|Home | About | Journals | Submit | Contact Us | Français|
The major capsular polysaccharide of Cryptococcus neoformans, glucuronoxylomannan (GXM), is recognized by Toll-like receptor 2 (TLR2), TLR4, and CD14. In these studies, mice deficient in CD14, TLR2, TLR4, and the TLR-associated adaptor protein, MyD88, were utilized to investigate the contribution of TLRs and CD14 to in vivo host defenses against C. neoformans. MyD88−/− mice had significantly reduced survival compared with wild-type C57BL/6 mice after intranasal (i.n.) and intravenous (i.v.) infection with live C. neoformans. CD14−/− mice had reduced survival when infected i.v., while TLR2−/− mice died significantly earlier after i.n. infection. Mortality was similar comparing TLR4 mutant C3H/HeJ mice and control C3H/HeOuJ mice following i.v. or i.n. challenge with C. neoformans. The course of pulmonary cryptococcosis was studied in more detail in the CD14−/−, TLR2−/−, and MyD88−/− mice. MyD88−/− mice infected i.n. had higher numbers of CFU in the lungs as well as higher GXM levels in the sera and lungs 7 days after infection than wild-type mice did. Surprisingly, there were no major differences in the levels of tumor necrosis factor alpha, interleukin-4 (IL-4), IL-10, IL-12p70, or gamma interferon in the lungs of C. neoformans-infected knockout mice compared with wild-type mice. Histopathologic analysis of the lungs on day 7 postinfection revealed minimal inflammation in all mouse groups. These studies demonstrate a major role for MyD88 and relatively minor roles for CD14 and TLR2 in the response to cryptococcal infection, with the decreased survival of MyD88−/− mice correlating with increased numbers of lung CFU and serum and lung GXM levels.
Cryptococcus neoformans is an encapsulated yeast that can cause life-threatening infections in persons with defects in T-cell-mediated immunity, particularly those with AIDS. Exposure to C. neoformans is thought to most commonly occur by inhalation. If the infection is not controlled in the lungs, the organism can disseminate via the lymphatics and bloodstream. While any organ can be affected, C. neoformans has a predilection for invading the central nervous system, leading to meningoencephalitis (45). In mouse models, optimal defenses against pulmonary cryptococcosis require CD4+ and CD8+ T cells and a Th1 cytokine response (30, 39). Protective roles for tumor necrosis factor alpha (TNF-α), interleukin-12 (IL-12), IL-18, and gamma interferon (IFN-γ) have been inferred based upon knockout mouse models and cytokine neutralization studies (30, 39).
The major virulence factor of C. neoformans is its capsule, which is composed primarily of the polysaccharide glucuronoxylomannan (GXM). GXM is shed during cryptococcosis and circulates in the blood and cerebral spinal fluid. The cryptococcal capsule has numerous immunomodulatory properties, including inhibition of phagocytosis, inhibition of leukocyte migration, alteration of cytokine production by leukocytes, inhibition of neutrophil anticryptococcal activity, and interference with dendritic cell maturation (10, 11, 37, 53, 54). Although encapsulated C. neoformans is poorly recognized by phagocytes in the absence of opsonins, when opsonized by complement or anticapsular antibodies, the yeast can be phagocytosed via complement receptors CD35 (CR1), CD11b/CD18 (CR3), and CD11c/CD18 (CR4) or by immunoglobulin Fc receptors, respectively (12, 32). Soluble GXM has been shown to bind CD14, CD11/CD18 heterodimers, TLR2, and TLR4 (9, 48). In addition, uptake of GXM by monocytes and neutrophils was demonstrated to involve CD14 and CD11/CD18, respectively (37).
Toll-like receptors (TLRs) are innate immune-pattern recognition receptors that recognize a wide range of microbes and their products. When these receptors are activated, a signaling cascade is initiated that results in an inflammatory response, including the upregulation of costimulatory molecules on antigen-presenting cells and the release of proinflammatory cytokines (2). MyD88 is an adaptor molecule that is critical for the signaling responses initiated through most TLRs as well as the IL-1 receptor (IL-1R) family, which includes IL-1R and IL-18R (1). CD14 is a TLR-associated pattern recognition receptor that is glycosylphosphatidylinositol anchored and thus has no direct signaling capabilities. Although GXM has been shown to bind CD14, TLR2, and TLR4 in vitro, the role of these receptors in vivo is unknown. Thus, we sought to determine the function of these receptors, as well as that of MyD88, in cryptococcal infection.
The C. neoformans serotype A encapsulated strain 145 (no. 62070; American Type Culture Collection, Manassas, Va.) was grown on Sabouraud dextrose agar (Remel, Lenexa, Kans.) at 30°C and passaged every 3 to 4 days. For each infection, a fresh culture of C. neoformans 145 was started from −80°C stocks. For the survival studies with C57BL/6J, CD14−/−, TLR2−/−, and MyD88−/− mice, C. neoformans 145 obtained following culture on Sabouraud dextrose agar was grown for 48 h at 30°C in yeast nitrogen broth (Difco, Detroit, Mich.). For the rest of the infections, yeast cells were harvested from cultures on Sabouraud dextrose agar between days 2 and 4. Prior to inoculation, the yeast cells were washed in phosphate-buffered saline (PBS), counted, and adjusted to the desired concentration.
Specific-pathogen-free C57BL/6J wild-type mice were purchased from the Jackson Laboratory (Bar Harbor, Maine). The CD14−/−, TLR2−/−, and MyD88−/− knockout mice were engineered as described and backcrossed at least five generations to a C57BL/6 background (22, 38, 52). The MyD88−/− and TLR2−/− mice were generously provided by Shizuo Akira (Osaka University, Osaka, Japan) via Douglas Golenbock (University of Massachusetts Medical School, Worcester, Mass.), and the CD14−/− mice were kindly provided by Mason Freeman (Massachusetts General Hospital, Harvard Medical School, Boston, Mass.). C3H/HeJ and C3H/HeOuJ mice were purchased from the Jackson Laboratory. C3H/HeJ mice have a point mutation in the TLR4 gene, which renders the receptor unable to signal (16). C3H/HeOuJ mice have a functional TLR4 and were used as controls for the C3H/HeJ mice. The mice were housed in microisolator cages at the Boston University Medical Center Laboratory Animal Sciences Center. For the survival studies, infected mice were sacrificed when moribund, which was determined by using objective criteria including periorbital edema, posturing, ataxia, and inability to feed; these criteria were approved by the Boston University Institutional Animal Care and Use Committee. The mice were between 7 and 14 weeks of age at the time of infection.
For intranasal (i.n.) infections, mice were lightly anesthetized with Halothane (Samuel Perkins, Quincy, Mass.), and 50 μl of a C. neoformans suspension was pipetted onto the nares. These conditions promote aspiration into the lungs (31). For intravenous (i.v.) infections, the mice received 100 μl of a yeast cell suspension in the lateral tail vein. For intraperitoneal (i.p.) infections, mice received the fungal suspension in a volume of 1 ml. Aliquots of the inoculum were plated on Sabouraud dextrose agar to confirm the number of CFU administered.
As previously described, mice were euthanized by CO2 asphyxiation (35); the lungs and brains were harvested, weighed, and placed in 14-ml sterile polypropylene tubes containing 2 ml of cold PBS supplemented with a protease inhibitor cocktail (Boehringer Mannheim, Mannheim, Germany). The organs were homogenized with a PowerGen model 700 tissue homogenizer (Fisher Scientific, Pittsburgh, Pa.) for 5 to 10 s. The homogenate was diluted in PBS containing 40 U of penicillin and 40 μg of streptomycin per ml and plated on Sabouraud dextrose agar to determine the number of CFU per gram of tissue.
The organs were obtained and homogenized as described above. The homogenates were spun down at 19,000 × g for 3 min; the supernatants were collected and stored at −80°C. TNF-α and IL-12p70 were quantified using murine enzyme-linked immunosorbent assay (ELISA) kits from R&D Systems (Minneapolis, Minn.). IL-4, IL-10, and IFN-γ were measured using murine ELISA kits from eBioscience (San Diego, Calif.).
The organs were obtained and homogenized as above. Blood was collected postmortem by cardiac puncture and spun at 19,000 × g for 5 min, and the serum was collected and stored at −80°C. The GXM ELISA was performed as described (13) using unlabeled anti-GXM monoclonal antibody 3C2 (a kind gift of Thomas Kozel, University of Nevada School of Medicine, Reno, Nev.) as the coating antibody and horseradish peroxidase-labeled 3C2 as the detection antibody. The GXM used to generate the standard curve was isolated as previously described (48) from C. neoformans serotype A strain 6 (no. 62066; American Type Culture Collection). The ELISA was sensitive at a range of concentrations from 1 to 100 ng of GXM/ml.
The mice were sacrificed by CO2 asphyxiation, and the lungs were inflated with 10% buffered formalin. The lungs and brains were then removed and fixed in 10% buffered formalin. Tissue was embedded in paraffin wax in an automated tissue processor at the Skin Pathology Laboratories (Boston, Mass.). Sections (5 to 6 μm thick) were cut and stained with periodic acid-Schiff (PAS) stain or hematoxylin and eosin using an automated slide processor. The sections were read in a blinded fashion by a veterinary pathologist and scored for the degrees of fungal infiltration and inflammation as shown in Table Table11.
Kaplan-Meier survival curves were compared using the log rank test (NCSS Statistical Software, Kaysville, Utah). All other statistical comparisons utilized the Student's t test. P values of <0.05 were considered significant. Bonferroni's correction was applied when making multiple comparisons.
To study the role of TLR/MyD88 signaling in cryptococcosis, C57BL/6J wild-type, CD14−/−, TLR2−/−, and MyD88−/− mice were infected i.n. or i.v. with the virulent C. neoformans serotype A strain 145. The i.n. route of infection was used because it mimics the presumed natural pulmonary route of infection, while the i.v. route of infection simulates the hematogenous spread observed when the organism disseminates. MyD88−/− mice infected i.n. or i.v. succumbed to infection significantly sooner than the wild-type mice (P value of <0.001 for both i.n. and i.v. infections) (Fig. 1A and B). Compared to wild-type mice, TLR2−/− mice were more susceptible to infection by the i.n. route (P = 0.009) but not by the i.v. route (Fig. 1B and C). However, the survival of the TLR2−/− mice infected i.n. was not as dramatically impaired as that seen with the MyD88−/− mice. The CD14−/− mice succumbed earlier than the wild-type mice when infected i.v. (P = 0.047) (Fig. (Fig.1B);1B); however, the statistical significance was lost after applying Bonferroni's correction. The survival of the CD14−/− mice was similar to that of the wild-type mice following i.n. infection (Fig. (Fig.1A1A).
To investigate the role of TLR4, TLR4 mutant C3H/HeJ and control C3H/HeOuJ mice were infected with C. neoformans i.n., i.v., or i.p., and survival was monitored. There were no significant differences in susceptibility to infection between TLR4 mutant and control mice when the mice were infected i.n. with 105 or 106 organisms or given 105 yeast cells i.v. (Table (Table2).2). Similarly, there was no difference in survival between the C3H/HeJ mice and the control mice when the mice were infected i.p. with 104 organisms, but when infected i.p. with 106 yeast organisms, the C3H/HeJ mice had significantly decreased survival (Table (Table2).2). Taken together, these data suggest that a functional TLR4 receptor does not contribute greatly to resistance to C. neoformans infection.
In order to examine the course of pulmonary C. neoformans infection in more detail, the sera, lungs, and brains were harvested from wild-type, CD14−/−, TLR2−/−, and MyD88−/− mice on day 7 after i.n. infection. The lungs were studied since they are the primary site of infection, and the brains were examined because of the propensity of C. neoformans to disseminate to the central nervous system. MyD88−/− mice displayed significantly increased numbers of CFU in the lungs at day 7, which correlated with the reduced survival of these mice (Fig. (Fig.2).2). As expected, the higher numbers of CFU correlated with higher GXM levels in the lungs of infected MyD88−/− mice (Fig. (Fig.3B).3B). The MyD88-deficient mice also had significantly higher serum GXM levels than the wild-type mice (Fig. (Fig.3A).3A). There were no significant differences in the numbers of lung CFU or lung or serum GXM levels in CD14−/− or TLR2−/− mice compared with wild-type mice (Fig. (Fig.22 and and33).
In contrast to the lungs, neither the numbers of CFU nor the GXM levels in the brain differed between knockout and wild-type mice at day 7 postinfection (data not shown). The numbers of brain CFU were highly variable, with only some mice having measurable dissemination to the brain. The GXM levels in the brains of infected mice were also variable. However, there was a direct correlation between numbers of brain CFU and GXM levels (data not shown).
We also examined numbers of CFU and GXM levels at day 21 postinfection in the CD14−/− and TLR2−/− mice, but not in the MyD88−/− mice, as they had begun to succumb to the infection by that point. The numbers of lung CFU were much higher on day 21 than day 7 but were comparable in CD14−/−, TLR2−/−, and wild-type mice. Similarly, GXM levels in the lungs, sera, and brains were higher on day 21 than on day 7 but were comparable in the knockout and wild-type mice. Almost all of the mice had measurable numbers of brain CFU by day 21, but there were no significant differences in number between the knockout and wild-type groups (data not shown).
One important consequence of TLR activation by microbes and microbial products is the stimulation of cytokine pathways, particularly ones leading to proinflammatory responses (3). Therefore, levels of the cytokines TNF-α, IFN-γ, and IL-4 in whole lung and brain homogenates were determined on days 7 and 21 after i.n. infection with C. neoformans. IL-12p70 and IL-10 levels were examined on day 7 postinfection as well. TNF-α, IL-12, and IFN-γ were chosen because they have been shown to be critical for effective host defenses in murine models of cryptococcosis (8, 14, 19, 28, 29). In contrast, IL-10 and IL-4 were selected for study because IL-10 was deleterious in a pulmonary model of cryptococcosis, while the role of IL-4 was dependent upon the mouse strain and route of infection (5, 6, 23). We found no differences in lung TNF-α, IFN-γ, or IL-4 levels between any of the knockout and wild-type mice at days 7 or 21 postinfection (Fig. (Fig.4)4) (data not shown). IL-12p70 and IL-10 levels 7 days after infection were similar in knockout and wild-type mice as well (Fig. (Fig.4).4). Interestingly, in all mouse groups, the only cytokine that was upregulated in the lungs after C. neoformans infection was IL-4. While lung IL-10 and IFN-γ levels were similar in uninfected and infected mice, there was a trend towards downregulation of TNF-α and IL-12p70 after cryptococcal infection, with significantly lower levels of IL-12p70 in wild-type and CD14−/− mice after infection.
This trend was also observed in the brain. The amounts of TNF-α, IL-12p70, and IL-10 were lower in the brains of day 7 infected wild-type and MyD88−/− mice than in uninfected mice. Again, however, the amounts of these cytokines were similar in wild-type and MyD88−/− mice (Fig. (Fig.55).
Lung sections were prepared from wild-type and knockout mice 7 days postinfection and were scored for fungal burden, colony size and alveolar morphology, extent of inflammation, severity of inflammation, and the degree of bronchus-associated lymphoid tissue (BALT) expansion as shown in Table Table1.1. While the scores of TLR2- and CD14-deficient mice were similar to those of wild-type mice, MyD88−/− mice had a significantly increased organism score (reflecting fungal burden and colony size) (Fig. (Fig.6).6). The inflammation score was similar for all mouse groups (Fig. (Fig.66).
Lung sections from wild-type mice contained individual or small groups of C. neoformans (Fig. (Fig.7A).7A). The organisms were most numerous in respiratory bronchioles and the immediately adjacent alveoli. In contrast, the lung sections from MyD88−/− mice contained large clusters of organisms within tertiary and respiratory bronchioles, and the colonies extended beyond adjacent alveoli to more peripheral regions of the lung (Fig. (Fig.7B).7B). Often, the colonies present within the alveolar spaces of MyD88−/− mice were so large that they caused lateral distension of the alveolar walls. Despite differences in the organism burden, wild-type and MyD88−/− mice had a similar, minimal degree of pulmonary inflammation (Fig. (Fig.7,7, panels C and D, respectively), characterized by minimal expansion of the alveolar walls by small numbers of neutrophils and lesser numbers of eosinophils and monocytes/macrophages (Fig. 7E and F). Sometimes the infiltrating leukocytes were present within alveolar spaces, and foamy alveolar macrophages were occasionally observed in proximity to the organisms. PAS staining revealed mild, focal goblet cell hyperplasia in some airways of both wild-type and MyD88−/− mice. This finding may correlate with the increase of IL-4 in the lungs of infected mice, as Th2 cytokines, including IL-4, have been implicated in airway goblet cell hyperplasia (57). However, the extent of hyperplasia was similar in wild-type and MyD88−/− mice.
These data identify a critical role for MyD88 during cryptococcosis, with relatively minor roles for TLR2 and CD14. We demonstrate that MyD88−/− mice succumbed to C. neoformans infection significantly earlier than wild-type mice when infected either i.n. or i.v. Mice lacking TLR2 had significantly reduced survival only after pulmonary challenge, while CD14−/− mice displayed a trend towards reduced survival after intravenous infection. Other studies of TLR/MyD88 knockout mice have also found that the requirement for these molecules can depend upon the route of infection. For example, MyD88−/− mice are highly susceptible to systemic (i.v.), but not pulmonary, Staphylococcus aureus infection (49, 51).
Our data showed that C3H/HeJ mice, which have defective TLR4 signaling responses, had mortality similar to that of the C3H/HeOuJ controls following i.n. or i.v. C. neoformans challenge. However, the C3H/HeJ mice had increased mortality following infection with an i.p. inoculum of 106, but not 104, organisms. Inoculum-dependent requirements for TLRs have been shown in other systems as well. One study found that TLR2−/− mice were more susceptible than wild-type mice to aerosolized Mycobacterium tuberculosis when infected with a high, but not a low, inoculum (46).
MyD88−/− mice infected i.n. had increased numbers of CFU in the lungs and increased GXM levels in the lungs and sera, which correlated with the reduced survival of these mice. The higher GXM levels in the lungs and sera of these mice may have exerted deleterious effects on the immune response, possibly by impairing leukocyte recruitment or the anticryptococcal activity of phagocytic cells (10, 37). The histopathological analysis reflected the CFU and cytokine data from C. neoformans-infected mice. The lungs of MyD88−/− mice demonstrated increased organism burden and colony size, but the inflammation seen was comparable to that in wild-type mice. Inflammation was mild in all infected mice, which corresponded to the lack of proinflammatory cytokines measured in lung homogenates. As all of the mouse groups had similar cytokine levels in the lungs after infection, it is unlikely that a blunted cytokine response led to the increased mortality of MyD88-deficient mice.
We did not observe an increase in lung TNF-α, IL-12, or IFN-γ after i.n. infection of C57BL/6J mice with C. neoformans 145. This perhaps should not be surprising considering that the immune response to C. neoformans in murine models varies depending upon the murine strain as well as the strain of C. neoformans used (5, 15, 18, 20). C57BL/6 mice do not develop a protective Th1 response to pulmonary C. neoformans; instead, they initiate a deleterious response that results in nonresolving pneumonia. Cells from C. neoformans-infected C57BL/6 mice produce less IFN-γ and IL-2, and more IL-5, than murine strains capable of clearing pulmonary infection, such as the C.B-17, CBA/J, and BALB/c mice (15, 17, 20). In agreement with those studies, we did not detect a Th1 response in our model. While we did not measure IL-5, an increase in IL-4 was detected after infection. Because C57BL/6 mice do not respond to C. neoformans with a sustained Th1 response, it may not have been possible to identify a role for TLRs/MyD88 in an anticryptococcal proinflammatory cytokine response. Such a response might have been found if the knockout mice were on a different genetic background. Although most studies have examined the role of TLRs in inducing Th1 responses, TLR signaling can regulate Th2 responses as well. However, the induction of lung IL-4 in our model was TLR2 and MyD88 independent. One important caveat to our cytokine data is that by using whole organ homogenates, potential differences in cytokine levels at foci of infection might not be detected.
The strain of C. neoformans used influences the immune response as well, with heavily encapsulated and highly virulent C. neoformans strains generally eliciting poor inflammatory responses (5, 18, 20, 26). CBA/J mice infected with highly virulent strains of C. neoformans, including strain 145, are not able to clear pulmonary infection (5, 7, 18). B6129F2/J mice infected intratracheally with C. neoformans 145 produced less TNF-α, IFN-γ, and MCP-1 in the lungs than mice infected with a less virulent strain (44). Similarly, BALB/c mice infected intratracheally with another highly virulent strain of C. neoformans, YC-11, expressed IL-4 and IL-10 in the lungs but little TNF-α and no IFN-γ (27). It is likely that the immunosuppressive properties of GXM contribute to the lack of proinflammatory cytokines in the lungs during pulmonary cryptococcosis. GXM has been shown to inhibit TNF-α release from human monocytes (54), which could help explain the downregulation of lung TNF-α and IL-12p70 levels we observed on day 7 postinfection. Additionally, C. neoformans 145 has been demonstrated to downregulate TNF-α production via the production of high levels of melanin (18). The reason for the lower levels of TNF-α, IL-12p70, and IL-10 in the brains of infected wild-type and MyD88−/− mice is unclear, as almost all of the mice had no CFU and low levels of GXM in the brain.
TLRs have been shown to recognize a number of fungal products in addition to GXM. For example, Aspergillus fumigatus and Candida albicans activate cells via TLR2 and TLR4 (21, 33, 36, 41, 42, 50, 56). Similarly, CD14 also recognizes fungi, including Blastomyces dermatitidis, A. fumigatus, and cryptococcal GXM (33, 43, 48, 56). The role of the TLR/IL-1R superfamily in the responses to C. albicans and A. fumigatus has been studied in vivo as well. TLR4 mutant C3H/HeJ mice were found to be more susceptible to disseminated C. albicans infection than wild-type mice (41). Conflicting results have been found concerning the role of TLR2 in resistance to disseminated candidiasis (40, 55). Interestingly, the study demonstrating that wild-type mice were more susceptible than TLR2−/− mice to C. albicans infection found severe impairment of IL-10 production in the TLR2−/− mice (40). Recently, consistent with our data on C. neoformans, Bellocchio et al. demonstrated that MyD88 is required for resistance to infection with C. albicans yeast and hyphae as well as A. fumigatus conidia (4). However, the individual contributions of TLR2, TLR4, TLR9, and IL-1R varied depending upon the species and morphotype of the fungus.
Although GXM has been demonstrated to be a ligand for TLR2, TLR4, and CD14 in vitro (48), our data do not support an absolute requirement for these receptors in the context of cryptococcal infection. One possible explanation for these findings is that stimulation with GXM does not activate the mitogen-activated protein kinase pathways necessary for proinflammatory cytokine release (48). It is important also to consider the complexity of antigens on whole microbes as well as the redundancy of innate immunity. When one receptor is knocked out, other receptors may compensate. Indeed, the modest phenotype of TLR2, TLR4 and CD14 mutant mice underscores the important roles of other receptors in immune recognition of C. neoformans. For example, complement and Fc receptors mediate the binding and uptake of opsonized C. neoformans, while mannose receptors recognize antigenic cryptococcal mannoproteins (34). In fact, these mannoproteins have been shown to be protective in a murine model of cryptococcosis (35).
The MyD88−/− mice were more susceptible to cryptococcosis than the TLR2 and TLR4 mutant mice. One explanation for this finding is that MyD88 serves not only as an adaptor protein for TLR signaling but also for signaling through IL-1R and IL-18R (1). In this regard, IL-18−/− mice are more susceptible to cryptococcosis, apparently due to impaired induction of IFN-γ (24, 25). In addition, IL-1 is produced both in vitro and in vivo following cryptococcal stimulation (27, 54). Finally, full-scale activation of macrophages by IFN-γ requires MyD88 (47), which raises the possibility that defective macrophage activation in MyD88−/− mice could impair the anticryptococcal response. Although the exact mechanisms remain to be elucidated, our studies demonstrate that MyD88 is critical for host defenses against C. neoformans.
We thank Xiuping Liu and Jianmin Chen for invaluable technical assistance, Douglas Golenbock, Mason Freeman, and Shizuo Akira for providing the mice, and Thomas Kozel for the 3C2 antibodies.
This work was supported in part by National Institutes of Health grants RO1 AI25780, RO1 AI37532, and T32 AI07309. S.M.L. is the recipient of a Burroughs Wellcome Fund Scholar Award in Pathogenic Mycology. M.K.M is the recipient of a Boston University School of Medicine Graduate Student Research Fellowship.
Editor: T. R. Kozel