Structural basis for pegRNA-guided reverse transcription by a prime editor

Abstract

The prime editor system composed of Streptococcus pyogenes Cas9 nickase (nSpCas9) and engineered Moloney murine leukaemia virus reverse transcriptase (M-MLV RT) collaborates with a prime editing guide RNA (pegRNA) to facilitate a wide variety of precise genome edits in living cells¹. However, owing to a lack of structural information, the molecular mechanism of pegRNA-guided reverse transcription by the prime editor remains poorly understood. Here we present cryo-electron microscopy structures of the SpCas9-M-MLV RTΔRNaseH-pegRNA-target DNA complex in multiple states. The termination structure, along with our functional analysis, reveals that M-MLV RT extends reverse transcription beyond the expected site, resulting in scaffold-derived incorporations that cause undesired edits at the target loci. Furthermore, structural comparisons among the pre-initiation, initiation and elongation states show that M-MLV RT remains in a consistent position relative to SpCas9 during reverse transcription, whereas the pegRNA-synthesized DNA heteroduplex builds up along the surface of SpCas9. On the basis of our structural insights, we rationally engineered pegRNA variants and prime-editor variants in which M-MLV RT is fused within SpCas9. Collectively, our findings provide structural insights into the stepwise mechanism of prime editing, and will pave the way for the development of a versatile prime editing toolbox.

Fig. 1. Cryo-EM structure of the prime editor in the termination state.

a, Two components of the prime editing system. The prime editor is composed of nSpCas9 and M-MLV RT, and the pegRNA comprises the sgRNA region and the 3′ extension region. The 3′ extension region consists of the PBS and the RTT. The sgRNA region, PBS and RTT are coloured red, pink, and yellow respectively. b, Schematic of the nSpCas9–M-MLV RT–pegRNA–target DNA complex. nSpCas9 nicks the NTS in guide RNA- and PAM-dependent manners, and then M-MLV RT binds to the PBS–NTS heteroduplex and reverse transcribes the RTT sequences. TS, target strand. c, In vitro prime editing assay using purified PE2, pegRNA containing 13-nt PBS and 28-nt RTT, and 5′-Cy5-labelled pre-nicked DNA substrates. The PE2–pegRNA complex was mixed with DNA substrates, and incubated at 37 °C for 10 min. The reaction products were separated on a 10% Novex PAGE TBE–urea gel, and the Cy5 fluorescence was then visualized. PegRNA-MM refers to a pegRNA designed with non-complementary sequences between the guide and the PBS. Untethered PE refers to a construct in which nSpCas9 and M-MLV RTΔRNaseH (RTΔRH) were purified separately. Untethered PE exhibits pegRNA-dependent reverse transcription activity comparable to that of PE2. The experiments were repeated three times with similar results. d, Domain structures of dSpCas9 (D10A/H840A) and RTΔRH (D200N/T306K/W313F/T330P). RTΔRH lacks the RNaseH domain (residues 498–671). PI, PAM-interacting domain. e, Cryo-EM density map of the SpCas9–RTΔRH–pegRNA–target DNA complex in the termination state. f, Overall structure of the SpCas9–RTΔRH–pegRNA–target DNA complex in the termination state. The disordered regions are indicated as dotted lines.

Fig. 2. Nucleic acid architecture.

a, Schematic of the pegRNA and target DNA in the termination state. Except for the 3′ stem loop (G82–C96), the scaffold region of the pegRNA is represented with a red line for simplicity. The disordered regions are coloured grey. b, Structure of the pegRNA and target DNA in the termination state. The disordered nucleotides G(−20*)–C(−17*) of the NTS are indicated with a dotted line. c, Close-up view of the M-MLV RT active site. The cryo-EM densities for the catalytic residues (YVDD), the RTT–newly synthesized DNA heteroduplex (U97-ddA6^†–G102-dC1^†) and the 3′ end of the stem loop (G95 and C96) are shown as grey meshes.

Fig. 3. Termination site of reverse transcription.

a, Surface representation of the SpCas9–RTΔRH–pegRNA–target DNA complex in the termination state. Although reverse transcription proceeds up to the end of the RTT, there is an approximate 10-Å separation between SpCas9 and RTΔRH, leaving sufficient space for further reverse transcription. b, Recognition of the 5′ end of the RTT sequence and 3′ end of the stem loop. The key residues Y64/L99 and R116, crucial for the processivity of reverse transcription, form van der Waals interactions and a hydrogen bond with C96, respectively. The hydrogen bond is depicted with a green dashed line. c, In vitro prime editing assay using PE2 or untethered PE, a pegRNA and 5′-Cy5-labelled pre-nicked DNA substrates. The RTT sequence of the pegRNA was designed to contain ‘U’ only at the 5′ end, enabling reverse transcription to stop at the end of the RTT sequence when using ddATP. The PE2/untethered PE–pegRNA complex was added to DNA substrates with dNTPs or with dCTP, dTTP, dGTP and ddATP (referred to as ddATP in c for simplicity), and incubated at 37 °C for 10 min. The reaction products were separated on a 15% Novex PAGE TBE–urea gel, and the Cy5 fluorescence was visualized. The experiments were repeated three times with similar results. d, Close-up view of the space sandwiched between SpCas9 and M-MLV RT. SpCas9 recognizes the stem loop region (G82–C96) through non-base-specific interactions, and M-MLV RT terminates reverse transcription at U97. There is sufficient space between SpCas9 and M-MLV RT for M-MLV RT to proceed up to U94.

Fig. 4. Cryo-EM structures of the prime editor in multiple states.

a–f, Cryo-EM densities (a,c,e) and overall structures (b,d,f) of the SpCas9–RTΔRH–pegRNA–target DNA complex in the initiation state (a,b), the SpCas9–pegRNA–target DNA complex in the pre-initiation state (c,d) and the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation state (16-nt) (e,f). The disordered regions are indicated as dotted lines. g, Mapping of the G1054, E1055, T1068 and G1069 residues in the RuvC domain onto the SpCas9–pegRNA–target DNA complex in the pre-initiation state. These residues are located close to the PBS–NTS heteroduplex. h, In vitro prime editing assay using wild-type PE2 (referred to as WT) and two prime editor variants, with the RTΔRH inserted between G1054 and E1055 (M1: G1054–RTΔRH–E1055) or T1068 and G1069 (M2: T1068–RTΔRH–G1069) in the RuvC domain. The experiments were repeated three times with similar results. i, Close-up views of the PBS–NTS heteroduplex and the PAM-distal duplex in the initiation (left), elongation (16-nt) (middle) and elongation (28-nt) (right) states. As the reverse transcription of the RTT sequence progresses, the PBS–NTS heteroduplex is pushed in the opposite direction to the RTΔRH (shown with a black arrow), resulting in the rearrangement of the PAM-distal duplex.

Fig. 5. Stepwise model of prime editing.

Structure-based stepwise model of prime editing. (1) NTS cleavage. nSpCas9 nicks the NTS in guide RNA-dependent and PAM-dependent manners. (2) PBS–NTS formation. The PBS of the pegRNA base-pairs with the nicked NTS to form the PBS–NTS heteroduplex on the marginal surface of the RuvC domain. (3) PBS–NTS recognition. M-MLV RT recognizes the PBS–NTS heteroduplex on the surface, initiating the reverse transcription of the RTT sequence. (4) Reverse transcription. M-MLV RT consistently engages in reverse transcription of the RTT sequence around the initiation site, and the RTT–synthesized DNA heteroduplex accumulates along the longitudinal surface of SpCas9, accompanied by the rearrangement of the PAM-distal duplex. (5) Termination. M-MLV RT does not terminate at the end of the RTT sequence, but instead invades the scaffold region of the pegRNA, extending the reverse transcription up to three nucleotides upstream of the RTT (U94). It is speculated that M-MLV RT dissociates from the pegRNA owing to steric hindrance with nSpCas9. (6) Edit incorporation. The newly synthesized DNA containing the desired edit is integrated into the genomic loci by a mechanism that is still not fully understood. Given that scaffold-derived incorporations are much less frequent in mammalian cells, endogenous exonucleases might be involved in this process. Further functional analyses are required to fully understand this mechanism.

Extended Data Fig. 1. Protein purification and schematic of in vitro prime editing assay.

a, SDS-PAGE gel of purified PE2, dSpCas9, RTΔRH, PE2 mutants, and PE6a–d proteins. b, Schematic of the in vitro prime editing assay. The pegRNA and DNA sequences used in the assay are summarized in Supplementary Table 1. c, The pegRNA sequences used for the in vitro prime editing assay and structural determinations in multiple states. The RTT sequence of the pegRNA was designed to contain only one “U”, enabling reverse transcription to stop at this “U” using ddATP. Recent studies showed that the complementarity between the guide sequence and PBS sequence in the pegRNA inhibits the prime editing activity^,. Thus, we performed the structural analysis using a pegRNA with non-complementary guide and PBS sequences.

Extended Data Fig. 2. Single-particle cryo-EM analysis of the termination state.

a, Size-exclusion chromatography profile of the SpCas9–RTΔRH–pegRNA–target DNA complex in the termination state. The peak fraction (indicated by a black bar) was analysed by SDS-PAGE and urea-PAGE, and then used for cryo-EM analysis. b, Representative cryo-EM image of the SpCas9–RTΔRH–pegRNA–target DNA complex in the termination state, recorded on a 300 kV Titan Krios microscope with a K3 camera. c, Single-particle cryo-EM image processing workflow. d,f, Overall (d) and local (f) cryo-EM density maps. e,g, Fourier shell correlation (FSC) curves for the overall (e) and local (g) 3D reconstructions. The gold-standard cut-off (FSC = 0.143) is marked with the black dotted line. h, Local-resolution cryo-EM density map.

Extended Data Fig. 3. Overall structure of the prime editor in the termination state.

a, Structural comparison of the SpCas9–pegRNA–target DNA complex with the SpCas9–sgRNA–target DNA complex (PDB: 7Z4L). The SpCas9–sgRNA–target DNA complex (light blue) is superimposed onto the SpCas9–pegRNA–target DNA complex. b,c, Surface representation (b) and electrostatic surface potential (c) of SpCas9 in the termination state. The PBS–NTS heteroduplex is located on the weakly positively charged surface facing the RuvC domain. The disordered regions are indicated as dotted lines. d, Ribbon representation of RTΔRH with cryo-EM density. Residues 1–23, 449–454, and 483–496 are disordered due to their flexibilities. e, Ribbon representation (left) and electrostatic surface potential (right) of RTΔRH with the PBS–NTS and RTT–synthesized DNA heteroduplexes. These heteroduplexes are accommodated within the positively charged groove of the RTΔRH. The catalytic YVDD motif is indicated by the grey circle.

Extended Data Fig. 4. pegRNA modification.

a, In vitro prime editing assay using pegRNAs containing 6-, 10-, 20-, and 28-nt RTT sequences. The PE2–pegRNA complexes were added to DNA substrates with dNTPs or with dCTP, dTTP, dGTP, and ddATP (referred to as ddATP for simplicity), and incubated at 37 °C for 10 min. The reaction products were separated on a 10% Novex PAGE TBE–urea gel, and then Cy5 fluorescence was visualized. The experiments were repeated three times with similar results. b, Close-up view of the 3′ stem loop in the pegRNA scaffold. Three base pairs (G82-C96–A84-U94) do not form base-specific interactions with SpCas9, while the phosphate backbone of C93 forms hydrogen bonds with K30. The hydrogen bond is depicted with a green dashed line. c, Schematic of the wild-type and modified pegRNAs. The modified pegRNA has an altered U94–C96 to complement the target locus and adjust A84–G82 to maintain the stem structure. d, In vitro prime editing assay using the wild-type pegRNA and three modified pegRNAs. The three modified pegRNAs were altered from GCA (82–84)-UGC (94–96) to CGU-ACG, UUU-AAA, and CCC-GGG, respectively. The experiments were repeated three times with similar results. e, The desired edit and the undesired incorporation efficiencies of the PE2 with five conditions (left) and the PE3 with four conditions (right), using the wild-type and modified pegRNAs in HEK293 cells. Data are mean ± s.d. (n = 3 biologically independent samples).

Extended Data Fig. 5. Single-particle cryo-EM analysis of the initiation state.

a, Size-exclusion chromatography profile of the SpCas9–RTΔRH–pegRNA–target DNA complex in the initiation state. The peak fraction (indicated by a black bar) was analysed by SDS-PAGE and urea-PAGE, and then used for cryo-EM analysis. b, Representative cryo-EM image of the SpCas9–RTΔRH–pegRNA–target DNA complex in the initiation state, recorded on a 300 kV Titan Krios microscope with a K3 camera. c, Single-particle cryo-EM image processing workflow. d,f, Overall (d) and local (f) cryo-EM density maps. The ambiguous density corresponding to RTΔRH is enclosed within a black circle in the overall map. e,g, Fourier shell correlation (FSC) curves for the overall (e) and local (g) 3D reconstructions. The gold-standard cut-off (FSC = 0.143) is marked with the black dotted line. h, Local-resolution cryo-EM density map.

Extended Data Fig. 6. Structure of the prime editor in the initiation state.

a, Schematic of the pegRNA and target DNA in the initiation state. Except for the 3′ stem loop (G82–C96), the scaffold region of the pegRNA is represented with a red line for simplicity. The disordered regions are coloured grey. b, Overall structure of the SpCas9–RTΔRH–pegRNA–target DNA complex in the initiation state. The disordered regions are indicated as dotted lines. c, Close-up view of the M-MLV RT active site in the initiation state. The cryo-EM densities for the RTT, the synthesized DNA, the PBS–NTS heteroduplex, and the 3′ end of the stem loop are shown as grey meshes. The middle of the RTT region (C98–C100) is disordered due to its flexibility. The disordered regions are indicated as dotted lines.

Extended Data Fig. 7. Structure of the prime editor in the pre-initiation state.

a, Size-exclusion chromatography profile of the SpCas9–pegRNA–target DNA complex in the pre-initiation state. The peak fraction (indicated by a black bar) was analysed by SDS-PAGE and urea-PAGE, and then used for cryo-EM analysis. b, Representative cryo-EM image of the SpCas9–pegRNA–target DNA complex in the pre-initiation state, recorded on a 300 kV Titan Krios microscope with a K3 camera. c, Single-particle cryo-EM image processing workflow. d, Fourier shell correlation (FSC) curves for the 3D reconstruction. The gold-standard cut-off (FSC = 0.143) is marked with the black dotted line. e, Local-resolution cryo-EM density map. f, Close-up view of the PBS–NTS heteroduplex in the pre-initiation state. The cryo-EM densities for the 3′ extension region of the pegRNA and the NTS are shown as grey meshes. The ambiguous densities corresponding to the single-stranded region of the NTS (dG[−20*]–dC[−17*]) and the RTT (A97–G124) are visible, but we were unable to build the model. g, Close-up view of the PBS–NTS heteroduplex in the initiation state. h, Comparison of the positions of the PBS–NTS heteroduplex between the pre-initiation and initiation states indicates a similar location in both states. The initiation complex (light blue) is superimposed onto the pre-initiation complex.

Extended Data Fig. 8. Structure of the prime editor in the elongation (16-nt) state.

a, Size-exclusion chromatography profile of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (16-nt) state. The peak fraction (indicated by a black bar) was analysed by SDS-PAGE and urea-PAGE, and then used for cryo-EM analysis. b, Representative cryo-EM image of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (16-nt) state, recorded on a 300 kV Titan Krios microscope with a K3 camera. c, Single-particle cryo-EM image processing workflow. d,e, Fourier shell correlation (FSC) curves for the overall (d) and local (e) 3D reconstructions. The gold-standard cut-off (FSC = 0.143) is marked with the black dotted line. f, Local-resolution cryo-EM density map. g, Comparison of the RTΔRH positions between the initiation and elongation states. The initiation complex (light blue) is superimposed onto the pre-initiation complex. h, The AlphaFold-prediction models of G1054–RTΔRH–E1055 (left) and T1068–RTΔRH–G1069 (right) PE2 variants. The nucleic acid of the initiation state was mapped onto these AlphaFold models. G1054, E1055, T1068, and G1069 are shown as orange space-filling models. The five amino-acid linkers are coloured black.

Extended Data Fig. 9. Structure of the prime editor in the elongation (28-nt) state.

a, Size-exclusion chromatography profile of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (28-nt) state. The peak fraction (indicated by a black bar) was analysed by SDS-PAGE and urea-PAGE, and then used for cryo-EM analysis. b, Representative cryo-EM image of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (28-nt) state, recorded on a 300 kV Titan Krios microscope with a K3 camera. c, Single-particle cryo-EM image processing workflow. d,e, Fourier shell correlation (FSC) curves for the overall (d) and local (e) 3D reconstructions. The gold-standard cut-off (FSC = 0.143) is marked with the black dotted line. f, Local-resolution cryo-EM density map. g,h, The cryo-EM density (g) and overall structures (h) of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (28-nt) state. The disordered regions are indicated as dotted lines.

Extended Data Fig. 10. Structure of M-MLV RT with substrates.

a, Domain structure of M-MLV RT. Our construct lacks the RNaseH domain (residues 498–671). b, Ribbon representation of M-MLV RT with the RNA–DNA substrate. The disordered regions are indicated as dotted lines. c–e, Structural comparison between the substrate-bound and unbound states (PDB: 4MH8) in the fingers (c), thumb (d) and connection (e) domains. The substrate-unbound state (light blue) is superimposed onto the substrate-bound state. f, In vitro prime editing assay using PE2 and PE6a–d proteins and the pegRNA containing the 28-nt RTT sequence. The PE2/PE6a–d–pegRNA complex was added to DNA substrates with dNTPs and incubated at 37 °C for 10 min (left) and 60 min (right). The 62-nt marker represents the product transcribed up to the end of the RTT (U97), while the 65-nt marker represents the product transcribed up to U94 in the pegRNA scaffold. The experiments were repeated three times with similar results. g, Overall structure of the SpCas9–RTΔRH–pegRNA–target DNA complex in the elongation (28-nt) state. The disordered regions are indicated as dotted lines. h, Close-up views of the PBS–NTS heteroduplex, the PAM-distal duplex, and the end of the guide RNA–target DNA heteroduplex. As the reverse transcription of the RTT sequence progresses, the PBS–NTS heteroduplex is pushed in the opposite direction from the RTΔRH (shown in black arrow), resulting in the rearrangement of the PAM-distal duplex. This rearrangement induces the disruption of the base pairs at the end of the guide RNA–target DNA heteroduplex (G1-dC20 and G2-dC19).