Sequence-selective encapsulation and protection of long peptides by a self-assembled FeII8L6 cubic cage

Self-assembly offers a general strategy for the preparation of large, hollow high-symmetry structures. Although biological capsules, such as virus capsids, are capable of selectively recognizing complex cargoes, synthetic encapsulants have lacked the capability to specifically bind large and complex biomolecules. Here we describe a cubic host obtained from the self-assembly of FeII and a zinc-porphyrin-containing ligand. This cubic cage is flexible and compatible with aqueous media. Its selectivity of encapsulation is driven by the coordination of guest functional groups to the zinc porphyrins. This new host thus specifically encapsulates guests incorporating imidazole and thiazole moieties, including drugs and peptides. Once encapsulated, the reactivity of a peptide is dramatically altered: encapsulated peptides are protected from trypsin hydrolysis, whereas physicochemically similar peptides that do not bind are cleaved.


Synthesis of zinc(II) tetrakis(4-(((6-formylpyridin-3-yl)oxy)methyl)-phenyl)porphyrin A
In a 2 mL round-bottomed flask 5-hydroxypicolinaldehyde (18 mg, 0.15 mmol) and potassium carbonate (25 mg) were placed. The flask was evacuated and refilled with nitrogen three times. Dry DMF (0.5 mL) was added under nitrogen. The mixture was heated to 70℃ and zinc(II) tetrakis(4-bromomethylphenyl)porphyrin 1-Zn (30 mg, 0.03 mmol) was added. The reaction mixture was heated at 70℃ for 12 hours. After this time the mixture was cooled down and diluted with water (5 mL) which was accompanied by precipitation of the crude product. The suspension was centrifuged, solution above the solid was removed and this procedure was repeated three times (3 x 3 mL of water) in order to remove residual DMF and potassium carbonate. The product was dried under vacuum. Yield: 28 mg (80%). The assembly should not be dried completely, otherwise it is not possible to re-dissolve it. The reason could be that filling the cavity with solvent is a kinetically slow process.

Stability of 1:
Cage 1 showed very high stability in acetonitrile; no degradation was observed even after several weeks at micromolar concentration ( Supplementary Fig. 21). Encouraged by this observation, we hypothesized that 1 might be stable enough to be used in a mixture of acetonitrile and water. Assemblies based on imine-bond formation have been observed to be most stable in water when the subcomponents are also water-soluble. When subcomponents are insoluble, the small equilibrium population of insoluble free subcomponents precipitate, driving the equilibrium towards decomposition of the cage. Remarkably, although subcomponent A was not observed to dissolve in either water or acetonitrile, NMR spectroscopy demonstrated 1 to be soluble and stable in a 1:1 mixture of water:acetonitrile. No degradation of a 80 µM solution of 1 in water:acetonitrile was observed after 24 h ( Supplementary Fig. 24), even in the presence of 100 equivalents of 1-methylimidazole ( Supplementary Fig. 25), which is a good ligand for Fe II . However, a small degree of degradation was observed at lower concentrations ( Supplementary Fig. 22). Around 5% of the UV-Vis signal intensity was lost after one hour at 1.5 µM concentration. Host-guest model: The experimental data obtained using UV-Vis spectroscopy for each guest was fitted using nonlinear analysis with DynaFit program (Biokin Software) 2 to the binding equation derived for 1:1 model porphyrin:imidazole group (non-cooperative model). This model is ideal for G1 (1-methylimidazole), which is so small that the binding to one porphyrin does not affect the interaction with the rest of porphyrins present in 1. Therefore, each porphyrin of 1 must bind to one G1 molecule in the saturation point, with the six affinity constants having the same value. The Kd value was 30 µM in acetonitrile, consistent with values obtained in previous studies 3 .
The decrease in absorption at 417 nm observed in the case of G1 was also observed in the cases of the other guests. As shown in the Supplementary Tables 1 and 2, the behavior observed was similar for all guests. We thus inferred that in all cases each porphyrin unit of 1 was interacting with one imidazole group in the saturation point of the titration. In this way, a non-cooperative model was used to obtain the intrinsic dissociation constant for the rest of the guests 4 .
Host-guest chemistry studied by fluorescence spectroscopy: In order to test our hypothesis that guest is interacting with the zinc porphyrins from within the cavity of 1, we decided to exploit the fluorescence properties of NDI moieties present in G5. The NDI exhibits aggregation-induced emission enhancement in water solutions 5 . Therefore, if three NDI units are stacked on top of each other in the cavity of the cage, as we foresaw for G5, an increase in the fluorescence signal of the NDI should be observed, accompanied by the appearance of the excimer band, as in the aggregation process. Indeed, when a solution of G5 was added to 1, the expected changes in the spectra were observed. The NDI fluorescence increased, and a new excimer band appeared at 470 nm. The solution was saturated with around 0.35 equivalents of the cage, as expected from the Kd of the ligand. The model (Supplementary Fig. 47) showing the interactions between the host and the guest molecules was prepared using the CAChe Workspace 6 . The MM2-optimized structure of the complex anticipates that the NDI moieties of the guests form a stack of three rings through aromatic stacking interactions.

Volume calculations
In order to determine the available void space within the 1, VOIDOO 7 calculations based on the optimized structures were performed. A virtual probe with the minimum radius such that it would not exit the cavity during the calculation was employed. The following parameters were changed from their default values: The flexibility of 1 was determined through the expansion and contraction of the structure (See Supplementary Fig. 57 for normal structure). In order to obtain the expanded structure, the positive charge of the zinc atoms present in the porphyrins was increased to +5. In this way, the high electrostatic repulsion between the porphyrins and the irons of the corner leads to the expansion of the cage after an energy minimization with MM2 level of theory in CAChe Workspace. Then, the charge of zinc atoms was returned to the correct value (+2), and the energy minimization was repeated (5) (Supplementary Fig. 58).
The contracted conformation was achieved using the following protocol: Firstly, porphyrins in opposite faces were connected through coordination with 4,4'-bipyridine. The structure was energy minimized. 4,4'-bipyridine was removed from the structure. Finally, the minimization step was repeated ( Supplementary Fig. 59).