Recent Advances in Immunosafety and Nanoinformatics of Two-Dimensional Materials Applied to Nano-imaging

Two-dimensional (2D) materials have emerged as an important class of nanomaterials for technological innovation due to their remarkable physicochemical properties, including sheet-like morphology and minimal thickness, high surface area, tuneable chemical composition, and surface functionalization. These materials are being proposed for new applications in energy, health, and the environment; these are all strategic society sectors toward sustainable development. Specifically, 2D materials for nano-imaging have shown exciting opportunities in in vitro and in vivo models, providing novel molecular imaging techniques such as computed tomography, magnetic resonance imaging, fluorescence and luminescence optical imaging and others. Therefore, given the growing interest in 2D materials, it is mandatory to evaluate their impact on the immune system in a broader sense, because it is responsible for detecting and eliminating foreign agents in living organisms. This mini-review presents an overview on the frontier of research involving 2D materials applications, nano-imaging and their immunosafety aspects. Finally, we highlight the importance of nanoinformatics approaches and computational modeling for a deeper understanding of the links between nanomaterial physicochemical properties and biological responses (immunotoxicity/biocompatibility) towards enabling immunosafety-by-design 2D materials.


INTRODUCTION
Two-dimensional (2D) materials constitutes an emerging class of nanomaterials, characterized mainly by their high surface-area-to-mass ratio due to a sheet-like morphology; responsible for their outstanding physicochemical properties (e.g., electronic, optical, mechanical, and magnetic) with a currently leading position in materials science and technology (1,2). Since the pioneering work of Novoselov et al. (3) in 2004, several 2D materials have been produced for many applications in energy, catalysis, composites, sensors, biomedicine, agriculture, and environmental remmediation (4)(5)(6)(7).
Among several applications, 2D materials have attracted special interest to be applied in the bioimaging field because of their high electrical and thermal conductivity, high degree of anisotropy, exceptional mechanical strength, and unique optical properties (18). Due to such properties, 2D materials have been developed to be applied in molecular imaging techniques, such as computed tomography (CT), magnetic resonance imaging (MRI), optical imaging (fluorescence and luminescence), and nuclear imaging including positron emission tomography (PET) and single photon emission computed tomography (SPECT) (19). Besides, 2D materials allow multimodal imaging by providing a variety of properties useful for more than one imaging technique and/or because of their facility to combine them to form nanocomposites and hybrid materials (20). Given the applicability and growing interests in 2D materials, unveiling their impact on the immune system is a key step towards safe use and responsible innovation (21,22). These materials' intrinsic characteristics, such as chemical composition, surface chemistry, functionalization, morphology, lateral size, purity, and crystallinity are directed related to their degradability, dispersion stability, and protein corona profile; hence, their adverse effects in a biological system (23)(24)(25)(26). Such parameters modulate the biotransformation and biodistribution of 2D materials under in vitro and in vivo models, influencing their interaction with the immune system, fate, and toxicological profile (27)(28)(29)(30).
Biocompatibility, biodegradability, and eliciting an adequate biological effect in the organisms are crucial to the applicability of 2D materials (22,24,31). Indeed, the complexity of toxicokinetic and toxicodynamic events of 2D materials under physiological conditions associated with a lack of harmonized protocols for experimental research represents majors challenges for clinical translation and safety regulation involving these emerging materials (32)(33)(34)(35). Therefore, combining systems toxicology and nanoinformatics is a foremost strategy in the integration of 2D material design on a safe and sustainable basis (36)(37)(38).
In this mini-review, we present the recent advances involving 2D materials, nano-imaging, and immunosafety. Briefly, the main findings associated with the adverse immunological effects were shown in in vitro and in vivo models. Finally, we highlight the great potential of nanoinformatics approaches towards immunosafety-by-design 2D materials ( Figure 1).

TECHNOLOGICAL APPLICATIONS AND INNOVATION OF 2D MATERIALS
A literature review on the Web of Science ™ database was performed, considering articles published from 2000 to 2021 (25/03/2021), and over these last 20 years, many 2D materials have been synthesized as exemplified in Figure 2A. The number of publications of 2D materials and their applications is growing, in which nano-imaging and drug release systems stand out and are present mostly in the health sector ( Figures 2B, D). For energy application, the structural and electronic properties of 2D materials have been shown to improve the energy accumulation in devices such as lithium-ion, metal-air batteries (LIBs) (9,39,49,50) and electrochemical devices (51,52). Moreover, these 2D materials are of particular interest as catalysts and nanoscale substrates, replacing transition, or noble metals normally used to catalyze an acid-basic reaction, producing metal free-catalysts (53,54). In environment, the 2D materials have been used as adsorbents for removing pollutants to treat contaminated water (55)(56)(57). Their atomic thickness and antibacterial activity contribute to superior water permeability and anti-fouling capacity in the development of membranes for desalination (58)(59)(60)(61)(62) and cleaning purposes (63)(64)(65). Sensing has covered both environmental and health sectors, contributing to the detection and monitoring of traces of pollutants (66,67) and blood biomarkers (68)(69)(70)(71). The thin structure, large surface area, chemical modifications and quenching ability of 2D materials provide high sensitivity, durability, stability, selectivity, and conductivity for sensors and biosensors (72)(73)(74)(75)(76)(77)(78)(79)(80)(81)(82).
Considering biomedical applications, 2D materials have been applied in bone tissue engineering, conferring improved mechanical characteristics and great osteoconductivity for scaffold design (83)(84)(85)(86)(87). However, due to the higher surface area of 2D materials and distinguish light-material interactions, research has mostly given attention to their usefulness in nanoimaging and therapeutics (theranostics) (88) (Figures 2B, C), including early detection, monitoring, and treatment of diseases, which are the main examples described in this mini-review (89). For example, in cancer, malign tumors are sensitive to heating when compared to healthy tissues. Graphene oxide decorated with gold nanoparticles (GO-AuNPs), TMDs (MoS 2 , WS 2 ), and MXenes (MoC 2 , Ti 3 C 2 ) have shown effective agents in photothermal and photodynamic therapy for inducing tumor necrosis (40,41,(90)(91)(92). 2D materials have been successfully modified with numerous polymers to enhance their cytocompatibility and dispersibility (90) and used as nanoplatforms carrying active molecules or imaging agents to improve their biological function (93) and clinical visualization for imaging-drug delivery guiding (12,94). MoS 2 and BNNs have been employed as effective fluorescence quenchers and associated with aptamers, substituting antibody-based therapy (69,(95)(96)(97). Compared to the other 2D inorganic materials, and in addition to the previous features, the ultrathin structure of the BP nanosheets results in an exceptional biodegradability in physiological media it shows promising in theranostics (98,99). Magnetic nanoparticles have been used as contrast agents and incorporated into 2D materials in MRI, in place of conventional ones (100,101). In this respect, 2D magnetic materials production can be very useful for accurate bioimaging and therapy of diseases in vivo using MRI and CT techniques (10,102).

2D MATERIALS AND THE IMMUNE SYSTEM: ADVERSE EFFECTS IN IN VITRO AND IN VIVO MODELS
As far as it is known, 2D materials have proven their significance and innovation perspective in almost all industrial areas and sectors, making it imperative to assess their environmental health risks and safety aspects (24,(103)(104)(105). However, toxicological studies, including immunotoxicity, are still in their infancy for GBMs and 2D inorganic materials (31). Table 1 is an extensive literature revision reporting major findings of 2D materials and their adverse effects in the immunological system considering in vitro and in vivo models. The terms used for the literature research is detailed in the supplementary material.
Studies have demonstrated that 2D materials can induce immunological system activation with a consequent induction of an inflammatory response (145). This immunological system activation showed itself to be dependent of the 2D materials' physicochemical properties, such as size (106)(107)(108)(109)144), surface chemistry (114,115,123), number of layers, shape (118,119), and functionalization (109,112,114,128,135,139). For example, Yue et al. (106) demonstrated that larger graphene oxide (GO) (2 µm) has induced a higher immunological activation than smaller GO (350 nm) both in vitro (peritoneal macrophages) and in vivo (C57BL/6 mice). Similarly, Ma et al. (107) showed a lateral-sizedependent pro-inflammatory effect of GO under in vitro and in vivo conditions, wherein the largest GO (L-GO; 750-1300 nm) elicit higher inflammatory response than smallest GO (S-GO; 50-350 nm). Moreover, the mechanism of inflammation has also differed according to the lateral size, with L-GO being more prone to plasma membrane adsorption and the toll-like receptors (TLRs) and nuclear factor-kB (NF-kB) pathways activation, whereas S-GO was mostly taken up by macrophages. In another study that investigated the effects of small GO (S-GO < 1 µm) and large GO (L-GO, 1-10 µm) on human peripheral immune cells, it was found that the S-GO has a more significant impact on the upregulation of critical genes implicated in immune responses and the release of cytokines IL1b and TNFa compared to L-GO (108). However, it is important to clarify here that the S-GO in this study presented similar lateral size of the L-GO in the previous studies cited, which means that all these studies are in agreement, and we may erroneously interpret them because attention to the lateral size was not devoted. Indeed, a nomenclature harmonization of GBMs is urgently needed to allow a clear understanding on the impacts of GBM physicochemical properties on their biocompatibility.
Besides to assess the effect of lateral size, Duarte and coworkers (109) investigated the impacts of two different surfaces functionalization: pegylated graphene oxide (GO-PEG, 200-500 nm) and flavin mononucleotide-stabilized pristine graphene with two different sizes (200-400 nm and 100-200 nm). Their results showed that the cellular uptake of GBMs was mainly influenced by their lateral size, with smaller particles showing greater internalization, while the inflammatory response depended also on the type of functionalization, with GO-PEG showing the lower pro-inflammatory potential. This study corroborates in number previous ones that also showed an increased biocompatibility of GO due to the pegylation (GO-PEG) (110,111). Similarly, Xie et al. (139) studied PEG coated 2D titanium nanosheets (TiNS-PEG) and reported no indication of inflammation and other negative impacts. Moreover, the material was promising for photothermal tumor therapy and presented a high contrast for in vivo imaging. However, Gu et al. (129) found that MoS 2 and PEGylated MoS 2 induced a robust macrophage immune response, with PEG-MoS 2 eliciting stronger cytokine secretion than the pristine MoS 2 . By performing molecular dynamics simulations, they  Live/dead assay, TNF-a, IL-6 and IL1b release; and macrophage polarization, NF-kB signaling activation.
All GO materials have induced a decrease in cell viability, and a production of cytokines. The L-GO significantly elicited higher response than S-GO. Higher macrophage polarization to the M1 phenotype by L-GO than S-GO. Both S-GO and L-GO have induced an inflammatory response by cytokines production and leukocytes recruitment, been the L-GO response higher than the S-GO response in all endpoints. PEG-GO did not induce the macrophage polarization towards the M1 pro-inflammatory phenotype, with a slight shift towards M2 reparative phenotype. Proinflammatory cytokine secretion (IL-1b, TNF-a and IL-6) and proliferation of immune cells.
Only GO-6PEG increased the secretion of TNF-a by RAW-264.7 macrophages without alteration of IL-6 and IL-1b levels. The treatment of primary splenocytes with GO-1PEG and GO-6PEG in the presence of concanavalin A, anti-CD3 antibody, and LPS, produced significant dose-dependent decrease of cell proliferation and IL-6 levels.  GO caused the impairment of cell membrane integrity and functions including regulation of membrane-and cytoskeleton-associated genes, membrane permeability, fluidity, and ion channels. The -NH 2 and -PAA showed similar toxicity to GO, but -PEG and -PAA significantly decreased the GO cytotoxicity.

cells into macrophage-like cells (THP-1a)
Both GO induced a dose-dependent loss in cell viability, an increase in ROS production, and a disruption of the F-actin cytoskeletons leading to the loss of the adherence ability of THP-1a and a reduction in the phagocytosis capability of THP-1a cells. GONP presented higher upregulation of HO-1 and SOD-2 expressions, and higher levels of IL-1b, TNF-a, IL-8, and MCP-1, compared to rGONP. rGONP exhibited a greater expression of NF-кB (p65), higher uptake and a higher decrease of Rho/ROCK expression than GONP.
Pristine graphene with 1% pluronic Increased transcription and secretion of cytokines and chemokines, which is triggered by activation of the NF-kB signaling pathway; The cytokines and chemokines secreted by graphene-exposed macrophages further impaired the morphology of naïve macrophages by affect the actin structures and podosomes expansion, decreasing the adhesion and phagocytosis.      Both materials significantly increased the secretion of cytokines such as IL-6, IL-12, TNF-a, IFN-g and MCP-1. Interestingly, MoS 2 -PEG was found to elicit stronger cytokine secretion than the pristine MoS 2 , particularly involving IL-6, TNF-a, IFN-g, and MCP-1.  Overall, there were no significant differences in cytotoxicity assays, however high doses could promote DC maturation as observed by the expression of CD40, CD80 and CD86 and enhanced secretion of IL-6 and TNF-a. Also, MSNs upregulate ROS generation in DCs, further promoting cytoskeletal rearrangement and promoting the local lymphoid homing ability of DCs. Cell viability and proliferation, ROS production, immunofluorescence to analyze cell morphology, inflammatory marker expression tested by LPS to analyzed cytokine generation (IL-10 and IL-6).
Black phosphorus did not show cytotoxicity on human mesenchymal stem cells and inhibits the metabolic activity of SAOS-2 cell line while inducing both proliferation and osteogenic differentiation in HOb cell and mesenchymal stem cells. Also, the presence of BP inhibits the ALP (an early marker of osteogenesis) expression in SAOS-2 cells and induces antiproliferative and apoptotic effects by increasing the production of ROS on SAOS-2 cells. Besides, increase the inflammatory cytokine generation but inhibits proinflammatory mediators for the co-culture of SAOS-2 and HOb. Cell viability, necroptosis, protein expression, cytokine generation (IL-6 and TNF-a) and hemocytolysis.
No obvious cytotoxicity was observed, also no significant hemolysis. For BPTT treatments it was observed that necroptosis play an important role, mediating death process in cancer cells. These results were confirmed by the expression of necroptosisrelated proteins, where it was observed a significantly expression of RIP1 and RIP3. Caspase-8 and Caspase-3 levels were not significantly changed.  At the dose of 750 or less no mice mortality nor any reaction was observed. The nanomaterial mainly accumulated in liver, spleen and kidney, however, the concentration decreases with time.
(Continued) Other types of functionalization have also been studied in respect to their biocompatibility to immune cells. For instance, Zhi et al. (112) reported that the polyvinylpyrrolidone (PVP) coating of GO has exhibited lower immunogenicity when compared with pristine GO in relation to the inducing differentiation and maturation of dendritic cells (DCs), provoking a delaying in apoptotic process of T lymphocytes and the anti-phagocytosis ability against macrophages. Surface chemistry has also been shown to influence on the immunotoxicity of 2D materials. Gurunathan et al. (114) reported that both GO and reduced GO (rGO) induced a dose-dependent loss of cell viability and proliferation, cell membrane damage, a loss of mitochondrial membrane potential, a decreased level of ATP, a redox imbalance, and an increased secretion of various cytokines and chemokines (IL1-b, TNF-a, GM-CSF, IL-6, IL-8, and MCP-1) by THP-1 cells. However, to all these toxic effects the rGO presented a significantly worse response compared to GO. In a previous study, Yan et al. (115) showed that different oxidation degrees resulted in the toxicity of monocytes via different signaling pathways, with GO nanoplatelets (GONPs) inducing the expression of antioxidative enzymes and inflammatory factors, whereas the reduced GO nanoplatelets (rGONPs) activated the NF-кB pathway. The contradictory results between these two studies, in relation to cytokine and chemokine expression, may Pd nanosheets have no effect on cell viability, apoptosis, ROS generation, or mitochondrial depolarization.
The in vivo results show that the particle is primarily trapped by reticuloendothelial system (RES). Also, no significant hepatotoxicity was induced by Pd nanosheets of different sizes. The activity of ALP, ALT, AST and BUN observed was within normal range and no apparent histopathological abnormalities or lesions were observed in any major organ. Cell viability, ROS generation. No significant cytotoxicity was observed for B-PEG NSs. However, when exposed to an 808 nm NIR laser (1 Wcm −2 ) for 5 min it was notices a strong concentration-dependent cytotoxicity. Also, when the B-PEG NSs were combined with DOX and NIR laser irradiation, over 95% of the cells died at a DOX concentration of 100 µg mL −1 . No obvious side effects were noted, also the levels of TNF-a, IL-6, IFN-g, and IL-12+P40 were similar to those in the PBS control group indicating that B-PEG NSs did not induce obvious cytokine response. Compared with the control group, there is no statistically significant difference of the NSs-treated groups with PBS-treated groups in all the parameters, no obvious induction on cytokine response, no change in biochemical parameter and no hematological toxicity, therefore, B-PEG NSs do not cause obvious infection and inflammation in the treated mice. Moreover, no noticeable signal of inflammation or tissue damage was observed in major organs. be due to differences in the GBMs studied (i.e. GO sheets versus GO nanoplatelets), and they raise the need for further investigation concerning the effects of the oxidative degree of GBMs on immune cells. In order to investigate the pristine graphene effects in vitro (THP-1 cell line) and in vivo (C57BL/6 strain mice), Schinwald et al. (118) have assessed the impacts of the shape of graphene nanoplatelets (GNPs) on their inflammatory potential. This large few-layer graphene presented as inflammogenic both in vitro and in vivo, which was attributed to its large size that led to frustrated phagocytosis. The authors highlighted that the potential hazard of GNPs could be minimized by producing GNPs small enough to be phagocytosed by macrophages. Moreover, the number of GO layers has been shown to affect its immunotoxicity, in which single-layer GO (SLGO) caused a more pronounced decrease in cell viability due to membrane damage of THP-1 cells, while multi-layer GO (MLGO) induced higher reactive oxygen species (ROS) and IL-1b production, leading to necrosis and apoptosis (120). In addition, the histological animal analysis revealed that SLGO and MLGO induced acute and chronic damage to the lungs and kidneys in the presence or absence of Pluronic F-127 (120).
Another important parameter, when approaching nanomaterial biosafety, is colloidal stability. Aggregation can influence the immunological response as observed by Wang et al. (127), when compared the toxicological profile of 2D MoS 2 versus aggregated MoS 2 in lung cells and mice. In their in vitro evaluation, in THP-1 and BEAS-2B cells, they found that aggregated MoS 2 induces strong proinflammatory and profibrogenic responses, while 2D MoS 2 have little or no effect. In agreement with in vitro results, an acute toxicity study in vivo showed that aggregated MoS 2 induced an acute lung inflammation, while 2D MoS 2 had no or a slight effect.
To increase the stability of 2D materials, studies have shown that proteins can be used as a dispersant agent. Lin et al. (142) studied silicene nanosheets modified with a bovine albumin serum protein corona (SNSs-BSA) and observed a significant increase in the colloidal stability in several physiological media (0.9% saline, phosphate buffered saline and Dulbecco's modified Eagle medium). Furthermore, SNSs-BSA did not cause significant toxicity in vitro neither significant acute toxicity in vivo. Only meaningless hematological changes were observed during the treatment duration, and no significant inflammation or infection were caused by the SNSs-BSA.
It is imperative that in a physiological environment, the nanomaterials will interact with biomolecules, forming a complex biomolecular corona. Those biomolecules (e.g., proteins, lipids, carbohydrates) can change the identity of the nanomaterials and influence their interaction with biological systems, causing an increase or decrease in internalization, toxicity, and biocompatibility as well as in colloidal stability over time. Thus, the biotransformation of nanomaterials in a physiological environment is an important parameter to be studied (146). The most common and highly studied component of biomolecular corona is the protein corona. In this sense, Mo et al. (132) studied the effect of the human plasma protein corona on the cytotoxicity of BP nanosheets and BP quantum dots (BPQDs) observing a reduction in cell viability for both nanomaterials when coated with proteins. However, protein corona facilitated BP nanosheet internalization and induced an increase in inflammatory cytokines (IL-1b, IL-6, IL-8 and IFN-g) and in ROS generation. Besides, it was observed that protein corona coated BP caused an induction on the nitric oxide (NO) and tumour necrosis factor. Further, Mo et al. (133) studied the effect of the human plasma protein corona in BP toxicity, and observed an increased macrophage polarization due to the adsorption of opsonins present in the plasma, increasing the uptake of BP and the interaction with stromal interaction molecule 2 (STIM2) protein facilitating Ca 2+ influx.
Similarly, Han et al. (126) studied the effect of plasma coronacoated 2D monoelemental nanosheets and observed that the protein corona decreases cytotoxicity and cell membrane damage for borophene, phosphorene, and graphene nanosheets. The corona coating induced the secretion of inflammatory cytokines (IL-1b, IL-6, IL-8, and IFN-g) for all three materials. Also, for BNNs, it was observed an increase in cellular uptake when the material was coronated, and therefore, the corona may promote phagocytosis. Baimanov et al. (31) also investigated the effect of four different blood protein coronas (human serum albumin (HSA), transferrin (Tf), fibrinogen (Fg), and immunoglobulin G (IgG) corona) on cell viability, uptake, and pro-inflammatory effects of MoS 2 nanosheets (NSs) in the macrophages cell line. Their results demonstrate that blood proteins contribute to uptake and inflammatory effects, as protein coated MoS 2 NSs increase cell viability and decrease cytoplasmic membrane damage when compared to non-coated MoS 2 NSs. Besides, it was observed that the type of protein influences cytokine secretion, as IgG-coated MoS 2 NSs causes more inflammatory cytokine secretion (TNF-a, IL-6 and IL-1b). The highest proportion of b-sheets on IgG led to fewer secondary structure changes on MoS 2 NSs, facilitating uptake and producing a stronger pro-inflammatory response in macrophages due to the recognition of an MoS 2 NSs−IgG complex by Fc gamma receptors and the subsequent activation of the NF-kB pathways. Another interesting finding is that in a serum-containing medium, cellular uptake of MoS 2 NSs−protein complexes was higher than that in a serum-free medium. Also, the MoS 2 NSs−Fg, and MoS 2 NSs-serum complexes had similar results in serum-free conditions and different results in a serumcontaining medium, suggesting the formation of the protein corona layer above the previously formed MoS 2 NSs−protein complexes. Those studies can help to elucidate the mechanisms in which protein corona can affects the toxicity of 2D materials.
One important ability of the immune system is the innate immune memory, where cells from the innate immune system react to secondary stimulus, which mostly includes an increased or decreased production of inflammation-related factors (147). With regard to 2D materials studies, there is yet a little research on this topic. Liu et al. (148) functionalize GO with lentinan (LNT) and observed that GO-LNT was able to promote macrophage activation by NF-kB and TLR signaling pathway, as well as enhance antigen protein processing after initial contact with macrophage. Moreover, the efficiency of this material was investigated, as a vaccine adjuvant for ovalbumin (OVA), in this sense GO-LNT induced robust long-term OVA-specific antibody responses due to the prolonged release of OVA. Besides this, GO-LNT was able to sustain a long-term immune response because it facilitated the uptake and slowed the release rate of antigen in macrophage. Further, Lebre et al. (149), demonstrated that pristine graphene can promote the innate immune training, enhancing the secretion of IL-6 and TNF-a and a decrease in IL-10 after toll-like receptor ligand stimulation 5 days after graphene exposure, indicating that pristine graphene can activate the immune innate memory.
Immune cells, such as macrophages and neutrophils, are one of the first line of defense of the immune system; they are capable of engulf the foreign material (or pathogen), degrading it and producing cytokines to enhanced the immune response (150). The uptake of 2D materials by immune system cells have been reported in various studies (31,109,115,126,132); however, there are few studies that address the degradation of those materials after internalization. Mukherjee et al. (151) studied the degradation of large and small GO by neutrophils and observed that not only both GO be degraded by neutrophils but also that the product of the degradation was non-toxic to human cells. Similarly, Moore et al. (152) studied the degradation of few-layer MoS 2 in human macrophage-like cells and observed that internalization occurred following 4 h of exposure and after 24 h the in vitro degradation of the material was confirmed, which occurred within lipidic vesicles and associated with enzymatic regions containing lysozyme.
As presented above, 2D nanomaterials may have an inflammogenic potential and immunotoxicity, which may impair their successful clinical translation; however, the immunological system activation can also be useful for theragnostic purposes. This application uses the immune responses to protect the body and eliminate cancer cells. The advantage of immunotherapy is that it engages the immune system to kill tumor cells without damaging healthy cells, additionally, it may induce immunological memory, causing long-lasting protection (153).

Nanoinformatics Approaches Toward Immunosafety-by-Design
In materials science, theory, computational modeling and informatics have a substantial role in accelerating and discovering new materials with interesting properties and applications (154)(155)(156). Due to the growing interest in 2D nanomaterials, computational approaches are extensively used in the discovery, development and application of these materials by detailed study of their structure/property relationships (156)(157)(158).
The nano-bio interface phenomena are directly related to the physicochemical properties of nanomaterials. However, tracing general correlations and delineating predictive models of nanomaterials biological effects remains challenging. Some issues include the complexity of nano-bio interactions, nanomaterials structural heterogeneity, lack of standard methodologies, absence of systematic studies and low-quality nanomaterial characterization (159)(160)(161). In this context, computational methods have been incorporated into the nanotoxicology field to support the understanding of the nano-bio interface to enable the development of safe-by-design principles applied to nanomaterials (162,163). Theoretical modeling (i.e., molecular dynamics, density functional theory) enables precise control of critical parameters of the nanomaterials surface to study their individual effects in nanobio interactions, providing mechanistic knowledge (164)(165)(166). On the other hand, machine learning (ML) techniques are used to assess datasets of nanomaterials biological outcomes in order to find patterns and correlations between physicochemical properties and biological effects, often undetectable through other types of analysis (167)(168)(169).
Applications of data-driven strategies include data filling, grouping, and predictive modeling. Quantitative nanostructureactivity relationships (QNAR) consist of the main strategy to delineate prediction models based on correlations between nanomaterial structural characteristics to their properties and biological activities (170,171). It is based on the assumption that nanomaterials in their properties present similar biological effects. Diverse algorithms can be used in QNAR models, including support vector machine (172), artificial neural network (173), and decision trees (174), among others, and depending of the level of algorithms interpretability may enable the outline of causal relationships.
The scarcity of quality data and comprehensive databases is the major bottleneck in the application of ML to predict nanomaterials immune reactions (175,176). Data-driven strategies have been making important advances in modeling biological phenomena that have potential usage to evaluate nano-immune interactions, such as predicting biomolecular corona compositions (177)(178)(179)(180)(181), and nanomaterials and cell interactions (e.g., cell uptake, cytotoxicity, membrane integrity, oxidative stress) (182)(183)(184)(185). Furthermore, the exploration of omics approaches (e.g., genomics, transcriptomics, and metabolomics) has promoting the development of ML models to process the complex data generated by these techniques and enables a better understanding of the molecular mechanisms of nanomaterials adverse effects in a systemic context, defining and predicting adverse outcome pathways (186)(187)(188)(189). The omics' potential of data generation is demonstrated by Kinaret et al. (190), who were able to connect immune responses to observed transcriptomic alterations in mouse airway exposed to 28 engineered nanomaterials. Together with cytological and histological analyses (imaging processing), they generated an extensive in vivo data set of nanomaterial adverse effects.
Allied with quality data infrastructure and processing, computational methods are sizeable to deal with complexity of nano-bio interface to assess and model the toxicity of nanomaterials in a variety of environments (163,(191)(192)(193)(194). To support safe-by-design approaches, international efforts have been made to provide data integration and sharing, modeling tools, standard protocols, and ontologies, to ensure Findable, Accessible, Interoperate, and Reusable (FAIR) data (195,196). For example, European projects, such as NanosolveIT and NanoCommons, and more recently CompSafeNano are initiatives facing on this direction (164,165,197,198). In accordance with these initiatives, Gazzi et al. (199) recently presented the nanoimmunity-by-design concept developed inside G-IMMUNOMICS and CARBO-IMmap projects, which aim to bridge the knowledge gaps in the immune characterization of carbon-based materials, integrating data-driven methodologies which are extendable to other 2D materials.

CONCLUSIONS AND FUTURE PERSPECTIVES
Two-dimensional materials are key elements for nanoscience and innovation in energy, health, and the environment. This can lead to a broad range of technological applications, especially nanoimaging, which has been growing exponentially in recent years. The wide number of 2D materials with different physicochemical properties make immunotoxicity and safety evaluation a challenge. There are therefore still gaps and controversial data in the literature. For example, within the same material category (i.e., graphene oxide) different properties were observed that might affect immunological and toxicological responses. It is imperative to evaluate the biological effects of biomolecular corona formation on 2D materials at nanobiointerfaces. Only by the identification of these material properties (intrinsic and extrinsic) and an integrated understanding on how they may influence its immunological response, we can manage immunotoxicity/biocompatibility and then benefit from their unique properties for many applications. Furthermore, it is very important to highlight the critical influence of endotoxin contamination prior immunological studies and toxicity testing. Special attention on this topic will avoid misinterpretation of immunosafety results involving 2D materials (148). In addition, it is important to advance in the understanding of the links between nanomaterials and the immune system across environmental species; this being a future challenge for immunosafety research associated with 2D materials (200). Nanoinformatics and computational modeling will have a decisive role on immunotoxicological studies with nanomaterials toward the practical implementation of immunosafety-by-design. However, it is very important to develop harmonized protocols, ontologies, and public databases to facilitate and promote a global research community for the collaboration and an exchange of knowledge in this field, focusing efforts on FAIR data principles.

AUTHOR CONTRIBUTIONS
All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication. GS and LFr: literature research, data curation, writing, and editing. RP, LFo, and MM: literature research and writing. DM, AF, and OA: funding acquisition, supervision, project administration, and writing. All authors contributed to the article and approved the submitted version.