The main compartment of interest for studying innate immune responses in bivalves is the circulating fluid, the haemolymph, and is in direct contact with tissues, since the circulatory system is open. Bivalve haemolymph contains circulating cells, the haemocytes (granulocytes and agranulocytes), whereas its soluble component (serum) carries a large range of humoral defence factors (e.g., lectins, hydrolytic enzymes, antimicrobial peptides). All these components interact in providing a coordinated innate immune response. Mussel haemocytes encompass several subpopulations, representing the progressive maturation stages of a single cell type, in particular R1 (large granular cells), R2 (small semi-granular cells) and R3 (small agranular or hyaline cells), involved in different immune defence mechanisms (García-García et al., 2008). The different cell types can show distinct sensitivity/reactivity to different stimuli. Whatever the exposure conditions, after immune challenge the whole haemolymph can be withdrawn from the posterior adductor muscle sinus via a non-invasive method. Besides haemolymph, immune responses and inflammatory reactions can take place in other tissues depending on the nature of the contaminant or pathogen. The tissues mostly involved in immune reactions are the gills and digestive gland, due to their roles in filtration, feeding, and intracellular digestion.

The immunotoxicity in mussels can be evaluated in the laboratory, either considering the whole animal through in vivo exposure or isolated haemocytes for in vitro testing.

Multiple routes of exposure can be used in in vivo laboratory experiments, the most common being through the water (Figure 1). This mode of exposure resembles the natural conditions and allows for controlling the concentrations of each stimulus (chemicals or micro-organisms) for each individual (Canesi et al., 2014; Capolupo et al., 2021). Exposure via a dietary route integrates complementary concepts, as both contaminants and micro-organisms can be associated with food (i.e., microalgae, particulate matter in suspension), although there is not clear evidence about the exact amount transferred to the animal (Duroudier et al., 2019). Finally, to ensure the direct contact with immune cells, potential chemical or biological immune modulators are directly injected via the sinus present in the posterior adductor muscle (Balbi et al., 2019; Rey-Campos et al., 2019).

For in vitro testing, Mytilus haemocytes represent a suitable tool for immune evaluation (due to their high antibacterial activity, haemocytes can be kept under non strictly sterile conditions). The main limitation of in vitro experiments with mussel haemocytes is represented by the need of using primary cell cultures, since no stable long-term cultures or cell lines have been established so far (Canesi et al., 2017). Therefore, assays are carried out on primary haemocyte cultures maintained for very short periods (up to few hours) in sterile artificial seawater or natural medium (haemolymph soluble fraction or serum) or over 24 h in synthetic basal media adjusted for ionic strength (Katsumiti et al., 2015; Auguste et al., 2018). This is however only a partial limitation, because mussel haemocytes have the characteristic of building rapid functional responses, which can be observed within 1 h from exposure. Assays can be carried out on haemocytes in suspension or in monolayers, taking advantage of the capacity of these cells to easily adhere to virtually any surface (such as plastic or glass). Adhesion capacity is particularly strong for granular haemocytes, which are the majority of circulating cells.

A range of immunological biomarkers has been identified, which provide a quantitative evaluation of the immune responses to chemicals and micro-organisms. These biomarkers can be evaluated at different levels of biological organization (from molecular to organism level).

General haemocyte parameters (cell viability or apoptosis) can be estimated first to get an overview of the effects caused by different contaminants, but alone do not provide details on specific immune-related mechanisms of action. In the circulating fluid, quantitative analysis of the total haemocyte count (THC), and in particular the percentage of the different subpopulations can allow identifying specific activities induced by different stimuli. Flow cytometric techniques, which can evaluate cell size and complexity, are robust and sensitive methods for quantifying these parameters (Renwrantz et al., 2013; Sendra et al., 2020).

The most common targets for immunotoxicity assessment in haemocytes are changes in the lysosomal compartment (such as membrane disruption, change in pH, and changes in number), effects on phagocytosis (index, rate), oxidative burst (release of reactive oxygen species -ROS- or nitric oxide -NO-), and production of lytic enzymes (e.g., lysozyme). Other parameters can be evaluated in the nucleus, such as DNA damage to assess the genotoxicity of some contaminants. A combination of microscopic and spectrophotometric techniques can be used to appraise these parameters (detailed in the section on assays).

Mussel immune cells express a wide range of inducible immune-related genes encoding extracellular recognition and effector proteins, including lectins, peptidoglycan recognition proteins, lipopolysaccharide (LPS) and β1,3-glucan-binding proteins, and antimicrobial peptides. In the absence of functional tests, the transcription of immune-related genes can be assessed by qRT-PCR (Auguste et al., 2020).

Several immune-related markers can be monitored in gills, a complex organ involved in respiration and feeding, and therefore in direct contact with the surrounding medium. In whole gill samples, microscopical techniques can reveal several responses (i.e., enlargement of gill lamellae, increase in infiltrating haemocytes, uptake of foreign objects or pathogens) (Katsumiti et al., 2015; Balbi et al., 2017). Similar to haemocytes, DNA damage or production of NO and ROS by gills can be monitored (Canesi et al., 2014; Balbi et al., 2017). In addition, transcriptomic analysis can show the activation of immune effector immune mechanisms in gills in response to bacterial infection (Saco et al., 2020).

A very important immune compartment is mucosal immunity. In bivalves, the mucus is produced by different epithelia, including the gills, and is involved in particle capture for both feeding and first defence against foreign material. Mucus not only is an excellent physical barrier, but it also contains hydrolytic enzymes, such as lysozyme or proteases, thus representing an efficient defensive system that traps and destroys/neutralizes potential pathogens before they enter the body (reviewed in Allam and Pales Espinosa (2016). Although these studies highlight the diversity of immune effectors based on mucosal proteomics, standardized methods are not yet available to assess mucosal immunity in bivalves.

Other tissues are involved in systemic immune responses, in particular the hepatopancreas or digestive gland. Immune responses in these tissues are generally evaluated by transcriptomics (reviewed in Balbi et al., 2021) rather than using specific functional immune assays.

Studies in bivalve molluscs on secondary immune response after in vivo challenge with a pathogen showed enhanced phagocytic capacity and higher bactericidal activity (Cong et al., 2008; Ciacci et al., 2009a; Zhang et al., 2014). It was only recently that the above-mentioned effects were attributed to innate memory-like processes, thereby fostering newly designed experimental approaches/protocols (Lanz-Mendoza and Contreras-Garduño, 2021). In bivalves, most studies investigated the responses to natural pathogens (Rey-Campos et al., 2019; Lafont et al., 2020), but evidence is also available on the ability to mount memory responses upon exposure to foreign objects (e.g., plastics particles) (Détrée and Gallardo-Escárate, 2018; Auguste et al., 2020).

This being a recent field of investigations, several aspects of the experimental procedure need a more thorough definition. Regarding the routes of exposure, particles are usually spiked directly in water, while pathogens (e.g., viruses or bacteria) are preferentially injected into the animal. Major differences also concern the choice of the time frame between priming and challenge, often depending on the model system (from few days to months), and the protocol for immune stimulation (e.g., same dose and type of agent vs. higher dose or different agent for challenge; priming with molecules that act like a “vaccine” and mimic the presence of a true pathogen). These issues still represent the major hurdles that affect comparison among different studies.

Concerning the techniques used to quantify the memory responses, new methods such as -omics approaches, with transcriptomics being the most common, can offer a large coverage of the putative mechanisms involved in innate immune memory. However, the use of more classical/conventional methods (e.g., haemolymph functional parameters, quantification of selected immune-related genes) has proven useful to unravel the biological processes regulated upon challenge.

Tunicate immune defence relies on blood cells endowed with phagocytic and/or cytotoxic activity and on humoral factors mediating chemotaxis, opsonization and inflammation (Pinto et al., 2003; Franchi and Ballarin, 2014). The sites of immune recognition and response in ascidians are the pharynx, the haemolymph, and the digestive tract. The pharynx is involved in respiration and food collection, and, with its abundance of mature and immature blood cell types, is also considered the main haemopoietic site (Ermak, 1976). The ascidian haemolymph, flowing through a lymph-like circulatory system, encompasses nutrients, haemocytes, and molecules involved in the immune process. Haemocytes, engaged in immune responses (immunocytes) (Cima et al., 2016), include phagocytes and cytotoxic cells, express most of the pattern-recognition receptors and actively transcribe genes required for immune defence (Shida et al., 2003).

For immunotoxicity assessment, animals should be exposed to the agents under study (chemical contaminants, particles, microplastics) in controlled conditions. Since tunicates are filter feeders, the conditions of exposure are crucial for obtaining meaningful results, and should encompass a relevant concentration of the contaminant in the water at any given time and a relevant exposure time, in order to reproduce the conditions occurring in real life (Galloway and Depledge, 2001).

The ascidian digestive tract displays a constitutive level of immune-related transcripts that are upregulated in response to bacterial lipopolysaccahride (LPS) (Parrinello et al., 2016), suggesting the direct involvement of the gut in the recognition and clearance of foreign material. This assumption is confirmed by the presence of transcripts of genes for Toll-like receptors (TLRs), MBLs and MBL-associated serine proteases (MASPs) in both the stomach and the intestine, in addition to haemocytes circulating in the alimentary tract (Sasaki et al., 2009; Skjoedt et al., 2010). Variable region-containing chitin-binding proteins (VCBPs) are also secreted in the gut lumen, hypothetically for stabilising the commensal gut microbial flora (Dishaw et al., 2011).

The approaches to analyse the immune defence activities of animals exposed to various agents in vivo include assessments on their haemocytes and haemolymph, starting from the hematopoietic proliferation index, ex vivo functional assays and transcriptomics (on haemocytes), biochemical and enzymatic tests (on cell-free haemolymph). Studies of innate/inflammatory response induced by LPS have shown an upregulation of cytokine-like genes such as TNFα-like (Parrinello et al., 2008) and IL-17 (Vizzini et al., 2015) in both the pharynx epithelium and haemocytes. Depending on the species, immunocytes can produce several defensive molecules, such as complement components (Franchi and Ballarin, 2014), hydrolytic enzymes (Cima et al., 2016), the precursor form of phenoloxidase (proPO) (Franchi and Ballarin, 2016), histamine and heparin (García-García et al., 2014).

Ascidian phagocytic and cytotoxic cells have been exploited to test in vitro or ex vivo the effect of biotic and abiotic stimuli (Menin et al., 2008; Franchi et al., 2014; Cima and Varello, 2020; Valsesia et al., 2021). While the use of cells under in vitro conditions is a valuable alternative to test potential toxicants, at present no long-term cell culture systems have been established for most marine invertebrates (Rinkevich, 2005; Cai and Zhang, 2014). A first attempt of long-term cell culture was performed with Botryllus schlosseri blood cells, which can be cultured for several months but that lost reactivity over time (Rinkevich and Rabinowitz, 2000). Successful long-term cell culture was obtained from pharynx explants of the solitary ascidian Styela clava, which could maintain viability and proliferative capacity for up to 80 days (Raftos et al., 1990; Sawada et al., 1994), whereas circulating haemocytes remained viable in culture for a shorter period (only 18% viability after 20 days, no proliferation) (Ermak, 1982), although they could still release haemagglutinin (Arizza et al., 1997).

Among tunicates, the ascidian C. robusta is a good model for assessing the establishment of immune memory. A classical in vivo experimental design consists of the sequential exposure of C. robusta adult individuals to a priming stimulus followed by a period of resting to return to basal conditions, and a challenge with a second stimulus in homologous or cross-stimulation (Melillo et al., 2019). Exposure of C. robusta to microbial agents induces a reaction that primes animals for developing a different (expectedly more protective) response to subsequent challenges, showing the effective establishment of an immune memory. The endpoints of immune activities measured are the same as mentioned above, i.e., the frequency of haemocyte subpopulations, the ex vivo functional assays on haemocytes (phagocytosis) and the regulation of transcription of immune-like genes in the pharynx and gut. As different stimuli have been used in homologous and cross-combination, it was possible to identify different mechanisms establishing immune memory and whether they were priming-specific, challenge-specific, or non-specific.

Other assays can be applied for assessing the establishment of immune memory, such as degranulation of granular haemocytes, and extent of C3 activation, which is a measure of inflammatory anaphylatoxin (C3a) and oposonin (C3b) in the haemolymph (Marino et al., 2002; Giacomelli et al., 2012).

Figure 2 summarises the patterns of immune response of C. robusta to external stimuli and toxicants.

In the sea urchin, the immune response is brought about by a heterogeneous population of freely circulating cells within all coelomic spaces, including the water-vascular system, tissues, and organs, of which macrophage-like phagocytes are the most abundant cell type (Smith et al., 2018). The sea urchin circulating cells represent a rich source of information on the physiological state of an individual. Cells feel environmental stimuli through an abundant repertoire of immune functional molecules (identified based on homology at the RNA level), including more than 700 innate pathogen recognition and 400 environmental sensing and response molecules (Goldstone et al., 2006; Hibino et al., 2006). Sea urchin immune cells can be employed as a tool for analysis of toxicity and safety prediction, and can also be used to provide useful data on the pollution status of marine ecosystems. The abundance of each cell type can change depending on health vs. Disease conditions and in response to an insult.

The use of sea urchin immune cells for the risk assessment of long-term exposure to sources of pollution has been extensively developed and validated (reviewed in Pinsino and Matranga, 2015). Links between the sea urchin immune state and environmental factors in the field have been established at first by analysing the phenotypic profiles of sea urchin cellular subpopulations.

In the natural environment, the sea urchin immune system is continuously stimulated by diverse challenges. Thus, wild animals are immunologically more activated than those maintained in laboratory tanks. Housed animals become immunologically quiescent after months of housing (Alijagic et al., 2021a), and can be stimulated again under controlled conditions (Nair et al., 2005). To ensure the representativeness of the results using wild animals (field studies), a significant number of specimens should be examined, whereas this number could be reduced under controlled conditions. In vivo laboratory studies include ingestion and injection as preferential routes of exposure (Falugi et al., 2012; Pinsino et al., 2015).

The heat shock protein 70 (HSC70/HSP70) has been used as biomarker to assess the immunological reactions to environmental threats, including changes in temperature, pH and osmotic pressure, exposure to contaminants, UV-radiation and natural insults (Matranga et al., 2005; Pinsino et al., 2008). However, we have to consider that the HSC70/HSP70 protein level regulation is complex, and the relation between stress and HSC70/HSP70 upregulation is not linear. Thus, the use of HSC70/HSP70 as an index of stress requires a higher degree of validation. Recent developments in immunotoxicity assessment in the sea urchin imply the use of a selected panel of biomarkers encompassing both gene expression and protein levels (e.g., p38 MAPK, ERK, TLR4-like, HSP70, Interleukin-6) or more global approaches using biochemical, cell biological, and -omics technologies (Pinsino et al., 2015; Migliaccio et al., 2019; Alijagic et al., 2020; Alijagic et al., 2021b). In this context, metabolism and antioxidant activity seem to represent a relevant target for immunotoxicity assessment (Migliaccio et al., 2019; Alijagic et al., 2021b).

Apoptosis assays and active caspase detection are not good prognostic tools/biomarkers for the investigation of the sea urchin immunological state in vivo. In general, both the number of dead cells and the level of caspase activity are low, because immune cells are capable of adjusting to environmental changes by activating robust mechanisms of protection, resistance, and immunological plasticity (Matranga et al., 2005; Rast et al., 2006). Sea urchins are known for the lack of neoplastic diseases, and immune cells can most efficiently repair hydrogen peroxide-induced damage and develop a massive antioxidant response upon stress (Reinardy and Bodnar, 2015). Live cell imaging with specific dyes, including Neutral Red (NR) and 3, 3′-dihexyloxacarbocyanine iodide, is a good tool for assessing the physiological state of subcellular organelles (e.g., phagolysosome formation, endoplasmic reticulum, Golgi and vesicle membranes) (Pinsino et al., 2015). The development of optimised culture protocols resulted in a maintaining adherent phagocytes in pure stable cultures for 2 weeks, with >80% viability and full functional capacity (Pinsino and Alijagic, 2019). Based on the high similarity between sea urchin and human immune-related genes (Rast et al., 2006), sea urchin primary phagocyte cultures can be used as a human phagocyte proxy, specifically relative to innate immunity and inflammation, in line with the guidelines on the responsible use of animals in biomedical research (the 3R principle) (Herrmann et al., 2019). Sea urchin immune cells are a large source of functional molecules, which are secreted into the extracellular environment (Alijagic et al., 2019; D’Alessio et al., 2021). Methods for studying intra- and extra-cellular immunological responses encompass most of the approaches used for human primary cell cultures, including imaging, biochemistry, cellular and molecular biology, cytofluorimetry, proteomics, transcriptomics, and metabolomics (Smith et al., 2019; Alijagic et al., 2020; Alijagic et al., 2021a; Alijagic et al., 2021b; D’Alessio et al., 2021), obviously optimized for the characteristics of the system (e.g., high salt concentration).

The sea urchin anti-graft immunity is an example of a secondary immune response, primed by past experience, that increases upon a second experience (Coffaro and Hinegardner, 1997). However, how and if sea urchin immune cells possess innate immune memory is an open issue that needs attention. Therefore, no methods for assessing immune memory have been proposed yet in the sea urchin.

Figure 3 depicts the features of the sea urchin immune response to external stimuli and toxicants.

At variance with invertebrates, human beings possess two parallel systems of immunity, the evolutionarily conserved innate immunity and the highly sophisticated and specific adaptive immunity. Since innate immunity is the first line of defence of the organism, and the innate cells are the first that could raise an inflammatory reaction upon interacting with harmful stimuli, the majority of immunotoxicological assays are based on assessment on innate immunity/inflammation. Innate immune cells include different subsets of granulocytes (neutrophils, eosinophils, and basophils), dendritic cells, monocytes, macrophages, mast cells and innate lymphoid cells. Soluble innate immune effectors include the complement system, innate cytokines and chemokines, toxic ROS and NOS, defensins and antimicrobial peptides, among others (Kenneth and Casey, 2017). All of them participate to the development of a defensive inflammatory reaction, with monocytes and macrophages playing a major role. Recently, it has also been described that the innate immune system is able to develop an immunological memory, i.e., the capacity to mount a potentiated (trained immunity) or decreased (tolerance) inflammatory reaction upon secondary challenges based on previous stimulations (Ifrim et al., 2014).

Since the in vivo immunotoxicological methods are not easy to perform in a controlled fashion, analysis of innate immune system activation in humans is mainly based on in vitro tests. In vitro models have been predominantly based on transformed cell lines or primary blood leukocytes, whose reactivity to toxicants (besides survival) is mainly assessed in terms of production of inflammatory and anti-inflammatory factors, cytokines and chemokines. It is however important that the in vitro models can realistically reproduce the specific immunological processes that occur in vivo (such as innate/inflammatory reaction and the innate immune memory re-programming process), in order to obtain predictive results.

The most reliable in vitro models are based on human primary monocytes or monocyte-derived macrophages (Li et al., 2016). Monocytes are key inflammatory cells that come from blood into tissues during an inflammatory reaction. The use of primary human cells allows us to study the mechanisms of innate/inflammatory or memory reactions in conditions that readily translate to human responsiveness in vivo, as opposed to mouse models or in vitro models based on transformed cell lines (such as THP-1 or U937, i.e., tumour cells that share some characteristics with primary monocytes; Gennari et al., 2004; Gennari et al., 2005). The use of immortalized cell lines offers several advantages: they are cost-effective, easy to use, provide a pure population of cells with unlimited proliferation, bypass ethical concerns associated with the use of human tissues, and finally provide reproducible data. However, they are scarcely relevant from a biological/functional point of view, since they may have lost many characteristics of the original cells (e.g., the capacity to produce some inflammatory factors in response to challenges) and developed other activities that are absent in primary cells (e.g., uncontrolled proliferation). Primary immune cells may better reproduce the behaviour of normal human cells in vivo, and also provide the opportunity to assess donors’ variability in immune reactivity, which is the basis for developing effective approaches of personalized medicine (Brodin and Davis, 2017).

Human innate/inflammatory reactivity can be measured with two well-established in vitro assays: the Whole Blood Assay (WBA) and Monocyte Activation Test (MAT) (Hartung and Wendel, 1995; Langezaal et al., 2002; Hoffmann et al., 2005; Schindler et al., 2009; European Pharmacopoeia, 2019; Hartung, 2021). The MAT in particular is a test adopted for the detection of pyrogens in pharmaceutical products and is, therefore, suitable for detecting immunotoxicity in terms of induction of inflammation. These are the simplest in vitro models used for assessing the direct capacity of different stimuli to activate innate responses, using as endpoint the production of the inflammatory factors (in general inflammatory cytokines) upon acute exposure (e.g., 4 and 24 h). The WBA is based on the use of anticoagulated whole blood and allows us to study the global reactivity of blood cells and to assess the production of a number of different mediators, pertaining to monocytes, granulocytes, lymphocytes, platelets, etc. The MAT is based on the production of inflammatory cytokines (IL-6, TNFα, IL-1β) by circulating monocytes or mononuclear cells (PBMC) isolated from blood. Both assays are simple but allow for a reliable evaluation of human cell innate/inflammatory activation as well as direct toxic effects by different compounds. A further simplification can be made in the MAT test, with the use of monocyte-like cell lines instead of primary blood cells. This has the advantage of reducing variability, which may be a problem when using cells coming from different donors. However, since cell lines are not normal cells, their reactivity to stimuli/toxicants can be significantly different from that of primary cells, thereby hampering the value of the test (Okemoto et al., 2006; Lankveld et al., 2010; Andreu et al., 2017; Studholme et al., 2019). Other assays have been described for assessing changes in innate cell reactivity, based on NK cells (testing their cytotoxic capacity) and dendritic cells (DC; testing their capacity to mature) (Hymery et al., 2006; Li, 2018).

While these can be useful tools in human immunotoxicology, it should be underlined that an inflammatory reaction is an expected defensive reaction to any challenge, and it only suggests that the innate immune system is efficient. Conversely, an immunotoxic reaction can be only identified by examining other parameters, such as the rapidity of response initiation, the length of its duration, the extent of its magnitude, the balance between inflammatory and anti-inflammatory factors. Thus, the simple WBA and MAT tests are good for identifying the activation of an innate response but they fall short of predicting immunotoxic effects.

In this perspective, a kinetic evaluation of the regulation of monocyte activation and macrophage functional differentiation during an inflammatory reaction is required. In vitro models have been developed, using human primary cells kept in culture in conditions that reproduce the kinetic changes in the tissue microenvironmental conditions experienced by monocytes participating to inflammation (Italiani et al., 2014; Italiani et al., 2020). These models describe in a simplified but reliable fashion the kinetic development of human monocytes reactivity and its modulation during the entire course of resolving or chronic inflammation, providing the basis for understanding whether changes in their transcriptomic and proteomic profiles upon exposure to toxicants may be predictive of health hazard (e.g., a shift from physiological to pathological profile) (Li et al., 2016).

Even though this approach is more accurate in reproducing the course of a human in vivo inflammatory reaction, thereby allowing for the predictive identification of immunotoxicity effects, it presents several limitations that make its use hardly feasible. Besides the inherent variability due to the use of primary cells from different donors (an issue common to all in vitro assays based on primary blood cells), the assay needs time (several days) and specialized skills. Another limitation is conceptual, since the model exclusively reproduces the reactivity of the first monocyte wave entering an inflamed tissue, while it is known that monocytes enter an inflammatory site continuously during the reaction. Also, the lack of other cells, including resident tissue macrophages (and their mechanical and biological interaction), extracellular matrix components and 3D architecture may influence the monocyte reactivity in unforeseeable ways (Jain and Vogel, 2018; Jain et al., 2019).

As a step forward in the validity of in vitro immunotoxicological assays, several co-culture assays are being developed and used (Leonard et al., 2010; Noonan et al., 2019; Drasler et al., 2020; Karri et al., 2021). For example, an in vitro model of the inflamed intestinal mucosa, encompassing a three-dimensional co-culture of human intestinal epithelial cells (the cell line Caco-2) with human primary monocytes and monocyte-derived dendritic cells, can be used for assessing the effect of drugs and toxicants on gut inflammation (Leonard et al., 2010). While many of these models aim at a more thorough evaluation of immune reactions within the context of a tissue, still they cannot reproduce the complexity of cell-cell and cell-stroma chemical and mechanical interactions that take place in vivo in healthy or disease conditions. An advancement in this direction is the development of “organs-on chip”, i.e., 3D tissue cultures set in microfluidic systems (Ulleras et al., 2005). Organs-on-chip use microfabrication technologies to replicate the tissue microenvironment in its 3D microarchitecture (i.e., the spatial distribution of different cell types), its biochemical microenvironment (including chemokine, growth factor and nutrient gradients) and its mechanical microenvironment (e.g., mechanical compression, cyclic strain and shear stress) (Bhatia and Ingber, 2014). Organs-on-chip can reproduce the microenvironment of different organs (e.g., gut-on-chip; Irimia and Wang, 2018) and in different conditions (e.g., inflammation-on chip; Kim et al., 2016), thereby allowing for realistically monitoring human immune cell reactivities (Herrmann et al., 2019). In general, however, when building co-culture models that encompass immune cells, caution should be taken as such cells may react against the other cells in the co-culture (in particular if they are transformed or tumour cells), thereby creating a background sactivation state that would affect the immunotoxicological assessment.

A summary of the immunotoxicity methods that can be adopted for assessing effects on human cells is reported in Figure 4.

The mechanisms underlying the innate immune memory observed in vivo (tolerance or potentiation) can be studied in vitro using cell-based models based on human peripheral monocytes, primed in culture with a given stimulus (e.g., microbial products) for a short period of time (usually 24 h). Subsequently, cells are incubated for 7 days in culture medium without any stimulation, a resting period in which the functional programme of the cells returns to their steady-state basal level. The return to steady-state functionally quiescent state is particularly relevant when examining innate memory, to prevent mistakenly evaluating adaptation or differentiation mechanisms (Divangahi et al., 2021). If the primary stimulus induces transcriptional and epigenetic changes in cells, after the resting period and the extinction of the primary response they will show a different reaction to homologous or heterologous secondary stimuli (Ifrim et al., 2014; Madej et al., 2017; Swartzwelter et al., 2020), which can be assessed as an increased or decreased production of inflammatory and anti-inflammatory factors.

The tolerance and potentiation memory responses assessed in vitro, however, cannot be taken as predictive of increased or decreased resistance to the infections or diseases in vivo, as they only reflect a decrease or increase in the reactivity of monocytes to stimuli, i.e., a re-programming of their responsiveness (Italiani and Boraschi, 2017). Since the functional outcome of innate reactivities is a balance between immunostimulatory/inflammatory responses and feedback mechanisms that control/down-regulate inflammation, a more predictive assessment of effects on innate memory implies examining the simultaneous production of inflammatory and anti-inflammatory factors, which would allow for predicting whether memory will induce a more inflammatory or less inflammatory response profile relative to the primary exposure to stimuli/toxicants (Della Camera et al., 2021).

Notably, with this approach it is possible to define a personalised memory profiling, which can allow defining the individual effects of toxicants/contaminants on innate memory and predict the adequacy of protective responses to future challenges (e.g., infections). During the development of a new immunization strategy, ex vivo stimulation assays could be used for studying innate immune priming and help assessing the effectiveness of a vaccine from the point of view of capacity for non-specific innate/inflammatory response amplification.

The assays described above are mainly based on monocytes, but they are adaptable to other types of innate cells such as macrophages, mast cells and Innate Lymphoid Cells (ILCs), although no specific immunotoxicity methods have been proposed yet to assess innate memory with these cells.