Methylene diphenyl diisocyanate (MDI) is hazardous and a well-recognized immune-sensitizing chemical. Nitrile gloves, protective clothing, and goggles should be used for personal protection.

Established methods for generating and characterizing MDI-GSH reaction products were followed.³²⁻³⁴ Reduced glutathione, GSH (CAS No. 70−18−8), and 4,4′-methylenebis(phenyl isocyanate) (CAS No. 101−68−8) were from Sigma-Aldrich (St. Louis, MO). Two hundred 50 microliters of 10% (w/v) MDI in acetone from JT Baker (Phillipsburg, NJ) was added dropwise with stirring to 25 mL of 10 mM GSH in 200 mM sodium phosphate, pH 7.4. The reaction mixture was rotated end-overend for 2 h at 37 °C, centrifuged at 10,000g, 0.2 μm filtered, cooled to 4 °C and used within 2 h of preparation. Control solutions were prepared by mixing MDI and GSH solutions after individual separate reactions, during which time MDI hydrolyzes and forms low molecular weight ureas.^35,36 An aliquot of each sample was analyzed by reverse phase UV-LC-MS/MS in positive ion mode to verify bis(GSH)-MDI was the dominate reaction product based on the A₂₁₀ and A₂₄₅ spectra, and retention times of the 865.25 and 433.13 m/z ions reflecting singly and doubly charged species, as previously described.^32,34

Balb/C mice 8 weeks of age were housed under pathogen-free conditions and fed ad libitum with automated water supply and 12 h day/night light cycles. Immunologic sensitization to MDI, which is required for the development of asthma pathology following respiratory tract exposure, was accomplished via repeated skin exposure, as previously described.^32,37 Briefly, a region on the back was shaved 24 h prior to application of 50 μL of 1% (w/v) MDI in acetone on day 0 and again on day 5. For respiratory tract exposure, 50 μL of GSH-MDI reaction products or control solution (GSH mixed with MDI after separate reactions) was delivered intranasally under isoflurane sedation as previously described.^32,34,37,38 Mice received respiratory tract exposure on days 15, 16, 19, and 20. The dose and timing of MDI exposure were chosen to reflect occupational exposure and are consistent with recent field studies.³⁹ The 1% MDI (w/v) skin exposure dose used for inducing systemic immune sensitization is roughly equivalent to the final MDI concentration in polyol/diisocyanate mixtures used to make polyurethane foam.¹ Each intranasal exposure delivered the potential equivalent of 10 μg of MDI, approximately 50-fold below the permissible exposure limit established by the U.S. Occupational Health and Safety Administration as previously calculated.^32,40 All animal studies followed ethical guidelines established in the “Guide for the Care and Use of Laboratory Animals” prepared by the Institute of Laboratory Animal Resources, National Research Council, and published by the National Academy Press⁴¹ and were approved by Yale University’s Institutional Animal Care and Use Committee.

BAL (3 × 0.8 mL) and lung tissue samples were obtained from N = 6 mice/group (control, MDI exposed nonsensitized and MDI sensitized and exposed mice) 48 h following the last respiratory tract exposure, as previously described.^32,37,38 BAL cells were pelleted by centrifugation at 800g, treated with RBC lysis buffer, washed, and resuspended in PBS for cytospin and cell counting. Total BAL cell numbers were calculated using a hemacytometer, and differential counts were performed on 200 cytospun cells that had been stained with diff quick (Polysciences Inc.; Warrington, PA). Lung tissue was obtained following PBS perfusion and immediately homogenized in RLT buffer from Qiagen (Germantown, MD). RNA was purified using an RNeasy kit from Qiagen, which included on-column digestion with RNase-free DNase.

Processing of RNA and microarrays were performed by the Yale Center for Genome Analysis.^42,43 Lung tissue RNA quantity and quality was assessed respectively by microvolume spectrophotometry on the Nanodrop 2000 (Thermo Scientific) and by on-chip capillary electrophoresis on a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). Absorbance ratio at 260 and 280 nm was ≥1.9 and the RNA integrity number was >9 for all samples. GeneChip WT Plus Reagent Kit (Affymetrix, Santa Clara, CA) was used for the preparation of samples and generation of ss-cDNA for hybridization, using 250 ng of total RNA as template. All the reactions and hybridizations were carried out according to the manufacturer’s protocol.⁴⁴ Affymetrix GeneChip Mouse Clariom-S arrays were washed using the GeneChip Fluidics Station 450 and scanned with the GeneChip Scanner 3000. Experiments were repeated with N = 6 age matched animals per group, with all exposures and tissue collection performed along the same timeline.

Each data sets’ quality was confirmed based on hybridization, labeling controls, internal control genes (house-keeping controls), and global array metrics. Raw probe measurements (.CEL files) were background-adjusted, normalized, and converted into summarized expression values using the signal space transformation robust multichip analysis (SST-RMA) algorithm in the Transcriptome Analysis Console (TAC) software (version 4.0.2, ThermoFisher Scientific, Waltham, MA, USA), which incorporates the limma differential expression portion of the Bioconductor package as well as modified versions of the ComBat function from its Surrogate Variable Analysis. Analysis of variance (ANOVA) was used to fit a linear model to each probeset independently of the others. An eBayes analysis was applied to correct variance of the ANOVA analysis using information from all the probesets to yield an improved estimate for the variance. Differentially expressed genes were calculated based on default settings; >2-fold difference in average fluorescence with ANOVA p value <0.05. False discovery rate p values were further calculated using the Benjamini and Hochberg method.⁴⁵ Details on data processing modules, algorithms, and statistics underlying TAC software differential gene expression analysis are available on line.⁴⁶

IPA software (Ingenuity Systems; Qiagen China Co., Ltd.) was also used for bioinformatics analysis, specifically canonical pathway analysis and identification of predicted upstream regulators. IPA uses a network generation algorithm to segment the network map between molecules into multiple networks and assign scores for each network.^47,48 The score is generated based on hypergeometric distribution, where the negative logarithm of the significance level is obtained by Fisher’s exact test at the right tail. For canonical pathway analysis, the −log (P value) > 2 was taken as threshold, the Z score >2 was defined as the threshold of significant activation, while a Z score < − 2 was defined as the threshold of significant inhibition. For upstream regulators, the P value of overlap <0.05 was set as the threshold. The algorithm used for calculating the Z scores and P values of overlap has been described previously.⁴⁹

We compared the present gene expression data on LMW diisocyanate chemically induced pathology with previously published data on the prototypic asthma model triggered by the HMW allergen, ovalbumin. We queried the NCBI GEO database using the keywords “ovalbumin”, “asthma”, “lung”, “tissue”, and “Balb”, which identified 47 entries. Six of these studies provided data of lung mRNA 48 h post-exposure. Of these, the study entitled “Effects of the OVA challenge in mouse model of asthma” (GSO Series GSE41665) best matched the present study design (48 h post-exposure time point, Balb/C mouse strain, paired control exposures, mRNA, similar timing of sensitization) and was used to evaluate potential differences in lung gene expression associated with asthma pathology induced by LMW vs HMW asthmagens. For analysis, all N = 8 each (ovalbumin exposed, control). CEL files from the 48 h time point of GSO Series GSE41665 were imported and processed using TAC to calculate the fold change of individual genes in ovalbumin exposed mice (which developed asthmatic pathology) vs controls. Two of the total (N = 16). CEL files from GSO Series GSE41665 that did not pass all quality control assessments by TAC software (3′ labeling controls outside threshold value) were excluded from analysis, leaving N = 7 samples each for ovalbumin exposed and controls. Venn diagrams, hierarchical cluster analysis, and bar graphs were used to identify and visualize individual genes with the greatest relative change in expression induced by pathologic diisocyanate vs prototypic HMW allergen exposure.

DNase treated total lung RNA from N = 8 animals/group was used to prepare cDNA with an iScript cDNA synthesis kit. The cDNA samples were amplified with iTaq universal SYBR green PCR supermix on a BioRad C1000 thermocycler interfaced with a CFX96 real-time PCR detection system with CFX Maestro Software, according to the manufacturer’s recommendations (Bio-Rad; Hercules, CA). Changes in gene expression were calculated based on the ΔΔCt approach, with beta-actin (ACTB) as a reference gene.⁵⁰ PCR primer pairs were Bio-Rad’s PrimePCR SYBR Green Assay for ZBTB16 (qMmuCID0022995), IRF7 (qMmuCED0040274), IIGP1 (qMmuCED0038266), IFI44 (qMmuCID0011454), IFIT3 (qMmuCID0041372), CLCA1 aka murine CLCA3 (qMmuCID0009928), and ACTB (qMmuCED0027505).

For each group of mice, the mean and standard deviation of the data were initially entered, calculated in an Excel spreadsheet, and then exported to GraphPad Prism for graphing and statistical analyses. Significant differences of data that were not normally distributed were determined using the nonparametric Kruskal−Wallis test with Dunns’s correction for multiple comparisons. Data that were normally distributed were evaluated by Welch ANOVA with Dunnet T3 correction for multiple comparisons. Detailed descriptions of statistical calculations for TAC and IPA programs are available online as described above.^45,51,52

We performed paired analysis of cellular airway inflammation and lung tissue gene expression in samples from nonsensitized and MDI-immune sensitized mice exposed to MDI via the respiratory tract and control mice. Following exposure, only MDI-sensitized mice contained significantly (P < 0.05) increased percentages of eosinophils (Figure 1) in BAL samples. However, lung tissue from both nonsensitized and MDI-immune sensitized mice contained large numbers of differentially (>2-fold) expressed gene transcripts compared to controls. In nonsensitized mice, 534 gene transcripts were differentially expressed (254 increased and 280 decreased), while in MDI-immune sensitized mice, 218 gene transcripts were differentially expressed (181 increased and 37 decreased), as depicted in Venn diagram analysis (Figure 2) and listed in the Supporting Information (Tables S1 and S2).

A core set of 53 genes was differentially expressed in all MDI exposed vs control mice, with up to 10-fold greater changes in immune-sensitized vs nonsensitized hosts (Table 1 and Supporting Information Table S3 and Figure S1). Of these, 51 were increased and 2 were decreased. Proteins corresponding to several of the differentially expressed lung gene transcripts were elevated in prior proteomic studies of airway fluid from MDI immune sensitized mice similarly exposed to MDI, providing orthogonal evidence supporting the present data. The gene transcripts most increased by MDI exposure have been linked to human asthma functionally and/or genetically.⁵³⁻⁵⁷

In total, 480 gene transcripts were differentially expressed (±2-fold, p < 0.05 vs controls) in nonsensitized, but not MDI immune-sensitized, mice following respiratory tract MDI exposure; 203 increased and 278 decreased (Figure 2). The top differentially expressed gene transcripts are shown in a heat map (Figure 3) and a complete list and volcano plots are included in the Supporting Information (Table S4, Figure S2). The differentially expressed transcripts are enriched in Type 1 interferon related genes, with the largest relative increase in interferon regulatory factor 7 (IRF7), a master regulator of type-I interferon-dependent immune responses.⁵⁸ The gene with the largest relative decrease in expression encodes the antibody joining (J) chain expressed exclusively by B-cells, which play a major role in HMW allergen pathogenesis via secretion of antigen-specific IgE.⁵⁹

In total, 164 genes were differentially expressed (±2-fold, p < 0.05 vs controls) in MDI immune-sensitized, but not nonsensitized, mice following respiratory tract MDI exposure; 130 increased and 35 decreased. The top differentially expressed genes are shown in Figure 3 (with a complete list in Supporting Information Table S5 and volcano plot Figure S2). The increased gene transcripts are dominated by molecules crucial to inflammatory responses, especially chemotaxis. Three of the four gene transcripts with the largest relative increases encode Chitinase-like molecules, which are secreted by alternatively activated macrophages among other cell types and attract eosinophils.⁶⁰⁻⁶² The most downregulated gene transcript encodes nuclear receptor 4A1 (NR4A1) aka Nur77, known for its role in restricting metabolic activation of autoreactive T-cells and increasing asthma pathology when absent.^63,64

Thus, differential expression of 646 total genes, 333 increased and 313 decreased (<3% of the genome), is associated with differential pathologic responses to respiratory tract MDI exposure in mice, depending on prior exposure and systemic immune sensitization.

IPA comparison of MDI exposed nonsensitized and immune-sensitized mice vs controls identified overlapping as well as unique features in the top predicted upstream regulators and canonical pathways (Tables 2 and 3 and Supporting Information Table S6). LPS and IFN-γ were the most significantly predicted upstream regulators of the MDI response in all exposed animals but were followed differentially by IFN-α, STAT1, and TLR3, in nonsensitized mice, vs IL-13, IL-4, and TCL1A in MDI-sensitized mice, consistent with observed pathology. The most significantly affected canonical pathway for all MDI exposed mice vs controls was “granulocyte adhesion and diapedesis”. However, marked differences between MDI immune sensitized vs nonsensitized animals were observed in other significantly affected canonical pathways, several driven by a combination of complement, CD80, and HLA molecules (see Supporting Information Table S6). Compared with controls, murine MHC class II H2-A gene transcripts were exclusively increased >2-fold in exposed MDI sensitized mice, while nonclassical MHC class 1b molecules (H2-T23, H2-M3) were exclusively increased >2-fold in exposed nonsensitized mice, consistent with differing antigenic underpinnings of the exposure response.

TAC analysis, which curates genes based on Wiki pathways, predicted the involvement of slightly different biological pathways in response to MDI exposure (Table 4). The Wiki pathway most significantly affected in nonsensitized mice vs controls is that expected to result from “Deficiency in RNA Editing by Adenosine Deaminase Acting Against RNA-1 (ADAR1)”.⁶⁵ In contrast, “Chemokine Signaling” is the pathway most significantly affected in MDI immune-sensitized mice.

Together, IPA and TAC analysis suggests MDI respiratory tract exposure induces marked changes in immune-related genes in both nonsensitized and immunologically sensitized hosts. Gene transcripts involved in chemokine signaling are altered in all MDI exposed mice (vs controls), but to a larger extent in immune sensitized vs nonsensitized hosts. Distinct differences in gene expression reflecting ADAR-1 deficiency occur only in nonsensitized mice while complement, dendritic, and B cell developmental pathways are more significantly impacted by exposure in immune sensitized hosts. Differences in MHC gene expression suggest further potential differences in underlying antigenic stimuli and responsive cell clonotypes in exposed nonsensitized vs sensitized mice. Based on the present gene expression and bioinformatic analysis, Figure 4 presents a hypothetical mechanism by which MDI might affect airway inflammation by inhibiting an active innate immune process for differentiating self-vs nonself.

To validate microarray data, beyond prior limited proteomic studies,³² real-time PCR was performed for a panel of selected genes (CLCA1, ZBTB16, IRF7, IFIT3, IIGP1, IFI44) with tissue from additional MDI exposed animals and controls. Quantitation of gene expression for animals in each group demonstrated relative changes in nonsensitized and immune-sensitized mice consistent with those measured by microarray (Supporting Information Figure S3). CLCA1 expression was significantly elevated vs controls, upon MDI exposure of nonsensitized and MDI-sensitized mice, although levels were >10 -fold higher in sensitized mice. In contrast, ZBTB16 transcript levels were significantly increased only in MDI exposed-immune sensitized mice, while IRF-7, IFIT3, IIGP1, and IFI44 were significantly increased in MDI exposed nonsensitized but not MDI sensitized mice.