ReviewCell Biology

The Emerging Role of Linear Ubiquitination in Cell Signaling

See allHide authors and affiliations

Science Signaling  20 Dec 2011:
Vol. 4, Issue 204, pp. re5
DOI: 10.1126/scisignal.2002187

Abstract

The covalent attachment of ubiquitin molecules to target proteins is a posttranslational modification that is involved not only in signaling processes leading to protein degradation but also in those resulting in activation, proliferation, and cell death. Ubiquitination is a versatile regulation mechanism: In addition to single ubiquitin molecules, chains consisting of several ubiquitin moieties can also be attached to target proteins. The functional outcome of polyubiquitination depends on the lysine residue within ubiquitin that is used for chain elongation. The reason for this is that the particular linkage between two ubiquitin moieties through a specific lysine residue of one ubiquitin and the C terminus of the other ubiquitin creates a unique binding surface that is specifically recognized by specialized ubiquitin-binding domains. New evidence indicates that besides the seven internal lysine residues of ubiquitin, the N terminus of ubiquitin can also be used as an attachment point, thereby generating linear or M1-linked polyubiquitin chains. An E3 complex consisting of HOIL-1, HOIP, and Sharpin specifically generates such M1-linked ubiquitin chains in the context of various cellular signaling pathways that regulate cell activation and death, and it was named linear ubiquitin chain assembly complex (LUBAC). In this Review, we focus on the biochemistry and physiological role of linear ubiquitin chains generated by LUBAC. We summarize the function of linear ubiquitin chains in signaling pathways downstream of diverse cellular signaling events and provide an outlook on promising future directions of research.

Introduction

Appropriate responses to external stimuli are essential for a cell to function properly and to regulate the balance between cell death and survival (14). The signaling processes that transduce an external signal to the interior of a cell consist of complex cascades of events that are tightly regulated. In the case of cytokine signaling, one of the first steps is the recruitment of cytoplasmic proteins to the intracellular domains of the receptor (5). The molecular interactions in these complexes often also induce conformational changes that are required for the activation of particular signaling proteins and enable proteins to modify themselves and nearby substrates.

Two well-studied modifications are phosphorylation and ubiquitination. Both can serve as recognition signals and can regulate, activate, or inactivate proteins within signal transduction cascades (6, 7). Whereas phosphorylation of proteins is catalyzed by kinases, the covalent attachment of ubiquitin to lysine residues or the N terminus of a target protein is mediated by the concerted action of three different classes of proteins: the ubiquitin activating enzyme (E1), the ubiquitin-conjugating enzyme (E2), and the E3 ubiquitin-protein ligase (810). First, ubiquitin is activated in an ATP–dependent manner by the formation of a thioester bond between the C-terminal carboxyl group of ubiquitin and the catalytically active cysteine residue of the E1 enzyme. Subsequently, ubiquitin is transferred to a cysteine residue within the active center of the E2 enzyme. Finally, ubiquitin is attached to the substrate protein in a process mediated by the E3 ligase. Depending on the class of E3 ligase, this may involve ubiquitin transfer onto the E3 ligase, as in the case of HECT (homologous with E6-associated protein C terminus)–domain–containing proteins (11), or directly to the target protein, with the E3 ligase facilitating the transfer from the E2 enzyme to the substrate. The latter mechanism is employed by E3 ligases of the RING (really interesting new gene) or U-box classes (12, 13). Because seven of the 76 amino acids of ubiquitin are lysines, it can itself be a target for ubiquitination, thereby enabling the formation of polyubiquitin chains, which differ in linkage type depending on the lysine residue (K6, K11, K27, K29, K33, K48, or K63) used in chain formation. This creates a multitude of potential modifications, of which the best-studied linkage types are K48- and K63-linked chains. Whereas K48-linked chains are involved in proteasomal degradation of target proteins (14), K63-linked chains play a role in protein trafficking, DNA repair, and other nonproteolytic signaling pathways (1517). All potential chain types exist (18), and they differ not only structurally but also functionally (1923). In addition to the internal lysine residues, the N terminus of ubiquitin can be used for chain formation, thereby generating linear or M1-linked chains (24). Because ubiquitin is encoded either by polyubiquitin genes or as a fusion with other proteins, translation could lead to the formation of M1-linked chains as well. However, these chains are cotranslationally cleaved into single ubiquitin moieties by specific deubiquitinases (DUBs) (2529). The realization that linear ubiquitin chains are present in cells and are indeed the product of the activity of a particular E3 ligase was therefore both unexpected and exciting.

Generation of Linear Ubiquitin Chains

The linkage type of a ubiquitin chain formed in RING E3 ligase–mediated processes is generally defined by the E2 enzyme involved in its generation (3032). It was therefore surprising when in 2006 Iwai and colleagues described a 600-kD E3 complex that exclusively generated linear ubiquitin chains independently of the E2 enzyme used and which they therefore named linear ubiquitin chain assembly complex (LUBAC) (24). Two components of this complex were initially characterized: heme-oxidized IRP2 ubiquitin ligase-1 (HOIL-1) and HOIL-1–interacting protein (HOIP) (24). We and others showed that this complex also contains the Shank-associated RH domain interactor (Sharpin) (3335) (Fig. 1). Sharpin and HOIL-1 share high sequence similarity (36), and it has been suggested that their genes emerged by fusion of a Sharpin-like gene with a RING in-between RING (RBR)–family gene (37).

Fig. 1

Schematic representation of the domain structures of Sharpin, HOIL-1, and HOIP. The UBL and the NZF at the C terminus of Sharpin and at the N terminus of HOIL-1, respectively, show sequence homology (45% identity) (36). ZnF, zinc finger; NZF, Npl4 zinc finger; UBL, ubiquitin-like domain; UBA, ubiquitin-associated domain; IBR, in-between RING domain.

CREDIT: Y. HAMMOND/SCIENCE SIGNALING

Both HOIL-1 and Sharpin can interact with HOIP through their respective ubiquitin-like (UBL) domains (24, 3335), making it likely that HOIP is the central protein of the complex. Considering the sizes of the three components (57 kD for HOIL-1, 120 kD for HOIP, and 40 kD for Sharpin), different combinations for the formation of a 600-kD complex seem possible. However, the actual stoichiometry of this complex has not been resolved, and the existence of complexes consisting of only two of the three proteins cannot be ruled out at this time. In addition, depending on the cell type, varying amounts of Sharpin and HOIL-1 are not associated with HOIP (25, 33, 35), and these proteins may therefore exert LUBAC-independent functions. In line with this notion, several studies investigating the effects of these components on their own in different cellular contexts have been published (3849). Although it is possible that the three proteins also have functions outside of their role in generating linear ubiquitin as part of LUBAC, these publications have not excluded a contribution of the other two LUBAC components in the observed effects. Thus, it will be interesting to investigate the role of LUBAC in these different settings. For the purpose of this Review, we will focus on the linear ubiquitin–generating activity of LUBAC.

The fact that LUBAC exclusively generates M1 ubiquitin linkages is supported by its ability to use lysine-less (K0) ubiquitin in in vitro assays (24), by its inability to generate ubiquitin chains from N-terminally tagged ubiquitin (33), and by data obtained from mass spectrometric analyses of in vitro–generated polyubiquitin chains (24). However, little is known about the role of the individual components of the complex in this process. Overexpression of HOIP together with HOIL-1, Sharpin, or both is capable of inducing activation of nuclear factor κB (NF-κB) (3335, 50, 51), which correlates with the ubiquitin-generating activity of these different combinations (33). Although both HOIP and HOIL-1 contain RBR domains (Fig. 1), only the RBR domain of HOIP is essential for LUBAC activity (24, 50, 51). A mutant LUBAC complex composed of wild-type HOIP and HOIL-1 with mutations in the RBR domain can generate ubiquitin chains (24) and activate NF-κB in a reporter assay, demonstrating that the RBR domain of HOIL-1 is not directly involved in LUBAC-mediated formation of M1-linked ubiquitin chains (50, 51). Ubiquitin conjugation activity cannot be detected for wild-type HOIL-1 in in vitro ubiquitination assays using wild-type monoubiquitin as a model substrate (24), suggesting that HOIL-1 does not act as an active E3 ligase on its own. This could be due to autoinhibition of HOIL-1, similar to that observed for Parkin, a protein that closely resembles HOIL-1 in domain structure (52). The UBL domain of Parkin interacts with the C-terminal portion, resulting in a closed, inactive conformation. It is therefore conceivable that HOIL-1 acts as an E3 ligase in cells after it is activated by the binding of a so far unknown cofactor. However, both Sharpin and HOIL-1 do not seem to qualitatively differ in their potential to activate the linear ubiquitin–generating capacity of HOIP, despite the absence of an RBR domain in Sharpin. In contrast, the zinc finger and UBL domains of HOIL-1 (51) and Sharpin (34) are important for LUBAC activity. It can therefore be concluded that although HOIP is the “active” part of the complex, it needs to interact with Sharpin, HOIL-1, or both proteins to generate linear ubiquitin chains. Thus, the role of Sharpin and HOIL-1 may be to activate HOIP, possibly by relieving an autoinhibitory effect, and they may also act by affording target specificity. However, to date experimental evidence for either of these mechanisms has not been reported. Further biochemical and structural studies are needed to clarify the role of the different LUBAC components and to explain the specificity of this complex in exclusively generating linear ubiquitin chains.

Data from Klevit and colleagues sheds light on how LUBAC may mechanistically promote the formation of ubiquitin chains (53). The RBR domain–containing proteins Parkin and human homolog of Ariadne (HHARI) both function as HECT-like E3 ligases and accept charged ubiquitin through a thioester intermediate before transferring the bound ubiquitin to a substrate. Whereas this enzymatic reaction requires a conserved cysteine residue in the second RING domain (RING2), the interaction with the E2 enzyme is mediated by the first RING domain (RING1) (53). The ability of HOIL-1 and HOIP to generate ubiquitin chains with UBE2L3 (24), an E2 enzyme that lacks lysine reactivity (53), suggests that LUBAC, like other members of the RBR family, employs a HECT-like mechanism. However, this mechanism does not seem to determine linkage specificity, because Parkin generates K48- and K63-linked ubiquitin chains (54, 55). In addition to HOIP, HOIL-1 and Sharpin also contain ubiquitin-binding domains, which could play a substantial role in controlling linkage specificity by placing the N-terminal methionine (M1) of the acceptor ubiquitin in a position that enables it to attack the thioester bond of the RING2-ubiquitin linkage.

Linear Ubiquitin in Protein Degradation

When linear ubiquitin generated by LUBAC was discovered by Kirisako et al., it was described as functioning in the degradation of a model substrate (24). This role of linear ubiquitin chains was further supported by studies linking it to the degradation of TRIM25 (tripartite motif–containing protein 25) in retinoic acid inducible gene I (RIG-I) signaling, which is involved in sensing viral RNA (56). LUBAC was suggested to inhibit RIG-I–induced type I interferon production by targeting TRIM25 for proteasomal degradation and by competing with TRIM25 for the interaction with RIG-I, hence suppressing the antiviral activity of this pathway. However, it remains to be seen whether LUBAC-generated linear chains are directly recognized by the proteasomal machinery to cause the degradation of TRIM25 or whether LUBAC is involved in the regulation of another E3 ligase that is responsible for the degradation of TRIM25 upon LUBAC overexpression.

The ability of linear polyubiquitin chains to act as a degradation signal was shown by fusing K63- or M1-linked ubiquitin chains to the N or C terminus of proliferating cell nuclear antigen (PCNA), a processivity factor for replicative DNA polymerases (57). This is in line with the finding that linear chains competitively inhibit the degradation of K48-polyubiquitinated substrates by binding to the same receptors in the proteasome (58). Although the proteasomal targeting potential of linear chains is less prominent than that of K48-linked chains, the fusion of a linear noncleavable tetraubiquitin chain to a model protein can elicit the degradation of the fusion partner (59).

It has not been resolved whether linear ubiquitin chains also function as degradation signals in signaling complexes in which ubiquitin-binding proteins and steric effects could shield target proteins modified by M1-linked chains from being recognized by proteasome-targeting factors. Although it can therefore not be ruled out that LUBAC exerts its effects, at least partially, by degradation of specific substrates, most studies to date support a nonproteolytic function for linear ubiquitin in the context of different signaling pathways triggered by several cytokines and Toll-like receptor (TLR) ligands.

Linear Ubiquitin in TNFR1 Signaling

Upon binding of tumor necrosis factor (TNF) to TNF receptor 1 (TNFR1), the receptor trimerizes and recruits TRADD and RIP1 to its intracellular death domain (6063). TRADD then serves as an adapter protein for the recruitment of TRAF2, which in turn associates with cIAP1 and cIAP2 (cIAP1/2) through its cIAP-interaction motif (CIM) (64, 65). Once this core complex is assembled, several ubiquitination events occur that promote full assembly and function of the TNF–receptor signaling complex (RSC). The E3-ligase activity of cIAP1/2 is required for the efficient recruitment of LUBAC (50). Together with the finding that HOIL-1, HOIP, and Sharpin can all bind different ubiquitin linkage types (Table 1), this indicates that cIAP1/2-generated ubiquitin chains provide the binding sites by which LUBAC is recruited to the TNF-RSC. HOIP, which forms a stimulation-independent complex with Sharpin and HOIL-1, has a central role in recruiting LUBAC, as shown by the absence of Sharpin and HOIL-1 in the TNF-RSC after HOIP knockdown by RNA interference (33).

Table 1

Affinities of TNF-RSC components for different types of polyubiquitin chains. A dash indicates no detectable interaction. BIR3, baculovirus IAP repeat; NA, not applicable; ND, not determined; IP, immunoprecipitation; NZF, Npl4 zinc finger; SPR, surface plasmon resonance; UBA, ubiquitin-associated; UBAN, ubiquitin-binding domain; WB, Western blotting; ZF, zinc finger.

View this table:

LUBAC and the other E3 ligases present in the complex modify several targets in the TNF-RSC, including themselves, with different types of ubiquitin chains (50, 6670). It is currently unclear which type of ubiquitin chain is synthesized by cIAP1/2 in cells in the context of the TNF-RSC. However, a model in which cIAP1/2 only attach K63-linked polyubiquitin chains to RIP1, which are in turn responsible for recruitment of the IκB kinase (IKK) and transforming growth factor β–activated kinase (TAK)–TAK1-binding protein (TAB) complexes, has been challenged. First, RIP1 is not essential for TNF-induced NF-κB activation (71). Second, TNF-induced degradation of IκBα and nuclear translocation of p65 are unchanged in murine embryonic fibroblasts (MEFs) deficient for the E2 ligase UBE2N (72), which was believed to function together with TRAF proteins or cIAP1/2 to synthesize K63-linked polyubiquitin chains (7377). Third, cIAPs also generate K11-linked ubiquitin chains on RIP1 (67). Fourth, cells incapable of generating K63-linked chains retain the ability to activate NF-κB after TNF stimulation (78). Finally, we discovered that both RIP1 and NEMO undergo linear ubiquitination in the native TNF-RSC and that RIP1 is simultaneously ubiquitinated by M1-, K11-, K48-, and K63-linked chains in this protein complex (33). The presence of at least four different ubiquitin linkage types in the native TNF-RSC suggests that they all serve discrete and specific functions (79). Like K48-linked chains, ubiquitin linked through K11 is reported to target proteins for proteasomal degradation in the context of cell cycle regulation (30, 32). It is therefore possible that K11 chains contribute to the degradation of RIP1. However, like K63- and M1-linked chains, they may also have a role in signaling as suggested by the binding of NEMO to K11-, K63-, and M1-linked chains with comparable affinities (67).

The specific effects of the different chain types are mediated by ubiquitin-binding domains (UBDs) that bind with different topology and affinity depending on linkage type (80). This leads to differential recruitment of UBD-containing downstream components such as NEMO and TAB2. The ubiquitin-binding ability of NEMO and TAB2 is essential for recruitment of the IKK and TAB-TAK complexes and the activation of NF-κB and mitogen-activated protein kinases (MAPKs) (66, 6870, 75).

In the absence of HOIL-1 and HOIP, recruitment of NEMO and IKKα to the TNF-RSC is inhibited (50), suggesting that binding of NEMO to linear chains is important for efficient recruitment of the IKK complex. In addition, data from cells with decreased abundance of one or two LUBAC components through genetic ablation or by RNA interference show that the absence of LUBAC components results in reduced phosphorylation and degradation of IκBα, impaired and delayed nuclear translocation of the NF-κB subunit p65, diminished overall gene induction, and increased TNF-induced cell death (3335, 50, 51). This suggests that the ability of NEMO to bind linear ubiquitin is important for efficient NF-κB activation. This notion is corroborated by the identification of mutations in the UBAN [ubiquitin binding in ABIN (A-20 binding inhibitor of NF-κB activation) and NEMO] motif of NEMO in patients suffering from X-linked ectodermal dysplasia and immunodeficiency caused by impaired NF-κB signaling (81, 82). The UBAN domain binds preferentially to linear diubiquitin with approximately 100 times greater affinity than to K63-linked diubiquitin (83, 84), providing further evidence that NEMO binding to linear ubiquitin chains plays an important role for functional NF-κB activation in vivo.

The mechanism by which linear ubiquitin binding influences the activation of NF-κB has not been entirely clarified. However, binding of NEMO to M1-linked ubiquitin, which differs from binding to K63-linked chains in the binding mode because of differences in the ubiquitin chain linker region (83, 84), may induce a conformational change in the protein (84) that translates into an overall structural change in the IKK complex and results in the activation of its kinase subunits. Alternatively, binding of NEMO to linear chains on RIP1 or NEMO may lead to clustering of IKK complexes, thereby enabling transautophosphorylation and activation.

However, recruitment of complex components is not the only mechanism by which linear ubiquitin chains influence signaling. The ubiquitination of NEMO is required for efficient signal induction: LUBAC can attach linear ubiquitin chains to NEMO in vitro (3335, 51), and TNF-induced linear ubiquitination of NEMO occurs preferentially on K285 and K309 (51). In cells expressing a NEMO K285R/K309R mutant that cannot be modified with linear ubiquitin, both LUBAC- and interleukin-1β (IL-1β)–induced activation of NF-κB were reduced as measured by luciferase-based NF-κB reporter assays (51). The induction of conformational changes within NEMO and the IKK complex resulting in its activation could again be a possible mechanism for the influence on NF-κB activation.

Differences in IKK activation in the presence and absence of linear ubiquitin may be mediated not only by direct effects on NEMO but also by an overall stabilization effect of LUBAC on the TNF-RSC (50), an effect that requires the activity of LUBAC, implying that the linear ubiquitin chains generated by LUBAC, and not merely the presence of LUBAC in the complex, cause this effect (50). By maintaining RIP1, TRAF2, cIAP1, and TAK1 in the complex for a longer period of time, the presence of LUBAC ensures that the signal-initiating complex stays intact and keeps the IKK complex in close proximity to its activating kinase TAK1 for an extended time, thereby allowing for more potent IKK activation to take place (Fig. 2).

Fig. 2

Model of LUBAC-dependent TNF-induced signaling cascades. LUBAC recruitment to the TNF-RSC in a TRADD-, TRAF2-, and cIAP-dependent manner (left panel) ensures full activation of the NF-κB and MAPK pathways and the inhibition of cell death induction. The absence of LUBAC (right panel) destabilizes the TNF-RSC and, consequently, pro-survival pathways are initiated less efficiently, and the balance of signaling pathways emanating from the TNF-RSC is shifted toward cell death.

CREDIT: Y. HAMMOND/SCIENCE SIGNALING

However, in the absence of LUBAC components, the TNF-RSC still forms, and the role of M1-linked ubiquitin chains in the activation of the IKK complex could potentially be replaced, at least partially, by K63- or K11-linked chains, which can also bind or be attached to NEMO. This is in line with the finding that when the abundance of different LUBAC components is decreased, TNF-induced activation of NF-κB and MAPKs can still occur, albeit to a lesser extent. Although these data indicate that LUBAC is not essential for the activation of these pathways, they demonstrate that LUBAC is required for full activation of these signaling outputs. However, additional experiments in HOIP-deficient cells, which are devoid of LUBAC activity, are needed to corroborate this idea.

The stabilization of the TNF-RSC mentioned above is also a potential mechanism by which linear ubiquitin chains might influence TNF-induced activation of NF-κB and MAPKs. However, the importance of LUBAC and LUBAC-generated M1-linked ubiquitin chains for the TNF-induced activation of JNK and ERK pathways is still controversial and not entirely accepted in the field. Whereas some studies reported that the activation of JNK and ERK is unchanged in the absence of either HOIL-1 or Sharpin (35, 51), other reports showed that JNK activation is impaired in Sharpin-deficient MEFs obtained from chronic proliferative dermatitis (cpdm) mice as well as in HOIL-1 knockdown cells (33, 34, 50). However, based on findings demonstrating that LUBAC is involved in MAPK activation in response to stimulation with CD40L, IL-1β, or lipopolysaccharide, the emerging picture is that LUBAC not only regulates activation of the IKK complex but also enables full activation of MAPKs in different settings (3335).

Apart from allowing for full activation of gene induction, LUBAC also confers resistance to TNF-induced cell death. Decreased abundance or the absence of one of the LUBAC components sensitizes cells, including primary keratinocytes in vitro, to TNF-induced cell death (3335, 50, 51). Therefore, the inflammatory phenotype of cpdm mice caused by the absence of Sharpin may be explained by proinflammatory effects resulting from keratinocyte cell death. The rescue of this phenotype by ablation of TNF (33) suggests that TNF-induced cell death in Sharpin-deficient keratinocytes causes the inflammatory phenotype observed in cpdm mice. Intriguingly, TNF-induced cell death in cpdm-derived MEFs and primary keratinocytes is only partly apoptotic, and a major proportion of this cell death is due to necroptosis (33), a form of cell death that is thought to be proinflammatory (85, 86). The roles of LUBAC in pro-survival signaling and cell death induction may be linked if LUBAC ensures sufficient expression of genes encoding pro-survival proteins, thereby protecting cells from TNF-induced cell death. However, it is also conceivable that LUBAC prevents the induction of cell death on a different level. With respect to TNF-induced cell death, one level of regulation may be exerted on the formation of the death-inducing cytoplasmic complex II. It is possible that in the absence of linear ubiquitination, deubiquitination of RIP1, a prerequisite for complex II formation (86), may occur more quickly or more efficiently. The lesser effect of the absence of HOIL-1 as compared to that of Sharpin on sensitization to TNF-induced cell death (34) suggests that the two proteins play differential roles in LUBAC-mediated and possibly LUBAC-independent mechanisms involved in cell death regulation. Therefore, investigating the mechanisms of how LUBAC and its individual components regulate the balance between pro-survival signaling and different modes of cell death induction by TNF and other cytokines represents a promising future field of investigation, which is likely to provide insight into the interplay between cell death and inflammation and their role in autoimmunity and cancer.

Linear Ubiquitin in CD40, IL-1β, and TLR Signaling

CD40 belongs to the same family of receptors as TNF-R1, the TNF receptor superfamily. Despite their different physiological functions, both ligand-receptor systems also share some similarities. Accordingly, common components such as TRAF2, cIAP1/2, the IKK and TAB-TAK complexes, and LUBAC are recruited to the two receptor signaling complexes (33, 41, 64, 68, 8791). As with the TNF-R1 system, cIAP1/2 are also required for LUBAC recruitment to the CD40-RSC (33). Hence, it is likely that cIAP-generated ubiquitin chains also provide the recruitment platform for LUBAC to the CD40-RSC. Although no mass spectrometric verification of the presence of linear ubiquitin chains in the CD40-RSC has been presented to date, NEMO is also a probable LUBAC target in this signaling complex. In line with this notion, CD40L-induced activation of NF-κB, JNK, and ERK is inhibited in the absence of Sharpin in primary B cells obtained from cpdm mice (3335, 41). Furthermore, HOIL-1–deficient splenic B cells show reduced activation of the NF-κB and MAPK pathways in response to CD40L stimulation (35), and once recruited to the CD40-RSC, the ubiquitin ligase activity of HOIP is required for full CD40-mediated activation of NF-κB (41). Further analysis will be needed to reveal whether the effects of LUBAC in CD40 signaling are solely mediated by linear ubiquitination of NEMO or whether other proteins within the CD40-RSC are also modified with linear chains.

Although the receptors for IL-1β and related cytokines and TLRs do not belong to the TNF receptor superfamily and rely on other proteins to mediate signaling, the absence of LUBAC also impairs signaling downstream of these receptors. RNA interference or the use of Sharpin-deficient MEFs, primary keratinocytes, or bone marrow–derived macrophages indicate that IL-1β– and TLR4-induced NF-κB and JNK activation is at least partially decreased in the absence of a complete LUBAC (3335). In addition, bone marrow–derived macrophages from homozygous cpdm mice show decreased proinflammatory IL-12 production in response to TLR2, TLR4, TLR7, or TLR9 activation (92).

Again, cIAP1/2, which are involved in MyD88-dependent TLR2 and TLR4 signaling (93), might be responsible for the recruitment of LUBAC to the signal-initiating complexes. Alternatively, it is conceivable that ubiquitin chains generated by TRAF6 might serve as a recruitment platform to recruit LUBAC to either intracellular or membrane-bound signaling complexes in the IL-1 and TLR system. After IL-1β stimulation, NEMO is modified with linear ubiquitin chains on residues K285 and K309 (51). However, at this moment it is unclear which proteins are linearly ubiquitinated in the context of TLR signaling. Although NEMO is a likely candidate, additional studies are required to determine the exact role of linear ubiquitination in IL-1β and TLR signaling.

Open Questions and Future Studies

Receptors of the TNF receptor superfamily and the IL-1 and TLR families have a common requirement for LUBAC in the activation of downstream signaling, although other components mediating signal transduction differ substantially. This gives rise to the question of whether linear ubiquitination is a general concept in the activation of the canonical NF-κB pathways and possibly in the initiation of MAPK signaling cascades. Indeed, the requirement for NEMO and its binding to and modification by ubiquitin (70, 94) in the canonical activation of NF-κB, together with the findings that NEMO binds linear ubiquitin chains with high affinity (Table 1) and that it is modified with linear chains by LUBAC, suggest that linear ubiquitination may be of general importance in this pathway, at least for its full physiological functionality. Currently, the only exception in this regard seems to be hypoxia-induced NF-κB activation, which was reported to be independent of LUBAC (95). Therefore, further experiments investigating different systems activating NF-κB will have to prove or disprove this hypothesis.

Another question is the mechanisms by which LUBAC mediates its effects on NF-κB and MAPK activation. Although NEMO may be both the target of linear ubiquitination and the receptor for linear ubiquitin chains, the identification of linearly ubiquitinated RIP1 in the native TNF-RSC shows that NEMO is not the only physiological substrate of LUBAC. It therefore seems likely that additional targets of linear ubiquitination are present in the TNF-RSC and also in other signaling complexes that result in the activation of NF-κB and MAPKs. In addition, although other linear ubiquitin-binding proteins have been identified (Table 1), the relevance of this binding in the context of signaling pathways in which these factors are implicated has not been sufficiently investigated.

Because most reports that study the binding specificity of signaling proteins to different ubiquitin chains are so far limited to the analysis of K48- and K63-linked chains (Table 1), future interaction studies should also include the analysis of unconventional linked ubiquitin polymers once the appropriate tools and reagents are available (96). In addition, the results of binding studies may vary depending on the method used and on whether full-length proteins or isolated UBDs are investigated (Table 1). It is therefore important to establish certain standards and criteria to standardize the methodology to be able to compare results from different studies. We also identified M1- and K11-linked ubiquitin chains in the native TNF-RSC (33), and it will be interesting to determine whether proteins so far thought to exclusively bind to K63- or K48-linked chains can also bind to other atypical ubiquitin chains. In this context, the UBAN domain of NEMO and the BIR3-RING domain of cIAP1, for example, also bind to K11 ubiquitin chains (67). Analyses determining the specificity of ubiquitin-binding proteins for certain linkage types may help to dissect the kinetics and chronology of recruitment events and therefore deepen our understanding of the formation and deconstruction of receptor signaling complexes and the balance of the diverse signaling pathways emanating from these complexes. It will also be necessary to determine the relative abundance of the different polyubiquitin chain topologies as well as the precise length of particular ubiquitin modifications, because the abundance and length of a chain type may regulate its effect on signaling.

Most ubiquitin linkage types can be generated by several combinations of E2 enzymes and E3 ligases, thus complicating the investigation of the relevance of a ubiquitin linkage type to a signaling pathway. For example, members of the UBE2D E2 enzyme family, which also generate K63-linked polyubiquitin chains (97, 98), can most likely substitute for UBE2N in certain signaling pathways or cell types. It will therefore be crucial to resolve which E2 enzymes are involved in the regulation of specific ubiquitin-dependent signaling events under physiological conditions in vivo. The ubiquitin replacement strategy described by Xu et al. will probably be helpful in defining the effects of specific linkage types on signaling, because it enables the replacement of endogenous ubiquitin with different ubiquitin lysine mutants (78). Because LUBAC appears to be the only E3 ligase involved in generating this type of ubiquitin linkage, the investigation of M1-linked ubiquitin chains may rely on more conventional genetic ablation techniques (24).

Conclusions and Outlook

We are only beginning to understand the complex network and functional interplay between different kinds of ubiquitin chains. The identification of linear ubiquitination as a physiologically important regulator of inflammation and a component of various immune signaling pathways (3335, 50, 51) has established this type of ubiquitin chain linkage as an important and crucial part of the ubiquitin network. The generation of different types of ubiquitin chain linkages encodes the signals that enable the specific and temporally controlled recruitment and precise positioning of individual components within signaling complexes. One approach in deciphering the underlying ubiquitin code will be to investigate how the different linkage types differ structurally and functionally, especially from chain types present in the same context. Because K63- and M1-linked chains are similar in their overall structure (20), it is likely that their distinct effects on signaling are mediated either by specific ubiquitin receptors (Table 1) or by differential sensitivity to cleavage by specific DUBs. In this context, it is interesting to mention that except for CYLD, DUBs that play a role in the regulation of NF-κB and MAPK activation do not cleave linear ubiquitin chains or process them less efficiently than other chain types (20). CYLD itself cleaves linear ubiquitin chains attached to a substrate, such as NEMO, less efficiently than unanchored chains (99); human isopeptidase T, the DUB responsible for disassembly of the majority of unanchored polyubiquitin chains in cells, cannot cleave polyubiquitin chains anchored to a substrate because it needs to bind to the C-terminal glycine residue of ubiquitin (100).

Therefore, the increased stability of the TNF-RSC in the presence of LUBAC might also be due to the refractory nature of linear chains to DUB-mediated disassembly. Based on this finding, it is tempting to speculate that linear ubiquitin chains stabilize complex I of TNF signaling, and prevent formation of the death-inducing complex II, because they are more resistant to the DUBs recruited to this complex. Again, further studies are required to complete our current picture of how linear ubiquitin chains influence signaling and how they differ from other ubiquitin linkage types.

LUBAC and its linear ubiquitin chain–generating activity are required for the full activation of signaling pathways emanating from signaling complexes induced by diverse ligand-receptor systems. Hence, linear ubiquitin chains, and most likely other atypical chains, will need to be considered from now on, in addition to K63- and K48-linked chains, in the study of cellular signal transduction and the resulting physiological and pathological consequences.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
View Abstract

Stay Connected to Science Signaling

Navigate This Article