Pneumocystis Infection in an Immunocompetent Host Can Promote Collateral Sensitization to Respiratory Antigens

Abstract

Infection with the opportunistic fungal pathogen Pneumocystis is assumed to pass without persistent pathology in immunocompetent hosts. However, when immunocompetent BALB/c mice were inoculated with Pneumocystis, a vigorous Th2-like pulmonary inflammation ensued and peaked at 14 days postinfection. This coincided with a 10-fold increase in the number of antigen-presenting cells (APCs) in the lung, and these cells were capable of presenting antigen in vitro, as well as greater uptake of antigen in vivo. When mice were presented with exogenous antigen at the 14-day time point of the infection, they developed respiratory sensitization to that antigen, in the form of increased airway hyperresponsiveness upon a later challenge, whereas mice not infected but presented with antigen did not. Like other forms of collateral sensitization, this response was dependent on interleukin-4 receptor signaling. This ability to facilitate sensitization to exogenous antigen has been previously reported for other infectious disease agents; however, Pneumocystis appears to be uniquely capable in this respect, as a single intranasal dose without added adjuvant, when it was administered at the appropriate time, was sufficient to initiate sensitization. Pneumocystis infection probably occurs in most humans during the first few years of life, and in the vast majority of cases, it fails to cause any overt direct pathology. However, as we show here, Pneumocystis can be an agent of comorbidity at this time by facilitating respiratory sensitization that may relate to the later development or exacerbation of obstructive airway disease.

INTRODUCTION

The AIDS epidemic of the 1980s brought to light the previously obscure atypical fungus Pneumocystis as the cause of a serious and often fatal pneumonia in immunocompromised patients (54). Little was known about this organism at the time, but years of focused research have elucidated likely mechanisms of pathology in Pneumocystis pneumonia (reviewed in reference 14). In contrast to this increased knowledge of Pneumocystis in the immunosuppressed host, there is still very little known about the nature of Pneumocystis infection in the immunocompetent host. It is widely assumed that Pneumocystis causes a mild respiratory infection in human infants that results in persistent immunity. This is supported by multiple studies showing that over the first 2 to 3 years of life, greater than 80% of children seroconvert to positivity for Pneumocystis (41, 55). Furthermore, the lack of any correlation between clinical illness and a common definitive diagnosis of Pneumocystis infection speaks to the relative mildness and apparent lack of pathology associated with these infections (30). Nonetheless, a recent retrospective study of young children hospitalized for respiratory infections found that 16% tested positive for Pneumocystis, although those testing positive were twice as likely to have an upper respiratory tract infection (URTI) as opposed to a lower respiratory tract infection (LRTI) (32).

In spite of the fact that most childhood respiratory illnesses would appear to be relatively benign, there has been considerable research (and controversy) about the role of early respiratory infections in the subsequent development of chronic conditions such as asthma (39) and bronchitis (3). The majority of these studies have focused on early respiratory syncytial virus (RSV) infections (26, 34), due to the more frequent serious nature of these infections and the symptomatic association with wheezing. There is also some support for a role of infections with atypical bacteria in the causation of asthma (20, 31). However, despite the ubiquity of Pneumocystis and perhaps because of the mildness of its manifestation, the possibility that Pneumocystis infections may have such an impact has not been explored.

We previously observed that immunocompetent mice infected with Pneumocystis exhibit no external signs of illness except mild dyspnea and clear the infection with no obvious pathology. However, histological observation reveals a period of vigorous pulmonary inflammation during the period in which the Pneumocystis infection is cleared. We report here that at its peak, this inflammatory response results in a 10-fold increase in the number of antigen-presenting cells (APCs) in the lung and that these cells are capable of presenting exogenous antigen to CD4 T cells. Also at this time, the pulmonary inflammatory environment is becoming Th2 in character, and access of inhaled antigen to the interstitial lung compartment has increased. Furthermore, when mice were given an intranasal application of ovalbumin (ova) during this peak inflammatory response, they developed sensitization to ova, in the form of airway hyperresponsiveness upon later antigen challenge, without the requirement of previous extrapulmonary sensitization. This sensitization proved to be dependent on interleukin-4 (IL-4) signaling, which has previously been described to be a hallmark of collateral sensitization (10). We also found that the degree of exposure to the exogenous antigen (ova) required to cause sensitization was less during a Pneumocystis infection than what has been reported for other early respiratory infections, such as those caused by RSV (45). These observations, together with previous evidence for the common nature of Pneumocystis infections in children, suggest that Pneumocystis may have a potential role in the early development of chronic obstructive pulmonary conditions.

MATERIALS AND METHODS

Animals.

Many of the BALB/c mice and all IL-4 receptor-knockout (IL-4r KO) mice were bred in the research animal facility here at Montana State University from stock originally obtained from Jackson Laboratories (Bar Harbor, ME). The IL-4r KO mice are on a BALB/c background, Jackson stock number 003514, genotype BALB/c-Il4ra^tm1Sz/J. Additional BALB/c mice were purchased from the National Cancer Institute (Fredrick, MD). Mice were housed in isolation rooms inside ventilated cages receiving HEPA-filtered air and given autoclaved mouse chow and acidified water. Regular screening of local and vendor-supplied mice over the course of these experiments indicated negative exposures to common murine viruses. Periodic examination for anti-Pneumocystis serum antibodies and the lack of Pneumocystis pneumonia in immunocompromised mice kept in the same colonies verified the absence of unintentional infection with Pneumocystis.

Antibodies.

Most antibodies used for flow cytometry were from commercial sources. Anti-mouse CD4-allophycocyanin-Cy7, CD19-allophycocyanin, CD25-phycoerythrin (PE), CD80-fluorescein isothiocyanate (FITC), B220-allophycocyanin-Cy7, and major histocompatibility complex class II (MHC-II) (I^a/I^e)-PE were purchased from Becton Dickinson (Mountain View, CA). Anti-mouse CD8-PE-Cy7, CD11b-peridinin chlorophyll protein-Cy5.5, CD45-PE-Cy7, CD86-Alexa Fluor 700, and streptavidin-PE-Cy7 were purchased from Biolegend (San Diego, CA). CD204-Alexa Fluor 647 was purchased from AbD Serotec (Raleigh, NC). Anti-mouse CD103-biotin was purchased from eBioscience (San Diego, CA). Anti-mouse CD40-biotin was purchased from R&D Systems (Minneapolis, MN). Anti-mouse Gr-1 and CD11c antibodies were grown in-house from the hybridomas RB6-8C5 (R. Coffman, DYNAX, Palo Alto, CA) and HB-224 (ATCC, Manassas, VA) and conjugated directly to either FITC or Atto633 (Innova Biosciences, Babraham, United Kingdom).

Infection with Pneumocystis.

Pneumocystis murina organisms were maintained by serial infection of C.B17 scid/scid mice kept in our colony here at Montana State University. For inoculation, lungs from infected source mice were homogenized, Pneumocystis organisms in the homogenates were quantified and inspected for bacterial contamination, and volumes were adjusted to 10⁸ Pneumocystis nuclei per ml. Experimental mice were lightly anesthetized with isofluothane, and 0.1 ml of homogenate was administered via intratracheal inoculation (51). All mice in a given experiment were infected from the same batch of Pneumocystis organisms. In the past, we have compared responses in sham-infected mice given homogenates from uninfected SCID mouse lungs and found only a mild transitory response for 1 to 2 days, such as a slight alveolar influx of macrophages and neutrophils. At 4 days after inoculation, there were no significant differences between sham-infected and untreated mice in terms of inflammatory cell numbers (macrophages, lymphocytes, neutrophils, or eosinophils), inflammatory cytokine levels in the bronchoalveolar lavage fluid (BALF; IL-4, IL-10, gamma interferon [IFN-γ], tumor necrosis factor alpha [TNF-α], IL-12, or monocyte chemoattractant protein 1), or measures of tissue damage, including BALF albumin and lactate dehydrogenase and airway hyperresponsiveness (AHR) (data not shown). For this study, we also determined whether the levels of MHC-II^hi CD11c⁺ cells were elevated in the lungs of sham-infected mice at 14 days postinoculation. We found that there was no significant difference between sham-infected and uninfected mice in this respect. For these reasons, we elected to use untreated mice as controls to avoid unnecessary killing of donor mice.

Serum collection and bronchoalveolar lavage.

After respiratory measurements were taken, mice were deeply anesthetized with pentobarbital and killed by exsanguination. The blood was collected and allowed to clot at room temperature for 20 min, after which the samples were centrifuged and serum was collected and frozen at −80°C until analysis. For bronchoalveolar lavage, the trachea was nicked with fine scissors, and then a 16-cm length of Micro-Line tubing attached to a 5-ml syringe was inserted. Five 1-ml aliquots of sterile Hanks balanced salt solution (HBSS) with 3 mM EDTA were then used to lavage the alveolar contents (21). Cytospin preparations (100 μl) were made from each lavage fluid sample using a cytospin centrifuge, and then the preparations were stained with Diff-Quick dye (Dade Behring, Newark, DE). Differential cell counts were later determined microscopically using a ×100 objective lens. Additional aliquots of the BALF were taken in order to count total BALF cells using a hemocytometer. The BALF supernatant was collected by centrifugation at 900 × g for 10 min, and aliquots were saved at −80°C for any subsequent assays. BALF cells were resuspended in 100 to 200 μl of Fc Block (Dulbecco's phosphate-buffered saline [DPBS] with 2% calf serum and anti-mouse Fc receptor antibody [Trudeau Institute, Saranac Lake, NY]) to block nonspecific binding.

Lung digestion.

To obtain pulmonary interstitial cells, each lung was removed and minced into 3-mm cubes on a sterile petri dish, and the cubes were placed into 10 ml of RPMI containing 0.2% collagenase (LS004196; Worthington, Lakewood, NJ) and 50 units DNase I (D4263; Sigma Chemical, St. Louis, MO). This mixture was placed in a sterile 50-ml Erlenmeyer flask with a sterile stir bar and agitated for 90 min at 37°C. The digested lung was then passed through a 70-μm-mesh-size nylon mesh and centrifuged at 900 × g for 10 min. The resultant cell pellet was resupended in 5 ml of ammonium-chloride-potassium (ACK) lysis buffer, incubated for 5 min at room temperature, diluted to 30 ml with 5% calf serum in DPBS, and centrifuged at 900 × g for 10 min. After one additional wash, the cells were resuspended in 0.5 ml of Fc Block and filtered through a 50-μm-mesh-size mesh, aliquots were removed for cell counts, and the remainder was used for flow cytometry.

Flow cytometry.

BALF cells were stained with a mixture of fluorophore-conjugated antibodies against the mouse CD antigens CD4, CD8, CD19, and CD25 and then examined on a FACSCantos or LSR II flow cytometer (Becton Dickinson, Mountain View, CA). Lymphocytes obtained from the spleens of control uninfected mice were used for instrument setup. Analysis of cytometry data was performed with FlowJo software (Ashland, OR). The numbers of relevant cell types (e.g., CD4 lymphocytes) in BALF were determined by multiplying flow cytometry data (percentage of a given cell type) with BALF cell counts. Lung digest cells were stained with a mixture of CD45, CD11c, and MHC-II, together with a combination chosen from CD11b, CD103, CD204, CD40, CD25, B220, Gr-1 (RB6), CD80, and CD86. For analysis, cells were first gated on the basis of CD45 (common leukocyte antigen) staining to discriminate immune cells, and then CD45⁺ cells were gated into CD11c-positive or -negative groups. CD45⁺ CD11c⁺ pulmonary cells were then analyzed for relative expression of the other markers described above.

Cytokine analysis.

Cytokine concentrations in undiluted samples of BALF were assayed using cytometry-based bead enzyme-linked immunosorbent assay (ELISA) kits (either mouse Th1-Th2 or mouse Th1-Th2-Th17; Becton Dickinson, Mountain View, CA). A standard ELISA was used to measure the levels of IL-5 and IL-13 (eBioscience). Standard ELISA techniques were also used to measure ova-specific IgE or IgG antibody titers in serum or BALF samples. Aliquots were added to plates precoated with ova, and then, after incubation and washing, biotinylated rat anti-mouse IgE (Becton Dickinson) or IgG (Sigma Chemical) was applied. Later, an appropriate streptavidin-horseradish peroxidase conjugate was applied, after which a substrate was added and developed until the optical density was read with a plate spectrophotometer.

In vitro cell proliferation assay.

To examine whether pulmonary CD11c⁺ cells were capable of antigen-specific stimulation, an in vitro assay was used. Isolated CD11c⁺ cells were obtained from pulmonary digests by labeling with PE anti-CD11c antibody, followed by sterile cytometric sorting using a FACSDiva cell sorter (Becton Dickinson). Ovalbumin-specific CD4 cells were obtained from the spleens of D011.10 mice (Jackson Laboratory, Bar Harbor, ME) using negative-selection columns (R&D Systems, Minneapolis, MN), following the manufacturer's instructions. Coculture of these cells (in triplicate) was performed in 96-well plates, following the method of Masten and Lipscomb (35). At the end of this period, fresh splenocytes were obtained from another mouse, cells in numbers over the range of a standard curve (0 to 1 × 10⁶ cells) were added in triplicate to the plate, and then all wells were immediately harvested and cells were enumerated using a Vybrant 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide cell proliferation assay (Molecular Probes, Eugene, OR). Magnetic bead isolation of CD11c⁺ cells (Miltenyi Biotec, Auburn, CA) also provided results comparable to those obtained when fluorescent-activated cell sorting was used for isolation.

Enumeration of Pneumocystis nuclei.

As necessary, Pneumocystis nuclei (cysts and trophozoites) in cytospin preparations of lung homogenates were enumerated as previously described (51).

Histology and morphometry.

One-third of the lung (the large left lobe) was fixed for 24 h in phosphate-buffered formalin, embedded in paraffin, sectioned at 5 μm, and stained either with hematoxylin-eosin (H&E) or in some cases with Alcian blue with periodic acid-Schiff stain using standard histological techniques.

Antigen exposure and challenge.

To examine the effects of Pneumocystis infection on sensitization to ovalbumin, the following procedure was used. At 10, 14, 21, and 28 days after inoculation, mice were lightly anesthetized and a single dose of 50 μg of ovalbumin in 50 μl sterile HBSS was applied intranasally (i.n.). At 30 to 35 days after this exposure, mice were challenged on three successive days with a single i.n. dose of 20 μg ova in 50 μl of HBSS. On the day after the last antigen challenge, mice were subjected to measurements of AHR during a methacholine challenge, as described below. For a positive control, one group of mice was injected intraperitoneally (i.p.) with 100 μg ova together with 4 mg alum adjuvant in 200 μl sterile HBSS. These mice were challenged 25 to 35 days later by the same procedure just described. Other control groups for comparison were mice that were not infected with Pneumocystis but that were given ova i.n. at the same time as the other group, mice that were infected with Pneumocystis but that were not exposed to ova before the final challenges, and mice that did not receive either Pneumocystis or ova before the final challenges.

Respiratory measurements.

Whole-body plethysmography was used to measure respiratory parameters, including enhanced pause (Penh), in unrestrained, nonanesthetized mice, as described elsewhere (46). Briefly, mice were placed in lucite plethysmograph chambers that were part of a flowthrough system (Buxco, Wilmington, NC). Mice successively received aerosol exposures to PBS and methacholine (Sigma Chemical) at concentrations of 2.5, 5, 10, and 20 mg/ml, with 3-min exposures and 5-min measurement periods. Respiratory parameters were continuously calculated by a computer using Biosystem XA software (Buxco). Mice were allowed to recover to near baseline Penh values between successive doses of methacholine. We are aware of the controversy over the use of Penh values as valid indicators of airway constriction (1, 4). While we recognize that Penh does not always correlate well with the single parameter of large airway diameters, the approach does measure a complex respiratory distress and is useful for noninvasive screening for certain respiratory responses (47). Until such time as a more descriptive term is coined to describe what Penh represents, we will use the commonly used term AHR.

Statistical analysis.

The software program GraphPad Prism (San Diego, CA), was used for all statistical tests of significance (i.e., P ≤ 0.05). One-way analysis of variance, followed by Tukey's post hoc pairwise comparisons, was performed when more than two groups were being compared. When only two groups were compared, we used a two-sided t test with Welch's correction for unequal variances as necessary.

RESULTS

Pneumocystis initiates a distinct immune response pattern in immunocompetent mice.

Early influxes of NK cells and neutrophils into the alveolar compartment are followed by gradual increases in the numbers of alveolar macrophages and both CD4 and CD8 T cells (Fig. 1). However, levels of Pneumocystis were not affected by this early response; as reported previously, most of the inoculum is cleared from the lung by innate mechanisms (7), but what fungus remains then increased constantly until day 14, which also represents the peak of the overall inflammatory response (Fig. 1). At this point, there was also a transition in the nature of the inflammatory response; B cells and eosinophils began to appear in higher numbers, and this continued for another 7 days (Fig. 1). At this time, a vigorous acquired immune response, including anti-Pneumocystis antibody, was developing and Pneumocystis numbers dropped precipitously by 21 days of infection, and by 28 days there was no detectable pathogen (Fig. 1). Inflammation rapidly subsided as well: most cell types returned to normal levels by this time, although alveolar levels of macrophages, CD4 cells, and especially CD8 cells showed a more gradual decline (Fig. 1). In this way, the infection is efficiently dispatched, with no obvious residual pathology, unlike the persistent and often deadly pneumonia seen in immunocompromised hosts.

Fig. 1.

Kinetics of the immune response to Pneumocystis in BALF of immunocompetent BALB/c mice. NK cells and neutrophils are the first immune cells to increase in large numbers in the BALF, followed closely by CD4 and CD8 T cells and then B cells and eosinophils, together with a gradual increase in alveolar macrophages. Anti-Pneumocystis antibody titers increase in BALF between 14 and 21 days, and this coincides with the clearance of Pneumocystis (arrow, assay detection limit). Values are means ± standard errors of the means (n = 5), representative of three independent experiments. O.D., optical density. *, P = 0.05 to 0.01; ***, P < 0.001.

It has previously been reported that the cytokines TNF-α and IFN-γ are secreted early in response to Pneumocystis infection in other mouse strains (29, 36, 50), and we observed that as well in this strain (data not shown). However, we also found moderate amounts of the Th2 cytokine IL-4 over the course of infection, and although the Th2 cytokine IL-5 was not elevated during the peak of inflammation, it was elevated very early after inoculation with Pneumocystis, unlike most of the other cytokines examined (Fig. 2). Although the amounts of these cytokines by themselves did not reflect Th2 domination in the lungs at the times of collection, later histological observations (see below) represent changes consistent with a Th2 differentiation in the lung.

Fig. 2.

The Th2 cytokine IL-4 is moderately elevated in BALF during the peak of inflammation (A), while the Th2 cytokine IL-5 is elevated only early after Pneumocystis inoculation (B). Values are means ± standard errors of the means (n = 4), representative of two independent experiments. *, P = 0.05 to 0.01; **, P = 0.01 to 0.001.

Histological observations also indicate the strong level of pulmonary inflammation found at 14 days after Pneumocystis inoculation: Fig. 3 shows large numbers of inflammatory cells, both mononuclear and granulocytic, present in the interstitial and alveolar compartments. The histological appearance also supports the suggestion that a transition to a Th2 environment has occurred, based on the presence of multinucleate giant cells and the hypertrophy of airway cells, which have been observed in other models of Th2 inflammation (58). Many of the pulmonary leukocytes are found in clusters in the perivascular and peribronchiolar areas. Some of these clusters are dominated by the presence of eosinophils, again reinforcing the likelihood of a Th2 transition being under way. In contrast, the other clusters of leukocytes consist almost exclusively of mononuclear cells and are consistent with the appearance of the organized areas of pulmonary leukocytes known as bronchus-associated lymphoid tissue (BALT). The localization of these structures presumably facilitates the productive interaction of antigen with APCs and APCs with lymphocytes, as discussed below.

Fig. 3.

Vigorous pulmonary inflammation in wild-type immunocompetent mice 14 days after inoculation with 10⁷ Pneumocystis organisms (A and B) compared to normal lung histology (C). In addition to influx of inflammatory cells, there is hypertrophy of airway epithelial cells (arrows) and formation of BALT near airways and blood vessels (arrowheads). A Th2 environment is suggested by the presence of multinucleate giant cells (open arrowhead). Staining is standard H&E. Magnification, ×200.

Antigen-presenting cells are abundant in the lung during Pneumocystis infection.

In tandem with the inflammatory changes that we observed in the alveolar space, where the fungus was localized, there were also significant changes in the interstitial compartment of the lung. As seen in Fig. 4A, at 14 days postinfection, there was a nearly 10-fold increase of CD45⁺ CD11c⁺ potential APCs in the lung tissue. To verify that these cells are capable of functioning as APCs, we isolated CD11c⁺ cells from total lung digests and cocultured them with CD4 T cells from ova-specific (D011.10) mice in the presence of ovalbumin. As can be seen in Fig. 4B, MHC-II^hi CD11c⁺ cells were capable of presenting ovalbumin to T cells, as indicated by CD4 cell proliferation. It should be noted that Pneumocystis infection resulted in only marginal increases in the ability of these cells to present antigen to CD4 cells compared to the ability for noninfected mice, but the vastly increased number of these cells would allow a much greater overall potential to present antigen during infection. Although in many tissues CD11c⁺ cells are uniformly described as dendritic cells, this is not as clear-cut in the lung, where alveolar macrophages also express CD11c (16). For this reason, we have refrained from referring to these cells as dendritic cells and instead refer to them simply as pulmonary CD11c⁺ cells. We did, however, perform detailed phenotypic analysis of these cells and assessed their capacity to present antigen. Not only was there a dramatic increase in the number of MHC-II^hi CD11c⁺ cells in Pneumocystis-infected lungs, but the level of expression of MHC-II on these cells was also elevated over that on comparable cells from noninfected mice (Fig. 4C). Furthermore, expression of the αMβ2 integrin CD11b on the MHC-II^hi CD11c⁺ cells was greatly increased (Fig. 4C), and while MHC-II^hi CD11c⁺ cells in both infected and noninfected cells expressed the αE integrin CD103, this expression was slightly elevated in Pneumocystis-infected tissues (Fig. 4C), as was the expression of the scavenger receptor CD204 (Fig. 4C). Pneumocystis infection also caused MHC-II^hi CD11c⁺ cells to upregulate expression of the costimulatory receptor CD40 and, to a much lesser extent, CD25, CD80, and CD86 (data not shown). Finally, although changes in expression of the B220 and Gr-1 antigens have been associated with specific subsets of CD11c⁺ dendritic cells, Pneumocystis infection did not cause consistent changes in their expression. We also verified that pulmonary CD11c⁺ cells will take up exogenous protein by intranasal applications of fluorescently labeled ovalbumin. When mice were given a single dose of Alexa Fluor 633-tagged ovalbumin i.n., by 24 h a considerable proportion of pulmonary CD11c⁺ cells were positive for Alexa Fluor 633, as determined by flow cytometry (Fig. 5). This was true for both alveolar and interstitial cells, but when Alexa Fluor 633-tagged ovalbumin was given to mice at 14 days after Pneumocystis infection, a much larger percentage of interstitial cells and a lower percentage of alveolar cells were ova positive. This suggests either that the permeability of the pulmonary barrier had increased to allow greater access to interstitial APCs or that alveolar CD11c⁺ cells which take up the labeled ova in infected mice actively migrated into the interstitial compartment. Our observations of elevated serum albumin levels in the BALF at this time (data not shown) support the former possibility, but we cannot rule out the latter possibility, and a combined effect is entirely possible.

Fig. 4.

Antigen-presenting cells in the lung during Pneumocystis infection. (A) Levels of putative APCs (MHC-II^hi CD11c⁺) peak in the lung at 14 days after Pneumocystis inoculation. Cells are from collagenase digestion of the lung and are measured as CD45⁺ MHC-II^hi CD11c⁺ using flow cytometry. Values are means ± standard errors of the means (n = 5), representative of three independent experiments. (B) In vitro proliferation of ova-specific (D011.10) CD4 lymphocytes when they were cocultured with CD11c⁺ cells obtained from collagenase digestion of cells from mice obtained 14 days after Pneumocystis inoculation or noninfected control mice incubated with increasing concentrations of ovalbumin. Data are representative of those from two independent experiments, with each data point measured in triplicate. (C) Relative expression of cell surface markers on APCs obtained by collagenase digestion and analyzed by flow cytometry. Dark histogram, expression on cells obtained 14 days after inoculation with Pneumocystis; light gray histogram, expression of the same molecule on cells from uninfected mice. Histograms are representative of staining seen in two independent experiments with 4 to 5 mice per group. ***, P < 0.001.

Fig. 5.

Percentages of CD11c⁺ cells that are positive for fluorescent label 24 h after i.n. inoculation with 50 μg of Alexa Fluor 633-tagged ovalbumin (OVA-633). Cells are either BALF cells (BAL) or cells from a collagenase digestion of the lung after alveolar lavage (Lung). Mice are either uninfected controls or mice tested 14 days after inoculation with Pneumocystis (PC 14d). Values are means ± standard errors of the means (n = 4), representative of two independent experiments. Bars indicate the groups being compared by Tukey's test. ***, P < 0.001.

Pneumocystis infection can facilitate sensitization to exogenous antigen.

One of the functional implications of these inflammatory changes is that when mice were exposed to antigen at the peak of this response, they became sensitized to that antigen. We found that exposure to a single i.n. dose of ova without adjuvant or any prior extrapulmonary exposure at the 14-day time point of a Pneumocystis infection resulted in sensitization to ova, in the form of elevated Penh values upon later challenge, including those at lower doses of methacholine (Fig. 6). This ability to become sensitized was only within a narrow time window; however, exposure to ova at 21 days after Pneumocystis inoculation, as the infection was waning, resulted in a diminished Penh response (Fig. 6), as did earlier or later exposures (data not shown). The magnitude of this elevated Penh response was equal to that shown by mice that had been previously immunized, while mice exposed to a single i.n. dose of ova in the absence of a Pneumocystis infection exhibited Penh values upon challenge identical to those of control mice not exposed to either Pneumocystis or ova before the final challenge (Fig. 6). Mice inoculated with Pneumocystis but not exposed to ova exhibited slightly elevated Penh values when they were challenged 30 days after inoculation (Fig. 6); however, these values were significantly lower than those of the ova-sensitized mice.

Fig. 6.

AHR (Penh) is elevated in BALB/c mice in response to antigen challenge after intranasal exposure to that antigen at 14 days after Pneumocystis inoculation. Methacholine response curves after challenge with ova are presented as Penh values. Three groups were inoculated with 10⁷ Pneumocystis organisms, and then the first group was given 50 μg ova i.n. at 14 days postinoculation (OVA 14d), the second group was given 50 μg ova i.n. at 21 days postinoculation (OVA 21d), and the third group was not exposed to ova until the final challenge (PC only). The other three groups were not inoculated, and then one group was sensitized to ova with an i.p. injection of ova-alum 21 days before challenge [positive control, (+) CON], no ova was given to the second uninoculated group (No PC-OVA), and one group was left untreated until the final challenge [negative control, (−) CON]. All groups received antigen challenge (3 successive days of 1 dose 20 μg ova i.n and then the methacholine challenge and Penh measurements on the subsequent day). Final challenges occurred 30 to 35 days after previous ova exposure. Values are means ± standard errors of the means (n = 4), representative of three independent experiments. P values (**, P = 0.01 to 0.001) are from comparison of the results for the group treated with ova at 14 days postinoculation to those for both the uninoculated group not receiving ova and the group inoculated with Pneumocystis only. Other significant differences are not shown here for purposes of clarity.

In other aspects, however, the phenotype of this antigen sensitization differed from that seen with prior i.p. immunization with adjuvant. In that situation, subsequent antigen challenge resulted in increased influx of eosinophils into the alveolar space after challenge, as well as significant titers of antigen-specific IgE in the serum (15, 43). In contrast, we observed only a slight, but nonsignificant, trend toward elevated eosinophils and ova-specific IgE in mice exposed to ova at 14 days after Pneumocystis infection (Fig. 7).

Fig. 7.

BALF eosinophil numbers and serum ova-specific IgE of mice that were exposed to ova at different times after Pneumocystis inoculation. Groups are the same as those described in the Fig. 6 legend. Values are means ± standard errors of the means (n = 4), representative of three independent experiments. *, P = 0.05 to 0.01. OD, optical density.

IL-4 signaling is required for sensitization to occur.

Because sensitization to respiratory antigens and, in particular, the type of collateral sensitization that we demonstrate here are associated with Th2 differentiation (11), we wished to examine the hypothesis that the Th2 differentiation pathway was necessary for sensitization to occur. We found that mice deficient in the IL-4 receptor α chain, unlike wild-type BALB/c mice, did not demonstrate sensitization to a single i.n. dose of ova at 14 days after Pneumocystis infection (Fig. 8). This lack of response to antigen was concomitant with a distinctly different inflammatory response at the histological level, even though the Pneumocystis burden at this time (log₁₀ of 6.52 ± 0.13) was not significantly different from that of wild-type mice. As seen in Fig. 8B, the signs of Th2 inflammation, such as airway hypertrophy, eosinophils, and multinucleate giant cells, were absent. In addition, while there was still some aggregation of inflammatory cells in peribronchiolar and perivascular areas, they were not nearly as structured as either the eosinophil clusters or the BALT structures seen in wild-type mice at the same stage of infection.

Fig. 8.

IL-4 signaling is necessary for antigen sensitization to occur during Pneumocystis infection. (A) AHR (Penh) of IL-4r KO mice and BALB/c mice in response to antigen challenge. PC, mice inoculated with 10⁷ Pneumocystis organisms. All groups except the positive and negative controls were given 50 μg ova i.n. 14 days postinoculation. One group was sensitized i.p. with ova-alum. The negative control group [(−)CON] was untreated until the final challenges. All groups received antigen and methacholine (Mch) challenges as described in the text. Values are means ± standard errors of the means (n = 4), representative of three independent experiments. **, P = 0.01 to 0.001. (B) Lung inflammation in IL-4r KO mice 14 days after Pneumocystis infection is absent the epithelial hypertrophy, multinucleate giant cells, and organized clusters of inflammatory cells seen in wild-type mice (Fig. 3).

DISCUSSION

Many infectious diseases of the respiratory tract cause pathological changes that persist well after the infection has been resolved. We show here that the fungal pathogen Pneumocystis is uniquely capable in this respect, in that it can facilitate sensitization to an exogenous antigen with only a single exposure, without the necessity of an additional adjuvant, provided the exposure occurs at the time of the peak inflammatory response.

This ability is largely due to the fact that the immune response during an initial Pneumocystis infection includes most of the mechanistic elements that are believed to be instrumental in infectious disease-associated antigen sensitization. The first of these factors related to increased potential for antigen uptake and presentation is the increase in the numbers of mature APCs. This phenomenon has been observed to various extents in many infectious diseases, such as those caused by RSV (5), influenza virus (6, 57), and atypical bacteria such as Chlamydia pneumoniae (44), as well as noninfectious perturbations, such as those caused by certain inhaled particulates (23, 28, 38). Dendritic cells are typically described to be the most important APC in pulmonary tissue (19, 35). This classification is complicated by ever increasing numbers of proposed dendritic cell subsets (13), the evidence of plasticity in the established dendritic cell phenotypes (33, 56), and the observation that alveolar macrophages also express CD11c, the prototypical dendritic cell marker. For this reason, we restricted identification as APCs only to CD11c⁺ cells that display a phenotype consistent with antigen-presenting capabilities, including a high level of expression of MHC-II, as well as elevated levels of expression of CD11b and costimulatory receptors such as CD40. We also verified that the CD11c⁺ cells that we found in lung digests were capable of antigen presentation in vitro and took up labeled exogenous antigen in vivo.

The second mechanistic element supporting antigen sensitization is the temporary structural changes that are occurring during Pneumocystis infection that allow increased access of inhaled antigens to the APCs. Our observation that at 14 days postinfection labeled antigen appears in a greater proportion in interstitial CD11c⁺ cells suggests that some aspect of permeability of alveolar membranes or trafficking of APCs to the alveolar membranes is occurring, although we cannot discriminate between the two at this time. Pneumocystis infection also seems to have profound effects on the enhancement of structures where APC interactions with CD4 T cells and other potential effector cells occur. The tracheal-bronchial lymph nodes that drain the lungs are significantly enlarged at the time of sensitization (14 days), although we did not observe significant translocation of intranasally delivered antigen to the draining lymph nodes at this time. However, these observations are in agreement with those of studies that show retention of antigen-laden APCs in the pulmonary tissue, concomitant with successful priming of T cells, presumably in the lung microenvironment and tissue-specific tertiary lymphoid tissue (8, 25). In fact, this fits what we observed about the effect of Pneumocystis infection on the induction of BALT. BALT is an organized lymphoid structure that is only transiently present in healthy humans and mice but that can be formed rapidly in pulmonary tissue during infection and inflammation (40). Most importantly, BALT is an active site where APCs interact with T cells, priming them for effector status. A recent study has shown that BALT can be an effective site for priming of naive T cells against an antigen (in that case, an ova peptide) that is distinct from the antigenic stimulus or infectious agent that induced the formation of the BALT (18). As shown above, Pneumocystis infection induces the formation of BALT, and at 14 days postinoculation, BALT is well developed. Therefore, the nonspecific nature of the BALT to facilitate sensitization toward any exogenous antigen and its proximity to the site of antigen uptake can be major factors by which Pneumocystis infection can facilitate antigen sensitization.

A third factor important to the promotion of collateral sensitization during Pneumocystis infection is a Th2 immune response. In our model of sensitization, some component of the Th2-related immune response is essential, as sensitization does not occur in the absence of IL-4 signaling. The role of Th2 immune responses in respiratory sensitization and the early development of asthma has been examined in several animal models. Studies linking early viral infections (such as RSV infections) to the inception of allergic sensitization have found a Th2 response to be necessary for the development of the sensitization (45). However, other studies have produced contradictory results and have led to the conclusion that the timing and dose of the viral infection and the initial antigen exposure can profoundly affect the outcomes achieved with these experimental models (49). Human studies also indicate that the time of viral infection is key as to whether the infection will promote a Th2 response that predisposes to allergic responses, as opposed to the predicted Th1 immune response. Indeed, those studies point to early infancy as the period wherein viral infections (24) and in some cases atypical bacteria (22) can initiate Th2-type responses. In contrast to those infectious agents, a Pneumocystis infection in an immunocompetent host (but not in a CD4-deficient host) always results in a Th2 immune response during organism clearance (12, 27, 37, 48). More recent studies have characterized a more general model in which inhalational sensitization occurs in the lung during an ongoing inflammation (10). Because this sensitization can occur to antigens unrelated to the original infection or inflammation, the authors have termed this “collateral sensitization.” Furthermore, these authors show, as we do here, that IL-4 receptor signaling is required for this collateral sensitization to occur.

It is not surprising that this transition to a Th2 phenotype in these immunocompetent mice is what most distinguishes this immune response from that seen in CD4-depleted Pneumocystis-infected mice. In those mice, there is an early recruitment of neutrophils, as well as some T cells, and this pattern continues throughout the infection, so that neutrophil and especially CD8 lymphocyte numbers are very high by the end stage of the disease (52). This is in contrast to the reduction in neutrophils, increases in eosinophils and B cells, and moderation in T cell numbers that coincide here with the mounting of an effective acquired immune response.

It is notable that the asthma-like phenotype that occurs as a result of i.n. administration of ovalbumin during Pneumocystis infection is somewhat different from that observed when the sensitization occurs via i.p. injection with adjuvant alum. In the latter case, subsequent antigen challenges result in distinct eosinophilia, elevated titers of antiovalbumin antibodies, and profound AHR. In contrast, with i.n. sensitization, eosinophilia and antiovalbumin titers are only moderately elevated, but AHR is comparable to that seen in the i.p. plus adjuvant sensitization. This is especially interesting, given the growing acknowledgment of the heterogeneous clinical expression of asthma in the human population (2). For example, anywhere from 25 to 50% of asthma patients do not exhibit elevated eosinophils in induced sputum samples, even though other symptoms can range from mild to severe (17). These observations, along with the findings of recent mouse studies, lend support to the idea that AHR and inflammation may be distinct and uncoupled events in asthma (53). The fact that in our Pneumocystis sensitization model eosinophilia was generally lower than what was reported in RSV- and influenza virus-associated sensitization also suggests that although the mechanisms described above may be similar in all those models, there may be distinct differences in the sensitization process. AHR development may be, in a sense, the lowest common denominator of sensitization, while eosinophilia and high specific antibody titers may develop only when sensitization has followed a specific path that predisposes to extreme levels of subsequent Th2 skewing, such as with added adjuvants.

The question that remains is whether the experimental model of Pneumocystis-associated antigen sensitization that we present here is relevant to the etiology of allergic asthma in humans. The putative connection between RSV infections in infancy and asthma was a result of clinical observations of wheezing in patients that could be definitively diagnosed with RSV infection. This led to animal studies to establish potential roles of RSV infection and infection with other respiratory viruses and atypical bacteria in both the pathogenesis and exacerbation of asthma. This is in stark contrast to the situation with Pneumocystis. The pathology of Pneumocystis infection has come under scrutiny only because of the serious nature of Pneumocystis pneumonia in immunocompromised individuals. There is still insufficient information about the normal life cycle of this pathogen in human hosts and of the nature of Pneumocystis infections in immunocompetent individuals. What few studies have been performed suggest that Pneumocystis is constantly present and perhaps transiently carried by individuals in amounts that are too low to stimulate a robust acquired response but enough to spread to previously uninfected hosts (9, 42). What is not known is whether the degree of inflammation that we find in mice that have been given a large intratracheal bolus of Pneumocystis organisms is also reached in immunocompetent infants that have received aerial exposures to small numbers of organisms. Undoubtedly, Pneumocystis is responsible for one of the many respiratory infections seen early in every child's life that may be mildly symptomatic but not usually severe enough to cause hospitalization and receive a discrete diagnosis, but the literature would suggest that the responses to the infections are highly variable. The fact that in mice a Pneumocystis infection can occur, exhibit all the mechanisms which predispose for sensitization described above, yet pass with no lasting pathology begs the question of whether this may happen unnoticed in humans as well. In all models of infectious disease-related sensitization to inhaled antigen, there is a distinct commonality of mechanisms: increased abundance and maturation of APCs, Th2-supported immune responses, and possible increases in permeability of the lung epithelial barrier. Sensitization to exogenous antigen, therefore, is probably a stochastic phenomenon, where the upregulation of the mechanistic machinery is more likely to take up and respond to an innocent bystander protein. What we show here is that Pneumocystis infection is uniquely capable of upregulating all of these putative mechanisms, to the extent that even a single respiratory exposure may result in sensitization, provided that the inflammatory response sufficiently recapitulates these mechanisms.