Protein–Protein Interaction Prediction for Targeted Protein Degradation

Protein–protein interactions (PPIs) play a fundamental role in various biological functions; thus, detecting PPI sites is essential for understanding diseases and developing new drugs. PPI prediction is of particular relevance for the development of drugs employing targeted protein degradation, as their efficacy relies on the formation of a stable ternary complex involving two proteins. However, experimental methods to detect PPI sites are both costly and time-intensive. In recent years, machine learning-based methods have been developed as screening tools. While they are computationally more efficient than traditional docking methods and thus allow rapid execution, these tools have so far primarily been based on sequence information, and they are therefore limited in their ability to address spatial requirements. In addition, they have to date not been applied to targeted protein degradation. Here, we present a new deep learning architecture based on the concept of graph representation learning that can predict interaction sites and interactions of proteins based on their surface representations. We demonstrate that our model reaches state-of-the-art performance using AUROC scores on the established MaSIF dataset. We furthermore introduce a new dataset with more diverse protein interactions and show that our model generalizes well to this new data. These generalization capabilities allow our model to predict the PPIs relevant for targeted protein degradation, which we show by demonstrating the high accuracy of our model for PPI prediction on the available ternary complex data. Our results suggest that PPI prediction models can be a valuable tool for screening protein pairs while developing new drugs for targeted protein degradation.     

Protein–Protein Docking with Large-Scale Backbone Flexibility Using Coarse-Grained Monte-Carlo Simulations

Most of the protein–protein docking methods treat proteins as almost rigid objects. Only the side-chains flexibility is usually taken into account. The few approaches enabling docking with a flexible backbone typically work in two steps, in which the search for protein–protein orientations and structure flexibility are simulated separately. In this work, we propose a new straightforward approach for docking sampling. It consists of a single simulation step during which a protein undergoes large-scale backbone rearrangements, rotations, and translations. Simultaneously, the other protein exhibits small backbone fluctuations. Such extensive sampling was possible using the CABS coarse-grained protein model and Replica Exchange Monte Carlo dynamics at a reasonable computational cost. In our proof-of-concept simulations of 62 protein–protein complexes, we obtained acceptable quality models for a significant number of cases.     

Challenges in Discovering Drugs That Target the Protein–Protein Interactions of Disordered Proteins

Protein–protein interactions (PPIs) outnumber proteins and are crucial to many fundamental processes; in consequence, PPIs are associated with several pathological conditions including neurodegeneration and modulating them by drugs constitutes a potentially major class of therapy. Classically, however, the discovery of small molecules for use as drugs entails targeting individual proteins rather than targeting PPIs. This is largely because discovering small molecules to modulate PPIs has been seen as extremely challenging. Here, we review the difficulties and limitations of strategies to discover drugs that target PPIs directly or indirectly, taking as examples the disordered proteins involved in neurodegenerative diseases.     

cpxDeepMSA: A Deep Cascade Algorithm for Constructing Multiple Sequence Alignments of Protein–Protein Interactions

Protein–protein interactions (PPIs) are fundamental to many biological processes. The coevolution-based prediction of interacting residues has made great strides in protein complexes that are known to interact. A multiple sequence alignment (MSA) is the basis of coevolution analysis. MSAs have recently made significant progress in the protein monomer sequence analysis. However, no standard or efficient pipelines are available for the sensitive protein complex MSA (cpxMSA) collection. How to generate cpxMSA is one of the most challenging problems of sequence coevolution analysis. Although several methods have been developed to address this problem, no standalone program exists. Furthermore, the number of built-in properties is limited; hence, it is often difficult for users to analyze sequence coevolution according to their desired cpxMSA. In this article, we developed a novel cpxMSA approach (cpxDeepMSA. We used different protein monomer databases and incorporated the three strategies (genomic distance, phylogeny information, and STRING interaction network) used to join the monomer MSA results of protein complexes, which can prevent using a single method fail to the joint two-monomer MSA causing the cpxMSA construction failure. We anticipate that the cpxDeepMSA algorithm will become a useful high-throughput tool in protein complex structure predictions, inter-protein residue-residue contacts, and the biological sequence coevolution analysis.     

alfaNET: A Database of Alfalfa-Bacterial Stem Blight Protein–Protein Interactions Revealing the Molecular Features of the Disease-causing Bacteria

Alfalfa has emerged as one of the most important forage crops, owing to its wide adaptation and high biomass production worldwide. In the last decade, the emergence of bacterial stem blight (caused by Pseudomonas syringae pv. syringae ALF3) in alfalfa has caused around 50% yield losses in the United States. Studies are being conducted to decipher the roles of the key genes and pathways regulating the disease, but due to the sparse knowledge about the infection mechanisms of Pseudomonas, the development of resistant cultivars is hampered. The database alfaNET is an attempt to assist researchers by providing comprehensive Pseudomonas proteome annotations, as well as a host–pathogen interactome tool, which predicts the interactions between host and pathogen based on orthology. alfaNET is a user-friendly and efficient tool and includes other features such as subcellular localization annotations of pathogen proteins, gene ontology (GO) annotations, network visualization, and effector protein prediction. Users can also browse and search the database using particular keywords or proteins with a specific length. Additionally, the BLAST search tool enables the user to perform a homology sequence search against the alfalfa and Pseudomonas proteomes. With the successful implementation of these attributes, alfaNET will be a beneficial resource to the research community engaged in implementing molecular strategies to mitigate the disease. alfaNET is freely available for public use at http://bioinfo.usu.edu/alfanet/.     

Applying Protein–Protein Interactions and Complex Networks to Identify Novel Genes in Retinitis Pigmentosa Pathogenesis

Retinitis Pigmentosa (RP) is a hereditary retinal disorder that causes the atrophy of photoreceptor rod cells. Since individual defective genes converge on the same disease, we hypothesized that all causal genes of RP belong in a complex network. To explore this hypothesis, we conducted a gene connection analysis using 161 genes attributed to RP, compiled from the Retinal Information Network, RetNet. We then examined the protein interaction network (PIN) of these genes. In line with our hypothesis, using STRING, we directly connected 149 genes out of the recognized 159 genes. To uncover the association between the PIN and the ten unrecalled genes, we developed an algorithm to pinpoint the best candidate genes to connect the uncalled genes to the PIN and identified ten such genes. We propose that mutations within these ten genes may also cause RP; this notion is supported by analyzing and categorizing the known causal genes based on cellular locations and related functions. The successful establishment of the PIN among all documented genes and the discovery of novel genes for RP strongly suggest an interconnectedness that causes the disease on the molecular level. In addition, our computational gene search protocol can help identify the genes and loci responsible for genetic diseases, not limited to RP.     

Functional Delineation of a Protein–Membrane Interaction Hotspot Site on the HIV-1 Neutralizing Antibody 10E8

Antibody engagement with the membrane-proximal external region (MPER) of the envelope glycoprotein (Env) of HIV-1 constitutes a distinctive molecular recognition phenomenon, the full appreciation of which is crucial for understanding the mechanisms that underlie the broad neutralization of the virus. Recognition of the HIV-1 Env antigen seems to depend on two specific features developed by antibodies with MPER specificity: (i) a large cavity at the antigen-binding site that holds the epitope amphipathic helix; and (ii) a membrane-accommodating Fab surface that engages with viral phospholipids. Thus, besides the main Fab–peptide interaction, molecular recognition of MPER depends on semi-specific (electrostatic and hydrophobic) interactions with membranes and, reportedly, on specific binding to the phospholipid head groups. Here, based on available cryo-EM structures of Fab–Env complexes of the anti-MPER antibody 10E8, we sought to delineate the functional antibody–membrane interface using as the defining criterion the neutralization potency and binding affinity improvements induced by Arg substitutions. This rational, Arg-based mutagenesis strategy revealed the position-dependent contribution of electrostatic interactions upon inclusion of Arg-s at the CDR1, CDR2 or FR3 of the Fab light chain. Moreover, the contribution of the most effective Arg-s increased the potency enhancement induced by inclusion of a hydrophobic-at-interface Phe at position 100c of the heavy chain CDR3. In combination, the potency and affinity improvements by Arg residues delineated a protein–membrane interaction site, whose surface and position support a possible mechanism of action for 10E8-induced neutralization. Functional delineation of membrane-interacting patches could open new lines of research to optimize antibodies of therapeutic interest that target integral membrane epitopes.     

Small Molecule Arranged Thermal Proximity Coaggregation (smarTPCA)—A Novel Approach to Characterize Protein–Protein Interactions in Living Cells by Similar Isothermal Dose–Responses

Chemical biology and the application of small molecules has proven to be a potent perturbation strategy, especially for the functional elucidation of proteins, their networks, and regulators. In recent years, the cellular thermal shift assay (CETSA) and its proteome-wide extension, thermal proteome profiling (TPP), have proven to be effective tools for identifying interactions of small molecules with their target proteins, as well as off-targets in living cells. Here, we asked the question whether isothermal dose–response (ITDR) CETSA can be exploited to characterize secondary effects downstream of the primary binding event, such as changes in post-translational modifications or protein–protein interactions (PPI). By applying ITDR-CETSA to MAPK14 kinase inhibitor treatment of living HL-60 cells, we found similar dose–responses for the direct inhibitor target and its known interaction partners MAPKAPK2 and MAPKAPK3. Extension of the dose–response similarity comparison to the proteome wide level using TPP with compound concentration range (TPP-CCR) revealed not only the known MAPK14 interaction partners MAPKAPK2 and MAPKAPK3, but also the potentially new intracellular interaction partner MYLK. We are confident that dose-dependent small molecule treatment in combination with ITDR-CETSA or TPP-CCR similarity assessment will not only allow discrimination between primary and secondary effects, but will also provide a novel method to study PPI in living cells without perturbation by protein modification, which we named “small molecule arranged thermal proximity coaggregation” (smarTPCA).     

Unraveling Protein Interactions between the Temperate Virus Bam35 and Its Bacillus Host Using an Integrative Yeast Two Hybrid–High Throughput Sequencing Approach

Bacillus virus Bam35 is the model Betatectivirus and member of the family Tectiviridae, which is composed of tailless, icosahedral, and membrane-containing bacteriophages. Interest in these viruses has greatly increased in recent years as they are thought to be an evolutionary link between diverse groups of prokaryotic and eukaryotic viruses. Additionally, betatectiviruses infect bacteria of the Bacillus cereus group, which are known for their applications in industry and notorious since it contains many pathogens. Here, we present the first protein–protein interactions (PPIs) network for a tectivirus–host system by studying the Bam35–Bacillus thuringiensis model using a novel approach that integrates the traditional yeast two-hybrid system and high-throughput sequencing (Y2H-HTS). We generated and thoroughly analyzed a genomic library of Bam35′s host B. thuringiensis HER1410 and screened interactions with all the viral proteins using different combinations of bait–prey couples. Initial analysis of the raw data enabled the identification of over 4000 candidate interactions, which were sequentially filtered to produce 182 high-confidence interactions that were defined as part of the core virus–host interactome. Overall, host metabolism proteins and peptidases were particularly enriched within the detected interactions, distinguishing this host–phage system from the other reported host–phage PPIs. Our approach also suggested biological roles for several Bam35 proteins of unknown function, including the membrane structural protein P25, which may be a viral hub with a role in host membrane modification during viral particle morphogenesis. This work resulted in a better understanding of the Bam35–B. thuringiensis interaction at the molecular level and holds great potential for the generalization of the Y2H-HTS approach for other virus–host models.     

Functional Roles of PARP2 in Assembling Protein–Protein Complexes Involved in Base Excision DNA Repair

Poly(ADP-ribose) polymerase 2 (PARP2) participates in base excision repair (BER) alongside PARP1, but its functions are still under study. Here, we characterize binding affinities of PARP2 for other BER proteins (PARP1, APE1, Polβ, and XRCC1) and oligomerization states of the homo- and hetero-associated complexes using fluorescence-based and light scattering techniques. To compare PARP2 and PARP1 in the efficiency of PAR synthesis, in the absence and presence of protein partners, the size of PARP2 PARylated in various reaction conditions was measured. Unlike PARP1, PARP2 forms more dynamic complexes with common protein partners, and their stability is effectively modulated by DNA intermediates. Apparent binding affinity constants determined for homo- and hetero-oligomerized PARP1 and PARP2 provide evidence that the major form of PARP2 at excessive PARP1 level is their heterocomplex. Autoregulation of PAR elongation at high PARP and NAD⁺ concentrations is stronger for PARP2 than for PARP1, and the activity of PARP2 is more effectively inhibited by XRCC1. Moreover, the activity of both PARP1 and PARP2 is suppressed upon their heteroPARylation. Taken together, our findings suggest that PARP2 can function differently in BER, promoting XRCC1-dependent repair (similarly to PARP1) or an alternative XRCC1-independent mechanism via hetero-oligomerization with PARP1.     

Red-Edge Excitation Shift Spectroscopy (REES): Application to Hidden Bound States of Ligands in Protein–Ligand Complexes

Ligand-protein binding is responsible for the vast majority of bio-molecular functions. Most experimental techniques examine the most populated ligand-bound state. The determination of less populated, intermediate, and transient bound states is experimentally challenging. However, hidden bound states are also important because these can strongly influence ligand binding and unbinding processes. Here, we explored the use of a classical optical spectroscopic technique, red-edge excitation shift spectroscopy (REES) to determine the number, population, and energetics associated with ligand-bound states in protein–ligand complexes. We describe a statistical mechanical model of a two-level fluorescent ligand located amongst a finite number of discrete protein microstates. We relate the progressive emission red shift with red-edge excitation to thermodynamic parameters underlying the protein–ligand free energy landscape and to photo-physical parameters relating to the fluorescent ligand. We applied the theoretical model to published red-edge excitation shift data from small molecule inhibitor–kinase complexes. The derived thermodynamic parameters allowed dissection of the energetic contribution of intermediate bound states to inhibitor–kinase interactions.     

New Insights into the Interaction of Class II Dihydroorotate Dehydrogenases with Ubiquinone in Lipid Bilayers as a Function of Lipid Composition

The fourth enzymatic reaction in the de novo pyrimidine biosynthesis, the oxidation of dihydroorotate to orotate, is catalyzed by dihydroorotate dehydrogenase (DHODH). Enzymes belonging to the DHODH Class II are membrane-bound proteins that use ubiquinones as their electron acceptors. We have designed this study to understand the interaction of an N-terminally truncated human DHODH (HsΔ29DHODH) and the DHODH from Escherichia coli (EcDHODH) with ubiquinone (Q₁₀) in supported lipid membranes using neutron reflectometry (NR). NR has allowed us to determine in situ, under solution conditions, how the enzymes bind to lipid membranes and to unambiguously resolve the location of Q₁₀. Q₁₀ is exclusively located at the center of all of the lipid bilayers investigated, and upon binding, both of the DHODHs penetrate into the hydrophobic region of the outer lipid leaflet towards the Q₁₀. We therefore show that the interaction between the soluble enzymes and the membrane-embedded Q₁₀ is mediated by enzyme penetration. We can also show that EcDHODH binds more efficiently to the surface of simple bilayers consisting of 1-palmitoyl, 2-oleoyl phosphatidylcholine, and tetraoleoyl cardiolipin than HsΔ29DHODH, but does not penetrate into the lipids to the same degree. Our results also highlight the importance of Q₁₀, as well as lipid composition, on enzyme binding.     

Probing the Mint2 Protein–Protein Interaction Network Relevant to the Pathophysiology of Alzheimer's Disease

The intracellular adaptor protein Mint2 binds amyloid precursor protein (APP) and presenilin‐1, which are both central constituents of the amyloidogenic pathway associated with Alzheimer's disease (AD). Additional interaction partners have also been suggested for Mint2; several of them are also pertinent to AD pathogenesis. However, no comparative mapping of the Mint2 protein–protein interaction network is available. Here we provide a systematic characterization of seven interaction partners and address their specificities towards the different binding domains of Mint2, which reveal domain‐specific and ‐nonspecific interaction partners. Moreover, we show that the last two C‐terminal amino acids of Mint2 are both important for the intramolecular interaction with the PDZ1 domain and for the stability of Mint2.     

Small‐Molecule Inhibitors That Target Protein–Protein Interactions in the RAD51 Family of Recombinases

The development of small molecules that inhibit protein–protein interactions continues to be a challenge in chemical biology and drug discovery. Herein we report the development of indole‐based fragments that bind in a shallow surface pocket of a humanised surrogate of RAD51. RAD51 is an ATP‐dependent recombinase that plays a key role in the repair of double‐strand DNA breaks. It both self‐associates, forming filament structures with DNA, and interacts with the BRCA2 protein through a common “FxxA” tetrapeptide motif. We elaborated previously identified fragment hits that target the FxxA motif site and developed small‐molecule inhibitors that are approximately 500‐fold more potent than the initial fragments. The lead compounds were shown to compete with the BRCA2‐derived Ac‐FHTA‐NH₂ peptide and the self‐association peptide of RAD51, but they had no effect on ATP binding. This study is the first reported elaboration of small‐molecular‐weight fragments against this challenging target.     

Elucidating the Interaction between Pyridoxine 5′-Phosphate Oxidase and Dopa Decarboxylase: Activation of B6-Dependent Enzyme

Pyridoxal 5′-phosphate (PLP), the active form of vitamin B6, serves as a cofactor for scores of B6-dependent (PLP-dependent) enzymes involved in many cellular processes. One such B6 enzyme is dopa decarboxylase (DDC), which is required for the biosynthesis of key neurotransmitters, e.g., dopamine and serotonin. PLP-dependent enzymes are biosynthesized as apo-B6 enzymes and then converted to the catalytically active holo-B6 enzymes by Schiff base formation between the aldehyde of PLP and an active site lysine of the protein. In eukaryotes, PLP is made available to the B6 enzymes through the activity of the B6-salvage enzymes, pyridoxine 5′-phosphate oxidase (PNPO) and pyridoxal kinase (PLK). To minimize toxicity, the cell keeps the content of free PLP (unbound) very low through dephosphorylation and PLP feedback inhibition of PNPO and PLK. This has led to a proposed mechanism of complex formation between the B6-salvage enzymes and apo-B6 enzymes prior to the transfer of PLP, although such complexes are yet to be characterized at the atomic level, presumably due to their transient nature. A computational study, for the first time, was used to predict a likely PNPO and DDC complex, which suggested contact between the allosteric PLP tight-binding site on PNPO and the active site of DDC. Using isothermal calorimetry and/or surface plasmon resonance, we also show that PNPO binds both apoDDC and holoDDC with dissociation constants of 0.93 ± 0.07 μM and 2.59 ± 0.11 μM, respectively. Finally, in the presence of apoDDC, the tightly bound PLP on PNPO is transferred to apoDDC, resulting in the formation of about 35% holoDDC.