Gas Sensing and Signaling in the PAS-Heme Domain of the Pseudomonas aeruginosa Aer2 Receptor

ABSTRACT

The Aer2 chemoreceptor from Pseudomonas aeruginosa contains a PAS sensing domain that coordinates b-type heme and signals in response to the binding of O₂, CO, or NO. PAS-heme structures suggest that Aer2 uniquely coordinates heme via a His residue on a 3₁₀ helix (H234 on Eη), stabilizes O₂ binding via a Trp residue (W283), and signals via both W283 and an adjacent Leu residue (L264). Ligand binding may displace L264 and reorient W283 for hydrogen bonding to the ligand. Here, we clarified the mechanisms by which Aer2-PAS binds heme, regulates ligand binding, and initiates conformational signaling. H234 coordinated heme, but additional hydrophobic residues in the heme cleft were also critical for stable heme binding. O₂ appeared to be the native Aer2 ligand (dissociation constant [K_d] of 16 μM). With one exception, mutants that bound O₂ could signal, whereas many mutants that bound CO could not. W283 stabilized O₂ binding but not CO binding, and it was required for signal initiation; W283 mutants that could not stabilize O₂ were rapidly oxidized to Fe(III). W283F was the only Trp mutant that bound O₂ with wild-type affinity. The size and nature of residue 264 was important for gas binding and signaling: L264W blocked O₂ binding, L264A and L264G caused O₂-mediated oxidation, and L264K formed a hexacoordinate heme. Our data suggest that when O₂ binds to Aer2, L264 moves concomitantly with W283 to initiate the conformational signal. The signal then propagates from the PAS domain to regulate the C-terminal HAMP and kinase control domains, ultimately modulating a cellular response.

IMPORTANCE Pseudomonas aeruginosa is a ubiquitous environmental bacterium and opportunistic pathogen that infects multiple body sites, including the lungs of cystic fibrosis patients. P. aeruginosa senses and responds to its environment via four chemosensory systems. Three of these systems regulate biofilm formation, twitching motility, and chemotaxis. The role of the fourth system, Che2, is unclear but has been implicated in virulence. The Che2 system contains a chemoreceptor called Aer2, which contains a PAS sensing domain that binds heme and senses oxygen. Here, we show that Aer2 uses unprecedented mechanisms to bind O₂ and initiate signaling. These studies provide both the first functional corroboration of the Aer2-PAS signaling mechanism previously proposed from structure as well as a signaling model for Aer2-PAS receptors.

INTRODUCTION

Pseudomonas aeruginosa is a common environmental bacterium and a significant cause of opportunistic human disease. It survives in complex environments with the aid of 26 chemoreceptors and four chemosensory systems that collectively sense environmental conditions and modify bacterial behavior. The roles of three of these chemosensory systems are known: one modulates type IV pili production and twitching motility (Pil-Chp system), another controls biofilm formation (Wsp system), and a third regulates flagellum-mediated chemotaxis (Che system) (1, 2). The role of the fourth chemosensory system, Che2 (PA0173-PA0179), is currently unknown. Che2 expresses a complete set of chemosensory proteins (CheY2, CheA2, CheW2, CheR2, CheD, and CheB2), including a chemoreceptor (PA0176) called Aer2 (previously called McpB). Aer2 was so named because it, along with classical Aer, was observed to mediate aerotaxis by P. aeruginosa (3). However, we and others have not observed Aer2-mediated chemotaxis or aerotaxis in P. aeruginosa (4, 5). Moreover, it is now understood that the response regulators (CheY proteins) of Che2-like systems do not bind to the bacterial flagellar motor protein, FliM, to modulate swimming behavior (6–8; K. J. Watts and E. Orillard, unpublished data). This suggests that the primary role of Che2 is something other than the control of chemotaxis or aerotaxis. Notably, a role for Che2 in virulence has been suggested (9, 10).

The Che2 chemoreceptor Aer2 has no membrane-spanning segments. However, during the early stationary phase of P. aeruginosa growth, Che2 proteins form a cluster at the cell pole that is held together solely by Aer2 (5, 10). Importantly, Che2 proteins do not colocalize with Che (chemotaxis system) proteins (5). Aer2 has an unusual architecture, with a PAS sensory domain sandwiched between three N-terminal and two C-terminal HAMP domains (Fig. 1a). These domains precede a kinase control module that is typical of methyl-accepting chemoreceptors. The kinase control module has four predicted methylation sites (QEEE) and a C-terminal pentapeptide (GWEEF) for binding the adaptation enzymes CheR2, CheB2, and CheD (Fig. 1a) (11). In P. aeruginosa, receptor deamidation/demethylation by CheB2 (and possibly CheD), as well as methylation by CheR2, is expected to fine-tune Aer2-mediated responses. The kinase control module of Aer2 shares significant sequence identity with the kinase control modules of the major Escherichia coli chemoreceptors. Thus, Aer2 is able to control the E. coli chemotaxis pathway through direct interactions with the E. coli adapter protein, CheW, and the histidine kinase, CheA (4). When Aer2 is expressed in otherwise chemoreceptor-less E. coli, it mediates repellent tumbling (signal-on) responses to O₂, CO, and NO (4). Gas-bound Aer2 causes rapid autophosphorylation of bound CheA with subsequent phosphotransfer to CheY. Phospho-CheY in turn binds to the E. coli flagellar switch protein, FliM, causing a directional change in flagellar rotation from counterclockwise to clockwise, resulting in E. coli tumbling.

FIG 1.

P. aeruginosa Aer2 and the structure of its PAS domain. (a) Model of an Aer2 dimer showing the PAS domain sandwiched between three N-terminal and two C-terminal HAMP domains. The C-terminal kinase control module has four predicted methylation sites (QEEE) and a C-terminal pentapeptide (GWEEF) for binding adaptation enzymes. (b) Crystal structure of the Aer2 PAS domain in cartoon form with heme cofactor (shown as red sticks) and bound cyanide (shown as spheres) (PDB entry 3VOL) (13). The Fe-CN bond angle is 137° (13). The side chains of three amino acids relevant to this study, H234, L264, and W283, are shown as sticks. For clarity, the PAS structure is shown rotated 180° around the x axis compared with the orientation shown in panel a. (c) Cyanide-bound heme and a structural overlay showing the locations of the L264 and W283 side chains in both the unliganded (Fe³⁺ heme, gray side chains; PDB entry 4HI4) (14) and liganded (Fe³⁺-CN heme; colored side chains) (13) Aer2 PAS domain. The position of the W283 nitrogen, which is predicted to bond with O₂, is shown in blue. Abbreviations: res, residue; CN, cyanide.

The Aer2 gas response is initiated in the PAS (Per-ARNT-Sim) domain, which itself binds pentacoordinate b-type heme (4). PAS domains are common sensing and signaling domains in nature. They have a broadly conserved structure that consists of a central antiparallel β-sheet with five β-strands (Aβ, Bβ, Gβ, Hβ, and Iβ) flanked by several α-helices (Cα, Dα, Eα, and Fα) (12). There are currently two structures for the Aer2 PAS domain: one contains cyanide bound to ferric heme (cyanomet, or Fe³⁺-CN; PDB entry 3VOL) (13) (Fig. 1b), and the other contains unliganded ferric heme (Fe³⁺; PDB entry 4HI4) (14). These structures revealed several unusual PAS features, including an extended Cα/Dα helix, a short 3₁₀ helix, called Eη (that replaces Eα), heme coordination via a His residue on Eη, and potential O₂ stabilization via the indole group of a Trp residue on Iβ (Fig. 1b). In contrast, other PAS domains with b-type heme, like those in E. coli DOS (EcDOS) or Sinorhizobium meliloti FixL (RmFixL), coordinate heme with a His residue on the Fα helix and stabilize O₂ binding via an Arg residue on Gβ (15). The two Aer2 PAS structures represent nonphysiological heme states (cyanomet and ferric heme), but they do represent structures with and without ligand, and overlaying these two structures highlights several residues that may be important for conformational signaling. In the absence of ligand, the Iβ Trp residue W283 appears to rotate ∼90°, and an adjacent Leu residue, L264 on Hβ, contracts toward the heme iron center to occupy the position where CN⁻ was bound (Fig. 1c) (14). The heme itself appears to shift ∼2.0 Å upon ligand binding, and the heme pocket adjusts accordingly (14).

The Aer2 PAS domain is flanked on either side by poly-HAMP units (Fig. 1a). Individual HAMP domains form parallel four-helix bundles that are commonly found in prokaryotic proteins as signal-transducing modules (16). In Aer2, there appears to be minimal PAS-HAMP interactions, and overall, Aer2 assumes a linear domain arrangement (Fig. 1a) (14). This contrasts with the aerotaxis receptor, Aer, where side-on PAS-HAMP interactions allow PAS to control the HAMP signaling state through direct interactions (17). For Aer2, several structures have been solved for the N-terminal HAMP domains (18, 19). Those structures show that HAMP1 is separated from HAMP2-3 by a helical extension (18, 19). HAMP1 is also largely dispensable for Aer2 function (4). In contrast, N-terminal HAMP2-3, and C-terminal HAMP4-5, each form integrated di-HAMP units that are indispensable for Aer2 function (4, 19). The HAMP1 and HAMP2 structures represent signal-on and signal-off states, respectively (18), lending support to the hypothesis that poly-HAMP chains relay signals by interconverting HAMP signaling states along the HAMP chain.

Based on experimental evidence, our current signaling model for Aer2 includes the following features: Aer2 PAS-heme binds O₂, generating a conformational signal that is transmitted to the PAS Iβ strand (4, 13, 14). N-terminal HAMP2-3 structures do not transmit signals but function to stabilize the PAS signaling state by altering their conformations in response to PAS ligand binding (4, 18). The PAS conformational signal is transmitted to C-terminal HAMP4-5, which together function as a unit to inhibit signaling from the kinase control module (4). Therefore, without a PAS ligand, the kinase control module conveys the signal-off state, but in the presence of PAS ligand, HAMP4-5 no longer inhibits the kinase control module, resulting in a signal-on output and the autophosphorylation of bound CheA2. Without HAMP4-5, the default state of the isolated kinase control module is signal-on (4). The purpose of the current study is to clarify the mechanisms used by the Aer2 PAS domain to bind heme, regulate ligand binding, and initiate conformational signaling. We provide evidence that (i) the Eη His coordinates heme binding, (ii) the hydrophobic heme pocket is crucial for stable heme binding, (iii) O₂ is the native ligand of the Aer2 PAS domain, (iv) the unprecedented Iβ Trp stabilizes O₂-binding but not CO-binding and plays a pivotal role in signal initiation, and (v) the Hβ Leu and other conserved PAS residues are important for heme binding, stable gas binding, and signal transduction.

RESULTS

Aer2 PAS coordinates heme with a uniquely positioned histidine residue.

Structural studies suggest that the Aer2 PAS domain from P. aeruginosa coordinates b-type heme via a His residue (H234) that resides on a short Eη helix (Fig. 1b) (13, 14). In contrast, other PAS domains coordinate b-type heme with a His residue on the PAS Fα helix (15, 20). Notably, Eη and Fα His residues are both highly conserved in Aer2-PAS homologs (Fig. 2a; see also Fig. S1 in the supplemental material). To test the contributions of each histidine to heme binding in P. aeruginosa Aer2, H234A (Eη His), H239A (Fα His), and H234A/H239A were introduced into the PAS peptide Aer2[173-289]. Aer2[173-289] is expressed with an N-terminal 6× His tag and contains all necessary PAS heme-binding components (4). The purified PAS-H234A peptide showed a significant heme-binding defect, whereas PAS-H239A retained wild-type (WT) heme content (Fig. 2b and c). This confirms that the Eη His is the predominant means of coordinating heme in Aer2. However, 20% of PAS-H234A molecules retained heme, and PAS-H234A/H239A exhibited a significant decrease in heme content versus H234A alone (6% heme, P < 0.05) (Fig. 2c). Thus, the dual His replacement peptide has a lower heme affinity, suggesting that H239 contributes to heme coordination in the absence of H234.

FIG 2.

Heme coordination in the Aer2 PAS domain. (a) Location of the Eη (H234) and Fα His (H239) side chains in the cyanomet structure of the Aer2 PAS domain (13). Aer2-PAS structures indicate that the Eη His coordinates heme (13, 14), whereas the Fα His coordinates heme in other PAS-heme proteins (15, 20). Both His residues are highly conserved in Aer2-PAS homologs (see Fig. S1). (b) Purified Aer2 PAS[173-289] peptides (imidazole bound; 2.6 to 4 mg ml⁻¹) showing less red color in Aer2-H234A and Aer2-H234A/H239A than in WT Aer2 and Aer2-H239A. (c) Heme content of PAS peptides with Eη and Fα His replacements, given as a percentage of WT PAS heme content, corrected for peptide concentration (see Materials and Methods). Error bars represent the standard deviations from multiple experiments. Abbreviations: CN, cyanide; WT, wild type.

To determine the effect of the His substitutions on Aer2 signaling, mutations encoding H234A and H239A were introduced separately into full-length aer2 in an E. coli expression plasmid. Both Aer2 mutants had steady-state expression levels comparable to that of WT Aer2 (Fig. 3a). When WT Aer2[1-679] is expressed in chemoreceptorless E. coli BT3388, it directs E. coli to tumble in the presence of O₂ because Aer2 signaling activates the E. coli chemotaxis cascade (4). When O₂ is replaced with N₂, Aer2 no longer signals, and after 5 to 10 s, BT3388 cells resume smooth-swimming behavior (∼2% of the cells tumble at any time) (4). BT3388 cells expressing Aer2-H239A behaved like cells expressing WT Aer2: cells tumbled in air (20.9% O₂) and had smooth-swimming behavior in N₂. In contrast, Aer2-H234A orchestrated tumbling in air like that of WT Aer2, but cells remained tumbling biased in N₂ (∼60% of cells tumbled in N₂ versus ∼2% for WT Aer2). Aer2-H234A therefore has a signal-on bias (Fig. 4a). The partial response may reflect the ability of a small proportion of heme-retaining molecules to respond to O₂ changes.

FIG 3.

Steady-state cellular levels of full-length Aer2 proteins and PAS peptides in E. coli. (a) Steady-state levels of full-length Aer2 proteins compared with WT Aer2[1-679] in E. coli BT3388. Aer2 expression was induced with 50 μM IPTG, and protein levels were determined from Western blots, an example of which is shown in the inset box. Lanes are from the same gel; intervening lanes are represented by white space. (b) Steady-state levels of PAS peptides compared with WT PAS[173-289] in E. coli BL21(DE3). Aer2 peptide expression was induced with 100 μM IPTG, and protein levels were determined from Western blots. (See the example in the inset box. Lanes are from the same gel; intervening lanes are represented by white space.) Error bars represent the standard deviations from multiple experiments.

FIG 4.

Aer2 mutant phenotypes in temporal assays. (a) Effects of amino acid substitutions on Aer2-mediated behavior in E. coli BT3388. Signal-off mutants exhibited smooth-swimming behavior (∼2% tumbling) in both air and N₂, whereas signal-on mutants tumbled constantly in both air and N₂. Neither signal-off nor signal-on mutants responded to the introduction or removal of O₂. Signal-off-biased mutants responded to the introduction of O₂ but adapted, unlike WT Aer2 in BT3388, which remained signal-on in the presence of O₂. Signal-on-biased mutants responded to the removal of O₂, but at least 50% of the cells continued to tumble in N₂. Residue substitutions marked by an asterisk resulted in receptors that could respond to CO, i.e., they directed cell tumbling in the presence of CO. CO responses could not be determined for signal-on-biased and signal-on mutants. (b) Alanine mutants mapped onto the cyanomet structure of Aer2. Original side chains are shown as color-coded sticks based on the O₂ responses listed in panel a.

When WT Aer2 is expressed in E. coli, cells respond to both O₂ and CO (4). To test for a CO response, E. coli BT3388 cells expressing Aer2 are monitored in CO temporal assays. In these assays, cells are perfused with N₂ (to remove O₂) until they resume smooth swimming, after which CO is perfused for 10 s. If the receptor can respond to CO, cells tumble and continue to tumble for up to 30 s after CO has been removed (4). To determine if Aer2-H239A can respond to CO, cells expressing Aer2-H239A were perfused with CO under anaerobic conditions. Similar to WT Aer2, cells expressing Aer2-H239A responded to CO by tumbling, and the tumbling persisted for ∼30 s after CO was removed. However, a CO response could not be determined for Aer2-H234A, because cells expressing Aer2-H234A tumbled too extensively in the absence of O₂.

PAS structures suggest a signaling mechanism.

Two PAS domain structures currently exist for Aer2, one with cyanide bound to ferric heme (cyanomet [Fe³⁺-CN]) (13) and another containing ferric heme without ligand (ferric heme [Fe³⁺]) (14). The ligand-bound structure suggests that Aer2 stabilizes O₂ binding via a Trp residue on the PAS Iβ strand (W283) (Fig. 1) (13). Moreover, an overlay of the two PAS structures suggests that both the Iβ Trp and an adjacent Leu residue on Hβ (L264) reorient in response to ligand binding. In the absence of ligand, the indole group of the Iβ Trp may rotate ∼90°, whereas the adjacent Hβ Leu residue contracts toward the heme iron center to occupy the position where CN⁻ was bound (Fig. 1c) (14). To determine the importance of W283 and L264 for Aer2 sensing and signaling, we performed site-directed random mutagenesis on each residue and analyzed the effects on receptor function, heme binding, and ligand binding.

The Iβ Trp is important for gas binding and signal initiation.

The PAS Iβ Trp is 100% conserved in 100 Aer2 PAS-like sequences (Fig. S1) and may be essential for stabilizing heme-O₂ binding. To determine the role of the Iβ Trp in P. aeruginosa Aer2, we performed site-directed random mutagenesis on the W283 codon in the construct that expresses full-length Aer2. Expression was induced in E. coli BT3388 with 200 μM isopropyl-β-d-thiogalactopyranoside (IPTG), and individual mutants were screened under the microscope for behavioral defects. Mutants with non-WT behavior were sequenced to determine the amino acid substitution at W283. After several rounds of mutagenesis and screening, 12 amino acid changes were identified at W283 that altered behavior (Fig. 4a). W283H and W283Y were not identified during the screen, but they were specifically engineered since these amino acids stabilize O₂-binding in other heme proteins (21–23). W283A was similarly not identified during screening but was created as part of the PAS alanine mutagenesis described below.

All 15 of the W283 mutant proteins were stably expressed in E. coli BT3388 (Fig. 3a), but 12 of the receptors were signal-off receptors that did not respond to the addition or removal of O₂ (Fig. 4a). Cells expressing these receptors swam smoothly in both the presence and absence of O₂, even after induction with 1 mM IPTG to produce high cellular levels of Aer2. In contrast, cells expressing Aer2-W283F, L, or I retained some functionality (Fig. 4a). Aer2-W283F and Aer2-W283L were signal-on-biased mutants that orchestrated tumbling in air like cells expressing WT Aer2, but when air was removed, 50 to 80% of the cells continued to tumble (Fig. 4a). Cells expressing Aer2-W283I had an inverted phenotype where ∼50% of the cells tumbled in N₂ but became smooth swimming after ∼30 s in air.

Heme-CO binding does not require amino acid stabilization and should not require W283. However, only two of the 15 W283 mutants, Aer2-W283I and Aer2-W283V, responded to CO (Fig. 4a). Cells expressing these mutants tumbled when CO was added in either N₂ or air. It was not possible to determine whether cells expressing Aer2-W283F or Aer2-W283L responded to CO, because cells expressing these mutants tumbled too extensively to determine a CO response.

To determine the O₂ and CO binding affinities of W283 mutants, W283-encoding mutations were transferred into the Aer2-PAS expression construct, Aer2[173-289], and the PAS peptides were purified on nickel-nitrilotriacetic acid (Ni-NTA)–agarose. W283 mutants that were analyzed included those that responded to O₂ or CO and 11 of the signal-off mutants; these were compared with WT Aer2 (which was determined to have an O₂ dissociation constant [K_d] of 16 μM and a CO K_d of 2 μM) (Fig. S2 shows WT O₂ and CO titrations). Unexpectedly, PAS peptides for nine of the 11 W283 signal-off mutants exhibited very low heme content when purified (3 to 37.5% of WT heme levels) (Fig. 5a), and gas-binding affinities could not be determined. To test whether these heme-binding defects were also present in full-length receptors, we purified full-length Aer2-W283Y and Aer2-L264N (see below), both of which had PAS peptides with heme-binding defects (Fig. 5a). Neither of the purified full-length receptors showed heme binding (data not shown). The other signal-off mutants, W283C and W283V, had sufficient heme content to analyze (Fig. 5a). Aer2-W283C and Aer2-W283V did not respond to O₂ and did not bind it; during O₂ titrations, both mutants exhibited Met-heme spectra with rapid oxidation from Fe(II) to Fe(III) heme (Met-heme spectra were independently verified by oxidizing proteins with potassium ferricyanide and comparing with the spectra from O₂ titrations) (Fig. S2c). However, both mutants bound CO with WT affinity, and Aer2-W283V was able to respond to it (Fig. 4a and 5b).

FIG 5.

PAS peptide heme content and gas-binding affinities. (a) Heme content of PAS peptides with amino acid substitutions, given as a percentage of WT PAS heme content, corrected for peptide concentration (see Materials and Methods). Values below 40% indicate a substantial heme-binding defect. Aer2[173-289]-L264P contained no measureable heme. Error bars represent the standard deviations from multiple experiments. (b) PAS peptide O₂ and CO binding affinities. A dash indicates that O₂- or CO-bound spectra were not observed, so binding affinities could not be determined.

Of the three W283 mutants that responded to O₂ in the behavioral assay (Aer2-W283F, -L, and -I), only Aer2-W283F appeared to bind O₂, and its O₂ affinity was similar to that of WT Aer2-PAS (Fig. 5b). In contrast, Aer2-W283L and Aer2-W283I both responded to O₂, but purified PAS peptides with these substitutions did not bind O₂ (Fig. 4 and 5b) and were rapidly oxidized from Fe(II) to Fe(III) heme. Sawai et al. similarly reported that purified full-length Aer2-W283L does not bind O₂ (13). It is possible that O₂ binding to these mutants is too transient to observe during in vitro O₂ titrations but is sufficiently stable in vivo to generate a behavioral response. All of the W283 mutants tested bound CO with affinities similar to that of WT Aer2 (Fig. 5b), irrespective of whether the corresponding full-length receptors responded to CO or not (Fig. 4). Overall, these data indicate that the Iβ Trp is important for stable heme and O₂ binding but not CO binding and is important for signal initiation in the Aer2 PAS domain.

Substitutions at the Hβ Leu alter gas binding and signaling.

Aer2 PAS structures suggest that the Hβ Leu residue, L264, is involved in initiating PAS signaling (14). The L264 side chain appears to occupy the PAS ligand-binding site but swings out of the site when ligand binds (Fig. 1c) (14). Hβ Leu is well conserved in Aer2 PAS-like sequences, but other hydrophobic amino acids are also found at the same position, primarily Val and, to a lesser extent, Ile (Fig. S1). To determine if Aer2 can function with Val at 264, Aer2-L264V was engineered by site-directed mutagenesis. Full-length Aer2-L264V mediated an O₂ response but exhibited a 30-s-delayed smooth-swimming response in N₂ and did not respond to CO. Therefore, Aer2 can function with Val at 264, but the behavioral response is restricted to O₂. Notably, PAS-L264V bound O₂ and CO with affinities that were similar to that of the WT (Fig. 5b).

To determine whether other replacements at L264 affect the O₂ response, we performed site-specific random mutagenesis on the L264 codon and screened for defective mutants, as outlined above for W283. L264A was not identified during the screen but was created as part of the PAS alanine mutagenesis described below. After several rounds of mutagenesis and screening, 12 amino acid substitutions were identified at L264 that altered the O₂ response (Fig. 4a). All of these mutants were stably expressed in E. coli BT3388 (Fig. 3a). Eight of the mutants were nonfunctional, signal-off mutants (Fig. 4a), even after induction with 1 mM IPTG. In contrast, Aer2-L264A was locked signal-on, causing cells to tumble constantly in air and in N₂ (Fig. 4a). The remaining three mutants, Aer2-L264F, -I, and -Q, were signal-off-biased mutants that responded to O₂ (Ile shows some conservation at this position) (Fig. 4a; Fig. S1). Cells expressing Aer2-L264I and Aer2-L264Q had WT O₂ responses, whereas cells expressing Aer2-L264F had a reduced tumble response to O₂ (∼60% of cells tumbled). However, all three mutants adapted to O₂ (cells became less tumble biased) over the course of several minutes. Only one of the L264 mutants, Aer2-L264F, tumbled in response to the addition of CO (this mutant also responded to O₂) (Fig. 4a). A CO response could not be determined for Aer2-L264A, because cells expressing this Aer2 variant tumbled constantly in the presence and absence of O₂.

To determine the O₂ and CO binding affinities of the L264 mutants, L264-encoding mutations were transferred to the Aer2 PAS peptide Aer2[173-289], and the peptides were purified on Ni-NTA–agarose. L264 mutants that were analyzed included those that responded to O₂ or CO, the signal-on mutant Aer2-L264A, and five of the signal-off mutants; these were compared with WT Aer2-PAS (Fig. 5). Four of the mutants expressed PAS peptides that contained very low heme content when purified. This included three of the signal-off mutants, L264N, -P, and -W, and one of the functional mutants, L264Q (0 to 37% of WT heme levels) (Fig. 5a). Gas-binding affinities could not be determined for PAS-L264N, -P, and -Q but were determined for PAS-L264W due to higher peptide purity. PAS-L264P showed no detectable heme spectra, even after scanning concentrated (135 μM) protein. There was no defect in the steady-state cellular expression level of PAS-L264P (Fig. 3b). Of the remaining L264 mutants, three bound O₂ and CO (L264F, -I, and -V), three bound CO but not O₂ (L264A, -G, and -W), and one bound neither gas (L264K) (Fig. 5b). The three L264 mutants that bound O₂ (L264F, -I, and -V) also responded to it. In contrast, PAS-L264G, -I, -V, and -W all bound CO but did not respond to it. Of the L264 mutants that did not stably bind O₂, PAS-L264A and PAS-L264G were rapidly oxidized by O₂ and PAS-L264K was slowly oxidized by O₂. In contrast, PAS-L264W showed no shift in its Soret maxima during O₂ titrations, suggesting the absence of heme-O₂ interactions. This finding lends support to the hypothesis that L264 must move out of the ligand-binding site to allow for O₂ binding. However, L264W still allowed CO binding, indicating that the CO-binding angle (perpendicular to the plane of the heme) was permitted (Fig. 5b).

Of all the mutants in this study, the signal-off mutant, Aer2-L264K, was the only mutant that did not bind either O₂ or CO. Moreover, the deoxy spectra of PAS-L264K contained a β band and a prominent α band (Fig. S2d), suggesting the formation of a hexacoordinate heme. This would entail coordination to H234 on the proximal side of the heme as well as coordination on the distal side, quite possibly by the amino group of lysine (e.g., Lys can coordinate heme in place of Met in cytochrome c-550 [24]). This differs from WT deoxy-Aer2, which contains pentacoordinate heme (Fig. S2d) (4).

Aer2 signaling is disrupted by alanine replacements at conserved residues.

To complement the mutagenesis experiments on W283 and L264, 16 residues that are highly conserved in Aer2 homologs (Fig. S1, marked by asterisks) were selected for site-directed alanine mutagenesis (A178 was replaced with Val). Conserved Gly residues that are structural elements at turns were excluded. The results for L264A and W283A were discussed above. Most of the mutants exhibited stable steady-state expression levels in E. coli BT3388 (Fig. 3a). However, 10 of the 16 Ala mutants were signal-off mutants and did not respond to either O₂ or CO (Fig. 4). This included the four heme cleft mutants that were tested: M187A, I195A, F220A, and F233A. In contrast, Aer2-A178V, Aer2-P237A, and Aer2-Q240A mediated WT responses to both O₂ and CO (Fig. 4). Aer2-D231A similarly orchestrated WT-like responses to O₂ and CO but had a 30-s-delayed smooth-swimming response in N₂ after O₂ was removed. Overall, the Ala mutants that had WT or signal-on-biased behavior congregated on the Eη and Fα helices (Fig. 4b, orange and yellow residues). Very few signal-on mutants were identified in this study. Aer2-T287A, like Aer2-L264A, was a signal-on mutant that caused cells to tumble constantly in both air and N₂. Because of this, a CO response could not be determined for Aer2-T287A. L264 and T287 both reside on the PAS β-sheet, which is the signal output surface of the PAS domain (12, 14).

To determine gas-binding affinities for the four heme-pocket mutants (M187A, I195A, F220A, and F233A), relevant mutations were transferred to the construct expressing Aer2[173-289] and the PAS peptides were purified. PAS peptides with I195A and F233A had severe heme-binding defects (Fig. 5a), even though neither of these mutants had steady-state expression defects (Fig. 3b). In contrast, PAS peptides with M187A and F220A both bound heme (Fig. 5a) and could bind CO (Fig. 5b), yet neither mutant responded to CO. Aer2-F220A neither bound nor responded to O₂, whereas Aer2-M187A bound O₂ with WT affinity but was also rapidly oxidized by O₂ (Fig. 5a). Sawai et al. similarly reported that full-length Aer2-M187A binds O₂ (13). Aer2-M187A was the only mutant in this study that bound, but did not respond, to O₂.

Signal-on behavior is independent of Aer2 methylation.

When WT Aer2 is expressed in E. coli, it does not adapt to O₂. This is because Aer2 is methylated by the E. coli methyltransferase CheR but is not demethylated by the E. coli methylesterase CheB (4). When WT Aer2 is expressed in an E. coli strain lacking CheR and CheB, Aer2 remains unmethylated and the cells have a low tumbling frequency (∼5% of cells tumble in O₂ [4]). Hence, robust signal-on behavior requires receptor methylation. In this study, two locked signal-on mutants (L264A and T287A) and three signal-on-biased mutants (H234A, W283F, and W283L) were identified. To determine if their phenotypes were dependent on receptor methylation, full-length receptors containing each of these amino acid substitutions were expressed in E. coli UU2610, which lacks all E. coli chemoreceptors, as well as CheR and CheB (25). In UU2610, the tumbling bias of Aer2-L264A decreased ∼20% in both air and in N₂. However, the tumbling biases of the other four mutants were not diminished by the lack of receptor methylation in UU2610. This suggests that the signal-on biases of these receptors are due primarily to the amino acid changes in the PAS domain and not receptor methylation status.

DISCUSSION

The Eη histidine coordinates heme in the Aer2 PAS domain.

Crystal structures of the Aer2 PAS domain identify the Eη His (H234) as the proximal heme-coordinating ligand (13, 14). This differs from the PAS domains of RmFixL, EcDOS, Acetobacter xylinum PDEA-1 (AxPDEA-1), and Burkholderia xenovorans RcoM, where, in each case, an Fα His residue coordinates b-type heme (15, 20). Although the Fα His is well conserved in Aer2-PAS homologs (see Fig. S1 in the supplemental material), in P. aeruginosa Aer2 it resides ∼10 Å from the heme Fe, in contrast to the Eη His, which lies ∼2 Å away. In the current study, the Fα His substitution, H239A, did not affect PAS heme content or behavioral responses. In contrast, the Eη His substitution, H234A, imposed substantial heme-binding and behavioral defects (Fig. 2), confirming that it is the proximal coordinating His of Aer2. However, Aer2-H234A had a signal-on bias, unlike all other heme-binding mutants in this study, which were signal-off or signal-off biased. In addition, the 80% heme loss in PAS-H234A is less than that observed for other PAS-heme domains when the proximal coordinating His is replaced. For example, His substitutions in RcoM result in <1% heme (20), and a His-to-Ala alteration in the Aer2 PAS-2 domain from Vibrio cholerae results in 2% heme (K. J. Watts, unpublished data). In Aer2-PAS, it is unclear how heme might be coordinated in the absence of H234, particularly since H239 lies at a more remote location at the entrance of the heme cleft. A dual PAS-H234A/H239A mutant had significantly less heme than PAS-H234A (Fig. 2), suggesting that Fα-H239 contributes to heme coordination in the absence of Eη-H234. Alternatively, the dual His replacements might distort the heme pocket in a way that prevents heme retention. Heme might instead be retained by PAS-H234A through hydrophobic pocket interactions. Hydrophobic interactions are sufficient to bind b-type hemes in the YybT family proteins from Bacillus and Geobacillus, which have no natural proximal heme-coordinating residue in their PAS domains (26, 27).

The hydrophobic heme cleft is critical for stabilizing heme binding in Aer2.

In Aer2-PAS, the heme cleft is a hydrophobic pocket in which the imidazole ring of H234 coordinates the heme-Fe, and H251 hydrogen bonds to the heme-7-proprionate (13). In this study, replacing heme-coordinating H234 caused a substantial heme-binding defect (PAS-H234A retained 20% heme) (Fig. 2). However, other amino acid substitutions in both the proximal (I195A and F233A) and distal (L264N, -P, -Q, and -W and W283A, -G, -H, -K, -P, -Q, -S, -T, and -Y) heme cleft likewise caused substantial heme-binding defects (0 to 37.5% of WT heme content) (Fig. 5a). Remarkably, some of these defects were greater than that caused by H234A. This suggests that hydrophobic heme cleft interactions are critical for stabilizing heme binding in Aer2. On the proximal side of the cleft, I195 and F233 both reside ∼4 Å from, and parallel to, the heme; the severe defects caused by Ala substitutions at these residues (0.6 to 3% heme content) shows that moderate perturbations in the PAS heme pocket can alter heme binding. Moreover, perturbations in O₂ stabilizing and signaling residues, W283 and L264, also affected heme binding. In Bradyrhizobium japonicum FixL, replacing the distal O₂-stabilizing Arg residue with Ala relaxes heme-protein coupling (28), and a similar event could be responsible for heme loss in PAS-W283 mutants.

Oxygen is the native ligand of the Aer2 PAS domain.

In the absence of imidazole, P. aeruginosa Aer2 purifies in the oxy-bound state (13). However, Aer can signal in response to the binding of O₂, CO, and NO (4). This is atypical for heme sensors; they often bind all three gases, but typically they respond to only one. For example, the histidine kinase FixL is inhibited by O₂ but not by CO or NO, even though it binds these two more tightly (29). Protohemes generally bind O₂, CO, and NO with relative affinities of 1, 10³, and 10⁶, respectively (30, 31). In this study, we determined that the O₂ and CO affinities of the isolated Aer2 PAS domain are 16 μM and 2 μM, respectively (Fig. 5b). The O₂ affinity of Aer2-PAS is comparable with the O₂ affinities of the PAS-heme O₂ sensors EcDOS (13 μM), AxPDEA-1 (∼10 μM), and RmFixL (31 μM) (15, 32). In the DOS, PDEA-1, and FixL PAS domains, a Gβ Arg residue interacts directly with bound O₂: substitutions at this residue substantially lower O₂ but not CO affinities and affect O₂-regulated behavior (28, 33). In Aer2, substitutions at the Iβ Trp likewise prevented O₂ but not CO binding (Fig. 5b, W283C, -I, -L, and -V). Because changes in highly conserved residues predominately affect O₂ binding properties and responses, these data suggest that O₂ is the native ligand of the Aer2 PAS domain. We did not test NO binding in this study. However, unlike Aer2, heme-NO sensors usually exclude O₂ binding; e.g., H-NOX (heme-nitric oxide/oxygen) proteins that lack hydrogen bond donors bind NO instead of O₂ (34, 35), the PAS-heme domain of YybT is rapidly oxidized by O₂ (27), and nitrophorins exclude O₂ by maintaining their heme in the Fe(III) state (36).

The role of Aer2 PAS residues in ligand binding and signal transduction.

The data from this study indicate that the Iβ Trp, W283, stabilizes O₂ binding to the Aer2 PAS-heme domain. Twelve of 15 W283 mutants, including those with amino acids that commonly stabilize O₂ binding to other heme proteins (His, Tyr, and Arg [37]), resulted in signal-off Aer2 receptors that did not respond to O₂. For W283 mutants that did not stably bind O₂ in vitro (W283C, -I, -L, and -V), the heme cofactor was rapidly oxidized upon exposure to O₂ (Fig. S2c). This has been observed similarly for EcDOS Gβ Arg mutants (33). Our data also indicate that W283 is critical for initiating conformational signaling from the PAS domain. This perhaps is not surprising given that the PAS Iβ strand connects directly to the C-terminal HAMP4 domain, and W283 resides close to the PAS DXT motif, which has been proposed to be involved in conformational signaling (12, 14). None of the 15 W283 mutants in this study retained WT function (Fig. 4a), and PAS-W283F was the only Trp mutant that preserved a WT O₂ affinity (Fig. 5b). The side chain of Phe is similar in size to the nitrogen-containing Trp pyrrole ring that is predicted to hydrogen bond with heme-bound O₂. However, Phe lacks a hydrogen-bonding moiety. One possibility is that O₂ binding is supported by a solvent molecule in the distal pocket that acts as a hydrogen bond donor; this scenario was observed in a DevS mutant from Mycobacterium tuberculosis when the O₂-stabilizing Tyr residue was replaced with Phe (38). In this study, Aer2-W283F retained partial functionality (as a signal-on-biased mutant), but Aer2 is clearly fine-tuned to use Trp for O₂ binding and signal initiation.

The hypothesis that the Hβ Leu, L264, moves out of the ligand-binding site when O₂ binds to heme (Fig. 1c) (14) was supported by the results of this study. Notably, the bulky Trp substitution, L264W, appeared to block O₂ binding, and L264K formed a hexacoordinate heme that did not bind O₂. However, the Hβ Leu itself was not specifically required for function: Aer2-L264F and Aer2-L264Q retained some functionality, and L264 substitutions that occur in other Aer2 PAS-like homologs, Val and Ile (Fig. S1), resulted in Aer2 receptors that bound and responded to O₂ (Fig. 4 and 5). Interestingly, smaller hydrophobic replacements, L264A and L264G, resulted in nonresponsive Aer2 receptors that did not bind O₂ and furthermore were rapidly oxidized by it. This suggests that (i) the size of the amino acid at the Hβ Leu is important and (ii) Hβ Leu is choreographed to move concomitantly with the Iβ Trp so that Trp can rotate into place and bond with O₂ (Fig. 1c). Notably, O₂ binding alone was not sufficient for PAS signaling. Aer2-M187A bound O₂ but was unable to respond to it. M187 resides on the Aβ strand, adjacent to W283 on Iβ, but it does not exhibit a significant conformational shift between the cyanomet and ferric PAS-heme structures. Still, the distortion caused by the Ala replacement at residue 187 could feasibly block PAS β-sheet signaling rearrangements that are required for downstream signaling.

Like Aer2-M187A, most of the mutants in this study were signal-off mutants that did not respond to O₂ or CO (Fig. 4a). The amino acid changes in these mutants most likely disengage PAS control of the downstream HAMP4-5 unit so that HAMP4-5 continues to inhibit the activity of the kinase control module. D285A specifically disrupts the conserved DXT motif that has a proposed role in conformational signaling between the PAS and C-terminal HAMP domains (12, 14). In Aer2-N199A, the Ala substitution could also disrupt interactions between the PAS N-terminal cap (N-cap, Aα′ helix) (Fig. 1b) and the PAS core (Cα helix); these interactions are required for both structural stability (N199A was the least stable receptor in this study) (Fig. 3a) and for N-cap reorientation during signaling (14, 39).

All of the PAS peptides tested in this study (with the exception of PAS-L264K) bound CO with an affinity that was the same as that of the WT peptide or better (M187A and L264G bound CO in the nanomolar range) (Fig. 5b). This is not surprising, since heme-CO binding does not require amino acid stabilization due to its high inherent affinity for heme (30). However, heme-CO binding did not predict function. Mutants that responded to CO bound it with WT affinity (W283I and V and L264F), but so did many of the mutants that did not respond to CO (Fig. 4 and 5). In the latter instances, CO binding apparently could not induce the conformational changes required for signal transduction.

These studies provide insight into the mechanisms used by the Aer2 PAS domain to regulate heme binding and ligand binding and initiate conformational signaling. The results support the model based on differences between the cyanomet and ferric PAS-heme structures, corroborating the roles of W283 and L264 in O₂ stabilization and PAS signaling (Fig. 1) (14). When O₂ binds to Aer2-PAS, it generates a conformational signal that is transmitted via the PAS Iβ strand to modulate the activity of the C-terminal HAMP and kinase control domains. Future studies will test an expanded model whereby O₂-mediated signaling evokes a PAS dimer-to-monomer transition (14), resulting in the signal-off conformation of HAMP4, the signal-on conformation of HAMP5 (18), and deinhibition of the kinase control module (4).

MATERIALS AND METHODS

Bacterial plasmids and strains.

Full-length P. aeruginosa PAO1 Aer2[1-679] (PA0176) was expressed from pLH1, a pProEX-derived plasmid that expresses Aer2 with an N-terminal His₆ tag (4). The Aer2 PAS domain, Aer2[173-289], was likewise expressed from pProEX with an N-terminal His₆ tag (4). Full-length Aer2 was expressed in chemoreceptorless E. coli strains BT3388 (tar tsr trg tap aer [40]) and UU2610 (tar tsr trg tap aer cheR cheB [25]). UU2610 also lacks the E. coli adaptation enzymes CheR and CheB. Aer2 PAS peptides were expressed in E. coli BL21(DE3).

Mutagenesis and cloning.

Site-directed mutagenesis was performed on pLH1 using site-specific primers and PfuUltra II Fusion DNA polymerase (Agilent Technologies, Santa Clara, CA). To replace native codons with Ala or Val codons, 30 amplification cycles were performed with an annealing temperature of 55°C. For site-directed random mutagenesis, primers containing an equimolar mix of all four nucleotides at the L264 or W283 codons were used; however, the amplification conditions described above consistently created DNA insertions following the primer site. To solve this problem, we tested stepwise annealing temperatures from 55°C to 68°C and analyzed the constructs created. The lowest proportion of DNA inserts occurred when 68°C was used as the annealing temperature and 20 amplification cycles were performed. These conditions yielded no obvious bias for codon replacements with one, two, or three nucleotide changes and subsequently were used to create most of the site-specific random mutants identified in this study. Site-specific mutagenesis products were treated with DpnI (New England BioLabs, Ipswich, MA) to remove template strands and then electroporated into E. coli. Aer2 expression was induced with 600 μM IPTG, and products of the correct size were confirmed by Western blotting with HisProbe-horseradish peroxidase (Thermo Scientific, Rockford, IL). All mutations were confirmed by sequencing the entire coding sequence of aer2.

To create Aer2-PAS peptides expressing specific amino acid changes in the PAS domain, the PAS coding region (residues 173 to 289) was PCR amplified from pLH1-derived plasmids using PfuUltra II fusion DNA polymerase. PCR products were ligated into the NcoI and SalI sites of pProEX. Peptide expression and DNA sequencing were performed as described above.

Steady-state cellular Aer2 levels.

The steady-state cellular levels of the full-length Aer2 mutants were compared with that of WT Aer2 after inducing BT3388 cells with 50 μM IPTG. In contrast, the cellular levels of the PAS peptides were compared with that of WT Aer2[173-289] after inducing BL21(DE3) cells with 100 μM IPTG. Samples were electrophoresed in duplicate, and experiments were repeated on two to four separate days. Bands were visualized on HisProbe Western blots and quantified on a BioSpectrum digital imager (UVP, Upland, CA).

Behavioral assays.

BT3388 cells were grown at 30°C in tryptone broth containing 0.5 μg ml⁻¹ thiamine and induced with 200 μM IPTG. At this induction level, the number of Aer2 receptors in E. coli is comparable with the total number of chemoreceptors in WT E. coli cells (4). Cells were placed into a gas perfusion chamber where the gas was toggled between air (20.9% O₂) and N₂, and cell behavior was analyzed (41, 42). Mutants that were signal-off (smooth swimming in air and in N₂) were retested after induction with 1 mM IPTG to produce higher cellular levels of Aer2. Behavioral responses to O₂ were repeated two or more times on at least two separate days. To determine CO responses, BT3388 cells (induced with 200 μM IPTG) were perfused with N₂ for 30 s prior to perfusing with CO gas (>99% purity; Sigma-Aldrich, St. Louis, MO), which was added through the open end of the chamber for 10 s. For Aer2-W283I and Aer2-W283V, CO responses were also tested while air was being perfused.

Protein purification.

WT Aer2[173-289]/BL21(DE3) and relevant mutants were grown in LB broth, Lennox, containing 25 μg ml⁻¹ 5-aminolevulinic acid (Sigma-Aldrich) to enhance heme synthesis and incorporation. After 3 to 5 h of induction with 600 μM IPTG, cells were centrifuged at 10,000 × g for 15 min and resuspended to 1% of their original volume in lysis buffer (50 mM Tris, pH 7.5, 500 mM NaCl, and 10 mM imidazole) containing 0.3 mg ml⁻¹ lysozyme, 1 μg ml⁻¹ DNase I, and 100 μl of protease inhibitor cocktail for His-tagged proteins (Sigma-Aldrich). The cells were lysed by freeze-thawing five times, followed by sonication. Soluble protein was acquired by removing cellular debris at low speed (10,000 × g for 20 min) and the membrane fraction at high speed (485,000 × g for 1 h). The high-speed supernatant was applied to an Ni-NTA–agarose column (Qiagen, Valencia, CA) and allowed to empty by gravity flow. The column was washed with 10 column volumes of wash buffer 1 (50 mM Tris, pH 7.5, 500 mM NaCl, and 20 mM imidazole), followed by 8 to 10 column volumes of wash buffer 2 (50 mM Tris, pH 7.5, 500 mM NaCl, and 50 mM imidazole). Aer2 peptides were eluted by adding 1 ml of elution buffer (50 mM Tris, pH 7.5, 500 mM NaCl, and 250 mM imidazole) to the column, but only the red-colored fraction was collected. For proteins with no obvious red color, two ∼0.4-ml elution fractions were collected. Aer2 peptides usually were most concentrated in the second eluted fraction. The concentration of eluted protein was determined using a bicinchoninic acid protein assay (Thermo Scientific), and the quality of the sample was determined by staining SDS-PAGE gels with Coomassie brilliant blue.

Heme binding.

The proportion of heme bound to the WT Aer2[173-289] PAS domain was determined using a pyridine hemochrome assay (43, with modifications communicated by M. Gilles-Gonzalez). Briefly, WT PAS peptide and hemin standards (10 to 50 μM) were added to an alkaline pyridine solution and scanned from 350 to 700 nm under both dithionite-reduced and ferricyanide-oxidized conditions. Heme concentrations were determined from the reduced minus oxidized spectra, using an extinction coefficient of 23.4 mM⁻¹ cm⁻¹ for the absorbance difference of A_{556 (reduced)} minus A_{539 (oxidized)}. The heme content determined from the pyridine hemochrome assay was used to standardize PAS-heme concentrations used in ligand-binding assays.

To determine whether purified PAS peptides had heme-binding defects, 10 μM imidazole-bound PAS peptides were scanned from 300 nm to 700 nm in a BioMate 3S spectrophotometer (Thermo Scientific). Samples were overlaid by zeroing at 700 nm, and the maximum absorbance of each Soret peak was determined. Maximum Soret absorbances were divided by the maximum Soret absorbance of the WT PAS peptide. Peptide concentrations were determined by electrophoresing 2.5 μg of each purified protein in duplicate on SDS-PAGE as outlined above, staining gels with Coomassie brilliant blue, and quantifying the density of each PAS peptide on a BioSpectrum digital imager. The average density of each PAS peptide was divided by the average density of the WT PAS peptide (which itself was usually 85 to 90% pure). The heme content/peptide ratio then was calculated for each mutant and averaged from multiple purifications. Ratios below 40% indicated a substantial heme-binding defect from which gas affinity constants generally were not determined.

Gas-binding affinities.

Deoxy-heme was created by adding 0.5 mM dithionite to 4 to 10 μM anaerobic PAS-heme in an anaerobic hood (Coy Laboratory Products, Grass Lake, MI). Deoxy-PAS was added to a quartz septum-sealed cuvette (Starna Cells, Atascadero, CA) and used directly for CO binding. For O₂ affinities, sufficient O₂ was added to the cuvette to oxidize the dithionite (as determined spectrophotometrically). To create gas-saturated buffers, buffer (50 mM Tris, pH 8.0, 50 mM KCl, and 5%, vol/vol, ethylene glycol) was perfused with either CO (Sigma-Aldrich) or air. To create 50% CO-saturated buffer, a volume of N₂-saturated buffer was added to an equal volume of CO-saturated buffer in a Reacti-Vial (Thermo Scientific) and used immediately. Gas solutions were transferred to gas-tight Hamilton syringes (Hamilton, Reno, NV) and titrated into the deoxy protein solution. Stepwise spectra were recorded on a Beckman DU 650 spectrophotometer (Beckman Coulter, Brea, CA) after each addition of buffer. The amount of bound gas was estimated from the UV-visible spectrum by linear interpolation of the unliganded (Fe²⁺) and liganded (Fe²⁺-O₂, Fe²⁺-CO) spectra. For WT PAS and most of the PAS mutants, the Soret maxima occurred at 428 to 432 nm for deoxy-heme, 414 to 416 nm for oxy-heme, and 421 to 422 nm for carbonmonoxy-heme. After O₂ titrations were complete, CO was perfused directly into the cuvette to differentiate O₂-bound protein from Met-heme protein. O₂-bound or ferrous protein, but not ferric protein, showed CO-bound spectra after the addition of CO.

Met-heme absorption spectra.

To create Met-heme, 50 μM purified PAS peptide was oxidized with 50 μM potassium ferricyanide at room temperature for 15 min. To evaluate Met-heme spectra, samples were purified on a Micro Bio-Spin column (Bio-Rad, Hercules, CA) and scanned spectrophotometrically.