The Pathogenesis of Fatal Outcome in Murine Pulmonary Aspergillosis Depends on the Neutrophil Depletion Strategy

Abstract

Aspergillus fumigatus causes invasive disease in severely immunocompromised hosts but is readily cleared when host innate defenses are intact. Animal models for evaluation of therapeutic strategies to combat invasive aspergillosis that closely mimic human disease are desirable. We determined optimal dosing regimens for neutrophil depletion and evaluated the course of infection following aerosol infection in mice by determining survival, organ fungal burden, and histopathology in mice in which neutropenia was induced by three methods, administration of granulocyte-depleting monoclonal antibody RB6-8C5 (MAb RB6), administration of cyclophosphamide, and administration of both agents. Administration of either individual agent resulted in a requirement for relatively high conidial inocula to achieve 100% mortality in both BALB/c and C57BL/6 mice, although the infection appeared to be somewhat more lethal in C57BL/6 mice. Death following induction of neutropenia with MAb RB6 occurred when a relatively low fungal burden was present in the lung and may have been related to the inflammatory response associated with neutrophil recovery. In contrast, administration of both agents reduced the lethal inoculum in each mouse strain by approximately 1 log₁₀, and C57BL/6 mice that received both agents had a higher fungal burden and less inflammation in the lung at the time of death than BALB/c mice or mice of either strain that received MAb RB6 alone. Our data suggest that the relationship among fungal burden, inflammation, and death is complex and can be influenced by the immunosuppression regimen, the mouse strain, and the inoculum.

Aspergillus fumigatus is a ubiquitous mold that causes invasive disease in severely immunocompromised patients. The important risk factors for development of invasive aspergillosis include neutropenia, defective neutrophil function, the use of corticosteroids or other immunosuppressive therapies, such as those used to prevent rejection following organ transplantation, and late-stage human immunodeficiency virus infection (5, 12, 13, 30, 41, 54). The organism, aerosolized from soil, air filtration, or water sources, is inhaled, and the most common site of primary infection is the lung (1, 40, 48, 58). The inoculum required for disease production is unknown and may be dependent upon the immune defect of the affected host (14). Innate immune responses control infection in a normal host. Conidia are phagocytosed by alveolar macrophages, but in the absence of effective macrophage function, the conidia germinate, and the organism resumes hyphal growth (51). Neutrophils are rapidly recruited to the lung in response to experimental pulmonary infection with A. fumigatus (17) and effectively kill hyphae that result from escape from conidial killing by macrophages (51). The importance of NK cells as effectors during the initial response to A. fumigatus has been demonstrated recently (39).

Models for invasive aspergillosis have been established in mice, rats, rabbits, and guinea pigs (3, 18, 31, 53). When possible, these models have incorporated pulmonary rather than intravenous (i.v.) deposition of conidia because of the route of infection in humans (6, 9, 35). Neutropenia, chemotherapy, and glucocorticoid use are important risk factors for invasive aspergillosis in humans. Consequently, investigators have established a number of experimental models with immunocompromised mice using a granulocyte-depleting monoclonal antibody (MAb), MAb RB6-8C5 (MAb RB6), cyclophosphamide, an alkylating agent that is commonly used for cancer chemotherapy, and a variety of glucocorticoid preparations (8, 11, 36, 38, 55). The effects of these agents on host immune responses are different. To understand the way in which the method of neutropenia induction alters the pathogenesis of experimental invasive pulmonary aspergillosis, we evaluated the effects of such agents on infection following aerosol exposure in a model patterned after that developed by Piggott and Emmons (47) and Brieland et al. (6).

MATERIALS AND METHODS

Immunosuppressive agents.

The rat hybridoma producing MAb RB6 was a generous gift from Robert L. Coffman (DNAX Research Institute of Molecular and Cellular Biology, Inc., Palo Alto, Calif.). Ascites containing MAb RB6 was produced by intraperitoneal (i.p.) injection of hybridoma cells into pristine-primed SCID mice (National Cancer Institute, Bethesda, Md.) and was sterilized by filtration with a 0.2-μm-pore-size filter. The concentration of MAb RB6 was determined by an enzyme-linked immunosorbent assay (ELISA) by using a purified rat immunoglobulin G2b (IgG2b) standard (PharMingen, BD Biosciences, San Diego, Calif.). Cyclophosphamide, triamcinolone, and methotrexate were obtained from Sigma (St. Louis, Mo.). Dexamethasone was obtained from ICN Biomedicals, Inc. (Irvine, Calif.).

Mice.

Specific-pathogen-free female BALB/c, C57BL/6, A/JCr, and DBA/2 mice were obtained from the National Cancer Institute. BALB/c and A/JCr mice have the Ly6.1 haplotype, while C57BL/6 and DBA-2 mice express the Ly6.2 haplotype (34, 44). The mice weighed 16 to 18 g at the time of experimentation, except where indicated below. The mice were housed in microisolator cages in a pathogen-free barrier facility at the Albert Einstein College of Medicine.

Mice were inoculated via the lateral tail vein or i.p. with MAb RB6 diluted in sterile phosphate-buffered saline (PBS). The mice were bled from the orbital sinus into Eppendorf tubes containing 2 μl of 10% EDTA (pH 7.4). The blood was diluted 1:20 in Turk's solution (1% glacial acetic acid and 0.01% gentian violet in distilled H₂O), and the total number of white blood cells (WBCs) was determined by using a hemacytometer (37). WBC differential counts were determined on smears of whole blood stained with the Hema3 staining system (Fisher Scientific, Biochemical Sciences, Inc., Swedesboro, N.J.) used according to the manufacturer's instructions. At least 100 WBCs were counted per slide. The total WBC counts and differentials were used for calculation of the absolute neutrophil count (ANC) and the absolute lymphocyte count (ALC).

Mouse serum was tested for MAb RB6-binding antibody by an ELISA. MAb RB6 obtained from concentrated cell supernatants was purified by protein G chromatography. Polystyrene plates were coated with a solution containing 5 μg of MAb RB6 per ml, and following incubation for 1 h at 37°C, wells were blocked with PBS containing 2% bovine serum albumin (BSA). Serum samples were added to the wells and then serially diluted, beginning with a 1:50 dilution. After incubation and washing, goat anti-mouse IgM or IgG was added at a concentration of 1 μg/ml. p-Nitrophenyl phosphate in carbonate substrate buffer was added to the wells, and the A₄₀₅ was determined by using a Multiskan microplate reader (Labsystems, Franklin, Mass.).

Flow cytometry.

Five mice were treated with 25 μg of MAb RB6, five mice were treated with 150 mg of cyclophosphamide per kg, and five mice were treated with both agents and then killed 1 day later. Five naïve mice served as a control. The spleens were removed, and single-cell suspensions were prepared by homogenization in Hanks' balanced salt solution. Following centrifugation at 290 × g for 10 min at 4°C and washing, the pellet was passed through a 70-μm-pore-size filter. Red blood cells were lysed by incubation of the cell suspension in 0.17 M NH₄Cl (pH 7.0) at 4°C for 10 min. Fluorochrome- or biotin-conjugated MAbs to cell surface markers or isotype-matched controls were produced in rats, except as indicated below, and were obtained from Pharmingen (San Diego, Calif.) (CD3-allophycocyanin (APC) [hamster IgG], CD4-fluorescein isothiocyanate (FITC) [IgG2b], CD8a-phycoerythrin (PE) [IgG2a], CD19-biotin [IgG2a], Mac-3-PE [IgG1], CD11c-APC [hamster IgG]) or Caltag Laboratories (Burlingame, Calif) (IgG1-biotin, IgG2a-FITC, IgG2b-APC, and hamster IgG-biotin). MAb RB6 was purified from concentrated cell supernatants by protein G chromatography, conjugated to Alexa-488 (Molecular Probes, Eugene, Oreg.) according to the manufacturer's instructions, and quantified by ELISA. Cell suspensions were distributed into four tubes, blocked with 10% goat, rat, and/or hamster serum, and then stained with the following combinations: (i) 1 μg of CD3 per ml, 2.5 μg of CD4 per ml, 10 μg of CD8a per ml, and 10 μg of CD19 per ml; (ii) 1 μg of Mac-3 per ml, 1 μg of CD11c per ml, 2 μg of MAb RB6 per ml, and 10 μg of hamster IgG per ml; and (iii) 10 μg of rat IgG1per ml, 10 μg of rat IgG2a per ml, and 10 μg of rat IgG2b per ml. The cell suspensions were incubated for 20 min on ice, and then 1 ml of PBS with 1% BSA was added and the cells were centrifuged for 5 min, suspended in a solution containing 1 μg of streptavidin conjugated to peridinin chlorophyll (Pharmingen) per ml, and incubated on ice for 20 min. The cells were washed again in 1 ml of buffer consisting of PBS with 1% BSA and suspended in 1 ml of PBS with 1% BSA. A fluorescence-activated cell sorting analysis was performed by using a FACSCalibur (Becton Dickenson, Franklin Lakes, N.J.). The analysis was performed by using the CellQuest software (Becton Dickinson).

A. fumigatus.

Strain ATCC 90906 was obtained from the American Type Culture Collection (Manassas, Va.). This strain was isolated from the blood of a patient with invasive aspergillosis (15). Stock solutions of conidia were maintained at −80°C. Sabouraud dextrose agar slants were inoculated with a loopful of frozen stock and grown at room temperature for 7 days. PBS with 0.05% Tween 20 was added to the slants, which were then gently scraped, and 4-ml portions of the resulting conidial suspensions were transferred to eight-arm mouse inhalation flasks (Ace Glass, Vineland, N.J.) containing 150 ml of Sabouraud dextrose agar (6). The flasks were incubated at room temperature for 14 days. For intratracheal (i.t.) infection experiments, conidia from cultures grown on agar as described above were collected in 0.1% Tween 20 and then passed twice through 12-μm-pore-size Isopore membrane polycarbonate filters (Millipore, Billerica, Mass.) and washed in Tween 20 prior to quantification by hemacytometer counting. Microscopic examination of organisms collected in this manner revealed that conidiophores and hyphal fragments were removed. Conidial viability was confirmed by plating appropriate dilutions on Sabouraud dextrose agar. Colony counts were determined after growth at room temperature for 4 days. To determine the effect of preparation of conidia in solutions containing Tween 20 on the surface charge, 20-day-old cultures were collected in 0.01 M NaCl with or without 0.2% Tween 20. The zeta potential of 10 randomly selected cells passing through the detection meter was measured for each sample by using a Zeta-Meter 3.0+ (Zeta-Meter, Inc., Staunton, Va.), which calculated the zeta potential, as described previously (43).

Infection models.

For i.t. infection, mice were anesthetized by i.p. injection of ketamine and xylazine, and then conidia (volume, 50 μl) were administered via a midline neck incision by using a bent 26-gauge needle attached to a tuberculin syringe, as described previously (19). The incisions were closed with Nexaband adhesive (Closure Medical Corporation, Raleigh, N.C.). For aerosol infection, mice were placed in the side arms of an inhalation flask, and then a cloud of conidia was created by pumping air into the flask by using a 60-ml syringe attached to Tygon tubing. After 1 min, the mice were removed from the flask. The inoculum was varied by altering the force with which air was pumped through the syringe. Before the flask was used for infection, a trial run was performed to evaluate the size of the conidium cloud produced by a given force. To determine the reproducibility of the infecting inoculum for mice infected in a flask, initially two separate experiments were performed, in which two flasks were used and two trials were done per flask. All eight mice in each trial were killed 2 h after infection, their lungs were homogenized in sterile PBS, and dilutions were plated for determination of the inoculum that reached the lung. The CFU were counted after incubation of plates at room temperature for 4 days. In all subsequent experiments, two mice from each flask run were killed 2 h after infection for inoculum determination based on the CFU in the lungs. For all groups that were compared to each other, equal numbers of mice from each group were included in each run of the flask. In experiments in which defined inocula reaching the lung were desired, when this determination was outside 0.5 log₁₀ of the targeted range, the data were excluded from further analysis. In all experiments, four additional mice received each immunosuppression treatment in conjunction with sham infection, in which flasks with agar that was not seeded with A. fumigatus were used.

Chitin measurement.

Lung chitin was measured as described previously (33). The A₆₅₀ was measured after 25 min. Lungs from uninfected mice were used as blanks. Glucosamine and distilled H₂O were used as standards. The limit of detection of this assay was 0.002 μg of glucosamine.

Histopathology.

In some experiments, the right upper lobe of the lung was used for histopathology. Formalin-fixed tissues were embedded in paraffin. Five-micrometer sections were stained with hematoxylin and eosin (H&E) or Gomori methenamine silver (GMS) and were examined by using an Axioplan 2 or Axiophot microscope (Carl Zeiss, Inc., Thornwood, N.Y.). At least six sections were examined for each stain for each mouse.

Morphometry.

Each lung section was viewed with a Stemi SV11 microscope (Zeiss) at a magnification of ×15.75 for determination of the total lung area examined. All areas of each section then were examined at a magnification of ×25 or ×90, and all areas with visible fungi were photographed. The area of a lung containing A. fumigatus was quantified by using ImageJ, version 1.31t, based on the black color of organisms on GMS-stained slides. The threshold of each photograph was adjusted to determine the number of pixels that were black and represented areas where there was fungus. Areas that were black but not associated with identifiable fungi were excluded. After determination of the relationship between pixel number and distance at each magnification, the area of lung containing hyphae was calculated for each section and divided by the total area of the lung examined.

Statistical analysis.

The statistical analysis was performed by using Sigmastat, version 3.0 (SPSS, Inc., Chicago, Ill.). For two-group comparisons, we used Student’s t test for normally distributed data or the Mann-Whitney rank sum test for data that were not normally distributed. For multiple comparisons, pairwise comparisons were made by using the Holm-Sidak test after demonstration of statistical significance by one-way analysis of variance (ANOVA) or Dunn's method for comparison after one-way ANOVA on ranks. The correlation coefficient for chitin and morphometry was determined by Spearman rank order correlation. Significance was defined by a P value of <0.05.

RESULTS

Aerosol infection model.

In initial experiments to assess reproducibility, in each of eight trials the standard deviation for the CFU retrieved from the lungs of infected mice was less than 0.25 log₁₀, demonstrating that mice were evenly infected in the different positions around the flasks. In these eight trials, the mean inocula ranged from log₁₀ 6.2 to 6.8, which is a range within which pathogenesis is unlikely to be altered. Similar results were obtained with delivery of smaller inocula, ranging from log₁₀ 4.7 to 5.7 per mouse (Fig. 1). In only 2 of 14 experiments did the mean inoculum deviate from the target inoculum by more than log₁₀ 0.5, and in none of the experiments was the deviation greater than log₁₀ 0.6. Inocula that were log₁₀ 4 or less could not be delivered reliably. In summary, our data show that reproducible results can be obtained with the aerosol model and that two of the eight mice in each trial can be used to establish the inoculum that reaches the lung.

FIG. 1.

Reproducibility of aerosol infection, demonstrating the relationship between the target inoculum and the inoculum achieved. Each circle represents the mean log₁₀ inoculum attained in experiments that included 2 to 10 trials. The horizontal lines indicate target inocula.

MAb RB6-induced neutropenia.

As noted previously, MAb RB6 may cross-react with other members of the Ly-6 family (20). Therefore, we sought to determine the smallest amount that induced absolute neutropenia and retained selectivity. The results of four representative experiments are shown in Fig. 2. We found that in BALB/c mice, administration of 1 μg of MAb RB6 significantly reduced the neutrophil counts 1 day later. Absolute neutropenia, defined as an ANC of less than 5 × 10⁵ cells/ml (500 neutrophils/μl), was induced by doses of ≥7.5 μg. At doses less than 50 μg, absolute neutropenia persisted for 1 day. Administration of 10 μg of MAb RB6 or more also resulted in significant ALC reductions, although variability was seen when mice were treated with doses ranging from 10 to 25 μg (Fig. 2). Mouse weight, the route of administration (i.v. or i.p.), and the mouse strain (C57BL/6, A/JCr, or DBA/2) did not influence the reproducibility, degree, or duration of neutropenia when 25 μg of MAb RB6 was given (data not shown). To assess other cell populations depleted by this MAb, we determined its effect on splenocytes (Table 1). As expected, the number of MAb RB6⁺ splenocytes was significantly reduced. Significant differences in other splenocyte populations were not detected despite a reduction in the peripheral ALC.

FIG. 2.

Summary of representative experiments to determine effect of dose of MAb RB6 on leukocyte populations. The bars indicate means, and the error bars indicate standard deviations. All experiments were performed with BALB/c mice, and treatments were administered i.v. Ten mice per group were used for the experiments whose results are shown in panel A, and five mice per group were used for the experiments whose results are shown in panels B, C, and D. In the experiment whose results are shown in panel D, two additional groups were treated with 25 or 200 μg of polyclonal rat IgG. In these groups, neither the ANCs nor ALCs differed from those obtained for groups that received PBS. Day 4 values were not obtained for mice that received 25 μg of MAb RB6. a, P < 0.05, as determined by the Holm-Sidak test, for comparison to the ANC on day 0 after one-way ANOVA; b, concomitant significant reduction in the ALC compared to the ALC on day 0 (P < 0.05, as determined by the Holm-Sidak test). c, P < 0.05, as determined by Dunn's method for comparison to the ANC on day 0 after one-way ANOVA on ranks; d, concomitant significant reduction in the ALC compared to the ALC on day 0 (P < 0.05, as determined by Dunn's method).

TABLE 1.

Mean leukocyte populations in the spleens of immunosuppressed mice^a

Population	Treatment group
	Naïve	MAb RB6	CPA	MAb RB6 + CPA
Total splenocytes (107)	4.0 ± 1.4	3.9 ± 0.5	0.9 ± 0.2b	1.1 ± 0.1
CD3+ (106)	16.2 ± 4.6	17.4 ± 1.4	4.7 ± 0.9b	6.5 ± 0.6
CD4+ CD8− (106)	9.2 ± 2.7	10.2 ± 1.1	2.7 ± 0.5b	3.8 ± 0.5
CD8+ CD4− (106)	6.9 ± 2.0	7.2 ± 0.5	2.0 ± 0.4b	2.6 ± 0.2
CD4+ CD8+ (105)	2.7 ± 2.1	3.3 ± 0.9	0.4 ± 0.1b	0.9 ± 0.4
CD19+ (106)	19.6 ± 8.2	19.3 ± 2.9	3.5 ± 0.7b	4.0 ± 0.5
MAC-3+ (106)	1.9 ± 0.9	2.4 ± 0.5	0.7 ± 0.1	0.8 ± 0.1
Ly6-G+ (105)	5.0 ± 1.5	0.2 ± 0.1b	1.5 ± 0.4	0.1 ± 0.0b

^aMice were treated 1 day prior to determination with 25 μg of MAb RB6 i.v., with 150 mg of cyclophosphamide (CPA) per kg i.p., or with both agents. The values are means ± standard deviations per spleen. Each group contained four mice.

^bP < 0.05, as determined by Dunn's method for multiple comparisons after Kruskal-Wallis one-way analysis of variance on ranks.

To determine whether sustained neutropenia could be induced by repeated injection of MAb RB6, mice were given 25 μg of MAb or PBS i.v. every other day for four doses, and one-half of the mice were bled on each subsequent day. The ANC was 4.5 × 10⁵ ± 4.2 × 10⁵ cells/ml in MAb RB6-treated mice by day 2 and returned to the baseline value by day 5. Thus, administration of repeated doses did not induce sustained neutropenia. To determine whether our inability to induce sustained neutropenia was due to the development of mouse anti-rat IgG2b, serum from these mice was used for an ELISA. No antibody binding to MAb RB6 was detected in samples from mice that received repeated injections of PBS or in mice that received MAb RB6 on days 0 through 3. However, IgM binding to MAb RB6 became detectable on day 4 and was detected in all mice from day 5 to day 10, by which time the titers were declining (data not shown). Low levels of murine IgG binding to MAb RB6 were detected in serum on day 10 in all three mice tested but not in serum from earlier times or from control mice. To determine whether a single injection of MAb RB6 induced an anti-rat IgM response, a subsequent experiment was performed in which four mice each received 25 μg of MAb RB6 i.v. and then were bled 7 and 14 days later. Seven days after administration, naïve mice had a geometric mean IgM titer of 1:60 (range, 1:50 to 1:100), while mice that received MAb RB6 had a geometric mean IgM titer of 1:400 (range, 1:200 to 1:800) (P = 0.03, as determined by the Mann-Whitney rank sum test). By 14 days, the IgM titer in these mice was not significantly different from that in naïve mice.

Pathogenesis of invasive aspergillosis following administration of MAb RB6.

In BALB/c mice that received 25 μg of MAb RB6, the minimum inoculum that reproducibly resulted in 100% mortality was ∼10⁶ conidia reaching the lungs. The inoculum required for mortality was the same in mice that received 200 μg of MAb RB6; however, deaths did not occur until 5 days after infection, compared to 3 days when the lower dose was administered. Mice that received PBS did not die following comparable exposures, which were the maximum that reproducibly could be delivered with a single pump when A. fumigatus strain ATCC 90906 was used (data not shown). Determination of the lung chitin concentrations revealed that the amounts of chitin present in lungs of mice that received 5 × 10⁵ conidia on days 1, 2, and 3 were not significantly different (Fig. 4A). Mice that received 1.7 × 10⁵ conidia had low levels of chitin on day 1, and there was a trend toward an increase on day 2 (P = 0.1, as determined by Student's t test) (Fig. 4B). On day 3, despite 100% mortality (by day 4) for the mice monitored for survival, in three of four mice only a low fungal burden was detected by either method, while the fourth mouse had a high fungal burden as determined by both methods, suggesting that there is heterogeneity in the ability of mice to control fungal growth. Examination of H&E-stained lung sections from these mice indicated that by day 3 inflammatory foci were frequently centered on large airways and bronchial lumens contained densely packed inflammatory infiltrates composed predominantly of neutrophils (Fig. 5). In some mice, thrombosis of nearby large blood vessels was observed. In most mice, intense mixed inflammatory cellular infiltrates, in which neutrophils were prominent, obliterated large portions of the distal airspace. GMS-stained lung sections demonstrated that foci of infection were predominantly bronchocentric. Collections of hyphae were present inside bronchi and invaded the surrounding parenchyma through the bronchial wall. However, in regions where there was intense inflammation in the terminal airways, hyphae were rarely observed. Faint GMS-positive material was present diffusely in these regions, as were dysmorphic hyphae. Occasionally, infectious foci bordered the pleura and were associated with pleural thickening with inflammatory cells.

FIG. 4.

Lung fungal burden in MAb RB6-treated BALB/c or C57BL/6 mice infected with an aerosol. (A) Sixteen mice of each strain received 25 μg of MAb RB6 1 day prior to infection with A. fumigatus. Four mice of each strain were killed 1, 2, or 3 days after they received 5.3 × 10⁵ conidia. Each symbol represents a single mouse. (B) Experiment with BALB/c mice, performed as described above for panel A, except that on day 3 the fungal burden was also determined by morphometry. The mean inoculum was 1.7 × 10⁵ conidia. (C) Ten MAb RB6-treated C57BL/6 mice received 2.6 × 10⁵ conidia, and the fungal burden was determined by chitin measurement (•) or by morphometry (○) in six of these mice on day 3. The mortality rate was 100% for additional mice infected with A. fumigatus that were monitored for survival (four, eight, and four mice in the experiments whose results are shown in panels A, B, and C, respectively). No sham-infected mice died in these experiments (four mice per experimental group).

FIG. 5.

Pathology of invasive pulmonary aspergillosis 3 days after infection in BALB/c (A and B) and C57BL/6 (C through F) mice immunosuppressed with MAb RB6. All images were obtained from mice used in experiments whose results are shown in Fig. 4B (BALB/c) and Fig. 4C (C57BL/6). (A) Large inflammatory focus in the pulmonary parenchyma. The arrow indicates a bronchial lesion in which hyphae are visible, demonstrating destruction of the bronchial wall and intraluminal hyphae. The asterisk indicates a thrombosed blood vessel. (B) GMS-stained section from a mouse with a relatively high fungal burden, demonstrating that hyphae are predominantly confined within the large airways. (C) Lung section from C57BL/6 mouse, demonstrating a similar pattern of dense inflammation. (D) GMS staining, showing that few hyphae are present. (E and F) Higher magnifications of regions in panels C and D, demonstrating the mixed inflammatory cellular pattern in which neutrophils are prominent. In the parenchyma of the distal airways, only hyphal fragments are present. (A through D) Scale bars = 50 μm; (E and F) Scale bars = 20 μm.

In C57BL/6 mice that received MAb RB6 1 day prior to infection, an inoculum consisting of 5 × 10⁵ conidia resulted in 100% mortality (Fig. 3B). The lung fungal burden rose somewhat more consistently in this mouse strain during the course of infection than in BALB/c mice (Fig. 4A and C). The histological appearance of sections from these mice was similar to the appearance in BALB/c mice treated with MAb RB6, and extensive inflammation was seen on day 3. No deaths occurred in sham-infected mice in any of the experiments described above for either mouse strain.

FIG. 3.

Survival of mice immunosuppressed with 25 μg of MAb RB6 and infected 1 day later in aerosol flasks. (A) BALB/c mice; (B) C57BL/6 mice. Each group contained six mice. The target inocula were 10⁵, 5 × 10⁵, and 10⁶ conidia. The actual inocula, determined by plating lung homogenates of two mice 2 h after infection, are indicated. The P value was 0.005 for a comparison of BALB/c mice infected with 6 × 10⁵ and 1.4 × 10⁶ conidia in panel A, as determined by log rank analysis. The data in panels A and B are representative of mortality rates obtained in five and four additional experiments, respectively.

i.t. infection.

We determined the minimum inoculum that reproducibly resulted in 100% mortality following i.t. infection and the number of conidia retrieved from the lung 2 h after infection. With strain ATCC 90906, 5 × 10⁶ conidia was the minimum lethal inoculum in BALB/c mice that had received 25 μg of MAb RB6 (data not shown). Plating of the lungs of two mice infected with 5 × 10⁶ conidia demonstrated that the mean number of retrievable conidia was 9.4 × 10⁵ ± 0.4 × 10⁵. Thus, administration of minimal consistently lethal inocula by the i.t. and aerosol routes resulted in retrieval of comparable numbers of CFU from the lungs. To determine whether the lack of demonstrable fungal growth was dependent on the infection method, lung fungal burden was determined by both chitin measurement and morphometry. Again, heterogeneity in the response to infection with a lethal inoculum was seen (Fig. 6).

FIG. 6.

Lung fungal burden in BALB/c mice immunosuppressed with 25 μg of MAb RB6 and infected i.t., expressed as the amount of chitin per lung (•) or by morphometry (○). Twenty mice were infected, and the lungs of two mice were plated for determination of the inoculum that reached the lung 2 h after infection. Four mice were monitored for survival, and the remainder were killed on days 1, 2, and 3 (five, five, and four mice, respectively) for determination of the fungal burden. The mortality was 100% in the group monitored for survival by day 4.

Administration of cyclophosphamide and corticosteroids.

A single i.p. injection of 150 or 200 mg of cyclophosphamide per kg resulted in severe leukopenia, and the neutrophil nadir was reached on day 4 after inoculation (Table 2). By 5 days after administration of a single dose, the total WBC counts and ANCs were at or near normal. Repeated administration of 150 mg of cyclophosphamide per kg produced neutropenia in most mice that was sustained for 16 days. Neutropenia induced by cyclophosphamide showed significant interanimal variation, while ALC depression was both more consistent and sustained. Administration of 200 mg/kg resulted in less interanimal variation (data not shown). Repeated administration of 150 mg/kg resulted in loss of ∼10% of the body weight, followed by weight stabilization, while administration of 200 mg/kg resulted in loss of 25% of the body weight (data not shown), which precluded use of this dose. Administration of 150 mg/kg markedly reduced all of the splenic lymphocyte populations tested (Table 1). A trend toward reduced numbers of Mac-3⁺ and Ly6-G⁺ splenocytes also was seen, but the differences were not statistically significant after correction for multiple comparisons.

TABLE 2.

Effect of cyclophosphamide on leukocyte populations^a

Expt	Day	WBC (106/ml)		ANC (105/ml)		ALC (106/ml)
		Distilled H2O	CPA	Distilled H2O	CPA	Distilled H2O	CPA
I	0	4.5 ± 1.4	4.6 ± 1.1	5.6 ± 2.9	6.2 ± 2.9	3.8 ± 1.0	3.7 ± 1.0
	2	5.3 ± 1.1	2.3 ± 0.8b	4.7 ± 3.0	6.9 ± 5.3	4.8 ± 1.1	1.5 ± 0.6b
	3	4.5 ± 2.2	2.0 ± 0.5	2.7 ± 1.4	3.1 ± 2.4	4.1 ± 2.0	1.7 ± 0.6
	4	10.7 ± 2.5	1.1 ± 0.7b	12.5 ± 4.4	0.4 ± 0.5c	9.3 ± 2.3	1.0 ± 0.7b
II	0	6.3 ± 2.6	3.5 ± 1.5	9.4 ± 7.2	5.0 ± 2.9	5.1 ± 1.8	2.9 ± 1.2
	3	6.7 ± 3.0	2.0 ± 1.1	9.3 ± 5.6	2.2 ± 1.8	5.8 ± 2.6	1.7 ± 0.9b
	4	5.1 ± 1.6	1.3 ± 0.6b	5.4 ± 2.3	0.1 ± 0.2c	4.4 ± 1.4	105 ± 0.8b
	5	9.9 ± 3.3	2.6 ± 0.9c	12.1 ± 8.2	3.1 ± 1.0c	8.5 ± 2.5	2.1 ± 0.8c
	6	8.5 ± 1.7	5.0 ± 0.9c	11.9 ± 4.9	10.8 ± 4.3	7.2 ± 2.0	3.8 ± 0.8c
III	0	NDf	4.8 ± 1.1	ND	6.3 ± 2.0 (6.5)	ND	4.0 ± 0.9
	4		2.9 ± 1.0d		0.6 ± 0.5 (0.7)e		2.8 ± 1.0d
	5		0.6 ± 0.3d		2.9 ± 5.0 (0.4)e		1.2 ± 0.3d
	8		1.7 ± 0.8d		2.5 ± 1.6 (3.2)		1.4 ± 0.6d
	12		2.0 ± 1.1d		2.0 ± 4.6 (0)e		1.8 ± 0.8d
	16		2.4 ± 0.5d		1.2 ± 1.7 (0)e		2.2 ± 0.5d
	20		3.3 ± 0.8d		6.6 ± 4.2 (6.6)		2.5 ± 0.7d

^aThe data are representative of the results obtained in three experiments. In experiments I and II, mice received 150 mg of cyclophosphamide (CPA) per kg on day 0, and each group contained four mice. The control mice received a comparable volume of distilled H₂O. In experiment III, the cyclophosphamide-treated mice received cyclophosphamide on days 0, 4, 8, 12, and 16, and each group contained five mice. The numbers in parentheses are the median values obtained for five mice.

^bP < 0.05 for a comparison to the group that received distilled H₂O on the same day as determined by Student's t test.

^cP < 0.05 for a comparison to the group that received distilled H₂O on the same day, as determined by the Mann-Whitney rank sum test.

^dP < 0.05 for a comparison to the day 0 value, as determined by the Holm-Sidak method for multiple comparisons after one-way ANOVA.

^eP < 0.05 for a comparison to the day 0 value, as determined by Dunn's method for multiple comparisons after Kruskall-Wallis one-way analysis of variance on ranks.

^fND, not determined.

When cyclophosphamide was administered on days −4, 0, 4, and 8 after infection, an inoculum of 10⁶ conidia that reached the lungs was required to reproducibly induce 100% mortality (data not shown), which was comparable to the results observed with MAb RB6. To identify a model in which death occurred with a lower inoculum and with a high lung fungal burden, we combined a single 35-mg/kg dose of triamcinolone, administered 1 day prior to infection, with repeated cyclophosphamide doses administered on days −4, 0, and 4. While 100% mortality occurred in mice that received <10⁴ conidia by day 4 or 5 after infection (data not shown), we consistently observed some deaths in mice in this model by day 7 after sham infection, despite addition of trimethoprim-sulfamethoxazole to the drinking water to prevent bacterial infection. Coadministration of dexamethasone at i.p. doses ranging from 2 to 10 mg/kg with cyclophosphamide did not render mice more susceptible to death from infection with lower numbers of conidia compared to administration of cyclophosphamide alone (data not shown).

Coadministration of MAb RB6 and cyclophosphamide.

For BALB/c mice, coadministration of 25 μg of MAb RB6 and 150 mg of cyclophosphamide per kg on days 0, 3, 6, and 9 resulted in significantly more severe neutropenia on day 3 than did administration of MAb RB6 or cyclophosphamide alone (the ANCs were 12.7 × 10⁵ ± 5 × 10⁵, 9.8 × 10⁵ ± 2.8 × 10⁵, and 0.5 × 10⁵ ± 0.3 × 10⁵ for groups treated with MAb RB6, cyclophosphamide, and both agents, respectively [means ± standard deviations; four mice per group; P < 0.05 for both comparisons, as determined by Dunn's method for multiple comparisons after Kruskal-Wallis one-way ANOVA on ranks]). By day 6, the ANCs were comparable for mice treated with cyclophosphamide alone and mice that received both agents, and by day 9, the ANCs were comparable to the ANCs on day 0 for all treatment groups (data not shown). Coadministration of 9 mg of methotrexate per kg did not alter the ANC in the peripheral blood of mice treated with MAb RB6 on day 3 (data not shown). In mice treated with both cyclophosphamide and MAb RB6, the trend toward a reduction in splenic leukocyte populations 1 day later paralleled the trend observed in mice treated with cyclophosphamide alone, but the differences were not statistically significant (Table 1). The number of Ly6-G⁺ cells was significantly reduced compared to the number in naïve mice. These results suggest that the effects of MAb RB6 and cyclophosphamide are additive. The anti-MAb RB6 titers in these mice were not different from those in naïve mice either 7 or 14 days after administration of a single dose of MAb RB6 with cyclophosphamide (data not shown).

To determine whether administration of cyclophosphamide allowed a response to repeat doses of MAb RB6, four C57BL/6 mice were given MAb RB6, cyclophosphamide, or both on day 0 and then repeated injections of MAb RB6 on days 3 and 6 after bleeding. Although, as in BALB/c mice, more severe neutropenia was seen on day 3 in mice treated with both agents, by day 6 there was no difference between the groups, and the ANCs were not different from the values obtained on day 0 (data not shown). Significant reductions in ALCs were seen in mice treated with cyclophosphamide alone or in mice treated with both agents on days 3, 6, and 9 compared to the values obtained on day 0.

Pathogenesis of invasive aspergillosis following coadministration of MAb RB6 and cyclophosphamide.

An inoculum of 10⁵ conidia that reached the lung resulted in 100% mortality in BALB/c mice (data not shown), while C57BL/6 mice died after they received 5 × 10⁴ conidia (Fig. 7A). In BALB/c mice that received an inoculum of 1.9 × 10⁵ conidia, heterogeneity in the fungal burden was seen (data not shown). In C57BL/6 mice that received 9 × 10⁴ conidia, the fungal burden 3 days after infection was more consistently higher than the fungal burden in mice that received MAb RB6 alone, based on both chitin and morphometry (Fig. 7B). In BALB/c mice, some lung sections contained areas which resembled the inflammatory lesions seen in mice treated with MAb RB6 alone, while other sections contained regions of necrotic lung with hyphal growth and dense inflammatory infiltrates bordering the regions of necrosis (Fig. 8). In C57BL/6 mice, inflammation was present in lung sections of three of six mice in areas smaller than those observed in MAb RB6-treated mice. Extensive necrosis of the lung parenchyma was present and was not surrounded by inflammation, although small areas of perivascular inflammation were observed in some sections. Although many infectious foci appeared to originate from bronchi, diffuse extension of hyphae throughout the lung parenchyma was observed. Hyphae commonly invaded blood vessels, and vascular thrombosis occurred.

FIG. 7.

Invasive pulmonary aspergillosis in C57BL/6 mice treated with MAb RB6 and cyclophosphamide 1 day prior to infection. (A) Survival analysis. Ten mice received a mean inoculum of 5.6 × 10⁴ ± 0.4 × 10⁴ conidia, while four mice received a sham aerosol infection. All sham-infected mice remained alive for at least 14 days. (B) Fungal burden. Each symbol represents a single mouse (n = 6). The experiment was repeated, and similar results were obtained.

FIG. 8.

Lung pathology 3 days after infection of mice immunosuppressed with MAb RB6 and cyclophosphamide 1 day prior to infection. (A) H&E-stained section from a BALB/c mouse that received an inoculum of 1.8 × 10⁵ conidia, showing bronchial lesions with necrotic lung surrounded by inflammatory infiltrate. (B) GMS staining of the region shown in panel A, showing invasion through the bronchus into the surrounding lung parenchyma. (C) In C57BL/6 mice, extensive hyphal spread through the lung parenchyma resulted in severe necrosis with inflammation perivascularly but not in the distal airspace and large areas of necrotic lungs. (D) GMS-stained section of the region shown in panel C. (E) H&E staining of an infectious focus in the lung of a C57BL/6 mouse, showing vascular thrombosis near a large bronchus. (F) GMS staining of the region shown in panel E, showing extensive hyphal penetration of the vessel. C57BL/6 sections were obtained from the mice described in the legend to Fig. 6. The arrows in panels E and F indicate a smaller, thrombosed blood vessel in which hyphae were visible. (A through D) Scale bars = 0.2 μm; (E and F) scale bars = 0.1 μm.

Effect of Tween 20 on conidial surface charge.

The zeta potential of strain ATCC 90906 conidia collected without Tween 20 was −20.2 ± 1.6 (mean ± standard deviation), while that of conidia collected in Tween 20 was −12.9 ± 2.0, representing a 36% reduction in the net negative surface charge (P < 0.0001, as determined by Student's t test).

DISCUSSION

Humans, like immunocompetent mice, are relatively resistant to invasive disease due to A. fumigatus, and disease occurs in the setting of host damage (7) resulting from disease states that impair immune function, immunosuppressive agents, or underlying structural lung disease. Pathogenesis, as reflected by analysis of histopathology, correlates with the specific underlying host defect (2). Study of invasive aspergillosis in neutropenic animal models has been emphasized (www.niaid.nih.gov/contract/archive/rfp0309amd1.pdf). The importance of neutropenia in murine models of invasive pulmonary aspergillosis is highlighted by the need to induce this defect to unmask contributions of other host components (8, 24, 39). Therefore, we studied the effects of agents commonly used to induce neutropenia and applied them to aerosol models of invasive aspergillosis to enhance our understanding of their relevance to human disease when they are applied to the study of pathogenesis. We used lethality, fungal burden, and histopathology as readouts for pathogenesis in mice immunosuppressed with MAb RB6 or cyclophosphamide, with or without corticosteroids. Because of limitations of these models, we then coadministered cyclophosphamide with MAb RB6 in an effort to prevent the development of an antibody response to this rat immunoglobulin.

The use of pulmonary infection for the study of invasive aspergillosis is preferable, as the lung is the site of primary infection. The systemic circulation, which has an immune environment that is different from that in the lung, likely is not exposed to conidia. In contrast to intranasal and i.t. infection models or models in which conidia are nebulized in aerosol chambers (52), the aerosol model used in the present study does not require collection of conidia in suspensions, to which Tween commonly is added to prevent agglutination. Although at present our understanding of the mechanisms by which the surface properties of conidia contribute to pathogenesis is limited, hydrophobicity is considered important, and other physiochemical properties likely affect the interaction of conidia with the host lung (32). Here, we demonstrate that the surface charge of conidia may be altered by Tween treatment, suggesting that this model has a potential advantage over the methods described above. Aerosol administration produced reproducible infection inocula and clinical outcomes. Furthermore, by consistently monitoring the inoculum obtained from infected mice, we could exclude trials in which more-than-desirable variability occurred. We, therefore, concur with the conclusion of Brieland et al. (6) that the aerosol flask model is attractive.

The methods of neutropenia induction have included administration of chemotherapeutic agents and MAb RB6. Because the latter agent had the potential to provide a relatively defined host defect, we explored the use of MAb RB6 in order to understand its effects more fully. This MAb cross-reacts with Ly6-C, which is found on subpopulations of CD8⁺ T cells and monocytes, as well as plasmacytoid dendritic cells (20, 29, 42). We were unable to identify a dose at which neutropenia consistently was selective and persistent for more than 1 day. Following administration of 200 μg, a dose commonly used for animal models of other infectious diseases (49, 50), marked lymphopenia occurred, a finding that has implications for the conclusions that can be drawn about the role of neutrophils based on depletion with this MAb. We confirmed that the duration of neutropenia induced by this MAb could not be extended by repeated doses (26), a feature that may be due to the development of the antibody response to this rat reagent. However, although coadministration of cyclophosphamide, an agent that induces B-cell tolerance, with MAb RB6 delayed the development of an anti-rat antibody response, our ability to achieve sustained neutropenia remained limited. We did not evaluate the effect of continued administration of cyclophosphamide in this context, as such manipulation would abrogate any selectivity advantage of MAb RB6.

As expected, administration of cyclophosphamide depleted a broad range of host cells, including neutrophils, and repeated administration allowed more sustained effects. Regardless of the lack of selectivity, this method of immunosuppression is relevant to human disease, since, despite the correlation between neutropenia and invasive aspergillosis, many affected patients have received similar drugs. The relative sensitivities of lymphocyte populations to cyclophosphamide-induced depletion vary, so that B cells are more susceptible than T cells and CD4⁺ T cells are more susceptible than CD8⁺ T cells. However, cyclophosphamide also can deplete regulatory T-cell populations (reviewed in reference 21) and, therefore, may have proinflammatory actions. The known effects of cyclophosphamide on local pulmonary immunity include inhibition of complement synthesis and reduction of chemotactic activity by guinea pig bronchoalveolar macrophages (45, 46). The contribution of each of these extended effects to the pathogenesis of invasive pulmonary aspergillosis in murine models is unknown.

The fungal inoculum is an important determinant of lethality in murine models (16). The inoculum required for production of invasive disease in immunocompromised hosts is unknown, but it is probably smaller than the inocula commonly administered intranasally or i.t. in murine infections, which typically range from 5 × 10⁶ to 2 × 10⁷ conidia. We found that in BALB/c mice immunosuppressed with MAb RB6, the lethal inoculum in this aerosol model was comparable to that in the i.t. model and the same as that required following cyclophosphamide administration. The inoculum required for 100% mortality in C57BL/6 mice was fivefold less in mice that received either MAb RB6 or both MAb RB6 and cyclophosphamide, suggesting that there are subtle differences between the mouse strains. Administration of a higher dose (200 μg) of MAb RB6 did not reduce the inoculum required for 100% mortality, although deaths did not occur until 5 days after infection (data not shown), which is consistent with the longer time to neutrophil recovery when this dose is given. Coadministration of MAb RB6 and cyclophosphamide resulted in a ∼10-fold decrease in the lethal inoculum in both mouse strains. Thus, this approach offers a less problematic alternative to the combined cyclophosphamide-corticosteroid regimens that have been used to develop lower-inoculum models (55). Although cyclophosphamide appeared to delay or prevent the development of the anti-rat antibody response, as noted above, we did not observe continued efficacy of repeated MAb RB6 doses. Rather, enhanced susceptibility may be related to the ability to sustain maximal neutropenia for the first 3 days after infection, as the nadir counts with either individual agent lasted only 1 to 2 days, despite a somewhat longer statistically significant duration of reduction. Again, given the broad range of immune cells affected by this agent, the possibility of contributions by depletion of other cells cannot be excluded.

The course of disease in both BALB/c and C57BL/6 mice made neutropenic with all of the immunosuppression regimens examined was acute, as all deaths occurred within 3 to 5 days after infection. When inocula smaller than those resulting in 100% mortality were administered, only occasional mice that survived longer than 5 days after infection subsequently died. Plating of lung homogenates from such mice obtained 7 days after infection only occasionally yielded growth of small numbers of A. fumigatus colonies (data not shown), suggesting that the surviving mice cleared the infecting inoculum.

Following induction of neutropenia with MAb RB6, death coincided with a return of the neutrophil counts to the baseline value, a variable pulmonary fungal burden, and a massive inflammatory response in the lung parenchyma in the absence of disseminated disease (data not shown). These features suggest that mice died as a result of the inflammatory response. Fatal pulmonary complications are associated with rapid neutrophil recovery, and in human immunodeficiency virus-negative patients, more than one-third of reported cases of immunorestitution disease occur in patients with invasive aspergillosis (10, 57). Thus, the time of restoration of immunity is a time of vulnerability in human disease. In some mice, infectious foci with intact hyphae were observed in large bronchi. Therefore, it is possible that these lesions were responsible for the death of some mice. Disease confined to large airways, a pattern that is observed less commonly than bronchopneumonia, may result in acute airway obstruction in humans and has been reported after resolution of neutropenia (4, 22). However, the presence of more distal lesions in the mice at earlier times after infection (data not shown) and the consistent finding of extensive parenchymal inflammation in which only hyphal remnants were observed suggest that these inflammatory lesions were sufficient to cause death. Furthermore, although significant bronchial infectious foci could have been missed by microscopic examination of limited numbers of sections, the strong correlation of whole lung chitin content with morphometry results supports the finding that death occurred in mice with undetectable fungal burdens. The correlation of the values obtained by these methods for the samples used in this study was highly significant (P < 0.0001; Spearman rank order correlation coefficient = 0.52; n = 68). Use of both methods for evaluation of the fungal burden provides additional confidence in the results obtained with a system in which CFU determinations are of questionable relevance.

In mice in which neutropenia was induced with both MAb RB6 and cyclophosphamide, infectious foci extended outward from bronchi and bronchioles. Bronchopneumonia is the most common pattern observed in human invasive pulmonary aspergillosis (58). Distal lung disease results either from aspiration of conidia into the respiratory bronchioles and alveolar ducts with direct spread or from colonization of the bronchial mucosa by invasion through the wall (25). Thus, the pathology observed in this model is consistent with that of human disease in neutropenic hosts. The pathogenesis of the target lesion, a hallmark of invasive aspergillosis, reflects centrifugal expansion of a small focus of bronchitis or bronchopneumonia, resulting in a spherical nodule of necrotic lung in which vascular invasion is present, while invasion of larger vessels may result in classic wedge-shaped infarcts (23). Angioinvasion, a feature that can occur in any pattern of invasive aspergillosis but that is particularly associated with disease in neutropenic patients (23), also was observed. Necrosis may result from vascular invasion or toxic fungal products (25, 28). In BALB/c mice, inflammatory infiltrates bordered necrotic lungs, and the fungal burden in lethally infected mice was inconsistent. However, in C57BL/6 mice, death appeared to occur independent of a vigorous inflammatory response and coincided with a higher fungal burden. The difference in the responses of these mouse strains could represent subtle variations in the degree or persistence of immunosuppression or species-related disparities in the inflammatory response to A. fumigatus products, such as those that translate into marked differences in cell-mediated immune responses to other pathogens (27, 56).

In summary, as in human disease, the method of neutropenia induction is an important determinant of pathogenesis in murine models of invasive pulmonary aspergillosis. The host response appears to be the primary cause of damage in mice immunosuppressed with MAb RB6, and models in which this agent is used may be the best models for studying the dynamics of pathogenesis of invasive aspergillosis in the setting of neutrophil recovery and the accompanying increase in host response. In C57BL/6 mice treated with a combination of MAb RB6 and cyclophosphamide, fungal growth may play a more prominent role. Differences in pathogenesis must be borne in mind when these models are used to study the role of either fungal molecules or host inflammatory responses to infection or during the evaluation of pharmacologic or immune modulatory agents.