Emerging roles of immunoproteasomes beyond MHC class I antigen processing

Abstract

The proteasome is a multi-catalytic protein complex whose primary function is the degradation of abnormal or foreign proteins. Upon exposure of cells to interferons (IFNs), the β1i/LMP2, β2i/MECL-1, and β5i/LMP7 subunits are induced and incorporated into newly synthesized immunoproteasomes (IP), which are thought to function solely as critical players in the optimization of the CD8⁺ T-cell response. However, the observation that IP are present in several non-immune tissues under normal conditions and/or following pathological events militates against the view that its role is limited to MHC class I presentation. In support of this concept, the recent use of genetic models deficient for β1i/LMP2, β2i/MECL-1, or β5i/LMP7 has uncovered unanticipated functions for IP in innate immunity and non-immune processes. Herein, we review recent data in an attempt to clarify the role of IP beyond MHC class I epitope presentation with emphasis on its involvement in the regulation of protein homeostasis, cell proliferation, and cytokine gene expression.

Introduction

First discovered more than 30 years ago, proteasomes have since been the subject of intense investigations across the fields of molecular, cellular, and developmental biology. The 26S proteasome is a 2.5-MDa multi-protein complex responsible for degradation of poly-ubiquitylated proteins in eukaryotic cells. Structurally, it consists of a 670-kDa catalytic 20S core particle and at least one 900-kDa regulatory 19S subunit attached to one of the core end that recognizes and unfolds substrates prior to degradation. The 20S core particle is a 28-subunit barrel-like particle composed of two outer α-rings and two inner β-rings which are each made up of seven structurally similar α- and β-subunits, respectively [1, 2]. Importantly, three of the seven β subunits (i.e., β1, β2, and β5) are fully processed during assembly and maturation and expose their N-terminal threonine residues, which serve as catalytic nucleophiles for peptide bond hydrolysis [3]. These β1, β2 and β5 subunits are responsible for the three major enzymatic activities of the eukaryotic 20S proteasome by exhibiting caspase-like, tryptic-like, and chymotryptic-like activities, respectively. The 19S regulatory subunit serves as the gate into the interior of the 20S core protease. It can be divided into two sub-structures, namely the base and the lid. The base caps the α-ring of the 20S core complex and is composed of 6 ATPases (Rpt1-Rpt6) and 2 Rpn subunits (Rpn1 and Rpn2). The lid contains up to ten non-ATPase subunits and is responsible for recognition and binding of protein substrates through their poly-ubiquitin chains [4–6]. As one of the main protein-destroying apparatuses, the 26S proteasome is involved in the regulation and maintenance of many essential cellular processes including cell growth, cell differentiation, cell signaling, DNA repair, gene transcription, apoptosis, and the generation of antigenic peptides presented by major histocompatibility (MHC) class I cell surface molecules to CD8⁺ T cells [1, 7–9].

IP molecular structure and subunit composition

Two decades ago, an alternative form of proteasomes was observed in WEHI-3 cells treated with IFN-γ. Interestingly, it was originally referred to as a “low-molecular-weight polypeptide (LMP)” complex, as it distinguished itself from the so-far-described standard form of proteasome by exhibiting a reactivity for the so-called LMP antigens [10, 11]. Historically, the LMP antigens were identified as a set of approximately 20 proteins ranging from 20 to 30 kDa that “co-precipitate” with MHC class I molecules [12]. The term LMP was specifically used to discriminate these proteins from the MHC molecules, as they all were much smaller in molecular size when loaded on a two-dimensional gel electrophoresis. Further studies have revealed that such ‘LMP complex’ was structurally very similar to the standard proteasome (SP) complex except for at least two proteins identified as β1i/LMP2 and β5i/LMP7 [11, 13–15]. Importantly, in contrast to the other proteasome genes, which are dispersed throughout the genome, the β1i/LMP2 and β5i/LMP7 genes are concentrated in the MHC class II region of the chromosome 6 (6p21.3) immediately upstream of each of the transporter-associated genes TAP-1 and TAP-2 [12, 16–18]. Like MHC class I and II cell-surface molecules, both of these two subunits are polymorphic (albeit to a lesser extent) and strongly inducible by the cytokine IFN-γ [17]. When expressed, β1i/LMP2 and β5i/LMP7 are incorporated into proteasomes in detriment of their homologous standard counterparts β1 and β5, respectively (originally designated X and Y), which are down-regulated by a post-transcriptional mechanism [19–22]. The observation that both β1i/LMP2 and β5i/LMP7 were encoded by MHC genes immediately stimulated interest in the possible function of these subunits in the regulation of the immune response by optimization of MHC class I-antigen processing. This possibility gained first support from in vitro digestion studies showing that incorporation of β1i/LMP2 and β5i/LMP7 qualitatively and quantitatively altered the set of peptides emerging from a 25-mer polypeptide containing the IE pp89 L^d-restricted YPHFMPTNL naturally processed dominant antigenic peptide [23]. Hence, to better reflect their potential function in the cellular immune response, it was proposed to rename such β1i/LMP2- and β5i/LMP7-positive complexes ‘immunoproteasomes’ [24]. A role for IP in the modulation of the pool of peptides available for MHC class I antigen presentation is also strengthened by biochemical data showing that β1i/LMP2 and β5i/LMP7 enhance substrate cleavage after basic and hydrophobic amino acid residues [25–27]. Besides, the critical involvement of these two inducible subunits in the improved generation of MHC class I ligands has been evidenced in β1i/LMP2- and β5i/LMP7-deficient mice [28, 29]. The absence of either β1i/LMP2 or β5i/LMP7 has also been reported to be frequently associated with impaired presentation of bacterial- or viral-derived epitopes to CD8⁺ cytotoxic T cells [30–32], supporting the notion that correct antigen processing requires the incorporation of these two inducible subunits into fully mature IP. A few years later, a third and last IFN-γ inducible abundant protein in IP was identified as β2i/MECL-1 and was found to replace the homologous standard subunit β2 [33, 34]. Unlike β1i/LMP2 and β5i/LMP7, β2i/MECL-1 is not linked to the MHC class II region and is encoded by a gene which is found on human chromosome 16 [35]. The contribution of β2i/MECL-1 together with β1i/LMP2 and β5i/LMP7 to the production of MHC class I epitopes has also been mostly documented in viral models [36, 37]. Since its discovery, the catalytic and peptide cleavage properties of the inducible subunits have been the main—if not the only—focus of a large number of studies all of which coming to the general conclusion that IP complexes are, in most instances, more efficient than SP in generating antigenic peptides.

Tissue expression, distribution, and regulation of IP

In immune cells

In line with the so-far-described role of the inducible subunits β1i/LMP2, β2i/MECL-1, and β5i/LMP7 in shaping the MHC class I-restricted peptide repertoire, it has been shown that the spleen has the highest IP expression and activity of any tissue [38]. This is in agreement with the observation that IP are primarily found in hematopoietic cells in vivo including professional antigen-presenting cells such as macrophages [39] and activated or resting B cells [40]. Likewise, and in contrast to most solid tumors, malignant cell lines of hematopoietic origin such as EBV-transformed B cells or multiple myeloma-derived cells are systematically enriched with the β1i/LMP2, β5i/LMP7, and/or β2i/MECL-1 subunits [41, 42]. Constitutive expression of IP has also been reported in monocyte-derived dendritic cells (DC), although there is disagreement with respect to its regulation in response to pathogens or inflammatory stimuli. Early reports argued that immature DC carry equal amounts of SP and IP, while mature DC contain only IP [43, 44]. Nonetheless, it was shown later that the relative abundance of the inducible subunits β1i/LMP2, β2i/MECL-1, and β5i/LMP7 was similar in both immature and mature DC [45, 46]. Other reports even described a down-regulation of the IP content in maturing DC following treatment with pro-inflammatory cytokines [47]. One can argue that some of the issues leading to confusion with respect to the IP expression profile in DC may be related to the lack of highly specific antibodies against the inducible subunits at the time the experiments were carried out. Recently, an interesting study re-evaluated the IP expression pattern in various cell types including DC by means of ELISA assays using specific monoclonal antibodies directed against each of the β1i/LMP2, β2i/MECL-1, and β5i/LMP7 subunits. In this work, the authors show that immature DC were mostly endowed with IP as well as with intermediate LMP7(β5i)-β1-β2 and LMP7(β5i)-LMP2(β1i)-β2 complexes whose levels remain unchanged following stimulation with either LPS or a combination of the pro-inflammatory cytokines IL-1β, IL-6, TNF-α, and PGE2 [48]. In our hands, β1i/LMP2, β2i/MECL-1, and β5i/LMP7 were only inducible at the mRNA level following engagement of Toll-like receptor (TLR) agonists such as LPS or Poly-IC through an IFN-β autocrine loop [49]. In agreement with the above-mentioned study, the increased transcription of the inducible subunits was not followed by a parallel rise in maturing DC at the protein level (Ebstein et al., unpublished observations). The unchanged β1i/LMP2, β5i/LMP7, and β2i/MECL-1 protein levels following maturation are surprising in light of the increased β1i/LMP2 and β5i/LMP7 levels in response to LPS and/or Poly-IC, as measured by quantitative real-time PCR. However, the finding may have several explanations, including increased breakdown of LMP2 (β1i)/LMP7 (β5i)/MECL-1 (β2i) in addition to increased synthesis. In any case, it is highly unlikely that the constitutive presence of IP in hematopoietic cells is the result of a persistent stimulation by pro-inflammatory cytokines in vivo, since these cells maintain their IP content in vitro in the absence of extra-cellular stimuli. The notion that the basal expression of the inducible subunits in these cells is independent of external signaling requirements is further supported by the observation that the spleen of IFN-γ-deficient mice exhibit similar levels of IP to wild-type mice [50]. The constitutive and basal expression of the inducible subunits may be explained by intracellular signaling mechanisms including a predominant unphosphorylated form of Stat1 in these cells. Indeed, it has been shown that unphosphorylated Stat1 binds IRF1 to form a complex which permanently occupies the GAS sequences of the LMP2 promoter to support transcription [51]. As part of spleen cells, it is generally thought that natural killer (NK) cells, basophils, or neutrophils express the inducible subunits in a constitutive way, although no or very little information is available in this regard.

In non-immune cells

As alluded to earlier, it was long assumed that most of the tissues of non-hematopoietic origin were normally devoid of IP, unless these tissues or those of their environment become confronted with immunological challenges, which ultimately lead to the up-regulation of each of the inducible subunits via autocrine or paracrine production of IFN-α/β or IFN-γ. Over the past few years, however, the classical paradigm of IP solely as means to an immunological end is becoming outdated, as an increasing number of studies have redefined IP expression in a wide array of non-immunological cells and organs (Table 1). This begins with the observation that the inducible subunit β5i/LMP7 is constitutively expressed in normal tissues including small intestinal epithelial cells [52], colon [53], liver [38, 54, 55], umbilical vein cells [56] as well as in placenta [57]. In addition, a fully matured IP containing all three subunits could also be purified from human kidney [58]. Noteworthy, in both colon and liver, LMP7 is accompanied by the up-regulation of the two other inducible subunits LMP2 and MECL-1 during the course of pathologies including Crohn’s disease [53] and cirrhosis [55], respectively, as a result of an alteration of the local inflammatory milieu. The nature of the stimulus leading to the basal and persistent expression of one or more of the three inducible subunits in these tissues under healthy conditions remains unclear. One could argue that such IP expression in vivo may depend on extra-cellular signals. Noteworthy, none of these cell types is a strong producer of IFN-γ, thereby excluding the possibility that the formation of IP is driven by permanent autocrine stimulation. It is more conceivable in this case that the prolonged expression of β1i/LMP2, β5i/LMP7, and/or β2i/MECL-1 in these tissues relies on the secretion of pro-inflammatory cytokines by neighboring cells such as resident macrophages and infiltrating T cells or NK cells, which secrete high levels of IFN-γ as well as infected fibroblasts producing type I IFN. However, the fact that the expression is sustained would imply a continuous secretion of IFN-γ by infiltrating DC or macrophages, making this possibility under non-pathological or non-inflammatory conditions very unlikely. Therefore, and as for hematopoietic cells, the constitutive expression of the IP subunits in these non-immune tissues may reflect a permanent activation of intracellular signaling pathways, rather than a response to changes in cytokine environment, although such notion remains to be formally addressed.

Table 1.

Tissue distribution of the inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1 under normal conditions or following pathological events

Tissue/cell type	Stimulus/disease	Constitutively expressed and/or induced subunits	Determined at	References
Rat spleen	None	LMP2, LMP7, and MECL1	Protein level	Noda et al. [38]
Human thymus	None	LMP2, LMP7, and MECL1	Protein level	Oh et al. [121]
Human macrophages	None	LMP2 and LMP7	Protein level	Haorah et al. [39]
Human B cells	None	LMP2 and LMP7	Protein level	Frisan et al. [40]
Transformed B cells	NNS	LMP2 and LMP7	Protein level	Arima et al. [78]
Human PBMC	Sjögren’s syndrome	LMP7	mRNA level	Egerer et al. [122]
	IgA nephropathy	LMP7 and MECL1	mRNA level	Coppo et al. [123]
Human monocytes	Atopic dermatitis	LMP7	mRNA level	Nagata et al. [124]
Human monocyte-derived DC	LPS, CD40L	LMP2, LMP7, and MECL1	Protein level	Macagno et al. [43]
	MALP-2	LMP2, LMP7, and MECL1	Protein level	Link et al. [44]
	LPS, Poly-IC	LMP2, LMP7, and MECL1	mRNA level	Ebstein et al. [49]
	None	LMP2, LMP7, and MECL1	Protein level	Guillaume et al. [48]
Mouse D1 DC cell line	None	LMP2, LMP7, and MECL1	Protein level	Ossendorp et al. [46]
Human small intestine	None	LMP2 and MECL1	Protein level	Visekruna et al. [125]
Mouse small intestine	None	LMP2, LMP7, and MECL1	Protein level	Kuckelkorn et al. [126]
Human colon	None	LMP7	Protein level	Visekruna et al. [53]
	Crohn disease	LMP2 and MECL1	Protein level	Visekruna et al. [52]
	Inflammation	LMP2	Protein level	Fitzpatrick et al. [127]
Human liver	None	LMP7	Protein level	Visekruna et al. [125]
Mouse liver	LPS	LMP7	Protein level	Seifert et al. [74]
Human liver	Cirrhosis, chronic liver disease	LMP2, LMP7	Protein level	Vasuri et al. [55]
Human liver	Steatohepatitis	LMP2, LMP7, and MECL1	Protein level	French et al. [77]
Rat liver	Refeeding following fasting	LMP2 and LMP7	Protein level	Ushiama et al. [128]
	None	LMP2, LMP7, and MECL1	Protein level	Noda et al. [38]
Mouse liver	DCC-induced MDB	LMP2, LMP7, and MECL1	Protein level	Bardaf-Gorce et al. [129]
Human umbilical veins	None	LMP7	mRNA level	Loukissa et al. [56]
Human placenta	None	LMP7	mRNA level	Roby et al. [57]
Human kidney	None	LMP2, LMP7, and MECL1	Protein level	Baldovino et al. [58]
Rat lung	None	LMP7	Protein level	Noda et al. [38]
Mouse retina	Injury	LMP2 and LMP7	Protein level	Ferrington et al. [61]
Human retina	Age-related macular degeneration	LMP2 and LMP7	Protein level	Ethen et al. [130]
Mouse brain	EAE	LMP7	Protein level	Seifert et al. [74]
Mouse brain	Injury	LMP2 and LMP7	Protein level	Ferrington et al. [61]
Human brain	HIV Infection human	LMP2	Protein level	Gelman et al. [131]
	Glioblastoma human	LMP2	Protein level	Piccinini et al. [62]
Human neurons	Alzheimer disease	LMP7, LMP2	Protein level	Mishto et al. [63]
		LMP7, MECL1	Protein level	Nijholt et al. [64]
	Aging	LMP7, LMP2	Protein level	Mishto et al. [63]
	Multiple sclerosis	LMP7, LMP2	Protein level	Mishto et al. [132]
	Epilepsy	LMP7, LMP2	Protein level	Mishto et al. [133]
Mouse neurons	Huntington’s disease	LMP7, LMP2	Protein level	Diaz-Hernandez et al. [65]
		LMP7, LMP2	Protein level	Diaz-Hernandez et al. [66]
Rat neurons	Aging	LMP7	Protein level	Gavilan et al. [67]
Human muscle	Myofibrillar myopathy	LMP7, LMP2, and MECL1	Protein level	Ferrer et al. [76]
Rat muscle	Aging	LMP7, LMP2	Protein level	Husom et al. [134]
Human endothelial cells	NO	LMP7, LMP2	Protein level	Kotamraju et al. [135]
Mouse ocular lens	None	LMP7, LMP2, and MECL1	mRNA level	Singh et al. [59]
Human skin	JASL	LMP7(G197V)	Protein level	Kitamura et al. [79]
Mouse heart	Enterovirus (CVB3)	LMP7, LMP2	Protein level	Opitz et al. [80]

Tissue distribution of the inducible subunits β1i/LMP2, β5i/LMP7 and β2i/MECL-1 under normal conditions or following pathological events

Most surprisingly, IP has been reported to be found or induced in immune-privileged sites including the eye [59–61] and the brain [62], although the literature on the precise cellular distribution of the inducible subunits in these organs remains somehow conflicting (Table 1). In particular, the capacity of neurons to induce or to form IP is still a matter of debate. Earlier studies originally suggested that IP are absent in neurons under normal conditions but may be formed over the course of neurodegenerative diseases such as Alzheimer [63, 64] and Huntington [65, 66]. This notion is further supported by subsequent reports emphasizing that the increased expression of each of the inducible subunits in neurons is a classical feature of brain aging [63, 67, 68]. It was later shown that IP formation in the brain in response to infection with LCMV occurs in microglia cells but not in neurons [69]. In this work, Kremer and colleagues demonstrate that, although being induced at a transcriptional level, β1i/LMP2, β2i/MECL-1, and β5i/LMP7 accumulate as precursors in inflamed neurons without being incorporated into mature IP. The basis of these discrepancies is unclear and might in part be related to the nature of the stimulus and/or disease. For instance, because of the increased reactivity of microglia, astrocytes and innate immune cells of the brain that occurs during normal aging [70], it is argued that chronic neuroinflammation involving the continuous production of multiple cytokines such as IL-1, IL-6, TNF-α, and TGF-β is the main underlying cause of IP induction in neurons over the course of neurodegenerative diseases. The inability of neurons to assemble the up-regulated β1i/LMP2, β5i/LMP7, and β2i/MECL-1 precursors into fully mature IP following LCMV infection is intriguing. Although the authors in this study did not find any substantial impairment of the expression level of the critical IP assembly factor POMP in these cells, these data suggest that neurons still lack other yet unidentified partners required for proper IP formation under LCMV infection. Alternatively, one could also argue that such defect may be seen as a unique feature of LCMV infection rather than a characteristic of neurons. In any case, the observation that IP are formed in neurons following chronic inflammation but not following LCMV infection suggests a role for the inducible subunits beyond MHC class I processing.

Functions of IP in innate immunity and non-immune processes

While classically linked to MHC class I antigen presentation, recent evidence from knockout mice and inhibition studies on the role of the β1i/LMP2, β5i/LMP7, and β2i/MECL-1 inducible subunits have substantially expanded the functional repertoire of IP. Indeed and as listed in Table 2, an involvement of these subunits has been described in unexpected cellular processes such as protein homeostasis, cell proliferation, and cell signaling (Table 2).

Table 2.

List of the so-far reported functions for the inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1, which are not directly connected to MHC class I antigen processing

Function	Cell type/tissue	Implicated subunit (s)	Shown in/by	Phenotype	References
Protein homeostasis	Brain and liver	LMP2	LMP2-/− mice	Increased level of protein oxidation	Ding et al. [93]
	Endothelial cells	LMP2 and LMP7	LMP2 and LMP7 antisense oligonucleotides	NO accumulation	Kotamraju et al. [135]
	Retina cells	MECL1 and LMP7	MECL1−/− and LMP7−/− mice	Oxidation-induced cell death	Hussong et al. [60]
	MEF/brain/liver	LMP7	LMP7−/− mice	Accumulation of oxidized and poly-ubiquitylated proteins	Seifert et al. [74]
	MEF	LMP2	LMP2 siRNA	Accumulation of oxidized proteins	Pickering et al. [92]
	NNS fibroblasts	LMP7	Endogenous expression of a PSMB8 (G201V) mutant	Accumulation of oxidized and poly-ubiquitylated proteins	Arima et al. [78]
	Transformed B cells	LMP7	Endogenous expression of a PSMB8 (G197V) mutant	Accumulation of poly-ubiquitylated proteins	Kitamura et al. [79]
	Cardiomyocytes/heart muscle	LMP7	LMP7−/− mice	Accumulation of oxidized and poly-ubiquitylated proteins	Opitz et al. [80]
Cytokine production	BM-derived DC	LMP2	LMP2−/− mice	Impaired production of IFN-α, IL-6, IL-1β and TNF-α upon infection	Hensley et al. [113]
	PBMC from patients with RA	LMP7	LMP7-specific inhibitor (PR-957)	Decreased production of IL-23, IL-6 and TNF-α	Muchamuel et al. [114]
	Macrophages	LMP2, LMP7, and MECL1	LMP2−/−, LMP7−/−, MECL1−/− and LMP7/MECL1−/− mice	Reduced NO levels following LPS stimulation	Reis et al. [117]
	Colon from mice with DSS-induced colitis	LMP7	DSS-induced colitis in LMP7−/− mice	Reduced expression of inflammatory cytokines mediated by NF-κB	Schmidt et al. [82]
		LMP7	LMP7-specific inhibitor (PR-957)	Impaired production of IL-23 and IL-1β	Basler et al. [81]
		LMP2, MECL1, and LMP7	LMP2−/−, MECL1−/− and LMP7−/− mice	Impaired induction of TNF-α, IL-1β, IL-23 and IL-17.	Basler et al. [81]
	Serum from patients with NNS	LMP7	Missense mutation G201V	Increased production of IL-6	Arima et al. [78]
	Skin from patients with NNS	LMP7	Missense mutation G197V	Increased production of IL-6	Kitamura et al. [79]
	Blood from patients with CANDLE syndrome	LMP7	Nonsense mutation C135X	Increased induction of IP-10, MCP-1 and RANTES	Liu et al. [118]
Cell proliferation	T cells	MECL1 and LMP7	MECL1−/− and LMP7−/− mice	Hyperproliferation in vitro in response to mitogens	Caudill et al. [105]
		MECL1	MECL1−/− mice	Enhanced CD4/CD8 T-cell ratios	Zaiss et al. [100]
		LMP2, MECL1, and LMP7	LMP2−/−, MECL1−/− and LMP7−/− mice	Hypoproliferation in WT hosts	Moebius et al. [102]
	Multiple myeloma cells	LMP2	LMP2-specific inhibitor (IPSI-001)	Decreased cell proliferation	Kuhn et al. [103]
		LMP7	LMP7-specific inhibitor (PR-924)	Decreased cell growth	Singh et al. [104]
Cardiac muscle regulation	Myocytes	LMP2	LMP2−/− mice	Diabetic cardiomyopathy	Zu et al. [136]
	Heart	LMP2	LMP2−/− mice	Lower cardiac mass and heart rate	Cai et al. [137]
Brain function	Brain	LMP2	LMP2−/− mice	Increased degree of motor function	Martin et al. [138]
Vision transmission	Retina	MECL1 and LMP7	MECL1−/− and LMP7−/− mice	Bipolar cell responses defects	Hussong et al. [139]
Apoptosis	Atherosclerotic lesion-derived cells	LMP7	LMP7 siRNA	Increased sensitivity to apoptosis to IFN-γ	Yang et al. [110]

List of the so-far-reported functions for the inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1, which are not directly connected to MHC class I antigen processing

A role for IP in the disposal of accumulating damaged proteins upon inflammation and/or cell stress

By removing abnormal proteins or proteins that are no longer needed in the cell, proteasomes play an essential role in the regulation of cellular proteostasis. As such, aberrant proteins often tend to aggregate when proteasome function is impaired, thereby leading to intracellular accumulation of large ubiquitin-positive structures in the cytoplasm. Although prolonged protein aggregation is a hallmark of aging or age-related diseases such as Alzheimer [63, 64], transient protein aggregation is a physiological process that takes place in various cell types following cytokine stimulation or cell stress. This is particularly obvious in professional antigen-presenting cells including DC and macrophages whose maturation programs following microbial stimuli are characterized by a transient accumulation of large cytosolic ubiquitin aggregates (also referred to as DALIS for dendritic cell aggresome-like induced structures) [71, 72]. A similar tendency was observed in response to stress and starvation in non-immune cells including fibroblasts whose accumulating ubiquitin protein-conjugates in this case were referred to as aggresome-induced like structures (ALIS) to distinguish them from DALIS [73]. Importantly, both DALIS and ALIS require protein synthesis and it is assumed that their formation occurs as a consequence of excess levels of defective ribosomal products (DRiPs) that temporary exceed the degradation capacity of proteasomes. For many years, the respective roles of SP and IP in the degradation of DRiPs remained unclear and it is only recently that increasing evidence supports the concept that IP are more efficient than SP at eliminating damaged ubiquitylated cellular proteins that accumulate upon physiological and/or pathological stimuli. Indeed, we have recently shown that mouse embryo fibroblasts (MEF) from β5i/LMP7^−/− mice are less prone than those of their wild-type counterparts in clearing ALIS following IFN-γ stimulation [74]. In this study, we demonstrate that the increased capacity of IP to cope with ALIS essentially relied on its higher ability to degrade K48-linked poly-ubiquitylated substrates, when compared to SP. Up to now, the mechanisms by which the incorporation of the inducible subunits allows IP to degrade ubiquitin substrates faster than SP are unclear and warrant further investigations. One possible explanation may be the increased chymotryptic activity observed following incorporation of β5i/LMP7. This, in turn, might be due to a conformational change that in a way favors the entry of poly-ubiquitylated substrates into the catalytic core, as previously suggested [75]. In any case, our findings point to an unexpected role for IP in the accelerated degradation of poly-ubiquitylated substrates, which implies that IP-expressing tissues are better protected from disruption of protein homeostasis than tissues endowed with SP.

The observation that DC and macrophages are enriched with the inducible subunits even at the immature stage is of high interest and does not necessarily reflect a predisposing role in the adaptive immune response by generating MHC class I peptides. Indeed, DC and macrophages establish the first line of defense against microbial pathogens and, as such, are frequently confronted with pathogen-associated molecular patterns (PAMP), which initiate maturation and DALIS formation following their binding to Toll-like receptors. It is tempting to speculate that the constitutive presence of IP even before any stimulation allows these cells to rapidly cope with DALIS following activation. This notion is reinforced by the fact that DALIS are apparently cleared much faster in DC than ALIS in MEF, which need to up-regulate IP [71, 74]. In spite of the availability of numerous knockout models for the three inducible subunits, the role for IP in the removal of DALIS remains elusive and has to be formally addressed. The fundamental role of the IP in innate immunity by preserving protein homeostasis upon infection or insult is also strengthened by the fact that B and T cells, which have a very limited role in the generation of MHC class I epitopes, constitutively express β1i/LMP2, β5i/LMP7, and β2i/MECL-1. An increasing number of studies have documented a correlation between IP over-expression and a flurry of diseases characterized by accumulations of abnormal proteins including myofibrillar myopathy [76] or steatohepatitis [77]. It is conceivable that the increased expression of the inducible subunits over the course of such diseases reflects the permanent need of the damaged tissues to restore protein homeostasis. Supporting this notion, we recently showed that mice lacking β5i/LMP7 fail to clear protein aggregates that accumulate in the brain following induction of experimental autoimmune encephalitis (EAE) and in the liver in response to LPS [74]. Further evidence also includes the recent report of a missense mutation in the β5i/LMP7 gene (PSMB8) (G197V and G201V in the isoforms E1 and E2, respectively), which is associated with IP formation disturbance and concomitant increased accumulation of poly-ubiquitylated proteins in patients with Nakajo-Nishimura syndrome (NNS) (also refereed as to Japanese auto-inflammatory syndrome with lipodystrophy, JASL) [78, 79]. Similar observations were made in other experimental diseases including viral models, whereby infection of β5i/LMP7^−/− mice with coxsackievirus B3 (CVB3) results in an exacerbated myocarditis, which occurs as a consequence of a lack of protection against CVB3-mediated protein damage [80]. These studies appear to be in conflict with earlier reports showing that IP deficiency is associated with attenuation of disease progression in an experimental colitis auto-immune model [81, 82]. These contrasting data suggest that, depending on both disease model and tissue-specificity, IP can exert either beneficial or negative effects.

Binding of PAMP to TLR stimulates not only the release of specific cytokines by DC and/or macrophages but also reactive oxygen species (ROS) and nitric oxide (NO) [83]. Due to their high abundance in the cell, proteins are major targets for such oxidative stress and become irreversibly modified through amino acid carbonylation, disulfide bond cross-linking and peptide backbone cleavage [84]. Of note, oxidative stress is not restricted to immune cells but occurs per se in all tissues at a basal level, mostly because of the side reactions of the mitochondrion electron transport that occur with molecular oxygen, thereby generating ROS such as superoxide anions (O₂ ⁻), hydrogen peroxide (H₂O₂) and hydroxyl radicals (HO) [85]. The removal of oxidized proteins by proteasomes is clearly established [86–91], and recent work suggests that IP are more prone than SP at eliminating them [92]. An important observation is also made in the brain and the liver of β1i/LMP2^−/− mice exhibiting higher levels of protein carbonyls upon aging than those of their wild-type littermates of the same genetic background [93]. An additional insight in this field has emerged from our recent study showing that protein carbonyls accumulating following IFN-γ treatment are simultaneously subjected to poly-ubiquitylation [74]. Interestingly, this notion is somehow in conflict with other reports arguing that the proteasome-mediated degradation of oxidized proteins may occur independently of that of poly-ubiquitylated substrates [92, 94, 95]. The basis for these discrepancies may be explained by the co-existence of distinct pools of oxidized proteins. However, the mechanisms whereby such substrates are sorted into either the 20S- or the 26S-dependent degradation pathway remain ill-defined. Most likely, the particularly short half-life of IFN-γ-induced oxidized proteins does not constitute a critical criterion for their targeting to the 26S complex, as short-lived oxidant-damaged proteins have also been shown to be efficiently degraded by 20S complexes [96–98]. Rather, the fate of oxidized proteins may be determined by the nature and/or strength of the stimulus itself. It is, indeed, conceivable that IFN-γ favors the breakdown of protein carbonyls by 26S proteasomes because of the concomitant increased activation of the ubiquitin-conjugation system in response to this cytokine. In any case, the precise contribution of each pathway to the disposal of these substrates warrants further investigations. Finally, a protective role of β1i/LMP2, β5i/LMP7 and β2i/MECL-1 against cellular oxidative damage is also reinforced by the observation that these subunits are constitutively expressed in cells which are frequently challenged with reactive oxygen radicals such as phagocytic cells (i.e., DC and macrophages) during the respiratory burst process.

A role for IP in cell growth and differentiation

Although it is clearly established that ubiquitin- and proteasome-dependent proteolysis is one of the key mechanisms underlying cell cycle control, the influence of the incorporation of β1i/LMP2, β5i/LMP7, and/or β2i/MECL-1 in this process remained relatively understudied over the past 10 years. It is only recently that gene knockout studies in mice have revealed a complex and unexpected role of the inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1 in regulating cell proliferation and differentiation. This is particularly obvious in cells and tissues that constitutively express IP including hematopoietic cells such as T cells. A prime example of the involvement of IP in cell proliferation originally comes from the observation that both β1i/LMP2- and β2i/MECL-1-deficient mice exhibit significant decreased numbers of naive CD8⁺ T cells, when compared to their wild-type counterparts [29, 99]. The assumption that the inducible subunits intrinsically regulate T-cell expansion is further supported by the detection of enhanced CD4/CD8 T-cell ratios in β2i/MECL-1^−/− mice [100]. Subsequent studies have also shown that T cells from β1i/LMP2^−/−, β5i/LMP7^−/− or β2i/MECL-1^−/− mice fail to proliferate and survive when transferred into virus-infected wild-type mice [99, 101, 102], thereby suggesting that IP expression may be required for expansion and maintenance of T cells. In line with the notion that IP favor cell proliferation is the fact that β5i/LMP7, β1i/LMP2 and to a lesser extent β2i/MECL-1 are often over-expressed in various tumors of both hematopoietic and non-hematopoietic origin [48]. These findings even provided a rationale for targeting the inducible subunits β1i/LMP2, β5i/LMP7, or β2i/MECL-1 as a potential therapeutic strategy for several types of IP-positive tumors. Indeed, the use of the β1i/LMP2 specific inhibitor (IPSI-001) showed some success in diminishing cell proliferation in myeloma patient samples in vitro [103]. Interestingly, similar observations were made in vivo when inhibiting the β5i/LMP7 subunit with the PR-924 inhibitor [104], confirming the close correlation between mammalian IP expression and carcinogenesis. Nevertheless, the view that IP expression unconditionally supports cell proliferation is hampered by an earlier report showing that T cells from double knockout mice lacking both β2i/MECL-1 and β5i/LMP7 hyperproliferate in vitro in response to polyclonal mitogens [105]. These findings are not necessarily conflicting, since depending on their target substrates, proteasomes may realize negative as well as positive control of cell cycle [106]. The eukaryotic cell cycle is precisely orchestrated and regulated by multiple cyclin and cyclin-dependent kinases (CDK) which form cyclin/CDK complexes that are negatively regulated by a number of CDK inhibitors including p27^Kip1 and p21^Cip1/WAF1 [107]. Proteasomes are involved in this process by regulating the protein turnover of the CDK inhibitors whose degradation favors the prolongation of cell cycle progression [108]. Given that IP destroy poly-ubiquitylated substrates and protein carbonyls faster than SP, it is highly likely that CDK inhibitors exhibit a particularly high protein turnover in cells and tissues constitutively expressing the inducible subunits, thereby resulting in increased susceptibility to cell proliferation. This notion is in agreement with the observation that defects or down-regulation of p21 have been linked to the development of various cancers and contribution to tumor progression [109]. Another major characteristic of malignant transformation of epithelial cells is the inactivation or down-regulation of p53, which is associated with the loss of cell cycle control and apoptosis evasion. To date, the role of IP in the breakdown of p53 remains unknown, but since p53 undergoes poly-ubiquitylation and/or is susceptible to oxidation, it is conceivable that it is degraded more quickly in tumor cells over-expressing β1i/LMP2, β5i/LMP7, and β2i/MECL-1, thereby participating in cell growth. The decreased capacity of cells expressing IP to undergo cell cycle arrest is also supported by a recent report showing that β5i/LMP7 gene silencing facilitates atherosclerotic lesion-derived cells to enter apoptosis in response to IFN-γ [110]. A high protein turnover in proliferating precursor cells also seems to be a major characteristic of the cell differentiation process. Indeed, it has been shown that rapid protein turnover of ERK3 by 26S proteasomes is required to achieve successful differentiation of PC12 and C2C12 cells into the neuronal and muscle lineage, respectively [111]. Such a finding suggests that IP may facilitate this process, a concept that would be in line with the fact that the high expression levels of β1i/LMP2, β5i/LMP7, and β2i/MECL-1 are maintained in fully differentiated cells such as DC and macrophages.

A role for IP in cell signaling and the regulation of cytokine production

A couple of studies hint at an unexpected role for the inducible subunits β1i/LMP2, β5i/LMP7 and β2i/MECL-1 in the regulation of the production of pro-inflammatory cytokines. Visekruna and colleagues have shown that the prominent expression of IP found in intestinal tissue from patients with Crohn’s disease or ulcerative colitis was associated with enhanced activation of the NF-κB cell signaling pathway [53]. The transcription factor NF-κB is a well-known master regulator of many inflammatory cytokine genes and its activation by proteasomes is clearly established. Indeed, NF-κB is actively inhibited and sequestered in the cytoplasm when bound to proteins of the IκB family IκBα, β and γ. NF-κB activation involves phosphorylation of IκB which promotes its K48-linked poly-ubiquitylation and degradation by the 26S proteasome, thereby unmasking a nuclear targeting sequence of the NF-κB protein [112]. The role of the inducible subunits in this process has long remained unclear but the findings of the above-mentioned study leads to the hypothesis that IP may degrade IκB proteins faster than SP resulting in enhanced nuclear translocation of NF-κB. In our recent study, we could confirm that the protein turnover of IκB was much higher in cells over-expressing the three inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1 than cells expressing predominantly SP [74]. Moreover, this notion is further strengthened by the observation that bone marrow-derived DC from β1i/LMP2-deficient mice exhibit substantial reduced levels of IFN-α, IL-6, IL-1β, and TNF-α upon infection with influenza A virus [113]. Likewise, an impaired activation of NF-κB has been observed in β5i/LMP7^−/− mice over the course of CVB3 infection [80]. Such results are of high interest, as they confirm our previous observations implying that the increased capacity of IP to breakdown K48-linked poly-ubiquitylated proteins is not restricted to IFN-γ-induced DRiPs and ALIS but can be extended to other proteasomal substrates. In addition, it was shown that the specific disruption of the β5i/LMP7 subunit by the PR-957 inhibitor results in a block of the secretion of IL-23, TNF-α, and IL-6 from stimulated peripheral blood mononuclear cells (PBMC), which is further accompanied by attenuation of the pathological symptoms and disease progression in a mouse model of rheumatoid arthritis (RA) [114]. Surprisingly, and in spite of the fact that NF-κB regulates multiple critical cytokines involved in the pathogenesis of RA [115, 116], the authors failed to detect any impairment of the NF-κB signaling pathway in mice treated with PR-957. A role for β5i/LMP7 in the regulation of cytokine gene expression in a NF-κB-independent manner is also supported by a later study in which β5i/LMP7 disruption correlates with impaired production of IL-23 and IL-1β in a DSS-induced colitis model [81]. These results, however, are in contrast to other reports where the impaired production of pro-inflammatory cytokines and chemokines in β5i/LMP7^−/− mice was clearly associated with a substantial defect in the activation of NF-κB [80, 82]. Possible explanations why these studies have conflicting findings might be the tissue-specific contribution to different IP-functions or the simultaneous activation of multiple signaling pathways by β5i/LMP7. In line with the latter concept, it has recently been shown that each of the inducible subunits is capable of influencing regulatory signaling molecules other than NF-κB including TRIF and TRAM [117]. Both proteins are adaptor molecules, which play an essential role in the MyD88-independent signaling of TLR-4 and, as such, are critical regulators of the LPS response. In this study, macrophages from β1i/LMP2^−/−, β5i/LMP7^−/−, β2i/MECL-1^−/− and LMP7^−/−(β5i)/MECL-1^−/−(β2i) knockout mice were far less able to produce NO in response to LPS, when compared to macrophages from wild-type mice. Therefore, the authors argued that the decreased activation of this pathway in IP-deficient mice is the result of a decreased degradation rate of both TRIF and TRAM, although this assumption remains to be proven. Interestingly, recent data also suggest a potential implication of IP in the regulation of the IFN signaling pathway, as determined by the elevated transcription of the IP-10, MCP-1, and RANTES genes detected in patients with CANDLE syndrome, a disease that has recently been associated with a nonsense mutation in the PSMB8 gene [118]. In addition, it seems that IP are also capable of influencing the p38 signaling pathway, as shown by increased phosphorylation of p38 and subsequent induction of IL-6 in response to PMA/ionomycin in cells over-expressing a non-functional β5i/LMP7 mutant [78, 79]. Again, the mechanisms whereby β5i/LMP7 regulates IL-6 production via p38 are far from being elucidated. It is proposed that, in contrast to the NF-κB pathway, the alteration of the p38 pathway is an indirect rather than a direct effect of β5i/LMP7 deficiency. Because p38 is influenced by the intracellular redox homeostasis [119], the investigators of this study suggest that the promotion of IL-6 via p38 occurs as a consequence of the inability of β5i/LMP7-deficient 20S and/or 26S proteasomes to cope with oxidized proteins. Taken together, these studies emphasize the role of the inducible subunits in the optimal activation of various signal transduction pathways by increasing the protein turnover of critical signaling factors. In this regard, and since IκBα is faster degraded by IP than by SP, it is tempting to speculate that the constitutive presence of IP in DC or macrophages allow these cells to quickly react to external stimuli by engaging rapidly the NF-κB transduction pathway. This, in turn, leads to the rapid secretion of NF-κB-dependent cytokines in response to pathogens. Another role of the inducible subunits in the acceleration of signal transduction may include T-cell activation which requires signaling through the T-cell receptor (TCR) and CD28 co-stimulatory receptor. Indeed, CD28 co-stimulation results in the poly-ubiquitylation and subsequent degradation of Cbl-β, which eliminates the negative regulators and allows the expression of pro-inflammatory cytokines and their receptors [120]. It is likely that such signaling is facilitated by the constitutive presence of IP in T cells, which quickly eliminate Cbl-β, thereby favoring rapid T-cell activation.

Summary and conclusion

Collectively, the findings summarized in this review highlight a critical role for each of the inducible subunits β1i/LMP2, β5i/LMP7, and β2i/MECL-1 far beyond simple MHC class I antigen processing, as originally assumed. Indeed, IP are involved in all functions already fulfilled by SP including the regulation of protein homeostasis, cell proliferation, cell differentiation, cell signaling, gene transcription, and MHC class I antigen processing. The important observation that IP are more efficient than SP at degrading poly-ubiquitylated proteins has considerably expanded our understanding of how the incorporation of the inducible subunits into proteasome-complexes can modulate these functions. As illustrated in Fig. 1, by accelerating the degradation rate of all poly-ubiquitylated substrates, IP do not only promote the supply of MHC class I peptides but also induce remarkable changes in various other cellular functions by affecting the protein turnover of critical regulators such as IκBα. Depending on the influence of the degradation of such substrates, IP may represent positive or negative regulators. As such, the β1i/LMP2, β5i/LMP7, and β2i/MECL-1 subunits become attractive targets for therapeutic purposes against the development of pathologies including cancer, viral diseases, and neurodegenerative diseases. A particular high degree of complexity in the regulation of the half-life of poly-ubiquitylated proteins is brought by the existence of IP intermediate types containing one or two of the three inducible subunits. Therefore, further studies will need to be conducted in order to elucidate the precise role of each inducible subunit in these processes.

Fig. 1.

Immune and non-immune functions of the incorporated β1i/LMP2, β5i/LMP7, and β2i/MECL-1 inducible subunits. The proteasomal degradation of misfolded proteins (DRiPs) as well as regulator factors involved in cell cycle and cell signaling is a permanent ongoing process. The SP is less efficient than the IP at eliminating poly-ubiquitylated substrates and its capacity can be overloaded, thereby resulting in the accumulation of harmful protein aggregates (ALIS). The constitutive presence of IP (or its formation in response to IFN-γ) is accompanied by an acceleration of the degradation rate of all poly-ubiquitylated proteins which restores (or preserves) protein homeostasis. The increased protein turnover is not limited to DRiPs but also deals with other poly-ubiquitylated proteins including IκBα, p53 and/or p21 cip1/WAF1 whose increased degradation rate has impacts on cell cycle regulation and cytokine gene expression

Abbreviations

ALIS

Aggresome-like induced structures

BM

Bone marrow

CVB3

Coxsackievirus B3

DALIS

Dendritic aggresome-like induced structures

DC

Dendritic cells

DSS

Dextran sulfate sodium

EAE

Experimental autoimmune encephalitis

EBV

Epstein–Barr virus

IP

Immunoproteasome

JASL

Japanese auto-inflammatory syndrome with lipodystrophy

LCMV

Lymphochoriomeningitis virus

LMP

Low-molecular-weight polypeptide

LPS

Lipopolysaccharide

MEF

Mouse embryo fibroblasts

NO

Nitric oxide

NNS

Nakajo-Nishimura syndrome

PA28

Proteasome activator 28

PAMP

Pathogen-associated molecular patterns

PBMC

Peripheral blood mononuclear cells

POMP

Proteasome maturation protein

RA

Rheumatoid arthritis

ROS

Reactive oxygen species

SP

Standard proteasome

TAP

Transporter associated with antigen processing

TLR

Toll-like receptor