 |
An example of our proposed system of complex architecture and modulatiry, derived using our automated
procedure to define complexes automatically from TAP/MS data.
The figure shows two distinct families of cap-binding proteins: the nuclear CBC (cap-binding complex) and the cytoplasmic eIF4F.
Proteins are coloured according to their localization. Lines corresponds to our socio-affinity indices.
TAP bait proteins are shown in bold and shaded circles around groups of proteins
indicate cores and modules. See Gavin, Aloy et al., Nature, 2006 (PubMed). |
 |
Phenotypic data mapped on to six complexes.
Shaded cells indicate a growth defect (slow growth or no growth relative to the control);
those boxed in red represent the phenotypic signature of the complex. See Gavin, Aloy et al., Nature, 2006 (PubMed). |
 |
Schematic outlining our appraoch for finding protein linear motifs that mediate protein-protein interactions. Sets of proteins (A-F) sharing an interaction partner
(X) are grouped and domains and homologous sequences are removed. We then search for 3-8 residue
motifs that are over-represented in the remaining sequence, and score these by a binomial probability
to give a ranked set of candiate motifs mediated the interaction with protein X.
See Neduva et al., PLoS Biology, 2005 (PubMed). |
 |
Example of a VxxxRxYS motif predicted to bind the DNA/RNA binding
protein translin in Drosophila. Proteins are shown as domain bubblegrams, and the motifs
are labelled with blue bars, expanded to show the sequence and the conservation in Drosophila psuedoobscura (if an orthologue is found).
See Neduva et al., PLoS Biology, 2005 (PubMed). |
 |
Scatchard (binding) curve showing that the VxxxRxYS peptide
binds specifically to translin. Two controls (a mutated peptide, and an arbitray one) are
also shown.
See Neduva et al., PLoS Biology, 2005 |
 |
Illustration of an Interaction Type, which we consider to be
analogous to a protein fold for a pair of interacting proteins. A set of eight interactions
with Cyclin Dependent Protein Kinase (CDK) are grouped into four types based on sufficient similarity
between the interacting proteins.
See Aloy & Russell, Nature Biotech, 2004 (PubMed). |
 |
Plot showing the increase in the total number of interactions of
known three-dimensional structure (light grey bars) as a function of time. The number of interaction
types is shown in medium grey, and the depositions of new types are shown in black bars (and in
the inset). Note that the number of interactions of known structure was comparatively small until
the advent of synchrotron radiation, and cryo-freezing techniques in the mid-late nineties.
See Aloy & Russell, Nature Biotech, 2004 (PubMed). |
 |
Phosphate-binding site shared between two proteins of different fold (left: survival protein E (SurE)
homolog (1l5x); right: phosphate-binding periplasmic protein (1a40).
See Stark et al., Structure, 2, 1405-1412, 2004 (PubMed). |
 |
Novel Phospho-Tyrosine-binding site in the archael protein Mth1187 (1lxn, white) found by active-site comparison
with the phosphotyrosine-binding site of an SH2-domain (1fyr, black).
See Stark et al., Structure, 2, 1405-1412, 2004 (PubMed). |
 |
Pie-chart showing the complementarity of active-site comparison (PINTS-Server) and structural alignment (Dali-Server).
See Stark et al., Structure, 2, 1405-1412, 2004 (PubMed). |
 |
Illustration of our procedure for building structures for protein complexes
and interaction networks. Dashed circles indicate complexes (identified by affinity purifications or
known in the literature), orange lines indicate interactions known from experiments such as the
two-hybrid system, and green lines denote those that can be modelled by similarity to a complex
of known structure. The bottom right of the figure illustrates how two interactions of known
structure can be combined through a homologous shared component to build a three-component complex.
See Aloy et al., Science, 303, 2026-2029, 2004 (PubMed). |
 |
Network of interacting proteins defined according to whether
interactions can be modelled based on a known three-dimensional structure. The blow-up shows
structural details for interactions between several transcription factor complexes. See Aloy et al., Science, 303, 2026-2029, 2004 (PubMed). |
 |
MicroRNAs let-7, lin-4, and bantam showing the pattern of basepairing in known targets. Yellow indicates a conventional basepair; orange a G:U basepair; blue a mismatch. Black bars show loop positions in the target.
See Stark et al., PLoS Biology, 1, 397-409, 2003 (PubMed). |
 |
Pattern of base-pairing for predicted targets for Drosophila microRNA miR-7 showing the pattern of basepairing with target sites in E(spl) and Brd complex genes sorted in order of predicted folding energy (colours as above).
See Stark et al., PLoS Biology, 1, 397-409, 2003 (PubMed). |
 |
Valine, Leucine, Isoleucine catabolic pathway (KEGG with
enzymes whose genes are predicted to targets for Drosophila MicroRNA miR-277 boxed and identified by CG number. This microRNA appears to be a genetic
switch for this pathway.
See Stark et al., PLoS Biology, 1, 397-409, 2003 (PubMed). |
 |
Example of interactions between ferredoxin-like folds that are similar, despite the fact that the proteins lie in different superfamilies (and thus not necessarily
thought to share a common ancestor (according to SCOP). See Aloy et al., J. Mol. Biol., 332, 989-998, 2003. (PubMed), or read more here. |
 |
Plot of interaction RMSD (iRMSD) versus percent sequence identity for comparing pairs of interactions involving the same folds. Points are coloured according to
the most distant similarity in the pair: blue = fold, green = superfamily, red = family, as defined by SCOP. See Aloy et al., J. Mol. Biol., 332, 989-998, 2003. (PubMed), or query the data yourself here. |
 |
Example of searches that one can perform with PINTS. Here
active site residues from LuxS (left) are compared to a database of other proteins to find the similarities shown on the right. See Stark & Russell, Nucl. Acids Res., 31, 3341-3344, 2003 (PubMed), or see the PINTS server. |
 |
Equation providing the statistical significance of the RMSD between local structural patterns (i.e. a cluster of amino acids in space). See Stark et al J. Mol. Biol., 326, 1307-1316, 2003 (PubMed), or try this out with PINTS server. |
 |
Three-dimensional structures of several protein complexes solved by X-ray crystallography with homologues in yeast. All illustrate some feature of interaction discovery methods. (a & b) Examples of indirect interactions detected by a two-hybrid screen: (a) CKS-cyclin, (b) proteins from small and large subunits of the ribosome. (c & d) Two interactions not seen, despite testing, in two-hybrid screens: (c) Yeast cytochrome bc1 complex, (d) Human electron transfer flavoprotein alpha and beta subunits. (e) Example of an interaction of known structure (actin/profilin) predited not to succeed in a two-hybrid owing to interference with the interface (the C-terminus of Actin is near the interface). See Aloy & Russell Trends Biochem. Sci., 27, 633-638, 2002 (PubMed). |
 |
Possible explanation for proteins seen in many affinity purifications. Hypothetical scenario. In (a) one of the complex components is misfolded because of interference from the tag and is bound to a protease and a chaperonin. In (b) a component is still attached to the ribosome and also bound to a chaperonin. Abbreviation: HSP, heat shock protein. See Aloy & Russell Trends Biochem. Sci., 27, 633-638, 2002 (PubMed). |
 |
Equilibrium of four different states of the human estrogen receptor alpha (ER). The N-terminal A-domain (a), the C-terminal helix 12 (b), corepressors (CoR, c) and coactivators (CoA, d) compete for the same interaction surface on the ligand binding domain. Interaction of the A-domain with this site represses transactivation and transrepression properties of the ER and is modulated by ligands. See Metivier et al., Molecular Cell, 10, 1019-1032, 2002 (PubMed). |
 |
Binding site for the N-terminal A-domain on the ligand binding domain of human estrogen receptor alpha (ER). This interaction represses transactivation and -repression and is modulated by ligands and competed by the C-terminal helix 12. Hydrophobic residues that were mutated to map the interaction site are highlighted. See Metivier et al., Molecular Cell, 10, 1019-1032, 2002 (PubMed) |
 |
Figure showing a potential source of artefacts in protein
interaction networks derived from two-hybrids. The points on the top right of the graph are
two-hybrid "bait" proteins that appear to have hundreds of interacting partners probably
reflect a known source of error in two-hybrids: that many bait proteins are able to
activate transcription without necessarily any interaction with prey. See
Aloy & Russell, FEBS Lett, 530 , 253-254, 2002 (PubMed) |
 |
Domain architecture of the Exosome showing equivalences with the bacterial
degradosome component PNPase.
Adapted from Aloy et al EMBO reports, 3, 628-635, 2002 (PubMed) |
 |
Predicted subunit arrangement of the Exosome core based on PNPase (above) using the method
discussed below (Aloy & Russell, 2002).
Adapted from Aloy et al EMBO reports, 3, 628-635, 2002 (PubMed) |
 |
Three-dimensional model of the exosome core based on PNPase. Circles denote predicted functional
sites, and labelled residues are polar and conserved across different species.
Adapted from Aloy et al EMBO reports, 3, 628-635, 2002 (PubMed) |
 |
Overlay of PNPase onto the electron microscopy image of the exosome, with the predicted
location of subunits labelled. For more details see Aloy et al EMBO reports, 3, 628-635, 2002 (PubMed) This figure adapted from Editors Choice in Science, 297, 899, 2002 (9th August 2002). |
 |
Example of how sequences vary at a protein-protein interaction interface of known three-dimensional structure. The two proteins are shown as ribbon diagrams above or below alignments showing the region around the interacting residues. Residues making side-chain to side-chain contacts are linked by lines, and those contacting the protein main-chain are boxed. Adapted from Aloy & Russell, Proc. Natl. Acad. Sci. USA, 99, 5896-5901, 2002 (PubMed). |
 |
Intersection between yeast protein-protein interactions from the MIPS database and complexes of known three-dimensional structure. Lines indicate how our method to predict protein-protein interactions (InterPreTS) performs (solid lines are statistically significant predictions, dashed lines are insigificant). Adapted from Aloy & Russell, Proc. Natl. Acad. Sci. USA, 99, 5896-5901, 2002 (PubMed). |
 |
Representation of side-chain to side-chain empirical pair potentials derived from protein-interfaces of known three-dimensional structure. Interaction preferences are shown by a scale that goes from blue (unlikely) to red (highly likely). These are for use in the program InterPreTS (Interaction Prediction Through Structure). This method is described in Aloy & Russell, Proc. Natl. Acad. Sci. USA, 99, 5896-5901, 2002 (PubMed). |
 |
Statistically significant links between sequences from various SMART domains that all share a common fold (3-helical DNA/RNA binding bundles). We measured sequence similarity after protein three-dimensional structure alignment. Adapted from Aloy et al, Protein Sci. 11, 1101-1116, 2002 (PubMed). |
 |
An example of a new potential superfamily uncovered after detecting a sequence bridge between two proteins sharing a common fold. Equivalent secondary structures (arrows/ribbons) are the same colour, and the ball-and-stick representation shows amino acids common to the two proteins. Adapted from Aloy et al, Protein Sci. 11, 1101-1116, 2002 (PubMed). |
 |
Topology diagram of HPr Kinase. Triangles are beta-strands, circles are alpha-helices. Secondary structures coloured red are thought to form part of an ancient core common to all P-loop containing proteins. Adapted from a paper discussing the similarity between this kinase and phosphoenolpyruvate carboxykinase. (Russell et al. FEBS Lett. 517, 1-6, 2002; PubMed) |
 |
The structure of a four helical cytokine coloured according to the location of exon boundaries. This is adapted from a paper about exon/intron structure conservation despite no sequence similarity (i.e. proteins that have the same structure). For more details see Betts et al EMBO J. 20, 5354-5360, 2001 (PubMed) |
 |
Hypothetical scenario of how present day proteins could have evolved by successive addition of antecedent domain segements (ADSs). Adapted from Lupas et al J. Struct. Biol., 134,191-203, 2001. (PubMed). |
 |
The cytochrome C haem attachment site. An example of a possible ADS: a short, linear, functional motif found in proteins with different folds. Adapted from Lupas et al J. Struct. Biol., 134,191-203, 2001. (PubMed). |
 |
Two proteins with different folds that contain the above motif (Lupas et al J. Struct. Biol., 134,191-203, 2001 PubMed). |
|
|