A hyper-immunogenic and slow-growing fungal strain induces a murine granulomatous response to cryptococcal infection

Many successful pathogens cause latent infections, remaining dormant within the host for years but retaining the ability to reactivate to cause symptomatic disease. The human opportunistic pathogen Cryptococcus neoformans is a ubiquitous yeast that establishes latent pulmonary infections in immunocompetent individuals upon fungal inhalation from the environment. These latent infections are frequently characterized by granulomas, or foci of chronic inflammation, that contain dormant cryptococcal cells. Immunosuppression causes these granulomas to break down and release viable fungal cells that proliferate, disseminate, and eventually cause lethal cryptococcosis. This course of C. neoformans dormancy and reactivation is understudied due to limited models, as chronic pulmonary granulomas do not typically form in most mouse models of cryptococcal infection. Here, we report that a previously characterized Cryptococcus-specific gene which is required for host-induced cell wall remodeling, MAR1, inhibits murine granuloma formation. Specifically, the mar1Δ loss-of-function mutant strain induces mature pulmonary granulomas at sites of infection dormancy in mice. Our data suggest that the combination of reduced fungal burden and increased immunogenicity of the mar1Δ mutant strain stimulates a host immune response that contains viable fungi within granulomas. Furthermore, we find that the mar1Δ mutant strain has slow growth and hypoxia resistance phenotypes, which may enable fungal persistence within pulmonary granulomas. Together with the conventional primary murine infection model, latent murine infection models will advance our understanding of cryptococcal disease progression and define fungal features important for persistence in the human host.

suggest that the combination of reduced fungal burden and increased immunogenicity 50 of the mar1 mutant strain stimulates a host immune response that contains viable 51 fungi within granulomas. Furthermore, we find that the mar1 mutant strain has slow 52 growth and hypoxia resistance phenotypes, which may enable fungal persistence within 53 pulmonary granulomas. Together with the conventional primary murine infection model, 54 latent murine infection models will advance our understanding of cryptococcal disease 55 progression and define fungal features important for persistence in the human host. strain stimulates a host immune response that contains mar1 mutant cells within well-112 circumscribed granulomas. From the host perspective, we find that host GM-CSF 113 signaling, a known contributor to granuloma formation (17,(24)(25)(26), is required for the 114 formation of these granulomas. Finally, in vitro studies demonstrate that the mar1 115 mutant strain has cell cycle defects that may contribute to a slow growth phenotype and 116 hypoxia resistance, two features which likely enable cryptococcal persistence within 117 pulmonary granulomas. Because MAR1 is a Cryptococcus-specific gene, this model 118 represents a unique addition to the limited tools available to study the reactivation 119 model of cryptococcal disease. 120

Macrophage activation analyses 211
Intracellular staining of markers of macrophage activation was performed as 212 described previously (29). Leukocytes isolated from infected mice as described above 213 were incubated with cell stimulation cocktail (eBioscience Cat. # 00-4970-03) according 214 to the manufacturer's recommendation and incubated at 37°C in 5% CO 2 in cRPMI for 215 BD Biosciences) was added according to the manufacturer's recommendations and 217 incubated for an additional four hours (6 hours total). Cells were washed with PBS and 218 stained with yellow Zombie viability dye in PBS at room temperature in the dark for 219 15 minutes. Cells were then washed with FACS buffer and incubated with Fc block (BD 220 Biosciences) diluted in FACS buffer for 5 minutes. For nitric oxide (iNOS) and Arginase 221 1 (Arg1) production in macrophages, cells were stained for surface markers CD45, 222 CD11b, CD64, F4/80, and CD24, and incubated at 4°C for 30 minutes. Cells were then 223 washed and fixed with 2% ultra-pure formaldehyde (Polysciences, Warrington, PA) for 224 20 minutes. Subsequently, cells were washed with 0.1% saponin buffer and stained with 225 antibodies for iNOS and Arg1 for 30 minutes at 4°C. Finally, cells were washed with 226 saponin buffer and fixed with 2% ultra-pure formaldehyde. Samples were processed 227 using a Cell Analyzer LSRII (BD Biosciences) using BD FACSDiva v8.0 software at the 228 UNTHSC Flow Core, and 100,000 events were collected for analysis using FlowJo 229 v.10.8 Software. Statistical significance between strains at each timepoint was 230 determined using Student's t test (GraphPad Software, San Diego, CA). 231

Titan cell assay and quantification 232
A previously described in vitro titanization assay was used here (30). In brief, the 233 WT (H99), the mar1Δ mutant (MAK1), and the mar1Δ + MAR1 (MAK11) strains were 234 incubated for 18 hours at 30°C, 150 rpm in 5 mL yeast nitrogen base (YNB) without 235 amino acids prepared according to the manufacturer's instructions plus 2% glucose. 236 Cultures were washed six times with PBS. An optical density at 600 nm (OD 600 ) of 0.001 237 for each strain was transferred to 5 mL 10% heat-inactivated fetal bovine serum (HI-238 FBS) in PBS and incubated at 37°C, 5% CO 2 for 96 hours. Cells were imaged by 239 differential interference contrast (DIC) microscopy using a Zeiss Axio Imager A1 240 microscope equipped with an Axio-Cam MRm digital camera. Cell diameter was 241 measured using the ImageJ software (FIJI), and cells with a diameter > 10 μm were 242 considered Titan cells. A minimum of 400 cells were analyzed across three biological 243 replicates for each fungal strain. Statistical significance was determined using one-way 244 analysis of variance (ANOVA) and the Tukey-Kramer test (GraphPad Software, San 245 Diego, CA). 246

Cellular morphology defect quantification 260
The WT (H99), the mar1Δ mutant (MAK1), and the mar1Δ + MAR1 (MAK11) 261 strains were incubated for 18 hours in YPD medium at 30°C with shaking at 150 rpm. An 262 OD 600 of approximately 0.2 for each strain was transferred to fresh YPD medium and 263 subsequently incubated at either 30°C or 37°C for 18 hours with shaking at 150 rpm. 264 Cells were then pelleted, washed with PBS, and imaged by differential interference 265 contrast (DIC) microscopy. DIC images were captured using a Zeiss Axio Imager A1 266 microscope equipped with an Axio-Cam MRm digital camera. A minimum of 500 cells 267 were analyzed across three biological replicates for each strain using the ImageJ 268 software (FIJI). Statistical significance was determined using two-way analysis of 269 variance (ANOVA) and the Tukey-Kramer test (GraphPad Software, San Diego, CA). 270

Growth curve analysis 271
The WT (H99), the mar1Δ mutant (MAK1), and the mar1Δ + MAR1 (MAK11) 272 strains were incubated for 18 hours in YPD medium at 30°C with 150 rpm shaking. 273 Cultures were normalized to an OD 600 of 0.01 in fresh YPD medium and added to wells 274 of a 96-well plate. Growth was then measured at an absorbance of 595 nm every 10 275 minutes for 40 hours with shaking between readings and incubation at 37°C. Control 276 wells containing YPD medium alone were also included to eliminate any background 277 absorbance. 278

Hypoxia resistance analyses 279
The WT (H99), the mar1Δ mutant (MAK1), the mar1Δ + MAR1 (MAK11), and the 280 sre1Δ mutant (HEB6) strains were incubated in YPD medium at 30°C until mid-281 logarithmic growth phase. Strains were washed once in PBS, normalized to an OD 600 of 282 0.6 in PBS, and serially diluted onto YES (0.5% [w/v] yeast extract, 2% glucose, and 283 225 µg/mL uracil, adenine, leucine, histidine, and lysine) medium agar plates with or 284 without cobalt chloride (0.7 mM) (31). Microaerophilic conditions were generated using 285 a sealed chamber (BD GasPak TM ) and two activated GasPak TM EZ Campy Container 286 System sachets (31). Plates were placed in the chamber (microaerophilic) or outside 287 the chamber (ambient air), incubated at 30°C, and imaged daily for 96 hours. 288 Mouse isolate recovery and phenotypic characterization 289 C57BL/6 female mice acquired from Charles River Laboratories were infected as 290 described above. At 61 DPI and 100 DPI, mice were sacrificed by CO 2 inhalation 291 followed by an approved secondary method of euthanasia and fungi were subsequently 292 isolated from the lungs as described above. Single fungal colonies were plated onto 293 YPD agar medium and subsequently frozen in separate wells of 96-well plates at -80°C. 294 Isolated fungi were stamped onto YPD agar medium incubated at 30°C, YPD agar 295 medium incubated at 37°C, YPD agar medium supplemented with nourseothricin (NAT) 296 (100 µg/mL) incubated at 30°C, and YPD agar medium buffered (150 mM HEPES) to 297 pH 8.15 incubated at 30°C. All plates were imaged daily. Mouse isolates were 298 determined to be mar1Δ mutant strain isolates based on growth on YPD + NAT medium 299 and dry colony morphology on YPD pH 8.15 medium (23). The original WT (H99) and 300 mar1Δ mutant (MAK1) strains were included on each plate as controls. 301

Ethical use of animals 302
All animal experiments in this manuscript were approved by the University of 303 Texas at San Antonio Institutional Animal Care and Use Committee (IACUC) (protocol 304 #MU021), the Texas Christian University and the University of North Texas Health 305 Sciences Center (UNTHSC) IACUC (protocol #1920-9), and the Duke University IACUC 306 (protocol #A102-20-05). Mice were handled according to IACUC guidelines. 307

Data availability 308
All fungal strains and reagents are available upon request. 309

RESULTS 311
Pulmonary granulomas are formed and maintained in mice infected with the 312 mar1 mutant strain. 313 Based on our recent observations that the mar1 mutant strain displays a highly 314 immunogenic cell surface, we hypothesized that the mar1 mutant strain would have 315 unique interactions with the host in vivo. We previously observed that the mar1 mutant 316 strain is hypovirulent compared to the wild-type (WT) strain in a murine inhalation model 317 of cryptococcosis (23). Highly immunogenic fungal strains often induce a 318 hyperinflammatory response that is detrimental to the host, resulting in hypervirulence 319 (32)(33)(34). We therefore explored in greater detail the mechanisms by which the highly 320 immunogenic mar1 mutant strain simultaneously activates and is controlled by the 321 host immune response. 322 As an initial investigation into the interactions between the mar1 mutant strain 323 and the host, we assessed the gross appearance of infected lungs from our previously 324 reported mar1 mutant strain murine inhalation infection experiment. At the time of 325 sacrifice, generally between 24-40 days post-inoculation (DPI), we observed that the 326 lungs of mar1-infected C57BL/6 mice displayed large, well-circumscribed inflammatory 327 foci surrounded by healthy-appearing lung tissue ( Figure 1A). This contrasts starkly with 328 WT-infected lungs, which typically exhibit uncontrolled fungal proliferation accompanied 329 by a diffuse inflammatory response. 330 We examined histopathological features of infected murine lungs at specific 331 timepoints throughout the course of infection to further characterize the unique 332 pathology observed in mar1-infected lungs. To do so, we replicated the experimental 333 approach used in Figure 1A; we inoculated C57BL/6 mice by inhalation with the WT immune response at timepoints relevant to granuloma formation. To do so, we 387 replicated the experimental approaches used in Figure 1; we inoculated C57BL/6 mice 388 by inhalation and harvested lungs for analysis throughout infection. We previously 389 reported a decrease in fungal burden in mar1-infected lungs compared to WT-infected 390 lungs as early as 1 and 4 DPI, despite identical doses being used for both strains (23). 391 In this work, at all tested timepoints (3, 7, 14, & 21 DPI), we find that mar1-infected 392 lungs have a significantly reduced fungal burden compared to WT-infected lungs. 393 Specifically, the mar1-infected lungs have a 10-fold reduction in fungal burden at 3 394 DPI, a 100-fold reduction in fungal burden at 7 DPI, and a >500-fold reduction in fungal 395 burden at 14 and 21 DPI compared to WT-infected lungs (Figure 2). These observations 396 support the reduced number of mar1 mutant cells observed at these same timepoints 397 in our histopathology analyses ( Figure 1B). As a result of the drastic reduction in 398 pulmonary fungal burden throughout infection, we observed that the mar1 mutant 399 strain rarely disseminates to the brain (Figure 2). When the mar1 mutant strain does 400 disseminate to the brain, the fungal burden is markedly lower than that of the WT strain 401 ( Figure 2). Together, these observations indicate that the mar1 mutant strain has 402 reduced fungal burden in the murine lung and brain, reinforcing our previous reports 403 that the mar1 mutant strain has reduced fitness in host-relevant conditions. 404 Based on the drastic differences in fungal burden observed between WT-infected 405 and mar1-infected lungs, we hypothesized that the immune microenvironment within 406 the lungs would also differ significantly. We replicated the experimental approaches 407 used in Figure 1; we inoculated C57BL/6 mice by inhalation and harvested lungs for Granuloma formation is dependent on GM-CSF signaling in the context of both 447 mycobacterial (24-26) and cryptococcal infections (17). GM-CSF is the only cytokine 448 that showed significant differential production in our cytokine analyses. Specifically, we 449 observed that the mar1 mutant strain induces more pulmonary GM-CSF production 450 than the WT strain at 3 DPI ( Figures 3A & S2). We therefore hypothesized that GM-CSF 451 signaling would also be required for the formation of pulmonary granulomas in our we observed mar1 mutant cell morphology during logarithmic growth phase. When 516 incubated at the permissive temperature of 30°C, the mar1 mutant strain displays an 517 increased incidence of cytokinesis defects (such as elongated cells, cells with wide bud 518 necks, and cells that fail to complete cytokinesis), compared to both the WT strain and 519 the mar1 + MAR1 complemented strain ( Figure 6A). The frequency of these 520 cytokinesis defects is significantly enhanced at the physiological temperature of 37°C 521 ( Figure 6A). We next determined the impact of these defects on the growth kinetics of 522 the mar1 mutant strain. We observed that the mar1 mutant strain displays a 523 reduction in growth during logarithmic phase at 37°C, compared to both the WT strain 524 and the mar1 + MAR1 complemented strain ( Figure 6B). These data demonstrate that 525 the mar1 mutant strain has a slow growth phenotype at the physiological temperature 526 of 37°C that is likely driven in part by cytokinesis defects. 527 Cell cycle regulation is also known to be related to fungal adaptation to hypoxia 528 (42-44). Because C. neoformans is an obligate aerobe, WT fungal cells undergo G 2 -529 arrest in response to hypoxia (45, 46). We assessed the ability of the mar1 mutant 530 strain to grow in an environment with reduced oxygen availability by observing growth in 531 the presence of CoCl 2 and in a microaerophilic chamber. In both cases, we observed 532 that the mar1 mutant strain displays enhanced growth compared to the WT strain and 533 the mar1 + MAR1 complemented strain ( Figure 6C). In these assays, the CoCl 2 -and 534 hypoxia-sensitive sre1 mutant strain was used as a control (31) ( Figure 6C). 535 Collectively, these observations suggest that the cell cycle defects of the mar1 mutant 536 strain may contribute to its ability to survive, slowly proliferate, and persist in the murine 537 granuloma environment. 538

DISCUSSION 540
Here, we report and characterize the host response to a chronic and indolent C. 541 neoformans lung infection, one distinguished by sustained granulomas. Using the 542 inhalation route of infection in C57BL/6 mice, we observe granuloma formation in 543 infections due to both the WT and mar1 mutant strains. However, the appearance, 544 development, and maintenance of these granulomas differ significantly. In WT 545 infections, small, immature granulomas form early in infection. As infection progresses, 546 these nascent granulomas begin to degenerate, leading to fungal proliferation 547 throughout the lungs, fungal dissemination to the brain, and eventually murine death. 548 This type of early, immature granuloma formation has been observed previously in 549 murine infections with other C. neoformans WT strains (16,17). In contrast, in mar1 550 mutant strain infections we observe mature pulmonary granulomas that develop over 551 several weeks in the absence of overt clinical symptoms. These granulomas differ from 552 the WT-induced granulomas because they appear later in infection, are typically larger, 553 and are more contained. The containment of these granulomas may be expected 554 because mar1-induced granulomas are associated with a significantly lower fungal 555 burden compared to WT strain infections, suggesting that the granulomas effectively 556 inhibit fungal proliferation throughout the lungs. Despite this drastic reduction in fungal 557 burden, the mar1 mutant strain induces a comparable pulmonary cytokine and 558 leukocyte response to that of the WT strain during early stages of infection. Previous 559 work reported by our group characterized the mar1 mutant strain as more 560 immunogenic than the WT strain, due to its poorly organized cell wall and impaired 561 polysaccharide capsule attachment (23). We posit that the combination of reduced 562 fungal burden and increased immunogenicity drives mar1-induced granuloma 563 formation: the increased immunogenicity results in an immune response that contains 564 the reduced number of mar1 mutant cells within granulomas during early stages in 565 infection. 566 We further observe that mar1-induced granulomas are maintained throughout 567 infection, from 14 DPI to as late as 100 DPI. We find that the immune microenvironment 568 associated with these granulomas has significantly reduced cytokine and leukocyte 569 responses. Previous work has implicated classically-activated macrophage polarization 570 in enhanced antifungal activity of macrophages (47-49). We find that mar1-infected 571 lungs have a comparable number of or fewer (depending on the timepoint) classically-572 activated (M1) and alternatively-activated (M2) macrophages compared to WT-infected 573 lungs, suggesting that differential polarization of macrophages does not contribute to 574 the reduced fungal burden and associated immune response in mar1-infected lungs. 575 Collectively, these observations demonstrate that mar1-induced granulomas are 576 largely a fungal-driven phenomenon, with the sustained reduction in mar1 mutant 577 strain fungal burden resulting in a dampened immune response compared with WT-578 infected lungs. Using these approaches, we have defined a detailed timeline of 579 granuloma formation, in both WT and mar1 mutant strain infections, and characterized 580 multiple fungal factors that contribute to granuloma formation ( Figure S5). 581 In addition to the fungal drivers of mar1-induced granuloma formation described 582 above, we have also confirmed the role of GM-CSF as a host driver of cryptococcal 583 granuloma formation. From our pulmonary cytokine analyses, we observed that GM-584 CSF is the only differentially produced cytokine in mar1-infected lungs compared to 585 WT-infected lungs. Specifically, GM-CSF is elevated in mar1-infected lungs at 3 DPI, 586 an early timepoint in infection at which the pulmonary immune response is being 587 actively developed. This increased GM-CSF production may be a result of increased 588 Dectin-1 activation by the mar1 mutant strain. We previously reported that the mar1 589 mutant strain is partially recognized by the pathogen recognition receptor Dectin-1, 590 likely due its increased exposed surface β-glucan and chitin (23). Dectin-1 has been 591 shown to be required for normal GM-CSF production in murine macrophages (50). 592 Additionally, GM-CSF production is known to result in an increase in Dectin-1 593 expression by murine macrophages (50, 51). We also report that granuloma formation 594 significance between strains at each timepoint was determined using Student's t test (*, 944 P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; no designation between strains, 945 not significant). Only a subset of data is shown; refer to Figure S2 for full analysis. B. 946 Pulmonary leukocyte infiltrates of female C57BL/6 mice inoculated with 1 x 10 4 cells of 947 the WT strain or the mar1 mutant strain were measured by flow cytometry throughout 948 infection: 1, 3, 7, and 21 DPI. Data shown are the mean ± of absolute cell numbers from 949 three independent experiments (n = 3) performed using five mice per group per 950 timepoint per experiment. Error bars represent SEM. Statistical significance between 951 strains at each timepoint was determined using Student's t test (*, P < 0.05; no 952 designation between strains, not significant). Only a subset of data is shown; refer to 953