Supramolecular Amino Acid Based Hydrogels: Probing the Contribution of Additive Molecules using NMR Spectroscopy

Abstract Supramolecular hydrogels are composed of self‐assembled solid networks that restrict the flow of water. l‐Phenylalanine is the smallest molecule reported to date to form gel networks in water, and it is of particular interest due to its crystalline gel state. Single and multi‐component hydrogels of l‐phenylalanine are used herein as model materials to develop an NMR‐based analytical approach to gain insight into the mechanisms of supramolecular gelation. Structure and composition of the gel fibres were probed using PXRD, solid‐state NMR experiments and microscopic techniques. Solution‐state NMR studies probed the properties of free gelator molecules in an equilibrium with bound molecules. The dynamics of exchange at the gel/solution interfaces was investigated further using high‐resolution magic angle spinning (HR‐MAS) and saturation transfer difference (STD) NMR experiments. This approach allowed the identification of which additive molecules contributed in modifying the material properties.


pH of suspensions
pH values were monitored upon the addition of other amino acids to suspensions of Phe in water, due to its importance in dictating noncovalent interactions and, therefore, self-assembly processes. No significant differences were found between the samples under study. Table S1. pH values of suspensions of Phe, Phe/Leu (5:1), Phe/Ser (5:1), Phe/Trp (5:1) and Phe/Tyr (5:1) and dissociation constants for the molecules under study.

Temperature of gelation
The temperature above which hydrogels were broken (i.e. loss of structural integrity reflected by ability to flow under inversion) was defined as the gel-to-solution transition (Tgel). Temperature of gelation was determined by heating up the hydrogel samples with a hot plate at a heating rate of 1 K min -1 . Hydrogelation was then assessed through the vial inversion test.

Thermogravimetric analysis
Thermogravimetric analysis (TGA) was carried out using a TGA Q5000 TA instrument by placing ca. 2 mg of dried samples onto platinum pans. Samples were heated from 298 to 423 K with a heating rate of 2 K min -1 . The weight loss observed in thermogravimetric analysis of dried hydrogels was assigned to evaporation of physisorbed water retained after the drying process.

Atomic force microscopy
Widths of fibrillar features were measured directly from AFM images in intermittent contact mode. n = 83 measurements, taken from 6 different images (5 x 5 μm, 512 pixels per line). A normal distribution fitted to this histogram had a peak value of 437 ± 16 nm. Figure S1. Histogram of measured widths in AFM experiments for the hydrogel of Phe, with an average width of 437 ± 16 nm.

Rheology
Phase angle, storage and loss moduli were monitored and recorded as a function of stress. All samples were subjected to stress amplitude sweeps in the range of 500 to 10000 Pa. An oscillatory torque was imposed with a fixed frequency over a range of shear stress amplitudes. The hydrogel showed a typical G' value, essentially constant below the critical value of oscillatory torque (''yield stress''). At this yield stress point, the sample starts to flow or there is slippage between the interface of the rheometer and the hydrogel. No trends or conclusions could be drawn from this data due to this latter fact. Hydrogel materials often exude water (syneresis) resulting in uncontrollable slippage. Attempts were made to act against this but inconsistent data with little or no trends was obtained. The data below show general results with gels showing little differences, within error of normal data collection on the rheometer utilised for experiments. Figure S2. Storage modulus (G') at increasing stress sweeps for the hydrogels of Phe, Phe/Leu (5:1), Phe/Ser (5:1), Phe/Trp (5:1) and Phe/Tyr (5:1).

Microscopy
Polarised light microscopy was carried out using a Leica DMLS2 with ×20 magnification coupled to a JVC colour video camera. 20 µL of hot solutions were pipetted onto a glass slide and allowed to gelate in situ.
The photomicrographs of the hydrogel of Phe/Tyr showed the presence of crystalline needles, which were attributed to insoluble crystals of Tyr. This phase was also detected using PXRD and 1 H-13 C CP/MAS NMR. 8. NMR spectroscopy 8.1. Solid-state NMR spectroscopy Hydrogels were prepared by pipetting 40 μL of hot solutions into Kel-F plastic inserts and allowed to cool down and gelate inside the insert. Dried hydrogels were packed inside zirconia rotors. 1 H-13 C CP/MAS NMR spectra of hydrogels were acquired using 8192 scans and an MAS rate of 8.5 kHz with a recycle delay of 20 s and contact time of 2 ms. 1 H-13 C CP/MAS NMR spectra of dried hydrogels were acquired using 2048 scans and an MAS rate of 10 kHz with a recycle delay of 20 s and contact time of 2 ms. 1 H-13 C CP/MAS NMR spectra were acquired on both wet and dried hydrogels, to compare the consequences of drying in the native tridimensional organisation of the hydrogels. Increased signal-to-noise was observed for dried materials. Interestingly, similar chemical shift values and peak splitting patterns were recorded in both physical states, indicating that experiments conducted with dried samples are able to reproduce the original structure of the hydrogel fibres in study.  C CP/MAS NMR spectra of hydrogels of Phe, Phe/Leu (5:1), Phe/Ser (5:1), Phe/Trp (5:1) and Phe/Tyr (5:1) acquired with MAS rates of 8.5 kHz. Carbon labelling scheme is shown for clarity.

Solution-state NMR spectroscopy
Peak intensities from 1 H-NMR spectra can be correlated with concentration of diluted species when long enough recycle delays are applied. This can be used to derive the ratio between gelator molecules in the isotropic phase and solution-state NMR "invisible" molecules forming the rigid fibres of supramolecular hydrogels. [2] The observed line broadening results from the anisotropy imposed to the gelator molecules by their incorporation in partially mobile structures. For these Phe-based hydrogels, 40-50 % of gelator molecules are structural components of the fibres. The calculation of fraction of 1 H peak intensity and variation of full width at half maximum are highly dependent on the time of the acquisition of the initial spectra and on the kinetics of gelation (which can be affected by sample volume and temperature of the NMR tube).

1 H-1 H 2D NOESY experiments
Negative nOe enhancements (blue) were detected in 1 H-1 H 2D NOESY NMR spectra for Phe protons in Phe-based hydrogels ( Figure  S9a-e). Negative cross-peaks are characteristic of large molecules which transfer magnetisation efficiently through dipolar interactions. [2] Since these Phe-based hydrogel systems are composed exclusively of LMW species, these findings indicated that molecules in solution contain information from the fibrous network due to their fast dynamics of exchange in the NMR frequency time scale. Strong negative cross-peaks were also recorded between Phe and Trp or Tyr ( Figure S9d and e), in hydrogels of Phe/Trp and Phe/Tyr, supporting that Trp and Tyr were in close proximity with Phe due to their incorporation in the hydrogel fibres. The presence of weak negative cross-peaks between Phe and Leu ( Figure S9b) indicated spatial proximity between both molecules. Despite the evidence for fast interaction of Leu with Phe at the gel/solution interfaces, the detection of positive spatial correlations between Leu protons (green), associated with small molecules, showed this molecule exists mainly in a free dissolved state. These findings were in agreement with the sharp Leu peaks ( Figure S8) and the unmodified Leu peak integral (Table S6) observed for Leu after gelation. No cross-peaks were observed between Phe and Ser in the hydrogel of Phe/Ser ( Figure S9c).

STD NMR experiments
STD NMR spectroscopy is applied frequently to identify the functional groups of a ligand responsible for binding to its receptor (a protein, typically). [3] This method relies on the transfer of saturation through cross-relaxation from a large saturated protein to a small bound ligand. [4] For amino acid based hydrogels, we can consider the network as the supramolecular entity that can be saturated selectively. Considering an analogous dependence of ηSTD with concentration as in the case of protein-ligand studies: -, S1 Where is the signal intensity from the difference spectrum, 0 is the signal intensity from the STDoff spectrum and is a dimensionless scaling factor. [5] We considered that the concentration of the network-bound gelator, Net-, and the total gelator tion, , were equivalent to the concentration of protein receptor-ligand complex, L , and the total ligand concentration, L , respectively. [5] Figure S11. Build-up curves of ηSTD in hydrogels of a) Phe, c) Phe/Leu (5:1), d) Phe/Ser (5:1), e) Phe/Trp (5:1) and f) Phe/Tyr (5:1) acquired at 298 K (STDon = 0 ppm and STDoff = 40 ppm). b) Initial slope values recorded from 298 to 338 K upon saturation of the network (STDon = 0 ppm and STDoff = 40 ppm) in the hydrogel of Phe.