Synthesis of β,β‐Dinaphthyl Amino Acids: Towards Molecular Gearing in Peptides and Mini‐Proteins

Abstract

We report the synthesis of β,β‐dinaphthyl amino acids by directed CH‐activation and their subsequent utilization in peptides. The synthesis of aromatic β‐branched amino acids is limited by sterical crowding for large aromatic ring systems. The amino acids are transformed into adequately protected building blocks for solid‐phase peptide synthesis by using our amide weakening strategy. Our aim was to study the interactions of the side chain aryl groups in solution and solid state. Based on quantitative analysis of ¹H NMR reference nuclei, we investigate conformational equilibria and compare them to established gearing systems.

Amino acids bearing two aromatic side chains are studied here for their ability to align in rotor‐like assemblies in a synthetic protein. Based on quantitative analysis of ¹H NMR reference nuclei, we investigate a correlated molecular rotor‐type behavior whose dynamic nature stands in remarkable contrast to established systems in the field of molecular bionics.

1. Introduction

Since the first description of the cogwheel effect by Kwart,^{[ 1 ]} researchers investigated functional molecular assemblies like molecular gears^{[ 2 ]} or ratchets,^{[ 3 ]} which today are even capable of unidirectional rotation.^{[ 4 ]} Common to many gears are interlocking aromatic rings in a sterically crowded molecular environment. The structure of these molecular machines developed from scratch is fundamentally different from the molecular machines found in nature which are based on polypeptides and perform significantly more complex functions such as the conversion of light into chemical energy that drives the unidirectional rotation of ATPase^{[ 5 ]} or the switchable bacterial flagellar motor,^{[ 6 ]} for example. Despite the differences, both types of machines obey the same physical rules of thermal motion, directed mastering of energy barriers, and more. A simple but inevitable obstacle between both regimes is solubility, because polyaromatic structures are insoluble in water, while, inversely, proteins unfold in organic solvents. Our aim here is to build a bridge by investigating correlated aromatic rotations within a peptide environment in an aqueous solution.

Ring flips of phenolic groups are fast at room temperature, visible by an AA´XX` observed by NMR spectroscopy.^{[ 7 ]} Adding a second aryl ring to the side chain as in diphenylalanine (Dip in Figure 1) does not significantly increase barriers to rotation, because geminal phenyl rings rotate independently without significant steric clashes. Even a larger 10π aromatic ring, which assumes a fixed orientation, does not significantly restrict the rotation of a geminal phenyl group (Wsf in Figure 1).^{[ 8 ]} Looking for a diaryl side chain with an intermediate energy barrier to rotation for two geminal aryl groups, we came up with the di‐2‐naphthyl side chain which appeared to be the best choice for studying barriers of intermediate height by NMR in solution. The incorporation of naphthyl groups provided interesting insights already in other disciplines like catalysis,^{[ 9 ]} material science,^{[ 10} , ^{11 ]} and medicinal chemistry.^{[ 11} , ^{12 ]} The interplay between protein environment and the side chain mobility should influence the diastereotopic naphthyl groups differently, yet not independently from each other, which should result in an allosteric effect between the environment of the proR naphthyl group and the proS group and vice versa. We expected an effective cooperative interplay for three dinaphthyl amino acids at the position of the phenolic groups in the wild‐type foldon—the C‐terminal trimerization domain of bacteriophage T4 fibritin.

Figure 1.

The increasing π‐character and steric demand allows for a systematic expansion of a single peptide side chain from phenylalanine (Phe) over diaryl amino acids (Dip, Wsf) to novel dinaphthyl amino acids (Njb, Dma).

Here we describe the synthesis of several dinaphthyl amino acids, their incorporation into peptides, and finally the synthesis and evaluation of the C₃‐symmetric miniprotein foldon equipped with six naphthyl rotors in the crowded environment of its N‐terminus, including a late‐stage demethylation.

2. Results and Discussion

2.1. Design

Figure 2 shows the relative energy barriers of rotation about χ² and χ^2′ for two geminal aryl groups. Two enantiotopic phenyl groups have fast interconvertible conformers from energy minima with minimal energy barriers in between (Figure 2a). The 2‐naphthyl group in the chiral environment of an amino acid or peptide however has four rotamers that are different in energy while the population of each minimum depends on the local environment (Figure 2b). A system in which the rotation of one aryl group leads to the rotation of the other and vice versa is in an allosteric equilibrium. This would require two energy minima for rotamers with opposite naphthyl orientations (Figure 2c). If the rotations are not only correlated but even concerted, this could lead to a molecular gearing system. Our goal is to design, analyze, and understand such a system from scratch by starting with allosteric equilibria in amino acids and inserting them into a protein environment.

Figure 2.

The first row shows molecular structures where the rotation around χ² and χ^2′ is indicated with blue arrows (χ²/χ^2′ are torsion angles Cα‐Cβ‐Cγ‐Cδ with respect to the proR and proS naphthyl group). Below are expected energy diagrams where the total energy is plotted against both rotations and transitions between local energy minima are highlighted by arrows. (a) Enantiotopic geminal phenyl groups. (b) Interlocked diastereotopic geminal 2‐naphthyl groups of an amino acid. (c) Conformational restriction in a designed peptide environment.

To increase the barriers between a reduced number of energy minima as shown in Figure 2c and simultaneously increase the solubility in an aqueous solution, we chose a synthesis method that allows not only the introduction of two identical side chains but also the stereoselective linking of two different naphthyl‐sized side chains. Directed CH‐activation appeared to be the method of choice because it is tolerant to at least one sterically demanding side chain and therefore we were confident to apply this method for naphthyl groups, too.

2.2. Synthesis

The amino acids were synthesized using a directed CH‐activation approach introduced by Corey^{[ 13 ]} for the synthesis of diphenylalanine, which gained a wide range of applications.^{[ 14} , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ^{19 ]} Our synthesis starts from alanine 1, which is protected with a Phthaloyl group (Phth) at the N‐term and forms an amide with 8‐aminoquinoline (8AQ) at the C‐term. The 8AQ acts as a bidentate directing group for the palladium catalyst, enabling regio‐ and stereoselective CH‐activation at the β‐methyl group. At lower temperatures, only one arylation takes place, thus enabling the introduction of a different aryl group in a consecutive reaction. Deploying three different aryl iodides in the first arylation step led to the naphthylalanines 2a‐c (Scheme 1a). These reactions were conducted at a temperature of 60°C with t‐amylol as a solvent and the less reactive silver trifluoroacetate (AgTFA) as an additive. The lower yield of 2c is explained by the lower reactivity of the corresponding naphthyl iodide, as the conversion was not quite complete upon workup. Despite the lower overall reactivity, the reaction takes place regioselectively at the 2‐iodine position, leaving the 6‐iodine group ortho to the methoxy group intact as a useful handle for further transformations. Similarly, 1‐iodonaphthalene barely reacts with 1 under various conditions and does not react at all with 2, which already bears a β‐phenyl group. This is mainly attributed to the C─I bond being sterically more shielded through the ortho annulated benzene.

Scheme 1.

Synthesis of amino acids by CH‐activation with 3‐letter‐codes given in parentheses (explanation in SI). (a) Consecutive β‐naphthylation with two different naphthyl groups. (b) Twofold β‐naphthylation with the same naphthyl group in one step with modifications: [a] toluene and AgOAc instead of t‐amylol and AgTFA; [b] Pd(OAc)₂ increased to 30 mol% during reaction.

The monoarylalanines 2a‐c were arylated with each of the other two aryliodides respectively under higher temperatures of 100°C in toluene and with silver acetate as additive to produce a set of three diastereomeric pairs of dinaphthyl amino acids (3ab‐cb). While the reaction also proceeds at 80°C, these molecules are quite stable and are more readily transformed at elevated temperatures.

It is possible to introduce the same aryl group again in the second arylation step. Nonetheless, it is more feasible to introduce it twice in a one‐pot reaction starting from alanine 1, with double the reagents and higher temperatures (Scheme 1b). Using the same aryl iodides as before, the β‐symmetric dinaphthyl amino acids 3aa, 3bb, and 3cc were obtained. While the reaction proceeded quite well for 3aa, more reaction time was necessary for full conversion to 3bb. However, the formation of 3cc was not observed under these conditions. Instead, full conversion to the monoarylated intermediate 2c was achieved without the diarylation product, demonstrating again a lack of reactivity of the diiodinated substrate.

To improve reactivity, we applied the reaction conditions of the second CH‐activation (from Scheme 1a) with double the amount of reagents and 5 equivalents of the aryl iodide. After three days, a 66% NMR yield of 3cc with residual starting material was observed and it only increased by 5% after one week. Repeating the reaction at 120°C led to full conversion of compound 1 after one night and 3cc was isolated with 62% yield. These experiments show that dinaphthalic compounds reach the limits of reactivity due to steric hindrance.

Upscaling of the successful transformations is necessary to obtain reasonable amounts of Fmoc‐amino acids. However, this can be challenging due to the heterogeneous nature of combining polar and nonpolar features into the same molecule. At a 15 mmol scale, it took 24 days at up to 110°C to achieve full conversion to 3bb with only 51% isolated yield. Losses are attributed to inclusion in salt precipitates. To synthesize 3cc at a 10 mmol scale, even higher temperatures (140°C) were necessary and additionally 10 mol% palladium‐catalyst was added during the reaction. At last, full conversion was achieved with a 66% yield of 3cc. All syntheses required the complete transformation of the educts because they are hardly separable from the diarylated products by column chromatography. The less‐than‐complete yields are due to losses during the workup.

After optimization of the arylation conditions, it was necessary to remove the protecting and directing groups to obtain a building block that could be used in solid phase peptide synthesis (SPPS). We recently showed that for larger (10π electrons) diaryl amino acids, the cleavage of the 8AQ group was a challenge utilizing the usual methods such as transesterification,^{[ 15} , ¹⁶ , ^{17 ]} N‐activation,^{[ 19} , ^{20 ]} oxidative or reductive cleavage.^{[ 18} , ^{21 ]}

For such cases of steric crowding or sensitivity to oxidation, we developed an alternative strategy that leads to Fmoc‐amino acids.^{[ 22 ]} It involves the cleavage of the phthalimide followed by protection of the amine with the SPPS‐compatible Fmoc‐group. Then the quinoline is reduced to a tetrahydroquinoline in a Hantzsch‐ester transfer hydrogenation. The newly formed amine and the amide are cyclized with triphosgene to a urea derivative. This finally leads to the pyramidalization of the former 8AQ amide nitrogen, and thus the amide bond is readily saponified by lithium hydroperoxide. As these steps have been discussed extensively in previous works, they are shown only as a summary in Scheme 2 (variations in scale, reaction conditions, and workup procedures were applied as needed, see Supporting Information for details).

Scheme 2.

Transformation of protecting groups as described previously.^{[ 22 ]} Isolated yields over 5 steps, 3‐letter‐codes given in parentheses (see Supporting Information for details).

Thus, the β‐symmetric dinaphthyl Fmoc‐amino acids 4aa and 4bb were synthesized in both their l‐ and d‐configuration. While the synthesis of both enantiomers is expected to proceed in exactly the same way and, by itself, gives no additional insight, both building blocks are useful when incorporated into a chiral environment, i.e., into a peptide. With this aim, all four diasteromeric amino acids with a phenyl/naphthyl substitution in β‐position 4da/4ad were synthesized, further demonstrating the ability to produce SPPS building blocks in a gram scale in all possible absolute and relative configurations. The yields of some products may appear low, at around 20%, but considering that this yield is over five steps, this corresponds to an average yield of 72% per step. While the first steps provided good yields, the quinoline reduction was, in most cases, the step where most of the yield was lost. However, a significantly better overall yield was achieved, when no column chromatography was performed after steps 3 and 4 in the case of d‐4aa. With these building blocks in hand, we proceeded to incorporate them into peptides to analyze their impact on conformation.

2.3. Conformational Analysis

The ³ J coupling of ca 11 Hz identifies the antiperiplanar alignment of Hα and Hβ caused by steric repulsion of the four substituents at the Cα─Cβ bond (Figure 3). A locked conformation about χ¹ is a requirement for the evaluation of the conformational equilibrium about the χ² torsions, which is exemplary discussed for 3aa subsequently. Each of the two diastereotopic naphthyl groups adopts two possible conformations, in which it lies in a plane with the Cβ─Hβ bond, due to the least steric repulsion, and is either facing towards Hβ (syn) or away from it (anti) (Figure 3a). In a 2D‐ROE‐NMR spectrum, both conformations show distinct NOE contacts between the ortho‐protons of the naphthyl groups (labeled as r1/r3 and s1/s3 for the proR and proS group respectively) to Hα and Hβ. If the conformations are present in a 2:1 ratio, this is reflected in the time‐averaged NOE pattern, with the NOE signals of major conformation having twice the integral (Figure 3b).

Figure 3.

(a) Expected ROE signals for both conformers of the proR‐naphthyl groups. (b) Expected time averaged ROE pattern for, e.g., a 2:1 ratio of both conformers, also shown for the proS‐naphthyl group. (c) Experimentally obtained ROESY spectrum (600 MHz, 300 K, CDCl₃) of 3aa with relative ROE integrals given in blue numbers. (d) Conformational equilibria between the four possible conformers of the dinaphthyl sidechains of 3aa with the observed allosteric equilibrium highlighted in blue. The figures visualize the diastereomorphic character of the two rotamers, with their two “arms” flipping together.

Overall, 3aa shows eight time‐averaged NOEs for both naphthyl groups (Figure 3c). Short mixing times allow the quantitative analysis of the cross‐peak intensities and for identical distances but different peak integrals the analysis of the rotamer distribution (Figure 3d).

3aa shows a dominating conformer that has the proR naphthyl group in the syn‐ and the other one in the anti‐orientation (highlighted in box). It stands in equilibrium with a less populated conformer, where both naphthyl wings are flipped, what we call an allosteric equilibrium. The syn/syn or anti/anti orientations are presumably less populated, as their presence would lead to a different NOE signal intensity distribution.

“Snapshots” of various sidechain orientations were observed in the crystal structures of different dinaphthylated amino acid derivatives. Figure 4 shows one example adopting exclusively syn/syn orientation and a second crystal with two different orientations in the asymmetric cell of the crystal. This demonstrates the highly dynamic nature of this dinaphthyl compound even in the solid state.

Figure 4.

Crystal structures (syn‐naphthalene carbons orange, anti‐naphthalene carbons magenta, other carbons grey; nitrogen blue; oxygen red; chlorine green; aliphatic hydrogens not shown except Hα, Hβ and in CHCl₃/CDCl₃; H‐bond dotted line) of (a) a 3aa derivative, where the phthaloyl group has been substituted by Fmoc and the quinoline is reduced to a tetrahydroquinoline, and (b) 3aa co‐crystallized with CDCl₃.

The dinaphthyl amino acid derived from 3aa (Njb) and several of its analogs were then incorporated into peptides, including the short ghrelin peptide 5 (Scheme 3a). It was observed that Njb fits well into the π‐helical secondary structure that the peptide adopts in an aqueous solution and may even stabilize it through π–π interactions with the neighboring residues, as the NMR solution structure suggests. In this peptide, the difference in the naphthyl orientation observed in 3aa was less pronounced. But in contrast to that, here the proR naphthyl group was in the opposite orientation with the β‐r1 NOE being stronger than the α‐r1 NOE (Scheme 3b).

Scheme 3.

(a) Synthesis of the Njb containing short ghrelin hexapeptide 5, illustrated as a computer model based on NMR restraints (naphthalene carbons red; other carbons grey; nitrogen blue; aliphatic hydrogens not shown). The interaction between the two naphthyl groups is indicated with a purple arrow. (b) Expansion from the ROESY spectrum (600 MHz, 300 K, DMSO‐d ₆) of 5, with ROE integrals given in blue numbers (compare with Figure 3).

To further increase the difference between the naphthyl orientations, we looked for a structure with increased steric crowding but still capable of incorporating amino acid‐bearing side chains comparable to the size of four phenylalanines. The trimeric foldon was a promising template because three tyrosines form a propeller at its N‐terminal tip.^{[ 23 ]} As a starting point, we chose a modified foldon sequence with residues 10 and 17 being replaced by d‐Ala and d‐Phe respectively, which avoids aspartimide formation during synthesis and still folds well.^{[ 24 ]} In addition, we introduced a C‐terminal amide to obtain the [ala¹⁰, phe¹⁷]‐NH₂ foldon (6a, foldon^af), which was synthesized by SPPS (Scheme 4a). Then, we incorporated Njb into the foldon instead of Tyr², but without Gly¹ to leave more space for the six naphthyl groups (6b). The solubility in an aqueous environment was limited, but manageable at pH = 3. This was expected after addition of six hydrophopic 10π aromatic rings to the tip of the protein, as visualized in Scheme 4b. The solubility improved after reinstalling Gly¹ (6c) and was even better with the di(methoxynaphthyl)alanine Dma in 6d. The latter foldon was then demethylated to yield its dihydroxy analog 7. While this late‐stage functionalization of a mini protein has the potential to be optimized, a sufficient amount of product was obtained for NMR analysis. Despite the increased local polarity, the overall solubility was not increased detectably. The ¹H NMR spectra of all synthesized foldon variants show strongly shielded methyl group signals, indicating a successful fold (Scheme 4c). In 6b, the unsymmetric syn/anti alignment of the proR and proS naphthyl groups is increased up to 4:1, which is a significant shift compared to the amino acid in the short peptide (Scheme 4d). These findings suggest a correlation between the naphthyl orientations. Further experiments are needed to confirm, whether the observed allosteric equilibria are caused by the naphthyl‐naphthyl interactions and if this can lead to a gearing system. More detailed NMR analyses and modeling are in progress.

Scheme 4.

(a) SPPS of foldon variants 6 (with 6a synthesized fully automated) and demethylation of 6d. (b) Model structure of 6b with three Njb (shown in red) at the position of Tyr² in the tetrahedral crystalline trimer based on PDB entry 2ww6. The interactions (indicated with arrows) between naphthyl rings of the same amino acid and between neighboring chains (colored green, blue and orange) form an assembly of interacting aromatic rings. (c) Expansion from ¹H NMR spectra (600 MHz, 300 K; 10 mM pH7 phosphate buffer/D₂O 9:1 for 6a, 50 mM pH3 phosphate buffer/D₂O for the rest) of foldon variants, showing the region of shielded methyl groups. (d) Expansion from NOESY spectrum (600 MHz, 300 K, 50 mM pH3 phosphate buffer (in D₂O)/D₂O 9:1) of 6b with NOE integrals given in blue numbers (compare Figure 3).

3. Conclusion

We investigated allosteric aromatic interactions in amino acids and designed peptide and mini‐protein structures with desirable properties. Dinaphthylated amino acids bring two 10π side chains into a peptide which is of interest for medicinal chemistry and for the installation of allosteric interactions in proteins in aqueous solution. Amino acid polarity can be adapted to aqueous solutions by solubilizing substituents like methoxy groups which are then transformed into phenolic side chains. Peptides have great potential for research into molecular gearing, even if this study is only a first step in this direction.

Supporting Information

Experimental procedures and analytical data are available in the SI to this article. Deposition Numbers 2 406 212 and 2 406 213 (for 3aa and a derivative therof) contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service. The authors have cited additional references within the Supporting Information.^{[ 26–36 ]}

Author Contributions

Timo Zimmer and Armin Geyer conceived and designed the project and wrote the manuscript. Timo Zimmer performed the synthetic experiments and analyzed the data. Sergei I. Ivlev performed the x‐ray crystallography.

Conflict of Interests

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.