Deubiquitylation and regulation of the immune response

CYLD

CYLD was originally identified as a tumour suppressor, the mutation of which predisposes patients for the development of tumours of the hair follicles (cylindromas)¹⁰. The mutations often occur in the carboxy-terminal portion of CYLD, which contains a DUB domain¹⁰, and the DUB function of CYLD was subsequently revealed by functional proteomics¹¹ and linked to signal transduction involved in the activation of NF-κB^12–14. So, it is probable that mutations of CYLD disrupt its DUB catalytic activity and cause dysregulated activation of NF-κB, which in turn promotes cell survival and tumorigenesis. A study using Cyld-knockout mice further indicates that CYLD deficiency might promote the transactivation function of NF-κB by increasing the ubiquitylation and nuclear translocation of an NF-κB co-activator protein, BCL-3 (B-cell lymphoma 3)¹⁵.

It is now clear that CYLD is a member of the USP family of DUBs that targets multiple signalling molecules, such as members of the TRAF (tumour-necrosis factor (TNF) receptor (TNFR)-associated factor) family, the IKK (inhibitor of NF-κB (IκB) kinase) regulatory subunit IKKγ (also known as NEMO)^{12–14,16,17}, the SRC protein tyrosine kinase LCK¹⁸, the NF-κB co-activator protein BCL-3 (Ref. 15), transforming growth factor-β (TGFβ) activated kinase 1 (TAK1)¹⁹, receptor-interacting protein 1 (RIP1)²⁰ and transient receptor potential channel A1 (TRPA1)²¹. Not surprisingly, CYLD regulates diverse biological functions, including host defence against infections^22,23, immune-cell development, activation and inflammation^18,19,24,25, cell survival^12–14, cell proliferation and tumorigenesis^15,24, microtubule assembly and cell migration²⁶, mitotic cell entry²⁷, calcium-channel function²¹, spermatogenesis²⁰ and osteoclastogenesis¹⁷.

Consistent with its substrate diversity, CYLD contains several protein-interaction domains, including three cytoskeleton-associated protein glycine-rich (Cap-Gly) domains, two proline-rich motifs and a zinc-finger-like domain¹⁰. The Cap-Gly domains typically mediate association with microtubule proteins, whereas the proline-rich motifs interact with proteins through SH3 domains. The first Cap-Gly domain of CYLD mediates its association with microtubules and the regulation of microtubule assembly²⁶. Interestingly, the tertiary structure of the third Cap-Gly domain of CYLD resembles that of an SH3 domain that is known to bind proline-rich sequences²⁸. The possession of both proline-rich motifs and an SH3-like domain could allow CYLD to interact with target proteins containing either SH3 domains or proline-rich motifs. Indeed, the third Cap-Gly domain of CYLD interacts with a proline-rich region in one of its targets, IKKγ (Ref. 28). In addition to these well-defined protein domains, new domains of CYLD have been defined, based on mutational analyses, for the interaction of CYLD with TRAF2, IKKγ and BCL-3 (Refs 13,15).

A prominent feature of the target proteins of CYLD is their conjugation with K63-linked ubiquitin chains, which seem to facilitate protein–protein interactions in the recruitment and activation of downstream signalling molecules by the CYLD target¹. In particular, the K63-linked ubiquitin chains of RIP1 form a platform that recruits the IKK complex and its upstream kinase, TAK1, leading to efficient activation of these kinases^29–31. Strong biochemical evidence indicates that CYLD is specific for K63-linked ubiquitin chains^13,15. Consistent with this, the USP catalytic domain of CYLD has a unique structure that differs from that of other USPs and seems to have specific determinants for mediating the deconjugation of K63-linked ubiquitin chains³². However, it is unclear whether CYLD is exclusively a K63-specific DUB, as CYLD has also been shown to have activity towards K48-linked ubiquitin chains in certain in vitro¹⁴ and in vivo^18,33 conditions. Results from two studies indicate that CYLD can also deconjugate K48-linked ubiquitin chains to prevent the proteasomal degradation of some target proteins^21,33. So, it is possible that the ubiquitin-chain specificity of CYLD is dependent on the target proteins. Moreover, recent evidence indicates that the function of CYLD might also involve adaptor proteins^17,34. CYLD binding to one of its targets, TRAF6, requires the adaptor protein p62, which promotes the deubiquitylation of TRAF6 by CYLD¹⁷ and probably also modulates the DUB activity of CYLD through induction of CYLD ubiquitylation³⁴.

A20

A20 (also known as TNFAIP3) is a zinc-finger-domain-containing protein that is required for terminating the NF-κB activation signal mediated by innate immune receptors, such as TNFR, Toll-like receptors (TLRs) and an intracellular pattern-recognition molecule, NOD2 (nucleotide-binding oligomerization domain protein 2)^35–38. The amino-terminal region of A20 contains an OTU domain, which has DUB activity towards several NF-κB signalling factors, including TRAF6, RIP1, RIP2 and IKKγ (Refs 8,37–40). A unique feature of A20 is its ability to function as both a DUB and an E3 ligase⁸. A20 catalyses the K48-linked ubiquitylation of RIP1 through its carboxy-terminal zinc-finger domain, an action that targets RIP1 for proteasomal degradation. Concurrently, A20 removes K63-linked ubiquitin chains from RIP1, which not only inactivates the signalling function of RIP1 but might also facilitate its K48-linked ubiquitylation and degradation⁸. It is important to note, however, that the A20-mediated K48-linked ubiquitylation and degradation of RIP1 might involve a more complex mechanism, as this function of A20 also requires ITCH (itchy homologue E3 ubiquitin protein ligase), a K48-specific E3 ligase that stably associates with A20 (Ref. 41). It is unclear whether both A20 and ITCH ubiquitylate RIP1 or whether ITCH regulates the E3 ligase activity of A20 through ubiquitylation of a regulatory protein.

How the DUB substrate specificity of A20 is regulated is also not completely understood. Although A20 has preferential DUB activity for removing K63-linked ubiquitin chains in vivo⁸, recombinant A20 or the A20 OTU domain cleaves both K63- and K48-linked ubiquitin chains in vitro with a preference for K48-linked ubiquitin chains^42,43. It seems that the substrate specificity of A20 is modulated by the conjugation of ubiquitin chains to specific target proteins. A recent study shows that A20 efficiently removes K63-linked ubiquitin chains when they are conjugated to TRAF6 (Ref. 43). Interestingly, however, A20 does not disassemble the K63-linked ubiquitin chains but instead removes the intact ubiquitin chains by cleaving the linkage between TRAF6 and the ubiquitin chains⁴³. It is therefore probable that the specificity of A20 for K63-linked ubiquitylated targets is not determined by its intrinsic DUB catalytic activity but possibly through its specific interaction with the ubiquitylated proteins. Accumulated evidence indicates that this function of A20 is in turn mediated through adaptor proteins. One such protein is ABIN1 (A20-binding inhibitor of NF-κB 1; also known as TNIP1), which physically associates with A20 and functions as an adaptor to recruit A20 to its target IKKγ (Ref. 40). Small interfering RNA (siRNA)-mediated knockdown of ABIN1 expression prevents A20 from deubiquitylating IKKγ (Ref. 40). Another protein that regulates the DUB function of A20 is TAX1-binding protein 1 (TAX1BP1)^44,45, a zinc-finger-domain-containing protein that was originally found to associate with the Tax protein encoded by human T-cell leukaemia virus type 1 (Ref. 46). TAX1BP1 also interacts with A20 (Ref. 47) and mediates the recruitment of A20 to RIP1 and TRAF6. Notably, both ABIN1 and TAX1BP1 contain a ubiquitin-binding domain, which seems to mediate their association with ubiquitylated TRAF6 and RIP1 (Refs 45,48). So, A20 interacts with its targets through ubiquitin-binding adaptor proteins.

Cezanne

Cezanne (cellular zinc finger anti-NF-κB; also known as OTUD7B) was identified on the basis of its sequence homology to A20 (REF. 49). Similar to A20, Cezanne has a DUB catalytic domain that belongs to the OTU family, and it negatively regulates activation of NF-κB in the TNF signalling pathway^49,50. In response to TNFR signalling, the expression of Cezanne is induced and it is recruited to the TNFR, where it seems to deubiquitylate RIP1 and thereby negatively regulate the activation of IKK^50,51. Cezanne might function, similar to A20, in the feedback control of NF-κB activation, although its functional relationship with A20 is currently unclear.

DUBA

Deubiquitylating enzyme A (DUBA; also known as OTUD5), an OTU-family member, was recently identified through siRNA screening as a negative regulator of type I interferon (IFN) induction⁷. Similar to A20, DUBA contains an OTU domain and selectively cleaves K63-linked ubiquitin chains in transfected cells. However, unlike A20 and CYLD, DUBA is not required for the negative regulation of NF-κB. Instead, DUBA seems to selectively regulate the activation of IFN-regulatory factor 3 (IRF3) and IRF7, which are transcription factors that regulate IFN expression.

Otubain-1

Another OTU-family member, otubain-1 (also known as OTUB1), was identified by affinity purification using biotin-conjugated ubiquitin aldehyde (a competitive DUB inhibitor)⁵². The DUB activity of otubain-1 was shown by its ability to cleave a tetraubiquitin substrate in vitro. In a separate study using a yeast two-hybrid screen, otubain-1 was isolated as a protein that associates with the E3 ligase GRAIL (gene related to anergy in lymphocytes)⁵³. An alternatively spliced isoform of otubain-1, ARF1 (alternative reading frame 1), shares the C-termimal portion with otubain-1 but lacks an intact OTU domain⁵³. Otubain-1 and ARF1 both bind GRAIL, but they have opposite effects in regulating the self-ubiquitylation and degradation of GRAIL. Surprisingly, although otubain-1 is a DUB, it promotes rather than inhibits the K48-linked self-ubiquitylation and proteolysis of GRAIL⁵³. This unexpected function of otubain-1 might be mediated through the inhibition of USP8, a DUB that binds to and deubiquitylates GRAIL; however, it is not known how otubain-1 might inhibit USP8 (Ref. 53). Nevertheless, it seems that the DUB activity of otubain-1 is important, because ARF1, which lacks the OTU domain and therefore the DUB activity of otubain-1, opposes the function of otubain-1 (Ref. 53). It has been proposed that the ratio between otubain-1 and ARF1 might have an important role in regulating the fate and function of GRAIL⁵³.

USP15

A recent study indicates that the USP-family member, USP15, might have a role in the regulation of NF-κB activation⁵⁴. In contrast to CYLD and A20, which target upstream signalling molecules in the NF-κB pathway, USP15 negatively regulates the K48-linked ubiquitylation of IκBα, a key downstream signalling step that triggers IκBα proteolysis and NF-κB nuclear translocation⁵⁵. This function of USP15 was discovered on the basis of its physical association with the COP9 signalosome (CSN), a multi-protein complex that has diverse functions, including the regulation of protein ubiquitylation and degradation⁵⁶. siRNA-mediated knockdown of a CSN subunit, which destabilizes the CSN complex, prolongs the TNF-stimulated ubiquitylation and degradation of IκBα, leading to sustained nuclear translocation of NF-κB⁵⁴. A similar, although much milder, phenotype was obtained by knockdown of USP15. It is unclear whether CSN associates with additional DUBs or has other mechanisms that interfere with IκBα ubiquitylation.

Cytokine-induced DUBs

The expression of a subfamily of DUBs is induced in lymphocytes by cytokines, such as interleukin-2 (IL-2), IL-3, IL-4, IL-5 and GM-CSF (granulocyte/macrophage colony-stimulating factor)⁵⁷. These cytokine-induced DUBs, which include mouse DUB1 (also known as USP36), DUB1A, DUB2 and DUB2A, and human DUB3 and USP17, might have a role in regulating the growth and survival of lymphocytes⁵⁷. For example, overexpression of DUB2 negatively regulates K48-linked ubiquitylation of the common cytokine-receptor γ-chain (γ_c) by counteracting the ubiquitin-ligase function of CBL⁵⁸. Consequently, DUB2 stabilizes γ_c, prolongs IL-2-induced activation of JAK–STAT (Janus kinase–signal transducer and activator of transcription) signalling and promotes the survival of lymphocytes^58,59. However, the physiological role of cytokine-induced DUBs in immune regulation has not yet been confirmed by in vivo models.

A20 as a global negative regulator of inflammation

The first DUB shown to have a role in innate immune regulation was A20. Genetic deficiency of A20 in mice causes persistent activation of NF-κB by TLRs and TNFR, leading to multiorgan inflammation, cachexia and neonatal lethality^36,37. These inflammatory disorders are independent of the adaptive immune system, as they are also detected in mice that lack both A20 and RAG1 (recombination-activating gene 1; which is crucial for both B- and T-cell development)³⁶. It seems that the loss of A20 breaks the tolerance of the innate immune system to the commensal intestinal microflora, leading to aberrant homeostatic TLR signalling and the production of pro-inflammatory mediators⁶³. Indeed, crossing A20-knockout mice with mice that lack the common TLR adaptor MyD88 (myeloid differentiation primary-response gene 88) markedly decreases the severity of inflammation and prevents premature lethality and cachexia of the A20-knockout mice⁶³.

The negative signalling function of A20 involves the deconjugation of K63-linked ubiquitin chains from TRAF6 and RIP1, which are central players of the TLR and TNFR signalling pathways¹. Recent evidence indicates that A20 also mediates the deubiquitylation of RIP2 and thereby negatively regulates the activation of NF-κB and the induction of pro-inflammatory cytokines by the intracellular PRR, NOD2 (Ref. 38). In addition, the E3 ligase activity of A20 also contributes to its negative signalling function. A20 catalyses the K48-linked ubiquitylation of RIP1, thereby targeting it for degradation. In the absence of A20, the signal-induced K63-linked ubiquitylated forms of TRAF6 and RIPs can accumulate, leading to prolonged activation of NF-κB and the aberrant expression of pro-inflammatory cytokines. So, A20 is a central player in the control of inflammatory responses (FIG. 1a).

CYLD in host defence against infections

Initial in vitro studies indicated that CYLD has a negative role in the regulation of TLR and TNFR signalling^12–14, but how CYLD regulates innate immune responses in vivo is still unclear. Unlike the A20-knockout mice, Cyld-knockout mice do not have severe abnormalities in innate immune-receptor signalling or multiorgan inflammation^15,18,23,24. Loss of CYLD in macrophages has moderate or undetectable effects on NF-κB activation by TLR ligands and TNF^18,19,24. As A20 is expressed in both a basal and inducible manner by macrophages³⁷, it is possible that CYLD is functionally redundant with A20. The fact that CYLD cannot compensate for A20 in terminating TLR and TNFR signals can probably be attributed to the phosphorylation-mediated inactivation of CYLD in stimulated cells⁶⁴.

Recent evidence indicates that CYLD might have a non-redundant role in the regulation of specific aspects of innate immunity. Using an animal model of pneumococcal pneumonia, it has been shown that loss of CYLD protects mice from acute lung injury and lethality induced by Streptococcus pneumoniae²³. CYLD seems to negatively regulate S. pneumoniae-induced expression of plasminogen activator inhibitor 1, a factor that protects the host during severe pneumococcal pneumonia by preventing tissue damage and decreasing bacterial translocation to the blood circulation. This function of CYLD is mediated through the negative regulation of the mitogen-activated protein kinase p38 (Ref. 23). In a similar study, CYLD was shown to negatively regulate NF-κB activation and lung inflammation in mice infected with non-typeable Haemophilus influenzae²².

The role of CYLD in regulating the host response to infections is also indicated by the finding that Cyld-knockout mice are more susceptible to the induction of colonic inflammation by dextran sulphate sodium (DSS)²⁴. DSS is known to cause colonic epithelial damage, leading to productive invasion by commensal bacteria and the activation of innate immune cells in the mucosa. The higher sensitivity of Cyld^−/− mice to DSS-induced inflammation indicates a role for CYLD in the negative regulation of the mucosal innate immune response to microorganisms.

Regulation of antiviral innate immunity by DUBA

The innate immune response to viral infections involves activation of TBK1 and IKKε, and the downstream transcription factors IRF3 and IRF7 (Ref. 65). A key intermediate molecule, TRAF3, connects TBK1 and IKKε to upstream signalling molecules^66,67. Interestingly, DUBA physically interacts with TRAF3 and inhibits the self-ubiquitylation of TRAF3 (Ref. 7) (FIG. 1b). As DUBA also interferes with the binding of TRAF3 to TBK1 (Ref. 7), it is conceivable that the K63-linked ubiquitin chains of TRAF3 might facilitate its binding to the TBK1–IKKε complex. Formation of this intermediate signalling complex also requires the adaptor protein TANK (TRAF-family-member-associated NF-κB activator) ⁶⁸. Analogous to IKKγ, TANK undergoes K63-linked ubiquitylation⁶⁹, but how TANK ubiquitylation regulates TBK1–IKKε activation is unclear. A recent study indicates that IKKγ targets TBK1–IKKε through TANK, thereby regulating the induction of type I IFNs⁷⁰. It remains to be examined whether the ubiquitylation of IKKγ and TANK is required for this newly characterized signalling axis.

Knockdown of DUBA by siRNA has little effect on NF-κB activation, which indicates that DUBA is either not involved or functionally redundant with other DUBs in the regulation of NF-κB signalling. The latter possibility is indicated by the finding that overexpression of DUBA potently inhibits the induction of an NF-κB-driven reporter gene stimulated by constitutively active forms of RIG-I and MDA5 (Ref. 7). As the ubiquitylation of RIG-I is crucial for its signalling function⁷¹, it will be important to examine whether DUBA regulates the ubiquitylation of this RNA sensor (FIG. 1c). However, a recent study indicates that CYLD is a DUB that negatively regulates the ubiquitylation of RIG-I and prevents aberrant activation of TBK1–IKKε (Ref. 72). Finally, the in vivo role of DUBA in the regulation of innate immunity awaits studies using animal models.

In vitro studies indicate that A20 also has a role in the negative regulation of antiviral immune responses mediated by TLR3 and RIG-I, in which A20 might target the adaptor protein TRIF (TIR-domain-containing adaptor protein inducing IFNβ) or downstream kinases, TBK1 and IKKε (Refs 73–75). However, a recent study using A20-knockout mice indicates that although A20 negatively regulates TRIF-mediated NF-κB signalling, this DUB is not important for the regulation of TBK1–IKKε activation, as indicated by the normal phosphorylation of the downstream target, IRF3, in A20-deficient macrophages⁶³.

T-cell development

T-cell development in the thymus involves ordered progression through distinct stages of double negative (DN), double positive (DP) and single positive (SP) cells, which are defined based on expression of the T-cell co-receptors CD4 and CD8 (Ref. 76). Signal transduction through the pre-T-cell receptor (pre-TCR) drives the proliferation of DN thymocytes and their progression to the DP stage, whereas progression from the DP to SP stage is tightly regulated by the strength of signals from the rearranged TCR. A TCR signal that is too strong, owing to the recognition of self antigens, leads to negative selection, whereas a signal that is too weak leads to death by neglect. Only those DP thymocytes that receive an adequate TCR signal strength are positively selected and further develop to SP T cells. So, many signalling factors that modulate TCR signalling have a role in T-cell development. However, our knowledge of how ubiquitylation and deubiquitylation regulate T-cell development is scant.

CYLD was the first DUB found to regulate thymocyte development¹⁸. Cyld-knockout mice are competent in producing early-stage thymocytes but have a decreased number of SP thymocytes and peripheral T cells. In contrast to its negative role in the regulation of NF-κB, CYLD positively regulates LCK¹⁸, a SRC-family protein tyrosine kinase that is required for TCR-proximal signalling and thymocyte development⁷⁷. CYLD is not required for LCK activation but instead controls the association of activated LCK with its target kinase ZAP70. Therefore, the CYLD-deficient thymocytes have a defect in TCR-stimulated phosphorylation of ZAP70 and downstream signalling events. Notably, CYLD selectively interacts with the active form of LCK. As activation of LCK involves its structural change from a closed to an open conformation⁷⁸, it is possible that the CYLD-binding domain in LCK is masked in its inactive, closed state.

The ubiquitylation of LCK by CBL (Casitas B-lineage lymphoma) is known to be a negative mechanism that regulates the signalling function of LCK⁷⁹. However, it is controversial whether this negative regulation of LCK by CBL is through the increased degradation of LCK. One study indicated that loss of CBL expression increases the constitutive level of LCK⁸⁰, but other studies have not shown this^80–82. What seems to be consistent is that CBL negatively regulates phosphorylation of the LCK target, ZAP70 (Refs 81,82). As the opposite phenotype is detected in CYLD-deficient thymocytes, CYLD might oppose some of the E3 ligase functions of CBL¹⁸. At least in transfected cells, CYLD inhibits CBL-induced ubiquitylation of LCK¹⁸. Notably, CYLD removes both K48- and K63-linked ubiquitin chains from LCK¹⁸, which is in contrast to the K63-specific activity of CYLD towards TRAFs and BCL-3 (Refs 13,15). However, as loss of CYLD does not alter the steady-state level of LCK in thymocytes, it is unclear whether CYLD regulates the K48-linked ubiquitylation of LCK in vivo (which targets LCK to the proteasome). Based on studies with both CYLD- and CBL-deficient mice, it seems possible that ubiquitylation of LCK might have a non-degradative effect. One such effect might be to prevent the translocation of LCK to lipid rafts, as LCK is excluded from lipid rafts when it is co-expressed with CBL⁸³. Because lipid rafts provide a microenvironment for concentrating TCR signalling molecules, the translocation of LCK to lipid rafts probably promotes its interaction with ZAP70, which would explain why the LCK–ZAP70 interaction is attenuated in the absence of CYLD¹⁸ and promoted in the absence of CBL⁸².

The finding that CYLD is not required for LCK activation but regulates LCK-mediated ZAP70 activation is consistent with the roles of LCK and ZAP70 in thymocyte development. Whereas LCK is involved in the late and early stages of thymocyte development, ZAP70 specifically regulates the production of SP cells⁷⁶. Therefore, it is likely that aberrant ubiquitylation of LCK, as a result of loss of CYLD, leads to decreased activation of ZAP70 and attenuated positive selection of SP thymocytes.

T-cell tolerance

The induction of T-cell tolerance occurs at two levels — central tolerance and peripheral tolerance^76,84,85. Central tolerance occurs in the thymus and involves negative selection of immature T cells to eliminate those that are self-reactive, as well as the generation of regulatory T (T_Reg) cells. Peripheral tolerance is mediated by several strategies that render mature T cells non-responsive to self antigens in the periphery. Ubiquitylation is a crucial mechanism that regulates both central and peripheral T-cell tolerance^3,9. Several E3 ligases that are involved in T-cell tolerance have been characterized; these include CBL-B, TRAF6, GRAIL, ROQUIN (also known as RC3H1) and ITCH^3,9. By contrast, relatively little is known about the DUBs that counteract the function of these E3 ligases.

Otubain-1 was the first DUB to be implicated in the regulation of T-cell anergy⁵³, an important type of peripheral tolerance that can be induced by TCR ligation in the absence of co-stimulatory signals⁸⁶. As noted already, otubain-1 and its alternatively spliced isoform ARF1 function conversely to control the ubiquitylation and stability of GRAIL⁵³, one of the main E3 ligases to be induced in T cells under anergic conditions⁹. The E3 ligases are thought to negatively regulate the function or expression of TCR-proximal signalling molecules, such as phospholipase Cγ1 (PLCγ1) and protein kinase C-θ (PKC-θ), thereby attenuating IL-2 production and T-cell proliferation⁹. GRAIL is also expressed by T_Reg cells and seems to have a role in the production of inducible regulatory T-cell populations in the periphery⁸⁷. Consistent with its ability to induce GRAIL degradation through K48-linked polyubiquitylation (see earlier), otubain-1 promotes T-cell activation when overexpressed in mouse T cells, whereas the overexpression of ARF1 inhibits T-cell activation⁵³. It should be noted, however, that a role for otubain-1 in the regulation of immune tolerance has not been examined in vivo.

An X-chromosome-encoded USP-family member, USP9X (also known as FAF), functions as a DUB that regulates the fate of ITCH⁸⁸, another E3 ligase that is required for the induction of T-cell tolerance⁸⁹. ITCH deficiency in mice causes the development of a lymphoproliferative disease associated with itching of the skin⁹⁰. Similar to GRAIL, ITCH undergoes self-ubiquitylation, which triggers its degradation by the proteasome⁸⁸. USP9X was identified as an ITCH-binding protein by tandem affinity purification and shown to inhibit the self-ubiquitylation and degradation of ITCH. The constitutive level of expression of ITCH is increased by overexpression of USP9X and decreased by RNAi-mediated USP9X knockdown⁸⁸. Whether USP9X regulates T-cell tolerance has not been investigated.

T-cell activation

Peripheral T-cell activation involves the formation of an intermediate signalling complex, composed of CARMA1 (caspase recruitment domain (CARD) membrane-associated guanylate kinase 1), BCL-10 and MALT1 (mucosa-associated-lymphoid-tissue lymphoma-translocation gene 1)⁹¹ (FIG. 2). On activation by the TCR, CARMA1 functions as a scaffold protein that recruits BCL-10 and MALT1 into the lipid raft, which in turn triggers the assembly of a ubiquitin-dependent signalosome containing the K63-specific E2 dimer, UBC13 (ubiquitin-conjugating enzyme 13)–UEV 1A (ubiquitin-conjugating enzyme E2 variant 1A), and probably also the E3 ligase TRAF6 (Refs 92,93). Within this signalosome, MALT1 is conjugated with K63-linked ubiquitin chains, which might function to recruit and activate TAK1 (Ref. 94), a kinase that mediates activation of the IKK complex and JNK in T cells^95–97. Activation of IKK by the CARMA1–BCL-10–MALT1 complex also requires K63-linked ubiquitylation of IKKγ, although how IKKγ ubiquitylation triggers the activity of IKKα and IKKβ is unclear^92,93. In addition, recent studies indicate that the catalytic activation of TAK1 requires its conjugation with polyubiquitin chains⁹⁸. Consistent with these biochemical results, genetic deficiency of Ubc13 results in the attenuation of TCR-mediated activation of TAK1 and downstream kinases⁹⁹. What is unclear, however, is which E3 ligase(s) mediates these K63-linked ubiquitylation events. MALT1 was shown to have E3 ligase activity and to catalyse K63-linked ubiquitylation of IKKγ (Ref. 93). By contrast, another study indicates that MALT1 induces IKKγ ubiquitylation by associating with and activating the E3 ligase TRAF6 (Ref. 92). Additional biochemical studies also support the role of TRAF6 as an E3 ligase in the TCR signalling pathway⁹⁴; however, this idea is challenged by the finding that T-cell-specific ablation of TRAF6 does not affect TCR-stimulated NF-κB activation, but rather promotes TCR signalling¹⁰⁰. It remains to be examined whether TRAF6 is functionally redundant with other TRAF-family members, such as TRAF2, in mediating TCR signalling.

Given the dynamic role of ubiquitylation in mediating the signalling function of the CARMA1–BCL-10–MALT1 complex, it is highly predictable that TCR signalling is subject to negative regulation by DUBs. So far, the role of A20 in the regulation of TCR signalling is unclear. Nevertheless, a recent study has shown that TCR and CD28 signals induce A20 cleavage in T cells, which is mediated by the paracaspase activity of MALT1 (Ref. 101). It is therefore possible that A20 has a negative role in regulating TCR signalling and that this function of A20 can be temporarily over-ridden during T-cell activation.

An important function of CYLD in controlling TCR signalling has been shown using Cyld^−/− mice^19,24 (FIG. 2). The loss of CYLD in T cells causes constitutive activation of TAK1 and its downstream signalling molecules, IKK and JNK (JUN N-terminal kinase). Although it is not yet clear precisely how CYLD regulates the function of the CARMA1–BCL-10–MALT1 signalosome, CYLD deubiquitylates TAK1 and thereby suppresses its catalytic activity¹⁹. The fact that loss of CYLD causes spontaneous activation of TAK1 implies that TAK1 ubiquitylation is a dynamic event that occurs even in resting T cells. However, it is unclear whether the activation of TAK1 in CYLD-deficient T cells bypasses the requirement for TCR-proximal signalling molecules.

In keeping with its crucial role in TAK1 regulation, CYLD is indispensable for maintaining normal T-cell activation. CYLD-deficient T cells are hyper-responsive to TCR and CD28 co-stimulation in vitro and have abnormalities in vivo^19,24. The Cyld-knockout mice were shown to either be predisposed to or spontaneously develop intestinal inflammation that was similar to human inflammatory bowel disease (IBD)^19,24, an autoimmune inflammatory disease caused by inappropriate response of the mucosal immune system to the intestinal microflora¹⁰². T cells might have an important role in this pathology, as the adoptive transfer of CYLD-deficient T cells to Rag1-knockout mice is sufficient to induce intestinal inflammation¹⁹. Notably, the CYLD gene lies adjacent to the NOD2 gene in both human and mouse genomes. NOD2 is known to regulate IBD by modulating the innate immune response¹⁰². As CYLD regulates T-cell responses, it is conceivable that NOD2 and CYLD might function cooperatively to prevent chronic mucosal immune responses to the intestinal microflora.