Impact of Hydrophobic Sequence Patterning on the Coil-to-Globule Transition of Protein-like Polymers

Understanding the driving forces for the collapse of a polymer chain from a random coil to a globule would be invaluable in enabling scientists to predict the folding of polypeptide sequences into defined tertiary structures. The HP model considers hydrophobic collapse to be the major driving force for protein folding. However, due to the inherent presence of chirality and hydrogen bonding in polypeptides, it has been difficult to experimentally test the ability of hydrophobic forces to independently drive structural transitions. In this work, we use polypeptoids, which lack backbone hydrogen bonding and chirality, to probe the exclusive effect of hydrophobicity on the coil-to-globule collapse. Two sequences containing the same composition of only hydrophobic “H” N-methylglycine and polar “P” N-(2-carboxyethyl)glycine monomers are shown to have very different globule collapse behaviors due only to the difference in their monomer sequence. As compared to a repeating sequence with an even distribution of H and P monomers, a designed protein-like sequence collapses into a more compact globule in aqueous solution as evidenced by small-angle X-ray scattering, dynamic light scattering, and probing with environmentally sensitive fluorophores. The free energy change for the coil-to-globule transition was determined by equilibrium denaturant titration with acetonitrile. Using a two-state model, the protein-like sequence is shown to have a much greater driving force for globule formation, as well as a higher m value, indicating increased cooperativity for the collapse transition. This difference in globule collapse behavior validates the ability of the HP model to describe structural transitions based solely on hydrophobic forces. ■ INTRODUCTION Protein folding is an inherently complex process involving a multiplicity of forces and interactions. Predicting a tertiary structure from a polypeptide sequence has presented a longstanding challenge to the scientific community. Dill and others have postulated that hydrophobicity serves as one of the most important driving forces for protein assembly.1−3 To this end, they have developed a computational framework known as the HP model in which only two types of monomers, hydrophobic (H) and polar (P), are considered. With this dramatically reduced set of interactions, polymer sequences can be computationally designed to fold into defined structures. A key experimental realization of this theory has been work by Hecht’s group to analyze the HP sequence patterns in combinatorial variants of existing proteins. For example, a sequence known to form a 4-helix bundle was randomized while still maintaining the same pattern of H and P residues. Overwhelmingly, the mutated proteins still formed 4-helix bundles, demonstrating that the particular side chain was not as Received: April 5, 2012 Revised: May 28, 2012 Published: June 8, 2012 Article pubs.acs.org/Macromolecules © 2012 American Chemical Society 5229 dx.doi.org/10.1021/ma300707t | Macromolecules 2012, 45, 5229−5236 important as the hydrophobic or polar character of the amino acid. Given that hydrophobicity has been shown to have such a strong effect on protein folding, efforts have focused on ways to design polypeptide sequences that include targeted hydrophobic and polar regions to induce folding.5−9 However, it is difficult to isolate the hydrophobic interactions, making de novo protein design a complex process in which the effect of different forces can be convoluted. Scientists have turned to simpler transitions such as globule formation to try and understand the forces that influence molecular assembly. Globule formation has been described as one of the first steps along the path to a folded protein, and hydrophobic interactions have been shown to have a large impact on this transition. In fact, theoretical works have shown that for many polypeptide sequences, the likelihood of the chain forming a compact globule is relatively high and that the easiest way to disrupt compact globule formation is to exchange polar and nonpolar residues. These works demonstrate that the position of hydrophobic and polar residues is very important for structure formation and folding. Many groups have used theoretical means to postulate how a monomer sequence in the HP model affects globule formation, including the effect of the relative fraction of hydrophobic monomers to polar monomers, the degree of hydrophobicity of the hydrophobic monomers, the overall length of the chain, the importance of large contiguous sections of H or P monomers, and even the ability of ionic interactions to impact the collapse process. One of the most extensive efforts has been the theoretical work performed by Khokhlov and Khalatur (KK), demonstrating that copolymers with blocky (or protein-like) distributions of monomers form more stable globules than corresponding random sequences where the monomers are more evenly distributed throughout the chain. They also predicted that the coil-to-globule transitions for the protein-like molecules were sharper and occurred at a higher temperature than those for the random sequences. Both results showed a clear difference in collapse behavior based solely on the sequence of the molecule. Experimental efforts in this area have lagged behind computational efforts21−24 due to the synthetic challenges inherent in generating precise sequence-specific polymers in quantity. The use of polypeptides would seem obvious, but due to their inherent chirality and backbone hydrogen bonding, it is difficult to directly study the impact of hydrophobic interactions. Alternatively, in most synthetic polymer systems where it might be possible to create a system with isolated HP interactions, it is nearly impossible to obtain the level of sequence control necessary to probe the effect of H and P monomer sequence on the coil-to-globule transition. Genzer and co-workers elegantly demonstrated the chemical labeling of surface exposed monomers in a collapsed polystyrene globule, generating protein-like sequences. However, assessment of exact chain sequence information was not possible. Given the difficulties mentioned above, polypeptoids or poly(N-substituted glycine)s are an ideal model system for understanding sequence effects on the hydrophobic collapse of polymers chains. Their stepwise submonomer synthesis is efficient, with 99% or greater conversion for most monomer additions and provides sequence control, allowing for the creation of sequence-specific chains in the 50 monomer range with excellent precision. In addition to their synthetic tractability, polypeptoids are known to possess flexible backbones and self-assemble into protein-like structures in aqueous solution,27−32 making them ideal candidates for studying fundamental self-assembly in the form of coil-toglobule transitions. The peptoid backbone is nearly identical to that of polypeptides, making chain measurements biologically relevant. Finally, and perhaps most importantly, polypeptoids also lack chirality and the ability to form intrachain hydrogen bonds (due to the absence of any backbone hydrogen bond donors), allowing the study of hydrophobic forces in isolation. This makes polypeptoids an excellent model system for experimentally validating the HP model. In this work, we designed and synthesized two polypeptoid 100mer sequences containing only the H and P monomers in order to probe their coil-to-globule transitions (Figure 1). One of the sequences was designed using the KK method to be a “protein-like” sequence that contained longer stretches of both the hydrophobic and polar monomers. The other “repeating” sequence contained the shortest possible stretches of both monomers. Growing interest in intrinsically disordered proteins including those with repeating and polar sequences has provided both experimental and theoretical results showing that they also undergo coil-to-globule transitions. Small-angle X-ray scattering (SAXS), dynamic light scattering (DLS), and environmentally sensitive fluorescence probe measurements were used to show that the protein-like sequence collapses into a tighter globule in aqueous solution than that formed by the repeating sequence. In addition, the transition of the protein-like molecule from the globule state to the coil state was shown to exhibit a significantly higher unfolding free energy and that the transition is more cooperative than that of the repeating sequence molecule, Figure 1. Protein-like and repeating sequence polypeptoid 100mers. The polypeptoids were synthesized by clicking two HPLC-purified 50mers together. Each monomer is represented by a circle where the red circles are the hydrophilic and polar N-(2-carboxyethyl)glycine (P) monomer while the blue circles are hydrophobic N-methylglycine (H) monomers. The protein-like sequence contains block sections of each type of monomer, while the repeating sequence has an even distribution of monomers. Both molecules have an identical composition of exactly 80 hydrophobic monomers and 20 hydrophilic monomers and a molecular weight of 8517 g/mol. Macromolecules Article dx.doi.org/10.1021/ma300707t | Macromolecules 2012, 45, 5229−5236 5230 indicating the importance of sequence on the folding behavior of a biomimetic molecule. ■ EXPERIMENTAL METHODS System and Sequence Design. The protein-like sequence generation was carried out according to methods previously published. The length and specific side chains were designed specifically for this synthetic system. The H and P monomer structures and their relative fractions were chosen to provide enough hydrophobicity to form a hydrophobic core and enough hydrophilicity to maintain solubility in aqueous solution. A sequence design procedure aimed at obtaining the protein-like polypeptide chain containing 20% aspartic acid residues denoted as “P” and 80% alanine residues denoted as “H” was carried out. The resulting polypeptide chain was then translated into a polypeptoid chain by using an Nmethyl side chain for the H residue and an N-2-carboxyethyl side chain as the P residue. First, a homopolymer globule from 100 H units was generated, using Accelrys Discovery Studio 2.5 and all-atom molecular dynamics with the AMBER96 force field. To construct a target protein-like H/P sequence, we perform globule surface “coloring”. By “coloring” we mean the change of a given monomer type: monomer units in the center of the globule are assigned to be H-type units, while monomer units belonging to globular surface are assigned to be P-type units. The “coloring” is applied to the units mostly exposed to water at the surface of the globule. After the formation of initial H−P sequence, we allow the macromolecule to undergo a coil-to-globule transition, and then we “recolor” a refolded globule. Each globular structure is obtained during 1 ns simulated annealing run (we start from 1000 K and then cool the system to 300 K). After that the system is again relaxed during another 1 ns run. In this way, we obtain a heteropolymer chain with a new primary sequence and all steps described above are performed again for the chain with a new sequence. After several attempts, we reach the regime when practically all of P units remain robustly located at the globular surface even after refolding, while the globular core is composed mostly of H units. This process is detailed in Figure 2. As a control sequence, a repeating sequence of the same monomer composition (80% H and 20% P) was chosen so as to minimize the length of any continuous region of H or P monomers. Thus, the repeating sequence consists of a simple pentamer repeat of (PHHHH), limiting the longest stretch of hydrophobic monomers to four residues. Synthesis and Conjugation. Polypeptoids were synthesized on a commercial robotic synthesizer Aapptec Apex 396 on 100 mg of Rink amide polystyrene resin (0.6 mmol/g, Novabiochem, San Diego, CA) using the procedure previously detailed. All primary amine monomers, solvents, and reagents were purchased from commercial sources and used without further purification. The β-alanine-OtBu·HCl was purchased from ChemImpex and used after freebasing by extraction from dichloromethane (DCM) and basic water. The resulting compound was confirmed by 1H NMR. The β-alanine submonomer was dissolved in N-methylpyrrolidinone at a concentration of 1.5 M, while the methylamine was used directly as purchased as a 40% w/v solution in water. All displacement times were 60 min for the first 15 monomer addition cycles, 90 min for the next 15 and 120 min for the remainder. All polypeptoids were acetylated on the resin and purified using reverse phase HPLC as previously described. The mass and purity were confirmed using reverse phase analytical HPLC and MALDI and representative traces and spectra are shown in Table 1 and Figure 3. Upon the synthesis of the 50mer components, the two molecules were then clicked together using an alkyne−azide reaction. An alkyne group was added to the N-terminus of the 50mer polypeptoid still attached to the polystyrene resin by the addition of a propargylamine submonomer in a 51st monomer addition cycle. For the other component, an azide group was added by first bromacylating the polymer while it was still attached to the polystyrene resin and then substituting for the bromine using sodium azide. The click reaction (Scheme 1) was performed by reacting the propargylated compound A (23 mM) with 2 equiv of azide-modified compound B in 200 μL of water. Copper(II) sulfate (5 equiv) and ascorbic acid (6 equiv) were added, and the solution was mixed at 70 °C for 24 h. The excess copper was removed by stirring over basic alumina for 2 h. The resulting compound was purified by a 0−30% acetonitrile gradient on reverse phase HPLC. Representative analytical HPLC and MALDI analysis of the 50mer and 100mer peptoids are shown in Figure 3. Self-Assembly Solutions. In order to probe aqueous assembly, the molecules were dissolved at 1 mg/mL (0.12 mM) in 25 mM Tris HCl pH 8 buffer. The solutions were sonicated for ∼30 s to ensure dissolution of the molecules and were then annealed at 70 °C for 4 h. Two-State Model. The equation below was fit to the Rg measurements with Rg,coil and Rg,globule calculated by averaging the three points at either end of the curve. The percentage of acetontrile in the solution is