Amelogenesis, the formation of dental enamel, is well understood at the histomorphological level, but the underlying molecular mechanisms are poorly characterized (Lacruz et al., 2017; Varga et al., 2018). Ameloblasts, derived from the oral epithelium, are known to secrete proteins and Ca²⁺ into the enamel extracellular matrix. They also regulate extracellular pH as the formation of hydroxyapatite crystals generates large quantities of protons that must be neutralized by HCO₃ ⁻ secretion to allow continued crystal growth. Genetic or environmental impairment of transport and regulatory processes (e.g. dental fluorosis) leads to the development of enamel defects such as hypomineralized enamel (Lacruz et al., 2017; Varga et al., 2018).

Ameloblast differentiation and ion transport activities have primarily been investigated by immunohistochemistry and characterization of the enamel phenotypes of knockout mouse models. However, to understand the mechanisms by which environmental factors, such as exposure to high levels of fluoride, affect enamel formation, in vitro model systems are needed (Klein et al., 2017).

Culture of primary enamel organ epithelial cells results in two distinct cell shapes: either polygonal or stellate (DenBesten et al., 2005). Stellate cells, which have a phenotype similar to mouse LS8 cells, grow rapidly in 2D culture but cannot be maintained in 3D Matrigel (Li et al., 2006). Polygonal cells are similar in morphology to both ameloblast-like mouse ALC (Nakata et al., 2003) and rat HAT-7 (Kawano et al., 2002) cells. The primary, enamel-originated cells grown in Matrigel alter their morphology in the presence of extracellular calcium (Li et al., 2006; Yan et al., 2006; Chen et al., 2009; He et al., 2010). However, these primary ameloblast-like cells can be passaged only 3 times, and so have limited use in studies of ameloblast function. When transfected by SV40 or telomerase (He et al., 2010; Zheng et al., 2013b; Zheng et al., 2013c), they maintain ameloblast-like characteristics through approximately eight passages (unpublished results), but do not form spheroid structures in Matrigel (Li et al., 2006). This then suggests that these transfected cells are also not a robust model to study ameloblast function (He et al., 2010; Zheng et al., 2013b; Zheng et al., 2013c). While these early studies offer the promise of advanced technologies to more completely regenerate the enamel organ epithelium, further advances are needed to identify cells that retain characteristics similar to primary cells, and that can be scaled up, expanded and standardized. This in turn will allow studies of tooth formation in vitro, and investigation of the mechanisms responsible for environmental effects on tooth enamel formation (Klein et al., 2017).

Scaling up of stem cells and progenitors can be performed using three different strategies: 1) monolayer cultures, 2) culturing cells on microspheres and 3) organoid/spheroid production (Silva Couto et al., 2020). Monolayer cultivation is the least effective method. To achieve higher yields of cell expansion, we can use microcarriers, which are principally designed for growing adherent cells in suspension culture systems. The application of microcarriers in dynamic culture systems may provide high yields but it is technically difficult to isolate the cells from their microcarriers (Chen et al., 2015; Lam et al., 2017; Cherian et al., 2020). In organoid/spheroid culture, three-dimensional cell aggregates can be produced under either static or dynamic conditions. This approach is thought to more closely mimic the three-dimensional in vivo environment (Wang et al., 2009; Bartosh et al., 2010). An obvious challenge of this strategy is to expand cells under controlled conditions without causing harmful effects on the cells and undesirable cell differentiation (Simoes et al., 2013; Yang et al., 2018a).

In the present study, we have investigated the effects of culture conditions on the three-dimensional growth of HAT-7 cells. These cells, which are derived from rat incisor cervical loop (Kawano et al., 2002), maintain a polygonal shape and exhibit characteristics of maturation-stage ameloblasts, such as the expression of the marker proteins kallikrein-4 and amelotin (Kawano et al., 2002). We have shown previously that HAT-7 cells grown on permeable supports form two-dimensional polarized epithelia which serve as a good experimental model for investigating the molecular physiology of ion transport in ameloblasts (Bori et al., 2016; Racz et al., 2017; Racz et al., 2018; Varga et al., 2018). Furthermore, this cell line has been successfully used in a large number of studies to investigate ameloblast function (Matsumoto et al., 2011; Zheng et al., 2013a; Jalali et al., 2017; Yang et al., 2018b; Takahashi et al., 2019).

In this report we demonstrate that HAT-7 cells can form spheroids similar to those previously obtained in primary cultures of ameloblast-lineage cells. We have compared their capacity to regulate intracellular pH and to show intracellular calcium responses to extracellular stimulation, to our previous findings in 2D culture. To explore the effects of fluoride exposure we have determined the dose at which fluoride exposure disturbs HAT-7 spheroid formation.

In the present study dissociated HAT-7 cells were cultivated within a Matrigel matrix in order to obtain single cell-derived 3D colonies. We found that the combination of Hepato-STIM epithelial cell differentiation medium and Matrigel induced cell expansion and 3D spheroid formation. The cells retained epithelial cell morphology and continued to express both ameloblast-specific and ion transport-specific gene expression. Furthermore, like HAT-7 polarized monolayers, the spheroid cells were able to regulate their intracellular pH and to show intracellular calcium responses to extracellular stimulation. Finally, we have demonstrated that HAT-7 spheroids may serve as a disease model for studying the concentration-dependent damaging effects of fluoride exposure on amelogenesis.

The choice of culturing protocols, culture medium and scaffold serve as important selection factors in epithelial cell expansion and 3D culturing (Bajaj et al., 2014). We initially tested three culture media: control medium (standard DMEM/F12 Ham medium supplemented with fetal bovine serum), differentiation medium (control medium supplemented with CaCl₂ and dexamethasone) and Hepato-STIM medium (also supplemented with fetal bovine serum). It quickly became apparent that the use of Hepato-STIM medium, rather than the other two media, resulted in a far better outcome, both regarding the number and size of the HAT-7 spheroids. Hepato-STIM is a specialised medium consisting of Williams’ E medium supplemented with dexamethasone, insulin, transferrin, selenium, EGF and 1.8 mM Ca²⁺. These components are all important factors in maintaining an epithelial phenotype (Szlavik et al., 2008a; Maria and Tran, 2011; Hegyesi et al., 2015). This medium was originally designed for the culture of hepatocytes, but has since been used successfully to grow both human primary salivary epithelial cells (Szlavik et al., 2008a; Maria and Tran, 2011; Hegyesi et al., 2015) and lacrimal acinar cells (Schonthal et al., 2000; Tiwari et al., 2018).

In the present work we used Matrigel as a scaffold to induce and promote spheroid formation by HAT-7 cells. Matrigel is a poorly defined basement membrane derivative extracted from Engelbreth–Holm–Swarm mouse sarcoma cells, thus its direct application in regenerative medicine is limited (Huang et al., 2020). However, its use is well established in the culture of spheroids and organoids from various tissues, especially those of epithelial origin. This is mainly because of three key factors: first, the Matrigel provides anchoring molecules such as the arg-gly-asp (RGD) sequence for cell adhesion; second, it has optimal stiffness for housing the cells; and third, it contains laminin-111, an important extracellular component that can independently provide biological signals for spheroid/organoid formation and growth (Huang et al., 2020). Using Matrigel, 3D spheroids and organoids have been created from intestinal epithelial cells (Huang et al., 2019), pancreas (Bakhti et al., 2019; Molnar et al., 2020), lacrimal glands (Schonthal et al., 2000; Massie et al., 2018) and salivary glands (Szlavik et al., 2008a; Szlavik et al., 2008b; Maria et al., 2011). Primary culture ameloblast-lineage cells also form 3D spheroids when cultured in Matrigel (Li et al., 2006).

When porcine and human ameloblast-lineage cells (Li et al., 2006; He et al., 2010) are grown in Matrigel, the cells form spherical, acinar-like structures that appeared similar to enamel pearls. Additionally, ameloblast-lineage cells (He et al., 2010) or human embryonic stem cell-derived epithelial cells (Zheng et al., 2013b; Zheng et al., 2013c) pre-cultivated in Matrigel and then cultured together with dental mesenchymal cells result in the formation of tooth-like structures. Finally, when mouse dental epithelial stem cells from the most posterior part of the cervical loop of incisors are isolated, dispersed, and then embedded into Matrigel (Natsiou et al., 2017) spheroids are produced (Natsiou et al., 2017). Our present success in reliably and reproducibly generating 3D HAT-7 spheroids, using Hepato-STIM culture medium and a Matrigel scaffold, are completely in line with those previous observations.

Tight junction proteins ZO-1 and claudins 1, 4, and 8 were all expressed at mRNA level in HAT-7 cells, regardless of whether they were grown on porous Transwell filters or in Matrigel. PCR data normalized to 2D HAT-7 cultivation revealed an increase in Cldn8 while the expression of ZO-1, Cldn1 and Cldn4 claudins were decreased in HAT-7 spheroids. These data are also in accordance with previous studies showing the expression of Cldn1, Cldn4 and Cldn8 in maturation-stage ameloblasts (Inai et al., 2008; Hata et al., 2010). The higher level of Cldn8 expression in the spheroids may reflect the elevated level of paracellular leakage observed when Cldn8 is upregulated (Amasheh et al., 2009).It is important to note that the expression of maturation ameloblast-specific Klk4 was considerably increased, suggesting a shift toward the maturation-type phenotype during 3D organization (Bronckers et al., 2016; Nunez et al., 2016).

The key electrolyte transporters and channels NHE1, AE2, NBCe1, pendrin and CFTR, which are key factors in HCO₃ ⁻ secretion and intracellular pH regulation, were all expressed in both HAT-7 monolayers and spheroids, although pendrin and CFTR were significantly decreased in the spheroids. Nonetheless these data are generally in line with previous studies which demonstrated NHE1, AE2, NBCe1, pendrin and CFTR expression in ameloblasts and their involvement in intracellular and extracellular pH regulation (Bronckers et al., 2011; Lacruz et al., 2013; Jalali et al., 2014). The fact that these transporters are present in HAT-7 spheroids suggests that this experimental model, like the polarized 2D HAT-7 monolayer, is suitable for studying ameloblast acid/base transport.

An important function of ameloblasts is to produce hydroxyapatite crystals, comprising Ca²⁺, phosphate and water, in a process whereby eight moles of protons are liberated during the formation of each mole of hydroxyapatite (Lacruz et al., 2017; Varga et al., 2018). While the Ca²⁺ transport mechanism in these cells is relatively well understood (Lacruz, 2017; Lacruz et al., 2017; Varga et al., 2018), much less is known about the mechanism of phosphate transport into the enamel matrix. Some possibilities include transport into the cells of the enamel via Na⁺-phosphate cotransporters of the SLC20 and SLC34 gene families (Merametdjian et al., 2018; Beck, 2019) followed by secretion across the apical membrane or into matrix vesicles (Pandya et al., 2017).

Functional models are needed to better understand how ameloblasts direct Ca²⁺ and phosphate transport into the enamel matrix, and how ameloblasts neutralize the acidity generated as hydroxyapatite crystals form (Lacruz et al., 2017; Varga et al., 2018). HAT-7 cells grown as monolayers on permeable supports express ion transport proteins that mediate the uptake of HCO₃ ^- ions across the basolateral membrane in order to drive HCO₃ ^- secretion across the apical membrane (Bori et al., 2016; Racz et al., 2017). Basolateral HCO₃ ^- uptake is achieved by the H₂DIDS-sensitive Na⁺-HCO₃ ^- cotransporter NBCe1, and indirectly by H⁺ extrusion via the amiloride-sensitive Na⁺/H⁺ exchanger NHE1 works in conjunction with intracellular carbonic anhydrase activity continuously producing H⁺ and HCO₃ ^-. The expansion of these cells into organoid culture systems is an important step toward further studies of ameloblast-mediated ion transport and matrix mineralization.

The pH_i measurements presented here confirm that these transporters are also active in HAT-7 spheroids and that, as in the HAT-7 monolayers, there is no evidence of a basolateral H⁺-ATPase contributing to the uptake of HCO₃ ^- ions. The Cl^- substitution experiments confirm the expression of a DIDS-sensitive Cl^-/HCO₃ ^- exchanger, most probably AE2, that can contribute to the basolateral uptake of Cl^- ions, which are also required for enamel formation (Bronckers et al., 2015b; Bronckers, 2017). And the RT-qPCR data confirm that all three transporters, NBCe1, NHE1, and AE2, are expressed in HAT-7 spheroids as they are in HAT-7 monolayers.

It was interesting to note that CFTR expression decreased in spheroid culture. In vivo, CFTR is highly expressed in maturation-stage ameloblasts with reduced expression in the earlier transition stage and almost no expression in the secretory stage (Bronckers et al., 2010; Bronckers et al., 2015a). In previous studies of primary ameloblast-derived cells in grown in Matrigel, we found increased expression and synthesis of amelogenin protein (He et al., 2010). Amelogenin is a secretory stage protein, and this suggests the possibility that in 3D culture, ameloblast-lineage cells may de-differentiate to an early stage of enamel formation. Further studies and culture media can be used to explore this interesting possibility.

Calcium signaling in HAT-7 spheroids was also investigated in the present study. According to the current classification, adenosine stimulates P1 purinergic receptors whereas the extracellular nucleotides ATP, ADP, UTP, and UDP activate P2 receptors, which are further divided into two subtypes: ionotropic (P2X) and metabotropic (P2Y). In the presence of external Ca²⁺, both ATP and UTP elicited biphasic Ca²⁺ responses in HAT-7 spheroids, suggesting the activation of metabotropic P2Y2 and/or P2Y4 receptors (Guns et al., 2005). Furthermore, it has been demonstrated that P2Y receptors could regulate the activity of the NCKX4 Ca²⁺ extrusion pathway (Yang and Lytton, 2013) that is known to play a critical role in dental enamel maturation (Nurbaeva et al., 2017). Since ionotropic P2X receptors are insensitive to UTP, the Ca²⁺ entry during the sustained phase is most likely due to opening of store-operated Ca²⁺ channels. Indeed Nurbaeva and colleagues have recently shown that ATP strongly stimulates store-operated Ca²⁺ entry in maturation but not secretory ameloblasts (Nurbaeva et al., 2018).

Environmental factors, such as exposure to excessive amounts of fluoride, can alter enamel formation. Most studies using ameloblast cell lines have found effects of fluoride only at millimolar concentrations (Li et al., 2006), whereas in primary ameloblast cell cultures fluoride can alter cell characteristics at micromolar levels (Yan et al., 2007; Zhang et al., 2007). In this study, we found that in the presence of 1 mM fluoride applied from day 4, spheroid size gradually decreased, and by day 7 they were clearly disaggregating. Longer term exposure to 1 mM fluoride, between days 2 and 7, resulted in the formation of only a few small spheroids and significant amounts of cellular debris. 3 mM fluoride killed all HAT-7 cells within 2 days.

Exposure to 0.1 and 0.3 mM fluoride for three or six days did not affect spheroid formation compared to controls. These findings are in line with our recent studies on the effect of fluoride using the 2D HAT-7 model, where cells were grown in monolayers on Transwell filters (Racz et al., 2017).

The delay in tight junction formation observed in the 2D monolayer model (Racz et al., 2017) also fits well with our finding that HAT-7 spheroids disintegrated after 3 days of exposure to 1 mM fluoride when isolated from Matrigel using even the most gentle spheroid isolation procedure. The data indicate that 1 mM fluoride not only affects spheroid formation, but it also weakens the cell-cell adhesion structures responsible of maintaining the spheroid structure.

Studies of the effects of micromolar, physiologic levels of fluoride on primary ameloblast culture show that, while there are effects on cell proliferation and apoptosis, there are few to no changes in gene expression (Kubota et al., 2005; Zhang et al., 2006; Yan et al., 2007; Zhang et al., 2007; Sharma et al., 2008; Zhang et al., 2016).

These findings highlight the importance of developing a robust and reproducible culture system to investigate questions such as how fluoride alone, or in combination with other environmental stressors, alters enamel formation and biomineralization. Our finding that HAT-7 cells can form spheroids similar to those formed by primary culture ameloblasts offers the possibility of scaling up this model system to better understand cell-cell and cell-matrix interactions and their importance in enamel formation.

In conclusion, this model system shows that HAT-7 cells cultivated within a Matrigel extracellular matrix form three-dimensional, multi-cellular, spheroidal structures that retain their functional capacity for pH regulation and intracellular Ca²⁺ signaling. They also express protein markers consistent with a maturation-stage ameloblast phenotype. This new 3D model will allow us to gain a better understanding of the molecular mechanisms involved in amelogenesis, not only in health but also in disorders of enamel formation, such as those resulting from environmental factors.