Plasmodium co-infection protects against chikungunya virus-induced pathologies

Co-infection with Plasmodium and chikungunya virus (CHIKV) has been reported in humans, but the impact of co-infection on pathogenesis remains unclear. Here, we show that prior exposure to Plasmodium suppresses CHIKV-associated pathologies in mice. Mechanistically, Plasmodium infection induces IFNγ, which reduces viraemia of a subsequent CHIKV infection and suppresses tissue viral load and joint inflammation. Conversely, concomitant infection with both pathogens limits the peak of joint inflammation with no effect on CHIKV viraemia. Reduced peak joint inflammation is regulated by elevated apoptosis of CD4+ T-cells in the lymph nodes and disrupted CXCR3-mediated CD4+ T-cell migration that abolishes their infiltration into the joints. Virus clearance from tissues is delayed in both infection scenarios, and is associated with a disruption of B cell affinity-maturation in the spleen that reduces CHIKV-neutralizing antibody production.


Results
Co-infection prevents severe CHIKV joint inflammation. In this study, we used the well-defined CHIKV joint-footpad mouse model where CHIKV infection alone induces measurable joint swelling that peaks at~6 days post infection (dpi) and lasts~14 dpi, with a viraemic profile of 10-12 dpi 20,21 . We also used two different species of rodent Plasmodium, PbA which causes fatal ECM, and Py17x which induces uncomplicated, non-lethal malaria. To understand the impact of Plasmodium infection on CHIKV-induced pathology, four different CHIKV co-infection scenarios were designed to reflect situations where co-infection of CHIKV and Plasmodium occur concurrently or sequentially [11][12][13][14][15][16][17] . In the first scenario, mice were pre-infected with PbA or Py17x, 4 days before CHIKV infection when mice mount acute Plasmodium-induced responses at the time of co-infection (Fig. 1a, b). In the second scenario, mice were concurrently co-infected with PbA or Py17x and CHIKV (Fig. 1c, d). In the third scenario, animals were infected with PbA or Py17x 4 days after CHIKV infection when mice mount acute CHIKV-induced responses at the time of co-infection ( Supplementary Fig. 1a, b). Lastly, in the fourth scenario, animals were infected with CHIKV 26 days after Py17x infection, when mice had recovered from Py17x-induced malaria ( Supplementary Fig. 1c).
Mice pre-infected (−4 dpi) with lethal PbA or non-lethal Py17x have abolished CHIKV-induced joint swelling and reduced or prevented viral load in the blood throughout the entire course of disease (Fig. 1a, b and Suplemenetary Fig S5). Consistent with previous findings 19 , 80% of the co-infected PbA (−4 dpi) + CHIKV mice succumbed to ECM 6-8 days after parasite infection. As such, data from the PbA (−4 dpi) + CHIKV coinfection scenario were not statistically significant from 4 dpi onwards (i.e. 8 days after parasite infection) (Fig. 1a). Concurrent CHIKV with PbA or Py17x co-infection suppressed peak joint swelling (~50%) with no effect observed for joint swelling or viraemia (Fig. 1c, d). No effects on joint swelling or viraemia were observed in mice infected with PbA or Py17x 4 days after CHIKV infection ( Supplementary Fig. 1a, b) or when mice were infected with CHIKV after recovery from prior Py17x infection (Supplementary Fig. 1c).
Together, pre-and concurrent Plasmodium co-infection protects against CHIKV-induced pathology to different degrees. Importantly, the impact of co-infection on CHIKV pathology was not limited to one Plasmodium species. Thus, all subsequent studies mimicking concurrent Plasmodium and CHIKV coinfection were performed using PbA 19 . Pre-Plasmodium (−4 dpi) and CHIKV co-infection were performed using Py17x due to the high death rate of PbA-infected mice 19 .
Co-infection delays CHIKV resolution in the joint. CHIKV replication persists in the joints for weeks after systemic viral load is resolved 20,22 . To understand the impact of co-infection on virus persistence in the joints, an infectious clone of CHIKV expressing a firefly luciferase marker that allowed tracking of virus dissemination and replication was used 20 .
Interestingly, concurrent CHIKV + PbA co-infection had no effect on early acute CHIKV replication (1-7 dpi) and levels of joint footpad viral RNA (2 dpi), but resulted in higher levels of virus persistence in the inflamed joints from 8-22 dpi (Fig. 2a-c and Supplementary Movie 1) despite similar viraemia levels to CHIKV-infection alone (Fig. 1c). Similarly, resolution of disseminated virus in the tails and contra-lateral footpad was delayed by~5 days in the co-infected mice (Supplementary Movie 1). Follow-up of tissue viral load beyond 22 dpi was not possible as co-infected mice succumb to PbA-induced hyperparasitaemia by 23-24 dpi 19 .
In contrast to concurrent PbA co-infection, pre-Py17x (−4 dpi) + CHIKV co-infection suppressed early CHIKV replication in the tissue from 1-3 dpi and levels of joint footpad viral RNA on 2 dpi (Fig. 2d-f and Supplementary Movie 2). Despite this early phenotype, pre-Py17x (−4 dpi) + CHIKV co-infection led to elevated tissue viral load from 5-18 dpi in the co-infected mice compared to CHIKV-only mice (Fig. 2d, e and Supplementary Movie 2). Viral load within the joints of co-infected mice remained significantly higher until 40 dpi, and virus clearance was delayed from 40 dpi to 45 dpi (Fig. 2g).
Co-infection suppresses acute CHIKV joint pathology. Histopathological assessment was next performed on the joint tissues from CHIKV + PbA (concomitant infection) and Py17x (−4 dpi) + CHIKV infected mice at 6 dpi, when peak joint swelling was suppressed in both models (Fig. 1). Haematoxylin and eosin  Fig. 1 Co-infection suppresses CHIKV-induced joint swelling and viral replication. a Joint inflammation and viraemia of CHIKV (n = 7) and PbA (−4 dpi) + CHIKV (n = 7) groups. Each blue " †" symbol represents one mouse that died of experimental cerebral malaria on the respective day. b Joint inflammation and viraemia of CHIKV (n = 5) and Py17x (−4 dpi) + CHIKV (n = 4) groups. For (a) and (b), Plasmodium infection was given 4 days prior to CHIKV infection, as shown in the schematic. c Joint inflammation and viraemia of CHIKV (n = 7) and CHIKV + PbA (n = 7) groups. d Joint inflammation and viraemia of CHIKV (n = 5) and CHIKV + Py17x (n = 5) groups. For (c) and (d), Plasmodium and CHIKV infection occurred concurrently, as shown in the schematic. All data were analyzed by Mann-Whitney two-tailed test (*P < 0.05, **P < 0.01 and ***P < 0.001), and are representative of two independent experiments. Data represent the means ± standard deviation. Parasitaemia were not altered by co-infection and data has been published in Teo et al. 19 . Abbreviations: CHIKV, Chikungunya virus; dpi, days post infection; PbA, Plasmodium berghei-ANKA; Py17x, Plasmodium yoelii 17× Co-infection delays virus resolution in tissues and suppress acute CHIKV-induced joint pathology. Tissue viral load in a the whole body and b the joints of CHIKV (n = 5) and CHIKV + PbA (n = 6) groups. Follow-up was terminated at 22 dpi as co-infected mice died of hyper-parasitaemia between 23 and 26 dpi. c Footpad viral RNA load on 2 dpi in the joints of CHIKV (n = 5) and CHIKV + PbA (n = 5) groups. Tissue viral load in d the whole body and e the joints of CHIKV (n = 7) and Py17x (−4 dpi) + CHIKV (n = 5) groups. f Footpad viral RNA load on 2 dpi in the joints of CHIKV (n = 5) and Py17x (−4 dpi) + CHIKV (n = 4) groups. g Footpad viral load of CHIKV (n = 6) and Py17x (−4 dpi) + CHIKV (n = 7) groups at 20 dpi, 30 dpi, 40 dpi and 45 dpi. Bioluminescence signal from the luciferase-tagged virus in both groups fell below the detection limit at 45 dpi. Representative pseudo-color images of bioluminescence signal at 20 dpi, 30 dpi, 40 dpi and 45 dpi are shown. Histopathological scoring of (h) oedema in the subcutaneous region and (i) muscle necrosis in the joints of CHIKV, CHIKV + PbA and Py17x (−4 dpi) + CHIKV (n ≥ 4 per group) at 6 dpi. j Total CD45 + cellular infiltration (flow cytometry) and infiltration of inflammatory cells in different regions of the joint (histological grading) of CHIKV, CHIKV + PbA and Py17x (−4 dpi) + CHIKV (n ≥ 4 per group) at 6 dpi. Data in a-g were analyzed by Mann-Whitney two-tailed test and h-j by one-way ANOVA with Tukey post hoc test (*P < 0.05, **P < 0.01 and ***P < 0.001). Data represent the means ± SD. Abbreviations: CHIKV, Chikungunya virus; CoIF, co-infected; dpi, days post infection; PbA, Plasmodium berghei-ANKA; Py17x, Plasmodium yoelii 17× (H&E) staining of the joint-footpad revealed that both scenarios of co-infection reduced subcutaneous oedema and severe muscle necrosis, and prevented inflammatory cell infiltration into the muscles, synovial membrane and subcutaneous region ( Fig. 2h-j and Supplementary Fig. 2). This phenomenon was also observed by the drastic reduction (4-5 fold) of CD45 + leukocyte infiltrates in the joints of co-infected mice when quantified by flow cytometry (Fig. 2j). Joint pathology was then assessed at a late time-point to observe for increased virus persistence in the joint as a result of co-infection. Joints from pre-Py17x (−4 dpi) + CHIKV coinfected mice were harvested for histopathological assessment at 45 dpi, when tissue viral load has been resolved in the co-infected mice (Fig. 2g). At 45 dpi, the acute CHIKV-induced joint phenotypes (extensive subcutaneous oedema, cellular infiltration and muscle necrosis) were resolved ( Supplementary Fig. 3a). Both CHIKV-only and pre-Py17x (−4 dpi) + CHIKV mice displayed equivalent levels of muscle fiber regeneration and synovial hyperplasia at the joint and tendon synovial lining suggesting similar states of joint regeneration despite the longer persistence of a higher viral load in the tissues of co-infected mice ( Fig. 2g and Supplementary Fig. 3).
IFNγ mediates acute CHIKV pathology in pre-Py17x coinfection. We next focused on identifying the mechanisms induced by Plasmodium that modulate CHIKV disease during Plasmodium co-infection. IFNγ is a dominant cytokine typically induced during acute Plasmodium infection 23 , and previous studies have shown that absence of IFNγ aggravates CHIKV viraemia in mice, suggesting a possible anti-viral role 20 . Therefore, the effects of concurrent CHIKV + PbA co-infection and pre-Py17x (−4 dpi) + CHIKV infection were investigated in IFNγ -/mice. In both WT and IFNγ -/mice, concurrent co-infection presented similar pathologies with reduced joint swelling (Supplementary Fig. 4), suggesting that IFNγ is not a mediator of pathological changes to CHIKV disease during concurrent PbA and CHIKV co-infection.
Although tissue viral load of pre-Py17x co-infected mice was similar to that in CHIKV-infected IFNγ -/mice on 1-4 dpi, levels were higher at the later phase (6-20 dpi) ( Fig. 3d-f). The absence of IFNγ only partially increased joint swelling in pre-Py17x coinfected mice at later stages of disease (5-10 dpi) to a similar degree as concurrently co-infected WT mice ( Fig. 1a and Fig. 3a).
To verify these effects observed in pre-Py17x co-infected IFNγ -/mice, we performed pre-Py17x co-infection of WT mice with IFNγ neutralization during the early acute phase (−4 dpi to 4 dpi) of the disease. Similarly, IFNγ neutralization restored early acute (1-4 dpi) CHIKV phenotypes (i.e joint swelling, viraemia, whole body and joint viral load) in the pre-Py17x co-infected mice ( Supplementary Fig. 5).
Co-infection limits CHIKV-neutralizing antibody production. B cells and CHIKV-neutralizing antibodies are essential for virus clearance 24 . As Plasmodium co-infections aggravate tissue viral load only in the later stages of CHIKV infection (Fig. 2), IgM and IgG antibody titers were quantified. At 6 dpi of concurrent PbA + CHIKV co-infection, higher CHIKV-specific IgM titers were observed (Fig. 4a), while IgG titers (Fig. 4b) and neutralizing capacities (Fig. 4c) were similar to mice infected with CHIKV alone. However, lower CHIKV-specific total IgG titers was detected coupled with a reduction in CHIKV neutralization in coinfected mice at 15 dpi (Fig. 4b, c and Supplementary Fig. 6a). In the pre-Py17x (−4 dpi) + CHIKV co-infection scenario, mice showed lower levels of CHIKV-specific IgM and IgG and reduced CHIKV neutralization compared to mice infected with CHIKV alone at both 6 and 15 dpi (Fig. 4d-f and Supplementary Fig. 6b).
Spleen dysregulation alters CHIKV antibodies production. Splenic B-cell response is disrupted during Plasmodium infection due to germinal center (GC) dysregulation [25][26][27][28][29] . To determine if disruption in the spleen is the primary reason for the suppression of CHIKV-specific antibodies in Plasmodium and CHIKV coinfected mice ( Fig. 4a-f), concurrent PbA + CHIKV co-infection was performed in splenectomised mice. Pre-Py17x co-infection was excluded due to the impact of Plasmodium induced IFNγ at the point of CHIKV infection that suppressed early acute CHIKV viral load (Fig. 3).
Interestingly, peak joint swelling at 6-7 dpi was increased by splenectomy in CHIKV-infected mice with no effect on viraemia and tissue viral load ( Fig. 4g-j). Concurrent co-infection reduced peak joint swelling in splenectomised mice as in nonsplenectomised controls ( Fig. 4g; right panel), with no effect on viraemia (Fig. 4h). However, splenectomy restored the ability of co-infected mice to resolve tissue viral load in 8-18 dpi to comparable levels as non-splenectomised CHIKV-infected mice ( Fig. 4i, j). CHIKV-specific IgM, total IgG titers and CHIKV neutralizing antibodies were also similar in both splenectomised CHIKV-infected and co-infected mice ( Supplementary Fig. 7a-c). Nonetheles, splenectomy also reduced parasitaemia in co-infected mice during the later phase (10-18 dpi) of the disease ( Supplementary Fig. 7d). The possible indirect impact of reduced parasitaemia on the tissue viral load in the co-infected splenectomised mice remains to be determined.

Co-infection delays germinal center B-cell generation.
PbA infection impairs the GC response 26,28 that could disrupt production of CHIKV antibodies. By adapting the flow cytometry gating strategy used in a recent study 30 , GC B cells were defined by B220 + IgD − GL7 + CD95 + CD38 − surface markers. Concurrent PbA and CHIKV co-infection suppressed early generation of GC B cells in the spleen at 6 dpi (Supplementary Fig. 8b; blue dots). Interestingly, a significant increase in the B220 + GL7 + CD95 + CD38 + B-cell subset was observed in co-infected mice at 6 dpi, which was not present in CHIKV-infection only controls (Supplementary Fig. 8b). This CD38 + B-cell subset was previously suggested to be either precursors of GC-independent memory B cells or GC B-cell 31 . At 15 dpi, the level of GC B cells in the spleen were similar between CHIKV-infected and co-infected mice ( Supplementary Fig. 8b). However, early reduction of GC B cells at 6 dpi was associated with a decline in GC-dependent CD73 + memory B cells 31 at 15 dpi in co-infected mice ( Supplementary  Fig. 8c).

Co-infection limits CD4 + T-cell infiltration in the joints.
To understand how co-infection alters the joint infiltrating immune subsets, immune cells were profiled by immunohistochemistry (IHC) staining and flow cytometry at 6 dpi during the peak of joint swelling. Both concurrent CHIKV + PbA and pre-Py17x (−4 dpi) + CHIKV co-infections reduced CD3 + T-cell infiltration into the subcutaneous region of the inflamed joints compared to CHIKV-infection alone (Fig. 5a, b). Flow cytometric profiling and ELISPOT assays of isolated CD4 + T cells also showed that Plasmodium co-infection abrogates infiltration of activated (LFA-1 + ) and IFNγ-secreting CHIKV-specific CD4 + T cells into the joints (Fig. 5c-e). Finally, co-infection also suppressed infiltration of other immune subsets, including neutrophils, NK cells and CD11b + Ly6C + monocytes/macrophages into the joints ( Fig. 5f-h).
Lymph node dysregulation alters CHIKV joint swelling. To first determine whether co-infection could impair CD4 + T-cell expansion via the lymph nodes and disrupt infiltration into the joints, LTα -/mice that are devoid of lymph nodes 32 were assessed. Within the context of concurrent co-infection, the spleen was not of primary focus as co-infection in splenectomised Yellow dotted box highlights acute suppression of the viral load in co-infected (CoIF) WT mice that is absence in CoIF IFNγ -/mice. Blue box in a, b, e and f denotes acute phenotypes (1-4 dpi) induced in CoIF WT mice that are ameliorated in CoIF IFNγ -/mice. Mann-Whitney two-tailed test was used to compare CoIF mice against their respective CHIKV-infected or Py17x-infected controls in the same genetic background. Comparisons between WT + CHIKV and WT + Py17x (−4 dpi) + CHIKV are represented by *P < 0.05 and **P < 0.01; comparisons between IFNγ -/-+ CHIKV and IFNγ -/-+ Py17x (−4 dpi) + CHIKV are represented by + P < 0.05 and ++ P < 0.01. Data represent the means ± SD and are representative of two independent experiments. Abbreviations: CHIKV, Chikungunya virus; dpi, days post infection; Py17x, Plasmodium yoelii 17× ; WT, wild type mice still reduced peak joint swelling when compared to CHIKV infected splenectomised mice (Fig. 4g). First, absence of lymph nodes delayed CHIKV-induced peak joint swelling from 6 dpi to 7-8 dpi (Fig. 6a). LTα -/mice with concurrent CHIKV + PbA coinfection showed equivalent joint swelling as CHIKV-infected only LTα -/animals with no effect on viraemia (Fig. 6a, b). Lack of lymph nodes also increased tissue viral load and caused extensive CHIKV dissemination from 6 dpi onwards in CHIKV-infected mice ( Co-infection induces CD4 + T cells apoptosis in the pLN. The popliteal lymph node (pLN) is the nearest lymph node to the site of CHIKV infection in the footpad. As immune regulation leading to suppressed joint swelling is dependent on the lymph node (Fig. 6a), expansion kinetics of CD4 + T cells in the pLN during concurrent co-infection was profiled. At 4 dpi, when pathogenic CD4 + T-cell expansion is observed, CHIKV-specific CD4 + T cells numbers were similar between the pLN of CHIKVinfected and Plasmodium + CHIKV co-infected mice (Fig. 6e). These data imply similar levels of CD4 + T-cell expansion in the early stages of infection (4 dpi).
Interestingly, co-infection reduced the numbers of total and CHIKV-specific CD4 + T cells in the pLN at 5 and 6 dpi (Fig. 6e, f). Plasmodium is known to induce T-cell apoptosis in the secondary lymphoid organs 33,34 . As early CD4 + T-cell expansion (4 dpi) was similar between single and co-infected mice, it is plausible that Plasmodium-induced apoptosis may reduce CD4 + T-cell numbers in the pLN of co-infected mice at 5 and 6 dpi. As expected, co-infection resulted in a higher proportion (~2-3 folds) of late apoptotic CD4 + T cells in the pLN at 5 and 6 dpi (Fig. 6g).
The reduction of CD4 + T cells from 5 dpi in the pLN on joint inflammation would depend on the timing of migration of pathogenic CD4 + T cells to the infected tissue. Therefore, CHIKV-specific CD4 + T-cell migration kinetics was assessed in the joints by flow cytometry and IFNγ ELISPOT assay. The majority of CHIKV specific CD4 + T cells infiltrated the joint between 5 and 6 dpi ( Supplementary Fig. 9), thus occurring after Plasmodium-induced CD4 + T-cell apoptosis. Together, the reduction of peak joint swelling in co-infected mice is associated with the elevation apoptotic of pathogenic CD4 + T cells in the lymph nodes prior to their migration into the joints.

Co-infection limits CXCR3-mediated infiltration into joints.
To assess whether reduced CD4 + T-cell numbers in the joints of co-infected mice could also be mediated by altered migration kinetics from the pLN in addition to increased levels of apoptosis, an in vivo migration assay was performed (Fig. 7a). This assay allows for the detection of any differences in cellular migration independent of the effects of increased Plasmodium-induced apoptosis at 5 dpi.
Concurrent Plasmodium and CHIKV co-infection inhibited migration of donor CD4 + T cells towards the joints of co-infected recipients by~6-fold (Fig. 7b). Furthermore, the increase of joint swelling induced by donor's CD4 + T cells in single CHIKVinfected recipients were absence in co-infected recipients (Fig. 7c). Immune mediators quantified from joint-footpad lysates at 6 dpi demonstrated a reduction of CCR5 ligands (MIP-1α, MIP-1β, RANTES) and CXCR3 ligand (CXCL10) in the joints (Fig. 7d). CHIKV infection in CCR5 -/mice and in WT mice with CXCR3 blockade was then performed to define the role of these two T-cell chemokine receptors. Joint swelling was suppressed at 2 dpi in CHIKV-infected CCR5 -/mice, but no effect was observed during peak swelling (mediated by CD4 + T cells) from 5-8 dpi (Supplementary Fig. 10a). No effect on viraemia was observed in CHIKV-infected CCR5 -/mice ( Supplementary  Fig. 10b). Conversely, CXCR3 blockade suppressed CHIKVinduced peak joint swelling from 5 to 7 dpi and reduced joint vascular leakage at 6 dpi with no effect on viraemia ( Fig. 7g and Supplementary Fig. 10c, d). Importantly, CXCR3 blockade was associated with the inhibition of CD4 + T-cell infiltration into the joints (Supplementary Fig. 10e). However, CXCR3 blockade in concurrent Plasmodium and CHIKV co-infected mice did not significantly reduce joint swelling or vascular leakage at 6 dpi as compared to co-infected mice with no blockade (Fig. 7e, g). CXCR3 blockade has minimal effect on viraemia (Fig. 7f). Together, these results show that CXCR3 and not CCR5, is functionally important for CD4 + T-cell migration during CHIKV infection, and co-infection sufficiently abrogates CXCR3mediated joint swelling.

Discussion
In this study, a CHIKV mouse model that recapitulates the parameters of CHIKV infection in human patients (in terms of viral load profile and joint pathology such as tenosynovitis, myositis, oedema, and cellular infiltration 21,22,24,35,36 ) was used to understand the effects of Plasmodium co-infection on virusinduced pathology. We showed that Plasmodium infection modulates early IFNγ production, B-cell maturation and CD4 + T cell responses toward CHIKV to suppress acute joint pathology and delay tissue viral resolution in the co-infected host.
We found that elevated IFNγ production, induced by prior acute Plasmodium infection, exerts an anti-viral effect, abrogating systemic viral load and suppressing early virus replication in the joint tissue, leading to abolished joint disease. Plasmodiuminduced IFNγ is produced by innate cellular subsets, including NK, NKT cells and γδ T cells 37 . We postulate that acute Plasmodium infection triggers high IFNγ production 23 that could "prime" cells to suppress initial viral replication in CHIKV-target cells such as fibroblasts. This effect would lead to the observed early reduction in viral load locally in the joints. Conversely, viral suppression in the monocytes/macrophages could also limit early viral dissemination to other tissues.
Disruption of germinal center architecture and germinal center B-cell maturation has been documented in experimental mouse and non-human primate models of Plasmodium [25][26][27][28][29] . Disturbance of B-cell homeostasis has been reported in Plasmodiuminfected patients 38 . Consistent with previous studies [25][26][27][28][29] , Plasmodium co-infection delays early (6 dpi) generation of GC B cells (B220 + GL7 + CD95 + CD38 − ), which leads to reduced GCdependent CD73 + memory B cells at a later stage of infection (15 dpi) 31 . This disruption to GC B-cell formation was originally proposed to be mediated by impaired T follicular helper-cell differentiation during Plasmodium infection 26 . Here, we found that changes to the splenic B-cell phenotype is associated with inefficient CHIKV-neutralizing antibody generation, delayed virus resolution in the joints and elevated virus dissemination during chronic stages of CHIKV infection. These changes in viral load in the joints of co-infected mice did not aggravate joint pathology at the earliest time-point of viral resolution in coinfected mice (45 dpi). Nonetheless, such disruption to the B-cell humoral response during co-infection may negatively impact the recall response against secondary CHIKV infection. Pathogenic virus-specific CD4 + T cells are the primary mediator of the peak of CHIKV-induced joint pathology 20,35 . Induction of Th1 cytokines in both CHIKV-infected patients and mice suggests that CD4 + T cells mediate joint pathology through a Th1-mediated mechanism 36,[39][40][41][42][43] . In fact, co-infection suppresses peak joint pathology of CHIKV infection by limiting the infiltration of pathogenic virus-specific IFNγ (Th1)-producing CD4 + T cells in the joints. Here, suppression is mediated in two stages. Plasmodium-infection induces apoptosis of a fraction of CD4 + T cells in the lymph node early after priming, and abrogates migration of the remaining live CD4 + T cells by suppressing chemotactic factors in the joints.
Prior to this study, the chemokines that regulate CD4 + T-cell migration during CHIKV infection were unknown. We show here that CXCR3, but not CCR5, is a critical mediator of CD4 + T-cell migration into infected joints. CXCR3 blockade in co-infected mice did not reduce joint swelling further, suggesting that the CXCR3-mediated pathway is already suppressed in co-infected mice. Suppresion of joint swelling was more pronounced in coinfected mice with or without CXCR3 blockade as compared to CHIKV-infected mice with CXCR3 blockade. This finding suggests that suppression of CHIKV-induced joint swelling in coinfected mice is only partially dependent on CXCR3.
These findings of chikungunya protection during co-infection with Plasmodium have important implications when considering malaria intervention programmes. Effective malaria control programmes have declined the global incidence of malaria over the past decade 10,44 and this is associated with the increasing trend of chikungunya incidence over a similar period of time 8 . Interestingly, abrogation of systemic viral load during coinfection would lower vector competence, as lower blood viral titres would reduce the proportion of vectors being successfully infected 45 . On the other hand, in the context of uncontrolled Plasmodium infection, the reduction of severe joint pathology by malaria co-infection would likely lower or mask CHIKV-related healthcare burden and prevalence. As the World Health Organisation scales up malaria intervention programmes to aggressively decline the incidence of malaria 46 , the severity and numbers of CHIKV clinical cases, and vector competence, could indirectly increase. Therefore, alphavirus management should be integrated as part of the ongoing malaria intervention programme. The orange box denotes suppression of joint inflammation (5-8 dpi) in co-infected WT mice compared to WT + CHIKV, which is lost in the co-infected LTa -/mice. Data from co-infected mice was compared with their respective CHIKV infected controls on the same genetic background using Mann-Whitney two-tailed analysis. Comparisons between WT + CHIKV and WT + CHIKV + PbA are represented by *P < 0.05 and **P < 0.01; comparisons between LTa -/-+ CHIKV and LTa -/-+ CHIKV + PbA are represented by + P < 0.05. Quantification of CHIKV-specific CD4 + T cells (ELISPOT) (e), total CD4 + T cells (flow cytometry) (f) and late apoptotic CD4 + T cells (g) in the pLN of mock, CHIKV and CHIKV + PbA (n ≥ 5 per group) at 2, 4, 5 and 6 dpi. e Representative ELISPOT images showing spots generated by virus-stimulated CD4 + T cells isolated from 100,000 pLN cells (4 and 5 dpi) and 200,000 pLN cells (6 dpi) in CHIKV-infected and co-infected mice. g Representative dot plots displaying late apoptotic CD4 + T cells gating on 6 dpi. Late apoptotic CD4 + T cells are defined as CD45 + CD3 + CD4 + FVD + AnnexinV + cells. Data shown are pooled from four independent experiments. Data (e-g) from the co-infected group were compared against CHIKV-infected controls at the respective time-point using Mann-Whitney two-tailed test (*P < 0.05 and **P < 0. Representative dot plots displaying recovered donor-cell gating in 10,000 CD45 + cells in the footpad. c Joint inflammation of CHIKV (no transfer controls), CHIKV donor transferred to CHIKV recipient, CHIKV + PbA (no transfer controls) and CHIKV donor transferred to CHIKV + PbA recipient (all groups n ≥ 5). d Concentration of MIP-1a, MIP-1b, RANTES and CXCL10 in the joint-footpad cell lysate of mock (n = 5), CHIKV (n = 5) and CHIKV + PbA (n = 6) at 6 dpi. e Joint inflammation and f viraemia of CHIKV + Rat IgG (n = 6), CHIKV + CXCR3Ab (n = 6), CHIKV + PbA + Rat IgG (n = 6) and CHIKV + PbA + CXCR3Ab (n = 8). g Joint vascular leakage for CHIKV + Rat IgG, CHIKV + CXCR3Ab, CHIKV + PbA + Rat IgG and CHIKV + PbA + CXCR3Ab (all groups n ≥ 6). Joint vascular leakage was determined by Evan's blue assay. All data were analyzed by Mann-Whitney two-tailed test (ns; not significant, *P < 0.05 and **P < 0.01). Each data point in the dot plot corresponds one mouse. Data represent the means ± SD. Abbbreviations: CHIKV, Chikungunya virus; dpi, days post infection; ns, not significant; PbA, Plasmodium berghei-ANKA

Methods
Mice. Male and female WT, CCR5 −/− , LTα −/− , and IFNγ −/− mice (aged 5-7 weeks) were maintained on a C57BL/6J background. All experiments were age and gender matched. All mice were bred and housed under specific pathogen-free conditions in the Biological Resource Centre, Agency for Science, Technology and Research, Singapore. All experiments and procedures were approved by the Institutional Animal Care and Use Committee (IACUC: 140968) of the Agency for Science, Technology and Research, Singapore, in accordance with the guidelines of the Agri-Food and Veterinary Authority and the National Advisory Committee for Laboratory Animal Research of Singapore.
Pathogens and infection. The lethal P. berghei ANKA (PbA) clone 15Cy1 was used to induce experimental cerebral malaria (ECM) 47 . The P. yoelii yoelii 17XNL clone 1.1 (Py17x) was used to induce non-lethal malaria 48 . Infected red blood cells (iRBC) were prepared by in vivo serial passage in C57BL/6J mice and blood samples stored in Alsever's solution in liquid nitrogen.
The CHIKV strain (SGP11) was previously isolated from blood samples taken from patients with CHIKV infection, admitted to the National University Hospital Singapore during the 2008 outbreak in Singapore 49 . The CHIKV variant expressing firefly luciferase (CHIKV-luc) or Zs-Green was constructed using a full-length infectious cDNA clone of the viral isolate CHIKV strain LR2006 OPY1 50 . CHIKV was propagated in Aedes albopictus C6/36 cultures (ATCC ® CRL-1660 TM ) and quantified by standard plaque assay in Vero E6 cells (ATCC ® CCL-81 TM ) as plaque forming units (PFU) 20 .
Mice were infected with PbA by intraperitoneal (i.p.) injection of 10 6 iRBC and with CHIKV by subcutaneous inoculation of 10 6 PFU CHIKV (in 30 μl PBS) in the ventral side of the right hind footpad towards the ankle.
CHIKV disease monitoring in mice. Viraemia was monitored daily from 1 day post-infection (dpi) until 8 dpi, and subsequently on every alternate day until 14 dpi. Briefly, 10 μL of blood collected from the tail vein was diluted in 120 μL PBS and 10 μL citrate-phosphate-dextrose solution (Sigma-Aldrich). Viral RNA was extracted with QIAmp Viral RNA kit (Qiagen) in accordance to manufacturer's protocol. To measure the levels of replicating virus the extracted samples, RNA copies of negative sense nsP1 were quantified using quantitative RT-PCR. Primers and probe sequences were adapted from Plaskon et al. 51 to target the same region of the negative sense nsP1 of the strain of CHIKV used in this study. The sequences of nsP1 forward and reverse primers are 5′GGCAGTATCGTGAATTCGATGC-GACACGGAGACGCCAACATT3′ and 5′AATAAATCATAAGTCTGCTCTCTG TCTACATGA3′. The sequence of probe used is 5′TGCTTACACACAGACGT3′. Both primers and probe targeting this region of the negative sense nsP1 has been previously validated to be highly strand specific 51 . To note, both free viruses and viruses present inside peripheral leukocytes were determined. Comparison of viral load using paired whole blood and serum samples shows that a significant portion of viral RNA would have been missed if measurement was only done using plasma or serum ( Supplementary Fig. 11).
Joint swelling of the footpad was scored daily from 0 to 14 dpi 20 . Measurements were done for both height (thickness) and breadth of the foot and were quantified as [height x breadth]. Degree of inflammation was expressed as relative increase in footpad size as compared to pre-infection (day 0) with the following formula: [(x−day 0)/day 0], where x is the quantified footpad measurement for each respective day.
Viral load and dissemination of luciferase-tagged virus (CHIKV-luc) into the organ tissues was tracked using an in vivo bioluminescence imaging system (IVIS; Perkin Elmer) 20 . Luciferase substrate, D-luciferin potassium salt (Caliper Life sciences), was dissolved in PBS at a concentration of 5 mg/ml. Mice were shaved and anesthetized in an oxygen-rich induction chamber with 2% isoflurane. Measurements were performed 2 min after subcutaneous injection of 100 μl of luciferin solution. Whole body imaging was performed with the animal in a ventral position. Foot imaging was performed with the animal in a dorsal position. Bioluminescence imaging was acquired with a field of view (FOV) of 21.7 cm and 13.1 cm for whole body (FOV-D) and foot (FOV-C) respectively. Exposure condition was an initial 60 s, followed by a 4 min delay before another exposure at 60 s. In the event that luminescence readings were above the detection limit of machine, the exposure time was reduced and kept consistent across groups. Bioluminescence signals taken pre-infection (0 dpi) were used for background subtraction. For bioluminescence quantifications, regions of interest were drawn using the software Living Image 3.0, and average radiance (p/s/cm 2 /sr) was determined.
Viral load determination in footpad. Mice were anesthetised with ketaminexylazine (150 mg/kg of ketamine; 10 mg/kg of xylazine), perfused intracardially with PBS and organs were harvested and preserved in 1 mL TRIZOL (Invitrogen) at -80°C. Tissue samples were transferred to gentleMACS M Tubes (Miltenyi Biotec) and homogenized using a rotor-stator homogenizer (Xiril Dispomix) 52 . Homogenized tissues were transferred to clean Eppendorf tubes and mixed thoroughly with 230 μL of chloroform, and allowed to stand for 2 min in room temperature. Tissue mixtures were centrifuged at 12 000 g for 10 min at 4°C. The aqueous phase was collected into a clean Eppendorf tube, and an equal volume of 70% ethanol was added in and mixed. Total RNA was subsequently extracted with RNeasy kit (Qiagen) following manufacturer's protocol. RNA copies of negative sense nsP1 were subsequently quantified using quantitative RT-PCR.
Plasmodium disease monitoring in mice. Parasitaemia was monitored by flow cytometry daily from 3 dpi to 12 dpi and subsequently on alternate days 53 . Briefly, 1 μl of blood was collected in 100 μl of PBS and stained in 1 μL of Hoechst (800 µM; Sigma), 0.5 μl of dihydroethidium (1 μg/ml; Sigma), 0.5 μl of PBS and 2 μl of monoclonal antibody CD45 coupled to Allophycocyanine (Miltenyi Biotech; Cat no. 130-102-544). The blood preparation was incubated for 20 min before addition of 400 μl PBS. Stained samples were acquired on a LSR II flow cytometer (BD Biosciences) using BD FACSDiva software and analyzed using Flowjo software. Plasmodium infected RBCs were determined as Hoechst + /dihydroethidium + /CD45-cells.
Splenectomy. Mice aged 4-weeks old were anesthetised with 80 μl ketamine: xylaxine (150 mg/kg:10 mg/kg). Incisions were made on the left flank, and the primary artery connecting the spleen to the other organs located within the fat was sealed. Post-surgery, enrofloxacin (5 mg/kg) and PBS (1 ml) was given and mice were monitored for recovery. A control, sham operation included similar incisions to the skin, without spleen removal and similar antibiotic treatment post-surgery. 5 weeks were given for mice to completely recover from surgery prior to further experiments, hence infection in these mice were delayed to 9 weeks old.
Histology of joint footpad tissues. Mice were euthanised by perfusion with PBS post anesthesia and tissues were harvested and fixed in 10% neutral buffered formalin, and cut into 5 μm-thick sections for hematoxylin and eosin (H&E) staining 35 . Histopathological examination was carried out in a blinded fashion based on the presence of oedema, inflammation, muscle necrosis, tendonitis, and synovitis in mouse joint. Severity of grading was assigned on the following scale: 1-minimal; 2-mild; 3-moderate, 4-marked & 5-severe as previously described 21 .
Immunohistochemistry (IHC) and image analysis. Immunohistochemical staining of the joint-footpad was performed using a polyclonal α-mouse CD3 antibody (1:200 dilution; Dako; Cat no. A0452), as previously described 35 . Stained sections were scanned using a Leica SCN400 slide scanner (Leica Microsystem GmbH, Germany) at a 20x magnification and exported to the SlidePath Digital Image system (Leica microsystem GmbH, Germany). CD3 expression was analyzed using cytoplasmic/membranous algorithms within the Leica SlidePath Tissue Image Analysis (tissue IA) software. The results among the different groups were computed and analyzed for statistical significance.
Quantification of joint vascular leakage by tracer assay. A Tracer 653 Assay (Molecular Targeting Technology, Inc) was performed as previously described 36 . Briefly, 100 μl tracer solution was injected (i.v) into each mouse, and the tracer solution that entered the joints was quantified using the IVIS. Regions of interest were drawn using Living Image 3.0 software, and average radiance [Log 10 (p/s/cm 2 / sr)/(μW/cm 2 )] was determined.
Leukocyte profiling in the joints, pLN and spleen. Joints, pLN and spleen were surgically extracted and processed to obtain a homogenous cell suspension. For spleens, tissue was dissociated in RPMI medium containing 10% FBS (complete RPMI) and passed through a 40 μm cell strainer (Fisherbrand), followed by RBC lysis with RBC lysis buffer (R&D system). For pLN, extracted tissue was incubated in 1 mL of digestion medium [dispase (2 U/mL; Invitrogen), collagenase IV (20 mg/ mL; Sigma-Aldrich), and DNase I mix (50 mg/mL; Roche Applied Science) in complete RPMI medium] for 30 min at 37°C. Cells were passed through 70 μm nylon mesh cloth (Sefar) followed by RBC lysis with RBC lysis buffer. For footpad, tissue were removed,de-skinned in 4 ml of digestion medium and incubated for 3 h at 37°C. Digested tissues were passed through a 40 μm cell strainer, followed by RBC lysis. Cells were then resuspended in 1 ml of complete RPMI and overlay on 35% v/v Percoll (Sigma-Aldrich)/RPMI medium, centrifuged at 1024 g for 20 min.
In vivo migration assay. At 6 dpi, pLN from singly CHIKV-infected donors were harvested. Total CD4 + T cells were isolated from these donors through negative selection using a CD4 + T-cell isolation kit (Miltenyi Biotec). Cells were subsequently labeled with 5 μM CFSE solution (Sigma-Aldrich) for 10 min, washed with PBS. A solution of 2 × 10 6 cells in 200 μl PBS was injected (i.v.) into a CHIKV only or co-infected recipient at 5 dpi. A portion of these donor cells were stained with α-CD45, α-CD3, α-CD4, α-CD8, and α-LFA-1 antibodies according to dilutions specified above and analyzed by flow cytometry to profile the numbers of total and LFA-1 + CD4 + T cells in 2 × 10 6 donor cells. The footpads were extracted from recipient mice 22 h after transfer, and processed to obtain a homogenous cell suspension. Extracted cells were stained with Live/Dead determination dye (Invitrogen) for 20 min, followed by α-CD45, α-CD3, α-CD4, α-CD8, and α-LFA-1 antibodies. The donor cells were identified by FITC expression using flow cytometry. The migratory capacity of the donor cells towards the joint-footpad was determined by the ratio of donor cells obtained from the recipient footpad divided by initial number of donor cells transferred.
Statistical analysis. Statistical analyses were performed according to the parametric or nonparametric distribution of the data using Prism 6 (GraphPad Software). Normality tests were performed by D'Agostino-Pearson omnibus normality test. All data with a normal distribution were analyzed by unpaired T-test or oneway ANOVA with Tukey's post-hoc test. All data that did not meet the normality requirements were analyzed by Mann-Whitney 2-tailed analysis or Kruskal-Wallis with Dunn's multiple comparisons. P values < 0.05 were considered statistically significant.

Data availability
The data that support the findings of this study are available from the corresponding author upon request.