Introduction
Rheumatoid arthritis (RA) is a systemic autoimmune disease characterized by chronic synovitis that is often pathogenic and destructive to articular cartilage and bone. To understand the complex pathogenesis and heterogeneous manifestations of autoimmune diseases including RA, DNA microarray has emerged as a powerful tool [
1‐
4]. We have shown in studies investigating the pathogenesis of juvenile idiopathic arthritis (JIA) and systemic lupus erythematosus (SLE) that DNA microarray can be even more effective when combined with bioinformatics techniques such as gene ontology databases and network pathway analysis software [
5,
6].
In RA pathology, fibroblast-like synoviocyte (FLS) has been shown to play an essential role in the chronic inflammation of RA joints [
7]. Therefore, a number of gene expression profiling studies have focused on synovial tissue or FLS to understand the aberrant biological pathways that contribute to the pathogenesis of RA [
1]. Others have focused on peripheral blood mononuclear cells (PBMC) from RA patients, either by comparing them with PBMC from healthy individuals or from patients with other autoimmune diseases [
1,
3]. Of greater interest to us is the accumulating evidence suggesting that abnormalities in the bone marrow (BM) have a significant role in RA inflammation [
2,
8,
9].
The BM contains three types of stem cells: hematopoietic stem cells (HSCs), which produce all the mature blood lineages for leukocytes, erythrocytes, and platelets; mesenchymal stem cells, which can differentiate into osteoblasts, chondrocytes, and adipocytes; and endothelial stem cells. The proliferations and differentiations of these heterogeneous cell populations are dependent on the BM microenvironment and are regulated by highly sophisticated networks, either through cell-cell interactions or cytokine networks. Indeed, a remarkable elevation in IL6 and IL8 levels in the BM serum from RA patients has been reported to relate to the synovial proliferation seen in multiple joints [
10]. Therefore, BM cells may be where the pathogenesis of RA originates, making the study of their abnormal regulatory networks very important.
In this study, we identify aberrant regulatory networks in BM cells from RA patients by analyzing differentially expressed genes based on their gene expression profiles with those of osteoarthritis (OA) patients. OA patients were chosen because the OA pathology is relatively well understood and the BM cells from these patients are far more readily available than those from healthy subjects.
Discussion
It is commonly known that autoimmunity plays a pivotal role in the pathology of RA. However, the exact etiology and pathogenesis are poorly understood. Our work, comparing the gene expression profiles of BMMC between RA patients and OA patients by microarray technology and gene ontology analysis, found abnormal immune responses in BMMC. This agrees with accumulating evidence indicating that abnormalities in BM cells may contribute to the pathogenesis of RA [
9]. To our knowledge, ours is the first report to combine DNA microarray with bioinformatics for describing gene expression profiles from RA BM cells and for revealing abnormal networks involving immune response- and cell cycle-related molecules in those cells.
Several reports have shown that peripheral blood from SLE patients has remarkably homogenous gene expression patterns and an overexpression of IFI genes [
6,
15‐
17]. The IFN signaling pathway is thought to play an important role in the pathogenesis of SLE. There is also one report of genomically profiled peripheral blood cells from 35 RA patients and 15 healthy controls that found a type I IFN signature in a subpopulation of RA patients [
3]. Here, we show that the IFN signaling pathway elevates in the BM cellular network pathway of RA patients similar to that in the peripheral blood of SLE patients, although to a lesser degree. The different IFN effects on RA and SLE may be because cytokines are pleiotropic in their biological activities and that they interact with each other in highly sophisticated networks. Along these lines, the effects of IFNβ treatment on arthritis were reviewed several years ago. An open, phase I study conducted on 12 patients with active RA and another pilot study performed on six children with juvenile RA have both shown that IFNβ treatment is in general well tolerated and leads to improvement [
18]. However, two other case reports claim RA can develop after the onset of IFNβ treatment in patients with multiple sclerosis [
18]. These suggest IFNβ therapy cannot be used universally to combat the development of arthritis.
Meanwhile, our finding that the MHC class I molecules HLA-E, HLA-F, and HLA-G, TAP, and TAPBP were all overexpressed in the BM cells of RA patients is also novel. All these genes relate to the antigen presentation pathway. For example, up-regulation of HLA-E is considered a potential marker for cancer. Additionally, its expression can confer resistance to NK cell-mediated lysis [
19,
20]. HLA-F has been recently reported to be a surface marker for activated lymphocytes [
21], while HLA-G has its highest expression during pregnancy and is thought to play a key role in modulating immune tolerance [
22]. There is a recently published study by Prigione et al. that reports a lower concentration of soluble HLA-G in sera may predispose to JIA and soluble HLA-E concentration in synovial fluid correlated with the number of affected joints [
23]. Nevertheless, the functions of these molecules in autoimmunity are still unclear and debated. In addition, we found TCR, IFNα, NFkB, p38MAPK, IL8, MyD88, and IFNγ play central roles in the immunoregulatory networks of BMMC in RA. Except for NFkB, we found all these genes to be overexpressed. MyD88, the Toll/IL-1 receptor (TIR)-containing adaptor, is used by almost all Toll-like receptors (TLRs) to activate a common signaling pathway that results in the activation of NFkB to express cytokine genes involved in inflammation, as well as IFN-inducible genes [
24,
25]. It is possible the up-regulation of MyD88 has a significant role on the aberrant immune response network seen in BMMC from RA patients. However, our data do not show a complementary up-regulation of TLRs, nor do they confirm that the up-regulation of MyD88 was caused by TLRs. It is interesting that Nagata reported up-regulation of IFN-inducible genes in DNase II-deficient mice, which develop a chronic polyarthritis resembling human RA, and they further found no involvement of a TLR system in the IFNβ gene activation in DNase II
-/- embryos [
26]. Kawane et al. also recently showed that when BM cells from the DNase II-deficient mice were transferred to the wild-type mice, they developed arthritis [
27]. Although the mechanisms of arthritis pathogenesis may be different between mice and humans, these mouse-model data do provide supportive evidence to our report.
Another interesting observation is that underexpressed genes were dominantly related to cell cycle and DNA metabolism. We are the first to report the suppression of cell cycle and DNA metabolism in BM cells from RA patients. Initially, there appear to be several possible mechanisms that can explain this result. One is a therapeutic effect caused by MTX, as MTX acts by inhibiting the metabolism of folic acid, which is needed for the
de novo synthesis of the nucleoside thymidine required for DNA synthesis. However, subsequent analysis showed MTX treatment does not correlate with the down-regulated gene expressions (data not shown). Alternatively, we considered the fact the BMMC samples in this study were isolated by using Ficoll-Paque, which may cause nucleated erythroblasts to be miscible in mononuclear cell proportions and thus affect cell cycle. Finally, a high concentration of serum IL6 in BM has been reported in RA patients [
10]. This is important because IL6 induces the secretion of hepcidin, a humoral factor regulating intestinal iron absorption and iron storage in microphages [
28,
29]. Hepcidin can contribute to low serum iron levels if up-regulated, which can then suppress erythroblast differentiation and proliferation in BM, as iron is a requisite element for this process. Furthermore, Colmegna et al. reported a defective proliferative capacity by peripheral blood hematopoietic progenitor cells from RA patients [
30]. They further showed that ATM deficiency in RA patients disrupts DNA repair and renders T cells sensitive to apoptosis [
31]. Together with their results and our finding that the ATM signaling pathway is repressed in the immunoregulatory networks of BMMC, we suggest that in RA patients, impairments in their immune response cells originally occur in the BM. However, more work is needed on a number of issues including why cell cycle and DNA metabolism were suppressed in the BM, how this suppression relates to RA, and whether defective BM cells relate to activated-immune responses in RA patients.
According to our unpublished data, the genes expressed in the peripheral blood cells of RA patients that correspond to cell cycle and DNA metabolism were not down-regulated as observed in BM cells, but the down-regulation for those in RNA metabolism- or translation-related genes were found. As all mature blood lineages in peripheral blood are produced from HSCs in the BM, the s abnormality in immune response and suppression of cell cycle in BM may contribute to the pathogenesis of RA.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
H-ML performed the data and statistical analysis, and drafted and revised the manuscript. HS and CA assisted with the acquisition of data and analysis. RS performed mRNA expression analysis with microarrays. YS and KO treated and recruited the patients for this study, and analyzed the clinical data of the patients. TO and NN made substantial contributions to the conception and design of the experiments, and analysis and interpretation of the data. All authors read and approved the final manuscript.