Impact of γ-Amino Acid Residue Preorganization on α/γ-Peptide Foldamer Helicity in Aqueous Solution

Abstract

α/γ-Peptide foldamers containing either γ⁴-amino acid residues or ring-constrained γ-amino acid residues have been reported to adopt 12-helical secondary structure in nonpolar solvents and in the solid state. These observations have engendered speculation that the seemingly flexible γ⁴ residues have a high intrinsic helical propensity, and that residue-based preorganization may not significantly stabilize the 12-helical conformation. However, the prior studies were conducted in environments that favor intramolecular H-bond formation. In this report, we use 2D-NMR to compare the ability of γ⁴ residues and cyclic γ residues to support 12-helix formation in more challenging environments, methanol and water. Both γ residue types support 12-helical folding in methanol, but only the cyclically constrained γ residues promote helicity in water. These results demonstrate the importance of residue-based preorganization strategies for achieving stable folding among short foldamers in aqueous solution.

TOC image

The study of unnatural oligomers that display biopolymer-like folding behavior offers a framework for interrogating relationships between covalent and noncovalent structure (constitution and conformation) in a way that transcends the deep understanding that has emerged from analysis of proteins and nucleic acids. For example, examination of “foldamers” that contain β-amino acid residues instead of or in addition to α residues (i.e., β-peptides or α/β-peptides) has revealed the strong influence of small-ring constraints on the identity and stability of H-bond-mediated secondary structure.¹ This issue is not pertinent to the folding of conventional polypeptides, containing exclusively α residues, because imposition of a cyclic constraint necessarily abolishes backbone H-bond potential and disrupts common secondary structures, as seen with proline residues.²

Insights gained from experimental correlations among residue identity, secondary structure and conformational stability have proven invaluable in the development of functional foldamers containing β residues. Stabilization of the β-peptide 14-helix by trans-cyclohexyl residues, for example, enabled development of a foldamer catalyst³ and fundamental single-molecule studies of hydrophobic interactions.⁴ Stabilization of α/β-peptide helices by trans-cyclopentyl residues has allowed the refinement of protein-protein interaction antagonists.⁵ These functional outcomes with β-containing foldamers highlights the importance of elucidating relationships between residue structure and conformational stability to other foldamer classes. Here we evaluate the impact of a specific cyclic constraint on helical secondary structure formed by α/γ-peptides.

Peptidic foldamers containing γ residues have been examined in many laboratories,⁶ but only a few studies⁷ have evaluated the relationships between γ residue structure and conformational stability. The α/γ-peptide 12-helix is probably the most extensively studied secondary structure in this foldamer class. This conformation, formed by oligomers with sequentially alternating α and γ residues, is defined by 12-atom C=O(i) → H−N(i+3) H-bonds between backbone amide groups. The α/γ-peptide 12-helix was initially observed by Balaram and coworkers in crystal structures of achiral oligomers containing gabapentin (Gpn) and α-aminoisobutyric acid (Aib) residues.^8,9 Guo et al. developed an efficient asymmetric synthesis leading to trisubstituted γ-amino acid residue I (Figure 1), and showed that α/γ-peptides containing I form the 12-helix in chloroform (based on NOE analysis) and in the crystalline state.¹⁰ The cyclic constraint and substitution pattern of I should limit conformational freedom. In particular, the Cα-Cβ and Cβ-Cγ bonds are predicted to favor a g⁺,g⁺ torsion angle sequence, as required for the 12-helix; this prediction is consistent with the many crystal structures of foldamers containing residue I.¹¹

Figure 1.

Structures of γ residues.

The groups of Gopi¹² and Balaram¹³ have reported a large set of crystal structures showing that α/γ-peptides containing exclusively γ⁴ residues (Figure 1) (4-mers to 16-mers) can adopt the 12-helix. In these structures the Cα-Cβ and Cβ-Cγ torsion angles favor a g⁺,g⁺ torsion sequence similar to that observed in 12-helices containing residue I. Balaram and coworkers further demonstrated via 2D NMR that 12-helicity is maintained by a 16-mer α/γ-peptide in chloroform.^13b These findings are striking because one would predict γ⁴ residues to be much more flexible than is γ residue I. As Balaram and coworkers noted, the extensive structural results with α/γ⁴-peptides lead one to question whether the constraint inherent in I stabilizes the 12-helical conformation. This uncertainty must be resolved because constrained residues generally require greater synthetic effort than do flexible residues.

To address this question, we moved to a new family of α/γ-peptide octamers (1-4; Figure 2) tailored for aqueous solubility. Water is the least favorable among common solvents for H-bonded conformations such as the α-helix or β-sheets formed by conventional peptides or helices formed by β- or α/β-peptide foldamers.¹⁴ Presumably the strong H-bond competition provided by water molecules diminishes the ability of intramolecular H-bonds to stabilize secondary structure. Admixture of an alcohol cosolvent, such as methanol, enhances the population of internally H-bonded secondary structures for which there is an intrinsic propensity,^14–15 but aqueous-alcohol mixtures do not guarantee folding. In contrast, chloroform allows intramolecular H-bonds to serve as a strong driving force for secondary structure formation. Qualitative comparisons based on data obtained in chloroform or crystal structures cannot decisively address the relative 12-helical propensities of I vs. a γ⁴ residue.

Figure 2.

Structures of water-soluble α/γ-peptides.

α/γ-Peptide series 1-4 features both γ⁴ residues and γ residue II (Figure 1), which is an analogue of I bearing a side chain that should be cationic at acidic pH. α/γ-Peptide 1 contains exclusively γ⁴ residues, which are progressively replaced with II, from N- to C-terminus, in analogues 2-4. The γ residue side chain complement varies at one position across the α/γ-peptide series: 1 and 2 have a glutamate-like side chain at residue 4, but 3 and 4 have a lysine-like side chain at this position. This variation was intended to promote ¹H NMR resonance dispersion. The four α residues are invariant among 1-4. Two aromatic side chains (Tyr-5 and Phe-7) were included to enhance dispersion of ¹H NMR signals.

Initial 2D NMR studies were conducted in CD₃OH, in which 1-4 were highly soluble. ¹H resonances were unchanged between 0.1 and 3 mM, suggesting that aggregation state, which we assume to be monomeric, does not vary in this range. In general, more constrained γ residues correlated with improved chemical shift dispersion. In order to detect 12-helical folding, we focused on three types of backbone-to-backbone i,i+2 NOE that are known to be characteristic of this secondary structure (Table 1).^{10, 13b}

Table 1.

Summary of crosspeaks observed in CD₃OH.^a

H atom pair	12-Helix distance (Å)b	NOE pairs	1	2	3	4
a	2.72	2 ⋯ 4	W	S	?	S
	2.70	4 ⋯ 6	?	M	S	S
	2.37	6 ⋯ 8	S	M	M	M
b	2.60c	2 ⋯ 4	W	S	S	?
	2.69c	4 ⋯ 6	?	W	M	S
	–	6 ⋯ 8	?	?	W	S
c	3.42	1 ⋯ 3	W	W	W	W
	3.64	3 ⋯ 5	VW	N.D.	W	W
	3.40	5 ⋯ 7	VW	VW	W	W

^aVW=very weak; W=weak; M=medium; S=strong; ?=ambiguous; N.D.=not detected.

^bFrom crystal structure of P3 in ref. 11c.

^cDistance to pseudoatom between diastereotopic H nuclei.

For fully flexible α/γ-peptide 1 in CD₃OH, partial chemical shift overlap precluded precise integration of several expected 12-helical crosspeaks in the 2D-ROESY spectrum. Nevertheless, unambiguous observation of six characteristic NOEs, uniformly distributed across 1, supported the conclusion that this α/γ-peptide is at least partially 12-helical in methanol (Table 1).

Improved ¹H signal dispersion enabled unambiguous assignment of almost all expected 12-helix NOEs for α/γ-peptide 2 in CD₃OH. The two 12-helical NOEs involving the cyclic γ residue (types a and b in Table 1) were more intense than those involving γ⁴ residues. The 12-helical NOEs involving the α residues (type c) were generally not very intense; consistent with this trend, protons that would give rise to type c NOEs are further apart in available 12-helix crystal structures than are protons that would give rise to type a or b NOEs.^12c α/γ-Peptide 3, containing two cyclic γ residues, showed excellent chemical shift dispersion, and numerous 12-helical NOEs were detected. As seen for 2, crosspeak intensity was lower for γ-to-γ NOEs involving exclusively γ⁴ residues relative to such NOEs involving at least one cyclic γ residue.

α/γ-Peptide 4, containing three cyclic γ residues, showed the most intense 12-helical crosspeaks among the four oligomers. Furthermore, numerous long-range NOEs between protons on side chains were detected that were not observed for 1-3. Collectively, the ROESY spectra for 1-4 in CD₃OH demonstrate a tendency of all α/γ-peptides to adopt the 12-helix conformation, but the trend of increasing crosspeak intensity and number with increasing number of γ⁴→II replacements suggests that the helical state becomes more stable as cyclic constraints are added.

We next characterized the folding behavior of α/γ-peptides 1-4 in water. All four α/γ-peptides where highly soluble (>3 mM), and chemical shifts did not vary between 0.1 and 3 mM. The data obtained in this competitive environment support our conclusion that the cyclic constraint of residue II stabilizes the 12-helix.

For the fully flexible α/γ-peptide 1, only two very weak i,i+2 ROESY crosspeaks consistent with the 12-helix conformation could be distinguished, with most others impossible to assign unambiguously because of poor resonance dispersion. For all ambiguous i,i+2 crosspeaks, the low intensity outside the overlapping region means that, at most, these NOEs are very weak. Three of the anticipated 12-helical i,i+2 crosspeaks would have occurred in relatively uncrowded regions of the 2D-ROESY spectrum but were not detected, which suggests that any tendency toward 12-helical folding by α/γ-peptide 1 is low in water. Three of the five non-sequential crosspeaks that could be unambiguously assigned were inconsistent with the 12-helix conformation (Figure 3). The i,i+2 NOEs of 1 in water fail to converge to a single mean structure and thus reflect a disordered conformational ensemble.¹⁶

Figure 3.

Summary of detected ROESY crosspeaks between protons on non-adjacent residues in aqueous solution.

The aqueous 2D-ROESY spectra of α/γ-peptides 2-4 traced the formation of helicity with increasing cyclic constraint content. As seen in CD₃OH, the ¹H resonances become more dispersed with increasing content of cyclic γ residue II, permitting unambiguous detection of an increasing fraction of the expected 12-helical ROESY crosspeaks as the content of II grows. All observed i,i+2 ROESY crosspeaks for 2-4 involve at least one residue II or occur between protons from α residues that are separated by a residue II. α/γ-Peptide 2, with a single residue II, illustrates the difference between cyclic and acyclic γ residues in terms of supporting local folding (Figure 4). The crosspeak of type a between II-2 and γ⁴Glu-4 is very intense, whereas the possible a crosspeak between γ⁴Glu-4 and γ⁴Lys-6 is not detected, even though this crosspeak would have occurred in an uncrowded region of the 2D spectrum. This trend continues for α/γ-peptides 3 and 4.

Figure 4.

Excerpt of the 600 MHz ROESY spectrum of 3 mM α/γ-peptide 2 in 9:1 H₂O:D₂O, 100 mM acetate, pH 3.8 at 5°C. i,i+2 NOEs indicated in red. Missing type a NOE shown with dashed line; observed a NOE shown with unbroken line.

Results of NOE distance-restrained simulated annealing calculations¹⁷ suggest an increasingly well-defined helical conformation in methanol as γ⁴ residues are replaced with constrained residues (Figure 5a). The calculations in water suggest an even more pronounced disorder-to-order transition across the series 1-4 than is evident in methanol (Figure 5b vs. Figure 5a). The non-12-helical NOEs of 1 in water result in a variety of simulated conformations that do not correspond to a regular helix. For α/γ-peptides 2 and 3 in water, partial ordering only at the segments containing cyclic γ residues is observed, which is expected since there was a consistent lack of characteristic i,i+2 ROESY crosspeaks for the segments containing γ⁴ residues. α/γ-Peptide 4 in water displays a relatively tight cluster of structures, each with the maximum number of 12-helix H-bonds. Increasing the number of γ⁴→II replacements across the series 1-4 improves the overlay of the NMR-derived structures with 12-helical conformations observed for crystalline α/γ-peptides, whether the oligomer crystallized contains constrained γ residue I or unconstrained γ⁴ residues; this trend is illustrated for a specific α/γ⁴-peptide crystal structure in Figure 6.

Figure 5.

Superposition of ten lowest-energy of 100 trial structures from simulated annealing calculations of 1-4. Mean pairwise backbone RMSD is shown below each structure. (a) Calculated ensembles using distance restraints derived from CD₃OH ROESY data. (b) Calculated ensembles using distance restraints derived from aqueous ROESY data.

Figure 6.

Comparison of NMR structures of α/γ-peptides 1-4 with a 12-helix crystal structure from ref. 11c. Black dots are backbone RMSDs of the 10 lowest-energy NMR structures. Mean pairwise RMSDs of each set of NMR structures vs. the crystal structure are shown with red bars.

The 2D-NMR data presented here for α/γ-peptides 1-4, as well as other ¹H-NMR data,¹⁸ support the hypothesis that the cyclic constraint in II stabilizes the 12-helix. γ⁴ Residues, intrinsically more flexible than is II, have a modest propensity for 12-helix formation, as indicated by previous reports^12–13 and by our NMR results in methanol. However, analysis of 1-4 in aqueous solution reveals a clear distinction between 12-helical propensities of II and γ⁴ residues, which decisively addresses questions raised by recent studies.^12–13 These results are significant since the impact of γ residue constraint on foldamer secondary structure has received very little scrutiny, in part because there has been very little previous study of γ-containing foldamers in aqueous solution.^{11c, 19} Our demonstration that the preorganization inherent in γ residue II favors a specific secondary structure should encourage fundamental research on new foldamer building blocks that favor adoption of discrete and diverse conformations.