Structure-Function Relationships in the HAMP and Proximal Signaling

JOURNAL OF BACTERIOLOGY, Mar. 2008, p. 2118–2127 0021-9193/08/$08.00⫹0 doi:10.1128/JB.01858-07 Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Vol. 190, No. 6

Structure-Function Relationships in the HAMP and Proximal Signaling Domains of the Aerotaxis Receptor Aer䌤 Kylie J. Watts, Mark S. Johnson, and Barry L. Taylor* Division of Microbiology and Molecular Genetics, Loma Linda University, Loma Linda, California 92350 Received 26 November 2007/Accepted 9 January 2008

Aer, the Escherichia coli aerotaxis receptor, faces the cytoplasm, where the PAS (Per-ARNT-Sim)-flavin adenine dinucleotide (FAD) domain senses redox changes in the electron transport system or cytoplasm. PAS-FAD interacts with a HAMP (histidine kinase, adenylyl cyclase, methyl-accepting protein, and phosphatase) domain to form an input-output module for Aer signaling. In this study, the structure of the Aer HAMP and proximal signaling domains was probed to elucidate structure-function relationships important for signaling. Aer residues 210 to 290 were individually replaced with cysteine and then cross-linked in vivo. The results confirmed that the Aer HAMP domain is composed of two ␣-helices separated by a structured loop. The proximal signaling domain consisted of two ␣-helices separated by a short undetermined structure. The Af1503 HAMP domain from Archaeoglobus fulgidus was recently shown to be a four-helix bundle. To test whether the Af1503 HAMP domain is a prototype for the Aer HAMP domain, the latter was modeled using coordinates from Af1503. Several findings supported the hypothesis that Aer has a four-helix HAMP structure: (i) cross-linking independently identified the same residues at the dimer interface that were predicted by the model, (ii) the rate of cross-linking for residue pairs was inversely proportional to the ␤-carbon distances measured on the model, and (iii) clockwise lesions that were not contiguous in the linear Aer sequence were clustered in one region in the folded HAMP model, defining a potential site of PAS-HAMP interaction during signaling. In silico modeling of mutant Aer proteins indicated that the four-helix HAMP structure was important for Aer stability or maturation. The significance of the HAMP and proximal signaling domain structure for signal transduction is discussed.

levels of CheY, which controls the direction of flagellar rotation. The result is the coupling of environmental conditions with bacterial swimming behavior. Although the mechanism of signaling in Aer is not completely understood, it is believed that reduction of FAD in the PAS domain initiates the signal that results in clockwise (CW) flagellar rotation. This is different from aerotaxis signaling in the Tsr chemoreceptor, which responds to changes in proton motive force (24). To date, the structures of at least 20 PAS domains have been determined, and 2 of these domains bind FAD: the NifL PAS domain from Azotobacter vinelandi (37) and the photoreceptor Vivid from Neurospora crassa (65). In both receptors, signaling is initiated by protonation of the flavin N-5 atom, leading to a change in the hydrogen bonding network within the PAS domain (37, 65). This constitutes a structural signal that is propagated to the surface of the PAS domain, generating an activated conformation. Among PAS domains, two different activated conformations have been observed: one in which the PAS FG loop has rigidified or moved (3, 28, 29, 31, 40, 47) and one in which an N-terminal (32, 54, 65) or C-terminal (33) helix has been displaced. Whichever PAS signaling mechanism is employed by Aer, the signal must ultimately be transduced to the HAMP domain. PAS and HAMP domains are present together in approximately 800 proteins from archaea, bacteria, and eukaryotes, most of which are putative or known histidine kinases or chemoreceptors (SMART; http://smart.embl-heidelberg .de/). HAMP domains consist of two amphipathic ␣-helices (AS-1 and AS-2) connected by a loop. This structure was first predicted by Le Moual and Koshland (42) and has since been

Aer, the aerotaxis receptor in Escherichia coli, senses environmental oxygen levels indirectly through the redox state of a flavin adenine dinucleotide (FAD) cofactor that is bound to an N-terminal PAS (Per-ARNT-Sim) (48) domain (9, 55). Unlike the sensing domains of the methyl-accepting chemoreceptors, which are periplasmic, the PAS redox sensor of Aer is cytosolic. Although the N-terminal sensing domains of the chemoreceptors and Aer are topographically different, each protein has two transmembrane helices; the second of these helices, TM2, is connected to a cytosolic HAMP (histidine kinase, adenylyl cyclase, methyl-accepting protein, and phosphatase) (6) domain, which precedes a highly conserved output signaling domain (Fig. 1). The signaling domain forms two long antiparallel helices that dimerize into a four-helix bundle (38, 51), and this bundle binds the histidine kinase, CheA, and coupling protein, CheW, to form a ternary complex (11, 27, 46). The minimal signaling unit is a dimer, although Aer and the chemoreceptors form mixed trimers of dimers (2, 30, 39) that act as a squad to amplify signals (2, 12, 23, 52). Within the chemoreceptors, signals are transmitted from the periplasm to the cytosol by a 1- to 2-Å piston movement of TM2 (18, 19, 25), but the mechanism of transduction through the HAMP and signaling domain is unknown. Ultimately, the conformational state of the receptor regulates the steady-state phosphorylation

* Corresponding author. Mailing address: Division of Microbiology and Molecular Genetics, Loma Linda University, Loma Linda, CA 92350. Phone: (909) 558-4881. Fax: (909) 558-4035. E-mail: bltaylor @llu.edu. 䌤 Published ahead of print on 18 January 2008. 2118

VOL. 190, 2008

Aer HAMP AND PROXIMAL SIGNALING DOMAINS

2119

silico Aer HAMP model, defining a possible PAS interaction surface that may be important for PAS-HAMP signaling. MATERIALS AND METHODS

FIG. 1. Domain map of Aer (left panel) and projected detail of the region scanned by cysteine in this study (right panel). Also indicated are substitutions between residues 210 and 258 that cause CW bias and null aerotaxis phenotypes (13, 43, 62; this study) (only null mutants that are not phenotypically rescued by other chemoreceptors are included). The null lesions were subsequently modeled in silico (see text for details); substitutions that were not permissible in PyMOL (italics) or caused structural defects when they were remodeled in Swiss-Model (underlined) are shown. TM, transmembrane; Sig., signaling.

supported by disulfide cross-linking of the Salmonella Tar HAMP domain (14) and by electron paramagnetic resonance measurements of the Natronomonas pharaonis HtrII HAMP domain (10). The recent nuclear magnetic resonance-derived structure of an Af1503 HAMP dimer from the archaeon Archaeoglobus fulgidus revealed a parallel four-helix coiled-coil arrangement for AS-1 and AS-2, with each connector packed into a groove between helices (36). Residues comprising the helical core of the HAMP domain were hydrophobic, as expected, although they were packed in an unusual knob-to-knob arrangement rather than the knob-into-hole conformation that is more commonly observed in coiled coils (36). The knob-toknob arrangement could be converted in silico into the nearly isoenergetic knob-into-hole conformation by a 26° rotation of the helices, which led the authors to propose that helical rotation is the signaling mechanism of HAMP domains (36). However, it is unclear whether the Af1503 HAMP domain should be used as a prototype for other HAMP-containing proteins because it contains no kinase or signaling domain and could therefore be a nonfunctional, evolutionary remnant of another pathway. If HAMP domains do signal by rotation, it is not obvious how different signals, such as the downward piston motion of TM2 in chemoreceptors or the lateral signal from the PAS domain of Aer, could contribute to HAMP rotation. To continue our exploration of signal transduction in Aer, we probed the structure of the Aer HAMP and proximal signaling domains by in vivo disulfide cross-linking and compared the results with structures from the same regions of Af1503 and Tar. Our data supported the proposed structure of a HAMP domain and suggested that the HAMP structure is conserved in functionally distinct proteins. In addition, lesions that promote the signal-on state of Aer clustered in one region of an in

Bacterial strains and plasmids. E. coli strains BT3312 (⌬aer-1 ⌬tsr-7028) (56), BT3400 (⌬aer-1 ⌬tsr-7028 recA::cat) (61), and BT3388 (⌬aer::erm ⌬tsr-7021 ⌬tartap-5201 trg::Tn10) (64) were used in this study. These strains are derivatives of RP437, an E. coli strain that is wild type for chemotaxis (53). The pDS7 plasmid expresses wild-type Aer from pACYC184 (17) using a tightly regulated sodium salicylate-inducible promoter (pnahG) (61). This plasmid also contains a p15A replication origin, allowing coexpression of genes with pTrc99A-derived plasmids. Plasmid pMB1 (44, 63) is a derivative of pGH1 that expresses cysteineless (C-less) Aer (Aer-C193S/C203A/C253A). Both pMB1 and pGH1 (wild-type Aer [55]) are derivatives of pTrc99A that express Aer under the control of an isopropyl-␤-D-thiogalactopyranoside (IPTG)-inducible ptrc promoter. Site-directed mutagenesis. Site-directed cysteine mutagenesis (Aer residues 210 to 290) was performed according to the instructions of a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) using pMB1 as the template. Aer expression was confirmed by Western blot analysis using antiAer2–166 antisera (56), and the expected mutations were confirmed by DNA sequencing. Mutant characterization. Aerotaxis phenotypes for each of the BT3312 Aer cysteine mutants were determined at 30°C in minimal soft agar (59) containing 30 mM succinate, 50 ␮g ml⫺1 ampicillin, and 0 to 1,000 ␮M IPTG. Nonaerotactic mutants were examined to determine their responses to oxygen in a gas perfusion chamber after induction with 200 ␮M or 1 mM IPTG as described previously (55, 59, 63). Aerotaxis phenotypes for the BT3388 Aer cysteine mutants were determined at 30°C in tryptone soft agar (59) containing 50 ␮g ml⫺1 ampicillin and 20 to 1,000 ␮M IPTG. To test for dominant or recessive behavior, BT3400 cells were cotransformed with pDS7 and the relevant pMB1-derived construct. Cotransformants were selected on LB agar containing 100 ␮g ml⫺1 ampicillin and 15 ␮g ml⫺1 tetracycline and then transferred to minimal soft agar containing 30 mM succinate, 50 ␮g ml⫺1 ampicillin, and 7.5 ␮g ml⫺1 tetracycline. Sodium salicylate (0.5 or 1 ␮M) and IPTG (between 0 and 0.6 mM) were also included in the semisoft agar in a series of titrations to vary gene expression levels. Actual expression levels were determined by Western blotting. The relevant constructs were also introduced individually into BT3400 and tested under the various conditions. In vivo cross-linking. BT3312 cells expressing individual Aer cysteine mutants were grown in H1 minimal salts medium supplemented with 30 mM succinate, 0.1% Casamino Acids, and 100 ␮g ml⫺1 ampicillin. Mid-log-phase cultures were induced for 3 h with 50 ␮M IPTG. Cross-linking was then performed at 25°C by exposing whole cells to 300 ␮M copper phenanthroline (CuPhe) for various time intervals (0, 1, 2, 5, 10, or 15 min), as described previously (5, 35, 44, 59). Western blots of cross-linked products were quantified using an Alpha Innotech digital imaging system. The percentage of cross-linking was calculated by dividing the intensity of the cross-linked dimer band by the sum of the intensities of the monomer and dimer bands and multiplying the result by 100. BT3312/pGH1 and BT3312/pMB1 were used as positive and negative cross-linking controls, respectively (44). Mid-log-phase cultures, induced with 50 ␮M IPTG, were also used to determine the steady-state cellular level of each mutant Aer protein. Cellular levels were inferred by measuring band intensities on Western blots and then comparing these intensities with the steady-state levels of C-less (pMB1) and wild-type (pGH1) Aer. In silico modeling. Homology models of wild-type and mutant Aer HAMP domains were created with the DeepView/Swiss-Model package (http://www .expasy.org/spdbv/, http://swissmodel.expasy.org/) using the coordinates of the A. fulgidus Af1503 HAMP domain and the sequence of the E. coli Aer HAMP domain. The stereochemical quality of the models was verified using WHAT IF (http://swift.cmbi.kun.nl/whatif/). PyMOL (http://pymol.sourceforge.net/) was used to view models, measure ␤-carbon distances, and “mutate” amino acid side chains. When side chains were mutated in PyMOL, all possible rotamers were sampled. ␤-Carbon distances were measured using a method analogous to that described by Careaga et al. (16). For glycines, which lack a ␤-carbon, side chains were mutated to cysteine and measured.

RESULTS Cysteine-scanning mutagenesis. To determine critical residues and study the secondary structure of the Aer HAMP and

2120

WATTS ET AL.

proximal signaling regions, Aer residues 210 to 290 were serially replaced with cysteine by mutating codons in pMB1, an otherwise C-less Aer expression construct. This region encompasses most of the HAMP domain (residues 210 to 253), the proximal signaling domain (residues 254 to 271), and the Nterminal portion of the signaling domain (residues 272 to 290) (Fig. 1). Multiple attempts to create a plasmid expressing AerT224C were unsuccessful. The 80 proteins containing single cysteine replacements were compared with wild-type Aer for the ability to support aerotaxis in BT3312, an E. coli strain that lacks the two aerotaxis receptors, Aer and Tsr. In succinate minimal soft agar without an IPTG inducer, cells expressing wild-type Aer from pGH1 exhibit three hallmark features: (i) a central dense zone of bacteria near the surface that is opaque to light, (ii) a dome-shaped, translucent zone surrounding the central dense zone, and (iii) an outer, well-defined ring formed by aerotactic bacteria at the bottom edge of the colony (Fig. 2A). When cells are not induced, the level of wild-type Aer expressed from pGH1 is approximately 10-fold higher than the chromosomal level or approximately 20% of the total number of cellular chemoreceptors. Of the 80 Aer cysteine mutants, 77 supported aerotaxis. In soft agar without IPTG, 57 of the mutants had swarm rates that were between 40 and 130% of the wild-type rate (Fig. 2B), and 55 of these mutants had colony morphologies similar to those of cells expressing wildtype Aer (e.g., V230C and N234C in Fig. 2A). An additional 18 mutants generated larger-than-normal “superswarming” colonies (e.g., V222C in Fig. 2A), which we have defined as colonies whose diameters are more than 1.3 times the diameter of BT3312/pGH1 colonies (43, 63) (Fig. 2B). Superswarming mutants generally had larger translucent zones than wild-type cells and were generated by cysteine replacements in all three HAMP subdomains and the proximal signaling and signaling domains (Fig. 2B). Cysteine replacements S261C, in the proximal signaling domain, and T288C, in the signaling domain, resulted in an impaired phenotype with partial aerotaxis. In soft agar, the corresponding colonies contained an outer ring of aerotactic bacteria, but the colonies were small, had no translucent zone, and were shaped like cylinders (e.g., S261C in Fig. 2A). These defects did not appear to be due to poor expression; when IPTG was added incrementally to increase the cellular levels of Aer-S261C and Aer-T288C, the colonies of both mutants became progressively smaller and lost the outer aerotaxis ring. All of the Aer HAMP mutants supported aerotaxis, indicating that the HAMP domain maintains a functionally native structure in the presence of introduced cysteines. However, three cysteine replacements in the proximal signaling domain, W255C, L256C, and Q263C, disrupted Aer-mediated aerotaxis. In succinate soft agar, Aer-L256C had a null aerotaxis phenotype, and the appearance and size of the colonies were similar to the appearance and size of the vector colonies (Fig. 2A and B). Aer-W255C and Aer-Q263C, on the other hand, generated larger, nonaerotactic colonies with fuzzy, undefined edges (Fig. 2A and B). When the expression of these three mutant proteins was induced incrementally with IPTG (at concentrations between 20 and 1,000 ␮M), none of them regained function. Steady-state protein levels. Since Aer HAMP mutants often produce unstable proteins (13, 43), the cellular steady-state

J. BACTERIOL.

FIG. 2. Influence of cysteine substitutions on Aer-mediated behavior. Plasmid-borne Aer mutants were tested in E. coli BT3312 (aer tsr). Colonies were transferred to succinate minimal soft agar containing 50 ␮g ml⫺1 ampicillin and then incubated at 30°C for 18 to 19 h and compared with positive (wild-type Aer, C-less Aer [Aer-C193S/C203A/ C253A]) and negative (pTrc99A vector) aerotaxis controls. (A) Representative Aer mutants with different colony phenotypes in soft agar, as follows: normal, Aer-V230C and Aer-N234C; large translucent zone, Aer-P211C (see text for details); superswarming, Aer-V222C; impaired function, Aer-S261C; and nonaerotactic, Aer-W255C, AerL256C, and Aer-Q263C. (B) Swarm rates for Aer mutants expressed as percentages of the wild-type Aer (pGH1) swarm rate. Residue numbers indicate the position of each cysteine substitution in the Aer protein. Mutants whose average colony expansion rate was more than 130% of the wild-type Aer rate were designated superswarming mutants, whereas mutants whose average colony expansion rate was less than 40% of the wild-type rate were either nonaerotactic (indicated by an asterisk) or had impaired function. C-less Aer and the vector control are indicated by a gray square and circle, respectively. WT, wildtype.

level of each cysteine mutant protein was measured after induction with 50 ␮M IPTG and compared with the level of C-less Aer. After induction with 50 ␮M IPTG, the steady-state level of C-less Aer was similar to the level of wild-type Aer (pGH1), which was approximately 50-fold higher than the chromosomal Aer expression level (and approximated the number of all other chemoreceptors in the cell). The steadystate levels of 4 of the 80 mutants were less than 60% of the C-less Aer level (Fig. 3A). These mutants were the P211C, R235C, G240C, and G277C mutants. Of these, the steady-state

VOL. 190, 2008

Aer HAMP AND PROXIMAL SIGNALING DOMAINS

2121

FIG. 3. Steady-state expression levels and in vivo cross-linking for Aer cysteine mutants. (A) Steady-state expression levels were measured in BT3312 (aer tsr) after induction with 50 ␮M IPTG and then expressed as percentages of the C-less Aer (pMB1) level. The levels of C-less Aer were 92.6% ⫾ 14.2% of the wild-type Aer (pGH1) levels (not shown), and mutants with steady-state levels less than 60% of the C-less Aer level (below the heavy dotted line) were considered unstable. An “X” indicates T224, where no cysteine mutant was made. (B) Percentage of dimer formation for each Aer cysteine mutant after incubation of cells with CuPhe for 10 min at 25°C. The data are from two or more independent cross-linking experiments. The secondary structure predicted by the PSA server (43) is shown beneath the graph (with helices indicated by cylinders and loops indicated by lines). (C) Helical wheel projections of the three proposed ␣-helices, HAMP AS-1 (top panel), HAMP AS-2/proximal signaling domain (middle panel), and the signaling domain (bottom panel), each modeled as 3.5 residues per turn. In Aer, W206 (underlined) is at the membrane-cytosol boundary and marks the end of TM2 (5). The positions of local cross-linking maxima (residues 210 to 290 from Fig. 3B and residues 206 to 209 from reference 5) are indicated by a black background. The unexpected maximum at position 211 rather than 212 (gray background) is discussed in the text. Prox., proximal; SD, signaling domain.

levels of Aer-P211C and Aer-R235C were the lowest, and both of them were 16% of the C-less Aer level. The instability of Aer-R235C has been recognized previously (13). Despite the low cellular levels of Aer-P211C and Aer-R235C, both mutant proteins supported aerotaxis in BT3312 when they were tested in soft agar without IPTG, although a large proportion of each colony consisted of the translucent zone (e.g., P211C in Fig. 2A). When the expression of either mutant was induced with 20 ␮M IPTG, the appearance in soft agar approached normal, possibly due to an increase in the number of mature Aer molecules per cell (not shown). With the exception of P211C and R235C, there was no apparent correlation between the different phenotypes observed in soft agar, such as superswarming or impaired function, and steady-state cellular levels.

Characterization of the nonaerotactic mutants. Although Aer-W255C, Aer-L256C, and Aer-Q263C did not support aerotaxis, the steady-state levels of these proteins were similar to that of C-less Aer (97.5, 115.5, and 106% of the C-less Aer level, respectively [Fig. 3A]). When expressed in BT3312 (with 200 ␮M IPTG) and examined using an aerotaxis temporal assay, Aer-W255C and Aer-Q263C mediated constant tumbling in air. When air was removed, both mutants showed a brief, weak inverted (smooth swimming) response. In addition, both mutants demonstrated phenotypic dominance when they were coexpressed with wild-type Aer. Aer-Q263C was phenotypically dominant over coexpressed wild-type Aer at all induction levels and resulted in a nonaerotactic phenotype with small, tight colonies in succinate soft agar. Aer-W255C, on the

2122

WATTS ET AL.

other hand, was partially dominant and impaired wild-type Aer-mediated colony size and ring formation only at high induction levels. As expected from its colony morphology in soft agar, AerL256C showed a true null aerotaxis phenotype in a temporal assay; cells exhibited random motility with no response to either an increase or decrease in oxygen at any level of induction up to 1 mM IPTG. Aer-L256C was also phenotypically recessive to wild-type Aer and did not inhibit the function of wild-type Aer when it was coexpressed at either low or high induction levels. Disulfide cross-linking. We used cysteine disulfide crosslinking (26) to analyze in vivo the secondary structure of the Aer HAMP and proximal signaling domains. In general, the rate and extent of disulfide cross-linking in response to an oxidant, such as CuPhe, reflect the proximity of introduced cysteines and/or the flexibility of the region being analyzed (5, 14, 35). If a protein forms a dimer, cognate cysteines that show the most extensive cross-linking are often located at a dimer interface (15). By comparing cross-linking rates for a large set of introduced cysteines, secondary structure can be inferred from characteristic periodicities or repeating patterns (14, 15, 41, 49). For example, a canonical ␣-helix has 3.6 residues per turn, a coiled-coil ␣-helix has 3.5 residues per turn, and a ␤-strand has 2 residues per repeat. The most probable secondary structure elements for Aer residues 210 to 290 were previously predicted using the PSA server (http://bmerc-www.bu .edu/psa/; 43) and are shown in the domain map in Fig. 3B. To confirm the predicted secondary structure, the single introduced cysteines spanning Aer residues 210 to 290 were cross-linked in intact cells after induction with 50 ␮M IPTG. The four less stable Aer mutants (P211C, R235C, G240C, and G277C) required 100 ␮M IPTG but were included in the analysis since the mature molecules made by these mutants were functional, indicating that their native conformation was not extensively disrupted. For these analyses, cells were treated with CuPhe at 25°C, after which cross-linked dimers were separated from non-cross-linked monomers by sodium dodecyl sulfate-polyacrylamide gel electrophoresis under nonreducing conditions and visualized by Western blotting. Under these conditions, the amounts of the dimer products increased linearly for 15 min for all the mutants tested except Aer-L256C, which cross-linked maximally by 5 min (not shown). Since the extent of in vivo cross-linking was low for most mutants after 5 min of exposure to CuPhe, comparisons were made at 10 min. Cross-linking in the HAMP domain. The Aer HAMP domain is predicted to consist of two ␣-helices, AS-1 and AS-2, that are separated by a connector with an undefined structure (43). Like AS-1 in other HAMP domains, AS-1 in Aer is predicted to be an amphipathic ␣-helix, but unlike AS-2 in other HAMP domains, AS-2 in Aer is predicted to be a buried ␣-helix with no obvious hydrophilic face (43). In this study, cross-linking in the HAMP AS-1 region (residues 207 to 224) was consistent with an ␣-helix (Fig. 3B) and could be modeled as either a 3.6-residue-per-turn ␣-helix (not shown) or a 3.5residue-per-turn ␣-helix (Fig. 3C, top panel). At the proximal end of AS-1, neither P211C (the apparent peak) nor I212C (the expected peak) cross-linked strongly (Fig. 3B and 3C). These results may have been influenced by the presence (in I212C) or absence (in P211C) of proline at residue 211, which

J. BACTERIOL.

typically introduces a bend into a helix (7). Such a bend might be required for the native conformation of some HAMP domains, since replacing the proline with cysteine in Aer (Fig. 3A) or in Tar (14) yields an unstable receptor. At the distal end of AS-1, there were similar rates of cross-linking at V222C (the expected peak) and at A223C (Fig. 3B and 3C), suggesting that the distal end of AS-1 is flexible. The region immediately following AS-1 (the connector, residues 225 to 234) could not be modeled as either an ␣-helix or a ␤-strand but was consistent with a structured loop (Fig. 3B). We previously defined the boundaries of AS-2 at residues 235 and 253 in Aer (62). However, in the recent solution structure of the Af1503 HAMP domain residues corresponding to Aer residues 235 to 237 are part of the connector, not AS-2 (36). In the present study, the cross-linking periodicity was consistent with an ␣-helix from residue 235 to residue 259, beyond AS-2 (Fig. 3B). This region was best modeled as a 3.5-residue-per-turn ␣-helix (Fig. 3C, middle panel). Moreover, in both AS-2 and AS-1, the positions with the greatest rates of cross-linking are occupied by hydrophobic amino acids in native Aer (with the exception of T242C) and correspond with the a and d positions of heptad repeating motifs (a-b-cd-e-f-g) (45) (Fig. 3C).This motif is commonly observed in coiled coils and four-helix bundles (20, 50). In AS-2, there are five additional hydrophobic positions that are not located at the dimer interface, perhaps explaining why AS-2 in Aer, but not in Tar, is predicted to be buried (14, 43). Cross-linking in the proximal signaling domain. As reported above, the heptad repeat pattern of AS-2 continued into the proximal signaling domain (residues 254 to 271 [43]) to residue 259 (Fig. 3B). This indicates that the N terminus of the proximal signaling domain forms a continuous helix with AS-2 (Fig. 3C, middle panel). Following residue 259, the periodicity of the helix was disrupted by consecutive cross-linking peaks that were four residues apart (Fig. 3B). This is consistent with the presence of a four- to eight-residue loop, similar to that predicted by the PSA server (43). However, precise boundaries could not be determined from the data because the rates of cross-linking at residues 263 and 264 were not significantly different (Fig. 3B). In addition, the structure of the Aer-Q263C receptor may be altered by the cysteine substitution at 263, since the mutant is a dominant, CW-biased mutant. This region of the proximal signaling domain was followed by a heptad repeat pattern that was phase shifted in comparison with AS-2. Like AS-2, this region of the signaling domain was best modeled as a 3.5-residue-per-turn ␣-helix (Fig. 3C, bottom panel); unlike AS-2, this helix contained predominately polar residues in the a and d positions. Model for HAMP structure. The HAMP domain structure that was recently solved for a putative transmembrane receptor (Af1503) from A. fulgidus is the first HAMP structure available (36). The Af1503 HAMP domain forms a dimeric, four-helix bundle with a parallel coiled-coil structure (36). To compare the AS-1/AS-1⬘ and AS-2/AS-2⬘ dimer interfaces of the Af1503 HAMP structure with the corresponding interfaces of Aer, we generated an Aer HAMP dimer model using DeepView/SwissModel and the coordinates of the A. fulgidus Af1503 HAMP domain (Fig. 4A). In the Aer HAMP model, residues predicted to be on the AS-1/AS-1⬘ and AS-2/AS-2⬘ interfaces were the same interface residues identified by cross-linking (with the

VOL. 190, 2008

Aer HAMP AND PROXIMAL SIGNALING DOMAINS

2123

FIG. 4. Distribution of residues with maximal disulfide cross-linking and Aer mutant phenotypes mapped onto an in silico model of the Aer HAMP domain. (A) Aer HAMP dimer model with helices indicated by blue (AS-1) or red (AS-2), connectors indicated by gray, and monomers distinguished by dark and light shading. (B and C) Residues with maximal cross-linking in AS-1 (B) and the AS-2/proximal signaling domain (C) mapped onto the in silico model (compare with Fig. 3). (D) Distribution of lesions that result in the following known phenotypes: CW bias phenotype (blue), null aerotaxis phenotype associated with unstable Aer protein (red), and null aerotaxis phenotype associated with comparatively more stable Aer protein (green). There are both signal-on (CW) and signal-off (null) mutants at residue W255. The Aer mutants analyzed are shown in Fig. 1.

exception of P211 and A223, as explained previously) (Fig. 4B and C). With few exceptions, residues in AS-1 or AS-2 that did not dimerize faced outward, away from the dimer interface of the model (not shown). ␤-Carbon distances for the AS-1/AS-1⬘ and AS-2/AS-2⬘ (and proximal signaling/proximal signaling⬘) residue pairs were calculated from the model and compared with the cross-linking data (Fig. 5). With few exceptions, the rate at which cysteine pairs in these helices cross-linked was inversely proportional to the predicted ␤-carbon distance. For example, L256C, which exhibited the greatest rate of cross-linking in the region encompassed by the model, had the shortest predicted ␤-carbon distance (4.41 Å). V260C also cross-linked strongly (approximately 70% of Aer-V260C formed dimer product in vivo after 15 min), but it is located outside the region of the in silico model and ␤-carbon distances could not be measured. In contrast, residues that formed less than 2% of cross-linked product in vivo were generally more than 14 Å apart in the model. Unlike AS-1 and AS-2, no correlation existed in the HAMP connector between predicted ␤-carbon distances and the crosslinking data. In fact, V230C, which showed the greatest rate of cross-linking in this region, had the greatest predicted distance between side chains (Fig. 5). The side chains of residues 230 and 230⬘ are predicted to face outward, on opposite sides of the HAMP domain model, suggesting that V230C might crosslink between adjacent dimers rather than within dimers. We tested this possibility using the double cysteine mutant AerV230C/V260C. V260C cross-links within dimers, so a dimer would be expected for Aer-V230C/V260C if V230C also crosslinks within a dimer. However, oligomers of Aer (larger than dimers) would be expected if V230C cross-links between dimers. In cross-linking experiments with temperatures up to 30°C for 15 min, Aer-V230C/V260C formed only dimers, suggesting that V230C cross-links exclusively within dimers. The possible explanations for these data include in vivo flexibility of the connector and differences between the structure and in silico model in this region. The modeled Aer HAMP connector is also one residue shorter than the connector in Af1503.

Distribution of CW-biased lesions in the folded HAMP domain. Null and CW-biased lesions were mapped previously onto a linear representation of the Aer HAMP and proximal signaling domains (13, 43, 62) (Fig. 1). With the creation of a

FIG. 5. Comparison between in vivo cross-linking (data from Fig. 3B) and predicted ␤-carbon distances for residues in the HAMP and proximal signaling domains. ␤-Carbon distances between cognate residues were determined from the Aer HAMP dimer model in PyMOL, as described in Materials and Methods.

2124

WATTS ET AL.

folded Aer HAMP model, it was possible to map these lesions onto a tertiary structure, an exercise that provided new insights into structure-function relationships in Aer. For example, mutations known to cause Aer to signal constantly in the absence of aerotactic stimuli (CW-biased mutants) are not contiguous in the linear sequence (Fig. 1) but are clustered together in the folded HAMP model at the distal end of the four-helix bundle (Fig. 4D). The finding that CW-biased HAMP lesions are localized suggests that this region plays a key role in signal transduction and is a prime candidate for interactions with the PAS domain. We previously observed that CW-biased Aer PAS mutants that are functionally rescued by chemoreceptors typically become superswarming mutants (63). However, in this study, not all superswarming mutants showed a CW bias in the absence of chemoreceptors. Only 6 of the 18 superswarming mutants in Fig. 2B (in BT3312; tsr aer) were nonaerotactic in a receptorless strain (BT3388; aer tsr tar trg tap), and just one of these, V222C, was CW biased (Fig. 1). Of note, V222 is located in the same region of the model occupied by lesions in the other CW-biased mutants (Fig. 4D). In silico mutagenesis and predicted effects on protein packing and stability. The HAMP domain is required for proper folding (34) and maturation (13) of the Aer protein, and HAMP null mutants (unlike CW-biased mutants) often produce an unstable Aer protein (13, 43). When these lesions (shown in Fig. 1 and in references 13, 43, and 62) were mapped onto the Aer HAMP model, they were dispersed throughout the structure, although the lesions that cause the most severe maturation defects were located in the HAMP domain (Fig. 4D) (13, 62), while those that cause less severe maturation defects were located in or near the proximal signaling domain (Fig. 4D). To gain insight into how these null mutants might affect the stability of Aer, we used PyMOL to “mutate” each of the 22 relevant side chains in silico and then reexamined the modeled structure. We found that 12 of the 22 substituted side chains were not permissible in any conformation because each rotamer caused steric hindrance and/or charge-charge repulsion (Fig. 1). To allow for readjustment of the backbone coordinates to accommodate side chain substitutions, we remodeled each of the 22 null mutants in DeepView/Swiss-Model using the coordinates of the Af1503 HAMP domain. When the mutants were remodeled, 10 of the side chain substitutions were accommodated (determined using WHAT IF), but the remaining 12 substitutions caused one or more defects, such as nonpermissible bond angles or bond lengths (Fig. 1). Ten of these nonpermitted substitutions overlapped with those determined by manually mutating side chains in PyMOL (Fig. 1). Seven of the eight substitutions that were permissible by either modeling method had fully surface-exposed side chains. Although these substitutions did not cause major structural deformities, they may disrupt stabilizing hydrogen bonds, salt bridges, or interactions with other parts of the Aer protein. DISCUSSION In this investigation we confirmed in vivo (Fig. 3B and C) the predicted secondary structure of the Aer HAMP and proximal signaling domains (42, 43). More importantly, the data were consistent with a four-helix bundle structure for the folded

J. BACTERIOL.

HAMP domain (Fig. 4). The predicted three-dimensional structure of the HAMP domain also revealed a clustering of CW lesions in Aer, defining a surface that is important for signaling between the PAS and HAMP domains (Fig. 4D). Although the PAS-HAMP interaction surface is yet to be defined, the proposed orientation of the PAS domain within an Aer dimer with respect to the HAMP and proximal signaling regions is shown in Fig. 6. In the proposed model, the PAS and HAMP domains are separated by a membrane module comprised of two transmembrane regions and a short periplasmic loop (4, 5). The PAS and HAMP domains are proposed to contact each other directly. This proposal is based on the results of second-site suppressor analysis (62) and on findings indicating that the PAS domain requires the HAMP domain for proper folding and FAD binding (34, 43). Interactions between the PAS and HAMP domains are also proposed to occur between cognate domains rather than between domains of the same monomer (61). Finally, the C terminus, which is critical for Aer function (61), is proposed to associate with the proximal signaling domain (1) (Fig. 6). Structure of the HAMP and proximal signaling domains. The A. fulgidus Af1503 protein is an atypical signaling protein that has no known role in sensory transduction. For this reason, it is important that the results of the present investigation support the hypothesis that there is a similar parallel four-helix fold in vivo in the HAMP domain of Aer, a protein with a well-documented role in aerotaxis. The evidence that there are similar HAMP structures in Af1503 and Aer includes the following results: (i) disulfide cross-linking independently identified the same residues at the dimer interface as the in silico Aer HAMP model that was generated using the coordinates of Af1503 (Fig. 4), (ii) the ␤-carbon distances measured on the HAMP model inversely correlated with the rate of cross-linking for residue pairs at the dimer interface (Fig. 5), and (iii) CW lesions were not contiguous in the linear Aer sequence (Fig. 1) but were clustered in one region of the folded HAMP model (Fig. 4D). Although this evidence is not definitive, it provides persuasive support for the hypothesis that there is a four-helix coiled-coil HAMP structure in Aer. While this paper was in preparation, evidence that the HAMP domain of the Tar chemoreceptor also forms a four-helix structure that is similar to the HAMP domain of Af1503 was published (57). Together, these findings increase our confidence that the solution structure of the Af1503 HAMP domain is a prototype for the structure of HAMP domains in general. The largest differences between the structure of the Aer HAMP domain and the structure of the Af1503 HAMP domain were in the connector that links AS-1 to AS-2. Both connectors are structured loops that begin with a conserved glycine (G225 in Aer) that is required for a U-turn toward AS-2 (Fig. 4A) (36). However, the Aer connector is one residue shorter, and the rates of cross-linking did not correlate with predicted ␤-carbon distances in the Aer HAMP model (Fig. 5, middle panel). The data for the proximal signaling domain (residues 254 to 271) were consistent with a predominantly coiled-coil structure formed by extending the flanking AS-2 and signaling domain helices (Fig. 3B and C). However, the helical structure of the proximal signaling domain was not continuous. There was a brief gap in the heptad repeat pattern after residue 259, and

VOL. 190, 2008

Aer HAMP AND PROXIMAL SIGNALING DOMAINS

2125

FIG. 6. Proposed model of an Aer dimer in cartoon format (left panel) and ribbon format (right panel). In the left panel, helices are represented by cylinders, loops are represented by lines, and PAS domains are represented by ellipses. In the right panel, the structures of the Aer PAS, HAMP, and signaling domains were modeled as described in Materials and Methods from the previously published coordinates of the NifL PAS (37), Af1503 HAMP (36), and MCP1143C signaling (51) domains, respectively. The F1(A. J. Campbell, K. J. Watts, and B. L. Taylor, unpublished data), membrane (4, 5), and proximal signaling domains (this study) were modeled from Aer cysteine cross-linking data. Also shown are the location of the glycine hinge (21) and the theoretical register between the proximal signaling domain and C-terminal tail of Aer (1).

although the structure of this break was not defined by crosslinking, it may be a short loop, as predicted by the PSA server (43). The heptad repeat pattern of the helix distal to this loop was also phase shifted in comparison with AS-2. A similar helical break at the end of the HAMP domain has been predicted for other chemoreceptors (42) and has been demonstrated for Tar (22). However, this helical break occurs at the end of the HAMP domain in chemoreceptors such as Tar (corresponding to residues 253 to 257 of Aer), whereas in Aer it is located within the proximal signaling domain after residue 259. The region scanned by cysteine substitutions in this study is a known locus (Fig. 1) for null lesions that lead to unstable Aer proteins (13, 43). However, in this study most cysteine substitutions caused only minor changes in structure or function (Fig. 2 and 3A). Only four of the mutants, P211C, R235C, G240C, and G277C, had low steady-state protein levels (Fig. 3A), suggesting that a cysteine substitution at most positions did not affect protein maturation or stability. All four cysteine mutants that had low Aer protein levels supported aerotaxis, indicating that in these Aer mutants a sufficient fraction of molecules folded into a functional conformation. Although none of the cysteine replacements in the HAMP domain eliminated aerotaxis, three replacements in the proximal signaling domain did disrupt aerotaxis (W255C, L256C, and Q363C) (Fig. 2) but did not appear to affect protein stability (Fig. 3A). Notably, cysteine replacements at equivalent positions in the

aspartate receptor (Tar-D263 [Aer-W255] and Tar-T264 [AerL256]) also eliminate receptor function (22). In silico mutagenesis of the HAMP domain was used to simulate null lesions that destabilize the Aer protein. More than one-half of the lesions that we modeled disrupted the four-helix bundle of the modeled HAMP domain. These in silico findings raise the following interesting possibilities: (i) disruptions in the four-helix HAMP bundle may destabilize the entire Aer protein, (ii) a four-helix bundle might be critical for folding of Aer in the maturation sequence, or (iii) the fourhelix bundle may shield residues sensitive to proteolysis. Signaling mechanisms. Aer and the E. coli chemoreceptors have a common signal output from a highly conserved signaling domain, but the initial signal transduction events differ. It is likely that different input signals are converted into a common signal at the HAMP domain, and the findings of this study suggest that the site of convergence for Aer is at the distal end of the HAMP domain, where a cluster of CW-biased (gain-offunction) lesions augments PAS-HAMP interactions (Fig. 4D). It has been proposed that the Af1503 four-helix HAMP bundle switches between the on and off signaling states by rotating neighboring helices between two nearly isoenergetic packing geometries (36). The interdigitating side chains of the bundle require that the neighboring helices rotate in opposite directions in the manner of a gear box with four cogwheels (36). This would result in less than a 3-Å change between most side chain distances in the HAMP domain, differences that would

2126

WATTS ET AL.

not be easily resolved by disulfide cross-linking. Since the Aer HAMP domain appears to have a structure similar to that of the Af1503 HAMP domain (Fig. 4 and 5), the four helices of the Aer HAMP domain might also rotate between kinase-on and kinase-off conformations. Aer is normally in the signal-off (counterclockwise) state until PAS-HAMP interactions, parallel to the plane of the membrane, promote the signal-on (CW) state (13, 61, 62). This is different from the chemoreceptors, which signal to the HAMP domain by a piston-type movement across the membrane. The general mechanism by which a HAMP domain transforms different input signals into a common signal output remains to be determined. The input signal to the HAMP domain is initiated by a redox change in the isoalloxazine ring of FAD bound to the PAS domain (24, 56, 58). How the resulting conformational change is propagated through the PAS domain to the HAMP domain remains to be determined. Known signal transduction mechanisms in other PAS domains include movement of the FG loop (in the direct oxygen sensor [40], PAS kinase [3], and FixL [28, 29, 31, 47]) and displacement of a PAS N-terminal helix (in Vivid [65] and the photoactive yellow protein [32, 54]) or C-terminal helix (in phototropin LOV2 [33]). Although there is no direct evidence for either mechanism in Aer, displacement of the PAS N-terminal helix (the N-cap) does appear to play a role in signaling, since removing the N-cap results in a conformation that mimics the signal-on state of Aer (63). This may be analogous to what has been observed for the resolved structure of the Avena sativa LOV2 domain, where the structure of the PAS domain in the light-induced (signal-on) state is similar to the structure obtained when the C-terminal helix has been removed (33). During signal transduction, the HAMP domain may rotate, but it is unlikely that the output from the signaling domain is also rotation. Fluorescence polarization measurements of yellow fluorescent protein tethered to the signaling domain of the Tsr chemoreceptor were consistent with lateral displacement between trimers of dimers but not rotation of the signaling domain (60). The signaling domain is proposed to flex at a glycine hinge (Gly330 and Gly331 in Aer) (Fig. 6) within the flexible-bundle subdomain (21). Aer apparently uses a similar signaling mechanism because the signaling domains of chemoreceptors are interchangeable with the Aer signaling domain (8, 56), and Aer can signal in mixed trimers of dimers with other chemoreceptors (30). If signal propagation involves rotation of the HAMP domain but flexing of the signaling domain, the “loop” at the end of the HAMP domain (in chemoreceptors) or in the proximal signaling domain (in Aer) may have a role in torque conversion. A mechanical (Lego) model constructed to simulate a four-helix coiled-coil domain is able to convert gear box rotation of the helices into bending of an attached helix, if a flexible hinge is attached to the distal end (R. Alexander and I. Zhulin, personal communication). If a short loop in the proximal signaling domain permits bending, a hinged proximal signaling domain could perform the torque conversion. The concept of bending is conjecture at this time, but torque conversion of a HAMP rotation into flexing of the signaling domain is a fruitful area for further investigation to determine whether torque conversion is the role of conserved proximal signaling domains in these sensory systems.

J. BACTERIOL. ACKNOWLEDGMENTS We thank Divya Amin for constructing the single-cysteine mutants for Aer residues 210 to 218 and Nathan Abraham for technical assistance. We also thank Andrei Lupas for helpful discussions concerning protein modeling. This work was supported by grant GM29481 from the National Institute of General Medical Sciences to B. L. Taylor. REFERENCES 1. Alexander, R. P., and I. B. Zhulin. 2007. Evolutionary genomics reveals conserved structural determinants of signaling and adaptation in microbial chemoreceptors. Proc. Natl. Acad. Sci. USA 104:2885–2890. 2. Ames, P., C. A. Studdert, R. H. Reiser, and J. S. Parkinson. 2002. Collaborative signaling by mixed chemoreceptor teams in Escherichia coli. Proc. Natl. Acad. Sci. USA 99:7060–7065. 3. Amezcua, C., S. Harper, J. Rutter, and K. Gardner. 2002. Structure and interactions of PAS kinase N-terminal PAS domain. Model for intramolecular kinase regulation. Structure 10:1349. 4. Amin, D. N., B. L. Taylor, and M. S. Johnson. 2007. Organization of the aerotaxis receptor Aer in the membrane of Escherichia coli. J. Bacteriol. 189:7206–7212. 5. Amin, D. N., B. L. Taylor, and M. S. Johnson. 2006. Topology and boundaries of the aerotaxis receptor Aer in the membrane of Escherichia coli. J. Bacteriol. 188:894–901. 6. Aravind, L., and C. P. Ponting. 1999. The cytoplasmic helical linker domain of receptor histidine kinase and methyl-accepting proteins is common to many prokaryotic signalling proteins. FEMS Microbiol. Lett. 176:111–116. 7. Barlow, D. J., and J. M. Thornton. 1988. Helix geometry in proteins. J. Mol. Biol. 201:601–619. 8. Bibikov, S. I., L. A. Barnes, Y. Gitin, and J. S. Parkinson. 2000. Domain organization and flavin adenine dinucleotide-binding determinants in the aerotaxis signal transducer Aer of Escherichia coli. Proc. Natl. Acad. Sci. USA 97:5830–5835. 9. Bibikov, S. I., R. Biran, K. E. Rudd, and J. S. Parkinson. 1997. A signal transducer for aerotaxis in Escherichia coli. J. Bacteriol. 179:4075–4079. 10. Bordignon, E., J. P. Klare, M. Doebber, A. A. Wegener, S. Martell, M. Engelhard, and H. J. Steinhoff. 2005. Structural analysis of a HAMP domain: the linker region of the phototransducer in complex with sensory rhodopsin II. J. Biol. Chem. 280:38767–38775. 11. Borkovich, K. A., N. Kaplan, J. F. Hess, and M. I. Simon. 1989. Transmembrane signal transduction in bacterial chemotaxis involves ligand-dependent activation of phosphate group transfer. Proc. Natl. Acad. Sci. USA 86:1208– 1212. 12. Bray, D., M. D. Levin, and C. J. Morton-Firth. 1998. Receptor clustering as a cellular mechanism to control sensitivity. Nature 393:85–88. 13. Buron-Barral, M. D. C., K. K. Gosink, and J. S. Parkinson. 2006. Loss- and gain-of-function mutations in the F1-HAMP region of the Escherichia coli aerotaxis transducer Aer. J. Bacteriol. 188:3477–3486. 14. Butler, S. L., and J. J. Falke. 1998. Cysteine and disulfide scanning reveals two amphiphilic helices in the linker region of the aspartate chemoreceptor. Biochemistry 37:10746–10756. 15. Careaga, C. L., and J. J. Falke. 1992. Structure and dynamics of Escherichia coli chemosensory receptors. Engineered sulfhydryl studies. Biophys. J. 62: 209–216. (Discussion, 62:217–219.) 16. Careaga, C. L., J. Sutherland, J. Sabeti, and J. J. Falke. 1995. Large amplitude twisting motions of an interdomain hinge: a disulfide trapping study of the galactose-glucose binding protein. Biochemistry 34:3048–3055. 17. Chang, A. C., and S. N. Cohen. 1978. Construction and characterization of amplifiable multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid. J. Bacteriol. 134:1141–1156. 18. Chervitz, S. A., and J. J. Falke. 1995. Lock on/off disulfides identify the transmembrane signaling helix of the aspartate receptor. J. Biol. Chem. 270:24043–24053. 19. Chervitz, S. A., and J. J. Falke. 1996. Molecular mechanism of transmembrane signaling by the aspartate receptor: a model. Proc. Natl. Acad. Sci. USA 93:2545–2550. 20. Cohen, C., and D. A. Parry. 1990. Alpha-helical coiled coils and bundles: how to design an alpha-helical protein. Proteins 7:1–15. 21. Coleman, M. D., R. B. Bass, R. S. Mehan, and J. J. Falke. 2005. Conserved glycine residues in the cytoplasmic domain of the aspartate receptor play essential roles in kinase coupling and on-off switching. Biochemistry 44: 7687–7695. 22. Danielson, M. A., R. B. Bass, and J. J. Falke. 1997. Cysteine and disulfide scanning reveals a regulatory alpha-helix in the cytoplasmic domain of the aspartate receptor. J. Biol. Chem. 272:32878–32888. 23. Duke, T. A., and D. Bray. 1999. Heightened sensitivity of a lattice of membrane receptors. Proc. Natl. Acad. Sci. USA 96:10104–10108. 24. Edwards, J. C., M. S. Johnson, and B. L. Taylor. 2006. Differentiation between electron transport sensing and proton motive force sensing by the Aer and Tsr receptors for aerotaxis. Mol. Microbiol. 62:823–837.

VOL. 190, 2008

Aer HAMP AND PROXIMAL SIGNALING DOMAINS

2127

25. Falke, J. J., and G. L. Hazelbauer. 2001. Transmembrane signaling in bacterial chemoreceptors. Trends Biochem. Sci. 26:257–265. 26. Falke, J. J., and D. E. Koshland, Jr. 1987. Global flexibility in a sensory receptor: a site-directed cross-linking approach. Science 237:1596–1600. 27. Gegner, J. A., D. R. Graham, A. F. Roth, and F. W. Dahlquist. 1992. Assembly of an MCP receptor, CheW, and kinase CheA complex in the bacterial chemotaxis signal transduction pathway. Cell 70:975–982. 28. Gong, W., B. Hao, and M. K. Chan. 2000. New mechanistic insights from structural studies of the oxygen-sensing domain of Bradyrhizobium japonicum FixL. Biochemistry 39:3955–3962. 29. Gong, W., B. Hao, S. S. Mansy, G. Gonzalez, M. A. Gilles-Gonzalez, and M. K. Chan. 1998. Structure of a biological oxygen sensor: a new mechanism for heme-driven signal transduction. Proc. Natl. Acad. Sci. USA 95:15177– 15182. 30. Gosink, K. K., M. del Carmen Buron-Barral, and J. S. Parkinson. 2006. Signaling interactions between the aerotaxis transducer Aer and heterologous chemoreceptors in Escherichia coli. J. Bacteriol. 188:3487–3493. 31. Hao, B., C. Isaza, J. Arndt, M. Soltis, and M. K. Chan. 2002. Structure-based mechanism of O2 sensing and ligand discrimination by the FixL heme domain of Bradyrhizobium japonicum. Biochemistry 41:12952–12958. 32. Harigai, M., Y. Imamoto, H. Kamikubo, Y. Yamazaki, and M. Kataoka. 2003. Role of an N-terminal loop in the secondary structural change of photoactive yellow protein. Biochemistry 42:13893–13900. 33. Harper, S. M., L. C. Neil, and K. H. Gardner. 2003. Structural basis of a phototropin light switch. Science 301:1541–1544. 34. Herrmann, S., Q. Ma, M. S. Johnson, A. V. Repik, and B. L. Taylor. 2004. PAS domain of the Aer redox sensor requires C-terminal residues for nativefold formation and flavin adenine dinucleotide binding. J. Bacteriol. 186: 6782–6791. 35. Hughson, A. G., and G. L. Hazelbauer. 1996. Detecting the conformational change of transmembrane signaling in a bacterial chemoreceptor by measuring effects on disulfide cross-linking in vivo. Proc. Natl. Acad. Sci. USA 93:11546–11551. 36. Hulko, M., F. Berndt, M. Gruber, J. U. Linder, V. Truffault, A. Schultz, J. Martin, J. E. Schultz, A. N. Lupas, and M. Coles. 2006. The HAMP domain structure implies helix rotation in transmembrane signaling. Cell 126:929– 940. 37. Key, J., M. Hefti, E. B. Purcell, and K. Moffat. 2007. Structure of the redox sensor domain of Azotobacter vinelandii NifL at atomic resolution: signaling, dimerization, and mechanism. Biochemistry 46:3614–3623. 38. Kim, K. K., H. Yokota, and S. H. Kim. 1999. Four-helical-bundle structure of the cytoplasmic domain of a serine chemotaxis receptor. Nature 400:787– 792. 39. Kim, S. H., W. Wang, and K. K. Kim. 2002. Dynamic and clustering model of bacterial chemotaxis receptors: structural basis for signaling and high sensitivity. Proc. Natl. Acad. Sci. USA 99:11611–11615. 40. Kurokawa, H., D. S. Lee, M. Watanabe, I. Sagami, B. Mikami, C. S. Raman, and T. Shimizu. 2004. A redox-controlled molecular switch revealed by the crystal structure of a bacterial heme PAS sensor. J. Biol. Chem. 279:20186– 20193. 41. Lee, G. F., G. G. Burrows, M. R. Lebert, D. P. Dutton, and G. L. Hazelbauer. 1994. Deducing the organization of a transmembrane domain by disulfide cross-linking. The bacterial chemoreceptor Trg. J. Biol. Chem. 269:29920– 29927. 42. Le Moual, H., and D. E. Koshland, Jr. 1996. Molecular evolution of the C-terminal cytoplasmic domain of a superfamily of bacterial receptors involved in taxis. J. Mol. Biol. 261:568–585. 43. Ma, Q., M. S. Johnson, and B. L. Taylor. 2005. Genetic analysis of the HAMP domain of the Aer aerotaxis sensor localizes flavin adenine dinucleotide-binding determinants to the AS-2 helix. J. Bacteriol. 187:193–201. 44. Ma, Q., F. Roy, S. Herrmann, B. L. Taylor, and M. S. Johnson. 2004. The Aer protein of Escherichia coli forms a homodimer independent of the signaling domain and flavin adenine dinucleotide binding. J. Bacteriol. 186: 7456–7459.

45. McLachlan, A. D., and M. Stewart. 1975. Tropomyosin coiled-coil interactions: evidence for an unstaggered structure. J. Mol. Biol. 98:293–304. 46. McNally, D. F., and P. Matsumura. 1991. Bacterial chemotaxis signaling complexes: formation of a CheA/CheW complex enhances autophosphorylation and affinity for CheY. Proc. Natl. Acad. Sci. USA 88:6269–6273. 47. Miyatake, H., M. Mukai, S. Y. Park, S. Adachi, K. Tamura, H. Nakamura, K. Nakamura, T. Tsuchiya, T. Iizuka, and Y. Shiro. 2000. Sensory mechanism of oxygen sensor FixL from Rhizobium meliloti: crystallographic, mutagenesis and resonance Raman spectroscopic studies. J. Mol. Biol. 301:415– 431. 48. Nambu, J. R., J. O. Lewis, K. A. Wharton, Jr., and S. T. Crews. 1991. The Drosophila single-minded gene encodes a helix-loop-helix protein that acts as a master regulator of CNS midline development. Cell 67:1157–1167. 49. Pakula, A. A., and M. I. Simon. 1992. Determination of transmembrane protein structure by disulfide cross-linking: the Escherichia coli Tar receptor. Proc. Natl. Acad. Sci. USA 89:4144–4148. 50. Paliakasis, C. D., and M. Kokkinidis. 1992. Relationships between sequence and structure for the four-alpha-helix bundle tertiary motif in proteins. Protein Eng. 5:739–748. 51. Park, S. Y., P. P. Borbat, G. Gonzalez-Bonet, J. Bhatnagar, A. M. Pollard, J. H. Freed, A. M. Bilwes, and B. R. Crane. 2006. Reconstruction of the chemotaxis receptor-kinase assembly. Nat. Struct. Mol. Biol. 13:400–407. 52. Parkinson, J. S., P. Ames, and C. A. Studdert. 2005. Collaborative signaling by bacterial chemoreceptors. Curr. Opin. Microbiol. 8:116–121. 53. Parkinson, J. S., and S. E. Houts. 1982. Isolation and behavior of Escherichia coli deletion mutants lacking chemotaxis functions. J. Bacteriol. 151:106–113. 54. Pellequer, J. L., K. A. Wager-Smith, S. A. Kay, and E. D. Getzoff. 1998. Photoactive yellow protein: a structural prototype for the three-dimensional fold of the PAS domain superfamily. Proc. Natl. Acad. Sci. USA 95:5884– 5890. 55. Rebbapragada, A., M. S. Johnson, G. P. Harding, A. J. Zuccarelli, H. M. Fletcher, I. B. Zhulin, and B. L. Taylor. 1997. The Aer protein and the serine chemoreceptor Tsr independently sense intracellular energy levels and transduce oxygen, redox, and energy signals for Escherichia coli behavior. Proc. Natl. Acad. Sci. USA 94:10541–10546. 56. Repik, A., A. Rebbapragada, M. S. Johnson, J. O. Haznedar, I. B. Zhulin, and B. L. Taylor. 2000. PAS domain residues involved in signal transduction by the Aer redox sensor of Escherichia coli. Mol. Microbiol. 36:806–816. 57. Swain, K. E., and J. J. Falke. 2007. Structure of the conserved HAMP domain in an intact, membrane-bound chemoreceptor: a disulfide mapping study. Biochemistry 46:13684–13695. 58. Taylor, B. L. 2007. Aer on the inside looking out: paradigm for a PASHAMP role in sensing oxygen, redox and energy. Mol. Microbiol. 65:1415– 1424. 59. Taylor, B. L., K. J. Watts, and M. S. Johnson. 2007. Oxygen and redox sensing by two-component systems that regulate behavioral responses: behavioral assays and structural studies of aer using in vivo disulfide crosslinking. Methods Enzymol. 422:190–232. 60. Vaknin, A., and H. C. Berg. 2007. Physical responses of bacterial chemoreceptors. J. Mol. Biol. 366:1416–1423. 61. Watts, K. J., M. S. Johnson, and B. L. Taylor. 2006. Minimal requirements for oxygen sensing by the aerotaxis receptor Aer. Mol. Microbiol. 59:1317– 1326. 62. Watts, K. J., Q. Ma, M. S. Johnson, and B. L. Taylor. 2004. Interactions between the PAS and HAMP domains of the Escherichia coli aerotaxis receptor Aer. J. Bacteriol. 186:7440–7449. 63. Watts, K. J., K. Sommer, S. L. Fry, M. S. Johnson, and B. L. Taylor. 2006. Function of the N-terminal cap of the PAS domain in signaling by the aerotaxis receptor Aer. J. Bacteriol. 188:2154–2162. 64. Yu, H. S., J. H. Saw, S. Hou, R. W. Larsen, K. J. Watts, M. S. Johnson, M. A. Zimmer, G. W. Ordal, B. L. Taylor, and M. Alam. 2002. Aerotactic responses in bacteria to photoreleased oxygen. FEMS Microbiol. Lett. 217:237–242. 65. Zoltowski, B. D., C. Schwerdtfeger, J. Widom, J. J. Loros, A. M. Bilwes, J. C. Dunlap, and B. R. Crane. 2007. Conformational switching in the fungal light sensor Vivid. Science 316:1054–1057.