Post-transcriptional air pollution oxidation to the cholesterol biosynthesis pathway promotes pulmonary stress phenotypes

The impact of environmentally-induced chemical changes in RNA has been fairly unexplored. Air pollution induces oxidative modifications such as 8-oxo-7,8-dihydroguanine (8-oxoG) in RNAs of lung cells, which could be associated with premature lung dysfunction. We develop a method for 8-oxoG profiling using immunocapturing and RNA sequencing. We find 42 oxidized transcripts in bronchial epithelial BEAS-2B cells exposed to two air pollution mixtures that recreate urban atmospheres. We show that the FDFT1 transcript in the cholesterol biosynthesis pathway is susceptible to air pollution-induced oxidation. This process leads to decreased transcript and protein expression of FDFT1, and reduced cholesterol synthesis in cells exposed to air pollution. Knockdown of FDFT1 replicates alterations seen in air pollution exposure such as transformed cell size and suppressed cytoskeleton organization. Our results argue of a possible novel biomarker and of an unseen mechanism by which air pollution selectively modifies key metabolic-related transcripts facilitating cell phenotypes in bronchial dysfunction.

LDH levels after exposure of BEAS-2B cells to the air pollution mixture (lower oxidative mixture in Table 1) for 1.5 h. Activity of LDH was measured by a colorimetric assay in the cell culture media (n = 3 independent experiments). Error bars are expressed as one standard deviation (s.d.), n.s. refers to no significance difference was determined by t-test (one-tailed homoscedastic) with significance established as p-value < 0.05. a b Supplementary Fig. 4 a Venn diagram shows the number of transcripts upregulated and downregulated in BEAS-2B cells following low-level exposure, and the overlap with oxidized transcripts after exposure. b Top ranked KEGG pathways enriched from the differentially expressed transcripts (upregulated and downregulated). We conducted transcriptomics analysis of the mRNAs (e.g., using a fraction of the input mRNA pool) to compare expression changes under exposed and control cells. This analysis shows differential expression of 878 mRNA transcripts with an adjusted p-value < 0.05. A lower p-value cutoff was used given the low variance in the transcriptome data as compared to the 8-oxoG pulldowns.
Of these, 336 transcripts exhibit increased expression with a fold change > 2, and 545 exhibit decreased expression with fold change < 0.5 (Supplementary Data 1 and 2). Terms ranked by the combined score in Enrichr. Genes associated to each pathway are presented in Supplementary Data 9 and 10.

Supplementary Fig. 5
Exposure of BEAS-2B cells to the high-level air pollution mixture (Table 1) for 1.5 h. Free 8-oxoG nucleosides from total RNA were quantified shortly after exposure (at t = 1.5 h) by ELISA (n = 3 independent experiments). Statistical difference was computed using t-test analysis (one-tailed homoscedastic) and significance is denoted as *** for pvalue < 0.0005. Error bars are expressed as one standard deviation (s.d.).

Supplementary Fig. 6
Detection of 8-oxoG in ex vivo exposure of total RNA to the high-level air pollution mixture. a Schematic depicting the direct exposure of RNA to air pollution. Cells were grown until reaching confluence. Cells were lysed with Trizol and then RNA was extracted and purify using spin column-based purification. 8 µg of total RNA was resuspended in 500 µl of TE buffer (pH 8.0) and exposed to air pollution for 90 min (using the higher concentrations of the VOC+O3 precursors in Table 1). RNA was purified and 8-oxoG was measured with ELISA. b Quantification of free 8-oxoG nucleosides from total RNA directly exposed to air pollution using ELISA (n = 3 independent experiments). Statistical difference was computed by t-test analysis and significance is denoted as ** for p-value < 0.001. Error bars are expressed as one standard deviation (s.d.).

Supplementary Fig. 7
Assessment of the anti-8-oxoG antibody (clone 15A3) demonstrating high specificity of the antibody. a Dot blotting of different RNA oligonucleotides containing common methylated and oxidized RNA modifications as described in the label. Decreasing amounts (indicated on top of the blot) of the oligos were spotted onto the membrane, UV crosslinked and probed with anti-8-oxoG antibodies. b Quantification of the signal detected for the 8-oxoG 25-mer with one modification (square) and the 6x 8-oxoG random 24-mer containing six modifications (circle). Quantification of the signal was conducted in CLIQS (TotalLab), with background subtracted. The signal ratio of 6x 8-oxoG random 24-mer/8-oxoG 25-mer of ~6 is proportional to the ratio of modifications in each oligomer at 10, 5 and 2.5 pmol of oligomers.

Supplementary Fig. 8
Log2-FC plot representing selected 20% of all detected 8-oxoG enriched transcript (5493 out of 27,269 transcripts) in exposed cells. Labeled transcripts refer to the ones in the subset of the identified 707 oxidized transcripts by air pollution. This plot demonstrates that after applying the comparisons in Fig. 2D, the minimum log2-FC of 6.7 provides a stringent threshold cutoff for removal of background noise from artifactual oxidation. a b Supplementary Fig. 9 a Fraction of oxidized transcripts out of all transcripts (within one expression bin) in BEAS-2B cells exposed to low-level air pollution (Table 1). b Volcano plot shows differential expression by comparing the input mRNA pool between air pollution vs clean air conditions at the low-level exposure (Table 1). Data points in blue are 8-oxoG enriched and data points in red are 8-oxoG enriched and differentially downregulated (Supplementary Data 3). Significant expression was established with a fold change < 0.5 or > 2 with a statistical confidence of α = 0.1. (C) Volcano plot shows differential expression by comparing the input mRNA pool between air pollution vs clean air conditions at the low-level exposure. Significant expression was established with a fold change < 0.5 or > 2 with a statistical confidence of α = 0.05. Supplementary Fig. 11 a Fold change changes for differential expression and 8-oxoG IP from BEAS-2B cells exposed at the high-level mixture as given by DESeq2. b Validation of the observed trends for FDFT1 was performed by qPCR quantification of 8-oxoG IP and differential expression. Importantly, the abundance patterns for the FDFT1-215 transcript identified in all three groups in the 8-oxoG analysis are replicated well by qPCR. Error bars are expressed as one standard deviation (s.d.). Supplementary Fig. S12 a Schematic of the reverse transcription truncation assay to validate the oxidation of the FDFT1-215 transcript via an antibody-free method. Chemical labeling of 8-oxoG with K2IrBr6 generated a covalent bond with an amine-terminated biotin with a polyethylene glycol linker (HN-R). This reaction yields a bulky moiety in 8-oxoG but not in G, which may cause reverse transcription stops. After reverse transcription of the labeled transcripts, PCR using primers near the 5' end (proximal) and the 3'end (distal) results in accumulation of proximal products compared with the distribution of distal products. The decrease in the ratio of distal/proximal PCR products represents the relative level of 8-oxoG oxidation as compared to the control. b To validate the reverse transcription truncation assay with 8-oxoG chemical labeling, normal RNA extracted from BEAS-2B cells was used for a proof of concept assay. Here, we treated a fraction of the purified RNA with the Fenton's reagents to induce RNA oxidation. Normal RNA and Fenton's oxidized RNA was then chemically labeled with the biotin-terminated amine. After biotin labeling, the samples were subjected to PCR with proximal and distal primers for FDFT1-215. GAPDH and PPIB amplifications were used as loading control and negative control respectively (these transcripts were selected because they were unaffected by exposure according to our 8-oxoG RIP-seq analysis). Results indicate a decrease in the distal/proximal ratio in the Fenton's oxidized RNA compared to normal RNA when both products are biotinylated. Interestingly, non-biotinylated RNA does not stop reverse transcription as demonstrated by the constant ratio between Fenton's oxidized and normal RNA. Error bars are expressed as one standard deviation (s.d.). c Degradation assay of total RNA treated with the chemical labeling of 8-oxoG in 1% agarose gel electrophoresis. This assay demonstrates that chemical labeling at lower temperatures prevent degradation of the RNA as seen by the presence of intact 28S and 18S rRNA in Fenton's reaction oxidized RNA and normal RNA treated at room temperature as compared with normal RNA treated at 75 °C. Normal RNA non-biotinylated (lane 4) was as positive control to depict intact RNA.

Supplementary Fig. 13
Cellular stress analysis by lactate dehydrogenase (LDH) release. We quantified LDH in basolateral side of BEAS-2B cells after 24-hr treatment with siRNAs by a colorimetric assay of LDH activity in the cell culture media. No statistical significance was found between conditions using t-test analysis with p-value < 0.05. Error bars are expressed as one standard deviation (s.d.).