New benchmark metrics for protein-protein docking methods

With the development of many computational methods that predict the structural models of protein-protein complexes, there is a pressing need to benchmark their performance. As was the case for protein monomers, assessing the quality of models of protein complexes is not straightforward. An effective scoring scheme should be able to detect substructure similarity and estimate its statistical significance. Here, we focus on characterizing the similarity of the interfaces of the complex and introduce two scoring functions. The first, the interfacial Template Modeling score (iTM-score), measures the geometric distance between the interfaces, while the second, the Interface Similarity score (IS-score), evaluates their residue-residue contact similarity in addition to their geometric similarity. We first demonstrate that the IS-score is more suitable for assessing docking models than the iTM-score. The IS-score is then validated in a large-scale benchmark test on 1562 dimeric complexes. Finally, the scoring function is applied to evaluate docking models submitted to the Critical Assessment of Prediction of Interactions (CAPRI) experiments. While the results according to the new scoring scheme are generally consistent with the original CAPRI assessment, the IS-score identifies models whose significance was previously underestimated.     

Rebuilding a macromolecular membrane complex at the atomic scale: Case of the Kir6.2 potassium channel coupled to the muscarinic acetylcholine receptor M2

Ion channel‐coupled receptors (ICCR) are artificial proteins built from a G protein‐coupled receptor and an ion channel. Their use as molecular biosensors is promising in diagnosis and high‐throughput drug screening. The concept of ICCR was initially validated with the combination of the muscarinic receptor M2 with the inwardly rectifying potassium channel Kir6.2. A long protein engineering phase has led to the biochemical characterization of the M2‐Kir6.2 construct. However, its molecular mechanism remains to be elucidated. In particular, it is important to determine how the activation of M2 by its agonist acetylcholine triggers the modulation of the Kir6.2 channel via the M2‐Kir6.2 linkage. In the present study, we have developed and validated a computational approach to rebuild models of the M2‐Kir6.2 chimera from the molecular structure of M2 and Kir6.2. The protocol was first validated on the known protein complexes of the μ‐opioid Receptor, the CXCR4 receptor and the Kv1.2 potassium channel. When applied to M2‐Kir6.2, our protocol produced two possible models corresponding to two different orientations of M2. Both models highlights the role of the M2 helices I and VIII in the interaction with Kir6.2, as well as the role of the Kir6.2 N‐terminus in the channel opening. Those two hypotheses will be explored in a future experimental study of the M2‐Kir6.2 construct. Proteins 2014; 82:1694–1707. © 2014 Wiley Periodicals, Inc.     

Solution structures of chicken parvalbumin 3 in the Ca(2+)-free and Ca(2+)-bound states

Birds express two β-parvalbumin isoforms, parvalbumin 3 and avian thymic hormone (ATH). Parvalbumin 3 from chicken (CPV3) is identical to rat β-parvalbumin (β-PV) at 75 of 108 residues. CPV3 displays intermediate Ca(2+) affinity--higher than that of rat β-parvalbumin, but lower than that of ATH. As in rat β-PV, the attenuation of affinity is associated primarily with the CD site (residues 41-70), rather than the EF site (residues 80-108). Structural data for rat α- and β-parvalbumins suggest that divalent ion affinity is correlated with the similarity of the unliganded and Ca(2+)-bound conformations. We herein present a comparison of the solution structures of Ca(2+)-free and Ca(2+)-bound CPV3. Although the structures are generally similar, the conformations of residues 47 to 50 differ markedly in the two protein forms. These residues are located in the C helix, proximal to the CD binding loop. In response to Ca(2+) removal, F47 experiences much greater solvent accessibility. The side-chain of R48 assumes a position between the C and D helices, adjacent to R69. Significantly, I49 adopts an interior position in the unliganded protein that allows association with the side-chain of L50. Concomitantly, the realignment of F66 and F70 facilitates their interaction with I49 and reduces their contact with residues in the N-terminal AB domain. This reorganization of the hydrophobic core, although less profound, is nevertheless reminiscent of that observed in rat β-PV. The results lend further support to the idea that Ca(2+) affinity correlates with the structural similarity of the apo- and bound parvalbumin conformations.     

Locally defined protein phylogenetic profiles reveal previously missed protein interactions and functional relationships

Phylogenetic profiles encode patterns of presence or absence of genes across genomes, and these profiles can be used to assign functional relationships to nonhomologous pairs of proteins (Pellegrini et al., Proc Natl Acad Sci USA 1999;96:4284-4288). Although it is well known that many proteins were created from combinations of domains, most of the existing implementations of phylogenetic profiles do not consider this fact. Here, we introduce an extension that considers the multidomain nature of proteins and test the method against the known interaction data sets. Whereas earlier implementations associated one entire sequence with one protein phylogenetic profile (Single-Profile), our method instead breaks the sequence into a set of segments of predetermined size and constructs a separate profile for each segment (Multiple-Profile). The results show that the Multiple-Profile method performs as well as the Single-Profile method. However, the two methods share, surprisingly, a small fraction of their predictions, indicating that the Multiple-Profile method can detect known interactions missed by the Single-Profile method. Thus, the Multiple-Profile method can be used with other methods to determine functional relationships on a genome scale with wider coverage.     

Structure‐guided design of a reversible fluorogenic reporter of protein‐protein interactions

A reversible green fluorogenic protein‐fragment complementation assay was developed based on the crystal structure of UnaG, a recently discovered fluorescent protein. In living mammalian cells, the nonfluorescent fragments complemented and rapidly became fluorescent upon rapamycin‐induced FKBP and Frb protein interaction, and lost fluorescence when the protein interaction was inhibited. This reversible fluorogenic reporter, named uPPI [UnaG‐based protein‐protein interaction (PPI) reporter], uses bilirubin (BR) as the chromophore and requires no exogenous cofactor. BR is an endogenous molecule in mammalian cells and is not fluorescent by itself. uPPI may have many potential applications in visualizing spatiotemporal dynamics of PPIs.     

Allosteric HIV‐1 integrase inhibitors promote aberrant protein multimerization by directly mediating inter‐subunit interactions: Structural and thermodynamic modeling studies

Allosteric HIV‐1 integrase (IN) inhibitors (ALLINIs) bind at the dimer interface of the IN catalytic core domain (CCD), and potently inhibit HIV‐1 by promoting aberrant, higher‐order IN multimerization. Little is known about the structural organization of the inhibitor‐induced IN multimers and important questions regarding how ALLINIs promote aberrant IN multimerization remain to be answered. On the basis of physical chemistry principles and from our analysis of experimental information, we propose that inhibitor‐induced multimerization is mediated by ALLINIs directly promoting inter‐subunit interactions between the CCD dimer and a C‐terminal domain (CTD) of another IN dimer. Guided by this hypothesis, we have built atomic models of inter‐subunit interfaces in IN multimers by incorporating information from hydrogen‐deuterium exchange (HDX) measurements to drive protein‐protein docking. We have also developed a novel free energy simulation method to estimate the effects of ALLINI binding on the association of the CCD and CTD. Using this structural and thermodynamic modeling approach, we show that multimer inter‐subunit interface models can account for several experimental observations about ALLINI‐induced multimerization, including large differences in the potencies of various ALLINIs, the mechanisms of resistance mutations, and the crucial role of solvent exposed R‐groups in the high potency of certain ALLINIs. Our study predicts that CTD residues Tyr226, Trp235 and Lys266 are involved in the aberrant multimer interfaces. The key finding of the study is that it suggests the possibility of ALLINIs facilitating inter‐subunit interactions between an external CTD and the CCD‐CCD dimer interface.     

Coevolutionary constraints in the sequence‐space of macromolecular complexes reflect their self‐assembly pathways

Is the order in which biomolecular subunits self‐assemble into functional macromolecular complexes imprinted in their sequence‐space? Here, we demonstrate that the temporal order of macromolecular complex self‐assembly can be efficiently captured using the landscape of residue‐level coevolutionary constraints. This predictive power of coevolutionary constraints is irrespective of the structural, functional, and phylogenetic classification of the complex and of the stoichiometry and quaternary arrangement of the constituent monomers. Combining this result with a number of structural attributes estimated from the crystal structure data, we find indications that stronger coevolutionary constraints at interfaces formed early in the assembly hierarchy probably promotes coordinated fixation of mutations that leads to high‐affinity binding with higher surface area, increased surface complementarity and elevated number of molecular contacts, compared to those that form late in the assembly. Proteins 2017; 85:1183–1189. © 2017 Wiley Periodicals, Inc.     

Solution structures of polcalcin Phl p 7 in three ligation states: Apo‐, hemi‐Mg2+‐bound,and fully Ca2+‐bound

Polcalcins are small EF‐hand proteins believed to assist in regulating pollen‐tube growth. Phl p 7, from timothy grass (Phleum pratense), crystallizes as a domain‐swapped dimer at low pH. This study describes the solution structures of the recombinant protein in buffered saline at pH 6.0, containing either 5.0 mM EDTA, 5.0 mM Mg²⁺, or 100 μM Ca²⁺. Phl p 7 is monomeric in all three ligation states. In the apo‐form, both EF‐hand motifs reside in the closed conformation, with roughly antiparallel N‐ and C‐terminal helical segments. In 5.0 mM Mg²⁺, the divalent ion is bound by EF‐hand 2, perturbing interhelical angles and imposing more regular helical structure. The structure of Ca²⁺‐bound Phl p 7 resembles that previously reported for Bet v 4—likewise exposing apolar surface to the solvent. Occluded in the apo‐ and Mg²⁺‐bound forms, this surface presumably provides the docking site for Phl p 7 targets. Unlike Bet v 4, EF‐hand 2 in Phl p 7 includes five potential anionic ligands, due to replacement of the consensus serine residue at –x (residue 55 in Phl p 7) with aspartate. In the Phl p 7 crystal structure, D55 functions as a helix cap for helix D. In solution, however, D55 apparently serves as a ligand to the bound Ca²⁺. When Mg²⁺ resides in site 2, the D55 carboxylate withdraws to a distance consistent with a role as an outer‐sphere ligand. ¹⁵N relaxation data, collected at 600 MHz, indicate that backbone mobility is limited in all three ligation states. Proteins 2013. © 2012 Wiley Periodicals, Inc.     

BioGPS: Navigating biological space to predict polypharmacology,off‐targeting,and selectivity

The structural comparison of protein binding sites is increasingly important in drug design; identifying structurally similar sites can be useful for techniques such as drug repurposing, and also in a polypharmacological approach to deliberately affect multiple targets in a disease pathway, or to explain unwanted off‐target effects. Once similar sites are identified, identifying local differences can aid in the design of selectivity. Such an approach moves away from the classical “one target one drug” approach and toward a wider systems biology paradigm. Here, we report a semiautomated approach, called BioGPS, that is based on the software FLAP which combines GRID Molecular Interactions Fields (MIFs) and pharmacophoric fingerprints. BioGPS comprises the automatic preparation of protein structure data, identification of binding sites, and subsequent comparison by aligning the sites and directly comparing the MIFs. Chemometric approaches are included to reduce the complexity of the resulting data on large datasets, enabling focus on the most relevant information. Individual site similarities can be analyzed in terms of their Pharmacophoric Interaction Field (PIF) similarity, and importantly the differences in their PIFs can be extracted. Here we describe the BioGPS approach, and demonstrate its applicability to rationalize off‐target effects (ERα and SERCA), to classify protein families and explain polypharmacology (ABL1 kinase and NQO2), and to rationalize selectivity between subfamilies (MAP kinases p38α/ERK2 and PPARδ/PPARγ). The examples shown demonstrate a significant validation of the method and illustrate the effectiveness of the approach. Proteins 2015; 83:517–532. © 2015 Wiley Periodicals, Inc.     

Soluble mimics of a chemokine receptor: chemokine binding by receptor elements juxtaposed on a soluble scaffold

Despite the broad biological importance of G protein-coupled receptors (GPCRs), ligand recognition by GPCRs remains poorly understood. To explore the roles of GPCR extracellular elements in ligand binding and to provide a tractable system for structural analyses of GPCR/ligand interactions, we have developed a soluble protein that mimics ligand recognition by a GPCR. This receptor analog, dubbed CROSS5, consists of the N-terminal and third extracellular loop regions of CC chemokine receptor 3 (CCR3) displayed on the surface of a small soluble protein, the B1 domain of Streptococcal protein G. CROSS5 binds to the CCR3 ligand eotaxin with a dissociation equilibrium constant of 2.9 +/- 0.8 microM and competes with CCR3 for eotaxin binding. Control proteins indicate that juxtaposition of both CCR3 elements is required for optimal binding to eotaxin. Moreover, the affinities of CROSS5 for a series of eotaxin mutants are highly correlated with the apparent affinities of CCR3 for the same mutants, demonstrating that CROSS5 uses many of the same interactions as does the native receptor. The strategy used to develop CROSS5 could be applied to many other GPCRs, with a variety of potential applications.     

Role of the 14-3-3 C-terminal loop in ligand interaction

14-3-3 proteins are a family of conserved dimeric molecules that interact with a broad range of target proteins, most of which contain phosphoserine/threonine. The amphipathic groove of 14-3-3 is the main structural feature involved in mediating its associations. We have studied another domain of 14-3-3, the C-terminal loop, to determine what role it plays in ligand interaction. A truncated form of 14-3-3zeta lacking this C-terminal loop was generated and found to bind with higher affinity than the wild-type 14-3-3zeta protein to the ligands Raf-1 and Bad. Interestingly, the truncated 14-3-3zeta also showed increased association with the 14-3-3 binding-deficient Bad/S136A mutant. Taken together, these data support a role for the C-terminal loop as a general inhibitor of 14-3-3/ligand interactions. This may provide a mechanism by which inappropriate associations with 14-3-3 are prevented.     

Enhanced pro‐protein convertase subtilisin/kexin type 9 expression by C‐reactive protein through p38MAPK‐HNF1α pathway in HepG2 cells

Plasma C‐reactive protein (CRP) concentration is associated positively with cardiovascular risk, including dyslipidemia. We suggested a regulating role of CRP on pro‐protein convertase subtilisin/kexin type 9 (PCSK9), a key regulator of low‐density lipoprotein (LDL) metabolism, and demonstrated the PCSK9 as a pathway linking CRP and LDL regulation. Firstly, experiments were carried out in the presence of human CRP on the protein and mRNA expression of PCSK9 and LDL receptor (LDLR) in human hepatoma cell line HepG2 cells. Treatment with CRP (10 μg/ml) enhanced significantly the mRNA and protein expression of PCSK9 and suppressed the expression of LDLR. Of note, a late return of LDLR mRNA levels occurred at 12 hrs, while the LDLR protein continued to decrease at 24 hrs, suggesting that the late decrease in LDLR protein levels was unlikely to be accounted for the decrease in LDL mRNA. Secondly, the role of PCSK9 in CRP‐induced LDLR decrease and the underlying pathways were investigated. As a result, the inhibition of PCSK9 expression by small interfering RNA (siRNA) returned partly the level of LDLR protein and LDL uptake during CRP treatment; CRP‐induced PCSK9 increase was inhibited by the p38MAPK inhibitor, SB203580, resulting in a significant rescue of LDLR protein expression and LDL uptake; the pathway was involved in hepatocyte nuclear factor 1α (HNF1α) but not sterol responsive element‐binding proteins (SREBPs) preceded by the phosphorylation of p38MAPK. These findings indicated that CRP increased PCSK9 expression by activating p38MAPK‐HNF1α pathway, with a certain downstream impairment in LDL metabolism in HepG2 cells.     

Host-guest scale of left-handed polyproline II helix formation

Despite the clear importance of the left-handed polyproline II (PPII) helical conformation in many physiologically important processes as well as its potential significance in protein unfolded states, little is known about the physical determinants of this conformation. We present here a scale of relative PPII helix-forming propensities measured for all residues, except tyrosine and tryptophan, in a proline-based host peptide system. Proline has the highest measured propensity in this system, a result of strong steric interactions that occur between adjacent prolyl rings. The other measured propensities are consistent with backbone solvation being an important component in PPII helix formation. Side chain to backbone hydrogen bonding may also play a role in stabilizing this conformation. The PPII helix-forming propensity scale will prove useful in future studies of the conformational properties of proline-rich sequences as well as provide insights into the prevalence of PPII helices in protein unfolded states.     

A generalized knowledge‐based discriminatory function for biomolecular interactions

Several novel and established knowledge‐based discriminatory function formulations and reference state derivations have been evaluated to identify parameter sets capable of distinguishing native and near‐native biomolecular interactions from incorrect ones. We developed the r·m·r function, a novel atomic level radial distribution function with mean reference state that averages over all pairwise atom types from a reduced atom type composition, using experimentally determined intermolecular complexes in the Cambridge Structural Database (CSD) and the Protein Data Bank (PDB) as the information sources. We demonstrate that r·m·r had the best discriminatory accuracy and power for protein‐small molecule and protein‐DNA interactions, regardless of whether the native complex was included or excluded, from the test set. The superior performance of the r·m·r discriminatory function compared with seventeen alternative functions evaluated on publicly available test sets for protein‐small molecule and protein‐DNA interactions indicated that the function was not over optimized through back testing on a single class of biomolecular interactions. The initial success of the reduced composition and superior performance with the CSD as the distribution set over the PDB implies that further improvements and generality of the function are possible by deriving probabilities from subsets of the CSD, using structures that consist of only the atom types to be considered for given biomolecular interactions. The method is available as a web server module at http://protinfo.compbio.washington.edu . Proteins 2009. © 2008 Wiley‐Liss, Inc.     

Staphylococcal superantigen-like protein 10 binds to phosphatidylserine and apoptotic cells

Staphylococcal superantigen-like proteins (SSLs) are a family of exoproteins that have structural similarities to staphylococcal superantigens. Although SSLs do not have superantigenic activity, some of them have been reported to bind to host immune related molecules and they have been implicated in immune evasion by S. aureus. In this study, we showed that SSL10 is capable of binding to phospholipids. SSL10 bound to phosphatidylserine (PS) containing liposome, but not to phosphatidylcholine liposome. SSL10, but not SSL7, bound to PS containing liposome, suggesting that SSL10 specifically binds to PS. Analysis of PS binding ability among recombinant truncated SSL10 fragments revealed that the β-barrel in the N-terminal oligonucleotide/oligosaccharide-binding (OB)-fold domain contributes to PS binding capacity. Fluorescein isothiocyanate labeled OB-fold of SSL10 stained hydrogen peroxide treated Jurkat cells. Annexin V is widely utilized for detection of apoptosis. Unlike annexin V, the OB-fold domain of SSL10 also bound to apoptotic cells in the presence of EDTA, suggesting that the OB-fold of SSL10 recognizes PS and apoptotic cells in a Ca(2+) independent manner. These findings suggest SSL10 and its derived peptides may be a novel detection tool for apoptotic cells.     

Score_set: A CAPRI benchmark for scoring protein complexes

Critical Assessment of PRedicted Interactions (CAPRI) has proven to be a catalyst for the development of docking algorithms. An essential step in docking is the scoring of predicted binding modes in order to identify stable complexes. In 2005, CAPRI introduced the scoring experiment, where upon completion of a prediction round, a larger set of models predicted by different groups and comprising both correct and incorrect binding modes, is made available to all participants for testing new scoring functions independently from docking calculations. Here we present an expanded benchmark data set for testing scoring functions, which comprises the consolidated ensemble of predicted complexes made available in the CAPRI scoring experiment since its inception. This consolidated scoring benchmark contains predicted complexes for 15 published CAPRI targets. These targets were subjected to 23 CAPRI assessments, due to existence of multiple binding modes for some targets. The benchmark contains more than 19,000 protein complexes. About 10% of the complexes represent docking predictions of acceptable quality or better, the remainder represent incorrect solutions (decoys). The benchmark set contains models predicted by 47 different predictor groups including web servers, which use different docking and scoring procedures, and is arguably as diverse as one may expect, representing the state of the art in protein docking. The data set is publicly available at the following URL: http://cb.iri.univ‐lille1.fr/Users/lensink/Score_set . Proteins 2014; 82:3163–3169. © 2014 Wiley Periodicals, Inc.     

Intrinsic disorder in protein sense‐antisense recognition

In addition to the well‐established sense‐antisense complementarity abundantly present in the nucleic acid world and serving as a basic principle of the specific double‐helical structure of DNA, production of mRNA, and genetic code‐based biosynthesis of proteins, sense‐antisense complementarity is also present in proteins, where sense and antisense peptides were shown to interact with each other with increased probability. In nucleic acids, sense‐antisense complementarity is achieved via the Watson‐Crick complementarity of the base pairs or nucleotide pairing. In proteins, the complementarity between sense and antisense peptides depends on a specific hydropathic pattern, where codons for hydrophilic and hydrophobic amino acids in a sense peptide are complemented by the codons for hydrophobic and hydrophilic amino acids in its antisense counterpart. We are showing here that in addition to this pattern of the complementary hydrophobicity, sense and antisense peptides are characterized by the complementary order‐disorder patterns and show complementarity in sequence distribution of their disorder‐based interaction sites. We also discuss how this order‐disorder complementarity can be related to protein evolution.     

Representing and comparing protein folds and fold families using three‐dimensional shape‐density representations

The question of how best to compare and classify the (three‐dimensional) structures of proteins is one of the most important unsolved problems in computational biology. To help tackle this problem, we have developed a novel shape‐density superposition algorithm called 3D‐Blast which represents and superposes the shapes of protein backbone folds using the spherical polar Fourier correlation technique originally developed by us for protein docking. The utility of this approach is compared with several well‐known protein structure alignment algorithms using receiver‐operator‐characteristic plots of queries against the “gold standard” CATH database. Despite being completely independent of protein sequences and using no information about the internal geometry of proteins, our results from searching the CATH database show that 3D‐Blast is highly competitive compared to current state‐of‐the‐art protein structure alignment algorithms. A novel and potentially very useful feature of our approach is that it allows an average or “consensus” fold to be calculated easily for a given group of protein structures. We find that using consensus shapes to represent entire fold families also gives very good database query performance. We propose that using the notion of consensus fold shapes could provide a powerful new way to index existing protein structure databases, and that it offers an objective way to cluster and classify all of the currently known folds in the protein universe. Proteins 2012. © 2011 Wiley Periodicals, Inc.