SRSF7 and SRSF3 depend on RNA sequencing motifs and secondary structures to regulate Microprocessor

Protein expression and purification

The coding sequences for SRSF1, SRSF2, SRSF3, SRSF5, SRSF7, SRSF8, and SRSF9 proteins in humans were amplified by PCR using cDNA obtained from total RNAs of human HCT116 cells. The PCR-amplified DNAs were inserted into human expression plasmid, pXG, used in previous studies (Nguyen et al, 2015). As a result, SRSF proteins were fused with GFP and 10×His-tag at their C-termini. The information of the primers used in the cloning experiments is presented in Table S1.

The seven SRSF proteins were expressed and purified in a similar way. pXG-SRSF plasmids were transfected into 50 of 100-mm dishes of HEK293E cells. The transfected cells were grown for 3 d and then collected by centrifuge. The collected cell pellets were resuspended in T500 buffer containing 20 mM Tris–HCl (pH 7.5), 500 mM NaCl, and 4 mM β-mercaptoethanol, with the addition of 2 µg/ml RNase A and protease inhibitors. The resuspended cells were sonicated and centrifuged at high speed (48,000g) for 20 min; thus, the clear supernatant (cell lysate) was obtained. The cell lysate was bound with 3 ml of Ni-NTA beads (pre-equilibrated with 30 ml of T500). The beads were then washed with 30 ml of T500 plus 40 mM imidazole and added with 10 ml of T500 containing 200 mM imidazole to elute SRSF proteins. The eluted protein solution was added with T0 containing 20 mM Tris–HCl (pH 7.5), 0 mM NaCl, and 4 mM β-mercaptoethanol to get the final NaCl concentration of 100 mM. The diluted protein solution was loaded onto 0.5 ml of Q-Sepharose beads, which were subsequently washed with 30 ml of T150 containing 20 mM Tris–HCl (pH 7.5), 150 mM NaCl, and 2 mM DTT. The SRSF proteins were eluted from Q-Sepharose beads with 5 ml of T500 containing 20 mM Tris–HCl (pH 7.5), 500 mM NaCl, and 2 mM DTT. The purified proteins were quantified, examined in SDS–PAGE, and stored at −80°C until being used. The eluted SRSF7 protein from Q-Sepharose step was concentrated using the Centricon (30-kD cutoff) and then loaded on the gel filtration column (a Superdex 200 10/300 Gl column). The column was run using the Bio-Rad NGC Chromatography Systems in T500 with 10% glycerol added. The peaked fractions from gel filtration were collected, quality-checked by SDS–PAGE, quantity-measured by Bradford assays, and eventually stored at −80°C until being used.

To obtain the human MP (NLSD3-DGCR8) complex and DROSHA (D3-G2), we used the similar protocols described in our previous studies (Nguyen et al, 2015, 2020; Li et al, 2020). NLS was the nuclear localization sequence, containing seven amino acids (PKKKRKV). D3 was the DROSHA fragment, containing amino acids 390–1,365. G2 was the DGCR8 fragment containing amino acids 701–773. In short, pXab-NLSD3 and pXG-DGCR8 plasmids, or pXab-D3 and pXG-G2 plasmids, were cotransfected in the HEK293E cells. The NLSD3-DGCR8 or D3-G2 complex was purified by Ni-NTA beads and Q-Sepharose beads. The D3-G2 complex was further purified using NGC Chromatography Systems.

Pre-miRNA synthesis

We synthesized pre-miR-142_CL-2 and CL0 as described in our previous study (Le et al, 2022). In short, we first generated the double-stranded DNA (dsDNA) containing the T7 promoter, the hammerhead (HH) ribozyme, a pre-miRNA sequence, and the hepatitis delta virus (HDV) ribozyme. Two oligos containing a part of HH ribozyme and a pre-miRNA sequence were annealed and then extended by Klenow Fragment (EP0421; Thermo Fisher Scientific) at 37°C for 2 h to make dsDNA. This dsDNA was amplified by the first PCR using two primers: one contained T7-HH ribozyme, whereas the other contained pre-miRNA and the 5′-half of HDV ribozyme sequence. Next, three other oligos were used to synthesize the dsDNA containing the HDV sequence, called HDV-dsDNA. This HDV-dsDNA and the dsDNA amplified in the first PCR were overlapped and amplified in the second PCR using the T7 primer and the HDV reverse primer to generate in vitro transcription (IVT)-dsDNA. Phusion Hot Start II DNA Polymerase (F549L; Thermo Fisher Scientific) was used for all PCRs, and the oligos were obtained from GENEWIZ and BGI Group. The PCR primer and template information are presented in Table S2.

Next, the IVTs were conducted with 200 ng of each dsDNA template using MEGAscript T7 Transcription Kit (AM1334; Invitrogen). The self-cleavage activity of the HH and HDV ribozymes released the pre-miRNAs, which contained 5′-OH and 2′-3′-cyclic phosphate. After being gel-purified, the 2′-3′-cyclic phosphate and 5′-OH were converted into 3′-OH and 5′-phosphate, respectively, by T4 polynucleotide kinase (EK0032; Thermo Fisher Scientific). Finally, the pre-miRNAs were purified using isopropanol and stored at −80°C until further use.

In vitro pri-miRNA cleavage assays for individual human pri-miRNAs

We synthesized the dsDNA containing the T7 promoter and DNA region coding for each pri-miRNA by PCR. The PCR primer and template information are presented in Table S2. The sequences of these synthesized dsDNAs were confirmed by Sanger sequencing. Next, ∼200 ng of each dsDNA was added in the IVT reactions containing the MEGAscript T7 Transcription Kit (AM1334; Invitrogen). The resulting RNAs were gel-purified, quantified, and finally stored at −80°C.

To conduct the pri-miRNA cleavage assays, 2 pmol of each pri-miRNA were incubated with the enzymes at 37°C in 10 µl reaction buffer containing 50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM MgCl₂, 1 mM DTT, 0.2 µg/μl BSA, and 10% glycerol. The amounts of enzyme were indicated in the figure and figure legends for each experiment. After 2 h, we treated the mixture with 20 µg of Proteinase K (Cat: 25530049; Invitrogen) for 15 min at 37°C and then 15 min at 50°C to stop the enzyme reaction. Finally, the mixtures were heated at 95°C for 5 min and immediately chilled on ice before being analyzed by a pre-run 12% urea-PAGE. The gel was run at 300 V for 90 min and was stained by SYBR Green II RNA Gel Stain (S7564; Invitrogen) for 5 min. The gel images were captured using the Bio-Rad Gel Doc XR+ system, and RNA band intensities were estimated using Image Lab v6.0.1. The relative pri-miRNA cleavage efficiency of MP or DROSHA was calculated as a ratio of the band density of cleaved products (F2) to that of the pri-miRNA substrate.

The electrophoretic mobility shift assay (EMSA)

2 pmol of pri-miR-142, 20 pmol DROSHA, and 26 pmol of SRSF7 were mixed in 12 µl of EMSA reaction buffer containing 50 mM Tris-HCl (pH 7.5), 125 mM NaCl, 10% glycerol, 0.2 mg/ml BSA, 1 mM DTT, and 2 mM EDTA. Next, the EMSA reactions were incubated on ice for 1 h, and then, 5 µl of each reaction mixture was loaded onto a 4% native PAGE. The gel was run in the electrophoresis apparatus for 1 h 20 min at 4°C in a pre-cooled 1× TBE buffer. After completion of electrophoresis, the gel was stained with ethidium bromide for 5 min. The images of stained gels were taken using the Bio-Rad Gel Doc XR + system.

The high-throughput pri-miRNA cleavage assay (HT-priCA)

We synthesized randomized pri-miR-142 and pri-miR-191 by the IVT. Each pri-miRNA possessed three randomized groups, containing four randomized base pairs in its CNNC position (group 1), and 4 nt upstream (group 2) or downstream (group 3) of its CNNC positions. Six groups of two randomized pri-miRNAs were synthesized similarly as described in the following for group 1 of pri-miR-142. First, two single-stranded DNAs, F-142lib (AAGAGTACACGCCGACGGACAGACAGACAGTGCAGTC) and R-142lib_1 (GGCAGCAGTGGCGTGATCTCCNNNNCCCACAGTACACTCATCCATAAAG), were used to amplify the dsDNA template of pri-miR-142 that was used in our previous study (Li et al, 2020) using Phusion Hot Start II DNA Polymerase (F549L; Thermo Fisher Scientific). The randomized nt were added in the R-142lib_1 primer. Next, the resulting dsDNA was amplified in the second PCR with F-T7-142lib primer (TAATACGACTCACTATA GGGAAGAGTACACGCCGACGGACAG) and R-142-RA3-1 (TTGGCACCCGAGAATTCCAGGCAGCAGTGGCGTGATCTCC). As a result, the two-round PCR-amplified dsDNA contained T7 promoter, pri-miR-142-group1–coding DNA, and RA3 sequence. These dsDNAs were subjected to the IVT reactions to produce the randomized pri-miRNAs. The IVT-synthesized RNAs were gel-purified, quality-checked by gel staining, quantity-checked by Nanodrop, and finally stored at −80°C until use. The PCR primer and template information are presented in Table S3.

In HT-priCA, 3 pmol of each randomized pri-miRNA group were incubated with 5 pmol of the MP with or without 10 pmol of SRSF3 or SRSF7 at 37°C for 2 h, in 10 µl of the standard reaction buffer as mentioned above. After the reaction, the cleaved RNA products were purified using phenol extraction followed by isopropanol precipitation. The purified RNAs were ligated with the RA5-6N adapter (GUU CAG AGU UCU ACA GUC CGA CGA UCN NNN NN), and then, the resulting RA5-6N–ligated RNAs were reverse-transcribed using R-RA3 primer and SuperScript IV Reverse Transcriptase (18090010; Invitrogen) to generate cDNAs. The cDNAs were finally amplified using PCR with the sequencing primers into DNA libraries of the cleaved products. The pri-miRNA substrates were cloned using a circularization method. The pri-miRNAs were first reverse-transcribed using SuperScript IV Reverse Transcriptase and cirRTP primer (5′Phos/NNN NNN GAT CGT CGG ACT GTA GAA CTC TGA AC/iSp18/CCT TGG CAC CCG AGA ATT CCA). The resulting cDNAs were circularized by CircLigase single-stranded DNA ligase (CL4115K; Lucigen). The circularized cDNAs were eventually amplified by PCR using the sequencing primers into DNA libraries of the pri-miRNA substrates. The DNA libraries were sequenced by NovaSeq 6000 sequencer, and the sequencing data were deposited in Gene Expression Omnibus (accession ID: GSE215790).

High-throughput pri-miRNA cleavage library analysis

We removed adapters using cutadapt (-a TGGAATTCTCGGGTGCCAAGG-A GATCGTCGGACTGTAGAACTCTGAAC-m 10) (Martin, 2011). Then, we joined two paired-end reads using fastq-join (Aronesty, 2013). The low-quality and duplicated reads were filtered out using fastq_quality_filter (-q 20 -p 90) and fastx_collapser, respectively (http://hannonlab.cshl.edu/fastx_toolkit/index.html). The 6 and 4-nt randomized barcodes were discarded from the cleaved products and OS, respectively. We mapped the resulting reads to the built-in sequence references containing all 1,532 pri-miRNA variants using BWA (Li & Durbin, 2010). Only unique mapped reads were collected. The raw read counts were normalized as count per million.

The basal canonical cleavage sites of MP in pri-miR-142 and pri-miR-191 were named as CL according to the 3′-ends of 3p miRNAs (miRBase.org, Kozomara et al, 2019). The cleavage sites upstream and downstream of CL0 were labeled as CL-1 and CL1, respectively. The cleavage sites of MP in HT-priCA were determined by the 5′-ends of the cleaved products, F3 fragments. The cleavage site of MP was CLx when the 5′-end of F3 fragments was mapped to the position x-1 in pri-miRNA. The basal cleavage sites ranged from CL-3 to CL3. The apical cleavage sites were in between CL8 and CL14.

For each variant, we obtained the normalized counts of the original substrate (N_S), all cleaved products (N_P) derived from this substrate, and the normalized counts of a cleaved product (N_X) cleaved at a cleavage site, CLx, in which x is ranging from −3 to 3.

The global cleavage efficiency score of a variant (E_P) was the cleavage efficiency of the MP estimated for all the cleaved products resulting from this variant and was calculated using the following equation: E_P = log₂(N_P + 0.1) − log₂(N_S + 0.1). A pseudocount of 0.1 was added.

The local cleavage efficiency score of a variant at the cleavage site of CLx (E_X) was estimated for only the cleaved product resulting from CLx cleavage and was calculated using the following equation: E_X = log₂(N_X + 0.1) − log₂(N_S + 0.1). A pseudocount of 0.1 was added.

The cleavage accuracy score of the cleavage site CLx (A_X) was calculated for each variant as follows: A_X = N_X/N_P.

The E_P, E_X, and A_X values were averaged using three scores from the three repeats for each cleaved product sample.

The relative cleavage accuracy and global and local cleavage efficiency scores for the MP + cofactor sample were calculated for each variant as follows:ΔAX[cofactor] =AX[MP+cofactor] – AX[MP],ΔEP[cofactor]=EP[MP+cofactor] – EP[MP],ΔEX[cofactor]=EX[MP+cofactor] – EX[MP] .

The structures of pri-miRNA variants were predicted using RNAfold (Lorenz et al, 2011).

Human pri-miRNA cleavage library analysis

We downloaded the cleavage efficiency and homogeneity scores of 1,881 pri-miRNAs cleaved by MP alone or MP and SRSF3 (Kim et al, 2021). In the original study, cleavage efficiency scores were calculated as the ratio of the productive cleaved products (±3 nt from the miRBase annotations of 5′-end of 5p miRNAs or 3′-end of 3p miRNAs) to the OS. The cleavage homogeneity scores were calculated as the maximum cleavage ratio of all productive sites. We selected 788 pri-miRNAs cleaved by MP and SRSF3 with cleavage efficiency and homogeneity scores higher than 1 and 0.1, respectively. The CNNC and 17-CRC motifs were detected from positions −15 to −22 and positions −17 to −19 on the 3′-strand of these 788 pri-miRNAs, respectively.

Human pri-miRNA structure analysis

We extracted 504 pri-miRNA sequences from the human genome reference (hg38; https://genome.ucsc.edu/) using the genome annotation from MirGeneDB v2.0 (Fromm et al, 2020). Each pri-miRNA sequence includes its pre-miRNA and 40-nt extensions at both ends of pre-miRNA. The structure of each pri-miRNA was predicted using RNAfold (Lorenz et al, 2011). We found 332/504 pri-miRNAs containing the CNNC motif from positions −15 to −22 on their 3′-strand. The CNNC structures of these 332 pri-miRNAs were examined.

SHAPE-MaP library construction

RNAs were incubated in a buffer containing 50 mM Tris–HCl (pH 7.5), 100 mM NaCl, and 2 mM MgCl₂ at 37°C for 20 min. Next, 1M7 (1-methyl-7-nitroisatoic anhydride) (908401; Sigma-Aldrich) was added to the concentration of 10 mM and incubated for 75 s at 37°C to modify RNAs. The control samples were prepared in parallel, where DMSO was added instead of 1M7. After modification, RNAs were precipitated by isopropanol and washed with ethanol before being ligated to the 3′ adapter (6N-RA3). Next, the reverse transcription was performed for 3 h at 42°C by SuperScript II (18064014; Invitrogen) using R-RA3 primer in the buffer containing 0.5 mM dNTPs, 50 mM Tris–HCl (pH 8.0), 75 mM KCl, 6 mM MnCl₂, and 10 mM DTT. The resulting cDNAs were later amplified two times, the first time by R-RA3 and the pri-miR-142 sequence-specific forward primers, and the second time by two sequencing primers, using Phusion Hot Start II DNA Polymerase. The final DNA products were subjected to NGS using the Illumina NovaSeq 6000 sequencer. The primer information for pri-miRNA cloning in SHAPE-MaP experiments is presented in Table S4. The sequencing data of SHAPE-MaP experiments were deposited in Gene Expression Omnibus (accession ID: GSE221996).

SHAPE-MaP analysis

We removed adapters, joined two paired-end reads, and removed low-quality and duplicated reads as described for the HT-priCA. Then, the 6-nt randomized barcodes in the 3′-end of the remaining reads were discarded. Next, ShapeMapper2 software was used to identify RNA structures of pri-miR-142 variants (shapemapper --min-seq-depth 1,000) (Busan & Weeks, 2018). The resulting SHAPE reactivity data from ShapeMapper2 were applied to detect RNA structures using the RNAstructure package (Reuter & Mathews, 2010).

iLIME-seq for pre-miRNA

1 µg of pcDNA3-CNNC-pri-miR-142 or pcDNA3-non-CNNC-pri-miR-142 plasmid were transfected into one well in a six-well plate of HCT16-DICER KO cells by Lipofectamine 3000 Transfection Reagent (L3000001; Invitrogen). The primer information for cloning pCDNA3-pri-miR-142 plasmids is presented in Table S1. The cells were harvested after 2 d. In another experiment, 1.5 µg of pXG, pXG-SRSF7, or pXG-SRSF3 were cotransfected with 1 µg of pcDNA3-CNNC-pri-miR-142 into one well in the six-well plate of HCT116-DICER KO cells. The cells were also harvested 2 d later. All the total RNAs were extracted using TRIzol Reagent (15596018; Invitrogen). We used iLIME-seq to clone and sequence pre-miR-142 from these total RNAs (Le et al, 2022). In short, 2.5 µg of total RNA were incubated with 16 units of T4 RNA ligase 1 (M0437M; NEB) in the ligation buffer containing 50 mM Tris–HCl (pH 7.5), 10 mM MgCl₂, 1 mM DTT, 16 units of SUPERase•In RNase inhibitor (AM2694; Invitrogen), and 20% PEG 8000. The mixtures were incubated at 25°C for 2 h. The ligated RNA products were then subjected to reverse transcription reactions using SuperScript IV (Reverse Transcriptase) at 50°C for 20 min. The resulting cDNAs were amplified the first time by pre-miR-142 sequence-specific primers and the second time by sequencing primers, both by Phusion Hot Start II DNA Polymerase (F549L; Thermo Fisher Scientific). The final DNA products from the second PCR were subjected to NGS using the Illumina NovaSeq 6000. The sequencing data were deposited in Gene Expression Omnibus (accession ID: GSE215791). The primer information for pre-miR-142 iLIME experiments is presented in Table S5.

We processed sequencing reads and mapped reads to build-in reference for iLIME-seq samples as described in the previous study (Le et al, 2022).