C3 deficiency predisposes to impaired pulmonary trained immunity.

(A) UMAP (Uniform Manifold Approximation and Projection) plots showing the identity of each cell cluster in human bronchoalveolar lavage (BAL) from BCG- and saline (Sal)-treated donors at Day 2 (BCG n=17,173; Sal n=23,71S) and Day 7 (BCG n=21,527; Sal n=29,540) post-vaccination. N=3 volunteers in each group. Cell clusters include: macrophage (Mo), activated macrophage (AcMo), non-resident macrophage (nrMo), NK cells, γ δ T cells, plasmacytoid DC (pDC), conventional DC (cDC1, cDC2), migratory DC (McDC), B cells, MAIT cells, CD4, CDS, CD4/S, cytotoxic-like CDS T cells (CtxCD4), terminally differentiated effector memory CD45RA-re-expressing CDS T cells (EMRA CDS), naÏfve T cells (NvCD4, NvCDS), T regulatory cells (Treg), neutrophils, mast cells, doublets, ciliated bronchial epithelial cells (GIBE), secretory bronchial epithelial cells (ScBE), and unidentified cells (Unknown). Data from Marshall et al, 2025. (B) Dot plot and bar chart showing C3 expression across all human alveolar macrophages (Mcp; Mo, AcMo, nrMo pooled) across conditions. Dot size reflects the percentage of cells expressing C3; dot color reflects mean normalized expression. Bar chart shows percentage of C3-expressing M φ per condition. (C) As in (B), for C3AR1. BCG vaccination increases the proportion of C3AR1-expressing alveolar macrophages at both timepoints relative to saline controls. (D) Schematic representing the training of mice via the intranasal route with heat-killed Pseudomonas aeruginosa (HKPA) and subsequent restimulation with lipopolysaccharide (LPS), followed by BAL and cytokine analysis. Created with BioRender. (E) WT untrained compared against WT-trained BAL levels of CXCL1, CXCL2, IL-6, and TNF α. Comparison of BAL C3a levels, similar to (E). (G) WT-trained versus C3-deficient (C3KO)-trained BAL concentrations of IL-6 and TNFα. WT-trained levels derived from (E) for comparison with C3KO-trained mice. Data were compared with two-sided unpaired t-tests with (E,G) or without (F) Holm-Sidak correction for multiple hypothesis testing. Each point represents a measurement from one mouse with at least n=4 in each group, mean± SD shown. *p < 0.05, **p < 0.01, ***p < 0.001.

C3 deficiency results in impaired trained immune responses in ex vivo alveolar macrophages (AMs).

(A) Schematic representing in vitro training of AMs with HKPA, with later stimulation by LPS and subsequent cytokine analysis of the supernatants. Created with BioRender. (B-C) Effects of HKPA-induced training in vitro on IL-6 and TNFα in supernatant from (B) WT AM, and (C) their comparison with C3KO-trained AMs. (D) Schematic representing in vitro training of AMs with HKCA, with subsequent restimulation by LPS and cytokines in supernatant. Created with BioRender. (E) Effects of heat-killed Candida albicans (HKCA)-induced training in vitro on cytokines in supernatant from WT AM. (F) Comparison of C3a levels post-HKCA training, similar to (B). (G) Comparison of IL-6 and TNFα post-HKCA training in WT versus C3KO AMs. WT-trained levels derived from (D) for comparison with C3KO-trained AMs. Data were compared with two-sided unpaired t-tests with (B,C,E,G) or without (F) Holm-Šidák correction for multiple hypothesis testing. Each point is a technical replicate from pooling AMs from at least n=4 mice in each group, with mean ± SD shown. Each experiment repeated twice. *p < 0.05, **p < 0.01, ***p < 0.001.

Alveolar macrophages take up C3 from airspaces of live alveoli in situ.

Cartoons in A-B show the experimental design of the confocal imaging studies shown in C-G. As indicated in A-B, we microinstilled alveolar airspaces of live, intact, perfused mouse lungs sequentially with: cell-permeant calcein red-orange dye (CR); Alexa Fluor 647 (AF), AF-tagged C3 (C3-AF), or AF-tagged C3a (C3a-AF); and phycoerythrin (PE)-tagged anti-CD11c Ab. The confocal image in C shows C3-AF fluorescence (yellow) in alveolar airspaces and CR fluorescence (magenta) in airspace-facing cells, including the alveolar epithelium and alveolar macrophages. alv, example airspace; mv, microvessel. Confocal images in D show the same alveoli, but CD11c fluorescence (cyan) now marks CD11c+ cells. Arrowheads point out example CD11c+ cells with intracellular C3-AF fluorescence. High power confocal images of CD11c+ cells (E-F) and group data (G) show C3-AF accumulated in cytosols of CD11c+ cells over time. In G, circles indicate mean ± SEM fluorescence in all of the CD11c+ cells present in imaging fields of at least 30 alveoli; n = 4 microinstillations in 2 lungs per group; *p < 0.05 versus C3-AF by ANOVA with post hoc Tukey testing. C3-AF, C3a-AF, and AF fluorescence in airspaces was normalized to C3-AF, C3a-AF, and AF fluorescence in glass micropipettes. Scale bars: 100 (C-D) and 10 (E-F) µm.

C3 uptake enhances trained immune responses in ex vivo alveolar macrophages (AMs) via the C3a receptor (C3aR).

(A) Schematic representing in vitro training of AMs with HKCA, with pre-treatment of C3 or C3a prior to induction of training, and later stimulation by LPS and subsequent cytokine analysis of the supernatants. Created with BioRender. (B) Effects of adding C3 prior to training on IL-6 and TNFα levels from C3KO AMs and their comparison with WT-trained AMs. (C)Effects of adding C3a prior to training, similar to (B).(D) Schematic representing addition of the C3aR antagonist prior to C3 treatment and in vitro training of AMs with HKCA, with later stimulation by LPS and subsequent cytokine analysis of the supernatants. Created with BioRender. (E) Effects of C3aR antagonism on IL-6 and TNFα levels from trained WT and C3KO AMs treated with exogenous C3. (F) Comparison of IL-6 levels post-HK-CA-training in C3aR-deficient (C3aRKO), C3KO and WT AMs treated with exogenous C3. Data were compared using one way ANOVA with Dunnett’s post hoc tests (B,C,F) or two-sided unpaired t-testing with Holm-Šidák correction for multiple testing (E). Each point is a technical replicate made by pooling AMs from at least n=4 mice in each group, with mean ± SD shown, and each experiment was repeated twice. *p < 0.05, **p < 0.01.

C3-C3aR axis is required for glycolysis as a part of trained immune responses in alveolar macrophages (AMs).

(A) Principal component analysis (PCA, left) and EnrichR analysis of 391 genes (right, Table S1) downregulated in HKCA-trained C3KO vs WT AM by filtering genes (FDR step up ≤0.05). Arrow shows metabolism gene set in EnrichR; bars ranked by p-value. (B) Schematic representing in vitro training of AMs with HKCA. Created with BioRender. (C) Extracellular acidification rate (ECAR) from Seahorse analysis representing full glycolytic activity, and basal and maximum glycolysis in untrained and HKCA-trained WT and C3KO AMs. (D) Schematic representing addition of the C3aR antagonist (SB290157) prior to C3 treatment and in vitro training of AMs with HKCA. Created with BioRender. (E) Seahorse analysis in the presence and absence of exogenous C3 supplementation and C3aR antagonism. Each point is a technical replicate of pooled AMs from at least n=4 mice in each group, with mean ± SD shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 when analyzed using one way ANOVA with Dunnett’s post hoc tests.