ABSTRACT
Bacteria sense environmental chemicals using chemosensor proteins, most of which are present in the cytoplasmic membrane. Canonical chemoreceptors bind their specific ligands in their periplasmic domain, and the ligand binding creates a molecular stimulus that is transmitted into the cytoplasm, leading to various cellular responses, such as chemotaxis and specific gene expression. Vibrio cholerae, the causative agent of cholera, contains about 44 putative sensor proteins, which are homologous to methyl-accepting chemotaxis proteins involved in chemotaxis. Two of them, Mlp24 and Mlp37, have been identified as chemoreceptors that mediate chemotactic responses to various amino acids. Although most of the residues of Mlp37 involved in ligand binding are conserved in Mlp24, these chemoreceptors bind the same ligands with different affinities. Moreover, they have distinct cellular roles. Here we determined a series of ligand complex structures of the periplasmic domains of Mlp24 (Mlp24p). The structures revealed that Ca2+ binds to the loop that forms the upper wall of the ligand-binding pocket. Ca2+ does not bind to the corresponding loop of Mlp37, implying that the structural difference of the loop may cause the ligand affinity difference. Isothermal titration calorimetry (ITC) measurements indicated that Ca2+ changes the ligand binding affinity of Mlp24p. Furthermore, Ca2+ affected chemotactic behaviors to various amino acids mediated by Mlp24. Thus, Ca2+ is suggested to serve as a cosignal for the primary signal mediated by Mlp24p, and V. cholerae fine-tunes its chemotactic behavior depending on the Ca2+ concentration by modulating the ligand sensitivity of Mlp24.
IMPORTANCE Mlp24 and Mlp37 are homologous chemoreceptors of Vibrio cholerae that bind various amino acids. Although most of the residues involved in ligand interaction are conserved, these chemoreceptors show different affinities for the same ligand and play different cellular roles. A series of ligand complex structures of the periplasmic region of Mlp24 (Mlp24p) and following ITC analysis revealed that Ca2+ binds to the loop of Mlp24p and modulates the ligand binding affinity of Mlp24p. Moreover, Ca2+ changes the chemotactic behaviors mediated by Mlp24. We propose that Ca2+ acts as a cosignal that modulates the affinity of Mlp24 for the primary signal, thereby changing the chemotactic behavior of V. cholerae.
INTRODUCTION
Many bacteria move to favorable environments and away from toxic matter by sensing chemicals in their surroundings to ensure survival. The environmental chemicals are sensed by chemoreceptor proteins (1). A canonical chemoreceptor is composed of an extracytoplasmic sensor domain, a transmembrane (TM) region, a HAMP domain, and a kinase control module, and specific ligands bind to the extracytoplasmic sensor domain. The canonical chemoreceptors are classified according to their sensor domain into four major groups by the periplasmic domain structure (2, 3): 4-helix bundle (4HB), helical bimodular (HBM), single CACHE (sCACHE), and double CACHE (dCACHE). 4HB consists of four antiparallel α-helices. Tar and Tsr of Escherichia coli belong to this family (4). HBM is composed of two 4-helix bundle domains tandemly arranged perpendicular to the membrane. McpS of Pseudomonas putida is a representative of this group (5). sCACHE comprises a single domain structurally analogous to the PAS domain, and dCACHE is made up of two CACHE domains tandemly arranged perpendicular to the membrane.
Vibrio cholerae, the etiological agent of cholera, has three sets of chemosensory pathways (Che system) and at least 44 proteins homologous to methyl-accepting chemotaxis proteins (MCPs), termed MCP-like proteins (MLPs) (6). Among the MLPs, Mlp24 and Mlp37 have been identified as major chemoreceptors involved in taxis to various amino acids (6, 7), but they also have other distinct roles. Mlp24 is required for production of cholera toxin (8), and Mlp37 is involved in biofilm formation (7, 9). Although their cellular roles are different, the periplasmic regions of Mlp24 and Mlp37 (Mlp24p and Mlp37p, respectively) show about 44% amino acid sequence identity (Fig. 1). Sequence analysis of Mlp24 and Mlp37 has predicted that they belong to the dCACHE family of chemoreceptors (3), and a recent crystal structure analysis of Mlp37p has confirmed that Mlp37 is a dCACHE-type chemoreceptor (7). Mlp24 and Mlp37 bind various l-amino acids, but the binding affinities for their ligands are different from each other (6, 7). Mlp37 mediates chemotactic responses to various amino acids, such as l-serine, l-alanine, l-cysteine, l-arginine, l-asparagine, l-threonine, l-lysine, glycine, l-valine, and taurine. Isothermal titration calorimetry (ITC) measurements indicated that Mlp37p binds l-alanine, l-arginine, l-serine, and taurine with similar affinity (7). Mlp24 is involved in chemotactic responses to l-serine, l-alanine, l-cysteine, l-arginine, l-asparagine, l-threonine, l-lysine, glycine, l-histidine, l-glutamine, and l-proline. In contrast to Mlp37p, Mlp24p shows different affinities for ligands (6). Moreover, Mlp24p does not bind taurine (7). These differences may be related to the distinct cellular functions of Mlp24 and Mlp37, but details are not known.
Amino acid sequence alignment of Mlp24p and Mlp37p. The secondary structures are shown above (Mlp24p) or below (Mlp37p) the sequences. Gray rectangles denote α-helices, and gray arrows denote β-strands. Residues involved in bonding and nonbonding contacts with the ligand amino acid are marked with circles (filled circles for the amino acid backbone and open circles for the side chain) and arrowheads, respectively. The residues coordinating Ca2+ are indicated by stars above the sequence of Mlp24p. Conserved residues are colored in red. CAD1, CAD2, the UJ loop, and the LJ segment are shaded in blue, orange, yellow, and green, respectively.
The structures of Mlp37p in complex with l-alanine, l-serine, and taurine have recently been revealed (7). Most of the residues involved in interaction with the ligands of Mlp37p are conserved in Mlp24p (Fig. 1), although the affinity and specificity of Mlp24p for the ligands differ from those of Mlp37p. Therefore, the molecular mechanism that produces the different affinities and specificities between Mlp24 and Mlp37 remains a mystery.
To elucidate the molecular mechanisms of ligand recognition by Mlp24, we have determined a series of ligand complex structures of Mlp24p. The structures revealed that Ca2+ binds to the loop that forms the upper wall of the ligand-binding pocket and interacts with ligands, but not that of Mlp37. ITC assays indicated that Ca2+ enhances the affinity of Mlp24p for ligands, especially for small amino acids. Consistent with this, chemotaxis assays demonstrated that Ca2+ changes the chemotactic behavior mediated by Mlp24. These lines of evidence suggest that V. cholerae changes the chemotactic behavior in response to the Ca2+ concentration by modulating the ligand sensitivity of Mlp24.
RESULTS
Overall structures of Mlp24p and its ligand complexes.We determined the crystal structures of Mlp24p and its five l-amino acid ligand complexes (l-arginine, l-asparagine, glycine, l-proline, and l-serine) at a 2.1- to 2.7-Å resolution (Fig. 2; Tables 1 and 2). The ligand complex crystals were prepared by cocrystallization and grew with different space groups and unit cell parameters under distinct conditions. The molecular arrangements in these crystals, therefore, differ from one another (Table 2). As expected from the sequence analysis (Fig. 1), the tertiary and quaternary structures of Mlp24p are very similar to those of Mlp37p (see Fig. S1 in the supplemental material). Each Mlp24p subunit is composed of a long N-terminal α-helix (H1), two CACHE domains (CAD1 and CAD2) tandemly arranged along H1, and a C-terminal helix (H6) (Fig. 1 and 2A). CAD1, the membrane-distal domain, consists of three α-helices (H2 to H4) and six β-strands (S1 to S6), five of which (S1 to S3, S5, and S6) form a central β-sheet (Fig. 2A). The domain has a mouth-like pocket, with the upper jaw formed by the loop of 16 residues connecting S2 and H4 (UJ loop) and the lower jaw formed by S4, the N-terminal half of S5, and the loop connecting S4 and S5 (LJ segment). CAD2, the membrane-proximal domain, comprises an α-helix (H5) and a β-sheet composed of five β-strands (S7 to S11). Although CAD2 shows a fold similar to that of CAD1, the ligands bind only to CAD1 (Fig. 2B to G), like other known dCACHE-type chemoreceptors except for TlpC (10–13).
Overall structure of Mlp24p. (A) A ribbon model of the Mlp24p monomer without ligand is colored in rainbow colors from the N terminus (blue) to the C terminus (red). The membrane-distal and -proximal CACHE domains (CAD1 and CAD2) are indicated with black and gray broken circles, respectively. The secondary structure elements are labeled. (B to G) Ribbon representation of the dimeric structures of Mlp24p (B), the Mlp24p–l-arginine complex (C), the Mlp24p–l-asparagine complex (D), the Mlp24p-glycine complex (E), the Mlp24p–l-proline complex (F), and the Mlp24p–l-serine complex (G). The ligand molecules are shown as ball models. Each subunit is in a different color (blue or red).
Summary of crystallization conditions with protein Mlp24p at 293 Ka
Summary of X-ray data collection and refinement statistics
All the crystal structures of Mlp24p show homodimers (Fig. 2B to G). The C-terminal region of H1, H2, and the N-terminal region of H3 contribute to the dimer interaction (see Fig. S2 in the supplemental material). L63, I70, and L73 in H1, F80, and V84 in H2, and L90 in H3 form a hydrophobic interaction core in the dimer interface. The dimer is further stabilized by hydrophilic interactions surrounding the hydrophobic interface. The side chain of N89 in H3 hydrogen bonds to S66′ in H1′ (a prime denotes the partner subunit), and the main-chain nitrogen atom of N89 hydrogen bonds to the side chain of E69′ in H1′. These dimer interactions are conserved in all the Mlp24p structures.
Although the six crystal structures show almost identical dimer interactions, they show different dimeric structures that can be classified into three groups (Fig. 2). Mlp24p, the l-arginine complex, and the l-asparagine complex show similar dimeric structures, in each of which the proximal ends of the two subunits are apart (Fig. 2B to D). They adopt very similar bending conformations in H1 (see Fig. S3A in the supplemental material), which is bent outward at S55, and therefore the N-terminal ends of H1 in the dimer are apart from each other to form a proximal open dimer conformation. In contrast, the H1 helices in the glycine and l-proline complexes have no bending and extend almost straight (Fig. 2E and F). Therefore, they adopt a nearly parallel dimer conformation. The conformation of H1 in the l-serine complex resembles those in the glycine and l-proline complexes (Fig. 2G). However, the dimer structure is slightly different (Fig. S3A), because the second subunit of the dimer tilts at the dimer interface at 12.5° around the axis perpendicular to the dimer axis. These differences in the dimer structures seem to be caused by different molecular packing in the crystals. The N-terminal regions of the two H1 helices of the dimer cross those of the neighboring dimer in the crystals of the proximal open dimers (Fig. S3B to D) but not in the other crystals (Fig. S3E to G).
l-Amino acid recognition of Mlp24p.The ligand amino acid is bound in the mouth-like pocket of CAD1. The pocket is composed of the UJ loop, H4, the LJ segment, and the central β-sheet. A calcium ion binds to the N-terminal region of the UJ loop and defines the orientation of the loop by forming a unique turn conformation. Ca2+ is coordinated by the side chain carboxy groups of E109 and E115, the main-chain carbonyl oxygen atoms of D111 and W114, and water molecules with octahedral geometry (Fig. 3A). These interactions bend the UJ loop around Ca2+ to form the turn structure. G113 at the corner of the turn is important to form the turn structure. This loop conformation is further stabilized by the hydrogen bond between the side chain carboxy group of D111 and the main-chain NH group of G113 (Fig. 3A). E109, G113, and E115 are not conserved in Mlp37p (Fig. 1 and 4A), and therefore Mlp37p does not bind Ca2+.
Close-up view of the ligand-binding pocket. (A) Ca2+ coordination to the UJ loop. (B to H) The ligand-binding pockets of Mlp24p in complex with l-arginine (B), l-asparagine (C), l-serine (D), l-proline (E), glycine (F), l-alanine (G), and acetate (H). l-Alanine is bound in chain B and acetate in chain A of the Mlp24p crystallized solution without l-amino acid ligands. (I) The ligand-binding pocket of Mlp37p in complex with l-serine. The ligand molecules are colored in orange. The protein residues are represented by gray stick models. The nitrogen and oxygen atoms are colored in blue and red, respectively. The bound water molecules are shown as red balls. Ca2+ is shown as a cyan ball. Possible hydrogen bonds are indicated by broken lines.
Comparison of the ligand-binding structures of CHD1. (A) Sequence alignment of Mlp24 with various MCPs and MLPs around the N-terminal region of the UJ loop. Residues identical or homologous to those of Mlp24 are shaded. The residues important for Ca2+ binding (E109 and E115) and for the turn structure (G113) are conserved in MLPs in marine bacteria (highlighted in red). Vc, Vibrio cholerae; Va, Vibrio alginolyticus; Vp, Vibrio parahaemolyticus; Pa, Pseudomonas aeruginosa; Bs, Bacillus subtilis. (B) CHD1 of the Mlp24p–l-arginine complex (pink), the Mlp24p–l-arginine complex (red), the Mlp24p–l-asparagine complex (orange), the Mlp24p-glycine complex (green), the Mlp24p–l-proline complex (purple), and the Mlp24p–l-serine complex (cyan) are superimposed onto CHD1 of the Mlp24p-acetate complex (gray). The residues involved in ligand recognition or Ca2+ coordination are shown in stick models and labeled. Ca2+ is shown as a cyan ball.
All the ligand amino acids bind to the pocket in the same orientation, and their α-amino and α-carboxy groups are recognized by Mlp24p in the same manner. The α-carboxy group forms hydrogen bonds with Y120 in the UJ loop and with R125 and W127 in H4. The α-amino group hydrogen bonds with D172 in S6 and with Y143 and D145 in the LJ segment (Fig. 3B to G). These three residues are triangularly arranged below the amino group. This arrangement is reasonable because the amino group is positively charged under our experimental conditions (pH < 9). These interactions are very similar to those in the Mlp37p-ligand complexes, although Y120 is replaced by W147 in Mlp37p.
In contrast to the ligand main chain, the ligand side chains are recognized in different manners, although the conformation of the residues involved in the ligand side chain recognition is almost identical in all cases. The guanidino group of the ligand l-arginine directly interacts with D111 in the UJ loop and D145 in the LJ segment, and the hydrophobic side chain arm of l-arginine interacts with F170 and W114 (Fig. 3B). The side chain of the ligand l-asparagine is recognized by the LJ segment. The side chain carbonyl oxygen of l-asparagine hydrogen bonds to the main-chain NH group of S147, but no specific hydrophilic interaction with the side chain NH2 group of l-asparagine was found. The β-methylene group of l-asparagine is in contact with F170 (Fig. 3C). Unlike the case for l-arginine and l-asparagine, the side chain hydroxy group of the ligand l-serine is indirectly recognized by D111 and D145 through the hydrogen bonding network mediated by three water molecules. The β-methylene group of l-serine interacts with F170 as in the case of l-asparagine (Fig. 3D). The side chain atoms of the ligand l-proline are in hydrophobic contact with F170 and W114 (Fig. 3E). The ligand-binding pocket of the glycine complex has a space between the bound glycine and W114 (Fig. 3F). The conformation of the UJ loop is almost identical in all the ligand complexes, and therefore the small ligands do not tightly contact the UJ loop in the pocket.
l-Amino acid binding induces domain closure of CAD1.The CAD1 domains of the dimer adopt different conformations in the structure of Mlp24p crystallized in solution without l-amino acid ligands. The LJ segment of CAD1 in one subunit of the Mlp24p dimer projects outward, and hence the entrance of the pocket is wide open (Fig. 4B). A small density is present in the CAD1 pocket (see Fig. S4A in the supplemental material) and is assigned as acetate because the shape of the density is like acetate and the crystallization buffer contained acetate. The carboxy group of the acetate interacts with Y120, R125, and W127 in the same manner as the α-carboxy group of the ligand l-amino acids (Fig. 3H). CAD1 in the other subunit of the Mlp24p dimer adopts a closed conformation. The LJ segment bends upward to close the entrance of the pocket. A clear density that can be assigned as l-alanine was observed in the pocket (Fig. 3G and S4B), although all solutions used in purification and crystallization contained no l-alanine. Such unexpected l-alanine binding also occurred in the structure analysis of Mlp37p (PDB no. 3C8C). The density corresponding to l-alanine was found in CAD1 in both subunits of the Mlp37p dimer without adding l-alanine in the crystallization solution. The CAD1 domains of all the l-amino acid complex structures of Mlp24p show a closed conformation. These observations suggest that l-amino acid binding to the CAD1 domain pocket induces bending of the LJ segment to close the entrance of the pocket. A similar conformational change has been observed in Tlp3 (14).
ITC analysis of the Mlp24 periplasmic fragment with and without Ca2+.The structures of Mlp24p in complex with the ligands revealed that Ca2+ stabilizes the unique turn conformation of the UJ loop. To explore the role of Ca2+ in ligand recognition, we carried out in vitro ligand-binding assays using isothermal titration calorimetry (ITC) (Table 3). We constructed a new plasmid encoding a His tag-free fragment of Mlp24p (Mlp24pNH) to avoid undesired binding of ions to the His tag and purified it for ITC measurements. The results demonstrated that Mlp24pNH binds Ca2+, although the affinity is relatively low (the Kd [dissociation constant] is in the submillimolar range) (Fig. 5A and B and Table 3). On the other hand, Mlp24pNH did not bind Mg2+ (see Fig. S5A in the supplemental material).
Ligand binding affinities of the Mlp24 periplasmic fragment for various amino acids and divalent cations
Binding of Ca2+ and asparagine to the periplasmic fragment of Mlp24 (Mlp24pNH). (A) Titration of the ITC measurement buffer with 10 mM CaCl2. (B to D) ITC measurements with 10 μM Mlp24pNH were carried out with 10 mM CaCl2 (B) or 10 mM l-asparagine in the absence (C) or presence (D) of 10 mM CaCl2. Enthalpy changes per mole are plotted as a function of the molar ratio of CaCl2 or asparagine to Mlp24pNH.
Next, we carried out ITC analyses of Mlp24pNH with various amino acids in the presence and absence of 10 mM CaCl2 in EDTA-free buffer. We found that Ca2+ significantly decreased the Kd values of Mlp24pNH for amino acids with smaller side chains (Table 3). For example, the Kd values of Mlp24pNH for l-asparagine in the presence and absence of Ca2+ were 6.3 μM and 49.5 μM, respectively, and those for serine were 18.3 μM and 184.2 μM, respectively (Fig. 5C and D, Fig. S5B and C, and Table 3). On the other hand, the Kd values for amino acids with long, positive side chains, such as l-arginine and l-lysine, were low even in the absence of Ca2+ and were slightly decreased by the addition of Ca2+ (Fig. S5D and E and Table 3). l-Glutamate, a negatively charged amino acid, did not bind to Mlp24pNH. These results indicate that Ca2+ increases affinities of Mlp24 for the ligands, especially for the amino acids with smaller side chains. However, Mlp24 still shows different affinities for ligands to some extent in the presence of 10 mM Ca2+ (Table 3).
Ca2+ enhances Mlp24-mediated chemotaxis to some amino acids in vivo.To examine whether Ca2+ actually plays a role in chemotaxis, we carried out capillary assays (6) of cells expressing Mlp24 as a sole major amino acid chemoreceptor: a strain (Vmlp201) lacking Mlp24 as well as Mlp37 (7) transformed with an mlp24-expressing plasmid (pMlp24). The mutant cells overexpressing Mlp24 were not attracted to the buffer (TMN) containing up to 100 mM CaCl2 (Fig. 6A), suggesting that Ca2+ binding to Mlp24 by itself did not elicit an attractant signal. We then examined responses to l-asparagine, l-serine, and l-arginine in the presence and absence of 10 mM CaCl2. The responses to 0.1, 1, and 10 mM l-asparagine were enhanced by Ca2+; the most dramatic effect was seen in the response to 1 mM l-asparagine (Fig. 6B). Ca2+ showed a milder effect on responses to 10 mM l-serine (Fig. 6C). In contrast, no significant effects were observed for the l-arginine response (Fig. 6D). These results are basically consistent with the results of ITC measurements. Taken together the results indicate that Ca2+ binding to Mlp24 potentiates chemotaxis to amino acids with smaller side chains by increasing its ligand-binding affinities.
Chemotactic responses of Vibrio cholerae cells overexpressing Mlp24 to various amino acids. Vmlp201 (Δmlp24 Δmlp37) cells carrying pMlp24 (the Mlp24-FLAG-expressing plasmid) (circles) or pAH901 (the empty vector) (triangles) were subjected to capillary assays with calcium (A) and with l-asparagine (B), l-serine (C), and l-arginine (D) in the absence (open symbols) or presence (closed symbols) of 10 mM CaCl2.
DISCUSSION
We have determined a series of ligand complex structures of Mlp24p and revealed the molecular basis of the ligand recognition of Mlp24. The crystal structures of the Mlp24p indicated that the N-terminal region of the UJ loop has a unique turn structure stabilized by a calcium ion (Fig. 3). Interestingly, no crystal was obtained without calcium ions. The turn structure restricts the conformation of the N-terminal region of the UJ loop and determines the orientation of the UJ loop (Fig. 3 and 4B). Therefore, the calcium ion is thought to be important for ligand binding. The ITC measurements confirmed that Ca2+ affects the binding affinity of Mlp24 for the ligands. The binding affinities for small amino acids are low without Ca2+ but are significantly increased by the addition of Ca2+ (Table 3). G113 is located at the corner of the turn, and the conformational flexibility of the glycine residue is essential to form the turn structure around the calcium ion (Fig. 3A). Without the bound Ca2+, the turn structure may be disrupted and the orientation of the UJ loop may not be well determined because of the high conformational flexibility of G113. Consequently, the upper wall of the pocket is not properly formed, and the ligands do not stably bind to Mlp24p. l-Arginine, however, strongly binds to Mlp24p even in the absence of Ca2+. This is probably because the guanidino group of arginine is recognized mainly by an electrostatic interaction with D145, and therefore, Ca2+ would not significantly alter the affinity for l-arginine.
ITC measurements have indicated that Mlp24p shows different affinities for the ligands in the presence of 10 mM Ca2+. The UJ loop of Mlp24p adopts almost the same conformation in all of the complex structures, the pockets of which are very similar in size and shape (Fig. 4B). The pocket of Mlp24p nicely accommodates l-arginine but is rather large for smaller amino acids. Therefore, the binding affinities of Mlp24p for smaller amino acids are relatively low, except for alanine.
The surface charge distribution of Mlp24p indicates that the entrance of the binding pocket is negatively charged (see Fig. S6 in the supplemental material). The negatively charged surface may attract positively charged amino acids, such as l-arginine and l-lysine, and repel negatively charged amino acids, such as l-glutamate and l-aspartate. This is probably another reason why Mlp24 strongly binds amino acids with basic side chains and does not bind those with acidic side chains (Table 3).
Taurine binds to Mlp37 but not to Mlp24 (7). The sulfonate group of taurine in Mlp37p hydrogen bonds to W141 through water molecules (PDB no. 5AVF). W114 of Mlp24p, which corresponds to W141 of Mlp37p, is too distant to form a hydrogen bond to the sulfate group of taurine, and hence Mlp24p does not bind taurine.
Mlp24 binds Ca2+ with a Kd of about 0.45 mM. The concentration of Ca2+ is about 10 mM in seawater and in the gastrointestinal tract (15). Therefore, 96% of the Mlp24 molecules bind Ca2+ in seawater and in the gastrointestinal tract, whereas 50% to 80% of the Mlp24 molecules release Ca2+ in freshwater (Ca2+ concentration, 0.025 to 0.5 mM). Ca2+ increases the binding affinity of Mlp24 to smaller amino acids to the levels close to those of l-arginine and l-lysine, and as a result, Mlp24 has broader substrate specificity. The chemotactic behavior of V. cholerae mediated by Mlp24 is also affected by Ca2+. Therefore, Ca2+ is considered to be a cosignal for the primary signal mediated by Mlp24p, and it is possible that V. cholerae fine-tunes its chemotactic behavior in response to environmental calcium concentrations by modulating the ligand sensitivity of Mlp24, although the physiological role of the calcium dependency of the amino acid sensing is still unclear. It is known that calcium affects biofilm formation, cell adhesion, and other virulence activities in V. cholerae as well as other Vibrio species (16–21).
We compared the amino acid sequence of Mlp24 with those of its homologues around the N-terminal region of the UJ loop. Interestingly, the residues important for Ca2+ binding (E109 and E115, which coordinate to Ca2+, and G113, which is important to form the turn structure) are conserved only in putative MLPs in marine bacteria, such as the products of VPA0511 of Vibrio parahaemolyticus and N646_4651 of Vibrio alginolyticus (Fig. 4A). As described above, seawater contains 10 mM Ca2+. The MLPs of marine bacteria may bind Ca2+ to form a specific loop structure, like Mlp24.
The acetate-binding and l-amino acid-binding structures of Mlp24p indicate that l-amino acid binding induces the domain closure of CAD1. Similar conformational changes caused by ligand binding have been observed in other dCACHE-type sensors, such as Tlp3 (14) and sensor kinase vpHK1s-Z8 (22). On the basis of the structures of Mlp24p, we propose a plausible mechanism of ligand binding by Mlp24. The ligand amino acid enters the pocket of CAD1 in the open conformation. The carboxy group of the ligand is bound to R125 in H4, which is the first step of ligand biding, and the interaction is further stabilized by Y120 and W127. This idea is supported by the acetate binding to CAD1 of the subunit in the open conformation. The amino group of the ligand backbone then interacts with D172 in S6 of the central β-sheet to adjust the orientation of the ligand. The positive charge of the ligand amino group induces the conformational change of the LJ segment to bind to Y143 and D145. Finally, the entrance of the pocket is closed, and the ligand side chain atoms are recognized by the residues in the pocket. Essentially the same mechanism may be exerted by Mlp37.
The ligand-binding signal should be transmitted into the cytoplasmic domain, but the mechanism is still under debate. Structural studies of the ligand-free and -binding structures of various dCACHE-type receptors have proposed possible structural changes involved in transmembrane signaling. Piston-like movements along the N-terminal helix have been proposed to be important for the signal transduction from the structures of Z2-Bistris complex and apo-Z3 (22). However, such movement was not observed in the acetate-binding and l-amino acid-binding structures of Mlp24p. The relative movement of the two CACHE domains caused by the bending of H1 is proposed to be another candidate involved in the signal transduction (22). However, the H1 bending in Mlp24p occurs independent of ligand binding. The Mlp24p, the l-arginine complex, and the l-asparagine complex structures show almost the same bending conformation in H1 and the same relative positions of the two CACHE domains. In contrast, other ligand-binding structures, such as glycine, l-proline, and l-serine complexes, show different relative positions of the two CACHE domains and different H1 conformations. Therefore, ligand binding to Mlp24p does not necessarily induce the movement of the CACHE domain. However, the structure of Mlp24p crystallized in solution without l-amino acids forms a heterodimer with a subunit that binds l-alanine in CAD1. Thus, the heterodimer can be regarded as a ligand-bound state. A recent study on Tlp3 (14) proposed that ligand binding to the distal domain induces opening of the proximal domain followed by the displacement of the C-terminal helix toward the membrane, which may cause a piston-like movement of the TM helix. This is, however, not the case in Mlp24. Mlp24p does not show any structural changes in CAD2 and the C-terminal helix by binding of ligands. No significant structural difference was observed between the acetate-binding Mlp24p and the l-arginine or l-asparagine complex except for the LJ segment of CAD1. In contrast, the ligand complex structures show large conformational variation in the dimer structures. This structural variation may be artifact caused by the crystal packing or the lack of membrane tethering. To elucidate the signal transduction mechanism across the membrane, we need both ligand-free and ligand-binding structures of the full-length protein or a larger fragment containing the membrane-spanning region.
MATERIALS AND METHODS
Bacterial strains and plasmids.The classical biotype strain O395N1 is wild type for chemotaxis (Che+), and its derivative strain Vmlp201 (Δmlp24 Δmlp37) lacks mlp24 and mlp37 (6). The E. coli strain HCB436 (23) lacks the methylesterase CheB and the methyltransferase CheR as well as the chemoreceptors. The E. coli B strain BL21(DE3) [F− ompT hsdS(rB− mB−) dcm gal λ(DE3)] (24) was used for expression of proteins. Plasmids pMlp24, encoding full-length Mlp24, and pVCP24, encoding the entire periplasmic fragment (residues 29 to 274) of Mlp24 (6), were used for in vivo functional assay and crystallization, respectively. The Mlp24 periplasmic fragment expressed from pVCP24 has a His6 tag at its C terminus. Plasmid pVCP24NH, encoding the Mlp24 fragment without a His6 tag, was constructed as follows. The DNA fragment encoding the entire periplasmic domain (residues 29 to 274) was amplified by standard PCR with primers GST-Mlp24PF (CGCGGATCCGTTCGTGAGGAGATTGAGTCACTCG) and GST-Mlp24RR (CCGGAATTCCTAAATCAATGAGCTGTGGCGTAAATCATCC) using pVCP24 as a template. The amplified fragment was then digested with EcoRI and BamHI and was ligated with the expression vector pGEX-6p-2 (GE Healthcare) digested with the same set of restriction enzymes. The cloned fragment was confirmed by standard DNA sequencing.
Expression and purification of the periplasmic fragments of Mlp24p and Mlp37p.Mlp37p was expressed and purified as described previously (7). Mlp24p was purified by a method similar to that used for the Mlp37p purification. Strain BL21(DE3) (Novagen) carrying pVCP24 was cultured in LB broth (Lennox) (Nacalai Tesque, Inc., Kyoto, Japan) containing 50 μg/ml of ampicillin at 37°C until the cell density had reached an optical density at 600 nm (OD600) of about 0.8. Isopropyl-β-d-1-thiogalactopyranoside (IPTG) was then added to a final concentration of 0.1 mM to induce protein expression, and the culture was continued for 12 h at 20°C. Cells were harvested by centrifugation, suspended in phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4·12H2O, and 1.8 mM KH2PO4 at pH 7.4), and disrupted by sonication. After removing cell debris by centrifugation, the cell lysate was loaded on a glutathione-Sepharose 4B column (GE Healthcare), followed by washing with PBS. Proteins were eluted with 50 mM Tris-HCl (pH 8.0) buffer containing 10 mM reduced glutathione. The N-terminal glutathione S-transferase (GST) tag was then cleaved using PreScission protease (GE Healthcare), and the reactant was dialyzed against 1.0 liters of dialysis buffer (50 mM Tris-HCl [pH 7.0], 150 mM NaCl, 1 mM dithiothreitol [DTT], and 1 mM EDTA) at 4°C for 12 h using a Spectra/Por dialysis membrane (molecular weight cutoff [MWCO], 6,000 to 8,000; Spectrum Laboratories, Inc.). The protein solution was loaded again onto a glutathione-Sepharose 4B column (GE Healthcare) to remove GST and unreacted protein and was further purified by size exclusion chromatography with a High Load 26/60 Superdex 200 column in 20 mM Tris-HCl and 150 mM NaCl (pH 8.0). The peak fraction was collected and concentrated to 10 mg/ml. The purity of the purified proteins was examined by SDS-PAGE. GST-fused Mlp24pNH (Mlp24p without the His6 tag) used for ITC analysis was expressed in HCB436 (23). Purification of GST-fused Mlp24pNH was carried out as described previously (6).
Crystallization.Initial crystallization screening was performed by the sitting-drop vapor diffusion technique. Crystallization drops were prepared by mixing 1 μl of protein solution (10 mg/ml) with 1 μl of reservoir solution. The ligand amino acids were added in the protein solution before preparing the drop to a final concentration of 10 mM except for l-arginine for Mlp24p (100 mM). CaCl2 was also added in the protein solution to a final concentration of 10 mM for crystallization of the Mlp24p-glycine and Mlp24p–l-serine complexes, because calcium binding to the UJ loop had been revealed before initial screening of these complexes. Initial screening was carried out using the Wizard Classic I and II (Rigaku Reagents, Inc.), Wizard Cryo I and II (Rigaku Reagents, Inc.), and Crystal Screen I and II (Hampton Research) screening kits. The crystallization conditions were optimized by the hanging-drop vapor diffusion method. Crystallization drops were prepared by mixing 1.5 μl of protein solution with an equal volume of reservoir solution. All crystals appeared within 3 days after optimization. Their final crystallization conditions, space groups, and unit cell dimensions are summarized in Tables 1 and 2.
X-ray data collection and structure determination.X-ray diffraction data were collected at synchrotron beamline BL41XU in SPring-8 (Harima, Japan), with the approval of the Japan Synchrotron Radiation Research Institute (JASRI) (proposal no. 2012B1462, 2013A1406, 2015A1097, and 2015B2097). The crystals of Mlp24p-glycine and Mlp24p–l-serine complexes were soaked in a 1:9 mixture of 2-methyl-2,4-pentanediol (MPD) and the reservoir solution for a few seconds before transfer into liquid nitrogen for freezing. All X-ray diffraction data were measured at 95 K under nitrogen gas flow. The diffraction data were processed with MOSFLM (25) or XDS (26) and scaled with SCALA (27), Aimless (27), or XDS (26). The initial phases were determined by molecular replacement (MR) with Phenix (28). A polyalanine model derived from the Mlp37p–l-alanine complex structure (PDB no. 3C8C) was divided into four regions (positions 38 to 55, 59 to 73, 77 to 174, and 178 to 258) and was used as a separate search model for the phasing of Mlp24p. Molecular replacement for the other Mlp24p crystal data was performed using the Mlp24p structure as a search model. The atomic models were constructed with Coot (29) and refined with Phenix (28). Data collection and refinement statistics are summarized in Table 2.
Capillary assay.Chemotactic ability was examined by a capillary assay as described previously (6, 7, 30). In brief, an overnight culture of V. cholerae cells grown in TG medium (1% tryptone, 0.5% NaCl, 0.5% glycerol) at 30°C was diluted 1:30 into fresh TG medium, shaken at 30°C for 6 h, harvested, and washed with TM buffer (50 mM Tris-HCl [pH 7.4], 5 mM glucose, 5 mM MgCl2). Cells were resuspended in TMN buffer (50 mM Tris-HCl [pH 7.4], 5 mM glucose, 100 mM NaCl, 5 mM MgCl2) with or without 10 mM CaCl2 (to an OD600 of 0.1). After preincubation of cells at 30°C for 1 h, a capillary containing an amino acid solution was inserted into the cell suspension and incubated for another 1 h. The number of bacteria in the capillary was estimated by plating serial dilutions on LB agar.
ITC.Isothermal titration calorimetry (ITC) measurements were performed using a VP-ITC microcalorimeter (MicroCal Inc., Northampton, MA) essentially as described previously (4, 6, 7) except that the purified fragment was dialyzed against “EDTA-free” cleavage buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl) with or without 10 mM CaCl2 at 4°C for 12 h.
Data availability.Atomic coordinates and structure factors have been deposited in the Protein Data Bank with accession codes of 6IOP (Mlp24p), 6IOQ (Mlp24p-glycine complex), 6IOR (Mlp24p–l-asparagine complex), 6IOS (Mlp24p–l-proline complex), 6IOT (Mlp24p–l-arginine complex), and 6IOU (Mlp24p–l-serine complex).
ACKNOWLEDGMENTS
We thank the beamline staff of SPring-8 for technical help with data collection and S. Kobashi and M. Kinoshita for technical support for purification of proteins.
This work was supported in part by JSPS KAKENHI grant numbers 15H02386 and 23370051 (to K.I.), 17J02169 (to Y.T.), 17K08842 (to S.N.), and 22390086 and 17KT0026 (to I.K.) and by MEXT KAKENHI grant number 23115008 (to K.I.).
FOOTNOTES
- Received 3 January 2019.
- Accepted 1 February 2019.
- Accepted manuscript posted online 11 February 2019.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00779-18.
- Copyright © 2019 American Society for Microbiology.
