Immunomodulatory effects of nanoparticles on dendritic cells in a model of allergic contact dermatitis: importance of PD-L2 expression

Nanoparticle (NP) skin exposure is linked to an increased prevalence of allergic contact dermatitis. In our prior studies using the mouse contact hypersensitivity (CHS) model, we reported that silica 20 nm (SiO2) NPs suppressed the allergic response and titanium dioxide NPs doped with manganese (mTiO2) exacerbated it. In this work, we conducted in vitro experiments using bone marrow-derived dendritic cells (BMDCs) to study the combinatorial effect of the potent 2,4-dinitrofluorobenzene (DNFB) hapten sensitizer with SiO2 and mTiO2 NPs on BMDC cytotoxicity, cytokine secretion and phenotype using the B7 family ligands. Results show that DNFB and mTiO2 behave similarly and exhibit proinflammatory characteristics while SiO2 promotes a naive phenotype. We observe that the B7-H3 (CD276) ligand is only expressed on CD80 + (B7-1) BMDCs. Results from adoptive transfer CHS studies, combined with BMDC phenotype analysis, point to the importance of PD-L2 expression in modulating the adaptive immune response. This work identifies metrics that can be used to predict the effects of NPs on contact allergy and to guide efforts to engineer cell-based therapies to induce hapten specific immune tolerance.


Effects of DNFB, SiO 2 and mTiO 2 on BMDC toxicity, cytokine secretion and co-stimulatory molecule expression
We performed cytotoxicity studies to establish concentration ranges to examine the effects of NPs and DNFB on BMDC phenotypes.We conducted single exposure studies to each stressor for 1 h to measure cytotoxicity as a function of concentration using the flow cytometry.Results (Fig. 1A) show a significant dose-dependent decrease in cell viability for DNFB and mTiO 2 but not for SiO 2 .Exposing cells for 1 h to DNFB (0.1 mM) or mTiO 2 (0.05 mg/mL) caused a ~ 50% decrease in cell viability whereas exposing cells to SiO 2 was not cytotoxic.The effects of DNFB exposure correlated with notable changes in the BMDC morphology viewed in bright-field and in TEM images, with cells becoming round and losing dendrites (Figs.S1-S2).NP uptake in endosomal vesicles was also confirmed (Figs.S3-S4).Non-cytotoxic concentrations of each stressor were selected and the BMDCs were exposed for 5 h, 15 h, and 24 h (Fig. 1B).Results again show a significant decrease in cell viability over time for DNFB (0.001 mM) and mTiO 2 (0.005 mg/mL) but not for SiO 2 (0.01 mg/mL).The dose and time dependent cytotoxicity flow cytometry results were consistent with results using the PrestoBlue assay (Fig. S5).
Cytotoxicity studies suggest that DNFB and mTiO 2 behave similarly and are likely potent proinflammatory stressors compared to SiO 2 .To further characterize interactions with BMDCs, we evaluated the key proinflammatory (IL-6, TNFα) and immunosuppressive (IL-10) cytokines secreted in the supernatant by ELISA.BMDCs were exposed to DNFB (0.001 mM), SiO 2 (0.01 mg/mL) and TiO 2 (0.005 mg/mL) as a function of time.Results showed that each compound increased IL-6 (Fig. 2A) and TNFα (Fig. 2B) secretion and exposure to mTiO 2 upregulated IL-10 (Fig. 2C).One-hour exposures as a function of concentration also showed that mTiO 2  TNFα, and IL-10 (Fig. S6).These results suggest that mTiO 2 is a potent proinflammatory stressor, behaving similarly to lipopolysaccharide (LPS) (50 ng/mL), that also upregulated IL-6, TNFα and IL-10 (Fig. S7).These observations were corroborated by intracellular flow cytometry staining for IL-10 and TNFα as a function of concentration and time (Figs.S8-S9) but in contrast to mTiO 2 , we observe that SiO 2 tended to upregulate IL-10 without upregulating TNFα which, may suggest an immunosuppressive potential for SiO 2 exposure.Next, we examined the effect of each stressor on the phenotype of the BMDCs by measuring the expression levels of the co-stimulatory molecules by flow cytometry.Single and double gating strategies, with FMO controls, were used (Fig. S10A-C).Single gating assesses the expression levels of each marker on the live cell population.Double gating analyzes the CD11c + MHCII+ subpopulation.To interpret results, we compared them to LPS exposure (50 ng/mL) over time which, for single gating (Fig. S11) and double gating (Fig. S12), induced the expression of CD86, CD80, and PD-L1.CD276 increased early (3-6 h) and then returned to baseline and no changes in PD-L2 expression were induced by LPS.For single gating, exposing BMDCs to DNFB (0.001 mM), SiO 2 (0.01 mg/mL), and mTiO 2 (0.005 mg/mL) caused differential changes in the CD11c + MHCII + phenotype as a function of time (Fig. 3) with DNFB producing effects most similar to LPS exposure.DNFB and mTiO 2  PD-L2 was expressed on ~ 50% of imDC and exposure to each stressor did not alter its expression.Exposure to SiO 2 did not alter any co-stimulatory molecule suggesting BMDC maintained a naïve phenotype.Taken together, the results suggest that DNFB and mTiO 2 induce an activated BMDC phenotype and that mTiO 2 may be a more potent stressor as it failed to upregulate CD80, which is important for binding CTLA-4 to promote regulation.The differential effects of each stressor on the expression of CD80 and CD86 led us to investigating the expression of PD-L1, PD-L2, and CD276 within the CD86/CD80 subpopulations over time (Fig. 4).Results showed that exposure to DNFB (0.001 mM) decreased the naïve CD80-CD86-double negative (DN) population and induced a steady significant rise in the activated CD80 + CD86+ double positive (DP) population which are trends expected for a proinflammatory stressor.Similarly, mTiO 2 (0.005 mg/mL) showed a trend toward upregulating the DP population, whereas SiO 2 (0.01 mg/mL) exposure over time did not.Both NPs tended to downregulate slightly the DN subpopulation up to 15 h but not as definitively as DNFB.This subtle decrease in the DN is due primarily to the upregulation of CD80+ single-positive cells (Fig. S13).The expression levels of CD276, PD-L1 and PD-L2 in the CD80/CD86 subpopulations also differed depending on the stressor, with DNFB and mTiO 2 exhibiting similar trends.Specifically, DNFB and mTiO 2 exposure up-regulated PD-L1 in the DN (Fig. 4) and the single positive subpopulations (Fig. S13) with no change in the DP subpopulations.In contrast, SiO 2 exposure down-regulated PD-L1 expression in the DP subpopulation over time with no change in DN subpopulation.No changes in PD-L2 expression in DP or DN subpopulations were induced by either stressor (Fig. 4) but decreases in PD-L2 expression in the CD86 and CD80 single positive cells were induced by SiO 2 only (Fig. S13).CD276 was only expressed CD80+ single (Fig. S13) and double positive cells (Fig. 4).Both NPs tended to increase CD276 on the activated DP cells at 24 h.In summary, this data shows that DNFB and mTiO 2 exposure promotes the upregulation of the activated DP phenotype, and increased PD-L1 expression in the naïve DN population over time.PD-L1 expression of an activated DP population was unchanged over time with exposure to DNFB and mTiO 2 , whereas SiO 2 exposure tended to decrease PD-L1 expression pointing again to the similarities between DNFB and mTiO 2 .

Effects of DNFB co-exposure with NPs on cytotoxicity, cytokine secretion and and co-stimulatory molecule expression
In prior studies, NPs co-exposed with DNFB modulated the in vivo CHS response in the challenge phase 21 .In this work, we were able to observe clear differences between SiO 2 and mTiO 2 in single exposure studies of cytotoxicity, cytokine production, and on phenotypic alterations of BMDC, with mTiO 2 behaving more similarly to DNFB.Here, we studied the effects of DNFB co-exposure with NPs on cytotoxicity and cytokine secretion compared to DNFB alone at 1 h (Fig. S14 and 24 h (Fig. 5).Consistent with earlier studies (Fig. 1), BMDC cultured with DNFB (0.001 mM) alone for 24 h was cytotoxic.Co-culture with SiO 2 (0.01 mg/mL) showed a protective effect (Fig. 5A).In contrast, co-culture with mTiO 2 (0.005 mg/mL) significantly exacerbated the toxic response (Fig. 5B).A cytoprotective effect of SiO 2 was also observed at 1 h exposure (Fig. S14) and is consistent with prior studies in keratinocytes 23 and fibroblasts 61 .We also tested the supernatant of each treatment group for IL-6, TNFα, and IL-10 by ELISA.Results show a downregulation of IL-6 with DNFB co-cultured with SiO 2 but not with mTiO 2 (Fig. 5C,D).Co-culture with either NP did not alter the levels of TNFα produced by DNFB exposure at 24 h (Fig. 5E,F) however, co-culture for 1 h with mTiO 2 , but not SiO 2 NPs, showed elevated IL-6 levels (Fig. S14D) and both NPs elevated TNFα above that produced by DNFB at 1 h (Fig. S14E,F).Co-culture with mTiO 2 , but not SiO 2 at 24 h (Fig. 5G,H), increased the secretion of IL-10, which is consistent with the mTiO 2 single exposure studies (Fig. 2).These co-culture studies demonstrate that SiO 2 exhibits a cytoprotective against effect DNFB exposure and exhibits an immunosuppressive effect as measured by a reduction in IL-6 secretion.
Next, the effects of DNFB co-exposure with NPs on the expression of the costimulatory molecules was investigated using flow cytometry (Fig. 6).Double gating on the CD11c+ MHCII+ subpopulation showed that co-exposing BMDC to DNFB (0.001 mM) with either SiO 2 (0.01 mg/mL) or mTiO 2 (0.005 mg/mL) similarly down-regulated the DNFB activation of CD86, CD80, CD276, and PD-L1.Co-exposure with mTiO 2 tended to downregulate PD-L2 expression to levels below the imDC levels.
Differentiating the effects of DNFB co-exposure with NPs on the expression levels of PD-L1, PD-L2 and CD276 in the BMDC CD80/CD86 subpopulations was less clear (Fig. 7).Consistent with single exposure studies at 24 h (Fig. 4), DNFB activates the BMDCs as evidenced by an increase in the C86+ CD80+ DP subpopulation and a decrease the naïve C86−CD80− DN subpopulation.Both NPs suppressed BMDC activation, maintaining the DP and DN subpopulations to imDC levels at 24 h.There were no statistically significant changes in CD276 expression.However, both NPs co-cultured with DNFB increased PD-L1 expression on the DP subpopulation at 24 h, which was also evident at 1 h for mTiO 2 , but not for SiO 2 , which exhibited a statistically significant decrease PD-L1 at 1 h relative to the imDC (Fig. S15).The most intriguing differential effect between the NPs co-cultured with DNFB was the observation that mTiO 2 induced a significant decrease in PD-L2 expression in the DP subpopulation at 24 h and SiO 2 did not (Fig. 7).This decrease was also evident after only 1 h of co-culture (Fig. S15).A second, potentially important difference between the NPs was observed at 1 h exposure in PD-L1 expression on the DP subpopulation where SiO 2 decreased expression whereas mTiO 2 upregulated it over DNFB levels (Fig. S15).However, at 24 h both NPs increased PD-L1 on the activated DP cells and decreased PD-L2 expression on the naive DN cells.Statistically significant changes in PD-L2 expression were not observed in the activated DP or the naive DN subpopulations in single-exposure studies (Fig. 4).The significant differences in PD-L1 and PD-L2 expression in the activated DP subpopulation caused by mTiO 2 co-cultured with DNFB and the opposing effects with SiO 2 co-culture maybe influential in the BMDC controlling the fate of the adaptive immune response.www.nature.com/scientificreports/

Effects of NPs on the sensitization and challenge phases using the in vivo DNFB contact hypersensitivity model
Analysis of cytotoxicity, cytokine secretion and co-stimulatory markers taken together suggest that DNFB + SiO 2 co-exposure tends to promote a more naïve or regulatory BMDC phenotype, whereas DNFB + mTiO 2 promoted a proinflammatory BMDC phenotype.To test this, we used the in vivo CHS mouse model with adoptive transfer.We first tested the difference between topical and subcutaneous (S.C.) DNFB sensitization.BMDCs were treated with DNFB (0.01 mM) for 1 h.After two wash steps, the cells were resuspended in sterile saline and injected S.C. to sensitize the mice.A second group of mice was topically sensitized by applying 20 µL of 0.05% DNFB in acetone:olive oil vehicle (4:1) 7,21,60 .Five days later, ear thickness was measured, and the mice were challenged with 0.2% DNFB on one ear and with vehicle on the other.On day 6, ear thicknesses were remeasured.Results of ear swelling response showed no difference between the topical sensitization and S.C. sensitization (Fig. 8A).
In prior work we observed an immunomodulatory effect of NPs in the challenge phase with DNFB topical sensitization 21 .To test if these effects are similarly observed with S.C. sensitization, we S.C sensitized the mice with BMDC treated with 0.01 mM DNFB for 1 h.We challenged the mice on Day 5 with 0.2% DNFB alone, or mixed with NPs and measured the ear thickness on Day 6. Results show a similar result to our previous data 21 indicating that SiO 2 NPs suppressed and mTiO 2 NPs exacerbated the allergic response relative to challenge with DNFB alone (Fig. 8B).Next, we compared S.C. sensitization using BMDC treated with DNFB (0.01 mM) alone or BMDC co-cultured with SiO 2 (0.01 mg/mL) or mTiO 2 (0.005 mg/mL) for 1 h.After two wash steps, we S.C. injected the cells to sensitize the mice.Upon challenge, we treated one ear with 0.2% DNFB and the other with vehicle.Results show that mice sensitized with BMDC treated with DNFB + SiO 2 measured a decreased ear swelling relative to DNFB alone.In contrast, mice S.C. sensitized with BMDC treated with DNFB + mTiO 2 showed an increase in ear swelling relative to DNFB alone (Fig. 8C).These results suggest that BMDCs co-cultured with NPs and a potent sensitizer can alter the BMDC phenotype to effect the efficiency of T cell priming and the intensity of the CHS response.In prior work we did not measure an effect of NPs co-exposed topically with DNFB in the sensitization phase 21 .This is most likely due to the inability of the NPs to breach the skin barrier to an appreciable extent to interact sufficiently with skin dendritic cells to alter their phenotype.

Discussion
Engineered NPs have broad applications in many industries and are extensively under development for biomedical use [62][63][64] .For example, NPs are being engineered for use in vaccine development where they act as adjuvants and/or carriers to generate antigen-specific tolerogenic adaptive immunity 64 .This is a superior therapeutic strategy compared to suppressing the entire immune system which, can cause long-term damage 65 .Hence, at the forefront of the nanomedicine and nanotoxicology fields, is the need to understand and control how NPs interact with the immune system 66 .This work expands on our previous studies that showed NPs can modulate the adaptive immune response in a CHS mouse model 21 .We showed that SiO 2 NPs suppressed the allergic response in the challenge phase and mTiO 2 NPs exacerbated it [21][22][23] .The mechanism of how these NPs can alter adaptive immune responses remains unclear, which motivated this investigation to examine how these NPs could impact dendritic cell phenotype and function by quantifying BMDC cytotoxicity, cytokine production, the expression of the B7 family co-stimulatory ligands, and the in vivo adoptive transfer CHS model.
Results of this study show that BMDC treated with DNFB or mTiO 2 , as a function of time and increasing concentration, are cytotoxic (Fig. 1) and they produce higher levels of proinflammatory cytokines (Fig. 2) compared to SiO 2 which exhibited a cytoprotective effect in DNFB co-culture studies whereas mTiO 2 exacerbated DNFB cytotoxicity (Fig. 5).TEM studies confirm that both NPs were taken up by the BMDCs and it showed that DNFB exposure caused BMDCs to lose dendrites and increase the presence of lysosomes (Fig. S2), which degrade exogenous materials 67 .mTiO 2 exposure induced a significant presence of lipid droplets whereas SiO 2 exposure showed only a small increase (Fig. S2).An increase in lipid droplets may result from oxidative stress 68 , or it may indicate upregulation in metabolic activity through glycolysis which also drives the secretion of inflammatory cytokines 69 .
Analysis of the B7 family of co-stimulatory ligands suggest that DNFB and mTiO 2 induce a proinflammatory BMDC (CD11c, MHCII+) phenotype by up-regulating CD86, CD80 and PD-L1 (Fig. 3) similar to LPS (Fig. S10) whereas, SiO 2 had little effect and in fact promoted a more naïve phenotype by inducing a decrease in the percent of CD86 + CD80+ cells over time (Fig. 4).Dendritic cells expressing low levels of CD80/CD86 present antigen poorly and may induce tolerance 47,48 .We observed that potent proinflammatory stressors (LPS, DNFB, mTiO 2 ) upregulate the immunosuppressive PD-L1 ligand (Fig. 3, Fig. S12) which, is a mechanism by which the PD-1/ PD-L1 pathway balances the pro-inflammatory effect by promoting the development of Foxp3 + Tregs to limit the immune responses 43,45,70,71 .The co-stimulatory molecule CD276 was found to be prominently expressed only on CD80+ cells (Fig. 4, Fig. S13) which, binds the inhibitory CTLA-4 receptor with high affinity 49,50 .Despite its link to promoting an immunosuppressive BMDC phenotype following activation of the arylhydrocarbon receptor 60 , CD276 expression was down regulated by both NPs in our DNFB co-culture studies (Fig. 6), suggesting that CD276 plays a minimal role driving the adaptive immune response in this CHS adoptive transfer model.Both NPs induced similar effects in modulating the expression levels of CD86, CD80, PD-L1 and CD276 induced by DNFB with the key exception of PD-L2, where mTiO 2 downregulated PD-L2 on the activated CD86+ CD80+ subpopulation and SiO 2 did not (Figs.6 and 7).This suggests an important role of PD-L2 in directing the efficiency T cell priming in the DNFB-CHS model where SiO 2 treated BMDC suppresses the allergic response in both the sensitization and challenge phases whereas, mTiO 2 exacerbated it (Fig. 8).This finding is consistent with studies of allergic asthma, that showed PD-L2 expression in the lung was protective against the initiation and progression of airway inflammation [72][73][74][75] .Our results point to the importance of PD-L2 expression in the sensitization phase with DNFB, a T H 1 skewing hapten 76,77 .Studies suggest that PD-L1 and PD-L2 participate in the differential regulation of T H 1 and T H 2 cells 78 .The PD-L1/PD-1 interaction causes a T H 2 response and an increased IL-4 secretion while the PD-L2/PD-1 interaction causes a T H 1 response and an increase in INF-γ secretion 73 .In future studies it would be important to analyze full cytokine panels that contain T cell polarizing signals as well as chemokines important for lymph node trafficking.Quantifying the expression of the chemokine receptor CCR7 on the engineered BMDC is important as upregulation directs their migration to T-cell zones in lymph nodes 79 .The percent of the S.C. injected BMDC that traffic to the lymph node could be determined using fluorescently labeled BMDC and correlated to CCR7 expression.Quantifying these metrics is important for engineering tolerogenic dendritic cell therapies for treating autoimmune and severe allergic disorders 65,80 .
While our studies point to differences in PD-L2 expression as a potential mechanism for the differential effects of SiO 2 and mTiO 2 on DNFB sensitization (Fig. 8), it is important to note that the adoptive transfer of ex vivo engineered BMDC preparations contains a heterogeneous mix of B7 phenotypes.Additional studies would be informative to examine the relative importance of each BMDC subset more fully in driving the allergic response.Specifically, the different CD86/CD80 subpopulations (DP, DN, single positive) could be sorted to test which phenotype induces potent allergic responses.Studies show that dendritic cells expressing high levels of CD80 but not CD86 are protective and can induce immune tolerance via promoting CD25+ regulatory T cells 81 .It is also important to investigate the effect of protein coronas that form on NPs exposed to biological fluids and cell culture media 82,83 .Differences in the corona protein composition or abundance could alter the NPs interaction with the BMDCs.Corona composition is highly dependent on surface charge 84 .SiO 2 and mTiO 2 are both negatively charged so we anticipate that similar compositional coronas would form but this should be confirmed in proteomics studies.It is also possible that culturing BMDC with DNFB may haptenize the cells making them directly antigenic.Injection of haptenized BMDC could activate endogenous antigen presenting cells (APCs) in the skin or in the lymph node.Sensitization via this mechanism could be confirmed using transgenic mouse models with deleted endogenous APCs, however, the striking differential effects of the NPs observed in sensitization phase suggest a role for the ex vivo engineered BMDC in directing the observed adaptive immune responses.
This study corroborated our earlier studies using topical DNFB sensitization that showed (Fig. 8B) these NPs affected the allergic response in the challenge phase with mTiO 2 exacerbating the ear swelling and SiO 2 suppressing it 21 .While the mechanism in the challenge phase remains unclear, it seems plausible that the NPs could modulate epidermal-derived signals that affect MC and/or DC activation.These signals could be alarmins produced by keratinocytes or the NPs could modulate of the endogenous expression of CD80 or CD86 on keratinocytes.Studies using transgenic mouse models that over expressed CD80 or CD86 on basal keratinocytes showed a differential ability induce a chronic inflammatory response in the DNFB CHS model 85 .The immunomodulation IL-10 cytokine was also persistently increased in the mouse ear skin in the CD80 transgenic mouse which is consistent with a potent proinflammatory response observed with LPS and mTiO 2 treatment in this study.
In summary, this work points to metrics that can be used to predict the effects of NPs on contact allergy and points to the novel use of NPs to engineer immunomodulatory responses in contact allergy.Given that skin contact allergy is on the rise 1,2 , as is the creation of novel engineered nanomaterials for industrial, biomedical and consumer use [62][63][64]86 , there is a need for assays that can predict the impact that NPs may have on the immune response in the context of skin allergic disease. Futher, immunoengineering is an important growing field for developing cell-based therapies to induce antigen specific immune tolerance 42,64,87,88 .

Animals
Hairless SKH mice back-crossed 6 generations with C57BL/6 mice were used in this study and all previous work [21][22][23][24] .All animal experimental protocols were reviewed and approved by the University of Rochester Committee on Animal Resources (UCAR #2010-24E).Experiments involving animals and reporting of data were carried out in compliance with the ARRIVE guidelines and all methods were carried out in accordance with relevant guidelines and regulations.

SiO 2 and mTiO 2 NP characterization
Using dynamic light scattering and zeta potential measurements the SiO 2 NPs (nanoComposix Cat# SISN20-25M) exhibited a hydrodynamic diameter of 33.5 nm (± 3.3 nm), zeta potential of − 21.9 mV (± 10.1 mV) and a PDI of 0.236 in ultrapure water (pH 6.5) 23 .The mTiO 2 NPs (Sigma-Aldrich Cat# 677469, < 100 nm) exhibited a hydrodynamic diameter of 556.4 nm (± 34.4 nm), zeta potential of −9.05 mV (± 1.3 mV) and a PDI of 0.296 in ultrapure water (pH 6.5) 22 .The lower surface charge on the mTiO 2 NPs suggest a greater tendency to agglomerate as is evidenced by the hydrodynamic diameter being larger than the vendor reported primary size particle size < 100 nm.Transmission electron microscope (TEM, Hitachi 7650) was also used to measure free NPs and NPs inside the cells.The average size of the NPs inside the cells, which for mTiO 2 NPs was found to be 51.6 ± 12.5 nm and SiO 2 NPs were found to be 20.6 ± 3.5 nm (Figs.S3, S4).

Generation of BMDCs exposed to DNFB, SiO 2 , and mTiO 2 NPs
The BMDCs were exposed to DNFB, SiO 2 , or mTiO 2 at different concentrations for 1 h and non-cytotoxic concentrations (0.001 mM DNFB, 0.01 mg/mL Si20 nm and 0.005 mg/mL mTiO 2 ) were chosen for the 24 h time studies and subsequent single exposure experiments.The co-exposure of DNFB with SiO 2 NPs did not give any discernible toxicity in vitro, for which, the concentration of DNFB was increased ten-fold (0.01 mM) for some experiments.The concentrations for the co-exposure of DNFB with mTiO 2 NPs remained at 0.001 mM and 0.005 mg/mL, respectively.

Flow cytometry
For cell staining and analysis of co-stimulatory molecule expression, the antibodies and the concentrations used per 1 M cells are summarized in Table S1.We used flow cytometry (Cytek Aurora, Cytek Biosciences) and FlowJo (v10.

Cytotoxicity and cytokine analysis
Cytotoxicity was measured by flow cytometry using the eFluor™ 780 viability dye that labels dead cells.We normalized the BMDCs live population against imDC control cells that were not treated with any stressor.In addition, we used the PrestoBlue cytotoxicity assay (Invitrogen, P50200) according to manufacture protocol, to confirm the trends observed using flow cytometry.The cell culture supernatant was collected and analyzed for the pro-inflammatory cytokines IL-6, TNFα and the immunosuppressive cytokine IL-10 by ELISA (Invitrogen Cat# 88-7064, 88-7324 and 88-7105, respectively) following manufacture instructions.These cytokines were selected as they represent important innate proinflammatory and immunosuppressive makers.

Contact hypersensitivity (CHS) mouse model
As in our prior work 21 , mice were sensitized topically on the back by applying 20 µL DNFB (0.05%) diluted in an acetone and olive oil vehicle in a 4:1 ratio.After 5 days, we performed challenge with 20 µL of 0.2% DNFB on one ear and vehicle (4:1 acetone and olive oil ratio) on the other ear.Alternatively, we sensitized the mice subcutaneously (S.C) by injecting BMDCs (2 × 10 6 ) treated for 1 h with DNFB (0.01 mM) only or DNFB (0.01 mM) plus NPs; SiO 2 (0.01 mg/mL) or mTiO 2 (0.005 mg/mL).After 5 days, we performed challenge with 20 µL of 0.2% DNFB on one ear, and the other ear treated with vehicle (4:1 acetone and olive oil ratio) or 0.2% DNFB plus SiO 2 or mTiO 2 NPs using the same doses and protocols in prior work 22,23 .The magnitude of the allergic response post challenge was quantified by measuring ear thickness using a digital caliper (Mitutoyo Cat# 209-931) with a resolution of 0.005 mm on day 5, before the application of the challenge dose (pre-challenge).On day 6, both ears were remeasured (post-challenge).Ear swelling (mm) was measured as: (post-challenge) − (pre-challenge).

Statistics
We used GraphPad Prism 9 to analyze all statistical analyses.Ordinary one-way or two-way analysis of variance (ANOVA) was used to compare expression levels of co-stimulatory molecules to imDC (*) and DNFB (#).imDC were not treated with any stressor and served as control.Two-tailed, unpaired with unequal variances, student's t-test was used to compare the ear thickness between two different sensitizing treatments in the CHS in vivo study.p-values < 0.05 were considered significant.All data are presented with standard deviation.The experiments were replicated at least three times.

Figure 1 .
Figure 1.Effects of DNFB, SiO 2 and mTiO 2 exposure on BMDC cytotoxicity as a function of concentration and time.BMDCs were harvested on day 8 and treated with DNFB, SiO 2 and mTiO 2 to study cytotoxicity as a function of concentration and time by flow cytometry.(A) Cell viability as a function of concentration for a 1 h exposure.(B) Cell viability following exposure to the lowest non-cytotoxic concentrations from (A) and exposed over a period of 5 h, 15 h and 24 h.Results indicate that cytotoxicity of DNFB and mTiO 2 NPs on BMDCs was dose-and time-dependent.Live population was normalized the imDC no treatment control.Ordinary oneway ANOVA was performed and compared to imDC.N = 3-5.Mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 4 .Figure 5 .
Figure 4. Co-stimulatory molecules from the B7 family quantified by flow cytometry gated on the CD11c+ MHCII+ and CD80/CD86 subpopulations following BMDC exposure to each stressor over time.Results for DNFB are colored in blue, SiO 2 in green and mTiO 2 in purple.The phenotypic characteristics of BMDCs following exposure to a low concentration of each stressor

Figure 8 .
Figure 8.Comparison of CHS allergic response for different models of sensitization and challenge.(A) Comparison of ear swelling response for DNFB topical (0.05%) vs. S.C. (0.01 mM DNFB) sensitization with 0.2% DNFB challenge on one ear vs. vehicle on the other.No differences between these sensitization methods was observed.(B) Comparison of ear swelling response for S.C. sensitization with DNFB only and challenge with DNFB or DNFB co-exposure with NPs and vehicle on the other.Ear swelling was exacerbated with mTiO 2 NPs but decreased with SiO 2 NPs compared to DNFB alone.Ordinary one-way ANOVA was performed and compared to imDC (*).N = 3-12.Mean ± SD. *p < 0.05, **p < 0.01.(C) Comparison of ear swelling response for S.C. sensitization with BMDC treated 1 h with DNFB only or DNFB + SiO 2 or DNFB + mTiO 2 and challenge with 0.2% DNFB on one ear vs. vehicle on the other.Ear swelling was exacerbated with mTiO 2 but decreased with SiO 2 compared to DNFB alone.
7.2) to analyze cells.The gating strategy is shown in Fig. S10A.For each experiment FMO controls with BMDC are used.Examples of the FMO gating are illustrated in Fig. S10B,C.