With increasing recognition as powerful agents of post-transcriptional regulation, the 170+ known RNA modifications—and methods to study them—have lately become subjects of intense focus. Earlier methods involved binding or precipitating fragments of RNA, followed by next-generation sequencing (NGS) to yield RNA segments (<200 nt) containing site(s) of modification. Recent advances are helping researchers to pinpoint modifications to single nucleotides, especially for more intensively studied modifications such as m6A, m5C, pseudouridine, and inosine. Definitive mapping of RNA modifications may expand our understanding of how they exert their effects, including their roles in mRNA stability and dysregulation in diseases. Here’s a look at tools used to study RNA modifications with regional and single-nucleotide resolution.
Binding and sequencing methods
Many current methods begin with RNA fragmentation, followed by either enriching fragments that contain modifications, or applying a chemical or enzyme that helps to distinguish modified from unmodified bases, and then obtaining a readout with NGS. (A comprehensive description of individual methods is beyond the scope of this article.) Single-nucleotide resolution is possible for some methods, but it’s tricky—in part because there are so many different types of modifications. “No single method reliably maps every RNA modification,” says Erbay Yigit, Associate Director of Research at New England Biolabs. “The right approach depends on the modification [type] and RNA class.” Methods often work by causing a perturbation—such as crosslinking or enzymatic activity—near modifications that results in a sequencing signature detectable by subsequent NGS.
For detecting oft-studied RNA modifications, there are simply more methods available. For example, one of the most prevalent modifications, m6A, can be pinpointed to single nucleotides with methods such as GLORI and eTAM-seq. The pseudouridine modification likewise can be detected with single-nucleotide resolution with the BID-sequence and BACS methods. Another method, miCLIP, “uses antibody-RNA crosslinking and the resulting RT signatures to identify individual m6A sites,” says Yigit. “It was an important step toward single-site mapping, but the sites are inferred from crosslinking signatures next to the m6A rather than read directly. It also can’t tell you what fraction of transcripts carry the modification at a given site, and results depend heavily on antibody quality and specificity.” Going forward, he expects to see significant progress in enzymatic methods, in which enzymatic activities are tuned to select for different RNA modifications.
Mass spectrometry
Tandem mass spectrometry (LC-MS/MS) is known as a gold standard method of measuring the total abundance of modifications in a sample. Typically, the RNA is digested completely into single nucleosides, which gives accurate total modification levels, but no information about where the modifications are located. “Alternatively, RNA can be digested into longer fragments, often with sequence-specific ribonucleases, allowing modifications to be mapped to individual positions,” says Yigit. Both approaches require relatively large amounts of input RNA, and mass spectrometry isn’t likely to become an everyday technique. “Mass spectrometry will remain the gold standard and the technology to validate other approaches, but not every lab has access to it, and it does not scale well,” he says.
Nanopore direct RNA sequencing
A newer method is direct RNA sequencing from Oxford Nanopore Technologies. This “third-generation sequencing” method reads the sequence (including modifications) directly from an individual native RNA molecule that passes through a nanopore. The electrical signal across the nanopore is measured, with variations correlating to different RNA modifications. “This information-rich signal is processed by deep-learning algorithms to output a highly accurate sequence, including methylation calls,” says Daniel Garalde, Director of Emerging Omics at Oxford Nanopore Technologies. Oxford Nanopore has base-calling models to sequence eight types of RNA modifications, and it is possible to detect others with models developed by academic research groups. “Modification models include the most commonly studied m6A, m5C, inosine, pseudouridine and all four 2’O-Me modifications,” says Garalde.
The method requires neither reverse transcription nor amplification, and it uses native RNA without antibodies, chemical conversions or crosslinking. It can also detect multiple types of RNA modifications on one molecule in a single run. Disadvantages of the direct RNA method—when compared to Oxford Nanopore’s standard cDNA sequencing—include lower throughput per run, which results in a higher cost per read. However, Garalde expects further progress in this relatively new method, noting that “Oxford Nanopore is continuing to improve RNA sequencing chemistry accuracy and throughput, and to expand the supported RNA modification base-calling models.”
Single bases vs multiple modifications
Although pinpointing specific RNA modifications to individual nucleotides in a sequence is valuable, the technology can come with trade-offs. For example, if starting sample material is precious or limited, splitting the sample and performing a separate assay for each modification may be impossible. “We believe that sensitivity, quantitative performance, and the ability to simultaneously measure multiple RNA modifications are equally important for understanding how the epitranscriptome changes across biological conditions,” says Gudrun Stengel, CEO and Co-founder of Alida Biosciences.
Alida Biosciences’ EpiPlex platform captures modifications with specific binders, and identifies them using proximity barcoding. “What really differentiates EpiPlex, however, is its ability to profile three modifications—m6A, inosine, and pseudouridine—alongside RNA expression in a single, highly sensitive assay,” says Stengel. The platform measures all three at once, at different resolutions: single-base for pseudouridine and inosine and regional for m6A. That saves time, funds and sample material compared with running separate assays. Depending on your experiments and workflows, the ability to assay multiple modifications simultaneously might outweigh the importance of single-base resolution. “Many studies are primarily interested in identifying changes in RNA modification patterns across cell types, disease states, or experimental conditions,” she says. “For these applications, being able to quantitatively profile RNA isolated from frozen and FFPE tissue, whole blood, plasma, or extracellular vesicles can be just as important as nucleotide-level precision.”
According to Stengel, the larger challenge is making RNA modification measurements reproducible, affordable, and routine. She expects the next major advances in epitranscriptomics to combine high sensitivity, low input requirements, and streamlined assay and analysis workflows. “Ultimately, the goal is not simply to generate increasingly detailed maps of RNA modifications,” she says. “It is to understand what those modifications do, how they change across cell states, and how they contribute to disease.”