Formalin fixation and paraffin embedding, better known as FFPE, is the most ubiquitous method for preserving clinical tissue specimens. One of the ways that the FFPE process stabilizes and maintains tissue architecture is through the addition of hydroxymethyl groups to RNA, which crosslinks RNA to itself and to other biomolecules. However, FFPE tissue preparation also induces RNA fragmentation via hydrolysis. And because standard RNA analysis methods—namely RT-PCR and RNA-seq—require long stretches of unmodified RNA, these methods are often inefficient and unreliable when applied to RNA from FFPE-treated specimens. These technical inadequacies motivated Dr Joel Credle and his colleagues in the lab of H Benjamin Larman at Johns Hopkins University (Baltimore, MD, USA) to establish a more sensitive and accurate methodology for analyzing gene expression in FFPE samples. With help from IDT, they developed a new technique for analyzing RNA in clinical FFPE specimens that offers numerous advantages over traditional methods. The technique, called Ligation in situ Hybridization (LISH), and its applications are described here.
Probe design. LISH begins with a pair of DNA probes that target a specific mRNA sequence of interest in the FFPE samples. As shown in Figure 1, the probe sequences are designed to be immediately adjacent, but non-overlapping. A crucial feature in probe design is to include 2 RNA bases at the 3′ end of the upstream probe and a phosphorylation modification at the 5′ end of the adjacent, downstream probe. Phosphorylation and the RNA bases are required for efficient probe set ligation by T4 RNA Ligase 2 (Rnl2). Additionally, the 5′ end of the upstream probe, and the 3′ end of the downstream probe are each flanked with a universal adapter sequence for downstream PCR amplification.
Assay workflow. Following probe hybridization to mRNA in a tissue section for about an hour, unbound probes are washed away. The LISH probe sets are then ligated in situ with Rnl2. This enzyme selectively ligates nicked, double-stranded nucleic acids, and is unique in that it naturally utilizes an RNA guide strand (mRNA in Figure 1). Ligation occurs between the adjacent bound probe sequences and results in a single continuous DNA/RNA strand (the “ligation product”). Further details of the probe ligation chemistry used in LISH were described in 2014 by Larman HB, Scott ER, et al. .
An important feature of the LISH 2-probe design and ligation step is that it dramatically improves specificity compared to probes that are simply hybridized and recovered. In addition, the high efficiency of the ligation reaction provides direct proportionality between the number of mRNA target molecules and the number of ligation products. Finally, because these ligation products are uniform in size and Tm, they can be subjected to PCR with negligible probe-to-probe amplification bias. Larman explains, “These 3 features of LISH ligation products—their high specificity, linear dependence on target abundance, and unbiased amplification—are what make them attractive for a range of FFPE tissue studies, particularly in cases where multiplex gene expression analysis is desired for small, distinct areas defined by microscopic analysis of the tissue. One exciting example is the tumor-immune microenvironment.”
The Rnl2 enzyme can ligate LISH probe sets annealed to complementary RNA or DNA strands (such as genomic DNA, gDNA). In LISH-based gene expression analysis, gDNA-bound ligation products are undesirable and should be eliminated from downstream analysis. The scientists address this by using RNase H1 (IDT) to specifically release RNA-bound ligation products into solution, where they can be recovered for downstream analysis. The gDNA-bound ligation products, on the other hand, remain hybridized in the nucleus, and are thus absent from downstream analysis. For applications involving ligation products that remain in the tissue, it is also possible to specifically destroy the gDNA-bound ligations products using RNase H2.
Figure 2 outlines the step-by-step workflow of the LISH assay.
Figure 2. LISH assay workflow. Step 1. Annealing: Pairs of chimeric 3′-diribonucleotide-containing and 5′-phosphorylated DNA probes are hybridized to formalin-fixed RNA (i) or DNA (ii) within a tissue section. Unbound probes are washed away. Step 2. Ligation: Rnl2 is added to ligate the adjacently annealed probe pairs in situ (i and ii). Step 3. Digestion: RNase H treatment (i) degrades the mRNA bound by ligation product, releasing the ligation products into solution for downstream analysis and/or (ii) cleaves the RNA dinucleotides in ligation product bound to gDNA, thus destroying those ligation products. Step 4. Amplification: Remaining ligation products are amplified by PCR using universal adapters: OF and OR.
Credle and colleagues first sought to establish whether RNA in FFPE human spleen was a viable template for LISH. Using end-point PCR analysis, LISH probe pairs targeting RPS19 and GAPDH transcripts were mixed together and found to successfully ligate in situ, and most importantly, in the correct combinations.
Targeting the same 2 genes, the group then compared the absolute sensitivity of LISH to that for RT-qPCR and showed they were comparable. "These experiments were particularly exciting, because they revealed LISH to be at least as sensitive as RT-qPCR on a per-gene basis, but with LISH + Illumina sequencing (“LISH-seq”), we can measure a much greater number of genes per analysis," Credle explained.
Subsequent experiments highlight several features of the LISH workflow that could offer unique benefits to future research studies. These include the ability to:
- Perform highly multiplexed (>100-plex) gene expression analysis of patient-derived FFPE samples
- Amplify ligation products with negligible bias, for downstream quantification using LISH-seq
- Apply LISH-seq to archived FFPE specimens (up to 10 years old were demonstrated in this study)
- Conduct sequential LISH analyses of different RNA targets in the same FFPE sample
- Use LISH in conjunction with histochemical staining
- Analyze exceedingly small tissue fragments, such as those obtained by laser capture microdissection (LCM)
Overall, these findings support the potential wide utility of the LISH method. In addition to the increased sensitivity LISH provides compared to traditional gene expression analysis in FFPE samples, LISH does not require RNA extraction, reverse transcription, or destruction of the tissue samples. Performing LISH does not necessitate highly technical knowledge or equipment. "Due to its broad utility and low barrier for adoption, Joel and I are looking forward to helping other labs incorporate LISH into their arsenal of RNA analysis techniques," says Larman.