MyD88 Signaling Regulates Both Host Defense and Immunopathogenesis during Pneumocystis Infection

Pneumocystis is a respiratory fungal pathogen that causes Pneumocystis pneumonia (PcP) in immunocompromised individuals. PcP-related morbidity and mortality continues to be a major health concern for patients with HIV infection as well as for those without HIV infection who are undergoing immunosuppression as a consequence of chemotherapy or organ transplantation (1, 2). New immunosuppressive therapies, such as anti–TNF-α therapy for Crohn’s disease and rheumatoid arthritis, are increasing the pool of “at risk” patients (3). In addition, Pneumocystis frequently colonizes patients with chronic obstructive pulmonary disease, which appears to exacerbate disease severity (4). Therefore, a better understanding of the mechanisms of PcP-related immunopathogenesis is key to improving upon current treatments.

Clinical observations and animal studies have indicated that lung injury during PcP is caused primarily by the host’s immune-mediated inflammatory response, and is not absolutely related to Pneumocystis burden (5–8). For example, in the CD4⁺ T cell–depleted model of PcP, physiological deterioration is associated with an increase in lung chemokine and cytokine levels, as well as the recruitment of large numbers of CD8⁺ T cells and neutrophils to the lung. Of interest, when CD4⁺ and CD8⁺ T cells are depleted simultaneously, fewer signs of inflammation, less cell recruitment, and improved lung function result, suggesting that CD8⁺ T cells are responsible for lung injury and respiratory impairment in this model of PcP (9).

Recent studies have focused on characterizing the mechanisms involved in generating pathogenic immune and inflammatory responses that damage the lung and other tissues. The TLR system is one of the most important host defense machineries involved in detection of invading pathogens. Upon recognition of pathogens, TLRs activate downstream kinases and transcription factors that induce the expression of genes involved in innate and adaptive immune responses. All TLRs, with the exception of TLR3, signal through the adaptor molecule myeloid differentiation factor 88 (MyD88). MyD88 is also critical for signaling through cytokine receptors that belong to the IL-1R family (10). A protective role for MyD88 in the control of fungal infections such as those caused by Candida albicans, Aspergillus fumigatus, Cryptococcus neoformans, and Paracoccidioides brasiliensis has been reported (11–14). Moreover, our laboratory and others have demonstrated that MyD88-dependent signaling is required for optimal alveolar epithelial cell (AEC) and alveolar macrophage (AM) cytokine responses to Pneumocystis or Pneumocystis cell wall components (15, 16). TLRs, including TLR2 and TLR4, have also been linked to Pneumocystis-stimulated AM cytokine responses (17, 18), but neither TLR2 nor TLR4 is required for AEC chemokine responses to Pneumocystis. Rather, the IL-1R is the upstream molecule required for the MyD88-dependent AEC response (16).

Although in vitro studies suggest that TLR-, IL-1R–, and MyD88-dependent responses are involved in the AEC and AM responses to Pneumocystis, the in vivo role of MyD88-dependent signaling events during active Pneumocystis infection remains undefined. In the current study, we used wild-type (WT) and MyD88-deficient mice to assess the role of MyD88 in host defense against Pneumocystis infection and/or the immunopathogenesis of PcP.

CB.17 SCID and C57BL/6 WT mice were bred at the University of Rochester, Rochester, NY. C57BL/6 MyD88^−/− mice were generously provided by Dr. S. Akira (Osaka University, Osaka, Japan) (19, 20). All animal protocols were approved by the University Committee for Animal Research at the University of Rochester Medical Center.

Pneumocystis was isolated from the lungs of heavily infected SCID mice and enumerated by Gomori’s methenamine silver staining, as previously described (21). Each Pneumocystis preparation was tested to ensure no bacterial contamination was present.

For the immunocompetent mouse model, 6- to 8-wk-old mice were anesthetized and 1e6 freshly isolated Pneumocystis (based on cyst count) were inoculated directly into the lungs via the trachea. For the immunosuppressed mouse model of PcP, CD4⁺ T cells were depleted by i.p. injections of 250 μg per mouse of mAb specific for mouse CD4 (clone GK1.5, ATCC TIB207). Injections were given twice per week, starting 1 wk prior to infection and continuing throughout the experiment. Depleted mice were inoculated with 5e5 freshly isolated Pneumocystis (based on cyst count).

Dynamic lung compliance and lung resistance were measured in live ventilated mice using a whole-body plethysmograph (BUXCO Electronics, Wilmington, NC) connected to a Harvard rodent ventilator (Harvard Apparatus, South Natick, MA), as previously described (22). Dynamic lung compliance was normalized to the peak body weight of each animal. Respiratory rates were measured using whole-body unrestrained chambers (BUXCO Electronics). Data were collected and analyzed using the Biosystems XA software package (BUXCO Electronics).

At indicated time points, mice were anesthesized, exanguinated, and perfused through the heart. Right lung lobes were lavaged with three 1-ml aliquots of 1× HBSS (Life Technologies, Invitrogen), and then snap frozen in liquid nitrogen and stored at −80°C. The bronchoalveolar lavage (BAL) fluid was centrifuged for 10 min at 150 × g at 4°C, and supernatants were frozen at −80°C. Recovered BAL cells were enumerated in the presence of trypan blue, using a hemacytometer (Hausser Scientific) and centrifuged onto glass slides for differential staining (Diff-Quick). Differential counts were performed by microscopic examination of the slides. Multiparameter flow cytometric analyses were also performed on BAL samples. Anti-CD4–Fluorescein (clone RM4-4) and anti-CD8a–Peridinin Chlorophyll-α Protein (clone 53-6.7) were purchased from BD Biosciences (San Diego, CA). The anti-CD4 clone RM4-4 was used to confirm CD4⁺ cell depletion in vivo in spleen samples because it is not blocked by the CD4-depleting Ab (clone GK1.5). Cells were analyzed on a FACSCalibur (BD Biosciences).

Real-time quantitative PCR method was used to quantify lung Pneumocystis burden in lung homogenate preparations. Right lung lobes were homogenized, subjected to three freeze/thaw cycles, boiled for 20 min, and then centrifuged at maximum speed. The lung homogenate supernatants were used for quantification of Pneumocystis burden. Quantitative PCR using TaqMan primer/fluorogenic probe chemistry was performed with a primer/probe set specific for the single-copy mouse Pneumocystis kexin gene, as previously described (22). The Applied Biosystems Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA) was used to quantify Pneumocystis kexin gene copies.

Pneumocystis-specific IgG and IgM in mouse serum were determined by ELISA, as described previously (23, 24). Flat-bottom microtitration plates (Flow Laboratories, McLean, VA) were coated with mouse Pneumocystis soluble total protein from the lungs of infected SCID mice (100 μg protein per milliliter) or protein from uninfected mouse lung. After blocking, using 5% nonfat dry milk in PBS, test sera were diluted 1:25 in PBS–0.05% Tween 20. After washing the plates, goat anti-mouse IgG + IgM (H + L) alkaline phosphatase (Jackson ImmunoResearch Laboratories) diluted 1:5000 was used as the secondary Ab. The plates were developed with p-nitrophenylphosphate as a substrate, and OD was measured at 405 nm. The background OD from uninfected lung protein–coated plates was subtracted from Pneumocystis-infected mouse lung protein–coated plates.

Left lung lobes were inflated with 15 cm gravity flow-pressure of 10% formalin (Sigma, St. Louis, MO). The lungs were fixed in situ for 10 min, removed from the mouse, and placed in fixative solution for 16 h at 4°C. Lung tissue was embedded in paraffin, and 4-μM sections were cut. Lung sections were stained with H&E to visualize tissue.

BAL fluid was collected and centrifuged at 12,000 × g for 10 min to remove cell debris. BAL fluids were stored at −80°C. Chemokine levels of CCL2 and CXCL2 and cytokine levels of TNF-α, IL-5, IL-17, IFN-γ, IL-1α, and IL-1β were determined by commercially available ELISA kits (R&D Systems) and used according to the manufacturer’s instructions.

Bone marrow transplant (BMT) chimeras were generated as previously described (25). Briefly, (donor → recipient) WT→WT, KO→ WT, WT→KO, and KO→KO (WT = C57BL/6; KO = C57BL/6 MyD88^−/−) BMT chimeras were prepared by radioablation by total-body irradiation of female recipients (6 Gy × 2 doses), followed by reconstitution with male donor bone marrow (Shepherd irradiator, 6000 Ci [¹³⁷Cs] source). To extract bone marrow from donors of the respective mouse strains, femurs and tibias were flushed into HBSS with 1% FCS. Cells were dispersed through a 21-gauge needle and pooled by strain. Erythrocytes were removed by hypotonic lysis. The cells were counted, resuspended, and delivered to recipient mice by tail vein injection of 1 × 10⁷ cells per mouse in a volume of 100 μL. After BMT, animals were allowed to reconstitute for 8–10 wk under microisolator conditions, supplied with high-efficiency particulate filtered air, sterilized food, acid water, and bedding. Fluorescent in situ hybridization to detect the Y chromosome was used on BAL cells from some mice to confirm the male origin of cells transplanted to female recipients, as previously described (26).

Data are presented as mean ± 1 SE measurement. In some experiments, differences between strains at different time points were analyzed using two-way ANOVA and the Bonferroni multiple comparison test as a posttest. Differences between infected animals and control uninfected animals were analyzed using one-way ANOVA using the Tukey multiple comparison test as a posttest. Differences were considered significant at p < 0.05. All data were analyzed using GraphPad Prism (version 5.00) software (GraphPad Software, San Diego, CA).

The MyD88 adaptor molecule functions in the generation and maintenance of innate and adaptive immune responses. However, the specific contribution of MyD88 to host defense against respiratory infection with Pneumocystis is unknown. To address this question, nonimmunosuppressed C57BL/6 WT and MyD88-deficient mice were infected with freshly isolated Pneumocystis organisms, and the organism burden over time was determined. As expected, immunocompetent WT mice harbored Pneumocystis organisms in the lung for ∼ 3 wk before an adaptive immune response cleared the infection (Fig. 1A, 1B). However, MyD88-deficient mice had significantly higher Pneumocystis burdens than WT mice on day 14 postinfection (Fig. 1A). At later time points no difference was observed between WT and MyD88-deficient mice with respect to lung burden, and the MyD88-deficient mice were competent to mount an effective immune response, which included the production of anti-Pneumocystis Ab (Fig. 1B). These results demonstrated that MyD88 plays a role in the control of Pneumocystis growth early postinfection but that MyD88 is not essential for effective adaptive immunity, the production of anti-Pneumocystis Ab, or the clearance of Pneumocystis from the lung.

In addition to monitoring fungal growth and clearance, pulmonary function was also measured to determine whether loss of MyD88 affected lung function independent of changes in Pneumocystis burden. As expected, immunocompetent WT mice showed little evidence of respiratory impairment during Pneumocystis growth and immune-mediated clearance. Furthermore, lung compliance and lung resistance measurements were very similar in WT and MyD88-deficient mice at all time points examined, and were not statistically different from those in uninfected control mice (Fig. 2A, 2B). In addition, neither group lost weight during Pneumocystis infection, which is another indicator of PcP-related respiratory disease (data not shown). These findings confirm that MyD88 is not required for effective host defense against Pneumocystis, and that loss of MyD88 does not render mice more susceptible to lung injury associated with clearance of Pneumocystis by the normal immune system.

Although no obvious respiratory consequences of Pneumocystis infection were observed in MyD88-deficient mice, it remained possible that the nature or magnitude of the immune response to Pneumocystis was altered by the absence of MyD88. Total cells recovered in the BAL fluid of WT and MyD88-deficient mice were enumerated on days 14, 18, 21, and 25 postinfection. Pneumocystis-infected WT mice had a significantly greater number of total BAL cells than did uninfected mice at day 14 postinfection, and this number decreased over time, coincident with Pneumocystis clearance (Fig. 3A). Of note, Pneumocystis-infected MyD88 mice showed a similar increase in the number of total BAL cells following infection. However, the number of BAL cells recovered from MyD88 mice, in contrast to WT mice, did not diminish over the course of the study (Fig. 3A). Pneumocystis-infected MyD88 mice had more lymphocytes, macrophages, eosinophils, and polymorphonuclear neutrophils (PMNs) than did Pneumocystis-infected WT mice at all time points. These differences reached statistical significance on day 21 postinfection for all cell types, and additionally on day 25 for PMNs (Fig. 3B–E).

To determine whether the altered immune response in Pneumocystis-infected MyD88-deficient mice was associated with an altered pattern of cytokine and chemokine production, protein levels were measured in BAL fluid of experimental mice. MyD88-deficient mice had statistically higher lung levels of the chemokine CCL2 than did WT mice on days 14 and 18 postinfection, although on day 25 both strains have similar baseline levels of the chemokine (Fig. 4A). These results correlated with increased numbers of lymphocytes and macrophages in the lungs of Pneumocystis-infected MyD88-deficient mice (Fig. 3B, 3C). IL-4 levels in the BAL of MyD88-deficient mice, compared with WT mice, were also significantly elevated on days 14, 18, and 21 postinfection but decreased to basal levels on day 25 (Fig. 4B), suggesting a Th2-associated immune response. IL-5 levels were also elevated relative to WT mice in the lungs of MyD88-deficient mice and peaked at day 21 (Fig. 4C), coincident with the peak number of eosinophils in the lungs (Fig. 3E). Furthermore, IL-17 levels were also elevated in the lungs of Pneumocystis-infected MyD88-deficient mice compared with WT (Fig. 4D), and were associated with increased numbers of neutrophils (Fig. 3D). No significant difference in TNF-α levels was noted, although MyD88-deficient mice showed a trend toward higher TNF-α levels on days 14, 18, and 21 postinfection (Fig. 4E). IFN-γ levels in the BAL fluid were very low and nearly at the limit of detection in both strains, and no difference was observed at any time point (Fig. 4F). These results further confirm that loss of MyD88 signaling alters the normal immune response to Pneumocystis, including more CCL2, more Th2-associated IL-4 and IL-5, and more Th17-associated IL-17. These results suggest that MyD88 helps to regulate the degree of inflammation during a normal immune response to Pneumocystis infection and that MyD88 deficiency results in exacerbated Th2 and Th17 responses following Pneumocystis infection.

MyD88 plays a role in the early inflammatory response to Pneumocystis (16). To investigate the role of MyD88 in the immunopathogenesis of PcP, a model of AIDS-related immunosuppression was used to induce susceptibility to PcP. WT and MyD88 deficient mice were depleted of CD4⁺ T cells and infected with freshly isolated Pneumocystis organisms. During the progression of the infection, body weights and respiratory rates were measured noninvasively to assess overall health and disease progression. WT and MyD88-deficient mice exhibited no weight loss over the first 4 wk of infection, but both began to show weight loss after 4 wk of infection. However, MyD88-deficient mice displayed a much more rapid weight loss than did WT mice at this time (Fig. 5A). Differences in respiratory rates were also noted. WT mice displayed increased respiratory rates compared with MyD88-deficient mice during the first 3 wk of infection (Fig. 5B). However, between 3 and 5 wk postinfection, the MyD88-deficient mice displayed a much more dramatic elevation in respiratory rate than did WT mice. The respiratory rates of MyD88-deficient mice rose ∼ 100%, compared with a 12% increase observed in WT mice over this period. By day 31 postinfection MyD88-deficient mice had higher respiratory rates than WT mice, with 514 ± 31 and 465 ± 46 breaths per minute, respectively.

To directly assess the contribution of MyD88 to PcP-related respiratory impairment, pulmonary function was measured on experimental mice at 4 and 5 wk postinfection. We chose two time points that reflected the distinct stages of disease in the MyD88 mice: prior to weight loss and respiratory rate increase (week 4), and when MyD88-deficient mice displayed signs of exacerbated disease compared with WT (week 5). At weeks 4 and 5 postinfection, WT mice showed significantly lower dynamic lung compliance and higher lung resistance relative to uninfected mice, which is characteristic of PcP (Fig. 6A, 6B). In contrast, at 4 wk postinfection MyD88-deficient mice were protected from the PcP-related lung function deficits observed in WT mice. The Pneumocystis-infected MyD88-deficient mice had significantly higher dynamic lung compliance than did WT infected mice, which was similar to lung compliance measurements in uninfected control mice (Fig. 6A). Similarly, MyD88-deficient mice also showed lung resistance measurements comparable to those of control uninfected mice at week 4 postinfection (Fig. 6B). However, between 4 and 5 wk postinfection, MyD88-deficient mice displayed a 50% drop in dynamic lung compliance and a 25% increase in lung resistance, signifying a dramatic acceleration in the severity of PcP in these mice over this period. These values were significantly different from those in uninfected mice, and worse than those in infected WT mice (Fig. 6A, 6B). Together these data demonstrate that MyD88 signaling contributes to PcP-associated respiratory impairment early during infection but protects at later stages of disease. Thus, MyD88 signaling likely plays distinct roles during different stages of PcP.

To determine whether MyD88 plays a role in controlling Pneumocystis growth in a CD4⁺ T cell–depleted model of PcP, lungs of WT and MyD88-deficient mice were tested for Pneumocystis burden using quantitative real-time PCR, as we have described (22). No difference in Pneumocystis burden was noted between WT and MyD88-deficient mice at 4 wk postinfection (Fig. 7). However, by 5 wk postinfection, the lungs of MyD88-deficient mice contained nearly 2-fold more Pneumocystis organisms than did those of WT mice (Fig. 7). These data show that MyD88-dependent mechanisms help to control fungal burden in CD4⁺ T cell–depleted mice.

To determine whether the differences observed in overall health and lung function between Pneumocystis-infected MyD88-deficient and WT mice correlated with the degree of pulmonary inflammation, total cell counts in BAL fluid were performed. Pneumocystis-infected MyD88-deficient mice had significantly fewer BAL cells than did infected WT mice at week 4 (Table I). Significantly fewer numbers of macrophages and CD8⁺ T cells were present in the lung of MyD88-deficient mice at this time. Both of these cell types are effectors of PcP-related immunopathogenesis, and reduced numbers in the lungs of MyD88-deficient mice are likely factors in the preserved lung function observed in these mice. At week 5 postinfection, no significant difference was observed in total cells or any specific cell type recruited to the lungs of Pneumocystis-infected WT or MyD88-deficient mice, suggesting that mice can compensate for the loss of MyD88, at least with respect to cell recruitment to the lung.

Histological examination of lung sections from experimental mice demonstrated increased cellular infiltration and inflammation in the alveolar region of Pneumocystis-infected WT mice at 4 wk postinfection, compared with uninfected lungs (Fig. 8A, 8B). In contrast, the lungs of Pneumocystis-infected MyD88-deficient mice displayed fewer alveolar infiltrates and less inflammation than WT mice at this time. Furthermore, although the cell infiltrates appeared dispersed throughout the alveoli in WT mice, they were localized mainly to peribronchial regions of the MyD88-deficient lungs, with less involvement of the alveolar regions. At week 5 postinfection, the lungs of Pneumocystis-infected MyD88-deficient mice showed more signs of inflammation and cell recruitment throughout the lung, similar to that observed in WT mice (Fig. 8C). Of interest, the lungs of MyD88-deficient mice displayed more foamy exudates filling the alveolar spaces than were evident in WT mice (Fig. 8C, 8D). These results demonstrate that MyD88 regulates cell recruitment and inflammation in the lung during PcP.

PcP-related immunopathogenesis is characterized by an increase in cytokine and chemokine production in the lungs. To determine whether the reduced cell-mediated pulmonary inflammation and preserved pulmonary physiology in the Pneumocystis-infected MyD88-deficient mice early during infection is associated with reduced chemokine and/or cytokine production in the lung, ELISA was used to evaluate BAL fluid from experimental mice. As expected, CCL2 levels were increased in the lungs of Pneumocystis-infected WT mice compared with uninfected controls. Also consistent with our earlier work (16), CCL2 levels were significantly reduced in the lungs of MyD88-deficient mice compared with WT mice at 4 wk postinfection (Fig. 9A), which likely explains our finding of fewer macrophages and CD8⁺ T cells in the lungs of MyD88-deficient mice at this time (Table I). However, by 5 wk postinfection CCL2 levels were similar in the lungs of MyD88-deficient and WT mice (Fig. 9A), indicating that CCL2 production can proceed via a MyD88-independent mechanism at later time points. Surprisingly, CXCL2, TNF-α, and IFN-γ levels were higher in the lungs of Pneumocystis-infected MyD88-deficient mice compared with levels in WT lungs (Fig. 9B–D), and elevated TNF-α and IFN-γ in the lungs at 5 wk postinfection likely contribute to the MyD88-independent inflammation observed at this time. IL-1α and IL-1β levels were also higher in infected MyD88-deficient mice than in WT mice (Fig. 9E, 9F), which is likely related to lack of negative feedback regulation owing to impaired IL-1R signaling. These data demonstrate that deletion of MyD88 causes dysregulated cytokine and chemokine responses in the lung during PcP.

To determine whether the MyD88-dependent mechanisms required for the control of Pneumocystis burden rely on lung resident cells or hematopoietic cells, we generated bone marrow chimera mice. WT and MyD88^−/− mice were lethally irradiated and reconstituted with WT or MyD88^−/− bone marrow cells. After 8–10 wk postreconstitution, mice were CD4 depleted and infected with Pneumocystis. At week 4 postinfection, no statistically significant differences in Pneumocystis lung burden were found between any of the groups, although WT mice reconstituted with MyD88^−/− cells [knockout (KO)→WT] showed higher organism burden (Fig. 10). On week 5 postinfection, irradiated WT mice that were reconstituted with WT bone marrow–derived cells (WT→WT) presented with Pneumocystis counts that were similar to those in MyD88^−/− mice reconstituted with WT cells (WT→KO). However, WT mice that received MyD88^−/− bone marrow–derived cells (KO→WT) were unable to control Pneumocystis growth, as demonstrated by significant increase in organism burden in the lungs of these mice (Fig. 10). MyD88^−/− mice reconstituted with MyD88^−/− (KO→KO) bone marrow cells showed slightly higher Pneumocystis burden than did WT→WT mice, but significantly lower burden than KO→WT (Fig. 10), suggesting that MyD88 signaling in hematopoietic and parenchymal cells may have opposing effects on control of Pneumocystis infection. These data suggest that the MyD88 signaling pathway in hematopoietic derived cells is important for control of organism burden during Pneumocystis infection in the immunocompromised host.

The MyD88 signaling pathway plays an important role in the generation of effective innate and adaptive immune responses to a variety of pathogens. In vitro studies have determined that MyD88 is required for AEC and AM chemokine and cytokine responses to Pneumocystis (15, 16). However, the role of MyD88-dependent signaling during Pneumocystis infection and immunity in vivo has not been determined. In this study, we expanded upon our previous work by 1) identifying a regulatory role for MyD88 during the normal adaptive immune response against Pneumocystis; and 2) demonstrating that MyD88 signaling participates in PcP-related immunopathogenesis. Although MyD88 signaling is not required for resistance to Pneumocystis infection in immunocompetent mice, it does help control fungal burden early during infection and limits the subsequent protective immune response. In the CD4-depleted AIDS-related model of active PcP, MyD88 contributes to inflammation, cell recruitment, and impaired pulmonary function during the early stages of PcP. However, at later stages of disease, MyD88 signaling limits inflammation and lung injury, and helps control Pneumocystis burden. These results suggest that MyD88 regulates the intensity of the immune response to Pneumocystis in immunocompetent hosts. However, MyD88 plays a dual role during active PcP by contributing to immunopathogenesis at early stages, but protecting against lung injury and uncontrolled growth of the organism at later stages.

Our results suggest that MyD88-dependent mechanisms control Pneumocystis burden early during the CD4⁺ T cell–mediated immune response. Even though slightly higher numbers of macrophages were found in the lungs of MyD88^−/− mice early, at day 14 postinfection, it seems that the MyD88-deficient macrophages were less efficient in controlling the fungal burden. MyD88 deficiency may prevent effective killing of Pneumocystis by AMs, either by defects in phagocytosis or through impaired production of antimicrobial agents such as reactive oxygen or nitrogen species. Other studies have shown similar increased susceptibility of Pneumocystis infection in dectin-1– or TLR2-deficient mice, and have suggested a role for reactive oxygen and nitrogen species in the clearance of Pneumocystis (27, 28). Of note, a similar pattern of increased fungal burden early during infection, followed by effective resolution, has been observed in MyD88^−/− mice infected with Aspergillus (12). Importantly, MyD88^−/− mice were able to clear Pneumocystis infection similarly to WT mice, suggesting that MyD88 is not required for host defense against this infection. The ability of MyD88^−/− mice to clear Pneumocystis is probably due to an effective adaptive response and production of Pneumocystis-specific Ab comparable to that found in WT mice. Production of Pneumocystis-specific IgG and IgM Ab has been shown to correlate with clearance of Pneumocystis organisms (23, 29, 30).

In addition, we found that MyD88 regulates the intensity and resolution of the Pneumocystis-specific immune response in nonimmunosuppressed mice. Although MyD88^−/− mice were able to clear the infection, the mice displayed an exacerbated immune response characterized by elevated CCL2, which correlated with higher numbers of lymphocytes and macrophages recruited to the lung, and an increase in the Th2-related cytokines IL-4 and IL-5, which may explain increased lung eosinophilia. Of interest, MyD88^−/− mice that were infected with the pathogens Legionella pneumophila, Salmonella tyhimurium, Chlamidya muridarum, or Leishmania major all presented polarized Th2 responses that were associated with enhanced disease (31–34). In addition, levels of the Th17 cytokine, IL-17, were also elevated in MyD88^−/− mice, which correlated with higher numbers of neutrophils recruited to the lungs. Increased inflammation and lung pathology in Pneumocystis-infected nude IFNγ^−/− mice correlated with higher levels of IL-17 in the lung (35). It should be noted that despite displaying an exacerbated immune response against Pneumocystis, nonimmunosuppressed MyD88^−/− mice did not exhibit reduced pulmonary function relative to WT mice. We speculate that the recruited MyD88-deficient inflammatory cells are less activated than WT cells, and thus do not have profound effects on lung function.

Recent studies have suggested a role for TLR2 and TLR4 in the in vivo responses to Pneumocystis in a CD4-depleted, AIDS-related model of PcP (18, 27). However, both studies suggest that neither TLR2 nor TLR4 is not the sole pathway responsible for immunopathogenesis during Pneumocystis infection. Our results using infected MyD88^−/− mice show an increase in lung TNF-α and CXCL2, compared with WT mice, which is consistent with results reported using a model of pulmonary Aspergillus infection (12). Conversely, our results show a decrease in CCL2 in the BAL of CD4-depleted MyD88^−/− mice with PcP. CCL2 (or MCP-1), is a member of the CC chemokine family that recruits macrophages and T cells to sites of infection (36–39). Macrophages and CD8⁺ T cells contribute to PcP-related immunopathogenesis, and reduced CCL2 in MyD88^−/− mice correlated with fewer macrophages and CD8⁺ T cells recruited to the lungs and preserved lung function (9, 40, 41). These results suggest that MyD88 mediates the inflammatory response and lung injury early during infection through CCL2-dependent recruitment of CD8⁺ T cells and macrophages. MyD88 also regulates other chemokines and cytokines, such as CXCL2, TNF-α, IFN-γ, IL-1α, and IL-1β (Fig. 9). Our data also suggest that additional MyD88-dependent receptors other than TLR2 or TLR4 may be involved in the response to Pneumocystis. For example, IL-1R was found to mediate the chemokine response to Pneumocystis in primary AECs, and may play a role in the immunopathogenesis of PcP (16). Of note, IL-1β is required for effective resolution of Pneumocystis in an immune-reconstitution model (42). Most likely, various TLRs and IL-1R play important and perhaps somewhat redundant roles in the immunopathogenesis of PcP. For this reason, our study attempted to determine the role during PcP of the adaptor molecule MyD88, which participates in the TLR/IL-1 pathways.

MyD88 is required for optimal AM TNF-α responses to Pneumocystis β-glucan (15), and the MyD88-dependent pattern recognition receptors TLR2 and TLR4 also contribute to AM chemokine and cytokine responses to Pneumocystis (18, 27). In vivo studies found that CD4-depleted TLR2^−/− mice harbored higher Pneumocystis burdens than did WT mice, similar to our findings in MyD88^−/− mice (27). AECs also respond to Pneumocystis, and we recently reported that the AEC chemokine response is dependent on MyD88 and IL-1R, but independent of TLR2 or TLR4 (16). It is interesting that our chimeric studies determined that loss of AEC MyD88 signaling did not affect fungal burden in CD4-depleted mice, whereas loss of hematopoietic cell MyD88 caused increased burden. These findings indicate that AEC responses may not directly contribute to control of Pneumocystis burden during active PcP in immunosuppressed mice. However, it remains possible that AEC responses contribute to inflammatory cell recruitment and PcP-related immunopathogenesis through MyD88-dependent mechanisms. Further studies are required to explore this possibility.

MyD88-independent pathways may also limit the inflammatory response during active PcP in CD4-depleted mice, especially at later stages of disease. At 5 wk postinfection, when MyD88^−/− mice developed severe inflammation and exhibited signs of impaired lung function, BAL TNF-α levels were significantly higher than in WT mice. Excessive TNF-α is associated with inflammatory cell recruitment, uncontrolled activation of proinflammatory pathways, and endothelial leakiness. A previous study from our group showed that TNFR signaling drives pulmonary inflammation and respiratory impairment in Pneumocystis-infected CD4-depleted mice (22). Although this study found that the presence of CD8⁺ T cells is required for maximal TNF production during PcP in CD4-depleted mice (22), CD8⁺ T cell–independent mechanisms of TNF production may also contribute to elevated lung levels in MyD88^−/− mice. For example, excessive lung injury in MyD88^−/− mice attributable to higher organism burdens at 5 wk postinfection could cause elevated TNF-α release. A similar outcome was observed in a study in which lymphocyte deficient SCID mice were infected with Pneumocystis. These mice produced very little lung TNF-α until the very late stages of disease, when lungs were injured and the Pneumocystis burden was very high (42). It is possible that various redundant pathways work in cooperation to detect and respond to Pneumocystis. For example, dectin-1 has been found to cooperate with TLR2 to stimulate TNF-α production in macrophages exposed to mycobacterial infection (43). In addition, macrophage mannose receptor–deficient mice infected with Pneumocystis, compared with WT mice, showed increased total protein, albumin, and inflammatory cells in lungs (44). Recently, a role for intracellular osteopontin in the control of Pneumocystis burden in immunocompromised mice was reported. The authors suggested that osteopontin clusters TLR2, dectin-1, and mannose receptor in macrophages exposed to Pneumocystis (45). It is possible that MyD88-independent pathways cooperate with the MyD88 pathway during the host response to Pneumocystis, although more studies are required to confirm these interactions.

Our data demonstrate that MyD88 signaling in hematopoietic cells is required to control Pneumocystis infection during active PcP in CD4-depleted mice. These results agree with recent findings showing that MyD88 signaling in the hematopoietic compartment was required for control of Legionella pneumophila (46). This appears to be dependent on the specific pathogen, because MyD88 expression in both resident and hematopoietic cells contributes to control of Klebsiella pneumoniae infection. In contrast, resident nonhematopoietic cells were important for control of Pseudomonas aeruginosa growth (47, 48). Although we have clearly shown that MyD88 signaling modulates pulmonary immunity and contributes to PcP-related immunopathogenesis, many questions remain. For example, neither the exact cell types that use MyD88 signaling during Pneumocystis infection nor the specific MyD88-dependent mechanisms that regulate host defense and immunopathogenesis have been defined. MyD88-dependent responses are known to regulate both innate and adaptive immunity, and many cell types, including epithelial cells, macrophages, neutrophils, lymphocytes, and DCs, use the MyD88 signaling pathway during the course of an immune response. Additional studies are needed to define the cell type–specific role of MyD88 in the complicated network of immune interactions that occur in response to Pneumocystis in vivo.

The blockade of MyD88 signaling is being explored as a possible therapeutic strategy to attenuate inflammatory disease (49, 50). Our preliminary findings suggest that MyD88 may represent a potential target for immunotherapy in Pneumocystis-infected patients. Because lack of MyD88 early during PcP limits inflammation and promotes better lung function, it is conceivable that blocking the MyD88 pathway, in combination with effective antibiotic treatment, may improve disease outcome. MyD88 represents a more specific pathway that might be blocked temporarily while antibiotic treatment to Pneumocystis is administered. This feature is in contrast to corticosteroids, which sometimes cause undesired systemic side effects. However, we also found that loss of MyD88 signaling could have negative effects on the severity of PcP at later stages of disease by increasing fungal burden, which raises concerns regarding the therapeutic blockade of MyD88 during PcP. In addition, the MyD88 and TLR pathways have been found to protect against lung injury through recognition of extracellular matrix components such as hyaluronan. Hyaluronan degradation products present during lung injury promote lung repair and protect the lung epithelium through interactions with TLRs (51, 52). Thus, we speculate that the later deterioration in CD4-depleted MyD88-deficient mice with active PcP may be partly related to loss of MyD88-dependent protective mechanisms that function during lung injury. Further studies using MyD88 blocking agents in vivo are required to thoroughly evaluate the potential of targeting MyD88 for anti-inflammatory therapy during PcP.

We thank Dr. Samir Bhagwat, Dr. Nabilah Khan, and Jane Malone for helpful comments and expert technical assistance.

The authors have no financial conflicts of interest.