Identification of SOX9 interaction sites in the genome of chondrocytes

Abstract

Background: Our previous work has provided strong evidence that the transcription factor SOX9 is completely needed for chondrogenic differentiation and cartilage formation acting as a "master switch" in this differentiation. Heterozygous mutations in SOX9 cause campomelic dysplasia, a severe skeletal dysmorphology syndrome in humans characterized by a generalized hypoplasia of endochondral bones. To obtain insights into the logic used by SOX9 to control a network of target genes in chondrocytes, we performed a ChIP-on-chip experiment using SOX9 antibodies.

Methodology/principal findings: The ChIP DNA was hybridized to a microarray, which covered 80 genes, many of which are involved in chondrocyte differentiation. Hybridization peaks were detected in a series of cartilage extracellular matrix (ECM) genes including Col2a1, Col11a2, Aggrecan and Cdrap as well as in genes for specific transcription factors and signaling molecules. Our results also showed SOX9 interaction sites in genes that code for proteins that enhance the transcriptional activity of SOX9. Interestingly, a strong SOX9 signal was also observed in genes such as Col1a1 and Osx, whose expression is strongly down regulated in chondrocytes but is high in osteoblasts. In the Col2a1 gene, in addition to an interaction site on a previously identified enhancer in intron 1, another strong interaction site was seen in intron 6. This site is free of nucleosomes specifically in chondrocytes suggesting an important role of this site on Col2a1 transcription regulation by SOX9.

Conclusions/significance: Our results provide a broad understanding of the strategies used by a "master" transcription factor of differentiation in control of the genetic program of chondrocytes.

Figure 1. Identification of SOX9 interaction sites in rat Col2a1 and Col11a2 genes by ChIP-on-chip.

ChIP-on chip using SOX9 antibody showed specific hybridization peaks, which were further modified by background subtraction to form smoothened peaks shown at the bottom of each panel. Exons are shown as solid bars. In the Col2a1 gene (left panel), we clearly detected two peaks, one in intron 1 and the other in intron 6. In the Col11a2 gene (right panel), we clearly detected two peaks, one in the promoter region and the other in intron 1. Exons shown in the right side of exon 1 in Col11a2 were those in the adjacent gene.

Figure 2. Validation of SOX9 binding motifs by EMSA.

EMSA was performed as described in Materials and Methods using SOX9 binding motifs found at or near the center of the hybridization peaks in each gene. Suppl. Table S1 lists the sequences of the probes used. The mobility of ³²P labeled probe bound to purified SOX9 was consistent with a complex containing mainly a SOX9 dimer. The amount of SOX9 used in each panel was 0, 20 or 40 ng.

Figure 3. Functional analysis of SOX9 interaction site in intron 6.

A: The 48 bp intron6 (+6676 to +6724) of rat was aligned with the corresponding sequences of mouse and human Col2a1 gene. The putative SOX9 binding sites were underlined. B: Left; Schematic representation of chimera constructs. The sequences of a 48 bp segment in intron 1 and intron 6 are shown above each construct in a and b. The bases written by bold letters indicated the putative SOX9 binding sites. The 5′ region in intron 1 contained dimeric direct repeat, and 3′ regions of both intron 1 and intron 6 contained inverted repeats. Right; Each chimera construct consisted of five tandem repeats followed by a Col2a1 minimal promoter (89bp) and the Firefly luciferase gene and transfected in RCS cells. The values of luciferase activity were normalized by adjusting the activity of construct (a) as 1.00.

Figure 4. Stimulation of enhancer activity of intron 1 by intron 6.

A: Schematic representation of constructs used. Construct (a) had a 3 kb promoter, exon 1, a 3 kb intron 1 of Col2a1followed by β-geo as previously shown . Construct (b) had 1kb corresponding to the entire intron 6 itself (+6489 to +7374) in addition to construct (a) just after the polA. In construct (c) the SOX9 binding sequence (48bp, +6676 to +6724) was deleted from intron 6. B: Each construct (500 ng) was co-transfected into RCS cells with 10 ng of a TK-Renilla luciferase plasmid that served as an internal control for transfection efficiency. The bars represent β-galactosidase activity of each construct normalized to the internal control (TK-Renilla luciferase plasmid).

Figure 5. Depletion of histone H3 in intron 6 in RCS cells but not in Rat-2 fibroblast.

Histone modification status in the chromatin at several introns of Col2a1 in RCS cells (A–C) and at the promoter of Col2a1 and Cyclin B1 (CNB1) and introns 1 and 6 of Col2a1 in Rat-2 fibroblasts (D) was examined by real time quantitative PCR. The ChIP DNA fragments were prepared as described in Materials and Methods using SOX9, H3 (A and C), H3K9ac, H3K14ac, H3K4me3 (B) and H3, H3K9ac (D) antibodies. The SOX9 hybridization peak in introns together with the partial map of the Col2a1 gene is shown in E.