Oxygen regulates epithelial stem cell proliferation via RhoA-actomyosin-YAP/TAZ signal in mouse incisor

Stem cells (SCs) possess the ability of self-renewal and differentiation into specialized cell types. This characteristic is essential to create new tissues and organs during development and to maintain tissue homeostasis (Wabik and Jones, 2015). SC proliferation is controlled by intrinsic (cell-autonomous) and extrinsic (environmental) cues, and collapse of the regulation causes serious events (Chen et al., 2012; Pietras et al., 2011). Thus, understanding the mechanisms governing SC proliferation is crucial.

In ectodermal appendages that undergo continuous cell turnover, such as teeth, hair follicles, mammary glands and nails, epithelial stem cells (ESCs) reside in a niche, a physiologically limited microenvironment, containing a heterogeneous cell population with different proliferation rates, such as slow-cycling cells (SCCs) and actively dividing transit-amplifying cells (TACs) (Harada et al., 1999; Hsu et al., 2014; Wabik and Jones, 2015).

In particular, the rodent incisor, which grows continuously throughout the life of the animal, provides an excellent model to investigate the characteristics and regulation of SCs in general, because teeth share many developmental mechanisms with other ectodermal organs. To counterbalance constant abrasion, the self-renewal of the dental epithelium located on the labial side covered with enamel is supported by the continuous supply of cells at the proximal end, called the apical bud (Harada et al., 1999) (Fig. 1A,B). Previously, it has been proposed that the incisors follow a classical SC paradigm. A few SCCs reside in the outer enamel epithelium (OEE) and the stellate reticulum (SR) as dental epithelial stem cells (DESCs). They asymmetrically divide to give rise to TACs in the inner enamel epithelium (IEE) through the basal epithelium (BE). Then, the IEE moves distally, ceases proliferation and undergoes differentiation into ameloblasts to form the enamel (Harada et al., 1999; Kuang-Hsien Hu et al., 2014) (Fig. 1C,D). This model is based on the lineage tracing of putative SC markers (Juuri et al., 2012; Seidel et al., 2017). Recently, it has been proposed that TACs in IEE undergo self-renewal and contribute to the formation of both SCCs and ameloblasts following single-cell RNA-sequencing studies and advanced computational techniques (Sharir et al., 2019). Thus, although the identity, hierarchies and kinetic of the DESCs are still under investigation, cell proliferation in the niche must be tightly regulated for proper development and renewal. To date, a number of signaling pathways regulating DESCs have been studied (Yu and Klein, 2020). However, the impact of niche microenvironment on their dynamics remains poorly understood.

Molecular oxygen (O₂) is involved in a wide range of cellular functions and metabolism. Recently, increasing evidence indicates that O₂ as a niche component plays an essential role in the maintenance of stem cells and in cell fate decisions. For example, in hematopoietic SCs residing in the sinusoidal hypoxic area distant from the blood vessels, hypoxia maintains their undifferentiated state and influences their proliferation and cell-fate commitment (Kubota et al., 2008). The hypoxic response is mediated by hypoxia-inducible transcription factors (HIFs) that are stabilized under hypoxia and regulate gene expression (Takubo et al., 2010) and intracellular signaling pathways (Choi et al., 2017; Gustafsson et al., 2005). In contrast, high O₂ concentration accelerates cellular aging, which leads to the loss of SC properties (Csete, 2005). These signaling cascades and transcriptional programs induced by O₂ appear to be shared by many SCs (Lin et al., 2008; Mohyeldin et al., 2010; Simon and Keith, 2008). However, in epithelial tissues, although the regulatory mechanism of SCs have been gradually elucidated (Morrison and Spradling, 2008), no comprehensive data on how O₂ is involved in the regulation of SC proliferation are available.

RhoA, the founding member of the Rho family GTPase, influences many biological processes, including cell cycle progression and proliferation (Coleman et al., 2004). RhoA is a molecular switch that cycles between inactive GDP-bound and active GTP-bound forms. The GTP-bound active RhoA associates with the downstream effector protein Rho-associated protein kinase (ROCK), which promotes the polymerization and stabilization of filamentous actin (F-actin), the phosphorylation of myosin light chain (MLC) and myosin ATPase activity, leading to actomyosin tension (Jaffe and Hall, 2005; Ridley, 1996; Riento and Ridley, 2003). Actomyosin in non-muscle cells controls cell shape and motility and functions as a mechanical stressor. Recently, we have demonstrated that RhoA signal regulates the cell proliferation of dental epithelial cells via the reorganization of actomyosin and the adherens junction (AJ) (Otsu et al., 2016, 2011), indicating that the RhoA signal is a potent regulator of cell proliferation of DESCs. However, its upstream and downstream components have yet to be identified.

Yes-associated protein (YAP) and its homolog, the transcriptional co-activator with PDZ binding motif (TAZ), are transcriptional cofactors that shuttle between the cytoplasm and nucleus and can be dependent on the Hippo pathway (Lei et al., 2008; Zhao et al., 2007). In the nucleus, YAP/TAZ associate with several promoters of specific transcription factors, which activate genes regulating cell proliferation for the control of organ size in both Drosophila and mammals (Dong et al., 2007). YAP/TAZ are also involved in cell-fate decisions as a stemness factor in different SC pools of the body (Camargo et al., 2007; Watt et al., 2010). In several tissues, YAP/TAZ regulate the proliferation and differentiation of stem/progenitor cells (Cao et al., 2008; Hu et al., 2017; Schlegelmilch et al., 2011). Recently, in rodent incisors, YAP/TAZ have been reported to control the proliferation and differentiation of TACs in response to integrin/FAK signaling (Hu et al., 2017). Further, the activity of YAP/TAZ has recently been linked to mechanical and cytoskeletal signals. Previous studies have shown that cortical actomyosin and stress fibers are involved in nuclear translocation of YAP/TAZ (Furukawa et al., 2017; Wada et al., 2011).

In this study, we explore the role of microenvironmental O₂ in ESCs and its molecular regulatory mechanism using the mouse incisor model. We hypothesized that RhoA-mediated actomyosin regulates YAP/TAZ activity to control the cell proliferation of DESCs. We found that microenvironmental O₂ is an important regulator of SC proliferation through the RhoA-actomyosin-YAP/TAZ signal. Our study offers a conceptual framework in understanding environment-mediated SC regulation, thus aiding in the development of SC-based regeneration therapy.

We first visualized the cell proliferation profile of the apical bud in the tetracycline-regulatable Histone 2B-Green Fluorescent Protein (H2B-GFP) transgenic mouse incisor model, in which the fusion protein of H2B-GFP is expressed after doxycycline (DOX) administration. GFP fluorescence was diluted twofold at each division, therefore it is retained preferentially in slowly dividing cells. In the lower incisor of postnatal day (P) 3 mouse after one DOX administration at embryonic day (E) 16.5, the GFP label retaining SCCs was found in the OEE and outside SR, whereas GFP was gradually decreased from the inside of SR to IEE through BE (Fig. 1E,F). Consistent with this observation, the number of Ki67-positive proliferating cells increased from the apical bud to TACs in P3 wild-type (WT) mice (Fig. 1G-K; Fig. S1A).

To estimate the oxygen concentration in tissues around the apical bud, we analyzed the blood vessel distribution. Stereomicroscopic images of the mouse incisor immediately after extraction showed that few blood vessels are visible around the apical bud (Fig. 2A,B, asterisk). To investigate the distribution of blood vessels inside the incisor, we analyzed the tissue sections of transgenic FLK1-GFP mice with GFP and the FLK1 (VEGFR-2; also known as Kdr) promoter. In both the frontal and sagittal sections, the apical bud was at a distance from the blood vessels (Fig. 2C-E), whereas the high-density blood vessels were close to highly proliferating TACs and pre-odontoblasts/odontoblasts (Fig. 2C,F, arrowheads). Further, HIF1α was strongly expressed in SCCs in the apical bud compared with TACs (Fig. 2G; Fig. S1B). These results suggest that more oxygen was supplied to TACs than to SCCs.

We further examined the difference of the metabolic program between SCCs and TACs. The expression of pyruvate dehydrogenase (PDH), which aerobically catalyzes the conversion of pyruvate to acetyl-CoA for use in mitochondrial metabolism (Harris et al., 2002), and of citrate synthase, the first and rate-limiting enzyme of the tricarboxylic acid (TCA) cycle for oxidative phosphorylation (OXPHOS) in mitochondria (Cheng et al., 2009), was higher in TACs than in SCCs (Fig. 2H,I). Moreover, scanning electron microscopy (SEM) revealed that the shape of mitochondria in H2B-GFP-positive SCCs was spherical, whereas that in GFP-negative TACs was tubular. The size of mitochondria in SCCs was smaller than that in TACs (Fig. 2J-M). These results indicate that SCCs preferentially use anaerobic glycolysis, whereas TACs undergo a metabolic switch towards OXPHOS to meet the cell energy demand for proliferation, thereby requiring more oxygen consumption and mitochondrial biogenesis. Taken together, these results strongly suggest that SCCs are maintained in a relatively hypoxic environment compared with TACs, dependent on the distance from blood vessels.

To ascertain whether cell proliferation in the apical bud is dependent on environmental O₂ concentration, we cultured mouse incisors under hypoxic condition. Stereomicroscopic images showed that, in hypoxia (5% O₂), the length of the dental epithelium was shorter and the area of the apical bud was smaller than in normoxia (21% O₂) (Fig. 3A-C). Consistent with the suppression of epithelial growth, the number of Ki67-positive cells was reduced by hypoxia in the apical bud (Fig. 3D,E; Fig. S2A) without inducing apoptosis (Fig. S3A). Then, we evaluated the association between environmental O₂ and YAP/TAZ expression and confirmed the strong immunoreaction of the antibodies against YAP/TAZ and YAP1 in TACs (Fig. 3F, arrows; Fig. S1C,D), whereas TAZ was predominantly expressed in the cytosol and showed no significant differences between TACs and the apical bud (Fig. 3G; Fig. S1E). The organ culture of mouse incisors showed that hypoxia reduced the number of nuclear YAP/TAZ- and YAP1-positive cells (Fig. 3H,I; Fig. S2B).

In addition, we investigated the effect of O₂ concentration on cell proliferation and the expression of YAP/TAZ in mHAT9d cells, originally established from the apical bud of a mouse incisor (Harada and Ohshima, 2004; Harada and Otsu, 2019; Kawano et al., 2002). Previous studies have reported that cell proliferation and YAP/TAZ expression in cultured epithelial cells were dependent on cell density (Zhao et al., 2007). In normoxia, low-density mHAT9d cells were actively proliferated, whereas Ki67, YAP/TAZ and YAP1 were strongly expressed in the nucleus (Fig. S4). In contrast, high-density mHAT9d cells were negative for Ki67, whereas YAP/TAZ and YAP1 were present in the cytosol (Fig. S4). We also demonstrated that the knockdown of YAP1 using small interfering RNA (siRNA) reduced cell proliferation of low-density mHAT9d cells, confirming YAP1 involvement in cell proliferation (Fig. S5).

Low-density mHAT9d cell culture under hypoxic conditions showed that cell proliferation and nuclear YAP/TAZ and YPA1 expression were significantly decreased compared with those under normoxic conditions, without inducing apoptosis (Figs S3B and S6A-F). Hypoxia also decreased the YAP1 mRNA expression, but not TAZ (Fig. S6G). Together, these results were consistent with the hypothesis that hypoxia repressed cell proliferation through the inhibition of YAP/TAZ activity in the apical bud.

We have previously demonstrated that the downregulation of RhoA/ROCK signal promotes cellular proliferation in the apical bud (Otsu et al., 2016, 2011). Thus, we hypothesized that the oxygen- and YAP/TAZ-dependent cell proliferation is mediated by RhoA signaling. Immunohistochemistry illustrated that the expression of active RhoA (GTP-binding form), phosphorylated myosin light chain (pMLC) and F-actin, a downstream event of RhoA, was higher in SCCs in the apical bud than in TACs (Fig. 4A,B; Fig. S1F-H). These indicated that RhoA signaling is more active in SCCs under hypoxic conditions in vivo. To determine whether the activity of RhoA in dental epithelial cells is dependent on environmental O₂ concentration, we cultured mouse incisors under hypoxic conditions and showed that hypoxia increased the expression of active RhoA, HIF1α, F-actin and pMLC in dental epithelium (Fig. 4C; Fig. S2C-E). Similarly, we have observed that hypoxia also increased active RhoA and HIF1α expressions in mHAT9d cells (Fig. 4D,E,H,J). Further, HIF1α knockdown decreased active RhoA expression during hypoxia in mHAT9d cells (Fig. 4E-G,I,J), indicating that HIF1α mediated hypoxia-induced RhoA activation.

We examined the detailed role of the RhoA/ROCK signal on actomyosin in mHAT9d cells. High-density mHAT9d cells exhibited a distinct cortical expression of F-actin and pMLC compared with low-density cells (Fig. S7). However, Y27632, an inhibitor of RhoA/ROCK-mediated actomyosin, diminished this distinct expression, induced a morphological change from cuboidal into spindle-shaped, and enlarged cell size (Fig. 5A,B), suggesting a decrease in the cortical tension in cells. Similarly, blebbistatin, a selective myosin II ATPase inhibitor, caused a reduction in cortical actomyosin and a morphological change (Fig. 5A,B).

Because cortical actomyosin is connected to the AJ and maintains its structure and function in epithelial cells (Lecuit and Yap, 2015), we next investigated the involvement of RhoA/ROCK-mediated actomyosin in AJ. AJ proteins, such as E-cadherin (E-cad), α-catenin, β-catenin and merlin (Nf2), were co-expressed at the cell periphery (Fig. S8), whereas Y27632 and blebbistatin diminished their expression in mHAT9d cells (Fig. 5C). Y27632 also decreased E-cad expression in the apical bud of the cultured mouse incisors (Fig. 5D). In addition, we have confirmed that E-cad and merlin are weakly expressed in TACs compared with the apical bud in vivo (Figs S1I,J and S9A, arrows), which coincided with the RhoA activation status (Fig. 4). Furthermore, in cultured incisors and mHAT9d cells, hypoxia increased expression of E-cad and merlin (Figs S2F,G and S9B,C), whereas HIF1α knockdown enlarged cell size and reduced the peripheral expression of E-cad, merlin and F-actin (Fig. S9D,E). Together, these results indicate that the hypoxia-RhoA signal is involved in the maintenance of cortical actomyosin and the AJ.

We also investigated whether the RhoA-mediated cortical actomyosin regulates cell proliferation and YAP/TAZ localization. Y27632 and blebbistatin increased cell proliferation (Fig. 6A) and induced the nuclear translocation of YAP/TAZ in high-density mHAT9d cells (Fig. 6B). Similarly, in cultured mouse incisor, Y27632 increased YAP/TAZ expression in the apical bud (Fig. 6C). Y27632 and blebbistatin reduced the cytosolic expression of phosphorylated YAP (pYAP; S127) and phosphorylated LATS1/2 (pLATS1/2) (Fig. 6D,E), both of which are essential for the Hippo pathway-dependent cytosolic retention of YAP/TAZ (Piccolo et al., 2014). Collectively, these results strongly support the notion that RhoA-mediated cortical actomyosin restrains cell proliferation through the Hippo-YAP/TAZ signal.

To elucidate the molecular mechanism of RhoA-cortical actomyosin-mediated YAP/TAZ nuclear translocation, we focused on merlin, a membrane-cytoskeleton scaffolding protein that interacts with the E-cad complex and F-actin downstream of the RhoA-actomyosin signal (Cooper and Giancotti, 2014). When we knocked down merlin mRNA (more than 30%) using two independent siRNA in mHAT9d cells, YAP/TAZ expression in the nucleus and cell proliferation were facilitated (Fig. 7A-D). Merlin knockdown also reduced cytosolic pYAP and pLATS1/2 expression (Fig. 7E,F), indicating the inactivation of Hippo signaling. These results suggested that merlin mediates between the RhoA-regulated cortical actomyosin and Hippo-dependent YAP/TAZ signal.

In this study, we have investigated the impact of microenvironmental O₂ on ESCs using the mouse incisor model and have revealed a signaling axis governing cell proliferation. We have shown that SCCs are located away from blood vessels and are therefore maintained in a relatively hypoxic environment, whereas TACs are found near blood vessels and exist in a relatively high O₂ environment with OXPHOS metabolism. In SCCs, hypoxia-induced HIF1α activated RhoA, which in turn stabilized the cortical actomyosin-AJ protein complex. Merlin at the AJ inhibited YAP/TAZ nuclear translocation through phosphorylation of YAP and LATS1/2, thus resulting in the retention of the slow-cycling state (Fig. 8). These results reveal a unique signal conversion from an environmental stimulus (O₂) to a biochemical transcriptional program (YAP/TAZ) that is integrated by a mechanical stressor (actomyosin), thus highlighting the important implication of blood-vessel geometry for the regulation of ESCs.

We have observed that SCCs were maintained under relative hypoxia in vivo compared with TACs. Interestingly, the apical bud was more distant from the blood vessels than TACs, implying that less O₂ was supplied to SCCs than to TACs. This finding is in line with previous studies that found slow-cycling HSCs residing in the sinusoidal hypoxic zone distant from the vascular niche near the capillaries (Kubota et al., 2008). In neural SCs, relief of hypoxia through the ingrowth of blood vessels promoted their differentiation (Lange et al., 2016). Thus, our findings have shown the important biological implication of the blood-vessel density and geometry for the regulation of ESC behavior, suggesting that the apical bud has a repulsive effect on angiogenesis. Indeed, we have previously demonstrated that the apical bud did not express semaphorin 4D, which has a pro-angiogenic effect (Otsu et al., 2016), whereas other studies have shown that dental epithelial cells (DECs) express semaphorin 3A and 3F, which are anti-angiogenic molecules (Shrestha et al., 2014; Sijaona et al., 2012). These studies strongly suggest the existence of an inter-tissue communication between the apical bud and blood vessels through molecular cues to maintain hypoxia. Therefore, further in vivo investigation on deletion/manipulation of blood vessels and direct measurement of oxygen tension will help to identify the correlation between distance from blood vessels and hypoxia, and elucidate the regulatory mechanism underlying blood-vessel density and geometry in the incisor.

TACs were shown to consume more oxygen in the mitochondria than SCCs, to guarantee that sufficient energy (ATP) for active proliferation is available. This suggests that a metabolic switch is activated between SCCs and TACs. Recent studies have shown that the energy requirements in certain SCs are altered depending on the cell lineage, which is supported by mitochondrial biogenesis and the functional change to ensure a suitable amount of ATP (Hu et al., 2016; Ito and Suda, 2014). Under hypoxia, the activity of the mitochondrial electron transport chain is decreased, energetic needs to shift from OXPHOX to glycolysis and, consequently, SCs enhance their self-renewal ability and maintenance of pluripotent capacity in vitro (Ježek et al., 2010). For most cell types, hypoxia has been found to decrease the levels of respiratory enzymes and oxygen consumption rate but increase the production of glycolytic enzymes and lactate, which eventually forces the cells to rely on glycolysis (Saini et al., 2013). For example, mesenchymal stem cells (MSCs) have been shown to express higher levels of glycolytic enzymes and lower levels of OXPHOS protein compared with their differentiated progeny (osteoblasts) (Chen et al., 2008). As O₂ concentration increased, MSCs adapted a state of high-oxygen consumption using OXPHOS, and their proliferation increased (Pattappa et al., 2011, 2012). Moreover, intestinal ESCs have been recently reported to exhibit a different metabolic program compared with differentiated cells (Rodríguez-Colman et al., 2017) and with our results. Thus, the importance of this metabolic switch and O₂ in ESCs and its regulatory mechanism should be clarified.

Immunofluorescence assays showed that SCCs resided in a hypoxic environment and exhibited higher HIF1α and active RhoA expression compared with TACs in vivo. Also, we have experimentally confirmed that hypoxia induced HIF1α and active RhoA expression, and that HIF1α inhibition reduced active RhoA expression in hypoxia. Other recent studies have also showed the link between hypoxia and the RhoA signal in angiogenesis, endocytosis and the motility of cancer cells (Dada et al., 2007; Gilkes et al., 2014; Turcotte et al., 2003). In normal differentiated cells, such as pulmonary arterial smooth muscle cells and chondrocytes, RhoA activation by hypoxia has been shown to be involved in actin remodeling and redifferentiation (Bailly et al., 2004; Öztürk et al., 2017). These studies noted that the responsiveness of RhoA to hypoxia was dependent on the cell age and culture condition, implying that impact of O₂ on RhoA signal is context dependent.

Regarding the mechanism by which HIF1α regulates RhoA, certain G-protein-coupled receptors (GPCRs), which modulate RhoA activity, have been shown to be HIF1α target genes (Recchia et al., 2011; Ronkainen et al., 2014), while various GPCR ligands are induced in hypoxia and stimulate RhoGEFs (Stenmark et al., 2006). Conversely, little is known on the involvement of HIF1α in the regulation of RhoGAPs and RhoGDI, which warrants further investigations.

The cortical actomyosin underlying AJ serves as a platform for integrating the chemical and mechanical signals to control many cellular behaviors, including proliferation. Here, we have shown that disruption of the RhoA-mediated cortical actomyosin promoted YAP/TAZ nuclear translocation and cell proliferation via Hippo signaling. We have further confirmed that SCCs, which exhibited high active RhoA, F-actin and pMLC expression, clearly displayed lower YAP/TAZ and YAP1 expression compared with TACs in vivo. These findings indicate the crucial role of RhoA-mediated actomyosin on the YAP/TAZ signal to control cell proliferation in DESCs. Consistent with our finding, in MDCK cells the formation of RhoA-mediated cortical actomyosin inhibited YAP/TAZ nuclear translocation and proliferation (Furukawa et al., 2017). On the other hand, other studies using adult transgenic mice showed that another small Rho GTPase, Cdc42, but not the RhoA-ROCK signal, regulated YAP/TAZ localization in DESCs (Hu et al., 2017). Because our study used neonatal mouse samples, this discrepancy suggests that the type of small GTPase that controls YAP/TAZ may differ depending on age. Further, other cell types have shown the opposite effect of Rho-mediated actomyosin on YAP/TAZ nuclear translocation in stress fibers and primary cilia (Dupont et al., 2011; Kim et al., 2015; Wada et al., 2011), suggesting that the regulatory mechanism of YAP/TAZ by RhoA-mediated actomyosin is cell- and actomyosin-type dependent.

Herein, we have demonstrated that YAP1 was strongly expressed in TACs compared with SCCs, whereas TAZ was expressed in cytosol with no differences between SCCs and TACs in vivo. In addition, hypoxia reduced YAP1 mRNA expression, but had no effect on TAZ. These results suggest that YAP and TAZ have non-redundant functions and that YAP1 predominantly contributes to our proposed mechanism under normal conditions. TAZ, expressed in the cytoplasm, is a compensatory effector when YAP activity is disrupted in the mouse incisor (Hu et al., 2017). Hepatic or intestinal deletion of YAP also resulted in TAZ nuclear localization, demonstrating that TAZ became activated in response to YAP loss (Moroishi et al., 2015). Thus, it is inferred that TAZ also play a compensatory role for YAP in our proposed mechanism.

We have showed that Y27632 and blebbistatin caused the disruption of the AJ protein complex, including merlin, and actomyosin breakdown, resulting in the promotion of YAP/TAZ nuclear translocation and cell proliferation. Knocking down merlin inhibited Hippo signaling and promoted YAP/TAZ nuclear translocation and cell proliferation. These results suggest that merlin mediates between RhoA-actomyosin and Hippo-YAP/TAZ signals in DESCs. Consistent with our results, merlin has been reported to regulate YAP/TAZ nuclear translocation through the Hippo pathway via LATS phosphorylation (Yin et al., 2013). However, a recent study has showed that merlin directly controls YAP/TAZ nuclear localization downstream of the cortical actin contraction via nucleocytoplasmic shuttling (Furukawa et al., 2017). This suggests the minor involvement of the direct regulation of YAP/TAZ by merlin in DESCs. In addition to merlin, the strong expression of E-cad in the apical bud has been reported to inhibit the proliferation of DESCs (Li et al., 2012), thus supporting the importance of AJ stability in DESC proliferation.

We have also observed that the expression patterns of HIF1α, active RhoA, F-actin and merlin were slightly different from pMLC and E-cad in the apical bud. pMLC and E-cad were strongly expressed, whereas the expression of Ki67, YAP/TAZ and YAP1 was weak in SR, which is consistent with our proposed mechanism. Meanwhile, the expression of HIF1α, active RhoA, F-actin and merlin in SR appeared to be relatively weak. This discordance of expression patterns suggests the minor involvement of our proposed mechanism in SCCs within the apical bud, including SR, OEE and BE. However, in comparison between the apical bud and TACs, the expression of HIF1α, active RhoA, pMLC, F-actin, E-cad and merlin in the apical bud were higher than that in TACs, whereas the expression of YAP1, YAP/TAZ and Ki67 were much lower in SCCs than in TACs. These results are in line with the proposed mechanism. Therefore, we concluded that the mechanism proposed in the present study significantly contributed to the regulation of cell proliferation between SCCs and TACs.

In summary, we have demonstrated that microenvironmental O₂ concentrations can fine-tune SC proliferation through RhoA-cortical actomyosin-YAP/TAZ signaling. This finding sheds light on the biological significance of blood-vessel geometry and signal transduction from an O₂ stimulus, mediated by the cytoskeleton, in SC proliferation. Our findings provide a conceptual advance on how SCs interact with their environment during tissue development and renewal, thus aiding in the development of SC-based therapeutic strategies for the regeneration of dental and other ectodermal tissues.

All animal experiments complied with the guidelines of the Ministry of Education, Culture, Sports, Science and Technology, the Ministry of the Environment, and the Science Council of Japan and were carried out in accordance with the Act on Welfare and Management of Animals. The protocol for the animal experiments was approved by the Institutional Animal Care and Use Committee (27-042, 28-028, SA00091) and the Institutional Recombinant DNA Experiments Safety Committee of Iwate Medical University and Niigata University (#308, 406, 423, SD00328).

Animals were maintained in plastic cages inside a room at an ambient temperature of 23±1°C with ad libitum access to water and standard laboratory pellet diet. Plug date was defined as E0.5. ddY male mice (RRID:MGI:5558113, Japan SLC, Shizuoka, Japan) were used as the wild type. Flk1-GFP BAC Tg mouse (RRID:IMSR_CARD:2408) that were administered with GFP combined with an Flk1 promoter (Ema et al., 2003, 2006; Ishitobi et al., 2010) were used to observe blood vessel distribution. TetOP-H2BGFP mice [B6;129S4-Gt(ROSA)26Sor^{tm1(rtTA*M2)Jae} Col1a1^{tm7(tetO-HIST1H2BJ/GFP)Jae}/J Mus musculus, RRID:IMSR_JAX:016836] were purchased from the Jackson Laboratory (Foudi et al., 2009). For GFP transgene expression, 2 mg/ml DOX (Merck, D9891, CAS: 24390-14-5) supplemented with 50 mg/ml sucrose was administered to a mother through drinking water to label cell pups at E16.5. Then 3-day-old labeled birthed animals were used to monitor the cell proliferation rate. For histological analysis, mouse jaws were dissected, fixed in 4% paraformaldehyde (PFA; Nacalai Tesque, 26126-25, CAS: 30525-89-4) and decalcified with Osteosoft (Merck, 01728). For Hematoxylin and Eosin (H&E) staining and immunostaining, the samples were dehydrated in a graded ethanol series, embedded in paraffin wax and sectioned (6-7 µm thick) using a microtome (RX-860, Yamato).

mHAT9d cells were established from the apical bud of a mouse incisor (Harada and Ohshima, 2004; Harada and Otsu, 2019; Kawano et al., 2002). The culture medium consisted of Dulbecco's modified Eagle's medium (DMEM)/F12 (11330-032; Life Technologies) supplemented with B-27 (Thermo Fisher Scientific; 17504044), 25 ng/ml basic fibroblast growth factor (FGF) (233-FB; R&D Systems), 20 ng/ml epidermal growth factor (EGF) (2028-EG; R&D Systems) and 1% penicillin-streptomycin (15140; Thermo Fisher Scientific). For comparing low- and high-density cultures, 4×10⁴ (low) or 3×10⁵ (high) cells for 24-well plates, and 3×10³ (low) or 5×10⁴ (high) cells for 96-well plates were seeded in Corning^® Primaria™ culture plates. The cells were treated with (-)-blebbistatin (13013, CAS: 856925-71-8; Cayman Chemical) or Y-27632 (253-00513, CAS: 331752-47-7; Wako). Dimethyl sulfoxide (DMSO; CAS: 67-68-5; Sigma-Aldrich, D2650) and H₂O were used as vehicles for blebbistatin and Y-27632, respectively. To induce hypoxia, the cells were cultured in hypoxic chambers (MCO-5 M, PHCbi) with 5% O₂, 5% CO₂ and 90% N₂. Nitrogen gas was supplied to the chambers to induce a controlled, reduced percentage of oxygen. For normoxia, the cells were cultured in incubators containing 5% CO₂ and 21% O₂.

Mouse incisor organ culture was performed as previously described (Otsu et al., 2016). Briefly, the incisors at P2 were dissected from the lower jaws of ddY mice and grown in a Trowell system. After cultivation with an inhibitor or under hypoxia, the images were taken using a stereomicroscope (Stemi 2000-C, Zeiss) and the explants were then fixed in 4% PFA in PBS, decalcified in Osteosoft, and embedded in paraffin for histological analysis. Image analyses were carried out using ImageJ (https://imagej.nih.gov/ij/).

Immunohistochemical and immunofluorescent staining were performed as described previously (Otsu et al., 2016, 2011). After blocking, the sections were reacted with the following antibodies (1:100): active RhoA (26904, RRID:AB_1961799; NewEast Biosciences), Ki67 (14-5698-82, RRID:AB_10854564; Thermo Fisher Scientific), E-cadherin (610182, RRID:AB_397581; BD Biosciences), F-actin (bs-1571R, RRID:AB_10859354, Bioss), HIF1α (ab16066, RRID:AB_302234; Abcam), p-LATS1/2 (SAB4504615; Merck), Merlin (HPA003097, RRID:AB_1079473; Merck), α-catenin (71-1200, RRID:AB_2533974; Thermo Fisher Scientific), PDH (MA5-14805, RRID:AB_10986167; Thermo Fisher Scientific), citrate synthase (ab96600, RRID:AB_10678258; Abcam), YAP/TAZ (8418, RRID:AB_10950494; Cell Signaling Technology), YAP1 (ab205270, RRID:AB_2813833, Abcam), TAZ (sc-518026; Santa Cruz Biotechnology), pMLC (3671, RRID:AB_330248; Cell Signaling Technology), β-catenin (sc-7963, RRID:AB_626807; Santa Cruz Biotechnology) and pYAP (D9W2I, RRID:AB_2650553; Cell Signaling Technology). DAPI (300 nM; D1306, RRID:AB_2629482), Alexa Fluor 488 (1:500) and Alexa Fluor 546 (1:500) secondary antibodies were procured from Thermo Fisher Scientific. To detect F-actin, the cells were stained using Alexa Fluor 546-conjugated Phalloidin (6.6 µM; A22283, RRID:AB_2632953; Thermo Fisher Scientific). TUNEL staining was conducted using in situ apoptosis detection kits according to the manufacturer's protocol (MK500; Takara Bio). Images were obtained under a fluorescence microscope (BX51, IX71, Olympus), laser-scanning confocal microscope (C1si, Nikon) or imaging cytometer (Cytell Cell Imaging System, GE Healthcare UK). Image analyses were carried out using ImageJ or the software provided with the microscope. For the analysis of cell proliferation, cells were imaged using at least 10 fields in every well under ×10 magnification using the Cytell Cell Imaging System, and the imaged nuclei (DAPI fluorescence) were enumerated as the cell number using an image analysis software (IN Cell Analyzer 1000 Workstation, GE Healthcare UK). Appropriate positive and negative controls were run during each experiment.

For comparing the mitochondria in SCCs and TACs, we applied array tomography (Micheva and Smith, 2007) with slight modifications. Briefly, the mandibles of TetOP-H2BGFP mice were fixed for 2 h at 4°C in 4% PFA dissolved in 0.1 M phosphate buffer (PB). After three washes with PB, the sample was dehydrated in an ascending series of ethanol (70, 80, 90, 95, 99, 100%), followed by substitution with 1:1 ethanol/LR white resin (AGR1280, CAS #:94188-59-7; Agar Scientific) overnight at room temperature (20∼25°C). Final infiltration with LR white resin was carried out. Polymerization was performed by incubation at 60°C for 3 days. The resin-embedded samples were cut into ultrathin sections (0.5 µm thickness) using a Histo Diamond Knife (Diatome) and then bonded to glass slides. GFP fluorescence in the sections was imaged using a laser-scanning confocal microscope (C1si). The same sections were then stained with 1% uranyl acetate in H₂O for 30 min and with lead stain solution (18-08-75-2; Merck) and imaged on an SU-8010 electron microscope (Hitachi High-Technologies). SCCs and TACs were classified based on the GFP intensity in the fluorescent image, and their mitochondria were analyzed using SEM images. Image analyses were carried out using ImageJ.

Immunoblotting was performed as previously described (Ida-Yonemochi et al., 2016). After incubation under hypoxic condition for 24 h, mHAT9d cells were lysed using RIPA buffer (08714-04; Nacalai Tesque). The lysates were separated using SDS-PAGE and transferred to PVDF membranes. The membranes were incubated with anti-HIF1α antibody (1:1000) or α-tubulin (1:1000; 66031-1-lg, RRID:AB_11042766; Proteintech). For immunodetection, the Envision+/HRP system (K4061; Dako) was used, and the antigens were detected using Western BLoT Quant HRP Substrate (T7102A; Takara Bio).

Sets of two 25-mer duplex siRNAs targeting genes were obtained from Thermo Fisher Scientific. All siRNA duplexes (10 nmol/l) were transfected into cells using Lipofectamine RNAiMAX (13778075; Thermo Fisher Scientific) according to the manufacturer's protocol. Stealth siRNA negative control duplexes (12935300; Thermo Fisher Scientific) were used to control for the sequence-independent effects following siRNA delivery. Transfection efficiency, which was monitored using a fluorescent oligonucleotide (14750100, BLOCK-iT fluorescent oligonucleotide; Thermo Fisher Scientific), was estimated to be between 80% and 90%. Inhibitory effects were observed in all siRNA duplexes, and two of the three siRNAs with strong inhibitory effect were used in three independent experiments.

Total RNA was extracted using an RNeasy Mini Kit (74104; Qiagen). Reverse transcription of the total RNA was performed using a PrimeScript RT reagent kit (RR037A; Takara Bio). The quantitative analysis of gene expression was performed through RT-qPCR using SYBR1 Premix Ex Taq (RR820S; Takara Bio) and the oligonucleotide primers specific for the target sequences on a Thermal Cycler Dice (Takara Bio) according to the manufacturer's protocol. Amplification conditions were as follows: 30 s at 95°C; 40 cycles of 95°C for 5 s and 60°C for 30 s; dissociation for 15 s at 95°C; and 30 s at 60°C. Target gene expression levels were normalized against the corresponding levels of Gapdh mRNA. The relative gene expression levels were calculated against the values of the negative control cultures using the comparative Ct (2^−ΔΔCt) method. Experiments were carried out in triplicate. Sequences of primers are as follows: Gapdh (Fw 5′-TGTGTCCGTCGTGGATCTGA-3′, Rv 5′-TTGCTGTTGAAGTCGCAGGAG-3′), E-cad (Fw 5′-CGTCCTGCCAATCCTGATGA-3′, Rv 5′-ACCACTGCCCTCGTAATCGAAC-3′), HIF1α (Fw 5′-GCGTGCATGTCTAATCTGTTCC-3′, Rv 5′-GATTCTGACATGCCACATAGCTC-3′), merlin (Fw 5′-GAAGCTCTTAGAAATCGCCACCA-3′, Rv 5′-CAGCAATAATGTCGAAGCTCGGTA-3′), YAP1 (Fw 5′-CATGGTGCGCCTTGTTATACCTC-3′, Rv 5′-TCAGGCCGTGATTCAAATTTAGTG-3′), TAZ (Fw 5′-TCACCGTCTCCAACCACCAG-3′, Rv 5′-AGCTCCTTGGTGAAGCAGATGTC-3′).

All data are mean±s.e.m. Differences were considered statistically significant if P<0.05 using unpaired two-tailed Student's t-test: *P<0.05, **P<0.01, ***P<0.001. To quantify the immunofluorescence intensity, at least 30 cells in every sample were analyzed in one experiment unless otherwise indicated.

We thank Naoyuki Nishiya (Department of Clinical Pharmacy, Iwate Medical University) for useful discussions regarding our study and for helpful insights. We also thank Tomohito Hanasaka and Eri Ishiyama (Technical Support Center for Life Science Research, Iwate Medical University), and Takafumi Fukasawa, Wataru Yasuno (Center for In Vivo Science, Iwate Medical University) and Aya Kikuchi for technical assistance, and Yukiko Onuma and Michiko Sugawara for secretarial support.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.194787.reviewer-comments.pdf