|
Figure 2.
Figure 2. (a) 15N-1H FHSQC spectrum of BeF[3]-CheY along
with (b) superpositions of backbone N, C^a, and C' coordinates
for BeF[3]-activated CheY and (c) comparison of BeF[3]-activated
and inactive CheY structures shown in stereoview. (a) In the
FHSQC, spectrum peaks are labeled with residue numbers.
Unassigned backbone resonances are labeled UA. Pairs of
side-chain NH[2] resonances are connected by horizontal lines.
Signals enclosed in boxes are folded in the 15N dimension. (b)
The 27 structures of BeF[3]-activated CheY. Backbone coordinates
for residues in the five helices and five-stranded b-sheet were
superimposed. (c) Superposition of the 27 structures of
BeF[3]-activated CheY (blue), with apo X-ray [Volz and Matsumura
1991] (gold), magnesium-bound X-ray [Bellsolell et al 1994]
(red), and mean magnesium-bound NMR [Moy et al 1994] (magenta)
structures. Superposition included backbone coordinates for
residues in H1, H2, b1, b2, and b3. Considering the mean
coordinates obtained from the family of magnesium-bound [Moy et
al 1994] and BeF[3]-activated NMR structures, backbone
superposition of H1, H2, b1, b2, and b3 yields an r.m.s.d. value
of 2.4 Å for the backbone coodinates of residues in H3,
b4, H4, b5, and H5. The Figure was produced with the program
MOLMOL [Koradi et al 1996]. Uniformly 15N and 15N/13C-labeled
samples were prepared by growth in M9 minimal medium
supplemented with biotin and either [15N]ammonium chloride or
[15N]ammonium chloride and [13C]glucose. The BeF[3]-activated
sample conditions were 4 mM CheY, 16 mM BeCl[2], 100 mM NaF, 20
mM MgCl[2], at pH 6.7, and 10 % 2H[2]O. NMR spectra were
recorded on AMX 600 and DRX 500 NMR spectrometers at 25 °C.
Backbone resonances were assigned with 3D 15N NOESY-FHSQC
[Talluri and Wagner 1996], HNCACB [Wittekind and Mueller 1993],
CBCA(CO)NH [Grzesiek and Bax 1992a], and HNCA [Grzesiek and Bax
1992b] spectra. Side-chain aliphatic 13C/1H pairs were assigned
with 3D 15N TOCSY-HSQC [Driscoll et al 1990], HCCH-TOCSY [Kay et
al 1993] and CBCA(CO)NH spectra. In each of the experiments
above, purge-type pulsed-field gradients were used to suppress
artifacts and the solvent signal [Bax and Pochapsky 1992].
Aromatic assignments were obtained from DQF-COSY [Rance et al
1983] and 13C/1H HMQC spectra [Bax et al 1990]. The assignment
process was also aided by making reference to published chemical
shifts for CheY [Bruix et al 1993 and Moy et al 1994]. Phi
torsion angle restraints were obtained from a 15N HMQC-J
spectrum [Kay and Bax 1990]. Stereospecific assignments for Val
and Leu methyl groups were obtained by comparison of ct-HSQC
spectra of uniformly 13C-labeled and 10 % uniformly 13C-labeled
samples [Neri et al 1989 and Szyperski et al 1992]. x1
restraints for the Val, Ile, and Thr residues were obtained from
ct-HMQC-J spectra [Grzesiek et al 1993 and Vuister et al 1993a].
NOEs identified in 3D NOESY-FHSQC, 4D 13C/15N HMQC-NOESY-FHSQC
and 4D 13C/13C HMQC-NOESY-HMQC (all recorded with a 100 ms
mixing time) [Vuister et al 1993b] spectra were classified as
strong (2.9 Å upper distance limit), medium (3.3 Å),
or weak (5.0 Å). A total of 972 non-trivial NOE restraints
(213 intraresidue, 271 sequential, 238 medium-range, and 250
long-range) were used as input to DYANA [Guntert et al 1997],
along with 78 phi torsion angle restraints and 17 x1 restraints
for the Val, Ile, and Thr residues. Once sets of 20 (of 60)
structures reached a backbone r.m.s.d. of 1 Å, 47 hydrogen
bonds (94 upper and 94 lower distance restraints (H-O distance
restraint 1.8-2.0 Å; N-O 2.7-3.0 Å)), identified on
the basis of slow amide proton exchange rates (protection
factors greater than 75) and short donor/acceptor distances were
included in the calculations. Structures resulting from DYANA
calculations with a pseudoatom (van der Waals radius 2.5
Å) corresponding to BeF[3]^ - attached to the side-chain
of Asp57 resulted in a backbone r.m.s.d. value of only 0.4
Å when compared to structures without the additional
pseudoatom. The 27 of 60 structures (BeF[3]^ - pseudoatom not
included) with residual target function values less than 1.0
Å2 (Table 1; target function before energy minimization
was 0.3(±0.2) Å2) were subjected to restrained
energy minimization using the AMBER94 forcefield [Cornell et al
1995] implemented in the program OPAL [Luginbuhl et al 1996].
Conjugate gradient minimization (1500 steps) included bond,
angle, dihedral, improper dihedral, van der Waals,
electrostatic, NMR distance, and NMR torsion angle terms. The
minimization was performed in a shell of water at least 6
Å thick, with the dielectric constant set to 1, and with
no cut-off for non-bonded interactions. PROCHECK analysis
[Laskowski et al 1993] of the structures revealed that 99 % of
the residues fall within the allowed or generously allowed
regions of the Ramachandran map. The 27 energy-minimized
structures are used to represent the solution structure of CheY
complexed with beryllofluoride and magnesium.
|
