Background
The hallmark of rheumatoid arthritis (RA) is chronic synovitis that affects multiple joints and invades cartilage causing bone erosions and joint destruction [
1]. As the synovium is the principal target of inflammation in RA, and the resident synoviocytes (fibroblast-like synoviocytes and macrophages-like synoviocytes) along with recruited cells (myeloid cells and lymphocytes) are implicated in the pathogenesis of synovitis, special interest has been given to the study of synovial tissue in this disease. These studies not only aim to clarify RA pathogenesis and provide insight into the mechanisms of action of therapeutic interventions [
1,
2], but are also a promising approach to search for biomarkers in the inflamed synovial tissue [
1]. Changes in the cellular infiltrate or biomarkers such as cytokines or growth factors in RA-affected synovial tissue have long been known to be associated with the clinical course of disease and have been used to identify specific responses to RA therapies [
1‐
4]. Recently there has been increasing interest in synovial biopsies to obtain inflamed synovial tissue from joints and thereby gain a better understanding of the pathogenic events in these diseases [
5]. Histopathotype and pathological pathways-based patient stratification prior to therapeutic intervention could be exploited to identify biomarker predictors of clinical outcomes and responses to therapy [
6,
7].
Tissue pathology and pathogenic pathways cannot yet be reliably explored through noninvasive circulating or imaging biomarkers. Given the complexity and heterogeneous nature of RA, it is unlikely that a single cytokine will provide sufficient discrimination between patients and thus be a good biomarker [
8,
9]. Global biomarker signatures may represent a more appropriate approach for improving treatment protocols and outcomes for patients with RA. Metabolomics is the science of identifying and quantifying the biochemical byproducts of metabolism in a cell, tissue, or organism [
10]. Metabolomics is an emerging field of biomedical research that can offer a better understanding of the mechanisms underlying disease and help to develop new strategies for treatment [
11]. Unlike genes and proteins, which are epigenetically regulated and post-translationally modified, metabolites are direct signatures of biochemical activity and thus it may be easier to test whether they are correlated with phenotype [
12].
The fundamental rationale in metabolomics is that perturbations caused by a disease in a biological system will lead to changes that are correlated with the concentrations of certain metabolites [
13,
14]. Metabolite patterns represent the final response of biological systems to disease status, or in response to a medical or external intervention [
12].
1H-nuclear magnetic resonance (NMR) can delineate patterns of changes in biomarkers that are highly discriminatory for the observed disease or intervention [
15]. We propose in this work, that the study of metabolomics in serum from patients with RA, using NMR, can be used to predict synovial pathology. We hypothesize that perturbations caused by inflamed synovial tissue will lead to changes that correlate with the concentration of certain metabolites in the synovium. These changes will then be reflected in blood serum and function as potential biomarkers of different synovial markers. Here we describe the first study that defines metabolite signatures in serum that correlate with gene expression profiling in synovial tissue from patients with active RA.
Discussion
Our increasing understanding of the pathogenesis of RA has transformed the therapeutic options available for people with this disease. The introduction of newer agents and novel treatment strategies has resulted in improved outcomes for patients. However, these successes have raised the bar for the goals of therapy. At present, disease remission, or low disease activity at the very least, has become the new goal of treatment for all patients. Therefore, there is still an unmet need in RA. Biomarkers employed in “personalized” medicine might be useful in an attempt to match a patient with the most appropriate biologic therapy, and thereby optimize outcomes. The accessibility of a biological biomarker is an important factor in this approach [
8]. Although sampling inflamed synovial tissue from joints might be critical to gain a better understanding of the pathogenic events of inflammatory arthritis, a biomarker that can be obtained in a minimally invasive manner is more attractive, particularly for patients in early stages of the disease, where mostly small joints are involved [
8]. In this study, we attempt, for the first time, to find serum metabolomics profiles that correlate with synovial marker gene expression.
Recent studies have indicated that metabolic regulation and cell signaling are tightly and ubiquitously linked with immune responses. Metabolomics studies that aim to improve biological understanding through the analysis of metabolite profiles of the underlying biological pathways are certainly relevant and have been successful in other fields, especially oncology. Though the application of metabolomics to RA is still in its infancy, early studies have yielded promising results [
19,
26‐
33]. A small number of metabolomics studies have focused on identifying metabolites associated with rheumatic diseases, primarily in the serum for diagnostic purposes [
30‐
32], but none have attempted to predict synovial pathology.
We hypothesized that perturbations caused by inflamed synovial tissue will lead to changes that correlate with the concentrations of certain metabolites in the synovium that will be then reflected in blood serum. A recent publication on a study of metabolic profiling in the synovial tissue reported altered glucose and choline metabolism [
34]. Both pathways have recently been involved in RA pathogenesis [
27,
34,
35]. Choline levels in patients from our cohorts are decreased in blood compared to the normal range; this, along with an increased uptake in the joints on choline C-11 PET scanning in inflammatory arthritis [
36] and high expression in fibrocyte-like synoviocytes (FLS) of choline like transporter (CTL)1 (high-affinity) and CTL2 (low-affinity) [
37], suggest increased circulating choline uptake and consumption by the inflamed synovium. Glucose levels were decreased, and lactate levels increased in serum from our cohort. Glucose is consumed through upregulation of aerobic glycolysis and when metabolized, gives rise to production of copious amounts of lactate, which must be extruded from the cell to prevent lactic acidosis [
38]. Several studies have highlighted the increase in glucose metabolism in the hypoxic joint [
27,
35]. Thus, our results in serum seem to agree well with recently described synovial studies [
34]. Of interest, both choline and glucose levels in the blood negatively correlated with TNF-α and CD3E gene expression in the synovium.
Literature in the field of oncology can help us to interpret some of our results. For instance, we observed an association of serine/glycine metabolism and aminoacyl-tRNA biosynthesis with TNF-α/CD3E and B/plasma cell signatures that suggest that lymphoid cells could be using these pathways after activation in the rheumatoid synovium. Although alterations in glucose and glutamine metabolism are central to metabolic transformation, recent studies have focused on the role of the nonessential amino acids serine and glycine in supporting tumor growth [
39]. In addition to their role in protein synthesis, serine and glycine contribute to anabolic pathways important for the generation of glutathione, nucleotides, phospholipids, and other metabolites [
40]. The requirement for intracellular serine and glycine for the support of cell growth and proliferation is clear. Other amino acids are also critical substrates that fuel mitochondrial metabolism and the biosynthesis of proteins, lipids, and other molecules. Of particular interest in cancer are key mitochondrial enzymes in the metabolism of glutamine, glutamate, proline, aspartate, and alanine [
41]. The branched chain amino acids (BCAAs) valine, leucine, and isoleucine are also highly metabolized by transaminases. By coordinating cellular bioenergetics and biosynthesis through the tricarboxylic acid (TCA) cycle, amino acid metabolism could be critical not only in tumor cells but also in lymphoid cell proliferation and survival as described recently [
42].
Another metabolite that correlates with several of our cytokine pathways is succinate. Succinate is an intermediate of the TCA cycle and plays a crucial role in adenosine triphosphate (ATP) generation in mitochondria. Recently, new roles for succinate outside metabolism have emerged. Succinate promotes expression of the pro-inflammatory cytokine IL-1β by inhibiting prolyl hydroxylases and stabilizing the transcription factor hypoxia-inducible factor-1α (HIF-1α) in activated macrophages, and stimulates dendritic cells via succinate receptor 1 [
38,
43]. Furthermore, succinate has been shown to post-translationally modify proteins. Of interest, the succinate level in blood positively associated with synovial IL-1β gene expression although it did not reach statistical significance.
The cluster comprising MMP1/MMP3/IL-6, which could represent a fibroblast-driven phenotype, was negatively correlated with ketone bodies. Acetoacetate is the common precursor of the two other circulating ketone bodies, acetone and 3-hydroxybutyrate [
44]. 3-hydroxybutyrate is the most abundant circulating ketone body and is less likely to degrade spontaneously into acetone than acetoacetate. One can speculate that rheumatoid fibroblasts require intracellular ketone bodies for the support of their invasive phenotype and that the increase in 3-hydroxybutyrate uptake and/or enzymes in this pathway could explain the negative correlation. Of note, the positive correlation between 3-hydroxybutyrate and IL-1β and IL-8 is also of interest, as 3-hydroxybutyrate, long viewed as a simple carrier of energy from the liver to peripheral tissues, also possesses signaling activities and is also an endogenous inhibitor of histone deacetylases (HDACs) [
45]. Moreover, recent research has shown that 3-hydroxybutyrate can block the NOD-like receptor pyrin containing 3 (NLRP3) inflammasome [
46]. Further studies are needed to understand the effect of these metabolites in the synovium in RA.
As mentioned above, metabolites can not only be biomarkers of perturbations caused by inflamed synovial tissue but also can have a pathogenic effect that would amplify synovial inflammation. Secondary roles have emerged for glucose metabolites, metabolic enzymes, and TCA cycle intermediates outside of metabolism. Not only succinate but also other metabolites including α-ketoglutarate, fumarate, and acetyl-CoA might be expected to accumulate in macrophages and FLS under hypoxic conditions, and are involved in eliciting important epigenetic changes, with unexplored potential for driving chronic inflammation [
47,
48]. Also, essential glycolytic enzymes have been shown to translocate to the nucleus or mitochondria where they function independently of their canonical metabolic roles in the regulation of cytokines and anti-apoptotic responses [
49,
50]. Thus, metabolomics studies have also the potential of defining the elements of synovial metabolic pathobiology.
Although NMR spectroscopy has less sensitivity compared to mass spectrometry instrumentation, NMR requires minimal sample preparation, and is not only non-destructive, inherently untargeted, highly reproducible [
51,
52], and intrinsically quantitative, but is also cheaper and more accessible than mass spectrometry [
53‐
55]. Depending on the biological samples, NMR can identify and quantify more than 200 metabolites in an untargeted fashion and more than 100 metabolites are uniquely identified by NMR [
56]. In this work, we also showed that the combination of only two or three metabolites identified in serum by NMR could discriminate between high or low levels of synovial TNF-α and CD3E gene expression. Studies in other cohorts of patients with active RA are needed to validate these results, yet the relationship between serum metabolic profiles and synovial biomarker profiling suggests that NMR may be a promising tool for predicting specific pathogenic pathways in the inflamed synovium in RA.
Although these findings are certainly promising, this study is not without limitations. Most importantly, we evaluated a small number of clinical samples. Despite similar clinical parameters for patient inclusion, large biological variance is expected in primary samples. In addition, patients had long-standing disease and were exposed to various therapies prior to the study, and were on methotrexate at the time of the study, which is reported to change several metabolic pathways including adenosine metabolism [
57]. Confirmation of our results in a larger sample size from a cohort of patients with new onset inflammatory arthritis before treatment initiation, studied prospectively, is necessary to strengthen our conclusions. Comparison with other arthritides or other systemic inflammatory diseases to determine if these changes in metabolite levels come from the joints or from different sources is also critical to interpret our results. One other confounder is the microbiome, which is altered in RA and can potentially cause metabolic changes in both serum and synovial tissues [
58‐
60]. In addition, further studies are needed to evaluate the relationship between circulating metabolites and synovial pathology. Metabolite profiles in blood, if they correlate with metabolic changes in synovial tissue, will certainly reveal more about RA etiology. We did not identify correlation between cytokine serum levels and cytokine synovial gene expression (Additional file
8: Figure S7), yet it remains unknown whether or not metabolic changes will display stronger correlation between blood and synovium.