Molecular Mechanisms for CFIm-Mediated Regulation of mRNA Alternative Polyadenylation

Abstract

Alternative mRNA processing is a critical mechanism for proteome expansion and gene regulation in higher eukaryotes. The SR family proteins play important roles in splicing regulation. Intriguingly, mammalian genomes encode many poorly characterized SR-like proteins, including subunits of the mRNA 3'-processing factor CFIm, CFIm68 and CFIm59. Here we demonstrate that CFIm functions as an enhancer-dependent activator of mRNA 3' processing. CFIm regulates global alternative polyadenylation (APA) by specifically binding and activating enhancer-containing poly(A) sites (PASs). Importantly, the CFIm activator functions are mediated by the arginine-serine repeat (RS) domains of CFIm68/59, which bind specifically to an RS-like region in the CPSF subunit Fip1, and this interaction is inhibited by CFIm68/59 hyper-phosphorylation. The remarkable functional similarities between CFIm and SR proteins suggest that interactions between RS-like domains in regulatory and core factors may provide a common activation mechanism for mRNA 3' processing, splicing, and potentially other steps in RNA metabolism.

Keywords: RNA-binding proteins; SR proteins; alternative polyadenylation; cleavage; mRNA 3′ processing; polyadenylation; splicing.

Figure 1. UGUA is not an essential cis-element, but an enhancer for mRNA 3′ rocessing

(A) Enrichment score (log2(frequency in PAS/frequency in random sequence)) for UGUA and A(A/U)UAAA. (B) RNA substrates used in this study: L3 and p14/Robld3 PAS. (C) Compare L3 wild type (WT), Δ1, Δ1-2, Δ1-3UGUA mutant PASs using in vitro mRNA 3′ processing assays. Top panel: coupled cleavage/polyadenylation assay. Bottom panel: cleavage assay. Quantification results are shown below of the gel: % processed= (5′cleavage product)/(pre-mRNA). (D) Compare p14 WT, 1x and 2xUGUA mutant PASs using in vitro mRNA 3′ processing assays. Results are shown similar to (C).

Figure 2. The enhancer activity of UGUA is position dependent

(A and D) Diagrams to show the design of PAS RNAs. Two tandem copies of UGUA (A) or one copy (D) were inserted at different positions in tL3-Δ1-3. (B and E) In vitro cleavage/polyadenylation assay using L3 PAS with 2x or 1xUGUA inserted at different positions. (C and F) Quantification of the results shown in (B) and (F): mean ± s.e.m (n=3). (G) Design of the pPASPORT reporter. CMV: promoter; Rluc: renilla luciferase; PAS: poly(A) site to be tested; IRES: internal ribosomal entry site; Fluc: firefly luciferase. (H) PAS activity (Rluc/Fluc): mean ± s.e.m (n=3).

Figure 3. Enhancer-bound CFIm promotes the assembly of mRNA 3′ processing complex

(A) Gel mobility shift assays with L3 or p14-derived PASs. P: mRNA 3′ processing complex; H: heterogenous complex. (B) A diagram showing the RNA affinity purification procedure. MBP-MS2: a fusion protein between maltose binding protein and MS2. (C) The complexes assembled on the 3MS2-tagged L3 or p14-derived PASs were purified and analyzed by western blotting. The red arrows mark the CFIm subunits. (D) mRNA 3′ processing complex assembly on L3-derived PASs as shown in Fig. 2(A). (E) The mRNA 3′ processing complexes assembled on the 3MS2-tagged L3 derivatives as shown in Fig. 2(A) were purified and analyzed by western blotting. (F) Quantification of western blot signals in (E) using ImageJ.

Figure 4. The RS-like domain of CFIm68/59 is necessary and sufficient for activating mRNA 3′ processing

(A) A diagram of the tethering assay. RRM: RNA recognition motif; PRR: proline-rich region; RS: arginine-serine repeat region. (B–E) Tethering assay results obtained by co-expressing the L3-2xBoxB reporter and the proteins as labeled. The CFIm25 mutant L218R was labeled vertically. The results were plotted as mean ± s.e.m (n=3). PAS activities for tagged and untagged proteins were compared. L218R was compared to the wild type CFIm25. *** indicates that the p-values<0.001 (t-test). All samples were compared with the vector and *** indicates that the p-values<0.001 (t-test).

Figure 5. The CFIm68/59 RS-like domain binds to Fip1

(A) HeLa nuclear extract (NE) or recombinant CFIm25-68 or CFIm25-59 complexes purified from baculovirus-infected Sf9 insect cells with or without alkaline phosphatase (CIP) treatment, were resolved by Phos-tag gel and analyzed by western blotting. The red arrows point to the phosphorylated proteins and the green arrows dephosphorylated proteins. (B) CFIm59 and CFIm68 RS domain sequences. (C) GST pulldown assay with GST, GST-RS(CFIm59) or (CFIm68) (purified from Sf9 cells) and in vitro translated ³⁵S-labeled individual CPSF and CstF subunits. GST pulldown samples were resolved on SDS-PAGE and visualized by phosphorimaging (top panel). The same pulldown assay was performed with 6xHis-Fip1 expressed in Sf9 cells and pulldown samples were resolved on SDS-PAGE and analyzed by western blotting (lower panel). (D) A diagram of the Fip1 domain/regions. The Fip1-N and -C fragments were marked. Pulldown assays were similar to (C) with in vitro translated and ³⁵S-labeled Fip1-N and Fip1-C. (E) Top panel: the sequences of the Fip1-RD and -RA peptides. Lower panel: Fip1-RD and –RA pulldown with purified 6xHis-CFIm25 (E. coli), 6xHis-CFIm25-59 (Sf9), and 6xHis-CFIm25-68 complexes (Sf9) and the bound proteins were resolved on SDS-PAGE and analyzed by western blotting. Negative control: streptavidin beads (beads). (F) Top panel: GST-RS(CFIm59/68) purified from E. coli were mock treated (−) or treated (+) with SRPK1 and then used in pulldown assays with 6xHis-Fip1. The pulldown samples were analyzed by western blotting. Lower panel: GST-RS(CFIm59/68) purified from Sf9 cells were mock untreated (−) or treated (+) with CIP, and then used in pulldown assays with purified 6xHis-Fip1. (G) Nuclear extracts from control, CFIm59-KO, or CFIm68-KO HEK293T cell lines were used for IP with anti-CFIm25 antibody and the IP samples were analyzed by western blotting. The red arrows mark the CFIm59 or CFIm68 that are absent in KO cell lines.

Figure 6. Mechanism for CFIm-mediated APA regulation

(A) Scatter plots to show an APA comparison between control HEK293T cells (y axis) and CFIm68 or CFIm59 KO cells (x axis). Genes with significant APA changes (FDR<0.05 and at least 15% change) were highlighted: red dots represent genes with distal to proximal (DtoP) APA changes while blue dots proximal to distal (PtoD). (B) Poly(A) site sequencing (PAS-seq) and PAR-CLIP data for Vma21 and Ddx3x genes. The proximal and distal PASs were marked by dotted boxes and labeled on the top. (C) UGUA distribution at the proximal (dotted lines) and distal (solid lines) of CFIm25 or CFIm68 target (red) and non-target (green) genes. The UGUA distribution curves at proximal and distal PASs were compared. ***: p value<0.001; n.s.: not significant (K-S test). (D) CFIm68 PAR-CLIP signals at the proximal (dotted lines) and distal (solid lines) of CFIm25 or CFIm68 targets (red) and non-target (green) genes. (E) Gel mobility shift assays to characterize interactions between CFIm25-68 complex and the specificied PASs. Free RNAs and RNA-protein complexes are marked. (F) Vma21 APA profiles were measured by RT-qPCR with one primer set for the common (comm) region and another for the extended (ext) 3′ UTR region. The over-expressed proteins are marked on the x-axis.

Figure 7. A unified activation mechanism for mRNA 3′ processing and splicing

The solid red line with arrow indicates that CFIm helps to recruits CPSF through direct interactions and the dotted red line with arrow indicates that CFIm promotes CstF recruitment indirectly (A–B). The dotted grey lines indicate the lack of RE/D regions in the yeast Fip1 and Snp1 (C–D). UE: U-rich elements. CFIm25-68 is a dimer, but shown as a monomer due to space limitation (A). The blue arrows represent cleavage and the widths of the arrows represent the frequencies of PAS usage.