MUC16-dependent Renal Vascular Adhesion of Candida Promotes Tissue Invasion and Predicts Clinical Outcome in Candidemia
Peter Williamson, Jin Qiu, Shellee Grim, Guowu Hu, Yoon-Dong Park, Melissa Johnson, Sunita Paaudel, Troy Stevens, John Perfect, Visiliki Matzaraki, Vinod Kumar, Frank van de Veerdonk, Mihai Netea, Michail Lionakis, Adebo Adebowale, Scott Filler, Nina Clark, Ernesto Cota
TL;DR
This study shows that the protein MUC16 helps Candida stick to blood vessels in the kidneys, leading to infection and worse outcomes in patients.
Contribution
The study identifies MUC16 as a key host factor mediating Candida vascular adhesion and tissue invasion through a ternary complex with Als3p and N-cadherin.
Findings
MUC16 deletion in mice reduced fungal vascular binding and neutrophil recruitment.
MUC16 expression is induced via MINCLE-, TLR9-, and Dectin-1–dependent signaling.
A MUC16 T10155I variant and elevated MUC16 levels correlate with higher 30-day mortality in candidemia.
Abstract
Bloodstream infection with Candida species is associated with high mortality, yet the molecular basis of vascular adhesion and invasion remains unclear. We identify the endothelial surface mucin MUC16 as a key host factor enabling fungal vascular adherence. The MUC16 SEA domain directly engages the Candida adhesin Als3p and endothelial N-cadherin to form a ternary complex that promotes endothelial attachment and vascular invasion. Genetic deletion of Muc16 in mice markedly reduced fungal vascular binding, renal fungal burden, and Ly6Ghi CD11b+ neutrophil recruitment following intravenous infection, and diminished oral colonization in an oral model. Inflammatory induction of MUC16 occurred via MINCLE-, TLR9-, and Dectin-1–dependent signaling, converging on CARD9. In clinical cohorts, a MUC16 T10155I variant and elevated circulating MUC16 were associated with increased 30-day mortality.…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAntifungal resistance and susceptibility · Parasitic Diseases Research and Treatment · Neutrophil, Myeloperoxidase and Oxidative Mechanisms
Introduction
Candida spp. are the fourth most common cause of nosocomial bloodstream infection in the US, with crude mortality of 25–40% and causing an estimated 400,000 life threatening systemic infections per year^1^ with an additive adverse impact during the COVID-19 era^2^. However, despite its significance, key elements of pathogenesis remain poorly understood, especially in disseminated bloodstream infections in which mortality remains high despite therapy. Risk factors for mortality include malignancy, presence of chronic indwelling central catheters and prolonged antibiotic therapy associated with bacterial infections^3^. Recently, genetic risk factors have been identified that include the single nucleotide polymorphisms (SNPs) in PLA2G4B with reactive oxygen species (ROS) and IL-6 production in response to Candida infections^4^ and relationships of CXCR1 and CX3CR1 with the acquisition of infections^5,6^. These studies suggest that better understanding of pathogenic molecular mechanisms may identify patients at highest risk of death from candidemia that would1) allow intensification of therapy in the clinical setting or 2) enable better stratification of patients in clinical treatment trials.
Transient Candida in the bloodstream may occur after colonization of and release from central venous catheters, or translocation through intestinal barriers after pathogenic epithelial cell damage^7^. Once in the bloodstream, dissemination throughout the host requires adherence followed by invasion through the vascular endothelium which occurs through endocytosis and is induced most efficiently by hyphae^8,9^. Successful endothelial endocytosis is dependent on fungal adhesins such as Als3p with a requirement for N-cadherin for fungal endocytosis^10^. Invasion through the endothelial layer into tissue, principally the kidney in mice^11^, results in significant tissue damage and is mediated by a plethora of factors-- proteases, lipases and haemolysins^12^ as well as toxins including the hyphae-specific inflammasome-activating candidalysin^13,14^, important for neutrophil recruitment^15^. Tissue invasion by neutrophils is thought to cause paradoxical collateral tissue damage during fungal killing, constituting a predominant mechanism of deep tissue injury^16^. However, despite the role of N-cadherin in the endocytosis of Candida hyphae that follows initial endothelial binding, the identity of the mammalian-endothelial cell adhesion molecule remains unknown.
In light of surface mucin’s role modulating infections by Staphylococcus^17^, which also expresses a bacterial adhesin with structural similarity to C. albicans Als3^18^, the present study sought a possible role for surface renal mucins in bloodstream infections with Candida albicans as the kidney has been identified as a key organ involved in initial pathogenesis^11^. Of the mucins expressed in endothelial cells (ECs) our studies found that MUC16 was both expressed on the surface of ECs by immunohistochemistry and flow cytometry of mouse kidney primary CD31+ ECs and required for endothelial binding of Candida hyphae. Additional mechanistic studies identified the formation of a Als3-SEA-N-cadherin ternary complex, facilitating sequential binding and endocytosis of the fungal pathogen. The presence of MUC16 in WT mice resulted in increased tissue burden and mortality compared to Muc16^−/−^ littermates in a disseminated candidiasis model, and increased weight loss and tissue burden in an oral candidiasis model. Further clinical studies identified a key T10155I heterozygous polymorphism to be associated with 30-day mortality in two independent cohorts of patients with candidemia and established this polymorphism as having a role in Candida-endothelial cell adherence in model studies. Interestingly, modeling of candidemia risk factors--exposure of ECs to bacterial and fungal products, resulted in increased expression of MUC16, mediated prominently by the receptors MINCLE, TLR9 and DECTIN-1 in a Card9-dependent fashion. Hypothesizing that increased MUC16 expression in patients via such mechanisms could result in poor outcomes, we tested and found in a cohort of 134 patients that increased serum concentrations of MUC16, measured by a commercial CA125 antibody at the time of diagnosis, demonstrated a strong association with 30-day mortality. In summary, these data point to a role for MUC16 in Candida binding to ECs and may provide a readily available biomarker for outcome in bloodstream infections.
Results
MUC16 is required for adherence of Candida albicans to purified primary renal ECs
Because of previously described roles of surface mucins, particularly MUC16 in modulating interactions with pathogens^17^ as well as interactions with surface cadherins^19^, we studied the role of mucins in Candida adherence using primary mouse kidney ECs^20^. As shown in Fig. S1a-c, purified primary mouse kidney ECs did not express CD45 and displayed typical properties of ECs^21–26 27^. Primary purified kidney ECs were then assessed for expression of renal-specific mucins identified in whole-genome expression studies^28^ in the BioGBS database^29^ which identified expression of Muc1, Muc14 and Muc16 that was suppressible by siRNA (Fig. S1e). Surface Muc16 protein was identified by flow cytometry which was not present in ECs purified from Muc16^−/−^ mice (Fig. S1d). Muc13 displayed little qRT-PCR signal (> 45 cycles). Confluent EC monolayers of each suppressed group were then incubated with hyphal elements of Candida albicans expressing fluorescent d-tomato followed by extensive washing^30 10^, which demonstrated reduced fungal adherence after Muc16 suppression but not Muc1, Muc13 or Muc14 vs non-targeted controls (Fig. 1a; Fig. S1f). In addition, primary EC monolayers from Muc16^−/−^ KO mice demonstrated reductions in adherence at 30, 60 or 90 minutes compared to WT litter-mate controls (Fig. 1b). As expected, incubation of WT ECs with the calcium chelator ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA) which inhibits subsequent N-cadherin endocytosis did not inhibit adherence in WT mouse kidney ECs (Fig. 1c).
Assays combining adherence and endocytosis were then conducted using a previously described dual assay utilizing a fluorescent-d-Tomato C. albicans strain followed by addition of a green-fluorescent Alexa 488-labeled non-permeable antibody against intact Candida, the latter of which labels surface adherent but non-endocytosed cells^10^. As shown in Fig. 1d, d-Tomato-labeled C. albicans hyphae exhibited both adherence to WT ECs, identified by green fluorescence (left panel, green arrows), as well as endocytosis demonstrated by red fluorescent hyphae not labeled by green fluorescence (left panel, red arrows). Endocytosis was inhibited by EGTA (+EGTA). In contrast, ECs from Muc16^−/−^ KO mice demonstrated little adherence or subsequent endocytosis of hyphae-induced fungal cells in the presence (+) or absence (-) of EGTA (Fig. 1d). Endocytosis intensity ratios calculated as described^10^ were significantly reduced by EGTA for WT ECs, as expected, but the small amount of residual unlabeled fungal cells found in the Muc16^−/−^ ECs were not further reduced by EGTA (Fig. 1e).
To molecularly probe interactions between fungal Als3, MUC16 and N-cadherin, a proximity ligation assay was utilized after incubation of primary murine ECs with recombinant Als3, the latter described previously^31^. As shown in Fig. 1f and g, strong signal was recorded evidenced by multiple visible puncta, linking Als3 to MUC16 in WT mouse ECs compared to equivalent assays using Muc16^−/−^ KO mouse ECs. Detectable signal was also exhibited between Als3 and N-cadherin in WT mice consistent with earlier studies^32^ which was abolished in the Muc16^−/−^ KO ECs. Signal was also registered between MUC16 and N-cadherin, which was absent in the Muc16^−/−^ KO EC’s. In summary, these data demonstrate a role for mammalian MUC16 in Candida adherence to mouse primary kidney ECs and is required for subsequent endovascular invasion by a complex linking fungal Als3 and mammalian MUC16 and N-cadherin.
Molecular mediators of MUC16-dependent binding of C. albicans to mouse primary ECs.
Additional studies sought to confirm a role for the C. albicans hypha-specific surface adhesin Als3^32–34^ in MUC16-specific adhesion to mouse kidney ECs. As shown in Fig. 2a, the normally non-adherent Saccharomyces cerevisiae fungal cells expressing ALS3 (+ALS3) were found to exhibit significant primary mouse kidney EC adherence compared to equivalent strains expressing empty vector (EV) alone. However, the same S. cerevisiae ALS3-expressing strains showed little augmentation of binding to Muc16^−/−^ KO ECs, suggesting a role of Als3 binding to MUC16-expressing ECs in mice. Some augmentation in ALS3-dependent binding to the Muc16^−/−^ KO ECs, suggesting additional minor constituents involved in ALS3-dependent adherence; however, MUC16 appears to be the dominant molecule involved in adherence.
Further molecular interaction studies sought to identify structural elements within the MUC16 molecule involved in adherence. As shown in Fig. 2b, MUC16 is a large surface-bound mucin with a significant number of N-terminal glycosylation and multiple conserved SEA domains (Fig. S2a), the latter so called because of a structural similarity to the sea urchin sperm protein, enterokinase and agrin^35,36^. Initial experiments utilized a combination of neuraminidase + b1,4 galactosidase to reduce N and O-labeled glycosylation of the surface mucin to determine if adherence might also be reduced. Surprisingly, this treatment increased adherence (Fig. 2c), which suggested an alternative domain that may have been sterically shielded by the extensively glycosylated N-terminus. Since autocatalytic processing of MUC16 is known to expose pauci-glycosylated SEA domains, we expressed a fusion protein of 8 murine SEA domains (a.a. 8682–8805; Fig. S2b), previously found to mimic the antigenic site of the anti-human CA125 antibody by NMR spectroscopy,^37^ with the affinity tag maltose-binding protein (MBP). As shown in Fig. S2c, the rMBP-SEA8 purified as a 145kD protein with some step-ladder proteolysis typical of the properties of this self-cleaving protein^38^, compared to the 42-kDa MBP. As shown in Fig. 2d (green arrows), rMBP-SEA8, but not rMBP alone was found to exhibit strong binding to hyphae of C. albicans but, interestingly, not the yeast forms of the fungus that are nonpathogenic during invasive disease and do not express Als3p. Flow cytometry confirmed the increased overall binding of the rMBP-SEA8 compared to rMBP alone (Fig. 2e,f). Addition of rMBP-SEA8 was also found to block endothelial adherence of C. albicans hyphae to levels equivalent to that of Muc16^/-^ ECs whereas rMBP alone showed no significant reductions in adherence (Fig. 2g). These results were duplicated using a single SEA domain protein fusion to MBP at either the N or the C-terminus (Fig S2d-g). rMBP-SEA1 binding to Als3p was further confirmed by co-immunoprecipitation experiments which showed little binding to rMBP alone (Fig. S2h, i). Interestingly, rMBP-SEA1 also retained binding to Als3-GK which has a mutation within the complement binding pocket described previously, suggesting an alternative novel binding site^31^. In summary, these data suggest a role of fungal Als3 in MUC16-dependent fungal adherence to primary mouse kidney ECs as well as a role for the MUC16 SEA domains in C. albicans-EC binding.
MUC16 is expressed in CD31+ renal ECs and co-localizes with renal vasculature and invading Candida hyphal elements.
While MUC16 is a well-known epithelial-related mucin, less is known about its role as a renal endothelial mucin. As shown in Fig. 3a, kidney tissue demonstrated robust expression of MUC16 as well as N-cadherin within glomeruli and surrounding structures of mouse kidneys by immunostaining. In addition, immunofluorescence (IFA) demonstrated colocalization of MUC16 with CD31+ staining ECs in vessel walls, without significant MUC16 staining in Muc16^−/−^ mutant mice (Fig. 3b). Furthermore, 2 hours after intravenous inoculation of C. albicans hyphae expressing fluorescent d-Tomato (Fig. 3c), fungal cells (red arrows) localized within CD31+ renal vascular structures in WT mice (green arrows), expressing MUC16 (center and left panel, purple arrow). Endothelial adherence was persistent at 4 days (Fig. 3d) with occasionally observed penetration of the endothelial layer by fungal hyphal structures (* and red arrows). In contrast, fungal organisms were not observed localized to vascular endothelium in the Muc16^−/−^ KO animals. In summary, these data suggest that MUC16 is expressed in renal ECs of intact mouse kidneys and intravenously-inoculated fungal cells co-localize to the MUC16-expressing EC layer, consistent with the MUC16-dependent adherence observed in purified mouse kidney ECs.
Deletion of MUC16 in mice conveys protection in an intravenous dissemination model of Candida albicans infections
The significance of MUC16 in systemic Candida pathogenesis was further studied using an intravenous dissemination candidemia model that results in mortality mediated primarily by fungal renal vascular invasion followed by pathogenic renal neutrophil invasion^11^. As shown in Fig. 4a, Muc16^−/−^ 129 × C57BL/6N strain F1 progeny exhibited significant protection after intravenous challenge with 1 × 10^5^ CFU of C. albicans compared to littermate WT F1 controls. To confirm the phenotypic difference, Muc16^−/^mice were then backcrossed to C57B6 mice 5x using a high-speed congenic strategy and first-wave male germ cells with 100% B6-homozygous loci (84 markers)^39^ and were again found to exhibit significant protection compared to WT littermate control mice at even a higher inoculum of 2.5 × 10^5^ CFU (Fig. 4b). Fungal burdens were shown to be reduced in the Muc16^−/−^ mouse kidneys by CFU (Fig. 4c) with WT excised kidneys demonstrating increased edema and hypopigmentation compared to Muc16^−/−^ mice (Fig. 4d) and histology (Fig. 4e) showing increased Grocott’s methenamine silver (GMS)-staining fungal elements in the WT mice compared to the Muc16^−/−^ mice. In addition, fungal burdens in Muc16^−/−^ mice showed reductions in brain, liver or lung compared to WT, though less consistent in brain (Fig. S4). Cellular studies also demonstrated reduced accumulation of ly6C^int^CD11b^+^Ly6G^hi^ neutrophils (Fig. 4f), Ly6C^hi^CD11b^+^ inflammatory monocytes (Fig. 4g) and MHCII^+^F4/F80^+^CD11c^−^ macrophages (Fig. 4h) in Muc16^−/−^ mice commensurate with the reduced fungal burden found in this organ and improved survivals. These data demonstrates a role for MUC16 in potentiating fungal renal vascular invasion, associated with increased neutrophil accumulation and mortality.
Deletion of MUC16 in mice conveys protection in an oral model of Candida albicans infections
Since MUC16 is well-known as an epithelial mucin^35^, the Muc16^−/−^ strain was also compared to littermate WT controls using two adapted oral models of C. albicans including one recently used to demonstrate a role for innate lymphoid cells in oral candidiasis^40^. Without immunosuppression, WT mice inoculated orally with a solution of C. albicans displays a transient reduction in weight (Fig. 5a). In addition, since corticosteroids are a known risk factor for severe mucocutaneous candidiasis^41^, an additional oral model of candidiasis in mice treated with cortisone acetate was utilized^42^, which demonstrated significant protection from infection as measured by percent of body weight loss and reduced tongue colonization measured by fungal CFU (Fig. 5b, c) or histology of GMS-stained tongues at 4 days (Fig. 4d). Commensurate with less fungal burden, Muc16^−/−^ mice demonstrated less cellular recruitment of neutrophils (Fig. 5e), monocytes (Fig. 5f) and macrophages (Fig. 5g) within the tongue tissue [summary statistics day 1 (Fig. 5h) and day 4 (Fig. 5i)]
Expression of MUC16 is induced via TLR and C-type lectin receptor ligands and is CARD9-dependent.
Since MUC16 plays a role in initiating and determining severity of candidal infection in mice, we hypothesized that MUC16 expression could be induced by factors associated with increased mortality of candidemia in patients, including pathogen products released during either transient seeding from central venous catheters or bacterial infections associated with antibiotic use^43^. Indeed, hematologic malignancies are well known to induce MUC16 measured by the CA125 levels in blood which are known risk factors for candidemia^44^. In the present studies intravenous exposure of mice with C. albicans increased kidney levels of MUC16 after 4 days (Fig. 6a) and incubation of fungal cells with purified mouse EC monolayers for 3 hours also induced MUC16 (Fig. 6b). Similarly, incubation of heat-killed E. coli resulted in a dose-dependent induction of Muc16 expression in primary mouse kidney ECs at 3 h (Fig. 6c). Since pathogen-associated molecular patterns (PAMPs) are prominent signaling ligands of pathogens, we assessed for induction of Muc16 by selected PAMPs within purified mouse kidney ECs for 1 hour and 3 hours and found a striking induction of this mucin after stimulation of a number of TLR and C-type lectin receptors (Fig. 6d), including Candida-associated ligands Mincle, TLR9 and Dectin1 (Fig. 6e). Induction was dependent on Card 9 (Fig. 6f), suggesting the role of this pathway in PAMP-dependent signaling of MUC16. In addition, adherence to primary mouse ECs was found to be reduced in Card9^−/−^ mice, almost to a degree found in the Muc16^−/−^ mice (Fig. 6g). In conclusion, MUC16 was found to be an inducible endothelial mucin, mediated by both TLR and C-type lectin receptors and is Card9 dependent. This could suggest mechanisms whereby inflammatory signals prime MUC16 to provide increased receptors to bind transient fungal cells released during illness, perhaps facilitating endothelial adhesion prior to endothelial endocytosis and tissue invasion.
MUC16-rs11670318 single nucleotide polymorphism and serum concentrations of MUC16/CA125 are associated with mortality in humans.
To identify possible human MUC16 SNPs that could contribute to severity of human disease, we analyzed human DNA samples from a prospective genetic discovery study of non-neutropenic patients with bloodstream infections of Candida. Over a 3-year period in 2005, 37 consecutive eligible patients were recruited at the University of Illinois at Chicago after being identified by the microbiological laboratory and informed consent obtained (see entry criteria in Table S6). Demographic data demonstrated similarity between patients that died versus survivors, although there was a difference in APACHE II score (Table S7). Characteristics of the pathogen obtained in cultures are described in Table S7. The predominant organism was Candida albicans, although there were also isolates of C. glabrata and other strains. Using an additive genetic model adjusted for age and gender, two SNPs (rs2547065, rs7508222) among 80 tested SNPs (Table S8) were associated with mortality with a nominal p < 0.05. These two SNPs remained significant after further adjustment for the APACHE II score. Validation of three of the top 4 SNPs from the discovery cohort were analyzed in a second previously-described cohort of candidemia patients^4^, selecting for patients matching that of the discovery cohort. The top candidate SNP associated with protection from mortality was identified as rs11670318-A with an adjusted p value of 0.07. In humans, the MUC16-rs11670318 SNP results in an adenine-to-guanidine substitution at position 19:9056982 that changes a threonine at position 10155 to an isoleucine, found to be a site of O-GalNAc modification by a whole genome study of the glycoproteome^45^.
To determine whether the rs11670318 polymorphism might have the capacity to affect MUC16 Candida adherence, a human EC line, HMEC-1 was subjected to CRISPR-cas9-directed substitution of the rs11670318 SNP utilizing the guide RNAs described in Methods. Three clones were selected by limiting dilution and isolated, verified to have similar observed growth and morphology at confluence and assessed by PCR followed by sequencing to establish the correct substitution. As shown in Fig. 6A, adherence assays demonstrated modest but significant reductions in fungal binding, that could contribute to less burden of renal disease after transient intravenous fungal exposure.
While serum was not collected in the discovery cohort I, it was available in Cohort II which allowed examination of serum concentrations of MUC16 protein in 134 patients by a standard CA125 assay of MUC16 (Quantikine ELISA Human CA125/MUC, R&D Systems, Minneapolis, MN). CA125 is a widely-used assay of the MUC16 soluble SEA domain that facilitates staging and follow-up of malignancies, principally ovarian cancer, whose prognosis and outcomes are strongly associated with levels of MUC16^46^. As shown in Fig. 6i concentrations of MUC16 in serum were highly associated with 30-day survival in non-neutropenic patients with candidemia (p = 0.01) with an ROC of 0.7 (Fig. 6j), suggesting that higher levels of MUC16 expression, perhaps mediated by risk conditions such as transient bacterial or fungal products, may have resulted in poor clinical outcome similar to the mouse models. Further study in additional cohorts may be required to delineate more fully the role of CA125 monitoring in patients with candidemia, but, to our knowledge, these data provide the first readily available commercial assay to stratify outcome in human candidemia.
Discussion
ECs are among the first host cells encountered by Candida during hematogenous dissemination, and adhesion is the critical initiating step for invasion and tissue damage. Previous studies demonstrated that endothelial endocytosis of Candida is calcium dependent and requires N-cadherin engagement, with Als3 functioning as a key fungal invasion^10 34^. However, the molecular mechanisms governing the initial attachment of Candida to endothelial surfaces have remained undefined. Our data identify MUC16 as a central mediator of this process, demonstrated through adherence studies, proximity ligation assays, siRNA knockdown, genetic deletion, and functional inhibition. Our findings support a model (Fig. 6k) in which infection-associated inflammatory cues induce endothelial MUC16 expression, thereby creating a permissive adhesive platform for fungal Als3-mediated binding, endocytosis, and tissue invasion. This work establishes MUC16 as a “pathogen-inducible endothelial adhesin” that links host risk factors, innate immune signaling, and fungal virulence mechanisms to disease severity.
MUC16 is a large molecule which contains a distal glycoprotein as well as 16 proximal SEA (sea urchin sperm protein, enterokinase and agrin)-domains internal to the glycoprotein which are self-proteolytically cleaved at specific intervals to yield soluble and residual cell surface pauci-glycosylated fragments containing one or more residual SEA domains^47^. The SEA domain is highly conserved evolutionarily and is known to assist or regulate binding a number of carbohydrate moieties^48^, E and L-Selectins^49^, as well as mammalian proteins including mesothelian, facilitating cell-cell adhesion and cancer metastasis^50,51^. In the present study, recombinant SEA domains bound preferentially to Candida hyphae, the morphological form responsible for endothelial adherence and invasion. These findings support a model in which exposed SEA domains provide a compatible binding interface for fungal adhesins.
Als3 is the best-characterized member of the ALS family of Candida adhesins and plays a central role in endothelial adherence and invasion^34^. Structural and functional studies of Als3 have demonstrated its capacity to engage multiple host ligands through distinct interfaces, enabling multivalent adhesion^32^. Interestingly, we observed SEA domain binding to the Als3 gatekeeper (GK) variant S179Y^31^, suggesting that binding of this ligan may be independent of the Als3 peptide binding cavity.
Clinically, Als3 is highly relevant, as vaccination with the Als3-derived NDV-3A reduces recurrent vulvovaginal candidiasis and induces protective antibody responses^52^. Our identification of MUC16 as an endothelial binding partner of Als3 extends these findings and provides a mechanistic basis for targeting host–fungal adhesion as a therapeutic strategy. Mechanistically, our data support a ternary adhesion complex involving fungal Als3, endothelial MUC16 SEA domains, and N-cadherin. N-cadherin mediates calcium-dependent homophilic adhesion and undergoes lateral clustering and polymerization at the EC surface, processes that are dynamically regulated during junction remodeling and endocytic uptake^53–55^. We propose that MUC16 facilitates stable initial attachment of Candida, positioning Als3 for productive engagement of N-cadherin and subsequent endocytosis. This multi-receptor model provides a ‘missing link,’ facilitating Als3–N-cadherin interactions required for invasion by supporting early adhesion events and recruitment of cadherins involved in endocytosis and tissue invasion^32^.
The kidney is a principal target organ in disseminated candidiasis, and endothelial interactions are critical during early infection. Using purified murine kidney ECs and intact mouse models, we demonstrate that MUC16 is expressed on renal vascular endothelium, co-localizes with CD31, and facilitates rapid fungal adherence and endocytosis. Genetic deletion of Muc16 resulted in significantly reduced renal fungal burdens, diminished neutrophil accumulation, and improved survival following intravenous Candida challenge. These findings provide mechanistic details supporting kidney-specific endothelial invasion as a driver of immunopathology and organ failure^11^.
In contrast, MUC16 plays a more limited role in oral candidiasis models, consistent with a plethora of additional epithelial-specific receptors and virulence mechanisms at mucosal surfaces, including EphA2-mediated β-glucan sensing and candidalysin-induced epithelial damage^13,15^. These observations highlight the compartment-specific nature of host–fungal interactions and underscore the unique importance of endothelial adhesion mechanisms in disseminated disease. However, a paucity of Candida-binding receptors in ECs is consistent with the primary property of Candida as a commensal organism that would likely gain little evolutionarily by facilitating vascular interactions that could kill the host. The present findings thus help to understand a long-known but poorly understood property of Candida pathogenesis.
A key finding of this study is that endothelial MUC16 expression is inducible by microbial and inflammatory signals associated with candidemia risk. Bacterial lysates, LPS, and multiple fungal PAMPs triggered robust MUC16 upregulation in ECs. Notably, induction by ligands for Dectin-1, Mincle, and TLR9 was CARD9 dependent. While CARD9 signaling downstream of C-type lectin receptors is well established, TLR9 has not previously been linked to CARD9-mediated pathways^56^ suggesting a novel signaling axis regulating endothelial adhesion molecule expression. Blocking this pathway in CARD9-deficient ECs abrogated Candida adherence, supporting a functional role for PRR/CARD9 MUC16 induction. These findings extend emerging concepts that ECs act as active innate immune sensors rather than passive barriers during fungal infection^57,58^ including recent evidence that serum bridging molecules can drive endocytosis through specific endothelial receptors^59^. These results also shed light on an important Candida-specific disease related to CARD9. For example, in a cohort of patients with CARD9 polymorphisms no significant association was identified between CARD9 variation and the prevalence of candidemia, even though the Dectin-1/CARD9 pathway was nonredundant in mucosal immunity to Candida albicans^60^ Such a result could be due to limited CARD9-dependent expression of endothelial receptors such as MUC16.
Given the clinical relevance of endothelial adhesion in dissemination, we investigated whether MUC16 variation influences outcomes in human candidemia in the most common risk group today, non-neutropenic patients with risk factors of intensive care residence or hematologic malignancies. In two independent cohorts, we identified three non-synonymous MUC16 polymorphisms associated with mortality in a discovery cohort and a fourth variant (T10155I) that showed consistent trends across cohorts. Functional validation using CRISPR–Cas9–engineered HMEC-1 cells demonstrated that the T10155I mutation significantly reduced Candida adherence, supporting a mechanistic link between genotype and endothelial function. While the observed p values did not uniformly meet conventional thresholds, the consistency across cohorts and functional corroboration argue for biological relevance, consistent with modern statistical perspectives that emphasize effect sizes and reproducibility over arbitrary cutoffs^61^.
In addition to genetic risk, we identified a strong association between elevated serum MUC16/CA125 concentrations and mortality in candidemia. This finding is particularly intriguing given that known risk factors for invasive candidiasis, including hematologic malignancies, inflammation, and prior infections, are themselves associated with increased MUC16 expression. We propose that these conditions may “prime” the endothelium by increasing MUC16 availability, thereby facilitating fungal adhesion and invasion once candidemia occurs (Fig. 6k). While cohort size limits definitive conclusions, further studies could suggest a role for CA125 in biomarker-stratified clinical designs where selection of higher risk patients may facilitate the use of smaller patient sample sizes and/or improved power as recently suggested for biomarker-driven clinical studies^62^.
In summary, this study identifies MUC16 as a pathogen-inducible endothelial adhesin that integrates host inflammatory signals, innate immune pathways, and fungal virulence mechanisms to drive disseminated candidiasis. By defining a novel Als3–MUC16–N-cadherin adhesion axis and linking endothelial biology to clinical outcomes, our findings open new avenues for host-directed therapies, anti-adhesion strategies, and biomarker-guided clinical trial design in a disease that continues to carry unacceptably high mortality.
Materials and Methods
Mice, yeast and bacterial strains
Wild-type C57BL/6 mice were obtained from Taconic or Jackson Laboratory. Muc16^−/−^ mutant mice in a 129 background were a gracious gift from the Department of Genetics, University of Texas M. D. Anderson Cancer Center, Houston, Texas and were backcrossed to C57B6 mice 5x using a high-speed congenic strategy and first-wave male germ cells with 100% B6-homozygous loci (84 markers)^39^ C57/Bl/6N card9^−/−^ mice were the generous gift of M. Lionakis.
Candida albicans strain SC5314 was obtained from American Type Culture Collection (Manassas, VA), C. albicans-dTomato was a kind gift of R. Wheeler,^30^, and E. coli strain 25922 was a generous gift from the NIH clinical center microbiology laboratory (Bethesda, MD).
Isolation of mouse primary kidney ECs
Mouse tissue ECs were isolated as described^20^ with some modification. Briefly, mouse kidneys were aseptically removed washed, dissected, finely minced and washed, digested with Liberase at 37°C with agitation for 25–35 min. The suspensions were filtered twice centrifuged at 300g for 10 minutes and the pellets washed three times in phosphate buffer saline (PBS) containing 0.5% bovine serum albumin (BSA) and 2 mM ethylenediamine-tetraacetic acid (EDTA). EC purifications were performed using CD146 MicroBeads positive selection kit according to the manufacturer’s instructions. Isolated cells were plated onto T25 flasks precoated with 1% gelatin and cultured at 37° C in a 5% CO2 in DMEM/F12 containing EC growth supplements. To verify the cell preparation, cells were tested for uptake of Dil-Ac-LDL according to the method of Voyta^27^, and cells were also stained with anti-CD45-PE-Cy7A, CD146 PE-A and Muc16 APC-A (Table S4) and subjected to flow cytometry. In addition, mouse kidney ECs demonstrated induction of E-selectin and VCAM-1^26,63^ as well as the endothelial marker, CD31/PECAM-1^22^.
Gene silencing of renal endothelial-associated mucins in mouse ECs
Gene silencing was performed using four siRNA plasmids targeting the mouse Muc16, Muc1, Muc14 and Muc13 genes and Accell Non-targeting Control Pool (D-001910-10-05, Dhar Macon) according to Materials and Methods. ECs were transfected with endotoxin-free siRNA plasmids containing the proprietary nucleotide sequences. Accell^™^siRNA delivery). Transiently transfected cells were assayed 72 h post-transfection.knockdown was quantified using real-time RT-PCR. Knockdown efficiencies ⩾70% were regarded as significant.
Adherence Assays of Candida to primary mouse kidney ECs
The method of Phan et al was utilized^10^. Briefly, 10^5^ mouse purified ECs were cultured in 6-well tissue culture plate until cells reached confluence. To determine Candida adherence, endothelial monolayers were washed twice with 2 ml of PBS, and 1 ml of the singlet Candida suspension (10^2^ cfu/ml of HBSS) was added to replicate monolayers. The inoculum size of each suspension was confirmed by culturing aliquots in YPD agar plates. Following incubation at 37°C, the nonadherent organisms were aspirated and the EC monolayers were rinsed with 10 ml of HBSS in a standard manner. Two milliliters of warm 2% glucose, 2% bactopeptone, 1% yeast extract agar (YPD) agar was poured over the surface and allowed to solidify. After incubation at 37°C for 24 h, colonies (representing adherent organisms) were counted, and the results were expressed as a percentage of the inoculum.
Measurement of endocytoses of C. albicans by ECs.
The number of organisms endocytosed by ECs was determined using the method of a previously described differential fluorescence assay with minor modifications^10^. Briefly, 10^3^ ECs were grown to confluency on 8 well chamber slides. The cells were rinsed once and infected with 10^2^ cells of dTomato-C. albicans in RPMI 1640 medium plus or minus 1.9 mM ethylene glycol tetraacetic acid (EGTA) for 45 min at 37 °C in 5% CO2. Next, the cells were rinsed twice with 0.5 ml of PBS in a standardized manner and then fixed with 3% paraformaldehyde. The adherent but non-endocytosed organisms were labeled with polyclonal rabbit anti-C. albicans antibodies (Table S2) that had been conjugated with green fluorescing AF488 (Molecular Probes). ECs were then permeabilized with 0.5% Triton X-100 (Merck-Sigma) in PBS, after which both the endocytosed and non-endocytosed organisms were visualized by microscopy using Imaris Software. The number of cell-associated organisms (endocytosed plus adherent organisms) was determined by counting the number of organisms that were labeled with Alexa 488. The number of endocytosed organisms was determined by subtracting the number of organisms labeled with red fluorescent dTomato from the number of cell-associated organisms.
Proximity Ligation Assay.
The method of Soderberg et al^64^ was adapted using the commercial Duolink reagents (Millepore Sigma/Merck KGaA, Darmstadt Germany). Briefly, purified mouse ECs were grown to 70–80% confluence on 12-well culture slides, rinsed once and 5×10^4^ hyphae of C. albicans SC5314 were incubated with EC monolayers in DMEM/F12 medium alone. Cells were rinsed twice with PBS and then fixed for 15 min at room temperature. For primary antibody incubation, the cells were treated with primary antibodies respectively for 2–3 hours in 3 groups: 1) rabbit anti-MUC16 antibody and mouse anti-Als3 antibody described previously^65^ (a generous gift from L Hoyer). 2) Rabbit anti N-cadherin Polyclonal Antibody (company PA5–19486) and mouse anti-Als3 antibody. 3) rabbit anti-MUC16 antibody and mouse anti-N-cadherin monoclonal antibody. Subsequently, probe incubation, ligation, and amplification were performed according to the manufacturer’s operating procedures. Puncta were captured by fluorescent (FITC) microscopy using Imaris Software.
Comparison of human and mouse MUC16 and expression and fungal binding of recombinant huSEA domains:
Recombinant SEA domains were expressed as a synthetic gene comprising 8 human SEA domains (aa 8682–8805; Fig. S2B) (GenScript, Piscataway, NJ), as a fusion protein with maltose-binding protein (MBP) by using the pIH902 expression system (New England Biolabs, Beverly MA) purified on amylose-Sepharose according to the manufacture’s directions. Alternatively, a single SEA domain was expressed with either an N-terminal or C-terminal MBP. Intact sequence was verified by Sanger sequencing prior to use.
SEA domain binding to C. albicans was performed as follows: d-Tomato-expressing C. albicans described above was induced for hyphal formation followed by incubation of 10^3^ CFU with 100 mg/ml of rSEA-MBP or MBP alone for 2 h at 37C followed by extensive washing and incubation with anti-MBP antibody (Table S2) and visualized by epifluorescence or by flow cytometry.
Isolated Kidney Perfusion and Tissue Clearing
Mouse kidney perfusions were performed as described^66^ with modifications. Briefly, wild-type mice were anesthetized. Mouse kidneys were perfused through the abdominal/renal artery with 10 ml of ammonium-chloride-potassium lysing buffer containing 10 unit of heparin followed by 1ml of 10^6^ CFU/ml of C. albicans-dTomato in PBS. After 2 h kidneys were perfused with PBS to remove unbound fungi. Excised mouse kidneys were post-fixed overnight and processed utilizing the method of Clear^T2^ tissue-clearing as described^67^. Briefly, kidneys were transferred to 25% formamide/10% polyethylene glycol (PEG) solution, 50% formamide/20% PEG solution, then clear in new 50% formamide/20% PEG solution. After sectioning (300 μm) on a vibrating microtome, the obtained kidney slices were blocked in blocking solution for 1 hour at room temperature. Goat anti-mouse CD31 and rabbit anti-mouse MUC16 were incubated overnight at 4°C. After washes AF488 conjugated anti-goat and AF647 conjugated anti-rabbit secondary antibodies were applied, incubated in 1%BSA/1%Tween/PBS overnight at 4°C. 4’6-diamindino-2-phenylindole (DAPI) was used for nuclear staining followed by washes in phosphate-buffered saline before epifluorescent imaging on an inverted SP8 microscope.
Mouse model of systemic candidiasis and oropharyngeal candidiasis (OPC)
The model of Huang et al was utilized^40^. Fungal burden was determined by excision of tissue and determination of colony forming units.
Histopathological and immunohistochemistry (IHC) for N-cadherin and Muc16 expression Studies
The method of Lionakis et al was adapted^11^. Briefly, control uninfected mice and infected mice were euthanized on days 1, 4, and 7 after infection and the kidney was removed, fixed with 10% formalin. The samples underwent sectioning with H&E or PAS staining. For IHC of N-cadherin and MUC16 expression, tissue sections were placed on poly- L -lysine-coated glass slides, deparaffinized in xylene and rehydrated. Following antigen retrieval in citrate buffer endogenous peroxidase was blocked and the slides were then incubated with 5% bovine serum albumin followed by washing with buffer containing sheep anti-N-Cadherin antibody or Rabbit anti-CA-125(MUC16) antibody. BSA (1%) without primary antibody was used as negative control. The slides were then washed twice and detection of immunoreaction was achieved using the streptavidin-HRP (horse radish peroxidase) containing the streptavidin-peroxidase complex after a 40-min incubation at RT. Color was developed with DAB 3 DAB (3,3’-diaminobenzidine tetrahydrochloride) substrate kit (Thermofisher, 34002) and slides were subsequently counterstained with hematoxylin, dehydrated, and mounted. Two different sections of each model were tested per experiment.
Flow cytometry
Control uninfected mice and infected mice were euthanized on days 1, 4, and 7 after infection. Then, the animals were perfused with normal saline and brain, spleen, liver and kidneys were harvested. Single-cell suspensions were obtained using methods previously described by Lionakis^11^. Further details including antibodies utilized can be found in Supplemental Materials. FACS was performed on an BD Fortessa (BD Biosciences) and data were analyzed using FlowJo software.
Induction of Muc16 in intact mice and in purified mouse kidney ECs.
For Candida induction of Muc16 in intact mice, C57/Bl/6 mice were inoculated intravenously with 2.5 × 10^5^ C. albicans, kidneys excised as described above and RNA assayed by qRT-PCR using primers described above. For Candida induction of Muc16 in ECs, 2 ×10^5^ CFU of C. albicans was incubated with 10^6^ ECs for 8 h in cell culture media and RNA assayed for Muc16 expression by qRT-PCR. For E. coli induction of Muc16 in ECs, E. coli strain 25922 (10^6^ CFU) were heat treated (65 C for 1h) and indicated equivalent amount of organism lysate dilution (0.01–100 organisms equivalent) incubated with 10^6^ of ECs and RNA assessed for Muc16 expression as above. For PAMP ligand induction, Pam3CSK4, Poly (I:C), lipopolysaccharide (LPS), fibroblast-stimulating lipopeptide (FSL), CpG oligodeoxynucleotide trehalose dibehenate FurFurman, Laminarin, Poly IC-LMW bacterial flagellin or single stranded RNA40 (see Table S5) were incubated with EC from either C57/Bl/6N or littermate C57/Bl/6N card9^−/−^ mice according to the manufacturer’s instructions for either 1 h or 3 h and RNA isolated and assayed for MUC16 expression by qRT-PCR.
CRISPR Cas9 mediated T10155I KI of MUC16 in HMEC1 cells.
To generate cells, ribonucleoproteins containing Cas9 protein and synthetically chemically modified sgRNA produced by the manufacturer (Synthego, Redwood City, CA) was electroporated into HMEC1 cells. Guide RNA cut location was chr19:8,946,320. Editing efficiency was assessed upon recovery and genomic DNA was extracted from a portion of the cells, PCR amplified and subjected to Sanger sequencing. The pool was subjected to repetitive dilutional cloning and clones screened for the appropriate KI primers and Sanger sequencing and equivalent growth rates and confluent monolayer productions to that of the wild-type cells.
Single cell clonogenic assays were performed as described^68^.
Ethics Statement
The human prospective observational cohort studies were approved by the respective institutional review board at each institution (Cohort 1, The University of Illinois at Chicago and Cohort 2, Duke University Health System). All subjects or their legally authorized representatives provided written informed consent, or had a waiver as approved by the IRB.
Study Design and Participants
The UIC candidemia discovery study (Cohort I) was a prospective, observational study involving non-neutropenic patients admitted to UIC over a three year period and developed Candida bloodstream infections reported by the UIC clinical laboratory. Inclusion criteria are described in Table S6. (Demographic data in Table S7). Further details are described in Supplemental Materials. All infected patients were followed for up to 12 weeks to assess for clinical outcome.
Case-association study of mortality in Cohort I candidemia patients.
DNA collected at entry was genotyped. Eighty non-synonymous alleles of the MUC16 gene (Table S8) were identified from the GnomAD database^69^, and assessed using DNA from 37 patients using Taqman (ThermoFisher) and Sequenom (San Diego, CA) platforms according to the manufacturer’s instructions.
Within-case association study of mortality in Cohort II candidemia patients
The genotyping, quality control, and imputation of the genetic data was previously described^4^ Briefly, DNA was genotyped using Illumina HumanCoreExome-12 v1.0 and HumanCoreExome-24v1.0 BeadChip Snp chips. Genotypes were imputed using the human reference consortium reference panel^70^ using the Michigan imputation server^71^. Quality control filters resulted in a dataset of 156 candidemia patients. Of these 156 patients, 148 had available mortality data. We tested for association with mortality at 14, 30 and 90 days within the patient cohort using 31 non-survivors and 117 survivors, 49 non-survivors and 99 survivors, and 65 non-survivors and 83 survivors respectively. The association between the genetic variants and mortality were tested by Fisher’s exact test using PLINK v1.9 (www.cog-genomics.org/plink/1.9/)^72^. The corresponding receiver operating characteristic (ROC) and precision–recall (PR) curves were generated using the cumulative prtC abundance as the predictor and actual disease status (based on the clinical metadata) as the ground truth.
Supplementary Material
This is a list of supplementary files associated with this preprint. Click to download.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Mc Carty T.P., White C.M. & Pappas P.G. Candidemia and Invasive Candidiasis. Infect Dis Clin North Am 35, 389–413 (2021).34016283 10.1016/j.idc.2021.03.007 · doi ↗ · pubmed ↗
- 2Seagle E.E., The landscape of candidemia during the COVID-19 pandemic. Clin Infect Dis (2021).
- 3Keighley C., Candidaemia and a risk predictive model for overall mortality: a prospective multicentre study. BMC Infect Dis 19, 445 (2019).31113382 10.1186/s 12879-019-4065-5PMC 6528341 · doi ↗ · pubmed ↗
- 4Jaeger M., A Genome-Wide Functional Genomics Approach Identifies Susceptibility Pathways to Fungal Bloodstream Infection in Humans. J Infect Dis 220, 862–872 (2019).31241743 10.1093/infdis/jiz 206PMC 6667794 · doi ↗ · pubmed ↗
- 5Lionakis M.S., CX 3CR 1-dependent renal macrophage survival promotes Candida control and host survival. J Clin Invest 123, 5035–5051 (2013).24177428 10.1172/JCI 71307 PMC 3859390 · doi ↗ · pubmed ↗
- 6Swamydas M., CXCR 1-mediated neutrophil degranulation and fungal killing promote Candida clearance and host survival. Sci Transl Med 8, 322ra 310 (2016).
- 7Allert S., Candida albicans-Induced Epithelial Damage Mediates Translocation through Intestinal Barriers. M Bio 9(2018).
- 8Sheppard D.C. & Filler S.G. Host cell invasion by medically important fungi. Cold Spring Harb Perspect Med 5, a 019687 (2014).25367974 10.1101/cshperspect.a 019687 PMC 4292075 · doi ↗ · pubmed ↗