Cryptococcus exploits delayed microglial activation, and microglial osteopontin (OPN/Spp1) impairs peripheral host control

SUMMARY

Cryptococcus, a neurotropic fungus classified as a critical-priority pathogen by the World Health Organization (WHO), causes cryptococcal meningoencephalitis (CM), the second leading cause of death in HIV/AIDS patients. Despite its clinical importance, host brain responses during CM remain poorly understood. In a mouse systemic infection model, Cryptococcus infiltrates the brain within a day. However, full activation of microglia and recruitment of leukocytes takes 14 days, a delay not observed in brain infections caused by Candida albicans. Microglia exhibit limited ability to directly detect Cryptococcus, and their activation depends on IFNγ from Th1 cells. Therefore, adaptive immunity (Th1 responses) precedes innate immune responses (microglial activation) in the brain during CM. Moreover, microglia-derived osteopontin (OPN/Spp1) exacerbates CM by altering peripheral immunity and increasing fungal loads in peripheral organs. These findings reveal a uniquely slow host cellular response to Cryptococcus brain infiltration, allowing the fungus an extended window to establish the infection.

Graphical Abstract

INTRODUCTION

CNS mycoses have increased, but clinical interventions remain exceptionally scarce, underscoring the urgent need to understand host-pathogen interaction mechanisms to develop novel treatments. Among CNS mycoses, CM caused by Cryptococcus neoformans, a fungal pathogen categorized as a critical priority by the WHO¹, remains a significant cause of morbidity and mortality, ranking as the second most common cause of death in HIV/AIDS patients². Although this poses a significant public health concern, our understanding of host brain cell responses during CM remains incomplete and limited.

CM is caused by an uncontrolled Cryptococcus pulmonary infection that spreads to the brain. Previous reports on CM have mainly highlighted the roles of leukocytes infiltrating the brain^3–7. Fewer studies reported brain-resident cells during Cryptococcal brain infection. Microglia provide survival niches for Cryptococcus, and their absence did not reduce brain fungal burden in acute CM⁸. We also showed that astrocytes, the most abundant glial cell type in the CNS, exhibit dysregulated aquaporin-4 levels that lead to lethal brain edema in a model of cryptococcal immune reconstitution inflammatory syndrome (C-IRIS)⁷.

While antifungal therapies are available, the rising threat of drug resistance and persistently high mortality associated with CM highlight the urgent need for improved diagnostics and treatment strategies⁹. The Olszewski lab laid foundational work in studying Cryptococcus in the brain^4,10, but the precise host responses during CNS infection remain poorly defined¹¹. A recent study suggested that microglia are not essential for initial resistance to acute Cryptococcus infection¹², but this also does not preclude their contribution to CM pathogenesis. In contrast, the mechanisms by which Cryptococcus enters the brain have been more extensively characterized. The fungus crosses the blood-brain barrier (BBB) via three main routes¹³: (1) the “Trojan Horse” mechanism, where it exploits phagocytes such as Ly6C^lo monocytes and neutrophils to gain CNS entry^14–16; (2) transcytosis, involving host cell uptake and release across brain endothelial cells, facilitated by hyaluronic acid–CD44 interactions¹⁷; and (3) paracellular traversal through disruption of tight junctions or endothelial damage^18–20. However, how CNS-resident cells respond to brain Cryptococcus infection after the fungal brain dissemination is still largely elusive.

Single-cell RNA sequencing (scRNAseq) studies have identified multiple microglial subsets, including disease-associated microglia (DAM)²¹, axonal tract microglia (ATM)²², proliferative-region-associated microglia (PAM)²³, CD11c⁺ microglia²⁴, and ‘microglia during acute demyelination²⁵.’ While DAM, CD11c⁺ microglia, and ‘microglia during acute demyelination’ were observed in neurodegenerative diseases^21–25, ATM and PAM were noted during microglia development^22,23. Notably, Spp1, encoding osteopontin (OPN), is consistently upregulated across these diverse subsets²⁶. OPN is broadly expressed in leukocytes under inflammatory conditions and exists in two translational isoforms with distinct roles: secreted OPN (sOPN), a well-studied extracellular ligand for CD44 and integrins, and intracellular OPN (iOPN), which functions as a cytoplasmic scaffold involved in immune signaling^27–35. Although OPN’s functions are highly context-dependent—ranging from protective to pathogenic—its roles in fungal infection and neuroimmunology remain underexplored.

In this study, we report markedly delayed host cellular responses in the brain, likely providing Cryptococcus with an extended window of opportunity to establish a brain infection by primarily using the Cryptococcus 52D strain (ATCC 24067; serotype D), because the commonly used H99 strain caused rapid mortality in mice, with potential confounding factors related to the acuteness of infection. The 52D strain was originally isolated from a human patient and accounts for approximately 5–8% of clinical Cryptococcus infections in Europe^36,37, and also has the advantage of producing a more gradual disease course in mice and modeling clinical conditions, particularly in the context of clinical CM. We found that even 52D Cryptococcus infiltrated the brain from circulation within a day, but microglia require significant time —up to two weeks— to fully respond to Cryptococcus infiltration, in stark contrast to their prompt response during Candida albicans (Ca) brain infection. Similarly, the full recruitment of leukocytes from circulation takes up to two weeks. The slow response of microglia is attributed to their dependency on IFNγ production by T cells recruited to the brain. We further found that OPN produced by microglia is detrimental to hosts by modulating peripheral immune responses.

RESULTS

Chronic CM model to study temporal dynamics of host responses in the brain

A limited number of studies have elucidated the role of CNS-resident cells during CM. We assessed the survival of C57BL/6 (B6) WT mice retro-orbitally (r.o.) infected with Cryptococcus neoformans H99 (serotype A) or Cryptococcus deneoformans 52D (serotype D), recently subclassified from C. neoformans38. H99 (10⁶ yeasts/mouse) is highly virulent, with no survival beyond 6-dpi. (Figures 1A, S1A). Even with a 1000-fold lower inoculum of H99 (10³ yeasts/mouse), all mice reached a humane endpoint by 17-dpi (Figures 1A, S1A). In contrast, most mice infected with 52D (10⁶ yeasts/mouse) survived at 28-dpi (Figures 1A, S1A) with a slight but significant increase in brain weight starting at 7-dpi (Figure S1B), indicating CM development. This increase in brain weight was also observed in older (16-week-old) mice following infection (Figure S1C), suggesting that the brain weight result is not specific to young mice. Importantly, Cryptococcus 52D infiltrated the brain within a day after infection with fungal loads increasing and plateauing after 14-dpi (Figure 1B). Cryptococcal lesions appeared throughout the brain parenchyma without region specificity (Figure 1C), and lesion sizes increased over time (Figure 1D, E). Despite numerous lesions, mice showed no overt abnormalities throughout the 28-day infection period (data not shown). These findings establish Cryptococcus 52D as a model of CM with minimal mortality over 4 weeks, mimicking the chronic progression typically seen in human CM. For subsequent studies of host cell responses during CM, we use the 52D strain to avoid the rapid mortality observed with H99, unless otherwise specified.

Figure 1. Cryptococcus infiltrates the brain and establishes CM.

A. Kaplan-Meier curve denoting survival of mice retro-orbitally (r.o.) infected with H99 (10³ yeast/mouse, n=6 mice, green square), H99 (10⁶ yeast/mouse, n=8 mice, purple triangle), and 52D (10⁶ yeast/mouse, n=21 mice, gray circle).

B. Brain fungal burdens in 52D-infected mice (10⁶ yeast/mouse), n=3–14 mice/group.

C-E. Representative images of 52D-infected brains (10⁶ yeast/mouse). Segmented lines denote edges of brain tissues, scale bar = 2 mm (C); inset box of cerebral cortex in C, scale bar = 200 μm (D). Parenchymal cryptococcal lesions; segmented lines denote the outline of a fungal lesion, scale bar = 100 μm (E). GXM: Glucuronoxylomannan (capsule component of Cryptococcus 52D, yellow), MBP: Myelin Basic Protein (blue).

F. CD45^hi cell counts at indicated time points, n=3–10 mice/group. Data combined from three independent experiments.

G. Counts of CD45^hi cells at 4-dpi with Cryptococcus 52D (10⁶ yeast/mouse, grey circles), H99 (10⁶ yeast/mouse, green squares) and C. albicans (10⁵ yeast/mouse, red circles).

Data presented as mean ± SEM. Statistical tests: One-way ANOVA with Dunnett’s multiple comparison test; ns, not significant; ****, p≤0.0001 (F, G).

See also Figures S1, S2.

Delayed leukocyte infiltration in the brain

To understand the development of the brain immune landscape during CM, we evaluated leukocyte infiltration in the brain. Mice were r.o.-infected with 52D, and brain cellularity of CD45^hi cells was examined at several timepoints by flow cytometry (Figures 1F, S1D). Cell counts of neutrophils, dendritic cells (DCs), monocytes/macrophages (mono/macs), CD4⁺ T cells, and CD8⁺ T cells did not show a statistically significant increase until 14-dpi (Figure S1E–I). B cells showed constant cell counts throughout the observed 28-day infection period (Figure S1J). We also compared cell numbers in the brains of H99-infected mice (10⁶ yeast/mouse). Despite the acute infection with H99 (Figure 1A), neutrophils, typically among the first recruited cells, did not increase in total numbers at 4-dpi (Figure S1K), two days before all mice reached a humane endpoint.

To investigate if delayed immune cell recruitment was Cryptococcus-specific, we r.o.-infected mice with Ca (10⁵ yeast cells, strain SC5314), also known to be neurotropic^39,40. Unlike Cryptococcus, Ca infection showed a significant increase in CD45^hi leukocyte recruitment to the brain by 4-dpi (Figures 1G, S1L–N). Specifically, neutrophils, mono/macs, and DCs were increased at this timepoint (Figure S1O–Q), while CD4⁺ and CD8⁺ T cell numbers were increased but to a lesser extent (Figure S1R, S). These results highlight effective leukocyte recruitment during Ca infection, contrasting with the delayed response in Cryptococcus infection.

During the brain harvest, mice were not perfused, and brain border tissues, including the meninges and choroid plexus, were not removed, as precise separation from CNS tissue was technically unfeasible in this context. Thus, phycoerythrin (PE)-conjugated anti-CD45 antibody was r.o.-injected to mice to stain circulating leukocytes (PE⁺), whereas leukocytes in brain parenchyma are not stained (PE⁻). By 14-dpi, a significant increase was found in the PE⁻ compartment for total lymphocytes, NK cells and total T cells (Figure S1T–V), while mononuclear phagocytes (MNPs) showed an increase in the PE⁺ compartment with a trend increase in the PE⁻ compartment (Figure S1W).

Our data suggest that Cryptococcus rapidly disseminates into the brain parenchyma within one day (Figure 1B), while leukocyte recruitment takes over 10 days. In contrast, leukocyte infiltration occurred much earlier in Ca infection at 4-dpi, pointing to a Cryptococcus-specific infection mechanism that delays host responses.

Early peripheral organ immune responses

In the 52D r.o. infection model, fungal burdens at 7-dpi in brains were significantly higher than those in lungs and spleens (Figure S2A). To assess the cellularity of these organs, we quantified immune cells using flow cytometry (gating strategies in Figure S2B, C). At 7-dpi, lungs and spleens exhibited a generally increased number of various myeloid cell types, except for monocytes (Figure S2D, E). In contrast, lymphocyte counts, including B, T, and NK cells, remained unchanged in both lung and spleen at 7-dpi (Figure S2F, G), suggesting that additional time is required for adaptive immune responses within these organs. Thus, peripheral organs exhibit increased numbers of myeloid cells at 7-dpi, contrasting with the delayed immune cell infiltration observed in the brain.

Delayed microglia activation in Cryptococcus infections but not in Ca

Microglia, the CNS-resident macrophage-like cell, are regarded as sentinel cells, serving as the first line of anti-microbial defense, detecting pathogens⁴¹. As an example of tissue-resident macrophages serving as immune sentinels, alveolar macrophages (AMs) in the lung rapidly respond to Cryptococcus within 4 hours by producing chemoattractants⁴². Therefore, we expected microglia to respond quickly to Cryptococcus in the brain. However, our data revealed otherwise.

Total microglia numbers remained constant throughout infection (Figure S3A), though their proportion decreased over time (Figure S3B) due to leukocyte infiltration (Figure 1F). As activated microglia upregulate CD11c and major histocompatibility complex class II (MHCII) expression⁶, MHCII^hiCD11c⁺ microglia were assessed. Based on these markers, activated microglia appeared at 10-dpi, peaking at 14-dpi (Figures 2A, B, S3C). We evaluated microglia activation during acute H99 Cryptococcus infection (10⁶ yeasts/mouse) at 4-dpi because no mice survived beyond 6-dpi (Figure 1A). The results showed unchanged numbers of microglia and their frequency in total CD45⁺ cells (Figure S3D), with no sign of microglia activation (Figure S3E, F). In contrast, Ca infection activated microglia by 4-dpi (Figure S3E, F).

Figure 2. Delayed microglia phenotype changes to Cryptococcus infection A and B.

Representative flow cytometry plots (A) and frequencies of activated microglia (B) at the indicated time points, n=3–13 mice/group.

C. Representative images of immunofluorescent staining of 52D-infected brains (10⁶ yeast/mouse) at the indicated timepoints. Shown are Cryptococcus (GXM, magenta), astrocytes (GFAP, yellow), and microglia (MG^tdT, cyan) within peri-lesional areas (defined in Figure S3I, scale bar = 200 μm).

D. Microglia density, indicated as proportion of tdTomato⁺ (cyan) areas quantified in 50 μm peri-lesional areas, as demonstrated in Figure S3I. One field includes 2–3 lesions, and 3 fields/mouse were quantified using ImageJ. A single datapoint was obtained from one mouse, n=3 mice/group.

E-J. Morphological assessment of microglia (MG^tdT) localized within 100 μm of lesion edges. Representative images of soma (white), processes (blue), and process endpoints (red) from the brains of mice infected with 52D- (E) or Ca-infected brains (F). Quantification of soma volume (G), soma sphericity (H), number of processes per microglia (I), and processes length sum (J). Analyses used 10–20 microglia/mouse. A single datapoint is obtained from one microglia. n=2 mice/group for 0-dpi, n=4–6 mice/group for other timepoints.

K-M. Gene expression levels in FACS-sorted microglia from 52D-infected mice (10⁶ yeast/mouse) at noted time points, n=3–4 mice/group.

Data presented as mean ± SEM. Data representative of two (K-M) or three (A and B) independent experiments. Statistical tests: One-way ANOVA with Dunnett’s multiple comparison test; ns, not significant; *, p≤0.05; **, p≤0.01; ***, p≤0.001; ****, p≤0.0001 (B, D, G-M).

See also Figures S3, S4 and Table S4.

These results demonstrate that microglia in Cryptococcus-infected brains required 10–14 days for activation despite the presence of the fungus in the brain at 1-dpi (Figure 1B). Conversely, Ca infection elicited microglial activation, reflecting an increase in leukocyte brain infiltration within just 4 days.

Microglia responses lag behind early fungal presence in the brain

Next, we investigated microglia localization in infected brains using Cx3cr1^CreERT2;Rosa26^LSL-tdTomato mice to label bona fide microglia and distinguished them from recruited myeloid cells. This model leverages microglia’s long half-life and unique yolk-sac origin⁴³. Tamoxifen (TAM) treatment followed by a 4–6-week “washout” period replaces TAM-exposed CX3CR1⁺ cells with hematopoietically de novo generated CX3CR1⁺ cells⁴³. We validated microglia-specific tdTomato expression in contrast to the minimal tdTomato signal in other cells in the brain, lung, spleen, and blood (Figure S3G, H). Using this system, we observed tdTomato⁺ microglia (MG^tdT) localization in peri-lesional areas within 50 μm of lesion borders (Figure S3I). MG^tdT were scarce near cryptococcal lesions at 4 and 7-dpi, but significant peri-lesional accumulation was identified at 14 and 21-dpi (Figure 2C, D). These histological analyses focused on cerebral cortex lesions, as this pattern was consistent throughout the brain.

We further investigated the morphology of microglia localized within 100 μm of lesion borders during CM. Representative microglial images for 52D and Ca infections are shown (Figures 2E, F, S3J, K), in which activated microglia are characterized by enlarged, less spherical soma and reduced number and length of processes. Quantitative analyses indicated that changes in these morphological readouts were not evident until 14-dpi (Figure 2G–J), along with values of soma surface areas and numbers of process terminal points and branches (Figure S3L–N). Again, with a stark contrast, microglia during Ca infection required only 4 days to exhibit significant changes in the morphological alterations observed in Cryptococcus-infected mice at 14-dpi (Figure 2E–J, S3J–N). However, transient hyper-ramification observed at 4-dpi with Cryptococcus is worth noting, characterized by increased process lengths and complexity without soma morphological changes (Figure 2I, J, S3L–N). Such hyper-ramification was previously reported under “psychological” stresses^44,45. Thus, microglia may have transiently responded to an unknown signal insufficient to induce full activation.

These findings confirm a delayed microglial response to Cryptococcus infection, with significant morphological changes at 14-dpi, compared to rapid changes within 4 days during Ca infection.

Delayed gene activation of microglia and astrocytes

Microglia gene expression has been extensively studied in neurodegenerative disorders, but data on fungal infections remain scarce. In this study, microglia were isolated at different infection timepoints by fluorescence-activated cell sorting (FACS) (gating strategy: Figure S1D), and gene expression levels were assessed by reverse transcription-quantitative polymerase chain reaction (RT-qPCR). A progressive decline in microglia homeostatic genes, represented with Cx3cr1 and P2ry12, was observed by 21-dpi (Figure 2K), resembling patterns of DAM in a mouse model of Alzheimer’s disease^21,46. Expression of proinflammatory genes, such as Tnf, Il6 (Figure 2L), and Il1b (Figure S3O), peaked at 8 to 10-dpi and then declined. In contrast, Spp1, encoding OPN and a hallmark DAM gene²¹, exhibited continuous upregulation through 21-dpi (Figure 2M).

Astrocyte responses were also analyzed during CM. Significant changes were absent until 14-dpi, when an intense GFAP signal, indicating astrogliosis, became apparent around lesions and persisted through 21-dpi (Figure S4A, B). To evaluate gene expression in astrocytes, ACSA-2⁺ magnetic bead sorting was used to enrich astrocytes⁴⁷, confirmed by astrocyte-specific gene expression (Gfap and Aldh1l1) and the absence of microglia-specific transcripts (Aif1 and Tmem119) (Figure S4C, D). Gfap transcript levels significantly increased at 14 and 21-dpi, with no changes at earlier timepoints (Figure S4E). Similarly, genes associated with astrogliosis and possible neurodegeneration (Cd44, Cxcl10, Cd274)^48,49, significantly increased at 14-dpi, with further elevation through 21-dpi (Figure S4F).

These data reveal that glial cells exhibit no significant gene expression changes until 14-dpi during CM, highlighting delayed reactivity in microglia and astrocytes.

Cryptococcal PAMPs are poorly recognized by microglia and astrocytes

Microglia, equipped with pattern recognition receptors (PRRs) for detecting pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), play crucial roles in CNS immunity¹³. To investigate their direct response to Cryptococcus, we assessed the expression of genes, including but not limited to Tnf, Cxcl2, Il6, and Il1b, at various timepoints (3, 6, 12, and 24 hrs), different MOIs, various sources of microglia and macrophages, with or without heat-killing (HK) fungi. The settings were on tissue culture with host cells without any preconditioning and various fungal strains or PRR ligands as positive controls. We presented detailed results of the expression of Tnf and Cxcl2, as representative genes, because they are often used in evaluating myeloid cell activation due to their rapid and robust induction by PRR stimulation. We also used them as markers to indicate the inflammatory response of alveolar macrophages to Cryptococcus lung infection⁵⁰. Microglia were co-cultured for 6 hours with HK fungal strains (Cryptococcus strains: 52D, H99, cap59Δ; Ca strain, SC5314; all MOI of 5, unless specified) or stimulated with PRR agonists (LPS for TLR4, Pam₂CSK4 for TLR2/TLR6, and curdlan for Dectin-1). Except for slight increases in Cxcl2 expression in primary microglia (CD11b⁺ from naïve B6 WT mice) with cap59Δ and Ca stimulation, other conditions detecting Tnf and/or in microglial cell lines (mouse BV2 and human HMC3) revealed no upregulation (Figure 3A–C). Results from other genes and timepoints were also shown -- Il6, Tnf, and Il1b at 3, 6, 12, and 24 hours (Figure S5A) and Cxcl2 at 3 and 12 hours (Figure S5B)—but Cryptococcus did not increase expression of these genes in BV2. The data suggested a poor response of microglia to Cryptococcus. The results with the cap59Δ strain, which is deficient in capsule formation, were unexpected because we initially considered the capsule, including glucuronoxylomannan (GXM), to suppress microglial activation, thereby contributing to delayed leukocyte infiltration. However, our findings ruled out the capsule or GXM as primary factors underlying the insensitivity of microglia. Moreover, a previous study demonstrated that GXM disrupts BBB integrity⁵¹, suggesting that GXM may instead facilitate leukocyte infiltration into the brain rather than delay it.

Figure 3. Limited potential of microglia to detect cryptococcal PAMPs and DAMPs.

A-D. Macrophages were cultured for 6 hours in the presence of HK-fungi (MOI of 5:1) or PRR ligands, LPS (10 ng/mL), Pam₂CSK4 (100 ng/mL), or curdlan (100 μg/mL). Tnfa and Cxcl2 mRNA levels determined by RT-qPCR in BV2 murine microglia cell line (A), CD11b⁺ primary microglia from naïve, perfused mice (B), HMC3 human microglia cell line (C), and BMDMs elicited with M-CSF (20 ng/mL) (D). A single datapoint was obtained as the average of triplicate wells for BV2 and HMC3 in tissue culture (A, C) and from one mouse for primary microglia and BMDM samples. Data representative of three independent experiments.

E and F. BV2 microglia were cultured for 2 hours in the presence of HK-fungi (MOI 5:1) followed by stimulation with either LPS (E) or Pam₂CSK4 (F) for 4 hours. HK-fungi remained with BV2 cells during PRR stimulation. A single data point represents one well, performed in triplicate.

G and H. Evaluation of pyroptosis in brains during CM. Images of brain sections showing Cryptococcus lesions (GXM, magenta) from ASC-Citrine reporter mice r.o.-infected with 52D (10⁶ yeast/mouse) (G), and active inflammasomes were quantified (H). Representative images show ASC specks (yellow) at 7 and 14-dpi with 52D and 7-dpi with Ca (upper panel scale bar = 100 μm; lower panels scale bar = 20 μm). A single datapoint in (H) represents the average value of 9 fields from one mouse, n=4–7 mice/timepoint.

I and J. Evaluating brain cell death using TUNEL staining. Representative images of 52D-infected brains at 14-dpi (I). Positive controls are images of brain sections treated with TACS nuclease. Cryptococcus (GXM, magenta) and TUNEL (green) are shown, scale bar = 100 μm. In the quantitative analysis result (J), a single datapoint denotes an average value of 9 fields from one mouse, n=5–7 mice/timepoint.

Data presented as mean ± SEM. Statistical tests: One-way ANOVA with Dunnett’s multiple comparison test; *, p≤0.05; **, p≤0.01; ****, p≤0.0001 (A-D, H, J). For E and F, ANOVA with Dunnett’s multiple comparison test was performed on PRR-stimulated conditions.

See also Figure S5 and Table S4.

As a comparison to microglia, macrophages were also evaluated first using bone marrow-derived macrophages (BMDMs) as a representative of circulating monocyte-derived macrophages. Although BMDMs responded to HK Ca, they did not respond to HK Cryptococcus 52D, H99, and cap59Δ strains (Figure 3D). With 10 times more HK or live yeast cells (MOI 50), the cap59Δ strain slightly induced expression of Tnf and Cxcl2 by BMDMs (Figure S5C). Additionally, peritoneal macrophages (PMs), a representative of tissue-resident macrophages, were evaluated. No HK Cryptococcus strain activated gene expression of Tnf, Cxcl2, Il1b, or Il6, but only live cap59Δ very slightly induced Cxcl2 and Il1b expression (Figure S5D). In contrast, PMs responded well to live Ca, particularly in Cxcl2 expression (Figure S5D). Therefore, macrophages generally appeared to be more sensitive than microglia in live Ca recognition, but live Cryptococcus was still poorly recognized.

To further explore the poor microglial response to Cryptococcus, we tested whether Cryptococcus desensitizes PRR signal transduction. In the BV2 microglia cell culture, Cryptococcus (52D or cap59Δ) yeasts were added 2 hours before stimulation with LPS or Pam₂CSK4. The results indicated that cell conditioning with Cryptococcus did not reduce the expression of representative genes (Figure 3E, F), except for Tnf expression with TLR2/6 stimulation. This data suggested that selective and slight inhibition of host cell signal transduction by Cryptococcus may be possible, but the inhibition is not global.

Astrocytes showed similar behavior to microglia. Using astrogliosis markers (Gfap, Cxcl10, Cd274, or Cd44), no fungal species induced expression in the C8S astrocyte cell line (Figure S5E).

These data suggest that Cryptococcus, even live capsule-deficient cap59Δ, is poorly recognized by microglia and astrocytes.

Absence of significant cell death in CM, suggesting host DAMPs not to be major drivers of microglial activation

Given the poor recognition of Cryptococcus PAMPs by microglia, we investigated whether DAMPs from dying host cells might drive microglial activation. We first examined pyroptosis, a form of inflammatory cell death triggered by inflammasome activation. Using inflammasome activation reporter mice previously validated in our lab^52,53, we visualized “ASC speck” formation, indicating active inflammasome assembly⁵². While Ca increased the number of ASC specks in brains by 7-dpi, Cryptococcus 52D infection did not significantly increase ASC specks at 4, 7, or 14-dpi (Figure 3G, H). Apoptosis analysis by TUNEL staining also showed minimal cell death by 14-dpi (Figure 3I, J). Thus, with negligible host pyroptosis and apoptosis induction by 14-dpi, microglial activation in vivo may not rely on DAMPs as significant triggers.

Microglia and lymphocyte communication at 14-dpi

We performed single-cell RNA-sequencing (scRNA-seq) analysis of brains at 0, 4, and 14-dpi (Figure S6A). After quality control (Figure S6B, C) and dimensionality reduction, cells were projected onto a uniform manifold approximation and projection (UMAP), identifying 19 clusters (Figures 4A, S6D). Cluster identities were confirmed by literature-based markers and unique expression profiles (Figure 4B). At 0 and 4-dpi, the majority of CD45⁺ cells in the brain were microglia, but by 14-dpi, approximately half of the CD45⁺ population consisted of brain-infiltrated leukocytes (Figure 4C), consistent with the cytometry analysis data (Figure 1F). Changes in CD45⁻ populations were more subtle, although oligodendrocyte precursor cells (OPCs) increased at 14-dpi (Figure 4D).

Figure 4. Delayed brain immune response indicated by single-cell RNA-seq analyses.

A. Uniform manifold approximation and projection (UMAP) of all single-cell transcriptomes and cell type annotation from 0, 4, and 14-dpi brains.

B. Dot plot showing marker gene expression for individual cell clusters. Abbreviations of cell types are defined in the STAR★METHODS.

C and D. Frequency (%) of indicated cell types in total CD45⁺ cells (C) and CD45⁻ cells (D) at 0, 4, and 14-dpi.

E. Putative interaction strengths of incoming (y-axis) and outgoing (x-axis) signals of indicated CD45⁺ cell types at 0, 4, and 14-dpi. Circle size indicates the number of significantly expressed genes, inferring receptor-ligand pathways of the noted cell populations.

F. Heatmap indicating putative cell-cell communication at 4-dpi and 14-dpi based on 0-dpi. The row represents the “source” while the column represents the “target” of the communication.

Data generated by CellChat (E and F).

See also Figure S6.

Among CD45⁺ cell types, microglia exhibited a significant increase in incoming and outgoing signals at 14-dpi, along with MNPs and T cells (Figure 4E), suggesting their enhanced role in communication at this stage. Some CD45⁻ cell types, particularly oligodendrocytes (OLG), Schwann cells (SC), and vascular and leptomeningeal cells (VLMC), maintained stably high communication levels regardless of infection status (Figure S5E). Intercellular interaction analyses highlighted increased role of microglia, as well as infiltrated T cells and MNPs, in communicating other cell types as both senders and receivers of signals at 14-dpi, both in interaction strength (Figure 4F) and numbers (Figure S6F). These findings indicate that microglia not only alter gene expression profiles but also increase communication with diverse brain cell types at 14-dpi.

Emergence of unique glial subpopulations at 14-dpi

To explore transcriptional changes in CNS-resident cells during CM, we investigated glial cell heterogeneity during infection. Cell type-specific gene expression confirmed annotations for microglia, astrocytes, and oligodendrocytes (Figure S7A). Gene expression profile of these glial populations was analyzed at 0, 4, and 14-dpi. UMAP clustering showed no significant changes between 0 and 4-dpi, but entirely new subclusters were present at 14-dpi (Figures 5A, S7B, C). For microglia, clusters 1 and 2 at 0 and 4-dpi were replaced by cluster 3 at 14-dpi (Figure 5A). Similarly, in astrocytes, cluster 3 replaced clusters 1 and 2 (Figure S7B). Oligodendrocyte clusters 1, 2, and 3 persisted from 0 to 4-dpi (early stage), and a novel cluster 4 was generated at 14-dpi (later stage) (Figure S7C). Pseudotime analysis highlighted that new glial cell clusters at 14-dpi were the most transcriptionally distinct compared to those at 0 and 4-dpi (Figure S7D). These findings demonstrate that Cryptococcus infection dramatically reprograms glial gene expression by 14-dpi, with little change observed between 0 and 4-dpi.

Figure 5. Gene expression profile of 14-dpi CM-microglia is distinct from DAM.

A. UMAP of microglia subsets present at 0, 4, and 14-dpi.

B. Volcano plots of DEGs in microglia at 14-dpi compared to 0-dpi.

C. Dot plot of the top 20 DEGs in microglia (CM-MG) showing average expression levels (color bar) and percentage of microglia expressing the gene (filled circles) at 0, 4, and 14-dpi.

D. Enriched hallmark pathways in microglia at 14-dpi compared to 0-dpi.

E. Relative expression levels of genes expressed in DAM (y-axis) and 14-dpi CM-MG (x-axis).

See also Figure S7 and Tables S1–S3.

Gene expression profile in CM-induced microglia is distinct from DAM

Microglia exhibited more differentially expressed genes (DEGs) than astrocytes and oligodendrocytes at both 4 and 14-dpi (Figure S7E). Limited number of DEGs were identified in microglia (Figure S7E, F) and astrocytes (Figure S7E, G) at 4-dpi compared to 0-dpi, but the number if DEGs significantly increased at 14-dpi (Figure S7E). In microglia, DEGs at 14-dpi included prominently upregulated Interferon Stimulated Genes (ISGs), including Irgm1⁵⁴, Gbp2⁵⁵, Igtp⁵⁶, and Iigp1⁵⁷, along with MHC-related molecules, such as H2 genes, B2m, and Cd74 (Figure 5B, C). DEGs in astrocytes and oligodendrocytes at 14-dpi also included these genes (Figure S7H, I). These upregulated DEGs in the glial cells represent the “IFNγ response gene” set (Figures 5D, S7J). DEGs in astrocytes also included astrogliosis-associated genes (Gfap, Cxcl10, and Cd274) at 14-dpi (Figure S7H), supported by RT-qPCR data (Figure S4E, F).

In contrast, major downregulated genes at 14-dpi in microglia included homeostatic genes Cx3cr1, P2ry12, and Tmem119^21,46 (Figures 5B, S7K). Downregulation of Cx3cr1, P2ry12, and Gpr34 started at 4-dpi (Figure S7K), but these were among a small number of genes (Figure S7E).

Further analysis revealed that DEGs in microglia at 14-dpi during CM (CM-MG) did not correlate well to those in DAM (Figure 5E, S7M). For instance, genes like Ccl5, Ccl24, and Gbp7 were upregulated in CM-MG but downregulated in DAM, while Trem2, Timp2, and Itga5 showed the opposite trend (Figure 5E, Tables S1 and S2). Notably, Trem2 is a DAM hallmark gene^21,58. Despite these differences, some similarities were observed, including upregulation of Itgax, Csf1, Cst7, Axl, Lyz2, Apoe, and Spp1 (Figure 5E, Table S3). These findings reveal microglial responses at 14-dpi in CM-MG are distinct from those in DAM.

T cell-microglia communication with IFNγ

Given the IFNγ-mediated gene expression pattern in glial cells at 14-dpi in CM, we predicted the source of IFNγ in the brain during infection. CellChat analysis identified T cells as the primary source of IFNγ at 14-dpi (Figure S8A), predominantly perceived by microglia, MNPs, OPC, and, to a lesser extent, VLMC (Figure 6A), which expressed Ifngr1 and Ifngr2, encoding the heterodimeric IFNγ receptor (Figure 6B). UMAP clustering confirmed Ifng and Tbx21 (encoding T-bet) expression in T cells at 14-dpi (Figure 6C), with minimal expression Il4 and Gata3 (Th2-associated genes; Figure S8B) or Il17a and Rorc (Th17-associated genes; Figure S8C). Indeed, IFNγ protein expression in CD4⁺ T cells was detected in over 40% of these cells at 14 and 21-dpi, while Th17 cells were scarce (Figure 6D).

Figure 6. Requirement of T cells for delayed microglia activation and immune cell recruitment into the brain.

A. Intercellular communication mediated by IFNγ at 14-dpi. Black arrows indicate direction of communication. The thickness of lines reflects the estimated interaction strength.

B. Violin plots showing gene expression levels of Ifngr1 and Ifngr2 in the indicated cell types at 14-dpi.

C. UMAPs of cell types expressing Ifng and Tbx21 (Th1 signature).

D. Representative flow cytometry plots depicting IFNγ and IL-17A production by CD4⁺ T cells isolated from 52D-infected brains at 14 and 21-dpi and frequency (%) of IFNγ- and IL-17A-producing CD4⁺ T cells in total CD4⁺ T cells, n=5 mice/group.

E-J. Comparison between WT and Tcra^−/− mice at 14-dpi with 52D. Representative flow cytometry gating for microglia, myeloid cells, and lymphocytes and quantification of total CD45^hi cells (E), MNPs (F), NK cells (G), B cells (H), n=6–10 mice/group. Microglia MHCII expression (I); black and blue dotted lines represent microglia MHCII MFI of naïve WT and Tcra^−/− mice, respectively, n=3–4 mice/group. Representative images of Iba-1⁺ cell distributions (cyan) near Cryptococcus lesions (GXM, magenta) in 52D-infected mice (10⁶ yeast/mouse) (J), scale bar = 200 μm.

K. Frequency (%) of Iba-1⁺ area in peri-lesional area, defined in Figure S3L. A single datapoint denotes a percentage value of the Iba-1⁺ area from 2–4 lesions from a mouse from n=4–6 mice/group.

L. Brain fungal burdens at 14-dpi from 52D-infected mice.

M-P. Microglia activation at 14-dpi with 52D in Tcra^−/− mice reconstituted with either naïve WT or Ifng^−/− knockout CD4⁺ T cells. Workflow for adoptive transfer (M). Frequency (%) and total cell counts of CD4⁺ T cells in the brain (N) and total microglia number (O). Representative histograms expression of MHCII MFI in microglia and corresponding quantification (P).

Data presented as mean ± SEM. Statistical tests: two-tailed unpaired Student’s t-test; **, p≤0.01; ; ***, p≤0.00; ****, p≤0.0001 (D-I, K, L, N-P). Data representative of two independent experiments (D-I, K, L).

See also Figure S8.

Furthermore, unbiased ligand-receptor interaction analyses revealed T cell-to-microglia communication networks beyond IFNγ, including interactions such as Ppia-Bsg and Ccl5-Ccr5, at 14-dpi (Figure S8D). Cyclophilin A (encoded by Ppia) binds CD147 (encoded by Bsg), driving inflammation, cell migration, and matrix metalloproteinase (MMP) production^59,60. Ccl5-Ccr5 connection suggests a role in chemotaxis and microglial activation^61,62. Additionally, microglia-to-T cell signals included CD86-Ctla4 and H2 genes-Cd4 combinations (Figure S8D), implicating immune checkpoint activity towards T cells. These findings emphasize bidirectional communication between CD4⁺ T cells and microglia at 14-dpi.

Th1 cell-driven microglial activation and fungal control during CM

To investigate the role of T cells during CM, we compared 52D infection (10⁶ yeasts/mouse) in WT and Tcra^−/− mice. T cell absence did not affect body weights in Tcra^−/− mice during infection (Figure S8E), but Tcra^−/− mice showed significantly reduced cell numbers and proportions of brain infiltrated leukocytes (Figures 6E, S8F), due to overall decreases both in myeloid cells and lymphocytes (Figure S8G), particularly MNPs and NK cells (Figures 6F, G, S8H,I). B cell counts also decreased, although their proportions remained stable in Tcra^−/− mice (Figures 6H, S8J). Brains in Tcra^−/− mice showed increased neutrophil frequency in CD45⁺ cells, but neutrophil numbers were comparable (Figure S8K). Microglia numbers also remained unchanged in Tcra^−/− mice (Figure S8L).

As T cells were predicted to communicate with microglia (Figures 6A, S8D), T cells were essential for microglial activation, evidenced by reduced MHCII expression in microglia from Tcra^−/− mice at 14-dpi (Figure 6I). Additionally, no Iba-1⁺ cells, including microglia, were recruited to fungal lesions without T cells (Figure 6J, K). Tcra^−/− mice also had significantly higher brain fungal burdens than WT mice at 14-dpi (Figure 6L). Tcra^−/− mice exhibited increased fungal burdens in the lung, but not in the spleen (Figure S8M). However, the magnitude of this increase was not comparable to that observed in the brain (Figure 6L), suggesting the antifungal impact of T cells is more pronounced in the brain.

T cell IFNγ as a driver of microglial activation

To determine whether IFNγ production by T cells is necessary for microglia activation during CM, we adoptively transferred CD4⁺ T cells (5 × 10⁵ cells/mouse) obtained from naïve WT or Ifng^−/− mice into Tcra^−/− recipients, followed by Cryptococcus 52D infection (Figure 6M). Frequency and counts of CD4⁺ T cells, as well as microglia numbers, in the brain were comparable between the two groups (Figure 6N, O). However, microglia in recipients transferred with Ifng^−/− CD4⁺ T cells were unable to upregulate MHCII expression (Figure 6P). Together, these results highlight the critical role of CD4⁺ T cells in microglia activation during CM.

Microglia-derived OPN extends its detrimental influence beyond the CNS

A recent study demonstrated that microglia depletion did not reduce brain fungal loads during Cryptococcus infection with the H99 strain¹². However, the role of microglia in modulating host immune responses during CM remains unclear. Here, we focused on OPN due to its increased expression in microglia during CM (Figure 2M, 5E), neurodegenerative diseases^32,63,64, and brain development^21–26.

We first evaluated the impact of OPN in CM by comparing WT and Spp1^−/− in CM, as OPN is widely expressed in various leukocyte cell types^{27–29,31–35,65–67}. However, global OPN knockout did not alter mouse weight loss (Figure S9A). Even an extended observation did not alter survival (Figure S9B). Spp1^−/− mice demonstrated decreased myeloid cell numbers (Figure S9C) and decreased brain weights (Figure S9D) at 14-dpi, but brain fungal burdens were not different between both groups at 7, 14, and 21-dpi (Figure S9E).

Next, to assess microglia-derived OPN, we generated microglia-specific Spp1 deletion mutant mice (Cx3cr1^CreERT2; Spp1^fl/fl, henceforth Spp1^ΔMG ) and treated them with TAM, followed by a 6-week washout period before r.o.-infection with the 52D strain. Control mice (Cx3cr1^CreERT2) also received TAM. The control group exhibited significantly greater weight loss than Spp1^ΔMG mice, but Spp1^ΔMG mice did not (Figure 7A), indicating that the absence of OPN in microglia protected infected hosts. In these experiments, control mice exhibited weight loss, not observed in WT mice, which do not have the Cre gene and tamoxifen treatment. Such effects appear to be attributable to the system itself, as tamoxifen can modulate immune responses and metabolism, and Cre expression may also influence estrogen-related signaling pathways. Thus, by comparing Cx3cr1^CreERT2; Spp1^fl/fl (test) mice with Cx3cr1 ^CreERT2 (control) mice—both carrying Cre and treated with TAM--we ensured that any observed differences were specifically attributed to OPN depletion in microglia.

Figure 7. Microglial OPN exerting detrimental effects through immunomodulation beyond the brain.

A. Comparison of total body weight change between control (Cr3cr1^CreERT2/WT, n=3) and Spp1^ΔMG male mice (Cr3cr1^CreERT2/WT;Spp1^fl/fl, n=7) r.o.-infected with 52D (10⁶ yeast/mouse). Area above the curve of percent weight change is shown for statistical analysis. Both sets of data show mean ± SEM.

B and C. Frequencies (%) in CD45⁺ cells and numbers of myeloid cells (B) and lymphocytes (C) in brains at 14-dpi.

D and E. Total numbers of microglia (D) and activated microglia (E) in brains at 14-dpi.

F and G. Fungal burdens in brains (F) and lungs (G) from 14-dpi mice.

H. Concentrations of IL-10 and IL-23 in mouse sera at 14-dpi.

Data presented as mean ± SEM. Statistical tests: two-tailed unpaired Student’s t-test; ns, not significant; *, p≤0.05; **, p≤0.01. Data representative of two independent experiments (A-H).

See also Figure S9.

Despite differences in weight loss, the weights of brains, lungs, and spleens did not differ between both groups (Figure S9F) and the numbers of recruited myeloid cells or lymphocytes in the brain at 14-dpi were comparable (Figures 7B, C, S9G, H), as were total microglia numbers (Figure 7D) and activated microglia (Figure 7E). Brain fungal burdens were also similar between Spp1^ΔMG and control mice (Figure 7F). However, Spp1^ΔMG mice exhibited significantly reduced fungal burdens in the lungs while spleen fungal burdens were comparable between groups (Figures 7G, S9I). Additionally, at 14-dpi, Spp1^ΔMG mice exhibited elevated serum levels of IL-10 and IL-23 (Figure 7H), along with trends of decreased IFNβ and increased IFNγ levels (Figure S9J). Together, these findings suggest that microglia-derived OPN plays a detrimental role in CM, influencing beyond the CNS and modulating peripheral host responses in a detrimental fashion.

DISCUSSION

This study used a Cryptococcus infection model mirroring CM with non-acute progression, commonly observed in humans. Under non-acute conditions, where mice survive for at least a month post-infection, Cryptococcus rapidly infiltrates the brain within one day. However, microglia required 14 days to fully activate, a delay that sharply contrasts with alveolar macrophages (AMs), which respond within hours by secreting CXCL2 to recruit neutrophils⁵⁰. Delayed microglial activation and leukocyte recruitment to the brain suggest Cryptococcus exploits slow host responses to establish infection.

Olszewski and colleagues previously reported the dual roles of CD4⁺ T cells, showing that they reduce brain fungal burdens while simultaneously increasing inflammation⁴. Our results extend previous findings by revealing a new role of CD4⁺ T cells as catalysts for microglial activation by showing IFNγ from brain-infiltrated T cells is the primary activator of microglia. At 14-dpi, ~40–50% of CD4⁺ T cells were Th1 cells, consistent with previous studies showing Th1 dominance in CM^5,6. We also demonstrated that T cells were essential for microglial clustering around Cryptococcus lesions and reducing brain fungal burden, highlighting the beneficial role of Th1 cells, as reported during peripheral cryptococcosis^68–71. However, excessive IFNγ can lead to cerebral edema, as observed in C-IRIS⁷. Balancing the Th1 response is crucial to optimize fungal clearance while minimizing adverse effects in the brain.

We demonstrated that Cryptococcus and Ca yeasts did not directly stimulate microglia, as shown by the failure to alter the expression of various genes. Even the cap59Δ mutant failed to elicit the responses, suggesting that microglia are not well-equipped for direct fungal detection. This insensitivity suggests the presence of complex mechanisms in fungal recognition. In contrast, BMDMs directly detected Ca but also failed to recognize the cap59Δ mutant, implying that distinct PRRs or signaling molecules effective against Ca are inadequate for Cryptococcus. Although the Cryptococcus’ capsule is well-known to serve as an immune evasion strategy⁷², our findings suggest an additional mechanism beyond its physical protection.

Our study showed unexpectedly poor microglial responses during CM, whereas a previous report described IL-1β and CXCL1 expression by microglia during C. albicans infection¹³, including in ex vivo stimulation assays. The seemingly conflicting results reflect differences in experimental design: the study employed various microglial priming or conditioning in the setting¹³, whereas our analyses used unprimed and unconditioned microglia. Thus, their findings also imply that additional stimulation is required for microglia to produce these cytokines, consistent with their statement that “Microglia produce IL-1β and CXCL1 in a candidalysin-dependent manner.”

Although microglia have limited capacity to detect Cryptococcus, our findings suggested that microglia respond once T cells are recruited to the brain. This raises an important question: Are microglia beneficial or detrimental to the host in CM? A previous study showed that brain fungal loads did not increase in the absence of microglia¹². However, our data suggested microglia influence host responses even at the systemic level. Specifically, we examined OPN, encoded by the Spp1 gene, which is induced in microglia under various conditions, including DAM²¹, axonal tract microglia (ATM)²², proliferative-region-associated microglia (PAM)²³, CD11c⁺ microglia²⁴, and during acute demyelination²⁵. Microglia-specific deletion of Spp1 protected against CM-mediated weight loss, indicating that microglial OPN is detrimental.

Notably, the detrimental effect was more pronounced in peripheral tissues than in the brain. While brain fungal loads were unaffected, fungal burdens in the lung and spleen were significantly reduced, suggesting that microglial OPN adversely affects peripheral host responses during CM. Despite the pathogenic role of microglial OPN, global OPN knockout did not affect outcomes. This may be due to the contextual nature of OPN function, which can be either beneficial or detrimental depending on the cellular environment, as demonstrated in numerous previous studies. In other words, the complete absence of OPN across all cell types may neutralize opposing effects, thereby masking phenotype differences. In addition, OPN exists with two isoforms: sOPN and iOPN. Due to the distinctive roles of the isoforms^{28–32,35,66,73,74}, the relative expression of balance of the two isoforms appeared to be cell type-dependent³³. Therefore, global Spp1 knockout across diverse cell types is likely to produce different outcomes compared to microglia-specific Spp1 deletion.

We found that OPN depletion in microglia increases serum levels of IL-10 and IL-23. Previous studies provide insights into the role of these cytokines during Cryptococcus infection. In one study, IL-23 knockout mice (p19^−/−) had slightly reduced survival times⁷⁵, suggesting IL-23 may play a beneficial role. Conversely, another study highlighted a detrimental role for IL-10 by demonstrating that antibody-mediated IL-10 depletion between 15- and 21-dpi enhanced immune responses and reduced lung fungal burdens⁷⁶. In our setting, it is possible that elevated serum IL-10 at 14-dpi, when antifungal responses in the brain are established, may work beneficially by controlling excessive inflammation.

Our findings also raise another question: How does microglial OPN exert systemic effects during CM? Since OPN is produced by most leukocytes, except naïve T cells^29,33, it is unlikely that sOPN produced by microglia leaks out of the brain and modifies peripheral immune responses. Instead, microglial OPN may alter the brain microenvironment, indirectly impacting peripheral immunity. For instance, sOPN from microglia interacts with integrins and CD44⁷⁷ to modulate cell behaviors in the brain⁷⁷. iOPN in microglia might influence their function as previously demonstrated⁶⁷.

Our data also show that global OPN loss in Spp1^−/− mice reduced brain inflammation, but this was not accompanied by changes in fungal burden or weight loss. This likely reflects the context-dependent nature of OPN, which can be either beneficial or detrimental depending on the cellular environment, the OPN-expressing cell types involved, and the relative contribution of different OPN isoforms (reviewed in^29,32). As a result, complete OPN deficiency in Spp1^−/− mice may mask the microglia-specific phenotype observed in Spp1^ΔMG mice.

In this study, we demonstrated that microglia have a limited capacity to detect Cryptococcus in the brain; thus, microglia activation requires Th1 cells to infiltrate the brain. Additionally, microglial OPN was detrimental in CM by altering peripheral immunity and increasing fungal loads in peripheral organs. These findings suggest that enhancing early microglial responses with IFNγ and regulating OPN expression may alter host responses against CM.

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mari L. Shinohara (mari.shinohara@duke.edu)

Materials Availability

Materials used in this study are listed in the Key Resource Table. All reagents and mice generated in this study are available from the lead contact upon completion of a Materials Transfer Agreement.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rat monoclonal anti-mouse CD170 PerCP-Cy5.5 (Clone: E50-2440)	BD	Cat# 565526; RRID: AB_2739281
Rat monoclonal anti-mouse CD45 Brilliant Ultraviolet 395 (Clone: 30-F11)	BD	Cat# 564279; RRID: AB_2651134
Rat monoclonal anti-mouse CD45 PE/Cyanine7 (Clone: 30-F11)	BioLegend	Cat# 103114; RRID: AB_312979
Armenian hamster monoclonal anti-mouse TCR-β chain PE/Cyanine5 (Clone:H57-597)	BioLegend	Cat# 109210; RRID: AB_313433
Armenian hamster monoclonal anti-mouse TCR-β Brilliant Violet 711 (Clone: H57-597)	BioLegend	Cat# 109243; RRID: AB_2629564
Armenian hamster monoclonal anti-mouse CD11c Brilliant Violet 510 (Clone: N418)	BioLegend	Cat# 117337; RRID: AB_2562010
Armenian hamster monoclonal anti-mouse CD11c PE (Clone: N418)	BioLegend	Cat# 117308; RRID: AB_313777
Rat monoclonal anti-human/mouse CD11b Brilliant Violet 605 (Clone: M1/70)	BioLegend	Cat# 101257; RRID: AB_2565431
Rat monoclonal anti-human/mouse CD11b APC (Clone: M1/70)	BioLegend	Cat# 101212; RRID: AB_312795
Rat monoclonal anti-mouse LY6G PE/Cyanine7 (Clone: 1A8)	BioLegend	Cat# 127618; RRID: AB_1877261
Rat monoclonal anti-mouse LY6G Brilliant Violet 421 (Clone:1A8)	BioLegend	Cat# 127628; RRID: AB_2562567
Rat monoclonal anti-mouse LY6G Brilliant Violet 510 (Clone: 1A8)	BioLegend	Cat# 127633; RRID: AB_2562937
Rat monoclonal anti-mouse LY6C Brilliant Violet BV711 (Clone: HK1.4)	BioLegend	Cat# 128037; RRID: AB_2562630
Rat monoclonal anti-mouse LY6C PE/Cyanine7 (Clone: HK1.4)	BioLegend	Cat# 128018; RRID: AB_1732082
Rat monoclonal anti-mouse CD19 APC/Cyanine7 (Clone: 6D5)	BioLegend	Cat# 115529; RRID: AB_830706
Rat monoclonal anti-mouse CD19 APC (Clone: 6D5)	BioLegend	Cat# 115512; RRID: AB_313647
Mouse monoclonal anti-mouse NK1.1 PE/Cyanine7 (Clone: PK136)	BioLegend	Cat# 108714; RRID: AB_389364
Rat monoclonal anti-mouse I-A/I-E FITC (Clone: M5/114.15.2)	BioLegend	Cat# 107606; RRID: AB_313321
Rat monoclonal anti-mouse I-A/I-E PE (clone M5/114.15.2)	BioLegend	Cat# 107608; RRID: AB_313323
Rat monoclonal anti-mouse I-A/I-E Alexa Fluor 700 (Clone: M5/114.15.2)	BioLegend	Cat# 107622; RRID: AB_493727
Mouse monoclonal anti-mouse CD24 PE (Clone: M1/69)	BioLegend	Cat# 101807; RRID: AB_312840
Mouse monoclonal anti-mouse CD64 APC (Clone: X54-5/7.1)	BioLegend	Cat# 139306; RRID: AB_11219391
Rat monoclonal anti-mouse CD4 APC/Cyanine7 (Clone: GK1.5)	BioLegend	Cat# 100414; RRID: AB_312699
Rat monoclonal anti-mouse CD8a Alexa Fluor 488 (Clone: 53-6.7)	BioLegend	Cat# 100723; RRID: AB_389304
Rat monoclonal anti-mouse CD8a Brilliant Violet 785 (Clone: 53-6.7)	BioLegend	Cat# 100722; RRID: AB_312761
Rat monoclonal anti-mouse IFNy Alexa Fluor 488 (Clone: XMG1.2)	BioLegend	Cat# 505813; RRID: AB_493312
Rat monoclonal anti-mouse IL-17a Brilliant Violet 421 (Clone: TC11-18H10.1)	BioLegend	Cat# 506925; RRID: AB_10900442
Mouse Polyclonal anti-GXM	From the laboratory of Dr. Arturo Casadevall (Johns Hopkins University)	N/A
Goat polyclonal anti-Iba1	Novus	Cat# NB100-1028; RRID: AB_3148646
Chicken monoclonal anti-Iba1	Synaptic Systems	Cat# 234-009; RRID: AB_2891282
Rabbit polyclonal anti-GFAP	Abcam	Cat# ab7260; RRID: AB_305808
Rabbit polyclonal anti-MBP	Abcam	Cat# ab40390; RRID: AB_1141521
Goat polyclonal anti-Mouse IgG (H+L) Cross-Adsorbed Secondary Antibody Alexa Fluor 488	Thermo Fisher Scientific	Cat# A11029; RRID: AB_2534088
Donkey polyclonal anti-Rabbit (H+L) Highly Cross-Adsorbed Secondary Antibody Alexa Fluor 488	Thermo Fisher Scientific	Cat# A21206; RRID: AB_2535792
Donkey polyclonal anti-Goat (H+L) Highly Cross-Adsorbed Secondary Antibody Alexa Fluor 488	Thermo Fisher Scientific	Cat# A11055; RRID: AB_2534102
Goat polyclonal anti-Rabbit (H+L) Highly Cross-Adsorbed Secondary Antibody Alexa Fluor 647	Thermo Fisher Scientific	Cat# A21244; RRID: AB_2535812
Goat polyclonal anti-Chicken IgG (H+L) Cross-Adsorbed Secondary Antibody Alexa Fluor 647	Thermo Fisher Scientific	Cat# A21449; RRID: AB_2535866
CD11b MicroBeads, human and mouse	Miltenyi Biotec	Cat# 130-049-601; RRID: AB_2927377
Anti-ACSA-2 MicroBeads Kit	Miltenyi Biotec	Cat# 130-097-678; RRID: AB_2894998
Myelin Removal Beads II, human, mouse, rat	Miltenyi Biotec	Cat# 130-096-733
Bacterial and virus strains
Biological samples
Chemicals, peptides, and recombinant proteins
Curdlan (from Alcaligenes faecalis, β-1,3 Glucan hydrate; Dectin-1 agonist)	Sigma-Aldrich	Cat# C7821-5G
Pam2CSK4 (TLR2/6 agonist)	InvivoGen	Cat# tlrl-pm2s-1
Lipopolysaccharide (LPS) (TLR4 agonist)	Sigma	Cat# L439101MG
Recombinant mouse (rm) M-CSF	BioLegend	Cat# 576406
Recombinant mouse (rm) IFNγ	BioLegend	Cat# 575306
Tamoxifen	Cayman Chemical	Cat# 13258
Corn Oil	Gibco	Cat# 405435000
Collagenase D Type IV (from Clostridium histolyticum)	Roche	Cat# 11088866001
Liberase TM	Roche	Cat# 5401119001
DNase I	Millipore	Cat# 260913-10MU
Prolong Gold Antifade Mountant	Thermo Fisher Scientific	Cat# P36934
Ionomycin (calcium salt from Streptomyces conglobatus)	Sigma-Aldrich	Cat# 10634
Phorbol 12-myristate 13-acetate (PMA)	Sigma-Aldrich	Cat# P1585
Brefeldin A solution	Biolegend	Cat# 420601
Isofluorane	Covetrus	Cat# 29405
Heparin	Sagent Pharmaceuticals	Cat# NDC 25021-400-30
Tissue-Tek OCT Compound	Sakura	Cat# 4583
TritonX100	Amresco	Cat# 0694-1L
Bovine Serum Albumin (BSA)	GeminiBio	Cat# 700-101P
Tween-20	Sigma	Cat# P1379-100ML
Fetal Bovine Serum - Premium, Heat Inactivated	Atlanta Biologicals	Cat# S11150H
HEPES 1M	Thermo Fisher Scientific	Cat# 15630080
Percoll	Cytiva	Cat# GE17-0891-01
10X Hanks’ balanced salt solution buffer (HBSS)	Thermo Fisher Scientific	Cat# 14065056
DAPI (4’,6-diamidino-2-phenylindole)	Sigma	Cat# D9542-10MG
Zombie UV™ Fixable Viability Kit	BioLegend	Cat# 423107
DMEM medium	Thermo Fisher Scientific	Cat# 11995073
EMEM	ATCC	30-2003
Penicillin-Streptomycin	Thermo Fisher Scientific	Cat# 15140122
L-glutamine	Thermo Fisher Scientific	Cat# 25030081
TRI Reagent	Millipore Sigma	Cat# 93289-100ML
1–bromo–3–chloropropane (BCP)	Acros Organics	Cat# AC106862500
1X Phosphate Buffered Saline (PBS)	Gibco	Cat# 10010049
10X Phosphate Buffered Saline (PBS)	Gibco	Cat# 70011044
RPMI 1640 medium	Thermo Fisher Scientific	Cat# 22400121
Critical commercial assays
CF640R TUNEL Assay Apoptosis Detection Kit	Biotium	Cat# 30074
TACS Nuclease	R&D	Cat# 4800-30-N
Papain Dissociation System	Worthington Biochemical Corporation	Cat# LK003153
qScript cDNA Mix	QuantaBio	Cat# 950048
PowerTrack SYBR Green Master Mix	Applied Biosystems	Cat# A46109
Easy Dead Cell Removal (Annexin V) Kit	STEMCELL Technologies	Cat# 17899
CellPlex Barcoding Kit	10x Genomics	CellPlex Kit Set A
LEGENDplex™ Mouse Inflammation Panel (13-plex) with V-bottom Plate	BioLegend	Cat# 740446
Cyto-Fast™ Fix/Perm Buffer Set	BioLegend	Cat# 426803
UltraComp eBeads™ Compensation Beads	Thermo Fisher Scientific	Cat# 01-2222-42
EasySep™ Mouse Naïve CD4+ T Cell Isolation Kit	STEMCELL	Cat# 19765
Deposited data
Raw scRNA-seq data for “Cryptococcus exploits delayed microglial activation, and microglial osteopotin (OPN/Spp1) impairs peripheral host control”	This paper	GEO: GSE284662
Available code for analysis of scRNA-Seq Data “Cryptococcus exploits delayed microglial activation, and microglial osteopotin (OPN/Spp1) impairs peripheral host control”	This paper	Dataverse - https://doi.org/10.7910/DVN/HQI3LD
Reanalyzed data for “Single cell RNA-seq identifies a unique microglia type associated with Alzheimer’s disease”	Keren-Shaul et al., 2017	GEO: GSE98969
Mouse reference genome, version GRCm38	Genome Reference Consortium	https://www.ncbi.nlm.nih.gov/grc/mouse
Experimental models: Cell lines
Mouse: BV-2	From the laboratory of Dr. Cagla Eroglu (Duke University)	RRID:CVCL_0182
Mouse: C8-S	ATCC	CRL-2535; RRID: CVCL_6381
Human: HMC3	ATCC	CRL-3304; RRID: CVCL_II76
Experimental models: Organisms/strains
Mouse: C57BL/6J	The Jackson Laboratory	Strain#: 000664; RRID: IMSR_JAX:000664
Mouse: CD11c-EYFP; B6.Cg-Tg(Itgax-Venus)1Mnz/J	The Jackson Laboratory	Strain#: 008829; RRID: IMSR_JAX:008829
Mouse: Cx3cr1 CreERT2; B6.129P2(C)-Cx3cr1tm2.1 (cre/ERT2)Jung/J	The Jackson Laboratory	Strain#: 020940; RRID: IMSR_JAX:020940
Mouse: Rosa26 LSL-tdTomato; B6.Cg-Gt(ROSA)26Sortm9(CAG-tdTomato)Hze/J	The Jackson Laboratory	Strain#: 007909; RRID: IMSR_JAX:007909
Mouse: Tcra−/−; B6.129S2-Tcratm1Mom/J	The Jackson Laboratory	Strain#: 002116; RRID: IMSR_JAX:002116
Mouse: Ifng−/−; B6.129S7-Ifngtm1Ts/J	The Jackson Laboratory	Strain#: 002287; RRID: IMSR_JAX:002287
Mouse: ASC-Citrine	From the laboratory of Dr. Douglas Golenbock (University of Massachusetts)	N/A
Mouse: Spp1fl/fl	From the laboratory of Dr. Natalia Nieto (University of Illinois, Chicago, with MTA)	N/A
C. neoformans (deneoformans) strain 52D	ATCC	Cat# 24067
C. neoformans strain H99	From the laboratories of Drs. J. Andrew Alspaugh and Joseph Heitman (Duke University)	N/A
C. neoformans strain cap59Δ	From the laboratories of Drs. J. Andrew Alspaugh and Joseph Heitman (Duke University)	N/A
C. albicans strain SC5314	From the laboratories of Drs. J. Andrew Alspaugh and Joseph Heitman (Duke University)	N/A
Oligonucleotides
Primers for RT-qPCR: see Table S4.	This paper
Recombinant DNA
Software and algorithms
R statistical programming environment	R Foundation for Statistical Computing	https://www.r-project.org
CellRanger	10X Genomics	https://www.10xgenomics.com/support/software/cell-ranger/latest
Loupe Browser V7	10X Genomics	https://www.10xgenomics.com/support/software/loupe-browser/latest
CellChat V2	Jin et al., 2021	http://www.cellchat.org/
Monocle3	Trapnell et al., 2014	https://cole-trapnell-lab.github.io/monocle3/
ImageJ	Schindelin et al., 2012	https://imagej.net/Fiji
Imaris Version 9.6	Oxford Instruments	https://imaris.oxinst.com/
FlowJo	BD	https://www.flowjo.com/solutions/flowjo
Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
Other

Data and Code Availability

Single-cell RNA-seq data have been deposited at GEO (accession number: GSE284662) and are publicly available as of the date of publication. All original code has been deposited at Dataverse and is publicly available (https://doi.org/10.7910/DVN/HQI3LD) as of the date of publication. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

STAR★METHODS

EXPERIMENTAL MODELS AND STUDY PARTICIPANT DETAILS

Mice

Six- to eight-week-old age- and sex-matched mice of the C57BL/6 (B6) background were used for experiments unless otherwise specified. We combined data from males and females when no difference in results was found between two sexes. The ASC-Citrine mice were generated and gifted by Douglas Golenbock (University of Massachusetts)⁵². Spp1^fl/fl mice are gifts from Dr. Natalia Nieto (University of Illinois, Chicago) with MTA. The following strains were obtained from the Jackson Laboratories: CD11c-EYFP (Cat. 008829), Cx3cr1^CreERT2 (Cat. 020940), Rosa26 ^LSL-tdTomato (Cat. 007909), Tcra^−/− (Cat. 002116), and Ifng^−/− (Cat. 002287). All mice were housed under specific pathogen-free conditions and were genotyped before being used for experiments. All animal experiments were performed as approved by the Institutional Animal Care and Use Committee at Duke University.

Fungal strains and cultures

Cryptococcus neoformans (deneoformans) strain 52D was purchased from ATCC (Cat. 24067). C. neoformans strains H99 and cap59Δ, and Candida albicans (Ca) strain SC5314 were gifts from Drs. J. Andrew Alspaugh and Joseph Heitman (Duke University). Fungal strains were kept in 30% glycerol stock stored at −80°C and plated on YPD plates for overnight growth at 30°C. A single colony was picked and further cultured in liquid YPD broth with shaking at 225 rpm overnight. Yeasts were then collected by centrifugation, washed twice with 1X PBS, and enumerated using a hemacytometer. When heat-killed (HK) yeasts were used, yeasts in 1X PBS were placed at 95°C for 1 hour. HK yeasts were validated by plating on YPD plates for no growth. To evaluate colony forming units (CFU), tissues from infected mice were homogenized in water. Mice were not perfused prior to CFU assessment. After serial dilutions, homogenates were plated on YPD plates. CFU count was performed after incubating plates at 30°C for 3 days.

Mammalian cell lines

C8-S murine astrocyte cell line was purchased from the ATCC (Cat. CRL-2535). BV2 murine microglia cell line was kindly provided by the laboratory of Dr. Cagla Eroglu (Duke University). HMC3 human microglia cell line was kindly provided by the laboratory of Dr. Xiao-Fan Wang (Duke University). All cell lines were maintained in a humidified incubator at 37°C with 5% CO₂.

METHOD DETAILS

Reagents, antibodies, and recombinant proteins

The following pattern recognition receptor (PRR) agonists were used for ex vivo experiments, unless otherwise noted: Curdlan (β–1,3 glucan hydrate, from Alcaligenes faecalis, Sigma-Aldrich) was used as a Dectin-1 agonist, Pam₂CSK4 (InvivoGen) as a TLR2/6 agonist, and lipopolysaccharide (Sigma, LPS) as a TLR4 agonist. Recombinant mouse (rm) M-CSF and IFNγ proteins were obtained from BioLegend. Detailed list of antibodies and reagents can be found in Key Resources Table.

Tamoxifen treatment for selective Cre expression by microglia

To achieve microglia-specific Cre expression in Cx3cr1^CreERT2 mice, we peritoneally injected 150 mg/kg tamoxifen (Cayman Chemical) in corn oil (Thermo Scientific Chemicals) into 6–10 weeks old mice 3 times, 24 hours apart between each infection, and rested them for 4–6 weeks (washout period), allowing for BM-derived myeloid cells to be replaced, as previously described⁷⁸. Cx3cr1^CreERT2 mice without a floxed locus, used for a control group, were also treated with tamoxifen.

Fungal infection of mice

Mice were exposed to isoflurane (Covetrus) followed by retro-orbital injection of fungal yeast in a max volume of 100 μL 1X PBS (Thermo Fisher Scientific). The syringe needle was introduced bevel down at a 30-degree angle into the retro-bulbar space, behind the globe of the eye. For Cryptococcus infection, we used 10⁶ yeasts/mouse unless otherwise noted. For Ca infection, we used 10⁵ yeast per mouse unless otherwise noted. Euthanasia was performed at noted timepoints or a humane endpoint, whichever came first, using CO₂ asphyxiation, followed by a secondary method, as recommended by the American Veterinary Medical Association.

T cell isolation and adoptive transfer

Splenic CD4⁺ T cells from naive WT or Ifng^−/− mice were enriched using the EasySep Mouse Naïve CD4⁺ T Cell Isolation Kit (STEMCELL Technologies) following the manufacturer’s protocol. Isolated cells were resuspended in 1X PBS (Thermo Fisher Scientific) and 5×10⁵ cells were r.o.-injected into Tcra^−/− mice. Mice were rested for 7 days prior to infection with Cryptococcus strain 52D.

Isolation of brain tissue for histological analysis

Following euthanasia, trans-cardiac perfusion was performed on mice with 10 mL of 1X PBS containing 0.1% Heparin (Sagent Pharmaceuticals) followed by 10 mL of ice-cold 4% paraformaldehyde (PFA). Brain tissue was extracted and fixed in 4% PFA overnight at 4°C, followed by dehydration in 30% sucrose for 1–3 days at 4°C (until the brain dropped to the bottom of the tube). Afterwards, brains were cut into sagittal halves and were embedded in Tissue-Tek OCT compound (Sakura) and stored at −80°C for long-term storage.

Immunofluorescent staining and imaging

Tissues were sectioned using a Cryostar NX50 (Thermo Fisher Scientific) at a thickness of 30–50 μm and rendered as floating sections. Sections were permeabilized with 0.25 % Triton X100 (Amresco) for 25 minutes and blocked overnight using 5% bovine serum albumin (BSA; GeminiBio). Floating sections in 1X TBS containing Tween-20 (Sigma) (1X TBS-T) and 5% BSA were incubated with primary antibodies at 4°C overnight, followed by fluorochrome-conjugated secondary antibodies for 2 hours at room temperature (Antibodies listed in the Key Resources Table). Washing was performed with 1X TBS-T. Stained sections were mounted onto slides and covered with ProLong^™ Gold Antifade Mountant (Thermo Fisher Scientific). TUNEL staining was performed with cryopreserved tissue sections using the CF 640R TUNEL Assay Apoptosis Detection Kit (Biotium). For positive controls of TUNEL, cryopreserved brain sections were treated with TACS Nuclease (R&D) to induce DNA fragmentation.

Brain imaging and quantification

Slides were imaged using the Andor Dragonfly Spinning Disk Confocal Microscope (Oxford Instruments) and analyzed using FIJI ImageJ⁷⁹ and Imaris (Oxford Instruments). Specific methodology and quantification of images are described in figure legends.

Cell isolation for flow cytometry analysis

To prepare brain samples, tissues within the skull were harvested from mice. Isolated tissues (brain parenchyma including a trace amount of border tissues) were minced for 1 minute using small scissors, digested in 1X PBS containing 5% heat-inactivated fetal bovine serum (FBS, Atlanta Biologicals), 1mM HEPES (Thermo Fisher Scientific, Cat. 15630080), Collagenase Type IV (Roche), and DNase I (Millipore) for 30 minutes at 37°C with constant shaking at 225 rpm. Following digestion, single-cell suspensions were prepared by passing through a 25G needle and filtered through a 70 μm cell strainer. Cells were then resuspended in 38% Percoll (Cytiva) and centrifuged at 532 x g for 30 minutes without brake. Following centrifugation, the lipid and debris layer was aspirated from the top of the tube, and the pelleted brain cells were resuspended in staining buffer. Splenocytes and lung cells were prepared as we previously described^50,80. All cells from tissue were treated with red blood cell lysis (RBC) lysis solution prior to staining for flow cytometry. Freshly isolated cells were used for all flow cytometry analysis, unless otherwise noted.

Antibody staining for flow cytometry

Optimal cell density for antibody staining was determined by counting cells using a hemocytometer. Cells were washed and stained in staining buffer (1X PBS 0.5% BSA). Cells were pre-incubated with FcBlock (CD16/32) for 10 minutes before antibody staining. Antibodies are listed in the Key Resources Table.

Intracellular cytokine staining for flow cytometry

Single cell culture from brains was prepared, and phorbol 12-myristate 13-acetate (PMA) (100 ng/mL) (Sigma Aldrich) and ionomycin (375 ng/mL) (Sigma Aldrich) were added to the culture. After 30 minutes, Brefeldin A (BioLegend) was added, and the cells were incubated for additional 3.5 hours. For intracellular cytokine staining, harvested cells were first stained with fixable LIVE/DEAD^™ (ThermoFisher Scientific), then cell surface proteins were stained with fluorochrome-conjugated antibodies. Antibodies are listed in the Key Resources Table. Next, cells were permeabilized and fixed with the Cyto-Fast^™ Fix/Perm Buffer Set (BioLegend) for intracellular cytokines staining following the manufacturer’s protocol.

Flow cytometry analysis

The following markers were used to identify brain cell populations: neutrophils (CD45^hiCD11b⁺Ly6G⁺), microglia (CD45^loCD11b⁺), activated microglia (CD45^loCD11b⁺CD11c⁺MHCII^hi), myeloid cells (CD45^hiCD11b⁺), dendritic cells (CD45^hiCD11b⁺Ly6G⁻CD11c⁺MHCII⁺, DC), monocytes/macrophages (CD45^hiCD11b⁺Ly6G⁻CD11c⁻, mono/mac), lymphocytes (CD45^hiCD11b⁻), B cells (CD45^hiCD19⁺TCRβ⁻CD11b⁻), T cells (CD45^hiTCRβ⁺CD19⁻CD11b⁻), CD4⁺ T cells (CD45^hiTCRβ⁺CD4⁺CD11b⁻), CD8⁺ T cells (CD45^hiTCRβ⁺CD8⁺CD11b⁻). Analyses were performed using fresh, live cells as negative of DAPI or Fixable Live/Dead staining, along with UltraComp eBeads (Thermo Fisher Scientific), using the BD Fortessa, BD Canto (BD Biosciences), or Cytek Aurora (Cytek Biosciences). Cell sorting was performed using FACS Beckman Coulter Astrios, and post-sorted cell purity was confirmed. FlowJo software (Treestar) was used for data analyses. Mice were not perfused prior to euthanasia for flow cytometry analysis unless otherwise noted.

Ex vivo cell stimulation with yeasts or PRR agonists

Primary cells and cell lines were cultured in cDMEM, supplemented with 10% FBS, 1% penicillin-streptomycin (Thermo Fisher Scientific), 2mM L-glutamine (Thermo Fisher Scientific). BMDMs, BV2, and C8-S were seeded a day before stimulation with indicated fungal yeasts or a PRR ligand, while PMs and primary microglia were seeded for 1–2 hours prior to the stimulation. Cells stimulation was performed for 6 hours in an incubator at 37°C with 5% CO₂. Concentrations of stimulants are described figure legends. Methods to prepare these cell types are as follows: BMDMs were generated by plating total BM cells in complete RPMI1640 media containing rmM-CSF (20 ng/mL, BioLegend) for 6 days, with replacement of medium on day 3 of culture. PMs were collected by peritoneal lavage from naïve mice and enriched using anti-CD11b MicroBeads (Miltenyi Biotec). To collect brain microglia, naïve adult mice were perfused using 1X PBS with 0.1% heparin to remove CD11b⁺ cells in the vasculature, and microglia were enriched using CD11b beads (Miltenyi Biotec).

Primary astrocyte and microglia isolation for evaluating gene expression

Astrocytes and microglia were isolated using beads and FACS, respectively. For astrocytes isolation, we followed a published method⁵³ with a few modifications. Briefly, whole mouse brains were minced and digested using the Papain Dissociation System (Worthington Biochemical Corporation). After myelin removal with Myelin Removal Beads II (Miltenyi Biotec), astrocytes were positively selected using anti-ACSA-2 Microbeads (Miltenyi Biotec). Microglia were sorted as CD45^loCD11b⁺ cells by FACS.

Real-time, quantitative polymerase chain (RT-qPCR) analysis

To evaluate gene expression, total RNA was prepared using TRI Reagent (Millipore Sigma) and reverse-transcribed using qScript cDNA Mix (QuantaBio) to obtain cDNA. qPCR was performed with PowerTrack^™ SYBR Green Master Mix (Applied Biosystems), using primers indicated in Table S4. Actb gene was an internal control, and relative expression for each sample was calculated using the standard -ΔΔCt method⁸¹. Fold changes are compared to the value from the highest-expressed sample or group for each gene.

scRNA-seq library preparation and sequencing

Age-matched female B6 WT mice were retro-orbitally infected with 10⁶ yeasts/mouse of Cryptococcus 52D and brains were collected at 0, 4, and 14-dpi. Samples from two mice were pooled for each time point, except for 14-dpi had data from one mouse. Brain cells were isolated using the Papain Dissociation Kit (Worthington Biochemical Corporation) and myelin was removed using Myelin Removal Beads II (Miltenyi Biotec). Live cells were enriched using the EasySep^™ Dead Cell Removal (Annexin V) Kit (STEMCELL Technologies). All cells isolated from brains were labelled with multiplexing oligos using the CellPlex barcoding kit (10x Genomics) as recommended by the supplier. Following barcoding, samples with the same cell numbers were combined for library preparation and sequencing using the Chromium Next GEM Single Cell 3’ Kit v3.1 (10x Genomics). Libraries were prepared by the Duke Sequencing and Genomic Technologies Shared Resource facility. Sequencing was performed by BGI Genomics using DNBSEQ-G400 sequencer.

scRNA-seq data analyses

Cell Ranger version 4.0.10 (10X Genomics) was used to process raw sequencing files into the fastq format. Reads were aligned with the mouse GRCm39 transcriptome containing all protein-coding and long non-coding RNA genes. Expression counts were processed using Cell Ranger to generate a matrix file for each sample. We combined 6 samples in one sequencing run. On average, we obtained 13,857 median reads per cell, 1,724 median genes per cell, and 3,652 median UMI counts per cell. Using Seurat version 4.3.0⁸², we calculated the percentage of mitochondrial genes, number of expressed genes, and number of counts per cell (Figure S6C). Following initial filtering, 15,966 cells from six samples were used for downstream analysis. Data normalization was performed using the NormalizedData and ScaleData functions in Seurat⁸³. PCA was run and the top 30 principal components were selected for UMAP (Figure 4A).

Clusters were identified by the k-nearest neighbor method and signature markers for each cluster were identified. We used the following abbreviations for cell types: neutrophils (NEUT), mononuclear phagocytes (MNPs), natural killer cells (NK cells), B cells, T cells, immature neurons (ImmN), Schwann cells (SC), oligodendrocyte precursor cells (OPC), oligodendrocytes (OLG), astrocytes (AST), microglia (MG), vascular and leptomeningeal cells (VLMC), vascular smooth muscle cells (VSMC), pericytes (PC), and endothelial cells (EC). Optimal clustering resolution was determined using a Louvain silhouette score-maximization approach⁸⁴. Initial cluster annotation was performed with SingleR to identify cell types. Cluster identities were validated to express canonical cell type-specific markers for each annotated population (Figure 4B). Individual clusters and timepoints were then analyzed, as described in this study. Microglia, astrocytes, and oligodendrocytes were subset and re-normalized as above to identify within cell-type differences in expression.

Differential gene expression was performed using the Seurat package FindMarkers function with the default Wilcoxon rank sum test and Bonferroni adjusted p-value. Intercellular communication networks between cell types were inferred, analyzed, and visualized using the CellChat package (Version 2)⁸⁵. Genes with adjusted p-values < 0.05 were selected for plotting of the normalized enrichment score and ranked by logFC. Gene-set Enrichment Analysis (GSEA) was performed using ranked gene lists. Hallmark gene sets from the Molecular Signatures Database (MSigDB) were used for pathway enrichment analysis⁸⁶. Correction for multiple comparisons was performed using the BH method, and pathways with an adjusted p-value < 0.05 were selected for plotting of the normalized enrichment score.

In the heatmap showing communication results among cell types at 4-dpi and 14-dpi compared to on 0-dpi, the y-axis denotes the “source” cell type while the x-axis denotes the “target” cell type. The bar plots in the “source” column represent the total absolute value of outgoing signals from the “source” cell type, while the top bar plots indicate the total absolute value of signals received by the “target” cell type. Taller bars indicate greater overall signaling changes to or from the cell type between timepoints. Red indicates increased signaling in the second dataset compared to the first dataset, while blue indicates decreased signaling. Intercellular communication networks between cell types were inferred, analyzed, and visualized using CellChat (Version 2).

Data comparison between DAM and microglia from cryptococcal meningoencephalitis (CM-MG)

DAM DEGs data were obtained from publicly available data (GSE98969)²¹ (Table S1) and were compared to total DEGs in CM-MG at 14-dpi from our analysis (Table S2). To generate a gene expression plot comparing two DEG datasets, gene expression was normalized by setting the fold increase of maximally upregulated and downregulated genes as 1 and −1, respectively (Table S3).

QUANTIFICATION AND STATISTICAL ANALYSIS

Data was analyzed and graphed using GraphPad Prism software (version 10.1). Statistical methods are indicated in the figure legends for each experiment in this study.

Supplementary Material

SupplTable-1

Table S1. Unique DEGs in DAM used for comparison to DEGs in CM-MG (reanalysis of publicly available data set GEO: GSE98969), Related to Figure 5.

SupplTable-2

Table S2. Unique DEGs in CM-MG at 14-dpi, Related to Figure 5.

SupplTable-3

Table S3. Comparison of normalized DEG, DAM vs. CM-MG, Related to Figure 5.

Document S1. Figures S1-S9, Table S4. Primer Sequences for RT-qPCR