Structural and Functional Characterization of Rv2966c Protein Reveals an RsmD-like Methyltransferase from Mycobacterium tuberculosis and the Role of Its N-terminal Domain in Target Recognition

Abstract

Nine of ten methylated nucleotides of Escherichia coli 16 S rRNA are conserved in Mycobacterium tuberculosis. All the 10 different methyltransferases are known in E. coli, whereas only TlyA and GidB have been identified in mycobacteria. Here we have identified Rv2966c of M. tuberculosis as an ortholog of RsmD protein of E. coli. We have shown that rv2966c can complement rsmD-deleted E. coli cells. Recombinant Rv2966c can use 30 S ribosomes purified from rsmD-deleted E. coli as substrate and methylate G966 of 16 S rRNA in vitro. Structure determination of the protein shows the protein to be a two-domain structure with a short hairpin domain at the N terminus and a C-terminal domain with the S-adenosylmethionine-MT-fold. We show that the N-terminal hairpin is a minimalist functional domain that helps Rv2966c in target recognition. Deletion of the N-terminal domain prevents binding to nucleic acid substrates, and the truncated protein fails to carry out the m²G966 methylation on 16 S rRNA. The N-terminal domain also binds DNA efficiently, a property that may be utilized under specific conditions of cellular growth.

Introduction

Mycobacterium tuberculosis is one of the most infectious pathogens worldwide. It causes about 2 million deaths and 10 million cases of new infections every year (1, 2). The major hurdle in the treatment of the disease is the emergence of drug resistant M. tuberculosis strains. Multiple drug resistant and extremely drug resistant strains of Mycobacterium are becoming more and more prevalent (3, 4). Understanding the molecular mechanism of drug resistance is an area of intense research now.

Ribosome is one of the most common targets of antibiotics in the cell. Mutations in different components of ribosome are responsible for drug resistance. Several mutations in 23 S and 16 S rRNA genes, rRNA methyltransferases, and ribosomal proteins in different mycobacterial strains have been shown to be responsible for resistance to different drugs. An A1400G mutation in 16 S rRNA (rrs) gene leads to resistance to amikacin-kanamycin in M. tuberculosis strain ATCC 35827 and 13 other clinical isolates (5). Substitutions of C1401A/T or G1483T led to kanamycin resistance (6). Mycobacteria belonging to M. tuberculosis complex have intrinsic resistance to macrolides (7, 8). The resistance has been attributed to erm genes, which are 23 S rRNA methyltransferases (9). Inactivation of tlyA gene, which encodes a 2′-O-methyltransferase and methylates nucleotide C1409 of 16 S rRNA and C1920 of 23 S rRNA of M. tuberculosis makes it resistant to capreomycin and viomycin (10). Loss of GidB, which is a 7-methylguanosine methyltransferase specific for 16 S rRNA, confers streptomycin resistance to the bacteria (11).

rRNA methyltransferases appear to be very strong candidates for drug targeting against mycobacteria, but surprisingly, there is very little information available about them. Erm, TlyA, and GidB are the only methyltransferases that have been studied in mycobacteria (9–12). In Escherichia coli, 10 methylated nucleotides in 16 S rRNA and 12 in 23 S rRNA are known with more than 20 rRNA methyltransferases responsible for these modifications (13). Alignment of 16 S rRNA sequences of E. coli and M. tuberculosis show very high conservation with as many as 9 of 10 methylated nucleotides being conserved. The very fact that each nucleotide in E. coli is methylated by a specific methyltransferase necessitates identification of the different methyltransferases in mycobacteria. Here, we have identified and characterized Rv2966c as an RsmD-like methyltransferase from M. tuberculosis by genome data base search, structural, and biochemical analysis. RsmD has m²G966 methyltransferase activity for 16 S rRNA in E. coli (14). We have shown that rv2966c can complement rsmD-deleted E. coli cells. We have also determined its crystal structure to 1.9 Å resolution, which reveals that Rv2966c is a structural homolog of RsmD. We also show for the first time the role of a minimalist N-terminal domain in recruiting Rv2966c to ribosome to carry out its methyltransferase function.

EXPERIMENTAL PROCEDURES

Recombinant Rv2966c Constructs: Cloning, Expression, and Purification

Open reading frame encoding Rv2966c was PCR-amplified from H37Rv genomic DNA using forward and reverse primers. Forward primer introduced a BamHI site at the start codon, whereas the reverse primer introduced an XhoI site 3′ to the stop codon. The PCR product was digested with BamHI and XhoI and cloned in similarly digested pET28-His₁₀-Smt3 vector to give the expression plasmid pMTase1. pMTase1 encodes full-length Rv2966c polypeptide fused to an N-terminal His₁₀-Smt3 tag. The tagged protein was expressed by growing BL21 (DE3) cells transformed with pMTase1 in the presence of 50 μg/ml kanamycin. Protein expression was induced by growing the cells at 37 °C to A₆₀₀ ≈ 0.5 and chilling them on ice for an hour followed by the addition of ethanol to 2% and 0.2 mm isopropyl 1-thio-β-d-galactopyranoside. Cells were further grown with constant shaking at 17 °C for 18 h, harvested by centrifugation, and stored at −80 °C. All the subsequent steps were performed at 4 °C. The cell pellet was thawed on ice and resuspended in buffer A (10 mm HEPES, pH 7.0, 10% glycerol, 0.1% β-mercaptoethanol, and 300 mm NaCl). The cells were lysed by sonication and centrifuged to remove insoluble material, and the supernatant was applied on a nickel-nitrilotriacetic acid-agarose (Qiagen) column pre-equilibrated with buffer A. The column was then washed with 20 column volumes of buffer B (10 mm potassium phosphate buffer, pH 7.4, 10% (v/v) glycerol, 0.1% (v/v) β-mercaptoethanol, 300 mm NaCl, 40 mm imidazole), and bound proteins were eluted with 3 column volumes of buffer B containing 250 mm imidazole. The eluted protein was mixed with Smt3-specific protease Ulp1 (Ulp1:protein ratio was 1:500) and incubated at 4 °C overnight to cleave the His₁₀-Smt3 tag. Tag-free protein was recovered in flow-through by passage of the mixture over the nickel-nitrilotriacetic acid-agarose column. Recombinant wild type Rv2966c (WT) thus purified was dialyzed against buffer C containing 10 mm potassium phosphate buffer, pH 7.4, 5 mm EDTA, 100 mm NaC1, 10% (v/v) glycerol, 0.1% (v/v) β-mercaptoethanol. The dialysate was applied onto an ion-exchange column to remove nucleic acids. Purified protein was pooled, dialyzed against buffer D (10 mm potassium phosphate buffer, pH 7.4, 5 mm EDTA, 10% (v/v) glycerol, 0.1% v/v β-mercaptoethanol, 50 mm NaCl), and concentrated using Amicon ultra 3-kDa molecular cutoff filter units and stored at 4 °C until further use.

N-terminal-truncated Rv2966c was expressed by PCR amplifying the truncated reading frames from pMTase1. Forward primers were designed to amplify the reading frame from the 19th or 24th amino acid residues of Rv2966c reading frame. The primers introduced a BamHI site at initiation of the reading frames. The reverse primer used was the same as used to amplify full-length Rv2966c. PCR products were digested with BamHI and XhoI and inserted between BamHI and XhoI sites of pET28-His₁₀-Smt3 to give the expression plasmid pMTase4 and pMTase5. The proteins were expressed and purified from E. coli BL21 (DE3) following the same procedure as described above for the full-length Rv2966c, resulting in truncated proteins Δ18RM and Δ23RM lacking residues 1–18 or 1–23, respectively.

Construct for Determining in Vivo Activity of Full-length and Truncated Rv2966c

In vivo activity of full-length and truncated proteins was determined by expressing them under an arabinose promoter in the pBAD24 vector-containing ampicillin selection marker (15). Plasmids pMTase1, pMTase4, and pMTase5 were digested with NcoI and XhoI, and the fragments containing the genes along with the fused N-terminal His₁₀-Smt3 tag were cloned between NcoI and SalI sites of pBAD24, giving the expression constructs pMTase3, pMTase6, and pMTase7, respectively. A control plasmid pBSmt2 was generated by digesting pET28-His₁₀-Smt3 vector with NcoI and XhoI and ligating the resulting 405-bp fragment between the NcoI and SalI sites of pBAD24. The control plasmid hence expresses His₁₀-Smt3 protein only without any additional fusion partner.

A construct of Rv2966c lacking positively charged residues in the NTD³ was made by mutating Arg-3, Arg-12, and Arg-13 in the NTD to Ala in pMTase3. Mutagenesis was carried out using the QuikChange site-directed mutagenesis kit (Stratagene). In the first cycle, both Arg-12 and Arg-13 were mutated, and the resultant construct was used as a template to make the third mutation, R3A, to give the plasmid pMTase8. The presence of the desired mutations was confirmed by DNA sequencing. pMTase8 was used to monitor the in vivo activity of triple Arg mutant of Rv2966c, 3RM. pMTase8 was digested with NcoI and HindIII, and the fragment containing rv2966c (triple Arg mutant) gene along with the fused N-terminal His₁₀-Smt3 tag was subcloned between NcoI and HindIII sites of pET28-His₁₀-Smt3 giving pMTase3RM. pMTase3RM hence expresses 3RM, which was purified as wild type Rv2966c.

Complementation of rsmD-deleted E. coli Cells

m²G966 methyltransferase activity of full-length Rv2966c, truncated, and mutant proteins was determined by transforming rsmD-deleted E. coli KL16 cells (KL16ΔrsmD) (16) with pMTase3, pMTase6, pMTase7, pMTase8, or pBsmt2. Cells were selected against 100 μg/ml ampicillin and 50 μg/ml kanamycin. Protein was expressed by growing the cells at 37 °C to an A₆₀₀ ≈ 0.5 and inducing with 0.2% arabinose for 3 h. Cells were harvested and stored at −80 °C.

Total RNA was isolated from the cells (RNeasy kit, Qiagen) and annealed to ³²P 5′-end-labeled primer P1 complementary to sequence 980–996 of E. coli 16 S rRNA. Primer extension was carried out using avian myeloblastosis virus reverse transcriptase (Fermentas) as per the manufacturer's protocol. The reaction was terminated with formamide/EDTA, and products were analyzed on 15% denaturing polyacrylamide, 7 m urea gel and visualized by phosphorimaging. Oligonucleotides were also synthesized complementary to sequence 965–996, 966–996, 967–996, and 968–996 (labeled ME1, ME2, ME3, and ME4, respectively) and used as markers to analyze the length of product of primer extension. The product size of primer extension would be 30 nucleotides long (equal to ME3: 967–996) if G966 was methylated.

Circular Dichroism Spectroscopy Measurements

Far-UV CD spectra were collected on a Jasco J815 CD spectrometer in a quartz cuvette with a path length of 0.1 cm at room temperature. Ellipticity data were collected in the range of 250–195 nm. Concentration of the proteins (Rv2966c, Δ18RM, or Δ23RM) for CD measurements was kept at 0.5 mg/ml in 5 mm HEPES, pH 7.4, and 50 mm KCl. Each spectrum was recorded as an average of four scans. In all experiments, contributions of the buffer to the spectra were subtracted, and mean residue ellipticities were determined before plotting the spectra.

Electrophoretic Mobility Shift Assay (EMSA)

EMSA with DNA was carried out with pUC18 DNA (linearized with HindIII). Varying concentrations of Rv2966c, Δ18RM, Δ23RM, or 3RM as indicated in figure legends were incubated with DNA in a buffer containing 50 mm Tris-HCl, pH 7.9, 250 mm NaCl, 10% glycerol, 1 mm EDTA, 0.5% Triton-X-100 for 60 min at 30 °C. For EMSA with RNA, RNA purified as above (RNeasy kit, Qiagen) was similarly incubated with Rv2966c, Δ18RM, Δ23RM, or 3RM in an identical buffer as for the DNA binding experiments. The protein-nucleic acid complex was analyzed on 0.8% agarose gel by ethidium bromide staining.

Western Blot Analysis

rsmD-deleted E. coli KL16 cells (16) transformed with pMTase3, pMTase6, pMTase7, or pMTase8 were grown at 37 °C to an A₆₀₀ ≈ 0.5 and induced with 0.2% arabinose for 1 h at 37 °C. Total cell extract was prepared from each of the induced cells and fractionated on 12% SDS-PAGE. The proteins were transferred onto nitrocellulose membrane and probed with anti-His (Qiagen) and anti-DnaK (Stressgen) at 1:5000 dilution for both the antibodies.

Ribosome Purification and in Vitro Methylation Assay

Ribosome was isolated from both KL16ΔrsmD as well as KL16ΔrsmD complemented with pMTase3, as has been described before (17). In brief, the cells were grown in 500 ml of LB medium in the presence of a suitable antibiotic(s) at 37 °C to an A₆₀₀ ≈ 0.6 and then induced with 0.2% arabinose. Cells were further grown for 3 h with constant shaking at the same temperature, centrifuged, washed with Tris buffer saline containing 10 mm Mg(OAc)₂, and stored at −80 °C. Pellets were suspended in 2 ml of resuspension buffer (20 mm HEPES, pH 7.5, 60 mm NH₄Cl, 10.5 mm Mg (OAc)₂, 0.5 mm EDTA, and 10 mm β-mercaptoethanol) for each gram of wet cell pellet and lysed by sonication. Insoluble fraction was removed by centrifugation at 15,000 rpm for 40 min. The supernatant was treated with DNase I on ice for 30 min and further centrifuged at 31,000 rpm in a Beckman SW32 rotor for 4 h to pellet ribosomes. Ribosome pellet was suspended in 1 ml of extraction buffer (same as resuspension buffer but with 1 m NH₄Cl) and centrifuged again at 44,000 rpm for 12 h in a Beckman SW28 rotor on an 18% sucrose cushion. Ribosome pellets thus obtained were resuspended in the extraction buffer containing 2 mm Mg(OAc)₂ to disassemble the ribosomal subunits. The subunits were separated by centrifuging through a 10–30% sucrose gradient for 12 h at 31,500 rpm in a Beckman SW28 rotor. Fractions containing 30 S ribosomes were collected, concentrated through Amicon Ultra50K, dialyzed against buffer containing 20 mm Hepes, pH 7.4, 120 mm NH₄Cl, and 10% glycerol and stored at −80 °C until further use.

In Vitro Methylation

An in vitro methyltransferase assay was done using a nonradioactive colorimetric assay kit (G-Biosciences) with a slight modification of the protocol as suggested by the supplier. The methyltransferase reaction was first carried out in a 20-μl reaction mixture containing 100 ng of enzyme (Rv2966c, Δ18RM, or Δ23RM), 4 μg of purified 30 S ribosome subunit, and 100 μm S-adenosylmethionine (SAM) and then incubated at 37 °C for 1h. 10 μl of the reaction mixture was added to 90 μl of “master mix” without SAM as suggested by the supplier. This reaction was further incubated at 37 °C. The methyltransferase reaction resulted in an increase in S-adenosyl homocysteine due to transfer of methyl group from SAM to substrate (16 S rRNA). S-Adenosyl homocysteine was rapidly converted to urate and hydrogen peroxide by coupled enzymatic reactions. The rate of production of hydrogen peroxide was monitored colorimetrically with 3,5 dichloro-2-hydroxy-benzene sulfonic acid at 510 nm to measure the methyltransferase reaction. A parallel control reaction to subtract background was done where methyltransferase assay was done in absence of SAM.

Protein Crystallization, Data Collection, and Structure Refinement

Rv2966c was crystallized by hanging drop vapor diffusion method by mixing 4 μl of 17 mg/ml Rv2966c (in 10 mm potassium phosphate buffer, pH 7.4, 5 mm EDTA, 10% (v/v) glycerol, 0.1% v/v β-mercaptoethanol, 50 mm NaCl, 100 μm SAM) and 2 μl of reservoir solution (100 mm Tris-HCl, pH 7.4, 3.0 m potassium acetate) at 24 °C. Long rod-shaped crystals were obtained in 2–3 days. The crystals were cryo-protected in a solution containing 25% ethylene glycol in addition to the crystal reservoir solution and flash-frozen in liquid nitrogen before data collection.

Diffraction data were collected at BM14 beam line at European Synchrotron Radiation Facility. The crystals belong to the space group P3₂21 and contain one molecule in the crystallographic asymmetric unit. Data were processed and integrated using MOSFLM (18) and scaled using SCALA (19). Data collection and refinement statistics are summarized in Table 1.

TABLE 1.

Data collection and model refinement statistics

Crystal and data collection
Space group	P32 2 1
Cell dimensions	a = b = 54.78 Å, c = 100.04 Å, α = β = 90°, γ = 120°
Resolution (Å)a	27.39-1.9 (2.0-1.9)
Number of unique reflections	14,232 (955)
Completeness (%)	99.94 (100.0)
Multiplicity	10.7 (10.8)
Rmeas (%)b	9.3 (33.6)
〈I〉/σ(〈I〉)〉c	4.9 (2.3)
Refinement
Resolution (Å)	27.39-1.9 (1.95-1.9)
Number of reflections working/test	13,511/712
R-work (%)d	17.8 (18.2)
Rfree(%)e	22.1 (19.8)
Protein residues/atomsf	182/1,374
Alternate conformations	4
Water molecules	181
Other atoms	2 acetates
Wilson B-factor (Å2)	17.1
B-factor (Å2)
Protein atoms (chain)	15.5
Water molecules	27.3
r.m.s.d.
Bond lengths (Å)	0.015
Bond angles (°)	1.53
Ramachandran plot
Most favored regions (%)	96.7
Disallowed regions (%)	0.0

^a Numbers in parentheses correspond to the highest resolution shell.

^b R_meas as described in Diederichs and Karplus (46).

^c 〈I〉/σ(〈I〉)〉 = mean I_h over the S.D. I_h averaged over all reflections in a resolution shell.

^d R_work = Σ ‖F_o| − |F_c‖/Σ|F_o|, where |F_o| is the observed structure factor amplitude, and |F_c| is the calculated structure factor amplitude.

^e R_free: R_factor based on 5% of the data excluded from refinement.

^f Total number of protein atoms, including those in alternate conformations.

To solve the structure of Rv2966c by molecular replacement, polyalanine models of several bacterial putative methyltransferases α(PDB IDs 2FPO, 2FHP, 2IFT) were used as templates in Phaser (19). The best solution was obtained with the polyalanine model of putative methyltransferase of Enterococcus faecalis (PDB ID 2FHP) as the starting model. The initial model was further improved by iterative cycles of model building in COOT (20) and refinement using Refmac5 (21). Structure comparisons were performed with the help of DALI server (22). A final R-factor of 17.8% and R_free of 22.1% was obtained (Table 1). The final coordinates have been deposited with PDB with the accession code 3P9N.

RESULTS

Identification and Sequence Analysis of Rv2966c of M. tuberculosis

A search of H37Rv genome data base for nucleic acid methyltransferases identified Rv2966c as one of the putative methyltransferases. Rv2966c consists of an open reading frame of 188 amino acid residues. A BLAST search of Rv2966c against non-redundant protein sequences shows a strong sequence homology with several putative uncharacterized methyltransferases of Actinobacteria.

A search for presence of conserved domains in the protein identified residues 17–182 to contain the domain belonging to Class I SAM-dependent methyltransferase superfamily. The protein contains a SAM binding motif ⁵¹YAGSG⁵⁵ (equivalent to conserved motif I, (Y/F)XGXG of DNA methyltransferases) and a ¹¹⁹DPPY¹²³ motif (equivalent to conserved motif IV, (D/N/S)PP(Y/F) of DNA methyltransferases) (Fig. 1A) (23–25). The ⁵¹YAGSGALG⁵⁸ sequence of Rv2966c represents a general nucleotide binding motif in the SAM binding pocket, whereas the “DPPY” motif is present at the active site in several DNA methyltransferases and N5-glutamine methyltransferases (26–28).

FIGURE 1.

Rv2966c; overall structure and comparisons. A, sequence alignment of Rv2966c homologs is shown. Sequence alignment of Rv2966c and its structural homologs are as identified from DALI. Sequences are shown in same color as their structural counterparts in C. The conserved (F/Y)XGXG motif in the SAM binding region and DPP(Y/F) in the catalytic region are marked in gray boxes. The sequential arrangement of the Target recognition region, SAM binding region, and catalytic region classifies the protein to be similar to the ζ-class of DNA methyltransferases (Malone et al. (23). M.Tb., M. tuberculosis. B, a stereo diagram of the overall structure of Rv2966c shows the NTD (magenta), linker region (yellow), and the CTD containing the SAM-MT-fold (cyan). Rv2966c is a SAM-dependent methyltransferase. The conserved YAGSG motif in the SAM binding region is shown in black. A SAM molecule, shown in sticks, has been modeled in the putative SAM binding pocket formed toward the C terminus of the parallel β-strands. Residues at N terminus (Ser), C terminus (Leu), and at the break in linker region (Thr) are labeled. C, structure superposition of Rv2966c homologs is shown. Superposition of Rv2966c with H. influenzae methyltransferase (PDB ID 2IFT (green)), E. coli RsmD (PDB ID 2FPO (red)), E. faecalis methyltransferase (PDB ID 2FHP (orange)), Streptococcus pyogenes methyltransferase (PDB ID 2ESR (purple)), T. thermophilus methyltransferase (PDB ID 1WS6 (gray)) shows that all the structures have a same overall fold. Rv2966c is shown in same color notations as Fig. 1B.

The stretch of first 16 amino acid residues at the N terminus of Rv2966c did not show the presence of any known conserved domain. This stretch appears to be variable in length in different organisms, although several residues in this region appear to be conserved (Fig. 1A). As described later in text, the N-terminal region of Rv2966c appears to play a role in target recognition. Rv2966c hence belongs to a distinct family of nucleic acid methyltransferase resembling the ζ-class of DNA methyltransferases consisting of three distinct regions in the following sequential order: a target recognition region, SAM binding region, and last, the catalytic region (23).

The closest characterized match to Rv2966c was found to be RsmD from E. coli (14) (e value of 10⁻¹⁸) with which it has 67 identical and 92 similar amino acid residues with 16 gaps in the sequence alignment. Other structural homologs, for instance Haemophilus influenzae putative methyltransferase (PDB ID 2IFT, unpublished work, e-value 10⁻²⁰) and E. faecalis putative methyltransferase (PDBID 2FHP, unpublished work, e-value 10⁻¹⁴) among several others could be identified, although their biochemical functions have still not been characterized. RsmD is a 16 S rRNA-specific methyltransferase that methylates G966 at the N2 position (14). Rv2966c hence may be a protein with similar role specific for modification of nucleotide equivalent to m²G966 in 16 S rRNA in mycobacteria.

Crystallization and Structural Determination of Rv2966c

Rv2966c was expressed by transforming plasmid pMTase1 in E. coli BL21 (DE3) cells. The protein was purified to homogeneity from soluble extract as His₁₀-Smt3 fusion at the N terminus over nickel-nitrilotriacetic acid and ion exchange columns followed by cleavage of the tag by Ulp1 protease as described under “Experimental Procedures” (supplemental Fig. S1). The purified protein was set up for crystallizations. The structure of Rv2966c was determined to 1.9 Å resolution by molecular replacement using the coordinates of E. faecalis putative methyltransferase as the search model and refined to a final R-work of 17.8% and R_free of 22.1%. The final model of Rv2966c consists of 182 residues of the entire molecule including an additional Ser at the N terminus obtained as a cloning artifact. Four residues at the C terminus and three residues in the linker region were disordered and could not be included in the final model. The overall structure of Rv2966c consists of two distinct domains, a short NTD comprising residues 1–16 and a larger C-terminal domain (CTD) comprising residues 25–188 resembling the SAM-MT-fold present in several SAM-dependent methyltransferases (29) (Fig. 1B). The two domains are linked to each other by a short linker region (residues 17–24), which is partially disordered in the present structure.

Several methyltransferases have a domain in addition to the SAM-MT-fold that defines its substrate binding ability. In Rv2966c, only a short NTD in addition to the C-terminal SAM domain is present. The NTD is a β-hairpin comprising a small two-stranded β-sheet separated by a long loop of nine residues. The NTD appears to be the only region present in the protein structure that is likely to interact with substrate molecules. The NTD is connected to the CTD via a flexible linker region (Fig. 1B). The linker region may enable large movements of the entire NTD to interact with nucleic acid substrates. This small stretch of residues, however, is too short to search for structural homologs in DALI.

In contrast to the variable substrate binding domain, the CTD is a highly conserved structure consisting of the classical SAM-MT-fold found in several SAM-dependent methyltransferases (29). It consists of a central eight-stranded β-sheet flanked by α-helices on both sides. The first five strands of the β-sheet are parallel to each other and encompass the SAM binding site toward their C terminus end (Fig. 1B), whereas the remaining strands of the β-sheet are antiparallel.

A search for structural homologs of Rv2966c was carried out in DALI using the final coordinates of Rv2966c. Rv2966c shows similarity to methyltransferases belonging to different classes. The top structural matches were found with E. coli RsmD (PDB ID 2FPO) and several putative methyltransferases with structural similarities encompassing both the domains of Rv2966c (Fig. 1C, supplemental Table ST1). Structures of these homologs have been determined as part of structural biology consortium efforts, but the function of these proteins is unknown. Interestingly, the linker region of Rv2966c was found to be disordered in all the homologous structures, except in one subunit of E. faecalis and Thermus thermophilus structures. An analysis of the coordinate files for each structure also indicated the absence of a bound SAM in any of these structures. Other methyltransferases showing structural similarities with that of Rv2966c belonged to diverse families of methyltransferases, e.g. Precorrin-6Y C5,15-methyltransferase (Z-score 18.8 and r.m.s.d. of 2.2 Å for 159 residues), 23 S rRNA m⁵C1962 methyltransferase (Z-score 18.9 and r.m.s.d. of 2.6 Å for 170 residues), catechol-O-methyltransferase (Z-score 18.2 and r.m.s.d. of 2.6 Å for 164 residues). However, the structural similarities with other methyltransferases were limited to the core SAM-domain but not the target recognition domain and hence are not discussed further.

SAM Binding Site

Superposition of Rv2966c structure with other SAM-dependent methyltransferases helped identify the SAM binding pocket. Residues in the YAGSGALG nucleotide binding motif are part of the SAM binding site on one side, whereas the linker region is present at the other end of the SAM binding pocket. Although SAM was included in the crystallization conditions of Rv2966c, an unambiguous density for SAM in this pocket could not be identified. The absence of SAM in the binding region appears to have resulted in a partially disordered binding pocket (missing density for linker region). However, SAM could be modeled very well into this binding pocket, replacing a network of water molecules present in the SAM binding pocket (Fig. 2). As shown in the figure, residues of the SAM binding motif as well as the catalytic motif are well positioned to interact with the SAM molecule.

FIGURE 2.

Stereo view of SAM binding pocket of Rv2966c. A, a network of waters, marked W, occupies the SAM binding pocket. B. SAM can be modeled into this pocket into positions occupied by waters in the current structure. The putative H-bonding interactions of the modeled SAM are shown as dashed lines. An acetate molecule, ACT, is well placed to interact with the six-amino group of the adenosine moiety of SAM. Tyr-51 of the (Y/F)XGXG motif stacks against the adenosine of SAM molecule. Residues, Asp-119, Pro-120, and Pro-121 of the DPP(Y/F) motif are present close to the SAM binding pocket. The side chain of Tyr-122 of the DPP(Y/F) motif is disordered in the present structure.

In Vivo Activity; Rv2966c Complements E. coli RsmD

To determine the in vivo activity of Rv2966c, pMTase3 was transformed into rsmD-deleted E. coli KL16 strain (16), and the protein was induced with 0.2% arabinose at 37 °C for 3 h. Total RNA was isolated from the induced cells and used for primer extension analysis as described under “Experimental Procedures.” Primer extension by reverse transcriptase stops 3′ to a base methylated at positions involved in Watson-Crick base pairing like m²G or m²A. Hence if G966 is methylated at N2, primer extension will stop at C967. As can be seen from Fig. 3, extension of primer by reverse transcriptase stops at nucleotide C967 when the cells are transformed with pMTase3 (lane 4) but not when it is transformed with vector alone (lane 2). rv2966c is hence able to complement E. coli KL16ΔrsmD strain by methylating G966 at N2 position and inhibiting extension of primer at C967. An identical result is obtained in wild type strain (Fig. 3, lane 5), whereas there is read-through in the rsmD-deleted strain (Fig. 3, lane 1). The primer extension reaction carried out on total RNA purified from E. coli KL16ΔrsmD and methylated in vitro with recombinant Rv2966c resulted in read-through at the G966 position (Fig. 3, lane 3), suggesting that Rv2966c failed to methylate G966 at N2 in this reaction.

FIGURE 3.

Primer extension analysis for in vivo activity of Rv2966c. Primer extension reaction was carried out using avian myeloblastosis virus reverse transcriptase to monitor the in vivo methylation carried out by Rv2996c. Products were analyzed by running a denaturing polyacrylamide gel and visualized by phosphorimaging. Primer extension products in the different lanes correspond to reactions carried out on total RNA isolated from E. coli KL16ΔrsmD (lane 1), E. coli KL16ΔrsmD containing pBSmt2 expressing only the His₁₀-Smt3 tag (lane 2), an in vitro reaction of total RNA from E. coli KL16ΔrsmD incubated with purified Rv2966c (lane 3), E. coli KL16ΔrsmD complemented with pMTase3 (lane 4), and parent E. coli KL16 strain (lane 5). Lanes 6–9 contain ³²P-end-labeled oligonucleotide markers ME1, ME2, ME3, and ME4 corresponding to sequence 965–996, 966–996, 967–996, and 968–996 of lengths 32, 31, 30, and 29 nucleotides, respectively, to ascertain the size of extension products (see “Experimental Procedures” for more details). The position and size of ME1 and ME4 are indicated.

In Vitro Activity of Rv2966c

E. coli RsmD as well as other nucleotide-specific RNA methyltransferases, viz. RsmC and RsmE that modify m²G1207 and m³U1498 of 16 S rRNA, respectively (30, 31), lack in vitro methyltransferase activity in the absence of ribosomal subunits. To monitor the in vitro activity of Rv2966c, Rv2966c was incubated with 30 S ribosomal subunits in the presence of SAM, and the methyltransferase activity was monitored colorimetrically. As shown in Fig. 4, an increase in absorbance at 510 nm, indicating the methyltransferase reaction, occurs when Rv2966c is incubated with purified 30 S ribosome from the rsmD-deleted E. coli KL16 strain. No in vitro methyltransferase activity was observed when Rv2966c was incubated with the 30 S ribosomal subunit of E. coli KL16ΔrsmD complemented with pMTase3, indicating that RNA was already methylated in vivo.

FIGURE 4.

In vitro activity of Rv2966c. In vitro methyltransferase activity of Rv2966c was monitored colorimetrically as described under “Experimental Procedures.” A methyltransferase reaction was carried out by incubating Rv2966c (filled diamonds), Δ18RM (filled square), or Δ23RM (open triangles) with purified 30 S ribosome subunit from E. coli KL16ΔrsmD and 100 μm SAM at 37 °C for 1 h. No in vitro methyltransferase activity was observed when Rv2966c was incubated with the 30 S ribosomal subunit of E. coli KL16ΔrsmD complemented with pMTase3 (open squares).

Although the necessity of ribosome for in vitro methylase activity is unclear, it is likely that the small N-terminal domain of Rv2966c is incapable to binding efficiently to RNA and requires additional factors for stable binding to carry out the methylase reaction. Another possibility is that Rv2966c has a preference for the short stem and loop structure-containing G966 nucleotide position in 16 S rRNA. Whether the base recognition is sequence, RNA secondary structure, or ribosomal position-specific remains to be ascertained.

The N Terminus of Rv2966c Is the Target Recognition Region That Helps in Binding to Nucleic Acids

Rv2966c has a conserved stretch of about 16 amino acid residues at its N terminus that is not part of the known “methyltransferase” domain (Fig. 1). We surmised that this conserved stretch of amino acid residues might have an important role to play that is not yet determined. We addressed this by making two truncations of 18- and 23-amino residues from the N terminus and purifying the recombinant truncated proteins Δ18RM and Δ23RM from their respective expression vectors, pMTase4 and pMTase5, as described under “Experimental Procedures.”

Quaternary structures of the full-length (WT) and truncated proteins were determined by doing size exclusion chromatography (data not shown). All the three proteins eluted from the column as monomers. The overall stability and correct folded state of the proteins was confirmed by circular dichroism (Fig. 5A).

FIGURE 5.

Role of NTD in binding and activity. A, CD analysis is shown. CD of Rv2966c, Δ18RM, and Δ23RM (top, middle, and bottom panels, respectively) indicates secondary structures, which is suggestive of folded protein. B, EMSA for nucleic acid binding is shown. Binding of Rv2966c and truncated proteins to DNA and RNA was analyzed by running the reaction mixture on agarose gel and visualizing by ethidium bromide staining. Full-length Rv2966c (WT) binds RNA (upper panel) and DNA (lower panel) (lanes 2–4) in a concentration-dependent manner. Truncated proteins lacking the N-terminal hairpin, Δ18RM (lanes 5–7), and Δ23RM (lanes 8–10) fail to bind the nucleic acids. A control with no protein is indicated in lane 1 in both panels. In the reaction mix, 100, 200, or 400 ng of full-length (Rv2966c) or truncated proteins (Δ18RM and Δ23RM) were incubated with RNA at 30 °C for 60 min (see “Experimental Procedures” for more details). For DNA binding, a similar reaction was carried out with 50, 100, or 200 ng of full-length (Rv2966c) or truncated proteins (Δ18RM and Δ23RM) to monitor gel shift. Position and sizes of linear DNA markers (in kilobases) are indicated in lane M. C, in vivo activity of truncated proteins is shown. In vivo activity of Δ18RM or Δ23RM was monitored by primer extension analysis using avian myeloblastosis virus reverse transcriptase and analyzing the reaction products on denaturing polyacrylamide gel. The gel was visualized by phosphorimaging. Reactions were carried out with RNA purified from E. coli KL16ΔrsmD (lane 1), E. coli KL16ΔrsmD transformed with pMTase6 expressing Δ18RM (lane 2), E. coli KL16ΔrsmD transformed with pMTase3 expressing Rv2966c (full-length) (lane 3), and E. coli KL16ΔrsmD transformed with pMTase7 expressing Δ23RM (lane 4). Lanes 5–8 contain ³²P-end-labeled oligonucleotide markers ME1, ME2, ME3, and ME4 corresponding to sequence 965–996, 966–996, 967–996, and 968–996 and lengths 32, 31, 30, and 29 nucleotides, respectively, to determine the size of extension products. Position and size of ME1 and ME4 are indicated. D, expression level of wild type, deleted, and mutant Rv2966c are similar. Upper panel, a Western blot of total cell extract of E. coli KL16ΔrsmD cells expressing full-length Rv2966c (WT; lane 1), Δ18RM (lane 2), Δ23RM (lane 3), and 3RM (lane 4) shows an equivalent level of expression for all proteins. The blot was probed with anti-His. Lower panel, the same blot was probed with anti-DnaK, which was used as loading control.

To explore the role of NTD in nucleic acid binding, nucleic acid binding ability of Rv2966c was determined by EMSA (Fig. 5B). Both DNA and RNA were used for binding assay, and the protein-nucleic acid complex was analyzed on agarose gel by ethidium bromide staining (Fig. 5B). Full-length Rv2966c (WT) bound to RNA, as can be seen by the gradual slowing of the migration of rRNA band on the gel (Fig. 5B, upper panel, lanes 2–4), and the binding was concentration-dependent. A surprising result was the formation of complex between DNA and Rv2966c (Fig. 5B, lower panel, lanes 2–4). Deletion of the N-terminal hairpin prevents binding to DNA or RNA (Fig. 5B, lanes 5–10, upper and lower panels).

Binding of protein to DNA was much stronger than it was to RNA, as is evident from the much slower migration of DNA substrate on gel even with less amount of protein (Fig. 5B). The ability of Rv2966c to bind to DNA with a higher apparent affinity than RNA is surprising as no methyltransferase activity onto substrate DNA molecules by Rv2966c has been identified so far.

N-terminal-truncated Proteins Lack in Vivo and in Vitro m²G966 Methyltransferase Activity

As the truncated proteins did not bind to RNA, we wanted to determine their m²G966 methyltransferase activity. To monitor the in vivo activity of Δ18RM and Δ23RM, plasmids pMTase6 and pMTase7 were transformed into E. coli KL16ΔrsmD. Activity of the truncated proteins was determined by doing a primer extension assay on total RNA isolated from the transformed strain as described before. The truncated proteins did not show any m²G966 methyltransferase activity as seen in Fig. 5C, lanes 2 and 4), as there was no inhibition of primer extension by reverse transcriptase at C967 in these lanes, whereas there was clear termination at the same position with wild type (full-length) protein (Fig. 5C, lane 3), as expected. A Western blot was done with the total cell extract of E. coli KL16ΔrsmD cells expressing Δ18RM and Δ23RM (or 3RM) to find out whether the lack of complementation was because of low levels of expression of the truncated (or mutated) proteins. As seen in Fig. 5D, expression levels of all proteins were found to be similar. Incubation of the 30 S ribosomes from E. coli KL16ΔrsmD with Δ18RM or Δ23RM indicated that the truncated Rv2996c lacked in vitro methyltransferase activity as well (Fig. 4).

Mutant Protein 3RM Does Not Have in Vitro DNA Binding Activity

Electrostatic surface representation of nucleic acid-binding proteins is a good indicator of the binding region on the protein. An electrostatic surface representation of Rv2966c shows a region of positive potential for the N-terminal domain that may interact with negatively charged backbone of nucleic acid substrates (Fig. 6A). An electrostatic surface generated for the N-terminal-truncated protein (Fig. 6B), on the other hand, shows a marked reduction in the positively charged region that may lead to loss of interaction with 16 S rRNA. To probe the role of positively charged residues of the NTD in nucleic acid binding, a triple Arg mutant of Rv2966c, 3RM, was made mutating conserved arginine residues at positions 3, 12, and 13 to alanine. The mutated protein 3RM did not bind to both DNA and RNA in vitro (Fig. 7A) and had no m²G966 activity in vivo (Fig. 7B).

FIGURE 6.

Molecular surface representation of Rv2966c. A, the electrostatic surface potential of Rv2966c shows the NTD (marked by an arrow) to have a positively charged surface. B, an electrostatic surface generated for Rv2966c lacking the NTD (deleted region marked by an arrow) shows a reduction in positive surface on the protein. The electrostatic potential was calculated with the program APBS (Baker et al. (45)). The surface is colored with a blue to red gradient from +3.5 to −3.5 K_bT/e.

FIGURE 7.

Role of conserved arginines in binding and activity. A, shown is an EMSA for nucleic acid binding. Binding of Rv2966c and 3RM to DNA and RNA was analyzed by running the reaction mixture on an agarose gel and visualizing by ethidium bromide staining. 3RM fails to bind RNA (upper panel, lanes 5–7) or DNA (lower panel, lanes 5–7). A control with no protein is indicated in lane 1 in both the panels, whereas concentration-dependent binding of nucleic acids with Rv2966c (WT) is seen in lanes 2–4. In the reaction mix 100, 200, or 400 ng of Rv2966c (WT) or mutant protein 3RM were incubated with RNA at 30 °C for 60 min (see “Experimental Procedures” for more details). For DNA binding, a similar reaction was carried out with 50, 100, or 200 ng of Rv2966c (WT) or mutant protein (3RM) to monitor gel shift. 3RM, like Δ18RM and Δ23RM (Fig. 5), fails to bind both DNA and RNA. A minor difference in appearance of bands from Fig. 5 is due to gel running artifact. Position and sizes of linear DNA markers (in kilobases) is indicated. B, in vivo activity of mutant protein. In vivo activity of 3RM was monitored by primer extension analysis using avian myeloblastosis virus reverse transcriptase and analyzing the reaction products on denaturing polyacrylamide gel as described earlier. Reactions were carried out with RNA purified from E. coli KL16ΔrsmD (lane 1), E. coli KL16ΔrsmD transformed with pMTase3 expressing full-lengthRv2966c (WT) (lane 2), and E. coli KL16ΔrsmD transformed with pMTase8 expressing 3RM (lane 3). Position and size of markers ME1 and ME4, corresponding to sizes 32 and 29 nucleotides, are indicated.

DISCUSSION

Ribosomal RNAs play very important roles in the structure and function of ribosomes. Both 23 S and 16 S rRNAs contain several highly modified nucleotides (32). Tertiary structure of ribosomes has revealed that the modified nucleotides are located at important functional sites underlying their importance in ribosome biochemistry (33). Here we have identified Rv2966c from M. tuberculosis as an ortholog of RsmD of E. coli. We show that Rv2966c can complement RsmD in a mutant E. coli strain. Rv2966c can also use 30 S ribosomes of E. coli as substrate and methylates N2-G966 of 16 S rRNA. The ability of Rv2966c to use E. coli ribosomes as substrate shows conservation of not only the tertiary structure of ribosomes and the proteins but also the nature of interactions between RsmD and the components of ribosomes between the two organisms. There is not as much conservation of the primary structures between the two proteins, as has also been seen with other Rsm homologs (24), but the structure and the amino acid residues at the conserved motifs are highly similar. The open reading frame of Rv2966c consists of 188 amino acid residues, with a conserved type I Rossman-fold SAM binding domain from 17 to 182 amino acid residues. This domain contains SAM binding motifs ⁵¹YAGSG⁵⁵ and a conserved active site, ¹¹⁹DPPY¹²³, present in nucleic acid methyltransferases (23). Homologs of Rv2966c are present in all the mycobacterial species including Mycobacterium leprae, a strain of Mycobacterium that has lost 33% of its genome. It shows that the protein might play an important role in mycobacterial physiology. rsmD-deleted E. coli cells do not show any significant phenotype except a modest decrease in growth rate (14). This could be because of the presence of modifications in other nucleotides in the same loop region of 16 S rRNA at the P-site binding pocket of the ribosome. As these modified bases are positioned together at important sites in ribosomes, the absence of one modification does not make any significant difference in cellular growth under normal conditions. These modifications might give growth advantage to the bacteria under stress. M. tuberculosis and several other mycobacterial species spend much of their life cycle within human (host) macrophages, and Rv2966c (or its homologs) might have a role in bacterial survival at that time.

Rv2966c shows no detectable activity on RNA in the absence of ribosomes or when ribosomes are isolated from wild type E. coli or rsmD-complemented cells where the RNA is already methylated. Substrate specificity of RsmD has been attributed to the interaction between the protein and ribosomes (14). The NTD seems necessary and sufficient to enable this interaction with ribosomes as seen by the loss of methyltransferase activity both in vivo and in vitro in both the truncated and mutant proteins. The conserved arginines play an important role in binding to the negatively charged nucleic acid substrates. The role of arginines in the NTD was confirmed, as the mutant protein 3RM did not bind to nucleic acids in vitro and also lacked any in vivo m²G966 activity.

Rv2966c, like RsmD, has a very short N-terminal extension in the form of a short β-hairpin comprising a small two-stranded β-sheet separated by a long loop. Other rRNA methyltransferases like RlmG and RlmL have longer N-terminal extensions that have been shown to be involved in RNA substrate binding (34, 35). β-Hairpins play an important role in interactions with several nucleic acids. Interestingly, Rv2118c, an m¹A58 tRNA methyltransferase, contains three-tandem β-hairpins as the only target recognition region in addition to the SAM-MT-fold (36, 37), although the nature of interactions of the hairpin with tRNA is unclear. Flexible β-hairpins of architectural proteins HU and IHF also take part in stabilizing interactions with unusual bent DNA (38, 39). The flexible linker region of Rv2966c is likely to allow large conformational movements of the N-terminal β-hairpin domain to enable specific interactions with RNA near the P-site of the ribosome.

The NTD of Rv2966c is important for binding not only RNA but also DNA substrates. Somewhat surprisingly, Rv2966c shows stronger binding to DNA than RNA under given conditions. Truncated constructs of Rv2966c lacking the N-terminal 18- or 23-amino acid residues and the mutant protein 3RM did not show any significant binding to either DNA or RNA (Figs. 5B and 7A). The proteins did not show any m²G966 methyltransferase activity, confirming an important role of the N-terminal region in target recognition and nucleic acid binding. It is worth noting that the stretch of amino acid residues at the N terminus is fairly conserved in RsmD homologs in different organisms including Mycobacterium (although annotated reading frames of many of these homologs including MSMEG_2413 and ML1664 need to be corrected by including additional amino acid residues at the N terminus) (supplemental Fig. S2). The NTD of Rv2966c represents the smallest functional domain to the best our knowledge.

It is provocative to think that Rv2966c and its orthologs are also involved in methylating DNA. Rv2966c has a strong binding to DNA, and the complex is stable even at 65 °C (data not shown) and also contains the conserved DPPY motif at its active site, which is present in many other DNA methyltransferases (26, 27). Moreover, it seems unlikely for cells to maintain Rv2966c or its homologs at high metabolic cost just for one function that has already been shown to be dispensable without significant phenotypic changes in E. coli. A single methyltransferase methylating more than one substrate is not without precedence. Eukaryotic Dnmt2 has been shown to methylate both tRNA and DNA (40). Methylation of specific nucleotides on tRNA in Drosophila protects them from stress-induced cleavage (41). Even the bacterial exocyclic DNA methyltransferases specific for adenine or cytosine methylates the other purine nucleotide (42, 43).

m²G966 is located at P-site of ribosomes and binds to the tip of tRNA anticodon (44). P-site in bacteria is the target of many antibiotics, and modifications of nucleotides at the site have been shown to play a role in drug sensitivity of the organism. We have started to learn more about their roles with the solution of tertiary structure of ribosomes. It will be interesting to further explore the role of methylation at G958 (equivalent position to m²G966 of E. coli 16 S rRNA) and that of Rv2966c in M. tuberculosis drug resistance.