The SMARTer approach to RNA-seq of FFPE tissues

FFPE samples: a double-edged sword

Tissue banks are a valuable source of material, but their utility in RNA sequencing experiments has been limited due to many of these samples being formalin-fixed and paraffin-embedded (FFPE). This preservation method has proven crucial to advancing genetic research: by allowing samples to be archived long-term without specialized storage equipment, researchers are able to retroactively examine large numbers of tissue samples for gene expression changes. However, this fixation process results in severe degradation of nucleic acids and renders FFPE samples a challenging source material for RNA-seq.

These difficulties spurred us to develop the SMARTer Stranded Total RNA-Seq Kit v2 - Pico Input Mammalian (referred to below as "Pico v2"), which leverages our proprietary template-switching SMART technology to enable the use of as little as 250 pg of input RNA of any quality (see workflow below). Our Pico v2 kit was put to the test against several competitors' kits in a head-to-head comparison of RNA-seq of FFPE samples. Spoiler alert: we passed with flying colors!

The challenge

At ABRF 2018, the Tighe group (University of Vermont) presented data where they used human thyroid tumor FFPE samples (n = 6) to generate RNA-seq data. The authors analyzed the performance of our kit (Pico v2) versus two competitors' kits (Diagenode CATS Total RNA-Seq Kit V2I and X Total RNA Library Prep Kit V1), directly comparing various sequencing quality metrics (Table 2) as well as transcript coverage (Figure 2 and 3).

The victor

All three kits exhibited per-sample variability in the number of total, rRNA, and filtered reads; however, in all six samples, our Pico v2 kit detected the most transcripts, with up to 763% more than the closest competitor (median of 143%) and superior expression profiling efficiency (Table 2). Furthermore, in five out of six cases, our kit demonstrated the highest uniquely mapped raw read count and percentage of uniquely mapped reads as well as the highest exonic rate. While our Pico v2 kit only had the lowest duplication rate in 2 out of 6 samples, we note that duplication rate is influenced by read depths and, as such, the differences in duplication rates are expected under the circumstances. Combined, these results indicate that our kit's performance was the least impacted by severe RNA degradation and low sample quality—a key metric, given the high degree of variability in FFPE samples.

The Pico v2 kit also demonstrated superior transcript coverage in both high- and low-expression transcripts (Figures 2 and 3, respectively). Our kit demonstrated more even coverage of both high- and low-expression transcripts (average of ~30X coverage), particularly compared to a competitor's kit that exhibited a marked bias towards the 3' end of transcripts.

The takeaway

Despite the difficulties of sequencing RNA in FFPE tissue samples caused by severe nucleic acid degradation, FFPE tissue samples will continue to be an essential sample source for genetic research. The SMARTer Stranded Total RNA-Seq Kit v2 - Pico Input Mammalian answers these challenges in a head-to-head competition by providing superior transcript detection and expression profiling efficiency, greater uniquely mapped reads, and the highest exonic rate of sequenced samples.

Schematic of technology in the SMARTer Stranded Total-RNA-Seq Kit v2 - Pico Input Mammalian. Random priming allows the generation of cDNA from all RNA fragments in the sample, while the use of SMART technology allows the preservation of strand-of-origin information. Following a first-round PCR amplification that adds full-length barcoded Illumina adapters, cDNA derived from ribosomal RNA is cleaved with our ZapR™ v2 technology prior to library enrichment by a second round of PCR amplification.