The aminoglycoside resistance methyltransferases from the ArmA/Rmt family operate late in the 30S ribosomal biogenesis pathway

Abstract

Bacterial resistance to 4,6-type aminoglycoside antibiotics, which target the ribosome, has been traced to the ArmA/RmtA family of rRNA methyltransferases. These plasmid-encoded enzymes transfer a methyl group from S-adenosyl-L-methionine to N7 of the buried G1405 in the aminoglycoside binding site of 16S rRNA of the 30S ribosomal subunit. ArmA methylates mature 30S subunits but not 16S rRNA, 50S, or 70S ribosomal subunits or isolated Helix 44 of the 30S subunit. To more fully characterize this family of enzymes, we have investigated the substrate requirements of ArmA and to a lesser extent its ortholog RmtA. We determined the Mg⁺² dependence of ArmA activity toward the 30S ribosomal subunits and found that the enzyme recognizes both low Mg⁺² (translationally inactive) and high Mg⁺² (translationally active) forms of this substrate. We tested the effects of LiCl pretreatment of the 30S subunits, initiation factor 3 (IF3), and gentamicin/kasugamycin resistance methyltransferase (KsgA) on ArmA activity and determined whether in vivo derived pre-30S ribosomal subunits are ArmA methylation substrates. ArmA failed to methylate the 30S subunits generated from LiCl washes above 0.75 M, despite the apparent retention of ribosomal proteins and a fully mature 16S rRNA. From our experiments, we conclude that ArmA is most active toward the 30S ribosomal subunits that are at or very near full maturity, but that it can also recognize more than one form of the 30S subunit.

INTRODUCTION

Aminoglycosides have been important antibiotics for treatment of serious bacterial infections, being especially effective against aerobic Gram-negative bacteria (Shakil et al. 2008). However, in a clinical setting bacteria have acquired several resistance mechanisms to these compounds, including (Magnet and Blanchard 2005): (1) deactivation of aminoglycosides by N-acetylation, adenylylation, or O-phosphorylation; (2) reduction of the intracellular concentration of aminoglycosides by changes in outer membrane permeability (Nikaido 2003), decreased inner membrane transport (Taber et al. 1987), active efflux (Moore et al. 1999; Magnet et al. 2001), and drug trapping (Menard et al. 1993; Magnet et al. 2003); and (3) alteration of the 30S ribosomal subunit by mutation (Musser 1995).

A fourth resistance mechanism is autoprotective in antibiotic producing strains of bacteria and confers resistance against the toxicity of the bacterium's own secondary metabolites through methylation of their ribosomes by genome-encoded methyltransferases (Kojić et al. 1996). In the presence of S-adenosyl-L-methionine (SAM), post-transcriptional methylation of specific nucleobases of rRNA in the aminoglycoside binding site of these strains blocks binding of these antibiotics and maintains faithful protein synthesis. Recently, six distinct but related plasmid-borne 30S rRNA methyltransferases that confer resistance have been identified in clinical bacterial strains: ArmA (Galimand et al. 2003), RmtA (Yokoyama et al. 2003), RmtB (Doi et al. 2004), RmtC (Wachino et al. 2006), RmtD (Doi et al. 2007), and RmtE (Davis et al. 2010). All methylate N7 of G1405 and thereby confer resistance to the 4,6-disubstituted aminoglycosides (Liou et al. 2006; Perichon et al. 2007). These six methyltransferases confer no known physiological advantage in the absence of the antibiotics, in contrast to the many endogenous housekeeping methyltransferases that are important for the structure and function of mature rRNA.

Housekeeping methyltransferases play a role in prokaryotic ribosome maturation and are synchronized with rRNA processing, ordered binding of ribosomal proteins, and conformational changes (Ofengand and Del Campo 2004; Andersen and Douthwaite 2006; Connolly et al. 2008). In 16S rRNA, many of the base modifications are clustered in the phylogenetically conserved and functionally essential decoding region of the 30S subunit (Berk et al. 2006). The aminoglycoside binding pocket and G1405, the target of ArmA, are located within the A-site of the decoding region and are flanked by several methylated nucleotides: C1402, C1407, U1498, A1518, and A1519. This tight clustering very likely imposes an evolved ordering of methylation steps to avoid methyltransferase crowding and possible obstructive stereoelectronic effects to downstream methylations introduced by the modifications. Such constraints would require synchronization of ArmA with the other housekeeping methyltransferases. The recent observation that expression in E. coli of an ArmA ortholog from an antibiotic producing strain decreases the methylation function of endogenous E. coli RsmF (Čubrilo et al. 2009) supports this inference. It is not known, however, if the presence of G1405 methylation and the absence of C1407 modification negatively affect ribosome function in the absence of gentimicin and related aminoglycosides.

Neither 16S rRNA alone nor 70S ribosome is a substrate for Arm/Rmt methyltransferases, while an assembled 30S subunit is methylated by purified ArmA in vitro (Liou et al. 2006). The sequestered position of the Arm/Rmt target nucleotide G1405 in the crystal structure of the 30S subunit (Decatur and Fournier 2002) and the heavy traffic of modifying enzymes converging on this region, suggest that conformational rearrangement of 16S rRNA is required for these enzymes to access their target nucleotide.

Since very limited biochemical data have been reported on the function of this group of enzymes, we undertook to determine their substrate specificity and to resolve the conundrum of how they interact with their target nucleobase. We find that very late steps in 30S ribosomal maturation are prerequisites for these methyltransferases to carry out their modification reaction, since the best substrate for ArmA is a highly structured ribonucleoprotein particle, very close in conformation to the fully mature 30S substrate. Conformational changes, either very late in the pathway of the 30S ribosomal subunit assembly or off-pathway, are necessary for Arm/Rmt methyltransferase to access the completely buried G1405.

RESULTS

Mg⁺² dependence of ArmA activity

ArmA is active in methylation of the wild-type 30S ribosomal subunit over a wide range of Mg⁺² concentrations (Fig. 1A). It is known that the conformational state of the 30S ribosomal subunits is sensitive to the concentration of Mg⁺² and NH₄Cl (Zamir et al. 1969, 1971). High Mg⁺² concentrations stabilize an active state of 30S, while low concentrations of Mg⁺² stabilize a translationally inactive conformation of the 30S subunits (Moazed et al. 1986). The broad plateau of maximal ArmA activity in the range of 10–15 mM Mg⁺² indicates that the enzyme methylates the translationally active form of the 30S subunit. However, ArmA retains one-half of its maximal activity at 4 mM Mg⁺², where the translationally inactive form of the 30S subunit predominates, suggesting that ArmA recognizes both forms of the 30S subunits.

FIGURE 1.

Effect of Mg⁺² on ArmA activity. In vitro methylation of wild-type 30S (A) and ksg^R 30S (B). Assays measured after 1 h contained either 10 pmol of wild-type 30S or ksg^R 30S, and 10 pmol enzyme, as described in Materials and Methods. The amount of [³H]-methyl groups incorporated was monitored by TCA and ethanol precipitation, followed by scintillation counting. Assays were performed in triplicate; error bars represent standard deviation.

Similar experiments were extended to the 30S subunits isolated from a ksgA null strain (Fig. 1B). The activity profile for the two forms of the 30S subunits remain similar (Fig. 1A,B), thus confirming that the inactive form of the 30S subunits is a suitable substrate. These data also suggest that prior modifications of A1518 and A1519 are not required for ArmA activity (see the following section).

KsgA effects on ArmA activity

The site of ArmA/Rmt methylation, G1405, makes numerous hydrogen bonds with A1518, one of two adjacent adenosines dimethylated by the rRNA methyltransferase KsgA (Schuwirth et al. 2005). To determine if ArmA/Rmt function is dependent upon KsgA function we made comparative measurements of ArmA activity using the 30S subunits from the wild-type E. coli and the 30S subunits from E. coli lacking KsgA and consequently methylation at A1518 and A1519. No differences in either the rate of incorporation or final level of methylation were observed in in vitro activity assays (Fig. 2), indicating that dimethylation of these adenosines or the presence of KsgA has little bearing on ArmA function.

FIGURE 2.

In vitro methylation of the 30S subunits. Time-course assays for ArmA. Diamonds indicate assays continaining 10 pmol of the wild-type 30S subunits and 10 pmol of ArmA. Squares indicate assays containing 10 pmol ksg^R 30S and 10 pmol ArmA. Assays containing 10 pmol of the 30S subunits and 10 pmol of enzyme were performed as described in Materials and Methods. The amount of [³H]-methyl groups incorporated was monitored by TCA and ethanol precipitation, followed by scintillation counting. Assays were performed in triplicate; error bars represent standard deviation.

To test this conclusion in vivo we relied on gentamicin sensitivity as a reporter of RmtA function in cells either expressing KsgA or lacking KsgA. RmtA was expressed via an arabinose inducible plasmid at two different levels of arabinose. We observed a 2X–8X reduction in the MIC of gentamicin when KsgA is not present (Table 1). This observation could be interpreted to mean that the absence of KsgA in some way reduces the ability of RmtA to methylate G1405 or that the methyl groups transferred by KsgA directly reduce gentamicin sensitivity of the cells. To test the latter hypothesis, we compared gentamicin MICs for wild-type E. coli and for the ksgA null strain in the absence of RmtA. We found that cells are 4X more sensitive to gentamicin when KsgA is absent, consistent with a direct connection between gentamicin activity and the presence of methyl groups at A1518 and A1519. We further conclude that ArmA/Rmt function is not strongly influenced by the prior methylation of A1518 and A1519, either in vivo or in vitro.

TABLE 1.

In vivo effect of RmtA and ΔksgA deletion on gentamicin cytotoxicity

To test whether RmtA can overcome this increased gentimicin sensitivity in the ΔksgA mutant, we did serial dilutions of E. coli carrying the RmtA plasmid and monitored cell growth of wild-type and ΔksgA strains in the presence of 64 μg/mL of gentamicin (Fig. 3). At high RmtA (0.2% arabinose), the two strains showed only a slight difference in gentimicin sensitivity. However, at low RmtA (0.05% arabinose) we observed a pronounced growth defect in the ΔksgA strain compared to the parental strain. We conclude that RmtA can overcome the inherent increased gentimicin sensitivity in the ΔksgA strain, but only at high levels.

FIGURE 3.

Overexpression of RmtA in the wild-type and ΔksgA cells from the Keio collection. Tenfold serial dilution of each cell type was plated on LB plates containing 100 μg/mL ampicillin, 64 μg/mL gentamicin, and either 0.2% (A) or 0.05% (B) arabinose at 37°C.

Site directed hydroxyl radical probing

To identify the site of interaction between ArmA and the 30S subunit, we used hydroxyl radical cleavage probing initiated from Fe(II)-modified ArmA proteins bound to the E. coli 30S subunits (Fig. 4). This technique was used to orient the methyltransferase KsgA on the 30S subunit (Xu et al. 2008). However, directed probing failed to produce readily interpretable cleavage patterns for ArmA, as would be expected for a single, strong binding complex (Supplemental Figs. 1, 2). A similarly ambiguous 16S rRNA footprint for the interaction of Sgm, an ArmA ortholog, with the 30S subunits was recently reported (Husain et al. 2010). An attempt to measure a binding constant between ArmA and the 30S subunits failed in our hands (data not shown). Taken together these reports suggest that ArmA family members bind 30S weakly, and possibly nonselectively.

FIGURE 4.

Directed hydroxyl radical probing of the 30S subunit from ArmA. (A) Ribbon structure (DeLano 2002) of ArmA (PDB file: 3FZG) (Schmitt et al. 2009) showing specific positions of cysteine residues mutated as previously described in Materials and Methods. (B) Directed hydroxyl radical cleavage sites from Fe(II)–ArmA shown as inside boxes on the secondary structure of 16S rRNA (Cannone et al. 2002). (C) Directed hydroxyl radical cleavage sites shown on the three-dimensional structure of the E. coli 30S structure (Schuwirth et al. 2005).

ArmA methylation of variant forms of the 30S ribosome subunits and 16S rRNA

ArmA can utilize fully formed 30S subunits as a substrate, but not 16S rRNA when fully depleted of ribosomal proteins (Liou et al. 2006). As an attempt to discover a substrate with intermediate complexity we used the well-known technique of washing the 30S subunits with varying concentrations of LiCl to deplete an ever increasing number of ribosomal proteins (Itoh et al. 1968). To our surprise, even at the low LiCl concentration of 0.75 M, we witnessed a sharp decrease in the capacity of the resulting particles to serve as a substrate of ArmA (Supplemental Fig. 3A). Two-dimensional polyacrylamide gel electrophoresis revealed that no single or group of ribosomal proteins is removed under such mild wash conditions (Supplemental Fig. 3B–E). Agarose gel electrophoresis confirmed that 16S rRNA remained intact at all concentrations of LiCl used (data not shown). Both results demonstrate that loss of methylation activity was not the result of a compositional change to the treated 30S subunit. Therefore, we are left to conclude that a 0.75 M LiCl wash induces a conformational change in 16S rRNA or within the set of ribosomal proteins that renders the complex unsuited for ArmA methylation. Further, annealing this particle at 42°C and cooling back to 37°C did not rescue its ability to serve as an ArmA substrate (Supplemental Fig. 4).

We also tested whether ArmA can methylate the pre-30S subunits, which do not assemble into 70S ribosome. These subunits contain 17S rRNA, a precursor of 16S rRNA. The fully formed 30S ribosomal subunits were separated from the preribosomal subunits as described in Materials and Methods, following the procedure of Connolly et al. (2008). For this experiment we used Keio collection cells where ksgA was deleted (Baba et al. 2006) to avoid possible inhibitory effects of bound KsgA on ArmA (see below; Connolly et al. [2008]). The pre-30S subunits were measured for methyl acceptor activity by ³H incorporation from ³H- SAM (Fig. 5) relative to a control of fully formed 30S ribosomal subunits prepared from 70S ribosomes isolated from the same ΔksgA strain. Our results show that these pre-30S subunits are suboptimal substrates for ArmA in vivo.

FIGURE 5.

Testing the pre-30S subunits as potential ArmA substrates. In vitro methylation of the pre-30S subunits and the 30S subunits from the 70S ribosomes at 37°C and 25°C. The amount of [³H]-methyl groups incorporated was monitored by TCA and ethanol precipitation, followed by scintillation counting. Assays were performed in triplicate; error bars represent standard deviation.

Competitors for ArmA binding and methylation of the 30S subunits

The 4,6-disubstituted aminoglycoside gentamicin binds to the A-site of the 30S ribosomal subunits, which is formed by the 16S rRNA and includes G1405 (Yoshizawa et al. 1998). It has been demonstrated that the 30S subunit methylated at G1405 cannot bind gentamicin (Liou et al. 2006). We measured in vitro ArmA activity toward the 30S subunit substrate that had been preincubated with various concentrations of gentamicin in both high [Mg⁺²] (8 mM) and low [Mg⁺²] (4 mM) reaction buffers. Our data show that gentamicin inhibits ArmA activity at both [Mg⁺²] concentrations, but it also suggests that the antibiotic binds to the low [Mg⁺²] form slightly better than to the high [Mg⁺²] (Fig. 6), since we observe a greater decrease in ArmA activity at a lower concentration of gentamicin in the presence of low [Mg⁺²].

FIGURE 6.

Effect of gentamicin on ArmA activity. In vitro methylation of wild-type 30S in the presence of low (4 mM) or high (8 mM) Mg⁺². Assays containing 10 pmol of the 30S subunits and 10 pmol of enzyme were performed as described in Materials and Methods. The amount of [³H]-methyl groups incorporated was monitored by TCA and ethanol precipitation, followed by scintillation counting. Assays were performed in triplicate; error bars represent standard deviation.

Since we have clear data for where the kasugamycin resistance methyltransferase KsgA binds to the 30S ribosomal subunit (Xu et al. 2008), we tested whether KsgA inhibits ArmA methylation as a possible indicator of where ArmA binds to the substrate. We measured ArmA methylation activity in the presence of KsgA at low (4 mM) Mg⁺² concentration, which stabilizes the translationally inactive conformation of the 30S subunits to which KsgA preferentially binds. To assure that the transfer of the methyl group measured in our assay is due to the ArmA activity and not KsgA, we used catalytically inactive histidine-tagged KsgA (E66A) (O'Farrell et al. 2004; Inoue et al. 2007) and 30S subunits prepared from the ksg^R strain described above. The ksg^R 30S subunits were preincubated with KsgA (E66A) for 10 min at 37°C prior to adding ArmA, and ³H-methyl group incorporation from labeled SAM was measured at intervals over a 2 h period (Fig. 7A). Our data show that in the presence of equimolar KsgA, methylation of translationally inactive (low [Mg⁺²]) 30S subunits by ArmA was reduced by 25%. In 10-fold excess KsgA, ArmA methylation activity was completely abolished (Fig. 7A). (As expected, in experiments to evaluate the inhibitory nature of KsgA the overall methylation level was lower than normal. This reduction reflects the lower Mg⁺² concentration used.) A model of the KsgA-30S ribosomal subunit complex shows the binding sites of KsgA on the 30S subunit to be along helix 44, where the ArmA target nucleotide G1405 lies (Xu et al. 2008), and would predict interference and mutual exclusivity of these two enzymes in binding at their target sites.

FIGURE 7.

In vitro methylation of 30S. (A) Time course competition assays of ArmA with KsgA E66A. Triangles indicate assays containing 10 pmol 30S ksg^R and 10 pmol of ArmA; diamonds indicate assays containing 10 pmol 30S ksg^R, 10 pmol of ArmA, and 10 pmol of KsgA E66A; squares indicate assays containing 10 pmol 30S ksg^R, 10 pmol of ArmA, and 100 pmol of KsgA E66A; circles indicate assays containing 10 pmol 30S ksg^R and 10 pmol of KsgA E66A. (B) Time course competition assays of ArmA with IF-3. Diamonds indicate assays containing 10 pmol 30S wild-type and 10 pmol of ArmA; squares indicate assays containing 10 pmol 30S wild-type, 10 pmol of ArmA, and 10 pmol of IF-3.

It has been shown that IF3 and KsgA compete for binding to the 30S subunit (Xu et al. 2008). Since KsgA and ArmA compete for the same site on the 30S subunit (Fig. 7A), we extended our competition experiments to determine whether the binding sites of IF3 and ArmA also overlap. We purified histidine-tagged IF3 as described in Materials and Methods and used it as a competitor in methylation assays for ArmA activity. The wild-type 30S subunits (10 pmol) were incubated with 10 pmol of IF3 in buffer containing 10 mM Mg⁺² for 10 min prior to the addition of ArmA (Fig. 7B). IF3 showed no effect on methylation by ArmA, from which we conclude that the common locus for IF3 and KsgA binding does not intersect that for ArmA (Dallas and Noller 2001; Xu et al. 2008). Confirmation of complex formation of IF3 with wild-type 30S was obtained by loading the sample onto a 10%–30% sucrose gradient whose fractions were then checked on SDS PAGE gel for the presence of IF-3 complex with the 30S subunits (data not shown).

DISCUSSION

The outlines of E. coli ribosome biogenesis were elaborated by Nomura and Nierhaus (Nomura 1973; Rohl and Nierhaus 1982), but new levels of complexity have been recognized in the overall process in recent years (Connolly and Culver 2009), particularly since the determination of high-resolution crystal structures of the ribosome and its subunits and complexes (Wimberly et al. 2000; Schuwirth et al. 2005). Transcription and processing of pre-16S rRNA is coordinated with the association of both ribosomal and transiently bound proteins, including RNA modification enzymes. While the overall pathway leading to the assembly of the mature ribosome is regulated and clearly ordered, there is evidence of parallel sequences of steps that converge downstream in the process (Talkington et al. 2005; Kaczanowska and Rydén-Aulin 2007).

The 16S rRNA undergoes covalent modification, primarily methylation, at a number of sites during maturation of the 30S ribosome subunit. For the methyltransferase RsmB, protein-depleted 16S rRNA can serve as an in vitro substrate (Tscherne et al. 1999). A larger number of methyltransferases, e.g., RsmC, RrmJ, KsgA (RsmA) (Ofengand and Del Campo 2004), RsmF (Andersen and Douthwaite 2006), RsmH, and RsmI (Kimura and Suzuki 2009) require a more complex substrate than the naked 16S rRNA for methylation to occur. In the E. coli 30S ribosomal subunit crystal structure (Schuwirth et al. 2005), many of the nucleotides that are methylated are clearly not accessible to modification enzymes, including those for which the mature 30S subunit serves as a substrate (Decatur and Fournier 2002). Some of these may be modified during the maturation process when subunit intermediates offer functional substrates to the modifying enzymes (Connolly et al. 2008). Others may reflect inherent lability in parts of the 30S subunit that permits transient access to sites that are otherwise sequestered in the most stable form of the assembly. The region of the 16S rRNA with the highest density of nucleotide methylation targets is around the mRNA decoding site, which is known to be structurally dynamic (Berk et al. 2006). This property is consistent with a requirement for local changes in structure of the 30S subunit in order for modification enzymes to bind and modify the mature 30S subunits.

ArmA has high specificity for the mature 30S subunit as a substrate and cannot methylate either the fully assembled 70S ribosome or protein-depleted 16S rRNA (Liou et al. 2006). The G1405 methylation target nucleotide in the structure of the mature 30S ribosomal subunit is inaccessible to an exogenous enzyme of the size of ArmA (Schuwirth et al. 2005; Husain et al. 2010) and the low k_cat value reported for ArmA (0.2 ± 0.06 min⁻¹) (Liou et al. 2006) suggests that the isolated, mature 30S subunits used in the in vitro assay of ArmA are not the optimal substrate of the enzyme. The reported k_cat value places ArmA in the group with RsmE (Basturea and Deutscher 2007), which also requires highly structured 30S subunits as substrates.

Our experiments tested whether other assembly and conformational states of the E. coli 30S subunit are more efficiently methylated by ArmA and sought to define the state of the target G1405 in the methylation-susceptible 30S ribosomal subunits. Our in vivo studies show that ArmA has a relatively narrow temporal window on the ribosome maturation pathway, but a wider window of conformational conditions in which to transfer the methyl group to G1405. We found that the enzyme could recognize and methylate, at lower levels than the intact 30S subunits, a free precursor of the 30S subunits from an E. coli strain lacking endogenous KsgA activity. This result, taken with the inhibition of ArmA by KsgA, indicates that ArmA operates after the release of KsgA from the free precursor 30S and the formation of the near-mature 30S subunits. Also, ArmA does not distinguish between the 30S forms, which have or lack methylated A1518 and A1519. In addition, the presence of IF3 does not affect ArmA activity, showing that the enzyme can transfer the methyl group to a fully assembled 30S subunit. Thus, ArmA binds to and methylates more than one conformation of 30S, but these conformations are close to that of the final 30S ribosomal subunit.

In the high [Mg⁺²] 30S subunit, the target G1405 nucleotide for Arm/Rmt methyltransferases is buried as a consequence of direct tertiary interactions with nucleotides A1518 and A1519 of helix 45 (Schuwirth et al. 2005). Binding of ArmA to this form of the 30S subunit requires a conformational rearrangement of 16S rRNA to allow access to G1405, most likely including the opening of the helix 44/helix 45 tertiary interactions. Since ArmA methylation activity is maximal for concentrations of Mg⁺² between 5 and 15 mM but persists at lower [Mg⁺²], the enzyme can methylate both the high [Mg⁺²] translationally active form and also the low [Mg⁺²] translationally inactive form of the 30S subunit. The differences between these two states do not affect ArmA access to G1405. This is reinforced by the abolition of ArmA activity in the presence of gentamicin at both high and low [Mg⁺²]. This result also implies that the gentamicin binding site, in which G1405 lies, is intact in both regimes of Mg⁺² concentration. ArmA contrasts with KsgA methyltransferase, which only binds the low Mg⁺² inactive form of the 30S subunits when methylating A1518 and A1519 (Desai and Rife 2006; Xu et al. 2008) and RsmE, which only methylates the higher Mg⁺² active form of the 30S subunits at U1498 (Basturea and Deutscher 2007). The differences in the two forms of the 30S subunit do not extend to the gentamicin binding site.

The inhibition of ArmA methylation by prebound KsgA at low [Mg⁺²] indicates overlap of the binding sites of KsgA and ArmA. However, this conclusion is qualified by the requirement for some conformational change in the 30S subunit for ArmA to access G1405. The conformational excursion for binding both KsgA and ArmA may be the same and each enzyme could be binding to a part of the 16S rRNA that is not accessible in the static, crystallographically determined the 30S subunit structure. A similar case is found in the sisomicin-gentamicin resistance methyltransferase Sgm, an ortholog of ArmA, which methylates G1405 in the A site and requires conformational change in order to access the buried target nucleotide (Husain et al. 2010).

We hypothesize that the ArmA family of 16S rRNA methyltransferases exploits a labile property of the mature 30S ribosomal subunit, rather than an intermediate on the pathway to subunit formation, to gain access to G1405. This lability almost certainly involves helices 44 and 45 and is consistent with the drop-off in ArmA activity at very high [Mg⁺²], where hyper-stabilization of the 30S subunit structure may be hindering the conformation change needed for ArmA to bind. Structural support for a significant conformation change in the helix 44/45 region comes from cryo-EM studies of the complex of RbfA with the 30S ribosomal subunits (Bylund et al. 1998; Xia et al. 2003).

MATERIALS AND METHODS

Cloning, expression and purification of ArmA

Acinetobacter baumannii pUCarmA1-ArmA construct was obtained from Dr. Yohei Doi of the University of Pittsburgh School of Medicine and was confirmed by sequencing. The armA gene was inserted into pET15b as an NdeI-XhoI fragment with an N-terminal His-tag coding region. The correct ArmA sequence was confirmed by the Nucleic Acid Research Core Facility at Virginia Commonwealth University.

ArmA was overexpressed in BL21 (DE3) cells transformed with pET15b-armA plasmid and grown in LB medium containing 100 μg/mL ampicillin at 37°C to an OD₆₀₀ of 0.6. Cell cultures were then induced with 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG), incubated for 4 h at 37°C, and harvested by centrifugation. Pellets were resuspended in lysis buffer (50 mM Tris pH 7.5, 300 mM NaCl, 10 mM imidazole, 10% glycerol, 1 mM TCEP) and broken with an Emulsiflex cell breaker (Avestin). Pellets were centrifuged at 15,000 rpm for 30 min to remove the cell debris and the lysate loaded onto a HiTrap Chelating column (Amersham) charged with Ni⁺². The column was rinsed with wash buffer (50 mM Tris pH 7.5, 300 mM NaCl, 10 mM imidazole, 10% glycerol, 1 mM TCEP) and the protein was eluted with elution buffer (50 mM Tris pH 7.5, 300 mM NaCl, 300 mM imidazole, 10% glycerol, 1 mM TCEP). This method yielded 9–11 mg/L of culture of ArmA.

Mutagenesis

Single mutant C115A ArmA and double mutants C115A/Y43C ArmA, C115A/Y59C ArmA, C115A/L65C ArmA, C115A/V190C ArmA, and C115A/G234C ArmA were produced using the QuickChange Mutagenesis Kit (Stratagene). The full sequence of all mutated genes was confirmed by sequencing. Mutated plasmids were transformed into BL21 (DE3) cells for protein expression and expression and purification of the mutant proteins were conducted as for wild-type ArmA.

Expression and purification of mutant E. coli KsgA and IF3

The pET25b-KsgA and pET28b-IF3 constructs were kindly provided by Dr. Heather O'Farrell and Dr. Matthew Hartman of VCU and were confirmed by sequencing. Proteins were expressed in BL21 (DE3) cells (Stratagene) grown at 37°C to an OD₆₀₀ of 0.6 in the presence of ampicillin (for KsgAE66A) or kanamycin (for IF3). Protein expression was induced with 0.1 mM IPTG and the cultures grown for a further 4 h. Cells were harvested and broken as for ArmA and KsgAE66A was purified on a HiTrap Ni⁺² chelated column equilibrated with 50 mM NaPO₄, pH 8.0, 300 mM NaCl, and 10 mM imidazole. KsgAE66A was eluted with the same buffer containing 250 mM imidazole. IF-3 was purified as previously described (Soffientini et al. 1994). Cells were lysed in 20 mM Tris- HCl, pH 7.7, 60 mM NH₄Cl, 10 mM MgCl₂, 10% glycerol, 5 mM BME, 0.1 mM PMSF, and 0.1 mM benzamidine and the cleared lysate loaded onto a HiTrap Ni⁺² column. IF-3 was eluted with 20 mM Tris-HCl, pH 7.7, 10 mM MgCl₂, 10% glycerol, 5 mM BME, 0.1 mM PMSF, 0.5 mM EDTA, and 0.3 M imidazole. All proteins were estimated to be >95% pure by SDS-PAGE analysis.

Preparation and purification of ribosomes and ribosomal subunits

The 70S, 50S, and 30S ribosome subunits were purified from E. coli MRE600. An E. coli strain lacking functional KsgA was constructed by growing cells on 200 μg/mL kasugamycin and selecting for loss of dimethylation at A1518 and A1519 (O'Farrell et al. 2006). The Ksg^R 30S subunits were prepared from Keio collection cells as previously described (Blaha et al. 2000) by sedimentation in a sucrose gradient (Maki et al. 2002). Clear lysate was loaded onto a 10%–40% sucrose gradient in buffer consisting of 40 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 100 mM NH₄Cl, and 1 mM DTE and spun in an SW28 rotor at 19,000 rpm for 17 h, at 4°C. The pre-30S subunits, 50S subunits, and 70S ribosomes were obtained by growing ΔksgA strains at 25 and 37°C. Clear lysate was loaded onto a 10%–40% sucrose gradient in a buffer containing 40 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 100 mM NH₄Cl, and 1 mM DTE) and spun in an SW28 rotor at 25,000 rpm for 16 h. The gradients were analyzed using a Biocomp Piston Gradient Fractionator with a Biorad Econo UV Monitor. Data were recorded using DataQ DI-158-UP data acquisition software. The purified 30S wild-type subunits were dialyzed into reaction buffer (40 mM Tris, pH 7.2, 40 mM NH₄Cl, 8 mM Mg(OAc)₂, 1 mM DTE) and stored at –80°C. The purified 30S ksg^R subunits were dialyzed into reaction buffer (40 mM Tris, pH 7.2, 40 mM NH₄Cl, 4 mM Mg(OAc)₂, 1 mM DTE) and stored at –80°C. The concentration of the 30S ribosomal subunits was estimated from the absorbance at 260 nm using an extinction coefficient of 1 OD at 260 nm per 67 pmol/mL of 30S.

In vivo effects of controlled expression of KsgA on ArmA activity

The Keio wild-type and ΔksgA cells were grown in LB media and ampicillin (100 μg/mL) at 37°C overnight. The saturated cultures were inoculated in LB media in the presence of either low arabinose (0.05%) or high arabinose (0.2%) and ampicillin (100 μg/mL) until they reached an OD₅₅₀ of 0.7–0.8. The cultures were diluted in 10-fold increments and 5 μL was spotted on LB plates containing 100 μg/mL ampicillin, 64 μg/mL gentamicin, and 0.05% (or 0.2%) arabinose. The plates were incubated at 37°C overnight.

LiCl salt washes of the 30S subunits

The protein-depleted 30S ribosomal subunits were obtained by treatment with increasing concentrations of LiCl according to Itoh et al. (1968). Briefly, the 30S subunits were diluted into the buffer (20 mM Tris-HCl, pH 7.8, 100 mM Mg(OAc)₂, and 1 mM DTE) at final LiCl concentrations of 0, 0.5, 0.75, 1, 1.25, 1.5, 1.75, or 2 M. The samples with concentrations of LiCl higher than 1 M were incubated overnight at 4°C; others were prepared just prior to centrifugation. The LiCl-washed particles were pelleted by centrifugation at 45,000 rpm in a Beckmann 70Ti rotor for 6.5h and a sucrose sedimentation profile was performed as described (Basturea and Deutscher 2007). Pellets were resuspended in buffer containing 20 mM Tris-HCl, pH 7.8, 0.1 mM Mg(OAc)₂, and 1 mM DTE and isolated on a 10%–30% sucrose gradient by centrifugation in a Beckmann SW28 rotor at 25,000 rpm for 17 h. The peak fractions were collected, concentrated by centrifugation through an Amicon Ultra 30K cutoff filter, and stored at −80°C.

Two-dimensional analysis of the 30S and 30S-like subunits

The peak containing LiCl-treated RNA–protein complex was also examined on 2D gels, following a protocol derived from Geyl et al. (1981) and modified by Lyon and Culver (unpubl.). In brief, the LiCl-treated subunits were isolated and precipitated using sodium acetate and ethanol. The samples were incubated at −20°C overnight and centrifuged at 10,000 pm for 30 min at 4°C. The pellets were then resuspended in 50 μL deionized H₂O. Mg-acetate/acetic acid was added to the mixture and centrifuged at 10,000 rpm for 30 min at 4°C. To the supernatant, five volumes of acetone were added, the samples vortexed and precipitated at −20°C, centrifuged at 10,000 rpm for 30 min and the pellets resuspended in deionized H₂O. The samples were run on 2D gels (Biorad) and ribosomal proteins were stained using SYPRO Ruby protein gel stain (Biorad).

The assembly of 70S from the 30S LiCl-treated subunits and the 50S ribosomal subunits was performed as follows: 30 pmol of the 30S subunits (MRE600) from different LiCl-treated concentrations (0, 0.5, 0.75, and 1.0 M LiCl) were incubated with 60 pmol of the 50S ribosomal subunits (MRE600) at 37°C for 1 h in buffer containing 40 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 100 mM NH₄Cl, and 1 mM DTE. The samples were loaded onto a 10%–40% sucrose gradient in a buffer containing 40 mM Tris-HCl, pH 7.5, 10 mM MgCl₂, 100 mM NH₄Cl, and 1 mM DTE and spun in an SW28 rotor at 19,000 rpm for 17 h. The gradients were analyzed as described above.

In vitro ArmA methyltransferase assay

The in vitro assay for ArmA-catalyzed methylation was adapted from Poldermans et al. (1979). The total reaction volume of 50 μL contained 40 mM Tris, pH 7.2, 40 mM NH₄Cl, 8 mM (or 4 mM when indicated) Mg Acetate, 1 mM DTE, 0.02 mM ³H-methyl-SAM (780 cpm/pmol, Perkin Elmer Life and Analytical Sciences), 10 pmol 30S subunits, and 10 pmol of ArmA (plus 10 or 100 pmol of KsgA or 10 pmol of IF3 where indicated). The 30S subunits in the appropriate buffer were prewarmed at 42°C before the reactions started. In the aminoglycoside competition experiment, different concentrations of gentamicin (Invitrogen) were preincubated with the 30S subunits at 37°C for 5 min in the buffers described above. At the designated time points, 10 μL of 100 mM unlabeled SAM (Sigma–Aldrich) was added to quench the reaction. The quenched reaction solutions were placed onto DE81 filter paper (Whatman), washed with ice-cold 5% TCA, and rinsed with 100% ethanol. The filters were dried for 1 h and then placed into 3 mL of scintillation fluid, and counted. For the competition assay of ArmA and KsgAE66A (or IF3), KsgAE66A (or IF-3) was preincubated with the corresponding 30S subunits at 37°C for 10 min before ArmA was added to the reaction.

Preparation of Fe(II)-BABE derivatized ArmA

Fe(II)-BABE complex was formed as previously described by Xu et al. (2008). In brief, 3 nmol ArmA (cysteineless or cysteine-substituted) were incubated with 70 nmol Fe(II)-BABE in BABE modification buffer (1 M KCl, 80 mM K⁺-Hepes, pH 7.6, 0.01% Nikkol) in 100 μL total volume at 37°C for 30 min. Excess Fe (II)-BABE was removed from derivatized ArmA by centrifugation at 5000 rpm in Microcon (10,000D cutoff) centrifugal concentrators at 4°C followed by additional washes with 400 μL BABE modification buffer.

Formation of the ArmA/30S subunit complexes

The formation of the ArmA/30S complex was done essentially as described (Xu et al. 2008). In brief, the 30 pmol submethylated 30S subunits were incubated with 150 pmol Fe(II)-ArmA in buffer containing 40 mM Tris, pH 7.2, 40 mM NH₄Cl, 8 mM MgOAc, 1 mM DTE in 100 μL total volume at 37°C for 1 h. Unbound Fe(II)–ArmA proteins were removed using Sephacryl S-200 spin columns at 2000 rpm for 3.5 min.

Primer extension analysis

Directed hydroxyl radical probing was done essentially as described by Culver and Noller (2000). Specific ArmA–30S complexes, 4 μL Fe-EDTA solution, 2 μL 5% hydrogen peroxide and 2 μL 500 mM ascorbic acid were added in a total volume of 100 μL. Reactions were performed on ice and quenched by adding 40 μL 0.1 M thiourea. Each experiment was performed at least three times to ensure that data were reproducible. Extracted 16S rRNA was analyzed by primer extension as previously described by Stern et al. (1988). Cleavage intensities were compared to the intensity of control sequencing lanes.

SUPPLEMENTAL MATERIAL

Supplemental material can be found at http://www.rnajournal.org.