Dinucleoside polyphosphates act as 5’-RNA caps in Escherichia coli

Abstract

Dinucleoside polyphosphates (Np_nNs), discovered more than 50 years ago,¹ are pleiotropic molecules present in almost all types of cells.² It has been shown that their intracellular concentration can under stress conditions increase from the μM to mM range^2,3. However, the cellular roles and mechanisms of action of Np_nNs are still speculative^4,5. They have never been considered as part of the RNA, even though they have similar chemical structures as already known RNA caps, such as the nicotinamide adenine dinucleotide (NAD)^6–8 and 7-methylguanylate cap⁹. Here, we show that both methylated and non-methylated Np_nNs serve as RNA caps in Escherichia coli (E. coli). Np_nNs are excellent substrates for T7 and E. coli RNA polymerases (RNAP) and efficiently initiate transcription. Further, we demonstrate that the E. coli decapping enzyme RNA 5’ pyrophosphohydrolase (RppH) is able to remove the Np_nNs-cap from the RNA. RppH was, however, not able to cleave the methylated forms of the Np_nN-caps, suggesting that the methylation adds an additional layer to the RNA stability regulation. Our work introduces an original perspective on the chemical structure of RNA in prokaryotes and the function of RNA caps. This is the first evidence that small molecules like Np_nNs can act in cells via their incorporation into RNA and influence the cellular metabolism.

The role and chemical structure of the 5’-end of prokaryotic RNA is still unclear. The discovery of NAD^{6, 10} and Coenzyme A (CoA)¹¹ as 5’ RNA caps changed the perception of the RNA structure. 5’-caps are usually cleaved by NudiX enzymes (NudC^6,12,13, RppH^{14, 15}), which can, besides their decapping role (eukaryotic Nudt16 and Dcp2^{16, 17}), cleave nucleoside diphosphate linked to another moiety (e.g. Np_nNs^{2, 18}). To investigate whether Np_nNs (Figure 1a) can serve as non-canonical initiating nucleotides (NCINs) similarly to NAD and CoA¹⁹, we performed in vitro transcription in the presence of different Np_nNs (Ap_3-6A, Ap_4-5G, Gp₄G, Figure 1b) with T7 RNAP. The resulting RNA was a mixture of capped and uncapped RNA (step 1 in Figure 1b, Extended data Fig. 1a). The presence of the cap was confirmed by electrophoretic analysis after subsequent treatment with 5’-polyphosphatase and Terminator™ 5’-phosphate-dependent exonuclease (terminator, step 2 and 3 in Figure 1b). The former dephosphorylated the 5’-triphosphate RNA (5’-ppp RNA) but not any capped RNA. The terminator then digested all RNA with 5’-monophosphate termini (5’-p RNA) and left the capped RNA intact. We observed that all tested Np_nNs were excellent substrates for T7 RNAP and served as NCINs for in vitro transcription (Figure 1c).

Figure 1:

Np_nNs are excellent substrates for RNAP. a, The chemical structure of Np_nNs. b, Scheme showing the in vitro transcription using T7 RNAP in the presence of Np_nNs (1.6 mM) and template DNA with A φ 2.5 and G φ 6.5 promotors yielding 35mer RNA with A or G at the 5’-end. The first step resulted in a mixture of capped and uncapped RNAs, which is then treated by 5’– polyphosphatase (P) cleaving a diphosphate group from the 5’-ppp RNA leaving intact the capped RNA. In the third step the 5’-p RNA is degraded by a terminator exonuclease (T). Red (dark blue) rectangles correspond to purine base, and light blue spots depict phosphates c, Polyacrylamide gel electrophoretic (PAGE) analysis of the products of the in vitro transcription with T7 RNAP followed by P and/or T treatment. If not specified, samples were treated with both P and T enzymes. d, PAGE analysis of the products of in vitro transcription with E. coli RNAP and template plasmid 458 with promotor rrnB P1, which lead to the production of 144 nt long RNA having A as the first nucleotide, followed by P and/or T treatment. e, Percentage of different types of capped RNAs produced by in vitro transcription with T7 RNAP calculated from PAGE analysis. The depth axis represents various concentrations of Np_nNs (0.2, 0.4, 1 and 1.6 mM, different shades) at a constant concentration of ATP (1 mM) and GTP (1 mM). The left panel shows the percentage of Ap_3-6A and NAD capped RNA, the right panel shows the percentage of Ap_4-5G and Gp₄G.

Subsequently, we tested E. coli RNAP that is known to accept NAD as NCIN in vitro and in vivo^19–21 (Extended data Fig. 1b). To confirm the existence of 5’-capped RNA products, we also treated them with 5’-polyphosphatase and the terminator. We found that Np_nNs are superior initiating substrates compared to NAD. The absence of NAD-capping (Figure 1d) was likely due to the different promoter sequence than previously published¹⁹.

To identify the best substrate for RNAP, we varied the concentrations of the Np_nNs in the presence of a constant (1 mM) ATP and GTP concentration (Figure 1e, Extended data Fig. 1c,d). The amount of capped RNA increased linearly with the concentration of Np_nN. When the ratio of ATP (GTP) to Np_nN was 1, we observed between 27% (for Ap₆A) and 46% (for Ap₄G) capped products. The NAD-capped RNA was produced in lower amounts compared to the majority of Np_nNs.

Next, we wanted to determine whether Np_nNs exist as 5’-RNA caps in vivo in E. coli. We established an LC-MS method for their detection in RNA. Because the intracellular concentration of Np_nNs is known to grow under stress conditions^2,3,22, we collected cells in exponential (EXP, OD=0.3) and late stationary phase of growth (STA). We focused on short RNA (sRNA) where NAD-cap¹⁰ and CoA-cap¹¹ have also been detected. The RNA was washed to remove all non-covalently interacting molecules and digested by Nuclease P1 into the form of nucleotides (Figure 2a). The negative control samples, where the addition of Nuclease P1 was omitted, did not show any signals of nucleotides or Np_nNs, which excluded the possibility of non-covalently bound contamination.

Figure 2:

LC-MS detection of naturally occurring Np_nN-RNA in E. coli. a, Scheme showing the RNA preparation for comparative LC-MS measurements. RNA was isolated from E. coli in various stages of growth. Short RNA was separated from long RNA by the RNAzol protocol and analyzed by Tape station. sRNA was separated from non-covalently bound small molecules by size exclusion chromatography and divided into two parts. One part was treated by Nuclease P1, the other was treated under identical conditions without addition of Nuclease P1 as negative control. Both samples were subjected to size exclusion chromatography again and the fraction of small molecules was analyzed by LC-MS. b, Table of detected m/z values in LC-MS analysis of digested sRNA from E. coli harvested in the exponential phase (EXP), the late stationary phase (STA) and after growth in minimal media with the sole source of nitrogen from ¹⁴NH₄Cl (¹⁴N) or ¹⁵NH₄Cl (¹⁵N). c, Relative quantification of dimethylated-Gp₄G in RNA from EXP and STA growth of E. coli. d-f, Structures of different RNA caps and MS spectra of the detected m/z in RNA from E. coli growth in minimal media containing ¹⁴N or ¹⁵N of methyl-Ap₃A (d, in ¹⁴N m/z 769.085 was shifted to m/z 779.056 in ¹⁵N), Ap₃G (e, in ¹⁴N m/z 771.069 was shifted to m/z 781.035 in ¹⁵N), dimethyl-Gp₄G (f, in ¹⁴N m/z 447.021 was shifted to m/z 452.006 in ¹⁵N).

In all the digested sRNA, we observed signals of Ap₃A ([M-H]^- at m/z 755.077), Ap₃G ([M-H]^- at m/z 771.065) and Ap₅A ([M-2H]^2- at m/z 457.008) (Extended data Fig. 2a,b,c). We also observed significant signals of mono- and dimethylated forms of Np_nNs, specifically methyl-Ap₃A ([M-H]^- at m/z 769.080), dimethyl-Gp₄G ([M-2H]^2- at m/z 447.018) and methyl-Ap₅G ([M-2H]^2- at m/z 472.018) (Extended data Fig. 3a,b,c). Besides the previously mentioned caps, in the STA, we detected signals of methyl-Ap4G ([M-2H]^2- at m/z 432.016), methyl-Ap₅A ([M-2H]^2- at m/z 463.992) and dimethyl-Ap₅G ([M-2H]^2- at m/z 479.003) (Figure 2b, Extended data Fig. 4a,b,c). We compared the amounts of dimethyl-Gp₄G-RNA at various growth stages. The level of this cap was more than two-fold higher in the STA compared to EXP (Figure 2c). In general, a higher number of Np_nNs were detected in this phase. This may indicate that the cells in the STA lack the nutrients and methylate the Np_nN-caps to preserve RNA.

To confirm the structure of the detected Np_nN-caps, we grew the E. coli in minimal media with the sole source of nitrogen from either ¹⁴NH₄Cl or ¹⁵NH₄Cl. We detected only three caps: methyl-Ap₃A, Ap₃G, and dimethyl-Gp₄G (Figure 2b,d,e,f Extended data Fig. 5a,b,c), because this type of growth represents another type of stress. This experiment confirmed the presence of ten nitrogen atoms in every detected molecule. To further verify the chemical structure of Np_nNs, we compared the LC-MS properties of standard Gp₄G with the isomeric p₃GpG. While half of the p₃GpG was fragmented in the ionization source to p₂GpG, the Gp₄G stayed intact (Extended data Fig. 6a,b). The same behavior was observed for the dimethyl-Gp₄G in the E. coli RNA sample, proving the internal polyphosphate chain. By linear ion trap LC-MS, we detected an intact triphosphate chain of Ap₃A confirming its structure (Extended data Fig. 7).

Since Np_nN-capped RNAs are produced in E. coli, degradation mechanisms of the capped RNA in E. coli must exist. Ap_4-6A have been reported to be in vitro substrates for the E. coli NudiX enzyme NudH (RppH^{18, 23}). RppH is an E. coli decapping enzyme of 5’-triphosphate^4,14,24 and 5’-diphosphate RNA¹⁵. To assess whether Ap_4-6A-capped RNA can be an RppH substrate in vivo, we prepared Np_nN- and NAD-capped RNAs by in vitro transcription and tested the products as substrates for RppH. First, we added RppH to transform 5’-cappped RNA into 5’-p RNA, which then served as substrate for the terminator (Figure 3a). Electrophoretic analysis showed that Np_nN-capped RNAs are cleaved into 5’-p RNA and subsequently degraded (Figure 3b,c), suggesting that Ap_4-6A and Ap_4-5G-capped RNAs are excellent substrates for RppH in vitro. However, Ap₃A and NAD⁶ were cleaved less efficiently. Surprisingly, the Np_nN-capped RNAs were always superior substrates for RppH compared to 5’-ppp RNAs. To understand the substrate specificity of RppH, we performed kinetic study of Np_nN-capped RNAs. The decapping reaction of Ap_4,5N RNA was almost quantitative within 5 min while the corresponding 5’-ppp RNA was decapped only by 50% within 40 min (Figure 3d,e, Extended data Fig 8).

Figure 3:

RppH cleavage of Np_nN-capped RNA. a, Scheme showing the degradation of capped and uncapped RNA from in vitro transcription using RppH and terminator exonuclease. b, PAGE analysis of in vitro transcribed RNA treated with RppH (+R) or without RppH (-R). c, PAGE analysis of RNA from b after the reaction with the terminator (T). d,e, Kinetic studies of RppH cleavage of Np_nN-capped and uncapped RNA stopped after 30 s, 1, 2, 5, 10, 20 and 40 min and analysed by PAGE. d, Ap_nA-RNAs in comparison with pppA-RNA. e, Gp_nA-RNAs in comparison with pppA-RNA.

As Np_nN RNAs are excellent substrates for RppH in vitro, we constructed a mutant E. coli strain KS47 (Δrpph) with knock-down RppH to enhance the in vivo concentration of Np_nN RNAs. The RNA isolated from the mutant strain did not show any significant difference in Np_nN-capping compared to wild type, possibly due to the redundancy of NudiX enzymes.²⁵

To reveal the effect of Np_nN-caps methylation on the RNA, we performed molecular dynamics (MD) simulations of the interaction between RppH and Np_nNs-capped RNA (Gp₄G-G and 2mGp₄G-G). The LC-MS analysis confirmed the presence of one methyl group per guanosine moiety in 2mGp₄G (Extended data Fig. 9). We speculated that these methylations could be in the positions N⁷ (m⁷G) and O2’ (Gm) similarly to eukaryotic RNA caps. We observed in MD simulations that the interactions with the arginines R28 and R86 were lost when the methylations were present (Figure 4a). These arginines are responsible for purines binding via cation-π stacking. This finding demonstrates that methylation of Np_nNs-caps in RNA can hamper decapping by RppH.

Figure 4:

Role of RppH in the cleavage of Np_nN-capped RNA in E. coli. a, Snapshots from molecular dynamics simulation of the interaction of RppH with Gp₄G-G and m⁷Gp₄Gm-G after 200 ns. b,c Relative abundance of non-methylated Np_nNs (left) and methylated Np_nNs (right) as derived from EIC in the sRNA fraction spiked with Gp₄G-RNA before (blue) and after 1 h RppH treatment (red) b and spiked with Gp₅A-RNA before (blue) and after 1 h RppH treatment (red) c. d, Hypothetic cellular processing of RNA in E. coli at different stages of growth.

To experimentally prove the MD findings, we added RppH into the mixture of isolated sRNA with a spiked model Gp₄G-RNA or Gp₅A-RNA to compare the activity of RppH on methylated and non-methylated substrates. We found that majority of the model capped RNAs were cleaved within 1 h while the amount of naturally present methyl-Ap₅G-RNA slightly decreased and dimethyl-Gp₄G/Ap₅G-RNA remained unchanged (Figure 4b,c) This confirms that the methylation stabilizes the NpnN capped RNAs against cleavage by RppH.

In summary, we identified Np_nNs as novel 5’-RNA caps, which are incorporated into RNA by RNAPs. We found that Np_nN RNAs were cleaved by the E. coli RppH decapping enzyme. Caps with long polyphosphate chains were cleaved most efficiently and are better substrates than 5’-ppp RNAs. This suggests that they may be the main cellular targets of RppH. In the cell, we detected the presence of Np_nNs in sRNA, including their methylated variants. MD simulations and experiments revealed that the methylation protects the caps from the RppH cleavage. The amounts of caps and their methylations increased in the STA. Hence, we propose that bacteria use methylated caps to stabilize some RNAs under stress (Figure 4d). In the exponential phase, the metabolism is efficient and the turnover of the macromolecules is fast. Therefore, the methylated caps are not necessary and were detected in low amount. In the stationary phase, cells lack nutrients so they need to find a strategy to save macromolecules. The methylation of the Np_nN-caps can be the way to preserve important RNA molecules. It is intriguing to conceive that many functions of Np_nNs can be explained via their RNA capping potential. Moreover, we show that RNA possesses at its 5’-termini previously unknown structures that may interact with a wide range of cellular partners and influence e.g. cellular reaction to starvation. Besides searching for methyltransferases responsible for the methylation of the Np_nN-caps, the biggest challenges lie in the development of specific techniques for identification of Np_nN-capped RNAs.

Author Contributions

O.H., R.B., M.H. and H.C. conceived the study, designed the experiments. O.H. and R.B. performed the experiments. J.C. L.R. and H.C. supervised the work. M.C. performed molecular dynamics study. O.H., R.B., M.C. and H.C. wrote the paper.

Author Information

The authors declare no competing financial interests. Correspondence and requests for materials should be addresses to H.C. ()

Acknowledgements

We thank Dr. K. Stříšovský for help with the preparation of mutant E. coli strain KS47 (Δrpph), Dr. L. Krásný for critical reading of the manuscript and advices, Dr. P. Reyes-Gutierrez and other members of Cahova lab for their help and discussion. This work was supported by the Ministry of Education, Youth and Sports (Czech Republic), program ERC CZ (LL1603).