Therapeutic blockade of PD-L1 and LAG-3 rapidly clears established blood-stage Plasmodium infection.

Infection of erythrocytes with Plasmodium species induces clinical malaria. Parasite-specific CD4(+) T cells correlate with lower parasite burdens and severity of human malaria and are needed to control blood-stage infection in mice. However, the characteristics of CD4(+) T cells that determine protection or parasite persistence remain unknown. Here we show that infection of humans with Plasmodium falciparum resulted in higher expression of the inhibitory receptor PD-1 associated with T cell dysfunction. In vivo blockade of the PD-1 ligand PD-L1 and the inhibitory receptor LAG-3 restored CD4(+) T cell function, amplified the number of follicular helper T cells and germinal-center B cells and plasmablasts, enhanced protective antibodies and rapidly cleared blood-stage malaria in mice. Thus, chronic malaria drives specific T cell dysfunction, and proper function can be restored by inhibitory therapies to enhance parasite control.

Introduction

Infection of red blood cells by Plasmodium species induces clinical malaria, a devastating global health problem that has been exacerbated by emergence of drug resistant parasites[1, 2]. Thus, new approaches to combat malaria, such as efficacious vaccines or other immune interventions, are desperately needed. Given the clear correlation between high parasite density and disease severity in children[3], much effort has gone into developing vaccination approaches that target the blood-stage of Plasmodium infection with the goal of reducing parasite burden and transmission. However, success has been limited and candidate subunit vaccines in clinical trials have thus far not proven highly efficacious[4, 5], although recent studies with killed blood-stage parasites and specific adjuvant show promise in mouse models[6]. One reason for the limited progress in anti-malarial vaccination likely relates to our incomplete understanding of how the parasite can evade adaptive immunity and the specific characteristics of cellular immune responses that can mediate protection against blood-stage Plasmodium infection. While it is well understood from both clinical human correlates[7-9], and experimental rodent models[10-13] that CD4+ T cells are a critical component of protective immune responses that arise following exposure to blood-stage Plasmodium parasites, very little is known about how Plasmodium-specific CD4+ T cell responses influence the balance between parasite clearance versus persistent blood-stage infection. Additionally, whether or how Plasmodium blood-stage infection influences the development of CD4+ T follicular helper cell responses, with subsequent and direct effects on humoral immunity, remains undefined.

In humans that survive Plasmodium falciparum infection without treatment, parasites can be detected in the blood for several weeks or months[14] and can also establish a chronic-relapsing blood-stage infection that can persist for years[15-17]. The former scenario is mimicked in mouse models by P. yoelii, which establishes patent infections lasting 30 days in immunocompetent hosts, whereas the latter is mimicked by P. chaubadi, which can establish persistent, subpatent infections lasting for several months[18]. Importantly, chronic infection of humans with viruses such as HIV or HCV drives the functional ‘exhaustion’ of anti-viral T cells[19-21], a concept first revealed through studies of CD8+ T cells in mice chronically infected with lymphocytic choriomeningitis virus (LCMV) clone 13 (ref. [22]). In the murine LCMV model, repeated antigen stimulation through the T cell receptor (TCR) drives the sustained expression of T cell inhibitory receptors including programmed death-1 (PD-1) and lymphocyte activation gene-3 (LAG-3) on virus-specific CD8+ T cells. Sustained signaling via these inhibitory receptors directly and indirectly induces transcriptional changes that negatively regulate proliferation and pro-inflammatory cytokine expression by virus-specific CD8+ T cells[23, 24]. Based on these collective observations, we tested the hypothesis that humans exposed to P. falciparum would harbor CD4+ T cells that exhibit phenotypic characteristics of T cell exhaustion, and that therapeutic blockade of T cell inhibitory receptor signaling in vivo would markedly improve clinical outcomes in models of rodent malaria.

Results

Plasmodium infection induces T cell exhaustion

To identify potential relationships between P. falciparum infection and exhaustion of circulating CD4+ T cells, we focused on a cohort study in Mali where the malaria season is intense and seasonal[25] and occurs during each six-month rainy period from July through December. Study participants consisted of children aged five to eleven years who presented as blood smear negative for P. falciparum at the end of the dry season and again seven days after the diagnosis and treatment of symptomatic P. falciparum infection (Before Malaria and After Malaria, respectively, ). Consistent with our hypothesis, we observed elevated percentages of PD-1 expressing CD4+ T cells in children after P. falciparum infection ( and ), suggesting that P. falciparum infection is associated with PD-1 T cell inhibitory receptor expression on CD4+ T cells in individuals presenting with clinical malaria.

To address the biological relevance of these data we turned to mouse models of blood-stage malaria. Initially we focused on the prolonged (>30 day) blood-stage infection induced by injection of mice with parasitized red blood cells (pRBC) harboring the normally non-lethal P. yoelii (Py) strain 17XNL. T cell depletion studies strongly suggested a critical role for CD4+ T cells in survival following Py pRBC challenge (). Additionally Aicda–/– mice, which harbor an addition deletion of the Igh μ secretory domain that contain mature, IgM+ B cells but cannot undergo isotype switching or secrete antibodies (hereafter called Aicda–/– μs–/–), also succumbed to Py blood-stage infection (). Although not conclusive, these data are consistent with a critical role for secreted antibody in survival following Py pRBC challenge. Of note, the paucity of identified epitopes has hampered efforts to define the precise characteristics of CD4+ T cells that either determine protection or correlate with persistent blood-stage Plasmodium infection. We recently applied a surrogate activation marker approach to evaluate the total CD8+ T cell response to attenuated whole sporozoite vaccines[26]. Importantly, this approach permits tracking of the total CD8+ T cell response to infection or vaccination in the absence of information about MHC restriction, epitopes or antigens[27]. To test our current hypothesis, we applied a modified surrogate activation marker approach, which relies on the coordinate upregulation of CD49d and CD11a on antigen-experienced CD4+ T cells, to directly identify Plasmodium-specific CD4+ T cells responding to blood-stage infection. This approach has been extensively validated for virus infections[28] and was confirmed by examining CD49d and CD11a expression patterns on LCMV- or Listeria monocytogenes (LM)-induced CD4+ T cell responses of known specificity. Indeed, all LCMV and LM epitope-specific, interferon-γ (IFN-γ) expressing CD4+ T cells exhibited the CD49dhiCD11ahi cell surface phenotype (). Of note, because not all CD4+ T cells responding to LCMV or LM were specific for the two examined dominant epitopes (GP61-80 and LLO190-201, respectively), and some antigen-specific CD4+ T cells may express cytokines other than IFN-γ, not all CD49dhiCD11ahi T cells expressed IFN-γ in these assays ().

Naïve C57BL/6 mice exhibited a small fraction (~1%) of CD49dhiCD11ahi circulating CD4+ T cells prior to infection, however infection of C57BL/6 mice with Py-parasitized red blood cells (Py pRBC) resulted in the appearance of a large population (>20%) of circulating CD49dhiCD11ahi CD4+ T cells by day 7 post-infection (p.i.) (). Importantly, CD4+ T cell expression of the CD49dhiCD11ahi phenotype required TCR crosslinking and was not a result of malaria infection-induced inflammation, because naïve transgenic CD4+ T cells specific for LCMV (GP61-80-specific, SMARTA CD4+ T cells)[29] did not upregulate CD49d or CD11a expression during P. yoelii blood-stage infection (). To address the specificity of these markers at the memory phase (>60 days post-Py infection), we adoptively transferred purified Py-specific CD49dhiCD11ahi memory CD4+ T cells, or CD49dloCD11alo naïve CD4+ T cells (both obtained >70 days p.i. from mice that had cleared Py infection) into new naïve, allelically marked recipients. Importantly, Py infection of recipient mice resulted in secondary T cell expansion only from transferred CD49dhiCD11ahi memory cells, but not CD49dloCD11alo naïve CD4+ T cells (). Thus, the CD49dhiCD11ahi phenotype faithfully and durably identifies Plasmodium-specific effector and memory CD4+ T cells.

Using this approach we determined that Py infection of mice results in substantial CD4+ T cell responses, and these cells exhibited sustained proliferation over the course of the 30-day period of patent blood-stage infection (). By day 30 p.i, we also observed that a large fraction (>25-40%) of Py-specific, CD49dhiCD11ahi CD4+ T cells expressed the T cell inhibitory receptors PD-1 and LAG-3, similar to CD4+ T cells responding to chronic, but not resolved, LCMV infection (). Of note, the fraction of CD4+ T cells expressing PD-1 in mice with clinical malaria (~3.75%) is in line with our observations in P. falciparum-infected humans, where we observed ~3.5% of all CD4+ T cells, on average, expressed PD-1 (). Furthermore, and consistent with their exhausted phenotype, Py-specific CD4+ T cells also exhibited impaired cytokine (IFN-γ, tumor necrosis factor (TNF) and interleukin 2, IL-2) production in response to phorbol ester (PMA) plus ionomycin stimulation, relative to virus-specific CD4+ T cells from mice either acutely or chronically infected with LCMV on day 31 p.i. (). Additionally, parasite-specific CD8+ T cells (identified by modulation of surface CD8α and CD11a[27]) also exhibited sustained proliferation throughout the infection and both the phenotypic and functional characteristics of T cell exhaustion at day 30 p.i. (). Collectively, these data show that prolonged Plasmodium blood-stage infection elicits dysfunctional parasite-specific T cells including the critical CD4+ T cell subset required for eventual clearance of infection.

Prolonged infection leads to T cell exhaustion

To directly evaluate whether prolonged Plasmodium blood-stage infection was responsible for the appearance of dysfunctional populations of T cells in mice with clinical malaria, we treated groups of P. yoelii infection-matched mice with injections of chloroquine, or vehicle control (PBS), on days 8 and 9 post-infection. Two successive chloroquine treatments resulted in rapid and substantial reductions in parasite burden without complete parasite clearance (data not shown). We found that ~60% of ex vivo-stimulated, parasite-specific CD49dhiCD11ahi CD4+ T cell express IFN-γ at day 8 p.i. (). The capacity to produce IFN-γ was reduced to ~15% of parasite-specific CD4+ T cells by d 24 p.i. in PBS control mice, however chloroquine treatment partially restored IFN-γ production (). Importantly, chloroquine treatment also partially reversed the exhausted phenotype at day 24 p.i., with ~40% and ~70% reductions in the fraction of parasite-specific CD4+ T cells expressing PD-1 or LAG-3 (). Similar to CD4+ T cells, chloroquine treatment reduced the fraction of CD8+ T cells expressing inhibitory receptors, as well as increased the fraction of CD8+ T cells capable of cytokine production (). Of note, these data also reveal that distinct phenotypic differences exist between CD4+ and CD8+ T cells undergoing functional exhaustion during chronic infection, as upregulation of the inhibitory receptors CD160 and 2B4 was only observed on exhausted CD8+ T cells ( and data not shown). Collectively, these data highlight that sustained high parasite burdens in mice with prolonged P. yoelii blood-stage infection elicit the phenotypic and functional attributes of parasite-specific T cell exhaustion. These data also show that specific functional attributes (for example, the capacity to express IFN-γ) of parasite-specific T cells progressively deteriorate during prolonged blood-stage Plasmodium infection.

Blocking PD-L1 + LAG-3 clears blood-stage malaria

The inhibitory receptor PD-1 interacts with PD-L1 and PD-L2, and in addition, PD-L1 can interact with CD80 (refs 30, 31). LAG-3 mediates negative regulation through interactions with MHC class II[32, 33]. Importantly, synergistic blockade of T cell inhibitory receptor interactions improves CD8+ T cell function and virus control during chronic LCMV cl13 infection[24]. Thus, to address a biological role for inhibitory receptor expression by parasite-specific T cells during prolonged Plasmodium blood-stage infection, we administered non-depleting monoclonal antibodies (anti-PD-L1 and anti-LAG-3) that prevent PD-1 and LAG-3 inhibitory receptors from functionally engaging their major ligands (PD-L1 and MHC class II, respectively)[23, 24, 33]. To mimic therapeutic intervention of clinical malaria, mice with matched, high parasite burdens (~25-30% of RBC harbored parasites) at day 14 p.i. were given regular injections of anti-PD-L1 and anti-LAG-3, or control rIgG, and monitored for parasite clearance from the blood. We observed that PD-L1 and LAG-3 blockade resulted in immediate control of parasite burdens and substantially accelerated parasite clearance (), which correlated with substantially increased parasite-specific CD4+ T cell numbers () and restored parasite-specific CD4+ T cell cytokine responses (). PD-L1 and LAG-3 blockade similarly restored both numerical and functional features of parasite-specific CD8+ T cell responses (). Importantly, therapeutic PD-L1 and LAG-3 blockade failed to improve parasite clearance in CD4+ T cell-depleted or Aicda-/- μs-/- mice (data not shown), consistent with a critical role for the CD4+ T cell–B cell axis in PD-L1 and LAG-3 blockade-enhanced parasite control. Of note, anti-PD-L1 blockade alone had a partial effect on parasite clearance, while anti-LAG-3 alone minimally influenced blood-stage infection but rather acted synergistically with anti-PD-L1 to reduce parasite burden and accelerate parasite clearance (). Collectively, these data show markedly enhanced parasite control during blood-stage Plasmodium infection following therapeutic blockade of PD-L1 and LAG-3. Additionally, these data suggest that PD-L1 and LAG-3, and their respective ligand-receptor interactions, act synergistically to inhibit T cell responses during persistent Plasmodium blood-stage infection.

PD-L1 + LAG-3 blockade prevents chronic infection

To address whether improved clinical outcomes following therapeutic PD-L1 and LAG-3 blockade are system-specific (P. yoelii and C57BL/6 mice) or generalizable, we next evaluated clinical efficacy in outbred Swiss Webster mice, which more closely mimic the genetic complexity of humans. Importantly, therapeutic blockade of PD-L1 and LAG-3 also rapidly reduced parasitemia and accelerated clearance of P. yoelii blood-stage infection in outbred Swiss Webster mice (), suggesting that both T cell exhaustion and successful therapeutic PD-L1 and LAG-3 blockade are independent of immunogenetics, background genes and the targeted parasite antigens or epitopes. Additionally, we evaluated the impact of therapeutic PD-L1 and LAG-3 blockade on the clinical course and persistence of Plasmodium chabaudi chabaudi (Pcc), a malaria parasite known to establish long-term, subpatent infections in mice, as determined by the ability of transferred blood to establish new infections in naïve mice[34-36]. The rationale for extending our studies to the P. chabaudi model was to address the impact of inhibitory receptor blockade on the clearance of persisting malaria parasites that are undetectable by blood-smear, which is an important aspect of human malaria that is not recapitulated in the P. yoelii model. For these reasons, Pcc blood-stage infection of rodents is widely believed to be the small animal malaria model most relevant for understanding persistent Plasmodium blood-stage infection of humans. Therapeutic PDL1 and LAG-3 blockade initiated at day 14 p.i. did not have a discernable effect on clearance of patent Pcc blood-stage infection in C57BL/6 mice (). Of note, neither control rIgG-treated nor anti-PD-L1 plus anti-LAG-3 blockade-treated mice showed detectable Pcc recrudescence from days 18 through 30, or on day 40 (limit of detection of parasitemia <0.02%). However anti-PD-L1 plus anti-LAG-3 blockade resulted in the complete elimination of persistent, subpatent Pcc infection in the majority of mice as determined by the failure of whole blood transfers to initiate blood-stage infection, which was observed for 100% of recipient mice receiving blood from rIgG-treated, Pcc infected mice (). Collectively, these data show that the clinical benefits of therapeutic PD-L1 and LAG-3 blockade are generalizable to outbred rodents and occur independently of both host immunogenetics and Plasmodium parasite species. Moreover, these data highlight that immunity to even low-grade, subpatent Plasmodium infection can be enhanced by PD-L1 and LAG-3 blockade to mediate sterilizing parasite clearance.

PD-L1 + LAG-3 blockade enhances TFH and plasma cells

Given our results suggesting a critical role for the CD4+ T cell-B cell axis in clearance of blood-stage Plasmodium infection ( data not shown), as well as recent data showing modestly enhanced humoral immunity following in vivo anti-PD-1 blockade[37, 38], we next evaluated the mechanistic effects of PD-L1 and LAG-3 blockade on T follicular helper (TFH) CD4+ T cells, which regulate germinal center B cell reactions necessary for the generation of high-magnitude and quality antibody responses[39]. Of note, blood-stage Py infection resulted in the induction of CD150loCXCR5+ TFH cells (), as was recently reported following chronic LCMV infection of mice[40]. However, we observed a 7-fold enhancement of TFH CD4+ T cell numbers () and a 50-fold enhancement of plasmablast (CD19loB220loCD138hiIgD-) B cell numbers () in Py-infected mice receiving combined PD-L1 and LAG-3 blockade. Thus, improved parasite control following in vivo PD-L1 and LAG-3 blockade is directly associated with enhanced TFH cell numbers and substantial induction of plasma cell differentiation.

Blockade enhances plasmablasts and antibodies

To formally investigate a mechanistic role for B cells and antibody responses during PDL1 and LAG-3 blockade, we next enumerated and characterized total and germinal center B cells as well as the functional production of anti-Plasmodial serum immunoglobulins. Relative to rIgG-treated control mice, P. yoelii-infection matched mice receiving anti-PD-L1 and anti-LAG-3 in vivo blockade therapy exhibited 10-fold greater absolute numbers of CD19+B220+ B cells (), 20-fold increases in the absolute number of peanut agglutinin-staining (PNAhi) germinal center B cells () and 30-fold increases in the total number of germinal center B cells that had undergone class switch recombination (). Of note, the marked expansion of the CD19loB220loPNAint subset of cells observed following PD-L1 and LAG-3 co-blockade () corresponds to B cell plasmablast populations (CD19loB220loCD138hiIgD–; ). Consistent with the flow cytometric data, immunostaining of cryosectioned spleens from d21 p.i. control rIgG- and blockade-treated mice revealed qualitatively improved splenic architecture marked by preservation of IgM+ B cell follicles and enhanced PNA+ germinal center formation (). Additionally, we directly evaluated titers of serum immunoglobulins against the blood-stage Plasmodium parasite antigen and human malaria vaccine candidate, merozoite surface protein 1 (MSP119)[11, 41]. In line with improved TFH and B cell responses, in vivo combinatorial blockade of PD-L1 and LAG-3 resulted in the generation of higher-magnitude parasite-specific anti-MSP119 IgG antibody responses (). Importantly, compared to serum from rIgG-treated mice, passive transfer of serum from mice that received anti-PD-L1 and anti-LAG-3 blockade therapy to naïve mice resulted in substantially better control and accelerated parasite clearance following challenge with Py blood-stage parasites (). Given that total anti-MSP119 IgG titers were only 2.5 fold different between control and blockade-treated mice, these observations suggest that specific anti-MSP119 antibody subclasses and/or antibodies targeting other blood-stage parasite proteins also significantly contribute to protection following PD-L1 and LAG-3 combinatorial blockade. Collectively, these data show that therapeutic PD-L1 and LAG-3 blockade during established clinical malaria also substantially improves the anti-Plasmodial humoral response via potent induction and secretion of protective antibodies.

Discussion

Here, we show that humans with clinical malaria upregulate expression of the inhibitory receptor PD-1 on their CD4+ T cells. We then show that CD4+ T cell dysfunction, which negatively impacts the induction of protective Plasmodium-specific antibody responses, is a major underlying factor for prolonged blood-stage Plasmodium infections in mice. We also found that administration of PD-L1 + LAG-3 specific blocking antibodies markedly improved effector and TFH CD4+ T cell responses and antibody producing B cell responses, resulting in rapid control of parasite burden and accelerated parasite clearance. Our results are consistent with a previous study showing PD-1 expression on T cells during P. yoelii blood-stage infection of mice[42], although the importance of this finding was not experimentally addressed. Of relevance, similar biologics targeting inhibitory pathways are already in clinical trial to improve T cell function during neoplastic disease[32, 43]. Thus, our results reveal new information relevant to the mechanisms of parasite persistence and also have potentially direct relevance for alternative immune-based strategies to combat the malaria global health epidemic.

Chronic or prolonged microbial infection, and subsequent repeated antigenic stimulation, has been associated with functional exhaustion of CD8+ T cells in both humans[20, 21] and mice[22-24]. Although CD4+ T cell exhaustion also likely occurs in humans persistently infected with HIV[19, 44] or HCV[45], and CD4+ T cell inhibitory receptor expression may contribute to persistent Mycobacterium tuberculosis infection of mice[46, 47], direct evidence the relevance of functional T cell exhaustion is limited to CD8+ T cells[48]. On the other hand, Plasmodium blood-stage infection provides unique opportunities to evaluate the biological significance of CD4+ T cell exhaustion, as these cells are critical for suppression of parasite replication and full resolution infection[10-13]. Indeed, here we report marked improvement of T cell function when blood-stage infection is truncated by chloroquine treatment, and we see functional improvements in parasite-specific T cells following PD-L1 and LAG-3 co-blockade. This latter point provides conclusive evidence that these inhibitory receptor interactions compromise clearance of blood-stage malaria. Blocking these interactions restores functionality to exhausted CD4+ T cells, resulting in accelerated clearance of infection. Thus, functional exhaustion of parasite-specific CD4+ T cells plays a critical role in the persistence of blood-stage Plasmodium infection and provides a unique target to enhance both the control and clearance of malaria parasites.

It is well established that the CD4+ T cell-antibody secreting B cells axis is critical for resolving non-lethal Plasmodium blood-stage infections in mice and is highly associated with protection against severe malaria in humans. While the precise mechanisms remain to be defined, PD-L1 and LAG-3 blockade results in robust numerical and functional enhancement of effector CD4+ T cell, TFH CD4+ T cell and antibody secreting B cell responses. Of interest, a previous report described modestly enhanced humoral immunity following in vivo PD-1 blockade in SIV-infected monkeys[37, 38]. However, here we directly demonstrate that therapeutic PD-L1 and LAG-3 blockade during clinical malaria resulted in ~50-fold increases in the number of plasmablasts capable of secreting potent Plasmodium-specific antibodies. Of note, therapeutic PD-L1 and LAG-3 blockade does not prevent lethal outcomes following P. yoelii infection of mice depleted of CD4+ T cells or in Aicda–/– μs–/– mice lacking secreted antibodies (data not shown). Thus, the major and relevant functional targets of therapeutic PD-L1 and LAG-3 blockade are likely CD4+ T cells and B cells.

Recent data revealed that chronic LCMV infection directs the differentiation of virus-specific CD4+ T cells into TFH cells[40], although the role these cells play in clearing virus remains unknown. Similarly, we also observe the induction of TFH cell differentiation during prolonged blood-stage Plasmodium infection. However, we also show that PD-L1 and LAG-3 blockade results in substantially expanded numbers of the TFH CD4+ T cell subset, which likely drive the attending increases in germinal center B cell numbers, B cell class switching to cytophilic (for example, opsonizing IgG2b) isotypes, and plasmablast differentiation that are not observed in control mice. Thus, the improved clinical outcomes of PD-L1 and LAG-3 blockade in mice with established malaria are directly associated with multifaceted enhancement of both cellular and humoral immunity.

Lastly, our data also highlight the generalizability and potency of therapeutic PDL1 and LAG-3 blockade during clinical malaria. Indeed, our data show improved clinical outcomes in outbred Swiss Webster mice, suggesting that therapeutic PD-L1 and LAG-3 blockade stimulates more potent anti-Plasmodial responses independent of MHC alleles or the parasite antigens and epitopes targeted by B cells and CD4+ T cells, respectively. Importantly, due to immunogenetic complexities, studies of phenotypic and functional CD4+ T cell exhaustion in outbred populations are only possible using the surrogate activation marker approach to identify and track Plasmodium blood-stage specific CD4+ T cells responses. Furthermore, in mice persistently infected with Pcc, which establishes low-grade, subpatent infection of mice, PD-L1 and LAG-3 blockade mediated sterilizing clearance of blood-stage infection, suggesting that even low numbers of parasites or antigen persistence negatively impacts the ability of the host immune response to eliminate Plasmodium parasites. This later observation is of particular interest as humans can harbor subpatent P. falciparum infections for months or even years in the absence of clinical intervention[14-17], and Plasmodium ovale and Plasmodium vivax spp. can persist as hypnozoites[49] capable of generating recurring blood-stage infection. Thus, in addition to well-described mechanisms of evolution and antigenic variation during human infection with Plasmodium parasites[50], our data reveal a mechanism by which malaria parasites can evade the protective host immune responses through dampening and dysregulating CD4+ T cell and B cell function.