Biomacromolecular speleology

Gerard J. Kleywegt & T. Alwyn Jones
Department of Molecular Biology
Biomedical Centre, Uppsala University
Uppsala - Sweden

Most substrate-enzyme and ligand-receptor interactions in proteins take place inside clefts or cavities. Studying cavities ("biomacromolecular speleology", for short) may give insight into the mechanism of such interactions and might thereby help in the design of new ligands, substrates or inhibitors.

Many programs exist with which one may detect and/or delineate and/or measure and/or display cavities (see ref. 1 and papers cited therein). However, since we needed a program which does all of this, does it quickly and interfaces to O (2), we were forced to write our own: VOIDOO (1). Since its conception, about a year ago, VOIDOO has evolved into a multi-functional tool which is in use on a routine basis in our lab and elsewhere.

In brief, VOIDOO has the following capabilities:

* detection of a specific void or all voids inside a biomacromolecular complex
* detection of certain cavities which are connected to the "outside world"
* delineation of cavities (i.e., finding their extent in space)
* measurement of cavity volumes
* generation of plot files for O which enable visualisation of cavities
* generation of molecular surface plot files for O
* measurement of molecular volumes

Detecting cavities which are connected to the "outside world" is non-trivial. We have implemented a method (1) which we call atomic fattening: if a certain cavity is "open" while the atoms have their normal Vanderwaals radii, we gradually increase all radii, until the cavity either vanishes or becomes a real void. In order to discern these different types of cavity, we use the following operational definitions:

* a void is a cavity which is completely surrounded by the protein and which is therefore closed off from the outside world;
* an invagination is a cavity which is connected to the outside world, but which would be closed off if the atomic radii were increased;
* a pocket is a cavity which is connected to the outside and which cannot be closed off by increasing the atomic radii.
VOIDOO will detect voids and invaginations, but it cannot pick pockets. Note that the definition of the extent of an invagination is subjective and therefore more or less arbitrary (where does the cavity end and the outside world begin?).

VOIDOO can be set to measure and display cavities in three different modes:

(1) Vanderwaals cavity: the cavity comprises the complement of the Vanderwaals surface of the surrounding atoms;
(2) Probe-accessible cavity: the cavity comprises all of space that can be accessed by the centre of a probe sphere (default);
(3) Probe-occupied cavity: the cavity comprises all of space that can be occupied by a probe sphere (slow).

The input to VOIDOO consists of three parts:

(1) a library which defines which residue types are considered to be part of the protein (or other molecule) as well the Vanderwaals radii of various atom types;
(2) an ordinary PDB file;
(3) parameters which determine what the program should do and how.

An example of part of a library file is given below:

REMA *** AMBER van der Waals radii ***
ELEM ' N' 1.75
ELEM ' C' 1.85
REMA CH2 = 1.925
SPAT 'SER* CB ' 1.925
REMA *** allowed residue types ***
We distribute two protein-oriented libraries, one containing Vanderwaals radii taken from the Amber force-field and one containing MS-like radii. Libraries for other types of molecules (DNA, oligosaccharides etc.) are easy to make.

The output of VOIDOO, in a cavity-detection run, also consists of three parts:

(1) normal screen output regarding the program's operation;
(2) a log file containing a summary of the results (extent of each cavity, its volume, its centre of gravity and a list of non-protein atoms that lie inside it as well as protein atoms that border on it);
(3) files for use with O, namely for each cavity: a map file which can be displayed and a macro which draws all residues inside the cavity as well as those that line the surface of the cavity. The map files are written in EZD-format (EZD ~ "easy density"; this is a formatted ASCII file which can be converted into DSN6-format or read into O directly).

Figure 1 is an example of an O display, showing the cavity as a chicken-wire contour (on SGI workstations, cavities may also be rendered as semi-transparent surfaces), the ligand inside the cavity (fat lines) and the residues that surround it. This example pertains to cellular retinol-binding protein, whose structure was recently solved in our lab (3).

VOIDOO is one in a series of "O-dalisques", i.e. programs that work in conjunction with O. VOIDOO runs on SGI, ESV and DEC ALPHA/OSF1 workstations. For more information, contact TAJ (preferably via E-mail: "").

(1) G.J. Kleywegt & T.A. Jones, "Detection, delineation, measurement and display of cavities
in macromolecular structures", submitted for publication.

(2) T.A. Jones, J.Y. Zou, S.W. Cowan & M. Kjeldgaard, "Improved methods for building
protein models in electron density maps and the location of errors in these models", Acta
Cryst. A47 (1991), 110-119.

(3) S.W. Cowan, M.E. Newcomer & T.A. Jones, "Crystallographic studies on a family of
cellular lipophilic transport proteins. Refinement of P2 myelin protein and the structure
determination and refinement of cellular retinol-binding protein in complex with all-trans-
retinol", J. Mol. Biol. 230 (1993), 1225-1246.

USF Latest update at 12 February, 1998.