Trinucleotide mRNA Cap Analogue N 6-Benzylated at the Site of Posttranscriptional m6Am Mark Facilitates mRNA Purification and Confers Superior Translational Properties In Vitro and In Vivo

Abstract

Eukaryotic mRNAs undergo cotranscriptional 5'-end modification with a 7-methylguanosine cap. In higher eukaryotes, the cap carries additional methylations, such as ^m6A_m─a common epitranscriptomic mark unique to the mRNA 5'-end. This modification is regulated by the Pcif1 methyltransferase and the FTO demethylase, but its biological function is still unknown. Here, we designed and synthesized a trinucleotide FTO-resistant N6-benzyl analogue of the ^m6A_m-cap-m⁷Gppp^Bn6A_mpG (termed AvantCap) and incorporated it into mRNA using T7 polymerase. mRNAs carrying ^Bn6A_m showed several advantages over typical capped transcripts. The ^Bn6A_m moiety was shown to act as a reversed-phase high-performance liquid chromatography (RP-HPLC) purification handle, allowing the separation of capped and uncapped RNA species, and to produce transcripts with lower dsRNA content than reference caps. In some cultured cells, ^Bn6A_m mRNAs provided higher protein yields than mRNAs carrying A_m or ^m6A_m, although the effect was cell-line-dependent. m⁷Gppp^Bn6A_mpG-capped mRNAs encoding reporter proteins administered intravenously to mice provided up to 6-fold higher protein outputs than reference mRNAs, while mRNAs encoding tumor antigens showed superior activity in therapeutic settings as anticancer vaccines. The biochemical characterization suggests several phenomena potentially underlying the biological properties of AvantCap: (i) reduced propensity for unspecific interactions, (ii) involvement in alternative translation initiation, and (iii) subtle differences in mRNA impurity profiles or a combination of these effects. AvantCapped-mRNAs bearing the ^Bn6A_m may pave the way for more potent mRNA-based vaccines and therapeutics and serve as molecular tools to unravel the role of ^m6A_m in mRNA.

Figure 1

mRNA 5′ cap structure. (A) Natural variants of cap structures carrying adenosine adjacent to the 5′ cap; (B) dynamic regulation of ^m6A_m presence in the cell; and (C) structure of m⁷Gppp^Bn6A_mpG (AvantCap).

Scheme 1. Chemical Synthesis of m⁷Gppp^Bn6A_mpG (1)

Reaction conditions: (i) benzyl bromide, tetrabutylammonium bromide, 1 M NaOH_aq, CH₂Cl₂; (ii) imidazole, 2,2′-dithiodipyridine, triphenylphosphine, triethylamine, DMF; and (iii) N7-methylguanosine 5′-diphosphate (m⁷GDP), ZnCl₂, DMSO.

Figure 2

m⁷Gppp^Bn6A_mpG initiates transcription from the T7 class III promoter (φ6.5) and GGG transcription start site. (A) Possible scenarios for the initiation of transcription occurring in the presence of *AG-type trinucleotides (such as m⁷GpppApG) and the studied DNA template; *A denotes N6-modified adenosine residue. (B) Capping efficiencies for short RNAs determined by gel electrophoresis. IVT reactions were performed in the presence of 1.25 μM template, 3 mM ATP, CTP, UTP, and 0.75 mM GTP, 6 mM cap analogue, at pH 7.9 (for details, see the Experimental Section). (C) Capping efficiencies (determined by two methods) and IVT yields (determined spectrophotometrically after initial purification) as a function of AvantCap concentration and pH. IVT reactions were performed in the presence of 25 mM MgCl₂, 40 ng/μL template, 5 mM ATP, CTP, UTP, 4 mM GTP, and 10 mM cap analogue (for optimizing pH) or various cap concentrations at pH 6.5.

Figure 3

N6-Benzyl-2′-O-methyladenosine (^Bn6A_m) within the 5′ cap acts as an mRNA purification handle. (A) RP-HPLC analysis of Gaussia luciferase (Gluc) mRNA (956 nt) obtained by IVT carrying various 5′ terminal structures: uncapped mRNA (gray), m⁷Gppp^m6A_mpG-capped mRNA (green), m⁷Gppp^Bn6A_mpG-capped mRNA (blue), and 1:1 mixture of m⁷Gppp^Bn6A_mpG-capped mRNA and pppG-mRNA (orange), revealing that the presence of ^Bn6A_m delays mRNA retention and enables the separation of capped mRNA (R_t = 19.4 min) and uncapped mRNA (R_t = 18.2 min). RP-HPLC conditions are given in the Experimental section. (B) RP-HPLC analyses of ∼1:1 mixtures of m⁷Gppp^Bn6A_mpG-capped transcripts of different lengths (1000, 1500, 2000, and 6000 nt) and corresponding uncapped mRNAs. RP-HPLC conditions are given in the Experimental Section.

Figure 4

N6-Benzyladenosine within the mRNA 5′ cap increases protein production in certain mammalian cells. (A) Firefly luciferase (Fluc)-dependent luminescence of murine colon carcinoma (CT26, 10⁴ cells per well), human embryonic kidney (HEK293T, 10⁴ cells per well) and human lung carcinoma (A549, 10 ⁴ cells per well) cells 6 h post transfection with 50 ng of Fluc mRNA; data show relative luminescence units (RLUs) means ± SD, n = 3, *P < 0.05, ****P < 0.001, ns—not significant, one-way ANOVA with Tukey’s multiple comparison test. (B) Human erythropoietin (hEPO) concentration in the culture medium of primary murine bone marrow (BM)-derived macrophages (mMΦ, 5 × 10⁴ cells per well), murine BM-derived dendritic cells (mDC, 5 × 10⁴ cells), and human embryonic kidney (HEK293T, 10⁴ cells per well) cells 24 h post transfection with 200 ng of hEPO mRNA; data show hEPO concentrations (ng/mL) with means ± SD, n = 3, **P < 0.01, ns—not significant, two-tailed unpaired t test. (C) Gaussia luciferase (Gluc)-dependent luminescence of human monocyte-derived dendritic cells (hDCs, 5 × 10⁴ cells per well) differentiated from monocytes of three healthy adult males, transfected with 5, 25, or 100 ng of Gluc mRNA; data show total protein production (sum of relative luminescence units [RLUs] in daily measures from day 1 until 6 post transfection) with means ± SD, n = 3, * P < 0.05, **P < 0.01, ns—not significant, two-tailed unpaired t test.

Figure 5

mRNA capped with m⁷Gppp^Bn6A_mpG (AvantCap) yields superior protein production in vivo. (A) Intravital bioluminescence at 4, 8, and 24 h in BALB/c mice injected intravenously (i.v.) with firefly luciferase (Fluc)-encoding mRNAs, capped with m⁷GpppA_mpG or m⁷Gppp^Bn6A_mpG, and formulated using different ionizable lipids (GenVoy-ILM, SM-102, and MC3). Flux [p/s] mean values ± SD, n = 5, GenVoy-ILM and MC3 data: *P < 0.05, **P < 0.01, multiple unpaired t test; SM-102 data: Mann–Whitney test, q values [false discovery rate (FDR)-adjusted P-values] are shown. Note the logarithmic scale on the y-axis. (B) Raw bioluminescence images of mice inoculated with Fluc encoding mRNA (scale for bioluminescent signals at the right). (C) Human erythropoietin (hEPO) serum concentrations 4, 8, 24, and 48 h in C57BL/6 mice injected i.v. with hEPO encoding mRNAs, capped with m⁷GpppA_mpG or m⁷Gppp^Bn6A_mpG, and formulated using different ionizable lipids (GenVoy-ILM, SM-102, and MC3). Data show mean values ± SD, n = 5, GenVoy-ILM and SM-102 data: Mann–Whitney test, q values (FDR-adjusted P-values) are shown; MC3 data: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 multiple unpaired t test.

Figure 6

mRNA capped with m⁷Gppp^Bn6A_mpG (AvantCap) shows therapeutic activity in cancer models. (A) Quantitative OT-I T-cell proliferation in response to in vivo delivery of SIINFEKL antigen/peptide-encoding m⁷GpppA_mpG or m⁷Gppp^Bn6A_mpG-capped mRNA. Mice were administered intravenously (i.v.) with 7.5 ng of mRNA in TransIT formulation. T-cell numbers were normalized to CountBright Absolute Counting Beads (ThermoFisher Scientific). Data show OT-I T-cell cell numbers in the spleen after 72 h of proliferation in vivo; data show means ± SD, n = 5, two-tailed unpaired t test. (B) C57BL/6 mice were inoculated with Lewis lung carcinoma (LLC) cells stably expressing a model antigen–ovalbumin (OVA) and weekly treated i.v. with 100 ng of OVA-encoding m⁷GpppA_mpG or m⁷Gppp^Bn6A_mpG-capped mRNA formulated in TransIT. Tumor volumes are shown as means ± SD, n = 8. Mixed-effect analysis with Dunnett’s multiple comparison test, m⁷Gppp^Bn6A_mpG-capped mRNA vs control: day 16: P = 0.0060; day 18: P = 0.0410; day 21: P = 0.0168; day 23: P = 0.0080. Control mice received PBS. m⁷Gppp^Bn6A_mpG-capped mRNA encoding human hEPO served as a tumor antigen-irrelevant control. TGI (tumor growth inhibition compared to controls). (C) BALB/c mice were inoculated with CT26 murine adenocarcinoma (CT26) cells stably expressing a human tumor antigen (NY-ESO1) and weekly treated i.v. with 100 ng of NY-ESO1-encoding m⁷GpppA_mpG or m⁷Gppp^Bn6A_mpG-capped mRNA formulated in TransIT. Tumor volumes are shown as means ± SD, n = 8. 2-way ANOVA with Dunnett’s multiple comparison test, m⁷Gppp^Bn6A_mpG-capped mRNA vs control: day 18: P = 0.0079; day 22: P = 0.0360; day 25: P = 0.0046. Control mice received PBS. TGI—tumor growth inhibition compared to controls.

Figure 7

Cap structure containing ^Bn6A_m is resistant to removal by FTO. ^Bn6A_m- or ^m6A_m-containing cap analogues were incubated with FTO, and the amount of N6-alkylated cap remaining in the mixture was assessed by RP-HPLC-MS. Reaction conditions: 20 μM cap and 2 μM FTO in 50 mM HEPES pH 7, containing 150 mM KCl, 75 μM Fe(II), 300 μM 2-oxoglutarate, 2 mM ascorbic acid. Similar data obtained for the 200 μM cap are shown in Figure S9.

Figure 8

(A) Binding affinities of AvantCap (1) and reference compounds for human translation initiation factor 4E (eIF4E) isoforms 1a and 3 and human 4E homologous protein (4EHP). (B) Relative translation efficiencies of mRNA capped with m⁷Gppp^Bn6A_mpG (1) and reference analogues in the rabbit reticulocyte lysate (RRL).

Figure 9

Susceptibility of short RNAs to decapping by the PNRC2-hDcp1/Dcp2 complex in vitro. Short 27-nt-capped RNAs (20 ng) were subjected to hDcp1/Dcp2 (13 nM) in complex with the regulatory peptide PNRC2 for 60 min at 37 °C. Aliquots from different time points were resolved by polyacrylamide gel electrophoresis (PAGE), stained with SYBR gold and analyzed by densitometry. (A) Representative PAGE gel from single experiment and (B) results from triplicate experiments ± SEM. Error bars are not visible if smaller than data points. Individual data for all replicates are shown in Figure S11.

Figure 10

Pull-down assay with protein extract from HEK293F cells. (A) The workflow of the experiment. (B) Chemical structure of trinucleotide affinity resins. (C) Dot plot representations of correlation between log2 fold change AR-3/AR-1 and log2 fold change AR-2/AR-1. (D) Volcano plots displaying the log2 fold change (log2 FC) against the t test-derived −log 10 statistical p-value (−log p-value) for the protein groups in the eluates from trinucleotide cap affinity resins AR-2/AR-1 (n = 3) and AR-2/AR-1 (n = 3). Student’s t tests (two-sided, unpaired) were performed to assess binding preferences of protein groups to AR-2 and AR-3 resins relative to AR-1 resin.