Background

RNA-targeting CRISPR systems that reduce specific transcripts without altering the genome are widely used in gene function studies and RNA therapeutic development. Representative systems include single-protein PspCas13b and RfxCas13d (CasRx), and multi-subunit Csm. The RNA knockdown efficiency of these systems is generally evaluated using reverse transcription quantitative polymerase chain reaction (RT–qPCR).

A common practice is to design amplicons that cross the cleavage site, presumably to prevent amplification of RNA fragments remaining after cleavage. However, the potential for interference between actual transcript reduction and measurement artifacts has not been adequately reviewed. Researchers at Johns Hopkins University began their investigation from the paradox that a catalytically inactive Csm also showed strong knockdown, prompting a re-evaluation of the quantification method itself.

Key Findings

The team compared wild-type Csm and a ribonuclease-deficient Csm3-D33A in HEK293T cells, targeting nuclear XIST and cytoplasmic BRCA1. The mutant variant, which lacks cleavage activity, showed similar knockdown in RT–qPCR as the wild-type, but no reduction was observed in protein analysis or mCherry flow cytometry. This suggests that the RNA was not actually reduced but the measurement was distorted.

The error varied significantly depending on primer location. For BRCA1, amplicons crossing the guide RNA binding site estimated a 78% knockdown for the ribonuclease-deficient variant and 42% for an upstream amplicon, but no reduction was observed downstream. Wild-type Csm showed 92% knockdown with crossing amplicons but only 64% downstream. In XIST, crossing amplicons showed approximately 90% knockdown for both wild-type and deficient variants, but downstream measurements were 35% and 14%, respectively. Similar results were observed with PspCas13b and CasRx.

The cause was the co-purification of guide RNA during RNA extraction. This RNA rebinds to the target transcript, blocking reverse transcriptase progression and reducing complementary DNA synthesis in regions near the binding site. Adding synthetic guide RNA to reverse transcription reactions induced concentration-dependent errors, which were not reversed by protease K treatment. The upstream interference effect decreased with distance and disappeared beyond approximately 500 nucleotides, with a decay constant of 170 nucleotides in the fitted curve.

Implications and Outlook

The team restored measurement accuracy by isolating the guide RNA using complementary 3′ dideoxycytidine-terminated antisense oligonucleotides (ddASO). Using the ultraMarathonRT (uMRT) reverse transcriptase, which has strong strand displacement and long continuous synthesis capabilities, the false knockdown in the deficient Csm and PspCas13b was eliminated, and the positional measurements of wild-type variants became more consistent.

In the laboratory, using a combination of processive and strand-displacing reverse transcriptases like uMRT with primers crossing the target site, or measuring downstream of the guide binding site with conventional enzymes, could mitigate the issue. Independent analyses such as protein expression, flow cytometry, and RNA sequencing should also be conducted. However, RNA sequencing may also be affected if complementary DNA synthesis occurs before RNA fragmentation.

These results do not imply that all RNA-targeting CRISPR efficiency values in existing literature are incorrect. The magnitude of the error depends on guide RNA quantity, binding affinity, extraction methods, and reverse transcription priming strategies. Phenol–chloroform extraction may concentrate small RNAs and exacerbate the problem, and oligo(dT) priming could affect the entire upstream region of the binding site. Similar position-dependent errors should be re-examined in studies involving small interfering RNA (siRNA), short hairpin RNA (shRNA), and antisense oligonucleotides (ASO).