Protein domain-based prediction of drug/compound–target interactions and experimental validation on LIM kinases

Predictive approaches such as virtual screening have been used in drug discovery with the objective of reducing developmental time and costs. Current machine learning and network-based approaches have issues related to generalization, usability, or model interpretability, especially due to the complexity of target proteins’ structure/function, and bias in system training datasets. Here, we propose a new method “DRUIDom” (DRUg Interacting Domain prediction) to identify bio-interactions between drug candidate compounds and targets by utilizing the domain modularity of proteins, to overcome problems associated with current approaches. DRUIDom is composed of two methodological steps. First, ligands/compounds are statistically mapped to structural domains of their target proteins, with the aim of identifying their interactions. As such, other proteins containing the same mapped domain or domain pair become new candidate targets for the corresponding compounds. Next, a million-scale dataset of small molecule compounds, including those mapped to domains in the previous step, are clustered based on their molecular similarities, and their domain associations are propagated to other compounds within the same clusters. Experimentally verified bioactivity data points, obtained from public databases, are meticulously filtered to construct datasets of active/interacting and inactive/non-interacting drug/compound–target pairs (~2.9M data points), and used as training data for calculating parameters of compound–domain mappings, which led to 27,032 high-confidence associations between 250 domains and 8,165 compounds, and a finalized output of ~5 million new compound–protein interactions. DRUIDom is experimentally validated by syntheses and bioactivity analyses of compounds predicted to target LIM-kinase proteins, which play critical roles in the regulation of cell motility, cell cycle progression, and differentiation through actin filament dynamics. We showed that LIMK-inhibitor-2 and its derivatives significantly block the cancer cell migration through inhibition of LIMK phosphorylation and the downstream protein cofilin. One of the derivative compounds (LIMKi-2d) was identified as a promising candidate due to its action on resistant Mahlavu liver cancer cells. The results demonstrated that DRUIDom can be exploited to identify drug candidate compounds for intended targets and to predict new target proteins based on the defined compound–domain relationships. Datasets, results, and the source code of DRUIDom are fully-available at: https://github.com/cansyl/DRUIDom.

Protein Domain-Based Prediction of Compound–Target Interactions and Experimental Validation on LIM Kinases

Predictive approaches such as virtual screening have been used in drug discovery with the objective of reducing developmental time and costs. Current machine learning and network-based approaches have issues related to generalization, usability, or model interpretability, especially due to the complexity of target proteins’ structure/function, and bias in system training datasets. Here, we propose a new computational method “DRUIDom” to predict bio-interactions between drug candidate compounds and target proteins by utilizing the domain modularity of proteins, to overcome problems associated with current approaches. DRUIDom is composed of two methodological steps. First, ligands/compounds are statistically mapped to structural domains of their target proteins, with the aim of identifying physical or functional interactions. As such, other proteins containing the mapped domain or domain pair become new candidate targets for the corresponding compounds. Next, a million-scale dataset of small molecule compounds, including the ones mapped to domains in the previous step, are clustered based on their molecular similarities, and their domain associations are propagated to other compounds within the same clusters. Experimentally verified bioactivity data points, obtained from public databases, are meticulously filtered to construct datasets of active/interacting and inactive/non-interacting compound–target pairs (~2.9M data points), and used as training data for calculating parameters of compound–domain mappings, which led to 27,032 high-confidence associations between 250 domains and 8,165 compounds, and a finalized output of ~5 million new compound–protein interactions. DRUIDom is experimentally validated by syntheses and bioactivity analyses of compounds predicted to target LIM-kinase proteins, which play critical roles in the regulation of cell motility, cell cycle progression, and differentiation through actin filament dynamics. We showed that LIMK-inhibitor-2 and its derivatives significantly block the cancer cell migration through inhibition of LIMK phosphorylation and the downstream protein cofilin. One of the derivative compounds (LIMKi-2d) was identified as a promising candidate due to its action on resistant Mahlavu liver cancer cells. The results demonstrated that DRUIDom can be exploited to identify drug candidate compounds for intended targets and to predict new target proteins based on the defined compound–domain relationships. The datasets, results, and the source code of DRUIDom are fully-available at: https://github.com/cansyl/DRUIDom.

NMR resonance assignments for a docking domain pair with an attached thiolation domain from the PAX peptide-producing NRPS from Xenorhabdus cabanillasii

Non-ribosomal peptide synthetases (NRPSs) are large multienzyme machineries. They synthesize numerous important natural products starting from amino acids. For peptide synthesis functionally specialized NRPS modules interact in a defined manner. Individual modules are either located on a single or on multiple different polypeptide chains. The “peptide-antimicrobial-Xenorhabdus” (PAX) peptide producing NRPS PaxS from Xenorhabdus bacteria consists of the three proteins PaxA, PaxB and PaxC. Different docking domains (DDs) located at the N-termini of PaxB and PaxC and at the C-termini of PaxA and BaxB mediate specific non-covalent interactions between them. The N-terminal docking domains precede condensation domains while the C-terminal docking domains follow thiolation domains. The binding specificity of individual DDs is important for the correct assembly of multi-protein NRPS systems. In many multi-protein NRPS systems the docking domains are sufficient to mediate the necessary interactions between individual protein chains. However, it remains unclear if this is a general feature for all types of structurally different docking domains or if the neighboring domains in some cases support the function of the docking domains. Here, we report the 1H, 13C and 15 N NMR resonance assignments for a C-terminal di-domain construct containing a thiolation (T) domain followed by a C-terminal docking domain (CDD) from PaxA and for its binding partner – the N-terminal docking domain (NDD) from PaxB from the Gram-negative entomopathogenic bacterium Xenorhabdus cabanillasii JM26 in their free states and for a 1:1 complex formed by the two proteins. These NMR resonance assignments will facilitate further structural and dynamic studies of this protein complex.

The solution structure of the complement deregulator FHR5 reveals a compact dimer and provides new insights into CFHR5 nephropathy

The human complement Factor H–related 5 protein (FHR5) antagonizes the main circulating complement regulator Factor H, resulting in the deregulation of complement activation. FHR5 normally contains nine short complement regulator (SCR) domains, but a FHR5 mutant has been identified with a duplicated N-terminal SCR-1/2 domain pair that causes CFHR5 nephropathy. To understand how this duplication causes disease, we characterized the solution structure of native FHR5 by analytical ultracentrifugation and small-angle X-ray scattering. Sedimentation velocity and X-ray scattering indicated that FHR5 was dimeric, with a radius of gyration (Rg) of 5.5 ± 0.2 nm and a maximum protein length of 20 nm for its 18 domains. This result indicated that FHR5 was even more compact than the main regulator Factor H, which showed an overall length of 26–29 nm for its 20 SCR domains. Atomistic modeling for FHR5 generated a library of 250,000 physically realistic trial arrangements of SCR domains for scattering curve fits. Only compact domain structures in this library fit well to the scattering data, and these structures readily accommodated the extra SCR-1/2 domain pair present in CFHR5 nephropathy. This model indicated that mutant FHR5 can form oligomers that possess additional binding sites for C3b in FHR5. We conclude that the deregulation of complement regulation by the FHR5 mutant can be rationalized by the enhanced binding of FHR5 oligomers to C3b deposited on host cell surfaces. Our FHR5 structures thus explained key features of the mechanism and pathology of CFHR5 nephropathy.

Chromophorylation of cyanobacteriochrome Slr1393 from Synechocystis sp. PCC 6803 is regulated by protein Slr2111 through allosteric interaction

Cyanobacteriochromes (CBCRs) are photochromic proteins in cyanobacteria that act as photosensors. CBCRs bind bilins as chromophores and sense nearly the entire visible spectrum of light, but the regulation of the chromophorylation of CBCRs is unknown. Slr1393 from Synechocystis sp. PCC 6803 is a CBCR containing three consecutive GAF (cGMP phosphodiesterase, adenylyl cyclase, and FhlA protein) domains, of which only the third one (Slr1393g3) can be phycocyanobilin-chromophorylated. The protein Slr2111 from Synechocystis sp. PCC 6803 includes a cystathionine β-synthase (CBS) domain pair of an as yet unknown function at its N terminus. CBS domains are often characterized as sensors of cellular energy status by binding nucleotides. In this work, we demonstrate that Slr2111 strongly interacts with Slr1393 in vivo and in vitro, which generates a complex in a 1:1 molar ratio. This tight interaction inhibits the chromophorylation of Slr1393g3, even if the chromophore is present. Instead, the complex stability and thereby the chromophorylation of Slr1393 are regulated by the binding of nucleotides (ATP, ADP, AMP) to the CBS domains of Slr2111 with varying affinities. It is demonstrated that residues Asp-53 and Arg-97 of Slr2111 are involved in nucleotide binding. While ATP binds to Slr2111, the association between the two proteins gets weaker and chromophorylation of Slr1393 are enabled. In contrast, AMP binding to Slr2111 leads to a stronger association, thereby inhibiting the chromophorylation. It is concluded that Slr2111 acts as a sensor of the cellular energy status that regulates the chromophorylation of Slr1393 and thereby its function as a light-driven histidine kinase.

Kinesin-12 motors cooperate to suppress microtubule catastrophes and drive the formation of parallel microtubule bundles

Human Kinesin-12 (hKif15) plays a crucial role in assembly and maintenance of the mitotic spindle. These functions of hKif15 are partially redundant with Kinesin-5 (Eg5), which can cross-link and drive the extensile sliding of antiparallel microtubules. Although both motors are known to be tetramers, the functional properties of hKif15 are less well understood. Here we reveal how single or multiple Kif15 motors can cross-link, transport, and focus the plus-ends of intersecting microtubules. During transport, Kif15 motors step simultaneously along both microtubules with relative microtubule transport driven by a velocity differential between motor domain pairs. Remarkably, this differential is affected by the underlying intersection geometry: the differential is low on parallel and extreme on antiparallel microtubules where one motor domain pair becomes immobile. As a result, when intersecting microtubules are antiparallel, canonical transport of one microtubule along the other is allowed because one motor is firmly attached to one microtubule while it is stepping on the other. When intersecting microtubules are parallel, however, Kif15 motors can drive (biased) parallel sliding because the motor simultaneously steps on both microtubules that it cross-links. These microtubule rearrangements will focus microtubule plus-ends and finally lead to the formation of parallel bundles. At the same time, Kif15 motors cooperate to suppress catastrophe events at polymerizing microtubule plus-ends, raising the possibility that Kif15 motors may synchronize the dynamics of bundles that they have assembled. Thus, Kif15 is adapted to operate on parallel microtubule substrates, a property that clearly distinguishes it from the other tetrameric spindle motor, Eg5.