cGAS-mediated control of blood-stage malaria promotes Plasmodium-specific germinal center responses

Malaria is a disease caused by parasites of the Plasmodium genus that kills approximately 430,000 persons per year (1). The humoral immune response is critical for both acute clearance of blood-stage malaria and protection against subsequent rechallenge (2), yet poor understanding of how to achieve protective humoral immunity hampers vaccine design. The immune response to malaria is initiated when malaria-associated pathogen-associated molecular patterns are recognized by host innate cells via pattern recognition receptors (PRRs) (3). Activation of PRRs has at least two roles in host immunity during blood-stage malaria infection: (a) direct control of parasite replication and/or parasite killing via innate immune effector mechanisms and (b) generation of cues that expand and differentiate antigen-specific CD4⁺ T cells and B cells (3–5). It was recently reported that the PRR cyclic GMP-AMP synthase (cGAS) was a critical innate signal in the context of a murine model of lethal malaria (6). We used a nonlethal murine model of blood-stage malaria (Plasmodium yoelii 17XNL) to determine the role that cGAS plays in the generation of the humoral immune response and, specifically, immunologic memory.

cGAS is a PRR that senses the presence of cytosolic DNA (7–9); it is important for initiating the host immune response to intracellular bacteria (10–12) and DNA viruses (13). After binding to cytosolic DNA, cGAS enzymatically generates cyclic dinucleotides (2′3′ cyclic GMP-AMP) that signal via the adaptor molecule stimulator of interferon genes (STING), with the signaling cascade thought to culminate primarily in the production of type I IFN (7, 8, 14). While cGAS has been demonstrated to affect T cell–independent B cell responses (15), the role of cGAS in driving the humoral response during infection is poorly understood. We therefore hypothesized that disruption of cGAS-mediated immune activation could lead to enhanced disease in a nonlethal model of blood-stage malaria.

We therefore utilized the P. yoelii 17XNL parasite to examine the differentiation of Plasmodium-specific CD4⁺ T cells and B cells in mice with and without cGAS during blood-stage infection. To examine antigen-specific CD4⁺ T cells responding to infection, P. yoelii were generated that constitutively express the Lymphocytic choriomeningitis virus–derived (LCMV-derived) glycoprotein (GP) epitope (GP_61–80). This allows for the identification and analysis of antigen-specific CD4⁺ T cells using previously described GP66:I-A^B tetramer enrichment strategies (16). B cell tetramers were additionally used to identify polyclonal P. yoelii–specific B cells (17). Using these tools, we demonstrated that cGAS deficiency had no effect on the early priming, differentiation, or expansion of Plasmodium-specific CD4⁺ T cells or B cells. We further observed that the absence of cGAS led to poor parasite control, suggesting an as yet unidentified cGAS-dependent mechanism of parasite control. Loss of parasite control was associated with poor germinal center (GC) responses, that could be rectified by pharmacologic clearance of the parasite, further supporting an indirect effect of ongoing parasitemia on the GC response. If parasite replication is controlled, however, cGAS is dispensable for the development of the humoral response. We conclude that manipulation of the cGAS pathway could be pursued as a therapeutic intervention to control parasitemia.

cGAS-STING–type I Interferon signals contribute to control of parasite burden during blood-stage malaria. To clarify the potential role of the cGAS-STING–type I IFN axis in non lethal blood-stage malaria, we infected cGAS^+/+ or cGAS^–/– littermate control mice with P. yoelii 17XNL infected erythrocytes and measured parasitemia daily via flow cytometry (18). As expected with P. yoelii, most infected erythrocytes were CD71⁺ reticulocytes (19) (Figure 1A). Parasitemia reached a peak of approximately 20% infected erythrocytes at approximately 15 days after infection in cGAS^+/+ P. yoelii–infected mice and was cleared 18–21 days after infection (20). In cGAS^–/–, mice there was an approximately 2-fold increase in parasitemia starting 7 days after infection that persisted until parasites were cleared in both groups around day 22 (Figure 1B). Higher parasite burden in cGAS^–/– mice was associated with worsened weight loss, increased anemia, and poor thermoregulation when compared with littermate controls (Supplemental Figure 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.94142DS1). We additionally recapitulated results reported in a lethal strain of P. yoelii YM (6) in which immunopathology driven by cGAS is ameliorated in its absence, leading to enhanced cGAS^–/– mouse survival (Supplemental Figure 1D).

Figure 1

cGAS, STING, and type I interferon contribute to control of blood-stage Plasmodium infection. (A) Flow cytometry gating scheme used to identify infected erythrocytes. Infected erythrocytes were defined as CD45^–, Ter119⁺, Hoechst⁺ cells. Immature red blood cells (reticulocytes) were identified by expression of CD71. (B) Male cGAS^+/+ and cGAS^–/– age-matched littermates between 6 and 10 weeks of age were infected with 10⁶ Plasmodium yoelii 17XNL–infected erythrocytes via the intraperitoneal route. The mean percentage of infected erythrocytes ± SD is displayed. *P < 0.05, as assessed by unpaired Student’s t test. Each group had at least 4 mice, and infection course was representative of 2 separate experiments. (C) IFNAR^+/+ and IFNAR^–/– age-matched littermates were infected with 10⁶ Plasmodium yoelii 17XNL–infected erythrocytes via the intraperitoneal route. The mean percentage of infected erythrocytes ± SD is displayed. *P < 0.05, as assessed by unpaired Student’s t test. Each group had at least 4 mice, and infection course was representative of 2 separate experiments.

To further explore the role of the cGAS-STING–type I IFN axis, we repeated our experiments in IFNAR^+/+ and IFNAR^–/– littermate controls, observing a similar phenotype of increased parasitemia in IFNAR^–/– mice (Figure 1C). We also infected STING signaling mutant mice (Tmem173^gt/gt), and similar to cGAS^–/– mice, these mice displayed an elevation in parasitemia starting at day 8 after infection and increased anemia, enhanced weight loss, and poor thermoregulation (W.O. Hahn et al., unpublished observations). Poor control of parasitemia in mice deficient in the cGAS–type I IFN axis demonstrates that this axis is important for the host response to blood-stage malaria.

Deficiency in cGAS is associated with altered type I IFN signature. cGAS has been most strongly linked with the production of type I IFN (7), so we assessed the effects of cGAS on gene programs associated with type I IFN signaling. We measured type I IFN and interferon-stimulated gene (ISG) mRNA expression relative to naive mice on days 4, 7, 9, and 14 in bulk splenocytes. We observed decreased expression (fold change relative to naive) of IFN-β in cGAS^–/– mice compared with WT mice both days 1 and 7 after infection (Figure 2A). Because we observed IFN-β mRNA, it was important to examine downstream candidate ISGs. We selected IRF7, IFIT1, and CXCL10, as they had previously been demonstrated to be transcriptionally upregulated in mice with activated cGAS signaling (21). Decreased expression of IRF7, CXCL10, and IFIT1 was observed in cGAS^–/– mice as compared with WT mice at day 7 and 9 (Figure 2A). To assess whether differences in ISG expression could be attributed to differences in IFN-γ production, we also measured IFN-γ protein in the serum by ELISA and IFN-γ mRNA expression in total splenocytes and observed no difference between cGAS^–/– mice and WT controls at any time point examined (W.O. Hahn, unpublished observations).

Figure 2

Deficiency in cGAS is associated with altered type I IFN signature. (A) Quantitative real-time PCR of indicated gene mRNA in bulk spleen tissue. Quantification was performed using the delta-delta CT method and normalized to a naive mouse, with HPRT as the designated housekeeping gene. Experiments were performed using 2 technical replicates of at least 6 biological samples with 2–3 separate experiments per time point. One representative experiment is shown. Since the data were nonparametric, statistical significance was assessed via Mann-Whitney U test. *P < 0.05, **P < 0.01. (B) Mean fluorescent intensity of PDCA-1 (CD317) on CD11b⁺ dendritic cells in representative flow plot. See Supplemental Figure 2 for full gating scheme. Representative data are shown. Statistical significance was assessed via Mann-Whitney U test. *P < 0.05. (C) L929-ISRE cells were plated at 5 × 10⁴ cells per well and cocultured for 6 hours at a ratio of 100 splenocytes to 1 reporter cell in a 96-well plate (mean ± SD). Cells were lysed and luciferase activity was measured. Statistical analysis was performed using Student’s t test.

We also examined IFN-dependent surface expression of PDCA-1 in CD11b⁺ DCs (22–24). Flow cytometric analysis of CD11c⁺CD11b⁺ dendritic cells after infection revealed no difference in the two experimental groups at 4 days after infection. By 7 days after infection, there was higher expression of PDCA-1 on CD11b⁺ dendritic cells in WT mice as compared with cGAS^–/– mice (Figure 2B and Supplemental Figure 2). To confirm that the mRNA and surface expression of ISG molecules reflected bioactive interferon in the spleen, we utilized an L929 cell line that stably expresses luciferase under a type I IFN–specific promoter (25). Coculture of splenocytes taken from mice at day 7 after infection revealed an elevated luciferase signal in WT mice compared with cGAS^–/– mice (Figure 2C). Together with our IFN-β and ISG expression data, these data support a model in which cGAS is required for optimal type I IFN expression.

cGAS deficiency does not affect early antigen-specific CD4⁺ T cell expansion or differentiation. CD4⁺ T cells are critical for the host response to blood-stage Plasmodium infection in addition to being critical for driving T-dependent B cell responses (26). Since our data suggested that there was a defect in the type I IFN cascade and type I IFN signaling has been reported to influence both the expansion and differentiation of CD4⁺ T cells (27, 28), we next sought to determine whether cGAS was associated with defects in early priming or expansion of antigen-specific CD4⁺ T cells during blood-stage malaria infection. To measure antigen-specific CD4⁺ T cell expansion, we utilized a P. yoelii parasite modified to stably express amino acids 51–84 of the LCMV GP under the Hep17 promoter (Supplemental Figure 3A). Expression under the Hep17 promoter localizes peptide expression to the parasitophorous vacuole, and this promoter is active during the erythrocytic stage of infection (29). This LCMV sequence encodes an immunogenic CD4⁺ T cell epitope presented on I-A^b MHC molecules (aa 61–80) (30, 31). Infection with P. yoelii expressing the GP₆₆ construct drove the expansion of GP₆₆-specific CD4⁺ T cells that could be identified with a GP₆₆:I-A^b tetramer (Figure 3A) (32). After infection, GP₆₆:I-A^b+ cells expanded from 328 ± 132 cells in a naive mouse to a peak of approximately 40,000 cells by day 9. We observed no differences in the numbers of GP₆₆:I-A^b+CD4⁺ T cells in cGAS^–/– mice as compared with WT controls on days 4, 7, or 9 after infection (Figure 3A), suggesting that cGAS does not play a role in the priming or expansion of antigen-specific CD4⁺ T cells during P. yoelii infection.

Figure 3

Absence of cGAS does not affect early expansion or differentiation of antigen-specific CD4⁺ T cells. (A) Quantification of GP₆₆I:A^b CD4⁺ T cells enriched from spleen and lymph nodes on indicated days after infection. Cells were gated on singlets, CD11b/CD11c/B220^–, CD4⁺ lymphocytes. Each time point represents data from at least 2 separate experiments with at least 6 mice per time point, except for the day 0 time point, where n = 3. Error bars represent SD. (B) Representative flow plots demonstrating CXCR5 and PD1 surface expression from CD44⁺ GP₆₆I:A^b+CD4⁺ T cells in the spleen and lymph nodes 7 days after infection with Plasmodium yoelii 17XNL. Data shown are representative of 4 separate experiments with at least 9 biological replicates. There were no statistical differences between any groups. Statistical analysis was performed using the unpaired Student’s t test.

Early type I IFN receptor blockade via antibody has recently been reported to enhance the formation of the total T follicular helper (Tfh) CD4⁺ T cell population in P. yoelii–infected mice (33), so we sought to determine if an absence of cGAS altered the differentiation of CD4⁺ T cells responding specifically to infection. As reported for the nonantigen-specific CD4⁺ T cell response in P. chabaudi (34), GP₆₆:I-Ab⁺CD4⁺ T cells formed in response to P. yoelii were predominantly CXCR5⁺ Tfh cells, with strong upregulation of CXCR5. By day 7, approximately 80% of CD44⁺GP66⁺CD4⁺ T cells had upregulated CXCR5 in both cGAS^–/– and WT mice (Figure 3B). Furthermore, we observed no differences in the intracellular expression of the Th1 master transcription factor Tbet on GP₆₆:I-A^b+CD4⁺ T cells in cGAS^–/– mice as compared with controls on day 7 after infection (data not shown). From these observations, we concluded that cGAS does not affect the early expansion and differentiation of antigen-specific CD4⁺ T cells.

cGAS does not affect early antigen-specific B cell expansion or differentiation. Previous investigations have demonstrated that antibodies play a critical role in the host response to acute blood-stage malaria infection with P. yoelii XNL (35), so an alternative hypothesis was that cGAS was affecting the early B cell response. Early production of antibodies can occur via a short-lived CD138⁺ plasmablast B cell subset (36, 37). Because cGAS has been reported to affect the T-independent B cell response (15), we hypothesized that the loss of parasite control we had observed in cGAS^–/– mice could be associated with a loss of early plasmablast formation in a T cell–independent fashion. To examine the antigen-specific B cell response, we used a B cell tetramer comprising the C-terminal portion of the merozoite surface protein 1 (MSP1) protein of P. yoelii (17, 38). Antibodies against MSP1 are associated with clinical immunity in endemic populations (39, 40) and murine models of Plasmodium infection (41, 42). Furthermore, memory B cells specific for this protein can be found in malaria-exposed individuals living in Plasmodium endemic regions (43, 44). To test the role of cGAS in the development of the humoral immune response, the differentiation of MSP1-specific B cells was assessed after P. yoelii infection of WT and cGAS^–/– mice (Figure 4A). As we found with CD4⁺ T cells, there were no differences in the expansion of total MSP1-specific B cells at day 7 after infection, a time point at which CD138⁺ plasmablasts and activated GC precursors can be found (43). The development of MSP1-specific CD138^–GL7⁺CD38⁺ activated precursor B cells (45) was also similar in cGAS^–/– mice and WT controls 7 days after infection (Figure 4C). We also measured the concentration of IgG2c and IgM antibodies in the sera of infected cGAS^–/– and WT mice at day 9 after infection. IgM production has been shown to be diminished in mice lacking cGAS compared with cGAS-sufficient mice exposed to a viral infection (15). IgG2c has been shown to be the predominant isotype formed in response to P. yoelii infection (46) and is most strongly associated with protection when administered to naive mice (47). We elected to measure serum antibodies on day 9, after differences in parasite load were observed in cGAS^–/– versus WT mice (Figure 1B). We reasoned that if serum antibody levels were similar, this would argue against antibodies contributing significantly to cGAS-mediated parasite control at this time point. Serum concentrations of both isotypes, as measured by ELISA, were not different 9 days after infection (Figure 4D). We also observed no differences in the percentage or number of plasmablasts formed by day 7 (Figure 4, B and E). Therefore, cGAS did not affect the early activation, differentiation, or antibody production of antigen-specific B cells responding to infection with P. yoelii 17XNL.

Figure 4

Absence of cGAS does not affect early activation or expansion of malaria-specific B cells or production of malaria-specific class-switched antibodies. (A) Flow cytometry gating scheme to quantify antigen-experienced MSP1⁺ B cells enriched from spleen and lymph nodes on days 4, 7, and 9 after infection. Antigen-experienced B cells were identified as singlets, Thy 1.2^–, CD138⁺, and/or B220⁺, MSP1⁺ tetramer⁺ cells. Error bars represent SD. (B) Representative plots of CD138⁺ and CD138^–MSP1⁺ B cells shown in A. There were no statistical differences between the frequencies of CD138⁺ or CD138^– B cells in experimental versus control mice. (C) GL7 by CD38 staining on CD138^– cells after infection. Cells were gated as singlets, Thy 1.2^–, decoy^–, MSP1⁺, CD138^– cells. Error bars represent SD. (D) IgM and IgG MSP-specific antibody titers 9 days after infection. The average titer from 11 mice pooled from 3 experiments is shown. Error bars represent SD. Statistical analysis was performed using the unpaired Student’s t test. (E) Representative plots demonstrating the isotype of MSP1⁺ B cells. Plots representative of 3 experiments are shown.

cGAS deficiency reduces GC formation late in infection. Malaria infection has been associated with impairment of GC formation in both natural infection and model systems. While we had observed no obvious differences in the formation of CD4⁺ T cell Tfh cells or activated GC precursor B cells in WT versus cGAS^–/– mice, we wanted to examine the effects of persistent infection on GC formation. GC formation was first assessed by histology. We defined a follicle by an area with clear T cell and B cell zones, and we defined PNA-expressing cells centrally located within a follicle as an active GC. On day 14 after infection, WT animals demonstrated a greater percentage of PNA⁺ follicles when compared with cGAS^–/– mice (Figure 5A). In keeping with a greater percentage of GCs in WT mice compared with cGAS^–/– mice, we also observed significantly higher concentrations of MSP1-specific IgG2c antibodies on day 14 after infection in WT mice as compared with cGAS^–/– mice, with no difference in the IgM response to MSP1 (Figure 5B).

Figure 5

Absence of cGAS impairs germinal center formation and decreases production of class-switched antibody. (A) Sections of spleen harvested 14 days after infection stained with antibodies against PNA, B220, and CD4 (original magnification, ×20). Follicles were counted, and the percentage of follicles with clear PNA⁺ staining was determined. 77% ± 10% PNA⁺ follicles were in WT mice as compared with 33% ± 9% in cGAS^–/– mice. P = 0.0052 by Student’s t test. Sections representative of 3 biological replicates from 3 separate experiments are shown. Scale bar: 1,000 μm (left); 100 μm (right). (B) Quantification of MSP1 IgG2c and IgM by ELISA in the serum 14 days after infection. Data represent 7 biological replicates pooled from 3 separate experiments. *P < 0.05, ***P < 0.01. (C) Representative flow plots of enriched CD44⁺GP66⁺CXCR5^– T effector cells (Teff), CXCR5^int Tfh cells, and CXCR5^hi GC Tfh cells. Data represent at least 9 biological replicates pooled from 3 separate experiments. (D) Representative flow cytometry plots used to identify CD138⁺MSP1⁺ B cells. Data represent at least 7 biological replicates pooled from 3 separate experiments. P = 0.0002. (E) Representative flow plots demonstrating CD38⁺GL7^– memory B cells, CD38⁺GL7⁺ GC precursors, and CD38^–GL7⁺ GC B cells. Data represent at least 7 biological replicates pooled from 3 separate experiments. Statistical analyses were performed using the unpaired Student’s t test. *P < 0.05, ***P < 0.01. Error bars represent SD.

Due to the obvious differences in GC histology and poor formation of antibodies in cGAS^–/– mice compared with WT mice, we hypothesized that there could be a defect in the generation of GC CD4⁺ Tfh (GC Tfh) cells in cGAS^–/– mice late in infection. GC Tfh cells can be identified by their high expression of both the B cell–homing chemokine receptor CXCR5 (48) as well as high expression of the inhibitory receptor PD-1 (49). An examination of the GP66:I-Ab⁺ cells 14 days after infection revealed a reduction of GC Tfh cells in the cGAS^–/– mice compared with WT mice associated with a concomitant decrease in the percentage and number of GP66:I-Ab⁺ CXCR5^– T effector cells (Figure 5C). Interestingly, in the absence of cGAS signaling, the majority of GP66:I-Ab⁺ cells resembled CXCR5^intPD1^lo Tfh cells, thought to localize to either the T-B border or in the extrafollicular region (50).

These Tfh cells have also been associated with the production of plasmablasts in both autoimmune disease and infection (26), so we therefore looked to determine if there were differences in antigen-specific B cell formation. While there were no differences in the absolute number of plasmablasts, cGAS^–/– mice still had a small but significant increase in the percentage of B220⁺CD138⁺ plasmablasts compared with WT controls (Figure 5D and Supplemental Figure 4). Similarly, there was an increased percentage of CD38⁺GL7⁺ MSP1-specific B cells — markers previously associated with an activated B cell population capable of generating both memory and GC B cells (“activated precursors”) of total activated precursors (27) — in cGAS^–/– mice compared with WT mice (Figure 4E and Supplemental Figure 4). We also observed a 2-fold decrease in the number and a decrease of approximately 10% in the frequency of GC B cells in cGAS^–/– mice compared with WT controls (Supplemental Figure 4 and Figure 5E). We can therefore conclude from these data that, in mice that lack cGAS, there are fewer antigen-specific GC Tfh cells and GC B cells, a greater percentage of plasmablasts and activated precursor B cells, and reduced antibody titers.

Persistent infection causes GC collapse and reduction in B cell memory. Because we had observed marked differences in the development of GC responses 14 days after infection in cGAS^–/– mice as compared with WT controls, we sought to determine the consequences for the development of cellular memory at a time point at which we had previously found MSP1-specific memory B cells to be present (38). Immunofluorescent staining of spleens from WT mice demonstrated ongoing PNA⁺ GCs, whereas there was an absence of PNA⁺ GCs in cGAS^–/– mice (Figure 6A). We further characterized the MSP1⁺ B cell populations by flow cytometry and determined that not only did cGAS^–/– mice have fewer total MSP1-specific cells, but more specifically they had fewer GC and memory B cells than WT mice (Figure 6B). Of interest, there were higher numbers of CD138⁺ cells in cGAS^–/– mice than WT mice (Figure 6B). This altered humoral response in cGAS^–/– mice was also reflected in lower levels of MSP1-specific IgG2C but not IgM (Figure 6, C and D). Expression of CD73, a marker associated with B cells that have interacted with T cells (51, 52), was markedly reduced on both IgM⁺ and IgM^– memory B cells of cGAS^–/– mice as compared with WT controls (Figure 6E). We therefore concluded that ongoing infection is associated with poor GC function and memory B cell output in cGAS^–/– mice.

Figure 6

Prolonged reduction in germinal center response is associated with reduced memory B cell formation in cGAS^–/– mice. (A) Spleen sections were stained with antibodies against PNA, B220, and CD4. Scale bar: 1,000 μm (first and third images); 100 μm (second and fourth images). (B) Quantification of MSP1⁺ B cells subsets 22 days after infection with Plasmodium yoelii 17XNL. Error bars represent SD. *P > 0.05, ***P < 0.001, ****P < 0.0001, by unpaired Student’s t test. (C and D) Quantification of serum levels of MSP1-specific IgG2c and IgM 22 days after infection. Data represent 7 biological replicates from 3 pooled experiments. Statistical analysis was performed using the unpaired Student’s t test. Error bars represent SD. **P < 0.01, ***P < 0.001. (E) Mean fluorescent intensity of CD73 on MSP1-specific memory B cells that are CD38⁺/GL7^– 22 days after infection. Representative data are shown representative of 2 separate experiments. The geometric mean intensity (GMI) of cGAS^–/– mice was 600 ± 151 as compared with versus 1,763 ± 112 in WT controls. P < 0.001 by unpaired Student’s t test. Error bars represent SD. *P < 0.05.

cGAS has no cell-intrinsic effects on CD4⁺ T cell or B cell GC development. Since we had observed differences in the formation of GCs in cGAS^–/– mice as compared with WT controls, we tested whether cGAS affected CD4⁺ T cells or B cell differentiation in a cell-intrinsic fashion by constructing congenically marked mixed bone marrow chimeric mice. Six weeks after a 1:1 input of 2.5 × 10⁶ bone marrow cells from either CD45.1 mice (WT) or cGAS^–/– (CD45.2) mice, chimeric mice were infected with P. yoelii 17XNL (Figure 7A). Fourteen days after infection, flow cytometric characterization was performed. No difference was observed in the ratio of congenically marked CD4⁺ T cells or B cells when compared with the ratio of congenically marked cells tested prior to infection (data not shown). No differences were observed in the percentage of GP66⁺-specific CD4⁺ GC T cells (Figure 7B). No differences were observed in MSP1-specific B cells (Figure 7C). Thus, we concluded cGAS deficiency has no cell-intrinsic effects at time points where we had previously observed differences in the adaptive response in cGAS^–/– mice as compared with controls.

Figure 7

cGAS has no cell-intrinsic activity in antigen-specific CD4⁺ T cells or B cells. (A) Representative flow plot demonstrating CD45.1 and CD45.2 chimerism among total MSP1-specific B cells 14 days after infection. (B) CD138⁺ plasmablast frequency (top row) and GL7⁺ precursor and GC B cells (bottom row) are similar in WT compared to cGAS^–/– MSP1⁺ B cells. Plots are representative of 6 mice from 2 separate experiments. Statistical analysis was performed by the paired Student’s t test. (C) Representative flow plot demonstrating CD45.1 and CD45.2 chimerism among total GP66⁺CD44⁺CD4⁺ T cells 14 days after infection. Frequencies of CXCR5^–Teff, CXCR5^int Tfh, and CXCR5^hi GC Tfh cells are similar between WT and cGAS^–/– populations. Plots are representative of 6 mice from 2 separate experiments. Statistical analysis was performed by the paired Student’s t test.

Clearance of parasite by atovaquone restores B cell memory and GC responses. We have previously observed profound effects of blood-stage infection on the development of liver-stage-specific germinal center and memory formation (38). Plasmodium infection has been associated with ongoing splenic disruption in mice, monkeys, and humans (52–56). We hypothesized that the reduced GC function we observed in cGAS^–/– mice could therefore be due to indirect effects from exacerbated infection. To test this possibility, we administered the drug atovaquone (MilliporeSigma) (which kills blood-stage parasites) to cGAS^–/– and control mice 7 days after infection. We chose this regimen because we had observed differences in parasitemia in WT and cGAS^–/– mice starting after day 7, and other antimalarial agents have known immunomodulatory properties (57, 58). Parasites were cleared from the blood within 2 days of starting atovaquone treatment (data not shown). We assessed the antigen-specific CD4⁺ T cells and B cells from experimental or control mice on day 22 after infection when we had previously seen differences in GC responses. Drug clearance of the parasite resulted in the complete restoration of both the number and percentage of GC Tfh cells (Figure 8A). Drug treatment reduced the percentage of CD138⁺MSP1⁺ plasmablasts to levels similar to WT mice (Figure 8B). Coincident with the reduction in plasmablasts, treatment with atovaquone restored the number and percentage of GC B cells in cGAS^–/– mice to levels similar to WT mice (Figure 8C). We therefore concluded that both the compromise in GC function and increased rate of plasmablast formation we had observed in cGAS^–/– mice were dependent upon ongoing infection and could be restored by clearing the infection with atovaquone. These data demonstrate that the effects of cGAS deletion on the GC response are indirect and reflect an important role for cGAS in reducing parasitemia but cGAS is dispensable for development of the GC response.

Figure 8

Clearance of parasite with atovaquone restores germinal center CD4⁺ T cells and B cell responses. (A) WT and cGAS^–/– mice were infected with 10⁶ Plasmodium yoelii 17XNL–infected erythrocytes via the intraperitoneal route. Starting at day 7, mice were treated with atovaquone for 5 days via daily intraperitoneal injection. GP66⁺CD4⁺ T cells enriched from spleen and lymph nodes were examined for GC Tfh differentiation using antibodies against CXCR5 and PD1. Because data were nonparametric with different SDs, statistical analysis for both number and percentage was performed using a Kruskal-Wallis test with Dunn’s post-hoc comparison. WT mice were selected as the comparator group. *P < 0.05, ***P < 0.001. (B) MSP1⁺ B cells were identified and CD138⁺ plasmablasts were quantified. Statistical analysis was performed using Kruskal-Wallis test with Dunn’s post-hoc comparison. WT mice were selected as the comparator group. In cGAS^–/– mice treated with atovaquone, there were CD138⁺MSP1⁺ 3.169% ± 3.6% (P = 0.92). *P < 0.05, ***P < 0.001. (C) MSP1⁺ GC B cells were quantified. *P < 0.05, ***P < 0.001. Statistical analysis was performed using Kruskal-Wallis test with Dunn’s post-hoc comparison. WT mice were selected as the comparator group. Data represent at least 6 biological replicates from 2 separate experiments. Error bars represent SD.

Our investigations reveal that cGAS contributes to parasite control, which is essential for the formation of GC-derived humoral immunity. The ability to simultaneously identify and analyze endogenous polyclonal antigen-specific CD4⁺ T cells and B cells responding to Plasmodium has revealed a profound effect of innate immune control of parasitemia. Control of parasitemia is important for timely development of the GC response, a response critical for both the efficient generation of high-affinity antibodies and durable immune memory (59–62). Furthermore, our ability to track parasitemia throughout the course of infection combined with the potential to rapidly clear blood-stage infection with atovaquone provided us with an elegant system in which the effects of innate control of parasite burden could be disentangled from the effects of early innate signals on the adaptive immune response. Specifically, the collapse of the GC response in cGAS^–/– mice as parasitemia persisted could be abolished with clearance of the parasite via drug treatment, emphasizing the destructive effects blood-stage Plasmodium infection can have on the host GC. From a translational standpoint, our findings suggest that ongoing blood-stage malaria infection is generally deleterious to the GC response, providing insight into how improved immunity may develop under chemoprophylaxis against blood-stage infection (63). Previously, it had been demonstrated that blood-stage infection compromised the development of liver-stage antibodies (38, 64). The concept that an ongoing blood-stage infection also reduces the quality of a blood-stage humoral response is, however, to our knowledge novel.

Importantly, our findings demonstrate that cGAS contributes to parasite control in response to a nonlethal P. yoelii infection. Our findings emphasize the importance of balancing immune-mediated protection and pathology during the primary immune response. In contrast to recent reports documenting increased survival in cGAS^–/– mice during infection with the closely related lethal strain P. Yoelii CM (6), we determined that absence of cGAS was associated with poor parasite control late during infection with the nonlethal P. yoelii 17XNL. We confirmed that cGAS^–/– mice had worsened outcomes with the lethal strain P. yoelii 17XL. While differences in type I IFN production at the protein level were present late during infection, cGAS may drive production of other cytokines. For example, increased serum levels of IL-12, TNF, and CCL2 have been reported in mice treated with exogenous cGAMP (the secondary messenger produced by cGAS) (65, 66).

This finding mirrors the potentially dichotomous role of type I IFN in lethal models as compared with nonlethal models of blood-stage malaria. In murine models of Plasmodium in which death is associated with immunopathology, such as P. berghei, the absence of IFNAR leads to reduced tissue pathology and reduced mortality (24, 67–69). Treatment with IFN-α or IFN-β, both prior to and during infection, however, reduces lethality, implying a role for early parasite control in contributing to lethality (70, 71). In nonlethal murine models, the role of IFNAR signals vary depending on the strain of parasite and experimental conditions. During P. chabaudi infection, IFNAR^–/– mice had elevated (72), unchanged (21), or reduced parasitemia (73, 74). The role of type I IFN in the control of P. yoelii (another model of nonlethal blood-stage malaria) is also variable, as disruption of type I IFN signals has been demonstrated to either increase (75) or decrease parasite burden (28, 33). Furthermore, increased type I IFN signals associated with knockout of the IRF3 repressor FOSL1 lead to reduced parasite burden (76).

Although our results differ from the reports of Zander et al. with respect to the effects of type I IFN signals on parasite control during P. yoelii 17XNL infection (33), the discrepancy may be due to the timing of the depletion of type I IFN, as they administered antibody on days –1, 2, and 4. Similar to others (75), we observed that type I IFN signals persist beyond day 4 after infection with P. yoelii and may be important for innate effector mechanisms. Sebina et al. also recently reported reduced parasitemia in IFNAR^–/– mice as compared with control mice infected with P. yoelii 17XNL starting at day 14 after infection (28). Another potential explanation for different experimental outcomes could be either mouse strain differences or differences in the microbiome, both of which have substantial effects on the host response to infection in murine models of blood-stage malaria (77–79). We used littermate controls to minimize the contribution of microbiome or strain background to our observations. Differences in parasite virulence may also account for some of the observed differences. We avoid using parasites passaged more than 3 times without recirculation through a mosquito, as repeated passage of murine strains of malaria has been demonstrated to increase virulence and substantially alter the host response (80).

Our observations that clearance of blood-stage malaria with atovaquone eliminates differences in the T cell–dependent GC response should prompt future studies to consider drug treatment as a control in models with divergent parasite burden. When infection is ongoing, there was impairment of both malaria-specific GC Tfh CD4⁺ T cells and GC B cells that could be restored with pharmacologic clearance of the parasite. Although antibodies are important for clearance of blood-stage malaria (81–84), whether manipulation of innate cytokines or PRRs directly affects the humoral response to blood-stage malaria infection should be determined after normalization of the parasite burden.

We propose that one mechanism for the poor GC response observed in the cGAS^–/– mice is due to increased formation of plasmablasts. This occurs in a cGAS-independent manner late after initial priming and expansion and occurs in conjunction with a reduction in the number of GC B cells. Since plasmablasts are thought to be short lived and not contribute to the memory pool (85), our data support epidemiologic observations that high levels of ongoing parasite transmission are associated with poor-quality, short-lived B cell responses (86–88). In contrast to the reduced number of MSP1-specific B cells, we determined that ongoing infection had no effects on the total number of endogenous antigen-specific CD4⁺ T cells but did affect the formation of GC Tfh CD4⁺ T cells. We have previously demonstrated that blood-stage infection disrupts the GC response to liver-stage antigens and that blood-stage infection is associated with alterations in chemokine expression, including CXCL10, that can be reversed by treatment with atovaquone (38). It is likely that the elevated parasite burden in cGAS mice leads to persistent alterations in chemokine signals. Differences in antigen burden may also affect the poor GC formation in the cGAS^–/– mice. Using model antigen systems (NP-CPG), it has recently been demonstrated that excess antigen diverts GC precursors toward a plasmablast fate (89).

We determined that there was no difference in CD4⁺ T or B cell development when the host environment was normalized using mixed bone marrow chimeric mice, demonstrating that cGAS does not act in a cell-intrinsic manner. Furthermore, elimination of the parasite with antimalarial drugs, restored late GC Tfh CD4⁺ T cell function in cGAS^–/– mice, demonstrating that the deficiencies in CD4⁺ GC Tfh formation we had observed were due to elevations in the parasite burden rather than an intrinsic programming defect.

The finding that ongoing blood-stage infection impairs the development of GC-derived humoral immunity to blood-stage antigens has clinical implications. For example, children in areas of high parasite transmission had worse vaccine response rates to the RTS’S vaccine than children in lower transmission areas, with no evidence of boosting of CSP-specific antibodies with natural infection or vaccination (90). A vaccine capable of reducing the magnitude or duration of blood-stage malaria could reasonably be anticipated to improve the quality of humoral response against Plasmodium and potentially other pathogens. Areas where malaria is eliminated consistently experience a decrease in mortality greater than can be attributed to malaria alone, typically associated with decreases in unrelated infections (91, 92). The role that persistent cGAS-mediated innate signaling plays in the control of parasitemia should remain an active area of exploration. Therefore, despite minimal effects on the adaptive system during acute priming of the response to blood-stage malaria, the cGAS-STING axis remains a target for manipulation for improving clinical outcomes in blood-stage malaria.

Mice. Male 6-to 8-week old C57BL/6J mice were purchased from Jackson ImmunoResearch Laboratories and were maintained and bred under specific pathogen–free conditions, per the institutional guidelines at the University of Washington. IFNAR^–/–, Tmem173^gt/gt (STING^–/–), and cGAS^–/– mice were provided by the MG and Dan Stetson (University of Washington). IFNAR^–/– and cGAS^–/– mice were bred as heterozygotes for experiments with littermate controls after genotyping as previously described (93, 94). All mice used for experimental purposes were male and aged 6–10 weeks.

Mixed bone marrow chimeras. Mixed bone marrow chimeras were generated as described previously (49). Briefly, bone marrow cells were depleted of T and NK cells, and congenically marked WT (CD45.1) cells and cGAS^–/– (CD45.2) cells were mixed in equal portions. Recipient mice (CD45.1/CD45.2) were lethally irradiated (10 Gy) and injected with total bone marrow cells and given enrofloxacin-treated water for 6 weeks.

Parasites. The 17XNL strain of P. yoelii was modified at a dispensable genetic locus (“p230p”) to stably incorporate an expression cassette for the P. yoelii hep17 gene tagged at the C-terminus with amino acids 51–84 of the GP from LCMV. Standard methods were used to create and integrate a linearized pDEF construct as well as to select transgenic parasites and to obtain two pure, clonal populations for further study. Confirmation of the presence of the transgenic locus was determined by genotyping PCR. P. yoelii 17XNL expressing the GP₆₆ epitope were maintained as frozen blood stocks and passaged through donor mice with no more than 3 passages prior to recirculation through mosquito inoculation. Infections were induced by intraperitoneal injection of 10⁶ parasitized red blood cells from the blood of donor mice at 1%–5% parasitized red blood cells.

Tetramer production. Biotinylated I-A^b LCMV GP 66–77 DIYKGVYQFKSV monomers were obtained from the NIH tetramer core and tetramerized with SA-APC as previously described (95). For antigen-specific B cell experiments, a 14-kDA truncated carboxy terminus of PyMSP1 was cloned, purified, biotinylated, and tetramerized with streptavidin-PE (Prozyme) as previously described (38, 96). Decoy reagent to detect B cells specific for tetramer components was constructed as previously described (17, 43).

Cell enrichment, flow cytometry, and antibodies. Single-cell suspensions of spleen and cervical, mediastinal, axillary, brachial, pancreatic, renal, mesenteric, inguinal, and lumbar lymph nodes (SLO) were prepared by mashing through Nitex mesh (Sefar, B0015H309U) and resuspending in 2% FBS and Fc block (2.4G2). Hemozoin was depleted via running cell suspension over magnetized LS columns (Miltenyi Biotec), and the resulting fraction was stained with GP₆₆I:A^b-APC tetramer and enriched by magnetic beads as previously described (95). The unbound fraction was then stained with decoy γ reagent at a concentration of 10 nM at room temperature for 15 minutes followed by MSP1 tetramer for 30 minutes on ice, washed, and then stained with anti-PE beads prior to a repeated magnetic enrichment. All bound cells then were stained with antibodies shown in Supplemental Table 1, detected on an LSRII Flow Cytometer (BD Biosciences), and analyzed using Flowjo 9.94 (Treestar).

ELISA. 96-well ELISPOT plates (Millipore) were coated overnight at 4°C with MSP1⁺ protein (BEI resources) at 1 μG/ml. Plates were then blocked with 5% dehydrated milk prior to sample incubation. Plates were incubated with serially diluted serum. Bound antibodies were detected using either IgM biotin (clone II/41) (Southern Biotech) or IgG2c biotin (clone 5.7) (Southern Biotech), followed by streptavidin-HRP (BD). Absorbance was measured at 450 nm using an iMark Microplate Reader (Bio-Rad). IFN-γ concentration was calculated using the Ready-SET-Go! ELISA kit (eBioscience) as per the manufacturer’s instructions.

Histology. Mouse spleens were flash frozen in liquid nitrogen prior to cutting on cryotome. 10-micron sections were cut and fixed for 10 minutes with ice-cold acetone. After washing, sections were blocked with 20% FBS for at least 2 hours. Then, sections were stained with CD4-bio (RM4.4), followed by SA-AF568, B220-AF647, and PNA-FITC. Sections were imaged using the Eclipse 90i Advanced Automated Research Microscope (Nikon). Scale bars were placed using Nikon software.

Quantitative PCR. Mouse spleens were stored at –80°C in RNAlater (Qiagen) prior to homogenization using a bead homogenizer and extraction with the RNeasy Mini kit (Qiagen). cDNA was then generated using a Superscript Vilo cDNA synthesis kit (Thermo Fisher). Quantitative PCR was then performed using PowerSYBR 2x master mix (Thermo Fisher) and primers indicated in Supplemental Table 2 on a CFX 1000 Thermal Cycler (Bio-Rad). All Ct values were normalized to HPRT, and quantitation was calculated using the delta-delta CT method.

Type I IFN bioassays of mouse cells. Luciferase activity was measured in L929 cells stably expressing an ISRE-luciferase reporter as previously described (25). 5 × 10⁴ L929-ISRE cells were plated at a ratio of 10 splenocytes to 1 reporter cell in a 96-well plate. Cells were lysed and luciferase activity was measured using the Luciferase Assay System (Promega) and Centro LB 960 Microplate Luminometer (Berthold Technologies).

Statistics. When data were parametric, unpaired, 2-tailed Student’s t tests were applied to determine the statistical significance of the difference between groups with Prism 6 software (GraphPad). For analysis of mRNA, a 2-tailed Mann-Whitney U test was utilized, as the data were nonparametric. For analysis of multiple groups with different SDs and nonparametric data, the Kruskal-Wallis test was used with the Dunn post-hoc test for multiple comparisons.

Study approval. All experiments involving animals were performed in accordance with the guidelines of and with the approval of the University of Washington Institutional Animal Care and Use Committee.