NUCLEAR MAGNETIC RESONANCE 119
were converted into distance constraints using the CALIBA module in
DYANA (see reference 25b ). The structural calculation began by subjecting
200 randomly generated structures to a simulated annealing torsional angle
dynamics procedure. The PBD: 1 GIW structural data were used again to set
the Fe – S distance for the Fe(II) heme ’ s met80 sulfur ligand at 2.50 Å. Consis-
tent violation for this bond during DYANA structural calculation caused the
researchers to relax the Fe – S bond length by 0.25 Å. Indeed, the fi nal met80
sulfur ligand – Fe(II) bond length is reported as S δ – Fe = 2.68 Å , an atypically
long Fe – S bond length. Stereospecifi c assignments for 41 diastereotopic fer-
rocytochrome c protons were made using the HABAS and GLO MSA modules
in DYANA. The fi nal calculation was performed with 2232 distance restraints
and 73 angle restraints on 750 randomly generated conformers. A family of 30
structures was used for subsequent energy minimization.
Restrained energy minimization (REM) was carried out using the SANDER
module in the AMBER 5.0 molecular modeling package. AMBER (Assisted
Model Building with Energy Refi nement) refers to two things: (1) a molecular
mechanical force fi eld for the simulation of biomolecules and (2) a package
of molecular simulation programs which includes source code and demos.^27
The ferrocytochrome c heme was generated using the coordinates and angles
for heme b, as available in AMBER, and was then modifi ed to represent heme
c. Force fi eld parameters for all residues and the heme were taken from the
AMBER database. The conformation of the met80 ligand was determined by
the NMR restraints used in the DYANA structure calculation. The charges on
the heme were obtained by an ab initio calculation using Gaussian 98. The
hypothetical model complex used a diethyl thioether to represent met80, a 2 -
methylimidazole to represent his18 and CH 3 – S – groups attached to the CAB
and CAC heme carbon atoms. (See Figures 7.25 and 7.32 to see representa-
tions of heme c and its atom identifi cation system.) The coordinates used were
again taken from the 1GIW NMR structure. The distance and angle restraints
for the energy minimizations were imposed using a square - well penalty func-
tion with force constants of 20 kcal mol − 1 Å − 2 and 70 kcal mol − 1 rad − 2 , respec-
tively. The energy minimizations were performed until the Erms was less than
0.4 kcal mol − 1. The average structure (PDB: 1LC1) was calculated from the 30
REM structures (PDB: 2LC1) obtained by this procedure. Finally, structure
validation for stereochemical quality was performed using PROCHECK and
PROCHECK - NMR. Figure 7.34 in Section 7.7 visualizes the ferrocytochrome
c structure resulting from these experiments and calculations.
To further extend the utility of structural methods, researchers compare
solid state X - ray crystallographic and solution - state NMR structures to defi ne
important differences. For instance, the Bertini group has studied the enzyme
matrix metalloproteinase 12 (MMP12), in the presence of its inhibitors.^28
Matrix metalloproteinases (MMPs) are involved in extracellular matrix deg-
radation, a fundamental step in tissue remodeling and repair. There are a great
variety of enzymes of this type, the one studied here is one of many found in
humans. Most MMPs have three domains: (1) a prodomain that is removed