New mechanistic insights into Prp22-mediated exon ligation and mRNA release

Abstract

The DExD/H-box RNA helicase Prp22 catalyzes messenger RNA (mRNA) release from the spliceosome, and has also been implicated in proofreading the 3′ splice site (3′SS), preventing exon ligation of mutant pre-mRNAs through an ATP-dependent mechanism. However, here we reveal an unexpected role for Prp22 in promoting exon ligation of both wild-type and mutant pre-mRNAs by stabilizing Slu7’s association with the spliceosome prior to exon ligation. Notably, ATP binding, rather than hydrolysis, by Prp22 inhibits exon ligation of 3′SS mutant pre-mRNA. Following exon ligation, Prp22-mediated ATP hydrolysis facilitates the dissociation of both Slu7 and mRNA from the spliceosome. Remarkably, Prp22 and Cwc22, which bind the 3′- and 5′-exons respectively, remain associated with the released mRNA, whereas Slu7 and Fyv6 dissociate independently. We propose that Prp22 facilitates exon ligation by stabilizing Slu7 binding, with binding of ATP by Prp22 potentially destabilizing that interaction, thereby weakening contacts between the 5′-exon and the 3′SS to inhibit exon ligation. After exon ligation, Prp22-driven ATP hydrolysis induces a conformational change in Prp8 that disrupts its interdomain interactions, enabling mRNA release through the domain interfaces, with Prp22 and Cwc22 remaining associated with the released mRNA.

Graphical Abstract

Graphical Abstract.

Introduction

RNA splicing is an energetically demanding process requiring ATP hydrolysis at multiple stages of the pathway. Eight DExD/H-box RNA helicases drive sequential structural transitions within the spliceosome, including during both catalytic steps [1–5]. In the first catalytic step, Prp2 first displaces the U2 snRNP components SF3a/b, removing proteins that occupy the region between the 5′ splice site (5′SS) and the branch site (BS) [6–11]. The branching reaction is facilitated by Yju2 and Cwc25, which stabilize the interaction between the 5′SS and the BS, and become tightly bound to the spliceosome after the reaction [12, 13]. To prepare for the second catalytic step, these proteins must be removed from the catalytic center, a process mediated by Prp16, to allow proper positioning of the 3′ splice site (3′SS) at the catalytic center [14]. The subsequent exon ligation reaction requires Slu7 and Prp18, which form a stable heterodimer [15–23]. After completion of the splicing reaction, Prp22 catalyzes the release of the mature messenger RNA (mRNA) from the spliceosome, along with Slu7–Prp18 and itself [24, 25]. Finally, spliceosome disassembly is driven by Prp43 [26, 27].

Several DExD/H-box proteins have been implicated in maintaining splicing fidelity [28–33]. Prp16 was initially identified as a suppressor to the ACT1 branchpoint (BP) brC mutation in a genetic screen for BS-binding proteins but was later found to be required for splicing only after the branching reaction [16,34]. Subsequent investigations led to the proposition that upon branch formation, Prp16 proofreads the BP sequence by rejecting defective spliceosomes, utilizing ATP hydrolysis for energy [28]. A later study suggested that Prp16 also proofreads slow 5′SS cleavage when a non-bridging oxygen of the U80 nucleotide of the U6 small nuclear RNA is replaced with sulfur in the pro-Sp configuration [35]. These findings imply a role for Prp16 in ensuring the accuracy at both the 5′SS and the BS during the branching reaction. However, recent research has yielded contrasting results, indicating that Prp16 promotes the branching reaction of pre-mRNAs harboring mutations at the 5′SS or the BP, leading to the utilization of aberrant 5′SS or BP [36]. Prp16 exerts this function by stabilizing the interaction of Cwc25 with the spliceosome independently of ATP. This raises the question as to whether Prp16 is actively involved in proofreading splice sites.

Prp22 has been shown to play a comparable role to Prp16 in that it proofreads the 3′SS [29]. In yeast, mutations at the 3′SS do not affect spliceosome assembly or the branching reaction, but impede exon ligation [37, 38], with Prp22 and Slu7 both remaining stably associated with the spliceosome [23, 39]. However, it was found that exon ligation could still proceed when the stalled post-branching spliceosome was isolated and re-incubated in splicing buffer lacking ATP, but not when ATP was present [29].

In contrast, here we show that the exon ligation reaction of 3′SS-mutated pre-mRNA depends on the presence of Prp22 and it is partially inhibited by ATP in a manner independent of ATP hydrolysis. The resulting mRNAs utilize aberrant 3′SSs, indicating that Prp22 facilitates, rather than prevents, aberrant exon ligation. Furthermore, Prp22 stabilizes binding of Slu7 to the spliceosome prior to exon ligation, but this stabilizing effect diminishes after exon ligation. Our findings indicate that Prp22 may promote exon ligation by stabilizing Slu7 on the spliceosome. We have also observed that although Prp22 is not essential for exon ligation of wild-type pre-mRNAs, it increases the reaction rate at early time points.

Cryogenic electron microscopy (Cryo-EM) structures of the spliceosome C* and P complexes reveal that Slu7 adopts a configuration, with multiple protrusions contacting Prp18 and several domains of Prp8, including the interfaces between the N and Endonuclease (Endo) domains, and between the Endo and RNase H-like (RH) domains [40, 41] (Fig. 1). These interactions suggest that Slu7 may contribute to maintaining the structural rigidity of Prp8, potentially stabilizing the close juxtaposition of the 5′SS and 3′SS for exon ligation. Notably, the mRNA protrudes outward from the spliceosome catalytic center through Prp8, with the 5′ exon positioned between the N and Linker domains, and the 3′ exon between the Linker, Thumb, and Reverse transcriptase-like (RT) domains. Strikingly, we found that Cwc22 and Prp22, which bind the 5′-exon and the 3′-exon, respectively, remain associated with the released mRNA, whereas Slu7–Prp18 dissociates from it independently. This observation challenges the prevailing model that Prp22-catalyzed ATP hydrolysis drives mRNA release through a pulling mechanism.

Figure 1.

The architecture of Prp8, Slu7, Prp18, Prp22, Cwc22, and mRNA in the spliceosome P complex. The circle indicates the catalytic center of the spliceosome.

Our findings elucidate the complementary roles of Prp22 and Prp16 in driving splicing progression. They remodel the spliceosome through their ATP-dependent activities, while their ATP-independent activities stabilize interactions between spliceosomal components, a function particularly pronounced in the context of mutant pre-mRNAs. Furthermore, we provide novel insights into the mechanism that governs mRNA release from the spliceosome.

Materials and methods

Yeast strains

The following yeast strains were used: BJ2168 (MATa prc1 prb1 pep4 leu2 trp1 ura3), YSCC227 (MATa prc1 prb1 pep4 leu2 trp1 ura3 CWC22-HA), YSCC618 (MATa prc1 prb1 pep4 leu2 trp1 ura3 FYV6-V5 PRP18-HA), and YSCC701 (MATa prc1 prb1 pep4 leu2 trp1 ura3 SLU7-V5).

Antibodies and reagents

Anti-hemagglutinin (anti-HA) monoclonal antibody 8G5F was produced by immunizing mice with a keyhole limpet hemocyanin-conjugated HA peptide. Anti-V5 monoclonal antibody was purchased from Serotec Inc. Anti-Ntc20, anti-Slu7, and anti-Cwc22 antibodies were raised by injecting rabbits with full-length recombinant proteins. Anti-Prp16 and anti-Prp22 antibodies were raised by injecting rabbits with recombinant proteins of amino acids 1–298 for Prp16 and 1–484 for Prp22. Protein A-Sepharose (PAS) was obtained from GE Healthcare Inc. Ni-NTA agarose was sourced from QIAGEN. Proteinase K was purchased from Cyrusbioscience Inc. SP6 RNA polymerase and RNasin were purchased from Promega.

In vitro splicing assay, immunoprecipitation, and immunodepletion

Splicing substrates were synthesized by in vitro transcription with SP6 RNA polymerase. ACT1 pre-mRNA and its derivatives were transcribed from EcoRI-linearized pSPAct6-88 plasmid [42]. Splicing assays were carried out according to the procedure described in Cheng et al. [42] at 25°C for 30 min using 40% (v/v) regular extracts or 50% (v/v) depleted extracts following incubation with antibody-conjugated PAS. In reactions supplemented with recombinant proteins, KPO₄ concentrations were adjusted to an equivalent ionic strength of 60 mM KPO₄. Immunoprecipitation of the spliceosome was performed as described previously [43]. For each 10–20 μl of splicing reaction mixture, 10 μl of PAS conjugated with crude sera were used, containing the following amounts of antibodies: 1.5 μl of anti-Ntc20, 2 μl of anti-Prp22, 5 μl of anti-Cwc22. For tagged proteins, monoclonal antibodies were used: 2 μl of anti-V5 for V5-Slu7 or 20 μl of anti-HA for Prp18-HA. Immunodepletion of specific proteins was performed under the condition of 50 μl of antibody-conjugated PAS for 100 μl yeast extracts, using 100 μl of anti-Prp22, 75 μl of anti-Prp16, 100 μl of anti-Slu7, or 10 μl of anti-V5 antibody for Fyv6-V5.

Crosslinking analysis

Ultraviolet (UV)-crosslinking of Cwc22, Prp22, and Prp8 to pre-mRNAs was performed as described by Chiang and Cheng [44]. Splicing reactions were conducted in Cwc22-HA extracts using 2 nM ACAC mutant pre-mRNA, or 2 nM wild-type ACT1 pre-mRNA in the presence of 5 nM prp22-S635A mutant protein. After UV irradiation, reaction mixtures were treated with denaturants and subsequently diluted 10-fold prior to immunoprecipitation.

Quantification of RNA from gels

Gels were exposed to an IMAGING PLATE (FUJIFILM Corp.), and RNA bands were visualized using a Typhoon^TM FLA 9000 system (GE Healthcare Life Sciences). Band intensities were quantified with ImageQuant TL7.0 (GE Healthcare Life Sciences). The value of each quantified RNA band was normalized to the number of uridines in each RNA species to obtain relative molecular amounts.

Sequence analysis of splice junctions using next generation sequencing

Splicing reactions were precipitated with anti-Ntc20 antibody, and the precipitated RNA was extracted and fractionated by electrophoresis on acrylamide gels. The mRNA band was excised and eluted from the gel for sequence analysis. RNA-seq libraries were prepared using the KAPA Hyper Prep Kit (Roche) following the manufacturer’s instructions. Sequencing was performed on an Illumina MiSeq and NextSeq 500 with paired read length of 150 bp. Sequencing reads comprising the BCL files were demultiplexed into FASTQ files using the Illumina bcl2fastq program v2.20 with default options allowing one mismatch of the sample barcode sequences. The RNA-seq experiments and initial bioinformatic analyses were performed by the Genomics and Bioinformatics Core at the Institute of Molecular Biology, Academia Sinica.

Results

Prp22 promotes aberrant 3′SS selection in 3′SS-mutated pre-mRNAs

Although it is widely recognized that Prp16 proofreads the BS and the 5′SS through its ATPase activity, our recent study revealed additional ATP-independent functions of Prp16 that facilitates utilization of aberrant 5′SS and BS in mutant pre-mRNAs by stabilizing 5′SS–BS interactions [36]. Given that both of the catalytic splicing steps share similar mechanistic principles, we speculated that Prp22 might play a comparable role in enabling erroneous 3′SS selection when obstacles arise during the second catalytic reaction.

Prp22 has been shown to proofread the 3′SS through its ATPase activity, similarly to how Prp16 proofreads the 5′SS and the BS. In that study, the 3′SS mutants of ACT1 pre-mRNA were used to accumulate splicing intermediates on the spliceosome, which was subsequently isolated by immunoprecipitation. When the purified spliceosome was incubated in splicing buffer with or without ATP, exon ligation occurred in the absence of ATP but was inhibited in its presence. Based on these observations and additional experiments, the authors concluded that Prp22 proofreads the 3′SS by preventing exon ligation through ATP hydrolysis [29]. We reasoned that if Prp22-mediated ATP hydrolysis impedes exon ligation in 3′SS-mutated pre-mRNAs, then depleting ATP should promote exon ligation in the splicing reaction mixture without the need for spliceosome purification. Adding glucose to cell extracts is an effective way to deplete ATP as hexokinase in the extract catalyzes the phosphorylation of glucose, converting ATP to ADP and glucose-6-phosphate [45]. Splicing was initiated using the 3′SS mutant of ACT1 pre-mRNA ACAC, in which the conserved intronic 3′SS dinucleotide and the first two nucleotides of the 3′ exon (AGAG) are mutated to ACAC (Fig. 2A) [38]. The reaction proceeded for 20 min before glucose was added to deplete ATP, and the splicing intermediates and products were subsequently monitored over a 60-min time-course (Fig. 2B). The results revealed an increase in spliced products following a prolonged incubation period (Fig. 2C, lanes 6–10). However, we also detected generation of spliced products even without ATP depletion, albeit at a slower rate (lanes 1–5). It is possible that ATP was consumed by the various ATPases in the extract, allowing exon ligation to occur during extended incubation even without addition of glucose. These findings suggest that ATP slows down, but does not completely block, the exon ligation reaction. Moreover, exon ligation proceeded very slowly even when ATP had been depleted.

Figure 2.

Prp22 promotes aberrant 3′SS selection in splicing of 3′SS-mutated of ACT1 pre-mRNA. (A) Sequence of the BS-3′SS regions of ACT1 and ACAC pre-mRNA. The BS is boxed, the BP and conserved intronic 3′SS dinucleotides are in bold, and nucleotide changes at the 3′SS in ACAC are underlined. (B) Schematic of the experimental flow chart. (C) Splicing was performed with ACAC pre-mRNA for 20 min. Following the addition (lanes 6–10) or not (lanes 1–5) of 10 mM glucose, the reaction mixtures were further incubated for the indicated time frames. (D) Splicing was performed with ACAC pre-mRNA for 20 min in mock-treated (lane 1) or Prp22-depleted (lane 4) extracts. Following the addition (lanes 3 and 6) or not (lanes 2 and 5) of 10 mM glucose, the reaction mixtures were further incubated for 60 min.

To determine whether Prp22 is essential for this reaction, we depleted Prp22 from splicing extracts using an anti-Prp22 antibody. Splicing was carried out for 20 min, with an additional 60-min incubation following the addition of glucose or ATP as a control (Fig. 2D). Exon ligation of the ACAC mutant pre-mRNA was markedly impaired in Prp22-depleted extracts, irrespective of ATP or glucose addition (lanes 4–6). These results demonstrate that Prp22 is required for efficient exon ligation of the mutant pre-mRNAs.

To rule out the possibility that this function of Prp22 in promoting exon ligation is specific to ACT1 pre-mRNA, we extended our investigation to include pre-mRNAs derived from the RPL23A and RPS6A genes, in which the 3′SS and the downstream dinucleotides AG had been mutated to AC so that no AG dinucleotide was present within a distance of 60 nucleotides (nt) downstream of the BP (Supplementary Fig. S1). Similar results were observed for these mutant pre-mRNAs as for ACT1 pre-mRNA, demonstrating that Prp22 is required for exon ligation of 3′SS mutant pre-mRNAs. To further explore the 3′SS utilized for exon ligation in these pre-mRNAs, we isolated the resulting mRNAs, and sequenced their splice junctions by means of next generation sequencing.

The 3′SSs for ACT1, RPL23A and RPS6A pre-mRNAs are located 43, 44, and 37 nt downstream of the BP-A, respectively. Sequence analysis revealed substantial variability in both distance from the BP and preferred dinucleotide sequences, with UG and AC being the most frequently used alternative 3′SS (Supplementary Fig. S1). Interestingly, there was a strong preference by ACT1 for the authentic site of AC located 43 nt downstream from the BP. In contrast, the RPL23A and PRS6A pre-mRNAs predominantly utilized sites 15–30 nt downstream of the BP with preferences for UG. These results demonstrated that Prp22 facilitates exon ligation by promoting the use of aberrant 3′SS with low sequence specificity.

Prp22 does not promote the release of Slu7 from the spliceosome before exon ligation

Our findings, together with previous studies, have demonstrated that ATP inhibits the exon ligation reaction of 3′SS mutant pre-mRNAs. It has been proposed that ATP hydrolysis by Prp22 either induces a conformational change in the spliceosome to prevent exon ligation of mutant pre-mRNAs or facilitates the rejection of spliceosomes that have already formed on the mutant pre-mRNA and undergone ligation [29]. Another possibility is that Prp22 facilitates Slu7 dissociation, thereby preventing exon ligation of mutant pre-mRNA, since Slu7 is essential for exon ligation, and is dissociated from the spliceosome following ligation upon ATP hydrolysis. Although Prp22 and Slu7 were shown to remain stably associated with the spliceosome formed on 3′SS mutant pre-mRNA, [23], it is possible that Slu7 dynamically cycles on and off the spliceosome in response to ATP hydrolysis when exon ligation is impeded, and is captured in the bound state during purification. To investigate if the release of Slu7 from the spliceosome is responsible for the inhibition of exon ligation, we isolated the spliceosomes stalled before exon ligation, and examined Slu7’s association with them after incubation with ATP. We first monitored the time course of the exon ligation reaction on the purified ACAC spliceosome to evaluate the amounts of the spliceosomes engaged in the exon ligation reaction in the presence or absence of ATP. Reincubation of the purified spliceosomes in the splicing buffer in the absence of ATP indeed triggered the exon ligation reaction, consistent with a report published previously (Fig. 3A, lanes 1–5) [29]. However, exon ligation also occurred when ATP was present during the incubation, albeit at a slower rate (lanes 6–10). This result aligns with our observations of the splicing reaction mixture without spliceosome purification.

Figure 3.

Prp22 does not promote release of Slu7 from the spliceosome before exon ligation. (A) Splicing was performed with ACAC pre-mRNA for 20 min and spliceosomes were precipitated with anti-Ntc20 antibody. Spliceosome were further incubated for 0–60 min without (lanes 1–5) or with (lanes 6–10) the addition of 2 mM ATP. (B) Splicing was performed with ACAC pre-mRNA in Slu7-V5 extracts, and spliceosomes were isolated by precipitation with anti-Ntc20 (lanes 1–5), anti-Prp22 (lanes 6–10), or anti-V5 (lanes 11–15) antibody. The spliceosomes were then incubated in the presence or absence of ATP, after which the supernatant and pellet fractions were separated. T, total; P, pellet; S, supernatant.

Next, we examined if Slu7 dissociates from the spliceosome upon incubation with ATP using ACAC mutant pre-mRNA and Slu7-V5-tagged extracts. Splicing reaction mixtures were immunoprecipitated with anti-Ntc20, anti-Prp22, or anti-V5 antibody, and the precipitates were then incubated for 10 min in the presence or absence of ATP (Fig. 3B). Despite a reduction in levels of spliced product levels in the presence of ATP, the Ntc20, Prp22, and Slu7 proteins remained associated with the spliceosome containing splicing intermediates (lanes 4, 5, 9, 10, 14, and 15). This indicates that none of these proteins dissociate from the C* complex. Thus, we conclude that the ATP-dependent inhibition of exon ligation is not caused by release of Slu7.

Prp22 is associated with mRNA upon release from the spliceosome

Interestingly, differential associations of lariat intervening sequence (IVS) and mRNA with these proteins were observed following ATP incubation. While Ntc20 remained tightly bound to the IVS, ∼40% of the mRNA dissociated from the Ntc20-associated spliceosome in this assay (Fig. 3B, lanes 4 and 5). Nearly 40% of IVS and mRNA were also separated from Slu7 (lanes 14 and 15). In contrast, all of the mRNA remained associated with Prp22 (lanes 9 and 10). These results indicate that upon splicing completion, Prp22 catalyzes the release of mRNA from the spliceosome while staying associated with the mRNA. Meanwhile, Slu7 dissociates from the spliceosome separately from the mRNA.

Given the minimal mRNA products generated from ACAC pre-mRNA, we analyzed mRNA derived from wild-type pre-mRNA arrested on the spliceosome by performing splicing in Prp22-depleted extracts. The role of Prp22 in exon ligation of normal pre-mRNAs has been controversial. One study reported that Prp22 depletion resulted in the accumulation of splicing intermediates, with reintroduction of recombinant Prp22 restoring exon ligation, suggesting that Prp22 is required for exon ligation. Nevertheless, shortening the BS-3′SS distance bypass the requirement for Prp22 in exon ligation [22]. Conversely, another study using purified spliceosomes showed that Slu7 and Prp18 alone were sufficient to dock the 3′SS at the spliceosome active site for exon ligation. Here, our Prp22 depletion experiments revealed that, unlike Prp16 or Slu7 depletion, Prp22 depletion did not impair exon ligation but instead led to intron accumulation (Fig. 4A), supporting the conclusion of the second study [39]. Notably, lariat introns are degraded only after spliceosome disassembly, which follows Prp22-mediated mRNA release. Thus, intron accumulation indicates a defect in either mRNA release or spliceosome disassembly. Additionally, our 4-thiouridine (4sU)-crosslinking experiments to explore the interaction of spliceosomal components with the splice site sequences corroborated this finding. We observed that Prp8 crosslinked to the 3′SS in the C* complex formed with ACAC mutant pre-mRNA. While depletion of both Slu7 and Prp22 abolished this crosslinking [46], depletion of Prp22 alone had no impact on Prp8’s interaction with the 3′SS (Supplementary Fig. S2). Collectively, these results indicate that although Prp22 can promote exon ligation in 3′SS mutant pre-mRNA, it is not essential for exon ligation in wild-type pre-mRNA.

Figure 4.

Prp22 is associated with the mRNA upon mRNA release from the spliceosome, whereas both Slu7 and Fyv6 dissociate independently. (A) Western blotting and splicing reactions of extracts depleted of no protein (lane 1), Prp16 (lane 2), Slu7 (lane 3), or Prp22 (lane 4). (B) Schematic of the experimental flow chart. (C) Splicing was performed with wild-type ACT1 pre-mRNA using Prp22-depleted Slu7-V5 extracts. Following ATP depletion, recombinant Prp22 was added to the reaction mixtures, and the spliceosomes were precipitated with anti-Ntc20 (lanes 1–5), anti-Prp22 (lanes 6–10), or anti-V5 (lanes 11–15) antibody. The spliceosomes were then incubated in the absence or presence of ATP, and the supernatant and pellet fractions were separated. T, total; P, pellet; S, supernatant. (D) Western blotting of Fyv6-V5/Prp18-HA double-tagged extracts following incubation with PAS (lanes 1 and 4), anti-HA (lanes 2 and 5), or anti-V5 (lanes 3 and 6) antibody and separation of the supernatant and pellet fractions. Sup, supernatant; PAS, protein-A Sepharose. (E) Splicing was performed with wild-type ACT1 pre-mRNA using Prp22-depleted Fyv6-V5/Prp18-HA double-tagged extracts. Following ATP depletion, recombinant Prp22 was added to the reaction mixtures, and spliceosomes were precipitated with anti-Ntc20 (lanes 1–5), anti-V5 (lanes 6–10), or anti-HA (lanes 11–15) antibody. The spliceosomes were then incubated in the absence or presence of ATP, and the supernatant and pellet fractions were separated. T, total; P, pellet; S, supernatant.

We then conducted the mRNA release assay by performing splicing in Prp22-depleted Slu7-V5 extracts to arrest the spliceosome at the P complex stage. Following ATP depletion, recombinant Prp22 was added back to the reaction, and mRNA release was assessed using the same procedure as described for ACAC pre-mRNA (Fig. 4B). Upon addition to the reaction, Prp22 bound to the P complex, as evidenced by the precipitation of the spliceosome with anti-Prp22 antibody (lane 6). Given substantial amounts of splice products generated in this assay, the results clearly demonstrate that Prp22 catalyzes the release of both mRNA and Slu7 from the spliceosome (Fig. 4C, lanes 4, 5, 14, and 15), while remaining bound to the mRNA itself (lanes 9 and 10), irrespective of mutations at the 3′SS. Although we cannot entirely rule out the possibility that Prp22 rebinds to the mRNA after its release, we consider it unlikely as only negligible amounts of mRNA are detected in the supernatant, and rebinding of mRNA to immobilized Prp22 is expected to be inefficient. In contrast, Slu7 is released from the spliceosome in a free form, not associated with either the mRNA or Prp22 (lanes 14 and 15). This is the first demonstration of Prp22 associating with mRNA following its release from the spliceosome.

Fyv6, the yeast homolog of the human FAM192A protein, has recently been identified as a novel second-step splicing factor that facilitates selection of the 3′SS distal to the BP [41, 47]. Cryo-EM structures of the P complex reveal that Fyv6 is positioned at the periphery of the spliceosome, where it contacts the RecA2 domain of Prp22. This interaction was proposed to support mRNA release [41]. To investigate the interplay between Fyv6 and Slu7–Prp18, we generated a PRP18-HA and FYV6-V5 double-tagged yeast strain for extract preparation. Immunoprecipitation with anti-HA antibody revealed coprecipitation of Slu7 but not Fyv6 (Fig. 4D, lane 5). Conversely, immunoprecipitation with anti-V5 antibody did not coprecipitate Slu7 or Prp18 (lane 6), indicating that Fyv6 is not tightly associated with Slu7–Prp18. An mRNA release assay showed that, like Slu7, both Prp18 and Fyv6 were separated from mRNA and the spliceosome upon ATP incubation (Fig. 4E, lanes 10 and 15), indicating that the interaction between Prp22 and Fyv6, observed on the spliceosome, is not maintained during mRNA release.

Cwc22 is associated with the released mRNA

Our discovery that Prp22 remains associated with the released mRNA prompted us to investigate other proteins potentially linked to the released mRNA. In a prior study employing 4sU-crosslinking, we demonstrated binding of Cwc21, Cwc22, and Snu114 to the 5′ exon spanning a region from nucleotides −20 to −16 [46]. Cryo-EM structures of the spliceosome P complex have further elucidated this arrangement, showing that the 5′ exon protrudes from the spliceosome’s catalytic center through a tunnel formed by the N and Linker domains of Prp8, with Cwc21, Cwc22, and Snu114 positioned outside this tunnel [40, 46, 48–51]. Given its role as an adaptor for the exon junction complex (EJC) binding to the 5′ exon of mRNA in the human spliceosome [52–54], Cwc22 likely associates with the released mRNA. UV-crosslinking analysis revealed that, like Prp22, HA-tagged Cwc22 predominantly crosslinked to the mRNA, whereas Prp8 exhibited the strongest crosslinking with the lariat-intron on the P complex (Fig. 5A, lanes 9–11). In the C* complex formed with ACAC pre-mRNA, Prp22 and Prp8 primarily crosslinked to intron-exon 2, whereas Cwc22 predominantly crosslinked to the 5′ exon (Fig. 5B, lanes 9–11), consistent with the 4sU-crosslinking data.

Figure 5.

Cwc22 is released from the spliceosome in association with mRNA. (A) Crosslinking of Cwc22 to the mature mRNA. Splicing was performed using Cwc22-HA extracts in the presence of V5-tagged recombinant Prp22-S635A mutant protein, followed by UV_{254 nm} irradiation. After denaturation, the reaction mixtures were diluted tenfold with NET-2 buffer containing 300-mM NaCl and immunoprecipitated with anti-Ntc20, anti-Prp8, anti-Prp22, or anti-HA antibody. Denat, denaturation; C22, Cwc22. (B) Splicing was performed with ACAC pre-mRNA using Cwc22-HA extracts, followed by UV_{254 nm} irradiation. Reaction mixtures were treated as in panel (A) prior to immunoprecipitation. Denat, denaturation; C22, Cwc22. (C) Splicing was performed with wild-type ACT1 pre-mRNA in Prp22-depleted extracts. After ATP depletion, recombinant Prp22 was added to the reaction mixtures, and the spliceosomes were isolated by precipitation with anti-Ntc20 (lanes 1–5), anti-Prp22 (lanes 6–10), or anti-Cwc22 (lanes 11–15) antibody. The spliceosomes were then incubated in the presence or absence of ATP, after which the supernatant and pellet fractions were separated. T, total; P, pellet; S, supernatant.

We then performed the release assay to examine Cwc22 associations during mRNA release (Fig. 5C). Compared to Ntc20 and Prp22, Cwc22 appeared to bind less tightly to the spliceosome, as a larger proportion of all RNA species dissociated from Cwc22 upon incubation in the absence of ATP (lanes 12 and 13). In the presence of ATP, >50% of the intron was separated from Cwc22, whereas <30% of mRNA did (lanes 14 and 15), indicating that Cwc22 predominantly associates with mRNA during its release. The cryo-EM structure of the spliceosome P complex reveals that the 5′ exon and the 3′ exon extrude out of the catalytic center through distinct Prp8 domain interfaces, i.e., between the N and the Linker domains for the 5′ exon and between the Linker, Thumb, and RT domains for the 3′ exon. Notably, both Cwc22 and Prp22, which bind to the 5′ and the 3′ exons, respectively, remain associated with the mRNA upon its release. These findings support that mRNA is not pulled out of the spliceosome from its catalytic center during release. Instead, Prp22 may facilitate mRNA release by destabilizing interactions between Prp8 domains.

Prp22 stabilizes binding of Slu7 to the spliceosome to promote exon ligation

Slu7, Prp18, and Prp22 are recruited to the spliceosome prior to Prp16-mediated remodeling [39]. They are conditionally required for the exon ligation reaction, depending on the distance between the BS and the 3′SS (referred to as the intron 3′-tail, or i3’T) on the pre-mRNA. When the i3’T is shorter than 7 nt, none of these proteins is essential for the exon ligation reaction [21, 55]. Slu7 becomes essential when the i3’T exceeds 7 nt, implying a pivotal role in positioning the 3′SS at the catalytic center of the spliceosome. Through 4sU-crosslinking experiments, we have previously demonstrated how Prp8’s interaction with the 3′SS on the C* complex formed with ACAC pre-mRNA depends on Slu7. Depletion of both Slu7 and Prp22 abolishes this interaction, whereas depletion of Prp22 alone does not impact crosslinking of Prp8 to the 3′SS [46] (Supplementary Fig. S3), further underscoring Slu7’s crucial involvement in positioning the 3′SS.

Slu7, Prp18, and Prp22 have been proposed to interact with the spliceosome in a coordinated and sequential manner [23]. Our immunoprecipitation analysis further revealed that Slu7 and Prp22 mutually stabilize each other on the C* complex assembled on the ACAC pre-mRNA (Fig. 6A). Depletion of either Slu7 or Prp22 individually led to little or no association of the other protein with the spliceosome, whereas supplementation with the corresponding recombinant proteins restored the interaction. However, Slu7 became more stably associated with the spliceosome following exon ligation (Fig. 6B and C). In the absence of Prp22, both mRNA and Slu7 remained bound to the spliceosome. Immunoprecipitation of the reaction mixtures showed that the association of spliced products with Slu7 was relatively unaffected, with 35% of Ntc20 detected compared to 14% for splicing intermediates derived from the ACAC pre-mRNA (Fig. 6C). These results indicate that, in addition to its ATP-dependent role in destabilizing Slu7 after exon ligation, Prp22 also contributes to stabilizing Slu7 binding to the spliceosome prior to exon ligation. This stabilization may promote exon ligation of 3′SS-mutant pre-mRNAs, leading to aberrant 3′SS selection.

Figure 6.

Prp22 stabilizes Slu7 on the spliceosome prior to but not after exon ligation. (A) Splicing was performed with ACAC pre-mRNA in Slu7-V5 extracts that had been mock-treated (lanes 1–4), depleted of Slu7 (lanes 5–12), or depleted of Prp22 (lanes 13–20), supplementing with recombinant Slu7 (lanes 9–12) or Prp22 (lanes 17–20). The reaction mixtures were then precipitated with anti-Ntc20, anti-V5 or anti-Prp22 antibody. α-20, anti-Ntc20 antibody; α-22, anti-Prp22 antibody. (B) Splicing was performed with ACAC (lane 1–8) or wild-type ACT1 pre-mRNA (lanes 9–12) in Slu7-V5 extracts that had been mock-treated (lanes 1–4) or depleted of Prp22 (lanes 5–12). The reaction mixtures were then precipitated with anti-Ntc20, anti-V5, or anti-Prp22 antibody. α-20, anti-Ntc20 antibody; α-22, anti-Prp22 antibody. (C) Bar graph summarizing quantification of the results from panel (B), displaying the percentages of RNA precipitated by anti-V5 or anti-Prp22 antibody relative to that precipitated by anti-Ntc20 antibody. The amounts of lariat-intron-exon 2 and lariat-intron were measured for ACAC and wild-type pre-mRNA, respectively. Data represent mean values from three experiments with standard deviation (SD) indicated.

Given Prp22’s role in stabilizing Slu7 to promote exon ligation in 3′SS mutant pre-mRNA, it is likely that Prp22 also facilitates exon ligation in wild-type pre-mRNA, even though it is not strictly essential for this process. To test that possibility, we carried out splicing for 20 min using wild-type pre-mRNA in extracts depleted of both Slu7 and Prp22 to arrest the pathway prior to exon ligation. Recombinant Slu7 and Prp18 proteins were then added, with or without Prp22, and exon ligation was assayed. We observed that exon ligation proceeded rapidly upon additions of Slu7 and Prp18, and the inclusion of Prp22 accelerated the reaction by >50% within the first 2 min (Fig. 7A and B). These results indicate that Prp22 promotes exon ligation independently of the 3′SS sequence, likely by stabilizing Slu7 binding to the spliceosome. In contrast, Fyv6 only modestly enhanced exon ligation regardless of Prp22’s presence, or 3′SS context (Supplementary Fig. S3).

Figure 7.

Prp22 accelerates the exon ligation reaction of wild-type pre-mRNA. (A) Splicing was performed with ACT1 pre-mRNA for 20 min in extracts depleted of Slu7 and Prp22 (lane 1). Following re-addition of Slu7–Prp18 without (lanes 3–6) or along with (lanes 7–10) Prp22, the reaction mixtures were further incubated for 1–10 min. (B) Graph of the exon ligation reaction kinetics from panel (A), calculated as the molar ratio of mRNA to the total of pre-mRNA, lariat-IVS-E2 and mRNA in each reaction. Data represent mean values from three independent experiments, with SD indicated.

ATP inhibits exon ligation independently of its hydrolysis

Although Prp22 promotes exon ligation, ATP inhibits the reaction both in reaction mixtures and on purified spliceosomes. The mechanism underlying this inhibition remains unclear. Notably, ATP hydrolysis is not required for the inhibitory effect. Nonhydrolyzable ATP analogs, including ATPγS, AMP-PNP or ADP, produce similar inhibition (Fig. 8), indicating that ATP binding alone might induce a conformational change in Prp22 that impairs its ability to promote exon ligation.

Figure 8.

ATP hydrolysis is not required for inhibition of exon ligation. Splicing was performed with ACAC pre-mRNA for 20 min and spliceosomes were precipitated with anti-Ntc20 antibody. Spliceosome were further incubated for 20–60 min without (lanes 2–4) or with the addition of 2 mM ATP (lanes 5–7), ATPγS (lanes 8–10), AMP-PNP (lanes 11–13), or ADP (lanes 14–16).

Discussion

In addition to its well-established role in catalyzing the release of mature mRNA [24, 25], it has also been reported that Prp22 is required for exon ligation in introns with extended 3′ tails [22]. However, a study employing a reconstitution system indicated that Prp22 may not be essential for exon ligation, as additions of Prp16, Slu7, and Prp18 to the purified post-branching spliceosomes were sufficient to drive exon ligation [39]. Consistently, we observed in this study that depletion of Prp22 has a minimal effect on exon ligation and on Prp8 crosslinking to the 3′SS, whereas depletion of Slu7 or simultaneous depletion of both Slu7 and Prp22 results in the inhibition of exon ligation and abolishment of Prp8 crosslinking to the 3′SS (Fig. 1 and Supplementary Fig. S3). However, further kinetics analysis revealed that Prp22 accelerates exon ligation at early time-points (Fig. 7), an effect likely missed in previous studies due to employing longer assay durations. Collectively, these findings indicate that Slu7 is crucial for proper positioning of the 3′SS at the catalytic center of the spliceosome, whereas Prp22 plays an auxiliary role [1].

Mutations at the 3′SS impair exon ligation, leading to the accumulation of lariat-intron-exon 2 intermediates. Prolonged incubations resulted in marginal amounts of mRNA, which increased further if ATP is depleted from the reaction. The resulting mRNAs utilize multiple alternative 3′SS for exon ligation. This observation aligns with current models suggesting that ATP hydrolysis inhibits exon ligation in 3′SS-mutant pre-mRNAs. However, we found that exon ligation is blocked in the absence of Prp22, indicating that Prp22 actively promotes exon ligation of 3′SS-mutant pre-mRNAs by using aberrant 3′SSs. This role is consistent with Prp22’s function in accelerating the exon ligation of wild-type pre-mRNA (Fig. 7). This mechanism parallels that of Prp16, which is generally dispensable for the branching reaction [16], but promotes the use of multiple alternative BPs when the authentic site is mutated [14]. These findings suggest that Prp16 facilitates, rather than prevents, aberrant BS selection in BP-mutated pre-mRNAs. We previously demonstrated that Prp16 promotes branching by stabilizing Cwc25 binding to the mutant pre-mRNAs [36]. Similarly, we have shown that Prp22 stabilizes Slu7 association with the spliceosome prior to exon ligation. Thus, by analogy, Prp22 likely facilitates exon ligation of 3′SS mutant pre-mRNAs by promoting stable Slu7 binding [2].

Recently, Fyv6 and its human counterpart FAM192A were identified as novel second-step factors that facilitate the selection of distal 3′SS [41, 47, 56]. Since Slu7–Prp18 and Prp22 are also differentially required for exon ligation of pre-mRNA with longer 3′ tails [21], accurate identification of the appropriate 3′SS is likely a crucial and shared function of second-step factors. In contrast to the first-step conformation, where first-step factors such as Yju2 and Cwc25 insert their N-termini into the catalytic core of the spliceosome to stabilize the 5′SS–BS interaction, all second-step factors are generally positioned outside the catalytic center, suggesting a different mechanism for stabilizing the 5′SS–3′SS interaction. The notable exception is the C-terminal tail of Prp22, which has been observed in some cryo-EM structures of the P complex, as extending along the inner face of the RT domain into the catalytic core, where it reaches and contacts the Linker domain and the α-finger [40, 41]. Notably, the C40 domain of Prp22 was not resolved in any of the C* complex structures. Previously, we observed strong 4sU crosslinking of Prp22 to the +8 position of the i3’T, as well as to the 3′ exon, in both C* and P complexes [46]. We speculate that this observed crosslinking at the +8 position reflects an interaction between the C40 domain of Prp22 and the intron, indicating that the C40 domain may be positioned within the catalytic center of the spliceosome even before exon ligation. However, deletion of the C40 domain does not completely abolish Prp22 function. Instead, it weakens Prp22’s association with the spliceosome (Supplementary Fig. S4A), substantially impairing its ability to promote exon ligation for both wild-type and 3′SS mutant pre-mRNA (Supplementary Fig. S4B). In contrast, its role in catalyzing mRNA release appears to be less severely affected (Supplementary Fig. S4D), suggesting that the primary function of the C40 domain is to stabilize the 5′SS–3′SS interaction and facilitate exon ligation [3].

Crosslinking analysis indicates that Prp8 is the only protein contacting both the 5′SS and the 3′SS in the C* and P complexes [46], suggesting that the second-step factors may function by stabilizing Prp8’s interaction with the splice sites. Slu7 adopts an extended conformation, making extensive contacts with multiple domains of Prp8, including its N, Linker, Endo, and RH domains, on the surface opposite the catalytic center of the spliceosome [40, 41, 49–51, 56–59]. Notably, two regions of Slu7, residues 158–196 and 241–290, are positioned at the interfaces between the N and Endo, and between the Endo and RH domains, respectively. Slu7 and Prp18 form a heterodimeric complex that associates with and dissociates from the spliceosome as a unit. In addition to interacting with Slu7, Prp18 also interacts with the Linker and RH domains of Prp8, nestling at the interface of these domains. These interactions indicate that the Slu7–Prp18 complex might help stabilize the structural integrity of Prp8 by reinforcing its domain interactions, thereby supporting proper positioning and engagement of the 3′SS. In the absence of Slu7, Prp8 domain interactions may become more dynamic, potentially disrupting stable alignment of the 5′SS and 3′SS required for exon ligation, as well as the retention of mRNA on the spliceosome following exon ligation. In this scenario, Slu7 may dissociate from the spliceosome before mRNA release, driven by Prp22 activity. Recent cryo-EM analysis of the P complex, using a helicase-deficient mutant of Prp22 to block mRNA release revealed distinct structural populations that lacked Slu7 and Fyv6, or exhibited more extensive loss of Slu7 density, while retaining Prp22 and mRNA [41]. These structures may represent intermediates in the process of mRNA release. Although efficient mRNA release requires both the ATPase and RNA helicase activities of Prp22, the mutant may weaken the interaction between Slu7 and Prp8 following ATP hydrolysis, leading to an incomplete release process [4].

Cryo-EM structures have also revealed that Prp22 occupies a similarly peripheral position relative to the catalytic center in both the P and C* spliceosome complexes [40, 41, 49–51, 56–59]. In those structural studies, the P complexes were stalled by using Prp22 mutants deficient in ATPase or helicase activity [40, 49, 50], whereas the C* complexes were generated by using modified 3′SS. These alterations likely arrest Prp22 in a pre-functional state, poised for action but impeded due to mutations at the 3′SS or within Prp22 itself. The structures also show that the 5′ exon and 3′ exon of the mature mRNA protrude outward from the spliceosome’s catalytic center through interfaces between the N and Linker, and between the Linker, Thumb, and RT domains of Prp8, respectively. We have shown herein that Cwc22 and Prp22, proteins that bind to the 5′ and 3′ exons, respectively, both remain associated with the mRNA following its release from the spliceosome. This suggests that Prp22 is unlikely to extract the mRNA directly from the catalytic center, which would unavoidably displace Cwc22. Instead, ATP hydrolysis by Prp22 may trigger conformational changes in Prp8 that disrupt its interaction with Slu7, Fyv6, and the mRNA. Consequently, the mRNA likely exits through the side of the Prp8 Linker domain opposite the spliceosome’s catalytic center, allowing Prp22 and Cwc22 to remain bound as the Prp8 domains assume a relaxed state. Notably, one recent structure of the yeast P complex resolved a part of the Prp22 N-terminal domain (amino acid residues 386–436) that exhibits a long helix bridging the Prp22 RecA2 domain and the C-terminal domain of Cwc22, the latter of which binds the 5′ exon [41]. This interaction may contribute to stabilizing the association of Prp22 and Cwc22 with the mRNA following their dissociation from the spliceosome [5].

Prp22 is thought to mediate mRNA release by docking onto the 3′ exon of the mRNA and translocating towards the catalytic center via ATP hydrolysis [25], a process accompanied by Slu7 dissociation. Cryo-EM structures of the spliceosome P complex confirm that Prp22 binds to the 3′-exon from a peripheral position, making contacts with the RT and Linker domains of Prp8, and its C-terminal tail extends into the catalytic center [40, 49, 50]. 4sU-crosslinking analyses have shown that Prp22 and Slu7 crosslink to an overlapping region of the 3′ exon, both before and after exon ligation [46]. However, Slu7 was not detected to directly interact with the exon or Prp22 in these structures. How Slu7 is stabilized by Prp22 prior to exon ligation or how it dissociates upon Prp22 action after exon ligation are not understood. Given that only ∼70% of the Slu7 sequence is resolved in these structures, it remains possible that the unresolved regions participate in additional interactions [6].

This study demonstrates that ATP hydrolysis does not trigger Slu7 release from the spliceosome prior to exon ligation. Instead, Prp22 promotes exon ligation of 3′SS mutant pre-mRNAs, enabling the selection of alternative 3′SSs. Prp22 functions potentially through stabilizing Slu7 association and, in turn, enhancing the interaction of Prp8 with the splice sites. Alternatively, Prp22 may act directly on Prp8 to stabilize its interaction with the splice sites. Notably, it is ATP binding, rather than hydrolysis, that inhibits the ligation reaction. These findings not only contradict the proposed proofreading role of Prp22 at the 3′SS, but also suggest that Prp22 facilitates usage of aberrant 3′SS in mutant pre-mRNAs, with ATP binding acting as a negative regulator of the ligation step. The precise mechanism by which ATP binding inhibits exon ligation remains unclear. One possibility is that ATP-bound Prp22 adopts a conformation that weakens its interaction with Prp8, thereby disrupting the alignment of the 5′SS and 3′SS for efficient exon ligation [7].

Notably, Prp22 is located at a similar peripheral position in the spliceosome both before and after exon ligation, with the overall architectures of the C* and P complexes being greatly alike. Our 4sU-crosslinking data also reveal highly similar crosslinking patterns of Prp8, Prp22, and Slu7 to the i3’T and 3′ exon regions in both complexes [46]. However, ATP hydrolysis promotes Slu7 release from the P complex, but not from the C* complex. This suggests a subtle conformational change, unresolved by the structural or crosslinking analyses, that modulates the activity of Prp22. Prior to exon ligation, ATP binding by Prp22 may only moderately destabilize Slu7, with no significant structural changes occurring following ATP hydrolysis. In contrast, after exon ligation, ATP hydrolysis by Prp22 triggers major structural rearrangements in the spliceosome, leading to release of the mRNA and Slu7. This marked functional difference is likely driven by subtle conformational changes in the spliceosome that occurs upon exon ligation. However, the precise mechanism underlying this process remains to be elucidated [8].

Our hypothesis that Prp22 mediates mRNA release through the Prp8 domains challenges the conventional “pulling” model. Prp22 is proposed to bind to the 3′ exon and translocate toward the spliceosome’s catalytic center to disrupt the interaction of the mRNA with the spliceosome [25]. A similar translocation mechanism has been observed for Prp16 during the second catalytic step, based on mapping Prp16-binding sites on the i3’T at different stages of its action [46]. Both Prp22 and Prp16 stall at the periphery of the spliceosome outside the catalytic center prior to exerting their function. Likewise, Prp2 binds to the i3’T at a comparable position during the first catalytic step, potentially loading onto the pre-mRNA via its prior interaction with Brr2 [8]. These localizations are supported by the cryo-EM structures [11, 40, 49, 50, 57, 60, 61]. Despite the conserved positions, how these DEAH box helicases displace their targeted components remains unclear. A pulling model has been proposed for Prp16 and Prp22 for promoting spliceosome rearrangement by anchoring at the periphery and pulling the RNA to disrupt its interactions with the spliceosome [62]. However, structural data show that the 5′ and 3′ exons of the mature mRNA exit the catalytic center through distinct regions of Prp8, making it unlikely that the mRNA could be pulled out by Prp22 from the catalytic center side without displacing proteins bound to the 5′ exon. Our finding that Cwc22 remains associated with the mRNA after release strongly argues against such a mechanism for mRNA release. Instead, we propose that Prp8 undergoes a conformational change that opens its structure, allowing the mRNA to exit on the side opposite the catalytic center and thereby retain Cwc22 on the released mRNA [9].

Human Cwc22 functions as an EJC adaptor by binding to the 5′ exon of mRNA within the spliceosome [52–54]. In yeast, Cwc22 is essential for cell viability and is required during the Prp2-mediated catalytic activation step of splicing [48]. Despite the absence of EJC in yeast, Cwc22 interacts with the 5′ exon in a manner similar to its human counterpart. This conservation suggests that yeast Cwc22 may represent an evolutionary remnant of the EJC adaptor. Our finding that Cwc22 remains associated with the released mRNA after splicing further supports a potential role in coordinating with the EJC for subsequent regulatory functions [10].

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

Academia Sinica and National Science and Technology Council (Taiwan) (111-2311-B-001-008-MY3). Funding to pay the Open Access publication charges for this article was provided by Academia Sinica.

Data availability

The data underlying this article are available in the article and in its online supplementary material.