S-adenosylmethionine directly inhibits binding of 30S ribosomal subunits to the S_MK box translational riboswitch RNA

Abstract

The S_MK box is a conserved riboswitch motif found in the 5′ untranslated region of metK genes [encoding S-adenosylmethionine (SAM) synthetase] in lactic acid bacteria, including Enterococcus, Streptococcus, and Lactococcus sp. Previous studies showed that this RNA element binds SAM in vitro, and SAM binding causes a structural rearrangement that sequesters the Shine–Dalgarno (SD) sequence by pairing with an anti-SD (ASD) element. A model was proposed in which SAM binding inhibits metK translation by preventing binding of the ribosome to the SD region of the mRNA. In the current work, the addition of SAM was shown to inhibit binding of 30S ribosomal subunits to S_MK box RNA; in contrast, the addition of S-adenosylhomocysteine (SAH) had no effect. A mutant RNA, which has a disrupted SD-ASD pairing, was defective in SAM binding and showed no reduction of ribosome binding in the presence of SAM, whereas a compensatory mutation that restored SD-ASD pairing restored the response to SAM. Primer extension inhibition assays provided further evidence for SD-ASD pairing in the presence of SAM. These results strongly support the model that S_MK box translational repression operates through occlusion of the ribosome binding site and that SAM binding requires the SD-ASD pairing.

Many regulatory mechanisms have been discovered recently in bacteria in which RNA transcripts sense a regulatory signal (1–6). These regulatory RNAs, usually called RNA sensors or riboswitches, act in cis to regulate expression of the downstream coding sequence(s) without a requirement for regulatory proteins. Typically, the regulatory signal is an effector molecule (tRNA, noncoding RNA, or small molecule) that binds the nascent RNA transcript, causing a change in the RNA structure. In many systems of this type, the structural change either causes or prevents formation of the helix of an intrinsic transcriptional terminator (reviewed in 1, 4). Similar structural rearrangements can sequester the ribosome binding site (RBS) to regulate at the level of translation initiation (7–9). Binding of the effector to the RNA can also catalyze self-cleavage of the RNA transcript, as in the glmS system (10). Thermosensor riboswitches, in contrast, have no effector binding domain but simply respond by temperature-dependent modulation of the RNA structure (11–13). Riboswitches generally exhibit conserved sequence and structural elements that are responsible for high specificity for their cognate molecular signal.

Many members of the Bacillus/Clostridium group of bacteria use the S box riboswitch system to control expression of genes involved in methionine metabolism (14). Included in this regulon is the metK gene, which encodes S-adenosylmethionine (SAM) synthetase, the enzyme responsible for the synthesis of SAM from methionine and ATP. The S box system uses SAM as the effector molecule both in vivo (15, 16) and in vitro (15, 17, 18), and regulation occurs primarily at the level of premature termination of transcription (14, 15). Translational regulation has also been predicted in certain S box genes, predominantly those found in Gram-negative organisms (refs. 2 and 6; F.J.G. and T.M.H., unpublished data). In contrast, the S box riboswitch is rare in members of the Lactobacillales branch of the Gram-positive bacteria. We investigated the metK genes of Lactobacillales sp. and found a conserved putative riboswitch element, which we named the S_MK box on the basis of its association with metK genes (19). Because metK expression in organisms that use the S box is controlled by negative feedback from SAM, it was likely that SAM was also the effector molecule for the S_MK box.

We demonstrated that the Enterococcus faecalis S_MK box RNA binds SAM but not S-adenosylhomocysteine (SAH), which differs from SAM by one methyl group (19). SAM binding causes a structural rearrangement in the RNA that includes pairing of a portion of the metK Shine–Dalgarno (SD) sequence to an anti-SD sequence (ASD) (Fig. 1). Unlike other metabolite-binding riboswitches, which usually contain an effector binding domain that is separable from the regulatory domain of the RNA, the S_MK box appears to require the SD-ASD pairing for SAM binding, as mutations that disrupt the pairing result in loss of SAM binding (19). The E. faecalis S_MK box was shown to confer translational repression of a lacZ reporter in Bacillus subtilis under conditions where SAM pools are elevated, and mutations of the S_MK box that result in loss of SAM binding in vitro result in loss of repression in vivo. These findings suggested that the S_MK box RNA represents a new class of SAM-responsive riboswitch that regulates metK expression at the level of translation initiation by preventing binding of the ribosome to the RNA when SAM is present. However, there was no direct evidence that the apparent sequestration of the SD sequence is sufficient to inhibit ribosome binding. Binding of other regulatory molecules, including proteins and antisense RNAs, directly to or near the RBS can inhibit translation initiation by preventing ribosome binding (20, 21). Alternatively, regulatory molecules can bind to an mRNA and “trap” the 30S subunit on the mRNA in an inactive initiation complex (22, 23).

Fig. 1.

Structural model of the S_MK box. Putative forms of E. faecalis metK RNA in the absence (a) or presence (b) of SAM (∗). Red, residues of the ASD sequence; blue, residues of the SD sequence; purple, AUG start codon; green arrows, location of stops observed in primer extension assays; black arrows, mutations in the E. faecalis S_MK box. The U14G substitution was introduced by including tandem G residues at the start-site for T7 RNAP transcription. Numbering of the sequence is relative to the predicted E. faecalis metK transcription start-site. The U22G and A93C substitutions were predicted to disrupt the SD-ASD pairing. Dotted lines represent potential pairings supported by phylogenetic and mutational analyses (ref. 19; A. Smith, R.T.F., F.J.G., and T.M.H., unpublished results). The position of SAM within the RNA is not known.

In this study, we used nitrocellulose filter binding and ribosomal toeprinting assays to test whether SAM inhibits the binding of 30S ribosomal subunits to S_MK box RNA in vitro. The results show that the metK RNA start codon is properly positioned in the P site of the 30S subunit in the absence of SAM, and that preincubation of the RNA with SAM reduced 30S subunit binding to the RNA and resulted in inhibition of processivity of reverse transcriptase near the SD region, which supports our previous observation that this area becomes more structured in the presence of SAM (19). These findings confirm our model for S_MK box RNA regulation, where SAM-dependent SD-ASD pairing directly inhibits translation initiation by blocking access of the ribosome to the RBS.

Results

SAM Inhibits 30S Subunit Binding to S_MK Box RNA.

Nitrocellulose filter binding assays were performed to determine whether the proposed pairing of the SD to the ASD in E. faecalis metK RNA is sufficient to inhibit binding of 30S ribosomal subunits to the mRNA. This assay took advantage of the ability of nitrocellulose filters to bind 30S subunits but not free nucleic acids. Because an E. faecalis metK-lacZ translational fusion exhibited a pattern of regulation in vivo in Escherichia coli, similar to that previously reported (ref. 19; data not shown), binding of highly purified E. coli 30S subunits to S_MK RNA was tested. The heterologous ribosomal subunits were selected because the S_MK system is absent in E. coli, reducing the probability that any additional cellular factors potentially required for regulation would co-purify with the ribosomal subunits. End-labeled in vitro-generated S_MK box RNA was heated, slow-cooled, and incubated in the presence or absence of SAM. E. coli 30S ribosomal subunits were added, in the presence or absence of tRNA^fmet, the samples were passed through a nitrocellulose filter, and the amount of radiolabeled RNA retained by the filter was determined by scintillation counting.

Incubation of RNA corresponding to positions 15–118 relative to the E. faecalis metK transcription start-site with 30S ribosomal subunits resulted in ≈30% retention of the RNA, whereas <0.03% retention was observed when ribosomal subunits were not added (Fig. 2a). The addition of SAM resulted in a 4-fold reduction in retention of the RNA by the 30S subunits, whereas the addition of SAH had no effect. Similar results were obtained when the RNA was preincubated with 30S subunits before the addition of SAM (Fig. 2b), suggesting that the RNA-ribosome complex can be disrupted by the addition of SAM. Retention of the RNA increased 2-fold when tRNA^fmet was included presumably because the tRNA stabilizes the interaction of the 30S subunit with the metK RNA. Incubation of the RNA with SAM before the addition of 30S subunits and tRNA^fmet reduced retention of RNA ≈2-fold, but SAM did not significantly reduce retention of the RNA when both 30S subunits and tRNA^fmet were added before the addition of SAM (Fig. 2b). The loss of the effect of SAM is not a result of SAM preferentially binding to tRNA^fmet over the metK RNA because tRNA^fmet did not retain a measurable amount of SAM in a binding assay (data not shown). These results suggest that SAM can compete with 30S subunit binding to the RNA, but the addition of the initiator tRNA to the complex results in a stable complex that is resistant to challenge by SAM and is therefore committed to translation.

Fig. 2.

Nitrocellulose filter binding assays. RNAs generated by T7 RNAP transcription were gel purified and end-labeled with [γ-³²P]ATP. Labeled RNAs were incubated with SAM or SAH followed by tRNA and 30S subunits (a) or tRNA and 30S subunits followed by SAM or SAH (b). Samples were passed through a nitrocellulose filter, and material retained by the filter was quantified by scintillation counting. Less than 0.03% of input RNA was retained by the filters in the absence of 30S subunits.

The effect of mutations in the SD-ASD pairing on ribosome binding was also tested. The U22G substitution in the ASD, which was predicted to disrupt pairing with residue A93 immediately downstream of the SD (positions 88–92), resulted in a 40-fold reduction of SAM binding in vitro (Fig. 3a), as was previously observed for other mutations in the ASD element (ref. 19; R.T.F., unpublished data). The U22G mutation also resulted in complete loss of SAM-dependent inhibition of 30S subunit binding (Fig. 3b). The A93C mutation, which is predicted to disrupt SD-ASD pairing without affecting the SD sequence itself, similarly resulted in loss of SAM binding in vitro (Fig. 3a). Combination of the U22G and A93C substitutions, which is predicted to restore pairing, restored both SAM binding and sensitivity to SAM in the filter binding assays, indicating that the A93C mutation could suppress the defect in response to SAM conferred by the U22G substitution. These results show that the reduction of 30S subunit binding in the presence of SAM occurs only when the RNA can bind SAM. The increase in 30S subunit binding exhibited by the U22G mutant RNA in the absence of SAM compared with either the wild-type or U22G/A93C double mutant RNA (Fig. 3b) suggests that SD-ASD pairing occurs to some extent even in the absence of SAM and that SAM binding shifts the equilibrium to the paired form. These experiments support the model that the SAM-dependent SD-ASD pairing can inhibit 30S ribosomal subunit binding to S_MK box RNA.

Fig. 3.

Effect of S_MK box mutations on binding of SAM and 30S ribosomal subunits. (a) SAM binding. RNAs generated by T7 RNAP transcription were gel-purified and then incubated in the presence of [methyl-¹⁴C]SAM. The RNA-bound SAM was separated from unbound SAM by size-exclusion filtration. Retention of SAM is expressed relative to that of the wild-type E. faecalis metK RNA extending from residues 15–118 (Fig. 1). (b) Nitrocellulose filter binding assays. RNAs end-labeled with [γ-³²P]ATP were incubated with SAM or SAH followed by tRNA and 30S subunits. Samples were passed through a nitrocellulose filter, and material retained by the filter was quantified by scintillation counting. Less than 0.03% of input RNA was retained by the filters in the absence of 30S subunits.

Lactobacillus plantarum metK, which also contains a SAM binding S_MK box element (19), showed the same pattern of 30S subunit binding as the E. faecalis S_MK box RNA, although the binding efficiency was somewhat lower (Fig. 3b). This result suggests that other S_MK box RNAs exhibit a similar effect of SAM on ribosome binding.

S_MK Box RNA Is Positioned Properly Within the 30S Subunit.

Although the filter binding studies indicated that 30S subunits bound the E. faecalis metK RNA, there was still a question as to whether binding was localized to the RBS. Primer extension inhibition (“toeprint”) assays (24) were used to localize the position of the 30S subunit on the RNA and assess the effect of SAM. A DNA primer complementary to a region of the E. faecalis metK RNA downstream of the SD was annealed to the RNA and then extended by reverse transcriptase. Extension by reverse transcriptase is inhibited when the enzyme encounters a bound obstacle (e.g., the 30S subunit) or significant secondary structure. The extension products were resolved by PAGE and identified by comparison to a DNA sequencing ladder.

Reactions containing the E. faecalis wild-type S_MK RNA (containing residues 15–208 relative to the transcription start-site) resulted in two extension products, one corresponding to a halt at position G77 and the other resulting from full-length extension to the 5′ end of the RNA (Fig. 4, lane 1). The G77 product was observed in all reactions and is probably due to secondary structure in the core region of the RNA (Fig. 1). A 5′ truncated RNA that began with residue 20, which retained SAM binding (Fig. 3a), also exhibited the G77 product (data not shown); this indicates that the conserved predicted pairing between positions 15–20 and 68–75 (ref. 19; Fig. 1a), which is absent in the 20–208 RNA, is dispensable for SAM binding and is not responsible for the G77 extension product.

Fig. 4.

Primer extension inhibition analysis of S_MK box RNA. E. faecalis metK RNAs containing positions 15–208 were generated by T7 RNAP transcription, gel-purified, and then annealed to a [γ-³²P]ATP-labeled DNA primer complementary to positions 180–203. Annealed RNAs were incubated in the presence of SAM or SAH followed by tRNA and 30S subunits. dNTPs and reverse transcriptase were added, and reactions were quenched by the addition of gel-loading buffer. Radioactivity at the position of the U118 stop in lanes 7 and 8 was compared with the value for lane 6, which was normalized to 100. Lanes 10 and 11 were compared with lane 9, and lanes 13 and 14 were compared with lane 12 in a similar manner. RT, readthrough to 5' end.

Incubation of the 15–208 RNA in the presence of SAM resulted in a new extension product corresponding to position A94 (Fig. 4, lane 2), which is two nt 3′ of the end of the SD (residues 88–92) and comprises the first position of the predicted SD-ASD pairing (19; Fig. 1b). Studies with avian myeloblastosis virus reverse transcriptase have shown that termination can occur when the enzyme encounters secondary structure, and the majority of these truncated products correspond to the first paired base in the structure or the last position before a cross-link (25, 26). Incubation of the RNA in the presence of SAH instead of SAM did not result in formation of the A94 product (Fig. 4, lane 3). The A94 extension product was also absent in reactions using the U22G mutant RNA (which did not bind SAM) and was present in the U22G/A93C double mutant in which both the SD-ASD pairing and SAM binding were restored (Fig. 4, lanes 10, 13). These observations support the conclusion that the A94 product depends on the RNA-SAM binding interaction and is likely to correspond to the SD-ASD pairing.

The addition of 30S subunits to a reaction containing the 15–208 RNA resulted in a set of low-abundance products clustered around U116–U122 (Fig. 4, lane 4). If both 30S subunits and tRNA^fmet were included, ≈90% of the primer extension products corresponded to position U118, which is 16 nt downstream of the A in the AUG start codon (Fig. 1). This distance is representative of the position at which reverse transcriptase stops when the start codon of an mRNA is positioned in the P site of the 30S subunit (27). The abundance of extension products corresponding to U118 was reduced ≈2-fold in the presence of SAM, whereas SAH had no effect. These results indicate that E. coli 30S subunits bind to the E. faecalis metK RBS and that ribosome binding is inhibited in the presence of SAM.

In contrast to what was observed for the wild-type RNA, the addition of SAM had no effect on the abundance of the U118 product for the U22G mutant RNA (Fig. 4, lanes 9–11), which is defective in SAM binding. Introduction of the compensating A93C mutation, which suppressed the SAM binding defect of the U22G mutation (Fig. 3a), resulted in restoration of sensitivity to SAM (Fig. 4, lanes 12–14), providing further support for the model that the SD-ASD pairing is required for both SAM binding and SAM-dependent inhibition of ribosome binding. This confirms the results from the filter binding assays where SAM inhibited the binding of 30S subunits to S_MK box RNA only when the RNA can bind SAM.

Discussion

In this study, we report that E. coli 30S subunits bind to S_MK box RNA in the absence of other factors and that SAM specifically inhibits ribosome binding to the RNA. Primer extension analysis demonstrated that in the absence of SAM, the 30S subunits are positioned with the AUG start codon of the RNA within the ribosomal P site, whereas the addition of SAM resulted in a significant reduction in the ribosomal toeprint. Under these conditions, the extension proceeded through the ribosomal toeprint site and yielded a new product corresponding to the 3′ end of the SD-ASD pairing. These findings further support the model that SAM binding promotes pairing of the SD to the ASD, which prevents translation initiation by blocking access of the ribosome to the RBS.

Most studies on the mechanism of action of riboswitches have focused on transcription termination systems. In systems of this type, binding of the effector to the riboswitch RNA causes the RNA to form an intrinsic transcriptional terminator or a competing antiterminator element. Structural probing, in vitro transcription, and mutational analyses have provided support for this mode of action for several of these riboswitch systems (reviewed in 1, 4). By contrast, most translational riboswitches have been analyzed in less detail, with most of the experimental work focused on regulation in vivo and effector-dependent modulation of RNA structure in vitro. Primer extension assays of the E. coli btuB riboswitch showed that the ability of 30S subunits to bind btuB RNA is reduced when adenosylcobalamin is present, providing a clear indication that regulation occurs at the level of translation initiation (7). One caveat is that the ribosomes used in the btuB experiments were prepared with a low-salt wash, which does not remove all ribosome-associated factors; binding was not observed with ribosomal subunits prepared with a high-salt wash, which was interpreted by the authors to indicate that other factors could be involved in translational control of btuB mRNA (7). The current study demonstrated a SAM-dependent effect on binding of 30S ribosomal subunits prepared with a high-salt wash from a heterologous host in which the S_MK box mechanism is not found, demonstrating that the observed effect of SAM on ribosome binding does not require additional cellular factors. The S_MK box is therefore the first translational control riboswitch for which the translational mechanism is clearly established by biochemical analyses.

Inhibition of 30S subunit binding to S_MK box RNA was observed regardless of whether SAM was added before or after the ribosomal subunits. However, the addition of the initiator tRNA to the complex resulted in a complex that was resistant to SAM, indicating that the complex was now committed to translation. This suggests that competition between the ribosome and SAM for binding to the mRNA occurs at the initial stages of formation of the initiation complex. One possible benefit of regulating in this manner is that the cell can quickly respond to a sudden drop in SAM levels by release of SAM from S_MK box mRNAs, allowing access to the RBS and expression of the gene. The importance of a rapid response may reflect the fact that metK is an essential gene for which expression is required at some level during all growth conditions. The possibility that the S_MK box represents a true reversible switch will require further analysis of regulation in vivo because reversibility depends on maintenance of a stable pool of the metK transcript.

Multiple mechanisms of translational regulation have been identified in bacteria. For example, the E. coli CsrA protein regulates the expression of cstA by directly binding the cstA RBS and preventing ribosome binding (20). A similar mechanism is used by the trans-acting OxyS RNA, which pairs with the RBS of the fhlA mRNA (21). Thermosensor RNAs, like E. coli rpoH and Listeria monocytogenes prfA, use intramolecular interactions that sequester the RBS at low temperatures and expose the RBS at higher temperatures (11–13). In contrast, regulation of the ribosomal proteins encoded by the α operon in E. coli results from “entrapment” of the 30S subunit in an unproductive complex by binding of the S4 repressor protein to the mRNA (23). Two examples of a “typical” metabolite-binding riboswitch translational mechanism are found in the ypaA gene in B. subtilis and the thiM gene in E. coli (9, 28, 29). In each case, the effector molecule, flavin mononucleotide for ypaA and thiamine pyrophosphate for thiM, binds to an aptamer domain in the 5′ untranslated region of the mRNA. This binding causes a downstream structural shift that sequesters the RBS, the regulatory target, and sequestration of the RBS was proposed to block ribosomal access, although no biochemical analyses were performed to test the proposed mechanism. The S_MK box also uses effector-dependent occlusion of the RBS to inhibit translation initiation but differs from these other systems in that the regulatory target is an intrinsic part of the effector binding domain, so that disruption of the SD-ASD pairing results in loss of SAM binding (Fig. 3a; ref. 19). The S_MK riboswitch therefore represents a special class of metabolite-binding regulatory RNAs.

The S_MK box represents one of the simplest known metabolite-binding riboswitch elements. The crystal structure of the much more complex S box RNA from Thermoanaerobacter tengcongensis has recently been reported (30), and the RNA was shown to envelop and interact with almost every functional group in SAM. The location and structure of the SAM binding pocket in the S_MK box RNA is not yet known, and the structural dissimilarity of S_MK and S box RNAs suggests that the molecular basis for SAM recognition may be very different.

The majority of riboswitches identified in low G+C Gram-positive organisms appear to regulate gene expression by premature transcription termination, whereas riboswitches from high G+C Gram-positive and Gram-negative organisms tend to regulate at the level of translation (1, 2, 5, 6). This divergence is readily observed with riboswitch elements that are found in both groups of organisms but regulate by different mechanisms depending on the host (e.g., the lysine-binding L box riboswitch; ref. 31). The S_MK box, which is found only in the Lactobacillales group of low G+C Gram-positive bacteria, is unusual in that it appears to function only at the level of translation. A third class of SAM-responsive riboswitch was recently identified in α-proteobacteria, but the mechanism of action has not yet been reported (32). The experiments in the current study were performed with E. coli components, whereas in vivo experiments in previous work were performed in B. subtilis (19). These results show that S_MK box riboswitches have the potential to function as gene regulators in a variety of nonnative organisms, including both Gram-positive and Gram-negative species. It will be important to demonstrate that the mechanism operates as predicted in the native host.

Materials and Methods

DNA Constructs.

DNA templates for transcription by T7 RNA polymerase (RNAP) were generated by combining pairs of complementary oligonucleotides as previously described (33). Sequence changes were introduced by replacement of a pair of oligonucleotides with a pair containing the changes. The metK constructs were derived from the sequences for E. faecalis strain V583 (NCBI AE016830) and L. plantarum strain WCFS1 (NCBI AL935263). All constructs were confirmed by DNA sequencing.

T7 RNAP Transcription Reactions.

RNA was generated by using a MEGA-shortscript T7 RNAP high-yield transcription kit (Ambion, Austin, TX). The RNA products were gel-purified as described previously (33).

SAM Binding Assays.

T7 RNAP-transcribed RNA (3 μM) in 1× transcription buffer (34) was heated to 65°C and slow-cooled to 40°C. [methyl-¹⁴C]SAM [ICN; 52 mCi mmol⁻¹ (1.92 GBq mmol⁻¹)] was added to a final concentration of 8 μM. After incubation for 10 min at room temperature, samples were passed through a Nanosep 10K Omega filter and washed four times with 75 μl of 1× transcription buffer. Radioactivity retained by the filters was measured in a Packard Tri-Carb 2100TR liquid scintillation counter.

Filter Binding Assays.

T7 RNAP-transcribed RNAs corresponding to positions 15–118 of the E. faecalis metK transcript were 5′ end-labeled with [γ-³²P]ATP (7,000 Ci mmol⁻¹) by using a KinaseMax kit (Ambion) and passed through a MicroSpin G-50 column (Amersham Biosciences, Piscataway, NJ) to remove excess [γ-³²P]ATP. Labeled RNAs (10 nM) in 1× binding buffer (10 mM Tris·HCl, pH 7.5, 10 mM MgCl₂, 60 mM NH₄Cl, and 6 mM 2-mercaptoethanol) were heated to 65°C and slow-cooled to 40°C. SAM (160 μM), SAH (160 μM), or water was added, and the samples were incubated for 10 min at 37°C. 30S ribosomal subunits (60 nM) isolated from E. coli MRE600 by using a high-salt wash (35) and provided by K. Fredrick (Ohio State University) were added, followed by a 30-min incubation at 37°C. tRNA^fmet (120 nM; Sigma–Aldrich, St. Louis, MO) was included as indicated. Samples were then loaded onto a nitrocellulose filter (0.45-μM pore size; Whatman, Clifton, NJ) that had been soaked in 1× binding buffer. A vacuum was applied, and filters were washed five times with 300 μl of 1× binding buffer. Radioactivity retained by the filters was measured in a Packard Tri-Carb 2100TR liquid scintillation counter.

Primer Extension Inhibition Assays.

A DNA primer complementary to positions 180–203 of the E. faecalis metK transcript was 5′ end-labeled as described above, except that a G-25 column was used to remove excess [γ-³²P]ATP. The labeled oligonucleotide (10 nM) was annealed to T7 RNAP-transcribed RNA (10 nM) corresponding to positions 15–208 of the metK transcript in 1× binding buffer by heating to 65°C and slow cooling to 40°C. SAM (160 μM), SAH (160 μM), or water was added, and the samples were incubated for 10 min at 37°C. 30S ribosomal subunits [60 nM; isolated from E. coli MRE600 by using a high-salt wash (35) and provided by K. Fredrick (Ohio State University)] were added, with or without tRNA^fmet (120 nM; Sigma–Aldrich), followed by a 30-min incubation at 37°C. Avian myeloblastosis virus reverse transcriptase (1 unit per rxn, Thermoscript RT-PCR; Invitrogen, Carlsbad, CA) and dNTPs (375 μM) were added followed by 10-min incubation at 37°C. Reactions were quenched by the addition of gel-loading buffer (Ambion) and resolved by using 10% denaturing PAGE. Products were visualized by using PhosphorImager analysis (Molecular Dynamics, Sunnyvale, CA) and quantified by using ImageQuant 5.2 software. A DNA sequencing ladder was generated by using a DNA Sequenase 2.0 Kit (USB), a DNA template containing positions 15–220 of the E. faecalis metK gene, and the same downstream primer as was used in the primer extension assays.

Abbreviations

SAM

S-adenosylmethionine

SAH

S-adenosylhomocysteine

RNAP

RNA polymerase

SD

Shine–Dalgarno

ASD

anti-SD

RBS

ribosome binding site.