Role of alkylated residues in the tetrapeptide self-assembly – A molecular dynamics study

Abstract

Designing peptide sequences that self-assemble into well-defined nanostructures can open a new venue for the development of novel drug carriers and molecular contrast agents. Current approaches are often based on a linear block-design of amphiphilic peptides where a hydrophilic peptide chain is terminated by a hydrophobic tail. Here, a new template for a self-assembling tetrapeptide (YXKX, Y = tyrosine, X = alkylated tyrosine, K = lysine) is proposed with two distinct sides relative to the peptide’s backbone: alkylated hydrophobic residues on one side and hydrophilic residues on the other side. Using all-atom molecular dynamics simulations, the self-assembly pathway of the tetrapeptide is analyzed for two different concentrations. At both concentrations, tetrapeptides self-assembled into a nanosphere structure. The alkylated tyrosines initialize the self-assembly process via a strong hydrophobic effect and to reduce exposure to the aqueous solvent, they formed a hydrophobic core. The hydrophilic residues occupied the surface of the self-assembled nanosphere. Ordered arrangement of tetrapeptides within the nanosphere with the backbone hydrogen bonding led to a beta sheet formation. Alkyl chain length constrained the size and shape of the nanosphere. This study provides foundation for further exploration of self-assembling structures that are based on peptides with hydrophobic and hydrophilic moieties located on the opposite sides of a peptide backbone.

Graphical Abstract

New template is proposed for a self-assembling tetrapeptide (YXKX, Y = tyrosine, X = alkylated tyrosine, K = lysine) with with two distinct sides relative to the peptide’s backbone: alkylated hydrophobic residues on one side and hydrophilic residues on the other side.

Introduction

Peptides, which can self-assemble into well-ordered nanostructures, have received increasing attention in recent years.¹ Their biocompatibility and multifunctionality make them ideal candidates for several biomedical applications, such as drug delivery², bioimaging³, tissue engineering⁴, and biosensors.⁵ Peptide self-assembly is a spontaneous process that is driven by non-covalent interactions, including the van der Waals forces, pi-pi stacking, and hydrophobic and electrostatic interactions.⁶ Individually, these intermolecular interactions are weak, but as a collective force, they can stabilize the structure. In addition, backbone atoms of a peptide strengthen self-assembled structures by forming hydrogen bonds between adjacent peptides. This process promotes a secondary structure formation that depends on a peptide’s amino acid sequence. The hydrophobic-to-hydrophilic ratio of amino acids in a particular peptide sequence determines an overall self-assembly pathway. A self-assembling peptide sequence can be designed based on peptide motifs isolated from natural proteins (top-down approach) or by using an amino acid library (bottom-up approach). A top-down approach that uses short peptide fragments derived from proteins involved in the amyloid formation of neurogenerative diseases has been extensively investigated by many research groups.^7–10 Further, it has been shown that short peptide diphenyalanine⁷ and its derivatives^8–9 form different kinds of nanostructures. The formation of these nanostructures depends on the peptide’s concentration and aromatic stacking interactions arising from phenylalanine residues that drive the self-assembly.

The bottom-up approach can result in an enormous number of possible combinations of 20 amino acids. Previously, the self-assembly propensity of short peptides, such as dipeptides¹¹ (400 combinations) and tripeptides¹² (8000 combinations), was predicted with the help of molecular dynamics (MD) simulations. Though hydrophobic peptides have higher tendency to aggregate compared to hydrophilic peptides, simulation results showed only a weak correlation between the aggregation propensity and hydrophilicity. But placing a particular amino acid depending upon its property in a more favorable position in the peptide sequence promoted the aggregation.^12–13 By having a clear understanding of the role of intermolecular interactions and the dynamic behavior of individual amino acids over the aggregation mechanism, one could design a peptide sequence for a programmed self-assembly according to a specific application.

The bottom-up approach often rely on amphiphilic peptides. This design is inspired by natural lipids as it entails a hydrophilic head and a hydrophobic tail.^14–15 Typically, the hydrophilic head is composed of charged amino acids while the hydrophobic tail can be designed in a number of different ways. A surfactant-like peptide contains a hydrophobic tail composed of a combination of six hydrophobic amino acids (G, A, V, L, I, and F).^16–17 Alternatvely, a saturated fatty acid tail can be conjugated to the N-terminal of long-chain peptides to introduce a hydrophobic moiety.^18–19 Zhang et al. designed an ionic-complementary self-assembling peptide, inspired by the Z-DNA binding protein zuotin.²⁰ This peptide is also referred to as the molecular lego peptide, as the hydrophobic and hydrophilic residues within the peptide sequence are arranged like the pegs and holes in a Lego brick. In this peptide, residues with positive and negative charges repeat periodically (Modulus I: -+-+-+-+, Modulus II: --++--++ and so on).^21–24 Switching hydrophilic residues in the peptide sequence has a significant effect on the self-assembly process with respect to the charge, but changing hydrophobic residues does not have a similar effect on the nanostructure formation.^25–26

Here we propose and evaluate a new template for amphiphilic self-assembling peptide where amino acid side chains in the peptide’s sequence are substituted with alkylated hydrophobic residues on one side of the peptide’s backbone and hydrophilic residues are located on the other side. In this design the hydrophobic and hydrophilic moities are located on the opposite sites of the peptide’s backbone that is different from the previously described linear configurations. For initial evaluation of this concept in formation of self-assembled nanostructures, we choose the tetrapeptide YXKX (Y = tyrosine, X = alkylated tyrosine, K = lysine), which contains alkylated tyrosine side chains in the second and the fourth positions (Figure 1). The major goal of this simulation study was to determine whether the alkylated hydrophobic amino acids can drive the peptide’s self-assembly and whether this hydrophobic/hydrophilic arragement of amino acids provide additional stabilization to the self-assembled structure. The same peptide sequence with all-natural amino acids was used for comparison.

Figure 1:

Tetrapeptide YXKX structure with alkylated tyrosine residues

All-atom molecular dynamics simulations were performed to observe the self-assembly of tetrapeptides with and without alkylated residues. The topology of the alkylated residues was determined by comparing them to known residues of the CHARMM36 force field.²⁷ We considered two systems with varying peptide counts (50 peptides and 100 peptides), to determine the effects of peptide concentration on self-assembly. The self-assembled nanostructures were analyzed according to several parameters, including the solvent-accessible surface area, the radius of gyration, free energy landscape, and radial density profile. The structure and dynamics of the tetrapeptides were determined from secondary structure analysis and hydrogen bonding analysis.

Methods

All-atom MD simulations were performed using the GROMACS 2018.3 package²⁸ with the CHARMM36 force field.²⁷ Tetrapeptide (Tyr-Tyr-Lys-Tyr) coordinates were constructed with Chimera software.²⁹ The secondary structure was defined as an antiparallel beta-strand with the respective phi-psi values (−139, 135). All atoms, including hydrogen atoms, were represented explicitly. For the new peptide design, two amino acids in the tetrapeptide were conjugated with alkyl chains. To obtain the symmetric structure with hydrophobic residues on one side and hydrophilic residues on the other side, the alkylated amino acids were placed in the alternative positions 2 and 4 in the tetrapeptide (Tyr-Tal-Lys-Tal) sequence (Figure 1). The alkylated tyrosine (O-dodecyl-tyrosine (Tal, X)) was formed by linking the hydrophobic alkyl chain (12 carbon length) with the hydroxyl group of tyrosine. The topology of the alkylated tyrosines were obtained by comparing them with known residues of the CHARMM36 force field (see supporting information for details).

In a cubic simulation box with dimensions of 10 × 10 × 10 nm³, 50 non-natural tetrapeptides (Tyr-Tal-Lys-Tal) were placed in random starting positions, resulting in a peptide concentration of 77 mg/mL (YXKX-low). To determine the effect of concentration on the self-assembly process, we placed 100 tetrapeptides (YXKX-high) in a 12 × 12 × 12 nm³ simulation box, which increased the concentration to 93 mg/mL. As a control, 50 natural tetrapeptides-YYKY50 (Tyr-Tyr-Lys-Tyr) were simulated in a cubic box of dimensions 9 × 9 × 9 nm³, resulting in a concentration of 72 mg/mL (similar to the non-natural tetrapeptide system). In general, the time scale for self-assembly simulations was around 1 μs. All-atom simulations for a larger system with longer time scales are computationally very expensive and time-consuming. Therefore, we simulated a smaller system with fewer peptides (i.e., 50 and 100). Both natural and non-natural tetrapeptide sequences were considered in their zwitterionic forms with a protonated N-terminal amino group (NH₃⁺) and deprotonated C-terminal carboxyl group (COO⁻). Tetrapeptides were immersed in a TIP3P water model³⁰, and the system was neutralized by adding chlorine ions.

Periodic conditions were applied in all directions. The steepest descent algorithm was used for energy minimization, with a 0.01 step size. The Particle-Mesh Ewald (PME) algorithm was used to calculate long-range electrostatic interactions with a cut-off of 1.2 nm. A Verlet cut-off scheme with a 1.2-nm cut-off distance was used for the short-range neighbor list. The cut-off for van der Waals interactions was shifted between 1.0 nm to 1.2 nm. The minimized system was equilibrated with position restraints in the constant volume ensemble (NVT) for 5,000,000 steps of 2 fs, and the temperature was maintained at 300 K using the v-rescale algorithm. This system was equilibrated further in the constant pressure ensemble for 5,000,000 steps of 2 fs, and the pressure was maintained around 1.0 bar, with a Berendsen weak coupling method. Bond lengths were constrained via the LINCS algorithm. With no restraints, 1300 ns production MD simulations were performed. During the production run, the v-rescale algorithm was used for temperature coupling, and a Parrinello-Rahman barostat was used to maintain pressure. Visual molecular dynamics (VMD) software was used to visualize the structures.³¹

The secondary structure of self-assembled peptides was analyzed with GROMACS built-in tools and in-house developed codes. Other programs, such as DSSP and STRIDE, predict the secondary structure of hydrogen bonding patterns and backbone geometry. However, they did not identify the terminal residues and considered them random coil.³² For our short peptide, the DSSP assignment failed to predict the secondary structure formation. Thus, we predicted the beta-sheet structure arrangement of the peptides using the following considerations⁸: (1) the distance between the main chains of two peptides should be less than 0.65 nm, with more than three contacts between the main chains; and (2) the angle between main chains of the two peptides should be less than 45° or greater than 135°. An angle between two peptides was calculated based on two vectors constructed from the N-terminal and C-terminal alpha carbon atoms of the respective peptides.

Results and discussion

Peptides were distributed randomly in the simulation box with water. Simulations were repeated five times for lower concentration and three times for higher concentration from different random initial starts (Table 1). In all cases, tetrapeptides first spontaneously aggregated into small clusters and then self-assembled into sphere-like nanostructures. Results of one such low concentration simulation (YXKX-low) showing the self-assembly process of tetrapeptides at various time points is presented in Figure 2 (See supporting information for the movie of YXKX-low self-assembly simulation).

Table 1:

Summary of simulation runs

Peptide concentration (mg/ml)	Number of MD runs	Self-assembled structure
77	5	Nanosphere
93	3	Nanosphere

Figure 2.

Self-assembly pathway of tetrapeptides (YXKX-low) during simulation (tyrosine, green; lysine, blue; alkylated tyrosine, grey). Water molecules are omitted for clarity.

Peptides initially aggregated into small clusters within the first 10 ns of the simulation. Peptide-peptide interactions were initiated by the strong hydrophobic effect arising from the alkyl sidechains. The small clusters then coalesced together and form ellipsoid like structure around 600 ns that slowly transformed into spherical structure at 800 ns. Simulations were further extended up to 1100 ns to confirm the structure formed was stable over the time. In the self-assembled nanostructure (Figure 2), hydrophobic alkyl tails (grey) were buried in the inner core to reduce the solvent exposure and the other hydrophilic residues lysine and tyrosine were arranged over the surface (blue and green).

To determine the morphology of the formed nanostructure, we calculated its moments of inertia (I) by aligning the nanostructure along its principal axes (Figure 3). Moment of inertia plot shows a steady convergence after 500 ns with a greatly reduced variance indicating that the nanostructure formed was stable. The moments of inertia calculated on the final structure along the x, y, and z axes confirmed that it was a nanosphere (I_x ≈ I_y ≈ I_z). The small surface area to volume ratio of the sphere allows the tight packing of hydrophobic amino acids inside the core region. By transforming into the nanosphere structure, more peptides can adopt an energetically favorable position in the aqueous solution by exposing less hydrophobic amino acids to water. This structural transformation stabilizes the moment of inertia that all three x, y and z components reach at the steady state once tetrapeptides form a sphere like structure.

Figure 3.

Moments of inertia with respect to simulation time of the self-assembled nanosphere

The solvent-accessible surface area (SASA) of the tetrapeptide system was computed using a probe radius of 1.4 Å (Figure 4). Within the first 10 ns simulation time, SASA decreases rapidly as small nanoclusters start to form and then gradually stabilizes around 500 ns as the system self-assembles into a nanosphere. After it remained in the same value with small fluctuations throughout the entire simulation. To understand the role of each residue in the process of self-assembly, we calculated the SASA of individual residues in the tetrapeptide (Figure 4). The rapid reduction in the SASA at 10 ns was primarily due to the Van der Waals interaction between the hydrophobic alkylated tyrosines (Tal2 and Tal4). In the tetrapeptide, lysine is the only hydrophilic polar residue. Tyrosine is an amphipathic residue with a polar hydroxyl group and an aromatic ring. The alkylated tyrosine has an unbranched hydrophobic alkyl chain that makes it more hydrophobic. Compared to the alkylated tyrosines, the first tyrosine has more hydrophilic character (green in Figure 4). At 500 ns SASA of the alkylated tyrosines is well reduced compared the more hydrophilic residues (i.e., Lys and Tyr). At this stage the hydrophobic alkylated tyrosines are well buried inside the nanosphere to reduce their solvent exposure that stabilizes the nanosphere structure.

Figure 4.

Time evolution of the solvent-accessible surface area (SASA) of the system YXKX-low and for the individual residues.

A two-dimensional free energy landscape (Figure 5) was constructed to identify the possible conformations adopted by the self-assembled peptides during the simulation. Gibbs free energy (S) was calculated from the probability distribution (P) of two reaction coordinates, radius of gyration (R_g) and SASA, using the equation S = − K_BT log P (R_g, SASA) where K_B is the Boltzmann constant and T is the temperature. Free energy surface shows minimum-energy basins populated by an intermediate structure with R_g=3.79 nm and SASA= 239 nm² that ultimately converges to a nanosphere with R_g = 2.2 nm and SASA = 199.92 nm² (Figure 5). The shape of free energy surface shows a more rapid initial reduction of SASA when compared to R_g. It is due to strong hydrophobic interactions arising from the alkyl chains which drive the tetrapeptide self assembly. This creates an intermediate minimum energy basin around 239 ns. As tetrapeptides starte arranging into a nanosphere structure, R_g is decreasing further untill reaching a steady state at a second free energy basin around 1000 ns.

Figure 5.

Two-dimensional free energy landscape of tetrapeptide self-assembly with reaction coordinates R_g and SASA.

The radial distribution function (RDF), calculated from the center of the mass of the nanosphere at 1100 ns is shown in Figure 6. The RDF curve of backbone atoms (Figure 6, magenta) at 2.4 nm serves as the reference point for separating the hydrophobic core (Tal2 and Tal4, grey) and hydrophilic corona (Tyr1, green, and Lys3, blue) regions of the nanosphere. The RDF curve of amphipathic residue Tyr1 overlaps with the backbone atoms. The hydrophilic Lys3 occupies the outer corona region and is the most exposed to the solvent. The alkylated tyrosines are tightly packed and buried inside the hydrophobic core (< 2.2 nm) that minimizes exposure to water.

Figure 6.

Radial distribution function (RDF) of hydrophobic residues Tal2 and Tal4 (grey), amphipathic Tyr1 (green) and hydrophilic Lys3 (blue) residues, tetrapeptide backbone (magenta).

The secondary structures of the self-assembled peptides were analyzed with a 2-D Ramachandran plot. Initially the phi-psi torsion angles of peptide backbones were assigned as −139, 135 corresponding to the anti-parallel beta structure. During the simulations, we did not impose any constraints on the backbone. Using the Ramachandran plot, we monitored the changes in the backbone dihedrals to determine their influence on the self-assembly. Figure 7 shows the backbone dihedrals of the middle residues Tal2 and Lys3 from the last frame of 1100 ns simulation; terminal residues are not included in the plot, as the phi angle of the N-terminal residue and the psi angle of the C-terminal residue were not identified. Though the phi-psi values of residues Tal2 and Lys3 varied from their initial values during the simulation, most of the residues remained in the beta-sheet region at the end of the 1100 ns simulation.

Figure 7.

Ramachandran plot of self-assembled tetrapeptides at 1100 ns representing the phi and psi angles of residues Tal2 (blue) and Lys3 (red). Note that the vast majority of the peptides clustered in the beta-sheet region.

As the secondary structure of tetrapeptides was stable and remained in the beta-sheet region (as beta strands), there is a possibility for beta-sheet structure formation with the neighboring peptides during self-assembly. To determine the beta-sheet structure formation, calculations were carried out as explained in the method section on the last frame of the 1100 ns simulation. Three sets of two-chain beta-sheet structures were identified out of 50 peptides, and the remaining beta strands were monomers. We also plotted the number of hydrogen bonds formed by the backbone atoms (N-H, C=O) between the peptides as a function of the simulation time (Figure 8). Our analysis shows that there are approximately 5 interstrand hydrogen bonds present in the self-assembled nanosphere. These interstrand hydrogen bonds formed by the backbone atoms connect the adjacent beta-strands and formed a beta-sheet structure. Though in the Ramachandran plot (Figure 7), most of the peptides (middle residues Tal2 and Lys3) are located in the extended beta-sheet region, only six peptides participate in the beta-sheet structure formation. All three, two chain beta-sheet structures formed were antiparallel, with interstrand hydrogen bonds established between the C=O and N-H backbone atoms. In some cases, we also observed interpeptide head (NH₃⁺) to tail (COO⁻) hydrogen bonds in addition to the backbone hydrogen bonds.

Figure 8.

Hydrogen bond evolution between the backbone atoms of tetrapeptides.

Self-assembling peptides can adapt different structural shapes under varying concentrations. For example, diphenylalanine peptides⁷ formed vesicle-like structures at low concentrations (toroid, ellipsoid, and discoid) while at higher concentrations they first self-assembled into bilayers and then turned into spherical or ellipsoid vesicles through bilayer closure. Experimental studies of cationic dipeptides showed that at high concentrations, they formed a nanotube structure and then spontaneously converted into vesicle-like structures at lower concentrations.³³ In the case of triphenylalanine (900 peptides), both nanosphere and nanorod structures were observed at the same concentration (120 mg/mL).⁸ Though the peptide-peptide and peptide-water interactions initiate self-assembly, the morphologic features of self-assembled peptides depend on their concentration as well.

We increased the peptide concentration to 93 mg/ml to study the effect of concentration on the self-assembly. As in the case of the lower concentration (i.e., YXKX-low), tetrapeptides initially aggregated into small clusters and then started to fuse together to form an ordered structure. Around 700 ns two ellipsoid structures formed and then slowly transformed into two sphere like structure at 1000 ns. We further continued the simulations up to 1300 ns but the size or shape of the nanospheres did not undergo any changes (See supporting information for the movie of YXKX-High self-assembly simulation). As seen in the earlier reports^7–8, we did not observe any concentration dependent structural changes in our self-assembled structures. We also simulated a self-assembly of natural tetrapeptides (YYKY50) with the same system size and concentration (72 mg/mL) as a control. They aggregated into clusters and were not able to self-assemble into an ordered nanostructure (Figure S1). It confirms that the alkylated tyrosines are the main driving force for the YXKX tetrapeptide self-assmebly.

Alkylated resiudes in the tetrapeptide bring out more detergent like character^34–37. Detergent monomers are able to form globular shapes like spheres, ellipsoids, cylinders depending upon the alkyl chain length and head group (phosphocholines, glucosides, maltosides, lysophosphatidyl glycerols)³⁶. The hydrophobic core and hydrophilic shell structure of nanosphere formed by the modified tetrapeptides is very similar to the detergent. The additional characteristics arising from the peptides like backbone hydrogen bonding, beta sheet structure strengthens the self-assembled structure. The number of tetrapeptides in the nanosphere varied between 35–57 from the different trials of low and high concentration simulations. To form a stable self-assembled structure with hydrophobic core and hydrophilic surface, the steric repulsion between the tetrapeptides should be minimum. As the inner core of the nanosphere is primarily occupied by the alkylated tyrosines, packing density of the tetrapeptides is limited by the steric repulsion between the alkyl chains. RDF analysis showed that more tetrapeptides are accommodated within nanospheres when the alkyl chains are fully stretched. Alkyl chain lengths of 2.0 nm and 2.2 nm were associated with nanospheres containing 35 and 57 tetrapeptides, respectively; note that the corresponding radii of nanosphere cores were the same as the alkyl chain lengths. This result confirms that packing of alkyl chains inside the core region determines the size and the shape of the self-assembled nanostructure.

Conclusions

We proposed a new design for self-assembling peptides consisting of two distinct sides relative to the peptide’s backbone - hydrophobic alkylated amino acids on one side and hydrophilic amino acids on the other side. We used YXKX tetrapeptide as a case study to investigate the role of alkylated residues in the self-assembly process using all-atom MD simulations. The alkyl tails attached to the aromatic tyrosines, initiated the aggregation process and the tetrapeptides self-assemble into densely packed nanospheres with a hydrophobic inner core. The other two amino acids, lysine and tyrosine, located on the surface of the self-assembled nanospheres. By forming a hydrogen-bonding network with the adjacent beta-strands, the tetrapeptides stabilized themselves in a beta-sheet structure.

The proposed peptide design exhibits the characteristics of a self-assembling peptide as well as a detergent nature from the alkylated residues. The size and the shape of the tetrapeptide nanosphere is constrained by the alkyl chain length and by the steric repulsion between the alkyl chains within the hydrophobic core. In addition to the formation of self-assembled nanostructures, the proposed peptide design could be used for solubilization and functionalization of hydrophobic nanoparticles such as oleic acid coated gold³⁸ or iron oxide.³⁹ In this case, the alkyl chains would interact with the surface of hydrophobic nanoparticles, and the peptide’s hydrophilic portion would provide functional groups for surface modifications. Further, these peptides could also be used for the solubilization of poorly soluble drugs for efficient in vivo administration.⁴⁰

Supplementary Material

Figure S1. Natural tetrapeptides (YYKY) in random positions at 0 ns and after 800 ns simulations. A few clusters formed at the end of 800 ns simulation, but there was no definite structure formation. (tyrosine, green; lysine, blue)

Figure S1. Natural tetrapeptides (YYKY) in random positions at 0 ns and after 800 ns simulations. A few clusters formed at the end of 800 ns simulation, but there was no definite structure formation. (tyrosine, green; lysine, blue)