Reconstructing host-pathogen interactions: analysis of the intercellular network of macrophages and other immune cells within an aspergillosis-on-chip model

Abstract

The human-pathogenic fungus Aspergillus fumigatus can cause life-threatening infections, particularly in immunosuppressed or neutropenic patients with hematological malignancies. The human antifungal immune response primarily relies on neutrophils with contribution from macrophages, while the roles of unconventional lymphocytes such as Natural Killer (NK) cells and mucosal-associated invariant T (MAIT) cells remain less understood. Therefore, using an advanced microfluidic “invasive aspergillosis-on-chip” (IAC) model, we investigated the coordinated antifungal response of various immune cell types in combination with macrophages within a human tissue-like 3D environment. The IAC model, initially incorporating monocyte-derived macrophages, was enhanced by perfusing either primary neutrophils, NK cells, or MAIT cells for overnight infection experiments. In this model, hyphal growth of A. fumigatus was prevented only when both macrophages and neutrophils were present. However, uncontrolled growth was detected in neutropenic patient scenarios. We observed key interactions such as conidial shuttling and metaforosis between innate immune cells, accompanied by elevated cytokine levels of e.g. IL-1β and TNFα. Importantly, NK cells applied to the vascular compartment of the IAC model significantly reduced hyphal length, confirming their antifungal activity. Similarly, MAIT cells in sufficiently high quantities reduced hyphal growth and branching, although this effect depended on the recognition of riboflavin-derived antigens, as shown by the lack of response to the A. fumigatus ΔriboB mutant strain. Together, our findings demonstrate that the advanced IAC model is a powerful tool to dissect human antifungal immunity. The interactions observed between macrophages and other immune cells highlight the importance of multi-layered defense strategies against A. fumigatus.

The human invasive aspergillosis-on-chip model demonstrates that the interplay of macrophages with neutrophils, NK, and MAIT cells contributes to the effective control of Aspergillus fumigatus infection.

Introduction

Invasive fungal infections are considered a global health threat, which was recognized by the World Health Organization in the recently published list of priority fungal pathogens. Aspergillus fumigatus (A. fumigatus) was assigned to the critical priority group (World Health Organziation 2022). Invasive aspergillosis affects ∼300 000 patients in Europe each year and has a high mortality rate of 30%–85% (Lass-Florl and Steixner 2023). At risk are immunocompromised individuals such as solid organ or hematopoietic stem cell transplant (HSCT) recipients, patients in neutropenic conditions or on long-term glucocorticoid treatment (Kosmidis and Denning 2015, von Lilienfeld-Toal et al. 2019) as well as the increasing population of immunocompromised hosts treated with targeted therapies (Giannella et al. 2024). New risk factors, such as infections by SARS-CoV-2 or influenza virus, and the emergence of drug-resistant molds, point to the need of a better understanding of the antifungal immune response as a basis for the development of new therapeutic strategies.

The investigation of antifungal immunity in conventional in vitro experiments can provide important initial insights. Although the use of established cell lines or primary cells allows the analysis of the interaction of fungal pathogens with specific immune cells, these methods are unable to mimic the complexity of human tissue. In addition, mouse experiments can only recapitulate rather limited aspects of the human pathophysiology because of the inherent difference between murine and human immunology (Mestas and Hughes 2004). To address this issue, we have developed a completely human cell-based microfluidic invasive aspergillosis-on-chip (IAC) model (Hoang et al. 2022), which represents a more complex cell system and furthermore allows the investigation of intercellular interactions between different immune cells under infection conditions. The IAC model comprises an air-filled “lung” compartment lined with human alveolar epithelial NCI-H441 cells as well as monocyte-derived macrophages and a medium-perfused “blood“ compartment lined by human umbilical vein endothelial (HUVEC) cells; the compartments are separated by a porous PET membrane (Deinhardt-Emmer et al. 2020). Aspergillus fumigatus infection is carried out by apical application of conidia to the lung side, germinating into hyphae and possibly even growing invasively through the membrane pores into the blood side. The described setup of the IAC model allows not only the evaluation of the role of various immune cells in the interaction with A. fumigatus, but also the assessment of the possible significance of their localization on lung and/or blood side as well as the recruitment of cells from the blood to the site of infection. Furthermore, it opens up the opportunity to study pathological scenarios like neutropenia or even to use cells retrieved directly from patients who have an increased susceptibility to invasive aspergillosis.

Due to their small size of 2–3 µm in diameter, conidia of A. fumigatus can reach the alveoli of the human lungs, where they encounter professional phagocytes, in particular macrophages and neutrophils which form a major part of the innate cellular immune defense. Although macrophages already possess a large repertoire of antifungal activities (Heinekamp et al. 2015, Wang et al. 2023), it is also true that conidia employ several strategies to escape the macrophage’s antifungal activity (Jia et al. 2024) which underlines the necessity of recruitment of additional immune cells. Neutrophilic granulocytes are recruited from the blood stream and can act against conidia and hyphae by a plethora of mechanisms that include phagocytosis, release of reactive oxygen species, antimicrobial peptides, neutrophilic extracellular trap formation, production of antimicrobial extracellular vesicles and cooperative behaviour such as swarming (Urban and Backman 2020, Brakhage et al. 2021).

Moreover, at later stages of infection, innate immune cells are supported by adaptive immune cells, including T cells and B lymphocytes (Heung et al. 2023). Recent evidence suggests that novel, unconventional lymphocyte subsets may bridge the gap between innate and adaptive immunity, as they have been identified as new players in antifungal immune responses (Dunne et al. 2021). One subset are Natural Killer (NK) cells with antifungal activity reported in in vitro studies against A. fumigatus (Schmidt et al. 2017). NK cells recognize A. fumigatus germ tubes and hyphae, but not conidia (Schmidt et al. 2011). Upon activation they release cytotoxic granules containing granzyme and perforin, which damage the hyphae of A. fumigatus (Schmidt et al. 2011) and/or IFN-γ, which interferes with the metabolic activity and growth of the fungus (Bouzani et al. 2011).

Another subset of unconventional lymphocytes consists of mucosal-associated invariant T (MAIT) cells. They are located in mucosal tissues and organs and comprise up to 10% of the T cell population in the peripheral blood (Dusseaux et al. 2011). MAIT cells exhibit an evolutionarily conserved semi-invariant T cell receptor, which recognizes riboflavin metabolites (Patel et al. 2013) and induces a fast effector response including the release of pro-inflammatory cytokines (TNF, IFN-γ) and the release of cytotoxic granules (granzymes, perforin). Activation of MAIT cells by fungi has been described for yeast, Mucormycetes and Aspergillus species including A. fumigatus (Le Bourhis et al. 2010, Dias et al. 2017, Jahreis et al. 2018, Bottcher et al. 2021). However, it remains unclear whether activation of MAIT cells also results in fungal damage.

In this study, we used our advanced IAC model to closely investigate the antifungal potential of MAIT cells, NK cells, as well as the role of neutrophils under physiological and neutropenic conditions.

Results

Analysis of neutrophils in the IAC model extends previous results on their behavior and allows quantification of their migration speed

Previous studies using the IAC model (Hoang et al. 2022) have investigated the role of macrophages introduced to the lung side of the chip in the defense against fungal pathogens. This report examines the complex cellular interplay in lungs infected with A. fumigatus in greater detail, as these interactions were more precisely mimicked by perfusion of additional innate leukocytes on the blood side of the model. This improvement of the IAC model makes it possible to analyze the recruitment of neutrophils from the blood side to the site of infection and to investigate how this affects the kinetics of infection. First, we perfused Cytopainter Green (CPG)-labelled neutrophils on the blood side, followed by infection with A. fumigatus conidia on the lung side which already contained monocyte-derived macrophages. Neutrophils and monocytes were isolated from peripheral blood of the same donor. By microscopical analysis, we observed numerous neutrophils attached to the blood endothelium (Fig. 1A) and, less frequently, on the lung epithelium (Fig. 1B) which supports the notion of migration of neutrophils. In accordance with microscopical analysis, only a few (20–100 cells) neutrophils were obtained from the lung epithelium when they were rinsed off for further flow cytometric analysis. The rinsed cells were stained on the surface for the neutrophil markers CD15 and CD16 as well as for the cell adhesion molecule CD62L. A strong reduction of CD15 and CD62 L expression after infection proved the activation of detached neutrophils (Supplemental Fig. 1A). Since perfusion causes mechanical stress for neutrophils, cell viability was also determined by flow cytometry, which demonstrated that ∼80% of neutrophils remained viable during the whole experiment (Supplement Fig. 1B).

Figure 1.

Analysis of neutrophils in the IAC model extends previous results on their behavior. (A, B) Representative confocal laser scanning microscopy images of neutrophils perfused on the blood side of the IAC model. Fluorescence-labelled neutrophils (green) were found numerously on (A) the endothelial side (labelled in orange by von Willebrand factor) as well as sparsely on (B) the lung side (epithelial cells labelled in orange by E-cadherin). Nuclei stained in blue by Hoechst 33258. White arrow indicates neutrophil. (C) Scheme of migrating neutrophils in the IAC model. (D) Live cell imaging was used to record the epithelial side of the IAC model for 60 min upon addition of freshly isolated primary neutrophils (red) in the absence or presence of A. fumigatus (green). The field of view after 60 min of imaging is shown. (E) The vertical z-positions (line represents mean) of neutrophils detected within a z-stack of CLSM images covering the membrane (estimated position at about 60 µm) and epithelial side of IAC models with or without A. fumigatus was analyzed using the Imaris software and plotted. Data from three independent donors are shown. Scale bar = 50 µm.

To further investigate neutrophil migration in the IAC system, we tracked cell movement within the first hour of infection using live-cell imaging as well as z-stacks (Fig. 1C and D). We applied a constitutively GnoA-eGFP-expressing A. fumigatus strain and Cytopainter Deep Red pre-stained neutrophils. In all three replicates, no more than six neutrophils per field of view (0.58 mm²) migrated from the blood to the lung side in the absence of fungal conidia (Fig. 1D). By contrast, this was increased after infection with A. fumigatus where we observed up to 24 neutrophils per field of view. In addition, individual neutrophils were tracked in the videos and analyzed regarding their position on their way from the blood side through the membrane pores toward the lung side (Fig. 1D). By evaluating the different z-layers, it was possible to track the migration of the neutrophils from three different donors and determine their individual axial position in the IAC model (Fig. 1E). The mean z-position (representing the axial position) of the majority of neutrophils in all replicates was above 60 µm correlating to the blood side. After infection, a larger number of neutrophils moved to a position below 60 µm, which represents positions either inside the membrane pores or on the lung side, than in absence of fungus.

A prerequisite for an efficient antifungal immune response is the uptake of conidia. To analyze phagocytosis by neutrophils and macrophages, we labelled the two populations with Cytopainter Deep Red (red, neutrophils) or CellTracker^TM Orange (orange, macrophages) and used live-cell imaging again (Supplemental video V1). Ninety minutes after infection neutrophils were perfused to the blood side and the chip was imaged. During the first 10 min of imaging, macrophages already internalized conidia while first neutrophils appeared on the lung side. Within the following 4 h the majority of conidia was phagocytosed, mainly by macrophages but also by neutrophils (Fig. 2A). Macrophages could only be tracked for 4 h post-infection due to a fading cell labelling. During this time their speed of movement was calculated (Fig. 2B). The fastest cells moved with 0.13 µm sec⁻¹ before phagocytosis and slowed down as the number of internalized conidia increased. Based on the speed of macrophages we observed that internalized conidia exhibited sub-clusters that could be classified as i) a few fast movers, ii) some with a medium speed, and iii) the majority of cells as slow movers (Supplemental Fig. 2). In the same experiment, neutrophils arrived at the site of infection (lung side) with a top speed of 0.1 µm sec⁻¹ (Fig. 2C). Upon phagocytosis of the first one or two conidia, the speed was reduced to 0.05 µm sec⁻¹. In the course of infection (4–8 h), the speed was further reduced upon internalization of additional conidia. During live cell imaging, the images of the lung side were captured at an interval of every 90 sec. Remarkably, throughout the live cell imaging, neutrophils were consistently observed to appear in several consecutive image frames with an elevated speed on the lung side even after 16 h after infection, despite their absence in the preceding image (Supplemental video V2). This indicates that newly recruited neutrophils were translocating from the vascular compartment to the pulmonary side throughout the entirety of infection.

Figure 2.

Quantification of cell migration speed upon phagocytosis. Live cell imaging of the epithelial side of the A. fumigatus-infected IAC model with macrophages and neutrophils was performed for 20 h. (A) The majority of phagocytosis of FITC-labelled conidia (green, green arrow) by CellTracker Orange-labelled macrophages (orange, orange arrow) or Cytopainter Deep Red-labelled neutrophils (red, red arrow) occurred during the first 4 h. After 4 h, macrophages had lost their fluorescent signal due to photobleaching. (B, C) The speed of each macrophage (B) or neutrophil (C) detected in the videos within the indicated time period was analyzed using the Imaris software and plotted against the respective number of phagocytosed conidia. Data are shown as fitted Gaussian distributions. The black line connects the mean values of the fitted gaussian functions, while the red dashed line represents the fitted curves of the top three speed values corresponding to the fastest immune cells. Cells moved expeditiously without but slowed down as the number of internalized conidia increased. Scale bars = 50 µm.

We further focused on co-operative processes of macrophages and/or neutrophils against A. fumigatus. Swarming, i.e. the accumulation of neutrophils along growing hyphae was observed during late stages of infection (16–20 h, Supplemental video V3). In addition, the “shuttling” of conidia already described in the literature (Jia et al. 2024) was verified in the IAC model, whereby conidia were transferred from neutrophils to macrophages or in general from one immune cell to another. As shown in the Supplemental video V4, a neutrophil loaded with multiple conidia moved in the vicinity of macrophages, attached to one of the cells, and transferred the spores to the macrophage. We also observed macrophage—macrophage interaction known as metaforosis, an event that has been well described (Shah et al. 2016, Armstrong-James et al. 2018, Camilli et al. 2021). Here, a dying macrophage transferred its conidia to a healthy macrophage thereby preventing the release and subsequent germination of these conidia (Supplemental video V5). None of these observations was further quantified or analyzed, as they only occurred sporadically within the observed part of the IAC models.

Only the addition of a number of physiologically relevant neutrophils leads to complete inhibition of fungal growth in the presence of macrophages

Previously, we showed that macrophages reduce fungal growth in IAC models, but cannot completely prevent it (Hoang et al. 2022). This raised the question of whether the addition of neutrophilic granulocytes is essential for controlling fungal growth and, consequently, for the manifestation of a pulmonary infection. Therefore, IAC models containing either one cell type (macrophages or neutrophils) or a combination of two types were analyzed for hyphal growth 14 h after infection (Fig. 3A). Formation of hyphae was only observed when just one type of immune cell was present. In IAC models with a combination of macrophages and neutrophils, hyphal growth was either nonexistent or severely impaired. The synergy between macrophages and neutrophils can effectively reduce the quantity of germinated fungal spores from 60%–70% in monocultures to 20% in the model with both cell types (Fig. 3B). Besides the advantageous effect of addition of neutrophils to the IAC model regarding the inhibition of hyphal growth, we investigated the modulation of cytokine secretion. While IL-1β, TNFα as well as the ROS-producing enzyme myeloperoxidase (MPO) were barely detectable in the absence of neutrophils, a strong increase of these cytokines was seen in IAC models with perfused neutrophils (Fig 3C). IL-6, which is also produced by endothelial cells, showed no significant differences when neutrophils were present. It has been shown that interferons produced by macrophages are important stimulators of neutrophilic activity upon A. fumigatus infection. Therefore, we investigated IFN-λ1, IFN-α and IFN-γ in our experiments but were only able to detect IFN-λ1 (Supplemental Fig. 1C).

Figure 3.

A physiologically relevant number of neutrophils leads to complete inhibition of fungal growth in the presence of macrophages. (A) After 14 h incubation, fungal hyphae (blue filaments) did not grow on the epithelium (stained for E-cadherin, orange) of IAC models in the presence of both macrophages (0.1 × 10⁶ cells, unstained) and Cytopainter Green-labelled neutrophils (2 × 10⁶ cells, green) but in presence of either macrophages only or neutrophils only. Nuclei were stained with Hoechst 33 258 (blue spheres). (B) Germination of A. fumigatus conidia on the lung side of IAC models in presence of both macrophages and neutrophils was significantly reduced compared to models containing only one cell type. Confocal laser scanning microscopy was used to quantify the percentage of germinated conidia on 3–5 tile scans per condition. Each dot represents a measurement from an individual donor (n = 3). (C) Secretion of pro-inflammatory cytokines in supernatants collected from the endothelial side of IAC models after overnight incubation in presence of macrophages and neutrophils or one cell type only. Each dot represents a measurement from an individual donor (n = 3). Statistical analysis for B and C was performed using the ordinary ANOVA test with Tukey’s post-test. * P < 0.05, ** P < 0.01 and *** P < 0.001 was considered significant. (D) Mimicking neutropenic conditions, fungal hyphae stained by calcofluor white (blue) were observed in all IAC models containing less than 2 × 10⁶ neutrophils. Nuclei were stained with Hoechst 33 258 (blue spheres). E) Image analysis-based quantification of hyphal length, number of branches and % invasive hyphae from IAC models with 0.5 × 10⁶ or 0.05 × 10⁶ neutrophils. With 2 × 10⁶ neutrophils, no hyphae were detected and hence could not be assessed. Data points represent individual hyphae from three independent experiments with individual donors (n = 3). Results were compared by the Mann–Whitney Test. The red line represents the median in all graphs. Scale bars = 50 µm.

An advantage of our advanced IAC model is the possibility to adapt the experimental conditions to simulate patient risk factors. Since neutropenia is the main risk factor for invasive aspergillosis, we mimicked these conditions by reducing the number of perfused neutrophils from 2 × 10⁶ (equaling 1818 per µl, which corresponds to a healthy neutrophil count defined to be above 1.5 × 10⁹ per liter, (Fioredda et al. 2023) in human blood) down to 0.5 × 10⁶ (= 454 per µl) and 0.05 × 10⁶ (= 45 per µl) neutrophils corresponding to severe neutropenia defined by a neutrophil count of less than 0.5 × 10⁹ per liter (= 500 per µl). While no hyphae were observed within the physiological setting at all, hyphal growth was detected under both neutropenic conditions (Fig. 3D). In addition to evaluating and quantifying hyphal growth in the context of lower neutrophil concentrations, we observed that the length and the number of branches of hyphae grown under neutropenic conditions were comparable. Moreover, invasive hyphae were observed under both conditions (Fig. 3E).

NK cells show antifungal activity in the IAC model

Besides neutrophils, lymphocytes play an important role in the further course of the infection (Heung et al. 2023). We therefore used our IAC model to investigate the role of NK cells in combating A. fumigatus in a more complex system than the classic in vitro confrontation assays. Due to the experimental design NK cells and MAIT cells were isolated from different donors than those used for neutrophil isolation. However, macrophages were always derived from the corresponding lymphocyte donor. CPG-prelabelled NK cells were perfused on the endothelial side of the IAC model and were detected on both the endothelial and epithelial side following infection. This way, the migration of NK cells from the blood to the site of infection was verified in the IAC model. For a better visualization, the endothelial and epithelial sides were stained accordingly, the endothelium for von Willebrand factor (red, Fig. 4A) and the epithelium for E-cadherin (orange, Fig. 4B).

Figure 4.

NK cells show antifungal activity in the IAC model. (A, B) Representative confocal laser scanning microscopy images of NK cells perfused on the blood side of the IAC model. Nuclei are stained with Hoechst 33 258 (blue); Cytopainter Green-prelabelled NK cells (green) are detectable on (A) the endothelial side stained for von Willebrand factor (red) as well as (B) the epithelial side stained with E-cadherin (orange). White arrows indicate NK cells. (C) Representative 3D confocal laser scanning microscopy images of hyphae in the IAC model (upper panel) and software-based reconstruction of hyphae made with Imaris (lower panel). Conidia were stained with FITC (green) and hyphae with calcofluor white (blue). Same colors were chosen for the reconstruction images. Left panel shows typical fungal growth without NK cells, while in presence of perfused NK cells (right panel) hyphae are shorter and less branched. (D) Bioinformatic analysis of reconstructed hyphae revealed reduced hyphal length in the presence of NK cells. Dots represent absolute values from n ≥ 5 measured hyphae per condition from five different donors. Results are shown as scatter plot with median in red. (E) After infection, supernatants of the endothelial side from the IAC model contained elevated levels of lytic enzymes analyzed with a bead-based immunoassay by flow cytometry for GrA (granzyme A), GrB (granzyme B), Perf (perforin), GrL (granulysin). Results from D and E were statistically evaluated using the Mann-Whitney test. * P < 0.05. All IAC models contained monocyte-derived macrophages from the corresponding NK cell donor. Scale bar = 50 µm.

Confocal laser scanning microscopy images revealed extensive hyphal growth and branching of A. fumigatus in the control condition with only alveolar macrophages present. In contrast, an addition of NK cells to the IAC model reduced hyphal length (Fig. 4C and D). Bioinformatic image analysis of reconstructed hyphae confirmed that the perfusion of NK cells on the blood side of the IAC model led to significantly shorter hyphae, whereas the number of branches was not affected (Supplemental Fig. 3A).

Furthermore, enhanced levels of granzyme A and B, granulysin and perforin on the endothelial side of the IAC model were detected when NK cells were present (Fig. 4E). These molecules were also slightly increased on the epithelial side of the IAC model containing NK cells, but absolute concentrations were lower compared to the endothelial side. Levels of the proinflammatory cytokines IL-6, IL-8, IL-1β, TNF as well as MCP-1 or IFN-γ remained unaffected by the presence of NK cells (data not shown).

MAIT cells contribute to antifungal activity in vitro and in the IAC model

As MAIT cells are activated by fungal pathogens, we investigated whether this activation may also influence fungal growth directly. To first assess the impact of MAIT cells on hyphal growth under static conditions, we developed a 2D infection assay in which macrophages were infected with A. fumigatus conidia in either the presence or absence of MAIT cells and the consequent hyphal growth was analyzed by fluorescence microscopy. MAIT cells were observed in close proximity to individual macrophages and reduced the hyphal length by 20% compared to coincubation of conidia with macrophages alone (Fig. 5A). However, the germination rate of conidia was not influenced by MAIT cells (data not shown). Previously, we found that a riboflavin deficient A. fumigatus mutant (ΔriboB) was unable to activate MAIT cells (Jahreis et al. 2018). Using this A. fumigatus mutant in the 2D infection model, MAIT cells were unable to inhibit hyphal growth whereas the hyphal length of the corresponding wild-type strain AfS77 was reduced in the presence of MAIT cells, as seen before with the wild-type strain CEA10 (Fig. 5B).

Figure 5.

MAIT cells contribute to antifungal activity in vitro and in the IAC model. (A) Macrophages were differentiated in chamber slides and infected with A. fumigatus conidia for 8 h in the absence or presence of MAIT cells. Slides were stained for confocal laser scanning microscopy and hyphal length was analyzed with Imaris. For each condition n ≥ 15 hyphae were imaged, experiments were performed with cells from five different donors (A) or seven different donors (B). Results are shown as scatter plot with median in red and statistically analyzed with Mann–Whitney test. (C–E) Representative confocal laser scanning microscopy images of MAIT cells in the IAC model perfused either on the endothelial part (C) or seeded on the epithelial side (D+E). Nuclei are stained with Hoechst 33 258 (blue). Cytopainter Green-prelabelled MAIT (green) cells were found on (C) the endothelial side stained for von Willebrand factor (red) as well as detectable on (D) the epithelial side stained for E-cadherin (orange), sometimes in close proximity to hyphae (stained with calcofluor white in blue) or (E) an alveolar macrophage (stained for CD68 in red). White arrows indicate MAIT cells. (F) Representative 3D confocal laser scanning microscopy images of hyphae in the IAC model in the absence or presence of MAIT cells perfused on the blood side or seeded on the lung side. Conidia were stained with FITC (green) and hyphae with calcofluor white (blue). (G) Quantification of reconstructed hyphae revealed reduced hyphal length and number of branches with increasing numbers of MAIT cells seeded on the lung side. Dots represent absolute values from n ≥ 5 measured hyphae per condition from three different donors. Results are shown as scatter plot with median in red and analyzed with Kruskal–Wallis test. * P < 0.05, ** P < 0.01, *** P < 0.001. All IAC models contain monocyte-derived macrophages from the corresponding MAIT cell donor. Scale bar = 50 µm

Given the inhibitory effect of MAIT cells on hyphal growth observed in our 2D infection model, we next applied MAIT cells to the IAC model to assess their role in the antifungal immunity in a more complex and physiologically relevant setting. We developed two different setups where MAIT cells were either applied to the lung (epithelial) side or to the blood (endothelial) side of our IAC model. MAIT cells seeded on the epithelial compartment remained on this side of the biochip after infection, sometimes in close proximity to hyphae or macrophages (Fig. 5C). Similar to NK cells, MAIT cells perfused on the blood side were first observed within the endothelium (Fig. 5D) and later migrated to the infected epithelial side. Microscopy-driven analysis suggested a trend toward impaired fungal growth when MAIT cells were present, independent of their localization within the IAC model. However, this observation was not statistically significant (Supplemental Fig. 3B). The effect was enhanced by application of higher numbers of MAIT cells to the epithelial side of the IAC model. Fivefold (0.5 × 10⁶) and ten-fold (1 × 10⁶) higher concentrations of MAIT cells significantly reduced hyphal length and the number of hyphal branches (Fig. 5G). Comparable to the 2D infection assay, we also tested the riboflavin deficient A. fumigatus mutant (ΔriboB) in the IAC model. The growth of this mutant was not affected, while the corresponding wild-type strain AfS77 had a significantly lower number of branches in the presence of MAIT cells and a trend towards shorter hyphae (Supplemental Fig. 3C).

Discussion

Research on the pathophysiology of invasive pulmonary aspergillosis in immunocompromised patients is currently limited by several factors, including the artificial setups during in vitro experiments and the inaccurate mimicking of human disease by mouse models. Microfluidic 3D tissue-like models can represent the human lung environment more accurately than traditional 2D cell culture, especially the organotypic structures and the blood vessels. These models are very important for the development of more effective lung disease treatments, personalized medicine, and for shifting away from animal testing (Park et al. 2024). Recently, animal-free setups such as “on-chip” systems, are pushed by the 3R principles (Replacement, Reduction and Refinement) in animal welfare (NC3Rs 2017). Thus, we developed the IAC model (Hoang et al. 2022), a human cell-based, immunocompetent and media-perfused microfluidic infection model of the alveolus to closely recapitulate the first 24 h of A. fumigatus infection in the human lung. Despite these improvements compared to classical confrontation assays, the IAC model still has limitations. In particular, these are the introduction of conidia to the system by suspension in media and not by air, as during inhalation and the inoculum of conidia per alveolar surface (1×10⁻³ per µm²) was orders of magnitude greater than what is normally inhaled by humans (7.4 × 10⁻⁵ per µm²) according to calculations (Pollmacher and Figge 2014). As described in detail in our previous report (Hoang et al. 2022), the infection conditions in live cell-imaging and endpoint experiments were designed to obtain a sufficiently large number of target structures (hyphae and immune cells) so that we were able to observe a statistically meaningful number of immune events even with the relatively small field of view of confocal microscopy. Therefore, an alternative approach could be to use light sheet microscopy that provides much larger fields of view acquired at a high rate and accompanied by only a moderately decreased spatial resolution. A further limitation of this model is the usage of the epithelial cell line NCI-H441. This cell line has been shown to behave similarly to primary normal human bronchial epithelial cells in a bacterial and viral co-infection model in regard to cellular response or permeability (Deinhardt-Emmer et al. 2020). However, NCI-H441 are of an AT-II type which does not reflect the in vivo situation where the alveolar epithelium is largely composed of AT-I cells. Although beyond the scope of this report, it is possible to integrate primary human alveolar epithelial cells (available from commercial source in healthy and diseased states). In prior experiments with the IAC model, hyphal growth of apically applied conidia was not prevented by human monocyte-derived macrophages (Hoang et al. 2022) because conidia have evolved to escape the macrophages’ antifungal activity (Jia et al. 2024). In addition, experiments in mice have long shed light on the necessity to recruit further immune cells, such as neutrophils, to successfully fight against A. fumigatus (Mircescu et al. 2009). Moreover, it is established that prolonged neutropenia is the major risk factor for patients to develop invasive fungal infection (von Lilienfeld-Toal et al. 2019). Compared to rare or no hyphal growth upon perfusion of a “healthy” neutrophil count, we also successfully mimicked the neutropenic situation by applying reduced neutrophil numbers in the IAC model and found frequent hyphal growth along with increased germination and decreased cytokine levels which underscores the model’s suitability for infection research.

In our model, neutrophils successfully migrated from the blood side of the IAC model to the infected lung side to interact with A. fumigatus at various stages, i.e. phagocytosis of conidia and germlings, or swarming against hyphae. Upon infection, both macrophages and neutrophils, migrated to the lung side, started to phagocytose conidia, with the majority of conidia being internalized within 4 h after addition of neutrophils, and all conidia being completely taken up after 8 h. Live cell imaging indicated that neutrophils migrated from the blood side to the site of infection within minutes after their addition which might be explained by a possible gradient of signals that macrophages already established to start an immune reaction upon phagocytosis prior to neutrophil addition. However, the intrusion of neutrophils was observed even in the absence of A. fumigatus and the level of increase of immigration upon infection was donor-dependent. Whether this indicates a routine “surveillance” of alveolar tissue by neutrophils rather than an exclusive migration based on distinct signals from macrophages that first encounter conidia, remains to be investigated. Importantly, we were able to measure the speed of both macrophages and neutrophils moving in the system. This is important data for bioinformatic models of human alveolar defense against inhaled fungi, as it has been shown that the random movement of immune cells would be too inefficient and slow to encounter fungal spores before they germinate (Pollmacher and Figge 2014, 2015, Blickensdorf et al. 2019, 2020, Saffer et al. 2023, 2024).

We also observed several other cooperative events of neutrophils and/or macrophages in live cell imaging, such as metaforosis and efferocytosis, which have already been described. Metaforosis, i.e. the lateral transfer of conidia from a dying macrophage to a healthy one, is a regular event in controlling fungal growth (Armstrong-James et al. 2018, Camilli et al. 2021). It has already been observed in in vitro infection assays of human monocyte-derived macrophages and human primary alveolar macrophages as well as in an in vivo zebrafish model (Shah et al. 2016). Importantly, metaforosis is calcineurin-dependent which has direct implications for patients who are immunosuppressed by calcineurin-inhibitors such as cyclosporin A, because macrophages capable of transmitting conidia are more successful to control fungal germination and thus fungal growth (Shah et al. 2016, Armstrong-James et al. 2018).

Especially in scenarios of neutropenia, other immune cells may become relevant for the host's remaining defense against A. fumigatus. Promising candidates with a record of antifungal activity are the innate-like unconventional lymphocytes such as NK and MAIT cells. Here, in our human-derived IAC model we confirmed the previously described antifungal effect of NK cells in in vitro and murine experiments (Schmidt et al. 2017). NK cells account for 5%–20% of peripheral lymphocytes, and a study reported 43–768 NK cells per µl blood with a median of 180 NK cells per µl in healthy adults (Merkt et al. 2021). Thus, we perfused a comparable number (200 000 NK cells in 1.1 ml medium equals 181 cells per µl) into the endothelial compartment of the IAC model 5 h before starting the infection with A. fumigatus conidia. Besides the peripheral blood, NK cells are also detectable in the lung parenchyma but were found to be hypofunctional towards target cell stimulation (Marquardt et al. 2017). For this reason, we decided against using NK cells on the epithelial side of the IAC model.

After seeding of conidia onto the epithelial cell layer, hyphal growth already started after around 4–6 h during a total infection period of 14 h. Consequently, NK cells, while unable to recognize conidia, were still able to sense fungal growth, migrated from the blood side to the side of infection and mounted a response against germlings or hyphae (Schmidt et al. 2011). As expected, fungal growth was further impaired in the presence of NK cells but interestingly the branching behavior of A. fumigatus hyphae was not influenced. The antifungal effect of NK cells is likely caused by the release of cytotoxic molecules such as granzyme A and granulysin, as we observed elevated levels on the endothelial side in accordance with Schmidt et. al (Schmidt et al. 2011). However, IFN-γ was not released in our setting, which may also mediate antifungal activity as observed in another in vitro study (Bouzani et al. 2011). A trend towards elevated levels of cytotoxic molecules was also observed on the epithelial side, but due to the air-liquid-interface at the lung side, the generation of supernatants with reliable protein concentrations was challenging.

In contrast to NK cells, the antifungal properties of MAIT cells are unknown and they need an additional antigen-presenting cell, e.g. macrophages, to be able to respond to pathogens. Therefore, we decided to start with an in vitro 2D infection assay and not directly investigated MAIT cells in the IAC model. In the in vitro setting, MAIT cells were able to reduce the hyphal length of wild-type strains CEA10 and AfS77, but as expected did not influence the riboflavin-deficient mutant ΔriboB. This confirms our previous observations that MAIT cells are activated in a T cell receptor-dependent manner by fungal riboflavin metabolites presented by an antigen-presenting cell (Jahreis et al. 2018).

Although a trend towards impairment of hyphal growth was visible when MAIT cells were present in the IAC model, this effect was not statistically significant. The blood of young Caucasian adults contains 10–183 MAIT cells per µl blood with a median of 71 MAIT cells per µl (Singh et al. 2022). To ensure comparability we decided to perfuse similar amounts of MAIT and NK cells to the blood side of the IAC model which correlates with high but still physiological MAIT cells numbers of 181 cells per µl. There is no information available about absolute MAIT cells numbers in the lung, thus we seeded half as many MAIT cells on the epithelial side as were perfused on the blood side. However, when applying higher concentration of MAIT cells to the lung side, these cells numbers could significantly reduce both the filament length and the branching behavior of the hyphae. This emphasizes the importance of using more complex models in addition to classical in vitro cell culture systems, as cellular interactions might influence the strength of an effect. The significant effect of higher cell numbers also implies that MAIT cells exert an antifungal effect although with less potency compared to NK cells. The actual number of MAIT cells in infected tissue is unknown but it is suggested that MAIT cells are recruited from the blood to the side of infection as a decreased MAIT cell count in the blood is reported for many infectious diseases like tuberculosis, influenza or septic shock. In addition, accumulation of MAIT cells in infected tissues was found e.g. in a mouse model with Salmonella infection (Salou et al. 2017).

Furthermore, the efficacy of MAIT cells depends on the capability of macrophages to sense, phagocytose and present riboflavin metabolites from pathogens (Tastan et al. 2018, Deng et al. 2025), while NK cells can directly recognize fungal pathogens. How MAIT cells inhibit fungal growth is not yet clear. Although they are equipped with cytotoxic granules, we could detect only slightly elevated levels of granzymes or granulysin in both the 2D infection assay as well as in the IAC model. For some donors, elevated IFN-γ levels were observed in the presence of MAIT cells in both settings, which could increase the fungicidal activity of macrophages, as reported in in vitro experiments for Candida (Marodi et al. 1993). Recently, a role of MAIT cells in tissue repair was discussed, e.g. in the skin and lung (du Halgouet et al. 2023, Zhang et al. 2025). Protecting the integrity of the epithelial barrier may also impede the growth of A. fumigatus hyphae.

Taken together, the advanced IAC model confirmed a strong antifungal effect of neutrophils, while the unconventional lymphocytes NK and MAIT cells also showed a measurable, albeit less pronounced effect against A. fumigatus. This finding is consistent with the fact that the latter two cell types are not professional phagocytes. Nevertheless, their antifungal activity may become important in immunosuppressed patients who miss functional neutrophils and/or macrophages. The model allows us to obtain quantitative data of hyphal morphology and of immune cells in response to fungal infections. This data is essential for the setup of bioinformatic infection simulations and for the global aim of improving infection prevention and patient therapy.

The IAC model can also be used to test new antifungal drugs not only regarding their efficacy but also regarding their toxicity against tissue and immune cells as shown before (Hoang et al. 2022). For a better understanding of the situation in patients prone to fungal infections, the IAC model can be modified by using immune cells from patients. A further step could be the testing of chimeric antigen receptor (CAR)-NK cells in the IAC model to analyze adoptive NK cell therapy options in a human cell-based environment. The potential of (CAR)-NK cells in relation to direct cytotoxicity against A. fumigatus has so far been mainly investigated in murine infection models (Morrison et al. 2003, Park et al. 2009, Soe et al. 2020, Seif et al. 2022).

Material and methods

Fungal strains

Aspergillus fumigatus wild-type (CEA10) and GnoA-eGFP mutant (Lapp et al. 2014), which is isogenic to CEA10, were cultured on malt agar plates for 5–7 days at 37°C. The ΔriboB mutant and the corresponding wild-type AfS77 were kindly provided by Hubertus Haas (Medical University of Innsbruck, Austria) grown on Aspergillus minimal medium, for ΔriboB supplemented with 100 g l⁻¹ riboflavin as described before (Dietl et al. 2018). Fungal conidia were harvested one day prior experiments by using sterile PBS with 0.01% (v/v) Tween20 and filtered through a 30 μm cell strainer to obtain a single-fungal spore suspension. If indicated, conidia were pre-swollen in RPMI 1640 + 10% (v/v) FCS for 1 h at 37°C in a rotator, washed three times with PBS with 0.01% (v/v) Tween20 and filtered through a 30 µm strainer. For fluorescent labeling, the conidia were incubated with 200 µg ml⁻¹ FITC (Sigma-Aldrich, Germany) at a final concentration of 10 µg ml⁻¹ in Na₂CO₃ (Merck, Germany) for 20 min at 37°C in a rotator, then washed three times in PBS with 0.01% (v/v) Tween20, centrifuged for 5 min at 2600 × g and stored in PBS at 4°C overnight.

Isolation of primary human blood cells

Fresh peripheral blood was taken from healthy donors after written and informed consent in accordance to University Hospital Jena ethics committee approval 2020–1965-Material and with the Declaration of Helsinki 1975, as revised in 2008. Serum was collected in separate vials and obtained by centrifugation.

Peripheral blood mononuclear cells (PBMC) were isolated by standard density gradient centrifugation and used for macrophage differentiation and isolation of lymphocytic cells. CD8+ cells were isolated by magnetic-activated cell sorting using human CD8 MicroBeads (Miltenyi Biotec, Germany) from PBMC and further separated on a fluorescence activated cell sorter into the MAIT cell fraction being CD3+CD8+CD161+ (with an average viability of 99% and 90% being TCRVa7.2+). Remaining CD8-depleted PBMC were used for NK cell separation using NK cell isolation kit (Miltenyi Biotec, Germany) with an average viability of 98% and a purity of 92% CD3- CD56+ cells. Lymphocytic cells were cultured in RPMI 1640 (ThermoFisher Scientific, Germany) supplemented with 10% (v/v) FCS (Biowest, France) and 50 U ml⁻¹ recombinant human IL-2 (PeproTech^TM, Germany).

Neutrophilic granulocytes were isolated freshly on the day of experiment from the same donor who provided the monocytes/macrophages using density gradient centrifugation by layering freshly drawn EDTA blood (mixed with PBS without Ca²⁺, Mg²⁺ in a 1:1 ratio) over Polymorphprep solution (Serumwerk Bernburg, Germany) and centrifugation at 500 × g for 35 min at room temperature (RT) without break. Neutrophils from the resulting ring were transferred to PBS (without Ca²⁺, Mg²⁺, Gibco, UK) and washed once, followed by erythrocyte lysis (15.5 mM NH₄Cl, 1 mM KHCO₃, 0.01 mM EDTA (all Merck, Germany) in sterile water) for 1 min and washed again in PBS without Ca²⁺, Mg²⁺. Prior to transferring neutrophils into the biochip, cells were labelled for tracking with Cytopainter Green© (CPG, Abcam, UK) according to the manufacturer’s instructions in RPMI 1640 + 10% (v/v) FCS + 100 U Penicillin/Streptomycin (P/S, ThermoFisher Scientific, Germany) + CPG solution (1:1000) for 10 min at RT in the dark.

Advanced invasive aspergillosis-on-chip model

Assembly of the basic IAC model

The assembly of lung-on-chip models has already been described (Deinhardt-Emmer et al. 2020, Hoang et al. 2022). Briefly, chips with a polyethylene-terephthalate membrane with 8 µm pores were purchased from Dynamic42 GmbH, Germany, and seeded on the upper side with human umbilical vein endothelial cells (obtained from the Institute for Molecular Cell Biology, University of Jena, Jena) in endothelial cell growth medium + supplements (EC+, all Promocell, Germany) + 100 U P/S constituting the endothelial side. Two days later, NCI-H441 cells (ATCC, USA) were seeded on the opposite (lower) side of the membrane in RPMI 1640 (ThermoFisher Scientific, Germany) + 10% (v/v) FCS (Biowest, France) + 100 U P/S + 1 µM dexamethasone (Sigma-Aldrich/Merck, Germany, starting on day 6) generating the epithelial side. On day 7, human monocytes obtained by PBMC isolation and subsequent differentiation in X-VIVO 15 media (Lonza, Switzerland), supplemented with 10 ng ml⁻¹ M-CSF + 10 ng ml⁻¹ GM-CSF (both PeproTech, Germany) + 100 U P/S for two days were introduced onto the H441 cells and afterwards co-cultured in RPMI 1640 + 20% (v/v) autologous serum from the macrophage donor and 10 ng ml⁻¹ GM-CSF (PeproTech^TM, Germany), 1 µM dexamethasone and 100 U P/S. On day 8, perfusion of the endothelial side was started using Reglo ICC peristaltic pump (Ismatec, USA) to supply EC+ medium from a reservoir at a flow rate of 21 µl min⁻¹. On day 13, medium was removed from the epithelial side to generate an air-liquid-interface. In parallel, perfusion was continued with endothelial cell growth medium with 20% (v/v) autologous serum, 10 ng ml⁻¹ GM-CSF and 100 U P/S. All media were changed daily until day 14, when chips were used for experiments.

Addition of neutrophils

In order to further optimize the existing model in line with the host-specific complexity, prior to infection, 2 × 10⁶ CPG-labelled neutrophils resuspended in infection medium (RPMI 1640 + 5% (v/v) autologous serum + 10 ng ml⁻¹ GM-CSF + 100 U P/S) were added to the reservoirs attached to the endothelial side of the lung-on-chip models and perfused for 4 h to determine effects of perfusion on neutrophil viability and/or activation. After the initial perfusion time, the cell suspension was collected from the reservoirs and replenished by fresh 2 × 10⁶ CPG-labelled neutrophils in infection medium for infection and subsequent overnight incubation.

Addition of MAIT cells

By contrast, 1 × 10⁵ MAIT cells were seeded either together with 1 × 10⁵ macrophages and cultured for another week in the presence of 50 U ml⁻¹ recombinant human IL-2 (PeproTech^TM, Germany) to allow integration into the tissue on the epithelial side of the biochip or 2 × 10⁵ CPG-prelabelled MAIT cells were perfused to the blood side for 5 h before infection as described above.

Addition of NK cells

On the day of infection, 2 × 10⁵ CPG-prelabelled NK cells were perfused to the blood side for 5 h before infection as described above.

Infection

Towards the end of the perfusion period of either neutrophils or lymphocytes, the epithelial side of the chip was flushed with infection medium and incubated for additional 15 min to extract cytokines and migrating neutrophils. The epithelial side was then infected with 1 × 10⁵ resting FITC-labelled A. fumigatus conidia and incubated upside-down for 90 min without perfusion to allow the attachment of fungal spores to the lung epithelium. After this incubation, the epithelial side was rinsed with fresh infection medium to remove nonattached conidia and the air-liquid-interface was re-established. The infected lung-on-chip models were reconnected to the pump for constant flow and incubated overnight for 14 h at 37°C and 5% CO₂. On the next day, media from the endothelial side and reservoirs were collected. The epithelial side of the chip was flushed with fresh medium (RPMI 1640 + 5% (v/v) autologous serum) and incubated for 15 min to extract cytokines and migrated cells from this compartment. All supernatants were immediately stored at −80°C until further analysis.

Live cell imaging

For optimization of imaging conditions during live cell imaging the workflow for the lung-on-chip setup as well as infection conditions were adjusted in the following way. First, macrophages were differentiated for 5 days (instead of 2 days) in X-VIVO 15 media (Lonza, Switzerland) supplemented with 10 ng ml⁻¹ M-CSF, 10 ng ml⁻¹ GM-CSF (both PeproTech^TM, ThermoFisher, Germany), 100 U P/S with media changes every other day. On day 10 of lung-on-chip assembly, macrophages were detached by a mix of 4 mg ml⁻¹ lidocaine (Sigma-Aldrich, Germany) and 5 mM EDTA in PBS without Ca²⁺, Mg²⁺ for 7 min at 37°C and labelled afterwards using 1 µM CellTracker^TM Orange (Invitrogen, USA) in X-VIVO medium for 30 min at 37°C. Labelled macrophages were seeded at a concentration of 0.2 × 10⁶ cells on the alveolar epithelium of each model and incubated in RPMI 1640 + 20% (v/v) autologous serum from the macrophage donor with 10 ng ml⁻¹ GM-CSF (PeproTech^TM, Germany), 1 µM dexamethasone and 100 U P/S. Perfusion was initiated the next day and the air-liquid-interface introduced on day 13. Green fluorescent A. fumigatus GnoA-eGFP conidia were swollen for 1 h before infection and 0.5 × 10⁶ conidia were added to the epithelial side for 90 min upside down. Before imaging, conidia suspension was removed, and fresh media was introduced into the lung chamber. The chip was then transferred to a sterile bag. Neutrophils were labelled with CytoPainter Cell Proliferation Staining -Deep Red (CPDR, Abcam, UK) at a 1:100 dilution in DPBS for 20 min at 37°C. Then, 1 × 10⁶ cells in 700 µl of media per model were stored in a reservoir in the same sterile bag as the biochip. After relocation to a Spinning Disk confocal laser scanning microscope (Carl Zeiss, Germany), the bag was manipulated from outside to connect the reservoir to the vascular chamber thus allowing entrance of neutrophils under gravity to the models. The chip inlets were closed and the chip was immediately mounted on the microscope to allow imaging of cell—conidia interaction without delay. Imaging was done over up to 20 h under static conditions at 37°C and 5% (v/v) CO₂ in the incubating chamber of Spinning Disk confocal laser scanning microscope using an EC Plan-Neofluar 10x NA0.3 objective with an aperture of 0.18 NA. Live-cell imaging was set up z-stacking with an axial increment of 1 µm and time series of 3 min per z-stack (1.5 min of imaging and 1.5 min of rest) were applied to track the movements of neutrophils and macrophages on the epithelium simultaneously. The pores were visualized using Zeiss microscope’s reflected light settings with laser illumination at 594 nm and epifluorescence detection at 570–605 nm.

2D infection assay

5 × 10⁵ CD8-depleted PBMC were seeded per well in 8 well chamber slides (ibidi, Germany) to differentiate monocytes into macrophages. Total 1.5 h after seeding non-adhering cells were removed by washing twice with PBS and cells were differentiated in X-VIVO+ medium, which consists of X-VIVO medium (Lonza, Switzerland) supplemented with 10% (v/v) autologous serum, 10 ng ml⁻¹ recombinant human GM-CSF, 10 ng ml⁻¹ recombinant human M-CSF (both PeproTech^TM, Germany) and 1% (w/v) P/S. After differentiation for seven days, 1 × 10⁵ MAIT cells pre-labelled with Cytopainter Green were added. The following day, the mixed cell population was infected with 5 × 10⁵ FITC-prelabelled resting A. fumigatus conidia per well for 8 h.

Immunofluorescence

For the 2D infection assay slides were washed twice with DPBS+ and fixed with 4% (v/v) formaldehyde in PBS for 10 min at RT. Hyphal staining was achieved by adding 100 µg ml⁻¹ calcofluor white (Sigma Aldrich, Germany) for 20 min at RT in the dark. After permeabilization with 100% methanol for 10 min at −20°C and blocking with DPBS + 3% (v/v) goat serum (Dianova, Germany) for 1 h at RT macrophages were stained with CD68 primary antibody (rabbit-anti-human, clone D4B9C, 1:800, Cell Signaling Technologies, Netherlands) at 4°C overnight. On the next day, Cy5 goat-anti-rabbit antibody (1:100, Abcam, United Kingdom) was applied for 45 min at RT, all wells were washed with DPBS+ and stored in 200 µl DPBS+ with 1% (v/v) P/S at 4°C until imaging.

The IAC model was stained as follows: after washing, chambers of the biochip were incubated with 4% (v/v) formaldehyde for 10 min for fixation. Membranes were removed and blocked with DPBS + 0.1% (w/v) Saponin (Sigma–Aldrich, Germany) + 3% (v/v) goat serum (Dianova, Germany) for 1 h at RT. For immunofluorescent staining the following primary antibodies were diluted in DPBS + 0.1% (w/v) Saponin + 0.3% (v/v) goat serum and incubated overnight at 4°C: CD68 (rabbit-anti-human, clone D4B9C, 1:800, Cell Signaling Technologies, Netherlands), E-Cadherin (mouse-anti-human, clone 36, 1:100, BD Biosciences, United States), von Willebrand factor (rabbit-anti-human, polyclonal, 1:200, DAKO, Germany). The next day, secondary goat anti-mouse or goat-anti-rabbit antibodies Cy3 or Cy5 (1:100–200, Abcam, UK and ThermoFisher Scientific, Germany) were incubated in DPBS + 0.1% (w/v) Saponin + 0.3% (v/v) goat serum together with 0.5 μg ml⁻¹ Hoechst 33 258 (ThermoFisher Scientific, Germany) for 45 min at RT. Fungal hyphae were stained with 2 μg ml⁻¹ calcofluor white for 7 min at RT. After washing, the membranes were mounted on glass slides with Roti®-Mount FluorCare (Carl Roth, Germany).

Imaging and data analysis

Ibidi® slides and IAC model membranes were imaged using a confocal laser scanning microscope LSM 780 (Carl Zeiss, Germany) with a 63x NA1.4 oil immersion objective using the ZEISS Efficient Navigation Black 2.3 software (Carl Zeiss, Germany). For the 2D infection assay four z-stack images with 2 µm of axial increments were acquired per condition and the hyphal length was determined using Imaris (Imaris version 10.2, Imaris, Andor Technology Ltd., Belfast) from at least 15 hyphae per condition.

Fungal growth in the biochip was analyzed as earlier described in detail (Hoang et al. 2022). In brief, z-stacks of at least 5 hyphae per condition were acquired of membranes stained only with calcofluor white. To capture the complete hyphae, up to 2 by 2 tile scans (one tile equals 225 µm x 225 µm) were taken depending on the planar hyphal growth, and z-stacks with increments of 0.5 µm–1.4 µm between lateral image layers were obtained. The axial range extended from the uppermost point of the hyphae on the epithelial side to the lowest point across the membrane on the endothelial side. Reflected light (excitation at 594 nm, detection at 570–605 nm) was used to visualize pores of the biochip membrane. The imaged hyphae were reconstructed using Imaris and corrected manually where necessary. First, FITC-labelled conidia were detected using the Spots objects of Imaris and then set as the starting points of the hyphae. Afterwards, the hyphal membranes were reconstructed as Surfaces objects and the semiautomated manual Filament tracer algorithm was used to reconstruct the hyphae. If necessary, a denoising algorithm was applied in advance. After reconstruction, the hyphal length in µm and number of hyphal branches were exported from Imaris. Hyphae without starting point were not reconstructed. To determine the number of internalized conidia, the spores were again segmented into identical 3D surfaces using the Spots tool in Imaris. The 3D surfaces were then used to create a mask: voxels outside the Spots objects (one object per conidium) were set to zero, whereas voxels inside the objects were set to a constant value, e.g. 10. The newly created channel served as an internalized conidia counter: the number of internalized conidia per cell were calculated by measuring the integrated intensity of the masked channel inside each macrophage and neutrophil, and dividing it by the sum of the voxel intensities inside an individual Spots object.

Flow cytometry analysis

Cell suspensions from both chambers of the IAC model were centrifuged at 200 × g for 5 min to obtain cell pellets for flow cytometry. Cells were washed once in PBS (without Ca²⁺ and Mg²⁺) + 2 mM EDTA (PE) and incubated with propidium iodide dye (Biolegend, USA) for 1 min to exclude dead cells from analysis and determine cell viability. Antibody mixes in PE were applied for 15 min at RT in the dark: CD15-FITC (clone HI98), CD16-APC/Cy (clone 3G8; both from Biolegend, USA) and CD45-BUV395 (clone HI30), CD62L-PE (clone DREG-56; both from BD Biosciences, USA). Cells were washed again in PE and analyzed in LSR Fortessa flow cytometer (BD Bioscienes, USA) using FACSDiva software. Neutrophils were gated based on their CPG signal and CD45 expression. Control samples were stained with isotype antibodies corresponding to the marker antibodies.

Cytokine analysis

Supernatants were collected from the endothelial and the epithelial side of IAC models as well as the chambers of the 2D infection assay after overnight infection and assayed with bead-based immunoassays (LEGENDplex™ Human CD8/NK Panel, Type 1/2/3 Interferon Panel, customized assay for TNF-α, IL-1β, IL-6, IL-8, all Biolegend, United States) in V-bottom 96-well plates according to the manufacturer’s instructions. Results were acquired on LSR Fortessa (BD Biosciences, USA).

Statistical analysis

Statistical analysis was performed in GraphPad Prism 10 using Mann-Whitney test to compare two groups and ordinary ANOVA test with Tukey’s multiple comparisons post-hoc test to compare more than two groups. P < 0.05 (*), p < 0.01 (**) and p < 0.001 (***) were considered significant.

Conflicts of interest

The authors declare no scientific or financial conflict of interest.

Funding

This work was supported by the German Research Foundation within the Collaborative Research Centre/Transregio 124–FungiNet (DFG project no. 210879364) with support to A.A.B. and M.v.L.T. (project A1), F.S. and A.A.B. (project A6) as well as M.T.F. (project B4). S.J., S.H., and M.v.L.T. are funded by Ruhr University Bochum (IF-037-24); S.J. was further funded by the Interdisciplinary Center of Clinical Research of the Medical Faculty Jena (AMSP 08). A.K.R. was supported by the Honours Programme for Future Researchers, Graduate Academy, Friedrich Schiller University Jena, funded through the Excellence Strategy of the German Federal and State Governments. C.Z. and M.T.F. were supported by the Collaborative Research Center 1278 “PolyTarget” (project number 316213987; project Z01). A.A.B. and M.T.F. further acknowledge financial support by the Cluster of Excellence “Balance of the Microverse” under Germany’s Excellence Strategy—EXC 2051–Project-ID 690 390713860. H.H. is funded by Austrian Science Fund (FWF) [Grant-DOI: 10.55776/DOC82].