Structure and Mechanism of Action of Sda, an Inhibitor of
the Histidine Kinases that Regulate Initiation of
Sporulation in Bacillus subtilis
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Rowland, Susan L. et al. “Structure and Mechanism of Action of
Sda, an Inhibitor of the Histidine Kinases That Regulate Initiation
of Sporulation in Bacillus Subtilis.” Molecular Cell 13.5 (2004):
689–701. Copyright © 2004 Cell Press
As Published
http://dx.doi.org/10.1016/S1097-2765(04)00084-X
Publisher
Elsevier
Version
Final published version
Accessed
Wed May 25 19:02:26 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/83848
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Molecular Cell, Vol. 13, 689–701, March 12, 2004, Copyright 2004 by Cell Press
Structure and Mechanism of Action of Sda,
an Inhibitor of the Histidine Kinases that Regulate
Initiation of Sporulation in Bacillus subtilis
Susan L. Rowland,1,4 William F. Burkholder,2,4,5
Katherine A. Cunningham,3 Mark W. Maciejewski,1
Alan D. Grossman,2 and Glenn F. King1,*
1
Department of Molecular, Microbial
and Structural Biology
University of Connecticut Health Center
263 Farmington Avenue
Farmington, Connecticut 06030
2
Department of Biology
Massachusetts Institute of Technology
31 Ames Street
Cambridge, Massachusetts 02139
3
Department of Biological Sciences
Stanford University
371 Serra Mall
Stanford, California 94305
Summary
Histidine kinases are used extensively in prokaryotes
to monitor and respond to changes in cellular and
environmental conditions. In Bacillus subtilis, sporulation-specific gene expression is controlled by a histidine kinase phosphorelay that culminates in phosphorylation of the Spo0A transcription factor. Sda
provides a developmental checkpoint by inhibiting this
phosphorelay in response to DNA damage and replication defects. We show that Sda acts at the first step
in the relay by inhibiting autophosphorylation of the
histidine kinase KinA. The structure of Sda, which we
determined using NMR, comprises a helical hairpin. A
cluster of conserved residues on one face of the hairpin mediates an interaction between Sda and the KinA
dimerization/phosphotransfer domain. This interaction stabilizes the KinA dimer, and the two proteins
form a stable heterotetramer. The data indicate that
Sda forms a molecular barricade that inhibits productive interaction between the catalytic and phosphotransfer domains of KinA.
Introduction
Histidine kinase signaling pathways are used extensively in bacteria to monitor and respond to signals that
communicate information about environmental conditions, stage of development, and metabolic status (Dutta
et al., 1999; Hoch, 2000; Robinson et al., 2000). They
also regulate key environmental and developmental responses in archaea, eukaryotic microoganisms, and
plants (Thomason and Kay, 2000). The cellular processes regulated by histidine kinases are diverse and
include metabolic adaptation, chemotaxis, cell cycle
progression, antibiotic resistance, and virulence.
*Correspondence: glenn@psel.uchc.edu
4
These authors contributed equally to this work.
5
Present address: Department of Biological Sciences, Stanford University, 371 Serra Mall, Stanford, California 94305.
Histidine kinase signaling pathways comprise a histidine kinase that senses the input signal and a response
regulator that determines the output response (Robinson et al., 2000). The kinase autophosphorylates at
a conserved His residue, and the response regulator
catalyzes transfer of the phosphate to its own conserved
Asp side chain, with the level of phosphorylated response regulator ultimately determining the strength of
the output response. In some cases, histidine kinase
signaling pathways consist of multiple components that
catalyze alternating His→Asp phosphotransfers. In Bacillus subtilis, the onset of sporulation-specific gene expression is tightly controlled by such a phosphorelay
(Burbulys et al., 1991). The sporulation phosphorelay
begins when histidine kinases (primarily KinA and KinB)
autophosphorylate at a conserved His. The phosphate
group is subsequently transferred to an Asp on Spo0F
and then to a His on Spo0B. In the final step, phosphate
transfer to an Asp on the transcriptional regulator Spo0A
(the response regulator) causes it to become activated.
Histidine kinase signaling pathways are often regulated by signals other than those that directly activate
the kinase (Sonenshein, 2000; Stock et al., 2000). This
allows cells to integrate disparate signals to exert fine
control over a particular cellular response. Phosphatases are commonly deployed to dephosphorylate response regulators or related Asp-containing phosphotransfer proteins (Perego and Hoch, 1996), and several
antikinases have been identified that inhibit specific histidine kinases through an interaction with their autokinase domain (Wang et al., 1997; Garnerone et al., 1999;
Pioszak et al., 2000.).
In response to defects in the initiation of DNA replication, B. subtilis prevents the onset of sporulation by
using Sda to inhibit the Spo0A phosphorelay (Burkholder
et al., 2001). KinA and KinB are the primary sources of
phosphoryl groups for this phosphorelay during sporulation (Trach and Hoch, 1993; LeDeaux et al., 1995), and
Sda exerts its effect by preventing the accumulation of
phosphorylated KinA and probably also KinB (Burkholder et al., 2001). Sda is a 46 residue polypeptide with
an amino acid sequence unlike any previously identified
histidine kinase inhibitor. In this study, we determined
the structure of Sda and examined how it regulates KinA.
We show that Sda is an antikinase that binds the KinA
phosphotransfer domain and acts as a molecular barricade that inhibits productive interaction between the
ATP binding site and the phosphorylatable His residue.
Results
Sda Binds the KinA Autokinase Domain
We used pull-down assays to delineate the region of
KinA that is bound by Sda. KinA can be considered a
three-domain protein, as depicted in Figure 1A. Residues 1–383 comprise the N-terminal sensor domain
which contains three PAS sensor modules. The C-terminal autokinase domain of KinA comprises an ATP
binding catalytic domain and a dimerization/histidine
Molecular Cell
690
Figure 1. Sda Forms a Stable Complex with
the KinA DHp Domain
(A) Cartoon of the domain architecture of
KinA.
(B) Sda (0.7 nmol) was incubated with naked
Ni2⫹-NTA beads or with beads that had been
decorated with full-length KinA autokinase
domain (His6-KinA383–606), the KinA catalytic
domain (His6-KinA456–606), or two different constructs encompassing the KinA DHp domain
(His6-KinA383–460 and His6-KinA383–465). Each reaction contained 3.1 nmol KinA, assuming all
KinA preincubated with beads was still bound
at the time of Sda addition. Beads were incubated at ambient temperature for 1 hr, and
then supernatant (lanes labeled “S”) and
bead-bound (lanes labeled “B”) fractions
were recovered and subjected to SDS-PAGE.
Gels were stained with Coomassie blue.
phosphotransfer (DHp) domain that contains the His405
autophosphorylation site (Wang et al., 2001).
We immobilized His6-KinA383–606 (comprising both the
catalytic domain and the DHp domain), His6-KinA456–606
(encompassing only the catalytic domain), and two
slightly different constructs of the DHp domain (His6KinA383–465 and His6-KinA383–460) on Ni2⫹ agarose beads and
used them as bait in pull-down experiments with Sda.
In these experiments (Figure 1B), any Sda which does
not bind KinA remains in the supernatant (S), while
bound Sda pellets with the beads (B). We found that Sda
bound to beads decorated with the complete autokinase
domain (Figure 1B, lanes 3 and 4) or either DHp domain
construct (Figure 1B, lanes 7–10), but not by naked
beads (Figure 1B, lanes 1 and 2) or beads decorated
with catalytic domain (Figure 1B, lanes 5 and 6). The
DHp domain was as effective as the entire autokinase
domain in binding Sda. Thus, Sda appears to be the first
example of an antikinase that interacts with a histidine
kinase DHp domain.
Sda Inhibits KinA Autophosphorylation
We previously demonstrated that Sda inhibits accumulation of KinAⵑP and Spo0FⵑP (Burkholder et al., 2001).
Sda could exert these effects by inhibiting KinA autophosphorylation, by acting as a phosphatase for KinAⵑP,
or by stimulating a latent autophosphatase activity of
KinA. Additionally, Sda might block phosphate transfer
from KinAⵑP to Spo0F. We performed several experiments to discriminate between these mechanistic possibilities.
Full-length KinA autophosphorylates in the presence
of ␥-32P-labeled ATP (Figure 2A, lane 1), but this accumulation of KinAⵑP was completely blocked by an ⵑ10fold molar excess of Sda (Figure 2A, lane 2). Sda also
inhibited autophosphorylation by the KinA autokinase
domain, indicating that the N-terminal sensor domain
is not essential for the Sda-KinA interaction (data not
shown). In contrast, we observed quantitative transfer of
phosphate from KinAⵑP to Spo0F when the two proteins
were mixed (Figure 2B, lane 2), and this reaction was
unaffected by Sda (Figure 2B, lane 3). We conclude
that Sda blocks KinA autophosphorylation but does not
inhibit phosphate transfer from KinAⵑP to Spo0F.
Next, we tested whether Sda stimulates dephosphorylation of KinAⵑP. We found that purified KinAⵑP was
stable for at least 30 min in both the absence and presence of a 10-fold molar excess of Sda (Figure 2C, compare lanes 2 and 3); indeed, over a period of 30 min,
we observed no release of inorganic phosphate in the
presence or absence of Sda (data not shown). This rules
out the possibility that Sda acts as a phosphatase for
KinAⵑP or that it stimulates a cryptic phosphatase activity of the autokinase domain, an activity that has been
demonstrated for some histidine kinases such as EnvZ
(Cai et al., 2003), FixL (Lois et al., 1993), and NRII/NtrB
(Pioszak et al., 2000).
Finally, we found that Sda blocked the reverse reaction (i.e., KinAⵑP ⫹ ADP → KinA ⫹ ATP). In the presence
of excess ADP, we observed robust transfer of phosphate from KinAⵑP to ADP (Figures 2C and 2D, lane 4);
this was expected, since the equilibrium for histidine
Structure and Mechanism of the Sda Antikinase
691
Figure 2. Regulation of KinA by Sda
Assays were performed using full-length
KinA, and levels of protein phosphorylation
were monitored using SDS-PAGE and autoradiography.
(A) Sda inhibits KinA autophosphorylation.
Autophosphorylation reactions contained
KinA1–606-His6 (ⵑ10 pmol), 0.5 mM ␥-32P-labeled
ATP, and the amounts of Sda indicated in
pmol above the gel. The KinA1–606-His6ⵑP
preparation contains autophosphorylated degradation products that are seen as lower molecular weight bands in the autoradiograms.
(B) Sda does not block phosphate transfer
to Spo0F. KinA1–606-His6ⵑ32P (ⵑ9 pmol) was
premixed on ice with Sda (90 pmol) or Sda
storage buffer prior to addition of kinase
buffer alone or buffer containing 150 pmol
Spo0F-His6. Quantitative transfer of 32P-phosphate to Spo0F was observed both in the
absence and presence of Sda.
(C and D) Sda inhibits the reverse reaction.
KinA1–606-His6ⵑ32P (ⵑ9 pmol) was premixed on ice with buffer A ⫾ Sda (ⵑ90 pmol). Reactions were started by addition of kinase buffer ⫾ 1
mM ADP. To monitor formation of 32P-ATP in the reverse reaction, 1 ␮l of each reaction mixture was spotted onto PEI-cellulose and developed
in 0.8 M LiCl /0.8 M acetic acid, using 32P-ATP as a standard. The histogram in (D) shows the proportion of transferred phosphate (i.e., that
released from KinA and bound to ATP) as a proportion of total phosphate (both ATP and KinA-bound) in each reaction. Each column in the
histogram corresponds to the reaction in the lane above it in (C).
kinase autophosphorylation strongly favors the unphosphorylated protein (Jiang et al., 2000; Stock et al., 2000).
Strikingly, however, the reverse reaction was strongly
inhibited by Sda (Figures 2C and 2D, lane 5).
We conclude that Sda shuts down the sporulation
phosphorelay by inhibiting KinA autophosphorylation.
Moreover, the fact that Sda blocks both autophosphorylation and the reverse reaction has important mechanistic implications. First, it indicates that Sda can bind
phosphorylated KinA, even though it does not affect
phosphate transfer from KinAⵑP to Spo0F or the rate
of KinAⵑP dephosphorylation. Second, it indicates that
Sda blocks all KinA reactions that require communication between the catalytic and DHp domains, whereas
no other KinA reactions are affected.
Three-Dimensional Structure of Sda
In order to gain further insight into the mechanism of
action of Sda, we determined its three-dimensional (3D)
structure and used structure-directed mutagenesis to
identify key functional residues. The solution structure
was determined using 3D triple-resonance NMR techniques and is represented by an ensemble of 25 conformers selected on the basis of lowest residual restraint
violations (Figure 3A). Because Sda is small, the heteronuclear NMR data were excellent, and as a result the
structure is highly precise with a root mean square (rms)
deviation of 0.13 ⫾ 0.05 Å over the backbone atoms of
the well-defined region (residues 5–42). There are no
distance or dihedral angle restraint violations greater
than 0.2 Å and 3⬚, respectively, and 91% of the structurally ordered residues (5–42) lie in the most favored regions of the Ramachandran plot. A full summary of structural statistics is given in Table 1.
The Sda monomer is dominated by two ␣ helices comprising residues Asp6–Met19 (helix A) and Arg23–Arg35
(helix B) that are linked by a highly structured interhelix
loop (residues Asn20–Asn22) to form a helical hairpin
(Figure 3B). The helices are antiparallel, and they pack
against one another with a crossover angle of ⵑ30⬚.
1
HN-15N residual dipolar couplings measured from Sda
samples aligned using Pf1 phage (Hansen et al., 2000)
support the orientation of the helices in the calculated
structures (data not shown). Residues Ser37–Val42, on
the C-terminal side of helix B, fold back against the
lower half of one face of the ␣ helices (the C-terminal
face), burying several hydrophobic residues (Leu38,
Ile41, and Ile42). Despite its small size, Sda is stabilized
by numerous hydrophobic and electrostatic interactions. The helical interface is hydrophobic (Figure 3C)
with the exception of an interhelical ion pair between
Asp6 at the N-terminal end of helix A and Arg36 at the
C-terminal end of helix B (Figure 3D). The individual
helices are further stabilized by several intrahelical ion
pairs (Marqusee and Baldwin, 1987) (Figure 3D). Many
of the residues involved in these ion pairs (Asp6, Glu11,
Lys15, Glu32, Arg36) are well conserved across Sda
orthologs (Figure 3E).
The extreme N and C termini of Sda are the least
conserved regions of the protein (see Figure 3E), and
the structure reveals that they are dynamic in solution;
however, their structural disorder is limited by weak
interactions with the helical hairpin. The ⑀-methyl group
at the tip of the Met1 side chain makes NOE contacts
with several hydrophobes (Phe25, Leu28, Ile29) on the
N-terminal face of the helices, thus limiting the inherent
disorder of Met1–Leu4. Similarly, the ␥-methyl groups
of Val44 make NOE contacts with Lys34 on the C-terminal face of the helices, limiting the disorder of Ser43–
Ser46.
The hydrophobic core of Sda is formed by two sets
of interactions. One set occurs at the interface between
the two helices and involves Leu9, Tyr13, Ala16, Ile26,
Ile29, and Ile33 (gray side chains in Figure 3C). The
second set of interactions occurs between several residues on the C-terminal helical face (Ile10, one face of
Molecular Cell
692
Figure 3. 3D Structure of Sda
(A) Stereoview of the ensemble of 25 Sda structures overlaid for lowest rmsd over the backbone atoms of the mean coordinate structure.
(B) Schematic of the Sda structure showing location and orientation of ␣ helices.
(C) The conserved hydrophobic core of Sda consists of interhelical interactions (gray side chains) and a second set of interactions between
residues at the base of helix A and residues in the C-terminal region (light green side chains). Tyr13 and Ile33 participate in both sets
of interactions.
(D) Putative ion pairs in Sda. Note the tripartite interaction at the base of the helices involving Asp6, Glu32, and Arg36. Cationic and anionic
side chains are colored blue and red, respectively.
(E) Sequence alignment of Sda orthologs. Numbering refers to B. subtilis Sda. Residues that are identical in at least four of the seven sequences
are colored yellow. The secondary structure determined in this study is given above the alignment. Hydrophobic core residues are indicated
by red arrowheads, while residues involved in ion pairs in B. subtilis Sda are indicated by blue squares. Green circles denote key functional
residues as determined in this study.
the Tyr13 aromatic ring, Phe14, and the ␥-methyl group
of Ile33) and Leu38, Ile41, and Ile42 in the C-terminal
tail (pastel green side chains in Figure 3C). The side
chains of Leu9, Ile10, Tyr13, Phe14, Ala16, Ile26, Ile29,
Ile33, Leu38, and Ile42 are largely buried in this hydrophobic core, along with the ␥-methyl groups of Thr17
and Ile41. The buried core residues are highly conserved
across known Sda orthologs (Figure 3E), and, consequently, the tertiary structure of Sda is likely to be similar
in all these bacteria.
Identification of Key Functional Residues in Sda
Comparison of the primary structure of Sda orthologs
(Figure 3E) allowed us to identify highly conserved surface-exposed residues in the structure of B. subtilis Sda.
Remarkably, almost all of these residues are located on
the N-terminal face of the helical hairpin (Figures 4A and
4B), suggesting that this region most likely contains the
binding site for KinA. We mutagenized each of these
conserved surface residues and assayed the ability of
each mutant to inhibit KinA autophosphorylation (Figure
Structure and Mechanism of the Sda Antikinase
693
Table 1. Structural Statistics for the Ensemble of 25 Sda Structures
Experimental restraints
Interproton distancesa
Hydrogen bondsb
Dihedral angles (38 φ, 34 ␺, 23 ␹1)
Mean rms deviations from
experimental restraints
NOE distances (Å)
Dihedral angles (deg.)
Mean rms deviations
from idealized geometryc
Bonds (Å)
Angles (deg.)
Impropers (deg.)
Mean X-PLOR energies (kcal mol⫺1)
ENOEe
Ecdihe
Ebond
Eimproper
Eangle
Erepel
Rms deviation to mean coordinate
structure (Å)
Backbone atoms (residues 5–42)
All heavy atoms (residues 5–42)
614
30
95
0.0210 ⫾ 0.0007
0.384 ⫾ 0.026
c
0.00237 ⫾ 0.00008
0.436 ⫾ 0.005
0.258 ⫾ 0.011
14.2
0.86
4.35
3.91
40.7
17.2
⫾
⫾
⫾
⫾
⫾
⫾
1.0
0.12
0.27
0.36
1.0
1.3
0.13 ⫾ 0.05
0.78 ⫾ 0.08
a
Only structurally relevant restraints are included.
Two restraints were used per hydrogen bond.
c
All statistics are given as mean ⫾ SD.
d
Idealized geometry is defined by the CHARMM force field as implemented in X-PLOR.
e
Final values of the square-well NOE and dihedral-angle potentials
were calculated with force constants of 50 and 200 kcal mol⫺1 Å⫺2,
respectively.
b
4D). Mutants E32A and R36A were insoluble and were
excluded from further analysis. These residues form part
of a conserved ion-pair network (see Figure 3D) which
might be critical for proper folding of Sda.
Most mutants (L8K, E11A, S12A, K15A, E18A, M19L,
D24A, R35E, and S45A) were equivalent to wild-type
Sda in their ability to inhibit KinA autophosphorylation,
whereas several mutants (M19A, M19I, and L28A)
showed a partial loss of inhibitory activity. Of most interest were four mutants (L21A, L21E, F25A, and F25H)
that showed a complete loss of function.
All but one of the soluble Sda mutants were correctly
folded. We assessed folding competency by comparing
the 1H-15N HSQC spectrum of each 15N-labeled mutant
protein with that of wild-type Sda (data not shown). The
HSQC spectrum is a structural fingerprint that contains
a single crosspeak for each backbone amide proton.
Changes in the backbone fold of a protein alters the
electronic environment of these atoms and perturbs
their chemical shift in the HSQC spectrum. The HSQC
spectrum of all but one mutant protein was similar to
that of wild-type Sda, with only minor chemical shift
perturbations for residues proximal to the mutation site.
However, the HSQC spectrum of F25A was vastly different from that of wild-type Sda, indicating that this mutant
is misfolded.
We conclude that only L21A, L21E, and F25H are true
loss-of-function mutants, while M19A, M19I, and L28A
are partial loss-of-function mutants. Remarkably, the
four residues affected by these mutations, Met19,
Leu21, Phe25 and Leu28, form a contiguous hydropho-
bic patch on the N-terminal surface of the helical hairpin
(Figure 4C). The residues that appear to be most critical
for Sda function, Leu21 and Phe25, are located in the
center of this patch.
Identification of KinA Binding Site on Sda
We took two approaches to mapping the KinA binding
site on Sda. First, we used pull-down assays to examine
the ability of each Sda mutant to bind KinA. All mutants
that were competent to inhibit KinA autophosphorylation were also able to bind KinA383–606 (Figure 4E). Mutant
Sda proteins that exhibited a partial loss-of-function
(M19I, M19A, and L28A) could still bind KinA383–606 in
the pull-down assay, as could the total loss-of-function
mutant L21A (Figure 4E). However, we suspect that more
quantitative binding experiments are likely to reveal subtle KinA binding defects for these mutants, as shown
below for L21A and M19I. L21E and F25H, two of the
three mutants that were completely defective in blocking
KinA autophosphorylation, showed defects in KinA
binding (bottom right panel in Figure 4E). This suggests
that Leu21 and Phe25 are major components of the KinA
binding site.
Next, we used NMR to examine the interaction between Sda and the KinA DHp domain (KinA383–465). When
unlabeled DHp domain was added to an equimolar solution of 15N-labeled Sda, the suppression of HSQC crosspeak intensity was so severe that chemical shift changes
could not be determined; however, the experiment provided unequivocal evidence for an interaction between
Sda and the ⵑ20 kDa DHp dimer. Although crosspeak
intensity was suppressed by an average of ⵑ60% when
DHp domain was added at a molar ratio of 1:2.7 (Figure
5A), chemical shift perturbations could be monitored for
most Sda residues, and they were greatest for Ser12,
Lys15, Leu21, Asn22, Asp24, and Phe25 (Figure 5B).
These residues form a spatially contiguous cluster centered around Leu21 and Phe25 on the N-terminal surface
of the Sda hairpin (red in Figure 5C).
Ser12, Lys15, and Asp24, which are spatially adjacent
to Leu21 and Phe25, undergo significant shift perturbations in the NMR experiment. However, they are unlikely
to be key binding residues since the S12A, K15A, and
D24A mutants exhibited no defects in autophosphorylation and binding assays. Similarly, although Asn22 suffers a major chemical shift perturbation upon DHp binding, this likely reflects proximity to the binding site rather
than a direct role in KinA binding since this residue
is poorly conserved across Sda orthologs (Figure 3E).
However, we cannot exclude the possibility that Asn22
plays a role in KinA binding. Note that the C-terminal
residues suffered the smallest shift perturbations, consistent with Sda contacting KinA solely via the N-terminal face of the hairpin.
Addition of DHp domain caused the crosspeak intensity for several N-terminal residues (Arg2–Ser5) to be
reduced to baseline levels. It is impossible to ascertain
whether the relative suppression of crosspeak intensity
was more severe than observed for other residues, since
these crosspeaks were weak even prior to addition of
DHp. However, as noted previously, the ⑀-methyl group
of Met1 makes NOE contacts with Phe25 and Leu28 in
the unbound conformation of Sda, and this interaction
Molecular Cell
694
Figure 4. Identification of Key Functional Residues
(A) Structure of Sda showing side chains of surface-exposed residues that are identical or conservatively substituted in all Sda orthologs
available at the time of this study (the O. iheyensis and B. cereus sequences were released after completion of the mutagenesis).
(B) Same schematic as in (A) but rotated ⵑ90⬚ clockwise around the long axis of the helical hairpin. All conserved surface residues, with the
exception of Ser37 and Ser45, are on the N-terminal face of the hairpin.
(C) Surface representation of Sda with location of key functional residues denoted in red. Molecular orientation is the same as (A).
(D) Assays of the ability of wild-type and mutant Sda proteins to inhibit KinA autophosphorylation. Sda concentrations (in pmol) are given
above each lane. Inhibition of KinA autophosphorylation is indicated by the lack of a KinAⵑP band(s) on the gel. Each small panel shows the
result for a single protein. All mutants were correctly folded with the exception of the F25A mutant labeled “Misfold.”
(E) Assays of the ability of wild-type and mutant Sda proteins to bind KinA autokinase domain. Sda proteins were incubated with Ni2⫹-NTA
agarose beads decorated with His6-KinA383–606, then unbound (lanes labeled “S”) and bead-bound (lanes labeled “B”) fractions were recovered
and analyzed as described in Experimental Procedures. In (D) and (E) the mutants are divided into three panels based on their ability to inhibit
KinA autophosphorylation.
is likely to be perturbed upon DHp binding. This might
alter the conformation of the relatively flexible N-terminal segment (Met1–Ser5), which could lead to signifi-
cant peak suppression for these residues in the presence of a substoichiometric amount of DHp.
We conclude that the hydrophobic surface patch
Structure and Mechanism of the Sda Antikinase
695
Figure 5. Sda Binds to and Stabilizes the DHp Dimer
(A) 1H-15N HSQC spectrum of Sda (260 ␮M) in the absence (red) and presence (blue) of 97 ␮M KinA DHp domain (KinA383–465). Residues that
suffered significant chemical shift perturbations upon binding DHp domain are labeled in black, while resonances whose intensities were
suppressed to an undetectable level are labeled in red. The inset highlights chemical shift perturbation for Phe25.
(B) Normalized chemical shift perturbation (⌬␦ ⫽ [(⌬␦H)2 ⫹ (0.17 ⫻ ⌬␦N)2]1/2) for Sda binding to KinA DHp domain. Residues with largest ⌬␦ are
labeled, while residues whose intensities were suppressed to baseline (thus precluding calculation of ⌬␦) are indicated by an asterisk.
(C) Molecular surface of Sda viewed from the N-terminal face (left) and C-terminal face (right) of the helical hairpin. Residues that showed
significant ⌬␦ upon binding DHp domain are colored red and labeled. Met19 and Leu28 are colored blue; these residues did not undergo a
large ⌬␦, but autophosphorylation and MALLS assays indicated they may participate in KinA binding (see text).
(D) Proteins were fractionated by SEC, and Mw was estimated using MALLS. Sda (trace A) and the KinA autokinase domain (trace B) were
chromatographed separately or in the molar ratios indicated (traces C–F). The peak at ⵑ16 min corresponds to aggregated KinA with an Mw
range of 1–11 MDa.
(E) MALLS analysis of the binding of Sda and various mutants to KinA autokinase domain. Molar ratios are given in parentheses to the right
of each trace.
formed by Leu21 and Phe25 represents the primary site
for interaction of Sda with the KinA DHp domain. These
residues are conserved in all Sda orthologs (Figure 3E).
Sda Enhances KinA Dimerization
We examined whether Sda alters KinA oligomerization
(Figure 5), since KinA is only active as a homodimer.
Although the C-terminal sensor domain facilitates KinA
dimerization, the autokinase domain still homodimerizes
in the absence of the PAS modules (Wang et al., 2001).
As for other histidine kinases, autophosphorylation is
presumed to occur in trans, with one monomer phosphorylating its partner in the dimer (Ninfa et al., 1993;
Qin et al., 2000). Thus, a plausible explanation for the
ability of Sda to block both the autophosphorylation and
reverse reaction is that it induces dissociation of the
KinA homodimer.
Proteins were fractionated using size exclusion chromatography (SEC), and molecular masses were estimated using multi-angle laser light scattering (MALLS).
Note that, unlike SEC, the molecular mass estimated
using MALLS is independent of protein shape and is
unaffected by protein interactions with the column matrix. Sda alone yielded a single peak with an estimated
weight-average molecular mass (MW) of 6.0 ⫾ 0.4 kDa
(Figure 5D, trace A). This is close to the theoretical mass
Molecular Cell
696
of an Sda monomer (5.6 kDa), indicating that Sda is
primarily monomeric under these conditions.
When applied alone to the column, the KinA autokinase domain eluted as two peaks (Figure 5D, trace B).
The peak with shortest retention time (RT ⵑ20.5 min)
had an MW value of 56.1 ⫾ 1.5 kDa, which is close to
the estimated mass of a KinA383–606 dimer (54.4 kDa).
We conclude that the early-eluting peak corresponds
to KinA dimer. The later eluting peak (RT ⵑ23 min) had
an MW value of 31.0 ⫾ 1.2 kDa, slightly higher than the
estimated mass of a KinA383–606 monomer (27.2 kDa).
However, because of its proximity to the dimer peak,
the estimated mass will be biased toward higher MW
by trailing dimers that overlap the monomer peak. We
therefore conclude that this peak corresponds to
KinA383–606 monomer. On the basis of the relative intensity
of the two peaks, we estimate that the KinA autokinase
domain is about 50% dimer/50% monomer under the
chosen experimental conditions.
If Sda functions by disrupting the KinA dimer, then
one would predict an increase in relative intensity of the
KinA monomer peak if increasing amounts of Sda were
added to a fixed concentration of KinA. We observed
precisely the opposite. As Sda was added to KinA383–606,
the intensity of the dimer peak increased at the expense
of monomer (Figure 5D, traces C–F). At approximately
equimolar concentrations of KinA383–606 and Sda (Figure
5D, trace E), most of the KinA was converted to dimer.
This indicates not only that Sda enhances KinA dimerization but that the binding affinity must be high. The small
amount of apparently monomeric KinA at a 5-fold excess
of Sda (Figure 5D, trace F) is probably either a contaminant or dimerization-incompetent KinA383–606 because we
were unable to convert it to dimer using a 10-fold excess
of Sda (data not shown).
Note that the KinA “dimer” peak shifts to shorter retention time as more Sda is added. This shift in retention
time would be expected if Sda bound tightly to KinA to
form a stable heteromeric complex with higher mass
than the KinA dimer. At a 5-fold excess of Sda (Figure
5F), the estimated mass of the Sda:KinA383–606 complex
is 66.7 ⫾ 0.5 kDa, which is close to the predicted mass
of a KinA dimer with two molecules of bound Sda (65.6
kDa). Thus, Sda and KinA appear to form a heterotetramer comprising two molecules of each protein. This
1:1 stoichiometry is consistent with the observation that
the KinA monomer is quantitatively converted to dimer
when equimolar amounts of Sda and KinA383–606 are
mixed (Figure 5D, trace E).
The F25H mutant, which failed to bind KinA in pulldown assays, also failed to affect KinA dimerization,
even when added in 5-fold molar excess (Figure 5E,
trace B). L21E behaved similarly (data not shown). Interestingly, while L21A appeared to bind KinA normally in
pull-down assays (Figure 4E), MALLS revealed a binding
defect. Whereas wild-type Sda fully bound to an equimolar amount of KinA and converted all KinA monomer to
dimer (Figure 5E, trace D), an equimolar quantity of L21A
failed to convert all KinA monomer to dimer, and a proportion of the L21A remained unbound; this weaker
binding manifests as a reduced shift in retention time
of the KinA dimer peak relative to the shift induced
by wild-type Sda (compare traces C and D in Figure
5E). This binding defect presumably reflects an in-
creased off-rate for KinA binding, which might explain
why L21A does not inhibit KinA autophosphorylation.
M19I showed a similar binding defect to L21A in the
MALLS assay, whereas L28A behaved similarly to wildtype Sda (not shown). Thus, the MALLS data indicate
that Met19, Leu21, and Phe25 all contribute to the
Sda:KinA interaction.
We conclude that Sda binds the DHp domain of KinA
via a cluster of conserved hydrophobic residues (Met19,
Leu21, Phe25, and possibly Leu28) and that this interaction enhances KinA homodimerization. This contrasts
with several organic compounds isolated in a chemical
screen for KinA inhibitors which appear to intercalate
into the four-helix bundle of the DHp domain in a manner
that disrupts the dimer and promotes aggregation of
KinA monomers (Stephenson et al., 2000).
Discussion
Sda Disrupts Communication between
KinA Domains
Our observation that Sda blocks both autophosphorylation and the reverse reaction (i.e., transfer of phosphate
from KinAⵑP to ADP) indicates that the primary lesion
caused by Sda is defective communication between
the ATP binding site in the catalytic domain and the
phosphorylatable His in the DHp domain. The model
that best explains these results is that Sda forms a
molecular barricade between the catalytic and DHp domains of KinA, thus preventing interdomain communication. This mechanism is consistent with Sda blocking
both the autophosphorylation and reverse reaction, and,
as outlined below, it can explain how Sda stabilizes the
KinA dimer.
The catalytic and DHp domains of histidine kinases
are separated by a flexible hinge which is thought to
facilitate kinase function by allowing the autokinase
module to exist in “open” and “closed” conformations.
The closed conformation is thought to bring the nucleotide binding site on one monomer into close proximity
with the phosphorylatable His on the other monomer
during the autophosphorylation and reverse reaction.
These regions are presumed to be further separated in
the open conformation to provide room for docking of
the cognate response regulator.
To the best of our knowledge, no structures are available of an intact autokinase domain, with the exception
of the unorthodox CheA histidine kinase which has a
different domain organization from KinA (Bilwes et al.,
1999). However, Inouye and coworkers (Cai et al., 2003)
recently derived a model of the complete EnvZ autokinase fragment (EnvZ-AK) by using targeted disulfide
crosslinking to determine the relative orientation of the
experimentally determined structures of the separate
catalytic (Tanaka et al., 1998) and DHp (Tomomori et
al., 1999) domains. NMR studies revealed a dynamic
relationship between the EnvZ catalytic and DHp domains (Cai et al., 2003), but targeted disulfide linking
specifically captures the conformation in which these
domains are in close proximity. Thus, the modeled EnvZAK structure (Figure 6B; PDB code 1NJV) presumably
represents the closed conformation of the autokinase
fragment that is associated with autophosphorylation.
Structure and Mechanism of the Sda Antikinase
697
Figure 6. Proposed Sda Binding Site and Mechanism of Action
(A) Alignment of EnvZ with KinA, B, and C. Numbering refers to KinA. Residues that are identical or conservatively substituted in at least three
of the four sequences are highlighted in yellow and orange, respectively. The experimentally determined secondary structure of the EnvZ
DHp (Tomomori et al., 1999) and catalytic (Tanaka et al., 1998) domains is given below the alignment. Linker regions are demarcated by red
lines. The percentage identity (I) and similarity (S) relative to KinA is indicated at the end of each sequence.
(B) Modeled structure of the EnvZ autokinase domain (PDB file 1NJV) (Cai et al., 2003). Domains and linkers are color-coded to match the
sequence alignment in (A). In (B)–(D) the side chain of the phosphorylatable His is colored orange.
(C) Schematic of the Spo0F-Spo0B cocrystal structure (PDB file 1F51). Only the N-terminal four-helix bundle of the Spo0B dimer is shown;
the C-terminal ␣/␤ domains have been omitted for clarity. The side chain of the active-site Asp residue in Spo0F is shown in red.
(D) Schematic of the structure of the EnvZ DHp domain (PDB file 1JOY). Highlighted in red is the OmpR binding site determined by NMR
chemical shift mapping (Tomomori et al., 1999).
(E) Alignment of KinA383–460 with Spo0B. The secondary structure of the four-helix bundle of Spo0B is indicated below the sequences. Residues
in Spo0B that contact Spo0F (Zapf et al., 2000) are indicated by red circles, and the active-site His residues are denoted by an asterisk (His405
in KinA, His30 in Spo0B). The predicted Spo0F binding site on KinA and the area available for Sda binding are indicated above the sequences.
(F) Schematic of the closed conformation of the KinA autokinase domain based on the EnvZ model structure. The two monomers are shown
in orange and blue, the phosphorylatable His405 is depicted as a green circle, and the approximate location of the ATP binding site on the
catalytic domain is indicated. The predicted Spo0F binding site and the area available for Sda binding are indicated.
(G and H) Two alternative models of the mechanism of Sda action. Sda could lodge under the linker region at the top of the DHp domain (G)
or bind exclusively to the linker region (H). Either orientation could explain why Sda enhances KinA dimerization (see text for details).
Molecular Cell
698
EnvZ is a type 1A histidine kinase and its H box, ATP
binding site, and hinge length are similar to that of KinA
and other type I histidine kinases. Although the autokinase domains of EnvZ and KinA are only 21% identical,
the extent of similarity is 49% when conservative replacements are considered (Figure 6A). Thus, for the
purposes of the following discussion, the modeled
EnvZ-AK structure is a good surrogate for the structure
of the KinA autokinase module. The EnvZ-AK structure
(Figure 6B) indicates that in the closed conformation the
ATP binding face of the catalytic domain packs against
two helices of the DHp domain, one provided by each
monomer. The ATP binding site on one monomer is
located close to the phosphorylatable His on the other
monomer, but distal to the corresponding His on the
same polypeptide chain, thus explaining why autophosphorylation occurs in trans.
In this study we showed that Sda binds KinAⵑP but
does not block phosphate transfer from the DHp domain
to Spo0F. This indicates that Sda and Spo0F can bind
simultaneously to nonoverlapping sites on the DHp domain. Although no structure is available of the KinASpo0F complex, we can infer the Spo0F binding site on
the KinA DHp domain from other studies. The crystal
structure of the Spo0F-Spo0B complex (Zapf et al.,
2000) indicates that Spo0F binds at the hairpin end of
the four-helix bundle formed by the Spo0B dimer (Figure
6C). The two proteins dock so that the Spo0F activesite Asp is perfectly positioned for phosphotransfer to
the phosphorylatable His on Spo0B. The Spo0F contact
surface extends from just above the His residue to the
end of the Spo0B hairpin. Although no structure is available of the complex between EnvZ and its cognate response regulator, OmpR, NMR chemical shift mapping
experiments (Tomomori et al., 1999) showed that it binds
to the hairpin end of the EnvZ DHp domain in a manner
analogous to docking of Spo0F to the Spo0B (Figure
6D). Thus, the Spo0F-Spo0B structure reveals a common mode of docking of response regulators to the
four-helix bundle of DHp domains (Varughese, 2002).
Alignment of KinA383–460 with Spo0B (Figure 6E) reveals
that, although the overall similarity is low, the Spo0B
residues that contact Spo0F (see Table 1 in Zapf et
al., 2000) are remarkably well conserved in KinA. Thus,
Spo0F most likely binds KinA in a manner similar to its
interaction with Spo0B, which enables us to pinpoint
the Spo0F binding site on KinA to residues 402–444 of
the DHp domain (Figures 6E and 6F).
Since Sda binds KinA383–460, its contact surface must
be restricted to residues 383–401 and 445–460 (Figure
6E), since its binding site cannot overlap with that of
Spo0F for the reasons outlined above. Thus, the Sda
binding site is contained within 35 residues that encompass the linker between the DHp and sensor domains,
the linker-proximal end of the four-helix bundle, and the
autokinase domain linker (Figure 6F). Sda is a small
protein and molecular modeling studies using EnvZ-AK
as a surrogate for KinA indicate that it could either lodge
under the linker region at the top of the DHp domain
(Figure 6G), or bind exclusively within the linker region
(Figure 6H). This interaction is presumably mediated by
the hydrophobic cluster delineated in our mutagenesis
and NMR studies (Figure 4C).
Note that the ATP binding face of the catalytic domain
packs against two helices of the DHp domain, one provided by each monomer (Figure 6F). Thus, if Sda bound
across the intermonomer face of the four-helix bundle
such that it made contact with ␣1 of one monomer and
␣2 of the other monomer (Figure 6G), it might be expected that Sda would stabilize the KinA dimer, as we
observed experimentally. An alternative possibility is
that Sda stabilizes the KinA dimer by binding the autokinase domain linker of one monomer and the linker to
the sensor domain on the other monomer (Figure 6H).
Regardless of its exact mode of binding, we propose
that Sda acts as a molecular barricade that stabilizes
an open conformation of the KinA autokinase domain,
thus precluding close contact of His405 with the ATP
binding site on the catalytic domain. This would explain
why Sda blocks both autophosphorylation and the reverse reaction, but not phosphotransfer to Spo0F. If our
hypothesis about the molecular basis of Sda function
is correct, then cocrystallization of a Sda:KinA-AK complex might provide a unique opportunity to obtain the
structure of an orthodox autokinase fragment, since Sda
binding might eliminate the dynamic relationship between the catalytic and DHp domains.
Comparison with Other Antikinases
Sda appears to have a very different mechanism of action from that of PII and GlnK, two homologous inhibitors
of the NRII/NtrB histidine kinase. PII/GlnK form large
homotrimers or mixed heterotrimers that bind the NRII/
NtrB catalytic domain (Pioszak et al., 2000). The active
form of the PII/GlnK trimer inhibits NRII/NtrB autophosphorylation and activates its phosphatase activity toward NRI/NtrCⵑP (Jiang and Ninfa, 1999).
The antikinase activity of Sda appears to be functionally analogous to KipI inhibition of KinA (Wang et al.,
1997) and probably also inhibition of the FixL sensor
kinase by FixT (Garnerone et al., 1999). However, the
molecular details of these kinase-antikinase interactions
are likely to be significantly different, as KipI (26.7 kDa)
and FixT (12.2 kDa) are larger that Sda, the proteins
share no amino acid sequence similarity, and, in contrast
to Sda, a single FixT monomer appears to bind each
FixL dimer (Garnerone et al., 1999). Nevertheless, it will
be interesting to see whether each of these antikinases
acts by blocking communication between the kinase
catalytic and DHp domains.
A New Paradigm for Histidine Kinase Inhibitors?
Histidine kinase signaling pathways are attractive antimicrobial targets, largely because of their role in mediating
antibiotic resistance and expression of virulence-associated factors (Matsushita and Janda, 2002; Stephenson
and Hoch, 2002) and their absence from the animal kingdom (Thomason and Kay, 2000). In recent years, much
effort has been expended in screening chemical libraries
for small-molecule inhibitors of histidine kinases. For
the most part, lead compounds from these screens have
proved disappointing (Stephenson and Hoch, 2002). A
more rational approach might be to develop mimics of
endogenous histidine kinase inhibitors. However, prior
to this study, the only available structure of such an
inhibitor was that of PII and its homolog GlnK (Xu et
al., 1998). PII/GlnK are large trimeric molecules that are
Structure and Mechanism of the Sda Antikinase
699
allosterically regulated by various ligands, and the molecular mechanism by which they alter NRII kinase activity is uncertain (Pioszak et al., 2000). Thus, at least at
the present time, they are not useful templates for medicinal chemistry.
Sda, on the other hand, might establish a new paradigm for rational design of histidine kinase inhibitors.
First, Sda inhibits its cognate histidine kinase via a
mechanism that has not been explored by medicinal
chemists. Second, Sda is the smallest antikinase protein
discovered to date, making it a tractable template for
rational design. This study has provided a first glimpse
of the Sda pharmacophore, and it appears to be restricted to a small hydrophobic cluster on one face of
the protein. The successful design of small mimetics
based on more complicated, discontinuous binding epitopes (Gadek et al., 2002) suggests that it should be
possible to use the Sda pharmacophore as a template
for rational design of histidine kinase inhibitors.
Experimental Procedures
Cloning
B. subtilis sda was PCR amplified using primers encoding 3⬘ BamHI
and 5⬘ EcoRI sites. The gene for Schistosoma japonicum glutathione
S-transferase (GST) was PCR amplified using a 3⬘ primer encoding
an NcoI site and a 5⬘ primer encoding BamHI and thrombin cleavage
sites. After digestion with relevant restriction enzymes the two fragments were inserted into NcoI/EcoRI-digested pET28a to create
plasmid pSLR65 which encodes a GST-Sda fusion with an intervening thrombin cleavage site. We used a similar approach to produce
pSLR70 which encodes a GST-sda fusion with codons modified for
optimal expression in E. coli. Sda mutants were produced by overlap-extension PCR of GST-sda using pSLR65 or pSLR70 as template.
Plasmid pBB202 was produced by ligating NdeI/EcoRI-digested
pET28b to a PCR product encompassing codons 383–606 of
B. subtilis kinA flanked 5⬘ by an NdeI site and 3⬘ by a TAA stop
codon and EcoRI site. This construct encodes KinA383–606 preceded
by residues MGSSHHHHHHSSGLVPRGSHM; the encoded thrombin
cleavage site (underlined) was not used during purification, so KinA
remained His-tagged. Plasmids encoding His6-KinA456–606, His6KinA383–460, and His6-KinA383–465 were constructed similarly.
Protein Production
GST-Sda fusion proteins were overexpressed in E. coli BL21(␭DE3).
Uniformly labeled proteins for NMR studies were obtained by growing cells in vitamin-supplemented M9 minimal medium with 15NH4Cl
and [U-13C]glucose as sole nitrogen and carbon sources, respectively. Fusion proteins were purified using GSH affinity chromatography. Sda was liberated from GST by on-column thrombin cleavage,
and then purified to ⬎98% homogeneity using SEC. Sda proteins
contained a vestigial N-terminal Gly-Ser from the thrombin cleavage
site. Sda was dialyzed against buffer B (50 mM NaPi, 300 mM NaCl,
0.02% NaN3 [pH 7.9]) for MALLS experiments.
Full length, C-terminal His6-tagged KinA and Spo0F (KinA1–606-His6
and Spo0F-His6) were produced as described (Burkholder et al.,
2001). His6-KinA383–606 was expressed from pBB202 in E. coli
BL21(␭DE3) and purified using His-bind Ni2⫹ resin. The eluted protein
was judged to be ⵑ99% pure based on Coomassie-stained SDSPAGE gels. Other His6-KinA constructs were expressed and purified
similarly. Protein concentrations were determined using a BCA
assay (Pierce).
Pull-Down Assays
Ni-NTA resin equilibrated in buffer A (20 mM NaPi, 100 mM NaCl
[pH 7.6]) was mixed with His6-KinA383–606 in buffer C (50 mM NaPi,
300 mM NaCl, 10% glycerol [pH 8]) containing 250 mM imidazole.
After dilution with buffer A to give an imidazole concentration of 6
mM, the sample was incubated at 4⬚C for 1 hr. The resin was pel-
leted, then supernatant containing excess unbound KinA was removed. The concentration of His6-KinA383–606 was 3.1 nmol/40 ␮l of
resin assuming all KinA bound the resin. Mutant or wild-type Sda
(0.7–2.8 nmol/3.1 nmol KinA) was added to resin, and the volume
was made up to 80–90 ␮l with buffer A. The resin mixture was
incubated at ambient temperature for 1 hr with rocking, then the
supernatant, containing unbound proteins, was decanted, and
beads were washed twice with buffer A. Thirty-two microliters of
buffer A containing 625 mM imidazole was added to each tube, and
then the resultant supernatant, containing proteins that had bound
to beads, was decanted. Pull-downs with other His6-KinA constructs
were performed similarly. Bound and unbound samples were analyzed using Tricine SDS-PAGE.
Autophosphorylation Assays
Autophosphorylation assays were performed essentially as described (Zapf et al., 1998). Reaction mixtures were assembled on
ice, and reactions were started by adding ATP (0.5 mM final concentration, 25 ␮Ci ␥-32P-ATP/reaction) and shifting the temperature to
25⬚C. Each reaction contained 10 pmol of KinA1–606-His6 and 0–100
pmol of Sda. Reactions were stopped after 5 min by adding EDTA
(pH 8) to 100 mM and placing samples on ice. Samples were electrophoresed on 10% glycine SDS-PAGE gels; then gels were washed
before being exposed for 12 hr on a Phosphorimager plate (Molecular Dynamics).
Dephosphorylation and Phosphotransfer Assays
P-labeled KinA1–606-His6ⵑP was prepared by incubating KinA1–606His6 (6.5 ␮g, 95 pmol) in a 50 ␮l reaction containing 0.5 mM ATP/
50 ␮Ci ␥-32P-ATP. KinA1–606-His6ⵑP was purified from the quenched
reaction mix by gel filtration at 4⬚C using a BioSpin-30 column (BioRad) prerinsed with 50 mM NaPi (pH 8), 200 mM NaCl, 5% glycerol,
0.1 mM EDTA. Assays of KinA1–606-His6ⵑP dephosphorylation and
KinA1–606-His6ⵑP-dependent phosphorylation of Spo0F-His6 were
performed in 25 ␮l reactions containing kinase buffer. KinA1–606His6ⵑP (ⵑ0.6 ␮g, 9 pmol/reaction) was premixed on ice with Sda
(500 ng, 90 pmol) or storage buffer. Reactions were started by adding kinase buffer alone or containing 1 mM ADP (final concentration)
or 2.1 ␮g Spo0F-His6 (150 pmol). Reactions were incubated for 30
min at 25⬚C and stopped by adding EDTA to 100 mM. To monitor
formation of 32P-ATP, 1 ␮l of each reaction was spotted onto PEIcellulose and developed in 0.8 M acetic acid/0.8 M LiCl. To monitor
formation of Spo0F-His6ⵑP, samples were processed for SDSPAGE and exposed on a Phosphorimager plate as described for
autophosphorylation assays.
32
MALLS
MALLS was performed as described (Folta-Stogniew and Williams,
1999) using a miniDAWN Tristar laser light scattering photometer
and Optilab DSP interferometric refractometer (both from Wyatt
Technology). SEC was performed with buffer B using a Superdex
75 HR 10/30 column (Pharmacia Biotech). All samples were injected
in a final volume of 200 ␮l to avoid volume-related retention-time
artifacts. MW estimates were determined using Debye fitting, and
reported errors are standard deviations (SDs) of the MW estimate.
Structure Determination
NMR data were acquired at 298 K on Varian INOVA 500 and 600
MHz spectrometers using 15N- or 15N/13C-labeled Sda (ⵑ1 mM in
buffer A [pH 6.85]). Data were processed with nmrPipe (Delaglio
et al., 1995), and spectra were analyzed using XEASY (Bartels et
al., 1995).
1
H, 15N, and 13C resonance assignments were obtained as described (King et al., 1999). Crosspeak intensities in 13C- and
15
N-edited 3D NOESY-HSQC spectra (␶m ⫽ 150 ms) were converted
to distance restraints using DYANA (Güntert et al., 1997). Thirty-four
(φ,␺) restraints were derived using TALOS (Cornilescu et al., 1999),
and errors were set to twice the estimated SD. φ restraints were
verified by 3JHNH␣ couplings measured from a 3D HNHA spectrum.
Four additional φ restraints of ⫺100 ⫾ 80⬚ were obtained from
H␣-HN NOEs (Clubb et al., 1994). Twenty-three ␹1 restraints and
stereospecific assignments for nine pairs of ␤-methylene protons
were derived from NOESY intensities and 3JHNH␤ couplings.
Molecular Cell
700
Sda was unstable when lyophilized. Thus, slowly exchanging amide protons were identified as those present in a 13C-edited NOESYHSQC spectrum acquired immediately after exchange of the sample
into 2H2O using a 3K Centricon. Fifteen slowly exchanging amide
protons were unambiguously assigned as H bond acceptors, and
corresponding H-bond restraints were applied as described (King
et al., 2000).
DYANA was used to calculate 500 structures from random starting
conformations. The best 50 structures (selected on the basis of final
penalty function values) were refined in X-PLOR (Brünger, 1992).
1
HN-15N residual dipolar couplings elicited by aligning Sda (ⵑ200
␮M) using ⵑ10 mg/ml Pf1 filamentous phage (Asla Biotech) were
used to ensure that the helices in the calculated structures had the
correct relative orientation. MOLMOL (Koradi et al., 1996) was used
for molecular graphics.
dyanamics for NMR structure calculation with the new program
DYANA. J. Mol. Biol. 273, 283–298.
Acknowledgments
King, G.F., Pan, B., Maciejewski, M.W., Rowland, S.L., Rothfield,
L.I., and Mullen, G.P. (1999). Backbone and side-chain 1H, 15N, and
13
C assignments for the topological specificity domain of the MinE
cell division protein. J. Biomol. NMR 13, 395–396.
This work was supported by GM41934 from the NIH. W.F.B. was
supported by a postdoctoral fellowship from the American Cancer
Society and the Merck/MIT Collaboration Program. We thank David
Wah for technical assistance and Scott Robson for help with MALLS.
Received: July 3, 2003
Revised: January 20, 2004
Accepted: January 20, 2004
Published: March 11, 2004
References
Bartels, C., Xia, T.-H., Billeter, M., Güntert, P., and Wüthrich, K.
(1995). The program XEASY for computer-supported NMR spectral
analysis of biological macromolecules. J. Biomol. NMR 5, 1–10.
Bilwes, A.M., Alex, L.A., Crane, B.R., and Simon, M.I. (1999). Structure of CheA, a signal-transducing histidine kinase. Cell 96, 131–141.
Hansen, M.R., Hanson, P., and Pardi, A. (2000). Filamentous bacteriophage for aligning RNA, DNA, and proteins for measurement of
nuclear magnetic resonance dipolar coupling interactions. Methods
Enzymol. 317, 220–240.
Hoch, J.A. (2000). Two-component and phosphorelay signal transduction. Curr. Opin. Microbiol. 3, 165–170.
Jiang, P., and Ninfa, A.J. (1999). Regulation of autophosphorylation
of Escherichia coli nitrogen regulator II by the PII signal transduction
protein. J. Bacteriol. 181, 1906–1911.
Jiang, P., Peliska, J.A., and Ninfa, A.J. (2000). Asymmetry in the
autophosphorylation of the two-component regulatory system
transmitter protein nitrogen regulator II of Escherichia coli. Biochemistry 39, 5057–5065.
King, G.F., Shih, Y.-L., Maciejewski, M.W., Bains, N.P.S., Pan, B.,
Rowland, S.L., Mullen, G.P., and Rothfield, L.I. (2000). Structural
basis for the topological specificity function of MinE. Nat. Struct.
Biol. 7, 1013–1017.
Koradi, R., Billeter, M., and Wüthrich, K. (1996). MOLMOL: a program
for display and analysis of macromolecular structures. J. Mol. Graph.
14, 51–55.
LeDeaux, J.R., Yu, N., and Grossman, A.D. (1995). Different roles
for KinA, KinB, and KinC in the initiation of sporulation in Bacillus
subtilis. J. Bacteriol. 177, 861–863.
Lois, A.F., Weinstein, M., Ditta, G.S., and Helinski, D.R. (1993). Autophosphorylation and phosphatase activities of the oxygen-sensing
protein FixL of Rhizobium meliloti are coordinately regulated by
oxygen. J. Biol. Chem. 268, 4370–4375.
Brünger, A.T. (1992). X-PLOR Version 3.1 (New Haven, Yale University Press).
Marqusee, S., and Baldwin, R.L. (1987). Helix stabilization by
Glu–…Lys⫹ salt bridges in short peptides of de novo design. Proc.
Natl. Acad. Sci. USA 84, 8898–8902.
Burbulys, D., Trach, K.A., and Hoch, J.A. (1991). Initiation of sporulation in B. subtilis is controlled by a multicomponent phosphorelay.
Cell 64, 545–552.
Matsushita, M., and Janda, K.D. (2002). Histidine kinases as targets
for new antimicrobial agents. Bioorg. Med. Chem. 10, 855–867.
Burkholder, W.F., Kurtser, I., and Grossman, A.D. (2001). Replication
initiation proteins regulate a developmental checkpoint in Bacillus
subtilis. Cell 104, 269–279.
Ninfa, E.G., Atkinson, M.R., Kamberov, E.S., and Ninfa, A.J. (1993).
Mechanism of autophosphorylation of Escherichia coli nitrogen regulator II (NRII or NtrB): trans-phosphorylation between subunits. J.
Bacteriol. 175, 7024–7032.
Cai, S.J., Khorchid, A., Ikura, M., and Inouye, M. (2003). Probing
catalytically essential domain orientation in histidine kinase EnvZ
by targeted disulfide crosslinking. J. Mol. Biol. 328, 409–418.
Perego, M., and Hoch, J.A. (1996). Protein aspartate phosphatases
control the output of two-component signal transduction systems.
Trends Genet. 12, 97–101.
Clubb, R.T., Ferguson, S.B., Walsh, C.T., and Wagner, G. (1994).
Three-dimensional solution structure of Escherichia coli periplasmic
cyclophilin. Biochemistry 33, 2761–2772.
Pioszak, A.A., Jiang, P., and Ninfa, A.J. (2000). The Escherichia coli
PII signal transduction protein regulates the activities of the twocomponent system transmitter protein NRII by direct interaction
with the kinase domain of the transmitter module. Biochemistry
39, 13450–13461.
Cornilescu, G., Delaglio, F., and Bax, A. (1999). Protein backbone
angle restraints from searching a database for chemical shift and
sequence homology. J. Biomol. NMR 13, 289–302.
Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J., and Bax,
A. (1995). NMRPipe: a multidimensional spectral processing system
based on UNIX pipes. J. Biomol. NMR 6, 277–293.
Dutta, R., Qin, L., and Inouye, M. (1999). Histidine kinases: diversity
of domain organization. Mol. Microbiol. 34, 633–640.
Folta-Stogniew, E., and Williams, K.R. (1999). Determination of molecular masses of proteins in solution: implementation of an HPLC
size exclusion chromatography and laser light scattering service in
a core laboratory. J. Biomol. Tech. 10, 51–63.
Gadek, T.R., Burdick, D.J., McDowell, R.S., Stanley, M.S., Marsters,
J.C., Jr., Paris, K.J., Oare, D.A., Reynolds, M.E., Ladner, C., Zioncheck, K.A., et al. (2002). Generation of an LFA-1 antagonist by
the transfer of the ICAM-1 immunoregulatory epitope to a small
molecule. Science 295, 1086–1089.
Garnerone, A.M., Cabanes, D., Foussard, M., Boistard, P., and Batut,
J. (1999). Inhibition of the FixL sensor kinase by the FixT protein in
Sinorhizobium meliloti. J. Biol. Chem. 274, 32500–32506.
Güntert, P., Mumenthaler, C., and Wüthrich, K. (1997). Torsion angle
Qin, L., Dutta, R., Kurokawa, H., Ikura, M., and Inouye, M. (2000). A
monomeric histidine kinase derived from EnvZ, an Escherichia coli
osmosensor. Mol. Microbiol. 36, 24–32.
Robinson, V.L., Buckler, D.R., and Stock, A.M. (2000). A tale of two
components: a novel kinase and a regulatory switch. Nat. Struct.
Biol. 7, 626–633.
Sonenshein, A.L. (2000). Control of sporulation initiation in Bacillus
subtilis. Curr. Opin. Microbiol. 3, 561–566.
Stephenson, K., and Hoch, J.A. (2002). Histidine kinase-mediated
signal transduction systems of pathogenic microorganisms as targets for therapeutic intervention. Curr. Drug Targets Infect. Disord.
2, 235–246.
Stephenson, K., Yamaguchi, Y., and Hoch, J.A. (2000). The mechanism of action of inhibitors of bacterial two-component signal transduction systems. J. Biol. Chem. 275, 38900–38904.
Stock, A.M., Robinson, V.L., and Goudreau, P.N. (2000). Two-component signal transduction. Annu. Rev. Biochem. 69, 183–215.
Tanaka, T., Saha, S.K., Tomomori, C., Ishima, R., Liu, D., Tong,
K.I., Park, H., Dutta, R., Qin, L., Swindells, M.B., et al. (1998). NMR
Structure and Mechanism of the Sda Antikinase
701
structure of the histidine kinase domain of the E. coli osmosensor
EnvZ. Nature 396, 88–92.
Thomason, P., and Kay, R. (2000). Eukaryotic signal transduction
via histidine-aspartate phosphorelay. J. Cell Sci. 113, 3141–3150.
Tomomori, C., Tanaka, T., Dutta, R., Park, H., Saha, S.K., Zhu, Y.,
Ishima, R., Liu, D., Tong, K.I., Kurokawa, H., et al. (1999). Solution
structure of the homodimeric core domain of Escherichia coli histidine kinase EnvZ. Nat. Struct. Biol. 6, 729–734.
Trach, K.A., and Hoch, J.A. (1993). Multisensory activation of the
phosphorelay initiating sporulation in Bacillus subtilis: identification
and sequence of the protein kinase of the alternate pathway. Mol.
Microbiol. 8, 69–79.
Varughese, K.I. (2002). Molecular recognition of bacterial phosphorelay proteins. Curr. Opin. Microbiol. 5, 142–148.
Wang, L., Grau, R., Perego, M., and Hoch, J.A. (1997). A novel histidine kinase inhibitor regulating development in Bacillus subtilis.
Genes Dev. 11, 2569–2579.
Wang, L., Fabret, C., Kanamaru, K., Stephenson, K., Dartois, V.,
Perego, M., and Hoch, J.A. (2001). Dissection of the functional and
structural domains of phosphorelay histidine kinase A of Bacillus
subtilis. J. Bacteriol. 183, 2795–2802.
Xu, Y., Cheah, E., Carr, P.D., van Heeswijk, W.C., Westerhoff, H.V.,
Vasudevan, S.G., and Ollis, D.L. (1998). GlnK, a PII-homologue: structure reveals ATP binding site and indicates how the T-loops may
be involved in molecular recognition. J. Mol. Biol. 282, 149–165.
Zapf, J., Madhusudan, M., Grimshaw, C.E., Hoch, J.A., Varughese,
K.I., and Whiteley, J.M. (1998). A source of response regulator autophosphatase activity: the critical role of a residue adjacent to the
Spo0F autophosphorylation active site. Biochemistry 37, 7725–
7732.
Zapf, J., Sen, U., Madhusudan, M., Hoch, J.A., and Varughese, K.I.
(2000). A transient interaction between two phosphorelay proteins
trapped in a crystal lattice reveals the mechanism of molecular
recognition and phosphotransfer in signal transduction. Structure
8, 851–862.
Accession Numbers
Chemical shifts for Sda have been deposited in BioMagResBank
(number 5847). Structure coordinates and experimental restraints
for Sda have been deposited in the Protein Data Bank (accession
code 1PV0).