Nanoparticle-induced immune response: Health risk versus treatment opportunity?

.

Exposure to NPs occurs through inhalation, ingestion, skin, or direct administration to the blood; which is followed by interactions with biological systems, tissues, and cells.In particular, interactions with the immune system are of critical importance.They will appear as foreign bodies to immune cells in bodily fluids and tissues (monocytes, phagocytes, platelets, leukocytes and dendritic cells), which will, in turn, engulf and eliminate them (Ilinskaya & Dobrovolskaia, 2016;Zolnik et al., 2010).Subsequently, the immune system may respond and lead to adverse effects, as hypersensitivity reactions and inflammation, at the tissue or body level.The majority of the immune responses towards NPs are undesired, and therefore, considerable effort has been made towards hiding NPs from the immune system.However, by clever design of the NPs, it is possible to channelise the immune response to our benefit, a strategy that can be exploited for developing vaccines and cancer immunotherapy.
The goal of this review is to discuss the different immune responses that are evoked by NPs and how these can be, on one hand, a threat to human health, but on the other hand, a chance to cure diseases.
We will first outline the different immune components that are known to be involved in interactions with NPs.Secondly, we will describe the methods used to study the immune responses and possible challenges associated with these assays to NPs.Subsequently, we will focus on frequently encountered NPs either engineered for medical applications or found in an hazardous manner in our environment, and elaborate on the specific immune responses towards particles with particular characteristics.Finally, several strategies will be described in which the immune response towards NPs can be exploited to benefit specific application such as vaccination and cancer immunotherapy.The aim of this review is not to discuss NP toxicity in detail; readers interested in this topic can consult other recent review articles (Ajdary et al., 2018;Boraschi et al., 2017;Chetyrkina et al., 2022;Yang et al., 2021).Similarly for reviews on nanomedicines or nanovaccines, we would like to refer to other recent articles (Abdel-Mageed et al., 2021;Das & Ali, 2021;Irvine & Dane, 2020;Mitchell et al., 2021;Shi et al., 2017).

Interactions between the immune system and nanoparticles
The immune system is responsible for protecting the human body from pathogens (e.g., viruses and bacteria), recognising and clearing of damaged or altered self (apoptotic and necrotic cells and protein aggregates) and non-self (e.g., NPs).Although engineered and often of nonnatural origin, NPs interacting with the immune system trigger the same pathways as natural pathogens (Boraschi et al., 2017;Fadeel, 2019).These pathways are initiated by proteins recognizing specific molecules or patterns on the surface of the external bodies.These recognized elements can be either part of the NPs themselves, or proteins adsorbed onto their surface after they have been introduced into the blood or tissue and during their journey in the human body.This protein-based "coating" is called the protein corona.Because of its critical role in the further interactions of the NPs with the immune system and their fate, it is presented in the first section in more detail (Capjak et al., 2017).
In the following sections we introduce the two pillars through which the immune system operates to ensure both short-and long-term efficiency: on the one hand, the innate immune system that is rapid and broadly specific and, on the other hand, the adaptive immune system, which is slow and highly specific.The combined activation of these two routes can lead to inflammation, hypersensitivity, or suppression of immune functions (immune-toxicity) (Ilinskaya & Dobrovolskaia, 2016).The different characteristics of NPs which influence their interactions with both pillars of the immune system are described followed by a section on methods to evade the immune system.

Protein corona
Upon their introduction in the body, NPs interact with proteins that adsorb on their surface, eventually forming a protein corona covering the entire particle.The formation of the protein corona is a dynamic process, driven by non-covalent forces such as electrostatic and hydrophobic interactions, affinity of the proteins towards the NP, other proteins already attached to the surface, and the protein structure thermodynamics, which indicates how easily the protein changes shape to interact with the NP (Capjak et al., 2017).Initially, proteins that are abundantly present in the body bind to the NP, this protein coating is partly replaced by those proteins that bind with stronger interactions and higher affinity towards the specific NPs, an effect known as the "Vroman effect" (Vroman et al., 1980).The protein corona can be divided into two parts: the hard corona consisting of proteins presenting high affinity for the NP surface that directly binding to it, and the soft corona, which is more dynamic, comprising proteins interacting via weaker interactions with the proteins in the hard corona (Fig. 1A) (Capjak et al., 2017).
The resulting protein corona determines largely the fate of the NPs; it affects their size, agglomeration, accumulation in specific tissues, distribution, circulation time, kinetics and possible toxicity (Lee et al., 2015).Receptors in the cell membranes can interact with specific proteins in the corona, and these receptor-protein complexes can impact the cell response, e.g., to engulf NPs, activate the apoptosis pathway, or send signals to other cells to come into action (Fig. 1B).
Three different scenarios can be identified for proteins interacting with NPs (Fig. 1C).First, a class of proteins known as opsonins directly interact with receptors on phagocytes, to facilitate phagocytosis.Secondly, dysopsonins, which in contrast to opsonins, have little affinity with receptors on the cells and the cell surface in general, effectively "hiding" the NPs from the phagocytes.Third, NP-protein interactions can cause alterations in the protein, by, e.g., changing its folding and shape, which is a strong marker for altered self, and thus elimination of the protein as well as the NP by the immune system (Capjak et al., 2017).
The most prominent opsonins are immunoglobulins (IgG, IgM) and complement proteins (C1q, C3b, C4b), which mark NPs for phagocytic uptake by the RES (reticuloendothelial system), and ultimately their clearance in liver, spleen or lungs.Tissue-specific opsonins can also determine the distribution pattern of NPs towards different RES organs; for example, fibronectin, which is abundant in the spleen, has specific affinity for positively charged NPs, which explains the high uptake of positively charged NPs by splenic macrophages (Rahman et al., 2013).
The dysopsonin clusterin (otherwise known as Apo J), was detected on metallic (Aoyama et al., 2016) and polystyrene NPs with various functional groups (-NH 2 , -SO 3 , -COOH) (Ritz et al., 2015).This protein has been identified to play a role in avoiding clearance by macrophages, though this effect is debated (Ritz et al., 2015).The same is applies to albumin, the most abundant serum protein, which is generally regarded as a dysopsonin; however, conformational changes in albumin after binding to a NP surface, potentially induced by the curvature or hydrophobicity of the NP, can lead to the exposure of a cryptic epitope within the albumin structure which is able to interact with scavenger receptors.This receptor interaction leads to rapid uptake of these albumin coated NPs (González-García et al., 2022;Mortimer et al., 2014).Therefore, only knowing the proteins in the corona without knowing the actual interactions with receptors is insufficient to know the effect these proteins have on the interactions between the NPs and the immune system.

Innate immune response towards NPs
The innate immune system includes the physical epithelial barriers, phagocytic cells (monocyte/macrophages, dendritic cells (DCs) and polymorphonuclear leukocytes), basophils, mast cells, eosinophils, natural killer (NK) cells as well as circulating plasma proteins, which are known as complement and which are always present in bodily fluids including blood, lymph, and tissues.The response of the innate immune system, which is quick and non-specific to any threat, is initiated by binding of opsonins, leading to phagocytosis and ultimately clearance.
The innate immune response towards NPs is initiated by binding of pattern recognition proteins (PRPs) and pattern recognition receptors (PRRs), which recognize molecular patterns commonly found on pathogens (pathogen-associated molecular patterns, PAMPs) or molecules released from damaged or dying cells (damage-associated molecular patterns, DAMPs).Common targets are molecular motives such as charge clusters, neutral sugars, vicinal hydroxyl groups or acetyl groups, which are present on NPs, before and after they have been covered with a protein corona (Li & Wu, 2021;Zolnik et al., 2010).To include all targets of PRRs, this PAMP-DAMP model has recently been expanded with HAMPs (homeostasis-altering molecular processes) (Liston & Masters, 2017) and NAMPs (nanoparticle-associated molecular patterns) (Fadeel, 2012).Most PRPs work through multiple low-affinity bindingor multivalent interactions -as single bonds are not strong enough to maintain the interactions, which explains the multimeric structure of many PRPs.Only once the PRR is strongly interacting with its target through multiple bonds, the immune cascade is activated (Li & Wu, 2021;Nayak et al., 2012).
The main innate immune response is governed by the complement system, which consists of over 40 soluble and cell surface proteins, which are normally present in an inactive form.These proteins are converted into an active state through three activation pathways, known as the classical pathway; the lectin pathway; and the alternative pathway, to opsonize targets (Fig. 2).Ultimately, complement activation induces phagocytosis by macrophages, which in turn release cytokines and chemokines such as Interleukin-1 (IL-1), IL-6 and tumour necrosis factor alpha (TNF-α), to recruit neutrophils and monocytes and induce an inflammatory response.
The PRP of the classical pathway is C1q, a charged PRP with 18 homologous polypeptide chains, ending in a globular head domain (C Fig. 1.A) The protein corona can be divided into two parts: the hard corona consisting of proteins presenting high affinity for the NP surface that directly binding to it, and the soft corona, which is more dynamic, comprising proteins interacting via weaker interactions with the proteins in the hard corona.B) The proteins in the corona affect the interaction with cell bound receptors.Depicted are the main protein-receptor interactions involved in immune responses towards NPs.Often specific individual proteins interact with a receptor, as in the case of albumin, fibrinogen, complement components such as C1q and C3, but other receptors recognize specific molecular patterns, for example pathogen-associated molecular patterns (PAMPs).C) Opsonins in the protein corona (eg.C1q and C3b) flag NPs for phagocytosis, dysopsonins lead to avoidance of phagocytosis, and therefore, increased circulation time of the NPs.Other proteins can lead to aggregation, which in turn results in accumulation in specific organs like the liver (Images created using Biorender).
terminal), which binds to IgGs, charge clusters or hydrophobic patches on targets (Kishore & Reid, 1999).After binding of C1q, proteases C1r and C1s are activated, which in turn, activate C4 and C2 forming C4b2a, or C3 convertase, which cleaves C3 to form C3b. C3b is a potent surface bound opsonin, which binds to the NPs and is gradually broken down to iC3b by Factor I (see under alternative pathway).While C3b interacts with C3b receptors on red blood cells (through CR1/CR35), phagocytic cells have receptors for iC3b (CR3 and CR4); as a consequence, NPs are transported by red blood cells and passed on to phagocytic cells to be further removed (Pondman et al., 2017b).The classical pathway continues by cleaving C5 into soluble C5a and C5b, which forms a complex with C6, C7, C8 and C9 (C5-9) called the membrane attack complex (MAC).The MAC forms a pore in the lipid bilayer of cells, causing lysis of the cell (Carroll & Sim, 2011;Moghimi & Simberg, 2017).
Mannose-binding lectin (MBL) is the PRP in the lectin pathway; it specifically binds to vicinal diols on sugars (mannose, fucose, glucosamine) or ficolins (Vang Petersen, 2001).For MBL to cause complement activation, multimers must form on the NP surface, these can only be formed if the spacing between targets is "just right" or fitting with MBL in order to trigger the cascade (Read et al., 2022).The multimers of MBL form a complex with MASP-1 and/or MASP-2 (Mannose-binding lectin-Associated Serine Protease), which acts similarly to C1r and C1s, catalysing the formation of C3 convertase, which next initiates the same cascade as in the classical pathway.
The alternative pathway does not include a specific PRP, but involves the continuous slow hydrolysis of C3 into C3(H 2 O) in solution.This process alters the shape of the protein allowing the formation of a complex with factor B, which in turn, allows factor D to cleave the bound Fig. 2. The three pathways of the complement system activated by NPs: classical pathway, lectin pathway and alternative pathway.Each pathway has a different recognition strategy.The classical and lectin pathways start with binding of their respective recognition proteins C1q and MBL onto the NPs, which act as opsonins marking the particle for phagocytosis, but also cleave C2 and C4 allowing the formation of C2a4b, the C3 convertase, leading to the proteolytic cleavage of C3 into C3a and C3b.The latter will form a complex with C2a4b to form C5 convertase.The alternative pathway is initiated by spontaneous hydrolysis of C3, resulting in binding of C3b onto the NPs.The formation of C3b allows Factor B to bind and Factor D to cleave Factor B into Ba and Bb.C3bBb is inherently unstable, until Properdin, the only known up-regulator of the alternative pathway, binds and a second C3b completes the newly formed C5 convertase.The pathway is negatively regulated (indicated by dashed arrows) by Factor H, which can form a complex with C3b, this complex is cleaved by Factor I, resulting in iC3b which is unable to form a convertase.Following the formation of both C5 convertases, all three pathways converge, the cleavage of C5 will allow binding of C6, C7, C8, C9 to C5b, and thereby, resulting in the formation of the Membrane Attack Complex (MAC).In the complement cascades C3a and C5a will be generated which are anaphylatoxins leading to extensive inflammatory responses and CARPA (Figure created using Biorender).
factor B into Ba and Bb.Bb remains bound to the target while Ba is released into the solution.Thus, the target NP is coated with C3bBb, which is stabilized by properdin (factor P) into C3bBbP.This latter complex acts as an enzyme able to generate more C3b to bind.To balance this cascade of ever-increasing opsonisation, factor H acts as a down regulator of the amplification loop (Wang et al., 2016).After binding of factor H, factor I is able to cleave C3b into iC3B, which is unable to form C3bBb (Kouser et al., 2013).
In the complement cascade, the potent anaphylatoxins, C3a and C5a, are formed; these proteins elicit specific physiological responses such as chemoattraction (to, e.g., attract phagocytes) and enhancement of the vascular permeability.These anaphylatoxins have a causal role in Complement-activation-related pseudo allergy (CARPA), which affects cardio-pulmonary functions, and in severe cases, leads to anaphylactic cardiogenic shock and death (Zhang et al., 2018).The effects of longterm or chronic complement activation are largely unknown (La-Beck et al., 2021), but there are strong indications of tumour-promoting roles of the complement system, especially for anaphylatoxin C5a (Markiewski et al., 2008).
Even without complement involvement, the cells of the innate immune system play an active role in recognising and eliminating NPs through their PRRs of which each cell may express as many as 50 distinct types (Silva et al., 2017).Dendritic cells (DCs) and neutrophils identify PAMPs, with receptors that are located either on their cell membrane, e. g., Toll-like receptors (TLRs) and C-type lectin receptors, or in their cytosol, e. g., Nucleotide-binding and oligomerization domain (NOD) -like receptors (NLR) and retinoic acid-inducible gene-I (RIG-I)-like receptors (RLR).Signalling PRRs (TLRs and NLRs) regulate inflammation and apoptotic response by activation of Nuclear Factor kappa-lightchain-enhancer of activated B cells (NF-kβ), mitogen-activated protein (MAP) kinase and interferon-regulatory factors, which leads to secretion of pro-inflammatory cytokines and chemokines.Endocytic PRRs can promote the interaction with pathogens and activate the engulfment and destruction of pathogens by phagocytes (Boraschi et al., 2017;Li & Wu, 2021).Although neutrophils are the most abundantly present immune cells in the blood, involvement of neutrophils in the immune response towards NPs is less studied compared to macrophage interactions, nevertheless these first responders during inflammation are involved in NP clearance (Fromen et al., 2017;Jones et al., 2013).Additionally, neutrophils can capture NPs by releasing neutrophil extracellular traps (NETs), a network of chromatin fibres with antimicrobial proteins (myeloperoxidase and neutrophil elastase) (Muñoz et al., 2016).Little information is available about NETosis in the context of NPs exposure, and high doses of NP seem to be required to induce such a response, which subsequently leads to NP agglomeration and potentially blood vessel occlusion and inflammation (Bilyy et al., 2018).
After their internalisation, NPs can undergo several processing events in phagocytes.If internalisation occurs in phagosomes or endosomes, the latter can fuse with lysosomes, which aim to neutralize the NPs through enzymatic digestion or by lowering the pH (Zaki & Tirelli, 2010).Cell-autonomous antimicrobial defence mechanisms, as autophagy can be detrimental in case of biopersistent NPs and lead to vesicle accumulation, enhancing cell death through mitochondrial dysregulation (Stern et al., 2012).However, if the cell is capable of compartmentalizing NPs into autophagosomes, isolating them from further interactions, toxic effects can be reduced, which is therefore beneficial for the cell as a stress coping mechanism (Stern et al., 2012).
By disrupting the cell membrane, reactive oxygen species (ROS) generation and ATP-receptor interactions, various NPs can activate NLRP3 inflammasomes.The activation of these intracellular multiprotein complexes, present in macrophages and neutrophils, leads to caspase-1 activation and the secretion of pro-inflammatory cytokine IL-1β (Baron et al., 2015).
The lungs are equipped with their own innate immune system, with surfactant proteins A and D (SP-A and SP-D), and alveolar macrophages as the main phagocytic cells.Both SP-A and SP-D interact with PAMP or DAMP materials in the lung as well as NPs, favouring their agglutination and phagocytosis, while modulating the inflammation (Kishore et al., 2006).Of note, the role of SP-A and SP-D as opsonic or dysopsonic agonists is under discussion; for example, SP-A was shown to increase the uptake of magnetite NPs and unmodified anionic PS NPs, while inhibiting the uptake of amino-terminated PS NPs (McKenzie et al., 2015;Ruge et al., 2011).A similar effect was found for SP-D, which favoured the uptake of carboxymethyl cellulose (CMC) -CNTs, but inhibited that of oxidised-CNTs (Pondman et al., 2017a).Similar to asbestos, NPs can cause long-term oxidative stress and pro-inflammatory effects, which most dominantly occur in already challenged lungs of COPD patients (Terzano et al., 2010).Research has to confirm whether these effects also lead to the diseases that are linked to asbestos, known as mesothelioma, pneumoconiosis, and silicosis (Shvedova et al., 2014).

Adaptive immune system response towards NPs
Involvement of the adaptive -or acquired-immune system following exposure to NPs requires the formation of antibodies/immunoglobulins against the NPs or parts thereof, a process which is either thymusdependent (TD) or thymus-independent (TI).
In the TD pathway, DCs form the cellular link between the innate and adaptive immune system.Immature DCs interact with (iC3b coated) NPs though their PRR (TLRs) and internalise them, which in turn, induces DC maturation towards a potent antigen presenting cell (APC).The activated APCs migrate to lymph nodes to present antigenic peptides in the context of major histocompatibility complex (MHC) I and MHC II molecules to T cells, initiating the adaptive immune cascade.Activated Tcells can interact with the antigens presented on the DCs with their T-cell receptors (TCRs).Subsequent interaction between T-helper cells and Bcells results in B-cell proliferation and differentiation towards plasma cells, which then start producing antibodies with high-affinity towards the targeted antigen.The TI mechanism does not involve T-cells; there, B-cell activation is triggered by direct recognition of elements in the antigen by B-cell receptors (BCRs), which results in the production of IgM (den Haan et al., 2014).
Fullerene-specific antibodies (IgG) were detected after immunization of mice with a C 60 fullerene conjugated to thyroglobulin.Although the thyroglobulin protein carrier was critical for the production of the antibodies, the latter were specific towards the core NP carbon structure, as was proven by the cross reactions with C 70 fullerenes (Chen et al., 1998) and Single Walled Carbon Nanotubes (SWNTs) (Erlanger et al., 2001).The identification of a human gold-NP specific antibody similarly suggests that antibody generation against other solid particles is conceivable (Watanabe et al., 2008).An explanation for this antibody production was provided by Zolnik et al (2010), who suggested that NPs behave as haptens (Fig. 3).BCRs interact with structural elements on the NPs, which initiates their uptake by the B-cells.Subsequently, the protein carrier is processed inside the B-cells resulting in the presentation of peptides to T-helper cells and in the production of NP specific antibodies (Zolnik et al., 2010).

Key factors influencing NP interaction with the immune system
Important characteristics of NPs influencing their protein corona and, hence, their interaction with both arms of the immune system are their size, shape, deformability, agglomeration (reversible) or aggregation (irreversible) state, and their surface properties, such as charge and hydrophobicity/hydrophilicity.In biodegradable particles or particles that can cross into different tissues, dynamic interactions can be expected depending on the state and location of the particle at a specific moment (Capjak et al., 2017;Huang et al., 2013;Papini et al., 2020).
The NP size and associated curvature are critical in binding of molecules and their uptake mechanism.In general, larger NPs adsorb larger proteins such as immunoglobulins and complement components.For example, opsonin C3b occupies an area of approximately 40 nm 2 , for which very small NPs do not offer sufficient space for binding (Pedersen et al., 2010).Also uptake mechanisms change with the size of NPs, small NPs (up to 200 nm) are internalised via clathrin-or caveolar-mediated endocytosis, while larger NPs more commonly taken up through phagocytosis, which is more influenced by the adsorption of opsonins (Oyewumi et al., 2010).Other studies have indicated that virus-like particles (20-200 nm) exhibit more B-cell interactions and are eliminated faster by the liver (association with liver sinusoidal endothelial cells), while larger NPs (>200 nm) are likely to remain in the interstitial space and be phagocytosed by DCs and macrophages (Markiewski et al., 2008).The NP geometry also impacts the cellular uptake efficacy, rodshaped NPs being internalized most readily, followed by spheres and cylinders, while cubes are not easily internalized (Gratton et al., 2008).
Hydrophobic particles are more likely to interact with host proteins and the cell membrane, while NPs exhibiting both hydrophilic and hydrophobic elements can act as surfactants, damaging cell membranes (Bosi et al., 2004).Surfaces with repetitive epitopes, as is often the case of polymeric or coated NPs, may trigger complement activation as the presence of a molecular pattern triggers binding of PRPs.Similarly, charge patterns can act as targets for PRPs, which are therefore more likely to bind both positively and negatively charged NPs compared to neutral particles.Cationic surfactants on both lipid and polymeric NPs have been shown to induce cell death in neutrophils, through lactate dehydrogenase (LDH), ROS and elastase release, suggesting oxidative stress as a key mechanism of cellular injury caused by these particles (Fang et al., 2015;Zolnik et al., 2010).
Looking further into surface chemistry, NPs (Si) containing either hydrocarbon chains or acid groups on their surface enhance the release of pro-inflammatory cytokines (IL-6, TNF-α) in a human monocyte cell line (THP-1), while only those functionalized with acid groups increase complement activation (González-García et al., 2022).Amine-rich coatings on NPs are often considered more likely to interact with proteins, especially opsonins, leading to their increased phagocytosis compared to other surface functionalities (Wahajuddin, & Arora, 2012).However, it has recently been shown that the increased uptake is followed by the production of anti-inflammatory cytokines, reducing the inflammatory effect of the NPs (González-García et al., 2022).

Stealth strategies for NPs
As an active stealth strategy, particles can be coated using cell membranes, e.g. from leukocytes (Parodi et al., 2013) or red blood cells Fig. 3.A hypothetical model for nanoparticle antigenicity proposed by (Zolnik et al., 2010).Antigenicity is induced by conjugation of a NP with a protein (e.g., HSA).After endocytosis of the NPs, the B-cell may present HSA peptides via their Class II human leukocyte antigen (HLA) molecule to T-helper cells through interaction with T-cell receptors (TCR).The activated T-cells then produce pro-inflammatory cytokines, and activated plasma B cells start producing antibodies directed towards the conjugated NPs (Image created using Biorender).(Xia et al., 2019), or by attaching "don't eat me markers" like CD47 onto their surface (Rodriguez et al., 2013).Most commonly, polyethylene glycol (PEG) is used as a stealth molecule; however, the application of PEG comes with its own challenges.PEGylation increases the hydrophilicity of NPs and prevents undesired interactions with proteins and opsonins through steric hindrance, causing "stealth" behaviour and longer circulation time.The effect of PEGylation is largely dependent on the polymer chain length, density, and thus, its conformation (Shi et al., 2022).Clusterin, a dysopsonin, was found to have a specific affinity towards PEGylated NPs, providing an additional explanation for their reduced uptake (Aoyama et al., 2016).However, PEGylation is not the ultimate solution to avoid the immune responses triggered by NPs.Sensitivity reactions and accelerated blood clearance (ABC) of PEGylated NPs upon repeated administration have also been reported (Verhoef et al., 2014).In ABC, both NP and PEG properties, such as their size, charge, PEG density, PEG length, PEG terminal groups, and cargo as well as the timing between doses play a role (Ilinskaya & Dobrovolskaia, 2016) and the overall response appeared to be linked to the presence of IgM antibodies against PEG.Hypersensitivity related to exposure to PEGylated NPs, ranging from local inflammation to CARPA, has been more commonly reported since the 1990s (Szebeni et al., 2018;Wenande & Garvey, 2016).In a recent study, PEG antibodies, both IgM and IgG, were found in 22-25 % of healthy blood donors (Garay et al., 2012).
While the pathways of immune-stimulation by PEGylated NPs are not completely clear, some studies have suggested that after binding of PEG-antibodies, both the lectin and alternative complement pathways are fully activated (Hamad et al., 2008).However, it has also been reported that the presence of PEG-antibodies does not correlate with complement activation (Neun et al., 2018).Readers interested in this topic can consult a recent extensive review on the immunological effects of PEGylated NPs (Shi et al., 2022).

Methods and models to evaluate immune response towards nanoparticles
Current methods to assess NP immuno-toxicity are overall derived from the field of chemical safety assessment.Still, of note, specific tests are needed to account for the unique properties of NPs compared to bulk materials, as for example their size, surface area (per weight unit), agglomeration and reactivity, which all influence their absorption, transportation through biological barriers, biodistribution, accumulation in specific organs, and circulation time.Efforts have been made to establish standardized preclinical testing for NPs designed for medical applications (Ramos et al., 2022).Yet, as of now, the FDA only offers a draft guidance (2022) on drug products containing nanomaterials (U.S.Department of Health and Human Services, 2017) and a guideline for the preclinical testing of liposome formulations (U.S.Department of Health and Human Services, 2018).
Immuno-toxicological studies on NPs are routinely performed either in vivo using different animal models, or in vitro using human or animalderived cells (immortalized cell lines or primary cells).Both approaches have specific advantages and limitations (Braakhuis et al., 2015), and by definition, no model is perfect.In immunology, in vitro/in vivo relations are often poorly correlated, due to the highly complex interactions between cells and molecules involved in the responses of the immune system (Dobrovolskaia & McNeil, 2013;Dubaj et al., 2022).Importantly, NP studies with the immune system come with another set of challenges.First, it is essential to select a representative model, with adequate complexity for the targeted endpoint, and include relevant positive and negative controls (Dobrovolskaia & McNeil, 2013;Ilinskaya & Dobrovolskaia, 2016).Additionally, it is important to fully characterise the NP sample which typically comprises more than the pristine NPs, as surfactants are often added to stabilize the NP suspension and since NPs can aggregate into larger clusters.This characterisation is critical because NPs and said surfactants can interfere with assays, for example by scattering or absorbing the light used in optical assays, and eventually affect the cell-NP interactions.
We summarise below various approaches which have been employed to study the interactions between NPs and the immune system, using models with increasing level of complexity, starting with their physicochemical characterisation including the presence of a protein corona, in vitro cellular assays and in vivo animal studies.Finally, we introduce a relatively new class of bioengineered in vitro models, or Organ-on-Chip (OoC) models, which enable the study of NP immunotoxicity following a 3D all-human approach.

Physicochemical characterisation, protein binding and complement activation
Before any biological analysis can be performed, the physicochemical characteristics of the NPs must be established.These include the particle size, charge, composition, solubility, architecture, and surface properties and surface ligand stability, leaching chemicals, and possible contaminations.Examples of routine particle-sizing techniques are dynamic light scattering (DLS) and transmission or scanning electron microscopy (TEM and SEM).Next, the NP zeta potential analysis of the NPs is measured to determine their charge state.Other assays and techniques to characterize NP physicochemical characteristics have been recently reviewed elsewhere (Joudeh & Linke, 2022;Ramos et al., 2022).Gathering information about the properties of a NP sample enables researchers to differentiate between effects due to the NPs or other components in the sample.For example, common surfactants such as Tween or Triton X-100 can cause disruption of cell membranes and interfere with protein binding on the NP surface (Dobrovolskaia & McNeil, 2013).Furthermore, contamination in the NP samples with bacterial lipopolysaccharide (LPS) can lead to a strong inflammatory response, which is not directly attributable to the NPs (Boraschi et al., 2017;Dobrovolskaia & McNeil, 2013;Li & Boraschi, 2016).Recently, Mukherjee et al showed that graphene-based NPs could interfere with the commonly used Lumilus amoebocyte lysate (LAL) assay, for the detection of endotoxins like LPS (Mukherjee et al., 2016); therefore, alternative methods must be developed.Endotoxin contamination must be avoided since even very small amount of LPS can cause severe immune responses e.g., inflammasome activation, pro-inflammatory cytokine release and inflammatory phenotype as could be expected from the NPs (Ngkelo et al., 2012).
Since the protein corona formed on NPs impacts their interaction with host cells, and the immune system, we will discuss methods to characterize the protein corona.Its presence and thickness can be measured by the same particle sizing techniques as described above (DLS, TEM, SEM), while the total protein amount can be quantified by colorimetric assays (e.g., Bradford assay).The composition of the protein corona is more difficult to determine.As for any protein analysis, it often starts with a 2D separation approach, where the proteins are separated from the NPs using, for instance, gel electrophoresis (SDS-PAGE: sodium dodecyl sulphate polyacrylamide gel electrophoresis) followed by mass spectrometry.Alternatively, 2D-liquid chromatography techniques can be employed instead of the 2D-gel electrophoresis, to separate proteins, before on-line mass spectrometric analysis.A comprehensive overview of methods to analyse the protein corona is available elsewhere (Carrillo-Carrion et al., 2017).Importantly, the methodology followed to create a model protein corona fundamentally impacts the outcome of the analysis in terms of detected proteins.First, protein binding or protein corona formation typically starts with incubation of the NPs with individual proteins or serum or plasma; here, the composition of the protein mixture, temperature, protein concentration, NP-to-protein ratio and incubation time are important parameters influencing the formation and eventual composition of the protein corona.Often, a washing step is performed before analysis, and the conditions of this washing step are known to influence the stability of the protein corona, and therefore, its composition (Böhmert et al., 2020).A limited number of studies have focussed on analyzing the protein corona in vivo NP samples; here, important parameters for the outcome of the analysis, include the administration route, exposure time, method used to recover the samples and as with in vitro studies the analysis method (Bai et al., 2021;Singh et al., 2021).
Using the techniques described above, the binding of complement proteins can also be confirmed.However, to verify complement activation, other experimental methods are employed, of which the CH50 method is the most common.This method relies on assessing the lysis of antibody-sensitized sheep erythrocytes, for which pathway-specific assays are available.In order to avoid direct haemolysis by NPs, it is critical to remove them from the serum before incubation with the erythrocytes to determine complement activation (Pondman et al., 2014;Salvador-Morales & Sim, 2013).Other complement activation strategies are based on immunoassays such as an enzyme-linked immunosorbent assay (ELISA), which detect the presence of targeted complement products, indicating the activation of specific pathways (e. g., C3a, iC3b, C4d, C5a, C5b-9).More recently, SPR (Surface plasmon resonance) has been employed, as a more reliable high-throughput alternative to ELISA (Coty et al., 2018).

In vitro cellular studies for immune interactions with NPs
Following the NP physicochemical characterisation and assessment of the protein corona, a broad range of in vitro cellular studies can be used to examine the interactions of NPs with cells in general, and immune cells, in particular.
Hemotoxicity assays are used to determine the effect of NPs on blood cells and clotting.Haemolysis, which is the lysis of red blood cells, is commonly evaluated using a colorimetric assay to analyse the released (plasma-free) haemoglobin in solution (Dobrovolskaia et al., 2008).Thrombogenicity, or the potential to cause blood clotting, can be assessed in various ways: by detection of the activation of the coagulation cascade, or by determining the clotting time of plasma or whole blood.Furthermore, the agglomeration and/or activation of platelets can be analysed.Of note, NPs can interfere with these assays through adsorption of free haemoglobin on their surface, reducing the free haemoglobin measured in solution (Dobrovolskaia et al., 2008), by sticking to the surface of red blood cells and platelets (Avsievich et al., 2019), and by adsorption of fibrinogen from plasma and protein unfolding, reducing the available functional protein involve in blood clotting (Deng et al., 2012).
Common immune cell lines used to study NP phagocytosis are HL-60 (human suspension leukocyte cells), THP-1 (human suspension monocyte cells), U937 (human suspension monocyte cells) and RAW264.7 (murine adherent macrophage cells).Alternatively, primary peripheral blood mononuclear cells (PBMCs) are employed, PBMCs comprise of, yet in small numbers, other immune cells, like primarily monocytes, (B-and T-) lymphocytes and dendritic cells.Of note, these cells are either in suspension (HL-60, THP-1, U937, lymphocytes in PBMCs) or adherent and grown as a monolayer (RAW264.7,monocytes in PBMCs and differentiated THP-1 cells) in a standard culture dish, which impacts the way they can interact with NPs.NPs are typically incubated with the cells in the absence or presence of serum/plasma to consider the impact of plasma proteins on the NP uptake, through the formation of a protein corona.Phagocytosis assays on live cells often use fluorescently labelled NPs, which in itself can alter the NP characteristics, and possibly their surface properties, depending on how and where the fluorescent dye is added.Alternatively, a luminol assay allows direct visualization of the phagocytic activity of live cells since this compound becomes fluorescent at low pH, as found in the phagolysosome, indicating thereby uptake of NPs.Phagocytosis can also be evaluated by SEM and TEM imaging after fixation of the samples (Skoczen et al., 2011).Often, a macrophage cell line (derived from THP-1 or selected from PBMCs) is activated towards a M1 or M2 phenotype to study the consequences of NP phagocytosis.It is however worth mentioning that this commonly used M1/M2 classification is an oversimplification of all possibly encountered phenotypes of macrophages in vivo and the pro/antiinflammatory effects they cause (Pajarinen et al., 2013).
In in vitro phagocytosis studies NPs are in fact artificially "fed" to the cells, which are thus not attracted to their presence and do not need to "search" for them.They are confronted with them as "food", especially in the case of monolayer-based models since NPs tend to sediment on their surface, and are therefore, more likely to be taken up by cells than in the human body, where interactions are more incidental (Gustafson et al., 2015).
Inflammatory responses, either pro-or anti-inflammatory or involving immunomodulation, as induced by NPs, are often examined by analysing the cytokine and chemokine release from the immune cells using immunoassays such as ELISA, global secretome analysis (e.g., using mass spectrometry) or even qPCR analysis at the gene expression level.It is important to note that the results depend largely on the type of cells, exposure time and selected cytokine profile so that results cannot be readily compared between studies (Brzicova et al., 2019).
Finally, immuno-toxicity and immuno-stimulation are often determined indirectly by assessing leukocyte proliferation.Toxicity to bone marrow, or myelosuppression, can be studied either by measuring the toxicity of the NPs in haematological cancer cell lines, or more sensitively by a colony-forming unit-granulocyte macrophage assay (CFU-GM), which characterizes the growth and differentiation of pluripotent bone marrow-derived stem cells (Bregoli et al., 2009).
Generally in vitro immuno-toxicity assays involving a single immune cell line are not very informative to predict the in-vivo response towards NPs.None of the immune cell lines accurately represent the wide variety of immune cells found in vivo (e.g., monocytes, macrophages, lymphocytes, NK-cells, DC cells); therefore, a better approach consists of performing a battery of assays including different immune cells.However, the immune system relies heavily on cells interacting with each other and without taking this complexity into account the value of the results will always be limited (Dobrovolskaia, 2015).

In vivo animal studies for immune interactions with NPs
In vivo immunological studies, which are typically pursued to gain a broader indication of an immune-toxic response at the systemic level, include hypersensitivity assays, NP distribution and accumulation studies, and histological evaluation of immune organs and tissues.Using representative in vivo models allows evaluation of the bi-directional communication between NPs and the immune system, including circulating and tissue-resident factors and cells, and its regulators, e.g.hormones and chemokines.Furthermore, in vivo exposure to NPs can be followed by haematological studies that examine haematocrit values (indicating haemolysis) and leukocyte counts (indicating leukopenia and immuno-toxicological effects).Lastly, effects on the adaptive immune system are deduced from antibody titres in blood which are determined by ELISA (Dobrovolskaia & McNeil, 2013).
While these in vivo studies yield more complete and systemic information, they cannot be readily extrapolated to human-beings due to essential physiological inter-species differences in the immune system.This can be illustrated by a phase I study where six healthy volunteers suffered from massive cytokine storm and multiorgan failure after (non NP) antibody treatment which was not found in a preclinical test in rats and nonhuman primates, but which could be reproduced using human peripheral blood nuclear cells (st. Clair, 2008).
There are differences between species not only in sensitivity towards different agents, but also, in the general composition and global organization of the immune system (Szebeni et al., 2007).Since mice and rats are resistant to endotoxins (Munford, 2010), more sensitive species such as rabbits are needed.This rabbit pyrogen test is also useful when a LAL assay might be inconclusive, e.g. when there are interactions between the assay and the NPs or endotoxins hidden inside (porous) NPs.
One of the main reasons to perform an in vivo test is to study CARPA.
This severe condition involves not only activation of the complement system, but also the stimulation of allergy mediating cells, (eg.mast cells, basophils) by anaphylatoxins and effects of allergic and vasoactive mediators on cells mediating the organ responses (e.g.smooth muscle cells and endothelial cells).Overall, the manifestation of CARPA can be influenced in nearly 40 different checkpoints, which can all vary between species and individuals.Dogs and pigs are more sensitive for CARPA (Szebeni et al., 2007), and display CARPA mainly through pulmonary hypertension, which might not only be due to complement activation, but also because of direct interactions between pulmonary intravascular macrophages and NPs (Mészáros et al., 2018;Szebeni, 2018).This can also be explained by the fact that where in humans, mice, rats, dogs and primates the Kupffer cells in the liver are the primary cells responsible for clearance of NPs; in sheep, cows, cats, goats and pigs the pulmonary intravascular macrophages are responsible for the bulk of the clearance (Brain et al., 1999).Furthermore, a model where pulmonary hypertension can be measured in rodents is currently not available (Zamboni et al., 2018).
A comparison between mouse, rat, dog, and human serum response to iron oxide NPs, revealed that not only the extent of complement activation varies between species, but also the pathway(s) through with activation occurs (Li et al., 2021), and a comparative study with pigs showed they have significantly lower complement activity through all pathways (Salvesen & Mollnes, 2009).A reduced function of complement inhibitor factor H in nonhuman primates leads to higher sensitivity of these species (Henry et al., 1997).
When studying the adaptive immune responses towards NPs in vivo, it is essential to consider the type of T-helper response, Th1 or Th2, as some laboratory animals are biased towards one of these types.This is helpful when studying either Th1 mediated autoimmunity, or Th2 mediated allergic responses, but needs to be considered when evaluating the results.
Due to the differences between human and animal models, humanized models must be considered to get a more relevant image of NP interactions with our immune system.An animal model can be humanized through either genetic or surgical methods.In genetically engineered murine models (GEMMs), human equivalents of genes are introduced into the murine host, which allows for fully immunological proficient models.For example, since MHCs in animals do not resemble human MHCs (Shiina et al., 2017) a humanized HLA (encoding for MHCs) transgenic mouse can be used to study NPs based vaccines (el Bissati et al., 2014).However, not all elements in an immune pathway can be humanized in these models leading to incomplete results.Alternatively, normal or diseased human cells or tissues can be grafted into an animal host, in these cell line or patient-derived xenografts (CDX/PDX) the histologic patterns of the human tissues are conserved, but to avoid rejection of the implanted tissue the model has to be immunodeficient (severe-combined immunodeficient (SCID) mice) (Zamboni et al., 2018).Attempts have been made to generate immunologically humanized mouse models by repopulating the immunologic compartment with either CD34+ hematopoietic precursor cells from human foetal cord blood, or adult peripheral blood mononuclear cells (PBMCs).Alternatively, human lymphoid organs are implanted into SCID mice; unfortunately these models still lack a sufficiently functional human immune system (Zhang et al., 2010).
The differences in response towards NPs between animal models and humans can be estimated by ex vivo phenotypic profiling of blood samples from animals and humans (van der Bol et al., 2010).For example, by analysing the ROS production and phagocytic activity of the blood samples towards NPs and selecting a model with similar response to humans, the toxicity and PK studies can be more accurate (Lucas et al., 2017).
Selecting an "ideal in vivo model" to study immune NP interactions is very difficult, as it requires multi-dimensional assessment of the model's similarities and differences to the human system, especially when immune-active NPs are examined, for example to be used in immunotherapy of cancers, the model should not only recapitulate accurately the human disease, but also include a fully competent immune system.

Advanced in vitro models to study immune responses towards NPs
To fill the gap between existing in vitro oversimplified cellular models and in vivo animal models and the human in vivo situation, advanced all-human in vitro models have been proposed for a variety of applications, including toxicity screening (Fröhlich, 2018).In a first and rather simple version, 3D cell clusters are created that comprise either only one cell type or various cell types, as found in human tissues, with possibly an immune component.These 3D models better emulate the in vivo situation than simple monolayers and cell suspensions since they incorporate cell-cell and cell-extracellular matrix interactions and can be engineered to present adequate tissue stiffness through more or less tissue compaction.To control size and composition of the 3D models, various engineering approaches have been proposed, using microwell arrays (Rivron et al., 2012;Sridhar et al., 2014), hanging drops (Tung et al., 2011), droplet microfluidics (McMillan et al., 2016;Sart et al., 2022), possibly arrayed in a microfluidic device (Saint-Sardos et al., 2020;Tomasi et al., 2020), surface patterning (Hardelauf et al., 2011), as reviewed elsewhere (Eke et al., 2022;Picollet-D'hahan et al., 2016).
In 3D cell models the cells have more realistic cell-cell interactions, leading to a larger variety of tissue types.For example, a dense and difficult to penetrate tissue can be formed using NIH-3T3 fibroblast and a loose tissue when 3D culturing A549 lung cancer cells.This variation in tissue type has an impact on the penetration depth and exposure to the NPs, leading to variations in toxicity profiles (Braakhuis et al., 2015;Sambale et al., 2015;Tchoryk et al., 2019).For example, in 2D and 3D cultures of Caco-2 cells, ZnO NPs showed very different toxic effects.In 3D, ZnO NPs induced an acute inflammatory response and toxicity to the cells which led to apoptosis, while in 2D Caco-2 culture exposure to ZnO NPs led to necrosis and not apoptosis (Wu et al., 2017).
At the next level, miniaturized organ models have been proposed in the form of organs-on-chip (OoC) (Morneau & Whitworth, 2022).OoCs can be defined as hybrid models combining, on one hand, microfabricated structures that aim to mimic key architectural or functional elements of an organ/tissue, with, on the other hand, cells, and possibly different cell types.The beauty of these OoC models is that they can be engineered for targeted applications, to include all necessary cells, extracellular components (e.g., extracellular matrix), functions relevant for said application as well as on-chip cell stimulation (mechanical (Paggi et al., 2020), electrical (Pavesi et al., 2015), or soluble).Furthermore, these models can be all humanized, by using only human cell lines or primary human cells, to come closer to the in vivo situation.They can even be prepared using cells isolated from specific human beings, since they require limited amounts of cells per model, or induced Pluripotent Stem Cells (iPSCs) (Ramme et al., 2019) to yield in both cases personalized models.As such, they allow examining at the level of an individual, the impact of NP exposure.
Different strategies can be pursued to create OoC models.In the simplest approach, a 3D cellular model, in the form of a cell aggregate or a cell-laden hydrogel, is cultured in a microfluidic device.The advantages of this microfluidic format for the culture of these 3D engineered tissues, compared to standard culture dishes are the presence of dynamic culture conditions with implementation of continuous or pulsatile flow, the possibility to accurately regulate physical and chemical parameters of the cell microenvironment, and the high level of confinement in the micrometer-sized structures found in the devices (Harink et al., 2013;Young & Beebe, 2010).The same approach applies to ex vivo pieces of (human) tissue that can be cultured in such miniaturized bioreactors (McLean et al., 2018), with the advantage that there is no need to engineer the tissue model.For instance, ex vivo pieces of testis have been cultured in a microfluidic reactor under continuous perfusion, while monitoring their secretory function (Sharma et al., 2022).This platform would allow in the future examining the impact of NP exposure to the integrity of the blood-testis barrier and, eventually, on the male reproductive function (Sharma et al., 2020).A commonly used format for OoC is a microfluidic Transwell system (Moraes et al., 2012), with two fluidic chambers representing the apical and basolateral compartments, separated by a horizontal porous membrane onto which epithelial cells are grown.The combination of this porous substrate and continuous flow has proven essential for the differentiation of epithelial cells into a threedimensional morphology, and the realization of various physiological barriers (Huh et al., 2010;Kim et al., 2012).This class of OoC is of particular interest to study interactions with exogenous NPs that must first cross a barrier (in the lung, intestine, placenta, skin, etc.) to enter the human body and eventually interact with the immune system (Braakhuis et al., 2015).In this process, the protein corona around the NPs and their overall characteristics are expected to change since differentiated epithelial cells secrete significant amount of mucus, comprising surfactants and surfactant proteins, that can all adsorb onto the NPs.In a seminal example developed by the Wyss institute, continuous stretching of a lung epithelium in a lung-on-chip model to emulate breathing motion, was found to significantly impact interactions of silica NPs with the blood-alveolar barrier (Huh et al., 2010).Of even higher value for immunological studies, in the same model blood borne immune cells were introduced in the bottom compartment in the form of neutrophils, which were shown to adhere onto the porous membrane and migrate through it and the epithelial layer in the compartment in which NPs had been added (Huh et al., 2010).This example powerfully illustrates the need for advanced allhuman models to better understand how NPs can interact with physiological barriers, and the immune component therein.A last approach to create OoC models is to customize the microfluidic structures to truly mimic key architectural elements of an organ.For instance, in many organs and tissues, cells are experiencing a soft and curved environment, while all culture dishes and microfluidic devices comprise flat and hard surfaces.To mimic such tubular structures in hydrogel materials resembling the native ECM, several strategies have been developed, as reviewed elsewhere (Virumbrales- Muñoz et al., 2020), notably to create cylindrical blood vessel models (Delannoy et al., 2022), kidney tubular structures (Venzac et al., 2018), among others.
The human in vivo response to NP exposure involves several organs and tissues: the tissue directly exposed to the NP, the immune system, but also the liver in which toxicants end up and are metabolized.Furthermore, immune cells can be recruited from different organs and over large distances.As such, understanding the impact of NP exposure calls for more comprehensive and systemic approaches modelling these different organs and communication between them.For this, multi-OoC models (Picollet-D'hahan et al., 2021)have been proposed that link individual OoC platforms in one closed circulation loop or integrate all required organ models in one plate, together with a pump to actuate fluid flow.A first example of a multi-OoC model including the immune system was proposed by Sasserath et al.; their model included three 3 separated yet connected compartments containing cardiomyocytes, skeletal muscle cells and liver cells, with recirculating THP-1 immune cells, which could infiltrate the various compartments.This multi-OoC was specifically proposed to study immunological effects of therapeutics, which could also include NPs (Sasserath et al., 2020).
An additional challenge of using these complex models to study NPs interactions with the immune system and other organs, is that they include many surfaces, often based on different materials, onto which NPs can adhere; these include, for example, the devices themselves but also the capillary tubing used to connect different parts and the device to an actuation system to apply controlled flow.Readers interested in other OoC platforms to study the efficacy and safety of NPs are referred to a recent review (Kang et al., 2021).

Nano-immunomodulation, a problem?
Activation of the immune system by NPs is likely to intervene with the intended use of NPs in the biomedical field, for example by removing the NP from circulation before the intended goal is reached.Furthermore, for both engineered and environmental particles, a response of the immune system towards the NPs can lead to severe adverse reactions, for example, inflammation or toxicity in liver or spleen where NPs are sequestered, causing depletion of the cells that allow appropriate functionality of the immune system.As discussed earlier, there are many factors that influence the interactions between the immune system and NPs, many of these are dependent on the type of NP, of which a broad range of types are available (Fig. 4).In this section, we will focus on four main categories of NPs, which are liposomes, polymeric, inorganic and carbon-based, each presenting unique characteristics, which can be exploited for medical applications.We will put special focus on how these differences influence their interaction with the immune system and hence immunotoxicity.Of note, while more key mechanisms and pathways of immune activation and suppression by NPs are regularly identified, many factors remain unknown.

Liposomes
Liposomes, are the most "cell-like" NPs since these spherical vesicles are encapsulated by a phospholipid shell.First introduced by Bangham et al. (Bangham et al., 1965), liposomes are now the most common clinically approved nanocarrier platform for drug and gene delivery.They are easily produced using a range of different phospholipids, commonly by sonication of (a thin film of) dried phospholipids in aqueous solutions, with great control over both their size and surface chemistry (Guimarães et al., 2021).Both hydrophilic and hydrophobic agents can be encapsulated in liposomes, either in their aqueous core or their hydrophobic lipid bilayer (Liu et al., 2022).In general, the cell membrane-like shell around the liposomes is poorly immunogenic, while there are exceptions (Inglut et al., 2020).Hypersensitivity reactions were reported in patients treated with PEGylated liposomal doxorubicin (Doxil®) and 99m Tc-HYNIC PEG liposomes, which could be attributed to the fact that these liposomes act as potent complement activators in vitro and in vivo (Szebeni et al., 2000;2002).Comparison of liposomes with similar size and composition, but without doxorubicin, revealed that a potential cause of the enhanced complement activation by Doxil® is surface-bound or membrane-trapped doxorubicin (Inglut et al., 2020;Szebeni et al., 2000;Wang et al., 2007).
Reports on the immune response towards liposomes themselves are scarce.Alving et al. reported that a group of human antibodies that recognize phosphate and sulphate esters, which are ubiquitous in phospholipids and thus cell membranes, was able to specifically bind phosphatidylcholine (PC)/Cholesterol (Chol), PC/Chol/dicetylphosphate (DCP) and phosphatidylserine (PS)/Chol liposomes (Alving, 1984;1986).This finding can be attributed to the presence of TLR-ligands on the liposome surface, which can stimulate B-cells, resulting in the expansion of B-cell clones producing IgM antibodies specific to the lipid constituents of the liposomes (Szebeni et al., 2000;2007).
Cationic liposomes are attractive vehicles for the delivery of negatively charged RNA.However, several employed cationic lipids (e.g., DOTAP or dioleoyl trimethyl ammonium propane) induce ROS, interact with APCs, and trigger inflammatory cytokine expression (Schwendener, 2014;Vasievich et al., 2011).Replacing these lipids with Dipalmitoylphosphatidylcholine-PEG (DPPC-PEG) or 1,2-Dipalmitoylsn-glycero-3-phosphoethanolamine-PEG (DPPE-PEG) reduces the zeta potential of the liposomes and leads to a reduction in immunogenic response.However, in general, compared to neutral liposomes and anionic liposomes, cationic liposomes are taken up by phagocytic cells more effectively and induce more ROS formation, leading to damage to mitochondria specifically as well as cells (Inglut et al., 2020).
While liposomes are an attractive instrument to reduce the side effects induced by many drugs in patients, the liposomes themselves can lead to immuno-toxicities that are still not fully understood.

Polymeric NPs
Polymeric NPs can be divided into two principal categories: those that are intentionally engineered to serve a specific purpose with predetermined and well-defined characteristics, and, then, those which are by-products or degradation products of larger polymeric objects, e.g.plastic bags, food containers and clothes with difficult-to-define characteristics.Exposure to this second category of polymeric NPs is not well-defined, rendering them a greater but largely unknown health hazard.Since polymers can be synthesized from a wide variety of monomers with different properties, engineered polymeric NPs are available in a range of shapes, sizes and with distinct (surface) properties, e.g., biodegradability or porosity, rendering them fully tuneable for their application, for example, in drug delivery (Begines et al., 2020).
Both natural and synthetic polymers are employed to produce NPs; synthetic polymers offer a larger variety of formulations and more precise batch-to-batch consistency.Biodegradable polymers such as polylactic acid (PLA), Poly(lactide-co-glycolic) acid (PLGA), Poly(L-Lysine) (PLL), and sugars such as chitosan or dextran are preferred for biomedical applications (Karlsson et al., 2018).However, biodegradability does not mean the particles are safe to use.For example, PLGA NPs do activate complement; the extent of activation depending on the NP surface functionalization and size (Fornaguera et al., 2015).Larger PLGA NPs attach to the cell membrane to constitute a potent inflammatory effect through NF-κβ translocation into the cell nucleus, while smaller PLGA NPs are readily taken up by phagocytic cells and DCs (Lutsiak et al., 2002).
In polymeric NPs drugs can be either encapsulated within their core, entrapped in or conjugated to their polymer matrix, or bound to their surface after formation of the particle.Finally, release of therapeutic agents can be triggered after incorporation of specific stimuli-responsive Fig. 4. Examples of NPs used and/or proposed for use in biomedical applications (center), and places they occur naturally, whether they are found in environmental pollution (e.g. from factories, cars and other traffic and volcano's but also degradation of large plastic items) or added in commercial products, which include food items as milk, yogurt, chocolate, fish and sea salt, but also sunscreens and other cosmetics (outer circle).The NPs are divided into four main categories based on their main material composition: Liposomes (lipids), polymers, inorganic and carbon.(Image created using Biorender).
elements (Karlsson et al., 2018).The specific method of drug encapsulation and whether the drug is exposed to the body will have an effect on the interaction with the immune system.
With the growing concern arising from the ubiquitous presence of micro-and nanoplastics (MNPs) in our environment, their possible impact on the immune system is currently getting well-deserved attention (Gruber et al., 2022).MNPs have been found everywhere in our environment including deep oceans and artic ice, in our food and drinking water, and even in the air we breathe (Lim, 2021;Sridharan et al., 2022).Recent studies have established that MNPs can enter our bodies, since they have been detected in all human tissues, including the placenta of unborn babies (Ragusa et al., 2021), and most recently, in human blood samples (Leslie et al., 2022).
The health risks of polymeric NPs are generally poorly understood; toxicity studies were only initiated in the last decade and their immunotoxicity is even less explored.However, the effects of MNPs to human health are being increasingly brought under the attention of researchers (Gruber et al., 2022).

Inorganic NPs
Inorganic NPs, consisting of gold, silver, iron, titania, and silica, are found in commercial products, such as sunscreens (silica) and cosmetics, or as antibacterial agent (nanosilver) (also in textiles), drug delivery vehicles (gold and silica) and contrast agents (gold and iron oxides) (Giner-Casares et al., 2016;Wiechers & Musee, 2010).As these particles are non-biodegradable, they are persistently present in cells and tissues.Some NP properties are related to the nature of the metal; for example, Ag NPs induce inflammasome formation, trigger IL-1β release and subsequent caspase-1 activation (Yang et al., 2012), while Cobalt (Co) NPs are rapidly internalized by leukocytes where they cause DNA strand breakage (Colognato et al., 2008).The main mode of metal and metal oxide NPs to cause damage is by inducing ROS generation, which can also be caused by metal ion release as has been found for quantum dots and silver NPs (Yu et al., 2020).ROS generation can, in turn, induce oxidative stress, biomolecule oxidation in proteins, phospholipids and DNA, ultimately resulting in cell death and inflammation (Manke et al., 2013;Nel et al., 2013;Yu et al., 2020).Many metallic NPs (TiO 2 , ZnO, ZrO 2 , and Ag) modulate immune responses via TLRs, through either direct association with TLRs or cooperation with small molecules, e.g., the LPS-binding protein where the complex of the NP with the protein activates TLRs (Luo et al., 2015).
Iron oxide NPs, which are magnetic and also known as SPIOs (superparamagnetic iron oxides), are commercially available, often with a sugar or other polymeric coating, as MRI contrast agents (Montiel Schneider et al., 2022).Other applications of these NPs include (magnetic) drug and gene delivery as well as hyperthermia treatment (Angelakeris, 2017).Dextran-coated SPIOs are associated with complement related side-effects, resulting in case of MRI-contrast agent Feridex TM in discontinuation from clinical use.Specifically, Dextrancoated SPIOs were found to activate complement through both the lectin pathway, and the alternative pathway and in some subjects as well via the IgM-dependent classical pathway (Banda et al., 2014).SPIOs can accumulate in U937 monocytes to induce stimulation of Th1 immune response (Park et al., 2015;Zhu et al., 2011).Again, the coating around the SPIOs is critical for their interactions with the immune system.PEGcoated iron oxide NPs could induce the expression of IL-1β, TNF-α as well as ROS-induced toxicity in THP-1 macrophages but did not affect their polarization (Escamilla-Rivera et al., 2016).PEI (Polyethyleneimine)-coated SPIOs caused M1 polarization of macrophages (as indicated by IL-12, CD40, CD80 and CD86 expression) through activation of TLR4-mediated signalling pathways and ROS production (Mulens-Arias et al., 2015).
As for other NPs for inorganic particles size is critical; Ag-NPs and not Ag-microparticles were reported to be damaging to red blood cells (erythrocytes) causing the release of free haemoglobin into the bloodstream (Choi et al., 2011).Haemolysis is dangerous in itself, since it reduces the number of erythrocytes in blood.Furthermore, haemoglobin is an opsonin, so that NPs binding the haemoglobin can be eliminated via scavenger receptor and phosphatidylserine-mediated phagocytosis (de la Harpe et al., 2019).
PEGylated gold particles, PEG-Au, are often described as ignored by the immune system, and therefore, generally useful in biomedical applications as drug delivery and contrast agents.However, this has been disputed by studies where accumulation in the liver and spleen were found and depletion of monocytes in these critical immune organs was detected indicating immuno-toxicity (Kozics et al., 2021;Veronese & Pasut, 2005).
Immuno-toxicity of inorganic NPs can be reduced by careful design of both the particle itself and its coating, however, currently insufficient (clinical) data is available to produce inorganic NPs which do not evoke any immune response (Ernst et al., 2021).

Carbon-based NPs
Carbon-based NPs vary largely in shape and composition, with spherical fullerenes, needle-like single-and multi-walled carbon nanotubes (SWNTs, MWNTs), and sheets of graphene and graphene oxide (GO).Although they are not yet commercially available, applications of carbon-based NPs are foreseen as anti-cancer drug carriers and contrast agents (Zhang et al., 2021).Environmental carbon NPs, also known as urban particulate matter that are by-products of traffic and industrial activities, come as a complex mixture of different types and sizes (Alemayehu et al., 2020).
Carbon-based NPs come with their own risks.Radomski et al. showed that many carbon-based NPs (C60s, SWNT, MWNTs and urban particulate matter) can cause platelet aggregation through activation of the glycoprotein integrin receptor GPIIb/IIIa via the protein kinase C (PKC) for larger particles and in a PKC-independent manner for nano-sized particles (Radomski et al., 2005).
GO NPs activate complement through the classical pathway in an oxidation state dependent manner, since higher oxidation grades of GO result in more complement activation, potentially due to their higher solubilization state (Wibroe et al., 2016).Both in vitro and in vivo studies demonstrated that GO NPs can promote pro-inflammatory polarization of macrophages (Ma et al., 2015;Orecchioni et al., 2016), this effect being triggered by activation of inflammasomes with subsequent IL-1β secretion (Mukherjee et al., 2017).GO NPs can align with the cell membrane of phagocytes, to promote their uptake while interfering with the function of the phagocytes by blocking large parts of their membrane (Russier et al., 2013).This alignment can also explain why GO NPs, and not fullerenes, were found to suppress antigen presentation to T-cells (Tkach et al., 2013).GO triggers NET formation in primary human neutrophils at low doses, where it is degraded in a myeloperoxidase dependent manner, indicating the capability of neutrophils to digest carbon-based nanomaterials (Mukherjee et al., 2017;Mukherjee et al., 2018b;Mukherjee et al., 2018c).
Carbon nanotubes (CNTs) bind C1q through its globular heads, which initiates the classical pathway of the complement system.Opsonisation of CNTs significantly enhances their uptake by phagocytic U937 cells and human monocytes.However, complement activation through the classical pathway is associated with down-regulation of proinflammatory cytokines and up-regulation of anti-inflammatory cytokines (Pondman et al., 2014).In a similar study, human properdin, an up-regulator of the alternative pathway, was also reported to enhance the uptake of CNTs by THP-1 cells; however, this uptake was followed by a pro-inflammatory immune response (Kouser et al., 2018).SWNTs are able to bind via hydrophobic interactions to TLR2/4 on primary human macrophages, both in the presence and absence of serum proteins, resulting in MyD88 (myeloid differentiation factor 88) dependent activation of NF-kB and secretion of chemokines (MIP-1α and RANTES) (Mukherjee et al., 2018a).
The asbestos-like shape of CNTs has triggered many studies due to their tendency to cause granuloma in lungs and mesothelioma (Dong, 2020;(Pondman et al., 2017a); Vietti et al., 2015).Depending on the aggregation state of CNTs used for lung exposure, diffuse interstitial fibrosis through the lungs or focal granulomas was found (Mercer et al., 2008).The exposure route did not influence mesothelioma development, as this severe condition was found after CNT exposure by inhalation, intratracheal intrapulmonary spraying, intrapleural injection, and intraperitoneal injection (Zhang et al., 2021).The inflammatory response to CNTs was shown to be both length-and rigidity-related, as more flexible MWNTs induced a lower airway allergic inflammation in mice (eosinophilia, mucus secretion, Th2 cytokines, pathogenicity) (Poland et al., 2008;Rydman et al., 2014).A recombinant fragment of human SP-D (rfhSP-D) containing trimeric neck and carbohydrate recognition domain (CRD) bound CNTs via its CRD region and enhancedphagocytosis by U937 and THP-1 cells and their pro-inflammatory response, suggesting that sequestration of SP-D by CNTs in the lungs could trigger an unwanted and damaging immune response (Pondman et al., 2017a)(.Thus, a pathophysiological contribution of SP-D in mesothelioma cannot be ruled out.

Modulating the immune system response towards NPs
As already discussed above, the physicochemical properties of NPs determine how they interact with the immune system.Therefore, smart design of the particles, e.g., by manipulating their size, shape, and surface charge accurately, can allow modulation of their interaction with the immune system.NPs with stealth properties have a longer circulation life and can therefore be used for drug delivery to specific locations.NPs lacking these stealth properties, are recognized and cleared directly by the immune system, thereby enabling targeting of phagocytic cells and modulating the actions of the immune system (Zamboni et al., 2018).
Examples are the use of very small NPs to avoid opsonisation (Oyewumi et al., 2010), PEG coating of NPs to induce stealth behaviour (Shi et al., 2022;Veronese & Pasut, 2005), or neutralization of the net charge to effectively suppress complement activation (Shan et al., 2009).Alternatively, complement activation by the alternative pathway (due to direct binding of C3) can be targeted by carboxylation of the NP surface (Thomas et al., 2011).
A simple approach to dampen the immune response evoked by NPs is by coating them with serum proteins to protect against inflammatory effects caused by "naked" SPIOs (Escamilla-Rivera et al., 2016).A more elegant strategy relies on specific attachment of C1q onto NPs, which enhances uptake of CNTs by macrophages and downregulates the proinflammatory immune response.A similar effect was successfully achieved by binding only the recombinant C1q globular heads onto CNTs, suggesting an additional level of complexity in receptor interaction (Pondman et al., 2015).Similarly, recombinant properdin TSR4 + 5 coated on the CNTs could inhibit complement activation through the alternative pathway by CNTs, suggesting that nanoparticle decoration with TSR4 + 5 can potentially be used to inhibit complement in several pathological contexts arising due to its exaggerated activation (Kouser et al., 2018).In the lungs, coating NPs with SP-D can be used to induce a desired pro-inflammatory immune response (Pondman et al., 2017a).Complement down regulator Factor H and an artificially engineered derivative mini-factor H were able to inhibit complement activation by liposomes (AmBisome) (Mészáros et al., 2016); however, when this strategy was applied to CNTs, only phagocytosis was reduced by factor H (Kouser et al., 2018).
Inhibition of complement activation by NPs can be achieved by their conjugation to complement inhibitors, such as recombinant C1 inhibitor, peptide-based C3 inhibitors (e.g., Compstatin, CD35, CD55) (Gaikwad et al., 2020), recombinant truncated CR-1 and C5aR1 antagonists or peptides that recruit factor H (Wu et al., 2011).Several pathogens (e.g., HIV-1) use Factor H binding to evade the immune defence, which is thought to be due to the presence of sialic acid on their surface.Kim et al. attached sialic acid onto NPs, which was found to not only allow evasion of the immune response, but also to increase targeting to tumour cells (Kim et al., 2017).
Other possibilities to modulate the immune response are direct conjugation of cytokines onto NPs or synthetic oligonucleotides, which can be either immune-suppressive or immune-stimulating.This will lead to interactions with TLRs, subsequent targeting, and activation/suppression of immune pathways (Klinman et al., 2016).
Immuno-suppression can be advantageous to prevent allergies, autoimmune diseases and rejection of transplanted tissues and organs.NPs can act as carriers to deliver immunosuppressive drugs, e.g., with the controlled release of steroids from PLGA NPs (Higaki, 2005) and liposomes containing clondronate to avoid skin graft rejection by reducing macrophage activity (Li et al., 2016).Allergen-loaded NPs are interesting to carefully dose an antigen and increase tolerance to it (di Felice & Colombo, 2017).Ryan et al. showed that by using fullerenes, the IgE receptor mediated signalling and degranulation of mast cells and basophils were decreased, while at the same time, histamine release was prevented, thereby reducing the effects of anaphylaxis in a mouse model (Ryan et al., 2007).

NP based vaccines
The vaccines that are currently developed can be classified into the conventional protein-based or more contemporary gene-based approach.Proteins can act directly as immunogens to activate an immune response, whereas gene-based vaccines are delivered via DNA or RNA vectors to a host cells, where they need to be expressed to produce the antigens in order to induce the immune response.When the antigen is produced, both strategies follow a similar pathway.In the inductive phase directly after administration, a release of inflammatory cytokines and immune cell recruitment (neutrophils, macrophages and APCs) occur.The APCs internalise the NPs and load the antigen on MHC molecules.The antigen-loaded APCs move to draining lymph nodes, were they present the antigens to T-cells, which differentiate into type 1 or type 2 T helper cells.Activated T helper cells interact with antigenspecific B lymphocytes, which stimulates their maturation into plasmablasts, plasma cells and memory B cells.The activated immune cells leave the lymph nodes and enter the bloodstream where plasma cells secrete antibodies (Guerrini et al., 2022).
NP-based technologies show great promise to overcome problems encountered with currently used protein or mRNA vaccines, which are their fast elimination from the body, and low immunogenicity of common antigens, which does not support a sufficiently strong T-cell immune response (Chen et al., 2016) Due to the ubiquitous presence of RNases, mRNA vaccines are degraded without the protection of a NP shell.Furthermore, the immune system recognises exogenous RNA molecules, which is a sign of viral infection, resulting in a strong innate immune response through interactions with PRRs.Therefore, to deliver a mRNA vaccine a NP has to be carefully designed to only deliver its cargo to the correct location (Park et al., 2021).Commonly, liposomes are used to deliver mRNA as these particles most easily encapsulate mRNAs and can enter into APCs (Cullis & Hope, 2017).However, the cationic lipids used in these liposomes destabilize cell membranes (Lv et al., 2006) and induce inflammation (Ma et al., 2005).Therefore, ionizable lipid-like materials are explored, which are positively charged at low pH to allow complex formation with mRNA, and neutral at physiological pH to reduce the toxic effects (Kauffman et al., 2015).
For protein-based vaccines, it is essential that the protein antigen is presented to an APC in the correct conformation.Self-assembling protein nanoparticles (SAPNs) contain replicates of recombinant proteins in a highly ordered manner, acting as a PAMP, to form three-dimensional nanostructures of similar size and architecture to viral capsids.The proteins in a SAPN are accessible as being "seen" by the immune system, and initiate the desired protective immune reaction through interactions with BCRs.In the shell of a SAPN, different immunogens can be selfassembled to bring about a strong immune response (Morales-Hernández et al., 2022).
For a vaccine to be efficient, it must deliver an antigen to DCs and activate them to trigger the adaptive immune response.DCs reside most commonly within the lymph nodes; therefore, the often-considered undesired uptake of NPs and transport into the lymph nodes are in fact beneficial for the development of vaccines.Of note, this effect can be further enhanced by binding TLR antagonists (TLR-7/8a) onto the NP surface (Lynn et al., 2015).
The size of the NPs carrying the antigen is also critical for its effectiveness.Smaller NPs with a size below 50 nm are often considered more effective in activating adaptive immune responses as they penetrate tissue barriers and traffic to draining lymph nodes more easily than larger NPs, which need to be taken up and trafficked to lymph nodes by DCs (Reddy et al., 2007).Using 50-nm PS NPs with ovalbumin as an antigen, phagocytosis by DCs was followed by a rapid transfer to acidic lysosomes and degradation of the antigen.While larger particles (500 nm and 3 µm) remained in a neutral environment after phagocytosis and led to an efficient cross-presentation onto MHC class I molecules (Tran & Shen, 2009).
Uptake of NPs by DCs and activation of T-cells resulting in generation of antigen-specific immune response is modulated by complement; therefore, local complement activation may be desirable for enhancing the antigen presentation to benefit vaccine efficacy.This effect was exploited by Reddy et.al. who coated nanovaccines with Pluronic® surfactant to intentionally activate complement.This led to increased maturation of DCs in the lymph nodes of mice, and after conjugation of ovalbumin to the NPs to a high antibody titer (Reddy et al., 2007).Another effective method to enhance uptake of NPs into DCs is targeting of DC-SIGN (DC-specific intercellular adhesion molecule-3-grabbin nonintegrin) by attaching specific antibodies towards this protein onto the surface of NPs (Cruz et al., 2011;2014).
In conclusion, NPs are very well suited for vaccine delivery as they can protect both protein and mRNA based vaccines from premature degradation, improve stability, assist in targeting APCs, and act as an adjuvant to stimulate the immune response.

NP-based cancer immunotherapy
Cancer immunotherapy is generating significant interest because of its potential to use the body's own defence system to fight tumours reducing the risk of undesired effects (Kraehenbuehl et al., 2022).
Cancer immunotherapies face several challenges which makes the subset of patients responding to therapy very low.This is partly due to the heterogeneity of solid tumours, which spiked the interest in developing patient-specific immunotherapies based on biomarker expression (Scheetz et al., 2019)..The main problem is the tumour microenvironment (TME) which consists not only of tumour cells but includes extracellular matrix, blood vessels, fibroblasts and immune cells, for example myeloid-derived suppressor cells and regulatory T-cells which are actively recruited by tumour cells (Nakamura & Smyth, 2020).The TME has inhibitory effects on immunity and it blocks the access of T cells, which could recognize and kill tumour cells (Chen et al., 2022;(Leko and Rosenberg, 2020;Morotti et al., 2021)).Besides efficacy issues, there are safety issues with cancer immunotherapy, as it can induce an autoimmune response, leading to attacks on healthy cells/tissues (Milling et al., 2017).
Administrating cancer immunotherapy in a more controlled manner could improve the safety and efficiency of the method.Therefore, over the past decades, the potential of NPs to boost host anticancer immunity has been explored (Irvine & Dane, 2020).As discussed earlier, NPs can deliver antigens, cytokines, chemokines and TLR antagonists to immune cells in a controlled and targeted manner.The first immunotherapies were based on recombinant cytokines, for example IFNα and IL-2, which lead to (partial) remission, but suffered from side effects and short therapeutic duration (Lee & Margolin, 2011).The application of PEGylated gold nanoparticles for the delivery of recombinant TNF-α successfully reduced the systemic toxicity (fever and hypotension) of this cytokine used to treat solid tumours while not leading to the formation of TNF-α-specific antibodies (Libutti et al., 2010).Also, the intrinsic properties of NPs, for example, their ability to generate a strong IL-2 and INF-γ production by leukocytes, can stimulate the immune response against tumours (Lizotte et al., 2016).
Immune checkpoint proteins are overexpressed in the TME, immune checkpoint therapy targets those immunosuppressive elements, by affecting e.g., cytotoxic T-lymphocyte associated antigen 4 (CTLA-4), programmed death-1 (PD-1) or programmed cell death-ligand 1 (PD-L1) (de Miguel & Calvo, 2020).NP-cancer immunotherapy combines several components to achieve an optimal effect.Successful colorectal tumour growth inhibition was achieved using liposomal NPs, which were complexed with a PD-L1 trap plasmid and cationic protamines, that were targeted to tumour tissue using aminoethyl anisamide ligands, combined with systemic treatment with oxaliplatin to activate DCs (Song et al., 2018).
Chen et al. developed a NP-vaccine which delivers a gene to block PD-L1 in cancer cells, ensuring interaction with APCs by including a mannose segment, which allows binding to mannose receptors overexpressed on DCs.In another example, ovalbumin was co-encapsulated as an antigen with unmethylated cytosine-phosphate-guanine (CpG) as an adjuvant.To allow delivery of the short hairpin RNA, shPD-L1, which silences PD-L1 protein expression, to the nucleus a p-toluylsulfonyl arginine element was included.The complex was shielded by hyaluronic acid, which is also a specific target for the CD44 receptor, which is overexpressed on tumour cells.By combining all these elements, playing with different immune components, a potentially effective strategy was successfully developed (Chen et al., 2022).
Reprogramming the polarization of tumour associated macrophages (TAMs) could be an effective method in cancer immunotherapy.Zanganeh found that even without any targeting moieties, Ferumoxytol, an FDA-approved SPIO, successfully induced the polarization of TAMs towards a pro-inflammatory (M1) phenotype, promoting ROS generation and eventually tumour cell killing (Zanganeh et al., 2016).Another strategy is to inhibit TAM function and survival, which was achieved using NPs containing scavenger receptor B type I targeting peptides and M2 macrophage binding peptides, which were subsequently loaded with anti-CSF-1R siRNA (anti-colony stimulating factor 1 receptor small interfering RNA) (Qian et al., 2017) .
Other approaches target circulating immune cells e.g.CD8 + T cells and reprogram them to attack tumour cells, for example by delivering nucleic acids, restoring T cell function, recruiting lymphocytes through TLR7/TLR8 agonists (Schmid et al., 2017), or encoding anti-tumour antigens within DC promoting T-cell responses through IFN-α activation (Kranz et al., 2016).Liposomes, conjugated with both the cytokine TRAIL, which engages death receptors on cancer cells to induce apoptosis, and vascular adhesion receptor E-selecting, could be bound onto the surface of circulating immune cells.The functionalised immune cells not only effectively killed circulating tumour cells, but also tumour cells within a solid prostate tumour.This shows that immune cells with NPs can still actively migrate into the TME to deliver its cargo tumours (Mitchell et al., 2014;Wayne et al., 2016).
In chimeric antigen receptor T cell (CAR T) therapy, T cells are collected from patient blood by leukopheresis and genetically engineered to express CARs that are specific for an antigen present on tumour cells.These engineered T cells are then re-administered in to the same patient where they will attack the tumour cells.However, the transplanted cells rapidly decline after administration; by functionalising the surface of T cells with NPs carrying an adjuvant stimulating the proliferation of T cells, T cell therapy could be improved in murine models (Stephan et al., 2010).To reduce the complexity of the procedure, as an alternative to in vitro genetic engineering, a NP was developed to re-programme T cells, through interactions with CD3, in circulation with leukaemia-recognizing CAR genes (Smith et al., 2017).