Light-induced unfolding and refolding of supramolecular polymer nanofibres

Unlike classical covalent polymers, one-dimensionally (1D) elongated supramolecular polymers (SPs) can be encoded with high degrees of internal order by the cooperative aggregation of molecular subunits, which endows these SPs with extraordinary properties and functions. However, this internal order has not yet been exploited to generate and dynamically control well-defined higher-order (secondary) conformations of the SP backbone, which may induce functionality that is comparable to protein folding/unfolding. Herein, we report light-induced conformational changes of SPs based on the 1D exotic stacking of hydrogen-bonded azobenzene hexamers. The stacking causes a unique internal order that leads to spontaneous curvature, which allows accessing conformations that range from randomly folded to helically folded coils. The reversible photoisomerization of the azobenzene moiety destroys or recovers the curvature of the main chain, which demonstrates external control over the SP conformation that may ultimately lead to biological functions.

curvature to construct supramolecular polymer of unusual shapes with good shape persistency. k, A large macrocyclized supramolecular polymer without significant twisting; l, An exceptionally macrocyclized supramolecular polymer with an unidirectional continuous up-down trajectory that indicates their formation in solution not in the surface during solvent drying. Scale bars, 50 nm for i−l.
Supplementary Figure 2. AFM cross-sectional analysis of three supramolecular polymers of trans-2. a-c, SP random ; d-f, SP linear ; g-i, SP spiral . Scale bars, 100 nm for all. These analyses show that the three supramolecular polymers have similar cross-sectional width (W) of ca. 9.7 ± 0.3 nm and thickness (T) of ca. 2.7 ± 0.2 nm, suggesting that they are formed based on the same selfassembly process, i.e., tilted stacking of hexamers.  Figure 5. a, Shapes used to approximate the looped structures observed in SP random and SP spiral . b, SAXS data for SP random , with toroid model fit (red line) shown. c, SAXS data for SP random , with hollow cylinder model fit (red line) shown. Regardless of the model used, the same value of r ave was obtained. d, SAXS plots of SP linear (c = 1 × 10 -4 M in MCH) showing absence of maxima/minima. Figure 6. a, Full UV-Vis spectra of 2 in MCH (c = 1.0 × 10 -4 M, SP spiral ) before (black), after irradiation of UV-light (red) and Vis-light (green) subsequently. After UV-and Visirradiation, a decrease and subsequent increase in absorbance at λ max were observed, from which the extent of trans-to-cis and cis-to-trans photoisomerization of azobenzene moiety could be estimated respectively. b, Partial UV-Vis spectra at λ = 413−433 nm which measures degree of internal order. Upon UV-irradiation, an increase in absorbance at 420 nm indicates reduction of internal order which mostly recovers after Vis-irradiation. Figure 7. a, Chemical structure of trans-2 highlighting two sets of benzylic hydrogens (H 1 and H 2 ) which are susceptible toward photoisomerization. b, Full 1 H NMR spectra of 2 (c = 1 × 10 -3 M) in CDCl 3 before (black) and after (red) irradiation of UV-light for 20 min to reach a PSS. c, Partial 1 H NMR spectra where the integration of the benzylic proton signals of trans-2 and cis-2 suggests that UV irradiation leads to the 50% photoisomerization (trans-2: cis-2 = 50:50). d, UV-Vis spectra of CDCl 3 solutions of 2 (c = 1 × 10 -4 M) prepared by diluting the solutions used for the above NMR measurements. The maximum absorption intensity of the azobenzene unit (initially at 370 nm) shows 35% decrease upon the UV-irradiation, which corresponds to 50% trans-to-cis isomerization according to the NMR analysis. This relationship has been used as a reference to calculate photoisomerization yield of azobenzene unit in supramolecular polymers in MCH. Figure 8. DLS analysis of the dilute solution (c = 2.5 × 10 -5 ) of SP spiral upon successive exposed to UV and Vis light, showing photoinduced changes in the distribution of the hydrodynamic diameters (D h ). Figure 9. SAXS data for (SP random ) Vis with 5% cis-2 content, which was achieved by thermal back-isomerization of the Vis-light irradiated supramolecular polymer solution at dark over 60 h. The plot showed weak contributions from the specific scattering derived from the spontaneous curvature, thus supporting an improved refoldability. Figure 10. AFM images showing a reversibility in the photo-interconversion between (SP linear ) UV and (SP random ) Vis . a, (SP linear ) UV obtained by irradiation of (SP random ) Vis with UV-light. b, (SP random ) Vis obtained by subsequent irradiation with Vis-light. c, (SP linear ) UV acquired by succeeding irradiation with UV-light. d, (SP random ) Vis obtained by subsequent exposure to Vislight. Scale bars, 400 nm for all images. Figure 11. a, Comparison of r distribution between SP random and (SP random ) Vis along with their r ave values. b, Comparison of θ distribution between SP random and (SP random ) Vis along with their θ ave values.  Figure 18. A schematic representation for stacking of heteromeric rosettes (involving two cis-2 and four trans-2 molecules) into very straight fibres with long-range domains (~ 500 nm) by aligning all cis-arms linearly in space due to their steric demand. Figure 19. UV-Vis spectra of 2 in MCH (c = 1.0 × 10 -4 M, SP helical ) before (black) UV-irradiation, after (blue) UV-irradiation and subsequent (green) Vis-irradiation. UV-irradiation results in ~31% cis-2, and subsequent Vis-irradiation (weak Vis-light, 20 cm distance between the light source and sample) takes around 20 minutes to reach PSS. The increase in absorbance at λ max estimates the extent of cis-to-trans back-isomerization (with 11.5 % cis-2) of azobenzene moieties in relative to the reference system. It can be noted that at λ = 413−433 nm, an increased absorbance by UV-irradiation, suggest a decrease in internal order of supramolecular polymers, while decreased absorbance by Vis-irradiation indicates recovery of the high degree of internal order of SP helical . Figure 20. a−d, AFM images showing the continuous transformation of (SP linear ) UV to (SP random ) Vis upon gradual decreasing cis-2 content controlled by irradiation with weak Vis-light. Exposure to the weak Vis-light is able to linearly recover the spontaneous curvature. e, Further decrease in cis-2 content was achieved by standing the supramolecular polymer solution at dark over 60 h. This resulted in further of recovery turning angle (θ), and thus reduction of the dispersity of supramolecular polymers on the surface. Scale bars, 200 nm for all. Figure 21. AFM images showing a spontaneous conversion of kinetically formed (SP linear ) Vis to (SP random ) Vis over 20 h. a, Just after exposure to strong Vis-light, b, After 1 h, c, After 5 h, c, After 20 h from the irradiation with strong Vis-light for 30 s. This suggests that quick recovery of trans-2 with the strong Vis light is not able to recover the encoded curvature instantly, and it recovers slowly over 20 h.

Supramolecular polymerization.
Supramolecular polymers of trans-2 were prepared according to the methods shown in Supplementary Figure 22. Supramolecular polymers with variable foldability were prepared by altering cooling rate or adding CHCl 3 as co-solvent that can push the supramolecular polymerization process under more thermodynamic conditions. The difference in monomer composition at 90°C between pure MCH (α agg = 0.46 at c = 1 × 10 -4 M) and 15:85 v/v% CHCl 3 -MCH (α agg = 0.00 at c = 1 × 10 -4 M) is not responsible for the resulting conformation of supramolecular polymers. Figure 22. Detailed schematic procedures for the preparation of supramolecular polymers.

Supplementary
Analysis of SAXS data. As mentioned in the main text, the samples SP random and SP spiral both exhibit maxima and minima in the SAXS data. In order to determine whether these might arise from the loops (average turning angle, θ av > 360°) apparent in the microscopy images, data analysis was carried out with models representing (a) toroids and (b) hollow cylinders. This approximation is used because to our knowledge a model describing the SAXS arising from a polymer chain containing multiple loops of similar size has not been reported, perhaps due to the rarity of such a structure. However, as shown in Supplementary Figure 5a, the model shapes (i.e. toroids or hollow cylinders) and the looped samples are similar. Given the reasonable quality of the fit and the good agreement between the values obtained by SAXS analysis and by AFM, it is highly likely that the maxima/minima arise from the looped sample structure. The analysis models are described mathematically as follows: The form factor for a torus with an elliptical cross-section, the scattering length density difference between the torus and the solvent Δρ, radius R, cross-sectional radius a and aspect ratio b is generally given as follows 1,2 :

(Supplementary Equation 1) (Supplementary Equation 2)
In equation (1), J 0 is the Bessel function of zero order. The overall scattering for a delta distribution of toroids with scale factor N is then given as:

(Supplementary Equation 3)
The form factor for a cylinder, with the scattering length density difference between cylinder and solvent Δρ, radius R, length L is generally given as follows: 2

(Supplementary Equation 4)
In equation (4), J 1 is the Bessel function of first order. The overall scattering for a delta distribution of hollow cylinders (ΔR = shell width) is then given as:

(Supplementary Equation 5)
In the above, ρ solv and ρ 2 are the scattering length densities of the solvent and electron dense parts f 2, as described below. After R and ΔR had been found using the SASfit 2 analysis software, the average loop center-to-center radii, r av was calculated as R + ½ ΔR. In equation 5, N cyl is a scale factor that accounts for the number density of toroids.
Finally, for the samples SP spiral and SP helical an additional peak function (Lorentzian) was needed to obtain an adequate fit, justified by the clear asymmetrical shape of the low Q maxima. This is given as follows:

(Supplementary Equation 6)
In the above, A is the peak amplitude, σ is the width and Q 0 is the peak centre. Judging from the microscopy images of SP helical and given the fitted position of peak centre (Q 0 = 0.499 nm −1 , so d = 2π/Q = 12.6 nm), the peak may represent the repeat distance between spirals within the helices. However, for SP spiral it less clear what this feature (at Q 0 = 0.477 nm −1 , so d = 2π/Q = 13.2 nm) in the scattering describes as no repeating distances of this length-scale are observed in the microscopy images. It may be that the structure of SP spiral is more unravelled in solution than when adsorbed on HOPG.
The scattering length density of the supramolecular polymer were approximated as arising solely from the electron dense parts of 2, including the barbituric acid moiety, naphthalene moiety, azobenzene moiety and phenyl + methoxy groups. Considering only these groups, approximating the physical density (~1.3 g mL −1 ) using the ACD ChemSketch plugin and inputting that into the NIST neutron activation and scattering calculator obtained a reasonable estimate of ρ 2 = 12.5 × 10 −6 Å −2 . Using the same calculator, the scattering length density of methylcyclohexane (ρ solv = 7.5 × 10 −6 Å −2 ) was found. These numbers were inputted into the SASfit software and held constant throughout the fitting process.

Supplementary Methods
General methods.

UV-Vis spectroscopy.
UV-Vis spectra were recorded on a JASCO V660 spectrophotometer with a Peltier device temperature-control unit. Screw-capped quartz cuvette (path length: 1 mm for 1 × 10 -4 M solution; 10 mm for 2.5 × 10 -5 M solution) was used for UV-Vis studies. UV-Vis measurements for light irradiation experiments were performed at a concentration (c) of 1 × 10 -4 M in MCH. For temperature dependent UV-Vis experiments, the lower concentration (c = 2.5 × 10 -5 M) was applied to obtain fully molecularly dissolved species at 90 °C, which is required for the plot of molar fractions of aggregated molecules (α agg ) against temperature. For AFM studies, the higher concentration (c = 1 × 10 -4 M) was used to increase the population of self-assembled nanostructures. There was no significant morphological difference between supramolecular polymers prepared with two different concentrations (1 × 10 -4 M and 2.5 × 10 -5 M). Also, no significant differences in morphology were noticed when a 1 × 10 -4 M MCH solution was cooling either from 90 °C or 100 °C.

Dynamic light scattering.
Dynamic light scattering measurements were conducted on Zetasizer Nano (Malvern Instruments). QS high precision cell (3 × 3 mm, Hellma Analytics) was used for the measurements. The temperature for measurements was kept at 20 °C.

Atomic force microscopy (AFM).
AFM images were obtained under ambient conditions using Multimode 8 Nanoscope V (Bruker Instrument) in Peak Force Tapping (Scanasyst) mode. Silicon cantilevers (SCANASYST-AIR) with a spring constant of 0.4 N/m and frequency of 70 kHz (nominal value, Bruker, Japan) were used. The samples were prepared by spin-coating (3000 rpm, for 1 min) of MCH solutions of supramolecular polymers onto freshly cleaved highly-oriented pyrolytic graphite (HOPG). 10 µL supramolecular polymer solution was injected on the HOPG (5 × 5 mm) for every measurement.

Transmission electron microscopy (TEM).
TEM images were acquired on JEM-2100F (JEOL) at an acceleration voltage of 120 kV. TEM samples were prepared by spin-coating of MCH solutions of supramolecular polymers onto carbon-coated STEM Cu 75P grid (SHR-C075, grade: super ultrahigh resolution carbon, mesh 339, whole size 75 µm) and dried under air for 1 h followed by drying under vacuum for 24 h.