A U2-snRNP–independent role of SF3b in promoting mRNA export

Significance

To ensure efficient and accurate gene expression, pre-mRNA processing and mRNA export need to be balanced. SF3b is a core component of U2 snRNP that participates in splicing and 3′ processing of pre-mRNAs. Here, we uncovered a role of SF3b in promoting mRNA export by directly binding with the mature mRNA and recruiting the key mRNA export adaptor THO. Intriguingly, this role is not dependent on U2 snRNP, but is rather enhanced upon U2 snRNP disruption as a result of enhanced SF3b–THO interaction and recruitment to the mRNA. Together, our work uncovers an important role of SF3b in promoting mRNA export and suggests that competition between U2 snRNP and the mature mRNP for SF3b contributes to balancing pre-mRNA processing and mRNA export.

Abstract

To ensure efficient and accurate gene expression, pre-mRNA processing and mRNA export need to be balanced. However, how this balance is ensured remains largely unclear. Here, we found that SF3b, a component of U2 snRNP that participates in splicing and 3′ processing of pre-mRNAs, interacts with the key mRNA export adaptor THO in vivo and in vitro. Depletion of SF3b reduces THO binding with the mRNA and causes nuclear mRNA retention. Consistently, introducing SF3b binding sites into the mRNA enhances THO recruitment and nuclear export in a dose-dependent manner. These data demonstrate a role of SF3b in promoting mRNA export. In support of this role, SF3b binds with mature mRNAs in the cells. Intriguingly, disruption of U2 snRNP by using a U2 antisense morpholino oligonucleotide does not inhibit, but promotes, the role of SF3b in mRNA export as a result of enhanced SF3b–THO interaction and THO recruitment to the mRNA. Together, our study uncovers a U2-snRNP–independent role of SF3b in mRNA export and suggests that SF3b contributes to balancing pre-mRNA processing and mRNA export.

In eukaryotes, the nascent pre-mRNA transcripts undergo multiple processing steps in the nucleus before mRNAs are exported to the cytoplasm for translation. Accumulating evidence suggests that pre-mRNA processing and mRNA export need to be balanced to ensure efficient and accurate gene expression (1–7). When splicing factors are limited or when mRNA export factors are present in excess, even unspliced pre-mRNAs are leaked to the cytoplasm (1–3). On the contrary, down-regulation of nuclear export factors results in nuclear retention of fully processed mRNAs that are ultimately subject to degradation (4–7). Thus, maintenance of the balance between pre-mRNA processing and mRNA export is of significant importance. However, how this balance is achieved remains largely unknown.

U2 snRNP is a core component of the spliceosome. It is comprised of U2 snRNA, multisubunit SF3a and SF3b complexes, U2-snRNP–specific proteins A′ and B″, as well as the seven Sm proteins common to the spliceosomal snRNPs (8–10). During splicing, SF3b proteins contact the pre-mRNA at and near the branch site (BS) in a sequence-independent manner, stabilizing the U2 snRNA/BS interaction (11–18), and thereby regulate BS recognition and selection. Except for splicing, U2 snRNP also functions in the 3′ processing of polyadenylated and nonpolyadenylated mRNAs (19, 20). On polyadenylated pre-mRNAs, U2 snRNP components including SF3b interact with the cleavage and polyadenylation specificity factor and enhance the rate of 3′ processing (19). On nonpolyadenylated histone pre-mRNAs, SF3b155, the largest SF3b subunit, together with Prp43, directly makes contact with the 7-nt motif, C/GAAGAAG, present in the coding region and facilitates 3′ processing as a component of U2 snRNP (20). To date, all known roles of SF3b are executed in the context of U2 snRNP, and whether it has a U2-snRNP–independent role remains unknown.

The highly conserved TREX complex plays key roles in mRNA export (21, 22). It mainly contains three parts: the THO subcomplex (THOC1/2/3/5/6/7) and proteins ALYREF and UAP56 (21, 23–26). ALYREF and THO serve as mRNA export adaptors that link the mRNA to the export receptor, a heterodimer of NXF1/NXT1 (27–29). In higher eukaryotes, splicing significantly promotes TREX recruitment and mRNA export (23, 24, 30). It has been thought that physical interaction(s) between splicing and mRNA export machineries mediate splicing-dependent TREX recruitment. However, these interactions remain to be identified.

In this work, we found that SF3b interacts with the THO complex in vivo and in vitro. Loss-of-function and gain-of-function data demonstrate a role of SF3b in promoting mRNA export by recruiting THO/TREX. In support of this role, SF3b binds intronless and spliced mRNAs in the cells. Intriguingly, disruption of U2 snRNP formation does not inhibit, but promotes, SF3b-mediated mRNA export by enhancing the SF3b–THO interaction and boosting THO recruitment to the mature mRNA. Thus, our work identifies SF3b as a linker between splicing and mRNA export machineries and suggests that the balance between pre-mRNA processing and mRNA export can be regulated through alteration of SF3b distribution in U2 snRNPs and mature mRNPs.

Results

SF3b Associates with THO in Vivo, and This Association Is Strengthened in the Absence of RNAs.

With the interest to understand the mechanism for splicing-dependent mRNA export, we examined the interactions of TREX components with splicing factors. In an attempt of THOC2 immunoprecipitation (IP) from RNase A-treated HeLa nuclear extract (NE), SF3b components, including SF3b155 and SF3b130, were specifically present in the THOC2 immunoprecipitate but not the control (Cntl; i.e., IgG) immunoprecipitate (Fig. 1A and Dataset S1). RRP6, a component of the exosome that is not known to interact with THO or SF3b, was not detected in either immunoprecipitate (Fig. 1A). Considering that THO and SF3b are tightly bound complexes (15, 21, 23, 26, 31), this result suggests that THO might interact with SF3b. To further investigate this interaction, we carried out reverse IPs by using an SF3b155 antibody from HeLa NE treated or not treated with RNase A, followed by MS. An antibody against all subunits of SF3a, another U2 snRNP component, was also included. As expected, SF3a and SF3b subunits, as well as the two other U2 snRNP components, A′ and B″, were present in both immunoprecipitates, and RNase A treatment weakened their coprecipitations, as indicated by reduced peptide enrichment (Fig. 1B and Dataset S2). In support of the SF3b–THO interaction, several THO proteins were present in the immunoprecipitate of SF3b155 but not that of SF3a. Intriguingly, the amounts of all these proteins apparently increased upon RNase A treatment (Fig. 1B).

Fig. 1.

SF3b interacts with THO in vivo and in vitro. (A) The THOC2 antibody coprecipitated SF3b. Western blot analysis to detect components of SF3b (SF3b155, SF3b130) and THO (THOC2, THOC5) complexes immunoprecipitated by the Cntl (i.e., IgG) and THOC2 antibodies. RRP6 serves as a negative control. (B) MS to detect proteins associated with SF3a and SF3b155. Total peptide counts of components of U2 snRNP and THO in the immunoprecipitates of SF3a and SF3b155 in the presence or absence of RNase A are shown. (C) The SF3b155 antibody coprecipitated THO. Western blot analysis to detect components of SF3b (SF3b155, SF3b130), SF3a (SF3a120, SF3a66, SF3a60), and THO (THOC2, THOC1, THOC5) complexes immunoprecipitated by the Cntl (i.e., IgG) and SF3b155 antibodies with or without RNase A treatment. RRP6 serves as a negative control. The numbers shows the ratio of each protein coimmunoprecipitated from NE in the presence (+) or absence (−) of RNase A. (D) Strep-tagged THO pulled down the SF3b complex. Inputs and proteins pulled down were detected by Coomassie staining and Western blotting. Strep-tagged MBP was used as a negative control. The white line delineates the boundary where the irrelevant lane has been removed from the same gel and blots. (E) Strep-tagged THO pulled down the SF3b core. Same as D, except that, instead of the whole SF3b complex, the SF3b core was used.

To validate the MS data, SF3b155 and Cntl IPs were repeated in the absence or presence of RNase A, followed by Western analysis using antibodies to subunits of SF3b (SF3b155 and SF3b130), SF3a (SF3a120, SF3a66, and SF3a60), and THO (THOC1, THOC2, and THOC5). The RRP6 antibody was used as a negative control. As shown in Fig. 1C, all of these proteins except RRP6 were apparently precipitated by the SF3b155 antibody, but not the Cntl. Consistent with the MS data, RNase A treatment resulted in reduced association of SF3b155 with SF3a subunits and concomitantly enhanced association with THO subunits (Fig. 1C). These data indicate that SF3b associates with THO and that this association is strengthened in the absence of RNAs.

SF3b Directly Interacts with THO via the Core Proteins.

We next asked whether SF3b interacts with THO in vitro. To this end, we expressed and purified THO and SF3b complexes in insect cells. In the THO complex, THOC1 was strep-tagged and used for pull-downs. Strep-tagged MBP was used as a negative control. Significantly, Western blotting showed that SF3b155 and SF3b130 were apparently pulled down by the THO complex but not by MBP, indicating that THO and SF3b interact in vitro (Fig. 1D). To further validate this interaction and narrow down the THO-interacting SF3b components, we next used the SF3b core (32), which is composed of SF3b155 (residues 454–1304), SF3b130, SF3b14b, and SF3b10, for pull-downs. Because the SF3b155 antibody recognizes two N-terminal fragments (residues 330–344, 450–463) that are completely or partially deleted in the SF3b core, we used only the SF3b130 antibody for Western blot analysis. As shown in Fig. 1E, SF3b130 was more apparently detected in the pull-down of strep-tagged THO even though significantly more strep-tagged MBP was pulled down. Together, these data indicate that SF3b directly interacts with THO via the core proteins.

Depletion of SF3b Results in Nuclear Retention of Bulk polyA RNAs and Individual Intronless mRNAs.

Considering the key roles of THO in mRNA export (21, 23, 26), we speculated that SF3b might be involved in this process as well. To test this, we first examined how the nucleocytoplasmic (N/C) distribution of polyA RNAs was affected by knockdown (KD) of SF3b subunits. Note that SF3b155 KD caused co-KD of other SF3b subunits, e.g., SF3b145 and SF3b130 (Fig. 2A). Significantly, treatment of the cells with SF3b155 or SF3b145 siRNA resulted in apparent nuclear retention of polyA RNAs (Fig. 2A and SI Appendix, Fig. S1 A and B). Given the critical role of SF3b in splicing, it is possible that the nuclear-retained polyA RNAs are mostly unspliced pre-mRNAs.

Fig. 2.

SF3b is required for intronless mRNA export. (A) Western blot analysis to examine the KD of SF3b155 and co-KD of SF3b145 and SF3b130. (B) KD of SF3b155 and SF3b145 inhibited nuclear export of the HSPA1A reporter mRNA. (B, Top) Schematic illustration of HSPA1A intronless mRNA reporter construct. The promoter, polyA site, and the FISH probe are shown. (B, Bottom) FISH was carried out to examine the distribution of the HSPA1A mRNA. SC35 immunofluorescence (IF) shows the localization of nuclear speckles. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 30, ***P < 0.01). (C) KD of SF3b155 inhibited nuclear export of the endogenous HSPA1A mRNA. FISH using a transcript-specific probe shows the distribution of endogenous HSPA1A mRNA. SC35 IF shows the location of nuclear speckles. Confocal microscopy was used to visualize the cells. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 10, ***P < 0.01).

To separate the impact of SF3b KD on mRNA export and splicing, we next examined nuclear export of naturally intronless mRNAs that do not undergo splicing. We microinjected reporter plasmids expressing intronless mRNAs, including HSPA1A, FOXC2, and CLDN3, into the nuclei of Cntl, SF3b155, and SF3b145 KD cells. At 20 min after microinjection, α-amanitin was added to terminate transcription, and, 2 h or 4 h later, FISH was carried out to detect N/C distribution of the corresponding mRNAs. Consistent with a previous study, the HSPA1A, FOXC2, and CLDN3 mRNAs were detected in nuclear speckles and the cytoplasm (Fig. 2B and SI Appendix, Fig. S1 C and D) (33). In marked contrast, they were largely accumulated in nuclear speckles in SF3b155- and SF3b145-depleted cells (Fig. 2B and SI Appendix, Fig. S1 C and D). Importantly, apparent nuclear retention of endogenous intronless mRNAs, including HSPA1A, ZXDB, and RHOB, was also observed in SF3b155-depleted cells compared with Cntl cells (Fig. 2C and SI Appendix, Fig. S1E). These data together indicate that SF3b is at least required for nuclear export of intronless mRNAs.

Tethering SF3b Enhances THO Recruitment and mRNA Export in a Dose-Dependent Manner.

The requirement of SF3b in intronless mRNA export does not necessarily indicate a role of SF3b in promoting mRNA export. It is also possible that SF3b depletion inhibited splicing and/or 3′ processing of pre-mRNAs encoding nuclear export factors, resulting in impaired mRNA export indirectly. A previous study identified a 7-nt SF3b-binding motif (SM for short), C/GAAGAAG, in replication-dependent (RD) histone mRNAs (20). We took advantage of this motif and asked whether SF3b binding promotes THO recruitment and mRNA export. To this end, we inserted one copy or three copies of SM into the AdML cDNA (cA) construct, which does not contain any SM (SI Appendix, Fig. S2). To exclude the possibility that SF3b binding impacts THO recruitment through promoting 3′ processing, we used in vitro transcribed mRNAs that do not undergo 3′ processing. The cA, cA-1SM, and cA-3SM mRNAs were incubated separately in HeLa NE, followed by IPs using antibodies to Cntl, SF3b155, and THOC2. As expected, the cA mRNA was barely bound by SF3b155, and insertion of SM progressively promoted SF3b155 binding as the number increased (Fig. 3A). Consistent with previous studies, the cA mRNA was not efficiently associated with THOC2 (24, 30). Significantly, 1SM slightly strengthened THOC2 binding and 3SM showed much more apparent effect (Fig. 3A). These results together indicate that tethering SF3b promotes THO recruitment in a dose-dependent manner. We next asked whether SF3b is required for efficient THO recruitment to SM-containing mRNAs by depleting SF3b155 from the NE. Consistent with the notion that SF3b is a tightly bound complex, SF3b155 depletion led to codepletion of other SF3b subunits, e.g., SF3b49 (SI Appendix, Fig. S1F). Significantly, SF3b depletion significantly reduced the binding of THO, but not that of CBP80, with the cA-3SM mRNA (Fig. 3B). Similarly, THO recruitment to the FOXC2 mRNA, which contains three copies of SM (SI Appendix, Fig. S2), was also apparently weakened by SF3b depletion (Fig. 3C). Thus, SF3b binding with the mRNA promotes THO recruitment.

Fig. 3.

Tethering SF3b enhances THO recruitment and mRNA export in a dose-dependent manner. (A) Insertion of SM to the mRNA enhanced THO recruitment. (A, Top) Schematic illustration of cA, cA-1SM, and cA-3SM mRNAs. (A, Bottom) RIPs of in vitro-transcribed cA, cA-1SM, and cA-3SM mRNAs from HeLa NE with indicated antibodies. IgG was used as the control. One fourth of the input was loaded. Relative enrichments are also shown (Bottom). Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01). (B) SF3b was required for the effect of SM on promoting THO recruitment. In vitro-transcribed cA-3SM mRNA was used for RIPs from mock- or SF3b155-depleted HeLa NE with indicated antibodies. The CBP80 antibody and the IgG were used as positive and negative controls, respectively. One fourth of the input was loaded. Relative enrichments are also shown (Bottom). Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01; n.s., not significant). (C) Same as B, except that, instead of the cA-3SM mRNA, the FOXC2 mRNA was used. (D) Insertion of SM to the mRNA enhanced its nuclear export. (D, Top) Schematic illustration of the cA, cA-1SM, and cA-3SM expression constructs. The promoter, polyA site, and probe are shown. (D, Bottom) FISH to detect the cA, cA-1SM, and cA-3SM mRNAs transcribed from microinjected constructs. DAPI staining was used to mark the nucleus. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 30, ***P < 0.01). (E) The effect of SM on promoting THO recruitment depended on SF3b and TREX. (E, Top) Schematic illustration of the cA-3SM expression construct. The promoter, polyA site, and probe are shown. (E, Bottom) FISH to detect the cA-3SM mRNA in Cntl, SF3b155, and UAP56 KD cells. DAPI staining was used to mark the nucleus. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 30, ***P < 0.01).

To examine whether SF3b binding promotes mRNA export, we microinjected the same set of cA reporter constructs into HeLa cell nuclei. As expected, the cA mRNA was exclusively nuclear at 2 h postinjection (Fig. 3D). In contrast, the cA-1SM mRNA could be detected in the cytoplasm, and more significant cytoplasmic accumulation was observed with the cA-3SM mRNA (Fig. 3D), indicating that SF3b binding indeed promotes mRNA export. Previously, it was shown that an SM-containing 22-nt RNA element promotes mRNA export by recruiting NXF1 through SR proteins (34, 35). Thus, it was also possible that the SM promoted mRNA export via SR-mediated NXF1 recruitment. This possibility was not supported by the observation that depletion of SF3b155 or the TREX component UAP56 blocked the cA-3SM mRNA in the nucleus (Fig. 3E). In addition to SF3b155, the SM also interacts with Prp43 (20). However, Prp43 is unlikely a key factor in SM-mediated mRNA export, as Prp43 KD showed only moderate impact on cA-3SM mRNA export (SI Appendix, Fig. S1 G and H). Together, these data indicate that SF3b promotes mRNA export by recruiting the THO/TREX complex.

SF3b Binds Intronless and Spliced mRNAs in the Cells.

Does SF3b actually bind intronless mRNAs in the cells? We analyzed the presence of SM in exons of intronless and intron-containing protein-coding genes from an Ensembl gene set (human hg19). This analysis revealed that one third of intronless genes contain at least one copy of SM, supporting the possibility that SF3b binds intronless mRNAs in the cells (Fig. 4A). Interestingly, two thirds of intron-containing genes harbor SM in their exon sequences (Fig. 4A), suggesting that SF3b might bind spliced mRNAs as well. To examine these possibilities, we analyzed SF3b155 enhanced cross-linking and IP (eCLIP) data (36) (SI Appendix, Fig. S3A). As expected, the vast majority (94%) of eCLIP peaks were mapped to protein-coding genes (SI Appendix, Fig. S3B), and most of these peaks (74.6%) were distributed in introns (Fig. 4B). Importantly, in agreement with the notion that SF3b binds on intronless and spliced mRNAs, a significant fraction (22.5%) of SF3b binding sites were mapped to intronless genes as well as exons of intron-containing genes (Fig. 4B). More importantly, eCLIP reads were apparently enriched at exon–exon junctions in both replicates (3.38% and 3.97%) compared with the input (1.31%; SI Appendix, Table S1). Motif analysis identified three apparently enriched motifs in SF3b155 binding peaks on exon sequences (SI Appendix, Fig. S3C). Although the top one GAGGA resembles the SM, the other two (GUUGGU and CGGGGGC) do not, suggesting that SF3b-binding sequences on exons are highly heterogenous. This provides an explanation for why no apparent enrichment of SF3b binding was detected on SM-containing intronless genes compared with SM-lacking ones (SI Appendix, Fig. S3D).

Fig. 4.

SF3b prevalently binds intronless and spliced mRNAs. (A) Pie charts show that more than one third of intronless mRNAs and two thirds of spliced mRNAs contain the SM motif in the exon sequences. (B) Pie chart shows the distribution of SF3b155 eCLIP peaks detected on protein-coding genes. (C) Western analysis to detect SF3b155 immunoprecipitated by the Cntl (i.e., IgG) and SF3b155 antibodies. (D) SF3b associated with U2 snRNA. RT-qPCRs to examine the associations of SF3b155 with U2 snRNA and 18S rRNA. RNAs enriched with Cntl were set as 1. Data represent the mean ± SEM (n = 3, ***P < 0.01; n.s., not significant). (E) SF3b associated with mature mRNAs. RT-qPCRs to examine the associations of SF3b155 with pre-mRNAs, intronless mRNAs, and spliced mRNAs. RNAs enriched with Cntl were set as 1. Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01).

To confirm SF3b binding on mature mRNAs, we next carried out RNA IPs (RIPs) with the SF3b155 antibody, followed by quantitative RT-PCRs (RT-qPCR; Fig. 4 C–E). As expected, SF3b155 was apparently enriched on the U2 snRNA but not the 18S rRNA (Fig. 4D). Consistent with previous studies (11–13, 20), pre-mRNAs, including UAP56 and URH49, and RD histone mRNAs, including H1-H2AG, H1-H3H, and H3-H2A, we tested were specifically enriched by the SF3b155 antibody (Fig. 4E). Significantly, SF3b association on nonhistone intronless mRNAs, including RHOB, HSPA1A, and HSPA6, and spliced mRNAs, including UAP56 and URH49, was also reproducibly detected, although to less of an extent compared with that on pre-mRNAs (Fig. 4E). Together, these data indicate that SF3b binds not only on pre-mRNAs but also on mature mRNAs.

Inhibition of U2 snRNP Function Does Not Inhibit, but Promotes, mRNA Export.

To date, all known roles of SF3b are executed in the context of U2 snRNP (8–10, 19, 20). To investigate whether this is also true for mRNA export, we used an antisense morpholino oligonucleotide (AMO) against U2 snRNA to inhibit U2 snRNP functions (37–39). Treatment of the cells with the U2 AMO, but not the Cntl AMO, significantly inhibited splicing of β-globin and Smad reporter pre-mRNAs (SI Appendix, Fig. S4A), indicative of inhibited U2 snRNP functions.

We next coinjected the FOXC2 expression construct with different amounts of Cntl or U2 AMO. To allow sufficient U2 snRNP inhibition, we incubated the cells for 8 h, with α-amanitin omitted to avoid cell toxicity resulting from long-term treatment. In cells treated with low or high Cntl AMO, the FOXC2 mRNA was largely detected in the nucleus, probably because of its continuous transcription (Fig. 5A). Unexpectedly, apparent cytoplasmic accumulation of the FOXC2 mRNA was detected with low U2 AMO, and the mRNA was mostly detected in the cytoplasm with high U2 AMO treatment, indicating that U2 snRNP inhibition did not impair, but enhanced, FOXC2 mRNA export (Fig. 5A). Similarly, U2 AMO treatment also promoted cytoplasmic accumulation of the CLDN3 reporter mRNA that contains one copy of SM (SI Appendix, Figs. S2 and S4B). Importantly, the effect of U2 AMO on enhancing mRNA export was also confirmed with endogenous intronless mRNAs, HSPA1A and RHOB. Although Cntl AMO did not apparently impact their N/C distribution, apparently enhanced cytoplasmic accumulation was observed in U2 AMO-treated cells compared with the surrounding nontreated cells (Fig. 5B and SI Appendix, Fig. S4C).

Fig. 5.

Inhibition of U2 snRNP promotes mRNA export. (A) U2 AMO treatment promoted FOXC2 mRNA export. FISH to examine the distribution of the FOXC2 mRNA derived from the construct coinjected with a low amount (1×) or high amount (3×) of Cntl or U2 AMO. The injection marker was used to mark injected cells. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 30, ***P < 0.01). (B) U2 AMO treatment promoted nuclear export of endogenous HSPA1A mRNA. Distribution of the endogenous HSPA1A mRNA was detected with its specific probe in cells injected with Cntl or U2 AMO. GFP was used to indicate injected cells. DAPI staining was used to mark the nuclei. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 30, ***P < 0.01). (C) The effect of U2 AMO treatment on promoting mRNA export required SF3b155. FISH to examine the distribution of the FOXC2 mRNA derived from the construct coinjected with Cntl or U2 AMO in cells treated with Cntl or SF3b155 siRNA. DAPI staining was used to mark the nuclei. (Scale bar, 20 μm.) Relative N/C ratios are also shown (Right). Data represent the mean ± SD from three independent experiments (n = 15, ***P < 0.01; n.s., not significant). (D) U2 AMO treatment enhanced nuclear export of spliced mRNAs. RT-qPCRs examining the cytoplasmic distribution of intronless and spliced mRNAs with Cntl or U2 AMO. mRNA levels in the cytoplasm (marked as “C”) and total cells (“T”) were examined, and the relative C/T ratios are shown. The C/T ratios in the Cntl AMO sample were set as 1. Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01). (E) Western analysis to assess the purities of the nuclear and cytoplasmic fractions. UAP56 and tubulin are markers for nucleus and cytoplasm, respectively. (F) SF3a and B″ did not associate with mature mRNAs. RT-qPCRs to examine the associations of SF3b155, SF3a, and B″ with pre-mRNAs, intronless mRNAs, and spliced mRNAs. RNAs enriched with IgG were set as 1. Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01; n.s., not significant).

Notably, KD of SF3b subunits, including SF3b155, SF3b145, and SF3b49, completely inhibited the effect of U2 AMO on mRNA export (Fig. 5C and SI Appendix, Fig. S4 D and E), indicating that U2 AMO treatment induced mRNA export promotion was dependent on the SF3b complex. It was possible that this promotion was caused by increased availability of nuclear export factors for intronless mRNAs as a result of reduced abundance of spliced mRNAs caused by splicing inhibition. However, although U1 AMO treatment also blocked splicing, it did not enhance cytoplasmic accumulation of the FOXC2 mRNA (SI Appendix, Fig. S5). Together, these data indicate that inhibition of U2 snRNP functions promotes intronless mRNA export.

As SF3b also associates with spliced mRNAs, we next sought to examine how nuclear export of spliced mRNAs was affected upon U2 snRNP inhibition. Considering the difficulty in distinguishing pre-mRNAs and spliced mRNAs based on FISH analysis, we carried out cell fractionation and RT-qPCRs. Consistent with the previously described FISH data, U2 AMO treatment resulted in apparently increased cytoplasm/total-cell (C/T) ratios of intronless mRNAs, including HSPA1A and RHOB. Significantly, the C/T ratios of spliced mRNAs we examined, including UAP56, URH49, FOS, FOXF2, and SF3b130, were all enhanced in U2 AMO-treated cells compared with Cntl cells (Fig. 5 D and E). Together, these data indicate that the role of SF3b in promoting mRNA export is not dependent on U2 snRNP, but is rather enhanced upon U2 snRNP inhibition. In further support of a U2 snRNP-independent role of SF3b in mRNA export, SF3a and B″ did not apparently associate with the intronless or spliced mRNAs we examined, although they were easily detected on the U2 snRNA and pre-mRNAs (Fig. 5F and SI Appendix, Fig. S6).

Disruption of U2 snRNP Formation Enhances the SF3b–THO Interaction and Their Recruitment to Mature mRNAs.

How does U2 AMO treatment enhance mRNA export? Our initial observation that RNase A treatment led to enhanced SF3b–THO interaction reminded us of the possibility that U2 AMO treatment might disrupt U2 snRNP. Indeed, IPs with the B″ antibody showed weakened bands corresponding to all SF3a and some SF3b subunits in the U2 AMO-treated NE compared with the Cntl NE on the silver-stained gel (SI Appendix, Fig. S7A). These weakened associations of SF3a and SF3b with B″ were validated by Western blotting (Fig. 6A). Further, the associations of SF3a, SF3b, and B″ with U2 snRNA were significantly reduced in the presence of U2 AMO (SI Appendix, Fig. S7 B and C). These data suggest that U2 AMO treatment disrupted the formation of a significant fraction of U2 snRNP. The disrupted U2 snRNP was further validated by a more direct method, the sedimentation of the U2 snRNPs on glycerol gradients. U2 AMO treatment caused an apparent shift of U2 snRNA and B″, but not U1 snRNA or the U1 snRNP protein U1A, to low-molecular weight (MW) fractions (SI Appendix, Fig. S8A). Similarly, U1 AMO treatment also caused a specific shift of U1 snRNA and U1-70K, a specific U1 snRNP component, to low-MW fractions (SI Appendix, Fig. S8B), suggesting that U1 snRNP was also disrupted by U1 AMO.

Fig. 6.

Disruption of U2 snRNP enhances the SF3b–THO interaction and their recruitment to the mRNA. (A) U2 AMO treatment reduced the association of SF3b and SF3a with B”. Western blot analysis to detect SF3b and SF3a subunits immunoprecipitated by the Cntl (i.e., IgG) and the B″ antibody from HeLa NE treated with Cntl or U2 AMO. (B) Western blot analysis to detect components of SF3b, SF3a, and THO complexes immunoprecipitated by the Cntl (i.e., IgG) and SF3b155 antibodies from HeLa NE treated with Cntl or U2 AMO. (C) U2 AMO treatment enhanced TREX recruitment to the FOXC2 mRNA. In vitro-transcribed FOXC2 mRNA was incubated under splicing condition with HeLa NE and treated with Cntl or U2 AMO, followed by IPs with the indicated antibodies, IgG was used as the Cntl. One fourth of the input was loaded. Relative enrichments are also shown (Bottom). Data represent the mean ± SEM (n = 3, *P < 0.05 and ***P < 0.01). (D) A model for two distinct pools of SF3b in U2 snRNP and the mature mRNP. In the normal condition, two distinct SF3b pools exist in the cells: SF3b bound on the pre-mRNA participates in its processing as a component of U2 snRNP, and SF3b bound on mature mRNAs promotes nuclear export by recruiting the THO. Upon U2 snRNP disassembly or disruption, SF3b is released and increasingly available for the assembly of export-competent mRNPs, resulting in reduced pre-mRNA processing and enhanced mRNA export.

In agreement with the view that U2 snRNP disruption leads to enhanced SF3b–THO interaction, reproducibly strengthened association of SF3b155 with THO proteins, including THOC2, THOC1, and THOC5, and reduced interaction with SF3a were detected in the U2 AMO NE compared with the Cntl NE (Fig. 6B). Further, SF3b155, THOC2, and ALYREF RIPs using in vitro-transcribed FOXC2 mRNA and HeLa NE treated with Cntl or U2 AMO revealed that U2 snRNP disruption promotes recruitments of SF3b and THO/TREX to the mRNA (Fig. 6C). Notably, when the cA and cA-1SM mRNAs were used for IPs, U2 AMO treatment enhanced the bindings of SF3b155 and THOC2 with the cA-1SM mRNA but not the cA mRNA (SI Appendix, Fig. S9), indicating that U2 snRNP disruption specifically promotes THO/TREX recruitment to mRNAs with SF3b-binding sites. Together, these data indicate that SF3b interacts with THO and promotes mRNA export independent of U2 snRNP, and suggest that the pool of SF3b in mature mRNPs competes with that present in U2 snRNP.

Discussion

To ensure efficient and accurate gene expression, nuclear RNA export needs to be tightly controlled. Accumulating evidence indicates that nuclear export machinery is limited to ensure only RNAs that are fully processed and properly assembled into mRNPs could be exported to the cytoplasm (3, 40). As the price, a fraction of mature mRNAs are retained in the nucleus and subject to degradation even in normal cells (7, 41). Producing a large number of mature mRNAs that ultimately undergo degradation is obviously not economic, and one way to avoid that is to ensure the balance between pre-mRNA processing and mRNA export.

Our study raises the possibility that the usage of common factors in pre-mRNA processing and mRNA export could provide a mechanism for balancing these processes. SF3b prevalently binds on pre-mRNAs, where it participates in splicing and 3′ processing in the context of U2 snRNP. Here we found that it also binds on mature mRNAs to promote nuclear export by recruiting THO/TREX through direct interactions. Intriguingly, the role of SF3b in mRNA export does not rely on, but competes with, that of U2 snRNP. These findings suggest that SF3b is limited and that alteration in SF3b distribution in U2 snRNPs and mature mRNPs impacts the balance between mRNA processing and export. In line with this, compared with the total abundance of U2 snRNA and mRNA (∼10⁶ and 2–4 × 10⁵, respectively; refs. 42–44), the concentration of SF3b is relatively low (∼3 × 10⁵; SI Appendix, Fig. S10). This SF3b-mediated balance could be important to ensure the majority of fully processed mRNAs are transported to the cytoplasm in normal condition, as only when released from mature mRNAs could SF3b enter the next cycle for pre-mRNA processing as a component of U2 snRNP. Also, it might be vital for guaranteeing mRNA export and protein expression when cells are under stressed conditions, such as heat shock, which causes U snRNP disassembly (45). Many other splicing factors, such as SR proteins, are also bound on mature mRNAs and promote mRNA nuclear export (22, 33–35, 46–51). It would be interesting to investigate whether these factors also contribute to balancing pre-mRNA processing and mRNA export. What other advantages could there be for utilizing SF3b, rather than U2 snRNP, in promoting mRNA export? It might be important to avoid pre-mRNA leakage by preventing THO recruitment to pre-mRNAs by the U2 snRNP. Also, U2 snRNP bound on mature mRNAs might be recognized by the surveillance machinery at the nuclear pore (52, 53) that likely precludes efficient mRNA export.

Our findings could also provide mechanistic insights into sequence- and splicing-dependent mRNA export. We found that SM exists in one third of naturally intronless mRNAs, and SF3b bindings are detected on them. Importantly, we provide evidence that this motif indeed promotes intronless mRNA export by enhancing THO/TREX recruitment. In addition to SM, we identified other motifs based on SF3b eCLIP data. In the future, it would be interesting to investigate whether these motifs are functional for SF3b binding and mRNA export. Although, in the present work, we could not directly examine whether SF3b promotes spliced mRNA export as a result of its critical role in splicing, we speculate that the SF3b–THO interaction possibly contributes to connecting mRNA export to splicing. SF3b is recruited to the BS region of pre-mRNAs as a component of U2 snRNP. During splicing, when SF3b is released from U2 snRNP (17, 18, 54–58), it might be handed over to spliced mRNAs via specific sequences present on exons and recruit THO (see model in Fig. 6D). Further studies are needed to investigate this hypothesis in future.

In recent years, recurrent mutations have been identified in SF3b155 in many types of cancers (32, 59, 60). Since this discovery, many efforts have been made to detect altered splicing events that are causative for tumorigenesis (2, 60–62). However, to date, no convincing evidence has been obtained supporting that SF3b155 mutations lead to cancer development as a result of splicing defects. The role of SF3b in mRNA export described here could open a new direction for investigation of the underlying mechanism for SF3b155 mutations causing cancers.

Materials and Methods

Cell Culture and Transfection.

HeLa cells were cultured in DMEM supplemented with 10% FBS and penicillin/streptomycin. For siRNA transfection, HeLa cells seeded in 12-well plates or glass-bottomed dishes were transfected with 10 pmol of siRNA per well using Lipofectamine RNAi Max (Invitrogen) following the manufacturer’s protocol. DNA transfection was performed by using Lipofectamine 2000 (Invitrogen) following the manufacturer’s protocol. AMOs (1 µM) were introduced to HeLa cells by using electroporation (NEPA21) according to the manufacturer’s recommendations. The siRNA and AMO target sequences are listed in SI Appendix, Table S2.

Pull-Downs.

Untagged SF3b or SF3b core complex (50 µg) was mixed with 50 µg of the strep-tagged THO complex or the strep-tagged MBP protein and 40 µL of Strep-Tactin Superflow beads (IBA) in a buffer of 50 mM Tris⋅HCl, pH 7.5, and 150 mM (for SF3b complex) or 500 mM (for SF3b core complex) NaCl and incubated for 4 h at 4 °C. Then, the beads were extensively washed four times, and the bound proteins were eluted with 5 mM d-desthiobiotin and separated on a 12% SDS/PAGE gel.

RNA Extraction and RT-qPCRs.

Total RNA was extracted by using TRI reagent (Sigma-Aldrich) and treated with RNase-free RQ1 DNase I (Promega) to remove genomic DNA. Random primer was used for reverse transcription by M-MLV reverse transcriptase (Promega). Quantitative PCR was carried out by using GoTaq Master Mix (Promega) according to the manufacturer’s instructions. The primers used for quantitative PCR are listed in SI Appendix, Table S3.

DNA Microinjections.

HeLa cells used for microinjection were plated on 20-mm coverslips at the bottom of 35-mm dishes. Plasmid DNA (100 ng/μL) was coinjected with FITC-conjugated 70-kDa dextran (injection marker). For each experiment, ∼300 cells were microinjected, followed by incubation at 37 °C. After 20 min of incubation, transcription was terminated with α-amanitin (4 μg/mL; Sigma-Aldrich), and incubation was continued for the desired period of time before fixation. For U1- or U2-snRNP disruption, control, U1, or U2 AMO was coinjected with plasmid DNA and α-amanitin was omitted.

Protein IPs.

To identify SF3a- and SF3b155-interacting proteins, the antibody to the whole SF3a complex or to SF3b155 was covalently cross-linked to nProtein A Sepharose (GE Healthcare) by dimethyl pimelimidate. A total of 350 μL of HeLa NE was treated or not treated with RNase A under splicing condition (6 mM Tris, pH 7.6, 30 mM KCl, 3.2 mM MgCl₂, 20 mM creatine phosphate, 0.05 mM ATP) at 30 °C for 20 min and then incubated with antibody cross-linked beads in IP buffer (1× PBS solution, 0.1% Triton, 0.2 mM PMSF, and protease inhibitor) overnight. The beads were washed with washing buffer (1× PBS solution, 0.1% Triton, 0.2 mM PMSF) three times, and the proteins were eluted with SDS loading buffer and analyzed by ion-trap MS. For protein IPs followed by Western blot, 75 μL of HeLa NE was used for IPs. For IPs following disruption of U2 snRNP formation, HeLa NE was coincubated with control or U2 AMO (4 μM) at 30 °C for 20 min. For THOC2 IP-Western experiment, the HeLa NE was diluted with IP buffer and treated with RNase A at 30 °C for 20 min and then incubated with antibody cross-linked beads overnight.

Immunodepletion and RIPs from NEs.

For immunodepletions, antibodies were covalently cross-linked to protein A Sepharose beads (GE Healthcare) at a 4:1 ratio (crude serum volume:packed beads volume) by using dimethyl pimelimidate (Sigma). A total of 100 μL of high-salt HeLa NE (350 mM) was mixed with 25 μL of beads and rotated for 2 h at 4 °C. The procedure was repeated three times. RIPs were carried out as previously described (30). Phosphorus-32–labeled in vitro-transcribed mRNAs were incubated in splicing reaction mixtures containing 30% HeLa NE for 2 h (63). To disrupt U2 snRNP formation, HeLa NE was coincubated with Cntl or U2 AMO at 30 °C for 2 h. For IPs, 5 μL of splicing reaction and 100 μL of binding buffer (20 mM Hepes, pH 7.9, 150 mM KCl, 0.1% Triton, 2.5 mM EDTA, and 5 mM DTT) were incubated with antibody-coupled protein A Sepharose beads at 4 °C for 2 h. After extensive washing with the binding buffer, immunoprecipitates were treated with proteinase K (Roche) at 37 °C for 10 min, and RNAs were recovered by phenol/chloroform extraction and ethanol precipitation. RNA was analyzed on denaturing polyacrylamide gels and visualized by PhosphorImager. One fourth of the input was loaded.

RIPs from Whole-Cell Lysates.

HeLa cell pellets were resuspended in 1 mL of NET-2 buffer [50 mM Tris⋅HCl, pH 7.4, 150 mM NaCl, 0.1% Tergitol (NP-40), 0.2 mM PMSF], followed by sonication and centrifuge. The lysates were incubated with the indicated antibodies for 2 h at 4 °C, followed by rotation with nProtein A Sepharose (GE Healthcare) for another 2 h at 4 °C. The immunoprecipitates were washed three times with the NET-2 buffer. One fifth of the immunoprecipitate was analyzed by Western blotting. The rest of the immunoprecipitate was treated with proteinase K, and RNAs were recovered by phenol/chloroform extraction and ethanol precipitation. For RIPs following disruption of U2 snRNP formation, HeLa lysates were coincubated with Cntl or U2 AMO at 30 °C for 20 min before incubation with the indicated antibodies.

Sedimentation Experiments.

A total of 400 μL of NE preincubated with Cntl, U1, or U2 AMO was layered on a 10–35% linear glycerol gradient containing 150 mM KCl, 20 mM Hepes/KOH, pH 8.0, and 1.5 mM MgCl₂. After centrifugation at 32,000 rpm for 17 h in a Beckman SW41 rotor, fractions were collected. The RNA of each gradient fraction was extracted by TRI reagent and visualized by SYBR Gold (Invitrogen) staining. The protein of each fraction was acetone-precipitated, followed by SDS/PAGE and Western blotting.