Assessing the Cytotoxicity of TiO2−x Nanoparticles with a Different Ti3+(Ti2+)/Ti4+ Ratio

Titanium dioxide (TiO2) nanoparticles are promising biomedical agents characterized by good biocompatibility. In this study, we explored the cytotoxicity of TiO2−x nanoparticles with a different Ti3+(Ti2+)/Ti4+ ratio and analyzed the efficiency of eryptosis indices as a tool in nanotoxicology. Two types of TiO2−x nanoparticles (NPs) were synthesized by the hydrolysis of titanium alkoxide varying the nitric acid content in the hydrolysis mixture. Transmission electron microscopy (TEM) images show that 1-TiO2−x and 2-TiO2−x NPs are 5 nm in size, whereas X-ray photoelectron spectroscopy (XPS) reveals different Ti3+ (Ti2+)/Ti4+ ratios in the crystal lattices of synthesized NPs. 1-TiO2−x nanoparticles contained 54% Ti4+, 38% Ti3+, and 8% Ti2+, while the relative amount of Ti4+ and Ti3+ in the crystal lattice of 2-TiO2−x nanoparticles was 63% and 37%, respectively. Cell viability and cell motility induced by TiO2−x nanoparticles were investigated on primary fibroblast cultures. Eryptosis modulation by the nanoparticles along with cell death mechanisms was studied on rat erythrocytes. We report that both TiO2−x nanoparticles do not decrease the viability of fibroblasts simultaneously stimulating cell migration. Data from in vitro studies on erythrocytes indicate that TiO2−x nanoparticles trigger eryptosis via ROS- (1-TiO2−x) and Ca2+-mediated mechanisms (both TiO2−x nanoparticles) suggesting that evaluation of eryptosis parameters is a more sensitive nanotoxicological approach for TiO2−x nanoparticles than cultured fibroblast assays. TiO2−x nanoparticles are characterized by low toxicity against fibroblasts, but they induce eryptosis, which is shown to be a promising tool for nanotoxicity screening. The Ti3+ (Ti2+)/Ti4+ ratio at least partly determines the cytotoxicity mechanisms for TiO2−x nanoparticles.


Introduction
Nanomedicine is an emerging field of nanotechnology, which is revolutionizing healthcare worldwide [1]. Among nanostructured materials, whose biomedical application is under investigation, nano-sized titanium dioxide seems to be promising. TiO 2 or titania nanoparticles (NPs) have been used as biosensors, implant coatings, antibacterial agents, cell imaging agents, drug delivery systems, agents for photodynamic therapy, etc. [2][3][4][5]. In addition to the effectiveness, NPs should be biocompatible and environmentally friendly [6]. For instance, Hashem and co-workers show that biocompatible TiO 2 quantum dots synthesized by a green and ecofriendly method using watermelon peel waste exhibit antimicrobial, antioxidant, and anticancer activities [7]. Ag@TiO 2 composites embedded within hydroxyethyl cellulose or bacterial cellulose cryogels increase their antimicrobial efficiency making them a promising environmental friendly material [8]. TiO 2 NPs embedded into an organic/inorganic cellulose composite have been shown to have a good antimicrobial activity and can be used as catalysts for photodegradation of pesticide intermediates [9].
Nano-sized TiO 2 is a non-stoichiometric oxygen-deficient compound (TiO 2−x ). TiO 2−x NPs have been reported to be characterized by a defective structure with a high amount of oxygen vacancies (V O ) stabilized by Ti 3+ or Ti 2+ ions [10]. The defective crystal structure of TiO 2 NPs with a high amount of titanium ions in the lower oxidation states (3+ and 2+) at the surface of NPs is considered to be responsible for their redox properties and ROS scavenging ability reported by some authors [11,12]. In addition to intrinsic properties of titania NPs, they can be modified by adding various dyes, inorganic dopants, targeting molecules, drugs, etc. [3].
Titanium dioxide NPs are currently among the top 5 most widely manufactured nanomaterials, which may contribute to the environmental pollution and increased exposure to them by humans [13]. This raises awareness of their toxicity and suggests evaluation of potential hazardous biological effects of titania NPs. Experimental data indicate that titanium dioxide nanomaterials are characterized by low toxicity and good biocompatibility, which is of huge importance regarding the possible application of nano-sized materials in clinical practice [2,14]. It has been suggested that the toxicity of titania NPs is significantly influenced by the size, shape, crystal structure, surface coating, UV irradiation, and light exposure [15,16]. In many studies, evidence has been provided that nano-sized titanium dioxide toxicity is mediated by reactive oxygen species (ROS) overproduction [16][17][18][19][20][21][22]. In addition, it has been stated that TiO 2 NPs induce cell necrosis, apoptosis, genotoxicity, and inflammation via ROS-mediated pathways [23][24][25]. However, there are no experimental data describing the effects of titanium dioxide NPs on eryptosis, a programmed non-lytic cell death mode of red blood cells.
Cultured fibroblasts are widely considered to be an informative model for detecting the cytotoxicity of metallic NPs [26][27][28]. However, the field of nanotoxicology requires novel, standardized approaches to detect hazardous bioeffects of nanomaterials and nanotoxicity mechanisms [29,30]. There is accumulating evidence that the assessment of eryptosis is a suitable model for determining biocompatibility of nanoparticles [31][32][33][34][35][36][37]. Furthermore, eryptosis is a convenient tool to detect mechanisms of nanoparticle-induced nanotoxicity. In nanotoxicology, it has been reported to be more sensitive than hemolysis [36]. Erythrocytes are easily isolated and handled with. They are characterized by simple morphology and biochemistry, lack of membranous organelles, and are prone to free radical reactions, which is important given that most metallic nanomaterials possess the ROS-mediated toxicity [38,39]. Thus, evaluation of changes in eryptosis indices under exposure to TiO 2−x NPs is of huge interest.
In this study, we evaluated the toxicity of TiO 2−x NPs synthesized by an original route and characterized by different Ti 3+ (Ti 2+ )/Ti 4+ ratios in their crystal lattices and analyzed the effectiveness of eryptosis parameters as a nanotoxicological model compared with widely used cultured fibroblast assays.

Synthesis and Characterization of TiO 2−x Nanoparticles
TiO 2−x NPs were synthesized by the modified Grätzel method [40] in a water-butanol mixture with tetrabutylorthotitanate (Ti(OC 4 H 9 ) 4 , 97.0%, Sigma-Aldrich, USA) as a precursor and nitric acid (HNO 3 , 65.0%, Sigma-Aldrich, USA) as a catalyst and peptizing agent. The Ti 3+ /Ti 4+ ratio was controlled by the different amounts of HNO 3 added. Initially, 12.5 ml of Ti(OC 4 H 9 ) 4 was added to 2 ml of n-butanol (CH 3 (CH 2 ) 3 OH, ≥ 99%, Sigma-Aldrich, USA). The obtained mixture was added dropwise for 3 min to 75 ml of distilled deionized water under vigorous stirring. To obtain 1-TiO 2−x NPs, while pouring dropwise Ti(OC 4 H 9 ) 4 during 10 min, 530 µL of 65.0% HNO 3 was added to the hydrolysis mixture, still stirring vigorously. In case of 2-TiO 2−x, 200 µl of 65.0% HNO 3 was added in the similar way. Then, the mixtures were stirred for 4 h at 80 °C in a water bath. Finally, 50 ml of stable 1-TiO 2−x and 2-TiO 2−x colloid water solutions were obtained. The colloids were cooled down to room temperature and diluted up to 100 ml with distilled deionized water. Final concentrations of 1-TiO 2−x and 2-TiO 2−x in water solutions were 20 g/L.
Ti 2+ , Ti 3+ , and Ti 4+ sites in 1-TiO 2−x and 2-TiO 2−x were determined by X-ray photoelectron spectroscopy (XPS) using a JSPM-4610 XPS 2400 photoelectron spectrometer (JEOL Co., Ltd.) fitted with a Mg Kα source (soft X-ray source at 1253.6 eV). The base pressure of the JSPM-4610 XPS 2400 system was below 1 × 10 −7 Pa. The XPS peaks were assumed to have Gaussian line shapes; a non-linear least squares procedure was used to resolve the individual components after proper subtraction of the baseline. The XPS Peak 4.1 program was used for data analysis.

Collection of Blood Samples and Incubation Conditions
Blood specimens were collected from 6 intact WAG rats in K 2 EDTA Vacutainer test tubes (Guangzhou, China). K 2 EDTA was used to prevent blood clotting. Dermal fibroblasts were isolated from skin of rat embryos with the gestational age of 19-20 days (n = 8). The animals were provided by the vivarium of Kharkiv National Medical University (Khrakiv, Ukraine). The rats were housed in cages for 2 weeks under standard and uniform temperature, humidity, and lighting conditions to ensure acclimatization. Free access to water and standard laboratory rat chow was provided. The EU Directive 2010/63/EU on the protection of animals used for scientific purposes was followed in this study.
The research was conducted in compliance with the EU Directive 2010/63/EU on the protection of animals used Following incubation with TiO 2−x nanoparticles (0-200 mg per L) for 24 h eryptosis parameters (cell shrinkage, phosphatidylserine exposure, intracellular Ca 2+ levels, and reactive oxygen species formation) and fibroblast characteristics (viability, monolayer structure, and cell motility) were assessed for scientific purposes and Ukrainian legislation and was approved by the Commission on Ethics and Bioethics at Kharkiv National Medical University (Kharkiv, Ukraine, minutes #3 dated August 28, 2020).

Evaluation of FSC Signaling to Detect Nanoparticles-Induced Cell Shrinkage
Forward scatter (FSC) signaling depends on the relative cell size. A weaker FSC signal is typical for shrunken red blood cells, while its stronger signal is observed in enlarged erythrocytes [41]. Cell shrinkage is a characteristic morphological alteration of eryptotic erythrocytes. To compare the effects of TiO 2−x NPs on cell shrinkage, the percentage of cells with weak FSC signaling was analyzed.

Detection of Cell Membrane Scrambling
Loss of membrane asymmetry in red blood cells treated with TiO 2−x NPs was analyzed by staining with FITC-labeled annexin V purchased from Becton Dickinson (USA). Erythrocytes resuspended in 100 µl of freshly prepared Annexin-binding buffer (Becton Dickinson, USA) were stained with 5 µl Annexin V-FITC. Annexin V binding to externalized phosphatidylserine was provided by incubation for 15 min in the dark. Thereafter, 400 µl of annexin-binding buffer was transferred to the test tubes and the mixtures were gently mixed. The erythrocytes exposed to hydrogen peroxide (0.1 mM) solution were considered to be a positive control. The cells not stained with annexin V-FITC were used as a negative control [31,42].

Detection of ROS Production in Erythrocytes by Flow Cytometry
Blood samples incubated with TiO 2−x NPs were washed twice with phosphate-buffered saline (PBS, PH 7.4, BD, USA) to obtain erythrocyte masses used further to obtain erythrocyte suspensions. Briefly, 2 µL of erythrocyte mass was added to 100 µL PBS. Thereafter, 2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA, Invitro-gen™, USA) staining was performed. H2DCFDA is a widely used ROS-detecting probe, which is converted in erythrocytes to 2′,7′-dichlorodihydrof luorescein (H2DCF) by esterases and then to dichrlorofluorescein (DCF) interacting with different types of ROS. Thus, the degree of DCF fluorescence is dependent on intraerythrocytic ROS levels.
To stain red blood cells treated with TiO 2−x NPs, the erythrocyte suspensions described above were incubated for half an hour in the dark with a H2DCFDA solution in PBS (5 µM). It was prepared from a H2DCFDA stock solution in DMSO (5 mM) [31,42].

Detection of Intracellular Calcium Ion Levels
BD Pharmingen™ Fluo-4 AM (Fluo-4 acetoxymethyl ester) purchased from Becton Dickinson (USA) was used as a calcium ion indicator. This cell-permeant dye is converted to Fluo-4 by intracellular esterases, and its fluorescence increases upon calcium ion binding. To stain cells, erythrocyte suspensions in PBS with 2.5 µM FLUO4 AM were prepared from the dye's 5 mM stock solution in DMSO and incubated in the dark for 30 min. A positive control sample was prepared from red blood cells treated with hydrogen peroxide (0.1 mM), while a negative control was made from the erythrocytes not exposed to the Ca 2+ -sensitive probe. To quantify intracellular calcium ion levels, the percentage of cells with high Fluo-4 fluorescence and MFI values of Fluo-4 were calculated [43].

Analysis of Flow Cytometric Data
BD FACS Canto™ II flow cytometer (Becton Dickinson, USA) was used to detect DCF, annexin V-FITC, and Fluo-4 fluorescence with excitation at 488 nm and emission detection at 525 nm. FlowJo™ (v10, BD Biosciences, USA) and BD FACSDiva™ (Becton Dickinson, USA) were used to process flow cytometric raw data.

Fibroblast Cultures
A complex evaluation of TiO 2−x NPs was conducted on rat embryonic fibroblasts using the previously described set of techniques [44]. The cells were isolated by an enzymatic method using 0.25% trypsin-EDTA (Biowest, France) and were cultivated in the SPL culture flasks (Republic of Korea) with 5% CO 2 at 37 °C in Dulbecco's modified Eagle's medium (DMEM, Biowest, France) with addition of 10% fetal bovine serum (Bio-Whittaker® reagents, Lonza, Belgium) up to the third passage. Every 3 days the medium was changed.
Either 1-TiO 2−x or 2-TiO 2−x NPs were added to the cells next day after seeding for 24 h at concentrations of 10-20-40-100-200 µg/ml. The cells killed by ice-cold ethanol were used as a positive control, while the intact fibroblasts with no added NPs were considered a negative control (Fig. 1b).

Morphological Evaluation of Cells
The size and shape of fibroblasts, the distance between them, and the confluence of monolayer were assessed visually using a Delta Optical IB-100 phase contrast microscope (Poland) equipped with Sigeta MCMOS 3100 3.1MP camera.

Cell Viability Assays
Cell viability of fibroblasts was determined using the neutral red uptake (NRU) and MTT assays as described earlier [45][46][47]. To perform the NRU assay, the cells were incubated with TiO 2−x NPs in 96-well plates, washed with PBS (pH 7.4).The neutral red-containing medium was added to the walls for 3 h. Then 96-well plates were washed with PBS, and the neutral red was extracted using 70° ethanol with 1% acetic acid. Absorbance was measured at 540 nm.

Scratch Assay
To evaluate the cell migration, the scratch assay was conducted [48]. Briefly, the confluent cell monolayer was scratched by a plastic pipette tip in 24-well plates. The plates were washed by PBS, and TiO 2−x NPs were added at the concentrations mentioned above. Every 24 h, the width of areas lacking cells was measured in 5 regions of each well.

Statistical Analysis
The Kruskal-Wallis and multiple comparison Dunn's tests were used to obtain biostatistical data represented as the median and 25th-75th percentiles. Calculations were carried out via GraphPad Prism 5.0 software (USA).

Characterization of Nanoparticles
TEM images reveal that 1-TiO 2−x and 2-TiO 2−x NPs were approximately 5 nm in size (Fig. 2), whereas XPS data provide evidence on the different Ti 3+ (Ti 2+ )/Ti 4+ ratios in their crystal lattice (see Fig. 1S, Supporting Information). In 1-TiO 2−x NPs, about half of titanium atoms were found to be in the lower valence states: 54% Ti 4+ , 38% Ti 3+ , and 8% Ti 2+ . The relative amount of Ti 4+ and Ti 3+ in the crystal lattice of 2-TiO 2−x NPs was calculated to be 63% and 37%, respectively, while Ti 2+ ions were not detected in this sample.

Cell Shrinkage
Erythrocyte forward scatter parameter was used to assess cell shrinkage. Low FSC signaling, which is typical for shrunken cells, allowed identifying shrunken cells (FSClow erythrocytes). Exposure to TiO 2−x NPs at concentrations of 100 mg/L and below did not statistically significantly increase the number of FSC-low cells suggesting no impact on cellular morphology. Higher concentrations of nanoparticles promoted cell shrinkage (Fig. 3).

Cell Membrane Scrambling
Changes in phosphatidylserine translocation and hence cell membrane phospholipid asymmetry in erythrocytes treated with TiO 2−x NPs were analyzed by the degree of annexin V binding. Incubation of cells at concentrations of 100 mg/L and above for both TiO 2−x NPs promoted the cell membrane scrambling associated with phosphatidylserine externalization. This was judged by an increase in the percentage of annexin V-positive cells and MFI values of annexin-FITC. Of note, 1-TiO 2−x NPs statistically significantly increased MFI values of annexin-FITC, but an elevation of the amount of phosphatidylserine-displaying red blood cells was statistically insignificant (Fig. 4).

ROS Production
In this study, to verify the intracellular ROS levels in erythrocytes exposed to two types of titanium dioxide nanoparticles, the cells were stained with H2DCFDA. Subsequent flow cytometric analysis revealed DCF fluorescence. DCF-detectable ROS levels were judged by the mean fluorescence intensity (MFI) values of DCF.
Our experimental data suggest that 1-TiO 2−x NPs at concentrations of 10-20-40-100 mg/L did not promote DCF-detectable ROS generation in erythrocytes treated with them for 24 h compared with the basal ROS production in untreated cells. This was verified by statistically insignificant (p > 0.05) changes in MFI values of DCF. However, treatment of erythrocytes with 200 mg/L 1-TiO 2−x NPs statistically significantly (p = 0.0002) increased ROS production, evidenced by a 2.5-fold increase in the studied parameter of DCF fluorescence (Fig. 5).
Exposure to 2-TiO 2−x NPs at all the concentrations used (10-20-40-100-200 mg/L) did not result in any statistically significant (p > 0.05) increase in MFI of DCF (Fig. 5) compared with the control samples. This indicated no changes in intracellular ROS levels in the red blood cells exposed to 2-TiO 2−x NPs.

Intracellular Ca 2+ Levels
In order to quantify intracellular Ca 2+ levels, erythrocytes were stained with Fluo-4 AM. The threshold concentration that promoted calcium ion elevation inside red

Fibroblast Morphology
Following 24 h after seeding, the cells formed a confluent monolayer, which was found to be somewhat denser following 48 h. The individual cells had a typical spindle shape. After the addition of NPs at any investigated concentration and incubation for 24 h, the size and shape of the cells did not visually differ from the control samples (Fig. 7).

Cell Viability Assays
As illustrated in Fig. 8, both cell viability assays used in this study demonstrated that TiO 2−x NPs did not reduce the viability of fibroblasts at any concentration used. However, the highest concentration of 1-TiO 2−x NPs (200 mg/L) statistically significantly increased the formazan formation indicating an accelerated metabolic activity of fibroblasts without reducing their viability. Figure 9 demonstrates the scratch assay findings used to measure cell migration. The initial width of scratch was measured in all samples (Fig. 9a). The scratches were of the same size (about 0.92 mm) with no statistically significant differences between the samples. Following 24 h, the cell-free gaps were completely covered with fibroblasts in

Discussion
Data from in vitro studies performed on fibroblast cultures indicate low cytotoxicity of TiO 2−x NPs without affecting the lysosomal activity (NRU assay) and slightly increasing the mitochondrial activity in case of 1-TiO 2−x NPs (MTT assays). These data are consistent with other studies that demonstrate good biocompatibility of titanium dioxide nanoparticles [15,[49][50][51]. In addition, TiO 2−x NPs promote the migration rate of fibroblasts. This feature suggests that they can accelerate extracellular matrix regeneration and remodeling improving wound healing, since the scratch assay is widely used as an in vitro model for evaluating wound healing capacities [52]. The detected ability of TiO 2 to accelerate fibroblast migration and wound healing has been reported by several research groups [53,54]. Moreover, their positive effects on wound healing can be partially attributed to antibacterial properties [55]. Our experimental data support the application of TiO 2−x NPs as wound healing agents. However, more studies are required to evaluate the effectiveness and safety of their use. Both titanium dioxide NPs trigger eryptosis accompanied by cell shrinkage, phosphatidylserine exposure, and Ca 2+ influx with the threshold concentration of 100 mg/L. Thus, we believe that eryptosis is more sensitive for detecting the toxicity of at least TiO 2−x NPs than fibroblast tests, and potentially, it can be used as a model for nanotoxicity screening.
The difference in the mechanisms of cytotoxicity against erythrocytes between two TiO 2−x NPs investigated in this study is mainly related to the involvement of ROS-mediated toxicity. In contrast to 1-TiO 2−x NPs, eryptosis induced by 2-TiO 2−x NPs is ROS-independent. It is generally accepted that cytotoxic effects of metal oxide NPs are associated with their tendency to generate excessive amounts of ROS [18,19,22,56,57]. Due to the strong oxidation potential, the excess ROS induced by NPs can result in the damage to biomolecules and cell organelles and lead to protein oxidative modifications, lipid peroxidation, DNA/ RNA breakage, and membrane structure destruction, which further cause necrosis, apoptosis, and mutagenesis [56].
In its turn, the ability of NPs to generate ROS is governed by their physicochemical characteristics, including particle size, surface charge, and chemical composition [57]. For such NPs as ZnO, CeO 2 , TiO 2 , and Ag NPs of specific size, which are deposited on the cellular surface or inside the subcellular organelles, the induction of oxidative stress signaling cascades eventually results in the development of oxidative stress [57].
Metal oxide NPs can contribute to intracellular ROS generation via different mechanisms. One of the possible mechanisms responsible for ROS bursts is a release of metal ions and mixing into redox cycling and chemocatalysis via the Fenton and Huber-Weiss reactions [56]. For oxygendeficient metal oxide NPs, in particular, TiO 2−x, oxygen vacancies compensated by Ti 3+ or Ti 2+ ions are considered to be reactive sites with an effective adsorption of O 2 , H 2 O 2 , H 2 O, and other small molecules and subsequent production of different kinds of ROS with a participation of electrons stored at V O or Ti 3+ and Ti 2+ ions [10].
Thus, the increase in ROS production observed in case of treatment of erythrocytes with 200 mg/L 1-TiO 2−x NPs could be ascribed to the mechanisms discussed above, since red blood cells lack mitochondria, which are major ROS generators in the cells. As mentioned above, 1-TiO 2−x NPs have more defective structure with a higher amount of titanium ions in lower oxidation states (i.e., higher amount of reactive sites) compared to 2-TiO 2−x nanoparticles, which could explain their stronger toxicity.
However, it should be noted that the concentration of 200 mg/L is 20-fold higher than that reported by Yu and co-workers [56] for 10 nm TiO 2 (10 µg/mL). We suppose that such a good biocompatibility of synthesized 1-TiO 2−x and 2-TiO 2−x NPs is associated with their ROS scavenging ability, as well. Based on the comparison of Ti 4+ /Ti 4+ and Ti 2+ /Ti 3+ redox potentials with those for some ROS decomposition reactions, we concluded that such process is also possible and predominated up to certain NPs concentrations in cells. This fact is very permissive for biomedical applications of the synthesized TiO 2−x NPs and should be studied in detail.
There is accumulating evidence that ROS mediate Ca 2+ influx in eryptosis, while calcium ions activate Gardos channels providing K + efflux and cell shrinkage. Moreover, elevation of Ca 2+ levels modifies the activity of flippase and scramblase to promote phosphatidylserine externalization [58]. However, in this study, Ca 2+ influx promoted by TiO 2−x NPs is not associated with ROS overproduction suggesting the involvement of ROS-independent mechanisms of Ca 2+ entry. However, more efforts should be made to assess the molecular mechanisms of TiO 2−x NPs-induced eryptosis, e.g., the role of eryptosisassociated signaling pathways, including protein kinase C, p38 MAPK, and casein kinase 1α.

Conclusions
This study reveals that TiO 2−x nanoparticles are characterized by quite good biocompatibility against fibroblasts. The stimulatory effects of TiO 2−x nanoparticles on fibroblast motility make them promising wound healing agents. TiO 2−x nanoparticle-mediated induction of eryptosis is attributed to Ca 2+ -dependent mechanisms.