Cryo-EM structures of different catalytic states in the Cte 1 intron splicing process
The Cte 1 RNA construct was prepared by in vitro transcription (IVT), during which the catalytic reactions have already occurred after cotranscriptional folding in the presence of Mg2+ (Extended Data Fig. 1a)17,21,22. We did not observe the BI band in denatured gel electrophoresis suggesting that both first and second catalytic reactions happened almost simultaneously, analogous to canonical group II intron splicing8,23. Cryo-EM analysis of the IVT product yielded reconstruction of the branched product at 2.90 Å resolution with the A127-G220 2'-5' phosphodiester linkage and D6 pointing towards D2, corresponding to the post-2S state with circular exons released (Fig. 1c, Extended Data Fig. 1b, Extended Data Table 1).
To capture the branching (1S) and ligation (2S) states, we purified the Pre RNA and replaced Mg2+ with Ca2+ in catalytic reactions that has previously been shown to inhibit splicing 8,23-25. Denatured gel electrophoresis showed accumulation of BI as splicing reaction proceeded, the 5-minute and 4-hour time points were subjected to cryo-EM analysis (Extended Data Fig. 2a). Three-dimensional (3D) classification and refinement yielded three reconstructions at 2.65 Å, 2.97 Å and 2.90 Å resolution, representing pre-1S, 1S and 2S states, respectively (Fig. 1c, Extended Data Fig. 2b, Extended Data Table 1). The 2S structure adopts almost identical conformation as the post-2S with extra density of 5'-exon bound to EBS1 and 3'-exon bound to EBS3. The pre-1S and 1S structures have the D6 helix translocated 65º away from the 2S and post-2S conformations, positioning the 2'-OH of bulged A127 in close proximity to the scissile phosphate at the 5'-SS. The 1S structure is distinguished from the pre-1S because bulged A127 in 1S is ready to cleave the docked 5'-exon.
Overall architecture and novel tertiary interactions that allosterically affect intron catalysis
The Cte 1 intron consists of novel P1 and P2 domains at the 5' end followed by D5 and D6 that establish the catalytic core, the intervened 3'- and 5'-exon await to be excised, and the scaffold D1-D3 with D4 at the 3' end forming a closing stem (Fig. 1d, Extended Data Fig. 3). P1 and P2 are found generally conserved and predicted to form pseudoknot (PK) interactions based on comparative genomics analysis14,26. Our Cte 1 cryo-EM structures reveal that P1 forms a PK with the D2 terminal loop, whereas P2 forms another PK with a terminal loop at the 3' end of D3 (Γ'-Γ and Δ'-Δ) (Fig. 2a-b).
The following D5 contains the conserved AGC triad and 2-nt bulge that collectively form a catalytic triplex core homologous to the spliceosomal U2/U6 small nuclear RNA1,27. Multiple tertiary interactions are identified between D5 and D1 (ψ'-ψ, λ'-λ, ζ'-ζ)28,29, D3 (κ'-κ, μ'-μ) and D4 (φ'-φ) in order to organize the catalytic core 30,31, consistent with those previously observed in canonical group II introns (Extended Data Fig. 4)9,22,32,33. D6 is downstream of D5 that contains the branchpoint A127 and a few previously identified tertiary interactions with D1 (π'-π) and D2 (η'-η, γ'-γ) that facilitates D6 dynamics between different catalytic states (Extended Data Fig. 5a-c)8,9,32,34. The 3'- and 5'-exon are presented next with intron binding sites (IBS3 and IBS1) that recognize EBS in D135,36.
The subsequent D1 (including D1a-D1d subdomains) forms numerous interactions with all domains of the intron to serve as the scaffold for correct folding, exon recognition, and catalytic core organization. Novel interactions have been observed in the peripheral regions, including three GAAA tetraloop/tetraloop receptor (TL/TLR) interactions between D1 and D1c (ξ-ξ') and D1b and D1c (Α-Α' and Β-Β'), a base triple between D1c and D2 (υ-υ'), and a PK between D1d and D3 (ο-ο') (Fig. 2c-g). Mutations that disrupted individual novel tertiary interactions in the peripheral regions resulted in significantly reduced catalytic activity compared to the wild-type (WT) sequence except for Δ'-Δ, suggesting that most of these tertiary interactions in the peripheral regions have allosteric effects on CP intron catalysis (Fig. 2h, Extended Data Fig. 5).
Additional interactions conserved in group II introns include D1c and D2 (θ-θ') interaction in order to secure the critical ε-ε' region in position for 5'-SS selection and catalysis8,37, the peripheral interaction between D1b and D1d (α-α')38, and interactions within D1d (ω-ω' and σ-σ') to reinforce positioning of D5 catalytic core (Extended Data Fig. 5d-h)9,22. D2 and D3 next to D1 are crucial for intron catalysis efficiency by forming the aforementioned interactions with D5 and D632,39. Moreover, junction J2/3 directly interacts with the catalytic triad in D5 to form the catalytic triplex, which is reinforced by an interaction between D2 and D3 (ρ-ρ') (Extended Data Fig. 5i)32,40,41. The 3' end of Cte 1 intron forms a short D4 as the closing stem with no IEP sequence. All interactions have been summarized in Extended Data Table 2.
Conformational dynamics of branchpoint A127 and the branching helix D6
Cryo-EM structures reveal substantial conformational changes and local shifts from branching to exon ligation of Cte 1 intron back-splicing. Different states are designated based on the connection of 5'-SS to IBS1 (pre-1S and 1S), the presence of both EBS1-IBS1 and EBS3-IBS3 ready for exon ligation (2S), and the A127-G220 2'-5' phosphodiester linkage density (2S and post-2S) (Extended Data Fig. 6). Previous studies have reported that branchpoint A could undergo dynamics either forming a 1-nt or 2-nt bulge 42,43. In 1S state, we observed the 1-nt bulged A127 pointing outside of D6 with 2'-OH in close proximity of 2.62 Å to 5'-SS scissile phosphate ready for the branching reaction, and further stabilized by stacking with 5'-SS G220 and forming a base triple with G101-C125, analogous to the bulged A recognition pattern recently reported in prebranching structures of both canonical group II introns and spliceosomes (Fig. 3a-c)8,9,44,45. In pre-1S state, the inactive A127 flips inside D6 and forms a non-canonical base-pair with C538 from the EBS loop while remaining stacking with G220, leading to local shifts of surrounding nucleotides in the bulge (Fig. 3d). In 2S and post-2S, A127 branched with G220 is translocated 20 Å away, facilitated by a 65º conformational change of D6, in order to allow docking of the 3'-exon in the catalytic core ready for ligation (Fig. 3e-f).
The branching helix D6 has been previously reported to undergo ~90º translocation enabled by IEP between two transesterification reactions in both self-splicing and transposition processes 8,9. In Cte 1 intron, the conformational change of D6 between first and second reaction is mostly facilitated by the η' interaction motif of D6 terminal T-loop. In pre-1S and 1S, we observed tertiary interaction anchored by a G111-A451 non-canonical base-pair between D6 T-loop and the terminal loop of Β' interaction motif in D1c, designated as η'-η'' interaction (Fig. 3g). After branching reaction (2S and post-2S), D6 translocates to allow 3'-exon docking enabled by interactions with D1c (π'-π) and D2 (η'-η) as previously observed in canonical group II introns (Fig. 3h-i), which also leads to local dynamics of the peripheral regions (Extended Data Fig. 7)8,9,23,32,33,46,47.
Unexpected major groove interaction that stabilizes EBS1-IBS1
Three exon binding sites (EBS1-3) have been previously identified to complementary base-pair with IBS1-3 in exons (EBS-IBS) in different classes of canonical group II introns35. The Cte 1 intron contains only EBS1 and EBS3 that is identical to group IIC introns1. EBS1 is found in a terminal loop in D1d that forms the 6-bp EBS1-IBS1 including a A543-A216 sheared pair in all structures except for post-2S (Extended Data Fig. 6a-d). In 2S when 5'- and 3'-exon are aligned for ligation, the single base-paired EBS3-IBS3 (G539-C134) that positions the 5' end of 3'-exon for transesterification is observed, also designated as δ-δ' interaction in group IIA introns 1 (Extended Data Fig. 6c). Intriguingly, a 3-nt strand (C535-G537) upstream of EBS3 in the same EBS loop forms a major-groove interaction of three base triples reinforcing the EBS1-IBS1, which is usually facilitated by IEP in canonical group II introns (Fig. 3j-l)8,9,23,33,47. After release of the circular exon product in post-2S, the EBS loop undergoes local rearrangement for stabilization by forming two internal base triples, C535-G537-C545 and U536-G539-A543 (Fig. 3m).
Metal ions enable folding and catalysis in Cte 1 intron
Metal ions are essential for ribozyme folding and catalysis48,49, and previous studies have systematically evaluated 32 monovalent metal ion-binding sites (K+) and 34 divalent metal ion-binding sites (M2+) in group IIC intron crystal structures25,50. More than 60 metal ions were identified in each cryo-EM structure of different states in Cte 1intron splicing, including designation of 7 K+ and 8 M2+ ions analogous to those previously identified in the group IIC intron50. The rest metal ions are numbered M35-M86 as K+ and M2+ are indistinguishable in cryo-EM density under such resolution (Extended Data Table 3). Similar to other group II introns, the majority of metal ions resides in D1, suggesting that correct folding of D1 is critical to precede proper folding of the entire Cte 1 intron (Extended Data Fig. 8a)51,52. A metal ion M64 is found to reside in the major groove of the newly identified ο-ο' interaction (Fig. 2g). In addition, we observed M5, K3 and K4 close to the catalytic core but not directly interacting with the reaction residues, which potentially have allosteric effects in catalysis by stabilizing the catalytic core as previously described (Extended Data Fig. 8b)50.
The catalytic core of Cte 1 intron is organized in D5 with J2/3 and a 2-nt bulge that crucially binds to divalent metal ions M1 and M253,54. In pre-1S prior to the branching reaction where bulged A127 flips away from the catalytic site, M1 and M2 coordinate with non-bridging oxygens of A60 and G61 in the catalytic triad, G83 in the 2-nt bulge, and C81 adjacent to λ'-λ interaction to establish the catalytic core. In the meantime, coordinations of M1 and M2 with G220 pro-Rp, and M2 with U219 bridging oxygen stabilize the leaving group (Fig. 4a). Additionally, K1 coordinates with A60, G61, G83 and G696 to further stabilize the catalytic triad, whereas K2 coordinates with G224 in D1 and A695 of the γ-γ' interaction that is critical for catalytic activity 34. An unexpected metal ion that has not been previously described, termed M35, is clamped by non-bridging oxygens of A422 and U423 next to λ-λ' interaction and coordinates with U218 to hold 5'-exon in place. Together these ions establish the heteronuclear metal ion center that is present in all states of Cte 1 intron catalysis and also conserved in canonical group II introns and spliceosomes (Fig. 4, Extended Data Fig. 6g-i)6,22,25.
In 1S state ready for branching, A127 flips toward the catalytic site and the 2'-OH nucleophile is coordinated with M1, whereas M2 remain coordinating with U219 and G220 to prepare for the nucleophilic attack (Fig. 4b). After branching in 2S, the A127-G220 linkage moves away from the catalytic site to allow 3'-exon docking, in which the C134 leaving group is stabilized by M1, while the 5'-exon U219 3'-OH nucleophile is coordinated by M2 to prepare for the second transesterification reaction (Fig. 4c). In post-2S after the ligation, M1, M2 and K1 remain in position and K2 shifts 4.3 Å to coordinate with C223 (Fig. 4d). The A695-U133 γ-γ' interaction is stacked by G221, G220 and A127 that brings the 3'-SS U133 closer to the catalytic core to coordinate with M1 and M35* (Fig. 4d, Extended Data Fig. 6d). M35* is another new metal ion unexpectedly appears in the catalytic core 17.1 Å from M35 to coordinate with U133 2'-OH, whereas M35 is no longer observed as 5'-exon is already released in post-2S. The interaction network of U133 in the catalytic core of post-2S is similar to 3'-SS in previous group II intron lariat forms ready for reverse splicing (Extended Data Fig.6 d-f)8,42.