CHIP inhibits odontoblast differentiation through promoting DLX3 polyubiquitylation and degradation Free

Teeth play important roles in our daily life and are essential for chewing, speaking, aesthetics and maintaining physical health (Zhai et al., 2019). Mouse molar development initiates at embryonic day 11.5 (E11.5), then goes through the bud, cap and bell stages from E12.5 to E18.5 (Thesleff and Sharpe, 1997; Yu and Klein, 2020). During the bell stage, dental mesenchymal cells differentiate into odontoblasts, which is an essential step for tooth development. Dentin, a major hard tissue of teeth, will be malformed if odontoblast differentiation is dysregulated.

Protein ubiquitylation is a dynamic post-translational modification involved in multiple biological processes. The 76 amino acid protein ubiquitin is attached to the substrates through sequential ubiquitylation processes, and E3 ubiquitin ligases play decisive roles during this process. According to the ubiquitin chains, ubiquitylation modification can be divided into polyubiquitylation and monoubiquitylation. Different ubiquitylation types play different roles in regulating the substrates, including their interactions with other proteins, subcellular localization, and their activity and stability (Hershko and Ciechanover, 1998; Swatek and Komander, 2016; Zheng and Shabek, 2017). Thus, ubiquitylation regulates many essential cellular processes. Dysfunction of ubiquitylation is implicated in the occurrence of many diseases, including neurodegeneration, cancer and immune disorders (Gilberto and Peter, 2017; Heaton et al., 2016; Popovic et al., 2014). Moreover, ubiquitylation modification regulates tissue and organ development. In tooth development, Smurf1 is the first identified E3 ubiquitin ligase that regulates odontoblastic differentiation of human dental pulp stem cells (hDPSCs) (Lee et al., 2011; Yang et al., 2014). Our recent studies have investigated two E3 ubiquitin ligases that regulate tooth development. Both murine double minute 2 (MDM2) and WW domain containing E3 ubiquitin protein ligase 2 (WWP2) can promote odontoblastic differentiation, the former through ubiquitylation of DLX3 and P53 (Zheng et al., 2020) and the latter through ubiquitylation of KLF5 (Fu et al., 2021).

Carboxyl-terminus of Hsc70 interacting protein (CHIP, also known as STUB1) has been identified as a co-chaperone of Hsc70. It has dual functions both as a co-chaperone and an E3 ubiquitin ligase (Ballinger et al., 1999; Murata et al., 2003). CHIP is involved in various cellular biological processes, such as protein refolding, degradation and trafficking (Paul and Ghosh, 2015). Accumulating evidence has demonstrated the essential role of CHIP in bone development. CHIP can target several tumor necrosis factor receptor-associated factor (TRAF) family proteins for ubiquitylation and degradation during osteoclastogensis, and Chip (Stub1)-deficient mice exhibit osteopenic phenotype with increased osteoclast formation and decreased osteoblast differentiation (Dai et al., 2003; Li et al., 2014, 2008; Min et al., 2008; Wang et al., 2018, 2020; Zhang et al., 2005). Tooth is another important mineralized organ. Tooth development involves odontoblast differentiation and dentin formation, which share some similarities with osteoblast differentiation and bone formation. However, whether CHIP is involved in regulating odontoblast differentiation and dentin formation remains unknown.

Distal-less-3 (DLX3) is a homeodomain transcription factor that is essential for tooth development. DLX3 is expressed in the dental mesenchymal cells starting from the cap stage (Zhao et al., 2000). At postnatal day 3 (P3), DLX3 is highly expressed in preodontoblasts and odontoblasts (Yang et al., 2017). Several studies have reported that mutations of DLX3 in human result in Tricho-Dento-Osseous (TDO) syndrome: a dysplasia characterized by hair, teeth and bone defects (Di Costanzo et al., 2011; Duverger et al., 2008; Haldeman et al., 2004; Price et al., 1998). Knockout of Dlx3 in mice leads to embryonic lethality between E9.5 and E10 due to placental defects (Morasso et al., 1999). Wnt1-Cre; Dlx3^F/LacZ mice develop severe dentin hypoplasia and dysplasia phenotypes, including shorter roots, thinner dentin and an enlarged pulp chamber, indicating the essential role of DLX3 during tooth development. DLX3 also binds to the Dspp promoter and positively regulates its transcription (Duverger et al., 2012). Our previous study revealed that DLX3 mediates the regulation of BMP2 in odontoblast differentiation through activating Dspp gene transcription (Yang et al., 2017). MDM2, the only known E3 ubiquitin ligase that targets DLX3, is able to monoubiquitylate DLX3 and upregulate its transcriptional activity on Dspp to facilitate odontoblastic differentiation (Zheng et al., 2020). Intriguingly, as a potential E3 ubiquitin ligase, CHIP has been predicted to interact with DLX3 at a high confidence level by the online integrated bioinformatics platform UbiBrowser (Li et al., 2017) (http://ubibrowser.ncpsb.org/ubibrowser/). Therefore, we hypothesized that CHIP might regulate tooth development by targeting DLX3. In this study, a series of in vitro and in vivo experiments were carried out to detect the expression pattern of CHIP and investigate its effects on odontoblastic differentiation of dental mesenchymal cells during tooth development, as well as the underlying mechanism.

To detect the expression pattern of CHIP during odontoblast differentiation, immunofluorescence was performed in P2 mouse incisors, which contain all stages of cells during odontoblast differentiation in one sample. In the incisor at P2, CHIP was widely expressed. It was strongly expressed in undifferentiated dental mesenchymal cells (Fig. 1Aa,b), where the two odontoblast marker genes DSPP and DMP1 were weak (Fig. S1). However, the expression of CHIP became moderate in preodontoblasts and odontoblasts (Fig. 1Ac,d), where DSPP and DMP1 were strongly expressed (Fig. S1). CHIP was also visible in both undifferentiated dental mesenchymal cells and odontoblasts at E16.5 and P2 molars (Fig. 1Ba-b′). Besides, we found that CHIP was also expressed in the inner enamel epithelium and ameloblasts (Fig. 1Ab,c and Bb). Negative control for immunofluorescence showed no positive signals (Fig. S2). Western blot using dental papilla tissues from E16.5, E18.5 and P0.5 molars further demonstrated that CHIP was downregulated during the in vivo differentiation process of odontoblasts (Fig. 1C).

RNAscope was also applied to detect mRNA expression levels of Chip. In incisors at P2, Chip transcript was visible in undifferentiated dental mesenchymal cells, preodontoblasts and odontoblasts (Fig. S3). In developing molars at both E16.5 and P2, Chip was expressed in the molar mesenchymal cells (Fig. 1Da,a′) and differentiating odontoblasts (Fig. 1Db,b′). We noticed that both mRNA and protein expression levels of CHIP were comparable in undifferentiated dental mesenchymal cells between molars at E16.5 and incisors at P2. In addition, real-time RT-PCR also demonstrated that mRNA levels of Chip were downregulated during the in vivo differentiation process of odontoblasts (Fig. 1E).

To investigate the possible function of CHIP during odontoblastic differentiation, gene overexpression and knockdown experiments were performed in mouse dental papilla cells (mDPCs) from E16.5 mouse molars. CHIP overexpression was performed using pCMV-Myc-DDK-Chip plasmids. There was a sixfold increase of Chip mRNA expression in overexpression cells compared with control (Fig. 2A). After being cultured in differentiation medium (DM) for 7 days, the expression levels of Dmp1 and Dspp were downregulated in the CHIP overexpression group compared with the control (Fig. 2B). ALP staining and ARS staining showed that ALP activities and the formation of mineralized nodules were decreased upon CHIP overexpression (Fig. 2C,D). Meanwhile, CHIP knockdown was also performed using Chip siRNA. The mRNA level of Chip was knocked down to 40% by Chip siRNA (Fig. 2E). In contrast to the overexpression experiments, Dmp1 and Dspp mRNA levels, as well as ALP activities and the mineralized nodule formation were all upregulated in CHIP knockdown group (Fig. 2F-H).

As CHIP was strongly expressed in the undifferentiated dental mesenchymal cells, whether CHIP regulates cell proliferation or apoptosis was examined. Results of Ki67 immunofluorescence and TUNEL assays showed that the proliferation and apoptosis of dental mesenchymal cells were not significantly changed in CHIP overexpression and knockdown groups (Fig. S4).

To confirm the in vivo function of CHIP in tooth development, Chip^−/− mice were obtained by crossing Chip^+/− mice. Real-time RT-PCR and immunofluorescence confirmed that Chip was completely depleted in molar tissues of Chip^−/− mice (Fig. S5). The phenotype of molars in Chip^−/− mice at P1 and P10 stages was observed. At P1, odontoblast-secreted predentin matrix was apparently increased in the molar cusp area of Chip^−/− mice (Fig. 3A-D). And the dentin layers of P10 molars were obviously widened in Chip^−/− mice (Fig. 3E-H′). Moreover, X-ray and micro-CT analysis of molars at P10 verified that hard tissue formation was increased in Chip^−/− mice (Fig. 3I-L).

To precisely calculate changes in the dentin formation in Chip^−/− mice, the widths of P10 maxillary and mandibular molar dentin were measured in one HE-stained section every five consecutive slices and compared with control (red lines in Fig. 3E-H). The thickness of dentin was determined as the ratio of the average of all five measurements to the body weight. Both the maxillary and mandibular molar dentin were widened (Fig. 3O). The dentin volume was also quantified using 3D reconstructed molars, which supported more crown dentin formation in P10 Chip^−/− mice (Fig. 3M,N,P).

Meanwhile, the phenotype of incisors upon Chip deletion was also observed by X-ray, HE staining and micro-CT (Fig. 3I,J; Fig. S6A-C). The results showed that the dentin was widened in the incisors of Chip^−/− mice, similar to the phenotype of molars.

To explore the in vivo regulation of odontoblast differentiation by CHIP, the expression levels of DSPP and DMP1, two odontoblast differentiation markers, were detected using immunofluorescence in mouse molars from E16.5 to P1. At E16.5, DMP1 signals were weak and comparable between Chip^−/− and control molars where DSPP was not yet expressed (Fig. 4A,B,I,J). At E18.5, odontoblasts and underlying dental mesenchymal cells at the cusp region started to express DSPP. DSPP expression in Chip^−/− mice was obviously stronger than that in control (Fig. 4C,D). However, the expression of DMP1 was comparable between Chip^−/− and control molars at E18.5 (Fig. 4K,L). In molars of Chip^−/− mice at P1, it was found that differentiating odontoblasts expressed more DSPP and DMP1 protein (Fig. 4E,F,M,N). Mean fluorescence intensity of dental mesenchyme in Fig. 4A′-L′ were calculated using ImageJ. The results showed that the protein levels of DSPP were increased in the dental mesenchyme of E18.5 and P1 Chip^−/− mice, and that of DMP1 was increased in the dental mesenchyme of P1 Chip^−/− mice (Fig. S7A). Apart from changes in odontoblasts, deletion of Chip also led to enhanced DSPP and DMP1 expression in ameloblasts (Fig. 4E,F,M,N).

mRNA expression levels of Dspp and Dmp1 in molars at P1 were also detected by RNAscope. Both Dspp and Dmp1 were increased in the odontoblasts of Chip^−/− mice (Fig. 4G,H,O,P). In addition, an increase of Dspp and Dmp1 mRNA was also observed in Chip-deficient ameloblasts.

Real-time RT-PCR using the dental papilla tissues from control and Chip^−/− mice at E16.5, E18.5 and P1 were also performed. The results showed that the expression levels of Dspp were increased in dental papilla tissues from E18.5 and P1 Chip^−/− mice, and that of Dmp1 was increased in P1 Chip^−/− mice (Fig. 4Q). These data suggest that Chip deletion could promote the differentiation of dental mesenchymal cells in vivo.

Possible changes in proliferation and apoptosis of dental mesenchymal cells in Chip^−/− mice were also analyzed. Cell proliferation in molars at E16.5 and P1 were detected by Ki67 immunofluorescence. The ratio of Ki67-positive dental mesenchymal cells in the Chip^−/− mice was comparable with controls at the stages examined (Fig. S7B-E′,N), while the negative control stained with IgG showed no positive signal (Fig. S7J-M). TUNEL assays of P1 molars showed that there was no significant difference in the apoptotic cells between Chip^−/− and control mice (Fig. S7F-I,O).

Among several proteins essential for odontoblast differentiation, DLX3 was predicted to interact with CHIP using the online integrated bioinformatics platform UbiBrowser (http://ubibrowser.bio-it.cn/ubibrowser/home/index). DLX3 protein level was increased with odontoblast differentiation as detected in P2 mouse incisors (Fig. S8A-A‴), which was contrary to CHIP. As Dspp is a direct downstream target of the transcription factor DLX3 in tooth development (Yang et al., 2017), and Dspp expression was upregulated in CHIP knockdown odontoblastic cells and in odontoblasts in Chip^−/− mice. We hypothesized that CHIP might target DLX3 protein for ubiquitylation and degradation.

First, the physical binding between CHIP and DLX3 was analyzed in cultured cell types and in mouse molars. Myc-DDK-Chip or/and His-Dlx3 plasmids were co-transfected into human embryonic kidney 293E (HEK293E) cells, co-IP assay showed that overexpressed CHIP interacted with DLX3 (Fig. 5A). In addition, interaction of endogenous CHIP and DLX3 was confirmed in mDPCs either in the absence or in the presence of differentiation induction (Fig. 5B). An in situ proximity ligation assay (PLA) using CHIP and DLX3 antibodies also demonstrated that CHIP interacted with DLX3 in the undifferentiated dental mesenchymal cells and the preodontoblasts of P1 molars (Fig. 5C).

Next, whether CHIP ubiquitylated DLX3 was investigated. Co-IP results showed that CHIP overexpression led to DLX3 ubiquitylation in HEK293E cells (Fig. 5D). In situ PLA further confirmed that ubiquitylated DLX3 was present in the undifferentiated dental mesenchymal cells and the preodontoblasts (Fig. 5E). But DLX3 ubiquitylation level was decreased when CHIP was knocked down (Fig. 5F). Ubiquitylation results in different consequences of the substrates, among which protein degradation is the most frequent. Therefore, whether DLX3 was targeted for degradation by CHIP was examined. Western blot showed that overexpression of CHIP in mDPCs and in HEK293E cells reduced the protein level of endogenous and overexpressed DLX3, respectively (Fig. 5G,H), while overexpression of CHIP^H261Q (mutation of His 261 to Gln leads to destruction of CHIP ubiquitin ligase activity; Yang et al., 2011) could not induce DLX3 degradation (Fig. 5I). Meanwhile, the mRNA level of Dlx3 was not significantly altered in CHIP overexpression group, which excluded transcriptional regulation of DLX3 by CHIP (Fig. 5J). To further confirm that CHIP promoted DLX3 degradation through ubiquitylation, cycloheximide (CHX) pulse-chase assays were carried out. Overexpression of wild-type CHIP reduced the half-life of DLX3, while overexpression of mutant CHIP^H261Q did not show this effect (Fig. 5K,L). The application of MG132 (a proteasome inhibitor) but not Bafilomycin A1 (Baf A1, a lysosome inhibitor) inhibited CHIP-induced DLX3 degradation (Fig. 5M), suggesting that CHIP triggered proteasomal rather than lysosomal degradation of DLX3. Immunofluorescent data confirmed that DLX3 protein level was increased in P1 Chip^−/− mice (Fig. 5N). Therefore, CHIP promotes degradation of DLX3 through ubiquitin-proteasome pathway. K48- and K63-polyubiquitin chains are the major signals for proteasome-mediated degradation (Grice and Nathan, 2016; Ohtake et al., 2018). In order to define the ubiquitylation type of DLX3 mediated by CHIP that results in proteasome-mediated degradation of DLX3, we co-transfected HEK293E cells with plasmids expressing CHIP, DLX3 and ubiquitin in which K48 or K63 is mutated to arginine, and then performed co-IP experiments. The results indicated that CHIP promoted K63-linked polyubiquitylation of DLX3 (Fig. 5O). Using K48- or K63-only ubiquitin plasmid (all lysines are mutated to arginines except for K48 or K63), we further confirmed that CHIP promoted K63-linked polyubiquitylation of DLX3 (Fig. 5P).

To confirm whether increased DLX3 accounted for the enhanced odontoblastic differentiation of mDPCs by CHIP deletion, a series of rescue experiments were performed using DLX3 siRNA. DLX3 mRNA levels were reduced to 20% by DLX3 siRNA-2 (Fig. 6A). After cell culture in DM for 5 days, CHIP knockdown upregulated Dspp, Dmp1 and Alp expression, while knockdown of DLX3 remarkably alleviated effects of CHIP knockdown on these genes (Fig. 6B). In addition, concomitant knockdown of DLX3 partially reversed changes in the formation of mineralized nodules, as well as the ALP activity (Fig. 6C-D′). Dspp is a direct downstream target of DLX3, so the transcriptional activity of the Dspp promoter was measured using the dual luciferase assay. The results showed that overexpression of CHIP alone was able to inhibit the transactivation of the Dspp promoter. Simultaneous overexpression of DLX3 and CHIP rescued the transcriptional inhibition on the Dspp promoter by CHIP overexpression (Fig. 6E). Therefore, DLX3 is a key mediator for the negative effect of CHIP on odontoblastic differentiation.

MDM2, a well-recognized E3 ubiquitin ligase, also targets DLX3 as a tooth-specific substrate. In contrast to CHIP, MDM2 plays a positive role during odontoblastic differentiation (Zheng et al., 2020). Therefore, whether CHIP or MDM2 interacted with and influenced each other raised our interest. Using the online integrated bioinformatics platform UbiBrowser (Li et al., 2017), MDM2 may interact with CHIP. Nevertheless, co-IP assay revealed that the interaction of MDM2 with CHIP did not exist in undifferentiated or differentiation-induced mDPCs (Fig. 7A,B). Meanwhile, CHIP overexpression did not change the protein level of MDM2 (Fig. 7C), and neither did MDM2 overexpression lead to CHIP degradation (Fig. 7D).

Whether CHIP and MDM2 competitively interacted with DLX3 was then investigated. Co-IP assay showed that with the increasing levels of CHIP expression, the interaction between MDM2 and DLX3 was decreased (Fig. 7E). The expression levels of MDM2 showed an increasing trend with the differentiation of odontoblasts (Zheng et al., 2020), which was opposite to that of CHIP. Therefore, the in vivo interaction of MDM2 or CHIP with DLX3 was explored by in situ PLA assay. CHIP-DLX3 complexes were found at a high level in undifferentiated dental mesenchymal cells but decreased in the secretory odontoblasts. On the contrary, the interaction between MDM2 and DLX3 was weak in undifferentiated dental mesenchymal cells but enhanced in the secretory odontoblasts (Fig. 7F,F′). Further investigation demonstrated that the competitive binding of MDM2 impeded the degradation effect of CHIP on DLX3 (Fig. 7G). Meanwhile, CHIP antagonized the promotion effect of MDM2 on the transcriptional activity of DLX3 and the activation of Dspp – the downstream target gene of DLX3 (Fig. 7H).

Ubiquitylation is an important form of post-translational modifications and influences a variety of developmental events. Several ubiquitin ligases have been identified to be involved in tooth development (Fu et al., 2022, 2021; Lee et al., 2011; Zheng et al., 2020). CHIP is an E3 ubiquitin ligase that plays crucial roles in various physiological and pathological processes through regulating the stability of substrate proteins (Cao et al., 2016). In this study, we aimed to illustrate the possible role of CHIP in tooth development and to elucidate the underlying molecular mechanisms. DLX3 is a key mediator of tooth development and regulates both proliferation of human dental papilla cells (hDPCs) and odontoblast differentiation (Li et al., 2012; Zhan et al., 2018). The present study demonstrates that CHIP inhibits odontoblast differentiation and dentinogenesis through targeting DLX3 for polyubiquitylation and subsequent proteasomal degradation.

The expression pattern of CHIP in tooth development was observed in mouse incisor at P2 and mouse molars from E16.5 to P1 using immunofluorescence. As mouse incisors contain dental mesenchymal cells at different differentiation stages, the expression of CHIP in undifferentiated to fully differentiated odontoblasts could be viewed in one sample. Immunofluorescence in P2 incisor and real-time RT-PCR results using dental papilla tissues showed that CHIP was highly expressed in the undifferentiated dental mesenchymal cells but decreased during odontoblast differentiation in vivo. Another E3 ubiquitin ligase MDM2 was slightly expressed in the undifferentiated dental mesenchymal cells but increased during odontoblast differentiation (Zheng et al., 2020).

Cell proliferation, differentiation and death are essential biological processes in dental mesenchymal cells during tooth development (Abramyan et al., 2021; Matalova et al., 2004; Sharpe, 2001). Ectopic expression and knockdown of CHIP revealed that CHIP could not alter cell proliferation or apoptosis but inhibit odontoblastic differentiation of mDPCs in vitro. However, MDM2 could promote odontoblastic differentiation in vitro (Zheng et al., 2020). To further investigate the function of CHIP in regulating tooth development, Chip^−/− mice were obtained. Chip^−/− mice have been demonstrated to exhibit increased sensitivity to stress-associated hyperthermia, severe defects in heart and neural tissues as well as in motor sensory, cognitive and reproductive function (Dai et al., 2003; Dickey et al., 2006; Ghoul-Mazgar et al., 2005; Zhang et al., 2005), leading to a markedly reduced lifespan (Min et al., 2008). Thus, Chip^−/− mice died soon after birth. Indeed, Chip^−/− mice mostly died before P10 in our study. The predentin at P1 and the dentin at P10 were significantly thicker in Chip^−/− mice than in controls. Increased DSPP and DMP1 expression levels demonstrated CHIP deletion resulted in enhanced odontoblast differentiation. Similar to in vitro experiments, CHIP could not alter proliferation or apoptosis of mDPCs and odontoblasts in vivo.

The expression of CHIP is also strongly expressed in the inner enamel epithelium and ameloblasts, and deletion of CHIP also increases the expression of DSPP and DMP1 in ameloblasts, suggesting that CHIP might also suppress ameloblast differentiation. Besides, it is well known that the reciprocal interactions and regulations between dental epithelium and mesenchyme play a pivotal role in tooth development (Balic and Thesleff, 2015). A previous study showed that conditional knockout of Wls in dental epithelium leads to defective ameloblast and odontoblast differentiation (Xiong et al., 2019). Conversely, deletion of mesenchymal Sufu results in the disruption of cell proliferation and programmed cell death in both dental epithelium and mesenchyme (Li et al., 2019). Therefore, it is possible that the changes on ameloblasts also affect odontoblast differentiation that is distinct from the direct effect of CHIP on odontoblast differentiation.

In contrast to its inhibitory role in odontoblast differentiation and dentinogenesis, CHIP was previously found to facilitate osteoblast differentiation and bone formation because Chip^−/− mice exhibited decreased osteoblast differentiation and osteopenic phenotype (Li et al., 2014; Wang et al., 2018). However, MDM2 was proved to promote odontoblast differentiation in vivo, as mice with MDM2 deletion in odontoblasts displayed decreased odontoblast differentiation and dentin formation ability (Zheng et al., 2022).

In most cases, ubiquitylation by CHIP influences protein stability and function through substrate degradation (Shi et al., 2020; Ullah et al., 2020; Xu et al., 2020; Yonezawa et al., 2017; Zhang et al., 2020). In our study, in vitro and in vivo experiments showed that CHIP interacted with and induced K63-polyubiquitylation of DLX3, leading to DLX3 degradation. Simultaneous knockdown of DLX3 and CHIP rescued the enhanced differentiation of mDPCs by knockdown of CHIP alone, suggesting that DLX3 mediated the function of CHIP in odontoblastic differentiation. DLX3 is a well-recognized transcription factor that is essential for odontoblast differentiation (Choi et al., 2010; Duverger et al., 2012). Inhibiting DLX3 results in inhibition of odontoblast differentiation (Duverger et al., 2012). During osteoclastogenesis, CHIP interacted with TRAF6, TRAF2 and TRAF5, and promoted their ubiquitylation and degradation. Thus, Chip^−/− mice exhibited increased osteoclast formation and osteopenic phenotype (Li et al., 2014; Wang et al., 2018). Protein degradation mediated by ubiquitylation has been found to depend on proteasomal and/or lysosomal pathways (Dikic, 2017). Application of MG132 and Baf A1 (the respective proteasome and lysosome inhibitor) demonstrated that CHIP promoted DLX3 degradation via proteasome system. Similarly, a previous study found that CHIP promoted degradation of nuclear factor of activated T cells 3 (NFATc3) through a ubiquitin-proteasome pathway in lipopolysaccharide-induced cardiac hypertrophy and apoptosis (Chao et al., 2019).

MDM2 has been verified to monoubiquitylate DLX3, promoting the transcriptional activity of DLX3 on its downstream gene Dspp (Zheng et al., 2020). CHIP and MDM2 both target DLX3 during odontoblast differentiation process. However, our experiments proved that CHIP and MDM2 did not interact with each other. Some previous studies have explored two E3 ubiquitin ligases that antagonistically regulate the same substrate. For example, TRIM67 antagonizes TRIM9-dependent VASP ubiquitylation by outcompeting TRIM9 for binding with VASP during axon guidance and accurate neuronal connection formation (Boyer et al., 2020). Therefore, it was speculated that CHIP and MDM2 may have a competitive effect on DLX3 during the odontoblast differentiation process. The results of co-IP assays confirmed this perspective. Interestingly, it has been shown that CHIP interacts with DLX3 mainly in undifferentiated dental mesenchymal cells while MDM2-DLX3 complexes are mainly seen in the odontoblasts. This trend is consistent with the expression patterns of CHIP and MDM2 during odontoblast differentiation. The competitive binding of MDM2 versus CHIP to DLX3 results in a decrease in DLX3 degradation and an increase in the expression of the DLX3 downstream gene Dspp.

Taken together, this study shows, for the first time, that CHIP expression is downregulated with odontoblast differentiation. In vitro and in vivo evidence indicates that CHIP inhibits odontoblast differentiation and dentin formation through interacting with and polyubiquitylating DLX3 for degradation. However, MDM2, which is upregulated during odontoblast differentiation, competitively binds to and monoubiquitylates DLX3, and increases DLX3 transcriptional activity on Dspp to promote odontoblast differentiation and dentin formation. Therefore, CHIP and MDM2 reciprocally regulate DLX3 activity by catalyzing distinct type of ubiquitylation (Fig. 8). Our findings suggest an important mechanism by which the differentiation of odontoblasts is delicately regulated at the transcription factor level through divergent ubiquitylation modifications.

Chip^−/− mice were obtained from Dr Chen's laboratory (Wang et al., 2018). All mice were housed in a specific pathogen-free (SPF) animal facility of Wuhan University. All animal experiments satisfied the requirements of protocols approved by the Animal Welfare and Ethics Committee of the School and Hospital of Stomatology at Wuhan University (S07919010B).

The primary mDPCs were isolated from the first mandibular molars of fetal Kunming mice at E16.5, as previously described (Xiao et al., 2021). The mDPCs and HEK293E cells were cultured in Dulbecco's Modified Eagle Medium (DMEM; Hyclone) containing 10% fetal bovine serum (FBS; Gibco) and 100 U/mL of penicillin/streptomycin (Hyclone) at 37°C in 5% CO₂. To induce odontoblastic differentiation of mDPCs, the cells were incubated in differentiation medium containing 10 mM sodiumβ-glycerophosphate (Sigma-Aldrich), 50 μg/ml of ascorbic acid (Sigma-Aldrich) and 10 nM dexamethasone (Sigma-Aldrich).

The mouse dental papilla were separated from E16.5, E18.5 and P1 Kunming mice. Briefly, the molar tooth germs were isolated under the stereomicroscope. The epithelial and mesenchymal tissues were separated after digestion in 0.5% trypsin.

siRNAs used for in vitro knockdown experiments were purchased from Genepharma (Suzhou, China), including siRNA-Chip (5′-3′ CCCUGUGCUAUCUGAAGAUTT, AUCUUCAGAUAGCACAGGGTT), siRNA-Dlx3 [(1) 5′-3′ GCUCCUCAGCAUGACUACUTT, AGUAGUCAUGCUGAGGAGCTT; (2) 5′-3′ GCCGUUUCCAGAAAGCCCATT, UGGGCUUUCUGGAAACGGCTT; (3) 5′-3′ GCCUCACACAAACACAGGUTT, ACCUGUGUUUGUGUGAGGCTT] and scramble siRNA (5′-3′ UUCUCCGAACGUGUCACGUTT, ACGUGACACGUUCGGAGAATT).

For in vitro overexpression experiments, pCMV-Myc-DDK-Chip plasmid was purchased from OriGene (plasmid MR204258), and pCMV-Myc-DDK-Chip (H261Q) (mutation of His 261 into Gln, which leads to E3 activity deficiency) was prepared by using Mut Express II Fast Mutagenesis Kit V2 (Vazyme) according to the manufacturer's protocol. pcDNA3.1(+)-Dlx3-6His was purchased from GeneCreate and pCMV-HA-Ub was purchased from Miaoling. pRK-HA-K48R, pRK-HA-K63R, pRK-HA-K48O and pRK-HA-K63O were kindly provided by Professor Hong-Bing Shu from Wuhan University. HEK293E cells were transfected using Lipofectamine 2000 Transfection Reagent (Invitrogen), and mDPCs were transfected using Lipofectamine 3000 Transfection Reagent (Invitrogen) according to the manufacturer's protocols.

Total RNA was extracted from mDPCs using a RNA extraction kit (Vazyme). cDNA was synthesized from 1 μg RNA by using the HiScript II Q RT SuperMix for qPCR (Vazyme). Real-time RT-PCR was performed using ChamQ SYBR qPCR Master Mix (Vazyme) under the BIORAD CFX RT-PCR system. The following PCR primers were used: Dmp1-F, ACCACAATACTGAATCTGAAAGCTC; Dmp1-R, TGCTGTCCGTGTGGTCACTA; Dspp-F, GTGGGATCATCAGCCAGTCAG; Dspp-R, TGCCTTTGTTGGGACCTTCA; Chip-F, CCTATGACCGCAAGGACCATT; Chip-R, CTCTACCCAGCCGTTCTCAG; Dlx3-F, CAGCAGCATCCTCACCGACATC; Dlx3-R, TGAACTGGTGGTGGTAGGTGTAGG; Gapdh-F, TGTGTCCGTCGTGGATCTGA; Gapdh-R, TTGCTGTTGAA-GTCGCAGGAG.

Mouse heads and mandibles at different developmental stages were isolated and fixed in 4% paraformaldehyde for 1 day at 4°C, then decalcified in 10% EDTA. The tissues were embedded in paraffin wax after graded alcohol dehydration and sliced into 5 µm sections. Tissue slides or cells were blocked with 3% BSA for 1 h at 37°C and incubated with primary antibodies at 4°C overnight. Primary antibodies were diluted in PBS as follows: CHIP (1:250, Abcam, ab134064), DMP1 (1:100, Abclonal, A16832), DSPP (1:100, Novus, NBP2-92546), Ki67 (1:100, Novus, NB110-89717), DLX3 (1:50, Proteintech, 13261-3-AP) and MDM2 (1:100, Abcam, ab226939). After incubation, the samples were washed three times with PBS and incubated with the fluorescence-conjugated secondary antibodies for 1 h at room temperature. The secondary antibodies were Alexa Fluor Red 594 donkey anti-rabbit IgG (1:200; Antgene, ANT030S) and Alexa Fluor Red 594 donkey anti-mouse IgG (1:200; Antgene, ANT029S). Sections or cells were counterstained with DAPI.

For RNAscope detection, the specific probes for target RNAs were designed by spatial FISH (Wuhan, China). Samples were fixed using 4% paraformaldehyde and made into slices. After dehydration and denaturation with methanol, samples were treated with the hybridization buffer and specific targeting probes and then incubated at 37°C overnight. Next, samples were washed then incubated with target probes in ligation mix at 25°C for 3 h. Subsequently, samples were washed and treated with rolling circle amplification using Phi29 DNA polymerase at 30°C overnight. The fluorescence detection probes were then applied. Finally, samples were dehydrated and mounted.

Alkaline phosphatase (ALP) staining was performed following the standard procedure using the ALP staining Kit (Beyotime). For Alizarin Red S (ARS) staining, cells were stained with 1% ARS solution (pH=4.8) made using ARS powder (Aladdin) to identify the presence of calcium deposits.

A co-immunoprecipitation (co-IP) assay was performed as described previously (Zheng et al., 2020). In brief, the cells were lysed by NP-40 lysis buffer (Beyotime) containing protease inhibitor cocktail (MCE). After centrifugation, cell lysates were incubated with the following antibodies: anti-CHIP (Abcam), anti-DLX3 (Proteintech), anti-HIS (Abcam) and anti-HA (Abcam).

Protein expression levels were analyzed by western blot as described previously (Fu et al., 2021). The primary antibodies used were as follows: anti-β-actin (1:10,000, Abclonal), anti-CHIP (1:1000, Abcam), anti-DLX3 (1:1000, Proteintech), anti-HA (1:1000, Abcam), anti-HIS (1:1000, Abcam) and anti-MYC (1:1000, Abcam).

To identify physical interaction between proteins, in situ PLA was performed as described previously (Alam, 2018). The slices of tissues were permeabilized with 0.1% Triton X-100 for 10 min and washed with PBS. After being blocked for 1 h at 37°C, the slices were incubated with primary antibodies overnight at 4°C. Then the PLA probe solution, ligation-ligase solution and amplification-polymerase solution were added to each sample in turn and incubated. Finally, the slices were mounted with coverslips using mounting medium with DAPI.

Statistical analyses of all data were performed with GraphPad Prism 8 (GraphPad Software) and are presented as mean±s.d. Comparisons were performed using a two-tailed Student's t-test or one-way analysis of variance (ANOVA). P<0.05 was considered statistically significant for all analyses. All experiments were independently repeated at least three times and n≥3 for all samples.

We thank Professor Hong-Bing Shu for advice.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.200848.reviewer-comments.pdf.