Reconstitution and structure of a bacterial Pnkp1-Rnl-Hen1 RNA repair complex

Abstract

Ribotoxins cleave essential RNAs for cell killing, and RNA repair neutralizes the damage inflicted by ribotoxins for cell survival. Here we report a new bacterial RNA repair complex that performs RNA repair linked to immunity. This new RNA repair complex is a 270-kDa heterohexamer composed of three proteins-Pnkp1, Rnl and Hen1-that are required to repair ribotoxin-cleaved RNA in vitro. The crystal structure of the complex reveals the molecular architecture of the heterohexamer as two rhomboid-shaped ring structures of Pnkp1-Rnl-Hen1 heterotrimer fused at the Pnkp1 dimer interface. The four active sites required for RNA repair are located on the inner rim of each ring. The architecture and the locations of the active sites of the Pnkp1-Rnl-Hen1 heterohexamer suggest an ordered series of repair reactions at the broken RNA ends that confer immunity to recurrent damage.

Figure 1. Reconstitution of the Pnkp1–Rnl–Hen1 RNA repair complex in vitro.

(a) Schematic view of the three proteins that constitute the new bacterial RNA repair system. The boundary of domains in each protein was determined based on the structure of the Pnkp1–Rnl–Hen1 heterohexamer. NTase, nucleotidyltransferase domain; Conn., connecting domain; MTase, methyltransferase. (b) Size-exclusion chromatography analyses of individual protein, the pairwise mixture and the three-protein mixture. The chromatographic curves of individual proteins are coloured the same as in a and the ones for the mixtures are in black. (c) Repair assay of a ribotoxin-cleaved tRNA by various combinations of Pnkp1, Rnl and Hen1 as indicated. SM, size marker.

Figure 2. Kinetics of an individual enzymatic reaction.

Enzymatic reactions of 5′-phosphoylation (a), 3′-dephosphoylation (b), RNA ligation (c) and RNA cleavage (d) were carried out at different time points and quantified, producing the time courses of the reactions. Of each panel of data, the curve marked with a circle represents the results of the left half of the panel, and the one marked with a square denotes the right half. The ratios of substrate to enzyme are 400 for the 5′-phosphorylation reaction shown in a, 4 for the 3′-dephosphoylation reaction shown in b, 4 for the ligation reaction shown in c and 1 for the cleavage reaction shown in d. The percentage in a represents the maximal value achieved in the experiment, which was estimated to be at least 90% of the 5′-OH RNA ends. The percentage in c represents the RNA ligation product formed, and the one in d denotes the tRNA that remains uncut. S, substrate; P, product. Representative results are shown.

Figure 3. Overall structure of the Pnkp1–Rnl–Hen1 heterohexamer.

(a,b) Ribbon representation of the top (a) and the side (b) views of the structure. One copy of Pnkp1, Rnl and Hen1 are coloured the same as in Fig. 1a, and the second copy of Pnkp1, Rnl and Hen1 are coloured sand, dark blue and ruby, respectively. (c,d) Surface of the Pnkp1–Rnl–Hen1 heterohexamer in the same colours and orientations as in a and b, respectively. Two arrows in d indicate the likely directions from which the damaged RNAs approach the Pnkp1–Rnl–Hen1 heterohexamer for RNA repair.

Figure 4. Details at the protein–protein interfaces.

(a) Details of the interface between the two kinase domains of the Pnkp1 homodimer. Cα-chains of the structure are represented and coloured the same as in Fig. 3a,b. The side chains are in stick and coloured orange except heteroatoms, which are coloured individually (nitrogen in blue, oxygen in red and sulfur in yellow). Hydrogen bonds and salt bridges (3.5 Å or less) are depicted with black dashed lines. (b–e) Details of the interface between the two phosphatase domains of the Pnkp1 homodimer (b), Pnkp1 and Rnl (c), Rnl and Hen1 (d), and Pnkp1 and Hen1 (e).

Figure 5. The presence of cofactors in all enzymatic active sites.

(a) ATP bound in the kinase active site of Pnkp1-a. Proteins are depicted and coloured the same as in Fig. 3a,b. ATP is in stick and coloured green with the exception of heteroatoms, which are coloured individually (nitrogen in blue, oxygen in red, phosphate in magenta and sulfur in yellow). The simulated annealing composite 2mFo-DFc omit density map contoured at 1.5σ is in black mesh. (b) A magnesium ion bound in the phosphatase active site of Pnkp1-a. Mg²⁺ depicted in sphere and coloured silver. A water molecule (in red sphere) was tentatively modelled on additional electron density 4.5 Å away from Mg²⁺, which could also be a phosphate group with a reduced occupancy. (c) SAH was modelled in the methyltransferase active site of Hen1-b. Because SAM or SAH was not added during crystallization or crystal soaking, the SAH modelled in the methyltransferase active site was likely obtained by Hen1 during its overexpression in E. coli. A significantly weaker omit map compared with other cofactors indicates that only a small fraction of SAH is retained by Hen1 after protein purification and crystallization steps. (d) ATP bound in the ligase active site of Rnl-b.

Figure 6. Docking model and proposed mechanism of RNA repair carried out by the Pnkp1–Rnl–Hen1 heterohexamer.

(a) Modelling the 5′ end (red) and the 3′-end (blue) of damaged RNAs into all four active sites required for RNA repair. The modelled RNAs are in stick, and the proteins are depicted the same as in Fig. 3c but with a closer view on the vacant space surrounded by the b-unit of the Pnkp1–Rnl–Hen1 heterotrimer (coloured cyan, green and magenta) and Pnkp1-a (coloured sand). (b) Schematic view of Pnkp1, Rnl and Hen1 that provide four active sites for RNA repair. Pnkp1-b is omitted for clarity. The locations of the active sites are indicated by cofactors (white circles), with ATP in the kinase and ligase active sites, a magnesium ion in the phosphatase site and SAM in the methyltransferase site. Arrows indicate the travel pathways for both the 5′ end and 3′ end of a damaged RNA for RNA repair.

Figure 7. Presence of the Pnkp1–Rnl–Hen1 RNA repair complex in microbes living in humans.

(a) Distribution of bacteria possessing the Pnkp1–Rnl–Hen1 RNA repair complex at five different locations of the human body. G. tract, gastrointestinal tract; U. tract, urogenital tract. (b) Distribution of the bacteria at different sub-locations of the human mouth. Supra. plaque, supragingival plaque; Sub. plaque, subgingival plaque; T. dorsum, tongue dorsum; A/K gingival, attached/keratinized gingival; P. tonsils, palatine tonsils; B. mucosa, buccal mocosa. (c) Distribution of six human-hosted bacterial species possessing the Pnkp1–Rnl–Hen1 RNA repair complex at three major sub-locations of the human mouth.