Purification of cross-linked RNA-protein complexes by phenol-toluol extraction

Abstract

Recent methodological advances allowed the identification of an increasing number of RNA-binding proteins (RBPs) and their RNA-binding sites. Most of those methods rely, however, on capturing proteins associated to polyadenylated RNAs which neglects RBPs bound to non-adenylate RNA classes (tRNA, rRNA, pre-mRNA) as well as the vast majority of species that lack poly-A tails in their mRNAs (including all archea and bacteria). We have developed the Phenol Toluol extraction (PTex) protocol that does not rely on a specific RNA sequence or motif for isolation of cross-linked ribonucleoproteins (RNPs), but rather purifies them based entirely on their physicochemical properties. PTex captures RBPs that bind to RNA as short as 30 nt, RNPs directly from animal tissue and can be used to simplify complex workflows such as PAR-CLIP. Finally, we provide a global RNA-bound proteome of human HEK293 cells and the bacterium Salmonella Typhimurium.

Fig. 1

PTex is a fast method to purify cross-linked RNPs. a In vivo cross-linking of HEK293 cells using UV light at 254 nm wavelength results in covalent bonds between RNA and proteins in direct contact. Cross-linked RNPs are indicated by an orange star. b Schematic of the separation principle of biphasic organic extractions used in PTex. Left panel: Phenol-Toluol (50:50) and neutral pH results in an accumulation of proteins and RNA in the upper aqueous phase (aq) while DNA and lipids are retained at the interphase (inter). Right panel: under acidic phenol and chaotropic conditions, non-cross-linked RNA accumulates in the aqueous phase (aq), non-cross-linked proteins in the lower organic phase (org), and cross-linked RNPs (clRNPs) are enriched at the interphase (inter). c Step-by-step analysis of proteins in 9 intermediary steps of the PTex protocol (3 extractions with 3 phases each). Western blot against HuR (ELAVL1, 35 kDa) demonstrates that UV-cross-linking-stabilised HuR-RNA complexes (upper edge/gel pocket of the blot) are largely enriched after PTex (step 3 interphase). A purified fly protein (Sxl RBD4) served as spike-in as 100% non-cross-linked RBP. d 5′-end radioactive-labelled RNA was subjected to PTex in vitro. e PCR with specific primers against exon 5 of the interleukin 3 (IL3) gene demonstrates efficient removal of genomic DNA after either full HEK293 cells (upper panel) or pre-purified genomic DNA (middle panel) were subjected to PTex. A PCR product derived from linear pUC19 DNA (lower panel) is also removed. f Enrichment of known RBPs by PTex tested by western-blot against PTBP1, FUS, or against non-classical RNA-binding enzymes Eno1 and GAPDH. Note that RNaseA treatment was performed after PTex as it removes partially shifted bands (smear) for some RBPs. g PTex enriches for cross-linked RBPs. RNase treatment before PTex strongly reduces recovery of known RNA-binders (PTBP1, FUS). Non-RBP controls Histone H3 and actin (ACTB) are efficiently depleted by PTex (c, f, g). For full gels/blots see Supplementary Figures 1–8

Fig. 2

Performance of PTex. a RNA interactome capture of HEK293 was used as 100% cross-linked input material and HuR recovery by PTex was assessed. b Quantification of a (HuR n = 6 biologically independent experiments). c Drosophila melanogaster RBP Sxl RBD4 was bound to RNA of several length carrying the Sxl U7 recognition site and UV cross-linked in vitro; minimal RNA length for efficient recovery by PTex is 30 nt. d Quantification of c (n = 5 biologically independent experiments). e PTex recovery of RNA (260 nm) and proteins (280 nm) determined by UV spectroscopy (n = 9; 5 biologically independent experiments; 2 of those with 3 technical repeats each). f, g Relative enrichment of cross-linked HuR (source Fig. 1c) and Sxl (source Fig. 2c, n = 3 biologically independent experiments) by PTex. Error bars represent SD. All data are in Supplementary Data 4. For full blots see Supplementary Figures 10–14

Fig. 3

RNP purification from animal tissue. Mouse brain tissue was cryo-grinded and UV-irradiated before (Hot)-PTex was performed. Western blot against HuR (ELAVL1) demonstrates recovery of cross-linked HuR from mouse tissue while beta-actin (ACTB) is efficiently depleted. For full blots see Supplementary Figure 15

Fig. 4

A fast PAR-CLIP variant employing phenolic extraction (pCLIP). a Schematic comparison of PAR-CLIP variants. b Read length distribution of uniquely mapping reads utilised for determine binding sites (cluster) of HuR (ELAVL1). PAR-CLIP samples were processed using PARpipe (see methods). c Relative proportion of PARalyzer-derived cluster annotation. d Heatmap of relative positional binding preference for intron-containing mRNA transcripts for each of the six HuR PAR-CLIP samples. Sample-specific binding preferences were averaged across selected transcripts (see methods). The relative spatial proportion of 5′UTR, coding regions and 3′UTR were averaged across all selected transcript isoforms. For TES (regions beyond transcription end site), 5′ splice site, and 3′ splice site, we chose fixed windows (250 nt for TES and 500 nt for splice sites). For each RBP, meta-coverage was scaled between 5′UTR to TES. The 5′ and 3′ intronic splice site coverage was scaled separately from other regions but relative to each other. e We applied de novo motif discovery for PARalyzer derived clusters using ZAGROS (left) and DREME (right). For Zagros, we found a T-rich motif scoring the highest in all cases. As ZAGROS does not return E-values we analysed the cluster sequences using DREME. For all but classic PAR-CLIP R2 we found a T-rich motif scoring the highest. For classic PAR-CLIP R2 however, the T-rich motif scored second with a similar E-value to a less frequent primary motif (Supplementary Fig. 16). f, g Genome browser shots of TUBB and SRSF6 example genes showing reproducible 3′UTR binding sites. Track y-axes represent uniquely mapping read count

Fig. 5

A global snapshot of RNPs in HEK293 cells. a Schematic of the experimental setup: HEK293 cells were UV-cross-linked using 0 (noCL), 0.015 (dark red), 0.15 (red) and 1.5 (dark yellow) J/cm² 254 nm light in triplicates. Total RNA from input (whole-cell lysate) and PTex-purified samples were analysed by RNA-Seq. b Deletions in RNA from input and PTex samples; frequency of mutations in transcripts correlate with higher UV doses. Boxplot centre line represents median, bounds are first and third quantile, and whiskers extend to 1.5 times the inter-quantile range (n = 3 biologically independent experiments except for input 0.015 J/cm² (n = 2). c Mutations (deletions = green, substitutions = blue) enriched in UV-irradiated samples were plotted to their position relative to AUG and Stop codon in coding sequences and serve as indicator for protein-binding sites. Note that we cannot delineate which protein bound to which position. Plots for PTex are shown; for input see Supplementary Fig. 17, 18. d–g Input (whole-cell lysate) and PTex-purified sample were analysed by label-free mass spectrometry. d Volcano plots of proteins enriched by PTex (FDR 0.01) under the three cross-linking conditions. e Overlap of PTex-enriched proteins (enriched in all 9 replicates, FDR 0.01) is 3037 (these PTex proteins are from here on coloured in orange). f Protein abundance (IBAQ intensities of input samples) does not correlate with PTex enrichment (log2-fold change of intensities [CL/-CL]). g PTex does not select for a subset of proteins based on general features such as molecular weight, pI or hydrophobicity. Boxplot centre line represents median, bounds are first and third quantile, and whiskers extend to 1.5 times the inter-quantile range. h, i PTex of individual predicted RNA-associated proteins. ATP-binding cassette sub-family F member 2 (ABCF2) and T-complex protein 1 subunit eta (CCT7) have not been reported to bind RNA. Both are enriched after PTex in a UV-irradiation-dependent fashion, indicating that they indeed associate with RNA in vivo. For full blots see Supplementary Figure 19

Fig. 6

Features of RNA-interacting proteins found by PTex. a Top 3 enriched GO terms (CC, MF, BP) and b enriched protein domains in PTex-purified proteins from HEK293 cells; p-value derived from a one-tail Fisher Exact test and FDR determined by the Benjamini–Hochberg method. c PTex-purified proteins overlap with well-described RBPs but not transcription factors. Recovery of Gerstberger RBPs and transcripiton factors (TFs) reviewed in ref. , a recent review on RBPs by ref. , HEK293 poly-A binders, RBPs found by RNA interactome using click chemistry (RICK; CARIC), P-body components and a recent prediction of candidate RBPs (SONAR,). d Distribution of previously identified HEK293 mRNA-binding proteins (green;) in PTex; each bin represents 10% of the 3037 PTex proteins from lowest to highest enrichment. e mRNA-binding proteins display a bimodal pI distribution pattern with peaks at pH 5.5 and 9.5^, . RNA-interactors in general peak at pI 5–6 as found by PTex and RICK. Proteins of the RNA exosome are prototype PTex proteins. f The RNA exosome core consists of 10 subunits: nine non-catalytically active proteins (Exo-9) forming a barrel-like structure and an additional RNase (Rrp44; Exo-10); modified from ref. . g *All exosome subunits are labelled “RNA-binding” (Uniprot.org); **green = identified via poly-A selection in ref. ; orange = enriched in PTex

Fig. 7

PTex recovers bacterial RNPs. a Salmonella Typhimurium SL1344 Hfq-FLAG was UV-cross-linked and HOT-PTex was performed to purify bacterial RNPs. b Western blot using an anti-FLAG antibody demonstrates recovery of Hfq monomers linked to RNA. Note that the physiologically active Hfq hexamer partially withstands SDS-PAGE conditions and that this complex is also enriched after PTex. c RNPs in Salmonella were purified by PTex globally. 172 Proteins enriched after UV-cross-linking (PTex CL) contain ribosomal proteins (transparent red), known RBPs (red) and DNA-binders (orange). Individual enriched proteins not known to associate with RNA before were used for validation (in parentheses). d Validation of PTex-enriched RNA-interactors: Salmonella strains expressing FLAG-tagged proteins were immunoprecipitated  ±UV irradiation. RNA-association is confirmed by radioactive labelling of RNA 5′ ends by polynucleotide kinase (T4 PNK) using autoradiography; a signal is exclusively detectable after UV-cross-linking and radiolabelling of precipitated RNA. CsrA-FLAG (pos. ctr.), YigA-FLAG (neg. ctr.), AhpC-FLAG, SipA-FLAG and YihI-FLAG are bound to RNA in vivo. e GO terms significantly enriched among the RNA-associated proteins; p-value derived from a one-tail Fisher Exact test. For full gels/blots see Supplementary Figures 25, 26