Prediction of protein-protein interactions using protein signature profiling

Protein domains are conserved and functionally independent structures that play an important role in interactions among related proteins. Domain-domain interactions have been recently used to predict protein-protein interactions (PPI). In general, the interaction probability of a pair of domains is scored using a trained scoring function. Satisfying a threshold, the protein pairs carrying those domains are regarded as "interacting". In this study, the signature contents of proteins were utilized to predict PPI pairs in Saccharomyces cerevisiae, Caenorhabditis elegans, and Homo sapiens. Similarity between protein signature patterns was scored and PPI predictions were drawn based on the binary similarity scoring function. Results show that the true positive rate of prediction by the proposed approach is approximately 32% higher than that using the maximum likelihood estimation method when compared with a test set, resulting in 22% increase in the area under the receiver operating characteristic (ROC) curve. When proteins containing one or two signatures were removed, the sensitivity of the predicted PPI pairs increased significantly. The predicted PPI pairs are on average 11 times more likely to interact than the random selection at a confidence level of 0.95, and on average 4 times better than those predicted by either phylogenetic profiling or gene expression profiling.     

Prediction of Protein-Protein Interactions Using Protein Signature Profiling

Protein domains are conserved and functionally independent structures that play an important role in interactions among related proteins. Domain-domain inter- actions have been recently used to predict protein-protein interactions (PPI). In general, the interaction probability of a pair of domains is scored using a trained scoring function. Satisfying a threshold, the protein pairs carrying those domains are regarded as "interacting". In this study, the signature contents of proteins were utilized to predict PPI pairs in Saccharomyces cerevisiae, Caenorhabditis ele- gans, and Homo sapiens. Similarity between protein signature patterns was scored and PPI predictions were drawn based on the binary similarity scoring function. Results show that the true positive rate of prediction by the proposed approach is approximately 32% higher than that using the maximum likelihood estimation method when compared with a test set, resulting in 22% increase in the area un- der the receiver operating characteristic (ROC) curve. When proteins containing one or two signatures were removed, the sensitivity of the predicted PPI pairs in- creased significantly. The predicted PPI pairs are on average 11 times more likely to interact than the random selection at a confidence level of 0.95, and on aver- age 4 times better than those predicted by either phylogenetic profiling or gene expression profiling.     

The carboxy-terminal cysteine of the tetraspanin L6 antigen is required for its interaction with SITAC, a novel PDZ protein

PDZ domains are protein modules that mediate protein-protein interactions. Here, we present the identification and characterization of a protein similar to the recently identified PDZ-containing protein TACIP18, which we have named SITAC (similar to TACIP18). SITAC is preferentially expressed in cells of the digestive tract, associated with intracellular membranes. Despite the high degree of sequence identity between the PDZ domains of TACIP18 and those of SITAC, none of the known ligands of the former shows interaction with the latter, as judged by two-hybrid analysis. SITAC interacts with peptides containing bulky hydrophobic amino acids two positions upstream of the C-terminal residue. Surprisingly, SITAC also shows interaction with peptides ending in C, a previously unacknowledged ability of PDZ domains. The sequence -Y-X-C-COOH, bound in vitro by SITAC, is present in the member of the tetraspanin superfamily, the L6 antigen. Coimmunoprecipitation experiments show that SITAC interacts with L6A, but not with an L6A C-terminal mutant, confirming the capacity of SITAC to interact with proteins ending in C. Confocal analysis shows that the interaction between L6A and SITAC is necessary for the precise colocalization of both molecules in the same subcellular compartment. In summary, the characterization of the protein SITAC has unveiled novel sequences recognized by PDZ domains, and it suggests that L6A is a natural ligand of this PDZ protein.     

D-AKAP2 Interacts with Rab4 and Rab11 through Its RGS Domains and Regulates Transferrin Receptor Recycling

Dual-specific A-kinase-anchoring protein 2 (D-AKAP2/AKAP10), which interacts at its carboxyl terminus with protein kinase A and PDZ domain proteins, contains two tandem regulator of G-protein signaling (RGS) domains for which the binding partners have remained unknown. We show here that these RGS domains interact with Rab11 and GTP-bound Rab4, the first demonstration of RGS domains binding small GTPases. Rab4 and Rab11 help regulate membrane trafficking through the endocytic recycling pathways by recruiting effector proteins to specific membrane domains. Although D-AKAP2 is primarily cytosolic in HeLa cells, a fraction of the protein localizes to endosomes and can be recruited there to a greater extent by overexpression of Rab4 or Rab11. D-AKAP2 also regulates the morphology of the Rab11-containing compartment, with co-expression causing accumulation of both proteins on enlarged endosomes. Knockdown of D-AKAP2 by RNA interference caused a redistribution of both Rab11 and the constitutively recycling transferrin receptor to the periphery of cells. Knockdown also caused an increase in the rate of transferrin recycling, suggesting that D-AKAP2 promotes accumulation of recycling proteins in the Rab4/Rab11-positive endocytic recycling compartment.     

Synaptotagmin interaction with the syntaxin/SNAP-25 dimer is mediated by an evolutionarily conserved motif and is sensitive to inositol hexakisphosphate

Synaptotagmins are membrane proteins that possess tandem C2 domains and play an important role in regulated membrane fusion in metazoan organisms. Here we show that both synaptotagmins I and II, the two major neuronal isoforms, can interact with the syntaxin/synaptosomal-associated protein of 25 kDa (SNAP-25) dimer, the immediate precursor of the soluble NSF attachment protein receptor (SNARE) fusion complex. A stretch of basic amino acids highly conserved throughout the animal kingdom is responsible for this calcium-independent interaction. Inositol hexakisphosphate modulates synaptotagmin coupling to the syntaxin/SNAP-25 dimer, which is mirrored by changes in chromaffin cell exocytosis. Our results shed new light on the functional importance of the conserved polybasic synaptotagmin motif, suggesting that synaptotagmin interacts with the t-SNARE dimer to up-regulate the probability of SNARE-mediated membrane fusion.     

Sgt1p is a unique co-chaperone that acts as a client adaptor to link Hsp90 to Skp1p

Sgt1p is a conserved, essential protein required for kinetochore assembly in both yeast and animal cells. Sgt1p has homology to both TPR and p23 domains, sequences often found in proteins that interact with and regulate the molecular chaperone, Hsp90. The presence of these domains and the recent findings that Sgt1p interacts with Hsp90 has led to the speculation that Sgt1p and Hsp90 form a co-chaperone complex. To test this possibility, we have used purified recombinant proteins to characterize the in vitro interactions between yeast Sgt1p and Hsp82p (an Hsp90 homologue in yeast). We show that Sgt1p interacts directly with Hsp82p via its p23 homology region in a nucleotide-dependent manner. However, Sgt1p binding does not alter the enzymatic activity of Hsp82p, suggesting that it is distinct from other co-chaperones. We find that Sgt1p can form a ternary chaperone complex with Hsp82p and Sti1p, a well characterized Hsp90 co-chaperone. Sgt1p interacts with its binding partner Skp1p through its TPR domains and links Skp1p to the core Hsp82p-Sti1p co-chaperone complex. The multidomain nature of Sgt1p and its ability to bridge the interaction between Skp1p and Hsp82p argue that Sgt1p acts as a "client adaptor" recruiting specific clients to Hsp82p co-chaperone complexes.     

Regulation of G protein-mediated signal transduction by RGS proteins

RGS proteins form a new family of regulatory proteins of G protein signaling. They contain homologous core domains (RGS domains) of about 120 amino acids. RGS domains interact with activated Galpha subunits. Several RGS proteins have been shown biochemically to act as GTPase activating proteins (GAPs) for their interacting Galpha subunits. Other than RGS domains, RGS proteins differ significantly in size, amino acid sequences, and tissue distribution. In addition, many RGS proteins have other protein-protein interaction motifs involved in cell signaling. We have shown that p115RhoGEF, a newly identified GEF(guanine nucleotide exchange factor) for RhoGTPase, has a RGS domain at its N-terminal region and this domain acts as a specific GAP for Galpha12 and Galpha13. Furthermore, binding of activated Galpha13 to this RGS domain stimulated GEF activity of p115RhoGEF. Activated Galpha12 inhibited Galpha13-stimulated GEF activity. Thus p115RhoGEF is a direct link between heterotrimeric G protein and RhoGTPase and it functions as an effector for Galpha12 and Galpha13 in addition to acting as their GAP. We also found that RGS domain at N-terminal regions of G protein receptor kinase 2 (GRK2) specifically interacts with Galphaq/11 and inhibits Galphaq-mediated activation of PLC-beta, apparently through sequestration of activated Galphaq. However, unlike other RGS proteins, this RGS domain did not show significant GAP activity to Galphaq. These results indicate that RGS proteins have far more diverse functions than acting simply as GAPs and the characterization of function of each RGS protein is crucial to understand the G protein signaling network in cells.     

PDZ domain protein GIPC interacts with the cytoplasmic tail of melanosomal membrane protein gp75 (tyrosinase-related protein-1).

Tyrosinase and tyrosinase-related proteins (TRPs) are a family of melanosomal membrane proteins involved in mammalian pigmentation. Whereas the melanogenic functions of TRPs are localized in their amino-terminal domains that reside within the lumen of melanosomes, the sorting and targeting of these proteins to melanosomes is mediated by signals in their cytoplasmic domains. To identify proteins that interact with the cytoplasmic tail of gp75 (TRP-1), the most abundant melanosomal membrane protein, we performed yeast two-hybrid screening of a melanocyte cDNA library. Here, we show that the cytoplasmic domain of gp75 interacts with a PDZ domain-containing protein. The gp75-interacting protein is identical to GIPC, an RGS (regulator of G protein signaling)/GAIP-interacting protein, and to SEMCAP-1, a transmembrane semaphorin-binding protein. Carboxyl-terminal amino acid residues, Ser-Val-Val, of gp75 are necessary and sufficient for interaction of gp75 with the single PDZ domain in GIPC. Although endogenous and transfected GIPCs bind efficiently to transiently expressed gp75, only a small amount of GIPC is found associated with gp75 at steady state. Using a strategy to selectively synchronize the biosynthesis of endogenous gp75, we demonstrate that only newly synthesized gp75 associates with GIPC, primarily in the juxtanuclear Golgi region. Our data suggest that GIPC/SEMCAP-1 plays a role in biosynthetic sorting of proteins, specifically gp75, to melanosomes.     

The solution structure of REF2-I reveals interdomain interactions and regions involved in binding mRNA export factors and RNA

The RNA binding and export factor (REF) family of mRNA export adaptors are found in several nuclear protein complexes including the spliceosome, TREX, and exon junction complexes. They bind RNA, interact with the helicase UAP56/DDX39, and are thought to bridge the interaction between the export factor TAP/NXF1 and mRNA. REF2-I consists of three domains, with the RNA recognition motif (RRM) domain positioned in the middle. Here we dissect the interdomain interactions of REF2-I and present the solution structure of a functionally competent double domain (NM; residues 1-155). The N-terminal domain comprises a transient helix (N-helix) linked to the RRM by a flexible arm that includes an Arg-rich region. The N-helix, which is required for REF2-I function in vivo, overlaps the highly conserved REF-N motif and, together with the adjacent Arg-rich region, interacts transiently with the RRM. RNA interacts with REF2-I through arginine-rich regions in its N- and C-terminal domains, but we show that it also interacts weakly with the RRM. The mode of interaction is unusual for an RRM since it involves loops L1 and L5. NMR signal mapping and biochemical analysis with NM indicate that DDX39 and TAP interact with both the N and RRM domains of REF2-I and show that binding of these proteins and RNA will favor an open conformation for the two domains. The proximity of the RNA, TAP, and DDX39 binding sites on REF2-I suggests their binding may be mutually exclusive, which would lead to successive ligand binding events in the course of mRNA export.     

MICALs,a family of conserved flavoprotein oxidoreductases,function in plexin-mediated axonal repulsion

Members of the semaphorin family of secreted and transmembrane proteins utilize plexins as neuronal receptors to signal repulsive axon guidance. It remains unknown how plexin proteins are directly linked to the regulation of cytoskeletal dynamics. Here, we show that Drosophila MICAL, a large, multidomain, cytosolic protein expressed in axons, interacts with the neuronal plexin A (PlexA) receptor and is required for Semaphorin 1a (Sema-1a)-PlexA-mediated repulsive axon guidance. In addition to containing several domains known to interact with cytoskeletal components, MICAL has a flavoprotein monooxygenase domain, the integrity of which is required for Sema-1a-PlexA repulsive axon guidance. Vertebrate orthologs of Drosophila MICAL are neuronally expressed and also interact with vertebrate plexins, and monooxygenase inhibitors abrogate semaphorin-mediated axonal repulsion. These results suggest a novel role for oxidoreductases in repulsive neuronal guidance.     

Recruitment of clathrin onto endosomes by the Tom1-Tollip complex

Tom1 (target of Myb 1) and its related proteins (Tom1L1/Srcasm and Tom1L2) constitute a protein family and share an N-terminal VHS (Vps27p/Hrs/Stam) domain and a following GAT (GGA and Tom1) domain, both of which are also conserved in the GGA family proteins. However, the C-terminal half is not significantly conserved between the Tom1 and GGA families or even between Tom1 and Tom1L1. We have previously shown that the GAT domain of Tom1 interacts with Tollip (Toll-interacting protein), which is associated with endosomes, to which it recruits Tom1. We here extend the previous data and show that the GAT domains of Tom1L1 and Tom1L2 also interact with Tollip, and the C-terminal regions of all the Tom1 family proteins interact with clathrin. Furthermore, when coexpressed with Tollip, all the Tom1 family proteins recruite clathrin onto endosomes. These results indicate that, in conjunction with Tollip, Tom1 family proteins play an important role in recruiting clathrin onto endosomes and suggest that they modulate endosomal functions.     

Calmodulin binds to a tubulin binding site of the microtubule-associated protein tau

Previous studies have demonstrated that the microtubule - associated proteins MAP-2 and tau interact selectively with common binding domains on tubulin defined by the low-homology segments a (430–441) and (422–434). It has been also indicated that the synthetic peptide VRSKIGSTENLKHQPGGG corresponding to the first tau repetitive sequence represents a tubulin binding domain on tau. The present studies show that the calcium-binding protein calmodulin interacts with a tubulin binding site on tau defined by the second repetitive sequence VTSKCGSLGNIHHKPGGG. It was shown that both tubulin and calmodulin bind to tau peptide-Sepharose affinity column. Binding of calmodulin occurs in the presence of 1 mM Ca 2+ and it can be eluted from the column with 4 mM EGTA. These findings provide new insights into the regulation of microtubule assembly, since Ca 2+/calmodulin inhibition of tubulin polymerization into microtubules could be mediated by the direct binding of calmodulin to tau, thus preventing the interaction of this latter protein with tubulin.     

Interactions between the PDZ domains of Bazooka (Par-3) and phosphatidic acid: in vitro characterization and role in epithelial development

Bazooka (Par-3) is a conserved polarity regulator that organizes molecular networks in a wide range of cell types. In epithelia, it functions as a plasma membrane landmark to organize the apical domain. Bazooka is a scaffold protein that interacts with proteins through its three PDZ (postsynaptic density 95, discs large, zonula occludens-1) domains and other regions. In addition, Bazooka has been shown to interact with phosphoinositides. Here we show that the Bazooka PDZ domains interact with the negatively charged phospholipid phosphatidic acid immobilized on solid substrates or in liposomes. The interaction requires multiple PDZ domains, and conserved patches of positively charged amino acid residues appear to mediate the interaction. Increasing or decreasing levels of diacylglycerol kinase or phospholipase D-enzymes that produce phosphatidic acid-reveal a role for phosphatidic acid in Bazooka embryonic epithelial activity but not its localization. Mutating residues implicated in phosphatidic acid binding revealed a possible role in Bazooka localization and function. These data implicate a closer connection between Bazooka and membrane lipids than previously recognized. Bazooka polarity landmarks may be conglomerates of proteins and plasma membrane lipids that modify each other's activities for an integrated effect on cell polarity.     

Sequential binding of cytosolic Phox complex to phagosomes through regulated adaptor proteins: evaluation using the novel monomeric Kusabira-Green System and live imaging of phagocytosis

We engineered a method for detecting intramolecular and intermolecular phox protein interactions in cells by fluorescence microscopy using fusion proteins of complementary fragments of a coral fluorescent reporter protein (monomeric Kusabira-Green). We confirmed the efficacy of the monomeric Kusabira-Green system by showing that the PX and PB1 domains of p40phox interact in intact cells, which we suggested maintains this protein in an inactive closed conformation. Using this system, we also explored intramolecular interactions within p47phox and showed that the PX domain interacts with the autoinhibited tandem Src homology 3 domains maintained in contact with the autoinhibitory region, along with residues 341-360. Furthermore, we demonstrated sequential interactions of p67phox with phagosomes involving adaptor proteins, p47phox and p40phox, during FcgammaR-mediated phagocytosis. Although p67phox is not targeted to phagosomes by itself, p47phox functions as an adaptor for the ternary complex (p47phox-p67phox-p40phox) in early stages of phagocytosis before phagosome closure, while p40phox functions in later stages after phagosomal closure. Interestingly, a mutated "open" form of p40phox linked p47phox to closed phagosomes and prolonged p47phox and p67phox retention on phagosomes. These results indicate that binding of the ternary complex to phagosomes can be temporally regulated by switching between adaptor proteins that have PX domains with distinct lipid-binding specificities.     

Role of the AP2 beta-appendage hub in recruiting partners for clathrin-coated vesicle assembly

Adaptor protein complex 2 alpha and beta-appendage domains act as hubs for the assembly of accessory protein networks involved in clathrin-coated vesicle formation. We identify a large repertoire of beta-appendage interactors by mass spectrometry. These interact with two distinct ligand interaction sites on the beta-appendage (the "top" and "side" sites) that bind motifs distinct from those previously identified on the alpha-appendage. We solved the structure of the beta-appendage with a peptide from the accessory protein Eps15 bound to the side site and with a peptide from the accessory cargo adaptor beta-arrestin bound to the top site. We show that accessory proteins can bind simultaneously to multiple appendages, allowing these to cooperate in enhancing ligand avidities that appear to be irreversible in vitro. We now propose that clathrin, which interacts with the beta-appendage, achieves ligand displacement in vivo by self-polymerisation as the coated pit matures. This changes the interaction environment from liquid-phase, affinity-driven interactions, to interactions driven by solid-phase stability ("matricity"). Accessory proteins that interact solely with the appendages are thereby displaced to areas of the coated pit where clathrin has not yet polymerised. However, proteins such as beta-arrestin (non-visual arrestin) and autosomal recessive hypercholesterolemia protein, which have direct clathrin interactions, will remain in the coated pits with their interacting receptors.     

Specific sequences of the Sm and Sm-like (Lsm) proteins mediate their interaction with the spinal muscular atrophy disease gene product (SMN)

The spinal muscular atrophy disease gene product (SMN) is crucial for small nuclear ribonuclear protein (snRNP) biogenesis in the cytoplasm and plays a role in pre-mRNA splicing in the nucleus. SMN oligomers interact avidly with the snRNP core proteins SmB, -D1, and -D3. We have delineated the specific sequences in the Sm proteins that mediate their interaction with SMN. We show that unique carboxyl-terminal arginine- and glycine-rich domains comprising the last 29 amino acids of SmD1 and the last 32 amino acids of SmD3 are necessary and sufficient for SMN binding. Interestingly, SMN also interacts with at least two of the U6-associated Sm-like (Lsm) proteins, Lsm4 and Lsm6. Furthermore, the carboxyl-terminal arginine- and glycine-rich domain of Lsm4 directly interacts with SMN. This suggests that SMN also functions in the assembly of the U6 snRNP in the nucleus and in the assembly of other Lsm-containing complexes. These findings demonstrate that arginine- and glycine-rich domains are necessary and sufficient for SMN interaction, and they expand further the range of targets of the SMN protein.     

Quality control in the endoplasmic reticulum: PDI mediates the ER retention of unassembled procollagen C-propeptides

Quality control within the endoplasmic reticulum (ER) is thought to be mediated by the interaction of a folding protein with one or several resident ER proteins [1]. Protein disulphide isomerase (PDI) is one such ER resident protein that has been previously shown to interact with proteins during their folding and assembly pathways [2, 3]. It has been assumed that, as a consequence of this interaction, unassembled proteins are retained within the ER. Here, we experimentally show that this is indeed the case. We have taken advantage of our previous finding that PDI interacts with procollagen chains early on in their assembly pathway [2] to address the role of this protein in directly retaining unassembled chains within the ER. Our experimental approach involved expressing individual C-propeptide domains from different procollagen chains in mammalian cells and determining the ability of these domains to interact with PDI and to be secreted. The C-propeptide from the proalpha2(I) chain was retained within the cell, where it formed a complex with PDI. Conversely, the C-propeptide from the proalpha1(III) chain did not form a complex with PDI and was secreted. Both domains were secreted, however, from a stable cell line expressing a secreted form of PDI lacking its ER retrieval signal. Hence, we have demonstrated directly that the intracellular retention of one substrate for ER quality control is due to an interaction with PDI.     

Interaction of DnaK with native proteins and membrane proteins correlates with their accessible hydrophobicity

Molecular chaperones are involved in protein folding, protein targeting to membranes, and protein renaturation after stress. They interact specifically with hydrophobic sequences that are exposed in unfolded proteins, and buried in native proteins. We have studied the interaction of DnaK with native water-soluble proteins and membrane proteins. DnaK–native protein interactions are characterized by dissociation constants between 1 and 50 μM (compared with 0.01–1 μM for unfolded proteins). This affinity is within the range of most intracellular protein concentrations, suggesting that DnaK interacts with a greater number of native proteins than previously suspected. We found a correlation between the affinity of native proteins for DnaK and their affinity for hydrophobic-interaction chromatography adsorbents, suggesting that DnaK interacts with exposed hydrophobic groups in native proteins. The interaction between DnaK and membrane proteins is characterized by DnaK's high affinity for detergent-solubilized membrane proteins, and its lower affinity for membrane proteins inserted in lipid bilayers, suggesting that the chaperone can interact with the hydrophobic sequences of the former, while it cannot penetrate the hydrophobic core of lipid bilayers. Thus, the specificity of DnaK for hydrophobic sequences is involved in its interaction with not only unfolded proteins, but also native water-soluble proteins and membrane proteins. All proteins interact with DnaK according to their exposed hydrophobicity.     

Mammalian CAP interacts with CAP,CAP2, and actin

We previously identified human CAP, a homolog of the yeast adenylyl cyclase—associated protein. Previous studies suggest that the N-terminal and C-terminal domains of CAP have distinct functions. We have explored the interactions of human CAP with various proteins. First, by performing yeast two-hybrid screens, we have identified peptides from several proteins that interact with the C-terminal and/or the N-terminal domains of human CAP. These peptides include regions derived from CAP and BAT3, a protein with unknown function. We have further shown that MBP fusions with these peptides can associate in vitro with the N-terminal or C-terminal domains of CAP fused to GST. Our observations indicate that CAP contains regions in both the N-terminal and C-terminal domains that are capable of interacting with each other or with themselves. Furthermore, we found that myc-epitope-tagged CAP coimmunoprecipitates with HA-epitope-tagged CAP from either yeast or mammalian cell extracts. Similar results demonstrate that human CAP can also interact with human CAP2. We also show that human CAP interacts with actin, both by the yeast two-hybrid test and by coimmunoprecipitation of epitope-tagged CAP from yeast or mammalian cell extracts. This interaction requires the C-terminal domain of CAP, but not the N-terminal domain. Thus CAP appears to be capable of interacting in vivo with other CAP molecules, CAP2, and actin. We also show that actin co-immunoprecipitates with HA-CAP2 from mammalian cell extracts. © 1996 Wiley-Liss, Inc.