Session D22: Physics of Proteins: Peptides and Small Proteins

Principles governing catalytic activity of self-assembled short peptides

Molecular self-assembly provides a chemical strategy for the synthesis of nanostructures by using the principles of nature, and peptides serve as the promising building block to construct adaptable molecular architectures. Recently, a series of hepta-peptides with alternative hydrophobic and hydrophilic residues were reported to form amyloid-like structures, which were capable of catalyzing acyl ester hydrolysis with remarkable efficiency. However, it remains elusive about the atomic structures of the fibrils: what is the origin of the sequence-dependent catalytic activity? How does the ester hydrolysis catalyzed by the fibril? In this work, the atomic structure of the aggregates was determined by using molecular modelling and further validated by solid-state NMR experiments, where the fibril with high activity adopts twisted parallel configuration within each layer, and the one with low activity is in flat antiparallel configuration [1]. The polymorphism originates from the interactions between different regions of the building block peptides, where the delicate balance between rigidity and flexibility plays an important role. We further show that the p-nitrophenylacetate (pNPA) hydrolysis reactions catalyzed by two different fibrils follow similar mechanism, and the difference in microenvironment at the active site between the natural enzyme and the present self-assembled fibrils should accounts for the different catalytic activities. The present work provides atomic understanding of the structure and function of self-assembled fibrils formed with short-peptides, and thus sheds new insight on designing aggregates with better functions.

References
[1] J. Am. Chem. Soc. 2019, 141, 223.   

Representation of the Conformational Ensemble of Peptides in Coarse Grained Simulations

Proteins/peptides can adopt to different environments by altering their conformation. Novel experimental techniques have enabled a quantitative characterization of this structural heterogeneity. In molecular dynamics simulations capturing this conformational ensemble quantitatively remains a major challenge. Even in atomistic simulations one has to find the best force field for the molecule of interest. With coarse grained (CG) simulations, where the aim is to reduce the degrees of freedom to reach the relevant length and time scales, representation of the conformational ensemble becomes even more problematic. Here, we revisit a recent CG model from our group, which was designed and tested for representing the aggregation driven conformational change of LKα14 peptide. We demonstrate here, that the structure/physics based approach used in the original parameterization of our CG model, strongly depends on the reference system chosen and excluded volume interactions. The updated model can recover the whole conformational ensemble in a quantitative manner, while maintaining the aggregation driven conformational transformation feature. This balanced parametrization leads to a sequence transferable CG model.   

Polarization of Intrinsically Disordered Proteins in the Presence of Charged Biopolymers: the Case of Tau Protein

Tau protein is an intrinsically disordered protein known to be associated with the progression of neurodegenerative diseases. In cells, tau interacts with charged macromolecules, including high charge density polyanions such as RNA and microtubules. Although attraction between tau protein and negatively charged biopolymers is thought to play an important role in cell function, the mechanisms leading to attraction between tau protein and negatively charged biopolymers remain poorly understood. We study tau protein conformations and polarization using molecular dynamics simulations. First, we study the electrostatically driven collapse of tau protein, and connect the resulting conformation to features of its charge sequence. We then study how the conformations of tau protein change in the presence of a constant electric field and in the presence of a negatively charged biopolymer. With the latter, we quantify polarization induced attraction of tau protein to biopolymers. The implications of conformational changes of tau protein as it interacts with polyanions inside the cell will be discussed.   

Strategic incorporation of fluorinated prolines can lead to faster folding and stable proteins

A single atom substitution in a proline residue of ribonuclease A can lead to accelerated folding into its 3D structure with increased thermostability. It has been suggested that this can be caused by stereo-electronic effects due to fluorination of amino acids. In this work we use quantum chemistry and QM/MM methods coupled with accelerated sampling to study the static and dynamic properties of proline containing peptides in water, with a focus on the effect of fluorination on the free energy surface and conformational properties. We compare the free energy surfaces of plain vs fluorinated dipeptides along the amide dihedral and the proline ring pucker reaction coordinates to examine the effect of fluorination. Finally, we characterize the important interactions in the system: namely intramolecular H-bonding, n -> pi* and gauche interactions in the presence of explicit solvent. We find that the experimental data can be understood from the change in the dihedral potential surface due to electronic structure differences due to fluorination. The impact of this on the conformational properties of peptides and self-assembly in solution is discussed.   

Low Temperature Protein Refolding Suggested by Molecular Simulation

The function of critical biological materials, such as proteins, is intrinsically tied to their structure, and this structure is in turn heavily dependent on the properties of the solvent, most commonly water. As water is known to exhibit anomalous properties, especially at supercooled temperatures, it is natural to ask how these properties might impact the thermodynamics of protein folding. To investigate this question, we use molecular simulation to explore the behavior of a model protein, Trp-cage, as low as 70 K below the freezing point of the solvent. Surprisingly, we find that while the expected cold denaturation of the protein is observed at moderate supercooling, further cooling to more than 55 K below the freezing point leads to cold refolding of the protein. Structural and hydrogen bonding analysis suggests that this refolding is driven by desolvation of the protein’s hydrophobic core, likely related to the pronounced decrease in density at this temperature. Beyond their intrinsic fundamental interest, these results have implications for cryo-microscopy and cryo-preservation, where biological materials are often transiently subjected to these extreme conditions.   

van der Waals Forces in Biomolecular Systems: from Solvation to Long-range Interaction Mechanisms

One decisive characteristic of the biomolecular machinery is the access to a rich set of coordinated processes within a small energy window. Most of these processes involve collective conformational changes and occur in an aqueous environment. Conformational changes of (bio)molecules as well as their interaction with water are thereby largely governed by non-covalent van der Waals (vdW) dispersion interactions. By virtue of their intrinsically collective nature, vdW forces also represent a key influence on collective nuclear behavior. Our understanding of vdW interactions in large-scale (bio)molecular systems, however, is still rather limited [Chem. Soc. Rev. 2019, 48, 4118]. Here, we employ the Many-Body Dispersion framework to investigate the vdW interaction in biomolecular systems and its spatial and spectral aspects. In particular, we show the role of beyond-pairwise vdW forces for protein stability and highlight the delocalized character of the protein-water vdW interaction. We further examine intermolecular electronic behaviors and reveal a coexistence of strong delocalization with spatially-separated yet correlated, local domains. This, ultimately, forms the basis for a potential, efficient long-range interaction mechanism for coordinated processes in biomolecular systems.   

The Assessment of MD Force Fields with Respect to Alanine Conformations in Aqueous Solutions

The ability to reproduce conformational ensembles of the central alanine residue in GAG and AAA obtained from experimentally-derived φ- and ψ-dependent J coupling constants and amide I’ profiles is examined for six molecular dynamics (MD) force fields (Amber ff14SB, Amber ff99SBnmr1, Amber ff03ws, OPLS-AA/L, OPLS-AA/M, and CHARMM36). Compared to the empirical Gaussian model, which is constructed to best fit the experimental data, MD-derived Ramachandran plots produce overly constricted polyproline II (pPII) basin, an overpopulated antiparallel β basin, and an underpopulated transitional β basin. Our results show that Amber ff14SB best reproduces the experimental J coupling constants and yields the highest pPII populations of the central alanine residue in GAG (55%) and AAA (63%), in a good agreement with the the Gaussian model (59% and 76%). The comparision between experimental and MD-derived results for GAG in water is extended to various water/ethanol mixtures in order to further evaluate Amber ff14SB, CHARMM36, and OPLS-AA/M. Amber ff14SB again outperforms CHARMM36 and OPLS-AA/M in reproducing experimental J coupling constants and amide I’ profiles.   

Developing an explicit solvent model for protein aggregation

Protein aggregation is responsible for amyloid formation and implicated in many neurological diseases. Since protein aggregation is a slow process, intermediate resolution protein models with implicit solvent have been developed to extend the time scale accessible to computer simulations. These models have been improved over many years and have become a valuable tool to support and interpret experimental work on protein aggregation. However, explicit solvent models are required to describe such processes as thermophoresis, which has recently become a tool to study fibril formation. In this work, we start from an off-lattice, mesoscale protein model with implicit solvent to develop an explicit solvent model that reproduces the equilibrium conformational and aggregation properties of the original model. To this end, we choose a solvent that is compatible with the protein model and perform simulations where the interaction parameters are adjusted during the simulation to match the desired properties. Applying the approach to short peptides in water, we find that solvent-solvent interactions are essential.   

A kinetic analysis of local fluctuations in ubiquitin by combining the LE4PD normal modes and Markov state modeling

Follwowing the conformational selection hypothesis, accurately determining the location and magnitude of fluctuations along a protein’s primary sequence is important in describing its mechanism of binding, so methods describing precisely the fluctuation dynamics of proteins can help reveal their biological function. Here, we model the fluctuation dynamics and kinetics of the protein ubiquitin using a coarse-grained description of the protein’s dynamics, the Langevin Equation for Protein Dynamics (LE4PD), which decomposes the dynamics of a protein into dynamical pathways that explore mode-dependent free-energy surfaces. Using as input to the theory statistics from a molecular dynamics simulation, we calculate the timescales of the slow LE4PD modes using Markov state models.The predicted timescales of these LE4PD modes can be elucidated using the committor function, and a version of the string method is used to extract real-space fluctuations of the protein from the mode representation. We find the dynamics predicted by the slow LE4PD modes correspond to motion in important binding regions of the protein. We also show that the fastest LE4PD modes correspond to localized, conserved fluctuations along the protein’s primary sequence.   

Non-additive effects of denaturing and protective cosolvents on protein stability

The conformational stability of the Trpcage protein, in the presence of pure and mixed solutions of two denaturants, urea and guanidinium chloride (GdmCl), and one protective osmolyte, trimethylamine N-oxide (TMAO), are studied using enhanced-sampling all-atomistic molecular dynamics simulations. We find that the pure and the mixed solutions of urea and GdmCl denature Trpcage as a whole, but remarkably, the helical segment ¹NLYIQWL⁷ of Trpcage is stabilized in mixed GdmCl-urea solutions. For this helical segment, we find that urea "over solvates" the peptide backbone by reorganizing water molecules from the peptide side chains to the peptide backbone and GdmCl strongly dehydrates the side chains. The effects of urea and GdmCl on the solvation structure of the peptide are non-additive and urea depletes Gdm⁺ from the surface of the peptide in mixed urea-GdmCl solutions. An intricate thermodynamic balance between these non-additive effects stabilizes the helix in mixed urea-GdmCl solutions. Interestingly, we find that the protective osmolyte TMAO also depletes Gdm⁺ from the peptide-surface in mixed TMAO-GdmCl solutions and the mixture of TMAO and GdmCl is less effective than GdmCl solutions in counteracting the urea-denaturation of the helix.   

Short peptides assemble to produce enzyme-like catalysts.

Design of a novel catalytic function in proteins and peptides, apart from its inherent practical value, is important for fundamental understanding of enzymatic activity. We will present applications of a minimalistic approach to design of artificial enzymes. We designed a series of short peptides that self-assemble into amyloid-like fibrils to act as Zn²⁺-dependent hydrolases. Zn²⁺ helps stabilize the fibril formation, while also acting as a cofactor to catalyze acyl ester hydrolysis and carbon dioxide hydration. These results indicate that amyloid fibrils are able to not only catalyze their own formation – they also can catalyze chemical reactions. Excitingly, the specific activities shown by these catalytic amyloids are on par with those shown by natural enzymes. Thus, amyloids might have served as intermediates in the evolution of modern-day enzymes. This work has implications for the design of self-assembling nanostructured catalysts including ones containing a variety of biological and non-biological metal ions.

Nature Chem. 2014, 6, 303-309; Proc. Natl. Acad. Sci. U.S.A. 2017, 114, 6191-6196; ACS Cat. 2018, 8, 59-62.