Structural Insights into Oxygen-Dependent Signal Transductionwithin Globin Coupled SensorsShannon Rivera,†,§ Paul G. Young,†,§ Eric D. Hoffer,‡ Gregory E. Vansuch,† Carmen L. Metzler,†

Christine M. Dunham,‡ and Emily E. Weinert*,†

†Department of Chemistry, Emory University, Atlanta, Georgia 30322, United States‡Department of Biochemistry, Emory University, Atlanta, Georgia 30322, United States

*S Supporting Information

ABSTRACT: In order to respond to external stimuli, bacteria haveevolved sensor proteins linking external signals to intracellularoutputs that can then regulate downstream pathways andphenotypes. Globin coupled sensor proteins (GCSs) serve to linkenvironmental O2 levels to cellular processes by coupling a heme-containing sensor globin domain to a catalytic output domain.However, the mechanism by which O2 binding activates theseproteins is currently unknown. To provide insights into the signalingmechanism, two distinct dimeric complexes of the isolated globindomain of the GCS from Bordetella pertussis (BpeGlobin) weresolved via X-ray crystallography in which differences in ligand-boundstates were observed. Both monomers of one dimer contain Fe(II)−O2 states, while the other dimer consists of the Fe(III)−H2O andFe(II)−O2 states. These data provide the first molecular insights intothe heme pocket conformation of the active Fe(II)−O2 form of these enzymes. In addition, heme distortion modes and heme−protein interactions were found to correlate with the ligation state of the globin, suggesting that these conformational changesplay a role in O2-dependent signaling. Fourier transform infrared spectroscopy (FTIR) of the full-length GCS from B. pertussis(BpeGReg) and the closely related GCS from Pectobacterium carotovorum ssp. carotovorum (PccGCS) confirmed the importanceof an ordered water within the heme pocket and two distal residues (Tyr43 and Ser68) as hydrogen-bond donors. Takentogether, this work provides mechanistic insights into BpeGReg O2 sensing and the signaling mechanisms of diguanylate cyclase-containing GCS proteins.

■ INTRODUCTION

Sensor proteins are used throughout biology to transmitchanges in signal concentrations throughout an organism.However, the mechanisms by which sensor proteins transmitthe ligand binding event into a downstream signal remainpoorly understood.1,2 As the presence of molecular oxygen(O2) is an important determinant of bacterial metabolism aswell as reactive oxygen species-mediated host immuneresponses, it is essential that bacteria be able to respond tochanging O2 levels.3−6 The globin coupled sensor (GCS)family of bacterial heme sensor proteins monitors O2 levels andmodulates downstream pathways for adaption,1 including O2-dependent biofilm formation,5,7−9 motility,7,9,10 and viru-lence.7,9−11 Because biofilms are a primary defense againstantibiotic treatment, host immune responses, and environ-mental stress, understanding the mechanisms by which O2levels regulate GCS activity has the potential to identify newantibacterial targets.1,2,12

GCS proteins are widely distributed throughout bacteria, aswell as in a number of archaea, suggesting their widespreadimportance in controlling O2-dependent bacterial pheno-types.13 These proteins consist of an N-terminal sensor globin

domain linked through a variable middle domain to one ofseveral output domains, such as methyl accepting chemotaxisproteins (MCP), diguanylate cyclases (DGC), phosphodies-terases, sulfate transport and anti-anti-σ factor domains, andkinases (Figure 1A).3,4,11,13−18 Within the GCS family, anumber of putative proteins are predicted to contain DGCoutput domains, which are the enzymes responsible forsynthesizing c-di-GMP, a bacterial second messenger that isa major regulator of biofilm formation.7,10 To date, ligandbinding to the heme within the sensor globin domain of GCSshas been shown to alter activity of all GCS proteins withenzymatic output domains.4−6,11,15,19,20

While the ligand-dependent enzymatic functions of theseproteins have been investigated, the mechanism of activationfor GCS proteins remains unknown.4,6,11,15,21 This is in partdue to subtle differences between closely related GCSproteins.1 GCS proteins containing a DGC domain mustdimerize to exhibit activity;22,23 however, GCS proteins fromBordetella pertussis (BpeGReg) and Pectobacterium carotovorum

Received: September 11, 2018Published: October 31, 2018

Article

pubs.acs.org/ICCite This: Inorg. Chem. 2018, 57, 14386−14395

© 2018 American Chemical Society 14386 DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

Dow

nloa

ded

via

EM

OR

Y U

NIV

on

July

16,

201

9 at

13:

47:1

8 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

(PccGCS) exist in different mixtures of oligomeric states (i.e.,monomer−dimer−tetramer and dimer−tetramer−octamer,respectively),4,6,19,24,25 while the GCS from Desulfotaleapsychrophila (HemDGC) is only found as a tetramer.6 Forboth BpeGReg and PccGCS, the tetrameric assemblies with O2bound exhibit the highest activity.4,25 Furthermore, differencesin dimerization affinity of the isolated globin domains weredemonstrated to alter O2 binding kinetics,

26 suggesting that theoligomeric state alters the heme pocket environment and thusplays an important role in signaling.1,24,25 Activation of GCSproteins is at least partially mediated by changes at the globindimer interface, and by extension changes in the oligomericstates of these proteins. This has been corroborated by workon the GCS from Anaeromyxobacter sp. Fw109-5 (AfGcHK),in which hydrogen/deuterium exchange coupled with massspectrometry (HDX-MS) was used to demonstrate that solventexposure of the globin dimer interface correlates with ligandbinding.27 Despite the variety of output domains and nativeoligomeric populations, the similar ligand-dependent responsesof these GCSs suggest a common signaling pathway.In previous work, crystallization of globin domains from

GCS proteins has been performed in either the Fe(II)−CO orFe(III)−CN ligation state to ensure extended heme stability,as the Fe(II)−O2 state is prone to autoxidation.20,24,27

However, the Fe(II)−O2 state is typically the high-activityform of GCS proteins, so this ligation state is the mostinformative for elucidating biologically relevant signalingmechanisms.1 BpeGReg and PccGCS have greater oxygenbinding stability than other GCS proteins, such as EcDosC,HemAT-Bs, and AfGcHK, as demonstrated by their loweroxygen dissociation rates and their ability to remain in the

Fe(II)−O2 state for extended periods.4,11,15,28,29 Furthermore,the isolated globin domains from BpeGReg and PccGCS(BpeGlobin and PccGlobin) have been shown to be even morestable than their full-length counterparts.26 For example,whereas BpeGReg and PccGCS can remain Fe(II)−O2 at 4°C for ∼18 h before showing significant autoxidation,BpeGlobin and PccGlobin remain fully Fe(II)−O2 after 72 hat 20 °C.4,25,26 Consequently, BpeGlobin and PccGlobin areexcellent systems to study the mechanism of O2 sensing inthese important signaling proteins.To probe the mechanism of O2-dependent activation and

the molecular basis for subtle differences between closelyrelated GCS proteins, X-ray crystallography and FTIR wereperformed on BpeGReg, PccGCS, and their protein variants.These studies provide novel insights into potential signalingpathways, the effects of ligand binding, the heme pocketenvironments, and key differences between the two relatedproteins in the GCS family.

■ METHODSProtein Expression. Codon-optimized genes encoding BpeGReg

and PccGCS in pET-20b4 served as templates for site-directedmutagenesis. Mutagenic primer sequences can be found in SupportingInformation Table S1. Isolated globin and full-length distal pocketmutants were produced by standard PCR protocol as previouslydescribed.26 Proteins were expressed in E. coli Tuner (DE3) pLysScells (Novagen) as previously described.25,26 Briefly, cells weretransformed with codon-optimized pET20b plasmids encoding eachprotein via heat shock and grown overnight on LB agar platescontaining chloramphenicol (30 μg mL−1) and ampicillin (100 μgmL−1). Transformants were then grown in Luria broth (LB) mediumwith the appropriate antibiotics at 37 °C with shaking at 225 rpm.These overnight cultures were used to inoculate yeast extractexpression media (RPI) and grown to an OD600 of 0.6. At thispoint, the temperature was dropped to 25 °C, and δ-aminolevulinicacid was added at a final concentration of 50 mM in each expressionmixture and left to shake. After 30 min, protein expression wasinduced by the addition of 100 μM IPTG for 6 h. Cells wereharvested by centrifugation (3500 × g, 4 °C, 20 min) and theresulting pellets were collected and stored at −80 °C.

Protein Purification. Cell pellets were resuspended in lowimidazole buffer (50 mM Tris, 50 mM NaCl, 1 mM DTT, 20 mMimidazole, pH 7.0), lysed using a homogenizer (Avestin), and partiallypurified by centrifugation (186 000 × g, 4 °C, 1 h). The supernatantswere loaded onto an equilibrated HisPur Ni2+-column (FisherScientific) and washed with low imidazole buffer. Proteins wereeluted with high imidazole buffer (50 mM Tris, 50 mM NaCl, 1 mMDTT, 250 mM imidazole, pH 7.0) and further purified and desaltedusing a S200 gel filtration column (50 mM Tris, 50 mM NaCl, 1 mMDTT, 5% glycerol (v/v), pH 7.0). Proteins were concentrated byultrafiltration (10 kDa MWCO filter, Millipore), aliquoted, flashfrozen in liquid nitrogen, and stored at −80 °C.

BpeGlobin Crystallization and Structural Determination.Purified protein was reduced using sodium hydrosulfite salt under ananaerobic atmosphere (Coy Laboratories).25 BpeGlobin was desalted,oxygenated by mixing with aerobic buffer, and concentrated to 5 mg/mL. Crystals formed in 5−20% polyethylene glycol (PEG) 3350, with0.2 M calcium acetate and 10 mM Tris, at pH 7.0. BpeGlobin wasmixed with the crystallization solution in equal and increasingamounts (i.e., 1 μL + 1 μL, 2 μL + 2 μL drops, etc.) in both hangingdrop and sitting drop fashion. Crystals appeared at 20 °C after 7 daysand were cryoprotected (100 mM Tris, pH 7.5, 400 mM CaOAc, 10%PEG 3350, 15% ethylene glycol) and flash frozen in liquid nitrogen.X-ray diffraction data were collected at the Southeastern RegionalCollaborative Access Team (SER-CAT) ID22 beamline at theAdvanced Photon Source (Chicago, IL). The X-ray diffraction datawere indexed, integrated, and scaled in XDS to 2.3 Å.30 The structure

Figure 1. Structure of BpeGlobin. (A) Schematic of GCS domainarchitecture. (B) Orientation of the two different dimeric forms. Eachmonomer forms associations via salt bridges along helices A, B, and Ein the asymmetric unit (solid pink, green, and blue). The ferrous-oxydimer is shown in green, and the mixed dimer is shown in pink andblue. (C) Close-ups for both Fe(II)−O2 (green insert, left) andFe(III)−H2O (pink insert, right) show heme pockets withoutsignificant differences (RMSD = 0.57 Å).

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14387

was solved by molecular replacement using an ensemble of four globinstructures (globins from two unique proteins; HemAT-Bs andEcDosC; PDB IDs 1OR4, 1OR6, 4ZVA, and 4ZVB) based onpredicted structure from the sequence as a search model in PHENIXAutoMR.31 The model was built in Coot32 for residues 13−168followed by refinement of xyz coordinates, occupancies, and B-factorsin PHENIX to a final Rwork/Rfree of 20.9/25.3% (Table 1). All figures

were made with PyMOL33 and Chimera.34 The structure wasdeposited with PDB ID 6M9A and can be accessed at https://deposit.wwpdb.org/deposition/ with deposition ID: D_1000236367.FTIR Characterization. Purified proteins were reduced using

sodium hydrosulfite under an anaerobic atmosphere (Coy Labo-ratories). Reduced proteins were then concentrated to >500 μMbefore being placed into 1 mL sealed reaction vials, removed from theanaerobic chamber and the headspace purged with CO gas. UV−visspectra of all proteins were taken before FTIR scans to ensureproteins were fully ligated.26 Protein and reference buffer were thenplaced in liquid FTIR cells with CaF2 windows (International CrystalLaboratory) and run on either a Varian 660 FTIR (Thermo Fisher) oron Nicolet iS10 (Thermo Fisher).The Nicolet iS10 was run with no additional modifications to the

instrument. Sample and reference spectra were collected as an average

of 2048 scans run with a 4 cm−1 resolution. The Varian 660 FTIRspectra were collected on a home-built, dry air purged external boxconnected to the FTIR spectrometer and equipped with a liquidnitrogen cooled MCT detector.35,36 Sample and reference spectrawere collected as an average of 2048 scans run with a 2 cm−1

resolution. For all spectra, the absorbance spectra (−log(Iprotein/Ibuffer)) were baseline corrected with a spline function.

D2O and H218O Buffer Exchange. A 1 mL sample of protein was

brought into the anaerobic chamber and split into 2 equal aliquots; 50μL of 500 mM sodium hydrosulfite was added to both aliquots. 500μL of 50 mM sodium phosphate, 50 mM NaCl, and 1 mM DTT ofeither D2O (pD = 7.0) or H2

18O (pH 7.0) were added to 1 aliquotand allowed to buffer exchange overnight on a 4 °C block in theanaerobic chamber. The isotopically labeled samples were thendesalted and concentrated to ∼200 μL using a 5 kDa spinconcentrator (Vivaspin), and additional isotopically labeled bufferwas then added to yield a 500 μL aliquot. The aliquot was incubatedfor an additional 1−2 h before repeating the concentration dilutionone more time. After the second dilution, the protein was incubatedfor 30 min to 1 h and then concentrated to ∼50−100 μL. The controlsample (protein in H2O) was treated in the same manner except thatthe buffer contained 50 mM Tris, 50 mM NaCl, and 1 mM DTT, pH= 7.0 in H2O. Both samples were placed in 1 mL sealed ReactiVials(Pierce), removed from the anaerobic chamber, and exposed to COgas by purging the head space of the vials. UV−vis spectra were takenbefore and after FTIR scans to ensure the protein samples were stillbound with Fe(II)−CO. FTIR spectra were obtained as describedabove.

■ RESULTS AND DISCUSSION

Structure of Ferrous-Oxy and Mixed BpeGlobinDimers. The globin domain of BpeGReg crystallized in thespace group C121 and contained three monomeric units ofBpeGlobin in the asymmetric unit (Figure 1, Table 1).However, the biological unit is a dimer as assessed by Pisa37

and based on previous biochemical studies.25,26 There are twospecies of biological dimers formed within the crystal lattice.The first dimer contains two monomers with O2 bound to theheme (hereafter referred to as the ferrous-oxy dimer) while thesecond dimer consists of one O2-bound species and oneFe(III)−H2O monomer (hereafter referred to as the mixeddimer) (Figure 1B and Figure S1). BpeGlobin adopts theclassic globin fold of an eight α-helix bundle with a heme heldin place by proximal His99 located on helix F. Interestingly, anordered water is found within the heme pocket of allmonomers regardless of oxidation/ligation status. This wateris located ∼4.1 Å away from the bound O2 and is engaged in adirect hydrogen bond with heme distal pocket residues,suggesting a potential role in ligand binding and stabilization(Table 2 and Table S2).This is the first reported crystal structure of a bacterial

sensor globin domain in the Fe(II)−O2 ligation state, offeringa novel opportunity to examine a signaling domain in its nativeactive form (Figure 1). Furthermore, both the ferrous-oxy andthe mixed dimer are present, allowing for the comparison ofthe native active heme pocket structure with a low-activityform. Previous work has shown that Fe(II)−O2 BpeGReg isstable for at least 18 h with no significant oxidation,aggregation, or oligomerization changes.4 The exceptionallylow autoxidation and O2 dissociation rates of BpeGReg likelyenabled the extended stability of the Fe(II)−O2 proteinnecessary for its structural determination.4

Comparison to Other Crystallized GCS Proteins. Toinvestigate the conservation of potentially relevant signalingresidues, the structure of BpeGlobin was compared to

Table 1. X-ray Structure Statistics

BpeGlobin

wavelength 1.000resolution range 35.97−2.3 (2.38−2.30)space group C121unit cell 145.16, 49.64, 68.14, 90, 110.59, 90total reflns 169919 (18748)unique reflns 18911 (2040)multiplicity 9.0 (9.2)completeness (%) 92.14 (99.66)mean I/σ(I) 26.08 (19.21)Wilson B-factor 38.06R-merge 0.07781 (0.1159)R-meas 0.08293 (0.1227)R-pim 0.02807 (0.03971)CC1/2 0.993 (0.994)CCa 0.998 (0.999)relns used in refinement 18882 (2033)reflns used for R-free 1890 (205)R-work (%) 16.14 (19.75)R-free (%) 25.27 (29.43)CC (work) 0.969 (0.917)CC (free) 0.953 (0.817)no. of non-hydrogen atoms 3986macromolecules 3664ligands 136solvent 186protein residues 469RMS (bonds) 0.008RMS (angles) 1.12Ramachandran favored (%) 98.27Ramachandran allowed (%) 1.73Ramachandran outliers (%) 0.0rotamer outliers (%) 1.32clashscore 5.08av B-factor 46.12macromolecules 45.94ligands 46.52solvent 49.25

aStatistics for the highest-resolution shell are shown in parentheses.

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14388

previously published structures of the globin domains fromAfGcHK,27 HemAT-Bs,20,38 and EcDosC24 (Figure S2).Alignments of the globin domains of HemAT-Bs, AfGcHK,and EcDosC with BpeGlobin show a similar helical arrange-ment as well as a conserved hydrogen-bonding Tyr residuewithin the distal pocket. The crystal structure of BpeGlobinreveals shorter distances between the distal hydrogen-bondingresidues (Tyr43 and Ser68) than EcDosC (Tyr43), HemAT-Bs(Tyr70 and Thr95), and AfGcHK (Tyr45 and Thr71; Table 2and Table S2). Additionally, the distances between BpeGlobinTyr43 and the ordered water, as well as Ser68 and the orderedwater, are within range for hydrogen bonding, implying a rolefor the water in ligand binding and signal transduction (Table2). The crystal structures of both HemAT-Bs and AfGcHKalso contain water molecules, which are ∼4.6 and ∼5 Å fromthe heme irons, respectively (Table S2).20,27,38 In the case ofHemAT-Bs, this ordered water has been shown to interact withbound O2 through resonance Raman studies.39 The watermolecule for AfGcHK has not been further examined as of yet.However, no water was found within the distal pocket ofEcDosC due to the presence of Leu92, which acts as ahydrophobic gate preventing the water from entering the hemepocket.29 Furthermore, EcDosC does not contain a seconddistal pocket hydrogen-bond donor, suggesting that the watermay contribute to the different catalytic, ligand binding, andoligomerization behaviors of the various GCS proteins.24

Heme Distortion in BpeGlobin Influenced by O2Binding. Heme deformation is a critical component in thesignaling mechanism of several heme-based sensor proteinfamilies.40−42 To determine if heme distortion could beinvolved in signal transduction within GCS proteins, the hemecofactors of different monomers of the BpeGlobin dimers werecompared to each other as well as to other GCS hemes (Figure2 and Tables S3 and S4). The heme group can traditionallyadopt different deformation forms including saddling, ruffling,and propellering, all of which can influence the surroundingheme pocket upon ligand binding to transduce the bindingsignal.40−42 All crystallized heme groups of BpeGlobin havesimilar saddling (B2u) and propellering (A1u) deformations(Figure 2A,B and Table 3). However, the direction andmagnitude of the doming (A2u) of the Fe(II)−O2 hemes aremore similar to each other than to the Fe(III)−H2O heme.Additionally, the change in ruffling value (B1u) of the O2-bound hemes from the ferrous-oxy dimer and the mixed dimersuggests that heme deformation for BpeGReg is ligand-dependent and that heme distortion modes may play a rolein GCS signaling. Since DGC activity is dimerization-dependent and the ligation state of the heme within GCSproteins affects oligomeric state, it is hypothesized that theferrous-oxy dimer represents the active, O2-bound state, whilethe mixed dimer represents a transition between active andinactive states (Fe(III)−H2O is a low-activity state of theenzyme). Specifically, the difference in heme conformationbetween Fe(II)−O2 and Fe(III)−H2O hemes suggests thatheme deformation propagates the ligand binding signalthrough interactions with heme pocket residues into theglobin fold, thereby influencing the dimer interface, oligomericstate, and activity of the output domain.Heme distortion within BpeGlobin is likely due to

interactions with heme pocket residues; in the proximalpocket, most residues located within 5 Å of the heme arehydrophobic, with the majority being isoleucines and valines(Figure 3). These residues are well-positioned on the F helix toinfluence the globin dimer interface, changes which mayrepresent a potential signaling mechanism.9,25,27 Such amechanism is consistent with the findings of Stranava et al.,who observed ligation state-dependent structural differences inthe heme proximal residues and the dimerization interface ofAfGcHK via HDX-MS.27 In addition to these hydrophobicproximal residues, Trp72 of BpeGlobin sits at the heme edgeon the E helix within an appropriate distance for van der Waalscontact with the heme. This Trp is conserved in EcDosC andAfGcHK, while HemAT-Bs contains a histidine residue in thesame position (Figure S2).The location of Trp72 on the E helix of BpeGReg suggests

another potential route for the changes in heme distortion tobe propagated through the protein by steric interaction. AsSer68 and Lys64 also are located on the E helix (Figure 3),steric repulsion between the heme cofactor and Trp 72 couldresult in a shift in conformation of the E helix, which couldthen lead to global protein rearrangements. Ser68 is involved instabilizing O2 binding

26 while Lys64 forms a hydrogen bondwith heme propionate 6 in only the Fe(II)−O2 monomers,suggesting a currently unexplored role for these residues inDGC activation. Recent cross-linking data also has demon-strated that the globin E helix of PccGCS makes a directinteraction with the DGC output domain near the active site(Walker et al., under review.) Given the relatively high degreeof identity between PccGCS and BpeGReg (36%), it is inferred

Table 2. Inter-Residue Distances in Å, Including NonligatedWater Molecules

residuechain AFe(II)

chain BFe(II)

chain CFe(III) chain D

BpeGReg (PDB: 6M9A)Fe−His99 2.1 ± 0.4 1.9 ± 0.5 2.0 ± 0.5Fe−Tyr43 4.5 ± 0.4 4.9 ± 0.5 4.7 ± 0.5Fe−Ser68 6.3 ± 0.4 6.2 ± 0.5 6.0 ± 0.5Fe−Trp72 8.1 ± 0.4 8.0 ± 0.5 8.3 ± 0.4Tyr43−Ser68 4.5 ± 0.4 4.1 ± 0.5 4.3 ± 0.5O2−Fe(II) 1.8 ± 0.4 1.9 ± 0.5O2−Tyr43 2.3 ± 0.4 2.4 ± 0.5O2−Ser68 5.1 ± 0.4 5.2 ± 0.5H2O−Tyr43 2.9 ± 0. 4 2.5 ± 0.4 2.5 ± 0.4H2O−Ser68 2.2 ± 0.4 2.1 ± 0.4 2.3 ± 0.4H2O−Fe 4.7 ± 0. 4 3.5 ± 0.4 4.2 ± 0.4

HemAT-Bs20,37 Fe(III)−CN (PDB: 1OR4)Fe(III)−His180 2.0 2.0Fe(III)−Tyr70 5.7 5.6Fe(III)−Thr95 7.1 6.9Tyr70−Thr95 5.7 5.7H2O−Tyr70 4.0 3.6H2O−Thr95 2.7 2.8H2O−Fe(III) 4.8 4.5

HemAT-Bs20,37 Fe(III) (PDB: 1OR6)Fe(III)−His180 2.8 2.1Fe(III)−Tyr70 5.8 6.3Fe(III)−Thr95 6.8 7.6Tyr70−Thr95 5.3 5.9

EcDosC23 Fe(II) (PDB: 4ZVB)Fe(II)−His99 2.2 2.2 2.2 2.2Fe(II)−Tyr43 7.0 7.2 7.0 7.0

EcDosC23 Fe(III) (PDB: 4ZVA)Fe(III)−His99 2.2 2.2Fe(III)−Tyr43 6.6 6.7

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14389

that the E helix of BpeGReg likely also interacts with the DGCoutput domain. Taken together, these findings imply thatligand binding and the heme distortions that follow result inmovements of Lys64, Ser68, and Trp72 that can then bepropagated through the E helix to regulate the output domain.Heme Distortion Comparison to Other GCS Proteins.

It has been noted previously that similar trends in hemedistortions may be observed for GCS proteins with the sameclass of output domain.40,41,43 Accordingly, the distortionsmeasured in BpeGlobin and EcDosC are more similar to eachother than to HemAT-Bs (Table S3). Specifically, the Fe(III)−H2O heme of BpeGlobin has highly similar Doop values to allunits of EcDosC Fe(III)−H2O (∼0.87 and ∼0.96, respec-tively). Additionally, heme groups from BpeGlobin andEcDosC both show high Doop and B1u (ruffling) values(∼0.90/0.85 and ∼0.92/0.70, respectively), whereas the valuesfor HemAT-Bs are lower (∼0.58 and 0.46 for Doop and B1u,respectively) indicating less overall distortion and ruffling. Itshould be noted that the Fe(III)−H2O form of BpeGlobin wasused in this comparison with other sensor globins in the

Fe(III)−H2O to control for the effects of different ligands andoxidation states on heme distortion. This analysis furthersupports previous findings that heme distortions are mostsimilar between proteins that have the same function40,44 andthat conserved heme pocket residues assist in signal trans-mission within DGC-containing GCS proteins,1,26 potentiallyby modulating heme ruffling (B1u) to transmit the ligationstatus of the heme-Fe into the globin fold. Indeed, acomparison of the pockets of PccGCS, EcDosC, andHemAT-Bs indicates that HemAT-Bs has the lowest sequenceidentity for these residues (12.5%, as compared to 75% forPccGCS and EcDosC) (Figure S2). In contrast, AfGcHKexhibits 16.5% sequence identity to BpeGReg and has a distinctoutput domain (histidine kinase). While some of the AfGcHKchains exhibit similar distortion trends to BpeGlobin and otherFe(III) globins, a universal trend within the AfGcHK chains isnot apparent. Taken together, these data imply a general modelfor transduction of the ligand binding signal from the globin toDGC domain in GCS proteins via ligand-dependent distortionof the heme cofactors.

Figure 2. Heme deformations in BpeGlobin. (A) O2-bound heme groups from both dimers exhibit a significant degree of ruffling, while theFe(III)−H2O heme exhibits doming in the opposite direction of the O2-bound hemes. The combination of all six types of deformation yields theoverall heme distortion shown. From left to right: Fe(II)−O2 heme from the ferrous-oxy dimer, Fe(II)−O2 heme from the mixed dimer, Fe(III)−H2O heme from the mixed dimer. The differences in heme deformation lead to changes in the orientations of propionate 6, which forms ahydrogen bond with Lys74 in only the O2-bound units. The 2Fo − Fc electron density is contoured at 1.5I/σ. (B) Heme cofactors can take on anumber of deformation modes, including saddling (B2u), ruffling (B1u), propellering (A1u), doming (A2u), or waving (Eg(x), Eg(y)) that may influenceligand-dependent signaling (inspired by work from Shelnutt et al.40).

Table 3. Heme Distortion Modes in BpeGlobin in Åa

protein Doop (over all) B2u (saddle) B1u (ruffle) A2u (dome) Eg(x) (wave) Eg(y) (wave) A1u (propeller)

ferrous-oxy dimer Fe(II)−O2 1.21 −0.22 1.13 −0.27 −0.24 −0.05 0.01mixed dimer Fe(II)−O2 1.11 −0.13 −1.03 −0.32 0.09 −0.20 <0.01mixed dimer Fe(III)−H2O 0.90 −0.18 0.85 0.15 −0.08 0.14 <0.01

aData shown is relative to a planar heme where all distortions are 0. Positive (+) numbers indicate distortion above plane; negative (−) indicatedistortion below plane. Numerical values represent distortion magnitude.

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14390

While previous studies on GCS proteins have not examinedthe role of heme distortion in signaling, work has been done toaddress the effect of high-spin/low-spin on activity of GCSsand other heme-based sensors.15,45−47 It has been noted thatlow-spin states of the heme are often correlated with the high-activity states of the enzymes; these findings agree with thedata on BpeGReg and PccGCS, which have identified theFe(II)−O2 state as having the highest enzyme activity.EcDosP, a PAS domain heme sensor, also exhibits highactivity in the Fe(II)−O2, Fe(II)−CO, and Fe(II)−NO states,but low activity in the high-spin Fe(II) unligated state.45,46

Previous studies on AfGcHK also support this trend andsuggest that it may be due to the observation that high-spinFe(II) heme in heme sensor proteins often has a shift of Feupward out of the heme plane by 0.4−0.5 Å, while ligationpushes the Fe back down toward the plane.47−49 However, asEcDosC, the GCS from E. coli, was found to exhibit highdiguanylate cyclase activity in the high-spin Fe(II) and low-spin Fe(II)−O2 and Fe(II)−CO states, the link between spinstate and activity is not universal.15 Taken together, theprevious studies and our results suggest that heme distortion,which is affected by ligand binding, plays an important role inGCS signaling and activity.Interactions between Distal Residues and Bound

Ligand. In order to verify putative interactions between boundligands and protein residues identified from the BpeGlobincrystal structure, FTIR was performed on CO-bound wild-typeand BpeGReg/BpeGlobin protein variants, as well as the closelyrelated GCS from P. carotovorum, PccGCS/PccGlobin (Figure4).4 PccGCS is often examined in tandem with BpeGReg dueto the relatively high degree of sequence identity between the

two proteins compared to other GCSs and the informationavailable on enzymatic activity and ligand binding.4,25,26

Furthermore, as PccGCS has been demonstrated to controlO2-dependent virulence of a plant pathogen, additional insightinto its mode of signal transduction is important fordevelopment of inhibitors.9 Although the sensor globins fromHemAT-Bs and EcDosC have previously been crystallized, theycontain lower sequence identity to BpeGReg as compared toPccGCS (14%, 34%, and 36%, respectively) (Figure S2).PccGCS exhibits the FTIR spectrum expected for a globin,with peaks at 1922 and 1961 cm−1 corresponding to COstretches with a hydrogen-bonding interaction and apolarpocket, respectively (Figure 4B and Table S5).50,51 Surpris-ingly, BpeGReg exhibits three CO stretching frequencies, 1925,1964, and 1972 cm−1 (Figure 4A and Table S5), which likelycorrespond to a hydrogen bond, apolar interaction, andelectrostatic interaction with a negative charge, respectively.51

FTIR scanning with isotopically labeled 13CO resulted in adownfield shift of all three peaks in the spectrum of BpeGReg(1925, 1964, and 1972 cm−1) and two peaks in the spectrum ofPccGCS (1922 and 1961 cm−1) (Figure 4A,B and Table S5),verifying that these stretching frequencies correspond to theCO stretch. In addition, FTIR spectra of full-length BpeGRegand PccGCS were consistent with the corresponding globinspectra (Figure 4C,D and Figure S3), demonstrating that thedeletion of the middle and DGC domains does notsubstantially affect interactions between bound CO and theproteins.The observation of a 1972/1973 cm−1 stretching frequency

for BpeGReg and BpeGlobin, respectively, was intriguing giventhe lack of charged amino acids within the distal pocket(Figure 1 and Figure S2). Furthermore, the observation of the1972/1973 cm−1 stretch only in BpeGReg/Globin suggestssubtle differences in domain structure and/or conformation, ascompared to PccGCS, supporting previously identified differ-ences in ligand binding kinetics.4 To probe interactions ofheme distal pocket residues with bound CO, protein variantswere generated with mutations to the distal hydrogen-bondingtyrosine and serine residues (Tyr43Phe and Ser68Ala forBpeGReg/BpeGlobin; Tyr57Phe for PccGCS; Tyr57Phe andSer82Ala for PccGlobin) (Table S6). For both proteins, onlymutation of the distal Tyr produced an appreciable change inthe FTIR spectrum; spectra of both BpeGReg/Globin(Tyr43Phe) and PccGlobin (Tyr57Phe) lack the stretches at1925 and 1922/1921 cm−1, respectively, suggesting that thedistal tyrosine provides a hydrogen bond to bound CO (Figure4D, Figure S3, and Table S7). Furthermore, the apolar stretchnear 1960 cm−1 shifted for both BpeGReg/Globin (Tyr43Phe)and PccGlobin (Tyr57Phe), indicating a change in the hemepocket environment upon mutation of Tyr43 (Figure 4D andFigure S3B). This shift may be due to subtle rearrangementsthat allow the partial positive charge in the center of the phenylπ-ring to interact with the C−O triple bond, increasing thepolarity and subtly changing the stretching frequency. Takentogether, these data demonstrate that the distal tyrosineresidue in these proteins makes a direct interaction with thebound CO ligand, while the distal serine does not. However,this work, along with previous studies,26 suggests that the distalserine plays a role in O2 binding, likely through a hydrogen-bond network within the heme pocket.

Examination of Structural Water. The observation of aBpeGReg/Globin CO stretching frequency at 1972/1973 cm−1

suggested that the bound water may be involved in modulating

Figure 3. Heme pocket residues in BpeGlobin. (A) The distalhydrogen-bonding residues Ser68 and Tyr43 are shown, along withthe proximal His99 and the heme pocket residues Lys64 and Trp72.(B) Hydrophobic residues on the heme proximal side are shown.These residues are thought to propagate ligand-mediated hemedistortions throughout the protein.

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14391

the distal pocket environment. While the crystal structureidentified a hydrogen bond between O2 and the distal pocketwater molecule, it was not clear which distal pocket residuesthe water interacts with to stabilize positioning within thepocket. To explore the role of the water observed in the hemepocket of the BpeGlobin crystal structure, FTIR was performedon wild-type BpeGlobin at pH values ranging from 5.0 to 8.0(Figure 5). As the buffer pH was decreased, the peak at 1973cm−1 decreased in intensity, suggesting that the distal pocket

ordered water exists in a partially deprotonated state,potentially stabilized through a hydrogen bond to the tyrosine.To further support the distal water as the moiety responsible

for the stretch at 1972 cm−1, FTIR spectra were taken in D2Oand H2

18O labeled water (Figure 6 and Figure S4A). Withinthe BpeGlobin FTIR spectrum, the 1973 cm−1 peak shifts andloses intensity when it is in the H2

18O buffer (Figure 6A). Asthis peak is tyrosine-dependent and is the only peaksignificantly affected by the change in pH, it is likely the

Figure 4. CO stretches for the GCS proteins. (A) BpeGReg Fe(II)−12CO (solid red) vs Fe(II)−13CO (dashed red). (B) PccGCS Fe(II)−12CO(solid black) vs Fe(II)−13CO (dashed black). (C) Fe(II)−12CO peaks of BpeGlobin, BpeGReg, BpeGReg (Y43F), and BpeGReg (S68A). (D)Fe(II)−12CO peaks of PccGlobin, PccGlobin (Y57F), and PccGlobin (S82A).

Figure 5. Effect of pH on CO stretching frequencies. (A) BpeGlobin Fe(II)−CO ranging from pH 5.0−8.0. (B) PccGlobin Fe(II)−CO rangingfrom pH 7.0 to 9.0.

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14392

result of the CO interaction with a partially deprotonated distalH2O, spatially oriented through interactions with the distalTyr43.While PccGCS/PccGlobin does not exhibit a peak in the

1970 cm−1 region like BpeGReg/BpeGlobin, the two proteinsshare high sequence identity (36% overall; 31% for the globindomain), suggesting they adopt a similar tertiary fold. Further,the high sequence identity between BpeGReg and PccGCSsuggests that PccGCS/PccGlobin may also be regulated in asimilar manner and thus likely also contains a distal pocketwater. To interrogate the presence of a structured water inPccGlobin, and thus PccGCS, FTIR was performed on wild-type PccGlobin over a pH range of 7.0−9.0 (Figure 5B). LikeBpeGlobin, PccGlobin remained stable at all tested pH values,although it was noted that the stability significantly decreasedabove pH 9. The spectrum of PccGlobin developed a peak at1975 cm−1 as pH increased, suggesting the deprotonation of ahydroxyl group in the heme pocket, potentially on either thedistal tyrosine or the putative ordered water. FTIR spectra ofPccGlobin were also taken in D2O and H2

18O labeled water tosupport the distal water moiety (Figures 5B and 6B, and FigureS4B). Unlike in the H2

18O spectrum of BpeGlobin, that ofPccGlobin does not demonstrate a significant shift in any of itspeaks due to the lack of structural water-dependent stretches inthe PccGlobin Fe(II)−CO spectrum (Figure 6). This differ-ence in the protonation state of the ordered water betweenBpeGReg and PccGCS, which must be due to differences inheme pocket environments, likely contributes to the observeddifferences in O2 binding kinetics (mono/biexponentialdissociation rates) and O2-dependent cyclase activity of theseGCS proteins.4

■ CONCLUSION

In conclusion, BpeGlobin is the first isolated GCS sensorglobin domain to be crystallized in the native active Fe(II)−O2ligation state. Differences in ligation state correlate with hemedistortion, suggesting a role in transmission of the O2 bindingsignal. The differences in ligation state and heme distortion arepotentially propagated through interactions with residues onthe E helix, allowing for signaling to the DGC output domain.For BpeGReg, a 1973 cm−1 peak is observed by FTIR anddemonstrated to be due to the presence of an ordered waterthat is at least partially deprotonated in the heme pocket; a

corresponding water in the PccGCS distal pocket remainsprotonated. In both proteins, the distal pocket Tyr is requiredfor the stretching frequencies in both the 1970 and 1920 cm−1

regions, indicating that the bound water is stabilized by the Tyrresidue and interacts with the bound CO ligand. Takentogether, these studies suggest roles for the distal water andheme distortion in transmission of the O2 binding signal inGCS proteins (Figure 7).

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.inorg-chem.8b02584.

Primers list, full table of heme distortion modes and ofdistance for all known crystallized globins, RMSD valuesfor alignment of the three BpeGlobin chain, along with asequence alignment figure to highlight key amino acidsfrom all the compared globin domains, and additionalFTIR spectra and all peaks (PDF)

Figure 6. FTIR examination of the structured water with difference spectra shown below. (A) FTIR spectra of BpeGlobin Fe(II)−CO in H2O(solid red) and H2

18O (dashed red) with the difference spectra shown below. (B) FTIR spectra of PccGlobin Fe(II)−CO in H2O (solid black) andH2

18O (dashed black) with the difference spectra shown below.

Figure 7. Overview of proposed heme pocket interactions involved inO2-dependent signal transduction. The tetramer, which has highdiguanylate cyclase activity (c-di-GMP depicted in green), exists inthe Fe(II)−O2 ligation state (right) with a deformed heme and aproposed hydrogen-bond network based on current and previouswork.4,15,24−26,28 Dissociation of O2 and/or heme autoxidation resultsin a less distorted heme and formation of a dimer with lowdiguanylate cyclase activity (left).

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14393

■ AUTHOR INFORMATION

Corresponding Author*E-mail: emily.weinert@emory.edu.

ORCIDEmily E. Weinert: 0000-0002-4986-8682Author Contributions§S.R. and P.G.Y. contributed equally to this work.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This work was supported by NSF CHE1352040 (E.E.W.) andFrasch Foundation Grant 824-H17 (E.E.W.). S.R. was partiallysupported by funds from the Emory University LaneyGraduate School. We thank members of the Weinert andDunham laboratories for assistance and helpful suggestions.Additionally, this research used resources of the AdvancedPhoton Source, a U.S. Department of Energy (DOE) Office ofScience User Facility operated for the DOE Office of Scienceby Argonne National Laboratory, under Contract DE-AC02-06CH11357.

■ REFERENCES(1) Walker, J. A.; Rivera, S.; Weinert, E. E. Mechanism and Role ofGlobin Coupled Sensor Signaling. Adv. Microb. Physiol. 2017, 71,133−169.(2) Shimizu, T.; Huang, D.; Yan, F.; Stranava, M.; Bartosova, M.;Fojtikova, V.; Martinkova, M. Gaseous O2, NO, and CO in signaltransduction: structure and function relationships of heme-based gassensors and heme-redox sensors. Chem. Rev. 2015, 115 (13), 6491−533.(3) Green, J.; Rolfe, M. D.; Smith, L. J. Transcriptional regulation ofbacterial virulence gene expression by molecular oxygen and nitricoxide. Virulence 2014, 5 (8), 794−809.(4) Burns, J. L.; Deer, D. D.; Weinert, E. E. Oligomeric state affectsoxygen dissociation and diguanylate cyclase activity of globin coupledsensors. Mol. BioSyst. 2014, 10 (11), 2823−2826.(5) Wu, C.; Cheng, Y. Y.; Yin, H.; Song, X. N.; Li, W. W.; Zhou, X.X.; Zhao, L. P.; Tian, L. J.; Han, J. C.; Yu, H. Q. Oxygen promotesbiofilm formation of Shewanella putrefaciens CN32 through adiguanylate cyclase and an adhesin. Sci. Rep. 2013, 3, 1945−1951.(6) Sawai, H.; Yoshioka, S.; Uchida, T.; Hyodo, M.; Hayakawa, Y.;Ishimori, K.; Aono, S. Molecular oxygen regulates the enzymaticactivity of a heme-containing diguanylate cyclase (HemDGC) for thesynthesis of cyclic di-GMP. Biochim. Biophys. Acta, Proteins Proteomics2010, 1804 (1), 166−72.(7) Romling, U.; Galperin, M. Y.; Gomelsky, M. Cyclic di-GMP: thefirst 25 years of a universal bacterial second messenger.Microbiol. Mol.Biol. Rev. 2013, 77 (1), 1−52.(8) Chan, C.; Paul, R.; Samoray, D.; Amiot, N. C.; Giese, B.; Jenal,U.; Schirmer, T. Structural basis of activity and allosteric control ofdiguanylate cyclase. Proc. Natl. Acad. Sci. U. S. A. 2004, 101 (49),17084−9.(9) Burns, J. L.; Jariwala, P. B.; Briggs, L.; Fontaine, B. M.; Weinert,E. E.; Rivera, S. Oxygen-Dependent Globin Coupled Sensor SignalingModulates Motility and Virulence of the Plant Pathogen Pectobacte-rium carotovorum. ACS Chem. Biol. 2017, 12, 2070−2077.(10) Hengge, R. Principles of c-di-GMP signalling in bacteria. Nat.Rev. Microbiol. 2009, 7 (4), 263−73.(11) Wan, X.; Tuckerman, J. R.; Saito, J. A.; Freitas, T. A.;Newhouse, J. S.; Denery, J. R.; Galperin, M. Y.; Gonzalez, G.; Gilles-Gonzalez, M. A.; Alam, M. Globins synthesize the second messengerbis-(3′-5′)-cyclic diguanosine monophosphate in bacteria. J. Mol. Biol.2009, 388 (2), 262−270.

(12) Kalia, D.; Merey, G.; Nakayama, S.; Zheng, Y.; Zhou, J.; Luo,Y.; Guo, M.; Roembke, B. T.; Sintim, H. O. Nucleotide, c-di-GMP, c-di-AMP, cGMP, cAMP, (p)ppGpp signaling in bacteria andimplications in pathogenesis. Chem. Soc. Rev. 2013, 42 (1), 305−341.(13) Vinogradov, S. N.; Tinajero-Trejo, M.; Poole, R. K.; Hoogewijs,D. Bacterial and archaeal globins - A revised perspective. Biochim.Biophys. Acta, Proteins Proteomics 2013, 1834, 1789−1800.(14) Borriello, G.; Werner, E.; Roe, F.; Kim, A. M.; Ehrlich, G. D.;Stewart, P. S. Oxygen limitation contributes to antibiotic tolerance ofPseudomonas aeruginosa in biofilms. Antimicrob. Agents Chemother.2004, 48 (7), 2659−64.(15) Kitanishi, K.; Kobayashi, K.; Kawamura, Y.; Ishigami, I.; Ogura,T.; Nakajima, K.; Igarashi, J.; Tanaka, A.; Shimizu, T. Important rolesof Tyr43 at the putative heme distal side in the oxygen recognitionand stability of the Fe(II)-O2 complex of YddV, a globin-coupledheme-based oxygen sensor diguanylate cyclase. Biochemistry 2010, 49(49), 10381−93.(16) Phippen, B. L.; Oliver, J. D. Role of anaerobiosis in capsuleproduction and biofilm formation in Vibrio vulnificus. Infect. Immun.2015, 83 (2), 551−9.(17) Roberts, R. R.; Hota, B.; Ahmad, I.; Scott, R. D., II; Foster, S.D.; Abbasi, F.; Schabowski, S.; Kampe, L. M.; Ciavarella, G. G.;Supino, M.; Naples, J.; Cordell, R.; Levy, S. B.; Weinstein, R. A.Hospital and societal costs of antimicrobial-resistant infections in aChicago teaching hospital: implications for antibiotic stewardship.Clin. Infect. Dis. 2009, 49 (8), 1175−1184.(18) Sommer, F.; Backhed, F. The gut microbiota-masters of hostdevelopment and physiology. Nat. Rev. Microbiol. 2013, 11 (4), 227−38.(19) Zhang, W.; Olson, J. S.; Phillips, G. N., Jr. Biophysical andkinetic characterization of HemAT, an aerotaxis receptor from Bacillussubtilis. Biophys. J. 2005, 88 (4), 2801−2814.(20) Zhang, W.; Phillips, G. N., Jr. Structure of the oxygen sensor inBacillus subtilis: signal transduction of chemotaxis by control ofsymmetry. Structure 2003, 11 (9), 1097−1110.(21) Fojtikova, V.; Stranava, M.; Vos, M. H.; Liebl, U.; Hranicek, J.;Kitanishi, K.; Shimizu, T.; Martinkova, M. Kinetic Analysis of aGlobin-Coupled Histidine Kinase, Af GcHK: Effects of the HemeIron Complex, Response Regulator, and Metal Cations onAutophosphorylation Activity. Biochemistry 2015, 54 (32), 5017−5029.(22) Tamayo, R.; Pratt, J. T.; Camilli, A. Roles of Cyclic Diguanylatein the Regulation of Bacterial Pathogenesis. Annu. Rev. Microbiol.2007, 61, 131−148.(23) Paul, R.; Abel, S.; Wassmann, P.; Beck, A.; Heerklotz, H.; Jenal,U. Activation of the diguanylate cyclase PleD by phosphorylation-mediated dimerization. J. Biol. Chem. 2007, 282 (40), 29170−7.(24) Tarnawski, M.; Barends, T. R.; Schlichting, I. Structural analysisof an oxygen-regulated diguanylate cyclase. Acta Crystallogr., Sect. D:Biol. Crystallogr. 2015, 71 (11), 2158−2177.(25) Burns, J. L.; Rivera, S.; Deer, D. D.; Joynt, S. C.; Dvorak, D.;Weinert, E. E. Oxygen and c-di-GMP Binding Control Oligomeriza-tion State Equilibria of Diguanylate Cyclase-Containing GlobinCoupled Sensors. Biochemistry 2016, 55, 6642−6651.(26) Rivera, S.; Burns, J. L.; Vansuch, G. E.; Chica, B.; Weinert, E. E.Globin domain interactions control heme pocket conformation andoligomerization of globin coupled sensors. J. Inorg. Biochem. 2016,164, 70−76.(27) Stranava, M.; Man, P.; Skalova, T.; Kolenko, P.; Blaha, J.;Fojtikova, V.; Martinek, V.; Dohnalek, J.; Lengalova, A.; Rosulek, M.;et al. Coordination and redox state-dependent structural changes ofthe heme-based oxygen sensor AfGcHK associated with intraproteinsignal transduction. J. Biol. Chem. 2017, 292, 20921−20935.(28) Aono, S.; Kato, T.; Matsuki, M.; Nakajima, H.; Ohta, T.;Uchida, T.; Kitagawa, T. Resonance Raman and ligand bindingstudies of the oxygen-sensing signal transducer protein HemAT fromBacillus subtilis. J. Biol. Chem. 2002, 277 (16), 13528−38.(29) Nakajima, K.; Kitanishi, K.; Kobayashi, K.; Kobayashi, N.;Igarashi, J.; Shimizu, T. Leu65 in the heme distal side is critical for the

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14394

stability of the Fe(II)-O2 complex of YddV, a globin-coupled oxygensensor diguanylate cyclase. J. Inorg. Biochem. 2012, 108, 163−70.(30) Kabsch, W. Integration, scaling, space-group assignment andpost-refinement. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2010, 66(2), 133−144.(31) Adams, P. D.; Afonine, P. V.; Bunkoczi, G.; Chen, V. B.; Davis,I. W.; Echols, N.; Headd, J. J.; Hung, L.-W.; Kapral, G. J.; Grosse-Kunstleve, R. W.; et al. PHENIX: a comprehensive Python-basedsystem for macromolecular structure solution. Acta Crystallogr., Sect.D: Biol. Crystallogr. 2010, 66 (2), 213−221.(32) Emsley, P.; Lohkamp, B.; Scott, W. G.; Cowtan, K. Featuresand development of Coot. Acta Crystallogr., Sect. D: Biol. Crystallogr.2010, 66 (4), 486−501.(33) Schrodinger, L. PyMOL The PyMOL Molecular Graphics System,Version: 2010.(34) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.;Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF Chimeraavisualization system for exploratory research and analysis. J. Comput.Chem. 2004, 25 (13), 1605−1612.(35) Greene, B. L.; Wu, C.-H.; Vansuch, G. E.; Adams, M. W.; Dyer,R. B. Proton inventory and dynamics in the Nia-S to Nia-C transitionof a [NiFe] hydrogenase. Biochemistry 2016, 55 (12), 1813−1825.(36) Greene, B. L.; Vansuch, G. E.; Wu, C.-H.; Adams, M. W.; Dyer,R. B. Glutamate Gated Proton-Coupled Electron Transfer Activity ofa [NiFe]-Hydrogenase. J. Am. Chem. Soc. 2016, 138 (39), 13013−13021.(37) Krissinel, E.; Henrick, K. Inference of macromolecularassemblies from crystalline state. J. Mol. Biol. 2007, 372 (3), 774−797.(38) Zhang, W.; Phillips, G. N., Jr. Crystallization and X-raydiffraction analysis of the sensor domain of the HemAT aerotacticreceptor. Acta Crystallogr., Sect. D: Biol. Crystallogr. 2003, 59 (4),749−751.(39) Ohta, T.; Yoshimura, H.; Yoshioka, S.; Aono, S.; Kitagawa, T.Oxygen-sensing mechanism of HemAT from Bacillus subtilis: aresonance Raman spectroscopic study. J. Am. Chem. Soc. 2004, 126(46), 15000−1.(40) Shelnutt, J. A.; Song, X. Z.; Ma, J. G.; Jia, S. L.; Jentzen, W.;Medforth, C. J. Nonplanar porphyrins and their significance inproteins. Chem. Soc. Rev. 1998, 27 (1), 31−41.(41) Ma, J. G.; Zhang, J.; Franco, R.; Jia, S. L.; Moura, I.; Moura, J.J.; Kroneck, P. M.; Shelnutt, J. A. The structural origin of nonplanarheme distortions in tetraheme ferricytochromes c3. Biochemistry 1998,37 (36), 12431−42.(42) Graves, A. B.; Graves, M. T.; Liptak, M. D. Measurement ofHeme Ruffling Changes in MhuD Using UV−vis Spectroscopy. J.Phys. Chem. B 2016, 120 (16), 3844−3853.(43) Ma, J. G.; Laberge, M.; Song, X. Z.; Jentzen, W.; Jia, S. L.;Zhang, J.; Vanderkooi, J. M.; Shelnutt, J. A. Protein-induced changesin nonplanarity of the porphyrin in nickel cytochrome c probed byresonance Raman spectroscopy. Biochemistry 1998, 37 (15), 5118−28.(44) Jentzen, W.; Ma, J. G.; Shelnutt, J. A. Conservation of theconformation of the porphyrin macrocycle in hemoproteins. Biophys.J. 1998, 74 (2), 753−763.(45) Tanaka, A.; Takahashi, H.; Shimizu, T. Critical role of the hemeaxial ligand, Met95, in locking catalysis of the phosphodiesterase fromEscherichia coli (Ec DOS) toward cyclic diGMP. J. Biol. Chem. 2007,282 (29), 21301−21307.(46) Tanaka, A.; Shimizu, T. Ligand binding to the Fe (III)-protoporphyrin IX complex of phosphodiesterase from Escherichiacoli (Ec DOS) markedly enhances catalysis of cyclic di-GMP: Roles ofMet95, Arg97, and Phe113 of the putative heme distal side in catalyticregulation and ligand binding. Biochemistry 2008, 47 (50), 13438−13446.(47) Kitanishi, K.; Kobayashi, K.; Uchida, T.; Ishimori, K.; Igarashi,J.; Shimizu, T. Identification and functional and spectral character-ization of a globin-coupled histidine kinase from Anaeromyxobactersp. Fw109−5. J. Biol. Chem. 2011, 286 (41), 35522−34.

(48) Miyatake, H.; Mukai, M.; Adachi, S.-i.; Nakamura, H.; Tamura,K.; Iizuka, T.; Shiro, Y.; Strange, R. W.; Hasnain, S. S. Ironcoordination structures of oxygen sensor FixL characterized by Fe K-edge extended X-ray absorption fine structure and resonance Ramanspectroscopy. J. Biol. Chem. 1999, 274 (33), 23176−23184.(49) Springer, B. A.; Sligar, S. G.; Olson, J. S.; Phillips, G. N.Mechanisms of Ligand Recognition in Myoglobin. Chem. Rev. 1994,94 (3), 699−714.(50) Pavlou, A.; Martinkova, M.; Shimizu, T.; Kitanishi, K.; Stranava,M.; Loullis, A.; Pinakoulaki, E. Probing the ligand recognition anddiscrimination environment of the globin-coupled oxygen sensorprotein YddV by FTIR and time-resolved step-scan FTIR spectros-copy. Phys. Chem. Chem. Phys. 2015, 17 (26), 17007−15.(51) Li, X.-Y.; Ibers, J. A.; Sessler, J. L.; Spiro, T. G.; Ray, G.B. Howfar can proteins bend the FeCO unit? Distal, polar and steric effects inheme proteins and models. J. Am. Chem. Soc. 1994, 116, 162−176.

Inorganic Chemistry Article

DOI: 10.1021/acs.inorgchem.8b02584Inorg. Chem. 2018, 57, 14386−14395

14395