De Novo Design of Functional Coassembling Organic–Inorganic Hydrogels for Hierarchical Mineralization and Neovascularization

Synthetic nanostructured materials incorporating both organic and inorganic components offer a unique, powerful, and versatile class of materials for widespread applications due to the distinct, yet complementary, nature of the intrinsic properties of the different constituents. We report a supramolecular system based on synthetic nanoclay (Laponite, Lap) and peptide amphiphiles (PAs, PAH3) rationally designed to coassemble into nanostructured hydrogels with high structural integrity and a spectrum of bioactivities. Spectroscopic and scattering techniques and molecular dynamic simulation approaches were harnessed to confirm that PAH3 nanofibers electrostatically adsorbed and conformed to the surface of Lap nanodisks. Electron and atomic force microscopies also confirmed an increase in diameter and surface area of PAH3 nanofibers after coassembly with Lap. Dynamic oscillatory rheology revealed that the coassembled PAH3-Lap hydrogels displayed high stiffness and robust self-healing behavior while gas adsorption analysis confirmed a hierarchical and heterogeneous porosity. Furthermore, this distinctive structure within the three-dimensional (3D) matrix provided spatial confinement for the nucleation and hierarchical organization of high-aspect ratio hydroxyapatite nanorods into well-defined spherical clusters within the 3D matrix. Applicability of the organic–inorganic PAH3-Lap hydrogels was assessed in vitro using human bone marrow-derived stromal cells (hBMSCs) and ex vivo using a chick chorioallantoic membrane (CAM) assay. The results demonstrated that the organic–inorganic PAH3-Lap hydrogels promote human skeletal cell proliferation and, upon mineralization, integrate with the CAM, are infiltrated by blood vessels, stimulate extracellular matrix production, and facilitate extensive mineral deposition relative to the controls.


Zeta potential (ζ)
All ζ-potential measurements were performed after resuspension of the PAs at a concentration of 1mM in ultrapure water. After loading the samples in folded capillary cells, measurements were performed at 25°C using a ζ-sizer instrument (Nano-ZS Zen 3600, Malvern Instruments, UK). For each PA, three separate samples were measured with at least five runs per sample.

Circular dichroism spectroscopy
Circular dichroism (CD) was measured with Chirascan™ circular dichroism spectrometer (Applied Photophysics Limited, UK) using quartz cell with 1 mm path length and the following parameters: data pitch -0.5 nm, scanning modecontinuous, scanning speed -100 nm/min, bandwith -2 nm and accumulation -5. All CD data are presented as ellipticity and recorded in millidegree (mdeg). CD measurements were performed on aqueous solutions of PAH3 (0.01% w/v), Lap (0.025% w/v) and their mixtures. CD spectra were obtained by signal integrating 3 scans, from 190 to 260 nm at speed of 50 nm/min. Data were processed by a simple moving average and smoothing method.
The hydrogels of PAH3_Lap were prepared by mixing equal volume of PAH3 (2% w/v) and Lap (5% w/v) in the cuvettes and gelation was allowed to proceed overnight. Synchrotron small-angle neutron scattering (SANS) measurements were performed on the fixed-geometry, time-of-flight LOQ diffractometer (ISIS Neutron and Muon Source, Oxfordshire, UK). A white beam of radiation with neutron wavelengths spanning 2.2 to 10 Å was enabled access to Q [Q = 4πsin(θ/2)/λ] range of 0.004 to 0.4 Å −1 with a fixed-sample detector distance of 4.1 m. The cuvettes were mounted in aluminium holders. Time taken for each measurement was approximately 30 minutes. All scattering data were normalized for the sample transmission, the backgrounds were corrected using a quartz cell filled with D2O and the linearity and efficiency of the detector response was corrected using the instrument-specific software.

S4
For all of the samples, the data at very low Q was ignored for the fitting. A Q-range of 0.005<Q<0.5 Å -1 was used to fit the data. The background level varied between samples. For the fittings, the background was generally fixed and not varied as part of the fit. The data for the Lap alone possesses a Q -1.8 dependency in the range 0.01<Q<0.1 Å -1 , which indicates a disk-shaped object (where one would expect a Q -2 dependency) (see graph Supplementary Figure S2). The data are similar to those reported elsewhere. 1 As discussed in this paper, there is also the start of a Q-independent plateau at approximately Q=0.012 Å -1 , implying that the largest particle dimension is approximately 500 Å. The data were fitted to a cylinder model, with the fit in agreement with thin disks with a thickness of around 12 Å. A SLD of 4.18x10 -6 Å -2 was used on the basis of previous data. Data for PAH3 were fitted to a cylinder model, although it is necessary to include some polydispersity in radius to satisfactorily fit the data. A SLD of 1.51x10 -6 Å -2 was used (from the NIST calculator). For PAH3 combined with Lap, the data can be fitted to a combination of two cylinders (one long and one flat (or disk) cylinder).
Details of the parameters for all the fittings are presented in Table S1.

Atomic Force Microscopy (AFM)
AFM was performed on the Bruker Multimode 8 AFM with a Nanoscope V controller using PeakForce Tapping mode with a ScanAsyst Air cantilever (spring constant 0.4N/m). The cantilever was calibrated using the automated 'no touch' calibration routine built into the software. Solutions of PAH3 (0.01% w/v, 40 µL), Lap (0.025% w/v, 40 µL) and PAH3/Lap mixtures were dropped onto freshly cleaved mica surface. The samples were air dried at room temperature for 24 h and imaged with a PeakForce setpoint of 500pN with a PeakForce amplitude of 30nm and frequency of 4kHz. Images were acquired at 512x512 pixels at a line rate of 2.8Hz. The height images were processed in the Nanoscope Analysis software after using 1 st order flattening to remove tilt. Images were processed in Nanoscope 1.7.

Transmission Electron Microscopy (TEM) and High Resolution TEM (HRTEM)
Aqueous solutions of PAH3 (0.01% w/v) and Lap (0.025% w/v, exfoliated with 0.0068% w/v ASAP) were dissolved ultrapure water. Similarly, mixtures of PAH3 (0.02% w/v) and Lap (0.05% w/v) were also prepared. Samples were mounted on copper TEM plasma etched holey carbon-coated copper grid (Agar Scientific, Stansted, UK). The grids were immersed in the sample solutions for five minutes. Excess was removed on filter paper before incubation with 2% uranylacetate solution for 30 seconds. Grids were then washed with ultrapure water for 30s S5 and air dried for 24h at room temperature. Bright-field TEM imaging was performed on a JEOL 1230 Transmission Electron Microscope operated at an acceleration voltage of 80 kV. All the images were recorded by a Morada CCD camera (Image Systems). At least three images were taken per samples for further analysis.
High resolution transmission electron microscope (HRTEM) images, selected area electron diffraction (SAED) patterns, scanning transmission electron microscope (STEM) images and energy dispersive X-ray spectroscopy (EDS) spectrum images were obtained with an FEI Talos F200X microscope equipped with an X-FEG electron source and Super-X SDD EDS detectors.
The experiment was performed using an acceleration voltage of 200kV and a beam current of approximately 1 nA. TEM Images were recorded with a FEI CETA 4k x 4k CMOS camera.
STEM images were acquired with HAADF and BF detectors.

Molecular dynamic simulations
The dynamic simulation was conducted according to the following steps: Step 1: Building layered cells

Layered sepiolite
A simulation box containing 2 layers of sepiolite was constructed using the Build Layer tool.
The structure was built as a crystal with a vacuum region of 20Å between the layers. A lattice with dimensions of 13.4x26.8x37.64 Å was obtained. The motion of the atoms of the lattice was constrained using the Edit Constraints dialog and choosing Fix Cartesian position. Thus, the XYZ position property in Cartesian space was held constant for the lattice atoms during the dynamic simulations, and only the PAs molecules, which were inserted later, were able to move.

Supercell layered sepiolite
To increase the surface area of the sepiolite crystal a supercell was built. A supercell has lattice vectors which are integral multiples of their equivalents in the original lattice, with a P1 (crystal) symmetry. The supercell was constructed from the sepiolite crystal using the Symmetry tool in the Built function with a factor of 2 in the (a) and (b) directions. Afterwards, a layered supercell sepiolite was constructed according to the procedure described above (at Layered sepiolite). A lattice with dimensions of 26.8x53.6x37.64 Å was obtained.

Insertion of PAs molecules into the layered sepiolite cells
Each of the three PAs examined (PAH3, PAK3 and PAE3) were inserted into the layered cell: • One molecule of each into the small Layered sepiolite cell S6 • Four molecules of each into the Layered sepiolite supercell • Ten molecules of each into the Layered sepiolite supercell After the insertion, the cells were optimized using the Geometry Optimization tool of Forcite module with COMPASII forcefield.
Step 2: Molecular Dynamics simulation Dynamic simulation was performed at 298 K. The cells were subjected to 10 6 dynamic steps of 1fs each at constant moles number, volume and temperature (NVT ensemble). All MD simulations were conducted using Forcite module with COMPASSII force field. Electrostatic and van der Waals terms were considered using Atom based summation methods with a repulsive cut-off of 12.5 Å.

Step 3: Analysis
The resulted dynamic trajectories (1000 frames) were analyzed using Forcite module analysis tools.

Interaction energy
The interaction energy between the clay and PAs can be calculated according to the equation:

E Interaction = E total -(E clay + E PA)
Etotal is the energy of the clay and the PA, Eclay is the energy of the clay without the PA and EPA is the energy of the PA without the clay. These calculations are all single point energies and they were undertaken whilst constraints are removed. A negative value indicates that the PA is binding to the clay.

Characterization of surface properties of xerogels
Nitrogen sorption isotherms of the lyophilized xerogels were measured at 77 K using an Autosorb-IQ system (Quantachrome Instrument, USA). Before measurements, the samples were degassed in a vacuum at 120 °C overnight. The specific surface areas (SBET) were calculated by the multipoint Brunauer-Emmet-Teller method using adsorption data in a relative pressure range from 0.04 to 0.2, and the pore-size distribution was calculated based based on quenched solid density function theory (QSDFT) using the adsorption branches of isotherms assuming slit and cylindrical pores geometries. By using the Barrett-Joyner-Halenda (BJH) model, the mesoporous surface areas (SBJH) were calculated from the adsorption line. The microporous surface areas (SDR) were calculated from the adsorption line by the Dubinin-Radushkevich (DR) model.

Biomineralization of hydrogels
The mineralizing solutions were prepared as previously reported by Elsharkawy et al. 2 Briefly, an aqueous suspension of hydroxyapatite powder (2 mM) and sodium fluoride (2 mM) was prepared in deionized water with continuous stirring. Then, 69% nitric acid was added dropwise to the suspension to aid a complete dissolution of the hydroxyapatite precipitates at pH 2.4. Thereafter, an aqueous solution of ammonium hydroxide (30%) was added dropwise to the hydroxyapatite solution until it reaches pH 6. Various hydrogels were then immersed in the hydroxyapatite solutions and incubated for eight days at 37 °C using a temperaturecontrolled incubator (LTE Scientific, Oldham, UK).

S8
All Raman analysis was carried out on a confocal WITEC Alpha300 system utilising a 785 nm laser and a 20× (S Plan Fluor, NA 0.45, ELWD) objective lens. Raman scatter was collected in a backscattering geometry. A small amount of each sample was placed on a microscope glass slide which had been previously cleaned with a methanol-soaked tissue, with a new slide used for each sample. The incident laser power was constant for all samples at 63 mW. No signal loss was observed, for example due to photobleaching or carbonisation, when samples were irradiated on the same spot in triplicate with integration times ranging from 10 s to 60 s. All spectra processing was performed using SpectraGryph 1.2 involving: 1. cosmic ray removal, 2. background correction and then 3. subsequent normalization. An advanced baseline correction protocol available in the SpectroGryph software was applied which fits a polynomial curve to the spectral regions where there is no Raman peak and enables subtraction of the variable y-offset associated with the luminescence background. To enable comparison of the relative changes in the Raman intensity of the 1047 cm -1 peak in Figures 6 and S10, all spectra were normalized with respect to the peak intensity in the 2800-3000 cm -1 region. This approach was adopted as the integration time was varied between samples to optimize the signal to noise alongside variation in background luminescence with mineralization times. However, the C-H vibrational spectral shape across 2800-3000 cm -1 remained relatively unchanged for each sample and the Raman peak intensity was also observed to change proportionally with integration time in this region. For each measurement, multiple spectra were acquired across the sample with the focus depth also optimized, which revealed good uniformity and ensured that the spectra presented are representative of the sample.

Synthesis and purification of peptide amphiphiles
The peptide amphiphiles (PAs) were synthesized using solid phase peptide synthesis (SPPS)

Implantation, extraction and Chalkley score
Animal studies were performed in accordance with the guidelines and regulations laid down in the Animals (Scientific Procedures) Act 1986. CAM model was carried out in accordance with Home Office Approval, UK (Project license -PPL P3E01C456). Chicken eggs were acquired from Medeggs (Norfolk, UK). Eggs were stored in Hatchmaster incubator (Brinsea, UK) at 37 °C in a 60 % humidified atmosphere and 1 h rotation. To ensure the maintenance of a humidified environment in the egg incubator deionised water (DW) was supplemented every two days. Implantation was carried out after 7 days of incubation. To assess embryo viability and development eggs were candled. A window of 1cm 2 was created with a scalpel onto the egg shell exposing the chorioallantoic membrane. Hydrogels were implanted and the window sealed with a sterile Parafilm strip (Bemis™, Parafilm M™, Laboratory Wrapping Film, Fisher Scientific, UK). Eggs were return to the Hatchmaster incubator for 7 days (37 °C in a 60 % humidified atmosphere) without rotation. Chalkley scoring was used as previously described 3 to quantify infiltration of blood vessels through the implanted scaffolds. Implants and blank controls were observed in situ under a stereo light microscope. A total of five independent counts obtained from the number of vessels fitting with the Chalkley graticule projected onto the samples were registered.

Histological analysis
Integrated hydrogel samples were extracted and fixed in 4 % paraformaldehyde (PFA) overnight. Samples were further embedded in optimum cutting temperature (OCT embedding matrix, CellPath, UK) and stored at -80 °C. Samples were sectioned using a Cryostat (CM 1850, Leica Biosystems, Germany) and 8µm thick sections were collected using Kawamoto's film method. 4 Stainings (Goldner's Trichrome and Von Kossa) were subsequently carried out on the cryotape. Sections were mounted using Super Cryomounting Medium (SCMM) type R3 (Section LAB, Co. Ltd. Japan) and UV cured for 30 min to photo-polymerise the SCMM.
Slides were imaged the following day using a Zeiss Axiovert 200 (Carl Zeiss, Germany).   Figure S2. HRTEM-EDS mapping showing bright field image of PAH3-Lap nanofiber-nanodisk co-assembly, which is evident from co-localization of nitrogen N (mainly from PAH3) and oxygen (O), magnesium (Mg), sodium (Na) and silicon (Si) from Lap. Supplementary