Chemoproteomic discovery of a human RNA ligase

RNA ligases seal RNA strands in diverse processes including intron-containing tRNA splicing, tRNA repair, mRNA splicing in the unfolded protein response, RNA recombination, and circular RNA biogenesis. 5'-3' RNA ligases catalyze ligation of 5'-phosphate and 3'-hydroxyl RNA ends via a classic three-step mechanism involving (1) ATP-dependent auto-AMPylation of a catalytic lysine, (2) transfer of AMP to the 5'-phosphate of RNA to form AppRNA, and (3) attack by the 3'-OH to form a phosphodiester with release of AMP. Although such ligases are common in viruses, fungi, and plants, no proteinaceous 5'-3' RNA ligase had been identified in vertebrates; in humans only a GTP-dependent 3'-5' RNA ligase (RtcB/HSPC117) was known. The authors study nucleotide-driven post-translational modifications, notably AMPylation, which participates in infection, UPR, and redox homeostasis. To probe the human AMPylated proteome, they designed an alkyne-modified diadenosine triphosphate (Ap3A) chemical probe to enrich AMPylated proteins. The central research question is whether an uncharacterized human protein catalyzing 5'-3' RNA ligation exists and can be discovered via chemoproteomics, and what role it might play in cellular stress responses.

Prior work established the diversity of proteinaceous RNA ligases and their mechanisms across life forms, including the three-step ATP-dependent 5'-3' ligases and the human GTP-dependent 3'-5' ligase RtcB involved in tRNA splicing and XBP1 mRNA splicing during UPR. A 5'-3' RNA ligase activity had been suggested in HeLa cells decades ago, but no vertebrate enzyme had been identified. AMPylation, a covalent addition of AMP to proteins, modulates numerous processes (pathogen infection, UPR, redox homeostasis). Chemical reporters and masked AMP analogues have been used to profile AMPylation in cells. Diadenosine polyphosphates (Ap3A/Ap4A) are stress-related alarmones and can serve as stable ATP analogs for proteomic enrichment. Structural work on Naegleria gruberi RNA ligase (NgrRnl) defined catalytic residues and domains; structural databases and AlphaFold predictions enable functional inference by homology.

• Chemical proteomics: Synthesized a C2,C2′-ethynyl-modified Ap3A probe (C2-eAp3A). Incubated with HEK293T and H1299 cell lysates to serve as an ATP surrogate during AMPylation. Incorporated alkyne allowed Cu(I)-catalyzed azide-alkyne cycloaddition to an azido-desthiobiotin tag for affinity capture. Enriched proteins were identified by LC-MS/MS-based activity-based protein profiling, and enrichment of C12orf29 was validated by immunoblotting.
• Recombinant protein production and characterization: Expressed wild-type and mutant C12orf29 in E. coli, purified AMPylated and deAMPylated forms under different buffer systems, and assessed masses by LC-MS to confirm auto-AMPylation (+329 Da). Auto-AMPylation was probed with ATP (including α-32P-ATP) and various divalent cations; Mg2+ supported the most efficient activity.
• Structural bioinformatics: Generated homology model (Phyre2) and used AlphaFold prediction (average pLDDT ~91.4); compared structures using Dali, identifying RNA ligases (e.g., NgrRnl) as top hits. Mapped conserved nucleotidyltransferase motif KX(D/H/N)G with catalytic Lys57 and other conserved residues (D59, E195, E250, K263).
• RNA ligation assays: Designed single-stranded RNA substrates forming open hairpins and dumbbell structures with defined 5'-phosphate and 3'-OH termini; 5' ends radiolabeled with 32P. Tested ligation by denaturing PAGE and phosphorimaging. Assessed substrate selectivity (RNA vs DNA), co-substrate specificity (ATP, GTP, dATP, Ap3A, Ap4A, NAD), and kinetic parameters by varying nucleotide concentrations with Michaelis–Menten analysis.
• Substrate scope probing: Varied nucleobase composition at the ligation junction and the length of 5' overhangs to evaluate efficiency. Prepared and tested point mutants (e.g., K57A, D59A, R77L, E123Q/D, E250A/Q, K263 variants) for effects on ligation.
• Processing of blocked 3' ends: Tested RNAs with 2',3'-cyclic phosphate or 2'-phosphate termini; combined with N-terminally truncated ANGEL2 (ANGEL2-AN) to hydrolyze 2',3'-cPO4 to 2'-OH-3'-OH, then assayed C12orf29-mediated ligation.
• Cell biology: Generated C12ORF29 knockout (KO) HEK293 cells via CRISPR/Cas and compared to WT cells. Assessed sensitivity to oxidative stress using menadione treatments across concentrations and times. Measured cell viability (CellTiter-Glo) and ROS (ROS-Glo H2O2) in parallel. Isolated total RNA and analyzed integrity by Agilent TapeStation, focusing on 28S and 18S rRNA patterns under increasing menadione concentrations.
• Proteomics data processing: LC-MS/MS data analyzed with MaxQuant (LFQ, match-between-runs) and Perseus; significant enrichment assessed by t-tests with FDR control. Data deposited to PRIDE (PXD038132).

• Discovery and validation: Chemoproteomics with C2-eAp3A enriched the uncharacterized human protein C12orf29 from HEK293T and H1299 lysates; enrichment confirmed by immunoblotting.
• Auto-AMPylation: Recombinant C12orf29 purified from E. coli carried a +329 Da mass consistent with covalent AMPylation (calc. WT 37,771 Da; found 37,772 Da; AMPylated calc. 38,100 Da; found 38,101 Da). Auto-AMPylation detected by anti-AMP immunoblot. Mg2+ is the most proficient cofactor; Mn2+ also supports activity. Mutation K57A abolishes auto-AMPylation, confirming the catalytic lysine.
• Structural homology: AlphaFold-based model aligns with the nucleotidyltransferase domain of NgrRnl (top Dali hit; Z-score ~7.4). Conserved residues (K57, D59, E195, E250, K263) map to the catalytic center.
• RNA ligase activity: C12orf29 catalyzes ligation of 5'-phosphate and 3'-OH termini within single-stranded RNA regions, producing linear ligation products, RNA-adenylate (AppRNA), and circular RNA. Strong selectivity for RNA: RNA:RNA constructs are ligated; DNA constructs are not detectably converted under the same conditions; nick sealing on splinted duplexes was not observed.
• Co-substrate specificity and kinetics: Efficiently uses ATP; also accepts GTP (less efficiently), Ap3A, Ap4A, and to a lesser extent dATP; NAD is not used. Kinetic parameters (Michaelis–Menten) for ligation: ATP: KM = 1.30 µM, kcat = 2.81 × 10^-1 s^-1, kcat/KM = 21.6 s/µM. GTP: KM = 37.7 µM, kcat = 3.14 × 10^-1 s^-1, kcat/KM = 0.83 s/µM. Apparent catalytic efficiency is ~29-fold higher with ATP than GTP, driven primarily by lower KM for ATP.
• Substrate requirements and preferences: Requires a non-phosphorylated 3' terminus (3'-OH). RNAs bearing 2',3'-cyclic phosphate or a 2'-phosphate at 3' are not ligated; however, ANGEL2-AN converts 2',3'-cPO4 to 2'-OH-3'-OH enabling subsequent C12orf29-mediated ligation. Purines at the ligation site enhance ligation efficiency, and longer 5' overhangs improve ligation rates.
• Mutational analysis: Most catalytic-site mutations (e.g., K57A, D59A, E250A/Q, K263 variants) abrogate ligation; E123D/E123Q are fully active; R77L retains ~12% activity.
• Cellular function under oxidative stress: C12ORF29-KO HEK293 cells are significantly more vulnerable to menadione-induced ROS than WT. After 3 h with 40 µM menadione, >85% of WT remain viable vs <10% of KO. Despite similar ROS levels over time between lines, KO shows markedly reduced viability. RNA integrity analysis shows earlier and stronger 28S rRNA degradation in KO (degradation at 40 µM) compared to WT (significant degradation at 100 µM), while 18S rRNA remains relatively stable. Findings implicate C12orf29 in maintaining RNA integrity during oxidative stress.

The study addresses the long-standing gap of a vertebrate 5'-3' RNA ligase by identifying and characterizing C12orf29 as a human enzyme that auto-AMPylates and ligates 5'-phosphate to 3'-OH RNA ends via the canonical three-step mechanism. Chemoproteomics using an Ap3A-based probe enabled enrichment of AMPylated proteins and led to C12orf29’s discovery. Structural homology to known RNA ligases supports its functional assignment. Biochemically, C12orf29 exhibits stringent RNA selectivity, requires a 3'-OH terminus, and prefers ATP as a co-substrate with much higher catalytic efficiency than GTP, while also accommodating Ap3A/Ap4A. The inability to process 2',3'-cPO4 ends directly, coupled with successful ligation after ANGEL2-AN treatment, suggests a coordinated repair pathway in which blocked RNA termini are first “healed” (e.g., by ANGEL2 and a 5'-OH kinase such as hClp1) to produce 3'-OH and 5'-PO4 ends that C12orf29 can “seal.” Cellular experiments show that loss of C12ORF29 sensitizes cells to ROS and accelerates rRNA degradation, indicating a role in preserving RNA integrity under oxidative stress. Together, these findings reveal a latent human RNA repair pathway complementary to the established RtcB-mediated 3'-5' ligation, with potential regulatory interplay involving ANGEL2 and other RNA processing enzymes. The Ap3A probe strategy also underscores a generalizable chemoproteomic approach for profiling NMPylation-dependent enzymology.

This work reports the chemistry-led discovery and mechanistic characterization of C12orf29 as a human 5'-3' RNA ligase (proposed name HsRnl). C12orf29 catalyzes ATP-dependent ligation of 5'-PO4 to 3'-OH RNA termini through sequential auto- and RNA-AMPylation, shows strong RNA specificity, defined substrate preferences, and relies on conserved nucleotidyltransferase residues. In cells, C12ORF29 deficiency compromises resilience to oxidative stress and coincides with increased rRNA decay, implicating C12orf29 in RNA maintenance and repair. The Ap3A-based chemoproteomic platform proved effective for uncovering AMPylated enzymes and may be extended to other NMPylation types. Future studies should identify endogenous interaction partners and RNA substrates of C12orf29, define its in vivo repair pathway and regulation, and determine high-resolution structures to fully elucidate its catalytic cycle and substrate recognition.

• The chemoproteomic probe (Ap3A) is symmetrical and may label proteins via phosphorylation as well as AMPylation, necessitating caution in interpreting enrichment data.
• Structural assignment relies on predicted AlphaFold models and homology; no experimentally determined C12orf29 structure was presented.
• Endogenous RNA substrates and interacting partners of C12orf29 were not identified; the in vivo repair pathway remains to be delineated.
• Cellular evidence links C12ORF29 loss to stress sensitivity and rRNA decay, but direct in vivo ligation events and pathway components (e.g., cooperation with ANGEL2, hClp1) require further validation.
• C12orf29 does not ligate nicked duplex substrates under tested conditions, indicating possible substrate constraints not fully mapped in vivo.