Chemoproteomic discovery of a human RNA ligase

Abstract

RNA ligases are present across all forms of life. While enzymatic RNA ligation between 5'-PO₄ and 3'-OH termini is prevalent in viruses, fungi, and plants, such RNA ligases are yet to be identified in vertebrates. Here, using a nucleotide-based chemical probe targeting human AMPylated proteome, we have enriched and identified the hitherto uncharacterised human protein chromosome 12 open reading frame 29 (C12orf29) as a human enzyme promoting RNA ligation between 5'-PO₄ and 3'-OH termini. C12orf29 catalyses ATP-dependent RNA ligation via a three-step mechanism, involving tandem auto- and RNA AMPylation. Knock-out of C12ORF29 gene impedes the cellular resilience to oxidative stress featuring concurrent RNA degradation, which suggests a role of C12orf29 in maintaining RNA integrity. These data provide the groundwork for establishing a human RNA repair pathway.

Fig. 1. RNA ligase mechanism and identification of C12orf29 by chemical proteomics using modified Ap₃A probe.

a Schematic display of the three-step mechanism of RNA ligation by a 5′–3′ RNA ligase. In step 1, the ligase is auto-AMPylated on the catalytic lysine using ATP as the co-substrate. In step 2, the AMP is transferred from the catalytic lysine to the 5′-PO₄ end of RNA (pRNA), giving the RNA-adenylate intermediate (AppRNA). In step 3, the ligated RNA is obtained upon the attack of 3′-OH to the AppRNA in the presence of the ligase along with the liberation of AMP. PPi, pyrophosphate. b Structures of Ap₃A analogues employed in this study. c Schematic display of the workflow for the identification of C12orf29. Cell lysates are incubated with C2-eAp₃A or controls. AMPylated proteins are expected to bear ethynyl functionalities that enable selective modification with an affinity tag desthiobiotin (DB) via CuAAC. Labelled proteins are enriched and identified by ABPP, and further verified by immunoblotting. d Structure of the azide-bearing desthiobiotin as affinity tag. e Affinity enrichment of C12orf29 from two cell lysates verified by immunoblotting (representative images of n = 3). Source data are provided as a Source Data file.

Fig. 2. Auto-AMPylation activity and structural prediction of C12orf29.

a (Top) immunoblotting of C12orf29^WT-AMP and C12orf29^K57A and LC-MS analysis of C12orf29^WT-AMP. C12orf29^K57A is deficient in auto-AMPylation activity (representative images of n = 3). Mass spectra indicating a 329 Da increase in mass upon auto-AMPylation of C12orf29^WT. C12orf29^WT-AMP: calc. 38,100 Da, found 38,101 Da. C12orf29^WT: calc. 37,771 Da, found 37,772 Da. (Bottom) preparation of C12orf29^WT and C12orf29^WT-AMP with different buffer systems (representative images of n = 3). b Divalent metal ion dependency of auto-AMPylation activity of C12orf29^WT. c Superimposition of the structure of C12orf29 predicted by AlphaFold (green)^, on the structure of NgrRnl-AMP (PDB ID: 5COT, grey). OB, oligonucleotide-binding. NT, nucleotidyltransferase. Structures were superposed in Coot using structural equivalent residues identified by the DALI webserver. d Enlarged view of the catalytic site of NgrRnl-AMP (PDB ID: 5COT, grey) and predicted C12orf29 (green). In the NgrRnl-AMP structure, AMP is covalently attached to the side chain of K170. D172, E227, and E312 bind Mn²⁺ via water-mediated contacts and K326 contacts the phosphate moiety of AMP. Corresponding residues in the putatively catalytic site of C12orf29 are indicated. Mn²⁺ and water molecules are depicted as yellow and red spheres, respectively. Atomic contacts are depicted as dashed lines. Source data are provided as a Source Data file.

Fig. 3. RNA ligase activity of C12orf29.

RNA/DNA oligonucleotides are schematically depicted in blue/green, respectively. The ³²P-labelled 5′-ends are depicted in red. The radioactive oligonucleotides were resolved by denaturing PAGE and analysed by phosphorimaging. a (Top) Schematic display of two nucleic acid substrates used. (Bottom) PAGE analysis of reaction products after incubation with C12orf29 for various time points as indicated. Ligation was observed for 5′-phosphorylated RNA (representative images of n = 3). b Reaction of C12orf29 with indicated RNA constructs using various nucleotides (representative images of n = 3). Besides ATP, C12orf29 is able to efficiently catalyse ligation by processing GTP, dATP, Ap₃A, and Ap₄A. Nuc., nucleotide. c (Left) C12orf29 promoted ligation reaction at various concentrations of ATP and GTP (representative images of n = 3). (Right) Michaelis-Menten fits to initial rates providing K_M, k_cat and k_cat/K_M for ATP and GTP. Plotted data represent the mean value ± SD for three biological replicates. d (Left) Proposed ligation scheme starting from 2′,3′-cyclic phosphorylated RNA by the sequential action of ANGEL2-ΔN and C12orf29^WT. (Right) PAGE analysis of the depicted reaction (representative images of n = 3). All oligonucleotide sequences are provided in Supplementary Table 1. Source data are provided as a Source Data file.

Fig. 4. Substrate scope of C12orf29.

RNA oligonucleotides that were used for intramolecular ligation resulting in cyclisation are schematically depicted. The ³²P-labelled 5′-ends were depicted in red. The radioactive oligonucleotides were resolved by denaturing PAGE and analysed by phosphorimaging. a, Investigation of the impact of the nucleobase composition at the 5′- and 3′- termini at the ligation site on the ligation efficiency of C12orf29 (representative images of n = 3). The depicted constructs were incubated under the same conditions for various time points. Most efficient ligation was observed when purines were at the ligation site. b Investigation of the impact of the length of the 5′-overhang on the ligation efficiency of C12orf29 (representative images of n = 3). c Impact of single site mutations on RNA ligase efficiency of C12orf29. Graph bars represent mean ligation efficiencies ± SEM and hollow circles represent individual data points for n = 3 biological replicates. All oligonucleotide sequences are provided in Supplementary Table 1. Source data are provided as a Source Data file.

Fig. 5. Effect of menadione-induced oxidative stress on WT and C12orf29-KO cells.

a Light microscopy of HEK293 WT and C12ORF29-KO cells. The cells were treated with 40 μM menadione for 3 h or only with the carrier as control (0 μM menadione). Scale bars are 100 μm. b Cell viability of HEK293 WT and C12ORF29-KO cells after 3 h treatment with different menadione concentrations. Hollow circles represent individual data points for n = 9 biological replicates. Error bars represent the ± SEM. Significance was calculated by two-way ANOVA with Sidak’s multiple comparisons test: ^nsP > 0.05; *P ≤ 0.05; ****P ≤ 0.0001. c Cell viability and corresponding H₂O₂ concentrations in HEK293 WT and C12ORF29-KO cells at different time points after treatment with 40 μM menadione. Hollow circles represent individual data points for n = 3 biological replicates. Error bars represent the ± SEM. Significance was calculated by two-way ANOVA with Sidak’s multiple comparisons test: ^nsP > 0.05; ****P ≤ 0.0001. RLU, relative light units. Source data are provided as a Source Data file.

Fig. 6. Analysis of total RNA from cell extracts of HEK293 WT and C12ORF29-KO cells treated with various concentrations of menadione.

(Top) Cells (left: HEK293 WT cells; right: C12ORF29-KO HEK293 cells) were treated with different menadione concentrations for 3 h. In turn, total RNA was isolated and subsequently analysed by TapeStation (version 4.1.1). The ratio of 28S and 18S rRNA intensities were listed below the electrophoregram (a.u., arbitrary units). (Bottom) Electropherograms of the TapeStation analysis of total RNA from cell extracts of HEK293 WT (in red) and C12ORF29-KO (in yellow) cells. Source data are provided as a Source Data file.