Fig 2: LIN28A expression is reduced in periodontal tissue biopsies and LPS-induced hPDLSCs. The expression of (A) LIN28A and (B) RUNX2 in periodontal biopsy tissues was examined by RT-qPCR. (C) The correlation between the mRNA expression level of LIN28A and RUNX2 in periodontal biopsy tissues was examined. (D) Cell morphology and number of hPDLSCs after treatment with different concentrations of LPS were observed by microscope. Scale bars, 100 µm. LIN28A expression in LPS-induced hPDLSCs was measured by (E) western blotting and (F) RT-qPCR. RUNX2 expression in LPS-induced hPDLSCs was determined by (G) western blotting and (H) RT-qPCR. *P<0.05, **P<0.01 and ***P<0.001 vs. Control. LIN28A, Lin-28 homeobox A; RUNX2, Runt-related transcription factor 2; RT-qPCR, reverse transcription-quantitative PCR; hPDSCs, human periodontal ligament stem cells; LPS, lipopolysaccharide.

Fig 3: EZH2 knockdown inhibits TLR4/MyD88/NF-κB signaling to facilitate PDLSC osteogenesis. (a) Western blot showed the phosphorylated p65, phosphorylated IKB, TLR4, and MyD88 protein level transfected with the indicated plasmids in PDLSCs. (b) IF showed the distribution of p65 in the indicated treatment. (c, d) ALP activity in PDLSCs after the indicated treatments. (e) Western blot showed the Runx2 and OCN protein level after the indicated treatment in PDLSCs. (f) qPCR showed the mRNA level of Runx2 and OCN. (g, h) Alizarin red staining showed the role of EZH2 and LPS on the mineralization in PDLSCs. Data are representative of three independent experiments. ∗P < 0.05; ∗∗P < 0.01; ns: not significant; LPS: lipopolysaccharide; shNC: sh-normal control; p-p65: phosphorylated p65; p-IKB: phosphorylated IKB; TLR4: toll-like receptor 4; MyD88: myeloid differentiation primary response gene 88; PMA: phorbol ester; ALP: alkaline phosphatase assay; OCN: bone gamma-carboxyglutamate protein; Runx2: RUNX family transcription factor 2.

Fig 4: Prehypertrophic differentiation is affected more in the anterior end than the posterior end of the braincase floor. (A) A diagram of the braincase floor in the immunohistochemical studies. (B) Immunodetection of OSX and RUNX2 in the three areas shown in A. The white boxed region was enlarged and shown. Arrows indicate Evc2 mutant chondrocytes with no nuclear localized OSX; arrowheads indicate Evc2 mutant chondrocytes with decreased nuclear localized OSX. Scale bar = 200 μm. Dashed lines indicate the boundary between resting and proliferating chondrocytes. (C) Number of cells with nuclear localized OSX are quantified and shown as percentage of controls, n = 3, **p < 0.01. (D) The intensity of nuclear localized OSX was quantified and shown as a percentage of controls, n = 4, **p < 0.01, error bars denote standard deviations. (E) The intensity of immunosignals for nuclear RUNX2 was quantified and shown as a percentage of control, n = 4, **p < 0.01, error bars denote standard deviations. (F) The percentages of resting chondrocytes with nuclear localized RUNX2 were quantified and shown, n = 3, **p < 0.01. Scale bar = 200 μm, bar in enlarged picture = 20 μm.

Fig 5: Knockdown of circ_AFF4 suppresses BM-MSC osteogenic differentiation (A) BM-MSCs were transfected with shcirc_AFF4_1–4# or shNC and the expression level of circ_AFF4 in shcirc_AFF4 or shNC transfected BM-MSCs was detected using RT-qPCR. BM-MSCs that were transfected with shcirc_AFF4 or shNC were cultured in OM or NM for 14 days. Afterward, different assays were performed to assess the osteogenic differentiation. (B) Representative images of ARS and ALP staining. Scar bar = 100 μm. (C) Representative RUNX2 immunofluorescence staining image. Scar bar = 100 μm. Gene and protein expression of osteogenic markers, ALP, OPN, RUNX2 and Colla1 were detected by (D) RT-qPCR and (E) western blot. * p < 0.05, ** p < 0.001 and *** p < 0.001. Each experiment was performed at least three times independently