A novel peptide stapling strategy enables the retention of ring-closing amino acid side chains for the Wnt/β-catenin signalling pathway† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c7sc02420g Click here for additional data file.

An alternative all-hydrocarbon stapling approach in which the amino acid side chains are retained at the stapled positions.


Introduction
Peptide drugs offer the prospect of enhanced potency, high specicity and low toxicity because of their large interacting surfaces. These features are particularly attractive for the disruption of protein-protein interactions (PPIs). A large group of peptides bind their receptors in the a-helical conformation but the helix proportion of linear peptides is usually low in solution. 1 The construction of peptide derivatives with a preorganized stable a-helix topology is expected to favour receptor binding and therefore has been widely investigated in recent years. [2][3][4] Generally, the stabilization of helices has been fullled through side-chain constraints by the covalent connection of two side-chain residues with disuldes, 5 lactams, 6 triazoles, 7 and others. 8 Among them, the all-hydrocarbon stapled peptide strategy developed by Verdine et al. 9 has been considered as one of the most promising stapling strategies for PPIs (Fig. 1). This strategy has been successfully used to improve the in vivo stability of peptides and enhance their membrane penetration against many biological macromolecular receptors associated with cancer, atherosclerosis, HIV and other diseases. [10][11][12] To avoid the intrinsic helix-destabilizing effect of D-congured amino acids while capitalizing on the helix-stabilizing effect of a,a-disubstituted amino acids, Verdine and coworkers introduced an a-methyl group into Fmoc-R 5 -OH and Fmoc-S 5 -OH, 9 which restricted the stapling position to Ala or mutated Ala. However, it is technically difficult to staple a peptide at the Ala position for every sequence, especially when each stapled peptide requires two stapling positions. Although reasonable stapled peptides can be rationally designed based on the Ala-scan and crystal structure of a peptide-protein Fig. 1 The difference between the previous method and the new method in this work. i ¼ stapling position; R 5 ¼ Fmoc-R 5 -OH; S 5 ¼ Fmoc-S 5 -OH; AA* ¼ new amino acids with modifications of the pentene groups on the a-carbon; R ¼ amino acid side chain.
complex, it remains interesting to explore whether the retention of the side chain of the stapling residue may bring about any extra benet (Fig. 1). This may be especially important for the cases where we cannot ignore the role of peripheral residues around PPIs. Moreover, another oen encountered drawback of the all-hydrocarbon stapling strategy is the solubility problem, especially in the cases where the native hydrophilic side chains of Ser, Lys or Arg are sacriced due to the stapling.
To solve the above problems, several methods have been developed to stabilize a-helices by the replacement of an internal (i, i + 4) hydrogen bond with a covalent linkage, such as an ethylene bridge developed by Alewood et al., 13

Results and discussion
To achieve the above goal, we rst need to make amino acids which contain both native side chains and stapling groups. For this purpose, seven classical N-Fmoc-a-pentene amino acids were tested as examples in the beginning, namely, Leu*, Met*, Ser*, Tyr*, Lys*, Arg* and Phe* (1a-g, as shown in Fig. 2A). A versatile de novo approach for the synthesis of these novel amino acids was developed by the optimization of a reported strategy for preparing Fmoc-R 8 -OH (Fig. 2B). 18a The conguration of the unnatural amino acids is critical and depends on a series of key intermediates 5a-g (Fig. 2B), oxazolidinone-like amino acid derivatives. In our study, steric The synthetic routes for the key intermediates 5a-g. (C) The synthetic route for Fmoc-AA*-OH 1a-g. Fmoc ¼ fluorenylmethyloxycarbonyl; hindrance caused by the N-benzyl protected group and oxazolidinone was exploited to acquire the product with correct stereochemistry. First, native a-amino acids (2a-f), which can be obtained easily from commercial resources or the reported methods (as shown in the ESI †), were protected at the nitrogen atom to provide N-t-Bu imide amino acids (3a-f). 18b Second, 3a-f were converted to the corresponding oxazolidinones 4a-f by Cbz-Cl in DCM. 19 Third, treatment of 4a-g with LiHMDS and 5iodo-1-pentene in a THF/HMPA solution at À78 C afforded the desired 5a-g in good yields. Particularly, the cyclization of 3g with BF 3 /Et 2 O and PhCH(OMe) 2 generated 4g in 71% isolated yield. Hydrolysis of the alkylated oxazolidinones with KOSiMe 3 afforded the free amino acids. Finally, the free amino groups of 6a-g were protected by a Fmoc group to give the key amino acids 1a-e, 20 which were ready for use in standard Fmoc SPPS. The azide of 7f was reduced by Zn and CH 3 COOH, and then we used AgNO 3 to promote the guanidinylation of the amino group, yielding 1f (Fig. 2C). 21 To assume our novel amino acids would not racemize, we also synthesized D-conguration Phe* (1g), and then examined the enantiopurity of 1g. As anticipated, our synthesis route enabled 1g to reach 95.82% enantiopurity.
As a typical PPI model, previous studies have suggested that the Axin-b-catenin interaction is suitable for targeting by hydrocarbon-stapled peptides. 22a Herein we incorporated our novel amino acids into a b-catenin-binding domain of Axin (469-482) and presented the rst stapled peptides with the retention of the native side chains. The scaffolding protein Axin, glycogen synthase kinase-3b (GSK-3b) and the adenomatous polyposis coli protein (APC) constructed a cytoplasmic protein complex, which can catalyze the phosphorylation of b-catenin in the absence of a Wnt signal. 23 It has been demonstrated that canonical Wnt/b-catenin signaling plays a critical role in the development of embryogenesis, the maintenance of adult tissue homeostasis and the control of tissue regeneration. A variety of human diseases, including osteoporosis, neurodegenerative diseases, diabetes and Joubert syndrome, are concluded to be closely related to the aberrant activation of the Wnt/b-catenin signalling pathway. 24 As shown in Fig. 4a, the Axin-b-catenin interface comprised both a rather hydrophobic interaction and some critical salt bridges and hydrogen bonds, of which Leu-473, Asp-474, and His-476 are critical residues and Ile-472, Val-477, Val-480 and Met-481 play moderately important roles. 22b Each of these residues is more or less involved in the binding interaction. To evaluate the new stapling strategy, we incorporated the amino acids Leu*, Met*, Ser* and Lys* combined with S 5 or R 5 into the b-catenin-binding domain (Axin (469-482)) ( Table 1). Solidphase peptide synthesis 25a and RCM 25b were carried out as previously reported, and thirteen stapled peptides containing Leu*/Met*/Ser*/Lys* and S 5 /R 5 were successfully synthesized ( Table 1).
The linear peptide was rst prepared using standard Fmoc SPPS procedures with Rink Amide MBHA resin as the solid support. The Fmoc protected amino acids were successively assembled onto the resin. Aer the peptide assembly was completed, the olen-containing peptide was stapled using Grubb's rst-generation catalyst. The peptide was cleaved off from the resin and globally deprotected with reagent K (82.5% TFA, 5% H 2 O, 2.5% EDT, 5% thioanisole and 5% phenol). Ether precipitation gave the crude peptides, which were puried by semi-preparative RP-HPLC. WNT-2d and WNT-7d failed to undergo RCM, even under high temperature, suggesting that steric hindrance of these two sequences can result in nonproductive conformations for the cyclization step. 8d The amino acid side chain of Lys* was initially an azide group, which was reduced to an amine on the resin by a nickel catalyst in the presence of NaBH 4 aer RCM (ESI, 3.1 †). To test the coupling efficiency of our unnatural amino acids, we monitored the coupling step of Met* and the subsequent elongation of Val in WNT-3a by analytical HPLC through micro-cleavage. According to the trace comparison of each intermediate in HPLC (Fig. S1 †), the coupling efficiency was not obviously compromised in spite of the introduction of an extra long alkene linker.
We next explored the effects of these peptides on Wnt/bcatenin dependent transcriptional activity. The Topash reporter, in which rey luciferase is transcriptionally activated by b-catenin, was employed. As shown in Fig. 3b, the new stapled peptides WNT-1a, -3a, -4a, -5a, -6a and -8a enhanced the Wnt3a-induced reporter expression more effectively than Axin (469-482), among which WNT-3a, -5a, and -6a treatment exhibited a three-fold enhancement. As expected, these stapled Table 1 The synthetic route and amino acid sequences of the stapled peptides a a *: the synthetic amino acid with the retention of the native side chain. WNT-2d and -7d: the peptide failed to undergo RCM. N/A: not applicable.
peptides promoted more Wnt3a-induced reporter expression than the corresponding regular stapled peptides WNT-3b, -3c, -5b, and -6b, demonstrating the benet of maintaining the side chains of Met-481, Ser-471 and Leu-473. Among them, WNT-3a showed a more signicant improvement than -3b and -3c, highlighting the inuence of the side chain Met-481. As we expected, the activity of WNT-3d, the uncycled WNT-3a, showed less promotion when compared to Axin (469-482) owing to the remarkable weakness of linear peptides (Fig. S2 †). All these results showcase the power of our new stapled peptides and the necessity of the retention of the specic side chain. It is worth noting that WNT-2d and -7d also signicantly promoted Wnt3ainduced reporter expression as uncycled linear peptides, of which the most potent, WNT-7d, with a modied Met compared with WNT-3a, exhibited a three-fold increase in potency compared to the wild-type peptide. The failure of cyclizing WNT-7d makes us unable to evaluate the activity of WNT-7a, a potential potent activator of the Wnt signalling pathway. Circular dichroism (CD) analysis of the peptides (Table 1 and Fig. 3a and S3 †) indicates that the helicity of Axin (469-482) is only 20%, while the helicity of WNT-1a to WNT-7d ranges from 22% to 71% corresponding to a 1.1-to 3.6-fold increase. Consistent with the previous report, 26 the "unstapled" analogue of WNT-2d and WNT-7d also displayed relatively high a-helicity, with an a-helical content value of 22% and 50%, respectively. These results demonstrate that our novel stapled peptide strategy can maintain the improvement of helicity when compared with the regular stapling strategy.
To test the protease stability of the stapled peptides with the retention of the side chains, we measured their susceptibility towards a-chymotrypsin-mediated degradation at room temperature in pH 7.4 PBS buffer containing 2 mM of CaCl 2 as monitored by HPLC. a-Chymotrypsin is a protease that predominantly cleaves at the carboxyl side of positively charged amino acids such as methionine and leucine. Under these conditions, the half-life of Axin (469-482) is 9 min (Fig. 3c). In sharp contrast, WNT-3a is, as expected, completely stable and no degradation was observed even aer 1 h. These results clearly demonstrate the inherent superiority of the new stapled peptides over linear peptides with respect to proteolytic stability.
We then assessed the cell permeability of the peptides by conjugating uorescein isothiocyanate (FITC) to Axin (469-482), WNT-3a and WNT-3c (Fig. 3d). Our data show that FITC-b-Ala-WNT-3a and -3c efficiently entered into the cytoplasm in the HEK 293T/17 cells. Unexpectedly, FITC-b-Ala-Axin (469-482) also penetrated the cell membrane but showed weaker reporter expression in the presence of Wnt3a compared to WNT-3a, demonstrating the critical role of the native side chain of Met-481. To explore the detailed binding mode of WNT-3a against b-catenin, we conducted a Z-dock of the WNT-3a-b-catenin complex to predict structural models of the protein-protein complexes. 27 As shown in Fig. 4a and b, subtle but important differences in the intermolecular interaction exist between WNT-3a and Axin (469-482). Owing to an introduction of the stapling group, Met-481 is closer to the Trp-338 of b-catenin, and thus forms a stronger interaction with that residue. In addition, a stronger salt bridge between the side chain carboxyl group of Asp-474 coordinated with the amino group of b-catenin Lys-292 and a more efficient hydrogen bond between the side chain of His-476 and b-catenin Asp-299 are constructed because of a closer distance geometrically. It is conceivable that these changes enhanced the conformational stability of WNT-3a in its bound state, thereby contributing to the improved binding activity.

Conclusions
In conclusion, we have developed a new series of stapling amino acids, which contain the native amino acid side chains and can be applied in the standard stapling chemistry for stabilized ahelical conformation peptides. In this study, we have made use of these amino acids and S 5 /R 5 , to prepare potent inhibitors of the Axin-b-catenin interaction. We have demonstrated that incorporation of AA* at the i position of an i, i + 3 or i, i + 4 stapled peptide is a tolerated modication that allows the allhydrocarbon staple to effectively mimic Axin, and exhibit superior protease stability. In this regard, this methodology is a more exible and alternative strategy for stapled peptides with potential impact for related PPIs.

Conflicts of interest
There are no conicts to declare.