PPARγ-Induced Global H3K27 Acetylation Maintains Osteo/Cementogenic Abilities of Periodontal Ligament Fibroblasts

Abstract

The periodontal ligament is a soft connective tissue embedded between the alveolar bone and cementum, the surface hard tissue of teeth. Periodontal ligament fibroblasts (PDLF) actively express osteo/cementogenic genes, which contribute to periodontal tissue homeostasis. However, the key factors maintaining the osteo/cementogenic abilities of PDLF remain unclear. We herein demonstrated that PPARγ was expressed by in vivo periodontal ligament tissue and its distribution pattern correlated with alkaline phosphate enzyme activity. The knockdown of PPARγ markedly reduced the osteo/cementogenic abilities of PDLF in vitro, whereas PPARγ agonists exerted the opposite effects. PPARγ was required to maintain the acetylation status of H3K9 and H3K27, active chromatin markers, and the supplementation of acetyl-CoA, a donor of histone acetylation, restored PPARγ knockdown-induced decreases in the osteo/cementogenic abilities of PDLF. An RNA-seq/ChIP-seq combined analysis identified four osteogenic transcripts, RUNX2, SULF2, RCAN2, and RGMA, in the PPARγ-dependent active chromatin region marked by H3K27ac. Furthermore, RUNX2-binding sites were selectively enriched in the PPARγ-dependent active chromatin region. Collectively, these results identified PPARγ as the key transcriptional factor maintaining the osteo/cementogenic abilities of PDLF and revealed that global H3K27ac modifications play a role in the comprehensive osteo/cementogenic transcriptional alterations mediated by PPARγ.

Keywords: PPARγ; histone acetylation; osteogenic differentiation; periodontal ligament.

Figure A1

The knockdown of PPARγ by 3 new independent siRNAs inhibits ALP activities and calcium deposition without affecting cell numbers. (A–C) PDLF-1 transfected with either si-control, si-PPARγ-2, si-PPARγ-3, or si-PPARγ-4. (A) PPARγ mRNA expression 24 h after transfection was analyzed using HPRT for normalization. (B) ALP activities were measured and cell numbers were enumerated by MTT every 3 days. ALP activities are normalized with MTT values. (C) Alizarin red S staining was performed on day 15. Data represent the mean ± SD of three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly lower than in cells transfected with si-control.

Figure A2

PPARγ agonists are unable to induce ALP activities in PDLF transfected with si-PPARγ. (A–C) BMMSC were transfected with si-control or si-PPARγ. (A) PPARγ mRNA expression 24 h after transfection was analyzed using HPRT for normalization. (B) ALP activities were measured and cell numbers were enumerated by MTT every 3 days. ALP activities are normalized with MTT values. (C) Alizarin red S staining was performed on day 21. Data represent the mean ± SD of three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly higher than in cells transfected with si-control.

Figure 1

PPARγ expression in PDL tissue and cells. (A,B) Demineralized 3-month-old maxilla sections were stained without the primary antibody or with mouse IgG or α-PPARγ and were also stained with Masson’s Trichrome stain and ALP. (C) Total RNAs of PDLF-1, PDLF-2, PDLF-3, PDLF-4, and BMMSC were collected to quantify the expression of PPARγ, PLAP-1/ASPN, COL1A1, OMD, and RUNX2. HPRT was used for normalization. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly higher than BMMSC cells. Scale bars correspond to 200 μm (A) and 50 μm (B). MT = Masson’s Trichrome staining.

Figure 2

PPARγ agonists enhance osteo/cementogenic abilities of PDLF-1. (A–D) PDLF-1 were cultured in induction medium for a maximum of 15 days. ALP activities were normalized by cell numbers (A), cell numbers were enumerated by MTT (B), gene expression changes in ALPL and COL1A1 by qPCR (C), and calcium deposition were visualized and quantified by Alizarin red S staining (D). (E) PDLF-1 were cultured in the presence of PPARγ agonists for 48 h and the global acetylation of H3K27 was analyzed by SDS-PAGE using whole histone H3 as the loading control. H3 = histone 3. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly higher than in cells treated with DMSO on the same day. D = DMSO, Pio = Pioglitazone, Tro = Troglitazone, Ros = Rosiglitazone, nTZ = nTZDpa.

Figure 3

Suppression of PPARγ expression eliminates osteo/cementogenic abilities of PDLF. (A) Cytoplasmic and nuclear fractions were isolated separately from PDLF-1 and PDLF-2 and loaded onto SDS-PAGE gels. (B–F) PDLF-1 and PDLF-2 were transfected with si-control or si-PPARγ. (B,C) PPARγ mRNA expression (B) and protein expression (C) 24 h after transfection was analyzed using HPRT for normalization (B) and GAPDH as the loading control (C). (D) ALP activities were measured and cell numbers were enumerated by MTT every 3 days. ALP activities were normalized by MTT values. (E) The gene expression levels of ALPL, COL1A1, BGLAP, PLAP-1/ASPN, and OMD were examined on day 6 for PDLFs and on day 12 for BMMSC. (F) Alizarin red S staining was performed on day 15 for PDLF-1 and on day 21 for PDLF-2. Data represent the mean ± SD of three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001 and ††† p < 0.001 significantly lower and higher in cells transfected in si-PPARγ, respectively.

Figure 4

PPARγ suppression inhibits the global acetylation of H3K9 and H2K27. PDLF-1 were transfected with si-control or si-PPARγ for 24 h and then recovered in normal growth medium for another 24 h. Total histones were collected and histone modifications, such as H3K27ac, H3K9ac, and H3K27me3, were examined using specific antibodies. Band intensities are normalized against those obtained using the antibody for whole histone 3. * p < 0.05 significantly lower than in cells transfected with si-control.

Figure 5

PPARγ suppression comprehensively reduces the expression of ECM- and ossification-related genes in PDLF. (A) PDLF-1 were transfected with si-control or si-PPARγ and then cultured for 6 days in induction medium. Total RNAs were collected before and after the culture and whole genomic transcriptional changes were assessed by RNA-seq. The upper scatterplot shows a comparison before (day 0: x-axis) and after (day 6: y-axis) the long-term culture of si-control-transfected PDLF. The color of each gene spot is linked with a gradient in proportion to its expression ratio (day 6/day 0), with a darker red color for spots having a higher ratio and a darker green color for those with a lower ratio. The middle color is set as white. The lower scatterplot shows a comparison between si-control-transfected PDLF (x-axis) and si-PPARγ-transfected PDLF (y-axis). The color of each gene spot was taken over from the upper scatterplot. (B–E) Pathway analyses of biological processes (B,D) and wikipathways (C,E). Enhanced terms on day 6 from those on day 0 (si-control-transfected cells) (B,C) and suppressed terms by si-PPARγ from si-control on day 6 (D,E).

Figure 6

Local genomic peaks retained by PPARγ include significantly higher numbers of RUNX2-binding sites. (A) Histograms represent the occurrence of H3K27ac peaks within 10 kb of the ‘si-control unique peak’ and ‘common peak’ centers. A heat map comparing the binding of technical duplicate tags from PDLF-1 transfected with either si-control or si-PPARγ within 10-kb windows surrounding the ‘si-control unique peak’ and ‘common peak’ centers. (B) qPCR analysis is shown as a bar graph. UCSC genomic browser tracks of each gene locus showed a gene locus (blue background) and H3K27ac peaks identified as ‘si-control unique peaks’ (red background). Arrows indicate transcriptional direction. (C) PDLF-2, PDLF-3, and PDLF-4 were transfected with si-control or si-PPARγ and SULF2, RCAN2, RGMA, and RUNX2 mRNA expression at 24 h after transfection was analyzed using HPRT for normalization. (D) PDLF-1 were cultured in induction medium for 6 days in the presence or absence of troglitazone (5 μM) to quantify the expression of SULF2, RCAN2, RGMA, and RUNX2. HPRT was used for normalization. (E) The number of PPARγ-binding elements (PPARE-BE), PPARα-binding elements (PPARα-BE), RUNX2-binding elements (RUNX2-BE), Smad4-binding elements (Smad4-BE), LEF1-binding elements (LEF1-BE), and TEAD-binding elements (TEAD-BE) in ‘si-control unique peak’ and ‘common peak’ were normalized as peaks/1 kbp. The average peak/1 kbp of each BE in ‘si-control unique peaks’ and ‘common peaks’ is shown. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly higher in PDLF-1 treated with troglitazone than in PDLF-1 treated with DMSO on day 6 (D) and in ‘si-control unique peaks’ than in ‘common peaks’ (E). n.s. = no significant difference.

Figure 7

Sodium acetate supplementation restores si-PPARγ -suppressed ALP activity. (A–C) PDLF-1 were transfected with si-control or si-PPARγ for 24 h and then cultured in induction medium for the indicated periods in the presence or absence of metabolic pathway intermediate products (A) and sodium acetate or parthenolide, a HDAC1 inhibitor (B). Calcium deposition was visualized and normalized by Alizarin red staining (C). H3 = histone 3, Pyr = pyruvic acid, Cit = citric acid, Iso = isocitric acid, α-k = α-ketoglutaric acid, Suc = succinic acid, Fum = fumaric acid, Mal = malic acid, Oxa = oxaloacetic acid. Data represent the mean ± SD of three independent experiments. * p < 0.05; ** p < 0.01; *** p < 0.001 significantly higher than in cells treated with neither sodium acetate nor parthenolide in the same siRNA group on the same day (B) and in si-PPARγ-transfected cells without sodium acetate supplementation (C). n.s. = no significant difference.