O-Mannosylation of Proteins Enables Histoplasma Yeast Survival at Mammalian Body Temperatures

ABSTRACT The ability to grow at mammalian body temperatures is critical for pathogen infection of humans. For the thermally dimorphic fungal pathogen Histoplasma capsulatum, elevated temperature is required for differentiation of mycelia or conidia into yeast cells, a step critical for invasion and replication within phagocytic immune cells. Posttranslational glycosylation of extracellular proteins characterizes factors produced by the pathogenic yeast cells but not those of avirulent mycelia, correlating glycosylation with infection. Histoplasma yeast cells lacking the Pmt1 and Pmt2 protein mannosyltransferases, which catalyze O-linked mannosylation of proteins, are severely attenuated during infection of mammalian hosts. Cells lacking Pmt2 have altered surface characteristics that increase recognition of yeast cells by the macrophage mannose receptor and reduce recognition by the β-glucan receptor Dectin-1. Despite these changes, yeast cells lacking these factors still associate with and survive within phagocytes. Depletion of macrophages or neutrophils in vivo does not recover the virulence of the mutant yeast cells. We show that yeast cells lacking Pmt functions are more sensitive to thermal stress in vitro and consequently are unable to productively infect mice, even in the absence of fever. Treatment of mice with cyclophosphamide reduces the normal core body temperature of mice, and this decrease is sufficient to restore the infectivity of O-mannosylation-deficient yeast cells. These findings demonstrate that O-mannosylation of proteins increases the thermotolerance of Histoplasma yeast cells, which facilitates infection of mammalian hosts.

IMPORTANCE For dimorphic fungal pathogens, mammalian body temperature can have contrasting roles. Mammalian body temperature induces differentiation of the fungal pathogen Histoplasma capsulatum into a pathogenic state characterized by infection of host phagocytes. On the other hand, elevated temperatures represent a significant barrier to infection by many microbes. By functionally characterizing cells lacking O-linked mannosylation enzymes, we show that protein mannosylation confers thermotolerance on H. capsulatum, enabling infection of mammalian hosts. KEYWORDS Histoplasma, glycosylation, mannose, phagocyte, thermotolerance F or dimorphic fungal pathogens, temperature provides an important cue for both fungal morphology and lifestyle. Elevated temperatures present a restrictive barrier to microbial growth, and mammalian body temperatures have been postulated as one reason for the rarity of fungal pathogens able to cause disease in humans (1, 2). However, for thermally dimorphic fungal pathogens, mammalian body temperature also serves as a cue triggering differentiation into a state better adapted for infection. At lower temperatures (e.g., Ͻ30°C), Histoplasma grows as saprobic hyphae. At elevated temperatures (e.g., 37°C), Histoplasma differentiates into a pathogenic yeast. The ability to transition to the yeast phase and the expression of yeast-phase genes are critical for Histoplasma virulence (3,4). Consequently, a large focus has been on the identification and characterization of yeast phase-specific factors that might facilitate Histoplasma pathogenesis (5)(6)(7)(8)(9)(10)(11)(12). Many of the extracellular proteins produced by Histoplasma yeast cells, but not mycelia, are heavily glycosylated (8), suggesting that glycosylation is important for virulence.
In fungi, mannosylation of proteins is the predominant form of both N-linked and O-linked glycosylation (13). N-linked glycosylation is characterized by the attachment of a large branched glycan structure to the asparagine residue of an N-X-S/T motif of substrate proteins. O-linked glycosylation in fungi is characterized by the attachment of mannose to a serine or threonine residue on substrate proteins, with additional saccharide extension into a linear mannan chain. For N-linked glycans, the extensive branching requires numerous enzymes to build the full glycan (13), while the linear O-linked mannan requires only two or three enzyme families, one for initial mannose attachment and, depending on the organism, one or two for mannan extension (13). For O-linked mannosylation, the initial mannose is attached to the protein by the protein mannosyltransferase (Pmt) family of proteins, with subsequent extension carried out by Ktr and Mnn1 family proteins (Saccharomyces cerevisiae) (14) or the Mnt family (Candida albicans) (15,16).
Given the correlation of protein glycosylation with the pathogenic phase of Histoplasma, we investigated the functional role of O-linked mannosylation in Histoplasma virulence. The Histoplasma genome contains genes encoding three Pmt family proteins (Pmt1, Pmt2, and Pmt4), as well as a single Mnt1 homologue. In this study, we characterize Histoplasma yeast cells deficient in these glycosylation enzymes. Yeast cells depleted of O-linked mannosylation are viable but are less tolerant of elevated temperatures. We show that these yeast cells have altered cell walls, which changes recognition by immune receptors. In vivo, glycosylation-deficient yeast cells are rapidly cleared from the lungs after infection because of loss of the ability to survive at mammalian body temperatures. These findings indicate that O-mannosylation of proteins contributes to Histoplasma's ability to thrive in a mammalian host by increasing the thermotolerance of pathogenic yeast cells.

RESULTS
Identification of Histoplasma protein mannosyltransferases. The correlation of extracellular protein glycosylation and Histoplasma growth at mammalian body temperature suggests a link between virulence and glycoprotein production (8). To begin investigations into the functional role of the O-glycosylation pathway in Histoplasma, we identified three protein-mannosyltransferase homologues (Pmt1, Pmt2, and Pmt4; see Fig. S1A in the supplemental material) and an Mnt1 homologue in the Histoplasma transcriptome. Despite the enriched protein glycosylation that characterizes yeast cells, PMT genes are transcribed at similar levels in yeast cells and mycelia (17). To probe the function of these enzymes, we isolated a PMT2 insertional mutant (T-DNA insertion between the third and fourth exons of the PMT2 gene) and also depleted functions by RNA interference (RNAi). RNAi-based knockdown of each identified gene (via sentinel gfp RNAi reduction [18,19]) indicates at least 90% knockdown of each PMT target (Fig. S1B). No major compensatory changes in transcript expression of PMT1 or PMT4 was found in the absence of PMT2 (Fig. S1C), suggesting that the lack of one PMT is not compensated for by another, consistent with studies of other fungi (20). Complementation of the pmt2 mutant with a DNA construct consisting of the native promoter and the PMT2 gene restored PMT2 expression (Fig. S1C).
Histoplasma protein mannosyltransferases glycosylate extracellular proteins. To confirm the function of Histoplasma Pmt proteins in protein glycosylation, secreted proteins were examined for the loss of protein glycosylation by reduction of protein molecular mass. The glycoprotein Cfp4 (8,21) and ␤-glucanases Eng1 and Exg8 (8) contain mucin-like domains in the primary amino acid sequence, suggestive of regions potentially modified by O-linked mannosylation (22). Consistent with O-mannosylation of Cfp4, the electrophoretic mobility of the Cfp4 protein is reduced by approximately 2 kDa when produced by Pmt2-deficient yeast cells, and the higher molecular mass is restored by PMT2 complementation (Fig. 1A). Histoplasma Eng1 and Exg8 also have increased electrophoretic mobility in the absence of Pmt2 function (23). We also examined the mobility of two proteins that naturally lack a mucin-like region, secreted proteins Sod3 (12) and Cbp1. Consistent with the lack of O-mannosylation, loss of Pmt2 function did not decrease the molecular mass of either Sod3 (Fig. 1B) or Cpb1 (data not shown). Together, these data show that Pmt2 is necessary for modification of extracellular proteins containing a mucin-like domain.
As further evidence that Pmt2 glycosylates protein, we quantified the saccharide content of extracellular culture filtrate proteins. The protein-associated saccharides from Pmt2-expressing and Pmt2-deficient yeast cells were hydrolyzed into monomeric sugars (e.g., glucose, mannose, and galactose) by acid hydrolysis and quantified by gas chromatography-mass spectrometry (GC-MS). Consistent with Pmt2 catalyzing O-linked mannosylation, there was a significantly smaller amount of mannose (72% reduction) on extracellular proteins from the pmt2 mutant than on those from the wild type (Fig. 1C). Additionally, loss of Pmt2 caused a 60% reduction in protein-associated galactose. Complementation of the pmt2 mutant restores both saccharides to wild-type levels.
Loss of Pmt2 function reduces the glycosylation of Cfp4. O-linked glycans were removed from purified Cfp4 by beta elimination, and the liberated glycans were characterized by matrix-assisted laser desorption ionization (MALDI) MS ( Fig. 1D and E). Spectra for one, two, and three hexose units were identified, and quantitation of the peaks relative to an internal standard showed that glycan levels were 53.9, 73.4, and 7.7% lower, respectively, on Cfp4 protein purified from the pmt2 mutant than on those from the wild type. As further evidence, electrospray ionization (ESI) MS of the Cfp4 tryptic peptides revealed the mucinregion peptide with variable hexoses (due to glycan fragmentation by ionization) (Fig. S2). The total hexose content on the Cfp4 protein isolated from culture filtrates from Pmt2deficient yeast cells was lower (23 hexoses; Fig. S2B) than that on Cfp4 from Pmt2expressing yeast cells (28 hexoses, Fig. S2A). Together, these data confirm that Pmt2 functions in O-glycosylation of extracellular proteins.
Loss of O-linked mannosylation alters the yeast surface and yeast recognition by lectin receptors. Since O-linked mannosylation has the potential to alter cell wall proteins, the consequences of the loss of mannosylation for cell wall structure and function were examined. To determine the impact of mannosylation on the Histoplasma cell wall, we calculated the 50% inhibitory concentrations (IC 50 s) of the cell wall-binding dyes Uvitex and Congo red (Table S2). Although the pmt2 mutant, deficient in O-linked mannosylation, was more susceptible to both compounds, the increase in sensitivity was relatively minor (1.5-fold reduction). Furthermore, the IC 50 s of other agents that induce cell wall stress (NaCl and sodium dodecyl sulfate [SDS]) were not affected by the loss of Pmt2 function (Table S2). No major structural defects in the electron-translucent cell wall were detected by electron microscopy of Pmt2-expressing ( Fig. 2A) and Pmt2-deficient yeast cells (Fig. 2B) grown in broth culture. Nonetheless, Pmt2-deficient yeast cells consistently showed increased electron density along the periphery of the cell wall, suggesting that Pmt2-dependent mannosylation alters the surface of the cell wall.
Given the important interaction of Histoplasma yeast cells with phagocytic cells and the altered cell wall in the absence of O-linked mannosylation, the recognition of yeast cells by immune receptors was tested. Dectin-1 (CLEC7A) and the macrophage mannose receptor (MR/CD206/MRC1/CLEC13D) are two of the major receptors on phagocytes that act as lectins to recognize fungal ␤-glucans and mannan, respectively (24,25). To determine if the altered cell surface of Pmt2-deficient yeast cells results in altered recognition by MR, yeast cells were added to human monocyte-derived macrophages (MDMs) with or without small interfering RNA (siRNA)-based depletion of MR (Fig. 2C). While there were no changes in binding to MDMs expressing MR (scramble siRNA), silencing of MR expression decreased the binding of pmt2 mutant yeast 3.5-fold, demonstrating increased recognition of Pmt2-deficient yeast by MR. To test the rec-   Pmt2-dependent size differences in Cfp4 (A) but not Sod3 (B) proteins were determined by immunoblotting of culture filtrates from wild-type (PMT2), Pmt2-deficient (pmt2), and complemented (pmt2/ PMT2) Histoplasma yeast cells following the removal of N-glycans by PNGase F treatment. The molecular masses (in kilodaltons) of protein standards are indicated on the left of the immunoblots. (C) Monosaccharide contents of glycans associated with culture filtrate proteins quantified by GC-MS. The ratio of the abundance of each sugar was calculated relative to the mean value of glucose. Error bars represent the standard deviations, and asterisks indicate statistically significant differences between Pmt-expressing and Pmt2-deficient strains determined by Student t test (n ϭ 3; **, P Ͻ 0.01; n.s., not significant). (D and E) Direct measurement of O-linked glycosylation of Cfp4 was determined by quantification of hexose oligomers released from purified Cfp4 by beta elimination and detection of ions corresponding to chains of hexose units by MALDI MS. Shown are the m/z values of the major peak and the corresponding hexose chains (Hex n ). Oligosaccharide ion intensities of Cfp4 from Pmt2-expressing (D) and Pmt2-deficient (E) yeast cells were compared after normalization to an internal ribose standard (pentose). Oligosaccharide abundance is shown as a percentage of that of the wild type. ognition of yeast cells by Dectin-1, yeast cells were added to Dectin-1-expressing fibroblasts. In this system, where only a single immune receptor is expressed, recognition of Pmt2-deficient yeast cells is 10-fold lower than that of yeast cells expressing Pmt2 (Fig. 2D). Competitive inhibition with soluble ␤-glucan eliminates the binding of Pmt2-deficient yeast cells, showing that binding is due to the recognition of cell wall ␤-glucans. Together, these data indicate that O-linked mannosylation alters yeast cell recognition by individual host lectin-type receptors; however, there is no overall change in the total association of yeast cells with phagocytes ( Fig. 2C, scramble siRNA cells; data not shown). Association assays were performed in the absence (red) or presence (blue) of 1 mg/ml laminarin to show the specificity of the interaction for Dectin-1 recognition of ␤-glucans. Associated yeast CFU counts were normalized to the average association of wild-type yeast cells with scramble siRNA-transfected MDMs (C) or binding of wild-type yeast cells to Dectin-1 in the absence of laminarin (D). Error bars represent the standard deviation of replicate assays (n ϭ 3), and asterisks indicate statistically significant differences determined by Student t test (n ϭ 3; **, P Ͻ 0.01; n.s., not significant).
Histoplasma Mannosylation Confers Thermotolerance ® Histoplasma survival in vivo requires protein mannosylation. To determine the functional role of O-linked mannosylation in the pathogenesis of Histoplasma, we examined the virulence of Histoplasma yeast cells lacking Pmt functions in a sublethal model of respiratory histoplasmosis. In contrast to wild-type yeast cells, whose lung fungal burden increased 20-fold over 3 days of infection, Pmt2-deficient Histoplasma yeast cells were strikingly reduced to levels below that of the inoculum (Fig. 3A). RNAi-based depletion of Pmt1 and Pmt2 similarly reduced lung infection 16-fold; however, depletion of Pmt4 had no effect on virulence (Fig. 3B). Depletion of Mnt1, which acts downstream of Pmt functions, caused a nearly identical reduction (16-fold) in lung fungal burdens, providing further evidence that O-mannosylation of proteins is required for Histoplasma pathogenesis (Fig. 3B).
To gain insight into the nature of the attenuation of Histoplasma Pmt2-deficient yeast cells early in infection, we investigated the phagocytes associated with Histoplasma yeast cells in the lungs. Mice were infected with fluorescently labeled Histoplasma yeast cells, and the lung phagocyte populations with fluorescent yeast cells  were characterized by flow cytometry at 24 and 48 h postinfection. After 24 h of infection, Ͼ90% of the yeast cells were associated with alveolar macrophages, with a small amount of yeast cells associated with neutrophils ( Fig. 3C and D). Pmt2expressing and Pmt2-deficient yeast cells showed similar phagocyte association profiles (although there was a 3-fold increase in the number of neutrophils associated with pmt2 mutant yeast cells, the difference did not reach statistical significance; Fig. 3D). At this early time point, yeast cells were not significantly associated with monocytes, conventional dendritic cells, or monocyte-derived dendritic cells (i.e., Ly6G Ϫ SiglecF Ϫ cells; Fig. 3C). At 48 h postinfection, similar trends were found. The 3-fold increase in neutrophils associated with pmt2 mutant yeast cells was statistically significant; however, the total neutrophil population increased 4-fold at 48 h ( Fig. S3) with no change in the number of yeast-associated neutrophils. These data indicate that loss of O-linked mannosylation does not substantially alter the association of yeast cells with the major phagocytic cell populations in the lung.

Control of Pmt2-deficient yeast cells in vivo is independent of macrophages and neutrophils.
Since the host cells associated with Histoplasma yeast cells in vivo were macrophages and neutrophils and O-linked mannosylation affects immune receptor recognition of yeast cells, we tested whether the lack of Pmt-dependent mannosylation affected the survival of yeast cells in these cell types. Surprisingly, Pmt2-deficient yeast cells are as viable as wild-type yeast cells following infection of either primary macrophages or neutrophils (Fig. S4). To test if macrophages or neutrophils mediated the reduced fungal burdens of Pmt-deficient yeast cells in vivo, we depleted mice of phagocytes prior to infection. Administration of liposomal clodronate (intranasal) reduced the number of alveolar macrophages by approximately 70 to 80% (Fig. S5A). This depletion of macrophages did not significantly affect host infection by wild-type yeast cells and importantly did not rescue the lung infectivity of Pmt2deficient yeast cells, indicating that macrophages are not the primary source of control of the pmt2 mutant (Fig. 4A). Administration of anti-GR-1 antibody to mice caused a 95% lower level of circulating neutrophils than in mice treated with a control antibody (Fig. S5B). As with macrophages, depletion of neutrophils did not restore the survival of Pmt2-deficient yeast cells in murine lungs (Fig. 4B). In addition to neutrophils, anti-GR-1 antibody also depletes inflammatory monocytes (26), indicating that inflammatory monocytes are also not the source of immune control of yeast cells lacking O-linked mannosylation. In contrast to depletion of individual phagocyte populations, immunosuppression of mice by administration of cyclophosphamide (90% depletion of circulating white blood cells; Fig. S5C) results in full recovery of the in vivo fitness of Pmt2-deficient yeast cells (Fig. 4C). Histoplasma Mannosylation Confers Thermotolerance ® O-mannosylation enables yeast tolerance of elevated temperatures. Pmt2deficient yeast cells are viable at 37°C, but growth at elevated temperatures requires O-linked mannosylation of proteins. On solid medium in vitro, the growth of Pmt2expressing and Pmt2-deficient Histoplasma yeast cells is the same at 37°C (Fig. 5A). Increasing the temperature to 38°C slightly impaired the growth of Pmt2-expressing yeast cells; however, the growth of Pmt2-deficient yeast cells was virtually eliminated (Fig. 5A). In broth culture at 37°C, Pmt2-deficient yeast cells grew with the same kinetics as Pmt2-expressing strains (Fig. 5B). At 38°C, the growth rate of all yeast cells decreased; however, the pmt2 mutant was even further attenuated (Fig. 5B). Strains depleted of Pmt1 and Mnt1 are similarly impaired at 38°C compared to 37°C, but depletion of Pmt4 function did not impair growth (Fig. 5C), paralleling the in vivo virulence defects (Fig. 3B). These results indicate that Pmt1-and Pmt2-dependent O-linked mannosylation, but not Pmt4-initiated glycan addition, increased Histoplasma survival of heat stress.
Since yeast cells deficient in O-linked mannosylation are sensitive to elevated temperatures, we tested if temperatures encountered during mammalian infection are the source of the decreased in vivo fitness of Pmt2-deficient yeast cells. Like other mammals, the core body temperature of mice varies with activity; the baseline core temperature of mice averaged 36.5°C during daytime (when mice are less active) but increased to 37.8°C during the night (when mice are active) (Fig. 6A). Surprisingly, sublethal infection with either Pmt2-expressing or Pmt2-deficient yeast cells did not change core body temperatures (Fig. 6A), despite the production of fever-inducing cytokines (e.g., tumor necrosis factor alpha [TNF-␣], interleukin-1␤ [IL-1␤], and IL-6) after infection (Fig. S6). Although infection did not produce a febrile response in host animals (mice maintained a daily average body temperature of 37°C), the roughly 1°C normal increase during periods of activity is sufficient to cause restrictive conditions for Pmt2-deficient yeast cells (Fig. 5A).
As Pmt2-deficient yeast cells have less thermotolerance and cyclophosphamide treatment fully restored the virulence of pmt2 mutant yeast cells (Fig. 4C), we investigated if the cyclophosphamide-mediated rescue was linked to the host animal body temperature. Independent of infection with Histoplasma, cyclophosphamide treatment of mice reduced the relative body temperature by an average of 1°C (during both the day and night oscillations) as early as 24 h after treatment (Fig. 6B) and the relative reduction persisted throughout the infection time period. The reduced body temperature following cyclophosphamide treatment correlated with reduced production of proinflammatory IL-1␤ (Fig. 6C), which can affect the core body temperature (27). The cyclophosphamide-induced drop in the core body temperature and rescue of the pmt2 mutant yeast cells' virulence indicate that the attenuation seen stems from reduced thermotolerance in the absence of Pmt-catalyzed O-linked protein mannosylation.

DISCUSSION
Glycosylation is one of the most common posttranslational modifications of extracellular proteins. Although the precise role of protein glycosylation remains unknown, it is essential for the health and survival of cells. Accordingly, glycosylation enzymes have been suggested as targets for antimicrobials; however, similarities between fungal and mammalian glycosylation pathways have hindered these efforts against eukaryotic pathogens. Loss of O-linked protein mannosylation in Histoplasma resulted in viable Histoplasma cells in vitro; however, fitness in a mammalian host was severely attenuated. Histoplasma has only one representative each of the Pmt1, Pmt2, and Pmt4 families, of which only Pmt1 and Pmt2 are required for Histoplasma infection. Attempts to create Pmt1 and Pmt2, as well as Pmt2 and Pmt4, double mutants were unsuccessful (data not shown), suggesting that there are overlapping functions or protein substrates among the Pmt proteins. Synthetic lethality of double PMT mutants of Cryptococcus (28,29) and Saccharomyces (30) also suggests overlapping functions. The protein substrate specificities of Pmt1, Pmt2, and Pmt4 enzymes have not been defined, but loss of either Pmt1 or Pmt2 function results in nearly identical phenotypes in Histoplasma, consistent with Pmt1 and Pmt2 acting as a complex, which has been suggested by genetic and biochemical studies of Saccharomyces (20). While the precise mannan configurations assembled on Histoplasma proteins are currently undefined, the significant reduction in mannose on extracellular proteins, as well as recapitulation of the phenotypes due to the depletion of Mnt1, the second enzyme in the mannosylation pathway, confirms the role of Pmt2 in the mannosylation of proteins. MS analysis of Cfp4 and the glycans liberated from this protein revealed glycans up to three units in length (Fig. 1D), although longer chains are likely to exist because of loss of hexose units through the beta elimination reaction (i.e., peeling). The glycosylated mucin-like region of Cfp4 contains at least 28 total hexoses (Fig. S2), but whether all sites are glycosylated and if they have identical glycan lengths cannot be determined, given 16 potential O-linked glycosylation sites within the peptide. Nonetheless, the biochemical and MS data indicate that Histoplasma O-linked glycosylation is largely homopolymeric, consisting of mannose similar to that observed in other fungi. Loss of O-linked mannosylation caused a reduction in the galactose content of the culture filtrate, which may suggest that some Histoplasma mannan structures could be capped by galactose (31). O-linked mannosylation in other fungi has been associated with the formation and function of the fungal cell wall. Mutants lacking Pmt functions are variably sensitive to cell wall-disrupting agents (e.g., calcofluor white and Congo red), can display cell separation defects following mitosis (32,33), and lack modification of cell wall glycan assembly proteins (e.g., ␤-glucan synthases and chitinases) (30,34). Histoplasma yeast cells lacking Pmt2 function maintain general cell wall integrity, showing only a minor increase in sensitivity to Uvitex (a polysaccharide-binding dye). Nonetheless, the outer surface of the yeast cell wall was altered (Fig. 2), as electron microscopy suggested an increase in exposed mannoproteins. This counterintuitive result (more mannan exposure with the loss of O-linked mannosylation) suggests that there may be compensation by increased expression of proteins modified by N-linked glycans or by increased N-linked glycosylation of the affected proteins. The latter situation has been shown to occur in at least one cell wall protein in S. cerevisiae, in which the cell wall protein Ccw5 is N-linked glycosylated in the Pmt4 mutant but not in wild-type S. cerevisiae (35). The increased mannan exposure due to loss of O-linked mannosylation is also supported by the MR-dependent recognition of Pmt2-deficient yeast cells. Together with the decreased ␤-glucan exposure, these results indicate that proper organization of the fungal cell wall requires O-linked mannosylation of proteins, likely those proteins that contribute to hydrolysis or formation of cell wall polysaccharide linkages, as these enzymes are often glycosylated (36,37).
Despite the altered cell wall and glycan recognition, phagocyte associations were not affected. This is not unexpected, as Histoplasma yeast cells primarily target ␤-integrins (i.e., CR3) (38-40) for stimulating phagocytic uptake and minimize recognition by signaling receptors (e.g., Dectin-1) (41). Investigation of phagocyte-centric aspects of Histoplasma pathogenesis revealed no defects to explain the substantial in vivo fitness attenuation of yeast cells deficient in O-linked mannosylation; there was no change in proinflammatory cytokines in vitro (data not shown) or in vivo (Fig. S6), and the survival of yeast cells in cultured polymorphonuclear leukocytes (PMNs) and macrophages was unaffected (Fig. S4). Furthermore, depletion of phagocyte populations in vivo did not rescue the attenuation, demonstrating that O-linked mannosylation is not required for defense against PMNs or macrophages.
O-linked protein mannosylation is necessary for thermotolerance of Histoplasma yeast cells. The elevated temperatures of mammals was postulated as a major restriction of the evolution of fungi as pathogens of mammals but widespread proliferation of fungi as pathogens of plants (42). Indeed, the ability to proliferate at 37°C is one of the classic virulence determinants of Cryptococcus (43). For Histoplasma infections, elevated temperature is both good and bad. Elevated temperature is an essential signal to trigger differentiation into the pathogenic program. This differentiation is necessary for the expression of virulence factors that enable yeast cells to survive host immune defenses (44). However, Histoplasma yeast cells must also be able to survive the elevated temperature of a mammalian host. While Histoplasma cells lacking O-linked mannosylation can maintain the pathogenic yeast state, they had significantly limited thermotolerance; wild-type Histoplasma yeast cells tolerated temperatures of up to 38°C in vitro, but strains deficient in O-linked mannosylation were arrested/dead at temperatures above 37°C (Fig. 5). Sensitivity to elevated temperatures also characterizes other non-thermally dimorphic fungal species with defects in O-linked mannosylation (28-30, 34, 45), although the threshold is not as narrowly defined as with Histoplasma. Cryptococcus neoformans Pmt2 and Pmt4 mutants both grew well at 30°C but were unable to grow at 37 and 39°C, respectively (28,29). In addition, C. albicans Pmt1 and Pmt2 are required for growth at 42°C (34) and Aspergillus fumigatus conidiation is severely attenuated at 50°C when lacking Pmt1 (45). While some Cryptococcus and Candida mutants have severe fitness deficiencies in mouse models, the attenuation in vivo has not been conclusively linked to temperature sensitivity; many of these mutants have significant structural deficiencies, and loss of Pmt4 in C. albicans results in virulence attenuation but not in temperature sensitivity (34).
This study demonstrated that O-linked mannosylation of proteins confers sufficient thermotolerance to Histoplasma yeast to enable infection of mammalian hosts. Although Histoplasma infection stimulated the production of pyrogenic cytokines (i.e., IL-1␤, IL-6, and TNF-␣), pyrexia is not induced in mice. However, maintenance of a core body temperature of 37 Ϯ 1°C was restrictive to Histoplasma yeast cells lacking O-linked mannosylation. Cyclophosphamide treatment artificially lowered the core body temperature, which rescued the in vivo fitness attenuation of yeast cells lacking Pmt2 function. Although cyclophosphamide affects multiple phagocyte populations, specific depletion of these cells did not rescue the attenuation, confirming that it is the cyclophosphamide-dependent reduction in body temperature that allowed Pmt2deficient yeast cells to survive and proliferate in the host. Furthermore, reduction of the average core body temperature was required to permit the growth of Omannosylation-deficient yeast cells, rather than prevention of fever, since treatment of mice with ibuprofen did not rescue the Pmt2 attenuation (data not shown). These findings, together with the perturbed organization of the cell wall, lead to the hypothesis that O-mannosylation maintains the stability and/or function of enzymes critical to the formation and integrity of the Histoplasma cell wall at elevated temperatures. Thus, O-linked mannosylation facilitates the thermotolerance of Histoplasma yeast cells and helps define how this thermally dimorphic fungus has become a successful fungal pathogen of mammals.

MATERIALS AND METHODS
Histoplasma strains and culture. The Histoplasma capsulatum strains used in this study were derived from wild-type strain G217B (ATCC 26032) and are listed in Table S1. Histoplasma yeast cells were grown in Histoplasma-macrophage medium (HMM) (46) supplemented with 100 g/ml uracil for growth of auxotrophs or with 25 M FeSO 4 for growth on solid medium.
Immunoblotting analysis. Culture filtrate proteins from wild-type (PMT2) or mutant (pmt2) yeast cells were treated with PNGase F (New England Biolabs) to remove N-linked glycans and then separated under reducing conditions by 10% SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Proteins were detected with monoclonal antibodies to Cfp4 (clone 2D20) (21) and Sod3 (clone 3J23) and visualized with horseradish peroxidase (HRP)-conjugated anti-mouse antibody and HRP chemiluminescent substrate (Millipore).
Saccharide composition analysis. The saccharide content of extracellular culture filtrate proteins was analyzed by the alditol acetate method (51) after exchange into phosphate-buffered saline (PBS). scyllo-Inositol (an internal saccharide standard) was added, and the glycans from 50 g of protein were hydrolyzed in 2 M trifluoroacetic acid (120°C for 3 h), followed by reduction with sodium borodeuteride (NaBD 4 ) and acetylation by acetic anhydride. The analytes were separated by GC (Trace GC Ultra; Thermo Scientific) with a 30-meter nonpolar capillary column (Restek) (210 to 240°C at 2°C/min in 30 min) and analyzed by MS (DSQII; Thermo Scientific). Peaks corresponding to monosaccharides (glucose, mannose, and galactose) were identified, and the total amount of each sugar was calculated relative to the peak area of scyllo-inositol.
O-Glycan analysis of Cfp4 was performed by MS of glycans released from Cfp4 protein purified from Histoplasma yeast culture filtrates. A Cfp4:hexahistidine fusion protein lacking N-linked glycosylation sites (21) was overexpressed by PMT2 or pmt2 Histoplasma yeast cells and purified from culture filtrates with HisPur cobalt spin columns (Thermo Scientific). Glycans associated with Cfp4 were released by nonreductive beta elimination and the addition of 1-phenyl-3-methyl-5-pyrazolone (PMP) (52). Briefly, 60 g of ribose (as an internal standard) was added to 400 g of Cfp4 and 0.5 M PMP and the samples were incubated at 50°C in 30% ammonia for 15 h. To recover the glycans, samples were extracted with CHCl 3 and the aqueous layer was dried and resuspended in 5% acetonitrile and further purified with a C 18 spin column (Pierce). Samples were analyzed with a MALDI Ultraflextreme mass spectrometer (Bruker) by using a matrix composed of 10:1 DHAP (2,6-dihydroxyacetophenone):DAHC (diammonium hydrogen citrate) mixed with oligosaccharide at 10:1. Samples were run in reflectron mode from 100 to 5,000 m/z.
Cell wall sensitivity assays. Yeast cells were grown in 96-well microtiter plates (53) with graded concentrations of the cell wall-destabilizing compounds Congo red (MP Biomedicals), SDS (Fisher), sodium chloride (RPI), and Uvitex 3BSA (54). Wells were inoculated with 2 ϫ 10 6 yeast cells/ml in HMM and incubated at 37°C. Turbidity was measured by determining the optical density at 595 nm (OD 595 ), and IC 50 s were computed by nonlinear regression of the dose-response data.
Infection of mice and determination of virulence in vivo. C57BL/6 mice (Charles River, Inc.) were Histoplasma infected by intranasal delivery. Their lungs were collected at various time points and homogenized, and serial dilutions of the homogenates were plated on solid HMM to determine the fungal burdens (CFU counts). Measurement of murine body temperature was performed with a probe thermometer (RET-3 probe; Kent Scientific) at 1100 and 2300 h each day. In vivo cytokine production was quantified with a proinflammatory multiplex panel (Mouse Proinflammatory Panel 1; Meso Scale Diagnostics) on lung homogenates 24 and 48 h after Histoplasma infection.
MR-dependent association of yeast cells with macrophages was quantified by knockdown of MR from human MDMs (56). Peripheral blood mononuclear phagocytes (PBMCs) were isolated from the buffy-coat layer after Ficoll-Paque PLUS sedimentation of human peripheral blood (400 ϫ g for 40 min at room temperature). MDMs were differentiated from monocytes (RPMI plus 20% autologous serum, 5 days). Day 5 PBMCs were transfected with MR-specific (GUGGUACGCAGAUUGCACGUU, AGUCCUUUCCGAUAUUUG, and AUUUAAAGUGGUGUUGCCC) or scrambled control small interfering RNAs (siRNAs) by using Amaxa Nucleofector (Amaxa Biosystems) and the Y010 nucleofection settings (56). After 48 h, MDMs were infected with Histoplasma yeast cells at an MOI of 1:1 (yeast-to-macrophage ratio) and unbound yeast cells were removed after 2 h. Associated yeast cells were quantified by hypotonic lysis of MDMs and plating on solid HMM for CFU counting. Depletion of MR by siRNA was confirmed by immunoblotting (SC-48758; Santa Cruz Biotechnology).
Transmission electron microscopy (TEM). Histoplasma yeast cells were fixed in 2% glutaraldehyde in cacodylate buffer (0.1 M cacodylic acid, pH 7.4), postfixed in 1% osmium tetroxide, rinsed with buffer, and embedded in 2% low-temperature-gelling agarose. One-cubic-millimeter blocks were incubated in 1% uranyl acetate for 90 min prior to being dehydrated in a series of graded ethanol washes. Samples were incubated in propylene oxide for 20 min and infiltrated with Eponate 12 resin. Seventy-nanometer sections were stained in 2% aqueous uranyl acetate and Reynolds lead citrate and observed with a transmission electron microscope (FEI Tecnai Spirit) at 80 kV.
Ethics statement. Animal experiments were performed in compliance with the National Research Council Guide for the Care and Use of Laboratory Animals and were approved by the Ohio State University (OSU) Institutional Animal Care and Use Committee (2007A0241). Human cells were obtained from healthy volunteers after Health Insurance Portability and Accountability Act research authorization and written informed consent were obtained in accordance with the Declaration of Helsinki. The human subject protocol was reviewed and approved by the OSU Biomedical Sciences Institutional Review Board (protocol number 2008H0242) under the OSU Office for Human Research Protections (Federalwide Assurance number 00006378).

ACKNOWLEDGMENTS
We thank Jordi Torrelles and Jesus Arcos for their assistance with biochemical quantification of saccharides, Darin Wiesner for developing the protocol for in vivo phagocyte identification and assisting with interpretation of flow cytometry data, the OSU Campus Microscopy and Imaging Facility for TEM processing, the OSU Center for Clinical and Translational Science for cytokine analysis, and the OSU Campus Chemical Instrument Center for assistance with MALDI and ESI MS. We have no conflicts of interest with respect to the contents of this article.
Funding for this research was provided by National Institute of Allergy and Infectious Diseases grants R21-AI117122 (C.R.), R01-AI035681 (B.K.), R01-AI093553 (M.W.), and R01-AI059639 (L.S.). A.G. was supported by predoctoral fellowship T32-AI112542 administered by the Center for Microbial Interface Biology (CMIB) at OSU. The cytokine analysis panel was funded by OSU CCTS CTSA grant UL1TR001070. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.