Defining substrate specificities of human RNA capping methyltransferases through quantitative assessment of independent yet cooperative activities

Abstract

Human RNA capping is critical for mRNA splicing, protection of RNA from 5′ exonucleases in the cytoplasm, and targeting to the ribosome. Human RNMT, CMTR1, and CMTR2 are RNA methyltransferases involved in the RNA capping process. They play a significant role in the proliferation and differentiation of embryonic stem cells and have been implicated in cancer. Substrate specificities of human RNA capping methyltransferases have been somewhat explored in a few studies. Here, we report on a comprehensive, systematic, and quantitative assessment of their substrate specificities along with SARS‐CoV‐2 counterparts, nsp14 and nsp16. We discovered novel cooperative activities of human enzymes. We designed and synthesized various RNA substrates with defined patterns of methylation to systematically assess the dependency or cooperativity of their activities using radiometric assays followed by mass spectrometry to verify RNA methylation status. We have tested all five enzymes in parallel against these substrates and determined kinetic parameters. Our data not only indicate that the catalytic activities of human RNMT, CMTR1, and CMTR2 are distinct and nonoverlapping, but also provide a novel quantitative assessment of their activities, indicating how and to what extent these proteins affect each other's function. Unlike nsp14 and nsp16, their functions are not necessarily sequential, but show significant cooperativity. Altogether, our data provide a comprehensive understanding of substrate specificities of human RNA capping methyltransferases, enabling the development of potential future anticancer therapeutics and assessment of antiviral therapeutics' selectivity.

1. INTRODUCTION

Addition of a noncoding nucleotide to the 5′ end of RNA is known as RNA capping (Ghosh and Lima 2010; Muthukrishnan et al. ¹⁹⁷⁸; Ramanathan et al. ²⁰¹⁶; Tsukamoto et al. ¹⁹⁹⁸). The presence of 5′ cap structures on eukaryotic mRNA is critical for mRNA splicing, protection of the RNA from 5′ exonucleases in human cytoplasm, and enabling them to get to the ribosome (Nencka et al. 2022; Shuman ²⁰⁰²; Shuman ²⁰¹⁵; Trotman and Schoenberg ²⁰¹⁹). Typically, it starts with the removal of an inorganic phosphate from the 5′‐triphosphate terminus of the newly transcribed RNA by an RNA triphosphatase, followed by guanylate cap formation catalyzed by an RNA guanylyltransferase (Ghosh and Lima 2010; Ramanathan et al. ²⁰¹⁶). In humans, RNA capping is associated with the methylation of the guanosine at N⁷ by RNA guanine‐N⁷ methyltransferase (RNMT), and at 2′‐O‐ribose of the first and second transcribed nucleotides by cap‐1 methyltransferase (CMTR1) and cap‐2 methyltransferase (CMTR2), respectively (Figure 1). The capped mRNA is then recognized by the human immune system as its own (Nencka et al. 2022). Coronaviruses mimic the human RNA capping process to evade the human immune system. Two SARS‐CoV‐2 non‐structural proteins, nsp14 and nsp16, are RNA methyltransferases that create cap‐0 and cap‐1 by methylating N⁷‐guanosine and 2′‐O‐ribose of the first transcribed nucleotide of viral RNA, respectively (Figure 1) (Devkota et al. 2021; Khalili Yazdi et al. ²⁰²¹; Nencka et al. ²⁰²²). Nsp14 is a bifunctional protein with a C‐terminal methyltransferase domain and an N‐terminal exoribonuclease domain (Bouvet et al. 2012; Eckerle et al. ²⁰¹⁰). Although it complexes with nsp10 for exonuclease activity, it does not need nsp10 to function as an RNA methyltransferase (Devkota et al. 2021). However, nsp16 is only active when in complex with nsp10 (Khalili Yazdi et al. 2021). A significant number of structures of human RNMT (Pearson et al. 2025; Varshney et al. ²⁰¹⁶), CMTR1 (Smietanski et al. 2014), and SARS‐CoV‐2 nsp14 (Czarna et al. 2022; Imprachim et al. ²⁰²³; Kottur et al. ²⁰²²) and nsp10–nsp16 (Inniss et al. 2023; Kalnins et al. ²⁰²⁴; Minasov et al. ²⁰²¹) RNA cap methyltransferases have been determined in complex with SAM, SAM analogues, and various cap RNA which provides an in‐depth understanding of substrate interactions (Figure 2).

FIGURE 1.

RNA capping in human and coronaviruses. Human mRNA capping includes methylation of the cap guanosine at N⁷ by RNMT (cap‐0), and methylation of the first and second transcribed nucleotides at 2′‐O‐ribose by CMTR1 (cap‐1) and CMTR2 (cap‐2), respectively. SARS‐CoV‐2 nsp14 also methylates the cap guanosine at N⁷, followed by methylation of the 2′‐O‐ribose of the first transcribed nucleotide by the SARS‐CoV‐2 nsp10–nsp16 complex (cap‐1) (reviewed in Nencka et al. ²⁰²²).

FIGURE 2.

Structural basis for RNA substrate selectivity of human and SARS‐CoV‐2 RNA methyltransferases. (a) Crystal structure of human RNMT (blue cartoon and surface) in complex with the SAM analogue, sinefungin and GMP‐PnP (purple sticks). The image was generated using PDB id 8Q9W (Pearson et al. 2025). Red arrow indicates the methylation site of RNMT. (b) Structural model of SARS‐CoV‐2 nsp14 in complex with SAH and mN⁷GpppG. The image was generated using various structural states of nsp14, available in the Protein Data Bank with accession codes 7QIF (Imprachim et al. 2023), 7TW7 (Kottur et al. 2022), and 7R2V (Czarna et al. 2022). The nsp14 exonuclease (ExoN) domain (residues 67–281) is colored teal, the methyltransferase (MTase) domain is colored yellow green (301–524), and the hinge domain (residues 282–300) connecting ExoN and MTase is colored yellow. SAH and mN⁷GpppG are shown as purple sticks. Two conformations of the SAM‐interacting loop (residues 352–362) are shown in yellow green and gray cartoon. In the presence of SAM or SAH, the loop assumes the conformation in gray. In the absence of SAM/SAH, the loop is in the conformation shown in yellow green. Red arrow indicates the methylation site of nsp14 (N⁷G). (c) Crystal structure of human CMTR1 (blue cartoon and surface) in complex with SAM and mN⁷GpppGAUC fragment (purple sticks) (PDB id 4N48) (Smietanski et al. 2014). Arrows indicate methylation sites, with the red arrow pointing at the CMTR1 methylation site (N1 ribose 2′‐OH). First three transcribed nucleotides downstream of the mN⁷Gppp cap are numbered 1, 2, and 3. Broken lines indicate electrostatic interactions between CMTR1 residues and RNA substrate. (d) Crystal structure of SARS‐CoV‐2 nsp10–nsp16 heterodimer in complex with mN⁷GpppmAUUA (cap‐1) and SAH. Nsp16 (green) and nsp10 (red) are shown in cartoon and surface representation, while SAH and RNA substrate are shown in purple sticks. Broken lines indicate electrostatic interactions between nsp16 and bound ligands. Black and red arrows indicate RNA methylation sites, with the red arrow pointing to the methylation site of nsp16. First four transcribed nucleotides downstream of the mN⁷Gppp cap are numbered 1–4. Structures of nsp16 in various states were used to show the loop conformations and rotamers of D6873 near the N2 ribose 2′‐OH methylation site. Structures with the following PDB accession codes were used to assemble the image: 7BQ7, 7L6R (Minasov et al. 2021), 8RV5 (Kalnins et al. 2024), 8F4Y (Inniss et al. 2023). (e) Surface and cartoon representation of the human CMTR2 homology model with modeled RNA substrate and SAM (Smietanski et al. 2014). Broken arrows indicate interaction between CMTR2 residues and RNA substrate, and interactions between the bases of N1 and N3 that participate in a non‐canonical cis‐Hoogsteen‐Hoogsteen base pairing. Arrows indicate methylation sites, with the red arrow pointing at the N2 ribose 2′‐OH group (CMTR2 methylation site). (f) Superposition of the experimental structure of human CMTR1 (residues 126–550, gray cartoon) and a previously published three‐dimensional homology model of human CMTR2 (residues 1–423, sky blue cartoon) (Smietanski et al. 2014). The respective ligands of CMTR1 are shown in gray sticks, while those of CMTR2 are shown in purple sticks. Numbers indicate transcribed nucleotides downstream of the mN⁷Gppp cap. Black arrows indicate the respective methylation sites of CMTR1 (N1 ribose 2′‐OH) and CMTR2 (N2 ribose 2′‐OH). The models superimpose with r.m.s.d. of 3.9 Å.

Both RNMT and CMTR1 are nuclear proteins with three and one nuclear localization signals (NLS), respectively (Liang et al. 2023; Shafer et al. ²⁰⁰⁵). CMTR2 activity was originally reported only in cytoplasmic fractions of HeLa cells (Langberg and Moss 1981). However, subsequent studies demonstrated that the enzyme is present in both the cytosol and the nucleus (Werner et al. 2011). RNMT is also involved in RNA recapping in the cytoplasm (Trotman and Schoenberg 2019). Cytoplasmic recapping by RNMT restores the mN⁷G cap to the de‐capped or endo‐nucleolytically cleaved mRNAs (Trotman and Schoenberg 2019). Although RNMT and CMTR1 mainly function co‐transcriptionally, CMTR2 methylates the RNA cap predominantly in the cytoplasm (Langberg and Moss 1981; Liang et al. ²⁰²³; Werner et al. ²⁰¹¹). Interestingly, coronaviral capping is a cytoplasmic process (Matsuda et al. 2024). The lack of RNA capping such as 2′‐O‐methylation of viral RNA would be detected by pattern recognition of cytoplasmic RNA sensors such as MDA5 and RIG‐I receptors, leading to the induction of type I interferon and the full human immune response (Zust et al. 2011). Interferon‐induced proteins with tetratricopeptide repeats, IFIT1 and IFIT5, play key roles in detecting single‐stranded RNAs lacking 2′‐O‐methylation in viral infection (Miedziak et al. 2020).

The function of human RNA capping enzymes and protein synthesis are tightly regulated processes, dysregulation of which is associated with the development and progression of various cancers (Aregger et al. 2016; Blagden and Willis ²⁰¹¹; Gonatopoulos‐Pournatzis et al. ²⁰¹¹). RAM, a small protein, is an activating subunit for RNMT at the transcription site. RAM is required for recruiting RNMT at the transcription site and is essential for cell survival (Gonatopoulos‐Pournatzis et al. ²⁰¹¹), but is not essential for its RNA methyltransferase activity (Perveen et al. 2024). In addition, CDK1‐cyclin B1 phosphorylates the RNMT regulatory domain during G2/M phase of the cell cycle, which activates the enzyme both directly and indirectly, leading to an increase in N⁷G methylation in G1 phase and elevated transcription (Aregger et al. 2016). Inhibition of RNMT phosphorylation reduces the cell proliferation rate, suggesting RNMT as a great target for development of cancer therapeutics (Aregger et al. 2016). Also, the regulation of RNMT and CMTR1 have been reported to affect embryonic stem (ES) cells differentiation. RNMT is highly expressed in ES cells and repressed during differentiation which leads to reduced N⁷G methylation. By contrast, CMTR1 is upregulated during differentiation (Grasso et al. 2016; Liang et al. ²⁰²²; Liang et al. ²⁰²³) which is important for maintenance of histone and ribosomal protein gene transcripts in differentiating ES cells (Grasso et al. 2016). These observations clearly indicate a very specific involvement of human RNA capping enzymes in regulation of expression, translation, and maintenance of needed proteins in proliferating and differentiating cells.

Methylation of N⁷ guanosine (cap‐0) by human RNMT and SARS‐CoV‐2 nsp14, 2′‐O‐ribose methylation of the first transcribed nucleotide (cap‐1) by CMTR1 and nsp10–nsp16 complex, and the second transcribed nucleotide (cap‐2) by CMTR2 have been well established (Nencka et al. 2022). Partial substrate specificities of CMTR1 (Belanger et al. 2010) and CMTR2 (Werner et al. 2011) have also been assessed using a combination of ³²P‐labelling of the cap, nuclease treatment, and thin layer chromatography (TLC).

Here, we aimed to better understand the patterns of human RNMT, CMTR1, and CMTR2 RNA methyltransferase, and SARS‐CoV‐2 nsp14 and nsp16 activities in vitro through a systematic and quantitative kinetic assessment of their substrate specificities and potential cooperative activities. We designed five RNA substrates with various patterns of methylation (Table 1), chemically synthesized them, and tested them with human RNMT, CMTR1, CMTR2, and SARS‐CoV‐2 nsp14 and nsp10–nsp16 complex in parallel using optimized radiometric assays that directly monitor product formation as well as using mass spectrometry to accurately verify RNA methylation status. We report more specific and quantitative details of distinct substrate specificities of these RNA capping methyltransferases, which may help better understand the roles they play in health and disease.

TABLE 1.

Sequences and abbreviations for RNA substrates.

Full substrate sequence	Abbreviation
5′‐GpppACCCCCCCCC‐Biotinylated 3′	GpppAC
5′‐[(N7 methylated)G] pppACCCCCCCCC‐Biotinylated 3′	mGpppAC
5′‐Gppp[A(2′‐o‐ribose methylated)] CCCCCCCCC‐Biotinylated 3′	GpppAmC
5′‐[(N7 methylated)G] pppA[C(2′‐o‐ribose methylated)] CCCCCCCC‐Biotinylated 3′	mGpppACm
5′‐[(N7 methylated)G] ppp[A(2′‐o‐ribose methylated)] CCCCCCCCC‐Biotinylated 3′	mGpppAmC

Note: The complete sequences of synthetic RNA substrates used to investigate substrate specificities of RNMT, CMTR1, CMTR2, nsp14, and nsp10–nsp16 in this study and their corresponding abbreviations are presented.

2. MATERIALS AND METHODS

2.1. Protein expression and purification

RNMT (Perveen et al. 2024), nsp14 (Devkota et al. 2021), and nsp10–nsp16 (Khalili Yazdi et al. 2021) were expressed and purified as previously described (Devkota et al. 2021; Khalili Yazdi et al. ²⁰²¹; Perveen et al. ²⁰²⁴). CMTR1 and CMTR2 were expressed in Sf9 cells and purified by multi‐step column chromatography. Detailed protocols for the purification of each protein are provided in Data S1, Supporting Information.

2.2. Activity assays

S‐adenosyl‐L‐methionine (³H‐SAM), Streptavidin‐coated SPA beads (Cat# RPNQ0007), and clear‐bottom plates (384‐well, Cat# 6007490; 96‐well, Cat# 6005040) were obtained from Revvity (Waltham, USA). Custom‐synthesized single‐stranded RNA substrates, hereafter referred to as RNA substrates, were sourced from BioSYNTHESIS (Lewisville, USA) or TriLink BioTechnologies (San Diego, USA). All RNA solutions were prepared by dissolving the substrate in nuclease‐free water with RNaseOUT™ recombinant ribonuclease inhibitor (Thermo Fisher, Cat# 10777019). S‐adenosyl‐L‐methionine (SAM) (Cat# QA‐2305) was purchased from Combi‐Blocks (San Diego, USA).

2.3. Characterization of methyltransferase substrate specificity

Methyltransferase activity assays were performed to characterize the substrate specificity of nsp14, nsp10–nsp16 complex, RNMT, CMTR1, and CMTR2 using optimized conditions for each protein (Table S1). In all experiments, the concentration of S‐adenosyl‐L‐methionine (³H‐SAM) was fixed at 1000 nM, while all RNA substrate concentrations were maintained at 800 nM. Each enzyme was titrated over a concentration range of 1 to 1000 nM to better detect even low activities and assess substrate preference. All reactions were stopped with 7.5M guanidine hydrochloride, followed by adding 10 mg of streptavidin‐coated SPA beads to each reaction mixture. The reaction volume was then topped up with 160 μL of 20 mM Tris–HCl (pH 7.5). Methylation levels were measured in 96‐well clear‐bottom plates using the MicroBeta² liquid scintillation and luminescence counter (Revvity, Waltham, USA).

2.4. Kinetic parameter determination

All experiments were performed under the optimized conditions for each enzyme (Table S1) with RNA substrates the enzyme showed activity with, as detailed below. A standard ³H‐biotin titration curve was used to calculate velocities and k _cat values.

In cases where reaction velocity curves were sigmoidal and did not fit to the classic Michaelis–Menten equation, we used the allosteric sigmoidal equation for fitting. We interpreted this as possible positive cooperativity. Kinetic parameters were thus extracted as K₁_/₂ values instead of K _m to more accurately capture the values from non‐hyperbolic patterns. This is not expected to affect the direct comparison of the substrate specificities.

2.4.1. nsp14

The MTase activity of SARS‐CoV‐2 nsp14 was measured using the radiometric methyltransferase assay monitoring the transfer of a tritiated methyl group from ³H‐SAM to a biotinylated RNA substrate GpppAmC. Kinetic parameters (K _m and k _cat) were determined by varying RNA (12.5–300 nM) while keeping ³H‐SAM fixed at 1.5 μM and nsp14 fixed at 1.5 nM. Data were analyzed using a Michaelis–Menten model in GraphPad Prism.

2.4.2. nsp16

The activity of SARS‐CoV‐2 nsp10–nsp16 methyltransferase (MTase) complex was assessed using a radiometric methyltransferase assay. The reaction was performed using a biotinylated RNA substrate mGpppACm with varying RNA concentration from (25–300 nM) and fixed ³H‐SAM at 8 μM in an optimized buffer with fixed concentration of nsp10–nsp16 at 250 nM. Kinetic parameters (K _1/2 and k _cat) were calculated using a sigmoidal model in GraphPad Prism.

2.4.3. Human RNMT

Radiometric methyltransferase assay was performed using biotinylated RNA GpppAmC and ³H‐SAM in an optimized reaction buffer. Kinetic parameters were determined by varying RNA concentrations (10–240 nM) while keeping ³H‐SAM fixed at 1.5 μM and RNMT concentration of 5 nM. Data were fitted to a sigmoidal model using GraphPad Prism to calculate K _1/2 and k _cat.

2.4.4. Human CMTR1

Radiometric methyltransferase assays to characterize the kinetic properties of CMTR1 activity were carried out in an optimized buffer with a fixed CMTR1 concentration of 50 nM. The enzyme kinetics were assessed by varying RNA substrate mGpppAC from 10 to 160 nM, while maintaining a fixed SAM concentration of 700 nM and varying SAM concentration (12.5–400 nM) at a fixed RNA concentration of 800 nM. Furthermore, for calculating kinetic parameters for additional substrates GpppAC and mGpppACm, SAM concentrations were fixed at 700 nM and RNA concentrations were varied from 5 to 160 nM and 10 to 200 nM, respectively. Kinetic parameters (K _1/2 and k _cat) were calculated using a sigmoidal model in GraphPad Prism.

2.4.5. CMTR2

To characterize CMTR2 kinetics, CMTR2 concentration was fixed at 500 nM while RNA concentration GpppAmC was varied from 50 to 800 nM and while SAM concentration was fixed at 4 μM. Additionally, for assessing the kinetic parameters of mGpppAmC, RNA substrate varied from 12.5 to 800 nM, while keeping the concentration of SAM and CMTR2 fixed at 4 μM and 100 nM, respectively. Also, for calculating SAM kinetic parameters, RNA concentration was fixed at 1 μM and SAM varied from 0.5 to 6 μM, while CMTR2 concentration was fixed at 100 nM. Kinetic parameters (K _1/2 and k _cat) were calculated using a sigmoidal model in GraphPad Prism. More complete protocols for all assays are detailed in Data S1.

2.5. Mass spectrometry sample preparation

All experiments were conducted under optimized buffer conditions for each enzyme (Table S1), supplemented with RNaseOUT™ recombinant ribonuclease inhibitor to prevent RNA degradation through the RNA sample purification and mass spectrometry. As the experiments were intended as Yes/No for confirmation of methylation of the new RNA substrates by specific protein(s), samples were incubated at 23°C for 4 h, ensuring sufficient reaction time. Since these reactions were intended to confirm methylation rather than quantify rate, linearity over time was not required. Non‐radiolabeled SAM was used for the preparation of samples for MS. All samples were prepared in triplicate.

After incubation of the reaction mixture, oligonucleotide purification and concentration were performed using the Oligo Clean‐Up and Concentration Kit (Norgen Biotek, Cat# 34100) according to the manufacturer's protocol. Oligonucleotide sample volume was adjusted to 50 μL, followed by the addition of 150 μL Buffer RL containing β‐mercaptoethanol (10 μL per 1 mL of Buffer RL) and 300 μL isopropanol (ratio of 1.3.6, respectively) then mixed by vortex for 10 s. The mixture was applied to a spin column, centrifuged at 14,000g for 1 min, and the flowthrough was discarded. For larger sample volumes, the binding step was repeated. The column was then washed three times with 400 μL Wash Solution A, each followed by centrifugation at 14,000g for 1 min. A final dry spin (2 min at 14,000g) was performed to remove residual ethanol. Elution was carried out by adding 30 μL Elution Solution A to the column, followed by centrifugation at 200g for 2 min, and then 14,000g for 2 min. Purified RNA oligonucleotides were stored at −70°C until further use.

The presence of methylated RNA products was assessed using liquid chromatography–mass spectrometry (LC–MS) analysis. Noncanonical substrates were tested against each enzyme, as described in Data S1.

3. RESULTS

Our lab previously reported the kinetic characterization of human RNMT (Perveen et al. 2024) and nsp14 (Devkota et al. 2021) methyltransferase activities using their known substrate, unmethylated N⁷‐guanosine cap RNA (GpppAC; Table 1). Similarly, we characterized nsp10–nsp16 complex (Khalili Yazdi et al. 2021) activity using N⁷‐methylated guanosine RNA substrate (mGpppAC). Human CMTR1 is also known to methylate the same RNA substrate at 2′‐O‐ribose of the first transcribed nucleotide (Belanger et al. 2010; Nencka et al. ²⁰²²), and human CMTR2 methylates the second nucleotide of “mGpppAmC” substrate at 2′‐O‐ribose (Figure 1) (Nencka et al. 2022; Werner et al. ²⁰¹¹). It may be an assumption that RNA capping enzymes work sequentially (Nencka et al. 2022). Although based on partial characterizations, human RNMT, CMTR1, and CMTR2 have been reported to be able to function independently of pre‐existing RNA methylation marks (Belanger et al. 2010; Werner et al. ²⁰¹¹), the effect and extent of such methylations on the function of each cap RNA methyltransferase has not been quantitatively evaluated. In this study, we explored the substrate specificity of the three human RNA capping methyltransferases as well as coronaviral nsp14 and nsp10–nsp16 complex in parallel systematically and further investigated potential overlap, dependency, or cooperativity between the enzymatic activities of these proteins quantitatively. We developed and used radiometric assays in 96‐ and 384‐well format that directly monitor product formation and used mass spectrometry to accurately verify RNA methylation status. In lack of kinetic data for CMTR1 and CMTR2, we first set out to determine their kinetic parameters for methyltransferase activity using mGpppAC and mGpppAmC as substrates, respectively (Table 2 and Figure 3).

TABLE 2.

Kinetic characterization of substrate specificities of RNMT, CMTR1, CMTR2, nsp14, and nsp10–nsp16 complex.

Protein	Substrate
	GpppAC		mGpppAC		GpppAmC		mGpppACm		mGpppAmC
	K m/K 1/2 (nM)	k cat (h−1)	K m/K 1/2 (nM)	k cat (h−1)	K m/K 1/2 (nM)	k cat (h−1)	K m/K 1/2 (nM)	k cat (h−1)	K m/K 1/2 (nM)	k cat (h−1)
nsp14	43 ± 15 a	48 ± 4 a	NA	NA	74 ± 24	121 ± 5	NA	NA	NA	NA
nsp16	NA	NA	1600 ± 400 a	16 ± 1.2 a	NA	NA	71 ± 4	0.7 ± 0.06	NA	NA
RNMT	158 ± 8 a	91 ± 1 a	NA	NA	50.1 ± 3.4	218 ± 12	NA	NA	NA	NA
CMTR1	48 ± 12	11 ± 1.6	56 ± 5	10.2 ± 0.5	NA	NA	55 ± 23	9 ± 1.3	NA	NA
CMTR2	NA	<0.1	NA	<0.1	103 ± 10	0.31 ± 0.01	NA	NA	113 ± 31	3.4 ± 0.4

Note: All values are from Figures 3 and 7. All experiments were performed in triplicate and values are presented as average ± standard deviation. NA, not applicable as no activity was observed.

^a The kinetic parameters for nsp14 (Devkota et al. 2021) and RNMT (Perveen et al. 2024) with GpppAC, and nsp16 (Khalili Yazdi et al. 2021) with mGpppAC were previously reported.

FIGURE 3.

Kinetic characterization of CMTR1 and CMTR2 RNA methyltransferase activities. Initial velocities were determined using mGpppAC as substrate for CMTR1, by (a(i)) varying concentrations of RNA at fixed concentration of SAM and (b(i)) varying SAM concentrations at fixed concentration of RNA substrate. The data from a(i) and b(i) were plotted and fit into sigmoidal equation to determine the K _1/2 of (a(ii)) RNA and (b(ii)) SAM and k _cat values as described in Data S1. Similarly, the initial velocities were determined for CMTR2 using mGpppAmC as substrate. Initial velocities in various concentrations of (c(i)) RNA and (d(i)) SAM were fit to sigmoidal equation (c(ii) and d(ii), respectively) to determine the kinetic parameters. All experiments were performed in triplicate and error bars represent the standard deviation.

The assay condition for CMTR1 was optimized (Figure S1) and was used for further studies at 50 nM of CMTR1 (Figure 3 and Table S1). The optimum MgCl₂ concentration for in vitro assays varies among RNA methyltransferases (Table S1). To determine the kinetic parameters (K _m/K _1/2 for RNA and k _cat), the initial velocities were determined by testing the activity of the enzyme at a fixed concentration of ³H‐SAM (700 nM) and varying concentrations of RNA (mGpppAC; Table 1) from 10 to 160 nM (Figure 3a(i)). Similarly, the linear initial velocities for CMTR1 were determined at a fixed concentration of RNA (800 nM) and increasing concentrations of SAM from 12.5 to 400 nM (Figure 3b(i)). Both sets of data for RNA and SAM were plotted, which fitted better to an allosteric sigmoidal equation (Figure 3a(ii),b(ii)) than the Michaelis–Menten equation. This may indicate some level of substrate binding‐related cooperativity. This curve fitting approach will not affect the conclusions made in this report; rather, it provides more accurate kinetic values for the direct comparison of the substrate specificities. The K _1/2 (K _m) values of 56.2 ± 4.8 nM for RNA (Figure 3a(ii)) and 95.6 ± 6.8 nM for SAM (Figure 3b(ii)) were determined. The k _cat (turnover number) for CMTR1 with this substrate was 10.2 ± 0.5 h⁻¹. The turnover number is the number of substrate molecules converted to product by a single enzyme molecule per unit of time at saturation of substrate. In addition to the RNA substrate with 10 transcribed nucleotides (Table 1), we also determined similar kinetic parameters for CMTR1 using a shorter RNA substrate with 6 transcribed nucleotides (RNA K₁/₂ of 44.9 ± 4.3 nM and k _cat of 13.8 ± 0.6 h⁻¹) (Figure S2), indicating the size of the RNA substrate we used is long enough for probing the substrate specificity of CMTR1. The kinetic parameters were also within the same range with a substrate with a different transcribed nucleotide sequence (m⁷GpppAUUAAA) (Figure S3), supporting the potential conclusion that CMTR1 could methylate the 2′‐O‐ribose of the first transcribed nucleotide (A) on RNA regardless of the neighboring nucleotide.

CMTR2 was active in CMTR1 assay conditions. Therefore, the same assay conditions were used for both proteins for better comparison of substrate specificities. Using mGpppAmC as a substrate for CMTR2 (100 nM), the RNA concentration varied from 12.5 to 800 nM at a fixed concentration of SAM (4 μM) (Figure 3c(i)), and the SAM concentration varied from 0.5 to 6 μM at a fixed concentration of RNA (1 μM) (Figure 3d(i)). Both sets of data were also fit to a sigmoidal equation, and K _1/2 values of 113 ± 31 nM (Figure 3c(ii)) and 1.3 ± 0.1 μM (Figure 3d(ii)) were determined for RNA and SAM, respectively. The k _cat for CMTR2 was 3.4 ± 0.4 h⁻¹.

Exploring the substrate specificity of human RNMT, CMTR1, and CMTR2, and SARS‐CoV‐2 nsp14 and nsp10–nsp16 complex, we then evaluated their ability to methylate five substrates with different patterns of pre‐methylation (Table 1 and Figure 4). We used highly pure synthetic cap RNA with 10 transcribed nucleotides (Table 1). The size was reasonable for this study as only the cap and a few first transcribed nucleotides (<10 nucleotides) have been typically shown to be involved in binding or catalytic activities of cap RNA methyltransferases in various species as described in detail in the discussion. These potential substrates were: (1) unmethylated RNA (GpppAC), (2) N⁷ methylated guanosine (mGpppAC), (3) methylated at the first tentative transcribed nucleotide, adenosine (GpppAmC), (4) methylated at both N⁷ guanosine and at 2′‐O‐ribose of the second tentative transcribed nucleotide, cytosine (mGpppACm), and (5) at both methylated N⁷ guanosine and 2′‐O‐ribose of adenosine (mGpppAmC). For simplicity, we are referring to each substrate in abbreviated forms as shown in Table 1. We first performed the radiometric assays at increasing concentrations of enzymes to ensure that we capture even the lowest levels of activities compared to each protein's canonical substrate. As expected, RNMT and nsp14 were active with the unmethylated RNA substrate GpppAC, which previously was used as a substrate for RNMT (Perveen et al. 2024) and nsp14 (Devkota et al. 2021). However, both enzymes were inactive with three RNA substrates with N⁷G pre‐methylation (mGpppAC, mGpppAmC, mGpppACm). This indicates that both RNMT and nsp14 solely methylate N⁷ guanosine within the RNA capping process. Interestingly, both were significantly more active with the substrate with methylation at the first transcribed nucleotide (GpppAmC) (Figure 4 and Table 2).

FIGURE 4.

Initial assessment of substrate specificity of RNA capping methyltransferases. The ability of human RNMT, CMTR1, CMTR2, and SARS‐CoV‐2 nsp14 and nsp10–nsp16 complex to methylate RNA substrates with various methylation patterns was assessed by monitoring their activities at increasing concentrations of enzymes using the radiometric assays as described in Data S1. For simplicity of referring to these substrates, we use the abbreviated forms as indicated in Table 1. Note that at low enough concentrations of enzyme in each case, the data fit a linear equation as expected. Here, we intentionally extended the range of protein concentrations to capture even the very low levels of activities. All experiments were performed in triplicate, and error bars represent the standard deviation.

CMTR1 was highly active with the unmethylated and methylated substrates only when 2′‐O‐ribose of adenosine (the first transcribed nucleotide) was not methylated (GpppAC, mGpppAC, and mGpppACm), indicating that this protein only methylates 2′‐O‐ribose of the first transcribed nucleotide independent of any other RNA capping methyltransferases (Figure 4 and Table 2). Altogether, these observations revealed that human RNMT and CMTR1 RNA methyltransferase activities are nonoverlapping, and they could function totally independently. Interestingly, RNA methylation at 2′‐O‐ribose of adenosine by CMTR1 further accelerated N⁷G methylation by RNMT, a cooperative effect. However, such an effect on nsp14 will not be observed during viral RNA capping as the nsp10–nsp16 complex is not capable of methylating 2′‐O‐ribose of adenosine in the absence of N⁷ methylated guanosine. CMTR1 was most active when the second transcribed nucleotide was methylated along with N⁷G (mGpppACm), another cooperative effect this time between CMTR1 and CMTR2 (Figure 4 and Table 2).

CMTR2 was highly active with the substrate methylated on both N⁷G and the 2′‐O‐ribose of the first transcribed nucleotide (mGpppAmC), but completely inactive with mGpppACm, which includes methylation of the second transcribed nucleotide. In the absence of N⁷G methylation, CMTR2 showed only low levels of activity, and more when the neighboring adenosine was methylated (Figure 4 and Table 2). These observations confirmed that CMTR2 only methylates the 2′‐O‐ribose of the second transcribed nucleotide with a significant multi‐level cooperativity effect between all, RNMT, CMTR1, and CMTR2. Altogether, the functions of the three human RNA capping methyltransferases are non‐overlapping, and they also could function independently, yet the levels of their activities are greatly affected by the patterns of RNA methylation of the substrates (Figure 4).

Based on the patterns of enzymatic activities of human RNMT, CMTR1, CMTR2, and SARS‐CoV‐2 nsp14 and nsp16 in the radiometric assays using the five substrates (Figure 4), we concluded that each of these RNA methyltransferases only monomethylates one specific nucleotide at a specific position, consistent with previous reports (Belanger et al. 2010; Werner et al. ²⁰¹¹). To further confirm these observations, we prepared non‐radiolabeled reactions with all the active substrates in parallel, purified the RNA product from these reaction mixtures using a commercially available kit as described in section 2, and evaluated the presence of methylated RNA products using liquid chromatography and high‐resolution mass spectrometry analysis (Figures 5, S4, and S5). Methylated RNA products were separated from unmethylated substrates using a reversed phase column and hexafluoro‐2‐isopropanol (HFIP) buffer system containing an alkylamine ion‐pairing reagent, followed by LC–MS analysis (Chen et al. 2021). High‐resolution mass spectrometry was employed to qualitatively identify the methylated products (Figure S4). All methylated products were chromatographically resolved from their corresponding unmethylated substrates (Figure 5).

FIGURE 5.

Chromatographic separation of methylated products from unmethylated substrate from LC–MS runs. (a) nsp14 with GpppAC, (b) nsp14 with GpppAmC, (c) nsp16 with mGpppAC, (d) nsp16 with mGpppACm, (e) RNMT with GpppAC, (f) RNMT with GpppAmC, (g) CMTR1 with GpppAC, (h) CMTR1 with mGpppAC, (i) CMTR1 with mGpppACm, (j) CMTR2 with mGpppAC, (k) CMTR2 with GpppAmC, and (l) CMTR2 with mGpppAmC. The chromatographic peak intensities indicate the abundance of the substrates or products. The high‐resolution mass spectrometric (HRMS) analysis of RNMT, CMTR1, CMTR2, nsp14, and nsp10–nsp16 enzymatic reactions with five RNA substrates with various patterns of methylation is presented in Figures S2 and S3. All experiments were performed in triplicate.

In most cases, methylated RNA products exhibited longer retention times (RT) than their unmethylated counterparts, suggesting reduced polarity upon methylation. However, an exception was observed for RNMT and nsp14, where the methylated products of using GpppAC and GpppAmC were eluted earlier than the unmethylated substrates. These observations are consistent with the known function of RNMT and nsp14, which methylate N⁷‐guanosine (Krafcikova et al. 2020; Nencka et al. ²⁰²²; Perveen et al. ²⁰²⁴). The resulting quaternary nitrogen increases the polarity of the methylated products, leading to their earlier elution (shorter RT) compared to the unmethylated substrates (Figure 6). The unmethylated substrate, GpppAC, was incubated with RNMT, nsp14, CMTR1, and CMTR2. High‐resolution mass spectrometry (HRMS) confirmed that each enzymatic reaction produced only one mono‐methylated product with the same molecular weight (Figures S4 and S5), supporting the notion that RNMT, CMTR1, and CMTR2 as well as nsp14 and nsp16 only monomethylate one specific target position (Figures 4 and 5). Interestingly, the methylated products generated by RNMT and nsp14 eluted earlier than the unmethylated substrate, whereas those from CMTR1, CMTR2, and nsp16 eluted later (Figure 6). This further revealed which enzyme methylates N⁷G versus 2′‐O‐ribose of the first and second transcribed nucleotides.

FIGURE 6.

Chromatographic separation of unmethylated GpppAC substrate and methylated product(s). The methylated product of RNMT (red) is more polar than its unmethylated substrate (blue), whereas the methylated product of CMTR1 (green) is less polar than its unmethylated substrate. The chromatographic peak intensities are adjusted according to the abundance of substrate or products. All experiments were performed in triplicate.

To assess these observed effects more quantitatively, we determined the kinetic parameters (K _m/K _1/2 and k _cat) for each protein with all confirmed noncanonical substrates (Figure 7 and Table 2). Consistently, RNMT showed higher k _cat (218 ± 12 h⁻¹) with GpppAmC as substrate than its known substrate GpppAC (91 ± 1 h⁻¹), with 7.6 times higher catalytic efficiency (k _cat/K _m), indicating that RNMT methylates the mRNA even more efficiently at N⁷‐guanosine if it has 2′‐O‐ribose methylation, a CMTR1 mark. Potentially, this could be a layer of re‐capped mRNA protection in the cytoplasm (Figure 7 and Table 2). A similar but less pronounced pattern was also observed for coronaviral nsp14. Interestingly, CMTR1 was equally active with similar catalytic efficiencies with GpppAC, mGpppAC, and mGpppACm as substrates (Figure 7 and Table 2). In the two CMTR1 inactive substrates, GpppAmC and mGpppAmC, the first transcribed nucleotide (A, adenosine) was methylated already. This further confirmed the first transcribed nucleotide specific 2′‐O‐ribose methylation of mRNA by CMTR1 independent of any other cap methylations. CMTR2 showed some low level of activity with all RNA substrates with unmethylated second transcribed nucleotide (C, cytosine), indicating specific methylation of the second transcribed nucleotide (Figure 7 and Table 2). Overall, the conclusions drawn from the data in Figure 4 were consistently confirmed.

FIGURE 7.

Kinetic characterization of noncanonical substrates of RNMT, CMTR1, CMTR2, nsp14, and nsp16. Initial velocities were determined by varying RNA concentrations at fixed SAM concentrations, a(i), b(i), c(i), d(i), e(i), f(i), and data were plotted and fit to the Michaelis–Menten equation for nsp14 a(ii) and the sigmoidal equation for the rest b(ii), c(ii), d(ii), e(ii), f(ii), as described in Data S1. The K _1/2/K _m and k _cat values are presented in Table 2. All experiments were performed in triplicate, and error bars represent the standard deviation.

4. DISCUSSION

Methylation of N⁷‐guanosine by RNMT (Liang et al. 2023; Nencka et al. ²⁰²²) and nsp14 (Bouvet et al. 2010; Devkota et al. ²⁰²¹), 2′‐O‐ribose of the first transcribed nucleotide by CMTR1 (Belanger et al. 2010; Inesta‐Vaquera and Cowling ²⁰¹⁷; Smietanski et al. ²⁰¹⁴) and nsp16 (Benoni et al. 2021; Bouvet et al. ²⁰¹⁰), and the second nucleotide by CMTR2 (Smietanski et al. 2014; Werner et al. ²⁰¹¹) have been previously reported (Nencka et al. 2022). The presence of the two RNA 2′‐O‐ribose methyltransferases catalyzing the cap‐1 and cap‐2 formation mainly in the nucleus and cytoplasm, respectively, was reported by Langberg and Moss (Langberg and Moss 1981). Belanger et al. identified and named the protein catalyzing the specific methylation of the 2′‐O‐ribose of the first transcribed nucleotide of a capped RNA transcript as hMTr1 (Belanger et al. 2010), which was previously called FTSJD2 and ISG95 (Haline‐Vaz et al. ²⁰⁰⁸). Using a combination of ³²P‐labelling of the cap, nuclease treatment, and thin layer chromatography (TLC), hMTr1 (human CMTR1) substrate specificity was evaluated. It was shown to methylate both GpppG and N ⁷‐methylated GpppG substrates (Belanger et al. 2010). The human protein encoded by FTSJD1 was also identified for RNA 2′‐O‐ribose methylation of the second transcribed nucleotide and was named HMTR2 (CMTR2) by Werner et al. (Werner et al. 2011). A combination of ³²P‐labelling of the cap, nuclease treatment, polyacrylamide gel electrophoresis, and TLC similar to the method previously reported for CMTR1 was used to characterize CMTR2 substrate specificity (Werner et al. 2011). It was later shown that CMTR2 is present in both the cytosol and the nucleus, and its function does not require prior methylation of the guanosine cap, and shows no preference for the presence of 2′‐O‐ribose methylation of the first transcribed nucleotide. However, it is less efficient for RNA without cap‐1 (Werner et al. 2011).

Our quantitative assessment of the substrate specificity of human RNA capping methyltransferases by direct measurement of product formation by radiometric assays, kinetic parameter determination combined with mass spectrometry analysis clearly revealed that RNA methyltransferase activities of RNMT, CMTR1, and CMTR2 are nonoverlapping, and each monomethylates only at a specific nucleotide position within the cap and could also function selectively and independently.

Our kinetic data also provide a quantitative assessment, which is consistent with previous qualitative reports (Belanger et al. 2010; Werner et al. ²⁰¹¹). RNMT and nsp14 only methylated N⁷‐guanosine, CMTR1 and nsp16 only methylated 2′‐O‐ribose of the first transcribed nucleotide, and CMTR2 methylated 2′‐O‐ribose of the second transcribed nucleotide. These observations are also consistent with independent RNMT and CMTR1 roles in ES cell differentiation (Grasso et al. 2016; Liang et al. ²⁰²²; Liang et al. ²⁰²³). Surprisingly, we also observed specific patterns of cooperativity such as RNA methylation by CMTR1 increasing the activity of RNMT (Table 2 and Figure 8). CMTR1 functions independently and does not need prior N⁷‐guanosine methylation by RNMT. Such substrate specificity and positive cooperativity combined with predominant nuclear localization of RNMT and CMTR1 (Liang et al. 2023; Shafer et al. ²⁰⁰⁵) further suggest that the levels of activities of these two RNA methyltransferases might have a significant regulatory role, as they both function co‐transcriptionally. In the absence of an experimental structure of human RNMT with a long RNA substrate or an RNA with a methylated first nucleotide but unmethylated guanosine, we can only speculate on a possible molecular mechanism for this increase in catalytic efficiency. Using the structure of RNMT with bound SAM analogue (sinefungin) and mN⁷Gppp (PDB ID: 8Q9W), we created a model where we extended the RNA chain with a 2′‐O‐methylated A base, followed by a U base (Figure 9a). In this model, the two transcribed bases AU are solvent exposed due to their open, putative binding site. This binding site is also amphiphilic. The polar, electropositive portion of this site, made of several positively charged residues such as K208, R239, R246, and R173, would be attractive to the transcribed nucleotides and their backbone phosphate groups. Further, above this site is a hydrophobic patch, or “roof,” that can engage in van der Waals interactions with a 2′‐O‐methylated nucleotide base (Figure 9). Thus, 2′‐O‐methylation of the RNA substrate that enables hydrophobic interactions could only enhance the existing charge–charge interactions between protein and RNA. This likely contributes to the observed 7.6‐fold increase in catalytic efficiency, although other unidentified factors could be at play. A crystal structure would be necessary to confirm this hypothesis. Our observation of such cooperativities between RNMT and CMTR1 could help better understand their roles in the regulation of gene expression, cell growth, and differentiation.

FIGURE 8.

Comparison of the turnover numbers and catalytic efficiency values of each RNA methyltransferase with its active substrates. (a) The k _cat values (turnover numbers) for each RNA methyltransferase with its active RNA substrates from Table 2 are represented, and the differences are statistically evaluated. The turnover number is the number of substrate molecules converted to product by a single enzyme molecule per unit of time at saturation of substrates (Copeland 2000). (b) Similarly, the catalytic efficiency (k _cat/K _m) values from Table 2 are represented, and the differences are statistically evaluated. Catalytic efficiency of an enzyme is a measure of how efficiently an enzyme converts a substrate to product (Copeland 2000). Note that catalytic efficiency values are plotted in units of “h⁻¹ μM⁻¹” for a better presentation than “h⁻¹ nM⁻¹” as in Table 2. RNMT shows a significantly higher k _cat (a) and catalytic efficiency (b) values with GpppAmC as a substrate than GpppAC (b). Although a similar pattern was observed for nsp14 with the same substrates, the increase in catalytic efficiencies was not statistically significant (b). CMTR1, with some minor differences in k _cat values (a), methylates GpppAC, mGpppAC, and mGpppACm efficiently with no significant differences (a, b). However, nsp16 is not active with GpppAC, has much lower k _cat with mGpppACm as a substrate (a), but its catalytic efficiencies with mGpppAC and mGpppACm are not significantly different (b). Error bars represent the standard deviation of three independent replicates. Statistical significance was assessed using unpaired t tests. Asterisks indicate significance levels as follows: p < 0.05 (*), p < 0.01 (**), p < 0.001 (***), and p < 0.0001 (****). ns: not significant.

FIGURE 9.

Structural models provide potential mechanism for increase in activity of RNMT by CMTR1 RNA methylation. (a) Model of human RNMT (green cartoon) bound to a SAM analogue (pink sticks) and mN⁷GpppAmU (orange sticks). The 2′‐O‐methyl group is indicated by an arrow. Broken black lines indicate hydrogen bonds/salt bridges between RNMT and RNA substrate. (b) Electrostatic potential of the human RNMT's RNA binding site. The protein orientation is the same as in (a). The binding site for the two outer, transcribed nucleotides is amphiphilic. Its blue surface indicates electropositive potential ideal for binding backbone phosphate groups and nucleotide bases. A white surface (inside broken circles) indicates hydrophobic patches, which could be stabilizing for methylated bases. The 2′‐O‐methyl group is indicated by an arrow and could be proximal to this hydrophobic patch in solution. The model is based on the published structure of human RNMT (PDB 8Q9W) (Pearson et al. 2025).

On the other hand, CMTR2 is fully active when N⁷G and 2′‐O‐ribose A are both methylated. Unexpectedly, CMTR2 is still active when only the 2′‐O‐ribose of the first transcribed nucleotide is methylated (Table 2 and Figure 8). This raises the possibility that CMTR2 potentially could add an extra level of RNA methylation to de‐capped human RNA in the cytoplasm even when N⁷G methylation is not present. These results further support the possibility of multi‐layers of regulation affecting mRNA processing in cells during translation.

On the contrary, coronaviral nsp14 and nsp16 function only in a sequential manner, with RNA methylation by nsp14 being essential for nsp10–nsp16 complex function, consistent with previous reports (Khalili Yazdi et al. 2021). Therefore, nsp14 appears to be a more critical target for the development of antiviral therapeutics. However, potentially dual inhibitors of nsp14 and nsp16 could be developed too (Devkota et al. 2021).

Structures of human (Pearson et al. 2025) and SARS‐CoV‐2 (Czarna et al. 2022; Imprachim et al. ²⁰²³; Kottur et al. ²⁰²²) methyltransferases support substrate specificities we observed (Figure 2). RNMT and nsp14 are N⁷G methyltransferases whose active sites can only accommodate SAM and the Gppp cap while leaving the rest of the RNA substrate, including the first transcribed nucleotides solvent exposed (Figure 2a,b). Their Gppp binding sites have well‐defined borders that limit the movement of RNA substrate, preventing methylation of downstream nucleotides. Given the tight space between the N⁷ atom of Gppp and SAM, RNMT and nsp14 are inactive towards pre‐methylated N⁷G RNA substrates, while methylation at downstream nucleotides is expected to have no effect on their enzyme activity. This is consistent with our observations that both proteins only methylate N⁷G. Interestingly, pre‐methylation of the first transcribed nucleotide (GpppAmC) significantly increased the N⁷G methylation by both RNMT and nsp14 (Figure 4 and Table 2). This effect will only be important for RNMT function as CMTR1 is able to methylate 2′‐O‐ribose of the first nucleotide in the absence of N⁷G methylation, but not the nsp10–nsp16 complex.

Human CMTR1 and SARS‐CoV‐2 nsp16 methylate the 2′‐O‐ribose of the first transcribed nucleotide of RNA substrates. Both have larger active sites than nsp14 and RNMT that bury SAM and the GpppN1 portion of the RNA substrates while leaving downstream nucleotides solvent exposed (Figure 2c,d). SAM is precisely positioned within a tight space to methylate the first transcribed nucleotide 2′‐O‐ribose such that pre‐methylation leads to steric clash with the methyl donor. In the crystal structures of both enzymes (Inniss et al. 2023; Kalnins et al. ²⁰²⁴; Minasov et al. ²⁰²¹; Smietanski et al. ²⁰¹⁴), the second transcribed nucleotide is more solvent exposed than the GpppN1 portion of substrate RNA, bordered by a nearby loop that can accommodate a 2′‐methoxy group on the ribose of the second transcribed nucleotide. In nsp16, the loop structure formed by residues 6868–6877 varies slightly in crystal structures, with the side chain of D6873 assuming a rotamer that makes space for the second nucleotide ribose 2′‐methoxy group (Figure 2d). The corresponding residue in CMTR1 is P280, which, although more conformationally restricted than an aspartate side chain, is smaller, leaving the second nucleotide ribose 2′‐methoxy group more solvent exposed (Figure 2c). Thus, both nsp16 and CMTR1 display significant activity towards RNA substrates with pre‐methylation of the second transcribed nucleotide ribose 2′‐OH groups (Figure 4 and Table 2).

Interestingly, nsp16 appears to be dependent on N⁷G pre‐methylation for activity but CMTR1 does not, as the latter remains active towards the unmethylated RNA substrate (GpppAC). In the crystal structure of CMTR1, the methylated N⁷G cap of RNA substrate fits snugly in the binding pocket, with the N⁷ methyl group facing the solvent space. Moreover, the mN⁷G cap is held in place by four electrostatic bonds with the base aromatic ring, encompassing its Watson‐Crick, Hoogsteen, and sugar edge, and a hydrophobic stacking interaction with the side chain of N439 (Figure 2c). Thus, N⁷G methylation might be less important to binding affinity for CMTR1 in the presence of these stabilizing interactions. On the other hand, in the nsp16 crystal structure, the aromatic base of mN⁷G is flipped 180° relative to that in CMTR1, with the Hoogsteen edge facing the hydrophobic interior of the pocket (Figure 2d). This orientation positions the N⁷ methyl group in a complementary binding pocket while also placing the base's aromatic ring in an optimal position for hydrophobic stacking with the side chain of Y6828. Stabilizing this configuration are only two electrostatic interactions with the peptide amino group of L6825 and the peptide carbonyl group of C6823. Thus, in nsp16, hydrophobic and van der Waals interactions that arise from shape complementarity dominate the enzyme's interaction with mN⁷G and underscore the role of the mN⁷G in binding stabilization. Curiously, the thiol group of C6823 in nsp16 and M214 in CMTR1 interacts directly with the N⁷G methyl group of substrate RNA, with an interaction distance of 3.6 Å between thiol S and N⁷G methyl C atoms. Cys and Met residues are known to stabilize aromatic groups in proteins (Orabi and English 2016). Met‐aromatic motifs have been shown to be necessary in high affinity ligand binding, and compared to purely hydrophobic interactions, yield 1.0–1.5 kcal/mol of additional stabilization (Valley et al. 2012). Further stabilization of mN⁷G binding in nsp16 and CMTR1 is presumably achieved with base methylation.

In the absence of an experimental structure, we also used a previous structural model of the CMTR2 methyltransferase domain in complex with SAM and substrate RNA (mN⁷GpppGGAA) to explicate its activity towards differently methylated RNA substrates (Smietanski et al. 2014). This model is essentially similar to the AlphaFold‐generated CMTR2 structure and superimposes with the experimental CMTR1 structure with an r.m.s.d. of 3.9 Å (Figure 2f). As previously described (Smietanski et al. 2014), the putative mN⁷G and SAM binding sites of the CMTR2 are conserved and align well with those of CMTR1, while the vicinity surrounding the proposed N1 binding site is relatively less conserved. Based on the location of SAM and mN⁷G in this model, methylation of the N2 ribose 2′‐OH group might require the RNA register to shift and result in a non‐canonical, cis‐Hoogsteen‐Hoogsteen base pair between N1(G) and N3(A) (Figure 2e). It is apparent that methylation of the second transcribed nucleotide ribose 2′‐OH would result in a steric clash with SAM, leading to an inactive RNA substrate. N⁷G methylation could also increase substrate affinity due to binding pocket complementarity, with the solvent‐facing N⁷G methyl group occupying a small pocket bounded by L82. Nevertheless, its positioning towards the solvent space could partly explain why it is dispensable for activity in our enzyme assays. On the other hand, methylation of the first transcribed nucleotide ribose 2′‐OH group appears to be important for activity. It is possible that the first transcribed nucleotide 2′‐methoxy group might contribute to substrate binding by stabilizing the bound RNA conformer through electrostatic interactions as a hydrogen bond acceptor to CMTR1 residue(s) and/or bridging solvent molecules, or with hydrogen bond donors within the RNA substrate itself. In the proposed model, a first transcribed nucleotide ribose 2′‐methoxy group could fill a space underneath the cis‐Hoogsteen‐Hoogsteen base pair, yielding a stabilization effect similar to those arising from base modifications in alternative RNA and DNA structures. Overall, structural models support the observed activities of the five human and SARS‐CoV‐2 RNA methyltransferases towards the differently methylated substrates used in this study.

Regulation of RNMT and CMTR1 expression has been reported to affect ES cell differentiation with low levels of RNMT and high concentrations of CMTR1 towards ES cell differentiation (Grasso et al. 2016; Liang et al. ²⁰²²; Liang et al. ²⁰²³). The impact of N⁷‐guanosine methylation and 2′‐O‐ribose methylation is gene‐specific (Liang et al. 2023). Unlike CMTR1, which interacts with RNA pol II directly through its WW domain, RNMT has no WW domain (Liang et al. 2023). These observations suggest possible distinct roles for these two human RNA methyltransferases which are nuclear and function co‐transcriptionally (Shafer et al. 2005). RNMT plays a significant role in tumorigenesis and various cancers including glioma (Chen et al. 2020). The regulation of RNMT has been reported to correlate with enhanced expression of specific genes. For example, RNMT is up‐regulated during T‐cell activation when rapid cell growth and proliferation require increased mRNA synthesis for adaptive immune responses (Galloway et al. 2021). RNMT and CMTR1 can be described as the major cap methyltransferases because the majority of mature mRNA in ES cells and other mammalian cells carries both methylations catalyzed by these two enzymes (Galloway et al. 2020; Wang et al. ²⁰¹⁹).

5. CONCLUSION

Our data confirm that human RNMT, CMTR1, and CMTR2 only monomethylate N⁷G, 2′‐O‐ribose of the first and the second transcribed nucleotides, respectively, with no overlap in their methyltransferase activities. All three enzymes also could methylate their respective nucleotide residues independently, and their functions clearly do not need to be sequential. Interestingly, the level of enzymatic activity of each of the three is affected by methylation by others. RNMT and CMTR1 are highly active independently, which could explain why they could have opposite roles in ES cell differentiation or other cellular functions. However, RNA methylation by CMTR1 increases the activity of RNMT, a novel finding that suggests a potentially new mechanism of action for this cooperative function. Future studies need to define such roles in cells. On the other hand, the coronaviral nsp10–nsp16 complex requires methylation of N⁷G to be active, and therefore, the cap methylation for coronavirus is sequential. The cap‐1 viral RNA could potentially be further methylated by human CMTR2. However, the normal concentration of CMTR2 may not be high enough to keep up with fast‐replicating viruses.

AUTHOR CONTRIBUTIONS

Fatemeh Taherian: Investigation; visualization; data curation; formal analysis; validation; writing – original draft; writing – review and editing. Mark F. Mabanglo: Formal analysis; visualization; writing – review and editing. Taraneh Hajian: Investigation; writing – review and editing. Sarah Tucker: Investigation; writing – review and editing. Emilija Kalinic: Investigation; writing – review and editing. Sumera Perveen: Investigation; writing – review and editing. Ahmed Aman: Investigation; data curation; formal analysis; writing – original draft; visualization; writing – review and editing. Masoud Vedadi: Conceptualization; supervision; funding acquisition; formal analysis; writing – original draft; writing – review and editing.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

Supporting information

Data S1. Supporting Information.

Taherian F, Mabanglo MF, Hajian T, Tucker S, Kalinic E, Perveen S, et al. Defining substrate specificities of human RNA capping methyltransferases through quantitative assessment of independent yet cooperative activities. Protein Science. 2026;35(1):e70428. 10.1002/pro.70428

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.