Leptospira interrogans Aer2: an Unusual Membrane-Bound PAS-Heme Oxygen Sensor

ABSTRACT

In this study, we provide the first characterization of a chemoreceptor from Leptospira interrogans, the cause of leptospirosis. This receptor is related to the Aer2 receptors that have been studied in other bacteria. In those organisms, Aer2 is a soluble receptor with one or two PAS-heme domains and signals in response to O₂ binding. In contrast, L. interrogans Aer2 (LiAer2) is an unusual membrane-bound Aer2 with a periplasmic domain and three cytoplasmic PAS-heme domains. Each of the three PAS domains bound b-type heme via conserved Eη-His residues. They also bound O₂ and CO with similar affinities to each other and other PAS-heme domains. However, all three PAS domains were uniquely hexacoordinate in the deoxy-heme state, whereas other Aer2-PAS domains are pentacoordinate. Similar to other Aer2 receptors, LiAer2 could hijack the E. coli chemotaxis pathway but only when it was expressed with an E. coli high-abundance chemoreceptor. Unexpectedly, the response was inverted relative to classic Aer2 receptors. That is, LiAer2 caused E. coli to tumble (it was signal-on) in the absence of O₂ and to stop tumbling in its presence. Thus, an endogenous ligand in the deoxy-heme state was correlated with signal-on LiAer2, and its displacement for gas-binding turned signaling off. This response also occurred in a soluble version of LiAer2 lacking the periplasmic domain, transmembrane (TM) region, and first two PAS domains, meaning that PAS3 alone was sufficient for O₂-mediated control. Future studies are needed to understand the unique signaling mechanisms of this unusual Aer2 receptor.

IMPORTANCE Leptospira interrogans, the cause of the zoonotic infection leptospirosis, is found in soil and water contaminated with animal urine. L. interrogans survives in complex environments with the aid of 12 chemoreceptors, none of which has been explicitly studied. In this study, we characterized the first L. interrogans chemoreceptor, LiAer2, and reported its unique characteristics. LiAer2 is membrane-bound, has three cytoplasmic PAS-heme domains that each bound hexacoordinate b-type heme and O₂ turned LiAer2 signaling off. An endogenous ligand in the deoxy-heme state was correlated with signal-on LiAer2 and its displacement for O₂-binding turned signaling off. Our study corroborated previous findings that Aer2 receptors are O₂ sensors, but also demonstrated that they do not all function the same way.

INTRODUCTION

Leptospira interrogans, the agent of leptospirosis, is a motile spirochete that is transmitted through direct contact with soil or water contaminated with animal urine. To survive in these complex environments, L. interrogans relies on 12 to 14 chemoreceptors and multiple copies of chemosensory proteins that are largely arranged into three gene clusters (MiST 3.0; https://mistdb.com [1]). Currently, there is little information on how L. interrogans utilizes its chemosensory proteins to sense, signal, and control cellular processes. Chemoattractants have been identified for both pathogenic and saprophytic Leptospira, and the structure of a chemoreceptor CACHE domain was recently solved (2–4). However, the ligand specificity of each chemoreceptor remains unknown.

The second chemosensory operon in L. interrogans (che2) encodes a chemoreceptor, two histidine kinases (HK and CheA2), a coupling protein (CheW3), two response regulators (RR and CheY2), two adaptation proteins (CheD1 and CheB3), and an anti-sigma factor antagonist (Fig. 1A). This operon is related to the operons that encode Aer2 receptors in Proteobacteria (5, 6), but the first two genes, encoding the HK and RR, are not usually associated with Aer2-containing systems. Transposon-mediated disruption of the HK gene results in cells with a reduced reversal frequency (7), although Aer2-containing systems are not directly responsible for flagella-based motility in other bacteria (8–10). In Pseudomonas aeruginosa, the Aer2-containing system crosstalks with the chemotaxis system through the adaptation protein CheD, which augments the methylation of Aer2 and several chemotaxis receptors (9). Given the homology between the che2 operon in L. interrogans and the Aer2-containing operons of other bacteria, we named the che2 chemoreceptor in L. interrogans LiAer2 (Fig. 1A).

FIG 1.

The L. interrogans che2 operon and LiAer2 structural models. (A) Organization of the che2 operon from L. interrogans serovar Pomona strain Pomona. Che2 encodes a chemoreceptor (Aer2), two histidine kinases (HK and CheA2), a coupling protein (CheW3), two response regulators (RR and CheY2), two adaptation proteins (CheD1 and CheB3), and an anti-sigma factor antagonist (σ*). (B) The structure of a LiAer2 dimer as predicted by AlphaFold. LiAer2 contains a potential periplasmic ligand-binding domain (LBD), followed by a transmembrane (TM) region, three cytoplasmic PAS domains, a di-HAMP unit, and a kinase control module with 34 heptad repeats (34H). (C) LiAer2 PAS3 dimer model based on the unliganded dimer structure of P. aeruginosa Aer2 PAS (PaPAS) with heme (PDB 4HI4) (16). PaPAS residues involved in heme- and O₂-binding (Eη-His and Iβ-Trp) are shown as sticks. (D) Proposed LiAer2 interactions and phosphotransfer reactions based on known interactions and reactions in the P. aeruginosa Che2 system (9) (see text for details). By analogy to the methylation sites of E. coli Tsr and Tar, LiAer2 has three predicted methylation sites (residues 729, 736, and 904), and one putative methylation site (residue 911, one heptad C-terminal to residue 904), all of which are surrounded by predicted CheD binding sites.

LiAer2 is an unusual Aer2 receptor. It is large (976 aa) and is predicted to contain a periplasmic ligand-binding domain (LBD), followed by a transmembrane (TM) region, three cytoplasmic PAS (Per-ARNT-Sim) domains, two HAMP domains, and a kinase control module with 34 heptad (34H) repeats (Fig. 1B). Aer2-type 34H receptors are present in both pathogenic and saprophytic Leptospires (Table S1). The Aer2 receptors from Proteobacteria contain kinase control domains with 36 heptad repeats and belong to the F7 “chemotaxis class” (11, 12). The Che2 system in L. interrogans belongs to the F8 class, although F8 phylogenetically groups with F7, and both systems are thought to share a common ancestor (13). The kinase control domain of LiAer2 contains four predicted methylation sites but lacks the C-terminal pentapeptide that is involved in adaptation enzyme binding in other Aer2 receptors (Fig. 1).

The Aer2 receptors from P. aeruginosa and Vibrio cholerae are well characterized. They are soluble O₂ sensors that localize to the cytoplasmic membrane near the cell pole where they form chemosensory complexes distinct from chemotaxis arrays (8, 12, 14). P. aeruginosa Aer2 (PaAer2) contains one PAS domain, whereas V. cholerae Aer2 (VcAer2) contains two PAS domains (5, 6). PAS domains are common globular sensing domains comprised of an antiparallel β-sheet (with strands Aβ, Bβ, Gβ, Hβ, and Iβ) and several flanking α-helices (Cα, Dα, Eα, and Fα) (15). Aer2 PAS domains differ slightly in that they contain an extended Cα/Dα helix and a short Eη helix in place of Eα (Fig. 1C). Aer2 PAS domains bind pentacoordinate b-type heme via a conserved His residue on the PAS-Eη helix and respond to O₂-binding by displacing an Hβ-Leu residue from the ligand-binding site while rotating an Iβ-Trp residue to bond with the ligand and initiate signaling ((6, 16–18); modeled in Fig. 1C).

In P. aeruginosa, Aer2 signaling in response to O₂ increases the autophosphorylation rate of the bound histidine kinase CheA2, which is subsequently dephosphorylated by the response regulator CheY2. Phosphotransfer from CheA2 to CheB2 enhances the phosphatase activity of CheB2, and Aer2 is demethylated to terminate signaling. In the absence of O₂, Aer2 does not signal and the adaptation protein CheD is freed from CheY2 to augment CheR2-mediated methylation of Aer2, thus enhancing the probability of signaling (9). The che2 operon in L. interrogans does not encode a methyltransferase, although two cheR genes are present elsewhere in the genome. Proposed Che2 protein interactions and phosphotransfer reactions in L. interrogans, based on those of P. aeruginosa Che2, are shown in Fig. 1D.

The exact role of Aer2-containing chemosensory systems remains elusive, but they are involved in stress responses and virulence (19, 20). However, Aer2-containing systems are also ancestral to the E. coli chemotaxis system and Aer2 receptors can hijack the E. coli chemotaxis system (E. coli CheA, W, and Y) to cause cell tumbling in the presence of O₂ (5, 6, 12). In E. coli, O₂-bound Aer2 promotes the formation of phospho-CheY, which binds to the flagellar motor and changes the direction of flagellar rotation to induce tumbling. This assay serves as a useful tool for assessing Aer2 function in vivo.

In this study, we analyzed LiAer2, the first L. interrogans receptor to be characterized. We show that LiAer2 is an unusual membrane-bound Aer2 receptor with three PAS-heme domains that each bind hexacoordinate b-type heme. LiAer2 responded to O₂ binding but signaling was inverted relative to other Aer2 receptors. For LiAer2, O₂ turned LiAer2 signaling off. This study corroborates previous findings that Aer2 receptors are O₂ sensors via their PAS-heme domains and demonstrates that they do not all function the same way. Potential signaling mechanisms are discussed.

RESULTS AND DISCUSSION

LiAer2 is an unusual, membrane-bound Aer2 receptor.

LiAer2 is an unusual Aer2 receptor because it contains two predicted membrane-spanning segments (TM1 and TM2, 19 and 20 amino acids, respectively, Fig. 1B). Most analyzed Aer2 receptors lack TM regions, are soluble, and instead associate with the cytoplasmic membrane near the cell pole (5, 6, 8, 14). We cloned the aer2 coding region from L. interrogans serovar Pomona and expressed it with an N-terminal His tag in chemoreceptorless E. coli BT3388 (tar, tsr, trg, tap, aer). Full-length LiAer2_1-976 (111.6 kDa) partitioned primarily into the insoluble fraction (the high-speed pellet) after lysed cells were centrifuged at low (10,000 × g) and then at high (407,000 × g) speed (Fig. S1A), supporting a membrane-embedded location.

LiAer2 hijacks the E. coli chemotaxis system to mediate O₂ and CO responses.

The Aer2 receptors from P. aeruginosa and V. cholerae both hijack the E. coli chemotaxis system in BT3388 and induce cell tumbling in the presence of O₂ (5, 6). To determine if LiAer2 can elicit a similar response, LiAer2_1-976 was expressed in BT3388 under the conditions previously used for PaAer2 (200 μM IPTG induction for 45 min) or VcAer2 (200 μM IPTG induction for 2 h) in the presence of 25 μg/mL 5-aminolevulinic acid (ALA). ALA enhances heme synthesis and incorporation. In BT3388, LiAer2_1-976 did not induce cell tumbling in air or N₂ nor did it induce chemotaxis ring formation in soft agar (Fig. S2A). In E. coli, native low-abundance chemoreceptors that lack a C-terminal pentapeptide for CheR/B binding require the assistance of a high-abundance receptor with a C-terminal pentapeptide (Tar or Tsr) for both their behavior and ability to adapt (21, 22). Unusual for an Aer2 receptor, LiAer2 also lacks a C-terminal pentapeptide for CheR binding. We, therefore, expressed LiAer2 in E. coli BT3312 (tsr, aer), which contains the high-abundance aspartate receptor, Tar, but lacks the aerotaxis receptors Aer and Tsr. After BT3312 cells were induced with 200 μM IPTG for 45 min, LiAer2_1-976 stimulated robust tumbling in N₂ ∼10 s after switching from air and continued tumbling for the 3 min they were observed, but stopped tumbling ∼10 s after N₂ was replaced with air (Fig. 2A). O₂, thus, turned LiAer2 signaling off. This response was inverted relative to the response generated by all previously tested Aer2 receptors, which orchestrated E. coli tumbling in air instead of N₂ (quantitated for PaAer2 in BT3312 in Fig. S2B). The direction of the LiAer2 response in BT3312 is the same as that observed in classic Aer-mediated aerotaxis (23). In Aer-mediated aerotaxis, though, Aer infers O₂ concentration through a PAS-flavin redox sensor, promoting cell tumbling in N₂ and smooth-swimming in O₂ (23, 24).

FIG 2.

LiAer2-directed behavior in E. coli BT3312. (A) The average percentage of E. coli BT3312 cells tumbling over a 1 s period, 30 s after switching to air or N₂. Results are shown for BT3312 expressing full-length LiAer2_1-976, truncated LiAer2 receptors, or cells containing the empty vector pProEXHTa (pProEX). Error bars represent the standard deviation from three to six independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001. There was no significant difference in percent tumbling for LiAer2_597-976 or pProEX in air or N₂ (P > 0.05). (B) E. coli BT3312 expressing LiAer2_1-976, LiAer2_203-976, or containing pProEX, in tryptone soft agar with 0, 50, or 100 μM IPTG. Plates were incubated at 30°C for 16 h.

PaAer2 is an O₂ sensor, but it also induces E. coli tumbling in the presence of carbon monoxide (CO) and nitric oxide (NO) whereas VcAer2 does not (5, 6). To determine if BT3312 cells expressing LiAer2_1-976 could respond to CO, CO was introduced into the gas perfusion chamber under anaerobic conditions. This introduction caused cells to immediately stop tumbling and commence smooth swimming. After the CO gas was removed, cells continued swimming smoothly for ∼25 s and then resumed tumbling. Once again, this response was inverted relative to that observed with PaAer2, for which CO induces E. coli tumbling (5). An NO response could not be verified because anaerobic BT3312 cells expressing LiAer2_1-976 responded by swimming smoothly when mixed with either the NO donor Proli NONOate or the 10 mM NaOH solution that was used to resuspend the NO donor (2.2 mM NaOH final concentration). LiAer2_1-976, therefore, responds immediately to an increase in pH, in contrast to PaAer2, which has a delayed tumbling response to NaOH (5).

The PAS3 domain of LiAer2 is sufficient for reversible O₂-mediated signaling.

Since LiAer2 responds to O₂ and contains four potential sensing domains (a periplasmic LBD and three cytoplasmic PAS domains), we next determined which of these domains are necessary for O₂ sensing. To do this, four N-terminally truncated LiAer2 receptors were created and tested in the temporal assay (Fig. 2A). All of the N-terminal truncations resulted in drastically increased steady-state cellular LiAer2 levels (Fig. S1B). Interestingly, removing the periplasmic domain, the TM region, PAS1 and PAS2 did not prevent LiAer2 signaling in BT3312 (Fig. 2A). PAS3 alone, attached to the di-HAMP and kinase control domains, was sufficient for a reversible O₂-mediated response. Notably, VcAer2, which has two PAS domains, cannot signal when the N-terminal PAS domain is removed (6), whereas Vibrio vulnificus Aer2, which has three PAS domains, can reversibly signal with only its C-terminal PAS domain (Stuffle and Watts, unpublished data). The reasons for these differences are unclear.

The aspartate receptor, Tar, assists LiAer2 signaling in E. coli.

Because LiAer2 responds to oxy-gases in E. coli BT3312 but not in BT3388, a chemosensory component present in BT3312 likely assists with LiAer2 signaling. For other Aer2 receptors, signaling in E. coli is potentiated by CheR-mediated methylation (5). Given that LiAer2 lacks a C-terminal pentapeptide for CheR binding, assistance neighborhoods with Tar may be necessary for LiAer2 signaling, as it is with E. coli’s low abundance chemoreceptors. However, it is also difficult to visualize LiAer2, with its three cytoplasmic PAS domains, di-HAMP region and 34H kinase control domain (Fig. 1B), aligning well with the cytoplasmic portion of Tar (comprised of just one HAMP domain and a 36H kinase control region) and forming mixed receptor teams. In tryptone soft agar, VcAer2 obliterates the chemotaxis rings formed by both E. coli Tsr and Tar with only 100 μM IPTG induction, most likely by titrating chemotaxis components away from E. coli’s chemoreceptors (6). In contrast to VcAer2, LiAer2 did not disrupt the aspartate ring generated by Tar at the edge of the BT3312 colony in tryptone soft agar at induction levels between 50 and 1000 μM IPTG (see Fig. 2B). This covered an ∼17-fold range of expressed LiAer2. Soluble LiAer2_203-976 did decrease the colony expansion rate (Fig. 2B), which may be due to the higher cellular levels of this truncated mutant (Fig. S1B). In addition, LiAer2 did not induce any new chemotaxis rings in BT3312 (Fig. 2B). These data indicate that Tar allowed LiAer2 responses to be observed in E. coli but there appeared to be no reciprocal enhancement or inhibition of Tar-directed behavior.

To further assess the role played by Tar and methylation in LiAer2 signaling, E. coli Tar and LiAer2 were expressed from compatible plasmids in chemoreceptorless E. coli, both in the absence (UU2610) and presence (UU2612) of CheR and CheB. As anticipated, LiAer2 alone did not induce tumbling in either of the CheRB⁺ or CheRB⁻ strains in air or N₂ (Fig. 3). Similarly, Tar did not induce LiAer2-mediated tumbling in the absence of CheR and CheB (Fig. 3). In contrast, when LiAer2, CheR, CheB, and Tar were present, the response mirrored that of LiAer2 in BT3312 with LiAer2 mediating a tumbling response in N₂ (Fig. 2A and 3). This suggests that LiAer2 methylation, as assisted by Tar, potentiates LiAer2-mediated signaling in E. coli in response to changes in O₂ concentration.

FIG 3.

LiAer2-directed behavior in E. coli UU2610 (CheRB⁻) and UU2612 (CheRB⁺) in the absence and presence of EcTar. LiAer2_1-976 and EcTar were expressed from compatible plasmids in isogenic, chemoreceptorless E. coli strains in the absence and presence of the adaptation enzymes CheR and CheB. Graphs represent the average percentage of E. coli cells tumbling over a 1 s period, 30 s after switching to air or N₂. Error bars represent the standard deviation from two to four independent experiments. ****, P < 0.0001.

The three LiAer2 PAS domains bind hexacoordinate b-type heme.

The three PAS domains of LiAer2 share substantial sequence identity. PAS1 has 73.5% sequence identity with PAS2 and 60% identity with PAS3, whereas PAS2 and PAS3 have 61% sequence identity (Fig. 4). All three PAS domains have ∼50% sequence identity with the PAS domain of PaAer2 (PaPAS). The three LiPAS domains contain the Eη-His residue that coordinates heme in other Aer2 PAS domains, and the Iβ-Trp residue that stabilizes O₂ binding (Fig. 4, boxed and colored dark blue and orange, respectively) (6, 18). These two residues are defining features of Aer2-type PAS-heme domains. Residues lining the heme pocket (based on the structure of PaPAS (17)) are conserved, hydrophobic residues in all three LiPAS domains (highlighted gray in Fig. 4A). To determine if the LiPAS domains bind heme, the coding regions for PAS1_219-333, PAS2_350-464, and PAS3_481-595 were cloned into pProEX, expressed with an N-terminal His tag, purified on nickel-nitrilotriacetic acid (Ni-NTA) agarose, and scanned in the UV/Vis spectrum for cofactors. PAS1 purified to ∼90% homogeneity on Ni-NTA agarose and was easier to work with than PAS2 or PAS3, which had to be purified from much larger cell volumes and routinely purified to only 40 to 50% homogeneity. Scanning UV/Vis spectroscopy revealed that each of the three PAS domains bound b-type heme (Fig. 5). The deoxy-heme spectra of all three LiPAS domains had Soret peaks between 425 and 430 nm and, uniquely for Aer2-type PAS domains, had β and prominent α bands indicative of low-spin hexacoordinate heme. The PAS-heme domains of PaAer2 and VcAer2 each bind pentacoordinate heme, characterized by Soret peaks at 433 nm and a single broad band replacing the α/β-bands (5, 6). An additional four PAS-heme domains from V. vulnificus Aer2 and Methylomicrobium alcaliphilum Aer2 likewise contain pentacoordinate heme (Stuffle, Orillard and Watts, unpublished data). The hexacoordinate PAS-hemes of LiAer2 are thus unique among studied Aer2 receptors. The spectral features of the LiPAS domains resembled those of EcDOS, which also has hexacoordinate heme in the deoxy state (25). In addition to binding hexacoordinate b-type heme, the three LiPAS domains bound O₂ and CO with affinities similar to each other and the PaAer2 and VcAer2 PAS domains (Fig. 5) (6, 18). Of note, PAS2 bound O₂ with the highest affinity of the three LiPAS domains, but, when titrated with O₂, it readily shifted from oxy-bound to oxidized heme with a Soret peak at 411 nm (unpublished data), indicative of unstable O₂-binding.

FIG 4.

LiPAS sequence alignment and structural elements. (A) An alignment of the Aer2 PAS domain sequences from L. interrogans serovar Pomona strain Pomona and P. aeruginosa PAO1 as generated by Clustal Omega. Stars indicate conserved residues, colons indicate similar amino acids, and periods indicate amino acids with weakly similar properties. The conserved Eη-His that coordinates heme in PaPAS, and the Iβ-Trp that stabilizes O₂-binding in PaPAS, are boxed and colored blue and orange, respectively. Residues that line the heme cleft of PaPAS are highlighted gray. The C-terminal “DxT” motifs that conformationally couple PAS to a C-terminal domain are boxed and colored red. Secondary structure elements are based on the solved structures of PaPAS (16, 17). (B) LiPAS3 cartoon from Fig. 1C (showing the right-side protomer rotated by −10° on the y-axis) with secondary structure elements labeled.

FIG 5.

PAS1-3 heme spectra and dissociation constants. Absorption spectra of 5 to 10 μM purified LiAer2 PAS domains in the reduced (deoxy), oxygen-bound (oxy), and carbon monoxide-bound (carbonmonoxy) states. The wavelengths of each absorbance maximum (the Soret and β and α peaks) are indicated. Inserts show an expanded view of the β and α peaks between 500 and 600 nm. The O₂ and CO affinities for each PAS domain are also shown.

Eη-His residues coordinate heme binding to PAS1, PAS2, and PAS3.

The PaAer2 and VcAer2 PAS domains coordinate iron via four protoporphyrin ligands and a single axial ligand from the His residue on Eη (6, 16–18). Because the Eη-His residue is conserved in the proximal heme pocket of all three LiPAS domains (Fig. 4), Eη-His to Ala mutants were created for PAS1_219-333, PAS2_350-464, and PAS3_481-595. The corresponding proteins were stable (Fig. S1C) but had substantial heme-binding defects relative to their corresponding WT PAS domain (1 to 4% of WT PAS-heme content, Fig. 6A). The strong binding defects indicate that Eη-His coordinates heme in all three LiPAS domains and the sixth coordination ligand does not contribute substantially to heme binding. To identify the sixth coordination ligand, we searched for residues in the distal heme pocket that are known to coordinate heme in other proteins, including His, Lys, Tyr, Met, and Cys (26). EcDOS, for example, coordinates deoxy-heme with a His residue in the proximal heme pocket (on PAS-Fα) and a Met side chain in the distal heme pocket (on the FG-loop) but swaps Met coordination for a water molecule in the ferric state (27, 28). The RcoM PAS-heme domain likewise utilizes an FG-loop Met as an axial ligand in the distal heme pocket, whereas the PAS-A domain of NPAS2 uses either a Gβ-His or a neighboring Cys residue, depending on the oxidation state of the heme (29, 30). Surprisingly few residues in the distal heme pockets of the LiPAS domains were identified that could potentially coordinate heme. A Met residue on PAS-Aβ most closely fit the criteria, although it lies 7.4 Å from the Fe atom in the PAS model. This residue was replaced with Ala in PAS1_219-333 (M233A), but deoxy-heme remained hexacoordinate (unpublished data). The heme position in the LiAer2 PAS domains may differ from our models, in which case a potential coordinating residue was overlooked due to its perceived location. The nature of the endogenous axial ligand that is displaced for gas binding in LiAer2 remains to be determined. In EcDOS, dissociation of the distal Met is the main energy barrier for gas binding (25, 28). Substituting the Met residue with Ile increases the O₂ affinity of EcDOS but at the expense of discriminating O₂ over CO (28). For LiAer2, future studies employing mutagenesis, resonance Raman spectroscopy, and structural methods, such as crystallography, would help in identifying the unknown ligand. Its identity is important because the signal-on state of LiAer2 occurs when that coordinating ligand binds heme, and its displacement for gas-binding shuts LiAer2 signaling off. Whether other Aer2 receptors exist that have hexacoordinate deoxy-heme and whether those receptors have inverted O₂ signaling remains to be seen.

FIG 6.

Analysis of PAS Eη-His and Iβ-Trp mutants. (A) Average heme content of PAS domains containing Eη-His or Iβ-Trp replacements, given as a percentage of WT PAS heme content and corrected for protein concentration. Error bars represent the standard deviation from three independent experiments. (B) Representative O₂ titration for 10 μM purified PAS1-W329L showing O₂ concentrations of interest. The “?” indicates that the heme is bound to an unidentified ligand.

LiAer2 Iβ-Trp mutants respond differently to O₂.

Aer2 PAS domains use their Iβ-Trp residue to stabilize O₂ binding and initiate signaling (6, 16, 18). The Trp residue rotates ∼90° to bond with the ligand, thus initiating a conformational signal that is transmitted to the downstream HAMP domains (31). The Iβ-Trp residue is universally conserved in Aer2-type PAS domains (18), including the three LiPAS domains (Fig. 4). Analysis of a PAS1_219-333 Iβ-Trp mutant (W329L) showed a limited impact on heme-binding (Fig. 6A), but O₂ titrations gave a different picture to what we have observed with all other Aer2-PAS Iβ-Trp mutants. Typically, O₂ titrations rapidly shift the spectra of Iβ-Trp mutants (including Trp to Leu mutants) from deoxy- to met-heme. An important exception to this is PAS1 from VcAer2, which uses either an Iβ-Trp residue or a Gβ-Tyr residue for O₂ binding (6). VcPAS1 Iβ-Trp mutants consequently have identical spectra to the WT PAS domain because the Trp mutation alone does not affect O₂ binding. In the case of PAS1_219-333-W329L, O₂ titration caused a slight spectral shift in the Soret peak from 424 nm (in the deoxy-state) to 421 nm, with a concomitant reduction in the height of the prominent α band and a slight redshift (Fig. 6B). These spectra do not match the oxy-bound spectra of PAS1 (which has a Soret peak at 417 nm, Fig. 5), and appear to be due to an unidentified ligand. None of the LiPAS domains contain the Gβ-Tyr that stabilize O₂-binding to VcPAS1. Lastly, CO affinity could not be determined for PAS1_219-333-W329L because the Soret peak for deoxy-PAS1_219-333-W329L was only 1 nm different from that expected for CO-bound spectra and CO titration showed no spectral shift.

LiAer2 is a membrane-bound PAS-heme O₂ sensor for which O₂ binding turns signaling off.

In this study, we showed that all Aer2 receptors were not created equal. LiAer2 is a membrane-bound Aer2 with three PAS-heme domains, of which only PAS3 was required for reversible O₂-mediated signaling. In VcAer2, PAS1 modulates O₂-mediated signaling from PAS2, a role that may require PAS1 to bind O₂ (6). Although LiAer2 can function without PAS 1-2, O₂-binding to PAS1 and/or PAS2 may regulate PAS3 function. Importantly, all of the LiPAS (as well as the PaPAS and VcPAS) domains are physically linked to downstream regions via conserved C-terminal “DxT” motifs (boxed and colored red in Fig. 4A). In PaAer2, O₂-mediated rotation of the Iβ-Trp moves the DxT regions closer together in the dimer, and the PAS N-terminus (the N-caps) move further apart (16, 31). The same movements may occur in the three LiPAS domains, although the signal output is ultimately inverted. Thus, O₂-regulated motions may be directly linked to the displacement of the sixth coordination ligand to invert signal output, or signal inversion occurs C-terminal to the PAS3 domain. In EcDOS, liberating the distal coordinating ligand from the heme releases a “catalytic lock” on protein function (32). It will be interesting to determine the choreography of the coordinating ligand and Iβ-Trp residue in LiPAS and how that influences O₂ binding and signaling.

An emerging theme from Aer2 studies is that Aer2 receptors contain a core composed of the most C-terminal PAS-heme domain, followed by the di-HAMP region and kinase control domain. The region N-terminal to the distal PAS domain differs depending on the species in which the receptor is found. LiAer2 is thus far unusual in that it includes a periplasmic domain that could potentially serve as a site for secondary signal input (Fig. 1B). The periplasmic domain of LiAer2 is not required for O₂-mediated signaling but it could potentially regulate O₂ signaling along with PAS1 and PAS2. The periplasmic domain of LiAer2 is a predicted four-helix bundle, with structural similarity to periplasmic LBDs like those found in Tsr and Tar (Fig. 1B) (33). However, the sequence appears to be specific to Leptospires. In chemoreceptors, four-helix bundles bind a multitude of small molecules like amino acids, citrate, benzoate, and nitrate (33). Whether the periplasmic domain of LiAer2 binds ligands, and how that affects O₂-mediated signaling, awaits further analysis.

In summary, LiAer2 is a membrane-bound Aer2 receptor with three PAS-heme domains and inverted O₂-based signaling compared with other Aer2 receptors. Future studies will be needed to tease out the unique signaling mechanisms of this receptor in L. interrogans.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

L. interrogans serovar Pomona (ATCC 23478) was grown in modified Leptospira medium (0.3 g/liter peptone, 0.2 g/liter beef extract, 0.5 g/liter sodium chloride, supplemented with 10% sterile rabbit serum [Sigma-Aldrich, St. Louis, MO] and 2.5 mL/liter of 0.05% hemin [Sigma-Aldrich, St. Louis, MO]) at 37°C. E. coli BT3312 (tsr, aer) (34), BT3388 (tar, tsr, trg, tap, aer) (35), and isogenic UU2610 (tar, tsr, trg, tap, aer, cheR, cheB) (36), and UU2612 (tar, tsr, trg, tap, aer) (36) were grown at 30°C or 37°C in lysogeny broth (LB; Lennox), tryptone broth (TB) or in tryptone soft agar, supplemented with 0.5 μg/mL thiamine and 50 or 100 μg/mL ampicillin or 25 μg/mL chloramphenicol as needed.

Mutagenesis and cloning.

The aer2 gene was amplified from the genomic DNA of L. interrogans serovar Pomona using PfuUltra II Fusion DNA polymerase (Agilent Technologies, Santa Clara, CA) and cloned into the NcoI and XbaI sites of pProEXHTa to express full-length LiAer2_1-976 (WP_000488891) with an N-terminal His₆ tag. To create constructs expressing truncated LiAer2 proteins or individual PAS domains, the appropriate DNA fragments were amplified from pProEXHTa-LiAer2 and cloned into the NcoI and XbaI sites of pProEXHTa. Site-directed mutagenesis was performed on plasmids containing the individual PAS domains using site-specific primers and PfuUltra II Fusion DNA polymerase. Amplification products were treated with DpnI (New England Biolabs, Ipswich, MA) to remove template strands, and electroporated into E. coli. For all constructs, products of the correct size were confirmed by Western blotting with 1:10,000 HisProbe-HRP (Thermo Scientific, Rockford, IL), and sequences were confirmed by DNA sequencing. The EcTar expression plasmid pLC114 was acquired from J.S. Parkinson. It contains a p15A origin of replication that is compatible with the pUC origin of replication in pProEXHTa.

Solubility assay.

BT3312 cells expressing full-length LiAer2_1-976 were grown in LB supplemented with 0.5 μg/mL thiamine and 25 μg/mL 5-aminolevulinic acid (Sigma-Aldrich). Protein expression was induced with 600 μM IPTG. Cells were lysed in lysis buffer (50 mM Tris, pH 7.5, 500 mM NaCl, 10 mM imidazole, 0.3 mg/mL lysozyme, 1 μg/mL DNase I, and 100 μL of Protease Inhibitor Cocktail for His-tagged proteins [Sigma-Aldrich]) and centrifuged at 10,000 × g for 20 min. The low-speed supernatant was then centrifuged at 407,000 × g for 1 h. The high-speed pellet (the insoluble fraction) was resuspended by sonication to the same volume as the supernatant (the soluble fraction) in 50 mM Tris, pH 7.5, 500 mM NaCl, and 10 mM imidazole. Samples were electrophoresed in duplicate by SDS-PAGE and bands were visualized on Western blots using 1:100,000 anti-Tsr antibody (a gift from J.S. Parkinson).

Steady-state cellular levels.

The steady-state cellular levels of the N-terminally truncated LiAer2 proteins were compared with full-length LiAer2_1-976 after inducing BT3312 cells with 200 μM IPTG. The steady-state cellular levels of the individual PAS mutants were compared with the corresponding WT PAS protein after inducing BT3388 cells with 200 μM IPTG. The cellular levels of LiAer2_1-976 were also assessed after inducing BT3312 cells with 50 to 1000 μM IPTG. All cells were grown at 37°C in LB with 0.5 μg/mL thiamine and 100 μg/mL ampicillin. Samples were electrophoresed in duplicate by SDS-PAGE and data were averaged from two to five independent experiments. Bands were visualized on HisProbe Western blots and quantified using a UVP ChemStudio (Analytik Jena, Upland, CA).

Behavioral assays.

Behavioral assays were performed in E. coli BT3388, BT3312, UU2610, or UU2612. Cells were grown at 30°C in TB supplemented with 0.5 μg/mL thiamine and 25 μg/mL 5-aminolevulinic acid. For cells containing the EcTar expression plasmid pLC114, 0.6 μM sodium salicylate was also included. This induction level provides optimal performance in behavioral assays (J.S. Parkinson, personal communication). Once cells had reached an optical density at 600 nm (OD₆₀₀) of 0.2 to 0.25, LiAer2 expression was induced for 45 min with 200 μM IPTG. BT3388/LiAer2_1-976 was also analyzed after inducing with 200 μM IPTG for 2 h. Assays were carried out in a gas perfusion chamber and toggled between air (20.9% O₂) and N₂ as previously described (5, 37). The gases were toggled at least twice in each assay and data were averaged from two to six independent experiments. Temporal responses were quantified by eye and the percentage of cells tumbling over a 1 s period, 30 s after switching to air or N₂, was measured using ImageJ (https://imagej.nih.gov/ij/). Statistical analyses were carried out using a Mann-Whitney U test. To determine a CO response, BT3312 cells (induced with 200 μM IPTG) were perfused with N₂ for 30 s before a 10 s perfusion with CO gas (>99% purity, Sigma-Aldrich). To determine an NO response, the NO donor Proli NONOate (Cayman Chemical, Ann Arbor, MI) was used as previously described (5). Swim plate responses were determined by inoculating 2 μL of overnight BT3388 or BT3312 cultures into tryptone soft agar containing 0 to 1000 μM IPTG and 50 μg mL⁻¹ ampicillin and incubating the plates at 30°C for 10 to 16 h.

Protein purification.

BT3388 cells expressing the individual PAS domains were grown in LB supplemented with 0.5 μg/mL thiamine and 25 μg/mL 5-aminolevulinic acid. Protein expression was induced with 600 μM IPTG, and proteins were purified on Ni-NTA agarose columns (Qiagen, Valencia, CA) as previously described (18). The concentration of the eluted proteins was determined in a BCA^TM Protein Assay (Thermo Scientific) using BSA as the standard. Sample quality was assessed by running 2.5 μg of each protein in duplicate on SDS-PAGE, followed by staining with Coomassie brilliant blue.

Heme binding, absorption spectra, and gas affinities.

The heme content of the purified PAS mutants (Soret heights and purity of 5 to 10 μM imidazole-bound proteins) were compared with the appropriate WT PAS protein as previously described (18). The deoxy, oxy, and carbonmonoxy spectra of the purified PAS-heme domains were similarly determined as previously described (5). Dissociation constants for O₂ and CO binding were estimated by linear interpolation of unliganded (Fe²⁺) and liganded (Fe²⁺-O₂, Fe²⁺-CO) spectra, as previously described (18). Experiments were repeated on at least three occasions, from which average affinities were determined and rounded to the nearest whole number.

In silico analysis.

Dimer models were predicted for LiAer2_1-976 in three overlapping fragments (res.1 to 349, 219 to 604, and 596 to 976) using AlphaFold2_advanced (38). The top-ranked models were used for the mid- and C-terminal segments of LiAer2, whereas model four out of five models was used for the N-terminal segment because it more closely simulated the structure of known chemosensory LBDs and TM regions. The boundaries of the LiAer2 TM helices were predicted using TMpred, PRED-TMR (http://athina.biol.uoa.gr/PRED-TMR/), and TMHMM (https://services.healthtech.dtu.dk/service.php?TMHMM-2.0). The TM boundaries listed in Fig. 1B most closely match the consensus of these three servers. A LiAer2 PAS3 dimer model (res.482 to 595) was created in SWISS-MODEL (https://swissmodel.expasy.org) (39) based on the unliganded dimer structure of P. aeruginosa Aer2 PAS with heme bound (PDB 4HI4) (16). To visualize possible heme placement in this model, the PAS3 model was overlaid onto the structure of PaPAS using the “fit” command in PyMOL (https://pymol.org/2/).