Selective depletion of rRNA enables whole transcriptome profiling of archival fixed tissue

Abstract

We report a method for Selective Depletion of abundant RNA (SDRNA) species from Human total RNA isolated from formalin-fixed, paraffin-embedded (FFPE) tissue, here demonstrating removal of ribosomal and mitochondrial transcripts from clinical FFPE tissue RNA archived up to 20 years. Importantly, SDRNA removes 98% of targeted RNAs while preserving relative abundance profiles of non-targeted RNAs, enabling routine whole transcriptome analysis of clinically valuable archival tissue specimens by Next-Generation Sequencing.

Figure 1. Read density of targeted regions.

Read density plotted across regions targeted for depletion and compared between different depletion methods. Regions targeted for depletion (SDRNA1 = green, SDRNA2 = red and green) are indicated at the bottom of each figure. Density values were computed using igvtools count (www.broadinstitute.org/igv/igvtools) and plotted using UCSC’s Genome Browser. FF – Fresh-frozen; FFPE – Formalin-fixed, paraffin-embedded.(a) Reads are mapped to chrUn_gl000220, an unplaced genomic contig from hg19 containing the 13 kb rRNA transcript described in Supplementary Figure 1a. (b) Reads are mapped to the human mitochondrion from hg18 (Supplementary Figure 1b).

Figure 2. Scatterplots of RefSeq transcript abundance.

Log₁₀ read counts are plotted. Transcripts with zero reads are excluded from analysis. An identity line is shown in each plot for reference. Sub-populations are indicated as follows: Black cross – sno-, sc-, sn- and mir-RNAs from RefSeq; Black and red circles – Genes homologous to SDRNA and SDRNA2 probes, respectively (Table S4); Light blue circles – All other RefSeq transcripts. FF – Fresh-frozen; FFPE – Formalin-fixed, paraffin-embedded. (a) FFPE RNA Pool: SDRNA1 vs SDRNA2. (b) FF breast RNA : polyA+ vs SDRNA1. (c) FF breast RNA : polyA+ vs SDRNA2.

Figure 3. Scatterplots of RefSeq transcript abundance: Depleted vs. Undepleted.

Log₁₀ read counts are plotted. Transcripts with zero reads are excluded from analysis. An identity line is shown in each plot for reference. Sub-populations are indicated as follows: Black cross – sno-, sc-, sn- and mir-RNAs from RefSeq; Black and red circles – Genes homologous to SDRNA and SDRNA2 probes, respectively (Table S4); Light blue circles – All other RefSeq transcripts. FF – Fresh-frozen; FFPE – Formalin-fixed, paraffin-embedded. (a) FF Breast RNA: not depleted vs polyA+ (b) FF Breast RNA: not depleted vs SDRNA1 and (c) FF Breast RNA: not depleted vs SDRNA2. (d) FFPE RNA pool: not depleted vs SDRNA1 and (e) FFPE RNA pool: not depleted vs SDRNA2.

Figure 4. Distribution of uniquely-mapped reads across broad RNA classes. rRNA class includes all reads mapping to the 13 kb transcript shown in Supplementary

Figure 1a . Mito class refers to all reads mapping to the mitochondrial genome shown in Supplementary Figure 1b. The Intergenic class encompasses all remaining reads not mapping to RefSeq annotations or their associated introns. FF – Fresh-frozen; FFPE – Formalin-fixed, paraffin-embedded. (a) polyA+ FF breast RNA.(b) SDRNA2 FF breast RNA. (c) SDRNA2 FFPE RNA pool.