Molecular Characterization of Trimellitic Anhydride–induced Respiratory Allergy in Brown Norway Rats

Abstract

To contribute to the hazard identification of low molecular weight (LMW) respiratory allergens, respiratory allergy induced by trimellitic anhydride (TMA) was characterized by whole genome analysis of lung tissue and blood proteomics in Brown Norway rats. Dermal sensitization (50% and 25% w/v) with TMA and an inhalation challenge of 15 mg/m³ TMA-induced apneas, laryngeal inflammation, increased numbers of eosinophils, neutrophils and macrophages in bronchoalveolar lavage (BAL), and increased immunoglobulin E levels in serum and lung tissue. Whole genome analysis of lung, sampled 24 hours after challenge, showed expression changes of not only genes belonging to several Gene Ontology groups with up-regulation of inflammatory-associated genes and those associated with lung remodeling but also genes involved in downsizing these processes. Blood proteomics reflected activation of inflammation-inhibiting pathways. Unsensitized animals challenged with TMA exhibited also an increased number of macrophages in BAL, but gene expression in the above-mentioned gene pathways was unchanged or down-regulated. The authors conclude that parameters for lung remodeling can be a valuable tool in hazard identification of LMW respiratory allergens.

Introduction

Asthma is present in 10% to 15% of the adult population, and it is estimated that up to 15% of all cases in industrialized settings are associated with workplace exposure (Bauer et al. 1994; Bauer et al. 1995). Diisocyanates and acid anhydrides are among the most common low molecular weight (LMW) classes of chemicals responsible for occupational asthma. The symptoms of LMW asthma are similar to those induced by high molecular weight (HMW) compounds. Asthma in man is defined as variable or intermittent airway obstruction associated with bronchial hyperresponsiveness (BHR) and is considered to be an inflammatory disorder. There is a relationship between magnitude of airway inflammation and BHR, but factors such as airway remodeling also influence hyperresponsiveness (Haley and Drazen 1998). The development of HMW asthma is associated with elevated immunoglobulin E (IgE) levels, the predominance of Th2 cytokine production and eosinophilic inflammation of the airways. However, IgE may not be essential for allergen-induced BHR and airway inflammation, and thus far, the physiological role of IgE in the airways has not been fully elucidated (Moffatt 2005). Asthma induced by the diisocyanates has a low association with atopy, low prevalence with specific IgE antibodies and a mixed Th1-Th2 response (Matheson et al. 2005; Jones et al. 2006). For acid anhydride allergens like trimellitic anhydride (TMA), the relationship between disease status and IgE is stronger but on a group not on an individual basis (man: Liss et al. 1993; rodents: Arts et al. 1998; Ban and Hettich 2001; Dearman et al. 2002).

Because of these uncertainties in the pathogenesis of asthma, there are, at present, no validated animal models to test the potential of a chemical to induce respiratory allergy. Candidate models for hazard identification investigate respiratory allergic potential in rodents on the basis of serum IgE or cytokine fingerprinting in skin and skin-draining lymph nodes following dermal application of the chemical (Dearman, Skinner, and Humphreys 2003). However, immune responses measured in blood, skin, and skin-draining lymph nodes may differ in various ways from those measured in the respiratory tract (Gould et al. 2006; Fritz et al. 2008) and particularly the lungs, for asthma is a disease of the smaller bronchi/bronchioles. Following intranasal exposure to diisocyanates, the cytokine profiles in the lymph nodes were not predictive of differences in responses to inhaled methacholine to measure BHR (Selgrade et al. 2006).

Some animal toxicity models include a challenge phase, in which changes are measured in breathing parameters and respiratory-tract inflammation on an inhalation challenge in sensitized animals (reviewed by Pauluhn and Mohr 2005; Arts and Kuper 2007; Kimber et al. 2007). These models resemble human disease in some but not all aspects. The mechanistic basis for species differences in response to a chemical may be found at any level of biological organization. Microarray technology enables us to compare the responses of test animals with those of humans at the global transcriptome level (Mattes 2006). Changes in mRNA (microarrays) that occur in the asthmatic lungs of man specifically refer to cytokines and chemokines and, to some extent, to genes related to lung remodeling (Ober and Hoffjan 2006; Bloemen et al. 2007), but there is a need to identify groups of genes related to the causative allergen (Holsapple et al. 2006). This will contribute to cross-species comparative toxicogenomics for safety assessment and to the identification of novel targets for in vitro toxicity tests.

This article aims to improve the hazard identification of respiratory allergens by molecular characterization of respiratory allergy in Brown Norway (BN) rats induced by the LMW respiratory allergen TMA, using a protocol that successfully identified chemical respiratory allergens in the high IgE-responding BN rat (Pauluhn et al. 2002; Zhang et al. 2004; Arts and Kuper 2007). Whole genome analysis (microarrays) of lung tissue was performed, and the results were related to functional (breathing) variables, allergic inflammation, and proteomics in blood.

Materials and Methods

Animals and Maintenance

Female, seven-to-eight-week-old, inbred BN rats were purchased from Charles River Deutschland GmbH (Sulzfeld, Germany). The animals were acclimatized for seven days before the start of the study. They were kept under conventional laboratory conditions and received the institute’s grain-based, open-formula diet and unfluoridated tap water ad libitum. All animal procedures were approved by the TNO Committee of Animal Welfare.

Treatment Schedules and Groups

The study was conducted with four groups of rats according to the scheme depicted in Table 1. Blood was collected one day before the start of the study. Female BN rats received 150 μl of TMA (purity 97%; Aldrich, Brussels, Belgium; 50% w/v) in a 4:1 (v/v) mixture of acetone and olive oil (AOO) as the vehicle in a 4:1 (v/v) mixture of acetone (Merck, Darmstadt, Germany) and refined olive oil (AOO; Sigma Diagnostics, St. Louis, MO, USA) as the vehicle on each flank (approximately 12 cm² each). Flanks had been shaved with an electrical razor at least two to three days earlier. Seven days after the first sensitization, they received 75 μl of TMA (25% w/v) on the dorsum of both ears. Controls received vehicle AOO. On day 21, basal breathing parameters (frequency, tidal volume, and pattern) were assessed, followed by TMA inhalatory challenge and assessment of the same breathing parameters during and after challenge. Animals were challenged by inhalation of a minimally irritating (based on breathing frequency) concentration of 15 mg/m³ of TMA for 15 minutes. On day 22, breathing parameters were assessed again, and afterward, necropsy was performed.

Atmosphere Generation and Analysis

During inhalation challenge, an all-glass nebulizer of the institute’s design was used to generate the test atmospheres from freshly prepared solutions of TMA in acetone. The acetone concentration was kept between 3,000 and 5,000 ppm (~7–12 g/m³), which levels are far below the level inducing sensory irritation (Alarie 1973; De Ceaurriz et al. 1981; Schaper and Brost 1991) and which indeed did not induce changes in breathing pattern in either sensitized or unsensitized rats (Arts et al. 1998). The challenge concentration of TMA was based on a previous study performed in BN rats (Arts et al. 2004). Atmospheric concentrations of TMA were determined gravimetrically by filter sampling, and those of acetone, by calculations based on the nominal concentration, assuming the usual 100% generation efficiency for this vapor. The particle size distribution of TMA in the test atmosphere was determined using a ten-stage cascade impactor (Andersen, Atlanta, GA, USA). Due to the small sampling air-flow rate and the large total volume required for analytical and particle-size determinations, samples were not collected during the challenge itself but immediately prior to or after challenge. The mass median aerodynamic diameter of the aerosolized TMA particles was between 0.9 and 1.2 μm with a geometric standard deviation between 2.5 and 2.9.

Inhalation Challenge and Measurements of Breathing Parameters

Rats were individually restrained in Battelle tubes, and each tube was placed into one of four whole-body plethysmographs connected to the central exposure unit. In this experimental setup, two TMA-sensitized (+/+) and two vehicle-treated (−/+) rats, at a time, were first exposed to fresh air for at least 25 minutes (prechallenge period) and then to the TMA atmosphere for exactly 25 minutes (challenge period), followed by a recovery (postchallenge) period of 30 minutes.

As a surrogate for changes in lung function, breathing parameters (frequency, pattern, and tidal volume; Arts and Kuper 2007) were monitored by means of recording the pressure signal in the plethysmograph before, during, and after challenge. Before challenge, breathing parameters were monitored approximately 20 seconds each minute, starting 6 minutes prior to challenge. During challenge, breathing parameters were monitored continuously, whereas after challenge, they were monitored approximately 20 seconds during each minute for the first 10 minutes, followed by 20 seconds each 3 minutes for 9 minutes, and followed by 20 seconds each 5 minutes for 10 minutes. Furthermore, breathing parameters were monitored approximately 20 seconds each minute for 6 minutes, 24 hours after challenge. Mean values were thus obtained based on six preexposure, fifteen exposure, fifteen postexposure (immediately after), and six postexposure (24 hours after) values.

Clinical Signs, Serum Collection, Body and Organ Weights, and Necropsy

The animals were observed daily and weighed shortly before the TMA dermal application, at weekly intervals thereafter, and just prior to necropsy. Individual serum samples were prepared from blood withdrawn via the orbital plexus prior to sensitization and via the abdominal aorta at necropsy. For blood proteomics, protease inhibitor cocktail was added to freshly obtained serum samples. The samples were stored at −80°C until analysis. For IgE analysis, the serum samples were stored at −20°C until analysis by Enzyme-Linked-Immuno-Sorbent Assay (ELISA). At necropsy, animals were anaesthetized with pentobarbital, killed by exsanguination from the abdominal aorta and examined grossly for abnormalities. Liver, kidneys, and the unlavaged left lung were weighed (after bronchoaleolar lavage had been performed on the right lung lobes; see below). The left lung was inflated with 50% Tissue Tek in saline, quick frozen on dry ice and kept at −70°C for microarray analysis or immunohistochemistry. The nasal tissues, trachea, and larynx were collected and fixated in neutral, phosphate-buffered 4% (v/v) formaldehyde for histopathological evaluation.

Bronchoalveolar Lavage

At necropsy, the right lung lobes were lavaged two times with a volume of 23 ml saline per kg bw, after binding of the left (main) lobe, which was used for histopathological evaluation (see below). The total amount of retracted lavage fluid was weighed and retained on ice. The bronchoalveolar cells were isolated from the supernatant by centrifugation (250 g) during 5 minutes at 4°C and resuspended in 0.5 ml saline to assess total cell and differential cell numbers. Total cell numbers were counted using an automated haematology analyzer (K-800, Sysmex, Toa, Kobe, Japan). The percentage of viable cells was determined using an acridine orange/ethidium bromide staining method in combination with fluorescence microscopic evaluation. For differential cell counts, cytospins were prepared and stained with May-Grunwald Giemsa. At least 200 cells were counted per animal to determine absolute numbers and percentages of macrophages or monocytes, lymphocytes, neutrophils, and eosinophils.

Serum and Lung Homogenate IgE Levels

Total IgE levels in serum or lung homogenate of snap-frozen left lung tissue were analyzed by means of an ELISA as described earlier (Arts et al. 1997). The concentration of IgE in the samples was calculated using a standard curve obtained with known quantities of monoclonal rat IgE and expressed as μg/ml serum.

Histopathology and Immunohistochemistry

The formalin-fixed nasal tissues, larynx, and trachea were embedded in paraffin wax and sectioned at 5 μm. Nasal tissues were cut at six levels according to Woutersen et al. (1994). The larynx was cut longitudinally through the epiglottis. The cranial part of the trachea was cut transversally and the caudal part longitudinally, together with the bifurcation and the two extra-pulmonary bronchi. The sections were stained with hematoxylin and eosin.

Cryostat sections were made from the deep-frozen left lungs of three animals per group (the other three were used for microarray analysis; see below) and from the lavaged right lung lobes. The 7 μm sections of the right lung lobes were stained with hematoxylin and used only for counting the number of small, medium, and large granulomas (Figure 1A). The sections of the three left lungs were stained for epsilon chain of rat IgE (MARE; Oxford Biotechnology, Oxford, UK; 1:500), CD4 (W3/25; Serotec, Oxford, UK; 1:800), CD8 (OX8; Serotec; 1:800), or CD161 (10/78; Serotec; 1:800), using a two-step indirect immunolabeling. After drying for one night at room temperature, the sections were fixed in cold acetone for 7 minutes at 4°C. Unspecific binding was blocked with 25% normal goat serum in PBS for 15 minutes. The sections were incubated for 60 minutes with the monoclonal antibodies listed above. Control sections were processed in the same way, except for the specific antibodies. For the second step, the sections were incubated with powervision poly HRP-anti-mouse IgG mixed with normal rat serum (100:5, first preincubation for 60 minutes) for 30 minutes. The sections were then treated with the chromogen NovaRed for 8 minutes and were additionally contrasted with hematoxylin for 20 seconds. All incubations were carried out at room temperature in a humid chamber and were followed by a three-times washing step with PBS. Positive cells were counted in the lung parenchyma at 5 square fields per left lung at 250x magnification, using a square grid of 0.4 mm², or counted as number of cells per terminal bronchiole transsection in the left lung (5 transsections per lung) and as number of cells per 0.2 mm epithelium of main intrapulmonary bronchioles/bronchi (3 sites per bronchiole).

Microarray Analysis

Total RNA was isolated from frozen left lung tissue of three rats per group, as described previously (Kuper et al. 2008). The RNA was labeled and hybridized to Affymetrix rat 230–2.0 GeneChip, containing probesets for 31,000 rat genes at ServiceXS (Leiden, the Netherlands). Data were loaded into GCOS software (Affymetrix, USA). GCOS data were transferred to GeneSpring (Agilent/Silicon Genetics, USA) for GC-RMA normalization. Log-transformed, normalized data were analyzed using Rosetta Resolver (RosettaBio, USA), Spotfire, and Excel (Microsoft, USA). Data from GCOS were annotated with Gene Ontology (GO) information via the SOURCE Web site (Stanford University, Palo Alto, CA, USA). Automated annotation updates were also performed with BRB ArrayTools. A subset of “expressed genes” was selected by filtering on expression level > 15 in at least three samples (18,624 probe-sets). To explore the variability in gene-expression groups, the explorative multivariate methods principal components analysis and hierarchical clustering were performed with Spotfire, Rosetta, and GeneSpring (demo version). Selection of differentially expressed genes was done by t-tests or by filtering on more than 25% up- or down-regulation in all three rats in a group.

A multiple testing correction (estimation of False Discovery Rate [FDR]) was incorporated into the statistical analysis of the microarray data. An analysis of variance (ANOVA) was performed to test for significant differences in expression among the four treatment groups. In addition, t-tests were performed to test for differences among the groups.

The microarray data set was also analyzed on the pathway or gene-group level. To infer gene groups that are significantly co-expressed we used T-profiler (Boorsma et al. 2005). T-profiler uses the unpaired t-test to score changes in average activity of predefined groups of genes. The gene groups are defined on the basis of GO categorization and have to contain at least six gene members (Ashburner et al. 2000; Boyle et al. 2004). T-profiler uses the log(2) ratios of gene expression between the treatment and the reference as input. After applying the unpaired t-test for each gene group, a p value was calculated from t using the t-distribution with n − 2 degrees of freedom, where n is the total number of genes tested. The p value was corrected for multiple testing by multiplying it by the number of gene groups that are being tested in parallel (Bonferonni correction) leading to an E value. To make the procedure more robust against outliers, the highest and lowest value of each gene group was discarded. This is similar to the jackknife procedure. A major advantage of T-profiler is that no parameters have to be tuned; the complete transcriptome is used for the calculation. The analysis is supported by the T-profiler Web site (http://www.t-profiler.org).

Blood Proteomics

Sample Preparation for Proteomic Analysis

The serum samples, obtained at necropsy, were pooled per group (−/−; +/−; −/+; +/+). Albumine and immunoglobulin G were removed (Albumine en IgG removal kit). Unless indicated otherwise, all reagents and equipment for blood proteomics were obtained from Amersham biosciences (Uppsala, Sweden). Per sample, 37.5 μl of serum was incubated for 1.5 hours at room temperature with 750 μl of antibody affinity resin from the kit. The mixture was then centrifuged for 5 minutes at 11,000 rpm and 4°C. The proteins in the supernatant were precipitated by adding four volumes of ice-cold acetone (1.6 ml). After 2 hours of incubation at −20°C, the mixture was centrifuged at 13,000 rpm and 4°C. The pellet was air dried and dissolved in 100 μl of DIGE buffer (8 M Urea, 4% w/v CHAPS, and 30 mM Tris), and the pH was adjusted to 8.5. The protein content was determined using the modified method of Bradford (Bio-Rad, Veenendaal, the Netherlands). Each sample (200 μg) was labeled with 4 μl of Cy 5 CyDye^™ DIGE fluors minimal dye (400 μM). After 30 minutes, the incubation was stopped by adding 1 μl of 10 mM Lysine. The labeled samples were further diluted with an equal volume of 2x sample buffer containing 8 M urea, 4% w/v CHAPS, 130 mM DTT (Sigma), and 2% Pharmalyte^™ 3–10. The internal standard included 140 μg of each sample (4 samples in total), labeled with CyDye 5. The two labeled samples were mixed, and the volume was adjusted to 350 μl with rehydration buffer containing 8 M urea, 4% w/v CHAPS, 2 mM tributyl phosphine (Fluka, Buchs SG, Switzerland), and 1% (v/v) immobilized pH gradient (IPG) ampholytes pH 4–7. Per gel, one sample (CyDye 3) and one internal standard (CyDye 5) were run. The gels, sixteen in total (4 per group), were processed in two separate 2-D experiments in which a duplicate of each group was performed simultaneously. The first dimension was carried out on an IPGphor system using pH 4–7 IPG gel strips of 18 cm. The isoelectric focusing was performed at 20°C under the following conditions: 12 hours at 30 V, 30 minutes at 150 V, 1 hour at 300 V, 1 hour at 1,500 V, and 6 hours at 8,000 V. After isoelectric focusing, the IPG strips were equilibrated for 30 minutes in 6 M urea, 30% (v/v) glycerol, 5 mM tributyl phosphine, and 2% (m/v) SDS in 0.05 M Tris-HCl buffer, pH 8.8. The second dimensional separations were carried out on custom-made 12% SDS-polyacrylamide gels and a Hoefer DALT electrophoresis system.

Gel Image Analysis

The gels were scanned using the Typhoon 9400 laser scanner at two different settings (Cy3, green laser 532 nm and 580 bp 30 filter; Cy5, red laser 633 nm and 670 bp 30 filter). Two images per gel were obtained (32 in total). One gel from group A was rejected and therefore omitted in further analyses. The scanned images were analyzed using Progenesis Workstation 2004 with the special cross-stain analysis module for analyzing multilabeled gels (Nonlinear Dynamics, Newcastle upon Tyne, UK). Spots were automatically detected and visually checked for undetected or incorrectly detected spots. The protein spots detected in each image were automatically linked between the two images per gel. All gels were matched to a digitized reference gel, containing all the protein spots. Per image the intensity levels were normalized by dividing the spot volume through the total intensity of all the spots in the image and multiplying it by the average of the total spot intensity of the gel images. Subsequently, the Cy3-labeled spot volumes were divided by the spot volume of the corresponding protein spot in the internal standard (Cy5) image. The obtained data set containing the expression ratios of each spot in each sample was analyzed by principal components discriminant analysis (PC-DA).

In-Gel Digestion and nLC-MS-MS

Preparative gels were run with 1 mg of protein and stained with RuBPS fluorescent staining. Spots were excised and sliced into small pieces. The gel pieces were washed twice with 100 mM ammonium bicarbonate and acetonitrile. Next, the pieces were dried in a vacuum centrifuge and digested overnight with 25 ng/μl trypsin (sequencing grade, Promega Benelux, Leiden, the Netherlands) in 100 mM ammonium bicarbonate and 2 mM dithiothreitol at 37°C. The peptide fragments were extracted twice with 5 μl water:acetonitrile:formic acid (5:14:1). After drying in a vacuum centrifuge, the lyophilized digest was dissolved in 25 μl of 4 M Urea buffered at pH 8.0 with 25 mM Tris. Nanobore chromatography was performed on an Ultimate nano LC system from LC Packings (Amsterdam, the Netherlands). Ten micro-liters of the peptide mixture was injected on a 300 μm ID X 0.5 mm Pepmap C₁₈ trap column (LC Packings) and washed at 30 μL/min for 10 minutes with 0.05% trifluoroacetic acid in water before the RP trap was switched online in back-flush mode to a 75 μm X 150 mm Pepmap C₁₈ nano LC column. Gradient elution of peptides was achieved at 300 nL per minute going from 95% mobile phase A (water:acetonitrile:formic acid, 97.9:2:0.08 v/v/v) and 5% mobile phase B (water:acetonitrile:formic acid, 19.9:80:0.1 v/v/v) to 45% B in 35 minutes and then to 60% B in 10 minutes. The nano LC column was coupled to a LCQ DECA ion trap MS (Thermo Electron, San Jose, CA, USA) via a nanoelectrospray interface from Proxeon (Odense, Denmark). Electrospray was performed by applying 1.3 kV to the electrospray pico tip (20 μm ID, 10 μm tip ID, distal coated from New Objective, Cambridge, MA, USA) via a Pt wire; ions were introduced in the mass spectrometer through a heated capillary kept at 180°C. The ion trap was operated in data-dependent mode, selecting top two ions for MS/MS scans at 35% collision energy units. Protein identification was performed by searching MS/MS spectra against the Swissprot database using Mascot search engine (www.matrix-science.com). Search parameters used were as follows: peptide mass tolerance 0.8 Da; MS/MS tolerance 1.2 Da; allowed missed cleavages 1; enzyme trypsin; taxonomy rattus; fixed modification, carbamidomethyl (C) variable modifications, oxidation (M), deamidation (NQ).

Statistics and Data Analysis

Body weights were analyzed by one-way analysis of covariance, followed by the two-sided Dunnett’s multiple comparison test. Organ weights, immunoglobulin levels, BAL biochemical parameters, and absolute cell numbers were determined by Anova-Dunnett’s test. Relative cell numbers in BAL were analyzed by Kruskall-Wallis nonparametric ANOVA followed by a Mann/Whitney U test. Differences were considered statistically significant if p < .05. Analyses were performed by the usage of Graphpad Prism (Version 3.0, San Diego, CA, USA).

Statistical analysis of the microarray data are described under “microarray analysis.” The statistics and multivariate data analysis of the proteomics data were as follows: The differences in spot ratios were analyzed by ANOVA followed by Student’s t-test, using Excel (Microsoft, Redmond, WA, USA) and PC-DA (PLS toolbox 3.0.2; 2003, Eigenvector Research in Matlab; Version 7.0, R14; Service Pack 1; 2004, The MathWorks).

Results

Clinical Signs, Body and Organ Weights, and Lung Function

Distinct clinical signs were absent. All animals gained approximately 27 grams between days 0 and 21; from day 21 to 22, animals of the sensitized/challenged (+/+) group showed a statistically significantly reduced body weight when compared with controls (Table 2). The weights of kidneys and liver were unaffected (data not shown). The mean absolute and relative weights of the left lung were increased by the challenge: the challenged groups (−/+ and +/+) differed statistically significantly from the unchallenged groups (−/− and +/−; Table 2). There was no additional effect of sensitization: the mean weights of the −/+ group did not differ significantly from that of the +/+ group.

TMA challenge in sensitized rats induced apneic periods—that is, irregularly lengthened pauses between varying number of breaths. Breathing frequency decreased to 50% of the original rate, starting within 1 or 2 minutes of the challenge, and returned to normal after the challenge but was increased at 24 hours after challenge (data not shown).

Serum and Lung Homogenate IgE Levels

Serum IgE levels were markedly increased in the sensitized (+/− and +/+) groups (4215.7 ± 616.6 and 2948.6 ± 532.4 μg/ml, respectively) when compared with the unsensitized (−/−and −/+) groups (346.4 ± 79.6 and 405.6 ± 116.8 ng/ml, respectively). IgE in homogenate of unlavaged left lung tissue was increased in the sensitized/challenged (+/+) group, compared with the unsensitized/challenged (−/+) group (Figure 2).

BAL

There was a slight reduction in total cell counts in the sensitized/unchallenged (+/−) group. The mean total cell number in BAL in the sensitized/challenged (+/+) group was about three times higher than that in the other three groups (Table 3). The increased total cell number in the +/+ group was largely due to an increase in the number of eosinophils and macrophages; the number of neutrophils was also increased. In the unsensitized/challenged (−/+) group, the total cell number tended to increase, and the increase was due to an increase in the number of macrophages. The increase in macrophages in the challenged (+/+ and −/+) groups was due to the challenge, and there was no additional effect of sensitization (the −/+ group did not statistically significantly differ from the +/+ group; Table 3).

Histopathology and Immunohistochemistry

The sensitized/challenged (+/+) rats exhibited a moderate laryngitis (Figure 1B), consisting of mucosal aggregates of predominantly macrophages with varying numbers of eosinophils and lymphocytes, with epithelial damage. No such changes were observed in rats of the three control groups. Histopathological changes were not observed in the nasal passages and trachea. The granulomatous inflammation in the lungs, specific for BN rats (Germann et al. 1998), did not differ between the groups but was slightly more pronounced in two naive (−/−) animals (animal nos. A1 and A3) and one sensitized/unchallenged (+/−) animal (animal no. B21).

Counts of immunohistochemically stained lung sections revealed an increase in IgE- and CD4-positive cells in the lung parenchyma of sensitized/challenged (+/+) rats (Table 4). The IgE staining was mostly membrane bound, and the cytology of these cells resembled that of lymphocytes (i.e., small, round cells). They were, therefore, considered to be predominantly B cells. In addition, there was an increase in IgE-positive cells around terminal bronchioles of sensitized/challenged (+/+) rats (Table 4; Figures 1C–E). Their staining was membrane bound or cytoplasmic, and they were considered to be B cells and plasma cells. The CD8, CD161 counts did not show differences between the groups (data not shown).

Microarrays

Data Exploration and Differentially Expressed Genes–Single Gene Comparison

Principal components analysis and hierarchical clustering showed a separation of the sensitized/challenged (+/+) group from the other three groups. Figure 3 depicts a part of the dendrogam (related to inflammation) after hierarchical clustering and illustrates the marked response in lung gene expression of the sensitized/challenged (+/+) group compared with the three control groups, despite the variability in granulomatous inflammation in the lungs, which was slightly more pronounced in animals A1, A3, and B21. Single-gene comparison between the groups (at a FDR threshold of 10%) resulted in the sensitized/challenged (+/+) group in a high number of differentially expressed genes above the number of differentially expressed genes by chance (Table 5). The genes expressed most distinctly in the sensitized/challenged group are listed in Table 6. These genes are associated with inflammation, including extracellular remodeling, as well as inflammation inhibition, but none of them is directly related to IgE. The effect of sensitization—that is, the number of genes expressed differently between the naive (−/−) and the dermally sensitized (+/−) group was low, if any, as was the number of genes expressed differently between the naive (−/−) and unsensitized/challenged (−/+) group. The last group reflects the possible toxic or irritating effects of the TMA challenge as well as the very early stage of sensitization by inhalation. Genes in this group were mostly down-regulated, but their differential expression was low. The mean fold changes were highest for the chemokines Ccl4, Ccl17, and Ccl20.

Gene Group Analysis (T-Profiler)

Seven GO groups/pathways of genes were expressed highly significantly different between the +/+ group and the other three groups (Table 7). The GO gene groups/pathways were “chemokine activity,” “chemotaxis,” “inflammatory response,” “extracellular space,” “extracellular region,” “cytokine-cytokine receptor interaction,” and “Toll-like receptor signaling pathway” (Figure 4). The GO gene groups/pathways were all up-regulated and had a t value of about 6.00 or more and e values of < .0001 in the +/+ group. In the +/− and −/+ group, most of these GO gene groups/pathways tended to be down-regulated; the only GO gene pathway that was significantly up-regulated was the Jak-STAT signaling pathway in the −/+ group.

Blood Proteomics

In total, 360 different spots were detected on each internal standard gel. The score plot showed a distinct protein pattern between the groups, with the −/− (A) and +/− (B) groups placed close to each other in the score plot. The +/+ (D) group is placed at a greater distance. Sixteen proteins were differentially expressed between the +/+ group and the three control groups (ANOVA followed by Student’s t-test) and were selected for identification. Figure 5 shows the differences between the expression levels of the sixteen proteins, as represented by the mean spot-intensity ratios on the DIGE gels. All spots could be identified, demonstrating six proteins (Table 8). Spots 660, 662, 665, 697, and 705 were found to consist of two proteins, Serine-like protease inhibitor 6 and Alpha-1 antiproteinase. The individual contribution of the two proteins to the spot intensity was not investigated. Many spots were spot trains, which indicates posttranslational modifications like phosphorylation or glycosylation (Swissprot database). This could also explain the difference between experimental and theoretical molecular mass.

Discussion

To contribute to the hazard identification of LMW respiratory allergens, respiratory allergy induced by TMA in the Th2-prone BN rat was characterized by whole genome analysis (microarrays) of lung tissue and proteomics in blood. In confirmation with previous studies, TMA-induced apneas and allergic laryngitis, increased inflammatory cell numbers in BAL, altered breathing frequency, and elevated IgE levels in serum and lung (Arts et al. 1998; Arts et al. 2004; Valstar et al. 2006).

The identification of inflammatory changes in the bronchial or bronchiolar mucosa and lung parenchyma of the allergic (sensitized/challenged or +/+) BN rats was hampered by the variable background inflammation in the lungs of BN rats (see Results section; Arts et al. 1998; Germann et al. 1998). This is reflected by the interanimal variability in expression of a number of inflammatory-associated genes in the lung (Figure 3). Nevertheless, a single TMA challenge in TMA-sensitized animals (+/+) led to a strong differential expression of single genes and GO gene groups/pathways in lung tissue (Tables 6–8).

There was a considerable consistency between the genes significantly up-regulated in the TMA-sensitized and TMA-challenged BN rats in the present study and data found in asthmatic humans: almost all cytokines, for which the genes were up-regulated in BN rats (Table 6), have been found in increased amounts in sputum, BAL, and exhaled air of asthmatic patients (reviewed by Bloemen et al. 2007). Moreover, the Toll-like receptor pathway, activated in the sensitized/challenged (+/+) BN rats, is considered to be a pathway that is commonly activated in a number of inflammatory conditions, including asthma, in man (Bauer, Hangel, and Yu 2007). However, of the list of about 100 genes associated with asthma in man (Ober and Hoffjan 2006), only Timp1 and Ccl2/MCP1 were up-regulated sufficiently in the BN rats to be included in Table 6.

Interestingly, a single challenge by TMA in the sensitized rats was sufficient to activate genes related to lung remodeling, which is featured by smooth muscle hypertrophy, transformation of fibroblasts into myofibroblasts, angiogenesis, and deposition of subepithelial collagen (Holgate et al. 2000). This early onset of lung remodeling is in agreement with the observation in man that remodeling is a consistent component of early asthma, rather than being a secondary phenomenon, developing late in the disease. Remodeling changes might even be of higher diagnostic value than eosinophilic infiltration (Davies et al. 2003). The most strongly up-regulated genes in the BN rat, involved in remodeling, were for arginase and glial fibrillary acid protein, which may lead to local synthesis of collagen (Zimmermann et al. 2003) and is linked to transformation of fibroblasts into myofibroblasts (Zhao and Burt 2007), respectively. In the present study there was indirect evidence for involvement of MMP9, namely, strong up-regulation of IL-1beta and IL-6, which are able to stimulate MMP-9 secretion (Lin et al. 2007), and the presence of high numbers of eosinophils in BAL, which are a source of MMP9 (Kay, Phipps, and Robinson 2004; Trivedi and Lloyd 2007). MMP9 selectively degrades extracellular matrix components (ECM) like type IV and V collagen (Vignola et al. 2003). An increase in MMP9 has been demonstrated in patients with acute severe asthma or following allergen challenge (Belleguic et al. 2002), supporting the idea that ECM and basement membrane degeneration by MMP’s and other proteinases may occur in the early phase of asthma. In addition, the GO pathways “extracellular space” and “extracellular region” were highly significantly expressed in the sensitized/challenged (+/+) rats, compared with the three control groups. The contact allergen dinitrochlorobenzene, tested according to an identical protocol, induced expression changes in a few allergy-associated genes for chemokines but did not up-regulate any of the genes for lung remodeling (Kuper et al. 2008). Thus, early lung remodeling may be useful in identification of compounds, which can induce asthma.

The single challenge with TMA in sensitized BN rats not only induced lung inflammation and lung remodeling but also activated mechanisms that down-regulate these processes. The increased IMP-1 production may have accompanied the increase in MMP to inhibit ECM degradation. Genes for serine protease inhibitors, especially 2c, were up-regulated as well. Interestingly, 24 hours after challenge, serine protease inhibitor 3 and 6 proteins were markedly increased in peripheral blood of the TMA-sensitized and TMA-challenged rats. Serine-like pro-tease inhibitor 6 has also been found in the BAL of BN rats, 24 hours after intratracheal exposure to lipopolysaccharide (Signor et al. 2004). The serine-like protease inhibitors are considered to be involved in tissue repair and directed to down-regulate inflammation (Ohkubo et al. 1991). It should be kept in mind as well that respiratory allergy-associated molecules may have dual actions–that is, stimulation but also inhibition of allergic manifestations. For example, IL-1beta can induce NF-kappaB, which, under certain circumstances, inhibits myofibroblast phenotype (Kuang et al. 2007), whereas IL-6 can act as an anti-inflammatory agent. Together, it supports the idea that sustained or chronic allergic diseases result not only from excessive or deranged activation of lung cells but also from deficiencies at the level of suppressing molecules (Ravetch and Lanier 2000).

The development of respiratory allergy, especially when induced by HMW allergens, is considered to be regulated predominantly by Th2-polarized immune reactions in which IgE plays a dominant role (reviewed by Boverhof et al. 2008). With TMA-induced respiratory allergy in BN rats, several parameters pointed to such a Th2-polarized mechanism. First, the increases in eosinophils in BAL, CD4+ lymphocytes in lung parenchyma, IgE+ cells in lung parenchyma and around terminal bronchioles, and in total IgE in serum and lung (and BAL; Valstar et al. 2006) are associated with Th2-mediated immune responses. Second, the strongly up-regulated gene for arginase-1 is thought to be the result of activation by the Th2-related cytokines Il-4 and Il-13 (Zimmermann et al. 2003), and it indicates that macrophages are activated via the alternative pathway (Mantovani et al. 2004). Macrophages, activated via this pathway, are involved in Th2-mediated allergy and are associated with changes in gene-expression profiles for the chemokines Ccl2, Ccl3, Ccl4, Ccl17, and Ccl22, which are also strongly up-regulated in the TMA-sensitized and TMA-challenged BN rats (Table 6). Such changes are not associated with classical macrophage activation. Classical activation of macrophages probably occurs in the unsensitized/challenged (−/+) BN rats. These animals also exhibited increased numbers of macrophages in BAL, albeit slightly less than the sensitized/challenged animals, but the above-mentioned genes were down-regulated, if at all. Arginase-1 is also linked with STAT6-dependent pathways, which are considered to be crucial in the induction of asthma (Zimmerman et al. 2004).

Although the GO groups/pathways for chemokines were highly significant in the sensitized/challenged rats when compared with the three control groups, no differential expression was observed of genes for those chemokines that are being used in cytokine profiling tests to try to distinguish typical respiratory from typical contact allergens (Dearman, Skinner, and Humphreys 2003) or those directly related to the IgE-cascade of events, including the Fc receptors for IgE that are determinants of the allergic reaction (Kraft and Novak 2006). This is probably related to the time of the tissue sampling—that is, 24 hours after challenge.

Whole genome analysis of lung tissue from TMA-sensitized and TMA-challenged BN rats provided a host of information, which at present cannot be put entirely in proper perspective with respect to chemical respiratory allergy. Nevertheless, it is clear that a single TMA challenge provokes a predominantly Th2-mediated gene profile in the lungs of TMA-sensitized BN rats. The gene-expression profile in lung tissue of the sensitized/challenged rats, 24 hours after challenge, is in agreement with the pulmonary inflammation—that is, increases in eosinophils, CD4+ lymphocytes, and activation of macrophages via the alternative pathway. In addition, the gene-expression profile points to early activation of lung remodeling. At the same time, inflammation downsizing mechanisms were activated, which is reflected by proteomics in peripheral blood. It is concluded that parameters for lung remodeling may not only be of diagnostic value for early asthma in humans but also be a useful tool for the hazard identification of respiratory allergens in test models.