AIM2 sensors mediate immunity to Plasmodium infection in hepatocytes

Abstract

Malaria, caused by Plasmodium parasites is a severe disease affecting millions of people around the world. Plasmodium undergoes obligatory development and replication in the hepatocytes, before initiating the life-threatening blood-stage of malaria. Although the natural immune responses impeding Plasmodium infection and development in the liver are key to controlling clinical malaria and transmission, those remain relatively unknown. Here we demonstrate that the DNA of Plasmodium parasites is sensed by cytosolic AIM2 (absent in melanoma 2) receptors in the infected hepatocytes, resulting in Caspase-1 activation. Remarkably, Caspase-1 was observed to undergo unconventional proteolytic processing in hepatocytes, resulting in the activation of the membrane pore-forming protein, Gasdermin D, but not inflammasome-associated proinflammatory cytokines. Nevertheless, this resulted in the elimination of Plasmodium-infected hepatocytes and the control of malaria infection in the liver. Our study uncovers a pathway of natural immunity critical for the control of malaria in the liver.

Keywords: Caspase-1; Malaria; innate immunity; liver.

Fig. 1.

P. falciparum infection induces inflammatory cell-death pathways in hepatocytes. (A) Representative pseudocolored fluorescent micrographs showing in vitro cultured primary human hepatocytes infected with Pf, 4 d p.i. The individual panels show staining with separate Pf-specific antibodies (GAPDH and merozoite surface protein-1, MSP-1) or DNA (DAPI), and fluorescence overlays. MSP-1 expression represents the liver-stage of Pf. (B and C) Transcriptional differences determined by scRNA sequencing of Pf-infected or uninfected primary human hepatocytes at 4 d p.i. Pf-infected hepatocyte transcripts were plotted for principal component analysis, which corroborated the unsupervised clustering of hepatocytes based on gross transcriptional changes (B) or as heat map to depict gross transcriptional differences (C). (D) Significantly enriched network of functional interactions evaluated by gene-set and biochemical pathway enrichment analysis comparing Pf-infected and uninfected human hepatocytes. The nodes represent the key functional outcomes predicted based on the transcriptional identity (represented by color coded proteins, see legend) of the Pf-infected primary human hepatocytes at 4 d p.i. (E) Heat map depicting transcriptional differences in the genes of the canonical inflammasome pathway determined to be up-regulated by unbiased biochemical pathway enrichment analysis in primary human hepatocytes, maintained ex vivo and coincubated with Pf for 4 d. The bar graph on right shows log fold change ± SD (logFC ± SD) in the differential expression of the indicated genes. Dotted line represents logFc of 1. (C and E) Data obtained from a total of four replicate infections (samples a–d). See SI Appendix, Fig. S1A for experimental details.

Fig. 2.

AIM2-Caspase-1 axis drives the control of liver-stage malaria. (A) Representative (of >10 cryosections) pseudocolored confocal image depicting Caspase-1 aggregates (arrows) in human hepatocytes in the liver of a humanized mouse inoculated with Pf (30 h p.i). Stained using DAPI for nucleus, hsp70 for Pf, and anti-hCaspase-1 antibody. Frequency (%) of association of Pf and hCaspase-1: 82.05 ± 2.69. Data presented as mean ± SEM compiled from ≥10 cryosections. Hsp70: heat shock protein 70. (B and C) Scatter plots showing relative liver-parasite burdens in the indicated mice inoculated with Py, at 36 h p.i. (D) Scatter plots showing relative parasite burdens at 36 h p.i in the whole livers of B6 or Casp1/11KO chimeric recipient mice reconstituted with B6 or Casp1/11KO bone-marrow and inoculated with Py i.v. (E) Scatter plots showing relative whole-liver parasite burdens at 36 h p.i. in B6 or AIM2KO chimeric recipient mice reconstituted with B6 or AIM2KO bone-marrow and inoculated with Py i.v. (B–E) The dots in the scatter plots represent individual mice, with the data combined from 3 separate replicate experiments and presented as mean ± SEM and analyzed using ANOVA with Dunnett’s corrections, yielding the indicated P values.

Fig. 3.

AIM2 receptor binds Plasmodium DNA in the infected hepatocytes. (A) Pseudocolored confocal image of a representative field (>10 fields) of BrdU⁺ Pb sporozoite stages (stained using anticircumsporozoite protein, CSP) derived from infected mosquitoes. CSP is located on the plasma membrane of the sporozoites. Arrows indicate BrdU incorporation. Frequency (%) of colocalization of CSP (sporozoites) and BrdU: 94 ± 2.0. Data presented as mean ± SEM from ≥10 microscopy fields obtained from at least 3 separate experiments. (B) Representative (>10 fields) pseudocolored confocal image of BrdU⁺ Pb exoerythrocytic form (EEF, stained using anti-Acyl carrier protein, ACP) in an in vitro cultured primary murine hepatocyte at 24 h p.i. ACP-stained Plasmodium EEF is indicated by the arrow. Frequency (%) of colocalization of ACP and BrdU in such Pb-infected hepatocytes: 91.6 ± 1.42. Data presented as mean ± SEM from ≥10 fields derived from at least 3 separate experiments. (C) ELISA comparing lysates of Pb sporozoites derived from BrdU-fed or control-infected mosquitoes to determine relative BrdU levels. Data presented as mean ± SEM from 3 technical and two biological replicate experiments, compared with t tests to yield the presented P value. (D) Immunoblot analysis for AIM2 after immunoprecipitation of the whole-cell lysates of primary mouse hepatocytes infected with BrdU⁺ Pb (24 h p.i.), using anti-BrdU antibodies. Hepatocytes were either infected with wild-type Pb (BrdU⁻), coincubated with salivary gland extracts from BrdU-fed mosquitoes, or remained uninfected/untreated to serve as controls. Data represent 4 separate replicate experiments. (E) Representative pseudocolored confocal image (>10 fields) of BrdU⁺ Pb exoerythrocytic form in an in vitro cultured primary murine hepatocyte at 24 h p.i. Arrow indicates the developing parasite. Frequency (%) of colocalization of AIM2 and BrdU: 58.50 ± 1.42. Data presented as mean ± SEM from ≥10 fields derived from at least 3 separate experiments.

Fig. 4.

Non-canonical processing of Caspase-1 in hepatocytes. (A and B) Immunoblot screen for Caspase-1 cleavage forms in primary mouse hepatocytes infected with Py (A) or treated with LPS+ATP (B) for the indicated time-frames. Murine BMDMs coincubated with LPS+ATP served as control. Caspase-1 p20 subunit-specific antibodies detect uncleaved procaspase-1 (p46) and the cleaved Caspase-1 products, p32 or p20. (C) Immunoblot analysis for cleaved Caspase-1 in human hepatocytes infected with Pf (Left) or treated with LPS+ATP (Right) for the indicated time-frames. (D) Immunoblot analysis confirming the identity of Caspase-1 pulled down from whole-liver lysates of liver-humanized mice infected with Pf (30 h p.i.), using with anti-p20-specific antibody. (E) Immunoblot analysis for Caspase-1 cleavage in primary mouse-hepatocytes infected with Py and probed with Caspase-1 p20 subunit (Left) or Caspase-1 p10 subunit (Right)-specific antibodies. Murine BMDMs coincubated with LPS+ATP served as the standard for conventional Caspase-1 cleavage pattern, indicating p20 and p10. (F) Immunoblot analysis in whole-cell lysates of primary B6 hepatocytes cocultured with Py for 24 h, immunoprecipitated with anti-mCaspase-1p20 and the precipitate probed with anti-mCaspase-1p10. (A–E) LC represents the protein loading controls from the SDS–PAGE, to provide an estimate of the representation of pro/Caspase-1 in the total protein content of the hepatocytes or BMDMs. Note that these are single exposure blots and due to the high signal intensity, the amount of total protein added from the BMDM lysates was insufficient for detection by Coomassie blue staining in the loading control. All data are representative of at least 3 separate experiments.

Fig. 5.

Caspase-1 activation induces GSDMD-mediated cell-death in hepatocytes. (A) Comparison of cytolysis determined by LDH release assay in ex vivo cultured primary hepatocytes derived from the indicated mice, coincubated with Py for the indicated times. Data presented as mean ± SEM, analyzed using ANOVA with Dunnett’s correction comparing each time point to the corresponding one in B6 mice, to yield the presented P-values. The dotted line indicates median cytolysis levels in B6 hepatocytes treated for 24 h with salivary gland extracts derived from uninfected mosquitoes. (B) Representative confocal time-lapse images showing primary hepatocytes from tdTomato⁺ B6 mice infected with Py (CellTrace Violet⁺, indicated by arrows), observed at 24 h and 32 h of coincubation. Hepatocyte #1 and #2 shown on left just prior to undergoing pyroptotic rupture; please see Movie S1 for the full sequence of events. (C) Immunoblot analysis for GSDMD cleavage in primary hepatocytes obtained from the indicated mice coincubated with Py, at 24 h. BMDMs treated with LPS+ATP (4 h) served as the positive control. LC: loading control. Bar graphs on the right depict the relative densities of cleaved GSDMD, presented as the mean ratio of cleaved: uncleaved GSDMD bands calculated from three separate immunoblots. Data presented as mean ± SEM, analyzed using ANOVA comparing each group to the B6 Py group to yield the presented P-values. (A–C): all data shown represent ≥3 separate experiments. (D) Scatter plots showing relative liver-parasite burdens in the indicated mice inoculated with Py, 36 h p.i. (E) Scatter plots showing relative liver-parasite burdens in the vehicle-treated (Rx ctrl) or Disulfiram-treated (−1 d, 0, 1 d p.i) mice inoculated with Py, 42 h p.i. (D and E) Data presented as mean ± SEM, analyzed with 2-tailed t tests to yield the presented P value and combined from three separate replicate experiments.