Solvent‐free liquid avidin as a step toward cold chain elimination

The temperature sensitivity of vaccines and therapeutic proteins forces the distribution of life‐saving treatments to rely heavily on the temperature‐controlled (usually 2–8°C) supply and distribution network known as the cold chain. Here, using avidin as a model, we demonstrate how surface engineering could significantly increase the thermal stability of therapeutic proteins. A combination of spectroscopic (Fourier transform infrared, circular dichroism, and ultraviolet‐visible) and scattering techniques (dynamic light scattering, small‐angle, and wide‐angle X‐ray scattering) were deployed to probe the activity, structure, and stability of the model protein. Temperature‐dependent synchrotron radiation circular dichroism spectroscopy was used to demonstrate a significant increase in thermal stability, with a half denaturation temperature of 139.0°C and reversible unfolding with modified avidin returning to a 90% folded state when heated to temperatures below 100°C. Accelerated aging studies revealed that modified avidin retained its secondary structure after storage at 40°C for 56 days, equivalent to 160 days at 25°C. Furthermore, binding studies with multiple ligands revealed that the binding site remained functional after modification. As a result, this approach has potential as a storage technology for therapeutic proteins and the elimination of the cold chain, enabling the dissemination of life‐saving vaccines worldwide.

challenge in the processing and handling of therapeutics: a small percentage of aggregated molecules can elicit undesired immunogenicity (Bhatnagar et al., 2007). To prevent aggregation, these biomolecules are refrigerated (2-8°C) or frozen (−20°C or −50°C), in the temperature-controlled supply chain from manufacturing to administration known as the "cold chain." Several challenges must be overcome for an effective cold chain, these include; poor temperature control and maintenance leading to the reduced potency of vaccines (Ashok et al., 2016), and the considerable cost of maintaining control temperatures, which can account for up to 80% of vaccination program costs (Pelliccia et al., 2016). As such, developing countries and resource-limited areas, with little infrastructure or refrigeration facilities, are unable to benefit fully from advances in the production of therapeutic antibodies and vaccines.
Many strategies to overcome the limitations of the cold chain have been developed to produce more thermally stable vaccines.
These include freeze-drying , immobilization of viral particles onto carbohydrate glass (Alcock et al., 2010), the addition of sugars to vaccine formulations (Alcock et al., 2010;Croyle et al., 1998;Evans et al., 2004;Gupta et al., 1996;Rexroad et al., 2003;Stewart et al., 2014), and encapsulation in silica (Y.-C. Chen et al., 2017). Despite the variety of strategies proposed, there are still significant hurdles as these technologies frequently do not raise the thermal stability sufficiently, and therefore still require storage between 2°C and 8°C (Rexroad et al., 2002), or come with undesirably complex reconstitution processes. In addition, when vaccines are exposed to freezing temperatures, the adjuvants in current formulations used to increase the efficacy of vaccines forms aggregates that can produce an unwanted immunological response (Hanson et al., 2017). Consequently, there is a necessity to develop new stabilization strategies.
Solvent-free biofluids are a new class of biomaterial involving the engineering of protein surfaces with a protective polymer-surfactant corona (A. P. S. Brogan et al., 2014aBrogan et al., , 2014bBrogan et al., , 2012Gallat et al., 2012;Perriman et al., 2010). These biofluids have been shown to retain both biomolecule three-dimensional structure and activity, and can be used as a solvent for the dissolution of anhydrous solutes (A. P. S. Brogan et al., 2018;A. P. S. Brogan & Hallett, 2016;A. P. S. Brogan et al., 2013Mukhopadhyay et al., 2018;K. Sharma et al., 2020). In addition, the biofluid conferred extremely high thermal stability on the protein, allowing for enzyme activity at temperatures as high as 150°C (A. P. S. Brogan et al., 2014b). We now wish to use the advantage of the thermophilic behavior of solvent-free biofluids to extend this technology to the long-term room temperature storage of therapeutic proteins.
Here, we demonstrate the feasibility of therapeutic biofluids using the globular binding protein avidin (Av) as a simplified model for therapeutic proteins (analogous ligand recognition and binding).
Antibodies can bind up to two ligands with high binding affinity and typically have a dissociation constant (K d ) between 1 × 10 −11 M and 2 × 10 −10 M (Landry et al., 2015). Av is a tetramer that can bind to the vitamin biotin with a dissociation constant K d of 6 × 10 −16 M (Green, 1990), the highest known noncovalent interaction between a protein and a ligand. Av has four independent binding sites with the same K d , which can be used to monitor any changes in binding strength due to its high affinity with biotin. We have successfully produced a proteinpolymer surfactant nanoconstruct from Av with a melting temperature of 139.0°C and retained secondary structure and binding activity after storage at 40°C for 56 days, equivalent to 160 days at 25°C. Our results show that the robust synthesis procedure provides the blueprint for engineering thermally stable antibodies and viruses for life-saving treatments worldwide, negating the requirement for the costly and flawed cold chain.

| RESULTS AND DISCUSSION
Av was successfully surface-engineered using established methods, with slight modifications for optimization of the protocol for the Av system, to produce discrete nanoconjugates of protein and polymersurfactant (A. P. S. Brogan et al., 2018;A. P. S. Brogan & Hallett, 2016;A. P. S. Brogan et al., 2014b. This two-step process first involved the cationization of Av (at an efficiency of 63%, Figure S1

| 593
Cationization caused little change in the secondary structure of avidin, with C-Av having a predominately β-sheet structure (46%), comparing well to Av (51%; Figure 2a; Table 1). However, after electrostatic complexation of surfactant molecules to the protein surface to produce [C-Av][S], there was a broadening of the negative band at 215 nm. In addition, the positive peak at 230 nm found in the native and cationized form disappeared, and the positive peak at 196 nm blue-shifted to 193 nm ( Figure 2a). Deconvolution of the spectra showed that β-sheet content reduced to 37%, which was concomitant with an increase in α-helical structure to 16%, indicating a shift in the global fold of Av (Table 1). Small-angle X-ray scattering F I G U R E 1 (a) Visual molecular dynamics (Humphrey et al., 1996) Table S1). Scattering plots for Av, C-Av, and   was incubated at 25°C and 40°C for 56 days that there was no structural degradation. The deconvolution of the CD spectra ( Figure   S9a,b) showed that the proportion of secondary structure remained unchanged after being incubated at 25°C and 40°C, further proof that the secondary structure of [C-Av][S] in the solvent-free form was retained. However, when incubated at 60°C, there was a slight but persistent decrease in Δ[θ] MRE over 56 days ( Figure S8c). The deconvolution of the CD spectra ( Figure S9c) showed that after incubation at 60°C for 56 days that the α-helix content decreased from 16% to 6% while the β-sheet and unordered secondary structure increased from 37% to 43% and 34% to 37%, respectively. A temperature coefficient (Q 10 ), the rate of change of degradation as a result of an increase in 10°C, is typically between 2 and 3 for most biological reactions (Reyes et al., 2008). Therefore, a Q 10 of 2 was assumed to determine the minimum long-term thermal stability at  To investigate potential binding site blockage further, we turned to the ligand displacement reaction between HABA and biotin (Figures 4c,d and S13-S15). The reversibility of the binding pocket was investigated by displacing HABA (λ max = 350 nm) from the Av-HABA complex (λ max = 500 nm) with biotin. The absolute data showed that Av possessed a K d of 1.12 × 10 −5 M (Table S6) Figure S11). Absolute (c) and normalized (d) HABA occupancy for HABA to binding site ratio up to 32 in 100 mM phosphate buffer pH 7 at 25°C (data from Figure S13). For aqueous Av (black), C-Av (

DATA AVAILABILITY STATEMENT
All data are archived at the Imperial College London online library.
Please consult the authors for details.