| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, March 2000, p. 1541-1550, Vol. 78, No. 3
and
Department of Chemistry, Department of
Biological Sciences, and *Institute of Molecular Biophysics, Florida
State University, Tallahassee, Florida 32306-4380, USA
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
REFERENCES
|
|---|
Creatine kinase (CK) and arginine kinase (AK) are related enzymes that reversibly transfer a phosphoryl group between a guanidino compound and ADP. In the buffering of ATP energy levels, they are central to energy metabolism and have been paradigms of classical enzymology. Comparison of the open substrate-free structure of CK and the closed substrate-bound structure of AK reveals differences that are consistent with prior biophysical evidence of substrate-induced conformational changes. Large and small domains undergo a hinged 13° rotation. Several loops become ordered and adopt different positions in the presence of substrate, including one (residues 309-319) that moves 15 Å to fold over the substrates. The conformational changes appear to be necessary in aligning the two substrates for catalysis, in configuring the active site only when productive phosphoryl transfer is possible, and excluding water from the active site to avoid wasteful ATP hydrolysis.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
REFERENCES
|
|---|
Phosphagen kinases are a family of enzymes that
include creatine kinase (CK) in vertebrates and arginine kinase (AK) in
arthropods and other invertebrates (Kenyon and Reed, 1983
; Strong and
Ellington, 1993). Their function is to catalyze the reversible transfer
of a phosphoryl group between a phosphorylated guanidino (phosphagen) compound and adenosine diphosphate (ADP). These enzymes play a central
role in cellular energy metabolism with the temporal buffering of ATP
levels in cells with fluctuating energy requirements (muscle, nerves,
etc.) and in shuttling energy between difference cellular compartments
(Wallimann et al., 1992). These enzymes are also paradigms for the
classical enzymology of bimolecular reactions that, from a structural
perspective, are less well characterized than their unimolecular
counterparts (Stroud, 1996).
This work is based on two phosphagen kinase structures that have been
determined recently: 1) an octameric vertebrate mitochondrial creatine
kinase (Mib-CK) in the apo form or as
a binary complex with ATP (but not Mg2+), both at
~3 Å resolution (Fritz-Wolf et al., 1996); and 2) a monomeric
arthropod arginine kinase (AK) crystallized as a transition state
analog (TSA) complex with Mg2+ADP, nitrate
(mimicking a planar phosphoryl in transition), and arginine (Zhou et
al., 1998), and determined at 1.86-Å resolution. These structures are
the fruits of extensive efforts in many laboratories over a period of
30 years, and, although other apo structures have since been
determined (Rao et al., 1998), there is still no sign that it will be
possible to crystallize creatine kinase in an active configuration
(Kabsch and Fritz-Wolf, 1997). Thus we turn to a comparison between
two different members of the family to understand the conformational
changes occurring on substrate binding. This is not ideal, because
there are other possible causes of conformational differences (detailed
below), but there is extensive independent biophysical evidence (see
below) consistent with the assertion that the observed structural
differences reflect substrate-induced conformational changes in both enzymes.
Phosphagen kinases share a common mechanism of direct, in-line
-phosphoryl transfer (Hansen and Knowles, 1981). The transition state structure of AK led to a substantially different understanding of
the catalytic mechanism (Zhou et al., 1998) in which substrate alignment appeared to be more important than other mechanisms, such as
acid-base catalysis. The direct visualization of precise prealignment
of substrates in any enzyme was of general interest, because it had
been predicted to be a significant contributor to the catalysis of many
bimolecular reactions, either through reduction of the entropic
component to the activation barrier (Page and Jencks, 1971), or through
alignment of the reactant orbitals on the optimal trajectory (Dafforn
and Koshland, 1971). The focus of this report is not the details of
catalysis, but the implications of differences between the two
structures in terms of the function of these enzymes.
Creatine and arginine are examples of a family of substrates used by
different phosphagen kinases. The substrates differ in size and
chemical properties, but all share a reactive guanidinium group. The
phosphagen kinases share ~40% amino acid identity (Babbitt et al.,
1986; Dumas and Camonis, 1993; Mühlebach et al., 1994; Suzuki and
Furukohri, 1994), and except in cases where there is evidence of gene
duplication (Stein et al., 1990; Suzuki et al., 1997a) or triplication
(Wothe et al., 1990), most have subunit molecular masses of ~40 kDa.
Although the core part of the structure appears to be conserved,
particular enzymes and isoforms have some side distinctive characters.
For example, unlike AK, Mib-CK has a C-terminal
region of 17 amino acids that is thought to mediate an adhesion between
inner and outer mitochondrial membranes via electrostatic interactions
(Rojo et al., 1991). Mib-CK octamers are
functionally and perhaps physically coupled to ATP/ADP translocators of
the inner mitochondrial membrane (Wyss et al., 1992). Thus it is
possible that some of the surface differences between the two
structures are due to specializations relevant to different quaternary
structure, intracellular compartmentalization, and possibly function.
However, there is much evidence, detailed below, that conformational
changes are directly involved in catalysis and that the mechanism is
shared by all phosphagen kinases. Furthermore, x-ray small-angle
scattering shows that it is the combination of
Mg2+ + ADP or ATP (substrates common to all
phosphagen kinases) that elicits the conformational change and not the
phosphagen (creatine, arginine, etc.) that is specific for each enzyme
(Forstner et al., 1998).
Prior evidence of substrate-induced conformational changes in both AK
and CK include spectroscopic studies of the active site (Reed and Cohn,
1972), tryptic susceptibility differences (Lui and Cunningham, 1966),
and x-ray solution scattering (Dumas and Janin, 1983; Forstner et al.,
1996, 1998). The conformational changes are induced by the presence of
MgATP or the full set of transition state analog (TSA)
components
Mg2+, ADP, nitrate (mimicking a
phosphoryl group in transfer), and creatine (or arginine)but not by
ATP alone. The apo-enzyme is considered to be in "open"
form, changing to "closed" when the substrates bind (Forstner et
al., 1998). Although differences in the x-ray scattering indicate
substantial changes, circular dichroism and sedimentation indicate that
the secondary and quaternary structures of Mib-CK
are unchanged (Forstner et al., 1996). NMR spectroscopy indicates that
the chemistry of the reaction is not rate limiting (Rao et al., 1976).
This leaves open the possibility that all or part of the conformational
change occurs on each reaction cycle and is rate limiting. AK's
catalytic turnover (kcat 
135 s
1) (Blethen, 1972) is at the slow end of those
measured for enzymes (1 to 107
s1) (Boyer, 1998), as are others undergoing
substrate-induced conformational changes, such as tryptophan synthetase
(2 s1) and lactate dehydrogenase (1000 s1) (Rhee et al., 1997; Gerstein and Chothia,
1991; Boyer, 1998).
The substrate-induced conformational changes that have been reported in several other systems are manifested as two kinds of motion:
1. Domain movements: Yeast hexokinase gave us the first
structural paradigm for induced fit. The active site is located in a
cleft between two globular domains that closes upon substrate binding
(Bennett and Steitz, 1980). Similar mechanisms were proposed for other
kinases (Anderson et al., 1979).
2. Loop movements: In other enzymes, a flexible loop folds
over the active site after substrate binding. Paradigms of this type of change include triosephosphate isomerase (Alber et al., 1987)
and RuBisCo (Schreuder et al., 1993), in which the catalytic Lys334 follows the substrate with a movement of
13 Å to complete the active site. Substrate-induced conformational
changes can be important in allosteric control mechanisms. However, it
has been pointed out that induced fit is potentially disadvantageous,
especially for nonallosteric enzymes, because the energy required to
make the protein conformational change adds to the enthalpy of the activation barrier and reduces catalysis (Fersht, 1985). This perhaps
accounts for the rarity of large structural changes during catalysis
and for the relatively low energy barrier to such changes when they do
accompany catalysis (Gerstein et al., 1994). In the examples given, it
appears that the energy-consuming conformational changes are a
necessary cost of excluding water from the active site to stop an
unwanted side reaction. The phosphagen kinases may be like other
kinases (Anderson et al., 1979) in needing to exclude water to avoid
the waste of energy with free (solvent-mediated) hydrolysis of ATP.
In this report, differences between the TSA complex (TSAC) AK structure and those of the apo Mib-CK structure are analyzed for their implications upon the conformational changes between open and closed forms of phosphagen kinases. They suggest that substrate binding induces a 13° domain rotation and large loop conformational changes that reconfigure critical elements of the active site and close the active site to exterior solvent during the reaction.
| |
MATERIALS AND METHODS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
REFERENCES
|
|---|
Coordinates and alignment
The TSAC-AK coordinates were as described by Zhou et al. (1998)
(PDB entry 1bg0). For alignment purposes, the small domain was defined
as residues 2-99 (Met1 is disordered) and the
large (C-terminal) domain as residues 100-357. For CK, most of the
comparisons were made with the apo-enzyme (Fritz-Wolf et
al., 1996) (PDB entry 1crk). The small domain was defined as residues
1-102 and the large domain as residues 103-363. The C-terminal
fragment 364-380 is unique to Mib-CK and is
thought to be associated with membrane attachment. The four subunits in
the Mib-CK asymmetrical unit have slightly
different conformations in the flexible N-terminal loop and disordered
loop 315-325. Subunit A was used for alignment and structural comparisons.
An approximate transformation for superimposing the CK structure on AK
was available from the cross-rotation and translation functions
calculated from the AK diffraction amplitudes and CK coordinates during
the AK structure determination (Zhou et al., 1999). This was used to
improve the pairing of AK and CK amino acids beyond that available from
prior sequence alignments (Dumas and Camonis, 1993; Strong and
Ellington, 1995). The alignment was least-squares optimized, using O
(Jones and Kjeldgaard, 1997) and Lsqkab (Kabsch, 1976) as implemented
in the CCP4 suite (Collaborative Computational Project, 1994). Domains
were aligned independently. Details of how the transformations were
defined are provided in Table 1.
|
Modeling the open state of AK and calculation of radii of gyration
Radii of gyration were calculated from the atomic coordinates
following the method of McDonald et al. (1979):
|
i are the atomic number and position
vector of protein atom i and 
is the
molecular center of mass. RG for the
closed state was calculated directly from the TSAC-AK coordinates. For
the open AK state, separate large and small AK domains were superimposed on the CK structure. RG
was then calculated from the AK coordinates that had been moved as
rigid domains only, or after loops and subdomains had also been moved
to CK-like configurations. Remodeled residues included residues
169-196 and 285-328. In the case of loop 309-319, the backbone
coordinates of the aligned CK residues 315-325 were used, and no
attempt was made to model the side chains. For residues 169-196 and
the rest of 285-328, the AK regions were rigid-body-superimposed on
the aligned MibCK coordinates.
Sequence realignment
AK loop 309-319 was realigned to MibCK
315-325 with a low gap penalty to allow the catalytically important AK
Glu314 to be aligned with a similar residue, CK
Asp321. The ends of the 
-strands to which the
loop is attached are highly conserved, leading to an unambiguous
alignment that was unchanged from prior reports (Dumas and Camonis,
1993; Strong, 1995). It was a nonconserved eight-residue section of the
loop that was realigned. The AK and CK structures were completely
different, giving no indication of how the sequences should be aligned.
The prior sequence alignments could be improved by pairing
Glu314 with Asp321 by
inserting a gap between AK Glu314 and its
conserved neighbor, Gly313. There were seven
positions in which a complementary single insertion could have been
made in the CK sequence to retain the alignment in the next conserved
region. The best was located by manually calculating the alignment
scores of the seven possibilities, using the blosum62 comparison matrix
(Henikoff and Henikoff, 1992).
Structure comparison and analysis
Coordinate differences and B-factors were analyzed using COMPAR
and BAVERAGE of the CCP4 suite (Collaborative Computational Project,
1994). Interactive graphical analysis used the program O (Jones and
Kjeldgaard, 1997). Structural figures were produced with MolScript
(Kraulis, 1991), Raster3D (Merritt and Bacon, 1997), and GRASP
(Nicholls, 1992). Accessible surface areas were calculated using MSP
(Connolly, 1983). From changes in surface area, estimation of changes
in the free energy of solvation followed the method of Xie and Chapman
(1996), using atomic solvation parameters (Eisenberg and McLachlan,
1986) that were adjusted to account for the usual ambiguities in
determining atom type from 2-Å resolution electron density.
| |
RESULTS AND DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
REFERENCES
|
|---|
Homology and magnitude of the differences between the Mib-CK and AK structures
Domain orientations
If entire subunits of AK and CK are optimally superimposed as rigid bodies, the correspondence is poor: 1.7 Å for main-chain atoms. The correspondence improves substantially when the large and small domains are treated independently and when loops are excluded that are in entirely different conformations. The rms main-chain difference is then 1.0 Å for the large domain and 0.9 Å for the small domain. If the large domains of AK and CK are superimposed, then the rms main-chain difference of 4.5 Å in positions of the small domains corresponds to a 13° rotation followed by a 0.2-Å translation. The axis of rotation and other parameters of the transformation are shown in Fig. 1 and Table 1. The translational component of the motion is small (<0.25 Å; Fig. 1 and Table 1) and mostly parallel to the axis of rotation. Consistent with the x-ray scattering-based terminology of open and closed states (Forstner et al., 1996), relative to the apo CK structure (Fritz-Wolf et al., 1996), the domains of the AK structure (Zhou et al., 1998) are
closed down upon the bound substrates. The axis of rotation passes
through Pro100. This is within a long extended
chain connecting the two domains (that is not highly conserved), so the
modest bending required could be accommodated with changes to several
torsion angles near Pro100.
|
Loop conformations
There are differences in several regions, but by far the most extensive differences are in AK residues 309-319, corresponding to Mib-CK residues 315-325 (Figs. 2 and 3). Relative to the open-state apo CK, the loop of the closed-state AK is folded and twisted over the bound substrate (Figs. 2 and 4) and is very much more ordered, with an average main-chain B-factor of 11.8 Å2 compared to 90.3 Å2 (Table 2). The mean coordinate differences for the loop are 6.6 Å and 8.8 Å for main-chain and side atoms, respectively, showing essentially no structural correspondence. Differences for individual amino acids are listed in Table 2, but the largest is 15.0 Å for Glu314, which we have aligned to Asp321 in CK (Table 3), based on function in the active site (Zhou et al., 1998).
|
|
|
|
|
-strand and two small
-helices that has an rms difference of 5.1 Å (Fig. 2). Another
region (residues 169-196), consisting of a three-turn helix, an
extended chain, and a single-turn helix, is 3.8 Å (rms) closer to the
active site in the transition state analog complex. The extended chain
(residues 95-115) that runs across both domains and contains the hinge
point near Pro100 (Fig. 1) shows rms differences
of 2.7 Å. With the exception of the immediate N-terminus, these are
the regions in which the biggest differences occur (Fig. 3) and are the
focus of this study.
Evidence that the conformational differences correspond to the substrate-induced conformational change
Crystallographic evidence of conformational changes in creatine kinase
Fritz-Wolf et al. (1996) noted several features of their
Mib-CK structure that led them to believe that
the conformations of their apo and ATP-bound forms would
have to change to become active. The substrate positions were not fully
consistent with prior spectroscopic results, active-site loops had high
thermal factors, and they had different conformations in different
subunits (Fritz-Wolf et al., 1996, and references therein). The sites
of change required for activity predicted from the
Mib-CK structure are entirely consistent with the
sites of greatest difference with the active-state AK. The substrate
nucleotide base, and phosphates, differ by an rms of 2.4 Å,
and the apo-CK disordered active-site loops have a
completely different configuration in AK (see above). Furthermore,
except for the immediate N- and C-termini, the regions of largest
conformational difference are exactly those that have the highest
temperature factors in CK (Fig. 3), suggesting that these regions are
flexible in the absence of a full complement of substrates. On a
different tack, the destruction of Mib-CK crystals with the addition of TSA components and the failure to date of
cocrystallization have also been interpreted as evidence for
substantial conformational changes (Kabsch and Fritz-Wolf, 1997).
Radii of gyration
The calculated radius of gyration (RG) of the closed state AK is 20.0 Å, agreeing with values of 19.7-20.0 determined experimentally from small-angle x-ray scattering (Dumas and Janin, 1983; Forstner et al.,
1998) for Mg-ATP complexes. If the AK domains are reoriented to the
open-state CK positions (Fritz-Wolf et al., 1996) with a rigid domain
movement, there is a predicted increase in
RG to 20.4 Å. Remodeling loops
169-196 and residues 285-328 to the obviously different backbone
configurations of the open-CK structure (Fritz-Wolf et al., 1996)
increases RG to 20.8 Å. This is in
reasonable agreement with experimental values of 21.2-21.5 Å (Dumas
and Janin, 1983; Forstner et al., 1998). The remaining modest
discrepancy between calculation and experiment may be due to subtler
changes in (for example) side-chain configuration that are not obvious
when comparing different enzymes, or to unrealistic assumptions in
RG calculation, such as spherical
shape. Considering such factors, agreement of predicted and measured
RG is good and supports approximate
modeling of open-state AK structure based on the currently known AK and CK structures.
Analysis of conformational changes by comparison of different structures
Given a history of difficulties in determining the structure of the active form of CK (Kabsch and Fritz-Wolf, 1997), for the foreseeable future, the only basis for understanding the impact of
conformational changes upon the mechanism of action will be through
comparison of different structures. Fortunately, the approach is
supported not only by the radii of gyration, but by the correlation of
disorder in Mib-CK (Fritz-Wolf et al., 1996) with
the AK-CK structural differences. Furthermore, it is not be the first
time that, for practical reasons, it has been necessary to infer
conformational changes by comparing structures from different species
and quaternary structure. Prior examples include RuBisCO and
phosphoribosyltransferases (Schreuder et al., 1993; Schumacher et al.,
1996).
Significance of the conformational changes
In other enzymes, substrate-induced conformational changes have
included both loop movements and domain realignments that can be either
primarily rotations about a hinge point or "shear" translations
(Gerstein and Krebs, 1998).
Domain movements
Analysis of the transformation shows a hinge point located at Pro100 with a 13° rotation about an axis roughly in the plane of the domain interface and a small 0.2-Å translation mostly parallel to the rotation axis (Fig. 1 and Table 1). While there are some notable examples of larger hinged domain motions (e.g., adenylate kinase), the 13° AK/CK rotation is of a magnitude similar to that of most hinged motions (e.g., formate and glutamate dehydrogenases and cAMP-dependent protein kinase; Gerstein and Krebs, 1998). The hinge point is within a long stretch of extended chain,
allowing domain motion with small changes in torsion angles, unhindered by secondary structure.
Loop movements
Fig. 4 highlights the loops that undergo substantial disorder-order and conformational transitions upon binding substrate and changing from open to closed forms. The loop with the greatest change (residues 309-319 in AK and 315-325 in Mib-CK) is also critically important in catalysis. AK-Glu314 binds the substrate arginine guanidinium group, holding it in precisely the right place for optimal catalysis (Zhou et al., 1998). AK-Glu314 is also
one of two candidate bases in catalysis (Zhou et al., 1998). In prior
sequence alignments (e.g., Dumas and Camonis, 1993),
AK-Glu314 is part of an eight-residue loop
segment with no sequence conservation. Based on the apparent importance
of a carboxylate (Zhou et al., 1998), the alignment has been
frame-shifted one amino acid with single insertions/deletions to align
AK-Glu314 with
Mib-CK-Asp321 (Table 3).
AK-Glu314 appears to be at the center of the loop
motion, with the largest main-chain difference of 15.0 Å for
Glu314, aligned as Asp321
in CK (Table 3). The parallels with RuBisCO's
Lys334 are extraordinary (Schreuder et al.,
1993). Motions of an active-site loop in each exceed 12 Å; there are
smaller motions or disorder-order transitions in other loops, and
either domain or subunit realignments. Focusing on the active-site
loop, as in RuBisCO and in contrast to triose phosphate isomerase
(Joseph et al., 1990), the loop movement is more than a rigid hinged
flap movement and involves the loop adopting a very different backbone
conformation. In both RuBisCO and AK, salt-bridge/hydrogen bond
interactions with a charged substrate group appear to lead the
conformational change. In RuBisCO an enzyme lysine interacts with a
substrate phosphate (Schreuder et al., 1993), and in guanidino kinases,
an enzyme carboxylate interacts with a guanidinium. In both cases, the
residue leading the change appears to be critical for completion of the active site and likely has a direct catalytic role.
In both RuBisCO and AK, the direct interactions of the moving loop with
the substrate may ensure that the conformational change is completed
only when substrate is bound. In AK, there are multiple interactions of
loop 309-319 with the substrates. Arg309 is one
of the five positively charged arginines that bind the negatively
charged and phosphates of ATP. With
Arg309 interacting with the nucleotide and
Glu314 with the substrate arginine, loop 309-319
bridges both substrates and may be partially responsible for bringing
them into correct juxtaposition. Furthermore, it is only with movement
of the loop (including side-chain movements of 18 and 6 Å for
Glu314 and Arg309) that the
active site is configured. Like RuBisCO, HGXPRTases, and protein
tyrosine phosphatase (Schreuder et al., 1993; Schumacher et al., 1996;
Schubert et al., 1995), phosphagen kinases support one of Koshland's
predictions (Koshland, 1958) that induced fit may initiate changes that
complete the active site.
Loop 309-319 is part of a larger C-terminal region (from residue
~281 to the C-terminus) that could be considered a subdomain undergoing a plastic hinged rotation. With an axis running near residues 280 and 300, most of the subdomain is rotated ~7° toward the active site. This appears to be accomplished with small changes in
the torsion angles of a number of amino acids. The region includes the
three C-terminal strands of the large domain sheet. The conformational difference is accommodated by an increase in the natural twist of the
sheet, with progressively larger movements from strand 4 to strand 8 of
the eight-stranded sheet. The conformational changes are plastic in
that there are local variations on the rigid rotation. The rigid motion
accomplishes only part of the movement of loop 309-319, which also
undergoes the larger nested hinged and refolding motion described
above. Differences near AK residue 295 are also larger (~8 Å), with
a different backbone conformation that could be as much the result of
quite different AK and CK sequences as of the presence of substrates.
The C-terminal helix is rigidly rotated through a smaller angle
than the rest of the subdomain (5° compared to 7°), which may be
due either to substrate binding or to the C-terminal
membrane-associated extension present in MibCK
(Kabsch and Fritz-Wolf, 1997) but not AK.
The next largest difference is a near-rigid shear translation of
169-196. A two-turn helix slides ~3.8 Å over the -sheet, and the
following extended chain moves toward the active site, making some
contact with the substrate nucleotide, and to other active-site loops
that have changed configuration. The conformational change seems to be
accommodated in a typical way (Gerstein and Krebs, 1998) by adjusted or
changed side-chain rotamers in the underlying -sheet, allowing the
hydrophobic side chains of the helix to slide past. The extended chain
following the helix parallels the end-to-end positions of the
substrates (Fig. 1). There are a few van der Waals contacts with the
substrates, but only one specific interaction, a hydrogen bond (3.0 Å)
between His185 N
and the
O2' hydroxyl of the nucleotide ribose. From the same region, the C-terminal end of the two-turn helix, there is a salt
bridge between Asp183 and
Arg312 of the closing active-site loop. Together,
these interactions may drive the conformational change. There is one
more possible specific interaction between the loops that move together
over the active sitea possible hydrogen bond between the side chain of Lys189 and the backbone carbonyl of
Glu314. Such interactions may help to zip the
occupied active site pocket closed. They also help to reduce the
solvent accessibility. Lys189 restricts access to
the Mg2+ region of the active site. The possible
Lys189-Glu314 hydrogen bond
and a van der Waals interaction between the side chain of
Leu195 and the C of
Glu225 may help to precisely position the
carboxylates of Glu314 and
Glu225 for their catalytic interactions with the
substrate guanidinium (Zhou et al., 1998).
With crystallographic evidence of motion in
MibCK, Kabsch and Fritz-Wolf (1997) have
predicted that regions corresponding approximately to the AK regions
309-319 and 169-196 described above might be changed in the active
configuration. They also suggested that a region corresponding to AK
residues 60-68 might move. It is moved closer to the active site by
the rigid-body small-domain rotation, but additional local
conformational differences are modest (Fig. 3). However, this loop is
the site of an insertion in enzymes like CK that work upon smaller
substrates (Suzuki et al., 1997b). The insertion has particularly high
B-factors, which might indicate more extensive motion in CK (Fritz-Wolf
et al., 1996) than AK.
Entrance to the active site
The molecular surface and electrostatic properties of the presumed open and closed state structures of AK are shown in Fig. 5. In the open state, the active site is wide open, and there is a region of strong positive charge that will bind the negatively charged ATP phosphates. With substrate bound, but with the open enzyme configuration, the accessible surface area of the-phosphate is ~10 Å2. With the change to
the closed state, the -phosphate is covered and becomes completely
inaccessible to solvent. Thus, as in some other kinases (Anderson et
al., 1979, and see below), one of the effects of the conformational
change is to exclude water and reduce the chance of wasteful hydrolytic
loss of ATP. By both configuring the catalytic elements and
simultaneously excluding solvent with a substrate-induced
conformational change, the enzyme appears to ensure that all ATP is
used for productive phosphoryl transfer. Closing of the enzyme active
site is mostly due to the folding down of a "flap" made by loop
309-319, but there are also contributions from the domain movement and
the movement of region 169 to 196.
|
Potential impact of conformational differences upon quaternary structure
Horseshoe crab AK does not form octamers as does MibCK. One of the MibCK interdimer contact points identified by Kabsch and Fritz-Wolf (1997) contains MibCK W264,
which is known to affect octamer dissociation (Gross and Wallimann,
1995; Gross et al., 1994), that is present only in octameric CKs, and
is within a region deleted from the AK sequence (MibCK residues
260-264). Differences between the apo-CK and TSAC-AK
structures provide a rationale for octamer dissociation in
MibCK. The second interdimer contact point is at
the N-terminus (Kabsch and Fritz-Wolf, 1997). Considering only a rigid
small domain reorientation, the N-terminal contact region (which is far
from the rotation axis) would move more than 6 Å, widening the
separation of the contact points and perhaps straining the octamer interactions.
Biological implications and conclusion
The phosphagen or guanidino kinases catalyze the buffering of
cellular ATP levels through the reversible transfer of the
-phosphoryl between ATP and a guanidino substrate (Wallimann et al.,
1992). They constitute one of the most intensively studied enzyme
families (Stroud, 1996). Our comparison of the substrate-bound
transition-state complex structure of arginine kinase (Zhou et al.,
1998) with that of apo-creatine kinase (Fritz-Wolf et al.,
1996) indicates that large conformational changes accompany substrate binding.
Role of the conformational changes
Movement of the 309-319 loop brings catalytic residues (e.g., E314) into contact with the reactive substrate groups. This component of the induced fit may minimize wasteful ATP hydrolysis by configuring the active site only when the appropriate specific substrates are present (Koshland, 1958). What about the domain movements and other
changes? Rationales of induced fit proposed for other enzymes (e.g.,
Anderson et al., 1979) are likely to be relevant here. Solvent
accessibility of the substrates is substantially reduced (Fig. 5),
further reducing wasteful ATP hydrolysis and reducing the dielectric.
As in hexokinase (Anderson et al., 1979) and adenylate kinase (Pai et
al., 1977), a low dielectric likely enhances the nucleophilicity of the
phosphoryl acceptor and increases the pK of potential carboxylate
catalytic bases (E225 and E314; Zhou et al., 1998). Likely associated
with the conformational changes is the sharp contrast between the
disordered ADP and active-site loops of the open-form
MibCK binary complex (Fritz-Wolf et al., 1996)
and the low B-factors of substrates and active site in the AK
transition-state analog complex (Zhou et al., 1998). It appears that
the folding of loop 309-319, the rotations of the small domain and
C-terminal subdomain, and the shearing motion of 169-196 together envelop the substrates, fixing them in catalytically optimal positions with lower entropy. Substrate alignment is precise enough (Zhou et al.,
1998) that it could be a significant part of the catalytic enhancement
of rate (Dafforn and Koshland, 1971; Jencks and Mage, 1974). The
participation of induced fit changes in catalysis may prove to be more
generally applicable as more transition state structures of
multimolecular enzyme reactions become available.
| |
ACKNOWLEDGMENTS |
|---|
We thank Wolfgang Kabsch for access to CK atomic coordinates before data base release; Golapakrishnan Parthasarathy, Scott Strong, Thayumanasamy Somasundaram, and Eric Blanc for participation in earlier parts of the structure determination; and Thayumanasamy Somasundaram for help in producing figures.
We gratefully acknowledge the National Science Foundation for support of methods development that allowed structure determination (MSC; BIR94-18741), and the American Heart Association, Florida Division (MSC; 9701728), the National Institutes of Health (MSC; 1R29GM55837), the National Science Foundation (WRE; IBN96-31907), and the Lucille P. Markey Charitable Trust for support of evolutionary and structural studies of arginine kinase.
Coordinates and structure factors for the arginine kinase transition state analog complex have been deposited with the Protein Data Bank (entry 1bg0).
| |
FOOTNOTES |
|---|
Received for publication 28 June 1999 and in final form 30 November 1999.
Address reprint requests to Dr. Michael S. Chapman, Institute of Molecular Biophysics, Florida State University, Tallahassee, FL 32306-4380. Tel.: 850-644-8354; Fax: 850-561-1406. E-mail: chapman{at}sb.fsu.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
MATERIALS AND METHODS
RESULTS AND DISCUSSION
REFERENCES
|
|---|
a square protein.
Curr. Opin. Struct. Biol.
7:811-818[Medline].
Biophys J, March 2000, p. 1541-1550, Vol. 78, No. 3
© 2000 by the Biophysical Society 0006-3495/00/03/1541/10 $2.00
This article has been cited by other articles:
 |
|
 |
|
  C. Tang and B. Ma GLYCOSIDATION OF AN ENDOTHELIN ETA RECEPTOR ANTAGONIST AND DICLOFENAC IN HUMAN LIVER MICROSOMES: AGLYCONE-DEPENDENT UDP-SUGAR SELECTIVITY Drug Metab. Dispos., December 1, 2005; 33(12): 1796 - 1802. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Compaan and W. R. Ellington Functional consequences of a gene duplication and fusion event in an arginine kinase J. Exp. Biol., May 1, 2003; 206(9): 1545 - 1556. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Borders Jr., K. M. MacGregor, P. L. Edmiston, E. R.K. Gbeddy, M. J. Thomenius, G. B. Mulligan, and M. J. Snider Asparagine 285 plays a key role in transition state stabilization in rabbit muscle creatine kinase Protein Sci., March 1, 2003; 12(3): 532 - 537. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. S. Yousef, S. A. Clark, P. K. Pruett, T. Somasundaram, W. R. Ellington, and M. S. Chapman Induced fit in guanidino kinases--comparison of substrate-free and transition state analog structures of arginine kinase Protein Sci., January 1, 2003; 12(1): 103 - 111. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Suzuki, H. Fukuta, H. Nagato, and M. Umekawa Arginine Kinase from Nautilus pompilius, a Living Fossil. SITE-DIRECTED MUTAGENESIS STUDIES ON THE ROLE OF AMINO ACID RESIDUES IN THE GUANIDINO SPECIFICITY REGION J. Biol. Chem., July 28, 2000; 275(31): 23884 - 23890. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
aCK;
Fritz-Wolf et al., 1996
, rms main-chain differences (Å) for each residue; ×,
Mean B-factors (Å2) for AK residues; +, CK (Fritz-Wolf et
al., 1996
