Impact of sequence on the molecular assembly of short amyloid peptides

The goal of this work is to understand how the sequence of a protein affects the likelihood that it will form an amyloid fibril and the kinetics along the fibrillization pathway. The focus is on very short fragments of amyloid proteins since these play a role in the fibrillization of the parent protein and can form fibrils themselves. Discontinuous molecular dynamics simulations using the PRIME20 force field were performed of the aggregation of 48‐peptide systems containing SNQNNF ( PrP (170–175 )), SSTSAA (RNaseA(15–20)), MVGGVV (Aβ(35–40)), GGVVIA (Aβ(37–42)), and MVGGVVIA (Aβ(35–42)). In our simulations SNQQNF, SSTTSAA, and MVGGVV form large numbers of fibrillar structures spontaneously (as in experiment). GGVVIA forms β‐sheets that do not stack into fibrils (unlike experiment). The combination sequence MVGGVVIA forms less fibrils than MVGGVV, hindered by the presence of the hydrophobic residues at the C‐terminal. Analysis of the simulation kinetics and energetics reveals why MVGGVV forms fibrils and GGVVIA does not, and why adding I and A to MVGGVVIA reduces fibrillization and enhances amorphous aggregation into oligomeric structures. The latter helps explain why Aβ(1–42) assembles into more complex oligomers than Aβ(1–40), a consequence of which is that it is more strongly associated with Alzheimer's disease. Proteins 2014; 82:1469–1483. © 2014 Wiley Periodicals, Inc.     

Osmolyte controlled fibrillation kinetics of insulin: New insight into fibrillation using the preferential exclusion principle

Amyloid proteins are converted from their native‐fold to long β‐sheet‐rich fibrils in a typical sigmoidal time‐dependent protein aggregation curve. This reaction process from monomer or dimer to oligomer to nuclei and then to fibrils is the subject of intense study. The main results of this work are based on the use of a well‐studied model amyloid protein, insulin, which has been used in vitro by others. Nine osmolyte molecules, added during the protein aggregation process for the production of amyloid fibrils, slow‐down or speed up the process depending on the molecular structure of each osmolyte. Of these, all stabilizing osmolytes (sugars) slow down the aggregation process in the following order: tri > di > monosaccharides, whereas destabilizing osmolytes (urea, guanidium hydrochloride) speed up the aggregation process in a predictable way that fits the trend of all osmolytes. With respect to kinetics, we illustrate, by adapting our earlier reaction model to the insulin system, that the intermediates (trimers, tetramers, pentamers, etc.) are at very low concentrations and that nucleation is orders of magnitude slower than fibril growth. The results are then collated into a cogent explanation using the preferential exclusion and accumulation of osmolytes away from and at the protein surface during nucleation, respectively. Both the heat of solution and the neutral molecular surface area of the osmolytes correlate linearly with two fitting parameters of the kinetic rate model, that is, the lag time and the nucleation rate prior to fibril formation. These kinetic and thermodynamic results support the preferential exclusion model and the existence of oligomers including nuclei and larger structures that could induce toxicity. © 2009 American Institute of Chemical Engineers Biotechnol. Prog., 2009     

Supersaturation-limited and Unlimited Phase Transitions Compete to Produce the Pathway Complexity in Amyloid Fibrillation

Although amyloid fibrils and amorphous aggregates are two types of aggregates formed by denatured proteins, their relationship currently remains unclear. We used β₂-microglobulin (β2m), a protein responsible for dialysis-related amyloidosis, to clarify the mechanism by which proteins form either amyloid fibrils or amorphous aggregates. When ultrasonication was used to accelerate the spontaneous fibrillation of β2m at pH 2.0, the effects observed depended on ultrasonic power; although stronger ultrasonic power effectively accelerated fibrillation, excessively strong ultrasonic power decreased the amount of fibrils formed, as monitored by thioflavin T fluorescence. An analysis of the products formed indicated that excessively strong ultrasonic power generated fibrillar aggregates that retained β-structures but without high efficiency as seeds. On the other hand, when the spontaneous fibrillation of β2m was induced at higher concentrations of NaCl at pH 2.0 with stirring, amorphous aggregates became more dominant than amyloid fibrils. These apparent complexities in fibrillation were explained comprehensively by a competitive mechanism in which supersaturation-limited reactions competed with supersaturation-unlimited reactions. We link the kinetics of protein aggregation and a conformational phase diagram, in which supersaturation played important roles.     

Burial of nonpolar surface area and thermodynamic stabilization of globins as a function of chain elongation

Proteins are biosynthesized from N to C terminus before they depart from the ribosome and reach their bioactive state in the cell. At present, very little is known about the evolution of conformation and the free energy of the nascent protein with chain elongation. These parameters critically affect the extent of folding during ribosome‐assisted biosynthesis. Here, we address the impact of vectorial amino acid addition on the burial of nonpolar surface area and on the free energy of native‐like structure formation in the absence of the ribosomal machinery. We focus on computational predictions on proteins bearing the globin fold, which is known to encompass the 3/3, 2/2, and archaeal subclasses. We find that the burial of nonpolar surface increases progressively with chain elongation, leading to native‐like conformations upon addition of the last C‐terminal residues, corresponding to incorporation of the last two helices. Additionally, the predicted folding entropy for generating native‐like structures becomes less unfavorable at nearly complete chain lengths, suggesting a link between the late burial of nonpolar surface and water release. Finally, the predicted folding free energy takes a progressive favorable dip toward more negative values, as the chain gets longer. These results suggest that thermodynamic stabilization of the native structure of newly synthesized globins during translation in the cell is significantly enhanced as the chain elongates. This is especially true upon departure of the last C‐terminal residues from the ribosomal tunnel, which hosts ca., 30–40 amino acids. Hence, we propose that release from the ribosome is a crucial step in the life of single‐domain proteins in the cell. Proteins 2014; 82:2318–2331. © 2014 Wiley Periodicals, Inc.     

Insight into the correlation between lag time and aggregation rate in the kinetics of protein aggregation

Under favorable conditions, many proteins can assemble into macroscopically large aggregates such as the amyloid fibrils that are associated with Alzheimer's, Parkinson's, and other neurological and systemic diseases. The overall process of protein aggregation is characterized by initial lag time during which no detectable aggregation occurs in the solution and by maximal aggregation rate at which the dissolved protein converts into aggregates. In this study, the correlation between the lag time and the maximal rate of protein aggregation is analyzed. It is found that the product of these two quantities depends on a single numerical parameter, the kinetic index of the curve quantifying the time evolution of the fraction of protein aggregated. As this index depends relatively little on the conditions and/or system studied, our finding provides insight into why for many experiments the values of the product of the lag time and the maximal aggregation rate are often equal or quite close to each other. It is shown how the kinetic index is related to a basic kinetic parameter of a recently proposed theory of protein aggregation. Proteins 2010. © 2010 Wiley‐Liss, Inc.     

Mechanisms of ataxin-3 misfolding and fibril formation: kinetic analysis of a disease-associated polyglutamine protein

The polyglutamine diseases are a family of nine proteins where intracellular protein misfolding and amyloid-like fibril formation are intrinsically coupled to disease. Previously, we identified a complex two-step mechanism of fibril formation of pathologically expanded ataxin-3, the causative protein of spinocerebellar ataxia type-3 (Machado-Joseph disease). Strikingly, ataxin-3 lacking a polyglutamine tract also formed fibrils, although this occurred only via a single-step that was homologous to the first step of expanded ataxin-3 fibril formation. Here, we present the first kinetic analysis of a disease-associated polyglutamine repeat protein. We show that ataxin-3 forms amyloid-like fibrils by a nucleation-dependent polymerization mechanism. We kinetically model the nucleating event in ataxin-3 fibrillogenesis to the formation of a monomeric thermodynamic nucleus. Fibril elongation then proceeds by a mechanism of monomer addition. The presence of an expanded polyglutamine tract leads subsequently to rapid inter-fibril association and formation of large, highly stable amyloid-like fibrils. These results enhance our general understanding of polyglutamine fibrillogenesis and highlights the role of non-poly(Q) domains in modulating the kinetics of misfolding in this family.     

Protein misfolding and aggregation research: Some thoughts on improving quality and utility

Once misfolded and aggregated proteins were as interesting as yesterday's trash, just a bothersome byproduct of productive activities. Today, they attract sustained interest from both basic researchers and practicing engineers. In the burgeoning biopharmaceutical industry, protein misfolding and aggregation pose significant challenges to the economic manufacture of safe and effective protein products. In the clinic, protein aggregates are believed to be pathological agents in a number of serious neurodegenerative disorders, such as Alzheimer's and Parkinson's. Over the past few years, the quantity of research into biotechnological aspects of protein misfolding and aggregation has skyrocketed. However, the quality of the published work is quite variable. In this brief opinion piece, we describe what we believe are some key features of high‐quality publications in protein aggregation. We focus on experimental studies that may also have a kinetic modeling component. © 2013 American Institute of Chemical Engineers Biotechnol. Prog., 29:1109–1115, 2013     

Cysteine inhibits amyloid fibrillation of lysozyme and directs the formation of small worm‐like aggregates through non‐covalent interactions

In this article, we discuss the effects of amino acids on amyloid aggregation of lysozyme. l ‐cysteine (Cys) dramatically inhibited fibrillation of lysozyme, whereas other amino acids (including l ‐arginine) did not. In the presence of Cys, the aggregation pathway of lysozyme shifted from fibrillation to the formation of the small worm‐like aggregates with unfolding. The interaction between Cys and lysozyme was observed to be non‐covalent, suggesting that the thiophilic interaction between the thiol group on the side chain of Cys and the core sequence of lysozyme significantly contributes to the inhibition of amyloid aggregation. These findings provide a new basis for the design of a biocompatible additive to prevent amyloid fibrillation. © 2014 American Institute of Chemical Engineers Biotechnol. Prog., 30:470–478, 2014     

Detection and reduction of microaggregates in insulin preparations

Insulin is an important biotherapeutic protein, and it is also a model protein used to study amyloid diseases, such as Alzheimer's and Parkinson's. The preparation of the protein can lead to small amounts of aggregate in the solution, which in turn may lead to irreproducible in vitro results. Using several pre‐treatment methods, we have determined that pH cycling and diafiltration of the insulin removes microaggregates that may be present in the solution. These microaggregates were not detectable with traditional biochemical methods, but using small‐angle neutron scattering, we were able to show that pH cycling reduces the radius of gyration of the insulin. Diafiltration removes the aggregates by size and pH cycling dissolves the aggregates by adjusting the pH through the pI of the protein. Pre‐treating the insulin with either pH cycling or diafiltration allowed reproducible kinetics of fibrillation for the insulin protein. Microaggregates are a common problem in protein production, formulation, and preparation; here we show that they are the main cause for inconsistent behavior and how pH cycling and diafiltration can mitigate this problem. Biotechnol. Bioeng. 2011; 108:237–241. © 2010 Wiley Periodicals, Inc.     

Biodegradable delivery system containing a peptide inhibitor of polyglutamine aggregation: a step toward therapeutic development in Huntington's disease

Huntington's and eight other neurodegenerative diseases occur because of CAG repeat expansion mutation culminating into an expanded polyglutamine tract in respective protein. In Huntington's disease (HD), a number of CAG repeats beyond normal repeat length (>36) lead to the formation of mutant protein, the proteolytic cleavage of which induces aggregation in polyglutamine length‐dependent manner. The neurodegeneration in this disease is linked to aggregation, and its inhibition is a potential approach for therapeutic development. Although peptides and other molecules have been developed for inhibiting aggregation, peptides in general are susceptible to degradation in vivo conditions. To understand their clinical significance, they also need to be delivered through blood–brain barrier. Here, for the first time, we have synthesized poly‐d ,l ‐lactide‐co‐glycolide nanoparticles containing a polyglutamine aggregation inhibitor peptide PGQ₉[P²], by nanoprecipitation method. This process yielded less than 200 nm spherical nanoparticles with uniform distribution. Characterization studies by infrared spectroscopy‐based and HPLC‐based assays show the presence of PGQ₉[P²] in nanoparticles. In vitro release kinetics demonstrates that nanoparticles release PGQ₉[P²] by erosion and diffusion processes. When the PGQ₉[P²]‐loaded nanoparticles are incubated with aggregation‐prone Q₃₅P₁₀ peptide, representing N‐terminal part of Huntingtin protein, it arrests the elongation phase of Q₃₅P₁₀ aggregation. These findings propose the first step toward delivery of a peptide inhibitor against polyglutamine aggregation in HD. Copyright © 2014 European Peptide Society and John Wiley & Sons, Ltd.     

Solution NMR studies reveal no global flexibility in the catalytic domain of CDC25B

The CDC25B phosphatase is a critical regulator of the cell cycle and has been validated as an important therapeutic target in cancer. Previous studies using molecular dynamics simulations have concluded that the catalytic domain of CDC25B may experience a significant degree of dynamics or be partially disordered in solution, a finding that has a pronounced impact on the structure‐based development of CDC25B inhibitors. We have probed the backbone dynamics of the CDC25B catalytic domain in solution using NMR relaxation experiments and found that the core of the protein is relatively rigid and does not experience any large‐scale dynamics over a broad range of time scales. Furthermore, based on residual dipolar coupling measurements we have concluded that the conformation in solution is very similar to that observed in the crystal form. Importantly, these findings rationalize the application of the CDC25B crystal structure in structure‐based drug development. Proteins 2014; 82:2889–2895. © 2014 Wiley Periodicals, Inc.     

Wild type and mutants of the HET‐s(218–289) prion show different flexibility at fibrillar ends: A simulation study

The C‐terminal segment (residues 218–289) of the HET‐s protein of the filamentous fungus Podosporina anserina is a prion‐forming domain. The structural model of the HET‐s(218–289) amyloid fibril based on solid‐state nuclear magnetic resonance (NMR) restraints shows a β solenoid topology which is comprised of a β‐sheet core and interconnecting loops. For the single‐point mutants Phe286Ala and Trp287Ala, slower aggregation rates in vitro and loss of prionic infectivity have been reported recently. Here we have used molecular dynamics to compare the flexibility of the mutants and wild type. The simulations, initiated from a trimeric aggregate extracted from the NMR structural model, show structural stability on a 100‐ns time scale for wild type and mutants. Analysis of the fluctuations along the simulations reveals that the mutants are less flexible than the wild type in the C‐terminal segment at only one of the two external monomers. Analysis of interaction energy and buried accessible surface indicates that residue Phe286 in particular is stabilized in the Trp287Ala mutant. The simulation results provide an atomistic explanation of the suggestion (based on indirect experimental evidence) that flexibility at the protofibril end(s) is required for fibril elongation. Moreover, they provide further evidence that the growth of the HET‐s amyloid fibril is directional. Proteins 2014; 82:399–404. © 2013 Wiley Periodicals, Inc.     

Quaternary conformational stability: The effect of reversible self‐association on the fibrillation of two insulin analogs

Under conditions relevant to the manufacturing of insulin (e.g., pH 3, room temperature), biosynthetic human insulin (BHI), and Lispro insulin (Lispro) require a nucleation step to initiate aggregation. However, upon seeding with preformed aggregates, both insulins rapidly aggregate into nonnative fibrils. Far ultraviolet circular dichroism (far‐UV CD) and second derivative Fourier transform infrared (2D‐FTIR) spectroscopic analyses show that the fibrillation process involves a change in protein secondary structure from α‐helical in native insulin to predominantly β‐sheet in the nonnative fibrils. After seeding, Lispro aggregates faster than BHI, likely because of a reduced propensity to reversibly self‐associate. Composition gradient multi‐angle light scattering (CG‐MALS) analyses show that Lispro is more monomeric than BHI, whereas their conformational stabilities measured by denaturant‐induced unfolding are statistically indistinguishable. For both BHI and Lispro, as the protein concentration increases, the apparent first‐order rate constant for soluble protein loss decreases. To explain these phenomena, we propose an aggregation model that assumes fibril growth through monomer addition with competitive inhibition by insulin dimers. Biotechnol. Bioeng. 2011;108: 2359–2370. © 2011 Wiley Periodicals, Inc.     

Suppression,disaggregation, and modulation of γ‐Synuclein fibrillation pathway by green tea polyphenol EGCG

Oligomerization of γ‐Synuclein is known to have implications for both neurodegeneration and cancer. Although it is known to co‐exist with the fibrillar deposits of α‐Synuclein (Lewy bodies), a hallmark in Parkinson's disease (PD), the effect of potential therapeutic modulators on the fibrillation pathway of γ‐Syn remains unexplored. By a combined use of various biophysical tools and cytotoxicity assays we demonstrate that the flavonoid epigallocatechin‐3‐gallate (EGCG) significantly suppresses γ‐Syn fibrillation by affecting its nucleation and binds with the unstructured, nucleus forming oligomers of γ‐Syn to modulate the pathway to form α‐helical containing higher‐order oligomers (~158 kDa and ~ 670 kDa) that are SDS‐resistant and conformationally restrained in nature. Seeding studies reveal that these oligomers although “on‐pathway” in nature, are kinetically retarded and rate‐limiting species that slows down fibril elongation. We observe that EGCG also disaggregates the protofibrils and mature γ‐Syn fibrils into similar SDS‐resistant oligomers. Steady‐state and time‐resolved fluorescence spectroscopy and isothermal titration calorimetry (ITC) reveal a weak non‐covalent interaction between EGCG and γ‐Syn with the dissociation constant in the mM range (K_d ~ 2–10 mM). Interestingly, while EGCG‐generated oligomers completely rescue the breast cancer (MCF‐7) cells from γ‐Syn toxicity, it reduces the viability of neuroblastoma (SH‐SY5Y) cells. However, the disaggregated oligomers of γ‐Syn are more toxic than the disaggregated fibrils for MCF‐7cells. These findings throw light on EGCG‐mediated modulation of γ‐Syn fibrillation and suggest that investigation on the effects of such modulators on γ‐Syn fibrillation is critical in identifying effective therapeutic strategies using small molecule modulators of synucleopathies.     

Feature selection and classification of protein–protein complexes based on their binding affinities using machine learning approaches

Protein–protein interactions are intrinsic to virtually every cellular process. Predicting the binding affinity of protein–protein complexes is one of the challenging problems in computational and molecular biology. In this work, we related sequence features of protein–protein complexes with their binding affinities using machine learning approaches. We set up a database of 185 protein–protein complexes for which the interacting pairs are heterodimers and their experimental binding affinities are available. On the other hand, we have developed a set of 610 features from the sequences of protein complexes and utilized Ranker search method, which is the combination of Attribute evaluator and Ranker method for selecting specific features. We have analyzed several machine learning algorithms to discriminate protein‐protein complexes into high and low affinity groups based on their K_d values. Our results showed a 10‐fold cross‐validation accuracy of 76.1% with the combination of nine features using support vector machines. Further, we observed accuracy of 83.3% on an independent test set of 30 complexes. We suggest that our method would serve as an effective tool for identifying the interacting partners in protein–protein interaction networks and human–pathogen interactions based on the strength of interactions. Proteins 2014; 82:2088–2096. © 2014 Wiley Periodicals, Inc.     

Heparin‐induced amyloid fibrillation of β2‐microglobulin explained by solubility and a supersaturation‐dependent conformational phase diagram

Amyloid fibrils are fibrillar deposits of denatured proteins associated with amyloidosis and are formed by a nucleation and growth mechanism. We revisited an alternative and classical view of amyloid fibrillation: amyloid fibrils are crystal‐like precipitates of denatured proteins formed above solubility upon breaking supersaturation. Various additives accelerate and then inhibit amyloid fibrillation in a concentration‐dependent manner, suggesting that the combined effects of stabilizing and destabilizing forces affect fibrillation. Heparin, a glycosaminoglycan and anticoagulant, is an accelerator of fibrillation for various amyloidogenic proteins. By using β₂‐microglobulin, a protein responsible for dialysis‐related amyloidosis, we herein examined the effects of various concentrations of heparin on fibrillation at pH 2. In contrast to previous studies that focused on accelerating effects, higher concentrations of heparin inhibited fibrillation, and this was accompanied by amorphous aggregation. The two‐step effects of acceleration and inhibition were similar to those observed for various salts. The results indicate that the anion effects caused by sulfate groups are one of the dominant factors influencing heparin‐dependent fibrillation, although the exact structures of fibrils and amorphous aggregates might differ between those formed by simple salts and matrix‐forming heparin. We propose that a conformational phase diagram, accommodating crystal‐like amyloid fibrils and glass‐like amorphous aggregates, is important for understanding the effects of various additives.     

Intermolecular disulfide bond formation promotes immunoglobulin aggregation: Investigation by fluorescence correlation spectroscopy

Protein aggregation generally results from association between hydrophobic regions of individual monomers. However, additional mechanisms arising from specific interactions, such as intermolecular disulfide bond formation, may also contribute to the process. The latter is proposed to be the initiating pathway for aggregation of immunoglobulin (IgG), which is essential for triggering its immune response. To test the veracity of this hypothesis, we have employed fluorescence correlation spectroscopy to measure the kinetics of aggregation of IgG in separate experiments either allowing or inhibiting disulfide formation. Fluorescence correlation spectroscopy measurements yielded a diffusion time (τ_D) of ～200 µsec for Rhodamine‐labeled IgG, corresponding to a hydrodynamic radius (R_H) of 56 Å for the IgG monomer. The aggregation kinetics of the protein was followed by monitoring the time evolution of τ_D under conditions in which its cysteine residues were either free or blocked. In both cases, the progress curves confirmed that aggregation proceeded via the nucleation‐dependent polymerization pathway. However, for aggregation in the presence of free cysteines, the lag times were shorter, and the aggregate sizes bigger, than their respective counterparts for aggregation in the presence of blocked cysteines. This result clearly demonstrates that formation of intermolecular disulfide bonds represents a preferred pathway in the aggregation process of IgG. Fluorescence spectroscopy showed that aggregates formed in experiments where disulfide formation was prevented denatured at lower concentration of guanidine hydrochloride than those obtained in experiments where the disulfides were free to form, indicating that intermolecular disulfide bridging is a valid pathway for IgG aggregation. Proteins 2015; 83:169–177. © 2014 Wiley Periodicals, Inc.     

A NanoLuc luciferase-based assay enabling the real-time analysis of protein secretion and injection by bacterial type III secretion systems

The elucidation of the molecular mechanisms of secretion through bacterial protein secretion systems is impeded by a shortage of assays to quantitatively assess secretion kinetics. Also the analysis of the biological role of these secretion systems as well as the identification of inhibitors targeting these systems would greatly benefit from the availability of a simple, quick and quantitative assay to monitor principle secretion and injection into host cells. Here, we present a versatile solution to this need, utilizing the small and very bright NanoLuc luciferase to assess the function of the type III secretion system encoded by Salmonella pathogenicity island 1. Type III secretion substrate–NanoLuc fusions are readily secreted into the culture supernatant, where they can be quantified by luminometry after removal of bacteria. The NanoLuc-based secretion assay features a very high signal-to-noise ratio and sensitivity down to the nanolitre scale. The assay enables monitoring of secretion kinetics and is adaptable to a high throughput screening format in 384-well microplates. We further developed a split NanoLuc-based assay that enables the real-time monitoring of type III secretion-dependent injection of effector–HiBiT fusions into host cells stably expressing the complementing NanoLuc–LgBiT.     

Discovery of binding proteins for a protein target using protein–protein docking‐based virtual screening

Target structure‐based virtual screening, which employs protein‐small molecule docking to identify potential ligands, has been widely used in small‐molecule drug discovery. In the present study, we used a protein–protein docking program to identify proteins that bind to a specific target protein. In the testing phase, an all‐to‐all protein–protein docking run on a large dataset was performed. The three‐dimensional rigid docking program SDOCK was used to examine protein–protein docking on all protein pairs in the dataset. Both the binding affinity and features of the binding energy landscape were considered in the scoring function in order to distinguish positive binding pairs from negative binding pairs. Thus, the lowest docking score, the average Z‐score, and convergency of the low‐score solutions were incorporated in the analysis. The hybrid scoring function was optimized in the all‐to‐all docking test. The docking method and the hybrid scoring function were then used to screen for proteins that bind to tumor necrosis factor‐α (TNFα), which is a well‐known therapeutic target for rheumatoid arthritis and other autoimmune diseases. A protein library containing 677 proteins was used for the screen. Proteins with scores among the top 20% were further examined. Sixteen proteins from the top‐ranking 67 proteins were selected for experimental study. Two of these proteins showed significant binding to TNFα in an in vitro binding study. The results of the present study demonstrate the power and potential application of protein–protein docking for the discovery of novel binding proteins for specific protein targets. Proteins 2014; 82:2472–2482. © 2014 Wiley Periodicals, Inc.     

PAIRpred: Partner‐specific prediction of interacting residues from sequence and structure

We present a novel partner‐specific protein–protein interaction site prediction method called PAIRpred. Unlike most existing machine learning binding site prediction methods, PAIRpred uses information from both proteins in a protein complex to predict pairs of interacting residues from the two proteins. PAIRpred captures sequence and structure information about residue pairs through pairwise kernels that are used for training a support vector machine classifier. As a result, PAIRpred presents a more detailed model of protein binding, and offers state of the art accuracy in predicting binding sites at the protein level as well as inter‐protein residue contacts at the complex level. We demonstrate PAIRpred's performance on Docking Benchmark 4.0 and recent CAPRI targets. We present a detailed performance analysis outlining the contribution of different sequence and structure features, together with a comparison to a variety of existing interface prediction techniques. We have also studied the impact of binding‐associated conformational change on prediction accuracy and found PAIRpred to be more robust to such structural changes than existing schemes. As an illustration of the potential applications of PAIRpred, we provide a case study in which PAIRpred is used to analyze the nature and specificity of the interface in the interaction of human ISG15 protein with NS1 protein from influenza A virus. Python code for PAIRpred is available at http://combi.cs.colostate.edu/supplements/pairpred/ . Proteins 2014; 82:1142–1155. © 2013 Wiley Periodicals, Inc.