1. Introduction
Fluoride therapy is the main method against dental caries (Centers for Disease Control Prevention, 2001; Iheozor-Ejiofor et al., 2015), and the fluoridation of communities’ water systems is widely used due to its efficiency and cost-effectiveness (Sampaio and Levy, 2011). Since the end of the 1980s, studies report that fluoride (F) controls caries acting in the oral cavity, in contact with the teeth by favorably interfering in de- and remineralization processes, rather than acting systemically (Buzalaf and Whitford, 2011; Pessan et al., 2011). This has led to a debate about the necessity of compulsory F exposure through the water supply. This is supported by the presence of F in different sources, such as food, beverages (tea, wine and milk), infant formula, and oral care products (Levy et al., 1998, 2001, 2003; Franzman et al., 2004; Ismail and Hasson, 2008; Hujoel et al., 2009; Wong et al., 2010). The safe F dosage for humans is around 0.7 to 3 mg/day for children and adults, respectively, and the combination of different sources results in overexposure that is related to health issues (Aoun et al., 2018; Buzalaf, 2018).
Although the main side effects due to F overexposure are dental and skeletal fluorosis (Ten Cate and Buzalaf, 2019), in the last decade a few studies showed neurological damage, such as cognitive decline and lower IQ scores (Choi et al., 2012, 2015; Grandjean and Landrigan, 2014; Malin and Till, 2015). However, such results remain controversial (Duan et al., 2018; Guth et al., 2020; Seddek and Ghallab, 2020). Recently, Johnston and Strobel (2020) suggested that a positive correlation between high F levels and low IQ does not mean a causal association. In the same way, our group reported in a systematic and meta-analysis review that F exposure under therapeutical does not cause neurological damage, while studies showing overexposure to high doses showed low levels of evidence mainly due to inappropriate methodologies and high bias (Miranda et al., 2021).
While studies with F-exposed humans seem to be controversial, animal models studies suggested alterations of the central and peripheral nervous system (Niu et al., 2009; Pereira et al., 2011; Melo et al., 2017; Dionizio et al., 2018, 2020), with structural damage in areas related to motor and sensory control, such as the hippocampus, motor cortex and amygdala (Bhatnagar et al., 2002; Pan et al., 2015). In vitro studies reported changes in central nervous system (CNS) cells related to oxidative stress, cell death, energy metabolism and DNA damage, both at neuronal and glial cell lines in humans and rodents (Zhang et al., 2007, 2008; Shuhua et al., 2012; Qian et al., 2013; Ghasemi et al., 2018; Puty et al., 2021). These results highlight the need for further investigations aimed to understand the mode of action of F in the CNS and to identify possible molecular targets of toxicity.
Furthermore, in vitro and in vivo experiments with glial cells points to it as a possible target for F toxicity with a central role in impairments of neural development (Rasband, 2016; Jäkel and Dimou, 2017) due to the involvement of these cells in CNS homeostasis as well as in neuronal cross-talk (Aschner, 2000; Aschner et al., 2003; Sidoryk-Wegrzynowicz et al., 2011; Jäkel and Dimou, 2017). As it is well known when the CNS is a target of xenobiotics, glial cells act as the primary line of cellular defense to protect the neurons (Aschner et al., 2002; Shanker et al., 2003; Noguchi et al., 2013; Sidoryk-Wegrzynowicz and Aschner, 2013; Ishihara et al., 2019). In this scenario, glial cells could be used as indicators of CNS drug-induced toxicity with molecular changes in these cells being biomarkers to monitor CNS impairment.
Recently, our group established a single-cultured F overexposure model using two cell lines, human glial-like (U87) and neuronal-like (IMR-32) cells. Once F toxicity is dose- and time-dependent, we simulated a continuous exposure for 10 days using a low and high F concentration (0.095 and 0.22 mg/ml) based on what it is usually found in plasma samples from people living in areas of endemic fluorosis. Our results showed that only exposure to 0.22 μg/ml induced signs of toxicity in U87 glial-like cells, due to lower cell viability, changes in cell energy metabolism, decreased reduced glutathione/oxidized glutathione (GSH/GSSG) ratio, and DNA damage (Puty et al., 2021). In the present study we aimed to show the molecular targets via gene and protein modulation in the human glial-like (U87) cells using the same F concentrations described above to provide molecular insight about F toxicity on glial-like cells. For that, we examined molecular changes underlying F-exposed U87 glial cells using the global gene and protein expression profiles. The biological targets provided by gene ontology and molecular pathways were also examined to give a general overview of cell responsiveness to F toxicity.
4. Discussion
Recent studies raised the debate about F-induced toxicity in the human CNS (Li et al., 2010; Flores-Méndez et al., 2014). Despite the lack of evidence supporting F as a potentially neurotoxic compound, studies suggested that F may be able to damage neurons, microglia, and glia (Zhang et al., 2007; Chen et al., 2017; Puty et al., 2021). Several mechanisms were pointed out as the F mode of action that results in neurotoxicity. The oxidative stress plays a role due to increasing levels of reactive oxygen species, and decreased levels of enzymatic and non-enzymatic antioxidants. Neuroinflammation, due to increasing levels of transcription factors and pro-inflammatory substances, and disorders in energy metabolism and mitochondrial dysfunction (Zhang et al., 2007, 2015; Chen et al., 2017; Yang et al., 2018). However, it is important to note that those studies often used high F concentrations which do not mimic human exposure.
We performed a large transcriptomic and proteomic analysis on a human glial-like cell model (U87) using a concentration that is similar to F plasma levels from individuals living in endemic areas for fluorosis. To our knowledge, this is the first study to investigate the effects of F in glial-like cells using omics analysis. Our results point to deep changes in the transcriptomic and proteomic profiles in a non-concentration-dependent manner, suggesting a major F effect on cell metabolism, death control, and cytoarchitecture. Several metabolic pathways, such as phosphate, macromolecule, and protein metabolic regulation were impaired, as well as proteins related to cytoarchitectures such as those belonging to tubulin and actin families.
We used F concentrations (0.095 and 0.22 μg/ml) that were close to those found in plasma samples of people living in areas of endemic fluorosis. Even when considering that the literature showed wide variation in human plasma F levels (0.017 to 1.43 μg/ml), depending on factors such as geographical region, cultural habits, and health status (Sener et al., 2007; Rafique et al., 2012; Fernando et al., 2020), it is important to highlight that the concentration of 0.095 μg/ml F is most likely to be observed in human plasma samples based on an animal model of F toxicity (Pereira et al., 2016; Dionizio et al., 2018; Miranda et al., 2018). In fact, our group has previously shown that plasma levels around 0.095 and 0.22 μg/ml, in animal models, are related to changes in the central nervous system. In this study, animals were exposed to 50 μg/ml F for 15 days and presented plasma F levels close to the concentration of 0.095 μg/ml triggering oxidative stress and proteomic modulation of the hippocampus (Ferreira et al., 2021). In previous work, we showed the F harm potential on U87 glial cells after 3, 5, and 10 days of exposure using 0.095 and 0.22 μg/ml. Signs of F toxicity were seen only after exposure to 0.22 μg/ml regarding a small decrease in cell viability, changes in energy metabolism, and cell cytoarchitecture (see Puty et al., 2021).
In the global microarray and protein profile analysis performed here, our results revealed that both 0.095 and 0.22 μg/ml F were able to modulate gene and protein networks of U87 glial-like cells. The functional classification of differentially expressed genes modulated by F exposure revealed similarities in the 0.095 and 0.22 μg/ml concentrations, suggesting molecular toxicity via regulation of several pathways related to cell death and metabolism control. The down-regulation of the endomembrane system was the top functional classification, according to GO annotation, modulated by F in a concentration-independent manner. The endomembrane system is a system derived from the endoplasmic reticulum that allows cell compartmentalization and a high degree of cell specialization and it is suggested to be modulated by F in different models of F toxicity (Arensdorf et al., 2013; Tabuchi et al., 2014; Søreng et al., 2018). Our data also showed that both F concentrations were able to modulate the MAPK, ERK1, and ERK2 cascades. These metabolic pathways are related to cellular proliferation/protection response, as well as cell death induction. In addition, the functional classification also identified pathways related to the cellular response to chemical stimuli, suggesting an adaptation mechanism to control F-induced stress over 10 days of F exposure.
MAPK plays an important role in F toxicity in different models, such as silkworm and ameloblast cells (Suzuki et al., 2015; Liu et al., 2021). MAPK up-regulation is strongly associated with different kinds of stress responses, and both MAPK and ERK are activated during the endoplasmic reticulum (ER) stress response (Hung et al., 2004; Li and Holbrook, 2004; Darling and Cook, 2014). The cell death initiated by ER stress after F exposure was demonstrated for different organisms (Kubota et al., 2005; Sharma et al., 2008; Wei et al., 2013; Zhou et al., 2013; Tabuchi et al., 2014) and may be induced by the activating transcription factor (ATF) family (Edagawa et al., 2014). Our data point to a significative modulation of different genes in the ATF4 pathway, as well as genes that are considered to be ATF4 targets. In this way, since ATF4 is involved in several classes of biological function, such as the biosynthesis, folding, and assembly of proteins, metabolism, oxidative stress, and apoptosis, we conclude that there is an important role of the ATF4 in the F exposure mode of action. The ATF4-target genes DDIT3, IL-8, IGFBP-1, ATF-3, and CEBPB were up-regulated after F exposure. The DDIT3 gene was modulated under F exposure on epithelial rat cells (ROE2) as an indicator of cell damage and is considered to be a key regulator of cell death mediated by ER stress (Tabuchi et al., 2014). Studies found that ROE2 cells exposed to F were modulated for EGR1, FOS, and IL6, which were up-regulated in the present study (Zinszner et al., 1998; Tabuchi et al., 2014). The regulation of those genes may be related to pro-cell death pathways, suggesting that concentrations such as the ones used in our study may be potentially harmful to U87 cells.
Another important result of the modulation of the ATF4 pathway was the up-regulation of TRIB1, CHAC1, and SESN2. The over-expression of CHAC1 and SESN2 is associated with human GSH depletion and changes in cell metabolism (Crawford et al., 2015; Garaeva et al., 2016). As we showed in a previous study, F leads to a decrease in ATP levels and cell death without activation of pro-apoptotic death as caspase3/7 (see Puty et al., 2021). We then conjecture that even the F concentrations used here may lead to ER disturbances mediated by the ATF4 pathway, causing physiological stress mediated by decreased levels of ATP and GSH/GSSG, and cell death.
The activation of the TRIB1 gene was associated with a reduction in protein synthesis (Jiang and Wek, 2005; Soubeyrand et al., 2016). Interestingly, our functional analysis revealed a wide modulation of mechanisms for protein synthesis/modification related to the extracellular matrix (see Figure 4). In order to better understand the changes in the protein profile we performed a global proteomic analysis. We identified only 17 up-regulated and 4 down-regulated proteins in 0.095 μg/ml vs. control and only 10 down-regulated proteins in 0.22 μg/ml vs. control. This poor correlation of transcriptomic and proteomic data may be attributed to posttranslational changes in the control of gene expression and the reduction of protein synthesis mediated by ATF4. However, we point out the importance of combining these two methods to investigate the effect of gene/protein changes in cellular and molecular processes after xenobiotics exposure (Udayan and Harihara, 2005). Although the regulation of some proteins was found to be different at both concentrations vs. control, the cellular behavior after F exposure seemed to be similar. We showed two main processes impaired by F exposure on the proteomic analysis: (1) cellular metabolism, via glycolytic processes, NADH regeneration, and gluconeogenesis, and (2) cellular cytoarchitecture, via structural constituents, and the organization of the cytoskeleton and nucleosome assembly. Despite the lack of an exact correlation between genes and protein modulation, the biological functional domains and their corresponding modulated pathways were the same.
Impaired energy metabolism induced by F was reported in a rodent model (Barbier et al., 2010; Pereira et al., 2018; Zuo et al., 2018). Recently, Araujo et al. (2019) showed changes in proteins related to glycolysis and gluconeogenesis in the mitochondria of rats exposed to F concentrations similar to the ones we used [50 μg/ml–since rats metabolize F 5–10 times faster than humans (Dunipace et al., 1995)]. The proposed mechanism of mitochondria impairment follows the decreased expression of fructose-biphosphate aldolase A and glyceraldehyde-3-phosphate dehydrogenase, both down-regulated in the present study, Table 1 (P0475- fructose-biphosphate aldolase A and P04406- glyceraldehyde-3-phosphate dehydrogenase).
We demonstrated alterations in the main cytoskeleton constituents that did not depend on F concentration. Despite the evidence in literature has suggested that the cytoskeleton alterations after F exposure occur in an indirect way via oxidative stress and metabolic changes (Wilson and González-Billault, 2015; Zepeta-Flores et al., 2018), our results point to a direct effect. In a previous study of our group, we showed that at the lower concentration used in the present study (0.095 μg/ml), there was no change in the oxidative stress parameters such as GSH, ROS, lipidic peroxidation and DNA integrity, and levels of cellular ATP (Puty et al., 2021).
The tubulin family is composed of α-, β- and γ-tubulins, and assembly and disassembly changes may have an impact on the morphological organization, cell migration, vesicle trafficking, cell compartmentalization, and lead cells to death (Lepekhin et al., 2001; Hohmann and Dehghani, 2019). The actin isoforms were modulated at 0.095 and 0.22 μg/ml F. Both concentrations led to the downregulation in the following actin isoforms: alpha isoforms (P68133; P68032; P62736), cytoplasmic-2 (P63261) and cytoplasmic 1 (P60709), beta isoforms (Q562R1; Q9BYX7) and gamma isoforms (P63267). However, regarding tubulin, our results pointed to strong differences in the impact of F exposure. While β-tubulins, 2A, 2B, and 3 (Q13885, Q9BVA1, Q13509, respectively) were up-regulated by 0.095 μg/ml, we observed a down-regulation on tubulin beta chain (P07437) and tubulin beta-4A chain (P04350) after 0.22 μg/ml. We also observed different classes of tubulin being expressed only in the control group compared to 0.095 μg/ml, such as tubulin alpha-1B chain (P68363), tubulin alpha-4A chain (P68366), tubulin alpha-3C chain (P0DPH7) and others (see Table 2). The tubulin beta chains 4A, 4B and 8 (P04350, P68371, Q3ZCM7, respectively) for 0.22 vs. 0.095 μg/ml were only expressed at the higher concentration (dramatically reduced at the lower concentration), and the tubulin beta 3 chain (Q13509) was down-regulated at 0.22 μg/ml. There is evidence showing a relationship between increasing beta tubulin levels and subsequent cell death (Weinstein and Solomon, 1992; Alvarez et al., 1998), which could be minimized by the co-overexpression of alpha-tubulin (Weinstein and Solomon, 1992; Alvarez et al., 1998). Besides the widely known effects of microtubule damage on the cell cycle and morphology, our findings suggest that the fluctuation of actin and tubulin expression under F exposure may trigger cell death by microtubule disassembly because the higher F concentration is, the lower the regulation of alpha subunits and higher beta subunits. Those results are in agreement with our previous work where we verified significant spreading and cell density changes for 0.22 μg/ml F followed by increased DNA fragmentation (see Puty et al., 2021). The dynamics can be observed in Figures 5, 6, where interactions in tubulin subunits were observed in two PPI.
In conclusion, our study provides valuable insights about the mode of action of F in U87 glial-like cells covering potential molecular alterations, corroborating the hypothesis of the F impact on cellular energy metabolism. We also showed F-mediated modulation of actin filament constituents, such as cytoskeleton and tubulin, suggesting an emerging target of F toxicity. Since our data were collected after a chronic model of F exposure it is important to point out that the overall gene and protein expression alteration may represent both F induced gene/proteins expression as well as adaptive changes after exposure. Finally, our results indicate that fluoride-induced damage on glial cells may happen only at overexposure to high levels of fluoride while there is no evidence of safety issues related to the optimal fluoride levels.