwent desialylation resulting in decreased signals (1–4) or possibly migration as monomers (6), while others corresponded to reticulocyte production (5–9 and re-apparition of signal 4). But most importantly, the phenomenon of desialylation throughout T. vivax infection, as described in the ageing process of erythrocytes (Bratosin et al., 1998), leads to galactose exposition recognized as a signal for phagocytosis by macrophages.
Trypanosomal TS are clearly involved in erythrophagocytosis and anaemia through erythrocytes desialylation
To establish a direct link between trypanosomal TS, desialylation and erythrophagocytosis and given the fact that inhibition of expression of TvTS enzymes with genetic tools is not possible in T. vivax, we used two methods. First, recombinant TS were injected into mice to observe potential effects on haematocrit and were used to treat erythrocytes in vitro to analyse their effect on erythrocyte sialylation and subsequent propensity to be phagocytosed (Fig. 7). We observed that injections of T. congolense or T. vivax TS resulted in haematocrit depression visible 24 h after the first injection and until 2 days after the final third injection (Fig. 7A). The most significant decrease (up to 43% of haematocrit) was observed 5 days after the first injection. These results implied that the presence of trypanosomal TS in the blood had a direct impact on haematocrit. Incubation of erythrocytes with recombinant TS in vitro resulted in modifications of SNA profiles which resembled those observed during infection especially when TvTS2 is used instead of bacterial neuraminidase (Fig. 7B). In fact, desialylation resulted in decreased SNA signals as observed for the gp2 signal 4 and increased presence of signal 6 (small molecular weights bands) and signal 7 that probably corresponded to glycophorin-modified migration as desialylation may facilitate their migration as monomers and may prevent multimer aggregation.
Figure 7. Effects of recombinant trypanosomal TS on the sialylation profile of erythrocytes and phagocytosis.
A. In vivo effect of trypanosomal TS injection on mouse haematocrit. Five to eight CD1 mice were injected with 650 μU of T. congolense TcoTS-A1 and TcoTS-D2, T. vivax TvTS2, commercial bacterial neuraminidase (NA) or BSA (control) via the retro-orbital sinus for 3 consecutive days (0, 1, 2) and haematocrit measured during 11 days. Each bar represents the mean of haematocrit values ± SD.
B. SNA lectin blot analysis of erythrocytes membranes after SA treatment. A total of 5·107 RBC from CD1 mice were treated or not (Control) with 140 μU TvTS2 or NA, for 6 h and 24 h before SDS-PAGE analysis on 10% (left) or 15% (right) acrylamide gels. Molecular weights (kDa) are indicated on the left.
C. The effect of recombinant trypanosomal TS erythrocytes treatment on erythrophagocytosis ratios. A total of 3·106 RBC were treated or not (Control) with different quantities of TvTS2, TcoTS-A1 or NA (indicated in μU) before the ex vivo erythrophagocytosis assay. Assay numbers are indicated by n.
Download figure to PowerPoint
After treatment with TS, erythrophagocytosis increased compared with non-treated erythrocytes (Fig. 7C). The increase was comparable between trypanosomal or bacterial enzyme. For TcoTS-A1 and bacterial neuraminidase we observed that the ratio of erythrophagocytosis increased in direct proportion to the quantity of enzyme (the specific activity of TvTS2 was too low to compare), highlighting the fact that increased erythrocyte desialylation is correlated with increased phagocytosis. Therefore, the concentration of TS in the host blood during infection may be critical to aggravate anaemia.
Second, we used known TS inhibitors to evaluate the impact of TS activity inhibition on erythrophagocytosis and anaemia during T. vivax infection. Arioka and colleagues described flavonoid and anthraquinone derivatives that inhibit T. cruzi TS with relative high specificity compared with human neuraminidase 2 and low IC50 value (Arioka et al., 2010). Therefore, we tested myricetin (flavonoid derivative) and rhein (anthraquinone derivative) in vitro on recombinant TvTS2 and obtained an equivalent inhibitory effect (11 μM and 24 μM for IC50 values of myricetin and rhein respectively). These compounds were subsequently injected (1 mg kg−1) into non-infected or T. vivax Y486-infected CD1 mice 1 day after infection for 4 consecutive days and parasitaemia and haematocrit were monitored and compared with non-treated T. vivax Y486-infected CD1 mice (Fig. 8 and B). Approximately the same results were obtained with the two compounds and no effect on haematocrit was observed when mice were not infected but injection of rhein seemed to be a slightly unpleasant for mice so we preferred to focus on myricetin. In infected mice, treatment decreased the anaemia but had no effect on the parasitaemia. Non-treated mice died after 6 days with a very low value of haematocrit (∼ 30%) whereas treated mice survived and developed a chronic infection (mice died after 3 weeks of infection). The prolonged survival of treated mice probably resulted from a slight recovery of haematocrit upon treatment. To evaluate the effect of myricetin on TvTS in vivo, we measured SA activity in the serum of treated infected mice versus non-treated infected mice and showed a strong decrease of SA activity upon treatment (Fig. S5A). In the same way, inhibition of TvTS2 by incubation with inhibitor prior to injection resulted only in a very slight haematocrit decrease compared with previously described decrease in Fig. 7 after injection of active TvTS2 (Fig. S5B). We subsequently measured ex vivo erythrophagocytosis with biological material from treated infected mice compared with non-treated infected mice. We observed that the erythrophagocytosis ratio with control mice (YYY/CCC) is 2.3 and 2.5 times lower when mice were treated (Fig. 8C). Levels of erythrophagocytosis are, therefore, strongly correlated with levels of anaemia and we proposed that inhibition of TS activity resulted in decreased erythrophagocytosis and consequently, in decreased anaemia. To corroborate these data we analysed sialylation profiles of erythrocytes from treated and non-treated mice and showed that desialylation is less pronounced in treated mice which correlated perfectly with the lower level of erythrophagocytosis and reinforced the role of erythrocyte desialylation by trypanosomal TS in anaemia during T. vivax infection (Fig. 8D).
Figure 8. Effect of sialidase inhibitors on anaemia and ex vivo erythrophagocytosis during experimental infection.
A and B. Effects of myricetin (Sigma) injection on parasitaemia (A) and haematocrit (B) during T. vivax Y486 infection in CD1 mice. Myricetin (1 mg kg−1) was injected intraperitoneally or in the retro-orbital sinus 1 day post infection (infection with 3–4·106 BSF of T. vivax Y486) for 4 consecutive days in five CD1 mice (treated mice). Results are shown for 7 days, after this period, treated mice survived and developed classical subchronic infections as described in Fig. 3. Non-treated CD1 mice (infected at the same time) died on day 5 or 6 post infection.
C. Comparison of ex vivo erythrophagocytosis ratio YYY/CCC of treated mice versus non-treated mice. Erythrocytes, serum and macrophages were collected 5 days post infection from four different CD1-infected mice.
D. Comparison of erythrocyte sialylation profiles on SNA blot for erythrocyte ghosts from control mice (non-treated and non-infected lane 1), infected and treated mice (lane 2 and 3) and infected and non-treated mice (lane 4). Erythrocytes were collected 5 days post infection. Treatment had no effect on the sialylation profile of non-infected mice compared with non-treated control mice (data not shown).
Download figure to PowerPoint