NEW YORK (GenomeWeb) – Stanford University researchers have devised a method for detecting RNA-protein binding events.
Described in a paper published this week in Nature Methods, the approach could enable more-efficient and less-biased in vivo measurements of RNA-protein interactions, said Muthukumar Ramanathan, a graduate student in the lab of Stanford professor Paul Khavari and first author on the study.
RNA-protein interactions are known to play a role in a variety of cellular functions and disease processes, but existing methods for studying these interactions have their limitations, Ramanathan said.
For instance, in one commonly used method researchers synthesize or otherwise isolate their RNA molecule of interest and then expose it to cell lysate to see what proteins bind.
This, Ramanathan noted, can lead to various biases due to alterations in the cellular environment upon cell lysis and the identification of binding events that may not be truly reflective of those occurring in vivo.
RNA-protein crosslinking experiments are another approach to studying these interactions. But while these experiments provide a look at the actual in vivo RNA-protein interactions, they require large numbers of cells, Ramanathan said. He added that the pull-down process in these experiments, which typically uses biotinylated DNA, is biased towards longer RNAs, making analysis of short RNA motifs more challenging.
To address these issues the Stanford team adopted a proximity labeling approach. Called RaPID (for RNA-protein interaction detection), it uses the biotin ligase BirA* attached to an RNA molecule of interest to biotinylate proteins bound to that RNA. The biotinylated proteins can then be pulled down via streptavidin capture and analyzed by mass spectrometry or other methods.
Such proximity labeling approaches are increasingly used for protein interaction studies. One method developed by Stanford researcher Alice Ting involves genetically inserting an ascorbic acid peroxidase (APEX) into proteins of interest. In the presence of hydrogen peroxide and biotin-phenol, peroxidase will generate a phenolic radical that will modify surrounding proteins with biotin. These biotin-tagged proteins can then be pulled down using streptavidin beads and identified and quantified using mass spectrometry, the idea being that the biotinylated proteins are in close proximity to the APEX-containing protein, and so are likely interactors.
The BirA*-based approach adopted by Ramanathan and his colleagues was originally devised by researchers at Sanford Children's Health Research Center in Sioux Falls, South Dakota and Singapore's Institute of Medical Biology. Ramanathan said that upon seeing researchers in his lab use the method for protein-protein interaction work, he realized it could also potentially be adapted to RNA-protein interaction studies.
The researchers tested the technique in an experiment looking at the RNA sequence EDEN15, which is known to bind the protein CELF1. Using RaPID to analyze protein binding to EDEN15 in HEK293T and Huh7 cells, the researchers identified EDEN15 as a binder of CELF1, with their interaction scoring a 0.9 according to the SAINT algorithm — a commonly used measure of confidence in molecule interactions in which 0 indicates no probability of a true interaction and 1 indicates the highest probability.
Ramanathan said he and his colleagues used random scrambled sequences of RNA as controls, which allowed them to identify false positives. They also screened their findings against the Contaminant Repository for Affinity Purification (CRAPome), a database that contains contaminants observed in protein-protein and protein-nucleic acid experiments.
"We used that [process] to make it as stringent as possible to eliminate any background proteins and identify only true interactors," he said. "We err on the side of more specificity. If we relaxed the threshold [the technique] might be more sensitive, but in general we prefer it to be more specific, and we find that with our cutoffs we get very specific results."
RNA-protein bindings events are associated with a variety of diseases, and with this in mind the researchers looked at several conditions to provide a proof of principle of the technique's usefulness. In the case of hereditary hyperferritinemia-cataract syndrome they were able to demonstrate that loss of binding between the IRE RNA motif due to point mutations and the IRE-binding protein IREB2 appeared linked to disease severity.
Looking at breast cancer and the previous observation by other groups that RNAs with SM1v1 motifs are decreased in advanced disease, they identified the protein RC3H1 as an interactor of these motifs. Pairing this finding with transcriptomic data from the Cancer Genome Atlas project indicating that upregulation of RC3H1 is associated with poorer survival in breast cancer, they hypothesized that, as they wrote, "overexpression of RC3H1 may lead to abnormal post-transcriptional control of RNAs bearing SM1v1-like motifs, potentially contributing to an altered disease course."
Ramanathan said this sort of cancer research is one of the main areas where he and his colleagues hope to apply the technique. He noted that while previous groups had observed the link between advanced breast cancer and decreased RNA SM1v1 motifs, "nobody really knew what was interacting with [those motifs]."
"With this method we could just plug-and-play and really discover that [interaction] quickly," he said. "So if you have an RNA motif that is upregulated or downregulated in cancer, it's really easy [using RaPID] to find out if there is a [relevant] protein interaction and what is the significance of the interaction."
"That's something we're looking into in the lab, trying to understand if are there other RNA sequences or motifs that are changing in cancer and what is the binding profile of those motifs and how that might affect cancer progression," he added.
Another area of interest is studying how drugs impact RNA-protein interactions, Ramanathan said. To help facilitate this and other work, the researchers engineered a new BirA* ligase from Bacillus subtilis that they determined enabled 1,000-fold faster kinetics and a 30-fold increase in signal-to-noise compared to the Escherichia coli-derived BirA* ligase previously used.
"As we study RNA-protein interactions, we want to be able to disrupt them or stabilize them [with drugs]," Ramanathan said. "And if you have a shorter time period, you can better study how fast the drug works and whether it works or not."