Targeting the cryptic sites: NMR-based strategy to improve protein druggability by controlling the conformational equilibrium

Which cryptic sites are feasible drug targets?

Cryptic sites can expand the space of druggable proteins, but the potential usefulness of such sites needs to be investigated before any major effort. Given that the binding pockets are not formed, the druggability of such sites is not well understood. The analysis of proteins and their ligands shows that cryptic sites that are formed primarily by the motion of side chains moving out of the pocket to enable ligand binding generally do not bind drug-sized molecules with sufficient potency. By contrast, sites that are formed by loop or hinge motion are potentially valuable drug targets. Arguments are provided to explain the underlying causes in terms of classical enzyme inhibition theory and the kinetics of side chain motion and ligand binding.

Prediction of protein-ligand binding affinity with deep learning

The prediction of binding affinities between target proteins and small molecule drugs is essential for speeding up the drug research and design process. To attain precise and effective affinity prediction, computer-aided methods are employed in the drug discovery pipeline. In the last decade, a variety of computational methods has been developed, with deep learning being the most commonly used approach. We have gathered several deep learning methods and classified them into convolutional neural networks (CNNs), graph neural networks (GNNs), and Transformers for analysis and discussion. Initially, we conducted an analysis of the different deep learning methods, focusing on their feature construction and model architecture. We discussed the advantages and disadvantages of each model. Subsequently, we conducted experiments using four deep learning methods on the PDBbind v.2016 core set. We evaluated their prediction capabilities in various affinity intervals and statistically and visually analyzed the samples of correct and incorrect predictions for each model. Through visual analysis, we attempted to combine the strengths of the four models to improve the Root Mean Square Error (RMSE) of predicted affinities by 1.6% (reducing the absolute value to 1.101) and the Pearson Correlation Coefficient (R) by 2.9% (increasing the absolute value to 0.894) compared to the current state-of-the-art method. Lastly, we discussed the challenges faced by current deep learning methods in affinity prediction and proposed potential solutions to address these issues.

NMR Characterization of the Papain-like Protease from SARS-CoV-2 Identifies the Conformational Heterogeneity in Its Inhibitor-Binding Site

Papain-like protease (PLpro) from severe acute respiratory syndrome Coronavirus 2 (SARS-CoV-2) is a prime target for the development of antivirals for Coronavirus disease 2019 (COVID-19). However, drugs that target the PLpro protein have not yet been approved. In order to gain insights into the development of a PLpro inhibitor, conformational dynamics of PLpro in complex with GRL0617, the most well-characterized PLpro inhibitor, were investigated using nuclear magnetic resonance (NMR) spectroscopy in solution. Although mutational analyses demonstrated that the L162 sidechain interaction is responsible for the affinity for GRL0617, NMR analyses revealed that L162 in the inhibitor-binding pocket underwent conformational exchange and was not fixed in the conformation in which it formed a contact with ortho-methyl group of GRL0617. The identified conformational dynamics would provide a rationale for the binding mechanism of a covalent inhibitor designed based on GRL0617.

Molecular mechanism of phosphopeptide neoantigen immunogenicity

Altered protein phosphorylation in cancer cells often leads to surface presentation of phosphopeptide neoantigens. However, their role in cancer immunogenicity remains unclear. Here we describe a mechanism by which an HLA-B*0702-specific acute myeloid leukemia phosphoneoantigen, pMLL747-755 (EPR(pS)PSHSM), is recognized by a cognate T cell receptor named TCR27, a candidate for cancer immunotherapy. We show that the replacement of phosphoserine P4 with serine or phosphomimetics does not affect pMHC conformation or peptide-MHC affinity but abrogates TCR27-dependent T cell activation and weakens binding between TCR27 and pMHC. Here we describe the crystal structures for TCR27 and cognate pMHC, map of the interface produced by nuclear magnetic resonance, and a ternary complex generated using information-driven protein docking. Our data show that non-covalent interactions between the epitope phosphate group and TCR27 are crucial for TCR specificity. This study supports development of new treatment options for cancer patients through target expansion and TCR optimization.

Mutual induced-fit mechanism drives binding between intrinsically disordered Bim and cryptic binding site of Bcl-xL

The intrinsically disordered region (IDR) of Bim binds to the flexible cryptic site of Bcl-xL, a pro-survival protein involved in cancer progression that plays an important role in initiating apoptosis. However, their binding mechanism has not yet been elucidated. We have applied our dynamic docking protocol, which correctly reproduced both the IDR properties of Bim and the native bound configuration, as well as suggesting other stable/meta-stable binding configurations and revealed the binding pathway. Although the cryptic site of Bcl-xL is predominantly in a closed conformation, initial binding of Bim in an encounter configuration leads to mutual induced-fit binding, where both molecules adapt to each other; Bcl-xL transitions to an open state as Bim folds from a disordered to an α-helical conformation while the two molecules bind each other. Finally, our data provides new avenues to develop novel drugs by targeting newly discovered stable conformations of Bcl-xL.

Discovery and Characterization of a Cryptic Secondary Binding Site in the Molecular Chaperone HSP70

Heat Shock Protein 70s (HSP70s) are key molecular chaperones that are overexpressed in many cancers and often associated with metastasis and poor prognosis. It has proven difficult to develop ATP-competitive, drug-like small molecule inhibitors of HSP70s due to the flexible and hydrophilic nature of the HSP70 ATP-binding site and its high affinity for endogenous nucleotides. The aim of this study was to explore the potential for the inhibition of HSP70 through alternative binding sites using fragment-based approaches. A surface plasmon resonance (SPR) fragment screen designed to detect secondary binding sites in HSP70 led to the identification by X-ray crystallography of a cryptic binding site in the nucleotide-binding domain (NBD) of HSP70 adjacent to the ATP-binding site. Fragment binding was confirmed and characterized as ATP-competitive using SPR and ligand-observed NMR methods. Molecular dynamics simulations were applied to understand the interactions with the protein upon ligand binding, and local secondary structure changes consistent with interconversion between the observed crystal structures with and without the cryptic pocket were detected. A virtual high-throughput screen (vHTS) against the cryptic pocket was conducted, and five compounds with diverse chemical scaffolds were confirmed to bind to HSP70 with micromolar affinity by SPR. These results identified and characterized a new targetable site on HSP70. While targeting HSP70 remains challenging, the new site may provide opportunities to develop allosteric ATP-competitive inhibitors with differentiated physicochemical properties from current series.

What Features of Ligands Are Relevant to the Opening of Cryptic Pockets in Drug Targets?

Small-molecule drug design aims to identify inhibitors that can specifically bind to a functionally important region on the target, i.e., an active site of an enzyme. Identification of potential binding pockets is typically based on static three-dimensional structures. However, small molecules may induce and select a dynamic binding pocket that is not visible in the apo protein, which presents a well-recognized challenge for structure-based drug discovery. Here, we assessed whether it is possible to identify features in molecules, which we refer to as inducers, that can induce the opening of cryptic pockets. The volume change between apo and bound protein conformations was used as a metric to differentiate chemical features in inducers vs. non-inducers. Based on the dataset of holo–apo pairs, classification models were built to determine an optimum threshold. The model analysis suggested that inducers preferred to be more hydrophobic and aromatic. The impact of sulfur was ambiguous, while phosphorus and halogen atoms were overrepresented in inducers. The fragment analysis showed that small changes in the structures of molecules can strongly affect the potential to induce a cryptic pocket. This analysis and developed model can be used to design inducers that can potentially open cryptic pockets for undruggable proteins.

Drugging Fuzzy Complexes in Transcription

Transcription factors (TFs) are one of the most promising but underutilized classes of drug targets. The high degree of intrinsic disorder in both the structure and the interactions (i.e., “fuzziness”) of TFs is one of the most important challenges to be addressed in this context. Here, we discuss the impacts of fuzziness on transcription factor drug discovery, describing how disorder poses fundamental problems to the typical drug design, and screening approaches used for other classes of proteins such as receptors or enzymes. We then speculate on ways modern biophysical and chemical biology approaches could synergize to overcome many of these challenges by directly addressing the challenges imposed by TF disorder and fuzziness.

Conformational exchange of the Zα domain of human RNA editing enzyme ADAR1 studied by NMR spectroscopy

Z-DNA binding proteins (ZBPs) play important roles in RNA editing, innate immune responses, and viral infections. Numerous studies have implicated a role for conformational motions during ZBPs binding upon DNA, but the quantitative intrinsic conformational exchanges of ZBP have not been elucidated. To understand the correlation between the biological function and dynamic feature of the Zα domains of human ADAR1 (hZα_ADAR1), we have performed the ¹⁵N backbone amide Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion experiments on the free hZα_ADAR1 at two different magnetic fields at 35 °C. The robust inter-dependence of parameters in the global fitting process using multi-magnetic field CPMG profiles allows us characterizing the dynamic properties of conformational changes in hZα_ADAR1. This study found that free hZα_ADAR1 exhibited the conformational exchange with a k_ex of 5784 s^-1 between the states "A" (89% population) and "B" (11% population). The different hydrophobic interactions among helices α1, α2, and α3 between these two states might correlate with efficient Z-DNA binding achieved by the hydrogen bonding interactions between its side-chains and the phosphate backbone of Z-DNA.

Systematic Approach to Find the Global Minimum of Relaxation Dispersion Data for Protein-Induced B-Z Transition of DNA

Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion spectroscopy is commonly used for quantifying conformational changes of protein in μs-to-ms timescale transitions. To elucidate the dynamics and mechanism of protein binding, parameters implementing CPMG relaxation dispersion results must be appropriately determined. Building an analytical model for multi-state transitions is particularly complex. In this study, we developed a new global search algorithm that incorporates a random search approach combined with a field-dependent global parameterization method. The robust inter-dependence of the parameters carrying out the global search for individual residues (GSIR) or the global search for total residues (GSTR) provides information on the global minimum of the conformational transition process of the Zα domain of human ADAR1 (hZαADAR1)-DNA complex. The global search results indicated that a α-helical segment of hZαADAR1 provided the main contribution to the three-state conformational changes of a hZαADAR1-DNA complex with a slow B-Z exchange process. The two global exchange rate constants, kex and kZB, were found to be 844 and 9.8 s-1, respectively, in agreement with two regimes of residue-dependent chemical shift differences-the "dominant oscillatory regime" and "semi-oscillatory regime". We anticipate that our global search approach will lead to the development of quantification methods for conformational changes not only in Z-DNA binding protein (ZBP) binding interactions but also in various protein binding processes.

Single-molecule analysis of interaction between p53TAD and MDM2 using aerolysin nanopores

Protein-protein interactions (PPIs) are regarded as important, but undruggable targets. Intrinsically disordered p53 transactivation domain (p53TAD) mediates PPI with mouse double minute 2 (MDM2), which is an attractive anticancer target for therapeutic intervention. Here, using aerolysin nanopores, we probed the p53TAD peptide/MDM2 interaction and its modulation by small-molecule PPI inhibitors or p53TAD phosphorylation. Although the p53TAD peptide showed short-lived (<100 ms) translocation, the protein complex induced the characteristic extraordinarily long-lived (0.1 s ∼ tens of min) current blockage, indicating that the MDM2 recruitment by p53TAD peptide almost fully occludes the pore. Simultaneously, the protein complex formation substantially reduced the event frequency of short-lived peptide translocation. Notably, the addition of small-molecule PPI inhibitors, Nutlin-3 and AMG232, or Thr18 phosphorylation of p53TAD peptide, were able to diminish the extraordinarily long-lived events and restore the short-lived translocation of the peptide rescued from the complex. Taken together, our results elucidate a novel mechanism of single-molecule sensing for analyzing PPIs and their inhibitors using aerolysin nanopores. This novel methodology may contribute to remarkable improvements in drug discovery targeted against undruggable PPIs.

Cryptic-site binding mechanism of medium-sized Bcl-xL inhibiting compounds elucidated by McMD-based dynamic docking simulations

We have performed multicanonical molecular dynamics (McMD) based dynamic docking simulations to study and compare the binding mechanism between two medium-sized inhibitors (ABT-737 and WEHI-539) that bind to the cryptic site of Bcl-xL, by exhaustively sampling the conformational and configurational space. Cryptic sites are binding pockets that are transiently formed in the apo state or are induced upon ligand binding. Bcl-xL, a pro-survival protein involved in cancer progression, is known to have a cryptic site, whereby the shape of the pocket depends on which ligand is bound to it. Starting from the apo-structure, we have performed two independent McMD-based dynamic docking simulations for each ligand, and were able to obtain near-native complex structures in both cases. In addition, we have also studied their interactions along their respective binding pathways by using path sampling simulations, which showed that the ligands form stable binding configurations via predominantly hydrophobic interactions. Although the protein started from the apo state, both ligands modulated the pocket in different ways, shifting the conformational preference of the sub-pockets of Bcl-xL. We demonstrate that McMD-based dynamic docking is a powerful tool that can be effectively used to study binding mechanisms involving a cryptic site, where ligand binding requires a large conformational change in the protein to occur.

Druggable Transient Pockets in Protein Kinases

Drug discovery using small molecule inhibitors is reaching a stalemate due to low selectivity, adverse off-target effects and inevitable failures in clinical trials. Conventional chemical screening methods may miss potent small molecules because of their use of simple but outdated kits composed of recombinant enzyme proteins. Non-canonical inhibitors targeting a hidden pocket in a protein have received considerable research attention. Kii and colleagues identified an inhibitor targeting a transient pocket in the kinase DYRK1A during its folding process and termed it FINDY. FINDY exhibits a unique inhibitory profile; that is, FINDY does not inhibit the fully folded form of DYRK1A, indicating that the FINDY-binding pocket is hidden in the folded form. This intriguing pocket opens during the folding process and then closes upon completion of folding. In this review, we discuss previously established kinase inhibitors and their inhibitory mechanisms in comparison with FINDY. We also compare the inhibitory mechanisms with the growing concept of “cryptic inhibitor-binding sites.” These sites are buried on the inhibitor-unbound surface but become apparent when the inhibitor is bound. In addition, an alternative method based on cell-free protein synthesis of protein kinases may allow the discovery of small molecules that occupy these mysterious binding sites. Transitional folding intermediates would become alternative targets in drug discovery, enabling the efficient development of potent kinase inhibitors.