(Received for publication, August 23, 1996, and in revised form, December 11, 1996)
From the Department of Microbiology and Immunology, University of Illinois, Chicago, Illinois 60612
Position 106 in CheY is highly conserved as an aromatic residue in the response regulator superfamily. In the structure of the wild-type, apo-CheY, Tyr106 is a rotamer whose electron density is observed in both the inside and the outside positions. In the structure of the T87I mutant of CheY, the threonine to isoleucine change at position 87 causes the side chain of Tyr106 to be exclusively restricted to the outside position. In this report we demonstrate that the T87I mutation causes cells to be smooth swimming and non-chemotactic. We also show that another CheY mutant, Y106W, causes cells to be more tumbly than wild-type CheY, and impairs chemotaxis. In the structure of Y106W, the side chain of Trp106 stays exclusively in the inside position. Furthermore, a T87I/Y106W double mutant, which confers the same phenotype as T87I, restricts the side chain of Trp106 to the outside position. The results from these behavioral and structural studies indicate that the rotameric nature of the Tyr106 residue is involved in activation of the CheY molecule. Specifically, CheY's signaling ability correlates with the conformational heterogeneity of the Tyr106 side chain. Our data also suggest that these mutations affect the signal at an event subsequent to phosphorylation.
How regulatory proteins are involved in signal transduction events is a fundamental question for both prokaryotic and eukaryotic intracellular signaling pathways. In most microorganisms, cells adaptively respond to a wide variety of environmental stimuli through "two-component" regulatory systems. During chemotaxis in Escherichia coli, the sensor, CheA, receives signals from transmembrane chemoreceptors with the help of CheW (1-4), and then transfers the signal to the response regulator, CheY, by transient phosphorylation (5, 6). Phosphorylated CheY (CheY-P) interacts with the switch motor to reverse the direction of flagella rotation from counterclockwise (CCW)1 to clockwise (CW). This causes the bacteria to tumble rather than swim smoothly (7, 8). CheY-P is deposphorylated by its autophosphatase activity, a reaction enhanced by CheZ (5, 6). CheY's return to its resting form results in normal CCW flagella rotation and smooth swimming of the bacteria.
More than 80 different two-component systems have been reported in bacterial and yeast signal transduction pathways, regulating many different activities such as chemotaxis, osmoregulation, sporulation, and virulence. All response regulators in this superfamily are homologous to CheY. CheY is the only member of the response regulator superfamily for which a detailed three-dimensional structure is known (9-11). Both genetic and structural studies demonstrate that the highly conserved residues in response regulators, such as Asp12, Asp13, Asp57, Thr87, and Lys109, play important roles in CheY activation (12-17). Mutagenesis studies indicate that substitutions of any one of those residues cause functional defects in CheY (13, 16, 17).
Position 106 is another conserved site in the CheY superfamily; this
position is occupied by an aromatic residue (tyrosine or phenylalanine)
greater than 80% of the time (9). When residue 106 is substituted with
a non-aromatic residue, the mutant CheY loses its signaling ability
(18). The three-dimensional structure of wild-type CheY
(Mg2+-free form) shows that Tyr106 is a rotamer
whose electron density is found both in an inside (solvent
inaccessible) and an outside (solvent exposed) position (10). When the
tyrosine side chain is in the inside, the hydroxyl group forms a
hydrogen bond with the 0 of the conserved
Thr87 residue through one intervening solvent molecule. In
the structure of the T87I mutant of CheY, the Tyr106 side
chain is well ordered, exclusively occupying the external position,
forced out by the increased bulk and hydrophobicity of the isoleucine
at position 87 (17). The functional defect caused by the T87I mutation
(17) correlates with this structural change.
We report here the behavioral effects, biochemical properties, and crystal structures of two very different CheY mutants. The first mutant has a tyrosine to tryptophan substitution at position 106 (Y106W). The second mutant bears a double mutation: T87I and Y106W (called T87I/Y106W). The Y106W mutant exhibits a hyperactive phenotype, while the T87I/Y106W double mutant causes a loss of activity. Both mutants are phosphorylatable in vitro. Structurally, the only significant differences between these molecules and wild-type CheY are the conformations of the Trp106 side chain. In Y106W, Trp106 is located in the inside position only, while in T87I/Y106W, Trp106 is forced to the outside position by the bulk of isoleucine at position 87, similar to that previously seen in the T87I mutant. Thus, results from our combined behavioral and structural studies support our proposal that movement of the side chain of residue 106 modulates the activation state of CheY.
Bacterial strains and plasmids are listed in Table I. pXYZ301, containing the CheY double mutant T87I/Y106W, was constructed by recombining appropriate restriction fragments containing the CheY mutations using standard techniques (19), then confirmed by DNA sequencing.
|
The plasmid-borne mutations were transferred into the host chromosome
of E. coli by gene conversion to obtain single copy mutations (20, 21). The EcoRI-SalI fragments of
either pYM31(T87I) or pXYZ301 (T87I/Y106W) containing cheB
and one third of cheY were replaced with pMC100
EcoRI-SalI fragments to extend to cheA upstream of cheY. The resulting plasmids pXYZ112 (containing
T87I) and pXYZ321 (containing T87I/Y106W), both carrying
cheAW tar tap cheRBYZ
, were used as
donors to cross the cheY alleles into the chromosome of D345
(cheB::rpsL::KanR::cheY377)
to create strain XYZ7 with T87I and strain XYZ9 with T87I/Y106W on
their chromosomes.
Cells were grown in tryptone broth at 30 °C to early post-exponential phase for all following behavioral assays as described previously (22). A spectrophotometric assay for measuring cell chemotactic ability was used (23), where a semi-solid chemoattractant was placed inside a cuvette containing a uniform cell suspension, and the increase in cell density near the chemoattractant was measured as a function of time. Bacterial tethering experiments were performed as described by Wang et al. (24).
Protein Phosphorylation AssaysPhosphotransfer reactions
from CheA to CheY were carried out as published (5, 17). The stability
of phosphorylated CheY was analyzed as described previously (17, 18),
in which the kinase protein CheA was coupled to Sepharose beads and
phosphorylated by [-32P]ATP in the presence of 0.5 mM Mg2+. CheY was then added to the CheA beads
to allow the phosphoryl-group transfer reaction to occur. Samples of
phosphorylated CheY were aliquoted at various times, where the
autodephosphorylation reaction was quenched by 2 × SDS-polyacrylamide gel electrophoresis sample buffer. Samples were
loaded directly onto SDS-polyacrylamide gels (15% polyacrylamide).
Radiolabeled proteins were visualized by autoradiography, and
radioactivity in protein bands was determined with an Ambis
-scanning system.
The Y106W and T87I/Y106W mutant CheY proteins were purified as described by Matsumura et al. (25). The purified mutant proteins were concentrated to 20 mg/ml with a Centriplus concentrator (Amicon) for crystallization trials. A screen method for crystallization (26) was used to obtain crystals. Fifty different conditions were screened using the hanging-drop method at 4 °C. Each droplet contained 2 µl of protein solution and 2 µl of reservoir solution. The Y106W crystal used for data collection was from a solution consisting of 28% polyethylene glycol 8000, 0.12 M calcium acetate, and 0.1 M cacodylate buffer at pH 6.0, with dimensions of 0.34 × 0.09 × 0.08 mm. The T87I/Y106W crystal was from a solution of 18% polyethylene glycol 3350 and 0.2 M ammonium sulfate, with dimensions of 0.2 × 0.05 × 0.05 mm.
Crystallographic Data Collection and ProcessingThe crystal
form of the Y106W mutant was the same as that of wild-type CheY (10),
being needles of the orthorhombic space group
P212121, with 1 molecule/asymmetric
unit. Diffraction data were collected from a single crystal at the area
detector facility at Abbott Laboratories. The crystal was exposed for a
total of 48 h in graphite monochromatized CuK
radiation from a rotating anode generator passing through a 0.3-mm
collimator. Data were recorded in two scans on an RAXIS II imaging
plate detector at a crystal to detector distance of 75 mm, with a swing
angle of 26.57° in 2
. Unit cell dimensions were a = 45.55 Å, b = 46.55 Å, and c = 53.33 Å, which is less than 1% different than the unit cell dimensions of
wild-type, apo-CheY. The final data set included a total 21,022 observations (I/
I > 1) of 7,433 unique
reflections, 92% complete to 2.00 Å resolution, with an overall
unweighted R-factor on intensities of 6.6% (Table
II).
|
Crystals of the T87I/Y106W mutant were also orthorhombic needles, but
of the space group P21212 with 2 molecules/asymmetric unit. Diffraction data were collected from a
single crystal at the area detector facility at the experimental
station X8C of the National Synchrotron Light Source, Brookhaven. The
crystal was flash frozen in liquid nitrogen, maintained near liquid
nitrogen temperature, and exposed for three scans within a total of
~6.5 h exposure of synchrotron radiation of 1.033 Å wavelength,
passing through a 0.3-mm collimator. Data were recorded at a
crystal to detector distance of 156 mm on a 2 × 2 CCD detector
array, with a swing angle of 5.0° in 2 for a theoretical upper
resolution limit of 2.30 Å. The unit cell dimensions were
a = 95.78 Å, b = 76.70 Å, and
c = 32.64 Å. The final data set included a total 28,137 observations of 7,861 unique reflections, 73% complete to 2.33 Å resolution, with an overall unweighted R-factor on
intensities of 5.1% (Table III).
|
Crystals of the Y106W mutant CheY were isomorphous with those of wild-type, apo-CheY, so phasing of the structure was initiated with a phase calculation using the wild-type structure with all solvent molecules excluded, and residue 106 modeled as an alanine.
Phasing of the T87I/Y106W double mutant CheY structure began with the
results of molecular replacement calculations (27) using the software
package MERLOT (28). The search model was constructed from the high
resolution, refined wild-type structure, where all solvent molecules
and the side chain atoms of polar, solvent-accessible residues were
excluded, and residues 87 and 106 were changed to alanines. A number of
resolution ranges and I/I cutoffs were
explored. The rotation function searches consistently gave one strong
peak, accompanied by a variety of minor peaks. A systematic
cross-translation function test of the strong rotation function peak
combined with each of the minor peaks finally yielded two convincing
solutions. These solutions were confirmed with packing analyses and a
polar rotation function calculation. R-value minimization
refinement of the rotation and translation parameters gave a 2-molecule
model with an R-factor at 39%.
The Y106W CheY mutant structure
refined without difficulty using the restrained least-squares method
(29), combined with 23 instances of manual intervention and partial
rebuilding with omit maps. The final R-factor was 18.5% for
the 7,156 reflections in the 10-2.0-Å range with intensities greater
than 2I (Table IV).
|
Refinement of the T87I/Y106W double mutant CheY was more problematic
due the poor molecular replacement model and the underdetermined ratio
of observations/variables of 0.90. Refinement was done using both the
conventional restrained least-squares method and the simulated
annealing method (30), interspersed with 37 rounds of partial rebuilds
based on omit maps. The final R-factor was 18.6% for the
7,698 reflections in the 10-2.3 Å range with intensities greater than
2I (Table V).
|
Calculations of electron density maps and other data processing were
done with the XTAL (31), CCP4 (32), and X-PLOR (33) software packages,
and several locally developed programs. Refinement was done with the
software packages PROTIN and PROFFT (29, 34) and X-PLOR. Display of
electron density maps and model rebuilding were done using the graphics
package QUANTA (35) on a Silicon Graphics Indigo2
workstation. All water molecules were built manually, using the following rules: 1) electron density must appear in both the
|Fo Fc| and
|2Fo
Fc| maps, 2) there must be proper hydrogen bonding distances and geometries, and 3)
molecules that refine to unreasonable temperature factors must be
removed. The presence of a SO42
ion in
the T87I/Y106W structure was indicated in the final rounds of
refinement by electron density greater than any other solvent density,
located in a positively charged environment, with hydrogen bond
distances too great for a water molecule.
All
strains were constructed in single copy to ensure wild-type levels of
expression. The resulting strains were XYZ7 (for T87I), XYZ3 (for
Y106W), and XYZ9 (for T87I/Y106W). The wild-type strain XYZ6 had the
same genotype as the above mutant strains except for its wild-type
cheY gene to serve as a control for wild-type chemotaxis. A
spectrophotometric assay was used to detect chemotactic ability for the
wild-type and all three mutant CheY strains. As shown in Fig.
1, the Y106W mutant showed about 80% chemotaxis of
wild-type strain XYZ6, while neither the T87I nor the T87I/Y106W mutant
had any chemotactic ability.
In wild-type cells, phosphorylated CheY interacts with the flagellar
motor to change the direction of rotation from CCW to CW and cause the
cells to make a tumbly motion. After phosphorylated CheY returns to its
resting state, the cells change back to a CCW flagellar rotation,
resulting in smooth swimming. Thus, normal chemotaxis behavior depends
on the balance of cell tumbling and smooth swimming. Since chemotaxis
is impaired by either excessive tumbling or excessive smooth swimming,
mutations in cheY can cause cells to be dominant tumbly (13,
18) or dominant smooth swimming (17, 18). To determine if the
reductions of chemotaxis in the cheY mutants were because of
extra tumbles or extra smooth swimming, the flagellar rotations of the
mutant strains were counted using the bacterial tethering method. In
comparison with the wild-type strain XYZ6, the Y106W mutant showed
increased CW-biased flagellar rotation, while both the T87I and the
T87I/Y106W mutants had extreme CCW rotation (Fig. 2).
These results show that Y106W is an activated mutant, which increases
cell tumble frequency and diminishes cell chemotaxis. On the other
hand, T87I and T87I/Y106W are non-activated mutants, which result in
smooth swimming cells with a non-chemotactic phenotype.
In Vitro Phosphorylation Biochemistry of the Mutants
CheY is
phosphorylated by CheA plus ATP, and dephosphorylated by its
autophosphatase activity and by CheZ (5, 6). A mutation in CheY that
abolishes its ability to be phosphorylated (e.g. D57C)
results in smooth swimming cells that are non-chemotactic (14, 15). In
order to determine whether the smooth-swimming phenotypes of the T87I
and T87I/Y106W mutants are due to defects in their phosphoryl-group
transfer abilities, the mutant molecules were combined with CheA and
ATP in vitro. Both T87I and T87I/Y106W were found to be
phosphorylatable to a level of wild-type CheY in vitro, as
well as the Y106W mutant (data not shown). Next, the stabilities of the
phosphorylated forms of the wild-type, T87I, Y106W, and T87I/Y106W
mutant CheYs were compared. Phosphorylated Y106W was 2 times more
stable than wild-type CheY-P, while phosphorylated T87I and T87I/Y106W
were 5 times more stable (Fig. 3). These results indicate different levels of autophosphatase activities of these mutants. Note, however, that the variation in the mutants'
phosphorylation half-lives correlates negatively with their signaling
abilities, so it is not a factor in this structure/function
analysis.
Description of Structural Results
The final electron density
maps for the refined structures of Y106W and T87I/Y106W were clear and
well defined for the entire backbones of all molecules. All side chain
atoms in the Y106W structure were located within interpretable electron
density, and only the side chains Lys91 of molecule A and
Arg19 of molecule B in the T87I/Y106W structure were not
identifiable. The final coordinate set for the Y106W mutant consists of
1,068 atoms, 83 of which are solvent, with one Ca2+ ion.
The T87I/Y106W mutant has 1,951 atoms for the two protein molecules in
the asymmetric unit, 177 solvent molecules, and one SO42 ion.
The crystal packing of the Y106W structure is the same as wild-type,
apo-CheY, the two being essentially isomorphous. This is surprising
given that the crystallization conditions were completely different
(calcium acetate/polyethylene glycol 8000 versus ammonium sulfate). For the T87I/Y106W structure, the molecules pack as dimers in
the asymmetric unit, with an almost perfect non-crystallographic two-fold axis parallel to the ab plane. The two molecules
superimpose with an r.m.s. deviation of 0.42 Å. The local dyad
involves close associations between hydrophobic residues of the
4 helices, plus extensive solvent-mediated interactions
of side chains from the
4-
5-
5 surfaces of the
molecules.
A least-squares superposition of -carbons of the Y106W mutant and
wild-type CheY structures yields an r.m.s. agreement of 0.33 Å. Thus,
their backbone conformations are the same, within the limits of error.
Comparisons of the T87I/Y106W molecules with wild-type require an
analysis similar to that for the previously reported T87I mutant
structure (17). The T87I molecules showed distinct backbone
conformational changes in the
4-
4 loop
that were directly attributed to that substitution (17). Those same localized changes also appear in the T87I/Y106W double mutant structure. Note that these changes do not occur in the Y106W mutant. This confirms the original assertion concerning the effects of the
threonine to isoleucine substitution at position 87, and provides further evidence that the Y106W mutation by itself does not introduce measurable backbone conformational changes. Therefore, as with the T87I
to wild-type comparison, the T87I/Y106W molecules can be divided into
NH2-terminal segments (residues 2-86) and
CO2-terminal segments (residues 87-129). The r.m.s.
deviations of
-carbon atoms from wild-type for the first segment are
0.42 Å for molecule A and 0.47 Å for molecule B. These differences
are not significantly greater than the 0.42-Å r.m.s. deviations from
the full-length comparison of T87I/Y106W molecule A versus
molecule B. Because of the relatively large (up to 2 Å) localized
changes in the
4-
4 loops, the
carboxyl-terminal segments cannot be least-square superimposed with any
validity.
One prominent feature of the structure of the Y106W mutant was the presence of a Ca2+ ion in the Mg2+ ion binding site (the crystallization conditions included 0.12 M calcium acetate). The Ca2+ ion has an octahedral coordination sphere similar to the Mg2+ ions in the two previously reported CheY·Mg2+ structures (36, 37). Its presence does not influence the conformation in the region of residue 106, which is more than 11 Å away. The active site structures of the two T87I/Y106W double mutant molecules are similar to wild-type, apo-CheY. There are minor differences in side chain conformations of Asp13, Phe14, and Asn59, due to intermolecular contacts. A detailed discussion of solvent structure is not warranted at these resolution limits.
Structural Differences at the Mutation SitesThe most obvious
and unambiguous results of these two structure determinations are the
repositionings of the side chain of residue 106 relative to wild-type
CheY (Fig. 4). In the Y106W mutant CheY, the tryptophan
side chain at position 106 is clearly in the g+
position (1 =
69°), neatly packed in a hydrophobic
cavity lined by residues Trp58, Met85,
Thr87, Ile95, and Ala98.
Restriction of the tryptophan ring to the internal cavity appears to be
driven by these hydrophobic interactions. There are van der Waals
contacts between the C
and C
atoms of
Thr87 and the six-membered ring of Trp106,
which appear to stabilize the inside position, but there are no obvious
steric barriers that would prevent the tryptophan ring from assuming
the external position in this crystal packing. In contrast, the
T87I/Y106W mutant shows the tryptophan side chain in the highly solvent
exposed g
position (
1 = 76°
and 79°) for both molecules of the asymmetric unit. Clearly the
presence of the isoleucine side chain at position 87 sterically
prevents the tryptophan ring from entering the cavity. The electron
density for the Ile87 side chains shows them to have the
same conformations as seen in the previously reported T87I mutant
structures, and the C
atom of Ile87 has
slight van der Waals contacts with the ring of Trp106 as it
occupies the outside position.
Our behavioral studies demonstrate that both the Y106W and T87I CheY mutations impair cell chemotaxis, but they cause the functional defect in opposite ways. Y106W is an activating mutant, which causes the cells to be more tumbly, while the T87I mutant loses signaling ability, resulting in completely smooth swimming.
Our previous structural reports showed that in wild-type CheY the side chain of Tyr106 was observed in both the inside and the outside positions (Ref. 10; Fig. 4a), and also that the side chain of Tyr106 in the T87I mutant was forced to the outside position by the bulk of the isoleucine side chain at position 87 (Ref. 17; Fig. 4b). Here we show that the crystal structure of Y106W has the same overall structure as wild-type CheY, except for the side chain of residue 106. The tryptophan side chain of the Y106W mutant was found buried in the internal cavity, apparently driven by hydrophobic forces, and assisted by local rearrangements (Fig. 4c). The functional relationship between residues Thr87 and Tyr106 suggested by those results led us to design the T87I/Y106W double mutant. As expected, the T87I/Y106W mutant is non-chemotactic with a smooth swimming phenotype like the T87I mutant. The side chain of residue 106 in T87I/Y106W is restricted to the outside position (Fig. 4d), as in T87I (Fig. 4b).
These combined results from both behavioral and structural studies suggest that the signaling state of the CheY molecule correlates with the nature of the side chain of Tyr106. The simplest model would be that the "in" position is the activated form, whereas the "out" position is the inactive form. But it may be the dynamic act of switching between the two conformations that is important in CheY activation. We propose that wild-type CheY can alter its signaling state by freely rotating the side chain of Tyr106 between inside and outside position, resulting in the optimum chemotaxis activity for the cells. Presumably, the position of the Tyr106 side chain would be modulated by the phosphorylation state of the molecule. In the Y106W mutant, the side chain of Trp106 could rotate to the external environment, but instead remains in the internal cavity, resulting in hyperactive signaling and impaired chemotaxis. In contrast, for both the T87I and T87I/Y106W mutants, this cavity is partially filled by the bulk of the Ile87 side chain, which forces residue 106 exclusively to the outside position, resulting in total loss of chemotaxis.
Very few activation mechanisms based on rotameric rearrangements have been proposed for other proteins. Some examples involving catalytic mechanisms include ribonuclease A, which has a mobile histidine residue in the active site whose conformation is modulated by pH (38, 39). A similar case of pH-dependent conformational mobility of a histidine in the active site of carbonic anhydrase II has been reported (40). An induced-fit type of conformational change occurs in ribonuclease T1 upon binding of guanosine to the recognition site (41). The importance of mobile aromatic residues in antibody-antigen recognition has also been discussed (42). Aromatic residues (tyrosine and tryptophan) have high occurrence in the antigen binding sites of antibodies, especially in the heavy chain. Movements of their side chains may play key roles in the variation of antibody-antigen recognition. One explanation for the specific use of aromatic residues is that their large rotatable volumes affect variation in the surface contours of the antigen binding site without alterations in backbone conformation. In addition, their amphiphilicity allows them to form hydrogen bonds, hydrophobic interactions, and electrostatic interactions with other residues, and contact easily with other molecules (42). Thus, the special role of aromatic residues may be a central component in protein-protein recognition, and CheY provides the best documented example.
Position 106 is occupied by an aromatic residue (tyrosine or phenylalanine) in more than 80% of known response regulators (9), which implies that an aromatic amino acid at this position is required for normal function of CheY. Indeed, our earlier paper (18) shows that when position 106 is substituted by a non-aromatic residue (such as Gly, Val, Leu, Ile, and Cys), the mutant molecules lose their signaling, and the cells become smooth-swimming and non-chemotactic. But when Tyr106 is substituted by either Trp or Phe, the mutants retain chemotaxis, although at a reduced level (18).
Involvement of the 106 Rotamer in CheY Activation Is a Post-phosphorylation EventAll three mutants presented here can be easily phosphorylated by CheA, which indicates that neither Thr87 nor Tyr106 are required for phosphorylation. Our data do suggest that Thr87 might be involved in the CheY phosphatase activity, since both the phosphorylated T87I and T87I/Y106W mutants show stability 5 times that of wild-type CheY-P. The Y106W mutant shows a smaller (2-fold) decrease in its phosphatase activity. This is consistent with the CheY structure: Thr87 is closer to the Asp57 phosphorylation site than Tyr106, and Thr87 connects with Asp57 through two bridging solvent molecules, while Tyr106 is away from the phosphorylation site, closer to the signaling surface defined by suppressor mutations (43, 44).
It is unlikely that the phosphatase defects in the T87I and T87I/Y106W molecules contribute to the loss of chemotaxis of these mutants. In the signaling flow of wild-type CheY, CheY is phosphorylated by CheA-P, and then CheY-P interacts with the switch motor to generate a tumble motion. Both the T87I and T87I/Y106W mutants have normal phosphorylation activities, but reduced dephosphorylation activities. One would expect a higher concentration of phosphorylated mutant CheY molecules in the cell, which would cause the cell to be more tumbly than wild-type cells. However, our behavioral data show that both mutants are exclusively smooth swimming. This implies that the local structural change of the side chain of residue 106 blocks the signal from phosphorylated CheY. In other words, the defect in signaling ability in these mutants is subsequent to the primary activation event of phosphorylation.
Our earlier paper (18) demonstrates that Tyr106 is not directly involved in switch binding, since mutants with different substitutions at position 106 bind to FliM as well as wild-type CheY. We could not measure the binding affinity of the phosphorylated T87I and T87I/Y106W mutants to FliM, since both mutants could not be phosphorylated by acetyl phosphate (Refs. 18 and 44; acetyl phosphate is the phosphorylating agent used in our in vitro binding assay for CheY-P·FliM association). However, the unphosphorylated forms of T87I and T87I/Y106W do bind to FliM as strongly as the unphosphorylated form of wild-type CheY (data not shown). Thus it is very likely that for wild-type, the conformational heterogeneity of residue 106 does not affect CheY-P's binding affinity for FliM, but acts to propagate the signal after CheY-P·FliM association. This is supported by the observations that no suppressors of fliG or fliM mutations map to positions 87 or 106 in CheY (43, 44). Further supporting evidence is that the non-phosphorylatable but constitutively active CheY mutant D13K generates a dominant tumbly signal without increased binding to FliM (45).
The atomic coordinates and structure factors (pdb5chy.ent, r5chy.ss, pdb6chy.ent, and r6chy.ss) have been deposited in the Protein Data Bank, Brookhaven National Laboratory, Upton, NY.
We are indebted to Dr. Cele Abad Zapetero at Abbott Laboratories and to the staff of experimental station X8C of the National Synchrotron Light Source, Brookhaven National Laboratory for assistance in collection of diffraction data. We also thank Dr. Frederick W. Dahlquist for generously providing strains and constructive suggestions.