SUPREM: an engineered non-site-specific m6A RNA methyltransferase with highly improved efficiency

Abstract

N 6-Methyladenine (m6A) RNA methylation plays a key role in RNA processing and translational regulation, influencing both normal physiological and pathological processes. Yet, current techniques for studying RNA methylation struggle to isolate the effects of individual m6A modifications. Engineering of RNA methyltransferases (RNA MTases) could enable development of improved synthetic biology tools to manipulate RNA methylation, but it is challenging due to limited understanding of structure-function relationships in RNA MTases. Herein, using ancestral sequence reconstruction, we explore the sequence space of the bacterial DNA methyltransferase EcoGII (M.EcoGII), a promising target for protein engineering due to its lack of sequence specificity and its residual activity on RNA. We thereby created an efficient non-specific RNA MTase termed SUPer RNA EcoGII Methyltransferase (SUPREM), which exhibits 8-fold higher expression levels, 7°C higher thermostability and 12-fold greater m6A RNA methylation activity compared with M.EcoGII. Immunofluorescent staining and quantitative liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis confirmed SUPREM's higher RNA methylation activity compared with M.EcoGII in mammalian cells. Additionally, Nanopore direct RNA sequencing highlighted that SUPREM is capable of methylating a larger number of RNA methylation sites than M.EcoGII. Through phylogenetic and mutational analysis, we identified a critical residue for the enhanced RNA methylation activity of SUPREM. Collectively, our findings indicate that SUPREM holds promise as a versatile tool for in vivo RNA methylation and labeling.

Graphical Abstract

Figure 1.

ASR of M.EcoGII. (A) Schematic maximum likelihood tree. The node values showed ultrafast bootstrap values calculated by IQ-TREE. (B, C) MTase-Glo assay of M.EcoGII and its ancestral proteins for (B) DNA and (C) RNA at a single time point. (B) The DNA methylation reaction was performed using 100 nM of methyltransferases, 50 nM DNA substrate (3569 bp containing 1819 adenine bases, Supplementary Table S2) and 10 μM SAM in 1× CutSmart buffer (NEB) at 37°C for 10 min. (C) The RNA methylation reaction was performed using 100 nM of the corresponding methyltransferases, 25 nM RNA substrate (1872 nt, containing 544 adenine bases, sequence in Supplementary Table S2) and 10 μM SAM in 1× CutSmart buffer (NEB) at 37°C for 60 min. Three independent experiments were performed, and each point showed an average of technical duplicates. P-values were determined by the one-way ANOVA test with the Dunnett test. (D) Time course experiments for RNA substrates. The condition of RNA methylation reaction was the same as panel (C) except for reaction time. Three independent experiments were performed, and each point showed an average of technical duplicates. SAH production amount was calculated by SAH standard curve. The lines represent simple linear regression analysis on the data from 0 to 120 min (SUPREM) or to 240 min (M.EcoGII, Anc291). (E) Normalized UV chromatograms of nucleoside digests of RNA treated with M.EcoGII or SUPREM at 254 nm. The full normalized UV chromatograms are shown in Supplementary Figure S5. (F) Fraction of m⁶A methylation in M.EcoGII- and SUPREM-treated samples (i.e. the amount of m⁶A as a fraction of total adenine, m⁶A + A, as determined by peak area at 254 nm), determined by LC–MS analysis. The RNA methylation reaction was performed using 5 μg of the corresponding methyltransferases, 1 μg RNA substrate (294 nt containing 80 adenine bases, sequence in Supplementary Table S2) and 320 μM SAM (NEB) in 1× CutSmart buffer (NEB) at 37°C for 120 min. Each point showed three independent methylation reactions. Error bars represent standard error of the mean.

Figure 2.

Methylation activity of M.EcoGII variants in HEK293T cells. (A) Representative images (63× objective with a 2× digital zoom) of HEK293T cells transfected with either eGFP, M.EcoGII-eGFP or SUPREM-eGFP constructs (reporter visible in green). m⁶A modifications were detected via immunofluorescence (in red) and the nucleus was stained with NucBlue™ (in blue). The scale bar is 10 μm. (B) Quantification of m⁶A fluorescence intensity in HEK293T transfected as described in panel (A). The bar chart indicates the normalized intensity by cell area. Statistical analysis was performed by one-way ANOVA followed by Tukey’s post-test (ns, not significant; *P <0.0332; **P <0.0021). (C, D) Analysis of m⁶A content in digested mRNA from cells transfected with eGFP, M.EcoGII-eGFP or SUPREM-eGFP by LC–MS/MS. (C) Extracted ion chromatogram at m/z = 282.11 corresponding to m⁶A. (D) Fraction of m⁶A methylation in eGFP-, M.EcoGII- and SUPREM-expressing HEK293T cells (i.e. the amount of m⁶A as a fraction of total adenine, m⁶A + A, as determined by calibration curve of m⁶A and A standards), determined by LC–MS analysis. Each point represents a different biological replicate (n = 3), which was analyzed in analytical duplicate by LC–MS. Statistical analysis was performed by one-way ANOVA followed by Tukey’s post-test (ns, not significant; **: P<0.0221; ***: P <0.0002).

Figure 3.

Analysis of RNA methylation site specificity of M.EcoGII and its variants by Nanopore DRS. (A) Sharkfin plots showing the logit P-value and absolute value of the logistic regression odds ratio for each 5-mer calculated by Nanocompore (n= 1456 adenine-containing 5-mers). Orange points indicate 5-mers that are significantly modified according to a stringent threshold (logit P-value <0.01 and absolute log odds ratio >0.5). Frequency of each base in modified 5-mers detected by Nanocompore using (B) a standard threshold (logit P-value <0.01) and (C) a stringent threshold (logit P-value <0.01 and absolute log odds ratio >0.5). For all three enzymes, a 5-mer was more likely to be methylated (according to the stringent threshold) if it contained multiple adenine bases (M.EcoGII, 3.6% of 5-mers containing >1 adenine were modified versus 0.3% of 5-mers containing 1 adenine; Anc291, 4.7% versus 0.8%; SUPREM, 8.7% versus 3.6%; P <0.0001 by Fisher’s exact test for all enzymes).

Figure 4.

Mutational and structural analysis of SUPREM. (A) Candidate amino acid residues for mutational analysis. (Left) M.EcoGII clade (clade I, 46 sequences) and SUPREM clade (clade II, 19 sequences). (Right) Comparison of amino acid sequence logo of mutational candidate sites based on multiple sequence alignment of clade I and clade II. The height of the logo represents the frequency of amino acid residues at each position. (B) RNA methylation activity of SUPREM variants. The RNA methylation reaction was performed using 100 nM methyltransferase and 25 nM RNA substrate (1872 nt, 544 adenine bases) with 10 μM SAM in 1× CutSmart buffer (NEB) at 37°C for 60 min. The luminescence was detected by MTase-Glo assay. Three or four independent experiments were performed, and each point shows an average of technical duplicates. P-values were determined by the one-way ANOVA test with the Dunnett test. (C) Size exclusion chromatogram of SUPREM. SUPREM with blue dextran (for estimation of void volume) was loaded into a Superdex 200 Increase 10/300 GL column, and the samples were eluted with 20 mM HEPES (pH 7.4) and 250 mM NaCl. The peak positions of low molecular weight markers were shown above. The two independent measurements were performed. (D) Dimer structural model of SUPREM generated by AlphaFold2 multimer program (version 2.2.1). Green spheres show residues selected for mutational analysis. (E) APBS calculation of the electrostatic surface potential of SUPREM (red: negative charge; blue: positive charge).