RNAs in the spliceosome: insight from cryoEM structures

Overview

Pre-mRNA splicing is essential for gene expression in all eukaryotes. In mammals, an average of 95% of the nucleotides in the primary transcript (pre-mRNA) of a protein-encoding gene are introns ¹. These introns need to be removed precisely by splicing before the mRNA can be transported out of the nucleus and translated. Alternative splicing greatly expands the gene coding capacity and is a fundamental component of eukaryotic gene regulation, influencing cell differentiation ², development ³, and many processes in the nervous system ⁴. Over 90% of human genes are alternatively spliced ⁵. Errors in splicing contribute to at least 30% of human genetic disorders ⁶. Aberrant splicing also plays a significant role in the onset and development of many other diseases including cancer ^6,7.

A typical intron contains a conserved 5ʹ splice site (5ʹ ss), a Branch Point Sequence (BPS) followed by a Polypyrimidine Tract (PYT), and a 3ʹ splice site (3ʹ ss) ^8,9. Introns are removed through two transesterification reactions catalyzed by the spliceosome, a huge RNA-protein complex. The spliceosome contains five small nuclear RNAs (U1, U2, U4, U5, and U6 snRNAs) that form five snRNPs with their associated proteins, plus numerous other protein splicing factors (the total number of proteins in the spliceosome is well above 100) ^10,11.

Spliceosome components assemble on the pre-mRNA in a step-wise manner (Fig. 1). The formation of the E complex involves the initial recognition of intron components by the spliceosome. In mammalian cells, the 5ʹ ss is recognized by U1 snRNP, while the BPS, PYT, and 3ʹ ss are recognized by SF1, U2AF65, and U2AF35, respectively ^12–15. The S. cerevisiae (yeast) 5ʹ ss, BPS, and the PYT are recognized by U1 snRNP, BBP, and Mud2, respectively ^16–20, and there is no obvious spliceosomal component recognizing the 3ʹ ss at this stage. How the splicing machinery accurately defines introns and exons in a large variety of pre-mRNAs is one of the most fundamental questions in pre-mRNA splicing. In lower eukaryotes whose pre-mRNAs typically contain small introns and large exons, the intron definition model (spliceosome assembly across an intron) seems to dominate ²¹. On the other hand, the exon definition model prevails in higher eukaryotes where small exons and large introns are prevalent ²¹. However, currently there is no direct structural evidence of either model and the molecular details governing these models are unclear.

Following initial intron recognition, the U2 snRNP replaces SF1 to form the A complex, assisted by RNA helicases Prp5 and Sub2 (UAP56 in higher eukaryotes) ^22,23. The U4/U6.U5 tri-snRNP then joins the spliceosome to form the Pre-B complex ²³. Helicase Prp28 disrupts the interaction between U1 snRNA and 5ʹ ss, releasing the U1 snRNP and forming the B complex ^24,25. Helicase Brr2 then unwinds the U4 and U6 snRNAs, releasing the U4 snRNP ²⁶. The NineTeen Complex (NTD) and NTC-related complex (NTR) join the spliceosome at this stage, stabilizing the newly formed catalytic core and forming the B^act complex ²⁷. Helicase Prp2 further rearranges the spliceosome to form the catalytically active B* complex that is ready for the first step reaction ²⁸. Following the first transesterification reaction (also referred to as the branching reaction), the C complex emerges, encompassing the newly freed 5ʹ exon and a lariat intermediate. The C complex further rearranges through the action of helicase Prp16 to form the C* complex, primed for the second transesterification reaction (also referred to as the ligation reaction) ^29,30. In the ligation reaction, the two exons are joined together, forming the post-catalytic P complex that contains the ligated exons and the lariat. The exons are then released through the action of helicase Prp22 ³¹, generating the intron-lariat spliceosome (ILS) complex, which only contains the lariat. ILS is further disassembled through the activity of the helicase Prp43 ^32,33.

The recent resolution revolution in cryoEM has brought about a slew of high-resolution structures of the spliceosome related complexes (including snRNPs), a previously unthinkable feat in the field (Table 1). The first high resolution cryoEM structure of the spliceosome was the S. pombe ILS complex determined by the Shi lab in 2015 ³⁴. In the following years, high resolution structures of the yeast U1 snRNP³⁵, tri-snRNP ^36,37, A ³⁸, pre-B ³⁹, B ⁴⁰, B^{act 41}, C ^42,43, C*,^44,45, P ^46–48, and ILS ⁴⁹ complexes, as well as the human B ⁵⁰, B^{act 51,52}, C ⁵³, and C* ^54,55 complexes were determined. These structures provided unprecedented details on the organization of RNA and protein components in the spliceosome and their functions in the splicing reaction. Below we will highlight insight into the structure and function of spliceosomal RNA components obtained from these cryoEM structures, with a focus on the yeast spliceosome.

The spliceosome is a ribozyme

Biochemical and genetic studies have long fueled the speculation that the spliceosome, like the ribosome, is a ribozyme (reviewed in ^56,57). In 1982, Cech and colleagues described the ribozyme as a complex RNA secondary structure that recognizes specific RNA sequences and catalyzes the splicing reaction within their active center ⁵⁸. Among the different classes of ribozymes, the group II introns catalyze RNA self-splicing through two sequential transesterification reactions, where the introns excise themselves from surrounding exons and stitch the resulting pieces back together ⁵⁹. When Sharp and others described the mRNA precursor splicing process through an intermediate lariat structure, the speculation that the spliceosome and group II introns shared a common evolutionary mechanism arose ^60–63. Since then, numerous biochemical and genetic studies have shed light on the possible catalytic roles of snRNAs in the splicing reaction. For example, the U2/U6 helix 1 and the U5 snRNA loop 1 configure the 5ʹ splice site for cleavage and play a role in exon ligation ^64–66 whereas U6 snRNA intra-stemloop (ISL) and AGC triad seem to be the principal actors in catalysis ^62,67. Five non-bridging oxygens in U6 (one each in A59, G60, G78 and two in U80) are metal ligands and may be involved in either branching, or exon ligation by stabilizing the leaving groups, or both ^68–70. Although such evidence supports the spliceosome being a ribozyme, the definitive proof of whether the spliceosome uses a RNA-based catalytic mechanism requires the visualization of the catalytic site of the spliceosome.

The recent high-resolution cryoEM structures of the spliceosome provided just such visualization, especially with the structures of the B^act, C, C*, and P complexes which capture the spliceosome before the first step reaction, after the first step reaction, before the second step reaction, and after the second step reaction, respectively. In the structure of the B^act and C* complexes, the branch point and the 3ʹ ss are not yet loaded and cannot be observed structurally. On the other hand, RNA components of the branching and ligation reaction have not yet left the active site in the C and P complexes after the reaction, providing the closest snapshots of various players in the two reactions.

The structure of the C complex reveals density likely corresponding to the two Mg²⁺ ions required for catalysis, M1 and M2, which are coordinated by nucleotides in U6 snRNA (M1 by G78 and U80, M2 by A59 and U80) ^42,43 (Fig. 2a). M2 is about 6 Å away from the lariat junction in the C complex (which may help prevent the reversal of the branching reaction) and it is likely much closer to the branch point right before the branching reaction (for example, in the B* complex whose structure is not yet known). Before the branching reaction, M2 can foreseeably activate the 2ʹ OH group of the branch point A for nucleophilic attack of the phosphodiester bond in the 5ʹ ss, while the M1 ion stabilizes the leaving group (Fig. 2a). The structure of the P complex reveals that the M1 ion is in a position to activate the 3ʹ OH group of 5ʹ ss, which attacks the 3ʹ ss, ligates the two exons and releases the intron ^46–48 (Fig. 2b). The M2 ion is not consistently observed in all three P complex structures, possibly reflecting the post-catalytic nature of the complex.

Although the spliceosome is clearly a ribozyme just like the ribosome, the spliceosome is much more protein rich compared to the ribosome. Of these proteins, Prp8 and Snu114 from the U5 snRNP, Cef1 and Syf2 from NTC, as well as Bud31, Cwc2, Cwc15, Ecm2, Prp45, and Prp46 from NTR form a protein core surrounding the active site that is fairly stable from the B^act to ILS complex (Fig. 3). Prp8, in particular, is in the heart of the spliceosome and makes extensive contacts with the snRNA and pre-mRNA. Prp8 is one of the largest (Prp8 from all species are over 2,000 amino acids in length) and most conserved protein in the nucleus, with 62% sequence identity between its yeast and human homologs. Previous biochemical and genetic experiments demonstrate that Prp8 is crosslinked to all key elements of the pre-mRNA involved in the splicing reaction and many Prp8 mutants either exacerbate or suppress pre-mRNA splicing mutants (reviewed in ⁷¹). Prp8 is composed of the N-terminal domain, the large domain (which can be further divided into the helix bundle, reverse transcriptase-like (RT), thumb/X, linker, and endonuclease-like (EN) domains), the RNase H domain, and the MPN domain ^34,72. The N-terminal, linker, and RT domains of Prp8 form a cavity that cradles the RNA active site. Multiple structural elements, including the alpha finger (residues 1570–1615), the switch loop, and the RNase H domain containing the beta finger change conformation dramatically in different complexes, often help to stabilize and guide the precise conformation of RNA in different stages of the splicing cycle (reviewed in ⁷³).

Spatial organizations of the snRNAs and Pre-mRNA through the splicing cycle

From the initial recognition of pre-mRNA by the U1 snRNA at the 5ʹ ss (E complex), several dramatic changes involving the snRNA occur to achieve spliceosomal activation. These changes include the recognition of the BPS through basepairing with U2 snRNA in the A complex, the joining of the tri-snRNP to form the pre-B complex, the removal of U1 snRNP to form the B complex, and the removal of U4 snRNP to form the B^act complex. Once the spliceosome is activated (starting from the B^act complex), although the 3ʹ end of U2 snRNA changes positions to a certain extent, the snRNA core surrounding the active site (U2/U6 helix 1 and 2, the U2 snRNA region that basepairs with the BPS, the U6 intra-stemloop (ISL) and ACAGAGA box, and U5 snRNP loop 1) remains largely unchanged through the rest of the splicing cycle (reviewed in ^73–77) (Fig. 4).

On the other hand, the pre-mRNA undergoes dramatic movements through the two transesterification reactions (Fig. 5). In the B^act complex, the branch point A is about 50 Å away from the 5ʹ ss. The branch point A presumably moves close to the 5ʹ ss in the B* complex and carries out the branching reaction. The intronic RNA downstream of the BPS remains in the active site in the C complex, which is moved away in the C* complex to make room for the incoming 3ʹ ss. The 3ʹ ss is then loaded into the active site for the ligation reaction, which remains in the active site after exon ligation in the post-catalytic P complex. Ligated exons are then released, forming the ILS.

The spliceosomal helicases are largely responsible for the movement of the pre-mRNA, either directly or indirectly. The pre-mRNA splicing process is remarkable in the large number of DExD/H-box RNA helicases required in the splicing cycle (Fig. 1). Four of these RNA helicases (Prp5, Sub2, Prp28 and Brr2) are involved in spliceosomal assembly and activation in early stages of the splicing cycle, up to the B^act complex. Interestingly, the four helicases (Prp2, Prp16, Prp22, and Prp43) in the later stages of the splicing cycle all belong to the DEAH-box sub-family with their helicase motif II taking the form of DEAH and the four proteins share similar domain organizations ^33,78–80. Prp2 is responsible for removing SF3b and exposing the BPS in preparation for the first catalytic reaction in the transition from the B^act to the B* complex ²⁸. Prp16 vacates the space near the active site in the C complex so that the 3ʹ ss can enter ²⁹. Prp22 acts in the P complex to release ligated exons ³¹, and Prp43 helps disassemble the ILS ^32,33. These helicases are recruited to specific stages of the spliceosome and discarded after their dedicated actions. For example, Prp2 is recruited to B^act, and released from B^act after ATP hydrolysis, allowing Prp16 to bind. The action of Prp16 induces the dissociation of Prp16 and a number of step one factors, allowing Prp22 and step two factors to bind and form C*. Prp22 remains bound in the P complex, whose action releases ligated exons and Prp22 is released with the exon, allowing Prp43 to bind. These observations suggest that the helicases have to be specifically recruited to the right place at the right time and their activities tightly regulated. In addition, it is clear from the P complex structure that Prp22 binds to the single stranded 3ʹ end of the 3ʹ intron, and pulls it from the 3ʹ to 5ʹ direction. This observation is consistent with previous biochemical observations that Prp22 uses a winching mode to fulfill its function in kinetic proofreading ⁸¹, and suggests that translocation on the RNA instead of unwinding by these helicases is important for their functions in splicing. It is interesting to note that three of the four helicases that are resolved in the cryoEM structures all bind to the periphery of the spliceosome and close to the 3ʹ end of their potential target RNAs (Prp2 is close to the 3ʹ end of intron ⁴¹, Prp22 binds to the 3ʹ end of 3ʹ intron ^46–48, and Prp43 is close to intron/U6 snRNA ⁴⁹). It is possible that these helicases fulfill their function through a similar mechanism of action.

RNA secondary structures in pre-mRNA

One intriguing observation in the P complex structure is that the region between the BPS and 3ʹ ss forms a stem-like secondary structure ⁴⁶ (Fig. 6). This secondary structure could potentially serve to bring the 3ʹ ss close to the 5ʹ ss and BPS (which are in close proximity of each other after the branching reaction) to facilitate the recognition of the 3ʹ ss and subsequent ligation reaction. Almost all the yeast introns with long BPS to 3ʹ ss distances are predicted to form some secondary structures ⁴⁶. The distance between the 5ʹ ss and BPS is typically much longer. It would not be surprising if secondary structures form in this region as well to bring the BPS and 5ʹ ss close together and facilitate the branching reaction.

The existence of secondary structures in introns of isolated genes that are important for splice site selection have previously been observed through biochemical and genetic experiments (reviewed in ⁸²). For example, intramolecular secondary structures in specific yeast introns are essential for spliceosome assembly and 5ʹ splice site selection ^83–85. Studies performed on UBC13 and RPS20B with long branch point (BP) to 3ʹ ss distances demonstrated impaired splicing when the proposed secondary structures are disrupted ^86,87. The structure of the P complex provided the first physical evidence of such secondary structures and points to the potentially general role of these secondary structures in facilitating splice site selection. Multiple methods have been developed in recent years to probe the genome-wide secondary structures of RNAs ⁸⁸, and these approaches may also be used to provide insight into the extent of secondary structures in all yeast introns and their functions. In addition to facilitating splicing by bringing important intron elements close together, secondary structures can also inhibit splicing by masking splice sites or blocking the binding of splicing factors ⁸⁹. Not surprisingly, secondary structures have been found to regulate alternative splicing in higher eukaryotes ⁹⁰ and can potentially be a point of modulation by small molecules in the future ⁸⁹.

RNA interactions in splice site recognition

The 5ʹ ss of pre-mRNA is largely recognized through Watson-Crick basepairing. It is initially recognized through basepairing with the 5ʹ end of the U1 snRNA, as can be seen in the structure of the A complex ³⁸ (Fig. 7a). During the spliceosomal activation process, U1 snRNP is removed, and the 5ʹ ss forms Watson-Crick basepairs with U6 snRNA instead in the B^act complex ⁴¹ (Fig. 7b). The BPS is initially recognized by the Branch Bridging Protein (BBP)¹³ and is then handed over to U2 snRNA which recognizes the BPS through Watson-Crick basepairing ³⁸ (Fig. 7c).

The CryoEM structure of the post-catalytic P complex reveals that the 3ʹ ss is mainly recognized through non-Watson-Crick basepairing between the 3ʹ ss and the 5ʹ ss as well as the BPS (Fig. 7d). The last nucleotide of the intron (G in the 3ʹ ss) forms a non-Watson Crick basepair with the first nucleotide of the intron (G in the 5ʹ ss), and the penultimate nucleotide of the intron (A upstream of the 3ʹ ss) forms another non-Watson Crick basepairing with the branch point A. These nucleotides are surrounded mainly by the U6 snRNA, Prp8, and Prp18.

The above structural observation helps explain the strong preference for the AG dinucleotide at the end of introns. In yeast, the last two nucleotides of all introns are strictly AG ⁹¹ and in humans, they are predominantly AG with some low frequency non-canonical 3ʹ ss ⁹². In the P complex structure, substituting A or G in the dinucleotide to the other three bases loses all or partial hydrogen bonds with the 5ʹ ss G (barring potential conformational changes around these nucleotides) (Fig. 7e, f), potentially explaining the conservation of the AG dinucleotide at the 3ʹ ss.

In summary, recent high resolution cryoEM structures revealed unambiguously that the spliceosome is a ribozyme. The U2, U5, and U6 snRNAs form a fairly constant active site core stabilized by proteins such as Prp8 once the spliceosome is activated. The pre-mRNA moves significantly to bring key elements to the active site for catalysis, and most of these movements are facilitated by RNA helicases. Both RNA structures and RNA-RNA interactions likely play important roles in splice site recognition. Although the core of the spliceosome generally has high resolutions in the cryoEM structures determined to date, resolutions of regions that are more flexible, like those in the periphery of the spliceosome, are typically much lower. For example, the helicases such as Prp2, Prp16, Prp22, and Prp43 that play important roles in RNA remodeling and spliceosome assembly/activation are located in the periphery of the spliceosome. Prp16 was not observed in the C complex structure, and the other three helicases also have much lower resolution than the core of the spliceosome, hindering a detailed mechanistic understanding of their action. As sample preparation and calculation methods continue to improve, we will likely to gain more high-resolution information on the RNAs in the spliceosome and the proteins that interact with them.