Three dimensional structure of the MqsR:MqsA complex: a novel TA pair comprised of a toxin homologous to RelE and an antitoxin with unique properties

Abstract

One mechanism by which bacteria survive environmental stress is through the formation of bacterial persisters, a sub-population of genetically identical quiescent cells that exhibit multidrug tolerance and are highly enriched in bacterial toxins. Recently, the Escherichia coli gene mqsR (b3022) was identified as the gene most highly upregulated in persisters. Here, we report multiple individual and complex three-dimensional structures of MqsR and its antitoxin MqsA (B3021), which reveal that MqsR:MqsA form a novel toxin:antitoxin (TA) pair. MqsR adopts an alpha/beta fold that is homologous with the RelE/YoeB family of bacterial ribonuclease toxins. MqsA is an elongated dimer that neutralizes MqsR toxicity. As expected for a TA pair, MqsA binds its own promoter. Unexpectedly, it also binds the promoters of genes important for E. coli physiology (e.g., mcbR, spy). Unlike canonical antitoxins, MqsA is also structured throughout its entire sequence, binds zinc and coordinates DNA via its C- and not N-terminal domain. These studies reveal that TA systems, especially the antitoxins, are significantly more diverse than previously recognized and provide new insights into the role of toxins in maintaining the persister state.

Figure 1. MqsR:MqsA are a bona fide toxin:antitoxin pair.

(A) The effect of mqsR, mqsA-F and mqsR-mqsA-F on colony formation for E. coli strain BW25113 containing pBS(Kan; empty plasmid), pBS(Kan)-mqsR, pBS(Kan)-mqsA-F, and pBS(Kan)-mqsR-mqsA-F. (B) Size exclusion chromatogram and SDS-PAGE gel of the MqsR:MqsA-F complex (Superdex 75 26/60 column; MqsA-F, 14.7 kD; MqsR, 11.5 kD). Elution positions of molecular weight standards are indicated by arrows. The complex elutes as a single peak at a position consistent with the formation of a dimer of dimers (two copies of MqsR and two copies of MqsA-F; expected MW of the complex  = 52.4 kD). Electrophoretic mobility shift assays: both the MqsR:MqsA-F complex (C) and MqsA-F alone (D) bind the mqsR promoter; MqsA-F also binds the mcbR promoter (E) and the spy promoter (F). PtomB is a control DNA fragment that we show in Figures S1F, S1G is not bound by either the MqsR:MqsA-F complex or MqsA-F.

Figure 2. The structures of MqsA and MqsR.

(A) Ribbon representation of the MqsA-F dimer. The MqsA-F dimer is composed of two chains (blue and orange), each of which have two distinct N- and C-terminal domains (illustrated in dark and light shades, respectively). MqsA-F dimerization is mediated by the C-terminal HTH-XRE domains (MqsA-C; light blue and light orange), with the N-terminal zinc binding domains (MqsA-N; dark blue and dark orange) projecting away from the dimer interface; the coordinated zinc ion is shown as a teal sphere. The local two-fold axis relating the MqsA-F monomers to one another is indicated by an arrow. The dimensions of the dimer (45×52×90 Å) are shown. (B) Ribbon representation of the complex between MqsR (magenta) and MqsA-N (dark blue). The MqsR:MqsA-N interface is composed of two sites, the primary site (1°) and the secondary site (2°). In (A) and (B), the panel on the right is rotated by 90° with respect to that on the left about the axis shown.

Figure 3. MqsA is a structured antitoxin.

(A) The MqsA-F monomer can be visualized as a ‘leg’: MqsA-C (the HTH-XRE domain; light blue) is the thigh, the flexible linker (centered on T68) is the knee and MqsA-N (the zinc binding domain; dark blue) is the ‘calf’ and ‘foot’; the zinc (teal sphere) is bound by the ‘toes’. (B) The zinc is coordinated by Cys3, Cys6, Cys37 and Cys40 (blue/yellow sticks). Sulfur-zinc distances are commensurate with values typical for structural zinc binding sites. Residues near the zinc binding pocket are also shown (grey sticks). (C) The MqsA zinc binding domain adopts a novel fold characterized by a long, twisted β-sheet buttressed by a five-turn α-helix with the loops coordinated by zinc. It is stabilized by two hydrophobic cores (core 1, sidechains shown as sticks in beige; core 2, sidechains shown as sticks in orange). (D) The MqsA-C dimerization interface is composed of the MqsA-C α-helices 3, 5, and 6, with the two monomers (chain A, grey; chain B, light blue) related by local two-fold symmetry. The sequence most highly conserved among MqsA proteins (residues 98–105, green) is predicted to bind DNA (see text). (E) The MqsA-C dimerization interface rotated by 90° with one monomer represented in surface representation (1600 Ų of buried SASA illustrated in beige) and the second in ribbon representation. (F) Residues buried at the MqsA-C dimer interface are shown as sticks (chain B) and the charge distribution is shown in surface representation (chain A; positive charge, blue; negative charge, red).

Figure 4. MqsA binds MqsR via its MqsA-N zinc binding domain and DNA via its MqsA-C HTH domain.

(A) Growth curves of the co-expression of MqsR with the N-terminal domain of MqsA (MqsR:MqsA-N, grey square) and the C-terminal domain of MqsA (MqsR:MqsA-C, black diamond) in BL21 (DE3) cells. Induction of protein expression using 0.5 mM IPTG corresponds to t = 0 (arrow; expression carried out at 18°C, following the same protocol used to produce the proteins for structural studies). Co-expression of MqsR with MqsA-C leads to growth arrest while co-expression with MqsA-N results in robust growth. Data shown represents the average of the two measurements with the standard deviation shown as error bars. (B) Size exclusion chromatogram (SEC) of the co-expressed MqsR:MqsA-N complex (dark grey, left) and the SDS-PAGE gel of the pooled peak. As can be seen, both MqsR (11.5 kD) and MqsA-N (8.5 kD) are present. The MqsR:MqsA-N SEC, with corresponding SDS-PAGE gel, is overlaid with that of MqsA-N alone (light grey, right), illustrating the shift in the elution position of MqsA-N upon complex formation. (C–E) EMSAs using only MqsA-N (C), only MqsA-C (D) or the MqsR:MqsA-N complex (E) with the mqsR promoter.

Figure 5. MqsR shares structural homology with the RelE and YoeB family of E. coli RNase toxins.

(A) MqsR (magenta) with labeled secondary structural elements, and those of its closest structural homologs YoeB (green; DALI Z-score = 5.1; PDBID 2A6Q [33]) and RelE (cyan, DALI Z-score = 4.0; PDBID 2KC8 [45]). RNase Sa (orange; PDBID 1RSN [48]), a canonical bacterial RNase, is also shown. The bacterial RNase fold is characterized by a central, twisted β-sheet and adjacent α–helix and is conserved among all four proteins. (B) Corresponding secondary structural elements of each protein; elements highlighted in green are structurally conserved between the proteins.

Figure 6. Identification of the MqsR functional site.

(A) Growth curves of cells over-expressing WT MqsR and seven MqsR mutants. MqsR mutants K56A, Q68A, Y81A and K96A show decreased toxicity compared to WT MqsR, as evidenced by their ability to grow following induction with 0.5 mM IPTG at t = 0 (arrow; expression carried out at 18°C, following the same protocol used to produce the proteins for structural studies). In contrast, induction of MqsR mutants Y55A, M58A and R72A, like WT, lead to growth arrest. (B) Western blots showing soluble expression of MqsR mutants K56A, Q68A, Y81A and K96A from the cultures used to generate the growth curves in (A) (WT MqsR arrests cell growth so rapidly that free MqsR is not detectable by Western Blot). All MqsR constructs include an N-terminal his₆-tag and protein expression was detected using a polyhistidine polyclonal antibody. Like WT, expressed protein for MqsR mutants Y55A, M58A and R72A was not detected.

Figure 7. MqsA recognition of MqsR.

(A) Stereoimage of the 1° interaction site between MqsR (magenta) and MqsA-N (dark blue; see also Fig. 2B), which buries 1537 Ų of solvent accessible surface area (SASA). The residues that participate in the interaction are shown as sticks. Those that are completely buried upon complex formation (>80% loss of SASA) are shown in pink (MqsR) and light blue (MqsA-N). The interface is centered on MqsA-N residue M45 and is predominantly hydrophobic. Electrostatic interactions, including the hydrogen bonds formed between MqsA-N β-strand 3 (S43-M45) and MqsR β-strand 2 (V22-T25) that function to extend the MqsR β-sheet with that of MqsA-N (β1_R-β3_R-β4_R-β5_R-β6_R-β2_R-β3_A-β2_A-β1_A), are indicated by dashed lines. (B) The 2° interaction site buries 479 Ų of SASA; colors and electrostatic interactions shown as in (A). (C) The MqsR:MqsA-N complex, with MqsA-N in surface representation (dark blue) and MqsR as a ribbon diagram (magenta). The residues that play a role in MqsR-mediated toxicity (Fig. 6A) are shown as sticks in yellow. With the exception of Q68, the residues that play a role in toxicity are accessible when MqsR is bound to MqsA-N.

Figure 8. MqsR is most similar to the bacterial toxin RelE.

Superposition of MqsR (magenta) with RelE (cyan; A), YoeB (green; B) and RNase Sa (orange; C) using corresponding β-strands (MqsR, β4/β5; RelE, β2/β3; YoeB, β3/β4; RNase Sa, β3/β4). Residues in RelE, YoeB and RNase SA that play a role in toxicity are shown as sticks with black labels while those in MqsR are shown as sticks with grey labels. MqsR lacks the histidine and glutamic acid residues that mediate catalysis in YoeB (E46, H83) and RNase Sa (H85, E54).

Figure 9. Model of the MqsR:MqsA₂:MqsR:DNA complex.

(A) The MqsA-F and MqsR:MqsA-N structures were used to generate the full MqsR:MqsA₂:MqsR complex. The MqsA monomers are shown in light blue and light orange, with conserved MqsA-F α–helix 4 in green. MqsR is in magenta. The local two-fold symmetry axis is indicated by a black arrow. Right and left panels are rotated by 90° about the horizontal axis. (B) The electrostatic surface of the MqsR:MqsA₂:MqsR complex (positive charges, blue; negative charges, red). The right panel is rotated about the horizontal axis, illustrating the extensive positively charged surface at the bottom of the MqsA C-terminal HTH-XRE domain. (C) Model of the MqsR:MqsA₂:MqsR:DNA complex, which is based on the observation that all HTH-XRE DNA binding domains bind DNA using the same helix. In MqsA, this corresponds to α-helix 4. Model generation described in the methods. Because the dimerization interface differs between MqsA and the P22 C2 repressor , the DNA bound by MqsA must bend (illustrated by dashed lines between the two DNA segments). Two views of the MqsR:MqsA₂:MqsR:DNA complex are shown. The left panel is the same orientation as that shown in the left panel of A. The image is rotated about the vertical axis. Very little of the MqsA-N zinc binding domain, which binds MqsR, interacts with the bound DNA. The pseudo two-fold symmetry axis is indicated by an arrow.