Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Deciphering novel common gene signatures for rheumatoid arthritis and systemic lupus erythematosus by integrative analysis of transcriptomic profiles

  • Neetu Tyagi,

    Roles Conceptualization, Data curation, Formal analysis, Visualization, Writing – original draft

    Affiliations Translational Bioinformatics Group, International Centre for Genetic Engineering and Biotechnology (ICGEB), New Delhi, India, Regional Centre for Biotechnology, Faridabad, India

    ⨯
  • Kusum Mehla,

    Roles Supervision, Visualization, Writing – review & editing

    Affiliation Translational Bioinformatics Group, International Centre for Genetic Engineering and Biotechnology (ICGEB), New Delhi, India

    ⨯
  • Dinesh Gupta

    Roles Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing

    dinesh@icgeb.res.in, dinesh.bioinfo@gmail.com

    Affiliation Translational Bioinformatics Group, International Centre for Genetic Engineering and Biotechnology (ICGEB), New Delhi, India

    ⨯

Abstract

Rheumatoid Arthritis (RA) and Systemic Lupus Erythematosus (SLE) are the two highly prevalent debilitating and sometimes life-threatening systemic inflammatory autoimmune diseases. The etiology and pathogenesis of RA and SLE are interconnected in several ways, with limited knowledge about the underlying molecular mechanisms. With the motivation to better understand shared biological mechanisms and determine novel therapeutic targets, we explored common molecular disease signatures by performing a meta-analysis of publicly available microarray gene expression datasets of RA and SLE. We performed an integrated, multi-cohort analysis of 1088 transcriptomic profiles from 14 independent studies to identify common gene signatures. We identified sixty-two genes common among RA and SLE, out of which fifty-nine genes (21 upregulated and 38 downregulated) had similar expression profiles in the diseases. However, antagonistic expression profiles were observed for ACVR2A, FAM135A, and MAPRE1 genes. Thirty genes common between RA and SLE were proposed as robust gene signatures, with persistent expression in all the studies and cell types. These gene signatures were found to be involved in innate as well as adaptive immune responses, bone development and growth. In conclusion, our analysis of multicohort and multiple microarray datasets would provide the basis for understanding the common mechanisms of pathogenesis and exploring these gene signatures for their diagnostic and therapeutic potential.

Citation: Tyagi N, Mehla K, Gupta D (2023) Deciphering novel common gene signatures for rheumatoid arthritis and systemic lupus erythematosus by integrative analysis of transcriptomic profiles. PLoS ONE 18(3): e0281637. https://doi.org/10.1371/journal.pone.0281637

Editor: Veena Taneja, Mayo Clinic Rochester, UNITED STATES

Received: October 31, 2022; Accepted: January 24, 2023; Published: March 16, 2023

Copyright: © 2023 Tyagi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and its Supporting Information files. In this study, we have used publicly available data from the NCBI Gene Expression Omnibus repository (accession numbers GSE15573, GSE4588, GSE17755, GSE1402, GSE56649, GSE93272, GSE68689, GSE11909, GSE50772, GSE22098, GSE61635, GSE24060).

Funding: This work was supported by the bioinformatics infrastructure grant from the Department of Biotechnology, Government of India (no. BT/PR40151/BTIS/137/5/2021).

Competing interests: On behalf of all authors, the corresponding author states that there is no conflict of interest.

Introduction

Autoimmune diseases are a family of more than 80 chronic, often debilitating, and sometimes life-threatening illnesses; some of which are well characterized such as Rheumatoid Arthritis (RA), Systemic Lupus Erythematosus (SLE), type 1 diabetes, multiple sclerosis, and psoriatic arthritis while some are rare and difficult to diagnose [1]. Epidemiological data provide evidence of a steady increase in autoimmune diseases globally, from an estimated prevalence of 3.2% between 1965 and 1995 to 19.1 ± 43.1 reported in 2018 [2, 3]. In a recent study, the risk of COVID-19 in patients with autoimmune diseases was reported to be significantly higher than in control patients [4].

RA is a multisystem chronic inflammatory disease characterized by erosive synovitis, autoantibody production (rheumatoid factor, RF), polyarticular inflammation of small joints of the hands, wrist, and feet, and associated stiffness and organ damage, leading to severe complications and poor quality life [5]. SLE is another chronic autoimmune disease with various clinical manifestations that affect multiple organs and tissues and involves a complex interaction between various immunological, environmental, hormonal, and genetic factors [6, 7]. Prior clinical and epidemiological studies provided evidence that both RA and SLE have overlapping clinical symptoms and shared genetic architecture [8]. They share certain clinical and pathogenic features, including activation of B and T cells, immune cell (macrophages and neutrophils) migration and infiltration of organs, production of a variety of pathogenic autoantibodies/inflammatory cytokines, and several susceptibility loci [9, 10]. The treatments are very similar for autoimmune disorders, except for cases involving organ damage or where the features of one disease dominate over the other. Thus, elucidating these shared genetic determinants would eventually contribute to identifying biomarkers [11, 12] and developing novel therapeutic strategies for combined diagnosis and prognosis of RA and SLE. The gene expression patterns analysis can provide valuable details for better understanding of molecular mechanisms in the diseases. To gain more rational and decisive results related to different autoimmune diseases, several studies have previously focused on analyzing integrated data from various studies for a single disease [6, 13–17]. Additionally, meta-analysis techniques offer tremendous opportunities to integrate data from different diseases to reveal novel common gene signatures, which may be missed in single disease meta-analysis studies. In the context of RA and SLE, Tuller et al. [18] analyzed the publicly available data from the PBMC samples of six different autoimmune diseases (SLE, multiple sclerosis, RA, juvenile RA, type 1 diabetes, Crohn’s disease and ulcerative colitis). The study aimed to understand the intra-regulatory mechanism in PBMC, which can be common to all autoimmune diseases or specific to any few of them. They found certain chemokines and interleukin genes were differentially expressed in the analyzed autoimmune diseases. Silva et al. [19] integrated the SLE and RA expression datasets and profiling modules for specifically induced or repressed and comodulated genes to uncover the coexpression patterns. Higgs et al. [20] conducted a study to analyze common signatures related to type 1 IFN by integrating data from SLE, myositis, RA and scleroderma. Toro-Domínguez et al. [21] uncovered the common signatures from SLE, RA and SjS (Sjogren’s Syndrome) PBMC patients. They conducted the gene expression meta-analysis using the publicly available gene expression datasets. Wang et al. [22] identified eight differentially expressed genes associated with many rheumatic diseases, including RA, SLE, ankylosing spondylitis, and osteoarthritis. Luan et al. [23] conducted a study integrating microRNA, methylation, and expression datasets to study the shared and specific mechanisms of four autoimmune diseases. In this study, they discovered shared and disease-specific pathways. A recent study by Wang et al. [24] identified the dysregulation of megakaryocyte expansion contributing to the pathogenesis of many autoimmune diseases, including RA and SLE. However, no study to date has focused on systematically identifying the common factors and their role in the underlying mechanism of the two most common systemic autoimmune diseases, RA and SLE. Therefore, the aim of this study is to reveal the commonly dysregulated genes and the significant gene networks associated with the two frequent chronic rheumatic autoimmune diseases.

In this study, we analyzed 1088 publicly available microarray samples of the two diseases belonging to different cell types, ages, sexes, platforms, and genetic backgrounds. To identify the common or specific gene expression signatures in these two diseases, we analyzed the large-scale multi-cohort gene expression microarray datasets of Peripheral Blood Mononuclear Cells (PBMCs), Whole Blood (WB), and other cell type samples obtained from SLE and RA patients. To our knowledge, this is the first large-scale study to report the meta-analysis of gene expression microarray datasets considering the biological and technical heterogeneity observed in the real-world patient population for the two systemic inflammatory diseases. We analyzed the common gene expression patterns, hub genes, commonly regulated important pathways, and regulatory biomarkers involved in the disease mechanism of RA and SLE.

Methods

Data collection

The microarray gene expression data was downloaded from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). The search terms used for the data retrieval include “rheumatoid arthritis” or “RA” and “systemic lupus erythematosus” or “SLE”, each with the filters: organism (Homo sapiens), study type (expression profiling by array) and entry type (dataset/series). As a result of the search, 538 datasets were retrieved. The retrieved datasets were filtered based on the presence of drug-treated samples, missing healthy controls, tissue type, unrelated/duplicated datasets, and summary Area Under the Receiver Operating Characteristic (AUROC) score to obtain a large and independent cohort of 1088 samples of RA and SLE patients. The dataset inclusion/exclusion criteria and detailed workflow of the study are shown in Fig 1. The selected datasets were downloaded from the GEO database using the GEOquery R package [25].

thumbnail
Download:
Fig 1.

Meta-analysis workflow of the study: (a) Details of datasets collection. (b) Data preprocessing and meta-analysis.

https://doi.org/10.1371/journal.pone.0281637.g001

Data preprocessing and meta-analysis

For meta-analysis, downloaded gene expression datasets from various studies were preprocessed using the quantile normalization method and imported into MetaIntegrator framework [26]. The MetaIntegrator-aided meta-analysis combines significance (P) values, Z-scores, ranks, or Effect Size (ES) across different studies and generates formal overall P values for each studied effect. It computes the Hedges g effect size for each gene in each dataset and pools these effect sizes across datasets from different studies.

Where:

J is the Hedges g correction factor; and are the mean expression values; S1 and S0 are the standard deviations; and n1 and n0 are the numbers of samples for case and control, respectively. The summary effect size gs was calculated using a random effect model using the equation given below. n is the no. of studies, gi is the hedges’ g of the gene within dataset i, Wi is the weight calculated by 1/(Vi + T2), Vi is the variance of the gene within a given dataset i, and T2 is the inter-dataset variation estimated by DerSimonian-laired method. MetaIntegrator computes the effect size for each data set independently, thus grabbing heterogeneity and avoiding the limitations of batch effect corrections. The random effect model employed in the above equation would provide more conservative results by extracting fewer Differentially Expressed Genes (DEGs) with more confidence. MetaIntegrator calculates Cochrane’s Q value and a combined p-value using Fisher’s method to account for the heterogeneity of ES estimates between the studies. In the second stage of filtering, we added/removed datasets one-by-one to optimize the AUROC score to generate a model of the pooled datasets with high summary/pooled AUROC scores while keeping in mind to balance the samples and datasets from different tissues. Detailed exclusion criteria followed in this study are given in S1 Table. In the process of data integration, patient samples from different sources were not segregated to reveal the common gene signature in these two autoimmune diseases. After the meta-analysis, a subset of the common DEGs was selected for downstream analysis using the filtering criteria: FDR <0.05, ES >0.40, and observed in at least four studies.

Hub genes and network analysis

To generate the Protein-Protein Interaction (PPI) networks, NetworkAnalyst was used. The networks for the common gene signatures and the RA and SLE-specific top DEGs were generated using the reference innateDB interactome database [27]. The identified common gene signatures and top 50 DEGs from the independent meta-analyses of RA and SLE were used to construct their respective networks for identifying hub genes.

Gene ontology and integrative pathway analysis

To identify over-represented biological terms and enriched pathways, we used the Enrichr R package [28]. The DEGs obtained from the independent meta-analyses of RA and SLE and the common gene signatures revealed in our study were used as input for Enrichr. Default settings were used for the functional annotation and the p-value was calculated using Fisher’s exact test. A significance threshold criterion of p-value <0.05 was used to identify significant gene ontology terms and biological pathways.

Results

Data preprocessing

The gene expression datasets, downloaded from the GEO, were manually checked to exclude duplicate and irrelevant studies. Out of the 538 studies, we selected the studies that reported gene expression in WB, PBMC, or blood cell components. We excluded studies. In the initial filtering step, total 83 studies were filtered out as the studies represented the effect of drug treatment in samples. From the remaining datasets, 29 studies were excluded as they lacked healthy controls. Further, we also removed studies involving other tissues, such as synovial fluid, chondrocytes or lung tissues. Finally, we were left with 38 datasets, representing 14 RA and 24 SLE studies. In the second filtering stage, the datasets that led to a decrease in summary AUROC were also excluded. After filtering, we were left with 14 definitive studies which included seven SLE datasets (GSE11909, GSE50772, GSE22098, GSE4588, GSE61635, GSE17755 and GSE24060) and seven RA datasets (GSE93272, GSE15573, GSE4588, GSE17755, GSE1402, GSE56649, and GSE68689). The resulting 14 datasets were biologically, clinically, and technically heterogeneous, representing five different countries, patients of different ages, different sample types (whole blood and PBMCs), and distinct technologies for gene expression profiling. A total of 1088 samples were used for identifying commonly dysregulated genes ideal for understanding the molecular pathogenesis of RA and SLE.

Meta-analysis and identification of common gene signatures in RA and SLE

We identified and downloaded the publicly available GEO gene expression datasets to achieve an extensive, unbiased study of the common signatures between RA and SLE. From the initially available public datasets, we selected 14 studies that passed the inclusion criteria (see methods). The chosen studies consisted of seven datasets for RA (4 PBMC; 2 WB; 1 CD4 T and B cells) and 7 for SLE (4 PBMC; 2 WB; and 1 CD4 T and B cells), which included 580 samples for RA (415 RA patients and 165 controls) and 508 for SLE (317 SLE patients and 191 controls) as shown in Fig 1. A detailed summary of the included datasets and samples is given in Table 1. From the meta-analysis, we identified 377 significant DEGs for RA (135 upregulated, 242 downregulated) and 1175 for SLE (566 upregulated, 609 downregulated) with the filtering criteria set to ES>0.4, FDR<0.05, number of studies (nstudies> = 4) and AUROC scores. The final selected dataset involving 14 studies given in Table 1, had more discriminatory power, as evidenced by high summary AUROC and was eventually used for predicting the DEGs involved in shared molecular mechanisms of the two diseases. The meta-scores distinguish patient samples from the healthy controls with an AUROC of 0.887 (95% confidence interval (CI): 0.70–1) and 0.927 (95% CI: 0.73–1) for RA and SLE, respectively (Fig 2A and 2B). Precision curves for the RA and SLE datasets are shown in S1 Fig.

thumbnail
Download:
Fig 2. Receiver operating characteristic curves of RA and SLE.

(a) RA datasets and (b) SLE datasets. A perfect classifier must have an AUROC of 1, while a random classifier has an AUC of 0.5. Here, the summary curve is a composite of the individual studies from PBMC, WB, and CD4 T and B Cells samples with AUROC scores of 0.887 and 0.927 for RA and SLE respectively.

https://doi.org/10.1371/journal.pone.0281637.g002

thumbnail
Download:
Table 1. Summary of the included RA and SLE datasets and their samples.

https://doi.org/10.1371/journal.pone.0281637.t001

We identified 62 genes common to both RA and SLE (see S2 Table). Fifty-nine genes (21 upregulated and 38 downregulated) out of the common had similar expression profiles in both the diseases. However, antagonistic expression profiles were observed for the remaining three genes (ACVR2A, FAM135A, and MAPRE1). List of 50 most significant up or downregulated genes for RA and SLE are provided in S3 Table. Of the common genes, 30 were defined as gene signatures for both RA and SLE, as their expression was reported across all the studies. The Venn diagram highlights the unique and common genes of RA and SLE (Fig 3). A heatmap of the 62 common genes between RA and SLE is shown in Fig 4. Heatmaps for highly differentially expressed RA and SLE genes are shown in S2 Fig. The complete description of the common gene signatures persistent across all datasets is given in Table 2.

thumbnail
Download:
Fig 3. Venn diagram of DEGs.

Comparison of DEGs (both upregulated and downregulated) obtained from individual meta-analyses of RA and SLE. The intersection showed the genes common to both diseases.

https://doi.org/10.1371/journal.pone.0281637.g003

thumbnail
Download:
Fig 4. Heatmap represent the effect size of the common differentially expressed gene across all datasets for RA and SLE.

Each column is a dataset and each row represents the expression level of the particular gene in all datasets. The colour scale represents the pooled effect size of that particular gene ranging from yellow (low expression) to red (high expression).

https://doi.org/10.1371/journal.pone.0281637.g004

thumbnail
Download:
Table 2. Detailed description of the common gene signatures persistent across both diseases, and cell types.

https://doi.org/10.1371/journal.pone.0281637.t002

Hub genes network analysis

We have generated three interaction networks as described in the methods section. The interaction network for common genes comprises 53 seeds with 907 connecting nodes and 1143 edges representing the interaction between these proteins. This analysis identified key hub genes among the common genes and the top DEGs specific to both RA and SLE.

The PPI network for the common gene signatures is shown in Fig 5. We analyzed the interaction network for the common gene signatures and found many hub genes based on the high degree of centrality and betweenness. Among these, the main hub genes were CDK1 (degree: 146, betweenness: 112412.8), RPS28 (degree: 94, betweenness: 66300.6), CCNA2 (degree: 77, betweenness: 46749.4), RBL2 (degree: 69, betweenness: 45832.2), EIF4B (degree: 54, betweenness: 44979.1) and MAPRE1 (degree: 42, betweenness: 34895.5).

thumbnail
Download:
Fig 5. Protein-Protein Interaction network of gene signatures common between RA and SLE.

The most highly ranked nodes were CDK1 (degree: 146, betweenness: 112412.8), RPS28 (degree: 94, betweenness: 66300.6), and CCNA2 (degree: 77, betweenness: 46749.4). The size and the colour of the nodes were layout by the degree and betweenness values.

https://doi.org/10.1371/journal.pone.0281637.g005

From the interaction network of top 50 DEGs for RA, the hub genes with the highest centrality degree and betweenness were SMURF2 (degree: 88, betweenness: 48783.5), CCNA2 (degree: 77, betweenness: 41466), and B2M (degree: 75, betweenness: 46273) for the upregulated genes and EWSR1 (degree: 212, betweenness: 178696.9), MAPK3 (degree: 127, betweenness: 107148.4) and G3BP1 (degree: 93, betweenness: 68231.1) for downregulated genes.

For the SLE interaction network, among the top 50 DEGs the notable hub genes included STAT1 (degree: 223, betweenness: 163468.8), ISG15 (degree: 188, betweenness: 156960.3), and PLSCR1 (degree: 84, betweenness: 61314.9) for the upregulated genes and CBL (degree: 216, betweenness: 214523.2), STUB1 (degree: 169, betweenness: 171140.1), MAPK8 (degree: 140, betweenness: 152176.3) for the downregulated genes.

A detailed description of the hub genes for the common gene signatures and disease-specific genes (RA and SLE) is provided in S4 Table. The PPI networks of RA and SLE for the top 50 DEGs are shown in S3 and S4 Figs. Forest plots were created for a few common genes (Fig 6) to represent the consistency of gene expression in both diseases across all datasets.

thumbnail
Download:
Fig 6.

Forest plots of genes with persistent expression in all studies of RA (a) and SLE (b). The x-axis shows the standardized mean difference (log2 scale) computed as Hedges’ g between disease and control samples for genes in multiple studies. The size of the blue box is inversely proportional to the standardized mean difference of the gene in each study. Whiskers represent 95% confidence intervals. The yellow diamond represents the combined mean difference for each gene and its width denotes the 95% confidence interval.

https://doi.org/10.1371/journal.pone.0281637.g006

Identification of over-represented biological pathways and gene ontology terms

The common gene signatures between RA and SLE were enriched with pathways related to TGF-beta signaling, viral carcinogenesis, citrate cycle, and cellular senescence. In RA, the NF-Kappa B signaling pathway, cytokine-cytokine receptor interaction, the IL-17 signaling pathway, and the rheumatoid arthritis pathways were observed for upregulated genes. In contrast, pathways such as the mTOR signaling pathway, the PI3K-Akt signaling pathway, and HIF-1 signaling pathways were observed for downregulated genes. In SLE, upregulated genes were enriched with pathways such as NOD-like receptor signaling, necroptosis, RIG-I-like receptor signaling, toll-like receptor signaling and many viral infections-related signaling pathways. In contrast, the downregulated genes were observed to show enrichment of pathways such as adipocytokine signaling, inflammatory mediator regulation of TRP channels, insulin-resistance and ubiquitin-mediated proteolysis. The top 10 Gene Ontology (GO) terms for each category viz. Molecular Function (MF), Cellular Component (CC), and Biological Process (BP) and enriched pathways for common genes are shown in Fig 7. Detailed information about significant GO terms, enriched pathways and genes involved for each category is provided in S5 Table.

thumbnail
Download:
Fig 7. Enriched GO terms and biological pathways related to common genes (P-value <0.05).

(a) The top 10 GO terms for each category (Molecular Function (MF), Cellular Component (CC), and Biological Processes (BP)) are shown. The X-axis represents the enriched GO categories and the Y-axis shows the gene counts. (b) The top enriched KEGG pathways. The X-axis represents the enriched KEGG pathways and the Y-axis shows the no. of genes present in the respective pathway.

https://doi.org/10.1371/journal.pone.0281637.g007

In enrichment analysis for common genes, BP terms such as tricarboxylic acid metabolic process, regulation of translational initiation, innate immune response in mucosa, mitotic chromosome condensation mucosal immune response, and neutrophil degranulation were observed (see Fig 7(A)).

Discussion

The etiology and pathogenesis of RA and SLE involve different types of cells such as macrophages, T and B cells, fibroblasts, and dendritic cells, in addition to various signaling pathways and immune modulators, which make it challenging to understand the underlying mechanism for the two diseases. The present study aimed to elucidate robust common, RA and SLE-specific gene signatures by integrating gene expression data from multiple heterogeneous sources leveraging the biological (samples from different cell types) and technical heterogeneity (data generated using diverse microarray platforms). To minimize the impact caused by differences in study design and platform usage among different datasets, MetaIntegrator calculated the combined effect size by applying a random effect model. It achieves more consistent and accurate results by considering the direction and magnitude of gene expression changes.

MetaIntegrator has been successfully applied to study various diseases, from cancer to many autoimmune diseases and a few of these study outcomes have been validated in clinical settings [29–36]. Using MetaIntegrator, we analyzed 14 datasets consisting of 1,088 samples that were collected from 5 countries, 9 research centres and represented different cell types such as WB, PBMCs, and CD4 T and B immune cells to identify gene signatures which are robust and consistently differentially expressed across all studies.

This is the first study to perform a combined analysis of RA and SLE in large heterogeneous data, revealing common gene signatures systemically expressed across different cell types. Our study would find potential applications in understanding the underlying disease mechanism and exploring new biological pathways and possible drug targets for further study, which will eventually improve the understanding and management of these diseases.

The role of neutrophils in the pathogenesis of the systemic autoimmune diseases appeared as an important regulator in innate and adaptive immune responses. Neutrophils act as phagocytic cells and their role has been intensively explored in defining the pathogenesis of RA and SLE [37–43]. We identified genes viz. TNFAIP6 (Tumor necrosis factor-inducible gene 6 protein), ANXA3 (Annexin A3), DEFA4 (Defensin Alpha 4), and CAMP (Cathelicidin Antimicrobial Peptide) as upregulated and IMPDH2 (Inosine Monophosphate Dehydrogenase 2), ALDOC (Aldolase, Fructose-Bisphosphate C) as downregulated which are related to neutrophil-mediated immunity, activation, and degranulation. TNFAIP6, which plays a critical role in osteogenesis and bone remodeling, has previously been explored to be up-regulated in the synovial fluid of patients with rheumatoid arthritis [44]. The Defensin Alpha4 gene (DEFA4) is a member of the alpha-defensin family, a part of antimicrobial peptides in the innate immune system. Variations in DEFA4 gene expression have been reported in different disorders such as diseases related to inflammation and immunity dysfunction, brain-related disorders, and various cancers [45].

Cytokines are the main modulators of immunity. We observed YTHDF2 (YTH N6-Methyladenosine RNA Binding Protein 2) and GPS2 (G Protein Pathway Suppressor 2) in our common gene signatures negatively regulate cytokine-mediated signaling pathway, which in turn regulates the expression of Polymorphonuclear neutrophils (PMNs) and plays an important role in host defense response and inflammation. Natural Killer (NK) cells are important cells of innate immunity and their role has already been explored in the pathogenesis and etiology of various autoimmune diseases [46]. We observed that NFIL3 (Nuclear Factor, Interleukin 3 Regulated), a key immunological transcription factor that is an essential component in developing precursor NK cells, was upregulated in our study for both diseases. Interferons are a category of functionally related cytokines implicated in the pathogenesis of several rheumatic diseases. Type 1 interferon pathway has been reported to be associated with increased inflammatory response in various rheumatic conditions in response to increased expression of type 1 Interferon Stimulated Genes (ISGs) [47]. In SLE, we found many interferon related genes such as IFI27 (Interferon alpha-inducible protein 27), IFI16 (Gamma-interferon-inducible protein 16), IFI27L1 (Interferon Alpha Inducible Protein 27 like 1), IFNAR1 (Interferon alpha/beta receptor 1), IFI6 (Interferon Alpha Inducible Protein 6), IFI44 (Interferon Induced Protein 44), IFIT1-3, 5 (Interferon Induced Protein with Tetratricopeptide Repeats) which were all upregulated. Additionally, Interferon Response Factors (IRF) such as IRF7 and IRF9 which coordinate type 1 interferon and ISGs expression were upregulated in SLE. However, in RA we observed normal expression levels for genes related to interferon. Reduced relative expression of ISGs in the circulation of RA patients as compared to SLE has already been reported [20, 48]. Even in SLE, Niewold et al. reported a wide range of serum interferon activity with 40–50% of SLE patients showing normal levels of serum interferons [49]. Therefore, status of Type 1 interferon signature as a predictive biomarker in various autoimmune conditions is debatable as it remains relatively stable in blood. The Type 1 interferon signature could play an important role in disease initiation rather than in predicting disease flares where other non-Type 1 interferon genes are reported to strongly correlate with disease activity [50].

Cell division in multicellular organisms is critical to developing and maintaining tissue homeostasis. Deregulation of cell functions leads to loss of tolerance and the development of autoimmunity [51]. Many cell cycle regulators, including cyclin-dependent kinase (CDK) and cyclins, are known for their crucial role in cell division [52]. In the common gene signatures, we identified CDK1 (Cyclin Dependent Kinase 1), CCNA2 (Cyclin A2) and MAPRE1(Microtubule Associated Protein RP/EB Family Member 1) genes that have an important role in cell division.

Bone mineralization is essential for the hardness and strength of the bone. Bone is the target tissue in inflammatory diseases, including rheumatic diseases such as RA, SLE, psoriatic arthritis and ankylosing spondylitis [53]. As bone loss has been found in both diseases, the regulation process of bone mineralization is important [54]. We found SRGN (Serglycin), known for negative regulation of bone mineralization, to be upregulated in both RA and SLE.

Ubiquitination is a key regulatory process that controls innate and adaptive immune responses. It is involved in the development, activation and differentiation of T-cells and B-cells, thus maintaining the efficient adaptive immune responses to pathogens and immunological tolerance to the self-tissues [55, 56]. In our study, we observed a negative regulator of protein polyubiquitination, GPS2, which could disrupt many aspects of immune functions and different intracellular signaling pathways.

The most striking observation was the antagonistic gene expression profiles for three genes, i.e. MAPRE1, ACVR2A (Activin a Receptor Type 2A), and FAM135A (Family with Sequence Similarity 135 Member A) in RA and SLE. MAPRE1 is an important gene believed to be involved in regulating microtubule structure and chromosome stability. Microtubules are important as they play an important role in maintaining cell structure [57] along with their recently identified roles in the innate and adaptive immune systems [58]. The PPI network for the common genes identified MAPRE1 as the hub gene. Hub genes produce proteins that can interact with many other proteins [59]. Hub genes play an important role in the pathogenesis and progression of many diseases; therefore, they can be targeted as diagnostic markers and candidate drug targets.

MAPRE1 was found to be upregulated in SLE whereas downregulated in RA. ACVR2A, involved in the TGF-beta signaling pathway, was also predicted to be the hub gene. TGF-beta signaling pathway plays a crucial role in immune regulation, tissue regeneration and many components of the immune system [60–63]. Malfunctioning of the TGF-beta signaling pathway can lead to immune dysregulation and other congenital effects. ACVR2A expression was elevated in RA, while it was repressed in SLE, and FAM135A followed a similar trend.

Contrary to our findings, ACVR2A expression was reported to be elevated in rheumatic diseases; however, conclusions were drawn from a small sample size of 60 patients [64]. This warrants further research into the role of these genes in the pathophysiology of autoimmune diseases. These discrepancies further emphasize the significance of using a rigorous integrated multi-cohort analysis approach. We created forest plots for ACVR2A, DEFA4, MAPRE1, TNFAIP6, and NFIL3 genes to represent the persistent gene expression patterns across all datasets of RA and SLE. However, it is apparent that some minor deviations existed for some of the datasets, which can be further validated via inclusion of more datasets.

This study has some limitations, as it relies on publicly available datasets, thus incorporating the inherent limitations of the experimental procedures and computational methods used for data analysis. For some cell types, the sample sizes were limited, making it hard to balance the samples. The gene signature set includes too many genes to be included in a simple diagnostic test. An accurate signature based on a small set of genes would be cost-effective and more technically feasible for diagnostic purposes. The performance of the identified gene signatures in a large, prospective cohort remains unknown and requires validation on larger sample datasets further to ensure the applicability of our findings in clinical settings.

Conclusions

With limited knowledge available about the etiology of RA and SLE, it becomes imperative to understand the precise molecular mechanisms underlying the pathophysiology of these autoimmune diseases. Many common DEGs such as TNFAIP6, DEFA4, YTHDF2, NFIL3, and SRGN predicted in our meta-analysis study have already been validated to potentially participate in the development and progression of both the diseases, which further strengthens the credibility of our results. Our study explored the novel common molecular mechanisms underlying the disease pathogenesis, and the predicted genes have the potential to be utilized as diagnostic and therapeutic targets when validated in a large prospective cohort.

Supporting information

S1 Table. Details of inclusion/exclusion criteria used in the study.

https://doi.org/10.1371/journal.pone.0281637.s001

(XLSX)

S2 Table. Expression details of the 62 common gene signatures in RA and SLE.

https://doi.org/10.1371/journal.pone.0281637.s002

(XLSX)

S3 Table. A list of 50 most significantly up or downregulated genes for RA and SLE.

https://doi.org/10.1371/journal.pone.0281637.s003

(DOCX)

S4 Table. Detailed summary of the hub genes for the common and disease-specific gene signatures.

https://doi.org/10.1371/journal.pone.0281637.s004

(XLSX)

S5 Table. The GO-term and enriched pathway details for common and disease-specific (RA and SLE) gene signatures.

https://doi.org/10.1371/journal.pone.0281637.s005

(XLSX)

S1 Fig.

Precision recall Curves for RA (a) and SLE (b). The average precision ranges from the frequency of positive examples ranging from 0.5 (for balanced data) to 1.0 (perfect model). Here, the precision-recall curves represent individual studies from PBMC, WB, and CD4 T and B cell samples.

https://doi.org/10.1371/journal.pone.0281637.s006

(TIF)

S2 Fig.

Heatmaps represent the effect size of differentially expressed gene signatures across all datasets (a) top RA DEGs and (b) top SLE DEGs. (Filtering criteria: Effect size > = 0.4 and FDR < = 0.05). Each column is a dataset and each row represents the expression level of the particular gene in all datasets. The colour scale represents the pooled effect size of that particular gene ranging from yellow (low expression) to red (high expression).

https://doi.org/10.1371/journal.pone.0281637.s007

(TIF)

S3 Fig. Protein-Protein Interaction networks of DEGs for RA.

(a) Upregulated RA genes, (b) Downregulated RA genes. The size and the colour of the nodes are layout by the degree and betweenness values.

https://doi.org/10.1371/journal.pone.0281637.s008

(TIF)

S4 Fig. Protein-protein interaction networks of DEGs for SLE.

(a) Upregulated SLE genes, and (b) Downregulated SLE genes. The size and the colour of the nodes are layout by the degree and betweenness values.

https://doi.org/10.1371/journal.pone.0281637.s009

(TIF)

Acknowledgments

We acknowledge ICGEB for providing the necessary infrastructure and facilities for the research. Senior Research Fellowship awarded to NT by GlaxoSmithKline (GSK, India), is duly acknowledged.

References

  1. 1. Karopka T, Fluck J, Mevissen HT, Glass Ä. The Autoimmune disease database: A dynamically compiled literature-derived database. BMC Bioinformatics. 2006;7: 325. pmid:16803617
  2. 2. Jacobson DL, Gange SJ, Rose NR, Graham NMH. Epidemiology and Estimated Population Burden of Selected Autoimmune Diseases in the United States. Clin Immunol Immunopathol. 1997;84: 223–243. pmid:9281381
  3. 3. Lerner A, Jeremias P, Matthias T. The World Incidence and Prevalence of Autoimmune Diseases is Increasing. Int J Celiac Dis. 2015;3: 151–155.
  4. 4. Akiyama S, Hamdeh S, Micic D, Sakuraba A. Prevalence and clinical outcomes of COVID-19 in patients with autoimmune diseases: a systematic review and meta-analysis. Ann Rheum Dis. 2021;80: 384–391. pmid:33051220
  5. 5. Mitchell DM, Spitz PW, Young DY, Bloch DA, McShane DJ, Fries JF. Survival, prognosis, and causes of death in rheumatoid arthritis. Arthritis Rheum. 1986;29: 706–714. pmid:3718563
  6. 6. Haynes WA, Haddon DJ, Diep VK, Khatri A, Bongen E, Yiu G, et al. Integrated, multicohort analysis reveals unified signature of systemic lupus erythematosus. JCI Insight. 2020;5: e122312. pmid:31971918
  7. 7. Pabón-Porras MA, Molina-Ríos S, Flórez-Suárez JB, Coral-Alvarado PX, Méndez-Patarroyo P, Quintana-López G. Rheumatoid arthritis and systemic lupus erythematosus: Pathophysiological mechanisms related to innate immune system. SAGE Open Med. 2019;7: 1–24. pmid:35154753
  8. 8. Lu H, Zhang J, Jiang Z, Zhang M, Wang T, Zhao H, et al. Detection of Genetic Overlap Between Rheumatoid Arthritis and Systemic Lupus Erythematosus Using GWAS Summary Statistics. Front Genet. 2021;12: 1–10. pmid:33815486
  9. 9. Tang S, Lui SL, Lai KN. Pathogenesis of lupus nephritis: An update. Nephrology. 2005. pp. 174–179. pmid:15877678
  10. 10. Lynn AH, Kwoh CK, Venglish CM, Aston CE, Chakravarti A. Genetic epidemiology of rheumatoid arthritis. Am J Hum Genet. 1995;57: 150–159. Available: pmc/articles/PMC1801237/?report=abstract pmid:7611283
  11. 11. Huang S, Xiang C, Song Y. Identification of the shared gene signatures and pathways between sarcopenia and type 2 diabetes mellitus. PLoS One. 2022;17: 1–15. pmid:35271662
  12. 12. Miao C, Chen Y, Fang X, Zhao Y, Wang R, Zhang Q. Identification of the shared gene signatures and pathways between polycystic ovary syndrome and endometrial cancer: An omics data based combined approach. PLoS One. 2022;17: 1–16. pmid:35830453
  13. 13. Song GG, Kim JH, Seo YH, Choi SJ, Ji JD, Lee YH. Meta-analysis of differentially expressed genes in primary Sjogren’s syndrome by using microarray. Hum Immunol. 2014;75: 98–104. pmid:24090683
  14. 14. Arasappan D, Tong W, Mummaneni P, Fang H, Amur S. Meta-analysis of microarray data using a pathway-based approach identifies a 37-gene expression signature for systemic lupus erythematosus in human peripheral blood mononuclear cells. BMC Med. 2011;9: 65. pmid:21624134
  15. 15. Olsen NJ, Sokka T, Seehorn CL, Kraft B, Maas K, Moore J, et al. A gene expression signature for recent onset rheumatoid arthritis in peripheral blood mononuclear cells. Ann Rheum Dis. 2004;63: 1387–1392. pmid:15479887
  16. 16. Afroz S, Giddaluru J, Vishwakarma S, Naz S, Khan AA, Khan N. A comprehensive gene expression meta-analysis identifies novel immune signatures in rheumatoid arthritis patients. Front Immunol. 2017;8: 74. pmid:28210261
  17. 17. Kröger W, Mapiye D, Entfellner JBD, Tiffin N. A meta-analysis of public microarray data identifies gene regulatory pathways deregulated in peripheral blood mononuclear cells from individuals with Systemic Lupus Erythematosus compared to those without. BMC Med Genomics. 2016;9: 1–11. pmid:27846842
  18. 18. Tuller T, Atar S, Ruppin E, Gurevich M, Achiron A. Common and specific signatures of gene expression and protein&ndash;protein interactions in autoimmune diseases. Genes Immun. 2013;14: 67–82. pmid:23190644
  19. 19. Silva GL, Junta CM, Mello SS, Garcia PS, Rassi DM, Sakamoto-Hojo ET, et al. Profiling meta-analysis reveals primarily gene coexpression concordance between systemic lupus erythematosus and rheumatoid arthritis. Ann N Y Acad Sci. 2007;1110: 33–46. pmid:17911418
  20. 20. Higgs BW, Liu Z, White B, Zhu W, White WI, Morehouse C, et al. Patients with systemic lupus erythematosus, myositis, rheumatoid arthritis and scleroderma share activation of a common type I interferon pathway. Ann Rheum Dis. 2011;70: 2029–2036. pmid:21803750
  21. 21. Toro-Domínguez D, Carmona-Sáez P, Alarcón-Riquelme ME. Shared signatures between rheumatoid arthritis, systemic lupus erythematosus and Sjögren’s syndrome uncovered through gene expression meta-analysis. Arthritis Res Ther. 2014;16: 1–8. pmid:25466291
  22. 22. Wang L, Wu LF, Lu X, Mo XB, Tang ZX, Lei SF, et al. Integrated analyses of gene expression profiles digs out common markers for rheumatic diseases. PLoS One. 2015;10: 1–11. pmid:26352601
  23. 23. Luan M, Shang Z, Teng Y, Chen X, Zhang M, Lv H, et al. The shared and specific mechanism of four autoimmune diseases. Oncotarget. 2017;8: 108355–108374. pmid:29312536
  24. 24. Wang Y, Xie X, Zhang C, Su M, Gao S, Wang J, et al. Rheumatoid arthritis, systemic lupus erythematosus and primary Sjögren ‘ s syndrome shared megakaryocyte expansion in peripheral blood. Ann Rheum Dis. 2022;81: 379–385. pmid:34462261
  25. 25. Sean D, Meltzer PS. GEOquery: A bridge between the Gene Expression Omnibus (GEO) and BioConductor. Bioinformatics. 2007;23: 1846–1847. pmid:17496320
  26. 26. Haynes WA, Vallania F, Liu C, Bongen E, Tomczak A, Andres-Terrè M, et al. Empowering Multi-Cohort Gene Expression Analysis to Increase Reproducibility Empowering Multi-Cohort Gene Expression Analysis to Increase Reproducibility. Pac Symp Biocomput. 2017;22: 144–153. pmid:27896970
  27. 27. Zhou G, Soufan O, Ewald J, Hancock REW, Basu N, Xia J. NetworkAnalyst 3.0: a visual analytics platform for comprehensive gene expression profiling and meta-analysis. Web Serv issue Publ online. 2019;47: W234–W241. pmid:30931480
  28. 28. Kuleshov M V, Jones MR, Rouillard AD, Fernandez NF, Duan Q, Wang Z, et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 2016;44: (W1):W90–7. pmid:27141961
  29. 29. Khatri P, Roedder S, Kimura N, De Vusser K, Morgan AA, Gong Y, et al. A common rejection module (CRM) for acute rejection across multiple organs identifies novel therapeutics for organ transplantation. J Exp Med. 2013;210: 2205–2221. pmid:24127489
  30. 30. Mazur PK, Reynoird N, Khatri P, Jansen PWTC, Wilkinson AW, Liu S, et al. SMYD3 links lysine methylation of MAP3K2 to Ras-driven cancer. Nature. 2014;510: 283–7. pmid:24847881
  31. 31. Chen R, Khatri P, Mazur PK, Polin M, Zheng Y, Vaka D, et al. A meta-Analysis of lung cancer gene expression identifies PTK7 as a survival gene in lung adenocarcinoma. Cancer Res. 2014;74: 2892–2902. pmid:24654231
  32. 32. Sweeney TE, Shidham A, Wong HR, Khatri P. A comprehensive time-course-based multicohort analysis of sepsis and sterile inflammation reveals a robust diagnostic gene set. Sci Transl Med. 2015;7: 1–16. pmid:25972003
  33. 33. Andres-terre M, Mcguire HM, Pouliot Y, Bongen E, Sweeney TE, Tato CM, et al. Integrated, Multi-cohort Analysis Identifies Conserved Transcriptional Signatures across Resource Integrated, Multi-cohort Analysis Identifies Conserved Transcriptional Signatures across Multiple Respiratory Viruses. Immunity. 2015;43: 1199–1211. pmid:26682989
  34. 34. Sweeney TE, Braviak L, Tato CM, Khatri P, Khatri and P. Genome-wide expression for diagnosis of pulmonary tuberculosis: a multicohort analysis. Lancet Respir Med. 2016;4: 213–224. pmid:26907218
  35. 35. Avey S, Cheung F, Fermin D, Frelinger J, Gaujoux R, Gottardo R, et al. Multicohort analysis reveals baseline transcriptional predictors of influenza vaccination responses. Sci Immunol. 2017;2: eaal4656. pmid:28842433
  36. 36. Li MD, Burns TC, Morgan AA, Khatri P. Integrated multi-cohort transcriptional meta-analysis of neurodegenerative diseases. Acta Neuropathol Commun. 2014;2: 1–23. pmid:25187168
  37. 37. Teng TS, Ji AL, Ji XY, Li YZ. Neutrophils and immunity: From bactericidal action to being conquered. J Immunol Res. 2017;2017: 9671604. pmid:28299345
  38. 38. Fresneda Alarcon M, McLaren Z, Wright HL. Neutrophils in the Pathogenesis of Rheumatoid Arthritis and Systemic Lupus Erythematosus: Same Foe Different M.O. Front Immunol. 2021;12. pmid:33746988
  39. 39. Kaplan MJ. Role of neutrophils in systemic autoimmune diseases. Arthritis Res Ther. 2013;15: 219. pmid:24286137
  40. 40. Fu X, Liu H, Huang G, Dai SS. The emerging role of neutrophils in autoimmune-associated disorders: effector, predictor, and therapeutic targets. MedComm. 2021;2: 402–413. pmid:34766153
  41. 41. Smith CK, Kaplan MJ. The role of neutrophils in the pathogenesis of systemic lupus erythematosus. Curr Opin Rheumatol. 2015;27: 448–453. pmid:26125102
  42. 42. Kim JW, Ahn MH, Jung JY, Suh CH, Kim HA. An update on the pathogenic role of neutrophils in systemic juvenile idiopathic arthritis and adult-onset still’s disease. Int J Mol Sci. 2021;22: 13038. pmid:34884842
  43. 43. Zhao Y, Marion TN, Wang Q. Multifaceted Roles of Neutrophils in Autoimmune Diseases. J Immunol Res. 2019;2019: 7896738. pmid:31016207
  44. 44. Tsukahara S, Ikeda R, Goto S, Yoshida K, Mitsumori R, Sakamoto Y, et al. Tumour necrosis factor α-stimulated gene-6 inhibits osteoblastic differentiation of human mesenchymal stem cells induced by osteogenic differentiation medium and BMP-2. Biochem J. 2006;398: 595–603. pmid:16771708
  45. 45. Basingab F, Alsaiary A, Almontashri S, Alrofaidi A, Alharbi M, Azhari S, et al. Alterations in Immune-Related Defensin Alpha 4 (DEFA4) Gene Expression in Health and Disease. Int J Inflam. 2022;2022: 9099136. pmid:35668817
  46. 46. Liu M, Liang S, Zhang C. NK Cells in Autoimmune Diseases: Protective or Pathogenic? Front Immunol. 2021;12: 624687. pmid:33777006
  47. 47. Muskardin TLW, Niewold TB. Type i interferon in rheumatic diseases. Nat Rev Rheumatol. 2018;14: 214–228. pmid:29559718
  48. 48. Hua J, Kirou K, Lee C, Crow MK. Functional assay of type I interferon in systemic lupus erythematosus plasma and association with anti-RNA binding protein autoantibodies. Arthritis Rheum. 2006;54: 1906–1916. pmid:16736505
  49. 49. Niewold TB, Hua J, Lehman TJA, Harley JB and CM. High serum IFN-α activity is a heritable risk factor for systemic lupus erythematosus. Genes Immun. 2012;23: 1–7.
  50. 50. Banchereau R, Hong S, Cantarel B, Baldwin N, Baisch J, Edens M, et al. Personalized immunomonitoring uncovers molecular networks that stratify lupus patients. Cell. 2016;165: 1548–1550. pmid:27259156
  51. 51. Balomenos D. Cell Cycle Regulation and Systemic Lupus Erythematosus. Systemic Lupus Erythematosus. Academic Press; 2011.
  52. 52. Liu S-T, Burgess A, Dick F, Jirawatnotai S, Laphanuwat P. Immunomodulatory Roles of Cell Cycle Regulators. Immunomodulatory Roles Cell Cycle Regul Front Cell Dev Biol. 2019;7: 23. pmid:30863749
  53. 53. Almoallim H, Cheikh M. Skills in Rheumatology. Skills in Rheumatology. 2021.
  54. 54. Lin FR, Niparko JK, Ferrucci and L. T-cells and B-cells in osteoporosis. Bone. 2014;23: 1–7.
  55. 55. Zinngrebe J, Montinaro A, Peltzer N, Walczak H. “Ubiquitylation: mechanism and functions” Review series Ubiquitin in the immune system. EMBO Rep. 2014;15: 142–154. pmid:24375678
  56. 56. Hu H, Sun SC. Ubiquitin signaling in immune responses. Cell Res. 2016;26: 457–483. pmid:27012466
  57. 57. Ilan Y. Microtubules: From understanding their dynamics to using them as potential therapeutic targets. J Cell Physiol. 2019;234: 7923–7937. pmid:30536951
  58. 58. Ilan-Ber T, Ilan Y. The role of microtubules in the immune system and as potential targets for gut-based immunotherapy. Mol Immunol. 2019;111: 73–82. pmid:31035111
  59. 59. Tsai C-J, Ma B, Nussinov R. Protein-protein interaction networks: how can a hub protein bind so many different partners? HHS Public Access. Trends Biochem Sci. 2009;34: 594–600. pmid:19837592
  60. 60. Batlle E, Massagué J. Transforming Growth Factor-β Signaling in Immunity and Cancer. Immunity. 2019;50: 924–940. pmid:30995507
  61. 61. Okeke EB, Uzonna JE. The pivotal role of regulatory T cells in the regulation of innate immune cells. Front Immunol. 2019;10: 1–12. pmid:31024539
  62. 62. Sanjabi S, Oh SA, Li MO. Regulation of the immune response by TGF-β: From conception to autoimmunity and infection. Cold Spring Harb Perspect Biol. 2017;9: 1–33. pmid:28108486
  63. 63. Gonzalo-Gil E, Galindo-Izquierdo M. Role of Transforming Growth Factor-Beta (TGF) Beta in the Physiopathology of Rheumatoid Arthritis. Reumatol Clínica (English Ed. 2014;10: 174–179. pmid:24685296
  64. 64. El-Gendi SS, Moniem AEA, Tawfik NM, Ashmawy MM, Mohammed OA, Mostafa AK, et al. Value of serum and synovial fluid activin A and inhibin A in some rheumatic diseases. Int J Rheum Dis. 2010;13: 273–279. pmid:20704626