MicroRNA-146a in autoimmunity and innate immune responses

Abstract

MicroRNA (miRNA) are approximately 22 nucleotide single-stranded RNA that regulate the stability of target messenger RNA by selective binding to specific sites at the 3′-untranslated regions (UTR). This triggers repression in translation and mRNA degradation. It has been estimated that approximately 60% of all mRNA are under the control of miRNA. Among the known hundreds of miRNA, some are considered master regulators controlling either a single or multiple cellular pathways. Some miRNA are known to affect development and cell differentiation, while others are implicated in immunity and autoimmune diseases. A very interesting example is miR-146a, which has been reported to be downregulated in systemic lupus erythematosus and upregulated in rheumatoid arthritis (RA). Several groups have recently focused their attention on miRNA in the pathogenesis of RA. Interestingly, the expression of miR-146a is upregulated in different cell types and tissues in RA patients. miRNA in RA could also be considered as possible future targets for new therapeutic approaches. This discussion will focus on the current understanding in the function of miR-146a in endotoxin tolerance and cross-tolerance, and how it may contribute to modulate the overproduction of known pathogenic cytokines, such as tumour necrosis factor α.

MICRORNA FUNCTIONS IN GLOBAL REGULATION

Gene expression starts from chromosomal DNA, which encodes genes of interest. Regulation by transcription factors, RNA polymerases and other regulatory factors controls transcription. The derived RNA transcripts mature by splicing and are transported to the cytoplasm to become mature messenger RNA and these steps are under global regulation.¹ Translation machinery then utilises mRNA as templates for protein synthesis. Transcriptional controls are thus logically considered the key to understand gene expression. However, once transcription is initiated, the next important control may depend on the stability of mRNA. If certain mRNA has a very long half-life, this allows continuous production of functional proteins no longer under the control of transcription. The regulation of mRNA half-life can be critical especially for mRNA that encodes cytokines, chemokines, or other polypeptides with significant biological functions affecting other cell types, as their overexpression can have detrimental effects on other cells and even the entire organism. Studies of AU-rich elements in the 3′-untranslated regions (3′UTR) have shown for many years their roles in fast degradation of many cytokine mRNA.² During the past 10 years, small RNA of approximately 20–25 nt are known to interfere with RNA stability by binding to 3′UTR leading to translational silencing and mRNA decay.^3–5 This RNA interference (RNAi) is now recognised as a major gene regulation mechanism in controlling the half-lives of specific mRNA. This class of small RNA is known as microRNA (miRNA). In fact, small interfering RNA (siRNA) can be designed as a complementary sequence to the coding region of mRNA, and this has become the most common method of gene knockdown in vitro whenever it is necessary to demonstrate the function of a given gene.

miRNA are important to help selectively control mRNA half-lives; all important mRNA are likely to be controlled such that when their function is not needed, their mRNA are selected for decay by miRNA-mediated pathway. The importance in RNAi is that siRNA, miRNA and anti-miRNA⁶ can become sequence-specific gene targeting agents for the development of therapeutics. There are clearly some concerns on the selectivity using these strategies but advances are continuing and new developments are available. For example, Obad et al⁷ show a new technique to inhibit the activity of entire miRNA families making use of eight nucleotide oligomers with a locked nucleic acid (LNA)-modified backbone, which they refer to as tiny seed-targeting LNA. These oligonucleotides are synthetised with a phosphorothioate backbone, which confers higher stability to the miRNA–LNA duplex. In a subsequent report, the overexpression of miR-21 in B and Tcells in B6.Sle123 lupus mice is targeted by in-vivo silencing using a tiny seed-targeting LNA. The silencing of miR-21 in vivo ameliorates autoimmune splenomegaly.⁸

CYTOPLASMIC GW BODIES LINK TO RNAi ACTIVITIES

Our laboratory has historically focused on human autoantibody research and it is from these studies that we began to work with miRNA and RNAi. Around the year 2000, our collaborator, Professor Marvin J Fritzler of the University of Calgary, became particularly interested in autoantibodies to some novel cytoplasmic foci. Autoantibody staining of HEp-2 cells displaying distinct cytoplasmic foci varied in number and size from cell to cell (figure 1A). Based on our experience in studies of anti-nuclear antibodies over the years, it was demonstrated quickly that these foci were distinct from other known subcellular structures including lysosomes, endosomes and the Golgi complex.⁹ A complementary DNA expression cloning project was initiated using these autoantibodies as probes to identify autoantigens that were enriched in these foci. A partial cDNA was cloned from an expression library and eventually the full-length cDNA encoded a protein of 182 kDa. The deduced protein was named GW182 because of the predicted molecular size and the discovery of a unique distribution of glycine (G)-tryptophan (W) dipeptide throughout the polypeptide.⁹ These foci were provisionally named GW bodies or GWB.⁹ Two experiments were conducted to show that the putative cDNA for GW182 was in fact a genuine marker for GWB. First, rabbit polyclonal antibodies, and later mouse monoclonals, raised to the recombinant protein were shown to recognise the same foci in co-staining experiments. Second, transfection of green fluorescence protein-fused GW182 cDNA into HeLa cells showed localisation to the same foci.⁹ The number and size of GWB were later demonstrated to be cell cycle dependent. GWB disassemble before mitosis and reassemble after the early G1 phase.¹⁰ The number and size of GWB increase from G1, S, to G2 cells. GWB in late S/G2 cells are the most abundant and brightest in intensity,¹⁰ as shown in figure 1A. Using immunogold electron microscopy, GWB are electron-dense cytoplasmic structures 100–300 nm in diameter and devoid of a lipid bilayer membrane.^9,10

Figure 1.

Structure and function of GW bodies (GWB) in health and disease. (A) Cell cycle-dependent expression of cytoplasmic foci GWB. Human prototype serum recognises GWB (green) in typical anti-nuclear antibody assay using HEp-2 cells. Nuclei are counterstained blue by DAPI. GWB are largely absent in early G1 cells, increase in number and size through the cell cycle, and are most abundant in late S/G2 cells. (B) Current known functions of GWB established for miRNA and siRNA. Transfected miRNA and siRNA are enriched or localised to these foci. Formation of GWB required miRNA biogenesis as cells interrupted at the DGCR8/Drosha processing step are largely depleted of these foci. Demonstrated activities correlated with increases in the number and size of GWB are shown. GWB enriched in Ago/GW182 complexes are clearly critical for translational repression and these foci may serve as transient storage of miRNA-mediated repressed mRNA. Currently both miRNA and siRNA pathways target mRNA, eventually leading to mRNA decay, which can take place within these foci as they are also enriched in certain specific nucleases. (C) Changes in expression of miRNA have been shown in many disease states. For example, underexpression of miR-146a in systemic lupus erythematosus may be responsible for overactivation of the type I interferon signalling pathway. Although supporting data are not yet available, changes in miRNA expression may have a significant effect on the presentation of self-antigens, such as Ago and GW182, leading to autoantibody production. References are shown in parentheses. LPS, lipopolysaccharide. This figure is only reproduced in colour in the online version.

GWB are associated with mRNA degradation and silencing

One of the first colocalisation studies showed that GWB are enriched in mRNA decapping enzymes Dcp1 and hLSm4.¹¹ This gave significant clues to the role of GWB in mRNA degradation. At about the same time, Dcp1-containing bodies were reported in yeast and named processing bodies or P bodies. Our laboratory showed that these were largely the same foci in mammalian cells, although there was some heterogeneity among these foci labelled by GW182 or Dcp1.¹² The function for GW182 remained unknown then. The next strategy taken to explore the function of GW182 was to determine its interaction partner proteins with the goal that known functional pathways could be identified. In brief, a 100 kDa polypeptide was shown to be co-immunoprecipitated with GW182 in HeLa cell lysate. Using mass spectrometry, the 100 kDa polypeptide interaction partner was identified as eIF2C2 or Argonaute2 (Ago2).¹³ As discussed below, Ago2 is the main polypeptide that binds and presents miRNA in a functional context primarily to recognise sequence elements in 3′UTR of its target mRNA.

Relationship of GWB and RNAi pathways

RNAi is known to involve either siRNA or miRNA pathways.¹⁴ The siRNA pathway requires the processing of cytoplasmic double-stranded RNA molecules by endonuclease Dicer to approximately 20–23 base pair siRNA duplex, which then bind Ago2 and are incorporated into the so-called RNA-induced silencing complex (RISC). The anti-sense strand siRNA-loaded RISC mediates recognition of cytoplasmic mRNA with sequence complementarity. The targeted mRNA is eventually degraded by further recruitment of 5′ to 3′ mRNA decay factors that include the deadenylase Ccr4, decapping factors (LSm1-7 ring, Dcp complex), and the 5′ to 3′ exonuclease Xrn1. siRNA generates siRNA–mRNA–protein complexes that are clearly components of GWB. As the process intensifies, smaller GWB aggregate with additional ribonucleoprotein complexes to become larger GWB. It is known that fluorescent tagged siRNA transfected into HeLa cells are localised to GWB.^13,15 Functional siRNA activity, requiring the presence of both siRNA and its target mRNA, correlates with increases in both the number and size of GWB.¹⁶ This siRNA-mediated increase in the number and size of GWB is abolished in cells with either Ago2 or GW182 knockdown, consistent with the increase correlating with siRNA activity.

In contrast, the miRNA pathway involves the transcription of miRNA genes encoded in chromosomes—currently approximately 1000 human miRNA genes are known—as primary miRNA transcripts, which are processed in the nucleus by the Drosha-DGCR8 microprocessor complex into approximately 80 nucleotide-long hairpin RNA known as pre-miRNA.¹⁴ These pre-miRNA are transported to the cytoplasm by exportin-5 and further processed by Dicer into mature miRNA of 20–23 base pairs.¹⁴ The matured miRNA is then loaded on Ago-RISC similar to the siRNA. The miRNA-loaded RISC then recognises elements in 3′UTR of target mRNA, leading to translational repression and delayed degradation or controlled degradation. Similarly, transfected fluorescent-tagged miRNA are highly enriched in GWB.¹⁷ When either Drosha or DGCR8 is knocked down in HeLa cells, thus inhibiting the primary miRNA processing step, mature miRNA and GWB are both depleted.¹⁷ Note that GWB reassemble when these knockdown cells are transfected with miRNA surrogates.¹⁷ The formation of GWB is thus a consequence of miRNA genesis.¹⁷

GWB are highly enriched in autoantigens

There are many known GWB components, including miRNA, mRNA, RNAi factors such as Ago2 and GW182 and RNA decay factors. Many of these factors are known human autoantigens.¹⁸ The most common clinical presentations for patients with anti-GW182 antibodies are neurological symptoms (ie, ataxia and motor and sensory neuropathy, 33%), Sjögren’s syndrome (31%), with the remainder being other diagnoses that include systemic lupus erythematosus (SLE), rheumatoid arthritis (RA) and primary biliary cirrhosis. Interestingly, approximately 50% of patients with anti-GWB antibodies also have autoantibodies to Ro52.¹⁹ Another GWB component Ge-1 is probably the most common autoantigen target of anti-GWB sera.¹⁹ The fact that GWB are enriched in ribonucleoprotein complexes is highly consistent with known characteristics of many other known autoantigens, such as small nuclear RNP (Sm and RNP), SS-A/Ro and SS-B/La, in systemic rheumatic diseases.^20,21 As Ago and GW182 are currently considered the core proteins of RISC,³ it is appropriate to recognise that RISC is a main target of human autoantibodies.

ABERRANT EXPRESSION OF miR-146a IN RA

Elevated expression of miR-146a and miR-155 in RA synovial fibroblasts and tissues has been reported, implicating their role in disease pathogenesis.^22,23 Our own data show increased expression of miR-146a in peripheral blood mononuclear cells (PBMC) of RA patients.²⁴ In contrast, Tang et al²⁵ showed decreased expression of miR-146a in PBMC of SLE patients and postulated that the expression of miR-146a is inversely correlated to type I interferon production. The difference in gene expression between SLE and RA is not unexpected.²⁶ More recent studies have validated the RA data and additionally shown that miR-146a is overexpressed in RA plasma,²⁷ CD4²⁸ and T-helper type 17 cells.²⁹ Table 1 summarises all of these reports to date. A recent report by Abou-Zeid et al³⁰ has basically completely reproduced our 2008 work demonstrating the elevated level of miR-146a in PBMC of 70 RA patients compared to 60 controls. Their data show that miR-146a expression is significantly higher in patients with RA than in those with osteoarthritis (n=45) and in controls (p<0.0001).³⁰ In patients with RA, miR-146a positively correlates with tumour necrosis factor (TNF) α (p=0.0003) and disease activity monitored by the erythrocyte sedimentation rate (p=0.022) and the disease activity score in 28 joints (p=0.009).³⁰

Table 1.

Elevated levels of miR-146a in blood and tissue cells from RA patients

	MiRNA-146a
RA samples	Samples	Increased expression (fold change)	References
Blood cells	PBMCs	2.6	24,30
	Plasma	NR	27
	Th17	1.43	29
	CD4 T cells	>2	28
Articular tissues	FLS	4	23
	FLS	NR	22

FLS, fibroblast-like synoviocytes; NR, exact fold change not reported; PBMC, peripheral blood mononuclear cells; RA, rheumatoid arthritis.

GWB ASSEMBLY CORRELATES WITH MIRNA ACTIVITY IN LIPOPOLYSACCHARIDE-STIMULATED HUMAN MONOCYTES

As GWB are important to RNAi functions, miRNA are linked to vital cellular processes, and autoantibodies to GWB are present in subjects with autoimmune diseases, it is reasonable to examine changes of GWB associated with immune stimulation. The model chosen is the classic lipopolysaccharide stimulation of human monocytes. The increase in the number and size of GWB after the addition of lipopolysaccharide to human monocytic cell line THP-1 is examined. The number of GWB is highest at 8 h after lipopolysaccharide stimulation compared to unstimulated controls. This implies a role of miRNA-mediated activity during lipopolysaccharide stimulation.³¹ In 2006, miR-146a, miR-132 and miR-155 were reported to be upregulated in response to lipopolysaccharide treatment of human monocytes.³² Further analysis of miR-146a induction by lipopolysaccharide reveals that this induction was nuclear factor κB (NF-κB) dependent and the mRNA targets of miR-146a include interleukin (IL)-1 receptor-associated kinase (IRAK1) and TNF receptor-associated factor 6 (TRAF6), each a key component involved in amplifying responses in Toll-like receptor (TLR)-4 signalling pathways.³²

CRITICAL FUNCTION OF miR-146a IN ENDOTOXIN TOLERANCE

As stated above, the implication of miR-146a in autoimmunity is apparent and perhaps it is important to examine its biological function further. Lipopolysaccharide is one of the potent stimulators of monocytes and macrophages in innate immunity, and it induces the production of a diverse array of inflammatory mediators, including TNFα both in vitro and in vivo.³³ To observe TNFα production in vitro, the most commonly employed monocytic cell line THP-1 is used.³⁴ Figure 2 summarises our current model of the critical role of miR-146a in lipopolysaccharide tolerance in THP-1 cells. Lipopolysaccharide binds to the lipopolysaccharide-binding protein, which in turn is coupled to CD14 on the cell surface of monocytes (figure 2A). Subsequently, lipopolysaccharide-CD14 interacts with TLR4 and forms a complex with another accessory protein MD-2. The TLR4 signalling cascade is initiated after binding with adaptor protein MyD88. A recent crystal structure study shows that the MyD88-IRAK4-IRAK2 complex, or Myddosome, reveals a left-handed helical oligomer that consists of six MyD88, four IRAK4 and four IRAK2 domains.³⁵ Assembly of this helical signalling tower is hierarchical; MyD88 recruits IRAK4 and the MyD88-IRAK4 complex recruits the IRAK4 substrates IRAK2 or the related IRAK1. Formation of these Myddosome complexes brings the kinase domains of IRAK into proximity for phosphorylation and activation of TRAF6. This chain of events triggers the activation and translocation of NF-κB, resulting in the transcription of immune-responsive genes and cytokines, such as TNFα and miR-146a.³²

Figure 2.

The role of miR-146a in the control of lipopolysaccharide (LPS) Toll-like receptor 4-mediated signal transduction in tolerised versus untolerised THP-1 cells. See text for details. IRAK, interleukin 1 receptor-associated kinase; NF-κB, nuclear factor κB; TNFα, tumour necrosis factor α; TRAF, tumour necrosis factor receptor-associated factor. This figure is only reproduced in colour in the online version.

Neutrophils and monocytes from sepsis patients are refractory to subsequent lipopolysaccharide challenge and no longer produce these cytokines.³⁶ This phenomenon is referred to as endotoxin tolerance, and is also a mechanism to prevent over-stimulation from the continuous exposure to the same danger signals in the environment. Endotoxin tolerance has been established for decades in vivo,³⁷ and has also been extensively investigated in vitro using primary monocyte/macrophage cells and cell lines.^38,39 To understand the endotoxin tolerance mechanism, changes of cell surface molecules, signalling proteins, pro-inflammatory and anti-inflammatory cytokines and other mediators have been studied. Despite intense investigations for decades into the hyporesponsiveness associated with innate immune cells in response to lipopolysaccharide priming, there is still no consensus on the primary mechanism responsible for its development.⁴⁰ Therefore, low-dose lipopolysaccharide-primed THP-1 cells produce TNFα rapidly and continue to do so for 4–6 h, then, as soon as regulatory miR-146a starts to increase, TNFα production decreases (figure 2B). At 18 h post-priming, a profound difference between miR-146a expression and TNFα secretion is established and, because the upregulated miR-146a acts negatively on IRAK1 and TRAF6 mRNA, the cells become tolerised (figure 2B) and do not respond to high-dose lipopolysaccharide challenges due to the high level of miR-146a, unlike the untolerised control that at this stage is responsive to lipopolysaccharide (figure 2C). It is well known that many regulatory genes (interferon regulatory factors⁴¹ and inhibitors IRAK-M, A20 and suppressor of cytokine signalling-1 (SOCS1))^42,43 participate in these pathways. To rule out other players in the complex TLR signalling pathway, the key points from our study⁴⁴ are that upregulation by transfection of miR-146a alone can mimic lipopolysaccharide priming to induce tolerance, and knockdown of miR-146a by transfection of specific antagomir (anti-miRNA) diminishes lipopolysaccharide tolerance. The findings thus fully support the dominant role of miR-146a in lipopolysaccharide tolerance.⁴⁴ This miRNA is highly upregulated in tolerised cells and acts as a tuning mechanism to prevent an overstimulated inflammatory state. It is interesting to speculate that modulating the level of miR-146a may be useful in therapeutic interventions for inflammation and protection against sepsis.⁴⁵

Beside homologous tolerance, lipopolysaccharide priming of the immune cells results in diminished cytokine response after subsequent stimulation with non-lipopolysaccharide heterologous TLR ligands.^46–49 This is known as lipopolysaccharide-induced cross-tolerance and has also been observed in cells from patients with sepsis.⁵⁰ Similarly, other TLR ligands such as peptidoglycan and flagellin, plus cytokines such as TNFα or IL-1β, have been shown to induce homologous tolerance in monocytes/macrophages and they can substitute for each other and sometimes mediate cross-tolerance both in vitro and in vivo.⁵¹ Lipopolysaccharide-induced tolerance and/or cross-tolerance are thought to play a broader role in host innate immunity. Because IRAK1 and TRAF6 are not only used by TLR4 for signalling, but also by other TLR, such as TLR2, 5, 7, 8 and 9, and the IL-1β receptor,⁵² they are considered the common and central adaptor kinases and, should their activity be diminished, cellular refractoriness may happen to other TLR (except for TLR3) signalling. This leads to the speculation that increased miR-146a expression during a lipopolysaccharide-primed state might play a part in a negative feedback pathway for other ligand–TLR interaction. Considering the ability of miR-146a to regulate IRAK1 and TRAF6, shared by all TLR (except TLR3), we show that miR-146a is involved in endotoxin-induced cross-tolerance against various microbial cargo sensed by other TLR. Our findings suggest that overexpression of miR-146a contributes to controlling pro-inflammatory cytokine production and confers cross-tolerance in innate immune cells, thus modulating our innate immunity to evade recurrent similar or different bacterial infections or both.⁵³

CONCLUSIONS

miR-146a is a dominant negative regulator of NF-κB activation primarily targeting IRAK1/TRAF6 of the MyD88-dependent pathway. Its ability to interfere with TLR ligand-induced cross-tolerance implies that this miRNA may have other targets that also affect MyD88-independent pathways but this will need to be examined in future studies. As its level has been consistently reported to be upregulated in RA, future studies will need to dissect its putative role in disease pathogenesis. It is intriguing that there is already a report of improvement on the inhibition of miR-146a in experimental therapy in a murine disease model.⁵⁴ However, more work is needed to understand the biological significance of elevated miR-146a in diseases as this does not appear to be specific to RA. An increase in level of miR-146a has also been reported in osteoarthritis,⁵⁵ Sjögren’s syndrome^56,57 and other diseases.^58–60 It should be noted that miR-146a has also been reported to play important roles in acquired immunity,⁶¹ and its level can be upregulated in certain cancers.^62–64