microRNA-mediated regulation of innate immune response in rheumatic diseases

miRNAs have been shown to play essential regulatory roles in the innate immune system. They function at multiple levels to shape the innate immune response and maintain homeostasis by direct suppression of the expression of their target proteins, preferentially crucial signaling components and transcription factors. Studies in humans and in disease models have revealed that dysregulation of several miRNAs such as miR-146a and miR-155 in rheumatic diseases leads to aberrant production of and/or signaling by inflammatory cytokines and, thus, critically contributes to disease pathogenesis. In addition, the recent description of the role of certain extracellular miRNAs as innate immune agonist to induce inflammatory response would have direct relevance to rheumatic diseases.

Introduction miRNAs are small endogenous noncoding RNAs, discovered nearly two decades ago. Our understanding of the biological importance of miRNAs has grown exponen tially recently owing to the tremendous breakthrough in research in the last several years. Mature miRNAs exhibit robust regulatory roles in almost all biological pro cesses by modulating the expression of their target genes. Not surprisingly, emerging studies have demonstrated the active role of miRNAs in regulating the development and function of immune cells and the association of aberrant expression of miRNA with disorders of the immune system. In this review, we will fi rst discuss several noteworthy features and new fi ndings in biology of miRNA, then focus on the function of miRNA in regulating innate immune response, and, fi nally, touch the evidence of dysregulation of this process in connection with rheumatic diseases.

New fi ndings in miRNA biology
miRNA biogenesis and action processes are subject to dynamic regulation miRNA genes are prevalent in multicellular organisms. Th ese genes often form clusters encoding multiple mature miRNAs that cooperatively regulate the same mRNA target or functionally related targets [1]. Most miRNAs are transcribed by RNA polymerase II. Th e cell type-specifi c or spatiotemporal expression patterns of miRNAs are primarily determined at the transcriptional level [2]. Th e primary transcripts of miRNA genes are sequentially processed by two nucleases, Drosha and Dicer, whose activities are assisted by a number of other protein cofactors, to generate the ~22 nucleotide-long miRNA duplexes [1]. Regulation of the expression and activity of these miRNA processors during diff erent develop mental stages or in response to environmental stimuli thus represents an intriguing post-transcriptional control of the miRNA expression profi le that accommodates the needs of shaping protein expression in a given cell [2]. For instance, activation of the mitogen-activated protein kinase/extracellular signal-regulated kinase signaling pathway mediates phosphorylation of a critical partner of Dicer, TRBP, which leads to increased stability of the processing complex and enhanced miRNA production [3]: a component of the processing complexes that regu lates the biogenesis of a subset of miRNAs is the KHtype splicing regulatory protein (KSRP) [4]; the activity of KSRP can be induced by signaling in an innate immune system [5,6]. Once cleaved, the guide strand of the miRNA duplex gives rise to mature miRNA, which is prefer entially incorporated into the RNA-induced silencing complex (RISC) and guides the complex to its target mRNAs. Th e passenger strand of the miRNA duplex gives rise to the rarely expressed star-form miRNA (miRNA*); however, the ratio of miRNA to miRNA* can be dynamically controlled in response to stimuli. miRNA* species also exhibit important regula tory function [6,7]. In most cases, the miRNA:target interaction is primarily Abstract miRNAs have been shown to play essential regulatory roles in the innate immune system. They function at multiple levels to shape the innate immune response and maintain homeostasis by direct suppression of the expression of their target proteins, preferentially crucial signaling components and transcription factors. Studies in humans and in disease models have revealed that dysregulation of several miRNAs such as miR-146a and miR-155 in rheumatic diseases leads to aberrant production of and/or signaling by infl ammatory cytokines and, thus, critically contributes to disease pathogenesis. In addition, the recent description of the role of certain extracellular miRNAs as innate immune agonist to induce infl ammatory response would have direct relevance to rheumatic diseases. mediated by base-pairing of the miRNA seed region (nucleotides 2 to 7) to the 3' UTR of the mRNA, resulting in target degradation and/or trans la tional repression [1]. Such a short sequence signature of individual miRNAs is readily found in the genomic transcripts, endowing them with the potential to target dozens or even hundreds of diff erent mRNAs. Moreover, multiple miRNAs can bind to the same mRNA and co ordinate its expression [2]. In addition, several other features aff ect miRNA:target interaction, including local AU content, the position of the binding site in the 3' UTR, and pairing at the 3' end of miRNA [8].
Similar to manipulation of miRNA biogenesis by regulating the components of the processing complex, miRNA eff ects can be enhanced or attenuated by positively or negatively regulating the levels and activity of RISC components [2]. For instance, in response to stress, the mitogen-activated protein kinase/p38 kinase signaling pathway mediates phosphorylation of serine-387 in AGO2, the core component of RISC, increasing its recruitment to processing bodies [9]. Another intriguing regulation of miRNA function on specifi c mRNAs depends on the interplay between RISC and other RNA binding proteins. Th e AU-rich element binding protein HuR is required by let-7/RISC for effi cient inhibition of c-Myc expression [10], whereas in other stress conditions HuR relieves miR-122-mediated repression of CAT-1 mRNA by promoting dissociation of RISC from the target RNA [2,11]. Th e released mRNA is recruited to polysomes for active translation, suggesting miRNAmediated repression is reversible [11].

Mature miRNA polymorphisms
A great number of polymorphisms other than those transcribed from genetic variants have been identifi ed in mature miRNA species [12]. First, Drosha and Dicer cleavage of some precursor molecules is not uniform and yields miRNA isoforms with shifting termini [2]. Second, the double-stranded segments in many miRNA precursor molecules are subject to RNA editing (adenosine to inosine) by adenosine deaminases that act on RNA. Th ose that occur in the mature miRNA corresponding region alter the sequence of the fi nal miRNA product [12,13]. RNA editing may also aff ect protein binding and, thus, alter the processing effi ciency or nuclear export of miRNA precursors [13]. Th ird, deep sequencing has revealed an abundance of untemplated additions of adeno sine or uracil residues to the 3' end of miRNAs [12]. Such polymorphisms can greatly aff ect the stability of mature miRNAs, and may direct the miRNA to diff erent target mRNAs if located in the seed region [2]. New technologies, such as deep sequencing, will promote the identifi cation of functional isoforms originating from a given miRNA gene and that precise quantifi cation of their expression levels in both physiological and disease settings.

Expanding the rules of miRNA behavior
Deep sequencing also revealed numerous miRNA binding sites that reside in coding sequences [14], in line with previous sporadic reports of functional miRNA target sites within the protein-coding region [15,16]. miRNA targeting can also be mediated by 11 to 12 contiguous perfect match to the center of the miRNA [17]. Th ere are also reports on miRNAs functioning as activators of translation [18,19]. In a recent study, a role for miR-328 as an RNA decoy to directly bind hnRNP E2 was ascribed, thus interrupting the protein's binding capacity and regulatory function towards select mRNAs [20]. Interaction between miRNAs and long noncoding RNAs or transcribed pseudogenes has also been reported [21,22]; such transcripts, along with mRNAs [23,24], can reciprocally control the level and function of miRNAs by dynamic binding to the same miRNA [25]. A growing body of evidence also shows the existence of miRNAs in body fl uids, which may be transferable and functional in recipient cells [26,27]. However, more studies are warranted to fully understand the miRNA regulatory network in maintaining homeostasis and the implication in human diseases.

Regulation of innate immune response by miRNAs
Since the initial observation of distinct miRNA expression patterns across the hematopoietic lineage [28], extensive studies have established critical roles for specifi c miRNAs in immune cell development and, equally importantly, in regulating their function during both innate and adaptive immune response [29][30][31]. Particularly, cells of the innate immune branch, such as monocyte/ macrophages, dendritic cells (DCs), and granulocytes, constitute the fi rst line of defense against invading pathogens. Toll-like receptors (TLRs), which constitute the major pathogen detection system, initiate rapid signaling upon engagement in innate immune cells to induce the transcription of a set of infl ammatory cytokines, such as TNFα and type I interferons, and to subsequently prime specifi c adaptive immune responses. Engagement of other pattern-recognition receptors such as nucleotide-binding oligomerization domain-like receptors (NLRs) and subsequent infl ammasome assembly leads to caspase-1 activation and, hence, the production of IL-1β and IL-18 to mediate infl ammatory response. Discoveries of novel miRNA players in regulation of innate immunity continue to emerge (Table 1).

miRNA regulates innate immune cell development
A circuitry involving mutual repression between three miRNAs and a key protein regulator has been described to control monocyte diff erentiation. AML1, the key transcription factor for the process, is directly targeted by miR-17-5p, miR-20a and miR-106a. Expression levels of the three miRNAs are thus downregulated during monocytic diff erentiation, allowing for the accumulation of AML1, which, in turn, can transcriptionally repress the expression of these miRNAs and promote cell diff eren tiation [32]. By contrast, miR-424 is upregulated by PU.1, another crucial transcription factor for mono cyte/macro phage diff erentiation, and facilitates the expression of diff erentiation-specifi c genes by suppres sing the protein level of the inhibitory transcription factor NFI-A [33]. Another study reported an increase in IKKα protein level during monocyte/macrophages diff er en tiation owing to Endotoxin sensitivity and tolerance [53] miR-106a IL-10 AML1 Monocyte diff erentiation, IL-10 production [32,77] miR-125b TNFα TNFα production by macrophages [55] miR-142-3p IL-6 IL-6 production by DCs [76] miR-145 MAL TLR2-mediated and TLR4-mediated signaling [63] miR substantial decrease in the expression of miR-15a, miR-16, and miR-223, which contributes to p52 production and prevention of the new macrophage from becoming overactivated [34]. Monocytes can also diff erentiate into DCs (monocytederived DCs), and miR-21 and miR-34a were shown to be important for this process by cooperatively targeting the mRNAs encoding JAG1 and WNT1 [35]. In addition, a handful of miRNAs are diff erentially expressed across DC subsets and regulate their fate decision, as miR-221 and miR-222 expression favors conventional DC development, whereas inhibition of the miRNAs skewed precursor cells toward plasmacytoid dendritic cell (pDC) commitment [36]. miRNAs also play important roles in granulocytes and natural killer (NK) cell development [30,37]. miR-155 is enriched in hematopoietic stem cells compared with more mature hematopoietic cells; enforced expression of the miRNA in mouse bone marrow cells caused granulocyte/monocyte expansion [38]. However, when miR-146a is depleted, proliferation myeloid cell lineage is observed [39,40].

miR-146a as a crucial negative regulator of innate immune response
miR-146a and miR-155 are the fi rst miRNAs induced during immune activation and profoundly regulate the innate immune response. In human and murine innate immune cells, transcription of both miRNAs is induced by engagement of several TLRs and infl ammatory cytokines or upon viral infection, although the extent and peak induction time may vary [6,[41][42][43][44][45]. miR-146a eff ectively suppresses NF-κB activation and downstream cytokine production (for example, IL-8 induction by IL-1) by various stimuli via a negative feedback loop [41,46]. Th e signaling adaptor proteins TNF receptor-associated family (TRAF)-6 and IL-1 receptor-associated kinase (IRAK)-1 were identifi ed as direct targets of miR-146a [41]. Because these molecules are also critical signaling components for type I interferon production, miR-146a has also been shown to be able to dampen type I interferon induction by TLR7 and the intracellular sensor retinoic acid-inducible gene-I pathway [43,47]. In this context, the transcription factor interferon regulatory factor-5 and another adaptor IRAK2 were also proved to be miR-146a targets, respectively [43,47]. miR-146a thus inhibits the type I interferon production by simultaneously targeting multiple key components of the induction pathway. In addition, miR-146a suppresses the expression of type I interferon-inducible genes in Akata cell line [44] and in peripheral blood mononuclear cells (PBMCs) via targeting signal transducer and activator of transcription (STAT)-1 [43]. Since these classes of molecules play essential roles in receiving and interpretation of the activation signals, relatively small reductions in their levels could greatly aff ect the func tional response [48]. Th e overall regulation of the type I interferon induction and action by miR-146a via multiple targets can thus produce great impact, although the inhibitory eff ect towards each individual target may be modest. Consistent with the in vitro fi ndings, miR-146a-defi cient mice display chronic NF-κB activation and develop autoimmune-like disease in aging animals [39,40].

Dual role for miR-155 and opposing action of miR-155/ miR-155* in diff erent settings
Th ere is extensive evidence supporting miR-155 as a negative regulator of innate immune or infl ammatory response. Th e adaptor protein MyD88 and the kinase IKKε were identifi ed as potential targets of miR-155 [49][50][51]. In monocyte-derived DCs, miR-155 attenuates TLR/IL-1R infl ammatory pathway activation by directly targeting the signaling molecule TAB2 [52]. On the contrary, miR-155 promotes infl ammatory response of macrophages and also type I interferon signaling via direct inhibition of the canonical negative regulator suppressor of cytokine signaling-1 [45,53]. Another study confi rmed SH2 domain-containing inositol phosphatase-1 (SHIP1) as a direct target of miR-155; repression of endogenous SHIP1 by miR-155 resulted in increased activation of the kinase AKT during macrophage response to lipopolysaccharide (LPS) [54]. Concordantly, miR-155 transgenic mice produced higher levels of TNFα when exposed to LPS [55]. During acute infl ammatory response, induction of miR-155 leads to the suppression of CCAAT/enhancer binding protein beta, which may be responsible for the upregulation of granulocyte colonystimulating factor [56]. Th e proinfl ammatory role of miR-155 was also evident in atherosclerotic plaques, where specifi c expression of miR-155 in macrophages directly inhibits the transcriptional repressor BCL6, leading to the expression of the chemokine CCL2, and thus recruitment of monocytes to the infl ammatory site [57]. In addition, several other proteins have been identifi ed as miR-155 targets in DCs, such as PU.1 [58], KPC1 [59], and c-Fos [60], indicating that miR-155 regulates many aspects of DC biology (reviewed in [61]). Indeed, miR-155-defi cient DCs fail to eff ectively activate T cells [61,62], exemplifying a role for miRNA in regulating the priming of adaptive immune response.
Interestingly, although miR-155 and miR-155* originate from the same precursor, they display opposite eff ects on the regulation of type I interferon production by pDCs [6]. In the initial stage of pDC stimulation by TLR7 agonist, the transcriptional activation of the miR-155/ miR-155* gene leads to rapid production of mature miR-155* versus miR-155. Th is results in the degradation of its target IRAKM, a negative regulator that blocks TLR7 pathway activation in resting pDCs, thereby facili tat ing type I interferon production. Simultaneously, both TLR7 stimulation and the autocrine/paracrine signaling of IFNα/β lead to gradual accumulation and activation of KSRP, which promote miR-155 maturation at the post-transcriptional level. In the later stage of activation, therefore, expression of miR-155 dominates whereas miR-155* levels decrease sharply. Targeting TAB2 by miR-155 in pDCs confers negative regulation of the activation signaling, thus maintaining type I interferon pro duction and pDC activation at a proper level [6].

Other miRNAs targeting innate immune signaling molecules
miRNA-mediated regulation of other molecules in the TLR signaling cascades can also eff ectively control or fi ne-tune the innate immune response. miR-145 was shown to target the bridging adaptor MAL [63]. miR-21 inhibits the expression of proinfl ammatory regulator PDCD4 after TLR4 engagement [64]. A couple of signaling proteins in the TLR4 pathway were predicted as potential targets for miR-200 family members (miR-200a/ b/c); however, a reporter gene screening showed that only the MyD88 3' UTR was targeted by miR-200b/c, which was confi rmed by mutational analysis [65]. Consequently, overexpression of miR-200b/c inhibited NF-κB reporter activity and TLR4-induced infl ammatory cytokine expression [65]. CaMKIIα is a major downstream eff ector of calcium and plays an important role in promoting TLR signaling-induced DC maturation and function. Upregulation of members of the miR-148 family (miR-148a/b and miR-152) in DCs by several TLR agonists leads to targeted inhibition of CaMKIIα, which results in suppression of cytokine production, reduced MHC II surface expression and DC-initiated antigenspecifi c T-cell proliferation [66], demonstrating a role for miRNAs other than miR-155 in regulating antigenpresenting capacity of DCs.
In a recent study, the concomitant regulation of TAB2, TAB3, and IKKα by miR-23b was reported, which is responsible for the critical suppression of NF-κB activation and infl ammatory cytokine production induced by IL-17, TNFα, or IL-1β [67]. Importantly, miR-23b is down regulated by IL-17 stimulation [67]. Despite the recent research focus on IL-17 as a T-cell-secreted cytokine, many innate immune cell populations release a high amount of IL-17 at the early stage of an immune response, which is central to the initiation of IL-17dependent immune responses [68]. Downregulation of miR-23b by IL-17 would therefore play a profound role in regulating the innate immune activation.
miRNA-mediated regulation is also dictated to targeting transcription factors that are instrumental in generating an innate immune response, as exemplifi ed by miR-146a (targeting interferon regulatory factor-5 and STAT1) dis cussed above. In macrophages, downregulation of miR-223 by TLR3 and TLR4 agonists results in de repression of its target STAT3 and, thus, in enhanced production of IL-6 and IL-1β but not TNFα [69]. miR-9 is induced by LPS in both monocytes and neutrophils and directly targets NFKB1 mRNA, representing another important feedback control of NF-κB-dependent responses [70]. In the IL-10-driven anti-infl ammatory response, miR-187 was shown to be induced to downregulate the production of several infl ammatory cytokines by activated monocytes. One relevant target identifi ed for miR-187 is IκBζ, which is a key transcriptional regulator of IL-6 and IL-12p40 [71].
miRNAs have also been shown to directly target mRNAs encoding individual TLRs [72]. TLR4 expression is thus inhibited by let-7e in macrophages [53], whereas TLR2 is targeted by miR-19a/b [73]. Th e miRNAmediated regulation of TLR signaling pathways is summar ized in Figure 1.
In the context of NLR-mediated infl ammatory response, two very recent studies independently reported the direct regulation of NLR family PYD-containing protein 3 (NLRP3) and, consequently, inhibition of IL-1β production from the NLRP3 infl ammasome by miR-223 [74,75].

Direct targeting of cytokine mRNAs
In addition to the preferential regulation via signaling molecules, several cytokine mRNAs also fall into direct control by miRNA [72]. In addition to indirect suppression of IL-6 and IL-12p40 by miR-187 discussed above, this miRNA also directly inhibits TNFα mRNA expression and translation in monocytes [71]. Th e 3' UTR of TNFα mRNA also harbors a binding site for miR-125b, and downregulation of the miRNA by LPS stimulation may help stabilize TNFα expression [55]. IL-6 mRNA is directly targeted by miR-142-3p; silencing of miR-142-3p leads to enhanced IL-6 production both in immature DCs and following LPS activation [76].
Th e results of another study revealed direct inhibition of IL-10 expression by miR-106a [77]. Ma and colleagues found that NK cells activated in vivo (that is, from mice infected with an intracellular pathogen) or in vitro by the innate immune ligand poly(I:C) downregulated their expres sion of miR-29 while producing a large amount of IFNγ [78]. Th ey further showed a direct interaction between miR-29 and IFNγ mRNA: in addition to evidence that mutation of the predicted miR-29 binding sites abolished its inhibitory eff ect on IFNγ 3'-UTR reporter gene activity, the authors detected elevated association of IFNγ mRNA with the Ago2-containing complex in cells transfected with synthetic miR-29a mimic using an immunoprecipitation approach with an antibody against Ago2 [78]. Th e importance of miR-29 in regulating the immune response to intracellular bacterial infection (via targeting IFNγ) was further demonstrated in vivo by competitive inhibition of miR-29 by transgenic expres sion of a sponge target [78].

miRNAs take action in host-virus interaction
Compelling evidence demonstrates that miRNAs are directly incorporated into host-virus interactions, providing another layer to the innate immune response [31,79]. For instance, host cell miR-32 can recognize and bind to fi ve viral mRNAs, contributing to the repression of the replication of the retrovirus primate foamy virus type 1 [80]. In addition to the induction of a plethora of well-known antiviral proteins, IFNβ is also found to stimulate the expression of several miRNAs that target the genome of hepatitis C virus [81]. Simultaneously, IFNβ suppresses the expression of miR-122, a host miRNA that is utilized by hepatitis C virus to promote its replication.
In the context of viral exploitation of miRNAs for their own advantage, some virus-encoded miRNAs target host mRNAs to evade immune surveillance or dampen the immune response. For instance, hcmv-miR-UL112 represses the expression of histocompatibility complex class I-related chain B and consequently impairs NK cell activation and killing infected cells [82]. Th e Epstein-Barr virus-encoded miRNA, miR-BART15, represses the and other infl ammatory cytokines. miRNAs exert pronounced control of the pathway activation at multiple levels to ensure the generation of proper immune response. The miRNAs preferentially target the common signaling components and transcription factors, but also directly act on receptors and cytokine mRNAs. In most cases, decreases in the concentrations of miRNA target proteins achieve eff ective negative regulation and therefore avoid detrimental immune activation. However, if the target protein itself is a negative regulator (IL-1 receptor-associated kinase (IRAK)-M, suppressor of cytokine signaling-1 (SOCS1), SH2 domain-containing inositol phosphatase-1 (SHIP1)), miRNA-mediated regulation will facilitate TLR signaling and the production of infl ammatory cytokines. ERK, extracellular signal-regulated kinase; IFR, interferon regulatory factor; IKK, I-kappa-B kinase; MAL, MyD88 adapter-like; MAPK, mitogen-activated protein kinase; TAB, TAK1-binding protein; TAK, transforming growth factor-beta activated kinase; TBK, TANK-binding kinase; TRAM, TRIF-related adapter molecule; TRAF, TNF receptor-associated factor; TRIF, TIR domain containing adaptor inducing IFNβ. expression of NLRP3, and thus IL-1β production [74]. Furthermore, this viral miRNA can be secreted and transferred via exosomes to inhibit the NLRP3 infl ammasome capacity in noninfected cells [74]. Such direct interactions between virus-encoded and host-encoded nucleic acids provide another dimension to innate immunity [79]. Since viral infection has also been implicated in rheumatic disease onset or fl are and Epstein-Barr virus is considered a major environmental risk factor for systemic lupus erythematosus (SLE) [83], the involvement of miRNA in host-virus interaction may also have some relevance to rheumatic disease pathogenesis.

Dysregulation of miRNA and innate immune response in rheumatic diseases
It becomes evident now that miRNAs mediate dynamic regulation at multiple levels that essentially controls innate immune cell development and activation, infl am ma tory cytokine pro duc tion and signaling, and antigen presentation. Dysregulated miRNA expres sion or func tion could severely aff ect the duration and extent of innate immune response and be detri mental. Indeed, emerg ing data underscore the role of excessive or protracted innate immune signaling in the pathogeneses of autoimmune and autoinfl ammatory rheumatic diseases [83][84][85], which has been linked to dysregulation of critical miRNAs.

Systemic lupus erythematosus
SLE is a prototypical autoimmune disease with a hallmark of chronic immune activation and multiple immunologic alterations. To identify dysregulated miRNAs in SLE, a profi ling analysis of 156 miRNAs was undertaken to compare their expression levels in the peripheral blood leukocytes from patients with SLE and healthy subjects. Th is led to the identifi cation of underexpression of miR-146a in SLE [43], which appeared to be a primary defect caused by lupus-associated germline variant in miR-146a promoter [86], rather than a consequence of disease onset or medication [43]. In the same study, a reverse correlation of miR-146a levels with disease activity and with interferon score was identifi ed, which refl ects the magnitude of type I interferon pathway activation in patients with SLE. Th is indicates that decreased expression of miR-146a would result in inadequate regulation of the multiple target proteins and consequently over production of type I interferons and unabated down stream activation [43]. Importantly, enforced expression of miR-146a in PBMCs from patients with active SLE attenuated the mRNA levels of several interferon-inducible genes [43], strongly supporting the contribution of miR-146a dysregulation to such molecular phenotype of SLE.
Other miRNAs with a known role in innate immune response and dysregulated in SLE include miR-21, miR-142-3p, miR-148a/b, and miR-155, all of which are up regulated in PBMCs in patients with SLE [87,88]. However, the contribution of these miRNAs in SLE patho genesis via dysregulated innate immune response still needs to be determined. Stagakis and colleagues reported that the expression of the miR-21 target gene PDCD4 [64] is correspondingly decreased in active SLE, but suggested that their interaction aff ects aberrant T-cell responses in SLE in humans [88].

Rheumatoid arthritis
Rheumatoid arthritis (RA) is a systemic autoimmune disease that causes irreversible joint damage. Investigation of the expression and contribution of miRNAs in RA is very active and has revealed the dysregulation of several miRNAs in various cells/tissues, including PBMCs, the synovial tissue, isolated fi broblast-like synoviocytes (FLS), and cell-free synovial fl uid (reviewed in [89]). Th e joint resident cells, FLS are unique for RA in that they, like innate immune cells, express several TLRs, are implicated in infl ammatory response, and play critical roles in osteoarticular destruction [73]. Stimulation of RA FLS with LPS or bacterial lipoprotein strongly induced TLR2 expression while suppressing the levels of miR-19a/b, which directly targets TLR2 mRNA [73]. Supporting a role for miR-19a/b in regulating RA infl ammation, transfection of the miRNA mimic significantly downregulated the release of IL-6 and matrix metalloproteinase-3 by TLR2-activated RA FLS [73]. Secretion of IL-6 and matrix metalloproteinase-1 also appears to be indirectly regulated by miR-203, which is highly expressed in RA FLS [90].
Compared with osteoarthritis, miR-155 is signifi cantly upregulated in RA FLS, whose expression can be further induced by TNFα, IL-1β, and by ligands of TLR2 through TLR4. Th is indicates that the infl amed milieu may be responsible for the altered expression of miR-155 in these cells [91]. Moreover, miR-155 is also highly expressed in synovial fl uid-derived monocytes/macrophages com pared with the peripheral blood counterparts from patients with RA [91,92], whereas both mRNA and protein levels of the miR-155 target SHIP1 are decreased [92]. Incu bation of peripheral blood CD14 + cells with RA synovial fl uid stimulated the expression of miR-155 and release of TNFα; the cytokine production was abrogated by transfection of miR-155 inhibitor [92]. Direct evaluation of the regulation in RA synovial CD14 + cells revealed inhibition of miR-155 augmented SHIP1 expression and downregulated TNFα production when these cells were reactivated by LPS [92]. Moreover, the authors of this study and another group independently showed that miR-155 knockout mice did not develop collageninduced arthritis (CIA) where signifi cantly lower production of many proinfl ammatory cytokines was observed [92,93]. One should note here that, in addition to the essential regulation of monocyte/macrophage activation and of DCs in priming the adaptive immune response, miR-155 is also directly required for proper function of T cells and B cells [30,62]. Clearly, the protective role of miR-155 defi ciency in the CIA model resulted from the combinatory eff ect on both innate and adaptive immune responses [92,93].
miR-223 is also signifi cantly overexpressed in RA FLS and synovial fl uid [89,94,95]. Intriguingly, when a lentiviral vector expressing the miR-223 target sequence was intraperitoneally administrated to mice with CIA to abrogate miR-223 function, the severity of experimental arthritis was markedly reduced. Th is suggested a potential therapeutic strategy [95], although the extent and contribution of miR-223 silencing in distinct cell types needs detailed examination.
To identify novel miRNAs associated with RA pathology, Pandis and colleagues started with a diff erent approach. Th ey fi rst applied deep sequencing to examine the miRNA expression profi le of FLS isolated from the human TNF transgenic mouse model (TghuTNF) [96]. A number of dysregulated miRNAs were identifi ed, including miR-155 and miR-223 that are known to be upregulated in FLS of RA patients. Th e expression levels of select miRNAs were further quantifi ed in patient biopsies, and the upregulation of miR-221, miR-222 and miR-323-3p was also consistently found to be associated in human RA [96].
Interestingly, miR-23b was found to be underexpressed in RA synovial tissue and in the joints of mice with CIA, in the kidneys of patients with SLE and the MRL/lpr mouse model, and in experimental autoimmune encephalo myelitis mice, which may be a result of IL-17mediated transcriptional inhibition [67]. Th e results of this study provide in vivo evidence that miR-23b could suppress autoimmune disease pathogenesis, although the expression of this miRNA in resident cells in infl ammatory lesions appears to be vital in this regard [67].

Upregulated expression of miR-146a in rheumatoid arthritis, Sjögren's syndrome, and myositis
In contrast to the decreased expression of miR-146a in SLE, patients with RA display higher expression of miR-146a in both FLS [91,97] and PBMCs [98], with a hint that the alteration primarily occurs in monocytes/ macrophages [98]. Although miR-146a does target IRAK1 and TRAF6 mRNA for degradation [99], their expression in PBMCs in patients with RA is similar to that in healthy subjects [98]. Th e results of a recent study revealed overexpression of miR-146a in PBMCs of patients with Sjögren's syndrome, which is also observed in PBMCs and the salivary glands in an animal model of the disease [99]. In another study, increased expression of both miR-146a and miR-146b in patients with Sjögren's syndrome was observed [100]. In PBMCs from patients with Sjögren's syndrome, the mRNA level of IRAK1 is decreased while that of TRAF6 is increased when examined in a small number of subjects (n = 9 for patients and n = 10 for healthy subjects, respectively [100]). Altered miRNA levels were also identifi ed in patients with myositis; the expression of miR-146a was found to be elevated, probably due to leukocyte infi ltration [101]. Although the 31 patients with myositis examined displayed a general signature of type I interferon pathway activation, six out of eight patients with dermatomyositis examined had reverse correlation between miR-146a levels and type I interferon gene signature [101]. Further studies are warranted to explore the reason for increased expression of miR-146a and its contribution to such rheumatic diseases.

Scleroderma
Plenty of studies have also been performed by Ihn's group to identify scleroderma-associated miRNAs -several miRNAs including miR-29a and miR-196a were found to be dysregulated, either in skin biopsy or fi broblast samples, or in the serum from scleroderma patients [102,103]. Given that recent studies have provided new insights into the role of the innate immune system in scleroderma [104], the potential contribution of miRNAmediated dysregulation of innate immune response to scleroderma pathology is yet to be explored.

Extracellular miRNAs: a missing link between innate immune response and rheumatic disease?
Th e presence of miRNAs in body fl uids attracts a lot of attention. Th ere are examples of the extracellular miRNAs entering into and maintaining their regulatory function in recipient cells [26,27], but further evidence is awaited. Another major focus of current studies is the identifi cation of certain circulating miRNAs as disease biomarkers.
Nevertheless, the results of two recent studies suggested an unconventional role for miRNAs and an intriguing link between miRNAs in body fl uids and innate immune signaling in disease settings. Lehmann and colleagues identifi ed an increase in let-7b levels in the cerebrospinal fl uid from individuals with Alzheimer's disease and provided in vivo evidence that extracellular let-7 acts as an RNA ligand to activate neuronally expressed TLR7 and induce neurodegeneration [105]. Th e results of another study showed that miR-21 and miR-29a in the cancer cell-derived exosomes are able to bind murine TLR7 and human TLR8 and to induce a prometastatic infl ammatory response [106].
Although the exact structural features in the sequence of such miRNAs that confer their capacity to activate TLR7/8 require more studies, they all appear to harbor a GU-rich motif, which is known to be present in TLR7/8-stimulating virus-derived RNAs [105,106]. Support ing this idea, miR-599, miR-147, and miR-574-5p, which also contain GU-rich motifs, similarly induce TLR7/8-dependent cytokine production [105,106]. Given the more direct relevance of such an innate immune pathway to rheumatic diseases, and given the dysregulated miRNA levels in the body fl uids of patients with such disorders systemically (in serum) and/or locally (such as in RA synovial fl uid) [89], it would be interesting to examine such a link in a specifi c rheumatic disease.
Conclusion miRNA appears to preferentially target signaling proteins and transcription factors (Figure 1), molecules that are instrumental for dictating the extracellular stimuli and driving the development and activation of innate immune cells. Some miRNAs simultaneously regulate the expression of multiple proteins (for example, targeting of IRAK1, TRAF6, interferon regulatory factor-5, and STAT1 by miR-146a), thus eff ectively controlling the activation of innate immune signaling cascade. Some other miRNAs bind to the same site (for example, targeting of CaMKIIα by miR-148 family members) or separate ones (for example, targeting of IKKα by miR-15a, miR-16, and miR-223) within a single mRNA and coordinately control the expression of a common target. In addition, miRNAs also directly target mRNAs encoding innate immune receptors, such as TLR4 and TLR2, or infl ammatory cytokines, such as TNFα, IL-6 and IFNγ. miRNAs may also exert their regulation through inhibition of some relevant targets that are previously not linked to the innate immune response or that display an important function in other cellular pathways, and thus one might expect a careful analysis of putative targets to lead to the identifi cation of novel genes involved in some aspects of innate immunity or to provide a missing link between innate immune and other cellular pathways [107].
In many cases, the expression of specifi c miRNAs is upregulated via transcriptional activation (for instance, NF-κB-dependent induction) to decrease the concen tration of their target proteins. In other scenarios, the miRNA expression is downregulated to allow for the accumulation of its target (for example, decreased miR-29 expression with increased IFNγ production during NK cell activation). Th e post-transcriptional regulation adds another layer of control of miRNA expression during an innate immune response, enabling selective modulation of levels of certain mature miRNAs, and ensuring miRNA-mediated regulation to be exerted more precisely, as in the case of KSRP-promoted maturation of miR-155 during pDC activation. Molecules mediating signaling activation are not only targeted by miRNAs, but several negative regulators are also under miRNAmediated control (for example, targeting of suppressor of cytokine signaling-1 and SHIP1 by miR-155). miRNAs thus regulate innate immune response at multiple levels. Depending on the nature of the target proteins, miRNAs can either suppress or facilitate distinct aspects of immune activation, and ultimately maintain the balance of innate immune response. Altered expression of critical miRNAs, such as miR-146a and miR-155, thus profoundly contributes to the pathogeneses of rheumatic diseases, where dysregulation of their target proteins leads to unabated infl ammatory cytokine production and signaling, and aberrant priming of adaptive immune response. With the application of new technologies, such as deep sequencing, one would expect that more miRNAs or functional isoforms will be identifi ed to have a role in regulating innate immune response and dysregulation in rheumatic diseases. Th is would particularly provide insight into autoinfl ammatory disorders, where activation of the innate immune system alone is suffi cient to induce the disease [83].
On the contrary, although the importance of miRNAmediated regulation of innate immune response should be highly appreciated, one should notice that a considerable fraction of miRNAs discussed here are also critical regulators of adaptive immune response (for example, miR-155, which also regulates T-cell and B-cell function, as evidenced by knockout mice). Th is is particularly important in disease settings, because dysregulation of adaptive immune response is considered indispensable in the pathogenesis of autoimmune rheumatic diseases [83]; in many studies, altered expression of miRNAs in PBMCs or diseased tissue with leukocyte infi ltration, instead of purifi ed innate immune cells, was observed. Moreover, some miRNAs may even simultaneously regulate processes beyond immune system but essentially related to disease pathogenesis. For instance, miR-155, miR-223 and miR-21 promote osteoclastogenesis [93,[108][109][110] while miR-146a inhibits it [111]. Th e eff ect of these miRNAs on local bone destruction in RA has been demonstrated in animal models [93,111]. miRNA knockout mice would thus provide unambiguous evidence for the physiological and pathological roles of specifi c miRNAs in the innate immune system and in other processes.
One should still pay attention to the strategy applied, however, as both the target miRNA and its star form partner will be depleted. For example, it would be interesting to hypothesize whether there is any phenotype reported for miR-155 defi ciency actually attributable to loss of miR-155*. Th is query would require a thorough investigation of the distinct contribution of critical targets of each miRNA, as occurred for demonstration of targeting activation-induced cytidine deaminase by miR-155 in vivo [112,113]. With a better under standing of the contribution of dysregulation of miRNAs to the aberrant immune activation and, consequently, pathogeneses of rheumatic diseases, we would further explore the promise that miRNAs hold for developing new therapeutic targets.