Bioinformatics Identification of Candidate Biomarkers Associated with Periodontitis in Cytokine-Treated Periodontal Ligament Stem Cells

Summary

The current study is aimed at exploring dysregulated genes via bioinformatics analysis in cytokine-treated periodontal ligament stem cells. Upregulated ERC2-IT1 and downregulated EPB41L4A-AS1 were identified as key molecules in these cells, which may serve as candidate biomarkers potentially associated with periodontitis.

Abstract

The primary objective of the current study was to investigate the potential molecular mechanism underlying cytokine-induced periodontal ligament stem cells (PDLSCs). Microarray data from the GSE260558 dataset were retrieved from the gene expression omnibus (GEO) database. After normalization, the dysregulated genes were identified. Then, the different expressed mRNAs (DE-mRNAs) were subjected to gene ontology (GO) and kyoto encyclopedia of genes and genomes (KEGG) enrichment analyses. Subsequently, hub genes were screened out based on protein-protein interaction (PPI) network. Finally, the competitive endogenous RNA (ceRNA) regulatory networks containing hub genes were established by using ENCORI website. A total of 739 different expressed lncRNAs (DE-lncRNAs) and 809 different expressed mRNAs (DE-mRNAs) were identified by using the limma package in the R programming language. Then, the GO and KEGG analyses demonstrated that the DE-mRNAs were closely associated with inflammation and extracellular matrix formation in cytokine-treated PDLSCs. A protein-protein interaction (PPI) network was constructed for DE-mRNAs, and hub genes were obtained by evaluating the degree. A total of 12 upregulated hub genes and 12 downregulated hub genes were screened. These upregulated and downregulated hub genes were then used to construct a ceRNA regulatory network, respectively. In these networks, upregulated ERC2-IT1 and downregulated EPB41L4A-AS1 were considered as key lncRNAs in periodontitis. In conclusion, the bioinformatics analysis revealed several key genes and a ceRNA regulatory network that participated in the progression of cytokine-induced PDLSCs. Although this study identified several candidate biomarkers related to periodontitis, further experimental validation in animal models or clinical samples is necessary.

Introduction

According to the World Health Organization, periodontitis is recognized as the sixth most prevalent, ubiquitous, and irreversible chronic inflammatory disease among humans^1,2. In the present era, it affects nearly 15% of the global population^1,2,3. It has been reported that periodontitis is strongly correlated with a variety of factors such as age, gender, oral hygiene practices, smoking status, ethnicity, lower socioeconomic status, and a range of systemic diseases^2,3,4,5. Periodontitis, which has a notably high prevalence among older adults, is pathologically instigated by dysbiotic alterations in the subgingival microbiota that adheres to the teeth^6,7. If left untreated, this condition ultimately leads to bone destruction, tooth movement, and the unfortunate loss of teeth⁸. Consequently, a comprehensive understanding of the molecular mechanisms underlying periodontitis and the identification of diagnostic markers for the disease are urgently needed.

The structural and interactive complexity of periodontal tissue is widely regarded as one of the significant factors that contribute to the substantial difficulty encountered in the process of periodontal regeneration^9,10. Among the diverse range of mesenchymal stem cells (MSCs) utilized for the purpose of periodontal regeneration, periodontal ligament stem cells (PDLSCs) have been thoroughly demonstrated to be a highly reliable source for the formation of new cementum-like structures in the in vivo environment^11,12. In the recent preclinical studies that have been conducted, the treatment based on PDLSCs has displayed extremely promising outcomes in the field of tissue regeneration^13,14. Compared to patient gingival tissue, the transcriptomics analysis of PDLSCs displayed stronger functional orientation and therapeutic transformation potential^13,14.

Bioinformatics technology and bioinformatics analysis methods empower researchers to identify potential biological processes and associations with proteomic, genomic, and metabolomic statistics^15,16. In addition, the high-throughput nature of bioinformatics technology also enables researchers to analyze large-scale biological data sets, such as those generated from next-generation sequencing and microarray experiments^17,18. This capability allows for the identification of potential biomarkers for disease diagnosis and prognosis^19,20. For instance, a disintegrin and metalloproteinase 28 (ADAM28) and a disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like-3 (ADAMTSL3) were identified as novel biomarkers in gingival tissues from periodontitis patients via a comprehensive analysis based on bioinformatics technology²¹. Besides, several ferroptosis-related genes, including nuclear receptor coactivator 4 (NCOA4), solute carrier transporter 1A5 (SLC1A5), and heat shock protein B1 (HSPB1), were reported to have better diagnostic value in periodontitis²². However, the molecular mechanism of periodontitis deserves further exploration. In the present study, researchers endeavor to utilize bioinformatics technology to investigate hub genes in cytokine-induced PDLSCs and to discern potential biological processes and pathways linked to periodontitis. Bioinformatics analysis of cytokine-treated PDLSCs captures inflammation-driven transcriptional programs relevant to PDLSC dysfunction, which addresses the gap between PDLSC dysfunction and periodontitis. These core genes and pathways in cytokine-induced PDLSCs offer valuable insights into the unified mechanisms underlying periodontitis.

Protocol

1. Dataset

This analysis used publicly available data and did not involve human or animal subjects, in compliance with institutional and JoVE ethical guidelines. All the information related to the used platforms is available in the Table of Materials.

1. Select a total of 6 samples (GSM8119437, GSM8119438, GSM8119439, GSM8119440, GSM8119441, and GSM8119442) from the GSE260558 dataset (National Center of Biotechnology Information Gene Expression Omnibus, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE260558) for bioinformatics analysis.
2. In these samples, culture periodontal ligament stem cells (PDLSCs) in normal medium or in medium containing inflammatory cytokines for 7 days.
3. Extract the total RNA from PDLSCs and subject it to high-throughput sequencing²³. The process of bioinformatic analysis is shown in Figure 1.

2. Identification of DEGs

1. Analyze the expression profiles of lncRNA and mRNA using the limma software (version: 3.46.0) as described previously²⁴.
2. Perform Pairwise comparisons between cytokines and normal groups.
3. Use the false discovery rate (FDR) correction with a standard cutoff of < 0.05 to identify significantly differentially expressed genes. Use the adjusted p-values (FDR) to control the proportion of false positives among statistically significant results.
4. Use an adjusted P-value cutoff of < 0.05 and a fold change threshold of > 1 to identify significantly dysregulated lncRNAs or mRNAs.
5. Present the differential gene expression in volcano plots (ggplot2: version 3.4.1), which were generated using the R software (version 4.0.4).
6. List the top 10 dysregulated lncRNAs and mRNAs in tables.

3. Functional enrichment analysis

1. Perform GO enrichment analysis of the DE-mRNAs by using the DAVID online database (Database for Annotation, Visualization and Integrated Discovery, https://david.ncifcrf.gov/).
   1. Submit the dysregulated genes to the website, and select GO terms, including molecular function (MF), biological processes (BP), and cellular components (CC), to perform GO analysis.
   2. Submit the dysregulated genes to the Cytoscape (version: 3.8.2) plugin ClueGO [16]+ and Cluepedia app.
   3. Apply the Benjamini-Hochberg correction for multiple comparisons and set the significance threshold at an adjusted P-value < 0.05.
   4. Download the results from Cytoscape. Rank the enrichments in descending order based on P-value, and present the top 5 enrichments in each category.
2. The kyoto encyclopedia of genes and genomes (KEGG) is a valuable database resource for comprehending the high-level functions and impacts of biological systems (http://www.genome.jp/kegg/).
   1. Perform the KEGG analysis similarly by choosing KEGG-PATHWAY on the website.
   2. Set the significance threshold at an adjusted P-value < 0.05., and present top 10 enrichments in this study. Perform these enrichment analyses according to reported studies ^25,26.

4. PPI network

1. Predict and construct protein-protein interactions (PPI) network by using the search tool for retrieval of interacting genes (STRING, version 12.0).
2. Submit 318 upregulated mRNAs or 465 downregulated mRNAs to the STRING website to construct the PPI network.
3. Export the interaction data from the PPI network, and submit the data to the Cytoscape software.
4. Hide the individual node in the PPI network. Download the PPI network present in this study.
5. Calculate the topological properties of the network, including degree distribution, using CentiScaPe app, a plug-in of Cytoscape.
6. Quantify the local topology of each gene by summing up the number of its adjacent genes, and count the interactions for a given node.
7. Rank all genes from the PPI network in descending order based on their degree of connectivity.
8. Select the top 12 upregulated and top 12 downregulated genes from this ranked list as the hub genes.

5. miRNA-mRNA regulatory network

NOTE: The ENCORI database (version 12.0, https://rnasysu.com/encori/) was used to predict regulatory associations between microRNAs (miRNAs) and DE-lncRNAs (or hub genes).

1. Submit the upregulated hub genes or downregulated hub genes to ENCORI, and set the cut-off criterion program ≥5 in total six programs, including PITA, RNA22, miRmap, miRanda, PicTar, and TargetScan, for the prediction analysis in ENCORI.
2. Download the predicted miRNAs and submit them to ENCORI to obtain potential lncRNAs.
3. Download the predicted lncRAs and identify the overlapped lncRNAs between step 5.2 and Figure 2.
4. Submit the hub genes, miRNAs obtained in section 5.1, and lncRNAs obtained in step 5.3 to the Cytoscape software (version 3.4.0) to construct the ceRNA network.

Results

Identification of DEGs
The limma R software was used to analyze the differential expression profiles of lncRNAs and mRNAs. The differences between the two groups were determined by a non-paired t-test. As depicted in the volcano diagram (Figure 2), a total of 739 different expressed lncRNAs (DE-lncRNAs, p value<0.05, |log2FC|>1) were identified. Among these DE-lncRNAs, 340 upregulated lncRNAs and 399 downregulated lncRNAs were identified in the cytokines g...

Discussion

Periodontal ligament stem cells (PDLSCs) are regarded as key cells that play a crucial role in suppressing periodontal damage throughout both the progression and recovery stages of periodontitis^32,33. A considerable amount of evidence has clearly demonstrated that incubation under an inflammatory condition has the potential to significantly accelerate the senescence of PDLSCs^34,35. However, it is worth no...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
R	R Foundation for Statistical Computing	4.0.4	Core open-source environment for statistical computing. Serves as the foundation for all subsequent analyses. Typically used with RStudio.
Limma	Bioconductor Community (Lead Dev: Walter & Eliza Hall Institute)	3.46.0	Specialized package for differential expression analysis of microarrays and RNA-seq.
Cytoscape	Cytoscape Consortium	3.8.2	Open-source platform for network visualization. Requires separate installation; supports plugins like ClueGO and cytoHubba via App Store.
STRING	European Molecular Biology Laboratory	12	The world's largest database for predicted and known protein-protein interactions (PPI). Accessed via web or Cytoscape plugin; provides confidence scores.
ENCORI	Sun Yat-sen University	3	Public database focusing on ceRNA networks (miRNA-mRNA-lncRNA).