Supramolecular polymers form tactoids through liquid–liquid phase separation – Nature

Materials

All chemicals and reagents were purchased from commercial sources at the highest purity available and used as received unless otherwise stated. UPy-Gly, BTA-EG₄ and BTA-Glc were ordered from SymoChem. mPEG-O-C12 (DPEG, PEG MW: 5,000), PEG-fluorescein isothiocyanate (FITC) (MW: 10,000) and dextran-FITC (MW: 550,000) were obtained from Creative PEGWorks. PEG (MW: 8,000), cetrimonium bromide (CTAB) and alginate and sodium salt from brown algae were purchased from Sigma-Aldrich. Dextran 500 was ordered from Pharmacosmos (pharmaceutical quality). Water was purified on an EMD Millipore Milli-Q Integral Water Purification system. UPy–COOH was synthesized following previous methods^52,53. The synthesis of BTA-Cy5 is also reported in our previous work⁵⁴. The synthesis route for UPy-Cy5 can be found in the Supplementary Information.

Sample preparation

UPy-Gly or UPy-COOH stock solution

The stock solution was always freshly made. To prepare 4 wt% of stock solution, either UPy-Gly or UPy-COOH solid was dissolved in the PBS buffer and a calculated amount of 1 M NaOH, to make the final pH higher than 11. The solution was then heated at 75 °C for 15 min. Subsequently, a calculated amount of 600 μM of UPy-Cy5 or 1 mM of Nile red in methanol was added and mixed with the stock solution to reach a final concentration of 7.2 µM (UPy–-Cy5) or 2 µM (Nile red). Finally, 1 M HCl was added to adjust the pH to the targeted pH.

UPy-Gly or UPy-COOH solution with dextran

The freshly made UPy-Gly or UPy-COOH stock solution was immediately added into the prediluted PBS solution and mixed. Then 10 or 20 wt% of dextran solution (with 0.08 mol% of dextran-FITC) was added to reach to the required concentration. The mixture was vortexed for 20–30 s.

UPy-Gly or UPy-COOH solution with PEG and dextran

The freshly made UPy-Gly or UPy-COOH stock solution was immediately added into the premixed PBS and PEG solution and mixed. To achieve the required concentration, 10 or 20 wt% of dextran solution (with 0.08 mol% of dextran-FITC) was then added. The mixture was vortexed for 20–30 s.

BTA-EG₄ solution with CTAB

A total of 4 mg ml⁻¹ of BTA-EG₄ stock solution was freshly made by mixing BTA-EG₄ solid with 0.1 mol equiv. of CTAB and water; it was stirred and heated at 80 °C for 15 min. After heating, 500 µM of BTA-Cy5 in methanol was added to get a final concentration of about 1 µM. A small volume of the concentrated CTAB solution (50 mg ml⁻¹) was added to adjust the BTA-EG₄/CTAB ratio later.

BTA-Glc solution with PEG and dextran

A total of 4 mg ml⁻¹ of BTA-Glc stock solution was prepared by dissolving BTA-Glc solid in water; it was stirred and heated at 75 °C for 15 min. After heating, 500 µM of BTA-Cy5 in methanol was added to get a final concentration of 1 or 2 µM. The stock solution was then mixed with the heated dPEG solution. Next, the preheated 20 wt% of PEG was added before adding 20 wt% of dextran solution (with 0.08 mol% of dextran-FITC). The volumes of the added solutions were adjusted to arrive at the designed concentrations. The solution was shaken or vortexed after each step. All solutions except for the UPy-Gly or UPy-COOH stock solution were either incubated in the Eppendorf tube or transferred to the imaging chambers immediately after preparation.

Zeta potential measurements

Zeta potentials were measured on a Malvern instrument Zetasizer, model Nano ZSP. Zetasizer software was used to analyse and process the zeta potential data. For the preparation of UPy-COOH and UPy-Gly samples, the solid samples were dissolved and annealed in basic 0.84 × PBS (110 mM NaOH) at a concentration of 4.4 wt% for 15 min at 75 °C. After dissolving, the samples were diluted with MiliQ and 1 × PBS and adjusted to the final pH varying from 4.6 to 9 with 1 M HCl, resulting in 0.4 mM of UPy-Gly or UPy-COOH with different PBS concentrations. A DTS1070 cuvette was used for measuring the zeta potential. The measurement duration time was automated, and automatic attenuation and voltage was selected. The samples were measured in triplo at RT with a 30 s equilibration time.

Dual asymmetric centrifugation

To break down UPy-Gly fibrils, dual asymmetric centrifugation was performed using a Hettich ZentriMix 380 R equipped with a ZentriMix rotor. Samples were added to 0.2 ml Eppendorf tubes and centrifuged for 60 min with a rotational speed of 2,500 rpm. The samples were subjected only to shear forces generated by the dual rotor set-up with no milling beads added.

Confocal laser scanning microscopy

Sample preparation method 1

A total of 8–9 µl of the fresh sample was loaded into a 120-µm-thick chamber built with two pieces of #1.5 cover glass with an imaging spacer (Grace Bio-Labs SecureSeal imaging spacer, 8 wells, diameter × thickness: 9 mm × 0.12 mm) in the middle.

Sample preparation method 2

A total of 45–48 µl of the fresh sample was loaded into an 800-µm-thick chamber (Grace Bio-Labs SecureSeal hybridization chambers, 8 wells, diameter × depth: 9 mm × 0.8 mm, port diameter: 1.5 mm) with the #1.5 cover glass at the bottom. The open ports were sealed after sample loading.

Sample preparation method 3

About 100 µl of the fresh sample was loaded into each well of the ibidi μ-slide (18 wells, no. 1.5H glass bottom, well size: 5.7 × 6.1 × 6.8 mm³) and covered with the lid.

Sample preparation method 4

About 5 µl of the sample was loaded onto the glass holder and covered with the #1.5 cover glass. The surroundings of the cover glass were sealed with nail polish to reduce evaporation-induced drifting or flowing.

The fluorescent images and videos were acquired with the Leica TCS SP8 microscope in the confocal mode with 40×/0.9 (dry), 63×/1.2 (water immersion) and 63×/1.3 (oil immersion) objectives at a resolution of 512 × 512, 1,024 × 1,024 or 2,048 × 2,048 pixels. The lasers used were 488 nm (for FITC), 552 nm (for Nile red) and 638 nm (for Cy5). The videos were taken with a time gap of 10 min. The z-stack images were collected with a gap of 0.5, 1.0 or 2.0 µm.

The bright field images were collected with the Leica TCS SP8 microscope in the BF mode.

The image analysis was conducted with ImageJ or the Leica microscope built-in software (only for some z-stack images).

The tracking of the phase-separation states (concentration ratios of UPy-Cy5 inside and outside the droplets) is realized by time-lapsed videos of CLSM. The tempo-spatially resolved change of the fluorescence intensities can be extracted from each frame of the video. To avoid a bleaching effect, we set a 10 min gap between each frame. Several samples were tracked simultaneously using the ‘mark & find’ function.

Fluorescence recovery after photobleaching

The FRAP was conducted on the Leica TCS SP8 microscope in the confocal mode with a 63×/1.2 (water immersion) numerical aperture objective at a resolution of 512 × 512 pixels. The samples were dyed with 1.8 µM of UPy-Cy5. One to three images were taken before bleaching with the imaging power of 0.5–1.0%. Subsequently, selected circular areas with diameters ranging from 0.6 to 7.0 µm were bleached for 10 cycles (0.3–0.7 s per cycle) at 50% of power with the 638 nm laser. The following images were captured with 0.5–1.0% of power every 0.3–5.0 s. The whole process was automated with the built-in software. Three or more measurements were conducted for each sample. The fluorescence intensities of the regions of interest (ROIs) were extracted by ImageJ and normalized with the intensities of the reference area following the equation below.

Where i(t) is the normalized fluorescence intensity at time t, I(t) and I(0) are the fluorescence intensities of the ROI at times t and 0, respectively. R(t) and R(0) are the fluorescence intensities of the reference area at times t and 0, respectively. The normalized fluorescence recovery curves were then fitted with a first-order exponential equation (equation (2)) where A is the amplitude of the recovery, τ is the critical recovery time and C is the intercept⁵⁵.

The half-life of the recovery (t_1/2) could then be determined by equation (3) and the apparent diffusion constant (D_app) could be calculated by factoring in the radius of the ROI (ω) following equation (4)^55,56.

Atomic force microscopy

Wet samples for force curve measurements

The freshly cleaved mica was first treated with 20 µM of CaCl₂ for 20–30 min and washed with water two or three times. The residual solutions were pipetted away from the mica, eventually leaving a thin layer of water on the top. Next, 1 µl of the sample was evenly added onto the mica by immersing the tip of the pipette into the thin water layer. The mica was then incubated at room temperature for at least half an hour. It was either covered or replenished with water to avoid drying during the incubation time. Finally, about 100 µl of the buffer, which contained the same concentration of dextran and PBS as the sample, was mounted onto the mica right before the measurements.

Dry samples for tapping mode imaging

Samples were first diluted in water with a final concentration of 10 µM. A total of 5 µl of the diluted sample was immediately spin-coated onto a piece of freshly cleaved mica, of size 1.0 × 1.0–1.5 × 1.5 cm², at 2,000 rpm for 1 min.

In situ force curve measurements on tactoids

AFM measurements were conducted on a Cypher Environmental Scanner equipped with a closed gas cell. An active heating and cooling stage was used to actively control the temperature at 20 °C. The measured force curves on the tactoids were recorded in contact mode using a superluminescent diode to reduce the signal-to-noise ratio. Spherical poly methyl methacrylate (PMMA)-based CP-CONT-PM probes (Nanotools, spring constant k  =  0.2 N m⁻¹; r = 750 nm) with a resonance frequency of 30 Hz (approximately 7 Hz in PBS buffer) were used for all measurements. The cantilever was thermally calibrated in solution with blueDrive using the ‘Get Real’ function in the Igor Pro software. Force curves were obtained using a slow scan rate of 0.061 Hz over a force distance of about 4 µm (velocity 414.11 nm s⁻¹) to prevent any perturbations in the PBS buffer that could cause the displacing of the tactoids. Furthermore, the trigger point was kept to between 10 and 15 nN. After obtaining the force curves, we applied the Hertz model (for spherical AFM tips) to extract the Young’s modulus of the tactoids (equation (5)).

where F is the applied force, E_s is Young’s modulus, V_s is the Poisson ratio (V_s = 0.4), r is the radius of the spherical AFM tip and δ is the indentation depth. For all samples, several force curves (6–26) were recorded, and mean values and standard deviations were extracted. Two-sample Student’s t-tests were performed to verify the difference between two populations.

Tapping mode imaging of fibrils

AFM measurements of dry samples in tapping mode (phase less than 90) were similarly recorded on the Cypher Environmental Scanner equipped with a closed cell and a normal laser diode. A heating and cooling stage was used to actively control the temperature at 20 °C. Silicon AC–160TS probes (Oxford instruments, spring constant k = 26 N m⁻¹; f = 300 kHz) with a tip height of 14 µm and a radius of 7 nm were used for all measurements and calibrated using the ‘Get Real’ function in the Igor Pro software. Height images of either 20 × 20 µm or 10 × 10 µm were acquired using a scan rate of 3.0 Hz and 1,024 × 1,024 pixels. Contrast of the images was further enhanced using first-order planefit and flattening using Gwyddion v.2.60. Subsequently, the lengths of the UPy-Gly fibrils in the processed images were measured using the Digimizer software. The histograms of the fibril length were fitted with the Gaussian distributions.

Cryogenic transmission electron microscopy

For cryoTEM measurements, Quantifoil grids (R 2/2, Quantifoil Micro Tools GmbH) or Lacey grids (LC200-Cu, Electron Microscopy Sciences) were used. Before sample addition, grids were treated with surface plasma at 5 mA for 40 s using a Cressington 208 carbon coater. A total of 3 µl of the sample was applied to the grid held in an automated vitrification robot (FEI Vitrobot Mark IV), operating at 22 °C with a relative humidity of 100%. Excess sample was removed by blotting for 3 s using the filter paper with a blotting force of −2. The thin film formed was vitrified by plunging the grid into liquid ethane just above its freezing point. Vitrified films were transferred into the vacuum of a CryoTITAN (Thermo Fisher) equipped with a field emission gun operated at 300 kV, a postcolumn Gatan energy filter and a 2,048 × 2,048 Gatan CCD camera. Virtrified films were observed in the CryoTITAN microscope at temperatures below −170 °C. Micrographs were taken at low-dose conditions, using defocus values including −10 µm, −5 µm and −2.5 µm at 24,000 magnification.

The fibril diameters were extracted from several cryoTEM images per sample with ImageJ. The histograms of the fibril diameters could be fitted with the Gaussion distribution. Two-sample t-tests were carried out to verify the difference of mean values between two populations.

Polarized fluorescence microscopy

Sample preparation

A total of 2 µl of the sample was added to the holding glass and covered with #1.5 cover glass. The sides of the cover glass were sealed with nail polish to avoid evaporation and sample drifting.

Optical polarization measurements

The samples were imaged using a commercially available inverted microscope (model: Nikon Eclipse Ti2). After the samples were loaded on the microscope stage, it was illuminated (in epi-illumination mode) using a 637 nm excitation laser (OBIS FP 637LX, Coherent). The excitation parameters (excitation power 0.1 mW) were controlled using the Coherent software. The collimated laser beam was first allowed to pass through a 640 ± 10 nm band-pass filter, followed by a zero-order half-wave plate (Thorlabs) which is mounted on a rotation mount. The excitation beam was reflected by a dichroic mirror (ZT640rdc, Chroma) and was directed to an oil-immersion objective (Apo-TIRF, 60×, 1.49 numerical aperture, Nikon). This set-up results in the sample being illuminated with a collimated light beam with a well-controlled polarization state. To eliminate the concern of the depolarization of the excited light passing through the optics, we mounted a polarizer on top of the objective lens and measured the laser intensity by tuning the angle of the half-wave plate. The minimum laser intensity detected was almost zero. This suggests almost 100% polarization of the excited light.

The fluorescence signal was collected by the same objective lens and passed through a 635 nm notch filter and a 700 ± 75 nm band-pass filter before being collected by the detector (Prime BSI Express sCMOS camera, Teledyne Photometrics). The wide-field fluorescence images from the samples were captured using NIS Elements (Nikon) with an exposure time of 100 ms. For each sample, several 225 s or longer videos were recorded with the half-wave plate manually rotated by 4° every 5 s. This resulted in a polarization rotation of 2θ when the half-wave plate was rotated by θ.

The fluorescence intensities of the tactoids were extracted using ImageJ and plotted against the polarization angle of the incident light beam (Fig. 3g and Supplementary Fig. 8). To fit the oscillating fluorescence intensity curves, we first studied the bleaching effect and found that the intensities decayed linearly over time and that the bleaching rate was roughly linear to the starting intensity (Supplementary Fig. 6). Subsequently, we modified the squared cosine function and fitted the oscillatory curves with the following equations using Origin.

where I₀ is the amplitude of the oscillation curve and b is a constant that is not sensitive to the angle of the linearly polarized light (θ). k is the decaying rate of the amplitude due to bleaching. θ₀ is related to the orientation of the dyes. The extent of alignment (α) of the fibrils within the tactoids can thus be calculated by equation (8).

The good correlation between the angles of the dyes (θ₀) and the relative orientations of the tactoids (θ_tactoids) obtained from the images confirmed correlated positioning between the dyes and the fibrils (Supplementary Fig. 7).

Instant molecular orientation microscopy based on a polarization camera

POLCAM is used to map the molecular orientation distribution in the tactoids³⁸. It is a simplified fluorescence orientation microscopy method based on a wide-field fluorescence microscope set-up. In the measurement based on POLCAM, we use the same microscope set-up as in normal linear polarizer-based measurement, as described above, except that instead of an sCMOS camera, we use a polarization camera DZK 33UX250 from The Imaging Source. A quarter-wave plate is installed in the light path to get circularly polarized light for the imaging of the background. The polarization camera is equipped with Sony IMX250MZR Polarsens sensor chip with on-chip polarizers with transmission axis in four directions (0°, 45°, 90° and −45°), thanks to which full Stokes parameters can be calculated on the basis of the fluorescence intensities measured from the four polarization channels. We define the Stokes parameters as follows.

where I₀, I₄₅, I₉₀ and I₁₃₅ are measured intensities in the four polarized channels from the polarization camera. Defined as such, the Stokes parameters describe the total intensity of the optical field (S₀), the intensity of linear horizontal or vertical polarization (S₁) and the intensity of linear 45° or −45° polarization (S₂). The pixel-by-pixel Stokes parameters are then used to calculate the in-plane orientation ϕ of the fluorescence emission coming from UPy-Cy5 in the tactoids. We define the in-plane orientation of the fluorescence emission by:

where AoLP stands for the angle of linear polarization.

To analyse the images from polarization camera measurement, we use the open-source software for the POLCAM method (POLCAM-SR) under the diffraction-limited mode.

Small-angle X-ray scattering

The SAXS measurements were conducted on a SAXSLAB Ganesha system, equipped with a GeniX-Cu ultra-low divergence source and a photon flux of 1 × 10⁸ photons s⁻¹. The wavelength of the X-ray is 1.5 Å. About 150 µl of solutions were loaded into each fixed, 2 mm quartz capillary. Two-dimensional intensity images were collected with a Pilatus 300 K silicon pixel detector with a measurement of 4 h and converted to one-dimensional plots with SAXSGUI. The q range covered by a single measurement was 0.12–6.90 nm⁻¹ and the absolute q value was calibrated with AgBeh. Solvent and capillary contributions to the scattering intensities were subtracted from the blank solution with the PRIMUS program from the ATSAS software package.

Dynamic and static light scattering

DLS and SLS experiments were performed on an ALV/CGS-3 compact goniometer system (ALV-GmbH), which consists of a single detector, ALV/LSE-5004 light-scattering electronics, several tau digital correlators and a Cobolt Samba 50 laser (laser wavelength 532 nm). DLS/SLS data were collected from 20° to 150° with a 10° step.