Immunomodulation and T Helper TH1/TH2 Response Polarization by CeO2 and TiO2 Nanoparticles

Immunomodulation by nanoparticles, especially as related to the biochemical properties of these unique materials, has scarcely been explored. In an in vitro model of human immunity, we demonstrate two catalytic nanoparticles, TiO2 (oxidant) and CeO2 (antioxidant), have nearly opposite effects on human dendritic cells and T helper (TH) cells. For example, whereas TiO2 nanoparticles potentiated DC maturation that led towards TH1-biased responses, treatment with antioxidant CeO2 nanoparticles induced APCs to secrete the anti-inflammatory cytokine, IL-10, and induce a TH2-dominated T cell profile. In subsequent studies, we demonstrate these results are likely explained by the disparate capacities of the nanoparticles to modulate ROS, since TiO2, but not CeO2 NPs, induced inflammatory responses through an ROS/inflammasome/IL-1β pathway. This novel capacity of metallic NPs to regulate innate and adaptive immunity in profoundly different directions via their ability to modulate dendritic cell function has strong implications for human health since unintentional exposure to these materials is common in modern societies.


Introduction
Nanoparticles (NPs) have become a ubiquitous staple of modern life, yet researchers have a less than complete understanding of how these materials affect human health. In fact, it is becoming increasingly clear that NP species with distinct physiochemical properties (size, shape, composition, solubility, surface chemistry, etc.) can interact with body systems in a variety of different ways. For instance, CeO 2 NPs have shown great promise at protecting tissues from oxidative stress and have been proposed as a modality to alleviate healthy tissue damage during cancer radiation therapy [1][2][3]. On the other hand, metallic NPs have been shown to negatively impact human health by inducing acute toxicity in the lung and kidneys [4]. As well, metallic NPs have been found to induce the production of pro-inflammatory cytokine when delivered in vivo, which suggests these materials likely can engage cells of the immune system.
Despite multiple studies detailing the influence of size, solubility, and surface modification on the biocompatibility of metallic nanoparticles [5], far fewer reports have directly examined how the varied physical characteristics of these NPs affect their interaction with the human immune system. Published work from our laboratory and other groups has suggested the inflammatory potential of metallic NPs is inversely proportional to their sizes [6][7][8]. Other studies have shown the highly charged surface of some metallic NPs can facilitate their binding to proteins and other molecules, leading to macromolecular complex formation and/or altered protein conformations that can be highly immunogenic [9]. Some authors have also suggested redox-active surface groups can directly influence NP interactions with immune cells, but the impact of these studies is dampened because catalytic NPs were not directly compared against NPs with opposite redox activities [10][11][12][13][14]. Indeed, few studies to date have examined whether antioxidant NPs affect immune function.
Considering the various NP physiochemical properties that could be considered impactful on immune function, redox activity is perhaps the most important since catalytic NPs have a unique capacity to directly modulate reactive oxygen species (ROS). (ROS are well-established regulators of immune reactions [15].) To formally address whether catalytic activity affects NP-immune interactions, we performed a comprehensive examination of the immunomodulatory potential of two metallic NPs (TiO 2 and CeO 2 ) with opposing redox activities in an in vitro model of human immunity. This system, which encompasses a number of modular constructs that permit the evaluation of different facets of immunity, has been shown in a variety of published studies to support the generation of responses that reflect known human in vivo immune profiles against a series of biologic compounds and vaccines [8,[16][17][18][19][20][21][22][23][24][25][26].
Specifically for this study, we have examined the effect of these unique NP species on human immune cell viability, phenotype, uptake, ROS production, and function in the in vitro cell culture model. Intriguingly, we noted that the reductive CeO 2 NPs were uniquely capable of stimulating DCs to produce IL-10, and when co-cultured with T cells, triggered a strong T H 2-biased/regulatory cytokine profile. In contrast, oxidative TiO 2 NPs induced DCs to produce IL-12 and polarized T cells toward a T H 1-biased program. As a whole, these data provide evidence that NPs have the potential to modulate human DC and T helper cell function with a directionality that is linked to surface redox properties and suggest a novel basis for modulating immunity via NPs with tunable surface chemistries.

NP Characteristics
To determine whether surface catalytic activity can affect the interaction of NPs with the immune system, we performed a parallel evaluation of the capacity of TiO 2 and CeO 2 NPs, which have opposite catalytic activities, to stimulate immune cell activation in an in vitro model of the human immune system. The physical properties of the CeO 2 and TiO 2 NPs included in this study are discussed in detail in the Materials and Methods section and are summarized in Table 1. Since we were specifically interested in understanding whether catalytic activity impacts the interaction of metallic NPs with the immune system, we first needed to ensure other physical features of these NPs, such as agglomeration and purity, did not contribute to changes in immune function in the in vitro model. As shown in Figure 1, the CeO 2 and TiO 2 NPs had a tendency to form soft agglomerates of 10 and 25 nm in diameter, respectively, when cultured for 24-hr in X-VIVO-15 tissue culture media (serum-free culture media used in all of the biological assays discussed below). Additionally, we confirmed both NP preparations were free of contaminating LPS (EU ,0.05) that could otherwise compromise the outcome of the subsequent immunoassays (see Figure S1).

DC Cytotoxicity and Maturation Resulting from NP Treatment
We recognize NPs can potentially interact with a variety of immune cell populations, but focused our initial evaluation on DCs since they are involved in many facets of innate and adaptive immunity. We previously established a dose range for TiO 2 NPs in our in vitro immune cell model [8]; here, we started here by establish whether assay-derived APCs had a similar tolerance forCeO 2 NPs. Following a 24-hr treatment of the DCs with NPs, the cells were labeled with a fluorescent apoptotic dye (PO-PRO), in combination with a vital dye (7-AAD), to discriminate between live, dead, and apoptotic cells. Unlike TiO 2 NPs, which triggered appreciable apoptosis and death of the cultured DCs in a dosedependent manner, we observed no increase in apoptosis or death in DCs cultured with CeO 2 NPs (Figure 2A). It is important to note that several published articles have shown these NPs do not interfere with these standard fluorescent readouts [3,8,27,28]. To further mitigate the risk of NP interference with these assays, the nanoparticles were diluted in, or delivered, in cultures maintained in the presence of protein containing media and the cells were thoroughly washed in protein containing buffers prior to their acquisition by any instrument. While our findings on TiO 2 NP cytotoxicity in human DCs are consistent with our previous work and the observations of others using cell lines [8,[29][30][31], we are unaware of other studies demonstrating a high tolerance of human DCs for CeO 2 NPs. Metallic NPs have previously been shown to activate/mature DCs towards an enhanced functional state [8,32]. To determine whether this DC immunostimulatory potential was driven, at least in part, by the oxidative activity of TiO 2 NPs, we directly compared DC activation/maturation triggered by TiO 2 and the antioxidant CeO 2 particles. As shown in Figure 2B, DCs treated with as little as 1 mM TiO 2 NPs increased their expression of surface receptors involved in T cell priming/activation (HLA-DR, CD80 and CD86) and migration (CCR7). TiO 2 -treated DCs also upregulated surface CD83, a phenotypic hallmark of DC maturation, but only at the highest treatment dose (100 mM). Interestingly, a 24-hour exposure of the DCs to even the highest dose of CeO 2 NPs had almost no effect on CCR7, CD83, CD80, CD86, or HLA-DR expression levels.
Besides triggering changes in surface marker expression, maturation stimuli also often cause DCs to produce a variety of soluble and membrane-bound cytokines that modulate many facets of innate and adaptive immunity. Indeed, TiO 2 particles stimulated a strong cytokine response from the DCs that was of a pro-inflammatory slant ( Figure 2C, IL-12, TNFa) and consistent with the phenotype changes highlighted in Figure 2B. Considering the lack of DC surface marker changes triggered by CeO 2 ( Figure 2B), we were surprised to find these NPs induced the APCs to produce significant quantities of the immunoregulatory cytokine, IL-10. However, the inability of CeO 2 NPs to activate DCs may not be surprising in light of the observation that antioxidants, such as N-acetylcysteine, do not induce DC maturation, and to some extent, have even been shown to mitigate DC maturation [33,34]. Furthermore, some published studies have also shown chemical antioxidants, like phenyl N-tertbutyl nitrone, have the propensity to induce IL-10 production in cultured DCs [35,36].

Redox Potential as a Regulator of DC Activation State
Considering evidence suggesting oxidative stress can result in cytotoxicity and inflammation [37], we suspected the differential responses generated by TiO 2 and CeO 2 NPs might be explained by their opposite surface reactivity. To rule out the possibility that these distinct responses could be explained simply by the differential uptake of TiO 2 and CeO 2 NPs by DCs, we used a highly sensitive inductively coupled plasma-mass spectroscopy (ICP-MS) technique [38,39] to examine whether the NPs were localized within the treated DCs. With this technique, we were able to rule out uptake as an explanation for the results of Figure 2 since Figure 3 shows uptake is dose-dependent and detectable by ICP-MS at concentrations above 50 mM for both NPs species. While previous studies examined APC-mediated uptake of TiO 2 and CeO 2 at a much higher dose ranges than those used in the current study [40][41][42][43], it should be noted that we used a lower treatment dose range because we wanted to ensure that the immune cells remained viable for subsequent functional assessments.
As noted in the Introduction section, catalytic NPs have a unique capacity to directly modulate reactive oxygen species (ROS). Given our findings thus far, and the known redox activity these materials possess, we felt it necessary to examine ROS as a possible mechanism to explain the unique and disparate DC activation/ maturation profiles triggered by TiO 2 and CeO 2 NPs. Towards this goal, we analyzed intracellular oxidative stress levels in NPtreated DCs with a specific dye, DCF-DA, which fluoresces upon contact with ROS. Figure 4A   Although ROS can act through a variety of downstream pathways to regulate/potentiate immune reactions, perhaps its most important feature is its ability to activate innate danger sensors, such as the NLRP3 inflammasome [44]. Since the detection of IL-1b has been routinely used as a readout of NLRP3 inflammasome activation [44], we used this cytokine as an indirect measure of whether TiO 2 and/or CeO 2 NPs activate the NLRP3 inflammasome in human DCs. Based on past studies demonstrating TiO 2 NPs activate the NLRP3 inflammasome in mice [44], we were not surprised to find DCs treated with these NPs were stimulated to secrete heightened quantities of IL-1b. In subsequent studies, we showed the selective NLRP3 inhibitor, glybenclamide (50 mM), abolished IL-1b production in these cultures. This provides further evidence that TiO 2 NPs act through the NLRP3 inflammasome to induce IL-1b production ( Figure 4C). In stark contrast to these results, we found CeO 2 NPs triggered no IL-1b production by the cultured DCs ( Figure 4C), which further supports our earlier conclusions that these anti-oxidant NPs induce a null or anti-inflammatory response in DCs.

NPs Drive CD4 + T Cell Proliferation and T H 1/T H 2 Polarization
Following our finding that CeO 2 and TiO 2 provide human DCs with distinct stimulatory/maturation cues, we questioned whether these differences would, in turn, translate into unique patterns of T cell responses resulting from stimulation with the NP-treated DCs. Prior to addressing this issue, we first investigated whether NPs directly activate lymphocytes in a 5-day stimulation assay where T cell proliferation serves as the primary readout of the response. To our surprise, TiO 2 had a modest immunostimulatory effect on the T cells, as demonstrated by their capacity to induce an increase in the divided (CFSE-low) lymphocyte population over the untreated control. Furthermore, the co-administration of TiO 2 NPs with the mitogens, PHA and PMA, synergistically increased the proliferative response ( Figure 5). CeO 2 NPs alone did not induce measurable T cell proliferation but, interestingly, did reduce the proliferative response when added with the mitogen cocktail ( Figure 5). Of note, neither of the particle types affected the viability of the T cells over a broad dose range (see Figure S2). As an additional measure to investigate the stimulatory effect these NPs have on T cells, we examined the expression levels of CD95 (FasR), which becomes upregulated under stress or disease conditions and is part of the programmed death response [45]. The expression of CD95 was unaffected by either NP treatment (see Figure S3). However, treatment with TiO 2 NP in addition to the mitogen cocktail, PHA/PMA, revealed the capacity for TiO 2 NPs to drive a much stronger level of CD95 expression as compared to CeO 2 NP and mitogen treated T H cells ( Figure S3). While this evidence doesn't tell us precisely how these NPs are interacting with T cells, the NPs are affecting T cell phenotype and function as measured by these assays.
To better define the impact of catalytic NPs on human adaptive immunity, we directly examined the capacity of NP-treated DCs to stimulate naïve T cell responses using our in vitro model of human immunity. With this approach, the engagement of TCR by foreign HLA class II molecules on the surface of mismatched DCs is sufficient to induce the activation of the lymphocytes in an antigen-independent fashion. Here, DCs were left untouched (iDC), matured with a maturation cocktail (mDC, positive control), or primed with CeO 2 or TiO 2 NPs before being co-cultured with allogeneic CD4 + T cells. After 5 days, the cells and culture supernatants were harvested for evaluation by flow cytometry (cell viability and proliferation) and Bio-Plex assay (cytokine production).
Although the CeO 2 NP-treated DCs had little influence on allogeneic naïve CD4 + T cell proliferation, TiO 2 NP-treated DCs boosted the magnitude of the proliferative response ( Figure 6). As well, we observed that both particles triggered cytokine responses, but the profiles were nearly opposite: TiO 2 NPs-pulsed DCs triggered a pro-inflammatory T H 1-biased cytokine response (IL-2, IFN-c) while DCs pulsed with CeO 2 NPs induced a naïve T cell response dominated by T H 2 cytokines (IL-4, IL-5, and IL-10) that are predominately anti-inflammatory and promote humoralskewed responses. Beyond their capacity to participate in the induction of a T H 2-biased T cell response, the CeO 2 NPs were even capable of eliciting the production of IL-4, IL-5 and IL-10 in T cell co-cultures stimulated with a strongly Th1-biasing mitogen ( Figure 7). While we might have anticipated that a well-described pro-inflammatory particle like TiO 2 could drive a type 1 immune response, the response profile induced by CeO 2 NPs, including IL-10 secretion by DCs ( Figure 2) and T H 2 polarization (Figure 6 and  7),suggest a unique functional property of metallic antioxidant NPs that, to our knowledge, has not previously been described.

Discussion
Despite the emergence and rapid adoption of NPs into modern life, a paucity of data exists on how these materials influence human physiology, including the immune system. In an earlier publication from our laboratory, we showed particle size had a profound impact on the ability of TiO 2 NPs to induce inflammation in an in vitro model of human immunity [8]. In the current study, we questioned whether other physiochemical features of NPs, specifically surface reactivity, might also influence the immunomodulatory potential of NPs. Towards this goal, we employed the same in vitro model employed above to examine whether oxidative TiO 2 and anti-oxidative/reductive CeO 2 NPs have altered capacities to influence human immune reactions.
In a series of experiments aimed at examining the impact of these NPs on innate responses, we demonstrated that TiO 2 NPs push human DCs towards a more activated/pro-inflammatory state while CeO 2 NPs triggered a more anti-inflammatory profile in these cells. Given these observations, we were not surprised to see the NP-treated APCs, in turn, triggered nearly opposite T helper cell response profiles (CeO 2 promoted a T H 2 profile while TiO 2 lead to a T H 1 pattern). Our current results with TiO 2 NPs were consistent with our published work and reports by others showing these NPs can induce oxidative stress and inflammation [8,46,47]. On the contrary, we did not anticipate CeO 2 NPs would induce such a pronounced T H 2-biased (IL-4, IL-5 and IL-10) T cell response and even blunt mitogen-induced T H 1 (IL-2, IFNc, and TNFa) cytokine production. Though the overall profile of cytokines produced by T H cells stimulated with CeO 2 -stimulated DCs is consistent with a T H 2 profile, it should also be noted that the strong IL-10 response might be reflective of T reg induction in these cultures. Additionally, we examined additional cytokines/ chemokines which were not activated in response to treatment (See Methods section). Our preliminary results did not indicate CeO 2 -stimulated cultures yield a higher frequency of T reg cells, but further experimentation will be necessary to fully investigate this possibility.
To date, few studies have detailed the capacity of NPs to polarize T H cell response and none have shown the pronounced NP-induced T H biasing demonstrated here. For example, Liu et al. showed poly-hydroxylated metallofullerenol NPs could induce T H cytokine responses, but only in a mixed fashion (both T H 1 and T H 2 cytokines were produced). In a second example, PLGA-NPs were shown to push T H cells towards a specific cytokine profile, but only in cases where the NPs were conjugated to known T H biasing peptides [48,49]. This unique and pronounced T H response polarization resulting from metal-oxide (TiO 2 and CeO 2 ) NP treatment could possibly be explained by the differences in the capacities of the two NP species to regulate ROS production, particularly since ROS can function as a second messenger and modulator of immunity [15,[50][51][52]. Going a step further, it is interesting to speculate the anti-oxidant redox activity of CeO2 NPs triggers significant IL-10 production by the DCs that ultimately leads to the strong IL-10 response by activated T cells in the DC/T cell cultures since this cytokine is a well-established regulator of T H cell differentiation [53][54][55]. This hypothesis is consistent with prior studies showing ROS-generating materials, like TiO 2 NPs, trigger downstream pro-inflammatory effects and antioxidants prevent the initiation of the innate immunity in LPSstimulated macrophages, as evidenced by the suppression of proinflammatory cytokine (TNF-a, IL-1b) secretion by the treated cells [56,57]. As well, it is supported by another study showing palladium NPs, a reducing agent with anti-oxidative properties similar to CeO 2 NPs, can trigger IL-10 production by human peripheral blood mononuclear cell (PBMCs) [58]. It should be noted that our observations suggesting NPs modulate immune function through an ROS pathway does not preclude the possibility that the particles act on APCs via other mechanisms.
While no current in vitro culture model can replicate all the intricacies and variables of the in vivo environment, we have assessed these materials to the best of our capacities using our in vitro model which has provided meaningful pre-clinical information on human immune responses shown to be reflective of human responses in prior publications [59,60]. However, since these materials are unable to be tested in a clinical setting, we are in the process of validating our in vitro results through the use of a murine model. Unfortunately, such evaluations are very complex and require a great deal of consideration across a number of experimental parameters including dosing schema, diluent, route of administration, number of treatments, kinetics, (disease) model, and possible readouts.
We speculate the distinct immunostimulatory potentials observed between CeO 2 and TiO 2 are likely explained by the distinct manner in which these materials are able to absorb photons. Here, the materials differ in that the photons have a tendency to migrate to the surface of TiO 2 NPs, where they are free to react with oxygen, water, or hydroxyls to form free radicals [61]. On the other hand, the CeO 2 NPs absorb these free photons where they remain isolated from the outside environment [61]. In fact, this chemistry leads to their distinct oxidant/antioxidant properties, as illustrated in Figure 4A and 4B, where ROS production by DCs increased linearly with TiO 2 NP dose, but remains absent in CeO 2 NPs-treated cultures. Moreover, CeO 2 actually inhibited ROS production induced by H 2 O 2 in a dose-dependent manner, which suggests this NP species is a very potent anti-oxidant.
Taken as a whole, the results of this study suggest differences in surface reactivity can profoundly affect how metallic NPs interact with the human immune system (Table 2). Specifically, these data suggest low-dose exposure of human immune cells to redox-active NPs have the propensity to modulate human innate and adaptive immunity, i.e, DC activation and primary CD4 T helper cell differentiation state. For this reason, CeO 2 NPs (and perhaps other anti-oxidant moieties) might offer researchers a unique opportunity to push adaptive responses in a focused direction away from a T H 1 bias and towards a T H 2/T reg bias. Alternatively, TiO 2 might serve as a potent Th1-promoting treatment during prophylaxis or disease treatment. On the contrary, the immunomodulatory potential of NPs could pose a considerable health risk if encountered in an uncontrolled environment. Specifically, the T H -skewing potential of NPs could possibly translate into effects on general inflammatory diseases, airway hyperresponsiveness, asthma, and autoimmunity. With further study, features like catalytic behavior may potentially be exploited for engineered NPs to meet a particular goal, such as enhancing immune responses during vaccination or mediating immune tolerance against allergies or autoimmune disease.

Subjects
This study included PBMC blood product from 10 healthy donors. This study was approved by the ethics committee of the Chesapeake Research IRB. Full documentation of application process, orientation attendance, and signed written informed consent forms were obtained from all donors prior to their participation and the study procedures were conducted in accordance with the Declaration of Helsinki (protocol CRRI 0906009). All applicants have met guidelines set forth in the approved IRB protocol, which includes (but is not limited to) restrictions regarding general health, disease screening, weight, and age. Blood collections were performed at Florida's Blood Centers (Orlando, FL), a state/federally regulated blood collection center, using standard techniques approved by their institutional review board. The PBMCs collected under our donor program are collected, stored, and later used for various immunological research projects at Sanofi Pasteur VaxDesign Campus. The donors' PBMCs used in this study were randomly selected from our cryo-bank.

Synthesis of NPs
TiO 2 NPs were synthesized by wet chemical synthesis as previously described [8]. Briefly, a 50:50 mixture of ultrapure ethanol (Sigma) and deionized water (18.2 M) was boiled to reflux. The pH of the boiling solution was adjusted to 3.0 with the addition of 1 N HCl. Titanium isopropoxide (Sigma) was added slowly to the refluxing mixture, which precipitates immediately to a white solution. The solution was then stirred at 85uC for 4 hours. The white solution was then cooled to room temperature and washed several times with ethanol until dry. The final preparation was mostly anatase (partially amorphous) TiO 2 . CeO 2 NPs were synthesized using wet-chemical synthesis as described previously [63]. Briefly, cerium nitrate hexahydrate was dissolved in deionized water (18.2 MV). A stoichiometric amount of hydrogen peroxide was added as an oxidizer and immediately resulted in the formation of cerium oxide NPs. The NP powder was obtained by washing the precipitate of CeO 2 NPs several times with acetone and water to remove the surfactant used in the synthesis process. The solution was aged further to allow the slow reduction of surface cerium from 4 + oxidation state to 3 + oxidation state in acidic medium by maintaining the pH of the suspension below 3.5 with nitric acid. Nanoparticle treatments investigated in this study are reported in molarity and the mass per volume is indicated in parenthesis as follows:   Characterization TiO 2 and CeO 2 NPs were analyzed using high-resolution transmission electron microscopy (HRTEM; Philips 300 TECNAI operated at 300 kV) to confirm their shape, size, and morphology. The HRTEM samples were prepared by dipping a polycarbon-coated copper grid into a dilute suspension of NPs dispersed in acetone. The surface area of the NPs were measured based on physical adsorption of ultra-high purity nitrogen gas at liquid nitrogen temperature using a Brunauer-Emmett-Teller (BET) Nova 4200e instrument manufactured by Quantachrome (Boynton  Table S1 and Table S2   Beach, FL). The samples were prepared in quartz tubes and degassed at 240uC in vacuum for 3 hours before actual measurement. The size of the NPs was determined by the dynamic light scattering method using the Zetasizer Nano manufactured by Malvern Instruments (Worcestershire, UK). The physical characterization of the materials is reviewed in Figure 1 and summarized in Table 1.

Evaluation of Endotoxin Contamination
All NP preparations were confirmed negative for the presence of endotoxin contamination using the FDA-approved Endosafe LAL colorimetric and turbidimetric assay system (Charles River Laboratories, Wilmington, MA). This data is shown in Figure S1.

PBMC Isolations
Within hours following their harvest from the donor, the enriched leukocytes were centrifuged over a Ficoll-plaque PLUS (GE Healthcare, Piscataway, NJ) density gradient [24,25]. PBMCs at the interface were collected, washed, and cryopreserved in IMDM media (Lonza, Walkersville, MD) containing autologous serum and DMSO (Sigma-Aldrich, St. Louis, MO).

Generation of Cytokine-Derived DCs
DCs used throughout the assays of this study were prepared using our previously published methodology [25]. Briefly, monocytes were purified from total PBMCs by positive magnetic bead selection (Miltenyi Biotec, Cologne, Germany) and cultured for 7 days in X-VIVO 15 (Lonza) serum-free media supplemented with GM-CSF (R&D Systems, Minneapolis, MN) and IL-4 (R & D Systems). In all assay conditions described below, treatments were delivered on day 6 followed by harvesting on day 7 for incorporation into the various assays.

ROS Determination
DCs were treated with serial dilutions of TiO 2 NPs and CeO 2 NPs for 24 h. Subsequently, the cultures were washed and treated at room temperature for 30 min with DCF at a final concentration of 10 mM. The cells were washed of excess dye with DPBS, harvested using cell-dissociation solution (Sigma), and washed again in DPBS. Fluorescence in the FITC channel from absorbed and oxidized DCF (indicative of peroxide levels) was analyzed by flow cytometry using an LSR II (Becton Dickinson). FlowJo software (Treestar, Ashland, OR) was used for data analysis.

DC Phenotype/cytokine Analysis
For flow cytometry analysis of surface molecule expression, DCs were washed in fluorescence-activated cell sorting buffer (FACS Buffer). Fc receptors were blocked with mouse serum (Jackson ImmunoResearch, West Grove, PA) to prevent nonspecific binding. DCs were then stained with a vital dye (LIVE/DEADH; Invitrogen). (Conversely, for determination of apoptosis DCs were stained with Po-Pro/7-AAD (Invitrogen) at the end of the surface antibody staining.) After washing away excess viability dye with PBS, the cells were then incubated with the appropriate antibody cocktail. The antibodies used in the staining panels include HLA-DR, CD14, CD40, CD80, CD83, CD86, CD19, CD3, CD209, and CCR7. All antibodies we purchased from eBioscience (San Diego, CA) with the exception of CD209 (BD Pharmigen, San Diego, CA). Following staining, cells were washed in FACS buffer and immediately acquired on a BD LSRII flow cytometer (Becton Dickinson), and data analyzed using FlowJo software V9.2 (Tree Star).

NP Uptake by DCs
Samples treated for 24-hrs with TiO 2 or CeO 2 NPs were harvested, washed and placed in 70% nitric acid overnight and then microwaved to digest the cellular material. The temperature of the cell harvest was steadily increased to 200uC over a 20-mins period and held constant at 200uC for an additional 20 minutes. The samples were then boiled down to less than 1 ml and reconstituted in water to an exact volume of 10 ml. Titanium and cerium levels were assessed using inductively coupled plasma mass spectroscopy (ICP-MS) using published techniques that have been optimized to minimize the possibility of surface-bound or aggregated NPs from being carried over from the washing steps [39].

CD4 + T Cell Proliferation Assay
Human CD4 + T cells were isolated from PBMCs by positive selection using EasySEP CD4 + T cell isolation kit II (Stem Cell Technologies, Vancouver, Canada). The purified CD4 + T cells were then carboxyfluorescein succinimidyl ester (CFSE)-labeled to follow proliferation and incubated either in the presence of the described NPs with or without PHA/PMA or without stimulation and left in culture for 5 days. The cells were harvested, and examined by flow cytometry using LIVE/DEAD AQUA and CFSE (Invitrogen) and antibodies specific for CD4 and CD25 (eBioscience) by flow cytometry.

In vitro Model of Human Immunity
Dendritic cells were generated using a 3-dimensional tissue engineered construct described previously [8,24]. These DCs were either untouched, matured with a cocktail of TNFa and PGE 2 as described previously as a positive control [24], or were exposed to various doses of NPs for 24 hours prior to being harvested. The treated DCs were harvested and added at an optimized ratio of 1:400 to allogeneic naïve CD4 + T cells isolated using EasySEP CD4 + T cell isolation kit II (Stem Cell Technologies) and labeled with CFSE (Invitrogen). After five days the cultures were harvested and stained for CD25, CD3, CD4, (eBioscience) and Live/Dead Aqua for viability (Invitrogen) and then acquired by flow cytometry using BD Pharmingen's LSR II as described above. Additionally, supernatant's were collected and examined for cytokine secretion by Bio-Plex array as previously described above. Here, PHA/PMA (1 mg/mL; 50 ng/mL) was used not only as a positive control for cytokine production, but also added in combination with NP-treated DC co-cultures where described.

Flow Cytometry, Data Plotting and Statistical Analysis
Cytometry data was analyzed using FlowJo software V9.2 (Tree Star). Each experiment was repeated with at least three donors or more, as described in the figure legends and plotted as an average (with S.D.) or displaying each data point. Analyzed statistical results were determined using a paired Student's t-test. Statistical significance was considered at p,0.05 or otherwise stated in figure legend. Tukey's honest significance test was employed, in conjunction with an ANOVA, to determine if the treatment groups (between CeO 2 and TiO 2 ) are significantly different from each other. All graphs and biostatistics were produced using GraphPad Prism software V5 (La Jolla, CA). Figure S1 Endotoxin levels of CeO 2 and TiO 2 NP measured ,0.05 EU/mL. The TiO 2 and CeO 2 NPs were diluted to 100 mM concentrations in sterile endotoxin-free water. The diluted preparations were then examined for endotoxin levels using an automated FDA-licensed endotoxin detection system by Charles Rivers Laboratories. No detectable (ND) levels of endotoxin were observed in the NP preparations. Three independent samples were run to generate average bar with S.D. Table S1 Statistical analysis of Figure 6 A. Tukey's honest significance test was employed, in conjunction with an ANOVA, to determine if the treatment groups (between CeO 2 and TiO 2 ) are significantly different from each other in relation to CFSE fluorescence.

(DOCX)
Table S2 Statistical analysis of Figure 6 B. As in Table S1, Tukey's honest significance test was employed, in conjunction with an ANOVA, to determine if the treatment groups (between CeO 2 and TiO 2 ) are significantly different from each other in relation to CD25 expression. (DOCX)