Introduction

Cycloaddition reaction is one type of bimolecular reaction in which two conjugated reactants react to yield cyclic products1. As a typical cycloaddition reaction, azide–alkyne reaction has been extensively investigated and utilized. However, cycloadditions involving azides are limited by their reaction speeds (impractically slow at ambient temperature) and low regioselectivities of the products. Although high temperature can speed up the reaction, the mixtures of 1,4- and 1,5-disubstituted triazoles are formed in the reaction when the reactants are alkynes2. Therefore, many catalysts including copper(I) and ruthenium(II) have been exploited to quicken the reaction and improve product selectivity3. However, these metal ions are not biologically friendly, which limits further applications of the azide–alkyne cycloaddition reaction in vivo. Thus, suitable conditions for this reaction which can yield regionally selective products in the absence of metal ions at room temperature have been continuously explored, and some progresses have been made (e.g., photocatalysis and strain-promoted cycloadditions)4,5,6.

Self-assembly is a spontaneous process driven by thermodynamic and kinetic7. It is achieved (or influenced) by the synergistic effect of various non-covalent interactions, such as hydrogen bonding, π-π stacking, electrostatics, hydrophobicity, and van der Waals force8. The thermodynamic stability and the state of minimum energy of the ultimately formed nanostructures are determined by the synergistic effect of these non-covalent interactions9,10. Under the non-covalent interactions, small building blocks spontaneously and hierarchically assemble into functional supramolecular materials11,12. For example, artificial enzyme formed by self-assembly can function as mimic peroxidase13. Sureshan and coworkers reported topochemical azide–alkyne cycloaddition reaction using assembly of several peptide and sugar derivatives in crystal state14. However, crystallization often has some limitations such as low yield or unsuitability for large-scale synthesis. Alternatively, they utilized supramolecular self-assembly (i.e., hydrogelation) to achieve metal-free azide–alkyne cycloaddition under heating at high temperature15,16. Thus, directly and exclusively applying supramolecular self-assembly or its non-covalent reactions to realize a specific selective chemical reaction remains a great challenge. Specifically, using supramolecular self-assembly to drive metal-free azide–alkyne cycloaddition reaction without the assistance of any other beneficial factors (e.g., heating) has not been reported.

Based on above-mentioned literature research, we are motivated to employ a typical supramolecular self-assembly system to realize a metal-free azide–alkyne cycloaddition reaction. Thus, we designed the self-assembling monomers Nap-Phe-Phe-Lys(azido)-OH (Nap-FFK-Azi) and Nap-Phe-Phe-Gly(alkynyl)-OH (Nap-FFG-Alk), as well as their cycloaddition product Nap-Phe-Phe-Lys(triazole)-Gly-Phe-Phe-Nap (Nap-FFK-Tria-GFF-Nap) (Fig. 1a). We speculate that, in aqueous solution, owing to the assembling ability of Nap-Phe-Phe (Nap-FF) moiety, Nap-FFK-Azi or Nap-FFG-Alk could spontaneously self-assemble into ordered nanostructures (always nanofibers). Along with the assembling process, Nap-FFK-Azi or Nap-FFG-Alk is drawn near to the nanofiber by the non-covalent interactions, driving the cycloaddition reaction to yield Nap-FFK-Tria-GFF-Nap (Fig. 1b). Hypothesis of the assembly-driven cycloaddition reaction process is summarized in Fig. 1c. Firstly, in the presence of two self-assembling monomers Nap-FFK-Azi and Nap-FFG-Alk, slim nanofibers are initiatively formed among the irregular aggregates through non-covalent interactions. Secondly, the distance between Nap-FFK-Azi and Nap-FFG-Alk gradually narrows, allowing effective molecule collision to realize azide–alkyne cycloaddition. Thirdly, accumulated product Nap-FFK-Tria-GFF-Nap transforms the slim nanofibers into wider ones (nanoribbons in this work).

Fig. 1: Regionally selective azide–alkyne cycloaddition reaction driven by assembly.
figure 1

a Chemical structures of Nap-FFK-Azi, Nap-FFG-Alk and Nap-FFK-Tria-GFF-Nap. b Diagram of the regionally selective azide–alkyne cycloaddition reaction driven by assembly. c Schematic illustration of the regionally selective azide–alkyne cycloaddition reaction driven by assembly.

Full size image

Results

Syntheses and characterizations of hydrogels

We first synthesized the three compounds Nap-FFK-Azi, Nap-FFG-Alk, and Nap-FFK-Tria-GFF-Nap. Nap-FFK-Azi and Nap-FFG-Alk were synthesized with solid phase peptide synthesis (SPPS), purified with high-performance liquid chromatography (HPLC), and then characterized with ESI-MS and 1H-NMR and 13C-NMR spectra (Supplementary Figs. 1–8). Nap-FFK-Tria-GFF-Nap was obtained from copper-catalyzed azide–alkyne cycloaddition (CuAAC) reaction (Fig. 2a). During the reaction, the mixture was injected into a HPLC system for analysis (Supplementary Table 1). As Fig. 2b shows, besides the desired product of Nap-FFK-Tria-GFF-Nap (retention time 20.3 min, 73.8%, Supplementary Figs. 9–11), the by-product 1,5-isomer of Nap-FFK-Tria-GFF-Nap (retention time 16.1 min, 26.2%, Supplementary Table 2, Supplementary Figs. 12 and 13) also appeared on the HPLC trace, suggesting poor regioselectivity of the copper-catalyzed azide–alkyne cycloaddition reaction.

Fig. 2: Cooper-catalyzed azide–alkyne cycloaddition (CuAAC) between Nap-FFK-Azi and Nap-FFG-Alk.
figure 2

a Synthetic route for Nap-FFK-Tria-GFF-Nap. Conditions: 20 mM Nap-FFK-Azi, 20 mM Nap-FFG-Alk, N,N-dimethylformamide (DMF), copper(I), 25 °C, and 6 h. b HPLC traces of Nap-FFK-Azi (green), Nap-FFG-Alk (blue), and the CuAAC reaction mixture in a (red). Wavelength for detection: 254 nm.

Full size image

To check the self-assembling ability of the three compounds, we measured their critical aggregation concentrations (CACs) in phosphate-buffered saline (PBS, pH 9). We chose this condition because these oligopeptides are hardly dissolved under pH 9. As shown in Supplementary Fig. 14, the CACs of Nap-FFK-Azi, Nap-FFG-Alk, and Nap-FFK-Tria-GFF-Nap were measured to be 40.6 μM, 263.8 μM, and 106.8 μM, respectively, suggesting Nap-FFK-Azi has the highest self-assembling ability among the three compounds. As we know, in aqueous solution, supramolecular hydrogelator containing Nap-FF self-assembling moiety always self-assembles into nanofibers to form macroscopic hydrogel17,18. In this work, we found that Nap-FFK-Azi and Nap-FFG-Alk hydrogels could be obtained via a heating-cooling method. As we mentioned above, high temperature can speed up the azide–alkyne cycloaddition reaction. Thus, to avoid the occurrence of the cycloaddition reaction between Nap-FFK-Azi and Nap-FFG-Alk while in the meantime assure the occurrence of their hydrogelations, we heated these two compounds at 100 °C in PBS and used HPLC to trace their cycloaddition product (i.e., Nap-FFK-Tria-GFF-Nap) at different times. The results in Supplementary Fig. 15 shows that 1 h heating of the two compounds at 2.5 mM did not yield the product Nap-FFK-Tria-GFF-Nap while 2 h heating did. Note that these oligopeptides are hardly dissolved in aqueous solutions at concentrations higher than 2.5 mM under above heating conditions. Thus, a hydrogelator concentration of 2.5 mM, as well as 1 h heating at 100 °C was chosen as the condition for following hydrogelation experiments. After obtaining the Nap-FFK-Azi and Nap-FFG-Alk hydrogels with heating-cooling method (heating at 100 °C for 1 h and cooling at 4 °C for 24 h, compound concentration: 2.5 mM), we used dynamic rheological measurements to evaluate their physical properties. As shown in Supplementary Figs. 16 and 17, the dynamic storage modulus (G′) of these two hydrogels were apparently higher than their corresponding loss modulus (G″), indicating both hydrogels can endure external shear force. However, upon the same heating-cooling treatment, 2.5 mM Nap-FFK-Tria-GFF-Nap remained a clear solution, indicating the compound might have self-assembled but could not form a hydrogel. Interestingly, when we applied above hydrogelation method on the mixture solution of Nap-FFK-Azi and Nap-FFG-Alk at 2.5 mM, a transparent hydrogel was obtained. And the Nap-FFK-Azi + Nap-FFG-Alk hydrogel had much higher G′ and G′′ values than above two hydrogels, respectively (Supplementary Fig. 18). Because the CAC value of Nap-FFK-Azi is much lower than that of Nap-FFG-Alk, we hypothesized that, in the Nap-FFK-Azi + Nap-FFG-Alk hydrogel, Nap-FFK-Azi precede Nap-FFG-Alk in self-assembling state. And HPLC analysis indicated that, at this hydrogelation condition, the cycloaddition product Nap-FFK-Tria-GFF-Nap was barely seen in the Nap-FFK-Azi + Nap-FFG-Alk hydrogel (Supplementary Fig. 19).

Monitoring of assembly-driven azide–alkyne cycloaddition reaction

To test our hypothesis of assembly-driven azide–alkyne cycloaddition, we used HPLC to analyze the chemical composition of the Nap-FFK-Azi + Nap-FFG-Alk hydrogel during its aging at 4 °C. As shown in Fig. 3a, after 7 d aging, desired 1,4-disubstituted cycloaddition product Nap-FFK-Tria-GFF-Nap (HPLC retention time 20.3 min, 14.7%, Supplementary Fig. 20 and Supplementary Table 3) was yielded in the hydrogel. Interestingly, the 1,5-isomer of Nap-FFK-Tria-GFF-Nap, which was also another product of the CuAAC between Nap-FFK-Azi and Nap-FFG-Alk (Fig. 2b), was not found in the hydrogel (Fig. 3a), suggesting excellent regioselectivity of this assembly-driven cycloaddition reaction. We noticed that two small shoulder peaks (17.5 min and 18.7 min) in the HPLC trace of Nap-FFK-Azi + Nap-FFG-Alk hydrogel (Fig. 3a). ESI-MS analyses indicated that they were the same substances as those at 17.3 min and 18.6 min, respectively (Supplementary Figs. 21 and 22). Above peak splitting may owe to the sample being prepared under alkaline conditions but tested in an acidic environment. The reaction was also monitored under different pH values ranging from 9 to 11 and concentrations ranging from 0.625 mM to 2.5 mM, and we found that pH variations affected the cycloaddition reaction very tiny while concentration did much. As shown in Supplementary Fig. 23, at pH 9–11, the reaction shared vicinal yields of 6.63–6.91%. In contrast, when the concentration was increased from 0.625 mM to 2.5 mM, the corresponding yield increased from 1.53% to 6.91% (Supplementary Figs. 23 and 24), indicating that the assembly-driven azide–alkyne cycloaddition reaction is concentration-dependent. We further monitored the reaction in the hydrogel till 36 d using HPLC analysis. As displayed in Supplementary Fig. 25, the reaction yield increased gradually and reached its plateau of 19.1% at 14 d.

Fig. 3: Characterizations of assembly-driven azide–alkyne cycloaddition reaction.
figure 3

a HPLC traces of 2.5 mM Nap-FFK-Azi (green), 2.5 mM Nap-FFG-Alk (blue), 2.5 mM Nap-FFK-Tria-GFF-Nap (red), 2.5 mM Nap-FFK-Azi + Nap-FFG-Alk hydrogel at 4 °C for 7 d (orange), and 2.5 mM Nap-FFK-Azi + Nap-FFG-Alk in DMF at 4 °C for 7 d (black). Wavelength for detection: 254 nm. b TEM images of 2.5 mM Nap-FFK-Azi hydrogel, 2.5 mM Nap-FFG-Alk hydrogel, 2.5 mM Nap-FFK-Tria-GFF-Nap solution, and 2.5 mM Nap-FFK-Azi + Nap-FFG-Alk hydrogel after aging at 4 °C for 7 d. scale bar = 500 nm. c CD spectra of the samples in (b).

Full size image

As we know, those necessary conditions for supramolecular self-assembly in aqueous solution (e.g., hydrophobic interaction, π-π stacking, hydrogen bonding, etc) do not exist in organic solvent such as N,N-dimethylformamide (DMF). Thus, to validate that above product Nap-FFK-Tria-GFF-Nap was yielded under assembly-induced space confinement effect, we incubated 2.5 mM Nap-FFK-Azi + Nap-FFG-Alk in DMF at 4 °C for 7 d and injected the mixture for HPLC analysis. Clearly, when Nap-FFK-Azi and Nap-FFG-Alk remained monomeric state in DMF, no cycloaddition product of Nap-FFK-Tria-GFF-Nap was found in the incubation (Fig. 3a), indicating that the product in above Nap-FFK-Azi + Nap-FFG-Alk hydrogel was indeed yielded under assembly effect. In order to eliminate the possibility that the solvent of Nap-FFK-Azi + Nap-FFG-Alk hydrogel (i.e., PBS) could induce the cycloaddition reaction between Nap-FFK-Azi and Nap-FFG-Alk, we added 2.5 mM Lys(azido)-OH (K-Azi) and Gly(alkynyl)-OH (G-Alk) to PBS, and applied mixture solution to above heating-cooling procedure. Without the self-assembling motif Nap-FF, neither K-Azi nor G-Alk could self-assemble into nanofibers in PBS. As expected, we did not find any fibrous structure from the transmission electron microscopy (TEM) image of above PBS solution (Supplementary Fig. 26). HPLC analyses consistently showed that, up to 14 d aging at 4 °C, no cycloaddition product between K-Azi and G-Alk (i.e., Lys(traizole)Gly, K-Tria-G) was found in above mixture PBS solution (Supplementary Figs. 27–30). Above results collectively indicate that, it is assembly but not the solvent that drives the metal-free azide–alkyne cycloaddition reaction to yield the regionally selective product.

We further used TEM observations to validate above assemble-driven cycloaddition in Nap-FFK-Azi + Nap-FFG-Alk hydrogel. As shown in Fig. 3b, after 7 d aging, while dense and slim nanofibers were observed in the TEM image of Nap-FFK-Azi hydrogel (average diameter: ~20 nm), helical ribbons with an average diameter of 40 nm were found in the TEM image of Nap-FFG-Alk hydrogel. In the TEM of Nap-FFK-Tria-GFF-Nap solution after same treatment, helical ribbons with an average diameter of 110 nm presented. Interestingly, in the TEM image of Nap-FFK-Azi + Nap-FFG-Alk hydrogel after 7 d aging, besides slim nanofibers, helical ribbons with an average diameter around 130 nm were clearly observed (Fig. 3b). To rule out the possibility that co-assembly of the two precursor peptides also result in the final nanoribbons, a nonreactive control Nap-Phe-Phe-Nva-OH (Nap-FF-Nva, by replacing the alkyne group of Nap-FFG-Alk with an alkyl group) was synthesized for comparative study (Supplementary Figs. 31–33). Cycloaddition cannot take place between Nap-FFK-Azi and Nap-FF-Nva as expected (Supplementary Fig. 34), while the possibility of co-assembly among them remains. Upon the same heating-cooling treatment, both 2.5 mM Nap-FF-Nva and 2.5 mM Nap-FFK-Azi + Nap-FF-Nva could form stable and transparent hydrogels (Supplementary Figs. 35 and 36). As shown in Supplementary Fig. 37, at day 7, similar nanoribbons to those in Nap-FFG-Alk hydrogel with an average diameter of 125 nm were also observed in Nap-FF-Nva hydrogel. However, only slim nanofibers (average diameter: ~18 nm) but not nanoribbons were found in Nap-FFK-Azi + Nap-FF-Nva hydrogel. Above observations indicated that co-assembly of the peptides most likely lead to the formation of nanofibers rather than nanoribbons in Nap-FFK-Azi + Nap-FF-Alk hydrogel. Since the assemblies of Nap-FFK-Tria-GFF-Nap were significantly wider than those of Nap-FFK-Azi or Nap-FFG-Alk, from the TEM image of Nap-FFK-Azi + Nap-FFG-Alk hydrogel we knew that cycloaddition reaction between Nap-FFK-Azi and Nap-FFG-Alk indeed occurred in the hydrogel.

To further validate the nanoribbons in Nap-FFK-Azi + Nap-FFG-Alk hydrogel were in accordance with those in Nap-FFK-Tria-GFF-Nap solution, we used circular dichroism (CD) to investigate the secondary structures of the assemblies in above four samples. As displayed in Fig. 3c, CD spectrum of Nap-FFK-Azi hydrogel exhibited a negative peak at 197 nm and a positive peak at 207 nm, indicating its nanofibers adopt β-turn-like structures. In contrast, a positive peak at 195 nm and a negative peak at 215 nm appeared in the CD spectrum of Nap-FFG-Alk hydrogel, suggesting the helical nanoribbons in the hydrogel adopt β-sheet-like structures. A positive peak at 197 nm and a negative peak at 217 nm appeared in the CD spectrum of Nap-FFK-Tria-GFF-Nap solution, suggesting β-sheet-like structures of its nanoribbons. Interestingly, a positive peak at 199 nm and a negative peak at 222 nm, both of which are closer to those of Nap-FFK-Tria-GFF-Nap solution than to those of Nap-FFG-Alk hydrogel, appeared in the CD spectrum of Nap-FFK-Azi + Nap-FFG-Alk hydrogel, suggesting the nanoribbons in the hydrogel are made of Nap-FFK-Tria-GFF-Nap. We noted that the CD spectra of these four samples all showed positive/negative peaks at 235 nm, which should be assigned to the different conformation of their chiral phenylalanine residues19. The negative peak at 235 nm of Nap-FFK-Azi hydrogel was induced by its β-turn-like nanofibers, while the positive peaks at 235 nm in other three groups were induced by their corresponding β-sheet-like nanostructures. We also performed CD analyses of Nap-FF-Nva hydrogel and Nap-FFK-Azi + Nap-FF-Nva hydrogel. CD spectrum of Nap-FF-Nva hydrogel exhibited a positive peak at 199 nm and a negative peak at 218 nm, indicating a β-sheet-like structure of the nanoribbons in the hydrogel. This observation was consistent with that of the Nap-FFG-Alk hydrogel. For Nap-FFK-Azi + Nap-FF-Nva hydrogel, two negative peaks at 194 nm and 222 nm and a positive peak at 208 nm appeared in its CD spectrum (Supplementary Fig. 38), suggesting a complex structure rather than a single secondary structure in the hydrogel. We believe this complex CD spectrum is more likely to be a spectral superposition. In contrast, Nap-FFK-Azi + Nap-FFG-Alk hydrogel presented a distinct secondary structure that is similar to that of Nap-FFK-Tria-GFF-Nap assemblies (Fig. 3c). Taken together, these results additionally confirmed that cycloaddition reaction between Nap-FFK-Azi and Nap-FFG-Alk indeed occurred in Nap-FFK-Azi + Nap-FFG-Alk hydrogel.

Mechanism of assembly-driven azide–alkyne cycloaddition reaction

To validate the mechanism of assembly-driven regionally selective azide–alkyne cycloaddition reaction shown in Fig. 1c, we used TEM observations to real-time monitor the formation of above nanostructures. TEM images in Fig. 4a show that, at 0.5 h, those raw materials at nanoscale for self-assembling into bigger nanostructures (e.g., nanoparticles for nanofiber formation, nanofibers for nanoribbon formation) started to appear in hydrogels of Nap-FFK-Azi and Nap-FFG-Alk, or solution of Nap-Trai-Nap. The final nanostructures (i.e., nanofibers in hydrogel Nap-FFK-Azi, nanoribbons in hydrogel Nap-FFG-Alk and solution Nap-Trai-Nap) started to form at 3 h and stabilize at 6 h. In contrast, TEM images of Nap-FFK-Azi + Nap-FFG-Alk hydrogel showed that dense and slim nanofibers (i.e., Nap-FFK-Azi nanofibers) formed at 3 h (Fig. 4b). After that, from 24 h to 4d, we found that nanoparticles (i.e., generation of the cycloaddition product Nap-Trai-Nap) were continuously generated along the nanofibers. Finally, the nanoparticles connected with each other to form helical nanoribbons at day 7. In contrast, the real-time formation of nanostructures in Nap-FFK-Azi + Nap-FF-Nva hydrogel displayed completely different assembly processes. As shown in Supplementary Fig. 39, an obvious nanoribbon-nanofiber transformation was observed in Nap-FFK-Azi + Nap-FF-Nva hydrogel. Compared with those in the reactive group (Nap-FFK-Azi + Nap-FFG-Alk), the wide nanoribbon structures in Nap-FFK-Azi + Nap-FF-Nva hydrogel formed merely by co-assembly were not stable and transformed into fine nanofibers at day 7, indicating they were sub-equilibrium intermediates during the co-assembly process. This result also indicates that occurrence of the cycloaddition reaction in Nap-FFK-Azi + Nap-FFG-Alk group improves the stability of the intermediate nanoribbon structure of the two precursor peptides to the final nanoribbons. In addition, different from that in Nap-FFK-Azi + Nap-FFG-Alk group, no nanoparticle structure was found in Nap-FFK-Azi + Nap-FF-Nva group from 24 h to 7d. Above TEM observations directly validate our hypothesis that, along with the self-assembly of nanofiber, Nap-FFK-Azi or Nap-FFG-Alk is drawn near to the nanofiber with space-confined effect, undergoing cycloaddition reaction to yield regionally selective product Nap-FFK-Tria-GFF-Nap.

Fig. 4: Real-time TEM observations of the assembly-driven regionally selective azide–alkyne cycloaddition reaction.
figure 4

a Real-time TEM images of 2.5 mM Nap-FFK-Azi hydrogel, 2.5 mM Nap-FFG-Alk hydrogel, and 2.5 mM Nap-FFK-Tria-GFF-Nap solution aging at 4 °C. b Real-time TEM images of 2.5 mM Nap-FFK-Azi + Nap-FFG-Alk hydrogel aging at 4 °C. scale bar = 500 nm.

Full size image

Influence of amino acid chirality on assembly-driven azide–alkyne cycloaddition reaction

In order to investigate the effect of amino acid chirality on this assembly-driven cycloaddition reaction, we designed and synthesized three sets of control compounds: (i) group C1: Nap-FFKd-Azi, Nap-FFGd-Alk, and Nap-FFKd-Tria-GdFF-Nap; (ii) group C2: Nap-FdFdK-Azi, Nap-FdFdG-Alk, and Nap-FdFdK-Tria-GFdFd-Nap; (iii) group C3: Nap-FdFdKd-Azi, Nap-FdFdGd-Alk, and Nap-FdFdKd-Tria-GdFdFd-Nap. All the compounds were synthesized and purified with similar protocols, and then characterized with ESI-MS, 1H-NMR and 13C-NMR spectra (Supplementary Figs. 40–60). CAC measurement showed that their CACs were much smaller than 2.5 mM (Supplementary Fig. 61), suggesting these are able to self-assemble under a heat-cooling process (Supplementary Figs. 62–66). However, none of the cycloaddition products (retention time 20.3 min) was monitored by HPLC in these chiral groups (Fig. 5a). Significantly shorter, fragile fibers with indistinct edges, either linear or banded, were observed in TEM images of the individual peptides (Fig. 5b). However, in the groups of mixed peptides (e.g., Nap-FFKd-Azi + Nap-FFGd-Alk), only a few fragile linear nanostructures were appeared in their TEM images while nanoribbons with larger size were not observed at all (Fig. 5b). These results suggested that, after the chirality of its amino acid was changed, self-assembling property of the peptide was greatly affected, and co-assembly between different peptides was almost impossible to occur. As a result, assembly-driven cycloaddition reaction was inhibited. This hypothesis was further validated by CD spectra of these peptide assemblies. As displayed in Fig. 5c, only the CD spectra of Nap-FFGd-Alk and Nap-FdFdG-Alk displayed obvious secondary structures, but they are not the typical structures of α-helix, β-turn, or β-sheet, indicating their assemblies were complex. In general, the change of chirality showed a significant inhibitory effect on the assembly process, as well as the assembly-driven azide–alkyne cycloaddition reaction.

Fig. 5: Characterizations of the assemblies of chiral peptides.
figure 5

a HPLC traces of 2.5 mM chiral isomers of Nap-FFK-Azi (green), 2.5 mM chiral Nap-FFG-Alk (blue), 2.5 mM chiral Nap-FFK-Tria-GFF-Nap (red), or 2.5 mM chiral Nap-FFK-Azi + Nap-FFG-Alk hydrogel (orange) at 4 °C for 7 d. Wavelength for detection: 254 nm. b TEM images of the chiral isomer Nap-FFK-Azi hydrogel, 2.5 mM chiral Nap-FFG-Alk hydrogel, or 2.5 mM chiral Nap-FFK-Azi + Nap-FFG-Alk hydrogel after aging at 4 °C for 7 d. scale bar = 500 nm. c CD spectra of the corresponding samples in (b).

Full size image

Discussion

In summary, we rationally designed three compound (Nap-FFK-Azi, Nap-FFG-Alk, and Nap-FFK-Tria-GFF-Nap) and validated that assembly could drive metal-free azide–alkyne cycloaddition reaction with excellent regioselectivity. Hydrogelation, TEM, and CD results showed that the nanoribbons in Nap-FFK-Azi + Nap-FFG-Alk hydrogel were similar to those in Nap-FFK-Tria-GFF-Nap solution. HPLC analyses indicated that, it was self-assembly but not other factors that drived the azide–alkyne cycloaddition between Nap-FFK-Azi and Nap-FFG-Alk to yield the pure product Nap-FFK-Tria-GFF-Nap. Real-time TEM observations validated that, Nap-FFK-Azi nanofibers were firstly formed in Nap-FFK-Azi + Nap-FFG-Alk hydrogel and then Nap-FFG-Alk is driven to react with the assemblies to yield the regionally selective product Nap-FFK-Tria-GFF-Nap. Previous studies have revealed that hydrophobic interactions and hydrogen-bonding could accelerate the Diers-Alder reaction20. This work demonstrated another example of using assembly to realize metal-free azide–alkyne cycloaddition with excellent regioselectivity. Currently, metal-free bioorthogonal click reactions show promising potential in a variety of biological applications including biomolecule labeling21, prodrug activation22, and cancer theranostics23. Encouraged by this work, we are applying supramolecular self-assembly to realize more reactions without metal ion catalysis, and those works are on the way.

Methods

Representative procedure for assembly-driven azide–alkyne cycloaddition

Nap-FFK-Azi (2.5 mmol, 1.0 equiv.) and Nap-FFG-Alk (2.5 mmol, 1.0 equiv.) were mixed in 1 mL PBS (10 mM, pH 9), thereafter the mixture solution was heated at 100 °C for 1 h and aging at 4 °C. During the period of aging, assembly-driven azide–alkyne cycloaddition evolved over time and afforded regioselevtive product.