Post-transcriptional gene silencing mediated by microRNAs is controlled by nucleoplasmic Sfpq

Abstract

There is a growing body of evidence about the presence and the activity of the miRISC in the nucleus of mammalian cells. Here, we show by quantitative proteomic analysis that Ago2 interacts with the nucleoplasmic protein Sfpq in an RNA-dependent fashion. By a combination of HITS-CLIP and transcriptomic analyses, we demonstrate that Sfpq directly controls the miRNA targeting of a subset of binding sites by local binding. Sfpq modulates miRNA targeting in both nucleoplasm and cytoplasm, indicating a nucleoplasmic commitment of Sfpq-target mRNAs that globally influences miRNA modes of action. Mechanistically, Sfpq binds to a sizeable set of long 3'UTRs forming aggregates to optimize miRNA positioning/recruitment at selected binding sites, including let-7a binding to Lin28A 3'UTR. Our results extend the miRNA-mediated post-transcriptional gene silencing into the nucleoplasm and indicate that an Sfpq-dependent strategy for controlling miRNA activity takes place in cells, contributing to the complexity of miRNA-dependent gene expression control.

Fig. 1

Sfpq, Pspc1, and NonO are components of Ago2 complex and interact with let-7a. a Overview of the proteomic method used to identify RNA-dependent proteins interacting with Ago2. b Silver staining of an SDS-PAGE analysis from the IP with the anti-Ago2 antibody, the protein G conjugated with the Dynabeads (DB), or the anti-Ago2 antibody alone incubated without cell lysate. The samples were untreated, treated with 10 μg ml⁻¹ RNase A for 30 min at room temperature (for partial digestion), or treated with 10 mg ml⁻¹ RNase A for 30 min at room temperature (for total digestion). c Radioactive images of a TBE-Urea gel showing signal from ³²P-labeled RNA fragments of samples untreated, treated with 10 μg ml⁻¹ RNase A for 30 min at room temperature (for partial digestion), or treated with 10 mg ml⁻¹ RNase A for 30 min at room temperature (for total digestion). d Scatter-plot of the log base 2 (−/+) RNase A ratios (abundance scores) plotted with the unique peptides for each identified protein. Each spot is a different protein. e Co-IP of endogenous Sfpq and Ago2, Pspc1, or NonO in RAW 264.7 and HEK293T cells. When indicated, cell lysates were incubated at room temperature with RNase A (10 mg ml⁻¹) for 30 min. f Sfpq and Pspc1 interact with mature let-7a but not with the precursor. RAW 264.7 cell extracts were immunoprecipitated with two different antibodies for each indicated protein. The RNA was purified and analyzed by Northern blotting

Fig. 2

Sfpq mediates the interaction between miRISC and Pspc1 or NonO. a Co-IP of endogenous Ago2 with Sfpq, Pspc1, or NonO from P19 cells. Cells were transfected with the indicated siRNAs and analyzed by Western blotting. Ago2 and tubulin served as controls. b RNA-IP of let-7a with the endogenous Sfpq, Pspc1, or NonO. RAW 264.7 cells were transfected with the indicated siRNAs. Cell extracts were immunoprecipitated with the indicated antibodies and the RNA was purified and analyzed by Northern blotting

Fig. 3

Sfpq, Pspc1, and NonO interact with Ago2 in the nucleoplasm. a Immunoblot analysis of chromatin, nucleoplasm, or cytoplasm from HeLa, HEK293T, P19, or RA-treated P19 cells using antibodies directed to the indicated proteins. b Co-immunofluorescence analysis of RAW 264.7 cells stained with the indicated antibodies. Scale bar corresponds to 10 μm. c Co-IP of endogenous Ago2 and the indicated proteins from nucleoplasmic or cytoplasmic extracts from RAW 264.7 or P19 cells

Fig. 4

Sfpq promotes miRNA targeting at selected binding sites. a Venn diagram of Ago2 or Sfpq RNA-IP-enriched miRNAs found by small RNA sequencing analysis. b Let-7a expression levels (upper panel) and RNA-IP of Ago2 and let-7a (lower panel) in control or siSfpq-transfected RAW 264.7 cells. RNA extracts were analyzed by RT-qPCR. Data are presented as the mean ± s.e.m. (n = 6) and normalized to U2 snRNA or the input, respectively. c Sfpq, Pspc1, and NonO interact with mature let-7a in RA-treated P19 cells. Cell extracts were immunoprecipitated with the indicated antibodies and RNA was purified and analyzed by Northern blotting. Genomic distribution of either the d Ago2-let-7a or e endogenous Ago2-miRNA peaks decreased upon Sfpq knockdown, according to the distance to Sfpq peaks by HITS-CLIP analysis. These data show the prevalence of the Sfpq-dependent Ago2 peaks in the 3′UTR when Sfpq binds closely (<500 nt). RNA-IP of Ago2 and the indicated 3′UTRs for either f let-7a or g the endogenously expressed miR-302b-binding sites. P19 cells were transfected with the indicated molecules. Nucleoplasm and cytoplasm fractions were separated. The indicated cell lysates were incubated with either 100 nM full-length recombinant wild-type Sfpq or the Sfpq-214–598 quadruple mutant (L535A, L539A, L546A, and M549A) for 30 min at room temperature. Before IP, the lysate was partially digested with 10 μg ml⁻¹ RNase A for 30 min at room temperature. RNA was purified from the immunocomplexes and from 5% of the input and analyzed by RT-qPCR using oligonucleotide probes surrounding the miRNA-binding sites identified by HITS-CLIP. Data are presented as the mean ± s.e.m. (n = 3) and normalized to their own inputs. Student’s t-test (for b) or one-way ANOVA followed by Tukey’s post hoc test (for f, g) with *p < 0.05 and **p < 0.01. ns not significant, TTS transcription termination site

Fig. 5

Sfpq promotes post-transcriptional silencing mediated by miRNAs in both nucleoplasm and cytoplasm. a Gene expression differences in let-7a-transfected P19 cells upon Sfpq knockdown and control. Differential expression is plotted for mRNAs containing canonical let-7a-binding sites (BS) localized in the 3′UTR within a close distance to Sfpq peaks (<500 nt) and reduced upon Sfpq knockdown. b P19 cells were transfected with the indicated molecules. Nucleoplasm and cytoplasm were separated to measure the expression levels of the indicated let-7a-target mRNAs. RNA was purified and analyzed by RT-qPCR. Data are normalized with U2 snRNA and presented as the mean ± s.e.m. (n = 3). c Immunoblot analysis of Lin28A and tubulin in P19 cells transfected with let-7a and/or siSfpq. d Quantitative RT-PCR analysis of the half-life of Lin28A transcript in let-7a-transfected P19 cells compared to control (upper panels) or siSfpq + let-7a-transfected cells compared to siSfpq control (lower panels). Total RNA from either nucleoplasm (left panels) or cytoplasm (right panels) was isolated at the indicated times after addition of actinomycin D. Data are normalized with β2-microglobulin and presented as the mean ± s.e.m. (n = 6). e Relative luciferase activity of reporter constructs containing the mouse Lin28A 3′UTR sequence in HEK293T cells transfected with let-7a and siSfpq as indicated. The data were normalized using Renilla activity and presented as the mean ± s.e.m. (n = 4). One-way ANOVA followed by Tukey’s post hoc test: *p < 0.05, **p < 0.01, ns not significant

Fig. 6

Sfpq aggregates onto long 3′UTRs to modulate the accessibility of miRNA-binding sites. a Genomic distribution of Sfpq peaks by HITS-CLIP analysis. b Scatter plot of length and frequency of Sfpq peaks in different genomic regions. c Distribution of the 3′UTR length of the indicated subpopulation of transcripts. d Result of the bioinformatic analysis for de novo search of Sfpq-binding motif from Sfpq HITS-CLIP data. e UV-crosslinking assay to analyze the interaction of the recombinant Sfpq (100 nM) with the ³²P-labeled RNA oligonucleotides containing two copies of the CUGU or CUGUA, respectively, but not to the CCCCG negative control sequence. f Top view of two topographic AFM images of typical Sfpq-Lin28A 3′UTR complexes. The color bar on the right represents the height scale with a maximum corresponding to 5 nm. Scale bar corresponds to 20 nm. g Spatial correlation of the recruited Ago2-let-7a peaks compared to the decreased ones. In the x-axis is reported the distance in nt between the reduced and the new recruited peaks on the 3′UTR, whereas in the y-axis is reported the percentage of recruited peaks. The black line is the data density, the red bars indicate that the number of recruited peaks is higher than that expected by chance, the blue bars indicate the opposite, whereas the white bars indicate no recruitment

Fig. 7

A model for the Sfpq-dependent control of miRNA targeting