2'-O-Methylation of the second transcribed nucleotide within the mRNA 5' cap impacts the protein production level in a cell-specific manner and contributes to RNA immune evasion

Abstract

In mammals, m7G-adjacent nucleotides undergo extensive modifications. Ribose of the first or first and second transcribed nucleotides can be subjected to 2'-O-methylation to form cap1 or cap2, respectively. When the first transcribed nucleotide is 2'-O-methylated adenosine, it can be additionally modified to N6,2'-O-dimethyladenosine (m6Am). Recently, the crucial role of cap1 in distinguishing between 'self' and 'non-self' in mammalian cells during viral infection was revealed. Here, we attempted to understand the impact of cap methylations on RNA-related processes. Therefore, we synthesized tetranucleotide cap analogues and used them for RNA capping during in vitro transcription. Using this tool, we found that 2'-O-methylation of the second transcribed nucleotide within the mRNA 5' cap influences protein production levels in a cell-specific manner. This modification can strongly hamper protein biosynthesis or have no influence on protein production levels, depending on the cell line. Interestingly, 2'-O-methylation of the second transcribed nucleotide and the presence of m6Am as the first transcribed nucleotide serve as determinants that define transcripts as 'self' and contribute to transcript escape from the host innate immune response. Additionally, cap methylation status does not influence transcript affinity towards translation initiation factor eIF4E or in vitro susceptibility to decapping by DCP2; however, we observe the resistance of cap2-RNA to DXO (decapping exoribonuclease)-mediated decapping and degradation.

Figure 1.

Tetranucleotide cap analogues act as initiators of in vitro transcription reactions. (A) Structure of the cap analogues used in this study; newly synthesized tetranucleotide and previously obtained trinucleotide (14) cap analogues are presented on the left and right side, respectively. (B) Comparison of the major transcription initiation events during the in vitro transcription reaction when either no cap analogue, a trinucleotide cap analogue or a tetranucleotide cap analogue was used as an initiator. Capped RNA obtained in the in vitro transcription reaction with a tri- or tetranucleotide cap analogue is 27 nt long, as uncapped RNA is 25 nt long. (C) Analysis of short RNAs obtained by in vitro transcription using T7 RNA polymerase in the presence of different cap analogues (DNAzyme-trimmed and HPLC-purified transcripts). The capping efficiency values (percentage) determined by densitometric quantification of the major bands corresponding to capped and uncapped RNA are shown at the top of the gel. Minor extra bands most probably arise from unspecific addition of nucleotides during in vitro transcription.

Figure 2.

Protein production levels are affected by cap methylation status. Relative total protein production after 72 h measured in the medium from culture of (A and B) A549 and (C and D) JAWS II cells transfected with IVT mRNAs encoding Gaussia luciferase bearing various cap analogues at their 5′ ends. Bars represent the mean value ± SEM normalized to transcripts with cap0(A). Statistical significance: NS, not significant; *P <0.05; **P <0.01; ***P <0.001; ****P <0.0001 (one-way ANOVA with Turkey's multiple comparisons test). (Raw data from three independent biological replicates are shown in Supplementary Figure S2; each independent biological replicate consisted of three independent transfections.) (E) Summary table presenting the influence of the cap methylation status on protein production levels in A549, JAWS II, THP-1 and 3T3-L1 cells relative to transcript with cap0(A), ↑, increase of protein production; ↓, decrease of protein production; double arrows indicate a large (at least 3-fold) increase/decrease; ↔, no significant change.

Figure 3.

2′-O-Methylation within the cap structure does not influence RNA affinity for the translational machinery. (A) Relative affinities of transcripts bearing different cap analogues for murine eIF4E determined using MST. Bars represent the mean value ± SEM from three independent replicates. Statistical significance: **P <0.01 (one-way ANOVA with Turkey's multiple comparisons test). (Representative MST curves and competitive binding curves obtained in the experiment are presented in Supplementary Figure S8.) (B) Comparison of apparent binding constant values K_D,app for capped RNAs in complexes with murine eIF4E measured with MST with dissociation constants of eIF4E–cap complexes obtained using time-synchronized fluorescence quenching titration (ts-FQT) (14). For all RNAs, three independent replicates were performed, besides cap2(A) and cap2-1(A) RNAs, for which data from two replicates are presented. (C) Competition of differently capped IVT mRNAs encoding Gaussia luciferase with endogenous mRNAs for the translation machinery. HEK 293 Flp-In T-REx cells expressing shmiRs targeting CMTR1, CMTR2 or both (a negative control was utilized in parallel) were transfected with IVT mRNAs, and medium was collected after 72 h for luciferase activity analysis. Bars represent the mean value ± SEM normalized to Gaussia luciferase activity measured for transcripts with cap0(A). Data for three independent experiments are presented (each biological replicate consisted of three transfections).

Figure 4.

2′-O-Methylation of the second transcribed nucleotide prevents RNA from decapping by DXO but not by DCP2. (A–C) Short capped RNAs were subjected to treatment with hDCP2 (wild type or mutant) over a 60 min time-course. Reactions without enzyme served as controls. Aliquots from the indicated time points were resolved on a polyacrylamide gel and bands corresponding to capped transcripts (27 nt long) and to RNAs decapped by hDCP2 action (26 nt long) were quantified densitometrically. (A) Representative polyacrylamide gel analyses obtained for all tested capped RNAs (two additional repetitions with wild-type hDCP2 of this experiment are shown in Supplementary Figure S12). (B and C) Quantitative results for all studied RNAs. The fraction of capped RNA remaining in the total RNA was plotted as a function of time. Data points represent mean values ± SD from triplicate experiments. (D) Short capped RNAs were subjected to treatment with hDXO (wild type or mutant) over a 60 min time-course. Reactions without enzyme served as controls. Aliquots from the indicated time points were resolved on a polyacrylamide gel. (E) Experimental set-up as in (D); however, a 2.5-fold higher hDXO concentration was used.

Figure 5.

Methylation of the first transcribed nucleotides in mRNA only partially prevents protein level decrease under IFN-induced stress. Relative protein production levels 72 h after a 5 h IFNα pre-treatment in (A and B) A549 and (C and D) JAWS II cells. Data were analysed for three independent replicates (each experiment consisted of three technical replicates). Bars for each transcript represent the mean value ± SEM normalized to protein production in mock-treated cells. Statistical significance: *P <0.05, **P <0.01, ***P <0.001, ****P <0.0001 (one-way ANOVA with Turkey's multiple comparisons test) (data presenting changes in relative protein production with increasing concentration of IFNα are shown in Supplementary Figures S15 and S16 for A549 and JAWS II cells, respectively). Co-purification of endogenous proteins from lysates of IFNα-treated (E and F) A549 (IFNα concentration was 500 U/ml) and (G and H) JAWS II (IFNα concentration was 5000 U/ml) cells with biotinylated RNA bearing cap0, cap1, cap2 or cap2-1 with (E and G) A or (F and H) m⁶A as the first transcribed nucleotide. Human eIF4E, IFIT1 and NCBP1 or murine counterparts (Ifit1b: lower band) were detected in precipitates by western blotting. A 1/15th and 1/21st volume of lysate used for each incubation with beads was loaded as input sample for (E–G) and (H), respectively; a 1/3rd volume of each eluate was loaded; shorter exposure is presented for input samples than for eluates; *non-specific band.