two distinct mechanisms of transcriptional regulation by ... · two distinct mechanisms of...

10
Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Lee a,1 , In-Gyun Lee a,1 , Ki-Young Lee a , Dong-Gyun Kim a , Hyun-Jong Eun a , Hye-Jin Yoon b , Susanna Chae a , Sung-Hyun Song c,d , Sa-Ouk Kang c,d , Min-Duk Seo e , Hyoun Sook Kim f , Sung Jean Park g , and Bong-Jin Lee a,2 a The Research Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Gwanak-gu, Seoul 151-742, Republic of Korea; b Department of Chemistry, College of Natural Sciences, Seoul National University, Seoul 151-742, Republic of Korea; c Laboratory of Biophysics, School of Biological Sciences, Seoul National University, Seoul 151-742, Republic of Korea; d Institute of Microbiology, College of Natural Sciences, Seoul National University, Seoul 151-742, Republic of Korea; e Department of Molecular Science and Technology & College of Pharmacy, Ajou University, Suwon, Kyeonggi 443-749, Republic of Korea; f Research Institute, National Cancer Center, Goyang, Gyeonggi 410-769, Republic of Korea; and g College of Pharmacy, Gachon University, 534-2 Yeonsu-dong, Yeonsu-gu, Incheon 406-799, Republic of Korea Edited by Gisela Storz, National Institutes of Health, Bethesda, MD, and approved July 11, 2016 (received for review March 17, 2016) For bacteria, cysteine thiol groups in proteins are commonly used as thiol-based switches for redox sensing to activate specific detoxifica- tion pathways and restore the redox balance. Among the known thiol-based regulatory systems, the MarR/DUF24 family regulators have been reported to sense and respond to reactive electrophilic species, including diamide, quinones, and aldehydes, with high specificity. Here, we report that the prototypical regulator YodB of the MarR/DUF24 family from Bacillus subtilis uses two distinct path- ways to regulate transcription in response to two reactive electrophilic species (diamide or methyl-p-benzoquinone), as revealed by X-ray crystallography, NMR spectroscopy, and biochemical experiments. Di- amide induces structural changes in the YodB dimer by promoting the formation of disulfide bonds, whereas methyl-p-benzoquinone allows the YodB dimer to be dissociated from DNA, with little effect on the YodB dimer. The results indicate that B. subtilis may discriminate toxic quinones, such as methyl-p-benzoquinone, from diamide to efficiently manage multiple oxidative signals. These results also provide evidence that different thiol-reactive compounds induce dissimilar conforma- tional changes in the regulator to trigger the separate regulation of target DNA. This specific control of YodB is dependent upon the type of thiol-reactive compound present, is linked to its direct transcrip- tional activity, and is important for the survival of B. subtilis. This study of B. subtilis YodB also provides a structural basis for the relationship that exists between the ligand-induced conformational changes adop- ted by the protein and its functional switch. YodB | MarR/DUF24 | transcriptional regulator | redox signaling | reactive electrophilic species R edox signaling in bacteria is an attractive field and has revealed how bacteria trigger defense mechanisms against environmental and host stresses to survive. For bacteria, the thiol groups of cysteines in proteins are commonly used as thiol-based switches in redox-sensing regulators to activate specific de- toxification pathways and restore the redox balance (13). Direct sensing and quick responses by transcription factors are regarded as an efficient way to promote the survival of a tiny bacterium and overcome oxidation stress (13). In addition, the redox sig- naling pathways have also been used as important virulence regulators that allow pathogenic bacteria to adapt to the host immune defense system (4). Among the thiol-based oxidation regulators, Escherichia coli OxyR (5, 6), Xanthomonas campestris OhrR (7, 8), Bacillus subtilis OhrR (9), Pseudomonas aeruginosa MexR (10), and E. coli NemR (11, 12) are representatives that demonstrate how well these regulators sense organic peroxide by forming disulfide bonds between two distantly located cysteines (10). OxyR oxi- dation leads to the formation of intramolecular disulfide bonds and alters the interaction between the OxyR tetramer and the DNA sites upstream from the OxyR-regulated genes, which enhances OxyR DNA recognition capability (5, 6). Similarly, intermolecular disulfide bonds are formed in response to the oxidation of X. campestris OhrR and reorient the winged-helix domains of the OhrR dimer, which reduces its DNA-binding affinity. Unlike a two-Cys type, such as X. campestris OhrR, B. subtilis OhrR, a one-Cys type, senses hydroperoxides by forming various reversible S-thiolations with the redox buffer bacillithiol (9). As with X. campestris OhrR, MexR forms intermolecular disulfide bonds in response to oxidative stress, which results in a rigid body rotation of the DNA-recognition helices, attenuating DNA-binding affinity (10). E. coli NemR possesses a redox switch that senses either electrophiles or reactive chlorine species by the formation of disulfide bonds (11) or a reversible sulfenamide bond (12), respectively. The representative redox regulators E. coli OxyR, X. campestris OhrR, B. subtilis OhrR, P. aeruginosa MexR, and E. coli NemR demonstrate how these proteins are structurally influenced by the formation of disulfide bonds that are induced by oxidative stress. In contrast, the MarR/DUF24 family regulators have been known to sense and respond to reactive electrophilic species (RES), including diamide, quinones, and aldehydes, with high specificity (1). Among the MarR/DUF24 family members, B. subtilis YodB is the prototypical regulator and is reported to be regulated by diamide and quinones, which induce intersubunit disulfide Significance Bacteria sense and protect themselves against oxidative stress using redox-sensing transcription regulators with cysteine resi- dues. Here, we investigate at the molecular level how the YodB protein, a transcription repressor in Bacillus subtilis, monitors and responds to different oxidative stresses. Diamide stress leads to the formation of disulfide bonds between cysteine residues, whereas the more toxic quinone compound methyl-p-benzoqui- none forms an adduct on a specific cysteine residue. These chemical modifications lead to considerably different changes in the YodB structure, causing the release of YodB from the DNA of antioxidant genes. The redox-sensing transcription regulator YodB allows B. subtilis to respond to multiple oxidative signals of differing toxicity by adopting different structures. Author contributions: S.J.L., I.-G.L., and B.-J.L. designed research; I.-G.L., D.-G.K., H.-J.E., S.C., and H.S.K. performed research; S.-H.S. and S.-O.K. contributed new reagents/analytic tools; S.J.L., I.-G.L., K.-Y.L., H.-J.Y., M.-D.S., and S.J.P. analyzed data; and S.J.L., I.-G.L., S.J.P., and B.-J.L. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. Freely available online through the PNAS open access option. Data deposition: The atomic coordinates and structure factors have been deposited in the Protein Data Bank, www.pdb.org (PDB ID codes 5HS7, 5HS8, and 5HS9). 1 S.J.L and I.-G.L. contributed equally to this work. 2 To whom correspondence should be addressed. Email: [email protected]. This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. 1073/pnas.1604427113/-/DCSupplemental. E5202E5211 | PNAS | Published online August 16, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1604427113

Upload: haque

Post on 29-Oct-2018

224 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

Two distinct mechanisms of transcriptional regulationby the redox sensor YodBSang Jae Leea,1, In-Gyun Leea,1, Ki-Young Leea, Dong-Gyun Kima, Hyun-Jong Euna, Hye-Jin Yoonb, Susanna Chaea,Sung-Hyun Songc,d, Sa-Ouk Kangc,d, Min-Duk Seoe, Hyoun Sook Kimf, Sung Jean Parkg, and Bong-Jin Leea,2

aThe Research Institute of Pharmaceutical Sciences, College of Pharmacy, Seoul National University, Gwanak-gu, Seoul 151-742, Republic of Korea;bDepartment of Chemistry, College of Natural Sciences, Seoul National University, Seoul 151-742, Republic of Korea; cLaboratory of Biophysics, School ofBiological Sciences, Seoul National University, Seoul 151-742, Republic of Korea; dInstitute of Microbiology, College of Natural Sciences, Seoul NationalUniversity, Seoul 151-742, Republic of Korea; eDepartment of Molecular Science and Technology & College of Pharmacy, Ajou University, Suwon, Kyeonggi443-749, Republic of Korea; fResearch Institute, National Cancer Center, Goyang, Gyeonggi 410-769, Republic of Korea; and gCollege of Pharmacy, GachonUniversity, 534-2 Yeonsu-dong, Yeonsu-gu, Incheon 406-799, Republic of Korea

Edited by Gisela Storz, National Institutes of Health, Bethesda, MD, and approved July 11, 2016 (received for review March 17, 2016)

For bacteria, cysteine thiol groups in proteins are commonly used asthiol-based switches for redox sensing to activate specific detoxifica-tion pathways and restore the redox balance. Among the knownthiol-based regulatory systems, the MarR/DUF24 family regulatorshave been reported to sense and respond to reactive electrophilicspecies, including diamide, quinones, and aldehydes, with highspecificity. Here, we report that the prototypical regulator YodB ofthe MarR/DUF24 family from Bacillus subtilis uses two distinct path-ways to regulate transcription in response to two reactive electrophilicspecies (diamide or methyl-p-benzoquinone), as revealed by X-raycrystallography, NMR spectroscopy, and biochemical experiments. Di-amide induces structural changes in the YodB dimer by promoting theformation of disulfide bonds, whereas methyl-p-benzoquinone allowsthe YodB dimer to be dissociated from DNA, with little effect on theYodB dimer. The results indicate that B. subtilismay discriminate toxicquinones, such as methyl-p-benzoquinone, from diamide to efficientlymanagemultiple oxidative signals. These results also provide evidencethat different thiol-reactive compounds induce dissimilar conforma-tional changes in the regulator to trigger the separate regulation oftarget DNA. This specific control of YodB is dependent upon the typeof thiol-reactive compound present, is linked to its direct transcrip-tional activity, and is important for the survival of B. subtilis. This studyof B. subtilis YodB also provides a structural basis for the relationshipthat exists between the ligand-induced conformational changes adop-ted by the protein and its functional switch.

YodB | MarR/DUF24 | transcriptional regulator | redox signaling |reactive electrophilic species

Redox signaling in bacteria is an attractive field and hasrevealed how bacteria trigger defense mechanisms against

environmental and host stresses to survive. For bacteria, the thiolgroups of cysteines in proteins are commonly used as thiol-basedswitches in redox-sensing regulators to activate specific de-toxification pathways and restore the redox balance (1–3). Directsensing and quick responses by transcription factors are regardedas an efficient way to promote the survival of a tiny bacteriumand overcome oxidation stress (1–3). In addition, the redox sig-naling pathways have also been used as important virulenceregulators that allow pathogenic bacteria to adapt to the hostimmune defense system (4).Among the thiol-based oxidation regulators, Escherichia coli

OxyR (5, 6), Xanthomonas campestris OhrR (7, 8), Bacillussubtilis OhrR (9), Pseudomonas aeruginosa MexR (10), andE. coli NemR (11, 12) are representatives that demonstrate howwell these regulators sense organic peroxide by forming disulfidebonds between two distantly located cysteines (10). OxyR oxi-dation leads to the formation of intramolecular disulfide bondsand alters the interaction between the OxyR tetramer and theDNA sites upstream from the OxyR-regulated genes, whichenhances OxyR DNA recognition capability (5, 6). Similarly,intermolecular disulfide bonds are formed in response to the

oxidation of X. campestris OhrR and reorient the winged-helixdomains of the OhrR dimer, which reduces its DNA-bindingaffinity. Unlike a two-Cys type, such as X. campestris OhrR,B. subtilisOhrR, a one-Cys type, senses hydroperoxides by formingvarious reversible S-thiolations with the redox buffer bacillithiol(9). As with X. campestris OhrR, MexR forms intermoleculardisulfide bonds in response to oxidative stress, which results in arigid body rotation of the DNA-recognition helices, attenuatingDNA-binding affinity (10). E. coli NemR possesses a redox switchthat senses either electrophiles or reactive chlorine species by theformation of disulfide bonds (11) or a reversible sulfenamide bond(12), respectively. The representative redox regulators E. coliOxyR, X. campestris OhrR, B. subtilis OhrR, P. aeruginosa MexR,and E. coli NemR demonstrate how these proteins are structurallyinfluenced by the formation of disulfide bonds that are induced byoxidative stress.In contrast, the MarR/DUF24 family regulators have been

known to sense and respond to reactive electrophilic species(RES), including diamide, quinones, and aldehydes, with highspecificity (1). Among the MarR/DUF24 family members, B. subtilisYodB is the prototypical regulator and is reported to be regulatedby diamide and quinones, which induce intersubunit disulfide

Significance

Bacteria sense and protect themselves against oxidative stressusing redox-sensing transcription regulators with cysteine resi-dues. Here, we investigate at the molecular level how the YodBprotein, a transcription repressor in Bacillus subtilis, monitors andresponds to different oxidative stresses. Diamide stress leads tothe formation of disulfide bonds between cysteine residues,whereas the more toxic quinone compound methyl-p-benzoqui-none forms an adduct on a specific cysteine residue. Thesechemical modifications lead to considerably different changes inthe YodB structure, causing the release of YodB from the DNA ofantioxidant genes. The redox-sensing transcription regulatorYodB allows B. subtilis to respond to multiple oxidative signals ofdiffering toxicity by adopting different structures.

Author contributions: S.J.L., I.-G.L., and B.-J.L. designed research; I.-G.L., D.-G.K., H.-J.E., S.C.,and H.S.K. performed research; S.-H.S. and S.-O.K. contributed new reagents/analytic tools;S.J.L., I.-G.L., K.-Y.L., H.-J.Y., M.-D.S., and S.J.P. analyzed data; and S.J.L., I.-G.L., S.J.P., and B.-J.L.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The atomic coordinates and structure factors have been deposited in theProtein Data Bank, www.pdb.org (PDB ID codes 5HS7, 5HS8, and 5HS9).1S.J.L and I.-G.L. contributed equally to this work.2To whom correspondence should be addressed. Email: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1604427113/-/DCSupplemental.

E5202–E5211 | PNAS | Published online August 16, 2016 www.pnas.org/cgi/doi/10.1073/pnas.1604427113

Page 2: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

formation between Cys6-Cys101′ or S-alkylation on Cys6, re-spectively (13, 14). However, little is known about the exact mo-lecular mechanisms that are responsible for both the diamide- andquinone-mediated signaling pathways of B. subtilis YodB, whichprompted us to initiate structural and functional studies onB. subtilis YodB.Here, we have revealed that the prototypical regulator YodB

of the MarR/DUF24 family from B. subtilis uses two distinctpathways to regulate transcription in response to two RES [di-amide or methyl-p-benzoquinone (MPBQ)], as revealed by X-raycrystallography, NMR spectroscopy, and biochemical experi-ments. Surprisingly, B. subtilis YodB possesses two distinctiveconformational states, depending on the types of thiol-reactivecompounds (diamide or MPBQ), which may steer the cells intoan efficient defense state. Following oxidative shock by eitherdiamide or MPBQ, B. subtilis YodB induces the formation ofdisulfide bonds between two YodB monomers or S-adducts intwo distinct ways. Thus, the regulatory mechanism of B. subtilisYodB should be dissimilar to those of known regulators, in-cluding X. campestris OhrR, B. subtilis HypR, or S. aureus QsrR.In addition, the different conformations of YodB may be relatedto an efficient method of transcriptional regulation to reduceoxidative stress. Our structural study provides the evidence thatdifferent thiol-reactive compounds induce dissimilar conforma-tional changes that trigger separate regulatory mechanisms on thetarget DNA. The specific control of YodB is dependent upon thetype of thiol-reactive compound, is linked with its direct tran-scriptional activity, and is important for the survival of B. subtilis.

ResultsAfter an extensive number of trials to crystallize variousB. subtilis YodB constructs, including the full-length construct,we obtained three types of crystals (B. subtilis YodBreduced,YodBdiamide, and YodBMPBQ) from the truncated YodB5–105construct. B. subtilis YodB was reacted with 1 mM diamide(or 1.5 mM MPBQ) to produce B. subtilis YodBdiamide (orYodBMPBQ), whereas 1 mM DTT was used to prepare B. subtilisYodBreduced. In B. subtilis, 2-methylhydroquinone (MHQ) isoxidized to MPBQ, which forms an S-adduct via 1,4-reductiveaddition of thiols to quinones (15). Due to the high toxicity ofMPBQ (Fig. S1), B. subtilis stimulates the azoreductases AzoR1and AzoR2, which convert MPBQ into MHQ to protect the cells(15–17). In our experiment, B. subtilis was not able to grow in thepresence of 1 mM MPBQ, whereas neither 1 mM MHQ nor1 mM diamide affected growth of either WT (YodBWT) orΔyodB mutant cells. In addition, the ΔyodB mutant was moreresistant to MHQ and diamide (Fig. S1). Therefore, we usedMPBQ to observe the quinone-induced change in the structureof YodB (YodBMPBQ), which was reported to be 50-fold moretoxic than hydroquinones, such as MHQ, in B. subtilis cells (13).The crystal structures of B. subtilis YodBreduced, YodBdiamide,and YodBMPBQ were determined at 1.7, 2.0, and 2.1 Å reso-lution, respectively (Table 1). The crystal structures of B. subtilisYodBreduced and YodBMPBQ contained two monomers (chains Aand B) in the asymmetric unit, whereas the crystal of B. subtilisYodBdiamide contains one monomer per asymmetric unit.

Table 1. Statistics for data collection, phasing, and model refinement

Data

Dataset

YodBreduced YodBdiamide YodBMPBQ

A. Data collectionSpace group P21 P62 P21Unit cell lengths, a, b, c, Å 40.78, 50.80, 50.35 94.98, 94.98, 25.42 38.94, 51.18, 48.39Unit cell angle, β, ° 95.49 93.46X-ray wavelength, Å 1.0000 1.0000 1.0000Resolution range, Å 50.0–1.70 (1.73–1.70)* 30.0–2.00 (2.03–2.00)* 50.0–2.10 (2.14–2.10)*Total/unique reflections 38,373/22,351 54,617/9,187 41,520/10,779Completeness, % 99.0 (98.4)* 99.9 (99.8)* 96.0 (95.8)*CC1/2

† 0.999 (0.921)* 0.997 (0.767)* 0.998 (0.935)*<I >/<σI> 44.5 (3.5)* 35.8 (2.3)* 36.9 (2.7)*Rmerge

‡, % 5.6 (45.1)* 6.8 (62.6)* 7.0 (62.9)*B. Model refinement

PDB ID code 5HS7 5HS8 5HS9Resolution range, Å 50.0–1.70 30.0–2.00 50.0–2.10Rwork/Rfree

§, % 19.7/22.1 20.6/24.8 24.4/27.6No. of nonhydrogen atoms/average B-factor, Å2

Protein 1,617/30.9 817/39.5 1,446/65.0Water 160/39.8 78/50.8 11/67.7Glycerol 6/29.2

Wilson B-factor, Å2 25.6 35.4 52.3Rms deviations from ideal geometry

Bond lengths, Å/bond angles, ° 0.012/1.54 0.011/1.26 0.012/1.48Rms Z-scores

Bond lengths/bond angles 0.589/0.720 0.696/0.569 0.603/0.710Ramachandran plot

Favored/outliers, % 96.0/0.0 99.0/0.0 99.0/0.0Poor rotamers, % 0.6 0.0 1.0

*Values in parentheses refer to the highest-resolution shell.†CC1/2 is described in ref. 46.‡Rmerge = Σh Σi j I(h)i – < I(h) > j/Σh Σi I(h)i, where I(h) is the intensity of reflection h, Σh is the sum over all reflections, and Σi is the sumover i measurements of reflection h.§R = Σ j jFobsj – jFcalcj j/Σ jFobsj, where Rfree and Rwork are calculated for a randomly chosen 5% of reflections that were not used forrefinement and for the remaining reflections, respectively.

Lee et al. PNAS | Published online August 16, 2016 | E5203

MICRO

BIOLO

GY

PNASPL

US

Page 3: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

Crystal Structures of the Reduced (YodBreduced), Diamide-Treated(YodBdiamide), and Quinone-Bound (YodBMPBQ) Forms of B. subtilisYodB. In all of the models, the C-terminal residues are disor-dered: three residues (Pro103–Asp105) in chain A and four resi-dues (Glu102–Asp105) in chain B of YodBreduced, two residues(Glu102–Pro103) of YodBdiamide, and five residues (Ser101–Asp105) of YodBMPBQ. In the YodBMPBQ model, the loop be-tween β2 and β3 is completely disordered in both the A and Bchains. The monomers of B. subtilis YodBreduced and YodBdiamideare composed of five α-helices and three β-strands as follows: α1(residues 7–17), α2 (residues 21–28), β1 (residues 33–34), α3(residues 35–41), α4 (residues 47–59), β2 (residues 63–68), β3(residues 74–79), and α5 (residues 81–100) (Fig. 1).The two monomer structures of YodBreduced are highly similar to

each other, with rms deviations of 0.7 Å for 96 equivalent Cα pairs.Additionally, the two monomer structures of YodBMPBQ are nearlyidentical to each other, with rms deviations of 0.7 Å for 88 equiv-alent Cα pairs. The overall monomer structures of YodBreduced arenearly identical to those of YodBdiamide (with rms deviations of1.3−1.5 Å for 87−91 equivalent Cα pairs) and YodBMPBQ (with rmsdeviations of 0.6−0.9 Å for 88−90 equivalent Cα pairs). Despite thehigh structural similarities between YodBreduced, YodBdiamide, andYodBMPBQ, large rms deviations are observed in several regions.For YodBdiamide, the N- and C-terminal α-helices (α1 and α5) andthe loop between β2 and β3 showed maximum deviations of 2.3 Å(Cys6 at the start of α1), 7.0 Å (Cys101 at the end of α5), and 3.8 Å(Pro72), respectively; for YodBMPBQ, the loop between β2 and β3was completely disordered, and the α3 helix and the loop betweenα3 and α4 deviated from the corresponding region of YodBreduced,with maximum deviations of 4.0 Å (Fig. 1C).When the monomer structures of YodBreduced and YodBdiamide

were analyzed to identify structural homologs using the Daliserver (18), the structures of Staphylococcus aureus QsrR (4)

[Protein Data Bank (PDB) ID codes 4HQE and 4HQM;Z-scores of 12.4−16.1, rms deviations of 1.3−3.3 Å, and a sequenceidentity of 38% for 96−100 equivalent Cα pairs], B. subtilis HypR(19) (PDB ID codes 4A5M and 4A5N; Z-scores of 13.7−14.6,rms deviations of 2.1−3.4 Å, and a sequence identity of 38% for95−100 equivalent Cα pairs), and E. coli MarR (20) (PDB IDcode 1JGS; Z-scores of 11.7−12.4, rms deviations of 1.9−3.3 Å,and a sequence identity of 14% for 90 equivalent Cα pairs) werethe most similar, as expected.

Diverse Features of the Dimer Structures of B. subtilis YodBreduced,YodBdiamide, and YodBMPBQ. The molecular masses of B. subtilisYodB5–105 (YodBreduced, YodBdiamide, and YodBMHPQ) were de-termined by size-exclusion chromatography with inline multianglelight scattering (SEC-MALS) and were ∼27−31 kDa (Fig. S2). Thesemasses correspond to the dimer of B. subtilis YodB in solution.Strikingly different features were observed in the dimer models

of each YodB protein (YodBreduced, YodBdiamide, and YodBMPBQ).Despite the high structural similarities between the YodBreduced,YodBdiamide, and YodBMPBQ monomers, the dimer models of eachYodB are discretely dissimilar to each other (Fig. 1). The refineddimer models of YodBreduced and YodBdiamide are essentially dis-similar, with rms deviations of 9.2 Å for 198 equivalent Cα pairs,whereas those of YodBreduced and YodBMPBQ display fewer con-formational differences, with rms deviations of 1.3 Å for 178equivalent Cα pairs (Fig. 1). The YodB dimer behaves in a dif-ferent mode, with respect to its two types of thiol-reactive com-pound-induced forms, giving rise to significantly different shiftsamong YodBreduced, YodBdiamide, and YodBMPBQ.The dimer structure of YodBreduced is nearly identical to the

structures of S. aureus QsrR and B. subtilis HypR, with the ex-ception of the local positions of the α-helix (α4), which participatedin the DNA binding, and the loop between β2 and β3 (Fig. S3).

Fig. 1. Overall structures of B. subtilis YodBreduced,YodBdiamide, and YodBMPBQ. (A) The structure of theB. subtilis YodBreduced dimer (chain A in cyan and chainB in blue) is presented in two views. The cysteine resi-dues are drawn as red and yellow spheres. The positionsof the cysteine residues are indicated as black rectan-gular boxes and the details are shown in D. (B) Thestructure of the B. subtilis YodBdiamide dimer (chain A inyellow and chain B in orange) is presented in two views.The two disulfides are indicated as black rectangularboxes (Right) and the details are shown in E. (C) Thestructure of the B. subtilis YodBMPBQ dimer (chain A inpink and chain B in magenta) is presented in two views.The disordered regions are drawn as black dotted lines.The conformational change related to the recognitionof the DNA by YodBMPBQ is marked by the black rect-angular box and the details are shown in F. (D) The twosets of reduced cysteine residues in B. subtilis (Cys6 andCys101′ or Cys6′ and Cys101) are drawn as red spheres.(E) The two disulfide bonds in B. subtilis YodBdiamide aredrawn as an mFo−DFc electron density map, which iscontoured at 2.0 σ (in blue mashes). (F) Modification ofCys6 by MPBQ induces conformational change of rec-ognition helix (α4′) and movement of Trp20 and Glu10.The residues Arg19, Trp20, Met44, and Ile45 are alsoinfluenced by MPBQ binding. The structures were con-structed using PyMOL (45).

E5204 | www.pnas.org/cgi/doi/10.1073/pnas.1604427113 Lee et al.

Page 4: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

Despite the high similarity between the monomer structures ofYodBreduced and E. coli MarR, their dimer structures and the in-terfaces between monomers are completely different (Fig. S3). Inthe reduced state (YodBreduced), the two sulfur atoms of Cys6/Cys101′ and Cys6′/Cys101 are well separated by 8.3 and 9.0 Å,respectively. After oxidation via the addition of diamide, disulfidebonds are formed with S–S distances of 2.2 Å (Fig. 1 D and E).The interfaces of YodBreduced, YodBdiamide, and YodBMPBQ per

dimers are 3,300, 2,240, and 2,700 Å2, respectively. The interfacesof YodBreduced are definitely larger than those of YodBdiamide andYodBMPBQ, which may indicate that YodBreduced forms a morefavorable dimer than the other forms. Although they have dis-similar interfaces, the residues involved in the hydrophobic inter-actions of each dimeric interface in YodBreduced, YodBdiamide,and YodBMPBQ are nearly identical (Fig. 2). However, the residuesthat contribute to the hydrophilic interactions in the dimer ofYodBdiamide are completely different from those in the YodBreducedand YodBMPBQ (Fig. 2).

Quinone-Induced Conformational Change in the YodBMPBQ Dimer.The mass spectra of the YodB (5–17) peptide show that theunmodified peptide peak at 1,413.6 Da is shifted to 1,535.7 Daafter the addition of MPBQ (Fig. 3). The observed increase of122.1 Da in the peptide mass corresponds to a covalent bondwith MPBQ (molecular mass, 122.1 Da) (Fig. 3). In YodBMPBQ,the electron density of MPBQ was not observed near the Cys6 orCys6′ residues, although MPBQ was covalently bound to YodBas shown in Fig. 3C. The Cys6 and Cys6′ residues showed highB-factors of 83 Å2 compared with 31 Å2 for the two Cys6 resi-dues in YodBreduced (the average B-factors of YodBMPBQ andYodBreduced are 65 Å

2 and 31 Å2, respectively). This result indicatesthat MPBQ may increase the mobility of the two Cys6 residues inthe dimer and affect a structural change in the dimer. In YodBMPBQ,the loop between α3 and α4 moves toward the α2 helix, whichcauses the side chain of Trp20′ to approach α1 and to interactwith Glu10 (Fig. 1F). The conformational change in the loopbetween α1′ and α2′ still maintains hydrophobic interactions withMet44′ and Ile45′ on the loop between α3′ and α4′. As a result,the MPBQ adducts on Cys6 and Cys6′ affect the DNA recog-nition helices α4 and α4′, causing them to move 3 Å (by mea-

suring distance between Lys48/Lys48′-Cα atoms) toward eachother, with ∼10° rotation (Fig. S4). In S. aureus QsrR, the con-formational change induced by the S-quinonization of Cys5 af-fects the DNA-binding region (α4 and α4′) and allows it to bedissociated from the DNA (4). For B. subtilisHypR, the diamide-and NaOCl-induced disulfide bond formation force the HypR tobind to the DNA and activate the transcription of hypO (19). TheS-quinonization of S. aureus QsrR moves α4 and α4′ of the QsrRdimer ∼11 Å in opposite directions, with a rotational change of∼28°, whereas the disulfide bond formed in the B. subtilis HypRdimer moves α4 and α4′ ∼4 Å toward each other, with a rota-tional change of ∼9° (Fig. S3). The conformation of B. subtilisYodBreduced most likely resembles that of QsrRoxidized(menadione)(or HypRreduced), rather than QsrRreduced (or HypRoxidized). Thereason for the opposite orientations of α4 and α4′ in QsrR and

Fig. 2. Detailed structures of the dimeric interface ofB. subtilis YodBreduced, YodBdiamide, and YodBMPBQ. (A) Theresidues in YodBreduced that contribute hydrophobic in-teractions at the interface are depicted. The detailed hy-drophilic interactions are labeled as a, b, and c, which areshown in D. (B) The residues in YodBdiamide that contrib-ute to the hydrophobic interactions at the interface aredepicted. The detailed hydrophilic interactions are labeledas a, which are shown in E. Because Glu92 and Glu92′ arepartly buried in the hydrophobic environment and formno favorable interactions with other residues, it is feasiblethat the carboxylate groups of Glu92 and Glu92′ areprotonated even at pH 5.7 (which is the measured pHvalue of the YodBdiamide crystallization condition) and in-teract with each other via hydrogen bonds with the dis-tance between OE1 (Glu92) and OE2 (Glu92′) [or OE1(Glu92′) and OE2 (Glu92)] of 2.6 Å. (C) The residues inYodBMPBQ that contribute to hydrophobic interactions atthe interface are depicted. The detailed hydrophilic in-teractions are labeled as a, b, and c, which are shown in F.(D−F) The residues that are involved in hydrophilic inter-actions at the dimeric interface of the three YodB dimers(YodBreduced, YodBdiamide, and YodBMPBQ) are shown in D−F, respectively. The hydrophilic interactions are indicatedby red dotted lines. All figures are presented in the sameorientation. For a better display, all residues of chain Aand chain B are presented in a surface view and ribbondiagram, respectively.

Fig. 3. Mass spectra of YodB before and after S-adduct formation withMPBQ. (A) The chemical structure and molecular mass of MPBQ. (B) Massspectra of the dissolved crystals of YodBC101S grown after incubating withMPBQ (Lower) and without treatment with MPBQ (Upper). An increase of125 Da indicates that MPBQ is covalently linked to Cys6 of the protein.(C) Mass spectra of the YodB (Met5−Gly17) peptide. The peptide mass isincreased by 121.1 Da by the formation of the S-adduct of MPBQ. The ad-dition of 1 mM DTT did not detach MPBQ from the peptide, indicating thatan irreversible S-adduct of MPBQ was formed.

Lee et al. PNAS | Published online August 16, 2016 | E5205

MICRO

BIOLO

GY

PNASPL

US

Page 5: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

HypR could be explained by the presence of a spacer betweentwo inverted repeats in the DNA. B. subtilis YodB binds to DNAcontaining a seven nucleotide spacer, whereas QsrR and HypRbind to DNA with five and two nucleotide spacers, respectively(Fig. S5). The other conformational changes in the YodBMPBQdimer, including shifts in the loop between α3 and α4, the loopbetween β2 and β3, and the α3 helix, were also observed in theQsrR dimer (Fig. S3).

Extensive Conformational Changes in the YodBdiamide Dimer. Sur-prisingly, the dimer structure of YodBdiamide is quite differentfrom any known structure, including S. aureus QsrR andB. subtilis HypR (Fig. 4 and Fig. S3). When we superimpose eachmonomer (monomer A) of YodBreduced and YodBdiamide, themovement of the other monomer (monomer B) of YodBdiamidecan clearly be analyzed. The conformational change in theYodBdiamide dimer is induced by the formation of two disulfidebonds between Cys6 and Cys101′ (or Cys6′ and Cys101) (Fig.4A). These disulfide bonds reorient one monomer (chain B) andchange the overall dimer structure, with a significant trans-location of 37 Å (Fig. 4A). The formation of a disulfide bondbetween Cys6 and Cys101′ is accompanied by a large movementof two helices (α1 and α5′) to produce a favorable disulfide bondbetween two cysteine residues. The formation of disulfide bondsinduces rotation of the α1 and α5′ helices by ∼10° and 30°, re-spectively, with a maximal shift of 11 Å at Glu81 on α5′ (Fig. 4B).To accommodate the formation of the other disulfide bond(Cys6′ and Cys101), the α1′ and α2′ helices are shifted in aperpendicular direction by the α5 axis, with a maximum distanceof 25 Å (Fig. 4B). The translational movement of α1′, α2′, andα5′ is accompanied by the reorientation of chain B, along withthe generation of a new dimer interface. As a result of the for-mation of the two disulfide bonds, the space between α1 and α1′increases from 5 Å to 23 Å, with a ∼50° rotational movement

(Fig. S6). One of the most striking features in the YodBdiamidedimer is a large movement of the DNA-recognition helices (α4and α4′). The distance between α4 and α4′ (between Lys48/Lys48′-Cα atoms) in YodBreduced is ∼39 Å, whereas that in theYodBdiamide is ∼60 Å for the equivalent Lys48-Cα pairs (Fig. 4C).The movement of the two α4 and α4′ helices of YodBdiamide isachieved by significant translational (37 Å) and rotational (56°)shifts following the addition of diamide. In addition, α1 and α1′are reoriented to face each other through a ∼68° rotationalmovement (Fig. 4D). Although there are significant translationaland rotational movements of the secondary structures in YodB,the dimensions of the YodBreduced dimer are nearly identical tothose of YodBdiamide, 35 × 30 × 80 Å. Disulfide bond formationbetween Cys6 and Cys108′ (14) does not seem to be favorablebecause the distance between Cys6 and Cys108′ is expected to bemuch larger than that between Cys6 and Cys101′.

NMR Study of YodBreduced, YodBdiamide, and YodBMPBQ in Solution. Theconformations of YodB in solution were monitored by NMRspectroscopy. Approximately 91.3% (85 of 93, excluding 6 pro-line residues) of the chemical shifts were assigned to individualresidues of B. subtilis YodB5–105 in the reduced state (in thepresence of 2 mM DTT). According to the chemical shift index,the secondary structure of the YodB protein in solution showedno deviation from the crystal structure.Following the addition of diamide, both chemical shift changes

and line broadenings on the backbone amide NH peaks wereobserved in the 2D [1H,15N] transverse relaxation optimizedspectroscopy–heteronuclear single-quantum correlation (TROSY-HSQC) spectra (Fig. 5A), indicating a conformational change inYodB in the intermediate and fast exchange mode on an NMRtime scale. Significant chemical shift changes (which deviate fromthe average by over 1σ using a 1:2 ratio of YodB:diamide) wereobserved for residues on the α1 (Cys6, Ser11, Ala12, Ser14, and

Fig. 4. Structural rearrangement of the YodBdiamide

dimer upon the formation of two disulfide bonds.(A) The large movement of chain B of YodBdiamide isindicated by an arrow in the superimposed view ofYodBreduced and YodBdiamide dimers. Two cysteineresidues (Cys6 and Cys101′) that form the disulfidebond are shown in red and yellow spheres. For abetter display, two chains of YodBreduced are pre-sented in a transparent view. (Right) Surface views ofthe two structures (YodBreduced and YodBdiamide) aresuperimposed. In the black rectangular box, surfaceviews of YodBreduced and YodBdiamide are presentedseparately. Following the addition of diamide, chainB of YodB is reoriented, with a large movement upto 37 Å and a rotation by 65°. The structural changesin YodBdiamide are depicted in B−D in detail. (B) Therotation of α1 and α5′ is caused by the disulfide bondbetween Cys6 and Cys101′. The translational move-ment of α1′ and α2′ is achieved by the disulfide bondbetween Cys6’ and Cys101. (C) The formation of twodisulfide bonds results in a large change in the dis-tance between the two DNA recognition helices (α4and α4′). (D) In chain B of YodBdiamide, α1′ is shiftedto a new position by a translational movementcoupled with a rotation of 68°.

E5206 | www.pnas.org/cgi/doi/10.1073/pnas.1604427113 Lee et al.

Page 6: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

Leu16), α2 (Gly22), and α5 helices (Ala88, Trp96, and Asp98), aswell as the loop between α1 and α2 (Trp20). In addition, signifi-cant line broadenings (top 10% of cross-peaks exhibiting largechanges in intensities using 1:2 ratio of YodB:diamide) were ob-served for residues on α1 (Phe13), α3 (Lys36), α4 (Gln47, Lys48,Ala51, and Glu58), the loop between α3 and α4 (Ser46), α5(Ala85 and Trp96), and the C-terminal loop (Gly104). Whenwe mapped the affected residues onto the crystal structure ofYodBreduced, the residues were clustered in the homodimericinterface of YodB and the DNA recognition helix α4 (Fig. 5 Band C). These results indicate that diamide induced a largestructural change in the homodimeric interface in solution, con-sistent with the crystal structure of YodBdiamide.Using a titration of MPBQ, the overall number of peaks was

significantly reduced in the 2D [1H,15N] TROSY-HSQC spectra.However, in contrast to the diamide titration experiment, mostof the cross-peaks did not exhibit significantly different chemicalshifts compared with those of YodBreduced following the addi-tion of MPBQ (Fig. 5A). The result suggests that YodB un-dergoes only minimal structural changes in solution followingMPBQ binding, which is consistent with the crystal structures ofYodBreduced and YodBMPBQ. Notably, a significant reduction inthe peak intensities (top 10% of cross-peaks exhibiting largechanges in intensities; brought by a 1:2 ratio of YodB:MPBQ)was identified for residues on α1 (Ala12, Phe13, and Ser14), α2(Gly22), and α5 (Thr84, Phe90, Asp98, and Gln99), and the loopbetween α1 and α2 (Lys18 and Arg19) (Fig. 5B). The observeddecrease in the peak intensities suggests that, in its MPBQ-

bound form, the residues on the homodimeric interface of YodB,which consists of the α1, α2, and α5 helices, undergo confor-mational changes at a rate that corresponds to the intermediateNMR time scale. Additionally, when we added a stoichiometricexcess of MPBQ compared with YodB (stoichiometric ratio ofYodB:MPBQ was 1:5), the overall cross-peaks were substantiallybroadened (Fig. S7). The results indicate that YodB may undergodynamic behavior upon MBPQ binding, which may explain theextremely low binding affinity of YodBMPBQ for DNA. Consistentwith this observation, the crystal structure of YodBMPBQ showedmuch higher average B-factors of 65 Å2 compared with 31 Å2

of YodBreduced.To further characterize the interaction between YodB and

DNA in solution, we investigated the binding mode betweenYodB and a 17-bp DNA containing the azoR1 promoter region(ATACTATTTGTAAGTAA) using NMR spectroscopy. Ingeneral, the overall cross-peaks in the 2D [1H,15N] TROSY-HSQC spectrum of B. subtilis YodB showed line broadeningsand not chemical shift changes upon DNA titration, even at thelow DNA:YodB ratio of 0.1:1 (Fig. S8A). This result demon-strated a long rotational correlation time, which is attributed tothe formation of the YodB–DNA complex in solution. In par-ticular, 10 cross-peaks (Ser11, Ala12, Ser14, Gly17, Arg19,Lys36, Met49, Ala51, Leu57, and Val75) in the 2D [1H,15N]TROSY-HSQC spectra were largely broadened upon DNAbinding, showing a more than 70% reduction in the peak in-tensities (Fig. S8B). The residues of the most affected peaks werelocalized in three α-helices (α1, α3, and α4) and one β-strand

Fig. 5. Diamide- and quinone-induced structuralchanges in B. subtilis YodB in solution. (A) (Left)Overlaid [1H,15N] TROSY-HSQC titration spectra ofthe 15N-labeled YodB titrated with different ratios ofdiamide. (Lower Left) The overlaid [1H,15N] TROSY-HSQC spectra of examples of peaks that exhibitchemical shift changes following diamide titration.The arrow in the diagram indicates the direction ofthe chemical shift. (Right) Overlaid [1H,15N] TROSY-HSQC titration spectra of the 15N-labeled YodB ti-trated with different ratios of MPBQ. (Lower Right)Examples of peaks that exhibited chemical shiftchanges following diamide titration. (B, Left) Theratio of the cross-peak intensities (orange) andchemical shift changes (purple) of residues in YodBproduced by diamide binding (2.0 equivalent) wasplotted against the residue number. The ratio of thepeak intensities was normalized to the ratio of thepeak intensity of Glu70. The asterisks indicatethe residues that did not exhibit observable peaksupon diamide titration due to severe broadening.The secondary structural elements of YodBreduced areshown above the plot, where the helices and strandsare indicated by cylinders and arrows, respectively.(B, Right) The ratio of the cross-peak intensities ofresidues in YodB produced by MPBQ-binding (2.0equivalent) was plotted against the residue number.The ratio of the peak intensities was normalized tothe ratio of the peak intensity of Thr68. (C, Left) Thereductions in the signal intensities and chemicalshift changes produced by diamide titration (2.0equivalent) were mapped onto the crystal structureof YodBreduced. The unassigned residues, includingprolines, are indicated as gray spheres. (B, Right)The reduction in the signal intensities produced byMPBQ titration (2.0 equivalent) was mapped ontothe crystal structure of YodBreduced. The unassignedresidues, including prolines, are indicated as grayspheres.

Lee et al. PNAS | Published online August 16, 2016 | E5207

MICRO

BIOLO

GY

PNASPL

US

Page 7: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

(β3), which possess a predominantly positive electrostatic po-tential (Figs. S8C and S9). The DNA binding site of YodBidentified here is consistent with that of DNA-bound S. aureusQsrR (PDB ID code 4QHE).

Insights into the Dissimilar DNA-Dissociation Modes of YodBdiamide

and YodBMPBQ. To investigate the influence of both thiol-reactivecompounds (diamide and MPBQ) on two cysteine residues (Cys6and Cys101) and their effects on the YodB–DNA interaction, weperformed a fluorescence polarization assay to measure the dis-sociation constant (Kd) of the YodBWT and mutant (YodBC6S,YodBC101S, and YodBC6S/C101S) YodB constructs following treat-ment with diamide or MPBQ. The average Kd values from multiplemeasurements are summarized in Fig. 6. The data for YodBWTshow that the reduced form of YodBWT binds to its cognate DNAwith high affinity (Kd of 1.0 ± 0.1 μM), showing preference for theTACT{7}AGTA consensus sequence (Fig. 7A and Fig. S10). Forthe diamide-treated and MPBQ-treated YodBWT, the binding af-finities to the cognate DNA were lower than that of the reducedform of YodBWT (Fig. 7A). In addition, diamide-treated YodB hasa relatively higher DNA-binding affinity (Kd of 5.0 ± 0.9 μM) thanMPBQ-treated YodB (Kd of 32.9 ± 8.1 μM), which may suggestthat the addition of MPBQ allows YodB to be easily dissociatedfrom the DNA compared with diamide-treated YodB. The dis-similarity of DNA-binding affinities between MPBQ-treated YodBand diamide-treated YodB is also consistent with our in vivotranscription assay of the YodB-controlled azoR1 gene (Fig. 7A).The result indicates that the transcription of the azoR1 gene isactivated by much lower concentration of MPBQ (0.01 mM) thandiamide (10 mM). Furthermore, the transcriptional activation ofazoR1 by MPBQ treatment lasts longer due to the irreversiblenature of the MPBQ-derived modification (Fig. 7B). Reversetranscription quantitative real-time quantitative PCR (RT-qPCR)analysis further confirmed that YodB responds to diamide andMPBQ with distinct mechanisms in vivo. The mutation of C101Sresulted in the decreased induction of azoR1 in response to

diamide, whereas the YodBC101S mutant still retained its re-sponsiveness to MPBQ (Fig. 7C). These results clearly show thatdiamide and MPBQ use different mechanisms in vivo, which isconsistent with our observations in vitro. For the C6S and C101Smutants, the binding affinities of the diamide-treated YodB mu-tants were three- to sevenfold higher than those of diamide-treatedYodBWT. However, the binding affinities of the diamide-treatedsingle mutants (C6S or C101S) were not similar to those of theYodBWT. The results deviated somewhat from our dimeric struc-ture of YodBdiamide, because the C6S (or C101S) mutant could notform disulfides to allow the large structural change that permitscomplete dissociation from the DNA. This result may indicate thatthe thiol group of Cys6 (or Cys101) is converted into sulfenic acidby diamide, which influences the dissociation of the YodB dimerfrom the DNA (21, 22). If sulfenic acid formed on the thiol groupof Cys6 (or Cys101), the intermediate state of YodB might tem-porarily stabilize the diamide-induced conformation until thedisulfide bonds are formed. In the YodBC6S/C101S mutant, the Kdvalue for the diamide-treated YodBC6S/C101S is nearly identical tothat for the untreated YodBC6S/C101S, as expected. Following theaddition of MPBQ, the Kd value of the YodBC101S is decreased by5.1-fold compared with that of the MPBQ-treated YodBWT,whereas the Kd of YodBC6S is decreased by 8.6-fold, indicating thatMPBQ has a larger effect on Cys6 than Cys101, as revealed inprevious reports (13). In the double mutant (C6S/C101S) of YodB,the Kd value for the MPBQ-treated YodBC6S/C101S (1.6 ± 0.3 μM)is still fivefold lower than the untreated YodBC6S/C101S (0.3± 0.1 μM),which indicates that an additional MPBQ-induced factor affectsthe oxidation of the YodB protein. To further reveal a minor ef-fect on YodB protein by MPBQ, we performed MALDI-TOF MSanalysis of the YodBC6S/C101S mutant to detect any possiblemodifications made by MPBQ (Fig. S11). The mass of C6S/C101Swas increased by 15 Da after the addition of MPBQ. This changemay result from an oxidation on a noncysteine residue that still af-fects DNA-binding affinity of C6S/C101S. Accordingly, we performedNMR experiments to study the interactions between the YodBC6s/101Sand MPBQ in solution (Fig. S11). Upon the addition of a fivefoldstoichiometric excess of MPBQ over YodB, the overall peakintensities were slightly decreased and the residues that showedline broadenings were distributed over a large fraction of protein.This result suggests that MPBQ can induce a minor but globalconformational change in YodB when in solution and decreasethe binding affinity of YodB to its cognate DNA.

Fig. 6. Fluorescence polarization assays for the DNA binding affinities of (A)the WT, (B) C101S, (C) C6S, and (D) C6S/C101S double mutants of B. subtilisYodB. The YodB binding assay was performed with the azoR1 promoter site.The experiment showed that diamide and MPBQ inhibit the DNA-binding af-finity of YodB through different mechanisms. The experiment with the reducedYodB proteins was performed in the presence of DTT, shown in black. The plotsof diamide-treated and MPBQ-treated YodB are shown in orange and blue,respectively. The average values of triplicate measurements and SDs are shown.

Fig. 7. Changes in the transcript levels for azoR1 in response to diamideand MPBQ in vivo, determined by RT-qPCR. WT or C101S mutant B. subtilisPS832 cells were treated with diamide (orange) or MPBQ (blue), and relativelevels for azoR1were determined. The data represent the mean + SD (n = 3).(A) MPBQ induced expression of azoR1 at a 1000-fold lower concentrationthan diamide. Total RNA was extracted 5 min after the treatment. Thetranscript level for azoR1 before the treatment was taken as 1, and therelative expression levels are shown. (B) WT B. subtilis cells were treated with10 mM diamide or 0.01 mM MPBQ, and total RNA was extracted after theindicated periods of time. The transcript level for azoR1 in the WT cellsbefore the treatment (0 min) was taken as 1, and the relative expressionlevels are shown. (C) C101S mutant cells were treated with 10 mM diamideor 0.01 mM MPBQ, and total RNA was extracted 5 min after the treatment.The transcript level for azoR1 in WT cells without the treatment was taken as1, and the relative expression levels are shown.

E5208 | www.pnas.org/cgi/doi/10.1073/pnas.1604427113 Lee et al.

Page 8: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

DiscussionIn B. subtilis, oxidative stress by RES such as diamide and qui-nones triggers the expression of regulons that are regulated bySpx, CtsR, PerR, CymR, and the MarR/DUF24 family of regu-lators (14, 23). B. subtilis YodB is a prototypical MarR/DUF24family transcriptional regulator that directly senses and respondsto both quinone and diamide (4, 14). Most of the MarR/DUF24family regulators possess a conserved cysteine residue near the Nterminus, and one or two cysteine residues with less-conservedpositions are located near the C terminus (14). Because B. sub-tilis YodB resembles two-Cys type regulators, including X. cam-pestris OhrR (sequence identity of 20.3%) and B. subtilis HypR(sequence identity of 31.0%), the molecular function of YodBwas expected to be very similar to these regulators (19). How-ever, our in-depth insights into the mechanism of B. subtilisYodB show that it undergoes different conformational changes(YodBdiamide or YodBMPBQ) that are individualized to differentredox-related signals. The diamide-mediated and quinone-mediated signaling pathways involving YodB process the stressinformation into distinct functional responses in B. subtilis butthey up-regulate the same subset of genes to inhibit the repressiveaction of YodB. Therefore, we propose the following redoxswitch mechanism for B. subtilis YodB: (i) diamide-mediatedsignaling pathway, which includes the possible diamide-inducedformation of sulfenic acid intermediates on Cys6 (or Cys101),formation of two diamide-induced disulfide bonds between Cys6

and Cys101′ (or Cys6′ and Cys101), full dissociation of YodB fromthe operator DNA, and activation of diamide-detoxification sys-tems and (ii) quinone-mediated S-alkylation on Cys6 of YodB,which includes the possible quinone-induced formation of sulfinicor sulfonic acid intermediates on Cys6 (or Cys101), formation of aquinone adduct on Cys6 and Cys6′, and full dissociation of YodBfrom the operator DNA (Fig. 8). It is surprising that two differentthiol-reactive compounds induce different conformational changesin the structure of the transcription factor YodB (YodBdiamide orYodBMPBQ), which activates gene expression in dissimilar ways.Diamide is sensed by B. subtilis YodB via the formation of in-termolecular disulfide bonds between two cysteine residues (Cys6and C101′/Cys6′ and Cys101) of opposing subunits with largestructural rearrangements. In contrast, MPBQ induces S-alkyl-ation on Cys6 of YodB and enables the dissociation of YodB fromthe target DNA, with minor structural changes that are similar toQsrR and HypR.In addition, the structural changes induced by MPBQ are

relatively minor and more responsive than those induced bydiamide. Therefore, it can be suggested that B. subtilis maydiscriminate two oxidative signals (MPBQ and diamide) usingone YodB regulator and respond to the more toxic compoundMPBQ at much lower concentration than diamide (Fig. S1). Inaddition to this delicate system, the different features or re-versibility of the two reactions (a reversible reaction by di-amide or an irreversible S-adduct formation by quinones) maycontribute to the equilibrated redox state in cells such as B. subtilisfor the efficient management of oxidative shocks. To the best ofour knowledge, our study on YodB provides the first insightsinto a redox regulator that responds to multiple oxidation sig-nals (via intermolecular-disulfide bonds or S-alkylation) withdistinct conformational changes and presents a possible regu-latory mechanism at the molecular level. In summary, this studyprovides structural insights into how B. subtilis YodB sensesmultiple signals and regulates gene expression in distinctpathways. It also provides a structural basis for the relationshipbetween ligand-induced conformational changes and its func-tional switch.

Materials and MethodsGene Cloning. The primers used in the study are listed in Table S1. The residuesMet5–Asp105 (YodB5–105) of the yodB (BSU19540) gene were amplified fromthe B. subtilis genomic DNA (strain 168) (24) by PCR using the primers yodB-F/yodB-R. The gene products for the three mutants (YodBC6S, YodBC101S, andYodBC6S/C101S) were obtained by PCR using the primers yodBC6S-F/yodB-R,yodB-F/yodBC101S-R, and yodBC6S-F/yodBC101S-R, respectively. The ampli-fied DNA was inserted into the pET-28a(+) expression vector (Novagen) thathad been digested with both NdeI and XhoI. The four constructs of YodBinclude a 21-residue hexa-histidine tag (MGSSHHHHHHSSGLVPRGSHM) atthe amino terminus of the recombinant protein to facilitate protein purifi-cation. The resulting constructs were verified by DNA sequencing.

Protein Expression and Purification. The purification of the WT and threemutant YodB5–105 proteins was nearly identical, except for the treatmentwith the oxidizing reagents (diamide or MPBQ). The YodB proteins wereoverexpressed in E. coli strain BL21(DE3). The cells were grown in LB culturemedium containing 50 μg/mL kanamycin at 37 °C until they reached anOD600 of 0.5, and protein expression was induced by treating the cells with0.5 mM isopropyl-β-D-thiogalactopyranoside for 4 h at 37 °C. The cells werethen harvested by centrifugation at 5,500 × g for 10 min at 4 °C. The cellpellet was resuspended and lysed by sonication in buffer A (50 mM Tris·HCl,pH 8.0, and 500 mM NaCl) containing 10% (vol/vol) glycerol and EDTA-freeComplete Protease Inhibitor Mixture (Roche). The cell debris was removedand discarded by centrifugation at 18,000 × g for 1 h at 4 °C. The super-natant was applied to a nickel-nitrilotriacetic acid-agarose affinity chroma-tography column (Novagen) that had been equilibrated in buffer A. Theprotein was eluted with buffer A containing 200 mM imidazole. To purifyYodBreduced, the eluted sample was diluted 10-fold with buffer B (20 mMTris·HCl, pH 8.0) containing 2 mM β-mercaptoethanol. The diluted proteinsample was loaded onto a Hiprep Q column (GE Healthcare) that had beenpreequilibrated with buffer B. The sample was eluted with a gradient of

Fig. 8. Proposed redox switch mechanism for B. subtilis YodB. The two path-ways of the YodB protein are depicted as (1) the diamide-mediated signalingpathway and (2) quinone-mediated S-alkylation, with each possible in-termediate form. The diamide-mediated signaling pathway is reversible,whereas the quinone-mediated S-alkylation is irreversible, as indicated by thearrows. YodB is regulated either by the formation of two diamide-induceddisulfide bonds between Cys6 and Cys101′ (or Cys6′ and Cys101) or the for-mation of a quinone adduct on Cys6 and Cys6′. In contrast to the reversiblereaction of the diamide-mediated pathway, low concentration of quinonesirreversibly dissociate YodB from the target DNA. The intermediate diamide-induced form may have oxidized the sulfur atoms on the cysteine residues,whereas the intermediate form indicated by an asterisk may be induced byunknown factors as well as the oxidation of cysteine residues. Finally, thederepression of the YodB protein on the target DNA by the two oxidativereagents allows the transcription of oxidation-detoxifying genes. The modelof the DNA-bound YodBreduced structure was generated using the High Am-biguity Driven Docking algorithm (HADDOCK) (43). Cysteine residues that areaffected by the oxidative signals are shown in orange.

Lee et al. PNAS | Published online August 16, 2016 | E5209

MICRO

BIOLO

GY

PNASPL

US

Page 9: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

50–800 mM NaCl in buffer B. To cleave the N-terminal hexa-histidine tag, 100units of thrombin from human plasma (Sigma-Aldrich) were added to 10 mgof the eluted protein and incubated for 12 h at 20 °C. The cleaved protein wasapplied to a HiLoad 16/60 Superdex 75 prep-grade column (GE Healthcare)that had been equilibrated with buffer C (20 mM Tris·HCl at pH 8.0 and150 mM NaCl) containing 1 mM DTT and 1 mM EDTA, as a final purificationstep. For crystallization, the purified protein was concentrated to 18.3 mg/mLusing an Amicon Ultra-15 centrifugal filter unit (Millipore).

The YodBdiamide protein was expressed and purified essentially as de-scribed for YodBreduced, with the exception of the reaction with diamide andbuffer composition. After purification on a nickel-nitrilotriacetic acid-aga-rose affinity column, the protein sample was incubated with 1 mM diamidefor 1 h at room temperature and then diluted 10-fold with buffer B. Thediluted protein sample was loaded onto a Hiprep Q column (GE Healthcare)that had been preequilibrated with buffer B. The sample was eluted with agradient of 50–800 mM NaCl in buffer B. The cleavage of the N-terminalhexa-histidine tag was identical to that for YodBreduced. The cleaved proteinwas applied to a HiLoad 16/60 Superdex 75 prep-grade column (GEHealthcare) that had been equilibrated with buffer C, as a final purificationstep. For crystallization, the purified protein was concentrated to 16.5 mg/mLusing an Amicon Ultra-15 centrifugal filter unit (Millipore).

The YodBMPBQ protein was expressed and purified essentially as described forYodBreduced, with the exception of the addition of MPBQ after purification. Topromote homogeneous conjugation following the addition of MPBQ, we usedthe YodBC101S mutant instead of the WT YodB protein, which is similar to theprocedure for menadione-modified QsrR (4). After purification by SEC, 600 μMpurified protein was incubated with 1.5 mM MPBQ for 1 h at room tempera-ture and was then further purified on a HiLoad 16/60 Superdex 75 prep-gradecolumn (GE Healthcare) that had been equilibrated with buffer C. For crystal-lization, the purified protein was concentrated to 24 mg/mL using an AmiconUltra-15 centrifugal filter unit (Millipore).

For NMR spectroscopy, the homogeneous 15N-labeled YodB5–105 proteinswere produced in E. coli BL21 (DE3) cells using M9 minimal media containing1 g/L 15NH4Cl. The 2H-, 15N-, and 13C-labeled proteins were producedby growing E. coli BL21 (DE3) cells in M9 minimal media containing 1 g/L15NH4Cl and 1.5 g/l 13C6-glucose in ∼99% D2O instead of H2O. The purifica-tion procedures for the uniformly labeled YodB5–105 proteins are identical tothose for YodBreduced. For the NMR measurements, the purified proteinswere concentrated using an Amicon Ultra-15 centrifugal filter unit (Millipore)and the buffer was exchanged to 20 mM MES, pH 6.5, containing 150 mMNaCl, 2 mM DTT, and 1 mM EDTA. Ten percent D2O was added to the samplebefore it was loaded into a Shigemi tube.

Crystallization and X-Ray Data Collection. YodBreduced, YodBdiamide, andYodBMPBQ were crystallized in 96-well crystallization plates at 293 K using thesitting-drop vapor-diffusion method. Each sitting drop was prepared by mixing0.5 μL each of the protein solution and the reservoir solution [0.2 M sodiumbromide and 20% (wt/vol) PEG 3,350 for YodBreduced; 0.1 M Bis-Tris at pH 5.5,200 mM lithium sulfate, and 25% (wt/vol) PEG 3,350 for YodBdiamide; and 0.2 Msodium citrate tribasic dihydrate and 30% (wt/vol) PEG 3,350 for YodBMPBQ] andwas placed over 70 μL of the reservoir solution. The crystals were vitrified using acryoprotectant solution that consisted of the reservoir solution supplementedwith 20% (vol/vol) glycerol. The crystals were soaked in the cryoprotectant solu-tion for a few seconds before being frozen in liquid nitrogen. A set of X-raydiffraction data for the YodBreduced crystal was collected at 100 K on a Quantum270 CCD area detector (Area Detector Systems Corporation) at the BL-7A exper-imental station of the Pohang Light Source, Korea. The YodBreduced crystal be-longs to the monoclinic space group P21, with unit cell parameters of a = 40.78 Å,b = 50.80 Å, c = 50.35 Å, β = 95.49°. The X-ray diffraction data for both theYodBdiamide and YodBMPBQ crystals were collected at 100 K on a Quantum 315rCCD area detector at the BL-5C experimental station of the Pohang Light Source,Korea. The YodBdiamide crystal belongs to the hexagonal space group P62, withunit cell parameters of a = b = 94.98 Å, c = 25.42 Å. The YodBMPBQ crystal belongsto the monoclinic space group P21, with unit cell parameters of a = 38.94 Å, b =51.17 Å, c= 48.39 Å, β = 93.46°. The raw datawere processed and scaled using theHKL2000 program (25). Table 1 summarizes the data collection statistics.

Structure Determination, Refinement, and Analysis. The crystal structures ofB. subtilis YodB were determined using the molecular replacement method inMOLREP (26), which used a monomer model of S. aureus QsrR as a searchmodel (4). Two models of B. subtilis YodB were further refined with theREFMAC (27) and PHENIX (28) programs, including bulk solvent correction. Themodel was manually constructed and water molecules were added usingthe Coot program (29). Five percent of the data were randomly set aside as thetest data for the calculation of Rfree (30). The stereochemistry of the final

structures was evaluated using MolProbity (31). The overall geometry of thefinal models of YodBreduced, YodBdiamide, and YodBMPBQ ranked in the 97th,100th, and 98th percentiles, withMolProbity scores of 1.29, 1.02, and 1.47, wherethe 100th percentile is the best among structures of comparable resolution. Thestructural deviations were calculated using Superpose (32). The solvent-accessiblesurface areas were calculated using PISA (33). The protein–protein interactionswere calculated using the Protein Interactions Calculator (34).

NMR Spectroscopy. All NMR experiments were conducted at 298 K on aBruker 800-MHz or 900-MHz NMR spectrometer equipped with cryogenicprobes. The backbone assignments of HN, N, C′, Cα, and Cβ were obtainedfrom the 3D HNCO, HN(CA)CO, HNCA, HN(CO)CA, HNCACB, and HN(CO)CACB spectra. The chemical shifts were externally referenced to DSS. All2D and 3D NMR datasets were processed with NMRPipe (35) and analyzedin NMRView (36). The titration of 15N YodB with diamide was performedwith 500 μM 15N YodB using stocks of 100 mM diamide in 20 mM MES,pH 6.5, 150 mM NaCl, 2 mM DTT, and 1 mM EDTA, which was used torecord the [1H,15N] TROSY-HSQC spectra. The titration of 15N YodB withMPBQ was performed with 500 μM 15N-labeled YodB using stocks of50 mM MPBQ in 20 mM MES, pH 6.5, 150 mM NaCl, 2 mM DTT, and 1 mMEDTA, which was used to record the [1H,15N] TROSY-HSQC spectra. Thebinding of YodBreduced to the double-strand 17-bp DNA containing theazoR1 promoter region (ATACTATTTGTAAGTAA) was investigated bycomparing the [1H,15N] TROSY-HSQC spectra of 500 μM 15N-labeled YodBin 20 mM MES, pH 6.5, 150 mM NaCl, 2 mM DTT, and 1 mM EDTA in thepresence or absence of DNA.

Fluorescence Polarization Assay. The 6-FAM–labeled 17-bp double-strand DNAcontaining the azoR1 promoter region (ATACTATTTGTAAGTAA) was pur-chased from Bioneer. To determine the DNA-binding affinities of YodB5–105 inthe reduced state, a 100 nM solution of the 6-FAM–labeled DNA was incubatedwith increasing amounts of the purified proteins in buffer D (20 mM Tris·HCl atpH 7.5 and 150 mM NaCl) containing 2 mM DTT for 30 min at 25 °C. To de-termine the DNA-binding affinities of YodB5–105 in the oxidized state, theproteins were incubated with 5 mM diamide or 5 mM MPBQ, which werefurther dialyzed against buffer D before incubation with a 100 nM solution ofthe 6-FAM–labeled DNA. The fluorescence polarization signals were recordedusing black 384-well plates (Greiner) on a SpectraMax M5e microplate reader(Molecular Devices) with a 485-nm excitation filter and a 520-nm emission filter.The data were analyzed with KaleidaGraph (Synergy Software) using theequation ΔFP = FP − FPfree = (ΔFPmax [protein])/Kd + [protein]), where FPfree isthe background polarization signal (no protein, measured), ΔFPmax is themaximum polarization change (calculated), and [protein] is the protein con-centration and Kd is the dissociation constant. The bound fractions were cal-culated as ΔFP/ΔFPmax. Each experiment was performed in triplicate.

SEC-MALS. SEC was performed on a BioSep SEC-s3000 size-exclusion column(Phenomenex) using a 1260 infinity HPLC system (Agilent Technologies), andMALS was measured inline using a miniDAWN-TREOS instrument with anemission at 657.4 nm (Wyatt Technology). The scattering data were analyzedwith ASTRA 6.0.1.10 software (Wyatt Technology). To determine the mul-timeric state of YodB5–105 in the reduced state, the protein was analyzed inbuffer C containing 2 mM DTT at room temperature. To determine themultimeric states of YodB5–105 in the oxidized state, the proteins were in-cubated with 5 mM diamide or 5 mM MPBQ at room temperature, whichwere further dialyzed against buffer C and loaded onto a column that hadbeen preequilibrated with buffer C.

Cell Growth Curve Measurement. B. subtilis strain PS832, a prototrophic de-rivative of strain 168, was used for the growth curve measurements. Thelong-flanking homologous recombination method was performed to deletethe yodB gene in B. subtilis (37) using the ΔyodB1_F/ΔyodB1_R andΔyodB2_F/ΔyodB2_R primers. For complementation of the WT yodB gene,the target DNA was amplified and included its native promoter and termi-nator using the pDG1730_yodB_F and pDG1730_yodB_R primers. The am-plified DNA was inserted into the pDG1730 vector that had been digestedwith both BamHI and EcoRI. The insertion of the vector containing the yodBgene into B. subtilis was performed as previously described (38). The clonedsamples were spread onto LB agar plates supplemented with 5 μg/mLchloramphenicol and 100 μg/mL spectinomycin, which were then incubatedovernight at 37 °C. For the growth curve measurements of B. subtilis strainPS832 and the ΔyodB mutant, the cells from each single colony were dilutedinto 20 mL prewarmed Spizizen minimal medium (39) in 50-mL conical tubes(SPL Life Sciences) and incubated at 37 °C with agitation at 180 rpm. Whenthe OD600 reached 0.4 (time 0), diamide, MHQ, or MPBQ from freshly

E5210 | www.pnas.org/cgi/doi/10.1073/pnas.1604427113 Lee et al.

Page 10: Two distinct mechanisms of transcriptional regulation by ... · Two distinct mechanisms of transcriptional regulation by the redox sensor YodB Sang Jae Leea,1, In-Gyun Leea,1, Ki-Young

prepared stock solutions were added to the medium and the OD600 was mea-sured every 30 min for up to 2 h (60 min over the next 2 h).

RT-qPCR Analysis. The yodB C101S mutant was generated using a site-directedmutagenesis kit (Stratagene) based on the complementation construct(pDG1730:yodB) with the pDG1730_C101S_F/ pDG1730_C101S_R primer set.B. subtilis strains were grown in LB medium at 37 °C overnight and diluted1:100 into LB medium. The cells were grown until they reached an OD600 of0.4, followed by the addition of diamide or MPBQ. Total RNA from bothtreated and untreated cells was isolated using an RNeasy Mini Kit (Qiagen)with additional treatment with RNase free DNase I (Qiagen) following themanufacturer’s instructions. RT-qPCR was performed using a One Step SYBRPrimeScript PLUS RT-PCR kit (Takara) in an Applied Biosystems 7300 RealTime PCR System with the following primers specific to azoR1 and 23s rRNA:RT-qPCR_azoR1_F/ RT-qPCR_azoR1_R and RT-qPCR_23s rRNA_F/ RT-qPCR_23srRNA_R, respectively. The 23s rRNA gene was used as an endogenous control,and the relative fold change in azoR1 gene expression was calculated usingthe comparative CT (2−ΔΔCT) method (40).

Determination of Minimum Inhibitory Concentration. The WT or ΔyodB mu-tant B. subtilis PS832 cells were tested using in vitro susceptibility tests[minimum inhibitory concentration (MIC)]. The MIC tests were performedusing the Clinical and Laboratory Standards Institute (CLSI) [formerlyNational Committee for Clinical Laboratory Standards (NCCLS)] brothmicroplate method (NCCLS, 2003) with a starting inoculum of ∼106 cfu/mLfor all isolates (41, 42). The cells were cultured in LB broth at 37 °C for 24 h.

MIC was defined as the lowest concentration of antimicrobial agent thatinhibited visible growth. The results of the MIC tests are summarized inFig. S1.

In Silico DNA-YodBreduced Docking. Lacking crystallographic data for the in-teraction between YodBreduced and its cognate DNA, an in silico moleculardocking study was performed using the High Ambiguity Driven protein–protein Docking algorithm (HADDOCK) (43). The coordinates for theYodBreduced protein were taken from the current crystal structurewithout modifications, and coordinates for a DNA molecule spanning 17 bp(sequence 5′-ATACTATTTGTAAGTAA-3′) were modeled ab initio using themodel.it server (44). The residues Gln47, Lys48, and Glu52 of the recognitionhelix (α4 and α4′) in the YodBreduced dimer as well as the base pairs thymine2–thymine5 and adenine13–adenine16 within two consecutive major grooves ofthe DNA were defined as “active residues,” which are required to have aninterface contact of ambiguous distance. Passive residues were defined auto-matically as residues around active residues.

ACKNOWLEDGMENTS. We thank the beamline (BL) staff members at thePohang Light Source, Korea (BL-5C and BL-7A); and Photon Factory, Japan(BL-5A, BL-17A, and NW12) for assistance with the X-ray diffraction experi-ments. This work was funded by Korea Ministry of Science, Information,Communication, Technology, and Future Planning and National ResearchFoundation (NRF) of Korea Grants NRF-2014K1A3A1A19067618 and NRF-2015R1A2A1A05001894 (to B.-J.L.) and NRF-2013R1A1A2062813 (to S.J.L.).This work was also supported by the 2014 BK21 Plus Project for Medicine,Dentistry, and Pharmacy.

1. Antelmann H, Helmann JD (2011) Thiol-based redox switches and gene regulation.Antioxid Redox Signal 14(6):1049–1063.

2. Imlay JA (2013) The molecular mechanisms and physiological consequences of oxi-dative stress: Lessons from a model bacterium. Nat Rev Microbiol 11(7):443–454.

3. Ortiz de Orué Lucana D, Wedderhoff I, Groves MR (2012) ROS-mediated signalling inbacteria: Zinc-containing Cys-X-X-Cys redox centres and iron-based oxidative stress.J Signal Transduct 2012:605905.

4. Ji Q, et al. (2013) Molecular mechanism of quinone signaling mediated throughS-quinonization of a YodB family repressor QsrR. Proc Natl Acad Sci USA 110(13):5010–5015.

5. Choi H, et al. (2001) Structural basis of the redox switch in the OxyR transcriptionfactor. Cell 105(1):103–113.

6. Kim SO, et al. (2002) OxyR: A molecular code for redox-related signaling. Cell 109(3):383–396.

7. Newberry KJ, Fuangthong M, Panmanee W, Mongkolsuk S, Brennan RG (2007)Structural mechanism of organic hydroperoxide induction of the transcription regu-lator OhrR. Mol Cell 28(4):652–664.

8. Panmanee W, Vattanaviboon P, Poole LB, Mongkolsuk S (2006) Novel organic hy-droperoxide-sensing and responding mechanisms for OhrR, a major bacterial sensorand regulator of organic hydroperoxide stress. J Bacteriol 188(4):1389–1395.

9. Fuangthong M, Helmann JD (2002) The OhrR repressor senses organic hydroperoxidesby reversible formation of a cysteine-sulfenic acid derivative. Proc Natl Acad Sci USA99(10):6690–6695.

10. Chen H, et al. (2010) Structural insight into the oxidation-sensing mechanism of theantibiotic resistance of regulator MexR. EMBO Rep 11(9):685–690.

11. Lee C, Shin J, Park C (2013) Novel regulatory system nemRA-gloA for electrophilereduction in Escherichia coli K-12. Mol Microbiol 88(2):395–412.

12. Gray MJ, Li Y, Leichert LI, Xu Z, Jakob U (2015) Does the transcription factor NemR usea regulatory sulfenamide bond to sense bleach? Antioxid Redox Signal 23(9):747–754.

13. Leelakriangsak M, et al. (2008) Regulation of quinone detoxification by the thiolstress sensing DUF24/MarR-like repressor, YodB in Bacillus subtilis. Mol Microbiol67(5):1108–1124.

14. Chi BK, et al. (2010) The redox-sensing regulator YodB senses quinones and diamidevia a thiol-disulfide switch in Bacillus subtilis. Proteomics 10(17):3155–3164.

15. Rodriguez CE, Fukuto JM, Taguchi K, Froines J, Cho AK (2005) The interactions of 9,10-phenanthrenequinone with glyceraldehyde-3-phosphate dehydrogenase (GAPDH), apotential site for toxic actions. Chem Biol Interact 155(1-2):97–110.

16. Antelmann H, Hecker M, Zuber P (2008) Proteomic signatures uncover thiol-specificelectrophile resistance mechanisms in Bacillus subtilis. Expert Rev Proteomics 5(1):77–90.

17. Cavelier G, Amzel LM (2001) Mechanism of NAD(P)H:quinone reductase: Ab initiostudies of reduced flavin. Proteins 43(4):420–432.

18. Holm L, Rosenström P (2010) Dali server: Conservation mapping in 3D. Nucleic AcidsRes 38(Web Server issue):W545-9.

19. Palm GJ, et al. (2012) Structural insights into the redox-switch mechanism of theMarR/DUF24-type regulator HypR. Nucleic Acids Res 40(9):4178–4192.

20. Alekshun MN, Levy SB, Mealy TR, Seaton BA, Head JF (2001) The crystal structure ofMarR, a regulator of multiple antibiotic resistance, at 2.3 A resolution. Nat Struct Biol8(8):710–714.

21. Gray MJ, Wholey WY, Jakob U (2013) Bacterial responses to reactive chlorine species.Annu Rev Microbiol 67:141–160.

22. Hillion M, Antelmann H (2015) Thiol-based redox switches in prokaryotes. Biol Chem396(5):415–444.

23. Chi BK, Kobayashi K, Albrecht D, Hecker M, Antelmann H (2010) The paralogousMarR/DUF24-family repressors YodB and CatR control expression of the catechol di-oxygenase CatE in Bacillus subtilis. J Bacteriol 192(18):4571–4581.

24. Kunst F, et al. (1997) The complete genome sequence of the gram-positive bacteriumBacillus subtilis. Nature 390(6657):249–256.

25. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in os-cillation mode. Macromolecular Crystallography Pt A 276:307–326.

26. Vagin A, Teplyakov A (2010) Molecular replacement with MOLREP. Acta Crystallogr DBiol Crystallogr 66(Pt 1):22–25.

27. Vagin AA, et al. (2004) REFMAC5 dictionary: Organization of prior chemical knowl-edge and guidelines for its use. Acta Crystallogr D Biol Crystallogr 60(Pt 12 Pt 1):2184–2195.

28. McCoy AJ, et al. (2007) Phaser crystallographic software. J Appl Cryst 40(Pt 4):658–674.29. Emsley P, Lohkamp B, Scott WG, Cowtan K (2010) Features and development of Coot.

Acta Crystallogr D Biol Crystallogr 66(Pt 4):486–501.30. Brünger AT (1992) Free R value: A novel statistical quantity for assessing the accuracy

of crystal structures. Nature 355(6359):472–475.31. Chen VB, et al. (2010) MolProbity: All-atom structure validation for macromolecular

crystallography. Acta Crystallogr D Biol Crystallogr 66(Pt 1):12–21.32. Krissinel E, Henrick K (2004) Secondary-structure matching (SSM), a new tool for fast

protein structure alignment in three dimensions. Acta Crystallogr D Biol Crystallogr60(Pt 12 Pt 1):2256–2268.

33. Krissinel E, Henrick K (2007) Inference of macromolecular assemblies from crystallinestate. J Mol Biol 372(3):774–797.

34. Tina KG, Bhadra R, Srinivasan N (2007) PIC: Protein Interactions Calculator. NucleicAcids Res 35(Web Server issue):W473-6.

35. Delaglio F, et al. (1995) NMRPipe: A multidimensional spectral processing systembased on UNIX pipes. J Biomol NMR 6(3):277–293.

36. Johnson BA, Blevins RA (1994) NMR View: A computer program for the visualizationand analysis of NMR data. J Biomol NMR 4(5):603–614.

37. Wach A (1996) PCR-synthesis of marker cassettes with long flanking homology re-gions for gene disruptions in S. cerevisiae. Yeast 12(3):259–265.

38. Kunst F, Rapoport G (1995) Salt stress is an environmental signal affecting degrada-tive enzyme synthesis in Bacillus subtilis. J Bacteriol 177(9):2403–2407.

39. Spizizen J (1958) Transformation of biochemically deficient strains of Bacillus subtilisby deoxyribonucleate. Proc Natl Acad Sci USA 44(10):1072–1078.

40. Livak KJ, Schmittgen TD (2001) Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25(4):402–408.

41. Jorgensen JH, Ferraro MJ (1998) Antimicrobial susceptibility testing: General princi-ples and contemporary practices. Clin Infect Dis 26(4):973–980.

42. National Committee for Clinical Laboratory Standards (2003) Methods for DilutionAntimicrobial Susceptibility Tests for Bacteria That Grow Aerobically (National Com-mittee for Clinical Laboratory Standards, Wayne, PA).

43. Dominguez C, Boelens R, Bonvin AM (2003) HADDOCK: A protein-protein dockingapproach based on biochemical or biophysical information. J Am Chem Soc 125(7):1731–1737.

44. Munteanu MG, Vlahovicek K, Parthasarathy S, Simon I, Pongor S (1998) Rod models ofDNA: Sequence-dependent anisotropic elastic modelling of local bending phenom-ena. Trends Biochem Sci 23(9):341–347.

45. DeLano WL (2002) The PyMOL molecular graphics system (DeLano Scientific, PaloAlto, CA).

46. Karplus PA, Diederichs K (2012) Linking crystallographic model and data quality.Science 336(6084):1030–1033.

Lee et al. PNAS | Published online August 16, 2016 | E5211

MICRO

BIOLO

GY

PNASPL

US