This article is mentioned in:

Introduction

Peri-implantitis is a destructive inflammatory disease that affects the tissues surrounding dental implants (1,2). It has been demonstrated that peri-implantitis is a crucial element in implant failure. In total, ~30% of patients that receive dental implants develop peri-implantitis (3). However, there are currently no effective therapeutic strategies against peri-implantitis. Moreover, the use of dental implants is constantly increasing, therefore an effective therapy to treat peri-implantitis is required. Thus, further investigations into the molecular pathophysiology of peri-implantitis are necessary in order to provide novel options for effective treatment (1).

Recently, a number of studies have investigated the pathological mechanisms underlying peri-implantitis progression. Becker et al (4) indicated that serglycin (SRGN) expression was significantly upregulated in peri-implantitis when compared with healthy individuals. It has been suggested that this gene may inhibit bone mineralization in vitro (5). Another study indicated that concentrations of the nuclear factor-κB (NF-B), soluble RANK ligand (sRANKL), osteoprotegerin (OPG) and sclerostin are significantly increased in patients with peri-implantitis (6). A number of typical bone matrix molecules, including collagen, type IX, α 1 (COL9A1), bone gamma-carboxyglutamate (Gla) protein (BGLAP) and secreted phosphoprotein 1 (SPP1) are decreased in the peri-implantitis tissues (1). Furthermore, it has been identified that fibroblasts are involved in the pathogenesis of peri-implantitis (7). The regulation of inflammatory mediators and matrix metalloproteinases (MMPs) in peri-implantitis fibroblasts function in the pathogenesis of the disease (8), and levels of the anti-inflammatory cytokine interleukin (IL)-10 are decreased in peri-implantitis (9). Furthermore, peroxisome proliferator-activated receptor γ (PPARγ) that can inhibit inflammation and promote osteoblast function is downregulated in the peri-implantitis tissues (10). However, other mechanisms associated with peri-implantitis have not been identified. Therefore, further research should focus on elucidating other potential mechanisms and investigate target genes for the treatment of peri-implantitis.

The microarray data of GSE57631 was used to confirm the similarities and differences of inflamed peri-implantitis tissues vs. normal peri-implantitis tissues at the mRNA level (1). In contrast to results from a previous study (1), the array data of GSE57631 was downloaded and the differentially expressed genes (DEGs) associated with peri-implantitis were analyzed using a biological informatics approach. In addition, functional enrichment analyses were performed for DEGs. In addition, a protein-protein interaction (PPI) network was established and four significant modules were analyzed. The present study aimed to identify the key genes and pathways of peri-implantitis, and identify possible significant mechanisms associated with it.

Materials and methods

Affymetrix microarray data

The array data of GSE57631 was downloaded from the Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) database, which was deposited by Schminke et al (1). Six samples of peri-implantitis tissues and two samples of normal tissues were included in the present study. The raw data and annotation files were downloaded for subsequent analysis, based on the GPL15034 platform (Affymetrix Human Gene 1.0 ST Array [HuGene10stv1_Hs_ENTREZ, Brainarray v14]; Affymetrix, Inc., Santa Clara, CA, USA).

Data preprocessing

The raw expression data was preprocessed using the robust multiarray average (11) algorithm by applying an oligo (12) in the R statistical software program in Bioconductor (http://www.bioconductor.org/). Background correction, normalization and a calculating expression were included in the process of preprocessing. A total of 18,977 gene expression values were obtained.

DEG analysis

The limma package (13) in Bioconductor was used to analyze DEGs in peri-implantitis samples compared with controls. In the process of the analysis, the P-values of DEGs were calculated using a t-test in the limma package. |log2FC|≥1 and P<0.05 were used as a cut-off criteria.

Gene Ontology (GO) and pathway enrichment analyses

GO is a tool that is used for gene annotation by collecting defined, structured, controlled vocabulary, which includes three main categories: Molecular function (MF), biological process (BP) and cellular component (CC) (14). Kyoto Encyclopedia of Genes and Genomes (KEGG) is a database used for associating related gene sets with their pathways (15). Moreover, DAVID, an integrated data-mining environment, is used to analyze gene lists (16).

GO annotation and KEGG pathway enrichment analyses were conducted for upregulated and downregulated genes by DAVID. Moreover, EASE ≤0.05 and gene counts ≥ were set as the threshold value.

PPI network analysis

The Search Tool for the Retrieval of Interacting Genes (STRING) (17) database provides information regarding the predicted and experimental interactions of proteins. The prediction method of this database came from neighborhood, gene fusion, co-occurrence, co-expression experiments, databases and text mining. Moreover, the interactions of protein pairs in the database are presented with a combined score. In the present study, DEGs were mapped into PPIs and a combined score of >0.9 was used as the cut-off value. In addition, PPI networks were constructed using Cytoscape software version 3.2.1 (18).

Topological properties of the PPI network, including degree (19), subgraph (20) and betweenness (21) centralities were determined using the R software package igraph (22), in order to analyze key genes in the network. A degree was used for describing the importance of protein nodes in the network. Subgraph centrality based on combining network topology and information of protein complexes was used to measure the importance of nodes in the network. The higher the degree and subgraph values were, the more important the nodes were in the network. Moreover, the betweenness centrality is an index describing the global topological properties of the network and could be used to describe how the nodes affect the connectivity between two nodes. Betweenness was defined as the ratio of the number of every path passed per node and the number of the shortest paths. The higher the betweenness values are, the greater the impact of the node in the network is. In addition, R software package igraph version 1.0.1 (22) was used for these three methods.

Module analysis

Network module was one of the characteristics of the protein network and may contain specific biological significance. The Cytoscape software package Molecular Complex Detection (MCODE) (23) was used to analyze the most notable clustering module. Next, the KEGG pathway enriched by DEGs in different modules was analyzed using DAVID online tool. EASE ≤0.05 and count ≥2 were set as the cutoff values.

Results

Data processing and DEG analysis

As shown in Fig. 1, a total of 819 DEGs including 759 upregulated and 60 downregulated genes, were identified in the peri-implantitis samples compared with the control ones. As a result, the number of upregulated genes was found to be significantly higher than that of the downregulated ones.

Figure 1. Heat map of differentially expressed genes. Green represents a lower expression level, red represents higher expression levels and black represents that there is no differential expression amongst the genes.

GO and pathway enrichment analyses

GO and KEGG pathway analyses were performed for upregulated and downregulated DEGs, respectively. The GO terms of upregulated DEGs were mainly associated with the proteasomal protein catabolic process, endoplasmic reticulum, melanosome and structural constituent of the cytoskeleton (Table IA). The downregulated DEGs were mainly enriched in the epidermis and ectoderm development, and the extracellular region part and cadmium ion binding (Table IB).

Table I. GO for differentially expressed genes.

Table I.

GO for differentially expressed genes.

A, Upregulated
Terms	Description	Counts	P-value
GO-BP
GO: 0010498	Proteasomal protein catabolic process	24	5.32×10−11
GO: 0043161	Proteasomal ubiquitin-dependent protein catabolic process	24	5.32×10−11
GO: 0032269	Negative regulation of cellular protein metabolic process	30	9.32×10−10
GO-CC
GO: 0005783	Endoplasmic reticulum	115	1.37×10−21
GO: 0042470	Melanosome	30	1.77×10−17
GO: 0048770	Pigment granule	30	1.77×10−17
GO-MF
GO: 0005200	Structural constituent of cytoskeleton	15	2.22×10−6
GO: 0070003	Threonine-type peptidase activity	8	1.13×10−5
B, Downregulated
Terms	Description	Counts	P-value
GO-BP
GO:0008544	Epidermis development	20	2.57×10−10
GO:0007398	Ectoderm development	20	9.84×10−10
GO:0006954	Inflammatory response	19	9.40×10−6
GO-CC
GO:0044421	Extracellular region part	35	1.05×10−5
GO:0005576	Extracellular region	54	1.33×10−4
GO:0005615	Extracellular space	25	2.79×10−4
GO-MF
GO:0046870	Cadmium ion binding	4	5.27×10−4
GO:0005507	Copper ion binding	6	6.22×10−3

[i] Terms represent the identification number of GO term; description represents the names of GO term; counts represent the number of genes enriched in GO terms. GO, Gene ontology; BP, biological process; CC: cellular component; MF, molecular function.

Two pathways that were significantly enriched by upregulated DEGs were Alzheimer's disease and the proteasome (Table II). Moreover, an important gene HSP90AA1 was significantly enriched in the antigen processing and presentation pathway and the nucleotide-binding oligomerization domain (NOD)-like receptor signaling pathway, and NFKB1 was enriched in NOD-like receptor signaling pathway in KEGG pathway enrichment analysis for upregulated genes. However, downregulated DEGs did not significantly enriched any pathways.

Table II. KEGG pathway enrichment analysis for upregulated differentially expressed genes.

Table II.

KEGG pathway enrichment analysis for upregulated differentially expressed genes.

Term	Description	Counts	P-value
hsa05010	Alzheimer's disease	25	1.69×10−5
hsa03050	Proteasome	12	5.36×10−5
hsa00190	Oxidative phosphorylation	18	1.19×10−3
hsa05012	Parkinson's disease	17	2.65×10−3
hsa05016	Huntington's disease	21	3.41×10−3
hsa00510	N-Glycan biosynthesis	9	4.31×10−3
hsa04142	Lysosome	15	7.11×10−3
hsa00480	Glutathione metabolism	9	7.27×10−3
hsa04612	Antigen processing and presentation	12	7.74×10−3
hsa04621	NOD-like receptor signaling pathway	10	8.63×10−3
hsa00970	Aminoacyl-tRNA biosynthesis	7	2.89×10−2
hsa00020	Citrate cycle (TCA cycle)	6	3.07×10−2
hsa01040	Biosynthesis of unsaturated fatty acids	5	3.51×10−2
hsa05130	Pathogenic Escherichia coli infection	8	4.45×10−2

[i] Term represents the identification number of the KEGG pathway; Description represents the name of the KEGG pathway; Counts represent the number of genes enriched in the KEGG pathway. KEGG, Kyoto Encyclopedia of Genes and Genomes; NOD, nucleotide-binding oligomerization domain-like; TCA, tricarboxylic acid.

PPI network analysis

Based on the STRING database, a total of 413 nodes and 1,114 protein pairs were obtained with a combined score >0.9 (Fig. 2). As shown in Fig. 2, the majority of the nodes in the network were upregulated DEGs in peri-implantitis samples. The top 20 most up- and dowregulated nodes are presented in Table III. A number of protein enzyme families, such as the proteasome subunit, α type 1 (PSMA1), proteasome subunit, beta type 1 (PSMB1) and proteasome subunit, α type 4 (PSMA4) were hub nodes based on the subgraph and degree centralities. Heat shock protein HSP90AA1 (90 kDa α, member 1), Ras-related C3 botulinum toxin substrate 1, NFKB1, Jun proto-oncogene were hub nodes with higher betweenness values.

Figure 2. Protein-protein interaction network of differentially expressed genes in peri-implantitis samples compared with the control ones. Red nodes, upregulated genes; green nodes, downregulated genes.

Table III. Nodes with higher values in subgraph centrality, betweenness centrality and degree centrality.

Table III.

Nodes with higher values in subgraph centrality, betweenness centrality and degree centrality.

Nodes	Subgragh	Nodes	Betweenness	Nodes	Degree
PSMA1	2.35×107	HSP90AA1	2.78×104	HSP90AA1	34
PSMB1	2.32×107	RAC1	1.77×104	JUN	27
PSMA4	2.31×107	NFKB1	1.38×104	RAC1	27
PSMB4	2.30×107	JUN	1.21×104	PSMA1	25
PSMA3	2.30×107	HIF1A	1.07×104	PSMA4	25
PSMB3	2.30×107	CDH1	9.21×103	PSMB1	25
PSMB6	2.30×107	SOS1	8.93×103	PSMC2	25
PSMC2	2.16×107	CANX	8.55×103	PSMB4	24
PSMD8	2.12×107	HDAC1	8.35×103	PSMA3	24
PSMD5	2.08×107	ATP5B	8.24×103	PSMB3	24
PSME4	2.08×107	GTF2B	8.17×103	PSMB6	24
PSMD10	2.04×107	HSPA5	7.84×103	PSMD10	24
PSME3	1.96×107	STAT3	7.46×103	PSMD8	23
UBE2D1	1.54×107	VCP	7.32×103	SEC61A1	23
CDC16	1.36×107	RPS5	7.14×103	RPN2	23
ANAPC5	1.36×107	NME1	6.31×103	NFKB1	23
BUB3	1.35×107	POLR2B	5.90×103	PSMD5	22
SEC61A1	1.23×107	P4HB	5.90×103	PSME4	22
SKP1	1.16×107	TXN	5.67×103	UBE2D1	22
HSPB1	1.05×107	RPN2	5.67×103	RPN1	22

[i] PSM, proteasome subunit.

Module analysis

In total, four modules (modules 1, 2, 3 and 4) with score >6 were detected by MCODE (Fig. 3). As shown in Table IV, PSMA1, PSMA4 and PSMB1 were hub nodes with higher node degrees in module 1, and ribophorin 1 (RPN1), ribophorin 2 (RPN2) and dolichyl-diphosphooligosaccharide-protein glycosyltransferase subunit (DDOST) were hub nodes in module 2. The hub nodes with higher node degrees in module 3 were pre-mRNA processing factor 8 (PRPF8), small nuclear ribonucleoprotein D1 polypeptide (SNRPD1; 16 kDa) and small nuclear ribonucleoprotein polypeptides B and B1 (SNRPB), and the hub nodes in module 4 were HSP90AA1, ATP synthase, H+ transporting, mitochondrial Fo complex, Subunit F2 (ATP5J2) and ATP synthase, H+ transporting, mitochondrial F1 complex and beta polypeptide (ATP5B). Furthermore, the proteasome pathway was identified as the most significant pathway in module 1.

Figure 3. Four significant modules identified from the protein-protein interaction network using the molecular complex detection method with a score of >6.0. Module 1: MCODE score=19.238; Module 2: MCODE score=11.333; Module 3: MCODE score=9 and Module 4: MCODE score=6.316.

Table IV. KEGG pathway enriched by differentially expressed genes in different modules (P<0.05).

Table IV.

KEGG pathway enriched by differentially expressed genes in different modules (P<0.05).

Term	Description	P-value	Genes
Module1
hsa03050	Proteasome	6.77×10−17	PSMB4, PSMA1, PSMB6, PSMB1, PSMC2, PSMA4, PSMB3, PSMA3, PSME3, PSME4, PSMD8
hsa04110	Cell cycle	9.03×10−3	ANAPC5, SKP1, CDC16, BUB3
hsa04120	Ubiquitin mediated proteolysis	1.16×10−2	ANAPC5, SKP1, CDC16, UBE2D1
Module2
hsa00510	N-Glycan biosynthesis	1.63×10−3	RPN1, RPN2, DDOST
hsa03060	Protein export	1.10×10−2	SRP68, SRP72
Module3
hsa03040	Spliceosome	5.19×10−6	PRPF8, SNRPD1, SNRPB, TXNL4A, SF3A3
hsa03020	RNA polymerase	3.26×10−2	POLR2H, POLR2B
Module4
hsa00190	Oxidative phosphorylation	1.68×10−5	ATP5J2, ATP5B, ATP5C1, ATP5G1, ATP5H, PPA1
hsa04612	Antigen processing and presentation	1.34×10−3	HSP90AB1, HSP90AA1, IFI30, HLA-DRB5
hsa05012	Parkinson's disease	4.63×10−3	ATP5B, ATP5C1, ATP5G1, ATP5H
hsa05010	Alzheimer's disease	9.07×10−3	ATP5B, ATP5C1, ATP5G1, ATP5H
hsa05016	Huntington's disease	1.19×10−2	ATP5B, ATP5C1, ATP5G1, ATP5H
hsa04514	Cell adhesion molecules	5.00×10−2	NCAM1, ICAM1, HLA-DRB5

[i] Term represents the identification number of the KEGG pathway; description represents the name of the KEGG pathway. KEGG, Kyoto Encyclopedia of Genes and Genomes.

Discussion

In the present study, the gene expression patterns obtained from the GEO database revealed a total of 819 genes, including 759 upregulated and 60 downregulated genes, that were differently expressed in peri-implantitis samples compared with controls. The results of the present study demonstrated that HSP90AA1, which had the highest degrees in the PPI network, was significantly identified in module 4. In addition, NFKB1 were also hub nodes with higher betweenness values in the PPI network. Moreover, the proteasome pathway was the most significant pathway in module 1, and may be key mechanisms associated with peri-implantitis progression.

HSP90, a member of the heat shock family of proteins, is essential in determining cell cycle control and survival, hormone and a number of signaling pathways (24). In addition, released HSP90, which functions as a danger signal, can elicit secretion of inflammatory cytokines (25). A cell-impermeable HSP90 inhibitor can prevent inflammatory responses, indicating that extracellular HSP90 is involved in mediating and initiating sterile inflammatory responses (26). Moreover, previous data indicate that HSP90-targeted agents may be helpful for the treatment of inflammatory disease, and that an HSP90 inhibitor can affect multiple signaling processes associated with inflammation (27). Furthermore, inhibition of HSP90 is able to reduce innate immunity responses and diminishes proinflammatory mediator production in immune-stimulated macrophages (28,29). In addition, analysis of inflammatory mediators in crevicular fluid can be used to distinguish peri-implantitis from normal tissue (30).

Systemic markers of inflammation are increased in patients with peri-implantitis (30) and high levels of inflammatory cytokines, such as IL-1β, are associated with signs of early peri-implantitis development (31–33). Furthermore, it has been suggested that fibroblasts express HSP90 that also participates in the pathogenesis of peri-implantitis (7,34). Moreover, HSPs together with vascular and inflammatory biomarkers may be useful as biomarkers of peri-implantitis development (35,36). In the present study, HSP90AA1, a hub node with higher node degrees in module 4, was enriched in antigen processing and the presentation pathway, as well as the NOD-like receptor-signaling pathway. Therefore, the results of the present study are in line with results from former previous studies and indicate that HSP90AA1 may be directly or indirectly important in peri-implantitis development.

In the present study, NFKB1 was shown to also have hub nodes with higher betweenness values in the PPI network. The NFKB1 gene is known to encode the NF-KB p105/p50 isoforms (37). The central pathological pattern of peri-implantitis is inflammatory osteoclastogenesis, which is mediated by proinflammatory mediators and performed by the regulators of osteoclastogenesis, including NF-B, sRANKL and OPG (38,39). A prior study indicated that the NF-B concentration increased 1.5–4-fold in patients with peri-implantitis compared to those with mucositis (6). Moreover, recent studies demonstrated that high levels of NF-B were associated with peri-implantitis (40,41). In addition, NF-B is upregulated in inflammatory bowel disease and is a transcription regulator of the immune response (42). Zou et al (43) reported that NFKB1 could regulate the transcription of genes in the immune response, and was also a key part in coordinating the immune system. Numerous studies have indicated that the NFKB1-94ins/del ATTG promoter polymorphism is associated with inflammatory disease (37,42,44,45). Furthermore, as mentioned in the aforementioned paragraph, inflammation is associated with the pathogenesis of peri-implantitis. Therefore, the results of the present study are in accordance with those from previous studies and suggest that NFKB1 may be a key gene associated with peri-implantitis development. Notably in the current study, HSP90AA1 and NFKB1 were enriched in the NOD-like receptor-signaling pathway. Although the important roles of the NOD-like receptor-signaling pathway have not been fully discussed, it is speculated that HSP90AA1 and NFKB1 may be involved in peri-implantitis progression via the NOD-like receptor-signaling pathway.

In addition to the two aforementioned genes, the proteasome pathway containing the proteasome subunits (PSM) A1, PSMA3, PSMA4, PSMB1, PSMB3, PSMB4, PSMB6, PSMC2, PSMD8, PSME3 and PSME4, were found to be enriched in the most significant pathway in module 1 and to have the highest P-value. Elliott et al (46) indicated that key proteins modulated by the proteasome were involved in controlling inflammatory processes. A prior study demonstrated that the activation of the ubiquitin-proteasome by macrophages is involved in an NF-B-dependent increase in inflammation (47). Moreover, previous studies have indicated that the proteasomal pathway as a targeting goal may be a potential way to treat autoimmune and inflammatory diseases (48–50). In addition, Wang and Maldonado (51) demonstrated that proteasome inhibitors could serve as novel drugs by regulating proinflammatory protein catabolism. Furthermore, as stated above, inflammation has been associated with the pathogenesis of peri-implantitis. Finally, the PSMA1, PSMA3, PSMA4, PSMB1, PSMB3, PSMB4, PSMB6, PSMC2, PSMD8, PSME3 and PSME4 proteins are essential subunits that contribute to the assembly of the proteasome complex (52–55). Therefore, the proteasome and its subunit genes may be important in peri-implantitis progression.

In conclusion, HSP90AA1, NFKB1 and the NOD-like receptor signaling pathway, as well as the proteasome pathway and its subunits genes may be important in the development of peri-implantitis. However, there were certain limitations in the present study, such as no experimental verification and the relatively small sample size used. Therefore, further research investigating the potential mechanisms involved in peri-implantitis are required.

References