Abstract

Background

   Osteoclast differentiation in the inflamed synovium of rheumatoid
   arthritis (RA) affected joints leads to the formation of bone lesions.
   Reconstruction and analysis of protein interaction networks underlying
   specific disease phenotypes are essential for designing therapeutic
   interventions. In this study, we have created a network that captures
   signal flow leading to osteoclast differentiation. Based on
   transcriptome analysis, we have indicated the potential mechanisms
   responsible for the phenotype in the RA affected synovium.

Method

   We collected information on gene expression, pathways and protein
   interactions related to RA from literature and databases namely Gene
   Expression Omnibus, Kyoto Encyclopedia of Genes and Genomes pathway and
   STRING. Based on these information, we created a network for the
   differentiation of osteoclasts. We identified the differentially
   regulated network genes and reported the signaling that are responsible
   for the process in the RA affected synovium.

Result

   Our network reveals the mechanisms underlying the activation of the
   neutrophil cytosolic factor complex in connection to osteoclastogenesis
   in RA. Additionally, the study reports the predominance of the
   canonical pathway of NF-κB activation in the diseased synovium. The
   network also confirms that the upregulation of T cell receptor
   signaling and downregulation of transforming growth factor beta
   signaling pathway favor osteoclastogenesis in RA. To the best of our
   knowledge, this is the first comprehensive protein–protein interaction
   network describing RA driven osteoclastogenesis in the synovium.

Discussion

   This study provides information that can be used to build models of the
   signal flow involved in the process of osteoclast differentiation. The
   models can further be used to design therapies to ameliorate bone
   destruction in the RA affected joints.

   Keywords: Synovium, Rheumatoid arthritis, Interaction network, Gene
   expression, Enriched pathways, Signaling pathways, Osteoclastogenesis,
   Osteoclast differentiation

Introduction

   Rheumatoid arthritis (RA) is a systemic autoimmune disease that
   primarily affects synovial joints. The disease is characterized by
   chronic inflammation in the joints, leading to synovial hyperplasia
   (pannus formation), destruction of the cartilage and erosion of the
   underlying bone. RA is a complex disease involving several molecular
   pathways across various cell types and tissues. Thus in order to
   elucidate the underlying cause of a particular phenotype associated to
   the disease, identification of the network consisting of differentially
   expressed genes (DEGs) in the interacting pathways is essential.
   Studies have used pathway analysis to identify affected pathways from
   lists of DEGs ([28]Hao et al., 2017; [29]Wang et al., 2017; [30]Lee et
   al., 2011; [31]Wu et al., 2010). The lists have also been used to
   create networks that are related to specific diseases or conditions.
   Earlier work using RA samples has focused on generating networks of the
   genes showing differential regulation ([32]Hao et al., 2017; [33]Wang
   et al., 2017) or the most enriched gene ontology (GO) ([34]The Gene
   Ontology Consortium, 2017) category in the DEG lists ([35]Lee et al.,
   2011). A comprehensive network describing molecular interactions across
   various RA affected tissues was created using publicly available
   microarray data by [36]Wu et al. (2010). Other groups have created gene
   regulatory networks (GRNs) using in vitro data from cultured
   fibroblasts and macrophages ([37]Kupfer et al., 2014; [38]You et al.,
   2014). [39]Kupfer et al. (2014) used time series data generated from RA
   synovial fibroblasts subjected to external stimulation to create a GRN.
   They simulated the network to analyze the behavior of genes involved in
   RA pathogenesis, in response to stimulation by RA associated cytokines
   and growth factors (GFs). [40]You et al. (2014) created a GRN and
   identified the critical interactions responsible for synovial
   fibroblast invasiveness in RA synovium. The creation of a detailed
   protein–protein interaction (PPI) network describing the connections
   between various pathways involved in any specific RA process, at the
   level of the synovial tissue, is yet to be attempted. In this study,
   using the publicly available gene expression data for RA synovial
   tissue and protein interactions and pathway databases, we created and
   analyzed a detailed phenotype-specific PPI network. We used
   differentially regulated genes to identify the altered pathways in the
   affected synovium. We identified the pathway of osteoclast
   differentiation as a phenotype connected to many of the altered
   pathways in the RA synovium. It is established that the RA synovium
   harbors osteoclasts, the cells responsible for bone degradation in the
   affected joints ([41]Schett, 2007). Therefore, a network of proteins
   participating in the interacting pathways underlying the RA associated
   process of osteoclast differentiation in the synovium was created for
   the first time. We report the upregulated signaling routes that drive
   osteoclastogenesis via the generation of reactive oxygen species (ROS)
   by neutrophil cytosolic factor (NCF) complex in the RA synovium. We
   demonstrate the contribution of elevated T cell receptor signaling in
   facilitating osteoclast differentiation in the affected tissue. In
   addition, we describe the importance of the canonical pathway of NF-κB
   activation and the transforming growth factor beta (TGFβ) pathway in
   connection to the process. Finally, the network reports all the
   possible routes by which the inflamed synovium promotes the
   differentiation of osteoclasts.

Materials and Methods

   This study involved two major steps: selection of a phenotype exhibited
   by the RA synovium, and construction and analysis of a PPI network for
   the selected phenotype. [42]Figure 1 shows the detailed workflow that
   was followed. Each step is described in detail in this section. The
   databases used in this study are summarized in [43]Table 1.

Figure 1. The workflow.

   [44]Figure 1
   [45]Open in a new tab

   (A) Selection of a KEGG pathway representing a phenotype exhibited by
   the RA synovium. The differentially expressed genes in the RA synovium
   were identified from the analysis of the microarray datasets obtained
   from GEO database. A KEGG pathway enrichment analysis was performed on
   the differentially expressed gene lists. The enriched pathways were
   categorized into process and signaling pathways. Based on the shared
   differentially expressed genes, a pathway overlap network was created.
   Osteoclast differentiation pathway was selected as the pathway of
   interest as it overlapped with most number of signaling pathways. (B)
   Construction and analysis of the osteoclast differentiation network.
   The proteins belonging to the KEGG osteoclast differentiation pathway
   were termed as “core proteins.” The core proteins were used as an input
   to the String DB to obtain the proteins interacting with them (first
   shell proteins). The interactions among all the proteins (core and
   first shell) were also extracted from String DB. The interactions were
   validated using PubMed. The directions of the protein interactions were
   also obtained. The proteins and the interactions were used to construct
   a network for osteoclast differentiation. The gene ontology term
   enrichment analysis was performed on the network. Differentially
   regulated genes were indicated in the network. The gene ontology term
   enrichment analysis and the differentially regulated genes were used to
   identify the important protein interactions that lead to osteoclast
   differentiation in the RA synovium. The protein targets of the drugs
   used in RA treatment were obtained from DrugBank database and were
   indicated in the network.

Table 1. The details of the databases used in this study.

   Database Type of data obtained for this study Features Rationale
   GEO ([46]Edgar, Domrachev & Lash, 2002) Microarray gene expression data
   GEO is a public repository with easy access to high throughput data,
   including microarray data and related metadata such as tissue type,
   disease state, etc. Microarray data from GEO database was used to
   identify DEGs in the RA synovium
   DAVID—Gene ID conversion tool ([47]Huang, Sherman & Lempicki, 2009b,
   [48]2009a) Gene ID types The DAVID knowledge base supports conversion
   between more than 20 gene ID types, including Affymetrix probe IDs The
   DAVID gene ID conversion table was used to convert Affymetrix probe IDs
   to Entrez IDs
   KEGG pathway ([49]Kanehisa et al., 2017) Molecular pathways KEGG
   pathways are manually drawn and frequently updated. References are