Sp6/Epiprofin is a master regulator in the developing tooth

Abstract

Tooth development involves the coordinated transcriptional regulation of extracellular matrix proteins produced by ameloblasts and odontoblasts. In this study, whole-genome ChIP-seq analysis was applied to identify the transcriptional regulatory gene targets of Sp6 in mesenchymal cells of the developing tooth. Bioinformatic analysis of a pool of Sp6 target peaks identified the consensus nine nucleotide binding DNA motif CTg/aTAATTA. Consistent with these findings, a number of enamel and dentin matrix genes including amelogenin (Amelx), ameloblastin (Ambn), enamelin (Enam) and dental sialophosphoprotein (Dspp), were identified to contain Sp6 target sequences. Sp6 peaks were also found in other important tooth genes including transcription factors (Dlx2, Dlx3, Dlx4, Dlx5, Sp6, Sp7, Pitx2, and Msx2) and extracellular matrix-related proteins (Col1a2, Col11a2, Halpn1). Unsupervised UMAP clustering of tooth single cell RNA-seq data confirmed the presence of Sp6 transcripts co-expressed with many of the identified target genes within ameloblasts and odontoblasts. Lastly, transcriptional reporter assays using promoter fragments from the Hapln1 and Sp6 gene itself revealed that Sp6 co-expression enhanced gene transcriptional activity. Taken together these results highlight that Sp6 is a major regulator of multiple extracellular matrix genes in the developing tooth.

1. Introduction

During tooth formation, a complex series of spatial and temporal events is initiated by interactions between undifferentiated epithelium and mesenchyme leading to specialized ameloblasts and odontoblasts, coordinating the synthesis of enamel and dentin components [1]. The most abundant tooth protein is amelogenin, which comprises 80–90% of the proteinaceous material produced during the secretory stage [2]. Amelogenin has a unique ability to self-assemble and promote apatite nanocrystals leading to enamel mineralization [2,3]. Two minor proteins, ameloblastin [4] and enamelin, and trace amounts of Mmp-20 protease [5] are made by ameloblasts. During the maturation phase, ameloblasts produce the Klk4 protease that remodels the matrix scaffold by protein proteolysis required for the mineralization process to proceed [6]. Odontoblasts also produce collagens, including Col1a2 and Col11a2, small amounts of amelogenin and a number of proteoglycans [7,8]. In odontogenesis, the timing and spatial complexity of matrix protein synthesis, removal and control of mineralization involves more than 1000 genes [9,10]. Proteomic studies [11–13] and gene expression analysis [14–16] have demonstrated a number of structural and regulatory proteins involved in developing teeth. Importantly, extracellular matrix molecules such as collagens and proteoglycans provide scaffolds for organizing the complex tissue required for proper enamel and dentin assembly.

Despite the identification of key tooth proteins, little is known about the transcriptional regulation of matrix proteins during development. Genetic and molecular studies have identified tooth-specific transcription factors including Sox2, Msx1, Msx2, and Dlx3 [17–20], however, the downstream targets of these transcription factors are poorly characterized. One transcription factor, Sp6, shows high levels of mRNA and protein expression in the developing inner dental epithelium and dental papilla [21]. Sp6 belongs to a family of transcription factors and is related to another member, Sp7/osterix, which regulates a number of osteoblast genes controlling bone formation [22]. Interestingly, murine Sp6 gene knockout studies have demonstrated profound defects in tooth development including severe reductions in the differentiation of dental epithelium into enamel matrix-producing ameloblasts [23,24]. However, little mechanistic information is known by which the loss of Sp6 contributes to tooth defects.

In this study, chromatin immunoprecipitation profiling DNA sequencing (ChIP-seq) of mouse molar tissue identified a Sp6 consensus DNA-binding motif, which was identified in a number of tooth-specific matrix proteins and transcription factors. Functional studies confirmed Sp6 increased transcription activity substantiating its role to regulate gene expression. Together these studies highlight the role Sp6 plays as a key transcriptional regulator of tooth formation.

2. Materials and methods

2.1. Sp6 chromatin immunoprecipitation-sequencing (ChIP-seq) and peak visualization

The Sp6 ChIP-seq was used to identify in vivo binding sites of genomic DNA from developing mouse tooth molar tissue. Briefly, dental mesenchyme and epithelium tissue from 30 newborn C57/BL6 P1 molars (Under approval from NIDCR Animal Care and Use Committee) were isolated and fixed with a 1% formaldehyde solution for 15 min and quenched with 0.125 M glycine. Additional processing of the tissue and ChIP-seq was performed by Active Motif Inc. (San Diego, California). Dental tissues were sonicated and precleared before antibody incubation. Rabbit polyclonal antibody against mouse Sp6 was generated using a synthetic peptide corresponding to mouse Sp6 amino acids 210–227. For each of the immunoprecipitation samples, approximately 40 million high quality reads were derived by Illumina sequencing. The full dataset of identified Sp6 peaks from both mouse molar epithelial and mesenchyme have been deposited with GEO accession GSE145909 (token mjszmskqvnublit).

Evaluation of the dataset using Model-based Analysis of ChIP-seq (MACs) with a cutoff p value of 10⁻⁷, identified 11,436 peaks. Some of the peaks (n = 2727) were not close enough to be assigned to a gene and were therefore excluded from the Excel files in GEO (GSE145909). The false discovery rate by MACs analysis was estimated to be <1%. Sp6-enriched immunoprecipitation peak sequences were compiled as binary analysis results (BAR) files and analyzed using the Integrated Genome Browser (https://bioviz.org).

2.2. Multiple expectation maximization for motif elicitation (MEME) for defining the Sp6 consensus binding motif

To determine the Sp6 consensus binding motif, 1400 DNA sequences were randomly subjected to MEME analysis [25]. WebLogo was used for visualization of the DNA consensus sequence. To identify genes from the ChIP-seq enriched in the tooth with the consensus motif, the ChIP-seq browser extensible MACs peaks in Excel format (GSM4339188) and the Pemberton et al. data set of 100 tooth-enriched genes [14] were joined by employing the Excel power query option. To elicit potential DNA motifs in tooth-enriched genes, 40 bp on either side of the MAX peak of the 37/100 shared genes (i.e., MACs peak and tooth-enriched) were analyzed by MEME.

2.3. Single-cell RNA sequencing (scRNA-seq) and UMAP analysis

scRNA-seq from mouse P7 incisor was described in detail previously [16] and only a brief description is presented. Here, scRNA was processed through the 10X Genomics workflow, cDNA libraries were prepared using a Nextera XT kit (Illumina) and sequencing performed on a NextSeq500 sequencer (Illumina). After Cell Ranger processing, removing low quality cell reads (cells with less than 700 identified genes and containing a mean value of greater than 9% mitochondrial genes) 6260 cells with 7356 median UMIs per cell were included in the analysis. Unsupervised clustering of cells populations was performed using Uniform Manifold Projection Approximation (UMAP) analysis from the Seurat R package [26].

2.4. Sp6 transcriptional regulation of the Hapln1 and Sp6 gene promoters

The 0.9 kb promoter region of the Hapln1 gene [27] was used to create a construct, pGL3-pHapln1, in the pGL3 promoterless firefly luciferase reporter plasmid (Promega). A mutant of the Hapln1 promoter (pGL3-pHapln1–4bp) was also constructed containing a 4 bp substitution in the core TAATTA sequence located at −219 to −218 and was changed to TAGCCG. Previous studies elucidated an important regulatory element in the Hapln1 promoter which was designated HLBS [28] and in this study, found to overlap the Sp6-enriched sequence. To study the transcriptional activity of this sequence, four direct repeats of this Sp6 target sequence (HLBS; 5’-GTTAGGCTGTAATTAGAGGA) were utilized driving a minimal promoter in pGL4.23 firefly luciferase reporter plasmid and was designated pGL4–4X-HLBS [28]. A series of Sp6 target mutants were constructed in a similar fashion in pGL4.23 including pGL4–4X-HLBS-Mut1 (5’- ACCGAGCTGTAATTAGAGGA), pGL4–4X-HLBS-Mut2 (5’-GTTAGATCAC AATTAGAGGA), pGL4–4X-HLBS-Mut3 (5’-GTTAGGCTGTGGCCGGAGGA), and pGL4–4X-HLBS-Mut4 (5’ GTTAGGCTGTAATTAAGAAG). A plasmid containing 53 bp (5’-CTCACCTTATAATTTATGCAAATCAACTCTGTAATTAACTA-TAATTTGGAGGC) from the Sp6 mouse gene itself located at −2015 was also subcloned in a plasmid designated pSp6–53bp. For analysis, HEK-293T cells were transfected with 1 μg of either empty vector (pGL4.23), wild type pGL4.23–4X HLBS, or the corresponding mutant targets along with 25 ng of pCMV-Sp, 25 ng of pCMV-Sp6-VP16 (a construct for enhanced cofactor recruitment), and 25 ng of CMV-Renilla luciferase. Transfections were performed in duplicate, and 48 h post-transfection, the cells were lysed in passive lysis buffer (Promega). Transcriptional activity normalized to transfection efficiency by sequentially assaying for firefly and Renilla luciferase activity with a luminometer (Berthold LB 940). Data was plotted comparing the pGL4.23 basal promoter co-transfected with a CMV expression vector without the gene insert.

3. Results

3.1. ChIP-seq analysis identifies Sp6 regulatory sites

To understand the gene networks controlled by the Sp6 in tooth development, genome-wide ChIP-seq analysis was utilized to identify in vivo genomic targets. Mouse mesenchymal and epithelial molar tissues were separately dissected, formaldehyde crosslinked, sonicated and used in subsequent ChIP analysis. Either input or affinity-purified rabbit anti-Sp6 antibody immunoprecipitated DNA pools were utilized, in which recovered DNA pools were sequenced to a depth of approximately 40 million reads per sample. Bioinformatic analysis was used to remove redundant sequences and identity candidate peaks in target genes based on the relative frequency of enriched genomic sequences. The ChIP-seq dataset derived from tooth mesenchyme, accession (GSM4339188), was utilized in this report.

From Sp6 peak evaluation from the tooth mesenchyme using MACs with a cutoff p-value of 10⁻⁷, identified 11,436 enriched peaks. Approximately 29% of the 11,436 peaks were located within 10 Kb of the transcriptional start sites of the target genes, in which a large cluster of the Sp6 sites were often found directly preceding the transcriptional start sites (Fig. 1A). From unbiased, bioinformatic analysis of the nucleotide sequences within the Sp6 peak binding regions of the 1400 target genes, a consensus sequence comprising the nine-nucleotide binding sequence of C-T-A/G-T-A-A-T-T-A was identified with high probability (4.0e-219, Fig. 1B). Evaluation of the dataset using the IGB genome browser revealed that many tooth genes contained Sp6 peaks in their gene promoters including amelogenin, (Amelx), ameloblastin (Ambn), enamelin (Enam), dental sialophosphoprotein (Dspp), membrane-type matrix metalloprotease-20 (Mmp20) and several matrix genes (Col1a2, Col11a2 and Hapln1 (Fig. 1C). Peaks for Ambn, Enam, Mmp20, and Dspp were all located within 300 bp upstream of the start of transcription, but for Amelx, the peak was located at 5 kb upstream (Fig.1C). No prominent peaks were observed in these genes or other evaluated Sp6 target genes using control input sequence (only the input tracing for Sp6 is shown, as well, see Supplementary Fig. 2). MEME analysis also defined the core sequence motif in Sp6, Hapln1, and Col11a2 (Fig. 1D, and Supplementary Fig. 1).

Fig. 1.

ChIP-Seq identifies Sp6 promoter binding sites and a consensus binding motif. In total, 11,436 high confidence Sp6-enriched peaks were identified from tooth mesenchymal tissue. A. 3305 ChIP-seq peaks were identified within 10 Kb of the transcriptional start sites of the target genes. B. MEME bioinformatic analysis of the ChIP-seq data set identified a Sp6 consensus binding motif. C. ChIP-Seq peaks are shown for Sp6 binding sites identified near the transcription start sites (arrow) for the following genes: Sp6, Amelx, Ambn, DSPP, Enam, Col1a2, Col11a2, Hapln1, Mmp20. The black tracing for each gene and the numbers above it represents the Sp6-enriched peak. For this figure, the Sp6 and Col1a2 enriched peak heights are not drawn to scale. D. Extracted peaks from MEME analysis shows CTGTAATTA at SP6, Hapln1, Col11a2, (Col1a2, a major tooth matrix protein did not occur in the Pendleton dataset was evaluated visually).

To evaluate the extent of Sp6 interactions in tooth-specific genes in an unbiased fashion, Sp6 ChIP peak data in browser extensible data (BED) format was correlated to 100 genes known to be highly expressed in the developing tooth determined by microarray analysis [14]. Comparison of the data files showed that 37 genes of the 100 tooth-enriched genes harbored Sp6 binding sites near their promoter or enhancer (Table 1).

Table 1:

Sp6 tooth-enriched genes.

Gene Symbol	Gene Name	Sp6 Peak Locationa
Ambn	Ameloblastin	−172
Amelx	Amelogenin	−5499
Asxl3	Sex combs like	−4873
Bmp7	Bone morphogenetic protein 7	+ 1314
Cldn1	Claudin 1	+3125
Col11a2	Collagen 11a2	−446
Cxadr	Coxsackie adeno receptor	−1520
Dlx2	Dlx homeobox 2	−1538
Dlx3	Dlx homeobox 3	+8935
Dlx4	Dlx homeobox 4	+425
Dlx5	Dlx homeobox 5	−124
Dspp	Dentin sialo phospho protein	−339
Dlx6os1	Dlx6 opposite strand	−562
Edg3	G-protein-receptor 3	+971
Enam	Enamelin	−168
Fras1	Fraser syndrome 1 homolog	+2274
Gal3st4	Galactose-3-O-sulfotransferase 4	−6722
Hapln1	Hyaluronan and proteoglycan binding link protein 1	+9622
Ibsp	Integrin binding sialoprotein	−221
Klk4	Kallikrein 4	−838
Maf	Avian fibrosarcoma homolog	−886
Mmp14	Matrix metallopeptidase 14	−264
Mmp20	matrix metallopeptidase 20	−135
Msx2	Msx2 homeo box	−203
Nedd4	Neural down-regulated gene 4	+4854
Papln	Sulfated glycoprotein	+749
Phex	Phosphate regulating gene	−270
Pitx2	Paired-like homeodomain	+460
Rnd3	Rho family GTPase 3	+23
Shh	Sonic hedgehog	+28
Slc39a8	solute carrier 39	+2372
Sox4	SRY-box4	−17
Sp6	Sp6	−2115
Sp7	Sp7	−84
Trp63	Transformation related protein 63	−28
Vangl2	Van gogh like-2	−1097
Wnt6	wingless-6	−7650

^aNumbers denote the location of the Sp6 target peak in either the promoter (negative number) or the first intron (positive number).

Besides major tooth matrix proteins, two tooth proteases were identified, kallikren4 (Klk4) and membrane-type matrix metalloprotease-14 (Mmp14), both of which are involved in enamel formation and dentinogenesis [6,29]. A large number of established tooth-specific transcription factors controlling dental tissue patterning and gene expression harbored Sp6 binding sites in their promoter or first intron regions including Dlx2, Dlx3, Dlx4, Dlx5, Sp6, Sp7, Pitx2, and Msx2 [18–20]. Additionally, genes for growth factors (Bmp7, Wnt6), and matrix proteins (Hapln1 (hyaluronan and proteoglycan binding link protein 1)) and Col11a2 had Sp6 binding sites in their promoters or first introns. The matrix protein Col11a2 was previously reported in tooth odontogenic cells [8]. Additional Sp6 ChIP-seq gene mapping data are found in Supplementary Table I and the Excel datasheets in GEO:GSE145909.

3.2. Evaluation of scRNA-seq from the developing tooth

Based on the finding that numerous tooth-specific genes had in vivo Sp6 binding sites near their transcriptional start sites, we investigated the mRNA expression of Sp6 by scRNA-seq from developing mouse incisor using the dataset (GSE146855). UMAP employing Seurat R-based package 3.0 was utilized. UMAP analysis showed Sp6 transcripts throughout multiple cell types of the incisor, including pre-ameloblast and ameloblast cell populations that co-expressed transcripts for Amelx, Ambn, DSPP, and Enam. Sp6 transcripts were also identified in a previously defined odontogenic cell population containing transcripts for the bone and tooth mineralization factor sphingomyelin phosphodiesterase 3 (Smpd3, [30]). Transcripts for Col1a2, Col11a2, Hapln1 (Fig. 2) and additional matrix proteins such as versican (not shown) were identified in the odontogenic cell populations.

Fig. 2.

A. Differential expression analysis of Sp6, Amelx, Ambn, Enam, MMP20, DSPP, Cola11a2, Hapln1, Col1a2 by UMAP analysis from mouse P7 incisor dataset (GSE146855). Gene signatures and inferred functions were previously described: IEE, inner enamel epithelium; Pre-Am, pre-ameloblast; Am, ameloblast; SI/SR/OEE, stratum intermedium, stellate reticulum, and outer enamel epithelium; DF, dental follicle; Od, odontoblast; PDL, periodontal ligament; Ery, erythrocyte; Leuko, leukocyte; Endo, endothelium; Neu, neural clusters [30].

3.3. Sp6 activation of the Hapln1 gene promoter

Based on the co-expression of SP6 with a number of tooth-specific target genes, we tested whether one such target gene, Hapln1, was regulated by Sp6. For these studies, cell transcription reporter assays were utilized employing a 900 bp Hapln1 genomic promoter fragment driving luciferase reporter activity. In Hapln1, the Sp6 consensus sequence is located at −220 bp in the promoter and overlapped a previously characterized HLBS target sequence [28]. Transfection of Sp6 or the Sp6-Vp16 chimeric activator expression vectors in HEK-293 cells activated the basal promoter activity of pGL3 control vector (Fig. 3A). Sp6 co-transfection with the pGL-Hapln1 promoter also demonstrated a 2-fold activation, and the Sp6-Vp16 chimeric construct produced 5-fold transcriptional activation (Fig. 3A). Importantly, a pGL-Hapln1 reporter with a 4bp substitution (i.e., from ATTA to GCCG) within the core Sp6 binding motif in the promoter demonstrated no significant activation by either Sp6 or the Sp6-Vp16 constructs (Fig. 3A). Lastly, a 53 bp fragment from the mouse Sp6 promoter region containing the Sp6 binding motif was also activated 3-fold by SP6 expression and increased over 30-fold by Sp6-Vp16 construct (Fig. 3A). These results demonstrate that the consensus Sp6 regulatory motif found in the Hapln1 and Sp6 promoters are functionally active in vivo and regulatable by Sp6.

Fig. 3.

Hapln1 and Sp6 gene promoters are transactivated by Sp6. A. Transfection experiment in HEK-293 cells were performed with luciferase reporter constructs for the promoterless pGL3 vector, 0.9 Kb promoter (pHapln1), a promoter with a 4 bp substitution (pHapln1-4bpSub) in the core AT-rich Sp6 binding sequence and with pSp6–53 bp, a construct containing 53bp of the rat Sp6 gene containing Sp6 target sequence. In these experiments, no addition, pCMV empty vector, pCMV-Sp6 and pCMV-Sp6-Vp16 expression vectors were utilized and following transfection, luciferase reporter activity was assessed. B. As described in the Material and Methods, transient transfection experiments were performed using pGL4.23 (a basal promoter luciferase reporter vector) and corresponding 4X wild type and mutant 4X HLBS reporter constructs.

Additional experiments were designed to test whether the Sp6-binding motif from the Hapln1 promoter was itself necessary and sufficient for activation. Four copies of the Sp6 target sequence (designated HLBS) were placed in front of the basal promoter and tested in co-expression experiments. As shown in HLBS reporter constructs were activated by Sp6 resulting in an up-regulation of promoter activity by 30-fold in HEK293 cells (Fig. 3B). Co-transfection experiments with other transcription factors, including Sox9 and β-catenin, were unable to activate the HLBS reporter (data not shown). Mutations within the HLBS sequence showed markedly less activity inducible by Sp6 transactivation, in which changing the core sequence within the motif resulted in no promoter activity (Fig. 3B). Specifically, two mutants pGL4–4X-HLBS-Mut2 and pGL4–4X-HLBS-Mut3 disrupting the AT-rich region within the consensus motif were essentially transcriptionally inactive (Fig. 3B). Taken together, these results confirm that Sp6 is a regulator of both the Sp6 and the Hapln1 gene and likely other matrix genes.

4. Discussion

Tooth development is a highly complex process involving the expression of many different specialized matrix proteins. Evaluating a list of the 100 most tooth-enriched genes revealed that 37 tooth-specific genes has Sp6 binding sites identified by ChIP-seq. Sp6 motifs were found in ameloblastin, amelogenin, enamelin and mmp20, representing some of the most abundant tooth proteins. Further evidence for a key regulatory role of the Sp6 transcription factor in tooth morphogenesis was the finding that several other critical tooth-enriched transcription factors including Dlx2, Dlx3, Dlx4, Dlx5, and Msx2, contained Sp6 binding sites in their promoter regions. In many tooth-related genes, the Sp6 binding sites were enriched in close proximity to the start of gene transcription. These findings along with the observation that Sp6 gene itself had two Sp6 binding sites in its promoter suggests that it is a major coordinator acting at a central hub of this gene network during tooth development. The large number of gene targets identified provides an extensive road map for more detailed analyses of tooth development.

Consistent with the study by Wang, X. et al., 2020 [30], a cell population with a gene expression odontogenic profile was identified. This cell population contained transcripts for matrix genes Col11a2, Col1a2, and Hapln1, which all harbor Sp6 binding elements occurred in the promoter or first intron. The finding of Hapln1 in the developing tooth is consistent with multiple other studies [13,14,31,32]. Here we further demonstrate that a 0.9 kb promoter region of the Hapln1 gene contains a single Sp6 site was regulated by Sp6 co-expression, and this regulation was lost when the conserved sequence in the core motif was mutated in the promoter. Sp6-dependent transcriptional activity was also substantiated when four copies of the Sp6 binding site from Halpn1 was placed in front of a basal promoter. An Sp6-binding site derived from the Sp6 gene itself was controlled by Sp6 expression highlighting a regulatory loop that may amplify matrix deposition during critical developmental stages. Although our in vitro analysis with these regulatory regions from Hapln1 and Sp6 suggest Sp6 acts as an activator, it is possible that the transcriptional milieu during in vivo development is much more complex, and Sp6 may function as either a repressor or an activator depending on the stoichiometry of other nuclear factors.

Taken together these studies shed light on the Sp6 transcription factor and its regulatory activity via the core motif containing the CTg/aTAATTA DNA sequence. Mechanistically, the Sp6 TAATTA identified contrasts to other well-characterized GC-rich binding sequences required for activation by Sp1-Sp4 clade members. The Sp6 consensus sequence identified in our study is nearly identical to the known binding motif reported for Sp7, a factor controlling osteoblast specification of bone formation [22]. Sp7 knockout mice exhibit skeletal defects in ossification through alterations in the regulation of mineralizing chondrocytes. Comparison of the Sp6 and Sp7 ChIP-seq datasets revealed multiple common matrix gene targets (e.g., Hapln1, Col1a1, Col11a2) found in both dental and bone cell types (Supplementary Fig. 2). These findings demonstrate that Sp6 and Sp7 members can share essentially identical in vivo DNA elements, but function in a tissue-specific fashion by expression in either odontoblast-like cells, or osteoblasts, respectively. Overall these observations provide potential mechanistic insight for the development of tooth and bone mineralized tissues.