Piezo1 and Piezo2 collectively regulate jawbone development Free

Piezo channels are relatively newly discovered mechanosensors that sense mechanical signals within cellular niches to enable tissue to adapt to various environmental conditions (Coste et al., 2010; Kim et al., 2012; Bagriantsev et al., 2014). Piezo1 shows broader expression in a wide range of tissue types and exerts crucial functions in a variety of biological processes, whereas Piezo2 is a mechanosensor primarily found in neurons (Bagriantsev et al., 2014; Volkers et al., 2015; Nie and Chung, 2022).

Owing to the lethality of conventional Piezo1 and Piezo2 knockout mice, conditional gene knockout (CKO) has been a popular approach (Ranade et al., 2014; Nonomura et al., 2017; Nie and Chung, 2022). Deleting Piezo1 and/or Piezo2 causes defects in multiple systems, including the skeleton (Wang et al., 2020; Zhou et al., 2020; Hendrickx et al., 2021; Nie and Chung, 2022). A deficiency of Piezo channels in osteoblasts or chondroblasts leads to skeletal malformations and a high incidence of spontaneous fractures of both the trunk and long bones (Li et al., 2019; Wang et al., 2020; Zhou et al., 2020; Hendrickx et al., 2021). Although multiple lines of evidence have shown that Piezo1 plays a more conspicuous role than Piezo2 in the skeleton, Piezo1 and Piezo2 may have partially redundant and synergistic roles during skeletal development and maintenance (Lee et al., 2014; Zhou et al., 2020; Hendrickx et al., 2021).

In comparison with the trunk and long bones, membranous bones are less affected by Piezo1 or Piezo2 deletion, which suggests specific regulatory mechanisms underlying different types of bones (Hendrickx et al., 2021; Nottmeier et al., 2023). As with the calvaria, jawbones are also primarily membranous bones, formed via direct condensation and differentiation of mesenchymal cells without using cartilage precursors, a mechanism known as intramembranous ossification. Development of jawbones is vigorously regulated by mechanical loading during embryogenesis and postnatal growth (Woronowicz and Schneider, 2019).

Unlike long bones and the calvaria, jawbones are derivatives of neural crest cells (NCCs), which are migratory and multipotent cells generated in the dorsal edges of the neural tube during early organogenesis (Erickson et al., 2023). Recent studies have demonstrated the elegant role of Piezo1 for NCC migration, supporting an early function in progenitor cell development (Canales Coutino and Mayor, 2021; Marchant et al., 2022). To better understand the functional roles of Piezo channels in NCCs and jawbones, we generated and analyzed NCC-specific Piezo1 and Piezo2 single CKO mice and Piezo1/Piezo2 double CKO mice.

Piezo1 is expressed in osteoblasts and their progenitors in developing long bones (Zhou et al., 2020). Here, we examined Piezo1 expression during jawbone development using a Piezo1 antibody. Piezo1 exhibited a high-level expression in the jawbones as early as embryonic day (E)13.5, when mineralization had just been initiated in the mandibular arches (Fig. 1A,B; Fig. S1). Robust expression was maintained throughout embryogenesis and in the postnatal period overlapping with alkaline phosphatase (Alp), an osteoblast marker (Fig. 1C-J; Fig. S1).

At three weeks postnatally, Piezo1 expression was observed in the periodontal ligament and periosteum of the alveolar bones (Fig. 1I,J), which are enriched with osteogenic cells. In contrast, Piezo1 staining was not detected in Wnt1-Cre; Piezo1^loxp/loxp CKO (hereafter Piezo1 CKO) mice, indicating a successful deletion of Piezo1 by the Wnt1-Cre driver (Fig. 1K,L).

Piezo2 displayed an expression pattern in the developing jawbones that overlapped with Piezo1, suggesting that Piezo1 and Piezo2 may act together in jawbone development (Fig. 1E-H; Fig. S1). Collectively, the robust expressions of Piezo1 and Piezo2 support their developmental roles in jawbones.

Early-stage Piezo1 deletion in skeletal progenitors leads to more striking phenotypes in mice than late time point deletion in osteocytes or chondrocytes, indicating that Piezo1 channels may play a role in regulating cell fate commitment and differentiation (Hendrickx et al., 2021; Nie and Chung, 2022). Wnt1-Cre is known for its solid and early expression in NCCs, during their specification and migration (Nie et al., 2018). The time point is far earlier than that of Runx2-Cre, Prx1-Cre, Ocn-Cre, Dmp1-Cre, Sp7-Cre, Col1a-Cre or Col2a-Cre (Li et al., 2019; Wang et al., 2020; Zhou et al., 2020; Hendrickx et al., 2021; Shen et al., 2021). Therefore, the Wnt1-Cre-mediated Piezo1 CKO mouse is a robust model for studying the roles of Piezo1 in the craniofacial skeleton as it spans the entire process of craniofacial development.

During postnatal development and growth, most Piezo1 CKO mice were viable, fertile and exhibited a normal appearance compared with controls (Fig. 2A,B,I). Genotyping at the 2 week stage showed that the ratio of each genotype agreed with what was expected, indicating no premature death before 2 weeks of age. However, we observed a small but significant number of CKO mice displaying craniofacial anomalies including malocclusions (MOs) and domed heads (DHs), which led to growth retardation (GR) and premature death (Fig. S2). Up to 6 weeks of age, incidences of MOs, DHs and GR were 2.6% (9), 1.7% (6), and 4.8% (17), respectively, from a total of 351 CKO mice, whereas none was seen in Cre-negative littermates (n=377). This contrasts with wild-type mice, which exhibit very low penetrance under laboratory conditions.

The penetrance of MO in laboratory mice varies from 0.002% to 0.089%, and the DH phenotype is seen at a penetrance of up to 0.36% depending on the genetic background (The Jackson Laboratory). There was also an excessive number of mice exhibiting GR that could not be explained by MO and DH, suggesting that additional defects existed (Fig. S2).

As most bones of the calvaria are not NCC derivatives nor directly affected by our genetic targeting, the DHs seen in Piezo1 mutants were likely caused by abnormal brain development, a condition conforming to hydrocephalus (Fig. S2). Previous studies using the Wnt1-Cre driver have also reported the DH phenotype, suggesting that Wnt1-Cre targets crucial domains affecting brain development (Dietrich et al., 2009; Zhou and Conway, 2016).

Histological examinations showed that the majority of the Piezo1 CKO mice were comparable with the Piezo1^loxp/loxp controls except the GR mice, which displayed a significant reduction in the thicknesses of the lamellar and trabecular bones in their jawbones at the 3 week stage (Fig. 2C-E; Fig. S2). We observed small clefts in the anterior palate of two GR mice, a condition conforming to a failed fusion of the primary palate with two bilateral palatal shelves (Fig. 2E). A micro-computed tomography (µCT) examination at an advanced stage was consistent with histology findings and further confirmed attenuated bone formation in the craniofacial skeleton of GR mice (Fig. 2F,G).

The Piezo1 CKO mice without GR survived to young adulthood and were indistinguishable from controls in appearance. Histology and µCT examinations revealed that their jaws were essentially normal compared with controls, aside from a slight reduction in the thicknesses of the laminar and trabecular bones in CKO mice (Fig. 2H,I,K). Of note, we also observed a small number of MOs that had manifested during adult stages (N=3 in 318 mice), suggesting that Piezo1 plays a role in maintaining tooth alignment in response to mechanical stimuli from mastication (Fig. 2H). TRAP staining showed comparable numbers of osteoclasts in control and Piezo1 CKO mice (Fig. 2J,L), suggesting that Piezo1 deletion did not significantly affect the osteoclast lineage in CKO mice, as has been observed in long bones (Wang et al., 2020).

We next performed Piezo1 and Piezo2 double staining and detected a more robust expression of Piezo2 in Piezo1 CKO mice than control, as indicated by an increase in the number of Piezo2⁺ cells in bone-lining cells (Fig. 2M-O). These data suggest a potential redundant and compensatory role of Piezo2 for the loss of Piezo1 in the jawbones and prompted us to examine Piezo1 and Piezo2 double CKO (dCKO) mice.

Previously reported dCKO mice, mediated by Runx2-Cre, Prx1-Cre and SP7-Cre, displayed more striking skeletal phenotypes than single CKO mice, suggesting functional synergy and redundancy (Zhou et al., 2020; Hendrickx et al., 2021). Unfortunately, owing to premature mouse mortality, the skeletal phenotype has not been fully explored, and jawbone development has not been previously examined in dCKO mice. Here, we sought to examine Wnt1-Cre-mediated dCKO mice. To achieve a thorough analysis, we also generated Wnt1-Cre; Piezo2^loxp/loxp CKO mice. As described in a previous report, Wnt1-Cre; Piezo2^loxp/loxp mice die at birth due to respiratory failure (Nonomura et al., 2017). Craniofacial development of Piezo2 CKO mice was comparable with littermate controls in gross and histological examinations (Fig. 3A,B; Fig. S3).

Piezo1 and Piezo2 dCKO mice were found at expected ratios at birth but failed to survive beyond the newborn stage and exhibited a spectrum of defects including small body size, attenuated bone formation, malformation of the tongue muscle and vasodilation (Fig. 3C-F). Cleft palates were also occasionally observed in dCKO mice (2/21; Fig. 3F). Attenuation of bone formation was evident in all dCKO mice at the neonatal stage (Fig. 3C-J). Rarefication and disorganization of the trabecular bones and thinned lamellar bone plates were also evident (Fig. 3C,D). Furthermore, the trabeculae of dCKO mice often failed to show the curvatures seen in controls, which may reflect bone remodeling following mechanical fluctuation of the bone marrow fluid (Fig. 3G,H). Alkaline phosphatase (Alp) and Col1a1⁺ cells were significantly reduced in dCKO mice, confirming compromised osteoblast differentiation and bone formation (Fig. 3G-J,Q). Levels of β-catenin, an intracellular mediator of the Wnt signaling pathway and a crucial regulator of osteogenesis, were markedly reduced (Fig. 3K,L,Q). Vasodilation suggested malformation of the vessel walls (Fig. 3M,N); indeed, α-smooth muscle actin staining revealed that the smooth muscle layers of the vasculatures were poorly formed in dCKO mice (Fig. 3M,N). TRAP staining revealed a comparable number of osteoclasts in dCKO mice (Fig. 3O-Q). These results indicate that compromised bone formation, but not bone resorption, accounted for the observed phenotype.

However, we did not detect any noticeable change in early craniofacial morphogenesis up to the E13.5 stage in CKO and dCKO mice, suggesting that initial NCC specification and migration were normal (Fig. S4). This is in sharp contrast to the roles of NCCs in Xenopus, where the loss of Piezo1 in cephalic NCCs leads to rapid migration due to the overactivation of Rac1 and abnormal cytoskeletal changes (Canales Coutino and Mayor, 2021). The discrepancy between mice and Xenopus in the requirement of Piezo channels during the early process of NCC development indicates a species-specific mechanism for NCC development. Alternatively, Piezo deficiency had not yet been manifested during the migration process.

Cell proliferation did not significantly change at E15.5 and later stages (Fig. 4A,B). However, we did observe excessive cell death in the maxillary and mandibular arches starting from the mid-gestation developmental period, when embryonic motility increases (Fig. 4C-F). The first wave of abnormal cell death was seen at E15.5 (Fig. 4C,D,F). Cell death was more conspicuous at the newborn stage, when the mandible begins to be regulated by vigorous mechanical loads (Fig. 4E-G). Apoptotic cells were observed in the alveolar bones and mandibular body (Fig. 4E,G; Fig. S4). As excessive cell death was not detected in Piezo1 and Piezo2 single CKO mice, it was therefore likely due to the loss of both Piezo1 and Piezo2 (Fig. 4E,G). TUNEL and Alp co-staining further demonstrated that the majority of apoptotic cells overlapped with Alp staining, suggesting that cell death was primarily seen in the osteogenic cell lineage (Fig. 4E). Excessive cell death at the newborn stage indicates that Piezo channels control the buildup of functional jawbones. Without Piezo signaling, many cells committed to osteogenic fate underwent apoptosis. A link between Piezo channel activity and the cell death pathways has been demonstrated in a variety of tissues (Kim and Hyun, 2023), and disruption of calcium homeostasis due to Piezo channel malfunction can lead to cell death through the P53 pathway (Kim and Hyun, 2023).

At the embryonic stages of jawbone development, Piezo1 channel activation may be low, as mechanical loading from oral activities is negligible. To test whether elevated Piezo channel activation promotes jawbone formation, we examined the early mineralization process in the first pharyngeal arches with and without the Piezo1 agonist Yoda1 using a tissue culture system (Alfaqeeh and Tucker, 2013). Results showed more robust mineralization in the first pharyngeal arch in the presence of Yoda1, relative to control samples cultured with vehicle only, supporting the idea that Piezo channel activation promotes bone formation during development (Fig. 4H,I).

Multiple lines of new evidence point to coupling of Piezo channels with the Wnt signaling pathway for bone regulation (Zhou et al., 2020; Hu et al., 2023). Here, we showed that Yoda1 caused a significant increase of β-catenin in cultured tissues of the mandibular arches compared with controls, suggesting that Piezo1 channels act, at least in part, via the Wnt signaling pathway (Fig. 4J-L).

In summary, jawbones were less affected by Piezo1 deletions, relative to the trunk and long bones where spontaneous fractures are prevalent (Zhou et al., 2020; Hendrickx et al., 2021). From a developmental perspective, the jawbones receive little cartilaginous contributions, which partially explains the minor defects seen in Piezo1 CKO mice. However, striking craniofacial defects in dCKO mice imply essential but partially redundant roles of Piezo1 and Piezo2 in craniofacial organogenesis including jawbone formation. Piezo channels may regulate multiple subpopulations of NCCs during organogenesis, particularly the microstructural establishment of jawbones through fine-tuning cell numbers by sensing mechanical loading. PIEZO1 and PIEZO2 mutations and malfunctions have also been implicated in an increasing number of human pathological conditions (Bagriantsev et al., 2014; Alper, 2017; Beech and Kalli, 2019; Nie and Chung, 2022). Therefore, establishing and phenotyping new models can provide insights into the roles of Piezo channels in relation to craniofacial anomalies and other pathological conditions.

Animal procedures were conducted within the National Institutes of Health (NIH) Guidelines for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the University of Maryland, Baltimore. Wnt1-Cre (Stock No. 022501), Piezo1^loxp/loxp (Stock No. 029213), Piezo2^loxp/loxp (Stock No. 027720) and wild-type mice (C57Bl6) were obtained from The Jackson Laboratory (Lewis et al., 2013; Woo et al., 2014; Cahalan et al., 2015). Genotyping was performed using regular PCR. We maintained the colony by breeding Wnt1-Cre; Piezo1^loxp/loxp mice with Piezo1^loxp/loxp mice. Piezo1^loxp/loxp mice were used as controls. To generate dCKO mice, we first established Wnt1-Cre; Piezo1^loxp/loxp; Piezo2^loxp/+ compound mice. Most compound CKO mice, like the Piezo1 CKO mice, survived and were fertile. We then crossed these mice with Piezo1^loxp/loxp; Piezo2^loxp/loxp mice to generate Piezo1 and Piezo2 dCKO mice. To generate Piezo2 single CKO mice, we crossed Wnt1-Cre; Piezo2^loxp/+ mice with Piezo2^loxp/loxp mice. Cre-negative mice were used as controls. For each stage, a minimum of three mice from each genotype were examined (Tables S1-S4).

Mouse mandibles were dissected and soft tissues were removed. Whole-mount Alizarin Red and Alcian Blue staining was performed as previously described (Nie et al., 2018).

Tissues of interest were fixed in 4% paraformaldehyde (PFA) overnight at 4°C and decalcified in 0.5 M EDTA for 2 weeks at room temperature. Decalcified tissues were further processed for embedding at the Histology Core of University of Maryland, Baltimore. Paraffin sections of 5 µm were used. Hematoxylin and Eosin (H&E) images were viewed and analyzed using Aperio Imagescope software (Leica Biosystems). Bone thickness measurements were taken on the trisection points of each bone plate. The average thickness of all bone plates was used to represent the lamellar bone thickness.

Paraffin sections of 5 µm were used. Heat-induced antigen retrieval was applied (Vector Laboratories, H-3301). Antibodies for Alp (Invitrogen, 702454), α-Smooth Muscle Actin (MilliporeSigma, A5228), β-catenin (Developmental Studies Hybridoma Bank, PY489-B-catenin), Col1a1 (Thermo Fisher Scientific, PA29569), Neurofilament (Developmental Studies Hybridoma Bank, 2H3), Phospho-Histone H3 (Millipore, 06-570), Piezo1 (Invitrogen, MA5-32876) and Piezo2 (MilliporeSigma, 587474) were used with 1/200 dilutions recommended by the manufacturers. Alexa Fluor 488-, Alexa Fluor 568- and Alexa Fluor 594-conjugated donkey anti-rabbit and goat anti-mouse secondary antibodies were used at 1/500 for signal detection (Thermo Fisher Scientific, A21206, A21207, A11029 and A11031). Each assay was repeated at least three times for each antibody. Sections were imaged and processed using a Zeiss Axiovert 200 M fluorescence microscope. Comparable sections were selected and analyzed for each genotype using ImageJ software (NIH).

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) was performed using the DeadEnd Fluorometric TUNEL System, following the manufacturer's protocol (Promega, G3250). For double staining, TUNEL was performed with a slightly extended proteinase K incubation (20 µg/ml), followed by Alp antibody incubation at 4°C overnight. Staining was repeated three times.

Tartrate-resistant acid phosphatase (TRAP) staining was performed on paraffin sections using a commercial kit (FUJIFILM Wako Pure Chemical Corporation, 294-67001) following the manufacturer's protocol with modifications (Wang et al., 2022). The Alp staining steps were skipped. The numbers of TRAP⁺ cells in the first molar areas were counted for comparison. Staining was repeated three times.

Mouse skulls and jawbones were fixed with 4% PFA overnight at 4°C, then dehydrated with a series of ethanol solutions and subjected to µCT scans (Siemens Inveon Micro-PET/SPECT/CT) (Wang et al., 2022). µCT data were analyzed using Analyze 14 software (AnalyzeDirect). Bone thickness measurements were performed at the trisection points of each bone plate at the level of the proximal cusp of the first molar. The average thickness of all measurements was used to represent the bone thickness.

Embryos at the E13.5 stage were collected from timed pregnant mice. Mouse heads were sliced into samples of 500 µm thickness. Mandibular and maxillary slices were transferred to a six-well Corning Transwell system (MilliporeSigma, TMO140644) and were cultured in DMEM containing 10% fetal bovine serum, with and without Yoda1 (MilliporeSigma, 448947-81-7) for up to 4 days (Alfaqeeh and Tucker, 2013). Yoda1 was first dissolved in DMSO (MilliporeSigma, 67-68-5) at a concentration of 10 mM and used at a concentration at 10 µM. Sample tissues were either stained with Alizarin Red or sectioned for immunofluorescence. Whole-mount samples were analyzed using ImageJ and relative mineralization areas were presented. This assay was repeated three times.

All data were generated from a minimum of three independent biological replicates with a minimum of three technical replicates. Statistical analyses were performed using GraphPad Prism 10 (Dotmatics). All data were presented as mean±s.d. The data were analyzed using unpaired two-sided t-tests between two groups and using one-way ANOVA with post hoc analysis for three and four groups. P<0.05 was considered statistically significant.

We thank Ms Sinu Kumari and Ms Cindy Zhou for their excellent technical assistance. We also thank the services provided by the Histology Core and Center for Translational Research In Imaging of University of Maryland Baltimore.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202386.reviewer-comments.pdf