Effects of nano-titanium dioxide on calcium homeostasis in vivo and in vitro: a systematic review and meta-analysis

Abstract With the extensive application of titanium dioxide nanoparticles (TiO2 NPs), their impacts on calcium homeostasis have aroused extensive attention from scholars. However, there are still some controversies in relevant reports. Therefore, a systematic review was performed followed by a meta-analysis to explore whether TiO2 NPs could induce the imbalance in calcium homeostasis in vivo and in vitro through Revman5.4 and Stata15.0 in this research. Fourteen studies were included through detailed database retrieval and literature screening. Results indicated that the calcium levels were significantly increased and the activity of Ca2+-ATPase was significantly decreased by TiO2 NPs in vivo and in vitro. Subgroup analysis of the studies in vivo showed that TiO2 NPs exposure caused a significant increase in calcium levels in rats, exposure to large-sized TiO2 NPs (>10 nm) and long-term (>30 days) exposure could significantly increase calcium levels, and the activity of Ca2+-ATPase showed a concentration-dependent downward trend. Subgroup analysis of the studies in vitro revealed that intracellular calcium levels increased significantly in animal cells, exposure to small-sized TiO2 NPs (≤10 nm) and high concentration (>10 μg/mL) exposure could induce a significant increase in Ca2+ concentration, and the activity of Ca2+-ATPase also showed a concentration-dependent downward trend. This research showed that the physicochemical properties of TiO2 NPs and the experimental scheme could affect calcium homeostasis


Introduction
Titanium dioxide nanoparticles (TiO 2 NPs) are a kind of metal oxide nanoparticles with particle size <100 nm (Sahoo et al. 2007). TiO 2 NPs have special surface effect, particle size effect, high chemical activity, excellent heat resistance and corrosion resistance, which can be used as excellent catalysts, catalyst carriers and adsorbents. They are also widely used in food, cosmetics, pigments, air purification, sewage treatment and biomedicine (Shahin and Mohamed 2017). Meanwhile, there are also many potential safety hazards.
The widespread application of TiO 2 NPs in various fields leads to their massive release into the natural environment, which increases the risk of exposure to humans (Sahoo et al. 2007). TiO 2 NPs can enter into body through the respiratory tract, gastrointestinal tract, skin contact and drug injection, and they are also easy to accumulate in brain (Vasantharaja and Ramalingam 2018), lung (Sagawa et al. 2021), kidney , placenta (Hong, Zhou, et al. 2017) and other organs or tissues due to their small particle size (Hong, Yu, et al. 2017). Existing studies have demonstrated that TiO 2 NPs can lead to the changes of calcium levels, oxidative stress, inflammatory response, tumorigenesis and cell apoptosis in vivo and in vitro (Feng et al. 2014;Wu et al. 2014;Shakeel et al. 2016).
Calcium ion concentration is the main influencing factor of cellular calcium homeostasis, and intracellular calcium ion concentration plays an important moderating role in differentiation, metabolism, intracellular transport, secretion, signal transduction and gene expression of cells (Brini et al. 2013). For example, calcium can affect the cell cycle by activating protein kinase C (PKC), calcium calmodulin-dependent protein kinases (CaMK) and mitogen-activated protein kinases (MAPK) (Berridge et al. 2003). Many environmental toxicants can directly or indirectly influence the concentration of Ca 2þ by promoting Ca 2þ influx, releasing Ca 2þ from cells, inhibiting the normal transport of Ca 2þ or preventing Ca 2þ outflow from cells (Zhang and Qiao 2019).
Studies have shown that TiO 2 NPs can lead to the changes of calcium ion concentration in bodies (Gao 2013;Wang 2014). Some scholars found that TiO 2 NPs can cause an increase in calcium ion levels in kidney tissue (Zhao et al. 2010), ovary and placenta (Hong, Zhou, et al. 2017) in animal experiments. It was reported that Ca 2þ concentration in cerebral cortex cells of poisoned rats only significantly increased in the high-concentration TiO 2 NPs exposure group, while there was a slight upward trend in the lowand medium-concentration TiO 2 NPs exposure groups (Feng et al. 2014). In vitro studies have also shown that the intervention of TiO 2 NPs can induce an increase in calcium levels in human lens epithelial cells (HLE B-3) and rat primary hippocampal neurons (Wu et al. 2014;Hong et al. 2015). However, the calcium ion concentration in primary normal human bronchial epithelial (NHBE) cells showed a significant downward trend after being exposed to TiO 2 NPs for 24 h (Kim et al. 2020). In addition, TiO 2 NPs had no significant effect on calcium levels in human differentiated HL60 neutrophil-like cell line (Johnston et al. 2015).
It can be seen that due to the differences in the selected research objects and research methods, there are conflicting results on the influences of TiO 2 NPs on calcium homeostasis. Therefore, whether TiO 2 NPs could cause an imbalance in calcium homeostasis in vitro and in vivo was investigated by the systematic review and meta-analysis in this research.

Search strategy
The studies were searched on PubMed, Embase, Web of Science, Scopus, CNKI, Wan Fang, VIP and Sinomed databases using the keywords 'titanium dioxide or TiO 2 ' and 'nano or nanoparticle or nano-sized' and 'calcium or Ca 2þ or calcium ions.' All studies, published in Chinese and English from 1 January 2000 to 19 March 2022, were included in this meta-analysis. This research protocol has been registered on PROSPERO, and the registration number is CRD42022331485.

Inclusion and exclusion criteria
The inclusion criteria of this study were as follows: (1) randomized controlled human cell or murine studies; (2) gender were unrestricted; (3) the control group was the blank control, the experimental group was the TiO 2 NPs exposure group, and the changes in calcium homeostasis were taken as the results.
Exclusion criteria: (1) without negative or positive controls; (2) non-original studies, such as reviews, comments, editorials, expert opinions or meeting abstracts; (3) the original article that cannot be obtained; (4) reports unrelated to calcium homeostasis; (5) without describing the physicochemical characteristics of TiO 2 NPs (such as particle size, crystal form or specific surface area); (6) combined treatments or composite materials; (7) the size of TiO 2 materials was not nanoscale (>100 nm); (8) overlapping publications or duplicate papers.

Paper quality assessment
In this research, the assessment of the included experiments in vitro was conducted by ToxRTool data quality reliability evaluation criteria, which were issued by the European Center for Validation of Alternative Methods (Schneider et al. 2009). And the quality evaluation of the included studies in vivo was conducted by the SYRCLE risk of bias tool, which was published by the Systematic Review Center for Laboratory Animal Experimentation (Hooijmans et al. 2014). The quality of the studies was independently assessed by two researchers (Yaqian Yang, Yiman Zhao), and a third reviewer (Guanling Song) would make a judgment if there was any dispute.

Data extraction
In this study, two researchers independently extracted the following information from each included study: (1) basic information: title, first author and publication year; (2) characterizations of nano titanium dioxide materials: crystal form (anatase, rutile, plate titanium and mixed form), particle size and specific surface area; (3) characteristics of subjects: species (rat, mouse, guinea pig, human cell, etc.), gender and organ; (4) interventions: exposure route, concentration and time; and (5) outcome indexes: calcium concentration and Ca 2þ -ATPase activity in vivo and in vitro. The outcome index results were extracted in the form of mean ± standard deviation (SD). Two researchers compared the data and resolved the disputes through discussion. Finally, Professor Guanling Song made the decision.

Statistical analysis
In this research, standard mean difference (SMD) was used as an effect size indicator, and data analysis was performed with the Review Manager Version 5.4 and Stata 15.0. SMD ¼ 0 means that there is no difference between the exposure and the control groups, while SMD > 0 indicates that the exposure group can cause an increase in calcium levels compared with the control group. The selection of the effect model was based on the heterogeneity that was analyzed by I-squared (I 2 ) statistic. The I 2 values of >50% were considered to be heterogeneous, and the random-effect model was selected to calculate the SMD and its 95% confidence interval (95% CI). Otherwise, the fixed-effect model was selected.
To determine the source of the heterogeneity, the subgroup analysis was carried out. And the subgroup analysis of studies in vivo was based on species (mice, rats), the exposure route of TiO 2 NPs (gavage, abdominal injection), particle size ( 10, >10 nm), the exposure time ( 30, >30 days) and the exposure concentration ( 50, >50 mg/kg). While the subgroup analysis of studies in vitro was based on the source of cells (animal, human), the particle size of TiO 2 NPs ( 10, >10 nm), the exposure time ( 12, >12 h) and the exposure concentration ( 10, >10 lg/mL). In addition, the funnel plot was used to explore the publication bias. Sensitivity analysis was implemented by gradually eliminating individual studies, which could evaluate whether one research affected the outcome. Egger's test was performed to analyze whether publication bias existed when the number of studies was more than 10 (Zhou et al. 2020). Two-tailed tests were used in this research and p < 0.05 was considered statistically significant.

Literature search
In this research, a total of 3397 eligible records were obtained through searching the PubMed, Web of Science, Embase, Scopus, SinoMed, CNKI, Wan Fang and VIP databases, 1072 of which were excluded because of duplicates. Based on the titles and abstracts of the remaining 2325 records, reviews, studies and conference articles unrelated to TiO 2 NPs or calcium homeostasis were excluded, and the remaining 115 records were downloaded. After reading the full text, 7 in vivo studies and 7 in vitro studies met the inclusion criteria and were finally included in this meta-analysis. The detailed literature screening process is shown in Figure 1.

Quality assessment
The quality of the 14 included studies was evaluated in this meta-analysis. Among them, 7 in vitro studies (supplemental Table S1) whose scores were presented as total quality scores (TQS) were assessed through the ToxRTool data quality reliability evaluation criteria, while 7 in vivo studies (supplemental Table S2) were assessed through the SYRCLE risk of bias tool. Two researchers evaluated the quality of the literatures respectively. If the results were controversial, a third researcher was required to review. Although five of seven in vitro studies lacked positive controls, they all had negative or blank controls, and the seven in vitro studies scored 15-18 (supplemental Table S1), which were of high quality. The risk of selection bias and detection bias was unclear in most in vivo studies (supplemental Table S3), mainly due to unspecific randomization methods or partial informational insufficiency, but the overall risk of in vivo studies was low.

Meta-analysis of the effects of TiO 2 NPs on calcium homeostasis in vivo
Results of meta-analysis showed that TiO 2 NPs exposure caused a significant increase in calcium levels in vivo (SMD ¼ 1.37; 95% CI, 0.40-2.35; p < 0.05) compared with the control group ( Figure 2). Results of subgroup analysis revealed that the sources of heterogeneity were species, particle size and exposure time (Table 1). In terms of species, calcium levels in vivo were higher in rats (SMD ¼ 2.67; 95% CI, 1.48-3.86; p < 0.05) than in mice, which indicated that rats were more sensitive to TiO 2 NPs than mice. We also found that exposure to large-sized TiO 2 NPs (>10 nm) (SMD ¼ 2.67; 95% CI, 1.48-3.86; p < 0.05) and long-term (>30 days) exposure (SMD ¼ 3.44; 95% CI, 1.93-4.96; p < 0.05) could significantly increase calcium ion concentration. While there were no significant differences in the indexes of calcium homeostasis in vivo between different exposure routes and different exposure concentrations of TiO 2 NPs groups.

Meta-analysis of the effects of TiO 2 NPs on Ca 2þ -ATPase activity in vivo
We conducted a meta-analysis on the activity of Ca 2þ -ATPase in vivo, and we also performed the subgroup analysis to explore the source of heterogeneity. Meta-analysis results indicated that the activity of Ca 2þ -ATPase in vivo was significantly decreased (SMD ¼ À7.78; 95% CI, À10.05 to À5.51; p < 0.05) ( Figure 3). The results of subgroup analysis illustrated that concentration of TiO 2 NPs was the source of heterogeneity ( Table 2). The SMD values of the concentration of TiO 2 NPs showed that it yielded lower levels at a concentration >50 mg/kg (SMD ¼ À10.60; 95% CI, À13.72 to À7.48; p < 0.05) than 50 mg/kg (SMD ¼ À6.55; 95% CI, À8.85 to À4.24; p < 0.05), which indicated that the activity of Ca 2þ -ATPase in vivo showed a concentration-dependent downward trend after the exposure of TiO 2 NPs. However, there were no significant differences in the index of Ca 2þ -ATPase activity in vivo between different species, exposure time and particle sizes of TiO 2 NPs groups.
Regarding sources of cells, calcium ion concentration in vivo was higher in animal cells (SMD ¼ 4.21; 95% CI, 1.62-6.80; p < 0.05) than in human cells, which indicated that animal cells were more sensitive to TiO 2 NPs than human cells. While the effects of TiO 2 NPs exposure on calcium homeostasis in vitro were similar between different exposure time groups.

Meta-analysis of the effects of TiO 2 NPs on Ca 2þ -ATPase in vitro
Meta-analysis results showed that the activity of cellular Ca 2þ -ATPase was significantly decreased when the cells were exposed to TiO 2 NPs (SMD ¼ À2.27; 95% CI, À3.40 to À1.15; p < 0.05) ( Figure 5). Results of subgroup analysis revealed that exposure concentration of TiO 2 NPs was the main source of heterogeneity (Table 4). The SMD values of the concentration of TiO 2 NPs showed that it yielded lower levels  at a concentration >10 lg/mL (SMD ¼ À3.51; 95% CI, À4.79 to À2.23; p < 0.05) than 10 lg/mL (SMD ¼ À1.01; 95% CI, À1.85 to À0.17; p < 0.05), which indicated that the activity of Ca 2þ -ATPase in vitro showed a concentration-dependent downward trend after the exposure of TiO 2 NPs. However, there were no significant differences in the activity of Ca 2þ -ATPase in vitro between TiO 2 NPs groups of different sizes, cell sources and exposure time.

Sensitivity analysis
In order to evaluate the robustness of these results, we performed the sensitivity analysis using the gradually eliminating individual studies method. All results in vivo and in vitro were distributed on both sides of the midline, and no individual study affected the combind results, which demonstrated that our results were relatively stable (supplemental Figure S1).

Publication bias
The funnel plots for the studies on calcium homeostasis in vivo and in vitro had good symmetry (supplemental Figure  S2), which indicated that there was no publication bias in this research (Egger's test, p > 0.05), and the results of this research were relatively reliable.

Discussion
At present, there are still contradictory opinions on the effect of TiO 2 NPs on calcium homeostasis in vivo and in vitro. In this research, the effects of TiO 2 NPs on Ca 2þ levels and the activity of Ca 2þ -ATPase in vivo and in vitro were evaluated by systematically analyzing the existing studies using a metaanalysis approach. The results of this research showed that the calcium levels were significantly increased, and the activity of Ca 2þ -ATPase was significantly reduced after the exposure of TiO 2 NPs both in vivo and in vitro. Subgroup analysis in vivo indicated that species, particle size and exposure time of TiO 2 NPs significantly influenced Ca 2þ concentration, while subgroup analysis in vitro illustrated that particle size, cell source and exposure concentration of TiO 2 NPs significantly affected Ca 2þ levels. In addition, the activity of Ca 2þ -ATPase in vivo and in vitro was associated with the exposure concentration of TiO 2 NPs.
This meta-analysis indicated that calcium homeostasis in vivo and in vitro could be significantly influenced after the exposure of TiO 2 NPs. Calcium homeostasis is the key to cellular Ca 2þ signal generation and transduction to complete a series of physiological functions (Berridge et al. 2003). Insufficient or excessive calcium content in the body can both affect growth, development and health. For example, Ca 2þ can directly act on nuclear transcription factors, which contributes to the activation and expression of pro-apoptotic genes such as Bax, caspase 3 and caspase 9 and down-regulation of the anti-apoptotic gene Bcl-2, ultimately leading to cell apoptosis (Xue et al. 2010;Jaeger et al. 2012;Li et al. 2013;Deng et al. 2022). Therefore, the maintenance of calcium homeostasis is critical to organismal well-being (Meng et al. 2010;Guo and Luo 2020).
Cellular Ca 2þ levels are mainly regulated by plasma membrane calcium transport and intracellular calcium pools, both of which work together to maintain intracellular calcium homeostasis. The regulation process of Ca 2þ levels involves  the carrier protein Ca 2þ -ATPase and multiple channel proteins, including L-type voltage-gated Ca 2þ channels (L-VGCC), Na þ /Ca 2þ exchanger (NCX), inositol triphosphate receptor (IP3R) and ryanodine receptor (RyR) (Kaplan et al. 2003;Brini 2009;Chen et al. 2012). TiO 2 NPs are known to induce the production of ROS, and ROS, as an important intermediate factor, can indirectly lead to disturbance of calcium homeostasis (Delamere et al. 1991;Zaidi and Michaelis 1999;Zaidi et al. 2009). Meanwhile, Ca 2þ concentration is also affected by other factors, such as the interaction of TiO 2 NPs with biological macromolecules.
After TiO 2 NPs enter the body through the respiratory tract, digestive tract or skin, they will induce cellular Ca 2þ overload by acting on calcium-related channel proteins (Kelly et al. 2008;Chen et al. 2011Chen et al. , 2012. It has been reported that TiO 2 NPs can activate L-VGCC on the cell membrane, allowing extracellular Ca 2þ to flow into the cytoplasm (Chen et al. 2012). Some scholars have also found that TiO 2 NPs can cause the influx of Na þ and the accumulation of cellular Na þ ([Na þ ] i ) due to an increase in the peak amplitude of sodium current (I Na ) (Hong et al. 2015), which may induce NCX to conduct Na þ -Ca 2þ reverse exchange (Xiao et al. 2002;Shattock et al. 2015;Abiko et al. 2016;Harper and Sage 2016;Rose et al. 2020), thus leading to increased intracellular Ca 2þ levels. IP3R mainly exists in non-excitable cell calcium pools such as endoplasmic reticulum (ER) or sarcoplasmic reticulum (SR) (Ferris et al. 1990;Maeda et al. 1991;Windhorst et al. 2012), and it can bind to inositol triphosphate (IP3) and have an effect on calcium concentration (Xu et al. 2007). Studies have shown that IP3R inhibitors inhibit the release of Ca 2þ from the ER by affecting the IP3-IP3R pathway, thereby significantly attenuating the elevated cellular Ca 2þ levels caused by TiO 2 NPs (Foskett et al. 2007;Mikoshiba 2007;Kawaai et al. 2009;Chen et al. 2012;Mikoshiba 2015). When cellular calcium concentration increases, Ca 2þ , as an intracellular signal, can activate the RyR2 channel on the membrane of the SR, releasing a large amount of Ca 2þ from the SR into the cytoplasm, which further augments cellular Ca 2þ levels. This process is called Ca 2þinduced Ca 2þ release (CICR) (Fill and Gillespie 2021;Mu 2013;Gong et al. 2019).
The results of this meta-analysis showed that TiO 2 NPs could reduce the activity of Ca 2þ -ATPase in vivo and in vitro. Ca 2þ -ATPase, an essential enzyme that regulates ion balance, exists in the membranes of cells and organelles such as SR/ ER and mitochondria (Jafarnejad et al. 2015), can catalyze the hydrolysis of ATP to generate energy and drive intracellular Ca 2þ to be pumped out of cells or stored in the intracellular calcium pools. Therefore, Ca 2þ -ATPase plays an important role in maintaining calcium homeostasis in vivo and in vitro.
In fact, TiO 2 NPs affect the activity of Ca 2þ -ATPase through multiple pathways. Among them, oxidative stress is  an important reason why TiO 2 NPs affect the activity of this enzyme (Guo et al. 2013;Xiong et al. 2013;Wang et al. 2018;Alena et al. 2019;Burgoyne et al. 2019). On the one hand, studies have revealed that TiO 2 NPs can increase the level of ROS , which may inhibit the activity of Ca 2þ -ATPase by changing its sulfhydryl groups (Zaidi and Michaelis 1999;Kaplan et al. 2003;Zaidi et al. 2009). On the other hand, excessive ROS will oxidize the membrane lipid bilayer structure of cells and organelles, followed by changes in the composition, structure, fluidity and permeability of the cell membrane and the formation of transient pores, which may also cause a decrease in the activity of Ca 2þ -ATPase (Delamere et al. 1991;Tang et al. 2006;Chen et al. 2012). In addition to ROS, interfering with the production of ATP and promoting the consumption of ATP are also two ways that TiO 2 NPs affect the activity of Ca 2þ -ATPase. Studies have indicated that the exposure of TiO 2 NPs can disrupt the mitochondrial electron transport chain (ETC), resulting in mitochondrial dysfunction, reducing ATP synthesis (Long et al. 2006(Long et al. , 2007Hong et al. 2015) and decreasing the activity of Ca 2þ -ATPase. The reduction of the activity of Ca 2þ -ATPase will lead to a high level of cellular calcium. When cellular calcium overload occurs, the activity of Ca 2þ -activated biological enzymes will abnormally increase, leading to ATP depletion, which can further inhibit the activity of Ca 2þ -ATPase (Dong et al. 2015;Yu et al. 2016;Aulestia et al. 2020).
In addition, TiO 2 NPs can also induce cellular calcium overload in some other ways. For instance, some scholars have found that TiO 2 NPs can interact with intracellular biomolecules such as lipids and proteins, which may trigger a series of adverse reactions in cells, including increased intracellular Ca 2þ (Wang et al. 2018). In addition, TiO 2 NPs may directly combine with the lipid bilayer and alter lipid packing, which can affect the influx of Ca 2þ as well (Jacobson et al. 2015). The possible mechanism of the effect of TiO 2 NPs on calcium homeostasis in vitro is shown in Figure 6.
Since cells are the basic units of organs and tissues, cellular Ca 2þ overload can also lead to increased calcium levels in vivo. Serum alkaline phosphatase (AKP) is known to be a Zn-dependent enzyme involved in the regulation of calcification by binding to interstitial calcium. Studies have shown that metallothionein (MT) can be induced by TiO 2 NPs, and then rapidly bind to Zn 2þ , resulting in a decrease in the concentration of Zn in blood, and Zn deficiency may reduce AKP activity (Hong, Zhou, et al. 2017), which may also increase serum calcium distribution (Yu et al. 2019). In addition, Mg usually competes with Ca to bind to Ca 2þ /Mg 2þ -ATPase, inhibiting the influx of Ca and activating the activity of the enzyme. One in vivo study has found that after TiO 2 NPs enter the blood, liver and brain tissues, they will inhibit the activity of Ca 2þ /Mg 2þ -ATPase by down-regulating the interaction between Ca 2þ and Mg 2þ , and eventually cause calcium homeostasis disorders (Hu et al. 2010).
According to the results of in vivo subgroup analysis on species, the disorder of calcium homeostasis was more likely to be observed in rats than in mice. This indicated that there were differences in the sensitivity of different species to the toxic effects of TiO 2 NPs. Therefore, the differences in species should be fully considered when extrapolating the experimental data from murine to humans. Generally, nanoparticles with a small size have high toxicity. However, we found that TiO 2 NPs with a large size (>10 nm) were more likely to cause an increase in calcium levels in vivo in this meta-analysis. This phenomenon can be explained in two ways. On the one hand, the toxicity may be associated with the absorption of nanoparticles. As described in some studies, the absorption of particles around 20 nm was much higher than other sizes (Bergin and Witzmann 2013;Wang et al. 2009;Jin et al. 2009), and an analytical model created by some scholars showed that the absorption peak of nanoparticles appeared between 20 and 30 nm (Zhang et al. 2009). These results showed that TiO 2 NPs with a particle size larger  than 10 nm, especially when the particle size was about 20 nm, had a higher absorption rate, and they were more likely to cause the disturbance of calcium homeostasis in vivo. On the other hand, in the subgroup analysis on particle size, we found a coincidental phenomenon that the particle sizes of TiO 2 NPs used in the included studies on rats were all larger than 10 nm, while those in the mouse experiments were all smaller than 10 nm. Combined with the results of the subgroup analysis on species, we speculate that the effects of TiO 2 NPs with different sizes on calcium levels in vivo may be changeable when species vary. Furthermore, according to the results of the subgroup analysis on exposure time, long-term (>30 days) exposure could significantly increase calcium ion concentration in vivo. Some researchers also found that the toxic effects of calcium overload induced by TiO 2 NPs are time-dependent (Attia et al. 2013).
Based on the results of the in vitro subgroup analysis, TiO 2 NPs with smaller particle size were more likely to increase the calcium level. This is because nanoparticles with a smaller particle size are more likely to be internalized by cells (Bergin and Witzmann 2013), thus leading to greater cytotoxicity. In this research, apart from size, other physicochemical characteristics of TiO 2 NPs, such as crystal form, specific surface area, zeta potential, coating, cannot be analyzed due to their inadequate description in the included studies. In fact, these physicochemical properties of TiO 2 NPs may also affect calcium homeostasis. The larger specific surface area of TiO 2 NPs increases their reaction area, which is more likely to cause disturbance of calcium homeostasis (Chen et al. 2012). However, if the TiO 2 NPs are coated or their zeta potential is unstable in the medium, and there is a tendency to agglomerate, it will be favorable for NP precipitation, which will reduce their effect on calcium homeostasis (Chowdhury et al. 2012;Carmo et al. 2018). One study showed that compared to anatase or P25, rutile TiO 2 NPs are the most difficult to disturb calcium homeostasis by entering cells through endocytosis (Gitrowski et al. 2014). TiO 2 NPs have been widely used in biomedical fields such as medical implantation, photodynamic therapy, drug carrier, biosensing and antibacterial drugs because of their unique physicochemical properties (Bhullar et al. 2021).
In addition, the higher the exposure concentration, the higher the calcium level in vitro. This indicated that the toxic effects of TiO 2 NPs on calcium levels were concentrationdependent. Moreover, the sensitivity of different cells to TiO 2 NPs varies due to species differences. In the subgroup analysis on cell source, calcium levels were significantly changed in animal cells but not in human cells after treatment with TiO 2 NPs, which suggested animal cells were more sensitive to TiO 2 NPs than human cells.
Surprisingly, in the subgroup analysis on exposure time, the effects of TiO 2 NPs on calcium levels in vitro were not significantly different between different exposure time groups, while some scholars found that the Ca 2þ concentration in NHBE cells increased after being exposed to TiO 2 NPs for 8 h but decreased after being exposed to TiO 2 NPs for 24 h (Kim et al. 2020). This may be because TiO 2 NPs cause NHBE cell death by inducing inflammation and immune responses in a short period of time (Yaghi et al. 2010), but over time, the more pronounced agglomeration effect of nanoparticles limits their internalization and reduces the toxicity of TiO 2 NPs (Bae et al. 2011).
According to the subgroup analysis on the activity of Ca 2þ -ATPase in vivo and in vitro, we found that only exposure concentration affected the effect of TiO 2 NPs on the activity of this enzyme. With the increasing exposure concentration of TiO 2 NPs, the activity of Ca 2þ -ATPase in vivo and in vitro gradually reduces, showing a concentration-dependent downward trend.

Limitations
The results of this study still have certain limitations. First of all, after the full-text screening, some results of calciumrelated reports were excluded because they were not presented in the form of quantitative results, and the data could not be extracted. Therefore, fewer studies were included in the research in the end. Secondly, some physical and chemical characteristics of TiO 2 NPs in the literature included in this meta-analysis, such as material crystal form, specific surface area and zeta potential, were partially missing. In addition, this study did not delve into the effect of TiO 2 NPs on calcium levels in intracellular calcium pools. In future research, it can contribute to better elucidating the effects of TiO 2 NPs on calcium homeostasis if the above issues can be further dissolved.

Conclusions
The results of this research showed that TiO 2 NPs exposure could significantly increase calcium levels and greatly reduce the activity of Ca 2þ -ATPase, which might lead to the disturbance of calcium homeostasis in vivo and in vitro. The calcium homeostasis in vivo was significantly affected by the particle size of TiO 2 NPs, exposure time and species. The calcium homeostasis in vitro was greatly influenced by the particle size of TiO 2 NPs, cell source and exposure concentration. In addition, the exposure concentration of TiO 2 NPs was the only factor affecting the activity of Ca 2þ -ATPase in vivo and in vitro. The results of this study can provide a theoretical supplement for related studies on the effects of TiO 2 NPs on calcium homeostasis, providing a certain reference for the safe and reasonable use of TiO 2 NPs in the future.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was supported by the National Natural Science Foundation of China [21966027,81560536].