The role of hydrophobic and polar sequence on folding mechanisms of proteins and aggregation of peptides

MINISTRY OF EDUCATION VIETNAM ACADEMY AND TRAINING OF SCIENCE AND TECHNOLOGY GRADUATE UNIVERSITY SCIENCE AND TECHNOLOGY ——————— NGUYEN BA HUNG THE ROLE OF HYDROPHOBIC AND POLAR SEQUENCE ON FOLDING MECHANISMS OF PROTEINS AND AGGREGATION OF PEPTIDES Major: Theoretical and computational physics Code: 9 44 01 03 SUMMARY OF PHYSICS DOCTORAL THESIS HANOI − 2018 INTRODUCTION The problem of protein folding has always been of prime concern in molecular biology. Under normal physiological conditions, most proteins acquire well defined compact three dimensional shapes, known as the native conformations, at which they are biologically active. When proteins are unfolding or misfolding, they not only lose their inherent biological activity but they can also aggregate into insoluble fibrils structures called amyloids which are known to be involved in many degenerative diseases like Alzheimer’s disease, Parkinson’s disease, type 2 diabetes, cerebral palsy, mad cow disease etc. Thus, determining the folded structure and clarifying the mechanism of folding of the protein plays an important role in our understanding of the living organism as well as the human health. Protein aggregation and amyloid formation have also been studied extensively in recent years. Studies have led to the hypothesis that amyloid is the general state of all proteins and is the fundamental state of the system when proteins can form intermolecular interactions. Thus, the tendency for aggregation and for- mation amyloid persists for all proteins and is a trend towards competition with protein folding. However, experiments have also shown that possibility of aggre- gation and aggregation rates depend on solvent conditions and on the amino acid sequence of proteins. Some studies have shown that small amino acid sequences in the protein chain may have a significant effect on the aggregation ability. As a result, knowledge about the link between amino acid sequence and possibility of aggregation is essential for understanding amyloid-related diseases as well as finding a way to treat them. Although all-atom simulations are now widely used molecular biology, the application of these methods in the study of protein folding problem is not feasible due to the limits of computer speed. A suitable approach to the protein folding problem is to use simple theoretical models. There are quite a number of models with different ideas and levels of simplicity, but most notably the Go model and the HP network model and tube model. Considerations of tubular polymer suggest that tubular symmetry is a fun- damental feature of protein molecules which forms the secondary structures of proteins (α and β). Base on this idea, the tube model for the protein was de- veloped by Hoang and Maritan’s team and proposed in 2004. The results of the tube model suggest that this is a simple model and can describes well many of the basic features of protein. The tube model is also the only current model that can simultaneously be used for the study of both folding and aggregation processes. 1 In this thesis, we use a tube model to study the role of hydrophobic and polar sequence on folding mechanism of proteins and aggregation of peptides. Spatial fill of the tubular polymer and hydrogen bonds in the model play the role of background interactions and are independent of the amino acid sequence. The amino acid sequence we consider in the simplified model consists of two types of amino acids, hydrophobic (H) and polar (P). To study the effect of HP sequence on the folding process, we will compare the folding properties of the tube model using the hydrophobic interaction (HP tube model) with tube model using the pairing interaction which is similar to the Go model (Go tube model). This comparison helps to clarify the role of non-native interactions in non-native interactions. To study the role of the HP sequence on aggregation of protein, we will compare the possibility of aggregation of peptide sequences with different HP sequences including the consideration of the shape of the aggregation structures and the properties of aggregation transition phase. In addition, in the study of protein aggregation, we propose an improved model for hydrophobic interaction in the tube model by taking into account the orientation of the side chains of hydrophobic amino acids. Our research shows that this improved model allows for obtaining highly ordered, long-chain aggregation structures like amyloid fibrils. 1. The objectives of the thesis: The aim of the studies is to gain fundamental understanding of the role of hydrophobic and polar sequence on folding mechanism of proteins and aggre- gation of peptides 2. The main contents of the thesis: The general understanding of protein and protein folding, protein aggregation is introduced in chapters 1, 2 of this thesis. Chapter 3 presents the methods used to simulate and analyze the data. The obtained results of role of HP sequence for protein folding are presented in chapter 4. The results of role of HP sequence for protein aggregation are presented in chapter 5. 2 Chapter 1 Protein folding 1.1 Structural properties of proteins Proteins are macromolecules that are synthesized in the cell and responsible for the most basic and important aspects of life. Proteins are polymers (polypep- tides) formed from sequences of 20 diffirent types of amino acids, the monomers of the polymer. The amino acids in the protein differ only in their side chains and are linked together through peptide bonds that form a linear sequence in a particular order. Under normal physiological conditions, most proteins acquire well defined compact three dimensional shapes, knows as the native conformations, at which they are biologically active. The amino acid sequence in the protein determines the structure and function of the protein. Proteins has four types of structure. Primary structure: It is just the chemical sequence of amino acids along the backbone of the protein. These amino acid in chain linked together by peptide bonds. Secondary structure is the spatial arrangement of amino acids. There are two such types of structures: the α-helices and the β-sheets. This kind of structure which maximize the number of hydrogen bonds (H-bonds) between the CO and the NH groups of the backbone. Tertiary structure: A compact packing of the secondary structures comprises tertiary structures. Usually, theses are the full three dimensional structures of proteins. Tertiary structures of large proteins are usually composed of several domains. Quaternary structure: Some proteins are composed of more than one polypep- tide chain. The polypeptide chains may have identical or different amino acid sequences depending on the protein. Each peptide is called a subunit and has its own tertiary structure. The spatial arrangement of these subunits in the protein is called quaternary structure There are a number of semi-empirical interactions that are introduced by chemists and physicists to describe interactions in proteins: disulfide bridges, 3 Coulomb interactions, Hydrogen bonds, Van der Waals interactions, Hydrophobic interactions. 1.2 Protein folding phenomenon Once translated by a ribosome, each polypeptide folds into its characteristic three-dimensional structure from a random coil. Since the fold is maintained by a network of interactions between amino acids in the polypeptide, the native state of the protein chain is determined by the amino acid sequence (hypothesis of thermodynamics). 1.3 Paradox of Levinthal Levinthal paradox which addresses the question: how can proteins possibly find their native state if the number of possible conformations of a polypeptide chain is astronomically large? 1.4 Folding funnel Based on theoretical and empirical research findings, Onuchic and his col- leagues have come up with the idea of the folding funnel as depicted in Figure 1.1. The folding process of the protein in the funnel is the simultaneous reduc- tion of both energy and entropy. As the protein begins to fold, the free energy decreases and the number of configurations decreases (characterized by reduced well width). N folding entropy g en er gy Figure 1.1: The diagram sketches of funnel describes the protein folding energy lanscape 4 Figure 1.2: Free energy lanscape in the two-state model. In this model, ∆F is the diference between the free energy of the folded and unfolded states. ∆FN and , ∆FD, ∆F are the height of barrier from the unfolded and folded states and free energy difference between the N and U states , respectively In the canonical depiction of the folding funnel, the depth of the well repre- sents the energetic stabilization of the native state versus the denatured state, and the width of the well represents the conformational entropy of the system. The surface outside the well is shown as relatively flat to represent the heterogeneity of the random coil state. 1.5 The minimum frustration principle The minimum frustration principle was introduced in 1989 by Bryngelson and Wolynes based on spin glass theory. This principle holds that the amino acid sequence of proteins in nature is optimized through natural selection so that the frustrated caused by interaction in the natural state is minimal. 1.6 Two-state model for protein folding Experimental observations suggest that the two-state model is a common mechanism used to characterize folding dynamics of the majority of small, globuar proteins. In a two-state model of protein folding, the single domain protein can occupy only one of two states: the unfolded state (U) or the folded state (N). The free energy diagram for two-state model is characterized by a large barrier separating the folded state and the unfolded state corresponding minima of the free energy of a reaction coordinate. The free energy difference between the N and U states (∆F ) characterize the degree of stability of the folding state called folding free energy. Rates of folding kf and unfolding ku obey the law Vant Hoff- 5 Arrhennius: kf,u = ν0 exp ( −∆FN,D kBT ) (1.1) For ν0 is constant, T is the temperature and kB is the Boltzmann constant. The change of such as temperature, pressure, and concentration may affect on the ∆F . 1.7 Cooperativity of protein folding Cooperativity is a phenomenon displayed by systems involving identical or near-identical elements, which act dependently of each other. The folding of proteins is cooperative process. In the protein, cooperativity is applied to the two- state process and is understood as the sharpness of thermodynamic transitions. In practice, cooperativity is determined by the parameter measured by the ratio between the enthalpy van’t Hoff and the thermal enthalpy. κ2 = ∆HvH/∆Hcal (1.2) High cooperativity means that the system satisfies the two-state standard and κ2 is closer to 1, the higher the co-operation and vice versa. 1.8 Hydrophobic interaction The hydrophobic effect is the observed tendency of nonpolar substances (such as oil, fat) to aggregate in an aqueous solution and exclude water molecule. The tendency of nonpolar molecules in a polar solvent (usually water) to interact with one another is called the hydrophobic effect. In the case of protein folding, the hydrophobic effect is important to understanding the structure of proteins. The hydrophobic effect is considered to be the major driving force for the folding of globular proteins. It results in the burial of the hydrophobic residues in the core of the protein. 1.9 HP lattice model In the HP lattice model, there are two types of amino acids with respect to their hydrophobicity: polar (P), which tend to be exposed to the solvent on the protein surface, and hydrophobic (H), which tend to be buried inside the globule 6 protein. The folding of the protein is defined as a random step in a 2D or 3D network. Using this model, Dill had design some HP sequence that the minimal energy state in the tight packet configurations was unique. The phase transition of the sequences is designed to be well cooperative. Research shows that aggregate due to hydrophobic interaction is the main driving force for folding. 1.10 Go model The Go model ignores the specificity of amino acid sequences in the protein chain and interaction potential is build based on the structure of the folded state. The basis of the Go model is the maximum consistent principle of protein interac- tions in the folded state. The results of the study show that the Go model for the folding mechanism is quite good with the experiment, especially in determining the contribution of amino acid positions in the polypeptide chain to the transi- tion state during protein folding. . Because the model is based on a native state structure, the Go model can not predict the protein structure from the amino acid sequence that is only used to study the folding process of a known structure. 1.11 Tube model Considerations of symmetry and geometry lead to a description of the pro- tein backbone as a thick polymer or a tube. At low temperatures, a homopoly- mer model as a short tube exhibits two conventional phases: a swollen essen- tially featureless phase and and a conventional compact phase, along with a novel marginally compact phase in between with relatively few optimal structures made up of α-helices and β-sheets. The tube model predicts the existence of a fixed menu of folds determined by geometry, clarifies the role of the amino acid se- quence in selecting the native-state structure from this menu, and explains the propensity for amyloid formation. 7 Chapter 2 Amyloid Formation 2.1 The structure of amyloid fibril (a) (b) Figure 2.1: 3D structure of the Alzheimer’s amyloid-β (1-42)fibrils has a PDB code of 2BEG (a) view along the direction of fibril axis (b) view perpendicular to the direction of fibril axis Amyloid fibrils possess a cross-β structure, in which β-strands are oriented perpendicularly to the fibril axis and are assembled into β-sheets that run the length of the fibrils (Figure 2.1). They generally comprise 24 protofilaments, that often twist around each other. Repeated interactions between hydrophobic and polar groups run along the fibril axis. 2.2 Mechanism of amyloid aggregation The formation of amyloid can be considered to involve at least three steps and are generally referred to as lag phase, growth phase (or elongation) phase and an equilibration phase. Seeding involves the addition of a preformed fibrils to a monomer solution thus increasing the rate of conversion to amyloid fibrils. Ad- dition of seeds decreases the lag phase by eliminating the slow nucleation phase. 8 Chapter 3 Methods and Models for simulations 3.1 HP tube model The backbone of the protein is models as a string of Cα atoms separated by an interval of 3.8A˚, forming a flexible tube of 2.5A˚ also has a constraint with both the tube’s three radii (local and non-local). Potential 3 objects describing this condition are given in figure 3.1) Vtube(i, j, k) = { ∞ if Rijk < ∆ 0 if Rijk ≥ ∆ ∀ i, j, k (3.1) The bending potential in the tube model is related to the spatial constraints of the polypeptide chain. The bending potential at position i given by (Figure 3.1) Vbend(i) =  ∞ if Ri−1,i,i+1 < ∆ eR if ∆ ≤ Ri−1,i,i+1 < 3.2 A˚ 0 if Ri−1,i,i+1 ≥ 3.2 A˚ . (3.2) eR = 0.3  > 0 and the unit  corresponds to the energy of a local hydrogen bond In the tube model, local hydrogen bonds are made up of atoms i and i+3 and assigned to energy equal to −. Non-local hydrogen bonds are formed between the atoms i and j > i + 4 and have the energy of −0.7 . The energy and geometric constraints of a local hydrogen bond between the atom i and the atom j are defined as follows:  j = i+ 3 ehbond = − 4.7 A˚ ≤ rij ≤ 5.6 A˚ |~bi ·~bj| > 0.8 |~bj · ~cij| > 0.94 |~bi · ~cij| > 0.94 (ri,i+1 × ri+1,i+2) · ri+2,i+3 > 0 . (3.3) The same for a non-local hydrogen bond: 9 Non local radius of curvature Hydrophobic interaction Local radius of curvature ݁ோ ݁ௐ ݎ௠௔௫ ݎ ݎ ݕ ݕ ݖ ݖ Figure 3.1: Sketch of the potentials used in the tube model of the protein. r, y are the local radius of curvature, nonlocal radius of curvature; z is distance between two amino acid residues; eR and eW are beding energy and hydrophobic energy  j > i+ 4 ehbond = −0.7  4.1 A˚ ≤ rij ≤ 5.3 A˚ |~bi ·~bj| > 0.8 |~bj · ~cij| > 0.94 |~bi · ~cij| > 0.94 . (3.4) In the tube model, hydrophobic interactions are introduced in the form of paring potential between non-continuous Cα atoms in sequence (j > i+ 1) given by Vhydrophobic(i, j) = { eW rij ≤ 7.5 A˚ 0 rij > 7.5 A˚ , (3.5) eW denotes the hydrophobic interaction energy for each contact, depending on the hydrophobicity of the amino acids i and j. In the most studies, these values were selected by eHH = −0.5 , eHP = ePP = 0. 3.2 Go tube model The Go tube model is a tube model in which hydrophobic interaction energy is replaced by the same energy interaction as the Go-like interaction model: E = Ebend + Ehbond + EGo . (3.6) Thus, the Go tube model retains the geometric and symmetric properties, the 10 bending energy and hydrogen bonds as in tube model. Go-type energy is built on the structure of the given native state. Interactive Go is given by: VGo(i, j) = { Cij eW rij ≤ 7.5 A˚ 0 rij > 7.5 A˚ , (3.7) where Cij are the elements of the native contact map. Cij = 1 if between i and j exist in the native state and Cij = 0 in the other case. An contact in the native state is defined when the distance between two consecutive Cα atoms is less than 7.5 A˚. 3.3 Tube Model with correlated side chain orientations we apply an additional constraint on the hydrophobic contact by taking into account the side chain orientation: ni · cij < 0.5 and −ni · cij < 0.5. Where ni and nj are the normal vectors of the Frenet frames associated with bead i and j, respectively, cij is an unit vector pointing from bead i to bead j. The new constraint is in accordance with the statistics drawn from an analysis of PDB structures 3.4 Structural protein parameters To study the protein folding to the native state, we examine the properties of the protein configurations obtained from the simulation through a number of characteristic features including folding contacts, root mean square deviation (rmsd) and radius of gyration (Rg) . 3.5 Monte Carlo simulation method For studying the folding and aggregation of protein, we carry out multiple in- dependent Monte Carlo (MC) simulations with Metropolis algorithm. The trans- fer of states of the systems in the models used is made by pivot, crank-shaft and tranlocation motion for protein aggregation and pivot, crank-shaft motion for protein folding. 3.6 Parallel tempering Parallel tempering , also known as replica exchange MCMC sampling, is a simulation method aimed at improving the dynamic properties of Monte Carlo 11 method simulations of physical systems, and of Markov chain Monte Carlo (MCMC) sampling methods more generally by exchanges configurations at different tem- peratures. Using Metropolis algorithm to swap two configurations kBA = min {1, exp [(βi − βj) (Ei − Ej)]} (3.8) For kBA is the probability of moving from A to B. This method is very effective to find the basic state simultaneously at each temperature still obtained balanced set and they are easily applied on parallel computers. 3.7 The weighted histogram analysis method The Weighted Histogram Analysis Method (WHAM) allows for optimal anal- ysis of data obtained from MC simulations as well as other simulations over a wide range of parameter