Machine learning in biological physics: From biomolecular prediction to design.

Machine learning has been proposed as an alternative to theoretical modeling when dealing with complex problems in biological physics. However, in this perspective, we argue that a more successful approach is a proper combination of these two methodologies. We discuss how ideas coming from physical modeling neuronal processing led to early formulations of computational neural networks, e.g., Hopfield networks. We then show how modern learning approaches like Potts models, Boltzmann machines, and the transformer architecture are related to each other, specifically, through a shared energy representation. We summarize recent efforts to establish these connections and provide examples on how each of these formulations integrating physical modeling and machine learning have been successful in tackling recent problems in biomolecular structure, dynamics, function, evolution, and design. Instances include protein structure prediction; improvement in computational complexity and accuracy of molecular dynamics simulations; better inference of the effects of mutations in proteins leading to improved evolutionary modeling and finally how machine learning is revolutionizing protein engineering and design. Going beyond naturally existing protein sequences, a connection to protein design is discussed where synthetic sequences are able to fold to naturally occurring motifs driven by a model rooted in physical principles. We show that this model is "learnable" and propose its future use in the generation of unique sequences that can fold into a target structure.

Unraveling the Nano-Bio Interface Interactions of a Lipase Adsorbed on Gold Nanoparticles under Laser Excitation.

The complex nature and structure of biomolecules and nanoparticles and their interactions make it challenging to achieve a deeper understanding of the dynamics at the nano-bio interface of enzymes and plasmonic nanoparticles subjected to light excitation. In this study, circular dichroism (CD) and Raman spectroscopic experiments and molecular dynamics (MD) simulations were used to investigate the potential changes at the nano-bio interface upon plasmonic excitation. Our data showed that photothermal and thermal heating induced distinct changes in the secondary structure of a model nanobioconjugate composed of lipase fromCandida antarcticafraction B (CALB) and gold nanoparticles (AuNPs). The use of a green laser led to a substantial decrease in the α-helix content of the lipase from 66% to 13% and an increase in the ß-sheet content from 5% to 31% compared to the initial conformation of the nanobioconjugate. In contrast, the differences under similar thermal heating conditions were only 55% and 11%, respectively. This study revealed important differences related to the enzyme secondary structure, enzyme-nanoparticle interactions, and the stability of the enzyme catalytic triad (Ser105-Asp187-His224), influenced by the instantaneous local temperature increase generated from photothermal heating compared to the slower rate of thermal heating of the bulk. These results provide valuable insights into the interactions between biomolecules and plasmonic nanoparticles induced by photothermal heating, advancing plasmonic biocatalysis and related fields.

Key aspects of the past 30 years of protein design.

Proteins are the workhorse of life. They are the building infrastructure of living systems; they are the most efficient molecular machines known, and their enzymatic activity is still unmatched in versatility by any artificial system. Perhaps proteins' most remarkable feature is their modularity. The large amount of information required to specify each protein's function is analogically encoded with an alphabet of just ∼20 letters. The protein folding problem is how to encode all such information in a sequence of 20 letters. In this review, we go through the last 30 years of research to summarize the state of the art and highlight some applications related to fundamental problems of protein evolution.

Grafting density induced reentrant disorder-order-disorder transition in planar di-block copolymer brushes.

By means of multiscale molecular simulation, we show that solvophilic-solvophobic AB diblock copolymer brushes in the semi-dilute regime present a re-entrant disorder/order/disorder transition. The latter is fully controllable through two parameters: the grafting density and the solvophobic to solvophilic ratio of the tethered macromolecules. Upon increasing density, chains first aggregate into patches, then further order into a crystalline phase and finally melt into a disordered phase. We demonstrate that the order/disorder transition can be explained through the peculiar properties of the aggregates: upon increasing density, the aggregation number grows as expected. On the contrary, their projection on the plane shrinks, thus melting the emergent ordered phase. Such a density dependent shrinkage, seen for the first time as the cause to an order/disorder phase transition, is as a consequence of the entropic/enthalpic competition that characterises the hierarchical self-assembly of the brush.

In Silico Evidence That Protein Unfolding is a Precursor of Protein Aggregation.

We present a computational study on the folding and aggregation of proteins in an aqueous environment, as a function of its concentration. We show how the increase of the concentration of individual protein species can induce a partial unfolding of the native conformation without the occurrence of aggregates. A further increment of the protein concentration results in the complete loss of the folded structures and induces the formation of protein aggregates. We discuss the effect of the protein interface on the water fluctuations in the protein hydration shell and their relevance in the protein-protein interaction.

Identification of Protein Functional Regions.

Protein sequence stores the information relative to both functionality and stability, thus making it difficult to disentangle the two contributions. However, the identification of critical residues for function and stability has important implications for the mapping of the proteome interactions, as well as for many pharmaceutical applications, e. g. the identification of ligand binding regions for targeted pharmaceutical protein design. In this work, we propose a computational method to identify critical residues for protein functionality and stability and to further categorise them in strictly functional, structural and intermediate. We evaluate single site conservation and use Direct Coupling Analysis (DCA) to identify co-evolved residues both in natural and artificial evolution processes. We reproduce artificial evolution using protein design and base our approach on the hypothesis that artificial evolution in the absence of any functional constraint would exclusively lead to site conservation and co-evolution events of the structural type. Conversely, natural evolution intrinsically embeds both functional and structural information. By comparing the lists of conserved and co-evolved residues, outcomes of the analysis on natural and artificial evolution, we identify the functional residues without the need of any a priori knowledge of the biological role of the analysed protein.

H-Bonding-mediated binding and charge reorganization of proteins on gold nanoparticles.

Once introduced into the human body, nanoparticles often interact with blood proteins, which in turn undergo structural changes upon adsorption. Although protein corona formation is a widely studied phenomenon, the structure of proteins adsorbed on nanoparticles is far less understood. We propose a model to describe the interaction between human serum albumin (HSA) and nanoparticles (NPs) with arbitrary coatings. Our model takes into account the competition between protonated and unprotonated polymer ends and the curvature of the NPs. To this end, we explored the effects of surface ligands (citrate, PEG-OMe, PEG-NH2, PEG-COOH, and glycan) on gold nanoparticles (AuNPs) and the pH of the medium on structural changes in the most abundant protein in blood plasma (HSA), as well as the impact of such changes on cytotoxicity and cellular uptake. We observed a counterintuitive effect on the ζ-potential upon binding of negatively charged HSA, while circular dichroism spectroscopy at various pH values showed an unexpected pattern in the reduction of α-helix content, as a function of surface chemistry and curvature. Our model qualitatively reproduces the decrease in α-helix content, thereby offering a rationale based on particle curvature. The simulations quantitatively reproduce the charge inversion measured experimentally through the ζ-potential of the AuNPs in the presence of HSA. Finally, we found that AuNPs with adsorbed HSA display lower toxicity and slower cell uptake rates, compared to functionalized systems in the absence of protein. Our study allows examining and explaining the conformational dynamics of blood proteins triggered by NPs and corona formation, thereby opening new avenues toward designing safer NPs for drug delivery and nanomedical applications.

Implementing efficient concerted rotations using Mathematica and C code⋆.

In this article we demonstrate a general and efficient metaprogramming implementation of concerted rotations using Mathematica. Concerted rotations allow the movement of a fixed portion of a polymer backbone with fixed bending angles, like a protein, while maintaining the correct geometry of the backbone and the initial and final points of the portion fixed. Our implementation uses Mathematica to generate a C code which is then wrapped in a library by a Python script. The user can modify the Mathematica notebook to generate a set of concerted rotations suited for a particular backbone geometry, without having to write the C code himself. The resulting code is highly optimized, performing on the order of thousands of operations per second.

Limiting the valence: advancements and new perspectives on patchy colloids, soft functionalized nanoparticles and biomolecules.

Limited bonding valence, usually accompanied by well-defined directional interactions and selective bonding mechanisms, is nowadays considered among the key ingredients to create complex structures with tailored properties: even though isotropically interacting units already guarantee access to a vast range of functional materials, anisotropic interactions can provide extra instructions to steer the assembly of specific architectures. The anisotropy of effective interactions gives rise to a wealth of self-assembled structures both in the realm of suitably synthesized nano- and micro-sized building blocks and in nature, where the isotropy of interactions is often a zero-th order description of the complicated reality. In this review, we span a vast range of systems characterized by limited bonding valence, from patchy colloids of new generation to polymer-based functionalized nanoparticles, DNA-based systems and proteins, and describe how the interaction patterns of the single building blocks can be designed to tailor the properties of the target final structures.

Assessment of the MARTINI 3 Performance for Short Peptide Self-Assembly.

The coarse-grained MARTINI force field, initially developed for membranes, has proven to be an exceptional tool for investigating supramolecular peptide assemblies. Over the years, the force field underwent refinements to enhance accuracy, enabling, for example, the reproduction of protein-ligand interactions and constant pH behavior. However, these protein-focused improvements seem to have compromised its ability to model short peptide self-assembly. In this study, we assess the performance of MARTINI 3 in reproducing peptide self-assembly using the well-established diphenylalanine (FF) as our test case. Unlike its success in version 2.1, FF does not even exhibit aggregation in version 3. By systematically exploring parameters for the aromatic side chains and charged backbone beads, we established a parameter set that effectively reproduces tube formation. Remarkably, these parameter adjustments also replicate the self-assembly of other di- and tripeptides and coassemblies. Furthermore, our analysis uncovers pivotal insights for enhancing the performance of MARTINI in modeling short peptide self-assembly. Specifically, we identify issues stemming from overestimated hydrophilicity arising from charged termini and disruptions in π-stacking interactions due to insufficient planarity in aromatic groups and a discrepancy in intermolecular distances between this and backbone-backbone interactions. This investigation demonstrates that strategic modifications can harness the advancements offered by MARTINI 3 for the realm of short peptide self-assembly.

Correction to "Targeting the Spike: Repurposing Mithramycin and Dihydroergotamine to Block SARS-CoV-2 Infection".

[This corrects the article DOI: 10.1021/acsomega.3c02921.].

Sequence controlled self-knotting colloidal patchy polymers.

Knotted chains are a promising class of polymers with many applications for materials science and drug delivery. Here we introduce an experimentally realizable model for the design of chains with controllable topological properties. Recently, we have developed a systematic methodology to construct self-assembling chains of simple particles, with final structures fully controlled by the sequence of particles along the chain. The individual particles forming the chain are colloids decorated with mutually interacting patches, which can be manufactured in the laboratory with current technology. Our methodology is applied to the design of sequences folding into self-knotting chains, in which the end monomers are by construction always close together in space. The knotted structure can then be externally locked simply by controlling the interaction between the end monomers, paving the way to applications in the design and synthesis of active materials and novel carriers for drugs delivery.

Bioinspired Polyethylene Glycol Coatings for Reduced Nanoparticle-Protein Interactions.

Nanoparticles (NPs) and other engineered nanomaterials have great potential as nanodrugs or nanomedical devices for biomedical applications. However, the adsorption of proteins in blood circulation or similar physiological fluids can significantly alter the surface properties and therapeutic response induced by most nanomaterials. For example, interaction with proteins can change the bloodstream circulation time and availability of therapeutic NPs or hinder the accumulation in their desired target organs. Proteins can also trigger or prevent agglomeration. By combining experimental and computational approaches, we have developed NPs carrying polyethylene glycol (PEG) polymeric coatings that mimic the surface charge distribution of proteins typically found in blood, which are known to show low aggregation under normal blood conditions. Here, we show that NPs with coatings based on apoferritin or human serum albumin display better antifouling properties and weaker protein interaction compared to similar NPs carrying conventional PEG polymeric coatings.

Targeting the Spike: Repurposing Mithramycin and Dihydroergotamine to Block SARS-CoV-2 Infection.

The urgency to find complementary therapies to current SARS-CoV-2 vaccines, whose effectiveness is preserved over time and not compromised by the emergence of new and emerging variants, has become a critical health challenge. We investigate the possibility of jamming the opening of the Receptor Binding Domain (RBD) of the spike protein of SARS-CoV-2 with small compounds. Through in silico screening, we identified two potential candidates that would lock the Receptor Binding Domain (RBD) in a closed configuration, preventing the virus from infecting the host cells. We show that two drugs already approved by the FDA, mithramycin and dihydroergotamine, can block infection using concentrations in the µM range in cell-based assays. Further STD-NMR experiments support dihydroergotamine's direct interaction with the spike protein. Overall, our results indicate that repurposing of these compounds might lead to potential clinical drug candidates for the treatment of SARS-CoV-2 infection.

Ultrasensitive detection of SARS-CoV-2 spike protein by graphene field-effect transistors.

COVID-19, caused by the severe acute respiratory syndrome-coronavirus 2 (SARS-CoV-2), originated a global health crisis, causing over 2 million casualties and altering human daily life all over the world. This pandemic emergency revealed the limitations of current diagnostic tests, highlighting the urgency to develop faster, more precise and sensitive sensors. Graphene field effect transistors (GFET) are analytical platforms that enclose all these requirements. However, the design of a sensitive and robust GFET is not a straightforward objective. In this work, we report a GFET array biosensor for the detection of SARS-CoV-2 spike protein using the human membrane protein involved in the virus internalisation: angiotensin-converting enzyme 2 (ACE2). By finely controlling the graphene functionalisation, by tuning the Debye length, and by deeply characterising the ACE2-spike protein interactions, we have been able to detect the target protein with an extremely low limit of detection (2.94 aM). This work set the basis for a new class of analytical platforms, based on human membrane proteins, with the potential to detect a broad variety of pathogens, even before their isolation, being a powerful tool in the fight against future pandemics.

Controlling the folding and substrate-binding of proteins using polymer brushes.

The extent of coupling between the folding of a protein and its binding to a substrate varies from protein to protein. Some proteins have highly structured native states in solution, while others are natively disordered and only fold fully upon binding. In this Letter, we use Monte Carlo simulations to investigate how disordered polymer chains grafted around a binding site affect the folding and binding of three model proteins. The protein that approaches the substrate fully folded is more hindered during the binding process than those whose folding and binding are cooperative. The polymer chains act as localized crowding agents and can select correctly folded and bound configurations in favor of nonspecifically adsorbed states. The free energy change for forming all intraprotein and protein-substrate contacts can depend nonmonotonically on the polymer length.

Telechelic star polymers as self-assembling units from the molecular to the macroscopic scale.

By means of multiscale molecular simulations, we show that telechelic-star polymers are a simple, robust, and tunable system, which hierarchically self-assembles into soft-patchy particles and mechanically stabilizes selected, open crystalline structures. The self-aggregating patchy behavior can be fully controlled by the number of arms per star and by the fraction of attractive monomeric units at the free ends of the arms. Such self-assembled soft-patchy particles while forming, upon augmenting density, gel-like percolating networks, preserve properties as particle size, number, and arrangement of patches per particle. In particular, we demonstrate that the flexibility inherent in the soft-patchy particles brings forward a novel mechanism that leads to the mechanical stability of diamond and simple cubic crystals over a wide range of densities, and for molecular sizes ranging from about 10 nm up to the micrometer scale.

Insights into Hepatitis C Virus E2core Interactions with Human Cellular Receptor CD81 at Different pHs from Molecular Simulations.

Hepatitis C virus (HCV) is the second viral agent that causes the majority of chronic hepatic infections worldwide, following Hepatitis B virus (HBV) infection. HCV infection comprises several steps, from the attachment to the receptors to the delivery of the viral genetic material and replication inside the cells. Tetraspanin CD81 is a key entry factor for HCV as it accompanies the virus during attachment and internalization through clathrin-mediated endocytosis. HCV-CD81 binding takes place through the viral glycoprotein E2. We performed full-atom molecular dynamics simulations reproducing the pH conditions that occur during the viral attachment to the hepatocytes (pH 7.4) and internalization (pH 6.2-4.6). We observed that changing the pH from 7.4 to 6.2 triggers a large conformational change in the binding orientation between E2core (E2core corresponds to residues 412-645 of the viral glycoprotein E2) and CD81LEL (CD81LEL corresponds to residues 112-204 of CD81) that occurs even more rapidly at low pH 4.6. This pH-induced switching mechanism has never been observed before and could allow the virus particles to sense the right moment during the maturation of the endosome to start fusion.

Protein design under competing conditions for the availability of amino acids.

Isolating the properties of proteins that allow them to convert sequence into the structure is a long-lasting biophysical problem. In particular, studies focused extensively on the effect of a reduced alphabet size on the folding properties. However, the natural alphabet is a compromise between versatility and optimisation of the available resources. Here, for the first time, we include the impact of the relative availability of the amino acids to extract from the 20 letters the core necessary for protein stability. We present a computational protein design scheme that involves the competition for resources between a protein and a potential interaction partner that, additionally, gives us the chance to investigate the effect of the reduced alphabet on protein-protein interactions. We devise a scheme that automatically identifies the optimal reduced set of letters for the design of the protein, and we observe that even alphabets reduced down to 4 letters allow for single protein folding. However, it is only with 6 letters that we achieve optimal folding, thus recovering experimental observations. Additionally, we notice that the binding between the protein and a potential interaction partner could not be avoided with the investigated reduced alphabets. Therefore, we suggest that aggregation could have been a driving force in the evolution of the large protein alphabet.

Multi-scale simulations provide supporting evidence for the hypothesis of intramolecular protein translocation in GroEL/GroES complexes.

The biological function of chaperone complexes is to assist the folding of non-native proteins. The widely studied GroEL chaperonin is a double-barreled complex that can trap non-native proteins in one of its two barrels. The ATP-driven binding of a GroES cap then results in a major structural change of the chamber where the substrate is trapped and initiates a refolding attempt. The two barrels operate anti-synchronously. The central region between the two barrels contains a high concentration of disordered protein chains, the role of which was thus far unclear. In this work we report a combination of atomistic and coarse-grained simulations that probe the structure and dynamics of the equatorial region of the GroEL/GroES chaperonin complex. Surprisingly, our simulations show that the equatorial region provides a translocation channel that will block the passage of folded proteins but allows the passage of secondary units with the diameter of an alpha-helix. We compute the free-energy barrier that has to be overcome during translocation and find that it can easily be crossed under the influence of thermal fluctuations. Hence, strongly non-native proteins can be squeezed like toothpaste from one barrel to the next where they will refold. Proteins that are already fairly close to the native state will not translocate but can refold in the chamber where they were trapped. Several experimental results are compatible with this scenario, and in the case of the experiments of Martin and Hartl, intra chaperonin translocation could explain why under physiological crowding conditions the chaperonin does not release the substrate protein.