Host–Guest Interaction Driven Peptide Assembly into Photoresponsive Two-Dimensional Nanosheets with Switchable Antibacterial Activity

Open AccessCCS ChemistryRESEARCH ARTICLE1 Jul 2021Host–Guest Interaction Driven Peptide Assembly into Photoresponsive Two-Dimensional Nanosheets with Switchable Antibacterial Activity Xiaoming Xie, Bo Gao, Zhiyuan Ma, Junqing Liu, Jianfeng Zhang, Jing Liang, Zhijun Chen, Lixin Wu and Wen Li Xiaoming Xie State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author , Bo Gao State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author , Zhiyuan Ma State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author , Junqing Liu School of Life Sciences, Jilin Province Innovation Platform of Straw Comprehensive Utilization Technology, Jilin Agricultural University, Changchun 130118 Google Scholar More articles by this author , Jianfeng Zhang School of Life Sciences, Jilin Province Innovation Platform of Straw Comprehensive Utilization Technology, Jilin Agricultural University, Changchun 130118 Google Scholar More articles by this author , Jing Liang School of Life Sciences, Jilin Province Innovation Platform of Straw Comprehensive Utilization Technology, Jilin Agricultural University, Changchun 130118 Google Scholar More articles by this author , Zhijun Chen State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author , Lixin Wu State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author and Wen Li *Corresponding author: E-mail Address: [email protected] State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun 130012 Google Scholar More articles by this author https://doi.org/10.31635/ccschem.020.202000312 SectionsSupplemental MaterialAboutAbstractPDF ToolsAdd to favoritesTrack Citations ShareFacebookTwitterLinked InEmail Selectively controlling the bioactivity of antimicrobial peptides is not only a fascinating scientific challenge but also a necessity in localized antibacterial therapy. Here, a smart antimicrobial system has been fabricated via host–guest driven dynamic self-assembly between a branched cyclodextrin and cationic linear peptides appended with azobenzene side chains. The self-assembly structure of the host–guest system could be controlled reversibly through the photoresponsive isomerization of azobenzene moieties. Notably, trans-azobenzene side chains of the cationic peptides can interact with the branched cyclodextrin and form microscale sheet-like structures with high surface potentials. The multivalent positive charges covering the surface of the sheet-like structures enable the antibacterial features. However, the UV-triggered cis-isomerization of azobenzene residues weakens the host–guest interactions between azobenzene and the branched cyclodextrin, resulting in the formation of small and inactive nanospheres. Thus, the selective regulation of the antibacterial activity of these peptide assemblies was achieved by delivering the light with spatiotemporal precision. This kind of photoresponsive peptide self-assembly system with switchable bioactivity may provide a new insight into the development of smart supramolecular antibacterial materials. Download figure Download PowerPoint Introduction Bacterial infection remains a major threat to global health and has become aggravated with the unceasing proliferation of antimicrobial resistance.1–3 The emergence of the antibiotic crisis demands the development of new approaches and antimicrobial agents, which are not affected by the conventional mechanisms of antibiotic resistance in bacteria.4–6 In this context, antimicrobial peptides (AMPs), serving as endogenous host defense molecules in many living organisms, are highly promising candidates.7–12 Unlike conventional antibiotics that exert the antibacterial function relying on a receptor-specific target manner, most AMPs carrying positive charges can attach to negatively charged bacteria cell membranes via nonspecific electrostatic interactions.13–15 The AMP segments binding on the surface of bacteria will accumulate and subsequently enter the bacterial cell to destroy the membrane, thereby causing the leakage of cytoplasmic components as well as the apoptosis of bacteria.16,17 Because AMPs have nonspecific action mechanisms, much research has commenced on the design and synthesis of various cationic short peptides for antibacterial treatment.18–21 However, even with these advances, the poor activity and stability of the short peptides under physiological conditions greatly plague their application. Peptide assembly has a significant role in the creation of bioactive materials with exquisite levels of efficiency.22–31 Recent studies report that self-assembling AMPs with multivalent nanostructures are highly effective in killing bacteria.32–36 The highly concentrated cationic groups covering the surface of the AMP nanostructures can enhance their binding affinity to bacteria, and lower the dose of AMPs required to treat bacteria.37 Additionally, molecular stacking within the nanostructures largely limits enzyme accessibility, thus improving the stability of AMPs against proteolytic hydrolysis.38 More importantly, the dynamic assembly and disassembly features of the AMPs provide great opportunity to engineer their antimicrobial efficacy on demand.39–41 For example, Chen et al.39 reported a guest-triggered AMP assembly system with switchable antibacterial activity. Yang et al.40 harnessed the overexpressed phosphatase in Escherichia coli to catalyze the assembly of short peptide or antibiotic-peptide conjugates,42 activating their biological functions within the bacteria cell. Chen et al.41 developed a pH-responsive AMP nanomaterial, which could be activated only in the acidic microenvironment of the bacterial infection area. These vivid examples further impel us to develop smart antibacterial agents with controllable bioactivities with spatiotemporal precision, which is crucial for implementing localized antibacterial therapy.43,44 Light is a noninvasive and bio-orthogonal trigger, and can be delivered with highly spatial and temporal selectivity.45–47 A pioneering study on the spatiotemporal control of antibacterial activity has been reported by Velema et al.48 and Wegener et al.49 via coupling conventional antibiotics with a photoactive azobenzene moiety. Recently, Babii et al. reported smart AMPs by introducing a diarylethene unit into a cyclic peptide backbone. The photoswitchable backbone conformation of the cyclic peptidomimetics enables selective control of antibacterial activity at the molecular level.50 However, the development of cyclic AMPs with photoresponsive behavior remains a great challenge and is difficult to generalize due to the rigorous design requirement. It also suffers from inherent problems associated with complicated synthesis and poor yield. As the self-assembling AMPs are based on dynamic noncovalent interactions, it is very logical to develop a simple but universal strategy to generate smart AMPs based on relatively simple linear peptides by taking advantage of their photoswitchable self-assembly at the supramolecular level. To address this issue, a series of azobenzene-containing linear peptides (P1–P4, Figure 1a) and a trigeminal β-cyclodextrin (tri-β-CD, Figure 1a) were designed and synthesized. It is expected that the azobenzene side chains of these peptides will recognize the tri-β-CD, forming cross-linking assemblies via host–guest interactions. The photoresponsive isomerization of azobenzene moieties allows us to dynamically regulate the assembly and disassembly between the synthesized peptides and tri-β-CD. The lysine side chains are introduced into the molecular design to provide nonspecific binding sites to bacteria, and the hydrophobic residues (such as valine, isoleucine, and alanine) facilitate peptides insertion into the cell membrane of the bacteria.15,17 Herein, we report the host–guest interaction driven self-assembly of the cationic peptides and tri-β-CD, and their enhanced antibacterial potency. We further demonstrate how the antibacterial activity of the peptides and tri-β-CD complexes could be controlled with spatiotemporal precision via photoswitchable self-assembly. Figure 1 | (a) Structures of peptides P1–P5 and the tri-β-CD. (b) The schematic drawing of the photoresponsive assembly between P1 and tri-β-CD. Download figure Download PowerPoint Experimental Methods Preparation of peptides/tri-β-CD complexes The peptides and tri-β-CD solutions were first prepared by dissolving the lyophilized peptides and tri-β-CD powders in deionized water, respectively. Then, the tri-β-CD solution was added dropwise into the stock solution of peptides, and the resulting solution was sonicated at 25 °C for 3 min. The molar ratio of peptides to tri-β-CD was maintained at 3∶2 and the solution was maintained at approximately 7.0 pH. The self-assembly peptides/tri-β-CD complexes were obtained after aging the mixed solution for ca. 3 h at 25 °C. The photoresponsive self-assembly was performed by exposing the aqueous solution of peptides/tri-β-CD to UV (365 nm) or visible light (470 nm) for 10 min. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) was captured on an Autoflex speed TOF/TOF (Brucker) operating in positive mode within a mass range from 700 to 5000 Da. Proton nuclear magnetic resonance Proton nuclear magnetic resonance (1H NMR) spectra were collected on a Bruker AVANCE 500 MHz spectrometer by dissolving P1, tri-β-CD, and P1/tri-β-CD in D2O at 25 °C. Two-dimensional (2D) 1H Nuclear Overhauser Effect Spectroscopy (NOESY) spectra were recorded on a Bruker AVANCEIII 600 MHz spectrometer, and the optimized mixing time was set to 200 ms before the acquisition of the free induction decay. The trans-P1/tri-β-CD (the molar ratio of P1 to tri-β-CD is 3∶2) solution was prepared freshly in H2O/D2O (v/v = 9∶1) and kept at room temperature for 2 h before NOESY testing. The cis-P1/tri-β-CD sample was prepared by exposing the trans-P1/tri-β-CD solution to UV light (365 nm) for 30 min before NOESY testing. The concentration of P1 in all the solution sample was kept at 8 mg mL−1. Dynamic light scattering Dynamic light scattering (DLS) experiments were recorded at 25 °C on a Malvern Zetasizer Nano ZS (Malvern Instruments, UK) using a detection angle of 173° and a 3 mW He–Ne laser operating at λ = 633 nm. The temperature equilibration time was set to 120 s in all cases, and the measurements were repeated at least three times. Circular dichroism Circular dichroism (CD) spectra were recorded on a JASCO model J-810 spectropolarimeter (25 °C, Xe lamp) under a constant flow of nitrogen gas during operation. The solution samples were loaded into a rectangular quartz cell with a 0.1 cm path length, and the CD spectra wavelength range was 260–190 nm with a step of 0.5 nm. The analyses were repeated five times and averaged. The JASCO software was used for background subtraction. Thioflavin T binding study The thioflavin T (ThT) titration was measured using the following procedure. A ThT solution was added into the aqueous solution of P1, and the final concentration of ThT was maintained at 10 μM. The resultant P1 solution (pH ∼7) was incubated at room temperature for 5 h. The fluorescence spectra of individual ThT (10 μM) and the P1/ThT solution samples were recorded on a 5301 PC spectrophotometer (Shimadzu, Tokyo, Japan), respectively, with an excitation wavelength of 420 nm. Zeta potential The zeta potentials of the P1/tri-β-CD solution samples were measured using the Zetasizer NanoZS instrument (Malvern Panalytical) at 25 °C. UV–visible spectra The UV–visible (UV–vis) spectra were performed on a Varian Cary 50 UV–vis spectrophotometer. The wavelength is in the range of 200–600 nm with a step of 1 nm. Transmission electron microscopy The transmission electron microscopy (TEM) images were acquired on a JEOL-2010 electron microscope (200 kV). The peptides/tri-β-CD solution samples were casted onto a carbon-coated copper grid and then dried completely in air. The individual peptide samples and the peptides/tri-β-CD specimens irradiated by UV light (365 nm) were stained, respectively, with 0.1 wt % uranyl acetate aqueous solution for 3 min, the excess amount of the uranyl acetate solution was removed by filter paper. The trans-peptides/tri-β-CD specimens without staining were used directly for TEM measurements. E. coli cells with and without peptides/tri-β-CD were washed with phosphate-buffered saline (PBS) at least three times. The centrifuged E. coli samples were redissolved in deionized water and casted on a carbon-coated copper grid for TEM measurements. Cryogenic TEM Cryogenic TEM (Cryo-TEM) was performed on a JEOL-JEM 2100 TEM instrument (approximately 90 K, 120 kV) equipped with a SC 1000 charge-coupled device (CCD) camera (Gatan, Inc., USA). A liquid droplet of P1/tri-β-CD (3 μL) was transferred to an ultrathin copper grid after hydrophilic treatment under controlled temperature and humidity (97–99%) to prevent evaporation of sample solution. Then, the superfluous liquid droplets were removed with filter paper, and the thin aqueous films were rapidly vitrified by plunging them into liquid ethane and cooled to approximately 90 K by liquid nitrogen. The excess amount of ethane was removed using blotting paper after the sample solution was frozen. Finally, the grid was inserted into a Gatan 626 cryo holder using a cryotransfer device for cryo-TEM measurements. Atomic force microscopy Atomic force microscopy (AFM) measurements were recorded on a Bruker Dimension 3100 instrument (Karlsruhe, Germany) using a tapping mode in air (25 °C). The AFM samples were prepared by casting the P1/tri-β-CD solution on the fresh surface of a mica wafer. After settling for 3 min, the excess amount of solution was removed by filter paper, and the air-dried samples were utilized for AFM tests. Laser scanning confocal microscopy Laser scanning confocal microscopy (LSCM) measurements were carried out using a FV1000 confocal microscopy. The P1/tri-β-CD solution (60 μM) was incubated with Rhodamine B solution (10 μM), a fluorescence probe exhibiting an increase of emission when it is adsorbed on the surface of P1/tri-β-CD nanosheets. The resultant solution was casted on glass for LSCM measurements with an excitation wavelength of 515 nm. In the case of the dead assay for cell viability, the E. coli cells were treated with propidium iodide (PI; 0.05 mg mL−1) and fluorescein diacetate (FDA; 0.02 mg mL−1) for 30 min, and then washed with PBS buffer at least three times before the test analysis. The obtained solution of double staining E. coli was casted on glass for LSCM observation with an excitation wavelength of 488 and 515 nm, respectively. The emission wavelengths were fixed at 520–550 nm (for FDA) and 600–650 nm (for PI), respectively. Scanning electron microscopy Scanning electron microscopy (SEM) images were obtained on a JEOL FESEM 6700F electron microscope (15 kV). E. coli cells treated with and without P1/tri-β-CD were washed with PBS buffer at least three times. The centrifuged E. coli was redissolved in deionized water and casted on a clean silica wafer adhered onto an aluminum sample holder and dried in the air, then sputter coated with platinum. Antibacterial assays Fresh Luria–Bertani (LB) liquid medium was obtained by mixing 2.5 g of peptone, 1.25 g of yeast extract, and 2.5 g of sodium chloride in 245 mL of deionized water and sterilized at 121 °C for 20 min. The LB agar plate was obtained by adding another 3.75 g agar powder into the LB liquid medium and then sterilizing at 121 °C for 20 min. Bacterial strains (Gram-negative: E. coli and Pseudomonas aeruginosa, Gram-positive: Staphylococcus aureusand Bacillus subtilis) were suspended in 25 mL of LB liquid medium (containing 25 μL of ampicillin), respectively. The obtained samples were cultivated in a constant temperature shaker (37 °C, 160 rpm) for 12–14 h to reach the stationary growth phase. The resulting bacterial broth was diluted to ∼0.035 of the OD600 (BioPhotometer plus instrument), then 2.0 mL of bacterial cell suspensions were added into four glass tubes containing deionized water (as the blank test), peptides, tri-β-CD, and peptides/tri-β-CD, respectively. The solution volume of each tube was kept at 2.5 mL, the concentrations of peptides were maintained at 5–60 μmol L−1, and the concentration of tri-β-CD was kept at 3.3–40.0 μmol L−1. The antibacterial activity of the samples was assessed through monitoring the OD600 values of the bacterial cell suspensions at indicated time points by the BioPhotometer plus instrument with a 10 s shaking step before each measurement. The antibacterial activity of the E. coli samples treated with P1/tri-β-CD was also evaluated by counting the colony-forming unit (CFU). The treated E. coli suspensions were serially diluted 1 × 10n fold with PBS buffer solution, then a 100 μL portion of the diluted solution was spread on the solid LB (supplemented with 1 μL mL−1 ampicillin) agar plate. After an incubation for 20 h at 37 °C, the final colonies of E. coli were counted. The tests were repeated three times and averaged. In the case of the photoresponsive antimicrobial test with temporal selectivity, the LB liquid medium containing trans-P1/tri-β-CD samples was first irradiated at 365 nm for 10 min in the dark prior to adding the E. coli cell suspension (OD600 ≈ 0.035), then the samples were irradiated by visible light (470 nm, 10 min) with different time intervals (0, 84, and 144 min). All samples after irradiation were still placed in an incubator shaker (160 rpm, 37 °C) to cultivate. In the case of the photoresponsive antimicrobial test with spatial selectivity, the LB liquid medium containing E. coli cells at the stationary growth stage was diluted to OD600 ≈ 0.035, and a 20 μL portion of the diluted E. coli was coated uniformly on the surface of LB agar plate containing trans-P1/tri-β-CD. After waiting for 15 min, the agar plate was kept in the dark and irradiated with UV light for 10 min (365 nm, 6 W). Next, the agar plate was illuminated with visible light (470 nm) through a mask placed on top of the plate. Afterwards, the plate was incubated in a constant temperature shaker (37 °C) overnight to allow for bacterial growth. For the antimicrobial test with spatiotemporal resolution, diluted E. coli culture droplets (OD600 ≈ 0.035) were coated uniformly on the surface of a LB agar plate containing cis-P1/tri-β-CD. The LB agar plate was kept in the dark and covered with a rotatable mask. The sample was incubated at 37 °C and irradiated (470 nm, 10 min) selectively at different areas by rotating the mask with different time intervals. After incubation in the dark for 24 h, patterned sectors of E. coli were observed. Cytotoxicity assay L-929 cells (mouse fibroblasts) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% fetal bovine serum and 1% penicillin/streptomycin. The assays were carried out in sterile 96-well flat-bottomed polystyrene microtiter plates. Plates containing 100 μL of cell suspension in each well (5000 to 10,000 cells/well) were preincubated for 24 h at 37 °C in a humidified environment with 5% CO2. The samples to be tested were two-fold diluted, and a 10 μL portion of the tested compounds (P1, tri-β-CD, and P1/tri-β-CD) was added to test plates in triplicate to get a final concentration of 60 μM (79.8 μg mL−1). The plates containing the tested compounds were incubated for 24 h. Subsequently, the plates were further incubated with 10 μL of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2,3-dihydro-1H-tetrazol-3-ium bromide (MTT) solution (2.5 mg mL−1) at 37 °C for 4 h. The top medium was then removed, and 100 μL of dimethyl sulfoxide (DMSO) was added to each well to dissolve the formazan crystals. The absorbance of the solution was measured with an enzyme standard instrument at 570 nm using a multiwall plate reader (AccuSkan, MultiSkan FC, Thermo Fisher Scientific). For fluorescence microscope examination, L-929 cells were inoculated in a 96-well flat-bottomed polystyrene microtiter plate with the number of cells at approximately 104, 5 × 103, and 103, respectively. The liquid medium was changed, replaced with fresh medium after inoculation for 24 h, and then P1/tri-β-CD was added. The live cell assay was monitored with different incubation time (1, 3, and 7 days, respectively). Afterwards, the cells were washed using PBS solution two to three times to remove the active esterase from the medium. Dye-working fluid (100 μL) containing 2 M calcein AM and 8 M PI of PBS was added to each hole, then incubated at 37 °C for 30 min. Finally, the dye-working fluid was extracted and visualized under a FV1000 confocal microscopy. The excitation wavelengths were fixed at 490 and 545 nm, respectively. Results and Discussion Host–guest interaction and 2D self-assembly The linear peptide P1, which consists of an alternating sequence of hydrophilic (lysine) and hydrophobic (valine) residues (Figure 1a), was synthesized by a standard 9-fluorenylmethoxycarbonyl (Fmoc) solid-phase method. Analogous peptides (P2–P4, Figure 1a) with various residues were also prepared to examine the effect of residue type on the antibacterial activity. Peptide P5, without an azobenzene terminus, was synthesized for comparison purposes. The purity of all the peptides was confirmed by high-performance liquid chromatography (HPLC; Supporting Information Figure S1) and MALDI-TOF-MS ( Supporting Information Figure S2). The tri-β-CD was synthesized per the reported procedure51 and was characterized by 1H NMR ( Supporting Information Figure S3) and MALDI-TOF MS ( Supporting Information Figure S4). DLS of the respective P1 aqueous solution (pH ∼7.0) showed a maximum hydrodynamic diameter (Dh) of ∼20 nm ( Supporting Information Figure S5). When a solution of P1 was cast onto a copper grid and then negatively stained with uranyl acetate, irregular spherical structures (Figure 2a) with a size of 10–40 nm were observed by TEM. The CD spectrum with a negative band at ∼199 nm ( Supporting Information Figure S6) revealed that pure P1 adopted a random coil conformation in aqueous solution (60 μM, pH ∼7.0).52–54 This disordered secondary structure of P1 was verified by ThT titration ( Supporting Information Figure S7). We believe that the random coil conformation of P1 stems from the strong electrostatic repulsion of the protonated lysine residues (pKa ∼10.5).55,56 The random coil conformation of cationic P1 remained unchanged upon the addition of tri-β-CD to an aqueous solution of P1 even with an approximate molar ratio of P1 to tri-β-CD of 3∶2, as observed from the CD spectrum ( Supporting Information Figure S6). However, DLS data of the P1/tri-β-CD sample (60 μM) showed a maximum Dh of ∼1.8 μm ( Supporting Information Figure S5), which is much larger than that of individual P1. TEM measurement of the P1/tri-β-CD sample further demonstrated the formation of large 2D assemblies at the micrometer scale (Figure 2b). The AFM image (Figure 2c) of the P1/tri-β-CD sample revealed the 2D architecture as an ultrathin nanosheet with a height of ∼1.85 nm. To avoid the effect of water evaporation on the assembled nanostructures, we performed the morphological study by in-situ techniques including cryo-TEM and LSCM. As shown in Supporting Information Figures S8 and S9, the cryo-TEM and the LSCM identified that the above nanosheets indeed formed homogeneously in the aqueous solution rather than assembled on a solid surface during the drying process.57,58 This spontaneous formation of nanosheets was not limited to P1/tri-β-CD because azobenzene-containing peptides P2, P3, and P4 also yielded similar ultrathin nanosheets when mixed with tri-β-CD ( Supporting Information Figure S10). In contrast, no large aggregates or nanosheets were observed in the case of P5/tri-β-CD due to the lack of azobenzene residues ( Supporting Information Figure S10). The above observations suggest that the extended nanosheets are initiated by the host–guest recognition between tri-β-CD and the azobenzene residues of the peptides. Figure 2 | (a) TEM image of P1. (b) TEM image of P1/tri-β-CD. (c) AFM image and cross-section analysis of P1/tri-β-CD nanosheet. Download figure Download PowerPoint This hypothesis was validated by 1H NMR. The detailed assignments of the proton peaks of tri-β-CD and P1 in D2O are listed, respectively, in Supporting Information Figures S3 and S11. Individual P1 molecules showed strong and sharp peaks at 7–8 ppm (Figure 3a), which were assigned to the protons of trans-azobenzene residues. In the case of P1/tri-β-CD sample, the corresponding azobenzene residues broadened (Figure 3a) due to the limited motion resulting from the formation of large nanosheets via the host–guest interactions between azobenzene moieties and tri-β-CD. To better understand the host–guest recognition of the P1/tri-β-CD sample, detailed 1H NOESY NMR characterizations were performed. As shown in Figure 3b, the cross-peaks located in the blue rectangle A are attributed to the intermolecular correlation between the protons of the azobenzene of P1 and the internal protons (2H–5H, or partially) in the cavity of β-CD, and the cross-peaks located in the black rectangle B correspond to the intermolecular correlation between the protons of the azobenzene of P1 and the internal proton (1H) in the cavity of β-CD.51,59 These results demonstrate that the trans-azobenzene moieties of P1 are deeply embedded in the hydrophobic cavities of tri-β-CD via host–guest interactions.51,59 Zeta-potential measurement revealed that the P1/tri-β-CD aqueous sample had a net potential (ζ) of + 44.8 mV (Figure 4a), thus implying that the surface of the nanosheets was covered by highly concentrated and protonated lysine residues of P1. Considering the random coil conformation of P1, the molecular length (0.8 nm) of P1 between the valine unit and the hydrophilic amine at the ɛ-position of lysine, the 1.53 nm size of β-CD, the 1.85 nm thickness of the nanosheets, and the high zeta potential, we propose a plausible assembly model. As shown in Figure 1b, the tri-β-CD segments act as cross-linkers to connect the P1 molecules via host–guest interactions, driving the supramolecular polymerization. However, the protonated lysine groups with highly local concentration on the surface of the assemblies have a strong propensity to inhibit close molecular packing due to the presence of multiple electrostatic repulsions among them. The synergy of these noncovalent interactions contributes to the formation of extended nanosheets with protonated lysine groups on both the top and bottom surface in aqueous solution. The proposed assembly mechanism was supported by the following series of experiments: (1) when P1 was mixed with monovalent β-CD under the same condition, small nanospheres (10–30 nm, Supporting Information Figure S12a) with low zeta potential (+8.7 mV; Supporting Information Figure S12b) were observed, indicating that the monovalent β-cyclodextrin is unable to initiate the cross-linking assembly and supramolecular polymerization; (2) the formation of the extended 2D nanosheets of P1/tri-β-CD in aqueous solution is pH dependent. As shown in Supporting Information Figure S13a, polydispersed structures with sizes of 70–135 nm were observed when the P1/tri-β-CD solution sample was prepared at pH ∼11. Notably, this solution sample was unstable and has strong propensity to coagulate into water-immiscible precipitate with cross-linked structures at the macroscopic level ( Supporting Information Figure S13b). However, with decreasing the pH to ∼7, the water-immiscible precipitate can redissolve into the parent solution within 5 min, and irregular spheres together with nanosheets ( Supporting Information Figure S13c) were