Regulation of inflammation by β-arrestins: Not just receptor tales

Abstract

The ubiquitously expressed, multifunctional scaffolding proteins β-arrestin1 and β-arrestin2 each affect inflammatory signaling in a variety of cell lines. In addition to binding the carboxyl-terminal tails of innumerable 7-transmembrane receptors, β-arrestins scaffold untold numbers of other plasma membrane and cytoplasmic proteins. Consequently, the effects of β-arrestins on inflammatory signaling are diverse, and context-specific. This review highlights the roles of β-arrestins in regulating canonical activation of the pro-inflammatory transcription factor NFκB.

β-arrestin1 and β-arrestin2 (also known as arrestin-2 and arrestin-3) were originally discovered and characterized in the context of desensitizing 7-transmembrane receptor signaling through heterotrimeric G proteins [1,2]. Subsequently, these 46–47 kDa, ubiquitously expressed proteins were found to function as multifunctional scaffolds that mediate receptor internalization [3,4] and that either diminish or activate signaling downstream of 7-transmembrane receptors, certain receptor protein tyrosine kinases [5–7], cytokine receptors [8–10], and ion channel receptors [11]. β-arrestins can interact with hundreds of proteins [12]; consequently, β-arrestins trigger or modulate myriad signal transduction pathways—including the activation of Src family tyrosine kinases [13], phosphatidylinositol-3-kinase/Akt [14], IRS-1 [15], RhoA [16,17] Reference numbers 17 and 27 are the sameNFκB [9,10,18,19], and the mitogen-activated protein kinase families comprising ERK, JNK, and p38 isoforms (reviewed in reference [20]). β-arrestin2 promotes anti-apoptotic [7,21–23] and chemotactic [24–27] activity in a variety of cells. Through these diverse activities, in part, β-arrestins play an important role in inflammation.

Because they share 78% amino acid identity [2], it is not surprising that β-arrestin1 and β-arrestin2 associate with many of the same proteins. Indeed, ~ 50% of the proteins thus far shown to bind to either β-arrestin isoform bind to both β-arrestin isoforms (often with differing affinities, however [4,20,28]) [12]. This overlapping but distinct β-arrestin binding specificity manifests in two distinctive signaling patterns. First, both β-arrestin isoforms may contribute to full activation of signaling, as is the case with G protein-independent ERK1/2 signaling elicited by the β₂-adrenergic receptor [29]. Second, one β-arrestin isoform may promote while the other β-arrestin isoform inhibits signaling, as exemplified by ERK1/2 signaling elicited by the angiotensin II AT₁ receptor [30], the protease-activated receptor-1 [31, 32], and lysophosphatidic acid receptor-1 and -2 [32]. This variability of β-arrestin-promoted ERK1/2 signaling manifests in inflammatory signaling, too.

Evidence for β-arrestin-dependent effects on inflammation has accumulated over the last 16 years—first in transfected model cell systems, and subsequently in cells and mice rendered β-arrestin-deficient genetically or by RNAi. Diverse biological systems and inflammatory stimuli have produced a multi-dimensional picture of β-arrestin-dependent regulation of inflammation.

Anti-inflammatory Roles for β-arrestins

The first use of βarr2^−/− mice to assess inflammatory activity investigated lymphocyte migration triggered by CXCL12 through the 7TMR CXCR4 [24]. Comparison of βarr2^−/− with cognate WT lymphocytes demonstrated that β-arrestin2 desensitized CXCR4-induced G protein-mediated signaling. However, βarr2 deficiency substantially diminished CXCL12-induced (CXCR4-dependent) migration of T and B lymphocytes [24]. Because CXCR4-dependent migration constrains lymphocytes to the bone marrow and thymus compartments [33], the ability of β-arrestin2 to augment CXCL12-induced migration ~ 2-fold could be envisioned as anti-inflammatory (a possibility not yet tested in vivo).

The role of β-arrestins in inflammatory signaling has been studied most extensively in the context of events promoting the activation of the pro-inflammatory transcription factor designated NFκB. Canonical activation of NFκB can be triggered through a variety of cytokine receptors [34], but we will focus here on agonist-mediated TLR4 or interleukin-1 receptor dimerization (Fig. 1). This event engenders MyD88-dependent activation of the E3 ubiquitin ligase TRAF6 (TNF receptor-associated factor 6), in a process that involves TRAF6 oligomerization, auto-ubiquitination, and subsequent synthesis of K63-linked polyubiquitin chains that are either covalently or noncovalently attached to other proteins [35] (Fig. 1). These K63-linked polyubiquitin chains activate the kinase TAK1, and co-localize TAK1 with inhibitor of NFκB (IκB) kinase (IKK) through noncovalent interactions [35]. Consequently, TAK1 phosphorylates and thereby activates IKKβ. IKKβ then phosphorylates the inhibitor α of NFκB (IκBα) and thereby triggers K48-linked polyubiquitination and proteasomal degradation of IκBα; subsequently, NFκB p65/p50 heterodimers are de-inhibited and can activate transcription of pro-inflammatory genes [35,36]. In this signaling model, βarr1 and/or βarr2 have been shown to inhibit NFκB activation at two levels: (i) by binding to IκBα and preventing its degradation [18,19], and (ii) by binding to TRAF6 and thereby impeding TRAF6 oligomerization and auto-ubiquitination [9].

Fig. 1. Model for βarr-mediated regulation of inflammatory signaling.

Pictured are NFκB activation pathways downstream of the 7-transmembrane receptor for the platelet product lysophosphatidic acid (LPA-R), the TLR4, and the TNF receptor-1 (TNFR1). LPA-R-mediated activation of NFκB requires a βarr2-scaffolded “signalosome” comprising CAR-MA3, Bcl10, MALT1, and TRAF6, which autoubiquitinates and generates K63-linked polyubiquitin that binds to NEMO to activate IκB kinase (IKK) and thence NFκB. By binding to IκBα, however, either βarr1 or βarr2 can inhibit IκBα degradation in the 26S proteasome, and thereby inhibit NFκB activation. TLR4 or interleukin-1 receptor dimerization triggers phosphorylation and thus activation of IRAK1 (IL-1 receptor-associated kinase), which then associates with and activates TRAF6, whose K63 polyubiquitin chains bind to NEMO and to the TAB2 subunit of the kinase TAK1, which autophosphorylates/activates and then phosphorylates and thereby activates IKK. TNFR1 activation recruits TRAF2 to at least 2 signaling complexes: (a) the classical complex involving TRADD and the ubiquitin E3 ligase cIAP1, with polyubiquitin- and phosphorylation-mediated activation of IKK similar to that described above; (b) a second complex involving βarr1 (but perhaps not βarr2), Src and Gq. “UUU” denotes polyubiquitin chains involved in promoting signaling; “+” denotes stimulation; “−” denotes inhibition. Schematic adapted from references 35, 44 and 46.

The association of βarr1 and βarr2 with IkBα was first discovered by yeast two-hybrid studies and confirmed by pull-down studies with purified proteins and by co-immunoprecipitation studies in mammalian cells [18,19]. In HeLa cells, however, tumor necrosis factor (TNF)-induced activation of NFκB increases consequent to siRNA-mediated silencing of βarr1, but not βarr2 [19]. In contrast, βarr2 silencing in HeLa cells augments TNF-induced expression of NFκB-dependent cytokines [18]. Furthermore, in HEK293 cells, silencing of βarr2 also augments NFκB activation triggered by TNF [18].

βarr1 and βarr2 associate with TRAF6 via the C-terminal 100 amino acids of the βarrs [9], rather than the βarr N-terminal domain that associates with IκBα [18]. Activation of TLR4 (by LPS), the interleukin-1β receptor, or the receptor activator of NFκB (RANK) promotes this βarr/TRAF6 association [9], which appears to inhibit NFκB activation—albeit in what seems to be a context-specific manner—in several cellular systems. In βarr1^−/−/βarr2^−/− MEFs, rescue expression of βarr1, βarr2, or their corresponding C-terminal minigenes reduces LPS-induced ubiquitination of endogenous TRAF6 [9]. However, LPS-induced secretion of the NFκB-dependent cytokine interleukin-6 (IL-6) by the same βarr1^−/−/βarr2^−/− MEFs was restored to WT levels by rescue expression of just βarr2, and not βarr1—perhaps because of interplay between TRAF6-dependent and ERK1/2-dependent signaling [37]. Congruently, in HeLa cells, siRNA-mediated silencing of βarr1 or βarr2 increases LPS-induced TRAF6 ubiquitination [9]. NFκB and AP-1 promoter/reporter studies in HEK293 cells revealed that NFκB activity is diminished or augmented, respectively, by βarr overexpression or silencing [9,37]. Silencing of βarr1 in THP-1 monocytic leukemia cells increases the LPS-induced secretion of NFκB-dependent cytokines (IL-6, IL-8), just as βarr2^−/− bone-marrow-derived macrophages show greater LPS-induced IκB kinase activity than WT macrophages [9]. Contrastingly, however, βarr2^−/− and WT macrophages—whether bone-marrow-derived or peritoneal—can show equivalent secretion of the NFκB-dependent cytokines TNF and IL-6 in response to LPS, even though βarr2^−/− splenocytes evince greater secretion of these cytokines in response to LPS [38]. Data from fibroblast-like synoviocytes suggests the possibility of βarr isoform-specific regulation of NFκB activity: the secretion of TNF and IL-6 in response to hyaluronan increases with βarr1 overexpression and decreases with βarr2 overexpression [39].

The inhibitory effect of βarrs on TRAF6 (and thus on NFκB activation) may reside in the ability of βarrs to prevent TRAF6 oligomerization, which is required for auto-ubiquitination: overexpression of βarr1, βarr2, or their corresponding C-terminal minigenes in HEK293 cells inhibits the association of co-transfected Flag- and HA-tagged TRAF6 constructs [9].

In the setting of acute inflammation in vivo, βarr2 deficiency appears to augment NFκB activation and thereby predispose to systemic disorders. Whether induced by cecal ligation and puncture [38] or by intraperitoneal injection of LPS [9], the systemic inflammatory response in βarr2^−/− mice substantially exceeds that in WT mice: βarr2^−/− mice demonstrate greater plasma levels of TNF and IL-6, as well as greater death rates over the ensuing 1–5 days than WT mice [9,38]. Arthritis induced over 5–14 days by anti-collagen IgG manifests more severely in βarr2^−/− than WT mice [39].

Pro-inflammatory Roles for β-arrestins

That β-arrestins promote inflammatory signaling was first appreciated in polymorphonuclear leukocytes, in which β-arrestins appear to promote degranulation. In model cell systems, IL-8 promotes βarr1-mediated CXCR1 internalization, subsequent βarr1-mediated activation of the Src-family tyrosine kinase Hck, translocation of the βarr-Hck complexes to granular compartments of the PMNs, and subsequent exocytosis of PMN granules [40].

β-arrestin2 plays a powerful pro-inflammatory role in allergic asthma, as judged by results obtained with an ovalbumin-promoted model in mice [41]. Indeed, ovalbumin-primed, methacholine-triggered bronchospasm observed in WT mice does not occur at all in βarr2^−/− mice, even though βarr2^−/− mice mount ovalbumin-specific IgG and IgE responses equivalent to those in WT mice. βarr2 deficiency abolishes ovalbumin-evoked T lymphocyte and eosinophil infiltration of the lungs, and eliminates Th2 (but not Th1) cytokine production in the lungs [41]. Mechanisms responsible for the diminished pulmonary infiltration by T lymphocytes appear to involve βarr2-mediated effects on several aspects of inflammation. First, βarr2 deficiency abrogates the secretion of the Th2 chemokine CCL22 by lung macrophages, airway smooth muscle cells, bronchiolar epithelial cells and mast cells [42] in response to aerosolized ovalbumin [41]. In this context, it is worthwhile noting that CCL22 is an NFκB-dependent gene product [43]; thus, the failure of ovalbumin to evoke pulmonary cell CCL22 secretion in βarr2^−/− mice may indicate that βarr2 activity augments NFκB activity in CCL22-secreting cells (see below). Second, echoing findings obtained with CXCL12 [24], βarr2 deficiency greatly reduces T lymphocyte migration toward CCL22. Thus, βarr2 activity is pro-inflammatory in ovalbumin-primed, methacholine-triggered allergic asthma, which develops over 3 weeks of antigen challenge and involves T cells as well as pulmonary macrophages and epithelial cells. In contrast, βarr2 activity has no effect at all in acute, LPS-promoted, methacholine-induced non-allergic asthma, which involves just pulmonary macrophages, epithelial cells, and immigrating neutrophils [41].

βarr1 plays a pro-inflammatory role downstream of TNF receptor activation in model adipocytes [44]. In these cells, βarr1 associates with TNF receptor-associated factor 2 (TRAF2), an adaptor/ubiquitin E3 ligase integral to TNF receptor-1 signaling [44]. Through binding to TRAF2, βarr1 augments TNF-induced lipolysis, JNK activation, and NFκB-dependent gene expression, as shown by studies employing siRNA-mediated βarr1 silencing [44]. Interestingly, this TRAF2/βarr1 association does not appear to occur in HEK293 cells [9].

βarr2—but not βarr1—also appears to promote inflammatory signaling downstream of lysophosphatidic acid receptors in mouse embryo fibroblasts (MEFs). Assessed by electrophoretic mobility shift assay and by IL-6 secretion, lysophosphatidic acid-induced NFκB activation observed in WT MEFs is virtually abrogated in both βarr1^−/−/βarr2^−/− double knockout and βarr2^−/− MEFs, yet it remains intact in βarr1^−/− MEFs [45]. Despite this remarkable βarr2-dependency for NFκB activation induced by lysophosphatidic acid, there is no βarr2-dependency for NFκB activation induced by tumor necrosis factor [45]. The locus of βarr2 activity appears to be upstream of IκB kinase activation [45], perhaps in promoting ubiquitination of IκB kinase-γ, or “NEMO” [46]. Rescue expression in βarr1^−/−/βarr2^−/− MEFs of βarr2, but not βarr1, restores lysophosphatidic acid-induced NFκB activation to WT levels.

In vascular inflammation, βarr1 and βarr2 appear to play reciprocal roles: βarr1 appears to be anti-inflammatory, and βarr2 pro-inflammatory, as demonstrated by endothelial denudation studies [32]. Endothelial denudation of the mouse carotid artery engenders platelet and neutrophil adhesion to the subendothelial basement membrane; subsequently, smooth muscle cells in the artery’s tunica media proliferate and migrate through the internal elastic lamina into the subendothelial tunica intima, thereby creating “neointimal hyperplasia” over a ~ 4-week time course. Thus, neointimal hyperplasia integrates inflammatory processes of leukocytes, platelets and smooth muscle cells. Whereas endothelial injury-induced neointimal hyperplasia is greater in βarr1^−/− than in WT mice, it is less in βarr2^−/− than in WT mice [32]. Coupled with bone marrow transplantation, these endothelial denudation studies demonstrate that neointimal hyperplasia is augmented by βarr2 activity not in bone marrow-derived cells, but in arterial smooth muscle cells. In smooth muscle cells, βarr2 activity promotes cell proliferation and diminishes apoptosis by augmenting the activation of ERK and Akt in response to diverse stimuli in vivo and to pro-inflammatory 7TMRs in vitro [32]. In contrast, βarr1 activity in smooth muscle cells accomplishes the reverse [32]. Furthermore, in smooth muscle cells βarr2 augments TLR4-dependent NFκB activation [10].

βarr2 also produces pro-inflammatory activity in atherosclerosis, a chronic vasculitis triggered by hyperlipidemia and responsible for the vast majority of strokes and myocardial infarctions [32]. Atherosclerosis fundamentally involves endothelial cells, smooth muscle cells, and several classes of leukocytes including monocyte/macrophages, T cells, and mast cells [47,48]. Atherosclerosis develops in Ldlr^−/− mice over 12–20 weeks, and thus represents a more chronic form of inflammation than other models discussed in this review. Compared with Ldlr^−/− mice, βarr2^−/−/Ldlr^−/− mice develop 40% less atherosclerosis, with 35% fewer smooth muscle cells in their atherosclerotic lesions [32]. βarr2 could theoretically exacerbate atherosclerosis, too, by decreasing endothelial barrier function [47]: βarr2 effects this process by augmenting the endocytosis of VE-cadherin, the protein that forms adherens junctions between endothelial cells [49].

Non-inflammatory Roles for β-arrestins

Neither βarr1 nor βarr2 appear to promote conventionally inflammatory processes in a bleomycin-provoked pulmonary fibrosis model. However, each βarr promotes pulmonary fibrosis, fibroblast invasiveness, and fibroblast expression of genes related to extracellular matrix production [50]. Bleomycin-induced mortality observed in WT mice does not occur at all in βarr1^−/− or βarr2^−/− mice—even though WT, βarr1^−/− and βarr2^−/− mice demonstrate equivalent populations of macrophages, neutrophils, and lymphocytes in bronchoalveolar lavage fluid after bleomycin challenge [50]. Furthermore, inflammatory signaling in lung fibroblasts appears to be βarr-independent when it is evoked by the cytokine transforming growth factor-β. Transforming growth factor-β signals, in part, via TRAF6 and TAK1 to activate IκB kinase, and thus NFκB [51]. In response to transforming growth factor-β1, WT, βarr1^−/− and βarr2^−/− lung fibroblasts expressed equivalent amounts of the NFκB-dependent gene products [52,53] plasminogen activator inhibitor-1 and hyaluronan [50].

Reconciliation of Apparently Discordant Data

The foregoing data offer often-discordant views of βarrs and their roles in inflammatory signaling and systemic diseases: in response to LPS, bone marrow-derived macrophages from βarr2^−/− mice show greater IκB kinase activity than WT macrophages [9], but they show comparable secretion of the NFκB-dependent gene products TNF and IL-6 [38]. Whereas βarr2 reduces secretion of NFκB-dependent gene products from fibroblast-like synoviocytes [39], it does not affect secretion of NFκB-dependent gene products from lung fibroblasts [50]. Furthermore, in SMCs βarr2 augments LPS-induced IκB degradation and inflammation-associated SMC proliferation [10,32]. βarr2 rescue experiments in βarr1^−/−/βarr2^−/− MEFs (performed by distinct groups) show that βarr2 either (a) increases LPA-induced activation of nuclear NFκB [45] or (b) decreases LPS-induced TRAF6 ubiquitination and IκBα phosphorylation [9]. βarr2 exerts similarly diverse effects in vivo on multiple read-outs regulated significantly by canonical NFκB activation [35,43,54]: βarr2 attenuates LPS-induced or septic shock, over a time course of hours to days [9,38]; however, βarr2 does not affect LPS-primed, non-allergic asthma [41] and βarr2 actually augments allergic asthma [41], arterial neointimal hyperplasia and atherosclerosis [32], which develop over many weeks. The biological dissonance of these findings becomes all the more remarkable when one considers that most investigators used the same βarr knockout mice and βarr1^−/−/βarr2^−/− MEFs.

To reconcile these apparently paradoxical findings, one could in some cases invoke the diversity of endpoints and signaling mechanisms in play: whether or not the endpoint in question integrates multiple βarr-dependent signaling cascades, or rather one βarr-dependent and multiple βarr-independent signaling cascades; whether the endpoint in question derives from acute versus chronic inflammation; and so on.

Another approach to understanding the diversity of inflammatory effects for βarr2, at least, involves further consideration of βarr2-mediated scaffolding: βarr2 forms a ternary complex with the E3 ubiquitin ligase TRAF6 and the ubiquitin-specific protease 20 (USP20), and can thereby facilitate the deubiquitination of TRAF6 by USP20 [10]. Silencing of βarr2 or USP20 augments TRAF6 ubiquitination [9,10,55], but silencing both βarr2 and USP20 does not produce an additive effect on TRAF6 ubiquitination; therefore, βarr2 and USP20 appear to reduce TRAF6 ubiquitination through a shared mechanism [10]. USP20 deubiquitinates both TRAF6 and βarr2—and it is deubiquitinated βarr2 that augments the anti-inflammatory effects of USP20. De-ubiquitinated βarr2 links USP20 to TRAF6 more efficiently than ubiquitinated βarr2 [10]. To associate in ternary complex with USP20 and TRAF6, the molar concentration of βarr2 must be only ~ 10–20% the concentrations of USP20 and TRAF6; otherwise, binary complexes of βarr2/USP20 and βarr2/TRAF6 become the prevalent species [10]. Thus, at least part of the variability in βarr effects on canonical NFκB activation can be attributed to cell type- and condition-specific concentrations of βarr2, USP20, and TRAF6, and the relative concentrations of ubiquitinated and de-ubiquitinated βarr2. βarr2 could be expected to augment NFκB activity when deubiquitination of βarr2 is relatively slow or impaired, so that βarr2 cannot serve to tether USP20 to TRAF6. Conversely, βarr2 could be expected to attenuate NFκB activity when βarr2-mediated scaffolding of USP20 is important for negatively regulating NFκB activation (Fig. 2).

Fig. 2. βarr2-mediated regulation of ubiquitin-dependent NFkB signaling.

Activation of TLR4 triggers ubiquitination of βarr2 and TRAF6 {Jean-Charles, 2016 #3951}. Because ubiquitinated βarr2 fails to scaffold USP20 to TRAF6 effectively, there is less deubiquitination of TRAF6 and greater NFκB signaling when βarr2 is ubiquitinated. Upon deubiquitination by USP20, however, βarr2 avidly scaffolds USP20 to TRAF6. Consequently, USP20 deubiquitinates TRAF6 and thereby inhibits NFκB signaling. This figure was originally published in P-Y Jean-Charles et al J Biol Chem. 2016 Apr 1;291(14):7450–64.

In light of the multiple paradoxes pertaining to βarr isoform effects on inflammation, it seems fair to infer that βarrs do not serve a unidimensional—or even a consistent—role in regulating pro-inflammatory signaling. That βarrs could be targeted therapeutically for systemic inflammatory disorders seems unlikely. Nonetheless, it remains to be determined whether βarr1 or βarr2 could possibly represent anti-inflammatory therapeutic targets for particular diseases or conditions.

Abbreviations

βarr

β-arrestin

IL

interleukin

IκBα

inhibitor α of NFκB

IKK

IκB kinase

MEF

mouse embryo fibroblast

LPA

lysophosphatidic acid

TLR4

Toll-like receptor-4

TNF

tumor necrosis factor

TNFR1

TNF receptor-1

TRAF

TNFR-associated factor

7TMR

7-transmembrane receptor