IFNAR1-Signalling Obstructs ICOS-mediated Humoral Immunity during Non-lethal Blood-Stage Plasmodium Infection

Parasite-specific antibodies protect against blood-stage Plasmodium infection. However, in malaria-endemic regions, it takes many months for naturally-exposed individuals to develop robust humoral immunity. Explanations for this have focused on antigenic variation by Plasmodium, but have considered less whether host production of parasite-specific antibody is sub-optimal. In particular, it is unclear whether host immune factors might limit antibody responses. Here, we explored the effect of Type I Interferon signalling via IFNAR1 on CD4+ T-cell and B-cell responses in two non-lethal murine models of malaria, P. chabaudi chabaudi AS (PcAS) and P. yoelii 17XNL (Py17XNL) infection. Firstly, we demonstrated that CD4+ T-cells and ICOS-signalling were crucial for generating germinal centre (GC) B-cells, plasmablasts and parasite-specific antibodies, and likewise that T follicular helper (Tfh) cell responses relied on B cells. Next, we found that IFNAR1-signalling impeded the resolution of non-lethal blood-stage infection, which was associated with impaired production of parasite-specific IgM and several IgG sub-classes. Consistent with this, GC B-cell formation, Ig-class switching, plasmablast and Tfh differentiation were all impaired by IFNAR1-signalling. IFNAR1-signalling proceeded via conventional dendritic cells, and acted early by limiting activation, proliferation and ICOS expression by CD4+ T-cells, by restricting the localization of activated CD4+ T-cells adjacent to and within B-cell areas of the spleen, and by simultaneously suppressing Th1 and Tfh responses. Finally, IFNAR1-deficiency accelerated humoral immune responses and parasite control by boosting ICOS-signalling. Thus, we provide evidence of a host innate cytokine response that impedes the onset of humoral immunity during experimental malaria.


Introduction
Although robust immunity to malaria is difficult to generate in humans through natural infection or vaccination [1,2], it is nonetheless clear that Plasmodium-specific antibodies offer the best known form of immunological protection against blood-stage parasites [3,4,5,6,7,8], and may also control liver-infective sporozoites [9,10]. Considering that a highly effective malaria vaccine remains elusive, it is important to understand how the onset of humoral immunity to blood-stage Plasmodium parasites is controlled, and whether this process can be boosted, to accelerate or otherwise enhance antibody-mediated immunity to malaria.
Mouse models of resolving, non-lethal blood-stage Plasmodium infection are useful for studying humoral immunity to malaria, since mice fail to control parasitemias and display increased disease severity in the absence of parasite-specific antibodies [4,11,12,13,14]. However, our understanding of how humoral immune responses develop in these models is currently modest. CD4 + T follicular helper (Tfh) cells and their associated cytokines, such as IL-21, and germinal centre (GC) B-cells are critical mediators of humoral immune responses in many systems [15,16], and appear to be similarly important during experimental malaria. For instance, an anti-parasitic role for T-cell-derived IL-21 was recently described during nonlethal Plasmodium chabaudi chabaudi AS (PcAS) infection [6]. Other recent studies using nonlethal Plasmodium yoelii 17XNL (Py17XNL) infection focused on co-stimulatory markers on CD4 + T-cells, and demonstrated that Programmed cell Death 1 (PD-1) and LAG-3 blockade, or stimulation via OX40 boosted Tfh and GC B-cell responses, with positive effects on parasite control [4,17]. With the exception of these reports, in vivo studies of Tfh cells and GC B-cells during experimental malaria remain sparse. Moreover, while these recent reports focused on molecules expressed by CD4 + T-cells themselves, less effort has been directed towards determining whether T-cell extrinsic factors, such as innate or inflammatory cytokines, can control humoral immunity.
It is becoming increasingly clear that inducible T-cell co-stimulatory (ICOS) receptor on CD4 + T-cells is vital for Tfh cell-dependent humoral immunity across numerous model systems [18,19]. ICOS has been implicated in Tfh differentiation via the stabilization of the transcription factor B-cell lymphoma-6 (Bcl-6) [18,20,21]. Importantly, ICOS supports interactions of emerging Tfh cells with ICOS ligand (ICOSL)-expressing bystander B-cells at the periphery of B-cell follicles, a pivotal process for GC B-cell formation and maintenance [22,23]. Moreover, ICOS facilitates the expression of CXCR5, a chemokine receptor essential for Tfh migration into B-cell zones [18,24]. Despite fundamental roles for ICOS on CD4 + Tcells in generating and optimizing B-cell responses and antibody production, its role during blood-stage Plasmodium infection was largely unexplored until recently [25], when Wikenheiser et al. described weaker Tfh and B-cell differentiation in ICOS-deficient mice after the first week of PcAS infection. Furthermore, although T-cell-intrinsic mechanisms have been defined for regulating CD4 + T-cell ICOS levels, for example via Roquin1 and 2 [26,27] and micro-RNA146a [28], whether or not T-cell extrinsic mechanisms can also modulate ICOS is unclear at present.
Type I interferon (IFN-I) signalling can amplify adaptive immune responses [29,30,31], and drive humoral immunity in vivo, particularly in the context of immunization [32,33], viral infection [34,35] and autoimmunity [29,36]. Furthermore, IFN-I-signalling was reported to induce Bcl6, CXCR5 and PD1 expression in naïve CD4 + T cells following TCR stimulation in vitro [37]. IFN-I-related immune responses have also been observed in PBMC from malaria patients [38,39,40]. Although their functional relevance in humans remains to be established, we recently showed in ex vivo cultures of PBMC from P. falciparum-infected humans, that signalling via IFNAR2 was immunoregulatory [41]. In addition, we showed in experimental mice that IFNAR1-signalling, acting via conventional dendritic cells (cDCs) [42], and employing the canonical IFN-I transcription factor, IRF7 but not IRF3 [43], suppressed Th1 responses and parasite control during experimental severe malaria caused by P. berghei ANKA (PbANKA). Recent data has suggested that increased Th1 responses might suppress Tfh cells via IFNγ-signalling in experimental malaria [17], and viral infection [44]. Therefore, the current literature supports a model in which IFNAR1/2-signalling suppresses Th1 responses yet promotes Tfhdependent humoral immunity during blood-stage Plasmodium infection. The aim of this paper was to determine the effect of IFNAR1-signalling on humoral immune responses during experimental malaria.
In this report, we investigated roles for CD4 + T cells, ICOS-and IFNAR1-signalling pathways in the development of humoral immune responses during blood-stage Plasmodium infection. We confirmed crucial roles for CD4 + T-cells and ICOS-signalling in controlling B-cell responses and anti-parasitic immunity. We showed that IFNAR1-signalling obstructed parasite control and antibody production, which was associated with regulation of numerous aspects of the humoral immune response including GC B-cell and plasmablast generation. In particular, IFNAR1-signalling acted early to limit proliferation and localization of activated CD4 + T-cells adjacent to and within B-cell follicles in the spleen. Finally, IFNAR1-deficiency boosted humoral immune responses and improved parasite control in an ICOS-dependent manner. Thus, we describe here the restrictive effect of an innate cytokine-signalling pathway on antibody-mediated immunity during experimental blood-stage malaria.

GC B-cell and plasmablast differentiation requires CD4 + T-cells and ICOS-signalling during blood-stage Plasmodium infection
CD4 + T-cells are critical for control and resolution of blood-stage Plasmodium infection [4,11,45], a phenomenon we first confirmed in Py17XNL-infected WT mice depleted of CD4 + cells (Fig 1A & 1B). Despite this, to our knowledge there remained no direct evidence that (C) Flow cytometric gating strategy employed to analyze splenic B-cell responses throughout the CD4 + T cells promoted B-cell responses during experimental malaria. To examine this, wildtype (WT) mice depleted of CD4 + cells or given control IgG, were infected with Py17XNL and examined for resulting splenic GC B-cell and plasmablast responses (Fig 1C). By day 10 postinfection (p.i), a timepoint just before the CD4-depleted mice began to succumb to infection, plasmablast differentiation was 80% lower in CD4-depleted mice compared to infected controls (Fig 1D), and GC B-cell formation (Fig 1E) was almost abrogated (95% reduction compared to controls), with a similar, substantial impairment in Ig-class switching ( Fig 1F). Together, these data formally demonstrated that GC B-cell and plasmablast generation was highly dependent upon CD4 + T-cells.
While previous studies in mice and humans demonstrated that ICOS expressed on CD4 + T-cells was critical for effective humoral responses [18,19,23,46,47], until recently, no such studies had been performed during Plasmodium infection [25]. Therefore, we first examined ICOS expression by CD4 + T-cells during Py17XNL infection (Fig 2A), revealing significant up-regulation by 5-7 days p.i. (Fig 2B), which progressively increased over the following 7-9 days. To examine a possible functional role for ICOS in the development of humoral immunity, WT mice were treated with α-ICOSL-blocking antibody (α-ICOSL) during Py17XNLinfection. Consistent with other experimental models [18,19,48], ICOSL-blockade impaired GC B-cell formation (Fig 2C), Ig-class switching (Fig 2D), Tfh differentiation (Fig 2E), and early production by day 16 p.i. of parasite-specific total IgG and IgG2b, but not IgM or IgG3 in the serum (Fig 2F). Day 16 p.i. was an important timepoint during Py17XNL infection because it marked the point at which the rate of increase in parasitemia began to slow in WT mice, thus indicating the beginning of the resolution phase of infection. Finally, ICOSLblockade over the first three weeks of infection exacerbated parasitemias and delayed resolution of infection ( Fig 2G). Taken together, these data indicated that ICOS-signalling promoted CD4 + T-cell dependent humoral immune responses and parasite control during Py17XNLinfection.

IFNAR1-signalling impairs B-cell and antibody responses during bloodstage Plasmodium infection
We next examined the impact of IFNAR1-signalling on parasite control and humoral immune responses during Py17XNL-infection. Ifnar1 -/mice displayed similar initial parasitemias compared to infected WT controls for the first two weeks of infection, but thereafter exhibited faster control of blood-stage parasites than WT controls ( Fig 3A). Similar effects were also observed during PcAS-infection (S1A Fig). Next, we noted increased parasite-specific IgM and total IgG levels in the sera of Py17XNL-infected Ifnar1 -/mice compared to WT controls at day 16 p.i. (Fig 3B & 3C). More specifically, levels of parasite-specific IgG1, IgG2b, IgG2c and IgG3 were all higher in Ifnar1 -/mice compared to WT controls (Fig 3B & 3C). Next, we noted that GC B-cell ( Fig 3D) and Ig-switched B-cell generation (Fig 3E) was limited by IFNAR1-signalling at day 16 p.i. In addition, IFNAR1-signalling impaired plasmablast formation at day 6 p.i. (Fig 3F). We also observed similar regulation of plasmablasts and emerging GC B-cells during PcAS-infection (S1B & S1C Fig). Finally, we explored the longer-term effect of IFNAR1-signalling on parasite-specific antibody production. At day 25 p.i. with Py17XNL, Ifnar1 -/mice maintained higher serum IgG levels, including IgG2b and IgG3, but not IgM, compared to WT controls ( Fig 3G). By day 40 p.i., total IgG, IgG2b and IgG3 concentrations had risen further to similar levels in WT and Ifnar1 -/mice, while IgM levels had dropped, again with no differences between groups (Fig 3G). Taken together, our data indicated that IFNAR1-signalling delayed parasite control, B-cell responses and the onset of antibody production during blood-stage Plasmodium infection.  Given that GC B-cell and plasmablast formation was dependent on CD4 + T cells (Fig 1D-1F), and likewise that Tfh differentiation depended on the presence of B-cells (S2 Fig), we next examined the impact of IFNAR1-signalling on splenic CD4 + T-cell responses (Fig 4). Compared to infected WT controls, Ifnar1 -/mice displayed increased Tfh proportions and numbers during Py17XNL infection (Fig 4A), and increased proportions during PcAS infection ( Fig  4B). We chose to assess emerging Tfh cells around day 6-8 p.i. during PcAS infection, since our previous work had suggested that strong CD4 + T helper cell responses were detectable around this time [43,49]. Enhanced Tfh differentiation was associated in both models with a substantial increase by 6 days p.i., in ICOS expression by splenic CD4 + T-cells (Fig 5A & 5B). ICOS expression by CD4 + T cells facilitates interaction with ICOSL-expressing B-cells at the periphery of B-cell zones, and sustains Tfh cells within B-cell follicles [19,22,23]. Therefore, we examined the impact of IFNAR1-signalling on CD4 + T-cell localization within the spleen ( Fig  5C). At day 5 p.i. we observed higher densities of ICOS + T-cells at the T/B border in Ifnar1 -/mice compared to WT controls (Fig 5C), with a similar although more modest effect within Bcell follicles themselves ( Fig 5C). Therefore, our data suggested that IFNAR1-signalling limited Tfh cell differentiation and the localization of activated CD4 + T-cells adjacent to and within Bcell follicles.
To rule out possible developmental or immune homeostatic defects in Ifnar1 -/mice accounting for phenomena described above, we next treated PcAS-infected WT mice with an IFNAR1-blocking antibody [42,43]. IFNAR1-blockade enhanced ICOS expression on CD4 + T-cells (Fig 6A), boosted Tfh cell responses but not Bcl-6 expression ( Fig 6B), and increased early plasmablast and GC B-cell formation (Fig 6C & 6D). Together, these data supported the hypothesis that IFNAR1-signalling regulated ICOS expression by CD4 + T-cells, limited Tfh differentiation, and restricted splenic B-cell responses.

IFNAR1-signalling limits both Th1 and Tfh differentiation
Elevated IFNγ responses during blood-stage malaria have been associated with impaired humoral immune responses [17,50,51]. Since IFNAR1-signalling had suppressed IFNγ production by Th1 cells in our previous reports using PcAS and PbANKA infection [42,43], we next determined the concurrent effect of IFNAR1-signalling on Th1 and Tfh differentiation. On day 6 p.i. with PcAS, as expected, Ifnar1 -/mice exhibited increased Th1 differentiation compared to WT controls [43] (Fig 7A). Importantly, emerging Tfh responses at day 6 p.i. were also higher in Ifnar1 -/mice compared to WT controls ( Fig 7B). This indicated that IFNAR1-signalling had simultaneously limited both Th1 and Tfh cell formation, rather than skewed CD4 + T cell responses to either helper subset. Moreover, IFNAR1-mediated regulation of ICOS expression in CD4 + T cells was observed in both the Th1 and emerging Tfh compartments during PcAS infection ( Fig 7C); and IFNAR1-signalling also reduced ICOS + Tfh cell numbers during Py17XNL infection ( Fig 7D).
Next, given the simultaneous effect of IFNAR1-signalling on Th1 and Tfh differentiation, we examined whether IFNAR1-signalling exerted a generalized effect on CD4 + T-cells. At day 6 p.i. during PcAS infection, the proportion of CD4 + T-cells that were proliferating, as assessed by Ki-67-staining ( Fig 7E), was modestly higher in Ifnar1 -/mice compared to WT controls. Similarly, in α-IFNAR1-treated Py17XNL-infected mice at day 6 p.i., CD4 + T-cells were more proliferative compared to isotype-treated controls ( Fig 7F). We also examined the effect of IFNAR1-signalling on CD4 + T-cell activation using the markers CD11a and CD49d [4,41]. At day 8 p.i. with PcAS, the proportion of CD4 + T-cells co-expressing CD11a and CD49d was modestly higher in Ifnar1 -/mice compared to WT controls, although absolute numbers in the spleen were not different (S3A Fig). We also studied Py17XNL-infected mice at day 6 p.i. after treatment with α-Ifnar1. We noted again that CD4 + T-cells were more activated in α-Ifnar1treated mice compared to isotype-treated controls (S3B Fig). Taken together, our data suggested that IFNAR1-signalling limited early CD4 + T-cell activation and proliferation.
Finally, we observed no substantial differences in the cellularity of the spleen, or bulk CD4 + T cell or B-cell numbers, in WT versus Ifnar1 -/mice during PcAS infection (Fig 7G), Py17XNL infection (Fig 7H), or in un-infected mice (S3C Fig). These data support the idea that any increases in Th1 or Tfh cells observed in our studies were not due to generalized increases in spleen cellularity, but instead were associated with specific regulation of CD4 + T cell activation and proliferation. Taken together, our data revealed that IFNAR1-signalling simultaneously regulated Th1 and Tfh cell formation, which was associated with restricted CD4 + T-cell activation, proliferation and ICOS expression.

IFNAR1-signalling via cDCs limits Tfh cell and GC B-cell responses
We next determined cell types in which IFNAR1-signalling had occurred, and focused on cDCs given our previous work using lethal PbANKA infection [42]. We generated mixed bone marrow chimeric mice as before [42], which harboured equal proportions of WT and Ifnar1 -/splenic cDCs (Fig 8A). These mice were infected with Py17XNL, and expression of co-stimulatory molecules was assessed at the peak of the cDC response (at 2 days p.i.) on splenic WT and Ifnar1 -/-cDC subsets (Fig 8B). Up-regulation of CD86 in particular, which we previously found was dependent on IFNAR1-signalling in cDC [42], was again substantially impaired in Ifnar1 -/-CD8 + and CD8 -cDC subsets compared to WT cDCs (Fig 8C). A similar pattern of expression was seen for PD-L1, while we observed minimal changes in ICOSL expression (Fig 8C). Interestingly, we also observed cDC subset-specific effects, with PD-L2 restrained by IFNAR1-signalling in CD8 -cDC subsets, and CD40 expression mediated by IFNAR1-signalling in CD8 + Taken together, our data demonstrated that similar to lethal PbANKA infection [42], IFNAR1-signalling in splenic cDC subsets influenced their upregulation of co-stimulatory molecules during non-lethal blood-stage Plasmodium infection.  To study the possibility of IFNAR1-signalling in B-cells, we next examined early plasmablast responses in WT:Ifnar1 -/mixed bone marrow chimeric mice during PcAS infection. At day 7 p.i., there was no consistent increase in plasmablast formation by Ifnar1 -/cells compared to WT counterparts (Fig 8D), suggesting that IFNAR1-signalling in B-cells played no major role in regulating plasmablast differentiation. Moreover, in these same mice equal proportions of WT and Ifnar1 -/-CD4 + T cells upregulated ICOS and CXCR5 (Fig 8E). This suggested that restriction of ICOS expression in emerging Tfh cells was not mediated by IFNAR1-signalling directly to CD4 + T cells.
Finally, we explored in vivo the effect of IFNAR1-signalling in cDC on emerging humoral responses in the spleen. CD11cCre +/-ifnar1 f/f mice and CD11cCre -/-ifnar1 f/f littermate controls were infected with Py17XNL, and splenic Tfh and GC B-cell differentiation assessed 6 days p.i. (Fig 8F and 8G). We noted that IFNAR1-signalling in CD11c hi cells limited both Tfh ( Fig 8F) and early GC B-cell (Fig 8G) differentiation. Taken together, our data strongly suggested that regulation of humoral immune responses by IFNAR1-signalling was mediated through cDCs, not B-cells or CD4 + T cells.

IFNAR1-deficiency boosts humoral immunity in an ICOS-dependent manner
Finally, we sought to explore molecular mechanisms by which IFNAR1-deficiency accelerated the onset of humoral immunity, and hypothesized a role for ICOS-signalling. To test this, we adopted a recent approach in which α-ICOSL blocking antibody was used to reduce ICOS-signalling in mice exhibiting higher than WT levels of ICOS [28]. We treated PcAS-infected Ifnar1 -/and WT mice with a moderate dose of α-ICOSL blocking antibody (100μg) and at day 8 p.i. examined antibody production and parasitemia. We saw no effect on early parasite-specific IgM production or parasitemias in WT mice (Fig 9A & 9B), with some impairment of emerging parasite-specific IgG responses (Fig 9A & 9B). In contrast, ICOS-signalling blockade in Ifnar1 -/mice abrogated any improvements in parasite-specific IgM or IgG production ( Fig  9A), and importantly, impaired parasite control compared to control-treated Ifnar1 -/mice ( Fig 9B). Consistent with these observations, we also noted that enhanced Tfh cell and GC Bcell formation in Ifnar1 -/mice was strongly dependent on ICOS-signalling (Fig 9C & 9D). Finally, ICOS blockade in Ifnar1 -/mice significantly reduced serum IFNγ levels compared to control-treated Ifnar1 -/mice (S5 Fig), consistent with enhanced ICOS-signalling during IFNAR1-deficiency being supportive of Th1 differentiation. Taken together, our data support a model in which IFNAR1-signalling limited humoral immune responses and parasite control by regulating ICOS-signalling.

Discussion
Here, using two mouse models of non-lethal blood-stage malaria, we have provided evidence that the onset of protective humoral immunity to Plasmodium can be influenced by an innate cytokine signalling pathway, in this case Type I Interferon-signalling via IFNAR1. Therefore,  we have demonstrated for the first time that innate immune cytokines can limit the onset of antibody production and B-cell responses to Plasmodium. Moreover, we found that release from IFNAR1-mediated immune-regulation enhanced humoral immune responses and parasite control in a manner dependent upon ICOS-signalling.
Although parasite-specific antibodies can control blood-stage Plasmodium parasite numbers in vivo, protective humoral immunity can take months or years to develop in humans. Reasons for this have focused on parasite mechanisms such as antigenic variation, rather than possible sub-optimal host immune responses. IFN-I responses have been well documented in malaria patients [38,52,53]. Polymorphisms in the Ifnar1 gene have been associated with reduced risk of severe malaria [38,52]. Although the location of polymorphisms did not indicate the direction of their effect on IFNAR1-signalling, the implication was that changes in IFNAR1-signalling could mediate improved parasite control. Indeed, our recent work using C57BL/6J mice suggested that IFNAR1-signalling via IRF7 but not IRF3 limited parasite control during PcAS infection [42,43], although other recent work, using a different route of infection [54] or a different genetic background [55] suggested more modest roles for IFN-Isignalling in mice. These different outcomes suggest possible context-dependent effects for IFN-I-signalling during PcAS infection.
It is likely that mouse models of malaria will be informative for studies of cytokine-mediated effects on humoral immunity, as recently epitomized in studies of T cell-derived IL-21 [6]. These reports and our new data complement studies of T-cell co-stimulatory molecules, such as PD-1 and LAG-3, and strengthen the idea that multiple molecular targets could be harnessed to boost humoral immunity to malaria [4,17]. Whether such strategies are more applicable to natural exposure events or vaccination scenarios remains to be studied. In addition, while we found no overt evidence of increased immune-pathology in Ifnar1 -/or α-Ifnar1 treated mice, it will be important to better test whether manipulation of IFNAR1-signalling triggers unwanted adverse events, as previously reported for IL-10-or IL-27-deficiency [56,57].
Much of our current understanding of cytokine-mediated control of humoral immunity derives from studies of viral infection or experimental immunization in mice [33,58,59,60,61]. However, it remains difficult to infer from these studies how cytokine-signalling impacts upon antibody responses during parasitic infection. In a recent report, Perez-Mazliah et al., demonstrated a pivotal role for T cell-derived IL-21 in mediating GC B-cell responses and IgG class switching but not for generating Tfh cells during blood-stage Plasmodium infection [6]. Our demonstration that IFNAR1-signalling restricts the onset of humoral immunity to malaria is the first description of cytokine-mediated suppression of Tfh, GC B-cell and plasmablast formation during parasitic infection. More generally, while the majority of viral and experimental immunization studies highlight a positive role for IFNAR1-signalling in driving humoral immunity, our data emphasizes that depending on the experimental context, IFNAR1-signalling can also limit humoral immune responses. Our previous studies using a lethal model of malaria demonstrated that IFNAR1-signalling occurred via cDC, which resulted in potent suppression of Th1-immunity, and was associated with effects on PDL1, PDL2 and IL-10 expression by cDC subsets [42]. However, the lethal model was not suitable for studying humoral immunity, since mice became moribund within the first week of infection. Here, using non-  lethal models, we similarly revealed that IFNAR1-signalling proceeding via cDCs, but not B-or T-cells, regulated the generation of Tfh and GC B-cell responses. Given that cDCs are critical for early Tfh differentiation and initiation of humoral immune responses [18,62,63], we propose that IFNAR1-signalling within splenic cDC regulates CD4 + T-cell proliferation, ICOS expression, and Th1 and Tfh differentiation.
One question that remains to be answered is how the interaction between cDC and CD4 + T cells is altered at a molecular level by IFNAR1-signalling. This work and our previous study using PbANKA infection [42] both revealed multiple changes in the co-stimulatory landscape on the surface of cDC, including a substantial shift in the ratio of PDL1:PDL2 expression on CD8 -cDC. Given a recent report that PDL2-signalling can compete against the regulatory effects of PDL1, and indeed, can protect against experimental malaria [64], we speculate that IFNAR1-signalling in cDC regulates CD4 + T cell activation by favouring PDL1-signalling over protective PDL2 signals.
It was recently reported during LCMV infection that IFN-I promoted Th1 responses, which then suppressed Tfh and GC B-cell responses via IFNγ [44]. A more recent report using a mouse model of malaria demonstrated that therapeutic release from PD-1 exhaustion, coupled with stimulation via OX40 dramatically increased IFNγ production by Th1 cells, which destabilized Bcl-6 in established Tfh cells and resulted in defective humoral immune responses [17]. Elsewhere, circulating 'Th1-like' Tfh cells were associated with impaired humoral responses in children living in malaria endemic areas [50]. Most recently, Ryg-Cornejo et al., implicated a combined effect of IFNγ and TNF in driving sub-optimal humoral responses during severe malaria infections in mice [51]. Together, these data suggest that Th1 responses might interfere with Tfh responses. However, in our studies, we found that deficiency in IFNAR1-signalling triggered a concurrent increase in both Th1 and Tfh responses. This apparent discrepancy between our work and recent studies could be explained by the differential kinetics of the elevated Th1 responses in our respective studies. For instance, OX40-and PD1-targeted therapy was initiated from the second week of infection, a period of time after which Th1/Tfh priming would likely have occurred [17]. In our studies, elevated Th1 responses, as a result of IFNAR1deficiency, occurred relatively transiently within the first week of infection. Therefore, we speculate that elevated Th1 responses in IFNAR1-deficient mice did not destabilize Tfh responses because they occurred early and were not prolonged.
Data from this and our previous studies [42,43,49] support a model in which abrogation of IFNAR1-signalling has the dual effect of boosting Th1 responses and antibody-production. These effects associated with improved control of Py17XNL, PcAS and PbANKA parasites; yet crucially we have not yet demonstrated a causal link between IFNAR1-deficiency and improved antibody-mediated parasite control. Nevertheless, we propose that depending upon host genetic background and that of the infecting Plasmodium species, IFNAR1-signalling can obstruct parasite control via a number of mechanisms including the regulation of Th1 cells and/or antibody production.
Numerous studies across a range of experimental systems have established a pivotal role for ICOS signalling in CD4 + T-cells in mediating T-cell dependent humoral immunity, via effects on Tfh generation, maintenance and trafficking [18,19,23,65]. The importance of ICOS-signalling is further highlighted by the existence of multiple layers of regulation within the T-cell for limiting its expression, for example via Roquin 1, Roquin 2 and microRNA146a [26, 27,28]. To date, however, evidence of T-cell extrinsic mechanisms for controlling ICOS levels on CD4 + Tcells has been limited. Our data reveals the existence of a cytokine signalling pathway, mediated by IFNAR1 which serves to limit the level of ICOS on CD4 + T-cells. Currently, the mechanism by which ICOS levels are modulated in our models by IFNAR1-signalling is unclear, but could theoretically involve Roquin1, Roquin 2 or microRNA146a.
Although we observed substantial early ICOS expression by activated CD4 + T-cells in our models, we noted minimal change in ICOSL levels on cDC (Fig 8C). This might suggest that boosting ICOS expression by CD4 + T-cells did not encourage further interaction with splenic cDC. Instead, given that IFNAR1-deficiency increased the frequency of ICOS + T-cells close to and within B-cell follicles, which were themselves essential for supporting Tfh responses (S2 Fig), and since ICOSL-expressing B-cells are located at the periphery of B-cell follicles [22], we propose that IFNAR1-signalling limits ICOS-mediated positioning of emerging Tfh cells adjacent to and within B-cell follicles during Plasmodium infection. However, further experiments will be required to examine the effect of ICOS-signalling on CD4 + T-cell trafficking in the spleen during experimental malaria.
In this study, we discovered that IFNAR1-deficiency accelerated early production of parasite-specific IgM and IgG. If such mechanisms could be induced in humans, this might improve control of parasite numbers and prevention of clinical malaria. However, the duration of this effect would require scrutiny, since in our mouse models, antibody levels normalized between Ifnar1 -/and WT mice by the seventh week of infection. Furthermore, whether acceleration in early antibody production would increase the rate of acquisition of immunity to clinical malaria is also unclear, particularly given the phenomenon of antigenic variation. However, we speculate that if subsequent infections were sufficiently similar from an antigenic perspective to previously encountered parasites, short-term elevations in parasite-specific antibodies could be beneficial. Finally, although we focused on studying the magnitude of parasitespecific antibody responses within a given antibody sub-class, neither antibody affinity nor avidity was examined. Therefore, further experimentation will be needed to determine whether beneficial changes in antibody affinity and avidity can be brought about via cytokine modulation.
In summary, we have demonstrated here that early cytokine-signalling during infection influences parasite-specific antibody production, and associated GC B-cell, plasmablast and Tfh differentiation in two models of non-lethal blood-stage malaria. Our work suggests that antibody-mediated immunity to malaria might be improved by targeting cytokine-signalling pathways, particularly in the context of natural infection.

Experimental Procedures Ethics statement
All animal procedures were approved by the QIMR Berghofer Medical Research Institute Animal Ethics Committee, under approval numbers A02-633M and A1503-601M, in accordance with the "Australian Code of Practice for the Care and Use of Animals for Scientific Purposes" (Australian National Health and Medical Research Council).

Parasites, infections and parasitemia
Non-lethal Plasmodium yoelii 17XNL (Py17XNL) and Plasmodium chaubadi chaubadi AS (PcAS) parasites were used following one in vivo passage in wild type C57BL/6J mice. Using parasitized red blood cells (pRBC) that were obtained from frozen/thawed stabilates, mice were infected i.v. with either 10 4 pRBCs (Py17XNL) or 10 5 pRBCs (PcAS). Blood parasitemia was measured in Diff-Quick (Lab Aids, Narrabeen, NSW, Australia) or giemsa-stained thin blood smears obtained from tail bleeds. Alternatively, a modified protocol of a previously established flow cytometric method was employed to measure parasitemia more rapidly [66]. Briefly, a single drop of blood, from a tail bleed or cardiac puncture, was diluted and mixed in 250μl RPMI containing 5U/ml heparin sulphate. Diluted blood was simultaneously stained with Syto84 (5μM; Life Technologies) to detect RNA/DNA, and Hoechst33342 (10μg/ml; Sigma) to detect DNA, for 30 minutes, in the dark at room temperature. Staining was quenched with 10x volume of ice cold RPMI, and samples were immediately analysed by flow cytometry, using a BD FACSCantoII analyser (BD Biosciences) and FlowJo software (Treestar, CA, USA). pRBC were readily detected as being Hoechst33342 + Syto84 + , with white blood cells excluded on the basis of size, granularity and much higher Hoechst33342/Syto84 staining compared to pRBC.

Preparation of crude parasite antigen
Crude antigen extract from Py17XNL or PcAS-infected RBC was prepared using an adapted version of a previously described protocol [67,68]. Briefly, mice were infected with Py17XNL or PcAS as described above. When parasitemias reached 20-30%, blood was collected by cardiac puncture into heparinized tubes. RBCs were washed once in RPMI at 1200rpm for seven minutes at room temperature, and then lysed using ultrapure water followed by four washes in ice-cold PBS at 16,000xg for 25 minutes at 4°C, as well as three cycles of freezing (two hours at -80°C) and thawing (30 minutes at room temperature). Extracts were also processed from RBCs of uninfected C57BL/6J mice, for use as negative controls in ELISA. The concentration of proteins in the purified extracts was determined by Bradford assay (Thermo Scientific). All extracts were stored at -80°C until use.

Detection of parasite-specific serum antibodies by ELISA
Costar EIA/RIA 96-well flat bottom plates were coated overnight at 4°C with 2.5μg of soluble antigen/ml in bicarbonate coating buffer (pH9.6). Wells were washed three times (all washes in 0.005% Tween in PBS) and then blocked for 1hr at 37°C with 1% BSA in PBS. Wells were washed three times, 100ul of sera diluted 1/400, 1/800, 1/1600 or 1/3200 was added and incubated for 1hr at 37°C. Following six washes, wells were incubated in the dark with biotinylated anti-IgM, total IgG, IgG1, IgG2b and IgG3 (Jackson ImmunoResearch) for 1hr at room temperature. Unbound antibodies were washed off (six times) prior to incubating wells in the dark with streptavidin HRP (BD pharmagen) for 30 minutes at room temperature. Wells were washed six times prior to development (100μl, OPD; Sigma-Aldrich) for five minutes in the dark before termination with an equal volume of 1M HCl. Absorbance was determined at 492nm using a Biotek synergy H4 ELISA plate reader (Biotek, USA). Data were analysed using Gen5 software (version 2) and GraphPad Prism (version 6).

In vivo cell depletion and blockade using monoclonal antibodies
Anti-CD4 depleting monoclonal antibody (clone GK5.1) and its isotype control were administered in 0.1mg doses, via intravenous (i.v.) injection in 200μl 0.9% NaCl (Baxter) one day before infection. CD4 + T-cell depletion was confirmed in PBMC over the first four days of infection, and absolute numbers subsequently remained >99% depleted in the spleen at day 10 p.i. compared to isotype-treated controls. For ICOSL blockade in Fig 2A-2F, α-ICOSL (clone HK5.3, BioXCell) and its isotype control (IgG2a, clone 2A3, BioXCell) were administered in 0.2mg doses, via i.v. injection in 200μl 0.9% NaCl (Baxter) one day prior to infection, and then every three days for up to 15 days p.i. (and for up to 21 days in Fig 2G). For ICOSL blockade experiments in Fig 9 and S5 Fig, α-ICOSL (clone MIL5733) and its isotype control (IgG2a, clone 1D10) were generated in-house and administered in 0.1mg doses, via i.v. injection in 200μl 0.9% NaCl (Baxter) one day prior to infection, and then every two days for up to 6 days p.i. (In Fig 9, the 100μg dose was employed since preliminary experiments indicated that 200μg reduced Tfh cell responses by~65%, 100μg reduced Tfh responses by 50%, and 25μg had no effect by day 8 p.i. with PcAS). For B-cell depletion, anti-CD20 (clone 5D2) or its isotype control antibody were produced and kindly provided by Genentech and administered in a single 0.5mg dose via i.p. injection in 200μl 0.9% NaCl (Baxter), five days prior to infection. For Ifnar1 blockade, α-Ifnar1 blocking monoclonal antibody (clone MAR1-5A3; Leinco Technologies Inc) and its isotype control mAb were administered in 0.1mg doses, via i.p. injection in 200μl 0.9% NaCl (Baxter) on the day of infection, and subsequently on days 2, 4 and 6 p.i.

Confocal microscopy analysis
Confocal microscopy was performed on 10-20 μm frozen spleen sections as previously described [71,72]. Briefly, tissues from infected and un-infected mouse spleens were snap frozen in embedding optimal cutting temperature (OCT) medium (Sakura) and stored at -80°C until use. Sections were fixed in ice-cold acetone for 10 minutes prior to labelling with antibodies against CD3-Biotin (clone-17A2), B220-PE (clone-RA3-6B2) and ICOS-APC (clone-C398.4A). Anti-CD3 was detected by streptavidin conjugated to Alexa Fluor 594. All antibodies were obtained from Biolegend (San Diego, CA). DAPI was used to aid visualization of white pulp areas. Samples were imaged on a Zeiss 780-NLO laser-scanning confocal microscope (Carl Zeiss Microimaging) and data analysed using Imaris image analysis software, version 8.1.2 (Bitplane). Cells were identified using the spots function in Imaris, with thresholds <10μm and intensities <150. T-B borders were defined by the region between CD3 + cells closest to the B cell follicle and B220 + cells furthest into the T-cell zone. All objects were manually inspected for accuracy before data were plotted and analyzed in GraphPad Prism (version 6)

Statistical analysis
Comparisons between two groups were performed using non-parametric Mann-Whitney (unpaired datasets) or Wilcoxon (paired datasets) tests. Where depicted, one-way or two-way ANOVA and Tukey's post-test were employed for multiple comparisons among three or more groups. p< 0.05 was considered significant (p<0.05 = Ã ; p<0.01 = ÃÃ ; p<0.001 = ÃÃÃ ; P<0.0001 = ÃÃÃÃ ). Survival graphs were assessed using Log-rank (Mantel-Cox) tests. Graphs depict mean values ± SEM, except where individual mouse data points are depicted, in which case median values are shown. All statistical analyses were performed using GraphPad Prism v6 or v7 software.