Gold Nanoparticle Aggregation as a Probe of Antifreeze (Glyco) Protein-Inspired Ice Recrystallization Inhibition and Identification of New IRI Active Macromolecules

Antifreeze (glyco)proteins are found in polar fish species and act to slow the rate of growth of ice crystals; a property known as ice recrystallization inhibition. The ability to slow ice growth is of huge technological importance especially in the cryopreservation of donor cells and tissue, but native antifreeze proteins are often not suitable, nor easily available. Therefore, the search for new materials that mimic this function is important, but currently limited by the low-throughout assays associated with the antifreeze properties. Here 30 nm gold nanoparticles are demonstrated to be useful colorimetric probes for ice recrystallization inhibition, giving a visible optical response and is compatible with 96 well plates for high-throughout studies. This method is faster, requires less infrastructure, and has easier interpretation than the currently used ‘splat’ methods. Using this method, a series of serum proteins were identified to have weak, but specific ice recrystallization inhibition activity, which was removed upon denaturation. It is hoped that high-throughput tools such as this will accelerate the discovery of new antifreeze mimics.


Synthesis and Characterisation of Poly(vinyl alcohol), PVA.
The synthesis of PVA used in this study, was conducted as described in Congdon et al 1 . The method is explained below and shown in Figure S1. Table S1 provides characterization details of the polymers selected for this study Figure S1. Schematic of the synthesis of PVA using RAFT polymerisation.
The yellow solutions were then cooled to room temperature. Poly(vinyl acetate) was then recovered as a yellow sticky solid after precipitation into hexane.

Determining Extent of Aggregation
The change in absorbance at 520 nm was identified as the simplest measurement of aggregation in this study. To account for scattering of aggregated samples, we conveniently measured the difference between the actual absorbance at 520 nm and that of a baseline define between 450 and 680 nm as show in Figure S4.

Variable volume gold particle aggregation experiments
An advantage of the AuNP method is that samples can be tested at a range of different volumes. The rationale for this was that larger volumes would take longer to freeze/thaw, which should increase the amount of ice recrystallization inducing more aggregation and enabling optimization so that there are clear differences between IRI active and non-active compounds. Furthermore, the standard 'splat' assay only uses 10 µL of liquid, which is far from the volumes employed in applications such as cryopreservation, and therefore larger volumes may provide a more predictive test for the additives ultimate application. To this end samples of 50, 100, 200, 500, 1000 and 2000 µL where prepared in 96 and 24 well plates and subjected to the same freeze/thaw cycle as detailed within the paper. The sample depth (which is crucial for freezing rate) verses the volume of the samples is plotted in Figure S4. From Figure S4, samples with lower well depth provide a generally lower AR (i.e. closer to 1), due to the more rapid thawing that would occur, thus demonstrating that  the thawing rate is important for this assay. Likewise the samples of thicker well depth also seem to show a higher AR possibly due to the depth of sample leading to a greater level of light absorbance. The depth-dependence also suggests that this assay is probing ice recrystallization and not ice shaping, which was previously suggested.
Overall the clearest difference between IRI active and no active compounds is found at a well depth of 3.118mm, which corresponds to 100 µL in a 96 well plate. It is therefore recommended that 50 µL of sample to 50 µL of AuNPs making an overall volume of 100 µL be used to investigate the IRI activity of compounds.

Comparison of Absorbance Change to Mean Largest Grain Size for Poly(Amino ethyl methacrylate)-co-succinic anhydride
Since PAEMA-co-SA has also been shown to exhibit IRI activity the absorbance has been plotted here against mean largest grain size (MLGS) (as a percentage of PBS buffer negative control), Figure S5. Figure S6. Comparison of PAEMA-co-SA polymers mean largest grain size (MLGS) and absorbance at 20mg/ml. MLGS = mean largest grain size relative to a PBS control, expressed as %. Error bars represent ± SD from a minimum of 3 repeats.
Plotting the MLGS absorbance change against each other on the same graph shows a linear correlation and reinforces the premise that the AuNP method is probing the same phenomenon as the "splat test" namely ice recrystallization inhibition.  From Figure S7 it can be seen there is a reduction in absorbance across the whole spectra as AuNP concentration decreases. However the peak at around 550 nm remains at all concentrations showing that PVA prevents the aggregation of AuNPs.
Furthermore if no PVA is added, this 550 nm peak disappears at all concentrations of AuNPs. To maximize the difference between 520 nm and 650 nm a minimum AuNP concentration of 80 µgmL -1 is optimal to avoid wastage and still function in the assay.