Modeling Macromolecular Complexes: A Journey Across Scales - Modeling in Computational Biology and Biomedicine - page 4

In both cases, the result is a 3D density map, where each voxel encodes the density
of matter. This density is in general very noisy due to the low electron doses used
to avoid damaging biological specimens. Choosing a density level for contouring
a surface (called the envelope) enclosing the model is non trivial, as the intensity
is generally high for globular domains of the proteins but low for unstructured
regions such as linkers connecting these domains. Typically, low (less than 10 A,
domains visible) to medium (around 5 A, secondary structure elements visible)
resolutions are achieved in cryo-EM. In favourable cases, fitting existing and/or
modelled structural elements into such maps yields atomic resolution models. The
PDB or its sister data bank, the EMDataBank ([ 40 ], see http://emdatabank.org ,
currently contains CryoEM models for over 1,000 biological complexes. These
resources contain detailed structural information ranging from the A scale, relevant
to small molecules and individual amino acids, to hundreds of nanometers and
beyond for large complexes.

1.1.2 Dynamics of Macromolecular Systems

Besides the 1,000-fold range of molecular dimensions, even more pronounced and
challenging differences exist in the characteristic timescales describing biological
processes. Biological timescales range from picoseconds for localized side-chain
transitions in amino acids at a protein surface, up to hundreds of nanoseconds or
microseconds for slow loop rearrangements, and from milliseconds to hours for
folding reactions and global conformational changes [ 1 ]. Different experimental
techniques provide dynamic information. The temperature factors in a crystal struc-
ture are obtained along with the atomic coordinates. Each reflects the spatial
dispersion of the electron density around a given atomic position and thus the
atom's mobility in the crystal, although static disorder and errors contribute as
well. Conformational dynamics is also measured using methods such as time-
resolved spectroscopy, which can provide exceptional detail on changes in structural
features such as bond lengths, coupled with some method of rapidly initiating a
change, such as laser temperature jump or photo-dissociation of a ligand. NMR,
in addition to macromolecular structure resolution, furnishes dynamic information:
in an external magnetic field, the interaction between two protons of a molecule
influences their rates of transition between magnetic energy levels. Such effects are
both time- and conformation-dependent, and can be exploited in different ways. For
example, in partially orienting solutions, incomplete rotational averaging allows one
to extract comprehensive information concerning conformational dynamics of the
macromolecule [ 59 ].
Understanding such dynamics in detail often entails the use of numerical simula-
tions. As mentioned above, for the processes we are considering here, no chemical
bond making or breaking takes place, for which quantum mechanical descriptions
would be necessary. The dynamics that occur in protein folding, conformational
changes, and association principally involve changes in weaker, non-covalent