Fungal microbiota sustains lasting immune activation of neutrophils and their progenitors in severe COVID-19

Abstract

Gastrointestinal fungal dysbiosis is a hallmark of several diseases marked by systemic immune activation. Whether persistent pathobiont colonization during immune alterations and impaired gut barrier function has a durable impact on host immunity is unknown. We found that elevated levels of Candida albicans immunoglobulin G (IgG) antibodies marked patients with severe COVID-19 (sCOVID-19) who had intestinal Candida overgrowth, mycobiota dysbiosis and systemic neutrophilia. Analysis of hematopoietic stem cell progenitors in sCOVID-19 revealed transcriptional changes in antifungal immunity pathways and reprogramming of granulocyte myeloid progenitors (GMPs) for up to a year. Mice colonized with C. albicans patient isolates experienced increased lung neutrophilia and pulmonary NETosis during severe acute respiratory syndrome coronavirus-2 infection, which were partially resolved with antifungal treatment or by interleukin-6 receptor blockade. sCOVID-19 patients treated with tocilizumab experienced sustained reductions in C. albicans IgG antibodies titers and GMP transcriptional changes. These findings suggest that gut fungal pathobionts may contribute to immune activation during inflammatory diseases, offering potential mycobiota-immune therapeutic strategies for sCOVID-19 with prolonged symptoms.

Extended Data Fig.1. Serological testing in Cohorts 1 and 2.

a-b, SARS-CoV-2 RBD IgG titration curves in HD (n=36, a) and sCOVID-19 (n=66, b). c, Plasma IgG antibody titers to SARS-CoV-2 RBD in HD (n=36), mCOVID-19 (n=25) and sCOVID-19 (n=66) (Supplementary Table. 1). The data are shown as endpoint titers normalized to ELISA reciprocal dilution as in a and b. The dotted line indicates the limit of detection. In the boxplots, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. Statistical significance was determined by the one-way ANOVA followed by Tukey's multiple-comparison. Related to Figure 1

Extended Data Fig.2.. Compositional analysis of gut mycobiota.

a-b, Ratio between Ascomycota and Basidiomycota (a) and relative abundance of fungal species (b) in ITS1 sequencing of fungal rDNA from stool samples of HD (n=10) and COVID-19 patients (n=10). Lower and upper hinges correspond to the first and third quartile; dots represent individual patients' samples. P values were calculated using a two-tailed Mann-Whitney testing between all groups. Related to Figure 2.

Extended Data Fig.3.. Immune responses to C.albicans isolates from COVID-19 patients.

a, Representative graphs depicting the flow cytometry gating strategy for defining murine neutrophil populations in tissues. b-d, Anti-C. albicans specific IgG titers (b), the amount of neutrophils in peripheral blood (c) and lung (d) in antibiotic-treated mice orally gavaged or not (PBS, n=7) with either C. glabrata (CgCOV3, n=7) or C. albicans (CaCOV1, CaCOV5, CaCOV2, n=7) isolated from COVID-19 patient's stool . Immune responses were assessed at 2 weeks after colonization. In boxplots in b, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. The bar graphs in c and d presented as mean ± SEM. The results were pooled from two experiments. P values were calculated using the one-way ANOVA followed by Tukey's multiple comparison. ns=not significant. Related to Figures 2 and 3.

Extended Data Fig.4.. Linear regression analysis of the Immune cell frequencies and levels of ACAL IgG in peripheral blood of mCOVID-19 and sCOVID-19 s.

a-e, Comparison of linear regression analysis of CD3⁺ T cell (a), CD19⁺CD20⁺ B cell (b), NK cell (c), CD4⁺ T cell (d) and CD8⁺ T cell (e) frequencies and levels of ACAL IgG in peripheral blood of sCOVID-19 (n=13) and mCOVID-19 (n=17), see Supplementary Table 1. Crosses indicate deceased patients. Red and black lines show linear fits within severe and low to moderate groups, respectively, with 95% confidence intervals shown in gray. Spearman correlation estimates (rho) and associated p values are shown in red and black for sCOVID-19 and mCOVID19, respectively. Related to Figure 3.

Extended Data Fig.5.. Strong correlation between ACAL IgG and GMPs among multiple progenitor cell types.

Correlation and linear regression of frequency of hematopoietic stem cells/multipotent progenitor cells (HSC/MPP), lymphoid-primed multipotent progenitor cells (LMPP), granulocyte-macrophage progenitor cells (GMP), erythroid progenitor cells (Ery), megakaryocyte-erythroid progenitor cells (MEP) with ACAL IgG (log10 titer) in enrichment of CD34⁺ HSPC from PBMC of HD(n=5), sCOVID-19 (n=12) and nonCOV-19 (n=5), see Supplementary Table 2. Blue line indicates the regression line for all patients. The associated linear regression equation, Pearson’s correlation. Coefficient, and significance are shown. Related to Figure 4.

Extended Data Fig.6.. HLA-related transcriptional signatures of GMPs form HD, sCOVID19 and sCOVID19 treated with tocilizumab.

a, Heatmap showing antigen presentation marker that are differentially expressed in GMP of HD(n=7) and sCOVID-19 who received(n=6) or not (n=7) IL-6R blockade treatment. Data are average of normalized expression for each gene in each group. b, Antibiotic treated mice were colonized with C.albicans CaCOV5 by oral gavage twice for two weeks prior to SARS-COV-2 challenge and were harvested at Day 20. Water with or without antifungal fluconazole was provided to mice two days after second oral gavage. Related to Figure 5.

Figure 1.. Antibodies to common intestinal fungi are elevated in COVID-19 patients.

a, Plasma antibody titers of circulating IgG to fungal mannan (Saccharomyces cerevisiae; ASCA) in Crohn's disease patients (CD, n=40), healthy individuals (HD, n=36) and COVID-19 patients (COVID-19, n=91). b, c, Plasma IgG antibody titers to C. albicans IgG (ACAL IgG), C. parapsilosis IgG and S. cerevisiae IgG (b) or A. fumigatus IgG, A. alternata IgG and M. restricta IgG (c) in HD (n=36), mCOVID-19 (n=25) and sCOVID-19 (n=66) (Supplementary Table 1). Mild and moderate patients were grouped together as mCOVID-19 for analysis against sCOVID-19 or HD. d, Plasma IgG antibody titers to C. albicans and A. fumigatus IgG in sCOVID-19 (n=28) and non-COVID-19 patients admitted in the intensive care unit (nonCOV, n=5) (Supplementary Table. 2). e. Longitudinal assessment of plasma IgG antibody titers to ASCA, C. albicans (ACAL) and A. fumigatus IgG in mCOVID-19 (n=6) and sCOVID-19 (n=6). Samples are collected at clinical diagnosis of infection or first administration (V1) and during the development of acute disease stage (V2; more than two weeks post-V1). All data are shown as endpoint titers normalized to ELISA reciprocal dilution. The dotted line indicates the limit of detection. For all boxplots, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. Statistical significance was determined by the one-way ANOVA followed by Tukey's multiple-comparison in a-c, a two-tailed Mann-Whitney test in d, and a two-tailed Wilcoxon matched-pairs singled-rank test in e. ns=not significant.

Figure 2.. Candida species of the gut mycobiota expand in sCOVID-19.

a, b, Stack plot of relative abundances at the genus level for the 17 fungal genera with the highest average abundance (a) or the 9 bacterial genera with the highest average abundance (b) assessed by ITS1 sequencing of fungal rDNA or 16S sequencing of bacterial rDNA from stool samples of HD (n=10) and sCOVID-19 (n=10) in cohort 3 (Supplementary Table 3 and 4). c,d, Relative abundance of Candida (c) or Clostridiales and Lachnospiraceae (d) as in a. Lower and upper hinges correspond to the first and third quartile. e, Linear regression analysis of Candida species (CFU/g of stool) in COVID-19 patient stool (n=10) and disease severity as defined by an eight-category ordinal scale score (see Methods). f, Representative flow cytometry plots for Multi-KAP-based assessment of fungal biomass (CFW⁺ Sybr Green⁺) and the percentage of CFW⁺ Sybr Green⁺ cells per total event from fecal samples of COVID-19 (n=10) and HD (n=10) in cohort 3 (Supplementary Table 3 and 4). Insets indicate percentage of cells within the gate in representative plots. g, C. albicans specific IgG titers in the serum of antibiotic-treated mice two weeks after oral gavage with either PBS (n = 7), C. albicans (COVCa5, n = 8) or C. glabrata (COVCa3, n = 7) strains isolated from COVID-19 patient's stool. The results were pooled from two experiments. For all boxplots, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. Plots represent individual patients (b, d, e, f) and mice (g). P values were calculated using a two-tailed Mann-Whitney test in b, d and f and the one-way ANOVA followed by Tukey's multiple-comparison in g. ns=not significant.

Figure 3.. Neutrophils and ACAL IgG differentiate mCOVID-19 from sCOVID-19, and link intestinal Candida overgrowth to proinflammatory immunity in the lung.

a, Calprotectin plasma levels in HD (n=36), mCOVID-19 (n=25) and sCOVID-19 (n=66) from cohort 1 (Supplementary Table 1). b-d, Scatter plots of lymphocytes and myeloid cells (b) and Lin⁻CD11b⁺CD14⁺CD15⁻ monocytes (MO), Lin⁻CD11b⁺CD14⁻CD15⁻ monocyte-derived cells (MC) and Lin⁻CD11b⁺CD14⁻CD15⁺ polymononuclear neutrophils/ granulocytes (PMNs) frequencies (c) in the blood (b,c) and calprotectin levels in the plasma (d) against levels of ACAL plasma IgG in sCOVID (n=13) and mCOVID-19 (n=17) from cohort 1(Supplementary Table 1). Crosses indicate deceased patients. Red and black lines show linear fits within severe and low to moderate groups, respectively, with 95% confidence intervals shown in gray. Spearman correlation estimates (rho) and associated p values are shown in red and black for sCOVID-19 and mCOVID-19, respectively. e. Representative plots and the number of CD45⁺Ly6G⁺CD11b⁺ neutrophils in peripheral blood and lung at week 2 after intestinal colonization of antibiotic-treated mice with either C. albicans (COVCa5, n=8) or C. glabrata (COVCa3, n=7), or oral gavage with PBS (n = 7). Insets indicate the percentage of cells within the gate. The results were pooled from two experiments. f, Serum calprotectin and the number of Ly6G⁺CD11b⁺ neutrophils in peripheral blood and lung at week 2 in mice that received oral gavage with PBS (n=5), COVCa5 (n=5) or COVCa5 followed by treatment with fluconazole two days after COVCa5 colonization (COVCa5 + fluconazole, n=5). Data in f are representative of two experiments. For boxplots in a, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. The bar graphs presented as mean ± SEM. Plots represent individual patients (a-d) and mice (e, f). Correlation analysis in b-d were assessed using the Spearman correlation coefficient with a two-tailed test. Differing correlations between severe versus mild/moderate groups were assessed by evaluating the significance of an interaction. P values were calculated using the one-way ANOVA followed by Tukey's multiple-comparison in a and Kruskal–Wallis test with a Dunn’s posttest in e and f. ns=not significant.

Figure 4.. Persistently altered function of neutrophil progenitors in patients recovering from sCOVID-19.

a, b, Plasma IgG antibody titers to C. albicans (ACAL IgG), A. fumigatus and flagellin (a) or calprotectin levels (b) in HD (n=8) and patients recovering from mCOVID-19 month 2-4 (n=7), mCOVID-19 month 4-12 (n=5), sCOVID-19 month 2-4 (n=10) and sCOVID-19 month 4-12 (n=18) (Supplementary Table 2). c, Subject-paired analysis of PBMCs (left) and HSPC subsets (right) with enrichment of CD34⁺ HSPC from PBMC followed by combined single-nuclei RNA/ATAC-seq (Multiome). Plots in HSPC were annotated for major progenitor cell types; erythroid progenitor cells (Ery), basophil-eosinophil-mast cell progenitor cells (BEM), lymphoid-primed multipotent progenitor cells (LMPP), megakaryocyte-erythroid progenitor cells (MEP), hematopoietic stem cells/multipotent progenitor cells (HSC/MPP), granulocyte-macrophage progenitor cells (GMP). d, Heatmap representation of differentially expressed genes involved in antifungal immunity and signaling by HSPC from HD (n=5) and sCOVID-19 (n=12). e, Boxplots representing the frequency of GMPs in enriched CD34⁺ HSPC from HD (n=5) and sCOVID-19 (n=12). The data in e and f represents normalized expression by sctranform. f, Correlation and linear regression of anti-fungal (C. albicans or A. fumigatus) or anti-bacterial (flagellin) IgG antibodies (log10 titer) or calprotectin levels (log10 concentration) with frequency of GMP. Each dot represents HD (5), sCOVID (n=12) and nonCOV (n=5). Blue line indicates the regression line for all patients. The associated linear regression equation, Pearson’s correlation. Coefficient, and significance are shown. g, Violin plots showing distribution of MPO expression in GMPs at month 2-4 (n=4) and month 4-12 (n=8) post hospital admission of sCOVID-19 versus HD (n=5). For boxplots in a, b and e, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. Dots represent individual patients (a, b, e, f) and cells (g). P values were calculated using the one-way ANOVA followed by Tukey's multiple-comparison in a, b and a two-tailed Mann-Whitney test in e and Kruskal–Wallis test with a Dunn’s posttest in g. ns=not significant.

Figure 5.. IL-6 mediates immune activation in severe COVID-19 patients and in a mouse model of COVID-19 during C. albicans-aggravated lung NETosis.

a, Representative dot plots of lung CD45⁺CD11b⁺Ly-6G⁺ neutrophils from COVCa5-colonized mice treated with IL-6R (n=7) or isotype (n=7) antibodies every other day for two weeks post-colonization or were left uncolonized and untreated (PBS, n=6). Insets indicate percentage of cells within the gate. b, Absolute numbers and c, percentage of CD45⁺CD11b⁺Ly-6G⁺ neutrophils in lung CD45⁺ cells, as in a. d, Plasma IgG antibody titers to C. albicans and flagellin in sCOVID-19 month 2-4 and sCOVID-19 month 4-12 patients who received IL-6R blockade treatment during acute infection (IL-6R Ab, n=10) or not (no treatment, n=7). The results were pooled from two experiments. e, Heatmap of genes involved in fungal recognition in blood GMPs isolated from sCOVID-19 who received IL-6R blockade treatment (IL-6R Ab sCOVID, n=6) or not (sCOVID, n=7) during acute COVID-19 infection or from HD (n=7). f-h, Number of Ly6G⁺CD11b⁺ cells in the peripheral blood as assessed by flow cytometry (f) and immunofluorescence images (g) and quantification (h) of H3Cit⁺, MPO⁺ and Ly-6G⁺ neutrophils from multiple fields (20×) of lung sections (g) at day 5 post intranasal inoculation with SARS-COV-2 in mice that were uncolonized (PBS, n=5) or colonized with COVCa5 at day -14 before infection (Extended Data Figure 6b) and left untreated (n=5) or treated with fluconazole (n=5) two days after colonization for the duration of the experiment. White scale bar (g) indicates 30 μm. Image analysis was conducted on acquired images as follows: two slides per animal of n = 3-5 animals. The data are representative of two experiments. For boxplots in c,d, the center is drawn through the median of the measurement, and the lower and upper bounds of the box correspond to the first and third quartile. The whiskers go down to the smallest value and up to the largest. The bar graphs in b, f and h presented as mean ± SEM. Dots represent individual patients (d) and mice (b, c, f, h). P values were calculated using the one-way ANOVA followed by Tukey's multiple-comparison in b, c, f and h and a two-tailed Mann-Whitney test in d. ns=not significant.