Molecular Profiling in Sarcoidosis

Abstract

Purpose of the Review

Sarcoidosis is a systemic disease characterized by granulomatous inflammation of unknown etiology. There is extensive heterogeneity between patients with respect to the number and types of organs involved, disease course, and response to therapy. Recent research in the field has leveraged “omics” techniques such as transcriptomics to identify important “molecular profiles” in the disease. These tools may help in identifying clinically useful biomarkers and targets for therapy.

Recent findings

Several studies have used gene expression profiling of predesignated lists or the entire genome to find genes and markers that differentiate sarcoidosis from healthy controls, but only a few have compared sarcoidosis patients based on disease phenotypes and organ involvement. The common gene pathways that have been repeatedly identified include those related to the interferon (IFN) response, T cell receptor (TCR) signaling, and the major histocompatibility complex (MHC).

Summary

While the molecular profiling studies to date offer the ability to compare sarcoidosis and health as well as across tissues, further longitudinal studies that include sarcoidosis patients with varying outcomes with respect to organ involvement and response to treatment are needed to identify clinically important phenotypes in the disease that can then be differentiated based on molecular features.

Introduction

Sarcoidosis is a disease of granulomatous inflammation that affects the lungs in over 90% of patients, but can also cause multiorgan involvement [1, 2]. The severity of disease varies greatly and outcomes vary by disease duration, including chronic and progressive courses that result in increased sarcoidosis mortality rates [3–5]. The heterogeneity in how sarcoidosis presents, the number of organs affected, longitudinal outcomes, and response to therapy is likely driven by underlying heterogeneous pathobiology. As in other lung diseases, the field is moving toward a precision medicine approach through molecular profiling, which utilizes “omics” methods to define specific “molecular phenotypes.” The utility of developing these methods includes predicting an individual patient’s disease severity, duration, and treatment responsiveness. This review will assess studies that have developed molecular phenotypes using gene expression methods. Studies that include patients with varying clinical outcomes based on disease severity, organs affected, and disease duration are crucial for determining useful molecular phenotypes, therefore, this review will highlight those that include sarcoidosis patients with differing clinical features.

Tissue Compartmentalization

Early research into the mechanisms of sarcoidosis-related inflammation focused on measuring expression of specific genes in lung samples, especially bronchoalveolar lavage (BAL) [6–10]. However, given the immune basis of sarcoidal inflammation, early work also utilized blood samples to understand, for example, human leukocyte antigen (HLA) associations with the disease and to compare blood and BAL immune cell populations [6, 7]. In other lung diseases, such as asthma and COPD, airway sampling is crucial for studying gene expression important to their pathobiology [11–15]. Similarly, obtaining lung samples is critical to understand the immune response in pulmonary sarcoidosis, but the fact that other organs can be involved lends to the complexity of tissues that can be sampled. The systemic nature of sarcoidosis also allows for the opportunity to use blood samples to gain insights into pathways that are globally affected. Rutherford et al. [16] used peripheral blood mononuclear cell (PBMC) samples to show differences in expression of the Bcl-2 family of genes and TNF-α-related pathways between sarcoidosis and heathy controls and later showed differences in expression of several genes including TNFA from PBMCs in sarcoidosis patients with differing disease courses [17]. In a study of PBMC, lung tissue, and lymph node samples, Rosenbaum et al. [18] identified a common set of genes regulated by STAT1 that were increased across all three tissue compartments relative to healthy controls. Subsequent studies have similarly shown how peripheral blood gene expression can differentiate sarcoidosis from health and other diseases and how PBMC and tissue gene profiles overlap [19–23]. This work underscores the utility of using peripheral blood to non-invasively phenotype patients and understand the underlying mechanisms that drive organ-specific inflammation and the resulting systemic response.

Interferon-Related Pathways

Given that a central mediator of granulomatous inflammation in sarcoidosis and tuberculosis (TB) infections includes the IFN response, several transcriptomic studies have compared the gene expression patterns in patients with these two diseases and have shown commonalities [19, 20, 24]. Three separate research groups [19, 20, 24], including our own at UCSF, found consistent overlap in the expression profiles between TB and sarcoidosis, especially with respect to type I and II interferons. Studies that have used lung sampling to identify gene expression profiles specific to tissue affected by granulomatous inflammation have also found an upregulation of pathways related to IFN and STAT1. Rosenbaum et al. [18] showed upregulation of STAT1-related genes including CCR5, and CXCL9 when comparing peripheral blood, lymph node, and lung samples. Crouser et al. [25] demonstrate that IFN-γ-related genes including STAT1, were increased in sarcoidosis lung tissue samples relative to healthy and also identified two IFN-independent genes MMP12 and ADAMDEC1 that were the were most highly differentially expressed. In an in vitro model of granuloma formation developed by Crouser et al. [26] wherein PBMCs from subjects with sarcoidosis and latent TB infection (LTBI) were exposed to PPD-coated beads, granuloma-like structures induced by samples derived from both sarcoidosis and LTBI samples also showed an upregulation of IFN-γ/STAT1 signaling pathways.

Studies that have compared expression profiles in sarcoidosis to healthy controls and TB have allowed for development of useful gene markers to identify phenotypes among sarcoidosis patients. Using a gene set that included those upregulated in our initial transcriptome analysis at UCSF [24], we found genes transcripts that differentiated sarcoidosis subjects based on their disease course as defined by decline in pulmonary function tests (PFTs) and persistent or increased immunosuppression use [27]. These genes included CXCL9 and three gene means identified using similar factor analysis methods employed in COPD and asthma [11, 14, 15, 28–31], which included IFN-related genes STAT1, STAT2, and GBP1 and TCR-related genes CD28, ITK, and LEF1. We subsequently found that the “TCR” factor was significantly different between those with stable PFTs, a decline in PFTs, and increases of PFTs [32], suggesting that those with and without subsequent dynamic changes in PFTs can be distinguished based on whole blood gene expression related to lymphocyte TCR signaling. We further showed how the serum protein levels of the chemokines CXCL9, CXCL10, and CXCL11 were associated with PFT deficits, PFT declines, and respiratory symptom severity as well as other IFN-related genes [33, 34]. Other groups have used transcriptomic analyses to differentiate sarcoidosis cases based on disease severity including in the study by Rutherford et al. [17] that found that the PBMC expression of HLA-DRB1*1501 DQB1*0602, TNFA, NFKB, CREM, and CD69 were increased in sarcoidosis with persistent as compared to self-limited disease. Lockstone et al. [35] compared sarcoidosis patients with self-limiting, active, and fibrotic disease and found that genes related to leukocyte activation and differentiation as well as pathways related to NF-κB and JAK-STAT were up-regulated in subjects with progressive disease and pulmonary fibrosis as compared to healthy controls and those with self-limited disease.

T Cell Receptor and T Cell Responses

T lymphocytes, especially CD4⁺ T helper (Th) cells, are critical in the formation and propagation of granulomas and multiple studies have established the importance of IFN-γ-producing Th cells in sarcoidosis [36–40]. There is an extensive literature on TCR [41–47] and MHC [48–54] genotyping in sarcoidosis that also supports the importance of T cells in sarcoidal inflammation. With respect to transcriptomic studies, Zou et al. [55] compared PBMCs in complicated (those with severe lung dysfunction or cardiac and neurologic involvement) or remitting (uncomplicated) sarcoidosis as well as healthy controls. They used a pre-set list of genes related to TCR signaling, JAK-STAT signaling and chemokine-cytokine receptor signaling to differentiate cases from healthy controls in both their dataset and datasets from UCSF [24] and Oregon [56]. Gharib et al. [57] found 86 gene sets that were significantly enriched in sarcoidosis BAL compared to healthy controls and many of these pathways mapped to inflammatory and immune-related processes including adaptive immunity, T-cell signaling (including IL-12 and IL-17). Similar findings were identified in a multiwall carbon nanotube mouse model of granulomatous inflammation wherein alveolar macrophages from these mice and patients with sarcoidosis had a common set of upregulated genes related to T-cell signaling, IL-12/IL-17 signaling, and oxidative phosphorylation [58].

Organ Involvement

One of the many challenges in sarcoidosis is determining the burden of systemic organ involvement and assessing severity of disease in specific organs. This was one of the motivations behind the GRADS (Genomic Research in Alpha-1 Antitrypsin Deficiency and Sarcoidosis) study [23], which included sarcoidosis subjects with specific organ involvement, published results from which are pending. To differentiate immune responses amongst distinct organs, several studies have compared granulomas related to cutaneous sarcoidosis to those with strictly pulmonary sarcoidosis. Judson et al. [59] found that Th17-related genes were dysregulated in skin lesions, including those for IL-12p40 (a component of IL-12 and IL-23) and the cytokine IL-21. They also found that the IFN-γ signaling pathway was the most prominent set of genes in whole blood and in sarcoidosis skin lesions, underscoring how peripheral blood can reflect inflammation at affected sites. Monast et al. [21] compared whole blood gene expression between those with cutaneous and pulmonary sarcoidosis and found distinct patterns related to interferon signaling. Rosenbaum et al. [22] compared lacrimal gland, orbital tissue, and whole blood profiles and found that all three tissues had increased expression of STAT1, JAK2, as well as GBP5, AIM2, SLAMF8, which are also induced by IFN-γ. Sheng et al. [59] used publicly available datasets for pulmonary sarcoidosis [25, 35] and cutaneous sarcoidosis [60] to differentiate genes related to each of these specific phenotypes. A combination of biomarkers (either CLDN7, NUDT11 and FAM73A; or TAP1 and TAP2) identified by feature selection using a support vector machine (SVM) classifier were able differentiate those with cutaneous versus pulmonary sarcoidosis with efficiencies of 87% and 92%, respectively. These studies show the common pathways in different sites of inflammation, but also demonstrate how molecular profiling can be used to distinguish those with different types of organ involvement.

MicroRNAs

MicroRNAs (miRNAs) are small noncoding RNA oligomers of typically 22–25 nucleotides that post-transcriptionally regulate other expressed mRNAs through an RNA-induced silencing complex based on complimentary sequence homology [61]. MicroRNAs are important regulators of gene expression and their differential expression can help identify gene pathways related to specific disease states. Recent studies have measured microRNA expression in sarcoidosis subjects [62–67], and while most of these studies assessed different pre-specified microRNAs, two microRNAs miR-146a and miR-150, were measured in more than one study [63, 68]. Kishore et al. [68] found increased levels of both miR-146a and miR-150 in extracellular vesicles from BAL sarcoidosis samples compared to healthy controls and in sarcoidosis subjects with Scadding Stage II as compared to Scadding Stage I radiographs; these microRNAs also negatively correlated with lung function including vital capacity and FEV1/FVC ratio. Dyskova et al. [66] found that BAL levels of miR-146a and miR-150 were also increased in sarcoidosis subjects relative to healthy controls, and a separate study confirmed this finding for miR-146a in serum [67]. miR-150 potentially is related to activation of CD3⁺CD4⁺ T cells through NOTCH3, while miR-146a is likely inhibitory given its increased expression due to TNF-α and IL-1β in the setting of NF-κB activation [68]. Other microRNAs related to angiogenesis and inflammation such as miR-27b, miR-192 and miR-221 have been found to be elevated in sarcoidosis with parenchymal involvement as compared to those with thoracic adenopathy only, while miR-16 and miR-20a expression was significantly higher in patients with lung volume restriction [62]. The micro-RNAs miR-33 and miR-34 have also been shown to be increased in sarcoidosis BAL and PBMCs, respectively; miR-33 is related to a lipid transporter regulator and miR-34 is involved in IFN-γ regulation [64, 65]. These studies suggest that microRNAs, especially miR-146a and miR-150, have potential as markers of higher inflammatory burden in sarcoidosis patients and should be explored further.

Microbiomics, Proteomics, Metabolomics, and Multi-Omics

Several groups have used 16S DNA sequencing to study the microbiome in sarcoidosis respiratory samples, although these studies have not found differences in microbial diversity between sarcoidosis and other ILD or autoimmune diseases [69–71]; conflicting studies have shown presence [72, 73] and absence [74] of respiratory microbial composition differences in sarcoidosis compared to healthy controls. Other “omics” techniques have included proteomics and metabolomics. Talwar et al. [75] employed a high-throughput T7 phage display protocol to obtain 1152 potential sarcoidosis antigens and compare them across patients with sarcoidosis, TB, and healthy controls. This type of approach brings the field closer to finding a specific antigen for the disease, especially if it can be coupled with advances in TCR sequencing and proteomics [46, 47]. In this study, they identified a combination of 32 antigens that was able to distinguish sarcoidosis from healthy controls, including the top 10 most different antigens, half of which were known antigens and half consistent with mimotopes (i.e. peptides similar to known epitopes). Mass-spectrometry techniques have been used by several groups to show differences in specific blood metabolites between sarcoidosis and healthy controls [76, 77] and between sarcoidosis subjects with and without pulmonary fibrosis [78]. To leverage multi-omics approaches, a recent study used established genomic, transcriptomic, and proteomics data from public datasets to perform a positional integration analysis [79]. This analysis identified several previously-recognized regions related to sarcoidosis (e.g. TNFA). Two regions on chromosome 6 were most highly upregulated in sarcoidosis, and the three main pathways that were involved included, MHC II (TAP1, TAP2, PSMB8 and PSMB9) and in endothelial cells (MIR1236, EGFL7). These findings confirmed the importance of genes such as TNFA and re-emphasized findings from other studies showed the potential utility of MHC-related proteins such as TAP1 and TAP2.

Response to Therapy

One of the main goals of personalized medicine is to be able to predict which patients may be responders to specific therapies. In asthma, for example, a combination of epithelial genes and cytokines have been used to define a “T2” phenotype characterized by high Th2 immunity, which has been associated with improved response to inhaled corticosteroids [11]. In sarcoidosis, therapeutic studies of infliximab have shown that higher serum levels of soluble IL-2 receptor (sIL-2R) are associated with increased risk of sarcoidosis relapse after discontinuing infliximab [80]. In our UCSF study, we found that those on higher doses of immunosuppression had lower levels of the CXCL9 and CXCL10 chemokines [34]. Arakelyan et al. [81] analyzed a previously published dataset of pulmonary sarcoidosis patients and healthy controls [25] along with datasets from studies using infliximab in other diseases such as ulcerative colitis. In a proof of concept study, they found that there were distinct signatures of genes in ulcerative colitis patients who responded to infliximab and these same set of genes were upregulated in sarcoidosis patients relative to healthy controls, suggesting that infliximab likely targets the same downstream genes in sarcoidosis. These types of analyses are becoming more feasible as a greater amount of transcriptomic and proteomic datasets are becoming publicly available, especially ones from clinical trials [82]. Given the complexity of sarcoidosis in terms of patients’ prognosis regardless of immunosuppression and the many different indications for therapeutic use, genetic, transcriptomic, and proteomic profiling will be crucial in identifying who will require therapy and to which therapy an individual patient will most likely respond.

Conclusion

There has been an impressive amount of genomic and proteomic profiling in sarcoidosis to date that has revealed a consistent pattern of genes that are upregulated in the disease including those related to STAT1 and IFN signaling, the TCR, and the MHC complex. As newer technologies offer faster and cheaper ways to increase the scale of “omics”-based research, the field must recognize the important clinical and pathobiologic issues that must be addressed. Most extensive profiling studies have focused on comparing sarcoidosis to healthy controls or other diseases, which does not address differences in which organs are affected, how patients respond to therapy, and clinical course. Longitudinal analysis of the immune responses in patients with varying clinical manifestations and trajectories along with combining transcriptional, genetic, and proteomic technologies to understand the inflammatory response in the context of specific genetic profiles is essential for future studies. With improved understanding of these complex interactions, more specific molecular profiles can be developed to phenotype and treat patients.

Key points:

• Gene expression profiling and other omics techniques including integrative analyses continue to identify specific pathways that are upregulated in sarcoidosis.

• The systemic aspects of sarcoidosis offer the ability to gain insights into sarcoidosis pathobiology by interrogating both the blood and affected organs.

• A common group of gene networks related to interferons, T cell receptor signaling, and the MHC have consistently been shown to distinguish disease from health and are associated with clinical features in the disease.

• MicroRNAs, microbiomics, metabolomics, proteomics, and integrative “multi-omics” approaches offers new avenues to phenotype patients and identify potentially effective therapeutics.

• Future research should incorporate evolving genomic technologies in longitudinal studies that include sarcoidosis patients with varying clinical courses to develop molecular signatures that can predict disease outcomes.