ADDovenom: Thermostable Protein-Based ADDomer Nanoparticles as New Therapeutics for Snakebite Envenoming

1. Introduction

Snakebite envenoming is a Neglected Tropical Disease that is annually responsible for up to 138,000 deaths and 400,000 disabilities in surviving victims [1], disproportionately affecting the most economically and medically disadvantaged communities of Asia, sub-Saharan Africa and Latin America [2]. The first-choice treatment for snakebite is intravenously-delivered antivenom, which is produced by hyperimmunising equines or ovines with subtoxic doses of venom(s) and then purifying immunoglobulins from resulting sera/plasma samples. Consequently, the resulting antivenoms are associated with numerous deficiencies, including limited cross-snake species reactivity, poor dose-efficacy, and high incidence of adverse reactions [3]. The efficacy of current antivenom is often restricted to the snake venom(s) used as immunogens in the manufacturing process due to the significant variation in toxin composition observed between the venoms of different snake species, though venom components can even differ between different individuals of the same species or over the lifetime of a single snake [4,5]. Snake venoms typically contain mixtures of functionally distinct protein isoforms encoded by relatively few toxin families, and are biochemically diverse secretions that cause variable pathologies in snakebite victims (i.e. haemotoxicity, neurotoxicity and/or cytotoxicity) [1]. Venom toxin variation therefore underpins the restrictive cross-snake species efficacy of antivenom, which in turn translates into the limited geographic utility, reduced economies of scale and poor commercial manufacturing incentives associated with these life-saving therapeutics [6].

Most regions affected by snakebite envenoming are home to several species of medically-important snakes, and therefore most antivenom manufacturers use venom from multiple snake species to produce polyspecific (polyvalent) antivenoms for the target region [7]. This inherently results in a smaller proportion of antibodies in the antivenom directed against the toxins from any one species in comparison to antivenoms raised against only one venom (“monospecific” or “monovalent” antivenom), and therefore much higher doses of polyspecific antivenoms are required to effectively treat envenoming [8]. Further compounding the poor dose-efficacy of antivenoms is evidence that only 5-36% of immunoglobulins purified from the animal sera/plasma are specific to the venom proteins used for immunisation [9,10,11,12], and only a proportion of those are likely to be neutralising antibodies. The weak potency of antivenoms, which often require multiple vials to effect cure, increases the risk of adverse reactions to the large quantity (often grams) of intravenously-administered, non-human immunoglobulin, while also resulting in increased costs to the patient. Another consideration regarding the accessibility of antivenoms is cold-chain transport and stable low temperature storage requirements, which frequently prevents their effective distribution in peripheral rural health centres close to the populations that have the greatest need [13].

Snakebite was recognised as a priority neglected tropical disease by the World Health Organisation (WHO) in 2017 due to the significant public health problem it presents across most of the rural tropics. A number of novel snakebite treatments are under development: monoclonal human/humanised antibodies (or cocktails thereof), non-antibody toxin-binding proteins or peptides, small molecule drugs, and aptamers (for reviews on these therapeutic modalities for snakebite see [14,15,16,17]). These treatments aim to be better tolerated, more affordable, and possess broader snake species-utility than existing antivenoms; however only two are currently in clinical trials [18,19]. Thus, there remains a strong need to explore alternative biological therapeutics that may provide the next breakthrough in snakebite treatment. Here, we describe the proposed development of ADDomer - a novel protein-based nanoparticle therapeutic for snakebite - in our so named ‘ADDovenom’ project.

2. ADDomer – a versatile thermostable nanoparticle

ADDomer is a nanoparticle scaffold derived from the Adenovirus penton base protein protomer [20]. In Adenovirus, the protomer forms pentons at the vertices of the capsid, providing a base for fibre protein attachment [21]. When expressed recombinantly in isolation, 60 protomers spontaneously self-assemble into pentons that form a dodecahedron, known as ADDomer (for Adenovirus-Derived Dodecamer) (Figure 1a,b). ADDomer-based nanoparticles have been developed as efficient vaccines, e.g. against Chikungunya, SARS-CoV-2 or Foot-and-mouth disease [20,22,23,24]. The penton base protein protomer comprises two domains, a jelly roll fold and a crown domain (Fig 1c). The jelly roll fold mediates multimerisation into the penton and dodecamer assembly. The crown domain comprises two loops on its surface, the variable loop (VL) and the RGD loop (Figure 1c). The VL and RGD loops are highly variable in sequence and length among the penton base proteins found in different Adenoviruses. Immunogenic epitopes and other sequences can be inserted into these loops without compromising efficient folding of the penton base protein and assembly into the ADDomer [20]. The immunogenic epitopes are then displayed on the surface of the ADDomer in 60, 120 or even more copies, depending on the number of epitopes inserted into VL and RGD loops [20].

ADDomer nanoparticles have a number of key advantages that make them uniquely suitable for deployment in developing countries:

• Thermotolerance up to >45 °C [20,25]. It was shown that the ADDomer nanoparticle can be stored for one month at ~20°C, it can be frozen and thawed and/or heated to 45 °C for 1 hour, virtually without losing its structural integrity [20]. Consistently, ADDomer does not require cold-chain storage which constitutes a key logistic advance.
• ADDomers can be lyophilised [25]. This manufacturing step increases storage life, providing a commercial incentive for manufacturers.
• ADDomers can be produced as recombinant protein-based nanoparticles with exceptionally good yields using a baculovirus insect cell expression system, allowing establishment of good manufacturing practices with stringent quality control.
• Importantly, of all the Adenovirus capsid proteins, the penton base protein is the least immunogenic. Therefore, ADDomer presents a low immunogenicity scaffold.

The aforementioned characteristics of the ADDomer scaffold – thermotolerance, shelf-life, recombinant production, low immunogenicity – are equally desirable for antivenoms. We found that the architecture of the crown domain with its highly variable loops has remote similarities with complementarity determining regions (CDR loops) in antibodies, used to recognise cognate antigens. We hypothesised that the crown domain could be functionalised to bind toxins. Then, the ADDomer could act as a superbinder ‘sponge’ to neutralise toxins. A prerequisite for such a superbinder is the feasibility of a binder molecule representing the crown domain of the protomer. The bipartite architecture of the protomer suggested a possibility to separate crown from jelly roll fold. Thus, we designed ADDobody, comprising only the crown domain and produced the prototype in Escherichia coli with excellent yields as a monomeric protein (Figure 1d) [26]. Next, the VL and the RGD loops of the crown domain were randomised in length and sequence, mimicking the CDR loops of antibodies (Figure 1d). The resulting synthetic ADDobody library is used for ribosome display in vitro selections (Figure 1e) against native and recombinantly produced toxins. Selected specific binders are biophysically characterised and tested for toxin neutralisation using enzymatic or cell-based assays. In a next step, the selected toxin-neutralising ADDobody can be re-joined with the jelly roll fold multimerisation domain to produce the ADDomer-based superbinders [26] (Figure 1). The resulting antitoxin nanoparticle displays 60 binding domains against the toxin target, and therefore may ensure high-avidity binding and highly efficient neutralisation of the snake toxin targets.

3. The EU-funded ADDovenom project

In the ADDovenom project, we will focus on neutralising venoms from two medically important groups of African snakes that cause severe life threatening envenoming across the continent, specifically the saw-scaled vipers (Viperidae: Echis spp.) and the mambas (Elapidae: Dendroaspis spp.) [27]. Echis saw-scaled vipers are responsible for the largest numbers of bites and deaths in the northern half of Africa, and their venom mainly causes life-threatening haemorrhage and coagulopathy and debilitating local tissue necrosis [28]. These effects are caused by toxins such as the snake venom metalloproteinases (SVMPs), phospholipases A₂ (PLA₂s), serine proteases, C-type lectin-like proteins and disintegrins [29,30]. Contrastingly, envenoming by Dendroaspis mambas causes neuromuscular paralysis, which may become rapidly lethal if respiratory paralysis occurs [27]. Their venom contains several distinct neurotoxins [31], which are mostly members of the three-finger toxin (3FTx) and Kunitz-type toxin families [32,33].

We seek to develop ADDomers and ADDobodies with neutralising capabilities against Echis SVMPs, PLA₂s, and disintegrins, and mamba 3FTx and Kunitz-type toxins, using approaches shown in Figure 2. First, venom mass spectrometry will be performed to define the venom toxin content of each species. Proteomics will identify the protein sequence of priority toxins to be targeted, their post-translational modifications, disulphide bonds, and relative abundance in each venom. Priority toxins will then be expressed as recombinant toxins using bacterial and eukaryotic expression systems. A naïve synthetic ADDobody library and recombinant toxins will be used to select for specific, high-affinity toxin-binders by ribosome display. Resulting ADDobodies and ADDomers (multimerised ADDobodies) will then be tested for their in vitro and in vivo neutralising capabilities against specific venom toxins and crude venom, and will also be assessed in vivo for their safety and pharmacokinetic characteristics. Finally, a good manufacturing practice (GMP)-compatible platform will be developed for the production of toxin-neutralising ADDomers.

3.1. Venom mass spectrometry and bioinformatics

One key challenge for the design of toxin-specific therapies is that it is often not apparent which specific isoforms within a family of toxins must be neutralised to prevent mortality and morbidity. The ADDovenom project will take advantage of the versality of mass spectrometry approaches [34,35,36,37] to determine the inventory of peptides and proteins present in the nine African snake venoms collected from saw-scaled vipers (E. ocellatus, E. pyramidum leakeyi, E. leucogaster, E. coloratus) and mambas (D. polylepis, D. angusticeps, D. viridis, D. jamesoni jamesoni and D. j. kaimosae) to be used in this project. From the peptide sequences, the prediction of toxin families and their related pharmacological activities will enable a ranking of toxins according to their toxicological importance to generate a list of ‘priority’ target toxins to be neutralised for each venom. A toxin database containing all this information will be generated to identify the target toxins for recombinant expression (see Section 3.2).

These objectives will be achieved through application of venomics, a methodology that integrates data generated from transcriptomics and proteomics [38,39,40]. This rational approach has already been demonstrated to be particularly powerful to explore the remarkable toxin arsenal of animal venoms. Snake venoms are typically composed of a combination of tens to hundreds of different components, mostly peptides and proteins (>90%), that vary between and within snake species [41,42]. Consequently, identifying the most potent toxins with proteomics can help identify targets for the production of recombinant antivenoms, which represent promising new approaches to treating snakebites.

In the ADDovenom project, we will characterise the crude venom complexity of nine medically important African snakes, as summarised in Figure 3. Each venom will be analysed by two-dimensional sodium dodecyl sulphate polyacrylamide gel electrophoresis, leading to a rough evaluation of their structural family. Accurate masses of the most concentrated toxins are obtained by analysing the venoms by Matrix-Assisted Laser Desorption/Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF MS). In addition to this visualisation of venom composition, the results will be exploited to roughly evaluate the abundance of each detected family of toxin aiming to extract information on the pharmacological activity in each venom that requires neutralising. Each venom is then subjected to the bottom-up proteomics strategy: peptides sourced from digested toxins are sequenced by liquid chromatography (LC)-MS/MS using Orbitrap analysers, exploiting their high resolution (70,000 at m/z 200), mass accuracy (less than 1 ppm at 2 m/z) and fragmentation efficacy (Higher Energy Collision Dissociation - HCD) for toxin sequencing and identification.

Several methodologies can be considered to increase the efficiency of the sequencing step. The first approach is based on bottom-up shotgun proteomics where each crude venom is chemically treated to reduce and alkylate the cysteines responsible for disulphide bonds formation. The samples are then digested by three enzymes (trypsin, chymotrypsin and GluC serine protease) to amplify the level of information gathered through the MS/MS process, by creating overlapping peptides [43]. The huge amount of proteomics data is then analysed by dedicated bioinformatics tools such as Peaks X (Bioinformatics Solutions) [44]. This software allows the prediction of sequences from MS/MS spectra (de novo sequencing), but also enables matching of protein annotations to the obtained mass spectra/sequences via protein databases extracted from the Uniprot or NCBI servers (e.g. “toxins” AND “snake”). In the case of ADDovenom, the transcriptomics data from mRNA sequencing of the snake venom glands are included to generate a more precise and complete protein database to match with the proteomics data [42,45].

Every toxin sequenced from the nine venoms is added to a home-made database containing; toxin name, Uniprot accession number (if known), sequence, mass, toxin family, relative toxin abundance and, where available, predicted/putative biologic activity. This database constitutes a critical tool essential for the identification and selection of the most potent toxins to be targeted by the ADDobodies/ADDomers produced in this project.

In addition, evaluation of what the produced antitoxin molecules are binding is of prime importance. A methodology allowing the determination at the molecular level of which families of toxins are captured by functionalised ADDomers (qualitative) and how many ADDomers are needed for neutralising a defined quantity of injected venom (quantitative) is desired. Using antivenomic principles [38,46], we will establish a high-throughput methodology based on magnetic beads coated, in the first instance, with antibodies sourced from gold standard antivenoms as a proof of concept, followed by specific ADDomers targeting toxins. After venom incubation, mass spectrometry will be applied to monitor which toxins are captured by the antibodies or ADDomers. This approach will provide rapid and robust evaluation of the potency of antitoxins to selectively and quantitatively bind to targeted toxins. Furthermore, it will provide an evidence base as to whether ADDomer constructs have higher avidities than classical antibody-based antivenoms.

3.2. Production of recombinant toxins as antigens for ribosome display selection

Current antivenoms are generated using crude or fractionated venom for immunisation. Recombinant toxin production has the advantage that no venomous snakes are required, tags can be added to facilitate downstream work, and high protein purity can be achieved, allowing the biochemical and biophysical characterisation of the recombinant toxins prior to their use as antigens in ribosome display selection experiments. Ribosome display is an in vitro selection and evolution method ideally suited for generating high-affinity binders from very large synthetic libraries (encoding antibody single-chain variable fragments, nanobodies or engineered proteins) [47,48,49,50,51,52]. While it is desirable to work with purified, well-characterised toxins, a key challenge is to establish expression protocols for the active toxins; in particular, SVMPs contain many disulphide bridges (some SVMPs comprise 40 cysteines) which make them difficult to express and fold correctly. Successful recombinant protein production, however, will remove the batch-to-batch biological variability in venoms and yield reliably pure protein with defined activity.

Only very few production protocols for toxins are available in the literature [53], and our objective is to develop new protocols, optimised for individual toxins, during this project using prokaryotic (E. coli) and eukaryotic (baculovirus insect cell expression) production systems. Moreover, to test the activity of recombinantly produced toxins, specific functional tests are required for each toxin family, confirming that the purified proteins are correctly folded and bioactive. For instance, the N-terminal and C-terminal residues of venom toxins can contribute to ligand binding and biological activity [54,55], and therefore functional tests must be performed to confirm that the addition of tags does not interfere with toxin activity.

In a first approach, all prioritised toxin types (see Section 3.1) will be produced in E. coli and purified following a high-throughput protocol that has been successfully validated for a previous EU project called VENOMICS [56] to purify thousands of toxins including 3FTxs, Kunitz-type toxins and disintegrins [56,57]. The toxins are produced in the periplasm of E. coli using an N-terminal hexahistidine-tagged redox-active DsbC as fusion tag. DsbC fusions increase solubility and contribute to oxidation of the cysteines, allowing efficient formation of correctly folded disulphide bridges [56]. The fusion protein can be cleaved off using TEV protease after affinity purification, restoring the native N-terminus in the toxin. For ribosome display selections, a non-cleavable C-terminal Avi-tag for biotinylation is fused to the toxins [58]. Importantly, folding of the toxins can be further improved by the co-expression of two cysteine chaperones, protein disulphide isomerase (PDI) and sulfhydryl oxidase Erv1p (the CyDisCo system, see [59]).

Larger toxins with enzymatic activity and composed of several domains, such as SVMPs, contain many disulphide bonds required for structural integrity and possibly also cysteines in the active centre for metal coordination. Due to this increased complexity, bacterial expression systems often fail to produce active enzymes. If the toxins cannot be produced in high yields in E. coli or are non-functional, our expression system of choice is MultiBac, a state-of-the-art baculovirus insect cell expression system [60]. The toxins are then expressed with a melittin signal sequence fused to the N-terminus to ensure that they are secreted into the medium [61] facilitating purification. Insect cell expression may lead to post-translational modification (PTM) of toxins, glycosylation and phosphorylation, and the impact of these PTMs on toxin activity and solubility are largely unknown to date.

A further strategy regarding toxin antigen production for selection experiments we will pursue is generation of ‘epitope strings’ comprising an N-terminal, highly produced fusion protein (e.g. maltose-binding protein) and the most conserved sequences from one toxin family fused via glycine-serine rich linker sequences [62] - the idea being that conserved regions are functionally important and binding these regions with ADDomers may interfere with their toxicity/function.

Our objective in ADDovenom by implementing these approaches is two-fold: to provide the toxin components as antigens for our selection studies, and to develop protocols and a knowledge base for how to comprehensively produce toxins of interest in the best suitable expression system.

3.3. Evaluation of the neutralising ability of ADDobodies and ADDomers

Ribosome display selected ADDobodies and corresponding ADDomers capable of binding recombinant toxins will be assessed for their ability to neutralise toxin-function using a panel of serological, phenotypic and functional in vitro assays against crude venom and recombinant toxins. First, serological methods such as end-point ELISA and immunoblots will be used to demonstrate recognition of the target toxin and venom proteins. Typically, an indirect ELISA format is used in which the venom/toxin of interest is coated on an ELISA plate, incubated with the test antibody (i.e., antivenom or other protein therapeutic), and a secondary antibody capable of binding the test antibody/protein used to detect binding. Although ELISAs generally show good correlation with in vivo neutralising ability, for some venoms a poor correlation is observed [63], highlighting the need for rigorous functional in vitro and in vivo testing for each candidate therapeutic beyond serological assays.

ADDobodies/ADDomers directed against Echis SVMP, PLA₂ and disintegrin toxins will be tested using specific enzyme activity assays and phenotypic assays, as indicated in Figure 4. Neutralisation of SVMP activity will be measured using the fluorogenic assay previously described [64]. Briefly, the quenched fluorescent peptide Mca-KPLGL-Dpa-AR-NH₂ (ES010, BioTechne) is cleaved by SVMP between the Gly and Leu residues, releasing the fluorescent Mca-containing fragment from the quenching Dpa-containing fragment (Figure 4a), resulting in increased fluorescence. Other assays of SVMP bioactivities, including enzymatic activity, cytotoxicity assays and endothelial cell tube formation assays are reviewed by Macedo and Fox (2016) [65]. The enzymatic activity (phospholipid cleavage) of Echis PLA₂ toxins will be determined using a fluorescent liposome-based PLA₂ assay (EnzCheck, ThermoFisher) [45]. The assay reagent is prepared by encapsulating a fluorescent dye (BODIPY PC-A2) in a liposome containing dioleoylphosphatidylcholine and dioleoylphosphatidylglycerol (Figure 4b). Liposomes cleaved by PLA₂ release the fluorescent substrate. Lower enzymatic activity is reported for S49-containing snake venom PLA₂s which exert cytotoxic activities in a calcium-independent manner [66] which can be measured in standard cell viability assays [67]. Echis venoms contain both D49 (enzymatically active) and S49 (enzymatically inactive) PLA₂ variants.

The haemostatic disturbances caused by Echis venoms can be measured using a phenotypic in vitro plasma clotting assay [68], in which the time for ‘normal’ plasma to clot is measured spectrophotometrically, and simultaneously disturbances to plasma clotting (either pro-coagulant or anti-coagulant) by venom toxins are detected by the differences in time to form a clot (Figure 4c). Disintegrin activity can be measured using several different methods that examine the effects of the toxins on platelet aggregation. Traditionally an aggregometer is used [65], but this approach requires specialist equipment and is low-throughput. Alternative approaches include a functional ELISA and flow cytometry, both employing antibodies that are selective for activated platelets, thus facilitating differentiation between activated and disintegrin-inhibited platelets (Figure 4d).

ADDobodies/ADDomers targeting the selected mamba toxins will be tested in cell-based assays for antagonism of muscle-type nicotinic acetylcholine receptors (nAChRs) and voltage-gated potassium channels, the targets of α-neurotoxins (a subclass of 3FTx) and Kunitz-type toxins respectively. To measure the effects of mamba long- and short-chain α-neurotoxins on nAChRs, we will use a functional assay that measures the ability of the receptors to activate in response to binding of the agonist acetylcholine (Figure 4e) [69]. This assay uses a cell line natively expressing muscle-type nAChR to measure the effects of α-neurotoxins, and complements an existing assay that measures the effects of neurotoxins on the neuronal α7-nAChR subtype [70], which are selectively targeted by long-chain 3FTx [71]. Traditional assays of the inhibitory activity of Kunitz-type toxins on voltage-gated potassium channels (K_v) use electrophysiological recording of Xenopus oocytes or mammalian cells engineered to express K_v channels [72]. Electrophysiology techniques are considered the gold-standard in ion channel research but require specialist equipment and training, and are low-throughput [73]. In the ADDovenom project, we are developing a higher-throughput (96-well format) assay to measure dendrotoxin-mediated block of K_v1 channels. HEK293 cells stably expressing tetrameric human K_v1.1 and 1.2 are stimulated by potassium sulphate, causing a change in membrane potential measured by a commercial membrane potential dye (FLIPR Membrane Potential Assay Kit, Molecular Devices), and antagonism of K_v channels by Kunitz-type toxins can therefore also be detected (Figure 4f).

Neutralising ADDobodies/ADDomers will subsequently be tested for venom neutralisation in vivo using murine models of envenoming. We will assess ADDobodies/ADDomers for their efficacy against the lethal systemic effects of Echis and Dendroaspis venoms using the WHO-approved ‘effective dose 50’ (ED₅₀) murine model of snakebite envenoming, where venom challenge and treatment are preincubated together ahead of intravenous co-administration [74]. Evidence of efficacy in this ‘best case’ in vivo model of envenoming will provide a compelling justification to apply more challenging animal models that better recapitulate an envenoming scenario, specifically with therapeutic administration occurring after venom challenge [75].

3.4. Scalable bioprocess for ADDomer production

ADDomer nanoparticles are produced in insect cells, a very competitive manufacturing host, that enables expression at high level of exogenous proteins encoded in a baculovirus genome. As eukaryotes, insect cells can process complex secondary structures, are easy to culture and can grow in serum-free media, require less energy than mammalian cells and have low biosafety requirements [76]. The most frequently used insect cells derive from Spodoptera frugiperda and Trichoplusia ni and both offer GMP compatible cell lines [77,78,79]. By testing a variety of culture modes and culture medium supplementation strategies, production yields can be increased. Even though the most used culture mode is the batch system, protein-based nanoparticles and other insect cell derived products can also be produced in fed-batch or perfusion [80,81], both to be tested for ADDomer production (Figure 5). A hybrid perfusion process will also be tested, considering the cost and benefit of using larger media volumes and implementing cell-medium separation devices on the production phase [81]. Use of stainless steel or of single use bioreactors will be considered [82], with future transfer of the final process to a GMP facility in mind, and the latter providing easier technology transfer to GMP. The choice of upstream parameters will be combined with downstream optimisation, ensuring a good compromise between purity and product loss along the process. Once more, ensuring process scalability and GMP compatibility are the main focus. Moving towards a continuous bioprocess has been a major interest as the higher automation will lead to process intensification, reduce steps and shorten the production cycle, thus reducing the production cost [83,84]. Downstream process optimisation encompasses an array of techniques ranging from bench top to industrial scale, however, not all can easily be performed under GMP standards. Centrifugation steps and chromatography with low loading capacity will be avoided (e.g., size exclusion chromatography). The first step will be to optimise the clarification via microfiltration, using for example hollow fiber tangential flow filtration systems. As ADDomer nanoparticles are an intracellular product, this system will be used to separate the cells from the supernatant followed by in situ lysis of cells and recovery of the clarified ADDomer particles (Figure 5). Purification strategies will combine different types of chromatography in bind/elute or flow through modes with anion exchange membranes, these last ones removing DNA, baculovirus and charged host cell proteins. The process will be finalised by ultrafiltration/diafiltration in tangential flow mode and sterile filtration to ensure sterility and stability in the defined storage buffer.

Quality control protocols will be established ensuring reliable efficacy of the antivenom and robust, large-scale ADDomer nanoparticle production. These will cover from the production phase to the final product. During production, virus titre and cell growth rate will be analysed prior to infection. During downstream processing, the intermediate products will be screened by high-performance liquid chromatography (HPLC) for particles quantification and purity analysis. The final product will be screened for the main properties of the ADDomer nanoparticle as an antivenom: size and morphology will be screened by transmission electron microscopy, presence of aggregates will be evaluated by multi-angle dynamic light scattering (MADLS), and HPLC will enable accurate quantification. Impurities will also be controlled by measuring the amount of residual host cell protein, endotoxin levels, host cell DNA concentration and mass spectrometry to identify main contaminants.

For vaccines, cold-chain logistical issues account for up to one third of the final cost, and amounts for up to 50% of yearly dose wastage [85,86]. It is likely that similar numbers apply to antivenoms. The need to reduce costs and wastage has driven several studies to develop thermostable, cold-chain free solutions in a variety of systems [87,88,89]. As mentioned in Section 2, ADDomer-based proteins are inherently very stable (Tm>45 °C) which will translate in a wider range of storage and transport temperatures and consequently reduce the associated cost and reduce reliance on cold-chain logistics. Buffer formulation can play an important role in nanoparticle stability. Even though stability of ADDomer is not a main concern, different buffers (e.g., phosphate-, citrate-based) and stabilisers reported as “generally regarded as safe (GRAS)” will be considered and the cost-benefit of their introduction evaluated. Non-animal origin stabilisers such as amino acids and sugars have been reported to increase protein stability by increasing solubility, preventing protein aggregation, or reducing oxidation [90]. ADDomer antivenom final formulation will be tested under different temperature settings as well as freeze-thaw cycles to evaluate long term stability.