ABSTRACT: Exposure to ambient particulate matter (PM) significantly increases cardiovascular morbidity and mortality in the general population. We hypothesized that some components of PM can affect the gene expression patterns in the hearts of rats exposed for 3 months to filtered air (FA), coarse (CP; 2.5 < dp < 10 μm), fine (FP; dp ≤ 2.5 μm) or ultrafine (UFP; dp ≤ 0.18 μm) components of PM. The median diameters of CP, FP, and UFP were 3 μm, 0.7 μm and 0.07 μm, respectively. Exposures (n = 8 per group) were performed using a particle concentrator system in Riverside, California, an area with high ambient levels of photochemically derived gaseous and particulate pollutants. At the end of the exposure, hearts were subjected to gene expression profiling by using Illumina RatRef-12 bead chips and levels of malonaldehyde (MDA), a biomarker of oxidative stress, were measured. Applying fold ratio >1.5 (for both up- and downregulated genes), we found three genes in the CP and nine genes in the UFP groups with significantly changed expression, compared with FA. No significant changes in gene expression patterns were observed in the FP group. In the UFP group, thioredoxin-interacting protein (Txnip), a negative regulator of an antioxidant enzyme thioredoxin, and cytochrome P450 (Cyp2e1), an enzyme involved in the metabolism of foreign substances, demonstrated significant up-regulation (fold ratios 1.79 and 1.57, respectively, with false discovery rate, FDR < 0.05). In the CP group, there was also a trend towards increased Txnip expression (fold ratio 1.43, FDR > 0.05) and significant increase in the Cyp2e1 expression (fold ratio 1.79, FDR < 0.05). Changes in the Txnip and Cyp2e1 expression showed statistically significant positive correlation to each other (p < 0.0009) and were confirmed by real-time PCR. In addition, Txnip and Cyp2e1 expression demonstrated statistically significant moderate correlation with the levels of MDA in the heart. Up-regulation of both Cyp2e1 and Txnip are observed in hearts of patients with certain cardiac diseases. Therefore, chronic exposure to CP and UFP directly affects expression of disease-relevant genes in the myocardium. Analysis was performed on 32 microarrays (eight microarrays in each group). After five chips with poor hybridization signals were removed, the final analysis included 27 microarrays (seven miocroarrays in the FA group, six microarrays in the CP group, six microarrays in the FP group and eight microarrays in the UFP group). The Bioconductor Lumi package was used to preprocess gene chip data with default settings (Du et al. 2007; Lin et al. 2008). Probes with all samples “Absent” (lower or around background levels) were removed from further analysis. After filtering the signals, 11,657 probes on the microarray (from a total of 22,226 probes) were kept for further analysis. To identify differentially expressed genes, we applied routines implemented in the Limma package (Linear model for microarray data) to fit linear models to normalized expression values. The variance used in the t-score calculation was corrected by an empirical Bayesian method (Smyth 2004). Genes with significantly changed expression (fold ratio >1.5 for both up- and downregulated genes) were selected. The gene list was further filtered by FDR (false discovery rate) of <0.05 and <0.25, which indicate that no more than 5% and 25% of all listed genes are false positives (an assumption that is not offered by a traditional p value-based analysis).
