A Trypanosoma brucei Kinesin Heavy Chain Promotes Parasite Growth by Triggering Host Arginase Activity

Background In order to promote infection, the blood-borne parasite Trypanosoma brucei releases factors that upregulate arginase expression and activity in myeloid cells. Methodology/Principal findings By screening a cDNA library of T. brucei with an antibody neutralizing the arginase-inducing activity of parasite released factors, we identified a Kinesin Heavy Chain isoform, termed TbKHC1, as responsible for this effect. Following interaction with mouse myeloid cells, natural or recombinant TbKHC1 triggered SIGN-R1 receptor-dependent induction of IL-10 production, resulting in arginase-1 activation concomitant with reduction of nitric oxide (NO) synthase activity. This TbKHC1 activity was IL-4Rα-independent and did not mirror M2 activation of myeloid cells. As compared to wild-type T. brucei, infection by TbKHC1 KO parasites was characterized by strongly reduced parasitaemia and prolonged host survival time. By treating infected mice with ornithine or with NO synthase inhibitor, we observed that during the first wave of parasitaemia the parasite growth-promoting effect of TbKHC1-mediated arginase activation resulted more from increased polyamine production than from reduction of NO synthesis. In late stage infection, TbKHC1-mediated reduction of NO synthesis appeared to contribute to liver damage linked to shortening of host survival time. Conclusion A kinesin heavy chain released by T. brucei induces IL-10 and arginase-1 through SIGN-R1 signaling in myeloid cells, which promotes early trypanosome growth and favors parasite settlement in the host. Moreover, in the late stage of infection, the inhibition of NO synthesis by TbKHC1 contributes to liver pathogenicity.


Introduction
The protozoan flagellate parasite Trypanosoma brucei is responsible for the diseases human sleeping sickness and nagana in cattle. In experimental murine models, the host immune response to this blood-borne pathogen involves antibody production against the Variant Surface Glycoprotein (VSG), as well as interferon-c (IFN-c)mediated activation of macrophages/myeloid cells into cells of the M1 phenotype. These engulf opsonized parasites and synthesize factors that interfere with trypanosome growth including tumor necrosis factor-a (TNF). However, uncontrolled IFN-c-induced immune responses including TNF and NO production as the infection persists induce tissue pathogenicity and death of the host [1][2][3][4][5]. Induction of IL-10 can attenuate the IFN-c/M1 response and hereby enables prolonged survival of T. brucei-infected mice [6][7][8].
Only few T. brucei-derived immunomodulatory factors have been identified. These include the glycosylphosphatidylinositol anchor of VSG, CpG oligodeoxynucleotides and the trypanosome suppressive immunomodulating factor (TSIF) which have been found to induce TNF-and NO-secreting M1 cells [9][10][11][12]. Factors inducing M1 cells favor the control of parasite development in the early stage of infection, but their continuous release in the late stage of infection can sustain inflammation responsible for tissue pathogenicity. Components found in the culture medium of T. brucei have also been shown to affect immune cells of the host. In particular, factors released by the parasite promote the degradation of L-arginine through increase of arginase activity in macrophages/myeloid cells, and antagonize NO synthases (NOS)-mediated conversion of L-arginine into NO in infected mice. Arginase induction appears to attenuate the innate response at the early stage of infection, and likely contributes to the synthesis of polyamines and the trypanosome anti-oxidant trypanothione, known to promote trypanosome growth and colonization of the host [5,[13][14][15]. We have identified TbKHC1, a kinesin heavy chain isoform, as a factor released by T. brucei to trigger host arginase-1 activity for promotion of its own growth in the host. Although arginase-1 expression is commonly associated with alternatively activated myeloid cell (M2) functions, TbKHC1induced arginase-1 was independent of IL-4Ra signaling but relied on a SIGN-R1 receptor-dependent IL-10 secretion.

Results
A T. brucei kinesin heavy chain induces arginase activity in myeloid cells T. brucei parasites were found to induce arginase activity in myeloid cells from non-infected mice (Fig. 1A). This induction was maintained when myeloid cells and trypanosomes were separated by a cell-retaining insert, indicating that soluble components from trypanosomes were involved (Fig. 1A). Parasite-released factors (PRF) were prepared under conditions leading to no detectable trypanosome death. PRF induced arginase activity, and this effect was abolished by heat-treatment (Fig. 1A). Monoclonal antibodies raised against T. brucei PRF (2C12) inhibited arginase activity induced by PRF, whereas other antibodies from the same hybridoma fusion (2A3, 2C6, 4B5) or of irrelevant specificity (5F6, 4J5) had no effect (Fig. 1B). The PRF fraction eluted after binding to a 2C12 antibody-based affinity column retained full arginase-inducing activity, confirming that this activity was directly targeted by 2C12 (Fig. 1C).
The 2C12 antibody was used to screen a cDNA expression library and identify the T. brucei arginase-inducing protein. Among the 6 positive clones identified, 5 encoded fragments of the same putative kinesin heavy chain isoform (geneDB accession number Tb927. 6.4390), which we termed TbKHC1 for T. brucei kinesin heavy chain 1 (Fig. S1in Text S1). The protein N-terminal domain contained the highly conserved kinesin motor domain with a typical ATP binding site. The region targeted by 2C12 antibody located downstream from the motor domain and was characterized by a predicted coiled structure (Fig. S1 in Text S1).
TbKHC1 knock-out (KO) parasites were generated from wildtype (WT) T. brucei. In contrast to PRF from WT parasites, PRF from TbKHC1 KO parasites did not trigger arginase activity in myeloid cells from non-infected mice (Fig. 1C).
TbKHC1 partially localizes in the endocytic compartment and is released from the parasite Pleomorphic trypanosomes, differentiating in the bloodstream from proliferative slender forms into quiescent stumpy forms, can induce long-standing infection in mammals and perform cyclical transmission in tsetse flies while differentiating into procyclic forms. In contrast, monomorphic trypanosomes, resulting from prolonged cultivation in vitro, are unable to develop long-lasting infection and cyclical transmission. The TbKHC1 gene was found to be expressed preferentially in slender bloodstream forms ( Fig. 2A,B). Parasite immunolabeling with both monoclonal 2C12 and polyclonal antibodies generated against recombinant TbKHC1 (rTbKHC1) vanished in TbKHC1 KO parasites or RNAi-mediated TbKHC1 knock-down (KD) parasites, while it increased in trypanosomes over-expressing TbKHC1 ( Fig. 2A-D), confirming that TbKHC1 is the genuine target of the anti-PRF 2C12 antibody.
Both antibodies revealed a scattered protein distribution primarily between the kinetoplast and the nucleus (Fig. 2C,D). TbKHC1 partially co-localized with tomato lectin, which detects linear chains of poly-N-acetyllactosamine typical of the endocytic compartment [16] (Fig. S2A in Text S1). However, TbKHC1 did not bind to tomato lectin in affinity chromatography assay ( Fig.  S2B in Text S1), suggesting that TbKHC1 is not an endocytic component. Moreover, a fraction of TbKHC1 was reproducibly found in the culture medium of parasites, as revealed by immunoprecipitation of detergent-free supernatants from metabolically 35 S-labelled T. brucei (Fig. 2E). In these experiments, trypanosome lysis was unlikely as the supernatants did not contain any evidence of the cytoplasmic markers actin [17] or protein disulfide isomerase 2 [18], but did contain enolase, a protein released from trypanosome-related parasites such as Leishmania [19] (Fig. 2E). TbKHC1 immunoprecipitated from the trypanosome supernatant did not associate with other 35 S-labelled proteins (Fig. 2E), as could have been expected in case of release as a complex.

IL-10 and mannose regulate TbKHC1-induced arginase activity
We screened for pathways altering the levels of PRF-induced arginase activity in myeloid cells. Co-incident with the induction of arginase activity, PRF induced myeloid cells to express the regulatory cytokine IL-10 (Fig. 3A). The arginase activity induced by PRF was inhibited by neutralizing anti-IL-10 antibody and by D-mannose, but not by D-galactose (Fig. 3B).

Author Summary
From the first invasive step onwards, African trypanosomes can efficiently undermine the protective immune response of their mammalian host to favor their survival within the host and successful transmission by its vector, the tsetse fly. Identifying the parasite factors affecting the protective immune response is thus critical to detail the immune evasion mechanisms of these organisms. We report here that during acute infection, a Trypanosoma brucei protein named Kinesin Heavy Chain 1 (TbKHC1) sustains the development of the first (most prominent) peak of parasitaemia in the blood and its control by the host. Mechanistically, TbKHC1 was found to interact with the SIGN-R1 molecule at the surface of immune cells. Hereby, TbKHC1 modulates arginine/NO metabolism in immune cells towards the production by the host of nutrients (polyamines) required for parasite growth via an IL-10dependent induction of arginase 1 and down-regulation of iNOS activities. Consequently, IL-10/arginase 1 producing immune cells are impaired in their capacity to destroy the parasite, favouring parasite settlement. Moreover, in the late stage of infection, the inhibition of NO synthesis by TbKHC1 increases liver pathogenicity that contributes to compromised host survival. Thus, targeting TbKHC1 may benefit the host protective immunity against T. brucei parasite.
rTbKHC1 mimicked the arginase-inducing effect of PRF, including its inhibition by 2C12, anti-IL-10R antibodies and Dmannose (Fig. 3C). The increased arginase activity induced by rTbKHC1 associated with increased expression of the Arg1, but not Arg2 gene, as well as impaired expression of the iNOS gene (Fig. 3D). Activation of myeloid cells by rTbKHC1 also resulted in increased expression of the IL-10 gene (Fig. 3D) and protein (Fig. 3C), the latter being inhibited by the 2C12 antibody and Dmannose. Thus, TbKHC1 interaction with myeloid cells involved a mannose-sensitive event triggering the synthesis of IL-10, resulting in enhanced Arg1 gene expression and activity, and inhibition of iNOS expression.
Although arginase-1 expression is commonly associated with alternatively activated myeloid cell (M2) functions, rTbKHC1 did not trigger Il4 and Il13 gene expression in myeloid cells from WT mice (Fig. 3D). Moreover, the induction of Arg1 and Il10 genes by rTbKHC1 was conserved in IL-4Ra KO mice (Fig. 3E), excluding that IL-10 synergistically enhanced IL-4Ra-induced Arg1 expression as observed in M2 cells [20,21]. Accordingly, rTbKHC1 did not trigger the expression of M2-associated genes characteristic of African trypanosome infection [22] (Fig. 3F). We also addressed whether the induction of Arg1 by rTbKHC1 required IL-6 and G-CSF in combination with IL-10 [23]. rTbKHC1 only induced Il6 gene expression (Fig. 3D). This expression was not affected following addition of anti-IL-10R antibody that inhibited the induction of arginase activity by rTbKHC1 (Fig. 3G), indicating that IL-10 induced by TbKHC1 is the main trigger of Arg1 expression in myeloid cells. TbKHC1 promotes parasite growth in mice by increasing ornithine availability In vitro, the parasite growth rate appeared to be identical between WT, TbKHC1 KD, TbKHC1 KO and TbKHC1-rescued parasites obtained by reinsertion of the TbKHC1 gene in one allele of its original locus in the TbKHC1 KO genome (Fig. 4A,B), indicating that TbKHC1 is not required for parasite growth. Moreover, TbKHC1 KO parasites motility or flagellum movement addressed by microscopic examination and sedimentation analysis [24] appeared normal (not shown). TbKHC1 KO parasites were able to perform the full parasite life-cycle, since procyclic transformants could cycle through tsetse flies to produce infective metacyclic forms.
In C57Bl/6 mice, the cumulative parasite load in the first peak of parasitaemia by TbKHC1 KO trypanosomes was reduced by .70% as compared to WT parasites (Fig. 4C). This reduced TbKHC1 KO parasite load was also observed under natural infection conditions in which infected tsetse flies were allowed to feed on mice (Fig. 4D). Reinsertion of TbKHC1 in TbKHC1 KO parasites reverted early parasitaemia to WT levels ( Fig. S3 in Text S1).
As compared to infection with WT parasites, at day 7 postinfection (p.i.) with TbKHC1 KO parasites, spleen myeloid cells showed a reduced induction of arginase activity that coincided with increased NO 2 accumulation, likely reflecting a shift towards NOS activity (Fig. 4E,F). Moreover, the induction of arginase activity was not observed in infected mice treated with a neutralizing anti-IL-10R antibody (Fig. 4E). Furthermore, mice infected with TbKHC1 KO parasites exhibited lower IL-10 serum levels than mice infected with WT parasites (Fig. 4G). The reduction of IL-10 secretion in TbKHC1 KO-infected mice could account for the higher inhibition of arginase activity observed in the presence of anti-IL-10R antibodies in these mice (Fig. 4E).
In TbKHC1 KO-infected mice, it is unlikely that the reduction of the first peak of parasitaemia resulted from increased NO production. Indeed, inhibition of NOS activity by N-(G)-nitro-Larginine methyl ester (L-NAME) or absence of iNOS activity in iNOS KO mice did not affect TbKHC1 KO parasitaemia, while it strongly reduced WT parasitaemia (Fig. 5A,B). The lower parasitaemia in TbKHC1 KO-infected mice might rather result from reduced arginase activity, which would limit nutrient availability for the parasite. Indeed, this enzyme converts Larginine to L-ornithine, a precursor of polyamines that are required for trypanosome growth [13]. Accordingly, in mice lacking arginase-1 in myeloid cells/macrophages following a cross between Arg1 loxP-targeted mice and LysM Cre or Tie2 Cre deleter mice [21], WT parasitaemia dropped to that of TbKHC1 KO parasitaemia (Fig. 5C,D). Moreover, treatment of mice with Lornithine increased the cumulative parasite load to a greater extent in TbKHC1 KO-than in WT-infected mice (,94% vs ,43%, respectively; Fig. 5E). In addition, spermine levels tended to increase although without reaching statistical significance in spleen myeloid cells and blood from mice infected with WT but not TbKHC1 KO parasites ( Fig. S4A in Text S1). Furthermore, an  increase in L-ornithine production, coinciding with the consumption of L-arginine and the induction of N-acetylputrescine production, was observed in the supernatant of myeloid cells from non-infected mice activated in vitro with rTbKHC1 ( Fig.  S4B in Text S1).

The effect of TbKHC1 on parasite growth involves the myeloid cell receptor SIGN-R1
Since the in vitro induction of arginase activity by PRF and rTbKHC1 was inhibited by D-mannose (Fig. 3B,C), we treated infected mice with D-mannose. This treatment reduced WT parasitaemia to the level of TbKHC1 KO parasitaemia, but did not significantly affect TbKHC1 KO parasitaemia (Fig. 5F). To follow up on this in vivo finding and the in vitro observation that arginase activation by PRF and rTbKHC1 was inhibited by D-mannose as well as by an antibody recognizing coils of TbKHC1 (Fig. 1B, 3C), we monitored the course of WT-and TbKHC1 KO-infection in mice deficient for myeloid cell receptors able to bind both mannose and peptidic coils, namely MMR (CD206) and SIGN-R1 (CD209b), using appropriate congenic control mice [25][26][27]. In MMR KO mice, WT and TbKHC1 KO parasitaemias were not affected as compared to infection in WT mice (Fig. 5G). In SIGN-R1 KO animals the differential control of the first peak of parasitaemia between WT and TbKHC1 KO parasites vanished, due to the reduction of the WT parasite level to that of TbKHC1 KO parasites (Fig. 5H). Thus, either the absence of SIGN-R1 in infected mice or the absence of TbKHC1 in the parasite decreased T. brucei parasitaemia similarly. These data suggested that TbKHC1 interacts with the SIGN-R1 receptor. To evaluate this hypothesis, we tested the in vitro activity of rTbKHC1 on myeloid cells from SIGN-R1 KO mice. In contrast to results obtained with myeloid cells from control mice, in myeloid cells from SIGN-RI KO mice rTbKHC1 did not stimulate IL-10 and Arg1 gene expression (Fig. 6A) and did not increase arginase activity (Fig. 6B). Conversely, in myeloid cells from MMR KO mice rTbKHC1 induced expression of Il10 and Arg1 genes as in WT mouse myeloid cells (Fig. S5 in Text S1). Therefore, TbKHC1 appeared to trigger arginase activity through SIGN-R1-mediated signaling.

TbKHC1 increases liver pathogenicity and reduces host survival by NO-dependent mechanisms
The survival time of TbKHC1 KO-infected mice was significantly prolonged as compared to WT-infected animals ( Fig. S6 in Text S1, Table 1). Death of mice infected with African trypanosomes is related to systemic inflammatory response syndrome (SIRS) that causes multiple organ failure [28]. Liver myeloid cells are critical to the clearance of trypanosomes from the bloodstream [29], hence maintaining liver integrity is fundamental to the host's ability to clear trypanosomes. At day 30 p.i. several indicators of liver damage differed between WT-and TbKHC1 KO-infected mice: (i) the serum levels of the liver damage marker alanine aminotransferase (ALT) were lower in TbKHC1 KO-than in WT-infected mice; (ii) anoxic infarcts were observed in none of the TbKHC1 KO-infected mice, but in 6 out of 8 WT-infected mice; (iii) lobular and portal mononuclear cell infiltrates were milder in all TbKHC1 KO-than in WT-infected mice ( Table 1, Fig. 6C). Concomitantly, NO production was higher in TbKHC1 KO-than in WT-infected mice, while IL-10 production was not induced and similar in all infected experimental groups (Fig. 6D). A moderate increase in iNOS mRNA expression level was also observed in myeloid cells from mice infected with TbKHC1 KO parasites (4.261.5 vs 1.3260.33 fold in mice infected with WT parasites, n = 3; p = 0.08). These data suggested that increased NO production protected liver integrity in TbKHC1 KO-infected mice. Accordingly, L-NAME treatment or knock-out of iNOS activity in TbKHC1 KO-infected mice reduced the survival time and increased ALT levels to those found during WT parasite infection (Table 1). L-NAME treatment or absence of iNOS activity in WTinfected mice did not affect the survival time, although it reduced the serum levels of ALT (Table 1). Therefore, a threshold of NOS activity, achieved in TbKHC1 KO-but not in WT-infected mice, might be required during T. brucei infection to protect liver integrity, thereby contributing to extended host survival. This hepatoprotective effect of NO was not due to difference in parasite load in mice infected with WT and TbKHC1 KO parasites (Fig. S7 in Text S1).
The survival times and ALT levels in WT and TbKHC1 KOinfected mice were similar in Arg1-deficient mice and control mice (Table 1). However, treatment of WT-infected mice with Lornithine slightly increased the serum level of ALT, while in TbKHC1 KO-infected mice L-ornithine treatment reduced the survival time and increased ALT levels to those found during WT parasite infection (Table 1). Therefore, L-ornithine as a polyamine precursor favored the development of liver injury and negatively affected the survival of T. brucei-infected mice.
As in the liver, brain injury characteristic of African trypanosomiasis was also lower in TbKHC1 KO-than in WT-infected mice. Indeed, no parasites could be detected in the choroid plexus of TbKHC1 KO-infected mice, sharply contrasting with the presence of numerous parasites, associated with immune infiltrates, in the choroid plexus of WT-infected mice (Fig. S8 in Text S1).
The survival times as well as ALT, NO and IL-10 levels in WTand TbKHC1 KO-infected mice were similar in SIGN-R1 KO and control mice (Table 1, Fig. 6E), suggesting that in late infection TbKHC1 signaling no longer operates primarily through SIGN-R1.

Discussion
We report that a particular isoform of the large KHC family of T. brucei is released by the trypanosome and influences both parasite growth and host pathogenicity.
T. brucei contains 51 sequences encoding kinesin-like proteins. The T. brucei isoform described here appears to be an orphan member without predicted function [30]. That a KHC isoform could be released in the extracellular environment is not unprecedented, since a KHC-like protein has been reported as secreted from the coagulating gland in rat [31]. TbKHC1 did not appear to be contained in high molecular weight complexes released from the parasite, for instance in exosomes as occurs in T. cruzi [32] and Leishmania [33]. Our immunoprecipitation data, rather, suggested that TbKHC1 is free in the medium, but the mechanism of release remains unclear.
TbKHC1 did not appear to be essential for T. brucei cellular proliferation or differentiation.
In vitro, TbKHC1 triggered arginase activity and, in contrast to the other documented T. brucei immunomodulators VSG, DNA or TSIF [9,10,12], inhibited NO synthesis by myeloid cells. These effects involved a region rich in coils in the TbKHC1 C-terminal domain, since they were inhibited by an antibody specifically recognizing the coiled region of TbKHC1.
Arg1 is often recognized as a prototypic M2 myeloid cell marker induced following STAT6 activation by IL-4/IL-13/IL-4Ra signaling. However, Arg1 can be expressed by M1 cells in a STAT3-dependent pathway triggered by IL-10, IL-6 and G-CSF [21,23]. In agreement, rTbKHC1 induced the expression of Il10 and Il6 genes in myeloid cells, and this neither involved IL-4Ra nor induced the expression of genes associated with M2 activation. Since Il6 expression was still induced in the presence of the Arg1 expression inhibitor anti-IL-10R antibody, we conclude that rTbKHC1-induced Arg1 expression primarily depended on IL-10 production.
In mice, TbKHC1 activity appeared to promote parasite growth in the early stage of infection. Reduced parasitaemia induced by TbKHC1 KO trypanosomes was associated with increased NO production, but did not directly result from this increase. Indeed, inhibition of NO production by L-NAME treatment or in iNOS KO mice did not improve TbKHC1 KO growth. Rather, in agreement with previous studies [34] L-NAME treatment or absence of iNOS activity in mice actually increased the control of WT parasitaemia. Together, these experiments suggested that the negative/braking effect of NO on the control of WT parasitaemia requires the expression/activity of TbKHC1. We propose that TbKHC1-mediated induction of host arginase activity fuels L-ornithine production and hereby the synthesis of polyamines, which are essential nutrients for growth of extracellular trypanosomes in the host. In bloodstream forms of T. brucei the arginase gene and activity are absent, but when exogenous Lornithine is lacking in body niches these parasites can use Larginine to produce L-ornithine through an arginase-independent mechanism [35,36]. However, the differential growth-promoting effect of exogenous L-ornithine during infection by WT or TbKHC1 KO parasites suggests that this arginase-independent mechanism is insufficient to provide all the amines necessary for the polyamine pathway. Similarly, the related but intracellular parasite Leishmania also requires host arginase induction for optimal growth, despite the presence of an arginase gene [37]. Therefore, optimal production of L-ornithine/polyamines for T. brucei growth would rely on host arginase activity induced by TbKHC1. In accordance, in two models of mice deficient for arginase-1 in myeloid cells/macrophages, WT T. brucei growth was reduced to the level observed with TbKHC1 KO parasites.
In T. brucei-infected mice, the TbKHC1-mediated effect on parasitaemia appeared to involve the myeloid cell receptor SIGN-R1. Indeed, improved parasite growth control was observed after either removal of TbKHC1 from the parasite or removal of SIGN-R1 from the host. Moreover, rTbKHC1-mediated induction of arginase activity was lost in myeloid cells from SIGN-R1 KO mice. The involvement of SIGN-R1 in TbKHC1-mediated IL-10 and arginase production is in keeping with previous data demonstrating the anti-inflammatory role of SIGN-R1 in various diseases [38][39][40]. Although LPS and intracellular mycobacteria infection can induce Arg1 in a MyD88/IL-10 dependent manner [21,23], we can exclude that TbKHC1 is simultaneously a ligand for SIGN-R1 and a trigger for MyD88 signaling in our extracellular infection model since the differential infection phenotype between WT and TbKHC1 KO T. brucei was conserved in Myd88 KO mice (data not shown).
D-mannose was found to decrease parasite growth during the first peak of parasitaemia by WT but not TbKHC1 KO parasites, and it inhibited rTbKHC1-mediated induction of IL-10 and arginase by myeloid cells. Similarly, the 2C12 antibody, which is specific to the coiled region of TbKHC1, inhibited IL-10 production and arginase activity triggered by rTbKHC1. Mannose-specific receptors such as SIGN-R1 and MMR are known to interact with coiled proteins, the complement protein C1q and collagen respectively [26,27]. Therefore, it is possible that mannose binding to SIGN-R1 could affect the binding of TbKHC1 peptidic coils to a second binding site of this receptor.
After the parasite growth-promoting effect occurring during the first peak of parasitaemia, a distinct effect of TbKHC1 consisted in reducing the survival time of infected mice while contributing to liver injury. Given the similar in vitro growth properties between parasites expressing or not TbKHC1, the mechanism by which this protein influenced the host survival time is likely to be independent from differences in parasite virulence. Parasite load throughout the infection, survival and pathogenicity are independent traits in experimental and natural infection with African trypanosomes [2,4,[6][7][8]28]. Therefore, the survival time difference between WT-and TbKHC1 KO-infected mice is primarily due to differences in tissue damage inflicted by inflammatory immune responses, rather than differences of parasite load in the first peak of parasitaemia. NO contributes to peripheral pathogenicity induced by African trypanosomes, while IL-10 exerts counterpathogenic activity [2,3]. In late stages of infection, IL-10 production was unaffected by the absence of TbKHC1, while NO production (and iNOS expression) was lower when TbKHC1 was present. Moreover, the reduction of liver injury in TbKHC1 KO-infected mice was erased by inhibition of NO synthesis upon L-NAME treatment or by infecting iNOS KO mice. Thus, TbKHC1-mediated depletion of NO might contribute to the liver damage associated with WT parasite infection. Accordingly, NO plays a pivotal role in vascular tone, and decreased NO bioavailability contributes to pathogenesis in chronic infectious diseases like malaria, and in liver ischemia reperfusion injury [41,42]. Our observation that after the first peak of parasitaemia parasite load was similar in WT-and TbKHC1 KO -infected mice, even upon inhibition of NO production, supports the notion that blood parasite load is not a major contributor to liver pathogenicity.
Besides NO, polyamines contribute to the pathogenicity of T. brucei infection [43]. Since our data suggest that induction of arginase activity by TbKHC1 could sustain the production of polyamines, these molecules could contribute to increase liver injury in WT-as well as in TbKHC1 KO-infected mice fed with Lornithine.
In the late stage of infection, the effects of TbKHC1 did not appear to depend on SIGN-R1 any longer, since the absence of SIGN-R1 did not affect the survival time and liver injury in WTand TbKHC1 KO-infected mice. A major difference between early and late infection is the evolution of the immune response driven by the continuous release of immunomodulatory/inflammatory factors by the parasite, such as VSG and TSIF, which induce pathogenic TNF-a and IFN-c [1,4,9] while switching off the IL-10 production [7]. The evolving context of host-parasite interaction could affect TbKHC1 signaling, and indeed the myeloid cell activation state and corresponding expression of receptors is known to vary during the time course of T. brucei infection [44]. Therefore, we cannot exclude that in infected SIGN-R1 KO mice, compensation by other myeloid cell receptors influences the immune response and outcome of pathogenicity in the late stage of the disease.
In conclusion, we have identified an unexpected function for a KHC of extracellular T. brucei, which is SIGN-R1-mediated activation of myeloid cells to produce IL-10 leading to increase in arginase activity, resulting in promotion of parasite growth in the host.

Ethic statement
Experiments, maintenance and care of mice complied with guidelines of the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes (CETS nu123) and were approved by the Ethical Committee for Animal Experiments of the Université Libre de Bruxelles, Brussels, Belgium (laboratory accreditation number LA2500482).
Mice were infected i.p. with 2,000 AnTat1.1E parasites [6]. Parasitaemia was monitored by tail blood puncture. When required, mice were injected i.p. 2 h before infection with 2 mg L-NAME, then 1 mg/ml L-NAME was added to drinking water. The less active enantiomer D-NAME was used as control. For Lornithine treatment, mice received 2 mg i.p. 2 h before infection, and then were given 5 mg/ml in drinking water. For D-mannose or D-galactose treatment, mice received 50 mg i.p. 2 h before infection, then every 2 days. Sugar treatment caused massive inflammation and the animals were euthanized 12 days p.i. To neutralise IL-10 activity, mice were treated once with 200 mg control antibody (rat IgG1, BD Biosciences) or 1B1.3a anti-IL-10R antibody at day 6 p.i. Injections of anti-IL-10R antibody at day 2, 4 and 6 p.i. killed WT-and TbKHC1 KO-infected mice within 48 h, as reported [6].
Parasite released factors, antibodies and PRF immunoaffinity column PRF was prepared as described [15]. Monoclonal antibodies were obtained following an immunoenzymatic screening that selected antibodies directed against PRF, followed by a test based on inhibition of arginase activity induction. rTbKHC1 C-terminal fragment was used for production of rabbit polyclonal antibodies. PRF was incubated overnight with 2C12 monoclonal antibody coupled to Sepharose 4B CNBr resin at 4uC in carbonate buffer (v/v). After loading on a column and washings, bound material was eluted with glycine-HCl (pH 3), then glycine-NaOH (pH 10), dialysed for 24 h against milliQ water, lyophilised and solubilized at 1 mg/ml in 50 mM phosphate buffer, 1 M NaCl (pH 6.2). The residual level of endotoxins was below 2 U/mg (LAL assay, Biowhittaker).

cDNA expression library screening
Approximately 10 5 plaques from a bloodstream T. b. gambiense LiTat1.3 cDNA expression library (lambda ZAP express, Stratagene) were screened by incubation with the mouse 2C12 monoclonal antibody. Clones targeted by an anti-mouse antibody/alkaline phosphatase conjugate were identified by incubation at pH 9.5 with the chromogenic substrate NBT/BCIP. Reactive clones were plaque purified by 2 additional rounds of screening. The positive clones were processed to obtain the circular pBK CMV plasmid version of the phage, according to the manufacturer instructions.

TbKHC1 KD and KO parasites
For TbKHC1 KD, a 500 bp fragment from the TbKHC1 open reading frame (ORF) was amplified by polymerase chain reaction (PCR) using the KD primers (Table S1 in Text S1), digested with XhoI and BamHI and ligated into the p2T7-177TiTA plasmid [46]. The vector was linearized with NotI and transfected into monomorphic bloodstream forms. Stable transformants were selected with 1 mg/ml hygromycin. Double stranded RNA production was induced by addition of 1 mg/ml doxycyclin. For TbKHC1 KO parasites, both alleles of the complete TbKHC1 ORF were replaced in procyclic forms by genes encoding resistance to neomycin and bleomycin, following transfection of these genes flanked by 75 bp sequences from the TbKHC1 UTRs. Primers used for PCR amplification of the resistance genes were KOneo and KObleo (Table S1 in Text S1). Stable transformants were selected with 15 mg/ml geneticin and 2.5 mg/ml phleomycin. Pleomorphic TbKHC1 KO parasites were obtained by cyclical transmission of TbKHC1 KO procyclic forms in tsetse flies.

Rescue of TbKHC1 KO parasites
TbKHC1 gene and UTRs were PCR-amplified (Rescue primers, Table S1 in Text S1) and cloned into pCR-XL-TOPO vector (Invitrogen). The neomycin resistance cassette was extracted from pLew114hyg59 [45] with AflII and StuI restriction enzymes and cloned into the AflII and EcoRV digestion of the pCR-XL-TOPO-TbKHC1 clones presenting the 59 sequence of TbKHC1 gene next to the pUC origin. pCR-XL-TOPO-TbKHC1-Neo was linearized at AvrII and transfected in vivo into TbKHC1 KO pleomorphic parasites as described [47].

TbKHC1 overexpressing trypanosomes
TbKHC1 ORF was amplified by PCR from the initial pBK-CMV plasmid clone 4 (primers OV, Table S1 in Text S1) digested with EcoRV and HindIII, and cloned in the pTSARib vector [48]. The construct was linearized with BglII before transfection.

rTbKHC1
The full length TbKHC1 and a 2,235 bp TbKHC1 fragment were amplified by PCR from the initial pBK-CMV plasmid clone 4 using primers FL and C-term (Table S1 in Text S1). The products were digested with SalI and NotI and cloned in-frame in pET 21d (Novagen). rTbKHC1 was extracted from IPTG-induced E. coli and purified on Ni-NTA agarose (Qiagen) using 60% isopropanol in the washing steps to remove endotoxins [49]. The protein was additionally purified by HPLC Superdex 200 (10/300 GL, Pharmacia) in 50 mM phosphate buffer, 1 M NaCl (pH 6.2). The residual level of endotoxins was below 2 U/mg (LAL assay, Biowhittaker).

Tomato lectin chromatography
Fractionation of trypanosome extracts on tomato lectin affinity column was performed as described [16].

Preparation of myeloid cells
Non elicited peritoneal cells were collected in 0.34 M sucrose. Spleen homogenates were incubated 10 min in cold 0.83% NH 4 Cl/0.01 M Tris-HCl, pH 7.2 to lyse erythrocytes. Peritoneal and spleen cells were washed twice in complete medium (RPMI 1640 supplemented with 10% heat-inactivated FCS, 5610 25 M 2-mercaptoethanol, 2 mM L-glutamine, 100 IU/ml penicillin, 100 mg/ml streptomycin, and 0.1 mM non essential amino acids; all from Invitrogen Life Technologies) and counted by trypan blue exclusion. To prepare myeloid cells, peritoneal cells adjusted to 5610 6 cells/ml in complete medium were dispensed by 1 ml in 6well plates and spleen cells adjusted to 10 7 cells/ml were dispensed by 10 ml in 10-cm tissue culture dishes. After 3 h at 37uC in 5% CO 2 , cells were washed five times with warm complete medium to remove non adherent cells. Adherent cells were collected in cold complete medium using a scraper, washed and viability was determined by trypan blue exclusion (usually .95%). Cells (CD11b + for .90% as determined by flow cytometry, data not shown) were suspended in complete medium at 10 6 cells/ml.

In vitro culture of myeloid cells
Peritoneal myeloid cells from non infected mice (5610 5 /ml) were cultured in 24-well plates in presence of 5610 5 /ml trypanosomes, 80 mg/ml PRF, 3 mg/ml eluate of anti-PRF antibody affinity column or 3 mg/ml rTbKHC1. When required, antibodies (monoclonal anti-PRF, JES5-2A5 anti-IL-10 (BD Biosciences), 1B1.3a anti-IL-10R or appropriate isotype-matched antibodies (3 mg/ml), D-mannose or D-galactose (50 mM) were added to PRF-or rTbKHC1-stimulated cells. Unless specified, gene expression, arginase activity and NO 2 and IL-10 secretions were determined after 24 h of culture. Alternatively, spleen myeloid cells from infected mice (10 6 /ml) were collected at day 7 or 30 p.i. and stored for arginase activity determination or cultured for 24 h before determining spontaneous NO 2 and IL-10 production in culture supernatants.

Myeloid cell metabolite extraction, HPLC-MS analysis and data processing
Spleen myeloid cells prepared as described above were cultured in modified complete medium (2% FCS, no 2-mercaptoethanol, no antibiotic) in 24-well plates in presence of 3 mg/ml rTbKHC1. After 12 h culture, supernatants were collected, rapidly cooled to 4uC in dry ice ethanol bath and centrifuged (3 min, 4uC, 13,000 g). Protein denaturation was achieved by adding 200 ml of 4uC chloroform:methanol:water (1:3:1) to 5 ml of cell supernatants. Samples were shaken for 1 h at 4uC and centrifuged again. The supernatants were extracted and frozen at 280uC until the samples were submitted to High Performance Liquid Chromatography-Mass Spectrometry (HPLC-MS) analysis at the Scottish Metabolomics Facility using High Performance Liquid Chromatography (Dionex UltiMate 3000 RSLC system) with a ZIC-HILIC 15064.6 mm, 5 mm columns (Merck Sequant). The mobile phase comprised 0.1% formic acid in water (phase A) and 0.08% formic acid in acetonitrile (phase B). A linear gradient was applied to B from 80-20% over 30 minutes, followed by an 8 minute wash with 5% B and concluded with 8 minute equilibration with 80% B. The flow rate was 300 mL/min and 10 mL was injected into the column. The column temperature was 20uC and the autosampler temperature was 4uC.
For data processing, LC-MS raw data were processed using IDEOM software (version 17) [50] which is a composite of XCMS [51], mzMatch [52], R and Microsoft Excel packages. Firstly, XCMS detected peaks with a peak width between 10 and 100 seconds; signal to noise ratio of 5 and mass deviation between scans of 2 parts per million (ppm). MzMatch then matched the peaks between the two samples, comparing their retention times (RT) and ion densities before calculating any deviations. Peaks were identified according to their mass-to-charge ratio and RT.
Arginase activity, NO 2 and IL-10 secretion, spermine levels Arginase activity was determined as described [53]. NO 2 quantification (reflecting NO production) was assayed by a Griess reaction. IL-10 was quantified with a specific sandwich ELISA (PharMingen) in accordance to the manufacturers' protocols. Spermine levels (Spermine ELISA Kit, Antibodies-online) were determined in blood serum and myeloid cell homogenates following the manufacturers' protocols.

Gene expression analysis
Quantitative real time PCR was performed as described [6]. Results of the PCR analyses were normalized against the housekeeping gene S12. Primers used are described in Table S2 in Text S1.

Microscopy, ALT levels
Brains and livers were fixed in 4% formaldehyde. Histological sections embedded in paraffin were stained with hematoxylineosin-saffron for microscopic evaluations. ALT was measured in serum samples using commercially available kits (Boehringer Mannheim).

Cumulative parasitaemia load, statistical analysis
GraphPad Prism 4.0 software was used to determine the cumulative parasitemia load (Area Under Curve function) and the statistical significance (Two-way ANOVA).

Supporting Information
Text S1 Contains Table S1 (Primers used to generate genetically modified parasites) and Table S2 (Primers used for gene expression analysis in myeloid cells) as supplement to Experimental procedures. The Text S1 contains also Fig. S1 (Schematic primary structure of T. brucei TbKHC1), Fig. S2 (Localization of T. brucei TbKHC1),