Fluoride induces endoplasmic reticulum stress in ameloblasts responsible for dental enamel formation

Abstract

The mechanism of how fluoride causes fluorosis remains unknown. Exposure to fluoride can inhibit protein synthesis, and this may also occur by agents that cause endoplasmic reticulum (ER) stress. When translated proteins fail to fold properly or become misfolded, ER stress response genes are induced that together comprise the unfolded protein response. Because ameloblasts are responsible for dental enamel formation, we used an ameloblast-derived cell line (LS8) to characterize specific responses to fluoride treatment. LS8 cells were growth-inhibited by as little as 1.9-3.8 ppm fluoride, whereas higher doses induced ER stress and caspase-mediated DNA fragmentation. Growth arrest and DNA damage-inducible proteins (GADD153/CHOP, GADD45alpha), binding protein (BiP/glucose-responsive protein 78 (GRP78), the non-secreted form of carbonic anhydrase VI (CA-VI), and active X-box-binding protein-1 (Xbp-1) were all induced significantly after exposure to 38 ppm fluoride. Unexpectedly, DNA fragmentation increased when GADD153 expression was inhibited by short interfering RNA treatment but remained unaffected by transient GADD153 overexpression. Analysis of control and GADD153(-/-) embryonic fibroblasts demonstrated that caspase-3 mediated the increased DNA fragmentation observed in the GADD153 null cells. We also demonstrate that mouse incisor ameloblasts are sensitive to the toxic effects of high dose fluoride in drinking water. Activated Ire1 initiates an ER stress response pathway, and mouse ameloblasts were shown to express activated Ire1. Ire1 levels appeared induced by fluoride treatment, indicating that ER stress may play a role in dental fluorosis. Low dose fluoride, such as that present in fluoridated drinking water, did not induce ER stress.

Fig. 1.. Toxic and antiproliferative effects of NaF treatment on LS8 cells.

A, limiting dilutions of LS8 cells were seeded into culture flasks, allowed to adhere for 18 h, and treated for 24 h with the indicated concentrations of NaF. After 8–9 days, the resulting colonies were stained and counted, and percent cell survival was calculated (number treated/untreated colonies) × 100. Colonies from three flasks were counted for each experimental treatment group, and three separate experiments were performed. Error bars represent the S.D. B, LS8 cells were seeded into 96-well plates and treated for either 24 or 48 h with the indicated concentrations of NaF. Reduction of MTT to an insoluble formazan dye by mitochondrial enzymes was quantified for each well by A₅₅₀ measurements, and the results were used to calculate percent cell proliferation (treated A₅₅₀/untreated A ₅₅₀ × 100). Six wells were assayed for each experimental treatment, and three separate experiments were performed. Error bars represent the S.D.

Fig. 2.. Role of caspases in NaF induced LS8 cell death.

Cells were pretreated or not with 50 μm z-VAD for 1 h followed by 48 h of treatment with 2 mm NaF. A, adherent cells were assayed for DNA fragmentation by an ELISA-based TUNEL assay that quantifies DNA strand breaks by measuring bound peroxidase activity (A₄₀₅). Two wells were assayed for each experimental treatment, and three separate experiments were performed. Error bars represent the S.D. Note that treatment of LS8 cells with NaF generated significant levels of DNA fragmentation that were almost completely eliminated by pretreatment with z-VAD. Cont, control. B, trypan blue dye exclusion was performed to obtain numbers of live cells after NaF treatment. Percent viability was calculated by counting live cells (number treated/number untreated × 100). Two wells of a 24-well plate were analyzed for each experimental treatment, and 3 separate experiments were performed. Error bars represent the S.D. Note that z-VAD did not significantly protect LS8 cells from the toxic effects of 2 mm, 48 h NaF treatment.

Fig. 3.. Treatment of LS8 cells with NaF induces the ER stress response.

A, representative Northern blots with 10 μg total RNA/lane. Treatment times and doses are indicated above each lane, and the identity of the mRNA assayed is listed beside each blot. The same blot was stripped and reprobed for each mRNA assayed. Treatment of LS8 cells with tunicamycin (Tun) served as the ER stress response positive control (cont). With the exception of the β-actin control, the stress response genes assayed displayed increased expression in a time- and dose-dependent manner after NaF treatment. B, Northern blot analysis of LS8 gene expression as a function of pretreatment with z-VAD (Z). The same filter was stripped and reprobed for each mRNA assayed. Tunicamycin (T) served as the positive control. Note that the absence of DNA fragmentation did not affect NaF-mediated gene transcription. C, Western blot of protein (30 μg) isolated from LS8 cell nuclei after 24 h of treatment with NaF (2 mm) or tunicamycin (0.5 μg/ml). Both GADD153 and the active form of Xbp1 were induced by each treatment. D, immunohistochemical staining of LS8 cell nuclei with a monoclonal antibody specific for GADD153. Cells were treated or not with 2 mm NaF for 24 h as indicated. Panels to the left were stained with both primary and secondary antibodies (Ab). Panels to the right were stained with secondary antisera only. Immunolabeling detected strong expression of GADD153 in the NaF-treated cells.

Fig. 4.. Treatment of LS8 cells with NaF activates the expression of CA-VI type B.

A, reverse transcription-PCR analysis for identification of CA-VI mRNA splice variant. Primers specific for type A (encodes a signal sequence) or type B (encodes no signal sequence) demonstrated the activation of only the type B variant after treatment with NaF (2 mm, 24 h) or tunicamycin (Tun, 0.1 μg/ml, 24 h). cont, control. B, Northern blot (10 μg of total RNA/lane) time-course analysis of CA-VI type B expression. Expression peaked after 24 h of 2 mm NaF exposure.

Fig. 5.. Attenuation of GADD153 expression increases NaF-induced DNA fragmentation in LS8 cells, whereas transient overexpression has little effect.

A, representative Western blots of GADD153 expression. Top, pretreatment with GADD153 siRNA (NaF, +) reduced the NaF-induced GADD153 up-regulation (NaF, −) to levels approximately equivalent to control cultures pretreated with nonspecific siRNA-only (cont, −). Each lane contains 30 μg nuclear protein (top). Transient transfection of either 0.75 μg (+, middle lane) or 1.5 μg (+, right lane) CMV-GADD153 vector dramatically increased GADD153 expression over that of cells treated with empty CMV vector alone (−, left lane). Each lane contains 5 μg of whole cell protein (bottom). B, ELISA-based TUNEL assays for quantification of DNA strand breaks. NaF-induced DNA fragmentation of cells pretreated with GADD153 siRNA (NaF, +) demonstrated significantly more strand breaks (left panel) than those treated with nonspecific siRNA and NaF (NaF, −) or with siRNA alone (cont, −). Results similar to the NaF treatments were obtained with 0.1 μg/ml tunicamycin (Tu) treatment for 24 h (right panel). C, ELISA-based TUNEL assay for quantification of DNA strand breaks. Cells were transfected with either empty CMV vector (−) or CMV-GADD153 vector (+) and were untreated (cont) or were treated with NaF (NaF). Regardless of NaF treatment, GADD153 overexpression had little to no effect on DNA fragmentation. All NaF treatments were 5 mm for 24 h, and for all TUNEL assays, two wells of a 96-well plate were analyzed for each experimental treatment, and three separate experiments were performed. Error bars represent the S.D.

Fig. 6.. Caspase-3 activity is responsible for increased levels of NaF-mediated DNA fragmentation in GADD153^−/− cells versus controls.

Mouse embryonic fibroblasts with (GADD153^−/−) or without (GADD153^+/+) homozygous GADD153 gene deletion were pretreated or not with either 50 μm z-VAD (top panel) or 2 μm z-DEVD (bottom panel) for 1 h followed by 24 h of treatment with 5 mm NaF. Adherent cells were assayed for DNA fragmentation by the ELISA-based TUNEL assay. Three wells were assayed for each experimental treatment, and three separate experiments were performed. Error bars represent the S.E. of the mean. Note that significantly more NaF-induced strand breaks occurred in the GADD153^−/− cells than in the control cells and that inhibition of caspase-3 with z-DEVD eliminated this difference. Cont, control.

Fig. 7.. High dose fluoride in drinking water causes increased enamel organ apoptosis and apparent induction of Ire1.

A, in situ TUNEL assay performed after 3–4 weeks of exposure to 150 ppm fluoride delivered ad libitum in drinking water. Untreated control secretory stage enamel organ with healthy columnar-shaped ameloblasts (left panel, bracket). Fluoride-treated secretory stage enamel organ (right panel). Note that the fluoride-treated enamel organ morphology was disrupted since no healthy ameloblasts can be observed. Also, several apoptotic cells were observed where the ameloblasts are normally located (right panel, bracket). Cont, control. B, LS8 cells were treated (right panel) or not (left panel) with 2 mm NaF for 24 h and processed and stained by immunohistochemical methods for the presence of active Ire 1. Note that staining was more intense in the rounded NaF treated cells when compared with the oval untreated cells. C, immunohistochemical staining for active Ire1 after 3–4 weeks of exposure to 75 ppm fluoride delivered ad libitum in drinking water. Shown is untreated maturation stage ameloblasts demonstrating low level staining for active Ire1 (left panel, bracket). Fluoride treated ameloblasts demonstrating increased staining for Ire1 (right panel, bracket). Note that some ameloblasts stained more strongly than others (bars, 50 μm). D, immunohistochemical staining of pancreatic islet cells with the same antisera specific for active Ire1 (left panel). Another mouse incisor section demonstrating active Ire1 staining after 3–4 weeks of exposure to 75 ppm fluoride delivered ad libitum in drinking water is shown.