Sample efficient reinforcement learning with active learning for molecular design.

Reinforcement learning (RL) is a powerful and flexible paradigm for searching for solutions in high-dimensional action spaces. However, bridging the gap between playing computer games with thousands of simulated episodes and solving real scientific problems with complex and involved environments (up to actual laboratory experiments) requires improvements in terms of sample efficiency to make the most of expensive information. The discovery of new drugs is a major commercial application of RL, motivated by the very large nature of the chemical space and the need to perform multiparameter optimization (MPO) across different properties. In silico methods, such as virtual library screening (VS) and de novo molecular generation with RL, show great promise in accelerating this search. However, incorporation of increasingly complex computational models in these workflows requires increasing sample efficiency. Here, we introduce an active learning system linked with an RL model (RL-AL) for molecular design, which aims to improve the sample-efficiency of the optimization process. We identity and characterize unique challenges combining RL and AL, investigate the interplay between the systems, and develop a novel AL approach to solve the MPO problem. Our approach greatly expedites the search for novel solutions relative to baseline-RL for simple ligand- and structure-based oracle functions, with a 5-66-fold increase in hits generated for a fixed oracle budget and a 4-64-fold reduction in computational time to find a specific number of hits. Furthermore, compounds discovered through RL-AL display substantial enrichment of a multi-parameter scoring objective, indicating superior efficacy in curating high-scoring compounds, without a reduction in output diversity. This significant acceleration improves the feasibility of oracle functions that have largely been overlooked in RL due to high computational costs, for example free energy perturbation methods, and in principle is applicable to any RL domain.

Determination of the Structure and Dynamics of the Fuzzy Coat of an Amyloid Fibril of IAPP Using Cryo-Electron Microscopy.

In recent years, major advances in cryo-electron microscopy (cryo-EM) have enabled the routine determination of complex biomolecular structures at atomistic resolution. An open challenge for this approach, however, concerns large systems that exhibit continuous dynamics. To address this problem, we developed the metadynamic electron microscopy metainference (MEMMI) method, which incorporates metadynamics, an enhanced conformational sampling approach, into the metainference method of integrative structural biology. MEMMI enables the simultaneous determination of the structure and dynamics of large heterogeneous systems by combining cryo-EM density maps with prior information through molecular dynamics, while at the same time modeling the different sources of error. To illustrate the method, we apply it to elucidate the dynamics of an amyloid fibril of the islet amyloid polypeptide (IAPP). The resulting conformational ensemble provides an accurate description of the structural variability of the disordered region of the amyloid fibril, known as fuzzy coat. The conformational ensemble also reveals that in nearly half of the structural core of this amyloid fibril, the side chains exhibit liquid-like dynamics despite the presence of the highly ordered network backbone of hydrogen bonds characteristic of the cross-ß structure of amyloid fibrils.

A Small Molecule Stabilizes the Disordered Native State of the Alzheimer's Aß Peptide.

The stabilization of native states of proteins is a powerful drug discovery strategy. It is still unclear, however, whether this approach can be applied to intrinsically disordered proteins. Here, we report a small molecule that stabilizes the native state of the Aß42 peptide, an intrinsically disordered protein fragment associated with Alzheimer's disease. We show that this stabilization takes place by a disordered binding mechanism, in which both the small molecule and the Aß42 peptide remain disordered. This disordered binding mechanism involves enthalpically favorable local π-stacking interactions coupled with entropically advantageous global effects. These results indicate that small molecules can stabilize disordered proteins in their native states through transient non-specific interactions that provide enthalpic gain while simultaneously increasing the conformational entropy of the proteins.

Conformational Entropy as a Potential Liability of Computationally Designed Antibodies.

In silico antibody discovery is emerging as a viable alternative to traditional in vivo and in vitro approaches. Many challenges, however, remain open to enabling the properties of designed antibodies to match those produced by the immune system. A major question concerns the structural features of computer-designed complementarity determining regions (CDRs), including the role of conformational entropy in determining the stability and binding affinity of the designed antibodies. To address this problem, we used enhanced-sampling molecular dynamics simulations to compare the free energy landscapes of single-domain antibodies (sdAbs) designed using structure-based (DesAb-HSA-D3) and sequence-based approaches (DesAbO), with that of a nanobody derived from llama immunization (Nb10). Our results indicate that the CDR3 of DesAbO is more conformationally heterogeneous than those of both DesAb-HSA-D3 and Nb10, and the CDR3 of DesAb-HSA-D3 is slightly more dynamic than that of Nb10, which is the original scaffold used for the design of DesAb-HSA-D3. These differences underline the challenges in the rational design of antibodies by revealing the presence of conformational substates likely to have different binding properties and to generate a high entropic cost upon binding.

Determination of intermediate state structures in the opening pathway of SARS-CoV-2 spike using cryo-electron microscopy.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the cause of COVID-19, a highly infectious disease that is severely affecting our society and welfare systems. In order to develop therapeutic interventions against this condition, one promising strategy is to target spike, the trimeric transmembrane glycoprotein that the virus uses to recognise and bind its host cells. Here we use a metainference cryo-electron microscopy approach to determine the opening pathway that brings spike from its inactive (closed) conformation to its active (open) one. The knowledge of the structures of the intermediate states of spike along this opening pathway enables us to identify a cryptic pocket that is not exposed in the open and closed states. These results underline the opportunities offered by the determination of the structures of the transient intermediate states populated during the dynamics of proteins to allow the therapeutic targeting of otherwise invisible cryptic binding sites.

A kinetic ensemble of the Alzheimer's Aß peptide.

The conformational and thermodynamic properties of disordered proteins are commonly described in terms of structural ensembles and free energy landscapes. To provide information on the transition rates between the different states populated by these proteins, it would be desirable to generalize this description to kinetic ensembles. Approaches based on the theory of stochastic processes can be particularly suitable for this purpose. Here, we develop a Markov state model and apply it to determine a kinetic ensemble of Aß42, a disordered peptide associated with Alzheimer's disease. Through the Google Compute Engine, we generated 315-µs all-atom molecular dynamics trajectories. Using a probabilistic-based definition of conformational states in a neural network approach, we found that Aß42 is characterized by inter-state transitions on the microsecond timescale, exhibiting only fully unfolded or short-lived, partially folded states. Our results illustrate how kinetic ensembles provide effective information about the structure, thermodynamics and kinetics of disordered proteins.

Small-molecule sequestration of amyloid-ß as a drug discovery strategy for Alzheimer's disease.

Disordered proteins are challenging therapeutic targets, and no drug is currently in clinical use that modifies the properties of their monomeric states. Here, we identify a small molecule (10074-G5) capable of binding and sequestering the intrinsically disordered amyloid-ß (Aß) peptide in its monomeric, soluble state. Our analysis reveals that this compound interacts with Aß and inhibits both the primary and secondary nucleation pathways in its aggregation process. We characterize this interaction using biophysical experiments and integrative structural ensemble determination methods. We observe that this molecule increases the conformational entropy of monomeric Aß while decreasing its hydrophobic surface area. We also show that it rescues a Caenorhabditis elegans model of Aß-associated toxicity, consistent with the mechanism of action identified from the in silico and in vitro studies. These results illustrate the strategy of stabilizing the monomeric states of disordered proteins with small molecules to alter their behavior for therapeutic purposes.

An intrinsically disordered motif regulates the interaction between the p47 adaptor and the p97 AAA+ ATPase.

VCP/p97, an enzyme critical to proteostasis, is regulated through interactions with protein adaptors targeting it to specific cellular tasks. One such adaptor, p47, forms a complex with p97 to direct lipid membrane remodeling. Here, we use NMR and other biophysical methods to study the structural dynamics of p47 and p47-p97 complexes. Disordered regions in p47 are shown to be critical in directing intra-p47 and p47-p97 interactions via a pair of previously unidentified linear motifs. One of these, an SHP domain, regulates p47 binding to p97 in a manner that depends on the nucleotide state of p97. NMR and electron cryomicroscopy data have been used as restraints in molecular dynamics trajectories to develop structural ensembles for p47-p97 complexes in adenosine diphosphate (ADP)- and adenosine triphosphate (ATP)-bound conformations, highlighting differences in interactions in the two states. Our study establishes the importance of intrinsically disordered regions in p47 for the formation of functional p47-p97 complexes.

A Practical Guide to the Simultaneous Determination of Protein Structure and Dynamics Using Metainference.

Accurate protein structural ensembles can be determined with metainference, a Bayesian inference method that integrates experimental information with prior knowledge of the system and deals with all sources of uncertainty and errors as well as with system heterogeneity. Furthermore, metainference can be implemented using the metadynamics approach, which enables the computational study of complex biological systems requiring extensive conformational sampling. In this chapter, we provide a step-by-step guide to perform and analyse metadynamic metainference simulations using the ISDB module of the open-source PLUMED library, as well as a series of practical tips to avoid common mistakes. Specifically, we will guide the reader in the process of learning how to model the structural ensemble of a small disordered peptide by combining state-of-the-art molecular mechanics force fields with nuclear magnetic resonance data, including chemical shifts, scalar couplings and residual dipolar couplings.

Effects of α-tubulin acetylation on microtubule structure and stability.

Acetylation of K40 in α-tubulin is the sole posttranslational modification to mark the luminal surface of microtubules. It is still controversial whether its relationship with microtubule stabilization is correlative or causative. We have obtained high-resolution cryo-electron microscopy (cryo-EM) reconstructions of pure samples of αTAT1-acetylated and SIRT2-deacetylated microtubules to visualize the structural consequences of this modification and reveal its potential for influencing the larger assembly properties of microtubules. We modeled the conformational ensembles of the unmodified and acetylated states by using the experimental cryo-EM density as a structural restraint in molecular dynamics simulations. We found that acetylation alters the conformational landscape of the flexible loop that contains αK40. Modification of αK40 reduces the disorder of the loop and restricts the states that it samples. We propose that the change in conformational sampling that we describe, at a location very close to the lateral contacts site, is likely to affect microtubule stability and function.

Metadynamic metainference: Convergence towards force field independent structural ensembles of a disordered peptide.

Metadynamic metainference has been recently introduced as a theoretical framework to determine structural ensembles by combining and weighting their noise multiple sources of experimental data with molecular mechanics force fields and metadynamics simulations. Here we build upon these initial developments to further extend and streamline the computational approach. We also show that metadynamic metainference can actually determine a structural ensemble for a disordered peptide that is essentially independent from the employed force field. We further show that it is possible to use a very computationally efficient implicit solvent force field in the place of very expensive state-of-the-art explicit solvent ones without a significant loss in accuracy.

Pharmacokinetic properties of peptidic radiopharmaceuticals: reduced uptake of (EH)3-conjugates in important organs.

UNLABELLED: The translation of radiolabeled tumor-targeting peptides into clinical routine is often hampered by an enhanced accumulation into the excreting organs. It has recently been reported that the (EH)3 purification tag is able to improve the biodistribution of Affibody molecules. Therefore, the aim of this study was to prove the positive influence of (EH)3 on the biodistribution of 2 peptidic radiopharmaceuticals, Glu-urea-Lys(Ahx)-HBED-CC and TATE-PEG2-HBED-CC (HBED-CC is N,N'-bis [2-hydroxy-5(carboxyethyl)benzyl] ethylenediamine-N,N'- diacetic acid, TATE is octreotate, and PEG2 is 8-amino-3,6-dioxaoctanoic acid spacer). METHODS: Both compounds were compared with their respective (EH)3-conjugated variants in cell-based in vitro assays and organ distribution. RESULTS: The introduction of (EH)3 to HBED-CC significantly changed the biodistribution profiles. In both cases, the uptake in several organs was reduced whereas tumor uptake was not affected. Most importantly, (EH)3 lowered the kidney and liver uptake of the prostate-specific membrane antigen inhibitor each by a factor of 2.8 and, in the case of octreotate, the liver accumulation by a factor of 51. CONCLUSION: The biodistribution data suggest that (EH)3 is able to improve the pharmacokinetic properties of peptidic radiopharmaceuticals, leading to reduced uptake in organs such as the liver, an important site of metastatic disease.

Discovery of highly potent and selective small molecule ADAMTS-5 inhibitors that inhibit human cartilage degradation via encoded library technology (ELT).

The metalloprotease ADAMTS-5 is considered a potential target for the treatment of osteoarthritis. To identify selective inhibitors of ADAMTS-5, we employed encoded library technology (ELT), which enables affinity selection of small molecule binders from complex mixtures by DNA tagging. Selection of ADAMTS-5 against a four-billion member ELT library led to a novel inhibitor scaffold not containing a classical zinc-binding functionality. One exemplar, (R)-N-((1-(4-(but-3-en-1-ylamino)-6-(((2-(thiophen-2-yl)thiazol-4-yl)methyl)amino)-1,3,5-triazin-2-yl)pyrrolidin-2-yl)methyl)-4-propylbenzenesulfonamide (8), inhibited ADAMTS-5 with IC(50) = 30 nM, showing >50-fold selectivity against ADAMTS-4 and >1000-fold selectivity against ADAMTS-1, ADAMTS-13, MMP-13, and TACE. Extensive SAR studies showed that potency and physicochemical properties of the scaffold could be further improved. Furthermore, in a human osteoarthritis cartilage explant study, compounds 8 and 15f inhibited aggrecanase-mediated (374)ARGS neoepitope release from aggrecan and glycosaminoglycan in response to IL-1ß/OSM stimulation. This study provides the first small molecule evidence for the critical role of ADAMTS-5 in human cartilage degradation.

Reviews Preclinical Safety and Immune-Modulating Effects of Therapeutic Monoclonal Antibodies to Interleukin-6 and Tumor Necrosis Factor-alpha in Cynomolgus Macaques.

Inhibitors of tumor necrosis factor-alpha (TNF-alpha) and interleukin-6 (IL-6) have been shown to be efficacious in a number of autoimmune diseases. In this study, the safety of long-term administration of anti-TNF-alpha and anti-IL-6 monoclonal antibodies (mAbs) was evaluated in cynomolgus macaques. Effects on the immune system were evaluated by analysis of lymphocyte subsets and histopathology of lymphoid tissues. To evaluate the functioning of the immune system, the ability of mAb-treated monkeys to mount a humoral immune response (IgG and IgM) to keyhole limpet hemocyanin (KLH) was evaluated. Treatment with the anti-TNF-alpha mAb produced no histopathological changes in any of the lymphoid tissues examined. There was a small (< 2-fold) elevation in circulating T-and B-lymphocytes during anti-TNF-alpha mAb treatment that was not considered to be toxicologically significant. The antibody response to KLH was unaffected by anti-TNF-alpha mAb treatment. Treatment with anti-IL-6 mAb resulted in a decrease in the size of germinal centers in the spleens of a minority of the animals and a modest but significant decrease in the IgG antibody response to KLH. Weekly intravenous treatment with the anti-IL-6 mAb and twice-weekly subcutaneous treatment with the anti-TNF-alpha mAb for up to 6 months was not associated with any signs of toxicity, and no animal developed an infection throughout the study period. This study demonstrates that the anti-IL-6 and anti-TNF-alpha mAbs produce specific modulating effects on the immune system without rendering the animals immune compromised.