Modulating Inflammation through the Negative Regulation of NF-κB Signaling

Abstract

Immune system activation is essential to thwart the invasion of pathogens and respond appropriately to tissue damage. However, uncontrolled inflammation can result in extensive collateral damage underlying a diverse range of auto-inflammatory, hyper-inflammatory, and neoplastic diseases. The NF-κB signaling pathway lies at the heart of the immune system and functions as a master regulator of gene transcription. Thus, this signaling cascade is heavily targeted by mechanisms designed to attenuate overzealous inflammation and promote resolution. Mechanisms associated with the negative regulation of NF-κB signaling are currently under intense investigation and have yet to be fully elucidated. Here, we provide an overview of mechanisms that negatively regulate NF-κB signaling through either attenuation of signal transduction, inhibition of post-transcriptional signaling, or interference with post-translational modifications of key pathway components. While the regulators discussed for each group are far from comprehensive, they exemplify common mechanistic approaches that inhibit this critical biochemical signaling cascade. Despite their diversity, a commonality among these regulators is their selection of specific targets at key inflection points in the pathway, such as TRAF family members or essential kinases. A better understanding of these negative regulatory mechanisms will be essential to gain greater insight related to the maintenance of immune system homeostasis and inflammation resolution. These processes are vital elements of disease pathology and have important implications for targeted therapeutic strategies.

Summary Sentence:

Review of how the NF-kB pathway is regulated to attenuate inflammation and promote resolution

1. Introduction

Inflammation can be described as a “double-edged sword”. On one side, inflammation is a robust and essential biological process that is necessary to protect the host from immediate threats and initiate long term protective immunity. Conversely, dysregulated inflammation creates highly chaotic conditions in the host and results in excessive tissue damage that ultimately underlies the pathobiology of inflammatory and autoimmune diseases. To date, a large number of studies have focused on deciphering the mechanisms associated with the initiation and propagation of inflammation and inflammatory signaling cascades. Following the initiation and propagation of inflammatory signaling cascades, dynamic mechanisms further ensue that have evolved to reign in overzealous immune responses, promote the resolution of inflammation, and ensure a return to immune system homeostasis once threats to the host have been eliminated.

Pattern recognition receptor (PRR) signaling is a critical component of immune system activation and is typically targeted by mechanisms designed to shut down inflammation and promote resolution. Broadly, PRRs are located within the cell cytosol, embedded in the cell membrane, or secreted extracellularly. Soluble extracellular PRRs include members of the complement system and the pentraxins that bind to phosphocholine in both a calcium dependent and independent manner [1, 2]. Membrane-associated receptors include the Toll-like receptor (TLR) and C-type lectin receptor (CLR) families, whereas cytosolic receptors include the Nod-like receptor (NLR), Rig-I-like helicase (RLR), and AIM2-like receptor (ALR) families (Figure 1). Recognition of a pathogen associated molecular pattern (PAMP) or damage associated molecular pattern (DAMP) by a particular PRR leads to rapid pleotropic cellular responses. These cellular responses are driven by a variety of interconnected, coordinated, and complex biochemical signaling pathways, including those driven by the NF-κB and IRF transcription factors. The broad and robust biological effects of these pathways both initiate and sustain inflammation and are essential components of the balanced host immune response. This includes the release of pro-inflammatory cytokines and chemokines that facilitate either cell death or proliferation, the transcription of regulatory elements that silence target genes regulating hematopoiesis and lymphopoiesis, and ultimately the activation of the adaptive immune system.

Figure 1: Convergence of Pattern Recognition Receptor Families on NF-κB Signaling.

The PRR families include membrane associated C-type lectin receptors (CLRs) and Toll-like receptors (TLRs), and the cytosolic receptors Nod-like receptors (NLRs), AIM2-like receptors (ALRs), and RIG-I-like receptors (RLRs); which all play a role in NF-κB signaling. CLRS, such as Dectin-1, recruit the adaptor Syk to their cytoplasmic domain upon binding a ligand at the Carbohydrate Recognition Domain (CRD). Upon ligand binding, TLRs hetero- or homo-dimerize, resulting in a conformational change in the receptor which leads to the recruitment of adaptor molecules MyD88, Mal, TRIF, and/or TRAM to their TIR domain. RLRs, such as RIG-I, have two N-terminal CARD domains and recruit MAVS via CARD-CARD interaction to activate NF-κB signaling upon recognition of RNA viruses. The NLRs are comprised of a central NACHT domain, LRR, and either CARD or PYD, or both in the case of NLRP1; ALRs such as AIM2 also possess a PYD. NLRs including NOD1 indirectly stimulate NF-κB, but the predominant role of NLRs and ALRs in NF-κB signaling is processing the output pro-IL-1β and pro-IL-18 into mature IL-1β and IL-18 by the multi-protein complex known as the inflammasome, which is dependent on the adaptors ASC and Caspase-1.

The intracellular signaling cascades associated with PRR signaling are subject to a stringent set of checks and balances that are necessary to avoid excessive inflammation. These negative regulatory mechanisms are typically self-generated and occur through negative feedback loops to restore homeostasis and promote resolution. In this review, we focus our discussion on a selection of mechanisms that converge to negatively regulate NF-κB signaling. As a master regulator of gene transcription, targeting NF-κB signaling is a highly effective strategy used by the cell to modulate significant and diverse biological processes, including inflammation. Indeed, there is a reinvigorated interest in better defining mechanisms associated with the negative regulation of NF-κB signaling, as evidenced by the increasing number of recent publications and reviews on this topic. For example, an excellent review by Afonina et al. was recently published that focused on the negative regulation of NF-κB signaling and the NLRP3 inflammasome [3]. Here, we do not significantly cover NLRP3 inflammasome regulation or modulation. However, we do discuss several similar post-transcriptional and post-translational mechanisms associated with NF-κB regulation. We further expand our discussion to provide a more extensive overview of both canonical and non-canonical NF-κB signaling and focus on additional mechanisms associated with pathway regulation not extensively included in other previous reviews on this topic. Together, we believe that the current review is an excellent complement to previous works focused on the regulation of NF-κB signaling and the individual molecules discussed. In the course of assembling this review, we identified over 200 proteins that have the ability to target and attenuate NF-κB signaling. This high level of regulation is likely due to the vast and essential biological functions associated with NF-κB signaling in different cell-type and temporal specific situations, and serves to underscore this particular pathway’s importance and complexity. For brevity, we chose to focus this work on several archetypical mechanisms that broadly regulate this signaling cascade at the transcriptional, post-transcriptional, and post-translational levels in order to provide a concise overview of a selection of major regulatory elements.

1.1. Inflammation and NF-κB signaling

NF-κB is an evolutionarily conserved transcription factor found in species ranging from Drosophilia to humans, which underscores its critical role in the host immune response. The last 3 decades of research have contributed greatly to our understanding of NF-κB. NF-κB functions as a prominent inducible transcription factor that regulates the immune system [4]. Due to its diverse and broad biological functions, strict regulation of NF-κB signaling is paramount to proper tissue allostasis. When NF-κB signaling is aberrant, several maladies can ensue, such as susceptibility to infections, autoimmunity and cancer [5, 6]. Thus, as with many major signal transduction pathways, several mechanisms tightly regulate NF-κB signaling and maintain the proper balance of activation and repression.

NF-κB in mammals consists of a total of five proteins that are predominantly present in an inactivated state in the cytosol as either homo- or hetero-dimers that include: RelA (p65), RelB, c-Rel, p105 and p100. The p105 and p100 proteins are unique because they must undergo post-translational processing through the proteasome to form the active subunits, p50 and p52, respectively [4]. All isoforms contain a common Rel homology domain, which is responsible for DNA binding and binding to the cytosolic inhibitory proteins, termed Inhibitor of –κB (IκB) [7]. Several families of cellular receptors signal through NF-κB to initiate gene transcription and are known to coordinate signaling through either the canonical NF-κB pathway or the non-canonical NF-κB pathway (which is also termed the “alternative” NF-κB pathway). Molecules that signal via the canonical NF-κB pathway include cytokines, such as tumor necrosis factor (TNF), interleukin-1 beta (IL-1β), and the majority of PRRs [4]. The non-canonical pathway is initiated by a much smaller repertoire of molecules that are members of TNF family, such as CD40, B-cell activating factor (BAFF), and lymphotoxin beta (LT-β) [8].

1.2. The Canonical NF-κB Signaling Pathway

Convergence on the canonical NF-κB signaling pathway occurs through the activation of several different families of receptors (i.e. TNFR, TLRs, NLRs, IL-1R) [9]. As a classical example of canonical activation, we will focus on MyD88-dependent TLR signaling (Figure 2), a process that occurs for the interleukin 1 receptor (IL-1R) and all TLRs with the exception of TLR3 [10]. IL-1/TLR signaling commences via binding of their respective ligands. Engagement of a PAMP to an extracellular TLR (i.e. TLR1, TLR2, TLR4, TLR5, TLR6) induces a conformational change and hetero- or homo-dimerization of the TLR. Dimerization leads to a change in conformation of the receptor, followed by recruitment of adaptor proteins to the toll/interleukin receptor (TIR) domain of the TLR. The adaptor molecule MyD88 is recruited to extracellular dimers composed of TLR5 receptors, whereas Mal followed by MyD88 are recruited by TLR1/TLR2, TLR2, TLR2/TLR6, TLR4 dimers [11–16]. MyD88 acts as a protein scaffold, as it contains both a TIR domain and a death domain (DD) in its protein structure. The DD of MyD88 recruits interleukin receptor associated kinase (IRAK) IRAK-4, followed by IRAK-1, all in a helical assembly to form a Myddosome complex [17, 18]. The close spatial proximity of IRAK-4 to IRAK-1 allows for IRAK-1 to become rapidly phosphorylated by IRAK-4, and then undergo auto-phosphorylation [17, 19]. Once phosphorylated, IRAK-1 leaves the receptor complex and interacts with TRAF6. Upon activation, TRAF6 associates with Uev1A and Ubc13 and results in TRAF6 lysine-63 (Lys-63)-mediated polyubiquitination [20, 21]. Lys63-mediated polyubiquitination of TRAF6 acts as an important scaffold resulting in the recruitment and docking of TAB2/3 and TGF-β activated kinase-1 (TAK1) [22, 23]. Close association of the TAB2/3/TAK1 complex with polyubiquitinated TRAF6 allows TAK1 to undergo auto-phosphorylation, resulting in TAK1 activation [22]. Once TAK1 becomes phosphorylated, it can mediate downstream signaling by phosphorylating the inhibitor of κB kinase (IKK) complex on IKKβ subunits [22, 24]. IKKβ phosphorylation results in the activation of the IKK complex, composed of NEMO/IKKα/IKKβ subunits, to phosphorylate IκB [25]. Phosphorylation of IκB results in its subsequent recognition by the SCF-βTrCP ubiquitin (Ub) ligase complex that covalently modifies IκB with Lys-48 poly-Ub chains, which ultimately leads to the degradation of IκB by the 26s proteasome [26]. Once IκB is degraded, NF-κB dimers are liberated and predominantly shuttle into the nucleus and bind to -κB promoter and enhancer elements involved in the regulation of immunity, cell migration, cell adhesion, cell death and inflammation (Figure 2).

Figure 2: Canonical NF-κB Signaling Pathway.

The canonical NF-κB pathway can be stimulated by a variety of receptors, but here we depict TLR4 activation. Upon binding LPS, TLR4 homo-dimerizes and the resultant conformational change recruits Mal and MyD88. This in turn recruits IRAK-4 and IRAK-1 and the proximity results in IRAK-1 phosphorylation by IRAK-4. Phosphorylated IRAK-1 leaves the receptor complex and engages with TRAF6, which interacts with the enzyme complex Uev1A and Ubc13. This leads to lysine-63 polyubiquinitaion of TRAF6, followed by the docking of TAB 2/3 and TAK1. TAK1 becomes auto-phosphorylated and activates the IKK complex by phosphorylating the IKKβ subunit, liberating p50 and RelA to translocate into the nucleus.

Another form of ubiquitination, termed linear ubiquitination, contributes to the activation of NF-κB [27, 28]. Linear ubiquitination, as the name suggests, forms straight poly-Ub chains through the addition of Ub subunits. Specifically, this occurs by the addition of the first Ub to the target substrate. Linear chains form through the covalent attachment of the N-terminal methionine of the first Ub to the α- carboxyl group on Glycine-76 of the next Ub that is to be added [29]. This has been extensively reviewed elsewhere [30] and is in contrast to the modifications in which Ub can form poly-Ub chains via its seven lysine residues. Downstream of TNFR, NEMO has the capacity to be targeted and modified via linear ubiquitination by linear Ub chain assembly complex (LUBAC) [28], which is essential for the activation of NF-κB [31]. LUBAC is composed of several subunit proteins (HOIL-1L, HOIP and SHARPIN) that contribute to its linear ubiquitinase activity [28, 32]. Indeed, when components of LUBAC are compromised, NF-κB activation is attenuated [31]. Experimental evidence has demonstrated the importance of linear ubiquitination downstream of TLR4 and IL-1R, as reduced function of LUBAC leads to attenuated NF-κB activity downstream of these receptors [27, 33]. These data suggest that linear ubiquitination is involved in the activation NF-κB downstream of TLR signaling [34], and future studies will aid to further define this link.

1.3. The Non-Canonical NF-κB Signaling Cascade

The activation of the non-canonical or alternative NF-κB signaling cascade is tightly regulated and has a much smaller group of ligands and receptors that can induce its activation [8]. Members of the TNF/TNFR superfamily, which include the lymphotoxin β receptor (LTβR), CD40, B-cell activating factor receptor (BAFFR), receptor activator of NF-κB (RANK), TNFR2 and CD27 have all been shown to signal through the non-canonical pathway [8]. Upon ligation of TNF family members to their receptors, the TRAF2-TRAF3-cIAP1/2 E3 complex is recruited to the receptor [8]. This recruitment leads to accumulation of TRAF2 molecules, allowing TRAF2 to shift cIAP1/2’s Lys-48 Ub ligase activity towards TRAF3 via Lys-63 polyubiquitination of cIAP1/2 [8, 35]. When TRAF3 is modified via this mechanism, it ultimately leads to proteasomal degradation of TRAF3, allowing for accumulation of MAP3K14 (also commonly known as NF-κB Inducing Kinase (NIK)), resulting in increasing the levels of cytosolic NIK [8]. Mechanistically, in an unstimulated state, TRAF3 acts as a binding partner for NIK bringing it close to the cIAP1/2 E3 ligase complex, which in this case leads to its degradation [36, 37]. NIK is essential for the processing of p100 into active p52 [7, 8], which is a defining feature of the noncanonical NF-κB signaling pathway (Figure 3). However, NIK has also been shown to phosphorylate IKKα, which can act as a secondary kinase that mediates the activation of p100 [38]. For this to occur, NIK must be present in relatively high concentrations inside the cell before interacting with IKKα. Thus, stabilization of NIK results in its accumulation and interactions with IKKα, culminating in phosphorylation and activation [8, 38].

Figure 3: Non-Canonical NF-κB Signaling Pathway.

When unstimulated, only small amounts of NIK are expressed, which are immediately degraded by the TRAF3-TRAF2-cIAP1/2 complex. TRAF3 binds to NIK while TRAF2 acts as a bridge between TRAF3 and cIAP1/2, allowing for cIAP1/2-mediated lysine-48 polyubiquitination of NIK and subsequent proteasomal degradation. However, when a TNFR family member is activated, the TRAF3-TRAF2-cIAP1/2 complex is recruited to the receptor. This leads to an accumulation of TRAF2, which results in cIAP1/2 lysine-63 polyubiquitination and a shift in ligase activity of cIAP1/2 towards TRAF3 via lysine-48. The polyubiquitination and degradation of TRAF3 leads to an increase in NIK. Stabilized NIK is then free to interact with and phosphorylate IKKα. p100 is phosphorylated by active IKKα, further processed by lysine-856 polyubiquitinatation and degraded, leaving active p52. p52 and RelB are then free to translocate into the nucleus.

It is thought that IKKα-mediated phosphorylation of p100 occurs on Ser-872 in vitro and Ser-866 and 870 in vivo [8]. While the cause of this discrepancy is currently unclear, presumably, IKKα acts as a site-specific kinase for p100. This phosphorylation step targets the C-terminal inhibitory domains (the PID and the ankyrin repeat domain) of p100 for degradation via the proteasome [8]. Ankyrin repeat domains (ARDs) have been shown to mask the nuclear localization sequence (NLS) of IκBα and interestingly, p100’s phosphorylation site contains a sequence similar to IκBα [8, 39]. Therefore, disruption of the ARD of p100 via proteasomal degradation may lead to unmasking of the NLS and subsequent translocation of p52/RelB to the nucleus [39]. Once in the nucleus, p52/RelB drives the transcription of a limited repertoire of genes, including CCL19, CCL21, CXCL12, and CXCL13. In contrast, when the TNFR family members are unstimulated, NIK is instead degraded via Lys-48 polyubiquitination [8]. Consequently, IKKα will not become activated and phosphorylate p100. Thus, in this scenario, the NF-κB dimer cannot enter the nucleus and initiate transcription of target genes. Due to its essential function, NIK represents a “bottleneck” in the non-canonical NF-κB signaling cascade and is frequently targeted by mechanisms that have evolved to negatively regulate this alternative pathway.

2. Transcription Inhibition

The canonical and non-canonical NF-κB signaling cascades are complex biochemical pathways that rely on a large number of diverse proteins to transduce signals to the cell nucleus. Given this intricacy, it is unsurprising that there are many different checkpoints within these cascades that inhibitory proteins routinely target. The IκB family of proteins are the most commonly associated proteins that negatively regulate NF-κB signaling and serve as a classical example of NF-κB inhibition. There are eight IκB proteins, including the inhibitory regions on both p100 and p105, that target NF-κB dimers. In the case of the canonical NF-κB pathway, the predominant inhibitory protein is IκBα. Negative regulation occurs via the binding of ankyrin repeats on IκBα to the Rel homology domain of p65. The binding of IκBα to NF-κB heterodimers results in the concealment of the NLS on RelA (p65), thus keeping NF-κB heterodimers mainly contained in the cytosol under normal conditions [9, 40]. Following activation of the canonical NF-κB pathway, the IKK complex participates in NF-κB activation by phosphorylating IκBα on two key serine residues (Ser-32 and Ser-36) [41]. Phosphorylation of IκBα on these particular serine residues is an irreversible tag that results in further processing by polyubiquitination and degradation by the proteasome. Once IκBα is degraded by the proteasome, the nuclear import signal on p65/p50 heterodimers is no longer blocked, allowing for NF-κB translocation to the nucleus to regulate gene transcription [42]. IκBα is induced by the transcriptional activities of NF-kB activation, and this mechanism of induction of IκBα demonstrates an auto-regulatory feedback loop [42]. NF-κB and IκBα shuttle into and out of the nucleus. When the majority of IκBα is bound to NF-κB, the cytosolic flux predominates and NF-κB is prevented from entering the nucleus. Conversely, when NF-κB is free of IκBα, the nuclear flux predominates, which allows for NF-κB to enter the nucleus and bind enhancer regions on target DNA [9].

Beyond the IκB family, a diverse array of proteins have evolved to attenuate NF-κB signaling by targeting essential “bottlenecks” in the pathway through protein-protein interactions. Here, we focus on two distinct families of proteins, the IRAKs and NLRs, that each have unique members who function to negatively regulate NF-κB signaling through archetypical mechanisms.

2.1. IRAK-M

There are four IRAK family members in mammals: three of which are known as IRAK-1, IRAK-2 and IRAK-4. These three family members act as positive regulators of NF-κB signal transduction [43–46]. One family member, IRAK-3 or IRAK-M, is unique from the other family members in both structure and function, most notably acting as an inducible negative regulator of the canonical NF-κB signaling pathway [47]. All four IRAK family members share an N-terminal DD that is important for homotypic protein-protein interactions with either MyD88 or with other IRAK family members. Likewise IRAK-1,−2, and -M contain a unique C-terminal domain that contains TRAF6 binding motifs [48]. Together, these domains are essential for signal transduction and all IRAK proteins contain kinase domains [49]. IRAK-1 and IRAK-4 were first described as being the only functional kinases in the family, with IRAK-2 and IRAK-M being inactive or pseudo-kinases [43–45, 47]. These designations were originally based on amino acid sequence predictions [50]. There are specific residues in the sequence of the kinase domain that are invariant for kinase function and these include residues in the kinase sub-domain VIb, which are known as the HRD and DFG motifs [50, 51]. The motifs contain invariant aspartic acid residues, which are essential for proper function and are mutated in IRAK-2 and IRAK-M to an asparagine and serine, respectively.

Early functional studies suggested that IRAK-M shared a redundant function with IRAK-1 and IRAK-2 [52], despite the later prediction that IRAK-M was inactive or a pseudo-kinase. In overexpression systems and luciferase reporter assays, IRAK-1, IRAK-2, and IRAK-M were shown to function as positive regulators of NF-κB signaling. However, with in vitro kinase assays, human IRAK-M was demonstrated to have very weak intrinsic kinase activity, especially compared to IRAK-1 [52]. Further, murine IRAK-M was shown to contain detectable kinase activity; albeit, at a lower level when compared to the robust auto-phosphorylation of human IRAK-1 [53]. These data suggest that the auto-phosphorylation, and therefore the kinase activity of IRAK-M is negligible for its function; however, it is tempting to speculate that IRAK-M may indeed function as an active kinase on a currently unknown cellular substrate following its induction.

Despite these initial findings and speculation, the prevailing literature suggests that IRAK-M functions as a negative regulator of NF-κB signaling following TLR activation. These data are based on findings utilizing genetically modified mice with targeted deletions of the Irak-m gene [47]. The Irak-m gene contains 12 total exons and in order to define the role of IRAK-M, exons 9–11 were targeted for deletion [47]. These exons encode the amino acids predicted to constitute the putative kinase domain. Full length IRAK-M was not detected when a western blot was performed with an antibody specific for the C-terminus of IRAK-M [47]. Using these genetically modified animals, Irak-m was found to be induced by NF-κB, along with IκB and A20 following TLR stimulation, suggesting a negative feedback mechanism. Subsequent co-immunoprecipitation experiments demonstrated that IRAK-M has the capacity to bind to TRAF6, leading to the current model that IRAK-M functions to inhibit IRAK-1 and TRAF6 interactions, which in turn inhibits downstream NF-κB activity [47]. The expression of IRAK-M can be induced by other transcription factors, such as C/EBP, Smad4, AP-1 and CREB [54], suggesting additional functions beyond the feedback mechanism currently described. Phenotypically, Irak-m^−/− mice do not display any abnormal fetal or post-natal development, but do develop osteoporosis later in life due to hyperactive osteoclast activity [47, 55].

These Irak-m^−/− mice were instrumental in defining IRAK-M as a negative regulator of NF-κB signaling and have been widely utilized to define its function. For example, consistent with increased NF-κB signaling, Irak-m^−/− bone marrow derived macrophages (BMDMs) display impaired endotoxin tolerance and hyper-production of inflammatory cytokines (i.e. IL-6, TNF and IL-12p40) following TLR stimulation with specific PAMPS, as well as, L. monocytogenes and S. typhimurium [47]. Consistent with the ex vivo BMDM studies, Irak-m^−/− mice were found to display enhanced small intestinal inflammation following in vivo exposure to S. typhimurium [47]. Interestingly, in these initial studies, Irak-m^−/− mice do not display enhanced morbidity or mortality following infection, despite the increased inflammation [47]. These findings are consistent with a more recent study that showed the Irak-m^−/− mice are protected in models of experimental colitis and colitis associated tumorigenesis [56]. Mice lacking IRAK-M were found to have a robust immune response to bacteria translocating from the lumen following chemical induced damage to the intestinal epithelial cell barrier [56]. The attenuation in pathogenesis was associated with large expansions of gastrointestinal associated lymphoid tissue (GALT), increased neutrophil function, and enhanced T-cell recruitment [56]. Complementary data revealed that the gastrointestinal (GI) tract of the Irak-m^−/− mice had a lower total colonic bacterial load compared to wild type counterparts, which could also contribute to attenuation of disease pathogenesis [57]. Mechanistically, these data suggest the immune system in Irak-m^−/− animals is primed and more prone to a robust inflammatory response. In the GI tract, this improves the efficiency of the host response to pathogenic and commensal components of the host microbiome that drive disease processes.

It should be noted that while the consensus data identifies IRAK-M as a negative regulator of NF-κB signaling, contrary data suggests a possible alternative mechanism. Recently, it was demonstrated that IRAK-M participates in TAK-1 independent NF-κB activation downstream of TLR7 through MEKK3 [58]. With the use of IRAK-1/IRAK-2 double deficient mice and IRAK-1/−2/-M triple deficient mice, this study demonstrated that under highly specific conditions, IRAK-M functions in the absence of IRAK-1 and IRAK-2 and can actually activate NF-κB signaling [58]. Mechanistically, IRAK-M interacts with IRAK-4 to form an IRAK-M myddosome complex in the absence of both IRAK-1 and IRAK-2 that modulates NF-κB signaling through TAK-1 independent mechanisms [58]. While these data are certainly intriguing, the study ultimately concluded that IRAK-M exerts an inhibitory effect under normal conditions by indirectly inducing inhibitory proteins (A20, SHIP-1, SOCS1 and IκB-α) [58]. Further clouding mechanistic insight, it has recently been revealed that the original Irak-m^−/− mice commonly used to characterize this protein may actually contain a truncated version of IRAK-M [56]. Detailed sequencing analysis of the Irak-m gene product was conducted following TLR stimulation of BMDM from genotype confirmed Irak-m^−/− mice [56]. Under these conditions, a splice variant of the Irak-m gene was identified that resulted from the splicing of exon 8 with exon 12, in essence splicing around the neo cassette [56]. This type of splice variant is a common occurrence in genetically modified animals where functional domains are targeted, as opposed to the gene’s start site. Typically, these truncated proteins are dysfunctional and degraded by the cell, which preserves the knockout status of the animals. It is also important to note that the truncated IRAK-M protein has not yet been detected in situ and may not exist in vivo [56]. However, overexpression of Irak-m^rΔ9−11 and functional studies using the recombinant protein revealed that this truncation mutant is significantly more potent at activating NF-κB signaling then the wild type version [56].

Considering these data, we agree with the consensus findings that IRAK-M functions to negatively regulate NF-κB signaling under normal conditions. However, the possibility remains that IRAK-M may actually have a dual role under certain cell type or temporal specific conditions to also activate NF-κB signaling. This could be directly correlated with the cellular concentration of other IRAK family members or other critical cellular substrates. Resolution of the crystal structure of IRAK-M would provide valuable insight into these questions. It should also be pointed out that the history of IRAK-M is similar to that of IRAK-2, which was first believed to function as an inactive pseudo-kinase [50]. However, it was later determined that IRAK-2 function was independent of IRAK-1 and plays a critical role in sustaining the late phase of NF-κB signaling through potentially functioning as an active kinase [46]. Future studies will indicate whether our interpretation into the function of IRAK-M will be modified, as we have previously seen with IRAK-2.

2.2. Regulatory NLRs

Members of the Nucleotide-binding domain and leucine rich repeat containing (NLR; Nod-like receptor) family function through the formation of multi-protein complexes and are potent cytosolic regulators of diverse biochemical pathways. The best known and characterized members of this family include proteins that form inflammasomes [59]. However, other sub-groups exist beyond these inflammasome-forming NLRs that function as regulators of diverse biochemical signaling cascades [60, 61]. These regulatory NLRs either positively or negatively modulate signaling pathways directly or following the activation of other PRRs [60, 61]. Negative regulatory NLRs, such as NLRP12, NLRX1, and NLRC3, are the least defined family members. While the mechanisms and pathways targeted by these three NLRs vary, all seem to converge on the NF-κB signaling cascade.

Of the three negative regulatory NLRs, NLRP12 is the most studied and was one of the first NLRs to be functionally characterized. Initial studies identified NLRP12 as an inflammasome forming NLR due to findings from overexpression studies that revealed this protein was capable of interacting with the inflammasome adaptor protein ASC [62, 63]. It does appear that NLRP12 can form an inflammasome under highly specific conditions, following infection with a selection of unique pathogens [64, 65]. However, multiple studies evaluating NLRP12 inflammasome formation using Nlrp12^−/− animals have shown that this NLR does not directly regulate caspase activation or IL-1β/IL-18 maturation under the majority of conditions [66–73]. Rather, NLRP12 functions to attenuate inflammation by inhibiting key components of the canonical and non-canonical NF-κB signaling cascades [66, 67, 70, 71, 73–79]. In non-canonical signaling, NLRP12 inhibits the pathway through association with TRAF3 and NIK to attenuate non-canonical NF-κB signaling [66, 75]. These interactions result in the degradation of NIK, which leads to the subsequent inhibition of p100 cleavage to p52. NLRP12 inhibits IRAK-1 phosphorylation in a similar mechanism to attenuate the canonical NF-κB signaling pathway [71, 74, 77] (Figure 4). Beyond these direct mechanisms, NLRP12 may also indirectly attenuate NF-κB signaling by also targeting components of the ERK signaling pathway [66, 71]. However, the exact mechanism of NLRP12 inhibition of ERK signaling and the relationship with NF-κB signaling has yet to be fully resolved [66, 71].

Figure 4: Transcriptional and Post-transcriptional Attenuation of NF-κB Signaling.

Both canonical and noncanonical NF-κB signaling cascades are tightly regulated, although here we are only depicting the regulation of the canonical arm. NF-κB activation leads to gene transcription of negative regulators IκBα, IRAK-M, and the regulatory NLRs NLRP12, NLRX1, and NLRC3. IκBα ankyrin repeats bind to the Rel-homology domain of RelA, concealing the nuclear localization sequence. The pseudo-kinase IRAK-M functions to inhibit TRAF6 and the phosphorylation of IRAK-1. NLRP12 inhibits IRAK-1 phosphorylation in the canonical arm, and also functions to inhibit the noncanonical arm by inhibiting p100 cleavage (not shown). NLRX1 interacts with TRAF6 until stimulated, and then binds to the IKK complex to inhibit NF-κB. NLRC3 functions to negatively regulate NF-κB signaling by delaying IκBα degradation via interactions with TRAF6. NF-κB is also regulated by post-transcriptional mechanisms, including MicroRNAs (miRNAs) and RNA binding proteins (RBPs). miRNAs target mRNA sequences for gene silencing. There is evidence suggesting miR-302b binds to IRAK-4, miR-146 silences IRAK-1 and TRAF6, miR-199a inactivates IKKβ, and miR-15, miR-16, and miR-223 target and silence IKKα. Tristetraprolin (TTP) is an example of an RBP that negatively regulates NF-κB signaling by leading to degradation of mRNA with AU-rich elements, including TNF and IL-6, functioning as a negative feedback loop.

Unlike other NLR family members, NLRX1 has an undefined N-terminal domain and has an intimate relationship with the mitochondria. This unique NLR was originally found to negatively regulate type-I interferon signaling through interactions with the mitochondria anti-viral signaling (MAVS) protein [80–84]. In these original studies, NLRX1 was shown to curtail inflammation and promote resolution following virus infection through inhibiting RIG-I specific signaling [81, 83]. However, beyond the attenuation of IFN signaling, and similar to NLRP12, NLRX1 also negatively regulates NF-κB signaling through interactions with TRAF proteins (Figure 4). Specifically, NLRX1 associates with TRAF6 and IKK though an activation signal-dependent mechanism [82]. NLRX1 interacts with TRAF6 until stimulation, where NLRX1 is rapidly ubiquitinated and disassociates from TRAF6 allowing it to bind to the IKK complex [82]. This ultimately results in the inhibition of canonical NF-κB signaling [82].

Similar to NLRX1, NLRC3 has also been shown to negatively regulate NF-κB and IFN-I signaling [85, 86]. Of the regulatory NLRs, NLRC3 is the least characterized and appears to inhibit IFN signaling through impeding STING-TANK-binding kinase 1 (TBK1) interactions and STING trafficking [87]. In the NF-κB pathway, NLRC3 appears to delay the degration of IκBα to temper TLR signaling through interactions with TRAF6 [85]. While our understanding of NLRC3 is significantly lacking compared to many of the other NLR family members, the characterization of this protein has yielded significant mechanistic insight pertinent to the other negative regulatory NLRs. Mechanistic studies with NLRC3 revealed prevalent interactions with TRAF proteins and support the emerging hypothesis that members of the negative regulatory NLR sub-group function through the formation of a multi-protein “TRAFasome” complex to negatively regulate NF-κB signaling [85]. These data are consistent with the findings associated with NLRP12 (TRAF3) and NLRX1 (TRAF6) and provides some mechanistic insight regarding the multiple pathways that seem to be modulated by members of this NLR sub-group. Continued systematic study will be necessary to better characterize the formuation of this complex and identify additional interacting proteins.

3. Post-transcriptional Inhibition

Once NF-κB mediated gene transcription has been initiated, there are several additional post-transcriptional mechanisms capable of attenuating signaling. One highly efficient mechanism takes advantage of alternative mRNA splicing of key inducible genes to disrupt signaling. For example, expression of the MyD88 short (MyD88s) splice variant, which lacks a key intermediate region separating the N-terminal DD and the C-terminal TIR domain, can ablate MyD88-dependent TLR signaling through competitive inhibition. The truncated MyD88s protein is capable of binding TLR4, but fails to interact with IRAK4, resulting in a dominant-negative effect and the termination of IL-1 and LPS driven NF-κB signaling [88]. Similarly, the mouse Irak2 gene encodes 4 distinct isoforms associated with alternative splicing at the 5’ ends of the gene. Two of these splice variants have been shown to negatively regulate NF-κB signaling. One of these isoforms, Irak2c, was found to contain a putative NF-κB binding site not present in other Irak2 isoforms [89]. Thus, this isoform is suspected to exert an inducible negative feedback effect on the NF-κB signaling cascade [89]. It should be noted that similar isoforms have not been characterized for human IRAK2 [89].

3.1. microRNA

In addition to alternative splicing strategies, targeting RNA for degradation is also a highly effective negative feedback mechanism to modulate NF-κB signaling. MicroRNAs (miRNAs) are a particular class of RNA with regulatory function that act in this manner. These miRNAs robustly regulate the stability and translation of mRNA transcripts encoding key components of the NF-κB signaling cascade. Prior to processing, miRNAs are transcribed from long primary transcripts, known as pri-miRNA. Following pri-miRNA transcription, hairpin-loops will form in the pri-miRNA that are then recognized by the enzyme Dicer, which aids in the further processing and cleavage of the pri-miRNA into the mature miRNA [90]. Once mature, the miRNA is loaded into the RISC complex and modulates gene silencing through post-transcriptional mechanisms via antisense base pairing to coding mRNA transcripts. Over 1000 human miRNAs have been identified, which makes this class of regulatory molecules one of the most abundant available to the cell to control signaling [91]. Indeed, a brief search of NF-κB pathway components for this review identified multiple miRNAs for almost every step in the cascade, including those shown in Figure 4. Due to the abundance of this class of regulatory molecules and the diversity of the respective targets, deciphering their biological functions and physiological relevance is essential. Of these miRNA, three have been well described in the context of NF-κB signaling, miR-146, miR-21, and miR-155 and are described in further detail below.

MicroRNA-146 (miR-146) is particularly relevant in macrophages and has two variants, miR-146a and miR-146b. These variants are strongly induced by NF-κB following TLR4 engagement by LPS and following pathogen encounter [92]. Once induced, miR-146 binds to the 3’UTR of both IRAK-1 and TRAF6, either promoting the degradation of their respective mRNA or physically preventing their translation [93]. The latter mechanism is supported by overexpression studies, where miR-146a/b prevented protein translation of IRAK-1 and TRAF6, with no observable decrease in mRNA transcripts [93]. This binding was sufficient to inhibit MyD88 dependent NF-κB signaling [92]. Additional studies using a miR-146a^−/− mouse, further confirmed these findings [94]. These mice display a vigorous hyperinflammatory response to low doses of LPS and most succumb to endotoxic shock at higher doses of LPS more rapidly than wild type control animals [94]. Further, mice that are aged beyond 6 months develop severe multi-organ inflammation, tumorigenesis, and eventual death [95]. These miR-146a^−/− mice develop malignancies that are consistent with constitutive NF-κB signaling, including lymphoma and myeloid sarcoma [60], and deletion of the p50 subunit of NF-κB resulted in suppression of these diseases [95]. It is interesting to note that both miR-146a/b and IRAK-M negatively regulate NF-κB by interfering with IRAK-1 and TRAF6. This demonstrates the importance of both IRAK-1 and TRAF6 as essential signaling nodes for NF-κB activation. Further, overexpression and hyperactivity aberrancies in IRAK-1 contribute to myelodysplastic syndrome and overexpression of TRAF6 results in osteosarcoma [96, 97], which is also consistent with the phenotype of the miR-146a^−/− mice and loss of negative regulation in the NF-κB pathway.

Similar to miR-146, miR-21 also regulates NF-κB signaling to limit excessive inflammation through a negative regulatory feedback loop [98]. Following TLR4 activation, miR-21 directly binds to the 3’UTR of programmed cell death-4 (PDCD4) and thereby counteracts the effects PDCD4 [99]. PDCD4 functions to assist in the activation of NF-κB and augments IL-6 production, while also limiting the production of the anti-inflammatory cytokine IL-10 [98]. In this regard, PDCD4 acts as a molecular switch that tips the balance towards a pro-inflammatory microenvironment. Therefore, by opposing the effects of PDCD4, miR-21 attenuates persistent inflammation, by negatively regulating NF-κB and IL-6 production, while simultaneously augmenting the production of IL-10 [98]. Consistent with this mechanism, Pdcd4^−/− mice display resistance to the lethal effects of endotoxic shock when administered LPS, compared to wild-type controls [98]. Cell type specific differences for the function of miR-21 have been described [98, 100]. In RAW 264.7 cells (macrophages) miR-21 acts as a negative regulator of NF-κB when stimulated with LPS [98]. Alternatively, in a cellular transformation model driven by estrogen receptor (ER)-Src in MCF10A cells (mammary gland epithelial cell), miR-21 is induced by STAT-3. In the MCF10A model, miR-21 directly inhibits PTEN in the PI3K-Akt pathway and results in the activation of NF-κB signaling necessary to maintain the transformed state of these cells [101].

Genetic ablation of miR-21 in mice has demonstrated both pro- and anti-oncogenic roles [102, 103], perhaps due to the cell type specific effects discussed above. In one model, miR-21^−/− mice display reduced tumorigenesis when skin cancer is induced by chemical means [102]. This was attributed to increased apoptosis, decreased proliferation of epithelial keratinocytes, and an overall inhibition of the Ras-pathway [102]. Interestingly, a different group demonstrated that allograft tumors, injected subcutaneously and allowed to proliferate, are increased in size in miR-21^−/− mice, suggesting a protective role against tumor growth in this model [103]. He et al. hypothesized that loss of miR-21 results in attenuated CD4+ and CD8+ T-cell responses, which was indeed demonstrated and shown to be regulated by mi-R21 in a PTEN-Akt dependent manner [103]. miR-21 is overexpressed in a wide variety of human cancers and has been proposed as a bridge connecting inflammation together with cancer progression [104–106]. Collectively, these studies demonstrate miR-21 as molecule with pleiotropic functions, and further research is needed to determine whether chemical modulation of miR-21 will result in successful prevention of inflammatory maladies due to aberrant NF-κB signaling and cancer.

Similar to the other microRNAs, miR-155 is also a key inducible regulatory miRNA that has a broad expression pattern in leukocytes. Following TLR stimulation, miR-155 expression is up-regulated in monocytes, dendritic cells, macrophages, T cells, and B cells [92, 107, 108]. In addition to broad expression patterns, miR-155 is also induced by a wide range of stimuli, including Poly(I:C), LPS, IFN-β, CpG DNA, and Pam3CSK4 [109]. Proximal signaling of either the MyD88-dependent or TRIF-dependent pathways leads to miR-155 induction; whereas, inhibition of the JNK pathway prevents the induction of miR-155 when stimulated [109]. Functionally, miR-155 directly binds to the 3’UTR of the inositol phosphatase SHIP1 and represses SHIP1 function [110]. In BMDMs deficient in miR-155, SHIP1 levels are significantly increased 24 hours after treatment with LPS [110]. These data illustrate the importance of miR-155 in the JNK and AKT signaling pathways following LPS exposure. However, the role of miR-155 on the NF-κB signaling pathway appears to be context dependent. When induced in Kupffer cells, in a mouse model of liver disease driven by chronic alcohol consumption, miR-155 was shown to decrease the levels of IRAK-M, SHIP1, PU.1 and increase TNF [111]. Likewise, overexpression of miR-155 in mouse B-cells, utilizing Eμ-miR155 transgenic mice, produce more TNF when challenged with LPS and are more susceptible to death from endotoxic shock [112]. In this later study, miR155 was found to modulate NF-κB signaling through targeting IKKε, FADD, and RIPK1 transcripts [112]. Mice lacking miR-155 have been generated and under typical physiologic conditions, these animals display normal growth and development [113]. However, in a chemically induced model of colitis utilizing DSS, loss of miR-155 contributes to resistance and protection [114]. This protection from colitis was attributed to a reduction in frequency of CD4+ T cells [114]. Likewise, diminished Th1 and Th17 cell numbers were observed, which also likely attenuated colitis progression [114]. It is worth noting that the over-expression of miR-155 skews T-helper cells towards a Th1 phenotype and proper antigen presentation from dendritic cells is lost in the absence of miR-155 [107, 115]. In hematopoietic stem cells, sustained expression of miR-155 results in severe myeloproliferative disorder [116]. When taken together, these data suggest that loss of miR-155 actually results in a global attenuation of immune cell numbers resulting in reduced inflammation and malignancy.

3.2. RNA Binding Proteins

In addition to miRNA, RNA binding proteins (RBPs) are also highly effective regulators of NF-κB signaling [3]. These proteins act by binding to specific RNA molecules and recruiting catalytic proteins, which perform a diverse range of functions to modulate the target RNA, such as facilitate alternative splicing, polyadenylation, assist with mRNA transportation, and degradation [3, 117]. Together, these post-transcriptional strategies offer a rapid and robust mechanism to modulate NF-κB signaling.

The degradation process is one of the most critical functions facilitated by RBPs. In the context of NF-κB regulation, one of the best characterized RBPs is Tristetraprolin (TTP). TTP is considered a prototypic member of the RBPs and regulates NF-κB signaling by facilitating mRNA degradation [3, 118]. Specifically, TTP binds to mRNAs rich in AU-rich elements (AREs), which includes TNF and IL-23, and induces degradation [119]. TTP deficient mice develop hyperinflammatory profiles that include arthritis, dermatitis, cachexia, and myeloid hyperplasia [119, 120]. Macrophages derived from TTP deficient mice have increased production of TNF after stimulation with LPS [119]. This increase in TNF is associated with improved TNF mRNA integrity directly associated with TTP deficiency [119]. While increased TNF would be anticipated to stimulate both canonical and non-canonical NF-κB signaling, luciferase reporter assay data suggests that TTP may actually attenuate the canonical NF-kB cascade and this attenuation appears to occur downstream of p65 [118]. In these studies, TTP does not affect TNF-induced IκBα degradation [118]. Rather, TTP-mediated inhibition occurs at the level of nuclear translocation, as truncation of the NLS on p65 removed the attenuation of luciferase activity [118]. Thus, TTP may impair the nuclear translocation of p65 via an ARE-independent mechanism [118].

TTP is not the only RBP that is known to negatively regulate inflammation. Regnase-1 was discovered to be important for the negative regulation of inflammation in a similar manner to TTP. Specifically, mice that lacked regnase-1 show signs of hyper-inflammation and autoimmunity, such as decreased survival rates, hyperimmunoglobulinemia, and increased tissue infiltration of immune cells [121]. In addition, peritoneal macrophages from Regnase-1^−/− mice show increased expression of pro-inflammatory cytokines, such as IL-6 and IFNγ, after stimulation with TLR agonists [121]. The increased expression of pro-inflammatory cytokines is associated with regnase-1 degradation of NF-κB target genes via its intrinsic RNase activity [121]. Regnase-1 contains RNase activity and exerts its effects through the AREs in the 3’ UTR regions of target mRNAs [121]. Mechanistically, it was shown that stem-loop (SL) structures are required for regnase-1 to bind with the target mRNA, but is not required for degradation [122]. Studies using Regnase-1^−/− macrophages revealed that regnase-1 works on translationally active mRNAs [122]. Specifically, regnase-1 knockdown using siRNA in HeLa cells followed by treatment with IL-1β, revealed increased Il6 and Ptgs2 mRNA expression in polysome fractions compared with the control cells [122]. This has led to the suggestion that regnase-1 is involved in regulating early inflammatory steps and likely involves either direct or indirect inhibition of NF-κB signaling. In addition to its RNase activity, regnase-1 has also been reported to show deubiquitinating enzyme (DUB) activity that can act to attenuate NF-κB activation. A study using Raw264.7 transfected with regnase-1 (or control vector) and stimulated with LPS showed that regnase-1 expression partially blocked IKKβ phosphorylation [123]. Further investigation revealed that regnase-1 cleaved K63-linked poly-Ub chains on TRAF2 and TRAF6 [123]. NF-κB activation has been shown to affect regnase-1 expression as well. As stated above, LPS stimulation in macrophages leads to an increase in regnase-1 expression. However, it was also found that regnase-1 shows similar expression kinetics, to that of IκBα when cells were stimulated with LPS. This lead to the hypothesis that regnase-1 may be controlled by modification via TLR-mediated NF-κB activation [124]. Specifically, stimulation of TLRs in macrophages showed a disappearance of regnase-1 via immunoblot [124]. Further experiments revealed that IKKβ phosphorylates regnase-1 leading to its ubiquitination and degradation via the proteasome [124]. Together, these data clearly show that the two most studied RBPs, TTP and regnase-1, are tightly linked to NF-κB activation and act as important players of negative feedback in this pathway to help control inflammation. However, more work is needed to better define mechanisms for these proteins.

In contrast to TTP and regnase-1, the RBP Roquin has been shown to target translationally inactive mRNAs [122]. This is interesting because both Roquin and regnase-1 affect a common stem loop structure, but in different subcellular locations [122]. Further evidence of Roquin’s ability to regulate inflammation has been shown in Roquin-1 deficient mice, as these animals develop severe multi-organ hyperinflammation [125]. Early experiments showed that Roquin targeted the T-cell inducible co-stimulator (ICOS) mRNA in the 3’UTR, thus regulating its expression via degradation [126]. Subsequent studies revealed that Roquin can inhibit Th17 differentiation and that this inhibition can be regulated via MALT1 (mucosa-asociated-lymphoid-tissue lymphoma-translocation gene 1) [127]. Specifically, in CD4⁺ T cells that were treated with PMA and ionomycin, Roquin is cleaved and this cleavage is blocked by MALT1 inhibitors [127]. Thus, Roquin also has the potential to attenuate NF-κB signaling, albeit through targeting translationally inactive mRNAs.

4. Post-Translational Inhibition

Post-translational modification of key proteins is another well-defined mechanism associated with the regulation of NF-κB signaling. Common modifications include phosphorylation, glycosylation, ubiquitination, methylation, acetylation, and proteolysis and many of these processes occur in a gene transcription independent manner that allows for rapid and specific modulation of complex biological signaling cascades. Of these modifications, phosphorylation and ubiquitination are commonly observed in the NF-κB signaling cascade. Phosphorylation is one of the best studied post-translational modifications and is commonly considered to be an activating signal in the NF-κB cascade. This process is typically reversible, allowing a high level of temporal or tunable control, and usually occurs on serine, threonine, or tyrosine residues. In the NF-κB signaling pathway, phosphorylation can induce conformational changes in proteins such as IRAK and the IKK kinases, resulting in the release of auto-inhibition and increased enzymatic activity [3]. Conformation changes associated with phosphorylation can also promote or facilitate protein-protein interactions, such as those associated with IκBα or caspase-recruitment-domain (CARD) proteins [3]. Thus, phosphatases can function as negative regulators of NF-κB signaling by de-phosphorylating select proteins in the cascade. Likewise, while phosphorylation is commonly considered an activating signal, phosphorylation of inhibitory proteins can result in a net inhibitory effect on the signaling cascade. For example, IKKα phosphorylation of key components of the A20 Ub-editing complex results in activation and subsequent attenuation of NF-κB signaling [128].

While phosphorylation is typically considered to be an activating post-translational modification, ubiquitination can be either an activating or an inhibitory signal, depending on the types of poly-Ub linkages. Ubiquitin is a small 8.5 kDa peptide found in eukaryotic cells that can be covalently attached to proteins to regulate their function post-translationally. This is accomplished via the addition of poly-Ub chains to amino acid residues on target proteins. Ub itself contains seven different lysine residues, and the particular lysine chain that is utilized to make the poly-Ub chain determines the regulatory status of the protein. The seven-lysine residues found on Ub are Lys-6, Lys-11, Lys-27, Lys-29, Lys-33, Lys-48, and Lys-63, with Lys-48 and Lys-63 linkages being the best characterized in the context of NF-κB regulation [129, 130]. Generally, when a target protein is modified by Lys-48 linked polyubiquitination, this serves as a molecular tag on the protein for recognition and further destruction by the 26S proteasome [131, 132]. Additionally, Lys-63 linked polyubiquitination is a post-translational modification that generally results in the activation of tagged proteins [129]. Similar to phosphorylation, ubiquitination is also reversible and is tightly controlled by a balance between the opposing actions of E3 Ub ligases and DUBs [3]. In the NF-κB signaling pathway, Ub moieties can serve as docking regions for proteins with Ub binding domains, such as NEMO recruitment to proteins modified by Lys-63 linked poly-Ub chains [3]. Likewise, ubiquitination can serve as a signal to break down specific proteins by marking them for proteasomal degradation, for instance Lys-48 polyubiquitination of IκBα [3]. Thus, mechanisms capable of manipulating ubiquitination are typically potent regulators of NF-κB signaling.

4.1. A20

A20 or TNF alpha-induced protein 3 (TNFAIP3) is a protein that is rapidly induced by NF-κB and functions through a feedback mechanism to halt the canonical signaling cascade. A20 is one of the most studied negative regulatory proteins targeting components of the NF-κB pathway and functions primarily by modifying positive regulatory proteins in the cascade [133] through its action as either a Ub ligase or as a protein de-ubiquitinase [134]. While having both DUB and E3 ligase activity seems paradoxical, these activities are highly regulated and function in a cooperative manner. For example, in its role as a Ub ligase, A20 inhibits canonical pathway signaling by adding Lys-48 linked poly-Ub chains to the receptor interaction protein 1 (RIP1) [134, 135]. RIP1 is an essential molecule that activates the canonical NF-κB pathway downstream of the TNF receptor. Addition of Lys-48 linked poly-Ub chains to RIP1 targets this protein for degradation, thus A20 effectively targets this positive signal and ultimately restores cellular homeostasis. Beyond the modulation of RIP1, A20 has been shown to inhibit the E3 ligase functions of TRAF6, TRAF2, and cIAP1 by disrupting interactions with the E2 Ub conjugating enzymes Ubc13 and UbcH5c [136]. The mechanism associated with this disruption is associated with A20 triggered ubiquitination of Ubc13 and UbcH5c, resulting in proteasome-dependent degradation [136]. As a DUB, A20 appears to preferentially deubiquitinate Lys-48-linked chains in vitro; whereas, in vivo it targets Lys-63-linked poly-Ub chains [137]. Interestingly, rather than disassembling the poly-Ub chains in a sequential manner, A20 cleaves the chain at the junction with the substrate [138]. A20 does not function as a general DUB, rather it targets very specific polyubiquitinated substrates, such as TRAF6, that regulate NF-κB signaling following TNF, IL-1, and PRR activation [138] (Figure 5).

Figure 5: NF-κB Inhibition by Post-Translational Modification.

NF-κB signaling is highly regulated via post-translational modification throughout the entire pathway. A20 can act as both a ubiquitin ligase or as a deubiquitinase. For example, here we depict its role in inhibiting ubiquitination of TRAF6, although it can also target TRAF2 and cIAP1. CYLD is a constitutively expressed deubiquitinase that targets RIP1, IKKγ, TRAF2, and TRAF6. Cezanne is another deubiquitinase that targets TRAF6 in the canonical NF-κB pathway, but also inhibits noncanonical NF-κB.

The importance of A20 as a key negative regulator of NF-κB signaling was demonstrated with the use of A20 knockout mice [133]. In the absence of A20, mice display a significant inflammatory phenotype in the liver, intestine, bone joints, skin and kidney [133]. Furthermore, severe cachexia occurs in several organs and mortality commences postnatally within a few weeks of age [133]. This increase in morbidity and mortality can be further augmented when mice are given low doses of TNF or LPS [133]. A20 does not appear to be crucial for fetal development, because mice display an appropriate Mendelian ratio when born. It was demonstrated in mouse embryonic fibroblasts (MEFs) that in the absence of A20, NF-κB transcriptional activity is sustained and IκBα mRNA is transcribed with similar, even possibly increased levels, compared to wild type animals. However, IκBα protein levels do not rebound following TNF treatment [133]. This was attributed to the continued activity of the IKK complex, because in the presence of the proteasome inhibitor MG-132, IκBα levels were restored. Furthermore, it was demonstrated that IκBα was continually phosphorylated by the IKK complex following translation, which resulted in its constant degradation in A20 null cells [133]. Similarly, A20-Tnf and A20-Tnfr1 deficient mice also develop severe spontaneous inflammation, with evidence supporting roles in both innate and adaptive immune system functions [139]. LPS stimulated A20-Tnf double deficient BMDMs display a strong production of IL-6 and nitric oxide (NO) compared to wild-type mice. These levels were attenuated compared to A20 deficiency alone [139]. Likewise, robust inflammation was observed in A20-Rag1 double knockouts, suggesting that A20 also plays a prominent role in negatively regulating immune pathways beyond those regulated by TNF and the adaptive immune system [139]. In addition to TNF signaling, A20 was also demonstrated to robustly regulate TLR signaling. Myd88-A20 double deficient mice do not develop severe multi-organ inflammation and cachexia, as seen in A20 knockout mice alone. Thus, if TLR mediated NF-κB activation is nonfunctional, then there is a reduced need for A20 inhibition.

Similar to the mouse studies, mutations in the human gene encoding A20 TNFIAP3, are strongly linked with a variety of hyperinflammatory, autoinflammatory, and malignant diseases in patients [140]. These disease include systemic lupus erythematosus (SLE), rheumatoid arthritis, inflammatory bowel disease (IBD), and type 1 diabetes [140]. In general, cells from patients with mutations in TNFIAP3 show increased degradation of IκBα, p65 nuclear translocation, and subsequent expression of a range of proinflammatory cytokines associated with canonical NF-κB signaling [141]. Interestingly, the host microbiome significantly contributes to the pathogenesis of the majority of the diseases linked to A20 dysfunction and proper NF-κB signaling is a key element required to keep the balance between commensal and pathogenic microbes, especially in the gut [142]. Thus, it is possible that dysfunctional A20 regulation of NF-κB signaling could contribute to dysbiosis and disease in human patients. Indeed, support for this hypothesis has already been reported in mouse models where spontaneous hyperinflammation in chimeric mice reconstituted with A20^−/− bone marrow were rescued using antibiotic microbiota ablation [143]. While there is still much to learn regarding the function of A20, together these mouse and human studies emphasize the importance of this key molecule in regulating NF-κB signaling in a variety of biological and clinically relevant conditions.

4.2. CYLD

CYLD is a constitutively expressed DUB that was originally described as a tumor suppressor gene [144]. Loss of function of CYLD in humans is associated with familial cylindromatosis, which are multiple benign skin cancers usually localized to the face and neck [144]. Additional mechanistic studies revealed that loss of CYLD contributes to an overall cellular resistance in apoptosis [145], which likely contributes to cancer pathogenesis. CYLD is a member of the Ub-specific protease family (USP), which represents the largest family of DUBs in the human genome [146]. As a bona fide cysteine protease, CYLD cleaves Lys-63 linked chains of poly-Ub on target proteins [147]. This serves as a repressive function because addition of poly-Ub on Lys-63 is generally a post-translational modification that leads to increased activity of modified proteins. Interestingly, in addition to Lys-63 mediated de-ubiquitination, CYLD has the ability to de-ubiquitinate via linear Ub chains [148]. It is interesting to note that CYLD and A20 cleave Ub chains on several of the same substrates that participate in the signal transduction cascade in the NF-κB pathway, including but not limited to RIP1, IKKγ, TRAF2 and TRAF6 [142, 145, 149, 150] (Figure 5). This overlap in negative regulatory function demonstrates the importance of signaling molecules, such as TRAF6, as primary nodes in the induction of the inflammatory cascade. The post-translational modification of several of the same molecular targets between A20 and CYLD could be due to steady-state and inducible negative regulation. Further kinetics and substrate avidity studies are necessary to explain the similar molecular targets between these two molecules.

Studies utilizing independently generated Cyld^−/− mice have yielded highly insightful findings, albeit somewhat conflicting. In one study, the Cyld^−/− animals demonstrated significant defects in NF-κB and JNK signaling in macrophages following TLR and CD40 stimulation [151]. However, this effect appeared to be stimuli specific, as TNF stimulation did not appear to be impacted by CYLD deficiency [151]. Importantly, this study observed normal B and T-cell development and function [151]. While there are several differences in the proposed mechanistic actions of CYLD, these findings are in general agreement with other groups using these and other independently generated Cyld^−/− animals [152, 153]. However, at least one additional group has not observed these phenotypes using another independently generated Clyd^−/− mouse line [154]. These animals did not recapitulate the defects in NF-κB, JNK, ERK, or p38 signaling [154]. Rather, these animals showed significant attenuation of T cell receptor (TCR) signaling in the absence of CYLD [154].

4.3. Cezanne

Cellular zinc finger anti-NF-κB (Cezanne) is a 100 kDa protein with sequence similarity to A20 [155, 156]. Because of this, its potential role in NF-κB attenuation has been extensively studied. Like A20, Cezanne is a negative regulator of NF-κB signaling, which has been shown using overexpression studies in HEK-293 cells [156]. In this model, overexpression of Cezanne reduced NF-κB activity after stimulation with TNF [156]. Cezanne was further found to be inducible in HEK-293 cells following stimulation with TNF, suggesting the presence of a feedback loop consistent with many of the negative regulators of NF-κB signaling [157]. Mechanistically, Cezanne was shown to interact with TRAF6 by co-immunoprecipation following overexpression [156]. Further studies revealed that Cezanne cleaves monomers from linear or branched poly-Ub chains and thus acts as a DUB [155]. In endothelial cells pretreated with specific Cezanne siRNAs, TRAF6 co-precipitates with Lys-63 poly-Ub chains, whereas this same result is not seen in control cells without the siRNAs [158]. Together, the data from these studies indicate that Cezanne can inhibit canonical NF-κB signaling through the de-ubiquitination of TRAF6 both in vitro and in vivo (Figure 5).

It should also be mentioned that Cezanne has also been reported to play a role in the negative regulation of the non-canonical NF-κB pathway [159]. Hu et al. (2013) showed that cells which were deficient in Cezanne had a significant increase in nuclear levels of p52, compared to wild type, after stimulating the non-canonical NF-κB activating receptors, LTβR, CD40 and BAFFR [159]. Furthermore, compared to wild type, cells that lacked Cezanne show a much higher level of TRAF3 Lys-48 ubiquitination, degradation and NIK accumulation after treatment with non-canonical NF-κB agonists [159]. Therefore, it can be said that Cezanne is an important negative regulator of inflammation via its ability to attenuate both the canonical and non-canonical NF-κB pathways.

5. Crosstalk between the canonical and non-canonical NF-κB pathways

The canonical and non-canonical NF-κB signaling pathways are distinct and separate entities. However, there is a significant amount of cross-talk that occurs between the two pathways. This crosstalk acts in both a positive and negative manner, allowing further control and regulation of signaling. In addition, this interplay occurs via direct and indirect mechanisms. Direct crosstalk occurs via the activation of pathway components that are shared between both the canonical and non-canonical cascades, including various TRAF proteins or IKKα. NF-κB dimer formation can also modulate the two signaling cascades. For example, in MEFs treated with TNF, RelA binds to RelB and represses its DNA binding function [160]. A later study also confirmed this RelA/RelB mechanism in human dendritic cells whereby cells were stimulated with dectin-1 leading to the nuclear translocation of both RelA and RelB individually and as RelA/RelB dimers [161]. This dimerization decreased the DNA binding ability of RelB, ultimately resulting in a shift towards RelA dependent canonical NF-κB signaling [161].

Further linkage between the two pathways occurs via the TRAF proteins. In general, these proteins have been shown to be positive regulators, including TRAF2, TRAF5, and TRAF6 that modulate canonical NF-κB activation. Conversely, TRAF3 is a negative regulator of non-canonical signaling in the context of NIK degradation. However, in addition to its better defined roles in the non-canonical pathway, TRAF3 may also contribute to the regulation of the canonical pathway. For example, in spleen and thymus extracts, both canonical and non-canonical NF-κB activation status is much higher in organs from Traf3^−/− mice compared to wild type animals [162]. Likewise, Traf3^−/− MEFs have similar increases in both canonical and non-canonical NF-κB signaling at baseline and following treatment with cytokines targeting the pathways (TNF and IL-1β) [162]. Mechanistically, it seems that an accumulation of NIK via the lack of TRAF3, may lead to an amplification of IKK activity and a subsequent increase in both canonical and non-canonical NF-kB activity [162].

Indirectly, both canonical and non-canonical NF-κB signaling pathways can communicate via many of the negative regulators discussed in this review. For example, the IκB family of molecules and the DUB cezanne clearly play important roles in negatively regulating both pathways. However, other molecules have also been shown to overlap as well, although sometimes having opposite pathway specific affects. A20 is an excellent example of this dichotomy. As discussed above, A20 is a potent regulator of canonical NF-κB signaling. However, in cells stimulated with non-canonical pathway agonists, processing of p100 to p52 is significantly reduced in cells treated with siRNAs for A20 [163]. This was further confirmed using A20^−/− MEFs stimulated with the same agonists, resulting in reduced NIK accumulation and p100 processing [163]. Interestingly, stimulation with CD40L induced NIK accumulation regardless of whether A20 was present or absent, suggesting an A20-independent mechanism for the non-canonical pathway under these conditions [163].

RIP-1 is another mediator of NF-κB signaling that has opposing effects on the two pathways. As mentioned earlier, RIP1 activates the canonical NF-κB pathway downstream of the TNF receptor. However, it also negatively regulates the non-canonical pathway via inhibition of TRAF2 degradation. TNF treatment in MEFs lacking RIP1 revealed TRAF2 degradation, which coincided with an increase in p100 processing [164]. Because TRAF2 plays a role in the recruitment of cIAP1/2 to the TRAF-cIAP1/2 complex, which is responsible for targeting NIK for degradation, it stands to reason that its degradation would lead to accumulation of NIK. Indeed, NIK was shown to be stabilized in MEFs lacking RIP-1 and treated with TNF compared to WT cells [164]. While TNF stimulation is widely accepted to activate the canonical pathway via recruitment of RIP-1 to TNFR1, this same series of events seems to inhibit TRAF2 degradation. Thus, NIK is unable to become stabilized, resulting in inhibition of the non-canonical pathway.

6. Concluding Remarks

The mechanisms associated with the attenuation and resolution of inflammation are still being defined. This stands in contrast to the extensive data pertaining to immune system activation and signaling. One can certainly argue that having an organized system of molecular checks and balances to prevent hyperactive immune system signaling is vital due to the collateral damage that uncontrolled inflammation can have at the tissue and organ level. NF-κB signaling lies at the center of the molecular mechanisms controlling inflammation, thus making this biochemical cascade a major target for both positive and negative regulatory pressure. Negative regulators of NF-κB signaling play a critical role in the extensive network of overlapping and often times redundant regulatory mechanisms associated with the control of inflammation. Indeed, many of the negative regulators are themselves controlled by NF-κB activation, which serves as a feedback mechanism to maintain immune system homeostasis. However, any shift in this delicate balance can have significant and far reaching biological effects. Furthermore, an additional level of control stems from crosstalk between these two pathways. Indeed, some molecules have been shown to limit the activation of one pathway while activating the other. However, much of the mechanism of this regulation is still unknown. Thus, further insight into these complex negative regulators will be critical to better understand their unique biological functions, elucidate clinical relevance, and improve therapeutic strategies aimed at controlling anarchic inflammation.

Abbreviations Page

(ALR)

AIM2-like receptor

(ARD)

Ankyrin repeat domain

(AREs)

AU-rich elements

(BAFFR)

B-cell activating factor receptor

(BMDMs)

bone marrow derived macrophages

(CARD)

caspase-recruitment-domain

(cIAP)

cellular inhibitors of apoptosis

(CLR)

C-type lectin receptor

(DAMP)

damage associated molecular pattern

(DD)

death domain

(DUBs)

deubiquitinating enzymes

(ER)

estrogen receptor

(GALT)

gastrointestinal associated lymphoid tissue

(IBD)

inflammatory bowel disease

(IκB)

Inhibitor of –κB

(IKK)

inhibitor of κB kinase

(IRAK)

interleukin receptor associated kinase

(IL-1β)

interleukin-1 beta

(LTβR)

lymphotoxin β receptor

(MAVS)

mitochondria anti-viral signaling

(MEFs)

mouse embryonic fibroblasts

(MALT1)

mucosa-associated-lymphoid-tissue lymphoma-translocation gene 1

(NIK)

NF-κB inducing kinase

(NLR)

Nod-like receptor

(NLS)

nuclear localization sequence

(PAMP)

pathogen associated molecular pattern

(PRR)

pattern recognition receptor

(PDCD4)

programmed cell death-4

(RIP1)

receptor interaction protein 1

(RANK)

receptor activator of NF-κB

(RLR)

Rig-I-like helicase

(SLE)

systemic lupus erythematosus

(TCR)

T cell receptor

(TAK1)

TGF-β activated kinase-1

(TNFAIP3)

TNF alpha-induced protein 3

(TRAF)

TNF-receptor-associated factor

(TIR)

toll/interleukin receptor

(TLR)

Toll-like receptor

(TTP)

Tristetraprolin

(TNF)

tumor necrosis factor

(Ub)

ubiquitin

(USP)

ubiquitin-specific protease