|
Figure 2.
Fig. 2. NMR solution structure of the Int^N–DNA complex.
(a) A cross-eyed stereo view showing the ensemble of 20
lowest-energy structures of the Int^N–DNA complex. The protein
(amino acids 1 to 55) and DNA backbone (nucleotides Cyt2 to
Ade10, and Thy15 to Gua23) are shown in blue and red,
respectively. (b) Ribbon drawing of the lowest-energy structure
of the complex. The strands in the beta-sheet and the helices
are labeled. The view in the left image is identical to that
shown in (a). The amino-terminal portion of the protein that
becomes ordered upon binding DNA is colored green. The solution
structure of the Int^N–DNA complex was determined in two
stages. The structure of Int^N in the complex was determined
using the ATNOS/CANDID software package, which identifies NOE
distance restraints by automatically assigning the NOESY NMR
data.^[24]^ and ^[25] Input spectra for the calculations
included a 3D ^13C-edited NOESY spectrum recorded using the
sample dissolved in 100% D[2]O and a ^15N-edited NOESY spectrum
of the sample dissolved in water. Chemical shift assignments for
residues Met1–Asp11 were excluded from the ATNOS/CANDID
calculations because long-range NOE signals from this part of
the protein were sparse and the software tended to misassign
these signals. After seven rounds of calculations, CANDID
yielded a converged bundle of conformers representing the
structure of Int^N. In a separate set of calculations, the
program NIH-XPLOR was used to calculate the structure of the
bound DNA molecule. Distance restraints for the DNA were
obtained by manually assigning 2D F1,F2 ^13C filtered NOESY
spectra of the complex. In addition, the structures were refined
using dihedral angle restraints obtained from the program TALOS
and loose DNA dihedral angle restraints for the DNA. The latter
maintained the DNA molecule in a B-form conformation and
facilitated convergence, but otherwise did not alter the
structure of the complex. NIH-XPLOR was then used to calculate
structures of the complex.^26 The structure was calculated using
the previously determined structure of the DNA molecule and the
protein in its unfolded state. The initial docking calculations
made use of a full set of distance restraints for the DNA and
protein, as well as a limited number of intermolecular NOEs to
orientate the protein and the duplex. The resultant structure
was then refined in an iterative manner by manually inspecting
the NMR data. During the refinement the program, QUEEN was used
to sort NOE restraints by decreasing information content.^27 The
50 most significant restraints were checked manually and this
process (restraint sorting by QUEEN and manual restraint
checking) was repeated until all of the most significant
restraints were correct. At the final stages of refinement,
intraprotein hydrogen bonds in regions of regular secondary
structure were identified by inspecting NOE data for
characteristic patterns. In addition to standard energy terms to
maintain appropriate covalent geometry and to account for
distance and dihedral angle data, a mean force potential was
employed to improve the structure of the DNA molecule.^[17]^ and
^[28] The final calculations produced 200 structures, 84 of
which completely satisfied the experimental data.
|
