Case study: Targeted RNA-sequencing of aged formalin-fixed paraffin-embedded samples for understanding chemical mode of action

Toxicol Rep. 2022 Apr 18:9:883-894. doi: 10.1016/j.toxrep.2022.04.012. eCollection 2022.

Abstract

Formalin-fixed paraffin-embedded (FFPE) samples are the only remaining biological archive for many toxicological and clinical studies, yet their use in genomics has been limited due to nucleic acid damage from formalin fixation. Older FFPE samples with highly degraded RNA pose a particularly difficult technical challenge. Probe-based targeted sequencing technologies show promise in addressing this issue but have not been directly compared to standard whole-genome RNA-Sequencing (RNA-Seq) methods. In this study, we evaluated dose-dependent transcriptional changes from paired frozen (FROZ) and FFPE liver samples stored for over 20 years using targeted resequencing (TempO-Seq) and whole-genome RNA-Seq methods. Samples were originally collected from male mice exposed to a reference chemical (dichloroacetic acid, DCA) at 0, 198, 313, and 427 mg/kg-day (n = 6/dose) by drinking water for 6 days. TempO-Seq showed high overlap in differentially expressed genes (DEGs) between matched FFPE and FROZ samples and high concordance in fold-change values across the two highest dose levels of DCA vs. control (R2 ≥ 0.94). Similarly, high concordance in fold-change values was observed between TempO-Seq FFPE and RNA-Seq FROZ results (R2 ≥ 0.92). In contrast, RNA-Seq FFPE samples showed few overlapping DEGs compared to FROZ RNA-Seq (≤5 for all dose groups). Modeling of DCA-dependent changes in gene sets identified benchmark doses from TempO-Seq FROZ and FFPE samples within 1.4-fold of RNA-Seq FROZ samples (93.9 mg/kg-d), whereas RNA-Seq FFPE samples were 3.3-fold higher (310.3 mg/kg-d). This work demonstrates that targeted sequencing may provide a more robust method for quantifying gene expression profiles from aged archival FFPE samples.

Keywords: AIC, Akaike Information Criterion; BMD, Benchmark dose; BMDA, traditional benchmark dose based on apical endpoint; BMDL, lower 95% confidence limit of benchmark dose; BMDT, transcriptomic benchmark dose; BMDU, upper 95% confidence limit of benchmark dose; Benchmark dose (BMD) analysis; CPM, counts per million; DCA; DCA, dichloroacetic acid; DEG, differentially expressed gene; DOs, detector oligos; Dichloroacetic acid; FDR, false discovery rate; FFPE, formalin-fixed paraffin-embedded; FROZ, frozen; Formalin-fixed paraffin-embedded (FFPE); GO, gene ontology; IRIS, Integrated Risk Information System; NAM, new approach method; Phred score, negative log10 of the base-calling error probability; RIN, RNA integrity number; RNA-Seq; RNA-Seq, RNA-Sequencing; TempO-Seq; TempO-Seq, Templated Oligo Sequencing.