A challenge of CLIP procedures is, again, their dependence on UV exposure to generate RNA-protein adducts. As outlined above, in vivo UV radiation is particularly inefficient: it usually takes minutes to complete and only about 1-5% of RNPs of interest are cross-linked28. Hence, some CLIP variants often require relatively large number of cells. Systematically poor cross-linking performance could be attributed to the impossibility of creating adducts with some amino or intrinsically limited UV penetration in certain cell types or growth media 2,15. Recently, we developed a much-improved UV cross-linker (Vari-X-linker; UVO3 1,21) that greatly increases the cross-linking efficiency in actively growing cells during markedly shorter time spans (seconds)1,21. However, the fastest and most efficient way to cross-link proteins to RNA is using UV lasers 29, although setting up such a system can be prohibitively expensive. Another major advantage of these rapid cross-linking devices is that they enable the monitoring of very dynamic changes in protein-RNA interactions, such as those occurring during stress responses, at high temporal resolution 1,21,29,30.
To quantify RNA-protein cross-linking efficiencies, most protocols attach radioactive phosphates or fluorescently labelled oligonucleotides to 5’ ends of the cross-linked transcripts. The RNP complexes are subsequently resolved by denaturing PAGE (Fig. 2) and the RNA can then be visualised by autoradiography or fluorescent imaging. Recently, we performed such radiolabelling analyses on a variety of RBPs identified in our S. aureus RBPome analyses. While the vast majority of the proteins tested detectably cross-linked to RNA, we learned that only for those proteins for which we observed very strong radioactivity signals (1 to 3-hour exposure of phosphoimager screen/film) were we able to obtain high-complexity complementary DNA (cDNA) libraries. It is also important to point out that the radiolabelling assay can generate false positive signals: radioactive labelling of cross-linked RNA involves an incubation step with T4 polynucleotide kinase (T4 PNK). However, it is possible that during this reaction the RBP of interest becomes radiolabelled by autophosphorylation or co-purified host kinases31. Consequently, we would always advise including control reactions with non-irradiated cells or leaving out PNK altogether (Fig. 2A).
As well as informing the choice of UV exposure, preliminary signal evaluation provides a strategy by which to filter out proteins with no or little RNA-binding activity. As outlined above, UV can also cross-link proteins to DNA 20. Accordingly, we strongly recommend performing control experiments where the cross-linked RNPs are incubated with increasing concentration of RNase I/A or DNase I. If the purified RNP indeed contains cross-linked RNA, higher concentrations of RNase should reduce the smearing of the bands corresponding to RNPs (Fig. 2B; PNPase). Alternatively, the intensity of those containing DNA-protein duplexes would only be resolved by DNase digestion. We also found that not all cross-linked proteins, such as the superoxide dismutase SodM, respond to either DNase or RNase treatment (Fig. 2B). This suggests that these proteins generally bind short nucleic acid fragments, which will make the library preparation steps more challenging.