The HEN1 methyltransferase from Arabidopsis thaliana modifies the 3′-terminal nucleotides of small regulatory RNAs. Although it is one of the best characterized members of the 2′-O-methyltransferase family, many aspects of its interactions with the cofactor and substrate RNA remained unresolved. To better understand the substrate interactions and contributions of individual steps during HEN1 catalysis, we studied the binding and methylation kinetics of the enzyme using a series of unmethylated, hemimethylated and doubly methylated miRNA and siRNA substrates. The present study shows that HEN1 specifically binds double-stranded unmethylated or hemimethylated miR173/miR173* substrates with a subnanomolar affinity in a cofactor-dependent manner. Kinetic studies under single turnover and pre-steady state conditions in combination with isotope partitioning analysis showed that the binary HEN1–miRNA/miRNA* complex is catalytically competent; however, successive methylation of the two strands in a RNA duplex occurs in a non-processive (distributive) manner. We also find that the observed moderate methylation strand preference is largely exerted at the RNA-binding step and is fairly independent of the nature of the 3′-terminal nucleobase, but shows some dependency on proximal nucleotide mispairs. The results of the present study thus provide novel insights into the mechanism of RNA recognition and modification by a representative small RNA 2′-O-methyltransferase.
- enzyme kinetics
- 2′-O-ribose methylation
- RNA modification
Small non-coding RNAs such as miRNAs, siRNAs and piRNAs (Piwi-interacting RNAs) are essential for post-transcriptional gene regulation in eukaryotic organisms including humans . Biogenesis of plant miRNAs and siRNAs or animal piRNAs and Argonaute2-loaded siRNAs involves methylation of small RNA molecules at the 2′-OH group of their 3′-termini [2,3]. Members of the small RNA 2′-O-methyltransferases family share a conservative catalytic domain and can be further divided in two classes across metazoan taxonomy. One class is represented by germline-specific enzymes from animals (approximately 400 amino acids in size), which modify single-stranded piRNAs and Argonaute2-associated siRNAs [4–7]. A paradigm of the other class, the large multidomain methyltransferase HEN1 from Arabidopsis thaliana, as well as its rice homologue WAF1 (WAVY LEAF1), catalyse methyl group transfer to RNA duplexes: miRNA/miRNA* and siRNA/siRNA* [8–11]. HEN1 displays a strong preference towards dsRNAs and efficiently modifies both strands on miRNA/miRNA* in vitro . This methylation type is crucial for plant small RNA stability in vivo since the abundance of miRNAs and ta-siRNAs (trans-acting siRNAs) is greatly reduced in hen1 and waf1 mutants [10,13].
Analysis of the structure of HEN1 has revealed two putative dsRNA-binding domains divided by a La-type winged helix–turn–helix motif in its N-terminal section (Figure 1B) . A Rossman-fold methyltransferase region, which includes conserved motifs characteristic for AdoMet (S-adenosylmethionine)-binding proteins, is present at the C-terminus . Although the sole methyltransferase domain is sufficient for dsRNA methylation, kinetics data have provided evidence for the N-terminal domain's significance in stabilizing the catalytic complex . According to steady-state kinetic studies, HEN1 exhibits a high catalytic efficiency towards unmethylated miRNA/miRNA* duplexes ; however, the mechanistic details of the catalytic cycle of dsRNA recognition and modification by HEN1 remained unclear. Complete HEN1-mediated modification of dsRNA substrates is attained through hemimethylated intermediates (Figure 1A). Therefore detailed analysis of the interaction of methyltransferase with various substrates and successive catalytic step is required for deeper understanding of the multistage mechanism of the action of HEN1.
To obtain mechanistic insights into the substrate interactions and contributions of the individual steps during HEN1 catalysis, we studied binding interactions and methylation kinetics using a series of unmethylated, hemimethylated and doubly methylated substrates. Kinetic studies under single turnover and pre-steady state conditions in combination with isotope partitioning analysis showed that binary HEN1–miRNA/miRNA* complexes are catalytically competent and successive methylation of the two strands in a RNA duplex occurs in a non-processive (distributive) manner. We also find that a moderate methylation strand preference is exerted at the binding step, but is not determined by interactions with the 3′-terminal nucleotide. Similar binding and methylation parameters observed with siRNA/siRNA* suggest that the HEN1 methyltransferase does not distinguish the two types of small non-coding RNAs in vitro.
EMSAs were used to analyse the binary HEN1–miRNA/miRNA* complex and ternary complex with AdoHcy (S-adenosyl-L-homocysteine) or AdoMet. A series of 2-fold HEN1 dilutions in a range from 240 nM to 29 pM were equilibrated with 50 pM 32P-labelled duplex RNA on the 5′-end of a selected strand. Incubation was carried out for 30–40 min at 25°C in the reaction buffer [50 mM NaCl, 10 mM Tris/HCl, 0.25 mM MgCl2 and 10% glycerol (pH 7.5)] with or without 100 μM cofactor. Complexes were separated electrophoretically on a non-denaturing 8% polyacrylamide gel (19:1 acrylamide/bisacrylamide) in 0.5× Tris-borate buffer at 25°C. The radioactive bands were then visualized by Fluorescent Image Analyzer FLA-5100 (Fujifilm). The fractions of bound and free duplex RNA were quantified with Multi Gauge version 3.0 software (Fujifilm) and fitted to a single-site binding equation using GraFit version 5 software in order to obtain the Kd values.
To assess the dissociation rates of the RNA duplexes from the binary and ternary assemblies, complexes were created by pre-incubating 150 pM 5′-32P-labelled miRNA/miRNA* and 250 nM HEN1 with 100 μM AdoHcy or AdoMet or in the absence of cofactor in a reaction buffer containing 0.05 mg/ml BSA for 30 min at 25°C. After an equilibrium was established between the reaction components, portions of the reaction mixture were collected at certain time points and diluted more than 104-fold by the addition of 2 μM corresponding double-stranded unlabelled RNA. The samples were resolved on polyacrylamide gels as described above. The dissociation rate constants were estimated by fitting the data to single- and double-exponential decay equations.
Methylation Reactions [I]–[IV] were performed by mixing chase and pulse solutions (10 μl of each) by hand. The pulse solution of Reaction [I] contained 0.5 μM HEN1 and 0.4 μM (0.8 μM targets) double-stranded miR173/miR173* with either guide or passenger strand 33P-labelled at the 5′-end, whereas the chase mix contained 5 μM AdoMet and 40 μM non-radioactive miR173/miR173*. In the positive control (Reaction [II]), 0.5 μM HEN1 and 0.4 μM miR173/miR173* of the pulse solution were mixed with 5 μM AdoMet; in Reaction [III] both pulse and chase fractions contained a pre-incubated mixture of labelled and unlabelled RNA in the ratio of 1:100 pooled with either 2 μM HEN1 in the pulse fraction or with 5 μM AdoMet in the chase fraction. The background control Reaction [IV] replicated Reaction [I], but was quenched by 0.7 M HCl or 1 mg/ml Proteinase K (Fermentas) in SDS buffer [20 mM Tris/HCl (pH 7.4), 0.5 mM EDTA, 10 mM NaCl and 0.5% SDS] prior to incubation. All reactions were carried out for 15 min at 37°C. Reactions [I]–[IV] were reproduced twice with both types of quenching solutions. Samples were treated with sodium periodate as described previously , pre-heated with an equal volume of denaturing 2× RNA Loading Dye (Fermentas) and resolved on 15% PAGE with 7 M urea. The reaction product formation was quantified using FLA-5100 Image Reader (Fujifilm) and MultiGauge version 3.0 software. The analogous reactions with hemimethylated substrates (miR173CH3/miR173* and miR173/miR173*CH3) were prepared and analysed in the same way.
Four exchange experiments [I]–[IV] were carried out as described above: chase and pulse solutions (10 μl of each) were mixed manually. For Reaction [I], the pulse mix contained 2 μM HEN1 and 5 μM [methyl-3H]AdoMet (5.65 Ci/mmol) and the chase mix contained 0.4 μM (0.8 μM targets) miR173/miR173* and 500 μM ‘cold’ AdoMet (0 Ci/mmol; Sigma); for Reaction [II], the pulse solution contained 2 μM HEN1 and 5 μM [methyl-3H]AdoMet (5.65 Ci/mmol) and the chase solution contained 0.4 μM miR173/miR173*. In Reaction [III] [methyl-3H]AdoMet (5.65 Ci/mmol) and unlabelled AdoMet were premixed in the ratio 1:100 in both the pulse and chase mixes (final specific activity of AdoMet was 1.07 Ci/mmol), and combined to 2 μM HEN1 in the pulse mix or with 0.4 μM miR173/miR173* in the chase mix. Reaction [IV] (background control) was identical with Reaction [I], but quenched immediately after mixing. All reactions were incubated for 30 min at 37°C and quenched by 0.4 mM ‘cold’ AdoMet and 1 mg/ml Proteinase K in Stop solution [20 mM Tris/HCl (pH 7.4), 0.5 mM EDTA, 10 mM NaCl and 0.5% SDS]. Samples were spread on 2.3 cm DE-81 filters (Whatman), washed four times with 50 mM Na3PO4 (pH 7.0), twice with RNAse-free water, twice with ethanol and once with acetone, air-dried, and counted in 3 ml of Rotiszint Eco lipophylic LSC Cocktail (Carl Roth) using a Beckman LS1801 scintillation counter. The background counts (400–600 d.p.m.) were subtracted from the sample counts. Enzymatic activity was estimated by analysing data from two replicates.
HEN1 pre-steady-state kinetics assays under [E]>[S] conditions
The methyl group transfer on miR173/miR173* under [E]>[S] conditions was studied using both types of reciprocal 33P-labelled unmethylated or hemimethylated (miR173CH3–miR173* and miR173/miR173*CH3) RNA duplexes. Reaction component (A), 0.2 μM RNA and 0.5 μM HEN1 in Reaction buffer [10 mM Tris/HCl (pH 7.5), 50 mM NaCl, 0.25 mM MgCl2 and 0.1 mg/ml BSA) was mixed by a Rapid Chemical Quench-Flow appliance RQF-3 (KinTek) with component (B), containing 40–4000 μM AdoMet in the same buffer. The final concentrations of RNA, HEN1 and AdoMet after combining equal volumes (15 μl of each) of components A and B were 0.1 μM, 0.25 μM and 20–2000 μM respectively. Proteinase K in Stop buffer [7 mM Tris/HCl (pH 7.4), 0.17 mM EDTA, 3 mM NaCl and 0.5% SDS] was added to a final concentration of 0.4 mg/ml to stop the reaction after certain period of incubation at 37°C. Sodium periodate-treated samples were processed and analysed as described above (RNA-exchange assay).
The substrate binding specificity of HEN1
Previous studies revealed that HEN1 preferentially transfers methyl groups on to double-stranded miRNA/miRNA* and siRNA/siRNA*, but not to DNA or ssRNA [8,12]. It remained unclear, however, whether this specificity is exerted at the substrate binding or the catalytic step. We have thus evaluated the binding activity of HEN1 with a series of ssDNA, dsDNA, ssRNA or dsRNA substrates (for the sequences see Supplementary Figure S6 at http://www.biochemj.org/bj/453/bj4530281add.htm) using EMSAs. Preliminary binding experiments showed no binding of 0.05 μM ssRNA samples, as well as ssDNA or dsDNA even at a protein concentration of 0.5 μM, whereas a clear mobility shift was observed with dsRNAs (Figure 1C and Supplementary Figure S1 at http://www.biochemj.org/bj/453/bj4530281add.htm). Evidently HEN1 exhibits high binding selectivity towards small RNAs, such as miRNA/miRNA* and siRNA/siRNA* as opposed to other cellular nucleic acids.
During the catalytic cycle, an unmethylated miRNA/miRNA* substrate is converted into a hemimethylated (two reciprocal variants) and then to a fully methylated RNA duplex (Figure 1A). Therefore the HEN1 binding of the miR173/miR173* substrates with the different combinations of modified and unmodified strands was studied under non-denaturing conditions. The full-length methyltransferase was found to form stable complexes with all of the tested RNA duplexes (Supplementary Figure S2 at http://www.biochemj.org/bj/453/bj4530281add.htm). Although the truncated protein represented by the catalytic domain of HEN1 (residues 666–942) methylated both strands of the miRNA/miRNA* duplex to completion , it did not show a protein-mediated RNA mobility shift under similar experimental conditions (Figure 1C). These results indicate that the N-terminal region is important for RNA binding.
For a detailed characterization of the binding efficiency, the dissociation constants (Kd) were derived from titration experiments by fitting to a single-site binding model (Figure 2 and Table 1). These experiments revealed similar subnanomolar affinities for the unmethylated miR173/miR173* and both hemimethylated substrates. The binding of the fully methylated duplex was approximately 20-fold weaker than the unmethylated or hemimethylated substrate, indicating that the stability of the binary complex is considerably reduced when both target residues are methylated (Figure 2C). This observation implies that HEN1 prefers RNA duplexes containing at least one unmodified strand (hemimethylated reaction substrates) over the doubly methylated product. This indicates that the enzyme binds preferentially to an unmethylated strand oriented in the catalytic site. Likewise, methylation of the catalytically bound strand destabilizes the complex and promotes the release of the protein in order to be reused. Furthermore, doubly methylated small RNAs are unlikely to sequester HEN1 in vivo for control of the intracellular enzyme activity.
HEN1 methyltransferase is involved in the biogenesis pathways of various cellular small RNAs that differ in their structure, sequence and length [17,18]. Since the majority of small RNAs are modified at the 3′-termini, different siRNAs compete with miRNAs for methylation in Arabidopsis . On the other hand, plant miRNAs rival with other small RNAs for loading into specific Argonaute complexes , and previous studies proposed that the length, as well as the asymmetry in structure of the miRNA duplexes, can affect their fate [20–22]. Therefore, to determine the role of substrate structure on the binding function of HEN1, we compared the affinity of the methyltransferase towards different small RNAs: miRNA type let-7a/let-7a* (22/22 nt) with two mispaired bases inside of a double-stranded part and one on the 5′-termini, miR173/miR173* (22/21 nt) with symmetric and asymmetric bulges of one unpaired base, and perfectly complementary siRNA type siR173/siR173* (23/23 nt) (Supplementary Figure S6). The measured Kd values (Supplementary Figure S3 at http://www.biochemj.org/bj/453/bj4530281add.htm) for the two different miRNAs duplexes and the siRNA were quite similar, revealing no considerable capability of HEN1 to distinguish small RNAs of a different size or structure.
AdoHcy, the reaction product produced upon the catalytic transfer of the methyl group from AdoMet, is part of the ternary reaction complexes involving hemimethylated or fully methylated RNA. The effect of cofactor on RNA binding was tested in similar experiments. The dissociation constants obtained with the unmethylated and hemimethylated RNA substrates were similar in the absence or presence of AdoHcy (Table 1), suggesting that the demethylated cofactor does not affect the HEN1 affinity towards incompletely modified RNA substrates. Notably, the affinity for the fully methylated RNA increased 4-fold in the presence of 100 μM AdoHcy, but no enhancement was detected upon addition of AdoMet (Figure 2C). Altogether, it could be concluded that AdoHcy binding leads to the formation of the ternary product complex HEN1–miRNACH3/miRNA*CH3–AdoHcy which should enhance the overall affinity of HEN1 for RNA. In the case of AdoMet, such a ternary complex is probably impossible due to the presence of an additional methyl group on the cofactor which would lead to a steric clash between the methyl groups of AdoMet and the 3′-terminal nucleotide in the active centre.
In order to estimate the kinetic stability of the HEN1–RNA complexes, the release of 32P-labelled miRNA–miRNA* was measured in displacement assays in the presence of a 13000-fold excess of unlabelled RNA substrate. Control reactions with no RNA competitor showed that the complexes were stable for at least 8 h (Supplementary Figure S4, the lanes marked with a dash at http://www.biochemj.org/bj/453/bj4530281add.htm). Approximately 50% of the radiolabelled RNA was retained in the binary HEN1–RNA complex after a 1 h incubation with ‘cold’ RNA (Figure 3 and Supplementary Figure S4). The dissociation profiles of the complexes were best described by a double-exponential model (Table 1). In the case of unmethylated RNA, the fast phase (t1/2=8 min), which accounts for about 30% of the total amplitude, was well distinguished from a slow (t1/2=144 min) dominant phase. The addition of 100 μM AdoHcy to the reaction mixture slightly enhanced the overall stability of the HEN1 complex with double-stranded RNA. The decay of the complex under these conditions was well fitted to a single-exponential model resulting in a koff value comparable with that of the slower phase observed in the binary complex. These results suggest that the binary HEN1–RNA complex exists in two conformational states, and that AdoHcy binding shifts the equilibrium towards the stable state.
Analogous behaviour was also observed in binding experiments with hemimethylated RNAs (Figure 3 and Table 1). The koff value of the binary complex for both hemimethylated miRNA substrates may resemble the decay of the unmethylated miR173/miR173* assuming that the interaction mode of the HEN1 complexes remains intact irrespective of the occurrence of a methyl group on one of 3′-terminal ends. The binary HEN1 complex with the fully methylated miR173CH3/miR173*CH3 duplex decomposed substantially faster. Two-exponential fits of the obtained decay curves revealed partial loss of the first phase (~75% of the amplitude is lost at the first time point of 10 min). The second phase, with a half-life of 60 min, was clearly discernible, but had a lower amplitude than that observed in the complex with unmethylated and hemimethylated RNAs. The addition of AdoHcy to the reaction mixture reduced the off-rate of the complex as demonstrated for the unmethylated RNA. Consistent with the binding studies mentioned above, HEN1 exhibited a weaker retention of the fully methylated RNA duplex compared with the methylatable substrates.
Catalytic competence of the binary HEN1 complexes
Further insights into the assembly of the ternary catalytic complex were obtained by analysis of the catalytic competence of the binary complexes using an isotope partitioning method [23,24]. For analysis of the binary HEN1–RNA complex, pre-incubated pulse solution consisting of 0.2 μM dsRNA miR173/miR173* 5′-33P-labelled on either strand and 1 μM HEN1 was rapidly mixed with an equal volume of chase solution containing 5 μM AdoMet and a 100-fold excess of ‘cold’ unlabelled miR173/miR173* (Figure 4, Reaction [I]). The final reaction mixture containing 0.1 μM 33P-labelled RNA, 20 μM cold RNA, 0.5 μM protein and 2.5 μM AdoMet was incubated for 15 min. The methylation levels of 33P-labelled miRNA strands from the pulse component were determined by the periodate method. If the binary complex is productive, extensive modification of the 33P-labelled RNA during the first turnover is expected followed by release of the enzyme and subsequent methylation of diluted (predominantly cold) RNA. If the HEN1–RNA complex dissociates before or upon AdoMet binding, the extent of methylation of the 33P-labelled miRNA is expected to occur at the same level as observed in the prediluted control Reaction [III] (10–12%). Furthermore this experiment tracks the modification levels of individual radiolabelled strands in the miRNA/miRNA* duplex and can therefore reveal the methyltransferase preference for a particular binding orientation. Experimental data obtained with unmethylated RNA show that 68% and 48% of the 33P-labelled miR173* and miR173 strand respectively, was modified in Reaction [I] (Figure 4A). The extent of RNA methylated during the first turnover was obtained by subtracting the counts that were incorporated in subsequent turnovers (Reaction [III]), which gives 56% and 38% respectively (Table 2). These numbers jointly represent approximately 94% of the modifications of the duplex indicating that the binary HEN1–RNA complex undergoes full methylation before exchanging the radiolabelled substrate with ‘cold’ miRNA from the chase portion.
In light of a nearly 2-fold strand bias (preference towards miR173*) observed in the single-turnover reaction, we also examined if the nature of the 3′-terminal nucleotide is critical for the catalytic competence of the complex. Terminal variations of the guide strand invariantly showed a 1.5–3-fold higher methylation of the miR173* strand (Table 2) indicating indiscriminatory action of HEN1 with respect to the 3′-terminal nucleotide. Similar experiments were also performed with hemimethylated miR173/miR173*CH3 or miR173CH3/miR173*. During the first turnover 56–57% of a 33P-labelled unmethylated strand was methylated (Figure 4B and Table 2) confirming that the binary HEN1 complex with hemimethylated RNA is catalytically competent. Interestingly, the hemimethylated substrates showed no apparent strand preference. Altogether our experimental evidence is consistent with a model whereby HEN1 near randomly binds a hemimethylated RNA duplex in two alternative orientations exposing either a methylated (unproductive complex) or unmethylated (productive complex) terminus in the catalytic centre.
The catalytic competence of the HEN1–AdoMet complex was studied in reciprocal isotope-partitioning experiments. Pulse solution containing 5 μM 3H-labelled AdoMet and 2 μM HEN1 was rapidly mixed with the chase component containing 500 μM unlabelled AdoMet and 0.4 μM miR173/miR173* (Figure 4C). The incorporation of tritium-labelled methyl groups into RNA turned out to be comparable with that observed in a control experiment in which labelled AdoMet was pre-diluted with a 100-fold excess of unlabelled cofactor. These results suggest that: (i) the KdAdoMet value is much higher than KmAdoMet value (2 μM) and a detectable amount of the binary HEN1–AdoMet complex is not produced in the pulse solution at practically attainable concentrations of radioactively labelled AdoMet (5 μM); or (ii) the HEN1–AdoMet complex is catalytically unproductive and requires AdoMet dissociation and rebinding to form a catalytic HEN1–AdoMet–RNA complex; or (iii) the rate of AdoMet exchange is much faster than the rate of RNA binding or the rate of catalytic methyl transfer. Our attempts to estimate the KdAdoMet value using equilibrium dialysis proved unsuccessful due to long-term instability of HEN1 in reaction buffers.
Methylation of individual strands in unmodified and hemimethylated substrates
To further elucidate the mechanism of HEN1-mediated modification, we have analysed the methylation kinetics of different strands in unmethylated and hemimethylated RNA duplexes. The level of microRNA modification at the 3′-termini was determined by their reactivity with sodium periodate. This method allowed us to follow the methylation time course of individual 33P-labelled strands at high micromolar concentrations of the cofactor, since non-radioactive AdoMet was used. Figure 5 shows the kinetics of HEN1 methylation of the RNA substrates under single-turnover ([E]>[S]) conditions. The methyl group transfer on the miR173* strand of unmethylated miR173/miR173* follows double-exponential kinetics (Figure 5B, upper panel, and Table 3), and variations of the AdoMet concentration significantly affects the rate, but not the amplitude, of the two stages. According to the branched reaction model (Supplementary Figure S5 at http://www.biochemj.org/bj/453/bj4530281add.htm) for individual strand methylation in duplex RNA the faster reaction phase (representing ~75–86% of the total reaction) most likely corresponds to direct modification of the passenger strand (kchemmiR173*). The slower phase for unmethylated substrate (less than 25% of the total amplitude) can be attributed to the passenger strand methylation occurring after the modification of complementary strand and subsequent complex dissociation and rebinding. Notably, this interpretation accords with the results of isotope-partitioning experiments, which suggest preferential modification of the miR173* strand in the HEN1–miR173/miR173* complex during the first round of methylation cycle (Table 2).
Kinetic analysis of the miR173 strand methylation revealed a single phase with a rate matching the slower phase of the miR173* strand (Figure 5A), and which showed little variation in response to changes of the AdoMet concentration from 20 μM to 100 μM. The simplest explanation of the coincidence of the two rates is that HEN1 to a large extent initially binds the complementary miR173* strand, and the modification of miR173 only takes place after protein dissociation from the complex with hemimethylated miR173/miR173*CH3 (according to the model in Supplementary Figure S5). This assumption is also consistent with our observation that HEN1 preferably binds the unmethylated duplex miR173/miR173* in such an orientation that the 3′-end of miR173* is associated with the HEN1 catalytic domain (Figure 4A).
The methylation time course analysis of the unmethylated strands of both hemimethylated substrates (miR173/miR173*CH3 and miR173CH3/miR173*) was performed at a constant (100 μM) AdoMet concentration. The results were similar to those obtained with corresponding strands of unmethylated RNA. The methyltransferase activity towards the miR173 strand can be well described by a single-exponential function, whereas the miR173* data fit better to a double-exponential model (Figures 5A and 5B, lower plots, and Table 3). Moreover, the modification rates and amplitudes derived for individual strands of hemimethylated RNA coincide well with the values obtained for miR173/miR173* under similar reaction conditions. As shown in Figure 5(C), the speed of the miR173* methylation fast phase (supposed methyl group transfer rate) are completely consistent in experiments with different concentrations of the cofactor. These results in combination with the binding experiments provide compelling evidence that the methylation of the 3′-terminal nucleotide on one strand of a miRNA/miRNA* duplex is not affected by the presence of the 2′-O-methyl group on the 3′-terminal nucleotide of the opposite strand.
The mouse HEN1 homologue with a similar catalytic domain structure exhibited different specific methylation activities on piRNA substrates depending on nature of the 3′-terminal nucleotides . To explore the impact of the target nucleoside on the Arabidopsis enzyme efficiency, we compared the modification rates of miR173 with 3′-terminal G, A, T or wild-type cytidine (Table 4 and Supplementary Figure S7 at http://www.biochemj.org/bj/453/bj4530281add.htm). The rates of methyl group transfer by plant HEN1 obtained with wild-type miR173-C were similar to the rates observed with miR173-G and miR173-U (terminal G and U correspondingly). Moreover, although the methylation of the miR173-A strand was characterized by a biphasic turnover, the rates of the fast and slow stages were comparable with those obtained with the miR173* strand.
To further characterize the HEN1-catalysed methylation of miRNA-like substrates, we studied the modification of individual strands in miR210/miR210* duplex (Figure 6). In spite of a different sequence and secondary structure, both miRNA/miRNA* substrates were modified with similar rates in vitro. Another large class of regulatory non-coding RNAs methylated in plants by HEN1 is siRNAs, which, in contrast with microRNAs, contain perfectly paired complementary strands. Therefore an siRNA-like structure, siR173/siR173*, was designed with a sequence similar to miR173/miR173. Again we only found subtle differences between the methylation profiles of siRNA- and miRNA-like substrates (Figure 6), disproving the importance of type-depending structure for preferential modification of small RNAs in vitro. However, no preferential methylation of individual siR173 or siR173* strands was observed in contrast with miR173/miR173*. These data support the idea that the presence of irregular helical structures (mispaired nucleotides or asymmetric bulges) in a terminal region may hamper the methylation of the corresponding strand and thereby modulate the function of the methyltransferase.
The microRNA/siRNA methyltransferase HEN1 from A. thaliana is a representative of a large class of RNA 2′-O-methyltransferases from various eukaryotic and prokaryotic organisms which share a homologous catalytic domain [3,7,25–27]. However, the plant enzymes are distinct owing to their specificity toward double-stranded small RNAs. Technically, the double-stranded nature of an RNA substrate adds some topological complexity as two targets on the same duplex are to be methylated, and thus full methylation of miRNA/miRNA* is achieved via intermediate hemimethylated states (Figure 1A). Such a mechanism is dissimilar with the interactions of most other types of RNA methyltransferases including the above mentioned homologues, and rather resembles interactions of monomeric DNA enzymes with their palindromic target sites. Therefore studies of HEN1 from A. thaliana establish a mechanistic paradigm for the terminal modification of double-stranded nucleic acids.
The HEN1 methyltransferase effectively methylates fully complementary (like siRNA/siRNA*) or partially unpaired (like miRNA/miRNA*) 21–24 nt dsRNAs, exhibiting preference towards duplexes with two nucleotide single-strand 3′-overhangs [8,9,12]. The results of the present study show that the binding selectivity determines the strand selection for methylation in dsRNA in vitro (Figure 4 and Table 2). Crystallographic evidence , steady-state kinetics  and deletion analysis (Figure 1) indicate that structural elements located upstream of the catalytic domain are essential for tight interaction with RNA. Moreover the N-terminal dsRNA-binding domain (dsRBD1) was distinguished as a key factor for the initial binding of RNA duplex (; S. Baranauskė and G. Vilkaitis, unpublished results). Nevertheless, full methylation of both strands of an miRNA duplex substantially decreases the binding affinity of the enzyme suggesting that interaction of the 3′-terminal RNA nucleotide also contributes to the interaction with the miRNA/miRNA* substrate. Moreover, binding of AdoHcy greatly enhances the stability of the HEN1–RNA complex. Taken together, these observations demonstrate the importance of the catalytic C-terminal domain in the stabilization of the HEN1–miRNA/miRNA* complex. On the other hand, HEN1 displayed a substantial strand preference (for the miR173* strand) in the binding and methylation experiments, but neither of these effects were dependent on the nature of the 3′-terminal base (Tables 2 and 4). One may therefore assume that some other (internal) structural features of the RNA duplex determine a preferential binding orientation of the enzyme. Notably, the mouse single-stranded piRNA-modifying methyltransferase mHEN1 showed a much stronger variation of methylation efficiency on substrates with different 3′-terminal nucleosides (A>C>U>G) , suggesting that microRNA/siRNA and piRNA methyltransferases may have different sensitivities to the nature of the target nucleotides.
To query the binding order and the catalytic competence of the reaction complexes along the reaction pathway we used isotope-partitioning approach. Experiments with HEN1–miRNA/miRNA* complexes showed that the amplitude of the methylation events on both miR173/miR173* strands was equal to approximately one turnover of the enzyme with an unmethylated duplex and 0.5–0.6 turnovers with hemimethylated RNA. The obtained results (Figure 4 and Table 2) clearly show that the HEN1–miRNA/miRNA* complex can bind the cofactor and turnover to products without dissociation and rebinding of the RNA indicating that the binary complex is catalytically competent. This observation excludes any ordered kinetic mechanism in which AdoMet is required to bind first. On the other hand, this means that the methylation of the other strand in the duplex is non-processive as this would give the methylation amplitudes of 2 and 1 enzyme turnovers respectively in the partitioning experiment. Therefore the HEN1–RNA complex does disintegrate after the first modification event and then rebinds again for the next cycle. From the structural perspective one can conclude that the methylation status of the 3′-terminal nucleotides determines the catalytic competence of the HEN1–RNA complexes such that complexes in which the catalytic domain contacts a methylated 3′-terminal nucleoside are catalytically incompetent (unproductive). The observed inability of HEN1 to switch over towards the opposite strand without fully releasing the bound RNA duplex in vitro may look somewhat unexpected in light of an interaction of bound RNA with both the catalytic and the N-terminal RNA-binding domains. Interestingly, a number of characterized DNA methyltransferases, which typically modify one strand at a time in their palidromic target sites, have been shown to act processively on adjacent target sites in DNA [28–31]. However, to date no examples of processive cross-strand methylation have been reported.
This work was supported by the Research Council of Lithuania [grant numbers MIP-99/2010 and MIP-028/2012 (to G.V.)] and a Student Research Fellowship Award from the Research Council of Lithuania (to A.O.).
The authors are grateful to L. Juknaitė for experimental assistance.
Abbreviations: AdoHcy, S-adenosyl-L-homocysteine; AdoMet, S-adenosylmethionine; piRNA, Piwi-interacting RNA; WAF1, WAVY LEAF1
- © The Authors Journal compilation © 2013 Biochemical Society