Quantitative identification of proteins that influence miRNA biogenesis by RNA pull-down SILAC mass spectrometry (RP-SMS)

RNA-binding proteins mediate and control gene expression. As some examples, they regulate pre-mRNA synthesis and processing; mRNA localisation, translation and decay; and microRNA (miRNA) biogenesis and function. Here, we present a detailed protocol for RNA pull-down coupled to stable isotope labelling by amino acids in cell culture (SILAC) mass spectrometry (RP – SMS) that enables quantitative, fast and speci ﬁ c detection of RNA-binding proteins that regulate miRNA biogenesis. In general, this method allows for the identi ﬁ cation of RNA-protein complexes formed using in vitro or chemically synthesized RNAs and protein extracts derived from cultured cells.


Introduction
miRNAs regulate gene expression and control a variety of biological processes, including developmental timing, differentiation, metabolism and neuronal patterning [1][2][3][4]. Importantly, a number of miRNAs are expressed in a tissue-specific manner, thereby contributing to cell differentiation and function [5]. Moreover, changes in the levels of a few miRNAs affect processes that include neural differentiation and the formation of induced pluripotent stem cells [6,7]. Finally, aberrant miRNA expression is linked to a variety of human pathological states including the initiation, progression and metastasis of cancer [8,9].
We have previously shown that hnRNP A1, a protein implicated in many aspects of RNA processing, binds to the conserved terminal loop of pri-miRNA-18a [20,21] and pri-let-7a-1 [22], stimulating and inhibiting their processing, respectively. Since then, the biogenesis of many miRNAs has been shown to be regulated by various RNA-binding proteins (RBPs) [23,24]. We have elucidated the mechanisms that regulate miRNA biogenesis in mammalian cells using a method that combines RNA pull-down with stable isotope labelling by amino acids in cell culture (SILAC) high-throughput mass spectrometry (which we call here RP-SMS), developed in our laboratory. We have identified RBPs that regulate the production of brain-enriched and brain-specific miRNAs, such as miRNA-7 and miRNA-9 [25][26][27], as well as factors responsible for the selective uridylation and degradation of miRNA precursors in embryonic cells [28]. Here, we describe a step-by-step RP-SMS protocol that can be used to identify miRNA biogenesis factors as well as any other RNA-protein complexes.

Overview
By combining RNA pull-down with SILAC [29] high-throughput mass spectrometry (RP-SMS) we can identify proteins that bind specifically to a given RNA (such as a pri-miRNA or pre-miRNA), as opposed to non-specifically to beads [25]. SILAC is a method that detects protein abundance using the incorporation of non-radioactive 'heavy' amino acid isotopes, which can be distinguished from naturally occurring 'light' amino acid isotopes by mass spectrometry. Importantly, we can also detect RNA-protein complexes derived from different cell types or the same cell type exposed to different environmental conditions [26,27]. Finally, using RP-SMS, we can compare two RNAs (e.g. with or without one or more mutations, or with or without chemical modification) and uncover differentially bound proteins [28].
The method involves producing cell extracts derived from SILAClabelled cells, obtaining RNA (by in vitro transcription or chemical synthesis), performing the RNA pull-down assay, and undertaking mass spectrometry analysis.

Incorporate cells with heavy or light isotopes
Grow cells in SILAC media, 'heavy' and 'light', supplemented with dialysed calf serum, for six passages. We use DMEM medium containing 13 C-labelled arginine and 2 D-labelled lysine (R6K4 (DC Biosciences)) as the heavy SILAC medium, and control DMEM containing unlabelled arginine and lysine (R0K0 (DC Biosciences)) as the light SILAC medium. -Add Trypsin buffer to just cover the gel pieces, and incubate on ice for 15 min. As the gel absorbs the buffer, add more Trypsin buffer to slightly above the gel pieces, and incubate at 37°C for 30 min. Add enough acetonitrile/ABC solution to slightly above gel pieces (to make sure the gel pieces do not dry out), and leave at 37°C overnight.

Verify incorporation of heavy isotopes
- -Following digestion and peptide purification, perform an LC-MS/MS analysis using an orbitrap mass spectrometer or a similar machine. Determine the efficiency of heavy label incorporation into peptides by manually examining randomly selected peptides from the raw file. Since arginine and lysine may have different labelling efficiencies, this needs to be done separately for arginine-and lysinecontaining peptides. Arginine-to-proline conversion should also be examined.

PCR
-Design primers that encompass your RNA of interest, appending a T7 promoter sequence (TAATACGACTCACTATAGG) to the 5′ end of the forward primer. Amplify the DNA using high-fidelity PCR, and confirm the size of the PCR product by electrophoresis in an agarose gel.

Wash away unbound protein
-Work on ice. Wash reactions three times with 1 ml Buffer G, spinning at 4°C and 1000 rpm for 2 min. Combine the "light" and "heavy" samples during the last wash.

Mass spectrometry analysis
-Cut out the band, proceed with in-gel digestion as described in Section 3.2.2, and perform an LC-MS/MS analysis using an orbitrap mass spectrometer or its equivalent. The MaxQuant software [30] platform is used to analyse the raw mass-spectrometry data in order to determine the ratio of the heavy-labelled peptides to the lightlabelled peptides. The samples can also be analysed by Western blotting for known proteins.
We incorporated R6K4 into HeLa cells constitutively expressing LIN28A, and confirmed the SILAC incorporation (Fig. 1). Protein extracts from 'heavy' (H) and 'light' (L) cells were generated by scraping adherent cells in Roeder D buffer, followed by cell lysis using sonication, and clearing the extracts using centrifugation. The RP-SMS protocol was followed for the pre-let-7a-1 as described above, providing us with a list of identified proteins and their H/L ratios, peptide counts and intensities, and many additional features (Supplementary Table 1).
Notably, a H/L ratio of 1 in the RNA pull-down compared to the beads-only control signifies that there is no enrichment of protein binding to RNA. H/L ratios of ≤0.5 indicate either preferential cellular protein binding to the beads-only control or protein contaminants that derive from the environment (such as the experimentalist's skin). H/L ratios of ≥2 indicate specific protein binding to the RNA.
H/L ratios revealed that 81 cellular proteins were enriched at least two-fold (H/L ratio ≥2) in the pre-let-7a-1 pull-down compared to the beads-only control (Fig. 2), among which were many known RNAbinding proteins, such as the helicase DHX9, several hnRNP proteins, LIN28A, the splicing factor SRSF1, and the novel RNA-binding protein TRIM25 (Supplementary Table 1). In contrast, the ratio observed for 828 other proteins was > 0.5 and < 2 when binding to RNA vs. binding to the beads-only control were compared, indicating non-specific binding. Of the 19 proteins enriched on beads alone (H/L ratio ≤0.5), there were several types of keratin (a common mass spectrometry contaminant).
To validate the specificity of our RNA pull-down assay, we introduced a well-known LIN28A-binding site (GGAG) into the terminal loop of pre-miR-16, which does not bind LIN28A (Fig. 3A) [27,28,33,35]. All pre-miRNA-16 mutants showed efficient LIN28A binding (Fig. 3B). DHX9, which recognises double-stranded RNA, also displayed binding to all pre-miRNA-16 transcripts (Fig. 3B). Finally, we introduced hnRNP A1-binding sites (UAGG) [20,35] into pri-miR-16 and showed that all mutants with UAGG sequences bind hnRNP A1 efficiently (Fig. 3B). It is important to note that introducing binding sites for miRNA biogenesis factors does not automatically mean that processing of the pri-miRNAs or pre-miRNAs is dependent on or regulated by these factors. Other sequences, structural elements, and protein factors cooperate to control the miRNA processing pathway.
As the next step towards identifying miRNA processing factors, one could take all significant hits from the RP-SMS experiment and perform loss-of-function (RNAi and/or CRISPR/Cas9) or gain-of-function (overexpression) experiments. This should be followed by in vitro processing assays and/or in vivo miRNA processing analysis using extracts from Distribution of H/L ratios among proteins identified in the pre-let-7a-1 pull-down. We used the RP-SMS method to identify protein factors that bind specifically to pre-let-7a-1 (H) or non-specifically to beads alone (L). Results reveal that although most proteins identified do not bind specifically to pre-let-7a-1 (i.e., they have H/L ratio of less than 2), 81 proteins are enriched 2-fold or more in the pre-let-7a-1 pull-down.

Concluding remarks
The regulation of RNA processing by RBPs is at the functional heart of all cells. RBPs mediate and control gene expression by regulating virtually all steps of RNA metabolism [36,37]. Consequently, they contribute to cellular homeostasis, development and disease. Furthermore, hundreds of novel RBPs have been recently identified [38][39][40][41]. Many of these proteins have not been previously known for their RNAbinding properties and do not contain canonical RNA-binding domains. This highlights the need for methods that detect physiologically relevant RNA-protein complexes.
Many methods to identify protein binding to specific RNAs have recently been established [42]. We have previously developed a method based on RNA-coupled agarose beads and RNase-assisted elution of RBPs [43]. Furthermore, high-throughput screens for regulators of miRNA processing have revealed numerous sequence-specific factors that bind to miRNA precursors and primary transcripts [24,35,44]. The RNA pull-down combined with SILAC high-throughput mass spectrometry (RP-SMS) protocol described in this paper allows for versatile experimental design and high levels of sequence specificity for the quantitative detection of RBPs.