Structure of a Thermostable Group II Intron Reverse Transcriptase with Template-Primer and Its Functional and Evolutionary Implications

Abstract

Bacterial group II intron reverse transcriptases (RTs) function in both intron mobility and RNA splicing and are evolutionary predecessors of retrotransposon, telomerase, and retroviral RTs as well as the spliceosomal protein Prp8 in eukaryotes. Here we determined a crystal structure of a full-length thermostable group II intron RT in complex with an RNA template-DNA primer duplex and incoming deoxynucleotide triphosphate (dNTP) at 3.0-Å resolution. We find that the binding of template-primer and key aspects of the RT active site are surprisingly different from retroviral RTs but remarkably similar to viral RNA-dependent RNA polymerases. The structure reveals a host of features not seen previously in RTs that may contribute to distinctive biochemical properties of group II intron RTs, and it provides a prototype for many related bacterial and eukaryotic non-LTR retroelement RTs. It also reveals how protein structural features used for reverse transcription evolved to promote the splicing of both group II and spliceosomal introns.

Keywords: LINE-1; RNA splicing; RNA-dependent RNA polymerase; RNA-seq; evolution; retrotransposon; retrovirus; reverse transcription; spliceosome; telomerase.

Figure 1. Comparison of GsI-IIC RT to Other RTs, Spliceosomal Protein Prp8, and HCV RdRP

The schematics show the domain organization, conserved RT-sequence blocks (0–7), and conserved sequences (below) in group IIC intron GsI-IIC RT (PDB:6AR1); group IIA intron Ll.LtrB RT (LtrA protein; GenBank:AAB06503); retrovirus HIV-1 RT (PDB:4PQU); Tribolium castaneum TERT (PDB:3KYL); non-LTR-retrotransposon human LINE-1 (UniProtKB:O00370) and Bombyx mori R2Bm RTs (GenBank:AAB59214); Saccharomyces cerevisiae Prp8 (PDB:4I43); and hepatitis C virus (HCV) RdRP (PDB:4WTA). APE, apurinic endonuclease domain; CTS, conserved carboxy-terminal segment; Cys, cysteine-rich conserved sequence; D or DB, DNA-binding domain; En, DNA endonuclease domain; REL, restriction endonuclease-like domain; TRBD, telomerase RNA-binding domain.

Figure 2. Structural Overview and Comparison of GsI-IIC RT to HIV-1 RT and HCV RdRP

(A) Structure of GsI-IIC RT bound to RNA template-DNA primer and dATP. α-helices and β-strands are labeled, and insert regions not present in retroviral RTs are demarcated with brackets. ‘N’ and ‘C’ denote N- and C-termini of the protein. The RNA template and DNA primer sequences are indicated below. Fingers, salmon; insertions, red; palm, dark blue; thumb, green; D domain, yellow; RNA template, purple; DNA primer, cyan; dATP (stick representation; yellow). See also Figure S1. (B) Schematic of protein-nucleic acid interactions. Bases, ribose rings, and phosphates are represented by rectangles, pentagons, and circles, respectively, and interactions between amino acids and nucleotides are indicated by a double black arrow (RNA 2′ OH H-bond), black line (polar interaction), or dashed black line (non-polar interaction). Amino acid names are color-coded according to their domain location within the RT as in panel A. Nucleotide n−1 is the templating RNA base. (C–E) Comparison of GsI-IIC RT to HIV-1 RT (PDB:4PQU) and HCV RdRP (PDB:4WTA), aligned via the palm subdomain, two views with 90° rotation (helices in cylindrical cartoon, colors as above). For HIV-1 RT, RNase H and p51 regions are faded yellow and orange, respectively. For HCV RdRP, regions homologous to RT insert regions are labeled in quotation marks and non-homologous N- and C-terminal regions are faded silver and yellow, respectively. A similar depiction for telomerase RT is shown in Fig. S2.

Figure 3. Comparison of Active-Site Regions of GsI-IIC RT, HIV-1 RT, and HCV RdRP

(A–C), (D–E), and (G–I) show three views highlighting different aspects for GsI-IIC RT, HIV-1 RT, and HCV RdRP, respectively. Regions are colored as in Fig. 2, and important side chains are labeled and shown in stick figure; catalytic Mg²⁺ (green) or Mn²⁺ (lime) are shown as spheres; the highly conserved YXDD and PQG motifs of the RT family and their homologs in RdRPs are circled with dashed lines. (A, D, G) dNTP-binding pocket showing the conformations of the YXDD/CGDD and PQG/ASG loops in the three polymerases. (B, E, H) Another view of the dNTP-binding pocket. The aromatic dNTP ‘gating’ residue below the nucleotide ribose moiety (F143 in GsI-IIC RT, Y115 in HIV-1 RT) is absent in HCV RdRP, but GsI-IIC RT and HCV RdRP have a nearby conserved Asp (D144 or D225, respectively), which H-bonds (green dashes) to the PQG or homologous ASG motif, potentially rigidifying the pocket. (C, F, I) Templating base (n−1)-binding pocket. While the HIV-1 RT templating base is held in place almost exclusively by weak hydrophobic interactions and is exposed in the major groove, in both GsI-IIC RT and HCV RdRP the RT0 motif forms a lid over n−1 in the major groove and H-bonds (green dashes) to the phosphates on either side of n−1.

Figure 4. Comparison of Template-Primer Binding between GsI-IIC RT, HIV-1 RT, and HCV RdRP

Three views for GsI-IIC RT, HIV-1 RT, and HCV RdRP are shown in panels A–C, D–F, and G–I, respectively. Nucleic acid and protein residues involved in binding the template-primer are depicted in stick figure; polar interactions are shown as green dashes; and the ‘fingertips’ β-hairpin loop is circled by a salmon dashed line (other colors as in Fig. 2). (A, D, G) Binding of the single-stranded 5′ RNA overhang. GsI-IIC RT H-bonds to the phosphate backbone of every nucleotide from n−3 to n+1, while HIV-1 RT H-bonds only to the 2′-OH groups of n−2 and n−1. The HCV-RdRP H-bonds to the n−1 and n+1 phosphates in a structure in which n−1 is the 5′ terminal nucleotide (structures with longer 5′ RNA overhangs unavailable). (B, E, H) Binding of the NTE in the major groove. Helix α1′ of the NTE in GsI-IIC RT projects into the major groove of the nucleic acid duplex, making contacts to both template and primer in a region free of contacts in HIV-1 RT. The HCV RdRP “NTE” behaves similarly, with H95 and R109 being potential primer-binding homologs of Q24 and K18 in GsI-IIC RT (structures with longer HCV RdRP primers unavailable). See also Figure S3. (C, F, I) Interaction of RT2a and neighboring regions with the RNA template (primer excluded for clarity). GsI-IIC RT forms 14 polar interactions (6 peptide-backbone H-bonds) with the RNA template, including 4 from RT2a (F110, R111, and N115 with 2′ OHs; G113 with phosphate backbone). HIV-1 RT makes only 6 polar contacts and one peptide backbone H-bond. HCV RdRP preserves the general shape of RT2a insert region but contacts primarily n−1 to n+3, leaving the remaining RNA nucleotides free of polar interactions (the crystallized HCV RdRP construct lacks a large C-terminal segment, which may make additional contacts to the RNA template).

Figure 5. Structure and Interactions of the GsI-IIC RT Thumb and D Domains with Template-primer

(A, B) Comparison of thumb domain interactions with the DNA primer between GsI-IIC RT and HIV-1 RT, respectively. Y of the YXDD motifs (Y221 in GsI-IIC RT and Y183 in HIV-1 RT), the FLG/WMG motifs, the G-(X)₃-Y/W motifs, nucleic acids, and dATP are labeled and shown in stick representation (colors as in Figure 2). (C) Structure of D domain and its interface with the thumb. Left, cartoon representation with potential nucleic acid-binding residues shown in stick figure, and a bound sulfate molecule shown as spheres (other colors as in Figure 2). Right, electrostatic surface potential (red, negative; white, neutral; blue, positive). Electrostatic surface potential for the entire protein is shown in Fig. S1C. Bottom: Amino acid sequence of the GsI-IIC RT thumb and a portion of D domain, with bars representing α-helices. The residues with side chains displayed in (C) are highlighted in blue. See also Figure S4.

Figure 6. RT0-Lid Mutations Inhibit Template-Switching Activity

(A) Model of the template-switching pocket. Space-filling format with semi-transparency around the RT0 loop and R85, colors as in Fig. 2. The bound template-primer substrate is depicted without n−2 to n+1 leaving a single-nucleotide 3′ DNA overhang (p+1) in a pocket for binding of the 3′ end of an incoming RNA template. (B) Electrostatic surface potential of the template-switching pocket, colors as in Fig. 5C. (C) and (D) Assays of template-switching and primer-extension activity respectively. Reactions were carried out with unlabeled template RNAs and 5′-end labeled DNA primers, as described in Methods, and cDNA products were analyzed by denaturing PAGE. Major products are indicated by arrow; lighter bands above the major template-switching product result from multiple end-to-end template-switches. Schematics of the reactions are shown beneath the gels.

Figure 7. Model of GsI-IIC RT Bound to a Group II Intron Lariat RNA and Adaptation of Group II Intron RT Regions for RNA Splicing

(A) Left, Model of GsI-IIC RT bound to a group IIC intron RNA lariat. The intron lariat (PDB:5J02) with the 5′ end of the intron linked to the branch-point adenosine (5′I-BP, teal triangle) is poised to use its 3′ OH (3′I-OH, orange circle) to attack the ssDNA exon junction (EJ). The latter is located downstream of a 5′-exon DNA hairpin (5′ HP) recognized by the GsI-IIC RT. The black arrow denotes the gap (~6-Å, or 1 nucleotide) between 3′ exon position +3 (from PDB:3IGI; 5′ exon, 5′ E; 3′ exon, 3′ E) and the 5′ end of the RNA template strand in the crystal structure. The group IIC intron lariat RNA is depicted in faded space-filling format with DVI, cartoon white. GsI-IIC RT/template-primer complex is shown in cartoon cylinder format with regions colored as in Fig. 2 and template and primer 5′ and 3′ ends labeled. Right, Schematic of intron lariat RNA (gray) at a single-stranded DNA target site before (top) and after (bottom) reverse splicing into the DNA strand, labels as at left. (B) NTE/RT0 interactions. Left, GsI-IIC RT NTE/RT0 bound to template-primer structure with intron DVI positioned as in the model of panel A; middle, Ll.LtrB RT NTE bound to DIVa(iii) from cryo-EM structure (PDB:5G2X); right, Prp8 NTE near the U2 branch-point recognition site bound to intron 3′ end (U2-BP Recog. Site/3′I, PDB:5LJ3). (C) RT3a interactions. Left, GsI-IIC RT RT3a and 3a loop near PQG motif bound to template-primer; middle, Ll.LtrB RT RT3a bound to intron DIVa(i) and (ii); right, Prp8 RT3a near U4 snRNA (PDB:5GAN) (D) Thumb and D domain interactions. Left, GsI-IIC RT thumb and D domain bound to template-primer with 5′ exon (5′ E, faded blue) model and IIC 5′-exon hairpin model (faded white) as in panel A; middle, Ll.LtrB RT thumb and D domain bound to 5′ exon and intron DI; right, Prp8 thumb and switch loop region bound to 5′ exon and U5 snRNA 5′-exon recognition site (PDB:5LJ3).