A sequence-specific RNA acetylation catalyst

Abstract

N4-acetylcytidine (ac4C) is a ubiquitous RNA modification incorporated by cytidine acetyltransferase enzymes. Here, we report the biochemical characterization of Thermococcus kodakarensis Nat10 (TkNat10), an RNA acetyltransferase involved in archaeal thermotolerance. We demonstrate that TkNat10's catalytic activity is critical for T. kodakarensis fitness at elevated temperatures. Unlike eukaryotic homologs, TkNat10 exhibits robust stand-alone activity, modifying diverse RNA substrates in a temperature, ATP, and acetyl-CoA-dependent manner. Transcriptome-wide analysis reveals TkNat10 preferentially modifies unstructured RNAs containing a 5'-CCG-3' consensus sequence. Using a high-throughput mutagenesis approach, we define sequence and structural determinants of TkNat10 substrate recognition. We find TkNat10 can be engineered to facilitate use of propionyl-CoA, providing insight into its cofactor specificity. Finally, we demonstrate TkNat10's utility for site-specific acetylation of RNA oligonucleotides, enabling analysis of ac4C-dependent RNA-protein interactions. Our findings establish a framework for understanding archaeal RNA acetylation and a new tool for studying the functional consequences of ac4C in diverse RNA contexts.

Graphical Abstract

Figure 1.

(A) RNA cytidine acetylation is catalyzed by the Nat10 family of enzymes. (B) Prior work has focused on characterization of monomeric bacterial RNA acetyltransferases (PDB 2ZPA) [10] that are highly specific or dimeric eukaryotic RNA acetyltransferases which use adapters to address 5′-CCG-3′ sequences in specific substrates. Homo sapiens NAT10 is an AlphaFold model superimposed on the cryogenic electron microscopy (cryo-EM) structure of Kre33 from C. thermophilum (PDB 6RXX) [52]. (C) This study characterizes an adapter-independent archaeal RNA acetyltransferase that modifies 5′-CCG-3′ consensus sequence in diverse substrates. Thermococcus kodakarensis Nat10 is depicted as a dimer based on AlphaFold Multimer predictions (Supplementary Fig. S1).

Figure 2.

(A) Growth rate of WT T. kodakarensis, TkNat10 deletion strain, and strain containing a TkNat10 H506A catalytic mutant at 85°C. (B) Differential growth of T. kodakarensis strains at 85°C for 11 h. Values represent ≥3 replicates, analyzed by two-tailed Student’s t-test (*P< .05, **P< .01). (C) Purification of 6xHis-MBP-tagged TkNat10 enzyme. (D) Schematic for qualitative analysis of TkNat10 activity by anti-ac4C dot blot. (E) Acetylation catalyzed by TkNat10 is enzyme, acetyl-CoA, ATP, and heat-dependent. (F) TkNat10 catalysis as a function of temperature.

Figure 3.

(A) Schematic for transcriptome-wide identification of recombinant TkNat10 substrates in total RNA isolated from a TK0754 KO strain. (B) Correlation of misincorporation percentages observed upon treatment of KO RNA with recombinant TkNat10 enzyme (x-axis) or in RNA isolated from wild-type T. kodakarensis (y-axis). (C) Example of an enzyme-catalyzed misincorporation site in iorA messenger RNA (mRNA) which was validated by amplicon-based Sanger sequencing. (D) Correlation of ac4C-seq misincorporation percentage with minimum free energy (MFE) of the 40 base region surrounding the ac4C site calculated by ViennaRNA.

Figure 4.

(A) Structure of 16S and 18S SSU helix 45 rRNA model substrates. (B) Schematic for high-throughput analysis of TkNat10 substrate-specificity using a hairpin substrate. (C) Simple, Thorough, Rapid, Enriched Motif Elicitation (STREME) analysis of 3-bp motifs differentially enriched in reaction (NaCNBH₃-treated) and control (mock-treated) libraries. Both libraries were subjected to equivalent ac4C sequencing reactions.

Figure 5.

(A) Heatmap depicting % misincorporation (‘% CCG mutation’) as a function of time and temperature for 16S rRNA substrate. Each column corresponds to a mutant sequence with the mutation indicated in the top row of the header and the position in the bottom row of the header. Data represent the average of two replicate experiments. Raw data are given in Supplementary Table S2. (B) Heatmap depicting % misincorporation (‘% CCG mutation’) as a function of time and temperature for 18S rRNA substrate. Each column corresponds to a mutant sequence with the mutation indicated in the top row of the header and the position in the bottom row of the header. Data represent the average of two replicate experiments. Raw data are given in Supplementary Table S2. (C) Predicted secondary structure of 18S SSU helix 45 rRNA substate as well as C1, C7, and U20 mutants. (D) double mutant across three conditions (3 h 55°C, 3 h 65°C, and 5 h 65°C). The width and intensity of each arc is scaled to the strength of the interaction, with mutations that stimulate ac4C modification in orange and those that reduce it in blue. Secondary mutations with a small effect (averaging 10% or less) were depicted in gray. (E) Simplified schematic indicating the potential for 18S helix 45 mutation to form alternative RNA structures by interacting with scaffolding RNA. Left: The expected helix 45 surrogate forms and is predictably modulated by single mutations and double mutants in the 3′ stem. Right: When weakened helix 45 variants (e.g. U20) acquire a second mutation in the 5′ stem—particularly at position 9—they favor an alternative structure in which the consensus sequence becomes fully paired with an upstream sequence, reducing susceptibility to TkNat10 modification. Detailed exemplary structures provided in Supplementary Fig. S8. (F) Correlation of ac4C-seq misincorporation percentage with MFE for 18S rRNA variants. To focus exclusively on the helix 45 hairpin this analysis was tailored to mutants not predicted to form the alternative structure and included (i) all single mutants, (ii) mutants that strengthen the helix by forming new base pairs (e.g. C1, A18), and (iii) position 15–16 or 19–21 mutants with a secondary mutation in stem II. The misincorporation is derived from incubation of TkNat10 and cofactors with RNA at 3 h 65°C. C7 mutants are uniformly not modified under these conditions and thus were omitted from the graph for clarity.

Figure 6.

(A) Schematic for MALDI-TOF assay assessing TkNat10 RNA oligonucleotide modification. (B) Top: MALDI-TOF analysis of a single-stranded RNA oligonucleotide treated with no enzyme (left), TkNat10 WT + acetyl-CoA (middle), or TkNat10 S623G + propionyl-CoA (right). Full MALDI traces are provided in the Supplementary Information. Bottom: Predicted structure of TkNat10, indicating bulky residues in active site. Mutation of homologous residues in the human GNAT histone acetyltransferase enzyme KAT2A expands cofactor utilization. (C) LC-MS analysis comparing proteins enriched by an unmodified biotinylated RNA oligonucleotide versus a TkNat10 acetylated RNA oligonucleotide. (D) Gene ontology analysis of proteins preferentially enriched by an unmodified biotinylated RNA oligonucleotide versus a TkNat10-modified acetylated RNA oligonucleotide. (E) Crystal structure of PCBP1 bound to cytidine-containing RNA oligonucleotide. Modeling of N4-acetylation indicates a potential steric clash caused by cytidine modification.