X-ray structure of an M. jannaschii DNA-binding protein: Implications for antibiotic resistance in S. aureus


  • Mj223 represents an NYSGRC target T88.

Introduction .

Bacterial resistance to commonly used antibiotics, such as the β-lactams and vancomycin, has become a major public health concern.1, 2 Research in the area of methicillin resistance revealed a number of auxiliary genes that augment the function of PBP2a, the gene product primarily responsible for methicillin resistance in Staphylococcus aureus.3 Although some of these auxiliary methicillin resistance proteins contribute to well-characterized biosynthetic pathways (e.g., cell wall metabolism), functions for several of these gene products remain elusive.4 This article describes the X-ray structure of a Methanococcus jannaschii protein that is related to one such auxiliary gene product, RUSA223 [Genbank Code: CAB60749].

Materials and Methods.

The S. aureus auxiliary gene product RUSA223 and its ortholog from M. jannaschii [Genbank Code: NP_248571, PSI-BLAST E-value = 10−25, sequence identity = 23%] were expressed in Escherichia coli BL21(DE3) with N-terminal His-tags. Following Ni-affinity chromatography, each protein was purified to homogeneity using an S Sepharose ion exchange column. Initial crystallization trials with the S. aureus protein yielded neither crystals nor promising leads. Dynamic light scattering studies demonstrated that RUSA223 was severely aggregated at submilligram concentrations in aqueous solution. Mj223 yielded crystals in a number of related conditions. Diffraction-quality crystals grown via hanging drop vapor diffusion against a solution containing 0.1 M Tris HCl 8.0, 30% PEG8000, and 10% dioxane belong to the hexagonal space group P65 (unit cell: a = 57.0 Å, c = 208.4 Å; two protomers per asymmetric unit). Selenomethionine-labeled Mj223, prepared as described above, yielded crystals under very similar conditions.

Following cryoprotection with added glycerol (25% v/v), X-ray data were collected at two wavelengths in the vicinity of the Se K-edge from a single crystal maintained at 100 K. Diffraction data were processed with DENZO/SCALEPACK5 (Table I), and the heavy atom substructure (four selenium sites/asymmetric unit) was elucidated with SnB.6 Se atomic positions were refined, and experimental phases were calculated with MLPHARE7 at 2.8 Å resolution (figure of merit = 0.64). Density modification yielded an interpretable electron density, which allowed tracing of the entire polypeptide chain (Met1-Asp151). This initial model was subjected to refinement with CNS8 against diffraction data obtained from an S-Met crystal, giving a final R-factor of 26.2% with R-free = 29.3% (Table I).

Table I. Crystallographic Data and Refinement Statistics
 λ1 (Se-Met peak)λ2 (Se-Met inflection)S-Met
  • a

    Rmerge = Σhkl Σi | I(hkl)i − <I(hkl)> | /ΣhklΣi < I(hkl)i>.

  • b

    Figure of merit calculated using MLPHARE.15

  • c

    Rcryst = Σhkl | Fo(hkl) − Fc(hkl) | /Σhkl | Fo(hkl) |, where Fo and Fc are observed and calculated structure factors, respectively.

  • d

    Computed with PROCHECK.16

  • Genbank code NP_248571

  • Crystal characteristics and data collection statistics

  • Unit cell dimensions a = 56.95 Å, b = 56.95 Å, c = 208.36 Å

  • Space group: P65; 2 moleculer per asymmetric unit

  • X-ray source: NSLS X12C beamline

Wavelength (Å)0.979350.979540.97962
Resolution (Å)30.0–2.830.0–2.830.0–2.8
Number of observations138373137127135799
Number of reflections945494699466
Completeness (%)99.599.799.7
(2.9–2.8 Å shell)97.499.699.9
Mean I/σ(I)38.437.546
(2.9–2.8 Å shell)7.96.910.3
R-merge on Ia0.0450.0480.038
(2.9–2.8 Å shell)0.2080.2110.146
Cutoff criteriaI < −3σ(I)I < −3σ(I)I < −3σ(I)
Figure of meritb0.68 (28.5–2.8 Å resolution) for 5484 reflections
Model and refinement statistics   
 Data set used in structure refinement S-Met 
 Resolution range 28.5–2.8 Å 
 Number of reflections 11,373 (10,317 in working set; 1056 in test set) 
 Completeness 96.8% (87.8% in working set; 9.0% in test set) 
 Cutoff criterion |F| > 0.0 
 Number of amino acid residues 151 
 Number of water molecules 121 
 Rcrystc0.262Rms deviations 
 Rfree0.293Bond lengths (Å)0.008
  Bond angles (°)1.38
  Luzzati error (Å)0.19
Ramachandran plot statisticsd   
 Residues in most favored regions  87.9%
 Residues in additional allowed regions  10.6%
 Residues in generously allowed regions  0.7%
 Residues in disallowed regions  0.7%
 Overall G-factor  0.2

Results and Discussion.

The N-terminal portion of Mj223 resembles a helix-turn-helix (HTH) winged-helix DNA binding motif,9 which suggests that it functions as a transcription factor. The C-terminal region of the protein is composed of two leucine-rich α-helices, which form a dimer with a neighboring protomer that is related by noncrystallographic two-fold symmetry [Fig. 1(a)]. Dynamic light scattering studies documented that the protein is indeed a dimer in solution (data not shown). The order of secondary structural elements in Mj223 is H1−H2−S1−H3-H4−S2−S3−H5−H6 (where H and S denote α-helices and β-strands, respectively). α-helices H5-H6 and H5′-H6′ (′ denotes the second protomer) dimerize to form an antiparallel four-helix bundle (buried surface area = 4551 Å2), which represents a novel dimerization motif not previously observed in dimeric transcription factors.

Figure 1.

Mj223 homodimer structure: (a) ribbon representation of the Mj223 dimer with labeled secondary structure elements, colored blue (α-helices) and orange (β-strands) in one protomer with the second protomer colored red; (b) ribbon representation of Mj223 docked with linear B-form DNA; (c) GRASP15 representation of the chemical properties of the solvent-accessible surface of Mj223, calculated using a water probe radius of 1.4 Å. Solvent-accessible surface has been color-coded for the calculated electrostatic potential (red and blue, respectively, represent electrostatic potentials <−8 and >+8 KBT, where KB is the Boltzmann constant and T is temperature). Calculations were performed with an ionic strength of zero and dielectric constants of 80 and 2 for solvent and protein, respectively. The surface of the winged-helix domain is both basic and hydrophobic.

The winged-helix domain is one of the most common DNA-binding motifs responsible for controlling transcription initiation in both bacteria and eukaryotes.9 We believe that the Mj223 dimer binds to a symmetric pair of half-sites by presenting α-helices H4 and H4′ to the major groove of duplex B-form DNA [Fig. 1(b)]. The calculated electrostatic potential for the putative DNA-binding surface is both basic and hydrophobic, and the separation of the two recognition α-helices is compatible with major groove binding [Fig. 1(c)]. The linear extent of the predicted DNA footprint is about 40 Å, which corresponds to two adjacent 7 base pair (bp) recognition sites related by two-fold symmetry [Fig. 1(b)].

Mj223 is structurally similar to the eubacterial transcription factors BmrR10 ( Bacillus subtilis, PDB Code 1EXI), and MarR11 (E. coli, PDB Code 1JGS). BmrR regulates expression of multidrug transporters when bacteria are challenged by lipophilic-cationic compounds.10 MarR is a transcription factor that is thought to regulate expression of the multidrug resistance proteins in the marRAB operon.11 BmrR, MarR, and Mj223 are structurally related in the sense that all three possess N-terminal winged-helix modules and C-terminal α-helical segments that support dimerization [Fig. 2(a)]. The C-terminal portion of BmrR is comprised of only one α-helix that forms an antiparallel coiled-coil structure followed by a “drug binding/sensing” domain. In contrast, MarR utilizes one N-terminal and two C-terminal α-helices for dimerization. Both modes of transcription factor dimerization are distinct from the strategy employed by Mj223.

Figure 2.

Comparison of Mj223, BmrR, and MarR: (a) ribbon representations of Mj223, BmrR, and MarR with the same color-coding as in Fig. 1(a); (b) superposition of the winged-helix domains of Mj223 (blue), BmrR (orange), and MarR (grey). The DNA recognition helices are labeled.

The DNA-binding domains of Mj223, BmrR, and MarR can be superimposed with pairwise root-mean-square deviations of 3.0–5.0 Å for equivalent α-carbon atoms [Fig. 2(b)]. Despite the similarities of Mj223, BmrR, and MarR, these proteins almost certainly make structurally distinct complexes with DNA. Superposition of the winged-helix domains yields quite different relative spatial dispositions for the dimerization segments. Both Mj223 and MarR appear to be capable of binding B-form DNA, with comparable footprints. BmrR binding, on the other hand, causes localized bp breaking and sliding that remodels the structure of the promoter and is thought to facilitate recruitment of σ-factor and RNA polymerase to genes that are upregulated upon exposure to lipophilic-cationic compounds.

Our structure of Mj223 provides insights into the function of the S. aureus RUSA223 gene products. We suggest that RUSA223 is a transcription factor that regulates expression of genes that contribute to antibiotic resistance. A recent comparison of archaeal and eubacterial transcription factors carried out by Koonin and coworkers suggests that HTH transcription factors are maintained through horizontal gene exchange, and that much of this exchange probably occurred before the divergence of the two bacterial kingdoms.12 This hypothesis is consistent with the structural similarity of BmrR (B. subtilis), MarR (E. coli), and Mj223 (M. jannaschii), suggesting that all three of these transcription factors belong to a general class of stress response proteins. Its is therefore likely that RUSA223 regulates a genetic response to exposure of antibiotic, which may be viewed as a chemical stressor.4

Bacterial HTH transcription factors can be subdivided into repressors and activators, which are structurally distinct. Those with an N-terminal HTH motif are usually repressors, whereas activators have predominantly C-terminal HTH motifs.13 This correlation suggests that RUSA223 and Mj223 are probably both repressor molecules. The X-ray structure of Mj223 provides additional insights into the functional role of RUSA223. Homology modeling of RUSA223 with Mj223 yielded a calculated model with good model score.14 DNA docking attempts suggest that RUSA223 also recognizes a symmetric 14 bp target (data not shown). Further experiments with RUSA223 to identify high-affinity DNA ligands and site-directed mutagenesis should permit full characterization of its putative promoter binding sites within the S. aureus chromosome. Similar comparative studies with Mj223 may well reveal chemical stress response genes that are common to both archaea and eubacteria.


We thank Dr. R.M. Sweet for access to Beamline X12C at the National Synchrotron Light Source, and Dr. G.A. Petsko and members of the New York Structural Genomics Research Consortium (NYSGRC) for useful discussions.