Overview of in vivo and ex vivo endpoints in murine food allergy models: Suitable for evaluation of the sensitizing capacity of novel proteins?

Abstract Significant efforts are necessary to introduce new dietary protein sources to feed a growing world population while maintaining food supply chain sustainability. Such a sustainable protein transition includes the use of highly modified proteins from side streams or the introduction of new protein sources that may lead to increased clinically relevant allergic sensitization. With food allergy being a major health problem of increasing concern, understanding the potential allergenicity of new or modified proteins is crucial to ensure public health protection. The best predictive risk assessment methods currently relied on are in vivo models, making the choice of endpoint parameters a key element in evaluating the sensitizing capacity of novel proteins. Here, we provide a comprehensive overview of the most frequently used in vivo and ex vivo endpoints in murine food allergy models, addressing their strengths and limitations for assessing sensitization risks. For optimal laboratory‐to‐laboratory reproducibility and reliable use of predictive tests for protein risk assessment, it is important that researchers maintain and apply the same relevant parameters and procedures. Thus, there is an urgent need for a consensus on key food allergy parameters to be applied in future food allergy research in synergy between both knowledge institutes and clinicians.


| INTRODUC TI ON
A variety of in vitro and in vivo models have been developed that address the factors and mechanisms involved in the sensitization to food proteins. [1][2][3][4] Currently, approaches are being developed using protein chemistry and in vitro and in silico methods to characterize food proteins and derivatives that arise during product processing and reformulation, which may explain why certain food proteins induce sensitization of the immune system, while others are tolerated. 5,6 However, elucidating the mechanisms underlying allergen sensitization is a complex, multidimensional problem that often requires a wide range of additional in vivo and ex vivo experimentation, 5 as a wide range of molecules, tissues, and cells play a role in the mechanisms underlying food allergen sensitization. 1 For instance, epithelial release of thymic stromal lymphopoietin (TSLP), granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-25, and IL-33 upon local epithelial stress support type 2 helper T (Th2) cell pathology by attracting IL-4 secreting lymphoid cells, basophils, and invariant natural killer T (iNKT) cells. 7 Il-4 promotes surface expression of Th2-costimulatory molecule OX40 ligand on dendritic cells (DCs) 8 and cytokine secretion by Th2 lymphoid cells (ILC2s), which further augments DC activity and suppresses allergen-specific regulatory T (Treg) cells. 9,10 This complexity, as depicted in Figure 1, illustrates the need for experimental food allergy models that integrate such complex cell-tissue communication to assess the sensitization potential of new protein sources. Murine food allergy models, even though they have their limitations, are currently the best predictive models available to evaluate the food-sensitizing capacity of new food proteins before introducing them into the human diet. Although researchers aim to reduce the use of experimental animals to address the 3R principle that guides animal experimentation to replace (alternative model), reduce (minimize number of animals), and refine (minimize animal pain and enhance animal welfare), there is a lack of replacement models such as in silico prediction models, in vitro primary cell assays, or tissue explants assays that are able to characterize and predict the human responses to food proteins. In the past, numerous experimental food allergy models have been developed to assess food allergenicity. However, interlaboratory differences in the models used with respect to sensitization and elicitation route, choice of endpoint parameters a key element in evaluating the sensitizing capacity of novel proteins. Here, we provide a comprehensive overview of the most frequently used in vivo and ex vivo endpoints in murine food allergy models, addressing their strengths and limitations for assessing sensitization risks. For optimal laboratory-to-laboratory reproducibility and reliable use of predictive tests for protein risk assessment, it is important that researchers maintain and apply the same relevant parameters and procedures. Thus, there is an urgent need for a consensus on key food allergy parameters to be applied in future food allergy research in synergy between both knowledge institutes and clinicians.

K E Y W O R D S
animal models, biomarkers, food allergy, prevention F I G U R E 1 Immune mechanisms of food allergy and its associated principal measured endpoints. A, Assessment of allergic symptoms (body temperature) after allergen challenge. B, Evaluation of immunoglobulin (IgE) in serum. C, Phenotyping of T-cell population. D, Cytokine production in response to allergen restimulation (ex vivo assay) adjuvant, clinical signs, genetic background of the animals, housing conditions, and microbiome composition and metabolic activity in the different vivaria often make it difficult to draw generalized conclusions. 5 It is important to note that almost all models (except genetic models) require adjuvants to trigger sensitization. Therefore, the choice of the adjuvants together with the exposure route are crucial points to consider. In addition, there are numerous in vivo, ex vivo, and in vitro parameters evaluated for the assessment of food allergy. Figure 2 illustrates the types of in vivo (inside a living organism) or ex vivo (outside an organism) methodology and endpoints used in experimental murine models of food allergy.
However, there is a need to establish a list of reliable, validated, and effective endpoint parameters to guide researchers working with animal models of food allergy. In this review, we describe a selective list of the most commonly used experimental applied endpoints in food allergies with a focus on milk, egg, and peanut allergens and critically evaluate their applicability for evaluating sensitization potency. Each endpoint was selected and critically described with strengths and limitation based on consortium experience and occurrence in literature.

| ME A SUREMENT OF BODY TEMPER ATURE
In murine-type models of food allergy to milk, eggs, and peanuts, a drop in the core body temperature is often observed after repetitive allergen challenge. This change in body temperature is an indicator of anaphylaxis (Table 1). Temperature is measured before and 30 minutes to 1 hour after allergen challenge, but this parameter can also be monitored over time. 21,22 Animals sensitized to a given food matrix or protein may display a significant reduction in body temperature (0.5-10°C) 3,4 compared with that of naive animals. For an adequate level of sensitivity, 5-16 animals per group should have their temperatures measured using a rectally inserted thermal probe, 29 but it is also possible to measure changes over time for individual animals using an electronic ID transponder implanted subcutaneously. 11,12 To refine, improve, and objectify the currently applied manual monitoring methods, an automatic imaging method has been developed. 14 It involves a noninvasive measurement of the whole-body surface temperature paired with assessment of activity (see also Data S1 about activity/behavior via camera). Anaphylaxis imaging has been used in three in vivo allergy mouse models for (a) milk allergy, (b) egg allergy, and (c) peanut allergy in proof-of-principle experiments and suggests that imaging technology represents a reliable noninvasive method for objective monitoring of small animals during anaphylaxis over time.
This method can be useful for monitoring diseases associated with changes in both body temperature and physical behavior.

| Strengths
• The measurement of core body temperature is a cost-effective, reliable assessment of the allergic reaction.
• Therapeutic or preventative strategies for the reduction of allergic reactions can be easily evaluated.
• Can be used to evaluate the severity of allergic shock and differences between allergens subjected to physical transformations (ie, native versus processed).

| Limitations
• The occurrence of anaphylaxis is dependent on the mouse strain used: Balbc or C3H mice are prone to develop anaphylaxis, whereas C57BL/6 or A/J mice necessitate stringent exposure protocols to achieve sensitization.
• The clinical score may be biased as a consequence of the laboratory environment, stress level, animal strain, and technical experimenter.
• A decrease in temperature is only observed after a food/allergen challenge after a previous sensitization event; this endpoint therefore contains no predictive value for the sensitization potential of a food protein.

| Technical recommendations
• Using a rectal probe, mice or rats must be acclimated to the experimental room at least 1 hour before starting the temperature measurements to obtain stable values.
• The rectal temperature must be evaluated 10 minutes to 1.5 hours after the challenge.
• The animal temperature can be registered over time using a programmable temperature transponder implanted subcutaneously.

| E VALUATION OF IMMUNOG LOBULINS IN S ERUM
While in vivo measurements are essential to assess the elicitation of an allergic response, they do not provide insight into de novo allergen sensitization. Therefore, blood, tissue, or organs must be collected and further analyzed by ex vivo methods. Serum immunoglobulin (Ig) content is the most common parameter measured when evaluating sensitization to food allergens in animal models, followed by fecal IgA (see Data S1), as antibody responses are considered a direct indicator of allergen sensitization together with mast cell and basophil degranulation. IgE is the most common Ig isotype measured when evaluating the allergenicity of food proteins and is regularly quantified in parallel with IgG1 (Table 2)  disease. 46,47 The avidity can be measured by means of simple potassium thiocyanate (KSCN) ELISAs which have shown that no general relationship exists between the level and avidity of specific Igs, 48

| Strengths
• Specific IgE antibody analysis is the most trustworthy measure of sensitization.
• Measures of specific IgE antibodies are often used to evaluate not only sensitization but also the potential severity of the allergic reaction after a second encounter.
• Measurements of antibodies can be performed without the use of advanced equipment such as a cytometer or robotics.

| Limitations
• Assays often need to be developed in-house, restricting the possibilities for comparison between laboratories.
• IgE only accounts for a fraction of all serum antibodies, requiring more advanced ELISAs for analysis of specific IgE.

| Technical recommendations
• Antibody-capture ELISAs should be used for the measurement of specific IgE.
• Other antibody parameters in addition to the amount of total and specific antibodies are relevant and should be measured, such as clonality and avidity.
• Measures of total and specific antibodies should always be expressed as titer values or as concentrations with no upper or lower limit for dilutions.
• Serum depleted of IgG using protein G columns before use in indirect ELISAs needs to be considered.

| Strengths
• Precise mechanistic insights into the cellular response in isolated organs and tissues support the sensitizing potential of food proteins when combined with additional readouts.
• Precise determination of the T-cell profile by using specific markers of the T-cell population.
• Quantitative evaluation of the infiltrating cell population by flow cytometry.

| Limitations
• Analysis of cell populations without the contribution of neighboring cell tissue (loss of microenvironment).
• Isolation of immune cells from tissues relies on enzymatic digestion protocols and may thus alter phenotypical and functional properties of the cells of interest.
• Difficulty with the separation of minor subpopulations.
• Sacrifice of the animal is required for organ and tissue sampling.
• Need for sophisticated equipment such as FACS.
• Type 2 immune response-associated mucus production in tissues makes cell isolation difficult and can create bias in cell phenotyping and frequencies.

| Technical recommendations
• Remove fat and store organs, tissues and cells at 4°C to avoid uncontrolled cell death or degradation of surface markers.
• Perform flow cytometry and culturing the same day as the animal kill.
• Phenotyping of T cells can be achieved by intracellular cytokine/ transcription factor staining using flow cytometry.

| C Y TOKINE PRODUC TI ON IN RE S P ON S E TO ALLERG EN RE S TIMUL ATI ON
The

| Strengths
• Precise assessment of the allergen specificity by restimulating cells with the same allergen used in the animal model.
• Class determination of the T-cell response by evaluation of cytokine production in the supernatant of sorted T cells.
• Higher production of cytokines can be obtained after proliferation and restimulation with the antigen than by direct measurement in serum.

| Limitations
• Restimulation with allergens can activate nonspecific T cells due to certain cross-reactivity.
• Difficult to obtain a level above the sensitivity threshold with cells isolated from naïve mice.
• Some mechanistic endpoints are not equally important in animals and humans.

| Technical recommendations
• For allergen presentation, presorted T cells need to be co-cultured with dendritic cells.
• MHC peptide-tetramers can be used to sort specific T cells and have better assessment of allergen specificity.
• Endotoxin levels within the allergen extract need to be controlled to prevent bias in restimulation responses.
• Ideally, when using gene expression sequencing data, this method should be confirmed with at least one other technology (eg, flow cytometry).
• As cells and mediators associated with immune responses change rapidly, longitudinal assessments of mechanistic endpoints will be more informative than single time point assessments. The timing of the measurements will depend on the research question, for example, sensitization mechanisms vs mechanisms of acute allergic responses following (re)challenge.

| FUTURE ANALYS IS OF FOOD ALLERGY MODEL S
To date, the methods to study intestinal pathophysiology are in vitro culture systems with cell lines or explanted mucosa grown in monolayers, 67,68 intestinal organoid cultures, 69,70 and "gut-on-achip" devices. 71 and metabolic activity of gut microbes can influence all aspects of innate and adaptive immune processes within the mucosa (see also Data S1 for stool consistency as a readout in food allergy assessment). Thus, focusing on the effect of diverse microbiota profiles and specific bacteria on immunological responses upon the introduction of allergenic proteins may lead to novel mechanisms, therapeutic targets, or predictive models. However, intra-and interlaboratory variability in microbiome composition and metabolic activity after birth as a result of the breeding environment is also a major underlying cause for conflicting results between experiments. This variability must be taken into account beforehand in the experimental design of an animal trial. 5 It is also noteworthy to consider the possible development of highly controlled chamber units for food allergy research used in combination with in vivo models to provide a new powerful strategy for studying mechanisms in the intestine.

| Strengths
• The tissue structure, cellular components, and neural system are highly preserved.
• The model provides the possibility to study immediate responses generated after the introduction of different molecules and microbes.

| Limitations
• Only short-term responses can be evaluated due to changes that can occur in the tissue over time.
• Currently, only intestinal segments from 12-to 14-day-old mice have been tested.
• Tissue preparation and assembly require specific skills.

| CON CLUS ION
The recent broadening of our knowledge of food allergy pathogenesis and development of murine food allergy models has enabled us to model the allergic elicitation reaction as well as the preceding sensitization events and observe relevant symptoms with different food proteins (milk, egg, and peanut). The principal endpoint parameters described in this review are critical parameters that should be evaluated in a correct manner so that they may be powerful in the different rodent models.
Characterizing a food allergy model using temperature, level of Igs, phenotyping of the cell infiltrate, and cytokine production gives an overview of the reaction while providing us insight into the degree of sensitizing capacity of the allergen used. Nevertheless, even though the in vivo measurements and the ex vivo experiments provide us with many answers about the immune response and the sensitization phase, we still do not have a complete overview of the immune mechanisms behind each reaction. There is still a strong need to better define the allergic reaction to predict the clinical outcomes of sensitization to novel food proteins. Although the current available models are suitable for studying the pathophysiology of food allergy, they still cannot predict the magnitude of the allergic potential of a particular allergen. Discovering and highlighting the molecules and cells involved in both sensitization and elicitation are necessary to improve risk assessment models and to facilitate the introduction of novel protein sources into our diet with a low risk of allergic sensitization.

ACK N OWLED G M ENT
The authors are all part of the COST Action FA1402 entitled: Improving Allergy Risk Assessment Strategy for New Food Proteins (ImpARAS).