Computational Design of a Self-Assembling β-Peptide Oligomer

Computational Design

Design of coiled-coils has played an important role in development of protein design and computational algorithms. Similarly, here we seek to design homo-oligomeric β-peptide coiled-coils. We used a variation of the Crick formula²⁹ to transform ideal β-3₁₄ helices into radially symmetric parallel coiled-coils with different core residues (Figure 1).

The packing of poly-β³ homoAlanine (hA), poly-β³ homoValine (hV), and poly-β³ homoIsoleucine (hI) was examined. These sequences were modeled into different stoichiometries. The hA and hV sequences showed particularly good packing. The interhelical Lennard-Jones³⁰ and electrostatic energies on a per helix basis were systematically evaluated as a function of a bundle radius, helix phase, and supercoil pitch (Table S5, Supporting Information). The results were ranked by normalized interhelical interaction energy (energy per helix relative to the minimal energy computed for the dimer) (Figure 2, Table S6, Supporting Information). hV showed a particularly strong minimum at n = 6, associated with tight packing of every third hV side chain in the core. This finding suggested that β-peptides with hV in the core could show good geometric specificity for this hexameric association state.

To complete the design of a water-soluble bundle, the optimized hydrophobic hV core was fixed, while the identity and conformations of exterior residues were allowed to vary. We introduced water-solubilizing residues and stabilizing interhelical salt bridges at these positions. While it is possible to stabilize the 3₁₄ helix through electrostatic interactions between residues one turn apart in the helix, such intrahelical interactions would not provide specificity for the desired tertiary structure versus other possible associations. Thus, to specifically stabilize the desired bundle rather than just the secondary structure, neutral residues were alternated with charged residues along a given face of the helices (Figure 3). Specifically, charged and neutral residues alternate at positions i, i + 3, i + 6, etc. in the sequence. A variety of sequence patterns consistent with this restraint were examined (Table S7, Figure S4, Supporting Information).

Experimental Characterization

The final sequence (hKhKhVhKhE-hVhFhFhVhK-hEhVhFhFhV-hKhEhVhYhK), nicknamed beta Vhex, contains C- and N-terminal β³-homoLysine (hK) residues to improve solubility. To examine the role of the hydrophobic hV residues, we have also prepared a sequence with hA in the core: betaAhex (hKh- KhAhKhE-hAhFhFhAhK-hEhAhFhFhA-hKhEhAhYhK). Furthermore, we have synthesized a corresponding α-peptide, alpha Vhex (KKVKE-VFFVK-EVFFV-KEVYK). Circular dichroism (CD) spectroscopy data show that the designed β-peptide beta Vhex adopts a 3₁₄-helical conformation in a concentration-dependent manner as expected for a self-associating system in which secondary structure formation and association are thermodynamically linked. At high peptide concentrations (more than ca. 360 µM, Figure 4) the spectrum of beta Vhex shows a minimum at 211 nm, consistent with that reported for β-3₁₄ helices.^31–36 The concentration dependence of the MRE at 211 nm conforms well to a highly cooperative monomer–hexamer equilibrium with the free energy of association of 27.8 kcal/mol·hexamer (1 M standard state) (Figure 4). By contrast, the negative controls betaAhex as well as alpha Vhex displayed CD spectra more typical of unordered polypeptides over the same concentration range(Figures S5 and S6, Supporting Information).

The hexameric oligomerization state of beta Vhex was confirmed by analytical ultracentrifugation (AUC). Equilibrium sedimentation data were well described by a monomer–hexamer equilibrium model. The free energy of hexamerization of 26.5 kcal/mol·hexamer obtained from the fit (Figure S7, Supporting Information) is in very good agreement with the CD data (27.8 kcal/mol·hexamer). Attempts to fit the data to other monomer-n-mer models (n = 4, 5, 7, 8) resulted in a poorer fit, confirming the hexameric state and the cooperativity of the assembly.

The thermal stability of the designed β³-peptide was monitored using CD spectroscopy. A sigmoidal melting curve was observed indicative of a cooperative unfolding process. The designed bundle demonstrated an enthalpic unfolding transition, similar to other recently published β³-peptides¹² (Figures S8 and S9, Supporting Information). The T_m (>96 °C) indicates a highly stable, well-folded β-peptide when compared to α -peptides (ubiquitin’s T_m = 90 °C).

The conformation of beta Vhex was also investigated by NMR. The 1D ¹H NMR spectrum of beta Vhex shows (Figure S10, Supporting Information) peptide amide resonances well dispersed over the range of ca. 1.5 ppm, suggesting formation of a distinct quarternary structure that provides a differentiated electronic environment, similarly to previously reported β-peptide bundles.^12,37

In summary, we have extended computational methods originally developed for α-peptides and proteins to β-peptides. Interestingly, while conformational search algorithms for α-peptides with core valines showed a strong preference for a trimeric association,^38,39 3₁₄ helical β-peptides showed optimal packing with six monomers. This reflects differences in shape and axial cross-sectional area between α- and β-peptides.

To test this computational result, we designed peptide beta Vhex, which was indeed found to form a hexameric coiled-coil structure. The difference in the secondary structure has a large influence on the preferred association state for α- vs β-peptides. A tendency for higher-order association was also observed by Schepartz et al. who synthesized β-peptides with hydrophobic sequence repeats patterned after those seen in 2- and 3-stranded α-helical coiled-coils.¹² In related studies, Gellman et al.⁹ have systematically introduced βAAs into α-peptide coiled-coils. Two interesting features were observed in this work. First, βAA residues could be inserted into an α-helix with minimal perturbation of its geometric parameters, and second, the packing patterns observed for the βAA core are quite different from those observed in α-helical coiled-coil quarternary structures. Therefore, we expect that utilizing intrinsic properties of βAA with this computational platform will pave a way to novel functional β-peptide-based molecules.

Supplementary Material