Identification of Candidate Predictors of Lupus Flare

Abstract

Systemic lupus erythematosus, the prototype systemic autoimmune disease, is characterized by extensive self-reactivity, inflammation, and organ system damage. Sustained production of type I interferon is seen in many patients and contributes to immune dysregulation. Disease activity fluctuates with periods of relative quiescence or effective management by immunosuppressive drugs, followed by disease flares. Tissue damage accumulates over time, with kidneys and cardiovascular system particularly affected. Identification of the underlying molecular mechanisms that precede clinical exacerbations, allowing prediction of future flare, could lead to therapeutic interventions that prevent severe disease. We generated gene expression data from a longitudinal cohort of lupus patients, some showing at least one period of severe flare and others with relatively stable disease over the period of study. Candidate predictors of future clinical flare were identified based on analysis of differentially expressed gene transcripts between the flare and non-flare groups at a time when all patients had relatively quiescent clinical disease activity. Our results suggest the hypothesis that altered regulation of genome stability and nucleic acid fidelity may be important molecular precursors of future clinical flare, generating endogenous nucleic acid triggers that engage intracellular mechanisms that mimic a chronic host response to viral infection.

INTRODUCTION

Systemic lupus erythematosus (SLE) is among the most complex of all diseases, involving multiple organ systems and characterized by impaired immune system regulation. The disease occurs more often in females than in males, with a 9:1 female:male ratio, and typically has its onset during the childbearing years. The disease is characterized by protean immunologic alterations, with T cell−dependent production of autoantibodies that target nucleic acids or nucleic acid binding proteins a common feature. In addition, increased production of some cytokines and chemokines contributes to altered immune function and inflammation. In that regard, production of type I interferon (IFN) is increased and sustained in many patients and contributes to many of the immunologic abnormalities observed (1,2).

Clinical manifestations of disease can be highly variable from one patient to another, but skin rash, arthralgias and arthritis, serositis, and fatigue are common. Glomerulonephritis, various central nervous system manifestations, and premature atherosclerosis can lead to significant morbidity and even mortality. Disease activity often fluctuates with periods of relative quiescence or effective management by immunosuppressive drugs followed by disease flares. Elevation of acute phase reactants, increased titers of anti-DNA antibodies, and complement activation typically are associated with disease flares, but the primary biologic precursors of clinical and serologic disease flares are not known.

The clinical features of lupus indicate important opportunities for research that can lead to new insights and advances in treatment for patients. Studies focused on dissecting the biologic correlates of disease heterogeneity, including the basis of the striking female predominance of lupus, are likely to identify important molecular pathways and therapeutic targets. Identifying biomarkers of disease activity and response to therapy will aid in drug development and medical management of patients. Most challenging and offering the greatest opportunity to control or even prevent disease, discovery of the biologic precursors of future clinical flare might elucidate those biologic mechanisms that are drivers of the altered immune system function, autoimmunity, and inflammation that are responsible for organ damage and disease.

TYPE I IFN: A CENTRAL MEDIATOR OF IMMUNE SYSTEM ALTERATIONS IN SLE

Type I IFN, comprising 13 IFN-α, IFN-β, and several other subtypes, was the first cytokine described, and extensive investigations have defined its many roles in anti-virus host defense (3,4). Reports of elevated serum IFN in lupus patients began to emerge in the 1970s, and clinical observations of the occasional induction of clinical manifestations consistent with lupus after administration of therapeutic recombinant IFN-α were later reported (5–7). Extensive studies of the biologic stimuli that induce production of IFN in lupus patients accompanied the demonstration of increased expression of hundreds of IFN-I−regulated gene transcripts in the blood of lupus patients, termed the IFN signature, suggesting a picture consistent with a virus infection, or alternatively, an anti-viral−like immune response driven by endogenous triggers (8–13). The IFN signature in lupus peripheral blood cells has been observed by numerous laboratories and is a feature of nearly all pediatric lupus patients and 60% to 70% of adult lupus patients. Increased expression of type I IFN−induced genes and expression of IFN-regulated proteins has also been observed at sites of tissue involvement, including skin, kidney, and synovial tissue (14,15). The clinical significance of type I IFN pathway activation is supported by data showing an association of IFN-induced gene expression with disease activity (16). The significance of elevated expression of type I IFN−regulated genes for altered immune function, autoimmunity, and organ damage is supported by the broad effects of sustained production of IFN-α that are seen in chronic viral infection, as shown in murine models of infection with lymphocytic choriomeningitis virus (17,18). In that situation, T cell function is skewed toward a T follicular helper cell phenotype, supportive of B cell differentiation, and proinflammatory mediators such as interleukin 6 and tumor necrosis factor are produced, contributing to tissue inflammation (19). The striking image of the IFN signature shown in “heat map” depictions of microarray gene expression data, along with knowledge of the contribution of prolonged activation of the type I IFN pathway to organ damage and disease, provide the rationale for investigation directed at identification of the upstream drivers of type I IFN production in lupus patients.

INSIGHTS FROM GENETIC STUDIES IN SLE

Like most autoimmune diseases, susceptibility to SLE shows a strong association with several major histocompatibility type II alleles, implicating a role for presentation of peptide antigens to T cells of the adaptive immune response. Deficiencies in certain early complement components are also associated with SLE, suggesting that impaired clearance of apoptotic debris might provide an antigenic stimulus for activation of a self-directed immune response.

Studies of common genetic variants as well as monogenic disorders with lupus-like clinical and serologic features have provided new insights into the biologic basis of the immune system activation that is responsible for clinical disease. Extensive studies of single nucleotide polymorphisms in thousands of lupus patients and healthy individuals have identified more than 50 genetic polymorphisms associated with a diagnosis of SLE, providing new insights into the molecular pathways that are important in generating or sustaining the altered immune function that underlies the disease (20). Some of those genetic variants encode kinases, phosphatases, and adaptors that might alter the threshold for lymphocyte activation, but most striking have been those variants that can be mapped to the innate immune response, including the endosomal Toll-like receptor (TLR) pathway responsible for recognition of DNA or RNA. In addition, single-gene mutations that result in clinical syndromes with similarities to lupus have suggested novel signaling pathways and endogenous drivers that might be relevant to lupus pathogenesis.

Aicardi-Goutieres syndrome (AGS) is a rare condition, typically diagnosed in infancy or early childhood and characterized by central nervous system disease, including seizures, encephalopathy or cerebral calcifications, distal skin lesions, and serologic features similar to those seen in patients with lupus, specifically elevated type I IFN (21). Many patients demonstrate autoantibodies specific for DNA- or RNA-binding proteins, as seen in SLE. Insight into the molecular pathways and mechanisms responsible for AGS derive from the discovery of mutations in any of several genes encoding proteins that regulate nucleic acid metabolism. Mutations in TREX1, encoding 3' repair exonuclease I (DNase III); SAMHD1, encoding an enzyme that degrades nucleotide triphosphates and also has nuclease activity against RNA in DNA/RNA hybrids; RNASEH2, encoding subunits of a complex with activity that removes single RNAs from DNA-DNA duplexes; and ADAR, encoding double-stranded RNA-specific adenosine deaminase and with RNA editing function, have all been associated with AGS (22,23). The association of altered expression or function of these gene products, each affecting nucleic acid editing or degradation, with a common clinical syndrome has implicated endogenous nucleic acids as drivers of the type I IFN pathway that is activated in AGS and has drawn attention to intracellular sensors of self-nucleic acids and the signaling pathways that they activate as possibly relevant to SLE pathogenesis (24–28). Additional support for that concept comes from an association of mutations in IFIH1, encoding MDA5, a cytoplasmic RNA sensor, with AGS and common polymorphisms in that gene with SLE (23,29,30). Both DNA and RNA cytoplasmic nucleic acid receptors trigger new gene transcription, including transcription of type I IFN, through Tank-binding kinase 1 (TBK1), and phosphorylation of TBK1 is dependent on the adaptor STING (stimulator of type I IFN gene) (31). Activating mutations in STING have also been associated with a severe clinical syndrome characterized by multi-organ system inflammation and vasculopathy, along with increased expression of type I IFN−induced gene transcripts and proteins (32,33). Although neither AGS nor the vasculopathy associated with mutations in STING generate clinical syndromes identical to SLE, both genetic diseases hold clinical and serologic similarities to lupus.

Taken together, the identification of single-gene mutations in regulators of nucleic acid metabolism, cytoplasmic nucleic acid sensors, and adaptors that are required for signaling new gene transcription downstream of innate immune system activation by cytoplasmic nucleic acids has revealed the TLR-independent pathway of innate immune system activation and its drivers as comprising multiple possible points of altered regulation relevant to lupus. The evolving concept is that alterations in intrinsic cellular regulation of nucleic acid integrity or degradation can generate innate immune stimuli that drive type I IFN production. Alterations in those mechanisms might represent early events in the pathogenesis of lupus that facilitate immune system activation but precede the development of autoantibodies (1).

Additional insight regarding potential triggers of innate immune system activation in lupus patients has come from genome-wide association studies of lupus patients and healthy donors. Those studies have identified genetic variants that suggest an important role for recognition of nucleic acids by the endosomal sensors of nucleic acids – TLR7, TLR8 and TLR9 – in lupus pathogenesis and particularly for induction of high level production of type I IFN (20,34). Common genetic variants in TLR7, an endosomal TLR that recognizes single-stranded RNA, and in IRF5, an IFN-regulatory factor that is activated to translocate to the nucleus and contribute to transcription of IFN-α and other proinflammatory mediators, are associated with a diagnosis of SLE. Our laboratory documented a significant association of the lupus risk variant of IRF5 with elevated levels of serum type I IFN activity in patients positive for autoantibodies specific for RNA-binding proteins (RBP, such as Ro, La, Sm or RNP) or DNA, but no association of the risk IRF5 variant with IFN production was seen in those patients who did not have measureable anti-RBP or anti-DNA antibodies (34). Our interpretation of those data is that the IRF5 risk genetic variant is only relevant to production of type I IFN in the presence of immune complexes containing stimulatory nucleic acids and the autoantibodies that can deliver those complexes to the relevant endosomal TLR by binding to Fc receptors. In vitro studies showed that immune complexes from patients with anti-RBP autoantibodies were particularly active in stimulating production of type I IFN by responder peripheral blood mononuclear cells (PBMC) or plasmacytoid dendritic cells from healthy donors. Together with our studies showing a strong association between presence of anti-RBP autoantibodies and high level expression of type I IFN−regulated gene transcripts in patient PBMC, these in vitro experiments support the hypothesis that RNA-containing immune complexes are particularly active in driving type I IFN production through RNA-sensing endosomal TLRs, with TLR7 likely to be most important (16). Studies in murine systems also show a significant role for TLR7 in immune complex−mediated activation of B cells and autoantibody production (35–37).

The current challenge presented by these data is to dissect the relative roles of the cytoplasmic nucleic acid sensing pathways and the endosomal TLR pathways in initiating and amplifying immune system activation and production of type I IFN and other mediators that can modulate immune function. It will be particularly important to characterize those cytoplasmic nucleic acids that might reflect altered regulation of nucleic acid metabolism and drive activation of cytoplasmic nucleic acids sensors.

CANDIDATE PRECURSORS OF LUPUS FLARE

Bearing in mind the potential for innate immune system activation and production of type I IFN to prime the immune system for altered function, development of autoimmunity, and inflammation, we designed a pilot study to gain insight into the biologic events that precede clinical disease flare and drive immune activation.

To identify gene transcripts that were differentially expressed between patients who would go on to flare from those patients whose disease activity remained low and stable, microarray gene expression analysis was performed on longitudinal samples of PBMC collected from 23 lupus patients over 2 to 3 years. Clinical data were obtained at the time of blood sample collection and flare status determined, as defined previously (38). Patients were characterized as either “flaring” (greater than one lupus flare/y) or “non-flaring” (less than one lupus flare/y). A gene expression data set was selected to represent each patient at a point in time at least 100 days before a documented lupus flare and when all patients were relatively quiescent clinically (Figure 1). Significantly differentially expressed transcripts between flaring and non-flaring patients were determined and distribution of those transcripts among the study patients was displayed using an unsupervised clustering algorithm (Figure 2).

Fig. 1.

Study design. Twenty-three patients meeting classification criteria for systemic lupus erythematosus were studied longitudinally over 2−3 years. From those, 10 patients showing > 1 lupus flare/y were designated as the flare group. Ten patients showing < 1 lupus flare/y were designated as the non-flare group. Microarray gene expression datasets derived from peripheral blood mononuclear cells collected at a point in time at which each patient was in a non-flare state, with low disease activity, were selected for analysis. For the flare group, the dataset selected was collected at least 100 days before a lupus flare.

Fig. 2.

Unsupervised clustering of differentially expressed gene transcripts between the flare group and non-flare group. Gene expression data were compared between flare (H) and non-flare (L) lupus patients. Transcripts significantly differentially expressed with a P value <0.05 were visualized in a heat map derived from unsupervised cluster analysis. Data from patients in each of the two groups generally sorted together. Gene names are indicated to the right of the heat map.

Differentially expressed transcripts generally clustered into two groups, defined by flare status (Figure 2). Review of the differentially expressed transcripts identified a transcript (LY6E) that has been previously associated with future lupus disease activity, supporting the validity of the analysis (39). Among the significantly differentially expressed transcripts that have been functionally characterized, APOBEC3B was expressed at a 3.9-fold higher level in the flare group compared to the non-flare group and is notable for its role in regulating retrotransposon expression (Table 1) (40). Other transcripts represent long noncoding RNAs, including one that inhibits transport of Alu-rich mRNAs from nucleus to cytoplasm (NEAT1) (41). A second noncoding RNA (HCG11) was highly significantly underexpressed in the flare group, is encoded in the major histocompatibility complex, and is uncharacterized with regard to function. Another differentially expressed transcript also is involved in regulation of mRNA transport from the nucleus (Midnolin) (42). Of potential importance was the observation of significantly decreased expression of MLH3, encoding a DNA mismatch repair protein that is involved in regulation of DNA recombination in meiosis and in immunoglobulin class switching and somatic hypermutation and may also bind to genomic repeat elements (43–46). Our interpretation of our pilot data is that the differentially expressed gene transcripts associated with lupus patients who go on to clinical flare may reflect a response to and/or basis of genomic events that generate potentially stimulatory nucleic acids. Altered expression of retrotransposon-bearing transcripts, impaired processing of DNA breaks, or altered regulation or transport of potentially stimulatory mRNAs, such as those rich in Alu sequences, from nucleus to cytoplasm might represent mechanisms that could result in an abundance of drivers that interact with cytoplasmic nucleic acid sensors and induce innate immune system activation. In that regard, our laboratory has studied the expression and stimulatory function of RNA encoded by Long Interspersed Nuclear Element−1 (LINE-1), and we have suggested that increased expression of those viral-like elements might represent one driver of type I IFN production (47). The differentially expressed transcripts identified in our study suggest future areas of investigation that might elucidate the mediators of innate immune system activation and type I IFN production observed in many lupus patients and provide candidate predictors of future lupus flare.

TABLE 1.

Candidate Predictors of Lupus Flare^a

Gene	Increase v Decrease	P Value	Proposed Function
APOBEC3B	Increase	.01	Inhibits LINE-1 retrotransposition
NEAT1	Increase	.0005	Long noncoding RNA; sequesters Alu-rich edited RNA in nucleus
Midnolin	Increase	.01	Expressed in ovary; may regulate mRNA transport from nucleus
MLH3	Decrease	.0006	DNA mismatch repair; mediates meiotic recombination; binds to genomic repeat elements
HCG11	Decrease	<.0001	Long noncoding RNA; function unknown

^aSelected gene transcripts significantly differentially expressed between flare and non-flare lupus patients are shown. Relative increased or decreased expression in the flare group, P value, and proposed function of the indicated gene transcript based on literature review are shown.

CONCLUSION

The identification of type I IFN as a central mediator of disease pathogenesis in patients with lupus has generated important new insights into the role of the innate immune response in the immune dysregulation, autoimmunity, and tissue damage that characterize disease in lupus patients. Mutations in genes encoding proteins that regulate nucleic acid degradation, cytoplasmic nucleic acid sensors, or components of the signaling pathways triggered by nucleic acids are associated with diseases that have characteristics similar to those of patients with lupus. Together, these observations suggest the hypothesis that alterations in genome integrity, nucleic acid metabolism or transport, or recognition of aberrant nucleic acids might represent significant mechanisms of immune system activation that contribute to lupus pathogenesis. Our pilot study aimed at identification of gene transcripts that are differentially expressed in those patients who go on to clinical flare support that hypothesis and identify gene candidates pointing to regulatory mechanisms that can be further investigated. Most intriguing is the potential for expression of virus-like genomic transposon elements to contribute to innate immune system activation and disease initiation or exacerbation in lupus. Through further studies, identification of the underlying molecular mechanisms that precede clinical exacerbations, allowing prediction of future flare, could lead to therapeutic interventions that prevent severe disease.

DISCUSSION

Rosenwasser, Kansas City: I am curious about the interferon alpha as being more than just a biomarker. Is it functionally active, and, therefore, would a monoclonal antibody as a biotherapeutic to interferon alpha be something to regulate here?

Crow, New York: I think that it is biologically active and important. Three weeks ago, I attended a meeting where I was charged with actually making that case, which I did citing quite a bit of human and mouse studies. So I think it is functionally important in lupus. You point out that there are studies ongoing using monoclonal antibodies to interferon alpha that have been not fully satisfactory. I think that what we are learning is that there is much more than interferon alpha that is included in that type 1 interferon pathway. I think that we probably have not sufficiently effectively targeted the interferons that are most relevant. Again, it is only a part of what I envision as being the mechanism.

Rosenwasser, Kansas City: I have one question about the intracellular regulation. You showed high- and low-risk IRF haplotype. Have you been able to stratify that to response to treatment or clinical characteristics at all?

Crow, New York: I am not aware of any data from the lupus literature that relates polymorphisms in susceptibility genes to responsiveness to therapy. But that is obviously something that would be important to do when we get a little further along.

Schuster, New York: I had a similar question about marker versus pathogenicity. Are there data on interferon alpha in the NZB or NZW mouse model? Does it modify the pathophysiology?

Crow, New York: Yes, actually the paper by Alfred Steinburg and Norm Talal that I mentioned back in 1969 was driving the production of type 1 interferon with polyinosinic-polycytidylic acid in the NZB/NZW model; disease was accelerated and exacerbated. In much more recent studies, interferon alpha delivered through an adenoviral vector accelerates disease and exacerbates disease. Type 1 interferon receptor deficient mice NZB — I think it's actually the BXSB model and to some degree the MRL/lpr model — have less disease in the absence of the interferon receptor. So I think there is growing data from the mouse literature that would support this also.

Stevenson, Palo Alto: You pointed out to us that lupus is predominately something that occurs in women, as do other autoimmune diseases. Not only are there cells from the fetus that are circulating in women during pregnancy, but there are also now ways that we can understand free DNA and free RNA. We can see the transcriptome of both the fetus and the mother. Could there be something going on very early in the way these women are responding to these challenges that might lead to some of the syndromes that you are seeing?

Crow, New York: Certainly lupus is seen in women who have never been pregnant. I guess they have their mother's cells available. So the idea of microchimerism I think has had its ups and downs. I don't favor it. In terms of the female susceptibility, it's just a giant wonderful challenging question out there. I don't think female hormones fully explain the female susceptibility. I mentioned to you at breakfast that interferon epsilon is now being recognized in female reproductive organs, which is very intriguing. Then in thinking about genome-integrity and genome-behavior in germ cells and in the ovary, there are very interesting events that deal with meiosis and methylation and demethylation that are intriguing in terms of perhaps producing stimulatory transcripts that might be driving this pathway.