I need to perform MD simulations for a system comprising a protein and a polymer using the GROMACS software. I have two questions regarding this:
In reality, the polymer is a part of the chromatograph bed. Is it a correct assumption to keep the polymer structure fixed by applying constraints even in the production step?
How can I determine the best protein position and orientation toward the polymer surface as the initial structure? In other words, I have a polymer surface larger than the protein structure, and I want to know how to scan the polymer surface to find the best position and orientation of the protein relative to the polymer.
I would appreciate it if you could share your thoughts and comments with me.
Molecular Dynamics-Guided Design of a Functional Protein–ATRP Conjugate That Eliminates Protein–Protein Interactions
Bibifatima Kaupbayeva
Abstract
📷
Even the most advanced protein–polymer conjugate therapeutics do not eliminate antibody–protein and receptor–protein recognition. Next-generation bioconjugate drugs will need to replace stochastic selection with rational design to select desirable levels of protein–protein interaction while retaining function. The “Holy Grail” for rational design would be to generate functional enzymes that are fully catalytic with small molecule substrates while eliminating interaction between the protein surface and larger molecules. Using chymotrypsin, an important enzyme that is used to treat pancreatic insufficiency, we have designed a series of molecular chimeras with varied grafting densities and shapes. Guided by molecular dynamic simulations and next-generation molecular chimera characterization with asymmetric flow field-flow fractionation chromatography, we grew linear, branched, and comb-shaped architectures from the surface of the protein by atom-transfer radical polymerization. Comb-shaped polymers, grafted from the surface of chymotrypsin, completely prevented enzyme inhibition with protein inhibitors without sacrificing the ability of the enzyme to catalyze the hydrolysis of a peptide substrate. Asymmetric flow field-flow fractionation coupled with multiangle laser light scattering including dynamic light scattering showed that nanoarmor designed with comb-shaped polymers was particularly compact and spherical. The polymer structure significantly increased protein stability and reduced protein–protein interactions. Atomistic molecular dynamic simulations predicted that a dense nanoarmor with long-armed comb-shaped polymer would act as an almost perfect molecular sieve to filter large ligands from substrates. Surprisingly, a conjugate that was composed of 99% polymer was needed before the elimination of protein–protein interactions.
This publication is licensed under
CC-BY-NC-ND 4.0.
📷
📷
📷
📷
Introduction
ARTICLE SECTIONS
Jump To
Proteins combine exquisite specificity and efficiency as catalysts, receptors, and transporters. In exchange for the perfect molecular structure, however, proteins can be unstable. When used therapeutically, proteins can have short lifetimes as a result of receptor-mediated clearance, immunogenicity, and clearance through the kidneys. (1,2) Thus, protein engineers have been interested in improving proteins for functional performance for decades. Successful strategies have included immobilizations on solid supports, (3) genetic engineering, (4) and surface engineering. (5) Taking inspiration from biology, where proteins are modified post-translationally to increase their functional diversity, scientists have used both natural (6,7) and synthetic (8,9) polymers to insulate proteins from their environment. However, even simple protein–polymer conjugates do not retain full functionality, because the polymer component does not completely insulate the protein surface from interaction with other large biomacromolecules. (10)
“Grafting-to” attachment of presynthesized polymers, such as poly(ethylene glycol) (PEG), (11,12) has been used for decades. Studies suggest that PEGylation reduces the immunogenicity of native proteins by shielding antibody binding epitopes on the protein surface. (13) Previous work has shown that the degree of molecular sieving is proportional to polymer grafting density. (14,15) Once a PEG is attached to a protein surface, it has a propensity to interact with the surface, including lysine residues, thereby decreasing the potential of more PEGs to attach. (16) Thus, PEGylation is not an effective tool to generate the high-density polymer coatings which could act as true molecular sieves.
An alternative method of producing protein–polymer conjugates is to grow polymers from protein surfaces. (8,9) The most common grafting-from method, atom-transfer radical polymerization (ATRP), grows polymers from initiators that have been added to a protein surface. The high grafting density and site-specific polymer growth of protein–ATRP has led to the rational synthesis of protein–polymer conjugates with dramatically enhanced solubility, (17,18) stability, (19) and functionality. (20) We have recently shown that one can further increase polymer grafting density by growing two polymer chains from each modified amino group by using a double-headed initiator. (15) These branched conjugates slow the rate of ligand binding by 10-fold versus their linear counterparts.
Although double-headed initiators limited the rate of ligand binding while retaining protein function, branched polymers were still not able to eliminate ligand binding completely. This inability to completely screen protein–protein interactions in functional conjugates remains a vexing problem. Two central problems prevent progress. The first is our inability to predict the structure–function dynamics of protein–polymer conjugates, and the second is our inability to fully understand the structure of the heterogeneous molecular chimeras that we synthesize. Therefore, we decided to perform molecular dynamics simulation to predict the size and shape of protein–ATRP conjugates that would control ligand binding, synthesize a family of conjugates, and analyze structure and function. A library of 7 protein–polymer conjugates was synthesized with chymotrypsin (CT) as a model protein. CT catalyzes peptide and ester hydrolysis but can also be inhibited by large peptide and protein inhibitors. Molecular dynamics-guided strategies were developed to synthesize linear, branched, and comb-shaped CT–polymer conjugates of varying polymer grafting density. CT function was monitored with detailed Michaelis–Menten kinetics, and asymmetric flow field-flow fractionation, coupled with multiangle laser light scattering and quasi elastic light scattering (better known as dynamic light scattering, DLS) (AF4-MALLS-DLS), was used to characterize size and shape. Remarkably, as predicted by molecular dynamics simulations, only extremely dense comb-shaped polymer configurations (99.1% of the conjugate was made up of polymer) retained catalytic activity while eliminating CT inhibition by protein inhibitors at even 10:1 stoichiometry (inhibitor to protein). Indeed, only in the presence of 100-fold excesses of inhibitor is it possible to observe any inhibition.
Discussion
ARTICLE SECTIONS
Jump To
Atomistic Implicit Solvent MD Simulations of Comb-Shaped CT Conjugates
We have previously simulated the dynamics of linear grafted-from protein–polymer conjugates with varied polymer charge, (21) finding that zwitterionic poly(carboxybetaine methacrylate) (pCBMA) stabilized chymotrypsin without compromising the function. (21,22) At first glance, the nonfouling properties of CBMA make it an ideal candidate for creation of nanoarmor to prevent protein–protein interactions. However, we hypothesized that due to weak pCBMA–protein and pCBMA–pCBMA interactions, pCBMA chains may be an efficient molecular sieve. (21) Therefore, we looked to explore, in silico, which polymer grafting density and architecture would be the most optimal to eliminate inhibition while retaining activity. After an initial series of synthetic experiments, we built in silico models of linear and comb-shaped conjugates. Previous studies have shown the importance of polymer grafting density in shielding protein surfaces. In an attempt to gain further molecular insight into the role of polymer grafting density in insulating protein surfaces, we simulated dynamics of 2 linear conjugates with 3 and 7 polymer chains per CT. pCBMA chains were attached to the N-terminus, K36, K202 for L-3 and the N-terminus, K36, K79, K90, K169, K177, and K202 for L-7 (Figure 1a and b). These locations were selected from tertiary structure-based predictions (23) that identified the most reactive amino groups with protein–ATRP initiators. Then, comb-shaped conjugates, C-18 (3 first-generation chains per molecule of CT and 15 second-generation chains per backbone chain, 114,744 atoms) and C-42 (7 first-generation chains per molecule of CT and 35 second-generation chains per backbone chain, 263,044 atoms) CT-pCBMA conjugates were built with first-generation copolymer chains (Figure 1c and d). In these conjugates, each copolymer chain of 50 repeat units had 45 zwitterionic monomers and 5 azide containing monomers. We then simulated dynamics in an implicit solvent with the protein backbone fixed in place. To account for random polymer growth, 5 independent systems were generated for each model with polymer branching at random locations with atactic polymers. Simulations were performed over 200 ns at 0.3 M at 37 °C.
Figure 1 📷Figure 1. Snapshots of CT conjugates during the implicit MD simulation showing the tunnel size leading to the CT active site. (a) L-3 overall conjugate view and tunnel view. (b) L-7 conjugate’s overall view and tunnel view. (c) C-18 overall conjugate view and tunnel view. (d) C-42 conjugate’s overall view and tunnel view. (e) Comparing relative diffusions of a probe with hydrodynamic diameter of 10 nm and with a probe of hydrodynamic diameter of 2 nm.
where Dn and D0 are the diffusion coefficients in the presence and absence of the “network” respectively, ε is the “mesh size” (Å) of the hydrogel, and Rh is the hydrodynamic radius of dextrans.Next, we sought to calculate the diffusivity of CT inhibitors through the polymer shell toward the CT surface. Wallace and colleagues demonstrated the diffusivity of differently sized dextrans in hydrogels. (24) The restriction of diffusivity imposed on dextrans by hydrogels was expressed as a diffusion quotient: the ratio of the diffusion coefficient in the presence of a hydrogel network to that in dilute solution. Treating dextrans as hard spheres, Iggo and colleagues derived the following scaling equation:📷(1)
In order to predict the simulated accessibility of the CT-surface to a large inhibitor versus a smaller substrate, we calculated the free space (tunnel) around the CT active site using the caver software (25) (Figure S2) and fit our data to eq 1. The relative diffusion of probes with hydrodynamic diameters of 10 and 2 nm were calculated (Figure 1e). We also calculated the Dn/D0 50% (the half-ratio of diffusion quotient) as a function of distance from the CT active site (Figure 1e inset). The L-3, L-7, and C-18 conjugates diffusion quotient50% were 0.3, 0.4, and 0.5, respectively. The smallest protein inhibitor of CT, aprotinin (Rg of 11 Å), had a path to the active site of CT for the linear and C-18 conjugates. Long, high-density simple polymers only decreased the diffusivity of an approaching aprotinin when it was ∼15–25 Å from the active site. Thus, the wide tunnel from the bulk solvent to the active site was short enough to enable the unfettered approach of the inhibitor to the enzyme surface. However, the tunnel calculations and relative diffusivity ratios predictions for the high density, long polymers with long arms (the C-42 conjugate) showed that the polymers slowed the rate of diffusion of the inhibitor toward the enzyme at approximately 40–100 Å from the active site. Thus, this particular molecular sieve design should not allow the surfaces of the enzyme and inhibitor to come into contact. We hypothesized that this sieve size may be small enough to selectively allow peptide substrates through but prevent the binding of aprotinin.
Protein–Polymer Conjugate Synthesis and Characterization
Given the molecular dynamics predictions we described above, we proceeded to synthesize a large family of conjugates with varied polymer architectures and grafting densities using grafting-from ATRP (Figure 2). Of the 17 targetable amino groups in CT (3 α-amino group and 14 ε-amino groups), our prior work had shown that up to 13 sites were accessible to simultaneous growth of polymers. (23) We modified chymotrypsin with increasing stoichiometries of a neutral single-headed ATRP initiator, (19) targeting an average of 3, 7, and 12 single-headed initiators per CT molecule. MALDI-ToF mass spectrometry analysis verified the modification degree (an average of 3 chains per molecule of CT for L-3, 7 chains per molecule of CT for L-7, and 12 chains per molecule of CT for L-12), before growing linear chains of pCBMA from the surface of the protein (Figure S3). Since the MD simulations had suggested that the long arms of the target degree of polymerization (DP) of 200 were needed to cover the free space around the protein in comb-shaped conjugates, we used a target DP of 200 for the linear conjugates. The MD simulations also predicted the relationship between density of modification and molecular sieving. We therefore used double-headed initiators to initiate the growth of two polymer chains per modified lysine. The maximum labeling of CT with the larger double-headed initiator was 11 initiators per molecule of CT, which after growth of pCBMA would lead to CT molecules carrying an average of 22 chains per molecule (B-22). Once again, the targeted DP was 200.
Figure 2 📷Figure 2. Synthetic scheme of CT-pCBMA conjugates: (a) linear, (b) branched, and (c) comb-shaped.
We next synthesized comb-shaped polymer architectures with long backbones and long side chains. We designed a straightforward three-step approach in which we used the CT-3, CT-7, and CT-12 macroinitiators to perform copolymer growth (targeted DP of 50) with azide and CBMA comonomers in a 1:4 molar ratio. The azide was then clicked to a DBCO-Br ATRP initiator prior to growth of long CBMA side chains from the backbone (targeted DP of 200). After each step, conjugates were purified via dialysis, lyophilized, and characterized with bicinchoninic acid (BCA) assay (to determine protein content and DP) and 1H NMR (to determine the number of azide groups and DBCO-Br groups per polymer chain) (Table 1, Figures S4–8). We named the highest-density comb-shaped conjugate C-72 (12 first-generation chains per molecule of CT and 60 second-generation chains per backbone chain).
Table 1. Characterization and Activities of Native CT and CT Conjugates
Understanding the true size and complexity of multicomponent disperse biohybrid structures has always been challenging. The molar masses and dispersities of almost all known protein–polymer conjugates have been determined by size exclusion chromatography (Figure S9, Table S1) using calibration standards. Unfortunately, to convert retention time into molar mass, calibration standards must be of identical conformation and density to the sample, (26) and there are no such standards for protein–polymer conjugates. We have routinely used a method that removes intact polymer chains from the conjugates prior to analysis, but ultimately, there can be no replacement for analysis of the absolute molar mass, dispersity, size, and shape of the conjugate itself. We have therefore turned to the use of asymmetric flow field-flow fractionation coupled with multiangle laser light scattering and dynamic light scattering (AF4-MALLS-DLS). AF4 is a channel-based separation technique that is nondestructive to the sample, and more accurately reflects physical properties of a sample in solution. (27) The sample-specific optimization of separation conditions allows for a detailed character
Molecular simulation of protein–polymer conjugates
Author links open overlay panel
,
https://doi.org/10.1016/j.coche.2019.02.006Get rights and content
Covalently bonded proteins and synthetic polymers allow for the design of materials of interest for practical biological and non-biological applications. Chemical functionality, specificity, selectivity, and stimuli response can be engineered through a fundamental understanding of protein–polymer interactions. An overview of the significant role that computer simulations, at the atomistic and mesoscale levels, have played in our understanding of protein–polymer conjugates is provided in this review. Challenges for further development of molecular simulation techniques are discussed, as well as the need for a close interrelation with systematic experimental studies.
Proteins, nature’s own polymers, can fold into unique structures and perform a wide variety of functions within organisms, including catalyzing reactions, selective binding, transporting molecules, responding to stimuli and acting as structural building blocks. While their chemical functionality, selectivity, specificity, and/or responsiveness to stimuli are of great interest to the materials and biomedical community, their practical applications are limited by their sensitivities to the environment. One promising approach in protein engineering is to combine proteins with synthetic polymers to achieve desired properties with increased complexity and modularity [1••,2,3]. Protein–polymer conjugates have been implemented in many biomedical applications, such as protein therapeutics, drug delivery, biomaterials, and biosensing [4, 5, 6, 7]. Modification of proteins with polymers has also been shown to generate conjugates with improved properties for use in non-biological applications such as in biocatalysis and biosensors [8,9].A fundamental understanding of properties, interactions, and mechanisms governing the behavior of protein–polymer conjugates is principal to the design of conjugates with specific desired functionalities. The effects of conjugation depend on a variety of factors, such as the nature of the polymer, the structure of the protein, the conjugation site, the grafting density, and the chemical environment. Computer simulations have played a significant role in our comprehension of these effects. In this review, we present a summary of simulation techniques at the atomistic and mesoscale levels that have been employed to study protein–polymer conjugates. We describe the use of these computational studies in many different approaches and demonstrate how such work has facilitated the understanding of structural and dynamic characterizations of conjugates through the investigation of interactions among protein, polymer, and their environment.
Section snippets
Molecular simulation techniques In atomistic molecular dynamics simulations (aMD), the motion of each atom is governed by Newton’s second law, and thus the position of each atom as a function of time can be obtained. [10, 11, 12]. Simulations can be carried out to refine structures, determine the properties at equilibrium states, and/or to understand the dynamic behavior of the system [13, 14, 15]. Perturbations of specific conditions (i.e. temperature, pressure, pH, initial structure, assembly) can be used to extract
Atomistic insights into improved performance of mono-PEGylated proteins as therapeutics PEGylation, the conjugation of a therapeutically active protein with a stabilizing poly(ethylene glycol) (PEG) polymer, is a well adopted strategy for improving the pharmacokinetic properties of the protein that has led to several clinically approved drugs [25,26]. Several model protein–polymer systems have been investigated to understand the underlying mechanism of enhanced stability and potency of pharmaceuticals upon PEGylation [27••,28,29]. It has been shown that the molecular weight of a
PEGylated peptide as biomaterials Other than PEGylation of therapeutically active proteins, attaching PEG to short peptides has also been investigated using computational simulations for two main reasons: (1) relatively simple peptides can serve as a model system for understanding protein–PEG interactions and (2) PEGylated peptides can be designed and fabricated as biomaterials with tailored properties and functionalities. Keten’s group investigated the effect of PEG conjugation on an α-helix using aMD and CG simulations based
Protein engineering to harvest protein function under non-native conditions While PEGylation is a commonly used approach to develop biocompatible protein therapeutics, there are also research efforts being conducted to go beyond the use of the PEG polymer [1••,2,5,36]. The design of new protein–polymer conjugates to maintain or improve performance of proteins under different conditions – temperature, pH, multi-conjugation sites and solvent – are in high demand for industrial applications. Understanding of the interactions between protein and polymers at an atomistic
New biomaterials based on protein–polymer conjugates Another growing application of protein–polymer conjugates is the design of new biomaterials for tissue engineering and drug delivery [40,41]. One of the fundamental questions regarding protein–polymer conjugates and their assembly revolves around understanding of the overall structure of the conjugate. A recent study combined model fitting and a genetic CG molecular dynamic simulation to interpret SANS data of a globular protein conjugate (see Figure 4) [42]. The red fluorescent protein was
Challenges and outlooks Protein–polymer conjugates not only present a new opportunity for the simulation community but also carry several challenges. For atomistic simulations, one obstacle rests in the compatibility and validity of the force fields developed for proteins and polymers in solution. For example, it has been shown that the use of different force fields for single polymer chains in (water) solutions can lead to very different, even contradictory, phase behaviors [44]. A good balance in the modeling of the
Conflict of interest statement Nothing declared.
References and recommended reading Papers of particular interest, published within the period of review, have been highlighted as: • of special interest •• of outstanding interest
Acknowledgement The authors acknowledge the financial assistance given by the University of Florida Preeminence Initiative.
References (45)
M. Karplus et al.Molecular dynamics simulations of biomolecules Nat Struct Biol (2002)
O. Valsson et al.Enhancing important fluctuations: rare events and metadynamics from a conceptual viewpoint Annu Rev Phys Chem (2016)
F.M. Veronese et al.The impact of PEGylation on biological therapies BioDrugs (2008)
C.N. Lam et al.The shape of protein-polymer conjugates in dilute solution J Polym Sci A: Polym Chem (2016)
J.E. Condon et al.Effect of conjugation on phase transitions in thermoresponsive polymers: an atomistic and coarse-grained simulation study Soft Matter (2017)
S.J. Rukmani et al.A molecular dynamics study of water-soluble polymers: analysis of force fields from atomistic simulations Mol Simul (2018)
A.J. Russell et al.Next generation protein-polymer conjugates AIChE J (2018)
I. Cobo et al.Smart hybrid materials by conjugation of responsive polymers to biomacromolecules Nat Mater (2015)
J.Y. Shu et al.Peptide-polymer conjugates: from fundamental science to application Annu Rev Phys Chem (2013)
Y. Wu et al.Protein-polymer therapeutics: a macromolecular perspective Biomater Sci (2015)
Cited by (19) Thermosensitive hydrogel microneedles for controlled transdermal drug delivery 2022, Acta BiomaterialiaCitation Excerpt :However, for GP which exhibits interaction with proteins and commonly for protein-loaded drug carriers, it is more desirable that the interaction between the carrier and the drug is not too strong because this may affect the molecular structure of the drug during long-term storage. Molecular simulation can provide researchers with details regarding the interaction between polymer materials and protein drugs from the perspective of intermolecular interactions and conformational alteration [49–51]. Silva et al. [52] used molecular simulations to investigate the electrostatic interactions between insulin–chitosan complexes.
Elucidating the mechanisms of the molecular sieving phenomenon created by comb-shaped polymers grafted to a protein – a simulation study 2022, Materials Today Chemistry
Molecular simulation of zwitterionic polypeptides on protecting glucagon-like peptide-1 (GLP-1) 2021, International Journal of Biological MacromoleculesCitation Excerpt :Despite this success, little is known about its microscopic protective mechanism. Molecular simulations, in particular, play a key role in understanding the relationship between structure and function of protein conjugations [25–31]. A hydrophobic interaction mode between PEG and insulin was firstly proposed by Liu and colleges [32].
Bioconjugates – From a specialized past to a diverse future 2020, PolymerCitation Excerpt :Molecular Dynamics. MD simulations have provided valuable insights into the specific interactions that occur between a conjugated polymer and protein [119,120]. Recently Colina and Russell collaborated to synthesize, analyze, and simulate chymotrypsin conjugates with polymers of differing charge states, with cationic pQA, anionic pSMA and zwitterionic pCBMA [100,101].
Polymer-enhanced biomacromolecules 2020, Progress in Polymer ScienceCitation Excerpt :PEGylation of insulin with larger PEG chains also decreased the solvent accessible surface area, thereby shielding the protein from protease hydrolysis and antibody binding. Molecular dynamics simulations are thought to be one of the best methods to determine the PEGylated protein conformation [102]. PEG can either be stretched away from the protein surface (dumbbell-like conformation) or PEG chain wraps around the protein or collapses on protein surface (core-shell conformation).
PEGylation of Insulin and Lysozyme To Stabilize against Thermal Denaturation: A Molecular Dynamics Simulation Study
Frontiers of Chemical Engineering: Molecular Modeling
Edited by
Randy Snurr, Jim Pfaendtner, Veronique Van Speybroeck
Other articles from this issue
View more articlesHighlights of (bio-)chemical tools and visualization software for computational science March 2019David Dubbeldam, …, Sofia Calero Recent advances in thermodynamics and nucleation of gas hydrates using molecular modeling March 2019Shivanand Kumar Veesam, …, Sudeep N Punnathanam Multiscale modeling and simulations of responsive polymers March 2019Zheng Chen, …, Jian Zhou
Molecular dynamics simulation of the glass transition in 4,4′-N,N′-dicarbazolylbiphenyl Chemical Physics Letters, Volume 633, 2015, pp. 41-46Alexey Odinokov, …, Alexander Bagaturyants
Protein-polyelectrolyte complexes: Molecular dynamics simulations and experimental study Polymer, Volume 113, 2017, pp. 39-45Alina A. Sofronova, …, Pavel I. Semenyuk
Atomistic molecular dynamics simulations of the LCST conformational transition in poly(N-vinylcaprolactam) in water Journal of Molecular Graphics and Modelling, Volume 90, 2019, pp. 51-58Oleksii S. Zhelavskyi, Alexander Kyrychenko
Thank you all for your very helpful and informative comments. As Abbas Alibakhshi says, docking is an approach, however usual docking tools like AutoDock does not work for this such big system. For whom are interested to this kind of system, I did the docking using GRAMM (https://gramm.compbio.ku.edu)(an online server tools for docking) and it works properly.