Tunable control of polyproline helix (PPII) structure via aromatic electronic effects: an electronic switch of polyproline helix.

Aromatic rings exhibit defined interactions via the unique aromatic π face. Aromatic amino acids interact favorably with proline residues via both the hydrophobic effect and aromatic-proline interactions, C-H/π interactions between the aromatic π face and proline ring C-H bonds. The canonical aromatic amino acids Trp, Tyr, and Phe strongly disfavor a polyproline helix (PPII) when they are present in proline-rich sequences because of the large populations of cis amide bonds induced by favorable aromatic-proline interactions (aromatic-cis-proline and proline-cis-proline-aromatic interactions). We demonstrate the ability to tune polyproline helix conformation and cis-trans isomerism in proline-rich sequences using aromatic electronic effects. Electron-rich aromatic residues strongly disfavor polyproline helix and exhibit large populations of cis amide bonds, while electron-poor aromatic residues exhibit small populations of cis amide bonds and favor polyproline helix. 4-Aminophenylalanine is a pH-dependent electronic switch of polyproline helix, with cis amide bonds favored as the electron-donating amine, but trans amide bonds and polyproline helix preferred as the electron-withdrawing ammonium. Peptides with block proline-aromatic PPXPPXPPXPP sequences exhibited electronically switchable pH-dependent structures. Electron-poor aromatic amino acids provide special capabilities to integrate aromatic residues into polyproline helices and to serve as the basis of aromatic electronic switches to change structure.

Aromatic amino acids play distinct structural and functional roles, because of the combination of hydrophobicity with the unique interactions possible via the negatively charged aromatic faces and the positively charged aromatic edges.[1−5] Aromatic quadrupoles, in contrast to simple aliphatic hydrophobic groups (e.g., cyclohexyl or tert-butyl), provide the possibility of specific structural interactions and well-defined geometric orientations for aromatic rings that are distinct from those possible solely via the hydrophobic effect.

Aromatic residues exhibit special interactions with proline residues, which promote cis amide bonds via local aromatic–proline interactions (either aromatic–cis-proline sequences or Hα-cis-proline–aromatic sequences) (Figure 1).[6−26] In the Protein Data Bank (PDB) and in model peptides, proline–proline and aromatic–proline sequences are the most likely to adopt a cis amide bond. In aromatic–proline sequences, the population of cis amide bonds correlates with aromatic electronics, consistent with a C–H/π interaction in which the partial positive charge on the hydrogen of a polarized proline C–H bond (i.e., Hα or Hδ, adjacent to the electron-withdrawing amide carbonyl and/or amide nitrogen) interacts with the negatively charged face of the aromatic ring to stabilize the cis amide bond, in a manner comparable to a classical cation−π interaction.[15,22,24,25,27−29] The aromatic–proline interaction also could potentially, with appropriate geometry, be additionally stabilized by interactions between the aromatic π orbitals and the σ* orbital of the C–H bond, which is consistent with the observation of sub-van der Waals distances in a significant number of C–H/π interactions, including those in small molecules, peptides, and proteins.[27,28,30−35] Thus, the aromatic–proline interaction can be considered electrostatic (δ–aromatic·δ+proline) and/or potentially stereoelectronic (πaromatic → σ*proline C–H) in nature, in addition to contributions from the hydrophobic effect. The strength of a C–H/π interaction is dependent on the electronics of the aromatic system, as is typical for cation−π and polar X–H/π interactions, with stronger interactions observed for more electron-rich aromatics and weaker interactions for electron-poor aromatics.[3,27,29,36,37]

Figure 1

Aromatic–proline interactions that favor cis amide bonds in proline-rich sequences. (a) Aromatic–cis-proline interaction. (b) Hα-cis-proline–aromatic interaction. Hα (i – 2)-cis-proline–aromatic interactions may occur with the Hα of any residue two residues prior to (i – 2 to) the aromatic residue but are most favorable with proline.[20] (c) Conformational heterogeneity in proline-rich sequences (PPFPP) due to multiple aromatic–proline interactions. Additional cis–trans isomerism is possible at other X-Pro sequences.

Proline-rich domains, as defined in ref (38), are among the most common domains in eukaryotes. Proline-rich domains have central functional roles, including in protein–protein interactions, in linking globular or other functional domains, and in responsiveness to phosphorylation.[39−43] Proline-rich domains are considered to be intrinsically disordered, rendering them resistant to structural analysis by X-ray crystallography. To understand protein structure within proline-rich domains, we recently analyzed the propensities of the 20 canonical amino acids to adopt the polyproline helix (PPII) conformation, via a host–guest model system that was designed to understand the conformational preferences in a typical proline-rich sequence (Ac-GPPXPPGY-NH2, where X is any canonical amino acid).[23] In that work, we found that the aromatic residues Phe, Tyr, and Trp strongly disfavor polyproline helix when they are present in proline-rich sequences, via the induction of very large populations of cis amide bonds (45–60% of all species contained at least one cis amide, compared to ≤10% of species with a cis amide bond for nonaromatic residues X). The very large population of cis amide bonds for peptides with aromatic residues in this context, which inherently prevents polyproline II helix formation in the residues involved in the cis amide bond, was primarily due to substantial induction of cis amide bonds at both the Pro2–cis-Pro3–aromatic4 and the aromatic4–cis-Pro5 sequences (Figure 1). In addition, bioinformatics analysis of proline-rich sequences from diverse eukaryotes (humans, mice, Drosophila, Caenorhabditis elegans, and Arabidopsis) demonstrated very strong biases against the incorporation of aromatic residues in proline-rich sequences, presumably because of substantial conformational heterogeneity that would result from multiple cis–trans isomerism events. Notably, collagen sequences (containing ProHypGly consensus repeats that assemble into a triple helix of polyproline helices) also exhibit very strong preferences against replacement of either the Pro or Hyp residues with aromatic amino acids, presumably because of similar problems of conformational heterogeneity and resultant slow kinetics in collagen folding and assembly.[44] Collectively, these data suggest that the very poor PPII propensity of aromatic residues in proline-rich sequences, in contrast to non-proline-rich sequences, is due to the special interactions of the proline residues with the aromatic rings inducing multiple cis amide bonds and non-PPII structure (Figure 1c).[45−50]

The polyproline helix is a fundamental secondary structure of proteins that is widely employed in molecular design because of its rigidity and lack of dependence on hydrogen bonding.[45,46,51−63] Notably, both aromatic residues and polyproline helices are, separately, broadly employed in molecular recognition.[1,2,5,39] The incorporation of aromatic amino acids within polyproline helices thus could be exploited in molecular recognition and biomolecular design. In peptides with aromatic–proline dipeptide sequences, the population of trans amide bonds can be increased using electron-poor aromatics, which weakens the aromatic–proline interaction. These data suggest that electron-deficient aromatic amino acids in proline-rich sequences could potentially promote polyproline helix. Herein, we examine the ability to control structure in proline-rich peptides using aromatic electronic effects.

Experimental Procedures

Peptide Synthesis

Peptides were synthesized via standard solid-phase peptide synthesis and purified to homogeneity via high-performance liquid chromatography (HPLC), as determined by the presence of a single peak on HPLC re-injection (>95% purity). Peptide synthesis and characterization details are given in the Supporting Information.

Circular Dichroism (CD)

CD experiments were conducted on a Jasco J-810 spectropolarimeter. Unless otherwise indicated, peptides were analyzed at 25 °C in aqueous buffer containing 5 mM phosphate and 25 mM KF. CD data are the average of at least three independent trials. Data are background-corrected but not smoothed. Error bars indicate the standard error.

Nuclear Magnetic Resonance (NMR) Spectroscopy

NMR experiments were conducted at 23 °C in 90% H2O/10% D2O containing 5 mM phosphate buffer (pH 4 or as indicated) and 25 mM NaCl, unless otherwise indicated. NMR experiments were conducted using watergate water suppression. Populations of species with all-trans amide bonds versus species with one or more cis amide bonds were determined via integration of all peaks associated with a resonance in the one-dimensional spectrum using assignments from TOCSY spectra or by analogy. Each calculated % cis (= 100 – % of the all-trans species = 100 × (sum of all species containing at least one cis amide bond divided by the total of all species, including the all-trans species)) involved integration of multiple sets of resonances to confirm the conformational identity. In peptides, % cis includes all populations of peptide species containing at least one cis amide bond, compared to the peptide species containing all-trans amide bonds. The error in each % cis is estimated to be ≤±3 percentage points.

Results and Discussion

To test whether the electronics of aromatic amino acids could be used to control cis–trans isomerism and polyproline helix structure within proline-rich sequences, a series of Ac-GPPXPPGY-NH2 peptides was synthesized, where X is one of a series of electron-rich to electron-poor aromatic amino acids. Peptides were synthesized using both commercially available amino acids, as well as derivatives of 4-thiophenylalanine, which were synthesized from the peptide containing 4-iodophenylalanine and a copper-mediated cross-coupling reaction of thiolacetic acid with the fully synthesized peptide on the solid phase. Thiolysis of the resultant peptide yielded 4-thiophenylalanine.[24]

All peptides were analyzed by NMR spectroscopy, and the populations of species with all-trans amide bonds versus those with one or more cis amide bonds were quantified (Figure 2 and Table 1). As expected, the peptide containing the electron-donating amino substituent (4-NH2-Phe) exhibited large populations of species with cis amide bonds, similar to those observed with Tyr. In contrast, increasingly electron-withdrawing substituents exhibited reduced populations of cis amide bonds, in a manner concomitant with aromatic electronics. These data are consistent with the interpretation that large populations of cis amide bonds for aromatic amino acids in proline-rich sequences are due to C–H/π interactions, which are weaker with less electron-rich aromatics. As expected for a π-facial interaction,[3,64−66] the effects observed were broadly independent of the position on the aromatic ring (2- vs 3- vs 4-substituted). These data demonstrate the ability to modulate cis–trans isomerism within a proline-rich sequence using aromatic electronics. Interestingly, by NMR, significant upfield shifts were observed in a subset of proline resonances (Figures S14–S24 of the Supporting Information), compared to the peptides for which X is alanine, consistent with aromatic–proline interactions that place these proline hydrogens near the face of the aromatic ring. Moreover, the extent of upfield proline chemical shifts correlated with aromatic electronics, with the largest effects in the X = tryptophan peptide, the smallest effects in the X = pentafluorophenylalanine peptide, and intermediate effects in peptides with other aromatic amino acids. Notably, the 3JαN coupling constant, which correlates with the backbone torsion angle ϕ,[67] of the aromatic guest residue (X) in the species with all-trans amide bonds was 6.9–7.9 Hz. These data indicate the compatibility of aromatic residues with polyproline helix when in the all-trans amide conformation, as has been observed previously for aromatic residues in glycine-rich sequences.[45−50]

Figure 2

1H NMR spectra (amide–aromatic region) of Ac-GPPXPPGY-NH2 peptides, where X is phenylalanine substituted as indicated. NMR spectra for peptides for which X is Phe, Tyr, or Trp are shown in ref (23). All other NMR spectra are given in the Supporting Information. In the peptide with pentafluorophenylalanine (bottom), the peak assignments for the major (all-trans amide bonds) species are δ 8.66 (Gly7 amide), 8.24 (F5-Phe amide), 8.13 (Gly1 amide), 7.74 (Tyr8 amide), 7.40 (-CONH2), 6.99 (Tyr aromatic), 6.90 (-CONH2), and 6.62 (Tyr aromatic). Analogous resonance assignments apply for the all-trans amide species in other peptides. For species with cis amide bonds, glycine amides are readily identified by their pseudotriplet (doublet of doublet) coupling patterns, while C-terminal carboxamides are readily identified as singlets, with the singlet at ∼7.40 ppm being a relatively isolated resonance that in most cases can be used to determine the number and relative populations of the species present. The glycine amide peaks downfield of the all-trans species at ∼8.5 ppm also are diagnostic of the relative populations of the species with cis amide bonds. Control experiments with Pro3 or Pro5 replaced with Leu or Hyp confirmed that the major cis–trans isomerization events were those indicated in Figure 1.

Table 1

Summary of 1H NMR Data for All Ac-GPPXPPGY-NH2 Peptidesa

X	% cis	Ktrans/∑cis
Trp	63	0.59
4-OH-Phe (Tyr)	58	0.74
4-S–-Phe	57	0.77
4-NH2-Phe	56	0.80
4-SH-Phe	56	0.80
Phe (4-H)	51	0.97
3,4-di-F-Phe	49	1.1
4-OAc-Phe (TyrAc)	49	1.1
2-F-Phe	46	1.2
4-I-Phe	46	1.2
3-F-Phe	45	1.2
3-Cl-Phe	45	1.2
4-S-(2-NO2Bn)-Phe	45	1.2
3,4-di-Cl-Phe	41	1.4
3,5-di-F-Phe	41	1.4
4-F-Phe	40	1.5
3-NO2-Phe	36	1.8
4-CF3-Phe	34	2.0
4-NO2-Phe	32	2.1
4-SAc-Phe	31	2.2
4-+NH3-Phe	17	4.9
pentafluoro-Phe	16	5.2
His(H+)	8	12.3

% cis is the percent of the total peptide species containing at least one cis amide bond. ∑cis is the sum of the population of all species containing at least one cis amide bond. K is the ratio of the population of the species will all-trans amide bonds divided by the sum of the populations of all species containing at least one cis amide bond.

Among the aromatic amino acids examined, the smallest populations of cis amide bonds were observed for peptides with electron-withdrawing substituents, including 4-nitrophenylalanine, 4-thioacetylphenylalanine, protonated 4-aminophenylalanine, and pentafluorophenylalanine. In contrast to most aromatic residues, the peptide with pentafluorophenylalanine exhibited only one major species by NMR, with populations of species with cis amide bonds that were close to the small populations of cis amide bonds observed in peptides with aliphatic amino acids (≤10% cis).[23] To identify the compatibility of electron-poor aromatic amino acids with PPII, the peptide with pentafluorophenylalanine was examined by CD. The CD spectrum of this peptide exhibited a significant positive band with a λmax of 223 nm and a larger minimum at a λmin of 204 nm, typical of PPII observed in peptides and proteins,[23,45,68−71] indicating that the peptide with pentafluorophenylalanine significantly adopts PPII (Figure 3).

Figure 3

CD spectra of Ac-GPPXPPGY-NH2 peptides. Left: X = pentafluorophenylalanine. Right: X = 4-NH2-phenylalanine (blue squares, pH 7) and X = 4-+NH3-phenylalanine (red circles, pH 3). Peptides for which X is Phe, Tyr, and Trp exhibited no difference in CD spectra at pH 3 vs pH 7 (Figure S1 of the Supporting Information). CD spectra of peptides for which X is one of the 20 canonical amino acids, including PPII-favoring Pro, Leu, and Ala and PPII-disfavoring Thr, Ile, and Val, are shown in ref (23). PPII in this context is indicated by the presence of a positive band at a λmax of ∼228 nm, with red-shifted and smaller maxima indicating lower PPII content.[23,45,68−71]

The peptides with thioacetate/thiol/thiolate substituents (4-thiophenylalanine, pKa = 6.4[24]) or ammonium/amine substituents [conjugate acid of aniline, pKa = 4.9; measured pKa within peptide of 4.7 ± 0.2 (Figure S7 of the Supporting Information)] exhibited NMR spectra that indicated that peptide structure could be controlled by chemical reactions (e.g., thiolysis of a thioester) or solution acidity. To examine the use of pH as an electronic switch of structure, the peptide Ac-GPPXPPGY-NH2 (X = 4-aminophenylalanine) was analyzed by CD in both the ammonium (pH 3) and amine (pH 7) protonation states (Figure 3). At pH 7 (Ar-NH2 protonation state), this peptide exhibited a diffuse CD spectrum with a positive band with a λmax of 240 nm, substantially red-shifted from that of an ideal PPII, and also consistent with a strong aromatic contribution to CD.[23,45,70,72,73] In contrast, at pH 3 (Ar-+NH3 protonation state), where small populations of cis amide bonds were observed, this peptide exhibited a CD signature consistent with a stable polyproline helix (λmax = 229 nm; λmin = 204 nm), similar to that observed for the peptide with pentafluorophenylalanine or peptides with nonaromatic guest residues.[23]

These data indicate that aromatic residues in proline-rich peptides can be used to switch conformational ensembles from species with large populations of cis amide bonds, with an inherently more compact conformation, to species with predominantly trans amide bonds and a relatively more extended polyproline helix conformation. PPII structures are stable in proline-rich sequences despite the absence of hydrogen bonds, because of the presence of n → π* interactions between adjacent residues and the absence of competing hydrogen-bonded structures.[59,61,74−78] We sought to examine whether alternating block sequences of proline and aromatic residues with two electronically distinct states could be used to develop electronic switches of peptide structure. Peptides were synthesized with one (Ac-PPXPPGY-NH2), two (Ac-PPXPPXPPGY-NH2), or three (Ac-PPXPPXPPXPPGY-NH2 and Ac-PPXPPXPPX-NH2) 4-aminophenylalanine residues embedded within proline-rich sequences. As a secondary structure, polyproline helix exhibits a 3.1 Å rise/residue. Thus, these peptides, if in a polyproline helix throughout their proline–aromatic sequence, would extend 16, 25, and 34 Å, respectively, with all aromatic residues along one PPII-helical face.

These peptides were analyzed by CD and NMR in both protonation states (Figure 4). At pH 7 (Ar-NH2 protonation state), by CD all peptides exhibited broad bands with maxima at ∼240 nm, similar to that observed in the Ac-GPPXPPGY-NH2 model peptide containing 4-NH2-Phe. By NMR, high populations of cis amide bond and exceptional conformational hetereogeneity were observed in all peptides with 4-NH2–Phe, consistent with this residue inducing substantial populations of cis amide bond around each aromatic residue via multiple aromatic-proline interactions. In contrast, at pH 3 (Ar-+NH3 protonation state), by CD all peptides exhibited a positive band at 225–230 nm, the loss of the intense band at 240 nm, and a more defined minimum at 205–210 nm, consistent with typical PPII CD spectra. NMR confirmed a substantially reduced population of cis amide bonds in these peptides, despite two, four, and six (for peptides with one, two, and three PPX sequences, respectively) potential cis-proline–aromatic or aromatic–cis-proline interactions in these peptides (pages S25–S27 and S34–S36 and Figures S25–S27 of the Supporting Information).

Figure 4

(a and b) NMR spectra of Ac-PPXPPGY-NH2 (X = 4-aminophenylalanine) at (a) pH 3 and (b) pH 7 at 4 °C. In these simplified NMR spectra, four prominent species are observed, including all-trans amide bonds, aromatic–cis-Pro, Pro–cis-Pro–aromatic, and presumably the peptide with both aromatic–cis-Pro and Pro–cis-Pro–aromatic species. (c–f) CD spectra of Ac-(PPX)PPGY-NH2 [(c) n = 1; (d) n = 2; (e) n = 3] and (f) Ac-PPXPPXPPX-NH2 peptides, X = 4-aminophenylalanine, at pH 3 (R-+NH3, red circles) and pH 7 (R-NH2, blue squares). These peptides all exhibit one major species by NMR at pH 3, but a mixture of species at pH 7 (Supporting Information). Additional comparative CD spectra are given in Figures S2–S5 of the Supporting Information. pH-dependent CD spectra for Ac-PPXPPGY-NH2 and Ac-(PPX)3PPGY-NH2 are shown in Figure S6 of the Supporting Information. Temperature-dependent CD data for Ac-PPXPPGY-NH2 at pH 3 and 7 are shown in Figure S8 of the Supporting Information. Temperature-dependent CD data for Ac-(PPX)3PPGY-NH2 at pH 3 and 7 are shown in Figure S9 of the Supporting Information. Temperature-dependent CD data for Ac-GPPPPPGY-NH2 are given in ref (23).

In the simplified NMR spectra of the Ac-PPXPPGY-NH2 peptides (Figure 4a,b), the equilibrium between the all-trans conformation and conformations with each of the individual cis amide bonds [aromatic–cis-Pro and Pro–cis-Pro–aromatic (Figure 1c)] could be determined. In the amine (Ar-NH2) protonation state, the individual equilibria of aromatic–proline interactions exhibited K = 2.5 and 1.6. In contrast, in the ammonium (Ar-+NH3) protonation state, K = 8.1 and 6.1 were observed, representing ΔΔG = −0.66 and −0.75 kcal mol–1, respectively, of increased preference for trans amide at each of these prolines in the peptides with an electron-withdrawing ammonium compared to an electron-donating amine. These peptides also exhibited temperature-dependent CD spectra that were consistent with the presence of PPII as the major species in the ammonium protonation state, but non-PPII structures as the major species in the amine protonation state (Figures S8–S10 of the Supporting Information). In total, these data demonstrate that aromatic electronic properties can be used to switch cis–trans isomerism in proline-rich sequences and polyproline helix conformation. Given the range of potentially electronically tunable and conditionally responsive aromatic amino acids, and the large distances that could be spanned by polyproline helices, including within collagen triple helices, these data suggest the application of aromatic electronics in proline-rich sequences as an intriguing strategy in macromolecular control.

We have previously demonstrated, in model peptides and in peptides derived from the microtubule-binding protein tau, that serine/threonine phosphorylation can induce a reversible conformational switch to PPII.[79,80] The basis of this switch is the modification of a residue with poor PPII propensity (Ser or Thr) to a residue with substantially greater PPII propensity (phosphoserine or phosphothreonine). Herein, we demonstrate the ability to use aromatic electronics to control polyproline helix conformation.[81,82] Electron-rich aromatic amino acids preceding or following proline residues strongly favor cis-proline amide bonds, and thus strongly disfavor polyproline helix, because of multiple favorable proline–aromatic C–H/π interactions. In contrast, in electron-poor aromatic residues, the reduced electron density in the π face (reducing the electrostatic driving force of a C–H/π interaction) significantly reduces the population of cis-proline amide bonds, resulting in substantially greater polyproline helix.

C–H/π interactions are typically observed when polarized C–H bonds interact with aromatic rings, with a favorable interaction between δ– of the aromatic ring and δ+ of the polarized C–H bond. C–H/π interactions are significantly electrostatic in nature, with substantial additional contributions from the hydrophobic effect and from dispersion forces.[30,32,33,36,83−85] The significantly electrostatic nature of C–H/π interactions is consistent with their geometric dependence, their common orientation toward the centroid of the aromatic ring, and the observation of stronger C–H/π interactions with more acidic hydrogens. Interestingly, surveys of the CSD and the PDB have revealed that many C–H/π interactions exhibit distances below the sum of the van der Waals radii (nonbonded H···C distances of <2.90 Å), and also orientations of the C–H bond away from the centroid of the aromatic ring.[15,27,28,30−33] The observation of sub-van der Waals distances is consistent with a potential additional role of stereoelectronic effects in stabilizing C–H/π interactions, with appropriate overlap of the donor π orbitals with the σ* orbital of the C–H bond. These orbital interactions are analogous to stereoelectronic effects observed in C–H···O interactions, with C–H σ* acceptors to an O lone pair donor.[35] Short C···H distances are found in both aromatic–cis-proline interactions and Hα-cis-proline–aromatic interactions (Figure 5;[86,87] see refs (15), (27), (28), (30−33), and (88) for additional examples). Notably, in these structures, the C–H bond is often oriented away from the centroid of the aromatic ring, positioning the C–H σ* orbital near the aromatic π HOMO. These distances and geometries, which are divergent from described idealized C–H/π interaction geometries based on optimized electrostatic interactions, are consistent with a potential role for stereoelectronic effects in stabilizing aromatic–proline C–H/π interactions.

Figure 5

Aromatic–proline interactions with short inter-residue C–H distances. Distances shown are between the indicated Pro Hα and the indicated carbons of the aromatic ring, with distances < 2.90 Å being less than the sum of the van der Waals radii of C and H. Hydrogens were added in Pymol. (a) Residues 191 and 192 of PDB entry 3wfh (sequence WP, 1.90 Å resolution, anti-prostaglandin E2 Fab). (b) Residues 71 and 72 of PDB entry 1h4i (sequence FP, 1.94 Å resolution, Methylobacterium extorquens methanol dehydrogenase).[87] (c) Residues 79–81 of PDB entry 1aoc (sequence PPF, 2.00 Å resolution, horseshoe crab coagulogen).[86]

Conclusion

Electron-rich aromatic amino acids strongly disfavor polyproline helix in proline-rich sequences because of cis–trans isomerism resulting from multiple aromatic–proline interactions. In contrast, electron-poor aromatic amino acids disfavor aromatic–proline interactions and favor polyproline helix. Electron-poor aromatic amino acids provide special capabilities to integrate aromatic residues into polyproline helices and to serve as the basis of aromatic electronic switches to change structure. The use of aromatic electronics to tune polyproline helix conformation could have broad applications in biomolecular design, medicinal chemistry, biomaterials, and engineering.