Introduction

Peptides are attracting great attention as therapeutics since they combine the high selectivity, potency and low toxicity of biologics with advantages such as the conformational restrictions and the reduced costs characteristic of small molecular entities1,2. In addition to the large number of commercialized peptides, many others are also found in clinical phases, thereby demonstrating their validity and application as active pharmaceutical ingredients3. However, the general use of peptides as drugs is severely hampered by their poor pharmacokinetic features. In this regard, it is widely believed that the characterization of protein–protein interactions, a fundamental issue in deciphering biological pathways, is better tackled through structurally defined small peptides with specific sequences. Therefore, there is a need for peptides with new topological architectures; however, these are difficult to obtain even using modern synthetic procedures4,5. To address these problems, general strategies for peptide macrocyclization have been developed to improve the properties (cell penetration, stability, selectivity and so on) and enhance the potential of peptides as therapeutics and bioprobes, with a special focus on poorly tractable targets6,7,8. In this context, peptides constrained via a non-amide sidechain-to-sidechain linkage (stapled peptides) provide a new structural paradigm because the conformational stabilized species display remarkably stronger biological activity. In this regard, the pioneering work of Verdine et al.9,10 on the basis of all-hydrocarbon staples through olefin metathesis represents a breakthrough in the field.

So far, stapled peptides are generated through a variety of strategies11, the main being the use of cysteine side chains to form disulfide bridges12 and thioether formation (crosslinking with α,α’-dibromo-m-xylene13 or aromatic nucleophilic substitutions with perfluoroaromatic reactants14); or with functionalized non-natural amino acids by means of biaryl linkages involving the borylated phenylalanine derivatives15,16,17,18, ring-closing metathesis19 and azide–alkyne cycloadditions (click chemistry)20, and so on, usually in expensive and long stepwise syntheses. A comparative table summarizes the major strengths and weaknesses of the main conventional stapling strategies (Fig. 1a). Owing to the increasing interest in stapled peptides and other conformationally restricted structures, many efforts have been made to develop new practical and general preparative methods for their generation.

Figure 1: Formation of stapled peptides.
figure 1

(a) Conventional stapling methods. (b) Phe/Tyr–Trp stapling via a selective Pd-catalysed C–H arylation process.

There is a considerable concern regarding the efficiency of synthetic aspects with respect to the preparation of complex bioactive compounds. In this context, the function-oriented synthesis approach to target simplified, although biologically meaningful, fragments of complex chemical entities is relevant, as it leads to practical syntheses of new specific drugs and probes, also in the field of peptides21.

Tryptophan (Trp) has a low relative abundance in peptide and protein sequences (≈1% of the amino acids); however, its presence is critical for the activity of these biomolecules. Therefore, the development of new synthetic methods for the selective and straightforward chemical modification of Trp is highly significant22.

Recently, metal-catalysed C–C coupling through direct C–H activation23,24,25,26,27,28,29 has become a fundamental process in modern organic syntheses, allowing the straightforward preparation of a plethora of new structural types. In this respect, the functionalization of indoles using this approach has been extensively examined30,31, including studies described in (refs 28, 32, 33, 34, 35) among others. Particularly relevant to our research was the methodology of Larrosa for indole arylation using Pd(OAc)2 in acidic media36. In particular, although the Pd-mediated C-2 arylation of Trp has been reported37,38,39, it has not yet been applied to staple true peptides. We recently disclosed the direct C-2 arylation of indoles in Trp-containing peptidic sequences in an intermolecular manner, without additional requierements37. Although the alternative N-arylation processes are conceivable, such Ullmann-type reactions take place normally with Cu and Pd catalysts, in the latter case usually requiring strong bases40. No experimental evidence of these transformations have been recorded in our systems. The main restriction is that lower conversions are obtained for sequences that comprise methionine, cysteine or histidine residues, presumably because of selective hydrolysis of the peptide bond catalysed by bidentate palladium coordination; on-going experiments along this line show that working in nonaqueous solvents allows reasonable arylations (unpublished results). Later we reported on the arylation of Trp-diketopiperazines41 and related transformations of Trp derivatives, leading to the generation of Fmoc-protected arylated Trps that are amenable to direct incorporation in solid-phase peptide synthesis (SPPS)42. These processes work well with N-protected peptides in dimethylformamide (DMF) or in aqueous environments.

In this work, we present a new stapling methodology involving Trp and Phe(Tyr) through a Pd-catalysed C–H activation process. The method is versatile, allowing the formation of constrained peptides of different ring sizes, amenable to solution- and solid-phase synthesis and can lead to the direct preparation of biologically meaningful peptide derivatives.

Results

Preliminary studies

Here we present a one-step process for the synthesis of Trp–Phe(Tyr)-stapled peptides directly from commercially available precursors. The method is based on an intramolecular Pd-catalysed C–H activation reaction between a Trp residue and an iodo-phenylalanine (or tyrosine) unit (Fig. 1b).

After a preliminary modelling study, we established the structural features for the intramolecular C–H arylation, placing the iodo-substituent for Phe or Tyr amino acids in the meta position. The preferential distances between these latter residues and Trp ranged from one to three amino acids in a series of linear N-terminal-acetylated sequences.

The initial experiments were carried out using the conditions previously applied in the arylation of the Trp diketopiperazine:41 Overall, 5 mol% Pd(OAc)2, AgBF4 (1.0 eq.) and o-nitrobenzoic acid (2-NO2BzOH) (1.5 eq.) in PBS:DMF (1:1) under microwave (MW) irradiation at 80 °C for 15 min (Table 1, entries 1–6). These preliminary observations suggested that these arrangements are suitable for stapled peptide formation, affording good to excellent conversions for Trp–Phe(Tyr) staples located at i−i+2, i−i+3 and i−i+4 positions (Table 1, entries 1–6). Furthermore, a series of representative sequences displaying from one to three amino acids between Trp and m-iodinated Phe were constrained in useful yields (Table 1, entries 7–10; Fig. 2; see Supplementary Fig. 1), in an anhydrous medium using trifluoroacetic acid (TFA) as the acid, applying the optimized conditions previously reported for the preparation of Fmoc-protected arylated Trps42. Although these transformations may lead to atropoisomeric diastereomers, only one stereochemically defined structure was observed in each case. Of note is the successful application of this technique to stapled peptides containing the Asn-Gly-Arg (-NGR-) and Arg-Gly-Asp (-RGD-) tumour-homing signalling sequences (2g and 2h, Table 1, entries 7 and 8, respectively).

Table 1 Results for the stapled bond formation of peptides 1 under microwave irradiation.
Figure 2: Structure of isolated locked peptides 2g–2k.
figure 2

Schematic representation: Phe (blue), Trp (purple) and staple bond (red).

Preliminary studies on the capacity of the RGD-containing compound 2h and its linear precursor 1h to inhibit cellular adhesion showed that these substances act selectively as antagonists for the ανβ3 in front of ανβ5 integrin receptors; compound 2h showing a moderate EC50 (6 μM) is more active than its linear precursor (1h, 26 μM, see Biochemical and Cellular Studies in Supplementary Methods). Compound 2h is considerably less potent than cilengitide, not surprisingly as this drug has been thoroughly optimized. However, it has to be considered that even linear analogues display a much lower potency (10–103 times, depending on the targeted integrin) than the parent cilengitide43,44. These preliminary results clearly show that this stapling technique preserves the activity of the natural sequence, while improving its potency, in line with pioneering experiments in macrocyclization techniques for medchem peptide development45,46.

We next tackled the preparation of the more challenging locked peptide i−i+1. Three sequences with adjacent Trp—(ortho-, meta- or para)-I-Phe were synthesized. A constrained C–C linked structure was obtained for the meta-derivative (60%), although it could not be properly characterized. With respect to the ortho analogue, only the reduced peptide was detected. Interestingly, the para-I-Phe peptide 1k was successfully reacted under the usual conditions to yield cyclodimer 2k (60%, Table 1, entry 11; see Supplementary Figs 2 and 3). Presumably, the putative monomeric structure 2k′ would be highly strained (molecular models display a nonplanar phenyl ring, see Supplementary Fig. 4) and the process evolves through a ditopic pathway, in a remarkable demonstration of the process versatility.

A detailed nuclear magnetic resonance (NMR) study was performed on dimethylsulphoxide to analyse the conformational behaviour of the peptides synthesized. The NMR spectra of linear sequences 1g–1j were indicative of flexible, unstructured peptides. Hα and 13Cα chemical shifts displayed values typical for random coil, and the NH temperature coefficients and 3J(HαNH) couplings were within the range expected for unstructured peptides47.

Staple bond formation caused a substantial modification of the peptide NMR spectra. Compared with their linear precursors, peptides 2g–2j showed larger Hα chemical shift dispersion (Fig. 3a and Supplementary Figs 5–7), indicating less conformational flexibility. However, broad resonances were observed at 25 °C for some backbone NH protons, suggesting that peptides 2g–2j show some flexibility. Significant differences in 13Cα chemical shifts were observed between stapled peptides 2g–2j and their linear counterparts 1g–1j (Fig. 3b and Supplementary Figs 5–7). The temperature dependence of the amide NH groups was also significantly affected by staple formation (Fig. 3c and Supplementary Figs 5–7). Stapled peptides showed a wider distribution of ΔδT values within the amino-acid sequence and also featured values characteristic of solvent-shielded NH groups (lower than −3 ppb/K). NOE connectivity was also affected by staple formation. In the case of peptides 2h, 2i and 2j, several nonsequential NOEs indicated that the peptides were constrained by intramolecular cyclization (Fig. 3c and Supplementary Figs 6 and 7). Incidentally, we detect all Nindole–H atoms, and also we find the corresponding indole Cα positions consequently substituted, therefore ruling out any interference by N-arylation processes. The aryl–aryl moieties show NOE correlations consistent with a defined single configuration in each case, which matches with the arrangement predicted from molecular modelling.

Figure 3: Peptide NMR spectra comparison between stapled peptide 2i and its linear counterpart 1i.
figure 3

(a) NMR Hα region of peptide 2i and its linear precursor 1i. (b) Plot of the 13Cα chemical shift differences (13Cα Δδcyclic–linear) between stapled peptide 2i and its linear counterpart 1i. (c) Summary of NOE connectivity and temperature coefficients of the NH amide protons (ΔδT) of peptide 1i (bottom left) and 2i (bottom right). The thickness of the bars reflects the intensity of the NOEs, that is, weak (), medium () and strong (▪). I-F: m-iodophenylalanine.

Moreover, circular dichroism measurements of stapled peptides 2g and 2h were performed in order to detect evidence of secondary structure and were then compared with their linear precursors 1g and 1h, respectively (See Supplementary Figs 8–11). Stapled peptides 2 show a positive maximum at 190 nm and minimum peaks at 206 nm, which indicate some levels of structuring. In contrast, linear peptides exhibit a more flattened profile typical of a flexible unfolded structure.

Extension to biologically relevant peptides

To further explore the potential of the methodology, we planned to make the stapled analogues of known linear bioactive peptides, and at the same time test the influence of new features, such as the presence of proline (Pro) in the sequence, the overall length of the chain (up to nine amino acids) and the possibility of performing the C–H arylation on the solid phase, which enables the transformation of unprotected N-terminal peptides. In this way, we prepared the stapled version of an active valorphin analogue48, which is a potent dipeptidyl peptidase III inhibitor, displaying Pro in a five amino-acid sequence containing also Trp and Tyr (see Supplementary Fig. 12). Interestingly, this peptide is closely related to spinorphin, an endogenous antinociceptive peptide, a potent and noncompetitive antagonist at the ATP-activated human P2X3 receptor49. The SPPS method was used to synthesize the Fmoc-protected sequence, which was intramolecularly arylated on resin. Subsequent N-terminal deprotection and cleavage, successfully afforded the stapled peptide 2l (Fig. 4a). This solid-phase protocol is fully compatible with the Pd chemistry involved and, interestingly, allows the preparation of arylated sequences having unprotected terminal amino groups. Futhermore, baratin50, a neurostimulating peptide, was stapled following a related procedure. In this way, the standard SPPS protocol was interrupted to perform the on-resin C–H activation step, to be continued with the incorporation of the final four amino acids. Afterwards, deprotection and cleavage afforded the desired derivative 2m (Fig. 4a, see Supplementary Methods and Supplementary Fig. 12).

Figure 4: Bioactive stapled peptides, biochemical and cellular studies.
figure 4

(a) Structure of stapled peptides 2l (valorphin analogue) and 2m (baratin analogue). (b) Proteolytic degradation assay of stapled peptides 2g and 2h and their linear precursors 1g and 1h. (c) Structure of labelled-stapled peptide 2j Bodipy (left) and the corresponding confocal microscopy image of SH-SY5Y cells treated with compound 2j Bodipy (750 nM). Scale bar, 25 μm (right).

Next, we determined the proteolytic stability of a couple of meaningful stapled peptides, in comparison with their respective linear counterparts. We followed a chymotrypsin-based protocol, previously used to evaluate stapled peptides51. Plotting the HPLC-MS profiles of the couples 2g/1g and 2h/1h (Fig. 4b, also Supplementary Figs 13–15) clearly shows an almost total protection of the stapled peptides towards enzymatic degradation, after 5–6 h, whereas the linear precursors suffered a rapid hydrolytic cleavage to remove the N-terminal amino acid, and completely disappeared after this time lapse.

Finally, we studied the controlled labelling of the stapled peptides. The goal was to selectively attach a fluorophore to the C-terminal amino acid to trace cellular permeabilization and localization. Thus, we linked a Bodipy residue to cyclopeptide 2j through an amino spacer (the novel Bodipy construct was designed and prepared for this purpose, see Supplementary Fig. 16 and Supplementary Methods), to get the desired labelled-stapled peptide 2j Bodipy (Fig. 4c) in a convenient manner. This compound exhibits low cytotoxicity at 750 nM after 24 h of incubation in SH-SY5Y cells (MTT assay; see Supplementary Fig. 17), and this concentration was used for the cell penetration studies. In a preliminary flow cytometry experiment (fluorescence-activated cell sorting assay), SH-SY5Y cells resulted in brightly stained sections on incubation for 30 min (see Supplementary Fig. 18). The cells treated in this manner were analysed using confocal microscopy. The stapled peptide 2j Bodipy was localized both in the membrane and in the cytoplasmatic region (Fig. 4c). Incidentally, the Bodipy-labelled linear precursor 1j Bodipy also penetrates into cells in a comparable extent; however, there are appreciable differences in toxicity and cell permeability with respect to the stapled analogue (see Supplementary Figs 19 and 20). Overall, these results enable the performance of systematic bioimaging studies on these compounds.

Intermolecular peptide conjugation through C–H activation

Chemical conjugation can effectively link drugs to carriers, and this technique is routinely used to modify biologics, especially antibodies with therapeutic indications52. We next explored peptide–peptide conjugation via C–H activation to achieve bismacrocyclic peptide constructs linked through a nonhydrolysable bond.

Peptides that inhibit the new blood vessel growth (angiogenesis) have become a promising tool for treating cancer. In this context, it has been reported that cyclic forms of NGR-containing sequences (tumour-homing peptides) lead to an improvement of the anticancer activity of an associated drug53. We selected the previously studied Asn-Gly-Arg (NGR) array and synthesized the corresponding Trp-containing macrocycle 3 to be conjugated to a synthetic p-I-Phe-cyclopeptide derivative of the cyclic depsipeptide sansalvamide A (4, Fig. 5a), a natural product whose synthetic analogues have demonstrated significant anticancer activity54. Using a standard Pd-catalysed reaction in an aqueous medium of PBS–DMF, the two partners were successfully coupled to yield the C–C conjugate 5 (Fig. 5a).

Figure 5: Macrocyclic conjugation via C–H activation.
figure 5

(a) Intermolecular conjugation of NGR cyclopeptide 3 with the sansalvamide derivative 4 via C–H activation. (b) Double conjugation of the NGR cyclopeptide 3 with 1,4-diodobenzene.

In a second example, the conjugation of two units of the NGR Trp-containing cyclopeptide 3 with a 1,4-diiodobenzene connector via a double C–H arylation was achieved. Again, the process yielded the expected bis-NGR-adduct 6 in a single step (Fig. 5b).

To evaluate the effects of macrocyclic conjugation, the NMR spectra of conjugated peptide 5 and its cyclic precursors 3 and 4 were compared (see Supplementary Fig. 21). After conjugation, small shifts were observed in the Hα resonances of the sansalvamide A analogue 4. In contrast, the Hα chemical shifts of the NGR-containing cycle were almost identical in peptides 5 and 3. Comparison of 13Cα chemical shifts showed that in both cycles the larger 13Cα chemical shift change corresponded to the residue involved in the intermolecular bond formation: Trp (1.2 p.p.m.) in the NGR-containing cycle and Phe (0.6 p.p.m.) in the sansalvamide A derivative. More modest changes were observed in the 13Cα of the remaining amino acids. The resemblance of the Hα and 13Cα chemical shifts, which are highly sensitive to conformational changes, suggests that conjugation provided a minimal perturbation of the overall structures of peptides 3 and 4. The NMR analysis was extended to peptide 6 and its precursor 3. The effects of conjugation in the Hα and Cα resonances are shown in Supplementary Fig. 22.

Unfortunately, in preliminary biological studies evaluating the conjugated peptide 5 and its macrocycle precursors 3 and 4 against several cancer cell lines, only the cyclopeptide 4 showed activity (IC50<10 μM). Nevertheless, the protocol seems general and offers new possibilities for peptide-based conjugation.

Stapled cyclopeptides

We then explored access to complex topologies, focusing on the synthesis of bicyclic peptide chemotypes. In a first approach, the synthesis of a stapled cyclopeptide was attempted by promoting the usual C–H arylation through an intramolecular Trp-I–Phe interaction. Preliminary studies showed that alanine (Ala) hexapeptides containing a m-I-Phe and Trp units, respectively, placed at positions i−i+2 and i−i+3 afforded the corresponding macrocycles after routine amide coupling in quantitative yields; however, the subsequent intramolecular C–H arylations did not take place under the standard conditions, probably because of a highly restricted conformation. Remarkably, when the stapling on the linear sequence anchored to the resin 7 was carried out first on the solid phase, we obtained the corresponding NH-free amino terminal-stapled peptide 9 after cleavage from resin. Finally, we isolated the desired bicyclic compound 10 by amide cyclization (Fig. 6a,b). Furthermore, in this way, we overcame the relative limitation of irreversible protection of the N-terminal amino group, as previously reported37. Incidentally, preliminary results showed that increasing the Pd(OAc)2 amount up to 0.2 eq. gave acceptable intermolecular arylations for NH-free amino terminal sequences (data not shown). The stapled cyclopeptide 10 was prepared in a suitable manner, involving a solid-phase arylation, thus facilitating the purification step, followed by a routine peptide coupling. This protocol may enable the synthesis of further derivatives of this attractive and unexplored structural class.

Figure 6: Synthesis of macrobicyclic peptides 10 and 12.
figure 6

(a) Synthesis of macrobicyclic peptide 10 through solid-phase stapling. Reaction conditions: (i) Pd(OAc)2 (0.05 eq.), AgBF4 (1.0 eq.), 2-NO2BzOH (1.5 eq.), DMF, MW 90 °C, 20 min; (ii) (1) 1% sodium diethyldithiocarbamate (DDC) in DMF. (2) Piperidine–DMF (1:4; 1 × 1 min, 2 × 5 min). (3) TFA-TIS-H2O (95:2.5:2.5), r.t, 1 h; (iii) PyAOP (2.0 eq.), DIEA (6.0 eq.), DMF, r.t, 1.5 h. (b) Minimized geometry of compound 10 generated by the Spartan ’14 suite. Hydrogens omitted for clarity. (c) Double C–H arylation to cyclic biaryl 12 (25% conversion, estimated by HPLC). Reaction conditions: (iv) 40 mol %Pd(OAc)2, AgBF4 (6.0 eq.), pivalic ac. (1.5 eq.), DMF, MW 90 °C, 20 min. (d) Minimized geometry of compound 12 generated by Spartan ‘14 suite showing the diagnostic NOE correlations (blue arrows).

The 1H NMR spectrum of the linear stapled peptide 9 was characterized by a large chemical shift dispersion of the NH and Hα resonances. Bicyclopeptide 10 showed a slightly larger chemical shift range for the Hα protons, as expected for a more rigid and structured peptide. Backbone cyclization was also reflected in a large splitting of the methylene Hα atoms of Gly-1 and Hβ atoms of Phe. In the case of peptide 9, the chemical shift difference between the two geminal protons was <0.2 (Hα, Gly-1) and 0.1 (Hβ, Phe) p.p.m., whereas for peptide 10 this difference increased to >0.9 (Hα, Gly-1) and >0.7 (Hβ, Phe) p.p.m. Backbone cyclization was further evidenced by significant changes in 13Cα chemical shifts and in the temperature coefficients of amide NH (Supplementary Fig. 23).

Double stapling of linear peptides

Biaryl bismacrocyclic peptide-derived natural products such as vancomycin55 and complestatin56 display very interesting bioactivity profiles and as synthetic targets they are extremely difficult to prepare57. Thus, the development of new methodologies to access simplified scaffolds is crucial to enable structural diversification, thereby allowing practical medicinal chemistry and biological studies. Hence, it was envisioned that intramolecular double C–H arylation of a sequence containing a diiodinated Tyr (commercially available) flanked by two Trp units would give raise to bicyclic peptide topologies with adjancent biaryl moieties in a straightforward manner. In order to establish the conditions for this transformation, we performed some preliminary experiments where the intermolecular C–H activation of a Ac-m,m′-I,I-Tyr(OAc)-OH unit with 2.0 eq. of Ac-Trp-OH was tested. An increase in the amount of the Pd catalyst (40%) and AgBF4 (6.0 eq.)58 and use of a mild excess of pivalic acid resulted in a productive reaction (see Supplementary Methods). Next, we designed the linear peptide sequence 11 (Fig. 6c), which was synthesized in a straightforward manner on the solid phase. Using the previous C–H activation conditions for the diodoTyr, we successfully obtained the double stapled peptide 12 directly from the corresponding linear precursor in one step in a 25% HPLC conversion, together with monoarylated cycles and dehalogenated derivatives, which were probably produced in competitive processes (Fig. 6c). This remarkable result is the proof of principle that even these complex peptidic topologies are accessible through the present methodology. The 1H NMR spectrum of the double stapled peptide 12 is characterized by a wide chemical shift range for the NH and Hα protons (see Supplementary Fig. 134). The observed pattern of NOEs, summarized in Fig. 6d, indicates that each side of the Tyr aromatic ring faces a distinct Trp-Ala-Gly motif.

As an optimization of this methodology, we have developed the exclusive use of standard proteinogenic amino acids (Tyr), which, once coupled, are modified in situ, thus simplifying the protocol and rendering it considerably more affordable. In this way, the corresponding linear precursor of peptide 11 can be obtained through routine amide couplings followed by on-resin iodination of the Tyr residue to yield derivative 11 (41% conversion, unoptimized, see Supplementary Fig. 24)59. This remarkable result enables the direct synthesis of bicyclopeptide 12 to be performed on solid-phase from commercially available reagents in a single sequence.

Discussion

Metal-catalysed arylations through C–H activation are suitable processes to gain access to minimalistic staples (two-electron) in relevant Trp-containing peptide sequences directly from Trp and iodo-phenylalanine (-tyrosine) precursors. The validation of the methodology includes the analysis of the scope of the transformation, compatibility with other amino acids and the applicability to SPPS. This approach has also been applied to constrain biologically active signalling sequences. In an intermolecular mode, the process efficiently links two Trp peptides to a benzene connector and is also useful to conjugate peptides through a C–C bond. All compounds showed a structured nature, as revealed by spectroscopic characterization. Finally, we have developed a simple protocol for the straightforward access to novel peptide topologies such as dimeric macrocycles, stapled bicyclopeptides and biaryl–biaryl species (see Supplementary Figs 4 and 25) from the corresponding linear precursors in only one step. These findings open up general access to a variety of novel constrained peptidic chemotypes, and we believe that this breakthrough will make a significant contribution to the development of a broad range of applications for peptides in biological and medicinal chemistry.

Methods

General

For abbreviations and detailed experimental procedures see Supplementary Methods. For NMR analysis of the compounds, see Supplementary Figs 26–148 and Supplementary Tables 1–40.

Peptide synthesis

All peptides were manually synthesized on a 2-Chlorotrityl, H-Rink-Amide Chemmatrix or TentaGel S NH2 resin using standard Fmoc chemistry for SPPS.

Linear peptide cyclization

The free-amine free-acid linear peptide (1.0 eq.) was dissolved in ACN/DMF or DMF (0.001–0.003 M) and N,N-diisopropylethylamine (DIEA) (6.0 eq.) and the corresponding coupling agents (1.5–3.0 eq., benzotriazol-1-yl-oxytripyrrolidinophosphonium hexafluorophosphate (PyBOP) with hydroxybenzotriazole (HOBt) or 1-[Bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-b]pyridinium 3-oxid hexafluorophosphate (HATU) and O-(benzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium tetrafluoroborate (TBTU) or (7-azabenzotriazol-1-yloxy)tripyrrolidinophosphonium hexafluorophosphate (PyAOP)) were added. The solution was stirred at r.t until the cyclization was complete (1–3 h). Workup was performed by extraction with aqueous solutions of NH4Clsat and NaHCO3sat. Organic layers were combined, dried over anhydrous sodium sulfate, filtered and concentrated under vacuum. When remaining protecting groups were present, the macrocycle was treated with a 95% TFA, 2.5% triisopropylsilane (TIS) and 2.5% H2O cocktail (3 h), washed with Et2O, dissolved in ACN:H2O and lyophilized to furnish the corresponding deprotected peptide. When necessary, the macrocycle was purified by flash column chromatography on silica gel or semipreparative RP-HPLC.

General protocol for the stapled bond formation in solution

The linear peptide (50 mg), AgBF4 (2.0 eq.), trifluoroacetic acid (1.0 eq.) and Pd(OAc)2 (0.05 eq.) were placed in a MW reactor vessel in DMF (1.2 ml). The mixture was heated under MW irradiation (250 W) at 90 °C for 20 min. The residue was filtered and purified using semipreparative RP-HPLC. This process was scaled up to 0.907 mmol of peptide, affording the stapled/locked peptides 2g2k in isolated yields ranging from 1 to 32%.

Procedure for the single macrocycle conjugation

Cyclopeptide 4 (40.0 mg, 0.056 mmol), cyclopeptide 3 (55.3 mg, 0.084 mmol, 1.5 eq.), AgBF4 (43.8 mg, 0.225 mmol, 4.0 eq.), pivalic acid (5.7 mg, 0.056 mmol, 1.0 eq.) and Pd(OAc)2 (1.4 mg, 0.077 mmol, 0.1 eq.) were placed in a MW reactor vessel in 2 ml of PBS:DMF (1:1). The mixture was heated under MW irradiation (250 W) at 90 °C for 20 min. The irradiation cycle was repeated after adding a new portion of Pd(OAc)2 and AgBF4. The residue was filtered and partially purified in a PoraPak Rxn reverse phase column (1.63 mg, 2% estimated using HPLC-MS). A pure fraction was obtained using analytic RP-HPLC to yield pure conjugate 5.

Procedure for the double macrocycle conjugation

1,4-diiodobenzene (35 mg, 0.106 mmol), macrocycle 3 (209 mg, 0.318 mmol, 3.0 eq.), AgBF4 (124 mg, 0.637 mmol, 6.0 eq.), pivalic acid (16.3 mg, 0.159 mmol, 1.5 eq.) and Pd(OAc)2 (9.5 mg, 0.042 mmol, 0.4 eq.) were placed in a MW reactor vessel in 2 ml of PBS:DMF (1:1). The mixture was heated under MW irradiation (250 W) at 90 °C for 20 min. The crude product was filtered, and the workup was carried out by washing with AcOEt and then precipitating by adding ACN to the aqueous phase. The resulting precipitate was washed with ACN, decanted and dried, obtaining 159 mg of crude product (pale solid, 42% estimated using HPLC-MS). A pure fraction was obtained with semipreparative RP-HPLC to yield pure conjugate 6.

Typical procedure for the stapled cyclopeptide formation on the solid phase

Once sequence 7 was synthesized on a TentaGel S NH2 resin via an AB linker, the peptide anchored to the resin (139 mg, 0.145 mmol), AgBF4 (28 mg, 0.144 mmol, 1.0 eq.), 2-nitrobenzoic acid (36 mg, 0.215 mmol, 1.5 eq.) and Pd(OAc)2 (1.6 mg, 7.1 μmol, 0.05 eq.) in DMF (2 ml) was placed in a MW reactor vessel. The mixture was heated under MW irradiation (250 W) at 90 °C for 20 min. Eight more batches were carried out following the same procedure and were then combined. The resin was treated with 1% DDC in DMF and the Fmoc group was removed. The peptide was cleaved from the resin with a 95% TFA, 2.5% TIS and 2.5% H2O cocktail (1 h), yielding the stapled sequence 9 (85% purity, estimated using HPLC-MS). Finally, the stapled peptide 9 was cyclized (see above for standard cyclization procedure) and the crude product was purified using semipreparative RP-HPLC to yield bicyclopeptide 10 (pale solid, 18% unoptimized).

One-step double arylation to biaryl–biaryl stapled peptide

Linear peptide 11 (50 mg, 0.044 mmol), AgBF4 (51 mg, 0.262 mmol, 6.0 eq.), pivalic acid (6.7 mg, 0.066 mmol, 1.5 eq.) and Pd(OAc)2 (3.9 mg, 0.018 mmol, 0.4 eq.) were placed in a MW reactor vessel in DMF (500 μl). The mixture was heated under MW irradiation (250 W) at 90 °C for 20 min. Three more batches were carried out following the same procedure and then filtered and combined (25% conversion, estimated using HPLC-MS). A pure fraction of bicyclopeptide 12 was isolated using semipreparative RP-HPLC.

Additional information

How to cite this article: Mendive-Tapia, L. et al. New peptide architectures through C–H activation stapling between tryptophan–phenylalanine/tyrosine residues. Nat. Commun. 6:7160 doi: 10.1038/ncomms8160 (2015).