Impact of Surface Modication on Cellular Uptake and Cytotoxicity of Silica Nanoparticles

Background: Silica nanoparticles (SiO 2 NPs) are widely used in industrial products as additives for rubber and plastics or as ller strengthening concrete, as well as being used in the biomedical eld for drug delivery and theranostic purposes. The present study investigated the effects of amino or carboxyl functionalization of rhodamine-labeled SiO 2 NPs on cellular uptake and cytotoxicity. Methods: Male mice were randomly divided into seven groups (n=6, each) and exposed to non-functionalized (plain), carboxyl or amino-functionalized rhodamine-labeled SiO 2 NPs at 2 or 10 mg/kg bw, or endotoxin-free water, by pharyngeal aspiration. At 24 hours after administration, the mice were euthanized and bronchoalveolar lavage uid (BALF) was collected for differential cell count and identication of silica nanoparticle uptake using confocal microscopy. In the in vitro studies, murine RAW264.7 macrophages were exposed to non-functionalized, amino- or carboxyl-functionalized Rhodamine-labeled SiO 2 NPs. Nonspecic caspase inhibitor and necrostatin-1 were used to determine the involvement of caspase or receptor-interacting protein 1 kinase domain in the cytotoxicity. Results: The in vivo study demonstrated that the neutrophil and macrophage counts and the percentage of macrophages with internalized particles was highest in the order of carboxyl >= amino- > > non-functionalized particles. The in vitro study demonstrated greater cytotoxicity for non-functionalized silica nanoparticles, compared to the others. Treatment with non-specic caspase or necroptosis inhibitor did not attenuate MTS cytotoxicity of non-functionalized silica nanoparticles. Conclusion: We conclude that carboxyl-functionalzed SiO 2 NPs are internalized by macrophages more eciently but less cytotoxic than plain SiO 2 NPs. The cytotoxic effect of plain SiO 2 NPs, which cannot be explained by apoptosis or necroptosis, can be avoided by carboxyl- or amino- functionalization.


Introduction
Nanoparticles (NPs) can be potentially used in various applications in the elds of nanotechnology, research and medicine. The small particle size and unique chemical and physical properties are highly exploitable in different consumer products as well as in the biomedical eld, such as drug delivery vectors, imaging purposes and/or cell tracking [1][2][3][4]. However, their low biodegradability, potential inhalation or prolonged skin contact with nanoparticles based on the small diameter and high surface reactivity are considered drawbacks for wider usage [4]. Therefore, before the application of NPs becomes widely accepted, it is necessary to establish their toxicological pro le and understand the potential adverse effects that might arise from human and environmental, accidental or desired exposure.
Silicon dioxide (silica) nanoparticles (SiO 2 NPs) are nowadays widely used in industrial products as additives for rubber and plastics or as strengthening llers for concrete. In addition, they are used in the biomedical eld for drug delivery or theranostic purposes [5][6][7][8][9]. The toxicological pro le of SiO 2 NPs has been widely studied recently, with increasing body of literature on the potential adverse effects of SiO 2 NP exposure. Silica materials exist in both crystalline and amorphous forms. The most common form of crystalline silica is quartz, whose toxicity has been studied for many years and is linked with chronic bronchitis, emphysema and silicosis [9][10][11]. Compared to the crystalline SiO 2 NPs, the amorphous form is less toxic [5,6]. With regard to the mechanism of toxicity, in vitro studies have shown that the toxic effect of SiO 2 NPs is mainly mediated by induction of oxidative stress and activation of intrinsic or mitochondrial (caspase-activated) apoptotic pathways [12][13][14][15][16][17]. Reactive oxygen species (ROS)-mediated cell death is considered one of the main mechanisms of action of many different types of nanomaterials (NMs), including silica NPs. In contrast, the results of a recent study indicate that total silanol content, cell membrane damage, and cell viability mediate the toxicity of SiO 2 NPs, rather than intracellular ROS [18].
The majority of the in vivo toxicological data are based on acute exposure studies, which usually include intra-tracheal instillation, intravenous injection or oral exposure [5,6]. Submicron amorphous silica particles were found to have greater in ammatory and cytotoxic potential compared to their bigger counterparts [19]. In the study of Morris et al. [20], C57BL/6 mice were intratracheally instilled with 4 or 20 mg SiO 2 NPs/kg body weight and signi cant effects were observed only at the high dose of 20 mg/kg. Twenty hours after instillation, approximately 20-and 10-fold increase in the cell number was observed in the bronchoalveolar lavage (BAL) of mice treated with the bare and amine-coated SiO 2 NPs, respectively, compared to the control mice; neutrophils were also increased about 30-and 20-fold, respectively [20]. Other studies demonstrated acute and chronic exposure to SiO 2 NPs aggravated airway in ammation [21][22][23][24][25].
One of the strategies to build "safe by design" NPs is to apply various types of surface modi cations to coat the NPs and modulate their surface reactivity, thereby decreasing their toxic effects. For example, surface modi cation of SiO 2 NPs was found to reduce their aggregation and nonspeci c binding [26], while functionalization with amino or phosphate groups was shown to mitigate their pro-in ammatory and immunomodulatory effects in allergic airway in ammation [21]. Interestingly, coating of SiO 2 NPs with polyethylene glycol polymer (PEG) was not found to be e cient in reducing the pro-in ammatory potential of these NPs in vivo [21,27]. Furthermore, few in vitro studies have shown that surface modi cation of SiO 2 NPs reduces their potential for in ammasome activation and cytotoxicity [28,29].
With this in mind, the present study was designed to determine the effect of surface modi cation of 25 nm amorphous SiO 2 NPs both in vivo (C57BL/6JJcl mice) and in vitro (murine macrophage RAW 264.7 cell line), with a special emphasis on their pro-in ammatory and cytotoxic potentials. Our data support the hypothesis that surface modi cation of SiO 2 NPs can modulate the uptake properties of the material, their pro-in ammatory potential and cytotoxicity. We showed that exposure to bare SiO 2 NPs with terminal Si-OH group decreased the number of macrophages collected from bronchoalveolar lavage uid (BALF) in vivo. On the other hand, both carboxyl-and amino-functionalized SiO 2 NPs were e ciently taken up by macrophages or neurtrophils in vivo, accompanied by increase in the number of neutrophils collected from BALF at the high dose used (10 mg NPs/kg bw). The results showed that SiO2 NPs cytotoxicity is not related with internalization of SiO2 NPs and neutrophil in ltration, although internalization of SiO2 NPs might be linked to neutrophil in ltration. The results of in vitro and in vivo experiments were complementary to each other, underlining the importance of surface modi cation of SiO 2 NPs in modulating their toxicological pro le, and suggesting that functionalization of 25 nm SiO 2 NPs with carboxyl groups in consumer products or for biomedical applications could lead to e cient cellular uptake, and that functionalization with the amino or carboxyl group reduced the toxicity of SiO 2 NPs.

Silica nanoparticles
Rhodamine-labeled SiO 2 NPs, "Sicastar", of 30 nm in diameter, functionalized with carboxy (Cat. #40-02-301), amino group (Cat. #40-01-301) or non-functionalized (Cat. #40-00-301) were purchased from Micromod Partikeltechnologie, Rostock, Germany. All NPs were dispersed in water at 25 mg/mL. The size of the NPs in water or in complete cell culture medium was characterized by DLS, Zeta potential analysis and TEM. Zeta-potential was measured with Photal LEZA-600 (Otsuka Denshi Co., Osaka). The uorescence intensities of the three types of silica NPs were measured at different concentrations using ARVO TM MX 1420 Multilabel Counter (Perkin Elmer, Waltham, MA). The slopes of the regression lines for independent variable of concentration and dependent variable of uorescence intensity were calculated to obtain the relative uorescence intensity between different types of silica NPs labeled with rhodamine (Supplementary le 1).

Animals
Forty-two male C57BL/6JJcl mice of 7-week-old were purchased from CLEA Japan Inc (Tokyo, Japan). All mice were housed and acclimatized to the new environment for one week in a pathogen-free animal room controlled at 23-25 °C and 55-60% humidity. Light was set within a 12-h light-dark cycle (on at 09:00 and off at 21:00), and food and water were provided ad libitum. This study was conducted according to the Japanese law on the protection and control of animals and the Animal Experimental Guidelines of Tokyo University of Science. The experimental protocol was also approved by the Animal Ethics Committee of Tokyo University of Science prior to the experiment.
Mice were randomly divided into seven groups of six each and exposed to plain SiO 2 NPs, carboxyl SiO 2 NPs or amino SiO 2 NPs at 2 or 10 mg/kg bw, which were equivalent to 40 or 200 µg per mouse if body weight was 20 g, or endotoxin-free water. These exposure levels were half of those adopted in a previous study, which demonstrated that exposure to SiO 2 NPs by intratracheal instillation increased the number of macrophages in BALF at 0.5 mg silica/mouse (20 mg/kg bw) but not at 0.1 mg silica/mouse (4 mg/kg bw) (Morris 2016). With regard to the relation of the above exposure levels by intratracheal instillation to those by inhalation, one study showed that 690 µg of titanium dioxide deposited in the rat lung after exposure to ultra ne titanium dioxide at 125 mg/m 3 for 2 hours (Osier 1997). Since the alveolar surface areas of rat and mice are 2970 [30] and 82.2 cm 2 [31], respectively, the estimated lung deposition of titanium dioxide in mice was 690 × 82.2/2970 = 19 µg when exposed to ultra ne titanium dioxide at 125 mg/m 3 for 2 hours. The higher dose of 200 µg in the present study is comparable to ultra ne particles deposited in the lungs after inhalation exposure to ultra ne particles at 125 mg/m 3 for 21 hours.
SiO 2 NPs dispersed in water at 25 mg/mL were vortexed and then further diluted with endotoxin-free water to obtain the NP solution at 1 and 5 mg/mL.
Mice were anesthetized with pentobarbital and then exposed to 40-µl aliquot of samples of SiO 2 NPs by pharyngeal aspiration, as described previously (Gabazza et al. 2004, Wu et al. 2015. The technique of pharyngeal aspiration involved placement of the nanoparticle suspension on the back of the tongue followed by pulling of the tongue to induce a re ex gasp with resultant aspiration of the droplets. At 24 h after administration, the mice were euthanized by intraperitoneal injection of pentobarbital. BAL was performed by cannulation of the trachea with 18-gauge needle, and infusion and collection of 5/6 mL of saline was repeated six times.

Total and differential cell counts in BALF
The recovered BALF was centrifuged (1,500 rpm, 5 min, 4 °C), and the cell pellet was mixed with 1 mL of ACK lysing buffer (Gibco-Thermo Fischer Scienti c, Waltham, MA) for hemolysis. Ten mL of Dulbecco's phosphate buffer saline (DPBS) was added and centrifuged at 1,500 rpm and 4 °C. The resultant pellet was re-suspended in DPBS for total and differential cell counts. The total cell count was measured using hematocell counter. Aliquots of 5 × 10 4 cells in 400 µl DPBS per slide were prepared for cytospins. The cell mixture was added to EZ Single Cytofunnel®(Thermo, UK and centrifuged for 10 min at 1,000 rpm with Cytospin, using cytoslides. The slides were dried overnight at room temperature and then stained with the Differential Quik Stain Kit (Modi ed Giemsa, Sysmex Co., Kobe, Japan) for differential cell count in 10 elds (20 x magni cation) of each slide.

Fluorescence immunocytology
A slide obtained by Cytospin was washed in DPBS three times and incubated with blocking agent (1% BSA) for one hour. The slide was further incubated with Biotin anti-Ly6G and Ly6C (Gr-1) (BD, Franklin Lakes, NJ), which was diluted 400 folds in 1% BSA, for one hour at room temperature, washed in DPBS three times and then incubated with 200-fold diluted Cy5 streptavisin (BioLegend, San Diego, CA) for 30 min at room temperature to stain the neutrophils. Ly6G and Ly6C (Gr-1) were used as markers of neutrophils, as their expression levels are known to correlate with differentiation and maturation of granulocytes [32,33] and are only expressed transiently on bone marrow cells in the monocyte lineage [33]. The slides were counterstained with Hochest33342 for 10 min at room temperature and enclosed #11360-070), 100 U/mL penicillin, 100 mg/mL streptomycin and 250 ng/mL amphotericin B (anti-anti, Gibco, Cat #15240-062), 2 mM glutamine and 10% FBS. All experiments were performed with cells from passages 4 to 15. Cells were grown in T25-asks (Violamo, AS ONE Co., Osaka) in monolayers. Exponentially growing cells were maintained in a humidi ed atmosphere of 5% CO 2 and 95% air at 37 °C and were passaged once every two days using a cell scraper (NEST 100071).

NP treatment
Depending on the experiment and after reaching 70-80% con uence, the cells were seeded onto appropriate cell culture plates and treated with silica nanoparticles of different surface modi cation. Before exposure to NP, the cells were rinsed with PBS to eliminate trace amounts of FBS. Treatments were performed under FBS-free condition for two reasons: 1) serum is reported to modulate NP uptake [34], and 2) to mimic in vivo condition whereby bronchial cells are not directly exposed to serum proteins. Stock of 30 nm rhodamine-labeled silica nanoparticles (25 mg/mL in water) was vortexed shortly before the preparation of the nal dilution for the treatment.

MTS assay
MTS assay was conducted using CellTiter 96® Aqueous One Solution Cell Proliferation Assay (Promega, Madison, WI) as described previously [35] and the instructions provided by the manufacturer. Brie y, RAW264.7 cells were seeded at 1.5 × 10 4 cells/well onto 96-well plates and incubated at 37 ºC in a humidi ed atmosphere of 5% CO 2 and 95% air for 24 hours. After incubation, the cell culture medium was The maximum concentration of silica nanoparticles 35.2 × 10 5 µg/cell (8.40 µg/cm 2 ) was determined to be equivalent to exposure level of 40 µg/mouse in vivo, given that the average number of macrophages collected in BALF was 1.14 × 10 5 cells/mouse. Flow cytometry RAW264.7 cells were seeded onto 6-well plates at 2.37 × 10 4 cells/cm 2 in complete cell culture medium and incubated for 24 h before treatment. After treatment with 3.0 mL/well of silica NPs at the preselected concentrations for 1 or 4 h in dark, the medium was removed, cultures were thoroughly washed with PBS three times and treated with 0.1% trypan blue for 1 min to quench the uorescence of rhodamine on the cell surface. The concentration of trypan blue for quenching the uorescence of rhodamine was determined beforehand by plot of trypan blue and rhodamine intensity in RAW264.7 cells exposed to bare rhodamine-labeled SiO 2 NPs ( Supplementary Fig. 2). The cells were washed with PBS, mixed with 500 µL of FACS buffer (PBS containing 0.5% FBS and 0.1% NaN 3 ) and harvested by cell scraper. Cell-associated uorescence was detected using FACSCalibur™ and results were analyzed with FlowJo software (BD, Franklin Lakes, NJ). The Mean Fluorescence Intensity (MFI) of rhodamine (excitation 488 nm, Filter range 564-606 nm) from three different size fractions indicated by the forward scatter (FS) was shown by the ow cytometer. The results are reported as the median of the distribution of cell uorescence intensity obtained by analyzing cells in the gate. To adjust the differences in the uorescence intensity relative to weight, the intensity of rhodamine uorescence was measured at different concentrations of three types of SiO 2 NPs using ARVOMx-a system (485 nm/535 nm 1.0 s).
LDH cytotoxicity assay LDH cytotoxicity assay was conducted using Pierce LDH cytotoxicity assay kit following the instructions provided by the manufacturer (Thermo Fisher Scienti c). Brie y, RAW264.1 cells were plated at 10 4 cells/well in 100 µl of medium in a 96-well tissue culture plate. To minimize the cytotoxic effect of lack of FBS, DMEM was replaced by Opti-MEM for LDH cytotoxicity assay and further investigation of the effect of pan-caspase or necroptosis inhibitor. After incubation at 37 ºC under 5% CO 2 for 24 hours, the cells were exposed to bare, carboxyl-functionalized, amino-functionalized rhodamine-labeled SiO 2 NPs at 5.85 µg /cm 2 (19.5 µg/mL). After the exposure for 1, 4, 12, 24, 36 and 48 hours, the supernatant of the

Statistical analysis
Data are expressed as mean ± standard deviation (SD). Differences with the control or between groups were analyzed respectively by Dunnett's or Tukey's multiple comparison following one-way ANOVA. A probability (p) of < 0.05 denoted the presence of a statistically signi cant difference. For the analysis of relative uorescence intensity among different types of particles, regression lines were obtained by forcing the intercept to zero using Excel 2016 (Microsoft, Redmond, WA). All statistical analyses were performed using JMP (version 14, SAS Institute, Cary, NC).

Physicochemical properties of SiO 2 NPs
We studied rst the in vitro and in vivo effects of the three types of 25 nm SiO 2 NPs. For this purpose, we used plain SiO 2 NPs with hydroxyl group (OH-SiO 2 NPs), and carboxyl-or amino-modi ed SiO 2 NPs (COOH-SiO 2 NPs or NH 2 -SiO 2 NPs). The diameter and surface charge of these NPs were determined by DLS and Zeta potential measurements both in water (since the materials were dispersed in endotoxinfree, ultra-pure water for intra-tracheal instillation in mice lungs) and in the cell culture medium relevant to the exposure of RAW264.8 cells. As indicated in Table 1, the mean diameters of the carboxyl and amino modi ed, and OH-SiO 2 NPs in water were 33.9 ± 0.1 nm, 34.8 ± 0.2 nm and 34.1 ± 0.2 nm, respectively, indicating a mono-dispersed size distribution of NPs. Similarly, the mean diameters of the three types of SiO 2 NPs in the cell culture medium were comparable, with 33.6 ± 0.6, 33.6 ± 0.3 and 36.3 ± 0.3 nm for the carboxyl and amino modi ed, and plain NPs, respectively. Interestingly, regardless the surface modi cation of these NPs using three different functional groups, their overall surface charge was neither modi ed in water nor in the cell culture medium ( Table 1). As indicated, the mean surface charges of COOH-, NH 2 -and OH-SiO 2 NPs in water were − 25.6 ± 4.05, -41.5 ± 3.02 and − 33.2 ± 1.85 mV, respectively, while the respective values in the cell culture medium were − 23.5 ± 4.0 mV, -23.8 ± 2.5 mV and − 23.3 ± 1.9, respectively. That the mean surface charge was more negative than − 20 mV indicates good colloidal stability of SiO 2 NPs both in water and in the cell culture medium. Data are mean ± SD of 4 and 3 measurements of diameter and zeta potentials, respectively.

In vivo toxicity
In order to address the in vivo toxicity of the three types of SiO 2 NPs, we measured body and lung weights of the test animals as a rough indication of disturbance of body and lung homeostasis following instillation of NPs. We also collected BALF for total and differential cell counts, which are reliable markers of lung in ammation induced by NPs. In order to better understand the localization of NPs after the instillation procedure, the macrophages and neutrophils were differentially stained and examined by confocal microscopy.

Body and lung weight
There was no signi cant difference in body weight between vehicle-treated control mice and mice treated with any of the three types of SiO 2 NPs at 2 or 10 mg NPs/kg bw (Table 2). On the other hand, there was a signi cant increase in the lung weight in the OH-SiO 2 NP -treated mice at 2 and 10 mg/kg bw.   BALF total and differential cell count Treatment with different types of SiO 2 NPs at 2 mg/kg bw did not increase the total cell number (Fig. 1A).
On the other hand, treatment with amino-and carboxyl-functionalized SiO 2 NPs at 10 mg/kg bw induced increase in BALF total cell count. Interestingly, there was no change in the total cell number after treatment with hydroxyl functionalized SiO 2 NPs at 10 mg /kg bw. Further analysis showed treatment with 10 mg/kg bw OH-SiO 2 NPs decreased the total number of BALF macrophages, compared to the untreated group (Fig. 1B). On the other hand, 10 mg/kg bw carboxyl functionalized SiO 2 NPs increased the number of macrophages (Fig. 1B), and carboxyl-or amino-functionalized SiO 2 NPs increased BALF neutrophils (Fig. 1C). The increase in the numbers of total cells and neutrophils was higher with COOH-SiO 2 NPs than NH 2 -SiO 2 NPs (Fig. 1A and C). Hydroxyl (plain) SiO 2 NPs increased the number of neutrophils at 2 mg/kg bw, but had no such effect at 10 mg/kg bw (Fig. 1C).
Interaction of SiO 2 NPs with BALF macrophages and neutrophils COOH-and NH 2 -SiO 2 NPs were both similarly internalized by cells regardless their dose, while only a few OH-SiO 2 NPs made contact with the cells (Fig. 2A). Internalization of NH 2 -and COOH-SiO 2 NPs into the macrophages was not in uenced by their dose (persistently ~ 60%), while only around 40% of the macrophages internalized OH-SiO 2 NPs (Fig. 2B and D). On the other hand, internalization of carboxyl and amino functionalized SiO 2 NPs into the neutrophils was dose-dependent (30-40% of cells internalized SiO 2 NPs at 10 mg/kg bw, Fig. 2C and E), with the numbers and percentages of such neutrophils signi cantly higher at 10 mg/kg bw of each of the above two types, compared to those neutrophils with internalized hydroxyl functionalized SiO 2 NPs (only 10% at 10 mg/kg bw).
In vitro toxicity of NPs MTS cell viability assay of SiO 2 NPs in RAW264.7 macrophage cell line The MTS assay was used to assess the cytotoxic potential of SiO 2 NPs, in which tetrazolium salts are reduced to formazan by metabolically active cells (probably through NAD(P)H-dependent dehydrogenases), producing measurable color changes proportionate to the number of viable cells. The percentage of live cells is then estimated by measuring absorbance of formazan at 490 nm, wavelength that is clearly distinguishable from rhodamine absorption pro le. Treatment of the cells with OH-SiO 2 NPs, but not NH 2 -and COOH-SiO 2 NPs, signi cantly decreased cell viability in time-and dose-dependent manners (Fig. 3). Exposure to OH-SiO 2 NPs signi cantly decreased cell viability at ≥ 3 µg/cm 2 (≥ 10 µg/ml) after 4 h of treatment and at ≥ 7.5 µg/cm 2 (≥ 25 µg/ml) after 24 h of treatment. The decrease in cell viability was also dose-dependent, with < 40% of the cells remaining metabolically active after 4 h exposure to 30 µg/cm 2 (100 µg/ml) of OH-SiO 2 NPs (Fig. 3A). The decrease in cell viability was also timedependent; <40% of the cells was still alive and metabolically active after 24 h of treatment with 7.5 µg/cm 2 (25 µg/ml) OH-SiO 2 NPs (Fig. 3B). The above in vitro data support the hypothesis that OH-SiO 2 NPs is the most toxic among the three types and suggest that the observed decrease in BALF macrophages (Fig. 1B) was mediated through cell death. Similarly, this cytotoxic effect of OH-SiO 2 NPs could explain the observed increase in BALF neutrophil count, compared to the control, though it but reached a plateau at ≥ 2 mg/kg bw (Fig. 1C), resulting in lack of statistical different in BALF total number of cells in the untreated mice. These data highlight the need to perform BALF differential cell count in order not to overlook subtle effects of these NPs on different cells in the lung.
Interaction of SiO 2 NPs with RAW264.8 macrophage cell line Next, we assessed the interaction of the three types of NPs with RAW264.8 macrophages in order to understand the mechanism of OH-SiO 2 NPs toxicity. COOH-SiO 2 NPs were e ciently internalized by RAW264.8 macrophages from 1 h after exposure, in dose-and time-dependent manners (based on the higher amounts of internalized NPs in RAW264.8 macrophages after 4 and 24 h of treatment, Fig. 4A).
Interestingly, hydroxyl plain NPs mainly co-localized with the plasma membrane irrespective of the dose used (Fig. 4A). This was remarkable already after 1 h of treatment at the highest concentration of NPs and even more prominent after 4 and 24 h of treatment.
Internalization of NPs by cells was further quanti ed by ow cytometry (Fig. 4B and Suppl Fig. 1). To eliminate the signal coming from NPs adsorbed on the surface of the cells, 0.1% trypan blue was added shortly before the analysis. The working concentration of trypan blue was optimized prior to the experiment (Suppl Fig. 1). Flow cytometry showed three populations of RAW264.4 cells (Suppl Fig. 2).
Due to difference in the slopes of the regression lines, with rhodamine intensity as the dependent variable and SiO 2 NPs concentration as the independent variable (Suppl Fig. 3), the uorescence of COOH-SiO 2 NPs and OH-SiO 2 NPs was each divided by the ratio of the slope for COOH-SiO 2 NPs and OH-SiO2NPs to the slope for NH 2 -SiO 2 NPs for normalization. Exposure to COOH-SiO 2 NPs increased the mean value of the normalized uorescence intensity in the population gated to A (Fig. 4B), B and C (Suppl Fig. 4) in dose-and time-dependent manners. Exposure to NH 2 -or OH-SiO 2 NPs increased the mean of normalized intensity in all examined populations to an extent lesser than exposure to COOH-SiO 2 NPs.

Cytotoxicity of OH-SiO 2 NPs
The LDH cytotoxicity test showed higher LDH release from the cells exposed to OH-NPs compared to the other types of NPs (Suppl Fig. 5). Pretreatment with pan caspase or necroptosis inhibitor did not attenuate the decrease in MTS cell viability induced by exposure to OH-SiO 2 NPs for 18 h (Fig. 5A and B), suggesting that the cytotoxicity of OH-SiO 2 NPs is unrelated to apoptosis or necroptosis.  NPs induced the mRNA expression of proin ammatory cytokines to a greater extent than COOH-SiO 2 NPs. It is possible that the relatively low recruitment of macrophages and lymphocytes in OH-SiO 2 NPstreated mice resulted from death of these cells, while exposure to OH-SiO 2 NPs induced a more extensive in ammatory response, as evident by the higher transcription levels of proin ammatory cytokines in the remaining cells and increased lung weight following exposure to OH-SiO 2 NPs.
Flow cytometry showed the uptake of COOH-SiO 2 NPs by RAW264.8 macrophages starting from 1 hr after exposure and was higher than those of NH 2 -and OH-SiO 2 NPs. The higher rate of uptake of COOH-SiO 2 NPs is in agreement with our recent study, which showed better interaction of COOH-SiO2 NPs with Caco-2 cells compared with NH 2 -and OH-SiO 2 NPs [35]. Increased lung weight can re ect pulmonary in ammation, which is associated with intense protein/ uid leakage from the vasculature into the alveoli mediated by in ammatory signals from pulmonary cells [36]. The results suggests that NH 2 -and COOH-SiO 2 NPs can recruit neutrophils only at the high dose of 10 mg/kg bw, with COOH-SiO 2 NPs being more potent in attracting neutrophils, while the lack of dose dependency in increasing the number of neutrophils and the decrease in macrophage numbers with larger dose of OH-SiO 2 NPs could be explained by cell death induced by these NPs. With the aim of con rming and further investigating these ndings, we analyzed the uptake of uorescently-labelled SiO 2 NPs by neutrophils and macrophages and performed in vitro experiments using RAW264.8 macrophage cell line.
The different patterns of interaction with the cells and subsequent cellular uptake observed in the present study could perhaps explain the differences in the toxicological pro le of the three types of SiO 2 NPs used in this study. While COOH-SiO 2 NPs were e ciently internalized into the cells, OH-SiO 2 NPs predominantly interacted with the cell plasma membrane, potentially damaging the membrane and inducing cell death. The zeta potentials of all types of SiO 2 NPs in water were negative, in the order of NH 2 -SiO 2 NPs, OH-SiO 2 NPs and COOH-SiO 2 NPs. On the other hand, the zeta potentials of all types of SiO 2 NPs in the FBS-free media were almost of the same negative values. These results are in agreement with our recent study, which showed greater interaction of Caco-2 cells or THP-1 cells with COOH-SiO 2 NPs than OH-or NH 2 -SiO 2 NPs [35]. The present ndings and our previous study show that the difference in the extent of internalization or cytotoxicity is not explained by the differences in zeta potentials. Further studies are needed to understand the mechanism(s) involved in cellular internalization and toxicity of SiO 2 NPs.

Conclusions
Carboxyl modi cation of SiO2 NPs increases their uptake into macrophages while carboxyl-or aminomodi cation of SiO 2 NPs reduces cytotoxicity of SiO 2 NPs. The cytotoxicity of OH-SiO 2 NPs could be related to the physical contact between NPs and the cell membrane, though further studies are needed to test this hypothesis and to determine the exact mechanism of SiO 2 NPs-induced cytotoxicity.

Declarations
Ethics approval This study was conducted according to the Japanese law on the protection and control of animals and the Animal Experimental Guidelines of Tokyo University of Science. The experimental protocol was also approved by the Animal Ethics Committee of Tokyo University of Science prior to the experiment. The approval number is Y14059.

Consent for publication
Not applicable Availability of data and materials All date generated or analysed during the current study are included in this published article and its supplementary information les.

Competing interests
The authors declare that they have no competing interests.