Candida albicans extracellular vesicles trigger type I IFN signalling via cGAS and STING

Abstract

The host type I interferon (IFN) pathway is a major signature of inflammation induced by the human fungal pathogen, Candida albicans. However, the molecular mechanism for activating this pathway in the host defence against C. albicans remains unknown. Here we reveal that mice lacking cyclic GMP–AMP synthase (cGAS)–stimulator of IFN genes (STING) pathway components had improved survival following an intravenous challenge by C. albicans. Biofilm-associated C. albicans DNA packaged in extracellular vesicles triggers the cGAS–STING pathway as determined by induction of interferon-stimulated genes, IFNβ production, and phosphorylation of IFN regulatory factor 3 and TANK-binding kinase 1. Extracellular vesicle-induced activation of type I IFNs was independent of the Dectin-1/Card9 pathway and did not require toll-like receptor 9. Single nucleotide polymorphisms in cGAS and STING potently altered inflammatory cytokine production in human monocytes challenged by C. albicans. These studies provide insights into the early innate immune response induced by a clinically significant fungal pathogen.

Candida albicans bloodstream infections are a dreaded complication of modern medicine, leading to mortality rates exceeding 50%^1,2. Treatment of this deadly infection is limited due to the eukaryotic nature of this pathogen and requirement of sterilizing immunity, coupled with a lack of mechanistic understanding of how C. albicans communicates with the host immune system. Transcriptomic analysis in human peripheral blood mononuclear cells (PBMCs) identified the type I interferon (IFN) signature profile in response to C. albicans infections³. In addition to bloodstream infections, mucosal infection models demonstrate dynamic IFN responses using multiple Candida species⁴. Known IFN-stimulated genes (ISGs), MDA5, and the IFN receptor, IFNAR, are involved in the immune response to C. albicans^5–7. The concept that C. albicans triggers a potent and seemingly antiviral-type IFN response is novel, and the mechanism of activation, including the full repertoire of fungal ligands involved, has not been discovered. The stimulator of IFN genes (STING) pathway is a major IFN-producing pathway known to respond to microbial pathogens^8,9. Recently, STING has been shown to bind to Src to inhibit Syk-mediated signalling in dendritic cells challenged with C. albicans¹⁰. STING pathway activation can also occur upon cyclic GMP–AMP synthase (cGAS) sensing of double-stranded DNA in the cytosol^11–18, triggering a signalling cascade that activates the IFN regulatory factor 3 (IRF3) transcription factor to transcribe early ISGs and type I IFNs (IFNβ and IFNα)¹⁹.

C. albicans and other pathogenic fungi secrete extracellular vesicles (EVs) that contain effector molecules, virulence factors, surface proteins and nucleic acids in order to transfer information and facilitate crosstalk between fungal cells^20–24. C. albicans EVs present themselves as strong candidates for the stimulating agent of the STING pathway as they contain nucleic acids, elicit a pro-inflammatory immune response, are major constituents of biofilms (BFs) and fungal masses, and can be internalized by immune cells^23,24. EVs containing DNA secreted from the gut microbiota are known to activate type I IFN and trigger peripheral cGAS–STING²⁵. We hypothesized that C. albicans EVs contain the ligand that triggers cGAS- and STING-dependent type I IFN signalling in the cytoplasm of macrophages before the host cell physically encounters a fungal organism. We further hypothesized that host cells, like macrophages, hijack fungal EVs to rewire and prime for an impending pathogen challenge.

In this Article, we reveal a strong connection between the STING innate immune signalling pathway in macrophages and C. albicans. Together, the data reveal a model of host cell re-programming that occurs as innate immune cells encounter C. albicans EVs before pathogen engagement. These findings are the first to present a mechanism of cGAS–STING pathway activation in macrophages by a fungal pathogen and identify a role for EVs in priming of host cells for an impending infection.

Results

cGAS and STING deletion improves survival

To determine if the STING pathway is critical for the host defence against C. albicans, we challenged wild-type (WT), cGas^−/− and Sting^−/− mice with intravenous C. albicans using an in vivo model of systemic candidiasis. Mice were infected with a lethal dose of C. albicans (SC5314), and survival was measured for >3 weeks (cGAS^−/−, 28 days; Fig. 1a, and STING^−/−, 22 days; Fig. 1b). In both experiments, C. albicans infection of WT mice resulted in 100% mortality before the conclusion of the experiment, whereas 60% of the mice lacking either cGAS or STING survived. These data demonstrate that mice lacking either of these integral components of the STING pathway had increased resistance to a lethal dose of C. albicans.

Fig. 1 |. Deletion of cGAS or STING renders mice more resistant to intravenous C. albicans.

a,b, WT, cGas^−/− (a), and Sting^−/− (b) mice were infected intravenously with 100,000 or 150,000 C. albicans yeast, respectively. The data are presented as Kaplan–Meier survival curves representative of three experiments (n = 14 for each group, and significance was determined using a log-rank test (Mantel–Cox) **P = 0.0041, ***P < 0.001 versus WT, data are presented as mean values ± standard deviation). c, Quantification of renal fungal burden following a C. albicans infection in WT and Sting^−/− mice (n = 8 for WT and 14 for Sting^−/− for each timepoint). d, Histological analysis of kidney sections from WT and Sting^−/− mice at day 5 following a C. albicans infection. Host immune cells and fungal cells were identified by H&E and GMS staining, respectively. Scale bar, 100 μm. e, Efficacy of fungal killing by WT, cGas^−/− and Sting^−/− immortalized macrophages to kill C. albicans for 2 h at varying ratios. Data calculated by two-way ANOVA and subsequent Dunnett’s multiple comparison test. Data are presented as mean values from three biological replicates ± standard deviation, ****P < 0.0001 versus WT.

STING facilitates renal fungal clearance

To examine the impact of STING pathway activation by C. albicans, we investigated the ability of WT and Sting^−/− mice to clear this pathogen by assessing fungal burden in the kidneys. We observed that C. albicans-infected Sting^−/− mice had higher fungal kidney burden compared with WT mice. This high fungal kidney burden persists up to day 28 in Sting^−/− mice that survived the course of infection (Fig. 1c). While some fungal organisms were detectable in other organs (liver, spleen and brain), this difference in fungal burden was exclusive to the kidneys of the surviving Sting^−/− mice (Extended Data Fig. 1a). WT and Sting^−/− mice displayed similar blood urea nitrogen (BUN) and creatinine levels over the course of infection, indicating that altered kidney function does not explain the enhanced survival of the Sting^−/− mice (Extended Data Fig. 1b,c). The increased presence of fungal organisms in Sting^−/− mice compared with WT was confirmed by Grocott’s methenamine silver (GMS) staining of kidney tissue to visualize C. albicans (Fig. 1d and Extended Data Fig. 1h). Kidney sections from day 5 of WT and Sting^−/− mice infected with C. albicans were also stained with haematoxylin and eosin (H&E) to assess immune cell infiltration. While the overall number of immune cells appeared similar between the two groups, we did observe altered organization of the immune infiltration with severe abscesses in the medulla of WT kidneys, but not in the Sting^−/− kidneys (Fig. 1d). To complement the histology, quantification of immune cell infiltrate in the kidneys was performed at day 5 post-infection using a multi-parametric flow-based immunophenotyping assay and subsequent gating strategy to identify immune cell populations (Extended Data Fig. 1d)^26,27. We observed similar immune cell recruitment in the WT and Sting^−/− kidneys (Extended Data Fig. 1e,f). In addition to the dramatic differences in immune cell localization in the kidneys, we observed that the effector functions of these cells differed in their ability to induce a known downstream output of the STING pathway, namely viperin (Extended Data Fig. 1g).

To further dissect the role of an active STING pathway in innate immune cells clearance of C. albicans, we tested fungal metabolic activity as a readout for viability following a 2 h co-culture with WT macrophages or macrophages lacking either cGAS or STING. We observed a significant increase in fungal cell death in the WT macrophage co-culture (Fig. 1e). Regardless of effector to target ratio, cGas^−/− and Sting^−/− macrophages killed C. albicans less effectively indicating a critical role for the STING pathway in fungal clearance. These paired in vivo and in vitro findings reveal an integral role for the STING pathway in reducing C. albicans burden during infection.

C. albicans EVs trigger the STING pathway

The critical role for cGAS and STING in responding to C. albicans suggested that fungal DNA activates the cytosolic sensor, cGAS. A known source of fungal nucleic acids is the extra-organismal content of EVs^22,23; hence, we examined the role of C. albicans EVs as the source of the DNA ligand for STING pathway activation. Induction of viperin, one of the well-studied ISGs, and IFNβ secretion was used as a model of activation of the cGAS–STING pathway and assessed in WT, cGas^−/− and Sting^−/− macrophages following a 6-h stimulation with C. albicans EVs^13,28. As ergosterol is a major constituent of the lipid bilayer of fungal EVs, we included an ergosterol only control²⁰. We observed a significant increase in both viperin and IFNβ levels in WT macrophages stimulated with EVs. We did not see this significant increase in cGas^−/− and Sting^−/− macrophages (Fig. 2a,b). Importantly, both WT and cGas^−/− macrophages produced high levels of IFNβ when stimulated with cGAMP (Fig. 2b), a downstream secondary messenger molecule produced when cGAS binds to DNA¹³. EVs from other Candida species, including Candida auris, also activated the STING pathway, as there was a significant increase in viperin and IFNβ levels in stimulated WT macrophages (Extended Data Fig. 2a,b). C. auris EVs grown from an agar plate, although smaller in size than C. albicans EVs grown in the same way, induced viperin and IFNβ to similar levels (Extended Data Fig. 2a–c).

Fig. 2 |. C. albicans EVs and BF DNA activate the STING pathway.

a, Immunoblot of viperin in WT, cGAS^−/− and Sting^−/− macrophages in response to PBS, cGAMP, C. albicans EVs or ergosterol. b, Secretion of IFNβ by WT, cGas^−/− and Sting^−/− macrophages stimulated with PBS, cGAMP, C. albicans EVs and ergosterol. c, Immunoblot of viperin induction in WT, cGAS^−/− and Sting^−/− macrophages following a stimulation with PBS, cGAMP and EVs extracted from either C. albicans BF grown for 48 h or C. albicans planktonic culture. d, IFNβ production by WT, cGas^−/− and Sting^−/− macrophages, following a stimulation with PBS, EVs extracted from either a C. albicans BF grown for 48 h, or C. albicans planktonic culture. e. Representative microscopy images of cGAS–GFP macrophages with and without stimulation with DiI-labelled C. albicans BF EVs (DIC = differential interference contrast). Macrophages were stimulated for 3 h before imaging. f. Quantification of nuclear versus non-nuclear cGAS–GFP localization patterns in unstimulated and stimulated cells. g,h, Immunoblot of viperin (g) and IFNβ production in response to non-transfected (NT) and transfected (TF) cGAMP, BF DNA and Benzonase (Benz.)-treated BF DNA (h). i, Immunoblot of phosphorylated IRF3, total IRF3, phosphorylated TBK1, total TBK1 and actin in response to TF cGAMP and BF DNA in WT, cGas^−/− and Sting^−/− macrophages. j, Immunoblot of viperin in WT cells treated with Dynasore or vehicle control of dimethyl sulfoxide (DMSO) before stimulation with PBS, C. albicans EVs and TF C. albicans BF DNA. Macrophages were stimulated for 6 h before processing. Two-way (b and d) and one-way (h) ANOVAs with subsequent Dunnett’s multiple comparison tests are shown for ELISAs. Data are presented as mean values from three biological triplicates ± standard deviation, with adjusted (adj.) P values. *adj. P = 0.0143, **adj. P = 0.012, ***adj. P = 0.0007, ****adj. P < 0.0001 versus untreated control (PBS or Lipofectamine). All western blots were done in biological triplicate, representative blot shown.

To assess the contribution of other fungal recognition receptors in the response to C. albicans EVs including toll-like receptor (TLR)9, Dectin-1 and caspase recruitment domain-containing protein (CARD)9^29–31, we stimulated Tlr9^−/−, Dectin1^−/− and Card9^−/− macrophages with C. albicans EVs and measured viperin and IFNβ. We observed a significant increase in both viperin and IFNβ compared with the phosphate-buffered saline (PBS) control indicating that TLR9, Dectin-1 and CARD9 are not needed to facilitate this signalling response (Extended Data Fig. 2d,e).

We next assessed the ability for EVs extracted from different sources to induce the STING pathway. We identified the same cGAS- and STING-dependent viperin and IFNβ induction when macrophages were stimulated with increasing concentrations of EVs collected from C. albicans biofilms (BF EVs) grown for 48 h and to a lesser extent in EVs collected from C. albicans yeast (planktonic) (Fig. 2c,d). Importantly, we also observed that both viperin and IFNβ production in response to BF EVs was dose dependent (Extended Data Fig. 2f,g). C. albicans BF EVs are, on average, smaller than EVs prepped from agar plates or planktonic culture (Extended Data Fig. 2c); however, the pathway activation they induce remains strong.

Due to the robust effect of BF EVs on pathway activation, we investigated the cellular changes occurring in response to macrophage internalization of EVs. cGAS localization is a dynamic and reactive process such that it translocates intracellularly from the nucleus to the cytosol to sense cytosolic DNA^32–34. In an unstimulated macrophage, cGAS is tethered tightly to the nuclear membrane to prevent autoreactivity³⁴. However, the localization of this DNA sensor changes in response to infection and stimulation³⁵. To assess whether C. albicans BF EV stimulation alters cGAS subcellular localization within macrophages, we labelled the lipid membranes of EVs with DiI and co-cultured them with WT macrophages expressing cGAS–green fluorescent protein (GFP). We observed a significant increase in non-nuclear localization patterns in macrophages with internalized EVs (Fig. 2e,f). We did not observe this phenotype when treating with a PBS sample exposed to DiI labelling (Extended Data Fig. 3). These data indicate that C. albicans EV stimulation results in the activation of the STING pathway and dynamic subcellular redistribution of cGAS within macrophages.

C. albicans BF DNA activates the STING pathway

Since the STING pathway senses foreign double-stranded DNA, we characterized C. albicans EV DNA. First, we assessed whether DNA was present on the exterior surface of EVs by treating with Benzonase (DNase)³⁶ before stimulation of WT macrophages. When comparing Benzonase-treated EVs with untreated EVs, viperin was not differentially induced indicating that the stimulating DNA is not freely exposed (Extended Data Fig. 4a). Next, we extracted DNA from EVs and transfected it into WT macrophages. Even at low concentrations, EV DNA was sufficient to induce viperin, supporting the hypothesis that C. albicans EVs contain DNA that triggers STING pathway activation (Extended Data Fig. 4a). To ensure that the DNA packaged within C. albicans EVs was indeed fungal, we sequenced DNA extracted from BF EVs. We identified the eight most abundant genes in this EV DNA (Supplementary Table 1). Importantly, these genes are C. albicans specific and none has murine or human orthologues, which provides unambiguous insight into the origin of this DNA.

Approximately 5% of C. albicans BFs is composed of extracellular DNA hypothesized to be packaged in EVs^37,38. Therefore, we extracted DNA from C. albicans BFs (BF DNA) grown for 48 h, 72 h and 96 h, then transfected concentrated BF DNA into WT macrophages for 6 h. We observed robust viperin induction and IFNβ production regardless of BF growth time (Extended Data Fig. 4b,c). Since the 72 h timepoint demonstrated the most robust induction, we conducted all subsequent BF DNA stimulations with BF DNA grown for 72 h.

We next assessed the transfection and DNA-dependency of viperin and IFNβ induction. We stimulated WT macrophages with cGAMP (transfected into cells or not), BF DNA (transfected or not) or BF DNA treated with Benzonase before transfection. Only cGAMP and transfected BF DNA induced viperin and IFNβ (Fig. 2g,h). To determine whether this induction was dependent on cGAS and STING, we assessed viperin and IFNβ levels in WT, cGas^−/− and Sting^−/− macrophages following transfection with C. albicans BF DNA. As with C. albicans EVs, we observed significant increases in viperin and IFNβ in WT macrophages stimulated with BF DNA and cGAMP, whereas the cGas^−/− and Sting^−/− macrophages failed to produce either output (Fig. 2i). Furthermore, BF DNA stimulation of WT macrophages resulted in phosphorylated TANK-binding kinase (TBK1) and IRF3 indicating that downstream elements of the STING pathway are activated in response to BF DNA. cGas^−/− and Sting^−/− macrophages failed to induce phosphorylation of either protein, as well as IFNβ production, indicating that the BF DNA pathway activation is cGAS- and STING-dependent (Fig. 2h and Extended Data Fig. 4d,e). Moreover, we observed significant increases in viperin and IFNβ levels in Tlr9^−/−, Dectin1^−/− and Card9^−/− macrophages stimulated with BF DNA, indicating that these proteins are dispensable for activation of this pathway (Extended Data Fig. 4d,e). Finally, pre-treatment of WT macrophages with Dynasore, an inhibitor of dynamin-dependent endocytosis, resulted in a complete ablation of viperin induction in cells, subsequently stimulated with both C. albicans EVs and transfected BF DNA (Fig. 2j). These data support a model where endocytosed EVs deploy their DNA cargo into the cytoplasm where it is then sensed by cGAS–STING.

Due to the potential for pleiotropic effects of a full knockout cell line, we treated WT macrophages with increasing concentrations of the STING inhibitor H-151^39,40. Similarly to the Sting^−/− macrophages, this specific drug treatment reduced levels of viperin, phospho-TBK1, and phospho-IRF3 in cells stimulated with cGAMP, BF DNA and EVs, but did not reveal as significant of a reduction in LPS-stimulated cells, as LPS triggers STING-independent IFN signalling (Extended Data Fig. 4f,g).

Host response to C. albicans EVs and BF DNA

The ability of both C. albicans EVs and BF DNA to elicit induction of known ISGs in a cGAS- and STING-dependent manner prompted us to assess the extent of the ISG induction. We stimulated WT, cGas^−/− and Sting^−/− macrophages with PBS and EVs for 6 h, extracted RNA from these immune cells and assessed the expression of 772 host response genes, which included 440 known ISGs. Using nCounter, we identified 73 genes induced by EVs as compared with the PBS control, including numerous early ISGs (that is, IFIT1, RSAD2 (viperin), OASL1, ISG15, IFIH1/MDA5, DHX58 and DDX58/RIG-I) (Fig. 3a and Supplementary Table 2a). Transcripts from early ISGs are known to be induced within 3–6 h of infection, independent of IFNβ production⁴¹. In parallel, RNA from macrophages stimulated with Lipofectamine reagents alone or with transfected BF DNA was analysed against the same host response panel. BF DNA specifically induced 18 genes in WT macrophages (Supplementary Table 2b), including the early ISGs: IFIT1, Rsad2RSAD2, OASL1, ISG15 and DDX58. Genes induced by EVs and BF DNA in WT macrophages overlapped greatly as demonstrated by the 13 shared genes, including early ISGs (Fig. 3a and Supplementary Table 2c).

Fig. 3 |. Early ISG upregulation in WT macrophages stimulated by C. albicans EVs and BF DNA.

a, Venn diagram of genes identified by NanoString analysis as induced by C. albicans EVs and BF DNA in WT macrophages as compared with their controls (PBS and Lipofectamine alone, respectively). The 13 overlapping genes are highlighted in the purple box and the ISGs are bolded. b, A heatmap is shown of normalized counts with z-scores and unsupervised clustering of these 13 overlapping genes in WT, cGas^−/− and Sting^−/− macrophages stimulated with PBS EVs, BF DNA or Lipofectamine (Lipo) alone. c–e, RT–qPCR analysis of RSAD2 (c), OASL (d) and IFIT3 (e) in WT, cGas^−/− and Sting^−/− macrophages after stimulation with PBS, EVs, Lipofectamine or BF DNA is shown, n = 3. All data are representative of at least two independent experiments. For RT–qPCR panels, error bars indicate the standard deviation, and statistical analyses were performed by means of unpaired two-tailed Student’s t-tests as compared with the appropriate control, with **P < 0.01; ***P = 0.0002; and ****P < 0.0001.

To determine if these 13 shared inflammatory genes are induced in a cGAS- and STING-dependent manner, we examined their expression levels in response to EVs and BF DNA in cGas^−/− and Sting^−/− macrophages. RSAD2, IFIT3 and ISG15 are highly induced by both stimulants in WT macrophages, but not in macrophages lacking cGAS or STING (Fig. 3b and Supplementary Table 2d). CXCL10 and DDX58 are induced by EVs, but not in a cGAS- and STING-dependent manner (Fig. 3b). All genes induced by BF DNA were upregulated in WT macrophages, but not in cGas^−/− and Sting^−/− (Fig. 3b). To validate our findings, we performed quantitative reverse transcription polymerase chain reaction (RT–qPCR) of RSAD2, OASL and IFIT3 induced by both EVs and BF DNA, which demonstrated potent upregulation of transcripts in a cGAS- and STING-dependent manner (Fig. 3c–e). Genes upregulated by EVs, regardless of cGAS and STING activity, included genes involved in the inflammasome pathway (for example, NLRP3 and CASP4) (Supplementary Table 2a,e,g), which is known to respond to cytosolic and double-stranded DNA^42–44. Furthermore, IL1B, CXCL2, Il1F6, LCP2, IRF9, CSF2RB and ADAR were induced by EVs in the cGas^−/− and Sting^−/− macrophages, but not in WT, suggesting activation of compensatory DNA sensing pathways not normally triggered by EVs (Supplementary Table 2).

cGAS and STING polymorphisms in human monocytes

To determine if cGAS and STING are relevant in human immunity to C. albicans infection, we investigated single nucleotide polymorphisms (SNPs) of both cGAS and STING in human monocytes exposed to the fungal pathogen and assessed whether these SNPs led to changes in cytokine levels. We probed a previously published large cohort of patients in which PBMCs were isolated from healthy volunteers and stimulated with C. albicans. In the published study, the authors found that genetic variation in type I IFN pathway genes (IRF1 and STATa) correlated with C. albicans-induced release of the following seven cytokines: tumour necrosis factor (TNF)α, interleukin (IL)1β, IL8, IL6, IL10, IFNγ and IL17³. We queried this existing database for variations in the genetic regions encoding cGAS and STING and correlations with production on the two cytokines known to be produced by STING-dependent activation of the nuclear factor κB transcription factor: IL6 and TNFα^8,45. Based on quantitative trait loci mapping, we identified that SNPs in the regions on chromosomes 5 and 6 (containing cGAS and STING, respectively) correlate with changes in production of TNF and IL6 (Fig. 4 and Extended Data Fig. 5). Specifically, we identified a significant SNP in genetic region MB21D1 (cGAS) with an associated genotype (GG) with decreased levels of cGAS (as evident by the negative z-score in Extended Data Fig. 5e), which correlates with decreased C. albicans-induced TNFα and IL6 (Fig. 4a,b). We also identified a significant SNP in genetic region TMEM173 (STING). This SNP has an associated genotype (CC) leading to increased levels of STING (as evident by the positive z-score in Extended Data Fig. 5e), which correlates with decreased TNFα and increased IL6 (Fig. 4c,d). We assessed whether these cGAS and STING-dependent changes in TNFα levels exist in our murine macrophage model when exposed to C. albicans. Indeed, TNF production is highly induced in murine macrophages lacking either cGAS or STING as compared with WT macrophages (Extended Data Fig. 5f). These data provide direct evidence that changes in expression in cGAS and STING alter cytokine levels secreted by both human monocytes and murine macrophages challenged by C. albicans.

Fig. 4 |. Functional consequence of activating the STING pathway.

a,b, Boxplots show the top SNP rs610913 at the MB21D1 locus (cGAS) stratified by genotypes for C. albicans-induced TNFα concentrations (a) and for C. albicans-induced IL6 concentrations (b). c,d, Individuals homozygous for the G allele (GG) on average produce less TNFα and IL6 than individuals carrying one G allele (GT) or homozygous for the T allele (TT). Boxplots show the top SNP rs28396805 at the TMEM173 locus (STING) stratified by genotypes for C. albicans-induced IL6 concentrations (c) and C. albicans-induced TNFα concentrations (d). On average, carriers of the C allele (CC and TC) produce less IL6 and TNFα than individuals homozygous for the T allele (TT). The minimum portion of the error bars denote the first quartile (25th percentile) and the maximum part of the error bars represent the third quartile (75th percentile). The black line in the middle denotes the median (50th) percentile. The black dots are whiskers that span between the minimum and maximum values. A total of n = 442 samples have genotype and cytokine data from the 500FG cohort⁸³.

Discussion

We demonstrate a novel connection between the innate immune STING pathway and EVs produced by the fungal pathogen, C. albicans. Specifically, we have identified that C. albicans activates the STING pathway in macrophages via EVs delivering fungal DNA to the cytosol of macrophages and triggering engagement of cGAS and initiation of downstream signalling. This signalling triggers the expression of many early ISGs, IFNβ and other innate immune processes historically associated with viral and bacterial infections^{8,11,14,16–18}. Furthermore, we identified that activation of this pathway clears the fungal infection, but at the detriment to the host. Importantly, STING activation is relevant in both mouse and human models of candidaemia as demonstrated through in vivo studies and SNP analysis, respectively. While type I IFNs had been implicated previously in candidemia in humans, little was known about the mechanism of this activation or the immune pathways involved^3,4,6,46. We have now identified a major innate immune pathway involved in this type I IFN response and further characterized how C. albicans interacts with its components.

EVs are integral in fungal biology as they assist in pathogen intracellular communication to develop BFs and coordinate community dynamics^25,47–49. C. albicans and other EV-producing microbial organisms use EVs to package important signalling and effector molecules to communicate with the surrounding community. These signals, packaged as cargo in these lipid bilayer membrane vesicles, can respond to changes in microenvironments by providing drug resistance, altering cell adhesion and disseminations properties, and enhancing BF matrix production^20,47. Mammalian cells also make EVs and are able to recognize and internalize them for their own community organization^50–55. This coordination can be useful for general growth and development purposes of healthy cells, but can also be usurped by tumour cells to prioritize proliferation of surrounding cancerous cells and metastasis^56–58. Macrophages endocytose and respond to microbial EVs^20,59. Our data support this by revealing a novel form of immune cell surveillance, in which macrophages internalize fungal EVs and co-opt them for the purposes of rewiring innate immunity to generate a robust type I IFN response, which can prime phagocytic cells to kill engulfed pathogens more efficiently. Specifically, we identified the STING pathway as integral to this immune response to DNA packaged inside C. albicans EVs, which might be linked to the known role of the STING pathway in macrophage metabolic re-programming following infection⁶⁰. For example, STING pathway activation via Brucella abortis infection increases intracellular succinate levels and shifts glycolytic profiles of inflammatory macrophages⁶⁰. Fungal EVs could similarly trigger this glycolytic switch to quickly clear invading fungal pathogen. It is also known that the ISG response promotes the phagocytic activity and microbial killing capacity of macrophages via increased acidity of the phagolysosome⁶¹. Therefore, EVs might be signalling to the macrophage of an impending viable microbial organism and triggering a drop in lysosomal pH. We demonstrated that following both EV and BF DNA stimulation, the type I IFN response is upregulated and that this has functional consequences for both the host cell and the live microbe. Future studies will investigate the extent to which this re-programming affects immune cell function.

We demonstrated that C. albicans and C. auris EVs rewire cells towards a type I IFN response. Historically, fungal recognition has been focused on C-type lectin receptors, TLRs, RIG-I-like receptors and the inflammasome sensing various cell wall epitopes^62–67. These receptors, while highly influential of disease outcome, were not explicitly linked to the type I IFN response. Studies have since observed the induction of a type I IFN signature profile in response to C. albicans in human PBMCs³ and dendritic cells⁵ and identified type I IFN-induced ISGs as essential for the host response to candidemia^7,46,68. Recently, a study demonstrated that Candida spp. elicit a mitochondrial-associated type I IFN response in vaginal epithelial cells⁴. Furthermore, clinical studies have found increased IFNα and IFNβ levels in the vaginal fluid of patients suffering from vulvovaginal candidiasis⁶⁹. One other study assessing the ability for nucleic acids from yeast to induce a type I IFN response identified that yeast DNA and RNA activate IFNβ responses in dendritic cells via signalling through TLRs, but did not observe this response in macrophages nor did they investigate the potential role of the STING pathway⁷⁰.

Our studies enhance our understanding of the Candida-induced type I IFN response by identifying the cGAS–STING pathway as integral for this innate immune response. We further identified that the source of the C. albicans nucleic acids sensed by cGAS is not from phagocytosed yeast, but from C. albicans EVs delivering double-stranded fungal DNA into the cytosol following endocytosis. While the exact mechanism is unknown, Aspergillus fumigatus also activates the STING pathway in a keratitis model of infection⁷¹. Perhaps A. fumigatus EVs function similarly to deliver DNA cytosolically^72,73. A. fumigatus has also been shown to activate a coordinated type I and III IFN response in the lung, warranting future studies to elucidate the ability of EVs from other pathogenic fungi to activate this signalling pathway^74,75. Additionally, investigating the ability for neurotropic pathogenic fungi such as Cryptococcus neoformans and Coccidioides posadasii to make EVs that can activate the STING pathway is merited, as there is evidence that microglia can activate the STING pathway⁷⁶. Expanding our studies past Candida species would be an exciting next step in elucidating the intimate relationship between fungal pathogens and the type I IFN response.

We have demonstrated a novel connection between C. albicans and the STING pathway and revealed fungal DNA packaged in released EVs as the mechanism of activation. Furthermore, we identified the functional consequences of this signalling in both mouse and human models, highlighting the physiologic relevancy of these findings. Overall, the results presented here have significantly enhanced our understanding of the type I IFN response initiated by a human fungal pathogen.

Methods

In vivo studies

In vivo infection studies were carried out as previously described⁷⁷. Briefly, C. albicans (WT strain SC5314) was grown in yeast extract peptone dextrose (YPD) medium for 16 h at 30 °C in a shaker incubator at 250 rpm. C. albicans was washed and resuspended at a concentration (500,000 yeast ml⁻¹) in ice-cold PBS. WT C57BL/6 ( JAX #000664), cGAS^−/− C57BL/6 (a gift from the Stetson Laboratory⁷⁸) and Sting^−/− B6(Cg) ( JAX no. 025805) mice were injected intravenously into the lateral tail vein with C. albicans (100,000 or 150,000 yeast). We chose to use this knockout mice strain instead of goldenticket, which carries a missense mutant allele of the Sting gene ( JAX no. 017537), which theoretically may carry residual function. Following the C. albicans challenge, mice were monitored twice daily for morbidity and mortality for up to 28 days. Mice displaying pre-specified criteria for distress (inability to feed or drink, laboured breathing, ruffled and/or matted fur, decreased activity, hunched posture and shivering) were euthanized by CO₂ asphyxiation. All mouse experiments were approved by the MGH Institutional Animal Care and Use Committee under protocol no. 2008N00078.

Organ burden quantification, kidney BUN and creatinine

Organs (kidneys, livers, spleens and brains) were collected from infected WT and Sting^−/− mice at different points during the infection. Organ weights were obtained, and each organ was homogenized using a sterilized and PBS-washed Dounce homogenizer for 30 s at a time until the organs formed a homogenate. Organ suspensions were then diluted in 1 ml of PBS and serially diluted. Then 100 μl of each dilution was plated onto YPD agar and allowed to grow at 30 °C for 48 h. Colony forming units were calculated by the following formula: (number of colonies × dilution factor × 10)/organ (g) (n = 8 for WT and 14 for Sting^−/− for each timepoint in Fig. 1c and n = 4 for WT and 4 for Sting^−/− for each time point in Extended Data Fig. 5a).

Kidney BUN and creatinine levels were measured from serum of infected and uninfected WT and Sting^−/− mice. Whole blood was collected from mice posthumously and left to clot for 45 minutes at room temperature. Blood was then spun at 12,000g for 10 min at 4 °C. Serum was then separated, frozen and given to the Clinical Pathology Laboratory in the Center for Comparative Medicine at Massachusetts General Hospital for analysis.

Immunophenotyping by flow cytometry and histopathology

WT and Sting^−/− mice (n = 6) were infected with 150,000 C. albicans yeast intravenously. Uninfected control mice were also processed for analysis. Mice were euthanized at 5 days post-infection, and immunophenotyping was performed as previously described^26,27. Briefly, kidneys were excised, placed into digestion solution (collagenase type IV and 100 U DNase I in 100 ml Roswell Park Memorial Institute (RPMI) 1640 medium), finely minced using scissors and incubated for 1 h with shaking at 37 °C. After 1 h, 800 ml of 0.5 M ethylenediaminetetraacetic acid was added to each sample to stop the collagenase activity. Kidney samples were passed through a 70 mM filter, pelleted, resuspended in a 40% Percoll-RPMI solution and under-layered with a 67% Percoll–RPMI solution. Samples were centrifuged at 650g for 20 min and leukocytes were collected at the interface between layers and washed twice with flow cytometry (FACS) buffer (PBS with 2% foetal bovine serum (FBS)). Cells were then treated with ammonium–chloride–potassium buffer to lyse the red blood cells for 5 min at room temperature (RT) and spun at 1200 rpm. All cells were stained for flow cytometry analysis. Fc block (anti-mouse CD16/CD32, Thermo Fisher Scientific, 14–0161–85) was added to cells at 1:100 dilution and incubated at RT for 15 min. Cells were then labelled with a mixture of the following fluorophore-conjugated antibodies for 1 h at 4 °C (all antibodies are from Biolegend, unless otherwise noted): anti-CD45 BV605 (Biolegend, 103139), anti-CD90.2 BV786 (Biolegend, 105331), anti-CD19 PE-Dazzle (Biolegend, 115553) and anti-Ly6G AF 488 (Biolegend, 127625). Cells were washed with FACS buffer, and 7-aminoactinomycin D (Stem Cell Technologies, 75001.1) was added for live-dead cell staining. Flow cytometry was performed using BD FACSCelesta and analysis was performed using FlowJo 10 software (BD). Quantification was performed in GraphPad Prism9 and FlowJo 10.9.0.

For histopathology, kidney tissue was collected and fixed in 10% neutral buffered formalin, trimmed, processed routinely, embedded in paraffin, sectioned (5 μm) and stained with H&E and GMS. Histology was performed by Tufts Comparative Medicine Services at Tufts University. Histopathology was evaluated by a board-certified veterinary pathologist. Images were obtained using 10× magnification.

Cell culture and macrophage cell line generation

All immortalized bone marrow-derived macrophages were cultured in complete Dulbecco’s modified eagle medium (cDMEM) (supplemented with 10% FBS, penicillin and streptomycin (Pen+Strep), L-glutamine and 1 M hydroxyethylpiperazine ethane sulfonic acid at 37 °C in 5% CO₂. For passage, macrophages were washed with PBS and lifted with 0.25% trypsin diluted at 1:10. The following cell lines were used: WT immortalized bone marrow-derived macrophages gifted by Douglas Golenbock (University of Massachusetts Medical School), cGas^−/− and Sting^−/− macrophages were generated as previously described⁷⁹, Dectin-1^−/− immortalized macrophages were a gift from Gordon Brown (University of Aberdeen) and Stuart Levitz (University of Massachusetts Medical School), Tlr9^−/− macrophages were generated as previously described⁸⁰, Card9^−/− macrophages were a gift from Ramnik Xavier (Harvard Medical School), and cGAS–GFP expressing macrophages were generated as previously described³⁵. All macrophage lines were immortalized from C57BL/6 mice.

PrestoBlue killing assay

The PrestoBlue assay was used to measure fungal viability of C. albicans as previously described⁷². WT, cGas^−/− and Sting^−/− macrophages were plated in triplicate in a 96-well plate at a density of 1 × 10⁴ cells per well. Following an overnight incubation, the cells were stimulated with either PBS or live C. albicans for 2 h at 37 °C and 5% CO₂. C. albicans only controls were also plated to determine the uninhibited rate of growth for end-point comparison. The ratio of fungal cell to macrophage was changed to allow for further interpretation of killing dynamics (1:1, 1:5 and 5:1). Plate was spun down at 1,200 rpm for 1 min before incubation to ensure contact of cells with fungi. Cells were then lysed with NP-40 to liberate phagocytosed yeast cells and placed on ice for 5 min. Fresh 3-(N-morpholino)propanesulfonic acid–RPMI pH 7 4 and PrestoBlue (Thermo Fisher Scientific, no. A13261) was added to each well and mixed thoroughly. The plate was then incubated at 37 ˚C for 10 min and then moved to 30 °C to incubate for 2 h in the dark. Fluorescence was measured using a SpectraMax i3× Multi-Mode Microplate Reader system (Molecular Devices) analysing the 560/590 nm wavelength one time. Percentage of live C. albicans compared with no macrophage control wells was determined using excel and statistical significance was assessed using GraphPad Prism9 software (GraphPad Software). P < 0.05 was considered significant.

EV isolation and purification from YPD agar lawns

Fungal EVs were isolated from C. albicans (SC5314) and C. auris (Biosample accession number SAMN05379624) as previously described⁸¹. Briefly, one colony of each fungus was inoculated in Sabouraud broth and incubated overnight at 30 °C under constant shaking. Yeast cells were transferred to YPD agar plates (~10⁷per plate) and spread using glass beads. The plates were incubated at 30 °C for 48 h. The cell lawns were carefully scraped from the plates and transferred to conical tubes containing PBS. Cell suspensions were centrifuged at 1,500g and the resulting supernatant was centrifuged again at 5,000g. The supernatant was filtered through a 0.8 μm membrane and then ultracentrifuged at 150,000g. After discarding the supernatant, EVs were suspended in PBS, and the ergosterol concentration was quantified using the Amplex Red cholesterol assay kit (Thermo Fisher). EV suspensions were stored at 4 °C. EVs were further quantified using the NanoSight core facility at Massachusetts General Hospital.

Isolation and characterization of C. albicans EVs from BF and planktonic culture

EVs were isolated from both planktonic and BF cultures from C. albicans (WT strain SN250). A volume of 1-litre batches of planktonic cultures were grown in two Erlenmeyer flasks at 30 °C at 200 rpm for 24 h and 48 h²⁴, whereas BF cultures were grown in the large-scale system employing polystyrene roller bottles at 37 °C at 4 rpm for 24 h and 48 h⁸². Planktonic culture supernatants were separated from cell biomass by centrifugation at 10,000g for 10 min at 4 °C and filter sterilized. BF culture media were carefully decanted, remaining suspended debris was removed by filtration through a paper filter and then filter-sterilized. Collected culture supernatants concentrated down to 25 ml using a Vivaflow 200 unit (Sartorius AG) equipped with a Hydrosart 30 kDa cutoff membrane. The samples were then centrifuged at 10,000g for 1 h at 4 °C to remove larger cellular/membrane-like particulates. The pellets were discarded, and the collected supernatants were next centrifuged in a Beckman Coulter Optima MAX-XP Ultracentrifuge equipped with an MLA-50 fixed rotor at 100,000g for 1.5 h at 4 °C. The resulting supernatants were subsequently discarded, whereas the pellets were resuspended in 5 ml of PBS (pH 7.2) and re-centrifuged in the ultracentrifuge utilizing an MLS-50 swinging bucket at 100,000g for 1 h at 4 °C. The collected EVs were then polished by flash size-exclusion chromatography on qEV/35 nm columns (Izon Science), filter sterilized and stored until further use at 4 °C⁴⁷.

Purified EVs were quantified using nanoparticle tracking analysis. EV samples were diluted in PBS to a final volume of 1 ml and pre-tested to obtain an ideal count of 30–100 particles per frame rate using a NanoSight NS300 system equipped with a sample assistant autosampler (Malvern). The following settings were applied: camera level was increased to 16 and camera gain to 2 until tested images were optimized and nanoparticles were distinctly visible without exceeding particle signal saturation. Each measurement consisted of five 1-min videos with a delay of 5 s between sample introduction and the start of the first measurement. For detection threshold analysis the counts were limited to 10–100 red crosses and no more than 5–7 blue crosses. Acquired data were analysed using the NanoSight Software NTA 3.4 Build 3.4.003. At least 1,000 events in total were tracked per sample to minimize data skewing based on single large particles²².

Analysis of DNA in C. albicans EVs

DNA was extracted from the purified BF EVs using the MasterPure Yeast DNA Purification Kit (Lucigen). Briefly, EVs were resuspended in 300 μl of yeast cell lysis solution and incubated at 65 °C for 15 min. The samples were placed on ice for 5 min, and then 150 μl of milk protein concentrate protein precipitation reagent were added, vortex mixed for 10 s and centrifuged in a table-top microcentrifuge for 10 min. The supernatants were transferred to new Eppendorf tubes followed by addition of 500 μl of isopropanol and mixed thoroughly by inversion. DNA was collected by centrifugation in a table-top microcentrifuge for 10 min at the maximum speed. The pellets containing the environmental DNA were washed with 500 μl of 70% ethanol, dried briefly at 42 °C and placed in 50 μl of TE buffer. DNA concentration was determined using the Qubit dsDNA HS Assay Kit (Life Technologies). Libraries were then prepared according to the Celero PCR Workflow with Enzymatic Fragmentation (Tecan Genomics). Both quality and quantity of the finished libraries were assessed using an Agilent Tapestation (Agilent) and the Qubit dsDNA HS Assay Kit, respectively. Paired-end 150 bp sequencing was performed using the Illumina NovaSeq6000 (Illumina).

C. albicans DNA extraction from BF

WT C. albicans (SC5314) culture was grown overnight from a frozen culture and rotated in fresh YPD media at 30 °C. From the overnight culture, an inoculum of C. albicans was washed 3× with 1× PBS and diluted in cDMEM at 1 × 10⁶ cells ml⁻¹. Inoculum was then plated to a 12-well tissue culture (TC)-treated plate at 1 ml per well in all 12 wells and spread to cover the bottom of each well. Plate was incubated at 37 °C with 5% CO₂ for 24 h, 48 h, 72 h or 96 h to grow BF before collection. BF DNA was extracted using the Norgen Biotek Biofilm DNA Isolation Kit (no. 62300) Briefly, using a cell lifter and sterile tweezers, 6 wells of the 12-well plate of BF were scraped and transferred to a microcentrifuge tube (200 mg wet weight). BFs were centrifuged at 20,000g for 1 min to collect the BF. Using tweezers, BF was transferred to a bead tube and protocol was followed per the kit instructions and the optional RNAse treatment was performed. Pure BF DNA was collected, measured by nanodrop and stored at −80 °C.

Immunoblot analysis

Macrophages were plated at a concentration of 1.5 × 10⁵ cells ml⁻¹ in a 12-well TC-treated plate and allowed to adhere overnight. For kidney homogenate experiments, kidneys were collected from mice at day 3 and 5 post-infection (left and right kidney were combined). Kidneys were homogenized using a sterilized and PBS-washed Dounce homogenizer for 30 s at a time until the kidneys formed a homogenate. Kidney suspensions were then diluted 1:20 and treated as ‘neat’ lysates for subsequent immunoblot analysis. For Dynasore treatment experiments, macrophages were plated at a concentration of 1 × 10⁵ cells ml⁻¹ and allowed to adhere overnight. The following day, cells were washed with PBS and serum-free media were added to the cells with or without 100 μM Dynasore Hydrate (Sigma-Aldrich D7693). Cells were treated at 37 °C and 5% CO₂ for 30 min before subsequent stimulation. For H-151 treatment experiments, macrophages were plated at a concentration of 1 × 10⁵ cells ml⁻¹ and allowed to adhere overnight. The following day, macrophages were washed and 5 μM/0.5 μM H-151 (Selleck Chemicals S6652) was added to treated wells for 24 h before stimulation. Macrophages were either stimulated with PBS, EVs, LPS or transfected with cGAMP, C. albicans DNA from BF or EVs. Lipofectamine reagents alone containing DNA elution buffer were also used in some experiments as a negative control. Transfections were completed using the Lipofectamine 3000 reagents and protocol (Invitrogen L3000015). Lysates were collected following a 6 h incubation at 37 °C and 5% CO₂ using mammalian protein extraction reagent lysis buffer (Thermo Scientific, no. 78501) with sodium orthovanadate and protease inhibitors. Lysates were spun down (14,000g for 5 min at 4 °C), transferred to a fresh tube and subsequently mixed with 4× NuPage lithium dodecyl sulfate loading buffer and 10× NuPage reducing agent. Western blot assays were performed using a 4–12% NuPage gel, with 2-[N-morpholino]ethanesulfonic acid running buffer (NuPage gels, Thermo Fisher Scientific), and transferred to methanol activated polyvinylidene difluoride membrane (Perkin Elmer, Waltham, MA) using transfer buffer (0.025 M Tris, 0.192 M glycine and 20% methanol) and electrophoretic transfer at 100 V for 1 h. For detection of proteins, polyvinylidene difluoride membranes were blocked for 1 h at RT in 5% milk in PBS 0.01% Tween 20 (PBST). To detect the viperin (Sigma MABF106), TBK1 (Cell Signaling 3504) and IRF3 (Cell Signaling 4302) proteins, blots were incubated for 1 h at RT in PBST, 1% BSA and primary antibody (1:5,000, 1:1,000 and 1:1,000, respectively). To detect phosphorylated proteins, phospho-TBK1 (Cell Signaling D52C2) and phospho-IRF3 (Cell Signaling E6F7Q), blots were incubated overnight at 4 °C in tris-buffered saline with Tween 20 (TBST), 5% BSA and primary antibody (1:1,000). For phosphorylation specific probes, TBST (Cell Signaling 9997S) was used for all incubations and washes. Following incubation with primary antibody, blots were subsequently washed in either PBST or TBST 3× and incubated with secondary swine anti-rabbit horseradish peroxidase conjugated antibody at 1:2,000 (Agilent DAKO, P0399) or secondary peroxidase AffiniPure goat anti-mouse IgG (H and L) ( Jackson ImmunoResearch) at 1:2,000 in 1% milk in PBST/TBST for 1 h at RT. To detect total protein via actin, blocked blots were incubated for 1 h at RT with simultaneous probing for actin (Cell Signaling 1:2,500) and secondary swine anti-rabbit horseradish peroxidase conjugated antibody. Membranes were washed 3× and then visualized using Western Lightning Plus ECL chemiluminescent substrate (Perkin Elmer) on Kodak BioMax XAR film (MilliporeSigma). Films were then scanned and processed using Adobe Illustrator. Any contrast adjustments were applied evenly to the entire image and adheres to standards set forth by the scientific community. All reported western blots/immunoblots were repeated in at least biological duplicate.

Benzonase treatment

For Benzonase treatments, we added eight units of Benzonase Nuclease ultrapure (E8263 Sigma) to 0.25 μg BF DNA and four units to 1 μM EVs based on vendor recommendation. Mixture was incubated for 30 min at 37 °C and 5% CO₂ before transfection or stimulation.

ELISA analysis

Macrophages were plated at a concentration of 1.5 × 10⁵ cells ml⁻¹ in a 12-well TC-treated plate for IFNβ enzyme-linked immunosorbet assays (ELISAs) and a concentration of 1 × 10⁴ cells ml⁻¹ in a 96-well plate for the TNF ELISA. For the IFNβ ELISA, macrophages were either stimulated with PBS, EVs or transfected with cGAMP, or C. albicans DNA from BF. Transfections were completed using the Lipofectamine 3000 reagents and protocol (Invitrogen L3000015). Supernatants were collected following a 6 h (IFNβ) and 2 h (TNFα) stimulation at 37 °C and 5% CO₂. IFNβ levels were assessed on these supernatants using ELISA (Duoset, R and D Systems) per the manufacturer’s instructions and read using an i3X Spectrophotometer (Molecular Devices, LLC). Results were analysed using PRISM9 software (GraphPad Software). All reported ELISAs were done in both technical and biological triplicate.

Microscopy

cGAS–GFP-expressing macrophages were plated on Nunc Lab-Tek eight-well chambered coverglass (Thermo Fisher Scientific). Macrophages were plated at a concentration of 1 × 10⁵ cells ml⁻¹ in 500 μl and incubated overnight (~18 h) at 37 °C and 5% CO₂. For stimulation, 50 μl of EVs (~10¹¹) were either dyed with 5 μM DiI lipophilic stain (Invitrogen D282) or untreated for 30 min at RT. Following staining, EVs were washed twice with PBS, using an ultracentrifuge to pellet the EVs at 100,000g for 1 h. A PBS plus DiI sample was also subjected to the same wash steps. Before stimulation, macrophages were washed three times with PBS to remove FBS and any mammalian EVs present from serum. Macrophages were then refreshed with DMEM without FBS and PBS; EVs, DiI-stained EVs or DiI-stained PBS were added to wells. Stimulated cells were then incubated for 3 h at 37 °C and 5% CO₂ to allow phagocytosis of EVs. Images were captured on a Nikon Inverted Microscope Eclipse Ti-E equipped with a CSU-X1 confocal spinning disk head (Yokogawa), and a Coherent 4 W continuous-wave laser (Coherent) excited the sample. A 100× high-numerical aperture objective (Nikon, 1003, 1.49 numerical aperture, oil immersion) was used. Images were obtained using an EMCCD camera (Hamamatsu Photonics). Image acquisition was performed using MetaMorph software (Molecular Devices). Raw images were then cropped by FIJI and assembled in Adobe Illustrator, version CS4 (Adobe Systems). Macrophage cGAS localization was quantified as either nuclear (only nuclear localization), non-nuclear (only cytosolic localization) or mixed (a combination of nuclear and cytosolic).

RNA extraction and NanoString analysis

Macrophages were plated at a concentration of 1.5 × 10⁵ cells ml⁻¹ in a 12-well TC-treated plate and incubated overnight (~18 h) at 37 °C and 5% CO₂. Macrophages were either stimulated with PBS, EVs, or transfected with Lipofectamine alone or C. albicans DNA from BFs grown for 72 h. Samples were collected following a 6 h incubation at 37 °C and 5% CO₂. NanoString (nanoString) profiling of transcriptional responses was performed using the nCounter Host Response Panel according to manufacturer’s instructions. Gene expression of host response genes was measured via NanoString Technology by first isolating RNA using the QIAGEN RNeasy Mini Kit. RNA was then measured via nanodrop and diluted to 25 ng per reaction. RNA was then hybridized in hybridization buffer at 65 °C for 18 h with the Host Response Gene Panel Reporter Codeset and Capture Probeset (nCounter). Hybridized samples were then loaded into a nCounter Cartridge and analysed using a nCounter Sprint Profiler for 6 h. Data were then normalized and analysed by nSolver. Subsequent heatmaps and graphs were made in nSolver, Prism and BioVenn. Each NanoString experiment was run in biological duplicate. Genes were normalized to the controls included in the plate for each genotype to account for variability, and Supplementary Table 2 includes these normalized ratios. RT–qPCR validation was performed on extracted RNA via Invitrogen EXPRESS One-Step Superscript (11781200). Primers for mouse RSAD2 (4453320, Mm00491265_m1), OASL (4448892, Mm00455081_m1) and IFIT3 (4453320, Mm01704846_s1) were ordered from Thermo Fisher.

Macrophage differentiation and stimulation and cytokine quantitative trait loci mapping

Details of the macrophage differentiation and stimulation experiments, as well as cytokine quantitative trait locus association analysis of the 500FG cohort, have been previously described⁸³. Briefly, monocytes were plated at a concentration of >5 × 10⁵ in flat-bottom plates with 10% human serum at 37 °C and 5% CO₂ for 6 days. After differentiation, the medium was removed and the differentiated macrophages were stimulated with heat-killed C. albicans for 24 h (strain ATCC MYA-3573, UC 820) in a concentration of 10⁶ CFU ml⁻¹. Supernatants were collected and stored at −20 °C until ELISA experiments were conducted to measure cytokines concentrations. Concentrations of IL6 and TNFα were profiled in macrophages. To explore the strength of association of genetic polymorphisms in cGAS and TMEM173 genes, we extracted genetic variants using a window size of 250 kb upstream and downstream of each gene, and results were visualized with the LocusZoom tool⁸⁴. The 500FG cohort consists of approximately 500 healthy individuals of European origin with age range of 18–75 years. We queried the most significant or top SNPs for statistical evidence of association with the target genes using the eQTLGen database (https://www.eqtlgen.org/).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Extended Data

Extended Data Fig. 1 |. Characterization of organs and serum from C. albicans-infected WT and Sting^−/− mice.

a. Quantification of fungal burden following a C. albicans infection in WT and Sting^−/− mice in the kidneys, liver, spleen, and brains (n = 4 for WT and Sting^−/− for each time point). Kidney BUN (b) and creatinine levels (c) were measured from serum of infected and uninfected WT and Sting^−/− mice over 28 days. n = 5 for each time point. d. Representative gating strategy for flow cytometry analysis showing selection of total kidney cells (SSC v FSC) → live white blood cells (WBC) (7AAD- and CD45 + ) → WBCs that are not B cells or T cells (CD90.2-, CD19-), and of these, how many are neutrophils (Siglec F + , Ly6G + ). e. Quantification of flow-based immunophenotyping of CD45+ WBCs and (f) Siglec F + , Ly6G+ neutrophils from single cell suspensions of kidneys from WT and Sting^−/− mice (n = 6 for each group). g. Immunoblot of viperin from kidney homogenates harvested from WT and Sting^−/− mice infected with 150,000 C. albicans at day 3 and day 5, compared to the positive control (immortalized WT macrophages stimulated with cGAMP for 6 h prior to lysis). h. Semi-quantification of fungal burden severity in GMS-stained kidney sections of WT and Sting^−/− mice (n = 12). Statistical analyses were performed on raw data by means of a non-parametric, two-tailed Mann Whitney U test = 38.5, p = 0.0364. Data are presented as mean values +/− SD.

Extended Data Fig. 2 |. Further characterization of the viperin and IFNß response to C. albicans EVs.

a. Immunoblot of viperin in lysates from WT macrophages stimulated with PBS, ergosterol alone, EVs from C. albicans or C. auris, or cGAMP. b. IFNß production in the supernatants from the same stimulated WT macrophages as in (a). Significance determined using a one-way ANOVA, ***adj. p = 0.0004, **adj. p = 0.004 c. Average sizes of EVs extracted from C. albicans (grown in different conditions) and C. auris. d. Immunoblot of viperin induction in WT, Card9^−/−, Dectin-1^−/−, and Tlr9^−/− immortalized macrophages following stimulation with PBS or C. albicans EVs. e. IFNß production by WT, cGAS^−/−, Sting^−/−, Card9^−/−, Dectin-1^−/−, and Tlr9^−/− macrophages following a stimulation with PBS, cGAMP, or C. albicans EVs. Significance determined using a two-way ANOVA and subsequent Dunnett’s multiple comparison test with the exception of Card9^−/−, which was an ordinary one-way ANOVA and subsequent Dunnett’s multiple comparison test, *adj. p = 0.0031, **adj. p < 0.01, ***adj. p < 0.001, ****adj. p < 0.0001 vs PBS f. Immunoblot of viperin and g. IFNß production by WT macrophages stimulated with graded amounts of EVs extracted from C. albicans 48 h BF. Significance determined using a one-way ANOVA and subsequent Dunnett’s multiple comparison test *adj. p = 0.0364, ***adj. p = 0.0008, ****adj. p < 0.0001 vs PBS. n = 3 biologically independent samples for all ELISAs and data are presented as mean values +/− SD. All Westerns were done in biological triplicate, representative blot shown. Macrophages were stimulated for 6 h prior to processing.

Extended Data Fig. 3 |. C. albicans EVs induce translocation of cGAS from the nuclear membrane to the cytosol.

Representative microscopy images of macrophages expressing cGAS-GFP and the localization of cGAS following no stimulation or stimulation with DiI only, C. albicans EVs only, or DiI-labeled C. albicans EVs. Macrophages were stimulated for 3 h prior to imaging. Size bar = 5 μm. Microscopy experiments were repeated with appropriate controls in biological triplicate, representative images shown of similar results.

Extended Data Fig. 4 |. Further characterization of the viperin and IFNß response to C. albicans BF DNA.

a. Immunoblot of viperin induction in WT macrophages stimulated with PBS, C. albicans EVs, Benzonase-treated EVs, and transfected EV DNA. b. Immunoblot of viperin induction and c. IFNβ production in WT macrophages stimulated with DNA extracted from C. albicans biofilms (BF) grown for 48 h, 72 h, 96 h and Lipofectamine reagents only. Significance was determined with an ordinary one-way ANOVA and subsequent Dunnett’s multiple comparison test. **adj. p = 0.0012, ***adj. p = 0.0002, ***adj. p < 0.0001 vs untreated control (Lipofectamine). d. Immunoblot of viperin induction in WT, Sting^−/−, Card9^−/−, Dectin-1^−/−, and Tlr9^−/− macrophages following a stimulation with PBS, cGAMP, and DNA extracted from C. albicans BF grown for 72 h. e. IFNß production by WT, cGas^−/−, Sting^−/−, Card9^−/−, Dectin-1^−/−, and Tlr9^−/− macrophages following a stimulation with PBS and DNA extracted from C. albicans BF grown for 72 h. Significance determined using a two-way ANOVA and subsequent Dunnett’s multiple comparison test. ****p < 0.0001 vs untreated control (PBS). f. Immunoblot of STING pathway component in WT, cGas^−/−, and Sting^−/− immortalized macrophages treated with increasing concentrations of the STING inhibitor, H-151. Cells were treated with H-151 1 h prior to stimulation with PBS, cGAMP, C. albicans BF DNA, or LPS (STING independent positive control). g. Immunoblot of viperin and actin in WT, cGas^−/−, and Sting^−/− macrophages treated with the H-151 1 h prior to PBS, cGAMP, C. albicans BF DNA, or C. albicans EV stimulation. All macrophages were stimulated for 6 h prior to processing. n = 3 biologically independent samples for all ELISAs and data are presented as mean values +/− SD. All Westerns were done in biological triplicate, representative blot shown.

Extended Data Fig. 5 |. Regional association plots for SNPs identified in cGAS and STING.

Regional association plot of SNPs around the MB21D1gene locus for Candida-induced TNF–α concentrations (a) and for Candida-induced IL-6 concentrations (b). Regional association plot of SNPs around the TMEM173 gene locus for Candida-induced TNF–α concentrations (c) and for Candida-induced IL-6 concentrations (d). The -log10 p-values of imputed SNPs are plotted on the y-axis against their physical position (NCBI build 36) on the x-axis. The most strongly associated SNPs in the regions are represented with purple diamond and surrounding markers are color coded according to their correlation coefficient (r²) with the top SNP using the hg19/1000 Genomes European samples. The light blue lines denote the estimated recombination rates. e. Evidence of the top SNPs regulation the expression levels of MB21D1 and TMEM173. We queried the most significant or top SNPs for statistical evidence of association with the target genes using the eQTLGen database and considered a p value of < 5 × 10⁻⁸ to be the threshold for significant cytokine QTLs. f. TNF production by WT, cGas^−/−, and Sting^−/− murine macrophages following a 2 h co-culture with live C. albicans. ****adj. p < 0.0001 vs. WT. n = 3 biologically independent samples for ELISA and data are presented as mean values +/− SD. Significance determined using an ordinary one-way ANOVA and subsequent Dunnett’s multiple comparison test.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Source data are provided with this paper.