 |
|
PQS Protein Quaternary Structure Query Form at the EBIQuaternary Structure is defined as that level of form in which units of tertiary structure aggregate to form homo- or hetero-multimers. Consideration of the presence of a quaternary state is important in the understanding of a protein's biological function. The Protein Quaternary Structure file server ( PQS ) is an internet resource that makes available coordinates for likely quaternary states for structures contained in the Brookhaven Protein Data Bank (PDB)[ref. 1] that were determined by X-ray crystallography. The crystallographic experiment on a particular macromolecule yields a set of coordinates that are not independent of the crystallographic symmetry (space group and unit cell). The deposited coordinates describe those atoms required for the purpose of refinement against the observed experimental data (i.e. the structure factors). However, these unique coordinates may not necessarily describe the complete molecule under study or may include multiple copies of the molecule. In a PDB entry the deposited coordinates usually consist of the contents of the asymmetric unit, (ASU), [i.e. that fraction of the crystallographic unit cell which has no crystallographic symmetry], from which the coordinates of the whole crystal system may be generated. An automatic procedure has been devised to recognise where multiple copies exist, and/or where symmetry is required to generate coordinate sets that potentially fully describe the particular macromolecule studied in the X-ray diffraction experiment. There are several possiblities for a relationship between the deposited crystallographic unique coordinates and the macromolecule studied: Method: A schematic outline of the method used is presented in Figure 1. For each individual PDB entry, all inter-chain atomic contacts less than 3.7 Ang for the required spacegroup symmetry operations, including each of the 27 possible unit cell translations per symmetry operation are calculated using modified CCP4[ ref. 2] and WHATIF [ref. 3] programs. A potential quaternary assembly is then built up by the progressive addition of monomeric chains that are considered to contribute to the assembly. Chain selection is based on the number of inter-chain contacts found and the number of residues in each chain. The procedure is recursive, allowing for the detection of quaternary structures where the contents of the ASU are not in contact will all other symmetry related members of the final assembly. Entries found to be monomeric are not further proccessed. All of these potential assemblies are available from the file server with annotation automatically assigned from a second process. The second step is an attempt to determine if the protein contacts found are specific (a true macromolecular oligomer) or non-specific (crystal packing). The structural characteristaion of protein-protein interfaces has been studied by many authors[refs. 4-8], and extensive reviews have been recently published[refs 9,10]. These authors have used many properties, including hydrophobicity, shape analysis, residue preferences, and number of hydrogen bonds, to characterise a protein-proein interface. To discriminate between crystal packing and a functional protein-protein interaction, the size of the solvent accessible surface area (asa)[ref. 11] buried in the interaction has been commonly used. The asa describes the extent to which a protein can form contacts with water. A DeltaASA was calculated for each potential quaternary complex found in the first step (Figure 1) as the difference between the sum of the asa of each of the partners contained within an assembly and the asa calculated for the complete assembly. Here DeltaASA represents this difference averaged over the number of chains that associate to form the complete assembly. For PDB entries where the quaternary state is known, the difference in accessible surface area per chain ranges from c.a. 370 to 4750 Ang**2 for homo-dimers and from c.a. 640 to 3230 Ang**2 for hetero-dimers[ref. 9]. The automatic file server described here uses a cutoff of 400 Ang**2 per chain as the difference in asa as one of the tests to classify the found protein-protein interactions into either crystal packing or a likely quaternary biological assembly. The full discrimination step uses an empirical weighted score with contributions from the DeltaASA, number of buried residues at the interface, a Delta-solvation energy of folding[ref. 12] number of salt bridges at the interface and the presence of di-sulphide bridges. FIGURE 1a A simplified scheme of the method used to derive and assign annotation to the macromolecular quaternary files.  FIGURE 1b 
Checking with On-line Annotation: Protein-protein dimeric interactions for the July 1993 version of the PDB (974 protein structures) have been reviewed[ref. 9]. In this review, protein dimers were identified by a combination of inspection of the individual PDB entries and by consulting the original journal articles. The PDB as of December 1997 contains 6739 structures, and it is impossible to check the literature for each entry to try to ascertain the 'true' oligomeric state. An automatic check was done by comparing the derived oligomeric state with annotation within the PDB entry and with L.Walsh's annotated listing [ref: 13 download a copy here] together with a list for nucleic acid complexes supplied by J. Westbrook from the Nucleic Acid Data Base group. Results from this check are illustrated in Figure 2 for the generated potential homo-dimers. The NCBI PubMed abstracts (http://www3.ncbi.nlm.nih.gov/PubMed/) were not searched for relevant annotation. FIGURE 2 A scheme showing the number of potential homo-dimers generated and the availability of on-line annotation to validate the automatic assignments. The listed number of mismatched dimers includes many examples similar to 1ASO28 (see text). 
No universal rule was found that could be used to differentiate a true oligomeric state and a crystal packing artefact unambiguously. Some of the dimers generated by this automatic process have a DeltaASA that allows the complex to be assigned as biologically significant, whereas the molecule is known to be monomeric. For example structures of monomeric T4 lysozyme are frequently assigned as dimeric, e.g. for the T4 lysozyme mutant 119L[ref. 14], DeltaASA is 885 Ang**2 per chain (Figure 3a). For this type of crystal packing interaction to give a pseudo-dimer, the surface involved may represent some unknown biological function, as suggested previously by other workers[ref. 15]. A further complication is that dimers found in the first pass but rejected in the check step are sometimes listed as true dimers in the PDB entry. For example, the dimeric form of Erabutoxin, 6EBX[ref. 16] is rejected by this procedure with a DeltaASA of only 290 Ang**2 per chain upon dimer formation. FIGURE 3 (a) PDB Entry 119L. An example of strongly associated crystal packing. The deposited molecule is shown in blue shades with a crystal symmetry related "dimer-like" partner in red shades; (b) PDB Entry 2DNJ. An example of a Protein/dna crystal symmetry related complex. The contents of the ASU are represented by the yellow/purple shaded protein ribbon and the green shaded cpk model of the dna chains; (c) PDB Entry 1QRD. The deposited coordinates are shown as non- crystallographic symmetry related crystal packing; (d) Entry 1QRD. One of the two independent crystal symmetry related biological dimers. The protein molecule shown in green shades is in the same orientation as the similarly shaded molecule in figure 3c; (e) PDB Entry 1GTO. An example of a stable crystal packing arrangement that is accepted by the file server as a quaternary state but may not have a biological function; (f) PDB Entry 1MVA. showing for an icosahedal symmetry viron particle the five-fold non- crystallographic pentamer. Figures 3a-f were drawn using molscript v2 by P.J. Kraulis (http://www.avatar.se/molscript) 
For some examples there is no readily available on-line information to validate the oligomeric state, including are the hexameric assemblies generated for both the nucleoside diphosphate kinase 1LWX[ref. 17] (3 chains deposited) and the inorganic pyrophosphatase 1IGP[ref. 18] (1 chain deposited). In addition, some of the available online annotation is in error, e.g. for the R-Phycoerythrin structure 1LIA[ref. 19] a dodecamer [A6B6] is generated, while the PDB entry describes the molecule as a hetero-octamer. Quinone reductase 1QRD[ref. 20] illustrates how complex this process is. The deposited co-ordinates, chains A and B are shown in Figure 3c to consist of two chemically equivalent chains related by non- crystallographic symmetry. In this case the contents of the ASU are not a biological unit but are an instance of crystal packing with a borderline DeltaASA of 581 Ang**2 . The PDB annotation indicates that there are two molecules in the asymmetric unit and the Walsh annotation describes this entry as "monomer - 2 chains given". However, the procedure described here generates two independent dimers, with each dimer consisting of one of the ASU chains and a partner related by a crystallographic symmetry operation (DeltaASA 3749 Ang**2 ). One of these dimers is shown in Figure 3d, consisting of chains A and a symmetry related chain A'. The description of the quaternary state for the files described here is complicated for many Protein/Nucleic acid complexes. For entries that contain more than 1 protein chain and more than 1 nucleic acid chain of bio-polymers, the protein chains may form a dimer in the absence of the nucleic acid chains or the entry may consist of several protein molecules complexed to the same nucleic acid chains but lacking protein-protein interaction. An example is the replication terminator protein/dna complex 1ECR[ref. 21]. This is a protein-two- chain nucleic acid complex that was found to be a symmetrical complex with only protein- protein contacts (DeltaASA 799 Ang**2). The deoxyribonuclease 2DNJ[ref. 22] is also a protein-two chain nucleic acid complex that is expanded by symmetry to a complex-dimer. Here the quaternary state has no protein-protein interaction (Figure 3b). The procedure also generates complex protein-protein arrangements that may only be stable in the solid state. An example is the 24meric assembly found for the transcription repressor protein, rop, 1GTO[ref. 23]. This PDB entry has a dimer for the biological unit, and the procedure automatically generates what the authors describe as a "hyperstable helical bundle" existing in the crystal structure (Figure 3e). Related arrangements were found for several protease inhibitors including serine protease inhibitor ci2, 1CIQ[ref. 24], a [A6B6] dodecamer (DeltaASA 140Ang**2) and for a Bowman-Birk proteinase inhibitor 1PI2[ref. 25] a hexamer (DeltaASA 1385 Ang**2). For the latter, evidence has been presented to show that the hexamer does exist in solution[ref. 26]. Finally virus entries are treated differently, for example with icosahedrally symmetric virus particles files are generated containing the complete virion, the 5-fold pentamer (e.g. the 5- fold for bacteriophage MS2 capsid 1MVA[ref. 27] is shown in Figure 3f), the 2,3-fold hexamer and a file containing all chains needed to describe all the unique protein-protein interfaces. For PDB entries of virus samples, the deposited coordinates are normally not the contents of the asymmetric unit, but rather the coordinates for the unique non-crystallographic repeat. For virus particles that have cylindrical polar symmetry (e.g. 2tmv), a 1FOO.mmol file of 49 repeats is produced together with a file, 1FOO.unique, that contains just those symmetry generated repeats that together describe all possible inter-chain contacts. The rod shaped particles are built up using the appropriate repeat distance for the particular entry. For helical virus entries [FILAMENTOUS BACTERIOPHAGE, e.g. 3ifm], the 1FOO.mmol file contains 35 repeats and the 1FOO.unique file contains the set of neighbours, Treatment of water molecules: For solvent atoms only, these atoms first moved to the closest symmetry related position with respect to the protein and if possible are expanded by symmetry to give a more complete solvation sphere. If appropriate the waters are also divided amongst the different oligomer files as second and third shell waters. Hence the total number of water atoms in all the split files may be greater than the number of water atoms deposited in the original file, as an individual water atom may be shared between more than oligomer. This is best illustrated by the following example:-
4INS Example of generation of MacroMolecule and positioning of watersThe following graphical files for PDB entry 4ins, which is the two-zinc insulin storage complex of the hormone insulin, illustrate the steps required to create a coordinate set that describes the actual MacroMolecule studied in the X-Ray diffraction experiment. (Click on any image to get a larger version)
PQS Atlas: This server provides a set of PDB coordinates that have been generated using a standardised method throughout with regard to, (i) orthogonalisation convention, (ii) positioning of all water molecules and (iii) oligomeric state. The status and number of PDB entries processed is given in Table 1. For those generated quaternary structure files that clearly do represent the biological active state, then the coordinates can be used in a more complete description of domain structures that involve more than one chain. Residues that are completely buried at the protein-protein interface will probably be conserved and may also be used in homology model building to improve an alignment in much the same way that structural information, such as cis-prolines and glycines that have specific dihedral angles, are used as focal points within a family of proteins. Multimers that may not represent a true biological molecule still contain information about specific surfaces that may be involved in some biological function. Of course some crystal contacts have no biological significance, such as those examples where the experiment was conducted on an apo-protein but a dimer results from part of a symmetry mate lying close to the active site. Caution should be exercised in these examples in comparing the apo-state to a bound state (e.g. ascorbate oxidase 1ASO[ref. 28] where 2 independent crystal symmetry dimeric like complexes were found). Availability: The Protein Quaternary Structure file server is available on a per PDB entry basis on the direct search URL pqs-quick.html, or via a hyperlink on the PDB 3db browser[ref. 29] atlas page for each entry on all the PDB WWW mirror sites. The PDB entries that are organised in a database are available via the URL ( PQS ). This is a web front end to a limited SQL query system that allows for searches other than by PDB Idcodes to a relational database. Status of number of PDB entries proccessed:The Current 61266 entries consist of: Downloads The procedure to generate the pqs files is not always straight forward and the matrices given in the rem350 files do not necessarily apply to the coordinates in the PDB. All entries are first re-orthogonalised and chains are re-position as close to the origin as a symmetry operation will allow. In some cases the sapce group is changed to a standard setting and non-crystallographic symmetry operations may have to be applied first to generate the asymmetric unit. Therefore in some cases it is not possible to get the pqs files direct from the PDB entry and the files in the rem350 directory. For example the PDB entry 1HZ5 gives a potential dimer that requires only one chain of the deposited coordinates to be moved, Chain A by (1.0 - X, Y - X, 2/3 -Z) and the deposited Chain A is not used. The resulting "dimer" may not be of biological significance, but in the PQS procedure this is the quaternary structure on the particular protein used in this particular structure determination. It is worthwhile to compare the two rasmol representations, PQS for 1HZ5 and the PDB for 1HZ5 (both here setup for rasmol), to get some understanding of what the PQS procedure carries out try doing a spacefill option with each. For 1HZ5 the "dimeric" arrangment in the PQS system completes the coordination for some zinc atoms - while others still require further symmetry related residues e.g. zinc atom A102 has the original His from chain A and now is given in a position bonding to a Glu from chain B, however it still needs further symmetry via the translational operation 1+x, 1+y, 1+z to give a tetrahedral coordination with 2 additional His residues from another Chain B. PDB entry is a fragment of the protein (SWALL Q51912) residues 111 to 173, PROTEIN L PRECURSOR (mature length 19 to 719) or for EMBL M86697. The mature protein binds kappa chain of immunoglobulins and has five such kappa-binding domains. The PQS representation is its most likely quaternary structure and may represent some of the protein surfaces that are used in inter-domain or kappa binding. The zinc ion may not be part of the mature protein but it is certainly part of the chemistry of the protein used in this experiment. Zinc coordination data1HZ5 Zinc Atom Donor Symm Dist Angles The resulting files are not available by ftp but one can download the files as fast as ftp using the following. First get the file lists from:
#!/usr/bin/perl
$HTTP="/pdbe/pqs/macmol";
use LWP::Simple;
if ($#ARGV != 0) {
die << EOT;
Usage: Vget idcode or if extn not mmol - e.g. virus 1ABC.5fold Vetm idcode 5fold or Vetm idcode water
EOT
}
$fext = ".mmol";
$fname = $ARGV[0];
if ( defined($ARGV[1])) {
$fext = ".".$ARGV[1] }
$ofile = $fname.$fext;
@result = split(/\n/, get("$HTTP/$ofile"));
if ( defined(@result)){
open (OUT, ">>$ofile") || die "Unable to open -> $ofile : $!\n";
foreach $mline (@result) {
print OUT "$mline\n"; }
}
This Perl script can be called from a shell script which contains the list of files to be downloaded. For example, create a shell script in the form given below. #!/bin/tcsh -f foreach pqs(1a32 1a8l 1a8o) echo $pqs ./script $pqs # script here is the name of the perl script end The shell script can be saved in the form filename.csh ensuring that the file is executable i.e., chmod +x filename.csh. The shell can then be run as ./filename.csh File Access : The files may be accessed from the pdb www service using the 3db browser. Once your query has identified a PDB ID and you have Retrieved the data, on the PDB Entry page under the heading: Data retrieval:if the above process has created a new file(s) then the link to the additional web page can be got by following the link: Macro Molecules for PDB entry: 1FOOFTP access to the files will be made available when EBI and BNL have coordinated the mirroring of these files. Acknowledgements: Part of the procedure outlined here uses a modification to the WHATIF program specially written for this purpose by Rob Hooft at EMBL Heidelberg. John Westbrook of the Nucleic acid Data Base group at the University of Rutgers (http://ndbserver.rutgers.edu) made available a machine readable list of the quaternary states for all nucleic acid and protein/nucleic acid complexes held in the PDB that was used in the annotation and checking stage described here.You can download a copy here. References: Related documents:
|
|
