OGT mediates O-GlcNAcylation of MEIS2 and affects palatal osteogenic development

Abstract

Post-translational modifications (PTMs) have been gradually elucidated in congenital malformations such as cleft palate. Among them, O-GlcNAcylation as a dynamic PTM of proteins regulates various critical biological processes including transcription, translation, and cell fate determination. In this study, a substantial decline in O-linked β-D-N-acetylglucosamine (O-GlcNAc) levels was detected within the palatine plates of all-trans retinoic acid (atRA)-induced cleft palate mice. The role of O-GlcNAc transferase (OGT), the sole enzyme responsible for catalyzing O-GlcNAcylation, was investigated in the process of palatal development. In a zebrafish model, the loss of O-GlcNAc resulted in an elevated prevalence of cleft palate and compromised palatal bone formation. Mechanistically, O-GlcNAcylation of myeloid ecotropic viral integration site 2 (MEIS2), which is mediated by OGT, was found to maintain osteogenic homeostasis by modulating its protein stability through inhibition of ubiquitination. Notably, the serine 237 residue (Ser237) was identified as a critical site for MEIS2 O-GlcNAcylation. Together, the present study uncovers the important function of MEIS2 O-GlcNAcylation in palatal bone development and establishes a novel theoretical framework for understanding the regulatory network of palatal development. This finding may provide novel avenues for the future diagnosis and prevention of cleft palate.

Introduction

As the most prevalent congenital deformity in the oral and maxillofacial region, cleft lip and palate (CLP) represents the most common craniofacial birth defect of human beings.^1,2 Its incidence varies across different populations and is relatively higher in Asian and Indigenous American populations and lower in African populations.³ Congenital cleft palate can be classified into syndromic and non-syndromic types based on whether it is accompanied by other deformities.⁴ Non-syndromic cleft palate, which constitutes 80% of all congenital cleft palate cases, manifests as an isolated malformation. In line with the embryogenesis and etiology of syndromic cleft palate or non-syndromic cleft lip, those of non-syndromic cleft palate involve multiple stages and factors resulting from the interplay of genetic and environmental influences,⁵ such as hyperglycemia and Benzo[a]pyrene which serve as contributing factors to palatal developmental abnormalities. It is hypothesized that disruptions occurring due to the aforementioned factors during the early embryonic stage of development may result in palatal dysmorphogenesis.^6,7

Secondary palate development primarily occurs between 6 and 12 weeks of gestation in humans but spans embryonic days (E) 13.5–15.5 in mice. This process is about palatal shelf growth, elevation and fusion.⁸ Bilateral palatal shelves undergo vertical growth alongside the tongue initially and reorient horizontally above it subsequently. The elevated shelves elongate toward the midline, come into contact with each other, and form the medial edge epithelium (MEE). Later, the MEE undergoes adhesion, disintegration and mesenchymal confluence, which ultimately contributes to the formation of an intact palate. Any disruption in these steps, such as failed shelf elevation, impaired contact and post-fusion breakdown, may result in cleft palate.^9–11 It has been demonstrated that the palatal epithelium and mesenchyme play critical roles, with the mesenchyme constituting the primary structural component essential for proper palatogenesis.¹² The anterior hard palate derives from maxillary palatal processes, while the posterior portion originates from palatine bones. The primary palate comprises a tiny triangular bony segment that extends from anterior maxillary processes to the incisor alveolar region.^8,13 Palatal osteogenesis takes place via intramembranous ossification, wherein condensed mesenchyme derived from the neural crest directly disintegrates into osteoblasts.¹⁴ The hard palate is actively ossified and vascularized from E14.5 to E16.5 in mice, with mesenchymal condensation, extracellular matrix (ECM) calcification, and the initiation of bone remodeling at E14.5, E15.5, and E16.5, respectively.¹³ Post-fusion defective osteogenesis may give rise to submucous cleft palate (SMCP). Despite its clinical prevalence, murine models of SMCP have only recently provided valuable insights into the formation of secondary palatal bones.¹⁵

Also called MRG1, myeloid ecotropic viral integration site 2 (MEIS2) is a homeobox gene that is part of the three-amino-acid-loop-extension (TALE) superclass. It harbors a conserved homeothorax (Hth) domain that mediates interactions with pre-B-cell leukemia homeobox proteins and facilitates deoxyribonucleic acid (DNA) binding.¹⁶ As an evolutionarily conserved transcription factor, MEIS2 is involved in a variety of developmental processes, including limb patterning, cardiogenesis, and neurogenesis. Prior studies have implicated MEIS2 in the pathogenesis of cleft palate potentially through haploinsufficiency or point mutations. Heterozygous loss-of-function variants in MEIS2 cause palatal defects, intellectual disability, and congenital heart disease.¹⁷ The neural crest-specific inactivation of MEIS2 results in cleft palate/SMCP and complete palatal bone loss. Mechanistically, Wang et al. demonstrated a functional interplay between short stature homeobox 2 (SHOX2) and MEIS2 in regulating palatal osteogenesis via chromatin immunoprecipitation followed by sequencing (CHIP-seq) assays.¹⁸ Nevertheless, the precise pathogenic mechanisms remain elusive.

O-linked β-D-N-acetylglucosamine (O-GlcNAc) is a prevalent post-translational modification (PTM) that targets serine (Ser or S)/threonine residues of nuclear, membranous, cytoplasmic, and mitochondrial proteins, including kinases, metabolic enzymes, transcription factors, signaling molecules, etc.¹⁹ The O-GlcNAc pathway functions as a cellular “nutrient sensor”, responds to metabolic status and modulates protein activity, energy metabolism, cell cycle progression, transcriptional regulation and protein stability. Unlike phosphorylation, O-GlcNAcylation is subjected to catalysis solely by O-GlcNAc transferase (OGT) or removal by O-GlcNAcase (OGA).^20,21 As revealed by recent studies, OGT conditional knockout (cKO) mice exhibited no overt long bone defects but severe mineralization deficits in flat cranial bones, which indicated the role of OGT in intramembranous ossification during the perinatal period. Additionally, OGT cKO mice demonstrated significant dental anomalies, which underscored the importance of OGT in craniofacial development.^22,23

In the present study, the role played by O-GlcNAcylation in palatal osteogenesis and progression was partly explored. O-GlcNAcylation was found to exhibit a marked decrease in the palatal plate of all-trans retinoic acid (atRA)-induced cleft palate mice, and O-GlcNAcylation was observed to inhibit the ubiquitination of MEIS2 at the Ser237 locus to maintain the stability of the MEIS2 protein, which thereby affected palatal osteogenic differentiation. To conclude, this study is an important resource for understanding the molecular mechanisms of the palate that will exert an impact on future CLP therapy.

Results

Down-regulation of O-GlcNAc in atRA-induced mice and its involvement in the regulation of palatal development

Prior research has demonstrated that the hard palate derived from the secondary palate includes the palatal process of the maxilla (ppmx) and the palatal process of the palatine (ppp). The period of active palatal ossification is concomitant with the developmental stages E14.5 to E16.5. We observed osteogenic patterns consistent with the above description through Von Kossa staining (Fig. 1a) and Masson’s trichrome staining (Fig. 1b), which highlighted the palate bone area and collagen deposition during these stages. IHC staining revealed that O-GlcNAc modification (Fig. 1c) and OGT (Fig. 1d) were highly expressed in the palatal epithelium and mesenchyme during mouse palatal development, with increased expression patterns observed at the later stages of palatal fusion, specifically after E14.5 (Figure S1A, B). The atRA-induced mouse model is a common tool for studying the mechanisms of cleft palate and defects in hard palate ossification²⁴. To investigate the molecular changes, Western Blot (WB) analyses were performed, which revealed that O-GlcNAc and OGT protein levels exhibited a significant down-regulation in the palatal shelf tissues of atRA-induced mice compared to controls (Fig. 2a, b). Further immunohistochemical (IHC) analyses confirmed these findings spatially. O-GlcNAc modification was significantly reduced in the atRA-induced mouse cleft palate model (Fig. 2c, Figure S1C, D). Collectively, OGT expression was also found to be down-regulated in the atRA group (Fig. 2d, Figure S1E, F). Overall, these results demonstrated the essential role of O-GlcNAc in palate development.

Fig. 1.

OGT and O-GlcNAcylation play critical roles in palatal development. a Von Kossa staining and (b) Masson’s trichrome staining on coronal sections from E14.5 to E16.5. The black dashed line indicates the palate bone area. c, d Representative Immunohistochemistry staining of OGT and global O-GlcNAcylation on coronal sections from E14.5 to E16.5. Scale bars: 100 μm

Fig. 2.

Down-regulation of O-GlcNAcylation in atRA-induced mice and its involvement in the regulation of palatal development. a Western blot analysis of OGT and O-GlcNAc expression in palatal shelf tissues in control mice and atRA mice. b Semi-quantification data for relative protein levels in (a). c Representative Immunohistochemistry staining of global O-GlcNAcylation on coronal sections at E15.5. d Representative Immunohistochemistry staining of OGT on coronal sections at E15.5. n = 3 biologically independent experiments. ppmx palatal process of the maxilla, ppp palatal process of the palatine

Cleft palate and impaired parasphenoid bone formation caused by the loss of O-GlcNAcylation in zebrafish

Tg(sox10: egfp) zebrafish was taken as an animal model to examine the effect of O-GlcNac on craniofacial and palatal development in vivo. ogt-knockdown zebrafish were generated by injecting antisense Mos (morpholines) targeting ogt (ogt MO) and standard control MO (con MO) into zebrafish embryos, and treated with OSMI-1 to pharmacologically inhibit O-GlcNAcylation in zebrafish. After the efficiency of ogt-knockdown and OSMI-1 inhibition of O-GlcNAc via RT-qPCR was confirmed (Figure S2A), the rate of palatal defects in zebrafish treated with ogt MO and OSMI-1 was far above that in the control group (Fig. 3a). The Alcian blue staining of craniofacial cartilage revealed a significantly higher rate of cleft palate in O-GlcNAc-inhibited zebrafish at 120 hpf compared to that in the control group (Figure S2B, C). Additionally, the Alizarin Red staining of craniofacial bones showed that the area of parasphenoid bones (Fig. 3b) in O-GlcNAcylation-inhibited zebrafish at 9 dpf was far below that in the control group (Figure S2D, E). Of note, co-injecting ogt mRNA with the ogt MO target sequence was effective in rescuing ogt mRNA levels and craniofacial and palatal phenotypes. Overall, O-GlcNAcylation may play an indispensable role in embryonic craniofacial development, particularly palatal development.

Fig. 3.

Cleft palate and impaired parasphenoid bone formation caused by the loss of O-GlcNAcylation in zebrafish. a Representative fluorescence ventral views of Tg(sox10: eGFP) transgenic zebrafish larvae. Palatal skeletons were dissected and flat-mounted. The anterior part is facing upwards. The white solid line represents the ethmoid palate. Scale bars, 100 μm. b Schematic diagram of the ventral (up) view of bone structures of 9 dpf zebrafish larvae. aa anguloarticular, br1 branchiostegal ray1, en ectopterygoid, m maxilla, n notochord, o opercle, p parasphenoid, br2 branchiostegal ray2, c cleithrum, cb ceratobranchial 5, ch ceratohyal, d dentary, hm hyomandibular, v vertebrae, bop basioccipital articulatory process

Negative impact of OGT knockdown in HEPM cells on in vitro osteogenic differentiation

Given that OGT plays a potential role in palatal bone development, further in vitro studies were conducted. First, mineralization was induced in HEPM cells, and WB analysis showed that OGT protein levels gradually up-regulated during early induction and maintained thereafter (Fig. 4a, Figure S3A). Subsequently, a lentivirus expressing a short hairpin targeting OGT mRNA was used to knock down OGT in HEPM cells. Compared with the control group, OGT knockdown had no significant effect on the proliferation of HEPM cells (Fig. 4b, Figure S3B) but inhibited osteogenic function, which was characterized by the down-regulation of osteogenic marker genes transcription and reduced mineralization capacity (Fig. 4c–f, Figure S3C). The above results indicated the negative impact of OGT knockout on in vitro palatal osteogenic differentiation.

Fig. 4.

Negative impact of OGT knockdown in HEPM cells on in vitro osteogenic differentiation. a WB analysis of OGT protein level during osteogenic differentiation. b Representative Edu staining images of in shCtrl and shOGT HEPM cells. shCtrl short hairpin control (control lentivirus), shOGT short hairpin OGT (OGT lentivirus). c Expression of osteogenic markers (COL1A1, RUNX2, OSX, and OCN) and the protein levels of HEPM cells treated as indicated and assessed by WB. d Expressions of osteogenic genes (COL1A1, RUNX2, OSX, and OCN) and the mRNA levels of HEPM cells treated as indicated and assessed by RT-qPCR. e, f ALP and ARS staining of HEPM cells treated as indicated. Data were indicated by mean ± SD. n = 3 biologically independent experiments. P-values were computed by two-tailed t-test or one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05, **P < 0.01

MEIS2 essential for palatal osteogenesis

Mounting evidence supports that MEIS2 is a candidate gene for cleft palate. However, the potential mechanisms underlying MEIS2-mediated palatal developmental defects, particularly abnormalities in palatine bone development are still unclear. Herein, the expression pattern of MEIS2 at critical time points during osteogenesis was first examined. Similar to previous studies, the strong expression of MEIS2 in epithelial and mesenchymal tissues was observed, with widespread expression in the palatine bone region (Fig. 5a). WB analysis showed that MEIS2 protein levels gradually increased over the period of 7-day mineralization induction (Fig. 5b, Figure S4A). Furthermore, a lentiviral vector expressing a short hairpin RNA targeting MEIS2 mRNA was used to knock down MEIS2 in HEPM cells. The protein and mRNA levels of a few osteogenic genes in both groups were noted, such as COL1A1, RUNX2, OSX and OCN. MEIS2 knockdown in HEPM inhibited the protein and mRNA levels of osteogenic markers (Fig. 5c, Figure S4B, C). Additionally, compared to the control group, the shMEIS2 group exhibited reduced ALP activity on the 5th day of induction and fewer calcium nodules on the 14th day of induction (Fig. 5d, e). These results confirmed the role of MEIS2 in regulating palatal osteogenic development.

Fig. 5.

MEIS2 essential for palatal osteogenesis. a Immunofluorescence staining of MEIS2 on coronal sections from E14.5 to E16.5. b WB analysis of MEIS2 protein levels during osteogenic differentiation. c Expression of osteogenic markers (COL1A1, RUNX2, OSX, and OCN) and the protein levels of HEPM cells treated as indicated and assessed by WB. d, e ALP and ARS staining of HEPM cells treated as indicated. n = 3 biologically independent experiments. Data were indicated by mean ± SD. P-values were computed by two-tailed t-test or one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05, **P < 0.01

O-GlcNAcylation of MEIS2 mediated by OGT on Ser237

The Uniprot database indicates that MEIS2 may have potential O-GlcNAcylation sites.²⁵ OGT is the sole enzyme responsible for catalyzing the O-GlcNAcylation of substrate proteins, so proteins that can be O-GlcNAcylated should interact with OGT. To confirm the mediation of MEIS2 O-GlcNAcylation by OGT. First, Co-IP and immunofluorescence staining were performed in HEPM cells, which verified the interaction between endogenous OGT and MEIS2 (Fig. 6a, b, Figure S5A). Furthermore, HA-tagged MEIS2 plasmids were transfected into HEK293 cells to overexpress MEIS2 (Figure S5B, C). IP assays showed that exogenous MEIS2 can also be O-GlcNAcated (Figure S5D). Regarding the succinylated wheat germ agglutinin (sWGA) affinity assay, it is an extensively used method of detecting O-GlcNAcylated proteins. HEPM cells were treated with the OGA inhibitor Thiamet G (TMG), and the O-GlcNAcylation levels of MEIS2 rose with overall cellular O-GlcNAcylation levels. While this process in HEPM cells was inhibited by adding OGT inhibitor OSMI-1. To further confirm the exogenous O-GlcNAcylation of MEIS2, we conducted similar experiments in HEK293 cells and observed that the O-GlcNAcylation of MEIS2 was enhanced or reduced upon treatment with TMG or OSMI-1, respectively (Fig. 6c, d). YinOYang (http://www.cbs.dtu.dk/services/YinOYang/), an online tool, was employed for predicting O-GlcNAc sites on MEIS2 using standard criteria, and Ser237 was identified as a high-scoring site. Interestingly, MEIS2 S237 is highly conserved across vertebrates, as Ser237 in Human and Mouse corresponds to Ser238 in Danio rerio (Fig. 6e). A MEIS2 mutant (MEIS2-S237A) was generated. The O-GlcNAcylation levels of the MEIS2 mutant in HEK293 and HEPM cells were validated. As shown by sWGA pull-down assays, the O-GlcNAcylation levels on MEIS2-S237A exhibited a marked decrease compared with those on MEIS2-WT. IP assays also showed significantly reduced O-GlcNAcylation levels on S237A in comparison with WT (Fig. 6f, g). In summary, Ser237 is the major site for O-GlcNAcylation on MEIS2.

Fig. 6.

O-GlcNAcylation of MEIS2 mediated by OGT on Ser237. a Co-IP confirmed the interaction between OGT and MEIS2 in HEPM cells. b Representative immunofluorescence staining was performed to determine subcellular OGT (green) and MEIS2 (red) localization in HEPM cells. Scale bar: 5 μm. c, d Exogenous MEIS2 in HEK293 cells and endogenesis MEIS2 in HEPM cells were observed by sWGA pull-down assays. Before harvesting, cells received 12 h treatment with 30 μΜ TMG or OSMI-1. e Cross-species protein sequence alignment of MEIS2. f sWGA pull-down assays were carried out in HEK293 cells. Plasmids encoding HA-MEIS2 (WT or S237A) transfected cells. g IB analyses of WCL and IP extracted from HA-WT or HA-S237A mutant-transfected HEK293 cells. Data were indicated by mean ± SD. n = 3 biologically independent experiments. P-values were computed by two-tailed t-tests, *P < 0.05, **P < 0.01

Stabilization of MEIS2 by O-GlcNAcylation through the suppression of its ubiquitination

To elucidate the specific mechanism through which OGT regulates MEIS2, it was found that the knockdown of OGT greatly reduced MEIS2 protein levels, but mRNA levels remained unchanged, which suggested that OGT may influence its PTM (Fig. 7a–c). Past research has demonstrated that the ubiquitin-proteasome degradation of MEIS2 affects cellular physiological functions.²⁶ Adding MG132, a proteasome inhibitor, could restore the OSMI-1-induced reduced MEIS2 protein levels in HEPM cells, which indicated that MEIS2 O-GlcNAylation influences the MEIS2 degradation process via the proteasome (Fig. 7d). Therefore, further experiments were conducted. Cells were treated with DMSO, TMG, and OSMI-1. Moreover, HEPM cells were added with cycloheximide (CHX) and gathered at designated time points. TMG treatment up-regulated the O-GlcNAcylation levels of MEIS2 and prolonged the half-life of MEIS2 protein, whereas OSMI-1 treatment showed the opposite effect. Compared with MEIS2-WT, S237A also substantially reduced the half-life of MEIS2, which suggested that OGT enhances the stability of MEIS2 protein by up-regulating its O-GlcNAcylation (Fig. 7e–g, Figure S6A–C). The role of O-GlcNAcylation in MEIS2 ubiquitination was further explored. TMG treatment reduced exogenous or endogenous MEIS2 ubiquitination, while OSMI-1 treatment showed the opposite effect. Additionally, MEIS2-S237A with lower levels of O-GlcNAcylation demonstrated higher ubiquitination levels. TMG treatment failed to attenuate MEIS2-S237A ubiquitination (Fig. 7h–j). These results indicated that O-GlcNAcylation stabilizes MEIS2 by inhibiting its ubiquitination.

Fig. 7.

Stabilization of MEIS2 by O-GlcNAcylation through the suppression of its ubiquitination. a, b WB and semi-quantitative analysis of MEIS2 protein levels in shCtrl and shOGT HEPM cells. shOGT is short for short hairpin OGT (OGT lentivirus). c qRT-PCR results of MEIS2 mRNA levels in shCtrl and shOGT HEPM cells. d HEPM cells received 8 h treatment with 10 μmol/L MG132 after 12 h treatment with DMSO, 30 μmol/L Thiamet G or 30 μmol/L OSMI-1. MEIS2 expression were observed by WB. e, f Half-life analysis of MEIS2 in HEPM cells. After an indicated period of treatment with 250 μmol/L CHX, MEIS2 levels in cells were observed by IB. Cells were treated with DMSO, 30 μmol/L OSMI-1 or 30 μmol/L Thiamet G. g Half-life analysis of HA-MEIS2 in HEK293 cells. After an indicated period of treatment with 250 μmol/L CHX, HA-MEIS2 levels in HEK293 cells were observed by IB. HA-MEIS2 (WT or S237A) plasmids transfected cells. h, i Cells received treatment with DMSO, 30 μmol/L Thiamet G or 30 μmol/L OSMI-1–12 h, endogenesis (h) or exogenous (i) MEIS2 ubiquitination was observed by anti-MEIS2 or anti-HA antibodies immunoprecipitated. j Cells were subjected to transfection with HA-WT or HA-S237A plasmids and treatment with 30 μmol/L Thiamet G 12 h before harvesting. MEIS2 ubiquitination levels were detected by IP assays. n = 3 biologically independent experiments. Data were indicated by mean ± SD. P-values were computed by one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05 and **P < 0.01

Regulation of SHOX2 expression by MEIS2 O-GlcNAcylation

To determine whether O-GlcNAcylation can influence the subcellular localization of MEIS2, MEIS2 expression in the cell nucleus was found to show a remarkable decrease after OGT knockdown. The overexpression of MEIS2 and S237A plasmids in HEPM cells rescued the reduced MEIS2 expression in the cell nucleus (Fig. 8a). Nevertheless, the groups were not significantly different in subcellular localization (Fig. 8b). The inactivation of SHOX2 widely expressed in the hard palate causes palatal defects in the secondary palatal cleft. Confirmed as a downstream MEIS2 target, SHOX2 jointly regulates osteogenic development in the palate. To investigate whether MEIS2 O-GlcNAcylation can regulate SHOX2 transcription, the online bioinformatics tool JASPAR was used to identify underlying transcription factors likely binding to the SHOX2 promoter region. As shown in Fig. 8d, there is a MEIS2 binding site in the SHOX2 promoter region, which indicates that MEIS2 is a potential transcription factor for SHOX2. To validate the transcriptional regulation of SHOX2 by MEIS2, OGT was knocked down in HEPM cells, and a decrease was observed in both MEIS2 and SHOX2 protein levels. By comparison, the overexpression of MEIS2 in cells increased MEIS2 and SHOX2 protein levels (Fig. 8c). The luciferase signal from the SHOX2 promoter exhibited a noticeable increase when MEIS2 was expressed, but a significant decrease when a mutation was introduced into the putative MEIS2 binding site of the SHOX2 promoter (Fig. 8d). Furthermore, SHOX2 promoter activity was reduced by OGT knockdown, while the overexpression of MEIS2 rescued SHOX2 promoter activity. However, this rescue was not achieved by S237A (Fig. 8e). These results indicated that MEIS2 O-GlcNAylation regulates SHOX2 expression.

Fig. 8.

Regulation of SHOX2 expression by MEIS2 O-GlcNAcylation. HEPM cells were subjected to transfection with OGT shRNA lentiviral vectors for inducing endogenous OGT knockdown and then infected with expressing MEIS2 (WT or S237A) plasmids. a Representative Immunofluorescence staining of MEIS2 in HEPM cells treated as described above. b Nuclear and cytosolic fractions underwent IB with anti-MEIS2, and the ratio of the nucleus to the cytoplasm in MEIS2 was quantified. n = 3 biologically independent experiments. c WB analysis of SHOX2 and MEIS2 protein levels in HEPM cells treated as described above. d Dual-luciferase reporter assays showed the effects of MEIS2 overexpression on relative SHOX2-promoter activity in HEPM cells. n = 3 biologically independent experiments. e The activity of SHOX2 promoter luciferase reporter constructs was measured by dual luciferase reporter assays. n = 3 biologically independent experiments. Data were indicated by mean ± SD. P-values were computed by one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05 and **P < 0.01

O-GlcNAcylation of MEIS2 promotes palatal osteogenic development in vitro and in vivo

Considering the significance of MEIS2 in palatal osteogenesis and the fact that OGT mediated MEIS2 O-GlcNAcylation enhances its protein stability, MEIS2 and S237A plasmids were overexpressed in OGT knockdown HEPM stable cell lines during induced mineralization. WB, ALP and ARS staining results revealed a marked reduction in the osteogenic capacity of OGT-knockdown cells. However, the re-expression of MEIS2-WT significantly restored osteogenic capacity, while that of MEIS2-S237A failed to significantly improve cellular osteogenic capacity (Fig. 9a–d). To further examine the influence of ogt on in vivo palatal osteogenesis, mRNA overexpressing meis2 and S238A (Ser237 in Human and Mouse corresponds to Ser238 in Danio rerio) were co-injected into zebrafish injected with ogt MO. The palatal defect rate of ogt MO zebrafish was significantly higher than that of the control group, and the mineralized area of parasphenoid bones was far below that in the control group. Nevertheless, embryos co-injected with ogt MO and meis2 mRNA rescued the phenotype of MO-injected embryos, but this rescue was not achieved by co-injecting S238A mRNA (Fig. 10a–d). To sum up, the O-GlcNAcylation of MEIS2 influences palatal bone osteogenesis both in vivo and in vitro.

Fig. 9.

O-GlcNAcylation of MEIS2 promotes palatal osteogenic development in vitro. HEPM cells were subjected to transfection with OGT shRNA lentiviral vector for inducing endogenous OGT knockdown and later infected with expressing MEIS2 (WT or S237A) plasmids. a, b WB and semi-quantitative analysis were performed to notice the expression of osteogenic differentiation marker proteins treated as described above in HEPM cells. c, d ALP and ARS staining of HEPM cells treated as indicated. n = 3 biologically independent experiments. Data were indicated as mean ± SD. P-values were computed by one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05 and **P < 0.01

Fig. 10.

O-GlcNAcylation of MEIS2 promotes palatal osteogenic development in vivo. a Images of 120 hpf zebrafish larvae stained with Alcian blue. The black dashed line represents the ethmoid palate. b Cleft palate rates in zebrafish embryos were quantified. n = 3 biologically independent experiments. c ARS staining images of 9 dpf zebrafish larvae. d Semi-quantitative analysis of parasphenoid mineralization IOD of (c). n = 10. Data were indicated as mean ± SD. P-values were computed by one-way ANOVA followed by Tukey’s multiple comparisons test, *P < 0.05 and **P < 0.01

In conclusion, our study demonstrated that OGT-mediated O-GlcNAcylation stabilizes MEIS2 by suppressing its ubiquitination, thereby regulating osteogenic differentiation during palatal development. Ser237 was identified as a critical O-GlcNAcylation site of MEIS2. These findings provide novel insights into the molecular diagnosis of cleft palate and propose the OGT–MEIS2 axis as a potential therapeutic target (Scheme 1).

Scheme 1.

Schematic model depicting the regulation of palatal osteogenesis by OGT-mediated MEIS2 O-GlcNAcylation. In embryonic palatal mesenchymal stem cells, OGT catalyzes the O-GlcNAcylation of MEIS2 at Ser237. This modification stabilizes MEIS2 by inhibiting ubiquitin-proteasome-mediated degradation. Stabilized MEIS2 binds to the SHOX2 promoter to activate transcription, thereby promoting osteogenic differentiation and ensuring normal palatal fusion

Discussion

The hypothesis was initially corroborated about the diminishment of O-GlcNAc expression in the palatine plates of atRA-induced cleft palate mice, possibly through nuclear receptor-mediated transcriptional regulation, redox homeostasis modulation, and direct enzyme inhibition. Evidence linking this reduction to palatal development, particularly osteogenic processes, was then provided. An O-GlcNAc-deficient zebrafish model was generated to demonstrate that this deficiency increases the incidence of cleft palate and palatine bone formation defects. Subsequent analysis identified MEIS2 as a potential novel PTM of O-GlcNAcylation based on evidence demonstrating its interaction with OGT in HEPM cells. Further results indicated the importance of O-GlcNAcylation in maintaining the protein stability of MEIS2 by suppressing ubiquitination. Furthermore, O-GlcNAcylation at Ser237 of MEIS2 has been demonstrated to play a pivotal role.

In recent years, a growing number of studies have indicated the crucial regulatory role of O-GlcNAc modification in embryonic development.^27–29 Additionally, OGT gene mutations are associated with X-linked intellectual disability, characterized by developmental delay and severe cognitive impairments, and accompanied by clinical phenotypes like craniofacial malformations.^30,31 We therefore hypothesise that O-GlcNAc may also have a potential role in craniofacial development. It is reported that palatal developmental abnormalities can be mediated by O-GlcNAcylation and ubiquitination under the influence of genetic and environmental factors. Also, a previous study reported that injecting morpholine targeting ogt transcripts to reduce O-GlcNAc levels in zebrafish embryos resulted in shortening of the embryonic axis, smaller heads and smaller or absent eyes. Similar shortened body axes and smaller heads were also observed in zebrafish larvae in our study. In addition, the in situ hybridization results of this study showed strong expression of OGT in the cranial neural crest.³² Zebrafish palatine bones formed 4 d post-fertilization (dpf) and were composed of paired cartilaginous laminae, sieve plates and dermal paraglottic bones. Therefore, the palatal phenotype was also examined. Interestingly, reduced O-GlcNAc increased the incidence of cleft palate and caused the abnormal ossification of parasphenoid bones. Increasing evidence suggests that the palatal bone development patterns in mammals are similar to those in zebrafish, which exhibits high evolutionary conservation.³³ Both zebrafish and mammals rely on cranial neural crest cells (CNCCs) and regulators like Sox9 and Runx2 to form palatal structures. Zebrafish retains ancestral chondrogenic mechanisms, while mammals evolved novel epithelial-mesenchymal transition (EMT) -based fusion for terrestrial feeding. Zebrafish develops pharyngeal cartilages via endochondral ossification rather than epithelial. While palatal shelves of mammals elevate horizontally and fuse via medial edge epithelium (MEE) apoptosis and ossify intramembranously.

The palatine bone and other craniofacial bones of zebrafish and mammals share a common embryological origin, arising from CNCCs and developing through intramembranous ossification.³⁴ Thus, the lentiviral knockdown of Human Embryonic Palatal Mesenchymal (HEPM) cells was used to confirm the crucial role of OGT in palatal osteogenesis during development in our study. The role of O-GlcNAc in craniofacial osteogenesis has also garnered increasing attention recently. The specific knockout of OGT in osteoblasts resulted in severe defects in craniofacial bone mineralization and tooth development in mice, which revealed the influence of OGT on intramembranous osteogenesis during the perinatal stage.²² Another study declared that OGT-mediated HDAC5 O-GlcNAcylation regulates the balance between nuclear entry and cytoplasmic proteolysis, and hence influences DNA epigenetic modifications and the Notch signaling pathway, and subsequently plays a part in osteogenesis.³⁵ The molecular mechanisms underlying the regulation of skeletal signaling events by O-GlcNAcylation in the future require further investigation.

MEIS2, an essential member of the TALE family of homing box transcription factors, has received extensive attention because of its significance in embryonic development, organ formation and disease pathogenesis.^36,37 Heterozygous missense mutations of MEIS2 or 15q14 microdeletions at MEIS2 gene loci feature cleft palate, intellectual disability and atrial or ventricular septal heart defects, referred to as MEIS2 syndrome.³⁸ In humans, substantial evidence supports MEIS2 as a candidate gene for cleft palate.^39,40 MEIS2 interacts with these osteogenic factors in a complex regulatory network during maxillary development.¹⁸ Recent research has suggested that MEIS2 functional activity is finely modulated by various PTMs, among which the ubiquitin-proteasome system is of importance in regulating MEIS2 stability.²⁶ In the current study, HEPM cells were used to confirm the role of MEIS2 in palatal osteogenesis, and its molecular mechanisms were further explored. How O-GlcNAc influences MEIS2 ubiquitination and thereby MEIS2 stability was investigated for the first time. Previous studies have identified SHOX2 as a downstream target of MEIS2.¹⁸ SHOX2 is a core gene for palatal development, which can regulate osteogenic gene expression and the hard palate by interacting with distal cis regulatory elements.⁴¹ In line with previous studies, MEIS2 acts as a transcription factor that binds to the SHOX2 promoter to regulate its expression. Furthermore, our results confirm that MEIS2 O-GlcNAylation regulates SHOX2 expression. These results expand our understanding of the post-translational modification mechanisms of the MEIS2 protein.

The functional mechanisms of O-GlcNAc modification remain controversial. Studies have shown that O-GlcNAcylation can suppress protein ubiquitination through competition with phosphorylation or other unheard mechanisms, which thereby stabilizes various non-histone proteins and prolongs their half-lives.^42,43 This modification is particularly evident in the fine-tuning of protein activity. In our study, the role of Ser237 of MEIS2 phosphorylation and its crosstalk relationship with O-GlcNAcylation were not explored.

Post-translational modifications (PTMs), a core regulatory layer in palatal development, integrate signal transduction, transcriptional regulation and ECM dynamics to ensure the precision of growth factor responses and epigenetic programming.⁴⁴ It has been proven that disruptions in PTM networks are strongly linked to the etiology of cleft palate. Such as serine and threonine residues are required for O-GlcNAcylation and phosphorylation, the extent of competition between PTMs depends on the cellular signaling regulatory network. Future studies should focus on expounding the crosstalk mechanisms between PTMs and systematically mapping the PTM interaction landscape in human palatal development using multi-omics approaches. Thus, new therapeutic targets for disease intervention can be identified.

Materials and Methods

Ethics

The Ethics Committees of Nanjing Medical University (NMU) examined and approved animal studies (IACUC-2503069 and IACUC-2303019) for zebrafish and mice, respectively). The handling of experimental animals followed the guidelines and protocols that gained the approval of the Animal Care Committee of NMU. Zebrafish and mice were euthanatized as per the American Veterinary Medical Association (AVMA) Guidelines for the Euthanasia of Animals.

Animals

Tg(sox10: egfp) zebrafish embryos were subjected to culturing on a 14 h:10 h light and dark cycle at the temperature of 28.5 °C in Jiangsu Key Laboratory of Oral Diseases, NMU. Euthanasia was performed on Zebrafish by fast chilling (2–4 °C). Wild-type (WT) C57BL/6 J mice aged eight weeks were supplied by the Animal Experiment Center of NMU and raised in the Animal Facility of NMU with a 12 h:12 h light-dark cycle at the temperature of 22 °C under a relative humidity of 40% to 60%. Water and food were freely available to mice. Female and male mice (at the age of 8–12 weeks; with a weight of 20–25 g) copulated at a ratio of 3:1 overnight. Embryonic day 0.5 (E0.5) of gestation was designated at 8 a.m. of the following day when a vaginal plug was found. For the animal model system, pregnant mice at E10.5 were randomly divided into groups treated with atRA and control groups. They were gavaged with atRA (100 mg/kg in corn oil; Macklin, 13497-05-7) or an equivalent amount of corn oil (Aladdin, 8001-30-7), respectively. An excessive amount of carbon dioxide (CO₂) was used to euthanize and sacrifice pregnant mice at E14.5, E15.5 and E16.5, and collect palatal shelf samples and fetal mouse heads. An overdose of CO₂ was used to euthanize all mice.

Antibodies

Primary antibodies utilized for immunoblotting (IB) were specific for anti-OGT (Proteintech, 11576-2-AP, 1: 1 000), anti-O-GlcNac (Abcam, ab2739, 1:1 000), anti-MEIS2 (HUABIO, ET7107-04, 1:1 000), anti-collagen, type I, alpha 1 (COLA1) (HUABIO, ET1609-68, 1:1 000), anti-Runt-related transcription factor 2 (RUNX2) (HUABIO, ET1612-47, 1:1 000), anti-Osterix (OSX) (UpingBio, YP-Ab-05206, 1:1 000), anti-osteocalcin (OCN) (Affinity, DF12303, 1:1 000), anti-HA (Abbkine, ABT2040, 1:1 000) and anti-SHOX2 (Immunoway, YT6364, 1:1 000). Primary antibodies utilized for immunofluorescence and immunohistochemical (IHC) staining were specific for anti-O-GlcNac (Abcam, ab2739, 1:50), anti-OGT (Proteintech, 11576-2-AP, 1:50) and anti-MEIS2 (Santa Cruz, sc-515470, 1:50).

Immunofluorescence and IHC staining

Immunofluorescence and IHC staining were employed for detecting the differences in phenotypes and the expression of related molecules, respectively. The palatal shelf samples of embryonic mice were gathered at E14.5, E15.5, and E16.5 under a stereomicroscope. Next, samples were fixed in 4% Paraformaldehyde Fix Solution (PFA) for no less than 24 h, followed by their dehydration, paraffin-embedding, sectioning at 4.5 μm and mounting on glass slides following standard histological procedures. Slice samples were hydrated again, and citrate buffer was applied to retrieve antigens in a pressure cooker. Then, slides were subjected to goat serum (Boster Biological Technology, AR0009) blocking and incubation with primary antibodies. On a subsequent day, slices were incubated with IHC secondary or fluorescence-conjugated secondary antibodies as appropriate and counterstained using either Diaminobenzidine (DAB) (MXB, DAB-0031) or 4’,6-diamidino-2-phenylindole (DAPI) (Beyotime, C1005). The process was replicated in consecutive staining rounds till the labeling of all antibodies. Finally, DAPI was adopted to stain the sections, and a fluorescence microscope (Leica Microsystems, Wetzlar, Germany) was used to capture images.

Masson’s trichrome staining

Masson’s trichrome staining was utilized for measuring fibrosis with the Masson kit (Leagene, DC0033). In brief, the abovementioned paraffin-embedded sample slices underwent 10 min incubation with Ponceau-Magenta. After 2-min treatment with phosphomolybdic acid, the slices were subjected to 1 min staining with aniline blue. Every step was carried out at room temperature (RT). A light microscope (Leica Microsystems) was used to obtain images.

Von Kossa staining

Mineral deposits were assessed by Von Kossa staining using the Von Kossa Kit (Leagene, DS0003). Slices from the palatal shelves of mice underwent 20 min incubation in 1% silver nitrate under ultraviolet light. The removal of excess silver was realized by a 5 min incubation in 5% thiosulfate sodium, followed by hematoxylin and eosin staining (Nanjing Jiancheng, D006-1-1). A light microscope (Leica Microsystems) was used to examine the slices.

Morpholinos and messenger ribonucleic acid microinjection

A micromanipulator (Nikon NARISHIGE, NT88-V3) was leveraged to inject 2 ng/μL morpholinos (MOs) into one-cell stage zebrafish embryos. MOs were provided by Gene Tools (Philomath, the United States of America (USA)): A standard control MO: 5’-CTAAAAGCA-GCAGGAGGCGATTCAT-3’, was utilized. A translation-blocking MO targeting zebrafish ogt with the sequence 5′-CCACGTTCCCCACCGAGCTTGCCAT-3′ was utilized for ogt knockdown. The sequences of zebrafish ogt and meis2 were offered by the National Center for Biotechnology Information Gene Database. After the cloning and ligation of ogt, meis2 and meis2-S238A complementary DNA (cDNA) into the pXT7 plasmid, the mMESSAGE mMACHINE T7 kit (Ambion, #AM1344) was used to linearize the plasmids with BamHI and transcribe them. One-cell stage zebrafish embryos were injected with the capped messenger ribonucleic acids (mRNAs) at 50 pg/embryo. Embryos were previously treated with 16 μmol/L (R)-N-(furan-2-ylmethyl)-2-(2-methoxyphenyl)-2-(2-oxo-1,2-dihydroquinoline-6-sulfonamido)-N-(thiophen-2-ylmethyl)acetamide 1 (OSMI-1) (MCE, HY-119738) until the experimental procedure was performed.

Alcian blue and Alizarin Red staining

Zebrafish embryos at 120 h post-fertilization (hpf) were collected at random and fixed in 4% PFA at 4 °C overnight before cartilage staining with Alcian Blue Kit (Muruibio, MR209-03). The samples were rinsed with a 60:40 glycerol/1% potassium hydroxide (KOH) solution till their full translucence and then kept in 100% glycerol. Bone mineralization levels were evaluated by staining 120 hpf zebrafish larvae with Alizarin Red Kit (Muruibio, MR210-03) as described previously.⁴⁵ Then, the larvae were rinsed and stored in a glycerol/KOH solution at 4 °C. Morphological craniofacial changes were observed using a Nikon microscope, and quantitative analysis was performed using ImageJ software. The cleft palate was judged through the revealing of a cleft in the ethmoid palate, in which a population of cells was absent in the median part.

Cell culture

Cells obtained from American Type Culture Collection (ATCC) included human embryonic kidney 293 (ATCC, CRL-11268) and human embryonic palatal mesenchyme (HEPM) (ATCC, CRL-1486). All cell lines were cultured in alpha-minimum essential medium (Gibco, C12440500BT) added with 10% fetal bovine serum (Vivacell, C04001-500) and 1% penicillin-streptomycin (Beyotime, C0222) at 5% CO₂ and 37 °C in a humidified incubator. The culture medium was replaced every third day. Short tandem repeat profiling was used to identify every cell line, and no mycoplasma contamination was found.

Plasmid constructs

The entire cDNA of Human OGT (NM_181672.2) and MEIS2 (NM_170675) was subjected to amplification by polymerase chain reaction (PCR) and subcloning into pcDNA3.1-HA or pcDNA3.1-3xFlag. MEIS2-S237A was constructed by overlapping PCR. Supplementary Table 1 provides all specific primers. The extraction of plasmids was performed by transforming DH5α (TIANGEN, CB101) with corresponding plasmids and utilizing FastPure Plasmid Mini Kit (Vazyme, DC201) as per the protocol of the manufacturer.

Construction of stable cell lines

Short hairpin RNA (shRNA) expression vectors were constructed for knocking down the expression of OGT and MEIS2. Supplementary Table 2 lists shRNA-targeted sequences. Cells expressing the constructs stably were chosen with puromycin (2 mg/mL) for seven days.

5-Ethynyl-2-deoxyuridine incorporation assay

In total, 5 × 10³ cells were subjected to seeding in plates with 96 wells and transfected at an appropriate confluence before the experiment. A Cell-Light EdU Apollo 567 Stain Kit (RiboBio, C10371-1) was used to assay the proliferative ability of cells. Additionally, 5 Ethynyl 2 deoxyuridine (EdU) was incorporated, and an EVOS® FL automated imaging system was used to image cells. Finally, ImageJ software was used to analyze the proliferative ability of cells.

Immunofluorescence staining

HEPM cells were seeded in plates with 24 wells and transfected at an appropriate confluence 24 h before the experiment. After being washed with phosphate buffer saline (PBS), HEPM cells were subjected to 20-min fixing with 4% PFA at RT and 20-min incubation in 0.5% Triton X-100 (Biosharp, BS084) for permeabilization. Cells were blocked with goat serum (Boster Biological Technology, AR0009) at 37 °C for 1 h and incubated with primary antibodies at 4 °C overnight. Goat anti-mouse Alexa Fluor 488 and goat anti-rabbit Cy3 (1:50, Beyotime, China) were selected as secondary antibodies for the combination of primary antibodies at 37 °C in the dark for 1 h. Finally, DAPI (Beyotime, C1005, China) was used to dye the nuclei blue, and a fluorescence microscope (Leica Microsystems, Wetzlar, Germany) was applied to capture fluorescent images.

Quantitative real-time PCR

An RNA extraction kit (Agbio, AG21024) was used to separate total RNA. Evo M-MLV RT Kit (Agbio, AG11705) was used for the reverse-transcription of RNA into cDNA. Subsequently, real-time PCR (RT-PCR) was conducted employing SYBR Green Premix Pro Taq HS qPCR Kit (Agbio, AG11701) on the ABI-7200 Real-Time PCR System (Applied Biosystems, CA, USA). The 2^−ΔΔCT method was adopted to calculate the data. Supplementary Table 3 shows quantitative PCR (qPCR) primers.

Alkaline phosphatase and Alizarin Red S staining

In this step, 5-Bromo-4-chloro-3-indolyl phosphate p-toluidine salt (BCIP)/nitroblue tetrazolium chloride (NBT) ALP Color Development Kit (Beyotime, C3206) was used to evaluate the activity of alkaline phosphatase (ALP). After 7-days incubation with conditional medium, HEPM cells were subjected to 15 min fixing with 75% ethanol at 4 °C and 15 min staining with the as-prepared BCIP/NBT solution without light. After being washed with PBS gently, images were captured by the use of an inverted microscope (Leica, Germany, DFC2900) and a scanner (EPSON, USA, A3 Transparency Unit).

HEPM cells were incubated with osteogenic inductive medium (100 mmol/L l-ascorbic acid (Sigma, USA), 1.8 mmol/L potassium phosphate monobasic (Sigma, USA), 100 mmol/L β-glycerophosphate (Sigma, USA) and 10 nmol/L dexamethasone (Sigma, USA) for 14 days. Then, they were subjected to 15-min fixing with 75% ethanol and 20-min incubation with 1% Alizarin Red S (ARS) solution (Leagene, DS0072). Eventually, superfluous dye was washed away via the addition of PBS, and the mineralized nodules were examined using a microscope and a scanner.

Western blot

Total protein was refined from cells and tissues and radio immuno precipitation assay buffer (Beyotime, P0013B) with 1 mmol/L protease inhibitor phenylmethylsulfonyl fluoride (Beyotime, ST505) was used for the 30-min culturing of cells on ice. The supernatants were collected after centrifugation (10 010 × g for 15 min). Specifically, immunocomplexes underwent 5 min boiling with 5 × sodium dodecyl sulfate (SDS) sample buffer (Beyotime, P0015) before western blot (WB) analysis. For WB, equivalent amounts of protein samples were isolated by SDS-polyacrylamide gel electrophoresis (Vazyme, E303-01) and moved onto polyvinylidene difluoride (PVDF) membranes (Millipore, ISEQ00010). After 3 h blocking in 5% milk in PBS at RT, membranes underwent at least 4 h incubation with antibodies at 4 °C. The next day, after washing with phosphate buffered saline with Tween-20 (PBST) three times, with 10 min for each, membranes underwent 1 h incubation with second antibodies. Enhanced chemiluminescence (ECL) reagents (Tanon) were used to develop specific bands, which were then visualized by Tanon-5200 Multi Chemiluminescent System.

Succinylated wheat germ agglutinin pull-down assays

For succinylated wheat germ agglutinin (sWGA) pull down assays, each lysate underwent denaturing in glycoprotein denaturing buffer and digestion with Peptide N-Glycosidase (New England Biolabs, P0704S) for the removal of N-linked glycoproteins. After being centrifuged, the supernatants underwent 2-h incubation with sWGA bound agarose beads (Vector Laboratories, AL-1023S) at 4 °C. Precipitates were then rinsed and immunoblotted with antibodies.

Immunoprecipitation and co-immunoprecipitation

For immunoprecipitation (IP) and co-IP, every cell lysate was added with Protein A/G magnetic beads (Vazyme, PB101-01) and rotated at 4 °C for one night after incubation with 2 μg antibody diluted in 500 μL of PBST. After the removal of the supernatant with a magnetic separator and washing with PBST, immunomagnetic beads conjugated with antibodies were prepared. After harvesting, 500 μL protein sample supernatants were lightly blended with antibody-conjugated immunomagnetic beads and underwent overnight rotation at 4 °C to create an immunomagnetic beads-antibody-antigen complex. The washing of beads with PBST three times was followed by the re-suspension of the complex in 60 μL of PBST and its use for detecting endogenous interactions between the targeted antibody and other proteins through WB.

Dual luciferase reporter assays

HEPM cells were cultured in plates with 24 wells (6.0 × 10⁴ cells per well). When growing to 70%, cells were co-transfected with pGL1Promoter luciferase reporter vectors covering the 5′-UTR fragment of SHOX2 or mutations, renilla vectors (pRluc-TK) and plasmids overexpressing MEIS2 or MEIS2-S237A using Lipofectamine 2000. After 48 h transfection, luciferase assays were conducted using a dual luciferase reporter assay system (Beyotime, RG088S).

Prediction of MEIS2 O-GlcNAcylation sites

MEIS2 O-GlcNAcylation sites were forecast on the YinOYang-1.2 website (https://services.healthtech.dtu.dk/services/YinOYang-1.2/).

MEIS2 protein stability assays

The stability of endogenous MEIS2 protein was detected by treating HEPM cells with 250 μmol/L cycloheximide (CHX) (Cayman Chemical,14126) and collecting them at 0, 1, 2 and 3 h. To detect the protein stability of exogenous Flag-MEIS2, HEK293 cells were subjected to re-expression with HA-WT or HA-S237A mutants, respectively, followed by the collection of cells at 0, 1, 2 and 3 h after 250 μmol/L CHX treatment. MEIS2 levels were analyzed by WB, and relative half-life was calculated.

Statistics

Every experiment was conducted no less than three times. GraphPad Prism 9 software or Excel was used to perform statistical analysis. Data were indicated by mean ± standard error of the mean (SEM). The significance of between-group differences was evaluated using two-tailed Student’s t-tests and one-way analysis of variance (ANOVA) followed by Dunnett’s multiple comparisons test or Turkey’s multiple comparisons test. *P < 0.05 and **P < 0.01.

Author contributions

Z.Z. performed experiments, analyzed data, and wrote the manuscript. Z.S., X.C., and Y.X. assisted with animal studies and histological experiments. L.M., Y.Z., and C.W. helped conceive the study. J.M. and L.Y. designed experiments and revised the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41368-026-00431-w.