Characterization of Odontogenic Differentiation from Human Dental Pulp Stem Cells Using TMT-Based Proteomic Analysis

Background The repair of dental pulp injury relies on the odontogenic differentiation of dental pulp stem cells (DPSCs). To better understand the odontogenic differentiation of DPSCs and identify proteins involved in this process, tandem mass tags (TMTs) coupled with liquid chromatography-tandem mass spectrometry (LC-MS/MS) were applied to compare the proteomic profiles of induced and control DPSCs. Methods The proteins expressed during osteogenic differentiation of human DPSCs were profiled using the TMT method combined with LC-MS/MS analysis. The identified proteins were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses. Then, a protein-protein interaction (PPI) network was constructed. Two selected proteins were confirmed by western blotting (WB) analysis. Results A total of 223 proteins that were differentially expressed were identified. Among them, 152 proteins were significantly upregulated and 71 were downregulated in the odontogenic differentiation group compared with the control group. On the basis of biological processes in GO, the identified proteins were mainly involved in cellular processes, metabolic processes, and biological regulation, which are connected with the signaling pathways highlighted by KEGG pathway analysis. PPI networks showed that most of the differentially expressed proteins were implicated in physical or functional interaction. The protein expression levels of FBN1 and TGF-β2 validated by WB were consistent with the proteomic analysis. Conclusions This is the first proteomic analysis of human DPSC odontogenesis using a TMT method. We identified many new differentially expressed proteins that are potential targets for pulp-dentin complex regeneration and repair.


Introduction
The development of dental-derived mesenchymal stem cells is an intriguing milestone of regenerative medicine, in view of their capability of differentiating into osteogenic, adipogenic, and chondrogenic lineages, representing a promising source for the bone and dentin mineralization treatment strategies in the future [1]. Dental pulp stem cells (DPSCs), a group of dental-derived mesenchymal stem cells derived from the neural crest, are considered important seed cells in dental tissue engineering for pulp-dentin complex regeneration [2,3]. When teeth are stimulated by dental caries, wear, or trauma, resident DPSCs migrate quickly to the injured site because of their suited location to secrete proregenerative cytokines to respond to the inflammatory microenvironment, then proliferate and differentiate into odontoblasts [4]. The formation of restorative dentin produced by odontoblasts could prevent disease progression to preserve dental pulp vitality [5,6]. When new regenerated dentin tissue is well integrated into the previously damaged teeth, clinical healing occurs [7,8]. This repair potential of dental pulp tissue provides a reliable biological basis for the study of pulp-dentin complex regeneration.
Proteomics can be used as an unbiased, global informatics tool to discover information about all protein expression levels and posttranslational modifications in cells or tissues [9]. The main quantitative techniques used in proteomics include gel-based proteomics (two-dimensional fluorescence difference gel electrophoresis (2D-DIGE), sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)) and gel-free proteomics (mass spectrometry-based) [10][11][12]. Quantitative proteomics is crucial to understand the comprehensive protein expression profile underlying the molecular mechanisms of biological processes and disease states [13]. Most quantitative proteomic techniques involve the isotopic labeling of proteins or peptides in two or more experimental groups, which can then be differentiated by mass spectrometry. At present, the technologies of isobaric tags for relative and absolute quantitation (iTRAQ) and tandem mass tags (TMTs), chosen according to the sample number, are two widely used quantitative proteome labeling techniques [14,15].
Wei et al. used 2D-DIGE and matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF-MS) technologies to explore the proteomic profile at the early stage (7 days) of odontogenic differentiation in dental pulp cells (DPCs). Twenty-three proteins were screened out in their study. The expression of heterogeneous nuclear ribonucleoprotein C, annexin VI, collagen type VI, and matrilin-2 was validated by quantitative real-time polymerase chain reaction (qRT-PCR) and western blotting (WB) [16]. In 2013, Kim et al. analyzed the secretome of human DPSCs after 3 days of odontogenic differentiation using SDS-PAGE/LC-MS/MS. The protein lysyl oxidase-like 2 (LOXL2) inhibited the odontogenic differentiation of DPSCs [17]. Gelbased techniques were applied in the above two studies. However, the gel-based techniques' low sensitivity, poor separation, and poor resolution for particular types of proteins and their lack of accuracy for an individual protein within a mixed spot undermine their prospects for profound and accurate proteomic research [18]. As an alternative, gel-free quantitative proteomics with greater accuracy and sensitivity is needed for studies about the protein profile of DPSCs during odontogenic differentiation.
Our study is the first investigation of proteomic profiles in the process of odontogenic differentiation of human DPSCs using TMT combined with LC-MS/MS and provides further insight into the molecular mechanisms in reparative dentinogenesis.

Materials and Methods
2.1. DPSC Isolation, Culture, and Identification. Healthy and intact premolars were extracted from 23 healthy individuals (13 females and 10 males in the 15-25 age range, mean age of 19.7) who were receiving orthodontic treatment at the Department of Stomatology, Nanfang Hospital, Southern Medical University. Teeth had been collected from April to December 2019. This project was approved by the Ethics Committee of Nanfang Hospital, Southern Medical University. DPSCs isolated from the pulp tissue of these premolars were cultured in routine media as we described previously [19].

Odontogenic
Induction. DPSCs were induced with an odontogenic differentiation medium which contains 100 nmol/L dexamethasone, 50 mg/mL ascorbic acid, and 10 mmol/L β-glycerophosphate (Sigma-Aldrich, St. Louis, MO, USA) in 6-well plates. DPSCs in the noninduced group were cultured in the DMEM+10% FBS. After 14-day culture, the cells were stained with Alizarin Red S (ARS A5533, Sigma-Aldrich). We observed and photographed the calcium nodules with a microscope (Crystal Violet, Amresco, Solon, OH). ALP staining was performed after 7 days of culture in the odontogenic differentiation medium following the protocol of the NBT/BCIP Staining Kit (Beyotime Biotechnology, Shanghai, China).

Preparation of Protein
Samples. Induced DPSCs were cultured for 14 days, and SDT lysate (4% SDS, 100 mM Tris-HCl, 1 mM DTT, and pH 7.6) was added. After ultrasound (80 W, 10 s per operation, 15 s intermittency, and 10 cycles), the cell lysates were bathed at 100°C for 15 min and then centrifuged at 14,000g for 40 min. The supernatant was kept, and a BCA kit was used for protein quantification.
2.4. SDS-PAGE Separation. Twenty micrograms of protein was taken from each sample, and 5x loading buffer (10% SDS, 0.5% bromophenol blue, 50% glycerol, 500 mM DTT, 250 mM Tris-HCl, and pH 6.8) was added. 12.5% SDS-PAGE electrophoresis (constant current 14 mA, 90 min) was performed after 5 min of boiling in a water bath, and the gel was then stained with Coomassie blue.

Filter-Aided Sample Preparation (FASP Digestion).
Thirty microliters of protein solution was taken from each sample. DTT (100 mM) was added separately, and the solution was cooled to room temperature after 5 min in a boiling water bath. We added 200 μL UA buffer (8 M urea, 150 mM Tris-HCl, and pH 8.0) and mixed it well, then transferred it to a 10 kD ultrafiltration centrifuge tube, centrifuged the tube at 14,000g for 15 min, discarded the filtrate, and repeated this centrifugation once. We added 100 μL IAA buffer (100 mM IAA in UA), oscillated the sample at 600 rpm for 1 min, let it react at room temperature in the dark for 30 min, and centrifuged it at 14,000g for 15 min. We added 10 μL UA buffer and centrifuged the sample at 14,000g for 15 min. This step was repeated twice. We next added 100 μL of 100 mM TEAB buffer and centrifuged the sample at 14,000g for 15 min. This step was also repeated twice. After 40 μL trypsin buffer (4 μg trypsin in 40 μL 100 mM TEAB buffer) was added, the sample was oscillated at 600 rpm for 1 min and placed at 37°C 2 BioMed Research International for 16-18 h. The collection tube was replaced, and the tube was centrifuged at 14,000g for 15 min. Then, 40 μL of 10fold diluted 100 mM TEAB buffer was added, and the sample was centrifuged at 14,000g for 15 min. The filtrate was collected, and the peptide was quantified for its OD280.
2.6. TMT Labeling. Each sample was labeled with 100 μg of peptide fragments according to the manufacturer's instructions for the TMT labeling kit (Thermo Fisher Scientific, Waltham, MA, USA). Peptides of the two groups were labeled with different TMTs: three biological repeats of the control group were labeled with TMT-126, TMT-127, and TMT-128, respectively, and three biological repeats of the exercise group were labeled with TMT-129, TMT-130, and TMT-131, respectively.

Peptide Fractionation.
After mixing the labeled peptide segments of each group in equal amounts, classification was performed using a high-pH RP spin column. After peptide labels were mixed and lyophilized, 100 μg was diluted with 300 μL of 0.1% trifluoroacetic acid and transferred to a high-pH RP spin column. The FT component was collected centrifugally, 300 μL of pure water was added, the wash component was collected centrifugally, and step gradient elution was started. After freeze-drying, the sample was redissolved with 12 μL of 0.1% formic acid, and the peptide concentration was calculated by determining the OD280.
2.8. High-Performance Liquid Chromatography (HPLC) and LC-MS/MS Analysis. Each fraction was injected for nano-LC-MS/MS analysis. Each sample was separated by a highperformance liquid-phase system, EASY-nLC with a nanoliter flow rate. The chromatographic column was balanced with 95% buffer A (0.1% formic acid aqueous solution). The sample was loaded onto the loading column (Thermo Scientific Acclaim PepMap 100, 100 μm × 2 cm, Nanoviper C18) by an automatic sampler and then separated by an analysis column (Thermo Scientific EASY-Column, 10 cm, ID75 μm, 3 μm, C18-A2) at a flow rate of 300 nL/min by IntelliFlow technology. Samples were separated by liquid chromatography and analyzed by a Q Exactive mass spectrometer. The analysis duration was 60/90 min, the positive ion mode was used for detection, the scanning range of the parent ions was 300-1800 m/z, the primary mass spectrum resolution was 70,000 at 200 m/z, the AGC target was 3e6, the primary maximum IT was 10 ms, the number of scan ranges was 1, and the dynamic exclusion was 40 s. The mass-tocharge ratio of polypeptides and polypeptide fragments were determined according to the following methods: 10 fragment patterns (MS2 scan) were collected after each full scan, the MS2 activation type was HCD, the isolation window was 2 m/z, the secondary mass spectrum resolution was 17,500 at 200 m/z (TMT6plex) or 35,000 at 200 m/z (TMT10plex), there was 1 microscan, the secondary maximum was 60 ms, the normalized collision energy was 30 eV, and the underfill was 0.1%.

Protein Identification and Quantitative
Analysis. MS/MS spectra were searched using the MASCOT engine (Matrix Science, London, UK; version 2.2) embedded into Proteome Discoverer 1.4. The search criteria were set as follows: all tryptic specificity was required; 2 missed cleavages were allowed; carbamidomethylation (C), TMT6plex (N-terminal), and TMT6plex (lysine, K) were set as the fixed modifications; oxidation (methionine, M) and TMT6plex (tyrosine, Y) were set as the variable modifications; peptide mass tolerances were set at 20 ppm for all MS1 spectra acquired; and fragment mass tolerances were set at 0.1 Da for all MS2 spectra acquired. The peptide false discovery rate (FDR) was set as ≤0.01. All peptide ratios were normalized by the median protein ratio. The thresholds were set at the ratio of exercise/control ≥ 1:2 and p value ≤ 0.05 for upregulation. Similarly, the thresholds were set at the ratio of exercise/control ≤ 0:83 and p value ≤ 0.05 for downregulation (refer to previous studies [21,22]).

Gene Ontology (GO) Function
Notes. The process of GO annotation of the target proteins set by Blast2GO can be roughly summarized into four steps: sequence alignment, GO item extraction, GO annotation, and supplementary annotation. First, the protein sequences of differentially expressed proteins (FASTA format) were retrieved in batches from the UniProtKB database (version 2016_10). NCBI BLAST client software (ncbi-blast-2.2.28-win32.exe) was used to carry out a local search on the retrieved sequences to find the homologous sequence neural network annotations. In this work, the first 10 BLAST values of each query sequence were retrieved if they were less than 1e − 3, and they were loaded into Blast2GO10 (version 3.3.5) for GO mapping and annotation. In the annotation process, the Blas-t2GO Command Line annotates the GO entries extracted in the entry extraction process to the target protein sequence by comprehensively considering the similarity between the target protein sequence and the alignment sequence, the reliability of the source of the GO entries, and the structure of the GO directed acyclic graph. After the annotation was completed, in order to further improve the annotation efficiency, conserved motifs found in the target protein sequence in the EBI database were searched through InterProScan, and the functional information related to the motifs was annotated to the target protein sequence. ANNEX was run to further supplement the annotation information, and links were established between different GO categories to improve the annotation accuracy. For each category, a two-tailed Fisher exact test was employed to test the enrichment of the differentially expressed protein against all identified proteins. The GO with a corrected p value < 0.05 is considered significant.

KEGG (Kyoto Encyclopedia of Genes and Genomes)
Pathway Notes. In the KEGG database, KO is the classification system of genes and their products. Orthologous genes with similar functions in the same pathway and their products are divided into a group, and the same KO (or K) marker is applied to them. When carrying out the KEGG pathway annotation on the target proteome, KASS (KEGG Automatic  BioMed Research International Annotation Server) software was first used to compare the target proteome with the KEGG GENES database. The target proteome sequence was KO-classified, and the path information related to the target proteome sequence was automatically obtained based on the KO classification. The results were filtered by the following criteria: a corrected p value < 0.05 and protein counts > 5. 2.14. Statistical Analysis. Band intensity in WB images was quantified with ImageJ software. Each data point is expressed as the mean ± standard deviation ðSDÞ, and the assay was repeated at least three times. Statistical analysis was performed by the t-test and one-way ANOVA using SPSS 17.0 for Windows (SPSS, Chicago, IL, USA). Statistical significance was defined as p < 0:05.

Results
3.1. Characteristics of Human DPSCs. Cells emerged from the tissue bulk adhering to the dish and preformed obvious fibroblast-like morphology (Figure 1(a)) after 14 days of culture. Using a limited dilution technique, we obtained the DPSCs (Figure 1(b)). The protein level of ALP increased with a rapid increase after 7 days of odontogenic induction (Figure 1(d)) compared with the control group (Figure 1(c)). After 14 days of induction, mineralized nodules were seen in the induced group by ARS staining (Figure 1(f)), but not in the control group (Figure 1(e)). Flow cytometry was used to The cells were identified to be positive for CD29 and CD44 and negative for CD34 and CD45 (Figure 1(g)), indicating the mesenchymal lineage of hDPSCs.

Differentially Expressed Protein Profile.
To get an overview of the data, the expression of endogenous proteins in three induced groups and three control groups was analyzed using a TMT-based quantitative proteomic approach. A flow diagram of the TMT-based quantitative proteomic platform applied to identify proteomic profiles is shown in Figure 2. A total of 223 proteins that were differentially expressed between the induced and control DPSC groups were identified using TMT analysis and are shown in Tables S1 and S2. Hierarchical clustering showed that the expression levels of proteins in the differentiated group differed significantly from those in the undifferentiated group according to the fold change (greater than 1.2 or less than 0.83) and p value thresholds (less than 0.05). Among these, 152 proteins were upregulated and 71 were downregulated (Figure 3). Tables 1  and 2 list the top 20 upregulated and downregulated proteins.

Functional Classification of the Differentially Expressed
Proteins. GO analysis with the assistance of DAVID Bioinformatics Resources was conducted to identify the functions of proteins identified using the TMT technique. The detailed functional classifications of the differentially expressed proteins are shown in Figure 4(a). Briefly, the classification by biological processes showed that the proteins were mainly  Figure 2: Flow diagram of the TMT-based quantitative proteomic platform applied to identify proteomic profiles. DPSCs were induced for 14 days or not, and whole cellular proteins were extracted from the two groups and quantified. Following trypsin digestion of equal amounts of protein, the resolved peptides were labeled with TMT6plex reagents, fractionated by HPLC, and analyzed by LC-MS/MS. 6 BioMed Research International involved in cellular processes, metabolic processes, biological regulation, regulation of biological processes, responses to stimuli, cellular component organization, and biogenesis (>40% for each class). On the basis of molecular function, the proteins in our study were implicated in binding, catalytic activity, transporter activity, molecular function regulator, transcription regulator activity, etc. In the cellular component ontology, we found that the majority of enriched categories were associated with the cell, cell part, organelle, organelle part, membrane, etc. We then performed KEGG analysis to investigate the enriched pathways that the differentially expressed proteins participated in during odontogenic differentiation. We found that a total of 223 altered proteins could be mapped to 238 signaling pathways (p < 0:05) ( Table S3). The top enriched pathways of the altered proteins were thermogenesis, Alzheimer's disease, oxidative phosphorylation, etc.

Protein Interaction Network
Analysis. STRING database analysis was used to build a protein-protein interaction (PPI) network concerning the process of odontogenic differentiation in DPSCs. Most of the differentially expressed proteins were implicated in physical or functional interaction. In this PPI network, we found that 223 proteins were mapped to 14 known protein-protein interaction networks, and 22 proteins had an interaction score of more than ten (Table 3). Among these proteins, cytochrome c oxidase subunit 5A (COX5A) was the most vital hub, interacting with 23 proteins. FBN1 and TGF-β2, which were reported to be involved in odontogenesis, were present in the most complex networks ( Figure 5).

Western Blotting Validation.
Two differentially expressed proteins, FBN1 and TGF-β2, involved in odontogenesis were selected and validated using western blotting. We found that the levels of FBN1 and TGF-β2 in induced cells were increased approximately 1.69-fold and decreased approximately 0.58-fold, respectively (Figures 6(a) and 6(b)). These validation results were consistent with the protein analysis data (Figure 6(c)).
Pleckstrin homology-like domain family B member 3 (PHLDB3) was the most upregulated protein (fold change: 8.93) among the differentially expressed proteins. PHLDB3 was once thought to be a tumor suppressor. Recent research found that PHLDB3 could increase tumor growth by inactivating p53 via a negative feedback loop in pancreatic, prostate, colon, breast, lung, and other common cancers [26]. There are few reports on PHLDB3 in cellular differentiation, and the potential role of PHLDB3 in odontogenic differentiation needs more research. The lowest-expressed protein was the cell growth-regulating nucleolar protein Ly1 antibody reactive (LYAR). LYAR is a zinc finger nucleolar protein that has been implicated in cell growth, self-renewal of ESCs, and medulloblastoma [27,28]. Li et al. reported that it is highly expressed in undifferentiated ESCs and plays a critical role in maintaining ESC identity. The reduced expression of LYAR in ESCs impairs their differentiation capacity [29]. The regulatory role of LYAR in ESC differentiation indicates that it might function in the odontogenic differentiation of DPSCs.
It was noted that there were significant differences in the expression of some proteins that are involved in the process of odontogenic differentiation, including FBN1 (upregulated fold change: 1.62) and TGF-β2 (downregulated fold change: 0.77). FBN1 was proven to be a key molecule forming the backbone of microfibrils [30]. More evidence has revealed that FBN1 plays an important role in the extracellular regulation of TGF-β as well as bone morphogenetic protein (BMP) activation and signaling, which are essential for odontogenic differentiation and reparative dentinogenesis [31]. Yoshiba et al. found that FBN1 upregulation was accompanied by wound healing in dental pulp tissue [32]. Our previous study found that the mRNA and protein expression of FBN1 was increased during the odontogenic differentiation of DPSCs, and the lncRNA-G043225/miR-588/FBN1 axis was involved in the odontogenic differentiation of DPSCs [19]. TGF-β2 was also identified to be an important regulator of DPSC differentiation [33]. Yu et al. induced the odontogenic differentiation of stem cells from dental apical papilla (SCAPs) and bone marrow (BMSCs) and tracked the expression of secretory proteins during early odontogenic differentiation using TMT combined with HPLC-MS/MS analysis [34]. The results revealed that TGF-β2 was significantly upregulated during the odontogenic differentiation of SCAPs and was significantly downregulated during the odontogenic differentiation of BMSCs. Tai et al. found that TGF-β2 possibly regulates the differentiation of pulp cells via an autocrine fashion by activation of the ALK/Smad2/3 signal transduction pathways at specific stages synergistically with other factors [33]. In our study, TGF-β2 was significantly downregulated during the odontogenic differentiation of DPSCs. Thus, we can conclude that TGF-β2 is a potentially important molecule with a distinct function in the regulation of odontogenesis. The exact regulatory mechanism of TGF-β2 in the odontogenic differentiation of DPSCs needs further in-depth research.
According to our GO analysis, the functions of proteins identified using the TMT technique included cellular processes, metabolic processes, binding, and catalytic activity, which were directly or indirectly related to cell differentiation. Functional annotation clustering and pathway analysis showed that oxidative phosphorylation, hypoxia-inducible factor-1 (HIF-1) signaling, and PI3K-Akt signaling were in the top 20 pathways. These three signaling pathways have been identified to regulate osteogenic/odontogenic differentiation through different underlying mechanisms [35][36][37][38].
Studying the interaction between proteins and the network formed by their interaction is of great significance to reveal the functions of proteins [39,40]. In PPI networks, proteins that interact directly with many other proteins are called hubs. A greater number of hubs indicate more 10 BioMed Research International importance to the whole system. Proteins with more interaction partners may play a key role in maintaining the balance and stability of the system, and they may be candidates for follow-up research [41]. In the PPI network constructed here, COX5A was the most vital hub. COX5A is a nuclear-encoded subunit of the terminal oxidase involved in mitochondrial electron transport [42]. Previous research indicated that the dysregulation of COX5A significantly affects COX function, thereby causing mitochondrial dysfunction in skeletal muscle, pulmonary arterial hypertension, lactic academia, and central nervous system diseases [43,44]. COX5A, as an enzyme involved in oxidative phosphorylation, may play an important role in the odontogenic differentiation of DPSCs. However, little information has been reported about the role of COX5A in odontogenesis. The function and regulatory mechanism of COX5A in the odontogenic differentiation of DPSCs require further exploration. The potential of odontogenic differentiation of DPSCs plays a crucial role in pulp-dentin complex regeneration in future clinical applications [2,3]. In our study, DPSCs were cultured with the odontogenic medium supplementing 10% FBS. However, the clinical application of DPSCs in regenerative medicine demands an in vitro expansion and in vivo delivery, which must deal with the biological safety issues about animal serum in the unique cell model. Marrazzo et al. reported a highly efficient in vitro reparative behavior of DPSCs cultured with platelet lysate. This novel model could apply platelet lysate as a valid candidate for FBS to culture and osteogenic-differentiate DPSCs [45].Therefore, we would like to refer to Marrazzo's protocol and establish an

Conclusions
Our study is the first to identify differentially expressed proteins related to the odontogenic differentiation of DPSCs using the TMT-based quantitative proteomic technique. Bioinformatics analyses suggest that a total of 223 proteins were differentially expressed during odontogenic differentiation of DPSCs and were mainly involved in cellular processes, metabolic processes, and biological regulation-related signaling pathways. Furthermore, FBN1 and TGF-β2, associating the odontogenic differentiation of MSCs, were confirmed to be differentially expressed, representing the potential regulation in the odontogenesis of hDPSCs. Our findings will facilitate a better understanding of the mechanisms of odontogenesis and provide a new perspective for research on pulp-dentin complex regeneration and repair.

Data Availability
The mass spectrometry proteomic raw data have been deposited to the ProteomeXchange Consortium via the PRIDE [46] partner repository with the dataset identifier PXD021279. The data supporting the research results can be obtained from the corresponding authors according to reasonable requirements.