Background
Immunoregulation is an essential part of the immune response and ensures that a comprehensive and protective response is elicited, but with limited damage to the host. Immunoregulation can be achieved by check-point proteins that either enhance or inhibit cell reactivity [
1]. Inappropriate expression of these proteins can therefore have detrimental consequences for immune responses to infection and also affect pathogenesis. Thus, blocking of check-point proteins to restore functional capacity of effector cells has been explored as potential immunotherapy for chronic viral infections and cancers [
2‐
8]. The expression of immunoregulatory proteins on conventional T cells has been recently shown in both malaria rodent infection models and in malaria-infected individuals, where inhibitory therapies of immunoregulatory proteins resulted in enhanced parasite clearance [
9‐
13]. However, the effect of continuous malaria exposure on immunoregulation among the
innate cell compartment remains a critically under-investigated aspect of malaria immunology.
γδ T cells are a subset of T cells that express a distinct T cell receptor (TCR). These cells are considered to be part of the innate/intermediate immune system due to their ability to respond rapidly to non-peptide antigens without the requirement of major histocompatibility complex (MHC) presentation. Substantial evidence indicates that γδ T cells mediate essential protection against a number of pathogens including
Plasmodium [
14‐
19] where
Plasmodium stimulation of γδ T cells involves metabolites of the 1-deoxy-
d-xylulose 5-phosphate (DOXP) pathway [
20]. While immunity to malaria requires a multifaceted network of cell interactions and cytokine production involving both innate and adaptive immune responses, γδ T cells have been shown to contribute to key processes associated with beneficial outcomes [
21,
22]. Mouse studies demonstrate that the frequency of γδ T cells is significantly increased during malaria infection and they provide protective immunity via interferon gamma (IFN-γ) production and control of parasitemia [
21,
23,
24]. Similarly, γδ T cells are an important early source of IFN-γ in malaria-infected individuals, which is associated with reduced risk of clinical disease [
25‐
30]. Furthermore, inhibition of intracellular parasite growth and granulysin-dependent cytotoxic activity against released blood stage merozoites have been demonstrated [
14,
31‐
33]. However, in addition to a protective role during malaria, γδ T cells were also suggested to contribute to pathogenesis. This is supported by observations that depletion of γδ T cells protected mice from developing cerebral malaria in a
P. berghei ANKA mouse model [
34] and that γδ T cells were found to be one of the predominant sources of cytokines and chemokines associated with severe malaria in malaria-infected individuals [
29]. Although numerous studies describe activation of γδ T cells in response to malaria, the understanding of how these cells are regulated is remarkably limited.
T-cell immunoglobulin domain and mucin domain 3 (TIM3) is a relatively recently described immunoregulatory protein that belongs to the TIM protein family. In humans this family consists of TIM1, TIM3, and TIM4, whereas mice have an additional protein, TIM2 [
35‐
38]. TIM3 is expressed by T cells, dendritic cells (DCs), natural killer (NK) cells, and monocytes and interacts with galectin-9 [
39]. TIM3 is generally referred to as a negative regulator, but TIM3 expression can affect different functions in the innate and the adaptive immune system and on different cells. In mice, engagement of TIM3 on conventional αβ T cells results in apoptosis and loss of effector T cells [
39], whereas TIM3 expression on human T cells is associated with functional exhaustion [
4,
39‐
41]. In contrast, TIM3 is thought to be a maturation marker on human NK cells [
42] and is also abundantly expressed on monocytes regulating cytokine production in these cells [
43,
44]. Recently TIM3 was observed to be upregulated in mice during acute
Plasmodium infection [
13,
45]. TIM3 was found to be expressed by conventional T cells and NK cells, and in vivo blocking of TIM3 resulted in enhanced parasite clearance [
13]. Furthermore, TIM3-expressing CD4+ and CD8+ αβ T cells were observed in individuals during acute
P. vivax infection but were undetectable following treatment [
46].
In contrast, the effect of TIM3 expression on γδ T cells has only recently started to receive attention [
47,
48]. In malaria, the biological relevance of TIM3 expression on γδ T cell function for clinical outcomes has not previously been investigated. Recent findings suggest that dysfunctional Vδ2 γδ T cells associated with malaria exposure induce tolerance to the
Plasmodium parasite [
47]. However, the precise immunological processes responsible for γδ T cell dysregulation remain elusive. Here, we specifically investigated associations between TIM3 and γδ T cell function during malaria as well as the factors that govern TIM3 upregulation. The roles TIM3 expression plays in the control of pathogenic mechanisms were also explored. Our main findings revealed that interleukin (IL)-12 in synergy with IL-18 are key factors required for TIM3 induction. Moreover, TIM3 expression renders cells functionally impaired, which is associated with reduced risk of clinical malaria. These findings provide novel insights into immune-specific processes involved in γδ T cell regulation and represent a major advancement in the field of γδ T cell biology.
Methods
Study site and subjects
Human blood samples from children aged 5–10 years were collected in a clinical trial (ClinicalTrials.gov registration: NCT02143934) conducted in 2009 and 2010 in five villages in East Sepik Province of Papua New Guinea (PNG), where both
P. falciparum and
P. vivax are endemic [
49]. Children were randomized into two treatment groups of directly observed treatment (DOT) over a total of 27 days. The first group of children received chloroquine (CQ, days 1–3 of DOT), artemether-lumefantrine (Coartem®) (AL, days 11–13 of DOT), and primaquine (PQ, days 1–20 of DOT; 0.5 mg/kg per dose). The second group of children received CQ (days 1–3 of DOT), AL (days 11–13 of DOT), and a placebo (days 1–20 of DOT). The drug treatment was implemented to be able to quantify the contribution of
P. vivax and
P. ovale relapses to infection and disease during follow-up in an epidemiological study of the cohort [
49]. The first treatment regime was designed to clear all parasites including
P. vivax hypnozoite stages, whereas the second treatment regime cleared only blood stages. Venous bleeds and peripheral blood mononuclear cell (PBMC) isolation were performed following completion of drug treatment, and PBMCs were cryopreserved.
Children were actively followed for 8 months with finger-prick (250-μl) blood samples collected every 2 weeks for the first 12 weeks and every 4 weeks for the remainder of the follow-up period. In addition passive surveillance measures were implemented at local health centers, aid posts, and via the village health volunteer network. Febrile children were tested with a rapid diagnostic test (RDT), and a blood slide was collected. Symptomatic infections (those with fever and who tested positive by RDT and/or light microscopy) during follow-up were treated with AL. For RDT-negative children, the slides were read within 12 h. If the slides were positive, the children were treated the next day. If the slide was negative, the result was recorded but no further action taken. The collected blood samples were screened for infection with
Plasmodium spp. by light microscopy and quantitative real-time PCR (qPCR). Slides were scored as light microscopy-positive for an individual
Plasmodium species if the species was detected independently by at least two microscopists and/or if subsequent qPCR diagnosis confirmed the presence of the species. Slide discrepancies were adjudicated by a World Health Organization (WHO)-certified level 1 (expert) microscopist [
49]. A generic qPCR was used to detect all
Plasmodium species occurring in PNG
, followed by subsequent species-specific qPCRs on
Plasmodium-positive samples [
50,
51].
A subset of PBMC samples from the children enrolled in the clinical trial (n = 132, of which n = 63 belonged to the primaquine drug-treated group and n = 69 belonged to the placebo group) were included in the current study. These children all had at least one Plasmodium falciparum infection verified by PCR during follow-up to ensure ongoing exposure. Of these, 50 individuals had a clinical episode during follow-up. A clinical episode of malaria was defined as febrile illness (axillary temperature of ≥37.5 °C, current or previous 48 h) plus the presence of P. vivax or P. falciparum parasites (any density) by light microscopy. PBMCs collected from 20 healthy blood donors by the Australian Red Cross were used as controls.
Mouse infections
Female C57BL/6 mice aged 6–8 weeks were infected with 5 × 10
4
Plasmodium chabaudi-infected red blood cells (iRBCs) intravenously or with 1 × 10
6
P. berghei ANKA iRBCs intraperitoneally.
P. chabaudi-infected mice were drug-treated on day 14 post-infection with CQ (6 μg/ml)- and pyrimethamine (70 μg/ml)-containing water for 5 days. Drug treatment on day 14 coincided with control of the infection and allowed for a whole infection cycle to be completed before drug treatment.
P. berghei-infected mice were treated at day 5 post-infection to avoid progression to cerebral malaria. Drug treatment consisted of an intraperitoneal injection of CQ (10 mg/kg) and pyrimethamine (10 mg/kg) followed by CQ- and pyrimethamine-containing water for 5 days as described previously [
52]. Livers and spleens were removed at different time points following completion of drug treatment. Untreated
P. chabaudi-infected mice establish a submicroscopic chronic infection with intermittent detectable parasitemia peaks. Mice that were chronically infected with
P. chabaudi were left untreated until day 98 after the initial infection. Mice infected multiple times (three consecutive infections (
P. chabaudi only) or two consecutive infections (
P. chabaudi and
P. berghei)) were drug-treated and then re-infected on day 14 post-completion of drug treatment to allow the immune cells to return to steady state (Additional file
1: Figure S1).
Parasite lines and cultures
Parasite lines 3D7 and XIE were cultured in human red blood cells, and trophozoite-stage parasites were isolated as described previously [
29,
53]. XIE parasites were snap frozen in 15% glycerol in phosphate-buffered saline (PBS) and thawed by sequential addition of 12%, 1.8%, and 0.9% NaCl and subsequently used for stimulation of cohort samples.
Flow cytometry
PBMCs (5 × 10
5) were stained with antibody cocktails in FACS buffer (PBS containing 0.5% bovine serum albumin and 2 mM ethylenediaminetetraacetic acid) on ice for 30 min. The human antibodies used were fluorescein isothiocyanate (FITC)-conjugated anti-γδTCR (clone 11 F2, BD Biosciences, San Jose, CA, USA), Qdot 605-conjugated anti-CD27-, Qdot 655-conjugated Streptavidin (Invitrogen, Carlsbad, CA, USA), PE-Texas Red (ECD)-conjugated anti-CD3 (clone UCHT1, Beckman Coulter, Brea, CA, USA), Brilliant Violet 421-conjugated anti-CD16-(clone 3G8), Brilliant Violet 711-conjugated anti-CD45RA (clone HI100, both from Biolegend, San Diego, CA, USA), and phycoerythrin (PE)-conjugated anti-TIM3 (clone FAB2365P from R&D Systems Minneapolis, MN, USA). The mouse antibodies used were FITC-comjugated anti-CD3 (clone 145-2C11), PE-conjugated anti-TIM3 (clone RMT3-23), and PerCPCy5.5-conjugated anti-γδTCR (clone GL3, all from Biolegend). Aqua amine reactive dye (Invitrogen) was used for dead cell exclusion. Samples were analyzed on a four-laser Fortessa flow cytometer (BD Biosciences). Data analysis was performed using FlowJo software (Tree Star, Ashland, OR, USA). Boolean gating was utilized to evaluate multiparametric expression, and fluorescence minus one (FMO) controls were used to set gates. The gating strategy is illustrated in Additional file
2: Figure S2.
Intracellular cytokine staining
PBMCs (2 × 105 cells/well in triplicate) were stimulated with either uninfected RBCs (uRBCs) or iRBCs (6 × 105/well) for 24 h or isopentenyl pyrophosphate (IPP, 3 μM, Sigma-Aldrich, St Louis, MO, USA) for 16 h. Brefeldin A (10 μg/ml, Sigma-Aldrich) and monensin (BD Biosciences) were added to the cells for the last 8 h of incubation. Assessment of γδ T cell cytokine production was performed by intracellular cytokine staining using allophycocyanin (APC)-conjugated anti-IFN-γ (clone B27, BD Biosciences) and PE-Cy7-conjugated anti-tumor necrosis factor alpha (TNF-α) (clone MAb11, eBioscience, San Diego, CA, USA), and cytotoxic capacity was assessed by Brilliant Violet 421-conjugated anti-CD107a (clone H4A3, Biolegend) staining in culture. A positive response was determined as the frequency of responding cells, which was twice above background and was ≥0.1% IFN-γ, TNF-α, or CD107a-positive γδ T cells of all γδ T cells or ≥0.5% positive γδ T cells of γδ T cell subsets following subtraction of background.
Cytokine and antigen stimulation of PBMCs
PBMCs (5 × 105 cells) from healthy individuals were stimulated with the following conditions: IL-6 (10 ng/ml), IFN-γ (10 ng/ml), IL-12/IL-18 (50 ng/ml each), IL-4 (10 ng/ml), IL-1β (1 ng/ml, all from Peprotech, NJ), TNF-α (10 ng/ml, Life Technologies, Carlsbad, CA, USA), iRBCs (3 iRBCs: 1 PBMC), lipopolysaccharide (LPS, 1 ug/ml, InvivoGen, San Diego, CA, USA), IPP (3 μM), or cells in medium only. After 24 h of incubation, the frequency of TIM3+ γδ T cells was assessed by flow cytometry.
Enzyme-linked immunosorbent assay (ELISA)
Plasma was assessed for IL-12p70 and IL-18 cytokine levels using ELISA (RayBiotech, Norcross, GA, USA) according to the manufacturer’s instructions. Plasma from healthy controls were included as negative controls for IL-12 and to measure the baseline plasma IL-18 concentration of healthy individuals.
Statistical analysis
Statistical analyses were performed using Prism 6.0 (GraphPad software) and STATA 12. Flow cytometry data were analyzed using the Student’s
t test or Kruskal-Wallis test followed by the Dunn post-test as indicated. Correlation coefficients were determined by Spearman rank correlation. Logistic regression was used to test whether recent infection was associated with an increased proportion of TIM3+ γδ T cells and whether TIM3 expression varied the odds of experiencing a clinical malaria episode during the follow-up period. Associations between molecular force of infection (
molFOI) and the proportion of TIM3-expressing cells were tested using negative binomial regression and in a Cox proportional hazards model for time to first clinical episode. In order to normalize the TIM3 expression levels, the data were power transformed. Clinical incidence was defined as the frequency of occurrence per time at risk of infections associated with a fever.
molFOI was determined by genotyping all samples for merozoite surface protein 2 (
msp2) using capillary electrophoresis for fragment sizing [
54,
55] in addition to using PCR conditions for highly purified DNA [
56].
molFOI was defined as the frequency of acquisition of new malaria infections per time at risk [
55]. New infections were defined as those where the detected allele had not been observed in a child at the previous two active or passive case detection visits [
49]. Time at risk was adjusted for missed visits, and children were censored from the analysis after they missed three or more active case detection visits. Multivariable analyses to determine the association of TIM3 levels and relevant covariates, including IL-18 levels, were conducted using general linear models (GLMs). Backwards elimination was applied to construct the most parsimonious model.
Discussion
The significant contribution of γδ T cells to the overall immunity during infection and cancer is increasingly appreciated. Although recent efforts to utilize γδ T cells as immunotherapy effector cells have produced promising results [
63], it is evident that a more comprehensive understanding of the biology related to immunoregulation of these cells is required to overcome the demonstrated induction of anergy and exhaustion upon repeat exposure to antigens [
64,
65]. Our data demonstrated that in an infectious disease setting with continuous exposure to malaria, TIM3 expression becomes upregulated on γδ T cells, and this process is controlled by environmental cues provided by the host immune response. IFN-γ, TNF-α, and cytotoxic activity were found to be absent in TIM3+ γδ T cells upon re-stimulation with malaria antigens, and increased frequencies of these cells were associated with reduced risk of clinical malaria.
Although
P. vivax is considered less virulent than
P. falciparum,
P. vivax is still a major cause of morbidity in endemic areas. Nevertheless, substantial differences in immunity to the two species have been noted, in particular in regard to naturally acquired immunity, where immunity to
P. falciparum is slower to develop than immunity to
P. vivax [
66]
. Few studies have compared innate responses and cytokine profiles between these species, though dissimilarities are plausible considering the distinct naturally acquired immunity patterns. We observed that recent infection with either
P. vivax or
P. falciparum affected the frequency of TIM3+ γδ T cells. However, the extent of TIM3 upregulation was different depending on the infective species. CD4+ T cells upregulate TIM3 following extended stimulation with IL-12 [
67]. Similarly, we also found that IL-12/IL-18 can promote γδ T cell upregulation of TIM3 in vitro where IL-18 plays an auxiliary role. While both
Plasmodium species are associated with increases in IL-12 production during acute infection,
P. falciparum infection has been demonstrated to result in higher plasma IL-12 levels during convalescence [
68,
69]. In the current cohort IL-12 was undetectable in the plasma, which is likely due to the absence of active infection for 28–30 days preceding the time point for PBMC collection. However, IL-18 plasma levels were readily detectable and were associated with TIM3 expression in
P. falciparum-infected children at enrollment. Thus, dissimilar cytokine profiles and concentration levels at the time of infection may contribute to the difference in TIM3+ γδ T cell frequencies observed between children recently infected with
P. vivax and
P. falciparum. Notably, IL-12/IL-18 can also induce IFN-γ production by γδ T cells. However, the effect of IL-12/IL-18 on TIM3 induction by γδ T cells may be dose-dependent, as was previously shown for TIM3 expression on CD4+ T cells [
67]. Thus, early production of IL-12/IL-18 may promote IFN-γ production, whereas accumulation of these cytokines in the plasma may induce γδ T cells to upregulate TIM3. Furthermore, while no association was observed between IL-18 plasma levels in children who were not infected with
P. falciparum and TIM3 expression, these children still have an increased population of TIM3+ γδ T cells. It is possible that phosphoantigen contributes to the observed TIM3 expression in these children, as IPP was also observed to induce TIM3 expression, albeit at lower frequency. However this remains to be determined.
Experimental mice infected with different
Plasmodium species resulted in significant TIM3 upregulation, and TIM3+ γδ T cells remained detectable after resolution of infection. Comparable findings were observed in the PNG children where TIM3 expression was associated with
molFOI, thus suggesting that exposure to distinct parasites is important for induction of TIM3 during repeated exposure. Cytokine responses to malaria have been reported to be influenced by the immune status of the host [
70,
71]. It is possible that the differences in cytokine profiles resulting from re-infection with the same parasite among immune hosts versus infection with a new clone or species may affect TIM3 expression in the innate cell compartment.
Continuous exposure to malaria is correlated with immunity to symptomatic disease and is likely to involve both antiparasitic mechanisms and regulation of cytokines implicated in immunopathogenesis [
72]. γδ T cells play a protective role as a major IFN-γ producer during malaria infection [
73] but are also a major source of cytokines and chemokines associated with disease [
29]. Typically, Vγ9+ Vδ2+ T cells are considered to be the malaria-antigen responsive cells and also represent the majority of γδ T cells in peripheral blood [
31,
74]. However, the functional roles for γδ T cells in general are continuously expanding, indicating that the contribution of other subsets of γδ T cells to both malaria immunity and immunopathogenesis may not be completely appreciated. Therefore, in contrast with the previous studies, we investigated TIM3 expression on the total γδ T cell population without distinguishing subsets based on TCR restriction. In the current study, we found that the TIM3+ γδ T cells from malaria-exposed individuals were effectively unresponsive to stimulation in vitro. The cells produced minimal IFN-γ and TNF-α and demonstrated low cytotoxic activity in response to iRBCs or phosphoantigen, thus indicating that these cells were functionally impaired. Interestingly, we observed that TIM3 was predominately expressed by CD16+ TEMRA γδ T cells. Our study is aligned with recent findings suggesting that dysfunctional CD16+ γδ T cells emerge in response to malaria exposure [
75]. While CD16 upregulation is most likely a consequence of prior TCR activation [
76], expression of CD16 alone does not explain why these cells are impaired. Here we identified TIM3 as a potential receptor responsible for the γδ T cell impairment associated with malaria infection.
It is worth noting that the CD16+ TEMRA γδ T cell subset is reported to be unresponsive to phosphoantigen stimulation [
61]. Instead CD16 allows γδ T cells to recognize opsonized targets. Farrington et al. (2016) proposed that accumulation of CD16+ cells represents a population which is preferentially stimulated through this receptor independently of the TCR [
75]. However, signaling through CD16 results in both TNF-α and IFN-γ production [
61,
76]. Given that TNF-α is a pyrogenic cytokine that significantly contributes towards malaria febrile disease [
77,
78], it indicates that inhibition of CD16+ TEMRA γδ T cells may be necessary to limit immunopathogenesis in the host. Notably, this study identifies specifically that the TIM3+ CD16+ TEMRA γδ T cell population was associated with reduced clinical incidence risk, which supports the concept that regulation of highly specialized subsets is important for reducing clinical malaria symptoms. Thus, increased TIM3 expression on this population was associated with less risk of febrile malaria and was associated with asymptomatic infections. While TIM3 may not directly inhibit CD16 signaling, a functional linkage between CD16 and TCR signaling has been reported [
79]. In αβ T cells, TIM3 inhibition of TCR signaling is regulated by Bat3 interaction potentially through binding of catalytically active Lck [
80]. Thus, TIM3 inhibition of the TCR may also affect the CD16-dependent response in this γδ T cell population, although this remains to be determined.
While we detected TIM3+ γδ T cells in the mouse model, in vivo assessment of their relative contribution to disease outcome using either a TIM3 knock-out mouse model or TIM3 depletion or blocking is hampered by the fact that TIM3 is known to be expressed by several different cell populations. These approaches would result in the inability to specifically assign effects on disease outcome to TIM3+ γδ T cells. Thus, a conditional knock-out mouse model would be required to address this. Furthermore, adoptive transfer of TIM3+ γδ T cells is potentially confounded by the presence of bound antibody, which may interfere with endogenous ligand interaction. Consequently, experimentally defining the role of TIM3+ γδ T cells for malaria disease outcome in vivo still remains unresolved.
Acknowledgements
We wish to thank the children and guardians for their participation in the study. We would like to acknowledge the staff at the Albinama Health Centre and Papua New Guinea Institute of Medical Research staff for their support and assistance. We acknowledge the efforts of the PNG Institute of Medical Research - Maprik field, administration, microscopy, and laboratory staff, in particular, Benson Kiniboro, Lawrence Rare, Danga Mark, and Heather Huaupe. We thank Ingrid Felger, Natalie E. Hofmann, and Rahel Wampfler at Swiss TPH as well as Andreea Waltmann and Jessica Brewster at Walter and Eliza Hall Institute of Medical Research (WEHI) for the molecular parasitology. We thank Amandine B. Carmagnac and Liana Mackiewicz at WEHI for technical assistance.